Lateral Hypothalamic Mc3R-Expressing Neurons Modulate Locomotor Activity, Energy Expenditure, and Adiposity in Male Mice

Abstract

The central melanocortin system plays a crucial role in the control of energy balance. Although the decreased energy expenditure and increased adiposity of melanocortin-3 receptor (Mc3R)–null mice suggest the importance of Mc3R-regulated neurons in energy homeostasis, the roles for specific subsets of Mc3R neurons in energy balance have yet to be determined. Because the lateral hypothalamic area (LHA) contributes to the control of energy expenditure and feeding, we generated Mc3r^cre mice to determine the roles of LHA Mc3R (Mc3R^LHA) neurons in energy homeostasis. We found that Mc3R^LHA neurons overlap extensively with LHA neuron markers that contribute to the control of energy balance (neurotensin, galanin, and leptin receptor) and project to brain areas involved in the control of feeding, locomotion, and energy expenditure, consistent with potential roles for Mc3R^LHA neurons in these processes. Indeed, selective chemogenetic activation of Mc3R^LHA neurons increased locomotor activity and augmented refeeding after a fast. Although the ablation of Mc3R^LHA neurons did not alter food intake, mice lacking Mc3R^LHA neurons displayed decreased energy expenditure and locomotor activity, along with increased body mass and adiposity. Thus, Mc3R neurons lie within LHA neurocircuitry that modulates locomotor activity and energy expenditure and contribute to energy balance control.

Obesity is a major global public health concern and is associated with pathological disorders, including type 2 diabetes, hypertension, cardiovascular disease, and cancer (1). Development of effective antiobesity therapies requires an understanding of the physiologic systems that regulate energy homeostasis. In this regard the melanocortin system is critical for energy balance via its regulation of food intake and energy expenditure (2). The central melanocortin system consists of the endogenous melanocortin receptor agonist proopiomelanocortin (POMC), the endogenous melanocortin receptor antagonist agouti-related protein (AgRP), and a family of five seven-transmembrane G protein–coupled melanocortin receptors (Mc1R to Mc5R) (2, 3). Mc3R and Mc4R are the predominant melanocortin receptors in the central nervous system (CNS). Mc4Rs are expressed widely in the CNS, and Mc4R mutations are associated with severe hyperphagic obesity in both rodents and humans (4–6). Mc3Rs are expressed in the hypothalamus and limbic structures of the brain that regulate feeding behaviors and autonomic function (7, 8), and Mc3r gene polymorphisms have been linked to childhood obesity (9–11). Deletion of Mc3r in mice increases fat mass, reduces lean mass, and increases susceptibility to diet-induced obesity. Loss of Mc3r also results in reduced locomotor activity and voluntary wheel running, implicating Mc3R-expressing cells in the regulation of locomotion (12–17). Selective re-expression of Mc3Rs in the CNS of an Mc3r-null background partially rescues the obesity and activity defects accompanying global Mc3r inactivation (16); although this indicates an important contribution of Mc3R-bearing neurons in energy balance control, the neural substrates and physiologic mechanisms by which these effects are mediated are not fully known.

Within the hypothalamus, Mc3R is expressed in the arcuate nucleus (ARC), the ventromedial nucleus of the hypothalamus (VMH), the lateral hypothalamic area (LHA), and the preoptic nucleus (7). The LHA is of particular interest, as it has long been appreciated for its crucial roles in the regulation of feeding, body weight maintenance, and metabolism. LHA lesions lead to reduced feeding, drinking, and body weight (18). Alternatively, electrical stimulation or glutamate receptor activation within the LHA potently stimulates food intake (19–21). LHA neurons are heterogeneous in regard to neuropeptide and neurotransmitter content and project widely throughout the CNS (22). Genetic inactivation of LHA glutamatergic neurons is associated with increased feeding and subsequent weight gain on a high-fat diet (HFD) (23), whereas activation of LHA γ-aminobutyric acid (GABA)ergic neurons drives consummatory behaviors with a positive valence (24). Multiple LHA cell types, including neurons expressing melanin-concentrating hormone (MCH), orexin (OX)/hypocretin, galanin, leptin receptors (LepRb), and neurotensin (Nts), have been implicated in energy balance control (25). MCH stimulates feeding and contributes to the long-term regulation of body weight, as MCH knockout mice are hypophagic and lean relative to controls (26). A subset of Nts-expressing LHA cells also expresses the LepRb, and leptin action in Nts^LHA neurons contributes to body weight regulation and reward behaviors (27). Nts^LHA neurons project to the ventral tegmental area (VTA) and modulate locomotor activity through the mesolimbic dopamine system (28–30). Additionally, LepRb/Nts-expressing LHA cells indirectly regulate local OX neurons, which may also contribute to feeding and locomotor activity control (25, 31). Subsets of galanin cells in the LHA (Gal^LHA) express LepRb and Nts, and Gal^LHA neurons contribute to energy balance control by modulating noncompulsive locomotion, energy expenditure, and food reward, but not consummatory behaviors per se (32, 33). Gal^LHA cells do not project to the VTA (34), but galanin has been shown to directly regulate the activity of OX neurons, presumably through local release in the LHA (31). Thus, LHA neurons marked by Nts, LepRb, and galanin contribute to energy balance control by modifying feeding behavior, food reward, and locomotor activity at least in part by modulation of VTA cells and OX neurons.

In the context of the melanocortin system, the LHA receives projections from both POMC and AGRP/NPY neurons (35), and overexpression of a melanocortin receptor antagonist in the LHA promotes obesity in response to an HFD (36). Additionally, optogenetic activation of AgRP terminals in the LHA stimulates feeding (37), suggesting a role for melanocortin receptor–bearing LHA cells in feeding behavior. Although re-expression of Mc4Rs only in the LHA of an otherwise null Mc4r background alters glucose homeostasis and sympathetic output, it does not alter feeding (38). The physiological importance of Mc3R^LHA neurons in the regulation of feeding, locomotor activity, and energy expenditure is largely unknown. To reveal the roles of Mc3R^LHA neurons in energy homeostasis, we generated a knock-in transgenic line that expresses Cre recombinase in Mc3R-containing cells (Mc3r^cre) and applied genetic and viral tools to identify and manipulate Mc3R^LHA neurons and determine their contributions to energy balance.

Materials and Methods

Experimental animals

Mc3R-T2A-Cre mice (Mc3r^cre) were generated using recombineering techniques as previously described (39). Briefly, the stop codon of Mc3r was replaced with a T2A-Cre fusion transgene that effectively tethers Cre recombinase to the Mc3r gene product via an 18–amino acid self-cleaving peptide sequence from the Thosea asigna virus. The final targeting construct containing the Mc3R-T2A-Cre fusion and 4 kb of flanking genomic sequence was electroporated into embryonic stem cells followed by neomycin selection. Appropriately targeted clones were identified by quantitative PCR and confirmed by Southern blot analysis. Targeted clones were expanded and injected into blastocysts by the University of Michigan Transgenic Core. Chimeric offspring were then bred to confirm germline transmission of the Mc3r^cre allele; the neomycin selection cassette was removed by breeding to a Flpe deleter strain (The Jackson Laboratory stock no. 003946). Offspring were genotyped using primers that spanned the stop codon of the Mc3r locus and internal to the Cre transgene sequence. Mc3r^cre mice were then bred to the Cre-dependent Ai9 Tomato reporter strain (The Jackson Laboratory stock no. 007909) to visualize Cre recombinase activity by fluorescent immunohistochemistry. Mc3r^cre × Ai9-TdTomato mice (Mc3r^cre-Tomato) were bred to the Gad1-GFP and Galanin-GFP backgrounds (33, 40) to generate animals expressing tdTomato in Mc3R neurons and GFP in Gad1- or Galanin-expressing cells, respectively. All mice were housed in a 12-hour light/12-hour dark cycle and cared for by the Unit for Laboratory Animal Medicine at the University of Michigan. Animals had ad libitum access to food and water, except in experiments when mice were fasted before perfusion or for timed feeding studies. All animal care and procedures were in accordance with the guidelines of, and approved by, the University of Michigan’s Committee on the Care and Use of Animals.

RNAscope in situ hybridization assays

Following rapid decapitation of mice under anesthesia, brains were flash frozen on 2-methylbutane at −20°C and stored at −80°C. The brains were cut to thickness of 16 μm on a cryostat and thaw mounted onto Superfrost Plus slides (Fisher Scientific). Slides were stored at −80°C until further processing. In situ hybridization was performed according to the RNAscope^® 2.5 HD duplex detection kit user manual for fresh frozen tissue (Advanced Cell Diagnostics). Probes used are as follows: RNAscope^® Probe-Mm-Mc3r-C1 (catalog no. 412541), RNAscope^® Probe-Mm-Mc4r-C2 (catalog no. 319181-C2), and RNAscope^® Probe-Cre-C2 (catalog no. 312281-C2).

Perfusion and immunohistochemistry

For some experiments, peptide colocalization required treatment with ICV colchicine (10 μg) 2 days prior to perfusion to trap neuropeptides in the cell soma. For perfusion, adult mice were euthanized with an overdose of IP pentobarbital injection and transcardially perfused with sterile PBS, followed by neutral buffered 10% formalin. Brains were removed, postfixed, and dehydrated in 30% sucrose before sectioning as 30-μm coronal slices using a freezing microtome (Leica). Coronal brain sections were collected as four representative series and stored at −20°C until further use.

For Fos immunohistochemistry (IHC), sections were pretreated with 0.3% H₂O₂, blocked in 3% normal donkey or goat serum and then incubated with primary antibodies for c-Fos (41) (1:10,000). Detection of primary Fos antibody was performed by the avidin-biotin/diaminobenzidine (DAB) method [Biotin-SP–conjugated donkey anti-rabbit; 1:200; Jackson ImmunoResearch Laboratories (42); ABC kit, Vector Laboratories; DAB reagents, Thermo Scientific). For hM3Dq–designer receptors exclusively activated by designer drug (DREADD)–mCherry and tract-tracing experiments, immunostaining was performed using primary antibody anti-dsRed (43) (Living Colors, rabbit, 1:1000), followed by secondary immunofluorescence detection with donkey anti-rabbit Alexa Fluor 568 (44) (1:200; Invitrogen). Peptide colocalization was performed using primary antibodies for anti-GFP (45) (Invitrogen A6455; rabbit; 1:4000), anti-Nts (46) (Millipore; rabbit; 1:1000), anti-OX (47) (Santa Cruz Biotechnology; goat; 1:1000), and anti-MCH (48) (Santa Cruz Biotechnology, goat, 1:1000) followed by secondary immunofluorescent detection with species-specific Alexa Fluor 488 or 568 antibodies (49, 50) (Invitrogen; 1:200) and processed for imaging. To identify neurons responsive to leptin, fasted mice were IP injected with sterile PBS or leptin (5 mg/kg) 2 hours before perfusion. For phosphorylated signal transducer and activator of transcription-3 (p-Stat3) IHC, sections were pretreated with 1% NaOH and 1% H₂O₂, 0.3% glycine, and 0.03% SDS, and then blocked with normal donkey or goat serum and incubated with primary antibody (51) (anti–p-Stat3; Cell Signaling Technology; rabbit; 1:1000). Detection of primary p-Stat3 antibody was done using the DAB method.

Stereotaxic virus injections

Adult (>8-week-old) Mc3r^cre mice were anesthetized with vaporized isoflurane and fixed in a stereotaxic apparatus. Coordinates used to target the LHA were A/P, −1.34 mm; M/L, ±1.13 mm; and D/V, −5.2 mm relative to bregma. Viral injections (200 nL) were performed using a pressurized Picospritzer system coupled to a pulled glass micropipette. After 5 minutes to allow for tissue absorption of virus, the glass micropipette was removed, the cranial access was filled with bone wax, and the surgical incision was closed with sutures. Mice were allowed to recover for at least 10 days before they were used in physiologic experiments. At the conclusion of all experiments, mice were perfused and brains were harvested for IHC to confirm appropriate viral targeting. Adenovirus Ad-in synaptophysin-mCherry was prepared by the University of Michigan Viral Vector Core with a titer of 2.7 × 10¹² genome copies/mL. Adeno-associated viruses (AAVs) used in these studies were purchased from the University of North Carolina Viral Vector Core with viral titers ranging from 2 to 5.9 × 10¹² genome copies/mL and included (i) Cre-inducible AAV8-Syn-DIO-hM3Dq-mCherry, (ii) Cre-inducible rAAV2-Flex-taCasp3-TEVP, and (iii) AAV8-CAG-flexGFP.

Energy balance studies

Acute neuronal activation studies

After recovery from bilateral stereotaxic delivery of Cre-dependent hM3Dq-DREADD-mCherry virus to the LHA, Mc3r^cre mice were habituated to daily IP injections for 5 days with sterile PBS prior to food intake assessments following DREADD-mediated activation of Mc3R^LHA neurons. To assess refeeding after a fast, food was removed at the onset of the dark cycle (6:00 pm) the day before the test. On the morning of the second day, mice were injected IP with 0.5 mL of vehicle or clozapine N-oxide (CNO; 0.3 mg/kg) and food was returned to the cage. Cumulative food intake was measured 2, 4, and 8 hours following injection. Each mouse was used as its own control using a cross-over experimental design to exclude batch effects.

Chronic neuronal inhibition studies

Cohorts of male Mc3r^cre mice were injected bilaterally in the LHA with 200 nL of AAV2-flex-taCasp3-TEVp virus (52). Weekly body weight and food intake were subsequently measured during the ensuing 12 weeks. For acute refeeding studies, food was removed at the onset of the dark cycle (6:00 pm) the night before the test; the next day, food was given back and cumulative food intake was measured at 2, 4, and 8 hours. At the end of the study, fat depots were collected and weighed at the time of perfusion. Brains were harvested following perfusion and processed by IHC to confirm appropriate targeting and cell ablation.

Metabolic and behavioral profiling

Energy expenditure and body composition

Energy expenditure and X-ambulatory activity were measured using a comprehensive laboratory monitoring system (Columbus Instruments) in the University of Michigan Small Animal Phenotyping Core. The comprehensive laboratory monitoring system housing maintained 12-hour dark/12-hour light cycles and an ambient temperature of 20 to 23°C. Prior to the start of data acquisition, mice were acclimatized to the sealed chambers for 2 days with free access to food and water before data were collected. Body fat and lean mass were determined using an nuclear magnetic resonance–based analyzer (Minispec LF90II; Bruker Optics) in the University of Michigan Small Animal Phenotyping Core.

Open-field activity

Separate cohorts of Mc3r^cre mice injected bilaterally in the LHA with the Cre-dependent stimulatory hM3Dq-DREADD-mCherry virus were used in an open-field activity test. Mice were removed from their home cages during the light cycle and acclimated for 2 hours in an open-field arena (ENV-017M; Med Associates) without access to food or water. After acclimation, open-field activity was recorded for 30 minutes at baseline, 30 minutes after saline injection, and 90 minutes after CNO injection (0.3 mg/kg, IP). To determine the importance of the mesolimbic dopamine system in these responses, additional animals were pretreated with the dopamine receptor-1 antagonist SCH23390 (0.1 mg/kg, IP) 30 minutes before CNO administration. Data are pooled for each genotype and treatment and expressed as total ambulatory movements. At the conclusion of the experiment, mice were perfused and brains were harvested and processed by IHC to confirm appropriate targeting of the AAV- hM3Dq-DREADD-mCherry in the LHA. Animals with hM3Dq-DREADD-mCherry expression outside the LHA were excluded from analysis.

Statistical analysis

Data were analyzed and graphs were generated using GraphPad Prism. Paired t tests and two-way ANOVAs were used to determine significance of the data collected and are noted in the figure legends. Results were considered significant with a P value <0.05 (*P < 0.05, **P < 0.01, ***P < 0.001).

Results

Mc3r^cre knock-in mice allow manipulation of Mc3R-expressing cells

To genetically identify and selectively manipulate Mc3R neurons, we generated knock-in mice that express Cre recombinase in Mc3R neurons (Mc3r^cre) [Fig. 1(a)]. To confirm the specific expression of Cre in Mc3R neurons, we crossed Mc3r^cre mice with the Ai9 Td-tomato reporter line and found tomato reporter expression in all brain regions known to express Mc3Rs [Fig. 1(b)]. To confirm that the expression of Cre activity recapitulates endogenous Mc3r expression in vivo, we used an RNAscope dual chromogen assay to detect Mc3r and Cre mRNAs. RNAscope in situ hybridization demonstrated nearly complete colocalization of Mc3R and Cre mRNAs in the VMH, ARC, and LHA [Fig. 1(c)–1(e)]. Melanocortin agonists and antagonists (POMC and AGRP peptides, respectively) are produced by neurons in the ARC and these neurons project to the LHA. To determine whether melanocortin peptides might target Mc3R^LHA neurons, we stained Mc3r^cre-Tomato brains for both AGRP and POMC immunoreactivities. Immunostaining revealed dense AgRP fibers in the LHA with AGRP boutons lying in close proximity to Mc3R^LHA neurons [Fig. 1(f)]. We also found dense POMC fiber boutons in apposition to Mc3R^LHA cell bodies [Fig. 1(g)], indicating that Mc3R^LHA neurons are positioned to receive and transmit melanocortin signals through LHA neurocircuitry that modulates energy balance and behavior.

Figure 1.

(a) Schematic of Mc3r^cre allele and cre-mediated Td-Tomato expression in Mc3R-expressing cells of Mc3r^cre-Tomato mice. (b) Immunofluorescent detection of Td-tomato (red) in the LHA, VMH, and ARC of Mc3r^cre-Tomato mice. (c–e) RNAscope dual chromogen staining shows colocalization of cre transcripts (red) with Mc3R- expressing neurons (blue) in the (c) VMH, (d) ARC, and (e) LHA from Mc3R^cre mice. (f) AgRP (green) immunostaining in the LHA of Mc3r^cre-Tomato (red) mice. (g) POMC (green) immunostaining in the LHA of Mc3r^cre-Tomato (red) mice. Scale bars: (b) 1 mm, (c–e) 20 μm, and (f and g) 100 μm.

Neurochemical characterization of LHA Mc3R-expressing neurons

As multiple LHA cell types have been shown to contribute to energy balance regulation (22, 53), we used Mc3r^cre × Ai9-TdTomato mice (Mc3r^cre-Tomato) to examine the expression of known neuropeptides in Mc3R^LHA neurons. Similar to LepRb-, Nts-, and galanin-expressing LHA neurons, we found little colocalization of Mc3R^LHA neurons with OX or MCH [Fig. 2(a) and 2(b)]. However, ∼45% of Mc3R^LHA neurons coexpressed the neurotransmitter Nts [Fig. 2(c)], whereas ∼28% of Mc3R^LHA neurons contain galanin [Fig. 2(d)]. To determine whether Mc3R^LHA neurons contain leptin receptors (LepRb), we treated fasted Mc3r^cre-Tomato mice with leptin and examined the leptin-induction of p-STAT3; we found that ∼42% of Mc3R^LHA neurons contained leptin-stimulated p-STAT3 [Fig. 2(e)]. Most Mc3R^LHA neurons appear to be GABAergic based on the extensive colocalization of Mc3r^cre-Tomato LHA neurons with GFP in a GAD1-GFP background, which labels GAD1-expressing GABAergic neurons (40) [Fig. 2(f)]. The overlap of Mc3R^LHA neurons with Nts, galanin, and LepRb suggests that Mc3R^LHA neurons lie within LHA neural circuits that modulate feeding, energy expenditure, and locomotor activity and that their activity may contribute to these energy balance parameters.

Figure 2.

(a and b) Immunofluorescent detection of Td-tomato (red) and OX [(a) green] or MCH [(b) green] in the LHA of Mc3r^cre-Tomato mice. (c) Immunofluorescent detection of Td-tomato (red) and Nts (green) in the LHA Mc3r^cre-Tomato mice 48 h after colchicine treatment. (d) Immunofluorescent detection of Td-tomato and GFP in Mc3r^cre-Tomato, Galanin^GFP mice. (e) Immunohistochemical detection of p-STAT3 (green) and Td-tomato (red) in the LHA of Mc3r^cre-Tomato mice 1 h after IP injection of leptin. (f) Immunofluorescent detection of Td-tomato (red) and GFP (green) in Mc3r^cre-Tomato, GAD1^GFP mice. Scale bars, 100 μm. Quantification of Mc3R colocalization shown in pie chart inset: red indicates percentage Mc3R neuron not colocalized; yellow indicates percentage Mc3R neuron colocalized with peptide of interest.

Determining the circuitry in which Mc3R^LHA neurons reside is essential for understanding how these neurons might contribute to metabolic function. To trace efferent projections from Mc3R^LHA neurons, we injected a Cre-dependent adenoviral synaptophysin:mCherry fusion construct unilaterally into the LHA of Mc3r^cre mice. The synaptophysin:mCherry fusion protein preferentially accumulates in neuron terminals, allowing identification of projection target areas (54) [Fig. 3(a)]. Unilateral synaptophysin:mCherry injection revealed dense terminal projections within the LHA itself [Fig. 3(b)]. Additionally, adenoviral synaptophysin:mCherry transduction of Mc3R^LHA neurons clearly labeled neural projection terminals in the paraventricular nucleus of the hypothalamus (PVH) and parabrachial nuclei (PBN) [Fig. 3(d) and 3(f)], brain regions known to regulate feeding (55). Mc3R^LHA cells also sent dense projections to the bed nucleus of the stria terminalis (BNST) and VTA [Fig. 3(c) and 3(e)], which play roles in reward seeking and addiction (56, 57). VTA neurons also play crucial roles in the control of locomotor activity and energy expenditure (29, 58, 59). [See Fig. 3(g) for injection site summary and Table 1 for relative projection densities to brain areas targeted by Mc3R^LHA neurons.] Activation of GABAergic-, Nts-, and Galanin-expressing LHA neurons have all been associated with alterations in locomotor activity, although galanin neurons do not project to the VTA. Although these populations do not overlap with OX neurons, LHA^LepRb neurons regulate OX neurons indirectly, and galanin has been shown to modulate OX neuron activity directly (31). These data suggest that local modulation of OX neuron activity by neighboring LHA cell types may also modify motivated behaviors and locomotor activity. Thus, the neurotransmitter content and projections of Mc3R^LHA neurons led us to hypothesize a role for these cells in the regulation of activity, energy expenditure, and energy homeostasis in part through the control of OX neurons and the VTA.

Figure 3.

(a) Schematic of cre-mediated expression of synaptophysin (syn)-mCherry in Mc3R^cre mice. (b) Immunofluorescent detection of mCherry in the LHA of Mc3R^cre mice following intra-LHA injection of Ad-syn-mCherry virus. (c–f) Immunofluorescent detection of syn-mCherry–containing projections in the (c) BNST, (d) PVH, (e) VTA, and (f) PBN after intra-LHA injection of Ad-syn-mCherry virus. Scale bars, 100 μm. (g) Summary of Ad-IN Syn-mCherry injection sites of Mc3R^Cre mice included in the study (n = 3). 3V, third ventricle; BNST, bed nucleus of the stria terminalis; lPBN, lateral parabrachial nucleus; PBN, parabrachial nuclei; PVH, paraventricular nucleus of the hypothalamus; scp, superior cerebellar peduncle; SN, substantia nigra.

Table 1.

Mc3R^cre LHA Projection Sites Using Cre-Dependent Viral Tracing (Ad-IN-syn-mCherry) in the LHA

Brain Area	Intensity
BNST	++++
Parabrachial nucleus, medial division, medial medial part	++++
PVH	+++
VTA	+++
Substantia nigra, reticular part	++
Periaqueductal gray	++
Nucleus of the solitary tract	+

Terminal projection sites and the relative intensity of projections in the different brain areas are shown.

Activation of Mc3R^LHA neurons increases locomotor activity and augments refeeding

To establish a role for Mc3R^LHA neurons in energy balance control, we used DREADDs technology to enable remote neuronal activation by the pharmacologically inert ligand CNO (60). The stimulatory DREADD (hM3Dq-DREADD-mCherry) couples through the Gq pathway to depolarize neurons and increase their firing rate in response to CNO. Mc3r^cre mice were injected stereotaxically in the LHA with the Cre-dependent, stimulatory AAV-hM3Dq-DREADD-mCherry vector (Mc3R^LHA-hM3Dq mice). Using Fos staining as an indicator for neuronal activation, we found that CNO treatment increased nuclear Fos immunoreactivity in Mc3R^LHA-hM3Dq mice, compared with vehicle controls [Fig. 4(a) and 4(b)], demonstrating that cell-specific DREADD expression permits the CNO-dependent activation of Mc3R^LHA neurons. A representation of LHA injection sites and viral transduction is found in Fig. 4(i). As we hypothesized that Mc3R^LHA cells may modulate the activity of OX neurons and cells of the VTA to exert their physiologic effects, we also examined Fos in OX neurons and the VTA. We found increased nuclear Fos staining [Fig. 4(c), 4(d), and 4(g)] in OX cells and in the VTA following Mc3R^LHA neuron activation [Fig. 4(e), 4(f), and 4(h)]. Because the activation of OX cells and the VTA is associated with alterations in locomotor activity and food intake under some circumstances (29, 61, 62), the increase in nuclear Fos expression seen following Mc3R^LHA activation suggests the potential modulation of these behaviors by Mc3R^LHA neuron activity.

Figure 4.

(a and b) Fos expression in the LHA of Mc3R^cre-DREADD mice after (a) vehicle or (b) CNO (0.3 mg/kg, IP) treatment; representative images of mCherry-IR (red) and Fos-IR (green). Scale bars, 100 μm. (c and d) Immunodetection of Fos (red) and OX (green) in the LHA of Mc3R^cre-DREADD mice injected with (c) vehicle or (d) CNO. Scale bars, 100 μm. (e and f) Immunodetection of Fos in the VTA of Mc3R^cre-DREADD mice injected with (e) vehicle or (f) CNO. Scale bars, 100 μm. (g) Quantification of OX and Fos colocalization in Mc3R^cre-DREADD mice treated with vehicle or CNO (SEM, n = 7; ****P < 0.0001). (h) Number of Fos-positive cells in the VTA of Mc3R^cre-DREADD mice treated with vehicle or CNO (SEM, n = 4; **P < 0.01). (i) Summary of AAV-3Dq-DREADD LHA injection sites of Mc3R^Cre mice included in the study (n = 9); injection sites are represented as two groups for clarity.

To determine the effect of pharmacologic Mc3R^LHA neuron activation on feeding, we analyzed the feeding responses of Mc3R^LHA-hM3Dq mice following Mc3R^LHA activation in both the fed state and following a fast. For the fasting–refeeding paradigm, mice were fasted overnight and then pretreated with vehicle or CNO before food was given back during the light phase. As shown in Fig. 5(a), the CNO-induced activation of Mc3R^LHA neurons in the fed state had little impact of feeding during an 8-hour period (vehicle, 0.52 ± 0.13 g; CNO, 0.49 ± 0.12 g). However, food intake following a fast was augmented with Mc3R^LHA activation within the first 2 hours of refeeding [Fig. 5(b)], indicating that Mc3R^LHA neuron activation can enhance feeding depending on metabolic state (fasted vs fed).

Figure 5.

(a) Food intake of Mc3R^cre-DREADD mice following vehicle or CNO treatment in the fed state (SEM, n = 9; **P < 0.01). (b) Food intake of Mc3R^cre-DREADD mice following vehicle or CNO treatment after a fast (SEM, n = 9; ***P < 0.001). (c) Locomotor activity following hM3Dq-mediated activation of Mc3R^LHA neurons in the absence of presence of by peripheral administration of a D1R antagonist, SCH23390 (0.1 mg/kg, IP). Activity (travel distance per 30 min) is binned for the 30 min after vehicle injection and averaged per 30 min for CNO treatment (SEM, n = 7; **P < 0.01). NS, not significant.

Activation of the mesolimbic dopamine system increases locomotor activity (63) and Nts action in the VTA is associated with increased locomotor activity and dopamine release in the nucleus accumbens (29). Because our viral tracing studies revealed Mc3R^LHA neuron projections into the VTA, we hypothesized that Mc3R^LHA neurons may directly modulate locomotor activity. To test this hypothesis, we determined changes in locomotor activity in Mc3R^LHA-hM3Dq mice in response to treatment with vehicle or CNO (0.3 mg/kg). CNO-mediated Mc3R^LHA neuron activation greatly increased the locomotor activity of these mice [Fig. 5(c)], suggesting a potentially important role for Mc3R^LHA cells in the control of activity. This increased locomotor activity was dopamine-dependent, as pretreatment with the dopamine receptor-1 antagonist SCH23390 significantly attenuated the locomotor response to Mc3R^LHA activation [Fig. 5(c)].

Mc3R^LHA neurons are required for energy homeostasis

To determine the contributions of Mc3R^LHA neurons to long-term energy homeostasis, we used the Cre-dependent virus AAV-flex-taCasp3-TEVp to conditionally express caspase and induce Mc3R^LHA cell death in adult animals (52). Injection of AAV-flex-taCasp3-TEVp into Mc3r^cre-Tomato mice selectively ablated Mc3R neurons in the LHA while leaving Mc3R neurons in the VMH and ARC essentially unperturbed [Fig. 6(a) and 6(b)]. Importantly, caspase ablation in the LHA was limited to Mc3R^LHA neurons, as cell counts within the LHA showed greatly decreased numbers of Mc3R^LHA neurons, whereas neighboring OX neurons were unaffected [Fig. 6(c)].

Figure 6.

(a and b) Immunodetection of Td-tomato in the Mc3r^cre-Tomato mice injected with (a) control or (b) AAV-flex-taCasp3-TEVp virus. (c) Quantification of Mc3r^cre-Tomato neurons and OX neurons in the LHA from Mc3r^cre-Tomato mice injected with control virus or AAV-flex-taCasp3-TEVp virus (AAV-Control, n = 5; AAV-Casp, n = 5). (d) Weekly food intake in Mc3r^cre-Tomato mice injected with control virus or AAV-flex-taCasp3-TEVp virus. (e) Food intake at 2, 4, and 8 h after a fast. (f) Weekly body weight change of Mc3r^cre-Tomato mice injected with control virus or AAV-flex-taCasp3-TEVp virus. (g) Percentage fat and lean mass of Mc3r^cre-Tomato mice injected with control virus and AAV-flex-taCasp3-TEVp virus. (h) Fat pad weights of Mc3r^cre-Tomato mice injected with control virus or AAV-flex-taCasp3-TEVp virus. Average values ± SEM are shown (AAV-Control, n = 8; AAV-Casp, n = 6). *P < 0.05; **P < 0.01; ****P < 0.0001. NS, not significant.

After establishing the effectiveness of Cre-dependent caspase ablation, a cohort of Mc3r^cre mice was stereotaxically injected with AAV-flex-GFP control virus or AAV-flex-taCasp3-TEVp virus to determine the necessity of Mc3R^LHA neurons in energy homeostasis. After recovery and allowing time for caspase expression to induce cell death, we measured food intake and body weight weekly for both groups of mice. We found no significant difference in either weekly food intake or acute fast–refeeding food intake between control and Mc3R^LHA-ablated mice [Fig. 6(d) and 6(e)]. Although the mean body weights of Mc3R^LHA-ablated mice were no different than the control mice 12 weeks after virus injection (Mc3R^LHA-ablated, 31.59 ± 5.5 g vs controls, 29.35 ± 4.9 g) owing to differences in weight at the time of injection (Mc3R^LHA-ablated, 27.61 ± 3.8 g vs controls, 28.85 ± 2.9 g), Mc3R^LHA-deleted mice gained weight more quickly than did the controls [Fig. 5(f)]. Furthermore, body composition analysis revealed an increased percentage of body fat in Mc3R^LHA-ablated mice, indicating a propensity for fat mass accrual [Fig. 6(g)]. Indeed, fat pad dissection at the time of perfusion revealed increased gonadal fat in Mc3R^LHA-deleted mice [Fig. 6(h)].

Because the accrual of excess fat mass in the setting of equivalent food intake suggested a defect in energy expenditure following Mc3R^LHA ablation, control and Mc3R^LHA-ablated mice were subjected to indirect calorimetry to assess parameters of energy expenditure and nutrient usage. Mc3R^LHA-ablated mice had a lower respiratory exchange ratio in the dark phase relative to controls, consistent with a shift toward fat utilization [Fig. 7(a)]. Energy expenditure normalized to body weight was lower in Mc3R-ablated mice [Fig. 7(b)]. To avoid confounding energy expenditure assessments by body weight normalization, we also analyzed absolute energy expenditure per animal as a function of body weight. Regression analysis of 24-hour energy expenditure as a function of body weight also showed a trend of lower energy expenditure in Mc3R-ablated mice [Fig. 7(c)]. Interestingly, mice with Mc3R^LHA neuron ablation demonstrated decreased horizontal movement (X-total activity) [Fig. 7(d)]. These data reveal that ablating Mc3R^LHA neurons decreases both locomotor activity and energy expenditure. Reductions in activity and energy expenditure would be predicted to enhance susceptibility to obesity when exposed to an HFD. Indeed, HFD-fed Mc3R^LHA-ablated mice gained weight more quickly than did controls despite similar food intake at baseline and with refeeding after a fast (Fig. 8).

Figure 7.

Comprehensive laboratory animal monitoring of Mc3R^LHA-ablated mice. (a) Respiratory exchange ratio (RER) and (b) energy expenditure normalized to total body weight measured by indirect calorimetry. (c) Regression analysis of total 24-h energy expenditure as a function of body weight. (d) X-total activity. Average values ± SEM are shown (AAV-Control, n = 8; AAV-Casp, n = 6). *P < 0.05. NS, not significant.

Figure 8.

Diet-induced obesity in Mc3R^LHA-ablated and control animals. (a) Body weight change and (b) absolute body weight of Mc3r^cre-Tomato mice injected with control virus or AAV-flex-taCasp3-TEVp virus. (c) Weekly HFD intake in Mc3r^cre-Tomato mice injected with control virus or AAV-flex-taCasp3-TEVp virus. (b) High-fat food intake at 2, 4, and 8 h of refeeding after a fast. Average values ± SEM are shown (AAV-Control, n = 6; AAV-Casp, n = 6). *P < 0.05; **P < 0.01.

Discussion

To selectively access and manipulate Mc3R-expressing neurons, we generated a novel Mc3r^cre transgenic line and used it in combination with viral vectors to probe the neurochemical composition, connectivity, and function of Mc3R^LHA neurons. We show that Mc3R^LHA neurons project to brain regions that regulate energy balance and express factors known to be involved in energy balance control. Indeed, acute activation of Mc3R^LHA neurons is sufficient to promote locomotor activity and augment refeeding after a fast and this is associated with activation of OX and VTA neurons. Conversely, ablation of Mc3R^LHA neurons results in decreased energy expenditure and activity, with increased body weight and fat mass, revealing that Mc3R^LHA neurons are required for energy homeostasis. Although Mc3R^LHA neurons are engaged by AgRP and POMC neurons anatomically [Fig. 1(f) and 1(g)], our studies do not test the role of Mc3R signaling in the LHA. We use Mc3R as a neuronal marker to genetically control the activity of the Mc3R^LHA neuronal subset and evaluate its role in energy balance control. Mc3r and Mc4r mRNAs overlap somewhat in the LHA (Fig. 9) with ∼18% of Mc3R^LHA neurons coexpressing Mc4r mRNA. Although MC4R signaling in the LHA contributes to glucose regulation and sympathetic output, it does not impact feeding (38). Whether the coexpression of Mc3r and Mc4r in the LHA has physiologic implications is not clear. Determining the role of LHA Mc3R signaling in energy balance control will require selective deletion or reactivation of Mc3R in the LHA.

Figure 9.

Colocalization of Mc3r and Mc4r mRNA in the LHA. (a) RNAscope dual chromogen staining of Mc3r transcripts (blue) with Mc4r transcripts (red) in the LHA from Mc3R^cre mice. Two representative LHA images are shown. Scale bars, 20 μm. (b) Quantification of colocalization of Mc3r and Mc4r transcripts in the LHA (n = 2 mice).

The LHA is a heterogeneous structure that contains a number of genetically distinct cell populations involved in energy homeostasis. Early lesion studies identified the LHA as an important feeding center (64), and cell-specific manipulations within the LHA profoundly alter feeding behaviors (24, 27, 33, 65, 66). Activation of GABAergic LHA neurons induces voracious feeding and increases the rewarding value of food (65, 67). Our studies reveal that Mc3R^LHA neurons are part of this GABAergic LHA population, express neuropeptides and receptors that regulate energy balance (Fig. 2), and project to multiple brain regions known to be involved in feeding control (Fig. 3). Direct chemogenetic activation of Mc3R^LHA cells (which are largely GABAergic) can promote feeding, but this effect is relatively small and is only observed in the fasted state. Activation of Gal^LHA neurons using DREADD technology enhances the motivation to work for food but does not alter overall food consumption (33). Activation of Nts^LHA neurons, alternatively, restrains feeding (68), suggesting that increased feeding following LHA GABAergic stimulation is not mediated by Nts. Thus subsets of Mc3R^LHA neurons that do not express either galanin or Nts are likely responsible for the feeding effects following Mc3R^LHA activation. Whether Mc3R^LHA activation modifies the willingness to work for food or is associated with positive valence will require additional studies.

Our viral tracing studies reveal dense Mc3R^LHA neuron projections to the BNST, VTA, PBN, and PVH (Table 1), brain regions associated with feeding, reward, locomotor activity, and energy expenditure. Selective activation of GABAergic LHA projections to the PVH stimulates feeding while suppressing grooming behaviors (69, 70). Our current studies do not determine the circuitry engaged by Mc3R^LHA neurons to augment refeeding. Selective activation of Mc3R^LHA terminals by optogenetics would be required to delineate the contribution of each projection pathway. It is of interest that Mc3R^LHA neuron activation induces nuclear Fos accumulation in OX neurons in the LHA. Although central administration of OX stimulates feeding acutely (71–74), direct activation of OX neurons using chemogenetics has given mixed results, with one group demonstrating robust feeding (75) whereas another showed no effect on feeding (76). These different outcomes may reflect the total number and/or location of OX neurons activated. Whether OX feeding effects reflect modulation of satiety pathways or a more generalized augmentation of foraging behaviors is not clear. The contribution of OX neuron activity in the physiologic responses to Mc3R^LHA activation remains to be determined and will be an important area of future investigation.

To determine the contribution of Mc3R^LHA neurons to body weight homeostasis, we selectively ablated these neurons using site-specific injection of Cre-conditional viral vectors. Ablation of Mc3R^LHA neurons was associated with more rapid weight gain with increased fat mass independent of changes in food intake. That Mc3R^LHA neuron ablation had no effect on feeding or refeeding after a fast indicates that Mc3R^LHA neurons are not required for normal feeding control. However, the observed increase in fat mass independent of food intake suggested a defect in energy expenditure. Indeed, indirect calorimetry of Mc3R^LHA neuron-ablated mice revealed differences in energy expenditure relative to controls either normalized to total body weight or plotted as a function of body weight. Consistent with an underlying energy expenditure defect, Mc3R^LHA neuron-ablated mice challenged with an HFD gained fat mass more rapidly than did controls with matched food intake. In contrast to these findings, a study by Begriche et al. (16) showed that the peripheral Mc3R, instead of central Mc3R, is important in diet-induced obesity. Again, it is important to recognize that our approach ablates the entire Mc3R-expressing neuron rather than just removing the Mc3R. Moreover, our manipulations were performed in adult animals and sidestepped issues of developmental compensation that may occur with germline gene deletion studies. Given the alterations in energy expenditure after Mc3R^LHA neuron ablation, we did expect a more pronounced weight gain with an HFD. The relatively modest weight gain seen in Mc3R^LHA- ablated animals following HFD exposure may reflect the redundancy of both central and peripheral systems (including peripheral Mc3R actions) that contribute to energy utilization and weight maintenance in the face of caloric challenge.

Ablation of Mc3R^LHA neurons is associated with decreased locomotor activity, whereas chemogenetic activation of Mc3R^LHA neurons drives locomotor activity. Alterations in locomotor activity following LHA cell manipulations have been previously reported. Activation of GABAergic-, Nts-, and galanin-expressing LHA neurons each alter locomotor activity. As highlighted by Qualls-Creekmore et al. (33), however, LHA cell-specific alterations in locomotor activity are not qualitatively equivalent. GABAergic LHA activation is associated with repetitive or compulsive behaviors, whereas activation of Gal^LHA neurons is associated with alterations across a range of behaviors that are not compulsive. A detailed characterization of the induced locomotor activity changes following Mc3R^LHA neuron activation will be necessary to determine whether these locomotor activity changes can be ascribed to defined subsets of Mc3R^LHA cells (e.g., galanin- vs Nts-expressing cells).

Chemogenetic activation of Mc3R^LHA neurons is associated with changes in Fos expression in OX neurons and neurons in the VTA, implicating these cells and brain region as possible effectors of Mc3R^LHA actions. As Mc3R^LHA neurons are predominantly GABAergic, the induction of Fos expression following Mc3R^LHA activation is likely to be indirect or mediated by other secreted factors. Immunohistochemical analysis of Mc3R^LHA neurons revealed colocalization with Nts, a neuromodulator that directly engages the dopamine system (77). Increased locomotor activity following Mc3R^LHA neuron activation was blunted by pretreatment with dopamine antagonists (Fig. 5), suggesting that Mc3R^LHA neurons might regulate the mesolimbic dopamine system through Nts release. Nts^LHA neurons project to the VTA, and Nts release in the VTA promotes physical activity (27, 30, 59). That a subset of Mc3R^LHA neurons express Nts [Fig. 2(c)] and that Mc3R^LHA neurons project to the VTA [Fig. 3(e)] suggest potential molecular and neuroanatomical mechanisms by which Mc3R^LHA neurons may regulate activity and link melanocortin action to the mesolimbic dopamine system. Importantly, note that a subset of Mc3R^LHA neurons expresses galanin and that Gal^LHA neurons modulate locomotor activity without directly targeting the VTA (33, 34). Thus, the changes in locomotor activity seen upon Mc3R^LHA activation may be a composite of effects driven through both Nts^LHA and Gal^LHA pathways, which target the VTA and other brain regions. Future studies designed to selectively delete Nts (or galanin) from Mc3R-expressing neurons are needed to definitively test these hypotheses.

In summary, Mc3R^LHA neurons play a crucial role in the regulation of energy expenditure and locomotor activity. Our studies demonstrate the necessity of Mc3R^LHA neurons in energy homeostasis, and they provide insight into cellular and anatomical pathways by which the LHA regulates feeding, energy expenditure, locomotor activity, and fat mass.

Glossary

Abbreviations:

AAV

adeno-associated virus

AgRP

agouti-related protein

ARC

arcuate nucleus

BNST

bed nucleus of the stria terminalis

CNO

clozapine N-oxide

CNS

central nervous system

DAB

diaminobenzidine

DREADD

designer receptors exclusively activated by designer drug

GABA

γ-aminobutyric acid

Gal^LHA

galanin cells in the lateral hypothalamic area

HFD

high-fat diet

IHC

immunohistochemistry

LepRb

leptin receptor

LHA

lateral hypothalamic area

MCH

melanin-concentrating hormone

Mc3R

melanocortin-3 receptor

Mc4R

melanocortin-4 receptor

Nts

neurotensin

OX

orexin

PBN

parabrachial nuclei

POMC

proopiomelanocortin

p-Stat3

phosphorylated signal transducer and activator of transcription-3

PVH

paraventricular nucleus of the hypothalamus

VMH

ventromedial nucleus of the hypothalamus

VTA

ventral tegmental area