Estrogen withdrawal increases postpartum anxiety via oxytocin plasticity in the paraventricular hypothalamus and dorsal raphe nucleus

Abstract

Background:

Estrogen increases dramatically during pregnancy, but quickly drops below prepregnancy levels at birth and remains suppressed during the postpartum period. Clinical and rodent work suggests that this postpartum drop in estrogen results in an “estrogen withdrawal” state that is related to changes in affect, mood, and behavior. How estrogen withdrawal impacts oxytocin neurocircuitry has not been examined.

Methods:

We used a hormone-simulated pseudopregnancy followed by estrogen withdrawal in Syrian hamsters, a first for this species. Ovariectomized females were given daily injections to approximate hormone levels during gestation and then withdrawn from estrogen to simulate postpartum estrogen withdrawal. Subjects were tested for behavioral assays of anxiety and anhedonia during estrogen withdrawal. Following sacrifice, neuroplasticity in oxytocin-producing neurons in the paraventricular nucleus of the hypothalamus (PVH) and its efferent targets was measured.

Results:

Estrogen-withdrawn females had increased anxiety-like behaviors in the elevated plus and open field, but did not differ from controls in sucrose preference. Furthermore, estrogen-withdrawn females had more oxytocin-immunoreactive cells and oxytocin mRNA in the PVH, as well as an increase in oxytocin receptor density in the dorsal raphe nucleus (DRN). Finally, blocking oxytocin receptors in the DRN during estrogen withdrawal prevented the high-anxiety behavioral phenotype in estrogen-withdrawn females.

Conclusions:

Estrogen withdrawal induces oxytocin neuroplasticity in the PVH and DRN to increase anxiety-like behavior during the postpartum period. More broadly, these experiments suggest Syrian hamsters as a novel organism in which to model the effects of postpartum estrogen withdrawal on the brain and anxiety-like behavior.

Introduction

During pregnancy and the postpartum period, many hormones fluctuate and likely contribute uniquely to peripartum changes in physiology, behavior, and mood/affect. Among these hormones, peripartum fluctuations in estrogen levels are particularly dramatic: estrogen, synthesized primarily in the placenta, rises 100-1000-fold during the third trimester (1). Following birth and the expulsion of the placenta, however, estrogen levels quickly drop to below pre-partum levels and remain suppressed until ovulation resumes weeks to months later (2). It has been hypothesized that the rapid and dramatic drop in estrogen during the postpartum period results in an “estrogen withdrawal state” that contributes to the etiology of postpartum psychological disorders (3). In support of this idea, several studies have found that symptoms of postpartum depression can be attenuated by estradiol treatment in women (4-6).

How estrogen withdrawal contributes to the etiology of peripartum psychological disorders is unclear. To better understand the impact of this dramatic endocrine change, some researchers have turned to rodent models. In particular, several groups have employed a hormone-simulated pseudopregnancy (HSP) model to directly test the impact of estrogen withdrawal on the brain and behavior. Initially developed in female Long-Evans rats (7), this model has successfully been replicated (8) and extended to Sprague-Dawley rats (9-13), and ICR mice (14,15). Consistent with the clinical literature, estrogen withdrawal following HSP results in a behavioral phenotype that may reflect depressed mood (7-12) and/or heightened anxiety (11,14,15) in rodents. In many cases, these behavioral changes can be prevented by continued administration of estradiol during the postpartum period, also matching the data from human studies.

An extreme, systemic endocrine shift like postpartum estrogen withdrawal is likely to impact multiple, distributed brain regions that have distinct effects on behavior. Despite this, the majority of studies that have used HSP to examine how postpartum estrogen withdrawal impacts the brain have focused on the hippocampus (8,11,13,14). Only one study to date has examined the impact of estrogen withdrawal outside of the hippocampus (15) and none have examined the paraventricular nucleus of the hypothalamus (PVH). The PVH is a highly-conserved hypothalamic subnucleus that secretes neuropeptides either into peripheral circulation via the neurohypophysial system, or to neural targets via centrally-projecting neurons (16). In particular, oxytocin (OT), synthesized primarily in the PVH and supraoptic nucleus (SON), is considered a key modulator of physiology, behavior, and affective processes associated with pregnancy and the postpartum period (17). Oxytocin receptors (OTR) are located in several estrogen-sensitive brain nuclei and are functionally implicated in maternal behaviors, anxiety, depression, social cognition, and sexual behavior (18,19). Furthermore, estrogen increases both the secretion of OT and the expression of OTRs in the brain (20), suggesting the oxytocinergic system as a likely target for neuroplasticity underlying behavioral changes during postpartum estrogen withdrawal.

Using Syrian hamsters, a well-established rodent model of reproductive and affective behaviors, we investigated the role of estrogen withdrawal following HSP on behavioral assays of mood and affect, as well as oxytocin neuroplasticity in the PVH and its estrogen-sensitive efferent targets. Initial experiments demonstrated that estrogen withdrawal increased oxytocin receptor binding uniquely in the dorsal raphe nucleus (DRN), an estrogen-sensitive efferent of the PVH that is known to regulate anxiety behavior. Subsequent experiments therefore focused on determining the role of oxytocin in the DRN in regulating anxiety behavior following estrogen withdrawal.

Methods and Materials

Subjects

For all experiments, adult female Syrian hamsters were purchased from Charles River Laboratories (Wilmington, Massachusetts) at approximately 55 days of age. The experimental groups and numbers of subjects for each experiment can be found in Table 1 and additional information can be found in the Supplementary Methods.

Table 1.

Experimental groups, numbers of animals, and dependent variables for each experiment.

Experiment	Groups (n)	Dependent Variables
1A: Effect of Estrogen Withdrawal on Anxiety-Like Behaviors	Withdrawn (n = 8) Sustained (n = 8) No Hormone (n = 8)	Elevated Plus Open Field Locomotor behavior
1B: Effect of Estrogen-Withdrawal on Anhedonia	Withdrawn (n = 8) Sustained (n = 8)	Sucrose Preference Sucrose Consumed
2A: Effect of Estrogen-Withdrawal on Oxytocin-Producing Cells	Withdrawn (n = 8) Sustained (n = 8) No Hormone (n = 8)	Number of OT-ir cells in PVH and SON and NeuN-ir cells in PVH
2B: Effect of Estrogen-Withdrawal on OT Gene Expression in PVH	Withdrawn (n = 6) Sustained (n = 6)	OT mRNA (qPCR) in PVH
3: Effect of Estrogen-Withdrawal on OTR density in PVH efferent targets	Withdrawn (n = 8) Sustained (n = 8) No Hormone (n = 8)	OTR density in MeA, BNST, NAc Core, NAc Shell, and Dorsal Raphe
4: Effect of Blocking OTRs in DRN during Estrogen-Withdrawal on Anxiety-Like Behaviors	Withdrawn VEH (n = 8) Sustained VEH (n = 8) Withdrawn OTA (n = 9) Sustained OTA (n = 7)	Elevated Plus Open Field

Estrogen Withdrawal Following Hormone-Simulated Pseudopregnancy (HSP)

Following recovery from ovariectomy (see Surgery), the HSP protocol (7) was initiated. In this model, ovariectomized females are given daily injections of estrogen and/or progesterone in doses that are known to reliably induce maternal behaviors in nulliparous ovariectomized rats (21-23). For these experiments, we have modified the protocol to fit the shorter gestational timeline (17d) of Syrian hamsters (24). All hamsters were administered daily subcutaneous injections of hormone dissolved in cottonseed oil vehicle (Sigma, St. Louis, MO, USA) over a 22-day period (see Supplementary Methods). In some experiments, both withdrawn and sustained females were compared to a “no hormone” condition, in which females received daily vehicle (oil) injections for all 22 days of the HSP.

Behavior Testing

Elevated Plus.

An elevated plus test was used to assess anxiety-like behavior (see Supplementary Methods) on days 20 or 21. This test was chosen because it has previously been used to measure anxiety-like behavior following HSP in rats and mice (11,14,15) and has been validated as a model for anxiety behavior testing in Syrian hamsters (25).

Open Field.

An open field test was used to measure anxiety-like behavior and spontaneous locomotor activity on days 20 or 21. This test was chosen because it has been used previously following HSP in rats and mice (11,14,15).

Sucrose Preference Test.

In a separate cohort of hamsters, a sucrose preference test was used to assess anhedonia, a core feature of depressed mood, on days 20 and 21 (see Supplementary Methods). This test was chosen because sucrose consumption and preference has previously been shown to decrease following HSP in rats (8).

Surgery

All surgery was conducted using aseptic surgical technique and under isoflurane anesthesia (2-3% vaporized in oxygen, Piramal, Bethlehem, PA, USA). Analgesic (Meloxicam, 2mg/kg, Portland, ME, USA) was administered subcutaneously immediately prior to the start of surgery and for three days postoperatively. For detailed descriptions of surgical procedures see Supplemental Methods.

Ovariectomy.

In order to remove the endogenous source of circulating estrogen and progesterone, females were bilaterally ovariectomized prior to the initiation of the hormone simulated pregnancy.

Dorsal Raphe Cannulation.

At least 1 week following ovariectomy, subjects in Experiment 4 were implanted with 26-guage stainless steel guide cannulae (Plastics One, Roanoke, VA, USA) targeting the dorsal raphe nucleus.

Drug Injections

Subjects were injected with either oxytocin receptor antagonist (OTA, [d(CH₂)₅¹, Tyr(Me)², Thr⁴, Orn⁸, des-Gly-NH₂⁹]OVT), a highly selective antagonist at OTRs (26), or sterile saline 10 min prior to testing in the elevated plus or open field. See Supplementary Methods for details.

Histology and Tissue Processing

Perfusion and Immunohistochemistry.

For experiments requiring immunolocalization of oxytocin or Neuronal Nuclei (NeuN) protein, subjects were sacrificed by intracardial perfusion. See Supplementary Methods for details.

Receptor Autoradiography.

For experiments measuring changes in oxytocin receptor density following a hormone simulated pregnancy, receptor autoradiography (see Supplementary Methods) was performed on eight regions of interest: the cingulate cortex (Cg1), prelimbic cortex (PrL), infralimbic cortex (IL), nucleus accumbens core (NAcC), nucleus accumbens shell (NAcSh), bed nucleus of the stria terminalis (BNST), medial amygdala (MeA), and dorsal raphe nucleus (DRN).

Cannulae Placement Verification.

At the conclusion of the DRN cannulation experiment, subjects were given an overdose of sodium pentobarbital (Beuthanasia-D Special), and an automated syringe pump was used to infuse 250 nL of India Ink (Dr. Ph. Martin’s, Salis International, Oceanside, CA, USA) to mark the location of the injection site. Subjects brains were removed and post-fixed in paraformaldehyde for 48 h. A cryostat (Leica CM1850, Leica Biosystems, Buffalo Grove, IL, USA) was used to collect 40-μm sections through the DRN. Sections were mounted onto slides and a brightfield microscope (Nikon) using SpotBasic software (Diagnostic Instruments) was used to acquire an image of the location of the ink injection.

qPCR.

On Day 22, subjects were euthanized via anesthetized rapid decapitation, brains were extracted, and tissue sections (1-mm thick) containing the PVH were taken using a cooled brain matrix. Bilateral tissue punches (1-mm diameter) of the PVH were rapidly collected from cooled tissue sections and stored in RNA-later (Quiagen, Valencia, CA, USA) for qPCR analysis (see Supplementary Methods).

Statistical Analysis

All statistics were run on Prism Software (Graphpad, San Diego, CA, USA). In Experiment 1, difference scores were calculated for the elevated plus (time spent in closed arms - time spent in open arms) and open field (time spent in periphery - time spent in center) and one-way ANOVAs were used to determine whether difference scores varied by experimental condition. Tukey’s HSD post hoc tests were used to examine significant omnibus tests for group differences. In Experiment 2, one-way ANOVAs or unpaired two-tailed t-tests were used to determine whether cell counts (OT or NeuN) varied by experimental conditions. Tukey’s HSD post hoc tests were used to examine significant omnibus tests for group differences. Likewise, unpaired two-tailed t tests were used to determine whether OT gene expression differed between experimental condition. In Experiment 3, one-way ANOVAs were used to determine whether optical densities (dpm/mg) varied by experimental conditions. Tukey’s HSD post hoc tests were used to examine significant omnibus tests for group differences. In Experiment 4, difference scores were calculated for the elevated plus and open field and two-way ANOVAS were used to probe for interactions and main effects of hormone condition (sustained vs. withdrawn) and drug condition (OTA vs. vehicle). t tests with Bonferroni corrections for multiple comparisons were used to explicate interactions.

Results

Experiment 1: Effect of estrogen withdrawal on behavioral measures of anxiety and depression

Elevated Plus.

There was a significant effect of hormone condition on the amount of time spent in the closed arms minus the amount of time spent in the open arms of the elevated plus (F(2,21) = 14.72, p = .0001). Post hoc tests revealed that sustained females did not differ from no hormone females with regard to where they spent their time in the elevated plus (p = 0.1184); animals in both of these groups spent roughly equivalent amounts of time in both arms. However, withdrawn females differed from both no hormone females (p = 0.0091) and sustained females (p < 0.0001); animals in the withdrawn group spent more time on average in the closed arms than the open arms, indicating a higher anxiety phenotype (Figure 1A).

Figure 1: Effect of Estrogen Withdrawal on Anxiety-Like Behaviors.

A) Withdrawn females showed a high-anxiety behavioral phenotype in the elevated plus compared to sustained females or no hormone controls. B) Withdrawn females showed a high-anxiety behavioral phenotype in the open field compared to sustained females, but did not differ from no hormone controls. These differences in anxiety-like behavior are not mediated by a more general locomotor deficit, as there is no effect of hormone condition on distance traveled (C and E) or average velocity (D and F) in either test. Data presented as mean ± SEM, * p < 0.05.

Open Field.

There was a significant effect of hormone condition on the amount of time spent in the periphery minus the amount of time spent in the center of the open field (F(2,21) = 5.457, p = 0.0123). Post hoc tests revealed that no hormone females did not differ from sustained females with regard to where they spent their time in the open field (p = 0.4781). Likewise, no hormone females did not differ from withdrawn females with regard to where they spent their time in the open field (p = 0.1176). However, sustained females differed from withdrawn females (p = 0.0100); animals in the withdrawn groups spent more time on average in the periphery than in the center, indicating a higher anxiety phenotype (Figure 1B).

Locomotor Behavior.

There was no effect of hormone condition on distance traveled or average velocity in either the elevated plus or open field (see Supplementary Results).

Sucrose Preference.

There was no effect of hormone condition on sucrose preference (see Supplementary Results and Supplementary Figure 1).

Experiment 2: Effects of estrogen withdrawal on oxytocin neurons in the PVH

The number of OT-immunoreactive (−ir) neurons in the PVH (Figure 2A) and supraoptic nucleus (SON, Figure 2B) was counted and compared between groups. There was a significant effect of hormone condition on the number of OT-ir cells in the PVH (F(2,21) = 22.67, p < 0.0001). Post hoc tests revealed that while the no hormone and sustained groups did not differ from each other (p = 0.7191), the withdrawn group had significantly more OT-ir cells than both the no hormone group (p < 0.0001) and the sustained group (p < 0.0001) (Figure 2C). There was no effect of hormone condition on the number of OT-ir cells in the SON, F(2,21) = 0.6547, p = 0.5299) (Figure 2D).

Figure 2: Effect of Estrogen-Withdrawal on Oxytocin-Producing Cells.

A) Representative OT-ir staining and counting domains in the PVH. B) Representative OT-ir staining and counting domains in the SON. C) In the PVH, the total number of OT-ir cells was significantly higher in withdrawn females than in sustained or no hormone controls. D) In the SON, the total number of OT-ir cells did not differ between hormone conditions. E) The total number of NeuN-ir cells in the PVH did not differ across hormone conditions. F) There was an approximately three-fold increase in OT gene expression in withdrawn females compared to estrogen sustained females. Scale bars = 200 μm, data presented as mean ± SEM, * p < 0.05.

There was no effect of hormone condition on the total number of NeuN-ir neurons in the PVH (F(2,21) = 0.2457, p = 0.7843), suggesting that the increase in OT-ir neurons is unlikely to be the result of neurogenesis (Figure 2E).

qPCR was used to assess OT mRNA in the PVH of a separate cohort of sustained and withdrawn females. There was a significant increase in OT mRNA in withdrawn animals compared to sustained animals (t(10) = 7.318, p < 0.001), suggesting that the increase in OT-ir cells is unlikely to reflect a decrease in OT release (Figure 2F).

Experiment 3: Effects of estrogen withdrawal on oxytocin receptors in PVH efferent targets

There was no effect of hormone condition on OTR density in Cg1, PrL, IL, NAcC, NAcSh, or MeA (Figure 3A-G, and supplementary results). However, there was a significant effect of hormone condition on OTR density in the DRN (F(2,21) = 13.43, p = 0.0002). While the no hormone group did not differ from the sustained group (p = 0.4521), OTR density was significantly higher in the withdrawn group than the no hormone group (p = 0.0002) and the sustained group (p = 0.0033, Figure 3H).

Figure 3: Effect of Estrogen-Withdrawal on OTR density in estrogen-sensitive PVH efferents.

OTR autoradiographic binding did not differ across hormone conditions in A) the cingulate cortex (Cg1), B) the prelimbic cortex (PrL), C), the infralimbic cortex (IL), D) the NAc Core, E) the NAc shell, F) the bed nucleus of the stria terminalis (BNST), or G) or the medial amygdala (MeA). H) In the dorsal raphe nucleus (DRN), however, withdrawn females had significantly higher OTR autoradiographic binding than sustained or no hormone controls. Representative autoradiographs with analysis boxes over the region of interest are shown next to each data set. Data presented as mean ± SEM, * p < 0.05.

Experiment 4: Effects of blocking oxytocin receptors in the DRN on anxiety-like behaviors

Cannulae Placement.

One hamster did not recover from cannulation surgery and the cannulae on eight hamsters did not remain intact and/or patent throughout the HSP protocol. In the remaining 51 hamsters, histological verification of cannulae placement using post-mortem ink injections showed that drug injections targeted the transition between the rostral and caudal DRN (5.4 - 5.7 mm posterior to Bregma, Figure 4A, (27)) in 32 subjects. Further, the angled cannulation approach resulted in bias for injections to target the lateral portion of the DRN. Subjects with cannulae outside of the DRN were misplaced either in the lateral periaqueductal gray (LPAG; n = 5), the fourth ventricle (n = 6), the lateral dorsal tegmental nucleus (n = 5) or, more rarely, the medial longitudinal fasciculus (n = 3); these animals were considered misses and removed from the analyses.

Figure 4: Effect of Blocking OTRs in DRN during Estrogen-Withdrawal on Anxiety-Like Behaviors.

A) Post-mortem localization of ink injections was used to determine whether placements accurately targeted the DRN (“hit,” filled circles) or were misplaced (“miss,” open circles) and therefore eliminated from analyses. OTA injection into the DRN reverses the high-anxiety behavioral phenotype seen in withdrawn females in B) the elevated plus and C) the open field. Although there was a main effect of hormone condition on distance traveled in the elevated plus, differences in anxiety-like behavior are unlikely to be mediated by a more general locomotor deficit, as there was no difference in distance traveled (D and E) or average velocity (F and G) between sustained and withdrawn females who received vehicle (VEH) injections or OTA injections. Scale bar = 100 μm, data presented as mean ± SEM, * p < 0.05.

Elevated Plus.

There was a significant interaction between hormone condition (sustained vs. withdrawn) and drug condition (OTA vs. vehicle) (F(1,28) = 8.916, p = 0.0058) in the elevated plus (Figure 4B). Post hoc tests revealed that in females who received vehicle injections into the DRN, the sustained and withdrawn groups differed from each other (t(28) = 3.532, p = 0.0029): sustained animals who received vehicle spent more time in the open arms of the elevated plus, whereas withdrawn animals who received vehicle spent more time in the closed arms of the elevated plus. In females who received OTA injections into the DRN, the sustained and withdrawn groups did not differ from each other (t(28) = 0.7018, p = 0.9772); both groups spent more time in the open arms of the elevated plus.

Open Field.

There was a significant interaction between hormone condition and drug condition (F(1,28) = 44.73, p < 0.0001) in the open field (Figure 4C). Post hoc tests revealed that in females who received vehicle injection into the DRN, the sustained and withdrawn groups differed from each other (t(28) = 6.701, p < 0.0001): sustained animals who received vehicle spent more time in the center of the open field, whereas withdrawn animals who received vehicle spent more time in the periphery of the open field. In females who received OTA injections into the DRN, the sustained and withdrawn groups also differed from each other (t(28) = 2.773, p = 0.0195): although both sustained and withdrawn females spent more time in the center of the open field, withdrawn females who received OTA spent more time in the center than did sustained females who received OTA.

Locomotor Behavior.

There was no significant interaction between hormone and drug condition on distance traveled in the elevated plus (F(1,23) = 0.01153, p = 0.9154, Figure 4D) or the open field (F(1,23) = 2.208, p = 0.1509, Figure 4E). Interestingly, there was a main effect of hormone condition on distance traveled in the elevated plus (F(1,23) = 8.002, p = 0.0095), but post hoc tests did not detect significant differences between sustained and withdrawn females who received vehicle injections (t(23)=1.887, p = 0.1437) or OTA injections (t(23)=2.119, p = 0.0902). There was no significant interaction between hormone and drug condition on average velocity in the elevated plus (F(1,23) = 0.008757, p = 0.9263, Figure 4F) or the open field (F(1,20) = 0.01354, p = 0.9085, Figure 4G).

Discussion

Here we demonstrate for the first time that estrogen withdrawal following HSP increases behavioral measures of anxiety in Syrian hamsters. Further, these increases in anxiety-like behavior are concurrent with increases in OT production in the PVH and increased OTR density in the DRN. Finally, blocking OTRs in the DRN during estrogen withdrawal prevents the increase in anxiety-like behavior in withdrawn females, but has minimal effect in females who continued to receive estrogen throughout the simulated postpartum period. Together, these results suggest a model in which postpartum estrogen withdrawal increases oxytocin transmission in the PVH and DRN, either independently or as part of a neural pathway, to increase anxiety-like behavior during the postpartum period.

These data add to a growing body of literature that uses HSP to model the impact of postpartum estrogen withdrawal, and expands the model to Syrian hamsters and to oxytocinergic neural circuitry. Although the HSP model does not provide a complete simulation of the hormonal changes that occur during the peripartum period, it allows researchers to isolate and manipulate postpartum estrogen withdrawal as a single independent variable, permitting causal inferences about resulting changes in the brain and behavior. Further, unlike in women, most common laboratory rodents experience postpartum and lactational estrus (28), during which estrogen levels rise sharply; in this way, estrogen withdrawal following HSP better models the human experience of postpartum estrogen withdrawal than would naturally-parturient rodents. It is noteworthy that the observed effects on behavior and neuroplasticity in this study required HSP treatment followed by estrogen withdrawal; this suggests that both the rise in estrogen, as would occur during pregnancy, and the subsequent withdrawal from estrogen, as would occur during the early postpartum period, are required. Future studies will need to address whether these effects are temporary or persist past the early postpartum period.

Differences in model organisms may influence behavioral outcomes of estrogen withdrawal. Similar to in the current study, estrogen withdrawal following HSP in Sprague-Dawley rats (11) and ICR mice (14,15) increases behavioral indicators of anxiety, suggesting that elevated anxiety following postpartum estrogen withdrawal is a robust behavioral phenomenon and may reflect an adaptive response that serves to increase vigilance during a vulnerable time. Although the elevated plus and open field are validated measures of anxiety-like behaviors in Syrian hamsters (25), our data demonstrate that control females do not always show a preference for the periphery of the open field or the closed arms of the elevated plus. It is perhaps even more notable then that estrogen withdrawal results in a very robust anxiety-like response in both behavioral assays, and that this behavioral response can be prevented by blocking oxytocin receptors in the DRN. In contrast, the current study did not find an effect of estrogen withdrawal on sucrose preference, despite this being reported in rats (8,12). This may reflect differences in the methods (e.g., 2% vs. 3% sucrose solution), or differences in the behavioral ecology of the two species. Although hamsters are capable of forming sucrose preferences (29), they consume less water per body weight than rats (30). As such, a drinking-based assay may not be the most robust test of anhedonia in this species. Future studies may wish to use other assays, such as intracranial self-stimulation (9), to test the effect of estrogen withdrawal on anhedonia in hamsters.

Although the HSP model has low ecological validity, the current findings are corroborated by data from postpartum rats (31,32), rabbits (33), and sheep (34), in which OT-immunoreactivity is increased in the PVH. Caldwell et al. also found increased OT-immunoreactivity in the SON of postpartum female rats, which differs from our findings using the HSP model; this suggests that while OT levels in the PVH and SON are both dynamic during the peripartum period, only PVH OT levels are likely to be mediated by estrogen withdrawal. Our finding that OTR autoradiographic binding is increased in the DRN of estrogen withdrawn females is also supported by recent data from naturally-parturient females showing that OTR autoradiographic binding and OT-ir fibers are increased in the DRN of postpartum rats (35). What’s more, viral-mediated knockdown of OTR mRNA in the DRN of postpartum rats decreases anxiety-like behavior in the elevated plus (35), mirroring our finding that pharmacological blockade of OTRs in the DRN decreased anxiety-like behavior in the elevated plus during estrogen withdrawal. When coupled with previous data, this suggests that 1) oxytocin in the DRN has anxiogenic effects on behavior and 2) these effects may depend on postpartum estrogen withdrawal. Our experiments did not address the effect of blocking OTRs in the DRN of no-hormone or cycling females; it is possible that oxytocin binding in the DRN may have effects outside of postpartum estrogen withdrawal, and/or that estrogen withdrawal potentiates a broader effect of OTA. Likewise, we focused our investigation of oxytocin receptor plasticity on PVH efferent targets that are known to regulate affective behavior and are sensitive to estrogen; it is likely, however, that other brain regions show changes in oxytocin receptor density in response to estrogen withdrawal and have differing effects on behaviors. Ultimately, these data join a growing body of literature challenging the dogma that OT is primarily anxiolytic (18,36) and suggests the need for more site-specific investigations of OT’s role in modulating affective behaviors.

The mechanism by which estrogen withdrawal increases OT-ir cells in the PVH is not known. The current experiments suggest that it is not an artifact of decreased OT release, as OT mRNA levels also increase substantially in withdrawn females, which would not be expected with decreased release. Further, it is unlikely to reflect an increase in the total number of cells, as we did not observe an increase in NeuN-immunoreactivity between hormone conditions. One intriguing possibility is that postpartum estrogen withdrawal is causing PVH cells to switch their neurochemical phenotype, either temporarily or permanently. This is not without precedent in the hypothalamus: in pregnant and lactating mice, tuberoinfundibular neurons in the arcuate nucleus begin synthesizing and releasing met-enkephalin rather than dopamine (37). The mechanism of this phenotypic switch depends on elevated prolactin levels during pregnancy suppressing dopamine synthesis in an opiate-dependent manner. It is possible that the massive fluctuations in estrogen during pregnancy and the postpartum period may similarly result in neurochemical plasticity at the level of individual neurons in order to facilitate behavioral changes during the postpartum period. Indeed, both OT (38) and OTRs (39) have steroid response elements, a putative target for how estrogen fluctuations during the peripartum period may impact transcriptional changes at OT- and OTR-containing cells.

We propose that estrogen withdrawal increases OT transmission between the PVH and DRN in order to increase anxiety during the postpartum period (Figure 5). The PVH sends moderate projections to the DRN, terminating primarily in the A10dc dopaminergic cell group (40). These dopaminergic DRN cells have been associated with arousal (41), and project to nuclei in the extended amygdala that are known to regulate anxiety (42,43). Recent work by Grieb et al (2019) demonstrates that approximately one third of these dopamine cells express OTRs, and the number and percentage of dopamine cells expressing OTRs are higher in postpartum rats than in diestrus virgin rats (35). In addition to dopaminergic cells, the raphe nuclei, including the DRN, are the primary source of forebrain-projecting serotonin neurons (44). Approximately half of raphe serotonin neurons contain OTRs, which modulate serotonin release to regulate anxiety (45). Given these neuroanatomical and functional substrates, it seems likely that estrogen withdrawal increases OT transmission between the PVH and the DRN to increase anxiety during the postpartum period, either by acting via direct synaptic contacts at dopaminergic neurons or via volume transmission (46) to serotonin neurons. Moving forward, experiments using chemogenetic inhibition of the PVH-DRN pathway are necessary to demonstrate its requirement for estrogen withdrawal-dependent increases in anxiety in the postpartum period. Ultimately, these data suggest a novel neural pathway in which peripartum estrogen fluctuations lead to neural plasticity that impacts postpartum affective behaviors.

Figure 5: Proposed Model.

Based on our data, we propose a model in which postpartum estrogen withdrawal increases oxytocin transmission between the PVH and DRN to increase anxiety-like behavior during the postpartum period. Specifically, we suggest that estrogen withdrawal leads to an increase in OT neurons in the PVH and a simultaneous increase in OTR density in the DRN, indicating a pathway-specific increase in OT neurotransmission. OT from the PVH may act via direct synaptic contact on dopamine (DA) neurons in the DRN to increases baseline arousal, or via volume transmission to serotonin (5-HT) neurons in the DRN to increase anxiety, during the early postpartum period.

KEY RESOURCES TABLE

Resource Type	Specific Reagent or Resource	Source or Reference	Identifiers	Additional Information
Add additional rows as needed for each resource type	Include species and sex when applicable.	Include name of manufacturer, company, repository, individual, or research lab. Include PMID or DOI for references; use “this paper” if new.	Include catalog numbers, stock numbers, database IDs or accession numbers, and/or RRIDs. RRIDs are highly encouraged; search for RRIDs at https://scicrunch.org/resources.	Include any additional information or notes if necessary.
Antibody	mouse monoclonal primary antibody against oxytocin	Millipore	Cat# MAB5296, RRID:AB_2157626
Antibody	rabbit polyclonal antibody against NeuN	Millipore	Cat# ABN78, RRID:AB_10807945
Chemical Compound or Drug	[d(CH2)51, Tyr(Me)2, Thr4, Orn8, des-Gly-NH29]OVT	doi: 10.1111/j.1365-2826.2012.02303.x	NA	provided by Maurice Manning to Elliott Albers
Chemical Compound or Drug	I125-labeled ornithine vasotocin analog Vasotocin, d(CH2)5[Tyr(Me)2,Thr4,Orn8,[125I]Tyr9-NH2	doi: https://doi.org/10.1111/jne.12882	NA	provided by Elliott Albers to Laura Been
Chemical Compound or Drug	Estradiol Benzoate	Sigma	Cat #P0130
Chemical Compound or Drug	Progesterone	Sigma	Cat #E8875
Sequence-Based Reagent	qPCR Primers for GAPDH	Eurofins Genomics	GenBank Accession ABD77188.1
Sequence-Based Reagent	qPCR Primers for OT	Eurofins Genomics	GenBank Accession HM357357.1
Commercial Assay Or Kit	Vectastain Elite ABC-Peroxidase Kit	Vector Laboratories	Cat# PK-6100, RRID:AB_2336819
Commercial Assay Or Kit	RNAeasy Mini Kit	Quiagen	Cat # 74104
Commercial Assay Or Kit	Transcriptor First-Strand cDNA Synthesis kit	Roche	Cat # 04379012001
Commercial Assay Or Kit	PowerUp SYBR Green Master Mix	Applied Biosystems	Cat # A25741