siRNA silencing of estrogen receptor-α expression specifically in medial preoptic area neurons abolishes maternal care in female mice

Abstract

The medial preoptic area has been shown to be intricately involved in many behaviors, including locomotion, sexual behavior, maternal care, and aggression. The gene encoding estrogen receptor-α (ERα) protein is expressed in preoptic area neurons, and a very dense immunoreactive field of ERα is found in the preoptic region. ERα knockout animals show deficits in maternal care and sexual behavior and fail to exhibit increases in these behaviors in response to systemic estradiol treatment. In the present study, we used viral-vector mediated RNA interference to silence ERα expression specifically in the preoptic area of female mice and measured a variety of behaviors, including social and sexual aggression, maternal care, and arousal activity. Suppression of ERα in the preoptic area almost completely abolished maternal care, significantly increasing the latency to pup retrieval and significantly reducing the time the moms spent nursing and licking the pups. Strikingly, maternal aggression toward a male intruder was not different between control and preoptic ERα-silenced mice, demonstrating the remarkably specific role of ERα in these neurons. Reduction of ERα expression in preoptic neurons significantly decreased sexual behavior in female mice and increased aggression toward both sexual partners and male intruders in a seminatural environment. Estrogen-dependent increases in arousal, measured by home cage activity, were not mediated by ERα expression in the preoptic neurons we targeted, as ERα-suppressed mice had increases similar to control mice. Thus, we have established that a specific gene in a specific group of neurons is required for a crucially important natural behavior.

Dating back to the 1970s, the preoptic region has been implicated in a variety of complex, natural, biologically crucial behaviors, ranging from sexual behavior to locomotion to maternal care. These roles for the preoptic region were first demonstrated using lesion studies; for instance, radiofrequency, electrolytic, and cytotoxic lesions of the medial preoptic area (MPA) were shown to disrupt many components of maternal care (1–6). Likewise, electrolytic lesions (7–9), chemical lesions (9–11), or knife-cuts disrupting preoptic afferents (12,13) all severely impair or abolish male sexual behavior, whereas electrical stimulation of this region enhances male copulatory behaviors (14). Additionally, c-Fos expression, a marker of neuronal activation, is significantly increased in the preoptic region following copulatory behaviors (15). With respect to locomotor activity, there is ample evidence demonstrating that electrical or chemical stimulation of preoptic neurons enhances locomotor activity in rats (16–19).

Given the importance of the preoptic region in mediating these complex biologically relevant hormone-dependent behaviors, the next step was to establish the cytochemical make-up of these populations of neurons. Before immunocytochemical techniques were routinely used, placement of hormone extracts was used to delineate the hormone and exact location of the action of these chemicals. The preoptic region was soon established as a prime site for estrogenic hormone action in the brain (20); for instance, the preoptic region is the most effective brain site for estradiol application to enhance running wheel activity (20, 21). With regard to sexual behaviors, application of estradiol to the preoptic region enhances copulatory behaviors (22), and preoptic implants of estrogen facilitate maternal care in female rats (23). Within the preoptic region, the most abundant receptor for estrogen is ERα, and binding of estradiol to ERα is necessary for facilitation of locomotion and maternal behaviors; in fact, animals with ERα knocked out exhibit significant impairments in maternal care and often display infanticidal behaviors (24). ERα knockout female mice also fail to exhibit increases in running wheel activity in response to systemic estradiol treatment (24), lack lordosis behavior, and have increased aggression toward males (24, 25).

Collectively, these findings suggest that ERα expression, specifically within the preoptic area (POA), may be involved in mediating these varied, interrelated, complex behavioral phenotypes. Specifically, we hypothesized that ERα expression in the MPA may mediate the increases in running wheel activity and generalized arousal in response to estradiol and the estradiol-dependent increases in sexual behavior, maternal care, and aggression.

The aim of this study was to use cutting-edge molecular biology techniques to specifically reduce ERα expression in the preoptic region and monitor the expression of complex, biologically relevant behaviors.

To test these hypotheses, we used adeno-associated virus (AAV) delivery of a short hairpin RNA (shRNA) directed against the mRNA for ERα in the MPA of female mice, and measured the behavioral effects that such silencing would have, on a variety of specific social behaviors, notably sexual and maternal behaviors, as well as indices of generalized central nervous system (CNS) arousal (26).

The main finding of this paper is that manipulation of a specific gene, encoding a ligand-activated transcription factor, in a specific group of neurons can drastically alter the expression of a complete, biologically crucial behavior. Furthermore, we demonstrate that these effects are remarkably specific, as even related behaviors are unaffected by these manipulations.

Results

Microinjection of shRNA ERα Drastically Reduced ERα Expression in MPA Neurons.

Only animals that had greater than an 80% decrease in immunoreactive ERα-positive neurons bilaterally in the MPA were included in the shRNA ERα group (Fig. 1, Right). The control group consisted of animals with correct bilateral EGFP expression in the MPA and intact ERα expression (Fig. 1, Left). All analyses (behavior and immunocytochemistry) were performed by individuals blind to the experimental treatment, and only animals that had adequate reduction of ERα expression were included in the experimental group.

Fig. 1.

Immunohistochemical evaluation of estrogen receptor-α (ERα) silencing in the medial preoptic region (MPA). Animals received bilateral MPA microinjections of either AAV shRNA LUC or AAV shRNA ERα. Only animals with greater than 80% reduction in ERα expression in the MPA were included in shRNA ERα group. (Left) AAV shRNA LUC MPA-injected mouse. (Right) AAV shRNA ERα MPA-injected mouse. DAB stained brain tissue: brown coloration (DAB), EGFP; Black nuclear staining (DAB with nickel), ERα. (Scale bar, 200 μm.)

Silencing of ERα in the MPA Greatly Decreased Maternal Care Behaviors.

Reduction of ERα expression in the MPA significantly decreased the amount of time spent licking and nursing pups, by 91% and 95%, respectively, compared with mice with intact ERα expression in the MPA [shRNA luciferase (LUC) group] (Fig. 2A). Average time spent licking the pups was decreased from 36.2 ± 10.4 to 2.9 ± 2.2 s (t test: P = 0.015); animals with unilateral or inefficient suppression of ERα in MPA neurons spent 29.2 ± 11.5 and 21.6 ± 13.3 s licking their pups, respectively. Average time spent nursing the pups was, likewise, decreased from 36.0 ± 8.6 to 1.6 ± 1.6 s (t test: P = 0.005), in mice with reduced ERα expression in MPA neurons (Fig. 2A). Similar to pup licking time, amount of time nursing the pups was not significantly different from controls in unilateral and inefficient mice (42.2 ± 16.8 and 33.2 ± 23.5 s, respectively). Loss of ERα in MPA increased latency to pup retrieval from 20 ± 2.5 s in shRNA LUC animals to 247 ± 53.5 s in shRNA ERα, and five of the seven mice lacking ERα in the MPA failed to retrieve the pups during the testing session (t test: P = 0.006) (Fig. 2B).

Fig. 2.

MPA suppression of ERα significantly reduced maternal behaviors but not maternal aggression. (A) Pup care. (B) Latency to pup retrieval. (C) Aggressive behaviors toward male intruder. Values represent mean ± SEM (SEM). *P < 0.05, **P < 0.01.

Maternal Aggression Was Unaffected by Silencing of ERα in MPA in Female Mice.

Surprisingly, maternal aggression toward a male intruder placed in the nursing mom’s cage was not different between the groups (t test: P = 0.624) (Fig. 2C), thus demonstrating a remarkable specificity of the effects mediated by ERα in the MPA that this study reveals. Litter sizes were not significantly different between the two groups (t test: P = 0.523).

Lowered ERα Expression in the MPA Reduced Sexual Behavior in Female Mice.

Bilateral suppression of ERα expression in the MPA significantly reduced the number of rejective episodes against a stud male from 16.1 ± 1.2 to 11.8 ± 0.9, compared with control shRNA LUC-injected animals, a decrease of 27% (t test: P = 0.009) (Fig. 3A). Decreases in rejective episodes in animals with reduced ERα expression in the MPA were predominantly due to decreases in kicking and boxing (active aggression), which were decreased by 36% and 52%, respectively, whereas rearing and fleeing (passive aggression) were not significantly different between shRNA LUC and sh-RNA ERα-injected mice (kicking: t test: P = 0.005; boxing: t test: P = 0.047) (Fig. 3B).

Fig. 3.

Suppression of ERα in the MPA significantly decreased rejective and receptive behaviors in female mice, specifically by decreasing the number of kicking and boxing episodes. (A) All sexual behaviors: rejective, proceptive and receptive. (B) Different components of rejective behavior: kicking, rearing, boxing, and fleeing. Values represent mean ± SEM. *P < 0.05, **P < 0.01.

Number of proceptive episodes, characterized by a still posture and dorsiflexion of the legs, was not different between the groups (Fig. 3A). Receptivity, or lordosis, was decreased in ERα-silenced animals by 57%, from an average of 1.6 ± 0.3 episodes per session to 0.7 ± 0.3 in shRNA LUC control animals (t test: P = 0.026) (Fig. 3A).

Sexual behavior differences between the female test groups were not due to differences in male behavior. Male mice were randomly assigned to each group, and there was no statistical difference between sexual behavior among the male groups.

Reduction of Preoptic ERα Decreased Aggression toward a Male Intruder in a Seminatural Environment (SNE).

Suppression of ERα expression in MPA decreased social investigation of other mice (both male intruders and female cohabitants) in the SNE by 71%, from 22.3 ± 4.7 to 6.6 ± 1.7 events/session (t test: P = 0.007) (Fig. 4A) and significantly decreased aggression by 77% from 39.4 ± 9.8 to 9.1 ± 3.4 events/session (t test: P = 0.012) (Fig. 4A). Similarly, animals with intact ERα expression in MPA exhibited a slightly higher number of dominant behaviors compared with shRNA ERα-injected mice (shRNA LUC 1.4 ± 0.5 vs. shRNA ERα 0.3 ± 0.2), but these trends did not reach statistical significance (P = 0.088) (Fig. 4A). Animals with lowered ERα expression in the MPA had a similar number of submissive behaviors as shRNA LUC-injected animals (t test: P = 0.560) (Fig. 4A).

Fig. 4.

Suppression of ERα in the MPA diminished social investigation and aggressive behaviors in a SNE and decreased aggression under food competition conditions. (A) Social interactions among females in the SNE and toward male intruders. (B) Social interactions in the SNE under restricted food availability. Values represent mean ± SEM. *P < 0.05, **P < 0.01.

In the food competition paradigm, loss of ERα in the MPA dramatically decreased aggressive behaviors toward other female mice by 72%, from an average of 4.2 ±0.9 events/session in control (shRNA LUC) mice vs. 1.2 ± 0.3 events/session in mice lacking ERα in the MPA (t test: P = 0.009) (Fig. 4B). First entry into the conical containing food and number of submissive behaviors were not significantly different between the groups of animals (first entry: t test: P = 0.998; submissive: t test: P = 0.554) (Fig. 4B).

Arousal Activity Following Estradiol Treatment Was not Mediated by ERα in MPA.

Estradiol benzoate-treated animals exhibited higher levels of behavioral arousal compared with oil-treated controls; this was seen in all three parameters of home cage activity: horizontal activity, total distance, and vertical activity (Fig. 5A). Like running wheel activity, these differences in activity were observed predominantly during the dark or active period. Overall, daily horizontal activity was increased after EB treatment (two-way ANOVA: EB treatment effect, P < 0.0001; LUC vs. ERα: P = 0.513) (Fig. 5A). Horizontal activity was increased by 36% from 85,250.4 ± 3,430.9 beam breaks/24 h in oil-treated animals to 115,665.4 ± 3,521.7 in EB-treated animals (t test: P < 0.001) (Fig. 5A). Likewise, total distance was increased by 41% from 14,001.0 ± 920.7 in control animals to 19,784.3 ± 1,391.0 cm/24 h in EB-treated animals (two-way ANOVA: EB treatment effect P = 0.028; LUC vs. ERα: P = 0.169; t test: P = 0.002) (Fig. 5B). Vertical activity followed the same trend, being increased by 65% after EB treatment (24 h totals – oil: 15,038.4 ± 1,148.8, EB: 24,717.9 ± 1,482.9) (two-way ANOVA: EB treatment effect: P = 0.0007; LUC vs. ERα: P = 0.605; t test: P < 0.001) (Fig. 5C). Suppression of ERα expression in the MPA did not elicit any changes in home cage activity of oil-treated animals during the dark or light periods, with the exception of horizontal activity during the light period, which was decreased by 34% (t test: P = 0.013) (Fig. 5). Reduction of ERα in the MPA did not change horizontal activity, total distance, or vertical activity during the dark or light periods in EB-treated animals (Fig. 5).

Fig. 5.

Suppression of ERα expression in the MPA did not prevent estradiol-mediated increases in home-cage activity. (A) Horizontal activity. (B) Total distance. (C) Vertical activity. All values summed across 12-h dark (black bar) and light periods (white bar). Values represent mean ± SEM. *P < 0.05, ***P < 0.001.

Discussion

Using a small interfering RNA directed against the coding region for ERα, specifically delivered to the MPA, the present study reveals that silencing ERα from MPA neurons abolishes maternal care. This study provides a unique example of how manipulating specific genes, in specific neurons within the brain, leads to unique, and highly specific, behavioral changes.

Preoptic ERα Expression and Maternal Care.

Viral-vector–mediated suppression of ERα in the MPA of female mice significantly decreased maternal care. In fact, maternal nurturing behaviors, including nursing and licking, were almost completely abolished in ERα-injected mice. Likewise, retrieval of pups was significantly reduced, and the majority of the mice (five out of seven) failed to retrieve the pups altogether. These results showed amazing specificity, since maternal care, demonstrated by nursing, licking and pup retrieval, was abolished, whereas maternal aggression, defending the nest from a male intruder, was completely unchanged.

Preoptic ERα Expression and Maternal Aggression.

Surprisingly, maternal aggression toward a male intruder was not significantly changed in animals with reduced levels of ERα in the MPA. This illustrates the specificity of the behaviors mediated by ERα in the MPA and demonstrates how using a combination of clever molecular and neuroanatomical techniques can unravel the intricacies of even the most complex behaviors. Animals with unilateral or inefficient bilateral suppression of ERα in MPA neurons exhibited intermediate phenotypes of maternal behavior.

Lesions of the preoptic region disrupt maternal behaviors. Female mice used in the current study were adults, and at this age, unlike in juveniles (27, 28), very small lesions in the POA are sufficient to elicit severe deficits in maternal behaviors (2, 3, 29). On the other hand, electrical stimulation of the POA leads to shorter onset to maternal behaviors (30). Neuronal activation within the preoptic region, as measured by expression of the immediate early gene c-Fos, is significantly increased in animals exhibiting maternal behaviors (31–39). Importantly, estradiol implants within the preoptic region enhance the expression of maternal behaviors or decrease the latency to displaying maternal behaviors (2, 23, 29, 40, 41). Particularly relevant to the current study, maternal behaviors specifically activate ERα-expressing neurons within the POA (33). Recent studies by Champagne et al. have further demonstrated that female rats that naturally exhibit a high degree of maternal care have significantly higher ERα mRNA and protein in the medial preoptic region than do females exhibiting naturally low amounts of maternal care (42). In addition, ERα knockout animals exhibit poor maternal care, with increased latencies to pup retrieval, and half of all females exhibit infanticidal behavior (24).

Using site-specific silencing of ERα, we unambiguously demonstrate that ERα-positive neurons within the MPA are of critical importance to the proper expression of a variety of complex components of maternal behaviors, including locomotor, nurturing, and aggressive responses.

Sexual Behaviors Mediated by Preoptic ERα Neurons.

As expected, sexual behaviors were significantly decreased following suppression of ERα in the MPA of female mice. Particularly in males, the POA has been shown to be critically important for the proper manifestation of sexual behaviors; electrical or chemical lesions of the medial preoptic region decrease or abolish male sexual behaviors (7, 8, 9, 43). In contrast, stimulation of the POA increases copulation in male mice (44), and the expression of c-Fos in the POA is increased in male mice (43, 45, 46). Within the POA, implants of estradiol promote sexual behaviors in male mice (22, 47, 48). In female mice, lesions of the POA decrease proceptivity and increase receptivity in female mice, whereas electrical stimulation of the MPA led to persistent decreases in lordosis (49, 50) without interfering with proceptivity (50). Microinjection of galanin bilaterally to the POA stimulates lordosis after EB priming, suggesting that an inhibitory tone on the MPA is necessary for lordosis to proceed (51). Previous findings demonstrated that an intact ERα signaling network is critical for the manifestation of sexual behaviors in female mice, as ERα-deficient mice do not exhibit female sexual behaviors (24, 25). The present study defines the MPA as a major site where ERα expression is critical for various aspects of aggressive behavior, including kicking and boxing of male sexual partners. This is in contrast to the responses observed in the ERα knockout females, who exhibited increased aggression toward male mice (24). Since rejective behaviors, or aggression toward a male sexual partner, were not increased following the silencing of ERα in the MPA, this suggests that perhaps this aspect of female sexual behavior may be mediated elsewhere. The current study implicates ERα signaling specifically in the MPA in estradiol-mediated increases in sexual behaviors.

Loss of Preoptic ERα Decreases Social Aggression and Food Competition Aggression in an SNE.

Complex social behaviors can only be studied by observing animals in an environment as close to their natural environment as possible. To this end, we developed a unique SNE that permits the housing and observation of dozens of mice simultaneously, thus enabling them to form complex social relations that small groups of mice in typical laboratory housing conditions are not known to exhibit (52, 53). In these environments, animals are allowed to interact with each over long periods of time, which permit the expression of social behaviors within the group, as well as interactions when challenges are presented to the group. By studying animals in this environment, we observed that ERα silencing in the MPA decreased social investigation and aggression in female mice. Furthermore, even under stressful conditions of food restriction, animals with reduced ERα in the MPA still exhibited significantly reduced aggression toward other inhabitants of the SNE.

General aggression in the SNE was invariably suppressed in animals with reduced ERα in the MPA; reduced aggression against a sexual partner reduced maternal aggression and reduced aggression toward other SNE inhabitants and intruders, even in conditions of food restriction. At first glance this may reflect a reduction in the generalized arousal level or reduction in motor activity of the animal; however, as can be seen with home cage activity, this is not the case. This reduction in aggression was a very specific, and robust, response that was observed in a variety of testing paradigms and environments.

Enhancements of Generalized Arousal Following Estradiol Treatment Surprisingly Not Mediated by Preoptic ERα.

Estrogens increase behavioral arousal, including running wheel activity (20, 21) and home cage activity (54, 55), which along with emotional reactivity are hypothesized to reflect a “generalized arousal state” of the central nervous system (56).

Confirming our previous studies, administration of estradiol to ovariectomized female mice increased locomotor activity within the home cage (57). The MPA is the brain location where application of estradiol is the most effective at enhancing motor activity (20). Furthermore, ERα signaling is essential to the enhancements of running wheel activity in response to EB, as ERα knockout animals fail to have increases after systemic EB treatment (58). Collectively, these findings led us to hypothesize that ERα within the MPA would be critical for the enhancement of locomotor activity by EB. To our surprise, this was not what we observed; in fact, home cage activity was not reduced in animals with bilateral silencing of ERα in MPA. Although ERα+ neurons are abundantly found throughout the MPA, we speculate that the preoptic locomotor region may lie more lateral than the MPA that we targeted in this study. A recent study by Spiteri et al., in rats, showed that animals with reduced ERα expression in the MPOA failed to have increases in running wheel activity in response to EB (59); these results could be attributed to different species being used (rats vs. mice), different coordinates, or the different EB regiment (dosage and chronic delivery via capsules vs. bolus injection). Overall motor activity in the home cage was not significantly different in EB-treated animals receiving either shRNA against LUC or ERα, indicating that these animals are not debilitated or sick and do not have lowered arousal or locomotor activity.

Experimental Procedures

Female Swiss-Webster mice 6–8 wk old from Taconic Farms were used. All animals were group housed with phytoestrogen-free food and water available ad libitum and a 12:12 h light:dark cycle (lights off at 1500 hours). For experiments investigating maternal, sexual, and aggressive behaviors, gonadally intact female mice were used (group I) (n = 24). For arousal studies, female mice were ovariectomized (OVX) under Nembutal anesthesia (50 mg/kg) and implanted with a silastic capsule containing either sesame oil (vehicle) (n = 20) or 17 β estradiol 3-benzoate 1.25 μg/capsule (n = 20) (group II). After surgery, animals were group housed and allowed to recover for 1 wk, before stereotaxic surgeries were performed (further details in SI Text).

All procedures were approved by The Rockefeller University Institutional Animal Care and Use Committee and followed the Public Health Services Policy on Humane Care and Use of Laboratory Animals.

Behavioral Testing.

Gonadally intact female mice were evaluated for sexual behavior, maternal care, maternal aggression, and social and food competition aggression (see SI Text for behavioral testing details).

Ovariectomized female mice were tested for generalized arousal (see SI Text for behavioral testing details).

Immunocytochemistry.

Following the completion of the behavioral tests, immunocytochemical evaluation of ERα expression in the MPA was performed in all mice (see SI Text for further details). Mice were included in the bilateral shRNA ERα if they had > 80% suppression of ERα expression—measured by intraclass correlation—in both MPA (for details see SI Text).

Statistical Analysis.

Comparisons between shRNA ERα- and shRNA LUC-microinjected animals were made using a one-way analysis of variance (ANOVA) for group I animals and a two-way ANOVA for animals from group II, followed by unpaired Student t test a posteriori. All statistical analyses were carried out using SigmaStat Software version 3.1, Systat.