Monomethylation of Lysine 27 at Histone 3 (H3K27me1) Confers Lifelong Susceptibility to Stress

SUMMARY

Histone post-translational modifications are critical for mediating persistent alterations in gene expression. By combining unbiased proteomics profiling and genome-wide approaches, we uncovered a role for mono-methylation of lysine 27 at histone H3 (H3K27me1) in the enduring effects of stress. Specifically, mice susceptible to early life stress (ELS) or chronic social defeat stress (CSDS) displayed increased H3K27me1 enrichment in the nucleus accumbens (NAc), a key brain-reward region. Stress-induced H3K27me1 accumulation occurred at genes that control neuronal excitability and was mediated by the VEFS domain of SUZ12, a core subunit of the polycomb repressive complex-2, which controls H3K27 methylation patterns. Viral VEFS expression changed the transcriptional profile of the NAc, led to social, emotional, and cognitive abnormalities, and altered excitability and synaptic transmission of NAc D1-medium spiny neurons. Together, we describe a novel function of H3K27me1 in the brain and demonstrate its role as a “chromatin scar” that mediates lifelong stress susceptibility.

Graphical Abstract

eTOC Blurb

Torres-Berrio et al. establish a novel function of a form of histone methylation – H3K27me1 – in nucleus accumbens, part of the brain’s reward circuitry, and demonstrate its role as a “chromatin scar” that mediates life-long behavioral, physiological, and transcriptional signatures of stress susceptibility in mice.

INTRODUCTION

A life history of stress is the strongest known risk factor for depression and anxiety, which are among the world’s leading causes of disability^1,2. Stress induces persistent transcriptional and cellular signatures in brain regions involved in reward and mood regulation^3-5. These effects can be observed long after stress exposure has ceased and are highly sensitive to windows of brain neuroplasticity, resulting in several behavioral abnormalities that range from heightened stress reactivity and social avoidance to cognitive dysfunction and anhedonia^2,6-8. Despite current efforts to uncover the mechanisms that underlie stress-induced phenotypes, there is still a lack of information about the molecular players that mediate these long-lasting abnormalities.

The nucleus accumbens (NAc), a core component of the brain’s reward circuitry, serves as a hub region that integrates information related to reward, motivation and complex cognitive function^9,10. By receiving inputs from multiple brain areas, the NAc not only fine-tunes behavioral outputs towards rewarding and away from aversive stimuli¹¹, but also determines appropriate responses to stressful events^3,12,13. These responses are shaped by two subtypes of medium spiny projection neurons (MSNs): D1 and D2 dopamine receptor-expressing cells (D1-MSNs and D2-MSNs)^14-16, which are segregated, both transcriptionally and electrophysiologically, and are critical for the development of susceptible vs resilient stress-related phenotypes^14-16.

Using genome-wide RNA-sequencing (RNA-seq) in postmortem brain tissue of individuals with depression and mouse models of chronic stress, our group has identified broad transcriptional changes in the NAc and its innervating regions^4,5,17. Interestingly, these changes are further exacerbated by a previous history of early life stress (ELS)^6,8, suggesting latent mechanisms that could promote or repress future transcriptional states. Converging evidence supports the hypothesis that histone post-translational modifications are critical for stress-induced changes in gene expression^2,6,18-20. Such regulation occurs through the addition or removal of methyl, acetyl or other groups at specific histone tail residues, which together orchestrate chromatin remodeling and ultimately dictate the availability of DNA for lasting transcription²¹. However, while certain histone modifications have been implicated in stress-related disorders, their influence is restricted to a relatively small number of genes and may not be recapitulated across different stress models^19,20. In this context, the use of unbiased proteomic approaches has become a powerful tool to unravel those histone modifications that are most strongly affected in NAc in response to stress and that may influence the broad transcriptional states in this brain region⁶.

Here, we combined mass spectrometry profiling, genome-wide approaches, and mouse models of stress to demonstrate the role of mono-methylation of lysine 27 at histone H3 (H3K27me1) in the NAc in conferring persistent vulnerability to stress. We report increased levels of H3K27me1 in the NAc of susceptible to chronic social defeat stress (CSDS), an effect recapitulated in adolescent and adult mice previously exposed to ELS. Stress-induced H3K27me1 enrichment in the NAc is associated with the induction of SUZ12, a core subunit of the polycomb repressive complex-2 (PRC2), which determines H3K27 methylation patterns²². Viral expression of the VEFS domain of SUZ12 in D1-MSNs mimics selective H3K27me1 induction and leads to behavioral, transcriptional, and electrophysiological signatures of stress susceptibility. This study thereby establishes a novel function of H3K27me1 in the brain and demonstrates its role as an important “chromatin scar” that mediates lifelong stress susceptibility.

RESULTS

Dynamic Methylation of H3K27 in Response to Social Stress

To identify global changes in histone modifications associated with stress susceptibility, we used mass spectrometry, an unbiased proteomic approach that allows for the quantification of hundreds of histone marks. We subjected adult male mice to CSDS²³, a validated model for the study of depression-like behaviors that separate mice into susceptible and resilient populations based on a social interaction test, which highly predicts other behavioral abnormalities²⁴. Bulk NAc tissue from control, susceptible and resilient male mice was collected 24 hours after the social interaction test and processed for mass spectrometry (Figure 1A-B).

Figure 1. H3K27 Dynamic Methylation in NAc following CSDS.

(A) Schematic and timeline of CSDS experiment. Susceptible (SUS), control (CON) and resilient (RES) mice. (B) Interaction ratio: One-way ANOVA: F_(2,13)=17.32, p=0.0002. Tukey’s test: ***p<0.001 (n=5-6 mice/group). (C) Heatmap for relative abundance of H3.3K27 modifications. Arrows indicate significance (yellow, increase: ^††p<0.0001; blue, decrease: ^†p<0.001). (D) H3.3K27me1: One-way ANOVA: F_(2,13)=65.21; p<0.0001. Tukey’s test: ****p<0.0001. (E) H3.3K27me1 and interaction ratio negative correlation. (F) H3.3K27me2: One-way ANOVA: F_(2,13)=14.15, p=0.0005. Tukey’s test: Different from CON (**p<0.01) and RES (***p<0.001). (G) H3.3K27me2 and interaction ratio positive correlation. (H) H3K27me1 intensity in MSNs (n=3-4 bilateral sections; 4 mice/group). Left: Coronal section showing H3K27me1 (magenta) in Drd1- (cyan) and Drd2- (yellow) MSNs. Right: Two-way ANOVA: Group: F_(1,12)=17.87; p=0.0012. Sidak’s Group effect: **Different from CON in Drd1-MSNs, p<0.0036. (I) H3K27me2 intensity in MSNs. Left: Coronal section showing H3K27me2 (red) in Drd1- and Drd2-MSNs. Right: Two-way ANOVA: Group: F_(1,12)=4.07; p=0.07; Cell-type: F_(1,12)=0.156; p=0.69, or interaction: F_(1,12)=0.15; p=0.705. Scale bar: 20μm. Data: Mean ± SEM.

Analysis of the relative abundance of single peptides revealed significant changes in a modest number of histone modifications in the NAc of susceptible and resilient mice, including mono-, di- or trimethylation (me1, me2, or me3) at lysines K9, K27, and K36 of H3 (Figure S1A; Table S1-2). Strikingly, susceptible mice displayed the greatest alterations of the methylation dynamics of K27 in the histone variant H3.3 (Figure 1C), the dominant form of H3 in adult brain neurons^25-27. Specifically, there was an elevated H3.3K27me1 relative abundance (Figure 1D), and a corresponding decrease of H3.3K27me2, in susceptible mice (Figure 1F). Furthermore, the abundance of both histone marks was highly correlated, albeit in opposite directions, with social interaction (Figure 1E-G). By contrast, H3.3K27me3, a repressive mark, and H3.3K27ac, an active enhancer mark, remained unchanged between groups (Figure 1C). Moreover, stress regulation of H3.3K27me1 and -me2 was male specific, as we did not observed differences in CSDS-exposed females (Figure S1B-F; Table S3-4). Therefore, we focused on male mice moving forward.

Because D1- vs D2-MSNs in NAc subserve distinct functions^14-16, we next assessed whether CSDS-induced changes in these histone marks are specific to either cell type. For this, we used H3K27me1 and H3K27me2 antibodies, which recognize K27me1 and K27me2 modifications in H3 and H3.3 (Figure S1G), and RNAscope probes to identify D1- or D2-MSNs in control and susceptible mice (Figure S1H). Susceptible mice exhibited increased H3K27me1 in D1-MSNs, with D2-MSNs showing a trend toward this effect (Figure 1H); whereas there was a non-significant trending decrease of H3K27me2 in both D1- and D2-MSNs (Figure 1I). These results confirm the opposite abundance of the two histone marks in NAc of CSDS-susceptible mice and indicate that H3K27me1 upregulation occurs primarily in D1-MSNs.

H3K27me1 Accumulates within Intragenic Regions of Genes Associated with Neuronal Excitability

To better understand the contribution of H3K27me1 and H3K27me2 to the effects of CSDS and determine their genomic distribution and putative target genes, we applied Cleavage Under Targets & Release Using Nuclease followed by DNA-sequencing (CUT&RUN-seq)²⁸ (Figure S2A-D) to NAc tissue from CSDS-exposed mice (Figure 2A-B). We first confirmed that H3K27me1 and H3K27me2 were deposited across non-overlapping sites within the mouse genome^29,30 at baseline. Indeed, the vast majority of H3K27me1 accumulated within gene bodies followed by promoters, whereas H3K27me2 was predominantly enriched across intergenic regions (Figure S2E-F).

Figure 2. Genome-wide Deposition of H3K27me1 and H3K27me2 in NAc under Control and CSDS Conditions.

(A) Schematic of CUT&RUN-seq experiment (n=4 samples, 4 mice/group pooled tissue). Antibodies validation in Figure S2. (B) Interaction ratio of CON, SUS and RES mice (n=16/group). One-way ANOVA: F_(2,45)=39.49; p<0.0001. Tukey’s test: ****p<0.0001. (C) Number of H3K27me1 and H3K27me2 differentially enriched (yellow) of depleted (blue) peaks in SUS vs CON, RES vs CON and SUS vs RES. (D) Enrichment heatmap for H3K27me1 differential peaks within ±1kb around the start (SS) and end (ES) sites in SUS vs CON. The blue-to-red gradient indicates low-to-high counts within specified regions. Heatmap for H3K27me2 in Figure S2H. (E) Genomic distribution of H3K27me1 differential peaks in SUS vs CON. (F). Gene ontology (GO) molecular functions of H3K27me1-enriched genes. (G) Integrative Genomics Viewer (IGV) tracks of ΔH3K27me1-enriched sites across introns 4, 5, 6 and 9 of the Gabrb2 gene in CON, SUS and RES. (H) Gabrb2 mRNA: One-way ANOVA: F_(2,20)=4.46, p<0.05. Tukey’s test: *p<0.05. Gabrg1 mRNA: One-way ANOVA: F_(2,20)= 3.7, p<0.05. Tukey’s test: *p<0.05. (I) GO molecular functions of H3K27me1-depleted genes. (J) IGV track of ΔH3K27me1-depleted sites within introns 1 and 9 of the Kdm4b gene. (K) Volcano plot of differential peaks in SUS vs RES. (L) H3K27me1 differential peaks distribution in SUS vs RES. (M) GO molecular functions of H3K27me1-enriched genes in SUS vs RES. (N) GO molecular functions of H3K27me1-depleted genes in SUS vs RES. Data: Mean ± SEM.

We next analyzed the differentially enriched or depleted sites of H3K27me1 and H3K27me2 in susceptible and resilient groups relative to controls. We found a greater number of H3K27me1 differential peaks in susceptible vs controls, with thousands of enriched (5254) and depleted (5180) sites (Figure 2C). By contrast, there was a much lower number of H3K27me2 differential sites in the susceptible comparison, than in resilient vs control (Figure 2C). The genomic distribution of H3K27me1 differential sites in the susceptible comparison revealed that ~90% of upregulated peaks were observed within intragenic or promoter regions, while ~30% of depleted sites occurred at promoters (Figure 2D-E). This suggests that some H3K27me1-enriched sites are more accessible than others to mechanisms that regulate K27 methylation³⁰. H3K27me2 changes, on the other hand, were mainly restricted to intergenic regions (~87%), with very little regulation seen within gene bodies or promoters (Figure S2G-H). We, therefore, focused on H3K27me1 for further downstream analyses.

Given the much larger number of H3K27me1 differential peaks in susceptible vs control, we performed gene ontology (GO) analysis to examine known molecular functions affected under these conditions. H3K27me1 accumulation was associated with terms such as GABA function, ion channel activity and overall neuronal excitability (Figure 2F, Table S5). Interestingly, some of the genes that encode for GABA receptor subunits are clustered within neighboring loci and exhibit high H3K27me1 deposition, with complete depletion of H3K27me2 (Figure S2F), indicating that stress-induced H3K27me1 accumulation may affect their transcriptional potential³¹. Indeed, expression of the beta 2 (Gabrb2) and gamma 1 (Gabrg1) subunits of the GABA_A receptor was higher in the NAc of susceptible mice (Figure 2G-H). By contrast, GO analysis of the gene displaying reduced H3K27me1 deposition in the susceptible comparison identified molecular functions involved in histone methyltransferase activity, and overall transcription and posttranscriptional regulation (Figure 2I), including several genes previously implicated in depression^32,33, such as lysine-specific histone demethylase-4B (Kdm4b, Figure 2J; Table S6).

Finally, we compared H3K27me1 differential peaks in susceptible vs resilient conditions. There was roughly the same number of H3K27me1-enriched (4358) and -depleted (4711) sites (Figure 2C) in susceptible vs resilient, while the genomic distribution of the sites revealed a similar pattern to the one observed in susceptible vs control conditions. Indeed, most (~80%) of the H3K27me1-enriched sites occurred intragenically (Figure 2M). GO analysis of H3K27me1 enrichment identified molecular functions involved in GABA-related channels, ion channel activity, and neuronal excitability (Figure 2N), whereas H3K27me1 depletion was associated with RNA regulation and catalytic activity on RNA (Figure 2O), which strikingly recapitulates the functions reported in the susceptible vs control comparison (Figure 2F; Figure S3). Collectively, our findings establish unique patterns of H3K27me1 regulation in the NAc of susceptible mice and indicate that H3K27me1 enrichment vs depletion dynamics affect different molecular functions and, in turn, contributes differentially to susceptible vs resilient phenotypes. Moreover, the finding of a similar number of sites at which H3K27me1 is enriched vs depleted in CSDS-susceptible mice, despite an increase in total tissue levels of H3K27me1, is consistent with numerous prior studies of other histone modifications⁶ and underscores the complexity of chromatin regulation and the importance of mapping histone genomic sites to understand the full range of their functional consequences.

VEFS Domain of SUZ12 Induces H3K27me1 in D1-MSNs and Confers Stress Susceptibility

Methylation dynamics of H3K27, including its H3.3 variant, are catalyzed by PRC2^34-36, a protein complex that displays preference for H3K27me1 and unmethylated substrates in vitro^37,38, and involves the coordinated action of three core proteins: suppressor of Zeste 12 (SUZ12), enhancer of Zeste 2 (EZH2), and embryonic ectoderm development (EED)^34,39,40, as well as accessory subunits such as RBBP4, AEBP2, and JARID2^22,41,42. To assess the role of PRC2 core subunits in stress susceptibility, we measured SUZ12, EZH2, and EED expression in the NAc of CSDS-exposed mice. SUZ12 was higher in susceptible mice compared to their control counterparts (Figure 3A), while no differences in EZH2 or EED were detected (Figure 3B-C). Interestingly, Suz12 mRNA expression was not different groups (Figure S4A), which suggests that post-transcriptional mechanisms elevate SUZ12 protein levels in the NAc of susceptible mice after CSDS.

Figure 3. VEFS Domain of SUZ12 Induces H3K27me1 in NAc D1-MSNs.

(A) SUZ12: One-Way ANOVA: F_(2,12)=3.62; p<0.05. Tukey’s test: *p<0.05. (B) EZH2: One-Way ANOVA: F_(2,15)=0.18; p=0.83. (C) EED: One-Way ANOVA: F_(2,14)=1.28; p=0.3. One empty well was left between samples (n=5-6/group). (D) Left: Schematic of the SUZ12 protein (yellow). Right: SUZ12 colored to highlight its interacting (VEFS, blue) and localization (ΔVEFS) domains. The VEFS domain determines SUZ12 and EZH2 interaction. Modified with permission from Højfeldt, et al. (2019)⁵⁸. Copyright Elsevier. (E) Schematic of VEFS expression viral construct. (F) Timeline of viral infection. (G) Amplification of VEFS or ΔVEFS domains (n=4/group). VEFS: t₍₄₎=3.95, p<0.05; ΔVEFS: t₍₆₎=0.57, p=0.58. (H) H3K27me1 expression in AAV-GFP-injected (top) or AAV-VEFS-injected (bottom) NAc D1-MSNs. Scale bars: 20μm. (I) Percentage of H3K27me1 fluorescence intensity in GFP-positive and GFP-negative neurons. Two-way ANOVA: Virus: F_(1,16)=6.43; p<0.05; Cell-type: F_(1,16)=4.52; p<0.05; Virus by Cell-type: F_(1,16)=4.52; p<0.05. Tukey’s test: *Different from AAV-GFP in GFP+ cells, p<0.05. (J) H3K27me1 abundance in AAV-VEFS: t₍₈₎=2.57; p<0.05 (n=5/group). (K) Enrichment heatmap for H3K27me1 differential peaks within ±1kb around the start (SS) and end (ES) sites in AAV-VEFS vs AAV-GFP (n=6 samples, 2 mice/group pooled tissue). The blue-to-red gradient indicates low-to-high counts within annotated regions. (L) Number of enriched (yellow) or depleted (blue) H3K27me1 peaks in AAV-VEFS vs AAV-GFP. (M) H3K27me1 differential peaks distribution in AAV-VEFS vs AAV-GFP. (N). GO molecular functions of H3K27me1-depleted genes. (O) IGV track of ΔH3K27me1 peaks across intron 8 of the Gabra4 gene. Data: Mean ± SEM.

In vitro studies have demonstrated that the C-terminal domain of SUZ12, known as the VEFS-BOX domain (Figure 3D), directly interacts with EZH2 and contributes to the catalytic actions of PRC2^34,39, while the ΔVEFS domain determines the precise localization of the complex⁴³. Notably, expression of the VEFS domain is sufficient to restore H3K27 methylation in SUZ12 knockout cells³⁰, and induce selective genome-wide H3K27me1 enrichment in vitro³⁰. We, therefore, developed a Cre-dependent and neuron-specific adeno-associated virus (AAV) construct expressing the VEFS domain of SUZ12 (AAV-VEFS, Figure 3E) to induce selective H3K27me1 accumulation in NAc D1-MSNs.

We microinfused the AAV-VEFS construct, or AAV-GFP control, into the NAc of adult D1-Cre mice and processed tissue for molecular and neuroanatomical validations (Figure 3F-G; Figure S4B-C). We observed that AAV-VEFS increased H3K27me1 (Figure 3H-I), but not H3K27me2 or H3K27me3 (Figure S4D-G), in D1-MSNs using immunofluorescence. Proteomic analysis confirmed greater H3K27me1 in AAV-VEFS-injected NAc tissue (Figure 3J; Figure S4H-J), an effect associated with reduced H3K27me3 (Figure S4N), indicating that VEFS expression affects H3K27 methylation dynamics in the NAc (Figure S4Q), without affecting other histone modifications (Table S7). By contrast, overexpression of full-length SUZ12 did not alter H3K27 methylation in D1-MSNs (Figure S5), which suggests that the VEFS domain is sufficient to selectively induce H3K27me1 in vivo³¹.

Next, we examined genome-wide changes of H3K27me1 following VEFS expression in the NAc using CUT&RUN-seq. Indeed, there was a stronger H3K27me1 peak signal and a larger number of H3K27me1-enriched sites in VEFS-expressing tissue compared to AAV-GFP (Figure 3K-L). Moreover, H3K27me1 accumulation occurred primarily within intragenic regions (~80%), followed by promoters and intergenic regions (Figure 3M); which confirms that VEFS expression induced H3K27me1 deposition at the expected sites. GO analysis of genes displaying H3K27me1 enrichment confirmed molecular functions related to inhibitory neurotransmission, or ion channel activity (Figure 3N-O), which resembles some functions reported for CSDS-susceptible mice (Figure S6).

To evaluate whether manipulating the VEFS domain confers stress-induced susceptibility, we exposed AAV-VEFS- or AAV-GFP-injected D1-Cre mice to subthreshold social defeat (SbD)²³, (Figure 4A). We chose SbD because this paradigm induces a rapid stress response but does not induce social avoidance to a CD1 aggressor in most adult wild-type mice²³. As anticipated, AAV-GFP-injected mice exposed to SbD displayed normal social interaction, evidenced by the interaction time with a novel CD1 mouse (social target) (Figure 4B). However, VEFS expression reduced the interaction time with a social target (Figure 4B), and increased the time spent in corners of the arena in mice subjected to SbD (Figure 4C), which led to a higher percentage of susceptible mice in this group (Figure 4D).

Figure 4. VEFS Domain of SUZ12 in NAc D1-MSNs Induces Susceptibility to Sub-Threshold Social Defeat Stress (SbD).

(A) Timeline for viral infection and SbD exposure. (B) Social interaction time. Three-way ANOVA: Virus: F_(1,68)=10.17; p<0.01; Session by Virus: F_(1,68)= 4.226; p<0.05. Tukey’s test: AAV-VEFS-SbD different from AAV-GFP-CON and AAV-GFP-SbD, *p<0.05 (C) Time in corners. Two-way ANOVA: Virus: F_(1,36)=7.22; p=0.01; Group: F_(1,36)=4.093; p<0.05. Sidak’s Virus effect: AAV-VEFS-SbD different from AAV-GFP-SbD, *p<0.05. (D) Percentage of SUS and RES mice per group. (E) Social interaction toward conspecifics. Three-way ANOVA: Social target (ST): F_(1,46)=81.89; p<0.0001; ST by Group: F_(1,22)=13.34; p<0.01, ST by Group by Virus: F_(1,22)=4.57; p<0.05. Tukey’s test: ST different from empty cage (EC) only in AAV-GFP-CON, AAV-VEFS-CON and AAV-GFP-SbD, ****p<0.0001. (F) Time in center. Two-way ANOVA: Virus: F_(1,36)=7.63; p=0.009; Group: F_(1,36)=4.091; p<0.05. Sidak’s Virus effect: AAV-VEFS-SbD different from AAV-GFP-CON, *p<0.05. (G-H) Acquisition of the reversal learning task (RLT). (G) Percentage of correct lever presses: Three-way ANOVA: Session: F_(2,111)=18.32; p<0.0001. Sidak’s Session effect: D6 different from D4, *p<0.05. (H) Sessions to learning: Two-way ANOVA: Non-significant Virus effect: F_(1,35)=0.033; p=0.85; Group: F_(1,35)=0.51; p=0.47, or interactions. (I-J) Reversal phase of the RLT. (I) Percentage of correct lever presses: Three-way ANOVA: Session: F_(4,180)=64.77; p<0.0001; Virus: F_(1,180)=51.36; p<0.0001; Session by Virus: F_(4,180)=3.084; p<0.05, Group by Virus: F_(1,180)=13.64; p<0.001. Tukey’s test: AAV-GFP-SbD different from AAV-GFP-CON in reversal day 4 (R4), *p<0.05. AAV-VEFS-CON different from AAV-GFP-SbD in R3 and R4, **p<0.01; AAV-VEFS-SbD different from AAV-GFP-CON in R5, *p<0.05 and AAV-VEFS-SbD different from AAV-GFP-SbD, during R3, R4 and R5, ****p<0.0001. (J) Sessions to reversal learning: Two-way ANOVA: Virus: F_(1,35)=8.16; p<0.01; Virus by Group: F_(1,35)=5.88; p<0.05. Tukey’s test: AAV-VEFS-SbD different from AAV-GFP-SbD, *p<0.01. Data: Mean ± SEM.

We next tested the effects of VEFS expression on social interaction towards a conspecific mouse, which provides a better indication of abnormal behavior than responses to CD1 aggressors typically used in CSDS protocols²². Thus, we placed experimental mice in an open field with one grid cage containing a conspecific juvenile mouse and one empty grid cage and compared the percentage of interaction time. Control mice infused with either AAV-GFP or AAV-VEFS, and SbD mice infused with AAV-GFP, displayed greater interaction towards a conspecific mouse compared to an empty cage. However, this social preference was reduced in AAV-VEFS mice exposed to SbD (Figure 4E). Moreover, this group exhibited lower social discrimination compared to their CON-VEFS counterpart (Figure S7B). Notably, there were no differences in the locomotor activity between groups (Figure S7C), but SbD mice infused with AAV-VEFS spent less time in the center area of an open field than AAV-GFP mice from the control group (Figure 4F), indicating an anxiogenic-like phenotype in this group.

VEFS Expression in NAc D1-MSNs Induces Stress Susceptibility in Complex Behaviors

Susceptible mice to CSDS, as defined by a simple social interaction test, exhibit deficits in cognitive flexibility when assessed in an operant reversal learning task (RLT) (Figure S8A-D). Moreover, chemogenetic-induced inactivation of NAc D1-MSNs impairs performance in this task (Figure S8E-H). We therefore assessed whether AAV-VEFS-injected mice subjected to SbD also displayed alterations in cognitive flexibility using the RLT. For this, SbD-subjected mice injected with either AAV-GFP or AAV-VEFS, and control counterparts, were trained to press one of two levers for a 0.2% saccharin-water reward (acquisition phase). Daily trials were given until all mice reached a learning criterion of 75% correct responses during three consecutive days (Figure 4G). As expected, there were no significant differences between groups during the acquisition phase, as all experimental mice took about the same number of sessions to achieve the learning criterion (Figure 4H). During the reversal phase, all experimental mice were trained to press the previously unrewarded lever to obtain the same saccharin reward. Interestingly, control mice injected with either AAV-GFP or AAV-VEFS exhibited similar reversal learning, with AAV-GFP-injected mice subjected to SbD displaying a higher percentage of correct responses during reversal days 3 and 4 compared to controls. This finding suggests an enhancing effect of SbD on reversal learning as seen previously for acute stressors^44,45. By contrast, AAV-VEFS mice exposed to SbD had lower correct responses by reversal day 5 relative to Control-GFP mice and by reversal days 3, 4, and 5, relative to AAV-GFP-injected SbD mice (Figure 4I). Indeed, AAV-VEFS-injected SbD mice took a greater number of sessions to complete the RLT than their AAV-GFP-injected SbD counterparts (Figure 4J). Overall, these findings indicate that VEFS expression rendered mice less prone to the cognitive enhancement induced by a subthreshold stressor.

We next assessed whether social alterations induced by VEFS expression can be observed in stress-naïve mice by testing a separate group of AAV-GFP- and AAV-VEFS-injected mice in an operant social reward task (Figure S9A). For this, experimental mice were trained to lever press for a saccharin reward, and subsequently trained to press a lever that activated a guillotine door and gave access to sensory, but not physical, interaction with a conspecific mouse. Mice from the AAV-GFP and AAV-VEFS groups displayed similar learning to lever press for a saccharin reward (Figure S9B). Interestingly, AAV-VEFS mice exhibited reduced motivation to lever press for social interaction as indicated by the lower number of breakpoints when evaluated on a progressive ratio schedule (Figure S9C). Together, our results from several behavioral procedures demonstrate that a selective increase of H3K27me1 in NAc D1-MSNs, as seen in susceptible mice after CSDS and induced experimentally via expression of the SUZ12 VEFS domain, exacerbates the emotional, social, and cognitive alterations induced by stress.

VEFS Expression Alters the Transcriptional Profiles of the NAc

To characterize the transcriptional signatures induced by VEFS expression in NAc, we processed tissue from AAV-GFP- and AAV-VEFS-injected mice exposed to SbD for RNA-seq. We identified a larger number of upregulated differentially expressed genes (DEGs) in AAV-VEFS-injected mice compared to AAV-GFP-injected mice (Figure 5A; Figure S10A-D), confirming that the VEFS manipulation regulates the transcriptional profiles of the NAc, and that stress exacerbates this impact (Figure S10E-F). GO analysis of VEFS-induced upregulated transcripts revealed molecular functions related to enzymatic activity and cytokine response (Figure 5B), consistent with pathways known to modulate SUZ12 and PRC2 activity⁴⁶.

Figure 5. VEFS Domain of SUZ12 Alters NAc Transcriptional Signatures.

(A) Volcano plot showing differentially expressed genes (DEGs) upon VEFS expression compared to GFP (DEG criteria: Log₂(Fold Change) >∣0.20∣, and p<0.05). Additional comparisons in Figure S10A-D (n=8 mice/group/virus). (B) GO molecular functions in VEFS-injected mice. (C-D) Rank-rank hypergeometric overlap (RRHO) plots showing coordinated transcription of H3K27me1-enriched genes (ΔH3K27me1-genes) between AAV-GFP-SbD vs AAV-GFP-CON and AAV-VEFS-SbD vs AAV-VEFS-CON (C) or between AAV-VEFS-CON vs AAV-GFP-CON and AAV-VEFS-SbD vs AAV-GFP-SbD (D). Arrows within quadrants indicate directionality of transcriptional overlap, and heat denotes strength of the overlap. ΔH3K27me1-genes selected according to the SUS vs CON comparison in Figure 2. (E) Union heatmap of differentially expressed ΔH3K27me1-genes in AAV-VEFS-SbD vs AAV-GFP-CON (bottom-row), ranked from lowest (blue) to highest (yellow) across comparisons. ΔH3K27me1-genes selected from the overlap shown in Figure S11C. Differential criteria: Log₂(Fold Change) >∣0.20∣, and p<0.05). (F) Heatmap of H3K27me1-enriched genes involved in neuronal excitability and synaptic transmission in AAV-GFP-SbD and AAV-VEFS-SbD compared to AAV-GFP-CON. *p<0.05, †p<0.01, pairwise comparisons.

We next compared the genes displaying H3K27me1 differential sites (ΔH3K27me1) in the NAc of CSDS-susceptible mice (Figure 2C) with their gene expression in AAV-VEFS- and AAV-GFP-injected mice subjected to SbD. For this, we first assessed the transcriptional concordance across ΔH3K27me1-genes in SbD relative to control (SbD vs CON) and in AAV-VEFS relative to AAV-GFP (VEFS vs GFP) through a threshold-free rank-rank hypergeometric overlap (RRHO) analysis. We found a weak effect of SbD on the transcriptional overlap between AAV-VEFS and AAV-GFP groups (Figure 5C), while AAV-VEFS induced a greater transcriptional similarity in control and SbD groups (Figure 5D). This finding is further supported by the observation that only 70 of 5,722 genes with H3K27me1 differential peaks displayed altered expression in AAV-SbD-GFP mice (Figure S11A), whereas 325 and 519 ΔH3K27me1-genes were altered in AAV-VEFS-CON or AAV-VEFS-SbD groups, respectively (Figure S11B-C), indicating that the VEFS domain primes gene expression in the NAc and exerts a strong regulation on ΔH3K27me1-genes following stress exposure. Indeed, there was robust gene upregulation in AAV-VEFS-SbD mice (Figure 5E; Figure S11D-E), an effect observed within genomic regions enriched with this mark (Figure S11F), and that involved voltage- and ligand-gated channels, such as potassium channels and subunits of GABA or glutamate receptors (Figure 5F; Figure S12). Overall, these findings support our hypothesis that stress-induced H3K27me1 enrichment leads to altered expression of genes that control neuronal excitability.

VEFS Expression Alters the Excitability and Synaptic Activity of D1-MSNs

Robust evidence demonstrates that susceptible mice display enhanced intrinsic excitability and altered inhibitory and excitatory synaptic activity in D1-MSNs of the NAc^13,15,47,48. Therefore, we hypothesized that VEFS expression would mimic stress-induced electrophysiological alterations in this cell type. We injected stress-naïve D1-Cre mice with either AAV-GFP or AAV-VEFS into the NAc and conducted whole-cell patch clamp recordings to examine the intrinsic excitability and miniature inhibitory and excitatory postsynaptic currents (mIPSCs and mEPSCs) of fluorescently-tagged neurons (Figure 6A). As expected, there was a dramatic increase in current-induced neuronal firing in AAV-VEFS-injected mice to AAV-GFP-injected mice (Figure 6B-D), an effect that associated with a decreased threshold to fire an action potential (Figure 6E), but with no change in resting membrane potential (Figure 6F). We also observed a small percentage (~8%) of AAV-VEFS-injected MSNs that fired spontaneously, while none of the AAV-GFP-injected MSNs exhibited spontaneous firing (Figure 6G)—consistent with the known lack of spontaneous firing of normal D1-MSNs in brain slices.

Figure 6. VEFS Domain of SUZ12 Alters Neuronal Excitability and Synaptic Transmission in NAc D1-MSNs.

(A) Schematic of electrophysiology recordings in D1-MSNs from AAV-GFP- and AAV-VEFS-injected mice (n=41-46 neurons/6–8 mice/virus). (B) Current-evoked spikes. Two-way ANOVA: Virus: F_(1,85)=6.86; p<0.01, Current: F_(30,2550)=127.4; p<0.0001, Virus by Current: F_(30,2550)=1.60; p=0.02. Sidak’s Virus effect: *Different from AAV-GFP; p<0.05. (C) Number of spikes at 200 pA current: t₍₈₅₎= 2.713; p<0.01. (D) Representatives of voltage traces. (E) Threshold potential: t₍₈₅₎=2.95; p<0.01. (F) Resting membrane potential: t₍₈₅₎=0.52; p=0.59. (G) Percentage of D1-MSNs exhibiting spontaneous firing. (H) Miniature inhibitory postsynaptic currents (mIPSCs) traces. (I) Cumulative fraction of mIPSC amplitude. Inset: Cells average: t₍₃₆₎=2.059; p<0.05. (J) Cumulative fraction of mIPSC interevent intervals (IEIs). Inset: Cells average: t₍₃₆₎=2.59; p<0.05. (K) Miniature excitatory postsynaptic currents (mEPSCs) traces. (L) Cumulative fraction of mEPSC amplitude. Inset: Cells average: t₍₄₉₎=1.602; p=0.11. (M) Cumulative fraction of mEPSC IEIs. Inset: Cells average: t₍₄₅₎=3.48; p<0.01. Data: Mean ± SEM.

To investigate functional changes in inhibitory and excitatory synapses induced by VEFS expression, we performed whole-cell patch clamp recordings of mIPSCs and mEPSCs in D1-MSNs of the NAc. Analysis of mIPSCs revealed increased amplitude (Figure 6H-I) and higher inter-event intervals (Figure 6J) in AAV-VEFS-expressing MSNs. By contrast, the amplitude of mEPSCs remained unchanged (Figure 6K-L), but there was a higher inter-event interval in AAV-VEFS-expression MSNs compared to GFP (Figure 6M). These results demonstrate that presynaptic release at both GABAergic and glutamatergic synapses is reduced by VEFS expression in MSNs which, along with the observed enhanced excitability in these neurons, support our sequencing and behavioral findings.

ELS Induces a Persistent Increase of H3K27me1 in the NAc

To evaluate whether changes in H3K27 methylation are observed across stress models and could serve as an epigenetic scar, we focused on ELS, a paradigm known to heighten stress vulnerability in adulthood^2,6-8 (Figure 7A). We used a recently generated mass spectrometry dataset of histone profiling in NAc tissue from mice exposed to ELS or their standard-raised (Std) counterparts⁶. Consistent with our results in CSDS-susceptible mice, we observed increased H3K27me1 abundance in the NAc of mice exposed to ELS. This effect was observed at PND21, immediately after ELS exposure, and lasted into adulthood, when H3K27me1 abundance is normally very low in Std-raised mice (Figure 7B). By contrast, we observed a significant decrease in H3K27me2 abundance at PND21 only (Figure 7C). These findings confirm our hypothesis that different forms of stress can induce H3K27me1 accumulation in NAc and that these effects can persist long after stressors have ceased.

Figure 7. Persistent H3K27me1 Increase in the NAc after ELS.

(A) Timeline of the ELS experiment. (B) H3K27me1 relative abundance. Two-way ANOVA: ELS: F_(1,8)=59.11; p<0.0001; Age: F_(1,8)=21.7; p<0.01; ELS by Age: F_(1,8)=9.349; p<0.05. Tukey’s test: ELS different from Std at PND21 (*p<0.05) and Adult (***p<0.001). Std in Adult different from Std at PND21, **p<0.01. (C) H3K27me2 relative abundance. Two-way ANOVA: ELS: F_(1,8)=7.675; p<0.05. Sidak’s ELS effect: ELS different from Std group at PND21, *p<0.05. (D) “Double-hit” paradigm: pups exposed to ELS and SbD in adulthood (n=8/group). (E) Social interaction. Two-way ANOVA: ELS: F_(1,28)=10.19, p<0.01; SbD: F_(1,28)=8.78, p<0.01; ELS by SbD: F_(1,28)=1.94, p=0.17. Sidak’s ELS effect: ELS-SbD different from Std-CON (***p<0.001), and ELS-CON (*p<0.05). (F) Number of differential enriched (yellow) of depleted (blue) H3K27me1 peaks in ELS-CON, Std-SbD, and ELS-SbD relative to Std-CON. (G) Enrichment heatmap for H3K27me1 differential peaks within ±1kb around the start (SS) and end (ES) sites in ELS-SbD vs Std-CON. Gradient blue-to-red color indicates low-to-high counts. (H) Average H3K27me1 density within differential peaks and their ±1kb flanking regions. (I) Volcano plot of ELS-CON vs ELS-Std differential peaks. (J) Volcano plot of ELS-CON vs ELS-Std differential peaks. (K) GO molecular functions in H3K27me1-enriched genes in ELS-SbD vs Std-CON. (L) IGV tracks of ΔH3K27me1 peaks across introns 1 and 3 of the Kcnn2 gene. (M) Enrichment heatmap for H3K27me1 in ELS-CON and ELS-SbD comparisons within H3K27me1 differential peaks in SUS vs CON. (N) Venn diagram showing H3K27me1 peak overlap in SUS vs CON and ELS-SbD vs Std-CON comparisons. (O) GO molecular functions of shared H3K27me1-enriched genes across stress models. Data: Mean ± SEM.

We next examined the effects of ELS on H3K27me1 deposition across the genome, and how this deposition is affected by a second hit of adult stress (Figure 7D). Pups were exposed to either ELS or Std conditions, and allowed to reach adulthood, when a subset of these mice was subjected to SbD. NAc tissue was collected 24 hours after the social interaction test and then processed for H3K27me1 CUT&RUN-seq. As expected, adult mice exposed to ELS alone or to SbD alone did not exhibit social avoidance, however, the two hits of stress (ELS-SbD) reduced social interaction compared to other conditions (Figure 7E), which indicates that ELS heightens vulnerability not only to adult exposure to chronic stress, as previously reported by our group^6-8, but also to subthreshold stressors.

We analyzed the differentially-enriched sites for H3K27me1 in the NAc of ELS-control, Std-SbD and ELS-SbD conditions relative to Std-controls. This approach revealed the largest number of differential peaks in the ELS-SbD vs Std-control comparison, and about the same number of peaks in mice exposed to ELS-control and Std-SbD conditions (Figure 7F). Therefore, exposure to ELS or SbD alone was insufficient to induce dramatic enrichment of H3K27me1, but the combination of the two stressors was effective at doing so. This finding was further confirmed by heatmaps, which showed the strongest H3K27me1 peak signals in the ELS-SbD vs Std-control comparison (Figure 7G-H), and robust depletion of H3K27me1 sites in ELS-control and Std-SbD compared to ELS-SbD (Figure 7I-J). Furthermore, GO analysis of H3K27me1-enriched genes in the ELS-SbD vs Std-control comparison revealed molecular terms related to neuronal activity (Figure 7K, Table S8), such as “Voltage-gated cation channel activity”, or “Potassium channel activity”, which indicates once again that genes involved in maintaining excitability of MSNs are a preferential target for H3K27me1 regulation in the NAc (Figure 7L).

Finally, we assessed whether deposition of this mark displayed common genes across early life and adult stress models by normalizing H3K27me1 differential enrichment across ELS-exposed mice to those genes identified in CSDS-susceptible mice (Figure 2). H3K27me1 enrichment predominated in the ELS-SbD condition (Figure 7M), and there was a significant overlap between the genes identified in the susceptible and ELS-SbD comparisons (Figure 7N), indicating that H3K27me1 accumulation at specific gene sites, including those encoding GABA or glutamate receptor subunits (Figure 7O), mediates the enduring effects of stress. Moreover, we observed a similar H3K27me1 enrichment profile in CSDS-susceptible mice relative to ELS-exposed mice (Figure S13). Collectively, our results demonstrate that H3K27me1 enrichment at genes involved in maintaining excitability and synaptic transmission is critical for conferring life-long stress susceptibility.

DISCUSSION

H3K27me3 has been widely explored and consistently associated with gene repression due to its preferential accumulation across heterochromatin sites and at promoter regions of silenced genes^21,49. By contrast, the roles of H3K27me1 and H3K27me2 have been poorly studied in brain despite their higher occurrence compared to H3K27me3²⁹, and little is known about the mechanism by which these marks influence transcriptional states, either across development or in response to environmental challenges. Here, we identified H3K27me1 as the most prominent histone modification induced in the NAc by CSDS and ELS, two extensively characterized mouse models of chronic stress that induce long-lasting behavioral abnormalities^7,24, and revealed the role of the SUZ12 VEFS domain as a critical mediator. Indeed, VEFS expression in D1-MSNs induced H3K27me1 accumulation, which enabled us to directly implicate this lasting histone modification in the behavioral, transcriptional and electrophysiological abnormalities induced by chronic stress. Our work is, therefore, the first study to establish a novel function of H3K27me1 in brain and supports the hypothesis that H3K27me1 in NAc D1-MSNs represents a chromatin scar that mediates stress susceptibility across the lifespan.

Evidence from in vitro studies shows that H3K27me1 effects on gene expression is complex and depends on its precise location within genomic regions or the cell type where it is enriched^29,34,50,51. For example, H3K27me1 is deposited over intragenic regions of active genes in neuroblastoma cells⁵¹, lymphocytes⁵² and mouse embryonic stem cells^29,34, whereas depletion of this mark is observed at promoters of actively transcribed genes in HeLa cells⁵⁰. Using bulk NAc tissue, we generated the first dataset of genome-wide enrichment of H3K27me1 and H3K27me2 in adult brain and confirmed previous in vitro observations reporting the preferential intragenic and promoter deposition of H3K27me1²⁹. Stress susceptibility was characterized by the intragenic accumulation of H3K27me1 across genes related to neuronal excitability, such as GABA receptors and potassium channel subunits, both of which have consistently been linked to depression and anxiety^13,53-56. By contrast, H3K27me1 depletion within the same class of genes was a hallmark of stress resilience. Collectively, these findings suggest that stress alters the pattern of H3K27me1 accumulation within particular genomic regions and thereby influences different molecular functions and transcriptional signatures. This may indicate that specific H3K27me1-enriched sites are more accessible to the activity of PRC2 and, in turn, may bias its regulatory actions towards those sites at the expense of other regions^30,37; therefore, the dynamic of enriched and depleted sites can be a critical factor in determining susceptible vs resilient phenotypes.

The pattern of H3K27 methylation can be shaped by the enrichment of SUZ12 at precise target regions, a process mediated by its C-terminal VEFS and N-terminal (ΔVEFS) domains^{30,34,36,43,57}. Thus, while the VEFS domain is sufficient to recruit PRC2 for H3K27 methylation, the ΔVEFS domain determines SUZ12 precise localization by recognizing specific genomic regions. Furthermore, expression of the VEFS domain, but not ΔVEFS, restores all three H3K27 methylation states in SUZ12 knockout cells, but induces aberrant accumulation of H3K27me1 only across active gene bodies^30,40,58. Our viral-mediated manipulations revealed that VEFS expression alters H3K27 methylation dynamics by elevating the levels of H3K27me1, but not H3K27me2 or H3K27me3, in D1-MSNs of the NAc. Moreover, H3K27me1 enrichment was observed across genes associated with neuronal excitability and inhibitory and synaptic transmission, consistent with the molecular functions identified in susceptible mice after CSDS or ELS. Notably, VEFS expression led to a larger number of upregulated differentially-expressed transcripts, including those enriched for H3K27me1, and induced several behavioral and cellular signatures of stress susceptibility, such as increased social avoidance, impaired cognitive flexibility and altered intrinsic excitability and inhibitory and excitatory synaptic transmission in D1-MSNs^{13,15,16,47,48,59}. Intriguingly, these signatures were specific to the VEFS domain, since we did not observe stress-related phenotypes by the full-length of SUZ12 (Figure S5). Our observation that CSDS induces SUZ12 in NAc, but that SUZ12 overexpression does not per se yield the stress-induced pattern of H3K27me1 regulation or downstream behavioral abnormalities, suggests that other SUZ12-interacting factors operate in D1-MSNs of stressed mice to induce the observed phenotypes. An improved understanding of PRC2 functioning in the brain, and identification of the rich regulatory mechanisms that control its activity and genomic targeting, will be required to identify such additional stress-related factors. The PRC2-related subunits, RBBP4 or JARID2, may serve as potential regulators of SUZ12 function in response to stress^31,42. The polycomb-like protein PHF1B, shown to regulate Gabrb1 gene expression by affecting H3K27me1 binding at the Gabrb1 promoter, may also play a role³¹.

Robust evidence demonstrates that D1-MSNs and D2-MSNs display distinct molecular and electrophysiological signatures and contribute differentially to stress-related phenotypes^15,16,47. Our previous work has shown that stress occurring in early life shapes the transcriptional profiles of the brain’s reward circuitry^6-8. This transcriptional regulation is mediated, in part, by enrichment of H3K79me2 in D2-MSNs of the NAc, and the coordinated action of its writer, DOT1L, and eraser, KDM2B, enzymes⁶. Using the same mass spectrometry dataset as in⁶, we identified H3K27me1 as a persistent histone modification induced by ELS (Figure 7), which, along with our proteomics, RNAscope and viral-manipulation experiments, indicate that these changes predominate in D1-MSNs. Despite the inability to obtain cell-specific histone proteomics data due to the amount of tissue required, we hypothesize that stress induces these two distinct histone modifications in these two distinct MSN cell types and thereby causes cell-type-specific transcriptional and physiological regulation. Furthermore, given that we did not observe CSDS-induced changes in H3K79me2 in male and female adult mice, or in H3K27me1 in adult females (Figure S1), we hypothesize that the influence of stress on histone modifications is age-dependent, sex-specific, and highly sensitive to developmental windows of brain plasticity. In this context, certain histone marks may be more likely to be enriched or depleted with high spatial and temporal specificity as compared to others, an effect that can be modulated by sex-specific actions of the PRC2 complex. Much remains unknown about whether stress-induced changes in H3K27me1 in D1-MSNs may lead to specific crosstalk between histone modifications⁶⁰, thus, facilitating the generation or depletion of other marks and, as a result, altering the epigenetic landscape in other cell types of the NAc. This is a provocative question given previous reports in cultured cells showing that, in the absence of H3K27me2, there is aberrant deposition of H3K27me1 at intragenic regions of poorly expressed genes and a global, but diffuse, increase in H3K27ac accumulation along the genome^29,40. Our observation that VEFS expression disrupted H3K27 methylation dynamics, shown via proteomics, and that stress depletes H3K27me1 within genes involved in histone methyltransferase activity, supports this idea.

In summary, our work establishes a novel function of H3K27me1 as a chromatin scar that mediates lifelong stress susceptibility and unravels the genome-wide targets for this mark in brain. We also introduce the VEFS domain of SUZ12 in driving the aberrant deposition of H3K27me1 in response to several forms of stress. The fact that epigenetic alterations can be reversible highlights the potential of the SUZ12 VEFS domain, and its interactions with other core proteins of PRC2, as potential targets for epigenetic therapeutics in stress-related disorders.

START Methods

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contacts, Dr. Eric J. Nestler (eric.nestler@mssm.edu).

Materials availability

The new reagents generated by this work (AAV-VEFS and AAV-SUZ12) will be made freely available to the scientific community upon request.

Data and code availability

Mass spectrometry data supporting the findings of this study can be found within the supplementary information of this paper. CUT&RUN-seq data, and RNA-seq data are deposited in NCBI’s Gene Expression Omnibus (GSE233419). Accession numbers are listed in the key resources table. This study did not generate original code.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-H3K27me1 - CUT&RUN-seq	Active Motif	Cat#61015
Anti-H3K27me1 - Immunofluorescence, RNAscope	Millipore Sigma	Cat#07-448
Anti-H3K27me2 - CUT&RUN-seq, Immunofluorescence, RNAscope	Cell Signaling	Cat#9728
Anti-H3K27me3 - Immunofluorescence	Cell Signaling	Cat#9733
IgG - CUT&RUN-seq	Cell Signaling	Cat#2729
Anti-GFP - Immunofluorescence	Aves Labs	Cat#GFP-1010
Anti-SUZ-12 - Western blot	Abcam	Cat#ab126577
Anti-EZH2- Western blot	Thermo Scientific	Cat#36-6300
Anti-EED - Western blot	Cell Signaling	Cat#85322
Anti-TATA-binding protein (TBP) - Western blot	Abcam	Cat#mAbcam 51841
Bacterial and virus strains
AAV9-hSyn-DIO-VEFS-eGFP	UMB Virus Vector Core	N/A
AAV9-hSyn-DIO-SUZ-12-eGFP	UMB Virus Vector Core	N/A
AAV9-hSyn-DIO-eGFP	UMB Virus Vector Core	N/A
pAAV-hSyn-DIO-hM3D(Gq)-mCherry	Addgene	Plasmid #44361
pAAV-hSyn-DIO-hM4D(Gi)-mCherry	Addgene	Plasmid #44362
pAAV-hSyn-DIO-mCherry	Addgene	Plasmid #50459
Chemicals, peptides, and recombinant proteins
Ketamine	Vedco	NDC 50989-996-06
Xylazine	AnaSed	NDC14043-700-50
Tetradotoxin	Tocris	Cat#1078
Gabazine	Tocris	Cat#1262
CNQX	Tocris	Cat#1045
Pentobarbital Sodium	Vortech	Cat#9373
Clozapine N-oxide	Tocris Bioscience	Cat#4936
BioMagPlus Concanavalin A beads	Polysciences	Cat#86057-3
SYBR green	Applied Biosystems	Cat#A25741
RNase-Free DNase Set (50)	Qiagen	Cat#79254
Microcystin	Enzo Biochem	Cat#ALX-350-012
Micrococcal nuclease	EpiCypher	Cat#15-1016
Trypsin	Promega	Cat#VA9000
2-methylbutane	Fisher Scientific	Cat#M32631
Co-Detection Target Retrieval Solution	Advanced Cell Diagnostics	Cat#323165
8-16% SDS-PAGE	Bio-Rad	Cat#4561103
0.5 mM EDTA, pH 8.0 solution	ThermoFisher	Cat#AM9260G
UltraPure (1M) Tris-HCl, pH 8.0	ThermoFisher	Cat#15568025
Blocker BSA	ThermoFisher	Cat#37520
RNase A, DNase and protease-free	ThermoFisher	Cat#EN0531
UltraPure Phenol:Chloroform:Isoamyl Alcohol	ThermoFisher	Cat#15593031
Proteinase K	ThermoFisher	Cat#EO0491
KCl (2M), RNase-free	ThermoFisher	Cat#AM9640G
NaCl (5M), RNase-free	ThermoFisher	Cat#AM9760G
MgCl2 (1M)	ThermoFisher	Cat#AM9530G
Triton X-100 solution	Sigma-Aldrich	Cat#93443
Glycogen	Sigma-Aldrich	Cat#10901393001
Complete EDTA-free protease inhibitor	Sigma-Aldrich	Cat#11873580001
Spermidine	Sigma-Aldrich	Cat#S2501
Complete, Mini, EDTA-free Protease Inhibitor	Sigma-Aldrich	Cat#11836170001
DL-Dithiothreitol solution	Sigma-Aldrich	Cat#646563
HEPES potassium salt	Sigma-Aldrich	Cat#H0527
HEPES sodium salt solution	Sigma-Aldrich	Cat#H3662
Trichloroacetic acid	Sigma-Aldrich	Cat#T0699
Saccharin	Sigma-Aldrich	Cat#109185
AEBSF solution	Sigma-Aldrich	Cat#SBR00015
Sodium butyrate	Sigma-Aldrich	Cat#B5887
IGEPAL	Sigma-Aldrich	Cat#I8896
Manganese (II) chloride solution	Sigma-Aldrich	Cat#M1787
Calcium chloride solution	Sigma-Aldrich	Cat#21115
Critical commercial assays
NEBNext Ultra II DNA Library Prep Kit	New England Biolabs	Cat#E7645
NEBNext® Multiplex Oligos for Illumina - Dual Index Primers Set 1	New England Biolabs	Cat#E7600S
SMARTer Stranded Total RNA-Seq Kit v3 - Pico Input Mammalian	Takara Bio USA	Cat#634487
SMARTer® RNA Unique Dual Index Kit – 96U Set A	Takara Bio USA	Cat# 634452
Multiplex Fluorescent Reagent Kit v2 protocol	Advanced Cell Diagnostics	Cat#323100
SNAP-ChIP K-MetStat Panel	EpiCypher	Cat#19-1001
NucleoSpin Gel and PCR Clean-Up Kit	Macherey-Nagel	Cat#740609
miRNeasy Micro Kit	Qiagen	Cat#217084
RNeasy Mini Kit	Qiagen	Cat# 74104
High-Capacity cDNA Reverse Transcription Kit	Applied Biosystems	Cat#4374967
High Sensitivity DNA Bioanalyzer assay	Agilent	Cat#5067-4626
Qubit dsDNA HS assay kit	Thermo Scientific	Cat#Q32854
High Sensitivity D1000 ScreenTape	Agilent	Cat#5067-5584
High Sensitivity RNA ScreenTape analysis	Agilent	Cat#5067-5579
Deposited data
Raw and analyzed data - CUT&RUN adult study	This paper	GEO: GSE233419
Raw and analyzed data - CUT&RUN ELS study	This paper	GEO: GSE233419
Raw and analyzed data – RNA-seq AAV-VEFS study	This paper	GEO: GSE233419
Raw and analyzed data - CUT&RUN AAV-VEFS study	This paper	GEO: GSE233419
Experimental models: Organisms/strains
Mouse: C57BL/6 wild-type mice	The Jackson Laboratory	N/A
Mouse: Drd1a-Cre hemizygote bacterial artificial chromosome transgenic	The Jackson Laboratory	Line FK150
Mouse: ERα-Cre Esr1-Cre knockin mice	The Jackson Laboratory	Line: 017911
Mouse: CD-1 retired breeder	Charles River Laboratories	N/A
Oligonucleotides
Primers for the VEFS-BOX subunit: Forward: 5’-GGA TCC TGA ATG GCT AAG AG-3’ Reverse: 5’-TAC AAA CAG CAT ACA GGC AT-3’	This paper	N/A
Primers for the ΔVEFS-BOX subunit: Forward: 5’-TGG TGA AGT CTT ACT CGT TG-3’ Reverse: 5’-AAT GTC TTT TCC CCA TCC TC-3’	This paper	N/A
Rab7 – Intron 3 Forward: 5’-AGT GGA GTA GTG ATG GCT GT-3’ Reverse: 5’- CGG CTT CAC TAT GAT GGC TT-3’	This paper	N/A
Primers for Gaphd: Forward: 5’-AGG TCG GTG TGA ACG GAT TTG-3’ Reverse: 5’-TGT AGA CCA TGT AGT TGA GGT CA-3’	This paper	N/A
TaqMan probe Gabrg1	Applied Biosystems	Mm00439047_m1
TaqMan probe Gabrb2	Applied Biosystems	Mm00433467_m1
TaqMan probe Gapdh	Applied Biosystems	Mm99999915_g1
Software and algorithms
ImageJ	Schneider et al. 65	https://imagej.nih.gov/ij/
EpiProfile 2.0	Yuan et al. 64	https://github.com/zfyuan/EpiProfile2.0_Family
Bowtie2	Langmead and Salzberg. 73	http://bowtie-bio.sourceforge.net/bowtie2/index.shtml
Samtools	Li et al. 74	http://samtools.sourceforge.net/
Bedtools	Quinlan et al. 77	https://github.com/arq5x/bedtools2
Cutadapt	Martin. 72	https://cutadapt.readthedocs.io/en/stable/
diffReps	Shen et al. 78	https://github.com/shenlab-sinai/diffreps
Region Analysis program	Shen et al. 78	https://github.com/shenlab-sinai/region_analysis
deepTools2	Ramírez et al. 79	https://deeptools.readthedocs.io/en/develop/
HISAT2 program	Kim et al. 67	https://ftp.ensembl.org/pub/release-90/fasta/mus_musculus/dna/
DESeq2 package version 1.28.1	Love et al. 68	http://www.bioconductor.org/packages/release/bioc/html/DESeq2.html
Rank-rank Hypergeometric Overlap (RRHO)	Plaisier et al. 71	github.com/RRHO2/RRHO2
IGV	Robinson et al. 80	https://software.broadinstitute.org/software/igv/trackline
MED-PC IV	Med Associates	N/A
EthoVision XT 11	Noldus	N/A
pClamp 11	Molecular Devices	N/A
MultiClamp 700A	Axon Instruments	N/A
BioRender	BioRender	https://www.biorender.com/

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Animals

Experimental procedures were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC) at the Icahn School of Medicine at Mount Sinai. All male and female mice used in these studies were maintained on a 12 hour light-dark cycle (lights on at 7:00) with ad libitum access to food and water throughout the experiments. All data derived from animal studies were analyzed by an experimenter blind to the conditions.

Male and female C57BL/6J wild-type mice (Postnatal day, PND, 75±15, Jackson Laboratory) served as experimental subjects in the social defeat stress paradigms. Male and female mice were group-housed before exposure to the stress procedures and single-housed at the completion of the last defeat session and before the social interaction test (SIT). Animals were randomized by cage prior to exposure to stress or before stereotaxic surgeries (e.g., each cage was assigned to each experimental condition or treatment: control or manipulation). The order of the animals was further randomized prior to behavioral tests.

Male CD-1 retired breeder mice (≥3 months old) previously screened for aggressive behavior were used as social aggressors for male stress exposure. CD-1 mice were obtained from Charles River Laboratories and were single-housed throughout the study.

ERα-Cre transgenic mice: Male ERα-Cre (Line: 017911, B6N.129S6(Cg)-Esr1tm1.1(cre)And/J0 were obtained from Jackson laboratory, and were crossed with CD-1 females to obtain F1 males, which were used as aggressors for female CSDS⁶¹.

D1-Cre transgenic mice: male Drd1a-Cre hemizygote bacterial artificial chromosome transgenic mice on a C57BL/6J background (Line FK150, http://www.gensat.org/cre.jsp) were bred in our colony room. Drd1a-Cre mice were group-housed until the beginning of social defeat experiments and single-housed at the completion of the last defeat session and before all behavioral testing and downstream experiments.

METHOD DETAILS

Stress Paradigms and Behavioral Tests

Chronic social defeat stress (CSDS):

CSDS was performed as in^23,62. Briefly, each adult male C57BL/6J experimental mouse was exposed to 5 minutes of physical aggression by a male CD-1 mouse. At the completion of the session, C57BL/6J experimental and CD-1 mice were housed overnight in a 2-compartment hamster cage and separated by a transparent divider with holes to provide sensory, but not physical, contact. The procedure was repeated for a total of 10 consecutive days, in which experimental mice faced a new aggressor every day. Control C57BL/6J mice were housed in 2-compartment cages with a different cage-mate every day. The CSDS protocol was conducted during the light cycle, between 11:00 and 14:00.

Female CSDS:

CSDS was performed as in⁶¹. In brief, females were exposed to 5 minutes of physical aggression by a male ERα-Cre mouse, for 10 consecutive days. During the CSDS, females were group-housed but single housed after the last defeat session.

Subthreshold social defeat (SbD):

SbD consisted of three sessions of social defeat occurring on a single day, as described in²³. In brief, adult C57BL/6J experimental mice were exposed to 5 minutes of physical aggression by a novel CD-1 mouse, and then housed with the same aggressor CD-1 mouse in a 2-compartment hamster cage for 15 minutes, followed by two additional defeat exposures. After the SbD sessions, C57BL/6J experimental mice were single-housed for 24 hours prior to the SIT.

Social interaction test (SIT):

Twenty-four hours after the last session of CSDS or SbD, C57BL/6J experimental mice were assessed in the SIT. This test consisted of 2 sessions in which defeated and control mice explored a squared-arena (44 x 44 cm) in the absence or presence of a novel aggressor CD-1 mouse (social target) for a period of 2.5 minutes each session. In the first session, an empty wire mesh enclosure 10 cm (length) × 6.5 cm (width) × 42 cm (height) was located against one of the walls of the arena to assess baseline exploration. In the second session, an unfamiliar CD-1 aggressor was placed inside the wire mesh enclosure. The area that surrounded the enclosure was designated as the social interaction zone (14 cm x 9 cm), whereas the corner areas of the walls opposite to the enclosure were designated as corners (9 x 9 cm) and represented the farthest point from the social interaction zone. The time (in seconds) of interaction with the social target was measured during both sessions. The social interaction ratio (time of interaction with social target present/the time in interaction zone with social target absent) was calculated to classify mice as susceptible (ratio<1) and resilient (ratio≥1) as in²³. This simple measure has been shown to correlate strongly with numerous other behavioral outcome measures.

Early life stress (ELS):

Two adult C57/BL6J wild-type female mice (PND 75±15) were mated with one adult male mouse from the same strain in our animal facilities. The male was removed after 1 week, and the females were separated into individual clean cages 2–3 days before the expected date of delivery (PND0). Litters were weighed and counted at birth and on PND7, and cages were cleaned on PND10, but were otherwise undisturbed. Only litters with 5 to 9 pups were included in this study, and the litter size was balanced across ELS and standard-raised conditions (Std).

The ELS paradigm consisted of a combination of both maternal separation, in which the entire litter was removed from the dam to a clean cage for 4 hours/day, and nesting material (crinkled Enviro-Dri paper) was reduced to one-fourth the regular amount during PND10 to PND17. We selected this developmental period based on our previous research which demonstrated enhanced susceptibility in adult mice exposed to this ELS paradigm without any overt behavioral abnormalities in the absence of this second hit of adult stress^6-8. Upon completion of the ELS protocol, nesting material was restored at PND17 and pups stayed with their dams until weaning at PND21. Std pups were monitored daily from PND10 to PND17, but cages contained regular amount of nesting material and pups were not separated from their dams.

Open field:

Open field test was measured in an open arena of 44 cm (length) x 44 cm (width) x 45 cm (height). The area surrounding the walls of the arena was designated as periphery and the central area of the arena was designated as center (16 x 16 cm). Mice were placed in the central area of the maze and allowed to explore for 600 seconds. The time spent in the central area vs periphery of the maze was recorded with an overhead video and analyzed online using EthoVision XT 11 (Noldus Leesburg).

Sociability test:

Sociability to a novel conspecific mouse was assessed in an open field (44 x 44 cm) which containing two enclosures placed in opposite corners of the arena. The area surrounding the enclosures (19 cm x 23 cm) was designated as interaction quadrants. During habituation, each experimental mouse was allowed to freely explore the open field and empty enclosures for 8 minutes. For the sociability test, a juvenile male conspecific (~4 weeks old) -a social target- was placed in one of the enclosures, while the other enclosure remained empty. The experimental mouse was allowed to explore the open field containing a social target on one side and an empty enclosure on the other side for 8 minutes. The percentage of time spent in each quadrant with the social target and the empty enclosure and the social discrimination index (time with social target minus time with empty enclosure divided by time with social target plus time with empty enclosure) were recorded.

Reversal learning task:

Reversal learning was conducted in mouse operant chambers (Interior dimensions: Interior: 55.69 cm x 38.1 cm x 40.64 cm; exterior dimensions: 63.5 cm x 43.18 cm x 44.45 cm, and walls: 1.9 cm from Med Associates (St. Albans). Operant chambers were enclosed in light and sound attenuating cubicles equipped with white house lights as well as fans to provide ventilation and to mask external noise. Each operant chamber contained two retractable levers, located on the right and left sides of a central reward magazine calibrated to deliver ~50 μl of liquid. Adult male mice were water deprived and given 4 hours of water access during the 3 days prior to the beginning of behavioral training. They received a single operant session every day and were given 2 hours of water access following each daily session throughout the course of the experiment. During the single 30 minutes pre-training session, mice explored the operant chamber and learned to introduce their noses into the central reward magazine to get 0.2% saccharin rewards, which were delivered every 60 seconds. Levers were retracted throughout this pre-training.

The reversal learning task consisted of acquisition and reversal phases. The acquisition phase was further divided into “ad libitum” and “restricted” sessions. The “ad libitum” sessions were given during the first three days of the acquisition phase and consisted of 30 minutes of free access to the active (correct) and inactive (incorrect) levers. Mice were allowed to freely press each lever, but only the active lever granted access to 10 seconds of 0.2% saccharin-water reward at a fixed-ratio 1 (FR1) schedule. During the “restricted” sessions, the active and inactive levers were available but mice had only 30 seconds to make a choice. Levers were immediately retracted once the choice was made and the reward was delivered if the correct lever was pressed. If mice did not press any of the levers during the 30 second trial or pressed the incorrect lever, they entered a 30 second timeout and an omission or an incorrect response was recorded. The number of correct and incorrect lever presses along with the number of earned rewards and response omissions were measured for each session, with a learning criterion of 75% correct presses for three consecutive days. During the reversal phase, only “restricted” sessions were given, in which the correct lever was switched, thus, all experimental mice would need to learn to press the previously inactive lever to obtain the 0.2% saccharin reward. The number of correct and incorrect lever presses along with the response omissions and the number of earned rewards were recorded, and a learning criterion of 75% correct presses for three consecutive days was used.

Operant social reward task:

An operant social reward task was conducted in the same mouse operant chambers described above with a few modifications. Specifically, each operant chamber contained two front retractable levers, located on the right and left sides of a central reward magazine calibrated to deliver ~50 μl of liquid, and two retractable levers located at the back of the chamber, opposite to the reward magazine, and next to a guillotine door (Figure 5A). During the first phase of the task, for four days, mice were trained to press one of the two front levers for 0.2% saccharin-water reward, as described above.

The social reward task was divided into two phases: social training at an FR1 schedule of reinforcement and a progressive ratio (PR) schedule. FR1 social training occurred over three consecutive days when mice were placed in the operant chambers for brief habituation during which the lights were off for 30 seconds. After habituation, lights were turned on and the guillotine door was lifted for 30 seconds to give access to a juvenile male mouse (social target, ~3-4 week-old C57Bl/6J) placed inside a chamber attached at the back of the operant chamber and separated from it with a wire mesh to allow sensory contacts, but no direct physical interaction. Following a period of 30 seconds, each experimental mouse was given 10 trials, separated by an intertrial interval (ITI) of 30 seconds, to press one of the two levers located at the back of the operant chamber. If the mouse pressed the active lever (social lever), the guillotine door was activated, the mouse was allowed to interact with the juvenile social target for 20 seconds, and a social response was recorded. If mice did not press any of the levers during the 30 second trial, they entered the ITI and an omission was recorded.

Social PR consisted of 1 hour sessions over five consecutive days, in which the active (social) and inactive levers were available, however, the number of responses required to activate the guillotine door was incremented following an exponential progression (R = 5 x e0.12*P - 5). Task performance was assessed by the reward-breakpoint during the PR session, defined as the maximum increment point at which each mouse stopped responding.

Tissue Dissection

For CSDS experiments, tissue was collected 24 hours after the SIT. In brief, experimental mice were euthanized by rapid decapitation. Brains were removed, and cooled with ice-cold PBS prior to slicing on a pre-defined brain matrix. Bilateral 14-gauge punches of the NAc were taken from 1 mm coronal sections starting on approximately plate 15 of the Paxinos & Franklin mouse atlas, frozen immediately on dry ice, and stored at −80°C until further use. For ELS experiments, tissue was collected at PND21 and adulthood. Bilateral 15-gauge punches were used for PND21 mice.

Mass Spectrometry

Nuclei isolation and histone purification:

Frozen tissue from the NAc was washed in a 2 ml pre-chilled dounce homogenizer containing 500 μl of nuclear isolation buffer (15 mM Tris, 60 mM KCl, 15 mM NaCl, 5 mM MgCl₂, 1 mM CaCl₂, 250 mM sucrose) with protease inhibitors and stabilizing agents (1 μl of 1 M DTT, 2.5 μl of 200 mM 4-benzenesulfonyl fluoride hydrochloride, AEBSF, 2 μl of 2.5 μM microcystin and 2 μl of 5 M sodium butyrate, for each 1 ml), and dissociated on ice in 500 μl of lysis buffer (nuclear isolation buffer + 0.2% IGEPAL) as previously described⁶³. For histone purification, nuclei were incubated in 0.2 M H₂SO₄ for 2 hours with constant rotation at 4°C, followed by centrifugation at 3,400 rcf for 5 minutes. Collected supernatant was added with chilled 33% trichloroacetic acid (Sigma-Aldrich), and left on ice at 4°C for 1 hour. After centrifugation at 3,400 rcf for 5 minutes, the supernatant was discarded and the histone pellet was rinsed first with 150 μl of ice-cold acetone + 0.1% HCl, and then with 150 μl of 100% ice-cold acetone. Histones were allowed to air-dry for 5 minutes, dissolved in 20 μl ddH₂O and stored at −80°C for further applications.

Histone derivatization and digestion:

Purified histones were treated with 20 μl of 50 mM NH₄HCO₃ (pH 8.0), and split into two aliquots for digestion into ArgC-like peptides (bottom-up) and intact histone N-terminal tails (middle-down). For the bottom-up preparation, samples were treated with 5 μl of acetonitrile followed by 5 μl of propionic anhydride and 14 μl of ammonium hydroxide and incubated for 20 minutes. Samples were then dried, resuspended in 20 μl of 50 mM NH₄HCO₃ and digested with Trypsin (Promega) at an enzyme:sample ratio 1:20 for 2 hours. The derivatization reaction was then performed again twice to derivatize peptide N-termini. Samples were then desalted by using in-house packed C18 Stage-tips and dried using a SpeedVac centrifuge prior to Liquid Chromatography–Mass Spectrometry (LC-MS/MS) analysis.

LC-MS/MS:

Samples were resuspended in 10 μl of water + 0.1% formic acid. A volume of 2 μl of histone peptide solution was injected onto a 75 μm ID x 25 cm Reprosil-Pur C18-AQ (Dr. Maisch Beim Brückle) nano-column packed in-house. The LC-MS setup consisted in a Dionex RSLC Ultimate 3000 (Thermo Scientific), coupled online with an Orbitrap Fusion Lumos (Thermo Scientific). The HPLC gradient was as follows: 2% to 28% solvent B (A = 0.1% formic acid; B = 95% MeCN, 0.1% formic acid) over 45 minutes, from 28% to 80% solvent B in 5 minutes, 80% B for 10 minutes at a flow rate of 300 nl/minute. The mass spectrometer was set to acquire spectra in a data-independent acquisition (DIA) mode. Briefly, the full MS scan was set to 300-1100 m/z in the orbitrap with a resolution of 120,000 (at 200 m/z) and an AGC target of 5x10e5. MS/MS was performed in the orbitrap with sequential isolation windows of 50 m/z with an AGC target of 2x10e5 and an HCD collision energy of 30 as described in⁶³. Extraction of the signal of the (un)modified peptides was performed by using EpiProfile 2.0, a computational platform that allows the accurate quantification of histone marks⁶⁴. From the extracted ion chromatogram, the area under the curve was obtained and used to estimate the abundance of each peptide. To achieve the relative abundance of post-translational modifications (PTMs), the sum of all different modified forms of a histone peptide was considered as 100% and the area of the particular peptide was divided by the total area for that histone peptide in all of its modified forms. The relative ratio of two isobaric forms was estimated by averaging the ratio for each fragment ion with different mass between the two species. The resulting peptide lists generated by EpiProfile 2.0 were exported to Microsoft Excel and further processed for a detailed analysis.

Neuroanatomical Experiments with Mouse Brain Tissue

Immunofluorescence:

Mice were anesthetized with a lethal dose of pentobarbital sodium (Vortech), and perfused intracardially with 0.9% saline, followed by 4% paraformaldehyde (PFA) in PBS (pH 7.4). Brains were removed and maintained in 4% PFA overnight. Coronal sections of the NAc were obtained at 30 μm using a vibratome (Leica VT1000S). For double-labeled immunofluorescence, NAc sections were blocked with 2% BSA + 0.2% Tween 20, and incubated overnight at 4°C with anti-H3K27me1 (1:500), anti-H3K27me2 antibodies (1:500), anti-H3K27me3 (1:500) antibodies, or Chicken anti-GFP antibody (1:1000). Immunostaining was visualized with Alexa 488-conjugated (Jackson Immunoresearch), and Cy3-conjugated (Jackson Immunoresearch) secondary antibodies raised in donkey.

RNAscope and immunofluorescence:

Mice were anesthetized with a lethal dose of pentobarbital sodium, and perfused intracardially with 50 ml of 0.9% saline, followed by 75 ml of ice-cold fixative solution (4% PFA in PBS; pH 7.4). Brains were immersed in 10%, 20% and 30% sucrose in 0.2 M PB at 4°C until sunk and flash frozen with cold 2-methylbutane (Fisher Scientific). Coronal sections of the NAc, spanning plates 15–24 of the Paxinos & Franklin mouse atlas, were obtained at 30 μm using a cryostat (Leica CM3050 S), mounted onto SuperFrost Plus (Fisher Scientific), and maintained in the −80°C freezer until processing. For immunofluorescence followed by RNAscope, tissue was processed using the Multiplex Fluorescent Reagent Kit v2 protocol (Advanced Cell Diagnostics), with minor modifications. In brief, brain sections were first acclimatized at −20°C for ~2 hours, and then at RT° for ~1 hour. Sections were incubated in 4% PFA for 30 minutes, rinsed three times with PBS and air-dried for at least 15 minutes. Slices were then baked in the HybEZ Oven at 60°C for 30 minutes, and dehydrated with xylene followed by ethanol 100%. Dried sections were treated with RNAscope hydrogen peroxide, and Co-Detection Target Retrieval Solution at 100°C for 15 minutes (Advanced Cell Diagnostics). After several washes with distilled water and PBS, sections were incubated overnight at 4°C in anti-H3K27me1 (1:300) or anti-H3K27me2 (1:300) antibodies diluted in Co-Detection Antibody Diluent. For RNAscope, tissue was washed and incubated in RNAscope Protease plus at 40°C for 30 minutes in the HybEZ Oven, followed by hybridization in Mm-Drd1a, and in Mm-Drd2 probes to label Drd1 and Drd2 MSNs, respectively. Slides were counterstained with DAPI for 5 minutes and cover-slipped with ProLong Diamond Antifade Mountant (Thermo Scientific). Images were taken within 2 weeks of staining on a Zeiss LSM780 confocal microscope and analyzed using a macro from ImageJ⁶⁵ to find intensity of Drd1, Drd2, H3K27me1 or H3K27me2 staining for each DAPI-positive nucleus. Images were captured at 512 × 512 pixels, and 140 × 140-pixel. Nuclei above the 75th percentile of Drd1 expression were deemed Drd1+ and likewise for Drd2+ nuclei. Oversaturated images were excluded from the analysis. The percentage of change in average H3K27me1 or H3K27me2 intensity was calculated for a total of four animals per group with at least three bilateral sections analyzed per animal.

Western Blotting

NAc tissue punches were processed for Western immunoblotting as before⁶⁶. Briefly, tissue samples were lysed using Radioimmunoprecipitation assay buffer (Thermo Scientific) with protein inhibitors (Roche) and sonicated with a Bioruptor system (Diagenode). Protein samples (30 μg) were separated on 8-16% SDS-PAGE (Bio-Rad) and transferred to a Trans-Blot Turbo 0.2 μm nitrocellulose membrane (Bio-Rad), followed by 2 hours incubation with 5% Non-fat dry milk blocking buffer and overnight incubation at 4°C with antibodies against SUZ12, EZH2, and EED (Key Resources Table). The TATA-binding protein (TBP) was used as loading control. A total of 5 to 6 biological samples were included per experimental condition with one empty well between samples to prevent bands from neighboring lanes to merge.

RNA Extraction and Quantitative Real-Time PCR

Total RNA was isolated from mouse frozen tissue with the miRNeasy Micro Kit protocol with minor modifications (Qiagen). All RNA samples were determined to have 260/280 and 260/230 values ≥1.8, using the NanoDrop One C system (Thermo Scientific). RNA integrity was assayed using an Agilent 2100 Bioanalyzer (Agilent). Reverse transcription was performed using High-Capacity cDNA Reverse Transcription kit (Applied Biosystems). Real-time PCR was carried out with SYBR green (Applied Biosystems) using an Applied Biosystems QuantStudio 5 Pro Real-Time PCR System. Data from target genes was analyzed by comparing C(t) values using the ΔΔC(t) method and Gapdh was used as reference gene (Key Resources Table). Real time PCR was run in technical triplicates.

Plasmids

Plasmids of full-length Suz12 and of the VEFS-BOX domain (VEFS), corresponding to the C-terminal portion of SUZ12, were kindly donated by Kristian Helin (Institute of Cancer Research, UK). The full-length Suz12 plasmid spans residues 1 to 739 and the VEFS plasmid spans residues 545 to 739 in the encoded proteins, as described in^30,58.

Viral Constructs

Viral constructs were obtained from the Virus Vector Core (VVC) of the University of Maryland. Adeno-associated virus (AAV9) expressing a Cre-dependent VEFS-eGFP (6.89 x 10¹¹ vg/ml) or SUZ12-eGFP (2.9 x 10¹² vg/ml) fusion protein under the control of the human synapsin 1 (hSYN) promoter were used to overexpress SUZ12 or VEFS in D1-MSNs selectively. A Cre-dependent construct fused to hSYN-eGFP (9.8 x 10¹¹ vg/ml) was used as control virus. Validation of the viral constructs was achieved via real time PCR using primers that recognize either the C-terminus domain of SUZ12 (VEFS) or the N-terminus domain (ΔVEFS) (Key Resources Table). For the experiment involving the silencing of D1-MSNs during the reversal learning task, the inhibitory DREADDs pAAV-hSyn-DIO-hM4D(Gi)-mCherry (7×10¹² vg/ml, Plasmid #44362) or the control pAAV-hSyn-DIO-mCherry (~7×10¹² vg/ml, Plasmid #50459) were obtained from Addgene. For female CSDS, the excitatory DREADDs pAAV-hSyn-DIO-hM3D(Gq)-mCherry (2.0×10¹² vg/ml, Plasmid #44361, Addgene) was used.

Stereotaxic Surgery

All surgeries were performed under aseptic conditions. Adult male D1-Cre mice were deeply anesthetized with an intraperitoneal injection of rodent cocktail (ketamine: 100 mg/kg and xylazine: 10 mg/kg, diluted in 0.9% saline) and placed in a stereotaxic apparatus (Kopf Instruments). Bilateral microinfusions were made using stainless-steel infusion cannulae (33 gauge) into the mouse NAc at the following coordinates: +1.6 mm (A/P), ±1.5 mm (M/L), −4.4 mm (D/V), and 10° angle relative to Bregma. For GFP, SUZ12 and VEFS overexpression experiments, a total volume of 0.5 μl of virus was delivered on each hemisphere over a 5 minutes period. The infusion cannulae were left inside the brain area during a 5 minutes pause to prevent virus reuptake. Mice recovered for 21 days before social defeat paradigms when transgene expression is maximal. For experiments involving DREADD-mediated inhibition, a total volume of 0.5 μl of either mCherry or hM4D(Gi)-mCherry viruses was injected into the NAc of each hemisphere, as described above. For aggressors used in female CSDS, ERα-Cre F1 mice were infused bilaterally with 0.5 μl of either mCherry or hM3D(Gq)-mCherry viruses into the ventromedial hypothalamus at the following coordinates: −1.5 mm (A/P), ±0.7 mm (M/L), −5.7 mm (D/V) from bregma. At the completion of the behavioral testing, experimental mice were euthanized for neuroanatomy and molecular experiments.

DREADD-Mediated Manipulation

For D1-MSNs inhibition, D1-Cre mice infused with either mCherry or hM4D(Gi)-mCherry were injected intraperitoneally with 1 mg kg⁻¹ of clozapine N-oxide (CNO, Tocris Bioscience) 20 minutes before each session of the reversal learning task. For female CSDS, ERα-Cre F1 mice infused with either mCherry or hM3D(Gq)-mCherry were injected intraperitoneally with 1 mg kg⁻¹ of clozapine N-oxide (CNO, Tocris Bioscience) 20 minutes before each defeat session.

Electrophysiology

Adult D1-Cre mice injected in the NAc with either GFP, VEFS or SUZ12 viruses were deeply anesthetized with isoflurane and decapitated. The brain was rapidly removed and chilled in artificial CSF (ACSF) containing: N-methyl-D-glucamine, HCl, KCl, NaH₂PO₄, NaHCO₃, HEPES, glucose, sodium ascorbate, thiourea, sodium pyruvate, MgSO₄, and CaCl₂, pH 7.4. The brain was embedded in 2% agarose and coronal slices (300 μm thick) were made using a Compresstome (Precisionary Instruments). Brain slices were allowed to recover at 33±1°C in ACSF for 30 minutes and thereafter at room temperature in holding ACSF containing: NaCl, KCl, NaH₂PO₄, NaHCO₃, HEPES, glucose, sodium ascorbate, thiourea, sodium pyruvate, MgSO₄, and CaCl₂, pH 7.4. After at least 1 hour of recovery, the slices were transferred to a submersion recording chamber and continuously perfused (2-4 ml/minute) with recording ACSF containing: NaCl, KCl, NaH₂PO₄, NaHCO₃, HEPES, glucose, MgSO₄, CaCl₂, 10 μM CNQX, and 100 μM gabazine, pH 7.4. All solutions were continuously bubbled with 95% O₂/5% CO₂. GFP-expressing MSNs were visually identified with infrared differential contrast optics (BX51; Olympus). Whole-cell patch-clamp recordings were performed at room temperature using a Multiclamp 700 A amplifier (Molecular Devices). All recordings and analyses were conducted in a blinded manner.

Intrinsic Excitability:

For excitability experiments, pipettes (3-5 MΩ) pulled from borosilicate glass were filled with solution containing: K-gluconate, HEPES, KCl, MgATP, MgCl₂, Na₂GTP, Na-phosphocreatine, and EGTA, pH 7.25. Data acquisition (filtered at 10 kHz and digitized at 10 kHz) and analysis were performed with pClamp 11 software (Molecular Devices). Cells were allowed to stabilize for 3 minutes after breakthrough in I=0 mode. Action potentials were evoked in current-clamp mode in response to 1 second depolarizing steps (30 steps in 10 pA intervals). For synaptic recordings, 100 μM gabazine and 1 μM TTX were added to ACSF at least 30 minutes before the recording. Only cells with stable input resistances were included in the analysis.

Miniature Postsynaptic Currents:

For recording miniature synaptic events, 10 μM CNQX + 1 μM TTX , or 100 μM gabazine + 1 μM TTX, were added at least 30 minutes before miniature inhibitory postsynaptic current (mIPSC) or miniature excitatory postsynaptic currents (mEPSC) recordings, respectively. The pipette solution contained: Cs-gluconate 122, HEPES 10, KCl 5, Mg ATP 5, Na₂GTP 0.5, QX314 1, and EGTA 1 mM, pH 7.25. Biocytin (2 mg/ml) was added to the pipette solution for post-recording immunohistochemical confirmation of the recorded cells.

Data acquisition (filtered at 10 kHz and digitized at 10 kHz) and analysis was performed with pClamp 11 software (Molecular Devices). Cells were allowed to stabilize for 3 minutes after breakthrough in I=0 mode. Action potentials were evoked in current-clamp mode by 1-s depolarizing steps (30 steps in 10 pA intervals). mEPSCs and mIPSCs were recorded in voltage-clamp mode at a holding potential of −90 mV and +10 mV respectively, for 5-10 minutes, beginning 3 minutes following the breakthrough. Only cells with stable input resistances were included in the analysis.

RNA-sequencing

Total RNA was isolated from frozen NAc tissue of AAV-GFP- and AAV-VEFS-injected D1-Cre mice exposed to subthreshold social defeat stress or control conditions with the RNeasy Mini Kit (Qiagen). All RNA samples were determined to have 260/280 and 260/230 values ≥1.8. RNA integrity (RIN) was assessed using the High Sensitivity RNA ScreenTape analysis (Agilent), and only samples with RIN values above 8 were used for RNA libraries. Libraries were prepared using a 7.5 ng concentration of purified RNA with the SMARTer Stranded Total RNA-Seq Kit v3 - Pico Input Mammalian (Takara Bio USA), according to the instructions from the manufacturer. The libraries were size selected and purified using AMPure XP beads (Beckman Coulter), and concentrations were measured by the High Sensitivity D1000 ScreenTape (Agilent). Libraries were submitted to sequencing with GENEWIZ/AZENTA (Azenta US Inc) on an Illumina HiSeq 4000 machine with a 2x150-bp paired-end read configuration at a minimum depth of 40 million paired-end reads per sample.

Analysis of RNA-seq data

Quality control was performed using FASTQC software (https://github.com/shenlab-sinai/NGS-Data-Charmer). Reads were aligned using the HISAT2 program⁶⁷, and count matrices were generated using the feature Counts function of the Subreads program (https://subread.sourceforge.net). For differential gene expression, raw reads were aligned to the mouse genome (mm10) and analyzed using the DESeq2 package⁶⁸. The following cutoffs were applied to identify differentially expressed genes: [Log₂(Fold Change)] >∣0.20∣ and a <0.05 p-value significance cutoff. Volcano plots were generated using the R package "EnhancedVolcano". Gene Ontology analysis was conducted with ShinyGO⁶⁹. The ontological categories for molecular function, cellular component and biological process were determined to be significant if they passed both the hypergeometric and binomial tests with a fold enrichment ≥1.5 and FDR Q<0.05. Significant GO terms with less than five genes were discarded. Venn Diagrams were generated using nVenn⁷⁰. P-values of Venn diagram area based on hypergeometric tests.

Rank-rank Hypergeometric Overlap (RRHO)

RRHO provides the ability to compare gene expression profiles of two conditions in a threshold-free manner to identify the degree and significance of overlap⁷¹. RRHOs were generated to compare transcriptomic overlap across DEG lists for two distinct experimental conditions. For example, SbD x CON compares DEG lists from AAV-GFP-SbD vs AAV-GFP-CON against DEG lists from AAV-VEFS-SbD vs AAV-VEFS-CON, whereas VEFS vs GFP compared AAV-VEFS-CON vs AAV-GFP-CON against AAV-VEFS-SbD vs AAV-GFP-SbD. RRHO plots were generated using the RRHO2 package with default settings (github.com/RRHO2/RRHO2).

Cleavage Under Targets & Release Using Nuclease (CUT&RUN)

The CUT&RUN procedure was conducted as previously described²⁸ with minor modifications.

Nuclei isolation:

Bulk NAc tissue punches were pooled from three to four mice from the same experimental group and homogenized with 1 ml lysis buffer (320 mM sucrose, 5 mM CaCl₂, 0.1 mM EDTA, 10 mM Tris-HCl, pH 8.0, 1 mM DTT, 0.1% Triton X-100, 1.5 mM spermidine) containing protease inhibitors (Roche). Nuclei were released using an ice-cold 2 ml dounce-homogenizer (30 strokes per pestle). The resulting homogenate was then filtered through a 40 μm cell strainer (pluriSelect) and transferred to an ultra-clear centrifuge tube (Beckman Coulter). A total of 6.5 ml of sucrose solution (1.8 M sucrose, 10 mM Tris-HCl, pH 8.0, 1 mM DTT, 1.5 mM spermidine, and protease inhibitors) was added underneath the nuclei homogenate for ultracentrifugation at 24,000 rpm for 1 hour at 4°C.

CUT&RUN:

Pellet containing nuclei was resuspended in 1 ml wash buffer (20 mM HEPES-NaOH, pH 7.5, 150 mM NaCl, 0.5 mM spermidine, 0.1% Triton X100, 0.1% Tween-20, 0.1% BSA, and protease inhibitors), and incubated with BioMagPlus Concanavalin A beads (Polysciences) in binding buffer (20 mM HEPES-KOH, pH 7.9, 10 mM KCl, 1 mM CaCl₂, and 1 mM MnCl₂) for 10 minutes. Nuclei-bound beads were reclaimed using a DynaMag Magnet (Thermo Scientific) and resuspended in antibody binding buffer (wash buffer with 2 mM EDTA) containing the indicated primary antibodies for an overnight incubation at 4°C. Antibodies for H3K27me1 (1:10), H3K27me2 (1:10) and IgG (1:10) were selected based on previous publications²⁹, and their specificity was confirmed with the SNAP-ChIP K-MetStat Panel (EpiCypher), using the primer set specific for each methylation state of H3K27 according to the manufacturer’s instructions (Figure S2A-C). After overnight incubation, nuclei-bound beads were washed with antibody binding buffer, and resuspended with 2.5 μl Micrococcal nuclease (EpiCypher) at 4°C for 1 hour. Beads were further rinsed with wash buffer, and low-salt buffer (20 mM HEPES-NaOH, pH 7.5, 0.5 mM spermidine, 0.1% Triton X100, 0.1% Tween-20, and protease inhibitor), followed by an incubation in ice-cold calcium buffer (3.5 mM HEPES-NaOH, pH 7.5, 10 mM CaCl₂, 0.1% Triton X-100, and 0.1 Tween-20) at 0°C for 5 minutes. The reaction was quenched by the addition of 200 ul EGTA-STOP buffer (170 mM NaCl, 20 mM EGTA, 0.1% Triton X-100, 0.1% Tween-20, 25 μg/ml RNase, and 20 μg/ml glycogen), and incubated at 37°C for 30 minutes in a Thermomixer at 500 rpm. Beads were centrifugated at 16,000 rcf for 5 minutes at 4°C, and the supernatant was collected for DNA clean-up using the NucleoSpin Gel and PCR Clean-Up kit (Macherey-Nagel).

Library preparation:

Sequencing libraries were prepared using the NEBNext Ultra II DNA Library Prep Kit (New England Biolabs), and NEBNext^® Multiplex Oligos for Illumina (Dual Index Primers Set 1) (New England Biolabs), according to the manufacturer's instructions. Libraries were quantified using Qubit dsDNA HS assay kit (Thermo Scientific) and the size distribution was determined by the High Sensitivity DNA Bioanalyzer assay (Agilent). Libraries were submitted to sequencing with GENEWIZ on an Illumina HiSeq 4000 machine with a 2x150-bp paired-end read configuration at a minimum depth of 40 million reads.

CUT&RUN analysis:

The NGS-Data-Charmer pipeline was used for preprocessing, QC and alignment of the FASTQ files (https://github.com/shenlab-sinai/NGS-Data-Charmer). In brief, initial adaptor trimming was achieved by using the Trim-Galore tool (v0.6.5) followed by a secondary trimming step through Cutadapt⁷² (v2.10). Reads of each histone modification were aligned to mouse mm10 genome assembly using Bowtie2⁷³ (v2.4.1), and duplicated read pairs were further removed using the ‘rmdup’ module of SAMtools^74,75 (v1.10).

Peak calling:

For each biological replicate and corresponding IgG control, peaks were called with MACS2⁷⁶ (v2.2.6) (macs2 callpeak -f BAMPE –p-value 0.05 --keep-dup). Region Analysis program⁷⁷ (https://github.com/shenlab-sinai/NGS-Data-Charmer/blob/master/Snakefile) was used to annotate peaks from MACS software. Differential region analysis between conditions was performed using diffReps package⁷⁸ (v1.55.4) with initial p-value settings of 0.0001. Detected differential regions merged and then further filtered with false discovery rate (FDR) of 10%. Annotation of called peaks and differential regions to their genomic features (promoters, gene bodies, intergenic, etc.) was performed using the Region Analysis program (https://github.com/shenlab-sinai/region_analysis).

Coverage heatmaps for read density across 1kb upstream and downstream of the peaks were obtained from combined bigwig files using Deeptools⁷⁹ through the ‘computeMatrix’ module and visualized with ‘plotHeatmap’ module. For the heatmaps, peaks were grouped according to their location in genomic features (promoters, gene bodies, intergenic, etc.). bigwig files were visualized with Integrative Genomic Viewer IGV tools (2.12.2)^80,81, and Venn diagrams were elaborated using nVenn⁷⁰

Union peak sets were created and converted to general feature format (GTF). Featurecounts extracted read counts for each of the peaks for both the H3K27me1 and the IgG bam files. The raw read counts were converted to CPM. For each treatment group (e.g., SUS) the CPM for IgG was subtracted from the CPM of the H3K27me1 sample. Negative values were then removed from these lists. To derive the relevant comparisons, a nominal value (0.0005) was added to each peak's IgG-removed CPM and the fold change of the sample was generated by dividing the experimental condition by the control condition (e.g., VEFS/GFP). This fold change was then transformed into Log₂ Fold Change. Union heatmaps were plotted using R.

Gene Ontology analysis was conducted with ShinyGO 0.77⁶⁹. The ontological categories for molecular function, cellular component, and biological process were determine to be significant if they passed both the hypergeometric and binomial tests with a fold enrichment ≥1.5 and FDR Q < 0.05. Significant GO terms with less than five genes were discarded.

QUANTIFICATION AND STATISTICAL ANALYSIS

Sample sizes (n) and statistical tests used are indicated in the figure legends. All values were represented mean ± S.E.M. Statistical analysis was performed using Graphpad Prism 6.0. A significance threshold of α<0.05 was used in all the experiments. Statistical differences between two groups were analyzed with Student’s t-tests with two-tailed analysis. Correlations were calculated using the Pearson correlation coefficient. Otherwise, one-way, two-way or three-way ANOVAs were performed, followed by Tukey’s or Sidak’s multiple comparison tests. Outliers were screened using the ROUT method. No statistical methods were used to determine the sample sizes, but the number of experimental subjects is similar to sample sizes routinely used in our laboratory and in the field for similar experiments. Data with normal distribution, and similar variance were analyzed with parametric statistics. Otherwise, nonparametric statistics were applied.

Supplementary Material

2

Table S1. Raw data of single posttranslational modifications in CON, SUS and RES male mice. Related to Figures 1C-G, and Figure S1A.

3

Table S2. Raw data of single and combinatorial peptides in CON, SUS and RES male mice. Related to Figures 1C-G, and Figure S1A.

4

Table S3. Raw data of single posttranslational modifications in CON, SUS and RES female mice. Related to Figures S1B-F.

5

Table S4. Raw data of single and combinatorial peptides in CON, SUS and RES female mice. Related to Figures S1B-F.

6

Table S7. Raw data of single posttranslational modifications in VEFS- vs GFP-injected mice. Related to Figure 3J, and Figure S4H-Q.

Highlights.

• Stress causes lifelong and cell-type-specific enrichment of H3K27me1 in nucleus accumbens

• H3K27me1 accumulates across genes involved in neuronal excitability and neurotransmission

• Expression of the VEFS domain of SUZ12 selectively induces H3K27me1 enrichment

• VEFS expression induces behavioral and transcriptional signatures of stress susceptibility

INCLUSION AND DIVERSITY

Several authors self-identify as underrepresented minorities and actively lead inclusion and diversity initiatives. We tirelessly work to increase minority representation in neuroscience and promote gender balance.