Sex-dependent changes in metabolism and behavior, as well as reduced anxiety after eliminating ventromedial hypothalamus excitatory output.

OBJECTIVES: The ventromedial hypothalamic nucleus (VMH) regulates energy homeostasis as well as social and emotional behaviors. Nearly all VMH neurons, including those in the sexually dimorphic ventrolateral VMH (VMHvl) subregion, release the excitatory neurotransmitter glutamate and use the vesicular glutamate transporter 2 (Vglut2). Here, we asked how glutamatergic signaling contributes to the collective metabolic and behavioral responses attributed to the VMH and VMHvl.
METHODS: Using Sf1-Cre and a Vglut2 floxed allele, Vglut2 was knocked-out in SF-1 VMH neurons (Vglut2 (Sf1-Cre) ). Metabolic and neurobehavioral assays were carried out initially on Vglut2 (fl/fl) and Vglut2 (Sf1-Cre) mice in a mixed, and then in the C57BL/6 genetic background, which is prone to hyperglycemia and diet induced obesity (DIO).
RESULTS: Several phenotypes observed in Vglut2 (Sf1-Cre) mice were largely unexpected based on prior studies that have perturbed VMH development or VMH glutamate signaling. In our hands, Vglut2 (Sf1-Cre) mice failed to exhibit the anticipated increase in body weight after high fat diet (HFD) or the impaired glucose homeostasis after fasting. Instead, there was a significant sex-dependent attenuation of DIO in Vglut2 (Sf1-Cre) females. Vglut2 (Sf1-Cre) males also display a sex-specific loss of conditioned-fear responses and aggression accompanied by more novelty-associated locomotion. Finally, unlike the higher anxiety noted in Sf1 (Nestin-Cre) mice that lack a fully formed VMH, both male and female Vglut2 (Sf1-Cre) mice were less anxious.
CONCLUSIONS: Loss of VMH glutamatergic signaling sharply decreased DIO in females, attenuated aggression and learned fear in males, and was anxiolytic in males and females. Collectively, our findings demonstrate that while glutamatergic output from the VMH appears largely dispensable for counter regulatory responses to hypoglycemia, it drives sex-dependent differences in metabolism and social behaviors and is essential for adaptive responses to anxiety-provoking stimuli in both sexes.

Introduction

The murine ventromedial hypothalamus (VMH) is molecularly and functionally complex as evidenced by gene expression patterns [1] and phenotypes arising from numerous genetic lesions. Physiological and behavioral functions associated with the VMH include metabolic homeostasis, reproduction, social behaviors, anxiety, and female-specific energy expenditure, all of which are presumably mediated by distinct VMH neuronal subpopulations. Although complete annotation of these functionally distinct VMH neuronal subsets has yet to emerge, nearly all VMH neurons express two markers, steroidogenic factor 1 (SF-1 encoded by Nr5a1), and the vesicular glutamate transporter 2 (VGLUT2 encoded by Slc17a6). The prominent expression of Vglut2 in the VMH [2], [3] suggests that excitatory, glutamatergic neurotransmission mediates multiple and diverse aspects of VMH functions. In addition, glutamatergic connections between the VMH and other metabolic brain centers, such as the arcuate nucleus have been documented [4], [5]

To date, numerous genetic perturbations of the VMH have been reported, which target either VMH development or general signaling components affecting metabolism. For example, disrupting VMH development by reducing or ablating SF-1 prenatally leads to obesity in both sexes on standard chow and accelerates diet-induced obesity (DIO) on high fat diet (HFD) [6], [7], [8]. Eliminating SF-1 postnatally, using the CamKII-Cre preserves the gross architecture of the VMH, but still promotes DIO as well as hyperglycemia in both fed and fasted states [6]. We previously reported that blocking development and migration of neurons to the sexually dimorphic ventrolateral VMH (VMHvl) subregion results in marked female-specific obesity on standard chow [9]. Several genetic lesions in SF-1 expressing neurons that manipulate metabolic signaling pathways are also reported to change DIO in both sexes. Reducing leptin sensitivity after deleting LepR, PI3K (p110α), or Ptpn1 with the Sf1-Cre increases susceptibility to DIO [10], [11], [12], [13]. On the other hand, deleting insulin signaling (Insr) using the same Sf1-Cre prevents insulin-mediated inhibition of VMH neuronal activity and decreases DIO, at least in males [14]. Whether loss of these general signaling components in other SF-1 expressing tissues contributes to the observed metabolic phenotypes remains unclear. It is also unclear whether disrupting excitatory neurotransmission from all VMH neurons recapitulates the selective genetic lesions described above and leads to increased food intake and DIO. Tong et al. previously generated a VMH knockout of Vglut2 using Sf1-Cre (Vglut2) in a mixed genetic background [15]. Surprisingly, only a modest increase in weight gain is observed in Vglut2 males and females in response to a high-sucrose, high-fat diet (58% kcal fat). However, mutant mice exhibit lowered serum glucose in the fasted but not fed state, suggesting that loss of VMH excitatory output blunts the counterregulatory response to hypoglycemia.

As mentioned above, the VMH also regulates fear and anxiety-like behaviors [16], [17], [18] as well as social behaviors that are regulated by the VMHvl, such as male aggression [19], [20], [21]. Inhibiting SF-1 neurons via pharmacogenetics decreases freezing to predator stimuli in both males and females, suggesting that the VMH is required for mounting appropriate defensive behaviors in both sexes [17]. In a CNS-specific knockout of SF-1 using the Nestin-Cre (Sf1), anxiety-like behaviors are elevated as evidenced by fewer entries and less time spent in the open arm in the elevated plus-maze (EPM) assay [18]. However, as with the global Sf1 knockout, Sf1 mice show gross abnormalities in VMH architecture. Hence, it is unclear if the increased anxiety phenotype noted in the Sf1 mice arises from organizational rather than functional VMH deficits.

To assess how blocking all excitatory VMH output might modulate the many physiological and behavioral functions associated with the VMH, we recreated the Vglut2 allele using Sf1-Cre and a Vglut2 floxed allele (Vglut2) [22]. As shown previously, targeted deletion of Vglut2 effectively blocks all synaptic release of glutamate from neurons that express this transporter protein [23]. Using multiple metabolic and behavioral assays, we examined the consequences of silencing glutamatergic signaling in both male and female mice bred in two genetic backgrounds. Analyses of these Vglut2 mice show that excitatory VMH output is an important factor in female metabolism, but surprisingly, less so for glucose homeostasis. Further, our results establish an essential role for VMH glutamatergic output in mounting adaptive behavioral responses to contextual and social cues.

Materials & methods

Animals

Vglut2 mice in a mixed C57BL/6, Sv129J background were provided by Dr. R.D. Palmiter, (University of Washington) [22]. Sf1 mice were provided by Dr. J.K. Elmquist (University of Texas Southwestern Medical Center) [11] and subsequently backcrossed into the C57BL/6 background (Taconic Biosciences), which was confirmed by microsatellite analysis. Vglut2 mice were generated and maintained on a mixed background or backcrossed for 10 generations into the C57BL/6 background. Experimental cohorts were obtained by crossing Sf1; Vglut2 with Vglut2 mice. We previously generated and characterized the Sf1 reporter mice [24].

Mice were maintained on a 12 h light–dark cycle with ad libitum access to water and to either a standard chow diet (5058; LabDiet, 4% fat) or a high fat diet (D12492; Research Diets, 60% fat), as indicated. All animal procedures were performed in accordance with the UCSF animal care committee's regulations and the Ingraham lab IACUC protocol of record.

Tissue collection and processing

Mice were deeply anesthetized with 2.5% Avertin and perfused with phosphate buffered saline (PBS) followed by 4% paraformaldehyde (PFA). Brains were dissected and post-fixed overnight in 4% PFA. Fixed tissue was cryoprotected in 30% sucrose and embedded in OCT (Tissue-Tek). Cryosections (20 μm) were collected on glass slides and stored at −80 °C until processing.

Immunofluorescence and in situ hybridization

Frozen brain sections were processed using standard procedures. Reporter GFP expression was detected by immunofluorescence using chicken anti-GFP antibody (1:2500; Aves Labs) and AlexaFluor 488-conjugated goat anti-chicken antibody (1:1000; Invitrogen). Detection of mRNA expression by in situ hybridization (ISH) was accomplished using probes against the carboxy-termini and 3′ UTR of Vglut1 and Vglut3 (nt:1589–2510 and nt:1511–1869, respectively); exon 2 of Vglut2 (nt:902–1230); and exons 9–18 of Gad67 (nt:1313–2267). DIG-labeled (Roche) riboprobes were transcribed according to the manufacturers specifications. ISH was performed using standard procedures, as previously described [1].

Double-label immunofluorescence in situ hybridization (DISH) was performed with the following modifications to standard IF/ISH protocols. Sections were permeabilized with 0.3% Triton X-100 (in PBS) for 10 min at room temperature. Vglut2 riboprobe was hybridized to the sections overnight at 72 °C. Sections were washed and blocked with 10% heat-inactivated donkey serum (HIDS) in buffer B1 (0.1 M Tris, pH 7.5; 0.15 M NaCl) for 3 h at room temperature. Primary antibodies (1:3000 Anti-Dig-AP, Roche; 1:2500 chicken anti-GFP) diluted in B1 with 1% HIDS were added to the sections and incubated overnight at 4 °C. Sections were washed and incubated with biotin-conjugated goat anti-chicken antibody (1:500) (Invitrogen) in B1 with 2% HIDS for 2 h at room temperature. After washing 3 × 5 min in buffer TNT (1 M Tris HCl, pH7.5; 5 M NaCl; 0.05% Tween 20), IF and ISH signals were detected sequentially using the TSA Biotin System (Perkin Elmer) and the HNPP Fast Red Detection Kit (Roche), respectively, according to the manufacturers' instructions. Sections were counter-stained with DAPI (1:1000) for 5 min at room temperature and mounted with aqueous mounting media.

Metabolic analyses

Male and female body weights were recorded weekly beginning at the time of weaning (3 wks of age). Fed and 24-hour fasting glucose levels were measured in 8–10 wk old mice on standard chow or after 6 wks on HFD. Glucose tolerance tests were performed after 6 wks on HFD. Mice were fasted overnight for 16 h and given glucose (2 g/kg) by IP injection. For both assays, blood was collected from the tail vein, and glucose levels measured using the OneTouch Ultra Glucometer (LifeScan).

Measurements of metabolic parameters and indirect calorimetry were conducted by the UCSF Diabetes Center Metabolic Core facility using the Oxymax Comprehensive Laboratory Animal Monitoring System (CLAMS) chambers. Prior to data collection, mice were acclimated to individual housing for 1 wk and subsequently acclimated to CLAMS chambers for 3 days. Chambers were maintained at room temperature, and mice had ad libitum access to standard chow and water.

Behavioral analyses

Levels of anxiety in adult male and female mice were analyzed by the UCSF Gladstone Mouse Phenotyping Core Facility (http://labs.gladstone.ucsf.edu/behavioral/home). For all assays described below, adult Vglut2 and Vglut2 control littermates in the C57BL/6 background were used. Tests were started during the light phase between 9:00–11:00. Mice were allowed to recover for at least 1 week between each test. The order of testing for each experimental cohort is as follows:

Rotarod

Subjects were transferred to the room containing the rotarod apparatus 30 min prior to testing. During the test, the speed of the rotating rod was increased from 4 to 40 rpm at 8 rpm/min. Latency to fall was recorded by the experimenter. Vertical spacers on the rod allowed five mice to be tested simultaneously. Each mouse was tested 3 times with an inter-trial interval of 15–20 min. The rotarod was cleaned with 70% ethanol between testing sessions.

Open field assay

The arena was a clear plastic chamber (41 × 41 × 30 cm). Activity and position measurements in the arena were recorded electronically using two 16 × 16 photobeam arrays (San Diego Instruments). Mice were habituated to the testing room under normal light for 30 min. During the testing interval, mice were allowed to explore the arena for 15 min. The arena was cleaned with 70% ethanol between testing sessions.

Elevated plus-maze

The testing apparatus consisted of a plus-shaped maze (Hamilton–Kinder) with two open and two closed arms positioned 63 cm above the ground. Mice were transferred to the testing room and acclimated to dim lighting for 30 min. At the start of the test, subjects were placed in the intersection of the plus-maze. Mice were allowed to explore the maze for 10 min, while activity and location measurements were recorded electronically. The maze was cleaned with 70% ethanol between testing sessions.

Corticosterone assay

Tail vein blood was collected in the beginning of the light phase to determine baseline corticosterone levels in control and mutant mice. The following day, an acute stress was induced by 15 min of restraint in the Broome Rodent Restrainer (Harvard Apparatus) or by a 5 min exposure to the elevated plus-maze. Following these stressors, a second blood sample was collected. Plasma hormone levels were quantified using the Corticosterone Enzyme Immunoassay (EIA) Kit (Cayman Chemical), according to the manufacturer's instructions.

Cued fear conditioning

Training and testing were performed in a sound attenuated clear, square chamber. Mice were transferred to the room containing the conditioning apparatus 30 min before training or testing. Initial levels of freezing were recorded of 5 min with no audible tone or shock presented. The next day, 3 min baseline freezing levels were recorded. The mice were then trained with three repetitions of a 30 s audible tone co-terminating with a 2 s foot shock. After each training trial, freezing levels were recorded for 90 s. On day 3, the context of the conditioning apparatus was altered (switched to a black, triangular, unlit chamber). After 2.5 min in the chamber, mice were presented with three 30 s exposures to the conditioned tone (100 s inter-tone interval). Freezing levels during each 30 s tone exposure were recorded.

Predator odor assay

Experimental mice were habituated to testing environment for 45 min prior to exposure of 30 μL of 10% solution of 2,3,5-Trimethyl-3-thiazoline (TMT) (Sigma) placed on filter paper. Behaviors were videotaped in the dark cycle with red light while the experimenter was out of room. Total exposure was 5 min, and responses scored for the final 3 min. Time spent engaged in freezing, sniffing, and grooming were scored by an observer, who was blinded to the genotype of mice.

Resident intruder assay

Experimental mice were singly housed for 1 wk. Individual males from a socially housed cohorts were placed into the home cages of the resident Vglut2 and Vglut2 male mice, and mice were allowed to interact for 15 min. Interactions were videotaped and scored offline. Attacks toward the intruder were scored twice by 2 independent observers, who were blinded to the genotype of mice.

Statistics

Data were analyzed using the GraphPad Prism statistical software package (version 6.0). Pairwise comparisons were made using unpaired, two-tailed t-tests. Comparisons involving multiple factors were made using 2-way ANOVA with repeated measures and Bonferroni post-test. Results were considered statistically significant when p < 0.05. Data are presented as means ± SEM. For all figures, * = p < 0.05, ** = p < 0.01, *** = p < 0.005, and **** = p < 0.001.

Results

Loss of Vglut2 expression in Vglut2 mice does not alter gross organization of the VMH

As expected, SF-1 and Vglut2 are coexpressed throughout the rodent VMH (Figure 1A). We used Sf1-Cre backcrossed into a pure C57BL/6 background [11], [24] and Vglut2 [22], [25] to conditionally eliminate VGLUT2 in neurons from the SF-1 lineage (Vglut2). As demonstrated previously, eliminating exon 2 renders VGLUT2 non-functional [23]. Within the CNS, Sf1-Cre-mediated loss of Vglut2 is selective for the VMH and does not diminish Vglut2 expression in other brain regions (Figure 1B). Furthermore, no compensatory upregulation of the other vesicular glutamate transporters, Vglut1 or Vglut3 is observed in the VMH of Vglut2 mice. Similarly, Vglut2 males and females show normal patterns of glutamate decarboxylase (Gad67) expression, which marks inhibitory neurons and is normally excluded from the VMH (Figure 1B). No differences in the pattern and intensity of major VMH projections were noted in Vglut2, as illustrated by the prominent ascending fibers through the ventral supraoptic commissure (VSOC) (Figure 1C) [24]. Taken together, these data establish that the gross exterior boundaries and major efferent projections of the VMH remain intact after deleting Vglut2 in the VMH.

Figure 1

Deletion of in the VMH does not alter gross anatomical organization or lead to compensatory upregulation of Vglut1 or Vglut3. A) SF-1+ neurons in the VMH were identified using Sf1. Expression of the SF-1 reporter (green) and Vglut2 exon 2 mRNA (red) in coronal sections at the level of the VMH was detected simultaneously using DISH (third ventricle = 3V). B) Expression patterns of Vglut2, Vglut1 and Vglut3, as well as Gad67 transcripts as detected by ISH. For Vglut2, higher magnification images of the boxed regions are shown in lower panels. C) Coronal sections showing the major VSOC efferent projection from the VMH in Vglut2 (top panel) and mutant Vglut2 (lower panel) mice. SF-1 neuronal projections are visualized as described in Materials and Methods and after crossing control and mutant mice with Sf1 reporter mice [24].

Eliminating Vglut2 in the VMH does not change fasting blood glucose and unexpectedly attenuates diet-induced obesity in females

Control and mutant mice were initially maintained in a mixed genetic background. When fed standard chow (4% kcal fat), no significant differences in body weight were observed between Vglut2 and Vglut2 males or females (Figure 2A, left panel). When placed on HFD (60% kcal fat) beginning at 10 wks of age, weight gain was equivalent in Vglut2 males compared to control littermates (Figure 2B, left panel). Vglut2 females gained slightly less weight over time on HFD with 2-way repeated measures ANOVA revealing a significant interaction between genotype and time on HFD (F7,98 = 3.116, p = 0.0052). Surprisingly, neither Vglut2 mutant males nor females exhibited significant differences in fasting blood glucose levels on standard chow or on HFD (Figure 2A, B, right panels). These data differ from the slight increase in DIO and the impaired glucose homeostasis observed in separately derived Vglut2 mice [15].

Figure 2

Fasting glucose levels on HFD are unaffected in on a mixed genetic background. Weekly body weight measurements of male (M) and female (F) Vglut2 and Vglu2 mice from 3 to 18 weeks of age. (A) Body weights of mice on standard chow (left panel) with corresponding blood glucose levels in each cohort measured pre- (Pr, solid bars) and post- (Po, hatched bars) a 24-hour fast (right panel). (B) Body weights of mice on standard chow and placed on HFD (arrow) at 10 weeks (left panel) with corresponding blood glucose levels measured as above and after HFD for 6 wks (right panel). Data are presented as means ± SEM. Statistical significance was determined by 2-way repeated measures ANOVA and Bonferroni post-test. Number of animals is indicated in each panel.

Given these discrepant results, we generated congenic Vglut2 and Vglut2 mice in the C57BL/6 background, a strain that exhibits increased weight gain and hyperglycemia on HFD [26], [27]. Similar to the mixed background, body weights between male Vglut2 and Vglut2 C57BL/6 mice were equivalent on standard chow. However, we observed a modest, but significant, decrease in female Vglut2 body weights (Figure 3A). Other metabolic parameters were analyzed in Vglut2 C57BL/6 mice using CLAMS chambers. The decreased weight of Vglut2 females correlated with a non-significant trend towards increased heat generation during both the light and dark periods (Figure 3B). Vglut2 females had normal expression of the thermogenic uncoupling protein 1 (Ucp1) in brown adipose tissue (BAT) at both warm and cold temperatures (Figure S1A). Further analysis of the CLAMS data revealed no significant alterations in ambulatory activity, food intake, oxygen consumption, respiratory exchange ratio, or lean mass between female Vglut2 and Vglut2 mice on standard chow (Figures 3C and S1B).

Figure 3

Attenuated Response to HFD in female mice on a pure C57BL/6 genetic background. A) Weekly body weights of adult male (M) and female (F) Vglut2 and Vglut2 mice on standard chow (NC). Indicated metabolic parameters measured in adult female Vglut2 and Vglut2 mice on standard chow using CLAMS chambers. B) Hourly heat production during a 24-hour period. C) Average hourly heat production, ambulatory activity, food intake, oxygen consumption (VO2), and respiratory exchange ratio (RER) during the same 24-hour period as shown in panel B. D) Weekly body weights of male (M) and female (F) Vglut2 and Vglu2 mice fed HFD beginning at 10 weeks of age (arrow). E) Glucose tolerance test in adult (16 wks old) female Vglut2 and Vglut2 mice on HFD for 6 wks. Data are presented as means ± SEM. For panels A, D, and E, the given p values represent the main effect of genotype as determined by 2-way repeated measures ANOVA, and for each time point results from Bonferroni post-test are indicated with * = p < 0.05; ** = p < 0.01; *** = p < 0.005; **** = p < 0.001. Number of animals is indicated in each panel.

As expected, DIO was readily apparent in the C57BL/6 mice on HFD compared with the mixed background, especially in females (Figure 2, Figure 3B). When compared to Vglut2 controls, weight gain in Vglut2 females was notably lower when placed on HFD at 10 wks of age (Figure 3D); however both genotypes consumed equivalent amounts of HFD (Figure S1C). In contrast, C57BL/6 mutant males showed no body weight differences on HFD. Consistent with the female-specific resistance to DIO, glucose clearance was improved in the Vglut2 females but not in mutant males as measured by an intraperitoneal glucose tolerance test performed at 16 wks of age (Figures 3C and S1D). The unexpected body weight reduction in Vglut2 females suggests that normal weight gain and sensitivity to DIO in females depends on glutamatergic output from the VMH. It should be noted that no gross deficits in reproduction were noted in Vglut2 females on both the mixed and C57BL/6 backgrounds, suggesting that the hypothalamic-pituitary-gonad axis controlling reproductive physiology was not impaired in these mutant mice.

Reduced anxiety-like behaviors in male and female Vglut2 mice

While conducting metabolic studies in which mice were housed individually, we noted a difference in the location of home cage nests built by Vglut2 and Vglut2 mice (Figure 4A). Nearly all Vglut2 control mice (96%) followed a stereotypic pattern and built their nests in the corner of the cage. In contrast, the majority of Vglut2 mice (62%) chose a central nest location. Based on these preliminary observations, anxiety-like behaviors in mutant and control mice were formally assayed in open field (OF) and elevated plus-maze (EPM) tests. Although the nesting phenotype was initially observed in mutant mice on the mixed background, the following behavioral tests were exclusively performed in C57BL/6 Vglut2 and Vglut2 mice. The C57BL/6 background is known to have intermediate levels of anxiety-like behavior and therefore allows detection of both anxiolytic and anxiogenic phenotypes [28]. An initial rotarod test confirmed that motor skills were equivalent between genotypes (data not shown). However, in an OF test both male and female Vglut2 mice exhibited significantly more ambulatory movement in the center of the field than control Vglut2 mice (Figure 4B). We also observed higher total ambulatory activity for Vglut2 males within the short OF test (15 min) (Figure 4B). Because total ambulatory movement was unchanged in mutant Vglut2 males when monitored over a 4 day period (Figure S1E), the increased movement in the OF test suggests a greater exploratory drive for mutant males when initially placed in a novel environment.

Figure 4

Male and female mutant mice have reduced levels of anxiety-like behaviors. A) Comparison of nest locations within the home cages of adult Vglut2 and Vglut2 mice. Percentages reflect the ratio of home cages with centrally located nests built by mutant (top panel) and control (bottom panel) mice. B) Activity in center of open field (upper panel, %) and total ambulatory activity (lower panel, beam breaks) measured in adult male and female Vglut2 and Vglut2 mice in OF test. C) Percentage of entries, time spent and distance in the open arm (OA) of the elevated plus-maze as well as total distance in open and closed arms (lower panel) for male and female Vglut2 and Vglut2 mice. Measurements were binned into two 5-min intervals as graphed. D) Serum corticosterone levels measured in adult male Vglut2 and Vglut2 mice at baseline (Pre) and after (Post) exposure to the indicated stressors. Data are presented as means ± SEM. Statistical significance was determined by unpaired t-tests (panel B) or by 2-way repeated measures ANOVA with Bonferroni post-test (panel C). Number of animals is indicated in each panel.

Exploratory and anxiety-like behaviors were also examined using the EPM assay. Entries into the anxiety-promoting open arm, time spent in the open arm, and distance covered in the open arm were analyzed across two consecutive intervals (0–5 and 5–10 min). Similar to the OF assay, Vglut2 males, but not females exhibited a higher level of exploratory drive and/or activity in the initial interval (0–5 min), as measured by the total distance traversed in the EPM (Figure 4C, last panel). During the final 5-min interval (5–10 min), both male and female Vglut2 mice displayed higher levels of all three open arm parameters relative to controls (Figure 4C). Despite the sustained open arm activity observed in mutants, total distance traveled was equivalent between genotypes. A significant interaction between genotype and time was observed for open arm entries (males: F1,22 = 17.33, p < 0.0005; females: F1,25 = 5.751, p < 0.05); this EPM behavior is thought to best reflect altered anxiety levels [29]. The increased open arm activity in the Vglut2 mice is consistent with their increased time in the center of the OF, and together these results demonstrate that loss of glutamatergic output from the VMH is anxiolytic in both males and females.

To confirm that Vglut2 mice could mount an appropriate physiological response to stressors, serum corticosterone levels were measured after 5 min in the EPM or 15 min of acute restraint. Despite lowered anxiety-like behavior in both the OF and EPM assays, corticosterone levels increased appropriately in Vglut2 mice (Figure 4D). For both wild type and mutant genotypes, two-way repeated measures ANOVA revealed a significant main effect of stressor only (EPM: F1,21 = 228.8, p < 0.0001; acute restraint: F1,21 = 420.13, p < 0.0001). These data show that the hypothalamic-pituitary-adrenal (HPA) axis is fully intact and unaffected by loss of glutamatergic transmission from the VMH.

Loss of Vglut2 in the VMH impairs fear conditioning and aggression in males

VMH neurons are part of the circuit linking the amygdala to the periaqueductal grey, both of which are essential for eliciting normal emotional behavior. Therefore, we asked if learned or innate responses to fear-provoking stimuli are affected in Vglut2 mice. Vglut2 mice were subjected to a cued fear-conditioning assay, which links painful stimuli with an auditory cue. While both male and female Vglut2 mice exhibited normal freezing responses to the aversive foot shock during training, mutant males exhibited a much weaker learned response during testing (F1,23 = 10.06, p < 0.005) (Figure 5A). In contrast, female Vglut2 mice had normal cued fear responses, with only a significant main effect of conditioning (F2,54 = 204.82, p < 0.0001). Given that female Vglut2 showed no difference in learned fear, their innate response to the predator fox odor, 2,4,5-trimethyl-3-thiazoline (TMT) was tested. Similar to the cued fear-conditioning, mutant females exhibited normal freezing responses to TMT (Figure 5B).

Figure 5

Male mice exhibit lowered learned fear and social aggression. A) Performance of male and female Vglut2 and Vglut2 mice in a cued fear conditioning assay that measures learning and memory associated with an aversive stimuli. Fear was quantified as the percent of total time spent freezing immediately after paired presentation of audible tone and foot shock (Train), and 24 h later in response to the conditioned tone (Test). B) Time engaged in indicated behaviors during exposure to an olfactory component of fox feces, TMT. C) Aggression in male Vglut2 and Vglut2 mice quantified as the number of attacks toward a resident intruder. Data are shown for all animals tested (All Animals) and only for male mice that displayed aggressive behavior (Aggressors). Data are presented as means ± SEM. Significance was determined by 2-way repeated measures ANOVA with Bonferroni post-test (panel A) or by unpaired t-test (panels B, C). Number of animals is indicated in each panel.

Finally, we examined the requirement for glutamatergic VMH output in a male-specific social behavior using the resident-intruder assay. Male resident Vglut2 mice were significantly less aggressive than the Vglut2 controls to a male intruder (Figure 5C). Moreover, among mice that exhibited aggressive behavior, Vglut2 males attacked far less frequently than their control Vglut2 littermates. Together, these data show that VMH glutamatergic output is essential for context-dependent behaviors in males and anxiety-like behavior in both sexes.

Discussion

Here, we created a conditional knockout of VGLUT2 in SF1-expressing cells to examine how excitatory VMH neurotransmission influences metabolism and behavior. Unexpectedly, Vglut2 mice fail to show increased susceptibility to DIO or impaired glucose homeostasis. Rather we find sex-dependent resistance to DIO that is accompanied by improved glucose tolerance in Vglut2 females. Conditional loss of VGLUT2 in the VMH also leads to a profound anxiolytic effect in both sexes and reduces aggression and conditioned-fear in males. Our results suggest that glutamatergic signaling in the VMH is largely dispensable for glucose homeostasis, but essential for sex-specific energy storage and for adaptive behavioral responses to anxiety and fear provoking stimuli.

Blocking VMH glutamatergic output results in female-specific resistance to DIO

The reduced body weight of congenic Vglut2 C57BL/6 female mice, even on standard chow, demonstrates that glutamatergic output from the VMH is an essential determinant of female energy balance. Although the VMH is a well-known regulator of metabolic homeostasis, this result differs from the increased sensitivity to DIO when leptin signaling in the VMH is ablated using Sf1-Cre [11]. These divergent metabolic phenotypes may reflect silencing of excitatory output from all VMH neurons (this study) versus disturbing a single signaling pathway in a subpopulation of VMH neurons. Clearly, the ability to genetically target and lesion different VMH subpopulations is needed for a complete functional understanding of the VMH and energy balance. Indeed, it is often overlooked that use of the Sf1-Cre, while specific for the VMH in the brain, is active in other tissues (such as the adrenal gland, pituitary, gonad, and spleen). Thus, it is possible that metabolic disturbances are influenced by depletion of signaling components in other SF-1-expressing tissues outside of the VMH. This issue is less of a problem with Vglut2 mice. With the exception of rat pituitary gonadotropes [30], the intersection of VGLUT2 and SF-1 expression is largely limited to the VMH. Further, because Vglut2 mice exhibit normal reproduction, we posit that all metabolic phenotypes observed in Vglut2 females result primarily from loss of glutamatergic signaling in the VMH.

Our study differs from the previous report that Vglut2 males and females are heavier when fed a high fat, high sucrose diet, and fail to regulate blood glucose during fasting [15]. One important difference between these two studies is the composition of the HFD. Our HFD contains a lower percent of kcal from carbohydrate and a lower total energy content than the high fat, high sucrose diet utilized by Tong et al. (2007) (20% versus 26% and 5.24 kcal/g versus 5.557 kcal/g, respectively). The source of dietary fat also varies substantially between the two studies (lard versus hydrogenated coconut oil). These nutritional factors could very well lead to differences in DIO outcomes, but fail to adequately account for differences in fasting glucose levels on standard chow. Genetic backgrounds introduced by the Vglut2 floxed allele and the Sf1-Cre line are quite different in our study and could potentially affect both DIO [26] and glucose homeostasis by influencing the degree of the hypoglycemic insult needed to reveal a defective counterregulatory response. For our analyses, the Sf1-Cre line was backcrossed on to the C57BL/6 background and exhibits robust and specific Cre-mediated recombination throughout the VMH, including the VMHvl [24]. Thus while VMH neurons are reported to control the counterregulatory response to hypoglycemia [31] and receive input from glucoregulatory CCK-expressing neurons [32], we conclude that under our experimental conditions glutamatergic output from the VMH does not appear to be the primary determinant of glucose homeostasis.

The sex-dependent metabolic phenotype observed in Vglut2 mice adds to the growing body of literature showing that the VMH is a critical regulator of energy balance and body weight in females. This aspect of VMH function is dependent on estrogen receptor alpha (ERα) in the VMHvl. Reducing ERα expression in the VMHvl [33], [34] or the number of ERα-expressing VMHvl neurons [9] leads to female-specific impairment of energy expenditure and increased body weight. In contrast, high estrogen levels or exogenous delivery of estrogen directly to the VMH increases energy expenditure and lowers body weight [35]. Surprisingly, the reduced body weight in the female Vglut2 mice, the origins of which remain unclear, mimics the metabolic consequence of elevated estrogen signaling. This result undermines the simple assumption that estrogen signaling potentiates VMH neurotransmitter output, suggesting instead that ERα signaling inhibits a circuit that involves glutamatergic VMH output, and that otherwise promotes energy storage. Because this brain region is tightly linked to reproductive behavior [36], we further speculate that estradiol-mediated regulation of glutamatergic VMH neurons in females maximizes fuel reserves in states of overnutrition (HFD) and ultimately improves reproductive fitness in times of undernutrition.

Glutamatergic output from the VMH controls innate and learned behavior

Our data also reveal that glutamatergic VMH output contributes greatly to excitatory signals and neural circuits that modulate emotional behaviors, including those associated with anxiety-like behavior, fear, and sex-specific social behaviors. Indeed, we find that glutamatergic signaling from the VMH is required for male aggression. Ablating VMHvl neurons dramatically inhibits male aggressive behaviors [21]. Conversely, artificially activating VMHvl neurons triggers male aggressive behaviors [20]. As predicted from these studies, we find that disrupting the excitatory VMH output decreases the amount and extent of attack behavior when male Vglut2 mice are presented with a resident intruder. That Vglut2 males, but not females, exhibit lowered behavioral responses to a conditioned fear-provoking stimulus and exhibit increased novelty-associated locomotion but have no metabolic phenotype expand on the theme that the VMH mediates important sex-differences. Moreover, these data imply that neural circuits controlling fear and anxiety-like behaviors and metabolic responses are not tightly coupled.

Similar to the anxiolytic phenotype described here, acute, postnatal ablation of Sf1-positive neurons, which would also extinguish VMH glutamatergic output, significantly decreases anxiety-like behavior in the EPM [37]. The anxiolytic phenotype observed in Vglut2 mice differs from the anxiogenic phenotype noted in Sf1 mutants. However, the latter genetic lesion not only eliminates SF-1 in the VMH, but also disrupts the overall integrity of the VMH. Taken together, we conclude that the differences in anxiety-like behaviors observed in Vglut2 and Sf1 result from impaired VMH neuronal function versus disrupted VMH organization, respectively. We further demonstrate that circulating corticosterone, which mediates the physiological response to stress upon activation of the HPA axis is unaffected in Vglut2 mutants. Interestingly, Sf1 mutants exhibit normal circadian fluctuations in corticosterone [18]. Thus, the role of the VMH in anxiety-like behavior appears distinct from regulation of the HPA axis

The periaqueductal grey (PAG) is a prominent downstream target of VMH projections [21], [24], [38], and regulates behaviors associate with anxiety and fear in response to glutamate. Consistent with the anxiolytic effect observed in mutant mice that have lost excitatory VMH output, delivering a glutamate receptor antagonist, 2-amino-7-phosphonoheptanoic acid, directly to the PAG in rats decreases anxiety-like behavior in the EPM [39]. Silencing VMH or PAG activity in mice using designer receptors exclusively activated by designer drugs (DREADDs) also impairs behavioral responses to fear-inducing stimuli, such as a predator or an aggressive conspecific [17]. Local application of a glutamate receptor agonist, N-methyl-d-aspartate, to the PAG in rats increases anxiety-like behavior in the EPM [40]. Collectively, we suggest that glutamatergic output from the VMH provides at least some of the excitatory input to the PAG that drives emotional behaviors associated with anxiety and/or fear.

Results from our behavioral assays indicate that not all responses to fearful stimuli are disrupted after inhibiting VMH glutamate release. For instance, female mutant and control mice display equivalent levels of freezing to a predator odor, and both male and female Vglut2 mice had normal freezing responses to foot shock during fear conditioning. However, during testing, male Vglut2 mice had a weaker conditioned response, indicating impairment in the formation or recall of the association between auditory tone and foot shock. Together, the lack of a uniform response to different anxiogenic and fear-provoking stimuli in Vglut2 mutants is broadly consistent with the idea that distinct neural circuits mediate behavioral responses to different sources of anxiety and/or fear [41].

Conclusions

In summary, our study establishes that excitatory glutamatergic output from the VMH is required for optimal energy storage in females and appropriate behavioral responses to novel or fear-evoking stimuli in both sexes. The phenotypes of Vglut2 mice reported here are consistent with emerging evidence that the VMH, specifically the VMHvl, regulates sex-dependent metabolic responses and social behaviors. Clearly, additional work is needed to define the precise anatomical and molecular pathways in the VMH needed to generate the complex emotional and metabolic responses associated with this neuroendocrine region.