Sex and stress govern the function and innervation of a basolateral amygdala to nucleus accumbens corticotropin releasing hormone/GABA expressing projection

Abstract

Motivated behaviors are executed by refined brain circuits. Early life adversity (ELA) is a risk for human affective disorders involving dysregulated reward behaviors. In mice, ELA causes anhedonia-like behaviors in males and augmented reward motivation in females, indicating sex-dependent disruption of reward circuit operations. We recently identified a long-range corticotropin-releasing hormone (CRH) expressing GABAergic projection from basolateral amygdala (BLA) to nucleus accumbens (NAc) that mediates reward-seeking deficits in adult ELA males—but not females. We verified a similar cell identity and electrophysiology of the projection across sexes. To probe the sex-specific role of this projection in reward behaviors, adult male and female CRH-Cre mice raised in control or ELA conditions received excitatory or inhibitory Cre-dependent DREADDs in BLA, and clozapine N-oxide or vehicle to NAc medial shell during reward behaviors. Using tissue clearing, light sheet fluorescence microscopy and deep learning pipelines, we mapped brain-wide BLA CRH⁺ axonal projections to identify potential sex differences in its innervation. Chemogenetic manipulations in male mice demonstrated inhibitory effects of the CRH⁺ BLA-NAc projection on reward behaviors, whereas neither excitation nor inhibition influenced female behaviors. Molecular and electrophysiological cell-identities of the projection did not vary by sex. By contrast, comprehensive whole-brain mapping uncovered significant differences in NAc innervation patterns that were both sex and ELA-dependent, as well as selective changes of innervation of other brain regions. The CRH⁺/GABA BLA-NAc projection that influences reward behaviors in males differs structurally and functionally in females, uncovering potential mechanisms for the profound sex-dependent impacts of ELA on reward behaviors.

Introduction

Early life adversity (ELA) arising from poverty, trauma, and chaotic environments affects millions of children worldwide (Merrick et al., 2018). Studies link ELA to adverse cognitive and emotional outcomes (Hailes et al., 2019; Short and Baram, 2019; Nelson et al., 2021; McKay et al., 2022), including disruptions in reward circuitry function (Pizzagalli et al., 2005; Bolton et al., 2018; Birnie et al., 2020; Kangas et al., 2022) and impaired development of associated brain regions (Gee et al., 2013; Callaghan and Tottenham, 2016; Brieant et al., 2023). Mental health disorders characterized by disruptions in the brain’s reward circuits often include anhedonia, a construct denoting decreased pleasure or desire for rewards (Pizzagalli, 2014). In humans, the relationship between ELA and dysregulated reward behaviors differs by sex (Bale and Epperson, 2015; Becker and Chartoff, 2019), with women more susceptible to craving comfort food and developing eating and opioid use disorders (Johnson et al., 2002; Evans et al., 2017) whereas men are more prone to alcohol use disorders (Colman et al., 2013; Evans et al., 2017). These differences may derive from sex-dependent characteristics of reward circuitry operations (Bale and Epperson, 2015; Becker and Chartoff, 2019) which may be coupled with sex-biased developmental vulnerabilities to ELA (Bale and Epperson, 2015; Hodes and Epperson, 2019; Bath, 2020; Luby et al., 2020; Parel and Peña, 2022). Establishing a direct causal link between ELA and anhedonia, and aberrantly augmented reward motivation in humans is challenging. Conversely, our rodent model of ELA consistently induces sex-specific disturbances in reward circuits, including anhedonia in males and augmented motivation for reward cues in females (Molet et al., 2016a, 2016b; Bolton et al., 2018; Levis et al., 2021, 2022; Birnie et al., 2023), providing a framework for addressing the mechanisms underlying these sex differences.

The brain is organized in circuits that execute complex behaviors, including those involved in reward. We hypothesized that circuits expressing stress-sensitive molecules such as the peptide corticotropin-releasing hormone (CRH), which is often involved in stress responses and adaptation (Roozendaal et al., 2002; Joëls and Baram, 2009; Hupalo et al., 2019), might be poised to be influenced by ELA. We initially mapped CRH projections to a key reward center in the brain, the nucleus accumbens (NAc), and identified a CRH-expressing (CRH⁺) GABA-ergic projection from the basolateral amygdala (BLA) to NAc shell (Itoga et al., 2019). This was exciting because CRH modulates reward-related behaviors in NAc, and its actions there are context-dependent, influenced by prior or ongoing stress (Peciña et al., 2006; Lemos et al., 2012). We then established that activating the CRH⁺ BLA-NAc projection reduced reward behavior in adult control male mice, and silencing it in ‘anhedonic’ ELA male mice restored reward behaviors to control levels (Birnie et al., 2023). This discovery is significant because it provides a mechanistic link between ELA and mood disorders (Chen and Baram, 2016; Molet et al., 2016a; Birnie and Baram, 2025).

Here, we investigate this novel, cell type-specific projection’s role in modulating motivated reward behaviors in female mice and the fundamental sex and ELA-dependent differences in the projection’s structure and function.

Methods and Materials

Mice

B6(Cg)-Crhtm1(cre)Zjh/J (CRH-ires-CRE, Jackson 012704) (Taniguchi et al., 2011; Chen et al., 2015) mice were bred in-house, weaned and group-housed by sex on a 12-hour light-dark cycle (lights on 7 am). Post-surgery, single-housed males and group-housed females were moved to a reverse light-dark cycle (lights on 12 AM). Mice used in each figure were separate cohorts and all were a CRH-ires-Cre strain (Chen et al., 2015) Procedures were approved by University of California-Irvine’s IACUC (AUP 24–104, 21–128) and adhered to NIH guidelines.

The Limited Bedding and Nesting ELA Model

To study ELA, we used a model of resource scarcity developed by our group and adopted world-wide, the limited bedding and nesting (LBN) protocol (Rice et al., 2008; Ivy et al., 2008; Molet et al., 2016a). Multiparous dams and pups (minimum 4–maximum 8; sex balanced) were randomly assigned to cages with either customary or limited bedding and nesting materials. LBN cages had a plastic-coated mesh platform (McNichols 4700313244) ~2.5 cm above the floor, and reduced bedding (150 ml), and nesting material (0.5 vs 1 nestlet). Cages were undisturbed during P2–P9 in a ventilated, quiet area. All mice returned to typical cages on P10 and weaned on P21.

Surgical Procedures

2–3 month old CRH-ires-Cre mice were anesthetized with 5% isoflurane and maintained at 1–1.5% on a stereotaxic apparatus with a heating pad. Crown fur was shaved, the skin sanitized (betadine and 70% ethanol), and skull exposed. Injection sites were drilled, and viral vectors were delivered via pulled pipettes (Drummond 2-000-010) at 100 nl/min, leaving the pipet in place for 5 min to minimize backflow. Post-surgery, mice received buprenorphine (Patterson Vet 07-894-9214) and were monitored for two days. To allow viral expression in axonal processes (Itoga et al., 2019), we waited 6 weeks prior to chemogenetics and 8 weeks prior to electrophysiology, optogenetics and brain clearing. For the latter, AAV1-EF1a-DIO-hChR2(H134R)-EYFP (Addgene 20298-AAV1) was injected into medial BLA bilaterally (0.2 μl/side; A/P −1.4mm, M/L ±4.25mm, D/V −4.60mm, 15° angle). For chemogenetic experiments, AAV2-hM3D(Gq)-mCherry (excitatory, Addgene 44361-AAV2) or AAV2-hM4D(Gi)-mCherry (inhibitory, Addgene 44362-AAV2) were bilaterally injected into medial BLA (0.2 μl/side). Guide cannulas were implanted bilaterally into medial NAc shell (A/P +1.2mm, M/L ±1.5mm, D/V −3.5mm, 12° angle). A dental cement headcap (Patterson Dental Orthojet Liquid 74594412; Powder 74598371) secured the cannulas with skull screws (Antrin AMS1201BINDSS).

Behavioral and Chemogenetic Experiments

Behavior tests were performed at ages 3–5 months. Experiments assessing the impact of rearing alone involved naïve mice. Chemogenetic experiments were performed on mice that underwent surgery for virus-mediated DREADDs and cannulas as described above. Experiments were initiated within the first 2 hr of the active, dark phase (Robins et al., 2020; Nelson et al., 2021) in a dimly lit room. All experiments were conducted in standard mouse cages (34 × 18 cm), either the home cage (males), or individual cages following habituation (females). For chemogenetic experiments, 0.2 μl clozapine- N-oxide (CNO, 1 mM, Hello Bio HB6149) (Mahler et al., 2014, 2019) dissolved in saline (VetOne V1510223) was bilaterally infused into the medial NAc shell via an indwelling cannula.

Palatable Food Task

This test determines palatable food consumption (Cocoa Pebbles, Post^R) over one hour in non food-restricted mice, providing a measure of reward motivation (Birnie et al., 2023). Mice were habituated to Cocoa Pebbles for three days to minimize novelty effects (Greiner and Petrovich, 2020) as follows: On Day 0, approximately 1 g of Cocoa Pebbles was placed in the home cage overnight; consumption was visually confirmed but not measured. On Days 1–3, mice were placed in individual cages in the testing room for 1 hour and provided with 1 g of Cocoa Pebbles. Consumption was measured after 1 hour. For baseline behavioral measurements, Day 4 (Test Day) followed the same procedure as Days 1–3. For chemogenetic manipulations, male mice, on Day 4 (Test Day 1), received counterbalanced infusions of CNO or saline into the NAc, then were placed in individual cages in the behavior room with 1 g Cocoa Pebbles, and consumption was measured after 1 hour. Following two days of rest and three additional days of habituation, the same procedure was repeated on Day 11 (Test Day 2) with the opposite treatment. For females, Day 4 (Test Day 1) followed the same infusion and testing procedure as males, but Test Day 2 occurred after a one-day rest (Day 6), without additional habituation, to minimize observed increases in average consumption with repeated exposure. Inter-test intervals (one day for females, one week for males) allowed sufficient washout of CNO or vehicle (Smith et al., 2016), and counterbalancing controlled for infusion order effects.

Scent of a Mouse (SoaM) Task (also known as the urine sniff test)

This task measures the active approach and duration of interest of mice toward a Q-tip scented with urine of a member of the opposite sex (scent of a mouse, SoaM) versus one scented with almond (neutral scent). Female subjects were exposed to Q-tips scented with male urine (Roberts et al., 2010), male subjects were exposed to Q-tips scented with estrus female urine (Birnie et al., 2023). Urine was collected on the test day or stored, capped, at 4°C for <3 days. Frequency of approaches toward the urine- or almond-scented Q-tip and the durations of sniffing were recorded. On Day 1, mice received CNO or saline (counterbalanced) into the NAc, then moved to cages with Q-tips (60 μl each of urine and almond scents) placed at opposite corners. Exploration of the SoaM (number of approaches to each Q-tip and total duration spent sniffing each Q-tip) was assessed manually over 3 min. After one rest day, the procedure was repeated with counterbalanced infusions (CNO or saline). While typically performed in receptive (estrus or proestrus) females (Cheetham et al., 2007; Miller et al., 2022; Ramm et al., 2008; Roberts et al., 2010) we conducted this test throughout the estrous cycle, based on our prior experience (see Fig.S1).

Electrophysiology Slice Preparation

Mice were deeply anesthetized with isoflurane and quickly decapitated. Acute horizontal slices (300 μm) encompassing NAc and BLA were obtained using a vibratome (Leica V1200S) in an ice-cold sucrose cutting solution containing (in mM): 228 sucrose, 11 glucose, 26 NaHCO3, 1.2 NaH2PO4, 2.5 KCl, 5 Na-ascorbate, 3 Na-pyruvate, 10 MgSO4–7H2O, and 0.5 CaCl2 (305–310 mOsm, pH 7.4). Slices equilibrated in a homemade chamber for 25–30 min (34°C) then 45 min in room temperature aCSF containing (in mM): 119 NaCl, 26 NaHCO3, 1 NaH2PO4, 2.5 KCl, 11 glucose, 10 sucrose, 1.3 MgSO4–7H2O, and 2.5 CaCl2 (290–300 mOsm, pH 7.4), before being transferred to a recording chamber. Solutions were continuously bubbled with 95% O2/5%CO2.

Whole-Cell Patch Clamp

Recordings were performed in the medial NAc shell using a Multiclamp 700B, Digidata 1550B, and Clampex 11 (Molecular Devices). Recordings were conducted in voltage-clamp mode at 31°C, low-pass filtered at 2 kHz, and digitized at 10 kHz. Borosilicate glass pipettes (3–4 MΩ, Molecular Devices) filled with internal solution (295–305 mOsm, pH 7.4, adjusted with CsOH) contained (in mM): 135 CsMeSO4, 8 CsCl, 10 HEPES, 0.25 EGTA, 5 Phosphocreatine, 4 MgATP, 0.3 NaGTP, and 1 mg/ml NeuroBiotin (295–305 mOsm, pH 7.4 with CsOH). Access resistance (Ra) was monitored, and recordings with Ra changes >20% were excluded. Cells were visualized with infrared DIC microscopy (Olympus BX51WI). Neurons were held at 0 mV to record optically evoked inhibitory postsynaptic currents (oIPSCs) using 488 nm LED light. After 5 minutes of stable oIPSCs, picrotoxin (100 μM) was superfused until oIPSCs were blocked. Neurons were then held at −60 mV to record optically evoked excitatory postsynaptic currents (oEPSCs). Stable oEPSCs were followed by the superfusion of CNQX (10 μM) and AP5 (50 μM). Slices were fixed in 4% PFA at 4°C overnight, then stored in 0.1M PBS for histological processing.

Immunohistochemistry

Concurrent immunolabeling of CRH and GAD was performed as described previously (Yan et al., 1998; Chen et al., 2015), and concurrent immunolabeling of CRH and virus-expressed fluorophore followed (Birnie et al., 2023). Briefly, sections were first incubated with a rabbit anti-CRH antiserum (PBL rC68, courtesy Dr. Paul Sawchenko, Salk Institute, La Jolla) (1:20,000, 7 days, 4°C), and then treated with HRP-conjugated anti-rabbit IgG (1:1,000, PerkinElmer) for 1.5 hours. Cyanine 3-conjugated tyramide (1:150, Akoya) was applied in the dark for 5–6 minutes. After CRH detection, sections were exposed to a mixture of mouse anti-GAD67 (1:250, Santa Cruz, sc-28376) and anti-GAD65 (1:1,000, Boehringer Mannheim, #1522 825) (3 days, 4°C), or to a mouse monoclonal anti-RFP (1:2,000, Rockland, #200-301-379) to detect AAV2-mCherry infected neurons, and visualized using anti-mouse IgG conjugated to Alexa Fluor 488 (1:400, Invitrogen).

Brain Clearing

Brains were processed using the Adipo-Clear protocol (Chi et al., 2018) with slight modifications. Mice were transcardially perfused and brains hemisected ~1 mm past the midline and postfixed overnight in 4% paraformaldehyde (Sigma Aldrich 30525-89-4) at 4°C. The following day, samples were dehydrated with a gradient of methanol (MeOH, Fisher Scientific 67-56-1)/B1n buffer (1:1000 Triton X-100, 2% w/v glycine, 1:10,000 NaOH 10N, 0.02% sodium azide) for 1 hr for each step (20%, 40%, 60%, 80%) on a nutator. Samples were then washed with 100% MeOH 2x for 1 hr each and then incubated in 2:1 dichloromethane (DCM):MeOH solution overnight. The following day, the samples were washed 2x for 1 hr in 100% DCM, followed by three washes of 100% MeOH (30 min, 45 min, 1 hr). Samples were bleached for 4 hr in 5:1 H2O2/MeOH buffer. A cascade of MeOH/B1n washes (80%, 60%, 40%, 20%; 30 min each) rehydrated the samples, followed by a 1 hr wash in B1n buffer. Tissue was permeabilized in 5% DMSO/0.3 M Glycine/PTxWH (1 hr then 2 hr). Samples were washed with PTxwH for 30 min and incubated in fresh PTxwH overnight. The following day, we performed two PTxwH washes (1 hr, then 2 hr). Samples were incubated in primary GFP antibody (GFP-1020, AVES Labs NC9510598) at 1:2000 in PtxwH shaking at 37°C for 11 days, washed in PtxwH (2× 1 hr, 2× 2 hr and then once per day for 2 days). Samples were then incubated in a secondary antibody (AlexFluor 647, ThermoFisher Scientific A78952) for 8 days shaken at 37°C. Samples were washed in PTxwH (same as after primary antibody). Samples were then washed in 1x PBS twice (1 hr, 2×2 hr, overnight). Samples were dehydrated in a gradient of MeOH:H₂O (20%, 40%, 60%, and 80%; 30 min each and then 100% 30 min, 1 hr, 1.5 hr). Samples were incubated overnight in 2:1 DCM:MeOH on a nutator, washed in 100% DCM (2× 1 hr). Samples were cleared in 100% DBE. DBE was changed after 4 hr. Samples were stored in DBE in a dark location at room temperature. Imaging took place at least 24 hr after clearing.

Whole-Brain Imaging

Brain samples were imaged on a light sheet microscope (Ultramicroscope II, LaVision Biotec) equipped with a sCMOS camera (Andor Neo) and a 2x/0.5 NA objective lens (MVPLAPO 2x) with a 6 mm working distance dipping cap. Image stacks were acquired at 0.8x optical zoom using Imspector Microscope v285 controller software. We imaged using 488 nm (laser power 20%) and 640 nm (laser power 50%) lasers. The samples were scanned with a step-size of 3 μm using the continuous light sheet scanning method with the included contrast adaptive algorithm for the 640 nm channel (20 acquisitions per plane), and without horizontal scanning for the 488 nm channel.

DeepTraCE Pipeline

Whole-brain image stacks were analyzed using python based DeepTraCE GUI (https://github.com/jcouto/DeepTraCE). Briefly, image stacks in the 640 nm channel were segmented using the 3D U-net based machine learning pipeline TrailMap (Friedmann et al., 2020) with a model trained to recognize clearly delineated axons (Model 1) described in Gongwer et al., 2023. Following segmentation, the 488 nm autofluorescence channel and axon segmentation were converted to 8-bit, scaled to 10 μm resolution and rotated to match the reference atlas. The autofluorescence channel was registered using elastix to the Gubra Lab light sheet fluorescence microscopy (LSFM) atlas average template, which has annotations that match the Allen CCF (Klein et al., 2010; Wang et al., 2020a; Perens et al., 2021). The same transformation was applied to the segmented brain using transformix and converted to 8-bit .tif format. Axons were then skeletonized (reduced to a single pixel thickness) for quantification, as described previously (Friedmann et al., 2020; Gongwer et al., 2023).

To account for variability in BLA viral expression across animals, we normalized all axon counts to the total number of labeled pixels across the whole brain (Fig. S2). Regional axon innervation was quantified by counting the number of skeletonized pixels in each brain region above a threshold, then dividing this pixel count by the total number of pixels in a region. Regions were defined by a collapsed version of the Gubra Lab LSFM atlas. The atlas was cropped on the anterior and posterior ends to match the amount of tissue visible in our data. Fiber tracts, ventricular systems, cerebellum, pons, medulla and olfactory bulb were excluded from analysis.

Axon Quantification

Medial-lateral axon distributions within regions were calculated in MATLAB by binning the whole-brain image into 100 μm voxels, calculating the percentage of segmented pixels within each voxel normalizing for total fluorescence as above. The averaged summation of axon counts from each group was then averaged and plotted along with the SEM.

Axon Visualization

To visualize axons in the NAc of representative brains (Fig. 3B), z-projections of raw light sheet were created in FIJI by scaling images to a 4.0625 μm space, virtually reslicing images in the coronal plane and performing maximum intensity z-projections of 30 μm depth followed by local contrast enhancement. To visualize axons at various coronal levels (Fig. 4A), representative brains from each group were resliced in the coronal plane, and then 80 μm thick z-projections of the skeletonized and registered axons were produced from the resliced brains.

Sex Considerations

To consider the possible effects of the estrous cycle and associated hormonal fluctuations on the function of the projection and female reward behaviors, (McCarthy et al., 2012; Shansky and Woolley, 2016), females were swabbed on the test day for a vaginal smear (<5 seconds).

Quantitative Approaches and Statistical Analyses

All statistical comparisons and analyses were performed in GraphPad Prism V10 or MatLab 2022a. The specific experimental design and statistical analyses can be found in the figure legends and Table S1 and Table S2.

Results

ELA influences reward behaviors in a sex-dependent manner.

As adults, male and female mice exposed to ELA (Fig. 1A) exhibited profound sex-dependent changes in reward behaviors. Male ELA mice consumed less palatable food (Fig. 1B) and explored the scent of a female in heat in the scent of a mouse task (SoaM) less than control male mice (Fig. 1C). In contrast, adult female mice that were reared in ELA conditions showed augmented reward behaviors compared to control females, evidenced by increased consumption of Cocoa Pebbles (Fig. 1B) and increased number of approaches towards the scent of urine from a male mouse (Fig. 1C). Of note, in males there was a significant difference between ELA and controls in both the number of approaches to this sex cue (Fig. 1C) and the duration of sniffing (Fig. 1D). In females, by contrast, the ELA and control groups did not differ in the duration of sniffing (Fig. 1D), a measure of consummatory pleasure. Rather, ELA females approached the male scent more frequently (Fig. 1C), a measure of motivation for reward. This suggests potential sex-specific reward behavior disruptions induced by ELA in male and female mice.

Fig. 1: Sex differences in the impact of early life adversity (ELA) on adult reward behaviors.

(A) Schematic of the limited bedding and nesting cages (LBN) model of ELA. Adult male mice raised in ELA cages exhibited “anhedonia-like” reward behaviors compared to typically reared mice (CTL) in (B) palatable food consumption, (C) number of approaches towards the scent of a mouse (SoaM) of the opposite sex, and (D) duration spent sniffing this sex cue. In contrast, adult ELA female mice had augmented motivated reward behaviors compared to control females, with increased (B) palatable food consumption and (C) number of approaches towards the scent of a male mouse but not increased duration of sniffing (D), a measure of consummatory pleasure. Cycle-phase or litter had no influence on data (Fig. S1). In B–D, dots represent individual mice, error bars represent mean ± SEM. 2-way ANOVA (B: F_{rearing × sex}(1,76)=35.2, p<0.01; F_sex(1,76)=18.14, p<0.01; F_rearing(1,76)=0.7117, p=0.40; n, CTL male=20 mice/7 litters, ELA male=17 mice/7 litters, CTL female=21 mice/5 litters, ELA female=22 mice/7 litters; C: F_{rearing × sex}(1,56)=17.1, p<0.001; F_sex(1,56)=2.74, p=0.103; F_rearing(1,56)=0.773, p=0.383; n, CTL male=15 mice/5 litters, ELA male=16 mice/9 litters, CTL female=12 mice/5 litters, ELA female=17 mice/6 litters; D: F_{rearing × sex}(1,56)=2.88, p=0.095; F_sex(1,56)=12.5, p<0.001; F_rearing(1,56)=5.72, p=0.020; n, CTL male=15 mice/5 litters, ELA male=16 mice/9 litters, CTL female=13 mice/5 litters, ELA female=16 mice/6 litters). See Table S1 for full statistical analyses and tests. CTL = control. ELA = early life adversity. LBN = limited bedding and nesting.

The CRH expressing GABAergic BLA-NAc projection mediates reward behaviors in male but not female mice.

In a recent paper, we discovered a CRH⁺/GABA BLA-NAc pathway which inhibits reward-seeking behaviors and mediates the effects of ELA on reward behaviors in male mice (Birnie et al., 2023). When the pathway was chemo- or optogenetically stimulated in control male mice, it suppressed reward behaviors. Here, we recapitulated these findings (Fig. 2A–B). Importantly, we report here the results of systematic manipulation of the projection in female mice, demonstrating profound sex differences. Chemogenetic activation of the CRH⁺ BLA-NAc pathway in control female mice did not affect palatable food consumption in contrast to control males (Fig. 2C). Similarly, exciting the projection in ELA females had little effect (Fig. 2C). Chemogenetic inhibition of the CRH⁺ BLA-NAc pathway in female control and ELA mice failed to significantly alter palatable food consumption (Fig. 2D), in stark contrast to its effect in rescuing reward deficits in ELA males (Birnie et al., 2023).

Fig. 2: Chemogenetic manipulation of the CRH+ BLA-NAc projection in vivo influences reward behavior in male but not female control and ELA-exposed adult mice.

(A) Viral strategy to chemogenetically target CRH⁺ BLA-NAc projection neurons. DREADD (Gq) excitation of CRH⁺ BLA-origin axon terminals in the NAc with 1 mM CNO reduced (B) palatable food consumption in control male mice but not in ELA male mice, in which the pathway is overactive (Birnie et al., 2023). (C) Excitation of the projection in control and ELA females had no effect. (D) Chemogenetic (Gi) inhibition of CRH⁺ BLA-origin axon terminals in the NAc with 1 mM CNO did not alter palatable food consumption in control or ELA female mice. In B–D, lines represent individual mice, bars represent mean. 2-way RM-ANOVA (B: F_{rearing × CNO}(1,13)=3.372, p=0.09; F_rearing(1,13)=5.182, p=0.04; F_CNO(1,13)=5.083, p=0.04; n, CTL male=9 mice/4 litters, ELA male=6 mice/3 litters; C: F_{rearing × CNO}(1,21)=0.1437, p=0.71; F_rearing(1,21)=5.187, p=0.03; F_CNO(1,21)=0.2984, p=0.59; n, CTL female=15 mice/5 litters, ELA female=8 mice/6 litters; D: F_{rearing × CNO}(1,15)=0.7343, p=0.40; F_rearing(1,15)=5.275, p=0.04; F_CNO(1,15)=1.851, p=0.19; n, CTL female=7 mice/3 litters, ELA female=10 mice/5 litters). CNO = clozapine N-oxide. CTL = control. ELA = early life adversity.

The CRH⁺ BLA-NAc projection is GABAergic in both male and female mice.

In male mice, CRH⁺ BLA-NAc neurons inhibited reward behaviors, whereas in females, activating the projection had no detectable effects on reward behaviors. Therefore, we tested if the projections might differ across sexes in terms of cell identity or electrophysiological properties. Specifically we tested whether, in females, the projection also made GABAergic synapses onto target NAc cells, as found in males. We injected CRH-ires-Cre mice with DIO-ChR2 in the BLA, and obtained whole-cell patch-clamp recordings from target neurons in the NAc medial shell. Inhibitory postsynaptic currents (IPSC) were optically evoked with a 2 ms light pulse in the NAc (Fig. 3). Representative IPSCs traces are shown in Fig. 3A and 3E. After 5 min, the GABA^A receptor antagonist picrotoxin was washed on, blocking IPSCs in both males and female slices across all groups as apparent in the time-course data of normalized oIPSCs amplitudes (Fig. 3B, 3F) and time-point analysis (Fig. 3C, 3G). Raw oIPSCs amplitudes pre and post picrotoxin are shown in Fig. 3D and 3H. These data indicate that in both sexes, optically evoked currents from CRH⁺ BLA-origin axons in NAc are inhibitory and are mediated by GABA^A receptors. In further support of the GABAergic nature of the CRH⁺ BLA projection, all CRH⁺ cells in the BLA expressed the GABAergic markers GAD 65/67 (Fig. 3I–K), including those projecting to the NAc (Fig. 3L–M).

Fig. 3: NAc-projecting CRH+ BLA neurons are GABAergic in both males and females, evident by both electrophysiological and neurochemical markers.

Whole-cell patch-clamp recordings were obtained from NAc medial shell neurons while inhibitory postsynaptic currents (IPSCs) were optically-evoked from CRH⁺ BLA neurons with a 2 ms light pulse at 488 nm. Representative traces of optically evoked IPSCs (oIPSCs) in males (A) and females (E). Time-course plot of normalized oIPSCs amplitudes before and after application of picrotoxin in (B) males and (F) females. Timepoint analysis of normalized oIPSC amplitudes at 5 min after start of recording compared to 15 min after picrotoxin application for (C) males and (G) females. (D,H) oIPSCs amplitudes pre and post picrotoxin. In B and F, dots and error bars represent mean ± SEM. In C–D and G–H, dots represent individual cells, bars represent mean. 2-way RM-ANOVA (C: F_{rearing × PTX}(1,15)=0.266, p=0.614; F_rearing(1,15)=0.107, p=0.748; F_PTX(1,15)=1408, p<0.001; D: F_{rearing × PTX}(1,15)=2.25, p=0.154; F_rearing(1,15)=2.31, p=0.149; F_PTX(1,15)=43.1, p<0.001; B–D: n, CTL male=8 cells/4 mice/2 litters, ELA male=9 cells/4 mice/2 litters; G: F_{rearing × PTX}(1,19)=0.721, p=0.406; F_rearing(1,19)=0.758, p=0.395; F_PTX(1,19)=939, p<0.001; H: F_{rearing × PTX}(1,19)=0.0844, p=0.775; F_rearing(1,19)=0.0328, p=0.858; F_PTX(1,19)=65.0, p<0.001; F–H: n, CTL female=10 cells/4 mice/2 litters; ELA female=11 cells/4 mice/2 litters). (I–K) Concurrent immunolabeling of CRH and both isoforms of GAD (67 and 65) in the BLA of a CTL male (J) and CTL female (K). High magnification photomicrographs indicate that all CRH⁺ cells (magenta) also express GAD (arrowheads), whereas numerous GABAergic cells are devoid of CRH expression (arrows). In I, scale bars = 100 μm. In J–K, scale bars = 35 μm, arrows = GAD67/65 only (green), arrowheads = colocalization of CRH and GAD67/65. (L–M) AAV2 infected cells in the BLA co-express endogenous CRH. Representative images (L) indicating that AAV2 injection is limited to the BLA. Cells in the CeA are not virus-infected. Boxed area in the BLA was magnified in the right panel (M) to show the colocalization between AAV2 and CRH expressions. In L, scale bars = 100 μm. In M, scale bars = 25 μm, arrowheads = AAV2 infected cells in the BLA co-expressing CRH. aCSF = artificial cerebrospinal fluid. PTX = picrotoxin. CTL = control. ELA = early life adversity.

Sex and rearing-dependent axonal innervation patterns of BLA origin CRH⁺ cells within the NAc.

The results above indicate that there are sex-dependent differences in the function of the CRH⁺ BLA-NAc projection that are not explained by neurotransmitter identity or by the electrophysiology of the originating cells. To investigate if structural differences in this pathway underlie the distinct sex-specific behaviors, we utilized a whole brain imaging approach to determine if sex or rearing alter the innervation of CRH⁺ BLA axonal projections within the NAc. Using a viral approach, we fluorescently labeled CRH⁺ BLA axons in adult male and female CRH-Cre mice who experienced typical (control) or ELA rearing (Fig. 4A). We then cleared the brains, imaged them using light sheet fluorescence microscopy and quantified the brain-wide axonal projections of CRH⁺ BLA neurons (Fig. 4B).

Fig. 4: Axonal innervation patterns of the BLA origin CRH+/GABA projection in the nucleus accumbens (NAc).

(A) Viral strategy for labeling axonal projections of CRH⁺ BLA neurons and (B) whole brain analysis pipeline. (C) Coronal view of 80 μm z-projections of skeletonized axons determined by DeepTraCE (axons, white) from example brains. In C, scale bars = 200 μm. (D) Differences in sex and rearing specific innervation patterns in NAc and quantification of differences in axonal innervation patterns among all groups. Results were not due to differences in the numbers of CRH⁺ neurons in the BLA (Fig. S2) and considered the possibility that regions with low axonal density might be more likely to show more significant effects (Fig. S3). In D, error bars represent mean ± SEM. (E) Quantification of axonal innervation density along the anterior-posterior axis of the NAc. 2-way ANOVA (D: F_rearing(1,11)=5.365, p=0.04; F_sex(1,11)=1.088, p=0.32; F_{rearing × sex}(1,11)=6.942, p=0.02; n, CTL male=4 mice/2 litters, ELA male=4 mice/1 litter; CTL female=4 mice/1 litter, ELA female=3 mice/1 litter). CTL = control. ELA = early life adversity. (F-K) The BLA origin CRH⁺/GABA projection targets both medium spiny neurons (MSNs) and interneurons in the NAc. (F). A low-magnification micrograph showing GFP, indicating channelrhodopsin-expression at the site of injection, the BLA, but not central amygdala nucleus (CEA). (G) Channelrhodopsin expressing fibers in the NAc (green), and CRH receptor type 1 expressing neurons (red). (H) high resolution confocal image showing the NLA-origin projection terminals (green) abutting dopamine receptor type 1(D1) expressing, presumed MSNs in the NAc. The frames in the main panel delineate the more detailed images on the right. Section co-labeled with DAPI (blue). (I) Dual immunohistochemistry with the channelrhodopsin axon terminals as above (green) and a parvalbumin antibody demonstrates targeting of parvalbumin-expressing NAc interneurons by the projection. Scale bars = 20μm in G;H and 40μm in I.

Given the surprising differences in the functional role of the CRH⁺/GABA BLA-NAc pathway in reward behavior, we first focused our analysis on the NAc, a well-established reward circuit hub. In this region, both sex and rearing influenced the density of CRH⁺ BLA axonal innervation (Fig. 4C–E). Overall, ELA reduced labeling of axons in the NAc of male mice, with no effect on females (Fig. 4C–D). Notably, when comparing innervation of the NAc along the anterior-posterior axis, we found that this reduction in ELA males was most notable in the anterior NAc, with negligible changes in the posterior NAc (Fig. 4E). These changes were not a result of differences in the numbers of CRH⁺ neurons in the BLA of the four groups studied (Fig. S2) and are in line with the sex-dependent functional role of the CRH⁺ BLA-NAc pathway in motivated behavior.

The significant reduction in density of CRH⁺ axon terminals emanating from the BLA in ELA males, in who the projection exerts a strong anhedonic effect was surprising. Therefore, we tested the possibility that inputs of the projection target both medium spiny neurons (MSNs) and inhibitory interneurons. A selective loss in ELA males, of innervation by the GABAergic/CRH projection of interneurons could augmented their effect on NAc function. Indeed, concurrent labeling for the projection (Channe-lrhodopsin), and dopamine receptor 1 (D1) or parvalbumin (PV) demonstrated that both MSNs and PV-expressing interneurons are targeted by the CRH/GABA BLA-NAc projection (Fig.4 F–K).

Sex and early life adversity influence brain-wide projection patterns of CRH⁺ BLA cells.

Our results above, showing a rearing effect in males but not in females, on axonal innervation of CRH⁺ BLA neurons within the NAc correlate with, and potentially contribute to, the reduction in reward behaviors in males who experience ELA compared to typically reared males. As no differences in axon labeling was observed in the NAc of female mice reared in ELA compared with controls, we speculated whether ELA might induce anatomical changes in CRH⁺ BLA neuronal innervation in other downstream (target) regions. To investigate this, we expanded our analysis to compare brain-wide projection patterns of CRH⁺ neurons in the BLA in typically reared and ELA mice of both sexes.

The axonal projection patterns of CRH⁺ BLA neurons were consistent with previous reports of BLA connectivity (Fig. 5A–B). For example, we found that these neurons project to the temporal association area, the hippocampal formation, including the entorhinal cortex, CA1, and CA3 (Yang and Wang, 2017; Wahlstrom et al., 2018; Hintiryan et al., 2021), the nucleus of the solitary tract, the piriform cortex, and the zona incerta (East et al., 2021; Arena et al., 2024). Unlike the overall BLA cell population, CRH⁺ cells also robustly innervated hypothalamic areas, including the paraventricular nucleus, anterior hypothalamic nucleus, ventral mammillary nucleus and the medial preoptic area.

Fig. 5: Visualization and quantification of brain-wide projection patterns of CRH+ BLA cells.

(A) 80 μm coronal optical sections showing CRH⁺ BLA axons registered to the standardized brain atlas. Each image represents one example brain from each group. Innervation of CRH⁺ BLA cells in CTL males, ELA males, CTL females and ELA females are shown in blue, green, pink and orange respectively. (B) Heatmap on the left shows the relative labeling density averaged across conditions (normalized to region volume and total labeling) of 120 regions defined by the Allen Brain Atlas. Right heatmap, p values from Aligned Rank Transform of axon innervation density comparing the effect of sex, rearing and their interaction for each individual brain region. (C–F) Sex and rearing effect on labeling density of axons in specific regions: (C) zone incerta, (D) superior colliculus, (E) auditory cortex, (F) retrosplenial cortex. 2-way ANOVA (C: F_rearing(1,11)=4.866, p=0.05; F_sex(1,11)=9.672, p<0.01; F_{rearing × sex}(1,11)=6.324, p=0.03; D: F_rearing(1,11)=12.90, p<0.01; F_sex(1,11)=9.853, p<0.01; F_{rearing × sex}(1,11)=10.27, p<0.01; E: F_rearing(1,11)=11.62, p=0<0.01; F_sex(1,11)=5.515, p=0.04; F_{rearing × sex}(1,11)=0.099, p=0.76; F: F_rearing (1,11)=22.76, p<0.01; F_sex(1,11)=21.97, p<0.01; F_{rearing × sex}(1,11)=3.022, p=0.11; n, CTL male=4 mice/2 litters, ELA male=4 mice/1 litter; CTL female=4 mice/1 litter, ELA female=3 mice/1 litter). See Table S2 for full statistical analyses and tests. CTL = control. ELA = early life adversity.

We next performed quantitative comparisons of axonal innervation density in each individual brain region and identified several regions in which CRH⁺ BLA axonal density was selectively altered in females that experienced ELA. Salient examples include the zona incerta, superior colliculus, auditory cortex and retrosplenial cortex. All of these regions are involved in various forms of sensory processing and memory formation (Letzkus et al., 2011; Wang et al., 2020b, 2020c; Cheng et al., 2024) (Wang et al., 2020b; Ye et al., 2023). Within the zona incerta and superior colliculus, ELA females displayed greater axonal innervation than control females and both male groups (Fig. 5C–D). In the auditory cortex males who experienced ELA displayed greater labeling compared to control males and both female groups (Fig. 5D). Lastly, within the restrosplenial cortex both and effect of ELA, increasing axon density, is apparent in both males and females. However, only in males are these increases significant, compared to controls and both female groups (Fig. 5E).

Discussion

The principal findings of the current studies are: (1) Unlike in males, chemogenetic manipulations of the CRH⁺/GABA BLA-NAc projection in females do not influence reward behaviors. (2) Optical stimulation and immunohistochemistry indicate that the basic cell types and synaptic functions of the CRH⁺/GABA BLA-NAc projection are similar in males and females. (3) The structural organization of the projection is sex-specific and is further modulated by early-life rearing conditions. (4) Beyond projections to the NAc, the innervation patterns of CRH⁺ BLA neurons suggest global sex-dependent differences in their brain-wide organization, which might account for both fundamental differences in their operations as well as sex-dependent differences in the outcomes of ELA.

In male mice, we previously discovered that the CRH⁺/GABA BLA-NAc projection mediates the anhedonic-like effects of ELA on reward behaviors (Birnie, 2023). Here we recapitulated these results, finding that chemogenetic excitation of the projection in typically reared adult male mice reduced palatable food consumption while in contrast, a similar manipulation in females did not affect consumption in either control or ELA groups. Further, while inhibiting the projection in ELA males rescued reward deficits (Birnie, 2023), this manipulation in control or ELA-adult females failed to influence palatable food consumption. These surprising results indicate profound, sex-specific differences in the functional roles of the projection in male and female mice.

Sex differences emerged in the specific consequences of ELA: Reward behaviors include motivation (‘wanting’) and consummatory (‘liking’) aspects (Berridge and Robinson, 2016). Testing a second reward seeking behavior, mouse responses to the scent of a mouse of the opposite sex, we found that ELA reduced behaviors towards the scent of an estrus female in male mice and enhanced these behaviors in ELA females presented with the scent of a male mouse. ELA reduced both consummatory behaviors (sniff times) and motivation (approaches to the scent-bearing Q-tip) in males. In contrast, in females, motivation for the male scent was augmented, with little change in sniff durations (Fig.1). Because motivational and consummatory aspects of reward behaviors have distinct anatomical substrates, these findings suggest a need for investigating the nuanced connectivity and potential functional or structural differences in the CRH⁺/GABA BLA-NAc projection in males and females.

In females, we found no sex or rearing effects on the basic functional properties of NAc-projecting CRH⁺/GABA cells. Using optogenetics paired with electrophysiology, we found that activating CRH⁺ cells in the BLA triggered inhibitory currents in the medial shell of the NAc across all groups, regardless of sex or rearing conditions (Fig.3). All CRH⁺ BLA cells were GABAergic, indicating that the fundamental cell-type and function of the projection are common across sexes and rearing. We then reasoned that the sex-dependent differences in the behavioral functions of the projection result from neuroanatomical differences in the organization and target innervation of CRH⁺ BLA cells.

We first determined the qualitative and quantitative aspects of the CRH⁺/GABA BLA-NAc projection within the NAc. As blocking terminals in the NAc restored typical reward responses to a palatable food cue in ELA males (Birnie, 2023), we hypothesized that in ELA mice, CRH⁺/GABA BLA neurons may have increased NAc axonal innervation, thus leading to increased synaptic activity that influences NAc operations. To our surprise, we observed a decrease in axon labeling in NAc of ELA male mice. The NAc contains both medium spiny neurons (MSN), the main output population, and GABAergic interneurons that regulate MSN activity. We found that CRH⁺ BLA-NAc axons innervate both MSNs and interneurons. This complexity allows reconciling the observed data in which activating this pathway in control males mimics the effects of ELA despite reduced axon density in the latter. Target-specific circuit reorganization, e.g. preferential loss of inputs onto inhibitory interneurons, may be at play. In that scenario, stimulating the GABAergic pathway in control males could inhibit MSNs, enhancing NAc inhibition thereby suppressing reward behaviors. By contrast, ELA males already exhibit reduced NAc activity, through reduced disinhibition (or heightened activity of the GABAergic pathway), which could occlude the effects of stimulating this pathway. In females we found no differences in axonal innervation in NAc. As for males, we excluded the possibility that any differences in signal density might result from differences in the number of BLA CRH⁺ cells (Fig.S2). We also considered potential influence of estrous cycle stage and established that cycle-phase had no influence on the data and there was no litter effect (Fig.S1). We also excluded misplacement of virus or cannulae as a potential explanation of this observation (Fig.S2).

Because ELA impacted reward behaviors in females and because NAc innervation by the BLA-origin CRH⁺/GABA cells did not differ in ELA and control females, we speculated that other projections of the same—apparently ELA-sensitive—cells might contribute to the observed behaviors effects of ELA. Indeed, our brain-wide analysis uncovers other projection targets of CRH⁺/GABA BLA cells that exhibit female-specific changes in ELA mice. One of these regions, the zona incerta, exhibited female-specific increases in CRH⁺ BLA axons following ELA. The zona incerta (ZI), an extension of the reticular formation of the thalamus, is a largely inhibitory nucleus containing genetically heterogeneous cell populations (Arena, 2024). Recent studies have found the ZI to be involved in integrating multisensory inputs, controlling motor behavior and importantly, driving motivational and appetitive behaviors (Wang, 2020b; Ye, 2023). Altered ZI activity contributes to anxiety-related and coping behaviors following chronic stress (Zhou, 2021; Ge, 2024), but the effects of ELA on ZI functions in the context of reward are poorly understood. This analysis paves the way for future studies elucidating the role of the CRH⁺ BLA-ZI pathway in reward behaviors and the effects of ELA.

Beyond the ZI, our brain-wide analysis revealed extensive target and sex specificity in the effects of ELA on CRH⁺ BLA innervation patterns. Notably, the retrosplenial cortex exhibits increased axon density in the ELA group, with significantly greater increases in males compared to females. The retrosplenial cortex is critically involved in spatial and contextual learning and memory (Vann, 2009; Cheng, 2024), and one study has reported male-specific impairments in the novel object placement task, a paradigm that depends on spatial learning, following ELA (Bath, 2017).

We also find male-specific increases in axon density within the auditory cortex, which supports the perception, interpretation, and memory of auditory cues (Rothschild, 2010; Letzkus, 2011), aligns with findings of male-specific deficits in auditory learning in adult rodents exposed to ELA (Hardy, 2023; Mazi, 2025). In contrast, a different sensory processing region, the superior colliculus, shows female-specific increases in CRH⁺ BLA innervation in ELA mice compared to controls. The superior colliculus integrates multisensory information, enabling the detection and prioritization of environmental stimuli (Stein, 1989; Perrault, 2005). Multisensory processing can be influenced by experience (Wang, 2020c) and altered sensitivity to external stimuli is a hallmark of several psychiatric conditions (Liss, 2005; Harrold, 2024). Whether these changes in sensory processing are sex-dependent and the neural mechanisms underlying these changes remain unclear but present a compelling direction for future investigation. Together these findings suggest that ELA, and perhaps the aspect of fragmented maternal care observed in our LBN ELA model, may alter sensory processing circuitry in a sex-specific manner. This may lead to enduring differences in how males and females process environmental cues, governing their reward behaviors.

Our observations of region-specific increases in axon labeling are consistent with previous studies that observed elevated BLA axon density in the prefrontal cortex of adolescent rats that experienced ELA (Honeycutt, 2020). We also considered the possibility that regions with low axonal density might be more likely to show more significant effects. To that end, we generated density plots and established that while numerous regions with low density indeed show significant effects, more than a third of the significantly altered regions have high axon densities (Fig. S3).

Several mechanisms may contribute to the increased axon labeling we observed. The increased density may result from augmented axon sprouting or a failure of developmental pruning mechanisms. The latter mechanism is supported by our prior work on the failure of synaptic pruning onto CRH⁺ cells in the hypothalamic paraventricular nucleus (Bolton, 2022). The deficient synaptic pruning was a result of ELA-induced microglial dysfunction. By contrast, in the hippocampus, ELA has led to selective deficits of neuronal arborization in specific hippocampal layers (Brunson, 2005; Ivy, 2010) thus, ELA-induced dysregulation of structural brain circuit maturation is a plausible common theme for the altered innervation observed here, and for the behavioral consequences (Birnie and Baram, 2022, 2025).

In sum, our study reveals that ELA causes sex-specific changes in adult reward behaviors which, in males are mediated, at least in part, by a GABA/CRH⁺ BLA-NAc projection. In contrast, projection function or innervation does not explain female-specific ELA-induced behavioral changes. Importantly, we find sex-and rearing related plasticity of brain-wide projections of CRH⁺ BLA neurons, suggesting that ELA can exert potent, lasting effects on brain organization of stress-sensitive cells and projections (Birnie & Baram, 2025). Our brain-wide datasets and the effects of sex and rearing will enable future investigations of how sex-specific BLA circuit alterations contribute to behavioral outcomes following ELA.

Significance Statement.

Early life adversity is a major public health issue linked to long-term mental health risks including depression. In adult mice with this history, reward behaviors are disrupted in a sex-specific manner. We identified a novel GABAergic and corticotropin releasing hormone-expressing projection from basolateral amygdala to nucleus accumbens that mediates poor reward motivation in early-adversity male mice, but has no effect in females. Here we studied why, and found important sex and experience-dependent change in the brain’s fine wiring. Our discoveries help explain sex-dependent consequences of early life adversity in people, and may inform targeted therapeutic strategies.