Bombesin-like receptor 3 (Brs3) expression in glutamatergic, but not GABAergic, neurons is required for regulation of energy metabolism

Objective Bombesin-like receptor 3 (BRS-3) is an orphan G protein-coupled receptor. Brs3 null mice have reduced resting metabolic rate and body temperature, increased food intake, and obesity. Here we study the role of Brs3 in different neuron types. Methods Mice able to undergo Cre recombinase-dependent inactivation or re-expression of Brs3 were generated, respectively Brs3fl/y and Brs3loxTB/y. We then studied four groups of mice with Brs3 selectively inactivated or re-expressed in cells expressing Vglut2-Cre or Vgat-Cre. Results Deletion of Brs3 in glutamatergic neurons expressing Vglut2 reproduced the global null phenotype for regulation of food intake, metabolic rate, body temperature, adiposity, and insulin resistance. These mice also no longer responded to a BRS-3 agonist, MK-5046. In contrast, deletion of Brs3 in GABAergic neurons produced no detectable phenotype. Conversely, the wild type phenotype was restored by selective re-expression of Brs3 in glutamatergic neurons, with no normalization achieved by re-expressing Brs3 in GABAergic neurons. Conclusions Brs3 expression in glutamatergic neurons is both necessary and sufficient for full Brs3 function in energy metabolism. In these experiments, no function was identified for Brs3 in GABAergic neurons. The data suggest that the anti-obesity pharmacologic actions of BRS-3 agonists occur via agonism of receptors on glutamatergic neurons.


INTRODUCTION
Obesity is an increasing world-wide problem with a profound effect on human health [1,2]. It is caused by an imbalance between energy intake and energy expenditure, which is regulated by the brain. Within the brain, the hypothalamus has a major role, integrating endocrine and peripheral nerve signals with inputs from other brain regions and sending signals centrally and peripherally. Bombesin-like receptor 3 (BRS-3) is an orphan G protein-coupled receptor (GPCR), first identified via sequence similarity to the gastrin releasing peptide and neuromedin B receptors [3e5]. To date, attempts to identify a mammalian endogenous BRS-3 ligand have been unsuccessful [6e9], although in chicken and spotted gar (a fish), gastrin releasing peptide and neuromedin B both appear to be endogenous ligands [10]. Mammalian BRS-3 may be conserved during evolution because it can modify the response pattern of other GPCRs or contribute to basal signaling [11]. Brs3 mRNA and BRS-3 binding activity are located in restricted regions of the brain, including portions of the hypothalamus, amygdala, and thalamus [12e19]. BRS-3 is also reported to be present at low levels in peripheral sites including pancreatic islets, developing testis, female reproductive tract, lung, and muscle, and in certain cancers [5,20e22] (https://gtexportal.org/ home/gene/BRS3). Insights into the function of BRS-3 were hugely advanced by development of a Brs3 knockout (KO) mouse [23]. The null phenotype includes obesity, increased food intake and meal size, and reductions in metabolic rate, resting body temperature, and resting heart rate [23e 27]. The development of potent and selective ligands has expanded our knowledge of BRS-3 ( [15], reviewed in [28,29]). Treatment with a BRS-3 antagonist increased food intake and body weight, while agonists reduced food intake, increased resting metabolic rate, body temperature, and heart rate [15,27,30,31]. One BRS-3 agonist, MK-5046, reached initial human studies but increased blood pressure and was discontinued [32]. It is currently unknown which cells mediate these functions of BRS-3. For example, it has not even been established if neuronal BRS-3 are the relevant receptors for regulation of body weight and energy homeostasis. Conditional deletion removes gene function in precisely defined subsets of cells, supplying evidence for the necessity of the gene (e.g., [33,34]). Conversely, reconstitution by conditional re-expression 1 identifies those cells that are sufficient for gene function. For example, these tools have been used to elucidate the hypothalamic MC4R neurons that regulate food intake [35]. To dissect the functions of Brs3, we developed mice that allow selective, conditional deletion or re-expression of Brs3. We use these mice to investigate the necessity and sufficiency of Brs3 in glutamatergic and GABAergic neurons for regulation of energy homeostasis and identify a role for Brs3 in glutamatergic but not GABAergic neurons.

Phenotyping
Body weight and food intake were measured weekly. Body composition was measured by time domain Echo MRI 3-in-1 (Echo Medical Systems, Houston, TX) every two weeks. Energy expenditure was estimated by energy balance of singly-housed mice in their home cage environment. In brief, estimated energy expenditure is calculated from the metabolizable caloric intake, corrected for the change in caloric content of the mouse (from the change in body composition over the measurement interval) [40]. Glucose and insulin tolerance tests, hormone and metabolite profiles: Intraperitoneal glucose (2 g/kg for chow-fed mice, 1 g/kg for HFD-fed mice, with AUC calculated from 0 mg/dl) tolerance tests were performed at 0900, following an overnight (16 h) fast. Glucose was measured with a Glucometer Contour (Bayer, Mishawaka, IN). Insulin (0.75 unit/kg, i.p.) tolerance tests were performed at 0900, in nonfasted mice, with AUC calculated from 0 mg/dl. Blood was collected at 0900 by tail bleed at 20 weeks of age for fed glucose and serum insulin, free fatty acid, cholesterol, and triglycerides measurement. Serum for other measurements was taken from anesthetized (100 mg/ kg ketamine and 10 mg/kg xylazine, ip) mice by retro-orbital bleed at euthanasia and frozen until assayed. Free fatty acids (FFA, Roche Diagnostics GmbH, Mannheim, Germany), triglycerides (Pointe Scientific Inc., Canton, MI), and cholesterol (Thermo Scientific, Middletown, VA) were measured using the indicated colorimetric assays. Leptin (R&D Systems, Minneapolis, MN) and insulin (Crystal Chem, Downers Grove, IL) were measured by ELISA. Treatment effect on food intake and energy expenditure: Food intake was performed as described [15]. In brief, mice were fasted for 4 h, dosed with drug or vehicle 30 min before lights off, and access to chow was provided at lights out. Food intake over the next 2 h was measured. Energy expenditure at thermoneutrality was measured as described [15]. The ratio of post-dosing (120e240 min) to predosing (À150 to À30 min) energy expenditure was calculated. Telemetric monitoring of body temperature: Core body temperature (Tb) and activity were continuously measured by telemetry using G2 E-mitters implanted intraperitoneally, ER4000 energizer/receivers, and VitalView software (Starr Life Sciences, Oakmont, PA) with data collected each minute. Only data collected at the same time were compared, to control for variation in environmental temperature, noise, and disturbances in the vivarium. We re-analyzed prior data [26] to identify a metric that distinguishes the WT and Brs3 null phenotypes. Using 24-hour (or multiples thereof) datasets with Tb measured every minute, the range (defined as the difference between the 95th and 5th percentiles) and the standard deviation were both sensitive and robust in discriminating WT and Brs3 null mice (Table S1). These metrics intrinsically minimize the effect of Tb accuracy errors, such as due to telemeter placement or calibration. The metric was not improved by selective analysis of light vs dark phase and/ or active vs resting intervals.
Quantitative PCR and RT-PCR: Tissue DNA and RNA were extracted (Qiagen Allprep DNA/RNA micro Kit, Germantown, MD). RNA was reverse transcribed (Roche Transcriptor High Fidelity cDNA Synthesis Kit, Indianapolis, IN). DNA and cDNA were quantified by real-time polymerase chain reaction (q-PCR, Applied Biosystems 7900HT, Foster City, CA) using SYBR green, normalized to 18S RNA. Experimental design and statistical analysis: Data are presented as mean AE SEM, and were analyzed by t-test, one-way, or two-way repeated measures ANOVA with Holm-Sidak post-hoc testing, using Sigmaplot 12.5 (Systat Software Inc) or Prism v7.03 (GraphPad). Statistical significance was defined as 2-tailed P < 0.05.

RESULTS
3.1. Generation of Brs3 fl/y mice To enable conditional deletion of Brs3, we generated a floxed allele (Brs3 fl ) by placing Cre recombinase recognition loxP sites flanking exon 2, which is required for Brs3 function [23e25] (Figure 1AeC). Insertion of loxP sites had no effect on body weight (not shown), indicating that these sequences are not deleterious.

Selective inactivation of
Brs3 by Vglut2-Cre reproduces the global knockout phenotype We hypothesized that neuronal BRS-3 was responsible for regulation of body weight and energy homeostasis and chose to gain more information than a pan-neuronal knockout would provide. Since Brs3 is expressed in both glutamatergic [17] and GABAergic ( [41] and unpublished observations) neurons, we used Brs3 fl/y mice to examine the effects of germline loss of Brs3 in glutamatergic neurons expressing Vglut2 or in GABAergic neurons expressing Vgat. The level of Brs3 deletion was quantified in hypothalamic tissue. In Brs3 fl/y ;Vglut2-Cre mice, we detected 37% lower Brs3 DNA levels and 46% lower Brs3 RNA levels ( Figure 1D). In Brs3 fl/y ;Vgat-Cre mice, there was 27% reduction of Brs3 DNA and 59% reduction in Brs3 mRNA levels ( Figure 1E). Thus, Cre recombinase successfully deleted the floxed Brs3 exon, indicating that Brs3 is expressed in Vglut2 and Vgat neurons in the hypothalamus. On a chow diet, body weight, fat mass, and lean mass of Brs3 fl/y ;Vglut2-Cre mice were remarkably similar to that of the Brs3 fl/y control mice (Figure 2AeC, Figure S1AeC). However, the matched body weight and composition were achieved with a 9.9% lower food intake and a 10.3% lower average metabolic rate over 10 weeks ( Figure 2D,E). On a HFD, Brs3 fl/y ;Vglut2-Cre mice developed obesity, with increased body weight and fat mass, having a 13% increase in food intake ( Figure 2FeI). Expressed per mouse, the metabolic rate appeared slightly (7.7%) greater than in the controls ( Figure 2J); however, if higher body weight is considered, the metabolic rate was lower than seen in control mice of matched body composition (not shown). Consistent with their increased fat mass, Brs3 fl/y ;Vglut2-Cre HFD-fed mice had heavier BAT, iWAT, and liver, and increased body length ( Figure S1DeF).  An insulin tolerance test showed insulin resistance, although the glucose tolerance was not different from control mice (Figure 3AeE). Serum cholesterol was increased 25% and triglycerides levels were unchanged ( Figure 3F). Mild behavioral phenotypes have been reported in Brs3 -/y mice, including a greater preference for saccharine and aversion to quinine [13], increased meal size [24], less anxiety in an elevated plus maze test and decreased response to social isolation [42]. We measured preferences for sucrose, saccharin, and quinine, but did not detect a clear difference between reference Brs3 -/y and control mice ( Figure S2AeC). The Brs3 -/y mice may have a slightly less strong preference for high fat diet over chow ( Figure S2D). Brs3 -/y and control mice learned similarly that treats would be presented for a limited time each day ( Figure S2E and F). In anxiety assays (open field and elevated plus maze), no phenotype was detected in Brs3 -/y mice (Figure S2Ge J). We concluded that any differences observed in Brs3 -/y mice with these assays were not robust enough to warrant testing mice with selective Brs3 inactivation. Brs3 -/y mice have a lower core body temperature (Tb) during resting periods in the light phase, but their Tb is comparable to WT mice during active intervals in the dark phase [26]. The 24-hour mean Tb of Brs3 -/y mice is 0.2e0.3 C below controls, which groups of 6e12 are typically not powered to detect (see [23,26] and Table S1). However, the Brs3 -/y Tb phenotype is robustly captured as an increase in the range (defined as the difference between the 95th and 5th percentiles) or an increase in the standard deviation of continuous 24-hour body temperature data ( Figure 4A, Table S1, Methods 2.2). Brs3 deletion in Vglut2-Cre neurons caused the null phenotype ( Figure 4B).

3.3.
Brs3 deletion by Vgat-Cre does not affect adiposity or body temperature Unlike the heavier Brs3 fl/y ;Vglut2-Cre mice, the HFD Brs3 fl/y ;Vgat-Cre mice had body weights and fat and lean masses that were near to that of controls, with similar food intake and metabolic rate (Figure 2PeT, Figure S1JeL). Brs3 fl/y ;Vgat-Cre mice on a HFD showed a slight worsening of glucose tolerance (Figure 3GeI), but no difference from controls in insulin tolerance (Figure 3JeK), serum cholesterol and triglyceride levels ( Figure 3L), or tissue adiposity ( Figure S1JeL). On a chow diet, Brs3 fl/y ;Vgat-Cre mice were slightly leaner than controls at 14e24 weeks of age, but were not different at 52 weeks (Figure 2Ke O, Figure S1GeI). Brs3 deletion in Vgat-Cre neurons had no effect on Tb ( Figure 4C). These data suggest that Brs3 in Vgat neurons makes little contribution to the food intake, energy expenditure, Tb, and body weight phenotypes observed in the global Brs3 null mice.
3.4. Loss of response to BRS-3 agonist in Brs3 fl/y ;Vglut2-Cre, but not Brs3 fl/y ;Vgat-Cre, mice The response to the BRS-3 agonist MK-5046 was studied in the chowfed Brs3 fl/y ;Vglut2-Cre and Brs3 fl/y ;Vgat-Cre mice. The Brs3 fl/y ;Vglut2-Cre mice were heavier than the controls and ate less in the acute food intake assay ( Figure 5A,B). MK-5046 reduced food intake in control mice and had no effect in Brs3 fl/y ;Vglut2-Cre mice ( Figure 5A). In contrast, the melanocortin agonist MTII reduced food intake similarly in the two cohorts (63% vs 65%) ( Figure 5B). Importantly, MK-5046 increased energy expenditure in control, but not in Brs3 fl/y ;Vglut2-Cre mice ( Figure 5C). In Brs3 fl/y ;Vgat-Cre mice, both MK-5046 and MTII reduced food intake similarly to their effects in controls ( Figure 5D,E). MK-5046 increased energy expenditure with a similar magnitude as in controls, although this was not statistically significant ( Figure 5F). Taken together, these data demonstrate that Brs3 expression in Vglut2, but not Vgat, neurons is necessary for proper regulation of food intake, metabolic rate, adiposity, adiposity sequelae (insulin resistance), and response to BRS-3 agonist.
3.5. Generation of Brs3 loxTB/y mice We next investigated if selective Brs3 re-expression was sufficient to reverse the phenotype of global Brs3 null mice. To do this, we generated a Brs3 allele with a floxed transcription-blocking sequence, which is removed by Cre recombinase to re-express Brs3 ( Figure 6A,B). To study the Brs3 re-expression, we generated five genotypes: the new allele (Brs3 loxTB/y ), alone and with Vglut2-Cre or Vgat-Cre, WT control (Brs3 þ/y ), and a previously characterized Brs3 KO (Brs3 -/y ) [24].

Selective re-expression of Brs3
Successful re-expression of Brs3 was confirmed in hypothalamic tissue. In Brs3 loxTB/y ;Vglut2-Cre mice, 30% recombination of Brs3 alleles occurred ( Figure 6C). This caused re-expression of Brs3 mRNA to 14% of the level in WT mice ( Figure 6D). In Brs3 loxTB/y ;Vgat-Cre mice, there was 24% recombination, producing re-expression at 24% of total WT Brs3 mRNA levels. No Brs3 mRNA was detected in Brs3 loxTB/y mice in the absence of Cre recombinase. These results confirm that the loxTB-Brs3 allele is a null in the absence of Cre and expresses Brs3 once recombined. On a HFD, reference Brs3 -/y KO mice were obese compared to WT mice with increased food intake and greater body weight and adiposity (Figure 7AeE). The Brs3 loxTB/y mice were similarly obese and hyperphagic as the reference Brs3 -/y mice, confirming that unrecombined Brs3 loxTB is a null allele. The re-expression of Brs3 in Brs3 loxTB/y ;Vglut2-Cre mice prevented the null phenotype of null mice, rescuing the body weight, fat mass, lean mass, food intake, and energy expenditure to levels that were remarkably similar to the WT mice. In contrast, re-expression of Brs3 in Brs3 loxTB/y ;Vgat-Cre mice had no effect, with the body weight, fat mass, and lean mass being similar to that of null (both Brs3 -/y and Brs3 loxTB/y ) mice. Consistent with the adiposity changes, Brs3 -/y , Brs3 loxTB/y and Brs3 loxTB/y ;Vgat-Cre mice had heavier BAT, iWAT and liver, longer body length, although they had decreased eWAT, compared to WT and Brs3 loxTB/y ;Vglut2-Cre mice (Figure 8JeL). Glucose homeostasis was evaluated in the HFD cohort. In each case the more obese mice (Brs3 -/y , Brs3 loxTB/y , and Brs3 loxTB/y ;Vgat-Cre) had similar values, which were different from the leaner mice (WT and Brs3 loxTB/y ;Vglut2-Cre). The obese mice had higher fasting glucose and insulin levels, glucose intolerance, and insulin resistance by insulin tolerance test (Figure 8AeF). The obese mice had higher leptin and cholesterol and lower free fatty acids (Figure 8GeI). On a chow diet, the phenotype was milder than in the HFD cohort, with a smaller increase in body weight and fat mass, and food intake and metabolic rate (Figure 7FeJ). Measures of glucose homeostasis were consistent with the adiposity changes, with the relatively obese groups (Brs3 -/y , Brs3 loxTB/y and Brs3 loxTB/y ;Vgat-Cre) having a tendency to impaired glucose and insulin tolerance and increased serum leptin levels, compared to the leaner groups (WT and Brs3 loxTB/y ;Vglut2-Cre). No consistent changes in serum TG, cholesterol, and FFA were observed ( Figure S3). The reduction in food intake caused by MK-5046 was similar to that in WT mice in Brs3 loxTB/y ;Vglut2-Cre, but lost in Brs3 loxTB/y mice ( Figure 8M). Suppression of food intake by melanocortin agonist MTII Figure 4: Effect of genotype on core body temperature. Tb was measured by telemetry, sampling for 24 h in singly-housed, free-ranging mice in their home cage. Ambient temperature was w22 C and each comparison is between groups studied simultaneously. We define the Tb range as the difference between the 95th and 5th percentiles. (A) Re-analysis of Tb in WT and global Brs3 -/y mice from [26] showed a 0.54 C larger range in the global Brs3 -/y mice compared to controls (p ¼ 0.00008). (B) Brs3 fl/y ;Vglut2-Cre (Vglut2-Cre) mice had a 0.59 C larger range than littermate Brs3 fl/y (flox/y) mice (p ¼ 0.008). (C) Brs3 fl/y ;Vgat-Cre (Vgat-Cre) and littermate Brs3 fl/y (flox/y) mice showed no difference in Tb range (p ¼ 0.78). Additional analyses of the Tb phenotype are in Table S1.  Treatment is compared to pre-injection baseline. (DeF) Brs3 fl/y ;Vgat-Cre (Vgat-Cre) and littermate control Brs3 fl/y (flox/y) mice were studied exactly as described in (AeC) (N ¼ 10e12/group for food intake studies, 6/group for energy expenditure study). * indicates P < 0.05 drug vs vehicle, within genotype by paired t-test. # indicates P < 0.05 vehicle between genotypes by unpaired t-test.

DISCUSSION
Determining the functions of Brs3 in mammals has been hampered by the lack of an identified endogenous ligand. This orphan status precludes the usual paradigm for unraveling a receptor's physiologydhistorically, having a hormone/ligand paved the way to identification of its cognate receptor, which often occurred after much of the biology was understood. In contrast, most of our knowledge to date about Brs3 has been acquired from genetics, by studying global KO mice. To refine the genetic approach, we developed mice that allow Cre-dependent selective ablation or re-expression of Brs3. Using these tools, we find that Brs3 expression in Vglut2 neurons is both necessary and sufficient for the identified functions of Brs3. Our observations focus on Brs3 regulation of metabolic rate, body temperature, food intake, and body weight. We also attempted to study heart rate because resting heart rate is lower in global Brs3 KO mice [27]. Indeed, with 4e6 mice/group, there was a hint that the heart rate effect is also mediated via Brs3 in Vglut2 neurons. However, these experiments are underpowered, so formally this remains unaddressed. Behavioral phenotypes have been reported for global Brs3 KO mice [13,42]. In our assays, the behavioral alterations of the global null mice were not sufficiently robust and reproducible for us to undertake behavioral characterization of the selective mutant mice. We also did not explore the regulation of glucose-dependent insulin secretion by Brs3, an action at the beta cell of the pancreatic islets [20]. A caveat is that our studies are loss of function or reconstitution of function, which are insensitive to redundant Brs3 functions. Studies combining BRS-3 agonist treatment with genetics could uncover additional functions of Brs3. Overall, the results are reassuringly concordant with prior phenotyping of the global null. One surprise was the lack of obesity in singly-housed, chow-fed Brs3 fl/y ;Vglut2-Cre mice. These mice did have a reduced metabolic rate but also ate less and, thus, did not become obese. Note that we would not have identified a phenotype in these mice without measuring long-term food intake. Interestingly, a cohort of group-housed, chow-fed Brs3 fl/y ;Vglut2-Cre mice did become obese (32.2 AE 1.4 g vs 26.4 AE 0.7 g at 13 weeks). It is possible that single vs group housing is the relevant difference, but we have not pursued this observation further. Brs3 mRNA and/or immunoreactivity is present in the CNS, but it is reported that BRS-3 is also found in peripheral sites, including muscle, islet beta cells, testis, ovary, uterus, lung, and cancers including carcinoid/neuroendocrine [5,20e22] (https://gtexportal.org/home/ gene/BRS3). The observation that Brs3 expression in Vglut2 neurons is sufficient to reverse the global null phenotype, demonstrates that Brs3 in this subset of neurons mediates the non-redundant activities of Brs3. Developing drug treatments for obesity has proven to be a very difficult task. Multiple nuclei in the brain integrate peripheral and central signals and send messages elsewhere in the brain and peripherally to regulate energy homeostasis. Because receptors often have different functions in different brain regions, even if complete molecular target selectivity is achieved with a drug, multiple drug effects are likely. To avoid this problem, there has been a hope that BRS-3 agonists might have anti-obesity effects via pharmacologic actions outside of the brain, avoiding some side effects [28,31,43]. However, our results strongly indicate that BRS-3 agonists achieve anti-obesity efficacy via activating glutamatergic neurons, likely chiefly in the brain.  Table S2. A potential confounder of our experiments is that Cre activity during development will cause gene deletion, even if the Cre is not present in the mature cells [44]. While we cannot completely rule this out, the complementarity of the deletion and re-expression phenotypes and the measured DNA deletion efficiencies indicate a role for Brs3 in mature Vglut2 neurons. Note that Brs3 also appears to be expressed in non-glutamatergic hypothalamic neurons, including GABAergic cells, because Brs3 mRNA is only partially reduced in Brs3 fl/y ;Vglut2-Cre mice and re-expressed in Brs3 loxTB/y ;Vgat-Cre mice. The functions of Brs3 in the non-glutamatergic cells remain to be determined. We have identified Brs3 neurons that express Vglut2 as functionally important in energy homeostasis. While many neurons express Vglut2, relatively few express Brs3. Brs3-expressing neurons are located in, among other brain regions, the preoptic area (POA), paraventricular nucleus of the hypothalamus (PVH), dorsomedial hypothalamus (DMH), medial amygdala, and parabrachial nucleus (PBN) [12e19]. Using in situ hybridization, Zhang et al. reported that many Brs3 neurons are glutamatergic, while few are GABAergic [17]. We (unpublished observations) and others [41] have also found Brs3 expression in GABAergic neurons. The RT-PCR data indicates that Brs3 is expressed in both Vglut2 and Vgat neurons in the hypothalamus. Brs3 may also have functions in Vgat neurons, with a slight reduction in body weight on chow diet and slightly impaired glucose tolerance on HFD in Brs3 fl/y ;Vgat-Cre mice.
We have speculated that low levels of a BRS-3 endogenous agonist signals an energy deficit, either acute or chronic (possibly analogous to cholecystokinin or leptin, respectively). Thus BRS-3 ligand deficiency signals an underfed state and evokes the physiological responses appropriate to that state [29]. A next research goal is elucidation of the specific glutamatergic Brs3 neurons responsible for regulation of food intake, metabolic rate, body temperature, and adiposity. Nuclei expressing Brs3 and also having glutamatergic neurons that inhibit food intake include the PVH [45], PBN [46], and lateral hypothalamus [47]. Glutamatergic neurons that activate BAT and increase energy expenditure and Tb are located in the DMH, PVH, and PBN [48e50]. In the POA, the best-studied thermoregulatory glutamatergic neurons decrease Tb but do not increase it [41,51]. Functional experiments, such as with optogenetics and chemogenetics, are required to further localize the specific nuclei and circuits where Brs3 regulates the different aspects of energy homeostasis.