Browse the corpus

Gene expression profiling technologies are used to analyze gene networks whose expression is associated with specific pathological conditions compared with normal tissue.1 For instance, in 1999, the high expression of a specific group of genes was identified in highly proliferative breast tumor cells that were compared with normal breast tissue samples.2 The development of effective tools for large-scale gene expression analysis has already provided new insights into the involvement of gene networks and regulatory pathways in various tumoral processes.3 Complementary DNA microarrays can be used to test the expression of thousands of genes at once, while real-time reverse-transcription polymerase chain reaction (RT-PCR) offers more accurate and quantitative information on smaller numbers of selected candidate genes.4–6

regulatory pathways in various tumoral processes.3 Complementary DNA microarrays can be used to test the expression of thousands of genes at once, while real-time reverse-transcription polymerase chain reaction (RT-PCR) offers more accurate and quantitative information on smaller numbers of selected candidate genes.4–6 We hypothesized that the histologically normal tissue usually used as normal controls in gene expression studies obtained in two different ways (that is, percutaneous or surgical liver biopsies), might have different gene expression patterns. We suspected that an acute gene response might be observed during surgery because of aggression and stress, despite the absence of any macroscopic injury. To confirm this hypothesis, real-time quantitative RT-PCR was used to quantify the messenger RNA (mRNA) expression of a large number of selected genes in pooled A (histologically normal tissue obtained percutaneously) specimens compared with pooled B (histologically normal tissue obtained surgically) specimens. The expression level of 240 genes known to be involved in various cellular and molecular mechanisms associated with response to stress was examined. We especially focused on the expression of genes related to early stress response, hypoxia, and inflammation.7–12 Genes of interest were further investigated in 14 individual group A specimens compared with 14 individual group B specimens. We then investigated whether the choice of histologically normal controls could lead to discordance or misinterpretation of specific pathological conditions such as chronic hepatitis C.

We hypothesized that the histologically normal tissue usually used as normal controls in gene expression studies obtained in two different ways (that is, percutaneous or surgical liver biopsies), might have different gene expression patterns. We suspected that an acute gene response might be observed during surgery because of aggression and stress, despite the absence of any macroscopic injury. To confirm this hypothesis, real-time quantitative RT-PCR was used to quantify the messenger RNA (mRNA) expression of a large number of selected genes in pooled A (histologically normal tissue obtained percutaneously) specimens compared with pooled B (histologically normal tissue obtained surgically) specimens. The expression level of 240 genes known to be involved in various cellular and molecular mechanisms associated with response to stress was examined. We especially focused on the expression of genes related to early stress response, hypoxia, and inflammation.7–12 Genes of interest were further investigated in 14 individual group A specimens compared with 14 individual group B specimens. We then investigated whether the choice of histologically normal controls could lead to discordance or misinterpretation of specific pathological conditions such as chronic hepatitis C. Materials and Methods We selected liver samples on the basis of a histologically normal pattern: no portal or lobular inflammation and/or necrosis; absence of portal, central, or perisinusoidal fibrosis; and no other significant abnormal features (steatosis <5%, no iron overload, no ballooning or liver cell clarification, no cholestasis or bile duct lesion).

samples on the basis of a histologically normal pattern: no portal or lobular inflammation and/or necrosis; absence of portal, central, or perisinusoidal fibrosis; and no other significant abnormal features (steatosis <5%, no iron overload, no ballooning or liver cell clarification, no cholestasis or bile duct lesion). Group A Group A comprised percutaneous normal liver biopsy specimens, obtained from 14 adults with mildly elevated serum alanine aminotransferase activity addressed to Beaujon Hospital (Clichy, France), in whom all causes of liver disease had been ruled out (medication, alcohol, chronic viral hepatitis, autoimmune processes, and metabolic disease). In these adults, liver biopsies were performed percutaneously under local anesthesia. A transparietal biopsy of a normal liver is illustrated in Fig. 1A. Fig. 1 (A) Transparietal biopsy of normal liver. The normal portal tract and central vein are shown. Hepatocytes are arranged in regular trabeculae (hematoxylin-eosin staining; magnification ×25). (B) Surgical sample of normal liver. The lobular architecture is well-preserved; the normal portal tract is present in the center (hematoxylin-eosin staining; magnification ×25).

rmal portal tract and central vein are shown. Hepatocytes are arranged in regular trabeculae (hematoxylin-eosin staining; magnification ×25). (B) Surgical sample of normal liver. The lobular architecture is well-preserved; the normal portal tract is present in the center (hematoxylin-eosin staining; magnification ×25). Group B Surgical liver biopsies of nontumoral livers were obtained from 14 adults during operations for liver metastasis of colorectal cancer (n = 7) or benign liver tumors (n = 7) under systemic/general anesthesia. For the purpose of this study, we sampled tissue fragments at least 3 cm from the nearest metastasis. Neither fragment showed portal distorsion or expansion, ductular proliferation, or cholestasis that could suggest a mass effect. A surgical biopsy of a normal liver is illustrated in Figure 1B. All 28 liver tissue specimens from group A and group B were histologically normal (absence of inflammation, fibrosis, and pathological pattern). For all cases, one fragment was frozen and used for mRNA extraction and another was formalin-fixed and paraffin-embedded. All these samples were carefully reviewed by two liver pathologists and considered normal. Chronic Hepatitis C Patients Percutaneous liver biopsy specimens obtained from 55 chronic hepatitis C patients, selected from a cohort of untreated patients with chronic hepatitis C followed at Beaujon Hospital (Clichy, France), were graded and staged (Metavir),13 and the gene expression was studied (A1F1 [n = 11], A2F1 [n = 9], A1F2 [n = 10], A2F2 [n = 10], A2F3 [n = 15]).

pecimens obtained from 55 chronic hepatitis C patients, selected from a cohort of untreated patients with chronic hepatitis C followed at Beaujon Hospital (Clichy, France), were graded and staged (Metavir),13 and the gene expression was studied (A1F1 [n = 11], A2F1 [n = 9], A1F2 [n = 10], A2F2 [n = 10], A2F3 [n = 15]). The study was approved by the local ethics committee and conformed to the 1975 Declaration of Helsinki. All patients gave informed consent prior to liver biopsy. Large-Scale Real-Time RT-PCR Theoretical Basis Reactions are characterized by the point during cycling when amplification of the PCR product is first detected, rather than the amount of PCR product accumulated after a fixed number of cycles. The larger the starting quantity of the target molecule, the earlier a significant increase in fluorescence is observed. The parameter Ct (threshold cycle) is defined as the fractional cycle number at which the fluorescence generated by SYBR green dye–amplicon complex formation passes a fixed threshold above baseline. The increase in fluorescent signal associated with exponential growth of PCR products is detected by the laser detector of the ABI-Prism 7900 Sequence Detection System (PerkinElmer Applied Biosystems, Foster City, CA), using PE Biosystems analysis software according to the manufacturer's instructions.

d above baseline. The increase in fluorescent signal associated with exponential growth of PCR products is detected by the laser detector of the ABI-Prism 7900 Sequence Detection System (PerkinElmer Applied Biosystems, Foster City, CA), using PE Biosystems analysis software according to the manufacturer's instructions. The precise amount of total RNA added to each reaction mix (based on optical density) and its quality (that is, lack of extensive degradation) are both difficult to assess. We therefore also quantified transcripts of two endogenous RNA control genes involved in two cellular metabolic pathways, namely TBP (Genbank accession number NM_003194), which encodes the TATA box-binding protein (a component of the DNA-binding protein complex TFIID), and RPLP0 (also known as 36B4 [Genbank accession number NM_001002]), which encodes human acidic ribosomal phosphoprotein P0. Each sample was normalized on the basis of its TBP (or RPLPO) content. Results, expressed as N-fold differences in target gene expression relative to the TBP (or RPLPO) gene, and termed Ntarget, were determined as Ntarget = 2ΔCtsample, where the ΔCt value of the sample was determined by subtracting the average Ct value of the target gene from the average Ct value of the TBP (or RPLP0) gene. The Ntarget values of the samples were subsequently normalized such that the median value of the percutaneous normal liver specimen Ntarget was 1.

Results, expressed as N-fold differences in target gene expression relative to the TBP (or RPLPO) gene, and termed Ntarget, were determined as Ntarget = 2ΔCtsample, where the ΔCt value of the sample was determined by subtracting the average Ct value of the target gene from the average Ct value of the TBP (or RPLP0) gene. The Ntarget values of the samples were subsequently normalized such that the median value of the percutaneous normal liver specimen Ntarget was 1. Primers and Controls We suspected that, during surgery, as during aggression or stress, an acute gene response would be observed despite the absence of macroscopic injury. Based on a study of the literature describing early gene expression changes during aggression (associated with stress), we selected 240 genes involved in various cellular and molecular mechanisms associated with response to stress and during hepatic stellate cell activation, because these cells participate in the remodeling of injured livers.7–12 These genes encode proteins involved in the immune response, extracellular remodeling, oxidative stress, signal transduction pathways, cell cycle control, apoptosis, angiogenesis, interferon signaling, and so forth. Approximately 10 to 20 genes were selected per pathway (Fig. 2). Fig. 2 List of the 240 genes studied. Primers for TBP, RPLP0, and the 240 target genes were chosen with the assistance of the Oligo 5.0 computer program (National Biosciences, Plymouth, MN).

Primers and Controls We suspected that, during surgery, as during aggression or stress, an acute gene response would be observed despite the absence of macroscopic injury. Based on a study of the literature describing early gene expression changes during aggression (associated with stress), we selected 240 genes involved in various cellular and molecular mechanisms associated with response to stress and during hepatic stellate cell activation, because these cells participate in the remodeling of injured livers.7–12 These genes encode proteins involved in the immune response, extracellular remodeling, oxidative stress, signal transduction pathways, cell cycle control, apoptosis, angiogenesis, interferon signaling, and so forth. Approximately 10 to 20 genes were selected per pathway (Fig. 2). Fig. 2 List of the 240 genes studied. Primers for TBP, RPLP0, and the 240 target genes were chosen with the assistance of the Oligo 5.0 computer program (National Biosciences, Plymouth, MN). We conducted searches in the dbEST and nr databases to confirm the total gene specificity of the nucleotide sequences chosen as primers and the absence of single nucleotide polymorphisms. In particular, the primer pairs were selected to be unique relative to the sequences of closely related family member genes or of the corresponding retro-pseudogenes. To avoid amplification of contaminating genomic DNA, one of the two primers was placed at the junction between two exons, if possible. In general, amplicons were between 70 and 120 nucleotides long. Gel electrophoresis was used to verify the specificity of PCR amplicons.

genes or of the corresponding retro-pseudogenes. To avoid amplification of contaminating genomic DNA, one of the two primers was placed at the junction between two exons, if possible. In general, amplicons were between 70 and 120 nucleotides long. Gel electrophoresis was used to verify the specificity of PCR amplicons. For each primer pair, we performed no-template control and no-RT control (RT-negative) assays, which produced negligible signals (usually >40 in Ct value), suggesting that primer–dimer formation and genomic DNA contamination effects were negligible. RNA Extraction Total RNA was extracted from frozen liver tissue samples using the acid-phenol guanidinium method. The quality of the RNA samples was determined via electrophoresis through agarose gels and staining with ethidium bromide, the 18S and 28S RNA bands being visualized under ultraviolet light. Complementary DNA Synthesis Total RNA was reverse-transcribed in a final volume of 20 μL containing 1× RT buffer (500 μM each deoxyribonucleotide triphosphate, 3 mM MgCl2, 75 mM KCl, 50 mM Tris-HCl [pH 8.3]), 20 U RNasin ribonuclease inhibitor (Promega, Madison, WI), 10 mM dithiothreitol, 100 U Superscript II ribonuclease H reverse transcriptase (Invitrogen, Cergy Pontoise, France), 3 μM random hexamers (Pharmacia, Uppsala, Sweden), and 100 ng total RNA. The samples were incubated at 20°C for 10 minutes and 42°C for 30 minutes, and reverse-transcription was inactivated by heating at 99°C for 5 minutes and cooling at 5°C for 5 minutes.

e H reverse transcriptase (Invitrogen, Cergy Pontoise, France), 3 μM random hexamers (Pharmacia, Uppsala, Sweden), and 100 ng total RNA. The samples were incubated at 20°C for 10 minutes and 42°C for 30 minutes, and reverse-transcription was inactivated by heating at 99°C for 5 minutes and cooling at 5°C for 5 minutes. PCR Amplification All PCR reactions were performed using an ABI-Prism 7900 Sequence Detection System (PerkinElmer Applied Biosystems) and the SYBR Green PCR Core Reagents kit (PerkinElmer Applied Biosystems). Ten microliters of diluted sample complementary DNA (produced from 2 ng of total RNA) was added to 15 μL of the PCR master mix. The thermal cycling conditions comprised an initial denaturation step at 95°C for 10 minutes, and 50 cycles at 95°C for 15 seconds and 65°C for 1 minute. Strategy of Analysis First, two pools of five liver specimens from each group were respectively constituted by mixing aliquots of equivalent amounts of RNA from each of the liver samples. We then determined the mRNA expression level of the 240 genes in each pool. Genes whose expression differed between pools by at least three-fold in group B versus group A were selected. This robust selection criterion ensures the identification of genes of marked interest.

amounts of RNA from each of the liver samples. We then determined the mRNA expression level of the 240 genes in each pool. Genes whose expression differed between pools by at least three-fold in group B versus group A were selected. This robust selection criterion ensures the identification of genes of marked interest. The expression level of these selected genes was then assessed in each of the 28 individual samples. Comparison of the pool values with the mean individual values showed that RNA pooling was an appropriate initial screening approach, significantly limiting the required number of PCR experiments. Using the same approach, we have previously shown the involvement of several altered molecular pathways in the genesis of hepatitis C virus (HCV) infection,4 breast cancer,14 and hepatitis C liver fibrosis.5 Statistical Analysis Relationships between the molecular markers and histological parameters (in both group A and group B and in chronic hepatitis C) were tested using the nonparametric Mann-Whitney U test.15 Differences between the two populations were judged significant at confidence levels above 95% (P < 0.05). To visualize the capacity of a given molecular marker to discriminate between two populations (in the absence of an arbitrary cutoff value), we summarized the data in a receiver operating characteristic (ROC) curve.16

erences between the two populations were judged significant at confidence levels above 95% (P < 0.05). To visualize the capacity of a given molecular marker to discriminate between two populations (in the absence of an arbitrary cutoff value), we summarized the data in a receiver operating characteristic (ROC) curve.16 The mRNA levels indicated in Tables 1 and 2 (calculated as described in Materials and Methods) show the abundance of the target relative to the endogenous control (TBP) to normalize the starting amount and quality of total RNA. Similar results were obtained with a second endogenous control, RPLP0 (also known as 36B4) (data not shown). Table 1 Significantly Dysregulated Genes in Surgical Nontumoral Liver Patients Relative to Percutaneous Normal Liver Patients

The mRNA levels indicated in Tables 1 and 2 (calculated as described in Materials and Methods) show the abundance of the target relative to the endogenous control (TBP) to normalize the starting amount and quality of total RNA. Similar results were obtained with a second endogenous control, RPLP0 (also known as 36B4) (data not shown). Table 1 Significantly Dysregulated Genes in Surgical Nontumoral Liver Patients Relative to Percutaneous Normal Liver Patients Gene Symbols Alternate Symbols Gene Name Gene Characterization Percutaneous Normal Liver (n = 14) Surgical Nontumoral Liver (n = 14) P Value* ROC- AUC Significantly up-regulated genes in surgical nontumoral liver patients PAI1 SERPINE1 Plasminogen activator inhibitor-1 Extracellular matrix 1.0 (0.2-3.6)† 29.7 (9.5-83.9) 0.0000067 1.000 THBS1 TPS1 Thrombospondin-1 Extracellular matrix 1.0 (0.3-1.9) 12.4 (5.6-81.2) 0.0000067 1.000 IL8 Interleukin-8 Growth factor/cytokine 1.0 (0.6-2.1) 97.9 (3.8-434.7) 0.0000067 1.000 PTGS2 COX2 Prostaglandin-endoperoxide synthetase-2 Angiogenesis 1.0 (0.4-1.4) 11.1 (2.5-40.7) 0.0000067 1.000 CXCR4 Chemokine (C-X-C motif) receptor-4 Growth factor receptor 1.0 (0.3-1.6) 5.9 (2.1-19.4) 0.0000067 1.000 JUN Jun oncogene Transcription factor 1.0 (0.2-1.9) 14.0 (3.7-22.5) 0.0000067 1.000 FOS Fos oncogene Transcription factor 1.0 (0.3-14.8) 57.9 (23.3-220.9) 0.0000067 1.000 CCL2 MCP-1 Chemokine (C-C motif) ligand-2 Growth factor/cytokine 1.0 (0.5-2.3) 13.6 (1.2-40.1) 0.000014 0.982 SOCS3 SSI-3 Suppressor of cytokine signaling-3 (SSI-3) Signal transduction 1.0 (0.2-2.4) 28.5 (1.2-71.9) 0.000035 0.959 CXCL1 GRO1 Chemokine (C-X-C motif) ligand-1 Growth factor/cytokine 1.0 (0.4-2.2) 9.2 (0,9-56.0) 0.000043 0.954 HIF1A Hypoxia-inducible factor-1, alpha Angiogenesis 1.0 (0.4-1.5) 2.5 (1.1-6.1) 0.000053 0.949 MMP9 Matrix metalloproteinase-9 Extracellular matrix 1.0 (0.3-6.0) 15.0 (0.8-74.2) 0.000094 0.934 CTGF Connective tissue growth factor Growth factor/cytokine 1.0 (0.1-4.6) 5.4 (0.7-16.8) 0.00020 0.913 HAS2 Hyaluronan synthase-2 Extracellular matrix 1.0 (0.5-2.1) 11.0 (0.1-41.8) 0.00024 0.908 IL6 Interleukin-6 Growth factor/cytokine 1.0 (0.3-7.9) 58.9 (0.2-338.9) 0.00048 0.888 EGR1 KROX-24 Early growth response-1 (KROX-24) Transcription factor 1.0 (0.2-16.4) 7.1 (2.9-18.9) 0.00048 0.888 CCL3 MIP-1A Chemokine (C-C motif) ligand-3 (MIP-1A) Growth factor/cytokine 1.0 (0.5-4.0) 6.7 (0.5-21.5) 0.00048 0.888 CCL4 MIP-1B Chemokine (C-C motif) ligand-4 (MIP-1B) Growth factor/cytokine 1.0 (0.3-2.4) 2.9 (0.3-10.3) 0.0028 0.832 PAI2 SERPINB2 Plasminogen activator inhibitor-2 Extracellular matrix 1.0 (0.0-11.4) 26.7 (0.0-165.2) 0.0035 0.824 CDKN1A P21 C

3 (MIP-1A) Growth factor/cytokine 1.0 (0.5-4.0) 6.7 (0.5-21.5) 0.00048 0.888 CCL4 MIP-1B Chemokine (C-C motif) ligand-4 (MIP-1B) Growth factor/cytokine 1.0 (0.3-2.4) 2.9 (0.3-10.3) 0.0028 0.832 PAI2 SERPINB2 Plasminogen activator inhibitor-2 Extracellular matrix 1.0 (0.0-11.4) 26.7 (0.0-165.2) 0.0035 0.824 CDKN1A P21 C yclin-dependent kinase inhibitor1A (p21 protein) Cell cycle regulation 1.0 (0.2-3.4) 2.7 (0.5-9.3) 0.0041 0.819 LIF Leukemia inhibitory factor Growth factor/cytokine 1.0 (0.3-4.2) 5.1 (0.3-17.5) 0.0051 0.811 CRP C-reactive protein Hepatic secretory protein 1.0 (0.3-14.1) 14.9 (0.4-132.8) 0.0067 0.801 MMP2 Matrix metalloproteinase-2 Extracellular matrix 1.0 (0.4-3.0) 1.9 (0.6-24.0) 0.039 0.730 CXCL5 ENA78 Chemokine (C-X-C motif) ligand-5 Growth factor/cytokine 1.0 (0.1-18.6) 3.5 (0.3-143,8) NS 0.702 COL1A2 Collagen, type I, alpha-2 Extracellular matrix 1.0 (0.5-2.9) 1.1 (0.1-42.5) NS 0.594 IL1A Interleukin 1, alpha Growth factor/cytokine 1.0 (0.0-4.4) 0.9 (0.0-13.0) NS 0.582 COL1A1 Collagen, type I, alpha-1 Extracellular matrix 1.0 (0.0-2.6) 0.8 (0.4-86.4) NS 0.467 Significantly down-regulated genes in surgical nontumoral liver patients IHH Indian hedgehog homolog Growth factor/cytokine 1.0 (0.28-2.01) 0.06 (0.01-0.20) 0.0000067 0.000 GPT Alanine aminotransferase Metabolic enzyme 1.0 (0.31-2.78) 0.38 (0.07-1.29) 0.00048 0.112 IREG1 SLC11A3, HFE4 Ferroportin-1 Iron metabolism 1.0 (0.55-1.72) 0.54 (0.21-1.66) 0.003 0.171 CYP2E1 Cytochrome P450 CYP2E1 Metabolic enzyme 1.0 (0.49-2.47) 0.69 (0.30-1.48) NS 0.283 GFAP Glial fibrillary acidic protein Cytoskeletal 1.0 (0.34-8.94) 0.45 (0.22-14.94) NS 0.298 Abbreviations: AUC, area under the curve analysis; NS, not significant; ROC, receiver operating characteristics.

4 (0.21-1.66) 0.003 0.171 CYP2E1 Cytochrome P450 CYP2E1 Metabolic enzyme 1.0 (0.49-2.47) 0.69 (0.30-1.48) NS 0.283 GFAP Glial fibrillary acidic protein Cytoskeletal 1.0 (0.34-8.94) 0.45 (0.22-14.94) NS 0.298 Abbreviations: AUC, area under the curve analysis; NS, not significant; ROC, receiver operating characteristics. * Mann-Whitney U test. † Median (range) of gene mRNA levels. Table 2 Genes Perfectly Discriminated Between Percutaneous Normal Liver and Surgical Nontumoral Liver Patients According to Nature of the Adjacent Tumor (Benign Versus Malignant) in the Surgical Nontumoral Group

4 (0.21-1.66) 0.003 0.171 CYP2E1 Cytochrome P450 CYP2E1 Metabolic enzyme 1.0 (0.49-2.47) 0.69 (0.30-1.48) NS 0.283 GFAP Glial fibrillary acidic protein Cytoskeletal 1.0 (0.34-8.94) 0.45 (0.22-14.94) NS 0.298 Abbreviations: AUC, area under the curve analysis; NS, not significant; ROC, receiver operating characteristics. * Mann-Whitney U test. † Median (range) of gene mRNA levels. Table 2 Genes Perfectly Discriminated Between Percutaneous Normal Liver and Surgical Nontumoral Liver Patients According to Nature of the Adjacent Tumor (Benign Versus Malignant) in the Surgical Nontumoral Group Gene Symbols Alternate Symbols Gene Name Gene Characterization Percutaneous Normal Liver (n = 14) Surgical Nontumoral Liver (n = 14) Surgical Nontumoral Liver Patients Adjacent to Benign (n = 7) Surgical Nontumoral Liver Patients Adjacent to Malignant (n = 7) P Value* ROC- AUC Genes up-regulated in surgical nontumoral liver patients PAI1 SERPINE1 Plasminogen activator inhibitor-1 Extracellular matrix 1.0 (0.2-3.6)† 29.7 (9.5-83.9) 19.0 (9.5-83.9) 31.1 (9.6-46.5) NS 0.633 THBS1 TPS1 Thrombospondin-1 Extracellular matrix 1.0 (0.3-1.9) 12.4 (5.6-81.2) 7.6 (5.6-81.2) 19.9 (9.4-25.7) NS 0.816 IL8 Interleukin-8 Growth factor/cytokine 1.0 (0.6-2.1) 97.9 (3.8-434.7) 80.1 (11.7-434.7) 115.7 (3.8-381.1) NS 0.388 PTGS2 COX2 Prostaglandin-endoperoxide synthetase-2 Angiogenesis 1.0 (0.4-1.4) 11.1 (2.5-40.7) 12.9 (5.5-34.4) 9.3 (2.5-40.7) NS 0.816 CXCR4 Chemokine (C-X-C motif) receptor-4 Growth factor receptor 1.0 (0.3-1.6) 5.9 (2.1-19.4) 6.0 (2.1-12.4) 4.2 (2.4-19.4) NS 0.490 JUN Jun oncogene Transcription factor 1.0 (0.2-1.9) 14.0 (3.7-22.5) 13.1 (6.2-21.5) 14.9 (3.7-22.5) NS 0.510 FOS Fos oncogene Transcription factor 1.0 (0.3-14.8) 57.9 (23.3-220.9) 69.0 (23.3-220.9) 38.6 (29.6-111.3) NS 0.388 Genes down-regulated in surgical nontumoral liver patients IHH Indian Hedgehog homolog Growth factor/cytokine 1.0 (0.28-2.01) 0.06 (0.01-0.20) 0.07 (0.02-0.20) 0.06 (0.01-0.16) NS 0.378 Abbreviations: AUC, area under the curve analysis; NS, not significant; ROC, receiver operating characteristics.

(29.6-111.3) NS 0.388 Genes down-regulated in surgical nontumoral liver patients IHH Indian Hedgehog homolog Growth factor/cytokine 1.0 (0.28-2.01) 0.06 (0.01-0.20) 0.07 (0.02-0.20) 0.06 (0.01-0.16) NS 0.378 Abbreviations: AUC, area under the curve analysis; NS, not significant; ROC, receiver operating characteristics. * Mann -Whitney U test (venign versus malignant). † Median (range) of gene mRNA levels. Results mRNA Expression of the 240 Genes in the Group B Pool Sample Relative to the Group A Pool Sample The mean TBP gene Ct (threshold cycle) values for the group A pool and the group B pool were 25.23 ± 0.24 and 25.43 ± 0.23, respectively. Seven (2.9%) of the 240 genes were detectable but not reliably quantifiable in both the group B and group A pools (Ct > 32). The mRNA expression of 32 (13.7%) of the remaining 233 genes showed at least a three-fold difference between the two pools; 27 (84.4%) genes were up-regulated and 5 (15.6%) were down-regulated in the group B pool sample compared with the goup A pool sample. mRNA Expression of the 32 Dysregulated Genes in 14 Group B Samples and 14 Group A Samples The expression level of the 32 dysregulated genes identified via pooled sample analysis was then determined individually in the 14 group B samples and 14 group A samples.

Seven (2.9%) of the 240 genes were detectable but not reliably quantifiable in both the group B and group A pools (Ct > 32). The mRNA expression of 32 (13.7%) of the remaining 233 genes showed at least a three-fold difference between the two pools; 27 (84.4%) genes were up-regulated and 5 (15.6%) were down-regulated in the group B pool sample compared with the goup A pool sample. mRNA Expression of the 32 Dysregulated Genes in 14 Group B Samples and 14 Group A Samples The expression level of the 32 dysregulated genes identified via pooled sample analysis was then determined individually in the 14 group B samples and 14 group A samples. Twenty-three (85.2%) of the 27 up-regulated genes identified by pooled sample analysis were significantly up-regulated in the 14 group B samples compared with the 14 group A samples (P < 0.05; Table 1). Three (60%) of the five down-regulated genes identified via pooled sample analysis were significantly down-regulated in the 14 group B samples compared with the 14 group A samples (P < 0.05; Table 1). The 23 up-regulated genes mainly encoded proteins involved in immune response (interferon pathway, growth factor, growth factor receptor, cytokine: IL8, CXCR4, CCL2, CXCL1, IL6, CCL3, CCL4, LIF, CXCL5, IL1A); and matrix remodeling (angiogenesis, extracellular matrix, extracellular matrix protease, inhibitors of matrix protease: PAI1, THBS1, PTGS2, HIF1A, MMP9, CTGF, HAS2, PAI2, MMP2, and COL1A1).

interferon pathway, growth factor, growth factor receptor, cytokine: IL8, CXCR4, CCL2, CXCL1, IL6, CCL3, CCL4, LIF, CXCL5, IL1A); and matrix remodeling (angiogenesis, extracellular matrix, extracellular matrix protease, inhibitors of matrix protease: PAI1, THBS1, PTGS2, HIF1A, MMP9, CTGF, HAS2, PAI2, MMP2, and COL1A1). The capacity of each of these 26 dysregulated genes (23 up-regulated and 3 down-regulated) to discriminate between group B and group A samples was then tested via ROC curve analysis. The overall diagnostic values of the 26 molecular markers were assessed in terms of their area under the curve (AUC) values (Table 1). Eight genes perfectly discriminated between groups A and B (AUC-ROC, 1.000): seven up-regulated genes (PAI1, THBS1, IL8, PTGS2, CXCR4, JUN, and FOS) and one down-regulated gene (IHH). Fig. 3 shows the mRNA levels of three of these genes (PAI1, THBS1, and IHH) in each of the 14 group B samples and the 14 group A samples. Fig. 3 Shown are the mRNA levels of three perfectly discriminatory genes (PAI1, THBS1, IHH) in the 14 percutaneous normal liver samples and the 14 surgical nontumoral liver samples. The median value (range) is indicated for each subgroup. NL, normal liver. Among the eight genes that discriminated perfectly between the group B and group A samples, there was no significant difference in samples from group B when they were compared for the nature (that is, benign or malignant) of the distant tumor (Table 2).

Fig. 3 Shown are the mRNA levels of three perfectly discriminatory genes (PAI1, THBS1, IHH) in the 14 percutaneous normal liver samples and the 14 surgical nontumoral liver samples. The median value (range) is indicated for each subgroup. NL, normal liver. Among the eight genes that discriminated perfectly between the group B and group A samples, there was no significant difference in samples from group B when they were compared for the nature (that is, benign or malignant) of the distant tumor (Table 2). mRNA Expression of IL8 in Different Stage of Chronic Hepatitis C in Comparison with Group B Samples and Group A Samples To determine whether the choice of histologically normal controls could lead to discordance or misinterpretation of specific pathological conditions such as chronic hepatitis C, we measured one (IL8) of the eight perfectly discriminating genes in five series of various grades of necroinflammation and stages of liver fibrosis (A1F1, A2F1, A1F2, A2F2, A2F3).

tologically normal controls could lead to discordance or misinterpretation of specific pathological conditions such as chronic hepatitis C, we measured one (IL8) of the eight perfectly discriminating genes in five series of various grades of necroinflammation and stages of liver fibrosis (A1F1, A2F1, A1F2, A2F2, A2F3). IL8 was investigated because it has been shown in culture cells that the HCV nonstructural 5A protein induces IL8.17 IL8 mRNA expression increases from mild chronic hepatitis C (A1F1) to severe liver lesions (A2F3) (Fig. 4). The results show an underexpression or overexpression of specific genes (such as IL8) in HCV infection depending on whether the controls were obtained percutaneously or surgically. It is interesting to note that in this example, group A seems to be the more appropriate control, because an increase in IL8 mRNA levels from mild (A1F1) to advanced disease (A2F3) is observed, suggesting a model with IL8 activation during fibrogenesis. Fig. 4 IL8 expression in different grade of necroinflammation and stage of fibrosis in chronic hepatitis C as compared with group B or group A.

IL8 was investigated because it has been shown in culture cells that the HCV nonstructural 5A protein induces IL8.17 IL8 mRNA expression increases from mild chronic hepatitis C (A1F1) to severe liver lesions (A2F3) (Fig. 4). The results show an underexpression or overexpression of specific genes (such as IL8) in HCV infection depending on whether the controls were obtained percutaneously or surgically. It is interesting to note that in this example, group A seems to be the more appropriate control, because an increase in IL8 mRNA levels from mild (A1F1) to advanced disease (A2F3) is observed, suggesting a model with IL8 activation during fibrogenesis. Fig. 4 IL8 expression in different grade of necroinflammation and stage of fibrosis in chronic hepatitis C as compared with group B or group A. mRNA Expression of Other Genes Involved in the Hedgehog-Gli Signaling Pathway in Group B and Group A Samples The only down-regulated gene that perfectly discriminated between groups A and B (IHH) is involved in the Hedgehog-Gli signaling pathway. To further explore the Hedgehog-Gli signaling pathway to discriminate between groups A and B, we tested the expression of six additional genes involved in this pathway (DHH, SHH, GLI1, GLI2, GLI3, and GLI4) in three high IHH-expressing percutaneous normal liver samples and three low IHH-overexpressing surgical nontumoral liver samples. The results are summarized in Fig. 5. DHH transcripts were detectable but not reliably quantifiable in both the group B and group A samples (Ct > 32). Total positive associations (AUC-ROC, 1.000) were found between IHH and three of the five expressed genes (SHH, GLI1, and GLI4).

ssing surgical nontumoral liver samples. The results are summarized in Fig. 5. DHH transcripts were detectable but not reliably quantifiable in both the group B and group A samples (Ct > 32). Total positive associations (AUC-ROC, 1.000) were found between IHH and three of the five expressed genes (SHH, GLI1, and GLI4). Fig. 5 Expression level of Hedgehog/Gli genes in three high IHH-expressing percutaneous normal liver samples and three low IHH-expressing surgical nontumoral liver samples. NL, normal liver. Discussion Gene expression profiling technologies are used to analyze gene networks whose expression is associated with specific pathological conditions compared with normal tissue.1 Generally, normal tissue for normal controls is obtained in various ways, including percutaneous and surgical biopsy.1–2,5,18

Fig. 5 Expression level of Hedgehog/Gli genes in three high IHH-expressing percutaneous normal liver samples and three low IHH-expressing surgical nontumoral liver samples. NL, normal liver. Discussion Gene expression profiling technologies are used to analyze gene networks whose expression is associated with specific pathological conditions compared with normal tissue.1 Generally, normal tissue for normal controls is obtained in various ways, including percutaneous and surgical biopsy.1–2,5,18 This study focused on the gene expression changes observed in the histologically normal liver in relation to the sampling method (percutaneous or surgical liver biopsy). We analyzed the gene transcriptional profiles of percutaneous normal liver specimens, obtained under local anesthesia from 14 adults with mildly elevated serum alanine aminotransferase activity in whom all causes of liver disease had been ruled out (medication, alcohol, chronic viral hepatitis, autoimmune processes, and metabolic disease) compared with nontumoral liver biopsies obtained from 14 adults during surgery for liver metastasis of colorectal cancer or benign liver tumors. All 28 liver tissue specimens (groups A and B) were histologically normal. For our study, we selected liver samples based on a histological normal aspect carefully analyzed by two liver pathologists.

liver biopsies obtained from 14 adults during surgery for liver metastasis of colorectal cancer or benign liver tumors. All 28 liver tissue specimens (groups A and B) were histologically normal. For our study, we selected liver samples based on a histological normal aspect carefully analyzed by two liver pathologists. The 26 genes that were significantly dysregulated (23 up-regulated and three down-regulated) in the group B samples mainly encoded proteins involved in immune response (interferon pathway, growth factor, growth factor receptor, cytokine: IL8, CXCR4, CCL2, CXCL1, IL6, CCL3, CCL4, LIF) and matrix remodeling (angiogenesis, extracellular matrix, extracellular matrix protease, inhibitors of matrix protease: PAI1, THBS1, PTGS2, HIF1A, MMP9, CTGF, HAS2, PAI2, and MMP2). The gene up-regulations in the surgical nontumoral biopsies were not due to tumor cell contamination or stroma cell activation, because similar expression levels were observed in the normal liver samples associated with distant malignant tumors compared with those associated with distant benign tumors.

PAI2, and MMP2). The gene up-regulations in the surgical nontumoral biopsies were not due to tumor cell contamination or stroma cell activation, because similar expression levels were observed in the normal liver samples associated with distant malignant tumors compared with those associated with distant benign tumors. Most of these genes belong to the acute phase response family and are up-regulated after “stress.”19 All living organisms need to sense and respond to conditions that stress their homeostatic mechanisms. The liver plays a central role in the body's response to injury.20 Expression of hepatic acute-phase and heat-shock genes probably contributes to restoring homeostasis after surgical procedures. Activation of the acute phase response can be due to different causes, such as hypoxemia, infection, surgery, and anesthesia. The acute phase response gene family includes and/or interacts with numerous family genes (inflammation, cytokines, extracellular matrix, and so forth). Systemic stressors can lead to regeneration.12 Hypoxia—a reduction in the normal level of tissue oxygen tension—occurs during acute and chronic vascular diseases, pulmonary disease and cancer.21 Another type of hypoxia known as acute or perfusion-limited hypoxia occurs when aberrant blood vessels are shut down, which also causes a reverse in blood flow. Closed vessels can be reopened, leading to reperfusion of hypoxic tissue with oxygenated blood. This leads to an increase in free radical concentrations, tissue damage, and activation of stress-response genes—a process known as reoxygenation injury. It should be noted that dysregulation of HIF1A, a gene playing a major role in hypoxia, was observed in this study.

perfusion of hypoxic tissue with oxygenated blood. This leads to an increase in free radical concentrations, tissue damage, and activation of stress-response genes—a process known as reoxygenation injury. It should be noted that dysregulation of HIF1A, a gene playing a major role in hypoxia, was observed in this study. What about surgical liver biopsies under general anesthesia? General anesthetics are known to transiently increase expression of mRNAs of immediate-early genes in the brain.22 Furthermore, anesthesia has been shown to mimic ischemic preconditioning,23 the process by which brief exposure to ischemia provides robust protection or tolerance against the injurious effects of longer-term ischemia via expression of acute phase response genes. Among the 26 dysregulated genes identified in this study, eight perfectly discriminated between groups A and B (AUC-ROC, 1.000): seven up-regulated genes (PAI1, THBS1, PTGS2, CXCR4, JUN, FOS, and IL8) all involved in the acute phase response, and one down-regulated gene (IHH) that codes one of the three mammalian Hedgehog (Hh) proteins playing a major role in vertebrate development and tumorigenesis.

ated between groups A and B (AUC-ROC, 1.000): seven up-regulated genes (PAI1, THBS1, PTGS2, CXCR4, JUN, FOS, and IL8) all involved in the acute phase response, and one down-regulated gene (IHH) that codes one of the three mammalian Hedgehog (Hh) proteins playing a major role in vertebrate development and tumorigenesis. THBS1, PAI1 THBS1 and PAI1 code molecules involved in matrix turnover. Thrombospondins form a family of secreted glycoproteins with pleiotropic functions and widespread expression.24–26 THBS1 is involved in the regulation of cellular responses to injury. It has been shown that THBS1 acts as a strong promoter of transforming growth factor β effects in hepatic stellate cells.27 Plasminogen activator inhibitor-1 (PAI-1) is the main physiological inhibitor of both the urokinase-type plasminogen activator and the tissue plasminogen activator and thereby plays an important role in regulation of the fibrinolytic system. PAI-1 has also been reported to act as an acute phase protein,28 and plasma PAI-1 levels rise markedly during disease states often associated with an acute phase response, including trauma, surgical procedures, and burn injury. The inflammatory response is a nonspecific reaction of the human body to trauma, injury, or infection, and the liver is a major site for synthesis of inflammatory and procoagulant mediators, including C-reactive protein, fibrinogen, interleukin-6, and PAI-1.29 CXCR4, IL8, and PTGS2 CXCR4, IL8, and PTGS2 code molecules involved in angiogenesis and inflammation.

THBS1, PAI1 THBS1 and PAI1 code molecules involved in matrix turnover. Thrombospondins form a family of secreted glycoproteins with pleiotropic functions and widespread expression.24–26 THBS1 is involved in the regulation of cellular responses to injury. It has been shown that THBS1 acts as a strong promoter of transforming growth factor β effects in hepatic stellate cells.27 Plasminogen activator inhibitor-1 (PAI-1) is the main physiological inhibitor of both the urokinase-type plasminogen activator and the tissue plasminogen activator and thereby plays an important role in regulation of the fibrinolytic system. PAI-1 has also been reported to act as an acute phase protein,28 and plasma PAI-1 levels rise markedly during disease states often associated with an acute phase response, including trauma, surgical procedures, and burn injury. The inflammatory response is a nonspecific reaction of the human body to trauma, injury, or infection, and the liver is a major site for synthesis of inflammatory and procoagulant mediators, including C-reactive protein, fibrinogen, interleukin-6, and PAI-1.29 CXCR4, IL8, and PTGS2 CXCR4, IL8, and PTGS2 code molecules involved in angiogenesis and inflammation. Stromal cell–derived factor-1 is a member of the C-X-C motif (CXC) chemokine family that binds to the seven-span transmembrane G-protein–coupled CXCR4 receptor, which has stromal cell–derived factor-1 as its unique ligand.30 CXCR4 is expressed by most leukocyte populations, endothelial cells, as well as epithelial and carcinomatous cells. In a recent study, hepatic regeneration was induced by treating rats with 2-acetylaminofluorene and followed by partial hepatectomy.31 CXCR4 mRNA expression, assessed by both quantitative RT-PCR and in situ hybridization, was increased during hepatic regeneration.

l cells, as well as epithelial and carcinomatous cells. In a recent study, hepatic regeneration was induced by treating rats with 2-acetylaminofluorene and followed by partial hepatectomy.31 CXCR4 mRNA expression, assessed by both quantitative RT-PCR and in situ hybridization, was increased during hepatic regeneration. PTGS2, also called COX-2, plays an important role in tumor and endothelial cell biology. Increased expression of PTGS2 occurs in multiple cells within the tumor microenvironment, which can affect angiogenesis. PTGS2 appears to play a key role in the release and activity of proangiogenic proteins.32 Interleukin-8, a cytokine of the CXC chemokine family, plays an important role in tumor progression and metastasis in a variety of human cancers, including lung cancers.33 Interleukin-8 biological activity in tumors and the tumor microenvironment may contribute to tumor progression through its potential function in the regulation of angiogenesis, cancer cell growth and survival, tumor cell motion, leukocyte infiltration, and modification of immune responses.

rs, including lung cancers.33 Interleukin-8 biological activity in tumors and the tumor microenvironment may contribute to tumor progression through its potential function in the regulation of angiogenesis, cancer cell growth and survival, tumor cell motion, leukocyte infiltration, and modification of immune responses. IL8 mRNA expression increases from mild chronic hepatitis C (A1F1) to severe liver lesions (A2F3). In prior immunohistochemical studies of HCV infection, IL8 protein was shown to be expressed in infiltrating cells in the portal tract and fibrotic septa and within hepatic lobules in patients.34 We have previously reported that there was a correlation between intrahepatic mRNA IL8 expression and hepatic fibrosis in HCV patients.5 Moreover, exposure of human umbilical vein endothelial cells to HCV-like particles resulted in increased IL8 production.35

and fibrotic septa and within hepatic lobules in patients.34 We have previously reported that there was a correlation between intrahepatic mRNA IL8 expression and hepatic fibrosis in HCV patients.5 Moreover, exposure of human umbilical vein endothelial cells to HCV-like particles resulted in increased IL8 production.35 JUN, FOS The AP-1 transcription factor is mainly composed of Jun, Fos, and/or ATF protein heterodimers. AP-1 mediates gene regulation in response to a plethora of physiological and pathological stimuli, including cytokines, growth factors, stress signals, and bacterial and viral infections, as well as oncogenic stimuli.36 Interestingly, a rat model after portal branch ligation produced atrophy of the deprived lobes (70% of the liver parenchyma), whereas the perfused lobes undergo compensatory regeneration; c-fos and c-jun expression were elevated during the first 2 hours in all the compartments.37 These findings suggest that the cellular and molecular changes that occur early in a regenerating liver are nonspecific, possibly stress-induced cellular responses. They do not indicate future progression toward atrophy or regeneration.

c-jun expression were elevated during the first 2 hours in all the compartments.37 These findings suggest that the cellular and molecular changes that occur early in a regenerating liver are nonspecific, possibly stress-induced cellular responses. They do not indicate future progression toward atrophy or regeneration. IHH and the Mammalian Hedgehog Proteins Among the 26 dysregulated genes, we identified eight that perfectly discriminated between group A and group B (AUC-ROC, 1.000), one of which is a down-regulated gene (IHH) that codes one of the three mammalian Hedgehog (Hh) proteins. Alteration of this unexpected pathway was confirmed via identification of an alteration of additional genes involved in this signaling pathway (one additional ligand [SHH] and two transcriptional factors [GLI1] and [GLI4]). The Hh pathway has been shown to direct the fate of neural and myofibroblastic cells during embryogenesis and during tissue remodeling in adults.38–39 Recent studies suggest a major role for the Hh pathway in hepatic stellate cell activation and viability40 and in the maintenance of hepatic progenitors during fetal development and adulthood.41 Fatty liver injury alters Hh activity in liver progenitors, and this might promote epithelial–mesenchymal transitions that result in liver fibrosis.42 Hh dysregulation is also observed in human hepatocarcinogenesis.43 Our results regarding Hh signaling could suggest qualitative or quantitative variations in hepatic stellate cells and/or hepatic progenitors between percutaneous and surgical normal liver tissues.

mesenchymal transitions that result in liver fibrosis.42 Hh dysregulation is also observed in human hepatocarcinogenesis.43 Our results regarding Hh signaling could suggest qualitative or quantitative variations in hepatic stellate cells and/or hepatic progenitors between percutaneous and surgical normal liver tissues. This study demonstrates that histologically normal liver tissue obtained in two different ways (percutaneous or surgical liver biopsy) has different gene expression patterns, though all specimens are histologically normal. The most notable changes in gene expression mainly occurred in the inflammatory response gene family. Therefore, this study emphasizes the importance of an adequate selection of histologically normal controls to prevent discordant or false results in gene expression profile analysis. It is difficult to state which is the best histologically normal control. In any study, the appropriate histological normal control should be obtained in the same technical way as the pathological sample. For instance, in a study of chronic hepatitis C in which liver samples are obtained percutaneously, the histologically normal samples should be obtained through percutaneous liver biopsy.4–6 In all cases, the controls used should be clearly described. Finally, the careful selection of controls is crucial, since the wrong selection could lead to misinterpretation of results. Abbreviations AUCarea under the curve Ctcycle threshold HCVhepatitis C virus HhHedgehog mRNAmessenger RNA PAI-1plasminogen activator inhibitor-1 PCRpolymerase chain reaction ROCreceiver operating characteristic

It is difficult to state which is the best histologically normal control. In any study, the appropriate histological normal control should be obtained in the same technical way as the pathological sample. For instance, in a study of chronic hepatitis C in which liver samples are obtained percutaneously, the histologically normal samples should be obtained through percutaneous liver biopsy.4–6 In all cases, the controls used should be clearly described. Finally, the careful selection of controls is crucial, since the wrong selection could lead to misinterpretation of results. Abbreviations AUCarea under the curve Ctcycle threshold HCVhepatitis C virus HhHedgehog mRNAmessenger RNA PAI-1plasminogen activator inhibitor-1 PCRpolymerase chain reaction ROCreceiver operating characteristic RT-PCRreverse-transcription polymerase chain reaction.

Persistent HCV infection can lead to liver disease such as hepatic steatosis, fibrosis, cirrhosis, and hepatocellular carcinoma, and is the leading indication for liver transplants in the United States.1 Despite this, the mechanisms of disease progression are poorly understood. With the development of the infectious hepatitis C virus (HCV) cell culture system (HCVcc),2-4 it became possible to study the entire virus infectious cycle and its effect on cellular gene and protein expression. Understanding the changes brought about by viral infection at the host cell level will allow a better insight into how current therapies work and to focus new therapeutics to the most promising areas. We used three techniques to investigate gene expression and proteomics changes following HCV infection in Huh 7.5 cells. We used the novel Solexa system (RNA-Seq) which uses whole-genome RNA sequencing technology5 to compare gene expression levels in infected and uninfected cells. RNA-Seq technology is likely to replace microarray technology as costs decrease.6, 7 The technology is reliable and reproducible8 and is increasingly used in transcriptome analysis.9, 10 However, RNA-Seq has not previously been used to study the impact of viral infection. We compared this method to conventional Affymetrix gene chip microarray, and two-dimensional gel electrophoresis (2DE)-based proteomics. In this multianalysis approach, we identified thousands of differentially expressed genes and proteins that allowed the dissection of the effects of HCV infection on a number of biofunctions and canonical pathways. These effects could have significant implications for HCV pathogenesis: if the profound cellular and metabolic modifications observed using the genotype 2 HCVcc system in vitro are confirmed in vivo and in different HCV genotypes, they could impact on disease pathogenesis and response to interferon treatment.11

thways. These effects could have significant implications for HCV pathogenesis: if the profound cellular and metabolic modifications observed using the genotype 2 HCVcc system in vitro are confirmed in vivo and in different HCV genotypes, they could impact on disease pathogenesis and response to interferon treatment.11 Materials and Methods Infection of Huh 7.5 cells with Jc1 HCV and X-31 Huh 7.5 cells were infected with Gt2a HCV J6CF-JFH1 (Jc1) at a multiplicity of infection (moi) of 0.02, or with X-31 influenza at moi of 1, or mock-infected with media, cultured as described12 and harvested when infection levels reached ≥ 90% (postinfection day 10). Immunofluorescence Huh 7.5 cells were fixed with paraformaldehyde, permeabilized with Triton X-100 and blocked with milk/phosphate-buffered saline (PBS) solution. Cells were subsequently incubated with anti-HCV core primary antibody (Cambridge Biosciences), followed by anti-mouse fluorescein isothiocyanate (Sigma). Each step was followed by PBS washes. DNA Microarray Analysis When HCV infection levels reached ≥ 90% total RNA was extracted from four infected and four noninfected replicates of Huh 7.5 cells, using the RNAeasy Mini Kit (Qiagen). Samples were prepared using the Affymetrix GeneChip WT sense target labeling and control reagents kit, and hybridized to the Affymetrix GeneChip Human Gene 1.0 ST Array containing 28,869 well-annotated genes. Chips were scanned on an Affymetrix Fluidics Station 450 and Scanner 3000.

using the RNAeasy Mini Kit (Qiagen). Samples were prepared using the Affymetrix GeneChip WT sense target labeling and control reagents kit, and hybridized to the Affymetrix GeneChip Human Gene 1.0 ST Array containing 28,869 well-annotated genes. Chips were scanned on an Affymetrix Fluidics Station 450 and Scanner 3000. Arrays were PLIER normalized and genes clustered in GeneSpring GX 9 (Agilent) using a Condition Tree and a Spearman correlation. Huh 7.5 cells were clustered into HCV infected and uninfected groups. Differentially expressed genes were identified using a Welch t test with a P value cut off of ≤0.05 and a fold-change difference between treatments of ≥1.5. Gene interaction networks and canonical pathways were analyzed using Ingenuity Pathways Analysis (IPA).13 RNA-Seq Analysis RNA was extracted from HCV infected and noninfected cells in the same way as for the microarray experiment. The poly-A containing messenger RNA molecules were purified using poly-T oligo-attached magnetic beads (Invitrogen). The messenger RNA was fragmented using divalent cations under elevated temperature (Ambion), and copied into first-strand complementary DNA (cDNA) using reverse transcriptase and random hexamer primers. Second strand cDNA synthesis was carried out using DNA polymerase I and RNase H. The cDNA fragments were prepared for sequencing on the Illumina Genome Analyzer using the Genomic DNA sequencing Sample Prep Kit (Illumina).

into first-strand complementary DNA (cDNA) using reverse transcriptase and random hexamer primers. Second strand cDNA synthesis was carried out using DNA polymerase I and RNase H. The cDNA fragments were prepared for sequencing on the Illumina Genome Analyzer using the Genomic DNA sequencing Sample Prep Kit (Illumina). The analyzer identified gene names backed up by a count of the number of times it appears. The number of counts and the Illumina counting tool determines fold-changes between the different samples. For samples with a fold-change of 1.5-2, we used a cutoff of 50 counts, for fold-change of >2 we used a cutoff of >15 counts and >8 counts for a fold-change >4. Gene interactions were analyzed with IPA.13

appears. The number of counts and the Illumina counting tool determines fold-changes between the different samples. For samples with a fold-change of 1.5-2, we used a cutoff of 50 counts, for fold-change of >2 we used a cutoff of >15 counts and >8 counts for a fold-change >4. Gene interactions were analyzed with IPA.13 Proteomic Analysis Sample analyses from HCV-infected (≥90%) and uninfected Huh 7.5 cells were analyzed using 2DE (n = 4) as previously detailed,14 except a 1.5-fold cutoff was used. Protein spots of interest were excised and digested in-gel. Tryptic peptides were eluted and analyzed by a Micromass Q-ToF liquid chromatography tandem mass spectrometry (LC-MS/MS) system (Micromass). Spectra processed using ProteinLynx Global Server (Waters) generated “.pkl” files which were searched against SwissProt version 56.9 using Mascot Daemon version 2.1 (Matrix Science). Searches were restricted to human taxonomy (20402 sequences) with carbamideomethyl cysteine as a fixed, and oxidized methionine as a variable modification. For confident protein identification, peptide ion cutoffs were chosen to include peptides showing identity or extensive homology (P < 0.05), and all data were checked manually.

e restricted to human taxonomy (20402 sequences) with carbamideomethyl cysteine as a fixed, and oxidized methionine as a variable modification. For confident protein identification, peptide ion cutoffs were chosen to include peptides showing identity or extensive homology (P < 0.05), and all data were checked manually. Detection of Sjogren Syndrome Antigen B Huh 7.5 cells were infected with HCV as above and infected and noninfected cells were lysed with sample buffer.14 Equal amounts of protein were loaded onto a 4%-12% precast sodium dodecyl sulfate polyacrylamide gel electrophoresis gel (Invitrogen); after western blotting, the presence of Sjogren syndrome antigen B (SSB) was probed for using mouse anti-SSB (Abnova). Effect of AY9944 on HCV Huh 7.5 cells were pretreated for 8 hours in six-well plates with or without AY9944 trans-1,4-bis(2-chlorobenzylaminomethyl)cyclohexane, (Calbiochem) at 2.5, 1.25, and 0.6125 μM. After 8 hours, the cells were infected with HCV at moi = 0.2 and grown in the continued presence of the inhibitor for 3 days, after which infected cells were detected by immunofluorescence and foci were counted.

ithout AY9944 trans-1,4-bis(2-chlorobenzylaminomethyl)cyclohexane, (Calbiochem) at 2.5, 1.25, and 0.6125 μM. After 8 hours, the cells were infected with HCV at moi = 0.2 and grown in the continued presence of the inhibitor for 3 days, after which infected cells were detected by immunofluorescence and foci were counted. Glucose Assay Cells were infected as above with either HCV or with influenza virus for 24 hours. HCV-infected and mock-infected cells were also subjected to 3 days treatment with 100 or 1000 IU/mL interferon (Sigma). Cells were harvested by trypsinization when infection levels reached ≥ 90%, and washed with PBS. Cells were lysed in radioimmunoprecipitation assay buffer and centrifuged for 5 minutes at 13500 g. The supernatant was collected and glucose determined using the Glucose assay kit (Sigma), and normalized to cellular protein content. Free Fatty Acid and Cholesterol Assay Cells were infected as above and when HCV infection levels reached ≥ 90% cells were harvested by trypsinization and washed with PBS. Cells were lysed in Triton X-100 in PBS and centrifuged for 10 minutes at 13500 g. The free fatty acid content of the supernatant was determined using the Free Fatty Acids, Half Micro Test Kit (Roche) and normalized to cellular protein content. Cholesterol content was determined using the Amplex Red cholesterol assay (Invitrogen).

ysed in Triton X-100 in PBS and centrifuged for 10 minutes at 13500 g. The free fatty acid content of the supernatant was determined using the Free Fatty Acids, Half Micro Test Kit (Roche) and normalized to cellular protein content. Cholesterol content was determined using the Amplex Red cholesterol assay (Invitrogen). Measurement of Reactive Oxygen Species Huh 7.5 cells were infected with HCV at moi of 0.02 for 1, 4, 24, and 48 hours, and 8 and 10 days. Mock-infected cells were prepared for the same time points. After harvesting, 50,000 cells were resuspended in 5 μM 2′,7′-dichlorofluorescein diacetate (Sigma), and incubated for 1 hour at 37°C in the dark. Fluorescence was measured on a NOVOstar plate reader, at excitation/emission filter wavelengths of 485 nm/530 nm. Results HCV-Induced Cellular Responses Analyzed by Microarray, RNA-Seq, and Proteomic Analyses HCV (Jc1; genotype 2a) infection of Huh 7.5 cells was monitored by immunofluorescence microscopy. When infection levels reached ≥ 90%, the cells were harvested for microarray, RNA-Seq, and proteomics analyses.

Measurement of Reactive Oxygen Species Huh 7.5 cells were infected with HCV at moi of 0.02 for 1, 4, 24, and 48 hours, and 8 and 10 days. Mock-infected cells were prepared for the same time points. After harvesting, 50,000 cells were resuspended in 5 μM 2′,7′-dichlorofluorescein diacetate (Sigma), and incubated for 1 hour at 37°C in the dark. Fluorescence was measured on a NOVOstar plate reader, at excitation/emission filter wavelengths of 485 nm/530 nm. Results HCV-Induced Cellular Responses Analyzed by Microarray, RNA-Seq, and Proteomic Analyses HCV (Jc1; genotype 2a) infection of Huh 7.5 cells was monitored by immunofluorescence microscopy. When infection levels reached ≥ 90%, the cells were harvested for microarray, RNA-Seq, and proteomics analyses. Microarray analysis identified 1351 genes, RNA-Seq identified 753 genes, and proteomics analysis identified 235 proteins which were differentially regulated in response to HCV infection. Although some overlap of genes/proteins identified by the three methods occurs, the majority of differential expression was, surprisingly, determined by a single method of analysis (Fig. 1A). IPA analysis of the genes detected by the microarray method identified 35 canonical pathways (Fig. 1B) and numerous biofunctions (Supporting Table 1). For the first time, HCV is shown to induce the differential expression of genes involved in PXR/RXR activation and LPS/IL-1 mediated inhibition of RXR function canonical pathways (Fig. 2). In addition gene networks identifying lipid and carbohydrate metabolism functions were identified (Fig. 3). By comparison; RNA-Seq gene expression analysis identified 78 canonical pathways (Fig. 1B), probably due to a lower redundancy rate in genes identified as compared to the microarray analysis. This method allowed identification of more than twice the number of canonical pathways determined by the microarray method and seven times more than by proteomics (Fig. 1B). In addition to supporting previous findings such as the impact of HCV on TGF-β signaling15 and hepatic fibrosis/hepatic stellate cell activation,16 previously unreported pathways were identified as being affected by HCV such as tight junction (TJ) signaling (Supporting Table 1). Of particular note in this pathway is the up-regulation of par-3 partitioning defective 3 homolog, which is thought to regulate TJ assembly.17, 18

rosis/hepatic stellate cell activation,16 previously unreported pathways were identified as being affected by HCV such as tight junction (TJ) signaling (Supporting Table 1). Of particular note in this pathway is the up-regulation of par-3 partitioning defective 3 homolog, which is thought to regulate TJ assembly.17, 18 Fig. 1 Comparative analysis of the number of genes and proteins found using RNA-Seq, microarray or proteomics analysis. (A) Numbers of genes and proteins identified by each method of analysis to be affected by HCV infection. Microarray analysis identified 1351 genes, RNA-Seq identified 753 genes and proteomics analysis identified 235 proteins (a full list can be seen in Supporting Fig. 1). (B) Analysis of the identified genes and proteins found in each study using IPA software determined the effect of HCV on a number of canonical pathways. The degree of crossover between the methods of analysis and canonical pathways identified using a cut off value of 1.5-fold change are identified in (B) and a selected number are shown in (C) (a full list can be seen in Supporting Fig. 1). † Indicates significance at the < 0.05 level. †† Indicates significance at the < 0.01% level. × Indicates not found. Fig. 2 Genes identified by microarray analysis to be involved in PXR/RXR activation (P < 0.005) and LPS/IL-1 mediated inhibition of RXR (P < 0.005). Genes involved in both pathways are highlighted in blue. Actual fold changes are shown in Supporting Fig. 1.

Fig. 1 Comparative analysis of the number of genes and proteins found using RNA-Seq, microarray or proteomics analysis. (A) Numbers of genes and proteins identified by each method of analysis to be affected by HCV infection. Microarray analysis identified 1351 genes, RNA-Seq identified 753 genes and proteomics analysis identified 235 proteins (a full list can be seen in Supporting Fig. 1). (B) Analysis of the identified genes and proteins found in each study using IPA software determined the effect of HCV on a number of canonical pathways. The degree of crossover between the methods of analysis and canonical pathways identified using a cut off value of 1.5-fold change are identified in (B) and a selected number are shown in (C) (a full list can be seen in Supporting Fig. 1). † Indicates significance at the < 0.05 level. †† Indicates significance at the < 0.01% level. × Indicates not found. Fig. 2 Genes identified by microarray analysis to be involved in PXR/RXR activation (P < 0.005) and LPS/IL-1 mediated inhibition of RXR (P < 0.005). Genes involved in both pathways are highlighted in blue. Actual fold changes are shown in Supporting Fig. 1. Fig. 3 Networks of connecting genes compiled by IPA from the microarray analysis. Genes highlighted in red are up-regulated, those in green are down-regulated. Expression levels of gray genes are not altered. Genes connected with blue filled lines have direct links; those connected with blue dotted lines have indirect links. Functional annotations for lipid and carbohydrate metabolism are indicated by filled gray lines.

red are up-regulated, those in green are down-regulated. Expression levels of gray genes are not altered. Genes connected with blue filled lines have direct links; those connected with blue dotted lines have indirect links. Functional annotations for lipid and carbohydrate metabolism are indicated by filled gray lines. Collating the genomic analyses can also provide further information; e.g., an additional five genes were added to the integrin-linked kinase (ILK) signaling canonical pathway (identified by RNA-Seq) by overlaying the microarray analyses (Fig. 4). Fig. 4 Impact of HCV infection on ILK signaling. A schematic of the ILK signaling canonical pathway produced using IPA and overlaid with genes identified by RNA-Seq and Microarray analysis. Genes highlighted in red are up-regulated, those in green are down-regulated. Gray genes are not detected to be altered. The 235 proteins identified by the 2DE-based proteomic study (Supporting Fig. 1) are in agreement with a previous study where changes in proteins involved in lipid metabolism, oxidative stress, and carbohydrate metabolism were observed.19 In addition we identified proteins involved in glutathione metabolism and numerous RNA binding proteins, including SSB, HNRPC, HNRPK, PCBP1, and PCBP2, of which SSB HNRPC and HNRPK are known to bind the untranslated regions of HCV RNA.20, 21 western blot analysis to confirm altered levels of SSB did not confirm its up-regulation by HCV. However, a cleavage fragment was detected in the HCV-infected sample only, illustrating an HCV induced change (Fig. 5).

1, and PCBP2, of which SSB HNRPC and HNRPK are known to bind the untranslated regions of HCV RNA.20, 21 western blot analysis to confirm altered levels of SSB did not confirm its up-regulation by HCV. However, a cleavage fragment was detected in the HCV-infected sample only, illustrating an HCV induced change (Fig. 5). Fig. 5 Proteomic analysis reveals the impact of HCV infection on SSB. Western blot analysis of equal amounts of protein extracted from HCV infected and mock-infected Huh 7.5 cells. Subsequent probing with a monoclonal antibody against SSB, detected the presence of a cleavage product of 35 kDa.

1, and PCBP2, of which SSB HNRPC and HNRPK are known to bind the untranslated regions of HCV RNA.20, 21 western blot analysis to confirm altered levels of SSB did not confirm its up-regulation by HCV. However, a cleavage fragment was detected in the HCV-infected sample only, illustrating an HCV induced change (Fig. 5). Fig. 5 Proteomic analysis reveals the impact of HCV infection on SSB. Western blot analysis of equal amounts of protein extracted from HCV infected and mock-infected Huh 7.5 cells. Subsequent probing with a monoclonal antibody against SSB, detected the presence of a cleavage product of 35 kDa. Impact of HCVcc Infection on Lipid and Cholesterol Metabolism Components of host lipid metabolism such as very low density lipoprotein and apolipoproteinB-100 are known to play key roles in HCV replication (for review, see Syed et al.22) and a more detailed investigation will aid our understanding of the mechanisms involved. Gene ontology analysis of the microarray and RNA-Seq data using IPA showed that biofunctions and canonical pathways associated with lipids and cholesterol are markedly affected by HCV infection (Supporting Table 1). We therefore investigated these pathways further using functional studies and biochemical assays. A significant (P < 0.01) increase in cholesterol and free fatty acid levels in HCV-infected Huh 7.5 cells is shown here for the first time, with cholesterol and free fatty acid levels more than a third higher in HCV-infected cells (Fig. 6). Increased levels of cholesterol are important for successful HCV replication as prophylactic use of the 3-beta-hydroxysterol Δ-7 reductase inhibitor AY9944, which prevents cholesterol synthesis,23 for 6 hours prior to infection, caused a reduction in HCV foci formation in a dose-dependent manner (Fig. 6). Statins have achieved similar effects in vitro.24, 25

rtant for successful HCV replication as prophylactic use of the 3-beta-hydroxysterol Δ-7 reductase inhibitor AY9944, which prevents cholesterol synthesis,23 for 6 hours prior to infection, caused a reduction in HCV foci formation in a dose-dependent manner (Fig. 6). Statins have achieved similar effects in vitro.24, 25 Fig. 6 Impact of HCV on cellular metabolism. At the same time point the gene and proteomics analyses were carried out (HCV infection levels ≥ 90%) the levels of free fatty acid, cholesterol and glucose in HCV infected Huh 7.5 cells were compared to mock-infected cells and significance measured using a Student t test (n = 4). (A) Free fatty acid levels in HCV infected cells were significantly higher than in control cells with average values of 320 μM in infected cells compared to 215 μM within uninfected cells. (B) Cholesterol levels are higher in infected cells with average levels of 1778 μg/mL in noninfected and 2774 μg/mL in HCV infected cells. (C) To demonstrate the requirement for cholesterol, Huh 7.5 cells were infected with HCV at a moi of 0.2, at 8 hours after pretreatment with 0, 0.6, 1.25, and 2.5 μM AY9944. At 3 days after infection, cells were analyzed by immunofluorescence and foci of infected cells counted. At increasing concentrations of AY9944 significant decreases in foci were exhibited compared to mock-infected control cells (student t test [n = 4]). (D) At day 10 after infection, glucose levels within HCV infected Huh7.5 showed a 9-fold drop (significance Student t test [n = 4]) compared to noninfected cells with levels of 4.6 μg/mL compared to 41.6 μg/mL of noninfected cells. This difference is ameliorated by the addition of interferon (which has no significant effect on mock-infected cells) in a dose-dependent manner. Infection of Huh 7.5 cells with X-31 influenza has no effect on glucose levels. *Indicates significance at the <0.05 level. **Indicates significance at the < 0.01% level. MI, mock-infected.

is ameliorated by the addition of interferon (which has no significant effect on mock-infected cells) in a dose-dependent manner. Infection of Huh 7.5 cells with X-31 influenza has no effect on glucose levels. *Indicates significance at the <0.05 level. **Indicates significance at the < 0.01% level. MI, mock-infected. In addition, we noted up-regulation of the pregnane X receptor (PXR)/retinoic acid receptor (RXR) ligand activation signaling pathway (identified by microarray and RNA-Seq data analysis, Fig. 2). This is of interest because PXR ligands may ameliorate human diseases such as cholestatic liver disease.26 Impact of HCVcc Infection on Cellular Glucose Metabolism Pathway analysis of the microarray data showed perturbation of the glycolysis and gluconeogenesis canonical pathways after HCV infection, with nine genes being differentially expressed. The SLC2A4RG gene (microarray), which induces the gene responsible for expression of the glucose transporter SLC2A4, is down-regulated 2.3-fold and another glucose transporter SLC2A8 (microarray, implicated in glucose homeostasis27) is up-regulated 1.9-fold. We therefore analyzed the glucose levels in HCV-infected Huh 7.5 cells, which were found to be nine-fold lower compared to noninfected cells (Fig. 6D). Glucose levels recovered in a dose-dependent manner, after interferon treatment for 3 days. In contrast, infection of Huh 7.5 cells with influenza virus (type A, strain X-31; infection level close to 100%), or treatment of uninfected cells with interferon, had no impact on glucose levels.

o noninfected cells (Fig. 6D). Glucose levels recovered in a dose-dependent manner, after interferon treatment for 3 days. In contrast, infection of Huh 7.5 cells with influenza virus (type A, strain X-31; infection level close to 100%), or treatment of uninfected cells with interferon, had no impact on glucose levels. Impact of HCVcc on Reactive Oxygen Species and Oxidative Stress HCV infection is characterized by increased levels of markers of oxidative stress, and lipid peroxidation products are increased in serum and liver specimens from patients.28 Analysis of HCV infection in the microarray experiment identified nine up-regulated metallothionein and five glutathione-related genes, the RNA-Seq experiment identified superoxide dismutase 3, and the proteomics analysis identified altered expression of superoxide dismutase, which are likely to be involved in alleviating oxidative stress, and counteracting production of lipid peroxidation and/or reactive oxygen species (ROS).29, 30 Measurement of ROS using 2′,7′-dichlorofluorescein diacetate showed a significant increase (P < 0.05) 48 hours after HCV infection, which increased further with time after infection; 10 days after infection (when gene and proteomics analyses were conducted) ROS levels were 56% (P < 0.01) higher in HCV-infected cells (Fig. 7).

ment of ROS using 2′,7′-dichlorofluorescein diacetate showed a significant increase (P < 0.05) 48 hours after HCV infection, which increased further with time after infection; 10 days after infection (when gene and proteomics analyses were conducted) ROS levels were 56% (P < 0.01) higher in HCV-infected cells (Fig. 7). Fig. 7 Measurement of ROS in the HCVcc system. Huh 7.5 cells were preinfected with HCV for 1, 4, 24, and 48 hours and 8 and 10 days. ROS were detected with 2′,7′-dichlorofluorescein diacetate. All regimes were compared to similarly treated mock-infected Huh 7.5 cells. *Indicates significance at the <0.05 level. **Indicates significance at the <0.01% level.

system. Huh 7.5 cells were preinfected with HCV for 1, 4, 24, and 48 hours and 8 and 10 days. ROS were detected with 2′,7′-dichlorofluorescein diacetate. All regimes were compared to similarly treated mock-infected Huh 7.5 cells. *Indicates significance at the <0.05 level. **Indicates significance at the <0.01% level. Discussion The RNA-Seq analysis, in combination with microarray and proteomics approaches, highlights hitherto unreported cellular responses to HCV infection at the gene level, adding to the recently published data by Walters et al.31 The RNA-Seq approach, based on new technological advances in massively parallel sequencing, has a number of advantages over conventional microarray techniques. The most important of these is that no prior selection of RNA is required, as is inherent in the use of probe/hybridization-based assays. Thus, an unbiased analysis of the total transcriptome can be performed, and the specific impact of HCVcc infection on Huh 7.5 cells in vitro could be analyzed. The limitations of the technique are the requirements for large quantities of high-quality RNA, the current cost and requirement for specialist sequencing equipment, and management of the huge data load. However, rapidly emerging data from this new field point to important potential benefits in terms of improved detection.6, 9, 32

ons of the technique are the requirements for large quantities of high-quality RNA, the current cost and requirement for specialist sequencing equipment, and management of the huge data load. However, rapidly emerging data from this new field point to important potential benefits in terms of improved detection.6, 9, 32 By collating the results from the different techniques it is possible to expand the information regarding the effect of HCV on gene and protein expression, increasing our chances of identifying potentially important genes and pathways. One such example is the observed increase of genes in the ILK signaling canonical pathway from 25 to 30. HCV induction of the ILK signaling canonical pathway (reported here for the first time) is likely to trigger actin rearrangement.33 Such cytoskeleton regulation is postulated to be important for HCV replication,34, 35 perhaps allowing viral movement to the tight junctions in a similar manner to Coxsackievirus.36, 37 The RNA-Seq analysis identified the tight junction signaling canonical pathway (Supporting Fig. 1), increasingly thought to be important in HCV entry and transmission.38, 39 A number of other canonical pathways affected by HCV infection including LPS/IL-1 mediated inhibition of RXR function, and PXR/RXR activation are reported here for the first time. HCV core protein has been shown to bind RXRα resulting in the up-regulation of certain lipid metabolism enzymes. As HCV replication is linked to increased lipid metabolism, this potentially increases viral replication.40 The “LPS/IL-1 mediated inhibition of RXR function” canonical pathway identified by IPA may also point toward a novel host antiviral response, as inhibition of RXR function could impair cholesterol and lipid metabolism41 required for viral replication.

ed lipid metabolism, this potentially increases viral replication.40 The “LPS/IL-1 mediated inhibition of RXR function” canonical pathway identified by IPA may also point toward a novel host antiviral response, as inhibition of RXR function could impair cholesterol and lipid metabolism41 required for viral replication. The combined analysis used in this study indicates that HCV infection is likely to cause disruption of glycolysis, gluconeogenesis and lipid metabolism. This is supported by the finding that interference of the cholesterol biosynthesis pathway causes a drop in HCV infectivity and replication. In addition, following HCV infection, cellular glucose levels are significantly decreased and both free fatty acid and cholesterol levels are significantly increased. An increase in cholesterol metabolism may lead to increased glycolysis and gluconeogenesis which could result in increased acetyl-CoA needed for cholesterol production, leading to decreased glucose levels and increased amino acid metabolism, perhaps for gluconeogenesis. Furthermore, free fatty acid metabolism is also increased. ACSL1, which converts free long-chain fatty acids into acetyl-CoA precursors, as well as CPT1A and CPT1B, which transport fatty acids across the outer mitochondrial membrane for subsequent production of acetyl-CoA, are up-regulated 2.5-fold, 2.3-fold, and 1.6-fold, respectively, which could support viral replication by increasing cholesterol production.

in fatty acids into acetyl-CoA precursors, as well as CPT1A and CPT1B, which transport fatty acids across the outer mitochondrial membrane for subsequent production of acetyl-CoA, are up-regulated 2.5-fold, 2.3-fold, and 1.6-fold, respectively, which could support viral replication by increasing cholesterol production. HCV suppresses cellular glucose uptake by down-regulation of surface expression of glucose transporters GLUT1 and GLUT2.42 Down-regulation of GLUT4 (SLC2A4) has been identified by Walters et al.31 and confirmed in this study, and we further identify down-regulation of GLUT3 (SLC2A3), indicating that it may be involved in a similar manner. Although infected cells may be attempting to counter the HCV-induced low cellular glucose levels by increasing gluconeogenesis, they appear to be unable to counteract both increased glycolysis as well as decreased glucose transport.

n-regulation of GLUT3 (SLC2A3), indicating that it may be involved in a similar manner. Although infected cells may be attempting to counter the HCV-induced low cellular glucose levels by increasing gluconeogenesis, they appear to be unable to counteract both increased glycolysis as well as decreased glucose transport. This is the first time ROS levels have been quantified in the infectious HCVcc system. The induction of ROS depends on the time after infection and on infection levels. ROS may contribute to the development of hepatocellular carcinoma due to triggering double-stranded DNA breaks43 and could also lead to fibrosis via induction of TGF-β.44 The superoxide dismutase protein identified by proteomics, and superoxide dismutase 3, whose up-regulated gene was identified by RNA-Seq analysis, as well as the metallothione and glutathione genes identified in the microarray analyses, could potentially protect against HCV-induced ROS and oxidative stress as well as suppress HCV RNA replication in a manner similar to antioxidants.45, 46

ide dismutase 3, whose up-regulated gene was identified by RNA-Seq analysis, as well as the metallothione and glutathione genes identified in the microarray analyses, could potentially protect against HCV-induced ROS and oxidative stress as well as suppress HCV RNA replication in a manner similar to antioxidants.45, 46 HCV induces differential expression of several RNA binding proteins including SSB (Supporting Table 1). SSB interacts with the HCV IRES in the 5′ untranslated region of the viral RNA, an area important for the translation of the HCV genome.47, 48 Western blot analysis probing for SSB indicates that HCV induces cleavage of SSB (Supporting Table 1). Using an HCV replicon system, Romero et al. demonstrate that granzyme H (from cytotoxic-lymphocyte granules) exerts an antiviral effect by cleaving SSB, thereby preventing IRES translation of the HCV genome.49 We do not know which factor is responsible for cleaving SSB in our study (HCVcc in the absence of cytotoxic lymphocyte granules), but it may hint at a novel host mediated antiviral response.

totoxic-lymphocyte granules) exerts an antiviral effect by cleaving SSB, thereby preventing IRES translation of the HCV genome.49 We do not know which factor is responsible for cleaving SSB in our study (HCVcc in the absence of cytotoxic lymphocyte granules), but it may hint at a novel host mediated antiviral response. We have only highlighted a key subset of the differentially regulated genes and proteins. Further analysis of the data sets will contribute to our understanding of virus-host interplay in HCV, and use of the RNA-Seq technology in combination with microarray and proteomics techniques could have a major impact in other infectious diseases. It will also be of great interest to compare and confirm our findings with other HCV infectious systems and genotypes as they become available. This study was performed using the most robust in vitro system available at the time comprising an HCV genotype 2 infected human hepatoma cell line, which has been widely used in similar studies.31, 42, 50 However, it will be important to define the significance of the transcriptional changes, especially in relation to different host and viral genotypes in similar studies and in vivo. We thank Professor R. A. Dwek for his kind support and advice during the project, David Chittenden for preparation of the 2D gels for the proteomic analysis and Bettina Kampa for digestion of the proteomic samples. Abbreviations 2DEtwo-dimensional gel electrophoresis AY9944trans-1,4-bis(2-chlorobenzylaminomethyl)cyclohexane cDNAcomplementary DNA HCVcchepatitis C virus cell culture system PXRpregnane X receptor

We thank Professor R. A. Dwek for his kind support and advice during the project, David Chittenden for preparation of the 2D gels for the proteomic analysis and Bettina Kampa for digestion of the proteomic samples. Abbreviations 2DEtwo-dimensional gel electrophoresis AY9944trans-1,4-bis(2-chlorobenzylaminomethyl)cyclohexane cDNAcomplementary DNA HCVcchepatitis C virus cell culture system PXRpregnane X receptor RNA-SeqRNA sequencing ROSreactive oxygen species RXRretinoic acid receptor SSBSjogren syndrome antigen B Supplementaary material

Hereditary hemochromatosis (HH) is one of the most common genetic diseases, and most patients are homozygous for the HH gene HFE Cys282Tyr (C282Y) mutation. Homozygosity for C282Y may be associated with excessive absorption of dietary iron in some cases. If untreated, progressive iron accumulation in key organs leads to toxicity, resulting in tissue damage and organ dysfunction.1–3 Patients with HH and cirrhosis are also at significant risk of developing hepatocellular carcinoma.4 Iron overload can be readily managed in most patients by therapeutic phlebotomy. However, patients with underlying anemias, severe heart disease, or reduced venous access may not tolerate this form of treatment.5 In addition, compliance may also be variable over time6 as a result of the inconvenience of frequent clinic visits and the discomfort associated with the procedure.

therapeutic phlebotomy. However, patients with underlying anemias, severe heart disease, or reduced venous access may not tolerate this form of treatment.5 In addition, compliance may also be variable over time6 as a result of the inconvenience of frequent clinic visits and the discomfort associated with the procedure. The once-daily oral iron chelator deferasirox (Exjade; Novartis Pharma AG, Basel, Switzerland) is indicated for the treatment of chronic iron overload due to frequent blood transfusions at approved doses of 10-40 mg/kg body weight/day. Clinical studies have demonstrated the efficacy, safety, and tolerability of deferasirox in patients with a variety of conditions associated with transfusional iron overload.7–11 Studies reporting iron chelation therapy in patients with primary (nontransfusional) iron overload are limited, but data are encouraging.12–16 As patients with HH generally have a lower relative iron burden than those with transfusional iron overload and do not accumulate iron at the same rate, the safety and efficacy of iron chelation therapy in this setting requires further evaluation. The current study is the first clinical trial to assess the safety and efficacy of deferasirox therapy (at the lower end of the approved dose range for patients with transfusional iron overload) in patients with HH who required de-ironing based on elevated serum ferritin levels.

his setting requires further evaluation. The current study is the first clinical trial to assess the safety and efficacy of deferasirox therapy (at the lower end of the approved dose range for patients with transfusional iron overload) in patients with HH who required de-ironing based on elevated serum ferritin levels. Patients and Methods Patient Recruitment The study was conducted across 18 sites in six countries (United States, France, Germany, Italy, Australia, and Canada). This was an open-label, multicenter, dose-escalation study designed to characterize the safety and efficacy of deferasirox using four dose levels (5, 10, 15, and 20 mg/kg/day). Male and female patients (aged ≥18 years) with homozygosity for the C282Y HFE mutation (as documented by molecular diagnostic testing) and iron overload (as shown by serum ferritin levels of 300-2000 ng/mL and serum transferrin saturation ≥45%) were included in the study. Key exclusion criteria were: males with hemoglobin concentrations <13 g/dL or females with hemoglobin concentrations <12 g/dL; treatment with phlebotomy within 2 weeks of screening; history of blood transfusion during the 6 months prior to study entry; current or previous treatment with deferiprone (Ferriprox; Apotex, Inc., Toronto, ON, Canada) or deferasirox; serum creatinine levels above the upper limit of the normal range (ULN); alanine aminotransferase (ALT) levels of ≥2× ULN; or known diagnosis of liver cirrhosis. All patients provided written, informed consent at the start of the core and extension periods of the study. The study was conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki.

range (ULN); alanine aminotransferase (ALT) levels of ≥2× ULN; or known diagnosis of liver cirrhosis. All patients provided written, informed consent at the start of the core and extension periods of the study. The study was conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki. Study Design After all eligibility criteria had been confirmed, at least eight patients were planned to be enrolled in each of the four dose levels in a sequential manner. Opening of the next dose level (dose escalation) followed a safety review by an independent safety monitoring committee; this was conducted once the sixth patient enrolled in that stratum had been treated for 4 weeks. Dose escalation was guided by a Bayesian logistic regression model17,18 for the most common severe adverse events (AEs; abdominal pain, nausea, vomiting, diarrhea, rash, increased serum creatinine, and increased aminotransferases) and clinical review. The recommended dose level at completion of the core study was based on an overall assessment of safety and efficacy, taking into account all tolerability and serum ferritin data collected at each dose level tested. At the end of the core period of 24 weeks, patients were invited to enter an extension study for a further 24 weeks of therapy with deferasirox. Patients who chose not to continue the study did not have to provide a reason.

aking into account all tolerability and serum ferritin data collected at each dose level tested. At the end of the core period of 24 weeks, patients were invited to enter an extension study for a further 24 weeks of therapy with deferasirox. Patients who chose not to continue the study did not have to provide a reason. Assessments The primary objective of the core study was to assess the safety of deferasirox therapy, with continued assessment in the 24-week extension phase. Safety assessments consisted of monitoring and recording all AEs (including serious AEs), the regular monitoring of biochemistry, hematology, urinalysis, electrocardiogram, audiometry, and ocular examination data. The effect of deferasirox therapy on serum ferritin levels was assessed as a secondary objective. If serum ferritin levels fell to <100 ng/mL at any visit, deferasirox was interrupted until levels rose to >300 ng/mL. Serum ferritin, serum iron, serum transferrin, calculated total iron-binding capacity, and transferrin saturation parameters were measured at screening and at each study visit.

s a secondary objective. If serum ferritin levels fell to <100 ng/mL at any visit, deferasirox was interrupted until levels rose to >300 ng/mL. Serum ferritin, serum iron, serum transferrin, calculated total iron-binding capacity, and transferrin saturation parameters were measured at screening and at each study visit. Statistical Methods All patients who received at least one dose of study drug, and had at least one post-baseline safety assessment, were included in the safety population. The per-protocol population comprised patients from the safety population who had no major protocol violations and was used for selected efficacy analyses. Median decrease in serum ferritin levels in the core and extension studies were analyzed based on the signed-rank test for the three respective dose cohorts. Changes in serum ferritin levels among the three dose cohorts from baseline to end of study were evaluated by performing an analysis of covariance, where the baseline serum ferritin measure was fitted as a continuous covariate. A longitudinal analysis was also performed by a linear mixed-effects model for serum ferritin. The statistical model for the dose escalation of deferasirox was based on a two-parameter logistic model.

y performing an analysis of covariance, where the baseline serum ferritin measure was fitted as a continuous covariate. A longitudinal analysis was also performed by a linear mixed-effects model for serum ferritin. The statistical model for the dose escalation of deferasirox was based on a two-parameter logistic model. Results Patient Characteristics A total of 49 Caucasian patients with HH were enrolled in the core study (Table 1) and received starting doses of deferasirox of either 5 (n = 11), 10 (n = 15), or 15 (n = 23) mg/kg/day for at least 24 weeks. The study was initiated on August 23, 2006, and completed on October 2, 2008, and March 19, 2009, for the core and extension study, respectively. Table 1 Patient Demographics and Characteristics at Baseline Characteristic 5 mg/kg/day (n = 11) 10 mg/kg/day (n = 15) 15 mg/kg/day (n = 23) All Patients (n = 49) Mean age ± SD, years 55.8 ± 12.8 47.8 ± 10.3 49.8 ± 16.4 50.6 ± 14.0 Male:female 9:2 11:4 13:10 33:16 Mean time since diagnosis ± SD, years 6.6 ± 7.0 1.3 ± 2.0 2.6 ± 3.8 3.1 ± 4.7 Median serum ferritin (range), ng/mL* 512 (376–1729) 859 (447–1792) 634 (357–1600) 645 (357–1792) Mean transferrin saturation ± SD, %* 78.1 ± 17.4 81.0 ± 12.7 74.1 ± 15.4 77.2 ± 15.2 * Based on the per-protocol population.

:2 11:4 13:10 33:16 Mean time since diagnosis ± SD, years 6.6 ± 7.0 1.3 ± 2.0 2.6 ± 3.8 3.1 ± 4.7 Median serum ferritin (range), ng/mL* 512 (376–1729) 859 (447–1792) 634 (357–1600) 645 (357–1792) Mean transferrin saturation ± SD, %* 78.1 ± 17.4 81.0 ± 12.7 74.1 ± 15.4 77.2 ± 15.2 * Based on the per-protocol population. After 49 patients had been enrolled in the study and 23 patients had been treated at the 15 mg/kg/day dose level, further dose escalation was stopped. The estimated mean rate of the most common severe AEs at 20 mg/kg/day was determined to be 34.9% compared with 17.4% at the 15 mg/kg/day dose level, suggesting that dose escalation should not continue beyond 15 mg/kg/day. All 49 patients were included in the safety analysis and 48 patients were included in the efficacy analysis (per-protocol; one patient was excluded due to clinical evidence of active hepatitis B/C). Thirty-seven (75.5%) patients completed the core study (10 [90.9%], 11 [73.3%], and 16 [69.6%] patients in the 5, 10, and 15 mg/kg/day cohorts, respectively). Twenty-six patients decided to continue into the extension study; 23 (88.5%) patients (nine [100%], six [100%], eight [72.7%] patients in the 5, 10, and 15 mg/kg/day cohorts, respectively) completed the 24-week extension.

90.9%], 11 [73.3%], and 16 [69.6%] patients in the 5, 10, and 15 mg/kg/day cohorts, respectively). Twenty-six patients decided to continue into the extension study; 23 (88.5%) patients (nine [100%], six [100%], eight [72.7%] patients in the 5, 10, and 15 mg/kg/day cohorts, respectively) completed the 24-week extension. Patient Discontinuations During the core and extension study, 15 patients discontinued treatment. Reasons for discontinuation in the core study were: protocol deviation (n = 1) in the 5 mg/kg/day cohort; AEs (n = 3) and consent withdrawal (n = 1) in the 10 mg/kg/day cohort; and AEs (n = 4), consent withdrawal, abnormal laboratory value and loss to follow-up (n = 1 for each) in the 15 mg/kg/day cohort (Fig. 1). In the extension study, three patients discontinued in the 15 mg/kg/day cohort primarily due to AEs (n = 2; Fig. 1). AEs leading to discontinuation were ALT elevation, increased bilirubin levels (attributed to preexisting Gilbert's syndrome; peak total bilirubin of 104.3 μmol/L) and diarrhea/nausea/backache/fatigue in the core 10 mg/kg/day cohort (n = 1 for each); severe rash, serum creatinine elevation, nausea, and ALT elevation in the core 15 mg/kg/day cohort (n = 1 for each); and diarrhea and increased levels of serum aminotransferases in the extension 15 mg/kg/day cohort (n = 1 for both).

and diarrhea/nausea/backache/fatigue in the core 10 mg/kg/day cohort (n = 1 for each); severe rash, serum creatinine elevation, nausea, and ALT elevation in the core 15 mg/kg/day cohort (n = 1 for each); and diarrhea and increased levels of serum aminotransferases in the extension 15 mg/kg/day cohort (n = 1 for both). Figure 1 Patient disposition in the core and extension study. *The 5 mg/kg/day dose reduced serum ferritin in only three patients; the other six patients in this dose cohort received 10 mg/kg/day at the start of the extension (all but one patient experienced serum ferritin levels <100 ng/mL when receiving 10 mg/kg/day). Safety Assessment The most common investigator-reported AEs after the start of deferasirox treatment (≥10% in all patients) during the core study were diarrhea, headache, nausea, and abdominal pain (Table 2). Common AEs suspected to be drug-related (≥10% in all patients) were diarrhea, nausea, and abdominal pain (Table 3). There were no serious AEs or deaths reported by the investigators. In the extension study, the incidence of newly reported AEs and drug-related AEs was reduced compared with the core study (Tables 2 and 3). Table 2 Most Common AEs (≥10% of All Patients in Core or Extension) n (%) 5 mg/kg/day 10 mg/kg/day 15 mg/kg/day All Patients

Safety Assessment The most common investigator-reported AEs after the start of deferasirox treatment (≥10% in all patients) during the core study were diarrhea, headache, nausea, and abdominal pain (Table 2). Common AEs suspected to be drug-related (≥10% in all patients) were diarrhea, nausea, and abdominal pain (Table 3). There were no serious AEs or deaths reported by the investigators. In the extension study, the incidence of newly reported AEs and drug-related AEs was reduced compared with the core study (Tables 2 and 3). Table 2 Most Common AEs (≥10% of All Patients in Core or Extension) n (%) 5 mg/kg/day 10 mg/kg/day 15 mg/kg/day All Patients Adverse Event Core (n = 11) Ext (n = 9) Core (n = 15) Ext (n = 6) Core (n = 23) Ext (n = 11) Core (n = 49) Ext (n = 26) Diarrhea 2 (18.2) 0 6 (40.0) 0 10 (43.5) 1 (9.1) 18 (36.7) 1 (3.8) Headache 1 (9.1) 0 2 (13.3) 0 7 (30.4) 1 (9.1) 10 (20.4) 1 (3.8) Nausea 0 0 2 (13.3) 0 6 (26.1) 0 8 (16.3) 0 Abdominal pain 0 0 3 (20.0) 0 4 (17.4) 0 7 (14.3) 0 Increased serum creatinine* 0 1 (11.1) 3 (20.0) 1 (16.7) 4 (17.4) 1 (9.1) 7 (14.3) 3 (11.5) Back pain 1 (9.1) 1 (11.1) 4 (26.7) 0 2 (8.7) 0 7 (14.3) 1 (3.8) Increased ALT* 0 1 (11.1) 3 (20.0) 0 3 (13.0) 0 6 (12.2) 1 (3.8) Rash 1 (9.1) 0 1 (6.7) 1 (16.7) 4 (17.4) 0 6 (12.2) 1 (3.8) Flatulence 2 (18.2) 0 1 (6.7) 0 2 (8.7) 0 5 (10.2) 0 Nasopharyngitis 1 (9.1) 0 1 (6.7) 0 3 (13.0) 0 5 (10.2) 0 Fatigue 1 (9.1) 1 (11.1) 1 (6.7) 1 (16.7) 2 (8.7) 1 (9.1) 4 (8.2) 3 (11.5) Arthralgia 1 (9.1) 0 2 (13.3) 2 (33.3) 1 (4.3) 1 (9.1) 4 (8.2) 3 (11.5) * Reported based on clinical signs or symptoms, considered clinically significant by investigator or requiring therapy as a result.

1 (9.1) 0 1 (6.7) 0 3 (13.0) 0 5 (10.2) 0 Fatigue 1 (9.1) 1 (11.1) 1 (6.7) 1 (16.7) 2 (8.7) 1 (9.1) 4 (8.2) 3 (11.5) Arthralgia 1 (9.1) 0 2 (13.3) 2 (33.3) 1 (4.3) 1 (9.1) 4 (8.2) 3 (11.5) * Reported based on clinical signs or symptoms, considered clinically significant by investigator or requiring therapy as a result. Table 3 Most Common Drug-Related AEs (≥10% of All Patients in Core or Extension) n (%) 5 mg/kg/day 10 mg/kg/day 15 mg/kg/day All Patients Adverse Event Core (n = 11) Ext (n = 9) Core (n = 15) Ext (n = 6) Core (n = 23) Ext (n = 11) Core (n = 49) Ext (n = 26) Diarrhea 1 (9.1) 0 4 (26.7) 0 9 (39.1) 1 (9.1) 14 (28.6) 1 (3.8) Increased serum creatinine* 0 1 (11.1) 3 (20.0) 1 (16.7) 4 (17.4) 1 (9.1) 7 (14.3) 3 (11.5) Nausea 0 0 2 (13.3) 0 4 (17.4) 0 6 (12.2) 0 Abdominal pain 0 0 2 (13.3) 0 3 (13.0) 0 5 (10.2) 0 Increased ALT* 0 1 (11.1) 3 (20.0) 0 2 (8.7) 0 5 (10.2) 1 (3.8) * Reported based on clinical signs or symptoms, considered clinically significant by investigator or required therapy as a result. Over the entire study (48 weeks), more patients in the 15 mg/kg/day cohort experienced increases in ALT or aspartate aminotransferase (AST) levels >3× baseline and greater than ULN than the lower dose cohorts (Table 4). All eight patients who had ALT >3× baseline and greater than ULN had their dose interrupted; four discontinued the study, three did not restart, and one was observed until the end of the core study but did not enter the extension. Of these patients with ALT events, only one patient had an increase in bilirubin greater than ULN.

patients who had ALT >3× baseline and greater than ULN had their dose interrupted; four discontinued the study, three did not restart, and one was observed until the end of the core study but did not enter the extension. Of these patients with ALT events, only one patient had an increase in bilirubin greater than ULN. Table 4 Serum Creatinine and Liver Function Tests Core and Extension* 5 mg/kg/day (n = 11) 10 mg/kg/day (n = 15) 15 mg/kg/day (n = 23) All Patients (n = 49) Serum creatinine, increase >33% above baseline and >ULN at 2 consecutive visits, n (%) 0 5 (33.3) 8 (34.8) 13 (26.5) AST Increase >3× baseline and >ULN, n (%) 1 (9.1) 2 (13.3) 5 (21.7) 8 (16.3) Increase >5× baseline and >ULN, n (%) 0 0 5 (21.7) 5 (2.0) ALT Increase >3× baseline and >ULN, n (%) 1 (9.1) 1 (6.7) 6 (26.1) 8 (16.3) Increase >5× baseline and >ULN, n (%) 0 0 5 (21.7) 5 (2.0) * Patients were counted only once in the core or extension.

se >3× baseline and >ULN, n (%) 1 (9.1) 2 (13.3) 5 (21.7) 8 (16.3) Increase >5× baseline and >ULN, n (%) 0 0 5 (21.7) 5 (2.0) ALT Increase >3× baseline and >ULN, n (%) 1 (9.1) 1 (6.7) 6 (26.1) 8 (16.3) Increase >5× baseline and >ULN, n (%) 0 0 5 (21.7) 5 (2.0) * Patients were counted only once in the core or extension. In the core study, two patients had ALT levels >5× ULN (both in the 15 mg/kg/day cohort; one patient had elevated levels at baseline), and 11 patients had serum creatinine levels >33% above baseline and greater than ULN on two consecutive occasions (five patients in the 10 mg/kg/day cohort and six patients in the 15 mg/kg/day cohort). In the extension study, one patient experienced an increase in ALT levels >5× ULN (15 mg/kg/day cohort; levels were elevated at baseline), and seven patients had serum creatinine levels >33% above baseline and greater than ULN on two consecutive occasions (two patients in the 10 mg/kg/day cohort and five patients in the 15 mg/kg/day cohort). Patients were managed with dose decreases and/or interruptions.

g/kg/day cohort; levels were elevated at baseline), and seven patients had serum creatinine levels >33% above baseline and greater than ULN on two consecutive occasions (two patients in the 10 mg/kg/day cohort and five patients in the 15 mg/kg/day cohort). Patients were managed with dose decreases and/or interruptions. Serum Ferritin and Transferrin Saturation Of the patients completing the core and extension study, 26 of 37 (70.3%) and 21 of 23 (91.3%) achieved serum ferritin levels of <500 ng/mL. Of the 23 patients completing 48 weeks of deferasirox treatment, 15 (65.2%) had serum ferritin levels of <250 ng/mL by the end of the extension study. Four (36.4%), two (13.3%), and eight (36.4%) patients (per-protocol) had serum ferritin levels of <100 ng/mL in the 5, 10, and 15 mg/kg/day cohorts, respectively, at some point during the 48-week treatment period. Reductions in serum ferritin levels were observed in the core study across all three dose cohorts with a median decrease of 31.1% (P = 0.08 [nonsignificant]), 52.8% (P = 0.0005), and 55.4% (P < 0.0001) in the 5, 10, and 15 mg/kg/day cohorts, respectively. The time course of the serum ferritin decrease was dose-dependent (Fig. 2A). In the patients who continued deferasirox therapy in the extension study, serum ferritin levels continued to decrease with a median reduction of 63.5% (P = 0.002), 74.8% (P < 0.0001), and 74.1% (P = 0.0004) from the core baseline in the 5, 10, and 15 mg/kg/day cohorts, respectively (Fig. 2B). However, it should be noted that six of the nine patients in the 5 mg/kg/day cohort received an increased dose of 10 mg/kg/day during the extension. Changes in serum ferritin levels at the end of the core study were not statistically different at a 0.05 level between the three dose cohorts. Longitudinal analysis over the core and extension study showed that the reduction in serum ferritin from baseline levels was significantly greater in the 10 and 15 mg/kg/day cohorts compared with the 5 mg/kg/day cohort (P < 0.05 for both). There was no significant difference in serum ferritin level decrease between the 10 and 15 mg/kg/day cohorts. Reductions in transferrin saturation were also observed in the core study across all three dose cohorts, with a mean decrease of 6.4%, 10.7%, and 2.1% in the 5, 10, and 15 mg/kg/day cohorts, respectively.

oth). There was no significant difference in serum ferritin level decrease between the 10 and 15 mg/kg/day cohorts. Reductions in transferrin saturation were also observed in the core study across all three dose cohorts, with a mean decrease of 6.4%, 10.7%, and 2.1% in the 5, 10, and 15 mg/kg/day cohorts, respectively. In the patients who continued deferasirox therapy in the extension study, transferrin saturation continued to decrease in the 10 and 15 mg/kg/day cohorts, resulting in a mean decrease from baseline to the end of the extension by 14.4% and 12.1%, respectively. However, the overall mean change from baseline in the 5 mg/kg/day cohort was +0.9%. Figure 2 Median serum ferritin in (A) patients enrolled in the core study and (B) patients who completed the core and continued into the extension study. BL, baseline. Note that the 5 mg/kg/day dose reduced serum ferritin in three patients; the other six patients in this dose cohort received 10 mg/kg/day at the start of the extension (all but one patient experienced serum ferritin levels <100 ng/mL when receiving 10 mg/kg/day). Relationship Between Serum Ferritin and Hepatic/Renal Parameters Subsequent analyses were performed to evaluate the relationship between baseline serum ferritin levels and changes in ALT and serum creatinine levels. These analyses suggested that ALT and serum creatinine increases may have been associated with relative deferasirox doses that were too high compared with baseline serum ferritin levels (Fig. 3).

erformed to evaluate the relationship between baseline serum ferritin levels and changes in ALT and serum creatinine levels. These analyses suggested that ALT and serum creatinine increases may have been associated with relative deferasirox doses that were too high compared with baseline serum ferritin levels (Fig. 3). Figure 3 Scatter plots of baseline serum ferritin versus worst ALT relative change (%) during (A) 24 weeks of treatment and (B) 48 weeks of treatment; and worst serum creatinine relative change (%) during (C) 24 weeks of treatment and (D) 48 weeks of treatment, by dose cohort and ULN. Five of the six patients in the 15 mg/kg/day cohort who had ALT levels >3× baseline and greater than ULN had relatively low serum ferritin levels at baseline (range 356-740 ng/mL). Two of these patients had sharp rises in ALT concentration associated with rapid reductions in serum ferritin to 70 and 108 ng/mL at week 6 and 36, respectively. The sixth patient had a baseline serum ferritin value of 1083 ng/mL, but experienced a rapid reduction at week 4 to 758 ng/mL. The patient in the 5 mg/kg/day cohort who had an ALT elevation of >3× baseline and greater than ULN had a baseline serum ferritin value of 740 ng/mL, and at the time of reaching the ALT elevation, had experienced a reduction in serum ferritin levels to 83 ng/mL following an increase in deferasirox dose (Fig. 4A). The patient in the 10 mg/kg/day cohort with increased ALT levels >3× baseline and greater than ULN did not show any distinct relationships between serum ferritin and ALT increases.

ng the ALT elevation, had experienced a reduction in serum ferritin levels to 83 ng/mL following an increase in deferasirox dose (Fig. 4A). The patient in the 10 mg/kg/day cohort with increased ALT levels >3× baseline and greater than ULN did not show any distinct relationships between serum ferritin and ALT increases. Figure 4 Individual patient examples of the relationship between serum ferritin and (A) ALT and (B) serum creatinine.

ng the ALT elevation, had experienced a reduction in serum ferritin levels to 83 ng/mL following an increase in deferasirox dose (Fig. 4A). The patient in the 10 mg/kg/day cohort with increased ALT levels >3× baseline and greater than ULN did not show any distinct relationships between serum ferritin and ALT increases. Figure 4 Individual patient examples of the relationship between serum ferritin and (A) ALT and (B) serum creatinine. Four of the five patients in the 10 mg/kg/day cohort who had serum creatinine levels >33% above baseline and greater than ULN at two consecutive visits had relatively low baseline serum ferritin levels (range 587-743 ng/mL); the remaining patient had a baseline serum ferritin value of 1316 ng/mL, but experienced a rapid reduction to 174 ng/mL by week 14. Baseline serum ferritin levels were similarly low in six of the eight patients in the 15 mg/kg/day cohort (range 490-896 ng/mL); the other two patients had rapid reductions in serum ferritin values from 1018 to 494 ng/mL at week 4 and from 1186 to 523 ng/mL at week 24, respectively. Overall in both dose cohorts, initial notable increases in serum creatinine levels corresponding with reductions in serum ferritin were observed early (within 4-6 weeks) in seven patients. The remaining six patients had serum creatinine increases by the end of the 24-week core study. One patient in the 15 mg/kg/day cohort had notable serum creatinine increases on several occasions in association with reductions in serum ferritin. On each of these occasions, the serum creatinine levels returned to baseline levels after deferasirox was temporarily discontinued (Fig. 4B).

nd of the 24-week core study. One patient in the 15 mg/kg/day cohort had notable serum creatinine increases on several occasions in association with reductions in serum ferritin. On each of these occasions, the serum creatinine levels returned to baseline levels after deferasirox was temporarily discontinued (Fig. 4B). Discussion This is the first clinical trial to demonstrate the safety and efficacy of deferasirox in a population of patients with C282Y homozygous HH and primary iron overload. There was higher enrollment of males than females as expected, because female C282Y homozygotes tend to have lower levels of iron overload than male homozygotes.19 Although phlebotomy is the first line of treatment for patients with HH, these data indicate that chelation therapy with deferasirox may offer an alternative approach to the management of iron accumulation for patients who are unwilling or unable to comply with de-ironing by phlebotomy. The results from the 24-week core study demonstrated that the AE profile was consistent with the known safety profile of deferasirox; all described gastrointestinal, renal, and hepatic AEs are known, labeled, and are clinically manageable with regular patient monitoring as reported in patients with transfusion-induced iron overload. The frequency of AEs was dose-dependent, being highest in the 15 mg/kg/day patient cohort; the estimated incidence of AEs at 20 mg/kg/day precluded dose escalation beyond 15 mg/kg/day. AEs generally occurred early during treatment as shown by the higher frequency reported during the core compared with the extension study. However, these data could be limited by the potential for self-selection of patients; for example, it is possible that patients experiencing AEs in the core phase of the study decided not to continue in the extension, reflecting the reduced frequency of AEs reported.

d during the core compared with the extension study. However, these data could be limited by the potential for self-selection of patients; for example, it is possible that patients experiencing AEs in the core phase of the study decided not to continue in the extension, reflecting the reduced frequency of AEs reported. Deferasirox doses of 5, 10, and 15 mg/kg/day reduced serum ferritin levels in this population of patients with HH, although it should be noted that six of the nine patients in the 5 mg/kg/day cohort who entered the extension study were dose escalated to 10 mg/kg/day. The percentage change in serum ferritin level was greater in the 10 and 15 mg/kg/day cohorts compared with the 5 mg/kg/day cohort and similar in the 10 and 15 mg/kg/day cohorts. When considered together with the safety results, where a higher frequency of AEs were reported in the 15 mg/kg/day cohort, a starting dose of 10 mg/kg/day appears to be the most appropriate in this population of patients. This is a starting dose lower than that generally recommended for patients with transfusional iron overload.7,10,20

gether with the safety results, where a higher frequency of AEs were reported in the 15 mg/kg/day cohort, a starting dose of 10 mg/kg/day appears to be the most appropriate in this population of patients. This is a starting dose lower than that generally recommended for patients with transfusional iron overload.7,10,20 Subsequent analyses of these data suggested that notable changes in renal and liver laboratory safety data were reflective of changes in serum ferritin levels over the course of the study, particularly in the 10 and 15 mg/kg/day cohorts. Several patients with low baseline serum ferritin levels, or those who experienced rapid reductions in serum ferritin, had notable increases in ALT or serum creatinine levels that required dose modifications and/or interruptions; there was no clear relationship between the occurrence of these laboratory abnormalities and time on study. A risk of AEs, including renal disorder following over-rapid chelation, has also been identified in transfusion-dependent patients and particular attention is advised for monitoring of serum creatinine levels in patients who are receiving high doses of deferasirox with low rates of transfusion.20,21 These data suggest that appropriate target serum ferritin levels should be predefined and rapid iron depletion is to be avoided in order to prevent the occurrence of AEs, particularly because many patients may be asymptomatic. This can be achieved by closely monitoring body iron levels and adjusting deferasirox doses accordingly. Serial measurement of serum ferritin levels is a convenient and inexpensive tool that has been widely used for monitoring body iron levels and the efficacy of chelation therapy in various transfusion-dependent anemias.10,20 Although no conclusive mechanistic explanation for the increases in serum creatinine observed in this study is available yet, it is likely related to modified renal hemodynamics as a result of deferasirox therapy.22,23 Although not an integral part of the protocol in the present study, serum creatinine levels in a limited number of patients who were monitored after treatment returned to normal. In addition, data from transfusion-dependent patient populations have shown that renal and hepatic events are generally nonprogressive and return to within normal limits following drug discontinuation.20,24,25

serum creatinine levels in a limited number of patients who were monitored after treatment returned to normal. In addition, data from transfusion-dependent patient populations have shown that renal and hepatic events are generally nonprogressive and return to within normal limits following drug discontinuation.20,24,25 The incidence of AEs, particularly in the core study, was slightly higher than that observed in patients with transfusional iron overload,20 although the lower number of patients included in this phase 1/2 trial should be considered. Our data suggest this difference may be related to the lower levels of iron burden in patients with HH, as well as lower rates of iron accumulation in the absence of ongoing red blood cell transfusions. This is supported by the higher rate of serum creatinine and ALT level increases in patients with lower serum ferritin levels or those with rapid reductions in serum ferritin following chelation therapy. Furthermore, as the patients in this study were generally asymptomatic, reported AEs may be comparatively higher due to the lack of medical conditions at baseline in the majority of patients, as opposed to other patient populations studied with additional underlying medical problems. The incidence of AEs during therapy with deferasirox may have implications for patient compliance and ultimately, the acceptance of iron chelation therapy for the treatment of patients with HH. In addition, the ease, effectiveness, and relatively low cost of phlebotomy therapy make it likely that phlebotomy will remain the treatment of choice for the vast majority of patients with HH.

cations for patient compliance and ultimately, the acceptance of iron chelation therapy for the treatment of patients with HH. In addition, the ease, effectiveness, and relatively low cost of phlebotomy therapy make it likely that phlebotomy will remain the treatment of choice for the vast majority of patients with HH. Although phlebotomy and iron chelation therapy are effective for reducing iron burden in patients with HH, natural history studies of untreated C282Y homozygotes have demonstrated that progressive iron accumulation does not always occur.26–29 As such, further consideration should also be given to the identification of suitable candidates for therapy30 by monitoring serum ferritin levels prior to and during any subsequent treatment. Some of the limitations of this study were that many of the patients had mild iron overload as assessed by serum ferritin levels, and a number of patients were treated with chelation therapy after having previously received phlebotomy. This was a phase 1/2 dose finding and safety assessment study of deferasirox in patients with HH, and as such the study involved a small number of patients. Therefore, further studies in larger numbers of patients with HH and assessments of the effects of de novo iron chelation therapy would be of value.

ed phlebotomy. This was a phase 1/2 dose finding and safety assessment study of deferasirox in patients with HH, and as such the study involved a small number of patients. Therefore, further studies in larger numbers of patients with HH and assessments of the effects of de novo iron chelation therapy would be of value. In conclusion, the findings reported here on the use of deferasirox to reduce iron in patients with nontransfusional iron overload due to HH are encouraging. Based on the safety and efficacy results, a starting dose of 10 mg/kg/day appears to be the most appropriate in this population; close monitoring of renal and hepatic function is required, especially in patients with lower body-iron burdens. Larger studies are warranted to more fully define the appropriate role of deferasirox for the treatment of selected patients. This study was sponsored by Novartis Pharmaceuticals Corp. Financial support for medical editorial assistance was provided by Novartis Pharmaceuticals. We thank Rebecca Helson for medical editorial assistance with this manuscript. Dr. Yves Deugnier and Dr. Fabrice Lainé from Pontchaillou University Hospital are also acknowledged for significant contributions to the conduct of this study. Abbreviations AEadverse event ALTalanine aminotransferase ASTaspartate aminotransferase C282Ycysteine-to-tyrosine mutation at position 282 HHhereditary hemochromatosis ULNupper limit of normal

Acute hepatitis C virus (HCV) infection in humans frequently progresses to chronicity,1 and virus persistence in the liver has been suggested to result, at least in part, from the ability of the virus to antagonize the interferon (IFN) system.2-5 Paradoxically, our ability to culture the virus for prolonged periods in differentiated primary hepatocytes in vitro has met with variable success.6-11 Use of the hepatoma line Huh-7 and its derivatives and adaptation of viral genomes to propagation in these cells has made possible the generation of high titer stocks of cell culture-derived HCV (HCVcc),12, 13 enabling the identification of cellular factors required for virus entry and replication.14-18 It has become apparent, however, that hepatoma lines may not fully recapitulate all aspects of HCV replication in the liver, and that host responses play an important part in determination of viral persistence or clearance. For example, nucleotide polymorphisms in or near the gene for the type III IFN, IL-28B, were recently shown to be predictive of resolution of acute HCV infection, or favorable response to IFN-alpha/ribavirin therapy in infected patients.19 The profound effect of these host polymorphisms may suggest a weak point in HCV's ability to evade the innate or adaptive immune response.

for the type III IFN, IL-28B, were recently shown to be predictive of resolution of acute HCV infection, or favorable response to IFN-alpha/ribavirin therapy in infected patients.19 The profound effect of these host polymorphisms may suggest a weak point in HCV's ability to evade the innate or adaptive immune response. In comparison to hepatoma lines, complex cultures of primary human hepatocytes from genetically diverse donors may provide a more informative environment for studying the virus life cycle and cellular mechanisms that may operate to limit virus spread. In the present study we examined the efficiency of HCVcc replication in primary human fetal liver cell cultures (HFLC). To investigate the possible role of the innate immune system in controlling productive HCV infection in these cultures we exploited the well-characterized ability of paramyxovirus (PMV) V proteins to counteract both IFN induction20 and antiviral signaling mediated by binding of the IFN receptor.21 All PMV genomes encode a unique open reading frame termed V. Although diverse in overall amino acid sequence (only ≍50% sequence identity between PMV family members) all V proteins share a conserved cysteine-rich C-terminus that interacts with the RNA helicase domain of the pattern recognition receptors (PRR) MDA5 and LGP2.22, 23In vitro, V proteins have been shown to block induction of type I IFN in response to stimuli that activate the MDA5 pathway.24, 25 Evidence to date indicates that V proteins do not engage or antagonize the related RNA helicase RIG-I.20, 22

A helicase domain of the pattern recognition receptors (PRR) MDA5 and LGP2.22, 23In vitro, V proteins have been shown to block induction of type I IFN in response to stimuli that activate the MDA5 pathway.24, 25 Evidence to date indicates that V proteins do not engage or antagonize the related RNA helicase RIG-I.20, 22 More extensively characterized is the ability of V proteins to potently inhibit cytokine signaling pathways by targeting STAT (signal transducer and activation of transcription).21 Receptor engagement by type I and type III IFNs results in the dimerization of STAT1 and STAT2 and assembly of the transcription complex ISGF3 that mediates expression of antiviral genes.21, 26, 27 The STAT molecules targeted by V proteins, and the mechanisms by which STAT signaling is inhibited, vary greatly between different PMV family members. The V protein of parainfluenza virus 5 (PIV5) targets STAT1 for proteosomal degradation by recruitment and assembly of components of the cellular E3 ubiquitin ligase machinery.28, 29 The V protein of measles virus (MV) targets both STAT1 and STAT2 and prevents their nuclear translocation in response to IFN receptor binding.30, 31

in of parainfluenza virus 5 (PIV5) targets STAT1 for proteosomal degradation by recruitment and assembly of components of the cellular E3 ubiquitin ligase machinery.28, 29 The V protein of measles virus (MV) targets both STAT1 and STAT2 and prevents their nuclear translocation in response to IFN receptor binding.30, 31 We introduced the V proteins of PIV5 and MV into cultured HFLC using bi-cistronic lentiviral vectors encoding a fluorescent reporter that permits direct visualization of HCV-infected cells.32 Our results show that V protein expression significantly enhances productive infection of HFLC with HCVcc, protects these cultures against the HCV-inhibitory effects of added type I and type III IFNs and antagonizes the induction of the type III IFN IL-29 in response to HCV infection. We also show by live-cell imaging that V protein expression dramatically enhances the early spread of HCV infection in these cultures. Patients and Methods Human Subjects All protocols involving human tissue were reviewed and exempted by the Rockefeller University Institutional Review Board.

We introduced the V proteins of PIV5 and MV into cultured HFLC using bi-cistronic lentiviral vectors encoding a fluorescent reporter that permits direct visualization of HCV-infected cells.32 Our results show that V protein expression significantly enhances productive infection of HFLC with HCVcc, protects these cultures against the HCV-inhibitory effects of added type I and type III IFNs and antagonizes the induction of the type III IFN IL-29 in response to HCV infection. We also show by live-cell imaging that V protein expression dramatically enhances the early spread of HCV infection in these cultures. Patients and Methods Human Subjects All protocols involving human tissue were reviewed and exempted by the Rockefeller University Institutional Review Board. Isolation and Culture of HFLC Deidentified fetal livers (16-24 weeks gestation) were procured through Advanced Bioscience Resources (ABR; Alameda, CA) or the Human Fetal Tissue Repository of the Albert Einstein College of Medicine (AECOM; New York, NY). Livers received on ice were washed with hepatocyte wash buffer (HWB) consisting of Williams' E Medium (WEM) plus 10 mM HEPES, 50 μg/mL gentamicin, 100 U/mL penicillin, and 100 μg/mL streptomycin (Invitrogen). Tissue was minced then resuspended in 20-40 mL warm digestion buffer consisting of Hanks Balanced Salt Solution plus 40 mM HEPES, 3.26 mM CaCl2, 2 U/mL DNase I Grade II (Roche), and 0.2% Collagenase type IV (Sigma). Tissue was digested for 30 minutes at 37°C, then diluted 1:1 with HWB and gently pushed through 70 μm cell-strainers (BD Biosciences). The suspension was centrifuged at 100g for 3 minutes and the cell pellet containing large hepatocytes was washed twice by resuspension in 50 mL HWB and centrifugation at 100g for 4 minutes. Hepatocytes were enriched by 1g sedimentation in 25 mL HWB for 1 hour at room temperature, followed by additional washing. In some experiments hepatocytes were further enriched by centrifugation through lymphocyte separation medium (Cellgro, Manassas, VA) as described.33 Hepatocyte yields ranged from 0.5 to 4 × 107 cells per tissue and cells were generally >80% viable as assessed by Trypan blue exclusion and collagen attachment. Hepatocytes were plated at ≍1 × 105/cm2 on 24- or 48-well collagen I-coated plates (BD Biosciences) in WEM containing 10% fetal bovine serum (FBS) (Omega Scientific, Tarzana, CA), 2 mM L-glutamine (Invitrogen), 1X ITS Plus (BD Biosciences) and antibiotics. After overnight incubation, adherent cells were washed with WEM, then maintained in Hepatocyte Defined Medium (HDM; BD Biosciences) plus L-glutamine and antibiotics. The culture medium was aspirated and replaced every 2 days.

ic, Tarzana, CA), 2 mM L-glutamine (Invitrogen), 1X ITS Plus (BD Biosciences) and antibiotics. After overnight incubation, adherent cells were washed with WEM, then maintained in Hepatocyte Defined Medium (HDM; BD Biosciences) plus L-glutamine and antibiotics. The culture medium was aspirated and replaced every 2 days. Lentiviral Vectors V proteins and control protein firefly luciferase (Fluc) were expressed from a hybrid albumin promoter in a bi-cistronic lentiviral vector34 modified to express the HCV-dependent fluorescence relocalization (HDFR) cassette TagRFP-NLS-IPS.32 In this cassette the fluorescent reporter TagRFP is fused to both a nuclear localization sequence (NLS) and the transmembrane domain of the mitochondrially tethered adapter protein IPS-1.32 Following HCV infection of HDFR-expressing cells, cleavage of IPS-1 by the viral NS3-4A protease2 leads to migration of the fluorescent reporter from mitochondria to the nucleus, enabling visualization of HCV-infected cells.32 Vector construction and source of protein-coding sequences are detailed in the Supporting Information and Supporting Fig. S1A. Transduction with Lentiviral Pseudoparticles (PP) PP were prepared by cotransfection of 293T cells with lentiviral and packaging plasmids as described.15, 32 HFLC were transduced 1-3 days postplating by incubation for 3-6 hours with PP stocks diluted 1:3 in HDM plus 20 mM HEPES and 4 μg/mL polybrene, then washed and fed with HDM.

n with Lentiviral Pseudoparticles (PP) PP were prepared by cotransfection of 293T cells with lentiviral and packaging plasmids as described.15, 32 HFLC were transduced 1-3 days postplating by incubation for 3-6 hours with PP stocks diluted 1:3 in HDM plus 20 mM HEPES and 4 μg/mL polybrene, then washed and fed with HDM. HCVcc Inocula HCVcc inocula used were the Gaussia luciferase reporter virus Jc1FLAG2 (p7-nsGluc2A)35 (JC1G) and J6JFH Clone 2,36 which lacks a reporter but replicates to higher titer in Huh-7.5 cells. Virus stocks were prepared by electroporation of in vitro transcribed RNA into Huh-7.5 cells as described35 and virus was collected in serum-free medium, or medium containing 1.5 or 10% FBS. For some experiments virus stocks were concentrated using Amicon Ultracel-100K filters (Millipore). Infectious titers of HCVcc inocula were determined by titration on Huh-7.5 cells as described12 and are expressed as 50% tissue culture infectious doses (TCID50). HFLC were infected for 3 to 6 hours with HCVcc diluted in HDM, then washed and fed with HDM. Detection of HCV RNA Total RNA was extracted from washed HFLC using RNEasy Kits (Qiagen). HCV RNA was detected by quantitative RT-PCR using the Eragen MultiCode-RTx method (Eragen Biosciences, Madison, WI) and primers directed to the 5′ untranslated region of the HCV genome. Quantitation of HCV RNA copy number was achieved using a synthetic RNA standard (Apath, Brooklyn, NY).

FLC using RNEasy Kits (Qiagen). HCV RNA was detected by quantitative RT-PCR using the Eragen MultiCode-RTx method (Eragen Biosciences, Madison, WI) and primers directed to the 5′ untranslated region of the HCV genome. Quantitation of HCV RNA copy number was achieved using a synthetic RNA standard (Apath, Brooklyn, NY). Cytokines and Drugs Recombinant human cytokines, IFN-beta (IFN-β), IL-28A, and IL-29, were from Peprotech (Rocky Hill, NJ). The HCV NS5B polymerase inhibitor 2′ C-methyl adenosine (2′CMA) was the gift of D. Olsen and S. Carroll (Merck Research Laboratories, West Point, PA). Immunofluorescence Analysis (IFA), Immunoblotting, and Enzyme-Linked Immunosorbent Assay (ELISA) IFA9 and immunoblotting14 were carried out as described. Antibodies and reagents are detailed in the Supporting Information. Human IL-29 was detected using ELISA kits from eBioscience (San Diego, CA). ELISA for human albumin is described in the Supporting Information. Time-Lapse Live Cell Imaging HFLC were plated on collagen-coated optical dishes and maintained in HDM. Images were acquired on a Zeiss Axiovert-200 inverted microscope equipped with an environmental chamber. Visualization of Tag-RFP in transduced HFLC was achieved by laser excitation and emission as described.32

Immunofluorescence Analysis (IFA), Immunoblotting, and Enzyme-Linked Immunosorbent Assay (ELISA) IFA9 and immunoblotting14 were carried out as described. Antibodies and reagents are detailed in the Supporting Information. Human IL-29 was detected using ELISA kits from eBioscience (San Diego, CA). ELISA for human albumin is described in the Supporting Information. Time-Lapse Live Cell Imaging HFLC were plated on collagen-coated optical dishes and maintained in HDM. Images were acquired on a Zeiss Axiovert-200 inverted microscope equipped with an environmental chamber. Visualization of Tag-RFP in transduced HFLC was achieved by laser excitation and emission as described.32 Results Cultured HFLC Are Long-Lived and Express HCV Entry Factors Like other investigators, we found human fetal liver-derived hepatocytes to be long-lived compared with cultured adult human hepatocytes.6, 7, 9 Figure 1A shows phase contrast morphology of cultured HFLC. Hepatocytes formed a tightly packed monolayer with refractive cell margins indicative of the formation of canaliculi. Hepatocytes retained this morphology and continued to secrete albumin for at least 1 month in culture (Fig. 1B).

ult human hepatocytes.6, 7, 9 Figure 1A shows phase contrast morphology of cultured HFLC. Hepatocytes formed a tightly packed monolayer with refractive cell margins indicative of the formation of canaliculi. Hepatocytes retained this morphology and continued to secrete albumin for at least 1 month in culture (Fig. 1B). Fig. 1 Cultured HFLC are long-lived, express markers of differentiated hepatocytes, and are positive for the expression of HCV entry factors. (A) Phase contrast microscopy of HFLC 2 weeks postplating. (B) Albumin secretion by HFLC during 1 month of culture (mean and SD of three cultures). (C) Antibody staining for cytokeratin 8 (CK8), cytokeratin 7 (CK7), vimentin (Vim), alpha-1 antitrypsin (AAT), alpha fetoprotein (AFP), and albumin (Alb) in cultured HFLC 1 week postplating. Bound antibody was detected with AlexaFluor (AF)-488-conjugated antibody to immunoglobulins. (D) AF-488 (green) detection of bound antibodies to the tight junction protein zona occludens 1 (ZO1) and the HCV entry factors occludin (OCLN) and claudin 1 (CLDN1) in cultured HFLC 1 week postplating. Hepatocyte staining for AAT (red) was detected with AF-594-conjugated antibody to goat IgG. Nuclei are counterstained with DAPI (blue). (E) Western blot for HCV entry factors in lysates of Huh-7.5 cells and HFLC 1 week postplating: Lane 1, 15 μg Huh-7.5 lysate; Lane 2, 30 μg Huh-7.5; Lane 3, 30 μg HFLC. We also examined entry factor expression in HFLC transduced with each of the three lentiviral vectors described below. Lanes 4-6 correspond to 30 μg lysate from HFLC transduced with pseudoparticles encoding: Lane 4, firefly luciferase; Lane 5, parainfluenza virus 5 V protein; Lane 6, measles virus V protein. Lysates were prepared 6 days posttransduction. SRBI, scavenger receptor BI. Note that HFLC express the ≍60 kDa form of OCLN that is found in adult human liver,37 and absent or weakly expressed in Huh-7.5 cells.

coding: Lane 4, firefly luciferase; Lane 5, parainfluenza virus 5 V protein; Lane 6, measles virus V protein. Lysates were prepared 6 days posttransduction. SRBI, scavenger receptor BI. Note that HFLC express the ≍60 kDa form of OCLN that is found in adult human liver,37 and absent or weakly expressed in Huh-7.5 cells. At 16 to 24 weeks of gestation, human fetal liver is considered to contain a mix of bi-potential hepatoblasts (capable of forming both hepatocytes and cholangiocytes) and their more committed progeny. Figure 1C shows IFA for the liver-derived proteins alpha-1 antitrypsin (AAT), alpha fetoprotein (AFP), and albumin, and for intermediate filament proteins in cultured HFLC. Hepatocytes stained uniformly positive for albumin and AAT and unevenly for AFP. CK8 expression was seen in both hepatocytes and cholangiocytes; the latter were visible as frequent clusters of CK7-positive cells within the hepatocyte monolayer. Unlike adult hepatocytes, fetal liver epithelial cells commonly coexpress mesenchymal markers such as vimentin in conjunction with hepatocyte markers.38 As shown in Fig. 1C, cultured HFLC showed strong staining for vimentin in cells with hepatocyte morphology and in cells with fibroblast morphology.

patocyte monolayer. Unlike adult hepatocytes, fetal liver epithelial cells commonly coexpress mesenchymal markers such as vimentin in conjunction with hepatocyte markers.38 As shown in Fig. 1C, cultured HFLC showed strong staining for vimentin in cells with hepatocyte morphology and in cells with fibroblast morphology. CLDN1 and OCLN, two of the four known entry factors for HCV, are tight junction proteins expressed preferentially at the canalicular domain in adult liver.16, 17 In cultured HFLC, CLDN1, OCLN, and the zona occludens marker ZO1 showed similar canalicular staining patterns (Fig. 1D). Immunoblot analysis of all four HCV entry factors (CLDN1, OCLN, SRBI, and CD81) in HFLC showed levels of expression comparable to those of the HCV-susceptible hepatoma Huh-7.5 (Fig. 1E). Taken together, these results show that cultured HFLC, although immature in phenotype, express all of the entry factors required for HCVcc infection. Cultured HFLC Are Susceptible to Infection with HCVcc We first tested the ability of cultured HFLC to support productive infection with HCVcc using the reporter virus JC1G, with Gaussia luciferase secretion as the readout for infection. Figure 2A shows that HFLC were susceptible to HCVcc infection and secreted low but measurable levels of luciferase, which persisted for at least 4 weeks of culture. Luciferase secretion was reduced by addition of the polymerase inhibitor 2′CMA during the first 6 days of culture, demonstrating the signal to be dependent on HCV RNA replication.

C were susceptible to HCVcc infection and secreted low but measurable levels of luciferase, which persisted for at least 4 weeks of culture. Luciferase secretion was reduced by addition of the polymerase inhibitor 2′CMA during the first 6 days of culture, demonstrating the signal to be dependent on HCV RNA replication. Fig. 2 Cultured HFLC are susceptible to infection with HCVcc. (A) Gaussia luciferase secretion by HFLC (ABR-15168, 23 weeks gestation) following infection with the reporter virus JC1G in the presence or absence of the polymerase inhibitor 2′CMA (2.2 μm). Mock-infected cells were incubated with culture medium alone. Polymerase inhibitor treatment was discontinued after day 6 of culture. Secreted luciferase was measured at each medium change (mean and SD of three cultures). RLU, relative light units. The dashed line indicates mean plus 2 SD of RLU detected in culture medium alone. (B) Doses of HCVcc required for establishment of detectable infection. HFLC (AECOM-052810, gestational age not available) were incubated with the indicated tissue culture infectious doses (TCID50) of JC1G for 5 hours, washed, and refed with HDM. Secreted luciferase was measured at each medium change (mean and SD of three cultures). (C) JC1G infection of four different HFLC preparations with ≍105 TCID50 of a single virus stock that was prepared in medium containing 10% v/v FBS.

e culture infectious doses (TCID50) of JC1G for 5 hours, washed, and refed with HDM. Secreted luciferase was measured at each medium change (mean and SD of three cultures). (C) JC1G infection of four different HFLC preparations with ≍105 TCID50 of a single virus stock that was prepared in medium containing 10% v/v FBS. Figure 2C shows the results of an experiment in which HFLC were infected with graded doses of JC1G ranging from 104-107 TCID50. Establishment of detectable infection required virus doses >105 TCID50 per culture (i.e., ≍1 TCID50 per seeded hepatocyte). Intriguingly, at all virus doses tested levels of secreted luciferase did not increase over time postinfection, but rather persisted at levels similar to those achieved during the first 2 days after virus inoculation (Fig. 2B). We also observed significant donor-to-donor variability in both the magnitude and duration of HCV replication between HFLC infected with equivalent doses of the same virus stock (Fig. 2C). In some experiments reporter virus replication declined in a manner suggestive of active viral clearance (Fig. 2C, and see below).

We also observed significant donor-to-donor variability in both the magnitude and duration of HCV replication between HFLC infected with equivalent doses of the same virus stock (Fig. 2C). In some experiments reporter virus replication declined in a manner suggestive of active viral clearance (Fig. 2C, and see below). These results are similar to those previously obtained with HCVcc infection of micropatterned adult human hepatocytes.9 They are in marked contrast to the exponential virus amplification obtained following HCVcc infection of Huh-7.5 cells.32 We hypothesized that, despite the documented ability of HCV to counteract innate antiviral sensing and signaling pathways,2-5 spread of HCV infection in primary HFLC may be limited by the IFN system. To investigate this hypothesis we exploited the ability of PMV V proteins to antagonize both IFN induction and IFN signaling by way of STAT.

, despite the documented ability of HCV to counteract innate antiviral sensing and signaling pathways,2-5 spread of HCV infection in primary HFLC may be limited by the IFN system. To investigate this hypothesis we exploited the ability of PMV V proteins to antagonize both IFN induction and IFN signaling by way of STAT. V Protein Expression Promotes Productive HCVcc Infection in HFLC We selected two PMV V proteins that differ in their STAT targeting specificity and mechanism of STAT inhibition. Expression of PIV5 and MV V proteins, and control protein Fluc, in HFLC was achieved by transduction with lentiviral PP encoding the HDFR cassette, which permits visualization of HCV-infected cells by nuclear translocation of RFP.32 Vector constructs and characterization of V protein expression are described in the Supporting Information and Supporting Fig. S1. Protein expression was found to be efficient in Huh-7 cells and hepatocytes (Supporting Fig. S1B-D), and vector transduction was found not to compromise HCV replication in HFLC (Supporting Fig. S1E).

structs and characterization of V protein expression are described in the Supporting Information and Supporting Fig. S1. Protein expression was found to be efficient in Huh-7 cells and hepatocytes (Supporting Fig. S1B-D), and vector transduction was found not to compromise HCV replication in HFLC (Supporting Fig. S1E). HFLC were either left untransduced or transduced with PP 1-3 days postplating, then infected with JC1G 4-6 days later. Figure 3A shows the effect of V protein transduction in one experiment, using both luciferase secretion (left panel) and infectious virus production (right panel) as the readouts for infection. Both luciferase secretion and virus release declined steadily between days 2 and 12 postinfection in control cultures. In contrast, cultures transduced with either PIV5 or MV V protein showed persistence of luciferase and maintenance of virus production over the same time period. Results from four independent infection experiments are shown in Fig. 3B. In each experiment, levels of infectious HCV recovered from HFLC supernatants at 2 weeks postinoculation were ≍1-2 logs higher in V protein-transduced cultures than control cultures. These results suggest that V protein expression serves to enhance productive infection, or “rescue” an abortive infection with HCVcc in these primary cells. Similar results were obtained in micropatterned cultures of adult human hepatocytes, indicating that the enhancing effects of V proteins on HCV replication are not limited to fetal hepatocytes (see Supporting Information and Supporting Fig. S2).

tion, or “rescue” an abortive infection with HCVcc in these primary cells. Similar results were obtained in micropatterned cultures of adult human hepatocytes, indicating that the enhancing effects of V proteins on HCV replication are not limited to fetal hepatocytes (see Supporting Information and Supporting Fig. S2). Fig. 3 PMV V protein expression enhances productive infection with JC1G in cultured HFLC. (A) One day postplating, HFLC (ABR-5591, 17 weeks gestation) were transduced with lentiviral vectors encoding firefly luciferase (Fluc), parainfluenza virus 5 V protein (PIV5 V), or measles virus V protein (MV V), or left nontransduced (NTD). Five days later cells were infected with 1 × 106 tissue culture infectious doses (TCID50) JC1G reporter virus. Washed cells were cultured in HDM and the medium was removed and replaced every 2 days. Harvested supernatants were assayed for luciferase (left panel) and for levels of infectious virus by titration on Huh-7.5 cells (right panel). RLU, relative light units. Values show mean and SD of three cultures. The dashed line indicates the lower limit of detection for each assay. (B) Levels of infectious virus recovered from cell supernatants 2 weeks post-JC1G infection of four different HFLC preparations. Graphs show mean and SD of 2-4 cultures. Log-transformed TCID50 values within each experiment were analyzed by 1-way analysis of variance (ANOVA) with Bonferroni multiple comparison posttest. Values that differ significantly from both nontransduced (NTD) and Fluc-transduced cultures are indicated: *P < 0.05; **P < 0.01; ***P < 0.001. Fold differences relative to nontransduced cultures are summarized below the graphs.

nalyzed by 1-way analysis of variance (ANOVA) with Bonferroni multiple comparison posttest. Values that differ significantly from both nontransduced (NTD) and Fluc-transduced cultures are indicated: *P < 0.05; **P < 0.01; ***P < 0.001. Fold differences relative to nontransduced cultures are summarized below the graphs. Direct Visualization of HCVcc Infection in Cultured HFLC We compared the effects of V protein expression on replication of both JC1G and J6JFH Clone 2, which replicates to comparatively higher titers on Huh-7.5 cells.36 Figure 4A shows levels of cell-associated HCV RNA detectable during the first 96 hours postinfection with an equivalent dose of each virus. In Fluc-transduced HFLC, levels of RNA for both viruses declined slightly between 8 and 96 hours postinfection. In V protein-transduced HFLC, HCV RNA levels increased over the same time period, with levels of J6JFH Clone 2 RNA increasing about one log during the first 24 hours. Figure 4C shows a comparison of these two viruses using RFP nuclear translocation32 as the readout for infection 7 days postinoculation. Nuclear translocation events were more readily quantifiable for J6JFH Clone 2, particularly in the context of V protein expression (Fig. 4B,C).

bout one log during the first 24 hours. Figure 4C shows a comparison of these two viruses using RFP nuclear translocation32 as the readout for infection 7 days postinoculation. Nuclear translocation events were more readily quantifiable for J6JFH Clone 2, particularly in the context of V protein expression (Fig. 4B,C). Fig. 4 J6JFH Clone 2 replicates to higher levels in HFLC than JC1G, and replication is enhanced by V protein expression. Four days posttransduction with lentiviral vectors, HFLC (AECOM-063010; gestational age not available) were incubated with 1 × 107 TCID50 JC1G or J6JFH Clone 2 for 6 hours, washed three times, and cultured in HDM. (A) Levels of cell-associated HCV RNA detected during the first 4 days postinfection (mean and SD of 2 assays on 2 cultures). (B) Representative images for HFLC infected with J6JFH Clone 2. (C) Numbers of RFP-positive nuclei per microscope field on day 7 postinfection (mean and SD of six fields from two cultures). As with JC1G, V protein expression enhanced J6JFH Clone 2 replication in HFLC, permitting sustained, 2′CMA-sensitive production of high titers of infectious virus for 14 days (Fig. 5A). Scoring for RFP translocation events during the first 72 hours postinfection revealed a steady increase in the numbers of HCV-infected cells in V protein-transduced HFLC, but not in Fluc-transduced cultures (Fig. 5B).

itting sustained, 2′CMA-sensitive production of high titers of infectious virus for 14 days (Fig. 5A). Scoring for RFP translocation events during the first 72 hours postinfection revealed a steady increase in the numbers of HCV-infected cells in V protein-transduced HFLC, but not in Fluc-transduced cultures (Fig. 5B). Fig. 5 Kinetics of J6JFH Clone 2 replication in cultured HFLC. (A) Five days posttransduction, HFLC (AECOM-031010, gestational age unavailable) were incubated with 1 × 107 TCID50 J6JFH Clone 2, washed three times, and refed with HDM with or without of 2.2 μm 2′CMA. Cells were washed again and refed 24 hours postinfection. Treatment with 2′CMA was discontinued on day 7. Infectious virus in culture supernatants was determined by titration on Huh-7.5 cells. Values are mean and SD of four cultures. (B) Kinetics of appearance of RFP-positive nuclei during the first 3 days of culture following infection with J6JFH Clone 2 (ABR-8338; 20 weeks gestation). Values are mean and SD of nuclei counts from four fields of two cultures. The latter finding prompted us to carry out time-lapse live cell imaging to monitor the appearance of new RFP translocation events after HCVcc infection (Fig. 6). The results showed little or no spread of infection in Fluc-transduced cultures at 48-119 hours postinfection, whereas V protein-transduced cultures showed numerous new translocation events. Representative examples of 10 independent fields are shown for each culture condition (Fig. 6). Video files are viewable in the online version of the manuscript.

spread of infection in Fluc-transduced cultures at 48-119 hours postinfection, whereas V protein-transduced cultures showed numerous new translocation events. Representative examples of 10 independent fields are shown for each culture condition (Fig. 6). Video files are viewable in the online version of the manuscript. Fig. 6 Live-cell imaging of J6JFH Clone 2 infection in PMV V protein- or Fluc-transduced HFLC. Five days posttransduction with lentiviral vectors, HFLC (ABR-3676; 19 weeks gestation) were infected with 1 × 107 TCID50 J6JFH Clone 2. Washed cells were cultured in HDM for 48 hours prior to initiation of imaging. For each culture condition, 10 fields containing at least one cell with nuclear translocation of RFP were selected to monitor possible spread of infection. The figure shows the first frame of representative fields of cells transduced with (A). Firefly luciferase (Fluc). (B) Parainfluenza virus 5 V protein (PIV5 V) or (C) measles virus V protein (MV V). Full videos for each field are shown in Fig. 6A_Fluc.avi, Fig. 6B_PIV5 V.avi, and Fig. 6C_MV V.avi and are viewable in the online version of the manuscript. The time stamp indicates elapsed time in hours beginning at 48 hours postinfection.

Parainfluenza virus 5 V protein (PIV5 V) or (C) measles virus V protein (MV V). Full videos for each field are shown in Fig. 6A_Fluc.avi, Fig. 6B_PIV5 V.avi, and Fig. 6C_MV V.avi and are viewable in the online version of the manuscript. The time stamp indicates elapsed time in hours beginning at 48 hours postinfection. These results suggest that a major impact of V proteins in HFLC is to promote virus spread during the early stages of infection. V protein expression did not measurably affect levels of HCV entry factors in HFLC (Fig. 1E), or entry of HCV-enveloped PP (Supporting Fig. S3), suggesting that their enhancing effect on infectious HCV spread is not mediated at the level of virus entry. We next tested which of the V proteins' known anti-IFN functions (i.e., inhibition of STAT signaling and inhibition of IFN induction) were active in HFLC. V Proteins Counteract the HCV-Inhibitory Effects of Added IFNs in HFLC Although they bind to distinct membrane receptors, both type I and type III IFNs induce antiviral signaling by way of activation of STAT1 and STAT2.26, 27 We tested the ability of V proteins to counteract the HCV-inhibitory effects of IFN-β and type III IFNs, IL-28A, and IL-29 using the JC1G reporter virus to measure productive infection of HFLC (Fig. 7).

ne receptors, both type I and type III IFNs induce antiviral signaling by way of activation of STAT1 and STAT2.26, 27 We tested the ability of V proteins to counteract the HCV-inhibitory effects of IFN-β and type III IFNs, IL-28A, and IL-29 using the JC1G reporter virus to measure productive infection of HFLC (Fig. 7). Fig. 7 V protein expression counteracts the HCV-inhibitory effects of added type I and type III interferons in HFLC. Four days posttransduction with lentiviral vectors, HFLC (ABR-1108; 19 weeks gestation) were incubated with 1 × 106 TCID50 per well JC1G reporter virus for 5 hours. Cells were washed three times then refed with HDM containing serial 10-fold dilutions of IFN-β, IL-28A, or IL-29. The culture medium was aspirated and replaced every 2 days. Cytokine treatment was discontinued after the second day of culture. (A) JC1G reporter virus-derived luciferase activity detected in culture supernatants during the first 6 days of culture (mean and SD of two cultures). RLU, relative light units. (B) Levels of infectious HCV detected in cell culture supernatants on day 14 of culture (mean and SD of two assays on two cultures). TCID50, tissue culture infectious doses; Fluc, firefly luciferase; PIV5 V, parainfluenza virus 5 V protein; MV V, measles virus V protein.

SD of two cultures). RLU, relative light units. (B) Levels of infectious HCV detected in cell culture supernatants on day 14 of culture (mean and SD of two assays on two cultures). TCID50, tissue culture infectious doses; Fluc, firefly luciferase; PIV5 V, parainfluenza virus 5 V protein; MV V, measles virus V protein. IFNs were added to the culture medium immediately after HCVcc infection and withdrawn after the second day of culture. In control-vector-transduced HFLC, all three IFNs inhibited HCV replication in a dose-dependent manner, with maximal inhibition occurring with 100 U/mL IFN-β, 100 ng/mL IL-28A, and 10 or 100 ng/mL IL-29 (Fig. 7A). With each dose of each IFN the level of HCV replication in HFLC transduced with PIV V protein was equivalent to that of cells cultured in the absence of added cytokine, indicating potent antagonism of STAT signaling and confirming the efficiency of lentiviral transduction in these cultures. HFLC transduced with MV V protein were likewise protected, and for most IFN doses, luciferase levels were equivalent to or greater than those of control cells cultured without added cytokine (Fig. 7A).

cating potent antagonism of STAT signaling and confirming the efficiency of lentiviral transduction in these cultures. HFLC transduced with MV V protein were likewise protected, and for most IFN doses, luciferase levels were equivalent to or greater than those of control cells cultured without added cytokine (Fig. 7A). To confirm that the 2-day IFN dosing schedule used was sufficient to reduce productive HCV infection, we measured levels of infectious virus in HFLC supernatants after 14 days of culture (i.e., 12 days after removal of added IFN) (Fig. 7B). Infectious virus was recovered from supernatants of V protein-transduced, but not Fluc-transduced HFLC, treated with the maximally inhibitory doses of each IFN defined above. In addition, levels of HCV produced by HFLC cultured without added cytokine were significantly higher for V protein-transduced cells (27-fold for MV V, and 15-fold for SV5) than control cells (Fig. 7B, left panel). These data suggest that V protein expression counteracts the inhibitory effects of added IFN and also antagonizes endogenous cytokines generated during the course of infection. V Protein Expression Inhibits IL-29 Protein Induction in HCVcc-Infected HFLC In the accompanying article we report that acute infection of HFLC with J6JFH Clone 2 results in the secretion of IL-29 into the culture supernatant and induction of mRNAs for IL-29 and IL-28B.39 Detectable induction of type III IFNs was less frequent in HFLC infected with the JC1G reporter virus, possibly reflecting the superior replication efficiency of J6JFH Clone 2 relative to JC1G.

th J6JFH Clone 2 results in the secretion of IL-29 into the culture supernatant and induction of mRNAs for IL-29 and IL-28B.39 Detectable induction of type III IFNs was less frequent in HFLC infected with the JC1G reporter virus, possibly reflecting the superior replication efficiency of J6JFH Clone 2 relative to JC1G. In light of these results, we determined whether PMV V protein expression would modulate IL-29 induction after acute infection of HFLC with J6JFH Clone 2. As shown in Fig. 8, both untransduced and control transduced HFLC showed a peak of IL-29 production at 48 hours postinfection with HCVcc, which declined by 72 hours. No detectable IL-29 production was seen in mock-infected HFLC (assay cutoff 50 ng/mL). In HFLC transduced with either PIV5 or MV V proteins, little or no IL-29 induction was seen at any timepoint postinfection with HCVcc. Fig. 8 V protein expression inhibits IL-29 protein induction following HCV infection. Four days posttransduction with lentiviral vectors, cultures of gradient-enriched HFLC (ABR-8338; 20 weeks gestation) were incubated for 5 hours with (A) 1 × 106 TCID50 per well J6JFH Clone 2, or (B) with an equivalent dilution of supernatant from mock-electroporated Huh-7.5 cells. Cells were washed three times then refed with HDM. The culture medium was removed and replaced daily and supernatants were assayed for IL-29 by ELISA (mean and SD of two cultures). Comparable results were obtained in two other experiments. NTD, no transduction; Fluc, firefly luciferase; PIV5 V, parainfluenza virus 5 V protein; MV V, measles virus V protein.

d with HDM. The culture medium was removed and replaced daily and supernatants were assayed for IL-29 by ELISA (mean and SD of two cultures). Comparable results were obtained in two other experiments. NTD, no transduction; Fluc, firefly luciferase; PIV5 V, parainfluenza virus 5 V protein; MV V, measles virus V protein. Discussion We have shown that primary cultures of HFLC reliably support productive infection with HCVcc. Virus replication was sensitive to inhibition by the polymerase inhibitor 2′CMA (Figs. 2A, 5A), and to added IFNs (Fig. 7). Levels of virus replication varied significantly between different donor cell preparations and frequently declined in a manner suggestive of active viral clearance (Figs. 2, 3). At present, we do not know whether this variability is due to differences in cellular composition or state of differentiation of hepatocytes within these cultures or to genetic differences between tissue donors. Studies to date have not revealed an association with IL28B genotype.39 V protein-transduced HFLC supported significantly enhanced (≍10 to 100-fold) levels of HCV infection relative to untransduced or control vector-transduced HFLC. Infection was assessed by measurement of virus-driven luciferase, by assays for infectious HCV and viral RNA, and by direct visualization of HCV-infected hepatocytes (Figs. 3-5). Similar results were obtained with micropatterned cocultures of adult human hepatocytes11 (Supporting Fig. S2).

ntrol vector-transduced HFLC. Infection was assessed by measurement of virus-driven luciferase, by assays for infectious HCV and viral RNA, and by direct visualization of HCV-infected hepatocytes (Figs. 3-5). Similar results were obtained with micropatterned cocultures of adult human hepatocytes11 (Supporting Fig. S2). Time-lapse live cell imaging of HFLC 48-119 hours postinfection demonstrated little or no spread of infection in the absence of V protein expression. In contrast, V protein-transduced HFLC showed numerous HCV infection events (Fig. 6). To our knowledge, this is the first report of visualization of HCV spread in primary cells. During the course of live cell imaging we observed considerable turnover of HCV-infected cells. Cell turnover could indicate a direct cytopathic effect of HCV infection in primary hepatocytes. However, more comprehensive studies will be required to determine the survival time of HCV-infected cells relative to uninfected hepatocytes in these cultures.

maging we observed considerable turnover of HCV-infected cells. Cell turnover could indicate a direct cytopathic effect of HCV infection in primary hepatocytes. However, more comprehensive studies will be required to determine the survival time of HCV-infected cells relative to uninfected hepatocytes in these cultures. V proteins did not measurably affect levels of expression of HCV entry factors in HFLC (Fig. 1), nor did they function to promote virus entry as assessed by experiments using HCV-enveloped pseudoparticles (Supporting Fig. S3), or measurements of cell-associated HCV RNA 8 hours postinfection (Fig. 4A). Taken together, these results suggest that V proteins exert their effect by mechanisms independent of an effect on virus entry. Consistent with their known role in counteracting innate immunity during PMV infection, we found V protein expression to efficiently antagonize the HCV-inhibitory effects of added IFNs in HFLC (Fig. 7).

ther, these results suggest that V proteins exert their effect by mechanisms independent of an effect on virus entry. Consistent with their known role in counteracting innate immunity during PMV infection, we found V protein expression to efficiently antagonize the HCV-inhibitory effects of added IFNs in HFLC (Fig. 7). Interestingly, induction of the type III IFN, IL-29, that follows acute HCVcc infection of HFLC39 was inhibited in V protein-transduced cultures (Fig. 8). Studies conducted predominantly in hepatoma lines have defined RIG-I as the primary PRR required for IFN induction after HCV infection.5 Because V proteins do not engage RIG-I, but do bind helicases MDA5 and LGP2,20, 22 our results may suggest that recognition of HCV RNA occurs differently in primary HFLC cultures, and that infection of these cells leads to the generation of RNA species that are capable of directly activating the MDA5 pathway. Alternatively, they may reflect V protein antagonism of the PRR up-regulation that has been shown to accompany IFN stimulation.40 Previous work from our laboratory has shown that overexpression of RIG-I or MDA5 is inhibitory for HCV replication in Huh-7 cells.34 Additional studies are required to further define the mechanisms for V protein-mediated enhancement of HCV replication in primary culture.

t has been shown to accompany IFN stimulation.40 Previous work from our laboratory has shown that overexpression of RIG-I or MDA5 is inhibitory for HCV replication in Huh-7 cells.34 Additional studies are required to further define the mechanisms for V protein-mediated enhancement of HCV replication in primary culture. Chimeric HCV genomes, such as those used in our study, enable the production of well-defined virus stocks for use in infection experiments. In contrast, HCV derived from patient plasma or tissues may be of variable infectious titer, and frequently complexed with virus-neutralizing antibody. Reports of successful infection of primary hepatocytes with such isolates are relatively rare, and the degree of replication achieved has varied widely between different culture systems.6-11 Few have reported sustained production of titratable infectious virus. Many factors may contribute to this variability, including the differentiation state of cultured hepatocytes and differences between HCV genomes in their relative dependence on cellular cofactors required for virus replication and spread. Our results with HCVcc suggest that the hepatocyte innate response to infection may provide an additional barrier to productive replication in primary culture. Strategies aimed at dampening this response may be key to the further development of robust HCV culture systems based on infection or transfection of viral genomes. Abbreviations 2′CMA2′C-methyl adenosine AATalpha-1 antitrypsin AFPalpha fetoprotein CKcytokeratin CLDN1claudin 1 Flucfirefly luciferase HCVhepatitis C virus

Our results with HCVcc suggest that the hepatocyte innate response to infection may provide an additional barrier to productive replication in primary culture. Strategies aimed at dampening this response may be key to the further development of robust HCV culture systems based on infection or transfection of viral genomes. Abbreviations 2′CMA2′C-methyl adenosine AATalpha-1 antitrypsin AFPalpha fetoprotein CKcytokeratin CLDN1claudin 1 Flucfirefly luciferase HCVhepatitis C virus HCVcccell culture-derived HCV HDMhepatocytes defined medium HFLChuman fetal liver cells HWBhepatocyte wash buffer IFNinterferon JC1GJc1FLAG2p7-nsGluc2A MVmeasles virus OCLNoccludin PIV5parainfluenza virus 5 PMVparamyxovirus PPpseudoparticle RLUrelative light units SRB1scavenger receptor B1 STATsignal transducer and activation of transcription TCID50fifty percent tissue culture infectious doses WEMWilliams'E medium ZO1zona occludens 1 Supplementary material Additional Supporting Information may be found in the online version of this article. Supporting Video 1: Firefly luciferase (Fluc) Supporting Video 2: Parainfluenza virus 5 V protein (PIV5 V) or (C). Supporting Video 3: Measles virus V protein (MV V).

After infection with hepatitis C virus (HCV), outcomes are variable: spontaneous resolution of the infection is observed in approximately 30% of individuals, but for others, chronic infection develops. Factors such as age, gender, and host genetic variants have been associated with different infection outcomes1,2 (reviewed by Rauch et al.3). Study cohorts that capture all individuals exposed to the virus, such as HCV single-source outbreak cohorts4,5 and cohorts of individuals who have a high risk of HCV exposure,6 have been particularly important in delineating relevant viral and host factors associated with the outcome of HCV infection. Such studies corroborate other studies indicating that a host's T cell response to HCV, including genes involved in regulating this response, is an important correlate of infection outcome.7-11

have been particularly important in delineating relevant viral and host factors associated with the outcome of HCV infection. Such studies corroborate other studies indicating that a host's T cell response to HCV, including genes involved in regulating this response, is an important correlate of infection outcome.7-11 T cell immune responses are stimulated by the presentation of processed viral peptides (epitopes) by human leukocyte antigen (HLA) molecules to CD4+ and CD8+ T cells. This host-virus interaction is dependent on the sequence of the viral epitope and surrounding regions, which play a role in peptide processing and presentation to T cells. Viral adaptations can reduce the binding affinity of the peptide to the HLA molecule and result in poor peptide cleavage or poor T cell recognition; these factors can subvert host immune control (reviewed by Bowden and Walker12). The importance of immune control in HCV infection has been illustrated in studies showing that mutations in CD8+ T cell epitopes contribute to viral persistence in both chimpanzees and humans.13,14 Accordingly, the extent to which the virus can adapt to the host's immune response is likely to be an important factor in determining infection outcome. These adaptations are dependent on the sequence of the incoming virus and the balance between the fitness cost incurred by these mutations15 and their benefit to the virus due to immune escape.

which the virus can adapt to the host's immune response is likely to be an important factor in determining infection outcome. These adaptations are dependent on the sequence of the incoming virus and the balance between the fitness cost incurred by these mutations15 and their benefit to the virus due to immune escape. It is unclear how much genetic diversity observed in HCV is the result of host immune pressures. Recent studies have suggested that viral adaptation can be observed at both the individual level16,17 and the population level.18,19 For example, genetic studies examining HCV sequences in the context of the HLA repertoire of a host population have shown associations between specific polymorphisms across the viral genome and HLA types within individuals in a host population.18,19 These HLA-associated viral polymorphisms are thought to represent viral adaptations and tag regions of the viral genome that are under in vivo T cell pressure. However, HCV evolution is shaped by evolutionary forces that include genetic drift and both positive and purifying selection pressures.20,21 It is likely that all these factors exert their influence simultaneously on the virus and affect the ability of the virus to adapt to new selection pressures and/or revert in a new host.

, HCV evolution is shaped by evolutionary forces that include genetic drift and both positive and purifying selection pressures.20,21 It is likely that all these factors exert their influence simultaneously on the virus and affect the ability of the virus to adapt to new selection pressures and/or revert in a new host. A previous study of an Irish HCV single-source cohort showed evidence of immune selection in known T cell targets.22 In this study, we compared HCV sequences from 63 individuals with genotype 1b infection from this single-source outbreak5 to identify sites likely representing new T cell targets in the HCV genome and to determine the extent to which host immune pressures on the virus affected sequence diversity in the cohort. Knowledge of the incoming viral sequence also allowed us to determine whether preexisting viral adaptations could predict beneficial or detrimental host HLA alleles within the cohort with respect to infection outcomes.

ine the extent to which host immune pressures on the virus affected sequence diversity in the cohort. Knowledge of the incoming viral sequence also allowed us to determine whether preexisting viral adaptations could predict beneficial or detrimental host HLA alleles within the cohort with respect to infection outcomes. Patients and Methods Study Population The study population was part of a cohort of women who had been infected with HCV between May 1977 and November 1978 in Ireland through the administration of anti-D immunoglobulin that had been contaminated with an HCV genotype 1b virus originating from a single individual.5 From this original cohort, we studied 63 individuals with chronic HCV infection; a subset (n = 15) was selected on the basis of the carriage of HLA-A*03, an allele that was previously shown to be protective in this cohort.8 A comparison of the HLA alleles found in this cohort and those in another Irish population is in the Supporting Information. Serum samples from the subjects were collected between 1996 and 2002 and were stored at −80°C. Written, informed consent was obtained from participants, and local institutional review board approval was obtained by all centers contributing to the study. Viral RNA Extraction Viral RNA was extracted from serum samples with the QIAamp Viral RNA mini kit (Qiagen) or the Cobas Amplicor HCV specimen preparation kit (version 2.0, Roche) according to each manufacturer's instructions. HLA Genotyping Two-digit resolution HLA class I (HLA-A, HLA-B, and HLA-C) typing was performed at St. James Hospital (Dublin, Ireland).8

Viral RNA Extraction Viral RNA was extracted from serum samples with the QIAamp Viral RNA mini kit (Qiagen) or the Cobas Amplicor HCV specimen preparation kit (version 2.0, Roche) according to each manufacturer's instructions. HLA Genotyping Two-digit resolution HLA class I (HLA-A, HLA-B, and HLA-C) typing was performed at St. James Hospital (Dublin, Ireland).8 Interleukin-28B (IL-28B) Genotyping Genotyping of the single-nucleotide polymorphism (SNP) rs12979860 upstream of the IL-28B gene was performed for 34 subjects as previously described.23 Bulk Viral Sequencing HCV sequencing was performed as previously described.18,19 Briefly, three separate reverse-transcription PCRs were performed which overlapped to cover the core to nonstructural (NS) 5B region. The first-round products were used as templates in nested second-round polymerase chain reactions containing generic or genotype-specific primers. Amplicons were bulk-sequenced with the BigDye Terminator version 3.1 cycle sequencing kit (Applied Biosystems) according to the manufacturer's recommendations, and electropherograms were edited with Assign (Conexio Genomics). Mixtures were identified in which the secondary peak was greater than 20% of the main peak. HCV sequences in this study have been submitted to GenBank (accession numbers HM106522 to HM106981). Supporting Information Table 1 lists the mean sequence coverage by protein. An analysis of the viral sequences for testing the single-source nature of this outbreak can be found in the Supporting Information.

Bulk Viral Sequencing HCV sequencing was performed as previously described.18,19 Briefly, three separate reverse-transcription PCRs were performed which overlapped to cover the core to nonstructural (NS) 5B region. The first-round products were used as templates in nested second-round polymerase chain reactions containing generic or genotype-specific primers. Amplicons were bulk-sequenced with the BigDye Terminator version 3.1 cycle sequencing kit (Applied Biosystems) according to the manufacturer's recommendations, and electropherograms were edited with Assign (Conexio Genomics). Mixtures were identified in which the secondary peak was greater than 20% of the main peak. HCV sequences in this study have been submitted to GenBank (accession numbers HM106522 to HM106981). Supporting Information Table 1 lists the mean sequence coverage by protein. An analysis of the viral sequences for testing the single-source nature of this outbreak can be found in the Supporting Information. Ultradeep Sequencing To identify minor quasispecies below the detection threshold of bulk sequencing methods, ultradeep sequencing was carried out with the 454 Life Sciences platform (Roche Applied Science) for two individuals (HLA-A*03+/HLA-B*08− and HLA-A*03−/HLA-B*08−). With the previously described amplification method, polymerase chain reaction templates were obtained that covered NS3 (positions 3494-4530) and NS5A to NS5B (positions 7335-8356). Amplicons were quantified and pooled for each individual. Library preparation and sequencing were performed according to the manufacturer's protocol. Data were collected and analyzed with Roche and public license software programs. All sequence reads were aligned to the source sequence (AF313916) with GS Reference Mapper software (Roche). The threshold for mixtures was set at 1% with 100-fold or greater coverage.

e performed according to the manufacturer's protocol. Data were collected and analyzed with Roche and public license software programs. All sequence reads were aligned to the source sequence (AF313916) with GS Reference Mapper software (Roche). The threshold for mixtures was set at 1% with 100-fold or greater coverage. HLA-Associated Viral Polymorphisms Associations between HLA alleles and amino acid distributions at each residue of the HCV proteins were assessed with Fisher's exact test for classification as consensus or nonconsensus amino acid. A false discovery rate analysis was carried out, and q values were obtained as reported previously.19 Only sequences with ≥50% sequence coverage for each respective protein were used. Analyses were carried out with Spotfire S+ 8.1 (TIBCO, Somerville, MA). Associations with a P value ≤0.01 for Fisher's exact test of consensus versus nonconsensus are reported. An assessment of possible confounding by founder effects via viral cluster stratification and the Mantel-Haenszel procedure, as described by Rauch et al.,19 indicated that no correction for significant associations was necessary, and this was consistent with the sequences originating from a single source. In addition, because P values associated with relatively small frequencies can be affected by small numbers of misclassified cases, we restricted our analysis to associations for which there were five or more nonconsensus amino acids and five or more carriers of the HLA allele.

quences originating from a single source. In addition, because P values associated with relatively small frequencies can be affected by small numbers of misclassified cases, we restricted our analysis to associations for which there were five or more nonconsensus amino acids and five or more carriers of the HLA allele. Sliding-Window Analysis In order to identify viral escape that might not be captured with a single amino acid approach, an analysis was conducted as described previously, except that adaptation was defined as nonconsensus at any residue within sliding windows of nine amino acids, which represented typical peptide sizes for HLA class I molecules. Significant sites of associations were identified as strings of significant values, whereas the window slid over any residues containing strong associations or combinations of associations. We restricted the analysis to cases that had all amino acids in the window. Associations with P ≤ 0.01 were reported. Covariation Residue covariation was assessed with Fisher's exact test for classification as consensus or nonconsensus amino acid. Covariation based on a sequence with ≥90% coverage was reported; covarying sites had P ≤ 0.001 for amino acid versus amino acid comparison and P ≤ 0.0001 for amino acid versus nucleotide comparison. Because of the exploratory nature of this part of the analysis, no adjustment was made for multiple comparisons.

no acid. Covariation based on a sequence with ≥90% coverage was reported; covarying sites had P ≤ 0.001 for amino acid versus amino acid comparison and P ≤ 0.0001 for amino acid versus nucleotide comparison. Because of the exploratory nature of this part of the analysis, no adjustment was made for multiple comparisons. Peptide Prediction for HLA-Associated Viral Polymorphism Sites Flanking sequences of the identified HLA-associated viral polymorphisms and sites of common divergence from the source sequence were entered into the epitope prediction software SYFPEITHI24 to identify putative epitopes based on a cutoff score of 20 with the highest scoring peptide reported. HLA-associated viral polymorphism sites were compared against published genotype 1 epitopes found in the Immune Epitope Database (http://www.immuneepitope.org). Viral Sequence Diversity Sequence diversity from the source sequence (AF313916) was determined with the Highlighter program (available at http://www.lanl.gov) for NS3 and NS5B to identify sites of synonymous and nonsynonymous substitutions for sequences with greater than 50% sequence coverage. Genetic diversity was determined with the Kimura two-parameter model, and differences in the rate of nonsynonymous and synonymous changes (ds/dn) were obtained with the modified Nei and Gojobori method with MEGA version 3.1.25

nymous and nonsynonymous substitutions for sequences with greater than 50% sequence coverage. Genetic diversity was determined with the Kimura two-parameter model, and differences in the rate of nonsynonymous and synonymous changes (ds/dn) were obtained with the modified Nei and Gojobori method with MEGA version 3.1.25 IL-28B–Associated Viral Polymorphisms We assessed associations between the presence or absence of the minor allele rs12979860 and consensus or nonconsensus amino acids at each residue of the HCV proteins via Fisher's exact test. Because of the smaller number of subjects with typing available for this part of the analysis, no assessment of false discovery rates was made, and P ≤ 0.01 was used to indicate significance.

e minor allele rs12979860 and consensus or nonconsensus amino acids at each residue of the HCV proteins via Fisher's exact test. Because of the smaller number of subjects with typing available for this part of the analysis, no assessment of false discovery rates was made, and P ≤ 0.01 was used to indicate significance. Results HLA-Associated Viral Polymorphisms: Putative Viral Adaptations in the New Hosts Reflecting Sites of Immune Pressure We determined whether there were associations between the expression of particular HLA alleles in subjects in this cohort and specific polymorphisms in their viral sequences (putative viral adaptations) reflecting areas under in vivo T cell immune pressure. We identified 29 HLA-associated viral polymorphisms with P ≤ 0.01 for 23 sites along the HCV genome (Table 1 and Supporting Information Fig. 3). In some instances, HLA alleles from different loci were associated with the same site, and we have previously shown that these associations can be explained in part by the linkage disequilibrium observed within the major histocompatibility complex (MHC).18 Among those associations shown in Table 1, three HLA-B/C combinations are associated with common MHC haplotypes. The q values for associations within some of the proteins are high with respect to others (particularly E2) and possibly reflect smaller sample sizes in these proteins (Supporting Information Table 1). Table 1 HLA Class I-Associated Viral Polymorphisms

Results HLA-Associated Viral Polymorphisms: Putative Viral Adaptations in the New Hosts Reflecting Sites of Immune Pressure We determined whether there were associations between the expression of particular HLA alleles in subjects in this cohort and specific polymorphisms in their viral sequences (putative viral adaptations) reflecting areas under in vivo T cell immune pressure. We identified 29 HLA-associated viral polymorphisms with P ≤ 0.01 for 23 sites along the HCV genome (Table 1 and Supporting Information Fig. 3). In some instances, HLA alleles from different loci were associated with the same site, and we have previously shown that these associations can be explained in part by the linkage disequilibrium observed within the major histocompatibility complex (MHC).18 Among those associations shown in Table 1, three HLA-B/C combinations are associated with common MHC haplotypes. The q values for associations within some of the proteins are high with respect to others (particularly E2) and possibly reflect smaller sample sizes in these proteins (Supporting Information Table 1). Table 1 HLA Class I-Associated Viral Polymorphisms Two HLA-associated viral polymorphisms fell within previously published epitopes (HLA-A*02 epitope in E2 404 SLLAPGAKQNV and HLA-A*03 epitope in NS5B 2518 SLTPPHSAK; Table 1). Furthermore, three HLA-associated viral polymorphisms fell within predicted epitopes as determined by the peptide binding prediction program SYFPEITHI24 (Table 1). The limited number of matches between known epitopes and putative viral adaptation sites may be the result of the small number of published HCV epitopes in the literature and its focus on common HLA types. Several of the putative viral adaptations are associated with HLA-C alleles for which there are either no or few known HLA-restricted epitopes or characterized binding properties.

viral adaptation sites may be the result of the small number of published HCV epitopes in the literature and its focus on common HLA types. Several of the putative viral adaptations are associated with HLA-C alleles for which there are either no or few known HLA-restricted epitopes or characterized binding properties. None of the associations shown in Table 1 overlap with the findings of our previous studies examining HLA-associated viral polymorphisms for genotype 1.18,19 However, the previous study had a much larger number of genotype 1a sequences in the data set than 1b sequences; because the sequences in this single-source cohort were all genotype 1b, it was likely that we would observe differential escape profiles similar to what we had seen between genotypes 1a and 3a but to a lesser extent between genotype 1 subtypes (1a and 1b). Furthermore, in contrast to the subjects in the previous cross-sectional studies, the subjects in this study were infected from a single-source strain.

at we would observe differential escape profiles similar to what we had seen between genotypes 1a and 3a but to a lesser extent between genotype 1 subtypes (1a and 1b). Furthermore, in contrast to the subjects in the previous cross-sectional studies, the subjects in this study were infected from a single-source strain. Window Analysis Identifies Additional Areas Under T Cell Pressure Areas under HLA-specific immune pressure that can accommodate more than one site of variation may not be detected by our initial single amino acid approach. Accordingly, a sliding-window analysis (with a size reflective of a typical HLA class I epitope) was also performed to examine areas under HLA-specific immune pressure in which more than one site might be relevant for escape. As expected, several of the HLA-associated viral polymorphisms identified with a single-site analysis were identified with the window analysis (Table 1). However, the single-site associations found in highly variable regions in E2 were not identified in the window analysis, probably because of the higher level of variation found in this region in comparison with other proteins that may occur in some cases when the variation is not related to adaptation (as tested here) and may hinder the ability to find specific HLA associations with any change(s) within a window. There were three examples [E2 and HLA-C*06 with a median position of 537, odds radio (OR) = 28; NS2 and HLA-B*08 within windows of 875-878, OR = 0.026-0.039; and NS5A and HLA-B*08 with a median position of 2132, OR = 26] for which the window analysis identified HLA-associated substitutions that were not found to be significant in the single-site analysis. These cases suggested that multiple sites within a target region may be under immune pressure (Supporting Information Fig. 4). This observation is consistent with our own study and other studies showing different escape profiles within epitopes, including the immunodominant HLA-B*08 epitope (1395-1403) in NS317 and the protective HLA-B*27 epitope (2841-2849) in NS5B.11

target region may be under immune pressure (Supporting Information Fig. 4). This observation is consistent with our own study and other studies showing different escape profiles within epitopes, including the immunodominant HLA-B*08 epitope (1395-1403) in NS317 and the protective HLA-B*27 epitope (2841-2849) in NS5B.11 Overall, the number of associations found with either the single-site analysis or the sliding-window analysis represented only a portion of the 184 variable sites across the viral genome that fit the inclusion criteria described in the methods (18 of 163 if the highly variable region in E2 is excluded because this area is likely to also be under other strong selective pressures). Source and Causes of Viral Adaptation We then examined the pattern of synonymous and nonsynonymous changes in these sequences to determine if purifying selection was acting across the HCV genome and potentially restricting the ability of the virus to adapt to new selection pressures or revert to unadapted forms. Figure 1 shows the pattern of these changes in each individual with respect to the source within the NS3 and NS5B proteins. It is apparent that there are a greater number of synonymous changes with respect to nonsynonymous changes in this region (indicating purifying or negative selection; dS-dN for NS3 = 0.080 and for NS5B = 0.061). Similar results were observed for other proteins (data not shown).

source within the NS3 and NS5B proteins. It is apparent that there are a greater number of synonymous changes with respect to nonsynonymous changes in this region (indicating purifying or negative selection; dS-dN for NS3 = 0.080 and for NS5B = 0.061). Similar results were observed for other proteins (data not shown). Fig. 1 Highlighter plot of synonymous and nonsynonymous substitutions in NS3 and NS5B with respect to the source sequence (AF313916). The plot was created with Highlighter (available at http://www.lanl.gov). Red lines denote nonsynonymous substitutions, green lines indicate synonymous substitutions, and gray regions show unsequenced sections.

of synonymous and nonsynonymous substitutions in NS3 and NS5B with respect to the source sequence (AF313916). The plot was created with Highlighter (available at http://www.lanl.gov). Red lines denote nonsynonymous substitutions, green lines indicate synonymous substitutions, and gray regions show unsequenced sections. Covarying Sites in the Genome Likely to Reflect Networks Within the HCV Genome As previously suggested, purifying selection may reflect the existence of covarying sites in the HCV genome.26 Here we identified sites of covariance by assessing amino acid sites in a pairwise manner per protein and genome-wide for sequences with greater than 90% sequence coverage. Only results with P < 0.001 were reported because adjustments for multiple comparisons were not made in this analysis. Thirteen of 25 paired sites of significant covariance were within the same protein, whereas 12 of 25 fell in different proteins. For the majority of pairs of covariant sites, one or both sites fell at a reported HLA-associated viral polymorphism site, within a known epitope, or at a common site of reversion from the source. Four of the 25 paired sites fell at an HLA-associated site in Table 1. In particular, two HLA-A*03–associated sites at positions 1087 and 1088 in NS3 fell within a confirmed HLA-A*03 epitope in which variation at both sites is required to restore replicative efficacy (K.F., unpublished data, 2010); this reflects the potential compensatory nature of these covariations.

ted site in Table 1. In particular, two HLA-A*03–associated sites at positions 1087 and 1088 in NS3 fell within a confirmed HLA-A*03 epitope in which variation at both sites is required to restore replicative efficacy (K.F., unpublished data, 2010); this reflects the potential compensatory nature of these covariations. Fig. 2 shows a linear trend for many covarying sites suggesting that many fell in close proximity to one another but not necessarily in the same protein. Interestingly, clusters of covarying sites appeared to connect sites across the genome and particularly other proteins with NS5A. One group contained sites in only one protein (NS3 sites 1644F/Y, 1647A/T, and 1656A/T), whereas another group contained sites in three proteins (NS2 908R/K, NS3 1173S/L, and NS5A 2279R/K). These links may further restrict the ability of the virus to adapt or revert quickly and suggest critical interactions between the HCV proteins. We extended this analysis to assess covariation at amino acid and synonymous sites to identify potential constraints on codon usage (and subsequent amino acid changes) and identified four amino acid sites associated with synonymous changes in other proteins.

y and suggest critical interactions between the HCV proteins. We extended this analysis to assess covariation at amino acid and synonymous sites to identify potential constraints on codon usage (and subsequent amino acid changes) and identified four amino acid sites associated with synonymous changes in other proteins. Fig. 2 Covarying sites (P < 0.001) in the HCV genome represented as coordinates. Open diamonds indicate that one or both sites fall within an epitope or at an association site, and dark diamonds indicate that the sites do not fall within either. Many covariant sites fall in close proximity to one another in the genome (illustrated by the linear trend); however, there are groupings that suggest strong covariation between residues within NS5A and residues within other proteins. Sequence coverage was not found to be a function of covariant site identification.

r. Many covariant sites fall in close proximity to one another in the genome (illustrated by the linear trend); however, there are groupings that suggest strong covariation between residues within NS5A and residues within other proteins. Sequence coverage was not found to be a function of covariant site identification. Relevance of Viral Adaptations in the New Hosts and Preexisting Ones in the Source in Infection Outcomes Although the host immune pressure is one of several forces shaping HCV diversity, it is likely that only a small number of selected viral adaptations in the sequence may affect infection outcomes. In this cohort, HLA-A*03 was shown to be protective,8 and we selected chronic HCV–infected individuals with HLA-A*03 for this study to identify viral adaptations in these individuals that may have affected their infection outcomes. Three viral polymorphisms were associated with HLA-A*03 in this study (Table 1). Two of the associations were in NS3 at positions 1087 and 1088 within a predicted epitope for HLA-A*03. As mentioned previously, this epitope was subsequently shown to be a true in vivo target of the immune response (NS3 1080 TVYHGAGTK; K.F., unpublished data, 2010; Fig. 3A) and reflected a drop in the SYFPEITHI-predicted binding score from 34 for the wild type to 21 for the putative escape peptide. Another HLA-A*03–associated viral polymorphism at position 2518 in NS5B was within the previously characterized genotype 1a epitope SLTPPHSAK (Fig. 3B). Half of the HLA-A*03 individuals had a polymorphism at these sites in both regions. These results suggest that these two viral epitopes are important immune targets and that escape within the targets may influence the outcome.

2518 in NS5B was within the previously characterized genotype 1a epitope SLTPPHSAK (Fig. 3B). Half of the HLA-A*03 individuals had a polymorphism at these sites in both regions. These results suggest that these two viral epitopes are important immune targets and that escape within the targets may influence the outcome. Fig. 3 HLA-A*03–associated viral polymorphisms at (A) positions 1087 and 1088 in NS3 and (B) position 2518 in NS5B. Sequences in regions of interest (from Table 1) are displayed for HLA-A*03+ and HLA-A*03− subjects. The sequence identity with the source sequence is identified by a dot. Amino acid mixtures at a site are separated by a forward slash. The number of individuals with a particular sequence is shown in the count column. The lysine (K) to arginine (R) substitution at 2518 (8 of 15 HLA-A*03+ subjects versus 4 of 47 HLA-A*03− subjects) resulted in a change in the SYFPEITHI-predicted binding score from 27 to 21. Only one HLA-A*03 individual with chronic infection did not have a polymorphism at the 1087 or 1088 site in NS3 or at the 2518 site in NS5B. Further analysis of the quasispecies at the NS3 1087 and 1088 sites in HLA-A*03+ and HLA-A*03− subjects was performed with ultradeep sequencing. Table 2 reveals the lack of a source sequence at amino acid position NS3 1088 in the HLA-A*03 subject with complete amino acid replacement but 100% retention of the source sequence in the HLA-A*03− subject. The two subjects had the same amino acid at position 1087 (unadapted), but codon usage was different between the two.

e 2 reveals the lack of a source sequence at amino acid position NS3 1088 in the HLA-A*03 subject with complete amino acid replacement but 100% retention of the source sequence in the HLA-A*03− subject. The two subjects had the same amino acid at position 1087 (unadapted), but codon usage was different between the two. Table 2 Ultradeep Sequencing Reveals a Lack of a Source Sequence at Putative Viral Adaptation Sites (NS3 1087 and 1088) in a Subject With HLA-A3 but 100% Maintenance of the Source Sequence in an HLA-A3− Subject Previous studies have found other HLA alleles to be associated with chronic infection that are specific to this cohort, such as alleles HLA-A*01, HLA-B*08, and HLA-C*078 (these alleles most likely correspond to a single MHC haplotype). It has been suggested that the association between infection outcomes and specific HLA alleles may be due to preexisting viral adaptations in the incoming virus that may facilitate the evasion of host immune responses with the corresponding HLA types.27 Here we tested this hypothesis by examining the source sequence for escape mutations within known epitopes as well as putative viral adaptations identified in our previous genetic study of chronic HCV infection.18,19

ming virus that may facilitate the evasion of host immune responses with the corresponding HLA types.27 Here we tested this hypothesis by examining the source sequence for escape mutations within known epitopes as well as putative viral adaptations identified in our previous genetic study of chronic HCV infection.18,19 Initially, we examined the immunodominant epitope for HLA-B*08 in NS3 (1395 HSKKKCDEL) and the protective HLA-B*27 epitope in NS5B (2841 ARMILMTHF). The region in the source containing the HLA-B*27 epitope in NS5B had the unmutated form. However, the HLA-B*08 epitope in NS3 in the source sequence had a preexisting viral adaptation in the epitope (arginine at position 3), which subsequently reverted in 8 of 11 subjects without HLA-B*08 and was retained in 5 of 8 subjects who expressed HLA-B*08. Although the numbers in the two groups were not significantly different (P = 0.18), they supported other studies showing reversion from an arginine to lysine at position 3 in this epitope when there was no immune pressure; this is suggestive of a fitness cost.15 This HLA-B*08 epitope was previously studied in this cohort with similar results.15,22 The fitness cost of this substitution was further supported by the results from the ultradeep sequencing of two HLA-B*08− subjects in this region, who showed complete reversion from the source escape mutation at position 3 of the epitope (Table 3). Table 3 Ultradeep Sequencing Reveals a Lack of a Source Sequence at Position 1397 in the Immunodominant HLA-B*08 Epitope in NS3 (HSKKKCDEL) in Two HLA-B*08− Subjects

Initially, we examined the immunodominant epitope for HLA-B*08 in NS3 (1395 HSKKKCDEL) and the protective HLA-B*27 epitope in NS5B (2841 ARMILMTHF). The region in the source containing the HLA-B*27 epitope in NS5B had the unmutated form. However, the HLA-B*08 epitope in NS3 in the source sequence had a preexisting viral adaptation in the epitope (arginine at position 3), which subsequently reverted in 8 of 11 subjects without HLA-B*08 and was retained in 5 of 8 subjects who expressed HLA-B*08. Although the numbers in the two groups were not significantly different (P = 0.18), they supported other studies showing reversion from an arginine to lysine at position 3 in this epitope when there was no immune pressure; this is suggestive of a fitness cost.15 This HLA-B*08 epitope was previously studied in this cohort with similar results.15,22 The fitness cost of this substitution was further supported by the results from the ultradeep sequencing of two HLA-B*08− subjects in this region, who showed complete reversion from the source escape mutation at position 3 of the epitope (Table 3). Table 3 Ultradeep Sequencing Reveals a Lack of a Source Sequence at Position 1397 in the Immunodominant HLA-B*08 Epitope in NS3 (HSKKKCDEL) in Two HLA-B*08− Subjects Viral adaptation in the source sequence at a site in the HLA-B*08 immunodominant epitope likely to incur a fitness cost suggests that the source may have been an HLA-B*08+ individual. We suggest that this could potentially reduce the ability of hosts with HLA-B*08 to control the virus via the reduction of good immune targets, and this reflects the association of this allele with poor outcomes in this cohort. Additional association sites with HLA-B*08+ individuals found in this study may represent alternative targets for HLA-B*08 along the HCV genome. Furthermore, Table 1 and Supporting Information Fig. 4 list HLA-associated viral polymorphisms that have an OR less than 1 and represent the maintenance of the consensus sequence (which for most sites in Table 1 is the same as the source) for the specific HLA type; this possibly reflects that the source sequence is pre-adapted at these sites. Interestingly, this occurs for alleles within the MHC haplotypes HLA-A*01, HLA-B*08, and HLA-C*07, which are associated with poor outcomes.

ensus sequence (which for most sites in Table 1 is the same as the source) for the specific HLA type; this possibly reflects that the source sequence is pre-adapted at these sites. Interestingly, this occurs for alleles within the MHC haplotypes HLA-A*01, HLA-B*08, and HLA-C*07, which are associated with poor outcomes. Other Selective Pressures Likely to Affect HCV Evolution In order to determine how other host immune pressures may affect HCV evolution, we assessed possible associations between HCV polymorphisms in this cohort and an SNP that tags the IL-28B gene encoding interferon-λ3 and recently has been associated with infection outcome.2 We found one significant association between homozygosity for the major allele of rs12979860 (associated with good outcome) and variation at position 849 in NS2 (P = 0.006). We also tested for additional effects of the IL-28B SNP on the HLA-associated polymorphisms. After adjustments for HLA, among the positions identified in Table 1, IL-28B was associated with a polymorphism (P = 0.036) only at position 2609 of NS5B, which harbors the strong HLA-B*35/HLA-C*04 association. The significance of the HLA-B*35 association with nonconsensus after adjustments for the IL-28B SNP is P = 0.00004, whereas for HLA-B*35 alone, the P value is 0.0001. There was no significant interaction between the effects of HLA-B*35 and IL-28B (P > 0.9), and this suggests that they act independently. Further studies examining the association between variations that tag IL-28B and HCV evolution are warranted and should be performed on larger cohorts including subjects with different treatment and infection outcomes.

between the effects of HLA-B*35 and IL-28B (P > 0.9), and this suggests that they act independently. Further studies examining the association between variations that tag IL-28B and HCV evolution are warranted and should be performed on larger cohorts including subjects with different treatment and infection outcomes. Discussion Here we illustrate that the incoming viral sequence, host immune pressure, and covariation play important roles in shaping HCV viral diversity. Specifically, we identified 29 significant HLA-associated viral polymorphisms (P ≤ 0.01; 23 sites) within the cohort that likely reflect viral adaptations. Some of these sites fall within published and/or predicted T cell epitopes. The use of a sliding-window analysis accounting for more than a single escape variant within a T cell target identified a small number of additional potential regions under T cell pressure, and this supported other studies showing that escape can require the accumulation of escape mutations28 or that viral escape sites are often mutually exclusive because of the fitness cost.15,18

single escape variant within a T cell target identified a small number of additional potential regions under T cell pressure, and this supported other studies showing that escape can require the accumulation of escape mutations28 or that viral escape sites are often mutually exclusive because of the fitness cost.15,18 The number of significant HLA-associated viral polymorphism sites identified in this study is only a small proportion of the sites (23/184) across the HCV genome showing variation in the cohort; this is possibly due to the relatively small sample size or suggests that the host immune pressure has a targeted influence on HCV diversity. This would be expected because the immune system sees the viral polyprotein as a set of peptides, and only a small number of these peptides are likely to be presented to the immune system. Furthermore, the lack of significant overlap with previously reported adaptations for genotypes 1a and 3a likely reflects the constraint of the incoming virus and differential viral adaptation pathways on genotype 1b versus other circulating genotypes due to the genetic distance between these strains. It should be noted that although we did not show HLA class II–associated viral polymorphisms, it is likely that, in addition to what we observed for HLA class I alleles, some of the variations correspond to the expression of specific HLA class II alleles.

ing genotypes due to the genetic distance between these strains. It should be noted that although we did not show HLA class II–associated viral polymorphisms, it is likely that, in addition to what we observed for HLA class I alleles, some of the variations correspond to the expression of specific HLA class II alleles. To appreciate the extent to which both positive and purifying selections influence HCV diversity, we examined the number of synonymous and nonsynonymous changes across the genome for this single-source cohort. An abundance of synonymous changes indicated purifying selection that would to some extent limit the plasticity of HCV. Covariations that become fixed across the HCV genome may also restrict the ability of HCV to adapt to the host's immune response and revert when it enters a new non–HLA-matched host. We examined the genome for covarying sites and showed that although covariation did occur locally within proteins, there were also a number of sites that were linked to sites more distant in the genome. Furthermore, several of these sites were putative viral adaptation sites.

when it enters a new non–HLA-matched host. We examined the genome for covarying sites and showed that although covariation did occur locally within proteins, there were also a number of sites that were linked to sites more distant in the genome. Furthermore, several of these sites were putative viral adaptation sites. Access to the source viral sequence from this single-source cohort allowed the identification of preexisting escape mutations across the genome. A known escape mutation at position 3 of the immunodominant HLA-B*08 NS3 epitope was found in the source sequence. This mutation was for the most part retained in HLA-B*08 subjects but had reverted in most HLA-B*08− subjects. Furthermore, deep sequencing revealed no traces of the escape mutant in two B*08− individuals, and this supports the fitness cost that may be incurred by the escape mutation. Importantly, existing adaptation in the incoming virus may affect infection outcomes in individuals expressing the appropriate HLA type. The pre-adaptation of the source sequence to HLA-B*08 may account for the observed lack of protection of HLA-B*08 in this cohort. The single-source cohort studied here has provided us an opportunity to obtain a better understanding of viral diversity and the ways in which different forces can shape viral diversity at the population level. The authors thank the patients and clinical staff who participated in this study and their colleagues at the Centre for Clinical Immunology and Biomedical Statistics of Murdoch University. Abbreviations HCVhepatitis C virus HLAhuman leukocyte antigen

The single-source cohort studied here has provided us an opportunity to obtain a better understanding of viral diversity and the ways in which different forces can shape viral diversity at the population level. The authors thank the patients and clinical staff who participated in this study and their colleagues at the Centre for Clinical Immunology and Biomedical Statistics of Murdoch University. Abbreviations HCVhepatitis C virus HLAhuman leukocyte antigen IL-28Binterleukin-28B MHCmajor histocompatibility complex NSnonstructural ORodds ratio SNPsingle-nucleotide polymorphism Supplementary material

Nonalcoholic fatty liver disease (NAFLD) is a highly prevalent condition estimated to afflict approximately one-third of the adult population in Western countries.1 NAFLD comprises a spectrum of hepatic disorders extending from hepatic steatosis to cirrhosis.2, 3 The hallmark of hepatic steatosis is the presence of triglycerides (TGs) stored as lipid droplets in the cytoplasm of hepatocytes. It has been estimated that 10% to 20% of those with steatosis develop inflammation (steatohepatitis), and the disease can progress to cirrhosis and perhaps to hepatocellular carcinoma.3 It is anticipated that NAFLD will soon overtake hepatitis C as the most common indication for liver transplantation. A major factor contributing to the increase in hepatic triglyceride content (HTGC) in the general population is the high prevalence of obesity and insulin resistance. Hepatic TG levels typically are low in lean, nondiabetic individuals, whereas approximately 50% of those with a body mass index (BMI) > 30 kg/m21 and approximately 75% of those with adult-onset type 2 diabetes have steatosis.3 Thus, although obesity and insulin resistance are important susceptibility factors for NAFLD, not all obese or insulin-resistant individuals develop hepatic steatosis. The mechanisms underlying the variation in susceptibility to NAFLD and the causal nature of the relationships between obesity, hepatic TG, and insulin resistance are poorly understood.

in resistance are important susceptibility factors for NAFLD, not all obese or insulin-resistant individuals develop hepatic steatosis. The mechanisms underlying the variation in susceptibility to NAFLD and the causal nature of the relationships between obesity, hepatic TG, and insulin resistance are poorly understood. Genetic association provides a powerful tool for dissecting the mechanistic relationships between variables that are correlated in complex diseases. In 2008, a single-nucleotide polymorphism (SNP) in the patatin-like phospholipase domain containing 3 protein 3 encoding gene (PNPLA3) was identified that was strongly associated with HTGC in Caucasians, Hispanics, and African Americans.4 The variant (rs738409) results in the substitution of methionine for isoleucine at residue 148 of PNPLA3. The PNPLA3-I148M variant is most common in Hispanics [minor allele frequency (MAF) = 0.48], the group with the highest prevalence of hepatic steatosis (45%), and is least common in African Americans (MAF = 0.14), who have the lowest incidence of steatosis (24%); the frequency in Caucasians is intermediate (MAF = 0.23). Hispanics who are homozygous for the variant have an approximately 2-fold increase in HTGC, whereas African American and European American homozygotes have 60% and 30% increases in median HTGC, respectively. According to the homeostatic model assessment of insulin resistance (HOMA-IR), despite the clear association between the variant and HTGC, PNPLA3-I148M is not associated with insulin resistance in either the Dallas Heart Study or the Atherosclerosis Risk in Communities (ARIC) study.4 The variant was independently identified in a genome-wide association study of serum liver enzyme levels.5 Subsequent studies in other populations have confirmed that PNPLA3-I148M is associated with increased liver fat content and elevated plasma levels of aminotransferases but not with BMI, insulin sensitivity, or plasma TG levels.4, 6-12

endently identified in a genome-wide association study of serum liver enzyme levels.5 Subsequent studies in other populations have confirmed that PNPLA3-I148M is associated with increased liver fat content and elevated plasma levels of aminotransferases but not with BMI, insulin sensitivity, or plasma TG levels.4, 6-12 Recently, two SNPs in the promoter region of the gene encoding apolipoprotein C3 [APOC3; rs2854117 (−482 C > T) and rs2854116 (−455 T > C)] were reported to be associated with an approximately 3-fold increase in median HTGC values and with insulin resistance.13 The two variants, which had previously been found to be associated with plasma TG levels in some studies,14-16 are located in a putative insulin response element17 located 5′ to exon 1 of APOC3. In vitro promoter studies have suggested that insulin binding to this site inhibits APOC3 transcription. The two variants prevent insulin binding and thus increase the levels of APOC3 messenger RNA and protein.17 APOC3 is transported on circulating lipoproteins and limits clearance of TG-rich particles.18 Petersen et al.13 proposed that the sequence variants lead to increased uptake of chylomicron remnants by the liver, and this results in NAFLD and hepatic insulin resistance.

se the levels of APOC3 messenger RNA and protein.17 APOC3 is transported on circulating lipoproteins and limits clearance of TG-rich particles.18 Petersen et al.13 proposed that the sequence variants lead to increased uptake of chylomicron remnants by the liver, and this results in NAFLD and hepatic insulin resistance. Here we examined the relationship between APOC3 genotypes, HTGC, insulin resistance, and fasting TG levels in the Dallas Heart Study, a probability sample of Dallas County19 in which HTGC was measured noninvasively with proton magnetic resonance spectroscopy (1H-MRS).1, 4, 20 We also tested for the combined effect of the PNPLA3 and APOC3 variants. To confirm our findings, we analyzed the relationship between the two APOC3 SNPs, fasting TG levels, and insulin resistance in the ARIC study.22 No association was found between either of the SNPs and indices of insulin resistance in the Dallas Heart Study or in the ARIC study.

effect of the PNPLA3 and APOC3 variants. To confirm our findings, we analyzed the relationship between the two APOC3 SNPs, fasting TG levels, and insulin resistance in the ARIC study.22 No association was found between either of the SNPs and indices of insulin resistance in the Dallas Heart Study or in the ARIC study. Patients and Methods Human Subjects The Dallas Heart Study is a multiethnic population-based probability sample of Dallas County residents (African Americans self-identified as black, 52%; individuals of mixed European descent self-identified as white, 29%; Hispanics self-identified as Hispanic, 17%; and other ethnicities, 2%). The sampling design and the study protocol have been described previously.19 The study was approved by the institutional review board of the University of Texas Southwestern Medical Center at Dallas, and all participants provided written, informed consent. Fasting venous blood samples were obtained from 3551 individuals, 3071 of whom completed a clinic visit. Alcohol consumption (g/day) was determined from responses to previously validated questions (Institute for Survey Research, Temple University, Philadelphia, PA, 1996). HOMA-IR was calculated from the fasting plasma glucose and insulin values, which were measured as described.21

s, 3071 of whom completed a clinic visit. Alcohol consumption (g/day) was determined from responses to previously validated questions (Institute for Survey Research, Temple University, Philadelphia, PA, 1996). HOMA-IR was calculated from the fasting plasma glucose and insulin values, which were measured as described.21 HTGC was measured with 1H-MRS.20 Localized spectra of the liver were obtained with a 1.5T Gyroscan Intera MR system (Philips Medical Systems, the Netherlands), and HTGC was calculated as described.1 Of the individuals who completed a clinic visit, 2349 underwent 1H-MRS. Some study participants failed to obtain an MRS study because of claustrophobia (n = 191), a medical contraindication (n = 49), equipment failure (n = 19), refusal (n = 74), or scheduling conflicts (n = 289). In addition, 58 extremely obese individuals (>145 kg) were excluded because of the weight limitations of the table.1 From the 2349 1H-MRS measurements, data of sufficient quality to determine HTGC were obtained for 2239 individuals, who included 1104 African Americans, 734 European Americans, and 401 Hispanics. Hepatic steatosis was operationally defined as a liver fat content of 5.5% or greater, which corresponds to the 95th percentile of the liver fat content in lean, healthy individuals from the Dallas Heart Study.20 The definition is based on population prevalence and is not coupled to a future risk of steatohepatitis and cirrhosis.

atosis was operationally defined as a liver fat content of 5.5% or greater, which corresponds to the 95th percentile of the liver fat content in lean, healthy individuals from the Dallas Heart Study.20 The definition is based on population prevalence and is not coupled to a future risk of steatohepatitis and cirrhosis. Genotyping Genomic DNA was extracted from circulating leukocytes. Genotypes for the rs2854117 (−482 C > T) and rs2854116 (−455 T > C) polymorphisms were determined with the TaqMan AD assay (Applied Biosystems). Oligonucleotides used for genotyping are shown in Supporting Table 1. A total of 3477 individuals from the Dallas Heart Study were successfully genotyped for the APOC3 rs2854117 variant; they included 2952 participants who completed a clinic visit and 2198 subjects with HTGC measurements (1081 African Americans, 726 European Americans, and 391 Hispanics). A total of 3399 subjects were successfully genotyped for APOC3 rs2854116, and these included 2880 participants who completed a clinic visit and 2150 individuals with hepatic TG measurements (1096 African Americans, 733 European Americans, and 401 Hispanics). Both genotypes were determined for 3336 of these subjects, who included 2113 with HTGC measurements. The two APOC3 SNPs were also genotyped in the ARIC study with the TaqMan AD assay (Applied Biosystems). Inferred Ancestry To account for a possible substructure in the self-identified ethnic groups, we inferred ancestry in the Dallas Heart Study participants with STRUCTURE23 under a linkage model with 2270 ancestry-informative SNPs24 as described.4

The two APOC3 SNPs were also genotyped in the ARIC study with the TaqMan AD assay (Applied Biosystems). Inferred Ancestry To account for a possible substructure in the self-identified ethnic groups, we inferred ancestry in the Dallas Heart Study participants with STRUCTURE23 under a linkage model with 2270 ancestry-informative SNPs24 as described.4 Statistical Analysis Associations between SNP genotypes and HTGC, HOMA-IR, fasting TG, alanine aminotransferase (ALT), and aspartate aminotransferase (AST) were tested with linear regression models with age, gender, ethnicity, and BMI as covariates. The relationships were also assessed in each ethnic group separately. To avoid confounding by population stratification, we repeated the analysis with the inferred ancestry score as a covariate instead of self-reported ethnicity. We applied a power transformation (λ = 1/4) to HTGC, a natural log transformation to plasma insulin levels, HOMA-IR, and BMI, and a log-log transformation to TGs, ALT, and AST before the analysis to render the error distributions approximately normal. Diabetic individuals were excluded from the analysis. Alcohol consumption was included as a covariate when we were testing for an association with HTGC. Genotypes were coded as 0, 1, or 2, and the association was tested under the assumption of an additive model. Deviations from Hardy-Weinberg proportions were tested with the exact test for Hardy-Weinberg equilibrium. All statistical analyses were performed with the R statistical language (R Development Core Team, 2008).

enotypes were coded as 0, 1, or 2, and the association was tested under the assumption of an additive model. Deviations from Hardy-Weinberg proportions were tested with the exact test for Hardy-Weinberg equilibrium. All statistical analyses were performed with the R statistical language (R Development Core Team, 2008). Web Resources The free software R can be downloaded from the Comprehensive R Archive Network (http://www.cran.r-project.org). Results Insulin Resistance and HTGC The liver TG content was strongly correlated with HOMA-IR in the three major ethnic groups in the Dallas Heart Study (Fig. 1), as indicated by Spearman's rank correlation coefficient (ρ), a measure that is invariant to monotonic transformations of the data. Both HTGC and HOMA-IR were strongly associated with BMI in this sample, as previously reported.1 After adjustments for age, gender, ethnicity, and BMI, HTGC could account for 10% of the variation in HOMA-IR in the combined sample. BMI could account for approximately 15% of the variation in HOMA-IR after adjustments for age, gender, ethnicity, and HTGC (data not shown). Thus, neither HTGC nor BMI could explain the major fraction of variation in HOMA-IR. Fig. 1 Relationship between HOMA-IR and HTGC with stratification by ethnicity in the Dallas Heart Study. The solid lines denote least squares regression lines.

Results Insulin Resistance and HTGC The liver TG content was strongly correlated with HOMA-IR in the three major ethnic groups in the Dallas Heart Study (Fig. 1), as indicated by Spearman's rank correlation coefficient (ρ), a measure that is invariant to monotonic transformations of the data. Both HTGC and HOMA-IR were strongly associated with BMI in this sample, as previously reported.1 After adjustments for age, gender, ethnicity, and BMI, HTGC could account for 10% of the variation in HOMA-IR in the combined sample. BMI could account for approximately 15% of the variation in HOMA-IR after adjustments for age, gender, ethnicity, and HTGC (data not shown). Thus, neither HTGC nor BMI could explain the major fraction of variation in HOMA-IR. Fig. 1 Relationship between HOMA-IR and HTGC with stratification by ethnicity in the Dallas Heart Study. The solid lines denote least squares regression lines. Allele Frequencies The frequencies of the APOC3 polymorphisms differed markedly among ethnic groups. The variant alleles [rs2854117 (T) and rs2854116 (C)] were most common in African Americans (66.4% and 71.2%, respectively) and less common in Hispanics (32.7% and 38.9%, respectively) and Europeans (25.5% and 36.6%, respectively). These estimates are consistent with allele frequencies reported previously.15 Within each group, the genotype distributions were in Hardy-Weinberg equilibrium (Supporting Tables 2 and 3). The SNPs belonged to a larger linkage disequilibrium block spanning the APOA5/APOA4/APOC3/APOA1 gene region (Supporting Fig. 1) and were in strong linkage disequilibrium (D′ = 0.98 and R2 = 0.75 in African Americans, D′ = 0.93 and R2 = 0.53 in European Americans, and D′ = 0.98 and R2 = 0.73 in Hispanics), as previously described.14

nged to a larger linkage disequilibrium block spanning the APOA5/APOA4/APOC3/APOA1 gene region (Supporting Fig. 1) and were in strong linkage disequilibrium (D′ = 0.98 and R2 = 0.75 in African Americans, D′ = 0.93 and R2 = 0.53 in European Americans, and D′ = 0.98 and R2 = 0.73 in Hispanics), as previously described.14 Genetic Association Between APOC3 and HTGC, HOMA-IR, and Plasma TG First, we examined the relationship between each APOC3 variant, HTGC, and HOMA-IR in the Dallas Heart Study. The APOC3 rs2854117 (C > T) variant was not associated with HTGC in the overall sample or in any of the three ethnic groups (Fig. 2A). In contrast to the findings of Petersen et al.,13 the rs2854116 variant allele (C) was associated with a slight but significant reduction in HTGC (median HTGC: 4.1% in C/C genotype carriers versus 3.2% in T/T genotype carriers, P = 0.04; Fig. 2B). No association was found between either variant and the fasting levels of plasma glucose or insulin (Supporting Tables 2-5) or HOMA-IR except in Hispanics where the variant was marginally associated with a lower HOMA-IR (Fig. 3). Fig. 2 Median HTGC values in the Dallas Heart Study subjects stratified by ethnicity and genotype: (A) APOC3 rs2854117 and (B) APOC3 rs2854116. HTGC values were power-transformed (λ = 1/4) before the analysis, and P values were determined using a linear regression model with adjustment for age, gender, BMI, and alcohol consumption. Diabetics were excluded from the analysis.

ects stratified by ethnicity and genotype: (A) APOC3 rs2854117 and (B) APOC3 rs2854116. HTGC values were power-transformed (λ = 1/4) before the analysis, and P values were determined using a linear regression model with adjustment for age, gender, BMI, and alcohol consumption. Diabetics were excluded from the analysis. Fig. 3 Median HOMA-IR values in the Dallas Heart Study participants stratified by ethnicity and genotype: (A) APOC3 rs2854117 and (B) APOC3 rs2854116. HOMA-IR values were log-transformed (natural logarithm), and P values were determined using a linear regression model with adjustment for age, gender, and BMI. Diabetics were excluded from the analysis. Neither APOC3 sequence variant showed consistent relationships with plasma TG concentrations among the three ethnic groups. The rs2854117 (T) allele was associated with a modest increase in fasting TG concentrations in African Americans (median TG level: 78 mg/dL in C/C versus 84 mg/dL in T/T, P = 0.014; Supporting Table 2) but not in the other ethnic groups. rs2854116 was not associated with fasting plasma TG levels (Supporting Tables 3 and 5). No associations were found between either APOC3 variant and the plasma levels of ALT or AST (Supporting Tables 2-5).

G level: 78 mg/dL in C/C versus 84 mg/dL in T/T, P = 0.014; Supporting Table 2) but not in the other ethnic groups. rs2854116 was not associated with fasting plasma TG levels (Supporting Tables 3 and 5). No associations were found between either APOC3 variant and the plasma levels of ALT or AST (Supporting Tables 2-5). In some previous studies, the association between these two APOC3 variants and plasma TG levels was apparent only in specific subgroups, such as healthy (nondiabetic) lean individuals13, 16 and nonsmokers.15 To investigate whether a relationship between APOC3 variants and HTGC was apparent only in a subset of individuals or was obscured by smoking, we confined the analysis to the subset of Dallas Heart Study participants who were nondiabetic, had a BMI less than 25 kg/m2, and consumed less than 30 g of alcohol per day (n = 468), and we tested for an interaction with smoking status. No relationship between APOC3 variants and HTGC or insulin resistance emerged in this subgroup (data not shown) in either men (n = 210) or women (n = 258). The results were not affected by smoking status. We next investigated the possibility of a combined action of the APOC3 polymorphisms through a comparison of the wild-type homozygotes (rs2854117 C/C and rs2854116 T/T) to carriers of one or more variant alleles [rs2854117 (T) and rs2854116 (C)], as was done previously.13 No significant relationships were observed between the two groups (carriers versus noncarriers for either variant) in HTGC, HOMA-IR, or fasting plasma TG levels (Fig. 4 and Tables 1 and 2).

117 C/C and rs2854116 T/T) to carriers of one or more variant alleles [rs2854117 (T) and rs2854116 (C)], as was done previously.13 No significant relationships were observed between the two groups (carriers versus noncarriers for either variant) in HTGC, HOMA-IR, or fasting plasma TG levels (Fig. 4 and Tables 1 and 2). Fig. 4 Median values of (A) HTGC, (B) HOMA-IR, and (C) plasma TG levels in noncarriers [individuals homozygous for the APOC3 reference allele (rs2854117 C-482 and rs2854116 T-455)] and carriers (individuals with 482T, 455C, or both alleles) in the Dallas Heart Study. Table 1 Clinical Characteristics of APOC3 Wild-Type Homozygotes (rs2854117 C/C and rs2854116 T/T) and Carriers of One or More Variant Alleles (rs2854117 T and rs2854116 C) in the Dallas Heart Study Characteristic Wild-Type Homozygotes Variant-Allele Carriers P Value n 586 1911 Female/male 301/285 1071/840 Age (years) 43.9 ± 9.9 43.8 ± 9.7 0.069 BMI (kg/m2) 29.3 ± 6.3 30.3 ± 7.4 0.986 Insulin (mIU/L) 11.7 (6.8-18.8) 11.8 (7.2-19.5) 0.3 Glucose (mg/dL) 92 (85-99) 91 (84-98) 0.522 HOMA-IR (U) 2.6 (1.5-4.3) 2.7 (1.6-4.5) 0.274 TGs (mg/dL) 104 (72-156.5) 89 (65-131) 0.398 HTGC (%) 3.8 (2.1-7.7) 3.3 (2.0-5.6) 0.175 AST (U/L) 21 (18-25) 21 (18-26) 0.376 ALT (U/L) 19 (15-28) 19 (14-27) 0.227 Means and standard deviations are presented for the age and BMI, and medians and interquartile ranges are presented for all other characteristics. P values were calculated with a linear regression model as described in the Patients and Methods section.

) 21 (18-25) 21 (18-26) 0.376 ALT (U/L) 19 (15-28) 19 (14-27) 0.227 Means and standard deviations are presented for the age and BMI, and medians and interquartile ranges are presented for all other characteristics. P values were calculated with a linear regression model as described in the Patients and Methods section. Table 2 Clinical Characteristics of Homozygotes for APOC3 Wild-Type Alleles (rs2854117 C/C and rs2854116 T/T) and Carriers of Variant Alleles (rs2854117 T and rs2854116 C) in the Dallas Heart Study African Americans European Americans Hispanics

) 21 (18-25) 21 (18-26) 0.376 ALT (U/L) 19 (15-28) 19 (14-27) 0.227 Means and standard deviations are presented for the age and BMI, and medians and interquartile ranges are presented for all other characteristics. P values were calculated with a linear regression model as described in the Patients and Methods section. Table 2 Clinical Characteristics of Homozygotes for APOC3 Wild-Type Alleles (rs2854117 C/C and rs2854116 T/T) and Carriers of Variant Alleles (rs2854117 T and rs2854116 C) in the Dallas Heart Study African Americans European Americans Hispanics Characteristic Noncarriers Carriers P Value Noncarriers Carriers P Value Noncarriers Carriers P Value n 102 1126 320 523 164 262 Female/male 55/47 647/479 162/158 263/260 84/80 161/101 Age (years) 45.5 ± 10.1 44.5 ± 9.7 0.332 45.1 ± 9.9 44.7 ± 9.7 0.575 40.7 ± 8.9 39 ± 8.3 0.039 BMI (kg/m2) 30.9 ± 7.3 31.2 ± 8 0.974 28.8 ± 6.3 28.6 ± 6.2 0.705 29.5 ± 5.2 30 ± 6.4 0.458 Insulin (mIU/L) 14.2 (8.7-22.0) 12.9 (7.7-20.7) 0.071 9.9 (6.0-16.4) 10.1 (6.0-16.6) 0.514 14.1 (8.5-20.5) 13.0 (7.6-19.8) 0.235 Glucose (mg/dL) 91 (84-99) 90 (83-97) 0.206 91 (84-97) 91 (83-97.5) 0.769 96 (87-101) 93 (87-100) 0.714 HOMA-IR (U) 3.1 (1.9-5.2) 2.9 (1.7-4.8) 0.059 2.3 (1.3-3.8) 2.2 (1.3-3.9) 0.534 3.3 (1.9-4.9) 2.9 (1.7-4.8) 0.244 TG (mg/dL) 78 (54.5-124) 80 (60-113) 0.312 105.5 (75-164) 105 (73-155.5) 0.975 117 (83-176) 111 (78-173) 0.446 HTGC (%) 3.6 (2.3-5.2) 3.0 (1.9-4.8) 0.106 3.5 (1.9-6.9) 3.5 (2.1-6.8) 0.771 4.4 (2.8-11.8) 4.3 (2.5-9.0) 0.34 AST (U/L) 20 (17-23) 21 (17-26) 0.108 21 (18-25) 21 (18-25) 0.827 21 (17-29) 21 (17-26) 0.927 ALT (U/L) 16 (12-23) 18 (13-26) 0.023 20 (15-27) 20 (16-27) 0.834 21 (15-33) 20 (15-32) 0.922 The subjects are stratified by ethnicity. Means and standard deviations are presented for the age and BMI, and medians and interquartile ranges are presented for all other traits. P values were calculated with a linear regression model as described in the Patients and Methods section.

0.834 21 (15-33) 20 (15-32) 0.922 The subjects are stratified by ethnicity. Means and standard deviations are presented for the age and BMI, and medians and interquartile ranges are presented for all other traits. P values were calculated with a linear regression model as described in the Patients and Methods section. Finally, to examine the combined effects of the APOC3 SNPs and the PNPLA3 rs738409 (C > G) polymorphism on HTGC, we fit a linear regression model including all three variants as predictors. The PNPLA3 rs738409 risk allele (G) remained highly significant, even when the two APOC3 variants were included in the model (P = 4.06 × 10−13). On the contrary, none of the APOC3 SNPs were significantly related to HTGC with PNPLA3 rs738409 in the model. In addition, we analyzed the relationship between APOC3 variants and liver fat content in PNPLA3-148M carriers and in PNPLA3-148I homozygotes. No significant associations with HTGC were found in either group (data not shown). In the Dallas Heart Study, the PNPLA3 rs738409 variant explained a variable proportion of HTGC in the three different ethnic groups, and this reflected the differences in allele frequencies between African Americans (MAF = 14%), European Americans (MAF = 23%), and Hispanics (MAF = 48%). The PNPLA3 polymorphism explains 12% of the variation in HTGC in Hispanics, 4% in African Americans, and only 2% in European Americans. Because African Americans comprised approximately 52% of the study participants, the PNPLA3 polymorphism explained 4% of the variation in HTGC in the entire Dallas Heart Study sample after adjustments for ethnicity, age, sex, BMI, and alcohol consumption. APOC3 variants accounted for no meaningful variation in HTGC in any of the ethnic groups.

comprised approximately 52% of the study participants, the PNPLA3 polymorphism explained 4% of the variation in HTGC in the entire Dallas Heart Study sample after adjustments for ethnicity, age, sex, BMI, and alcohol consumption. APOC3 variants accounted for no meaningful variation in HTGC in any of the ethnic groups. Confirmation Study To confirm the prevalence of APOC3 variants and the results of the association analysis, we examined the two SNPs in the ARIC study. HTGC was not measured in the ARIC study; therefore, we tested each variant for an association with fasting plasma TG levels and with HOMA-IR. Clinical characteristics of the ARIC population stratified by the APOC3 genotype are shown in Supporting Tables 6 to 8. The frequencies of the APOC3 rs2854117 and rs2854116 polymorphisms in the ARIC population were similar to those observed in the Dallas Heart Study in African Americans and Caucasians (Supporting Tables 6-8). The linkage disequilibrium block structure between the variants in the APOA5/APOA4/APOC3/APOA1 gene cluster mirrored that calculated with the Dallas Heart Study data (Supporting Fig. 2). The variant allele at rs2854117 (T) was associated with higher plasma TG levels in ARIC whites (P = 0.001) but not in African Americans. Neither variant was significantly associated with HOMA-IR. No associations were observed when the analysis was confined to nondiabetic, lean (BMI < 25 kg/m2) individuals who consumed <30 g of ethanol per day (684 African Americans, including 329 men, and 3715 Caucasians, including 1167 men).

t not in African Americans. Neither variant was significantly associated with HOMA-IR. No associations were observed when the analysis was confined to nondiabetic, lean (BMI < 25 kg/m2) individuals who consumed <30 g of ethanol per day (684 African Americans, including 329 men, and 3715 Caucasians, including 1167 men). Discussion The goal of the present study was to elucidate the nature of the relationship between the liver TG content and insulin resistance through an examination of the metabolic sequelae of two sequence polymorphisms in APOC3 that were associated with the liver TG content in a previous study.13 Here we found that one of the variants (rs2854116) was weakly associated with HTGC (P = 0.041) but in the direction opposite to that of the previous report.13 Carriers for this variant had lower HTGC values than noncarriers. The rs2854117 polymorphism had no detectable effect on levels of liver fat. Neither variant was associated with HOMA-IR in the Dallas Heart Study or in the ARIC study, even after the analysis was restricted to lean, nondiabetic individuals who consumed <30 g of ethanol per day. Thus, the minimal effect of the variants on the liver TG content and the lack of an effect on insulin resistance could not be explained by confounding due to ethnicity, obesity, or alcohol use. Taken together, our findings do not support a model in which a genetic polymorphism in the putative insulin response element of APOC3 leads to increased accumulation of TG in the liver, which in turn results in insulin resistance.

sistance could not be explained by confounding due to ethnicity, obesity, or alcohol use. Taken together, our findings do not support a model in which a genetic polymorphism in the putative insulin response element of APOC3 leads to increased accumulation of TG in the liver, which in turn results in insulin resistance. Rather, our data are consistent with previous findings showing that genetic variation in APOC3 expression has little effect on insulin sensitivity in humans25 or in mice.26

sistance could not be explained by confounding due to ethnicity, obesity, or alcohol use. Taken together, our findings do not support a model in which a genetic polymorphism in the putative insulin response element of APOC3 leads to increased accumulation of TG in the liver, which in turn results in insulin resistance. Rather, our data are consistent with previous findings showing that genetic variation in APOC3 expression has little effect on insulin sensitivity in humans25 or in mice.26 The reasons for the differences between our results and those of Petersen et al.13 are not clear. One possibility is that secondary factors such as obesity, exercise training, and alcohol use obscured the relationship between the APOC3 polymorphism and insulin resistance in our population. However, the two polymorphisms were not associated with insulin resistance even when these factors were excluded. An alternative explanation is that the differences are due to ethnic differences in the two populations. The initial association between the sequence variations in APOC3 and HTGC was identified in 95 Asian Americans, a group not represented in our study. It may be that the nucleotide substitutions on the variant allele of APOC3 are not directly responsible for the associations with liver fat and insulin resistance observed in their Asian-Indian subjects but are in linkage disequilibrium with causal variants in this population that are not present in the ethnic groups (Caucasians, African Americans, and Hispanics) that we studied. However, Petersen et al. reported essentially identical findings in a second group composed mainly of Caucasian men. Therefore, it seems highly unlikely that differences in allele structure can account for the different effects of APOC3 alleles observed in the two studies. Recently, Pollin et al.25 reported that 5% of the Lancaster Amish are heterozygotes for a null allele (R19X) of APOC3. Carriers of the mutation had a 50% reduction in circulating levels of APOC3, but their plasma glucose and insulin levels were indistinguishable from those of noncarriers. Thus, an approximately 50% reduction in circulating levels of APOC3 is not associated with changes in insulin sensitivity, at least in this population. Similarly, mice overexpressing APOC3 did not have insulin resistance despite markedly elevated plasma TG levels.18

els were indistinguishable from those of noncarriers. Thus, an approximately 50% reduction in circulating levels of APOC3 is not associated with changes in insulin sensitivity, at least in this population. Similarly, mice overexpressing APOC3 did not have insulin resistance despite markedly elevated plasma TG levels.18 The rs2854117 and rs2854116 variants were reported to be in linkage disequilibrium with a polymorphic SstI site in the 3′ untranslated region of APOC3 that was strongly associated with hypertriglyceridemia in several small studies.14, 27 In subsequent studies using larger samples, the association was inconsistent. Russo et al.28 reported that the SstI variant was not significantly associated with plasma TG levels in 1219 men and 1266 women in the Framingham Heart Study. Conversely, Waterworth et al.29 found a modest but statistically significant association between the SStI polymorphism and plasma TG levels in 2745 Caucasian men from the Second Northwick Park Heart Study. In the present study, the rs2854117 polymorphism was significantly associated with plasma TG levels in 9799 Caucasian men and women from the ARIC study. The effect on plasma TG levels of the rs284117 variant and the SstI polymorphism was comparable in all three studies: homozygotes for the minor allele had plasma TG levels that were approximately 10% higher than those of homozygotes for the common allele. Thus, a single allele altered plasma TG levels by approximately 5%. In contrast, the APOC3 nonsense allele (R19X) reported by Pollin et al.,25 which presumably decreased APOC3 expression by 50% in heterozygotes, caused a 45% reduction in plasma TG levels. Taken together, these results indicate that the rs2854117 allele has very modest effects on APOC3 expression and plasma TG levels.

. In contrast, the APOC3 nonsense allele (R19X) reported by Pollin et al.,25 which presumably decreased APOC3 expression by 50% in heterozygotes, caused a 45% reduction in plasma TG levels. Taken together, these results indicate that the rs2854117 allele has very modest effects on APOC3 expression and plasma TG levels. Hepatic steatosis is now recognized as one of the adverse metabolic consequences of obesity that constitute metabolic syndrome.30 Despite extensive investigation, the mechanistic couplings between the various components of the syndrome have not been fully defined. In cross-sectional studies, the liver fat content has been correlated with indices of insulin resistance.3, 31, 32 These findings are consistent with the proposal that high liver TG levels cause insulin resistance.31, 32 However, data from humans with genetic variations that cause a primary increase in the liver fat content are incompatible with this view. Mutations in APOB lead to hepatic steatosis but are not associated with increased HOMA-IR33 or reduced insulin-mediated glucose disposal.34 Similarly, the I148M allele of PNPLA3 has been systematically associated with increased liver TG content but not with insulin resistance as determined by HOMA-IR or euglycemic clamp studies.4 Thus, increased liver TG content per se does not lead to insulin resistance. The authors thank Tommy Hyatt for excellent technical assistance and Jay Horton for helpful discussions. Abbreviations ALTalanine aminotransferase APOapolipoprotein ARICAtherosclerosis Risk in Communities ASTaspartate aminotransferase BMIbody mass index

Hepatic steatosis is now recognized as one of the adverse metabolic consequences of obesity that constitute metabolic syndrome.30 Despite extensive investigation, the mechanistic couplings between the various components of the syndrome have not been fully defined. In cross-sectional studies, the liver fat content has been correlated with indices of insulin resistance.3, 31, 32 These findings are consistent with the proposal that high liver TG levels cause insulin resistance.31, 32 However, data from humans with genetic variations that cause a primary increase in the liver fat content are incompatible with this view. Mutations in APOB lead to hepatic steatosis but are not associated with increased HOMA-IR33 or reduced insulin-mediated glucose disposal.34 Similarly, the I148M allele of PNPLA3 has been systematically associated with increased liver TG content but not with insulin resistance as determined by HOMA-IR or euglycemic clamp studies.4 Thus, increased liver TG content per se does not lead to insulin resistance. The authors thank Tommy Hyatt for excellent technical assistance and Jay Horton for helpful discussions. Abbreviations ALTalanine aminotransferase APOapolipoprotein ARICAtherosclerosis Risk in Communities ASTaspartate aminotransferase BMIbody mass index HOMA-IRhomeostatic model assessment of insulin resistance HTGChepatic triglyceride content MAFminor allele frequency MRSmagnetic resonance spectroscopy NAFLDnonalcoholic fatty liver disease PNPLA3patatin-like phospholipase domain containing 3 ρSpearman's rank correlation coefficient SNPsingle-nucleotide polymorphism TGtriglyceride

BMIbody mass index HOMA-IRhomeostatic model assessment of insulin resistance HTGChepatic triglyceride content MAFminor allele frequency MRSmagnetic resonance spectroscopy NAFLDnonalcoholic fatty liver disease PNPLA3patatin-like phospholipase domain containing 3 ρSpearman's rank correlation coefficient SNPsingle-nucleotide polymorphism TGtriglyceride Supplementary material Additional Supporting Information may be found in the online version of this article. Supporting Figure 1. Linkage disequilibrium (R2) between single nucleotide polymorphisms in APO A5/A4/C3/A1 gene cluster on chromosome 11q23 in the African-Americans (top), European-Americans (middle) and Hispanics (bottom) in the Dallas Heart Study. Gene and SNP positions correspond to NCBI genome build 34 (July 2003). SNPs were genotyped using a custom oligonucleotide genotyping array (Perlegen Sciences). Correlations were calculated from genotype data of 1,730 African Americans, 996 European Americans and 581 Hispanics. Only common variants (minor allele frequency (MAF) of ≥ 5%) are shown in the correlation map for each ethnic group. Supporting Figure 2. Linkage disequilibrium (R2) between 41 single nucleotide polymorphisms in the APO A5/A4/C3/A1 gene cluster on chromosome 11q23 in African-Americans (top), European-Americans (bottom) in the ARIC study. Gene and SNP positions correspond to NCBI genome build 36 (March 2006).

Supporting Figure 1. Linkage disequilibrium (R2) between single nucleotide polymorphisms in APO A5/A4/C3/A1 gene cluster on chromosome 11q23 in the African-Americans (top), European-Americans (middle) and Hispanics (bottom) in the Dallas Heart Study. Gene and SNP positions correspond to NCBI genome build 34 (July 2003). SNPs were genotyped using a custom oligonucleotide genotyping array (Perlegen Sciences). Correlations were calculated from genotype data of 1,730 African Americans, 996 European Americans and 581 Hispanics. Only common variants (minor allele frequency (MAF) of ≥ 5%) are shown in the correlation map for each ethnic group. Supporting Figure 2. Linkage disequilibrium (R2) between 41 single nucleotide polymorphisms in the APO A5/A4/C3/A1 gene cluster on chromosome 11q23 in African-Americans (top), European-Americans (bottom) in the ARIC study. Gene and SNP positions correspond to NCBI genome build 36 (March 2006). Supporting Table 1. Oligonucleotides used for genotyping APOC3 SNPs in the Dallas Heart Study (DHS) population and in Atherosclerosis Risk in Communities (ARIC) study. Supporting Table 2. Clinical characteristics of the Dallas Heart Study participants stratified by APOC3 rs2854117 genotype and ethnicity. Supporting Table 3. Clinical characteristics of the Dallas Heart Study participants stratified by APOC3 rs2854116 genotype and ethnicity. Supporting Table 4. Clinical characteristics of the Dallas Heart Study participants stratified by APOC3 rs2854117 genotype Supporting Table 5. Clinical characteristics of the Dallas Heart Study participants stratified by APOC3 rs2854116 genotype. Supporting Table 6. Clinical characteristics of Atherosclerosis Risk in Communities study (ARIC) participants stratified by APOC3 rs2854117 genotype and ethnicity. Supporting Table 7. Clinical characteristics of Atherosclerosis Risk in Communities study (ARIC) participants stratified by APOC3 rs2854116 genotype and ethnicity. Supporting Table 8. Clinical characteristics of APOC3 wild-type homozygotes (rs2854117 C/C and rs2854116 T/T) compared to carriers of one or more variant alleles (rs2854117 T and rs2854116 C) in Atherosclerosis Risk in Communities study (ARIC) participants, stratified by ethnicity.

Genetic predisposition influences the development of primary sclerosing cholangitis (PSC).1 The co-occurrence of inflammatory bowel disease and classical autoimmune diseases in patients with PSC suggests that loss of immune tolerance contributes to the pathogenesis. The strongest genetic risk factors in PSC are found in the human leukocyte antigen (HLA) complex at chromosome 6p21.2 Many of the genes in this region are immune-related, and variants tend to be inherited together on extended haplotypes, i.e., they are in strong linkage disequilibrium (LD). Deciphering the contribution of the various genes in the region is a major challenge in disease genetics. As for most HLA-associated diseases, a multitude of HLA class I and class II gene associations have been reported in PSC, most consistently for alleles that are components of the extended ancestral haplotypes AH8.1 (i.e., HLA-B*08-DRB1*03 [serological DR3]) and AH7.1 (i.e., HLA-B*07-DRB1*15 [serological DR2]), along with various less conserved HLA class II haplotypes, namely, DRB1*13:01, DRB1*04, and DRB1*07.3-6 In genome-wide association studies,2,7 strong associations near HLA-C, HLA-B, and MICA suggest a role for these loci in modifying PSC risk. The mechanism could involve an effect of alleles carried by the AH8.1 and AH7.1 haplotypes on the activation level of natural killer cells and T cells.8-11 However, associations detected for HLA class II haplotypes appear to have a significant influence on PSC, in addition to the effect of HLA class I.7

fying PSC risk. The mechanism could involve an effect of alleles carried by the AH8.1 and AH7.1 haplotypes on the activation level of natural killer cells and T cells.8-11 However, associations detected for HLA class II haplotypes appear to have a significant influence on PSC, in addition to the effect of HLA class I.7 The class II genes encode heterodimers consisting of an α and a β chain (e.g., the HLA-DR molecule is encoded by HLA-DRA and HLA-DRB1) which present peptides to CD4-positive T cells. The sequences encoded by the second exon of class II genes determine the properties of the peptide-binding groove. In several autoimmune diseases HLA class II associations have been attributed to particular amino acids in the molecule that critically determine the binding of disease-specific antigen(s). One example is the protective effect in type 1 diabetes of HLA-DQβ1 chains with aspartic acid in residue 57,12 which induces distinct characteristics of the peptide-binding groove of the HLA-DQ molecule.13 Determination of the structural and electrostatic properties of the molecules associated with disease may help in identifying the disease mechanism. In primary biliary cirrhosis and autoimmune hepatitis, specific residues have been suggested to explain associations with HLA-DRB1 alleles.14,15 In PSC, an association with leucine in residue 38 of the HLA-DRβ chain was proposed by Farrant et al.,16 whereas a later study considered residues 55 and 87 of the HLA-DQβ chain as more likely candidates.3 A consistent peptide-binding motif for the class II molecules associated with PSC has not been defined, and no attempts have been made to model how specific amino acids affect the structure and the electrostatic properties of the peptide-binding groove.

esidues 55 and 87 of the HLA-DQβ chain as more likely candidates.3 A consistent peptide-binding motif for the class II molecules associated with PSC has not been defined, and no attempts have been made to model how specific amino acids affect the structure and the electrostatic properties of the peptide-binding groove. The portal inflammation in PSC livers is dominated by T cells, which seem to exhibit a restricted T-cell receptor repertoire.17 It would be of importance to identify characteristics of the HLA molecules that determine the specificity of these T-cell responses. Strong LD in the HLA class II region makes it difficult to determine at the genetic level which loci are most relevant. However, several minor observations suggest that HLA-DRB1 could be the determinant of PSC risk; (1) The HLA-DQA1 and DQB1 alleles encoded on the AH8.1 haplotype are associated with PSC only on this haplotype and not when encoded on different haplotypes.4 (2) The protective DRB1*04 haplotypes may carry different DQB1 alleles.4 (3) A recent study in African-Americans confirms the association with DR13,18 which in Northern Europe forms the DRB1*13:01-DQB1*06:03 haplotype,16 whereas in African-Americans both DRB1*13:01-DQB1*06:03 and DRB1*13:01-DQB1*05:02 are common haplotypes.19 The HLA-DRB1 association is also more consistent than the association with the closely related (paralogous) HLA-DRB3 gene; e.g., PSC-associated HLA-DRB1*13:01 haplotypes may carry either the HLA-DRB3*01:01 or DRB3*02:02 alleles.4 Given this background we aimed to explore how HLA-DRB1 variation affects the molecular characteristics of HLA-DR and susceptibility to PSC.

than the association with the closely related (paralogous) HLA-DRB3 gene; e.g., PSC-associated HLA-DRB1*13:01 haplotypes may carry either the HLA-DRB3*01:01 or DRB3*02:02 alleles.4 Given this background we aimed to explore how HLA-DRB1 variation affects the molecular characteristics of HLA-DR and susceptibility to PSC. Materials and Methods Subjects Scandinavian PSC patients (n = 356, Table 1) were recruited from Oslo University Hospital, Rikshospitalet, Oslo, Norway, and Karolinska University, Hospital Huddinge, Stockholm, Sweden. Diagnosis of PSC was based on accepted criteria with typical cholangiographic appearance. Ethnically and gender-matched healthy controls (n = 366) were randomly selected from the Norwegian Bone Marrow Registry. All participants gave informed consent. The study was approved by the Regional Committee for Research Ethics in South-Eastern Norway and the Ethics Committee of Karolinska Institutet. Table 1 Characteristics of Included Individuals PSC Healthy Controls N 356* 366 Male, n (%) 254 (71) 256 (70) Age at diagnostic cholangiography, years, median (range) 36 (12-75) — Concomitant inflammatory bowel disease, n (%) 290† (82) — Ulcerative colitis/Crohn's disease/indeterminate, % 81/12/7 — Cholangiocarcinoma, n (%) 50 (14) — Endpoint (tx or death), n (%) 210 (59) — Follow-up, years, median (range) 10 (0-34) — * Norway n = 230, Sweden n = 126. † Missing information about the intestine in three patients.

PSC Healthy Controls N 356* 366 Male, n (%) 254 (71) 256 (70) Age at diagnostic cholangiography, years, median (range) 36 (12-75) — Concomitant inflammatory bowel disease, n (%) 290† (82) — Ulcerative colitis/Crohn's disease/indeterminate, % 81/12/7 — Cholangiocarcinoma, n (%) 50 (14) — Endpoint (tx or death), n (%) 210 (59) — Follow-up, years, median (range) 10 (0-34) — * Norway n = 230, Sweden n = 126. † Missing information about the intestine in three patients. HLA-DRB1 Data Four-digit HLA-DRB1 genotypes were available from a previous study.20 Peptide sequences of all HLA-DRB1 alleles in IMGT/HLA database release 2.23 (October 2008) were aligned, and each individual was assigned two amino acids (one encoded by each chromosome) for each polymorphic residue.

† Missing information about the intestine in three patients. HLA-DRB1 Data Four-digit HLA-DRB1 genotypes were available from a previous study.20 Peptide sequences of all HLA-DRB1 alleles in IMGT/HLA database release 2.23 (October 2008) were aligned, and each individual was assigned two amino acids (one encoded by each chromosome) for each polymorphic residue. Statistical Methods Stepwise logistic regressions were performed in the statistical package R v2.10.0 (http://www.r-project.org/) assuming an “allele dosage” model, entering the count of all amino acids at a given residue as covariates. A model with all observed combinations of amino acids (“genotypes”) at a given residue entered as covariates was applied to control the validity of the model. Some combinations of amino acids were rare and after testing several criteria, combinations with a frequency of n < 2 in cases or controls at a given residue were grouped in order to avoid empty cells. In both models the reference was randomly chosen, thus no assumptions were made on which amino acid or pair of amino acids constituted high or low risk. Comparisons of allele and carrier frequencies were performed in Microsoft Excel (Redmond, WA) and PASW v. 18 (SPSS, Chicago, IL). P < 0.05 was considered statistically significant. P-values of novel HLA-DRB1 allele associations were Bonferroni corrected according to the number of alleles present in the dataset (n = 32).

omparisons of allele and carrier frequencies were performed in Microsoft Excel (Redmond, WA) and PASW v. 18 (SPSS, Chicago, IL). P < 0.05 was considered statistically significant. P-values of novel HLA-DRB1 allele associations were Bonferroni corrected according to the number of alleles present in the dataset (n = 32). 3D Protein Structure Modeling of HLA-DR Molecules The atomic coordinates of the most common HLA-DR molecules were determined using comparative protein structure modeling by satisfaction of spatial restraints as implemented in the MODELLER computer algorithm.21 HLA-DR proteins of known structure suitable as modeling templates were identified in the Protein Data Bank (PDB; http://www.rcsb.org/pdb/) and evaluated for structural quality. Accordingly, seven structures were selected as templates (PDB entries: 1KLU, 2G9H, 1D5Z, 1D5M, 2Q6W, 1PYW, and 2IPK). The amino acid sequences of the target HLA-DR molecules were obtained from the IMGT/HLA database. Multiple sequence alignments were performed with CLUSTAL_X v.1.8322 and manually corrected when indicated. The alignment files were then used as input to the MODELLER program. In brief, MODELLER generates the 3D atomic coordinates of the target sequences by satisfying spatial restraints, obtained from the templates, and by CHARMM23 energy terms enforcing proper stereochemistry. Optimization is then carried out by employing methods of conjugate gradients and molecular dynamics with simulated annealing.24 All calculations were performed in the absence of antigenic peptides to enable direct comparison of the structural and physiochemical characteristics of the peptide-binding groove among different molecules. The stereochemical quality of the modeled structures was verified using the PROCHECK25 and WHAT_CHECK26 algorithms and by assessment of Ramachandran plots. In addition, the structures were examined for protein folding quality using empirical energy potentials as implemented in the ProSA algorithm.27 Modeled coordinate sets are available upon request.

e modeled structures was verified using the PROCHECK25 and WHAT_CHECK26 algorithms and by assessment of Ramachandran plots. In addition, the structures were examined for protein folding quality using empirical energy potentials as implemented in the ProSA algorithm.27 Modeled coordinate sets are available upon request. Electrostatic Potential Calculations The electrostatic potential around the 3D structures was computed by numerically solving the Poisson Boltzmann equation using the finite difference method implemented in the DelPhi program within Discovery Studio 2.1 (Accelrys, San Diego, CA). Essential hydrogens were added to the structures. To determine the protonation state of titratable amino acid side chains the titration curves and residue pKa were calculated for each molecule (dielectric constant of 10 for the protein interior and 80 for the solvent) and titratable residues were protonated at a pH of 7.4. The protonated protein molecule was subsequently used to compute the electrostatic potential. The low dielectric protein interior (dielectric constant of 2) was embedded in a high dielectric continuum environment (water exterior, dielectric constant of 80). A solution with charged ions was simulated with an assigned ionic strength of 0.145, typical of the conditions at a pH of 7.4. The dielectric boundary between the protein and the solvent was defined by calculating the solvent-accessible surface generated by a rolling probe sphere of 1.4 Å radius. Atomic radii and partial atomic charges were taken from the CHARMM parameter set.23 An ion exclusion layer (Stern layer) for the solvent ions was defined around the solvent-accessible surface using an ionic radius of 2 Å. The layer has an ionic strength of 0.0 and determines the maximum distance that an ion can approach the solvent-accessible surface. The system was mapped into a 3D cubical grid and the electrostatic potential at each grid point was calculated iteratively starting from the Debye-Hückel boundary conditions. The accuracy of the calculations was improved by using a method of grid focusing; in the first run the coarse grid was allowed to be filled by 50% by solute and the calculated grid point potentials were used in the second run where the fine grid was filled by solute by 90%. The grid dimensions were set at 251 grid points per axis (spacing 0.3 Å / grid point). The solvent accessible surface was colored according to its calculated electrostatic potential and visualized using the Discovery Studio interface.

tentials were used in the second run where the fine grid was filled by solute by 90%. The grid dimensions were set at 251 grid points per axis (spacing 0.3 Å / grid point). The solvent accessible surface was colored according to its calculated electrostatic potential and visualized using the Discovery Studio interface. Results Statistical Modeling: Identification of Position 37 and 86 in the DRβ1 Chain as PSC-Associated Residues The amino acid sequence encoded by exon two of HLA-DRB1 was determined from the genotypes of each individual. Thirty residues were polymorphic, i.e., two or more different amino acids were observed at these positions. In the first step, a logistic regression was performed for each polymorphic residue. The counts (0, 1, 2) of the observed amino acids were included as covariates and the overall effect of the residue was tested with a likelihood ratio test. The strongest PSC associations were detected for residue 37 (P = 1.2 × 10−32, Table 2). In a second step, two-residue models were fitted containing the amino acid covariates of both the investigated residue and residue 37 and compared with the single-residue model of residue 37. The only residue that remained strongly associated with PSC in these two-residue models was 86 (Table 2). When performing a similar two-residue test for additional effects on top of 86, several residues (i.e., also residue 37) were found to contribute significantly (Table 2). No other residues showed significant disease association when included in three-residue models with residues 37 and 86 (Table 2).

s was 86 (Table 2). When performing a similar two-residue test for additional effects on top of 86, several residues (i.e., also residue 37) were found to contribute significantly (Table 2). No other residues showed significant disease association when included in three-residue models with residues 37 and 86 (Table 2). Table 2 Association Analyses Between Amino Acid Variation in the HLA-DRβ1 Chain and PSC* Single-Residue LR Two-Residue LR† Three-Residue LR‡ Basic Model: - Basic Model: Residue 37 Basic Model: Residue 86 Basic Model: Residues 37+86

s was 86 (Table 2). When performing a similar two-residue test for additional effects on top of 86, several residues (i.e., also residue 37) were found to contribute significantly (Table 2). No other residues showed significant disease association when included in three-residue models with residues 37 and 86 (Table 2). Table 2 Association Analyses Between Amino Acid Variation in the HLA-DRβ1 Chain and PSC* Single-Residue LR Two-Residue LR† Three-Residue LR‡ Basic Model: - Basic Model: Residue 37 Basic Model: Residue 86 Basic Model: Residues 37+86 Residue Observed Amino Acids P-value Rank P-value Rank P-value Rank P-value 9 Glu, Lys,Trp 0.051 27 0.59 22 0.0086 20 0.29 10 Glu,Gln,Tyr 2.4 × 10−11 14 0.54 21 0.0017 17 0.48 11 Asp,Gly,Leu,Pro,Ser,Val 1.2 × 10−18 8 0.23 15 1.7 × 10−8 3 0.43 12 Lys,Thr 2.7 × 10−12 12 0.64 24 0.00046 14 0.74 13 Phe,Gly,His,Arg,Ser,Tyr 9.6 × 10−25 2 0.22 14 2.1 × 10−11 2 0.25 14 Glu,Lys 0.0014 20 0.36 17 0.74 28 0.83 16 His,Tyr 3.2 × 10−5 17 0.88 29 0.00053 15 0.78 25 Gln,Arg 0.0014 21 0.36 18 0.74 29 0.83 26 Phe,Leu,Tyr 2.6 × 10−19 6 0.024 2 3.9 × 10−8 4 0.23 28 Asp,Glu,His 2.8 × 10−5 16 0.13 8 0.0053 19 0.10 30 Cys,Gly,His Leu,Arg,Tyr 0.00012 18 0.12 7 2.0 × 10−5 11 0.20 31 Phe,Ile,Val 0.31 29 0.078 5 0.0046 18 0.12 32 His,Tyr 1.2 × 10−21 4 0.38 19 6.6 × 10−8 6 0.64 33 His,Asn 1.4 × 10−15 11 0.40 20 5.0 × 10−8 5 0.31 37 Phe,Leu,Asn,Ser,Tyr 1.2 × 10−32 1 — 3.1 × 10−16 1 — 38 Ala,Leu,Val 0.029 26 0.17 11 0.0014 16 0.14 40 Phe,Tyr 0.77 30 0.17 12 0.52 27 0.14 47 Phe,Tyr 2.4 × 10−21 5 0.11 6 0.00024 13 0.29 57 Ala,Asp,Ser,Val 2.6 × 10−5 15 0.78 26 0.16 22 0.44 58 Ala,Glu 0.00055 19 0.85 28 0.035 21 0.84 60 His,Ser,Tyr 0.0046 24 0.59 23 0.28 24 0.29 67 Phe,Ile,Leu 0.0018 22 0.80 27 0.45 25 0.22 70 Asp,Gln,Arg 0.063 28 0.16 10 0.49 26 0.32 71 Ala,Glu,Lys,Arg 1.4 × 10−15 10 0.043 3 5.4 × 10−6 8 0.32 73 Ala,Gly 5.5 × 10−12 13 0.14 9 1.7 × 10−5 10 0.34 74 Ala,Glu,Leu,Gln,Arg 1.7 × 10−18 9 0.25 16 9.4 × 10−6 9 0.59 77 Asn,Thr 5.8 × 10−19 7 0.060 4 4.8 × 10−7 7 0.34 78 Val,Tyr 0.023 25 0.77 25 0.25 23 0.20 85 Ala,Val 0.0037 23 0.21 13 7.2 × 10−5 12 0.13 86 Gly,Val 1.8 × 10−22 3 2.0 × 10−5 1 — — * Stepwise logistic regressions were performed assuming an “allele dosage” effect.

Glu,Leu,Gln,Arg 1.7 × 10−18 9 0.25 16 9.4 × 10−6 9 0.59 77 Asn,Thr 5.8 × 10−19 7 0.060 4 4.8 × 10−7 7 0.34 78 Val,Tyr 0.023 25 0.77 25 0.25 23 0.20 85 Ala,Val 0.0037 23 0.21 13 7.2 × 10−5 12 0.13 86 Gly,Val 1.8 × 10−22 3 2.0 × 10−5 1 — — * Stepwise logistic regressions were performed assuming an “allele dosage” effect. † P-values of likelihood ratio tests of whether residue n improves the logistic regression model when added to a model with one other residue (37 or 86). ‡ P-values of likelihood ratio tests of whether residue n improves the model when added to a model with both residues 37 and residue 86. LR: likelihood ratio; Rank: residue rank according to P-value, lowest P-value is highlighted in bold.

† P-values of likelihood ratio tests of whether residue n improves the logistic regression model when added to a model with one other residue (37 or 86). ‡ P-values of likelihood ratio tests of whether residue n improves the model when added to a model with both residues 37 and residue 86. LR: likelihood ratio; Rank: residue rank according to P-value, lowest P-value is highlighted in bold. In the logistic models used above the effect of a single amino acid was assumed to be additive on the log-scale: The log-odds ratio of having PSC given two copies of the amino acid is two times the log-odds ratio when having one copy. The advantage with this model is that it keeps the number of covariates to a minimum, leading to more powerful tests as long as the model assumptions are approximately true. In order to confirm the results obtained with this model, we also performed regressions where we allowed each observed combination (“genotype”) of amino acids to have a potential effect. In these “genotype” model analyses, residue 37 remained the most significantly PSC-associated residue (P = 6.9 × 10−32, Table 3), with an independent contribution from residue 86 still observed (P = 1.2 × 10−5). Several other residues contributed on top of residues 37 or 86 in two-residue models, as well as in three-residue models with both residues 37 and 86 included (Table 3). When inspecting the distribution of amino acid combinations (“genotypes”) in the dataset, it became apparent that the extra associated residues 26, 70, 71, 73, 74, and 77 (Table 3) reflected a large number of patients homozygous for HLA-DRB1*03:01 (n = 62 patients versus n = 3 healthy controls), meaning that it was not possible to determine the part of HLA-DRB1*03:01 that confers this additional risk.

aset, it became apparent that the extra associated residues 26, 70, 71, 73, 74, and 77 (Table 3) reflected a large number of patients homozygous for HLA-DRB1*03:01 (n = 62 patients versus n = 3 healthy controls), meaning that it was not possible to determine the part of HLA-DRB1*03:01 that confers this additional risk. 3 Summary of Association Analyses Between Amino Acid Variation and PSC Assuming a “Genotype” Model* Single-Residue LR Two-Residue LR†, Three-Residue LR‡ Basic Model: - Basic Model: Residue 26 Basic Model: Residue 37 Basic Model: Residues 77 Basic Model: Residues 86 Basic Model: Residues 37 + 86 Residue P-value P-value P-value P-value P-value P-value 26 6.2×10−20 — 4.2×10−7 0.19 1.3×10−8 4.1×10−5 37 6.9×10−32 3.3×10−19 — 1.5×10−19 4.8×10−17 — 70 0.013 0.23 0.023 0.083 0.055 0.036 71 1.6×10−17 3.8×10−14 0.00034 3.4×10−13 7.1×10−8 0.014 73 1.1×10−12 0.15 0.00075 0.17 8.1×10−6 0.0038 74 1.6×10−18 0.14 6.0×10−5 0.096 1.3×10−6 0.00093 77 7.3×10−21 0.80 8.5×10−8 — 7.8×10−9 3.4×10−6 86 2.0×10−21 1.3×10−9 1.2×10−5 2.2×10−9 — — * Stepwise logistic regressions were performed assuming a “genotype” effect, entering all different pairs (combinations) of amino acids at a given residue. Only residues contributing significantly to the model when added to residue 37 and 86 in the three-residue regressions are shown. † P-values of likelihood ratio tests of whether residue n improves the logistic regression model when added to a model with one other residue (26, 37, 77, or 86).

Residue P-value P-value P-value P-value P-value P-value 26 6.2×10−20 — 4.2×10−7 0.19 1.3×10−8 4.1×10−5 37 6.9×10−32 3.3×10−19 — 1.5×10−19 4.8×10−17 — 70 0.013 0.23 0.023 0.083 0.055 0.036 71 1.6×10−17 3.8×10−14 0.00034 3.4×10−13 7.1×10−8 0.014 73 1.1×10−12 0.15 0.00075 0.17 8.1×10−6 0.0038 74 1.6×10−18 0.14 6.0×10−5 0.096 1.3×10−6 0.00093 77 7.3×10−21 0.80 8.5×10−8 — 7.8×10−9 3.4×10−6 86 2.0×10−21 1.3×10−9 1.2×10−5 2.2×10−9 — — * Stepwise logistic regressions were performed assuming a “genotype” effect, entering all different pairs (combinations) of amino acids at a given residue. Only residues contributing significantly to the model when added to residue 37 and 86 in the three-residue regressions are shown. † P-values of likelihood ratio tests of whether residue n improves the logistic regression model when added to a model with one other residue (26, 37, 77, or 86). ‡ P-values of likelihood ratio tests of whether residue n improves the model when added to a model with both residues 37 and residue 86. LR: likelihood ratio. In conclusion, residues 37 and 86 were consistent determinants of PSC susceptibility irrespective of statistical model, whereas it was difficult to exclude additional risk associated with other parts of the β chain encoded by HLA-DRB1*03:01.

‡ P-values of likelihood ratio tests of whether residue n improves the model when added to a model with both residues 37 and residue 86. LR: likelihood ratio. In conclusion, residues 37 and 86 were consistent determinants of PSC susceptibility irrespective of statistical model, whereas it was difficult to exclude additional risk associated with other parts of the β chain encoded by HLA-DRB1*03:01. Residue 37 Influences the Electrostatic Properties of Pocket P9 of HLA-DR The amino acid frequencies at residues 37 (Table 4) showed that the highest and lowest risks of PSC were observed for carriers of asparagine (Asn37) (odds ratio [OR] = 5.7, 95% confidence interval [CI] 4.0-8.0) and tyrosine (Tyr37) (OR = 0.25, 95% CI 0.18-0.34), respectively. Table 4 Frequencies of Different Amino Acids at HLA-DRβ1 Residues 37 and 86 Allele Frequency, n (%) Carrier Frequency, n (%)

Residue 37 Influences the Electrostatic Properties of Pocket P9 of HLA-DR The amino acid frequencies at residues 37 (Table 4) showed that the highest and lowest risks of PSC were observed for carriers of asparagine (Asn37) (odds ratio [OR] = 5.7, 95% confidence interval [CI] 4.0-8.0) and tyrosine (Tyr37) (OR = 0.25, 95% CI 0.18-0.34), respectively. Table 4 Frequencies of Different Amino Acids at HLA-DRβ1 Residues 37 and 86 Allele Frequency, n (%) Carrier Frequency, n (%) Residue Amino Acid PSC Healthy Controls PSC Healthy Controls 37 Asparagine (Asn) 398 (56) 199 (27) 292 (82) 163 (45) Leucine (Leu) 4 (1) 16 (2) 4 (1) 15 (4) Phenylalanine (Phe) 35 (5) 67 (9) 34 (10) 62 (17) Serine (Ser) 192 (27) 208 (28) 162 (46) 175 (48) Tyrosine (Tyr) 83 (12) 242 (33) 80 (22) 197 (54) 86 Glycine (Gly) 185 (26) 368 (50) 164 (46) 284 (78) Valine (Val) 527 (74) 364 (50) 335 (94) 282 (77) The specificity of the peptide-binding groove on an HLA class II molecule is governed by the properties of pockets in the groove that accommodate the amino acid side chains of the bound peptide, typically pockets for peptide residues 1 (pocket P1), 4, 6, and 9. Residue 37 of the HLA-DRβ1 chain is integral to pocket P9.28 Figure 1 shows the structural and electrostatic characteristics of pocket P9 on representative HLA-DR molecules. Significantly, HLA-DR carrying the risk residue Asn37 in the β chain (e.g., HLA-DRB1*03:01, *09:01, *13:01, *14:02; Fig. 1B) formed P9 pockets with similar structural architecture and consistently positive surface electrostatic potential (the only exception was HLA-DRB1*13:02, further discussed below). In contrast, HLA-DR molecules expressing the protective Tyr37 residue in the β chain (e.g., HLA-DRB1*04:01, *10:01, *11:01, *03:25; Fig. 1C) formed P9 pockets with consistently negative electrostatic potential. The distinct P9 pocket electrostatic patterns were conserved both among molecules that differed at several amino acid sequence positions and between structures where residue 37 constitutes the only disparity (e.g., HLA-DRB1*03:01 and -DRB1*03:25). Interestingly, a database search for peptides eluted from HLA-DR molecules showed that the presence of Asn37 restricted the amino acid preferences at position 9 (e.g., only tyrosine, leucine, and phenylalanine are defined as P9 anchors in HLA-DRB1*0301), whereas most amino acids may be P9 anchors in HLA-DRB1*0401 which carries Tyr37 (http://www.syfpeithi.de).29

es eluted from HLA-DR molecules showed that the presence of Asn37 restricted the amino acid preferences at position 9 (e.g., only tyrosine, leucine, and phenylalanine are defined as P9 anchors in HLA-DRB1*0301), whereas most amino acids may be P9 anchors in HLA-DRB1*0401 which carries Tyr37 (http://www.syfpeithi.de).29 Fig. 1 Structure and molecular surface electrostatic potential of pocket P9. (A) The structure and electrostatic potential of HLA-DRB1*03:01. The area within the frame is depicted in expanded form in (B,C). All structures were superimposed on HLA-DRB1*03:01 and therefore show the same view. HLA-DR carrying the risk residue Asn37 in the β chain had P9 pockets (arrows) with positive charge (B), whereas molecules expressing Tyr37 had P9 pockets (arrows) with consistently negative charge (C). Potentials less than −5 kT/e are colored red, those greater than 5 kT/e blue, and neutral potentials (0 kT/e) are colored white. Linear interpolation was used to produce the color for surface potentials between these values.

ereas molecules expressing Tyr37 had P9 pockets (arrows) with consistently negative charge (C). Potentials less than −5 kT/e are colored red, those greater than 5 kT/e blue, and neutral potentials (0 kT/e) are colored white. Linear interpolation was used to produce the color for surface potentials between these values. Residue 86 Defines Opposite Effects of HLA-DRB1*13:01 and *13:02 on PSC Risk At the dimorphic residue 86, the highest risk was observed for carriers of valine (Val86) (OR = 4.8, 95% CI 2.9-7.9), whereas glycine (Gly86) appeared protective (OR = 0.25, 95% CI 0.18-0.34). Residue 86 of the HLA-DRβ1 chain is integral to pocket P1.28 In contrast to pocket P9, modeling of the P1 pocket of several HLA-DR molecules showed that the glycine/valine dimorphism at residue 86 had a minimal physiochemical effect. The majority of HLA-DR molecules examined had P1 pockets with an overall neutral charge (Fig. 2). Even though a steric effect (i.e., an effect on the volume of the pocket) imposed by the side chain of Val86 cannot be excluded, the results of the present analysis argue against a significant role of residue 86 on the choice of peptide residue accommodated by pocket P1.

pockets with an overall neutral charge (Fig. 2). Even though a steric effect (i.e., an effect on the volume of the pocket) imposed by the side chain of Val86 cannot be excluded, the results of the present analysis argue against a significant role of residue 86 on the choice of peptide residue accommodated by pocket P1. Fig. 2 Structure and molecular surface electrostatic potential of pocket P1. (A) The structure and electrostatic potential of HLA-DRB1*03:01. The area within the frame is depicted in expanded form in (B). All structures were superimposed on HLA-DRB1*03:01 and therefore show the same view. Structural modeling and calculation of the electrostatic potential at the P1 pocket (arrows) of representative HLA-DR molecules showed that the Gly/Val dimorphism at position 86 had a minimal physiochemical effect (B). The majority of HLA-DR molecules examined had P1 pockets with an overall neutral charge. HLA-DRB1*03:01 and -DRB1*14:01 express Val86 whereas -DRB1*01:01 and -DRB1*04:01 express Gly86. Potentials less than −5 kT/e are colored red, those greater than 5 kT/e blue, and neutral potentials (0 kT/e) are colored white. Linear interpolation was used to produce the color for surface potentials between these values.

HLA-DRB1*03:01 and -DRB1*14:01 express Val86 whereas -DRB1*01:01 and -DRB1*04:01 express Gly86. Potentials less than −5 kT/e are colored red, those greater than 5 kT/e blue, and neutral potentials (0 kT/e) are colored white. Linear interpolation was used to produce the color for surface potentials between these values. Further analysis, however, led to an interesting observation. As mentioned above, HLA-DR molecules expressing the risk residue Asn37 in their β chain possess electropositive P9 pockets, with the exception of HLA-DRB1*13:02 where an electronegative P9 pocket was observed (Fig. 3). Notably, when looking at the allele frequencies, HLA-DRB1*13:02, as opposed to other Asn37 encoding alleles (like the established PSC risk allele HLA-DRB1*13:01), was more frequent in healthy controls than in PSC patients (Pcorrected = 0.040, Table 5), suggesting that HLA-DRB1*13:02 may protect against PSC. This statistical observation is therefore in agreement with the protective effect associated with HLA-DR molecules expressing electronegative P9 pockets, as shown above for Tyr37 encoding alleles. Intriguingly, HLA-DRB1*13:02 and DRB1*13:01 have otherwise overall similar structural architecture and electrostatic properties (Fig. 3) with the main disparity observed at pocket P9. Because the only amino acid sequence difference between these alleles is at position 86 it may be suggested that the Gly86Val substitution may influence the choice of presented peptides through long-range electrostatic modification of pocket P9.

ostatic properties (Fig. 3) with the main disparity observed at pocket P9. Because the only amino acid sequence difference between these alleles is at position 86 it may be suggested that the Gly86Val substitution may influence the choice of presented peptides through long-range electrostatic modification of pocket P9. Fig. 3 Electrostatic modification of pocket P9 by the Gly86-to-Val86 substitution. The structure and electrostatic potential of the peptide-binding groove is shown for (structures were superimposed) HLA-DRB1*13:01 (left figures) and -DRB1*13:02 (right figures); these molecules have a single amino acid sequence difference at position 86 (Val86 and Gly86, respectively). HLA-DRB1*13:02 has an electronegative pocket P9 despite the presence of Asn at position 37, suggesting a long-range effect of the Val86-to-Gly86 substitution. The molecular surface is colored according to the calculated electrostatic potential, as for Figs. 1 and 2. Table 5 Frequency of HLA-DRB1Alleles Encoding Asn37 PSC (2n=712) Healthy Controls (2n=732) Allele n (%) n (%) OR (95%CI)* P-value† Residue 86 03:01 254 (36) 106 (14) 3.3 (2.5-4.2) 1.3 × 10−20 Val 09:01 12 (2) 6 (1) 2.0 (0.8-5.0) 0.14 Gly 13:01 117 (17) 47 (6) 2.8 (2.0-4.1) 2.0 × 10−9 Val 13:02 15 (2) 39 (5) 0.4 (0.2-0.7) 0.0013‡ Gly 14:02 0 (0) 1 (0) 0.3 (0.0-3.8) 1.0 Gly * Calculated with Woolf's formula with Haldane's correction. † Not corrected. Calculated with chi-square tests or Fisher's exact test where appropriate. ‡ Pcorrected = 0.040. OR = odds ratio. CI = confidence interval.

Allele n (%) n (%) OR (95%CI)* P-value† Residue 86 03:01 254 (36) 106 (14) 3.3 (2.5-4.2) 1.3 × 10−20 Val 09:01 12 (2) 6 (1) 2.0 (0.8-5.0) 0.14 Gly 13:01 117 (17) 47 (6) 2.8 (2.0-4.1) 2.0 × 10−9 Val 13:02 15 (2) 39 (5) 0.4 (0.2-0.7) 0.0013‡ Gly 14:02 0 (0) 1 (0) 0.3 (0.0-3.8) 1.0 Gly * Calculated with Woolf's formula with Haldane's correction. † Not corrected. Calculated with chi-square tests or Fisher's exact test where appropriate. ‡ Pcorrected = 0.040. OR = odds ratio. CI = confidence interval. Discussion By exploring variation in the amino acid sequence of the HLA-DRβ1 chain in PSC, we show that residues 37 and 86 distinguish disease susceptibility alleles and protective alleles. Investigations into the HLA-DR molecular structure revealed that the electrostatic properties of pocket P9 are determined by residue 37 and, indirectly by residue 86, suggesting that the P9 pocket is crucial for PSC risk.

we show that residues 37 and 86 distinguish disease susceptibility alleles and protective alleles. Investigations into the HLA-DR molecular structure revealed that the electrostatic properties of pocket P9 are determined by residue 37 and, indirectly by residue 86, suggesting that the P9 pocket is crucial for PSC risk. In the HLA-DR molecule, residue 37 of the β chain appeared to be a key determinant of the electrostatic properties of pocket P9, which may be related to disease risk. The situation is reminiscent of type 1 diabetes, where amino acids at residue 57 of the HLA-DQ β chain associated with disease risk contribute to a larger volume of pocket P9 and a positive charge, allowing, e.g., glutamate residues from insulin peptides at position 9.30 In HLA-DR, Asn37 would restrict the range of amino acids at anchor position 9 of the peptide, and thereby which peptides may be presented. This is supported by data from peptide elusion experiments, where HLA-DR molecules with Asn37 and Tyr37 exhibit different ranges of amino acids at P9.29 Direct experimental observations focusing on pocket P9 variation and T-cell responses are scarce, but it has been shown that modification of only residue 37 (on DR4 molecules) is sufficient to alter recognition by the T-cell receptor, e.g., by neutralizing the T-cell-activating potential of the peptide-DR-complex.31,32 It should therefore be considered highly likely that characteristics of pocket P9 of the HLA-DR molecule facilitate particular immune responses.

residue 37 (on DR4 molecules) is sufficient to alter recognition by the T-cell receptor, e.g., by neutralizing the T-cell-activating potential of the peptide-DR-complex.31,32 It should therefore be considered highly likely that characteristics of pocket P9 of the HLA-DR molecule facilitate particular immune responses. Pocket P1 of HLA-DR was found to have an overall neutral electrostatic potential in the present study irrespective of whether glycine or valine was present at position 86. This fits with the observation that this pocket has a preference for hydrophobic amino acid side chains, and that the range of amino acids in position 1 of presented peptides is largely overlapping.33,34 However, pocket P1 with Gly86 in the β chain (e.g., as encoded by HLA-DRB1*13:02) has a tendency to accept larger (aromatic) side chains than when Val86 is present (e.g., encoded by DRB1*13:01); this has been attributed to the lack of a side chain on glycine allowing for a larger pocket volume.33,34 A more remarkable difference between HLA-DRB1*13:01 and DRB1*13:02 encoded HLA-DR molecules was that the amino acid substitution at residue 86 affected the electrostatic properties of pocket P9, in another part of the molecule. HLA-DRB1*13:02 was the only allele which contributed to a HLA-DR molecule with a negative pocket P9 with asparagine at position 37 of the β chain. Intriguingly, this allele exhibited a significantly reduced frequency in PSC patients. Taken together, our findings suggest that the association of residues 37 and 86 with PSC primarily reflects the properties of pocket P9.

ed to a HLA-DR molecule with a negative pocket P9 with asparagine at position 37 of the β chain. Intriguingly, this allele exhibited a significantly reduced frequency in PSC patients. Taken together, our findings suggest that the association of residues 37 and 86 with PSC primarily reflects the properties of pocket P9. Although HLA-DRB1*13:01 is a well-established PSC risk allele,4 this study is the first identifying HLA-DRB1*13:02 as a protective allele. This observation was significant even when correcting for multiple comparisons. Interestingly, similar contrasting effects have been observed in autoimmune hepatitis in Latin America, where risk is associated with HLA-DRB1*13:01 and protection with DRB1*13:02.15 HLA-DRB1*13:01 has also been associated with a protracted course of hepatitis A virus infection, which has been postulated to be a trigger of autoimmune hepatitis.35 To what extent these parallel observations are relevant for the specificity of the immune response in PSC can currently only be speculated.

DRB1*13:02.15 HLA-DRB1*13:01 has also been associated with a protracted course of hepatitis A virus infection, which has been postulated to be a trigger of autoimmune hepatitis.35 To what extent these parallel observations are relevant for the specificity of the immune response in PSC can currently only be speculated. Given the complexity of the HLA associations in diseases such as PSC, it is not unlikely that other alleles besides the most strongly associated ones modify the disease risk. Two previous studies of HLA-DR in PSC evaluated selected residues encoded by haplotypes associated with disease,3,16 and suggested that the presence of leucine at position 38 (Leu38) of the β chain may confer risk. Leu38 is rarely present in DRβ1 (most often encoded by DRB1*12 alleles). An explanation for the conflicting results is that the previous studies included alleles at both the HLA-DRB1 locus as well as those at other, paralogous, HLA-DRB loci. Several HLA haplotypes carry a second HLA-DRB gene besides HLA-DRB1, e.g., HLA-DRB1*03:01 and *13:01 haplotypes typically also carry an allele encoded by HLA-DRB3; DRB1*04 and *07:01 carry an allele encoded by HLA-DRB4, and the DRB1*15:01 haplotype carries an allele encoded by HLA-DRB5. These β chains couple with DRα and also have a role in antigen presentation.36 They are generally observed at several-fold lower expression levels than DRβ1.37,38 However, in diseases where the second DRB gene has been shown to be of actual relevance, the association seems to be specific to the gene in question and not due to shared sequence motifs with DRB1.39–41 These facts, along with the more consistent PSC associations with HLA-DRB1 rather than HLA-DRB3,4 make it likely that the present focus on DRB1 is valid, even though an effect of other DRB loci cannot be formally ruled out at this stage.

ific to the gene in question and not due to shared sequence motifs with DRB1.39–41 These facts, along with the more consistent PSC associations with HLA-DRB1 rather than HLA-DRB3,4 make it likely that the present focus on DRB1 is valid, even though an effect of other DRB loci cannot be formally ruled out at this stage. Given the LD in the HLA complex, we cannot exclude the possibility that causal variants at other loci may be associated with the distribution of amino acids observed at given positions in HLA-DRβ1. The strong LD is particularly important in relation to the neighboring HLA-DQ genes and HLA-DRB paralogs, but it is also difficult to formally exclude an association with the nearby BTNL2 gene, which has been associated with inflammatory bowel disease,42 or even genetic variants further away. When applying a “genotype” model, in addition to residue 37 and 86 we could not exclude a residual association that could be attributed to being homozygous for HLA-DRB1*03:01. This may be speculated to relate to effects of a recessive variant outside HLA-DRB1,4 potentially related to the AH8.1 haplotype which is associated with multiple autoimmune diseases and probably contains several genetic variants in strong LD contributing to disease.43

attributed to being homozygous for HLA-DRB1*03:01. This may be speculated to relate to effects of a recessive variant outside HLA-DRB1,4 potentially related to the AH8.1 haplotype which is associated with multiple autoimmune diseases and probably contains several genetic variants in strong LD contributing to disease.43 In conclusion, this study shows that variation in PSC associated residues encoded by HLA-DRB1 impose distinct structural and physiochemical characteristics on the HLA-DR peptide-binding groove, suggesting that PSC risk molecules likely present a restricted peptide repertoire. The findings are highly relevant for and important to evaluate in future experimental studies of antigen presentation in PSC. The amino acid sequence and structural observations did not apply uniformly to all PSC patients, suggesting multiple pathogenetic mechanisms, as might be expected for a disease with the clinical heterogeneity observed in PSC. We thank Bente Woldseth and Hege Dahlen Sollid for expert technical assistance and the Norwegian Bone Marrow Donor Registry (NORDONOR) at Oslo University Hospital Rikshospitalet for contributing healthy controls. Abbreviations AHancestral haplotype HLAhuman leukocyte antigen LDlinkage disequilibrium PSCprimary sclerosing cholangitis

Since the publication of the American Association for the Study of Liver Diseases (AASLD) practice guidelines on the management of hepatocellular carcinoma (HCC) in 2005, new information has emerged that requires that the guidelines be updated. The full version of the new guidelines is available on the AASLD Web site at http://www.aasld.org/practiceguidelines/Documents/Bookmarked&percnt;20Practice%20Guidelines/HCCUpdate2010.pdf. Here, we briefly describe only new or changed recommendations.

information has emerged that requires that the guidelines be updated. The full version of the new guidelines is available on the AASLD Web site at http://www.aasld.org/practiceguidelines/Documents/Bookmarked&percnt;20Practice%20Guidelines/HCCUpdate2010.pdf. Here, we briefly describe only new or changed recommendations. Surveillance and Diagnosis In the previous guideline, groups were specified for which surveillance was likely to be cost-effective because the hepatocellular carcinoma (HCC) incidence was high enough. New data on defining HCC risk have emerged for hepatitis B virus,1,2 hepatitis C virus,3 and autoimmune hepatitis.4 Surveillance is deemed cost-effective if the expected HCC risk exceeds 1.5% per year in patients with hepatitis C and 0.2% per year in patients with hepatitis B. Analysis of recent studies show that alpha-fetoprotein determination lacks adequate sensitivity and specificity for effective surveillance (and for diagnosis).5,6 Thus, surveillance has to be based on ultrasound examination. The recommended screening interval is 6 months. Diagnosis of HCC should be based on imaging techniques and/or biopsy.The 2005 diagnostic algorithm has been validated and the diagnostic accuracy of a single dynamic technique showing intense arterial uptake followed by “washout” of contrast in the venous-delayed phases has been demonstrated.7-9 Contrast-enhanced US may offer false positive HCC diagnosis in patients with cholangiocarcinoma and thus, has been dropped from the diagnostic techniques. The diagnostic algorithm is shown in Fig. 1. The application of dynamic imaging criteria should be applied only to patients with cirrhosis of any etiology and to patients with chronic hepatitis B who may not have fully developed cirrhosis or have regressed cirrhosis. Interpretation of biopsies and distinction between high-grade dysplatic nodules and HCC is challenging. Expert pathology diagnosis is reinforced by staining for glypican 3, heat shock protein 70, and glutamine synthetase, because positivity for two of these three stains confirms HCC.10

osis or have regressed cirrhosis. Interpretation of biopsies and distinction between high-grade dysplatic nodules and HCC is challenging. Expert pathology diagnosis is reinforced by staining for glypican 3, heat shock protein 70, and glutamine synthetase, because positivity for two of these three stains confirms HCC.10 Fig. 1 Diagnostic algorithm for suspected HCC. CT, computed tomography; MDCT, multidetector CT; MRI, magnetic resonance imaging; US, ultrasound. Staging and Treatment of HCC The BCLC staging system (Fig. 2)11 has come to be widely accepted in clinical practice and is also being used for many clinical trials of new drugs to treat HCC. Therefore, it has become the de facto staging system that is used. Fig. 2 The BCLC staging system for HCC. M, metastasis classification; N, node classification; PS, performance status; RFA, radiofrequency ablation; TACE, transarterial chemoembolization.

Staging and Treatment of HCC The BCLC staging system (Fig. 2)11 has come to be widely accepted in clinical practice and is also being used for many clinical trials of new drugs to treat HCC. Therefore, it has become the de facto staging system that is used. Fig. 2 The BCLC staging system for HCC. M, metastasis classification; N, node classification; PS, performance status; RFA, radiofrequency ablation; TACE, transarterial chemoembolization. The recommendations for liver transplantation have not changed. No new data have emerged that can be used to define a new limit for expanding the patient selection criteria. The usefulness of portal pressure measurement to predict the outcome of patients and define optimal candidates for resection has been validated in Japan.12 Thus, resection should remain the first option for patients who have the optimal profile, as defined by the BCLC staging system. Although resection can be performed in some of these patients with advanced liver disease, the mortality is higher and they might be better served by liver transplantation or ablation. A cohort study of radiofrequency ablation demonstrated that complete ablation of lesions smaller than 2 cm is possible in more than 90% of cases, with a local recurrence rate of less than 1%.13 These data should be confirmed by other groups before positioning ablation as the first-line approach for very early HCC.

ion. A cohort study of radiofrequency ablation demonstrated that complete ablation of lesions smaller than 2 cm is possible in more than 90% of cases, with a local recurrence rate of less than 1%.13 These data should be confirmed by other groups before positioning ablation as the first-line approach for very early HCC. The recommendations regarding patient selection and method of administration of chemoembolization are unchanged. Radioembolization, i.e., the intra-arterial injection of yttrium-90 bound to glass beads or to resin, has been shown to induce tumor necrosis, but there are no data comparing its efficacy to transarterial chemoembolization or to sorafenib treatment for those with portal vein invasion. However, for patients who have either failed transarterial chemoembolization or who present with more advanced HCC, new data indicates the efficacy of sorafenib (a multikinase inhibitor with activity against Raf-1, B-Raf, vascular endothelial growth factor receptor 2, platelet-derived growth factor receptor, c-Kit receptors, among other kinases) in prolonging life.14,15 Sorafenib induces a clinically relevant improvement in time to progression and in survival The magnitude of the improvement in survival compares with other established molecular targeted therapies for other advanced cancers, and the associated toxicity is easily managed without treatment-related mortality. The most frequent adverse events were diarrhea (sorafenib versus placebo: 11% versus 2%) and hand–foot skin reaction (sorafenib versus placebo: 8% versus <1%), fatigue, and weight loss. Sorafenib is now considered first-line treatment in patients with HCC who can no longer be treated with potentially more effective therapies.

t frequent adverse events were diarrhea (sorafenib versus placebo: 11% versus 2%) and hand–foot skin reaction (sorafenib versus placebo: 8% versus <1%), fatigue, and weight loss. Sorafenib is now considered first-line treatment in patients with HCC who can no longer be treated with potentially more effective therapies. In summary, in the past decade HCC has gone from being an almost universal death sentence to a cancer that can be prevented, detected at an early stage, and effectively treated. Physicians caring for patients at risk need to provide high-quality screening, proper management of screen-detected lesions, and provision of therapy that is most appropriate for the stage of disease. Abbreviations AASLDAmerican Association for the Study of Liver Diseases BCLCBarcelona Clinic Liver Cancer HCChepatocellular carcinoma

Liver-directed gene therapy has the potential to treat metabolic diseases and plasma protein deficiencies, including hemophilia.1–3 Moreover, hepatic gene transfer can favor the induction of immune tolerance, so it may also be considered for the prevention or treatment of autoimmune diseases.4,5 Because there is no need to overcome preexisting cellular or humoral immunity to viral vector components, the use of lentiviral vectors (LVs) represents a promising approach for liver gene transfer and immune tolerance induction.6–8

so it may also be considered for the prevention or treatment of autoimmune diseases.4,5 Because there is no need to overcome preexisting cellular or humoral immunity to viral vector components, the use of lentiviral vectors (LVs) represents a promising approach for liver gene transfer and immune tolerance induction.6–8 In previous studies, we showed that immune tolerance induction with LVs is dependent on stringent targeting of transgene expression to hepatocytes, which can be achieved by a combination of transcriptional and posttranscriptional microRNA-based regulation.6,8 Transcriptional control elements derived from hepatocyte-specific genes were incorporated into the vector design to preferentially express the transgene in hepatocytes. Because LVs readily transduce professional antigen-presenting cells,9 target sequences for a hematopoietic-specific microRNA—microRNA 142 (miR-142)–were incorporated downstream of the transgene to suppress any residual expression in antigen-presenting cells.7 This stringent regulation suppressed the induction of immunity against the transgene product in the injected mice and induced antigen-specific immunological tolerance, which could not be reversed by vaccination.6 Using this strategy, we also obtained long-term coagulation factor IX (FIX) expression in hemophilia B mice, which resulted in immune tolerance to FIX and correction of the disease phenotype.8

product in the injected mice and induced antigen-specific immunological tolerance, which could not be reversed by vaccination.6 Using this strategy, we also obtained long-term coagulation factor IX (FIX) expression in hemophilia B mice, which resulted in immune tolerance to FIX and correction of the disease phenotype.8 Although these studies have demonstrated the feasibility of gene replacement therapy and immune tolerance induction with LVs, concerns remain about the potential long-term adverse effects of vector integration due to insertional mutagenesis. Although advanced LV design potentially reduces this risk,10,11 the impact of LV integration in the liver is largely unexplored. Moreover, it is not known whether the tolerogenic outcome of microRNA-regulated LV delivery depends on sustained high levels of transgene expression within hepatocytes, which requires substantial levels of LV integration and would limit the application of this finding outside gene replacement strategies for the correction of monogenic diseases.

whether the tolerogenic outcome of microRNA-regulated LV delivery depends on sustained high levels of transgene expression within hepatocytes, which requires substantial levels of LV integration and would limit the application of this finding outside gene replacement strategies for the correction of monogenic diseases. Integrase-defective lentiviral vectors (IDLVs) are typically generated by the packaging of the vector with catalytically inactive human immunodeficiency virus (HIV) integrase.12,13 The class I D64V mutation in the integrase catalytic site substantially reduces integration (102- to 103-fold) without compromising other steps in the transduction pathway and was thus adopted for this study.12,14-18 Upon transduction, IDLVs can support transgene expression from the nonintegrated proviral forms. Because this episomal DNA is progressively lost in actively dividing cells, transgene expression is only transient.14,15,19 In contrast, IDLVs have been reported to support sustained transgene expression in quiescent mouse tissues, such as the retina and central nervous system, likely because the episomal vector DNA is retained within the postmitotic nucleus.17,18,20 Despite these advances, the ability of IDLVs to achieve therapeutically meaningful transgene expression after hepatic gene delivery and to induce immune tolerance has never been studied. Moreover, the nature of any persistent IDLV genome in transduced cells is poorly defined because a comprehensive characterization of the residual IDLV integration profile is lacking. This study addresses all these outstanding questions and provides evidence supporting IDLVs as an emerging platform technology for liver gene transfer and immune tolerance induction.

enome in transduced cells is poorly defined because a comprehensive characterization of the residual IDLV integration profile is lacking. This study addresses all these outstanding questions and provides evidence supporting IDLVs as an emerging platform technology for liver gene transfer and immune tolerance induction. Materials and Methods Vector Construction Plasmids pCCLsin.cPPT.ET. cFIX.142T, pCCLsin.cPPT.PGK.OVA, and pCC Lsin.cPPT.ET.OVA.142T (ET = enhanced transthyretin, OVA = ovalbumin, and PGK = phosphoglycerokinase) were constructed with standard cloning techniques. Details are available upon request. Mouse Experiments Green fluorescent protein (GFP)–expressing integrase-competent lentiviral vectors (ICLVs) and IDLVs were administered to 7-week-old female BALB/c mice by tail vein injection. Six-week-old female and male C57BL/6 FIX-knockout mice were injected intravenously on 2 consecutive days with a total p24 dose of 260 μg (2 × 500 μL) in FIX-expressing ICLVs or IDLVs supplemented with 40 mg/mL polybrene. For 70% partial hepatectomy, mice were anesthetized with isoflurane. Liver sections were snap-frozen for genomic DNA extraction (Qiagen DNA extraction kit, Qiagen, Belgium). FIX activity in citrated plasma was quantified (Biophen factor IX chromogenic activity assay, Hyphen Biomed, France) according to the manufacturer's instructions with normal dog plasma as a reference (detection limit > 0.1%). All animal experiments were approved by the Animal Ethics Committee of the University of Leuven or the San Raffaele Institutional Animal Care and Use Committee (321).

genic activity assay, Hyphen Biomed, France) according to the manufacturer's instructions with normal dog plasma as a reference (detection limit > 0.1%). All animal experiments were approved by the Animal Ethics Committee of the University of Leuven or the San Raffaele Institutional Animal Care and Use Committee (321). Integration Site (IS) Analysis To analyze ICLV and IDLV integration, we used standard and nonrestrictive, 5′-long terminal repeat (5′-LTR)–mediated and 3′-LTR–mediated linear amplification–mediated polymerase chain reaction (LAM-PCR) as previously described.21,22 Further details can be found in the Supporting Information.

genic activity assay, Hyphen Biomed, France) according to the manufacturer's instructions with normal dog plasma as a reference (detection limit > 0.1%). All animal experiments were approved by the Animal Ethics Committee of the University of Leuven or the San Raffaele Institutional Animal Care and Use Committee (321). Integration Site (IS) Analysis To analyze ICLV and IDLV integration, we used standard and nonrestrictive, 5′-long terminal repeat (5′-LTR)–mediated and 3′-LTR–mediated linear amplification–mediated polymerase chain reaction (LAM-PCR) as previously described.21,22 Further details can be found in the Supporting Information. Results IDLVs Efficiently Transfer Episomal Genomes Into Human Hepatocytes In Vitro but Drive Lower Levels of Transgene Expression Than Their Integration-Competent Counterparts We generated GFP-expressing LVs with either integrase-defective (IDLV) or integrase-competent (ICLV) packaging constructs and compared their transduction efficiency and transgene expression levels in human cell lines and primary hepatocytes in vitro. To drive transgene expression, we used a hepatocyte-specific chimeric promoter (designated as ET).8 The vectors carried target sequences for miR-142 in the transgene 3′-untranslated region (3′-UTR; ET.GFP.142T; Fig. 1A). The physical particle content was determined by HIV-1 group-specific antigen (Gag) p24 quantification. To determine infectivity, which we defined as transducing units per physical particle, we designed an ad hoc quantitative polymerase chain reaction (PCR) that selectively amplified the reverse-transcribed vector genome and discriminated it from plasmid DNA, which was carried over from the transfection used to produce the vectors (Supporting Information Fig. 1). IDLVs and ICLVs had similar infectivity in several human cell lines (Supporting Information Table 1 and data not shown). We transduced the hepatocyte Huh7 cell line and measured vector genomes and GFP expression at 3 days and 2 weeks post-transduction (Fig. 1B). Initially, comparable amounts of reverse-transcribed vectors were present in IDLV- and ICLV-transduced cells. However, the frequency of GFP+ cells and the mean fluorescent intensity (MFI) of GFP were lower in IDLV-transduced cells versus ICLV-transduced cells (P < 0.01, n = 3), and this indicated less efficient expression from the former vector. Analysis of the same cultures 2 weeks after transduction showed a nearly complete loss of vector genomes and GFP+ cells in the IDLV-transduced cells, as expected for an episomal form.

LV-transduced cells versus ICLV-transduced cells (P < 0.01, n = 3), and this indicated less efficient expression from the former vector. Analysis of the same cultures 2 weeks after transduction showed a nearly complete loss of vector genomes and GFP+ cells in the IDLV-transduced cells, as expected for an episomal form. Fig. 1 IDLV performance in hepatocytes culture (A) Schematic representation of the third-generation self-inactivating vector used for these studies. SD: splicing donor site. SA: splicing acceptor site. ψ: packaging signal (including 5' portion of GAG gene (GA)). RRE: Rev responsive element. cPPT: central polypurine tract. Wpre: woodchuck hepatitis virus post-regulatory elements. 142T: miR-142 target sequence made of 4 tandem copies of a sequence perfectly complementary to miR-142. Vectors were produced with integrase-competent (ICLV) or integrase-defective (IDLV) construct. The green fluorescent protein (GFP), coagulation Factor IX (FIX) and ovalbumin (OVA) were driven from the hepatocyte-specific ET promoter composed of synthetic hepatocyte-specific enhancers and transthyretin promoter or the ubiquitously expressed phosphoglycerokinase (PGK) promoter (B) Percentage of GFP+ cells and mean fluorescence intensity of GFP (MFI; left axis) and vector copies/diploid genome (vector copy number – VCN; right axis) in Huh7 cells transduced with ICLV or IDLV at the indicated multiplicity of infection (MOI) and analyzed 3 days or 2 weeks after transduction by flow cytometry. Black bars correspond to ICLV-transduced cells, grey bars to IDLV-transduced cells. Circles show VCN. The results are presented as mean ± standard error of the mean (SEM; n = 3). (C) Representative images of human primary hepatocytes transduced as indicated or left untreated (UNT) and analyzed by live fluorescence microscopy 1 week after transduction. Nuclei are stained with Hoechst. (D) Percentage of GFP+ cells and MFI of GFP (5 fields per sample; left axis) and VCN (circles, right axis) in cultures from quiescent human primary hepatocytes. The results are presented as mean ± range (n = 2). Abbreviations: MOI, multiplicity of infection; UNT, untreated.

ransduction. Nuclei are stained with Hoechst. (D) Percentage of GFP+ cells and MFI of GFP (5 fields per sample; left axis) and VCN (circles, right axis) in cultures from quiescent human primary hepatocytes. The results are presented as mean ± range (n = 2). Abbreviations: MOI, multiplicity of infection; UNT, untreated. We then tested the same vectors on human primary hepatocytes, which do not proliferate in culture, and showed that IDLVs attained similar levels of transduced vector genomes vector copy number (VCN) 1 week after transduction but expressed GFP at substantially lower levels in comparison with matched doses of ICLVs (n = 2; Fig. 1C,D). Overall, these data indicate that IDLVs efficiently transfer episomal vector genomes into hepatocytes. However, these forms provide a less proficient substrate for the transcriptional machinery than the integrated proviral vectors.

substantially lower levels in comparison with matched doses of ICLVs (n = 2; Fig. 1C,D). Overall, these data indicate that IDLVs efficiently transfer episomal vector genomes into hepatocytes. However, these forms provide a less proficient substrate for the transcriptional machinery than the integrated proviral vectors. IDLVs Support FIX Expression From Hepatocytes in Mice To ascertain that IDLVs could be used to transduce hepatocytes in vivo, we injected increasing doses (5-, 20-, and 40-μg HIV-1 Gag p24 equivalents, n = 8) of GFP-expressing IDLVs (ET.GFP.142T) intravenously into adult mice and measured GFP expression in hepatocytes and vector DNA contents in the liver 5 weeks post-injection (Fig. 2A). A vector dose–dependent increase in hepatic transduction was apparent; the yield of GFP+ hepatocytes in the treated livers was as high as 13% according to an analysis by GFP-specific immunostaining. Subsequently, we administered matched doses (20-μg HIV-1 Gag p24 equivalents) of GFP-expressing IDLVs and ICLVs (ET.GFP.142T) intravenously to adult mice (n = 20 for IDLV mice and n = 4 for ICLV mice in three independent experiments) and measured GFP expression and vector DNA contents in the liver at different times post-injection (Fig. 2B,C). One week post-injection, GFP-expressing hepatocytes were readily detectable in IDLV-treated mice, although the frequency and the intensity were substantially lower than those observed in mice treated with matched ICLV doses. The contents of reverse-transcribed vector genomes were only slightly lower in mice treated with IDLVs versus mice treated with ICLVs and reached up to 1.5 vector copies per diploid genome. However, although the GFP expression and the vector content remained stable in the ICLV-treated mice, the frequency of GFP+ hepatocytes and the vector content progressively decreased with increasingly longer times post-injection for the IDLV-treated groups.

ICLVs and reached up to 1.5 vector copies per diploid genome. However, although the GFP expression and the vector content remained stable in the ICLV-treated mice, the frequency of GFP+ hepatocytes and the vector content progressively decreased with increasingly longer times post-injection for the IDLV-treated groups. Fig. 2 IDLV performance in the mouse liver (A) Morphometric analysis of GFP-expressing hepatocytes and VCN in the liver of mice injected with the indicated HIV-1 p24 Gag equivalents of IDLV.ET.GFP.142T and analyzed at the indicated time post-injection. GFP+ cells were identified either by immunostaining or direct fluorescence. Results are presented as mean ± SEM of n = 3 mice per time point, 5-10 optical fields scored from 5-10 non-consecutive GFP-immunostained (filled bars) or unstained (open bars) liver sections per mouse (left axis). Circles show VCN (right axis), as mean ± SEM. (B) Same analysis performed for IDLV- (grey bars) vs. ICLV- (black bars) injected mice, n = 3-7 mice per IDLV time point from 3 different experiments. (C) Representative images of unstained liver sections from mice reported in (B). Nuclei were stained with TOPRO-3. Abbreviation: UNT, untreated.

ht axis), as mean ± SEM. (B) Same analysis performed for IDLV- (grey bars) vs. ICLV- (black bars) injected mice, n = 3-7 mice per IDLV time point from 3 different experiments. (C) Representative images of unstained liver sections from mice reported in (B). Nuclei were stained with TOPRO-3. Abbreviation: UNT, untreated. We then administered matched high doses (260-μg HIV Gag p24 equivalents per mouse) of IDLV- or ICLV-expressing canine FIX complementary DNA (ET.FIX.142T) to adult hemophilia B mice (n = 15 for IDLV mice and n = 8 for ICLV mice; Fig. 3). IDLV delivery resulted in prolonged production of FIX in mouse plasma at levels considered within the therapeutic range (up to 1.5% of normal levels). In agreement with the different efficiencies of expression observed for the GFP marker, the FIX levels supported by IDLVs were up to 15-fold lower than those supported by the cognate ICLVs. Twelve weeks post-injection, the VCN in transduced livers was 0.22 ± 0.16 per diploid genome for the IDLV group and 4.36 ± 0.24 per diploid genome for the ICLV group. At this time, some of the IDLV-treated mice (n = 7) and some of the ICLV-treated ones (n = 3) were subjected to 70% partial hepatectomy. We validated the concept that partial hepatectomy resulted in an increased incorporation of bromodeoxyuridine consistent with de novo induction of hepatocyte proliferation (data not shown). In the recovering mice, FIX levels remained comparable to those measured before hepatectomy in the ICLV-treated group (Fig. 3A). In contrast, although FIX levels in the IDLV-treated mice were relatively stable before partial hepatectomy, they significantly declined shortly afterwards in the interval between weeks 12 and 20 (P < 0.05; Fig. 3B). Instead, there was no statistically significant decline in FIX expression in another IDLV-treated cohort that was not subjected to PHX in the same time interval (weeks 12-20). Nevertheless, FIX expression had become subtherapeutic after 1 year (Fig. 3C).

ards in the interval between weeks 12 and 20 (P < 0.05; Fig. 3B). Instead, there was no statistically significant decline in FIX expression in another IDLV-treated cohort that was not subjected to PHX in the same time interval (weeks 12-20). Nevertheless, FIX expression had become subtherapeutic after 1 year (Fig. 3C). Fig. 3 Functional FIX expression in hemophilic mice FIX KO mice were injected with ICLV.ET.FIX.142T (A) or IDLV.ET. FIX.142T (B) and FIX activity was monitored in the mouse plasma collected by retro-orbital bleeding. Partial hepatectomy (PHX) was performed at week 12. In both panels, grey lines correspond to untreated (UNT) FIX KO mice. (C) FIX activity in FIX KO mice injected with IDLV.ET.FIX.142T monitored for 1 year post injection and not subjected to partial hepatectomy. Abbreviations: PHX, partial hepatectomy; UNT, untreated. Overall, these data indicate that IDLVs, though less efficient than their ICLV counterparts at expressing transgenes from human and murine hepatocytes, can support therapeutically relevant levels of transgene expression in vivo. This finding warranted a further assessment of the risk/benefit ratio of the IDLV platform. The posthepatectomy drop in FIX levels observed in IDLV-treated mice suggested that FIX expression could be attributed mainly to nonintegrated vector episomes. Thus, we investigated the residual IDLV integration frequency.

vivo. This finding warranted a further assessment of the risk/benefit ratio of the IDLV platform. The posthepatectomy drop in FIX levels observed in IDLV-treated mice suggested that FIX expression could be attributed mainly to nonintegrated vector episomes. Thus, we investigated the residual IDLV integration frequency. IDLVs Integrate Only to Background Levels and With Features Incompatible With Residual Catalytic Activity of the HIV Integrase Because the D64V mutation reduces LV integration by 2 to 3 log (see also Fig. 1),13 low-level integration still occurs with this mutant. We therefore assessed residual IDLV integration in vitro and in the treated livers. First, two murine cell lines were transduced with IDLV.GFP and were subsequently cultured for 8 weeks to dilute out the episomal forms. The GFP+ cells were then selectively enriched by fluorescence-activated cell sorting (FACS). We then applied nonrestrictive and standard, 5′- and 3′-LTR–mediated LAM-PCR strategies21,22 along with 454 pyrosequencing to analyze IDLV integration in the bulk positive cell populations and in single cell–derived clones. A comprehensive, large-scale analysis of more than 800 unique, mappable IDLV ISs on the mouse genome revealed close to random genomic integration of IDLVs without any preference for gene coding regions (Fig. 4A). In contrast, the 3317 unique ICLV ISs that could be mapped to the genome showed the characteristic lentiviral IS profile with gene coding regions as preferred targets. To distinguish noncanonical integration from residual integrase activity–mediated integration, we screened our IDLV and ICLV integration data for the presence of the characteristic 5-bp direct repeat of host DNA flanking the proviral terminal CA dinucleotide as a hallmark of integrase activity. For IDLVs, we retrieved the 5′- and 3′-vector–host genome junctions in 22 instances from the bulk cell populations (statistical considerations make it extremely unlikely that such junctions could come from two distinct integration events) and in 2 instances from single cell–derived clones. Seventeen of the 24 integrants showed loss of the CA nucleotide at least at one end of the vector. All 24 integrants revealed a partial deletion of LTR and/or genomic sequences, and no 5-bp direct repeat could be detected in any of these vector-genome junctions (Fig. 4B).

in 2 instances from single cell–derived clones. Seventeen of the 24 integrants showed loss of the CA nucleotide at least at one end of the vector. All 24 integrants revealed a partial deletion of LTR and/or genomic sequences, and no 5-bp direct repeat could be detected in any of these vector-genome junctions (Fig. 4B). In contrast, 18 of 19 ICLV integrations for which both sequence ends were mapped revealed neither CA dinucleotide deletions nor LTR sequence deletions, and they harbored the typical 5-bp direct repeat without genomic deletions. In agreement with these data, screening for LTR sequence deletions in our complete IS data sets showed that one-third of all IDLV integrations had a partial loss of the LTR sequence, whereas LTR deletions accounted for only approximately 2% of all ICLV sequences. Interestingly, our LAM-PCR screening for potentially broken linear and circular vector forms, which used vector internal primers binding to the GFP transgene or vector backbone, did not show any sign of integration (data not shown). Our data provide direct molecular evidence that the background integration of D64V IDLVs is not mediated by residual catalytic activity of the mutant integrase.

rcular vector forms, which used vector internal primers binding to the GFP transgene or vector backbone, did not show any sign of integration (data not shown). Our data provide direct molecular evidence that the background integration of D64V IDLVs is not mediated by residual catalytic activity of the mutant integrase. Fig. 4 Assessment of residual IDLV integration in vitro and in the liver (see also Supporting Figure 2-4 and Table 1) (A) Distribution of integration sites determined by LAM-PCR that were located in gene coding regions of ICLV- (black bars) or IDLV- (grey bars) transduced mouse SC-1 and C1498 cells (SC-1 cells: 533 IDLV IS, 2419 ICLV IS; C1498: 271 IDLV IS, 898 ICLV IS). (B) LTR deletions (red triangles) and genomic deletions (blue triangles) identified in IDLV and ICLV integrants from which both the 5' and 3' genomic flanking region were identified by LAM-PCR. The terminal CA dinucleotide (underlined, GT at end of 5' strand LTR, complementary to CA at end of 3' strand LTR) is shown if exist. The 5 bp direct repeat of host DNA flanking the provirus is delineated in bold. The number of deleted bp in host DNA is indicated (Del.). LTR: long terminal repeat. LV: lentiviral vector. B2-PCR product gel image for primer set 1 (C) and primer set 2 (D). The arrows indicate the position of the respective PCR bands; densitometric analysis of the B2-PCR bands received with primer set 1 (E) and primer set 2 (F) in the untreated (UNT), IDLV- and ICLV-treated liver samples. Statistically significant differences (*) were found between ICLV and IDLV (t-test: P < 10-4 and P < 10-11) as well as between UNT and IDLV (t-test: P < 0.002) with both primer sets 1 and 2, respectively. Abbreviation: UNT, untreated.

E) and primer set 2 (F) in the untreated (UNT), IDLV- and ICLV-treated liver samples. Statistically significant differences (*) were found between ICLV and IDLV (t-test: P < 10-4 and P < 10-11) as well as between UNT and IDLV (t-test: P < 0.002) with both primer sets 1 and 2, respectively. Abbreviation: UNT, untreated. To determine the residual IDLV integration frequency in the treated livers, we first performed a semiquantitative nested PCR to amplify host DNA–vector junctions with short interspersed (SINE) nuclear B2 repeat sequences that are scattered throughout the mouse genome (Supporting Information Fig. 2). Although the livers from ICLV-treated mice yielded intense bands for the 5′- and 3′-vector–host genome junctions, the livers from IDLV-treated mice showed bands of much lower intensity consistent with low-level residual integration (Fig. 4C-F). We further retrieved vector sequences from the treated livers at 6 to 12 weeks post-injection by highly sensitive LAM-PCR and deep sequencing. The number of unique ISs found in the IDLV-treated liver samples was significantly lower than the number in the ICLV samples. In the IDLV liver samples (n = 16), a total of 35 unique, mappable ISs could be recovered versus 785 ISs in the ICLV samples (n = 4, P < 10−9) with an almost 100-fold lower frequency of retrieval (Table 1). Accordingly, we retrieved an excess of 2-LTR junctions from IDLV-treated liver samples versus ICLV-treated ones, and this was suggestive of the presence of episomal vector forms (2-LTR junctions: 54.5% in IDLV samples and 2.8% in ICLV samples; Table 1 and Supporting Information Fig. 4). Consistent with the in vitro IS data, the few retrieved integrations did not show enrichment for genes and frequently showed deletion in the LTR ends (25.7% in IDLV samples and 0.6% in ICLV samples; Table 1 and Supporting Information Fig. 3).

samples and 2.8% in ICLV samples; Table 1 and Supporting Information Fig. 4). Consistent with the in vitro IS data, the few retrieved integrations did not show enrichment for genes and frequently showed deletion in the LTR ends (25.7% in IDLV samples and 0.6% in ICLV samples; Table 1 and Supporting Information Fig. 3). Table 1 Deep Sequencing of ICLV- and IDLV-Transduced Liver Samples

samples and 2.8% in ICLV samples; Table 1 and Supporting Information Fig. 4). Consistent with the in vitro IS data, the few retrieved integrations did not show enrichment for genes and frequently showed deletion in the LTR ends (25.7% in IDLV samples and 0.6% in ICLV samples; Table 1 and Supporting Information Fig. 3). Table 1 Deep Sequencing of ICLV- and IDLV-Transduced Liver Samples Vector Raw Sequence Reads 1-LTR Amplicon/Internal Control 2-LTR Amplicon Reads With an LTR+, ≥20-nt Genomic Sequence Unique Mappable ISs LTR Deleted at the Genome Junction by >3 nt ICLV 2787 211 51 1688 152 2 (10, 11 nt) 2179 13 0 244 244 1 (22 nt) 3169 896 53 101 101 2822 250 258 288 288 2 (8, 23 nt) IDLV 1451 284 760 340 0 971 530 237 186 2 1227 373 82 767 0 1365 466 154 26 2 1 (21 nt) 2264 371 838 841 2 2752 400 996 1322 1 1349 221 795 308 4 1173 193 587 277 3 1 (11 nt) 2809 347 1538 502 1 1 (8 nt) 1722 215 801 421 4 2 (19, 24 nt) 3494 350 2124 916 4 2 (14, 21 nt) 3260 223 2397 500 2 3291 316 1799 621 4 1 (25 nt) 2771 330 1963 232 2 2882 275 2795 318 0 3251 241 1787 808 4 1 (8 nt) This table shows 454 sequencing results for ICLV- and IDLV-transduced liver samples. GFP-transduced samples are indicated in bold, and FIX-transduced samples are indicated in regular type. The 1-LTR amplicon and vector internal control fragments (resulting from the 3′-LTR-U3 LAM-PCR primer annealing at 5′-LTR-U3) could not be distinguished by sequencing. Notably, the 2-LTR amplicon LAM-PCR product (restriction enzyme Tsp509I, AATT) exhibited a length of only 18 bp without an LTR and linker sequence, and this resulted in the overrepresentation of sequenced 2-LTR amplicons versus the 1-LTR amplicon and the internal control (Supporting Information Fig. 4).

by sequencing. Notably, the 2-LTR amplicon LAM-PCR product (restriction enzyme Tsp509I, AATT) exhibited a length of only 18 bp without an LTR and linker sequence, and this resulted in the overrepresentation of sequenced 2-LTR amplicons versus the 1-LTR amplicon and the internal control (Supporting Information Fig. 4). Overall, these studies indicate abrogation of the HIV integrase–dependent integration in IDLVs. The background integration in treated livers exhibits molecular features reminiscent of those described for plasmids and other types of episomal DNA. This genomic IS analysis further underscores the minimal risk of insertional mutagenesis by IDLVs.

ndicate abrogation of the HIV integrase–dependent integration in IDLVs. The background integration in treated livers exhibits molecular features reminiscent of those described for plasmids and other types of episomal DNA. This genomic IS analysis further underscores the minimal risk of insertional mutagenesis by IDLVs. IDLV Delivery Can Tolerize the Recipient to Foreign Antigens We then investigated whether hepatocyte-targeted expression by IDLVs induces transgene-specific immunological tolerance. We monitored liver infiltration by lymphocyte subpopulations after the injection of IDLV.ET.GFP.142T or a control IDLV expressing GFP from a constitutive PGK promoter, which was expected to induce a GFP-specific immune response6 (n = 20 for the IDLV.142T group, n = 19 for the IDLV group, and n = 9 for the untreated group). CD8+ T cell infiltration was well detected 1 week after IDLV treatment independently of the presence or absence of miR-142 regulation. However, although the CD8+ T cells persisted at a high frequency in the livers of mice treated with the unregulated vector, they decreased substantially 3 weeks post-injection in IDLV.142T-treated mice (Fig. 5A). These findings were paralleled by the induction of GFP-specific CD8+ T cells (Fig. 5B). The frequency of CD4+CD25+FOXP3+ regulatory T cells (Tregs; FOXP3 = forkhead box P3) increased in the liver after treatment with both IDLV types 3 weeks after vector administration; however, the ratio of Tregs to CD8+ effector cells was much higher in the mice treated with miR-142-regulated vector than in control vector treated mice (Fig. 5C,D).

OXP3+ regulatory T cells (Tregs; FOXP3 = forkhead box P3) increased in the liver after treatment with both IDLV types 3 weeks after vector administration; however, the ratio of Tregs to CD8+ effector cells was much higher in the mice treated with miR-142-regulated vector than in control vector treated mice (Fig. 5C,D). Fig. 5 Kinetics of liver lymphocytes infiltrate after IDLV gene transfer Mononuclear cells (MNC) were isolated from the liver of mice injected with matched doses of hepatocyte targeted ET.GFP.142T IDLV (IDLV.142T) or control PGK.GFP IDLV (IDLV) at the indicated time post-injection (IDLV.142T n = 20, IDLV n = 19, untreated n = 9 in 3 indipendent experiment) (A) Frequency of CD8+ T, (B) GFP200-208-pentamer-positive, and (C) CD4+CD25+Foxp3+ Tregs infiltrating cells was evaluated by flow cytometry. Data are expressed as the mean % ± SD (n = 3 per group per time point; a representative experiment out of 3 is shown). (D) Ratio of T regs to CD8+ T cell. Abbreviation: CTL, cytotoxic T lymphocyte.

D8+ T, (B) GFP200-208-pentamer-positive, and (C) CD4+CD25+Foxp3+ Tregs infiltrating cells was evaluated by flow cytometry. Data are expressed as the mean % ± SD (n = 3 per group per time point; a representative experiment out of 3 is shown). (D) Ratio of T regs to CD8+ T cell. Abbreviation: CTL, cytotoxic T lymphocyte. To investigate whether active tolerance was established, we challenged the mice by intramuscular vaccination with GFP-encoding plasmids and evaluated the frequency of GFP-specific CD8+ T cells in the spleen and liver (Fig. 6A,B). Although the low frequency of GFP-specific CD8+ T cells induced in the spleens and livers of IDLV.142T-treated mice did not increase upon rechallenge, there were many more antigen-specific effectors in the control, unregulated IDLV-treated mice after the primary (IDLV) and secondary challenges (intramuscular plasmid). In addition, residual levels of vector genomes and GFP expression were still detectable in the livers of mice previously treated with IDLV.142T after rechallenge (Fig. 6B and data not shown).

fectors in the control, unregulated IDLV-treated mice after the primary (IDLV) and secondary challenges (intramuscular plasmid). In addition, residual levels of vector genomes and GFP expression were still detectable in the livers of mice previously treated with IDLV.142T after rechallenge (Fig. 6B and data not shown). Fig. 6 Transgene-specific tolerance after IDLV liver gene transfer (A) Secondary immune response to the vector-encoded antigen was assessed by frequency of IFN-γ-producing, GFP-specific CD8+ T cells in the spleen of mice subjected to antigen re-challenge (by intramuscular vaccination with GFP-encoding plasmids) 6 weeks after IDLV treatment. Single values are plotted and the mean ± SD number of GFP-specific CD8+ T cells per 106 total CD8+ T cells is shown (n = 6 per group). Representative wells are shown on top. (B) Quantification of the CD8+ GFP200-208-pentamer-positive T cells infiltrating the liver of mice treated with the indicated IDLV (IDLV.PGK n = 9; IDLV.PGK.142T n = 6; IDLV.ET.142T n = 3) after antigen re-challenge. Single values are plotted and mean ± SD is shown. Mean ± SD VCN in the liver is indicated. (C) Morphometric analysis of GFP+ hepatocytes and VCN in the liver of immune-deficient mice previously injected with ET.GFP.142T ICLV and adoptively transferred (AT) with cells derived from naïve, IDLV.PGK- and IDLV.ET.142T-treated mice (from 5A-C). Mean ± SEM of n = 3 mice (left axis). Circles show mean ± SEM VCN (right axis). A representative field is shown on top. Note that residual vector DNA in the liver of mice in which GFP-expressing hepatocytes were completely cleared can be ascribed to the persistence of vector genomes in transduced macrophages and endothelial cells that did not express GFP. (D) Plasma samples from the various groups of animals were screened for anti-FIX neutralizing antibodies using an aPTT mixing assay using a positive control of 2.6 Bethesda Units (BU) and a known negative control. Samples were assayed in a blinded fashion. Groups include FIX KO mice treated with the indicated ET.FIX.142T vector or untreated (UNT) 12 weeks after vector administration, with or without challenge with canine FIX and IFA. Only untreated mice receiving FIX + IFA had detectible neutralizing anti-FIX antibodies and all were greater than 2.6 BU. Abbreviations: Ag, antigen; ns, not significant; UNT, untreated.

icated ET.FIX.142T vector or untreated (UNT) 12 weeks after vector administration, with or without challenge with canine FIX and IFA. Only untreated mice receiving FIX + IFA had detectible neutralizing anti-FIX antibodies and all were greater than 2.6 BU. Abbreviations: Ag, antigen; ns, not significant; UNT, untreated. To further confirm that IDLV.142T is able to direct the immune system toward the induction of immune tolerance to transgene antigens, we reconstituted immune-deficient recipients through the adoptive transfer of immune cells derived from either immunized (IDLV-treated) or tolerized (IDLV.142T-treated) mice (Fig. 6C). Rag2−/−γ-chain−/− mice (Rag2 = recombination activating gene 2; γ-chain = cytokine receptor common subunit gamma) were, therefore, first transduced with hepatocyte-targeted ICLVs (ET.GFP.142T) to obtain robust GFP expression in hepatocytes. A pool of splenocytes and liver lymphocytes derived from naive, control IDLV-treated or IDLV.142T-treated animals was transferred 1 month later. Although comparable amounts of GFP-expressing hepatocytes were detected in the mice reconstituted with naive cells or cells derived from tolerant (IDLV.142T-treated) mice, a complete clearance of GFP+ hepatocytes was observed in recipient mice reconstituted with cells derived from control, unregulated IDLV-treated mice, and this was accompanied by a loss of vector DNA. These results indicate that only those cells derived from control mice exhibited the full effector potential upon adoptive transfer, whereas the effector cells derived from tolerized mice (IDLV.142T-treated) were kept in check by the induced regulatory compartment. Taken together, these data indicate that GFP-specific immunological tolerance was established.

ells derived from control mice exhibited the full effector potential upon adoptive transfer, whereas the effector cells derived from tolerized mice (IDLV.142T-treated) were kept in check by the induced regulatory compartment. Taken together, these data indicate that GFP-specific immunological tolerance was established. To assess whether immune tolerance could be induced by IDLVs toward a therapeutically relevant secreted antigen, we evaluated the anti-FIX immune response in hemophilic mice treated with FIX-encoding IDLV.142T. Activated partial thromboplastin time (aPTT)–based Bethesda assays showed no anti-FIX inhibitory activity (Fig. 6D). Importantly, when these IDLV.142T-treated mice were challenged with FIX protein in incomplete Freund adjuvant (IFA), they remained negative for anti-FIX neutralizing antibodies. In contrast, hemophilic mice that were not injected with any vector but were immunized with FIX and IFA showed the induction of neutralizing antibodies (the Bethesda titer was >2.6 Bethesda units in these mice, whereas it was below the level of detection in all recipient mice of the other cohorts). These results indicate that FIX-specific immunological tolerance was established.

vector but were immunized with FIX and IFA showed the induction of neutralizing antibodies (the Bethesda titer was >2.6 Bethesda units in these mice, whereas it was below the level of detection in all recipient mice of the other cohorts). These results indicate that FIX-specific immunological tolerance was established. IDLV Delivery Induces Antigen-Specific Tregs To determine whether the antigen specificity of liver-infiltrating Tregs is affected by IDLV.142T administration, the antigen-driven conversion of naive CD4+FOXP3− T cells into FOXP3+ Tregs was evaluated. To this end, we adoptively transferred naive CD4+ T cells obtained from double-transgenic mice for an OVA-specific T cell receptor (major histocompatibility complex II–restricted, OTII) and a GFP reporter knock-in downstream of the Foxp3 promoter. OTII CD4+ FOXP3-GFP Ly5.2 T cells were isolated, FACS-sorted to deplete FOXP3+ cells, and adoptively transferred into naive Ly5.1 recipient mice before the injection of OVA-encoding IDLVs or IDLV.142T (Fig. 7A). Three weeks later, the recipient mice were euthanized, and GFP expression, used as a surrogate marker for FOXP3, was determined in OTII Ly5.2 CD4+ T cells. A significantly higher frequency of OVA-specific Tregs was detected specifically in the IDLV.142T-treated mice versus the untreated or control, unregulated IDLV-treated mice (Fig. 7B). These results indicate that the pattern of OVA expression driven by miR-142–regulated IDLVs in the liver favored the conversion of transgene-specific naive CD4+ T cells into transgene-specific FOXP3+-induced Tregs (Fig. 7B,C).

V.142T-treated mice versus the untreated or control, unregulated IDLV-treated mice (Fig. 7B). These results indicate that the pattern of OVA expression driven by miR-142–regulated IDLVs in the liver favored the conversion of transgene-specific naive CD4+ T cells into transgene-specific FOXP3+-induced Tregs (Fig. 7B,C). Fig. 7 Induction of transgene-specific Tregs by IDLV treatment CD4+ cells isolated from OTII Ly5.2 Foxp3-GFP transgenic mice were FACS-sorted to remove GFP+ cells obtaining an homogeneous population of CD4+ non-regulatory T cells with a unique antigen specificity (OVA323-339 presented in IAb molecule). (A) Tregs-depleted OTII CD4+GFP- (2.5&times;106/mouse) were adoptively transfer intravenously into naïve C57BL/6 Ly5.1 recipient mice one day before the injection of IDLV.PGK (n = 3) or IDLV.ET.142T (n = 3) encoding for OVA. Three weeks after IDLV administration livers were harvested and infiltrating lymphocytes isolated. OVA-specific induced Tregs (iTregs) were measured as GFP+ cells gated on CD4+Ly5.2+. (B) a representative histogram and (C) mean % induced Tregs ± SEM is reported. Overall, these data indicate that hepatocyte-targeted expression of foreign antigens, including intracellular or secreted, therapeutically relevant (FIX) and model antigens (GFP and OVA), by IDLVs can result in a state of transgene-specific immunological tolerance due to a strong contraction of the transgene-specific effector compartment and the induction of the transgene-specific regulatory compartment.

ncluding intracellular or secreted, therapeutically relevant (FIX) and model antigens (GFP and OVA), by IDLVs can result in a state of transgene-specific immunological tolerance due to a strong contraction of the transgene-specific effector compartment and the induction of the transgene-specific regulatory compartment. Discussion IDLVs are emerging as an attractive platform for transgene expression for several purposes.23 This platform harnesses the pantropism and proficiency of LV transduction without relying on integration and permanent modification of the cellular genome. Here we show that IDLVs can be used to express transgenes for a window of time in the liver as long as the expression is stringently targeted to hepatocytes with transcriptional and miR-142–mediated regulation. Although the IDLV expression efficiency is lower than that observed for ICLVs, the expression levels are sufficient to induce immune tolerance and achieve prolonged therapeutic effects in a clinically relevant disease model. Moreover, the nonintegrating feature of the platform provides important safety advantages because of the low risk of genotoxicity and the reversibility of transgene expression.

e expression levels are sufficient to induce immune tolerance and achieve prolonged therapeutic effects in a clinically relevant disease model. Moreover, the nonintegrating feature of the platform provides important safety advantages because of the low risk of genotoxicity and the reversibility of transgene expression. The declining transgene expression of IDLVs in proliferating cells has been mostly ascribed to the progressive loss of episomal DNA from the cell nucleus during mitosis versus the stably integrated provirus of ICLVs. Here, we have reliably quantified the contents of reverse-transcribed vector genomes and the transgene expression level in transduced dividing or quiescent cells, and we have shown that the expression efficiency per genome copy is significantly lower for IDLVs versus ICLVs in hepatocytes. This may be due to diffusion of the episomes to nuclear areas that are not involved in active transcription, inefficient chromatin deposition, or enrichment with histone modifications typical of transcriptionally silenced chromatin.24 This is in contrast to what has been reported in other quiescent tissues, such as the retina and the central nervous system, and may indicate tissue-specific factors affecting the expression and stability of episomal IDLVs. It is conceivable that the incorporation of additional genomic elements into the IDLV backbone may improve the expression proficiency and nuclear stability.

cent tissues, such as the retina and the central nervous system, and may indicate tissue-specific factors affecting the expression and stability of episomal IDLVs. It is conceivable that the incorporation of additional genomic elements into the IDLV backbone may improve the expression proficiency and nuclear stability. We have shown that IDLV background integration occurs by mechanisms incompatible with residual activity of the mutant integrase because they often present deletions in the LTR ends and lack the typical flanking genomic repeats at the insertion site. These events may be mediated by nonhomologous end joining of linear episomes to sites of chromosomal breakage.25 Although we cannot rule out a contribution to the observed sustained FIX expression by these integrated IDLVs, the significant decline of FIX expression with the induction of hepatocyte proliferation after partial hepatectomy strongly suggests that in vivo expression is mostly mediated by the episomal forms.

reakage.25 Although we cannot rule out a contribution to the observed sustained FIX expression by these integrated IDLVs, the significant decline of FIX expression with the induction of hepatocyte proliferation after partial hepatectomy strongly suggests that in vivo expression is mostly mediated by the episomal forms. The significantly lower expression of IDLVs versus ICLVs represents a limiting factor for their application to stable therapeutic gene replacement in the liver, at least in the current design. However, IDLVs may be considered whenever reversible gene transfer is preferable (e.g., the testing of a new gene therapy approach), especially if the clinical setting imposes high safety bars in the face of existing treatment options (e.g., hemophilia) or the biological effects of gene-based delivery are difficult to predict or may entail substantial toxicity. Here we demonstrate the therapeutic potential of inducing a prolonged window of FIX expression in the plasma of hemophilic mice. Apparently, FIX expression was more prolonged than GFP expression after hepatic transduction with IDLVs. This possibly reflects the higher vector doses and/or differences in the detection limits of the assays used to quantify the expression of the respective transgene products. IDLVs may also be used for hepatic expression of therapeutic proteins, such as interferon (IFN) and other cytokines, in chronic viral hepatitis or hepatic tumors; gene-based delivery may provide therapeutic concentrations of the factor at the disease site with limited systemic exposure and only for a defined window of time.26,27

also be used for hepatic expression of therapeutic proteins, such as interferon (IFN) and other cytokines, in chronic viral hepatitis or hepatic tumors; gene-based delivery may provide therapeutic concentrations of the factor at the disease site with limited systemic exposure and only for a defined window of time.26,27 Hepatic gene transfer has been associated with the induction of immunological tolerance to the transgene product with several vector platforms.28-30 In contrast to other viral vectors used in gene therapy, most subjects are immunologically naive to the IDLV vector components, so it is unlikely that IDLV-transduced hepatocytes would be recognized by vector-specific cytotoxic T cells. Here we demonstrate a major accompanying benefit of hepatocyte-targeted IDLV gene transfer: the induction of transgene-specific Tregs and active tolerance to the transgene product. Most importantly, this response may extend beyond the duration of vector-mediated transgene expression and improve the efficacy of existing protein replacement therapy by preventing the induction of neutralizing antibodies, which represent one of the major hurdles of this treatment. Indeed, the in vivo induction of antigen-specific Tregs by microRNA-regulated IDLVs may well represent their most attractive feature to date, and it provides an intriguing contrast to the immunogenic nature of unregulated IDLV delivery, which is currently being explored for the design of improved virus-based vaccines.31,32 Although IDLVs are less efficient at expressing the transgene in hepatocytes in comparison with their integrating counterparts, they are equally efficient at inducing transgene-specific tolerance, and this suggests that the pattern of transgene expression (and not the level) plays a crucial role in directing the immune system response in this setting.

at expressing the transgene in hepatocytes in comparison with their integrating counterparts, they are equally efficient at inducing transgene-specific tolerance, and this suggests that the pattern of transgene expression (and not the level) plays a crucial role in directing the immune system response in this setting. A broad application of hepatocyte-targeted expression by ICLVs for immune modulation is currently limited by the concerns associated with integration in the target cell genome. IDLVs are advantageous for this purpose and could be exploited in inverse vaccination strategies to tolerize individuals to protein replacement or gene therapies and prevent the development of autoimmune disease in at-risk individuals.33 The present study sets the stage for further testing in preclinical models. The authors thank V. Gillijns, L. Sergi Sergi, and T. Jofra for their technical assistance; A. Lombardo, B. D. Brown, and R. Mazzieri for their scientific discussion; and I. Verma (Salk Institute, La Jolla, CA) and L. Wang (University of Pennsylvania, Philadelphia, PA) for the hemophilia B FIX-knockout mice. Abbreviations Agantigen aPTTactivated partial thromboplastin time CTLcytotoxic T lymphocyte ETenhanced transthyretin FACSfluorescence-activated cell sorting FIXcoagulation factor IX FOXP3forkhead box P3 γ-chaincytokine receptor common subunit gamma Gaggroup-specific antigen GFPgreen fluorescent protein HIVhuman immunodeficiency virus ICLVintegrase-competent lentiviral vector IDLVintegrase-defective lentiviral vector IFAincomplete Freund adjuvant IFNinterferon ISintegration site

FACSfluorescence-activated cell sorting FIXcoagulation factor IX FOXP3forkhead box P3 γ-chaincytokine receptor common subunit gamma Gaggroup-specific antigen GFPgreen fluorescent protein HIVhuman immunodeficiency virus ICLVintegrase-competent lentiviral vector IDLVintegrase-defective lentiviral vector IFAincomplete Freund adjuvant IFNinterferon ISintegration site LAM-PCRlinear amplification–mediated polymerase chain reaction LTRlong terminal repeat LVlentiviral vector MFImean fluorescence intensity miR-142microRNA 142 MOImultiplicity of infection nsnot significant OVAovalbumin PCRpolymerase chain reaction PGKphosphoglycerokinase PHXpartial hepatectomy Rag2recombination activating gene 2 Tregregulatory T cell UNTuntreated UTRuntranslated region VCNvector copy number Supplementary material

Hepatic insulin resistance associated with nonalcoholic fatty liver disease (NAFLD) is a major factor contributing to the pathogenesis of type 2 diabetes.1-3 Hepatic fat accumulation is the result of an imbalance between lipid delivery and de novo lipogenesis and lipid disposal either by oxidation or export in very low density lipoproteins (VLDL). Apolipoprotein CIII (APOC3) has been shown to be an important factor in regulating plasma triglyceride concentrations; in both rodents and human subjects increased expression of APOC3 is associated with higher plasma triglyceride levels and a null mutation with reduced triglyceride levels.4, 5 APOC3 predominantly affects VLDL-triglyceride metabolism and greater APOC3 production can increase VLDL-triglyceride production and reduce hepatic clearance of triglyceride rich lipoproteins (TGRLs).6 APOC3 stimulates VLDL synthesis and secretion in cultured hepatocytes7 and elevated circulating APOC3 levels correlate with VLDL production8 and higher postprandial hyperlipidemia in humans.9 Furthermore, APOC3 expression was proinflammatory in endothelial cells10 and adipose tissue.11 Elevated circulating APOC3 levels are associated with many diseases including metabolic syndrome,12 coronary artery disease,5 and insulin resistance9, 13 both in humans and animal models.

stprandial hyperlipidemia in humans.9 Furthermore, APOC3 expression was proinflammatory in endothelial cells10 and adipose tissue.11 Elevated circulating APOC3 levels are associated with many diseases including metabolic syndrome,12 coronary artery disease,5 and insulin resistance9, 13 both in humans and animal models. Recently, we found that common rs2854116 [T-455C] and rs2854117 [C-482T] single nucleotide polymorphisms in an insulin-response element of the promoter region in the APOC3 gene were associated with increased plasma APOC3 concentrations and an increased prevalence of NAFLD and whole-body insulin resistance in healthy lean men.9 These results suggest that increased plasma APOC3 concentrations may predispose lean individuals to NAFLD and hepatic insulin resistance, although the underlying mechanisms for hepatic lipid accumulation and insulin resistance are unclear. In order to examine these questions we examined: (1) whether transgenic mice with hepatic overexpression of human APOC3 (ApoC3Tg) are prone to develop nonalcoholic steatohepatitis (NASH); (2) whether NAFLD is associated with hepatic insulin resistance in this model; 3) the underlying mechanism by which these mice develop NAFLD; and (4) the cellular mechanisms by which NAFLD in these mice leads to hepatic insulin resistance.

human APOC3 (ApoC3Tg) are prone to develop nonalcoholic steatohepatitis (NASH); (2) whether NAFLD is associated with hepatic insulin resistance in this model; 3) the underlying mechanism by which these mice develop NAFLD; and (4) the cellular mechanisms by which NAFLD in these mice leads to hepatic insulin resistance. Materials and Methods Animals The ApoC3Tg mice, under the control of its native promoter in the C57BL/6J background,4, 6 were further backcrossed with female C57BL/6J mice over five generations in the Yale animal facility. Male ApoC3Tg mice and age-matched littermate control male mice (WT) were studied at the age of 4 to 8 months. Mice were individually housed under controlled temperature (23°C) and lighting (12/12-hour light/dark) with free access to water and fed ad libitum on regular chow (RC, 2018S, Harlan Teklad) and a high-fat diet (HFD, 55% calories from fat, TD93075, Harlan Teklad). Body composition was assessed by 1H magnetic resonance spectroscopy (Bruker BioSpin). Whole-body energy metabolism including VO2, VCO2, energy expenditure, locomoter activity, and food intake were measured for 72 hours by indirect calorimeter (CLAMS, Columbus Instrument). The study was conducted at the NIH-Yale Mouse Metabolic Phenotyping Center. All procedures were approved by the Yale University Animal Care and Use Committee.

ism including VO2, VCO2, energy expenditure, locomoter activity, and food intake were measured for 72 hours by indirect calorimeter (CLAMS, Columbus Instrument). The study was conducted at the NIH-Yale Mouse Metabolic Phenotyping Center. All procedures were approved by the Yale University Animal Care and Use Committee. Plasma Parameters Blood samples were taken from the tail vein after an overnight fast. Plasma triglyceride and nonesterified fatty acid concentrations were determined using standard commercial kits according to the manufacturer's instructions (Diagnostic Chemicals and Wako Chemicals, respectively). Plasma insulin and adiponectin levels were measured by radioimmunoassays (Linco, Billerica, MA). Plasma glucose was measured with a glucose analyzer (Beckman Coulter, Brea, CA). Human APOC3 concentrations were measured with Cobas Mira plus (Roche Diagnostic, Indianapolis, IN). Mouse cytokine array (RayBio) was conducted as described in the manufacturer's description. Liver Histology Liver sections for histology were obtained after overnight fasting, fixed in 10% formalin, and stained with hematoxylin-eosin or Masson trichrome. Histology was read by two independent pathologists blinded to experimental design and treatment groups. Briefly, steatosis (0-3), lobular inflammation (0-3), ballooning (0-3), and fibrosis (0-4) were scored separately on a scale of 0-4 as described.14, 15 NAFLD activity score is expressed as the sum of each scoring.15

Histology was read by two independent pathologists blinded to experimental design and treatment groups. Briefly, steatosis (0-3), lobular inflammation (0-3), ballooning (0-3), and fibrosis (0-4) were scored separately on a scale of 0-4 as described.14, 15 NAFLD activity score is expressed as the sum of each scoring.15 Hyperinsulinemic-Euglycemic Clamp Study The mice were maintained on an HFD for ≍2 months. After an overnight fast, [3-3H]-glucose (high-performance liquid chromatography [HPLC] purified; PerkinElmer, Waltham, MA) was infused at a rate of 0.05 μCi/min for 2 hours to assess a rate of basal glucose turnover, followed by a 140-minute hyperinsulinemic-euglycemic clamp with a primed/continuous infusion of human insulin (31.5 mU/kg prime for 3 minutes, 4.5 mU/(kg-min) infusion; Novo Nordisk, Copenhagen, DK) as described.16, 17 Rates of basal and insulin-stimulated whole-body glucose fluxes and tissue glucose uptake were determined by bolus injection of 10 μCi of 2-deoxy-D-[1-14C] glucose (PerkinElmer). Tissue Lipid Measurements Tissue triglyceride was extracted using the method of Bligh and Dyer18 and measured using a DCL Triglyceride Reagent (Diagnostic Chemicals). Hepatic ceramide species and hepatic cytosolic DAG species were measured after overnight fasting using liquid chromatography and tandem mass spectrometry as described.19 Total DAG and ceramide are expressed as the sum of individual species.

nd Dyer18 and measured using a DCL Triglyceride Reagent (Diagnostic Chemicals). Hepatic ceramide species and hepatic cytosolic DAG species were measured after overnight fasting using liquid chromatography and tandem mass spectrometry as described.19 Total DAG and ceramide are expressed as the sum of individual species. Membrane Translocation of Protein Kinase C-ε (PKCε) and Phosphorylated AKT PKCε membrane translocation in the liver protein extracts after overnight fasting and insulin phosphorylation of AKTSer473 were assessed both after overnight fasting and after the hyperinsulinemic euglycemic clamp study using the methods described.20, 21 Images were analyzed and quantified with ImageJ (NIH). Membrane translocation of PKCε was expressed as the ratio of membrane to cytosol bands.

ting and insulin phosphorylation of AKTSer473 were assessed both after overnight fasting and after the hyperinsulinemic euglycemic clamp study using the methods described.20, 21 Images were analyzed and quantified with ImageJ (NIH). Membrane translocation of PKCε was expressed as the ratio of membrane to cytosol bands. Tissue Lipid Clearance and Uptake Plasma lipid clearance and tissue uptake were assessed using mouse [3H]-labeled triolein and endogenously dual labeled chylomicrons (CM), including [3H]-retinyl ester and [14C]-triglyceride. Animals fed HFD for ≍3 months were cannulated (jugular vein) 7 days before the injection day. After collection of overnight fasting blood samples, a bolus injection of Liposyn (20%, 0.75 mL/kg, Abbott Laboratories, North Chicago, IL) conjugated with 10 μCi of [9,10-3H(N)]-triolein was administered over 1 minute through the jugular vein. Blood samples were collected at 2.5, 5, 7.5, 10, 15, 20, 30, and 60 minutes from tail vein. Endogenously dual-radiolabeled CM were prepared as described.22 After collection of blood at 0 minutes, 2 × 105 dpm of [14C]-triglyceride-CM and 6 × 105 dpm of [3H]-RE-CM were injected. Blood was collected at 2.5, 5, 10, and 15 minutes after injection. Tissue and plasma organic phase were extracted using the method of Bligh and Dyer18 and 3H/14C radioactivity was measured by beta-counter and the plasma and tissue triglyceride concentration was determined using a DCL Triglyceride Reagent (Diagnostic Chemicals). Tissue-specific triglyceride uptake rates [mg/(kg tissue-min)] were calculated by the following equation: Triglyceride uptake rate [mg/(kg tissue-min)] = plasma triglyceride (mg/ml) × {tissue radio-activity [dpm/(kg tissue-min)]/area under curve (AUC) of plasma radioactivity (dpm/ml)}.

e Reagent (Diagnostic Chemicals). Tissue-specific triglyceride uptake rates [mg/(kg tissue-min)] were calculated by the following equation: Triglyceride uptake rate [mg/(kg tissue-min)] = plasma triglyceride (mg/ml) × {tissue radio-activity [dpm/(kg tissue-min)]/area under curve (AUC) of plasma radioactivity (dpm/ml)}. Liver Triglyceride Production Mice were maintained on RC or HFD for 3 months. After an overnight fast, plasma triglyceride levels were assayed prior to (0 hours) and 1, 2, 3, and 4 hours after poloxamer407 injection using an enzymatic method (Diagnostic Chemicals). The VLDL-triglyceride production rate was calculated by the increase in plasma triglyceride level from baseline to 4 hours after 1 g/kg of poloxamer407 intraperitoneal injection and the data were expressed as micromoles of triglyceride produced per hour per kg of body weight. Plasma apolipoprotein B (apoB) levels were determined by western blotting using rabbit polyclonal antibody (Abcam) in 4%-12% gradient gel (Invitrogen). The same membrane was stained with Coomassie-blue to show equal loading. Intraperitoneal Glucose Tolerance Test (IPGTT) Weight-matched mice were maintained on an HFD for 6 weeks. After an overnight fast mice were weighed and received a bolus intraperitoneal injection of glucose (1 g/kg). Blood samples were obtained from a tail vein at baseline (0) and 15, 30, 45, 60, 90, and 120 minutes after glucose challenge for determination of plasma glucose and insulin concentrations.

n an HFD for 6 weeks. After an overnight fast mice were weighed and received a bolus intraperitoneal injection of glucose (1 g/kg). Blood samples were obtained from a tail vein at baseline (0) and 15, 30, 45, 60, 90, and 120 minutes after glucose challenge for determination of plasma glucose and insulin concentrations. Reverse-Transcription Polymerase Chain Reaction (RT-PCR) Tissue samples were obtained from mice maintained on RC or HFD for 3 months. The detailed methods, gene expression data, and sequences of PCR primers used are described in Supporting Table 2. Statistics All values are expressed as mean ± SEM. The significance of the differences in mean values among two groups was evaluated by two-tailed unpaired Student's t tests. More than three groups were evaluated by analysis of variance (ANOVA) followed by post-hoc analysis using Bonferroni's Multiple Comparison Test. P < 0.05 was considered significant.

as mean ± SEM. The significance of the differences in mean values among two groups was evaluated by two-tailed unpaired Student's t tests. More than three groups were evaluated by analysis of variance (ANOVA) followed by post-hoc analysis using Bonferroni's Multiple Comparison Test. P < 0.05 was considered significant. Results APOC3 Overexpression Promotes the Development of Diet-Induced Hepatic Steatosis and NAFLD in Mice Increases in plasma human APOC3 were associated with significantly higher levels of plasma triglyceride in ApoC3Tg mice (Table 1). Both on RC and HFD, body weight, body composition, whole-body energy metabolism (VO2, VCO2, respiratory quotient, energy expenditure, food intake, and locomotor activity) were identical (Table 1) between WT and ApoC3Tg mice. Both groups showed identical increases in body weight and there were no differences in body composition while consuming a HFD (Table 1). Fasting plasma fatty acid concentrations were ≍70% higher in ApoC3Tg mice, compared with WT mice (Table 1). The ApoC3Tg mice fed the HFD for ≍3 months exhibited more severe hepatic steatosis with lipid accumulation around the Zone 3 area (Fig. 1A) than the WT, accompanied by ≍60% increased liver triglyceride levels (Fig. 1C), which was similar between WT and ApoC3Tg mice on RC (Supporting Fig. 1). Ballooned cells, characterized by ≍2 times the size of adjacent hepatocytes and pyknotic nucleus, were increased in ApoC3Tg mice compared with WT mice after HFD feeding (Fig. 1A). The increased liver triglyceride content in the ApoC3Tg mice was associated with an ≍80% increase in serum aspartate aminotransferase levels and a tendency for higher serum alanine aminotransferase levels (Fig. 1D). Fasting plasma proinflammatory cytokines were markedly increased both in WT and ApoC3Tg mice fed the HFD compared with RC fed mice and there were further increases in plasma tumor necrosis factor alpha (TNF-α) and interferon gamma (INF-γ) (both ≍70%) in ApoC3Tg mice compared with the WT mice after HFD feeding (Fig. 1E, Supporting Fig. 2). We applied a system of histological scoring designed for use in humans to systematically compare the differences in histology between the two groups. The NAFLD activity score (NAS) of ApoC3Tg mice was ≍4 (Fig. 1B) after HFD feeding, and NAS of ≥5 has been correlated with a diagnosis of NASH.15 In contrast, there were no significant differences in plasma concentrations of interleukin (IL)-1β, IL-10, IL-12p70, IL-8, and IL-6 levels (Supporting Fig.

een the two groups. The NAFLD activity score (NAS) of ApoC3Tg mice was ≍4 (Fig. 1B) after HFD feeding, and NAS of ≥5 has been correlated with a diagnosis of NASH.15 In contrast, there were no significant differences in plasma concentrations of interleukin (IL)-1β, IL-10, IL-12p70, IL-8, and IL-6 levels (Supporting Fig. 2A) or hepatic nuclear factor kappaB (NF-κB) p65, IkappaBα, or c-Jun N-terminal kinase (JNK) phosphorylation (Supporting Fig. 2A,B) between genotypes fed the HFD. Table 1 Basal Characterization for Animals RC HFD

een the two groups. The NAFLD activity score (NAS) of ApoC3Tg mice was ≍4 (Fig. 1B) after HFD feeding, and NAS of ≥5 has been correlated with a diagnosis of NASH.15 In contrast, there were no significant differences in plasma concentrations of interleukin (IL)-1β, IL-10, IL-12p70, IL-8, and IL-6 levels (Supporting Fig. 2A) or hepatic nuclear factor kappaB (NF-κB) p65, IkappaBα, or c-Jun N-terminal kinase (JNK) phosphorylation (Supporting Fig. 2A,B) between genotypes fed the HFD. Table 1 Basal Characterization for Animals RC HFD WT ApoC3Tg WT ApoC3Tg Body weight (g) 29.6 ± 0.3 (n=8) 29.7 ± 0.4 (n=8) 39.5 ± 1.2† (n=8) 40.7 ± 1.1† (n=8) Liver weight (g) 0.93 ± 0.03 (n=4) 1.0 ± 0.06 (n=4) 1.72 ± 0.12† (n=6) 2.18 ± 0.18† (n=6) Epididymal WAT weight (g) 0.28 ± 0.05 (n=4) 0.23 ± 0.03 (n=4) 2.9 ± 0.23† (n=6) 3.1 ± 0.26† (n=6) Body fat (%) 8.98 ± 0.9 (n=8) 9.7 ± 0.74 (n=8) 25.98 ± 1.3† (n=8) 26.31 ± 1.1† (n=8) Lean body mass (%) 74.45 ± 0.85 (n=8) 74.29 ± 0.71 (n=8) 62.76 ± 1.43† (n=8) 63.35 ± 0.98† (n=8) Fasting glucose (mg/dL) 118.3 ± 9.4 (n=6) 122.7 ± 11.4 (n=6) 156.4 ± 6.7† (n=8) 149.1 ± 10.9† (n=8) Fasting insulin (μU/mL) 8.3 ± 2.0 (n=6) 11.0 ± 1.5 (n=6) 30.3 ± 6.1† (n=8) 37.3 ± 9.9† (n=8) Fasting NEFA (mg/dL) 1.07 ± 0.11 (n=4) 1.80 ± 0.15** (n=4) 0.64 ± 0.05† (n=6) 0.97 ± 0.08*,† (n=6) Fasting plasma triglyceride (mg/dL) 92.3 ± 6.8 (n=8) 1342 ± 221*** (n=8) 88.1 ± 14.1 (n=12) 617 ± 96.3***,† (n=12) Human plasma APOC3 (mg/dL) ND 209.2 ± 54.1 (n=8) ND 106.4 ± 18.9† (n=8) Fasting plasma adiponectin (μg/mL) 39.9 ± 3.5 (n=4) 31.4 ± 2.3 (n=4) 26.2 ± 1.9† (n=4) 31.9 ± 4.0 (n=4) Whole body VO2 [Kcal/(kg-hr)] 3077 ± 52 (n=8) 3039 ± 56 (n=8) 2775 ± 77† (n=8) 2743 ± 52† (n=8) Whole body VCO2 [Kcal/(kg-hr)] 2662 ± 60 (n=8) 2627 ± 62 (n=8) 2121 ± 57† (n=8) 2112 ± 44† (n=8) Food intake [Kcal/(kg-hr)] 17.8 ± 1.2 (n=8) 17.5 ± 1.3 (n=8) 12.67 ± 0.52† (n=8) 13.72 ± 1.28† (n=8) Respiratory quotient 0.86 ± 0.01 (n=8) 0.86 ± 0.01 (n=8) 0.76 ± 0.01† (n=8) 0.77 ± 0.01† (n=8) Activity (counts) 143.6 ± 14.5 (n=8) 176 ± 25.5 (n=8) 91.0 ± 21.9 (n=8) 110.1 ± 17.4 (n=8) WT and ApoC3Tg mice fed a RC and high fat diet for 2-3 month and fasted overnight prior to experiments.

.52† (n=8) 13.72 ± 1.28† (n=8) Respiratory quotient 0.86 ± 0.01 (n=8) 0.86 ± 0.01 (n=8) 0.76 ± 0.01† (n=8) 0.77 ± 0.01† (n=8) Activity (counts) 143.6 ± 14.5 (n=8) 176 ± 25.5 (n=8) 91.0 ± 21.9 (n=8) 110.1 ± 17.4 (n=8) WT and ApoC3Tg mice fed a RC and high fat diet for 2-3 month and fasted overnight prior to experiments. Data are expressed as mean values ± SEM. * P < 0.05, ** P < 0.01, *** P < 0.001 by unpaired Student's t test compared to WT versus ApoC3Tg group. † P < 0.05 by unpaired Student's t test compared to RC versus HFD in each group. ND, not detected. NEFA, nonesterified fatty acid. Fig. 1 APOC3 promote diet-induced hepatic steatosis and liver damage. Animals fed an RC and HFD for 2-3 months and fasted overnight prior to taking liver and blood samples. (A) Representative histological section stained with Masson-trichrome staining, examined in ×400 light microscopy field after HFD feeding for 2-3 months (insert ×200 magnifications). Arrow indicates ballooned hepatocyte. (B) Histological scoring of NAFLD in liver section after HFD feeding for 3 months. (C) Triglyceride level in the liver of WT and ApoC3Tg mice after RC and 2-3 months of HFD feeding (n > 10). (D) Plasma ALT and AST concentration after 3 months of HFD feeding (n = 4). (E) Plasma TNF-α concentration in the WT and ApoC3Tg mice after RC and 3 months of HFD feeding (n = 3, equal amounts of 4 mice plasma were pooled into each sample). Data are expressed as mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 by Student's t test for (B,D) and by ANOVA with post-hoc analysis for (C,E).

4). (E) Plasma TNF-α concentration in the WT and ApoC3Tg mice after RC and 3 months of HFD feeding (n = 3, equal amounts of 4 mice plasma were pooled into each sample). Data are expressed as mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 by Student's t test for (B,D) and by ANOVA with post-hoc analysis for (C,E). ApoC3Tg Mice Manifest Hepatic Insulin Resistance In order to assess the tissue-specific contributions of liver, skeletal muscle, and adipose tissue to whole-body insulin resistance in the HFD-fed ApoC3Tg mice, we performed hyperinsulinemic-euglycemic clamp studies combined with 3H/14C-labeled glucose infusions.16, 17 Overnight fasting plasma glucose and insulin concentrations were similar between the two groups (Table 1) as were basal rates of endogenous glucose production (EGP) (Fig. 2E). During the hyperinsulinemic-euglycemic clamp the glucose infusion rates required to maintain euglycemia (Fig. 2A) in the ApoC3Tg mice were 34% lower than those of WT mice (Fig. 2B), indicating that the ApoC3Tg mice were more insulin resistant than the age-weight-matched WT mice. The ApoC3Tg mice displayed marked hepatic insulin resistance as reflected by the lack of suppression of EGP during the hyperinsulinemic-euglycemic clamp compared with the WT mice (Fig. 2E,F). Rates of insulin-stimulated whole-body glucose uptake were also slightly decreased by 13% in the ApoC3Tg mice compared with the WT mice (Fig. 3C), whereas muscle 2-deoxy-glucose uptake was not significantly different (Fig. 2D). Fasting plasma fatty acid concentrations decreased in a similar fashion during the hyperinsulinemic-euglycemic clamp study in both groups (% suppression: WT, 42% ± 7%; ApoC3Tg, 45% ± 6%), indicating that adipose tissue insulin responsiveness was similar between the two groups. Plasma insulin concentrations were similar between WT and ApoC3Tg mice during the clamp study (Supporting Table 1).

uring the hyperinsulinemic-euglycemic clamp study in both groups (% suppression: WT, 42% ± 7%; ApoC3Tg, 45% ± 6%), indicating that adipose tissue insulin responsiveness was similar between the two groups. Plasma insulin concentrations were similar between WT and ApoC3Tg mice during the clamp study (Supporting Table 1). Fig. 2 Hepatic steatosis in ApoC3Tg mice is associated with hepatic insulin resistance. (A) Plasma glucose concentration and glucose infusion rate in WT (n = 6) and ApoC3Tg (n = 7) mice during hyperinsulinemic-euglycemic clamp study following 2 months of HFD. (B,C) Glucose infusion rate and whole-body glucose uptake rate during steady status. (D) Insulin stimulated muscle 2-deoxy-D-[1-14C] glucose (2DOG) uptake in gastrocnemius muscle. (E) Basal EGP and insulin suppressed (clamped) EGP and (F) percentage of suppression during the hyperinsulinemic-euglycemic clamp study. Data are expressed as mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 by Student's t test except (A) (2-way ANOVA with post-hoc analysis).

e (2DOG) uptake in gastrocnemius muscle. (E) Basal EGP and insulin suppressed (clamped) EGP and (F) percentage of suppression during the hyperinsulinemic-euglycemic clamp study. Data are expressed as mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 by Student's t test except (A) (2-way ANOVA with post-hoc analysis). Fig. 3 Hepatic insulin resistance in the ApoC3Tg mice is associated with increased hepatic diacylglycerol content and PKCε activation. Animals fed an RC and HFD for 2 months and fasted overnight prior to experiments. (A) Intracellular cytosolic diacylglycerol after overnight fasting. (B) Membrane translocation of PKCε was analyzed in the liver of WT and ApoC3Tg mice after hyperinsulinemic-euglycemic clamp study. (C) Akt phosphorylation on Ser473 was analyzed in the liver of overnight-fasted and hyperinsulinemic-euglycemic clamped WT and ApoC3Tg mice after 2 months of HFD feeding. Data are expressed as mean ± SEM. (A), n = 6; (B), n = 4; (C), n = 6-7 per group. *P < 0.05, **P < 0.01 by Student's t test.

-euglycemic clamp study. (C) Akt phosphorylation on Ser473 was analyzed in the liver of overnight-fasted and hyperinsulinemic-euglycemic clamped WT and ApoC3Tg mice after 2 months of HFD feeding. Data are expressed as mean ± SEM. (A), n = 6; (B), n = 4; (C), n = 6-7 per group. *P < 0.05, **P < 0.01 by Student's t test. ApoC3Tg Mice Have Increased Hepatic DAG Content, PKCε Activation, and Hepatic Insulin Resistance Increased hepatic triglyceride content and hepatic insulin resistance was associated with ≍45% increase in hepatic cytosolic DAG content in the liver from ApoC3Tg mice compared with the WT mice fed the HFD (Fig. 3A), whereas hepatic ceramide content was not different between genotypes (Supporting Fig. 3C,D). Furthermore, this increase in liver DAG content was associated with an ≍90% increase in the membrane/cytosol ratio of PKCε (Fig. 3B), reflecting an increase in PKCε activity, and a 32% reduction in insulin-stimulated AKT2 activity in the livers of ApoC3Tg mice compared with WT mice (Fig. 3C). Taken together, these data are consistent with the hypothesis that increased plasma levels of APOC3 predispose animals to NAFLD and hepatic insulin resistance, which in turn can be attributed to DAG-induced PKCε activation resulting in decreased insulin signaling.

livers of ApoC3Tg mice compared with WT mice (Fig. 3C). Taken together, these data are consistent with the hypothesis that increased plasma levels of APOC3 predispose animals to NAFLD and hepatic insulin resistance, which in turn can be attributed to DAG-induced PKCε activation resulting in decreased insulin signaling. ApoC3Tg Mice Display Increased Hepatic Triglyceride Uptake To address the underlying mechanism for the increased hepatic triglyceride content in the ApoC3Tg mice we assessed hepatic triglyceride uptake by intravenous injection of dual-labeled chylomicrons containing [3H]-retinyl ester (RE) and [14C]-triglyceride into ApoC3Tg and age-weight-matched WT littermate mice after 3 months of HFD feeding. Plasma decay of tracers in WT mice was 0.55 ± 0.15 pools/min for triglyceride and 0.19 pools/min for RE. In ApoC3Tg mice, the plasma clearance of triglyceride was reduced to 0.27 pools/min, whereas that of RE was not significantly altered (0.17 pools/min). These data suggest that the major effect of the APOC3 transgene was on the initial clearance of triglyceride.

0.15 pools/min for triglyceride and 0.19 pools/min for RE. In ApoC3Tg mice, the plasma clearance of triglyceride was reduced to 0.27 pools/min, whereas that of RE was not significantly altered (0.17 pools/min). These data suggest that the major effect of the APOC3 transgene was on the initial clearance of triglyceride. We then assessed the uptake of tracer into tissues with a focus on the ratio of the two labels. The uptake of both [3H]-RE-CM and [14C]-triglyceride-CM was increased by approximately twofold in the liver of ApoC3Tg mice, and the liver was the major site of TGRL uptake, compared with skeletal muscle and white adipose tissues (Fig. 4A,B). Although, in WT mice, liver triglyceride/RE uptake was 1.00 ± 0.05, indicative of greater fatty acid uptake due to lipolysis, this ratio was reduced to 0.74 ± 0.11 in the ApoC3Tg mice. Triglyceride/RE uptake was 19.4 ± 9.67 in skeletal muscle from WT mice, as would be expected due to greater fatty acid uptake after lipolysis. This ratio decreased to 6.1 ± 2.12 in the ApoC3Tg mice. APOC3 has been reported to alter TGRL metabolism either by inhibiting lipoprotein lipase (LpL) activity or interfering with receptor-mediated hepatic uptake of remnant lipoproteins.23 Our data suggest a block in peripheral triglyceride lipolysis consistent with LpL inhibition.

creased to 6.1 ± 2.12 in the ApoC3Tg mice. APOC3 has been reported to alter TGRL metabolism either by inhibiting lipoprotein lipase (LpL) activity or interfering with receptor-mediated hepatic uptake of remnant lipoproteins.23 Our data suggest a block in peripheral triglyceride lipolysis consistent with LpL inhibition. Fig. 4 Amount of hepatic triglyceride uptake was increased in ApoC3Tg mice both on RC and HFD. Animals fed an HFD for 3 months and fasted overnight prior to experiments (A,B). After collection of blood at 0 minutes, then endogenously dual-radiolabeled chylomicron, 6 × 105 dpm of [3H]-RE-CM and 2 × 105 dpm of [14C]-triglyceride-CM were injected. Blood was collected at 2.5, 5, 10, and 15 minutes after injection (A,B). With a separate batch of animals fed RC or HFD for 3 months, after 6 hours fasting a bolus injection of 9 μCi of [3H]-triolein was intravenously administered and blood samples were collected at 2.5, 5, 7.5, 10, 15, 20, 30, and 60 minutes from tail vein (C). Tissue-specific uptake rate of (A) [3H]-RE-CM (B) [14C]-triglyceride-CM were assessed in various tissues and (C) hepatic [3H]-triolein uptake rate were assessed after RC and HFD. N = 4-5. *P < 0.05; **P < 0.01; ns, not significant by Student's t test. CM, chylomicron. RE, retinyl ester. WAT, epididymal white adipose tissue.

e-specific uptake rate of (A) [3H]-RE-CM (B) [14C]-triglyceride-CM were assessed in various tissues and (C) hepatic [3H]-triolein uptake rate were assessed after RC and HFD. N = 4-5. *P < 0.05; **P < 0.01; ns, not significant by Student's t test. CM, chylomicron. RE, retinyl ester. WAT, epididymal white adipose tissue. Because ApoC3Tg mice developed hepatic steatosis when fed an HFD diet but not when fed a regular chow diet, we also compared hepatic triglyceride uptake in ApoC3Tg and age-weight-matched WT littermates fed the RC and HFD. Surprisingly, we found that hepatic triglyceride uptake was increased in ApoC3Tg mice fed either the RC or HFD compared with WT mice, and that there was no further increase in hepatic triglyceride uptake in the HFD-fed ApoC3Tg mice compared with the RC-fed ApoC3Tg mice (Fig. 4C).

ched WT littermates fed the RC and HFD. Surprisingly, we found that hepatic triglyceride uptake was increased in ApoC3Tg mice fed either the RC or HFD compared with WT mice, and that there was no further increase in hepatic triglyceride uptake in the HFD-fed ApoC3Tg mice compared with the RC-fed ApoC3Tg mice (Fig. 4C). Hepatic VLDL-Triglyceride Secretion Was Decreased in ApoC3Tg Mice During HFD Next, we assessed VLDL-triglyceride secretion in WT and ApoC3Tg mice fed either the RC or HFD by injection of the lipoprotein lipase inhibitor poloxamer407. APOC3 has been shown to stimulate VLDL synthesis and secretion in cultured cells7 and elevated circulating APOC3 levels correlate with VLDL production8 and postprandial hyperlipidemia in humans.9 Consistent with these findings, we observed an ≍70% increase in VLDL-triglyceride production in ApoC3Tg mice fed a RC diet (Fig. 5A,B), which was associated with marked increases in circulating apoB48 concentrations in ApoC3Tg mice (Fig. 5C). In contrast, we observed a 53% reduction in VLDL-triglyceride production rate in HFD-fed ApoC3Tg mice compared with the HFD-fed WT mice (Fig. 5B). The decreased VLDL-triglyceride production was accompanied by a marked decrease in circulating apoB100 level as well as a slight decrease in apoB48 concentrations in ApoC3Tg mice during HFD feeding (Fig. 5C). Liver microsomal triglyceride transfer protein (MTP) expression (Supporting Fig. 4) and messenger RNA (mRNA) expression of MTP (Supporting Table 2) were not significantly different between WT and ApoC3Tg mice on HFD. Thus, the greater uptake of TGRLs in the ApoC3Tg mice was not compensated by increased liver secretion of apoB-containing lipoproteins.

ansfer protein (MTP) expression (Supporting Fig. 4) and messenger RNA (mRNA) expression of MTP (Supporting Table 2) were not significantly different between WT and ApoC3Tg mice on HFD. Thus, the greater uptake of TGRLs in the ApoC3Tg mice was not compensated by increased liver secretion of apoB-containing lipoproteins. Fig. 5 Hepatic VLDL-triglyceride secretion was decreased in ApoC3Tg mice during HFD. Animals fed an RC and HFD for 3 months and fasted overnight prior to experiments. (A) Net increase of plasma triglyceride concentration after poloxamer407 (p-407) injection from basal (0 minutes, before p-407 injection) (n = 4-6). (B) Triglyceride production rate during 4 hours after p-407 injection, expressed as micromoles of triglyceride produced per hour per kg of body weight (n = 4-6). (C) Plasma apoB determined by western blotting using 4%-12% gradient gel. The same membrane stained with Coomassie-blue was provided as loading control (n = 3, equal amount of 4 mice plasma were pooled into each sample). Data are expressed as mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 by two-way ANOVA post-hoc analysis (A) and **P < 0.01 by two-tailed t test (B,C). ns, not significant.

same membrane stained with Coomassie-blue was provided as loading control (n = 3, equal amount of 4 mice plasma were pooled into each sample). Data are expressed as mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 by two-way ANOVA post-hoc analysis (A) and **P < 0.01 by two-tailed t test (B,C). ns, not significant. Other possible contributors to greater hepatic triglyceride content are de novo synthesis and reduced fatty acid oxidation. Interestingly, mRNA expression of hepatic sterol regulatory element binding protein-1c (SREBP-1c) and LpL were significantly higher in ApoC3Tg mice compared with WT on HFD (Supporting Table 2), indicating a possible contribution of increased hepatic lipogenesis to the increased hepatic lipid accumulation in ApoC3Tg mice during HFD feeding. There were no differences in whole-body oxygen consumption, respiratory quotients (Table 1), and liver mRNA expression of peroxisome proliferator-activated receptor-α (PPAR-α) and carnitine palmitoyltransferase 1 (CPT1), between WT and ApoC3Tg mice both on RC and HFD (Supporting Table 2), suggesting that alterations in hepatic fatty acid oxidation were not likely responsible for the observed net increases in hepatic triglyceride content observed in the HFD-fed ApoC3Tg mice.

ceptor-α (PPAR-α) and carnitine palmitoyltransferase 1 (CPT1), between WT and ApoC3Tg mice both on RC and HFD (Supporting Table 2), suggesting that alterations in hepatic fatty acid oxidation were not likely responsible for the observed net increases in hepatic triglyceride content observed in the HFD-fed ApoC3Tg mice. ApoC3Tg Mice Have Reduced Hepatic ApoB100 Expression Secondary to Postprandial Hyperinsulinemia Recent studies have demonstrated that insulin suppresses apoB secretion in cultured hepatocytes24 and in vivo.25-27 In order to examine the potential role of hyperinsulinemia in causing reduced hepatic apoB100 expression in ApoC3Tg mice, we measured hepatic apoB100 protein levels in HFD-fed ApoC3Tg and WT mice at the end of the hyperinsulinemic-euglycemic clamp and found that insulin rapidly suppressed hepatic apoB100 expression both in the liver of WT and ApoC3Tg mice (Fig. 6A). We next examined whether ApoC3Tg mice exhibited increased plasma insulin concentrations following an intraperitoneal glucose tolerance test. Plasma glucose concentrations were similar between HFD-fed WT and ApoC3Tg mice after glucose challenge (Fig. 5B). In contrast, ApoC3Tg mice had an ≍120% increase in peak and an ≍70% increase in the AUC of glucose-stimulated plasma insulin concentrations compared with WT mice (Fig. 5C), reflecting whole-body insulin resistance as well as increased postprandial insulin secretion in ApoC3Tg mice on HFD. Taken together, these data suggest that the increase in net hepatic triglyceride content in the HFD-fed ApoC3Tg mice can likely be attributed to both an increase in hepatic triglyceride uptake in combination with decreased hepatic VLDL secretion, due to suppression of hepatic apoB expression from chronic postprandial hyperinsulinemia.

these data suggest that the increase in net hepatic triglyceride content in the HFD-fed ApoC3Tg mice can likely be attributed to both an increase in hepatic triglyceride uptake in combination with decreased hepatic VLDL secretion, due to suppression of hepatic apoB expression from chronic postprandial hyperinsulinemia. Fig. 6 Insulin suppressed hepatic apoB100 expression and postprandial insulin secretion was increased in ApoC3Tg mice after HFD feeding. (A) Hepatic apoB100 protein expression in the liver of WT and ApoC3Tg mice from overnight fasted basal and after hyperinsulinemic-euglycemic clamp study following 2 months of HFD feeding (n = 3 for RC, n = 5 for HFD group). (B) Plasma glucose concentration and area under curve of glucose, (C) plasma insulin level and area under curve of insulin after 1 mg/kg of glucose challenge during IPGTT in WT, and ApoC3Tg mice after 6 weeks of HFD feeding (n = 7). Data are expressed as mean ± SEM. †P < 0.05 by two-way ANOVA analysis. *P < 0.05 and **P < 0.01 by two-tailed t test. ns, not significant.

ve of glucose, (C) plasma insulin level and area under curve of insulin after 1 mg/kg of glucose challenge during IPGTT in WT, and ApoC3Tg mice after 6 weeks of HFD feeding (n = 7). Data are expressed as mean ± SEM. †P < 0.05 by two-way ANOVA analysis. *P < 0.05 and **P < 0.01 by two-tailed t test. ns, not significant. Discussion Previous studies have found that whole-body insulin-mediated glucose disposal was unchanged without any ectopic lipid deposition in ApoC3Tg mice fed an RC diet.6, 28 In contrast to these findings, we found that when ApoC3Tg mice were challenged with an HFD they developed NASH associated with marked hepatic insulin resistance compared with WT littermate mice fed the same diet. These data provide important genetic verification of a recent study showing that healthy lean individuals carrying variants in the insulin response element of the APOC3 gene, leading to increased APOC3 concentrations in plasma, are predisposed to developing NAFLD associated with insulin resistance.9

fed the same diet. These data provide important genetic verification of a recent study showing that healthy lean individuals carrying variants in the insulin response element of the APOC3 gene, leading to increased APOC3 concentrations in plasma, are predisposed to developing NAFLD associated with insulin resistance.9 Interestingly, we found that ApoC3Tg mice were more prone to develop hepatic steatohepatitis as reflected by the histologic NAFLD score and increased serum aspartate aminotransferase (AST), TNF-α, and INF-γ concentrations. These data are consistent with previous data demonstrating that APOC3 can cause inflammation in various cells, including endothelial cells,10 monocytes,29 and adipose tissue,11 and support the hypothesis that increased plasma levels of APOC3 may also make individuals more prone to develop NASH in addition to simple steatosis.9 However, the increased circulating cytokines were not associated with hepatic NF-κB p65, JNK phosphorylation, and ceramide content, indicating the hepatic inflammatory signals may not be the major factors for aggravating the hepatic insulin resistance in ApoC3Tg mice. In contrast, we found that the NAFLD and hepatic insulin resistance could be attributed to increased hepatocellular DAG content and activation of PKCε, which we have previously demonstrated causes decreased insulin signaling at the level of the insulin receptor kinase.20, 30 This finding is in contrast to another recent report of a PNPLA gene variant that was associated with NAFLD but was not associated with insulin resistance.31

ar DAG content and activation of PKCε, which we have previously demonstrated causes decreased insulin signaling at the level of the insulin receptor kinase.20, 30 This finding is in contrast to another recent report of a PNPLA gene variant that was associated with NAFLD but was not associated with insulin resistance.31 Furthermore, we found that hepatic triglyceride uptake was surprisingly increased in ApoC3Tg mice fed either the RC or HFD despite the absence of hepatic steatosis when the ApoC3Tg mice were fed the RC diet. We went on to show that ApoC3Tg mice fed the HFD have reduced hepatic VLDL production, which could be attributed to decreased hepatic apoB100 expression from chronic hyperinsulinemia. Taken together, these findings suggest that increased hepatic expression of APOC3 promotes increased hepatic triglyceride uptake that is compensated for by increased hepatic VLDL production in ApoC3Tg mice fed RC. In contrast, ApoC3Tg mice fed an HFD are unable to compensate with increased VLDL production due to postprandial hyperinsulinemia, leading to suppression of hepatic apoB100 and VLDL production and to net hepatic triglyceride accumulation. In addition, we also found an ≍6-fold and ≍2-fold increase in expression of LpL and SREBP1c mRNA, respectively, in the livers of the ApoC3Tg mice following HFD, which may also have contributed to the development of hepatic steatosis in these mice. These findings may also explain the mechanism of hepatic lipid accumulation in the other mouse models of hepatic steatosis and steatohepatitis that showed severe hyperlipidemia such as apolipoprotein E (APOE)-deficient mice and LDL receptor-deficient mice.32-34

ted to the development of hepatic steatosis in these mice. These findings may also explain the mechanism of hepatic lipid accumulation in the other mouse models of hepatic steatosis and steatohepatitis that showed severe hyperlipidemia such as apolipoprotein E (APOE)-deficient mice and LDL receptor-deficient mice.32-34 Many of the postulated effects of APOC3 on peripheral and tissue lipid metabolism were confirmed in our studies. As had been reported,6, 35 using endogenously labeled chylomicrons we confirmed that one reason for the hypertriglyceridemia in these animals is reduced plasma clearance. In vitro studies by these investigators found that VLDL from the APOC3 transgenic mice were lipolyzed normally by LpL but had reduced uptake into cells6; this latter finding is what one would expect from nonlipolyzed TGRLs that interact poorly with lipoprotein receptors. Our data showing that skeletal muscle uptake of triglyceride is much more impaired than its uptake of RE is more consistent with APOC3 functioning as an inhibitor of lipolysis. In addition, the observations that APOC3 overexpression leads to hypertriglyceridemia in APOE knockout mice36 and causes less hypertriglyceridemia in apoB48 only mice that are more dependent on non-LDL receptor-mediated lipid uptake37 are consistent with LpL inhibition.

h APOC3 functioning as an inhibitor of lipolysis. In addition, the observations that APOC3 overexpression leads to hypertriglyceridemia in APOE knockout mice36 and causes less hypertriglyceridemia in apoB48 only mice that are more dependent on non-LDL receptor-mediated lipid uptake37 are consistent with LpL inhibition. In summary, we found that transgenic mice with hepatic overexpression of human APOC3 were more prone to develop hepatic steatosis associated with hepatic insulin resistance compared with WT littermates fed the same HFD diet. This study provides strong genetic evidence in support of the hypothesis that increased plasma APOC3 concentrations predispose lean individuals to NAFLD associated with hepatic insulin resistance.9 Importantly, the development of NAFLD associated with hepatic insulin resistance in response to dietary alterations exemplifies how gene-environment interactions contribute to the pathogenesis of complex phenotypes. We thank to Irena Ignatova-Todorova, Xiaoxian Ma, Mario Kahn, Christopher M. Carmean for expert technical assistance withthe studies and Dr. Jan Breslow (Rockefeller University, New York, NY) for providing the ApoC3Tg mice. Abbreviations apoBapolipoprotein B APOC3apolipoprotein CIII ApoC3Tgtransgenic mice with hepatic overexpression of human APOC3 APOEapolipoprotein E AUCarea under curve CMchylomicron DAGdiacylglycerol EGPendogenous glucose production HFDhigh-fat diet IPGTTintraperitoneal glucose tolerance test JNKc-Jun N-terminal kinase LpLlipoprotein lipase MTPmicrosomal triglyceride transfer protein NAFLDnonalcoholic fatty liver disease NASHnonalcoholic steatohepatitis

ApoC3Tgtransgenic mice with hepatic overexpression of human APOC3 APOEapolipoprotein E AUCarea under curve CMchylomicron DAGdiacylglycerol EGPendogenous glucose production HFDhigh-fat diet IPGTTintraperitoneal glucose tolerance test JNKc-Jun N-terminal kinase LpLlipoprotein lipase MTPmicrosomal triglyceride transfer protein NAFLDnonalcoholic fatty liver disease NASHnonalcoholic steatohepatitis NASNAFLD activity score NF-κBnuclear factor kappaB PKCεprotein kinase C-ε PPAR-αperoxisome proliferator-activated receptor RCregular chow REretinyl ester SREBP-1csterol regulatory element binding protein-1c TGRLstriglyceride rich lipoproteins TNF-αtumor necrosis factor-alpha VLDLvery low density lipoproteins WTwildtype littermates Supplementary material Additional Supporting Information may be found in the online version of this article.

Hepatic steatosis is estimated to affect >20% of the Western population, with raising incidence partly caused by excess nutrition and a lack of exercise.1 Steatosis as a hallmark of nonalcoholic fatty liver disease (NAFLD) is connected to obesity, insulin resistance, and type II diabetes.2 A strong correlation between steatosis and insulin resistance has been demonstrated in human patients and animal models of NAFLD.1, 3-6 Persistent hepatic lipid accumulation contributes to chronic inflammation with progression to nonalcoholic steatohepatitis (NASH), cirrhosis, and hepatocellular carcinoma (HCC).7 Steatosis results from excessive free fatty acid (FFA) synthesis relative to oxidative clearance8, 9 and/or elevated lipid hydrolysis in adipose tissues. FA synthesis, clearance, and release are, among others, regulated by neuroendocrine factors, such as growth hormone (GH) or glucocorticoids (GCs), whose levels vary under conditions of changing energy supply. Both signaling pathways have been implicated in the development of NAFLD and metabolic syndrome.10, 11 Animal studies have revealed that the transcription of distinct signal transducer and activator of transcription 5 (STAT5) target-gene subsets requires cofactor function of the glucocorticoid receptor (GR).12, 13 The interaction of STAT5 and GR ensures the proper transcription of genes implicated in postnatal body growth, such as insulin-like growth factor-1 (IGF-1).12, 13 As serum IGF-1 levels negatively regulate the release of GH in the pituitary, an impairment of this autoinhibitory GH/STAT5/IGF-1 feedback loop leads to GH resistance. This is of clinical interest, because it is tightly associated with metabolic syndrome.14 Mice lacking STAT5 or the GH receptor (GHR) in the liver acquire characteristics of GH resistance and develop steatosis and insulin resistance.3, 4 Importantly, hepatic STAT5 deficiency contributes to CCl4−induced liver fibrosis and HCC development.15 Furthermore, hepatocyte-specific deletion of JAK2 also results in GH resistance and the development of hepatic steatosis. However, these mice harbor no defects in glucose and insulin homeostasis.16

in resistance.3, 4 Importantly, hepatic STAT5 deficiency contributes to CCl4−induced liver fibrosis and HCC development.15 Furthermore, hepatocyte-specific deletion of JAK2 also results in GH resistance and the development of hepatic steatosis. However, these mice harbor no defects in glucose and insulin homeostasis.16 We aimed to investigate whether the regulation of hepatic lipid homeostasis (1) requires synergism of STAT5 and GR signaling or (2) both signaling cascades affect lipid metabolism independently. We confirm previous findings3, 4, 17 that STAT5 deficiency causes steatosis, insulin resistance, and glucose intolerance. However, the combined deletion of hepatic STAT5 and GR led to severe fatty liver disease resulting from a combination of hepatic GH resistance and hypercortisolism. The former resulted from the liver-specific ablation of STAT5, and the latter was from the deletion of the GR in hepatocytes. A combination of both conditions, as found in compound STAT5/GR mutants, induced peripheral lipodystrophy, additional liver lipid accumulation, and, subsequently, tumorigenic transformation of hepatocytes.

resulted from the liver-specific ablation of STAT5, and the latter was from the deletion of the GR in hepatocytes. A combination of both conditions, as found in compound STAT5/GR mutants, induced peripheral lipodystrophy, additional liver lipid accumulation, and, subsequently, tumorigenic transformation of hepatocytes. Materials and Methods Mice Mice with a hepatic deletion of STAT5 and/or the GR were generated as described.13 Littermates not expressing Alfp-Cre recombinase served as controls. For experimental procedures, we used male mice, if not stated otherwise. Mice were kept at the Decentralized Biomedical Facilities, Medical University of Vienna (Vienna, Austria), under standardized conditions. All animal experiments were carried out according to an ethical animal license protocol, and our contract was approved by university and Austrian Ministry authorities. Western Blotting Liver homogenates were prepared as previously described.13 Blots were incubated with antibodies against STAT5b (rabbit polyclonal antibody, epitope aa775-788), pY-STAT5 (#71-6900; Invitrogen, Carlsbad, CA), heat shock cognate 70-kDa protein (HSC-70) (sc-7298; Santa Cruz Biotechnology, Santa Cruz, CA), GR (sc-1004; Santa Cruz Biotechnology), and antibodies against total levels and the phosphorylated isoforms of p38, extracellular signal-regulated kinases 1 and 2 (ERK1/2), and c-Jun N-terminal kinases 1 and 2 (JNK1/2) (mitogen-activated protein kinase [MAPK] Sampler Kits #9926 and #9910; Cell Signaling, Beverly, MA).

GR (sc-1004; Santa Cruz Biotechnology), and antibodies against total levels and the phosphorylated isoforms of p38, extracellular signal-regulated kinases 1 and 2 (ERK1/2), and c-Jun N-terminal kinases 1 and 2 (JNK1/2) (mitogen-activated protein kinase [MAPK] Sampler Kits #9926 and #9910; Cell Signaling, Beverly, MA). Other Materials and Methods Animal and histology procedures, quantitative reverse-transcription polymerase chain reaction (qRT-PCR), serum biochemistry, determination of hepatic triglyceride levels, immunohistochemistry, and the measurement of reactive oxygen species (ROS) levels are described in the Supporting Materials and Methods. Statistical Analyses Results are presented as mean ± standard error of the mean. Statistical analyses were performed by analysis of variance, followed by Dunn's or Tukey's post-hoc tests. Data were considered statistically significant (*P < 0.05; **P < 0.01; ***P < 0.001).

Other Materials and Methods Animal and histology procedures, quantitative reverse-transcription polymerase chain reaction (qRT-PCR), serum biochemistry, determination of hepatic triglyceride levels, immunohistochemistry, and the measurement of reactive oxygen species (ROS) levels are described in the Supporting Materials and Methods. Statistical Analyses Results are presented as mean ± standard error of the mean. Statistical analyses were performed by analysis of variance, followed by Dunn's or Tukey's post-hoc tests. Data were considered statistically significant (*P < 0.05; **P < 0.01; ***P < 0.001). Results Deletion of Hepatic STAT5 and the GR Causes Steatosis and Lipodystrophy To investigate whether hepatic lipid homeostasis would require STAT5-GR synergism or whether the two transcription factors would affect lipid metabolism independently, we conditionally deleted the GR (GRKO), STAT5 (S5KO), or STAT5 and the GR (double knockout [DKO]) in hepatocytes. Efficient deletion was confirmed by western blotting analyses (Supporting Fig. 1A). Macroscopic hepatomegaly and steatosis were first evident in 2-month-old S5KO and DKO mutants, as compared to GRKO and control mice. Although hepatomegaly in S5KO mutants remained stable, DKO mice displayed progressive fatty liver disease, with a 4-fold increase in liver mass by 12 months of age (Fig. 1A, B) and an 8-fold rise in hepatic triglyceride (TG) content as early as 2 months of age (Fig. 1C). Strikingly, a dramatic depletion of white adipose tissue (WAT) was observed exclusively in DKO mice (−58%; Fig. 1A,B). Histological examination revealed a significantly increased mean score of steatosis in young (83% versus 49%), but not in aged, DKO mice, compared to age-matched S5KO mutants (77% versus 53%; Supporting Fig. 1B). Micro- and macrovesicular steatosis in S5KO (Fig. 1D, c, g, and k) and DKO mice (Fig. 1D, d, h, and l) was associated with elevated serum levels of alanine aminotransferase (ALT) and alkaline phosphatase (ALP) as indicators of liver injury (Fig. 1E). In contrast, histological analysis revealed a normal liver architecture in control (Fig. 1D, a, e, and i) and GRKO animals (Fig. 1D, b, f, and j) at all time points analyzed. A summary of the histological analysis is given in Supporting Table 1. Taken together, the STAT5-dependent fatty degeneration of hepatocytes is severely aggravated upon additional GR deletion in the liver.

rchitecture in control (Fig. 1D, a, e, and i) and GRKO animals (Fig. 1D, b, f, and j) at all time points analyzed. A summary of the histological analysis is given in Supporting Table 1. Taken together, the STAT5-dependent fatty degeneration of hepatocytes is severely aggravated upon additional GR deletion in the liver. Fig. 1 DKO mice develop severe steatosis, hepatomegaly, and lipodystrophy. (A) Macroscopic appearance of livers and epigonadal WAT in mutant and control mice at indicated time points. (B) Liver weight (LW)/body weight (BW) and WAT/BW ratios of mutant and control mice at indicated time points (n = 8/genotype/time point). (C) Hepatic triglyceride content in 2-month-old mice (n = 5/genotype). (D) Liver histology of livers from 6-month-old mice. (a-d) Liver sections were stained with hematoxylin and eosin. (e-h) Lipid accumulation in livers was visualized by Oil Red O on cryosections. (i-l) Electron microscopy analysis of fat distribution in livers of 2-month-old mice (cytoplasmic lipid droplets, green arrows; intrahepatic glycogen granules, black arrows). (E) Serum liver-damage parameters ALT and ALP of 2-month-old mice (n ≥ 5/genotype). *P < 0.05; **P < 0.01; ***P < 0.001.

l Red O on cryosections. (i-l) Electron microscopy analysis of fat distribution in livers of 2-month-old mice (cytoplasmic lipid droplets, green arrows; intrahepatic glycogen granules, black arrows). (E) Serum liver-damage parameters ALT and ALP of 2-month-old mice (n ≥ 5/genotype). *P < 0.05; **P < 0.01; ***P < 0.001. STAT5 Regulates Hepatic De Novo Lipogenesis Independently of GR Cofactor Interaction Hyperglycemia, hyperinsulinemia, and elevated resistin levels in both STAT5-deficient lines suggested hepatic insulin resistance upon STAT5 loss (Supporting Fig. 2A). Oral-glucose and insulin-tolerance tests confirmed insulin resistance and glucose intolerance3 (Supporting Fig. 2B). At the molecular level, defects in insulin receptor (IR) signaling, such as reduced tyrosine phosphorylation of the IR, IR substrates 1 and 2 (IRS-1 and -2), and serine phosphorylation of AKT were evident in both STAT5-deficient lines upon insulin administration (Supporting Fig. 2C). We detected increased transcript and protein levels of sterol regulatory element binding protein 1 (SREBP-1) and peroxisome proliferator-activated receptor gamma (PPAR-γ) in S5KO and DKO livers. In line with this, the gene expression of SREBP-1 (Fasn and Scd2) and PPAR-γ targets (Dgat1, Dgat2, and Cd36) was found to be increased. Transcript levels of Fgf21, which negatively control SREBP-1 maturation and activation, were decreased in single-knockout and DKO livers. Furthermore, messenger RNA (mRNA) and protein levels of lipogenic CCAAT enhancer binding protein (C/EBP)α and C/EBPβ were elevated (Supporting Fig. 3A,B; Supporting Table 2). Using chromatin immunoprecipitation analysis, a significantly enriched binding of GH-activated STAT5 to the Srebp-1a and Srebp-1c promoter was observed (Supporting Fig. 3C,D). Furthermore, GH-induced STAT5 activation led to a marked, time-dependent decrease of Srebp-1a and Srebp-1c mRNA levels in control livers (Supporting Fig. 3E).

n immunoprecipitation analysis, a significantly enriched binding of GH-activated STAT5 to the Srebp-1a and Srebp-1c promoter was observed (Supporting Fig. 3C,D). Furthermore, GH-induced STAT5 activation led to a marked, time-dependent decrease of Srebp-1a and Srebp-1c mRNA levels in control livers (Supporting Fig. 3E). Fig. 2 Combination of hepatic GH resistance and hypercortisolism causes peripheral lipodystrophy in DKO animals. (A) Histology of WAT. Epigonadal WAT of 12-month-old mice was stained with hematoxylin and eosin. (B) Quantification of WAT cell density. Hematoxylin and eosin–stained sections were used to analyze cell density using HistoQuest image analysis (TissueGnostics GmbH, Vienna, Austria). (C) Levels of FFA were determined in 2-month-old mice using colorimetric assays. (D) Levels of GH and IGF-1 were determined by ELISA. Levels of corticosterone (Cort) and ACTH were determined by radioimmunoassay (n = 8/genotype/time point). (E) Representative western blotting analysis of WAT homogenates from 2-month-old mice. Expression levels of STAT5 and GR proteins were determined using specific antibodies. HSC-70 served as the loading control. (F) Relative mRNA levels of Hsl, Atgl, and Plin were quantified by qRT-PCR in WAT from 6-month-old mice. Ct values were normalized to GAPDH (ΔCt method; n = 4/genotype). *P < 0.05; **P < 0.01; ***P < 0.001.

levels of STAT5 and GR proteins were determined using specific antibodies. HSC-70 served as the loading control. (F) Relative mRNA levels of Hsl, Atgl, and Plin were quantified by qRT-PCR in WAT from 6-month-old mice. Ct values were normalized to GAPDH (ΔCt method; n = 4/genotype). *P < 0.05; **P < 0.01; ***P < 0.001. Fig. 3 Impact of GR agonist or antagonist treatment on WAT lipolysis. (A) Macroscopic appearance of livers and WAT from S5KO mice following 14 days of dexamethasone (Dex) or mock treatment (phosphate-buffered saline). (B) LW/BW (left) and WAT/BW (middle) ratios of 6-month-old mutant mice of indicated genotypes and treatment. (C) Histological analysis of liver and WAT using hematoxylin and eosin–stained sections from control and S5KO mice after Dex treatment. (D) Macroscopic appearance of livers and WAT from DKO mice after 14 days of RU486 or mock (Oil) treatment. (E) LW/BW (left) and WAT/BW (middle) ratios of 6-month-old mice of indicated genotypes and treatment. (F) Levels of FFA were determined after RU486 or mock treatment of control and DKO mice using a colorimetric assay. For Dex and RU486 treatment: n ≥ 4/genotype/treatment. *P < 0.05; **P < 0.01; ***P < 0.001.

treatment. (E) LW/BW (left) and WAT/BW (middle) ratios of 6-month-old mice of indicated genotypes and treatment. (F) Levels of FFA were determined after RU486 or mock treatment of control and DKO mice using a colorimetric assay. For Dex and RU486 treatment: n ≥ 4/genotype/treatment. *P < 0.05; **P < 0.01; ***P < 0.001. Hepatic GH Resistance and Hypercortisolism Triggers Lipolysis of Adipose Tissue in DKO Mice Phenotypically, DKO mice display an aggravation of liver phenotype, compared to S5KO mice. One explanation for the increased hepatic TG load in DKO might be alterations in whole-body lipid homeostasis, resulting in enhanced lipolysis of adipocytes and elevated hepatic FFA delivery (Fig. 1A,B). Analysis of epigonadal WAT, brown adipose tissue, and subcutaneous fat from DKO mice revealed a severe reduction in fat depots and adipocyte cell size, compared with control and single-mutant mice (Fig. 2A,B and data not shown). Accordingly, elevated levels of circulating FFA were found in DKO mice (Fig. 2C). As expected, both STAT5-deficient lines showed high serum levels of GH secondary to the loss of negative IGF-1 regulation3 (Fig. 2D, left row). Yet, unexpectedly, GRKO and DKO mutants developed hypercortisolism, that is, elevated serum levels of corticosterone and its positive regulator, adrenocorticotropic hormone (ACTH; Fig. 2D, right row). In line with this, adipocytes from DKO mice showed increased STAT5 and GR activation (Fig. 2E and data not shown). It was demonstrated that the simultaneous activation of GH-STAT5 and GC-GR signaling stimulates lipolysis in human adipocytes.18 Therefore, we quantified the transcript levels of major WAT lipases, that is, adipose triglyceride lipase (Atgl) and hormone-sensitive lipase (Hsl). We observed a significant up-regulation of Atgl and Hsl transcripts accompanied by the reduced gene expression of Perilipin (Plin), a major coating protein of adipocytes, exclusively in DKO WAT (Fig. 2F). To confirm that synergistic adipose GH/STAT5/GC-GR activation accounted for the induction of lipases and concomitant lipolysis in DKO mice, we pharmacologically mimicked the combination of GH resistance and hypercortisolism. Therefore, we administered the GR agonist, dexamethasone (Dex), to 6-month-old S5KO mice for 14 days. Although Dex treatment had no effect in control animals, Dex treatment severely aggravated hepatomegaly and steatosis in S5KO livers, accompanied by a decrease in WAT size (Fig. 3A,B).

resistance and hypercortisolism. Therefore, we administered the GR agonist, dexamethasone (Dex), to 6-month-old S5KO mice for 14 days. Although Dex treatment had no effect in control animals, Dex treatment severely aggravated hepatomegaly and steatosis in S5KO livers, accompanied by a decrease in WAT size (Fig. 3A,B). Histological analysis further confirmed the aggravation of hepatic steatosis and lipodystrophy in Dex-treated S5KO mice (Fig. 3C). To determine whether systemic GR inhibition would protect from increased WAT lipolysis in DKO mutants, we treated 6-month-old DKO mice with the GR antagonist, RU486, for 14 days. Macroscopically, no considerable changes in liver and WAT size of RU486-treated DKO, compared to control, animals could be observed (Fig. 3D,E). Yet, RU486 treatment of DKO mice normalized the amount of serum FFA to levels comparable to RU486- and vehicle-treated control mice (Fig. 3F). Thus, the combination of hepatic GH resistance and hypercortisolism in DKO mice results in a generalized depletion of adipose stores, which, in turn, aggravates the STAT5-dependent fatty liver phenotype.

O mice normalized the amount of serum FFA to levels comparable to RU486- and vehicle-treated control mice (Fig. 3F). Thus, the combination of hepatic GH resistance and hypercortisolism in DKO mice results in a generalized depletion of adipose stores, which, in turn, aggravates the STAT5-dependent fatty liver phenotype. Spontaneous Development of Hepatocellular Carcinomas in DKO Mice Despite negligible fibrotic changes and mild inflammatory infiltration (Supporting Table 1), we observed the development of spontaneous liver tumors in DKO mice. Macroscopical examination of livers from 9-month-old DKO mice revealed atypical nodules (in 2 of 5), whereas S5KO and control mice showed no evidence of hepatic tumorigenesis at all time points analyzed. Furthermore, detailed phenotypic examination revealed no significant differences between control and GRKO mice (data not shown). Histological analysis of atypical nodules displayed distinct dysplastic lesions with vacuoles of accumulated fat compressing adjacent parenchyma (Supporting Fig. 4A, a-d). At 12 months of age, 35% (6 of 17) of DKO mice displayed dysplastic nodules and 59% (10 of 17) HCCs (Supporting Fig. 4B; Fig. 4A, a-b). Histological examination revealed well to moderately differentiated HCCs either of a (1) nonfatty and solid or (2) lipid-laden-tumor type, both of which displayed nuclear poly- and pleomorphism. Malignant hepatocytes were either growing in solid sheets or tended to aggregate in disorganized laminae (Fig. 4A, c-d; Supporting Fig. 4A, e-f). Periodic acid Schiff (PAS) staining revealed no significant necrotic degeneration of hepatocytes. Interestingly, solid/nonfatty tumors, in particular, displayed increased fibrous, pericellular collagen depositions, as illustrated by Chromotrop Anilinblue (CAB) staining (Fig. 4A, e-h and i-l). Hepatocyte proliferation was enhanced in DKO livers (∼8%), compared to S5KO (∼3%) and control livers (∼1%), as demonstrated by elevated numbers of Ki67-positive hepatocytes (Fig. 4A, m-p). However, no change in apoptotic rates of DKO livers was observed (Fig. 4A, q-t), and gene-expression levels of apoptosis regulating Bcl-2 family members Bcl-2, Bcl-xL, and Bax were only slightly changed (Supporting Fig. 4C). At the time point analyzed, ALT levels were similarly increased in DKO and S5KO mutants, indicating potent hepatocyte damage in both groups.

livers was observed (Fig. 4A, q-t), and gene-expression levels of apoptosis regulating Bcl-2 family members Bcl-2, Bcl-xL, and Bax were only slightly changed (Supporting Fig. 4C). At the time point analyzed, ALT levels were similarly increased in DKO and S5KO mutants, indicating potent hepatocyte damage in both groups. Next, we determined serum levels of the proinflammatory and tumor-promoting cytokines, tumor necrosis factor alpha (TNF-α) and interleukin (IL)-6. TNF-α was strongly elevated in DKO mice, whereas IL-6 levels were unchanged (Fig. 4B and data not shown). On the transcriptional level, we observed a strong up-regulation of Tnf-α and, to a lesser degree, Il-6 mRNA in DKO livers, whereas hepatic Il-1β transcript levels were unchanged (Fig. 4C). Collectively, these data suggest that aggravation of the STAT5-dependent fatty liver phenotype caused by the additional deposition of extrahepatic lipids facilitates liver tumorigenesis.

gulation of Tnf-α and, to a lesser degree, Il-6 mRNA in DKO livers, whereas hepatic Il-1β transcript levels were unchanged (Fig. 4C). Collectively, these data suggest that aggravation of the STAT5-dependent fatty liver phenotype caused by the additional deposition of extrahepatic lipids facilitates liver tumorigenesis. Fig. 4 Spontaneous development of liver tumors in DKO mice. (A) Hepatocellular carcinoma (HCC) formation in 12-month-old DKO mice. (a) Control liver. (b) Macroscopic view of representative DKO liver. Arrows indicate tumors and atypical nodules. Representative hematoxylin and eosin–stained sections showing two different types of HCCs as either (c) solid and nonfatty or as (d) tumors containing lipid droplets. (e-h) Representative PAS staining for glycogen deposition. (i-l) Representative CAB staining for collagen deposition. (m-p) Quantification of Ki67-positive hepatocytes by immunohistochemistry showing enhanced proliferation of DKO livers. Ki67-positive hepatocytes were quantified using HistoQuest image analysis (n ≥ 5/genotype; TissueGnostics GmbH, Vienna, Austria). (q-t) Representative immunohistochemistry for cleaved caspase 3–positive hepatocytes showing no increase in apoptosis of DKO livers. (B) Serum liver-damage parameters ALT and TNF-α levels of 12-month-old mice (n ≥ 5/genotype). (C) Relative mRNA levels of proinflammatory cytokines were quantified by qRT-PCR in livers from 12-month-old mice and normalized to Gapdh (n = 6/genotype). *P < 0.05; **P < 0.01; ***P < 0.001.

rease in apoptosis of DKO livers. (B) Serum liver-damage parameters ALT and TNF-α levels of 12-month-old mice (n ≥ 5/genotype). (C) Relative mRNA levels of proinflammatory cytokines were quantified by qRT-PCR in livers from 12-month-old mice and normalized to Gapdh (n = 6/genotype). *P < 0.05; **P < 0.01; ***P < 0.001. Mechanistic Insights in Tumor Development in DKO Mice Progressive fatty degeneration of hepatocytes is associated with oxidative stress and subsequent hepatocyte damage, a process shown to contribute to tumorigenesis.19 Global gene-expression and subsequent gene set enrichment analysis revealed deregulated expression levels of several antioxidant genes already in 2-month-old DKO animals (Supporting Fig. 4D). Thus, we measured ROS production and release in liver mitochondria. Extramitochondrial ROS levels in DKO livers were increased ∼4-fold over control and ∼2-fold over S5KO livers (Fig. 5A). The transcription of inducible nitric oxide synthase, Nos2, which leads to ROS and reactive nitrogen species generation, was also up-regulated in DKO livers. At this time point, transcript levels of the DNA damage-responsive gene, Gadd45a, were strongly elevated, whereas the expression levels of two major antioxidant genes, Sod1 and Sod2, were unchanged (Fig. 5B; Supporting Fig. 4E). Consistent with observed oxidative stress, DKO livers showed increased DNA damage, compared to control and S5KO livers, as assessed by the emergence of phoshorylated histone residues (pH2AX; Fig. 5C). To gain molecular insight in the processes governing the malignant transformation of hepatocytes in DKO animals, we determined the activation of the major stress-dependent MAPK-signaling pathways. These are triggered by continuous liver damage and are known to be involved in the pathogenesis of HCC. Tumor-bearing DKO mice exhibited elevated levels of JNK1 activity in the liver, which was almost absent in control and S5KO hepatocytes. In contrast, activation of ERK1/2 was unchanged between DKO and control livers, whereas p38 activation was reduced in DKO animals (Fig. 5D; Supporting Fig. 4F). HCCs displayed a modest increase in STAT3 phosphorylation, which was recently linked to hepatic tumorigenesis in the setting of chronic liver disease.15, 20 In nontumor liver tissue, however, STAT3 activity was almost not detectable (Fig. 5E).

eas p38 activation was reduced in DKO animals (Fig. 5D; Supporting Fig. 4F). HCCs displayed a modest increase in STAT3 phosphorylation, which was recently linked to hepatic tumorigenesis in the setting of chronic liver disease.15, 20 In nontumor liver tissue, however, STAT3 activity was almost not detectable (Fig. 5E). On the transcriptional level, livers from tumor-bearing mice exhibited significant up-regulation of Myc, Jun, Mmp9, and Vegfa that might contribute to an increased incidence of tumorigenesis (Fig. 5F). In summary, the development of HCCs in DKO mice coincides with oxidative stress and the activation of tumor-promoting JNK1- and STAT3-signaling cascades.

s from tumor-bearing mice exhibited significant up-regulation of Myc, Jun, Mmp9, and Vegfa that might contribute to an increased incidence of tumorigenesis (Fig. 5F). In summary, the development of HCCs in DKO mice coincides with oxidative stress and the activation of tumor-promoting JNK1- and STAT3-signaling cascades. Fig. 5 Oxidative stress-dependent hepatocyte damage and tumor-promoting signaling in DKO livers. (A) Extramitochondrial ROS production. ROS was determined using the 1-hydroxy-3-carboxy-pyrrolidine spin-trap method (n = 4/genotype). (B) Relative mRNA levels of Nos2 and Gadd45a were quantified by qRT-PCR in livers from 12-month-old mice and normalized to Gapdh (n = 6/genotype). (C) DNA damage in DKO mice. Liver sections were stained with antibodies against phoshorylated H2AX. Positive hepatocytes were quantified using image analysis (n ≥ 5/genotype). (D) Representative western blotting analysis showing protein expression and activation of JNK1/2, p38, and ERK1/2 in 12-month-old mice. HSC-70 served as the loading control. (E) STAT3 activation in DKO HCCs. Liver sections were stained with antibodies against phoshorylated STAT3. Positive hepatocytes were quantified in control, S5KO, and DKO nontumor and DKO tumor tissue using HistoQuest image analysis (n ≥ 5/genotype; TissueGnostics GmbH, Vienna, Austria). (F) Relative mRNA levels of Myc, Jun, Mmp9, and Vegfa were quantified by qRT-PCR in livers from 12-month-old mice and normalized to Gapdh (n = 6/genotype). *P < 0.05; **P < 0.01; ***P < 0.001.

, and DKO nontumor and DKO tumor tissue using HistoQuest image analysis (n ≥ 5/genotype; TissueGnostics GmbH, Vienna, Austria). (F) Relative mRNA levels of Myc, Jun, Mmp9, and Vegfa were quantified by qRT-PCR in livers from 12-month-old mice and normalized to Gapdh (n = 6/genotype). *P < 0.05; **P < 0.01; ***P < 0.001. Discussion Hepatic GH- and GC-signaling cascades influence metabolic functions under conditions of altered energy balance and stress. Defects in either of the signaling pathways have been implicated in NAFLD development, including children with NAFLD progressing to end-stage liver disease.10, 11, 21-23 On the molecular level, steatosis is often associated with enhanced expression of the prolipogenic transcription factors, SREBP-1c and PPAR-γ. Recent studies have revealed an important role of STAT5 in the prevention of steatosis. This was partly linked to the observation that impairment of hepatic GH-STAT5 signaling causes enhanced gene expression of Pparγ and its target gene, Cd36, which can also be directly regulated by STAT5.3, 17 Additionally, hepatic GHR deficiency resulted in enhanced Srebp-1c expression.4 Further studies have implicated SREBP-1c in Pparγ transcription,24 whereas a bidirectional, inhibitory cross-talk between STAT5 and PPAR-γ was postulated.25 SREBP-1 is most likely activated in response to decreased expression of Fgf21 and Insig2. Both transcripts are reported to be severely decreased upon liver-specific STAT5 deletion as well as upon systemic impairment of GHR signaling.17 Accordingly, we detected a decline in Fgf21 mRNA levels in S5KO livers, which was found to be even more severe upon additional GR deficiency. However, the expression of liver X receptor (LXR) isoforms, which are known to regulate Srebp-1 expression, was not significantly changed (data not shown). Additionally, we show that GH-activated STAT5 interacts with the promoter of both Srebp-1 isoforms, which results in down-regulated expression. Taken together, these observations confirm a GH-STAT5-dependent regulation of hepatic lipogenesis on the transcriptional level, where STAT5 might repress Srebp-1 isoforms. The induction of prolipogenic transcription factors and steatosis was not observed in GRKO livers, and it is also not present upon deletion of the N-terminal GR-interaction domain of STAT5 (as in STAT5ΔN mice)15 (data not shown). Thus, we consider the induction of SREBP-1 and PPAR-γ-mediated lipogenesis as the primary effect of hepatic STAT5 deficiency.

cription factors and steatosis was not observed in GRKO livers, and it is also not present upon deletion of the N-terminal GR-interaction domain of STAT5 (as in STAT5ΔN mice)15 (data not shown). Thus, we consider the induction of SREBP-1 and PPAR-γ-mediated lipogenesis as the primary effect of hepatic STAT5 deficiency. Moreover, hepatocyte-specific deletion of JAK2 causes massive steatosis and GH resistance. Interestingly, the phenotype was linked to increased peripheral GH-induced lipolysis and cluster of differentiation 36–mediated hepatic uptake of FFA, which could be rescued by the abrogation of GH secretion or partially normalized by antagonistic PPAR-γ action.16

ion of JAK2 causes massive steatosis and GH resistance. Interestingly, the phenotype was linked to increased peripheral GH-induced lipolysis and cluster of differentiation 36–mediated hepatic uptake of FFA, which could be rescued by the abrogation of GH secretion or partially normalized by antagonistic PPAR-γ action.16 Other conditions associated with steatosis are insulin resistance and glucose intolerance.1, 6 On the molecular level, insulin resistance is characterized by defects in IR signaling, which is observed upon the hepatocyte-specific deletion of the GH receptor or STAT5,3, 4 but not upon ablation of hepatic JAK2.16 Insulin resistance might be explained by observations made by others and similarly by us as follows. (1) A decrease in IRS-2-mediated signal transduction was accompanied by increased SREBP-1c associated with insulin resistance.26 Furthermore, when mice on a high-fat diet were treated with ezetimibe, a selective inhibitor of intestinal cholesterol absorption, down-regulation of hepatic SREBP-1c and reversed insulin resistance (IR) was a consequence, which was associated with increased pY-IRS-2 and pS-AKT.27 Similarly, we observed increased SREBP-1c mRNA and protein level in S5KO and DKO livers as well as impaired IR signal transduction. (2) The absence of insulin resistance in mice deficient for hepatic JAK2 might hint at a role of hepatic STAT5 in propagating IR signal transduction. It was shown that STAT5 is a physiological substrate of the IR in vitro and in tissues sensitive to insulin. Importantly, signaling through STAT5 upon insulin stimulation is JAK2 independent,28, 29 and insulin-stimulated STAT5 was shown to bind to the glucokinase promoter.30

STAT5 in propagating IR signal transduction. It was shown that STAT5 is a physiological substrate of the IR in vitro and in tissues sensitive to insulin. Importantly, signaling through STAT5 upon insulin stimulation is JAK2 independent,28, 29 and insulin-stimulated STAT5 was shown to bind to the glucokinase promoter.30 Hepatic GH-STAT5 signaling influences GH/IGF-1 and insulin levels in the circulation,3, 4 whereas GR signaling counteracts the effects of high stress-hormone (e.g., corticosterone and ACTH) levels. GCs have been shown to suppress hepatic CBG (Serpina6) expression. Hence, GR knockout mice display high basal CBG levels that are not suppressed by Dexamethasone.31 Accordingly, we observed elevated Serpina6 expression levels in GR-deficient livers, whereas expression of the GC-level regulating enzyme, 11β-HSD1, was unchanged (Supporting Fig. 5). This might suggest that upon hepatic GR deficiency, increased CBG expression results in elevated total serum GC levels. However, the increased CBG expression might lead to decreased unbound, active GCs and a subsequent decrease in negative feedback regulation, followed by enhanced ACTH and GC secretion.32, 33 Consistent with the notion of a strong induction of adipose tissue lipolysis by combined action of GH-STAT5 and GC-GR signaling, the enhanced shuttling of peripheral lipids to the liver was observed only in DKO mice. This process was associated with down-regulation of Plin34 and increased expression of Hsl and Atgl35, 36 in WAT, which triggered lipid mobilization from adipose tissue.

lipolysis by combined action of GH-STAT5 and GC-GR signaling, the enhanced shuttling of peripheral lipids to the liver was observed only in DKO mice. This process was associated with down-regulation of Plin34 and increased expression of Hsl and Atgl35, 36 in WAT, which triggered lipid mobilization from adipose tissue. Recent studies revealed that hepatic GH-STAT5 signaling not only prevents steatosis, but also has protective functions in the context of genetically or chemically induced liver fibrosis and cancer development15, 37 (Friedbichler et al., unpublished). In addition, STAT5 counterbalances unscheduled cellular proliferation by inducing the cell-cycle inhibitors, Cdkn2b and Cdkn1a.38 This suggests that STAT5-deficient livers are more sensitive to hepatocyte damage and malignant transformation. Coupling the preexisting steatosis in STAT5-deficient livers with increased adipose tissue-derived lipid fluxes causes the spontaneous development of liver tumors in DKO mice. Other studies have applied genetic, chemical, and dietary-based liver insults to mimic chronic liver disease, and demonstrated that these conditions facilitate HCC development.20, 39, 40 However, the onset of HCCs in our model occurred in settings of progressive steatosis, despite minor inflammation and fibrosis. This was also reported in genetically obese mice, which develop spontaneous hepatic hyperplasia and harbor an age-dependent risk of HCC formation in the absence of apparent inflammation and fibrosis.6 Notably, also, in patients, the development of HCC is increasingly observed in the absence of advanced liver injury, with metabolic syndrome as the only identified risk factor.41 We suggest that tumorigenesis in DKO livers is a direct effect of the massive lipid accumulation causing persistent liver damage partly via increased mitochondrial ROS production and leakage. STAT5b deficiency is further associated with increased PPAR-α-dependent FFA oxidation in peroxisomes,42 which, possibly, contributes to additional ROS accumulation. It is well known that high ROS levels and concomitant DNA damage predisposes hepatocytes to malignant transformation.

chondrial ROS production and leakage. STAT5b deficiency is further associated with increased PPAR-α-dependent FFA oxidation in peroxisomes,42 which, possibly, contributes to additional ROS accumulation. It is well known that high ROS levels and concomitant DNA damage predisposes hepatocytes to malignant transformation. Oxidative stress and a subsequent vicious cycle of hepatocyte damage, apoptosis, and cellular replenishment were shown to contribute to liver tumorigenesis.40 However, as observed in the high-fat diet or obesity-induced liver cancer,6, 20 the increased ROS formation did not lead to enhanced hepatocyte apoptosis, whereas tumor-tissue proliferation was elevated. Excessive hepatic lipid accumulation and accompanying hepatocyte damage might activate tumor-promoting MAPK signaling during HCC development.20, 43 ERK1/2 and p38 MAPK signaling was not induced, whereas JNK1 activity was enhanced in tumor-bearing mice. Increase in FFA, TNF-α, and ROS levels (as observed during the onset and progression of NAFLD) are potent activators of JNK1 and are all found elevated/activated upon the development of murine and human HCC.19, 20, 40 Moreover, the activity of STAT3 that is frequently activated in human HCC and implicated in the development of chemically and obesity-induced HCC,15, 20, 44 was significantly enhanced in DKO tumors. The latter might be explained by (1) compensatory GH-dependent STAT3 activation under conditions of hepatic STAT5 deficiency3 (Supporting Fig. 6) and/or (2) elevated systemic and liver TNF-α levels in DKO mice, which can lead to IL-6 production45 and subsequent STAT3 activation. Finally, known tumor-promoting downstream effectors of JNK1 and STAT3 displayed enhanced expression (e.g., Myc, Jun, Mmp9, and Vegfa), which was restricted to DKO livers. In summary, our results underline the importance of hepatic GH-STAT5 and GC-GR signaling in the maintenance of systemic lipid homeostasis, where these pathways protect hepatocytes from metabolic stress and HCC development.

played enhanced expression (e.g., Myc, Jun, Mmp9, and Vegfa), which was restricted to DKO livers. In summary, our results underline the importance of hepatic GH-STAT5 and GC-GR signaling in the maintenance of systemic lipid homeostasis, where these pathways protect hepatocytes from metabolic stress and HCC development. The authors thank Michael Trauner, Jan Tuckermann, and Lothar Hennighausen for their helpful discussions and critical reading of the manuscript for this article and Jelena Marjanovic for her excellent technical support. Abbreviations ACTHadrenocorticotropic hormone ALPalkaline phosphatase ALTalanine aminotransferase CBGcorticosteroid binding globulin C/EBPCCAAT enhancer binding protein Dexdexamethasone DKOdouble knockout ERK1/2extracellular signal-regulated kinases 1 and 2 FFAfree fatty acids GCsglucocorticoids GHgrowth hormone GRglucocorticoid receptor HCChepatocellular carcinoma HSC-70heat shock cognate 70 kDa protein IGF-1insulin-like growth factor 1 ILinterleukin iNOSinducible nitric oxide synthase IRinsulin receptor IRSinsulin receptor substrate JNK1/2c-Jun N-terminal kinases 1 and 2 LXRliver X receptor MAPKmitogen-activated protein kinase mRNAmessenger RNA NAFLDnonalcoholic fatty liver disease NASHnonalcoholic steatohepatitis Plinperilipin PPAR-γperoxisome proliferator-activated receptor gamma qRT-PCRquantitative reverse-transcription polymerase chain reaction ROSreactive oxygen species STAT5signal transducer and activator of transcription 5 SREBP-1sterol regulatory element binding protein 1 TGtriglyceride TNF-αtumor necrosis factor alpha WATwhite adipose tissue.

Plinperilipin PPAR-γperoxisome proliferator-activated receptor gamma qRT-PCRquantitative reverse-transcription polymerase chain reaction ROSreactive oxygen species STAT5signal transducer and activator of transcription 5 SREBP-1sterol regulatory element binding protein 1 TGtriglyceride TNF-αtumor necrosis factor alpha WATwhite adipose tissue. Supplementary material Additional Supporting Information may be found in the online version of this article.

This practice guideline has been approved by the American Association for the Study of Liver Diseases (AASLD) and endorsed by the Infectious Diseases Society of America, the American College of Gastroenterology and the National Viral Hepatitis Roundtable. Preamble These recommendations provide a data-supported approach to establishing guidelines. They are based on the following: (1) a formal review and analysis of the recently published world literature on the topic (MEDLINE search up to June 2011); (2) the American College of Physicians' Manual for Assessing Health Practices and Designing Practice Guidelines;1 (3) guideline policies, including the AASLD Policy on the Development and Use of Practice Guidelines and the American Gastroenterological Association's Policy Statement on the Use of Medical Practice Guidelines;2 and (4) the experience of the authors in regard to hepatitis C.

actices and Designing Practice Guidelines;1 (3) guideline policies, including the AASLD Policy on the Development and Use of Practice Guidelines and the American Gastroenterological Association's Policy Statement on the Use of Medical Practice Guidelines;2 and (4) the experience of the authors in regard to hepatitis C. Intended for use by physicians, these recommendations suggest preferred approaches to the diagnostic, therapeutic, and preventive aspects of care. They are intended to be flexible, in contrast to standards of care, which are inflexible policies to be followed in every case. Specific recommendations are based on relevant published information. To more fully characterize the quality of evidence supporting recommendations, the Practice Guidelines Committee of the AASLD requires a Class (reflecting benefit versus risk) and Level (assessing strength or certainty) of Evidence to be assigned and reported with each recommendation (Table 1, adapted from the American College of Cardiology and the American Heart Association Practice Guidelines).3,4 Table 1 Grading System for Recommendations

Intended for use by physicians, these recommendations suggest preferred approaches to the diagnostic, therapeutic, and preventive aspects of care. They are intended to be flexible, in contrast to standards of care, which are inflexible policies to be followed in every case. Specific recommendations are based on relevant published information. To more fully characterize the quality of evidence supporting recommendations, the Practice Guidelines Committee of the AASLD requires a Class (reflecting benefit versus risk) and Level (assessing strength or certainty) of Evidence to be assigned and reported with each recommendation (Table 1, adapted from the American College of Cardiology and the American Heart Association Practice Guidelines).3,4 Table 1 Grading System for Recommendations Classification Description Class 1 Conditions for which there is evidence and/or general agreement that a given diagnostic evaluation procedure or treatment is beneficial, useful, and effective Class 2 Conditions for which there is conflicting evidence and/or a divergence of opinion about the usefulness/efficacy of a diagnostic evaluation, procedure, or treatment Class 2a Weight of evidence/opinion is in favor of usefulness/efficacy Class 2b Usefulness/efficacy is less well established by evidence/opinion Class 3 Conditions for which there is evidence and/or general agreement that a diagnostic evaluation, procedure/treatment is not useful/effective and in some cases may be harmful Level of Evidence Description Level A Data derived from multiple randomized clinical trials or meta-analyses Level B Data derived from a single randomized trial, or nonrandomized studies Level C Only consensus opinion of experts, case studies, or standard-of-care Introduction The standard of care (SOC) therapy for patients with chronic hepatitis C virus (HCV) infection has been the use of both peginterferon (PegIFN) and ribavirin (RBV).

a derived from a single randomized trial, or nonrandomized studies Level C Only consensus opinion of experts, case studies, or standard-of-care Introduction The standard of care (SOC) therapy for patients with chronic hepatitis C virus (HCV) infection has been the use of both peginterferon (PegIFN) and ribavirin (RBV). These drugs are administered for either 48 weeks (HCV genotypes 1, 4, 5, and 6) or for 24 weeks (HCV genotypes 2 and 3), inducing sustained virologic response (SVR) rates of 40%-50% in those with genotype 1 and of 80% or more in those with genotypes 2 and 3 infections.5-7 Once achieved, an SVR is associated with long-term clearance of HCV infection, which is regarded as a virologic “cure,” as well as with improved morbidity and mortality.8-10 Two major advances have occurred since the last update of treatment guidelines for chronic hepatitis C (CHC) that have changed the optimal treatment regimen of genotype 1 chronic HCV infection: the development of direct-acting antiviral (DAA) agents11-17 and the identification of several single-nucleotide polymorphisms associated with spontaneous and treatment-induced clearance of HCV infection.18,19 Although PegIFN and RBV remain vital components of therapy, the emergence of DAAs has led to a substantial improvement in SVR rates and the option of abbreviated therapy in many patients with genotype 1 chronic HCV infection. A revision of the prior treatment guidelines is therefore necessary, but is based on data that are presently limited. Accordingly, there may be need to reconsider some of the recommendations as additional data become available. These guidelines review what treatment for genotype 1 chronic HCV infection is now regarded as optimal, but they do not address the issue of prioritization of patient selection for treatment or of treatment of special patient populations.

ed to reconsider some of the recommendations as additional data become available. These guidelines review what treatment for genotype 1 chronic HCV infection is now regarded as optimal, but they do not address the issue of prioritization of patient selection for treatment or of treatment of special patient populations. Direct-Acting Antiviral Agents There are multiple steps in the viral lifecycle that represent potential pharmacologic targets. A number of compounds encompassing at least five distinct drug classes are currently under development for the treatment of CHC. Presently, only inhibitors of the HCV nonstructural protein 3/4A (NS3/4A) serine protease have been approved by the Food and Drug Administration (FDA).

resent potential pharmacologic targets. A number of compounds encompassing at least five distinct drug classes are currently under development for the treatment of CHC. Presently, only inhibitors of the HCV nonstructural protein 3/4A (NS3/4A) serine protease have been approved by the Food and Drug Administration (FDA). Protease Inhibitors The NS3/4A serine protease is required for RNA replication and virion assembly. Two inhibitors of the NS3/4A serine protease, boceprevir (BOC) and telaprevir (TVR), have demonstrated potent inhibition of HCV genotype 1 replication and markedly improved SVR rates in treatment-naïve and treatment-experienced patients.12,13,16,17 Limited phase 2 testing has shown that TVR also has activity against HCV genotype 2 infection but not against genotype 3.20 With regard to BOC, there are limited data indicating that it too, has activity against genotype 2 but also against genotype 3 HCV infection.21 However, at this time, neither drug should be used to treat patients with genotype 2 or 3 HCV infections, and when administered as monotherapy, each PI rapidly selects for resistance variants, leading to virological failure. Combining either PI with PegIFN and RBV limits selection of resistant variants and improves antiviral response.15

ime, neither drug should be used to treat patients with genotype 2 or 3 HCV infections, and when administered as monotherapy, each PI rapidly selects for resistance variants, leading to virological failure. Combining either PI with PegIFN and RBV limits selection of resistant variants and improves antiviral response.15 Patients Who Have Never Received Therapy (Treatment-Naïve Patients) Boceprevir The SPRINT-2 trial evaluated BOC in two cohorts of treatment-naïve patients: Caucasian and black patients.12 The number of patients in the black cohort was small in comparison to that of the Caucasian cohort and may have been insufficient to provide an adequate assessment of true response in this population. All patients were first treated with PegIFN alfa-2b and weight-based RBV as lead-in therapy for a period of 4 weeks, followed by one of three regimens: (1) BOC, PegIFN, and RBV that was administered for 24 weeks if, at study week 8 (week 4 of triple therapy), the HCV RNA level became undetectable (as defined in the package insert as <10-15 IU/mL), referred to as response-guided therapy (RGT); if, however, HCV RNA remained detectable at any visit from week 8 up to but not including week 24 (i.e., a slow virological response), BOC was discontinued and the patient received SOC treatment for an additional 20 weeks (2) BOC, PegIFN, and RBV administered for a fixed duration of 44 weeks; and (3) PegIFN alfa-2b and weight-based RBV alone continued for an additional 44 weeks, representing SOC therapy.12 The BOC dose was 800 mg, given by mouth three times per day with food. The overall SVR rates were higher in the BOC arms, (63% and 66% respectively) than in the SOC arm (38%), but differed according to race (Fig. 1). The SVR rates among Caucasian patients were 67% in the RGT, 69% in the fixed duration, and 41% in the SOC arms, respectively.12 In black patients, the SVR rates were 42% in the RGT, 53% in the fixed duration, and 23% in the SOC arms, respectively (Fig. 1).12 A total of 54% of Caucasian recipients of BOC experienced a rapid virological response (RVR; HCV RNA undetectable, <10-15 IU/mL at week 8, this interval selected because of the 4 week lead-in). By contrast, only 20% of black recipients of BOC experienced an RVR.

ration, and 23% in the SOC arms, respectively (Fig. 1).12 A total of 54% of Caucasian recipients of BOC experienced a rapid virological response (RVR; HCV RNA undetectable, <10-15 IU/mL at week 8, this interval selected because of the 4 week lead-in). By contrast, only 20% of black recipients of BOC experienced an RVR. Regardless of race, among those patients who became HCV RNA negative at week 8 (∼57% in both BOC arms and 17% in SOC arm), the SVR rates were 88% in the RGT arm, 90% in the fixed duration arm and 85% in the arm treated by SOC, compared to SVR rates of 36%, 40%, and 30%, respectively, if HCV RNA remained detectable at week 8 (Fig. 2).12 Fig. 1 Sustained virological response (SVR) rates, overall and according to race, in treatment-naïve patients with genotype 1 chronic HCV infection: Boceprevir (BOC) plus peginterferon (PegIFN) and ribavirin (RBV) versus standard of care (SOC). All patients were first treated with PegIFN + RBV for 4 weeks as lead-in therapy followed by one of three regiments: (1) BOC/PegIFN/RBV RGT - triple therapy for 24 weeks provided HCV RNA levels were negative weeks 8 thorugh 24 – response guided therapy; those with a detectable HCV RNA level between weeks 8 and 24 received SOC for an additional 20 weeks; (2) BOC/PegIFN/RBV fixed duration - triple therapy for a fixed duration of 44 weeks; and (3) SOC - consisted of PegIFN and weight based RBV administered for 48 weeks.12

ive weeks 8 thorugh 24 – response guided therapy; those with a detectable HCV RNA level between weeks 8 and 24 received SOC for an additional 20 weeks; (2) BOC/PegIFN/RBV fixed duration - triple therapy for a fixed duration of 44 weeks; and (3) SOC - consisted of PegIFN and weight based RBV administered for 48 weeks.12 Fig. 2 Sustained virological response (SVR) rates, overall and based on a rapid virological response (RVR, undetectable HCV RNA at week 8 [week 4 of triple therapy]) in treatment-naïve patients with genotype 1 chronic HCV infection: Boceprevir (BOC) plus peginterferon (PegIFN) versus standard of care (SOC). All patients were first treated with PegIFN + RBV for 4 weeks as lead-in therapy followed by one of three regiments: (1) BOC/PegIFN/RBV RGT - patients who achieved an RVR (undetable HCV RNA at week 8 [week 4 of triple therapy]) continued treatment for an additional 24 weeks (RGT - response guided therapy); if an RVR did not develop, treatment with triple therapy continued to week 28 followed by SOC treatment for 20 weeks. SOC treatment consisted of PegIFN and RBV administered for 48 weeks.12 Note that the combined numbers of RVR-positive and RVR-negative patients are not equivalent to the total number of patients enrolled, presumably because of missing HCV RNA values at the week 8 time point.

k 28 followed by SOC treatment for 20 weeks. SOC treatment consisted of PegIFN and RBV administered for 48 weeks.12 Note that the combined numbers of RVR-positive and RVR-negative patients are not equivalent to the total number of patients enrolled, presumably because of missing HCV RNA values at the week 8 time point. In subgroup analysis, SVR rates were higher in BOC-containing regimens across all the pretreatment variables that had been identified in previous studies to influence response to SOC therapy, including advanced fibrosis, race, and high pretreatment HCV viral load. Moreover, the SVR rate in subgroups was similar in both the RGT and fixed duration arms and therefore, the AASLD and the FDA support the use of RGT for treatment-naïve patients without cirrhosis. The FDA recommends that patients with compensated cirrhosis should not receive RGT, however, this is based on limited data and requires further study. Of note, if the virological response did not meet criteria for RGT, i.e., a slow virological response, the FDA recommends (based on modeling) triple therapy for 32 weeks preceded by the 4 weeks of SOC treatment), followed by 12 weeks of PegIFN and RBV alone; a strategy that differs from the phase 3 trial design. All therapy should be discontinued if the HCV RNA level is ≥100 IU/mL at week 12 or ≥10 to 15 IU/mL at week 24.

e FDA recommends (based on modeling) triple therapy for 32 weeks preceded by the 4 weeks of SOC treatment), followed by 12 weeks of PegIFN and RBV alone; a strategy that differs from the phase 3 trial design. All therapy should be discontinued if the HCV RNA level is ≥100 IU/mL at week 12 or ≥10 to 15 IU/mL at week 24. Telaprevir Two phase 3 trials evaluated the efficacy of TVR in combination with PegIFN alfa-2a and RBV in treatment-naïve patients with genotype 1 chronic HCV infection.16,22 Black patients were included but not as a separate cohort and were insufficient in number to provide an adequate assessment of true response in this population. In the ADVANCE trial, patients received TVR together with PegIFN and RBV for either 8 (T8PR) or 12 (T12PR) weeks followed by PegIFN and RBV alone in a response-guided paradigm.16 The TVR dose was 750 mg given by mouth every 8 hours with food (in particular, a fatty meal). Patients in the T8PR and T12PR groups who achieved an “extended RVR” (eRVR)—which for this drug was defined as undetectable (<10-15 IU/mL) HCV RNA levels at weeks 4 and 12—stopped therapy at week 24, whereas those in whom an eRVR did not occur received a total of 48 weeks of PegIFN and RBV. All patients in the control group received PegIFN and RBV therapy for 48 weeks. The overall SVR rates among patients in the T8PR and T12PR groups were 69% and 75%, respectively,16 compared with a rate of 44% in the control group (Table 2 and Fig. 3). Using the RGT approach, 58% and 57% of patients in the T12PR and T8PR groups, respectively, attained an eRVR, 89% and 83% of whom ultimately achieved an SVR.16 Thus, developing an eRVR appears to be the strongest predictor that an SVR will occur.

16 compared with a rate of 44% in the control group (Table 2 and Fig. 3). Using the RGT approach, 58% and 57% of patients in the T12PR and T8PR groups, respectively, attained an eRVR, 89% and 83% of whom ultimately achieved an SVR.16 Thus, developing an eRVR appears to be the strongest predictor that an SVR will occur. Fig. 3 Sustained virological response (SVR) rates, overall and according to race, in treatment naïve patients with genotype 1 chronic HCV infection: Telaprevir (TVR) plus peginterferon and ribavirin (PR) treatment for 8 (T8PR) or 12 (T12PR) weeks versus standard of care (SOC). Patients in the triple therapy arms who developed an eRVR (extended rapid virological response; defined as undetectable HCV RNA at weeks 4 and 12) stopped treatment at week 24 (response-guided therapy, RGT); if eRVR did not develop, treatment continued to 48 weeks. SOC treatment consisted of PegIFN and RBV administered for 48 weeks.16 Table 2 Comparison of Protease Inhibitors in Combination with Peginterferon Alfa (PegIFN) and Ribavirin (RBV) in Treatment-Naive Subjects

with fresh medium supplemented with 10% fetal bovine serum and penicillin-streptomycin. Forty-eight hours later, cells were used for RNA extraction. Reverse transcription (RT) and real-time polymerase chain reaction (Q-PCR) were carried out to examine gene expression levels of mouse Pxr, human UGT1A1 and mouse Cyp3a11. Immunoblot Analysis and Real-Time PCR Mice were sacrificed and livers were perfused with ice-cold 1.15% KCL and microsomes prepared as outlined.14 All western blots were performed using NuPAGE BisTris-polyacrylamide gels as described.15 For real-time quantitative Q-PCR analysis, ≍100 mg of liver tissue was homogenized into 1 mL of TRIzol and RNA prepared. Using iScript Reverse Transcriptase (BioRad), 1 μg of total RNA was used for the generation of complementary DNA (cDNA) as outlined by the manufacturer in a total volume of 20 μL. Following synthesis of cDNA, 2 μL was used in real-time PCR conducted with a QuantiTect SYBR GreenPCR kit (Qiagen, Valencia, CA) using a MX4000 Multiplex Q-PCR (Stratagene, La Jolla, CA) programmed to take three fluorescence data points at the end of each annealing plateau. All PCR reactions were performed in triplicate as outlined.14Ct values were normalized to mouse cyclophilin (CPH). The specific primers used to quantitate the respective gene transcripts18 are listed in Supporting Table 1.

Fig. 3 Sustained virological response (SVR) rates, overall and according to race, in treatment naïve patients with genotype 1 chronic HCV infection: Telaprevir (TVR) plus peginterferon and ribavirin (PR) treatment for 8 (T8PR) or 12 (T12PR) weeks versus standard of care (SOC). Patients in the triple therapy arms who developed an eRVR (extended rapid virological response; defined as undetectable HCV RNA at weeks 4 and 12) stopped treatment at week 24 (response-guided therapy, RGT); if eRVR did not develop, treatment continued to 48 weeks. SOC treatment consisted of PegIFN and RBV administered for 48 weeks.16 Table 2 Comparison of Protease Inhibitors in Combination with Peginterferon Alfa (PegIFN) and Ribavirin (RBV) in Treatment-Naive Subjects Variable Boceprevir (BOC)12 Telaprevir (TVR)16 Study design RCT RCT 4-Week lead-in PegIFN/RBV Yes No Duration of triple therapy 24 or 44 weeks in combination with PegIFN/RBV* 12 weeks followed by 12 or 36 weeks PegIFN/RBV† Response-guided therapy (RGT) Yes Yes Eligible for response-guided therapy (%) 44 58 SVR (%) BOC44/PR: 66 T8PR: 69 BOC/PR/RGT: 63 T12PR: 75 SOC: 38 SOC: 44 End of treatment response (%) BOC44/PR: 76 T8PR: 81 BOC/PR/RGT: 71 T12PR: 87 SOC: 53 SOC: 63 Relapse (%) BOC44/PR: 9 T8PR: 9 BOC/PR/RGT: 9 T12PR: 9 SOC: 22 SOC: 28 Treatment emergent resistance (%) 16 12 Adverse event more frequent in triple therapy arm compared to SOC Anemia, dysgeusia Rash, anemia, pruritus, nausea, diarrhea Adverse events leading to treatment discontinuation (%) NA 12 Serious adverse events study drug vs SOC (%) 11 vs 9 9 vs 7 NA, not available; PR, peginterferon plus ribavirin; RCT, randomized, controlled trial; SOC, standard of care; SVR, sustained virological response.

dysgeusia Rash, anemia, pruritus, nausea, diarrhea Adverse events leading to treatment discontinuation (%) NA 12 Serious adverse events study drug vs SOC (%) 11 vs 9 9 vs 7 NA, not available; PR, peginterferon plus ribavirin; RCT, randomized, controlled trial; SOC, standard of care; SVR, sustained virological response. * All patients were first treated with PegIFN alfa-2b and weight-based RBV as lead-in therapy for a period of 4 weeks, followed by one of three regimens: (1) BOC/PR/RGT: BOC, PegIFN, and RBV that was administered for 24 weeks if, at study week 8 (week 4 of triple therapy), the HCV RNA level became undetectable (as defined in the package insert as <10-15 IU/mL), referred to as response-guided therapy (RGT); if, however, HCV RNA remained detectable at any visit from week 8 up to but not including week 24 (i.e., a slow virological response), BOC was discontinued and the patient received SOC treatment for an additional 20 weeks; (2) BOC44/PR: BOC, PegIFN, and RBV administered for a fixed duration of 44 weeks; and (3) SOC: PegIFN alfa-2b and weight-based RBV alone continued for an additional 44 weeks.

ot including week 24 (i.e., a slow virological response), BOC was discontinued and the patient received SOC treatment for an additional 20 weeks; (2) BOC44/PR: BOC, PegIFN, and RBV administered for a fixed duration of 44 weeks; and (3) SOC: PegIFN alfa-2b and weight-based RBV alone continued for an additional 44 weeks. † Telaprevir (TVR) plus peginterferon and ribavirin (PR) treatment for 8 (T8PR) or 12 (T12PR) weeks versus standard of care (SOC). Patients in the T8PR and T12PR groups who achieved an “extended RVR” (eRVR), which for this drug was defined as undetectable (<10-15 IU/mL) HCV RNA levels at weeks 4 and 12, stopped therapy at week 24, whereas those in whom an eRVR did not occur received a total of 48 weeks of PegIFN and RBV. All patients in the control group received PegIFN and RBV therapy for 48 weeks.

“extended RVR” (eRVR), which for this drug was defined as undetectable (<10-15 IU/mL) HCV RNA levels at weeks 4 and 12, stopped therapy at week 24, whereas those in whom an eRVR did not occur received a total of 48 weeks of PegIFN and RBV. All patients in the control group received PegIFN and RBV therapy for 48 weeks. SVR rates were higher in TVR-containing regimens compared to SOC treatment among patients with disease characteristics found previously to be associated with a poorer response to SOC treatment. Although few black patients and other difficult-to-treat patient populations were included in the TVR phase 3 trials, an improved SVR rate was observed regardless of race, ethnicity, or level of hepatic fibrosis. With regard to race, treatment with a TVR-based regimen significantly improved SVR rates in black patients (T8PR, 58% and T12PR, 62%) compared to the SVR rates achieved in those treated with the SOC regimen (25%) (Fig. 3). Moreover, the SVR rate was >80% among black patients who achieved an eRVR on a TVR-based regimen. A total of 62% of patients in the T12PR group and 53% in the T8PR group with advanced fibrosis achieved an SVR, the rate improving to >80% among those with an eRVR. In the T12PR group, the impact of high versus low viral load (>800,000 or <800,000 IU/mL) on SVR rates was minimal; the SVR rate was 74% in patients with a high viral load and 78% in those with a low viral load.

the T8PR group with advanced fibrosis achieved an SVR, the rate improving to >80% among those with an eRVR. In the T12PR group, the impact of high versus low viral load (>800,000 or <800,000 IU/mL) on SVR rates was minimal; the SVR rate was 74% in patients with a high viral load and 78% in those with a low viral load. The ILLUMINATE trial focused on defining the utility of RGT in patients with an eRVR. All patients received an initial 12 weeks of TVR-based triple therapy followed by PegIFN and RBV therapy alone.22 Those who achieved an eRVR were randomized at week 20 to receive either an additional 4 or an additional 28 weeks of PegIFN and RBV whereas those who failed to achieve an eRVR were not randomized and received an additional 28 weeks of PegIFN and RBV. The overall SVR rate for all patients was 72% (Fig. 4), similar to the 75% rate found in the ADVANCE trial.22 Among the 65% of patients who achieved an eRVR and received either an additional 4 or 28 weeks of PegIFN and RBV, SVR rates were 92% and 88%, respectively (Fig. 4). By contrast, the SVR rate was only 64% among patients who did not achieve an eRVR.22 These data suggest that a response-guided strategy based on eRVR permits a shortened duration of therapy without jeopardizing the SVR response rate and may be appropriate for up to two-thirds of patients with genotype 1 chronic HCV infection. The use of RGT may, however, be unsuitable for patients with cirrhosis, but at present the data are insufficient to guide management in this difficult-to-treat population. Therapy should be discontinued in all patients if HCV RNA levels are ≥1,000 IU/mL at weeks 4 or 12 and/or >10-15 IU/mL at week 24.

ic HCV infection. The use of RGT may, however, be unsuitable for patients with cirrhosis, but at present the data are insufficient to guide management in this difficult-to-treat population. Therapy should be discontinued in all patients if HCV RNA levels are ≥1,000 IU/mL at weeks 4 or 12 and/or >10-15 IU/mL at week 24. Fig. 4 Sustained virological response (SVR) rates in treatment naïve patients with genotype 1 chronic HCV infection: Telaprevir (TVR) plus peginterferon and ribavirin (PR) results overall and among those who did or did not achieve an eRVR (extended rapid virological response; undetectable HCV RNA at weeks 4 and 12). Patients who achieved an eRVR were randomized at week 20 to receive an additional 4 or an additional 28 weeks of SOC therapy; those who did not develop an eRVR were not randomized and all received an additional 24 weeks of SOC therapy.22 Recommendations: The optimal therapy for genotype 1, chronic HCV infection is the use of boceprevir or telaprevir in combination with peginterferon alfa and ribavirin (Class 1, Level A). Boceprevir and telaprevir should not be used without peginterferon alfa and weight-based ribavirin (Class 1, Level A). For Treatment-Naïve Patients: The recommended dose of boceprevir is 800 mg administered with food three times per day (every 7-9 hours) together with peginterferon alfa and weight-based ribavirin for 24-44 weeks preceded by 4 weeks of lead-in treatment with peginterferon alfa and ribavirin alone (Class 1, Level A).

Boceprevir and telaprevir should not be used without peginterferon alfa and weight-based ribavirin (Class 1, Level A). For Treatment-Naïve Patients: The recommended dose of boceprevir is 800 mg administered with food three times per day (every 7-9 hours) together with peginterferon alfa and weight-based ribavirin for 24-44 weeks preceded by 4 weeks of lead-in treatment with peginterferon alfa and ribavirin alone (Class 1, Level A). Patients without cirrhosis treated with boceprevir, peginterferon, and ribavirin, preceded by 4 weeks of lead-in peginterferon and ribavirin, whose HCV RNA level at weeks 8 and 24 is undetectable, may be considered for a shortened duration of treatment of 28 weeks in total (4 weeks lead-in with peginterferon and ribavirin followed by 24 weeks of triple therapy) (Class 2a, Level B). Treatment with all three drugs (boceprevir, peginterferon alfa, and ribavirin) should be stopped if the HCV RNA level is >100 IU/mL at treatment week 12 or detectable at treatment week 24 (Class 2a, Level B). The recommended dose of telaprevir is 750 mg administered with food (not low-fat) three times per day (every 7-9 hours) together with peginterferon alfa and weight-based ribavirin for 12 weeks followed by an additional 12-36 weeks of peginterferon alfa and ribavirin (Class 1, Level A). Patients without cirrhosis treated with telaprevir, peginterferon, and ribavirin, whose HCV RNA level at weeks 4 and 12 is undetectable should be considered for a shortened duration of therapy of 24 weeks (Class 2a, Level A).

The recommended dose of telaprevir is 750 mg administered with food (not low-fat) three times per day (every 7-9 hours) together with peginterferon alfa and weight-based ribavirin for 12 weeks followed by an additional 12-36 weeks of peginterferon alfa and ribavirin (Class 1, Level A). Patients without cirrhosis treated with telaprevir, peginterferon, and ribavirin, whose HCV RNA level at weeks 4 and 12 is undetectable should be considered for a shortened duration of therapy of 24 weeks (Class 2a, Level A). Patients with cirrhosis treated with either boceprevir or telaprevir in combination with peginterferon and ribavirin should receive therapy for a duration of 48 weeks (Class 2b, Level B). Treatment with all three drugs (telaprevir, peginterferon alfa, and ribavirin) should be stopped if the HCV RNA level is >1,000 IU/mL at treatment weeks 4 or 12 and/or detectable at treatment week 24 (Class 2a, Level B).

Patients with cirrhosis treated with either boceprevir or telaprevir in combination with peginterferon and ribavirin should receive therapy for a duration of 48 weeks (Class 2b, Level B). Treatment with all three drugs (telaprevir, peginterferon alfa, and ribavirin) should be stopped if the HCV RNA level is >1,000 IU/mL at treatment weeks 4 or 12 and/or detectable at treatment week 24 (Class 2a, Level B). Patients Who Have Previously Received Therapy Three categories have been defined for persons who had received previous therapy for CHC but who had failed the treatment. Null responders are persons whose HCV RNA level did not decline by at least 2 log IU/mL at treatment week 12; partial responders are persons whose HCV RNA level dropped by at least 2 log IU/mL at treatment week 12 but in whom HCV RNA was still detected at treatment week 24; and relapsers are persons whose HCV RNA became undetectable during treatment, but then reappeared after treatment ended. Taking these categories into account, phase 3 trials have been performed also in treatment-experienced patients with genotype 1 chronic HCV infection using BOC and TVR in combination with PegIFN and RBV. The BOC trial design included a 4-week lead-in phase of PegIFN and RBV and compared response-guided triple therapy (BOC plus PegIFN and RBV for 32 weeks; patients with a detectable HCV RNA level at week 8 received SOC for an additional 12 weeks) and a fixed duration of triple therapy given for 44 weeks (total 48 weeks of therapy), to SOC therapy.13 The TVR trial design consisted of three arms: in the first arm, patients received triple therapy for 12 weeks followed by SOC treatment for 36 weeks; in the second arm, patients received lead-in treatment with SOC for 4 weeks, followed by triple therapy for 12 weeks, ending with SOC treatment for 32 weeks; the third arm consisted of SOC treatment for 48 weeks.17 In both trials, an SVR occurred significantly more frequently in those who received the triple therapy regimens than in those who received the SOC therapy. In the BOC trial (RESPOND-2 Trial), the SVR rates were 66% and 59% in the two triple therapy arms compared to 21% in the control arm, prior relapsers achieving higher SVR rates (75% and 69%, respectively) than prior partial responders (52% and 40%, respectively) compared to the rates attained in the SOC arm (29% and 7%, respectively); null responders were excluded from this trial (Table 3 and Fig.

iple therapy arms compared to 21% in the control arm, prior relapsers achieving higher SVR rates (75% and 69%, respectively) than prior partial responders (52% and 40%, respectively) compared to the rates attained in the SOC arm (29% and 7%, respectively); null responders were excluded from this trial (Table 3 and Fig. 5).13 Similarly, the SVR rates in the TVR trial (REALIZE Study) were 64% and 66% in the TVR-containing arms (83% and 88% in relapsers, 59% and 54% in partial responders, and 29% and 33% in null responders) and 17% in the control arm (24% in relapsers, 15% in partial responders and 5% in null responders) (Fig. 6).17 Thus, the response to the triple therapy regimen in both the BOC and TVR trials was influenced by the outcome of the previous treatment with PegIFN and RBV which highlights the importance of reviewing old treatment records to document previous treatment response. In the BOC trial, the SVR rate was higher in those who were relapsers than in those who were partial responders. In the TVR trial also, the highest SVR rate occurred in prior relapsers, a lower rate in partial responders, and the lowest rate in null responders (defined as patients who had <2 log10 decline in HCV RNA at week 12 of prior treatment) (Table 3 and Fig. 6).17

ho were relapsers than in those who were partial responders. In the TVR trial also, the highest SVR rate occurred in prior relapsers, a lower rate in partial responders, and the lowest rate in null responders (defined as patients who had <2 log10 decline in HCV RNA at week 12 of prior treatment) (Table 3 and Fig. 6).17 Fig. 5 Sustained virological response (SVR) rates, overall and among relapsers and partial responders, in treatment experienced patients with genotype 1 chronic HCV infection: Boceprevir (BOC) plus peginterferon and ribavirin (PR) versus standard of care (SOC). All patients were first treated with PegIFN and RBV for 4 weeks as lead-in therapy followed by one of 3 regimens: (1) BOC/PR48 triple therapy for 44 weeks. (2) BOC RGT triple therapy for 32 weeks if an eRVR was achieved (undetecatble HCV RNA at week 8 and 12). If an eRVR was not achieved, but HCV RNA became undetectable at week 12, BOC was stopped at week 32 and patients received an additional 12 weeks of SOC treatment (total 48 weeks of therapy). (3) SOC treatment consisted of PegIFN and RBV administered for 48 weeks.13 Fig. 6 Sustained virological response (SVR) rates, overall and among relapsers, partial responders, and null responders, in treatment-experienced patients with genotype 1 chronic HCV infection. T12PR48: Telaprevir (TVR) plus peginterferon and ribavirin (PR) administered for 12 weeks followed by 36 PR for 12 weeks followed by PR for 32 weeks; SOC consisted of PegIFN and RBV administered for 48 weeks.17

ial responders, and null responders, in treatment-experienced patients with genotype 1 chronic HCV infection. T12PR48: Telaprevir (TVR) plus peginterferon and ribavirin (PR) administered for 12 weeks followed by 36 PR for 12 weeks followed by PR for 32 weeks; SOC consisted of PegIFN and RBV administered for 48 weeks.17 Table 3 Comparison of Protease Inhibitors in Combination with Peginterferon Alfa (PegIFN) and Ribavirin (RBV) in Treatment-Experienced Patients

ial responders, and null responders, in treatment-experienced patients with genotype 1 chronic HCV infection. T12PR48: Telaprevir (TVR) plus peginterferon and ribavirin (PR) administered for 12 weeks followed by 36 PR for 12 weeks followed by PR for 32 weeks; SOC consisted of PegIFN and RBV administered for 48 weeks.17 Table 3 Comparison of Protease Inhibitors in Combination with Peginterferon Alfa (PegIFN) and Ribavirin (RBV) in Treatment-Experienced Patients Variable Boceprevir (BOC)13 Telaprevir (TVR)17 Study design RCT RCT 4-Week lead-in PegIFN/RBV Yes Yes/No* Duration of triple therapy 32 or 44 weeks in combination with PegIFN and RBV** 12 weeks followed by 36 weeks of PegIFN and RBV*** Response-guided therapy (RGT) Yes No Eligible for RGT (%) 46 NA Prior response to PegIFN/RBV (%) Relapser 64 53 Partial responder 36 19 Null responder NA 28 Efficacy, SVR (%) Relapser BOC/PR48: 75 T12/PR48: 83 BOC/RGT: 69 LI-T12/PR48: 88 PR48: 29 PR48: 24 Partial responder BOC/PR48: 52 T12/PR48: 59 BOC/RGT: 40 LI-T12/PR48: 54 PR48: 7 PR48: 15 Null responder NA T12/PR48: 29 LI-T12/PR48: 33 PR48: 5 Overall relapse (%) 12-15 NA Relapser NA T12/PR48: 7 LI-T12/PR48: 7 PR48: 65 Partial responder NA T12/PR48: 21 LI-T12/PR48: 25 PR48: 0 Null responder NA T12/PR48: 27 LI-T12/PR48: 25 PR48: 60 Adverse events Discontinuation (%) 8-12 NA SAE (%) 10-14 11-15 Adverse event more frequent in triple therapy arm Anemia, dysgeusia Rash, anemia, pruritus, nausea, diarrhea NA, not available; PR, peginterferon plus ribavirin; RCT, randomized, controlled trial; SAE, serious adverse event; SVR, sustained virological response.

rse events Discontinuation (%) 8-12 NA SAE (%) 10-14 11-15 Adverse event more frequent in triple therapy arm Anemia, dysgeusia Rash, anemia, pruritus, nausea, diarrhea NA, not available; PR, peginterferon plus ribavirin; RCT, randomized, controlled trial; SAE, serious adverse event; SVR, sustained virological response. * A lead-in arm was included in the telaprevir retreatment trial but the FDA approved regimen did not include a lead-in strategy. † The BOC trial design included a 4-week lead-in phase of PegIFN and RBV and compared response-guided triple therapy and a fixed duration triple therapy given for 44 weeks to peginterferon and ribavirin therapy. BOC/RGT response-guided therapy patients who achieved an eRVR (undetectable HCV RNA at week 8 [week 4 of triple therapy]) received an additional 24 weeks (total 32 weeks of therapy). If an eRVR was not achieved but HCV RNA became undetectable at week 12, BOC was stopped at week 32, and patients received an additional 12 weeks of SOC treatment (total 48 weeks of therapy). BOC/PR48: 4-week lead-in with peginterferon and ribavirin followed by a fixed duration of triple therapy for 44 weeks; PR48: PegIFN and RBV administered for 48 weeks.

NA became undetectable at week 12, BOC was stopped at week 32, and patients received an additional 12 weeks of SOC treatment (total 48 weeks of therapy). BOC/PR48: 4-week lead-in with peginterferon and ribavirin followed by a fixed duration of triple therapy for 44 weeks; PR48: PegIFN and RBV administered for 48 weeks. ‡ Telaprevir (TVR) plus peginterferon and ribavirin (PR) administered with and without a 4 week SOC treatment lead in versus standard of care (SOC). T12PR48: TVR administered for 12 weeks followed by 36 weeks of peginterferon and ribavirin; LI-T12/PR48: peginterferon and ribavirin for 4 weeks followed by TVR plus peginterferon and ribavirin for 12 weeks, followed by peginterferon and ribavirin for 32 weeks; PR48: peginterferon and ribavirin administered for 48 weeks.

2PR48: TVR administered for 12 weeks followed by 36 weeks of peginterferon and ribavirin; LI-T12/PR48: peginterferon and ribavirin for 4 weeks followed by TVR plus peginterferon and ribavirin for 12 weeks, followed by peginterferon and ribavirin for 32 weeks; PR48: peginterferon and ribavirin administered for 48 weeks. Thus, the decision to re-treat patients should depend on their prior response to PegIFN and RBV, as well as on the reasons for why they may have failed, such as inadequate drug dosing or side effect management. Relapsers and partial responder patients can expect relatively high SVR rates to re-treatment with a PI-containing triple regimen and should be considered candidates for re-treatment. The decision to re-treat a null responder should be individualized, particularly in patients with cirrhosis, because fewer than one-third of null responder patients in the TVR trial achieved an SVR; there are no comparable data for BOC because null responders were excluded from treatment. In addition, a majority of null responders developed antiviral resistance. The FDA label, however, indicates that BOC can be used in null responders but, given the lack of definitive information from phase 3 data, caution is advised in the use of BOC in null responders until further supportive evidence becomes available. Accordingly, any potential for benefit from treating nonresponders must be weighed against the risk of development of antiviral resistance and of serious side effects, and the high cost of therapy.

phase 3 data, caution is advised in the use of BOC in null responders until further supportive evidence becomes available. Accordingly, any potential for benefit from treating nonresponders must be weighed against the risk of development of antiviral resistance and of serious side effects, and the high cost of therapy. Response-guided therapy, based on achieving an eRVR, was evaluated for retreatment in the BOC trial. Shortening the duration of therapy from the standard 48 weeks to 36 weeks in patients who received triple therapy and who achieved an eRVR (which for this drug was defined as HCV RNA negative weeks 8 through 20) did not significantly lower the SVR rate (59% for RGT versus 66% for fixed duration treatment).13 In patients with cirrhosis, however, the SVR rate was statistically lower in those who received RGT therapy than in those who were treated for the full 48-week duration (35% versus 77%, respectively).13 The emergence of BOC resistant variants was more common among patients who responded poorly to interferon treatment (<1 log decline in HCV RNA level) during the lead-in phase and who were treated with RGT compared to those with >1 log decline in HCV RNA level and treated for 48 weeks (32% and 8%, respectively).13 There are no comparable data for RGT using TVR. Nonetheless, SVR rates are at least as high in relapsers as in treatment-naïve patients, and TVR exposure is 12 weeks with both RGT and 48-week treatment options. Accordingly, although there are no direct data to support the recommendation that relapsers could be treated with TVR using an RGT approach, the FDA does endorse such a recommendation, as is the case for BOC.

psers as in treatment-naïve patients, and TVR exposure is 12 weeks with both RGT and 48-week treatment options. Accordingly, although there are no direct data to support the recommendation that relapsers could be treated with TVR using an RGT approach, the FDA does endorse such a recommendation, as is the case for BOC. Utility of Lead-In There is uncertainty about the benefit of a lead-in phase. Theoretically, a PegIFN and RBV lead-in phase may serve to improve treatment efficacy by lowering HCV RNA levels which would allow for steady-state PegIFN and RBV levels at the time the PI is dosed, thereby reducing the risk of viral breakthrough or resistance. In addition, a lead-in strategy does allow for determination of interferon responsiveness and on-treatment assessment of SVR in patients receiving either BOC or TVR. Patients whose interferon response is suboptimal, defined as a reduction of the HCV RNA level of less than 1 log during the 4-week lead-in, have lower SVR rates than do patients with a good IFN response during lead-in treatment.12 Nevertheless, the addition of BOC to poor responders during lead-in still leads to significantly improved SVR rates (28% to 38% compared with 4% if a PI is not added) and thus a poor response during the lead-in phase should not be used to deny patients access to PI therapy.

good IFN response during lead-in treatment.12 Nevertheless, the addition of BOC to poor responders during lead-in still leads to significantly improved SVR rates (28% to 38% compared with 4% if a PI is not added) and thus a poor response during the lead-in phase should not be used to deny patients access to PI therapy. A direct comparison of the lead-in and non-lead-in groups in the BOC phase 2 study, however, did not show a significant difference in SVR rates for either the 28 week regimen, 56% and 54%, or the 48 week regimen, 75% and 67%, treated with and without lead-in, respectively.11 Combining data across all treatment groups in the phase 2 trial demonstrated a trend for a higher rate of virological breakthrough in the BOC-treated patients without a lead-in, 9%, than in those who received lead-in treatment, 4%, (P = 0.06). However, because all the phase 3 data were based on the lead-in strategy, until there is evidence to the contrary, BOC should be used with a 4-week lead-in. A lead-in strategy was not evaluated in the phase 3 TVR treatment-naïve trial, and therefore no recommendation can be made for this drug. Recommendations: For treatment-experienced patients: Re-treatment with boceprevir or telaprevir, together with peginterferon alfa and weight-based ribavirin, can be recommended for patients who had virological relapse or were partial responders after a prior course of treatment with standard interferon alfa or peginterferon alfa and/or ribavirin (Class 1, Level A).

ents: Re-treatment with boceprevir or telaprevir, together with peginterferon alfa and weight-based ribavirin, can be recommended for patients who had virological relapse or were partial responders after a prior course of treatment with standard interferon alfa or peginterferon alfa and/or ribavirin (Class 1, Level A). Re-treatment with telaprevir, together with peginterferon alfa and weight-based ribavirin, may be considered for prior null responders to a course of standard interferon alfa or peginterferon alfa and/or weight-based ribavirin (Class 2b, Level B.) Response-guided therapy of treatment-experienced patients using either a boceprevir- or telaprevir-based regimen can be considered for relapsers (Class 2a, Level B for boceprevir; Class 2b, Level C for telaprevir), may be considered for partial responders (Class 2b, Level B for boceprevir; Class 3, Level C for telaprevir), but cannot be recommended for null responders (Class 3, Level C). Patients re-treated with boceprevir plus peginterferon alfa and ribavirin who continue to have detectable HCV RNA > 100 IU at week 12 should be withdrawn from all therapy because of the high likelihood of developing antiviral resistance (Class 1, Level B). Patients re-treated with telaprevir plus peginterferon alfa and ribavirin who continue to have detectable HCV RNA > 1,000 IU at weeks 4 or 12 should be withdrawn from all therapy because of the high likelihood of developing antiviral resistance (Class 1, Level B).

Patients re-treated with boceprevir plus peginterferon alfa and ribavirin who continue to have detectable HCV RNA > 100 IU at week 12 should be withdrawn from all therapy because of the high likelihood of developing antiviral resistance (Class 1, Level B). Patients re-treated with telaprevir plus peginterferon alfa and ribavirin who continue to have detectable HCV RNA > 1,000 IU at weeks 4 or 12 should be withdrawn from all therapy because of the high likelihood of developing antiviral resistance (Class 1, Level B). Adverse Events Adverse events occurred more frequently in patients treated with PIs than in those treated with PegIFN and RBV therapy alone. In the BOC trials, anemia and dysgeusia were the most common adverse events, whereas in the TVR trials, rash, anemia, pruritus, nausea, and diarrhea developed more commonly among those who received TVR than who received SOC therapy.12,16 In the phase 3 TVR trials, a rash of any severity was noted in 56% of patients who received a TVR-based regimen compared to 32% of those who received a PegIFN and RBV regimen.16 The rash was typically eczematous and maculopapular in character, consistent with a drug-induced eruption. In most patients, the rash was mild to moderate in severity but was severe (involving >50% of the body surface area) in 4% of cases. The development of rash necessitated discontinuation of TVR in 6% and of the entire regimen in 1% of the cases. The Stevens Johnson Syndrome or Drug-Related Eruption with Systemic Symptoms (DRESS) occurred in <1% of subjects but at a higher frequency than generally observed for other drugs. The response of the rash to local or systemic treatment with corticosteroids and oral antihistamines is uncertain. Pruritus, commonly but not always associated with rash, was noted in ∼50% of patients who received TVR therapy.16

urred in <1% of subjects but at a higher frequency than generally observed for other drugs. The response of the rash to local or systemic treatment with corticosteroids and oral antihistamines is uncertain. Pruritus, commonly but not always associated with rash, was noted in ∼50% of patients who received TVR therapy.16 Anemia developed among recipients of both PIs. Hemoglobin decreases below 10 g/dL (grade 2 toxicity) occurred in 49% of patients who received a BOC regimen compared to 29% of those who received the SOC regimen, whereas 9% had a hemoglobin decline of <8.5 g/dL (grade 3 toxicity).12 Among patients treated with T12PR, hemoglobin levels of <10 g/dL were observed in 36% of patients compared to in 14% of patients who received SOC, and 9% had hemoglobin decreases to <8.5 g/dL.16 Because hematopoietic growth factors were not permitted during the TVR trials, there was a 5%-6% higher rate of treatment discontinuation among those who developed anemia than among those who did not. However, neither anemia nor RBV dose reduction adversely affected the SVR rate. Of note is that in the BOC trial, SVR rates in patients managed by RBV dose reduction alone were comparable to those in patients managed with erythropoietin therapy.23 Similarly, in the TVR trials, dose reduction of RBV had no effect on SVR rates, and therefore dose reduction should be the initial response to management of anemia.24 Because the duration of BOC therapy (24 to 44 weeks) is longer than the duration of TVR therapy (12 weeks), the frequency of anemia is likely to be greater in BOC-containing regimens, leading to more RBV dose reductions and consideration of erythropoietin use. However, the potential benefits of erythropoietin must be weighed against its potential side effects, the fact that its use in HCV therapy is not approved by the FDA, and its considerable cost. If a PI treatment–limiting adverse event occurs, PegIFN and RBV can be continued provided that an on-treatment response had occurred. There are no data to help guide substitution of one for the other HCV PI. If a patient has a serious adverse reaction related to PegIFN and/or RBV, the PegIFN and/or RBV dose should be reduced or discontinued. If either PegIFN and/or RBV are discontinued, the HCV PI should be stopped. Additional information on management of other adverse events can be found in the package insert.

her HCV PI. If a patient has a serious adverse reaction related to PegIFN and/or RBV, the PegIFN and/or RBV dose should be reduced or discontinued. If either PegIFN and/or RBV are discontinued, the HCV PI should be stopped. Additional information on management of other adverse events can be found in the package insert. Drug–Drug Interactions Because patients with CHC frequently receive medications in addition to those used to treat HCV infection, and because the PIs can inhibit hepatic drug-metabolizing enzymes such as cytochrome P450 2C (CYP2C), CYP3A4, or CYP1A, both BOC and TVR were studied for potential interactions with a number of drugs likely to be coadministered. These included statins, immune suppressants, drugs used to treat HIV coinfection, opportunistic infections, mood disorders, and drug addiction support medications. Both BOC and TVR, were noted to cause interactions with several of the drugs examined, either increasing or decreasing pharmacokinetic parameters. It is particularly important, therefore, that the medical provider review this information as listed in the package insert for each of the drugs before starting treatment for CHC. This information can be obtained at the FDA Web site: http://www.accessdata.fda.gov/scripts/cder/drugsatfda/index.cfm. Other helpful sites are: http//:222.drug-interactions.com and http://www.hep-druginteractions.org

is information as listed in the package insert for each of the drugs before starting treatment for CHC. This information can be obtained at the FDA Web site: http://www.accessdata.fda.gov/scripts/cder/drugsatfda/index.cfm. Other helpful sites are: http//:222.drug-interactions.com and http://www.hep-druginteractions.org Viral Resistance and Monitoring Emergence of antiviral-resistant variants during PI-based therapy has been observed during all trials and is associated with virological failure and relapse (Tables 2 and 3). Mutations that confer either high or low level resistance to BOC and TVR cluster around the catalytic site of the NS3/4A serine protease. Similar variants were detected in both BOC and TVR-treated subjects, suggesting that some degree of cross-resistance exists between the two PIs. In both phase 3 studies, sequence analysis of the NS3/4A region was performed in all subjects at baseline and for all subjects who failed to achieve an SVR. Antiviral resistant variants were detected in a small proportion of patients at baseline, 7% in the BOC studies and 5% in the TVR trials, but did not appear to impact response to either PI.25,26 Therefore, there is currently no clinical indication for baseline resistance testing.

subjects who failed to achieve an SVR. Antiviral resistant variants were detected in a small proportion of patients at baseline, 7% in the BOC studies and 5% in the TVR trials, but did not appear to impact response to either PI.25,26 Therefore, there is currently no clinical indication for baseline resistance testing. Among treatment-naïve patients receiving a BOC regimen, antiviral resistant variants developing during treatment were observed overall in 16% of patients (Table 2).12 During treatment, TVR-associated antiviral variants occurred in 12% of treatment-naïve patients and 22% of treatment-experienced patients (Tables 2 and 3).16,17 A majority (80%-90%) of patients who experienced virological breakthrough or incomplete virological suppression on therapy, or virological relapse after discontinuation of PI therapy, were found to have antiviral resistant variants. In the BOC studies, poor response to interferon (<1 log decline in HCV RNA during the lead-in phase) was associated with a higher rate of development of resistance.12 Among TVR-treated patients, population sequencing has suggested that high-level resistance develops more commonly when virological failure occurs during the initial 12 weeks of treatment, whereas low-level resistance variants are more likely when virological failure occurs later, during treatment with PegIFN and RBV alone. These observations highlight the importance of response to interferon for the prevention of emergence of antiviral resistance.

failure occurs during the initial 12 weeks of treatment, whereas low-level resistance variants are more likely when virological failure occurs later, during treatment with PegIFN and RBV alone. These observations highlight the importance of response to interferon for the prevention of emergence of antiviral resistance. The clinical significance of antiviral resistant variants that emerge during PI therapy is uncertain. In longitudinal follow-up of patients enrolled in phase 2 trials, BOC-resistant variants were detected in 43% of subjects after 2 years of follow-up. Similarly, among patients with documented TVR-resistant variants who were enrolled in the TVR phase 3 trials, 40% still had detectable resistant variants after a median follow-up period of 45 weeks.27 In general, the decline or loss of variants appears to be related to their level of fitness.

fter 2 years of follow-up. Similarly, among patients with documented TVR-resistant variants who were enrolled in the TVR phase 3 trials, 40% still had detectable resistant variants after a median follow-up period of 45 weeks.27 In general, the decline or loss of variants appears to be related to their level of fitness. Further data are needed to determine whether selection of these variants during and after PI therapy affects subsequent treatment choices. In phase 3 studies, the emergence of resistant variants and virological breakthrough was more common in patients infected with HCV subtype 1a than 1b, a result of a higher genetic barrier required for selection of resistant variants in HCV subtype 1b compared to 1a.28 Thus, HCV subtyping may play a role in helping to select future treatment regimens and predict the development of resistance. Finally, minimizing development of compensatory mutations would involve early discontinuation of PI therapy when antiviral therapy is unlikely to succeed. Although viral stop rules varied widely in the phase 2 and 3 trials, week 4 and 12 time points on triple therapy are still key decision points for stopping therapy based on HCV RNA levels. Current data suggest that for patients receiving BOC, therapy should be stopped at week 12 if the viral level is >100 IU/mL or >10-15 IU/mL at treatment week 24 and, for TVR, therapy should be stopped at either week 4 or 12 if the viral level is >1,000 IU/mL or if week 24 HCV RNA is detectable.

based on HCV RNA levels. Current data suggest that for patients receiving BOC, therapy should be stopped at week 12 if the viral level is >100 IU/mL or >10-15 IU/mL at treatment week 24 and, for TVR, therapy should be stopped at either week 4 or 12 if the viral level is >1,000 IU/mL or if week 24 HCV RNA is detectable. Recommendations: Patients who develop anemia on protease inhibitor-based therapy for chronic hepatitis C should be managed by reducing the ribavirin dose (Class 2a, Level A). Patients on protease inhibitor-based therapy should undergo close monitoring of HCV RNA levels and the protease inhibitors should be discontinued if virological breakthrough (>1 log increase in serum HCV RNA above nadir) is observed (Class 1, Level A). Patients who fail to have a virological response, who experience virological breakthrough, or who relapse on one protease inhibitor should not be re-treated with the other protease inhibitor (Class 2a, Level C).

Patients on protease inhibitor-based therapy should undergo close monitoring of HCV RNA levels and the protease inhibitors should be discontinued if virological breakthrough (>1 log increase in serum HCV RNA above nadir) is observed (Class 1, Level A). Patients who fail to have a virological response, who experience virological breakthrough, or who relapse on one protease inhibitor should not be re-treated with the other protease inhibitor (Class 2a, Level C). Role of IL28B Testing in Decision to Treat and Selection of Therapeutic Regimen The likelihood of achieving an SVR with PegIFN and RBV and of spontaneous resolution of HCV infection differ depending on the nucleotide sequence near the gene for IL28B or lambda interferon 3 on chromosome 19.18,19 One single-nucleotide polymorphism that is highly predictive is detection of the C or T allele at position rs12979860.18 The CC genotype is found more than twice as frequently in persons who have spontaneously cleared HCV infection than in those who had progressed to CHC. Among persons with genotype 1 chronic HCV infection who are treated with PegIFN and RBV, SVR is achieved in 69%, 33%, and 27% of Caucasians who have the CC, CT, and TT genotypes, respectively; among black patients, SVR rates were 48%, 15%, and 13% for CC, CT, and TT genotypes, respectively.29 The predictive value of IL28B genotype testing for SVR is superior to that of the pretreatment HCV RNA level, fibrosis stage, age, and sex, and is higher for HCV genotype 1 virus than for genotypes 2 and 3 viruses.29,30 There are other polymorphisms near the gene for IL28B that also predict SVR, including detection of the G or T allele at position rs8099917, where T is the favorable genotype, and essentially provides the same information in Caucasians as C at rs12979860.31,32

V genotype 1 virus than for genotypes 2 and 3 viruses.29,30 There are other polymorphisms near the gene for IL28B that also predict SVR, including detection of the G or T allele at position rs8099917, where T is the favorable genotype, and essentially provides the same information in Caucasians as C at rs12979860.31,32 In one study, as well as in preliminary analyses of the phase 3 registration data, IL28B genotype remained predictive of SVR even in persons taking BOC or TVR.33 In Caucasian patients randomized in the SPRINT 2 trial to take BOC for 48 weeks, SVR was achieved by 80%, 71%, and 59% of patients with CC, CT, and TT genotypes, respectively.34 In Caucasian patients randomized in the ADVANCE trial to take TVR for 12 weeks, SVR was achieved by 90%, 71%, and 73% of patients with CC, CT, and TT genotypes, respectively.35 IL28B genotype also predicts the likelihood of qualifying for RGT. In treatment-naïve Caucasian patients randomized in SPRINT 2 to BOC, the week 8 HCV RNA threshold was achieved in 89% and 52% of patients with CC and CT/TT genotypes, respectively.34 In treatment-naïve Caucasian patients randomized in the ADVANCE study to TVR, eRVR was achieved in 78%, 57%, and 45% of patients with CC, CT, and TT genotypes, respectively.35 Although the IL28B genotype provides information regarding the probability of SVR and abbreviated therapy that may be important to provider and patient, there are insufficient data to support withholding PIs from persons with the favorable CC genotype because of the potential to abbreviate therapy and the trend for higher SVR rates observed in the TVR study. In addition, the negative predictive value of the T allele with PI-inclusive therapy is not sufficiently high to restrict therapy for all patients, because SVR was achieved by more than half of Caucasians with the TT genotype.34,35

tial to abbreviate therapy and the trend for higher SVR rates observed in the TVR study. In addition, the negative predictive value of the T allele with PI-inclusive therapy is not sufficiently high to restrict therapy for all patients, because SVR was achieved by more than half of Caucasians with the TT genotype.34,35 In summary, these data indicate that IL28B genotype is a significant pretreatment predictor of response to therapy. Consideration should be given to ordering the test when it is likely to influence either the physician's or patient's decision to initiate therapy. There are insufficient data to determine whether IL28B testing can be used to recommend selection of SOC over a PI-based regimen with a favorable genotype (CC) and in deciding upon the duration of therapy with either regimen. Recommendation: IL28B genotype is a robust pretreatment predictor of SVR to peginterferon alfa and ribavirin as well as to protease inhibitor triple therapy in patients with genotype 1 chronic hepatitis C virus infection. Testing may be considered when the patient or provider wish additional information on the probability of treatment response or on the probable treatment duration needed (Class 2a, Level B).

lfa and ribavirin as well as to protease inhibitor triple therapy in patients with genotype 1 chronic hepatitis C virus infection. Testing may be considered when the patient or provider wish additional information on the probability of treatment response or on the probable treatment duration needed (Class 2a, Level B). Special Populations There is a paucity of information for many of the subgroups with the greatest unmet need for treatment (e.g., patients coinfected with HIV and HCV, those with decompensated cirrhosis, and those after liver transplantation). Data from phase 1 and 2 trials have provided interim information that may guide related treatment issues. BOC and TVR undergo extensive hepatic metabolism, BOC primarily by way of the aldoketoreductase (AKR) system but also by the cytochrome P450 enzyme system, whereas TVR is metabolized only by the cytochrome P450 enzyme system, and the main route of elimination is via the feces with minimal urinary excretion. Thus, no dose adjustment of BOC or TVR is required in patients with renal insufficiency. No clinically significant differences in pharmacokinetic parameters were observed with varying degrees of chronic liver impairment in patients treated with BOC and therefore, no dosage adjustment of this drug is required in patients with cirrhosis and liver impairment. Although TVR may be used to treat patients with mild hepatic impairment (Child-Turcotte-Pugh class A, score 5 or 6), it should not be used in HCV-infected patients with moderate to severe hepatic impairment, because no pharmacokinetic or safety data are available regarding its use in such patients. As noted above, BOC and TVR are both inhibitors of CYP3A4, and concomitant administration of medications known to be CYP3A4 substrates should be done with caution and under close clinical monitoring. Pharmacokinetic interactions have particular implications in HIV-coinfected and transplant populations, where drug–drug interactions will complicate treatment paradigms, so that any use of BOC or TVR in transplant or HIV-coinfected populations of patients with HCV should be done with caution and under close clinical monitoring. TVR and BOC are not recommended for use in children and adolescents younger than 18 years of age, because the safety and efficacy has not been established in this population. Thus, whereas BOC and TVR have great promise for improved SVR in special populations, many complex treatment issues remain to be evaluated in further phase 2 and 3 testing.

nded for use in children and adolescents younger than 18 years of age, because the safety and efficacy has not been established in this population. Thus, whereas BOC and TVR have great promise for improved SVR in special populations, many complex treatment issues remain to be evaluated in further phase 2 and 3 testing. This practice guideline was produced in collaboration with the Practice Guidelines Committee of the AASLD. This committee provided extensive peer review of the manuscript. Members of the Practice Guidelines Committee include Jayant A. Talwalkar, M.D., M.P.H. (Chair), Adrian M. Di Bisceglie, M.D. (Board Liaison), Jeffrey H. Albrecht, M.D., Hari S. Conjeevaram, M.D., M.S., Amanda DeVoss, M.M.S., PA-C, Hashem B. El-Serag, M.D., M.P.H., David A. Gerber, M.D., Christopher Koh, M.D., Kevin Korenblat, M.D., Raphael B. Merriman, M.D., M.R.C.P.I., Gerald Y. Minuk, M.D., Robert S. O'shea, M.D., Michael K. Porayko, M.D., Adnan Said, M.D., Benjamin L. Shneider, M.D., and Tram T. Tran, M.D. External review was provided by Gary Davis, M.D., Chair, AASLD HCV Special Interest Group, and the American College of Gastroenterology, the Infectious Diseases Society of America, and the National Viral Hepatitis Roundtable. Senior officials at the Division of Viral Hepatitis of the Centers for Disease Control and Prevention, the Office of HIV/AIDS Policy, U.S. Department of Health and Human Services, and the Public Health Strategic Health Care Group, U.S. Department of Veterans Affairs were provided an opportunity to review and comment on the manuscript. Abbreviations AKRaldoketoreductase

This practice guideline was produced in collaboration with the Practice Guidelines Committee of the AASLD. This committee provided extensive peer review of the manuscript. Members of the Practice Guidelines Committee include Jayant A. Talwalkar, M.D., M.P.H. (Chair), Adrian M. Di Bisceglie, M.D. (Board Liaison), Jeffrey H. Albrecht, M.D., Hari S. Conjeevaram, M.D., M.S., Amanda DeVoss, M.M.S., PA-C, Hashem B. El-Serag, M.D., M.P.H., David A. Gerber, M.D., Christopher Koh, M.D., Kevin Korenblat, M.D., Raphael B. Merriman, M.D., M.R.C.P.I., Gerald Y. Minuk, M.D., Robert S. O'shea, M.D., Michael K. Porayko, M.D., Adnan Said, M.D., Benjamin L. Shneider, M.D., and Tram T. Tran, M.D. External review was provided by Gary Davis, M.D., Chair, AASLD HCV Special Interest Group, and the American College of Gastroenterology, the Infectious Diseases Society of America, and the National Viral Hepatitis Roundtable. Senior officials at the Division of Viral Hepatitis of the Centers for Disease Control and Prevention, the Office of HIV/AIDS Policy, U.S. Department of Health and Human Services, and the Public Health Strategic Health Care Group, U.S. Department of Veterans Affairs were provided an opportunity to review and comment on the manuscript. Abbreviations AKRaldoketoreductase BOCboceprevir CYPcytochrome P450 DAAdirect-acting antivirals eRVRextended rapid virological response FDAFood and Drug Administration HCVhepatitis C virus IL28Binterleukin-28B NS3/4AHCV nonstructural protein 3/4A PegIFNpeginterferon PIprotease inhibitor RBVribavirin RGTresponse-guided therapy RVRrapid virological response SOCstandard of care

Abbreviations AKRaldoketoreductase BOCboceprevir CYPcytochrome P450 DAAdirect-acting antivirals eRVRextended rapid virological response FDAFood and Drug Administration HCVhepatitis C virus IL28Binterleukin-28B NS3/4AHCV nonstructural protein 3/4A PegIFNpeginterferon PIprotease inhibitor RBVribavirin RGTresponse-guided therapy RVRrapid virological response SOCstandard of care SVRsustained virological response TVRtelaprevir

Although foreign pathogens and their products, pathogen-associated molecular pattern (PAMP) molecules, have been shown to activate the innate immune system, the mechanisms by which damaged tissues notify the immune system remain to be fully elucidated. The recognition of a group of endogenous damage-associated molecular pattern (DAMP) molecules that serve a similar function to PAMPs has provided a framework for understanding the overlap between the inflammatory responses activated by pathogens and injury. It appears that certain pattern recognition receptors (PRRs), such as the family of Toll-like receptors (TLRs), serve as a common pathway for immune recognition of microbial invasion and tissue injury. By recognizing either PAMP or DAMP molecules, PRRs alert the host to tissue damage by activating the innate immune system. Initially, this process is manifested by the production of inflammatory mediators that allow the host to respond appropriately to infectious or noninfectious insults. When excessive, this inflammatory response can contribute to severe organ damage and dysfunction.

o tissue damage by activating the innate immune system. Initially, this process is manifested by the production of inflammatory mediators that allow the host to respond appropriately to infectious or noninfectious insults. When excessive, this inflammatory response can contribute to severe organ damage and dysfunction. Several DAMPs have been identified, including heat shock proteins, S100 proteins, hyaluronan, heparin sulfate, RNA, DNA, and high-mobility group box 1 (HMGB1).1 Endogenous DAMPs are elevated and contribute to poor outcomes in several inflammatory models, both infectious and noninfectious, including sepsis,2 acute lung injury,3 pancreatitis,4 burns,5 and trauma.6, 7 These molecules are recognized by PRRs on various cell types, and in turn, drive the inflammatory response. We previously identified the endogenous DAMP HMGB and the PRR TLR4 to be major mediators of organ damage in hepatic ischemia/reperfusion (I/R).8 Recently, DNA and TLR9 have also been shown to play critical roles.9 Although studies by our group and by others have investigated the function of circulating HMGB1 and DNA in I/R injury, the involvement of histone proteins, which are closely associated with both HMGB1 and DNA in the nucleus, has not yet been examined. Traditionally, histones have been examined in the context of regulating nuclear architecture; however, recent reports have also examined the role of extracellular histones.10–12 Most importantly, Xu et al.13 identified extracellular histones as major mediators of death in sepsis.

the nucleus, has not yet been examined. Traditionally, histones have been examined in the context of regulating nuclear architecture; however, recent reports have also examined the role of extracellular histones.10–12 Most importantly, Xu et al.13 identified extracellular histones as major mediators of death in sepsis. I/R injury consists of a series of pathophysiological events that involve deprivation of blood and oxygen followed by their restoration. Liver I/R injury occurs unavoidably after elective liver resection, organ transplantation, trauma, and hypovolemic shock.14 Subsequent organ damage occurs as a result of both direct cellular damage, including hepatocyte necrosis and apoptosis, as well as delayed organ dysfunction from activation of the innate immune system and propagation of the inflammatory response.15–18 Although much is known, the roles of endogenous DAMPs and PRRs have not been fully delineated.

curs as a result of both direct cellular damage, including hepatocyte necrosis and apoptosis, as well as delayed organ dysfunction from activation of the innate immune system and propagation of the inflammatory response.15–18 Although much is known, the roles of endogenous DAMPs and PRRs have not been fully delineated. The aim of this study was to determine the initial mechanisms by which ischemic tissues alert the innate immune system in sterile tissue damage. We used I/R as a model of acute, noninfectious tissue injury to the liver. We demonstrate that extracellular histones are released from liver parenchymal cells in vitro and in vivo. Although exogenous administration of histones exacerbates I/R injury, neutralizing antibodies prevent hepatocellular damage and suppress the activation of the inflammatory cascade. These effects are mediated through TLR9 and MyD88. Furthermore, in addition to directly activating TLR9, extracellular histones also enhance nucleic acid–mediated inflammation through TLR9. Materials and Methods Materials Antibodies to histone H3 (LG2-1) and histone H4 (BWA-3) were provided by Jun Xu.13 Immunoglobulin G (IgG) (item I5006), calf thymus histones (H9250), and Dnase I (D5319) were obtained from Sigma-Aldrich. TLR9 agonist CpG (ODN1826), antagonist CpG (ODN2088), or control CpG were obtained from Invitrogen. A commercial kit based on histone release was purchased from Roche.

A-3) were provided by Jun Xu.13 Immunoglobulin G (IgG) (item I5006), calf thymus histones (H9250), and Dnase I (D5319) were obtained from Sigma-Aldrich. TLR9 agonist CpG (ODN1826), antagonist CpG (ODN2088), or control CpG were obtained from Invitrogen. A commercial kit based on histone release was purchased from Roche. Animals Male wild-type (C57BL/6) mice (8-12 weeks old) were purchased from Jackson ImmunoResearch Laboratories. TLR9CpG/CpG mutant, MyD88−/− and MyD88+/+ mice were provided by Timothy Billiar (University of Pittsburgh Medical Center, Pittsburgh, PA). Animal protocols were approved by the Animal Care and Use Committee of the University of Pittsburgh, and the experiments were performed in adherence to National Institutes of Health guidelines for the use of laboratory animals. Liver I/R A nonlethal model of segmental (70%) hepatic warm ischemia and reperfusion was used as described.19 Experimental Design Mice received anti-histone H3 or H4 antibodies (20 mg/kg),13 or control IgG (Sigma-Aldrich) intravenously 30 minutes before ischemia. Histones (5 mg/kg and 25 mg/kg), CpG, ODN2088 (100 μg per mouse), or phosphate-buffered saline (PBS) were injected intraperitoneally immediately after ischemia. Sham animals underwent anesthesia, laparotomy, and exposure of the portal triad without hepatic ischemia. Animals were sacrificed at predetermined time points (1-6 hours).

mg/kg and 25 mg/kg), CpG, ODN2088 (100 μg per mouse), or phosphate-buffered saline (PBS) were injected intraperitoneally immediately after ischemia. Sham animals underwent anesthesia, laparotomy, and exposure of the portal triad without hepatic ischemia. Animals were sacrificed at predetermined time points (1-6 hours). Isolation and Culture of Hepatocytes and Nonparenchymal Cells Hepatocytes and nonparenchymal cells (NPCs) were isolated from normal wild-type (C57BL/6) mice as described.8, 20 Hepatocytes (3 × 106) and NPCs (50 × 106) were plated as described.8, 20 In Vitro Coculture Assays Wild-type hepatocytes were rendered necrotic by incubation at 60°C for 60 minutes as described.9 Supernatants from necrotic hepatocytes were harvested after a 12-hour incubation period at 37°C and were used as conditioned media in subsequent coculture assays. Liver Damage Assessment Serum alanine aminotransferase (sALT) levels were measured using the DRI-CHEM 4000 Chemistry Analyzer System (Heska). Sodium Dodecyl Sulfate–Polyacrylamide Gel Electrophoresis and Western Blotting Western blot analysis for histone H3 and H4 (Cell Signaling Technology), functional TLR9 (1:1,000; eBioscience), extracellular signal-regulated kinase, p38, c-Jun N-terminal kinase, and nuclear factor κB (NF-κB) p65 (1:1,000; Cell Signaling Technology) were performed as described.21, 22

trophoresis and Western Blotting Western blot analysis for histone H3 and H4 (Cell Signaling Technology), functional TLR9 (1:1,000; eBioscience), extracellular signal-regulated kinase, p38, c-Jun N-terminal kinase, and nuclear factor κB (NF-κB) p65 (1:1,000; Cell Signaling Technology) were performed as described.21, 22 SYBR Green Real-Time Reverse-Transcription Polymerase Chain Reaction Total RNA was extracted from the liver using the RNeasy Mini Kit (Qiagen). Messenger RNA (mRNA) for tumor necrosis factor α (TNF-α), interleukin-6 (IL-6), and β-actin was quantified in duplicate by SYBR Green (reverse-transcription polymerase chain reaction [PCR]). PCR reaction mixture was prepared using SYBR Green PCR Master Mix (PE Applied Biosystems) using described primers.21 Immunofluorescent Staining Cells were incubated with the specific primary antibodies for histone H3 and H4 (1:500; Abcam) in 1% bovine serum albumin for 1 hour, washed 4 times, and incubated with secondary antibody (1:500; AlexaFluor 488 goat anti-rabbit, Invitrogen). F-actin was stained with rhodamine phalloidin (Invitrogen). Cells were mounted with Vecta-Shield Mounting media with DAPI nuclear stain. Slides were viewed with Olympus Provis and Leica TSL-SL immunofluorescent microscopes.

mes, and incubated with secondary antibody (1:500; AlexaFluor 488 goat anti-rabbit, Invitrogen). F-actin was stained with rhodamine phalloidin (Invitrogen). Cells were mounted with Vecta-Shield Mounting media with DAPI nuclear stain. Slides were viewed with Olympus Provis and Leica TSL-SL immunofluorescent microscopes. Coimmunoprecipitation Immunoprecipitation was performed with 1 μg of antibodies against histone H4 in 300 μg whole lysate protein or 40 μL of serum diluted in immunoprecipitation buffer [50 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid, 0.5% Nonidet P-40, 150 mM NaCl, 10% glycerol, 1 mM ethylene diamine tetraacetic acid]. Normal rabbit IgG was used as a negative control. Precleared lysates were incubated with anti-TLR9 overnight, and then incubated for 2 hours with protein A/G-agarose. Samples were washed four times with PBS and subjected to western blot analysis. Statistical Analysis Results are expressed as the mean ± SE. Statistical analysis was performed using the Student t test or one-way analysis of variance. All statistical analyses were performed using Sigma Stat version 3.5 (Systat Software, Inc.). Graphs were generated using Sigma Plot version 10 (Systat Software, Inc.). P < 0.05 was considered statistically significant.

n ± SE. Statistical analysis was performed using the Student t test or one-way analysis of variance. All statistical analyses were performed using Sigma Stat version 3.5 (Systat Software, Inc.). Graphs were generated using Sigma Plot version 10 (Systat Software, Inc.). P < 0.05 was considered statistically significant. Results Neutralizing Extracellular Histones Is Protective in Hepatic I/R Injury Nuclear histone proteins associate closely with DAMP molecules HMGB1 and DNA, both of which are known to impact I/R injury. Thus, to determine whether extracellular histones contribute to hepatic organ damage after hepatic I/R, neutralizing antibodies to histone H3 or H4 were administered to mice before I/R. Anti-H3 and anti-H4 antibodies had no effect on sham-treated animals but conferred significant protection after I/R (Fig. 1A,B). IgG-treated animals exhibited 18.7 ± 4.5% necrotic hepatocytes compared with 6.7 ± 2.1% in the anti-H3 group or 5.8 ± 2.6% in the anti-H4 group (Fig. 1B).

ere administered to mice before I/R. Anti-H3 and anti-H4 antibodies had no effect on sham-treated animals but conferred significant protection after I/R (Fig. 1A,B). IgG-treated animals exhibited 18.7 ± 4.5% necrotic hepatocytes compared with 6.7 ± 2.1% in the anti-H3 group or 5.8 ± 2.6% in the anti-H4 group (Fig. 1B). Fig. 1 Pretreatment with neutralizing antibodies to histones protects against liver I/R injury. (A) Sham or I/R-treated mice were given anti-histone H3 or anti-histone H4 antibody (20 mg/kg body weight) or control antibody intravenously 30 minutes before ischemia. Data represent the mean ± SE (n = 6 mice per group). *P < 0.05. (B) Quantification of necrotic hepatocytes in hematoxylin and eosin–stained liver sections from control and anti-histone antibody-treated animals 6 hours after reperfusion. The graph is representative of liver sections from six mice per group. (C) Hepatic TNF-α and IL-6 mRNA expression after 6 hours of I/R. Results are expressed as relative increase of mRNA expression compared with sham-treated animals. Data represent the mean ± SE (n = 6 mice per group). *P < 0.05. Tissue levels of TNF-α and IL-6, two cytokines shown to be important in hepatic I/R,21 were also significantly decreased in the anti-histone antibody–treated I/R groups (Fig. 1C), with parallel results obtained from mouse serum (data not shown). Taken together, these results suggest that histones play a key role in the inflammatory response and organ injury following hepatic I/R.

ortant in hepatic I/R,21 were also significantly decreased in the anti-histone antibody–treated I/R groups (Fig. 1C), with parallel results obtained from mouse serum (data not shown). Taken together, these results suggest that histones play a key role in the inflammatory response and organ injury following hepatic I/R. Extracellular Histones Exacerbate Hepatic I/R Injury To further explore the role of extracellular histones in I/R, exogenous histones were administered to mice. We found that a dose of 25 mg/kg did not cause elevation of sALT in sham-treated animals 7 hours after injection; consequently, this dose was used for all experiments. We also chose to administer histones immediately after ischemia, because we hypothesized that this approach was most physiologically relevant; additionally, we found that preinjection of histones 60 minutes prior to ischemia had little effect of sALT levels (Supporting Fig. 1). sALT levels measured 6 hours after reperfusion in mice that were given exogenous histones immediately after ischemia were significantly greater than in PBS-treated animals (Fig. 2A) (sALT 5,579 ± 340 IU/L versus 2,400 ± 87 IU/L). Liver histology also confirmed the hepatotoxic effect of exogenous histone administration. Histone-injected mice displayed 50.1 ± 10.2% necrotic hepatocytes versus PBS-treated mice at 13.4 ± 7% (Fig. 2B). Parallel tissue levels (Fig. 2C) and serum levels (data not shown) of TNF-α and IL-6 were observed. Because the dose of histones was nontoxic in sham-treated animals, but intensifying in I/R, these results suggest that administration of exogenous histones exacerbates organ injury only after I/R, serving to amplify the inflammatory cascade.

s (Fig. 2C) and serum levels (data not shown) of TNF-α and IL-6 were observed. Because the dose of histones was nontoxic in sham-treated animals, but intensifying in I/R, these results suggest that administration of exogenous histones exacerbates organ injury only after I/R, serving to amplify the inflammatory cascade. Fig. 2 Treatment of exogenous histones mixture exacerbate liver I/R injury. (A) Sham-treated mice and mice that underwent ischemia and 1, 3, and 6 hours of reperfusion were treated with a nonlethal dose of exogenous histone mixture (25 mg/kg body weight) or vehicle PBS immediately after ischemia. sALT levels were analyzed. Data represent the mean ± SE (n = 12 mice per group). *P < 0.05. **P < 0.01. (B) Quantification of necrotic hepatocytes in hematoxylin and eosin–stained liver tissue. The graph is representative of liver sections from six mice per group. (C) Hepatic TNF-α and IL-6 mRNA. Data represent the mean ± SE (n = 6 mice per group). *P < 0.05.

the mean ± SE (n = 12 mice per group). *P < 0.05. **P < 0.01. (B) Quantification of necrotic hepatocytes in hematoxylin and eosin–stained liver tissue. The graph is representative of liver sections from six mice per group. (C) Hepatic TNF-α and IL-6 mRNA. Data represent the mean ± SE (n = 6 mice per group). *P < 0.05. Extracellular Histones Are Released from the Liver In Vivo After Hepatic I/R Injury and from Hepatocytes After Hypoxia In Vitro To investigate whether histone-dependent injury was associated with extracellular changes in histone protein levels in vivo, enzyme-linked immunosorbent assay was performed on serum from animals subjected to liver I/R. After I/R, extracellular histone protein expression increased (Fig. 3A). We have shown hepatocytes to be the major source of the DAMP molecule HMGB1 after I/R,23 and necrotic hepatocytes are hypothesized to be a key source of endogenous DNA24; therefore, we explored hepatocytes as a potential source of endogenous extracellular histones. Immunofluorescent staining was performed in sham-treated livers and livers that underwent I/R. Histones localized to the nucleus of hepatocytes in sham-treated animals, and after I/R, histone-positive staining was observed in the cytoplasm of hepatocytes along with several areas that lacked nuclear histone staining (Fig 3Bb). We observed similar results in hepatocytes in vitro after hypoxia (Fig 3D,E), suggesting this translocation results in the heightened release of extracellular histones from hepatocytes to worsen I/R injury.

observed in the cytoplasm of hepatocytes along with several areas that lacked nuclear histone staining (Fig 3Bb). We observed similar results in hepatocytes in vitro after hypoxia (Fig 3D,E), suggesting this translocation results in the heightened release of extracellular histones from hepatocytes to worsen I/R injury. Fig. 3 Extracellular histones are released from hepatocytes after hypoxia in vitro and from the liver in vivo after hepatic I/R. (A) Systemic histones levels were assessed by way of serum enzyme-linked immunosorbent assay. Data represent the mean ± SE (n = 6 mice per group). *P < 0.05. (B) Immunofluorescent stain of histone H3 from sections of normal liver and (C) and liver 6 hours I/R (original magnification ×400). Images are representative liver sections from six mice per group. Red, histone H3; blue, nuclei; green, F-actin. (D) Cultured mouse hepatocytes were exposed to hypoxia (1% O2) from 0 to 48 hours. Media were subjected to western blot analysis of histone H3 and H4. The blots shown are representative of three experiments with similar results. (E) Immunofluorescent stain of histone H3 and H4 from cultured mouse hepatocytes were exposed to hypoxia (1% O2) overnight (original magnification ×600). Images are representative of three experiments with similar results. Green, histone H3 or H4; blue, nuclei; red, F-actin.

e of three experiments with similar results. (E) Immunofluorescent stain of histone H3 and H4 from cultured mouse hepatocytes were exposed to hypoxia (1% O2) overnight (original magnification ×600). Images are representative of three experiments with similar results. Green, histone H3 or H4; blue, nuclei; red, F-actin. Exogenous Histones Modulate Inflammatory Signaling Pathways To determine how released extracellular histones might affect the inflammatory response to hepatic I/R injury, the role of histones in the activation of mitogen-activated protein kinases was evaluated. After 1 hour of I/R, phosphorylation of c-Jun N-terminal kinase, p38, and extracellular signal-regulated kinase increased, and these effects were further augmented with histone treatment (Fig 4A). Conversely, histone neutralization with anti-H4 or anti-H3 decreased phosphorylation of these proteins after I/R (Fig 4B). Finally, we observed an increase in NF-κB activation after 1 hour of I/R in liver tissue after histone treatment by phosphorylation at serine 536 of the p65 subunit (Fig. 4C).

e treatment (Fig 4A). Conversely, histone neutralization with anti-H4 or anti-H3 decreased phosphorylation of these proteins after I/R (Fig 4B). Finally, we observed an increase in NF-κB activation after 1 hour of I/R in liver tissue after histone treatment by phosphorylation at serine 536 of the p65 subunit (Fig. 4C). Fig. 4 Extracellular histones modulate inflammatory signaling pathways. (A) Mitogen-activated protein kinase activation was determined in sham-treated mice and mice that underwent ischemia and 1 hour of reperfusion. Animals were treated with exogenous histones mixture or vehicle PBS. (B) Mitogen-activated protein kinase activation was determined. Hepatic protein lysates from ischemic lobes were obtained; each lane represents a separate animal. The blots shown are representative of three experiments with similar results. (C) Phosphorylation at serine 536 of the p65 subunit of NF-κB after 1 hour of I/R.

Mitogen-activated protein kinase activation was determined. Hepatic protein lysates from ischemic lobes were obtained; each lane represents a separate animal. The blots shown are representative of three experiments with similar results. (C) Phosphorylation at serine 536 of the p65 subunit of NF-κB after 1 hour of I/R. Extracellular Histones Mediate Hepatic I/R Injury Through TLR9 Recent studies have demonstrated that TLR9-deficient mice are significantly protected from hepatic I/R injury,9 and our group has reported the role of TLR4.8 Thus, it is known that pattern recognition receptors play a critical role in sterile inflammation initiated by I/R injury. To determine whether the PRRs TLR9, TLR4, or TLR2 are involved in histone recognition during hepatic I/R injury, exogenous histones were administered to TLR9 mutant (TLR9CpG/CpG), TLR4 knockout (KO), TLR2 KO, or TRIF KO mice and their wild-type (WT; C57BL/6) counterparts. As expected, significant protection was observed in TLR9 mutant mice compared with their WT counterparts. However, exogenous histones failed to enhance liver damage in TLR9 mutant mice, whereas damage was significantly increased in TLR9 WT mice (Fig 5A). Whereas TLR9 mutant mice failed to respond to exogenous histones, we observed increased damage in TLR4 KO, TLR2 KO, and TRIF KO mice (Supporting Fig. 2). Additionally, administration of anti-H3 or anti-H4 to TLR9 mutant mice after I/R also failed to confer further protection, and examination of liver histology (Fig 5B) and tissue cytokine levels (Fig 5D) corroborated our sALT findings.

observed increased damage in TLR4 KO, TLR2 KO, and TRIF KO mice (Supporting Fig. 2). Additionally, administration of anti-H3 or anti-H4 to TLR9 mutant mice after I/R also failed to confer further protection, and examination of liver histology (Fig 5B) and tissue cytokine levels (Fig 5D) corroborated our sALT findings. Fig. 5 Extracellular histones mediate hepatic I/R injury through TLR9. (A) Serum ALT levels in TLR9 mutant and WT mice after I/R with either anti-histone antibodies or exogenous histone administration. Data represent the mean ± SE (n = 4-6 mice per group). *P < 0.05. (B) Hematoxylin and eosin–stained liver sections (original magnification ×100). Images are representative liver sections from six mice per group. The dashed line indicates the necrotic area. (C) Hepatic TNF-α and IL-6 mRNA in TLR9 mutant and WT mice after histone administration. (D) Hepatic TNF-α and IL-6 mRNA expression after histone neutralization. Results are expressed as the relative increase of mRNA expression compared with sham-treated animals. Data represent the mean ± SE (n = 4-6 mice per group). *P < 0.05. N.S., not significant.

in TLR9 mutant and WT mice after histone administration. (D) Hepatic TNF-α and IL-6 mRNA expression after histone neutralization. Results are expressed as the relative increase of mRNA expression compared with sham-treated animals. Data represent the mean ± SE (n = 4-6 mice per group). *P < 0.05. N.S., not significant. MyD88 is downstream of TLR9 signaling, hence MyD88 KO and WT mice were treated with exogenous histones after I/R. Extracellular histones also failed to enhance liver damage in MyD88 KO mice (Fig. 6A,B), revealing that histones function through MyD88–TLR9 signaling. To further confirm the role of TLR9 in histone recognition, we used the TLR9 antagonist ODN2088. Exogenous histones were administered to both ODN2088-treated and ODN2088 control–treated animals. Histones had no effect in TLR9 antagonist–treated WT mice (Fig. 6C). Liver histology was also consistent with these results (Fig. 6D).

r confirm the role of TLR9 in histone recognition, we used the TLR9 antagonist ODN2088. Exogenous histones were administered to both ODN2088-treated and ODN2088 control–treated animals. Histones had no effect in TLR9 antagonist–treated WT mice (Fig. 6C). Liver histology was also consistent with these results (Fig. 6D). Fig. 6 Extracellular histones-mediated hepatic I/R injury involves TLR9 signaling cascade. (A) Serum ALT levels in MyD88 KO and WT mice after 6 hours of I/R with or without histone administration. Data represent the mean ± SE (n = 4-6 mice per group). *P < 0.05. N.S., not significant. (B) Hematoxylin and eosin–stained liver sections (magnification ×100). Images are representative liver sections from six mice per group. The dashed line indicates the necrotic area. (C) sALT levels in ODN2088-treated mice after 6 hours of I/R. Data represent the mean ± SE (n = 4-6 mice per group). *P < 0.05. N.S., not significant. (D) Hematoxylin and eosin–stained liver sections from ODN2088-treated mice (magnification ×100). Images are representative liver sections from six mice per group. The dashed line indicates that necrotic area. (E) TLR9 activation by western blot analysis. The blots shown are representative of three experiments with similar results. (F) Coimmunopreciptation of histone H4 and cleaved TLR9 from liver cell lysated.

0). Images are representative liver sections from six mice per group. The dashed line indicates that necrotic area. (E) TLR9 activation by western blot analysis. The blots shown are representative of three experiments with similar results. (F) Coimmunopreciptation of histone H4 and cleaved TLR9 from liver cell lysated. The truncated form of TLR9, rather than the full-length form, functions to recruit MyD88 upon activation.22 Therefore, cleaved TLR9 was assessed after 1 hour and 6 hours of I/R. Cleaved TLR9 was enhanced compared with sham-treated mice and increased by way of histone treatment at both time points (Fig. 6E), demonstrating that histones further can enhance the activation of TLR9. Finally, functional TLR9 was immunoprecipitated from whole liver tissue lysates in mice treated with PBS or exogenous histones. A physical interaction was detected between TLR9 and histone H4 in both PBS-treated and exogenous histone treated–mice after hepatic I/R, whereas no interaction was observed in sham-treated mice (Fig. 6F), further implicating the TLR9 signaling pathway in the mechanism of extracellular histone– mediated I/R injury. Taken together, these results suggest that TLR9 and its downstream signaling molecule MyD88 are involved in histone recognition during sterile inflammatory injury induced by I/R.

n sham-treated mice (Fig. 6F), further implicating the TLR9 signaling pathway in the mechanism of extracellular histone– mediated I/R injury. Taken together, these results suggest that TLR9 and its downstream signaling molecule MyD88 are involved in histone recognition during sterile inflammatory injury induced by I/R. Extracellular Histones Enhance Nucleic Acid–Mediated Inflammation Through TLR9 DNA from necrotic hepatocytes has recently been shown to increase hepatic NPC cytokine production in a TLR9-dependent manner.9 We used conditioned media from necrotic hepatocytes to stimulate hepatic NPCs. IL-6 mRNA significantly increased in NPCs after treatment with conditioned media, and this effect was reduced by treatment with DNAse. We observed a similarly significant effect when anti-H4 was added to the conditioned media, suggesting that histones also contribute to the stimulatory potential of necrotic cell supernatants (Fig 7A). Furthermore, the combination of DNAse and anti-H4 completely abolished IL-6 production. We also treated NPCs with low, nonstimulatory doses of CpG (ODN1826), histones, or both. With CpG or histones treatment alone, we observed minimal increase in proinflammatory cytokine IL-6 mRNA. However, cotreatment with CpG and histones led to a dramatic increase in IL-6 mRNA. Thus, we found a synergistic effect between nonstimulatory doses of CpG-DNA and exogenous histones (Fig 7B). Taken together, these results suggest that the inflammatory effects of histones may also occur by enhancing the DNA–TLR9 innate immune activation.

t with CpG and histones led to a dramatic increase in IL-6 mRNA. Thus, we found a synergistic effect between nonstimulatory doses of CpG-DNA and exogenous histones (Fig 7B). Taken together, these results suggest that the inflammatory effects of histones may also occur by enhancing the DNA–TLR9 innate immune activation. Fig. 7 Extracellular histones enhances the nucleic acid–mediated damage after hepatic I/R. (A) IL-6 mRNA expression was obtained in NPCs cocultured overnight with media from necrotic hepatocytes. NPCs were treated with vehicle PBS, Dnase I, or anti-histone H4 antibody. Results are expressed as the relative increase of mRNA expression compared with PBS treatment. Data represent the mean ± SE and are representative of three experiments with similar results. *P < 0.05. (B) IL-6 mRNA expression was observed in NPCs that were treated with vehicle PBS, CpG, exogenous histones, or both. Results are expressed as relative increase of mRNA expression compared with PBS treatment. Data represent the mean ± SE and are representative of three experiments with similar results. *P < 0.05.

< 0.05. (B) IL-6 mRNA expression was observed in NPCs that were treated with vehicle PBS, CpG, exogenous histones, or both. Results are expressed as relative increase of mRNA expression compared with PBS treatment. Data represent the mean ± SE and are representative of three experiments with similar results. *P < 0.05. Discussion Nuclear histones are small, highly abundant proteins traditionally known as an essential component of nucleosomes in eukaryotic cells.25 Recently, Xu et al.13 showed that extracellular histones are involved in the pathogenesis of sepsis by contributing to endothelial cytotoxicity and triggering an inflammatory and thrombotic response in vivo. In addition, the cytotoxicity of histones appears to be regulated by proteolytic inactivation by activated protein C. Interestingly, activated protein C is protective in cardiac and liver I/R models,26, 27 and the proteolytic inactivation of histones represents a novel mechanism by which activated protein C protects against end organ damage. TLR9 is an intracellular molecule that functions as a sensor of DNA, and it was originally reported that TLR9 KO mice failed to respond to bacterial DNA, which is rich in unmethylated CpG.28 Subsequently, TLR9 was shown to recognize endogenous DNA from mammalian cells.29 Recently, the critical role of TLR9 expressed on liver nonparenchymal cells was reported in the pathogenesis of liver I/R injury.9 DNA released from necrotic hepatocytes is thought to be the activating ligand of TLR9 signaling, although this has not yet been substantiated in vivo.

endogenous DNA from mammalian cells.29 Recently, the critical role of TLR9 expressed on liver nonparenchymal cells was reported in the pathogenesis of liver I/R injury.9 DNA released from necrotic hepatocytes is thought to be the activating ligand of TLR9 signaling, although this has not yet been substantiated in vivo. Previously, we reported that HMGB and TLR4 are critical in I/R injury.21, 23 Here, we hypothesized that histones play a similar role as HMGB1 during liver I/R. We only observed hepatotoxic effects after I/R, suggesting that histones serve as cofactors to amplify other circulating pathogenic signals. This concept has been explored for other DAMPs, including HMGB1; HMGB1 combines with several endogenous and exogenous danger signals to amplify their effect.30 Thus, histones may also function to amplify the effect of circulating DNA through TLR9 activation. Regardless, the role of endogenous molecules released from necrotic and/or apoptotic cells remains uncertain. For acute, sterile inflammation induced by I/R, we hypothesize that the innate immune response is dependent on in vivo circulating complexes, including DNA, histones, HMGB1, and others (Fig. 8). Fig. 8 Hepatic I/R induces release of extracellular histones from hepatocytes. The model shows the release of extracellular histones after hepatic I/R.

Previously, we reported that HMGB and TLR4 are critical in I/R injury.21, 23 Here, we hypothesized that histones play a similar role as HMGB1 during liver I/R. We only observed hepatotoxic effects after I/R, suggesting that histones serve as cofactors to amplify other circulating pathogenic signals. This concept has been explored for other DAMPs, including HMGB1; HMGB1 combines with several endogenous and exogenous danger signals to amplify their effect.30 Thus, histones may also function to amplify the effect of circulating DNA through TLR9 activation. Regardless, the role of endogenous molecules released from necrotic and/or apoptotic cells remains uncertain. For acute, sterile inflammation induced by I/R, we hypothesize that the innate immune response is dependent on in vivo circulating complexes, including DNA, histones, HMGB1, and others (Fig. 8). Fig. 8 Hepatic I/R induces release of extracellular histones from hepatocytes. The model shows the release of extracellular histones after hepatic I/R. This study shows that endogenous serum histone levels increase significantly after I/R. However, whether extracellular histones are passively released from necrotic cells or actively secreted from damaged but viable cells is not yet known. Our group recently found that decreased nuclear histone deacetylase activity in hepatocytes after liver I/R contributed to the hyperacetylation and release of HMGB1.20 Further studies will determine whether posttranslational modifications of histones can contribute to extracellular histone-mediated toxicity.

yet known. Our group recently found that decreased nuclear histone deacetylase activity in hepatocytes after liver I/R contributed to the hyperacetylation and release of HMGB1.20 Further studies will determine whether posttranslational modifications of histones can contribute to extracellular histone-mediated toxicity. In conclusion, the present study shows that extracellular histones similar to HMGB1 and DNA contribute to hepatic I/R injury by functioning as DAMPs, activating immune responses leading to inflammation and organ damage. The protective effects of blocking extracellular histones and the detrimental effects of exogenous histones in hepatic I/R are dependent on the activation of TLR9 signaling. Extracellular histones may also mediate sterile inflammation after hepatic I/R injury by enhancing the DNA-activated TLR9 signaling cascade. Thus, neutralization of extracellular histones may be a novel, effective strategy to minimize organ damage in the setting of ischemic liver injury. We thank Nicole Hays for technical assistance in preparing the manuscript. We also thank Marc Monestier at Temple University for providing hybridoma cells for producing LG2-1 and BWA-3. Abbreviations DAMPdamage-associated molecular pattern HMGB1high-mobility group box 1 IgGimmunoglobulin G IL-6interleukin-6 I/Rischemia/reperfusion KOknockout mRNAmessenger RNA NF-κBnuclear factor κB NPCnonparenchymal cell PAMPpathogen-associated molecular pattern PBSphosphate-buffered saline PCRpolymerase chain reaction PRRpattern recognition receptor sALTserum alanine aminotransferase TLRToll-like receptor TNF-αtumor necrosis factor α WTwild-type

IgGimmunoglobulin G IL-6interleukin-6 I/Rischemia/reperfusion KOknockout mRNAmessenger RNA NF-κBnuclear factor κB NPCnonparenchymal cell PAMPpathogen-associated molecular pattern PBSphosphate-buffered saline PCRpolymerase chain reaction PRRpattern recognition receptor sALTserum alanine aminotransferase TLRToll-like receptor TNF-αtumor necrosis factor α WTwild-type Supplementary material Additional Supporting Information may be found in the online version of this article.

Pregnancy results in dramatic surges in hormones during gestational development, with increases in steroids such as estrogens, progestins, and the glucocorticoids. These steroids have been linked to the control and regulation of xenobiotic drug metabolism, primarily through activation of the nuclear xenobiotic receptors such as the pregnane X receptor (PXR)1-3 and the constitutive androstane receptor (CAR).4, 5 PXR and CAR have been classified as steroid and xenobiotic sensors, so it can be anticipated that significant fluctuations in the pregnancy hormones will modulate transcriptional control of target genes such as those involved in xenobiotic or drug metabolism.

(PXR)1-3 and the constitutive androstane receptor (CAR).4, 5 PXR and CAR have been classified as steroid and xenobiotic sensors, so it can be anticipated that significant fluctuations in the pregnancy hormones will modulate transcriptional control of target genes such as those involved in xenobiotic or drug metabolism. Clinical findings indicate that steroid fluctuations lead to changes in xenobiotic glucuronidation during pregnancy. For example, circulating unconjugated bilirubin is cleared from the circulation solely through UDP glucuronosyltransferase 1A1 (UGT1A1) metabolism.6 During pregnancy, total serum bilirubin (TSB) levels are lower in women,7 indicating that bilirubin metabolism is accelerated through induced UGT1A1. Labetalol, an antihypertensive agent, which is metabolized primarily by UGT1A1 glucuronidation, shows increased clearance in the second and third trimesters of pregnancy compared to the postpartum period.8 Lamotrigine, an antiepileptic agent metabolized by UGT1A3 and UGT1A4, has a 50% decreased elimination half-life with an increased clearance of over 200% during pregnancy, leading to a closely correlated higher incidence of epileptic seizures.9 Labetalol and lamotrigine clearance during pregnancy indicates that UGT1A1, UGT1A3, and UGT1A4 are induced and clearance is accelerated. It has been suggested that up-regulation of UGT1A4 during pregnancy may be mediated by 17β-estradiol and the estrogen receptor alpha (ERα).10 Clearly, hormonal sensors during pregnancy are leading to induction of human glucuronidation capacity.

s that UGT1A1, UGT1A3, and UGT1A4 are induced and clearance is accelerated. It has been suggested that up-regulation of UGT1A4 during pregnancy may be mediated by 17β-estradiol and the estrogen receptor alpha (ERα).10 Clearly, hormonal sensors during pregnancy are leading to induction of human glucuronidation capacity. The exact opposite is occurring during neonatal development, which is evident by the very high incidence of hyperbilirubinemia in newborn children. Because bilirubin is metabolized exclusively by UGT1A1,6 hyperbilirubinemia develops from the inability of liver glucuronidation to match the early rise in serum bilirubin that forms from the abundance of red blood cells needed to carry oxygen. Senescence of the erythrocytes leads to an accumulation of hemoglobin that is rapidly metabolized into bilirubin and released into the circulation, where it is transported to the liver for excretion following UGT1A1-dependent glucuronidation. Jaundice is directly linked to inadequate glucuronidation of serum bilirubin stemming from reduced expression of liver UGT1A1.11-13 It is unclear if the reduced expression of UGT1A1 in neonates is a controlled event through transcriptional silencing or simply a result of limited epigenetic factors that are eventually produced to positively regulate the UGT1A1 gene in a developmental fashion.

in stemming from reduced expression of liver UGT1A1.11-13 It is unclear if the reduced expression of UGT1A1 in neonates is a controlled event through transcriptional silencing or simply a result of limited epigenetic factors that are eventually produced to positively regulate the UGT1A1 gene in a developmental fashion. In this report it will be demonstrated that PXR is linked to both pregnancy-induced expression of the UGT1 locus as well as repression of the UGT1A1 gene in neonatal development. These findings were generated through the development of humanized UGT1 (hUGT1) mice,14, 15 which express the entire human UGT1 locus in a murine Ugt1-null background.16 Taking advantage of the power of reverse genetics, it will be shown that PXR plays a crucial role in pregnancy-induced glucuronidation in addition to the early development of hyperbilirubinemia in neonatal hUGT1 mice. Materials and Methods Animals The generation of Tg(UGT1A1*1)Ugt1−/− (hUGT1*1) and Tg(UGT1A1*28)Ugt1−/− (hUGT1*28) mice has been reported.15Pxr−/− mice were generated as described17 and Car−/− mice were generously provided by Dr. Masahiko Negishi (NIEHS). All genetically modified strains were bred for over five generations with C57BL/6 wildtype mice before inbreeding. To generate hUGT1/Pxr−/− mice, hUGT1*1 mice were crossed with Pxr−/− mice, producing Tg(UGT1A1*1)Ugt1+/−Pxr+/− mice. These mice were backcrossed in brother/sister matings to generate Tg(UGT1A1*1)Ugt−/−Pxr−/− (hUGT1*1/Pxr−/−) mice. The same breeding strategy was used to generate hUGT1*1/Car−/− mice.

e before inbreeding. To generate hUGT1/Pxr−/− mice, hUGT1*1 mice were crossed with Pxr−/− mice, producing Tg(UGT1A1*1)Ugt1+/−Pxr+/− mice. These mice were backcrossed in brother/sister matings to generate Tg(UGT1A1*1)Ugt−/−Pxr−/− (hUGT1*1/Pxr−/−) mice. The same breeding strategy was used to generate hUGT1*1/Car−/− mice. Primary Hepatocyte Isolation and PXR-Targeted Specific Small Interfering RNA (siRNA) Regulation Hepatocytes were isolated as described.14 The hepatocytes were then cultured in 6-well collagen-treated plates (Discovery Labware, Bedford, MA) in 2 mL of Dulbecco's modified Eagle's medium (DMEM) containing penicillin/streptomycin and supplemented with 10% fetal bovine serum. siRNA duplexes specific for mouse PXR were provided by Bioneer (Alameda, CA) and Santa Cruz Biotechnology. Four hours after primary hepatocytes were isolated from 14-day-old hUGT1*1 mice, cells were transfected in the presence of 20 nM of either siRNA or control RNA with Lipofectamine 2000 (Invitrogen) in a final volume of 0.5 mL of OPTI-MEM. After 5 hours cells were changed with fresh medium supplemented with 10% fetal bovine serum and penicillin-streptomycin. Forty-eight hours later, cells were used for RNA extraction. Reverse transcription (RT) and real-time polymerase chain reaction (Q-PCR) were carried out to examine gene expression levels of mouse Pxr, human UGT1A1 and mouse Cyp3a11.

olla, CA) programmed to take three fluorescence data points at the end of each annealing plateau. All PCR reactions were performed in triplicate as outlined.14Ct values were normalized to mouse cyclophilin (CPH). The specific primers used to quantitate the respective gene transcripts18 are listed in Supporting Table 1. Chromatin Immunoprecipitation (CHIP) CHIP analysis was performed using the modified protocol based on the EZ-CHIP kit (Millipore). Liver tissue (100 mg) was minced and cross-linked in DMEM (Invitrogen) containing 1% formaldehyde. The procedures for cell lysis and sonication to shear DNA were followed according to the manufacturer's protocol (EZ-CHIP kit, Millipore). One mL of cell extract was precleared by incubation with 60 μL of protein A Agarose/Salmon sperm DNA overnight at 4°C. The cleared cellular extract was incubated with anti-PXR antibody (Santa Cruz, sc-25381) for 2 hours at 4°C. Following precipitation with protein A agarose, the antibody-chromatin complex was then washed as outlined.19 The protein-DNA complexes were eluted in 200 μL elution buffer and DNA was then reverse cross-linked and released from the complex as indicated in the EZ-CHIP instructions. Following the DNA purification with spin columns (Qiagen), the purified DNA was further analyzed by PCR with a pair of primers (forward 5′-TTGTGGGGCAATACACTAGTA-3′, reverse 5′-GTCCGGGTTTCAGGTTATGTA-3′) for the amplification of the UGT1A1 promoter region containing the PXR binding site.3

ted in the EZ-CHIP instructions. Following the DNA purification with spin columns (Qiagen), the purified DNA was further analyzed by PCR with a pair of primers (forward 5′-TTGTGGGGCAATACACTAGTA-3′, reverse 5′-GTCCGGGTTTCAGGTTATGTA-3′) for the amplification of the UGT1A1 promoter region containing the PXR binding site.3 Results Expression of the Human UGT1A Genes in TgUGT1 Mice During Pregnancy Heterozygous female TgUGT1*28 mice were mated with wildtype mice and the presence of the vaginal plug in the morning was set as gestation day 1 (GD1). Starting at GD4, gravid mice were sacrificed at various times during pregnancy and liver microsomes used for western blot analysis where the UGT1A proteins were identified with a human anti-UGT1A antibody.20 Total UGT1A protein expression was induced significantly by the end of the second trimester (GD14), and remained at a high level of expression throughout the entire third trimester (Fig. 1A). This anti-UGT1A antibody also detects murine UGT1A proteins,16 but with wildtype mice as controls, no induction of murine UGT1A proteins during pregnancy was observed (Fig. 1B).

icantly by the end of the second trimester (GD14), and remained at a high level of expression throughout the entire third trimester (Fig. 1A). This anti-UGT1A antibody also detects murine UGT1A proteins,16 but with wildtype mice as controls, no induction of murine UGT1A proteins during pregnancy was observed (Fig. 1B). Fig. 1 Induction of the UGT1 locus in TgUGT1 mice. Age-matched female TgUGT1 mice were mated with wildtype mice. The following morning, female mice with the presence of a vaginal plug were removed, housed separately, and timed as gestation day 1. Pregnant mice were sacrificed at gestation days (GD) 4, 7, 10, 14, 16, and 19. Samples from at least three nonpregnant female TgUGT1 mice were used as controls. Liver microsomes and RNA from the nonpregnant control and pregnant mice were prepared. (A) Immunoblot detection of UGT1A proteins. Liver microsomes prepared from pregnant TgUGT1 mice at progressive stages of pregnancy were analyzed on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and the UGT1A proteins detected on immunoblots by using an anti-UGT1A antibody and anti-GAPDH antibody (Santa Cruz Biotechnology). Human liver microsomes (HLM) were used as a positive control. (B) Immunoblot of UGT1A proteins from liver microsomes prepared from female wildtype and TgUGT1 mice that were pregnant for 16 days. Control samples were prepared from nonpregnant wildtype and TgUGT1 mice. (C) Total RNA was isolated from liver samples taken from the pregnant TgUGT1 mice and used in RT and Q-PCR analysis to examine murine UGT1A1 RNA (Ugt1a1) expression and human UGT1A1 RNA expression. Student's t test was used to evaluate the statistical significance (**P < 0.01). (D) The same RNA samples were used to quantitate murine Ugt1a6 and human UGT1A6 gene expression (**P < 0.01, t test).

mice and used in RT and Q-PCR analysis to examine murine UGT1A1 RNA (Ugt1a1) expression and human UGT1A1 RNA expression. Student's t test was used to evaluate the statistical significance (**P < 0.01). (D) The same RNA samples were used to quantitate murine Ugt1a6 and human UGT1A6 gene expression (**P < 0.01, t test). Two of the UGTs that are highly conserved in function between mice and humans are UGT1A1 and UGT1A6. Expression of the murine Ugt1a1 and Ugt1a6 genes along with the human UGT1A1 and UGT1A6 genes were evaluated by RT and Q-PCR analysis using species specific primers for each of these genes. Throughout pregnancy, we observed no induction of murine Ugt1a1 or Ugt1a6 gene expression (Fig. 1C,D), findings that correlated with the lack of wildtype UGT1A protein expression. Consistent with induction of UGT1A protein in TgUGT1*28 mice during pregnancy, UGT1A1 and UGT1A6 gene expression is induced at GD14, with the induction being sustained throughout the remainder of the gestational period. Introduction of the human UGT1 locus and the UGT1A genes in TgUGT1*28 mice is regulated throughout pregnancy in a pattern that is not replicated by the murine Ugt1 locus. Emerging principles of regulatory evolution strongly favor genetic diversity in cis-regulatory DNA and not trans-regulation of gene expression to explain interspecies differences in gene expression.21, 22 The differences in transcriptional regulation between the murine and human UGT1 locus during pregnancy may be credited to important genetic differences in the regulatory regions of these genes.

n cis-regulatory DNA and not trans-regulation of gene expression to explain interspecies differences in gene expression.21, 22 The differences in transcriptional regulation between the murine and human UGT1 locus during pregnancy may be credited to important genetic differences in the regulatory regions of these genes. Induction of the UGT1 Locus in Humanized UGT1 Mice We generated hUGT1*1 and hUGT1*28 mice, which differ predominantly in expression levels of UGT1A1 in adult liver.15 Adult hUGT1*28 mice are hyperbilirubinemic, with TSB levels that average 1 mg/dL. During pregnancy and late gestation, the TSB levels in hUGT1*28 mice averaged 0.4 mg/dL, over 50% lower than in nonpregnant mice (Fig. 2A). The reduction in TSB can be accounted for by elevated levels of liver UGT1A1.

A1 in adult liver.15 Adult hUGT1*28 mice are hyperbilirubinemic, with TSB levels that average 1 mg/dL. During pregnancy and late gestation, the TSB levels in hUGT1*28 mice averaged 0.4 mg/dL, over 50% lower than in nonpregnant mice (Fig. 2A). The reduction in TSB can be accounted for by elevated levels of liver UGT1A1. Fig. 2 Regulation of the UGT1 locus during pregnancy. (A) Serum bilirubin levels in female hUGT1*28 mice were determined before pregnancy and at GD18 (*P < 0.05, Student's t test). (B) Quantitation of liver UGT1A gene expression by RT and Q-PCR in adult female and male mice. The relative values are expressed as a ratio value comparing female to male expression. (C) Quantitation of liver UGT1A gene expression in hUGT1*1 mice pregnant for 16 days. Total liver RNA was prepared and used in RT and Q-PCR analysis with specific UGT1A gene oligonucleotides. Fold induction was calculated relative to those values obtained in nonpregnant mice. (D) Immunoblot analysis of human UGT expression in nonpregnant and pregnant hUGT1*1 mice. Microsomes were prepared from nonpregnant and pregnant hUGT1*1 mice and subjected to 4%-12% polyacrylamide gel electrophoresis. Following blotting, protein expression was determined using a UGT1A antibody, or isozyme specific UGT1A1, UGT1A4, and UGT1A6 antibodies.

GT expression in nonpregnant and pregnant hUGT1*1 mice. Microsomes were prepared from nonpregnant and pregnant hUGT1*1 mice and subjected to 4%-12% polyacrylamide gel electrophoresis. Following blotting, protein expression was determined using a UGT1A antibody, or isozyme specific UGT1A1, UGT1A4, and UGT1A6 antibodies. In hUGT1*1 mice, expression of UGT1A1, -1A3, -1A4, and -1A6 are 2 to 4-fold greater in female liver than male liver (Fig. 2B). The sole exception to the female dominance is UGT1A9, which has minimal expression in female liver. Examination of the fold increase in gene expression of each of the UGT1A genes shows that UGT1A9 expression increases to nearly 70-fold over nonpregnant values (Fig. 2C). This large increase in UGT1A9 expression accounted for by fold induction during pregnancy results in part from the very low basal levels observed in nonpregnant female mice. Along with UGT1A9 gene expression, UGT1A1 and UGT1A6 gene expression are found to dominant the induction process during pregnancy. These increases are also reflected in microsomal protein abundance as determined by western blot analysis using isoform specific antibodies (Fig. 2D).

s observed in nonpregnant female mice. Along with UGT1A9 gene expression, UGT1A1 and UGT1A6 gene expression are found to dominant the induction process during pregnancy. These increases are also reflected in microsomal protein abundance as determined by western blot analysis using isoform specific antibodies (Fig. 2D). Pregnancy Steroids and Induction of UGT1A1 We hypothesized that hormone surges in late pregnancy play an important role in the gestational regulation of the human UGT1 locus. To examine if selective steroids are capable of regulating the UGT1A genes, primary hepatocytes from hUGT1*1 mice were isolated, placed in culture, and exposed to 17β-estradiol, progesterone, or the synthetic glucocorticoid, dexamethasone (DEX). We measured induction of the UGT1A1 gene because the other UGT1A genes are refractive to expression in hepatocytes in culture. Progesterone (50 μg/mL) and estradiol (20 μg/mL) exposure for 24 hours resulted in a minimal 2 to 3-fold induction of the UGT1A1 gene (Fig. 3A). These concentrations were found to be optimal for UGT1A1 induction. The synthetic glucocorticoid DEX (10 μM) induced UGT1A1 gene expression up to 60-fold (Fig. 3B). This induction of gene expression and UGT1A1 induction is dose-dependent, with transcriptional induction of the UGT1A1 gene in hepatocytes being substantial at 0.1 μM. Because corticosterone is the primary form of glucocorticoids in mouse, 20 μM of corticosterone was used to treat freshly isolated hepatocytes (Fig. 3C). Twenty-four hours after exposure, increased UGT1A1 protein levels were detected in whole cell lysates by western blot analysis, indicating that glucocorticoids play an important role in UGT1A gene expression.

of glucocorticoids in mouse, 20 μM of corticosterone was used to treat freshly isolated hepatocytes (Fig. 3C). Twenty-four hours after exposure, increased UGT1A1 protein levels were detected in whole cell lysates by western blot analysis, indicating that glucocorticoids play an important role in UGT1A gene expression. Fig. 3 Induction of UGT1A1 in primary hepatocytes by pregnancy-related hormones. (A) Primary hepatocytes were isolated from hUGT1*1 mice and treated with either estrogen (20 μg/mL), progesterone (50 μg/mL), or DEX (1 μM). After 24 hours RNA was isolated and UGT1A1 gene expression was determined by RT and Q-PCR. Ct values from RT and Q-PCR were normalized to the housekeeping gene CPH and calculated as fold induction over the expression levels of cells treated with vehicle control. (B) Primary hepatocytes were exposed to different concentrations of DEX. After 24 hours RNA was isolated and UGT1A1 gene expression monitored by RT and Q-PCR. A sample of total cell extract was also prepared and used for immunodetection of UGT1A1 and GAPDH by western blot analysis. (C) Isolated hepatocytes were exposed to DEX (1 μM) or corticosterone (20 μM) for 24 hours, followed by RT and Q-PCR analysis and western blot analysis to detect human UGT1A1 expression, and GAPDH blotting was used as loading control.

o prepared and used for immunodetection of UGT1A1 and GAPDH by western blot analysis. (C) Isolated hepatocytes were exposed to DEX (1 μM) or corticosterone (20 μM) for 24 hours, followed by RT and Q-PCR analysis and western blot analysis to detect human UGT1A1 expression, and GAPDH blotting was used as loading control. Xenobiotic Receptors and Induction of the UGT1 Locus During Pregnancy The progestins, corticosterone, and estradiols are low-affinity substrates for PXR23-25 and 17β-estradiol has been shown to activate CAR.4 Thus, experiments were conducted to examine precisely the role of PXR and CAR toward induction of the UGT1 locus during pregnancy.

s and Induction of the UGT1 Locus During Pregnancy The progestins, corticosterone, and estradiols are low-affinity substrates for PXR23-25 and 17β-estradiol has been shown to activate CAR.4 Thus, experiments were conducted to examine precisely the role of PXR and CAR toward induction of the UGT1 locus during pregnancy. To undertake these studies, hUGT1*1 mice were crossed with Car-null mice to create hUGT1*1/Car−/− mice. On GD16, liver samples from hUGT1*1/Car−/− mice were processed for total RNA along with microsomal extracts. RT and Q-PCR analysis for liver UGT1A gene products were conducted with specific oligonucleotide primers for each gene (Fig. 4A). In hUGT1*1/Car−/− mice, gestational induction of the UGT1A1, -1A3, and -1A6 genes was found to be similar to that observed in hUGT1*1 mice. Using western blot analysis to examine UGT1A1 expression in liver microsomes, UGT1A1 was induced during pregnancy in hUGT1*1 and hUGT1*1/Car−/− mice (Fig. 4B). However, CAR does play a role in the induction of UGT1A4 and UGT1A9. We observed approximately a 50% reduction in the 8-fold increase in UGT1A4 RNA accumulation observed in hUGT1*1 mice. When UGT1A9 expression was analyzed, the robust induction in hUGT1*1 pregnant mice was reduced over 75% during pregnancy in hUGT1*1/Car−/− mice, indicating an important role for CAR in the induction of the UGT1A9 gene (Fig. 4A).

a 50% reduction in the 8-fold increase in UGT1A4 RNA accumulation observed in hUGT1*1 mice. When UGT1A9 expression was analyzed, the robust induction in hUGT1*1 pregnant mice was reduced over 75% during pregnancy in hUGT1*1/Car−/− mice, indicating an important role for CAR in the induction of the UGT1A9 gene (Fig. 4A). Fig. 4 The impact of PXR and CAR deletion on gestational regulation of the human UGT1 locus in liver tissue. Humanized UGT1/Pxr−/− or hUGT1/Car−/− mice were obtained by backcrossing hUGT1*1 mice with Pxr−/− mice or Car−/− mice. Female hUGT1*1, hUGT1/Pxr−/−, and hUGT1/Car−/− mice at 8 weeks old were used for timed pregnancy experiments. Age-matched nonpregnant female mice from each strain were used as controls. Mice were sacrificed at GD16. (A) RNA was isolated from pooled liver samples followed by RT and Q-PCR analysis. Primers specific for human UGT1A1, UGT1A3, UGT1A4, UGT1A6, and UGT1A9 gene products were used (Supporting Table 1). Q-PCR results from pregnant mice were normalized by the housekeeping gene CPH and described as fold of induction over the nonpregnant control. (B) Western blot analysis using liver microsomes and anti-UGT1A1 and anti-GAPDH antibodies. (C) CHIP analysis of PXR associated with the human UGT1A1 gene in nonpregnant and 16 day pregnant hUGT1 mice, and DEX-treated adult females.

housekeeping gene CPH and described as fold of induction over the nonpregnant control. (B) Western blot analysis using liver microsomes and anti-UGT1A1 and anti-GAPDH antibodies. (C) CHIP analysis of PXR associated with the human UGT1A1 gene in nonpregnant and 16 day pregnant hUGT1 mice, and DEX-treated adult females. When hUGT1*1 mice are placed into a Pxr-null background, there was substantially reduced induction of each of the UGT1A genes during pregnancy when compared to expression in hUGT1*1 mice (Fig. 4A). The UGT1A1 gene, robustly induced around 15-fold in hUGT1*1 and hUGT1*1/Car−/− mice during pregnancy, displays reduced expression at GD16 in hUGT1*1/Pxr−/− mice. A similar pattern of expression was observed when UGT1A1 was detected by western blot analysis, showing little expression in hUGT1*1/Pxr−/− mice (Fig. 4B). An important role for PXR binding to the UGT1A1 gene during pregnancy was reinforced when we examined PXR binding by CHIP analysis to the PXR binding site that flanks the UGT1A1 promoter.3 PXR is activated during pregnancy and binds to the UGT1A1 gene as demonstrated by CHIP analysis (Fig. 4C), indicating that endogenous ligands are participating in regulation of this gene. Coupled with CHIP analysis showing induced binding of PXR to the UGT1A1 gene following DEX treatment (Fig. 4C) along with previous experiments demonstrating that PXR binding to this region of the UGT1A1 gene stimulates transactivation of the promoter,3 these findings confirm that induction of UGT1A1 is closely linked to activation of PXR during pregnancy.

duced binding of PXR to the UGT1A1 gene following DEX treatment (Fig. 4C) along with previous experiments demonstrating that PXR binding to this region of the UGT1A1 gene stimulates transactivation of the promoter,3 these findings confirm that induction of UGT1A1 is closely linked to activation of PXR during pregnancy. Glucocorticoids Induce the UGT1 Locus in a PXR-Dependent Fashion Because regulation of the UGT1 locus during pregnancy is linked to PXR, we examined if the genes associated with the UGT1 locus in humanized mice could be activated in a PXR-dependent fashion by glucocorticoids. We treated 8-week-old hUGT1*1 and hUGT1*1/Pxr−/− female mice by the intraperitoneal route with 20 mg/kg DEX for 4 days and measured UGT1A gene expression in liver 48 hours after treatment (Fig. 5). Each of the five UGT1A genes expressed in liver was induced in hUGT1*1 mice. Although activation of the glucocorticoid receptor by DEX has been shown to activate UGT1A1 reporter gene constructs in HepG2 cells,26 DEX treatment had no effect on induction of UGT1A1 in hUGT1*1/Pxr−/− mice, indicating that induction of the UGT1A genes by glucocorticoids is facilitated solely by activation of the PXR in vivo.

activation of the glucocorticoid receptor by DEX has been shown to activate UGT1A1 reporter gene constructs in HepG2 cells,26 DEX treatment had no effect on induction of UGT1A1 in hUGT1*1/Pxr−/− mice, indicating that induction of the UGT1A genes by glucocorticoids is facilitated solely by activation of the PXR in vivo. Fig. 5 Induction of the UGT1 locus by DEX in adult hUGT1*1 and hUGT1/Pxr−/− mice. Adult hUGT1*1 and hUGT1/Pxr−/− mice were treated with DEX by intraperitoneal injection for 4 consecutive days at 20 mg/kg per dose. Nontreated mice received solvent and were used as controls. Twenty-four hours after the last dose, liver RNA was prepared and RT and Q-PCR analysis of human UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A9, and murine Cyp3a11 gene expression was performed. Fold induction reflects the change in gene expression between solvent-treated and DEX-treated mice.

olvent and were used as controls. Twenty-four hours after the last dose, liver RNA was prepared and RT and Q-PCR analysis of human UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A9, and murine Cyp3a11 gene expression was performed. Fold induction reflects the change in gene expression between solvent-treated and DEX-treated mice. In neonatal hUGT1 mice, UGT1A1 expression controls the levels of TSB, with significant hyperbilirubinemia developing due to limited expression of hepatic UGT1A1. We examined if glucocorticoids could regulate UGT1A1 gene expression during the neonatal period in a PXR-dependent fashion. In hUGT1*1 mice, TSB peaks 14 days after birth and ranges from 12-15 mg/dL (Fig. 6A). Within the same litters, half of the newborn hUGT1*1 mice were treated with 8 mg/kg DEX by oral gavage, whereas the other half received just vehicle. Serum bilirubin levels were determined 48 hours after treatment. DEX treatment led to a dramatic reduction in TSB (Fig. 6A). Analysis of UGT1A1 gene expression in hepatic tissue following DEX treatment confirmed a 40-fold induction of RNA in hUGT1*1 neonatal mice (Fig. 6B). When we examined TSB levels in response to 8 mg/kg DEX treatment in hUGT1*1/Pxr−/− mice, the serum bilirubin levels did not change when compared to vehicle-treated neonatal mice (Fig. 6A). There was also no induction of hepatic UGT1A1 as determined by RT and Q-PCR analysis (Fig. 6B). Thus, DEX induces hepatic UGT1A1 leading to bilirubin metabolism through a PXR-dependent mechanism.

in hUGT1*1/Pxr−/− mice, the serum bilirubin levels did not change when compared to vehicle-treated neonatal mice (Fig. 6A). There was also no induction of hepatic UGT1A1 as determined by RT and Q-PCR analysis (Fig. 6B). Thus, DEX induces hepatic UGT1A1 leading to bilirubin metabolism through a PXR-dependent mechanism. Fig. 6 Induction of UGT1A1 by DEX in neonatal hUGT1*1 and hUGT1/Pxr−/− mice. Humanized UGT1*1 and hUGT1/Pxr−/− mice at 12 days after birth were treated by oral gavage with 8 mg/kg DEX for 3 days. (A) Serum bilirubin levels in untreated (Ctrl) and DEX-treated mice were evaluated in 14 day old mice (***P < 0.001, n.s., no significant difference, t test). (B) After DEX treatment, fold induction of UGT1A1 gene expression in liver was determined by RT and Q-PCR analysis. PXR Represses Human UGT1A1 Gene Expression in Neonatal hUGT1 Mice As we undertook these experiments, we also noted that TSB levels in neonatal hUGT1*1/Pxr−/− peak to only 4-6 mg/dL, almost 10 mg/dL lower than observed in hUGT1*1 mice (Fig. 6A). When we examined liver UGT1A1 gene expression at 14 days after birth, there was over a 10-fold increase in gene expression in hUGT1*1/Pxr−/− mice when compared to hUGT1*1 mice (Fig. 7A). This finding indicates that nonliganded PXR during development in hUGT1*1 mice is serving to repress liver UGT1A1 gene expression, because in PXR-deficient mice we observe induction of liver UGT1A1, which correlates with a reduction in serum bilirubin.

ion in hUGT1*1/Pxr−/− mice when compared to hUGT1*1 mice (Fig. 7A). This finding indicates that nonliganded PXR during development in hUGT1*1 mice is serving to repress liver UGT1A1 gene expression, because in PXR-deficient mice we observe induction of liver UGT1A1, which correlates with a reduction in serum bilirubin. Fig. 7 PXR represses UGT1A1 gene expression in neonatal hUGT1*1 mice. (A) Comparison of the hepatic expression levels of the UGT1A1 gene in hUGT1 and hUGT1/Pxr−/− mice 14 days after birth. (B) CHIP analysis of PXR associated with the human UGT1A1 gene in neonatal and adult hUGT1*1 mice.

ion in hUGT1*1/Pxr−/− mice when compared to hUGT1*1 mice (Fig. 7A). This finding indicates that nonliganded PXR during development in hUGT1*1 mice is serving to repress liver UGT1A1 gene expression, because in PXR-deficient mice we observe induction of liver UGT1A1, which correlates with a reduction in serum bilirubin. Fig. 7 PXR represses UGT1A1 gene expression in neonatal hUGT1*1 mice. (A) Comparison of the hepatic expression levels of the UGT1A1 gene in hUGT1 and hUGT1/Pxr−/− mice 14 days after birth. (B) CHIP analysis of PXR associated with the human UGT1A1 gene in neonatal and adult hUGT1*1 mice. To further examine the developmental properties of PXR on UGT1A1 gene repression, we performed PXR CHIP analysis by using liver samples from both neonatal and adult hUGT1*1 mice. As shown in Fig. 7B, intensified PXR signals were observed in neonatal livers in comparison to adult livers. Abundant PXR binding to the UGT1A1 gene is concordant with reduced UGT1A1 gene expression, indicating that PXR is repressing gene expression. To examine this possibility, we isolated primary hepatocytes from 14-day-old hUGT1*1 mice and transfected them with PXR-specific siRNAs from two sources (Fig. 8). Forty-eight hours later, UGT1A1 gene expression was quantitated by using RT and Q-PCR. The fold induction was tied to the extent of the PXR mRNA knockdown. When the PXR knockdown was 50% (siPXR_b), ≍2-fold induction of human UGT1A1 expression was observed. When the PXR knockdown was 70%, UGT1A1 gene induction was 4-fold. In contrast, Cyp3a11, another PXR target gene, showed no change. These findings confirm that transcriptional silencing of the UGT1A1 gene by PXR occurs during neonatal development in liver tissue.

2-fold induction of human UGT1A1 expression was observed. When the PXR knockdown was 70%, UGT1A1 gene induction was 4-fold. In contrast, Cyp3a11, another PXR target gene, showed no change. These findings confirm that transcriptional silencing of the UGT1A1 gene by PXR occurs during neonatal development in liver tissue. Fig. 8 Knockdown of PXR and derepression of UGT1A1 gene expression in primary hepatocytes. Hepatocytes were isolated from 14-day-old neonatal hUGT1*1 mice. Cells were transfected with mouse siRNA oligonucleotide manufactured by Bioneer (siPXR_b) and Santa Cruz Biotechnology (siPXR_sc). Forty-eight hours later, cells were harvested for RNA preparation. The relative gene expression levels of mouse Pxr, human UGT1A1, and mouse Cyp3a11 were determined by RT and Q-PCR (**P < 0.01, t test).

ansfected with mouse siRNA oligonucleotide manufactured by Bioneer (siPXR_b) and Santa Cruz Biotechnology (siPXR_sc). Forty-eight hours later, cells were harvested for RNA preparation. The relative gene expression levels of mouse Pxr, human UGT1A1, and mouse Cyp3a11 were determined by RT and Q-PCR (**P < 0.01, t test). Discussion Employing recently developed TgUGT1 mice with the UGT1A genes expressed in a Ugt1-null background, the function of PXR and CAR during pregnancy was investigated using reverse genetics to examine induction of the UGT1A genes in xenobiotic receptor-defective mice. Previous experiments with TgUGT1 and hUGT1 mice have shown that chemical treatment with activators of either PXR or CAR leads to induction of UGT1A1, -1A3, -1A4, -1A6, and -1A9.14 The mechanisms by which PXR and CAR control the induction of all these genes have not been determined, although PXR and CAR-responsive elements have been shown to play an important role in PXR/CAR binding and induction of the UGT1A1 gene. In primary hepatocytes from hUGT1 mice, selective treatment with 17β-estradiol, progesterone, and DEX each led to induction of the UGT1A1 gene, with glucocorticoid treatment maximizing the induction response. In timed pregnancy experiments, each of the maternal liver-specific UGT1A genes was induced in hUGT1 mice, with UGT1A1, UGT1A6, and UGT1A9 gene expression being the most prominent. During pregnancy, gestational induction of the UGT1A genes was mostly conserved in hUGT1/Car−/− mice but greatly diminished in hUGT1/Pxr−/− mice, suggesting that PXR participates in a global fashion to regulate the UGT1 locus during fetal development. The sole exception to this appears to be with UGT1A9 gene expression, which displayed reduced expression during pregnancy in hUGT1/Car−/− mice. Because both PXR/CAR appear to be necessary for induction of UGT1A9 during pregnancy, there may be crosstalk occurring between the two xenobiotic receptors to facilitate regulation of the UGT1A9 gene.

is appears to be with UGT1A9 gene expression, which displayed reduced expression during pregnancy in hUGT1/Car−/− mice. Because both PXR/CAR appear to be necessary for induction of UGT1A9 during pregnancy, there may be crosstalk occurring between the two xenobiotic receptors to facilitate regulation of the UGT1A9 gene. The contribution of PXR toward induction of the human UGT1 locus is not conserved with the murine Ugt1 locus, because pregnancy has no effect on regulation of the murine Ugt1a genes as determined by gene expression profiling and protein accumulation. This was surprising because it has been demonstrated previously that overexpression of the human PXR in humanized PXR mice or treatment of mice with PXR ligands, such as PCN, leads to induction of the murine Ugt1a1 gene.3, 27, 28 Our results indicate that PXR is the central modulator of the human UGT1A genes during pregnancy, but additional regulatory events specifically toward control of the human UGT1 locus, and not the murine Ugt1 locus, are in place during pregnancy. Because an increase in UGT1A-dependent glucuronidation occurs in humans during pregnancy, the ability to reproduce this event in transgenic mice indicates that the genetic sequence specific to the UGT1 locus is largely responsible for directing the transcriptional program of the human UGT1A genes during pregnancy. Thus, the differences observed in the induction patterns between the human UGT1A genes and the murine Ugt1a genes are not the result of interspecies differences in epigenetic machinery or the cellular environment. From this result we can infer that transcriptional factors play a secondary role in dictating the differences observed in the human UGT1 and murine Ugt1 locus during pregnancy. Such a pattern expressing conserved genes between humans and mice has been demonstrated,27-29 reinforcing the hypothesis that differences in gene expression between species are controlled by changes in cis-acting transcriptional binding sequences.21, 22, 30

rved in the human UGT1 and murine Ugt1 locus during pregnancy. Such a pattern expressing conserved genes between humans and mice has been demonstrated,27-29 reinforcing the hypothesis that differences in gene expression between species are controlled by changes in cis-acting transcriptional binding sequences.21, 22, 30 Among the UGT1A isoforms, UGT1A1 is of special physiological importance because it is the only enzyme that catalyzes the glucuronidation of bilirubin.6 Accumulation of bilirubin leads to benign levels of hyperbilirubinemia shortly after birth, but if bilirubin levels continue to rise, the more serious symptoms associated with bilirubin-induced neurological dysfunction (BIND) can develop.31, 32 Phenobarbital, a CAR agonist, has been used clinically for the treatment of neonatal hyperbilirubinemia in infants at risk for severe jaundice (TSB levels more than 16 mg/dL), therefore reducing the need for exchange transfusion.33-35 However, phenobarbital treatment is not effective immediately, and it diminishes the oxidative metabolism of bilirubin, increasing the risk of neurotoxic effects.36 Glucocorticoids have also been used to treat hyperbilirubinemia.37 The initial intent of glucocorticoid therapy is to help fetal lung maturation and reduce neonatal mortality in women at high risk for preterm labor before 35 gestational weeks. During these treatments, it has been observed that hyperbilirubinemia is significantly lower in the DEX-treated groups compared to untreated control groups. Our studies indicate that PXR serves as a major regulator following glucocorticoid treatment by inducing liver UGT1A1 expression, leading to reduction of hyperbilirubinemia. Identification of PXR as a key regulator of the UGT1A1 gene during neonatal development can be exploited as a potential therapeutic target in the treatment of hyperbilirubinemia.

rves as a major regulator following glucocorticoid treatment by inducing liver UGT1A1 expression, leading to reduction of hyperbilirubinemia. Identification of PXR as a key regulator of the UGT1A1 gene during neonatal development can be exploited as a potential therapeutic target in the treatment of hyperbilirubinemia. Analysis of TSB levels in neonatal hUGT1 and hUGT1/Pxr−/− mice demonstrate that PXR plays a key role in controlling serum levels during development. Neonatal hUGT1 mice develop severe hyperbilirubinemia due to a reduction in liver UGT1A1 gene expression.15 In hUGT1/Pxr−/− mice, liver UGT1A1 is induced when compared to expression in hUGT1 mice. The increased levels of liver UGT1A1 in hUGT1/Pxr−/− mice lead to reduced levels of TSB. This finding indicates that in the absence of endogenous/exogenous ligands, the physiological role of PXR leads to repression of the UGT1A1 gene during early development. It may also be an underlying regulator of the UGT1A1 gene in newborns and responsible in part for neonatal hyperbilirubinemia. Hence, PXR acts as a repressor of UGT1A1 expression in the absence of ligand. It is known that the repressive function of PXR works in part through the recruitment of the corepressor Silencing Mediator of Retinoid and Thyroid Hormone Receptors (SMRT).38, 39 SMRT binds to nuclear receptors in the absence of ligand and alters the chromatin structure through histone modification.40, 41 Clearly, deletion of PXR releases the repression (de-repression) allowing for spontaneous induction of UGT1A1 gene expression. This finding may be useful in future studies to identify PXR modulators that might directly influence bilirubin homeostasis and accelerate bilirubin metabolism and clearance in children with abnormally high levels of TSB.

eases the repression (de-repression) allowing for spontaneous induction of UGT1A1 gene expression. This finding may be useful in future studies to identify PXR modulators that might directly influence bilirubin homeostasis and accelerate bilirubin metabolism and clearance in children with abnormally high levels of TSB. Abbreviations CARconstitutive androstane receptor CHIPchromatin immunoprecipitation ERαestrogen receptor alpha GDgestational day hUGT1humanized UGT1 mouse PXRpregnane X receptor TgUGT1transgenic UGT1 mouse TSBtotal serum bilirubin UGTUDP glucuronosyltransferase UGT1human UGT1 locus Ugt1murine Ugt1 locus Supplementary material Additional Supporting Information may be found in the online version of this article.

Chronic liver disease and cirrhosis result in an estimated 800,000 deaths each year worldwide.1 In the United States alone, it is the ninth leading cause of death, with about 30,000 deaths each year.2 An additional 30 million Americans have chronic liver impairment.3 Hospitalizations of these patients are frequent and substantial proportions of these admissions include stays in the intensive care unit (ICU).4–6 The estimated number of ICU admissions related to cirrhosis in the United States alone is in excess of 26,000 per year with an estimated cost of $3 billion.7 A major cause of ICU admission among patients with cirrhosis is sepsis.5, 8–10 The incidence of sepsis is estimated to be at least 30%-50% of hospital admissions in this group.11, 12 Cirrhosis-associated septic shock stands out in terms of presentation, outcome,13 and therapeutic options.14 One of the key questions is whether modifiable practice-related factors contribute to the poor outcome in this group of patients. Limited data are available about the appropriate application of the newer options that have emerged in the management of sepsis over the last decade15–17 in this high-risk group,18 as patients with cirrhosis-related septic shock are often excluded from clinical trials. In addition to this paucity of evidence-based information,19 the Surviving Sepsis Campaign guidelines do not provide a clear direction for this group of patients.20

ent of sepsis over the last decade15–17 in this high-risk group,18 as patients with cirrhosis-related septic shock are often excluded from clinical trials. In addition to this paucity of evidence-based information,19 the Surviving Sepsis Campaign guidelines do not provide a clear direction for this group of patients.20 In a heterogeneous patient population with septic shock, the early initiation of appropriate antimicrobials and combination antibiotics (for bacterial septic shock) is associated with higher survival rate.21–23 However, few data exist on the use of antibiotics and outcome from septic shock among patients with cirrhosis.11 Such lack of information may adversely affect decision-making about patient management and prognostication. We conducted this study to examine the relationship between the aspects of early, initial empiric antimicrobial therapy and hospital mortality in patients with cirrhosis and septic shock. Patients and Methods Patients and Setting This was a nested cohort study within a retrospective database on septic shock conducted in 28 medical centers in Canada, the United States, and Saudi Arabia by the Cooperative Antimicrobial Therapy of Septic Shock (CATSS) Database Research Group between 1996 and 2008. The details of the setting have been described elsewhere.22 Data were extracted for all adult patients with cirrhosis (biopsy-proven cirrhosis, documented variceal hemorrhage or portal hypertension, hepatic ascites, or encephalopathy). Approval was obtained from the Institutional Review Boards of all participating institutions.

ls of the setting have been described elsewhere.22 Data were extracted for all adult patients with cirrhosis (biopsy-proven cirrhosis, documented variceal hemorrhage or portal hypertension, hepatic ascites, or encephalopathy). Approval was obtained from the Institutional Review Boards of all participating institutions. Measurements We collected baseline patient characteristics including demographics and comorbid conditions. The following data were obtained on day 1 of septic shock: serum lactate, bilirubin, creatinine, and bicarbonate levels, platelet count, international normalized ratio and white blood cell (WBC) count, and acute physiology and chronic health evaluation (APACHE) II24 score. We calculated the model for end-stage liver disease (MELD) score on day 1 as described previously.25 Outcomes Hospital mortality was considered the primary outcome variable. Secondary outcomes were ICU mortality and hospital and ICU length of stay.

Measurements We collected baseline patient characteristics including demographics and comorbid conditions. The following data were obtained on day 1 of septic shock: serum lactate, bilirubin, creatinine, and bicarbonate levels, platelet count, international normalized ratio and white blood cell (WBC) count, and acute physiology and chronic health evaluation (APACHE) II24 score. We calculated the model for end-stage liver disease (MELD) score on day 1 as described previously.25 Outcomes Hospital mortality was considered the primary outcome variable. Secondary outcomes were ICU mortality and hospital and ICU length of stay. Definitions Septic shock was defined using the 1992 American College of Chest Physicians/Society of Critical Care Medicine guidelines.26 As per that definition, case patients were required to have documented or suspected infection, persistent hypotension requiring therapy with vasopressors, and two of the following four elements: (1) a heart rate of >90 beats/minute; (2) a respiratory rate of >20 breaths/minute or arterial partial pressure of carbon dioxide (PaCO2) of <32 mm Hg; (3) a core temperature of <36°C or >38°C; and (4) a WBC count of <4,000/μL or >12,000/μL or bands >10%. An episode of hypotension was considered to represent the initial onset of septic shock when hypotension persisted from onset despite fluid (2 L of saline or equivalent) administration (persistent hypotension) or hypotension was only transiently improved (hypotension resolution for <1 hour) with fluid resuscitation (recurrent hypotension).21 Predetermined rules were used to define documented and suspected infections and to assign significance to clinical isolates (see Supporting Information). For patients with multiple isolated organisms, we identified the primary organism that was likely responsible for the infection. We documented the following multidrug-resistant organisms: Methicillin-resistant Staphylococcus aureus, carbapenem-resistant gram-negative organisms, vancomycin-resistant enterococci and extended-spectrum beta-lactamase–producing Enterobacteriaceae. Nosocomial infection-related septic shock was defined as septic shock caused by any infection developing >48 hours after hospital admission. Cases not meeting this definition were considered to be septic shock associated with community-acquired infections. We used the term “immunocompromised patients” for a subgroup of patients who had one of the following comorbidities: acquired immunodeficiency syndrome, acute or chronic lymphoma, acute or chronic leukemia/multiple myeloma, metastatic solid cancer, immunosuppressive chemotherapy, or long-term steroid therapy (>10 mg prednisone equivalent daily). Other patients were labeled as “non-immunocompromised.”

e of the following comorbidities: acquired immunodeficiency syndrome, acute or chronic lymphoma, acute or chronic leukemia/multiple myeloma, metastatic solid cancer, immunosuppressive chemotherapy, or long-term steroid therapy (>10 mg prednisone equivalent daily). Other patients were labeled as “non-immunocompromised.” Predetermined rules used to assess the appropriateness and delays of initial antimicrobial therapy21–23 are summarized below and detailed in the Supporting Information. For culture-positive septic shock, initial antimicrobial therapy was considered appropriate if an antimicrobial with in vitro activity appropriate for the isolated pathogen or pathogens was the first new antimicrobial agent given after the onset of recurrent or persistent hypotension or was initiated within 6 hours of the administration of the first new antimicrobial agent. Otherwise, the initial therapy was considered inappropriate.22 For culture-negative septic shock, initial therapy was considered appropriate when an antimicrobial agent consistent with broadly accepted norms for empiric management of the typical pathogens for the clinical syndrome was the new antimicrobial agent given after the onset of recurrent or persistent hypotension or was initiated within 6 hours of administration of the first new antimicrobial agent.22 At each participating institution, infectious disease physicians/microbiologists were consulted to account for the local community and nosocomial flora when considering appropriateness of empiric therapy during the period covered by data collection. Otherwise, appropriate empiric therapy of culture-negative infections leading to septic shock was based on the recommendations listed in the “Clinical Approach to Initial Choice of Antimicrobial Therapy” in the Sanford Guide to Antimicrobial Therapy 2004 (34th edition).27 For the purposes of this study, antibiotic monotherapy was defined as the administration of any single, appropriate, intravenous, preferably bactericidal antibiotic at any point after the onset of recurrent or persistent hypotension. Combination therapy was defined as the concomitant use of two or more antibiotics of different mechanistic classes with activity for the isolated or suspected pathogens.

ation of any single, appropriate, intravenous, preferably bactericidal antibiotic at any point after the onset of recurrent or persistent hypotension. Combination therapy was defined as the concomitant use of two or more antibiotics of different mechanistic classes with activity for the isolated or suspected pathogens. The second agent had to be started within 24 hours of the first antibiotic or within 24 hours of the onset of hypotension (if the first agent was initiated before hypotension was documented).23 Patients with septic shock due to yeast, anaerobic, or mycobacterial infection were excluded from this analysis of single versus combined antibiotics. Statistical Analysis Continuous variables are reported as the mean ± SD and median (interquartile range) as appropriate. Categorical variables are reported as numbers and percentages. The Student t test, Mann-Whitney U test, chi-square test, and Fisher's exact test for comparison between groups were used, as appropriate. To study the association between appropriateness, timing, and combination of antimicrobial/antibiotic therapy and hospital mortality (dependent variable), forward step-wise logistic regression analyses were performed. The following independent variables were included based on their significance in the univariate analysis: APACHE II score, MELD score, immunocompromised (versus non-immunocompromised), bacteremia (versus no bacteremia), community-acquired (versus nosocomial), and culture-positive (versus culture-negative).

e performed. The following independent variables were included based on their significance in the univariate analysis: APACHE II score, MELD score, immunocompromised (versus non-immunocompromised), bacteremia (versus no bacteremia), community-acquired (versus nosocomial), and culture-positive (versus culture-negative). To determine the predictors of inappropriate antimicrobial and single antibiotic therapy, we performed forward stepwise logistic regression analyses. In the first analysis, inappropriate antimicrobial therapy was the dependent variable, and the independent variables were age, sex, BMI, APACHE II score, MELD score, serum lactate, bilirubin, creatinine and bicarbonate levels, platelet count, international normalized ratio, WBC count, heart rate, temperature, respiratory rate, blood pressure, community-acquired (versus nosocomial), organ failures, activated protein C, steroids, multidrug-resistant organisms, and comorbidities. A similar analysis was performed for single antibiotic therapy as the dependent variable. A third analysis was performed to assess the predictors of delayed antimicrobial therapy using forward stepwise linear regression analysis with the same independent variables listed above.

To determine the predictors of inappropriate antimicrobial and single antibiotic therapy, we performed forward stepwise logistic regression analyses. In the first analysis, inappropriate antimicrobial therapy was the dependent variable, and the independent variables were age, sex, BMI, APACHE II score, MELD score, serum lactate, bilirubin, creatinine and bicarbonate levels, platelet count, international normalized ratio, WBC count, heart rate, temperature, respiratory rate, blood pressure, community-acquired (versus nosocomial), organ failures, activated protein C, steroids, multidrug-resistant organisms, and comorbidities. A similar analysis was performed for single antibiotic therapy as the dependent variable. A third analysis was performed to assess the predictors of delayed antimicrobial therapy using forward stepwise linear regression analysis with the same independent variables listed above. We performed subgroup analyses for the following categories: septic shock with documented and suspected infection, culture-positive and culture-negative, bacteremia and no bacteremia, community-acquired and nosocomial infections, gram-positive and gram-negative infections, pneumonia, intra-abdominal infection, immunocompromised and non-immunocompromised, and country of origin (Canada, United States, and Saudi Arabia). To adjust for the impact of potential changes in practice over time, we divided the study time period into four quartiles and compared outcomes across the four periods. To examine the possibility of effect modification, we tested for interaction of the above-mentioned subgroups with the appropriateness, timing and combination of antimicrobial in the related multivariate models.

er time, we divided the study time period into four quartiles and compared outcomes across the four periods. To examine the possibility of effect modification, we tested for interaction of the above-mentioned subgroups with the appropriateness, timing and combination of antimicrobial in the related multivariate models. For all multivariate analyses, we checked for multicollinearity among covariates by evaluating the variation inflation factors. Missing data were handled using the mean and median imputation method.28 Results were reported as adjusted odds ratios (aOR) and 95% confidence intervals (CI). P < 0.05 was considered significant. SAS software (SAS Institute, Cary, NC) was used for statistical analyses. Results Among the 8,670 patients with septic shock within the CATSS database, we identified 635 (7.3%) patients with cirrhosis (385 men [60.6%], 250 women [39.4%]). The mean age (± SD) was 55.5 ± 12.7 (Table 1). The first day mean APACHE II and MELD scores were 28.2 ± 8.2 and 26.7 ± 11.1, respectively. The frequencies of chronic comorbidities among the patient cohort are presented in Table 1. Nearly half of the patients suffered from chronic alcohol abuse. The most common sites of infection were lung (37%), intra-abdominal (35%), and primary bloodstream (7.9%) (Table 2). Positive cultures were obtained in 473 (74.5%) patients. The most common isolated pathogens were gram-negative (35.1%) followed by gram-positive (26.5%) and fungi (9.3%). Thirty-one (4.9%) patients had multidrug-resistant organisms (Table 3).

lung (37%), intra-abdominal (35%), and primary bloodstream (7.9%) (Table 2). Positive cultures were obtained in 473 (74.5%) patients. The most common isolated pathogens were gram-negative (35.1%) followed by gram-positive (26.5%) and fungi (9.3%). Thirty-one (4.9%) patients had multidrug-resistant organisms (Table 3). Table 1 Baseline Characteristics of Patients With Cirrhosis and Septic Shock

lung (37%), intra-abdominal (35%), and primary bloodstream (7.9%) (Table 2). Positive cultures were obtained in 473 (74.5%) patients. The most common isolated pathogens were gram-negative (35.1%) followed by gram-positive (26.5%) and fungi (9.3%). Thirty-one (4.9%) patients had multidrug-resistant organisms (Table 3). Table 1 Baseline Characteristics of Patients With Cirrhosis and Septic Shock Characteristic Value Age, years 55.5 ± 12.7 Sex, men/women 385 (60.6)/250 (39.4) BMI, kg/m2 27.8 ± 7.8 APACHE II score 28.2 ± 8.2 MELD score 26.7 ± 11.1 No. of organ failures on day 1 4.7 ± 1.7 Mechanical ventilation 471 (74.2) Laboratory findings on day 1 Lactate, mmol/L 6.4 ± 4.8 Bilirubin, μmol/L 142 ± 171 Creatinine, μmol/L 215 ± 167 Bicarbonate, mmol/L 17.3 ± 6.2 Platelet count, ×109/L 128 ± 119 INR 2.4 ± 1.5 WBC count, ×109/L 15.7 ± 12.4 Vital signs Heart rate, beats/minute 114 ± 28 Respiratory rate, beats/minute 27 ± 9 Temperature, °C 36.9 ± 1.8 Mean arterial pressure, mm Hg 56 ± 14 Vasopressor use 635 (100) Renal replacement therapy 56 (8.8) Activated protein C 9 (1.4) Steroids 192 (30.2) Comorbidities AIDS 19 (3.0) Acute or chronic lymphoma 10 (1.6) Acute or chronic leukemia/multiple myeloma 9 (1.4) Metastatic solid cancer 26 (4.1) Immunosuppressive chemotherapy or long-term steroid therapy (>10 mg prednisone equivalent daily) 54 (8.5) Neutropenia (>500 cells/L) 12 (1.9) New York Heart Association class IV heart failure 34 (5.4) COPD (requiring medication or oxygen) 27 (4.3) Chronic renal failure* 74 (11.7) Chronic dialysis dependence 31 (4.9) Diabetes mellitus (medication-dependent) 100 (15.8) Diabetes mellitus (insulin-dependent) 54 (8.5) Alcohol abuse 278 (43.8) Elective surgery 64 (10.1) Emergency surgery/trauma 33 (5.2) Bacteremia/fungemia 245 (38.6) Community-acquired infection 357 (56.2) Nosocomial infection 278 (43.8) Continuous variables are expressed as the mean ± SD. Categorical variables are expressed as no. (%).

s (insulin-dependent) 54 (8.5) Alcohol abuse 278 (43.8) Elective surgery 64 (10.1) Emergency surgery/trauma 33 (5.2) Bacteremia/fungemia 245 (38.6) Community-acquired infection 357 (56.2) Nosocomial infection 278 (43.8) Continuous variables are expressed as the mean ± SD. Categorical variables are expressed as no. (%). Abbreviations: AIDS, acquired immune deficiency syndrome; BMI, body mass index; COPD, chronic obstructive pulmonary disease; INR, international normalized ratio. All percentages are out of the total number of patients in the cohort (n = 635). * Serum creatinine >1.5 the upper limit of normal. Table 2 Clinical Sites of Infection Among the Patient Cohort

Abbreviations: AIDS, acquired immune deficiency syndrome; BMI, body mass index; COPD, chronic obstructive pulmonary disease; INR, international normalized ratio. All percentages are out of the total number of patients in the cohort (n = 635). * Serum creatinine >1.5 the upper limit of normal. Table 2 Clinical Sites of Infection Among the Patient Cohort Site of infection Value Lung 235 (37.0) Pneumonia 228 (35.9) Empyema 7 (1.1) Intra-abdominal 222 (35.0) Intra-abdominal abscess 14 (2.2) Ascending cholangitis 17 (2.7) Cholecystitis 8 (1.3) Ischemic bowel/bowel infarction 25 (3.9) Bowel perforation/peritonitis 23 (3.6) Spontaneous bacterial peritonitis 112 (17.6) Clostridium difficile enterocolitis/toxic megacolon 7 (1.1) Others 16 (2.5) Skin and soft tissue 29 (4.6) Cellulitis 6 (0.9) Necrotizing soft tissue infections 19 (3.0) Others 4 (0.6) Genitourinary 41 (6.5) Intravascular catheter infection 18 (2.8) Primary bloodstream (bacteremia/fungemia without identifiable source) 50 (7.9) Systemically disseminated (including yeast and tuberculosis) 22 (3.5) Septic arthritis 7 (1.1) Data are expressed as no. (%). All percentages are out of the total number of patients in the cohort (n = 635). Clinical sites of infection were documented in 548 (86.3%) patients and suspected in 87 (13.7%) patients. Table 3 Primary Organisms Among the Patient Cohort

Site of infection Value Lung 235 (37.0) Pneumonia 228 (35.9) Empyema 7 (1.1) Intra-abdominal 222 (35.0) Intra-abdominal abscess 14 (2.2) Ascending cholangitis 17 (2.7) Cholecystitis 8 (1.3) Ischemic bowel/bowel infarction 25 (3.9) Bowel perforation/peritonitis 23 (3.6) Spontaneous bacterial peritonitis 112 (17.6) Clostridium difficile enterocolitis/toxic megacolon 7 (1.1) Others 16 (2.5) Skin and soft tissue 29 (4.6) Cellulitis 6 (0.9) Necrotizing soft tissue infections 19 (3.0) Others 4 (0.6) Genitourinary 41 (6.5) Intravascular catheter infection 18 (2.8) Primary bloodstream (bacteremia/fungemia without identifiable source) 50 (7.9) Systemically disseminated (including yeast and tuberculosis) 22 (3.5) Septic arthritis 7 (1.1) Data are expressed as no. (%). All percentages are out of the total number of patients in the cohort (n = 635). Clinical sites of infection were documented in 548 (86.3%) patients and suspected in 87 (13.7%) patients. Table 3 Primary Organisms Among the Patient Cohort Organism Value Gram-negative* 223 (35.1) Escherichia coli 95 (15.0) Klebsiella species 46 (7.2) Pseudomonas aeruginosa 26 (4.1) Enterobacter species 13 (2.0) Haemophilusinfluenzae 12 (1.9) Acinetobacter species 7 (1.1) Serratia species 5 (0.8) Stenotrophomonas maltophilia 6 (0.9) Other gram-negative organisms 13 (2.0) Gram-positive* 168 (26.5) Staphylococcus aureus 74 (11.7) Streptococcus pneumoniae 37 (5.8) Streptococcusfaecalis 12 (1.9) Group A Streptococcus species 8 (1.3) Other β-hemolytic Streptococcus species 12 (1.9) Viridans Streptococcus species 9 (1.4) Streptococcus faecium 12 (1.9) Other gram-positive organisms 4 (0.6) Yeast/fungus 59 (9.3) Candida albicans 40 (6.3) Candida glabrata 8 (1.3) Candida tropicalis 5 (0.8) Other Candida species/yeast 6 (0.9) Anaerobes 12 (1.9) Clostridium difficile 7 (1.1) Bacteroides fragilis 2 (0.3) Other Clostridium species 1 (0.2) Other anaerobes 2 (0.3) Other organisms 11 (1.7) Total culture-positive 473 (74.5) Total culture-negative 162 (25.5) Multidrug-resistant 31 (4.9) Methicillin-resistant Staphylococcus aureus 17 (2.7) Carbapenem-resistant gram-negative bacteria 8 (1.3) Vancomycin-resistant enterococci 3 (0.5) ESBL-producing Enterobacteriaceae 3 (0.5) Data are expressed as no. (%). All percentages are out of the total number of patients in the cohort (n = 635).

g-resistant 31 (4.9) Methicillin-resistant Staphylococcus aureus 17 (2.7) Carbapenem-resistant gram-negative bacteria 8 (1.3) Vancomycin-resistant enterococci 3 (0.5) ESBL-producing Enterobacteriaceae 3 (0.5) Data are expressed as no. (%). All percentages are out of the total number of patients in the cohort (n = 635). Abbreviation: ESBL, extended-spectrum beta-lactamase. * Includes multidrug-resistant organisms. The ICU and hospital mortality rates were 61.6% and 75.6%, respectively. The mean ICU length of stay was 9.9 ± 11.5 days, and the mean hospital length of stay was 17.8 ± 25.2 days, respectively. Hospital mortality was similar over the four study period quartiles. Hospital nonsurvivors had higher APACHE II and MELD scores (Table 4) and were more likely to receive inappropriate initial empiric antimicrobials (30.6% versus 5.2%) and delayed appropriate empiric antimicrobial therapy (median (interquartile range)) (10.0 (4.9–23.8) versus 3.2 (1.3–6.8) hours) than survivors. Nonsurvivors with bacterial septic shock were also more likely to have been treated with empiric mono-antimicrobial therapy (77.0% versus 63.4%). Table 4 Comparison of Patient Characteristics Between Hospital Survivors and Nonsurvivors Among the Patient Cohort Variable Hospital Survivors Hospital Nonsurvivors P No.

The ICU and hospital mortality rates were 61.6% and 75.6%, respectively. The mean ICU length of stay was 9.9 ± 11.5 days, and the mean hospital length of stay was 17.8 ± 25.2 days, respectively. Hospital mortality was similar over the four study period quartiles. Hospital nonsurvivors had higher APACHE II and MELD scores (Table 4) and were more likely to receive inappropriate initial empiric antimicrobials (30.6% versus 5.2%) and delayed appropriate empiric antimicrobial therapy (median (interquartile range)) (10.0 (4.9–23.8) versus 3.2 (1.3–6.8) hours) than survivors. Nonsurvivors with bacterial septic shock were also more likely to have been treated with empiric mono-antimicrobial therapy (77.0% versus 63.4%). Table 4 Comparison of Patient Characteristics Between Hospital Survivors and Nonsurvivors Among the Patient Cohort Variable Hospital Survivors Hospital Nonsurvivors P No. of patients 155 480 — Age, years 54.5 ± 12.8 55.9 ± 12.7 0.22 Sex, men/women 103 (66.5) 282 (58.8) 0.09 BMI, kg/m2 27.8 ± 7.9 27.8 ± 7.8 0.96 APACHE II score 22.8 ± 6.5 29.9 ± 8.0 <0.0001 MELD score 22.2 ± 10.1 28.1 ± 11.0 <0.0001 Laboratory findings on day 1 Lactate, mmol/L 5.5 ± 4.2 6.7 ± 5.0 0.47 Bilirubin, μmol/L 85 ± 117 160 ± 182 <0.0001 Creatinine, μmol/L 201 ± 177 220 ± 163 0.22 Bicarbonate, mmol/L 19.0 ± 6.0 16.6 ± 6.2 0.0006 Platelet count, ×109/L 156 ± 143 119 ± 108 0.004 INR 2.0 ± 1.5 2.5 ± 1.5 0.0006 WBC count, ×109/L 17.3 ± 12.8 15.2 ± 12.2 0.09 Vital signs Heart rate, beats/minute 114 ± 27 115 ± 29 0.73 Respiratory rate, beats/minute 25 ± 10 28 ± 9 0.009 Temperature, °C 37.4 ± 1.6 36.7 ± 1.9 0.0002 Mean arterial pressure, mm Hg 60 ± 16 55 ± 14 0.05 Activated protein C 2 (1.3) 7 (1.5) 1.00 Steroids 48 (31.0) 144 (30.0) 0.82 Inappropriate antimicrobials 8 (5.2) 147 (30.6) <0.0001 Single appropriate antibiotic 59 (63.4) 167 (77.0) 0.01 Delay in effective antimicrobials, hours 3.2 (1.3–6.8) 10.0 (4(.9–23.8) <0.0001 Comorbidities AIDS 1 (0.7) 18 (3.8) 0.06 Acute or chronic lymphoma 2 (1.3) 8 (1.7) 1.00 Acute or chronic leukemia/multiple  myeloma 1 (0.7) 8 (1.7) 0.70 Metastatic solid cancer 5 (3.2) 21 (4.4) 0.53 Immunosuppressive chemotherapy or long-term steroid therapy (>10 mg prednisone equivalent daily) 9 (5.8) 45 (9.4) 0.17 Neutropenia (>500 cells/L) 1 (0.7) 11 (2.3) 0.31 New York Heart Association class IV heart failure 9 (5.8) 25 (5.2) 0.77 COPD (requiring medication or oxygen) 7 (4.5) 20 (4.2) 0.85 Chronic renal failure* 14 (9.0) 60 (12.5) 0.24 Chronic dialysis dependence 6 (3.9) 25 (5.2) 0.50 Diabetes mellitus (medication-dependent) 26 (16.8) 74 (15.4) 0.69 Diabetes mellitus (insulin-dependent) 13 (8.4) 41 (8.5) 0.95 Alcohol abuse 74 (47.7) 204 (42.5) 0.25 Elective surgery 12 (7.7) 52 (10.8) 0.27 Emergency surgery/trauma 8 (5.2) 25 (5.2) 0.98 Culture-positive 106 (68.4) 367 (76.5) 0.05 Bacteremia 57 (36.8) 188 (39.2) 0.59 Community-acquired infection 104 (67.1) 253 (52.7) 0.002 Nosocomial infection 51 (32.9) 227 (47.3) 0.002 Continuous variables are expressed as the mean ± SD or median and interquartile range.

Emergency surgery/trauma 8 (5.2) 25 (5.2) 0.98 Culture-positive 106 (68.4) 367 (76.5) 0.05 Bacteremia 57 (36.8) 188 (39.2) 0.59 Community-acquired infection 104 (67.1) 253 (52.7) 0.002 Nosocomial infection 51 (32.9) 227 (47.3) 0.002 Continuous variables are expressed as the mean ± SD or median and interquartile range. Categorical variables are expressed as no. (%). Abbreviations: AIDS, acquired immune deficiency syndrome; BMI, body mass index; COPD, chronic obstructive pulmonary disease; INR, international normalized ratio. * Serum creatinine >1.5 the upper limit of normal. Of the 635 patients with cirrhosis and septic shock, inappropriate initial antimicrobial therapy was administered in 155 (24.4%) (Table 5). Forty-six (7.2%) patients never received appropriate antimicrobials before death. The median (interquartile range) time to administration of antimicrobials was 7.3 (3.2–18.3) hours after documentation of hypotension associated with septic shock. Two hundred twenty-six (72.9%) of the 310 patients with eligible bacterial septic shock who could potentially receive combination antibiotic therapy received a single antibiotic. There were no significant differences in these antibiotic-related variables across the three countries (see Supporting Information), except for a higher proportion of patients receiving combination therapy during the course of shock in United States hospitals (25.3% of eligible patients in Canada, 50% in the United States, 24.6% in Saudi Arabia; P = 0.03). Table 5 Descriptive Analysis of Antimicrobial Determinants and Patient Outcomes

Of the 635 patients with cirrhosis and septic shock, inappropriate initial antimicrobial therapy was administered in 155 (24.4%) (Table 5). Forty-six (7.2%) patients never received appropriate antimicrobials before death. The median (interquartile range) time to administration of antimicrobials was 7.3 (3.2–18.3) hours after documentation of hypotension associated with septic shock. Two hundred twenty-six (72.9%) of the 310 patients with eligible bacterial septic shock who could potentially receive combination antibiotic therapy received a single antibiotic. There were no significant differences in these antibiotic-related variables across the three countries (see Supporting Information), except for a higher proportion of patients receiving combination therapy during the course of shock in United States hospitals (25.3% of eligible patients in Canada, 50% in the United States, 24.6% in Saudi Arabia; P = 0.03). Table 5 Descriptive Analysis of Antimicrobial Determinants and Patient Outcomes Total Mortality P* Appropriateness of initial antimicrobial therapy, no. (%) Inappropriate 155 (24.4) 147 (94.8) <0.0001** Culture-positive 128 (20.2) 120 (93.8) Culture-negative 27 (4.3) 27 (100.0) Appropriate 480 (75.6) 333 (69.4) Culture-positive 345 (54.3) 247 (71.6) Culture-negative 135 (21.3) 86 (63.7) Timing of first appropriate antibiotic, no. (%) Prior to hypotension onset 113 (17.8) 84 (74.3) 0.68 After hypotension onset 476 (75.0) 349 (73.3) Hours after hypotension, median (IQR) 7.3 (3.2–18.3) Appropriate antimicrobial therapy during the course of shock, no. (%) Never received appropriate antimicrobials 46 (7.2) 46 (100.0) <0.0001 Received appropriate definitive therapy 589 (92.8) 434 (73.7) Potential candidates for combined antibiotic therapy 310 (48.8) Received single therapy 226 (72.9) 167 (73.9) 0.01 Received combination therapy 84 (27.1) 50 (59.5) Abbreviation: IQR, interquartile range.

appropriate antimicrobials 46 (7.2) 46 (100.0) <0.0001 Received appropriate definitive therapy 589 (92.8) 434 (73.7) Potential candidates for combined antibiotic therapy 310 (48.8) Received single therapy 226 (72.9) 167 (73.9) 0.01 Received combination therapy 84 (27.1) 50 (59.5) Abbreviation: IQR, interquartile range. * P values are for the comparisons of mortality. ** Comparison of mortality between inappropriate and appropriate initial antimicrobial therapy. Impact of Inappropriateness of Initial Antimicrobial Therapy on Hospital Mortality The use of inappropriate antimicrobials as initial therapy was associated with a significant increase in mortality (aOR, 9.5; 95% CI, 4.3–20.7) (Fig. 1A). Tests of interaction indicated that this finding was consistent in all tested subgroups of patients, whether with documented or suspected infections, culture-positive or culture-negative, bacteremia or no bacteremia, community-acquired or nosocomial infections, gram-positive or gram-negative infections, pneumonia, intra-abdominal infection, immunocompromised or non-immunocompromised, and across countries (Canada, United States, and Saudi Arabia) and the four study periods. Results of the interaction tests are presented in the Supporting Information. Figure 1A shows the aOR and 95% CI for selected subgroups.

nfections, pneumonia, intra-abdominal infection, immunocompromised or non-immunocompromised, and across countries (Canada, United States, and Saudi Arabia) and the four study periods. Results of the interaction tests are presented in the Supporting Information. Figure 1A shows the aOR and 95% CI for selected subgroups. Fig 1 Association of inappropriate antimicrobial therapy (A), hours of delay in effective antimicrobial therapy (B), and use of single versus combined antimicrobial therapy (C) with hospital mortality across various subgroups of patients using multivariate analyses. The following independent variables were entered in the model: APACHE II score, MELD score, immunocompromised (versus non-immunocompromised), bacteremia (versus no bacteremia), community-acquired (versus nosocomial), and culture- positive (versus culture-negative). The results are shown as aOR and 95% CI on a logarithmic scale. Impact of Timing of Initial Antimicrobial Therapy on Hospital Mortality The delay in use of appropriate antimicrobials was associated with a significant increase in mortality (aOR, 1.1; 95% CI, 1.1–1.2 per hour of delay) after onset of shock (Figs. 1B and 2). Tests of interaction indicated that this finding was also consistent in all tested subgroups of patients (see Supporting Information). Figure 1B shows the aOR and 95% CI for selected subgroups.

with a significant increase in mortality (aOR, 1.1; 95% CI, 1.1–1.2 per hour of delay) after onset of shock (Figs. 1B and 2). Tests of interaction indicated that this finding was also consistent in all tested subgroups of patients (see Supporting Information). Figure 1B shows the aOR and 95% CI for selected subgroups. Fig 2 aOR and 95% CI of hospital mortality (on logarithmic scale) by the time from the onset of hypotension to the antimicrobial therapy in hours. Adjustments were made for the following independent variables: APACHE II score, MELD score, immunocompromised (versus non-immunocompromised), bacteremia (versus no bacteremia) community-acquired (versus nosocomial), and culture-positive (versus culture-negative). Impact of Combination Antibiotic Therapy on Hospital Mortality The use of a single antibiotic for bacterial septic shock was associated with a significant increase in mortality (aOR, 1.8; 95% CI, 1.0–3.3) (Fig. 1C). Tests of interaction indicated that this finding was also consistent in all tested subgroups of patients (see Supporting Information). Figure 1C shows the aOR and 95% CI for selected subgroups.

otic for bacterial septic shock was associated with a significant increase in mortality (aOR, 1.8; 95% CI, 1.0–3.3) (Fig. 1C). Tests of interaction indicated that this finding was also consistent in all tested subgroups of patients (see Supporting Information). Figure 1C shows the aOR and 95% CI for selected subgroups. Clinical, Laboratory, and Microbiological Predictors of Suboptimal Antimicrobial Therapy We identified the multidrug-resistant organisms (aOR, 3.1; 95% CI, 1.5–6.4) as a predictor of inappropriate antimicrobial therapy. The following were the predictors of delay in initial antimicrobial therapy: patients who had a lower presenting temperature (P = 0.003), higher initial serum bicarbonate concentration (P = 0.02), nosocomial infections (P = 0.0009), and who were female (P = 0.05). We did not find any significant predictors of single versus combined antimicrobial therapy on multivariate analysis. When appropriateness, timing, and combination of antimicrobials were compared according to the micro-organisms, fungal infections were found to be more likely associated with inappropriate and delayed antimicrobial therapy (P < 0.001 for both). We found the following variables to be significantly associated with the development of fungal infections: higher MELD score (P = 0.007), chronic renal failure (P = 0.009), higher bilirubin (P = 0.002), lower heart rate (P = 0.03), and lower body mass index (P = 0.05) on univariate analysis.

erapy (P < 0.001 for both). We found the following variables to be significantly associated with the development of fungal infections: higher MELD score (P = 0.007), chronic renal failure (P = 0.009), higher bilirubin (P = 0.002), lower heart rate (P = 0.03), and lower body mass index (P = 0.05) on univariate analysis. Discussion In this study, the hospital mortality of patients with cirrhosis and septic shock was high. We found that inappropriate and delayed appropriate initial empiric antimicrobials were associated with significant increase in mortality. We also found that the empiric use of a single appropriate drug (compared with combination therapy with two or more antibiotics active for proven or suspected pathogens) for bacterial septic shock was associated with a significant increase in mortality. These findings were consistent among various clinically relevant subgroups.

ound that the empiric use of a single appropriate drug (compared with combination therapy with two or more antibiotics active for proven or suspected pathogens) for bacterial septic shock was associated with a significant increase in mortality. These findings were consistent among various clinically relevant subgroups. Our study describes some of the unique features of septic shock in patients with cirrhosis. We found that patients with cirrhosis and septic shock were younger and had higher APACHE II scores compared with the general cohort of patients described.21–23 Similarly, body temperature at the time of presentation with septic shock was lower.21–23 Spontaneous bacterial peritonitis was present in 17.6% of the patients at the time of onset of septic shock. Escherichia coli and Staphylococcus aureus were the most common bacterial pathogens, and a strikingly high number of fungal infections (9.3%) were found. The use of activated protein C (1.4%) and corticosteroids (30.2%) were relatively low compared with that in the general cohort of patients described.21–23

ptic shock. Escherichia coli and Staphylococcus aureus were the most common bacterial pathogens, and a strikingly high number of fungal infections (9.3%) were found. The use of activated protein C (1.4%) and corticosteroids (30.2%) were relatively low compared with that in the general cohort of patients described.21–23 We found a significant increase in mortality with inappropriate initial antimicrobial use. This is consistent with the findings of Kumar et al. in a heterogeneous cohort of septic shock patients.22 Although a sizable amount of literature is available on this topic,29 we are unaware of other similar work exploring such associations in a cohort of patients with cirrhosis. The strength of this finding is that it was consistent across various subgroups, including patients with gram-positive and gram-negative infections, pneumonia, and intra-abdominal infections. As such, inappropriate initial empiric antimicrobial therapy appears to be a strong yet modifiable determinant of outcomes of septic shock in patients with cirrhosis.

t was consistent across various subgroups, including patients with gram-positive and gram-negative infections, pneumonia, and intra-abdominal infections. As such, inappropriate initial empiric antimicrobial therapy appears to be a strong yet modifiable determinant of outcomes of septic shock in patients with cirrhosis. Delay in the initial empiric administration of appropriate antimicrobials is also associated with higher mortality. Over the last two decades, a large body of knowledge has emerged showing that timely antimicrobial administration is associated with significant improvement in outcomes.30, 31 This has been demonstrated in severe sepsis/septic shock,21 pneumonia,32 meningitis,33 bacteremia, and fungemia.31 Timely administration of antibiotics is a measure of the quality of care for community-acquired pneumonia. However, similar data are not available for patients with cirrhosis and septic shock. Given the high mortality of this group of patients, it is entirely possible that earlier administration of appropriate antimicrobial therapy would have resulted in better outcomes.

the quality of care for community-acquired pneumonia. However, similar data are not available for patients with cirrhosis and septic shock. Given the high mortality of this group of patients, it is entirely possible that earlier administration of appropriate antimicrobial therapy would have resulted in better outcomes. The final significant finding is that the empiric use of a single appropriate antibiotic (monotherapy) is associated with increased mortality in bacterial septic shock in this cohort. To our knowledge, no previous studies have addressed the exact question. A small study by McCormick et al.34 evaluated the efficacy and incidence of renal impairment with netilmicin plus mezlocillin compared with ceftazidime among 128 patients with cirrhosis and sepsis. Mortality rates were similar in the two groups. The reasons for our finding of a survival advantage with combination antibiotic therapy are not clear. This finding cannot be explained by a higher rate of coverage with combination therapy resulting from a high incidence of resistant bacteria in patients with cirrhosis who are often on prophylactic antibiotics. This is because combination therapy was defined as two or more antibiotics that were active for the isolated or suspected (in culture negative cases) pathogens. Additionally, the survival advantage was consistent in patients with or without multidrug-resistant organisms. We believe this is a novel finding that needs further exploration, given the high mortality and morbidity associated with septic shock in patients with cirrhosis.

spected (in culture negative cases) pathogens. Additionally, the survival advantage was consistent in patients with or without multidrug-resistant organisms. We believe this is a novel finding that needs further exploration, given the high mortality and morbidity associated with septic shock in patients with cirrhosis. Our data support a paradigm shift in the way we think about the natural progression of patients with cirrhosis. The natural course of cirrhosis has been considered to be irreversible and often fatal, except for patients who receive a liver transplant.7 Acute on chronic liver failure (ACLF) is a newly defined entity in which an acute insult in a previously compensated liver disease leads to deterioration and organ failure,35 which is partially reversible when identified early and patients receive early and appropriate intensive care support.7 A second principle in defining ACLF is the presence of an identifiable precipitating event, which in most cases is infection and sepsis.7 To improve outcomes in cirrhosis, early identification and management of these events is essential.7 We believe that our study contributes to this emerging field by guiding key aspects of antimicrobial therapy. Although the focus has been on therapies with unproven efficacy and safety profiles and low cost-effectiveness, such as liver support systems, the answer may be in redesigning the way we deliver routine care such as antimicrobial therapy to these patients.

emerging field by guiding key aspects of antimicrobial therapy. Although the focus has been on therapies with unproven efficacy and safety profiles and low cost-effectiveness, such as liver support systems, the answer may be in redesigning the way we deliver routine care such as antimicrobial therapy to these patients. Although the mortality of patients with cirrhosis who develop septic shock is very high, the diagnosis and treatment of this combination has been poorly studied. Part of the problem is the overlapping findings in sepsis and cirrhosis. Patients with cirrhosis have low baseline blood pressure, higher baseline heart rate, higher baseline breathing rate, and tend to not mount a vigorous febrile response.18 As such, identification of systemic inflammatory response syndrome in cirrhosis may prove difficult.36 Our study reflects this difficulty, as 25.5% of patients met the criteria for septic shock but were culture negative. Nevertheless, tests of interaction showed that the associations of appropriateness and timing of initial antimicrobial therapy and the use of single versus combination therapy were similar in culture-positive and culture-negative patients and in patients with documented or suspected infection. Furthermore, little progress has been made in the management of patients with cirrhosis and sepsis, as these patients tend to be excluded from studies of therapeutics in severe sepsis, such as the study of activated protein C in severe sepsis.15 Because the baseline central and mixed venous oxygen saturation tends to be higher in patients with cirrhosis,18 this specific goal for early goal directed therapy16 may not be applicable to these patients. Patients with cirrhosis are more prone to hypoglycemia and are not suitable candidates for intensive insulin therapy, either. Although antibiotics are commonly used in prophylaxis and treatment, the choice, timing, combinations and dosing have not been well studied. Our study addresses some of these points. The data suggest that appropriate and timely antimicrobial therapy and combination antibiotic therapy should be initiated before or, at the latest, concurrent with the onset of the hypotension of septic shock that typically happens several hours prior to ICU admission.

studied. Our study addresses some of these points. The data suggest that appropriate and timely antimicrobial therapy and combination antibiotic therapy should be initiated before or, at the latest, concurrent with the onset of the hypotension of septic shock that typically happens several hours prior to ICU admission. Possible explanations for the observed patterns of antimicrobial use may include process of care–related variations. Factors that influence prescription, transcription, preparation, dispensing and administration of antimicrobials among patients with cirrhosis and septic shock need to be investigated further as possible root causes. However, our study was not designed to delve into these issues. Our study should be interpreted in light of its strengths and limitations. The strengths include the inclusion of patients from 28 ICUs based in three geographic regions. This lends the results of the study wide generalizability. To our knowledge, this is the first study specifically addressing the impact of various aspects of antibiotic use on outcomes among patients with cirrhosis and septic shock.

ngths include the inclusion of patients from 28 ICUs based in three geographic regions. This lends the results of the study wide generalizability. To our knowledge, this is the first study specifically addressing the impact of various aspects of antibiotic use on outcomes among patients with cirrhosis and septic shock. In terms of limitations, the results were not from a randomized controlled study. As such, the findings only highlight associations, and cause–effect relationships cannot be inferred. However, the combination and monotherapy groups, appropriate and inappropriate groups, and delay and early groups were generally comparable, and severity of illness as measured by APACHE II were not different. Observational studies such as ours are susceptible to confounding.37 Regression analysis is one way of adjusting for this in the statistical analysis.37 There are a number of possible factors that may influence outcome in acutely ill medical patients. Although we adjusted for severity of illness as measured by APACHE II and MELD scores, we cannot rule out residual confounding. However, the consistent and robust findings across various subgroups make it very unlikely that these findings are related to confounders alone. Furthermore, our classification of community-acquired and nosocomial infection followed the definitions used at the time of initiation of the database, and as such did not utilize the more recent concept of health care–associated infections that was introduced later.38 Nevertheless, we do not think this point affects the overall findings of the study, because the associations were observed in both groups.

the definitions used at the time of initiation of the database, and as such did not utilize the more recent concept of health care–associated infections that was introduced later.38 Nevertheless, we do not think this point affects the overall findings of the study, because the associations were observed in both groups. In conclusion, this study shows that the inappropriate and delayed empiric antimicrobial therapy and single initial antibiotic therapy in patients with cirrhosis and septic shock is associated with significant increase in hospital mortality. Efforts need to focus on improving the choice and timing of empiric antibiotic therapy in this high-risk group. We thank Christine Mendez, Sheena Ablang, Debbie Friesen, and Lisa Halstead for data entry.

In conclusion, this study shows that the inappropriate and delayed empiric antimicrobial therapy and single initial antibiotic therapy in patients with cirrhosis and septic shock is associated with significant increase in hospital mortality. Efforts need to focus on improving the choice and timing of empiric antibiotic therapy in this high-risk group. We thank Christine Mendez, Sheena Ablang, Debbie Friesen, and Lisa Halstead for data entry. Appendix Additional members of the CATSS Database Research Group include: Phillip Dellinger, M.D., Cooper Hospital/University Medical Center, Camden, NJ; Peter Dodek, M.D., St. Paul's Hospital, Vancouver, BC, Canada; Paul Ellis, M.D., University Health Network, Toronto, ON, Canada; Dave Gurka, M.D., Rush-Presbyterian-St. Luke's Medical Center, Chicago, IL; Jose Guzman, Harper Hospital, Detroit, MI; Sean Keenan, M.D., Royal Columbian Hospital, New Westminster, BC, Canada; Andreas Kramer, M.D., Brandon General Hospital, Brandon, MB, Canada; Aseem Kumar, Laurentian University, Sudbury, ON, Canada; Denny Laporta, M.D., Jewish General Hospital, Montreal, QC, Canada; Kevin Laupland, M.D., Foothills Hospital, Calgary, AB, Canada; Bruce Light, M.D., Winnipeg Regional Health Authority, Winnipeg, MB, Canada; Dennis Maki, M.D., University of Wisconsin Hospital and Clinics, Madison, WI; John Marshall, M.D., St. Michael's Hospital, Toronto, ON, Canada; Greg Martinka, M.D., Richmond General Hospital, Richmond, BC, Canada; Yazdan Mirzanejad, M.D., Surrey Memorial Hospital, Surrey, BC, Canada; Gourang Patel, Pharm.D., Rush-Presbyterian-St. Luke's Medical Center, Chicago, IL; Charles Penner, M.D., Brandon General Hospital, Brandon, MB, Canada; Dan Roberts, M.D., Winnipeg Regional Health Authority, Winnipeg, MB, Canada; John Ronald, M.D., Nanaimo Regional Hospital, Nanaimo, BC, Canada; Dave Simon, M.D., Rush-Presbyterian-St. Luke's Medical Center, Chicago IL; Sat Sharma, M.D., Winnipeg Regional Health Authority, Winnipeg, MB, Canada; Yoanna Skrobik, M.D., Hôpital Maisonneuve Rosemont, Montreal, QC, Canada; Kenneth E. Wood, DO, University of Wisconsin Hospital and Clinics, Madison, WI; Sergio Zanotti, M.D., Cooper Hospital/University Medical Center, Camden, NJ.

, Chicago IL; Sat Sharma, M.D., Winnipeg Regional Health Authority, Winnipeg, MB, Canada; Yoanna Skrobik, M.D., Hôpital Maisonneuve Rosemont, Montreal, QC, Canada; Kenneth E. Wood, DO, University of Wisconsin Hospital and Clinics, Madison, WI; Sergio Zanotti, M.D., Cooper Hospital/University Medical Center, Camden, NJ. Associate members of the CATSS Database Research Group include: Muhammed Wali Ahsan, M.D., Winnipeg, MB, Canada; Mozdeh Bahrainian, M.D., Madison, WI; Rob Bohmeier, University of Manitoba, Winnipeg, MB, Canada; Lindsey Carter, BA, Winnipeg, MB, Canada; Harris Chou, BSc, of British Columbia, Vancouver, BC, Canada; Sofia Delgra, R.N., King Abdulaziz Medical City, Riyadh, Saudi Arabia; Winnie Fu, University of British Columbia, Vancouver, BC, Canada; Catherine Gonzales, R.N., King Abdulaziz Medical City, Riyadh, Saudi Arabia; Harleena Gulati, M.D., University of Manitoba, Winnipeg, MB, Canada; Erica Halmarson, M.D., University of Manitoba, Winnipeg, MB, Canada; Ziaul Haque, M.D., Montreal, QC, Canada; Johanne Harvey, R.N., Hôpital Maisonneuve Rosemont, Montreal, QC, Canada; Farah Khan, M.D., Toronto, ON, Canada; Laura Kolesar, R.N., St. Boniface Hospital, Winnipeg, MB, Canada; Laura Kravetsky, M.D., University of Manitoba, Winnipeg, MB, Canada; Runjun Kumar, University of Toronto, Toronto, ON, Canada; Nasreen Merali, M.D., Winnipeg, MB, Canada; Sheri Muggaberg, University of Manitoba, Winnipeg, MB, Canada; Heidi Paulin, University of Toronto, Toronto, ON, Canada; Cheryl Peters, R.N., M.D., University of Manitoba, Winnipeg, MB, Canada; Jody Richards, Camosun College, Victoria, BC, Canada; Christa Schorr, R.N., Cooper Hospital/University Medical Center, Camden, NJ; Norrie Serrano, R.N., King Abdulaziz Medical City, Riyadh, Saudi Arabia; Mustafa Suleman, M.D., Concordia Hospital, Winnipeg, MB; Amrinder Singh, M.D., Winnipeg, MB, Canada; Katherine Sullivan, University of Manitoba, Winnipeg, MB, Canada; Robert Suppes, M.D., University of Manitoba, Winnipeg, MB, Canada; Leo Taiberg, M.D., Rush Medical College, Chicago IL; Ronny Tchokonte, M.D., Wayne State University Medical School, Detroit, MI; Omid Ahmadi Torshizi, M.D., Montreal, QC, Canada; Kym Wiebe, R.N., St. Boniface Hospital, Winnipeg, MB, Canada.

, MB, Canada; Robert Suppes, M.D., University of Manitoba, Winnipeg, MB, Canada; Leo Taiberg, M.D., Rush Medical College, Chicago IL; Ronny Tchokonte, M.D., Wayne State University Medical School, Detroit, MI; Omid Ahmadi Torshizi, M.D., Montreal, QC, Canada; Kym Wiebe, R.N., St. Boniface Hospital, Winnipeg, MB, Canada. Additional Supporting Information may be found in the online version of this article. ACLFacute on chronic liver failure aORadjusted odds ratio APACHEacute physiology and chronic health evaluation CATSSCooperative Antimicrobial Therapy of Septic Shock CIconfidence interval ICUintensive care unit MELDmodel for end-stage liver disease WBCwhite blood cell.

Preamble The present version of the American Association for the Study of Liver Diseases (AASLD) Position Paper represents a thorough overhaul from the previous version of 2005. In addition to two new additional authors, the revision includes updated expert opinion regarding (1) etiologies and diagnosis, (2) therapies and intensive care management, and (3) prognosis and transplantation. Because acute liver failure (ALF) is an orphan disease, large clinical trials are impossible and much of its management is based on clinical experience only. Nonetheless, there are certain issues that continue to recur in this setting as well as growing consensus (amidst innovation) regarding how to maximize the ALF patient's chance of recovery. The changes in ALF management are not global in nature, but are more consistent with incremental experience and improvements in diagnosis and intensive care unit management.

continue to recur in this setting as well as growing consensus (amidst innovation) regarding how to maximize the ALF patient's chance of recovery. The changes in ALF management are not global in nature, but are more consistent with incremental experience and improvements in diagnosis and intensive care unit management. Introduction The diagnosis of ALF hinges on identifying that the patient has an acute insult and is encephalopathic. Imaging in recent years has suggested “cirrhosis,” but this is often an overcall by radiology, because a regenerating massively necrotic liver will give the same nodular profile as cirrhosis.1 It is vital to promptly get viral hepatitis serologies, including A-E as well as autoimmune serologies, because these often seem to be neglected at the initial presentation. Fulminant Wilson's disease can be diagnosed most effectively not by waiting for copper levels (too slow to obtain) or by obtaining ceruloplasmin levels (low in half of all ALF patients, regardless of etiology), but by simply looking for the more readily available bilirubin level (very high) and alkaline phosphatase (ALP; very low), such that the bilirubin/ALP ratio exceeds 2.0. 2 The availability of an assay that measures acetaminophen adducts has been used for several years as a research tool and has improved our clinical recognition of acetaminophen cases when the diagnosis is obscured by patient denial or encephalopathy. 3 Any patient with very high aminotransferases and low bilirubin on admission with ALF very likely has acetaminophen overdose, with the one possible exception being those patients who enter with ischemic injury. Obtaining autoantibodies should be routine and a low threshold for biopsy in patients with indeterminate ALF should be standard, given that autoimmune hepatitis may be the largest category of indeterminate, after unrecognized acetaminophen poisoning. 4

ible exception being those patients who enter with ischemic injury. Obtaining autoantibodies should be routine and a low threshold for biopsy in patients with indeterminate ALF should be standard, given that autoimmune hepatitis may be the largest category of indeterminate, after unrecognized acetaminophen poisoning. 4 Advances in Management of ALF The medical management of ALF has not been extensively studied and remains poorly defined. In the absence of evidence-based clinical trials, experts from 23 centers in the United States have proposed detailed management guidelines by consensus.5 Since the last AASLD Position Paper, several noteworthy advances have been made in assessing the risk of developing, and managing, specific complications of ALF.

ed. In the absence of evidence-based clinical trials, experts from 23 centers in the United States have proposed detailed management guidelines by consensus.5 Since the last AASLD Position Paper, several noteworthy advances have been made in assessing the risk of developing, and managing, specific complications of ALF. A detailed analysis of serum ammonia in patients with ALF identified a concentration of 75 μM as an important threshold below which patients rarely develop intracranial hypertension (ICH).6 Conversely, arterial ammonia levels of >100 μM on admission represent an independent risk factor for the development of high-grade hepatic encephalopathy, and a level of >200 μM predicts ICH. The risk of developing ICH is decreased by raising the serum sodium to 145-155 mEq/L with hypertonic saline. 7 Once established, however, the medical treatment of ICH must bridge patients to liver transplantation, because no treatment permanently reverses cerebral edema. In cases of ICH refractory to osmotic agents (e.g., mannitol and hypertonic saline), therapeutic hypothermia (cooling to a core temperature of 32°C-34°C) has been shown to bridge patients to transplantation, 8 but is associated with a theoretical risk of impairing liver regeneration.

ermanently reverses cerebral edema. In cases of ICH refractory to osmotic agents (e.g., mannitol and hypertonic saline), therapeutic hypothermia (cooling to a core temperature of 32°C-34°C) has been shown to bridge patients to transplantation, 8 but is associated with a theoretical risk of impairing liver regeneration. To optimize neurological recovery after ALF, mean arterial pressure (MAP) and cerebral perfusion pressure (CPP) must be raised to avoid cerebral underperfusion and anoxia. In hypotensive patients with ALF, intravascular volume should be repleted first with normal saline, and vasopressors should be administered subsequently to titrate the MAP to >75 mmHg and CPP to 60-80 mmHg. Vasopressin, or its analog, terlipressin, is often added to norepinephrine in critically ill patients who remain hypotensive on norepinephrine, but was reported to increase intracranial pressure (ICP) in patients with ALF.9 More recent data suggest, however, that vasopressin and analogs increase cerebral perfusion without increasing ICP and may be used safely as an adjunct to norepinephrine. 10

critically ill patients who remain hypotensive on norepinephrine, but was reported to increase intracranial pressure (ICP) in patients with ALF.9 More recent data suggest, however, that vasopressin and analogs increase cerebral perfusion without increasing ICP and may be used safely as an adjunct to norepinephrine. 10 It is generally accepted that patients with ALF have a bleeding diathesis based upon elevation of the international normalized ratio (INR). Concern about the safety of inserting ICP monitors and other invasive devices has prompted the use of recombinant factor VIIa,11 although the practice has been associated with thrombotic complications in patients with ALF. 12 However, a recent study has suggested that global hemostasis assessed by thromboelastography usually remains normal, suggesting that the perceived bleeding risk based upon INR may be overstated. 13 Prognosis and Transplantation To date, it often remains difficult to predict which ALF patients will ultimately require transplantation. Newer models, including the model for end-stage liver disease (MELD) score, have not improved our accuracy. In fact, the discriminative power of the MELD was not found to be superior to that of the INR or the King's College Hospital criteria.14 In addition, equating transplantation with death, in many models, inflates the positive predictive value of a particular system. The King's College Criteria remain the most clinically useful, with a sensitivity of 68%-69% and a specificity of 82%-92%. 15 However, reliance entirely upon any set of guidelines cannot be recommended.

, equating transplantation with death, in many models, inflates the positive predictive value of a particular system. The King's College Criteria remain the most clinically useful, with a sensitivity of 68%-69% and a specificity of 82%-92%. 15 However, reliance entirely upon any set of guidelines cannot be recommended. Despite great early interest in liver support systems, the field has had little forward movement since our last publication. Both artificial (i.e., sorbent-based) and bioartificial (i.e., cell-based) systems have been tested. There has been no good evidence that any artificial support system reliably reduces mortality in the setting of ALF.16, 17 Thus, the currently available liver support systems cannot be recommended outside of clinical trials. Liver transplantation remains the only definitive treatment for patients who fail to demonstrate recovery. The 1-year survival after cadaveric liver transplant for ALF is less than that observed in patients transplanted for chronic liver failure.18 However, after the first year, this trend had reversed and ALF patients have a better long-term survival. The use of live donor liver transplantation and auxiliary liver transplant remain controversial. 19 Urgent cadaveric liver transplantation remains the standard of care in the setting of ALF. Developing effective methods of liver support or other alternatives to transplantation and better prognostic scoring systems remain key goals to further improve overall survival rates and avoid unnecessary transplants. Abbreviations AASLDthe American Association for the Study of Liver Diseases

Liver transplantation remains the only definitive treatment for patients who fail to demonstrate recovery. The 1-year survival after cadaveric liver transplant for ALF is less than that observed in patients transplanted for chronic liver failure.18 However, after the first year, this trend had reversed and ALF patients have a better long-term survival. The use of live donor liver transplantation and auxiliary liver transplant remain controversial. 19 Urgent cadaveric liver transplantation remains the standard of care in the setting of ALF. Developing effective methods of liver support or other alternatives to transplantation and better prognostic scoring systems remain key goals to further improve overall survival rates and avoid unnecessary transplants. Abbreviations AASLDthe American Association for the Study of Liver Diseases ALFacute liver failure ALPalkaline phosphatase CPPcerebral perfusion pressure ICHintracranial hypertension ICPintracranial pressure INRinternational normalized ratio MAPmean arterial pressure MELDmodel for end-stage liver disease.

High plasma total and low-density lipoprotein (LDL) cholesterol levels are linked to an enhanced risk of developing premature atherosclerosis. Thyroid hormone (TH) is an important regulator of cholesterol metabolism and hyperthyroidism is commonly associated with decreased—and hypothyroidism with increased—plasma cholesterol concentrations.1–3 TH is known to exert a number of beneficial effects on cholesterol and lipoprotein metabolism,4 and promising results have recently been reported from the clinical development of liver-selective TH analogs, such as eprotirome.5, 6 One of the mechanisms by which TH may lower plasma cholesterol is by an increased secretion of biliary cholesterol,7 a main route for elimination of cholesterol from the body.8

bolism,4 and promising results have recently been reported from the clinical development of liver-selective TH analogs, such as eprotirome.5, 6 One of the mechanisms by which TH may lower plasma cholesterol is by an increased secretion of biliary cholesterol,7 a main route for elimination of cholesterol from the body.8 An important and presumably rate-limiting step in the process of biliary secretion of cholesterol is mediated by the half-transporters ATP-binding cassette, subfamily G (WHITE), member 5 (ABCG5) and ATP-binding cassette, subfamily G (WHITE), member 8 (ABCG8). By heterodimerization with each other, these structures form a functional complex that promotes the transport of cholesterol and plant sterols from liver cells into bile9–11 at the apical plasma membrane of hepatocytes.10, 11 Disruption of either one12, 13 or both11, 14–16 genes reduces biliary cholesterol concentration and secretion in mice. In contrast, induction of hepatic ABCG5/G8 gene expression is associated with increased biliary cholesterol concentration and secretion.15–17 In a previous study,18 biliary cholesterol secretion was strongly reduced in hypophysectomized rats as compared with intact animals, a finding associated with markedly reduced hepatic ABCG5/G8 gene expression. The administration of TH increased biliary cholesterol secretion and, concomitantly, hepatic ABCG5/G8 gene expression levels were increased. This suggests that TH-induced stimulation of biliary cholesterol secretion may be mediated by ABCG5/G8.

finding associated with markedly reduced hepatic ABCG5/G8 gene expression. The administration of TH increased biliary cholesterol secretion and, concomitantly, hepatic ABCG5/G8 gene expression levels were increased. This suggests that TH-induced stimulation of biliary cholesterol secretion may be mediated by ABCG5/G8. Hepatic gene expression of ABCG5/G8 is not always concurrent with biliary cholesterol secretion,16, 19–21 however, and there are indications that other pathways, independent of ABCG5/G8, promote cholesterol transfer into bile.11, 22, 23 Furthermore, it is unclear whether the stimulation of biliary cholesterol secretion is a direct effect of TH. It may well be mediated by nuclear liver x receptor-alpha (LXRa), the expression of which has recently been reported to be positively regulated by TH receptor-beta (TRb) in the mouse.24 LXRa regulates the transcription of several genes involved in cholesterol metabolism25 and the administration of the LXR agonist T0901317 to mice increases hepatic ABCG5/G8 gene expression and biliary cholesterol concentration and secretion.15, 17, 26 Thus, experimental evidence indicates that LXRa may mediate effects of TH on cholesterol metabolism.27 Here, we investigated whether the induction of biliary cholesterol secretion by TH is dependent on the ABCG5/G8 complex or if other mechanisms are involved. Furthermore, the question of if LXRa is important for the effect of TH on biliary cholesterol secretion was explored.

Hepatic gene expression of ABCG5/G8 is not always concurrent with biliary cholesterol secretion,16, 19–21 however, and there are indications that other pathways, independent of ABCG5/G8, promote cholesterol transfer into bile.11, 22, 23 Furthermore, it is unclear whether the stimulation of biliary cholesterol secretion is a direct effect of TH. It may well be mediated by nuclear liver x receptor-alpha (LXRa), the expression of which has recently been reported to be positively regulated by TH receptor-beta (TRb) in the mouse.24 LXRa regulates the transcription of several genes involved in cholesterol metabolism25 and the administration of the LXR agonist T0901317 to mice increases hepatic ABCG5/G8 gene expression and biliary cholesterol concentration and secretion.15, 17, 26 Thus, experimental evidence indicates that LXRa may mediate effects of TH on cholesterol metabolism.27 Here, we investigated whether the induction of biliary cholesterol secretion by TH is dependent on the ABCG5/G8 complex or if other mechanisms are involved. Furthermore, the question of if LXRa is important for the effect of TH on biliary cholesterol secretion was explored. We present three novel findings: (1) Biliary cholesterol secretion induced by TH is predominantly excerted by ABCG5/G8; (2) this TH-induced biliary cholesterol secretion is independent of LXRa; and (3) a minor part of the TH-induced stimulation of biliary cholesterol secretion occurs independently of the ABCG5/G8 complex.

We present three novel findings: (1) Biliary cholesterol secretion induced by TH is predominantly excerted by ABCG5/G8; (2) this TH-induced biliary cholesterol secretion is independent of LXRa; and (3) a minor part of the TH-induced stimulation of biliary cholesterol secretion occurs independently of the ABCG5/G8 complex. Materials and Methods Animals and Treatments Male mice (3–5 months of age) were used in the experiments. In the first experiment, mice homozygous for the disruption of the ABCG5 gene (Abcg5−/−) and their wild-type (WT) counterparts (Abcg5+/+) were divided into the following groups: Abcg5+/+ (n = 6); Abcg5+/+ T3 (n = 5); Abcg5−/− (n = 5); and Abcg5−/− T3 (n = 6). In the second experiment, mice homozygous for the disruption of the LXRa gene (Lxra−/−) and their WT counterparts (Lxra+/+) were divided into the following groups: Lxra+/+; Lxra+/+ T3; Lxra−/−; and Lxra−/− T3 (all groups: n = 7). For detailed descriptions of how knockout mice were generated, see previous reports.13, 21

In the second experiment, mice homozygous for the disruption of the LXRa gene (Lxra−/−) and their WT counterparts (Lxra+/+) were divided into the following groups: Lxra+/+; Lxra+/+ T3; Lxra−/−; and Lxra−/− T3 (all groups: n = 7). For detailed descriptions of how knockout mice were generated, see previous reports.13, 21 Animals were housed in a temperature-controlled environment, with lights on from 6 a.m. to 6 p.m. They had free access to drinking water and mouse chow. Groups treated with T3 (Abcg5+/+ T3, Abcg5−/− T3, Lxra+/+ T3, and Lxra−/− T3) received drinking water supplemented with 0.5 μg of T3/mL (3,3′,5-triiodo-L-thyronine; Sigma-Aldrich, St Louis, MO) and 0.01% albumin (bovine serum albumin; Sigma-Aldrich). After 14 days of treatment, mice were anesthetized by an intraperitoneal injection of Hypnorm (fentanyl/fluanisone, 1 mL/kg) and diazepam (10 mg/kg). Bile was collected for 30 minutes from cannulated gallbladders, as previously described,21 and blood was collected by heart puncture at the end of the bile-collection period. After animals had been killed by cervical dislocation, livers were removed and immediately frozen in liquid nitrogen and stored at −80°C. All experimental procedures were approved by the Local Ethical Committee for Animal Experiments of the University of Groningen.

art puncture at the end of the bile-collection period. After animals had been killed by cervical dislocation, livers were removed and immediately frozen in liquid nitrogen and stored at −80°C. All experimental procedures were approved by the Local Ethical Committee for Animal Experiments of the University of Groningen. RNA Isolation and Real-Time PCR Measurements Total RNA was extracted from individual samples of liver and proximal small intestine using TRIzol Reagent (Invitrogen, Carlsbad, CA), according to the manufacturer's instructions. cDNA synthesis was performed using Omniscript reverse transcriptase (Qiagen, Hilden, Germany). Quantitative real-time PCR was performed with SYBRGreen PCR MasterMix on a 7500 Fast Real-Time PCR System, and primers were designed using Primer Express Software 2.0 (Applied Biosystems, Foster City, CA). Glyceraldehyde-3-phosphate dehydrogenase (Gapdh) and hypoxanthine guanine phosphoribosyl transferase (Hprt) were used as endogenous controls, and the comparative Ct method was used to quantify the results. The following primer sequences were used: Gapdh forward 5′-tgtgtccgtcgtggatctga-3′; Gapdh reverse 5′-cctgcttcaccaccttcttgat-3′; Abcg5 forward 5′-aatgctgtg aatctgtttccca-3′; Abcg5 reverse 5′-ccacttatgatacaggcca tcct-3′; Abcg8 forward 5′-tccatcctcggagacacgat-3′; Abcg8 reverse 5′-gctgatgccgatgacaatga-3′; Lxra forward 5′-gctct gctcattgccatcag-3′; Lxra reverse 5′-tgttgcagcctctctactt gga-3′; Hprt forward 5′-ggtgaaaaggacctctcgaagtg-3′; Hprt reverse 5′-atagtcaagggcatatccaacaaca-3′; Cyp7a1 forward 5′-agcacctaaacaacctgccagtacta-3′; Cyp7a1 reverse 5′-gtccggatattcaaggatgca-3′; Hmgcr forward 5′-tgattggagttggcaccat-3′; Hmgcr reverse 5′- tggccaacactga catgc-3′; Ldlr forward 5′-ggatggctatacctacccctcaa-3′; and Ldlr reverse 5′-cacatcgtcctccaggctg-3′.

gaagtg-3′; Hprt reverse 5′-atagtcaagggcatatccaacaaca-3′; Cyp7a1 forward 5′-agcacctaaacaacctgccagtacta-3′; Cyp7a1 reverse 5′-gtccggatattcaaggatgca-3′; Hmgcr forward 5′-tgattggagttggcaccat-3′; Hmgcr reverse 5′- tggccaacactga catgc-3′; Ldlr forward 5′-ggatggctatacctacccctcaa-3′; and Ldlr reverse 5′-cacatcgtcctccaggctg-3′. Assay of Biliary Cholesterol Concentration and Secretion 25 μL of bile was used for this assay. After Folch extraction, dried samples were hydrolyzed with 1 mL of 0.5 M KOH at 70°C for 90 min. Samples were extracted by the addition of 1 mL of H2O and 5 mL of hexane. After centrifugation at 3,000 rpm for 5 min, the upper phase was evaporated under nitrogen and silylated with pyridine/hexametyldisilazane/chlorotrimetylsilane (3:2:1, v/v/v) at 60°C for 30 min. After evaporation, the product was redissolved in hexane and analyzed using gas chromatography/mass spectrometry (GC/MS). D7-cholesterol was used as internal standard. Biliary cholesterol secretion was calculated for each individual by multiplying the cholesterol concentration by the volume of bile secreted per minute and per 100 g of body weight. Assay of Biliary Phospholipid Concentration and Secretion Phospholipids were extracted from individual bile samples, as previously described.28 The concentration was subsequently determined as in Böttcher et al.29 Secretion of phospholipids was calculated for each individual by multiplying the concentration of phospholipids by the volume of bile secreted per minute and per 100 g of body weight.

extracted from individual bile samples, as previously described.28 The concentration was subsequently determined as in Böttcher et al.29 Secretion of phospholipids was calculated for each individual by multiplying the concentration of phospholipids by the volume of bile secreted per minute and per 100 g of body weight. Assay of Biliary Bile Acid Concentration and Secretion 2 μL of bile were hydrolyzed with 0.5 mL of 5 M NaOH in 90% EtOH at 67°C for 90 min. Then, 0.5 mL of H2O and 3 mL of cyklohexane were added and samples were centrifuged at 2,000 rpm for 10 min before upper phase was removed. This was repeated once before acidification of samples with 200 μL of 6 M HCl. Ether was added to extract bile acids (BAs) and H2O was added to collected ether extracts, which were centrifuged at 2,000 rpm for 10 min before the upper phase was collected and evaporated under nitrogen at 60°C. Methylation was carried out at room temperature for 10 min by adding 400 μL of toluene, 100 μL of MeOH, and 25 μL of trimethylsilyldiazomethane, and samples were then dried under nitrogen at 60°C. Samples were silylated with pyridine/hexametyldisilazane/chlorotrimetylsilane (3:2:1, v/v/v) at 60°C for 30 min and thereafter dried under nitrogen, redissolved in hexane, and analyzed using GC/MS. D4-labeled BAs were used as internal standards. BA secretion was calculated for each individual by multiplying the sum of concentrations of specific BAs by the volume of bile secreted per minute and per 100 g of body weight.

or 30 min and thereafter dried under nitrogen, redissolved in hexane, and analyzed using GC/MS. D4-labeled BAs were used as internal standards. BA secretion was calculated for each individual by multiplying the sum of concentrations of specific BAs by the volume of bile secreted per minute and per 100 g of body weight. Statistical Analyses Data show means ± standard error of the mean (SEM). The significance of differences between groups was tested by 1-way ANOVA, followed by post-hoc comparisons according to Tukey's test, using GraphPad Prism software (GraphPad Software Inc., San Diego, CA). Results Effects of T3 on Hepatic Gene Expression in Abcg5+/+ and Abcg5−/− Mice Hepatic ABCG5 and ABCG8 gene expressions were both increased 1.5-fold in T3-treated Abcg5+/+ mice (Fig. .1). ABCG8 gene expression was unaltered in Abcg5−/− control and in T3-treated Abcg5−/− mice. Hepatic LXRa gene expression was unaltered in Abcg5−/− mice, while reduced in T3-treated Abcg5+/+ and Abcg5−/− mice (by 23% and 10%, respectively), as compared to respective controls. Compared to untreated Abcg5+/+ mice, CYP7A1, hydroxymethylglutaryl coenzyme A reductase (HMG CoA red), and LDLr gene expressions were increased 4.6-, 3.7-, and 1.6-fold, respectively, in T3-treated Abcg5+/+ mice, whereas they were unaltered in untreated Abcg5−/− mice. In T3-treated Abcg5−/− mice, gene expressions of HMG CoA red and LDLr were unaltered, whereas CYP7A1 gene expression was 2.9-fold increased.

HMG CoA red), and LDLr gene expressions were increased 4.6-, 3.7-, and 1.6-fold, respectively, in T3-treated Abcg5+/+ mice, whereas they were unaltered in untreated Abcg5−/− mice. In T3-treated Abcg5−/− mice, gene expressions of HMG CoA red and LDLr were unaltered, whereas CYP7A1 gene expression was 2.9-fold increased. Figure 1 Effects of T3 treatment on hepatic gene expression of ABCG5 (A), ABCG8 (B), LXRa (C), CYP7A1 (D), HMG CoA red (E), and LDLr (F) in Abcg5−/− mice and in their WT counterparts (Abcg5+/+). Number of animals (n) per group: Abcg5+/+ n = 6; Abcg5+/+ T3 n = 5; Abcg5−/− n = 5; and Abcg5−/− T3 n = 6. Data are presented as mean ± SEM. ***P < 0.001; **P < 0.01; *P < 0.05. Biliary Cholesterol, Phospholipids, BAs, and Bile Flow in T3-Treated Abcg5+/+ and Abcg5−/− Mice In Abcg5−/− mice, biliary cholesterol and phospholipid concentrations were reduced by 75% and 46%, respectively (Table 1). In T3-treated Abcg5+/+ mice, biliary cholesterol and phospholipids were increased 1.8- and 1.3-fold, respectively. Compared to untreated Abcg5−/− mice, cholesterol and phospholipids were unaltered in T3-treated Abcg5−/− mice. Table 1 Effects of T3 Treatment on Body Weight, Biliary Lipids, and Bile Flow in Abcg5−/− Mice and in Their WT Counterparts (Abcg5+/+)

Biliary Cholesterol, Phospholipids, BAs, and Bile Flow in T3-Treated Abcg5+/+ and Abcg5−/− Mice In Abcg5−/− mice, biliary cholesterol and phospholipid concentrations were reduced by 75% and 46%, respectively (Table 1). In T3-treated Abcg5+/+ mice, biliary cholesterol and phospholipids were increased 1.8- and 1.3-fold, respectively. Compared to untreated Abcg5−/− mice, cholesterol and phospholipids were unaltered in T3-treated Abcg5−/− mice. Table 1 Effects of T3 Treatment on Body Weight, Biliary Lipids, and Bile Flow in Abcg5−/− Mice and in Their WT Counterparts (Abcg5+/+) No. of Animals Abcg5+/+ (n = 6) Abcg5++ T3 (n = 5) Abcg5−/− (n = 5) Abcg5−/− T3 (n = 6) Body weight, g 28 ± 1 31 ± 1 28 ± 1 32 ± 1 Cholesterol, nmol/mL 180 ± 10 320 ± 10* 50 ± 5*§ 70 ± 10*§ Phospholipids, nmol/mL 6,060 ± 320 7,890 ± 420‡ 3,300 ± 180*§ 4,460 ± 400‡§ BAs, nmol/mL 48,000 ± 9,770 44,800 ± 8,530 22,100 ± 3,290 18,700 ± 1,110‡ Ratio cholesterol/phospholipid 0.03 ± 0.003 0.04 ± 0.002† 0.01 ± 0.000*§ 0.02 ± 0.002*§ Ratio cholesterol/BA 0.004 ± 0.001 0.008 ± 0.001† 0.002 ± 0.000§ 0.004 ± 0.001‖ Ratio phospholipid/BA 0.14 ± 0.02 0.20 ± 0.03 0.16 ± 0.03 0.24 ± 0.02‡ Bile flow, μL/min 2.0 ± 0.1 3.9 ± 0.3* 2.1 ± 0.2§ 3.7 ± 0.3*¶ Data are presented as mean ± SEM. * P < 0.001, † P < 0.01, and ‡ P<0.05 versus Abcg5+/+. § P < 0.001 and ‖ P < 0.01 versus Abcg5+/+ T3. ¶ P < 0.001 versus Abcg5−/−.

No. of Animals Abcg5+/+ (n = 6) Abcg5++ T3 (n = 5) Abcg5−/− (n = 5) Abcg5−/− T3 (n = 6) Body weight, g 28 ± 1 31 ± 1 28 ± 1 32 ± 1 Cholesterol, nmol/mL 180 ± 10 320 ± 10* 50 ± 5*§ 70 ± 10*§ Phospholipids, nmol/mL 6,060 ± 320 7,890 ± 420‡ 3,300 ± 180*§ 4,460 ± 400‡§ BAs, nmol/mL 48,000 ± 9,770 44,800 ± 8,530 22,100 ± 3,290 18,700 ± 1,110‡ Ratio cholesterol/phospholipid 0.03 ± 0.003 0.04 ± 0.002† 0.01 ± 0.000*§ 0.02 ± 0.002*§ Ratio cholesterol/BA 0.004 ± 0.001 0.008 ± 0.001† 0.002 ± 0.000§ 0.004 ± 0.001‖ Ratio phospholipid/BA 0.14 ± 0.02 0.20 ± 0.03 0.16 ± 0.03 0.24 ± 0.02‡ Bile flow, μL/min 2.0 ± 0.1 3.9 ± 0.3* 2.1 ± 0.2§ 3.7 ± 0.3*¶ Data are presented as mean ± SEM. * P < 0.001, † P < 0.01, and ‡ P<0.05 versus Abcg5+/+. § P < 0.001 and ‖ P < 0.01 versus Abcg5+/+ T3. ¶ P < 0.001 versus Abcg5−/−. The concentration of total BAs was unaltered in Abcg5−/− mice. T3 treatment of Abcg5+/+ and Abcg5−/− mice did not significantly change biliary BA concentration compared to respective controls. Both the C/PL ratio and C/BA ratio were increased (1.4- and 1.9- fold, respectively) in T3-treated Abcg5+/+ mice, whereas the PL/BA ratio was unaltered. In Abcg5−/− mice, the C/PL ratio was decreased by 40%, and the C/BA and PL/BA ratios were unaltered. T3 treatment of Abcg5−/− mice did not alter the ratios. Under basal conditions, bile flow was the same in Abcg5+/+ and Abcg5−/− mice. T3 treatment increased bile flow to similar extents in Abcg5+/+ (1.9-fold) and in Abcg5−/− (1.8-fold) mice.

C/PL ratio was decreased by 40%, and the C/BA and PL/BA ratios were unaltered. T3 treatment of Abcg5−/− mice did not alter the ratios. Under basal conditions, bile flow was the same in Abcg5+/+ and Abcg5−/− mice. T3 treatment increased bile flow to similar extents in Abcg5+/+ (1.9-fold) and in Abcg5−/− (1.8-fold) mice. Effects of T3 on Biliary Composition of BAs in Abcg5+/+ and Abcg5−/− mice T3 treatment decreased the biliary proportion of deoxycholic acid (DCA) in Abcg5+/+ and Abcg5−/− mice by 56% and 55%, respectively (Table 2). The proportion of cholic acid (CA) tended to be reduced in T3-treated animals. Chenodeoxycholic acid (CDCA) was increased 1.8-fold by T3 treatment in Abcg5−/− mice, and there was a trend to an increased proportion of CDCA in the Abcg5+/+ mice (P = 0.05). Alpha-muricholic acid (α-MCA) was increased by T3 treatment in both Abcg5+/+ and Abcg5−/− mice 2.1- and 3.4-fold, respectively. Biliary proportions of β-muricholic acid (β-MCA), ursodeoxycholic acid (UDCA), and litocholic acid (LCA) were unaltered. Table 2 Effects of T3 Treatment on Biliary BA Composition in Abcg5−/− Mice and in Their WT Counterparts (Abcg5+/+)

Effects of T3 on Biliary Composition of BAs in Abcg5+/+ and Abcg5−/− mice T3 treatment decreased the biliary proportion of deoxycholic acid (DCA) in Abcg5+/+ and Abcg5−/− mice by 56% and 55%, respectively (Table 2). The proportion of cholic acid (CA) tended to be reduced in T3-treated animals. Chenodeoxycholic acid (CDCA) was increased 1.8-fold by T3 treatment in Abcg5−/− mice, and there was a trend to an increased proportion of CDCA in the Abcg5+/+ mice (P = 0.05). Alpha-muricholic acid (α-MCA) was increased by T3 treatment in both Abcg5+/+ and Abcg5−/− mice 2.1- and 3.4-fold, respectively. Biliary proportions of β-muricholic acid (β-MCA), ursodeoxycholic acid (UDCA), and litocholic acid (LCA) were unaltered. Table 2 Effects of T3 Treatment on Biliary BA Composition in Abcg5−/− Mice and in Their WT Counterparts (Abcg5+/+) No. of Animals Abcg5+/+ (n = 6) Abcg5++ T3 (n = 5) Abcg5−/− (n = 5) Abcg5−/− T3 (n = 6) CA nmol/mL 23,500 ± 3,700 20,100 ± 3,330 11,500 ± 2,150‡ 7,540 ± 840†¶ % of total 51 ± 3 46 ± 2 51 ± 3 40 ± 3 CDCA nmol/mL 170 ± 10 260 ± 30‡ 100 ± 10§ 160 ± 20‖ % of total 0.4 ± 0.05 0.7 ± 0.09 0.5 ± 0.06 0.9 ± 0.08*** α-MCA nmol/mL 1,990 ± 410 4,220 ± 1,210 430 ± 70‖ 1,190 ± 70¶ % of total 4.0 ± 0.4 9.0 ± 1.1* 2.0 ± 0.1§ 6.0 ± 0.4‡¶# β-MCA nmol/mL 21,300 ± 5,720 19,700 ± 4,400 9,420 ± 1,200 9,490 ± 800 % of total 42 ± 3 44 ± 3 44 ± 3 51 ± 3 DCA nmol/mL 510 ± 80 200 ± 10† 470 ± 70¶ 180 ± 30††† % of total 1.0 ± 0.1 0.5 ± 0.1‡ 2.0 ± 0.1*§ 1.0 ± 0.2# UDCA nmol/mL 570 ± 100 350 ± 20 200 ± 30† 170 ± 20* % of total 1.0 ± 0.2 1.0 ± 0.1 1.0 ± 0.1 1.0 ± 0.1 LCA nmol/mL 20 ± 0 20 ± 1 20 ± 0 20 ± 1 % of total 0.02 ± 0.02 0.02 ± 0.02 0.1 ± 0.00†‖ 0.1 ± 0.00*‖ Data are presented as mean ± SEM.

± 10† 470 ± 70¶ 180 ± 30††† % of total 1.0 ± 0.1 0.5 ± 0.1‡ 2.0 ± 0.1*§ 1.0 ± 0.2# UDCA nmol/mL 570 ± 100 350 ± 20 200 ± 30† 170 ± 20* % of total 1.0 ± 0.2 1.0 ± 0.1 1.0 ± 0.1 1.0 ± 0.1 LCA nmol/mL 20 ± 0 20 ± 1 20 ± 0 20 ± 1 % of total 0.02 ± 0.02 0.02 ± 0.02 0.1 ± 0.00†‖ 0.1 ± 0.00*‖ Data are presented as mean ± SEM. * P < 0.001, † P < 0.01, and ‡ P < 0.05 versus Abcg5+/+. § P < 0.001, ‖ P < 0.01, and ¶ P < 0.05 versus Abcg5+/+ T3. # P < 0.001, ** P < 0.01, and †† P < 0.05 versus Abcg5−/−. Importance of a Functional ABCG5/ABCG8 Complex for the Stimulation of Biliary Cholesterol Secretion by T3 Biliary cholesterol secretion was increased 3.1-fold in T3-treated Abcg5+/+ mice (Fig. .2). Basal secretion of biliary cholesterol in Abcg5−/− mice was only 28% of that observed in untreated Abcg5+/+ mice. In T3-treated Abcg5−/− mice, biliary cholesterol secretion was unaltered compared to Abcg5−/− mice, and did not differ from that of untreated Abcg5+/+ mice. Biliary cholesterol secretion in T3-treated Abcg5−/− mice was 79% lower than in T3-treated Abcg5+/+ mice. Biliary phospholipid secretion was unaltered in Abcg5−/− mice. T3 treatment increased phospholipid secretion 2.3-fold in Abcg5+/+ mice and 2.1-fold in Abcg5−/− mice, compared to respective controls. Total BA secretion was unaltered in Abcg5−/− mice. T3 treatment of Abcg5+/+ and Abcg5−/− mice tended to increase BA secretion, but the differences did not reach statistical significance.

treatment increased phospholipid secretion 2.3-fold in Abcg5+/+ mice and 2.1-fold in Abcg5−/− mice, compared to respective controls. Total BA secretion was unaltered in Abcg5−/− mice. T3 treatment of Abcg5+/+ and Abcg5−/− mice tended to increase BA secretion, but the differences did not reach statistical significance. Figure 2 Effects of T3 treatment on secretion of biliary cholesterol (A), phospholipids (B), and BAs (C) in Abcg5−/− mice and in their WT counterparts (Abcg5+/+). Number of animals (n) per group: Abcg5+/+ n = 6; Abcg5+/+ T3 n = 5; Abcg5−/− n = 5; Abcg5−/− T3 n = 6. Data are presented as mean ± SEM. ***P < 0.001; **P < 0.01; *P < 0.05. Effects of T3 on Hepatic Gene Expression in Lxra+/+ and Lxra−/− Mice LXRa gene expression was unaltered in T3-treated Lxra+/+ mice, whereas ABCG5 and ABCG8 gene expression levels were both increased 2.1- and 1.5-fold, respectively (Fig. .3). Gene expressions of ABCG5 and ABCG8 were unaltered in Lxra−/− mice, whereas they were increased in T3-treated Lxra−/− mice (1.8- and 1.7-fold, respectively), as compared to Lxra−/− mice. Gene expressions of ABCG5/G8 in proximal small intestine were unaltered (data not shown). Hepatic CYP7A1, HMG CoA red, and LDL receptor (LDLr) gene expressions were unaltered in Lxra−/− mice and in T3-treated Lxra+/+ mice. In T3-treated Lxra−/− mice, HMG CoA red and LDLr gene expressions were unaltered, whereas CYP7A1 gene expression was increased 4.1-fold, compared with untreated Lxra−/− mice.

not shown). Hepatic CYP7A1, HMG CoA red, and LDL receptor (LDLr) gene expressions were unaltered in Lxra−/− mice and in T3-treated Lxra+/+ mice. In T3-treated Lxra−/− mice, HMG CoA red and LDLr gene expressions were unaltered, whereas CYP7A1 gene expression was increased 4.1-fold, compared with untreated Lxra−/− mice. Figure 3 Effects of T3 treatment on hepatic gene expression of ABCG5 (A), ABCG8 (B), LXRa (C), CYP7A1 (D), HMG CoA red (E), and LDLr (F) in in Lxra−/− mice and in their WT counterparts (Lxra+/+). Number of animals per group: n = 7. Data are presented as mean ± SEM. ***P < 0.001; **P < 0.01; *P < 0.05. Effects of T3 Treatment on Biliary Cholesterol, Phospholipids, BAs, and Bile Flow in Lxra+/+ and Lxra−/− Mice Biliary cholesterol and phospholipids were unaltered in Lxra−/− mice, and T3 treatment of Lxra+/+ or Lxra−/− mice did not alter biliary cholesterol or phospholipid concentrations, as compared to respective controls (Table 3). Further, the total concentration of biliary BAs did not differ between groups. C/PL and C/BA ratios were unaltered in T3-treated Lxra+/+ mice, whereas the PL/BA ratio was 1.4-fold increased. None of the ratios were altered in the Lxra−/− mice. T3 treatment of Lxra−/− mice increased both the C/BA and PL/BA ratio (1.6- and 1.4-fold, respectively), whereas the C/PL ratio was unaltered, as compared to untreated Lxra−/− mice. Bile flow was unaltered in Lxra−/− mice. However, T3 treatment increased bile flow by 2.4-fold in Lxra+/+ and 2.1-fold in Lxra−/− mice.

ce. T3 treatment of Lxra−/− mice increased both the C/BA and PL/BA ratio (1.6- and 1.4-fold, respectively), whereas the C/PL ratio was unaltered, as compared to untreated Lxra−/− mice. Bile flow was unaltered in Lxra−/− mice. However, T3 treatment increased bile flow by 2.4-fold in Lxra+/+ and 2.1-fold in Lxra−/− mice. Table 3 Effects of T3 Treatment on Body Weight, Biliary Lipids, and Bile Flow in Lxra−/− Mice and in Their WT Counterparts (Lxra+/+) No. of Animals Lxra+/+ (n = 7) Lxra+/+ T3 (n = 7) Lxra−/− (n = 7) Lxra−/− T3 (n = 7) Body weight, g 34 ± 1 40 ± 1† 36 ± 1 38 ± 2 Cholesterol, nmol/mL 190 ± 30 320 ± 60 340 ± 80 470 ± 50‡ Phospholipids, nmol/mL 5,870 ± 610 6,710 ± 930 6,670 ± 710 7,610 ± 320 BAs, nmol/mL 49,200 ± 5,390 43,900 ± 9,560 53,400 ± 7,330 41,900 ± 3,480 Ratio cholesterol/phospholipid 0.03 ± 0.00 0.05 ± 0.00 0.05 ± 0.01 0.06 ± 0.05‡ Ratio cholesterol/BA 0.004 ± 0.001 0.008 ± 0.001 0.007 ± 0.001 0.011 ± 0.001*¶ Ratio phospholipid/BA 0.12 ± 0.01 0.17 ± 0.01‡ 0.13 ± 0.01 0.19 ± 0.01†¶ Bile flow, μL/min 1.7 ± 0.1 4.1 ± 0.5* 1.7 ± 0.2§ 3.5 ± 0.2*‖ Data are presented as mean ± SEM. * P < 0.001, † P < 0.01, and ‡ P < 0.05 versus Lxra+/+. § P < 0.001 versus Lxra+/+ T3. ‖ P < 0.001 and ¶ P < 0.05 versus Lxra−/−.

No. of Animals Lxra+/+ (n = 7) Lxra+/+ T3 (n = 7) Lxra−/− (n = 7) Lxra−/− T3 (n = 7) Body weight, g 34 ± 1 40 ± 1† 36 ± 1 38 ± 2 Cholesterol, nmol/mL 190 ± 30 320 ± 60 340 ± 80 470 ± 50‡ Phospholipids, nmol/mL 5,870 ± 610 6,710 ± 930 6,670 ± 710 7,610 ± 320 BAs, nmol/mL 49,200 ± 5,390 43,900 ± 9,560 53,400 ± 7,330 41,900 ± 3,480 Ratio cholesterol/phospholipid 0.03 ± 0.00 0.05 ± 0.00 0.05 ± 0.01 0.06 ± 0.05‡ Ratio cholesterol/BA 0.004 ± 0.001 0.008 ± 0.001 0.007 ± 0.001 0.011 ± 0.001*¶ Ratio phospholipid/BA 0.12 ± 0.01 0.17 ± 0.01‡ 0.13 ± 0.01 0.19 ± 0.01†¶ Bile flow, μL/min 1.7 ± 0.1 4.1 ± 0.5* 1.7 ± 0.2§ 3.5 ± 0.2*‖ Data are presented as mean ± SEM. * P < 0.001, † P < 0.01, and ‡ P < 0.05 versus Lxra+/+. § P < 0.001 versus Lxra+/+ T3. ‖ P < 0.001 and ¶ P < 0.05 versus Lxra−/−. Effects of T3 on Biliary Composition of BAs in Lxra+/+ and Lxra−/− Mice CDCA was increased by T3 treatment in Lxra+/+ and Lxra−/− mice (2.0- and 1.7-fold, respectively) (Table 4). α-MCA was increased by T3 treatment in Lxra+/+ and Lxra−/− mice (2.2- and 1.8-fold, respectively). There was a tendency toward decreased biliary proportions of DCA and CA in T3-treated Lxra+/+ and Lxra−/− mice, whereas the biliary proportions of β-MCA, UDCA, and LCA were unaltered. Table 4 Effects of T3 Treatment on Biliary BA Composition in Lxra−/− Mice and in Their WT Counterparts (Lxra+/+)

Effects of T3 on Biliary Composition of BAs in Lxra+/+ and Lxra−/− Mice CDCA was increased by T3 treatment in Lxra+/+ and Lxra−/− mice (2.0- and 1.7-fold, respectively) (Table 4). α-MCA was increased by T3 treatment in Lxra+/+ and Lxra−/− mice (2.2- and 1.8-fold, respectively). There was a tendency toward decreased biliary proportions of DCA and CA in T3-treated Lxra+/+ and Lxra−/− mice, whereas the biliary proportions of β-MCA, UDCA, and LCA were unaltered. Table 4 Effects of T3 Treatment on Biliary BA Composition in Lxra−/− Mice and in Their WT Counterparts (Lxra+/+) No. of Animals Lxra+/+ (n =7) Lxra+/+ T3 (n =7) Lxra−/− (n =7) Lxra−/− T3 (n =7) CA nmol/mL 21,900 ± 3,320 18,900 ± 4,170 28,300 ± 4,700 19,300 ± 2,200 % of total 44 ± 4 43 ± 3 51 ± 3 45 ± 2 CDCA nmol/mL 670 ± 50 1,120 ± 180‡ 730 ± 80 1.050 ± 80 % of total 1.0 ± 0.1 3.0 ± 0.4† 2.0 ± 0.2‖ 3.0 ± 0.3‡†† α-MCA nmol/mL 2,860 ± 390 5,810 ± 1,560 3,970 ± 790 5,200 ± 370 % of total 6.0 ± 0.5 13.0 ± 0.9* 7.0 ± 0.6§ 13.0 ± 0.9*# β-MCA nmol/mL 21,400 ± 2,610 16,300 ± 3,610 18,100 ± 2,930 14,900 ± 1,210 % of total 43 ± 3 37 ± 2 34 ± 4 36 ± 2 DCA nmol/mL 1,320 ± 130 710 ± 110‡ 1,370 ± 200¶ 630 ± 70†** % of total 3.0 ± 0.4 2.0 ± 0.2 3.0 ± 0.7 2.0 ± 0.2 UDCA nmol/mL 1,000 ± 90 1,060 ± 220 910 ± 110 830 ± 30 % of total 2.0 ± 0.1 2.0 ± 0.1 2.0 ± 0.1 2.0 ± 0.1 LCA nmol/mL 60 ± 2 60 ± 3 60 ± 2 50 ± 1 % of total 0.1 ± 0.01 0.2 ± 0.02 0.1 ± 0.03 0.1 ± 0.02 Data are presented as mean ± SEM. * P < 0.001, † P < 0.01, and ‡ P < 0.05 versus Lxra+/+. § P < 0.001, ‖ P < 0.01, and ¶ P < 0.05 versus Lxra+/+ T3. # P < 0.001, ** P < 0.01, and

No. of Animals Lxra+/+ (n =7) Lxra+/+ T3 (n =7) Lxra−/− (n =7) Lxra−/− T3 (n =7) CA nmol/mL 21,900 ± 3,320 18,900 ± 4,170 28,300 ± 4,700 19,300 ± 2,200 % of total 44 ± 4 43 ± 3 51 ± 3 45 ± 2 CDCA nmol/mL 670 ± 50 1,120 ± 180‡ 730 ± 80 1.050 ± 80 % of total 1.0 ± 0.1 3.0 ± 0.4† 2.0 ± 0.2‖ 3.0 ± 0.3‡†† α-MCA nmol/mL 2,860 ± 390 5,810 ± 1,560 3,970 ± 790 5,200 ± 370 % of total 6.0 ± 0.5 13.0 ± 0.9* 7.0 ± 0.6§ 13.0 ± 0.9*# β-MCA nmol/mL 21,400 ± 2,610 16,300 ± 3,610 18,100 ± 2,930 14,900 ± 1,210 % of total 43 ± 3 37 ± 2 34 ± 4 36 ± 2 DCA nmol/mL 1,320 ± 130 710 ± 110‡ 1,370 ± 200¶ 630 ± 70†** % of total 3.0 ± 0.4 2.0 ± 0.2 3.0 ± 0.7 2.0 ± 0.2 UDCA nmol/mL 1,000 ± 90 1,060 ± 220 910 ± 110 830 ± 30 % of total 2.0 ± 0.1 2.0 ± 0.1 2.0 ± 0.1 2.0 ± 0.1 LCA nmol/mL 60 ± 2 60 ± 3 60 ± 2 50 ± 1 % of total 0.1 ± 0.01 0.2 ± 0.02 0.1 ± 0.03 0.1 ± 0.02 Data are presented as mean ± SEM. * P < 0.001, † P < 0.01, and ‡ P < 0.05 versus Lxra+/+. § P < 0.001, ‖ P < 0.01, and ¶ P < 0.05 versus Lxra+/+ T3. # P < 0.001, ** P < 0.01, and †† P < 0.05 versus Lxra−/−.

No. of Animals Lxra+/+ (n =7) Lxra+/+ T3 (n =7) Lxra−/− (n =7) Lxra−/− T3 (n =7) CA nmol/mL 21,900 ± 3,320 18,900 ± 4,170 28,300 ± 4,700 19,300 ± 2,200 % of total 44 ± 4 43 ± 3 51 ± 3 45 ± 2 CDCA nmol/mL 670 ± 50 1,120 ± 180‡ 730 ± 80 1.050 ± 80 % of total 1.0 ± 0.1 3.0 ± 0.4† 2.0 ± 0.2‖ 3.0 ± 0.3‡†† α-MCA nmol/mL 2,860 ± 390 5,810 ± 1,560 3,970 ± 790 5,200 ± 370 % of total 6.0 ± 0.5 13.0 ± 0.9* 7.0 ± 0.6§ 13.0 ± 0.9*# β-MCA nmol/mL 21,400 ± 2,610 16,300 ± 3,610 18,100 ± 2,930 14,900 ± 1,210 % of total 43 ± 3 37 ± 2 34 ± 4 36 ± 2 DCA nmol/mL 1,320 ± 130 710 ± 110‡ 1,370 ± 200¶ 630 ± 70†** % of total 3.0 ± 0.4 2.0 ± 0.2 3.0 ± 0.7 2.0 ± 0.2 UDCA nmol/mL 1,000 ± 90 1,060 ± 220 910 ± 110 830 ± 30 % of total 2.0 ± 0.1 2.0 ± 0.1 2.0 ± 0.1 2.0 ± 0.1 LCA nmol/mL 60 ± 2 60 ± 3 60 ± 2 50 ± 1 % of total 0.1 ± 0.01 0.2 ± 0.02 0.1 ± 0.03 0.1 ± 0.02 Data are presented as mean ± SEM. * P < 0.001, † P < 0.01, and ‡ P < 0.05 versus Lxra+/+. § P < 0.001, ‖ P < 0.01, and ¶ P < 0.05 versus Lxra+/+ T3. # P < 0.001, ** P < 0.01, and †† P < 0.05 versus Lxra−/−. Biliary Cholesterol Secretion Is Induced by T3 independent of Lxra Biliary cholesterol secretion was similar in untreated Lxra+/+ and Lxra−/− mice (Fig. .4). In response to T3 treatment, it increased 3.5-fold in Lxra+/+ and 2.6-fold in Lxra−/− mice, to similar levels. Phospholipid secretion was unchanged in Lxra−/− mice. In T3-treated Lxra+/+ mice and Lxra−/− mice, phospholipid secretion increased 2.3- and 2.2-fold, compared to respective controls. Secretion of total BAs was unchanged in the groups, although there was a trend to an increased secretion in T3-treated mice.

lar levels. Phospholipid secretion was unchanged in Lxra−/− mice. In T3-treated Lxra+/+ mice and Lxra−/− mice, phospholipid secretion increased 2.3- and 2.2-fold, compared to respective controls. Secretion of total BAs was unchanged in the groups, although there was a trend to an increased secretion in T3-treated mice. Figure 4 Effects of T3 treatment on secretion of biliary cholesterol (A), phospholipids (B), and BAs (C) in Lxra−/− mice and in their WT counterparts (Lxra+/+). Number of animals per group: n = 7. Data are presented as mean ± SEM. **P < 0.01; *P < 0.05.

lar levels. Phospholipid secretion was unchanged in Lxra−/− mice. In T3-treated Lxra+/+ mice and Lxra−/− mice, phospholipid secretion increased 2.3- and 2.2-fold, compared to respective controls. Secretion of total BAs was unchanged in the groups, although there was a trend to an increased secretion in T3-treated mice. Figure 4 Effects of T3 treatment on secretion of biliary cholesterol (A), phospholipids (B), and BAs (C) in Lxra−/− mice and in their WT counterparts (Lxra+/+). Number of animals per group: n = 7. Data are presented as mean ± SEM. **P < 0.01; *P < 0.05. Discussion TH exerts a number of important regulatory effects on cholesterol, lipid, and lipoprotein metabolism.4 These include stimulation of hepatic lipase activity, induction of hepatic LDL receptors, promotion of cholesterol breakdown to BAs, and cholesterol excretion into bile. Furthermore, there is evidence that TH may promote reverse cholesterol transport through stimulation of high-density lipoprotein (HDL) clearance.4, 30 Many of the positive actions of TH in lipid metabolism are constrained to the liver, and the recent demonstration of the possibility to achieve pronounced lipid-lowering effects in humans by selectively stimulating TRb in the liver has revitalized the interest for understanding the molecular effects of TH.4–6 We here explored by which mechanisms TH exerts its powerful effects on biliary cholesterol secretion by specifically analyzing the role of the ABCG5/G8 half-transporter complex in mice. This complex has been shown to be of major importance for sterol excretion into bile, but there are also data indicating that ABCG5/G8-independent mechanisms may promote cholesterol secretion. First, biliary cholesterol secretion/concentration is not completely abolished in single12, 13 and double11, 14–16 ABCG5/G8 knockout models. Second, hepatic overexpression of scavenger receptor class B, member 1 (SR-BI) in Abcg5−/− mice can restore their initially decreased biliary cholesterol secretion to WT levels.23 And third, since transintestinal cholesterol efflux occurs in Abcg5−/− 22 and Abcg8−/−31 mice via additional pathways not yet defined, such mechanisms may operate also in the liver.

er receptor class B, member 1 (SR-BI) in Abcg5−/− mice can restore their initially decreased biliary cholesterol secretion to WT levels.23 And third, since transintestinal cholesterol efflux occurs in Abcg5−/− 22 and Abcg8−/−31 mice via additional pathways not yet defined, such mechanisms may operate also in the liver. To determine to which extent the strong stimulation of biliary cholesterol secretion induced by TH is mediated by the ABCG5/G8 complex, we treated Abcg5−/− and WT mice of the same genetic background (Abcg5+/+) with T3. In line with previous results,18 TH treatment increased hepatic gene expression of ABCG5/G8 in Abcg5+/+ mice, but failed to increase ABCG8 gene expression in Abcg5−/− mice. This lack of response may be the result of a disruption in a regulatory region of Abcg8 caused in the procedure of disrupting Abcg5. The ABCG5 and ABCG8 genes are orientated in a head-to-head manner in the genome within 400 base pairs of each other. This implies that putative binding sites for transcription factors for one gene may be positioned within the opposite gene. Therefore, the insertion of the LacZ/Neo cassette used to disrupt the ABCG5 gene has been shown to also indirectly influence the expression of the other gene (ABCG8).13

0 base pairs of each other. This implies that putative binding sites for transcription factors for one gene may be positioned within the opposite gene. Therefore, the insertion of the LacZ/Neo cassette used to disrupt the ABCG5 gene has been shown to also indirectly influence the expression of the other gene (ABCG8).13 Biliary cholesterol secretion was strongly reduced in Abcg5−/− mice, to only 28% of that in Abcg5+/+ mice. T3 treatment increased biliary cholesterol secretion 3.1-fold in Abcg5+/+ mice, whereas in Abcg5−/− mice, this response was blunted. These results demonstrate that stimulation of biliary secretion of cholesterol by T3 treatment of mice is largely dependent on an intact ABCG5/G8 complex. However, T3 treatment restored the low biliary secretion of cholesterol in Abcg5−/− mice up to the basal rate observed in Abcg5+/+ mice. This suggests that, although a functional ABCG5/G8 complex is required for the major stimulation of biliary cholesterol secretion by T3, there is also an additional, ABCG5/G8-independent, mechanism.

atment restored the low biliary secretion of cholesterol in Abcg5−/− mice up to the basal rate observed in Abcg5+/+ mice. This suggests that, although a functional ABCG5/G8 complex is required for the major stimulation of biliary cholesterol secretion by T3, there is also an additional, ABCG5/G8-independent, mechanism. The increased secretion in Abcg5−/− mice occurred simultaneously with a T3-induced doubled flow rate of bile, regardless of the genetic background of the animals. Thus, one explanation for the non-ABCG5/G8 driven cholesterol secretion could be that it reflects the combined results of simple diffusion of cholesterol and the biliary capacity to bind cholesterol. The T3-induced flow rate of bile would then modulate the total output of diffusible lipophilic compounds such as cholesterol and phospholipids, as observed, and may in turn be related to circulatory effects exerted by the hormone. In addition to the markedly (3-fold) increased secretion of cholesterol, gene expression of the rate-limiting enzyme in BA synthesis, cholesterol 7α-hydroxylase (cytochrome P450 [CYP]7A1), was 4.6-fold increased by T3 treatment in Abcg5+/+ mice. These changes were accompanied with increased gene expression levels of the LDLr and HMG CoA red, the rate-limiting enzyme in cholesterol synthesis (2- and 4-fold, respectively), suggesting that the increased hepatic turnover of cholesterol is balanced by an increased de novo synthesis of cholesterol and by an increased uptake of cholesterol from the circulation. Consistent with previous results,13 the concentration of total BAs in bile was unchanged in Abcg5−/− mice. In spite of an increased bile flow rate, and in contrast to the effect on the secretion of cholesterol, the secretion of total BAs was unaltered by T3 treatment. T3 treatment decreased the proportion of DCA and CA, whereas the proportions of CDCA and α-MCA increased. These results are in line with the concept that TH suppresses the BA synthetic enzyme, sterol 12-α-hydroxylase (CYP8B1), as has previously been shown.32–34

the secretion of total BAs was unaltered by T3 treatment. T3 treatment decreased the proportion of DCA and CA, whereas the proportions of CDCA and α-MCA increased. These results are in line with the concept that TH suppresses the BA synthetic enzyme, sterol 12-α-hydroxylase (CYP8B1), as has previously been shown.32–34 Activation of LXR by selective agonists has similar effects on hepatic ABCG5/G8 gene expression levels and biliary cholesterol secretion as TH.15, 17 It has been reported that LXRa is positively regulated at the transcriptional level by TRb.24 We therefore investigated the role of LXRa in the TH-induced stimulation of biliary cholesterol secretion. For this purpose, Lxra−/− and Lxra+/+ mice with the same genetic background were treated with T3. Biliary cholesterol secretion rates did not differ between T3-treated Lxra+/+ and Lxra−/− mice. These results clearly indicate that the stimulation of biliary cholesterol secretion in response to T3 is independent of LXRa. Mean levels of CYP7A1, HMG CoA red, and LDLr gene expressions were higher in T3-treated Lxra+/+ mice, compared to the controls. However, as opposed to the response in T3-treated Abcg5+/+ mice, mean levels were not statistically significantly different. Because LXRa agonists have been shown to possess adverse side effects,35–37 the apparent absence of LXRa involvement in TH-induced responses on biliary cholesterol is promising from a therapeutic point of view, since available novel thyromimetics, such as eprotirome, should thus not be expected to present such side effects.

se LXRa agonists have been shown to possess adverse side effects,35–37 the apparent absence of LXRa involvement in TH-induced responses on biliary cholesterol is promising from a therapeutic point of view, since available novel thyromimetics, such as eprotirome, should thus not be expected to present such side effects. In conclusion, we have demonstrated that the ability of TH to stimulate the secretion of cholesterol into bile is largely mediated by the ABCG5/G8 complex, whereas LXRa does not seem to be of importance for this effect. The authors thank Rick Havinga for expert technical assistance. ABCG5ATP-binding cassette, subfamily G (WHITE), member 5 ABCG8ATP-binding cassette, subfamily G (WHITE), member 8 α-MCAalpha-muricholic acid ATPadenosine triphosphate BAbile acid β-MCAbeta-muricholic acid CAcholic acid CDCAchenodeoxycholic acid CYPcytochrome P450 DCAdeoxycholic acid Gapdhglyceraldehyde-3-phosphate dehydrogenase GCgas chromatography HDLhigh-density lipoprotein HMG CoA redhydroxymethylglutaryl coenzyme A reductase Hprthypoxanthine guanine phosphoribosyl transferase LCAlitocholic acid LDLlow-density lipoprotein LDLrLDL receptor LXRaliver x receptor alpha MSmass spectrometry T3triiodothyronine THthyroid hormone TRbthyroid hormone receptor beta UDCAursodeoxycholic acid WTwild type

Fibrosis of the liver is characterized by excessive extracellular matrix (ECM) deposition. One of the major cell types responsible for this is the hepatic stellate cell (HSC).1, 2 In response to injury, HSCs become activated into proliferative myofibroblasts, migrate into the surrounding parenchymal cells, and secrete tissue-damaging ECM, the major component of which is type 1 collagen (COL1). In addition, the ECM contains a complex mix of proteins that promote cell proliferation, migration, and differentiation. One ECM component with such roles is the matricellular glycophosphoprotein, osteopontin (OPN), also known as secreted phosphoprotein 1. OPN is detected in a wide range of tissues and body fluids, with physiological roles during development (e.g., in bone, bile duct formation, and during vascular remodeling), immune system regulation, and wound repair.3 However, it is also associated with pathological processes relating to cancer and inflammation.3, 4 The ability of OPN to mediate such diverse cellular functions is likely related to its extensive post-translational modifications and ability to signal through several integrin and CD44 variant receptors.3, 5

nd repair.3 However, it is also associated with pathological processes relating to cancer and inflammation.3, 4 The ability of OPN to mediate such diverse cellular functions is likely related to its extensive post-translational modifications and ability to signal through several integrin and CD44 variant receptors.3, 5 OPN contributes to wound scarring in skin6 and has been implicated in lung, kidney, and heart fibrosis.7-9 It has previously been detected in activated HSCs, where it is thought to mediate cell migration.10 More recently, OPN levels have been highlighted as a potential biomarker of liver disease, levels correlating with the severity of disease,11-13 and has been reported to promote the progression of fibrosis in nonalcoholic steatohepatitis.14 The latter study, and others,15 has demonstrated regulation of OPN expression by Hedgehog (Hh) signaling, mediated by the member of the glioblastoma (GLI) family of transcription factors, GLI1, binding to an upstream element of the OPN promoter.15 There are three GLI transcription factors, with different activator and repressor forms of GLI2 and GLI3 generated by alternative splicing of the parent transcripts.16 Previously, we have shown that the transcription factor, sex-determining region Y-box 9 (SOX9), becomes ectopically expressed in activated HSCs, where it is responsible for COL1 production.17 During development, SOX9 has diverse roles regulating the expression of a number of genes encoding ECM proteins.18 SOX9 has also been associated with fibrotic pathologies affecting the skin, kidney, and heart.18-23

9 (SOX9), becomes ectopically expressed in activated HSCs, where it is responsible for COL1 production.17 During development, SOX9 has diverse roles regulating the expression of a number of genes encoding ECM proteins.18 SOX9 has also been associated with fibrotic pathologies affecting the skin, kidney, and heart.18-23 In this present study, we show that OPN and SOX9 colocalize to biliary cells in the healthy liver and to the same regions of fibrotic tissue. Both are markedly increased during HSC activation, when it appears unlikely that GLI1 regulates OPN. Instead, we demonstrate that SOX9 lies downstream of GLI2 and is responsible for OPN expression. These data support a role for SOX9 during the progression of liver fibrosis as a regulator of key fibrotic ECM components, and suggest that the manipulation of SOX9 or its downstream targets may be a means of developing antifibrotic therapies. Furthermore, the identification of other ECM targets of SOX9 may have additional utility as biomarkers of fibrotic severity in liver disease similar to recent studies on OPN.11, 12 Materials and Methods Human Tissue and Serum Collection Human fetal material was collected under guidelines issued by the Polkinghorne Committee, as described previously.17, 24 Ethical approval was granted by the North West Regional Ethics Committee. Freshly isolated adult liver was purchased after resection (Invitrogen Ltd., Warrington, UK) and processed as previously described.17, 24

material was collected under guidelines issued by the Polkinghorne Committee, as described previously.17, 24 Ethical approval was granted by the North West Regional Ethics Committee. Freshly isolated adult liver was purchased after resection (Invitrogen Ltd., Warrington, UK) and processed as previously described.17, 24 Animal Models of Liver Fibrosis Liver fibrosis was induced by 5-week treatment of adult male Sprague-Dawley rats with CCl425 or in C57Bl/6 mice fed a methionine- and choline-deficient diet for 7 weeks.

material was collected under guidelines issued by the Polkinghorne Committee, as described previously.17, 24 Ethical approval was granted by the North West Regional Ethics Committee. Freshly isolated adult liver was purchased after resection (Invitrogen Ltd., Warrington, UK) and processed as previously described.17, 24 Animal Models of Liver Fibrosis Liver fibrosis was induced by 5-week treatment of adult male Sprague-Dawley rats with CCl425 or in C57Bl/6 mice fed a methionine- and choline-deficient diet for 7 weeks. Immortalized and Primary Cell Culture Primary rat hepatic stellate cells (rHSCs) were isolated as described previously.17, 25 Human LX2 cells were a gift from Prof. Scott Friedman (Mount Sinai School of Medicine, New York, NY).26 Primary human HSCs (hHSCs) were isolated after liver resection (see Supporting Materials and Methods) under ethical approval from the Nottingham Research Ethics Committee, activated in culture, and fixed for immunocytochemistry (ICC).17 All cells were cultured in monolayer at 5% CO2 and 37°C in Dulbecco's modified Eagle's medium plus L-glutamine, Na-pyruvate, and antibiotics supplemented with fetal bovine serum: 1% (low serum) or 10% (high serum) for LX2 cells, as indicated, or 16% for rHSCs and 10% for hHSCs.17 Gene silencing was carried out transiently using short interfering RNA (siRNA) (see Supporting Table 1) with HiPerFect (LX2 cells) or Nucleofection for HSCs (Amaxa Biosystems GmbH, Cologne, Germany), as described previously.17 To interrogate Hh signaling, LX2 cells and rHSCs were treated with 5 μM of 3-Keto-N-(aminoethyl-aminocaproyl-dihydrocinnamoyl)/cyclopamine (CYC) (Merck Chemicals Ltd., Nottingham, UK) or 100 and 50 nM of smoothened agonist (SAG; Merck Chemicals Ltd.) for LX2 cells and HSCs, respectively. SAG treatments were performed in serum-free conditions. Overexpression experiments were carried out in LX2 cells. Plasmid transient transfections were achieved using Transfast reagent (Promega, Madison, WI), as described previously,17 in the presence or absence of CYC (described above). Briefly, 0.5 μg of expression plasmids (see Supporting Table 2) containing SOX9 or myc-tagged constitutively active GLI2 (GLI2ΔN)27 or active GLI3 (GLI3A-myc)28, 29 were transiently transfected into LX2 cells. After 24 hours, cells were then treated with CYC or vehicle control for an additional 24 hours and assayed for protein expression. All experiments were carried out with the appropriate empty vector (EV) control.

utively active GLI2 (GLI2ΔN)27 or active GLI3 (GLI3A-myc)28, 29 were transiently transfected into LX2 cells. After 24 hours, cells were then treated with CYC or vehicle control for an additional 24 hours and assayed for protein expression. All experiments were carried out with the appropriate empty vector (EV) control. Gene Expression, Protein Analysis, and Chromatin Immunoprecipitation Assays Antibodies used are listed in Supporting Table 3. Tissue preparation, immunohistochemistry (IHC), ICC, immunoblotting, and quantification were performed as described previously.17 For quantitative reverse-transcription polymerase chain reaction (qPCR), RNA was isolated from cells using the RNeasy kit (Qiagen Ltd., West Sussex, UK). Complementary DNA (cDNA) was synthesized from 1 μg of RNA with a RNA-to-cDNA kit (Applied Biosystems Ltd., Cheshire, UK). qPCR reactions were carried out on a StepOnePlus Real-Time PCR system (Applied Biosystems Ltd) using 1 μL of cDNA, intron-spanning primers, wherever possible (Supporting Table 4), and SYBR green (PrimerDesign Ltd., Southampton, UK). GusB was used as a housekeeper control for gene expression, as described previously.30 Changes in messenger RNA (mRNA) expression were calculated using ΔΔCT methodology. Chromatin immunoprecipitation (ChIP) assays were performed as described previously.31, 32 Further details are described in the Supporting Materials and Methods.

sed as a housekeeper control for gene expression, as described previously.30 Changes in messenger RNA (mRNA) expression were calculated using ΔΔCT methodology. Chromatin immunoprecipitation (ChIP) assays were performed as described previously.31, 32 Further details are described in the Supporting Materials and Methods. Statistical Analysis Statistical significance was determined by the two-tailed Student t test. All experiments were carried out three times or more (n = 3). For rHSCs, experimental data arise from three different preparations of stellate cells from different animals. Error bars in graphs show the standard error for each experiment.

nalysis Statistical significance was determined by the two-tailed Student t test. All experiments were carried out three times or more (n = 3). For rHSCs, experimental data arise from three different preparations of stellate cells from different animals. Error bars in graphs show the standard error for each experiment. Results Expression of SOX9 and OPN in Biliary Duct and Liver Fibrosis in Humans and Rodents SOX9 was detected in the round nuclei of biliary epithelial cells in fetal and adult livers, where it demonstrated cellular colocalization with OPN (Fig. 1 and Supporting Fig. 1). Previous data have independently identified OPN11, 12, 14 and SOX917 in areas of liver fibrosis in animal models. Here, in rat and mouse models of liver fibrosis, nuclear Sox9 localized to desmin-positive cells, confirming its presence in HSCs (Fig. 2A). Opn localized with Sox9 to spindle-shaped HSCs with elongated nuclei in areas of fibrosis as well as to biliary cells (Fig. 2B). In vitro, Opn was barely detected in quiescent rHSCs that lacked Sox9 (Fig. 3A,B and Supporting Fig. 2A,B). However, as reported by others,10, 14Opn expression was induced ∼60-fold and its protein increased as rHSCs became activated on tissue culture plastic over 2 weeks, paralleling the induction of Sox9 and the sequential increase in Col1 (Fig. 3A,B). Similar results were gained using the human cell line, LX2, an in vitro model of stellate cells.26 In high-serum conditions, which mimic stellate cell activation, OPN was increased along with SOX9 (Fig. 3C-E). Final confirmation of OPN cellular colocalization with SOX9 in both activated rHSCs and hHSCs was demonstrated in vitro using immunofluorescence (IF). Nuclear SOX9 is shown surrounded by OPN or α-smooth muscle actin (α-SMA) (Fig. 4). These data led us to question whether SOX9 was capable of regulating OPN.

(Fig. 3C-E). Final confirmation of OPN cellular colocalization with SOX9 in both activated rHSCs and hHSCs was demonstrated in vitro using immunofluorescence (IF). Nuclear SOX9 is shown surrounded by OPN or α-smooth muscle actin (α-SMA) (Fig. 4). These data led us to question whether SOX9 was capable of regulating OPN. Fig. 1 IHC of SOX9 and OPN in healthy liver. Consecutive 5-μm sections of healthy liver in rat and human (at 18 weeks post-conception [wpc] and adulthood) stained for SOX9 and OPN (brown) and counterstained with toludine blue. Note detection only in the round nuclei (SOX9) and cytoplasm (OPN) of biliary epithelial cells. Size bar represents 50 μm. Fig. 2 IHC of SOX9, OPN, and desmin in fibrotic liver. (A) Dual IF in fibrotic tissue from rat and mouse showing nuclear Sox9 (red) in biliary cells (asterisks) and in cells with cytoplasmic staining for desmin (green) (white arrowheads). (B) Consecutive 5-μm tissue sections shown from fibrotic rat and mouse liver stained with Sox9 and Opn (brown) counterstained with toluidine blue. Note similarly located staining for Sox9 and Opn in cells with more spindle-shaped nuclei (arrows) as well as in biliary cells (asterisk). Size bars represent 100 μm.

eads). (B) Consecutive 5-μm tissue sections shown from fibrotic rat and mouse liver stained with Sox9 and Opn (brown) counterstained with toluidine blue. Note similarly located staining for Sox9 and Opn in cells with more spindle-shaped nuclei (arrows) as well as in biliary cells (asterisk). Size bars represent 100 μm. Fig. 3 SOX9 and OPN expression in activated HSCs. (A-E) Quantification of SOX9 and OPN in quiescent and activated rHSCs and LX2 cells by qPCR (A and C) and immunoblotting (B, D, and E) (in [A], Sox9 was up-regulated 8.0-fold). In (B), induction of Sox9, Opn, and Col1 is shown during activation of rHSCs in culture (relative to quiescent; day 0). Representative immunoblotting images for (B) and (D) are shown as inset (B) or as an individual image (E), respectively. All immunoblotting quantification was normalized to β-actin. *P < 0.05; †P < 0.005, compared to quiescent day 0 cells (A and B) or 0% serum (C and D). Fig. 4 SOX9 and OPN expression in activated hHSCs and rHSCs. IF showing nuclear SOX9 (red) and cytoplasmic α-SMA (green; left panel) or OPN (green; right panel) in rat and human activated HSCs (aHSCs). Size bar represents 20 μm.

Fig. 3 SOX9 and OPN expression in activated HSCs. (A-E) Quantification of SOX9 and OPN in quiescent and activated rHSCs and LX2 cells by qPCR (A and C) and immunoblotting (B, D, and E) (in [A], Sox9 was up-regulated 8.0-fold). In (B), induction of Sox9, Opn, and Col1 is shown during activation of rHSCs in culture (relative to quiescent; day 0). Representative immunoblotting images for (B) and (D) are shown as inset (B) or as an individual image (E), respectively. All immunoblotting quantification was normalized to β-actin. *P < 0.05; †P < 0.005, compared to quiescent day 0 cells (A and B) or 0% serum (C and D). Fig. 4 SOX9 and OPN expression in activated hHSCs and rHSCs. IF showing nuclear SOX9 (red) and cytoplasmic α-SMA (green; left panel) or OPN (green; right panel) in rat and human activated HSCs (aHSCs). Size bar represents 20 μm. Sox9 Is Responsible for Opn Expression in Activated HSCs To determine whether Sox9 regulates Opn expression, we abrogated Sox9 using siRNA in activated rHSCs. Reducing Sox9 by 70%-80% caused a commensurate 50%-70% decrease in Opn transcript and its encoded protein (Fig. 5A,B). Similar results were detected in the LX2 HSC line (Fig. 5C). In silico analysis of the OPN 5′ flanking region identified a conserved SOX9 binding motif ∼3 kilobase pairs upstream of the transcriptional start site (Fig. 5D). ChIP demonstrated that Sox9 was enriched at this site in both activated rHSCs and human LX2 cells (Fig. 5E; negative control data for GAPDH shown in Supporting Fig. 3). These data indicate that OPN is a novel downstream target of SOX9. Because others have implicated the Hh pathway in liver fibrosis33 and as a regulator of OPN expression,14, 15 and because in different circumstances SOX9 has been reported downstream of Hh signaling,18 we were curious to investigate whether SOX9 might be regulated by Hh in stellate cells as a means of determining OPN production.

hers have implicated the Hh pathway in liver fibrosis33 and as a regulator of OPN expression,14, 15 and because in different circumstances SOX9 has been reported downstream of Hh signaling,18 we were curious to investigate whether SOX9 might be regulated by Hh in stellate cells as a means of determining OPN production. Fig. 5 SOX9 regulation of OPN in HSCs. (A-C) siRNA abrogation of Sox9 in activated rHSCs (A and B) and LX2 cells (C) standardized against scrambled siRNA control for mRNA (A) and protein (B and C). Example immunoblotting is shown as inset in (B) and (C). *p < 0.05; **P < 0.01; †P < 0.005; ‡P < 0.001, compared to control. (D) Alignment of the upstream OPN enhancer region with conserved SOX9-binding motif highlighted in black (human sequence shown is −3,886 to −3,842 base pairs relative to transcriptional start site). Conserved nucleotides indicated by asterisk (*). The core SOX-binding motif is CAAT with increased binding affinity for SOX9 conferred by additional flanking nucleotides.49 (E) ChIP assay for SOX9-binding element in conserved upstream OPN enhancer element in LX2 cells cultured in high serum and activated rHSCs. Negative control is immunoglobulin G (IgG), and positive control is input (diluted 10-fold).

ith increased binding affinity for SOX9 conferred by additional flanking nucleotides.49 (E) ChIP assay for SOX9-binding element in conserved upstream OPN enhancer element in LX2 cells cultured in high serum and activated rHSCs. Negative control is immunoglobulin G (IgG), and positive control is input (diluted 10-fold). Hh Signaling Regulates SOX9 and Its Downstream Target, OPN Serum-activated LX2 cells and rHSCs activated in culture for 10 days were incubated with the Hh antagonist, CYC, or agonist, SAG (Fig. 6A-D and Supporting Fig. 4A,B). Both SOX9 and OPN proteins were significantly decreased by 45%-60% in response to CYC and increased ∼2- to 3-fold after SAG treatment in both stellate cell models. These data demonstrate that both OPN and SOX9 are positively regulated by Hh signaling in stellate cells. To intimate a role for SOX9 as the mediator of Hh's effect on OPN production, we used siRNA in LX2 cells after SAG augmentation of Hh signaling (Fig. 6E and Supporting Fig. 4C). SAG induced increases in both SOX9 and OPN protein, compared to dimethyl sulfoxide (DMSO) control, which were unaffected by control siRNA. However, siRNA abrogation of SOX9 prevented the Hh agonist from increasing OPN levels above DMSO control values. To perform the converse experiment, transient transfection of an expression vector containing the human SOX9 coding sequence was carried out to overexpress SOX9 in LX2 cells (Supporting Fig. 4D). Overexpression of SOX9 rescued the inhibitory effect of CYC on OPN production (Fig. 6F). Collectively, these data implicate SOX9 as a positive regulator of OPN production downstream of Hh signaling in stellate cells.

g the human SOX9 coding sequence was carried out to overexpress SOX9 in LX2 cells (Supporting Fig. 4D). Overexpression of SOX9 rescued the inhibitory effect of CYC on OPN production (Fig. 6F). Collectively, these data implicate SOX9 as a positive regulator of OPN production downstream of Hh signaling in stellate cells. Fig. 6 Hh regulates SOX9 and OPN expression in HSCs. (A-D) SOX9 and OPN protein levels quantified from immunoblotting of activated rHSCs and LX2 cells after 24-hour treatment with the Hh antagonist, CYC, or the Hh agonist, SAG. (E) Protein levels for SOX9 and OPN after treatment with SAG for 24 hours and knockdown of SOX9 (by 87%) using siRNA or scrambled control in LX2 cells. (F) Quantification of OPN protein after overexpression of SOX9 in LX2 cells in the presence or absence of CYC. Example immunoblotting image shown in inset. Change in expression is compared to vehicle treated cells (DMSO) for all experiments and, in the case of (F), EV control. Experiments standardized against β-actin. *P < 0.05; **P < 0.01; †P < 0.005; ‡P < 0.001, compared to control.

X2 cells in the presence or absence of CYC. Example immunoblotting image shown in inset. Change in expression is compared to vehicle treated cells (DMSO) for all experiments and, in the case of (F), EV control. Experiments standardized against β-actin. *P < 0.05; **P < 0.01; †P < 0.005; ‡P < 0.001, compared to control. The Hh Mediator, GLI2, Regulates SOX9 Expression The GLI family of transcription factors is known to mediate the effects of Hh signaling.16 To determine which GLI factor might be responsible for Hh's effect on SOX9 expression, we first investigated the expression of family members in quiescent and activated rHSCs. By qPCR, Gli1 was poorly detected in quiescent HSCs and unaltered upon activation (Fig. 7A). In contrast, Gli2 and Gli3 mRNAs were increased ∼6- and ∼50-fold, respectively, in activated cells. Although, by this analysis, GLI3 appears the more likely candidate for the regulation of SOX9 in stellate cells, detection of mRNA is not indicative of protein levels, especially given the potential for both repressor or activator forms of GLI2 and GLI3. Several commercial and published antibodies were available to us29, 34, 35; however, we found them unhelpful in detecting or distinguishing the different forms by immunoblotting. Therefore, we investigated the potential contribution of GLI2 and GLI3 to SOX9 and OPN expression by using siRNA in LX2 cells (Fig. 7B,C). Diminution of GLI2 transcripts by ∼67% significantly reduced SOX9 and OPN expression by ∼43% and ∼64%, respectively (Fig. 7B). In comparison, although achieving more robust reduction in GLI3 expression (∼86%) with siRNA, SOX9 expression was less affected and OPN was unaltered (Fig. 7C). Moreover, overexpression of constitutively active GLI2 (GLI2ΔN) was able to rescue, at least partially, the antagonistic effects of CYC on SOX9 and OPN production (Fig. 8A,B and Supporting Fig. 5A). In contrast, overexpressing the activator form of GLI3 (GLI3A) in the presence of CYC had little or no positive effect on SOX9 or OPN production (Fig. 8C,D and Supporting Fig. 5B). These data imply that GLI2 is the predominant regulator of SOX9 expression in HSCs. In keeping with these results, nuclear immunoreactivity against Gli2 was detected in activated rHSCs in vitro and in regions of fibrosis after CCl4 treatment in vivo (Fig. 8E). Interestingly, Gli2 was also detected in the round nuclei of biliary epithelial cells similar to Sox9 and Opn (Fig. 8E, arrows).

ssion in HSCs. In keeping with these results, nuclear immunoreactivity against Gli2 was detected in activated rHSCs in vitro and in regions of fibrosis after CCl4 treatment in vivo (Fig. 8E). Interestingly, Gli2 was also detected in the round nuclei of biliary epithelial cells similar to Sox9 and Opn (Fig. 8E, arrows). In contrast, despite detecting nuclear immunoreactivity for Gli3 in control brain tissue during human fetal development, such staining was not apparent in fibrotic rat tissue (data not shown). Fig. 7 Gli2 mediates the expression of Sox9 and Opn in HSCs. (A) Expression of Gli factors in quiescent and activated rHSCs by qPCR. (B and C) siRNA for GLI2 (B, 67% knockdown) and GLI3 (C, 86% knockdown) or scrambled control in LX2 cells, followed by qPCR for GLI2 (B) or GLI3 (C), SOX9, and OPN. *P < 0.05; †P < 0.005; ‡P < 0.001, compared to scrambled siRNA treatment. Fig. 8 Gli2 overexpression rescues antagonistic effects of CYC on the expression of Sox9 and Opn in HSCs. (A-D) Quantification of SOX9 and OPN protein after overexpression of constitutively active GLI2 (GLI2ΔN; A and B) or active GLI3 (GLI3A; C and D) in LX2 cells in the presence or absence of CYC. (E) IF showing nuclear Gli2 (red) and cytoplasmic α-Sma (green) in activated rHSCs (aHSCs) and bright-field IHC showing nuclear Gli2 (brown staining) in CCl4-treated fibrotic rat liver. Arrows indicate Gli2 expression in a bile duct. *P < 0.05; **P < 0.01; †P < 0.005, ‡P < 0.001, compared to EV transfection. Size bar represents 20 μm.

uclear Gli2 (red) and cytoplasmic α-Sma (green) in activated rHSCs (aHSCs) and bright-field IHC showing nuclear Gli2 (brown staining) in CCl4-treated fibrotic rat liver. Arrows indicate Gli2 expression in a bile duct. *P < 0.05; **P < 0.01; †P < 0.005, ‡P < 0.001, compared to EV transfection. Size bar represents 20 μm. Discussion OPN has been implicated as an important mediator, by which the inflammatory response ultimately leads to scarring and fibrosis in various organs,6-10, 14 with the potential that its presence in the circulation can be used as a biomarker of disease progression.11-13 Previously, we demonstrated a novel role for the transcription factor, SOX9, in models of liver fibrosis. Under the influence of transforming growth factor-beta (TGF-β) signaling, SOX9 became expressed in activated HSCs, where it was responsible for the production of the profibrotic collagen, COL1.17 In this study, we have demonstrated a more diverse role for SOX9 by regulating OPN expression. In the liver, SOX9 and OPN colocalized in the regions of fibrosis. The onset of OPN production during rHSC activation, its reduction in activated HSCs after Sox9 abrogation, and the binding of SOX9 to an upstream OPN enhancer element infers that the transcription factor is required for OPN expression during liver fibrosis.