Sex-specific effects of the histone variant H2A.Z on fear memory, stress-enhanced fear learning and hypersensitivity to pain

Emerging evidence suggests that histone variants are novel epigenetic regulators of memory, whereby histone H2A.Z suppresses fear memory. However, it is not clear if altered fear memory can also modify risk for PTSD, and whether these effects differ in males and females. Using conditional-inducible H2A.Z knockout (cKO) mice, we showed that H2A.Z binding is higher in females and that H2A.Z cKO enhanced fear memory only in males. However, H2A.Z cKO improved memory on the non-aversive object-in-place task in both sexes, suggesting that H2A.Z suppresses non-stressful memory irrespective of sex. Given that risk for fear-related disorders, such as PTSD, is biased toward females, we examined whether H2A.Z cKO also has sex-specific effects on fear sensitization in the stress-enhanced fear learning (SEFL) model of PTSD, as well as associated changes in pain sensitivity. We found that H2A.Z cKO reduced stress-induced sensitization of fear learning and pain responses preferentially in female mice, indicating that the effects of H2A.Z depend on sex and the type of task, and are influenced by history of stress. These data suggest that H2A.Z may be a sex-specific epigenetic risk factor for PTSD susceptibility, with implications for developing sex-specific therapeutic interventions.

stressor, often used to induce SEFL, potentiates pain sensitivity to a similar extent in both sexes, but that this sensitization is achieved through distinct molecular mechanisms 16 , suggesting that unique biological risk factors may regulate sensitivity to stress-induced alterations in pain sensitivity.
Indeed, sex differences have been described for the cellular and molecular mechanisms of fear memory, stress, and pain responses. For fear circuitry, differences exist in synaptic function, circuit activity, and in the size and volume of hippocampus and amygdala subregions, even when memory performance is similar (reviewed in 17,18 ). Similarly, acute stress exposure results in widely distinct transcriptional responses in males and females 19 and stress exposure may sensitize pain responses through sex-specific mechanisms. For example, knocking out the nociceptin/orphanin FQ receptor (NOP) protects against stress-potentiated pain sensitivity only in males 16 , indicating that molecular underpinnings of stress-potentiated pain sensitivity may be sex-specific.
Epigenetic modifications emerged as strong candidate mechanisms for understanding stable behavioral adaptations, such as PTSD, because of their potential to persist over long periods of time and their vital role in the maintenance of enduring fear memories [20][21][22][23] . Indeed, epigenetic mechanisms are involved in adapting to stressors and are emerging as regulators of pain responses in human patients 24 , making them especially promising as potential links between fear memory, PTSD, and pain. Although the epigenetic basis of fear potentiation in SEFL is not clear, stressors utilized in SEFL paradigms result in extensive epigenetic changes across the genome, including genes with known roles in stress regulation and neural plasticity, such as the glucocorticoid-receptor encoding gene, as well as plasticity-related genes like Bdnf and the postsynaptic density gene, Dlgap2 [25][26][27][28] . Moreover, stressful experiences change transcriptional and epigenetic responses to subsequent stressors 29 , suggesting that stress exposure may sensitize fear learning at least in part through epigenetic factors.
Sex differences in the epigenetic basis of fear learning and SEFL have not been widely studied, but there is evidence for sex differences in the binding and expression of various epigenetic markers in the brain 30,31 , which have functional implications for behavior. For example, sex differences in fear extinction have been linked with sex-differences in epigenetic marks in the medial prefrontal cortex 32 and sex-specific differences in DNA methylation have been linked with risk for PTSD 33 . Together, these data indicate that sex differences in the epigenome may contribute to sex differences in fear memory and associated conditions, including PTSD.
Most epigenetic studies relating to stress and fear memory have focused on DNA methylation and posttranslational modifications of histones, but recent studies from our lab and others show that histone variants are a novel epigenetic regulators of memory that have a unique relevance in neurons 22,[34][35][36][37][38] . DNA is packaged into the nucleus by wrapping in 147 base pairs segments around an octamer of histone proteins (2 copies of histones H2A, H2B, H3 and H4) to form nucleosomes, the building blocks of chromatin. This packaging creates a physical barrier to transcription that is altered by post-translational modifications of histone proteins, but also by replacing canonical histones with their functionally and structurally distinct variants 39 . In contrast to replication-dependent canonical histones (H2A, H2B, H3, and H4), their non-allelic histone variants are replication-independent, and as such, they become the primary source of histones in post-mitotic neurons [38][39][40] . Histone variants are of particular interest in studies of behavioral regulation because of their higher potential level of selectivity as therapeutic targets. Specifically, variants represent individual protein targets that have a more sparse distribution compared to DNA methylation and post-translational modifications of histones, which are widely distributed across the genome and various histone types and as such, are much less specific targets. Indeed, H2A.Z represents approximately 10% of total H2A 41 and tends to be strongly positioned at nucleosomes flanking transcription start sites 34 , which theoretically makes this variant a more precise target for therapeutic intervention. Moreover, histone variants form an important part of the histone code through interactions with other epigenetic modifications, including a well characterized exclusion of H2A.Z from methylated DNA in nonneuronal cells 40,42 , indicating that a comprehensive understanding of epigenetic regulation of neural plasticity necessitates the study of histone variants. In contrast to post-translational modifications, which tend to be short lasting 21,43 , histone H2A.Z incorporation into nucleosomes is stably altered after fear learning 22 , suggesting that histone H2A.Z may also be relevant for regulating stable phenomena, such as PTSD.
H2A.Z is a variant of the canonical histone H2A that acts as a negative regulator of fear memory, whereby virally-mediated reduction of H2A.Z in the hippocampus of male mice enhances associative fear memory 34,35 . However, it is not clear whether this effect holds true in females, or if the effect of H2A.Z on memory is unique to fear memory. It is also not known whether improved memory observed with H2A.Z reduction represents a general improvement in associative learning, or if the loss of H2A.Z could be maladaptive by also potentiating fear-related symptoms in models of PTSD. To address these questions, the goal of the present study is to investigate sex differences in H2A.Z binding and functional differences in fear memory, SEFL, and associated changes in pain sensitivity using an inducible-conditional H2A.Z knockout mouse with selective loss of H2A.Z in CamKIIα-positive cells.

Results
H2A.Z binding is higher in female than in male mice. Previous studies have exclusively quantified H2A.Z binding in the hippocampus of male mice. With the aim of characterizing potential sex differences in H2A.Z occupancy, we utilized chromatin immunoprecipitation (ChIP) in area CA1 to assess H2A.Z binding on proximal promoters of several genes that previously showed evidence of H2A.Z binding in genome-wide studies conducted in the hippocampus 34 . To ensure that any differences in H2A.Z binding were not caused by differences in total nucleosome occupancy in male and female mice, all H2A.Z data were normalized to H3 levels at the same locus. There were no differences in H2A.Z binding on Fos, Arc, or Gadd45b, but females had significantly higher H2A.Z binding on beta-2-microglobulin (B2m; t 17 = 2.56, p = 0.02), FK506 Binding Protein 5 (Fkbp5; t 15 = 2.60, p = 0.02), and Tyrosine Hydroxylase (Th; t 17 = 2.48, p = 0.024), all of which are implicated in neural plasticity and memory [44][45][46] . There were no sex differences in H3 binding for any of the genes exam-Scientific RepoRtS | (2020) 10:14331 | https://doi.org/10.1038/s41598-020-71229-x www.nature.com/scientificreports/ ined (data not shown). These data suggest that H2A.Z exhibits gene-specific occupancy differences in male and female mice, with females having higher H2A.Z binding compared to males (Fig. 1).  (Fig. 2d). Note that H2A.Z deletion was restricted to CaMKIIα-positive cells, and as such, we do not expect a dramatic loss of H2A.Z binding in a mixed cell population. To exclude potential loss of nucleosomes as a source of reduced H2A.Z binding at these loci, we also conducted ChIP against histone H3 and found no changes in H3 binding at any of the above genes (Fig. 2e), suggesting that reduced H2A.Z binding occurs without a loss of nucleosomes. Females have higher H2A.Z binding than males in a gene-specific way. ChIP was conducted to assess binding of H2A.Z, normalized for H3 binding at the same loci, at several genes related to neural plasticity. There were no sex differences in H2A.Z binding at Fos, Arc, or Gadd45b, but females had significantly more H2A.Z at B2m, Fkbp5, and Th compared to males. Data are shown as Mean ± SEM, *p ≤ 0.05. N = 8-10/sex. www.nature.com/scientificreports/ H2A.Z has sex-specific effects on fear memory, but not on object-in-place memory. Given that depletion of H2afz alone enhanced fear memory in males 34,35 , we tested whether dual deletion of both H2A.Zencoding genes affects classical fear memory in male and female mice. Analysis of baseline (pre-training) freezing behavior before shock delivery found that control females froze 5 s more on average than control males (p = 0.001), whereas this differences was not seen in H2A.Z cKO mice (Sex × Genotype interaction: F 1,68 = 5.58, p = 0.02). The lack of a sex difference in H2A.Z cKO mice was driven by reduced freezing in H2A.Z cKO females compared to control females (p = 0.05), which brought H2A.Z-depleted females to the same baseline freezing level as males (Fig. 3a).

Male
To test the effects of H2A.Z cKO on fear memory, freezing behavior was assessed 24 h after a one-shock (0.5 mA) contextual fear conditioning paradigm. Effects of H2A.Z deletion depended both on sex and genotype (Sex × Genotype interaction: F 1,68 = 4.35, p = 0.04). Consistent with studies of H2afz depletion alone, males with dual H2A.Z cKO had enhanced fear memory compared to controls (t 45 = 3.30, p = 0.02), indicating that the loss of both H2A.Z encoding genes has similar effects on fear memory as loss of H2afz alone in male mice. In contrast, H2A.Z deletion had no influence on fear memory in females (Fig. 3b). Given that tamoxifen, which is used to induce Cre activity in H2A.Z cKO mice, acts on the estrogen receptor, we conducted additional experiments to exclude any potential influence of tamoxifen in masking the effects of H2A.Z deletion in females. To this end,   www.nature.com/scientificreports/ additional Cre-positive females were treated with oil (control mice) or tamoxifen (H2A.Z cKO) to control for potential effects of tamoxifen. However, this approach also failed to uncover an effect of H2A.Z cKO on fear memory in females, suggesting that H2A.Z has sex-specific effects on fear memory (Fig. 3c). Given that fear conditioning is a particularly robust form of learning, we investigated the possibility that a memory-enhancing effect of H2A.Z in females might be revealed under a weaker training paradigm. Objectin-place (OiP) is a particularly challenging memory task in which mice are familiarized with 4 objects that remain the same during training and recall, but 2 out of 4 objects change location during recall, such that the preferential exploration of objects that were moved compared to objects that remain in the same location is taken as an index of memory. All H2A.Z cKO mice performed better than controls irrespective of sex (Main effect of Genotype: F 1,54 = 20.29, p < 0.001), and females had better overall memory than males (Main effect of Sex: F 1,54 = 5.11, p = 0.028) (Fig. 3d). We also assessed the time spent exploring novel compared to familiar objects and found a significant interaction between genotype and novelty (F 1,53 = 20.31, p < 0.0001), whereby H2A.Z cKO, but not control mice, exhibited a significant difference between time spent interacting with objects in familiar vs novel locations (t 23 = 5.78, p < 0.0001), suggesting that only H2A.Z cKO mice exhibited significant learning. H2A.Z deletion improves memory for a non-aversive learning task in both sexes, suggesting that H2A.Z may act as a general memory suppressor in both sexes, but that it has sex-specific effects selectively for fearful stimuli.

H2A.Z has sex-specific effects on stress-enhanced fear learning (SEFL).
Given that H2A.Z selectively impacted fear memory and that there are extensive sex differences in fear-related disorders, such as PTSD, we next investigated whether sex differences in acute fear memory also extend to stress-enhanced fear learning (SEFL), an animal model of PTSD-like symptomatology. SEFL is a model of potentiated fear learning that occurs in PTSD patients and involves trauma-induced potentiation of subsequent fear learning in a novel context 7,9,47,48 . Following a one-hour session with 10 randomly distributed shocks in Context A (mimics traumatic experience), mice were fear conditioned the next day in Context B. We did not find any group differences before shock delivery in Context B (Fig. 4), suggesting that any carry-over effect from receiving shock in context A was not affected by sex or genotype. However, average freezing was ~ 30%, which is higher than the ~ 10% freezing in mice undergoing fear conditioning for the first time ( Fig. 3), suggesting that the 1 h SEFL paradigm did produce some degree of generalization to context B. However, this effect did not differ across groups and overall, freezing was ~ 45% lower than it was after training in context B. Specifically, during the 24 h recall test for Context B, H2A.Z cKO females (t 18 = 4.68, p < 0.0001), but not males, had reduced fear memory compared to controls (Genotype × Sex interaction: F 1,40 = 4.62, p = 0.04), although this difference was not evident after 7 days (Fig. 4). Thus, even though H2A.Z did not affect acute fear memory in females, it did inhibit potentiation of fear memory by prior stress, an effect that was not observed in males. Moreover, freezing decreased between 24 h and 7 day recall tests irrespective of sex or genotype, indicating that any extinction occurring between days was similar across conditions. Overall, these data suggest that H2A.Z regulates distinct aspects of fear learning in males and females: In males, H2A.Z regulates acute fear learning without altering stress-mediated fear sensitization, whereas the opposite is true in females.

H2A.Z and SEFL regulate the development of pain hypersensitivity. PTSD patients exhibit
altered pain sensitivity compared to the general population 12 , suggesting that sex-specific effects of H2A.Z on fear memory in SEFL-exposed mice may influence pain responses. As with fear memory, pain is subject to sensitization by a variety of noxious stimuli, including an intra-hind paw injection of complete Freund's adjuvant (CFA), which potentiates sensitivity to thermal and mechanical stimuli for several days 49 . As such, we used the CFA model of chronic inflammatory pain to compare effects of H2A.Z on sensitivity to thermal and mechanical stimulation in mice exposed to SEFL and non-stressed controls.
Mechanical withdrawal thresholds. CFA-induced hypersensitivity and recovery to mechanical stimuli was measured using the von Frey test before and over several days after CFA hind paw administration. In control males without SEFL exposure (Days, Sex × SEFL interaction: F 4,244 = 3.83, p = 0.005; Days × Genotype interaction: F 4,244 = 4.01, p = 0.004), CFA treatment reduced withdrawal thresholds to a mechanical stimulus on Days 1, 3, and 7 (all p < 0.05), with recovery occurring by day 10. Control males that were exposed to SEFL had similar response patterns without attaining recovery on day 10 (Day 10: p < 0.0001 vs baseline) and SEFL did not influence overall withdrawal threshold on any day (Fig. 5), indicating that SEFL had minimal effect on CFA-induced mechanical hypersensitivity in control males. Similarly, CFA resulted in reduced mechanical withdrawal thresholds in male H2A.Z cKO mice without SEFL exposure on days 1, 3, and 7 (all p < 0.05) and in SEFL exposed H2A.Z cKO mice on days 1, 3, and 10 (p < 0.05). Notably, SEFL-exposed mice had higher mechanical thresholds  www.nature.com/scientificreports/ on day 1 compared to mice without SEFL (t 19 = 3.69, p = 0.002), suggesting that stress may reduce mechanical pain hypersensitivity in H2A.Z cKO males (Fig. 5).
In females, CFA reduced mechanical thresholds in all mice (control and H2A.Z cKO mice, with and without SEFL) on each of the 4 test days (days 1, 3, and 10: p < 0.05; day 7: p = 0.053 for control without SEFL, p = 0.050 for H2A.Z cKO without SEFL). Additionally, control females exposed to SEFL had lower pain thresholds on day 1 compared to controls without SEFL exposure (t 13 = 2.16, p = 0.05), suggesting that SEFL enhanced mechanical pain hypersensitivity in females. However, SEFL had no effect in H2A.Z cKO females on any day, suggesting that H2A.Z cKO may protect against SEFL-induced potentiation of mechanical pain (Fig. 5).
Open field and elevated plus maze (EPM). To exclude potential differences in generalized anxiety as a source of sex and genotype differences in fear memory, additional male and female mice were tested in the open field and the EPM. There were no sex or genotype differences in total distance traveled in the open field, but an analysis of time in the centre identified a significant interaction (F 1,47 = 4.16, p = 0.047). Although H2A.Z cKO did not influence time in the centre, H2A.Z cKO males spent significantly more time in the centre than H2A.Z cKO females (t 21 = 3.13, p = 0.005). Neither sex nor genotype influenced any measure of anxiety-like phenotypes in the EPM (time in open and closed arms, number of open or closed arm entries), suggesting that differences in general anxiety are not contributing to differences in fear memory or pain sensitivity (Fig. 6a-e). These data are consistent with evidence that altered pain sensitivity in PTSD patients is not observed in patients with anxiety disorders 12 . www.nature.com/scientificreports/ sion was higher in H2A.Z-depleted compared to control males (all p < 0.05) and H2A.Z depletion had no effect in females. In contrast, Bdnf IV expression was lower in H2A.Z cKO compared to controls only in females (Sex × Genotype interaction: F 1,24 = 7.30, p = 0.01). Similarly, only cKO females had reduced expression of the synaptotagmin 1-encoding gene Syt1 (Sex × Genotype interaction: F 1,24 = 5.23, p = 0.03) (Fig. 7b). Several genes were equally affected by H2A.Z cKO in male and female mice, including Gadd45b (Main effect of Genotype: F 1,24 = 4.23, p = 0.05) and the tyrosine-hydroxylase encoding gene Th (Main effect of Genotype: F 1,21 = 11.97, p = 0.002), whereas the expression of other genes varied by sex, but was not affected by H2A.Z. Specifically, the immediate early gene Fos (Main effect of Sex: F 1,24 = 11.17, p = 0.003) was was higher in males than in females. In contrast, Fkbp5, a gene implicated in depression and PTSD, was higher in females than in males (Main effect of Sex: F 1,24 = 7.00, p = 0.01). Other genes were not affected by H2A.Z depletion or sex, including Egr1, B2M, Syp (encodes synaptophysin), and Nr3c1 (encodes glucocorticoid receptor).
Estrous cycle. We did not track estrous cycle during any behavioural tasks, even though such factors were previously shown to contribute to cognitive function 50 . To indirectly evaluate the potential contribution of estrous cycle in our behavioral outcomes, we conducted additional analyses to compare the degree of variability within males and within females on each behavioral measure in the manuscript using Levene's test for the equity of variances. The assumption here was that if estrous cycle is causing a major behavioral effect in females, it would be observed as increased variability in behavioral outcomes within females than within males 50 . No measure resulted in a significant Levene's statistic (all p > 0.21), thus, we did not observe greater variability in female than in male mice on any of the behaviors reported here, suggesting that estrous cycle was not a major driver of behavior compared to the effect of genotype.

Discussion
Our data show that hippocampal H2A.Z binding is higher in area CA1 of the hippocampus of females compared to males and that conditional inducible H2A.Z deletion has sex-specific effects on fear memory, SEFL, and hypersensitivity to thermal and mechanical pain stimuli. Importantly, we showed that the depletion of both H2A.Z-encoding genes (H2afz and H2afv) in CamKIIα-positive cells produces similar memory-enhancing effect in males as does the selective depletion of H2afz restricted only to the hippocampus 34,35 , suggesting that H2A.Z depletion in adulthood is a well-tolerated method for improving memory. In contrast to our previous studies that selectively depleted H2A.Z in area CA1 or mPFC 34,35 , the deletion of H2A.Z across brain regions in the current study precludes our ability to attribute our findings to specific brain regions. However, given that effects varied based on the aversive nature of the task and stress history, likely candidates include the hippocampus, amygdala and the mPFC. Although we did not find a similar enhancement of fear memory in females with H2A.Z depletion, both male and female mice exhibited improved memory for the non-aversive object-in-place test, indicating that H2A.Z depletion may provide a strong therapeutic target for improving hippocampus-dependent memory in both sexes.
Notably, sex differences were most prominent in aversive tasks, whereas H2A.Z depletion on the non-aversive OiP task produced improved memory irrespective of sex. This is particularly notable because this task is more difficult than standard object recognition and location tasks, as it includes only three 5-min training phases that include exposure to the same 4 objects used during testing. Indeed, this procedure resulted in particularly weak learning in our control mice, which may be related to the memory-impairing effects of tamoxifen injections administered across conditions 51,52 . Contribution of such factors to subthreshold learning created conditions that reveal a memory-enhancing effect of H2A.Z, consistent with use of subthreshold learning paradigms to reveal enhancing effects of various treatments, including estrogens and histone deacetylase (HDAC) inhibitors 53,54 . As such, these data suggest that H2A.Z depletion may be especially beneficial for reducing the threshold for the formation of non-aversive memories.
H2A.Z had sex-specific effects on SEFL, whereby females with H2A.Z depletion had reduced fear potentiation compared to controls, suggesting that lower levels of H2A.Z may protect against PTSD in females. Interestingly, the protective effect of H2A.Z depletion in females extended to pain sensitivity, whereby H2A.Z cKO females exhibited more rapid recovery from SEFL-induced hypersensitivity to thermal stimuli. This is notable because the beneficial effects of H2A.Z deletion on fear and pain sensitization were not observed in male mice despite similar effects on object-in-place memory, indicating that males and females utilize H2A.Z in different ways when adapting to aversive stimuli.
How differences in behavior relate to higher levels of H2A.Z in females is not clear, but elevated H2A.Z binding in our study is consistent with emerging work on sex differences in the epigenome. Specifically, we previously showed that trimethylation of histone H3 at lysine 4 (H3K4me3) has a high level of overlap with H2A.Z binding 34 , which is consistent with evidence that females have more H3K4me3 binding compared to males 55 . Additionally, DNA methylation, which is inversely related to H2A.Z binding in non-neuronal cells 42,56 is higher in males than in females 57 . This is particularly relevant because studies in non-neuronal cells showed that artificially elevating DNA methylation results in reduced levels of H2A.Z binding and vice versa 42 , suggesting that sex differences in H2A.Z binding may be associated with lower levels of DNA methylation in females. Although the upstream drivers of this epigenetic shift are not currently known, the sex difference in H2A.Z binding may reflect broader sex difference in the epigenetic landscape. Such differences in the epigenome are consistent with evidence that males and females exhibit widely different transcriptional responses to acute stress within the hippocampus 19 , generating the hypothesis that sex differences in the epigenome help drive unique adaptations to stress in males and females.
Sex differences in molecular adaptations to stress may be especially important for understanding why the effect of H2A.Z on fear memory varied with history of stress, as evidenced by distinct H2A.Z effects to the same Scientific RepoRtS | (2020) 10:14331 | https://doi.org/10.1038/s41598-020-71229-x www.nature.com/scientificreports/ fear conditioning procedure when it followed SEFL (Fig. 4) compared to mice without SEFL (Fig. 3). Notably, H2A.Z binding in males can be stably altered after fear conditioning 22 , indicating that experience-induced alterations in H2A.Z may mediate adaptations and responses to subsequent events. Some evidence shows that stress exposure has opposite effects on histone modifications in the frontal cortex and hippocampus of males and females 31 , suggesting that unique adaptations to stress can produce unique responses to subsequent fear-related learning in each sex. With respect to H2A.Z, the shift in its effects under different conditions may reflect its complex role in transcriptional regulation, whereby H2A.Z may facilitate various types of responses in different contexts rather than playing a single and specific role in gene regulation. To this end, H2A.Z has been characterized as a factor that integrates upstream signals to produce variable transcriptional outcomes 58 , consistent with the ability of H2A.Z to recruit both repressive and activating complexes to DNA during development 59 . Moreover, its effect on the activity of associated genes has been linked with the history of the gene's activity 60 , suggesting that H2A.Z may influence unique responses before and after stress exposure based on interactions with distinct transcriptional regulators under different conditions. Its role may be further influenced by its interactions with sex hormones, which strongly regulate H2A.Z binding in non-neuronal cells and may interact in the same way in the brain to regulate sex-specific outcomes 61 . Indeed, our gene expression data (Fig. 7) show that the reduction of H2A.Z levels can produce one of 4 genespecific outcomes on basal transcription in area CA1: (1) an effect on transcription only in males; (2) an effect on transcription only in females; (3) an effect on transcription irrespective of sex; (4) no effect on transcription. This mixed profile is consistent with the view of H2A.Z as a "molecular rheostat" for transcriptional regulation 58 and its ability to promote the binding of activating and repressive transcriptional regulators to chromatin 59 to produce varied transcriptional and behavioral outcomes. As such, our behavioral, transcriptional, and H2A.Zbinding data are consistent with a view that H2A.Z integrates diverse upstream signals to facilitate contextrelevant outcomes, which appear to vary with history of stress exposure. This contextual specificity of H2A.Z is particularly relevant for the relationship between sex differences in H2A.Z binding (Fig. 1) and sex differences in gene expression produced by H2A.Z depletion on gene expression (Fig. 7). Indeed, there does not appear to be a one-to-one relationship between higher H2A.Z levels in female mice and sex differences in gene expression, indicating that these regulatory differences may emerge specifically in response to particular upstream signals. Moreover, the variable pattern of gene expression with H2A.Z depletion is consistent with our previous reports of gene-specific changes in expression when H2A.Z levels are reduced, whereby some genes are upregulated and others are downregulated in response to H2A.Z depletion 34,35 .
Additionally, we cannot exclude potential interactions of H2A.Z with sex hormones, as reported in nonneuronal cells 62 , as key contributors to sex-specific effects of H2A.Z. This is particularly noteworthy because females in our study were freely cycling and as such, estrous phase may contribute to behavioral outcomes 50 . Although behavioral performance exhibited similar levels of variability in males and females, suggesting minimal contribution of estrous phase compared to the effect of genotype, we did find that certain genes exhibited sex differences in expression irrespective of sex differences in H2A.Z binding. Thus, H2A.Z is a vital contributor to diverse behavioral and transcriptional profiles in males and females, but its role is selective and appears to be modulated by the diverse epigenetic and hormonal landscape that contribute to complex behavioral outcomes. Other examples of sex-specific mechanisms in behavioral regulation have been reported. For example, knockout of glutamate receptor 1 has sex-specific effects on fear memory despite producing similar effects on spatial memory, and epigenetic editing of Cdk5 selectively regulates fear memory in females and not in males 17,63 . Similar effects were recently reported for stress-mediated adaptations to pain, whereby knocking out the nociception/ orphanin FQ receptor (NOP) is protective against PTSD-potentiated pain sensitivity in males without having an effect in females 16 . Combined with the results of the present study, these data indicate that stress-induced changes, particularly maladaptive fear and pain sensitization, may be mediated by unique molecular underpinnings in males and females, which has important implications for developing sex-appropriate therapeutic interventions.
This study contributes to growing evidence from our lab for therapeutic potential of H2A.Z inhibition. Specifically, these data expand on our prior findings that selective H2A.Z inhibition in the hippocampus results in enhanced fear memory in males 34,35 by demonstrating that forebrain-wide reduction of H2A.Z in CaMKIIαpositive cells also improves fear memory in males and OiP memory irrespective of sex. As such, H2A.Z inhibition may be especially beneficial for improving non-aversive memory. However, our data also suggest that potentially beneficial effects of H2A.Z reduction in PTSD and pain treatment may be sex-specific, as benefits of H2A.Z cKO for reducing SEFL-related outcomes were strongest for females. Although there are no H2A.Z-targeting drugs available today, we did demonstrate that pharmacological inhibition of Tip60, which is part of a complex that deposits H2A.Z into chromatin, regulates memory and reduces H2A.Z binding 22 . As such, emerging data are pointing to the need to develop pharmacological strategies that target H2A.Z directly, or indirectly through its upstream regulators.

Conclusion
Overall, our data extend previous findings that H2A.Z acts as a memory suppressor 22,34,35 and extend these findings to a non-aversive object-in-place memory task in male and female mice. However, our data show sex-specific effects of H2A.Z on fear memory that depend on history of stress exposure in the SEFL paradigm, indicating that H2A.Z has context-specific effects on regulation of fear memory and related disorders. Specifically, our data suggest that higher levels of H2A.Z in female mice may be a risk factor for PTSD and associated increases in pain sensitivity, and simultaneously point to H2A.Z inhibition as a potential mechanism for reducing risk for these conditions in females. Growing evidence points to unique molecular factors as regulators of risk in males and females, and epigenetic factors may be especially important contributors to this sex differences, as epigenetic changes are evident even when other factors, such as changes in neuronal morphology, are not 64  Genotyping. Genotyping was carried out at weaning for all mice used in behavioral studies. Animals were ear notched and DNA was extracted using the "HotSHOT" method 65 . The Cre transgene was identified by amplifying a unique portion within the Cre transgene, as specified by Jackson Labs. Presence of LoxP sites was confirmed by amplifying each target gene (H2afz and H2afv) with primers specified by Riken (see Table 1), such that the floxed genes resulted in a larger PCR product than wild type genes.
Tamoxifen. Tamoxifen (Sigma T5648-5G, final concentration 2 mg/mL) was dissolved in ethanol (40 mg/ mL) and mixed with corn oil (375uL tamoxifen in EtOH/750uL corn oil) in 1.5 mL microcentrifuge tubes. After thorough vortexing, ethanol was evaporated by spinning in Eppendorf Vacufuge Plus on V-AL (vacuumalcohol) setting at 3 °C for 20 min (or until all EtOH had evaporated). Mice received 5 daily injections of tamoxifen (200 mg/kg, IP) beginning at 60-90 days of age and were given 10-14 days to recover before performing additional procedures or testing. All animals (except where specified) were treated with tamoxifen, such that floxed mice without CamkIIα-CreER T2 were used as controls. During tamoxifen treatment, mice were housed in ventilated cages (5 × 7 × 14 in.) in a closed-circuit atmosphere rack (Techniplast Easy Flow #BOX110EFUL) and transferred to standard housing thereafter.

Contextual fear conditioning.
To facilitate habituation to the experimenter/non-associative aspects of fear conditioning, mice were transported to the testing room and handled for 30-60 s daily for 3 days before testing, as previously reported 22,34 . On the training and test days, mice were transported to the testing room and placed into test chambers (9.8 in boxes; designed for mice) equipped with an electrified grid floor (Coulbourn Instruments, Holliston, MA, USA). During training, mice were given 2 min to explore the apparatus before receiving a single foot shock (0.5 mA, 2 s), followed by an additional minute of exploration. Memory was assessed 24 h later by replacing the mouse into the training apparatus without shock and measuring freezing behavior for 3 min. Freezing was recorded with a camera placed directly in the chamber ceiling and scored by automated software (FreezeFrame, Coulbourn Instruments).

Stress-enhanced fear learning (SEFL).
The mouse version of the SEFL paradigm was adapted from Rajbhandari et al. (2018). The chambers and electric shock boxes were purchased from Coulbourn Instruments (Holliston, MA, USA) and the chambers were modified to create 2 distinct training and testing contexts. Context A was differentiated by the following cues: the chamber was wiped down and scented with 1% acetic acid and illuminated by white light. Green plexiglass inserts were placed within the chamber such that the chamber had a green hue and a triangular shape (17.5 cm × 17.5 cm × 23 cm diagonal), and the furniture within the testing room was placed in a triangular format. Context B was illuminated by a red light and cleaned and scented with 5% ammonium hydroxide (Sigma# 221228-500ML-A), the furniture was rearranged along the walls, and no inserts were used, such that the chamber was in an original manufacturer shape (9.8 inch boxes; designed for mice). Mice were handled for 3 days before testing, which began in Context A. Mice were placed into transfer cages (7.5 × 11.5 × 5 in) and transported on a metal cart to a room (also illuminated with white light) across from the testing room and allowed to habituate for 30 min. They were then placed into individual black wooden boxes (8.5 × 8.5 × 8.5 in.) and transported to the testing room, similar to context A in Rau et al. 9 to reduce potential  -CTA GGC CAC AGA ATT GAA AGA TCT 5′-GTA GGT GGA AAT TCT AGC ATC ATC C 324 bp   Cre  5′-AGC TCG TCA ATC AAG CTG GT  5′-CAG GTT CTT GCG AAC CTC AT  184 bp   H2A.Z1  5′-CGC CTT GGT AAT TCT ATC TTC TCC  5′-CGC CAG TTA ACA CAC ATG TGATC  326 bp (WT)  448 bp (flox)   H2A.Z2  5′-GCC TCA GAT CAT CCA GTC  5′-GGC TCT GAA TTC CCA ATG TAG  570  www.nature.com/scientificreports/ generalization between the two contexts. Mice were then placed into the shock chamber and exposed to 10 randomly spaced 1 mA (1 s) shocks over 1 h to mimic a traumatic experience, after which mice returned to home cages. On day 2, mice were trained with a mild fear conditioning protocol in Context B. Mice remained in their home-cages and were transported on a plastic cart to the hallway outside of the testing room and allowed to habituate for 30 min. Mice were then transported to the testing room two at a time and placed in Context B for 2 min of exploration, followed by a single foot shock (0.5 mA, 2 s), then another minute of exploration. Memory for context B was tested on days 3 and 9 (24 h and 7 day Recall, respectively), wherein mice underwent the same procedures as during the Context B Training day but were exposed to Context B without shock. In all cases, percent freezing was scored by automated software (FreezeFrame, Coubourn Instruments).
Object in place (OiP). Mice were transported to the testing room in their home cages and allowed to acclimatize for 30 min. Before behavioral testing, all mice were extensively handled and habituated to an empty testing apparatus (open-field; 45 × 45 × 30 cm) for 10 min on two consecutive days. Mice underwent three 5-min sample phases with 5-min in between each sample. In all sample phases, mice were presented with 4 distinct objects, placed in each corner of the open-field, away from the walls. After a 24 h delay, mice underwent a 5-min choice phase, in which they viewed the same four objects, but two objects had switched locations (right or left, counterbalanced) creating a 'novel side' . Memory is inferred from the preferential exploration of the objects on the novel compared to the familiar side. The novelty preference was quantified by calculating a discrimination ratio [DR = (novel object exploration -familiar object exploration)/(total object exploration)].

Pain behavior. Complete Freund's adjuvant (CFA) preparation and administration.
Complete Freund's adjuvant (CFA; 50%; Sigma), used to induce hypersensitivity to pain, was injected subcutaneously in a volume of 20 μl into the left or right plantar hind paw using a 100-μl microsyringe with a 30-gauge needle. Mice were tested for mechanical (von Frey) or thermal (radiant heat withdrawal) sensitivity of both hind paws as described below. Mechanical and thermal testing occurred before, 1, 3, 7 and 10 days post-CFA injection.
Von Frey test. Mice were placed into Plexiglas chambers (7 × 6 × 11 cm) on a wire mesh and allowed to acclimate for approximately 1 h. The automated version of this test was used (Ugo Basile Dynamic Plantar Aesthesiometer), whereby a single motor-driven fiber aimed at the animal's hind paw was applied until the paw was withdrawn. Force was recorded to the nearest 0.1 g. If a cut-off of 20.0 g of force was reached, the probe automatically descended and 20.0 g was recorded as the threshold. Each paw was tested four times with at least 3 min between successive trials, with paw stimulations counterbalanced between animals.
Radiant heat paw withdrawal test. Mice were placed into Plexiglas chambers (7 × 6 × 11 cm) on a glass surface and allowed to acclimate for approximately 2 h. The commercial device (IITC model 336) was set to 25% active intensity. Latency was recorded to the nearest 0.01 s. Heating was automatically stopped if the cut-off value of 40.00 s was reached without withdrawal, and 40.00 s was recorded as the latency. Mice were left in the chambers and the contralateral paws were then stimulated. This was repeated four times for each animal, with a minimum of 6 min between same-paw stimulations. Paw stimulations were counterbalanced between mice in an alternating fashion. The mean of the four measurements was considered as the pain threshold for subsequent analyses.

Elevated plus maze (EPM).
Mice were transported to the testing room in their home cages and were allowed to habituate to for 30 min. They were then placed in the EPM apparatus (arm width 5 cm, arm length from the middle to end: 35 cm, closed arm wall height: 15.25 cm, elevation from floor: 63 cm) facing one of the closed arms and allowed 10 min to explore. Behavior was video recorded by a Sony HD HandyCam (HDR-CX405) and scored manually by an experimenter blind to the experimental conditions and sex of the animals.

Open field (OF).
To acclimatize mice to transportation and handling, animals were transported to the behavior testing room in their home cages for 30 min of habituation followed by 30-60 s of handling daily for 3 days before testing. On the test day, mice were transported to the testing room in their home cages and allowed to habituate for 30 min before being placed into the testing chamber (45 cm × 45 cm × 45 cm) for 10 min. Video was recorded using a camera (Microsoft LifeCam Studio) and scored using EthoVision XT 8.5. Time spent in the outer region along the wall (10 cm from the wall) and in the centre (25 cm × 25 cm) was scored using EthoVision XT 8.5 (Noldus Information Technology, Wageningen, The Netherlands). Light levels were recorded to be 10.4 lx at the bottom of the chamber.
Tissue collection. For RNA and western blot tissue analysis, mice were cervically dislocated and decapitated. Brains were snap frozen in isopentane and stored at − 80 °C until processing, as in previous reports 22,34 . Western blotting. Tissues were homogenized using a Dounce homogenizer in RIPA buffer (50 mM Tris HCl pH 7.4, 150 mM NaCl, 10% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with Protease Inhibitor Cocktail (Cell Signaling, Danvers, MA, USA), as previously reported 22,34 . Homogenates were incubated for 20 min on ice and flicked every 5 min, centrifuged at maximum speed at 4 °C for 20 min, and the supernatant Table 2. qPCR primers.

Statistics. Analyses were conducted with SPSS Version 24 and consisted of independent-samples t-test
when only a single variable was manipulated or two-way ANOVA when multiple variables were manipulated. For pain studies, data were analyzed with mixed measures ANOVA, with Days after CFA injection as the withinsubjects factor and Genotype and SEFL as the between-subjects factors. When the omnibus test was significant, follow up analyses were conducted with LSD post-hoc, independent-samples t-test or paired-samples t-test, when appropriate.