Sexually dimorphic oxytocin receptor-expressing (OXTR) neurons in the anteroventral periventricular nucleus (AVPV) in the postpartum female mouse are involved in maternal behavior

Abstract

Maternal care is crucial for the survival and development of offspring. Oxytocin modulates maternal behavior by binding to oxytocin receptors (OXTRs) in various parts of the brain. Previously, we showed that OXTRs are expressed in the anteroventral periventricular nucleus (AVPV) of female, but not male mice. Because the AVPV is involved in the regulation of maternal behavior and oxytocin enhances its induction, this finding leads to the hypothesis that the female specific population of OXTR neurons in the AVPV regulates maternal behavior. To address this hypothesis, OXTR-Venus reporter mice were used to assess if expression levels of OXTR in the AVPV are changed during the postpartum period. The total number of OXTR-Venus neurons was significantly greater in postpartum dams compared to virgin females. To assess efferent projections of the AVPV-OXTR neurons, a Cre-dependent fluorescent protein (tdTomato) expressing a viral vector was injected into one side of the AVPV of female OXTR-Cre mice. Fibers expressing tdTomato were found in hypothalamic areas containing oxytocin neurons (the supraoptic and paraventricular nuclei) and the midbrain areas (the ventral tegmental area and periaqueductal gray) that are involved in the regulation of maternal motivation. To assess if activity of the AVPV-OXTR neurons is involved in the regulation of maternal behaviors, a chemogenetic approach was employed. Specific inhibition of activity of AVPV-OXTR neurons completely abolished pup retrieval and nest building behaviors. Collectively, these findings demonstrate that AVPV-OXTR neurons in postpartum female mice constitute an important node in the neural circuitry that regulates maternal behavior.

Introduction

The neurohypophysial hormone, oxytocin, is known for its critical role in female reproductive physiology, specifically uterine contraction during labor and milk ejection while nursing. Oxytocin is synthesized by neurons located primarily in the paraventricular (PVN) and supraoptic (SON) nuclei of the hypothalamus. Oxytocin neurons send long axonal projections into the neurohypophysis where oxytocin is released into the general circulation in response to physiological demands. Oxytocin is also released in the central nervous system and influences several aspects of behaviors, such as social recognition, maternal behavior, pair bonding, and grooming (2–6). These behaviors are influenced by the binding of oxytocin to oxytocin receptors (OXTR) located in various parts of the brain (7, 8).

In our previous study, OXTR neurons in virgin female mice were found throughout the entire medial preoptic area (MPOA); however, ovariectomy abolished the OXTR neurons in the ventral half of the periventricular zone approximately 130 μm from the wall of the third ventricle without affecting the number of OXTR neurons outside of this area within the MPOA (9). A sex difference in the expression of OXTR was also found as there were virtually no OXTR neurons in the periventricular region of males as in ovariectomized females (9). Since this region reasonably corresponds to the area described as the anteroventral periventricular nucleus (AVPV) in other studies (10–16), the periventricular region where OXTR neurons were eradicated after ovariectomy was defined as the AVPV as in our previous (9) and present studies.

The AVPV is known as the sexually dimorphic brain area consisting of a cluster of neurons critically involved in coordinating sexual dimorphism in reproductive physiology and behavior (13). For examples, corticotropin releasing hormone receptor 1 (CRFR1)-expressing neurons (11, 17), progesterone receptor neurons (16), and tyrosine hydroxylase immunoreactive (TH⁺) neurons (13–15) were more abundant in the AVPV of female mice than in male mice. In particular, sexually dimorphic TH⁺ neurons in the mouse AVPV were found to regulate maternal behavior in a sex-specific manner (15). Since most of TH⁺ neurons in the AVPV express DOPA (3,4-dihydroxyphenylalanine) decarboxylase (DDC), the enzyme responsible for the formation of dopamine from DOPA (15), these neurons are likely dopaminergic neurons described as A14 cell groups located in the periventricular region (18–20). The A14 dopaminergic cell groups arborize extensively and innervate several hypothalamic nuclei, such as the PVN (19, 20) and SON (21). Because dopamine signaling participates in oxytocin secretion (22–27) and enhance mother–pup interactions (28, 29), TH⁺ neurons may interact with oxytocin neurons to regulate maternal behavior.

During pregnancy, estrogen and progesterone progressively increase; however, prior to birth, progesterone levels fall while estrogen remains elevated until birth (30). This estrogen surge following progesterone withdrawal at parturition is required for proper onset of maternal behavior (30–35). High levels of the estrogen receptor alpha (ERα) were found in the MPOA (36), and the nuclear estrogen receptor binding in the MPOA increased at the onset of maternal behavior (37). Moreover, treatment with estrogen in virgin female rats resulted in a significant increase in OXTR binding in the MPOA (38). An estradiol implant into the MPOA stimulated maternal behavior (39) whereas treatment with an OXTR antagonist directly into the MPOA blocked that behavior (40, 41). Together with the presence of ERα in OXTR neurons in female AVPV (9), these findings suggest that oxytocin mediates the effect of estrogen on maternal behaviors via OXTR neurons in the AVPV within the MPOA.

The present study was conducted to investigate the contribution of sexually dimorphic estrogen-dependent OXTR neurons in the AVPV (AVPV-OXTR neurons) in the regulation of maternal behaviors. In addition, because the location and morphology of TH⁺ neurons (15) resemble that of OXTR neurons in the AVPV, the possibility that TH⁺ neurons express OXTR was also investigated. Using a quantification of AVPV-OXTR neurons, cell-type specific neural tracing, and a chemogenetic approach to specifically manipulate the activity of AVPV-OXTR neurons in vivo, our data show that a sexually dimorphic cluster of AVPV-OXTR neurons is a critical node in the neural circuit regulating maternal behavior in postpartum dams.

Materials and Methods

Animals

OXTR-Venus mice in which a part of the OXTR gene was replaced with Venus (a variant of the green fluorescent protein) cDNA (42) were originally provided by Dr. K. Nishimori of the Tohoku University in Japan. A colony was established in a facility at Louisiana State University by backcrossing OXTR-Venus mice with C57BL/6J (# 000664, Jackson Lab, Bell Harbor, ME) mice for more than 10 generations. OXTR^{tm1.1(cre)Hze} mice, also known as OXTR-T2A-Cre-D, in C57BL/6J background were purchased from Jackson Laboratory (# 031303, Bell Harbor, ME). OXTR-T2A-Cre-D mice has a T2A sequence and a Cre recombinase gene inserted immediately at the OXTR translational stop codon so that the endogenous promoter drives Cre expression in OXTR-expressing neurons (43). The OXTR-T2A-Cre-D mice were used for Cre dependent viral expression of Designer-receptors exclusively activated-by-designer drugs (DREADDs). OXTR-T2A-Cre-D mice were also crossed with a floxed tdTomato reporter strain (Ai14, # 007914, Jackson Lab, Bell Harbor, ME) for a set of experiments. For genotyping, genomic DNA was isolated from tail snips by incubating the tissues with REDExtract-N-Amp^™ Tissue PCR Kit (Sigma, St. Louis, MO). The isolated DNA was subsequently genotyped using the following sets of primers: F (5’-GTTGGGAACAGCGGTGATTA-3’) and R (5'-GGCTCAGGCTTTCTCTACTT-3') for OXTR-Venus mice; F (5’-CATGGATCTACATGCTCTTCA-3’), R (5'-ACGTCTCCGCATGTCAGAAGACT-3'), and R (5'-GACCCTGTCATCTCCCATGGA-3') for OXTR-T2A-Cre-D mice. Only heterozygous mice were used for the experiments. Males and females were housed in separate cages (maximum 4 mice/cage) in the same room on a 12:12 h light/dark cycle (lights off at 15:00) with access to food and water available ad libitum. All protocols and animal experiments were approved by the Institutional Animal Care and Use Committee at Louisiana State University.

All experiments used young adult mice of the same age and weight (8–11-week-old, 20–25 g) from our breeding colony to prevent age from becoming a variable in the studies. All lactating dams were the first-time dams to avoid previous experience with pups becoming a variable. Animals were randomly assigned to treatment groups for all experiments.

Immunohistochemistry

Mice were deeply anesthetized with Ketamine-Xylazine (9:1; 100 mg/kg; i.p.) and transcardially perfused with 0.01M sodium phosphate buffered saline (PBS; pH 7.2), followed by 4% paraformaldehyde in 0.1 M sodium phosphate buffer (PB; pH 7.2). The brains were extracted and postfixed overnight in the same fixative. After infiltration with 20% sucrose in 0.1 M PB for cryoprotection for 12 hr., coronal sections were transected at 40 μm with a sliding microtome (Leica SM2010R; Mannheim, Germany).

To enhance the signal of Venus, immunocytochemical localization of Venus with anti-green fluorescent protein (GFP) antibody (ab13970; abcam, Cambridge, UK) was conducted in the brain sections obtained from adult virgin male, virgin female, and postpartum female OXTR-Venus age-matched siblings. Free-floating brain slices were incubated overnight with primary antibodies against GFP at dilutions of 1:10,000 in PBS containing 0.5% Triton X-100 (PBST) with continuous gentle agitation at 4°C. Brain sections were subsequently incubated with a secondary antibody (goat anti-chicken) conjugated with Alexa Fluor488 (Code# 103-545-155, Jackson ImmunoResearch, West Grove, PA) for 4 hr at room temperature. The anti-GFP antibody showed no specific binding in brain sections from wild-type animals in our previous study (9). For double-fluorescence imaging of Venus and immunohistochemistry of tyrosine hydroxylase (TH), brain sections were processed overnight with polyclonal anti-TH antibody (Cat#AB152; EMD Millipore, Billerica, MA) at dilutions of 1:5,000 in PBST with continuous gentle agitation at 4°C. The sections were subsequently incubated with a secondary antibody conjugated with fluorescence marker (AffiniPure Goat Anti-Rabbit IgG DyLight594; Jackson ImmunoResearch, West Grove, PA) at 1:250 dilution for 2–4 hr at room temperature. The sections were mounted in polyvinyl alcohol (PVA) with the anti-fading agent 1,4-Diazabicyclo[2.2.2]octane (DABCO) that consists of 4.8 g PVA, 12 g glycerol, 12 ml dH₂O, 24 ml 0.2 M Tris-HCl and 1.25 g DABCO. Confocal fluorescence microscopic images (1,024 × 1,024; 1 μm optical section thickness) were acquired with a confocal microscope (Leica TCS SP8 spectral confocal microscope, Mannheim, Germany).

To visualize the cellular distribution of Cre expressing neurons, male and female brains from OXTR-T2A-Cre-D mice crossed with Ai14 (OXTR cre^+/tdTomato) were sectioned and mounted as described above. For a subset of experiments, the sections were also processed overnight for the anti-TH antibody at dilutions of 1:5,000 in PBST with continuous gentle agitation at 4°C. The sections were subsequently incubated for 4 hr at room temperature with a secondary antibody conjugated with fluorescence marker (AffiniPure Goat Anti-Rabbit IgG Alexa Fluor488; Jackson ImmunoResearch, West Grove, PA) at 1:250 dilution. Fluorescence microscope images (1280 × 1024) were acquired digitally (Nikon Eclipse 80i equipped with DS-QiMc digital camera, Nikon, Tokyo, Japan).

Cell Count and colocalization analysis

OXTR-Venus and TH⁺ neurons in the AVPV started appearing in the area surrounding the organum vasculosum laminae terminalis (OVLT) at the ventral end of the third ventricle, bregma 0.5 mm in the mouse brain atlas (44) and was continuously found in the next 20–25 consecutive 40 μm-thick coronal sections. All brain sectioned (40 μm-thick) from 5 sections (200 μm) anterior and 20 sections (800 μm) posterior to the level of the anterior commissure (AC) were collected, processed for immunohistochemistry, and digitally imaged. The number of OXTR-Venus neurons in the AVPV on every brain section was counted from the level of the OVLT to the level of the suprachasmatic nucleus (SCN) where most of OXTR-Venus/ TH⁺ neurons were found. Detection of colocalization was performed using the ImageJ Plugin application, JACoP (Bethesda, NIH). A Li’s IC value between 0.3 – 0.5 was considered as colocalization.

Viral Vectors

Cre-recombinase-dependent adeno-associated virus (AAV) containing a double-floxed inverted open reading frame encoding a fluorescent protein (mCherry) with Designer-receptors exclusively activated-by-designer drugs (DREADD) was used for neuron-specific silencing of OXTR-Cre neurons. pAAV-hSyn-DIO-mCherry (control, 2.1× 10¹³ vg/ml) and pAAV-hSyn-DIO-hM4D(Gi)-mCherry (hM4Di, 2.3× 10¹³ vg/ml, (45)) were gifts from Bryan Roth (Addgene viral prep #50459-AAV9 and #44362-AAV9, respectively). AAV-FLEX-rev-ChR2-tdtomato (2.1× 10¹³ vg/ml, (46)) was a gift from Scott Sternson (Addgene viral prep #18917-AAV9) and used for the neural tracing study. All viruses were stored in aliquots at −80°C until use.

Stereotaxic Injection

Female mice (8-week-old) were deeply anesthetized with 1.5–2% isoflurane (4% for induction) in oxygen at a rate of 2L/min, and their heads were immobilized within the frame of a stereotaxic instrument (#940, David Kopf Instruments, Tujunga, CA). Prior to operation, mice received Carprofen (5 mg/kg) and lactated ringers for hydration subcutaneously. Following sterilization, an incision was made to expose the skull. Access holes were drilled and a guide cannula (#C315G/SPC, Plastics One, Roanoke, VA) was inserted into the AVPV (anteroposterior, 0.55 mm; mediolateral, ±0.15 mm; dorsoventral, −5.4 mm) according to the mouse brain atlas (44). An injector filled with virus and attached to a microliter syringe (#7000.5, Hamilton Company, Reno, NV) was threaded into the guide cannula, and a volume of 120 nl/site was slowly infused into the target area at a rate of 20 nl/30s. The guide cannula and injector remained in place for 5 min before removal to prevent backflow. Skull access was sealed with bone wax (#DYNJBW25, Medline Industries, Northfield, IL), and the incision was sutured. Buprenorphine (0.1 mg/kg) was injected subcutaneously every 12 hr as needed for pain management. After the surgery, mice were housed singly for two weeks to allow for recovery and viral expression before pairing them with males for breeding or for histological analysis. Cannula placement and viral spread were verified at the end of all experiments.

To validate the site of injection, brains from dams were harvested and sectioned immediately after the behavior tests. Free-floating sections were mounted as described above, and fluorescent microscopic images (1280 × 1024) were acquired digitally (Nikon Eclipse 80i with DS-QiMc digital camera, Nikon, Tokyo, Japan). Digital images were minimally altered in dynamic range with ImageJ software (Bethesda, NIH).

For the neuronal tracing experiment, AAV-FLEX-rev-ChR2-tdtomato was unilaterally injected into the AVPV of OXTR-Cre female mouse following the stereotaxic injection procedure above. After the surgery, mice were housed singly for two weeks to allow for recovery and co-housed with another female for an additional week to allow for viral expression. Brains were collected and imaged as described above.

Behavior Analysis

Animal behaviors were video recorded from above during the pup retrieval test and from the side of the cage for the home cage test. A CCD monochrome camera (Stingray F-145B, Allied Vision, Exton, PA) with commercial video acquisition software (StreamPix 8, Norpix, Quebec, Canada) was used in a semi-dark room with infrared illumination (< 5 lux) at a frame rate of 16 frames/s. All tests were performed within the first 4 hrs of dark cycle.

After becoming visibly pregnant, the female mice were individually housed (cage length 32 cm, width 17.5 cm, and height 14 cm) and monitored for pup birth. The pup retrieval assay was performed in the female's home cage, which contained bedding and a plastic dome. On the test day, animals were moved into the testing room and left untouched for an hour followed by an i.p. injection of saline or Clozapine-N-Oxide (CNO, Cat. #4936, Tocris, Minneapolis, MN). CNO was dissolved in 0.9% saline and used at a dose of 3 mg/kg to activate inhibitory DREADD (hM4Di). The same volume of saline was used as vehicle control. After 30 min post-injection, the mother and pups were briefly removed from their home cage, and the nest was disturbed with nesting material scattered throughout the cage. A new dome was placed in one corner and three pups were scattered on the opposite side. The dam was then reintroduced into the cage and her behavior was video recorded for 10 min. Failure to retrieve all three pups within 10 min resulted in termination of the retrieval test was reported as maximum duration (600 sec). All the pups from the litter were returned at the end of the retrieval test. Latency to retrieve the 1st, 2nd, and 3rd pup was quantified during a 10-min period. Following the pup retrieval test, the animals were left to acclimate for 5 min before the home cage observations were captured. The mother’s behavior was recorded for a 1 hr period without disturbing the home cage. Quantified were the time spent inside the nest, pup directed licking, crouching over the pup (lactating position), self-grooming, eating, and drinking. Two individuals blind to the study performed manual behavioral annotation on a frame-by-frame basis. The description of the annotation for each maternal behavior can be found in publications (47, 48). The data collected was compared to the annotations from the experimenter for consistency.

Ex vivo Electrophysiology

Slice preparation:

Following behavioral recordings, some mice were euthanized for whole-cell patch clamp electrophysiology. Mice were deeply anaesthetized using Ketamine-Xylazine (9:1, 100 mg/kg i.p.) and perfused through the heart with cold artificial cerebral spinal fluid (ACSF) in which NaCl was replaced by equiosmolar sucrose. Brains were then removed and coronal slices (200 μm) containing the hypothalamic periventricular nucleus were collected using a vibrating microtome (Leica VT1200; Leica, Mannheim, Germany). Brain slices were kept in ACSF (in mM: 125 NaCl, 2.5 KCl, 1 MgSO₄, 1.25 NaH₂PO₄, 26 NaHCO₃, 20 D-glucose, 2 CaCl₂, 0.4 Ascorbic acid, pH of 7.3–7.4, with an osmolality of 290–300 mOsm/kg H₂O).

Whole cell patch clamp recording:

Patch clamp electrophysiology recordings were performed on virus injected slices containing the AVPV region. Labeled neurons in the AVPV were identified with an epifluorescence microscope (Olympus BX51WI, Tokyo) equipped with a 40x water immersion lens (0.8 n.a.) and a CMOS sensor camera (ORCA-Flash4.0LT, Hamamatsu Photonics, Hamamatsu). Whole-cell recordings were obtained with an Axopatch 700B amplifier (Molecular Devices, Foster City, CA). Traces were acquired digitally at 20 kHz and filtered at 5 kHz with a Digidata 1440A in conjunction with PClamp 10 software (Molecular Devices, Foster City, CA). Patch electrodes from borosilicate capillary glass tubing (1.1 mm ID, 1.5 mm OD, Sutter Instruments, Novato, CA) were drawn with a Flaming/Brown micropipette puller (Model P-1000, Sutter Instrument Novato, CA) to have a resistance of 4–8MΩ when filled with the pipette solution that contained (in mM): 140 K-Gluconate, 1 MgCl₂, 10 HEPES, 10 CaCl₂, 2 ATP(Mg⁺⁺), and 0.4 GTP (Na⁺) and 11 EGTA. ACSF was saturated with 95% O₂/5% CO₂ and was warmed to 33–34°C during the recordings. Picrotoxin (100 μM) and 6,7-dinitro-quinoxaline-2,3(1H,4H)-dione (DNQX; 10 μM) were also added to ACSF to block the synaptic activity. Activation of DREADD was obtained using 5 μm CNO applied in the bath. The data presented are not corrected for liquid junction potential estimated to be +9.2mV.

Data Analysis:

To observe the effect of oxytocin and CNO on OXTR-DREADD expressing neurons, membrane potentials were recorded in the presence and absence of oxytocin (100 nM) or CNO (500 nM). Depending on the cell, a steady state depolarizing or hyperpolarizing current was injected to hold the membrane potential slightly below threshold. Mean membrane potentials were obtained by averaging the membrane potential in 30 sec time frames. These time frames were captured at 30 sec before the application of OT, two min after the oxytocin application and 4 min after bath application of CNO. These sampling durations and points were selected because the peak response or plateauing phase of the membrane potential occurred at these time points.

Statistical Analysis

Statistical analyses were performed using JMP 16.0 (SAS Institute Inc). Data are reported as mean ± SEM. Box and whisker plots represent numerical data. The box extends to the quartiles of the data points, and the whiskers extend to the farthest data points. The mean and median are represented by a filled circle and a line in the box, respectively. The total number of OXTR-Venus, TH⁺, TH⁺-OXTR cells in the AVPV from virgin female, virgin male, and postpartum female mice were analyzed with one-way ANOVA. The differences in the total cells number between groups were assessed by the Tukey-Kramer Honestly Significant Difference (HSD) multiple comparisons test. Behavioral assays to compare between wild type and OXTR-Cre mice were analyzed using the studenťs t test. In the electrophysiology recordings, the changes in membrane potential in response to oxytocin and CNO were independently analyzed using the one-way repeated measures multivariate analysis of variance (MANOVA). For AAV injected mice, the difference in behavioral measures among experimental groups under the same condition (saline or CNO injection) were analyzed by a one-way ANOVA. The post-hoc Tukey-Kramer HSD compared differences between each group within the condition. The repeated measures MANOVA with group as a fixed factor was used to assess the effect of CNO. Differences were considered to be statistically significant at p <0.05.

Results

Distribution of OXTR-Venus and TH immunoreactive (TH⁺) neurons in the AVPV of virgin male, virgin female, and postpartum female mice

Our previous study found that OXTR-Venus neurons in the AVPV (AVPV-OXTR-Venus neuron) were found almost exclusively in females (9). We first assessed if the number of AVPV-OXTR-Venus neurons in postpartum females (dams) differed from that of virgin females. Since TH expressing neurons in the AVPV regulate maternal behavior in a sex-specific manner (15), the possible presence of TH immunoreactivity in AVPV-OXTR-Venus neurons was also assessed.

Brains from heterozygous (+/−) OXTR-Venus virgin females (n=8), postpartum dams (n=7), and virgin males (n=6) were coronally sectioned at 40 μm, and all sections were collected from 5 sections (200 μm) anterior and 20 sections (800 μm) posterior to the level of the anterior commissure (AC). OXTR-Venus and TH immunoreactivity positive (TH⁺) neurons in the AVPV were observed in brain sections from both virgin females and dams (Fig. 1A and B). Some of the OXTR-Venus neurons were TH⁺ in both virgin females and dams (Fig. 1A and B). Unlike in females, a few to no OXTR-Venus neurons were found in the AVPV of virgin males (Fig. 1 C), consistent with our previous work (9). Unlike OXTR-Venus neurons, TH⁺ neurons were observed in the AVPV of virgin males (Fig. 1 Cii), but none of the TH⁺ neurons in males were OXTR-Venus neurons (Fig. 1 Ciii).

Fig. 1. Double confocal fluorescence photomicrographs of OXTR-Venus and TH immunoreactivity in the AVPV from a virgin female, postpartum dam, and virgin male OXTR-Venus (+/−) mouse.

OXTR-Venus neurons were seen in the AVPV of the virgin female (Ai) and postpartum dam (Bi), but not in the virgin male (Ci). TH⁺ neurons were seen in the AVPV of the virgin female (Aii), postpartum dam (Bii), and virgin male (Cii). The merged images show some OXTR-Venus neurons were TH⁺ in the virgin female (Aiii) and postpartum dam (Biii) whereas no TH⁺ neurons were OXTR-Venus neurons in the virgin male (Ciii). Scale bar = 100 μm. 3V: the third ventricle.

To compare the overall distribution and number of OXTR-Venus and TH⁺ neurons between the groups, the number of OXTR-Venus neurons, TH⁺ neurons, and OXTR-Venus neurons that were also immunoreactive to TH (TH⁺ OXTR-Venus neurons) in the AVPV on each brain section were counted and plotted from anterior to posterior order (Fig. 2 Ai, Bi, and Ci, respectively). The OXTR-Venus neurons first appeared anteriorly at the level of the organum vasculosum laminae terminalis (OVLT, section #−5), and the number of the neurons in both virgin females and dams increased and peaked at section #2 (80 μm posterior to the AC). In virgin females, the mean number of OXTR-Venus neurons/section gradually declined thereafter until section #20 (800 μm posterior to the AC). In dams, the mean number of OXTR-Venus neurons/section declined to section #11 (440 μm posterior to the AC) and increased slightly again to section #13 (520 μm posterior to the AC), forming a small secondary peak, then declined thereafter. Although the mean number of OXTR-Venus neurons in each brain section was significantly larger in dams than in virgin females, the difference was greater in the anterior half of the AVPV (sections #−5 to 8) compared to the posterior half (sections #8–20). In virgin males, no OXTR-Venus neurons were found in approximately 40% of brain sections (9.5 ± 2.26 sections out of 25 total). When OXTR-Venus neurons were present in males, the maximum number observed per section was three.

Fig. 2. Distribution and number of OXTR-Venus and TH⁺ neurons in the AVPV.

A serial brain sections from 5 sections anterior to and 20 sections posterior to the anterior commissure (AC) was obtained from each animal. The average number of OXTR-Venus neurons (Ai), TH⁺ neurons (Bi), and OXTR-Venus neurons that were TH⁺ (TH⁺-OXTR-Venus neurons: Ci) in the AVPV/section from virgin females, postpartum dams, and virgin males were plotted with individual standard error bars. Aii, Bii, and Cii show the mean total number of OXTR-Venus, TH⁺, and TH⁺-OXTR-Venus neurons, respectively. Post-hoc multiple comparison test was used to compare the differences between groups.

The total number of OXTR-Venus neurons among all groups was significantly different (F_{(2, 18)} =146.3, P<.0001, one way ANOVA). Post-hoc multiple comparison test (the Tukey HSD test) found that the total number of OXTR-Venus neurons was significantly greater in dams (1371 ± 89 neurons) than in virgin females (644 ± 23 neurons, p<0.0001) and virgin males (25 ± 2 neurons, p<0.0001). Virgin males had significantly less total number of OXTR-Venus neurons compared to virgin females (p<0.0001, Fig. 2 A ii).

Like OXTR-Venus neurons, TH⁺ neurons in the AVPV first appeared anteriorly at the level of the OVLT (section #−5). In virgin females, the number of the TH⁺ neurons increased and peaked at section #2 (80 μm posterior to the AC) where the number of OXTR-Venus neurons/section also peaked. However, in dams, the number of TH⁺ neurons did not reach its peak until section #5 (200 μm posterior to the AC). In both virgin females and dams, the number of TH⁺ neurons/section declined by approximately the same rate from their peaks. In contrast to OXTR-Venus neurons, the difference in the number of TH⁺ neurons/section was greater in the posterior half of the AVPV (sections #8 to 20) compared to the anterior half (sections #−5 to 8). In virgin males, considerably more TH⁺ neurons than OXTR-Venus neurons were found in each section. The increase in the number of TH⁺ neurons/section in virgin males was gradual and did not reach its peak until section #8 (320 μm posterior to the AC). The decline of the neuron number from its peak (section #8) was similar to that in virgin females, and there was no significant difference in the number of TH⁺ neurons/section between virgin males and females in sections #8–20. The mean total number of TH⁺ neurons among groups was significantly different (F_{(2, 18)} =30.2, P<.0001, one-way ANOVA). The total number of TH⁺ neurons was significantly greater in virgin females (953 ± 34 neurons) than in virgin males (622 ± 18 neurons; p=0.004, the Tukey HSD test, Fig. 2 B ii). Significantly more total TH⁺ neurons (1316 ± 97 neurons) were present in dams compared to both virgin females (p=0.001, the Tukey HSD test, Fig. 2 B ii) and males (p<0.0001, the Tukey HSD test, Fig. 2 B ii).

The total number of OXTR-Venus neurons (1371 ± 89) and TH⁺ neurons (1316 ± 97) were similar among dams; however, only ~25% (339 ± 41) of OXTR-Venus neurons were immunoreactive to TH (TH⁺ OXTR-Venus). Significantly more TH⁺ neurons (953 ± 34) than OXTR-Venus neurons (644 ± 23) were found in virgin females, and approximately 30% (198 ± 7) of the OXTR-Venus neurons were immunoreactive to TH. A significant difference in the number of TH⁺ OXTR-Venus neurons/section between virgin females and dams was observed in the anterior one third (section #−5 to 1) and posterior one third (section #12–20) of the AVPV. The total number of TH⁺ OXTR-Venus neurons among all groups was significantly different (F_{(2, 18)} =45.16, P<.0001, one way ANOVA, Fig. 2 Cii). Post-hoc multiple comparison test (the Tukey HSD test) found that the total number of TH⁺ OXTR-Venus neurons was significantly greater in dams (339.0 ± 40.9) than in virgin females (197.5 ± 6.6, p=0.0012) and virgin males (3.5 ± 0.5, p<0.0001). In addition, virgin males had significantly less total number of TH⁺ OXTR-Venus neurons compared to virgin females (p<0.0001).

Projection sites of OXTR neurons from the AVPV

To assess efferent projections of the AVPV OXTR cells, a Cre-dependent viral vector containing a double-floxed inverted open reading frame encoding a fluorescent protein (tdTomato) was injected into a side of the AVPV of virgin female OXTR-Cre mice (n=3). Histological analysis following a 3-week period to allow viral expression showed a robust tdTomato expression in neurons in the AVPV and fibers in various parts of the brain. At the level of the OVLT (Fig. 3 A; Bregma 0.62 mm), a dense group of labeled fibers was observed in the OVLT and the MPOA (Fig. 3 Ai). A few sparse labeled fibers were also found in the ventral pallidum (Fig. 3 Ai; VP) and nucleus accumbens (Fig. 3 Aii; NAc). At the level of the AC (Fig. 3 B; Bregma 0.14 mm), labeled fibers were mostly restricted within the AVPV (Fig. 3 Bi). However, a few sparsely distributed fibers were observed in the MPOA (Fig 3 Bii), and posterior and ventral bed nucleus of the stria terminalis (Fig. 3 B; pBNST and vBNST). Some labeled fibers were also observed in the AVPV and MPOA on the contralateral side (Fig. 3 Bi and Bii). The majority of labeled fibers that arose from the injected site in the AVPV coursed caudally through the periventricular zone of the hypothalamus. At the level of the supraoptic nucleus (SON; Fig. 3 C; Bregma −0.58 mm), labeled fibers were identified coursing parallel to the optic tract, projecting towards the SON (Fig. 3 Ci). Dense clusters of labeled fibers were observed in the anterior hypothalamic area (AHA) and suprachiasmatic nucleus (SCN: Fig. 3 Cii). In addition, projections to the SCN crossed to the contralateral side at the ventral end (Fig. 3 Cii). At the level of the retrochiasmatic area (RCA; Fig. 3 D; Bregma −0.94 mm), a robust tdTomato expression was observed at the ventral edge of the third ventricle (3V) in the RCA (Fig. 3 Dii). Labeled projections passing through this region were found to travel dorso-laterally towards the magnocellular preoptic nucleus (MCPO; Fig. 3 Dii) on the ipsilateral as well as contralateral side. At the level of the median eminence (ME; Fig. 3 E; Bregma −1.58 mm), robust labeled projections were observed in the ME, arcuate nucleus (Arc), and in the ventrolateral part of the ventromedial hypothalamus (VMH; Fig. 3 Ei). Sparsely distributed labeled projections were observed coursing dorso-laterally along the cerebral peduncle (cp) towards the stria terminalis (st) and ventral posterolateral thalamic nuclei (Fig 3 Eii; VPL). Finally, in the midbrain (Fig. 3 F), labeled projections were observed in the periaqueductal gray (PAG; Fig. 3 Fi) and ventral tegmental area (VTA: Fig. 3 Fii).

Fig. 3. Projections of OXTR neuron in the AVPV.

Fluorescent photomicrographs (A-F in the left panel) of whole coronal brain sections of OXTR-Cre mice unilaterally injected with a Cre-dependent tdTomato expressing viral vector into the AVPV. Photomicrographs in the middle (Ai-Fi) and right (Aii-Fii) panels show higher power views of regions where tdTomato expressing nerve fibers were observed in the photomicrographs in the left panel. Scale bars = 1 mm in the left panel, = 100 μm in the middle and right panels.

OXTR-Cre model replicates sex difference in AVPV OXTR neurons without impacting maternal care

Because specific expression of DREADDs in OXTR neurons depends on a Cre recombinase-dependent adeno-associated virus (AAV) vector and specific expression of Cre recombinase by the endogenous OXTR promoter sequences, the expression of Cre-recombinase in the brain and the maternal behavior of OXTR-T2A-Cre-D knock-in (OXTR-Cre) mice was first assessed. To visualize the expression of Cre recombinase expressing neurons, OXTR-Cre mice were crossed with homozygous Ai14 mice, a Cre reporter tool strain. When Ai14 mice are bred to mice that express Cre recombinase, the resulting offspring will have expression of a CAG promoter-driven red fluorescent protein variant (tdTomato) in Cre recombinase expressing neurons (49). Brains from age matched (10-week-old) OXTR-Cre x Ai14 virgin female (n=5) and male (n=5) siblings were coronally sectioned and the expression of the tdTomato in neurons were observed. A sexually dimorphic distribution of the tdTomato-expressing neurons occurred in the AVPV where tdTomato-expressing neurons were observed in females, but not in males (Fig. 4 A & B). To assess whether OXTR-Cre mice express normal maternal behaviors, behaviors of OXTR-Cre dams (n=5) were compared to that of wild-type mice (WT, n=6). Both OXTR-Cre and WT dams successfully retrieved all three pups scattered in the opposite side of the cage. No difference was found in the cumulative latency between OXTR-Cre and WT to retrieve the 1^st (t₍₉₎=0.925, p=0.417), 2^nd (t₍₉₎=0.986, p=0.395) and 3^rd pup (t₍₉₎=0.811, p=0.481, Fig. 4 C & Table 1). None of the parameters measured in an hour-long home cage test showed a difference between OXTR-Cre and WT dams (Fig. 4 D and Table 1). The parameters measured were the total time spent: inside the nest (t₍₉₎=1.06, p=0.159), building the nest (t₍₉₎=0.44, p=0.671), crouching over the pups (t₍₉₎=0.601, p=0.563), grooming the pups (t₍₉₎=0.19, p=0.856), and grooming themselves (t₍₉₎=0.711, p=0.495). Moreover, there was no difference in the average number of pups born per litter between OXTR-Cre (7.1 ± 0.37 pups) and WT (7.4 ± 0.37 pups) (Fig. 4 E, t₍₃₈₎=0.583, p=0.282). The average number of pups that survived at weaning from OXTR-Cre (6.35 ± 0.29 pups) was not different from WT (6.6 ± 0.41 pups) (Fig. 4 E, t₍₃₄₎=0.494, p=0.312). Lastly, there was no difference between the body growth of pups from OXTR-Cre and WT mice (Fig. 4 F, t₍₃₈₎=0.355, p=0.362). Taken together, these data showed that OXTR-Cre dams exhibit normal maternal behaviors, and OXTR-Cre mouse strain is an appropriate model to use for the present DREADD study to investigate the contribution of AVPV OXTR neurons to maternal behavior.

Fig. 4. Characteristics of OXTR-Cre mice.

A. Fluorescent photomicrographs of coronal sections from a female and a male OXTR-Cre-reporter mice expressing tdTomato under the control of OXTR promoter. B. Higher power view of photomicrographs in A showing the AVPV. OXTR-promotor driven tdTomato expressing neurons were seen in the female, but not in the male. C. The mean latency to retrieve pups of wild type (WT, n=6) and OXTR-Cre mice (n=5) did not differ. D. Results from an hour-long home cage test: There is no significant difference between WT (n=6) and OXTR-Cre (n=5) mice in mean time spent inside nest, building nest, crouching over pups, grooming pups, and self-grooming. E. The average litter size and number of pups weaned did not differ between WT (n=20) and OXTR-Cre (n=20) groups. F. The average body weight gain of pups did not differ between WT (n=20) and OXTR-Cre (n=20) groups. In C, D, E, and F, the box extends to the quartiles of the data points, and the whiskers extend to the farthest data points. The mean and median are represented by a filled circle and a line in the box, respectively. The numerical values between WT and OXTR-Cre groups were compared using the student’s t test.

Table 1.

Behavior test results from Wild Type (WT) and OXTR-Cre mice

Genotype	Pup retrieval	Inside nest	Building nest	Crouching over pups	Grooming pups	Grooming themselves
WT	94.5 ± 42.6 s	33.3 ± 7.7 min	7.8 ± 2.9 min	23.9 ± 7.5 min	16.0 ± 2.7 min	2.8 ± 0.8 min
OXTR-Cre	59.8 ± 4.5 s	22.4 ± 5.8 min	9.8 ± 3.6 min	18.0 ± 6.2 min	17.3 ± 6.4 min	2.1 ± 0.7 min

Highly targeted expression of DREADD occurred in OXTR neurons in the AVPV

To assess whether activity of the OXTR neurons is involved in expression of maternal behaviors, a chemogenetic approach with DREADDs was employed. Specific expression of DREADDs in OXTR neurons in the AVPV was achieved by the stereotaxic injection of a Cre recombinase-dependent AAV vector into OXTR-Cre mice. AAV expressing inhibitory DREADD and mCherry (hM4Di: pAAV-hSyn-DIO-hM4D(Gi)-mCherry) or mCherry only (CTRL: pAAV-hSyn-DIO-mCherry) under control of a hSyn promoter was injected into both sides of the AVPV to modulate OXTR neurons (Fig. 5 A). To avoid surgery during pregnancy, stereotaxic injections of the AAV viral vectors were performed on 8-week-old female mice 1–2 weeks before they were housed with males for breeding (Fig. 5 B). Following the successful delivery of pups, the maternal behaviors of dams were recorded on postpartum day 1. After the behavior tests, the brains were harvested for evaluation of DREADD expression and injection sites by the localization of the reporter molecule, mCherry, on the brain slices (Fig. 5 C). A robust expression of mCherry in neurons on both sides of the AVPV (bilateral) was observed in some animals, whereas such expression was observed only on one side (unilateral) in others. To limit the lateral diffusion of the AAV, a relatively small amount of AAV was injected into each side since the AVPV is a narrow strip of the periventricular region of the third ventricle (3V). The resulting mCherry expressing neurons formed a cluster that appeared in 3 consecutive 40 μm-thick brain sections. Therefore, the DREADD expression in the AVPV was estimated to be up to 120 μm in the anterior-posterior direction. The injection was targeted 80 μm posterior to the AC where the highest population of OXTR neurons were found (Fig. 2A) to recruit as many OXTR-Cre neurons as possible by a single injection. However, it is clear that the DREADD expression did not occur in the entire population of AVPV-OXTR neurons. Dams were grouped based on the type of DREADD and whether bilateral or unilateral injection occurred: 1) bilateral hM4Di (n=6); unilateral hM4Di (n=5); unilateral CTRL (n=5). In some cases, the harvested brains were also used for whole-cell patch-clamp electrophysiology to assess the effect of oxytocin and hM4Di on the mCherry expressing neurons (Fig. 5 D). A bath application of oxytocin caused membrane depolarization and a repetitive firing of action potentials, whereas the application of an engineered ligand of hM4Di, clozapine-N-oxide (CNO), caused significant membrane hyperpolarization in all mCherry expressing neurons recorded (Fig. 5 E&F). The mean membrane potential (Vm) measured in the presence of 100 nM oxytocin (−55.32 ± 1.35) was significantly depolarized (F_{(1, 7)}=3.31, p=0.0019, repeated measures) from the mean Vm measured in artificial cerebrospinal fluid only (ACSF, −61.59 ± 1.98). The mean Vm measured in the presence of CNO (−80.6 ± 2.69 mV) was significantly more hyperpolarized than the mean Vm measured in ACSF only (F_{(1, 7)}=4.59, p=0.0008, repeated measures). Therefore, this approach conferred highly targeted expressions of hM4Di in OXTR neurons in the AVPV.

Fig. 5.

A. A schematic illustration of a mouse brain in a coronal section injected with pAAV-hSyn-DIO-hM4D(Gi)-mCherry) or pAAV-hSyn-DIO-mCherry into the AVPV. B. An illustration of the overall experimental timeline. C. A low power fluorescent photomicrograph of a coronal section showing the DREADD injection site (a). The dashed lines represent the injection track. Higher power views of the AVPV b and c were taken from the same brain but at the different planes. A large group of mCherry expressing neurons are observed only on the left side of the third ventricle (3V), whereas only a small group of mCherry expressing neurons are seen at the ventral part (arrowhead) at this plane (b). In the 60 μm posterior plane (c), a large group of mCherry expressing neurons is observed on the right side of the 3V, whereas no mCherry expressing neurons was observed on the left side. Scale bars: 100 μm. D. A photomicrograph showing a patched mCherry expressing neuron in the AVPV during the whole-cell patch-clamp recording. Scale bar = 25 μm. E. A representative recording from a mCherry expressing neurons showing the effect of oxytocin and CNO on the membrane potential. F. The membrane potentials of mCherry expressing cells (n=4) measured in artificial cerebrospinal fluid (ACSF) and in the presence of 100 nM oxytocin (OT) and CNO in ACSF. The student’s t test was used to assess independently the effect of OT and CNO on the membrane potential. The box extends to the quartiles of the data points, and the whiskers extend to the farthest data points. The mean and median are represented by a filled circle and a line in the box, respectively.

Activity of OXTR neurons in AVPV is necessary for pup retrieval behavior

Dams received an intraperitoneal (IP) injection of saline as a vehicle control 30 min prior to the pup retrieval test. During the test, three pups were introduced into a cage, and pup retrieval behavior was recorded for 10 min. Following the pup retrieval test, behaviors of dams in their home cage (the home cage test) were recorded for an additional hour. After the first home cage test, the same dams received an IP injection of CNO and underwent a second set of the same behavior tests. The mean ± SE of pup retrieval test results from each group before (saline) and after CNO injections are summarized in Table 2.

Table 2.

Latency to retrieve pups from Control, Unilateral hM4Di, and bilateral hM4Di groups before (saline) and after CNO injection.

Group	Treatment	1st pup	2nd pup	3rd pup
Control	Saline	12.8 ± 2.4 s	30.8 ± 6.9 s	50.0 ± 13.0 s
	CNO	15.4 ± 5.3 s	21.4 ± 5.2 s	32.2 ± 6.1 s
Uni. hM4Di	Saline	55.8 ± 11.6 s	72.0 ± 15.6 s	129.4 ± 29.6 s
	CNO	did not retrieve	did not retrieve	did not retrieve
Bi. hM4Di	Saline	61.0 ± 11.6 s	96.8 ± 10.7 s	139.0 ± 23.7 s
	CNO	did not retrieve	did not retrieve	did not retrieve

To examine the effects of AVPV-OXTR inhibition on pup retrieval, a repeated measure MANOVA with group as a fixed factor and pup retrieval as the dependent variable was used. The most striking result of chemogenetic inhibition of AVPV-OXTR neurons was found in the pup retrieval test. Activation of hM4Di by CNO injection completely blocked the pup retrieval behavior of the dams (F_(1,13)= 35.94; p<0.0001, repeated measures). No postpartum dam from hM4Di groups, both bilateral (n=6, F_(1,5)= 75.72; p<0.0001, repeated measures) and unilateral (n=5, F_(1,4)= 63.16; p<0.0001, repeated measures) hM4Di, retrieved a pup after CNO injection, whereas all dams regardless of groups that received the saline injection retrieved all three pups within the duration of the pup retrieval test. Compared to hM4Di groups, CNO injection did not obstruct CTRL dams (n=5) from retrieving pups or affect the latency to retrieve 1^st pup (F_(1,4)=0.15; p=0.72, repeated measures), 2^nd pup (F_(1,4)=0.81, p=0.42, repeated measures) or 3^rd pup (F_(1,4)=0.96, p=0.38, repeated measures) (Fig. 6).

Fig. 6.

The mean latency to retrieve pups of CTRL (n=5), unilateral hM4Gi (n=5), and bilateral hM4Gi (n=6) groups prior to and after the CNO injection. DNR: Did Not Retrieve. The maximum time (600 sec) was given to latency to retrieve 3^rd pup for dams that did not retrieve pups. The box extends to the quartiles of the data points, and the whiskers extend to the farthest data points. The mean and median are represented by a filled circle and a line in the box, respectively. The post-hoc Tukey-Kramer HSD was used compared differences between each group.

The difference in the pup retrieval behavior among groups before CNO injection was also assessed (Fig. 6), because such difference may indicate the effect of hM4Di without activation by CNO. Before injection of CNO, there were already significant differences in the mean latency to retrieve pups among groups (1^st pup: F_(2,13)=6.93, p=0.009; 2^nd pup: F_(2,13)=8.38, p=0.005; 3^rd pup: F_(2,13)=4.28, p=0.04, ANOVA). The bilateral hM4Di group exhibited significantly longer latency to retrieve the 1^st pup (p=0.011), 2^nd pup (p=0.004), and 3^rd pup (p=0.043) than that from the CTRL group. The unilateral hM4Di group also exhibited significantly longer latency to retrieve the 1^st pup than that from the CTRL group (p=0.028); however, the significant differences were not observed in latency to retrieve the 2^nd and 3^rd pups. There was no difference in the mean latency to retrieve pups between unilateral and bilateral hM4Di groups.

Home cage test

The mean ± SEM of home cage test results from each group before and after CNO injections are summarized in Table 3. Before CNO injection, all dams regardless of the group spent some time inside of the nest. After injection of CNO, however, three out of five dams in the unilateral hM4Di group and one out of six dams in the bilateral hM4Di group failed to spend any time in the nest, whereas most of the others spent the time almost entirely (>50 min) in the nest. Because of this polarized response to CNO injection among the hM4Di groups, repeated measures analysis did not find significant change in the mean duration of time spent inside nest before and after CNO injection (unilateral: F_(4,1)=0.09, p=0.83; bilateral: F_(5,1)=0.05, p=0.57). In contrast to hM4Di groups, such a drastic change in response to CNO injection was not observed in the duration of time spent in the nest in the CTRL group (F_(1,4)=0.02, p=0.90, repeated measures).

Table 3.

Home cage test results in minutes from Control, Unilateral hM4Di, and bilateral hM4Di groups before (saline) and after CNO injection.

Group	Treatment	Inside nest	Crouching over pups	Feeding pups	Grooming pups	Building nest	Eating or drinking	Self-grooming
Control	Saline	43.0±8.3	37.1 ± 7.9	37.0 ± 7.9	13.0 ± 3.6	7.3 ± 2.6	3.6 ± 3.1	3.1 ± 1.4
	CNO	41.5±8.1	29.6 ± 7.7	29.6 ± 7.7	12.9 ± 5.0	10.9 ± 3.7	7.0 ± 3.7	2.6 ± 1.2
Uni. hM4Di	Saline	23.6 ± 1.9	7.5 ± 2.9	7.8 ± 3.1	10.7 ± 2.3	3.1 ± 1.0	18.0 ± 3.6	4.0 ± 1.8
	CNO	32.1 ± 13.1	21.5 ± 9.8	27.2 ± 11.3	6.7 ± 5.5	0.3 ± 0.3	0.02 ± 0.02	0.7 ± 0.3
Bi. hM4Di	Saline	25.3 ± 3.9	17.1 ± 5.2	11.1 ± 5.4	14.0 ± 7.1	2.5 ± 1.0	15.5 ± 3.1	7.1 ± 2.9
	CNO	28.2 ± 10.9	23.6 ± 7.7	21.8 ± 8.5	6.7 ± 5.5	0.6 ± 0.4	0.2 ± 0.2	7.5 ± 3.1

In the nest, dams regardless of the group spent most of their time (70–90%) crouching over pups before CNO injection (Fig. 7 B). After injection of CNO, two out of five dams from the unilateral hM4Di and one from bilateral hM4Di groups completely stopped crouching over and feeding the pups, while other dams spent much longer (6–50 folds increase) time in these behaviors. Due to the polarized response, repeated measures analysis did not find significant change in time spent crouching over pups (unilateral: F_(1,4)=1.26, p=0.33; bilateral: F_(1,5)=0.41, p=0.55) and feeding pups (unilateral: F_(1,4)=1.86, p=0.24; bilateral: F_(1,5)=1.42, p=0.29) in response to the CNO injection in the hM4Di groups. The time spent crouching over the pups also did not differ before and after CNO injection in the CTRL group (F_(1,4)=0.61, p=0.48, repeated measures). Because dams were feeding pups most of the time when they were crouching over them, similar results were found in time spent feeding pups (Fig. 7 C).

Fig. 7.

Quantification of behaviors of CTRL (n=5), unilateral hM4Gi (n=5), and bilateral hM4Gi (n=6) groups prior to and after the CNO injection in their home cage. Time spent (in min) inside nest (A), crouching over pups (B), feeding pups (C), grooming pups (D), building nest (E), eating/drinking (F), self-grooming (G). The box extends to the quartiles of the data points, and the whiskers extend to the farthest data points. The mean and median are represented by a filled circle and a line in the box, respectively.

In the nest, when dams were not crouching over and feeding pups, they were grooming pups (5–15 min, Fig. 7 D) or exhibiting nest building behavior (2–7 min, Fig. 7 E). Prior to injection of CNO, the mean duration of time spent grooming pups among CTRL, unilateral hM4Di, and bilateral hM4Di groups were comparable (F_(2,13)=0.18, p=0.84, ANOVA). Although CNO injection did not cause change in the duration of time spent grooming pups in the CTRL group (F_(1,4)=0.001, p=0.97, repeated measures), it did cause significant reduction in pup grooming behavior in the bilateral hM4Di group (F_(1,5)=10.02, p=0.03, repeated measures). For reference, two out of 6 dams in the bilateral hM4Di group did not groom their pups at all. In contrast to the bilateral hM4Di group, polarized responses to the injection were observed among dams in the unilateral hM4Di group. Two out of five dams spent zero to short (<30 s) time grooming pups after CNO injection, whereas two others spent longer time after injection compared to that before injection. The mean duration of time spent grooming pups among the unilateral hM4Di group after CNO injection was not significantly different compared to before CNO injection (F_(1,4)=0.31, p=0.61, repeated measures).

Before injection of CNO, all dams showed nest building behavior regardless of groups, and there was no significant difference in the duration of time spent building the nest among groups (F_(2,15)=2.49, p=0.12, ANOVA). After injection of CNO, three of six unilateral hM4Di dams and three of six bilateral hM4Di dams did not exhibit nest building behavior, whereas all CTRL dams exhibited this behavior. Results from repeated measure analysis to assess the effect of CNO injection on nest building behavior for each groups are: unilateral hM4Di (F_(1,4)=12.06, p=0.03); bilateral hM4Di (F_(1,5)=5.81, p=0.06); and CTRL (F_(1,4)=0.74, p=0.44).

The duration of time spent eating/drinking among dams varied from less than 1 min to 25 min regardless of groups. After injection of CNO, however, most dams (4/5 in unilateral and 4/6 bilateral hM4Di group) completely stopped eating and drinking. Among the three dams (1/5 in unilateral and 2/6 in bilateral hM4Di group) that spent time eating/drinking, only one ate (77 sec), and the remaining two just sipped water (1 and 6 sec). A repeated measures analysis found both unilateral and bilateral hM4Di groups spent significantly less time eating/drinking after CNO injection compared to before (F_(1,5)=23.26, p=0.005 and F_(1,4)=24.87, p=0.008, respectively). There was no difference in time spent eating/drinking before CNO injection compared to after in the CTRL group (F_(1,3)=0.35, p=0.6, repeated measures).

All dams in both unilateral and bilateral hM4Di groups showed several minutes of self-grooming behavior before CNO injection. However, after CNO injection, most of the dams (4/5) in the unilateral hM4Di group did not exhibit self-grooming behavior at all, whereas one dam exhibited self-grooming behavior for less than 2 min. The mean duration of time spent self-grooming in the unilateral hM4Di group was significantly shorter (F_(1,4)=8.25, p=0.05, repeated measures) after CNO injection compared to before. Three out of six dams in the bilateral hM4Di group also did not exhibit self-grooming behavior. However, because two dams from the bilateral hM4Di group spent considerably longer (11 and 18 min) time in self-grooming, the duration of time spent self-grooming after CNO injection (F_(1,5)=1.1, p=0.84, repeated measures) did not differ compared to before injection. CNO injection did not affect the duration of time spent self-grooming the dams in the CTRL group (F_(1,4)=1.1, p=0.35, repeated measures).

The differences in behaviors measured in home cage test among groups before CNO injection were also assessed. There were significant differences in the mean duration of time spent in nest (F_(2,13)=3.98, p=0.05), crouching over pups (F_(2,13)=6.62, p=0.01), feeding pups (F_(2,13)=7.31, p=0.008), and eating/drinking (F_(2,13)=5.34, p=0.02) among groups prior to CNO injection. There were no significant differences in the mean duration of time spent in grooming pups and self-grooming among groups (F_(2,13)=0.18, p=0.84 and F_(2,13)=0.2, p=0.81, respectively). Subsequent each paired t-test revealed that both unilateral and bilateral hM4Di groups already exhibited significantly shorter duration of time spent inside the nest, crouching over pups, feeding pups, and eating/drinking compared to the CTRL group. The statistical results of comparison for each pair are summarized in Table 4.

Table 4.

Statistical results from comparisons for each pair.

Comparison	Inside nest	Crouching over pups	Feeding pups	Building nest	Eating or drinking
CTRL vs Uni. hM4Di	t(8)=2.27, p=0.03*	t(8)=3.5, p=0.004*	t(8)=3.44, p=0.004*	t(8)=2.82, p=0.004*	t(8)=3.03, p=0.01*
CTRL vs Bi. hM4Di	t(9)=1.92, p=0.03*	t(9)=2.10, p=0.03*	(9)=2.71, p=0.01*	(8)=2.73, p=0.003*	t(9)=2.73, p=0.02*
Uni. hM4Di vs Bi. hM4Di	t(9)=0.38, p=0.82	t(9)=1.61, p=0.25	t(9)=0.52, p=0.69	t(9)=0.44, p=0.91	t(9)=0.53, p=0.6

^*p<0.05

Discussion

Our study revealed that the activity of OXTR neurons in the AVPV is necessary for maternal behavior in postpartum mice dams. The heightened interaction with pups typically observed in postpartum dams may be instigated, at least partly, by increased AVPV-OXTR neurons. Because OXTR is not expressed in the AVPV of males, the findings support the hypothesis that AVPV-OXTR neurons constitute an important node in female specific neural circuitry that regulates maternal behavior. Overall, the data provide significant insights into the regulation of maternal behavior in mice.

A reproductive state dependent expression of OXTR in the AVPV

The number of OXTR-Venus neurons in the AVPV from dams on postpartum day 1 was more than two-fold greater compared to their virgin siblings. Since the expression of OXTR in the AVPV is estrogen dependent (9), the increased number of AVPV-OXTR-Venus neurons was likely caused by an increased level of estrogen during the time of parturition and early postpartum (30). Importantly, the expression of OXTR-Venus was intact in most other areas including the MPOA outside of the AVPV (9) suggesting the estrogen-dependent expression of OXTR is brain area or cell type specific in a reproductive state-dependent manner. Tissue and reproductive state-dependent expression of OXTR was recognized in the uterus and mammary glands where transcription is regulated by gonadal steroids and profoundly elevated during parturition. Transcription of the mouse Oxtr in the mammary glands and uterus is epigenetically regulated by DNA methylation of specific sites in its promoter including an estrogen response element in a tissue-specific manner (50). In the mouse brain, the promoter region of the Oxtr is methylated differentially among distinct regions causing different levels of receptor expression (51). Therefore, DNA methylation of the promoter may play a distinct role in the Oxtr transcription in the AVPV under different reproductive conditions.

Interaction of OXTR and TH⁺ neurons in the AVPV

OXTR-Venus neurons and their processes were observed intermingling with TH⁺ neurons implying that OXTR neurons and TH⁺ neurons are in close or direct contact (i.e., synapses). In addition, approximately 30% of the OXTR-Venus neurons were TH⁺. The TH⁺ neurons send monosynaptic inputs to oxytocin neurons in the PVN, and optogenetic stimulation of the AVPV-TH⁺ neurons increased circulating oxytocin and enhanced maternal care (15). Thus, oxytocin may influence maternal behavior by interacting with the TH⁺ neurons in the AVPV. A greater number of AVPV-TH⁺ neurons during postpartum period imply that the interaction of oxytocin and TH⁺ neurons is enhanced when the physiological demand for maternal behavior is high.

Inhibition of AVPV-OXTR neurons blocked pup-retrieval behavior.

Pup retrieval in rodents is a major characteristic of maternal behavior shown to initially require oxytocin signaling for long-term neural plasticity (40, 52–56) and is enhanced during the postpartum period (57). A striking finding from the present study is that the inhibition of AVPV-OXTR neurons caused a complete blockade of pup retrieval. More surprisingly, unilateral inhibition was strong enough to entirely block the behavior demonstrating that AVPV-OXTR neurons are necessary for pup retrieval to occur. This starkly contrasts the inhibition of TH⁺ neurons in the AVPV where specific ablation of TH⁺ neurons by bilateral injection of neurotoxin significantly impaired, but did not block, pup retrieval behavior (15).

Postpartum dams display pup retrieval behavior, nest building, nursing behavior, and pup grooming (58) with an increased motivation to interact with pups (57). Pup retrieval and nest building behaviors are more indicative of maternal motivation because these behaviors are a voluntary act likely driven by attraction toward pup-related stimuli (59). In contrast, nursing behavior is indicative of a more reflexive maternal response because the behavior can be elicited by proximal pup stimuli that are regulated largely by brainstem mechanisms (59). Compared to pup retrieval and nest building behaviors, changes in the nursing behaviors (crouching over pups, feeding pups, and grooming pups) in response to inhibition of AVPV-OXTR neurons varied considerably. Approximately half of the hM4Di dams that received CNO did not spend any time with pups, indicating that they were not motivated to interact with them. The other half of the hM4Di dams spent time almost entirely in the nest and showed unusually prolonged nursing behaviors. This may be reflexive nursing postures elicited by proximal pup stimuli when dams were in close contact with pups with no interest in interacting with them. These findings indicate AVPV-OXTR neurons are involved in the induction of the maternal motivation, but not in the reflexive nature of the nursing behaviors.

Neurons in the MPOA send projections to the ventral tegmental area (VTA) (60, 61), a central structure in the mesolimbic dopamine system. Because the system is recognized for its central role in motivated behaviors, the presence of these projections suggests that the MPOA regulates maternal motivation by interacting with the mesolimbic dopamine system (reviewed in (62, 63)). The present tracing study found that AVPV-OXTR neurons project to the VTA. Interestingly, ERα expressing neurons in the MPOA were found to send projections to the VTA (39). More recently, a study found that chemogenetic inhibition of ERα neurons in the MPOA impaired pup approach and retrieval behavior, whereas optogenetic activation of these neurons stimulated pup retrieval (64). As all OXTR neurons in the AVPV, located within the MPOA, express ERα (9) it further supports our finding that AVPV-OXTR neurons directly project to the VTA. These findings suggest that estrogen and oxytocin act on AVPV-OXTR neurons to stimulate the mesolimbic dopamine system to regulate maternal motivation.

The present study also found a projection from AVPV-OXTR neurons to the periaqueductal gray (PAG). A portion of MPOA neurons that were activated (increase in Fos expression) following interaction with pups was determined to terminate in the PAG (61). This finding indicates that the MPOA-PAG pathway is also involved in the regulation of maternal behavior. According to the approach-avoidance model of the onset of maternal behavior (65), maternal behavior occurs because hormonal events at the end of pregnancy act to increase attraction toward pup-related stimuli while simultaneously decreasing avoidance responses to infant-related stimuli. In an attempt to match the actual neural circuit regulating maternal behavior to the approach-avoidance model, Numan (66) suggested that the MPOA to PAG projections depress avoidance responses toward pup-related stimuli, since the PAG is involved in defensive behavior, fearfulness, and avoidance/escape responses (58, 67–69). Because maternal behavior in lactating rats stimulated c-Fos in GABA neurons in the MPOA (70), it is possible that a portion of the AVPV-OXTR neurons that project to the PAG are inhibitory GABAergic and inhibit neurons involved in initiating pup avoidance responses. Nevertheless, the projection of AVPV OXTR neurons to the PAG suggests that oxytocin acts at the end of pregnancy to inhibit avoidance and rejection of pup-related stimuli – behaviors common in virgin females previously unexposed to pups (28, 71–74).

Inhibition of AVPV-OXTR neurons suppressed eating and drinking

The inactivation of AVPV-OXTR neurons significantly decreased the time spent eating and drinking for post-partum dams. This finding raised a concern that the CNO treatment caused disturbances in some physiology such as nausea and cardiac/autonomic alterations, which affect maternal behavior in hM4Di groups. However, these mice did not show any abnormal posture such as hunched posture (general sign of pain or disease) or behavior including back-arching, belly-pressing, twitching, and staggering. Therefore, it is less likely that the treatment caused pathophysiologic condition that affect eating, drinking, and maternal behavior.

The behavioral test used was not appropriately designed to assess food intake behavior in detail as such an investigation was beyond the scope of the study. Whereas the changes in eating and drinking time may be simply the time allocation change due to the changes in time spent for maternal behavior, oxytocin may be involved in a natural state of physiological hyperphagia that is known to occur to increase weight and adiposity during pregnancy and lactation (75, 76). Intracerebroventricular injection of oxytocin reduced food intake in both male and female rats (77); but, oxytocin increased, rather than decreased, food intake during pregnancy (78). It is highly speculative at this point, but AVPV-OXTR neurons may be involved in the mechanism underlying the change in food intake during pregnancy and lactation.

Use of DREADD

The significant differences in maternal behavior between CTRL and hM4Di groups were already existed before CNO injection. This indicates that a certain number of hM4Di channels were active in the absence of CNO. This is a matter of concern for anyone who uses the DREADD technique. However, all hM4Di dams retrieved all three pups within the duration of the test period indicating that, in the absence of CNO, the activity of hM4Di was subtle enough not to block the behavior. Therefore, the method used was reasonable to assess the effect of AVPV-OXTR neurons on the maternal behaviors.

Embedded Video 1.

Download video file ^{(57.9MB, mov)}

A video of a bilateral hM4Di postpartum dam during pup retrieval test. The dam received a vehicle control (saline) injection 30 min prior to the test. This dam successfully retrieved all 3 pups within 2 min.

Embedded Video 2.

Download video file ^{(325.7MB, mov)}

A video of the same bilateral hM4Di postpartum dam as in Embedded Video 1, but the dam received a CNO injection this time 30 min prior to the pup retrieval test. This dam did not approach pups but spent the entire 10 min all by herself in the plastic dome.

Significance Statement.

Oxytocin is known to modulate many aspects of social behaviors including social recognition, pair bonding, and maternal behavior. Oxytocin influences behaviors by binding to oxytocin receptors (OXTR) in various parts of the brain. Our previous study identified a group of neurons that express OXTR only in the brain of female mice but not male mice. The purpose of the present study was to understand the function of this sexually dimorphic OXTR neurons. The study revealed that the activity of the sexually dimorphic OXTR neurons is necessary for maternal behavior in female mice. Understanding the function of sexually dimorphic OXTR neurons will provide useful insight into sex-specific pharmacological interventions that treat the sex typical psychiatric disorders such as postpartum depression.