Urocortin-3 neurons in the perifornical area are critical mediators of chronic stress on female infant-directed behavior

Abstract

Infant avoidance and aggression are promoted by activation of the Urocortin-3 expressing neurons of the perifornical area of hypothalamus (PeFA^Ucn3) in male and female mice. PeFA^Ucn3 neurons have been implicated in stress, and stress is known to reduce maternal behavior. We asked how chronic restraint stress (CRS) affects infant-directed behavior in virgin and lactating females and what role PeFA^Ucn3 neurons play in this process. Here we show that infant-directed behavior increases activity in the PeFA^Ucn3 neurons in virgin and lactating females. Chemogenetic inhibition of PeFA^Ucn3 neurons facilitates pup retrieval in virgin females. CRS reduces pup retrieval in virgin females and increases activity of PeFA^Ucn3 neurons, while CRS does not affect maternal behavior in lactating females. Inhibition of PeFA^Ucn3 neurons blocks stress-induced deficits in pup-directed behavior in virgin females. Together, these data illustrate the critical role for PeFA^Ucn3 neuronal activity in mediating the impact of chronic stress on female infant-directed behavior.

Introduction

Decades of research have focused on the neurobiology of maternal behavior, revealing common mechanisms and pathways involved in infant caregiving across a variety of species^{1, 2}. Studies converged on the critical role of the medial preoptic area (MPOA) in orchestrating the behavioral responses of mothers to their young in frogs, fish, birds, and rodents^3-6. Recently, it has been appreciated that these mechanisms may also be involved in paternal behaviors, suggesting existence of a core circuitry across sexes that promotes caregiving^7-9. Moreover, it has been well-documented that neural plasticity underlies facilitation of infant caregiving, including alloparental care, particularly in females².

In the absence of caregiving, it is possible to observe neglect or aggression toward infants by adults. Studies have identified circuit nodes in the brain, including the medial and posterior amygdala, the bed nucleus of stria terminalis, and the perifornical area of hypothalamus (PeFA), that modulate expression of infant-directed neglect and aggression^10-13. We wondered if anti-parental circuitry may be active in neglectful animals including virgin or stressed females.

Clinical research shows that stress is a critical risk factor for postpartum mental illnesses including postpartum depression or anxiety which affect up to 25% of women and 10% of men annually in the U.S.^{14, 15}. However, few preclinical studies examine the neurobiology underlying reduced parent-infant bonding in animals^16-18. Pre- or postnatal maternal stress has been shown to have a profound impact on the behavioral, physiological, and neurobiological development of offspring^19-25. During lactation, females are hyporesponsive to stress due to hormonal changes that impact the Hypothalamic-Pituitary-Adrenal (HPA) axis regulation^{26, 27}. HPA axis hyporesponsivity is thought to protect caregiving and stress-coping in mothers^{28, 29}. However, chronic stress has been documented to have a long-lasting impact on HPA axis regulation, leading to reduced parenting, though the underlying neurobiology remains poorly understood^30-32.

Urocortin-3 (Ucn3) is a member of the corticotropin releasing factor (CRF) family of peptides and has the highest endogenous binding affinity for CRF receptor 2 (CRFR2). Overexpression of Ucn3 in the brain leads to increased anxiety- and depression-related behaviors and results in a blunted HPA response to stress³³. Social discrimination is altered in a sex-specific manner in total Ucn3 knockout mice³⁴. Furthermore, Ucn3 expression has site-specific behavioral functions such as aiding in recovery of acoustic trauma in the auditory brainstem³⁵, increasing preference for novel conspecifics in the medial amygdala³⁶, and enhancing anxiety-like behaviors in the PeFA³⁷. Previous studies of PeFA Ucn3 (PeFA^Ucn3) neurons suggest that they are sensitive to stress and modulate vasopressin expression levels in the paraventricular hypothalamus (PVH)^{38, 39}. Taken together, these studies suggest that PeFA^Ucn3 cells mediate stress-induced behavioral changes.

We recently discovered that PeFA^Ucn3 neurons govern infant neglect and infanticide in both sexes¹⁰. Thus, we hypothesized that chronic stress negatively affects infant-directed behavior in females and this disruption is dependent on activation of PeFA^Ucn3 neurons. We set out to determine if PeFA^Ucn3 neurons were more active in stressed females, and if we could recover parenting deficits by blocking activity in PeFA^Ucn3 cells. We find that enhanced parenting is accompanied by decreased activity in PeFA^Ucn3 neurons and inhibiting these cells enhances alloparenting in virgin females. Chronic stress reduces alloparenting and this stress-induced behavioral effect is blocked by inhibition of PeFA^Ucn3 cells. In stressed lactating females, parenting is preserved and PeFA^Ucn3 neurons are less active. Together, these data reveal a critical role for PeFA^Ucn3 neurons in the expression of alloparenting in female mice under normal and pathological conditions.

Materials and Methods

Animals

Mice were maintained on a 12h:12h dark light cycle (10:30am-10:30 pm dark phase) with ad libitum food and water access. All experiments were performed following NIH and ARRIVE ethical guidelines and approved by Albert Einstein College of Medicine Institutional Animal Care and Use Committee (protocol 20180110; 20180111; 00001386). C57BL/6J virgin and pregnant female (E14) mice were ordered from Jackson Laboratories (Bar Harbor, ME) aged 8-10 weeks. Female Ucn3::Cre BAC transgenic line (STOCK Tg(Ucn3-cre) KF43Gsat/Mmcd 032078-UCD; obtained from Catherine Dulac, Harvard University) were also used, aged 2-5 months. Experimental animals were randomly allocated to experimental groups. See Supplement for details.

Corticosterone Measurement

Trunk blood was collected during sacrifice and serum was isolated by centrifugation. A high-sensitivity corticosterone enzyme immunoassay (EIA) was conducted per manufacturer’s instructions (Immunodiagnostic Systems Ltd, Fountain Hills, AZ, USA) as previously described⁴⁰. See Supplement for details.

Chronic restraint stress (CRS) model

Virgin females were stressed for 20 days and lactating females for 17 days (before pup weaning). Lactating females were restrained starting ~postpartum day 2 (PP2). All females (with their litters) were brought to a test room under dim red light during their dark cycle, weighed, and either placed back into their home cage (and reunited with their litters) or placed into a 50 mL conical tube containing holes for ventilation for 1 hour. Humidity and temperature in the room was recorded each day. On the last day of stress, females remained in the room for 1-2 hours before exposure to a foreign-born pup (see Parental Behavior). For the stressed virgin female group, mice were excluded from the stressed group (n=4) based on open field behavior. For the groups that received viral injections, females with inadequate recombination were excluded (n=2, Figure 2; n=0 Figure 6). See Supplement for details.

Intruder stress model

Lactating females were stressed starting ~PP2. All females were brought to a test room under dim red light during their dark cycle, weighed, and control females were placed back into their home cage while stressed females had an adult male intruder introduced into their home cage for 10 minutes as described previously in rat dams^{32, 41}. See Supplement for details.

Chronic variable stress (CVS) model

Lactating females were stressed starting ~PP2. All mice were brought to a test room under dim red light during their dark cycle, weighed, and either placed back into their home cage or underwent stress. CVS was adapted from a previous study, for details see Supplement⁴⁰. On the last day of stress, females remained in the test room for 1-2 hours before exposure to a foreign-born pup (see Parental Behavior).

Behavior assays

Mice were individually housed for at least 1 week before testing. Experiments were conducted during the dark phase under dim red light. Tests were recorded by Fly Capture cameras (Point Grey, Richmond, BC, Canada) and behaviors were scored by an observer blind to experimental conditions using Observer XT13 Software or Ethovision XT 13 (Noldus Information Technology, Leesburg, VA, USA). Animals underwent 1 behavioral test per session with at least 24 hours between sessions.

Parental behavior

Parental behavior tests were conducted in the mouse’s home cage as previously described⁹. Mice were habituated to the testing environment for 10 minutes. One to two C57BL6/J pups 1-4 days old were presented in the opposite corner to the nest and behavior was recorded for 10-15 minutes (see Supplement for details).

Open field

Mice were assessed for activity in a 45cm x 45 cm open field at 40 lux for 5 min as previously described⁴⁰. Center was considered 15 cm x 15 cm and borders were 5 cm around the perimeter of the box. Time and frequency in center and border, distance, and velocity were calculated using Ethovision XT13.

Fluorescence in situ hybridization (FISH)

FISH was performed as recommended by ACD Bio (Newark, CA, USA) using V1 RNAscope reagents with Fos (Cat No. 316921), Ucn3 (Cat No. 464861), and Crh (Cat No. 316091) probes (see Supplement for details).

Single-cell RNAseq (scRNAseq)

Detailed methods including analysis were previously published⁴². Raw reads and output of the CellRanger pipeline are available on GEO (GSE113576). See Supplement for details.

Multiplexed Error-robust FISH (MERFISH)

A panel for MERFISH comprising 155 genes was chosen to discriminate major cell classes and mapscRNAseq clusters in tissue sections. Probes were generated as described previously⁴³. MERFISH staining and analysis was performed as described previously⁴². See Supplement for details.

Immunostaining

To visualize c-Fos protein and DREADD, perfused tissue was sliced on a freezing microtome at 30 μm, and every third PeFA section was stained. Primary antibody chicken anti-mCherry (Millipore AB3566481) and rabbit anti c-Fos (Cell Signaling 2250S) diluted at 1:1000 and secondary anti-chicken-A594 (Sigma CF594) and anti-rabbit-A647 (Life Technologies A31573) diluted at 1:200 and 1:1000 respectively were used. For details, see Supplement.

Chemogenetics

Ucn3::Cre virgin female mice (or Cre− littermates as controls) 8-20 weeks old were used. We stereotaxically injected ~225 nL of conditional inhibitory designer receptor exclusively activated by designer drug (DREADD; AAV1-hSyn-DIO-hM4D(Gi)) virus bilaterally into the PeFA (AP −0.6mm, ML±0.3mm, DV −4.2mm).

Cre+ and Cre− females were administered intraperitoneally (i.p.) either 1x PBS (vehicle) or 0.3mg/kg clozapine-n-oxide (CNO) and habituated to the testing environment for 2-3 hours. Females were presented 1-2 C57BL6/J pups in their home-cage and parental behaviors were recorded for 10-15 minutes.

Data analysis and Statistics

For colocalization, we used Fisher’s exact test and a two-tailed Welch’s t test. Pup retrieval percentages are analyzed by Kolmogorov-Smirnov (2 groups) or Friedman test (3 groups). For experiments with one manipulation (i.e. stress or neuronal inactivation), we used two-tailed t-test for normally distributed data and Mann-Whitney test for non-normally distributed data. One-way and two-way repeated measures ANOVA tests were used followed by post-hoc correction. Data were analyzed by Graphpad Prism 9.0 or Matlab scripts by an investigator blinded to condition. Sample sizes were based on previous experiments ^{8-10, 40}. For complete details, see Supplement. P-values reported as follows: * P<0.05, ** P<0.01, *** P<0.001, **** P<0.0001. All data are expressed as mean±SEM.

Image analysis

Images were exported from ZenBlue software and cells were manually counted by an observer blinded to condition using FIJI Cell Counter. Graphpad Prism 9 was used for statistics.

Fiber Photometry

Ucn3-Cre animals were injected with 150 nL of AAV1-syn-jGCaMP7f-WPRE and 225 nL of AAV1-hSyn-DIO-hM4D(Gi)-mCherry and implanted with a 200μm fiber optic cannula into PeFA (ML: 0.3mm; AP −0.6mm; DV 4.2mm). Animals recovered at least 3 weeks before behavioral experiments. Animals were brought up to the test room in dim red light and injected with either vehicle (session 1) or 0.3mg/kg CNO (session 2) i.p. Two hours later, mice were tethered to the fiber optic cable (Doric) and habituated 10 minutes before recording. Using a multi-channel photometry system (Neurophotometrics LTD), a 470nm LED and 415nm LED (isosbestic control) alternately illuminated at 60μW by a 20X objective and fluorescence emission was collected using a CMOS camera at 40 Hz. After 1-2 minutes of recording, animals underwent 6 tail suspensions (~5 seconds). Data were acquired by Bonsai. For analysis details, see Supplement.

Code availability

Custom Matlab code for analysis of photometry data is available at github.com/babdelme/Autrylab_photometry. MERFISH image analysis scripts are available at github.com/ZhuangLab/MERFISH-analysis.

Results

PeFA^Ucn3 neuronal activation after pup interaction depends on physiological context

To identify activation levels of rostral PeFA^Ucn3 neurons, we exposed virgin females and lactating females (postpartum day 2) to either a foreign pup (P0-P4) or ~25 mg of fresh bedding (control) in their homecage. Animals were sacrificed 30 minutes later (Figure 1A). To control for the number of pups and foreign pup discrimination^{44, 45}, we utilized two groups of lactating females that either had their litter removed 10 minutes before pup introduction or kept their litter (Figure 1A). We observed an increased proportion of PeFA^Ucn3 neurons colocalized with Fos in virgin females exposed to a pup compared to controls (Figure 1B, C; Supplemental Figure 1A, D; see Supplemental Figure 1E for FISH validation). However, in lactating females, the proportion of PeFA^Ucn3 neurons colocalized with Fos with pup exposure decreases with litter removal (Figure 1B, D; Supplemental Figure 1B, D) but increases if the litter is present (Figure 1B, E; Supplemental Figure 1C, D). These results suggest that PeFA^Ucn3 cells respond to pups similarly in virgin females and lactating females with their litter, while we observe an opposite impact on PeFA^Ucn3 cell activity in lactating females with litter removal.

Figure 1. PeFA Urocortin-3 neuronal activation levels in response to foreign pups depends on physiological context.

(A) Schematic of behavioral paradigm. C57 virgin females were exposed to a newborn pup and sacrificed 30 minutes after pup exposure or addition of fresh bedding into home cage (control). Lactating females either had a litter removed or litter remained intact and exposed to a foreign-born pup or fresh bedding. (B) Rostral perifornical area cells containing Urocortin-3 and Fos RNA were counted for colocalization in each group (scale bar 100 μm; arrows for colocalization). (C) Quantification of percentage of Ucn3+ cells colocalized with Fos across groups. (Left) Welch’s t test on percentage of PeFA^Ucn3 cells co-expressing Fos over total Ucn3 reveals significant enhancement in pup exposed virgin females (Welch’s t test; Control n=5 mice; pup exposed n=8 mice; p=0.0024, df=8.041; represented in dots) and (right) fisher exact test on proportion of Ucn3+/Fos+ cells within group compared to Ucn3+/Fos− reveals significant increase (Control n=541 cells; pup exposed n=655 cells; p<0.0001; represented in bars) (D) Lactating females with litters removed had a significant reduction in the proportion of Ucn3+/Fos+ when exposed to a foreign pup (Fisher exact test: Control n=731 cells; pup exposed n=427 cells; p=0.0001; represented in bars) (E) Lactating females with litters undisturbed both has a significant increase in the percentage of Ucn3+/Fos+ neurons (Welch’s t test: Control n=6 mice; Pup exposed n=7 mice; p=0.0341, df=6.657 represented in dots) and a significant upregulation in the proportion of Ucn3+/Fos+ when exposed to a foreign pup (Fisher exact test: Control n=537 cells; pup exposed n=638 cells; p<0.0001; represented in bars). (F-I) Single cell transcriptomic analysis of rostral Ucn3 expressing cells. (F) Cartoon depicting the rostro-caudal extent of the dissection used to conduct single cell transcriptional profiling of the rostral hypothalamus. (G) Experimental scheme for single cell RNA sequencing and MERFISH based spatial transcriptomic profiling of the rostral hypothalamus. (H) t-SNE plot of single cell RNA sequencing clustering of excitatory neurons indicating Z-scored expression of Ucn3. Single cluster expressing Ucn3 expression is marked with a dotted gray circle. (I) MERFISH image of the cluster expressing Ucn3 showing localization to the PeFa (upper panel) and violin plot showing expression profile of important genes within the cluster (lower panel). *All figures generated from previously published data in Moffitt, Bambah-Mukku et al, 2018.

To determine if this activity difference is explained by multiple subsets of PeFA^Ucn3 neurons, we analyzed previously collected single-cell sequencing and spatial sequencing (MERFISH) data⁴². Cells were collected for both workstreams from the MPOA and rostral PeFA (Fig. 1F, G). After clustering the single cell dataset, Ucn3 gene expression is highest in one subset of neurons. This single subset of Ucn3 cells is localized in the rostral PeFA, the same region we examined for behavior (Fig. 1I). This unique population of cells is characterized by expression of thyrotropin-releasing hormone (Trh), vesicular glutamate transporter 2 (vglut2; or Slc17a6), and several other markers, consistent with previous literature defining this rostral, excitatory population of PeFA^Ucn3 neurons^{10, 46, 47}.

Chemogenetic inhibition of PeFA^Ucn3 neurons enhances alloparenting

Because virgin females are not as parental as lactating females^48-51 and PeFA^Ucn3 neurons are activated by infanticide in females¹⁰, we wanted to test if suppression of PeFA^Ucn3 neuronal activity enhances alloparenting in virgin females. To accomplish this, we used a conditional viral strategy to express inhibitory DREADD in Ucn3::Cre+ and Cre− animals in the PeFA of virgin females naïve to pups. Two-three weeks after viral injection, both groups of animals were administered CNO (0.3 mg/kg intraperitoneally) and 2-3 hours later exposed to two pups for fifteen minutes in their homecages (Figure 2A; Supplemental Figure 2-1). We confirmed recombination for inclusion in subsequent analyses (Supplemental Figure 2A). Cre+ females retrieved more pups in a shorter amount of time relative to Cre− controls (Figure 2B-D), with no difference in latency to retrieve the 2^nd pup (Figure 2E). Furthermore, Cre+ animals spent more time in the nest with pups and started nest-building earlier compared to controls (Figure 2G, I). While other measures of caregiving were not affected (Figure 2 F, H, J-K; Supplemental Figure 2B-E), suppression of PeFA^Ucn3 neurons improved certain aspects of alloparenting (Figure 2L-M), particularly pup retrieval and time spent in the nest.

Figure 2. Inhibition of PeFA^Ucn3 neurons enhances alloparental behavior in virgin female mice.

(A) Schematic of viral injection strategy and behavior timeline (n=10 Cre− females; n=8 Cre+ females). (B) Percentage of pups retrieved by Cre+ females is significantly increased compared to Cre− females (Kolmogorov-Smirnov test p=0.0048). (C) Latency to retrieve each of two pups presented during the trial is significantly faster in Cre+ females compared to Cre− females (Two-tailed Mann-Whitney test; p=0.0033), as well as (D) latency to retrieve the first pup (Two-tailed Mann-Whiney test; p=0.0259), but there was no difference between groups in (E) latency to retrieve the second pup. (F) Time spent investigating pups trended lower in Cre+ animals (Unpaired t test; p=0.0854, df=16). (G) Time spent in nest with pups significantly increased in Cre+ animals (Unpaired t test; p=0.0289, df=16). (H) Time spent crouching did not yield any significant changes. (I-J) Latency to nest build was significantly faster in Cre+ animals (Two-tailed Mann-Whiney test; p=0.0152) but time spent nest building was not significantly different between groups. (K) Cumulative time spent parenting was unchanged between groups. (L-M) Representative behavior trace of a Cre− animal and a Cre+ animal during pup assay (time 0 is when pup was added to home cage).

Chronic restraint stress differentially impacts virgin and lactating female parenting

To understand the impact of chronic stress on alloparenting in virgin females, we employed a chronic restraint stress (CRS) paradigm (Figure 3A). Stressed females weighed significantly less than controls (Figure 3B). On day 19, females were tested for exploratory behavior in the open field task (Figure 3C-F). Stressed females spent less time in the center of the field compared to controls (Figure 3C). On the last day of stress, females were exposed to a newborn pup (P0-4) for 15 minutes and their alloparental behavior was recorded and analyzed (Figure 3G-P). Only 2 out of 9, or 22%, of stressed females retrieved pups compared to controls (5 of 9, or 55%) (Figure 3G). Virgin stressed females did not show any significant changes in other quantitative measures of alloparenting (Figure 3H-N; Supplemental Figure 3). However, 2/9 of the stressed females aggressively handled the pups (data not shown) and some stressed virgin females showed less caregiving behaviors (Figure 3O, P). Altogether, we find that CRS significantly reduces pup retrieval in virgin females.

Figure 3. Chronic restraint stress dampens alloparental behavior in virgin female mice.

(A) Schematic of timeline using restraint stress paradigm and behavioral testing in virgin females (n=9 control; n=9 stress). (B) Weights taken from each group during the period of chronic restraint stress. Stressed females have a significant difference in weight compared to controls (Two-way repeated measures ANOVA; main effect of interaction of stress x time F_(10,220)=22.92 p<0.0001; main effect of time F_{(10, 220)} = 20.35; Bonferroni’s multiple comparisons test) (C) Time spent in center of open field was significantly reduced for stressed females compared to controls (unpaired t test; p=0.0289, df=16) (D-F) but no other parameter in open field was changed. (G) Chronic restraint stress significantly decreased cumulative pup retrieval in females (Kolmogorov-Smirnov test; p=0.0059) (H-N) Chronic restraint stress did not significantly change other parenting measures such as retrieval latency, pup investigation, time in nest, crouching, nest building, and total time spent parenting. (O-P) Representative behavior trace for a control female and stressed female during pup assay. (*p=0.05; **p=0.01; ***p=0.001; ****p=0.0001).

Next, to understand the impact of chronic stress on maternal behavior, we utilized CRS in lactating females from PP2-18 (Figure 4A). Like stressed virgin females, stressed lactating females weighed significantly less than controls (Figure 4B), indicating that CRS induced physiological changes. On day 16 of CRS, females were tested in the open field task (Figure 4C-F). Surprisingly, stressed females spent more time in the center and less time in the borders (Figure 4C, D). Stressed females also showed increased velocity and distance traveled relative to controls (Figure 4E, F). On the last day of stress, females had their litters removed and 10 minutes later we tested infant-directed behavior (Figure 4G-P). We observed no difference in number of pups retrieved or retrieval latency between groups (Figure 4G, H). Stressed lactating females showed similar levels of parenting toward pups as controls (Figure 4I-P; Supplemental Figure 4-1). We also implemented a chronic social stress paradigm that was reported to impact parenting in lactating rats³², and we did not observe any changes in weight or parental measures (Supplemental Figure 4-2).

Figure 4. Chronic restraint stress does not induce changes in parental behavior in lactating females.

(A) Schematic of timeline using restraint paradigm and behavioral testing in lactating females (Control n=12; Stressed n=13). (B) Weights taken from each group during the period of chronic restraint stress from postpartum day 2 to 18. Stressed females have a significant difference in weight compared to controls (Two-way repeated measures ANOVA; main effect of interaction of stress x time F_(17,368)=8.286 p<0.0001; main effect of time F_{(7.579,174.3)}=75.28 p<0.0001; Sidak’s multiple comparisons) (C) Time spent in center of open field was significantly increased for stressed females compared to controls (unpaired t test; p=0.0459, df=8) and (D) time spent in borders decreased (unpaired t test; p=0.0459, df=8) accompanied by (E) increased distance traveled (unpaired t test; p=0.0089, df=8) and (F) velocity (unpaired t test; p=0.0089, df=8) (G-N) Chronic restraint stress did not significantly change any parenting measures as retrieval, pup investigation, time in nest, crouching, and nest building compared to females that did not receive stress. (O-P) Representative behavior trace for a control female and stressed female during pup assay.

To control for habituation to homotypic stress, we employed a chronic variable stress (CVS) paradigm for 10 days from PP2-12 (Supplemental Figure 4-3). Lactating females subjected to CVS weighed significantly less than controls (Supplemental Figure 4-3B), but we did not observe any differences in the open field test (Supplemental Figure 4C-F). In parenting, most parameters were unchanged between groups (Supplemental Figure 4G-N; Q). However, latency to nest-build was reduced (Supplemental 4-3O) and time spent nest-building was increased (Supplemental 4-3P) in lactating females with CVS.

Chronic restraint stress differentially impacts molecular substrates in virgin and lactating females

Next, we investigated molecular and physiological impacts of CRS in virgin or lactating females. We observed a significant increase in the proportion of PeFA^Ucn3 neurons colocalized with Fos in stressed virgin females (Figure 5A, B; Supplemental Figure 5-1A, C). In control virgin females, the percentage of PeFA^Ucn3+/Fos+ colocalization was negatively correlated with time spent parenting, indicating that activity of PeFA^Ucn3 neurons may reduce alloparenting (Figure 5E). Because activation of CRF cells in the PVH (PVH^CRF) is postulated to disrupt maternal behavior and is critical for physiological stress responses^{52, 53}, we also quantified PVH^CRF neurons colocalization with Fos (Figure 5A, C; Supplemental Figure 5-1 A, C). We found that the proportion of PVH^CRF neurons colocalized with Fos was significantly reduced in stressed virgin females, and, like PeFA^Ucn3, PVH^CRF neuronal activation is negatively correlated with parental behaviors in control virgin females (Figure 5F). CRS did not affect circulating corticosterone levels in virgin females (Figure 5D).

Figure 5. Chronic restraint leads to contrasting molecular changes between virgin and lactating females in parenting.

(A) Rostral perifornical area cells containing Ucn3, CRF, and Fos RNA were counted for colocalization in each virgin female group (scale bar 100 μm). (B) Colocalization of Ucn3 and Fos in perifornical area reveals increased (right) proportion of PeFA Ucn3+/Fos+ neurons in stressed virgin females compared to control virgin females (Fisher exact test: Control n=490 cells; Stressed n=472 cells; p=0.038; represented in bars). (C) Colocalization of CRF and Fos in paraventricular nucleus of the hypothalamus reveals both decreased percentage of PVN^CRF+ Fos neurons (Welch’s t test: Control n=9 mice; stressed n=9 mice; p=0.0087, df=8.481 represented in dots) and proportion of Ucn3+/Fos+ in stressed virgin females compared to control virgin females (Fisher exact test: Control n=1305 cells; Stressed n=1272 cells; p<0.0001; represented in bars). (D) Serum corticosterone levels were measured using ELISA in virgin females. Chronic restraint stress did not contribute to altered corticosterone in virgin females. (E) Plotting time spent parenting against PeFA^Ucn3 activation levels shows marked negative correlation between parenting and PeFA^Ucn3 neuron activation in controls but this relationship is abolished in stressed virgin females (Control R²=0.51 Stressed R²=0.13; difference in slope F= 10.73 DFn=1, DFd=14; p=0.0055). (F) PVH^CRF activation levels in negatively correlated with time spent parenting in control virgin females but this relationship is indiscernible in stressed virgin females (Control R²=0.68 Stressed R²=0.01). (G) Rostral perifornical area cells containing Ucn3, CRF, and Fos RNA were counted for colocalization in each lactating female group (scale bar 100 μm). (H) Proportion of PeFA Ucn3+/Fos+ are reduced in stressed lactating females (Fisher exact test: Control n=338 cells; Stressed n=360 cells; p=0.0005; represented in bars). (I) Colocalization of CRF and Fos in paraventricular nucleus of the hypothalamus reveals both decreased percentage of PVN^CRF+ Fos neurons (Welch’s t test: Control n=6 mice; stressed n=5 mice; p<0.0001, df=7.807) and proportion of Ucn3+/Fos+ in stressed lactating females (Fisher exact test: Control n=919 cells; Stressed n=2268 cells; p<0.0001). (J) Chronic restraint stress significantly reduced corticosterone in lactating females (unpaired t test; p=0.0198, df=22). (K) Linear regression analysis reveals a trend towards negative correlation between time spent parenting and PeFA^Ucn3 activation levels in control lactating females but not in the stressed group (Control R²=0.27 Stressed R²=0.01). (L) PVH^CRF activation is positively correlated with time spent parenting (Control R²=0.45 Stressed R²=0.39).

In lactating females, CRS led to a decrease in the proportion of PeFA^Ucn3 neurons colocalized with Fos compared to control lactating females, opposite of virgin females (Figure 5G, H; Supplemental Figure 5-1 B, C). Like virgin females, however, PVH^CRF cell activation was significantly decreased in chronically restraint stressed lactating females (Figure 5G, I; Supplemental Figure 5-1 B, C), consistent with previous literature^54-56. We observed a similar negative trend for correlation of PeFA^Ucn3 neuronal activation and parental behaviors in control lactating females that we observed in virgin females, but the trend for PVH^CRF cell activation is positively correlated in lactating females (Figure 5 K, L). We observed that corticosterone levels were significantly decreased in stressed lactating females (Figure 5J), suggesting adaptive habituation to repeated stress. In our intruder stress experiment, we did not observe molecular changes in PeFA^Ucn3 activation or circulating corticosterone levels, consistent with no changes in weight or parenting (Supplemental Figure 5-2). Altogether, CRS induces differential PeFA^Ucn3 activation patterns in response to pups in virgin and lactating females, while CRS reduces PVH^CRF neuronal activation in both groups.

CVS increased the proportion of PeFA^Ucn3 neurons colocalized with Fos in stressed lactating females (Supplemental Figure 5-3B). We did not observe a difference in corticosterone levels between groups (Supplemental Figure 5-3C). When time spent parenting was plotted against the percentage of activated PeFA^Ucn3 cells, no significant correlation was observed in either condition (Supplemental Figure 5-3D). Latency to nest build was negatively correlated with PeFA^Ucn3 activation levels (Supplemental Figure 5-3E) while time spent nest building was positively correlated in stressed females (Supplemental Figure 5-3F). Thus, CVS leads to behavioral changes in indirect parenting measures that correlate with PeFA^Ucn3 cell activity in lactating females.

Inhibition of PeFA^Ucn3 neurons rescues stress-induced deficits in female alloparenting

To confirm that our inhibitory DREADD strategy blocks the activity of PeFA^Ucn3 neurons, we injected the PeFA of Ucn3::Cre+ females with conditional inhibitory DREADD virus (Supplemental Figure 6-1A). After three weeks, animals were injected intraperitoneally either with vehicle or CNO. CNO administration significantly reduced the proportion of DREADD+/c−Fos+ cells compared to animals injected with vehicle (Supplemental Figure 6B, C). To assess if PeFA neuronal activity is reduced in real time, we injected Ucn3::Cre+ females with conditional inhibitory DREADD with a non-conditional GCaMP, which colocalized in the PeFA (Supplemental Figure 6-1 D, F). Three weeks later, we treated animals with either vehicle or CNO and did 6 tail suspensions for 5 seconds each (Supplemental Figure 6-1 E). We found that CNO administration reduced PeFA GCaMP signal across trials (Supplemental Figure 6-1 G, H) and with a significantly reduced area under the curve and average percent change of fluorescence in tail suspension (Supplemental Figure 6-1 I, J).

Because our CRS paradigm dampened alloparenting in virgin females and increased PeFA^Ucn3 neuronal activation, we aimed to ameliorate behavior deficits by inhibiting PeFA^Ucn3 neurons. To accomplish this, we injected virgin Ucn3::Cre+ and Cre− females with inhibitiory DREADD virus and started CRS 1 week after recovery from surgery. After 16 days of CRS, we tested females with vehicle treatment to observe the impact of stress on pup-directed behavior. The following day we administered CNO to determine if blocking PeFA^Ucn3 neuron activity altered alloparenting. We then observed open field behavior using randomized CNO/Vehicle trials using a crossover design in which each animal received vehicle or CNO on the first day and the alternate drug in the next session. On day 32, we performed another pup exposure with vehicle treatment (Figure 6A). This trial structure was selected based on our previous observation that allowing a pro-parental interaction either through natural pup exposure or via inhibition of PeFA^Ucn3 neurons leads to lasting enhancements in infant-directed behavior¹⁰. Stressed females weighed significantly less than controls (Figure 6B). During open field, CNO administration did not alter exploratory behaviors in either group (Figure 6C-F). In the pup exposure assay, both stressed and control virgin females had improved cumulative retrieval with CNO treatment, which we did not observe in the Cre− group (Figure 6G, H; Supplemental Figure 6-2 A). Strikingly, stressed females given CNO displayed improved latency to retrieve, time spent crouching, time spent in nest and overall time spent parenting which did not occur in unstressed controls or Cre− controls, or in the final vehicle session (Figure 6I, L, M, N; Supplemental Figure 6-2). No significant changes were observed in pup grooming (Figure 6K). Interestingly, pup investigation was significantly decreased in both groups potentially due to familiarity (Figure 7J)^57-59. CNO administration did not improve any parenting measures in stressed Cre− females (Supplemental Figure 6-2). Altogether, inhibition of PeFA^Ucn3 neuronal activity leads to enhanced alloparenting in stressed virgin females (Figure 6O, P).

Figure 6. PeFA^Ucn3 inhibition ameliorates parenting deficits in stressed virgin females.

(A) Schematic of viral injection strategy and behavior timeline (n=7 stressed Cre+ females; n=7 control Cre+ females). (B) Stressed females had a significant difference in weight compared to control females (2-way repeated measures ANOVA, main effect of stress x time F_(16,192)=4.347 p<0.0001). (C-F) Open field results show no difference in time spent in center, border, distance moved, or velocity in control or stressed mice with or without CNO treatment. (G) Cumulative pup retrieval in control animals improves significantly with CNO injection (Friedman’s test; p<0.0001). (H) Cumulative pup retrieval in stressed animals improves with CNO administration (Friedman’s test; p<0.0001). (I) CNO treatment decreases latency to retrieve in stressed females but not in control animals (Two-way repeated measures ANOVA, main effect of drug treatment F_(2,24)=8.978 p<0.01; Sidak’s multiple comparisons test post-hoc effect significant for Cre+ stressed group vehicle versus CNO p<0.0037 and for Cre+ control group vehicle versus vehicle p<0.0223). (J) Pup investigation dramatically reduces in both animal groups with CNO (Two-way repeated measures ANOVA, main effect of drug treatment F_(2,24)=82.86 p<0.0001; Sidak’s multiple comparisons test post-hoc effect significant for both groups vehicle versus CNO p<0.0001 and vehicle versus vehicle p<0.0001). (K) Pup grooming is unchanged with CNO treatment. (L) Time spent crouching increases with CNO administration in stressed females (Two-way repeated measures ANOVA, main interaction effect of stress and drug F_(2,24)=3.506 p<0.0462₎; Sidak’s multiple comparisons test post-hoc effect significant for Cre+ stressed group vehicle versus CNO p<0.0346). (M) Stressed females spent significantly more time in the nest with CNO administration (Two-way repeated measures ANOVA, main effect of drug treatment F_(2,24)=9.399 p=0.001; Sidak’s multiple comparison’s test test post-hoc effect significant for Cre+ stressed group vehicle versus CNO p<0.0008). (N) Cumulative time spent parenting increases with CNO treatment in stressed females (Two-way ANOVA, main effect of drug treatment F_(2,24)=3.942 p=0.0331; Sidak’s multiple comparisons post-hoc effect significant for Cre+ stressed group vehicle versus CNO p<0.0038). (O-P) Representative behavior traces showing induction of more parenting behaviors with CNO in stressed females. (Significant post-hoc comparisons noted as follows: *p=0.05; **p=0.01; ***p=0.001; ****p=0.0001).

Discussion

Previously, we observed that virgin females showing alloparental care toward pups have low-level of activation of PeFA^Ucn3 neurons while strongly activating these cells with chemogenetic or optogenetic methods leads to infant-directed neglect and aggression¹⁰. Here, we further explored the role of PeFA^Ucn3 neurons during female alloparental and maternal behavior. We find that around 20% of PeFA^Ucn3 neurons are active during alloparenting in virgin females, replicating our previous findings¹⁰. However, our data revealed that control bedding exposure and experimental pup exposure conditions impact PeFA^Ucn3 neural activity differentially in lactating females depending on whether her litter is present. Our results indicate that PeFA^Ucn3 neurons may be sensitive to specific social or physiological contexts; in the future, it will be important to tease apart whether this heightened baseline PeFA^Ucn3 neural activity after litter removal may be related to an aversive or stressed state, or possibly a social motivation set point that is altered postpartum. Indeed, previous studies have shown that maternal separation can impact both a mother and offspring’s behavior in measures related to anxiety, social behavior, and cognition^60-62.

A related possibility is that virgin female mice find pup encounters aversive and this valence changes peripartum. Indeed, rat females avoid offspring before becoming pregnant^{63, 64}. In laboratory conditions, virgin female mice show approach and interaction with pups^{8-10, 65, 66} and infant-directed behavior can be improved by guidance from lactating females⁴⁸. This willingness to approach pups could be due to domestication, given that wild or outbred female mice show much higher rates of infanticide^67-70. However, aggression does not necessarily indicate aversion toward a social cue, as aggression toward adult conspecifics can be appetitive^{71, 72}. More research is necessary to specifically address whether adults across sexes and physiological states interpret pup cues as appetitive or aversive and the correspondence with infant-directed behavior.

One potential limitation of our study is that we did not assess estrous phase during behavior testing in virgin females. Studies examining open field behavior indicate that estrous phase does not significantly affect exploration^73-75. Literature on the role of estrogen signaling in female infant-directed behavior in mice is controversial. For example, global knockout of estrogen receptor alpha (Esr1) increases infanticide in virgin female mice and knockdown of Esr1 in the MPOA reduces maternal behavior in lactating females^{76, 77}. Conversely, estrogen supplementation of virgin female mice whether ovariectomized or not can uncover infanticide or inhibit pup retrieval^78-80. Other studies show that infant-directed behavior is enhanced by experience, rather than hormones, in virgin and lactating female mice. For example, virgin females tested across estrous phases respond to natural pup calls at the same rate with similar levels caregiving behavior, on par with parturient mothers^{81, 82}. Conflicting evidence on the role of circulating ovarian hormones on infant-directed behavior in females is likely due to differences in when estrogen signaling is manipulated, whether constitutively, at birth, or during adulthood, due to the different impacts of estrogen signaling on specific brain regions throughout development^{83, 84}. Now that we have established that stress impacts alloparenting, it becomes critical to study how stress intersects with estrous phase in future research. Postpartum females undergo a brief period of fertility within 24 hours of parturition but afterward experience anestrus throughout lactation, so this limitation does not apply to our lactating female data^{85, 86}.

Our data clearly demonstrate that the method for testing infant-directed behavior in lactating feamles impacts observations. In the literature, there are several approaches for studying infant-directed behavior in rodent mothers that are aligned with the parameter of interest⁸⁷. For example, to study ultrasonic vocalizations, experimenters often use altered environmental conditions^{88, 89}, litter separation^{90, 91}, or pup handling^{92, 93} to elicit pup calls. In naturalistic studies, the litter may be removed with some or all pups returned to the homecage, or simply scattered to elicit pup retrieval^94-96, or undisturbed homecage behavior may be monitored continuously over long periods of time^{97, 98}. To study maternal motivation, an operationalized design may be used such as introducing a barrier to climb over or a lever to press to gain access to pups^{8, 99, 100}. With increased interest in the neural circuits underpinning parenting, the field should move toward adopting standardized approaches (experimental design, reporting, and analysis) for studying infant-directed behaviors across sex and physiological states.

Furthermore, our data reveal differences in the baseline activation of PeFA^Ucn3 neurons to pup exposure in early lactating females (Fig. 1) relative to late lactating females (Fig. 5), suggesting that there may be adaptive changes in PeFA^Ucn3 neuron activity corresponding to lactation phase. Previous research shows altered responsivity of PVH oxytocin-expressing neurons to alpha-melanocyte-stimulating hormone between non-pregnant and lactating rats¹⁰¹, suggesting that lactation influences neuronal responses in the hypothalamus. Additionally, lactation stage influences behavioral threat responses in rat dams¹⁰². Future study examining activity in relevant circuitry over time will be necessary to determine if molecular or behavioral changes occur over phases of lactation.

These changes could also be associated with dynamics of the PeFA^Ucn3 neuronal population. We showed using scRNAseq data and MERFISH spatial sequencing that these neurons comprise a single subset⁴². Previous research has established that PeFA^Ucn3 neurons consist of a rostral and caudal subdivision, with the rostral neurons co-expressing Trh^{10, 46, 47}. Importantly, the scRNAseq and MERFISH datasets were generated using mice of both sexes. However, there could be shifts in the molecular signature of PeFA^Ucn3 neurons associated with mating status or physiological state such as pregnancy, lactation phase, or stress that these data do not account for. For example, our data reveal that many receptors sensitive to hormonal regulation and physiological state like pregnancy and lactation are expressed by PeFA^Ucn3 neurons, including Esr1, progesterone receptor (Pgr), and oxytocin receptor (Oxtr). Additionally, there are other definitions of cell type besides molecular expression, including morphology, physiology, or projection patterns¹⁰³. We have previously shown that PeFA^Ucn3 neurons across the rostro-caudal axis bifurcate extensively, but further quantification of the projection pattern and examining the spontaneous and evoked firing properties of these neurons may reveal further insight into their function across physiological states.

Our data with CRS suggests that both virgin and lactating females show habituation based on reduced PVH^CRF neuronal activity in line with previous research showing reduction in Fos colocalization and intrinsic excitability of PVH^CRF neurons after CRS in rodents^{56, 104}. However, this habituation to stress manifests differently in terms of impact on infant-directed behavior across physiological states in females, with virgin females showing reduced alloparenting, but lactating females showing preservation of parenting (Supplemental Figure 6-3). This result was surprising because we anticipated a similar behavioral impact on stressed virgins and mothers. However, previous studies show that lactating females display changes in HPA axis responsivity to stress^{53, 105, 106}, which may account for the behavioral differences. In assessing several stressors in lactating females, we consistently observed a preservation of maternal behavior with different impacts on PeFA^Ucn3 neuronal activation. With CRS and intruder stress, we observed a decrease or trend toward decreased PeFA^Ucn3 neuronal activity which we interpret as a mechanism for preserving parenting. However, in CVS, we observed a positive correlation of PeFA^Ucn3 cell activation with nest-building behavior (Supplemental Figure 6-3). Taken altogether, we interpret that PeFA^Ucn3 neuronal activity may be reflective of stress coping behavior, as the alternative name for the gene, stresscopin¹⁰⁷, implies. Indeed, previous studies have identified neural circuitry that is preferentially involved in active versus passive behavioral coping mechanisms^{108, 109}. We plan to further dissect the relationship of the PeFA^Ucn3 circuitry to stress-coping and parenting in the future.

Conclusion

Overall, we find that PeFA^Ucn3 neuronal activity is higher in females showing less alloparenting. Chronic stress leads to reduced alloparenting accompanied by higher numbers of active PeFA^Ucn3 neurons in virgin females. Blocking activity in PeFA^Ucn3 neurons rescues infant-directed behavioral deficits in virgin females. Together with previous research¹⁰, our results suggest that the level of activity in PeFA^Ucn3 neurons determines the expression of pro- and anti-parental behavior (Supplemental Figure 6-3), and strengthens the idea that these neurons are critical mediators of stress responses in virgin female mice^37-39. These results support the critical role for PeFA^Ucn3 neurons in the neural circuitry controlling female alloparenting and the sensitivity of this behavior to stress.