Browse the corpus

Major depressive disorder (MDD) affects approximately 17 percent of the population, and is predicted to be the number two cause of illness worldwide by the year 20203. The symptoms include cognitive impairment and loss of memory, implicating synaptic dysfunction in the pathophysiology of MDD4–6. This possibility is supported by studies demonstrating a reduction of dendritic spine number and function of neurons in PFC in animal models of depression7,8. Postmortem studies also report morphometric changes in PFC of MDD subjects, including a reduction in neuronal body size1,9. This is also consistent with brain imaging studies reporting a reduction in the volume of PFC subregions in MDD subjects2,10. Together, these studies suggest a loss of spine number and dendritic arbor although there is no direct evidence demonstrating this type of neuronal atrophy in the brains of MDD subjects.

ize1,9. This is also consistent with brain imaging studies reporting a reduction in the volume of PFC subregions in MDD subjects2,10. Together, these studies suggest a loss of spine number and dendritic arbor although there is no direct evidence demonstrating this type of neuronal atrophy in the brains of MDD subjects. To gain insight into the mechanisms that underlie neuronal atrophy and reduced volume of PFC, we have re-analyzed the data from a prior microarray gene expression study that we conducted of the dlPFC (Brodmann area 9) of MDD subjects and matched controls11. We classified the microarray data with a 5% FDR11and found that about 30% of the down-regulated genes in the MDD group could be related to some aspect of synaptic function. A more extended range (20% cutoff) identified an additional subset of down-regulated synaptic function-related genes in dlPFC of MDD brains (Table 1). The synaptic categories include regulation of synaptic vesicles [calmodulin 2, synapsins I and III, Rab3A, amphiphysin, and synaptogyrin 3]12–16regulation of synaptic strength [neurogranin]17dendritic spine formation [Rab4B]18, and axonal outgrowth and regeneration [tubulins]19. It is notable that other genome-wide expression studies of MDD have reported gene alterations of similar functional categories, including cytoarchitecture, vesicular transport, and synaptic transmission, of some of the same synapse-related genes20–23or isoforms of the genes identified here (e.g., subtypes of amphiphysin, synaptogyrin, synapsin, and the Ras superfamily22–24.

of MDD have reported gene alterations of similar functional categories, including cytoarchitecture, vesicular transport, and synaptic transmission, of some of the same synapse-related genes20–23or isoforms of the genes identified here (e.g., subtypes of amphiphysin, synaptogyrin, synapsin, and the Ras superfamily22–24. PCR analysis demonstrated significant decreases for 5 of the 10 genes, and trends for all but one (amphiphysin) of the remaining genes in the dlPFC of MDD subjects (Table 1). In situ hybridization analysis of the five confirmed genes demonstrates enriched expression in gray matter of dlPFC with a laminar distribution in the middle (synapsin I) or middle and deep layers (calmodulin 2, Rab3A, Rab4B and β-tubulin 4) of dlPFC. Quantitative analysis confirms that levels of these five genes are significantly decreased in MDD subjects compared to controls (Fig. 1a–e). Studies in rodents demonstrate that chronic unpredictable stress (CUS), considered one of the most valid rodent models of depression, decreases the expression of synapsin I, calmodulin 2, Rab3A, and Rab4B, but not beta-tubulin 4, in the PFC (Supplementary Fig. 1), suggesting that the decreased levels of these synapse-related genes in MDD result from chronic stress exposure that could contribute to depressive behaviors.

id rodent models of depression, decreases the expression of synapsin I, calmodulin 2, Rab3A, and Rab4B, but not beta-tubulin 4, in the PFC (Supplementary Fig. 1), suggesting that the decreased levels of these synapse-related genes in MDD result from chronic stress exposure that could contribute to depressive behaviors. Decreased expression of synapse-related genes shown here, and a previous report of reduced neuronal cell body size in the dlPFC of MDD subjects1, suggest a reduction in synaptic density, although there is no direct evidence to support this possibility. Examination of dendritic morphology by microtubule-associated protein 2 (MAP2) immunohistochemistry revealed decreased staining of dendritic processes in dlPFC layers III through V of MDD subjects relative to controls, which was confirmed by MAP2 immunoblotting of dlPFC micro-punches (Fig. 1f,g). Using electron microscopic stereological analysis, a marked decrease in spine synapse number was observed in MDD subjects compared to controls (Fig. 1h). Co-factor analysis revealed no significant effects of medication status, age of first episode (<40 vs. ≥40 years old), or suicide on synapse number (Supplementary Figs. 2, 3, 5).

on microscopic stereological analysis, a marked decrease in spine synapse number was observed in MDD subjects compared to controls (Fig. 1h). Co-factor analysis revealed no significant effects of medication status, age of first episode (<40 vs. ≥40 years old), or suicide on synapse number (Supplementary Figs. 2, 3, 5). Examination of the transcription factor binding motifs in the promoter regions of the decreased synapse-related genes identified 3,266 upstream regulatory elements for 218 transcription factors (TRANSFAC scoring matrix, Supplementary Table 1). Twelve transcription factor-binding sites are localized to the upstream regulatory domain of all of the MDD and CUS-altered synapse-related genes (Fig. 2a). The rat homologues of these synapse-related genes include eight of the twelve transcription factor regulatory elements (Fig. 2a and Supplementary Table 2). Analysis of the microarray data revealed that one of these transcription factors GATA1 is significantly increased in MDD patients (Fig. 2b). Other transcriptions factors were either not significantly changed in MDD (GKLF and KID31.0- and 1.1-fold, respectively) or were not included in the microarray gene set. Levels of the related GATA-2, 3, and 4 isoforms were not significantly altered in MDD compared to controls. Co-factor analysis revealed no effect of medication status on GATA1 expression, and no significant differences between all MDD subjects and groups categorized by age of first episode, number of episodes, and suicide, although the number of subjects per subgroup was small (Supplementary Figs. 2–5). Studies in the CUS rodent model of depression demonstrate increased Gata1 expression in the PFC, which was completely reversed by chronic administration of fluoxetine (Fig. 2c). The ability of antidepressant treatment to normalize Gata1 expression in rodents but not in MDD could be due to the small number of the medicated and un-medicated subgroups, as well as treatment resistance and heterogeneity of the subjects. The binding activity of Gata1 to the promoter of the synapse related genes was confirmed by chromatin immunoprecipitation (ChIP) with a Gata1 antibody followed by PCR for the Gata1 binding region of each gene (Fig. 2d). For most of the genes, Gata1 antibody ChIP resulted in an enrichment of the promoter compared to mock, control ChIP.

vity of Gata1 to the promoter of the synapse related genes was confirmed by chromatin immunoprecipitation (ChIP) with a Gata1 antibody followed by PCR for the Gata1 binding region of each gene (Fig. 2d). For most of the genes, Gata1 antibody ChIP resulted in an enrichment of the promoter compared to mock, control ChIP. The possibility that elevated GATA1 underlies decreased expression of the synapse-related genes and the atrophy of dendritic processes was examined in primary neuronal cultures. Initial studies focused on Rab4B because this class of small GTP-binding protein is required for endosomal recycling that is critical for maintenance of spine size18 and because RAB4B showed the greatest reduction (Table 1). Expression of a GFP-tagged rAAV-GATA1 vector in cultured cortical neurons (Fig. 3a,b) significantly decreased the expression of Rab4b (Fig. 3c). Expression of a cofactor target gene, FOG1, that is positively regulated by GATA125,26, was significantly increased by viral expression of GATA1 (not shown). To analyze dendrite morphology, neurons were fixed and labeled with anti-GFP (Fig. 3d,e), as well as anti-MAP2 antibodies (Fig. 3f). Viral expression of GATA1 decreased the complexity of the dendritic arbor (Fig. 3d), the number of spines (Fig. 3e), and the intensity of MAP2 staining (Fig. 3f), which was confirmed by western blot analysis (Fig. 3g). Sholl analysis demonstrated that viral expression of GATA1 significantly decreased the number of dendrite intersections, indicating decreased complexity (Fig. 3h).

ritic arbor (Fig. 3d), the number of spines (Fig. 3e), and the intensity of MAP2 staining (Fig. 3f), which was confirmed by western blot analysis (Fig. 3g). Sholl analysis demonstrated that viral expression of GATA1 significantly decreased the number of dendrite intersections, indicating decreased complexity (Fig. 3h). Next the effects of GATA1 expression on behavior in rodent models of depression were examined. Control or GATA1 viral vectors were infused into the PFC (Fig. 4a), and the expression and location were confirmed by GFP expression (Fig. 4b). Infusion of rAAV-GATA1 produced depressive-like behaviors in two established rodent models. In the forced swim test, rAAV-GATA1 increased the time spent immobile, a measure of behavioral despair that is reversed by antidepressant treatment (Fig. 4c). In the learned helplessness model, exposing animals to inescapable stress causes escape deficits that are reversed by antidepressant treatment. Infusions of rAAV-GATA1 increased the number of escape failures during the initial block of active avoidance testing, similar to the effects of inescapable stress exposure (Fig. 4d). During the second block of active avoidance testing, there was no significant effect (Fig. 4d), indicating that GATA1 delays responding, but does not produce a sustained effect in this model. Infusion of rAAV-GATA1 did not influence locomotor activity (not shown), indicating that there was no generalized effect on ambulation. Further studies showed that the effects of rAAV-GATA1 in the forced swim test were not reversed by the antidepressant imipramine as expected, since drug treatment would not influence viral expression of GATA1 (Supplementary Fig. 6).

or activity (not shown), indicating that there was no generalized effect on ambulation. Further studies showed that the effects of rAAV-GATA1 in the forced swim test were not reversed by the antidepressant imipramine as expected, since drug treatment would not influence viral expression of GATA1 (Supplementary Fig. 6). The influence of GATA1 on depressive behavior caused by chronic stress was examined with a viral knock down strategy, using a small hairpin RNA (shRNA) targeted to GATA1 (rAAV-GATA1shRNA) (Fig. 4e). The ability of rAAV-GATA1shRNA to effectively decrease Gata1 mRNA was confirmed in cultured cells and rat PFC (Supplementary Fig. 7). The rAAV-GATA1shRNA or scrambled control (rAAV-ScrshRNA) was infused into the PFC (Fig. 4f) of rats that were then subjected to the CUS paradigm shown to increase Gata1 mRNA (Fig. 2c). This paradigm results in anhedonia, a core symptom of depression that can be measured by preference for a sweetened solution27. CUS exposure significantly decreased sucrose preference in control rats infused with rAAV-ScrshRNA and this effect was completely blocked by infusion of rAAV-GATA1shRNA. There was no effect of rAAV-GATA1shRNA in non-stressed rats indicating that basal levels of GATA1 are low and not sufficient to suppress basal rates of sucrose preference.

antly decreased sucrose preference in control rats infused with rAAV-ScrshRNA and this effect was completely blocked by infusion of rAAV-GATA1shRNA. There was no effect of rAAV-GATA1shRNA in non-stressed rats indicating that basal levels of GATA1 are low and not sufficient to suppress basal rates of sucrose preference. Using three complimentary approaches we have identified and confirmed a subset of synapse-related genes that are repressed in dlPFC of MDD subjects. At the presynaptic level, synapsin I, Rab3A and calmodulin 2 regulate the size, number, and targeting of synaptic vesicles, and β-tubulins are involved in axonal outgrowth and regeneration19, 28–31. Postsynaptically, Rab4B regulates endosomal recycling that is required for spine maintenance and neurotransmitter receptor recycling18,32. Consistent with the role of these genes in synaptic function and structure, our results also demonstrate a reduction in synapse numbers in dlPFC. Together these findings demonstrate molecular and cellular alterations that could underlie the reduction in neuronal cell body size and volume of PFC in MDD patients1,10. Decreased expression of these genes in response to chronic stress exposure also suggests an etiological relationship to MDD, which is often associated with severe life stress and trauma33,34.

lar and cellular alterations that could underlie the reduction in neuronal cell body size and volume of PFC in MDD patients1,10. Decreased expression of these genes in response to chronic stress exposure also suggests an etiological relationship to MDD, which is often associated with severe life stress and trauma33,34. GATA1 was also identified as a transcriptional repressor that is increased in MDD and has binding elements in the promoter regions of the synapse related genes. GATA1 is a member of a zinc finger family of transcription factors that are evolutionarily conserved and play important roles in embryonic development35. Although originally characterized in hematopoietic and cardiac tissues, GATA transcription factors are also expressed in endocrine tissue and brain, and are reported to regulate neuronal differentiation during development35–38. Although GATA1 levels are low in controls, expression is increased in MDD and in response to chronic stress. Over expression of Gata1 was sufficient to cause dendrite atrophy and decreased synaptic protein expression in cultured cortical neurons. Moreover, expression of Gata1 in PFC was sufficient to produce depressive behaviors, while knock down of Gata1 completely blocked depressive, anhedonic behavior caused by CUS exposure. Together, these studies demonstrate that expression of Gata1 in the PFC is sufficient and necessary for the development of depressive behaviors in multiple animal paradigms. Further evidence for a role of GATA1 is provided by a recent report of a polymorphism in the GATA1 binding site of the promoter for interleukin 6, an inflammatory cytokine dysregulated in depression39.

Gata1 in the PFC is sufficient and necessary for the development of depressive behaviors in multiple animal paradigms. Further evidence for a role of GATA1 is provided by a recent report of a polymorphism in the GATA1 binding site of the promoter for interleukin 6, an inflammatory cytokine dysregulated in depression39. Together the results describe molecular and cellular mechanisms that are sufficient to cause a reduction in synapse numbers and dendritic complexity and could contribute to neuronal atrophy and decreased cortical volume reported in MDD patients. The results suggest that approaches that block or reverse neuronal atrophy in the PFC could be effective antidepressant treatments. This possibility is supported by recent studies demonstrating that the rapid antidepressant actions of NMDA receptor blockade are associated with increased spine number and function, and increased synaptogenesis in the PFC40. Online Methods Human Subjects and Microarray Analysis The description of postmortem tissue collection, subjects, and microarray analysis are exactly as described in our previous study11.

Together the results describe molecular and cellular mechanisms that are sufficient to cause a reduction in synapse numbers and dendritic complexity and could contribute to neuronal atrophy and decreased cortical volume reported in MDD patients. The results suggest that approaches that block or reverse neuronal atrophy in the PFC could be effective antidepressant treatments. This possibility is supported by recent studies demonstrating that the rapid antidepressant actions of NMDA receptor blockade are associated with increased spine number and function, and increased synaptogenesis in the PFC40. Online Methods Human Subjects and Microarray Analysis The description of postmortem tissue collection, subjects, and microarray analysis are exactly as described in our previous study11. Quantitative Real Time PCR An aliquot of the total RNA that was previously extracted from prefrontal cortex punches was used for secondary validation using real time PCR. Five hundred ng of total RNA was used for cDNA synthesis using oligo dT primers and SuperScript II reverse transcriptase (Invitrogen), and subsequently diluted with nuclease-free water to 10 ng ul−1 cDNA. Gene-specific high-melt temperature primers for genes of interest were designed using Primer 3 software (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi; Whitehead Institute for Biomedical Research) and expressed sequence information obtained from GenBank (NCBI). PCR reactions were conducted on an ABI 7900 Sequence Detection System (Applied Biosystems) using a hotstart SYBR-green based method (Quantitect, QIAGEN) followed by melt curve analysis to verify specificity of product. The CT value (cycle number at threshold) was used for calculations of relative amount of mRNA molecules. The CT value of each target genes was normalized by subtraction of the CT value from multiple housekeeping genes. This value is the ΔCT. The difference in ΔCT between Control and MDD represents the ΔΔCT, and the relative quantitative change was showed as 2−ΔΔCT. All of the genes of interest were normalized to the housekeeping gene, cyclophilin.

t genes was normalized by subtraction of the CT value from multiple housekeeping genes. This value is the ΔCT. The difference in ΔCT between Control and MDD represents the ΔΔCT, and the relative quantitative change was showed as 2−ΔΔCT. All of the genes of interest were normalized to the housekeeping gene, cyclophilin. Electronic Microscopic Stereology The number of spine synapses in dlPFC of postmortem brains was counted as previously described41,42. Briefly, postmortem samples were immersed overnight in a mixture of 2% glutaraldehyde + 4% paraformaldehyde dissolved in phosphate buffer. Vibratome sections of 100 µm were cut throughout the tissue blocks and embedded in Durcupan (Electron Microscopy Sciences, Fort Washington, PA). Using the embedded sections, at least ten sampling areas were randomly selected from each brain, and ultrathin sections (~70 nm) from each sampling area were prepared with a Reichert Ultracut E ultrotome. Digitized electron micrographs of each sampling area were taken in a Tecnai 12 transmission electron microscope (FEI company, Hillsboro, OR). Synapses were counted using the dissector technique. Synapse numbers for each brain were obtained by averaging data from at least ten dissectors.

a Reichert Ultracut E ultrotome. Digitized electron micrographs of each sampling area were taken in a Tecnai 12 transmission electron microscope (FEI company, Hillsboro, OR). Synapses were counted using the dissector technique. Synapse numbers for each brain were obtained by averaging data from at least ten dissectors. In situ hybridization Sections of human brain were cut at 14 µm from blocked dlPFC. Rat brains were also sectioned at 14 µm, after animals were euthanized and brains stored at −80 °C. Brains were mounted onto ‘probe-on plus’ glass slides (Fisher Scientific). To prepare complimentary RNA probe (cRNA) for in situ hybridization histochemistry, the fragments of cDNA were amplified by PCR using T7 promoter attached primer and PCR products were used to generate 35S-radiolabeled riboprobes using T7 RNA polymerase in vitro transcription (Ambion, T7-MEGAshortscript). After fixation and acetylation, human or rat sections were hybridized with 35S-CTP incorporated RNA probes at 60 °C for 16 hr in the hybridization solution. Sections were then washed at 60 °C, dehydrated with graded alcohols, air-dried and exposed to film.

polymerase in vitro transcription (Ambion, T7-MEGAshortscript). After fixation and acetylation, human or rat sections were hybridized with 35S-CTP incorporated RNA probes at 60 °C for 16 hr in the hybridization solution. Sections were then washed at 60 °C, dehydrated with graded alcohols, air-dried and exposed to film. Stress and antidepressant model For chronic unpredictable stress, rats were subjected to chronic unpredictable stress (CUS) for 35 days as previously described40; fluoxetine was administrated from day 15–35 (twice a day, 15mg kg−1 per treatment). Control rats were handled every other day for the first 15 days or then daily for vehicle (saline) administration and brains were harvested for in situ hybridization analysis 4 hr after the last vehicle or drug treatment. For stress study with rAAV-GATA1shRNArats were subjected to 3 weeks of chronic unpredictable stress after the viral infusion. Transcription factor binding motif search The scoring matrix of TRANSFAC (http://www.biobase-international.com) was used for the binding motif search in promoter regions (1 kb of the 5’-flanking sequence and 200 bp on downstream from transcription start site) of our candidate genes. The over-represented TFBSs were determined in the set of promoters with the parameter of exact matches of core sequences (Core Score = 1) at 0.95 matrix similarity cut-off (Matrix Score ≥ 0.95) in the vertebrate matrix.

5’-flanking sequence and 200 bp on downstream from transcription start site) of our candidate genes. The over-represented TFBSs were determined in the set of promoters with the parameter of exact matches of core sequences (Core Score = 1) at 0.95 matrix similarity cut-off (Matrix Score ≥ 0.95) in the vertebrate matrix. Chromatin Immunoprecipitation Cultured primary cortical neurons (DIV 7) were incubated in 1% formaldehyde for 10 min at RT for cross-linking of DNA and DNA binding proteins. Crosslinked neurons were lysed in SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl, pH 8.1) including protease inhibitors and phosphatase inhibitors. Chromatin was sheared by sonication and separated by centrifugation for 10 min at 13000 rpm. The supernatant was collected and immunoprecipitated with dynabead (Dynal, 112-01D) conjugated 1 ug of anti-GATA1 (Santa Cruz, sc-265) or normal horse serum overnight at 4 °C. After reverse cross-linking, DNA was eluted using Qiagen minelute purification kit and quantified by real time PCR.

ion for 10 min at 13000 rpm. The supernatant was collected and immunoprecipitated with dynabead (Dynal, 112-01D) conjugated 1 ug of anti-GATA1 (Santa Cruz, sc-265) or normal horse serum overnight at 4 °C. After reverse cross-linking, DNA was eluted using Qiagen minelute purification kit and quantified by real time PCR. Primary cortical neuron culture Embryonic brains were prepared from fetal SD rat at 18 d gestation and mechanically triturated as described previously43. Dissociated cells were plated on 12 well plates (0.5x105 cells per well) in plating medium consisting of Neurobasal media supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 1 mM sodium pyruvate, 1 mM HEPES, 2% pen-strep and 2% B27 supplement. Proliferation of non-neuronal cells was halted by replacing with plating medium lacking fetal serum in the next day (DIV 1). Cultures were then fed once a week with plating medium lacking fetal serum and maintained at 37 °C in a humidified 5% CO2 atmosphere.

ium pyruvate, 1 mM HEPES, 2% pen-strep and 2% B27 supplement. Proliferation of non-neuronal cells was halted by replacing with plating medium lacking fetal serum in the next day (DIV 1). Cultures were then fed once a week with plating medium lacking fetal serum and maintained at 37 °C in a humidified 5% CO2 atmosphere. Generation of rAAV vectors To generate rAAV-GATA1-GFP, mouse GATA1 cDNA (1801 bp) was subcloned in the mammalian expression vector (pXM-GATA1) and inserted into the pAAV-EGFP-pA (5403 bp) vector by enzyme restriction and ligation. Hairpin RNA was designed to specifically target GATA1 mRNA. A 24-bp oligonucleotide sequence within the coding region of the GATA1 gene was identified. Hairpins were designed such that the antisense strand came before the sense strand during transcription. Two sets of oligonucleotides were synthesized (Integrated DNA Technologies, Coralville, Iowa) for cloning: GATA1shRNA top: 5’-TTTGAAGATGCTGTATCCAAACTCTCACTTCCTGTCATGAGAGTTTGGATACAGCATCTTATTTTT-3’; bottom: 5’-CTAGAAAAATAAGATGCTGTATCCAAACTCTCATGACAGGAAGTGAGAGTTTGGATACAGCATCTTC-3’; The oligonucleotides had Sap1 and Xba1 overhangs to allow for ligation into mU6pro region of the modified pAAV-MCS vector, pAAV-shRNA. The pAAV-shRNA plasmid was designed to coexpress small hairpin RNA (shRNA) and enhanced green fluorescent protein under the control of an independent RNA polymerase II promoter for enhanced green fluorescent protein and U6 promoter for shRNA. All final clones were verified for sequencing. For virus preparation, HEK293 cells were co-transfected with pAAV-GATA1-EGFP or pAAV-EGFP or pAAV-ScrshRNA or pAAV-GATA1shRNA, pAAV-RC and pHelper using calcium phosphate precipitate method. After 72 hours, vectors were purified from cell lysates by iodixanol gradient centrifugation followed by column concentration. Behavioral experiments using rAAV-GATA1shRNAwere performed after ensuring efficient knockdown of GATA1 mRNA in human embryonic kidney cells (HEK293). For in vitro validation, plasmids expressing scrambled shRNA (pScrshRNA) and GATA1shRNA (pGATA1shRNA) were transfected intor HEK cells using lipofectamine (Invitrogen) following manufacturer’s instructions. 24 hrs after the transfection, RNA was purified using RNaqueous kit (Ambion) and cDNA was synthesized for qPCR. For in vivo validation, animals were perfused and medial PFC was microdissected after behavioral testing.

re transfected intor HEK cells using lipofectamine (Invitrogen) following manufacturer’s instructions. 24 hrs after the transfection, RNA was purified using RNaqueous kit (Ambion) and cDNA was synthesized for qPCR. For in vivo validation, animals were perfused and medial PFC was microdissected after behavioral testing. Total RNA was extracted from the PFC using RNAeasy FFPE kit (Qiagen) and quantitative RT-PCR was performed as described above. Western blot Cultured neurons were lysed in RIPA buffer. Protein levels were measured using BCA Protein Assay Kit. Proteins were separated by SDS-PAGE and transferred to the nitrocellulose membranes. After blocking with 10% skim milk, the blots were incubated with primary antibodies overnight at 4 °C followed by horseradish peroxidase-conjugated secondary antibodies for 1 hr at RT. Blots were visualized by enhanced chemiluminescence and exposed on films. Immunohistochemistry Sections were fixed in 4% paraformaldehyde-PBS for 10 min and followed by 0.6% hydrogen peroxide treatment for 5 min. Sections were incubated in a blocking solution (2.5% BSA in PBS) for 30 min, reacted with rabbit polyclonal anti-GFP (abcam, 1:500) or mouse monoclonal antibody recognizing MAP2 (Sigma, 1:200) in PBS (0.25% Triton X, 1% BSA) and then reacted with secondary antibody for 1 hr at RT.

hydrogen peroxide treatment for 5 min. Sections were incubated in a blocking solution (2.5% BSA in PBS) for 30 min, reacted with rabbit polyclonal anti-GFP (abcam, 1:500) or mouse monoclonal antibody recognizing MAP2 (Sigma, 1:200) in PBS (0.25% Triton X, 1% BSA) and then reacted with secondary antibody for 1 hr at RT. Sholl analysis Dendritic branching was investigated by Sholl analysis44. A transparent grid showing concentric circles was placed over the dendritic image, the smallest circle was centered in the soma and the distance between each circle was equivalent to 10 µm apart. Total dendritic length was estimated by counting the total number of circle intersections and the density of dendrites was defined by counting the number of intersections on each circles. The confocal images of neurons were analyzed using Image J software (www.rcb.info.nih.gov/ij) which has plugged in Sholl analysis.

. Total dendritic length was estimated by counting the total number of circle intersections and the density of dendrites was defined by counting the number of intersections on each circles. The confocal images of neurons were analyzed using Image J software (www.rcb.info.nih.gov/ij) which has plugged in Sholl analysis. Stereotaxic surgery and infusions Rats were anesthetized with xylazine (6 mg kg−1, i.m., Lloyd laboratories, Shenandoar, IA) and ketamine (80 mg kg−1 i.m., Fort Dodge Animal Health, Overland Park, KS). Bilateral viral injections were performed with coordinates –3.5 mm (anterior/posterior), −0.5 mm (lateral), and −4.0 mm (dorsal/ventral) relative to the Bregma. A total of 3 µl of purified virus was delivered at a rate of 0.1µl min−1. Needles were removed and the scalp incision was closed with wound clips. After the behavioral testing was performed, animals were perfused with 4% paraformaldehyde. The brain was kept overnight in 4% paraformaldehyde and then transferred to 30% sucrose. 40 µm sections were cut using a microtome for visualization of GFP. Learned Helplessness (LH) paradigm The LH procedure was performed in custom built, 2-chambered shuttle boxes (Med Associates, Vermont) as previously described45,46. For active avoidance testing, animals were exposed to 30 escape trials using an FR1 schedule in which a single crossing terminated the footshock. Numbers of escape were automatically scored. Results are expressed as number of escape failures observed.

huttle boxes (Med Associates, Vermont) as previously described45,46. For active avoidance testing, animals were exposed to 30 escape trials using an FR1 schedule in which a single crossing terminated the footshock. Numbers of escape were automatically scored. Results are expressed as number of escape failures observed. Forced swim test (FST) FST was conducted as previously described47. On the test day, rats were placed for 10 min in a clear cylinder with water (24 ± 1°C, 45 cm depth). The sessions were recorded from the side and the immobility time was scored by a blind observer. Immobility was defined as floating or remaining motionless without leaning against the wall of the cylinder. Locomotor Activity (LA) To determine if there are general alterations in ambulation, locomotor activity was assessed. Rats were placed in clear plastic chambers fitted with automated activity meters (MED Associates, St. Albans, VT). Locomotor activity was recorded in 5-min bins for a total of 30 min. Sucrose Preference test (SPT) SPT was performed on day 21 after CUS. Animals were habituated to 1% sucrose solution (2%, Sigma, St. Louis, MO) for 3 d, followed by 6 h fluid deprivation on the test day. Rats were presented with two identical bottles, one containing 1% sucrose and the other water for 1 h and the consumption of each solution was measured. Sucrose preference was calculated as the percentage of sucrose consumed over total fluid consumption. List of PCR primers used in this study for validation mRNA expression

Sucrose Preference test (SPT) SPT was performed on day 21 after CUS. Animals were habituated to 1% sucrose solution (2%, Sigma, St. Louis, MO) for 3 d, followed by 6 h fluid deprivation on the test day. Rats were presented with two identical bottles, one containing 1% sucrose and the other water for 1 h and the consumption of each solution was measured. Sucrose preference was calculated as the percentage of sucrose consumed over total fluid consumption. List of PCR primers used in this study for validation mRNA expression Gene Forward (5'->3') Reverse (3'->5') CALM2 GGAATTAGTCCGAGTGGAGAG TGCTCTTCAGTCAGTTGGTCA SYN1 AGATTTTTGGGGGACTGGAC TGACCACGAGCTCTACGATG SYN3 CCAGATAGCTCCACCAGCTC GGATCCTGGAGAGGAGAAGG SYNGR3 AGCTGCTCCTCTCTCTGTGG TAGGACAGGGCTGGTGTCTT AMPH CCAACCATGAAGAGGAAGGA GAGGGTCCTGGATAGGCTTC NRGN GGCTCGGTAGGAGGAGTCTT AGCACAGTCACATGCACACA RAB3A GGTGGTGTCATCAGAACGTG GTCCAACGACTCGGACATCT RAB4B GGTCAGTGACGCGGAGTTAT AGGGTCCAGGTCCTTCTTGT TUBB2 AGCTCACCCAGCAGATGTTT CATTTGCTCATCCACCTCCT TUBB4 CTCGAGGCTTCTGACCTTTG CCAGGGTCGGAGATGAAGTA GATA1 CCAAGCTTCGTGGAACTCTC AGGCGTTGCATAGGTAGTGG PPIA CTCGTGCCGTTTTGCAGAC TGCAAACAGCTCAAAGGAGA Calm2 AGCGAGTCGAGTGGTTGTCT CCGTCGGCATCTACTTCATT Syn1 CCAGCTCAACAAATCCCAGT GGTGTCAGTCGGAGAAGAGG Rab3a GGTGGACGTGATCTGTGAGA GAAGGGAAGTGTGGTTCAGC Rab4b ACTATTGGCGTGGAGTTTGG CACAGAGGATGACCACGATG Tubb4 GGGGACCTCAACCACCTAGT ATCCTGGCATGAAGAAGTGG Gata1 GATGGAATCCAGACGAGGAA CTCTCCGCAATTCCCACTAC Ppia TTGCAGACGCCGCTGTCTCTT TGGAACTTTGTCTGCAAACAGCTCG GATA1 binding site Gene Forward (5'->3') Reverse (3'->5')

Gene Forward (5'->3') Reverse (3'->5') CALM2 GGAATTAGTCCGAGTGGAGAG TGCTCTTCAGTCAGTTGGTCA SYN1 AGATTTTTGGGGGACTGGAC TGACCACGAGCTCTACGATG SYN3 CCAGATAGCTCCACCAGCTC GGATCCTGGAGAGGAGAAGG SYNGR3 AGCTGCTCCTCTCTCTGTGG TAGGACAGGGCTGGTGTCTT AMPH CCAACCATGAAGAGGAAGGA GAGGGTCCTGGATAGGCTTC NRGN GGCTCGGTAGGAGGAGTCTT AGCACAGTCACATGCACACA RAB3A GGTGGTGTCATCAGAACGTG GTCCAACGACTCGGACATCT RAB4B GGTCAGTGACGCGGAGTTAT AGGGTCCAGGTCCTTCTTGT TUBB2 AGCTCACCCAGCAGATGTTT CATTTGCTCATCCACCTCCT TUBB4 CTCGAGGCTTCTGACCTTTG CCAGGGTCGGAGATGAAGTA GATA1 CCAAGCTTCGTGGAACTCTC AGGCGTTGCATAGGTAGTGG PPIA CTCGTGCCGTTTTGCAGAC TGCAAACAGCTCAAAGGAGA Calm2 AGCGAGTCGAGTGGTTGTCT CCGTCGGCATCTACTTCATT Syn1 CCAGCTCAACAAATCCCAGT GGTGTCAGTCGGAGAAGAGG Rab3a GGTGGACGTGATCTGTGAGA GAAGGGAAGTGTGGTTCAGC Rab4b ACTATTGGCGTGGAGTTTGG CACAGAGGATGACCACGATG Tubb4 GGGGACCTCAACCACCTAGT ATCCTGGCATGAAGAAGTGG Gata1 GATGGAATCCAGACGAGGAA CTCTCCGCAATTCCCACTAC Ppia TTGCAGACGCCGCTGTCTCTT TGGAACTTTGTCTGCAAACAGCTCG GATA1 binding site Gene Forward (5'->3') Reverse (3'->5') Calm2 GGGAAGAAGATCCATGGAGAA GGAGAATTCAGCCTGGAAAA Syn1 ACAGCCTGCATCTCTCCATT ATGTCTGCAAGGGTCCAAAG Rab3a CTGTCAAGGTCACAGCTTGG GTAGCAGGACCACACACAGG Rab4b GGCCTGTGGGGTTAGAATTT AAAAATCGGAGCAGAGACGA Tubb4 CTGCCTCTTCGTCTCTAGCC ACCGCTCAGGCTCACCTT Supplementary Material 1 2 3 Acknowledgments This work is supported by USPHS grants MH45481 (RSD), 2 P01 MH25642 (RSD), the Connecticut Mental Health Center (RSD), MH67996 (CAS), P20 RR17701 (CAS), and a National Research Foundation of Korea (NRF, No. 2011-0028317) (HS). We acknowledge the invaluable contributions made by the families consenting to donate brain tissue and be interviewed. We also thank the Cuyahoga County Coroner and staff, Cleveland, Ohio, for their willing assistance. The contributions of James Overholser, Ph.D., George Jurjus, M.D., Herbert Meltzer, M.D., Lisa Konick, Lisa Dieter, Nicole Herbst, Gouri Mahajan, Heather Kooiman and Justin Cobb in the psychiatric assessment and human tissue dissection and preparation are gratefully noted. We thank Dr. Tae Hoon Kim (Yale University) for his advice and suggestions regarding the binding motif analysis and chromatin immunoprecipitation. The pXM-GATA1 plasmid was kindly provided by Dr. Alan B. Cantor (Harvard) and pAAV-EGFP-pA vector was provided by Dr. Ralph J. Dileone (Yale). The GATA1 antibody was provided by Dr. Emery H. Bresnick (University of Wisconsin).

dvice and suggestions regarding the binding motif analysis and chromatin immunoprecipitation. The pXM-GATA1 plasmid was kindly provided by Dr. Alan B. Cantor (Harvard) and pAAV-EGFP-pA vector was provided by Dr. Ralph J. Dileone (Yale). The GATA1 antibody was provided by Dr. Emery H. Bresnick (University of Wisconsin). Author Contributions

dvice and suggestions regarding the binding motif analysis and chromatin immunoprecipitation. The pXM-GATA1 plasmid was kindly provided by Dr. Alan B. Cantor (Harvard) and pAAV-EGFP-pA vector was provided by Dr. Ralph J. Dileone (Yale). The GATA1 antibody was provided by Dr. Emery H. Bresnick (University of Wisconsin). Author Contributions H.J.K was involved in planning and conducting all aspects of the research, including analysis of microarray data, confirmation of results, molecular and cellular experiments, and construction of the viral vectors; H.J.K. also prepared the first draft of the manuscript. B.V. designed viral vectors, conducted the behavioral studies and was involved in the analysis and interpretation of rodent behavioral studies, C.A.S. and G.R. were responsible for generation of relevant human subjects’ information tables and methodology, and for preparation of human tissue and microdissections. T.H. conducted electron microscopy analysis of synapse number in postmortem tissue. M.B. with the technical help of A.L. carried out the rat CUS studies and prepared tissues for in situ hybridization analysis, animal surgeries and supervised behavioral studies. P.L. was involved in the viral vector preparations and H.S. was involved in viral vector experiments, including surgical infusions, behavioral studies, and immunohistochemistry. M.S.M and L.S.J synthesized the GATA1 specific inhibitor, K-7174. R.S.D. was responsible for overseeing the study, including all aspects of study design, data analysis, interpretation of results and preparation of the manuscript and figures. All authors discussed the results presented in the manuscript.

histochemistry. M.S.M and L.S.J synthesized the GATA1 specific inhibitor, K-7174. R.S.D. was responsible for overseeing the study, including all aspects of study design, data analysis, interpretation of results and preparation of the manuscript and figures. All authors discussed the results presented in the manuscript. Competing Financial Interests The authors declare no financial interests. Figure 1 Synaptic function-related genes, the number of spine synapses and MAP2 expression are decreased in dlPFC of MDD subjects. In situ hybridization analysis of five of the dysregulated genes that were confirmed by PCR analysis, including (a) calmodulin 2, (b) synapsin I, (c) Rab3A, (d) Rab4B, and (e) beta-tubulin 4. Representative images of the control (top) and MDD (bottom) autoradiographs (scale bar, 5 mm), and quantitative analysis (bar graphs) are shown. Results represent means ± S.E.M.s (n = 5). *P < 0.05 compared to control (unpaired t-test). (f) Levels of MAP2 were quantified by western blot analysis. Representative blot and quantitative results are shown. Results represent the mean ± S.E.M. (n = 4). *P < 0.05 compared to control (unpaired t-test). (g) Representative images of immunohistochemistry of MAP2 in dlPFC of control (upper) and MDD subjects (lower) (scale bar, 2 mm). (h) High power representative electron micrograph. Arrowheads point to examples of spine synapses (scale bar, 500 nm). Synapses were quantified in layer II/III of dlPFC (PFC II/III), and the results represent the mean ± SD (n = 5), *P < 0.05 compared to control (unpaired t-test).

MDD subjects (lower) (scale bar, 2 mm). (h) High power representative electron micrograph. Arrowheads point to examples of spine synapses (scale bar, 500 nm). Synapses were quantified in layer II/III of dlPFC (PFC II/III), and the results represent the mean ± SD (n = 5), *P < 0.05 compared to control (unpaired t-test). Figure 2 Identification of a transcriptional repressor of synapse-related genes. (a) Summary of transcription factor binding motif analysis. To determine the common promoter elements in the 5 synapse related genes confirmed by qPCR, overlapping binding motifs were identified by Transfac analysis. Eight elements including GATA1 are present in the promoter regions of all the synapse related genes of both human and rat. (b) Expression of GATA1 mRNA is increased in dlPFC of MDD subjects. Results from microarray and qPCR analysis of GATA1 are shown, and represent the mean fold change relative to matched controls (n = 5 microarray pairs, and n = 10 pairs for PCR). *P <0.05 compared to control (unpaired t-test). (c) Influence of chronic unpredictable stress (CUS, 35 d) on the expression of Gata1 in rat PFC was determined by in situ hybridization (representative image shown, scale bar, 1 mm). Increased expression of Gata1 was reversed by fluoxetine administration for the last 21 d of CUS (C+F). Fluoxetine (FLX) alone did not influence Gata1 expression. Results represent the mean ± S.E.M (n = 4); *P <0.05 compared to control (unpaired t-test). (d) Chromatin immunoprecipitation (ChIP) with GATA1 antiserum and PCR analysis of the putative GATA1 binding sites on the promoter regions of the synapse-related genes. Chromatin was immunoprecipitated using a GATA1 antibody and the presence of each gene was determined by PCR; lanes 1, no DNA; 2, horse serum ChIP (negative control); 3, GATA1 antibody ChIP; 4, total chromatin (positive control).

s of the putative GATA1 binding sites on the promoter regions of the synapse-related genes. Chromatin was immunoprecipitated using a GATA1 antibody and the presence of each gene was determined by PCR; lanes 1, no DNA; 2, horse serum ChIP (negative control); 3, GATA1 antibody ChIP; 4, total chromatin (positive control). Figure 3 GATA1 over-expression in cortical neurons decreases Rab4b expression and reduces dendrite branching. (a) rAAV-GATA1-EGFP vector construction. ITR, inverted terminal repeats; CMV, cytomegalovirus promoter; hGH-polyA, human growth hormone polyA signal (b) Levels of Gata1 protein, determined by western blot, are increased in response to rAAV-GATA1 infection; note that the size is shifted due to expression of conjugated GATA1-EGFP. (c) rAAV-GATA1 expression decreases levels of Rab4bdetermined by qRT-PCR. Cells were transfected at DIV5 and extracted at DIV19. Results of qPCR analysis of Rab4b represent the mean ± S.E.M. (n = 4). *P <0.05 compared to vector alone (unpaired t-test). (d, e) Images of cortical neurons infected with control virus (left panels) or rAAV-GATA1-EGFP (right panels) and stained with antibodies to GFP (d, scale bar = 20 µm; e, scale bar = 10 µm). (f) Cells were infected with control virus (left panel) or rAAV-GATA1-EGFP (right panel) as described above but were stained for the dendritic marker MAP2 (scale bar = 20 µm), which was also examined by (g) western blot analysis (n = 3). *P <0.05 compared to vector alone (unpaired t-test). (h) Representative picture of Sholl analysis and quantitative results of the number of neuronal process intersections (scale bar = 20 µm). Results represent the mean ± S.E.M. (n = 7 for control vector, n = 4 for GATA1). *P <0.01 compared to vector alone (unpaired t-test).

.05 compared to vector alone (unpaired t-test). (h) Representative picture of Sholl analysis and quantitative results of the number of neuronal process intersections (scale bar = 20 µm). Results represent the mean ± S.E.M. (n = 7 for control vector, n = 4 for GATA1). *P <0.01 compared to vector alone (unpaired t-test). Figure 4 Viral expression of GATA1 in rodent PFC causes depressive behavior. (a) Schematic diagram of the experimental schedule. (b) Representative low power (left panel) image of GFP immunohistochemistry (IHC) demonstrating the location of the viral infusion site in the medial PFC. The right panel shows the labeled neurons at higher magnification. (c,d) Influence of rAAV-control and rAAV-GATA1 on behavior in the forced swim test (FST, c) and learned helplessness (LH, d) paradigm. There was no effect on levels of locomotor activity (LA) (not shown). Results represent the mean ± S.E.M. (n = 9 for rAAV-control and n = 10 for rAAV-GATA1). * P < 0.05 compared to rAAV-control (t-test). (e) Schematic diagram of Gata1 shRNA knock down in the chronic unpredictable stress (CUS) paradigm. (f) Representative image of GFP immunohistochemistry demonstrating the location of viral infusion in the same region of the PFC as shown in b. (g) Influence of rAAV-ScrshRNA and rAAV-GATA1shRNA on sucrose preference in control and CUS exposed rats. Results represent the mean ± S.E.M. (Control: n = 6 for rAAV-ScrshRNA; n = 8 for rAAV-GATA1shRNA; CUS: n = 7 for rAAV-ScrshRNA; n = 8 for rAAV-GATA1shRNA). ** P < 0.05 compared to Control group (ANOVA and Fishers PLSD post-hoc analysis).

crshRNA and rAAV-GATA1shRNA on sucrose preference in control and CUS exposed rats. Results represent the mean ± S.E.M. (Control: n = 6 for rAAV-ScrshRNA; n = 8 for rAAV-GATA1shRNA; CUS: n = 7 for rAAV-ScrshRNA; n = 8 for rAAV-GATA1shRNA). ** P < 0.05 compared to Control group (ANOVA and Fishers PLSD post-hoc analysis). Table 1 Gene Genbank ID Description Fold Change P value Fold Change P value (microarray) (MAANOVA, FDR) (RT-PCR) (t-test) CALM2 BC008437 Calmodulin 2 0.55 0.004 0.81 0.03* SYN1 NM_133499 Synapsin I 0.60 0.50 0.75 0.01* SYN3 AF046873 Synapsin III 0.79 0.10 0.79 0.09 SYNGR3 NM_004209 Synaptogyrin 3 0.68 0.01 0.75 0.12 AMPH U07616 Amphiphysin 0.77 0.10 0.89 0.62 NRGN X99076 Neurogranin 0.80 0.10 0.84 0.09 RAB3A BC011782 Ras-related GTP-binding protein 3A 0.75 0.40 0.80 0.00* RAB4B AF217985 Ras-related GTP-binding protein 4B 0.54 0.01 0.79 0.01* TUBB2 NM_006088 Tubulin beta 2 0.60 0.01 0.79 0.06 TUBB4 BC000748 Tubulin beta 4 0.55 0.04 0.61 0.03* Synaptic enriched genes were decreased in MDD. The results from the microarray were confirmed by real time PCR. All of the genes of interest were normalized to cyclophilin A. The results are presented as the mean ± SEM, n = 15 (microarray), n = 8 (RT-PCR), P values represent comparison to controls (unpaired t-test).