Chronic variable mild stress alters the transcriptome and signaling properties of the anterodorsal bed nucleus of the stria terminalis in a sex-dependent manner

Abstract

Chronic stress is a physiological state marked by dysregulation of the hypothalamic-pituitary-adrenal axis and high circulating levels of stress hormones, such as corticosterone in mice or cortisol in humans. This dysregulated state may result in the development of mood disorders but the process by which this occurs is still unknown. The bed nucleus of the stria terminalis (BNST) serves as an integration center for stress signaling and is therefore likely an important area for the development of mood disorders. This project utilized a chronic variable mild stress (CVMS) paradigm to persistently stress mice for 6 weeks followed by RNA-Sequencing of the anterodorsal (ad) BNST and electrophysiology of corticotropin releasing hormone-expressing cells in the adBNST. Our results show significant sex-biases in the transcriptome of the adBNST as well as effects of CVMS on the transcriptome of the adBNST specifically in males. Female biased genes are related to synaptic transmission while male biased genes are related to RNA processing. Stress sensitive genes in males are related to synaptic transmission and synapse formation. Additionally, electrophysiology data showed that CVMS suppressed the M-current in males but not females. However, CVMS increased the strength of excitatory post-synaptic currents in females but not males. This suggests significant differences in how males and females process chronic stress. It also suggests that the BNST is more sensitive to chronic stress in males than in females.

1. Introduction

In the central stress response, the major signaling molecules include corticotropin releasing factor/hormone (CRF/H) and adrenergic neurotransmitters. By contrast, the major stress molecule in the periphery is cortisol in humans and corticosterone in rodent models. When presented with a stressor, the brain must first recognize the stressor through the integration of multiple processing centers, notably the amygdala, paraventricular hypothalamus, and locus coeruleus (Herman et al., 2005; Poe et al., 2020). The amygdala and paraventricular hypothalamus produce CRF/H, while the locus coeruleus produces norepinephrine and responds to CRF/H. CRH from the paraventricular hypothalamus controls the hypothalamic-pituitary-adrenal axis.

Chronic stress refers to state of dysregulation of these stress-related pathways. During chronic stress the hypothalamic-pituitary-adrenal axis is overactivated resulting in constant stress signaling in both the brain and periphery. This can result in permanent changes, such as the development of mood disorders in the brain. While chronic stress and acute stress activate similar signaling pathways, it is known that the two states result in different signaling processes in the brain. For example, it has been shown that acute stress leads to an increase in CRH receptor expression, possibly related to sensitization, but chronic stress does not cause this (Bonaz & Rivest, 1998). Similarly, chronic and acute stress have differing effects on the expression of CRF in various regions of the brain (Chappell et al., 1986). This has also been shown to happen to adrenergic pathways as well (Hellriegel & D'Mello, 1997). Thus, it is important to understand how chronic stress is processed and how that processing is changing the function and neuroanatomy of the brain,

The bed nucleus of the stria terminalis (BNST), a part of the limbic system or extended amygdala, is a region that also produces CRF-mediated output to mood regulation. Many of the previously discussed stress-related regions directly innervate the BNST and as such, it also can serve as an integration center for the central stress response (Herman et al., 2005). This heterogeneous area expresses receptors for various stress-related proteins, including CRF receptors and adrenergic receptors (Becker et al., 2008). The BNST is of particular interest for studying chronic stress processing as previous studies suggest the BNST is only activated through long-context stressors, as demonstrated by an increase in c-fos expression in the BNST only after long-context stressors (Perez-Villalba et al., 2005). Further research has shown that when the BNST is lesioned, only the mice exposed to long-context stressors showed impairment to their behavior adaptations and conditioning (Waddell et al., 2006; Hammack et al., 2015).

However, long-context stressors are not the same as chronic stress and, therefore, more evidence is necessary to show that the BNST is involved in chronic stress. This has been shown both molecularly and behaviorally. In one study, chronic stress significantly increased the dendritic arborization of the BNST and this appeared to be specific to this region as the central amygdala, another region of the brain important for stress signaling, had no difference in its arborization (Vyas et al., 2003). Furthermore, another study testing the differing effects of chronic vs. acute stress found that only chronic stress resulted in alterations of the long-term potentiation of the BNST that was paired to changes in their behavior (Conrad et al., 2011). Other studies have shown changes to electrophysiological properties and the transcriptome of specific populations of cells in the BNST after chronic stress (Daniel et al., 2019).

There are multiple ways to study chronic stress in animal models. Common ones include direct corticosterone injections, social dominance, physical restraint, and pain administration. Each chronic stress paradigm has benefits and drawbacks and should be selected based on the experimental question the researcher is addressing. A chronic variable mild stress (CVMS) paradigm, often also called chronic unpredictable mild stress, was developed in the 1980’s and has become a standard of chronic stress research. CVMS uses a randomized schedule of multiple different types of stressors with them frequently being mild in nature though occasionally stronger stressors are used in the schedule (Willner et al., 1987). This paradigm is attractive to study chronic stress as it often models naturalistic stressors that a rodent would encounter, such as uncomfortable environments, threats of nearby predators, and social stress. This improves its translatability to humans. Additionally, this paradigm has been validated as a great model for mood disorders as CVMS causes depression-like behaviors in rodents such as learned helplessness or anhedonia which are reversible by treatment with antidepressants (Kompagne et al., 2008; Wiborg, 2013; Papp et al., 2014; Farhan & Haleem, 2016). Furthermore, this study serves as an expansion and continuation of a previously published projects that utilized the CVMS paradigm to better elucidate the effects that this paradigm has in mice (Pu Hu et al., 2020; Degroat et al., 2024).

When studying how chronic stress develops into a mood disorder, one must also consider how variability across the sexes influences responses to chronic stress. In clinical research, it has been shown that most mood disorders have a strong gender bias, with major depressive disorder being twice as prevalent in cisgender women than cisgender men (Bromet et al., 2011). In animal models, differing responses to chronic stress has also been observed between the sexes. In our previous publication, we found that the response to the CVMS paradigm elicited different responses in avoidance behavior across the sexes and across the estrous cycle (Degroat et al., 2024). We also previously found that the CVMS paradigm, in a study only using male mice, altered the M-current and excitatory post-synaptic currents (EPSCs) in BNST CRH neurons (Pu Hu et al., 2020), but that this paradigm did not alter the M-current in BNST NPY-expressing cells in either sex (Degroat et al., 2024). Further, when using bulk RNA-Sequencing of the anterodorsal (ad) BNST, we found no effects of CVMS on the transcriptome 18 h after the last stressor (Degroat et al., 2024). In the current study, we examined the influence of CVMS on CRH excitability, M-current activity, and EPSCs, in both male and female mice, and, due to the transient nature of the transcriptome, on the transcriptome of the adBNST 1 h after the last stressor (wet bedding). We postulate that the earlier collection post-stress would provide better insight into how stressors alter the transcriptome of the adBNST in a sex-specific matter that may lead to the sustained differences in neurophysiology. Our main goal is to characterize how the neurons of the adBNST, especially CRH neurons, respond to chronic stressors across the sexes.

2. Materials & Methods

2.1. Animals

All procedures were in accordance with National Institutes of Health standards and approved by the Rutgers Institutional Animal Care and Use Committee and agree with ARRIVE’s guidelines for reporting animal research. Mice used for all cohorts were adult male and female C57BL/6J mice. Mice used for electrophysiology were bred in house, mice used for RNA-Sequencing were purchased at 3 weeks of age from Jackson Laboratories (Cat# 000664). For electrophysiology, mice were cross-bred between an Ai14 parent and a Corticotropin Releasing Hormone (CRH)-Cre parent so that offspring had mutations for both Ai14 and CRH-Cre, resulting in exclusive expression of td-Tomato in CRH-expressing cells. Both Ai14 and CRH-Cre breeding mice were purchased from Jackson Laboratories (Cat # 007914 & 012704, respectively). Control or stress conditions were started at 7–10 weeks of age. Except for stressors, mice were housed in a temperature and humidity-controlled room (22°C, 30–70% humidity), on a 12/12 h light/dark cycle, and provided food and water ad libitum. Sample sizes of n=7–9 cells for electrophysiology allowed for detection of affects with a power = 0.8 at α = 0.05. This was based on previously published studies in our lab (Hu et al., 2016; P. Hu et al., 2020; Degroat et al., 2024). RNA-Sequencing was conducted via pooling 3 biological replicates into a single sample with 15 mice per group, resulting in an ultimate sample of n=5 per group. For electrophysiology mice, they were sacrificed the morning after the last stressor, approximately 18 hours, at 9:00am-10:00am. For the RNA-Sequencing cohorts, mice were euthanized 1 hour after the last morning stressor at 10:30am-11:30am. Body weight was collected in the RNA-Sequencing cohort once a week to monitor health and the effects of the CVMS paradigm.

2.2. Chronic Variable Mild Stress Paradigm

The Chronic Variable Mild Stress (CVMS) paradigm is 6-week schedule of 1–2 minor stressors every day, typically with one morning stressor and one evening stressor. This paradigm follows the same procedure as outlined in our previous publication (Degroat et al., 2024). Briefly, mice were subjected to the following stressors: 3 or 8 hours of humid bedding, 8 hours of no bedding, 8 hours of replacing bedding with room temperature water, placing the cage on a 45° tilt for 8 hours, isolation for 8 hours or overnight, exposure to novel mice’s dirty cages for 8 hours, lights on overnight, 2 or 3 hours of dark during typical light cycle, 5 or 6 bedding changes in a single day, 2 minute swim stress in 4°C water, 4 minute swim stress in room temperature water, 15 minute cold stress, and 15 minutes of predator sounds, and 1 hour of restraint stress.

2.3. RNA-Sequencing

To collect tissue for RNA-Sequencing, mice were first anaesthetized via ketamine followed by decapitation. Mice were injected with 0.1 mL of ketamine and mice were monitored until anesthetization was ensured. All mice were euthanized within 5 minutes of ketamine injection. Brains were dissected out, sliced, and then microdissected using a dissecting scope and scalpel to isolate the adBNST. The adBNST was collected and pooled bilaterally. Tissue samples from 3 biological replicates were then pooled prior to RNA extraction and were stored in RNAlater at −80°C. RNA was extracted using an RNAqueous Micro Isolation kit (Invitrogen; AM1931). Integrity, concentration, and quality of the RNA samples were analyzed with an Agilent 2100 Bioanalyzer prior to being sent to the JP Sulzberger Columbia Genome Center (New York, NY) for sequencing. All samples sent to JP Sulzberger Columbia Genome Institute were analyzed on an Agilent 2100 Biolanalyzer to ensure that the RNA integrity number (RIN) was greater than 8. No samples were discarded due to low RIN. Libraries were constructed using Illumina TruSeq Stranded mRNA Library Prep Kit with KAPA HiFi HotStart Ready Mix for the final PCR step and were sequenced on an Element AVITI with 100bp paired-end reads at a depth of 40 million. Resulting fastq files were then analyzed as previously described (Degroat et al., 2024). RNA-Sequencing had a sample size of n=5 for each group with no outliers. Raw sequencing data can be accessed at GEO GSE281507.

2.4. Electrophysiology

Electrophysiology was conducted as previously described (Degroat et al., 2024). Briefly, mice were decapitated without anesthetization between 8:30am and 10:00am. The brain was then dissected out and sliced on a VT1000 S vibrating blade microtome (Leica) at 250 nm thickness in a high sucrose artificial cerebral spinal fluid (aCSF) that was prepared in-house (containing in mmol: 208 sucrose, 2 KCl, 26 NaHCO3, 10 glucose, 1.25 NaH2PO4, 2 MgSO4 1 MgCl2, 10 HEPES with a pH of 7.35 and osmolarity of 290–310 mOsm). Slices were then allowed to acclimate at room temperature in a bath of continuously oxygenated aCSF (containing in mmol: 124 NaCl, 5 KCl, 2.6 NaH2PO4, 2 MgCl2, 2 CaCl2, 26 NaHCO3, 10 glucose with a pH of 7.35 and osmolarity of 290–310). Once acclimated, individual slices were placed in the bath of the electrophysiology rig that contained the same aCSF recipe as the acclimatization chamber but held at 35°C. Individual cells were then selected by fluorescence of the tdTomato (a representative image of the fluorescing CRH+ neurons is available in Fig. 3l). A borosilicate pipette backfilled with an internal solution (containing in mmol: 10 NaCl, 128 K-gluconate, 1 MgCl2, 10 HEPES, 1 ATP, 1.1 EGTA, and 0.25 GTP with a 7.35 pH and 290–300 mOsm) and used to patch onto the cell. Membrane values and standard protocol were recorded for every cell and throughout the duration of the patching to ensure the health of the cell. Data collection was done using 1550B Axon Instruments Digidata acquisition system, 700B Axon Instruments Multiclamp amplifier, and pCLAMP software (version 11.1) from Molecular Devices.

Recordings and analysis of the M-current was conducted as previously described (Degroat et al., 2024). Briefly, slices were first bathed in an aCSF solution that contained 1μM tetrodotoxin (TTX) (Tocris, Cat. # 1069) and allowed to perfuse for 5 minutes before recording. Under voltage-clamp, standard deactivation protocol was used to measure K+ currents elicited during 500-millisecond voltage steps from −30 to −75 mV in 5-mV increments after a 300-msec prepulse to −20 mV. After this baseline recording, XE-991 (Tocris, Cat. # 2000), a selective inhibitor of KCNQ channels was, introduced to the bath at 40μM and allowed to perfuse for 10 minutes. The same deactivation protocol was then recorded. M-current data was analyzed by Clampfit 11. The M-current is an inhibitory current that acts through voltage-gates K+ channels (Brown & Adams, 1980; Robbins, 2001). As it is one of the few currents that is active at physiologically-relevant voltages, it has an impact on the likelihood for a cell to fire an action potential (Marrion, 1997). In summation, a strong M-current suggest a cell that is less likely to fire, whereas a weak M-current suggests a cell that is more likely to fire. Representative traces of the M-current can be found in Figure 3a.

Excitatory post-synaptic current (EPSC) was analyzed on Easy Electrophysiology. EPSCs were detected using the template event detection feature using the deconvolution method with a theta of 4.5. Events were then individually checked to ensure noise was not accidentally counted. Events were then averaged, and number of events, average amplitude, and average charge (Q), analyzed as area under the curve for the EPSCs, were recorded. Two 5-minute recordings were done for EPSCs, one in which the slice was in vehicle conditions and one in which Stressin 1 (Tocris, Cat. #1608), a selective agonist of the CRH receptor 1, was introduced to the bath. For this second recording, Stressin 1 was introduced to the bath at a 100 micromolar concentrations and allowed to incubate on the slice for 5 minutes before recording. Representative traces of current recording for EPSCs can be found in Figure 4a.

M-current tests had sample sizes of n=9 for control males from 6 mice, n=10 for stressed males from 8 mice, n=8 for control females from 6 mice, and n=9 for stressed females from 8 mice. There were no outliers for M-current. EPSC tests had sample sizes of n=7 for control males from 6 mice, n=8 for stressed males from 7 mice, n=8 for control females from 7 mice, and n=9 for stressed females from 8 mice. A table of outliers for EPSCs can be found as Supplementary Table 1. Females were tested only in diestrus to ensure consistent hormone profiles between mice, diestrus was confirmed by vaginal cytology the morning of the electrophysiology recording. Unless otherwise specified, all chemicals were purchased from Sigma-Aldrich.

2.5. Statistics

Unless otherwise specified, all tests were analyzed using Prism (GraphPad Prism, Version 10, Dotmatics) and used a p-value or adjusted p-value of <0.05 for significance. Body weight tracking and electrophysiology tests were first checked for outliers by the Grubbs’ test using a critical value of 0.05 as the cut off EPSC data were analyzed by a two-way ANOVA. RNA-Sequencing data was analyzed by DESeq2 on R Studio. Cumulative body weight gain and the IV plots of the M-current recordings were analyzed by a repeated-measures two-way ANOVA. Max peak of the current at −35mV, resting membrane potential, and change in input resistance were analyzed by a two-way ANOVA followed. All ANOVAs were followed by Tukey post-hoc comparisons.

3. Results

3.1. Body Weight

Body weight gain was impacted by CVMS exposure in both the males and females; however, the sexes responded differently. In males, body weights between control and stressed counterparts remained consistent for the first three weeks, diverging thereafter [main effect of treatment: F (1, 28)=10.94, p=0.0026; interaction effect (Weeks × Treatment): F (6, 68)=9.498, p<0.0001], Fig. 1a). After week one, the control males gained 2.6% of their body weight and stressed males gained 3.3% (not significant (ns)). After week 2, control males gained 7.9% while stressed males gained 7.6% (ns). After week three, control males gained 9.3% while stressed males gained 7.9% (ns). After week four, control males gained 13.8% while stressed males gained 9.8%, (trending, p= 0.054). After week five, control males gained 15.5% while stressed males gained 11.0% (p<0.0001). After the final week of stressors, control males gained 19.0% while stressed males gained 12.4% (p<0.0001). This data indicates that CVMS inhibited weight gain, presumably through changes in feeding or metabolism, in male mice.

Figure 1:

Cumulative %change in body weight during the 6-week stress paradigm in (A) male and (B) female mice. Analyzed by repeated measures two-way ANOVA. (*=0.05–0.01, **=0.01–0.001, ***=0.001–0.0001).

In females, body weight gain was different at the beginning [main effect of treatment: F (1, 28)=8.820. p=0.0061; interaction effect (Weeks × Treatment): F (6,168)=11.42, p<0.0001], then converged towards the end of the paradigm (Fig. 1b). After week one, control females gained 1.7% of their body weight stressed females gained 6.8% (p=0.0004). After week two, control females gained 5.6% while stressed females gained 12.1% (p<0.0001). After week three, control females gained 8.4% while stressed females gained 12.0% (p=0.0135). After week four, control females gained 10.1% while stressed females gained 12.9% (ns). After week five, control females gained 12.7% while stressed females gained 17.3% (p=0.0057). After the final week of stressors, control females gained 15.9% while stressed females gained 14.7% (ns). For females, data shows that they have a transient increase in body weight gain after exposure to stress, which plateaus under chronic conditions.

When comparing sexes, there are differences in both the control and stress groups. In the control group, males gained weight quicker than females, as expected [sex effect: F (1, 28)=5.732, p=0.0236; interaction effect (Weeks × Sex): F (6,168)=3.847, p=0.0013] (Fig. 1c). However, in the stressed groups, the females gained more weight quicker than the males [sex effect: F (1, 28)=13.17, p=0.0011; interaction effect (Weeks × Sex): F (6, 168)=3.901, p=0.0011] (Fig. 1d). This shows the differing effects of chronic stress on the sexes as the females gained weight quicker because of the stress whereas the male mice gained weight slower due to stress.

3.2. adBNST Transcriptome

The principal component analysis of all samples used for RNA-Sequencing illustrates that there are no outliers for any groups as all samples fit within their respective 95% confidence interval ovals (Fig. 2a). The analysis also finds that males and females are strongly separated, and control male mice are separated from stressed male mice. However, the control and stressed female samples are almost entirely overlapping. There were 248 genes found to be significantly differentially expressed between males and females and a further 216 that were trending (0.10>adj. p>0.05) (Fig. 2b). Of the significant genes, 116 were male biased while 132 were female biased.

Figure 2:

Bulk RNA-Sequencing of the anterodorsal bed nucleus of the stria terminalis. (A) Principal component analysis of all samples. (B) Volcano plot comparing control males and control females. A positive fold change represents a female biased gene while a negative fold change represents a male biased gene. (C & D) Volcano plots comparing control and stressed animals in (C) males and (D) females. A positive fold change represents an upregulation due to stress while a negative fold change represents a downregulation due to stress. (E & F) Gene ontology analyses comparing control males and control females, (E) pathways enriched in control males (F) pathways enriched in control females. (G) Gene ontology analysis comparing control males and stressed males showing pathways upregulated due to stress. A is analyzed by PCAexplorer; B-D are analyzed by DESeq2; and E-F are analyzed by enrichGO. Significant DEGS are highlighted in red in B-D.

Of the genes varying across sex, male biased genes of interest included: Hcrtr1 (log₂FC=1.075, adj. p=0.0240), the orexin receptor (Scott et al., 2011), Sgk1 (log₂FC=1.045, adj. p<0.0001), a modulator of neuronal excitability (Lang et al., 2010), Tac2 (log₂FC=0.783, adj. p=0.0167), the tachykinin/neurokinin B receptor (Al Abed et al., 2021), Mc4r (log₂FC=0.712, adj. p=0.0012), a melanocortin receptor (Tao, 2010), Crh (log₂FC=0.572, adj. p=0.0109), Insyn2a (log₂FC=0.280, adj. p=0.0297), a modulator of post-synaptic potential (Krueger-Burg et al., 2017), and Gna13 (log₂FC=0.198, adj. p=0.0254) which enables dopamine signaling (Yamaguchi et al., 2000). These differences suggest sex-biases in the type and strength of signaling of adBNST neurons as many genes are related to receptors or signaling cascades. A further list of selected male-biased DEGs of interest along with their log₂FC, adjusted p-value, and functional relevance can be found in Table 1. A list of all male-biased genes can be found in Supplementary Table 2.

Table 1:

Genes that are male-biased in the adBNST when comparing control males and control females. Genes are listed in order of the log2 fold change. A full list of all male-biased genes can be found in Supplementary Table 1.

Male Biased Genes
Gene	Log2 fold change	Adjusted p-value	Function
Ddx3y	10.335414	2.25E-58	Highest male-biased fold change
Hcrtr1	1.075102	0.02403575	Receptor
Plekhf1	1.0685986	0.00092238	Cell signaling
Sgk1	1.0453092	4.45E-06	Synaptic signaling
Tac2	0.7833121	0.01666949	Receptor
Mc4r	0.7118213	0.00124686	Receptor
Tsc22d3	0.6430774	1.78E-06	Transcription Factor
Crh	0.5717431	0.01086472	Neuropeptide
N4bp2	0.4978987	0.00402842	Neuronal excitability
Slc5a3	0.470757	9.80E-07	Ion channel
Prr5	0.4149862	0.00186334	Cell signaling
Fam126a	0.3768818	5.52E-05	Cell signaling
Mat2a	0.301764	0.00124686	Cell signaling
Insyn2a	0.2799577	0.02967483	Synaptic signaling
Irs2	0.2523133	0.00774428	Cell Signaling
Dpysl2	0.2273524	0.00159526	Synaptic signaling
Gna13	0.1976531	0.02541766	Synaptic signaling

Female-biased genes of interest included: Gsg1l (log₂FC=0.603, adj. p=0.0131) and Cacng3 (log₂FC=0.428, adj. p=0.0483), both regulators of AMPA receptors (Milstein & Nicoll, 2008; Gu et al., 2016), Tgfa (log₂FC=0.530, adj. p=0.0113), a regulator of neurogenesis (Jun Ma et al., 1992), Camk2n1 (log₂FC=0.443, adj. p=0.0046), an inhibitor of long-term potentiation (Astudillo et al., 2020), Cplx1 (log₂FC=0.281, adj. p=0.0423) and Cplx2 (log₂FC=0.350, adj. p=0.0297), both involved in synaptic vesicle exocytosis (Yoon et al., 2008), Anks1b (log₂FC=0.303, adj. p=0.0405), a modulator of synaptic plasticity (Younis et al., 2019), Comt (log₂FC=0.215, adj. p=0.0433), catecholamine metabolism (Harrison & Tunbridge, 2008), and Slc1a2 (log₂FC=0.181, adj. p=0.0397), a glutamate transporter (Bjørnsen et al., 2014). This suggests a difference in both the signaling and synaptic plasticity in the adBNST between males and females. A further list of selected female-biased DEGs of interest along with their log₂FC, adjusted p-value, and functional relevance can be found in Table 2. A list of all female-biased genes can be found in Supplementary Table 3.

Table 2:

Genes that are female-biased in the adBNST when comparing control males and control females. Genes are listed in order of the log2 fold change. A full list of all female-biased genes can be found in Supplementary Table 2.

Female Biased Genes
Gene	Log2 fold change	Adjusted p-value	Function
Xist	12.0753291	4.74E-247	Highest female-biased fold change
Scn4b	1.06144811	0.01592177	Ion channel
Unc13c	0.69298312	0.0003445	Synaptic signaling
Gsg1l	0.60262271	0.01313623	Synaptic signaling
Rgs7bp	0.56399579	0.00181925	Synaptic signaling
Tgfa	0.52973038	0.01129631	Growth factor
Pcp4	0.51704562	0.00270595	Synaptic signaling
Kcnab1	0.4574967	0.01979391	Ion channel
Dgkb	0.45423177	0.02057618	Synaptic signaling
Camk2n1	0.44342939	0.00457921	Synaptic signaling
Cacng3	0.42791418	0.04830886	Synaptic signaling
Pcsk2	0.41706433	0.00572449	Synaptic signaling
Gabra4	0.40890884	0.00790706	Receptor
Gpr158	0.39768693	0.01796807	Receptor
Spock3	0.38562695	0.00673688	Calcium regulation
Lrrtm3	0.36543401	0.01125902	Synaptic signaling
Cplx2	0.34982939	0.02967483	Synaptic signaling
Neto2	0.32994357	0.01532972	Synaptic signaling
Cdh8	0.32613862	0.00786683	Cellular adhesion
Anks1b	0.30290019	0.04046966	Synaptic signaling
Adcy1	0.29883662	0.01387116	Synaptic signaling
Cacnb4	0.28553624	0.04816953	Ion channel
Cplx1	0.28103691	0.04227382	Synaptic signaling
Tmem132b	0.26624608	0.00249415	Receptor
Comt	0.21502622	0.04332065	Synaptic signaling
Gpm6b	0.19125504	0.03356488	Synaptic signaling
Slc1a2	0.18134176	0.03972835	Transporter

Comparing control females and stressed females, only 1 gene, Rbm3, was found to be differentially regulated and only 6 further genes were trending (Fig. 2d). Rbm3 was upregulated due to stress (log₂FC=0.436, adj p<0.0001). Of the genes altered by stress in males, 217 were differentially regulated with 57 being downregulated by stress and 160 upregulated and another 301 trending (Fig. 2c). Those downregulated by stress included: Sgk1 (log₂FC=−0.842, adj. p=0.0050) (Lang et al., 2010), Slc1a6 (log₂FC=−0.552, adj. p=0.0232), a glutamate transporter (Gegelashvili et al., 2000), Crh (log₂FC=−0.485, adj. p=0.0420), Vamp1 (log₂FC=−0.424, adj. p=0.0196), involved in synaptic vesicle exocytosis (Liu et al., 2011), Nts (log₂FC=−0.409, adj. p=0.0295), neuropeptide receptor binding (Mustain et al., 2011), Arrb2 (log₂FC=−0.313, adj. p=0.0348), a modulator of β-adrenergic and dopamine signaling (Beaulieu et al., 2005), and Sorcs3 (log₂FC=−0.225, adj. p=0.0232), a regulator of long-term synaptic depression (Breiderhoff et al., 2013). Those upregulated due to stress included: Plxnd1 (log₂FC=0.695, adj. p=0.0263), involved in synapse assembly (Wang et al., 2015), Drd2 (log₂FC=0.650, adj. p=0.0390), a dopamine receptor (Noble, 2000), Unc13c (log₂FC=0.580, adj. p=0.0147), involved in glutamatergic transmission (Padmanarayana et al., 2021), Adra2c (log₂FC=0.530, adj. p=0.0484), an adrenergic receptor (Bylund, 1992), Tgfa (log₂FC=0.459, adj. p=0.0431) (Jun Ma et al., 1992), Gabra4 (log₂FC=0.455, adj. p=0.0067), a subunit of the GABA receptor (Mulligan et al., 2012), Camk2n1 (log₂FC=0.413, adj. p=0.0196) (Astudillo et al., 2020), Cplx2 (log₂FC=0.380, adj, p=0.0230) (Yoon et al., 2008), and Gpr158 (log₂FC=0.361, adj. p=0.0431), a GPCR implicated in depression (Sutton et al., 2018). Taken together, this suggests that stress causes major changes in synaptic signaling in the adBNST in males. A further list of selected DEGs of interest that are downregulated or upregulated due to stress along with their log₂FC, adjusted p-value, and functional relevance can be found in Tables 3 and 4 respectively. A list of all stress-regulated genes can be found in Supplementary Table 4.

Table 3:

Selected genes that are downregulated by stress in the adBNST in males. Genes are listed in order of the log2 fold change. A full list of all genes regulated by stress in males can be found in Supplementary Table 3.

Male Stress Downregulated
Gene	Log2 fold change	Adjusted p-value	Function
Sgk1	−0.8415739	0.00501306	Most downregulated gene
Slc1a6	−0.5518564	0.02320938	Transporter
Akap12	−0.5109548	0.01745418	Synaptic signaling
Crh	−0.4851822	0.04204631	Neuropeptide
Tsc22d3	−0.4394077	0.01335128	Transcription factor
Vamp1	−0.4242631	0.01961082	Synaptic signaling
Nts	−0.4087929	0.0294921	Neuropeptide
Slc5a3	−0.3474649	0.00607686	Ion channel
Otof	−0.3466157	0.04284925	Synaptic signaling
Prr5	−0.3443065	0.01854398	Cell signaling
Alk	−0.3421647	0.03482863	Synaptic signaling
Arrb2	−0.3132115	0.01075937	Synaptic signaling
Epha5	−0.2674514	0.04311296	Receptor
Grk3	−0.2649624	0.04374024	Synaptic signaling
Sema4a	−0.2401795	0.04735542	Synaptic signaling
Sorcs3	−0.22523	0.02320938	Synaptic signaling
Dpysl2	−0.1670872	0.04284925	Synaptic signaling

Table 4:

Selected genes that are upregulated by stress in the adBNST in males. Genes are listed in order of the log2 fold change. A full list of all genes regulated by stress in males can be found in Supplementary Table 3.

Male Stress Upregulated
Gene	Log2 fold change	Adjusted p-value	Function
Scn4b	1.08954915	0.01961082	Most upregulated gene
Rasd2	0.87684047	0.02308458	Synaptic signaling
Ppp1r1b	0.7098293	0.03707918	Cell signaling
Plxnd1	0.69541227	0.02628802	Synaptic signaling
Itpr1	0.6703348	0.03462583	Receptor
Drd2	0.65004132	0.03902923	Receptor
Ano2	0.61227969	0.00607686	Ion channel
Gng7	0.59981204	0.03850164	Receptor
Musk	0.5834368	0.02217479	Synaptic signaling
Unc13c	0.58046104	0.01465707	Synaptic signaling
Slc18a3	0.57407102	0.0426947	Transporter
Adra2c	0.52976636	0.04836964	Receptor
Penk	0.50832335	0.03386401	Neuropeptide
Ddn	0.50724373	0.01961082	Synaptic signaling
Rgs7bp	0.50103361	0.01817878	Synaptic signaling
Dapp1	0.48901824	0.01854398	Cell signaling
Kcnab1	0.48442864	0.01961082	Ion channel
Lzts3	0.4828807	0.03043355	Synaptic signaling
Kcna4	0.47734195	0.04524505	Ion channel
Tgfa	0.45877183	0.04311296	Growth factor
Gabra4	0.45498929	0.006678	Receptor
Filip1	0.45307766	0.03310177	Synaptic signaling
Mctp1	0.43760904	0.04204631	Synaptic signaling
Epha7	0.42772808	0.04204631	Receptor
Camk2n1	0.41345675	0.0195909	Synaptic signaling
Pcsk2	0.40659244	0.01681495	Protein modification
Shank3	0.39997231	0.04311296	Synaptic signaling
Cplx2	0.38001612	0.02296941	Synaptic signaling
Gpr158	0.36139748	0.04311296	Receptor
Spock3	0.34781457	0.02857924	Calcium regulation
Igsf11	0.29617933	0.03138945	Synaptic signaling
Adcy1	0.26192307	0.04311296	Synaptic signaling
Tmem123b	0.25428606	0.01075937	Receptor
Frmpd4	0.23296982	0.01681495	Synaptic signaling

Comparing control males and females for gene ontology enrichment, pathways that were male biased were mostly related to RNA processing and protein localization (Fig. 2e) whereas pathways upregulated in females are related to neuronal activity, such as neurogenesis and regulation of the synapse (Fig, 2f). Comparing control males to stressed males, pathways that are upregulated due to stress are mostly related to synaptic activity and neurotransmission (Fig. 2g). There were no pathways that were significantly downregulated due to stress in males.

3.3. Electrophysiology

3.3.1. M-current

In examination of stress-induced changes in excitability in the adBNST from CRH-expressing neurons, we found no differences in the resting membrane potential between the experimental groups. The average RMP in CRH neurons in control males was −69.2 ± 2.2 mV, in stressed males, −69.0 ± 1.9 mV, in control females, −68.7 ± 1.9 mV, and in stressed females, −67.7 ± 2.3 mV (ns, Fig, 3b). We also found no differences in the change in input resistance (Rin) due to XE-991, the KCNQ channel blocker. The average change in Rin in control males was −0.39 ± 0.18 pA, in stressed males, −0.38 ± 0.09 pA, in control females, −0.43 ± 0.11 pA, and in stressed females, −0.22 ± 0.14 pA (ns, Fig. 3c).

Figure 3:

Recordings of M-current activity of BNST Corticotropin Releasing Factor-expressing neurons from stressed and non-stressed mice. (A) M-current protocol and representative traces. (B) Resting membrane potential of the neurons. (C) Change in input resistance after perfusion with XE-991. I-V plots from −75 to −30 mV after XE-991 perfusion in (D) Control male mice and (E) CVMS-exposed male mice. (F) Current at max peak in male mice. (G) I-V plot of XE-991 sensitive current in male mice. (H-K) Control and CVMS- exposed female mice. Data are presented as mean +/− SEM. (a=0.5–0.1) (**=0.01–0.001, ****<0.0001).

As we have previously characterized the M-current in oval BNST (ovBNST) CRH neurons in male mice (Hu et al., 2020), we next examined the M-current in CRH neurons across the sexes. We found that CRH neurons from males and females did not have significant differences in the strength of their M-currents in control conditions (not shown). However, we did find an effect of CVMS in the males, as we previously found (Hu et al., 2020), but not CRH neurons from female mice. XE-991 perfusion suppressed the M-current activity in control male mice [interaction effect (Voltage × XE-991): F (9,126)=5.728, p<0.0001; −75 mV: 0.0250; −50 mV: 0.0118; −45 mV: p=0.0028; −40 mV: p=0.0029; −35 mV: p=0.0017; −30 mV: p=0.0131] (Fig. 3d). However, in stressed males, while they had a significant interaction effect, their current traces differed at −75 mV [interaction effect (Voltage × XE-991): F (9, 144)=2.871, p=0.0038; −75mV: p=0.0383 (Fig. 3e). The max peak, defined as the strength of the current at −35mV, in CRH neurons from males was significantly decreased after exposure to XE-991 [main effect of XE-991: F (1, 30)=1.547, p=0.0222] (Fig. 3f). However, the traces of the XE-991 sensitive current were significantly decreased in CRH neurons from stressed males in comparison to control males [interaction effect (Voltage × Treatment): F (9,135)=2.698, p<0.0001; −50 mV: p=0.0279; −45 mV: p=0.0125; −40 mV; p=0.0281; −35 mV: p=0.0115; −30 mV: p=0.0468] (Fig. 3g).

The M-current was suppressed by XE-991 perfusion in CRH neurons from both control and stressed female mice (Fig. 3h & i). Max peak was trending reduced in both control and stressed groups after XE-991 perfusion [main effect of XE-991: F (1, 30)=4.112, p=0.0515), indicating that the M-current is active in these neurons. Further, CVMS did not alter the XE-991 sensitive current in CRH neurons from female mice.

3.3.2. EPSCs

We next examined the influence of CVMS on EPSC in both sexes. Previous studies in our lab suggest that CVMS augments EPSCs in male mice (Hu et al., 2020). In the current study, CVMS did not change the frequency, amplitude, or charge of EPSCs in male mice. CRH neurons from control males in vehicle conditions exhibited an average of 159.4 ± 47.5 EPSCs events while in stressed males, CRH neurons exhibited an average of 145 ± 43.8 EPSCs (ns, Fig. 4b). In the same neurons, the average EPSC amplitude was −8.65 ± 1.0 pA from control males and the average EPSC amplitude was −9.8 ± 1.0 pA in neurons from stressed males (ns, Fig. 4c). The charge of EPSCs in neurons from controls males was an average Q of −82.7 ± 12.6 pA*ms and in neurons from stressed males, an average of −71.2 ± 15.3 pA*ms (ns, Fig. 4d).

Figure 4:

Recordings of EPSCs in CRF-expressing neurons of the adBNST from control and stressed mice. (A) Representative traces of current recordings for EPSCs in vehicle and after exposure to Stressin I. Number of EPSCs detected in a 5-minute voltage-clamp recording in vehicle and after exposure to Stressin I in (B) male and (E) female mice. Average amplitude of EPSCs detected in (C) male and (F) female mice. Average area under the curve of EPSCs detected in (D) male and (G) female mice. (H) Comparison of average amplitude in males and females in control mice . (I) Comparison of average amplitude in male and female mice in stressed mice. (J) Comparison of average amplitude between males and females and control and stressed when cells are in vehicle conditions. Data is presented as mean +/− SEM. Analyzed via two-way ANOVA (*=0.05–0.01, **=0.01–0.001).

After initial recording of EPSCs, we perfused each CRH neuron with Stressin 1 to determine if CRH acts in a feedback loop in within the adBNST CRH neuronal population, either in an autocrine or paracrine fashion. After perfusion with Stressin I, the frequency of EPSCs in CRH neurons from control males was an average of 131.5 ± 42.5 events and from stressed males, was an average of 115 ± 21.7 events (ns, Fig, 4b). The amplitude of EPSCs in CRH neurons from control male was an average of −9.0 ± 0.8 pA whereas in CRH neurons from stressed males, an average of −10.3 ± 0.9 pA (ns) (Fig. 4c). The average charge (Q) of EPSCs from control males was −68.3 ± 2.9 pA*ms and from stressed males, –78.0 ± 12.6 pA*ms (ns, Fig. 4d).

CRH neurons from female mice exhibited differences in the amplitude of their EPSCs due to stress. The frequency of EPSCs in CRH neurons from control females in vehicle conditions was an average of 73.8 ± 29.5 events and from stressed females, was an average of 75.0 ± 11.2 EPSCs (ns, Fig. 4e). The amplitude of EPSCS in CRH neurons from control females in vehicle, was an average of −13.7 ± 1.6 pA whereas from stressed females, was an average of −8.5 ± 1.0 pA [main effect of stress: F (1, 28)=12.77, p=0.0013] (Fig. 4f). The charge (Q) of EPSCS in CRH neurons from control females in vehicle, was an average of −81.7 ± 13.3 pA*ms and in neurons from stressed females, was an average of −62.7± 5.7 pA*ms (Fig. 4g).

After perfusion with Stressin I, EPSC frequency in CRH neurons from control females was 121.1 ± 33.7 and in CRH neurons from stressed females was 87.8 EPSCs ± 16.6 (ns, Fig. 4e). EPSC amplitude was −13.1 ± 1.8 pA in CRH neurons from control females and −8.6 ± 1.0 pA from stressed females [main effect of stress: F (1, 28)=12.77, p=0.0013] (Fig. 4f). The average Q was −72.4 ± 12.6 pA*ms in CRH neurons from control females and −61.9 ± 4.9 pA*ms in CRH neurons from stressed females (ns, Fig. 4g).

When comparing males and females, we do see differences in the amplitude of the EPSCs. Both in vehicle and Stressin I, EPSC amplitude was significantly lower in CRH neurons from females with a main effect of sex (F(1,24)=9.587, p=0.0049, Fig. 4g). We do not observe differences in the stressed animals, suggesting that CVMS causes the female CRH neurons to behave similarly to CRH neurons in males, though this cannot be confirmed (Fig. 4h). This suggestion is further supported by the interaction effect in Fig. 4i which indicates that the EPSCs in CRH neurons from females and males are responding differently to stress [F(1,26)=7.114, p=0.0130], which appears to be mainly driven by the stronger amplitude of EPSCs in the control females (pairwise comparison to control males: p=0.0398; pairwise comparison to stressed females: p=0.0180).

4. DISCUSSION

Our previous research using CVMS suggested that the transcriptome of the adBNST was not affected by stress, nor did it affect the M-current in NPY neurons of the adBNST (Degroat et al., 2024). This seemingly contradicted our hypothesis that the BNST was a major processing center for chronic stress. However, these current finding necessitate further research into the BNST and its relations to stress signaling, mood disorders, and sex variability. In contrast to our past research, this study suggests that there are transient changes in the transcriptome due to stress in males but not females suggesting that adBNST neurons and their signaling properties are altered by chronic stress, but in mechanisms that vary across the sexes.

We found an abundance of genes that were differentially expressed between males and females and regulated by stress in the males. Interestingly, we found no such differences in the females, already suggesting that stress is affecting this area differently in males and females. The lack of response to CVMS in females also suggests that the adBNST is an important mediator of stress in males but not in females and that the major central stress regulating region in females may be elsewhere. This hypothesis is further supported by the fact that presynaptic EPSCs were affected by stress in the females but the not males.

Alternatively, this data could suggest that CVMS is not effective on the adBNST transcriptome or CRH neurophysiology in females, although, our previous publication found that females were sensitive behaviorally to the same CVMS paradigm, which was dependent on the estrous cycles (Degroat et al., 2024). As we did not see differences due to stress in final cumulative BW gain or adBNST transcriptome, this suggests there may be a resistance to CVMS. Historically, literature has been contradictory in how sex relates to sensitivity of CMVS in mice (Hodes, 2018). This issue is further compounded by the limited research of CVMS in mice compared to rats, the limited research of CVMS in females, and potential strain differences which also show differing sensitivities to stress (Mineur et al., 2006). Further research should control for estrous stage in the RNA-sequencing, along with electrophysiology tests in differing estrous stages to test this. We believe this, along with our data showing increased EPSCs strength in stressed females, suggests the need for further research in the effectiveness of CVMS in females.

In terms of sex differences in the transcriptome of the adBNST, the gene ontology data shows that male-biased genes are related to transcriptional regulation while female-biased genes are related to synaptic processing, suggesting that the BNST in the females may be more plastic, especially at their synapse. For example, female-biased genes include Gsg1l and Cacng3 which both regulate AMPA signaling. Gsg1l acts to negatively regulate AMPA receptors by slowing the deactivation and desensitization of AMPA receptors (Greger et al., 2017). Cacng3 is a transmembrane AMPA regulatory protein (TARP) which is a family of proteins that positively regulate AMPA receptors. TARPs act by promoting AMPA receptor trafficking to the synapse and stabilization in the postsynaptic density (Coombs & Cull-Candy, 2009). With the transcripts of both negatively and positively regulating proteins being female-biased, suggests that the neurons in the female adBNST are more primed for synaptic plasticity than males. This hypothesis is further supported by other genes involved in synaptic plasticity and axonogenesis, such as Anks2b and Pcp4, are female-biased.

However, while the female adBNST may be primed for synaptic plasticity, the data suggests that only the transcriptome of the male adBNST is sensitive to stress. Of the genes that are regulated by stress, many are related to synaptic signaling, as evidenced by the gene ontologies and specific genes. For example, many downregulated genes are signaling molecules, their receptors, or regulators of their receptors, such as Crh, corticotropin releasing hormone, Slc1a6, a glutamate transporter (Gegelashvili et al., 2000), and Arrb2, and inhibitor of arrestin receptors (Beaulieu et al., 2005). Additionally, some are related to synaptic maintenance and signaling in general, such as Vamp1, which is involved in synaptic vesicle exocytosis (Liu et al., 2011), Otof, which is involved in synaptic vesicle priming (Johnson & Chapman, 2010), and Dpysl2, which is involved in axon growth (Inagaki et al., 2001). Upregulated genes, also fall a similar pattern of affecting synaptic transmission and catecholamine signaling. Examples include: Drd2, a dopamine receptor (Noble, 2000), Adra2c, an adrenergic receptor (Bylund, 1992), and Camk2n1 and Igsf11, which are both involved in long-term potentiation (Jang et al., 2016; Astudillo et al., 2020). This suggests that chronic stress causes major changes in the regulation of the synapse in the adBNST of males.

However, one must ask if this is truly a reflection of chronic stress or acute stress. Our previous publication showed no transcriptomic differences due to CVMS 18 h post-stress (Degroat et al., 2024) but in the current study, we found over 200 DEGs in the males 1 h post-stress. However, the transcriptional effects of chronic stress did lead to sustained (17–24 h post-stress) neurophysiological outcomes in males. It is possible that our results are due specifically to chronic stress and the transcriptome is so transient that difference are lost quickly when there is no stress, or it could be equally possible that this is due to the mild acute stress. One study using chronic mild stress paradigms found sustained differences in the nucleus accumbens 24 h after the last stressor (Ma et al., 2019). Two other studies found stress sensitive genes in the amygdala, hippocampus, and hypothalamus after chronic stress but waited to collect tissue until after multiple days of behavior testing (Karisetty et al., 2017; Yang et al., 2023). Comparisons between chronic stress studies can be difficult though as the chronic stress paradigms often differ significantly between labs and the transcriptome of different brain regions or cell types may have differing reaction times to stress. Additionally, Karisetty et al. and Yang et al. used behavior tests to ensure the mice expressed stress-induced behavioral changes before transcriptomic analysis, which we did not do, further complicating comparisons. Further experiments may be necessary, such as measuring the adBNST transcriptome in samples collected one hour after the typical last stressor (1 h wet bedding) without the chronic stress paradigm.

As the transcriptome data suggest that only the male adBNST is sensitive to stress, this is also supported by chronic stress affecting the M-current in males but not females. Chronic stress significantly decreased the strength of the M-current in males, supporting our previously published findings (P. Hu et al., 2020). The M-current is an intrinsic property of the neuron and is produced by voltage-gated K⁺ channel, the KCNQ channels. This current is inhibitory and is significant as it is one of the few currents that is active at physiologically relevant voltages and as such has a strong influence on the resting membrane potential of the cell (Marrion, 1997). This weakening of the M-current by stress infers that these neurons are more likely to fire, possible leading to increased signaling by the BNST and increased release of CRH or GABA. This further supports our hypothesis that CVMS directly affects the adBNST neurons in males but not females.

The females suggest there may be different areas of the brain that are more important for chronic stress processing. As the transcriptome data suggests, the adBNST in females is not sensitive to CVMS as the only gene affected by it was Rbm3, a gene that is related to cellular stress (Hu et al., 2022). Additionally, the electrophysiology data supports examining other regions critical to stress or steroid hormone signaling as we saw significant differences in the females to EPSCs. EPSCs being an extrinsic property shows that neurons other than the CRF-expressing adBNST neurons are possibly being affected by stress and we are seeing that downstream through their synaptic transmission (Egli & Winder, 2003). However, it is important to recognize that the strength of EPSCs can be modulated either in the pre-synapse or in the post-synapse. In the pre-synapse, EPSC strength can be modulated by the loading of vesicles (Wu et al., 2007), the docking position of vesicles (Pulido & Marty, 2017), and the number of vesicles primed for release (He et al., 2009); whereas the post-synapse can modulate the strength of EPSCs by altering the number of receptors at the synapse, modulating the effects of those receptors, and increasing dendritic arborization (Auger & Marty, 2000). While both sites can affect the strength of EPSCs, computational neuroscience suggests that the pre-synapse has the greater control of the strength of EPSCs (Auger & Marty, 2000; Ventriglia & Di Maio, 2002, 2003). This has also been supported experimentally which showed that mature neurons typically use pre-synaptic methods to modulate EPSC strength while younger neurons utilize post-synaptic methods (Han & Stevens, 2009). There are limitations to these studies though as they have been done almost exclusively in hippocampal neurons or at the neuromuscular junction as these synapses are well-characterized. It is possible, that in the BNST, post-synaptic modulation is more important., however, since we do not see any changes in receptors or related proteins in the RNA-sequencing data, we infer that this change in females is related to the pre-synapse. More research should be conducted to confirm this though. Furthermore, our data is supported by the lack of response to CRH receptor activation either in the M-current or EPSCs, demonstrating that adBNST CRH neurons do not respond to other CRH neurons. However, this does not preclude other types of neurons in the BNST causing these changes, One potential site is the posterior BNST which expresses high levels of steroid receptors and send projections to the adBNST (Becker et al., 2008; Sun et al., 2023).

Our data suggests that chronic stress is processed differently in the brains of male and female mice. However, surprisingly, it also suggests that they’re not only processed differently but potentially processed by separate areas of the brain. Further research is needed to elucidate which other areas are important and how these differences in signaling are related to the development of mood disorders. Additionally, this experiment only used one type of stress paradigm. It is well known that different types of stress are processed differently, thus further studies are also needed to be done to determine how the processing of psychosocial stress may differ from physiological stress for example.