Leptin action through hypothalamic nitric oxide synthase-1-expressing neurons controls energy balance.

Few effective measures exist to combat the worldwide obesity epidemic(1), and the identification of potential therapeutic targets requires a deeper understanding of the mechanisms that control energy balance. Leptin, an adipocyte-derived hormone that signals the long-term status of bodily energy stores, acts through multiple types of leptin receptor long isoform (LepRb)-expressing neurons (called here LepRb neurons) in the brain to control feeding, energy expenditure and endocrine function(2-4). The modest contributions to energy balance that are attributable to leptin action in many LepRb populations(5-9) suggest that other previously unidentified hypothalamic LepRb neurons have key roles in energy balance. Here we examine the role of LepRb in neuronal nitric oxide synthase (NOS1)-expressing LebRb (LepRb(NOS1)) neurons that comprise approximately 20% of the total hypothalamic LepRb neurons. Nos1(cre)-mediated genetic ablation of LepRb (Lepr(Nos1KO)) in mice produces hyperphagic obesity, decreased energy expenditure and hyperglycemia approaching that seen in whole-body LepRb-null mice. In contrast, the endocrine functions in Lepr(Nos1KO) mice are only modestly affected by the genetic ablation of LepRb in these neurons. Thus, hypothalamic LepRb(NOS1) neurons are a key site of action of the leptin-mediated control of systemic energy balance.

Commensurate with the diverse processes controlled by leptin, specialized types of LepR-b neurons lie in multiple brain regions involved in energy balance, including the brainstem, midbrain, and hypothalamus[10-13]. LepR-b knockdown or deletion in the hindbrain interferes with satiation, although these manipulations only slightly alter body adiposity[8,9]. Within the midbrain ventral tegmental area (VTA) and substantia nigra (SN), a subset of dopamine (DA) neurons contain LepR-b; leptin action via these neurons contributes minimally to body weight control, but plays a role in DA-mediated behaviors, including those linked to anxiety[14-17]. Midbrain serotonin neurons, while initially reported to play an important role in leptin action, neither express LepR-b nor contribute to leptin action[18,19]. In contrast, ablation of hypothalamic LepR-b produces a profound metabolic phenotype, demonstrating the importance of hypothalamic LepR-b signaling for leptin action[20].

Within the hypothalamus, the specific set(s) of LepR-b neurons responsible for the control of energy balance by leptin remain incompletely defined. Direct leptin action via proopiomelanocortin (Pomc) neurons of the hypothalamic arcuate nucleus (ARC), ARC Agouti-related peptide (Agrp) neurons, and steroidogenic factor-1 (Sf-1) neurons in the ventromedial hypothalamic nucleus (VMH) contributes only modestly to overall energy balance[5-7]. LepR-b neurons in the lateral hypothalamic area (LHA), including those that contain neurotensin, mediate leptin action on orexin neurons and the mesolimbic DA system, but deletion of LepR-b from these neurons only modestly increases adiposity[21-23]. Thus, the identity of the hypothalamic LepR-b neurons responsible for the majority of leptin action on energy balance has remained unclear.

Nos1-expressing LepR-b (LepR-bNOS1) neurons represent a relatively small population of LepR-b neurons mainly restricted to the hypothalamus, where they are distributed in areas poised to impact output from the paraventricular hypothalamic nucleus (signaling by which mediates much of the hypothalamic control of energy balance)[24-26]. To study LepR-bNOS1 neurons, we inserted an internal ribosome entry site (IRES) plus the coding sequences for cre recombinase into the 3’-UTR of Nos1 in mice to promote Nos1-restricted cre expression (Nos1)(Figure 1a). We bred Nos1 mice to the cre-dependent ROSA26-EGFP reporter line, generating Nos1 mice. EGFP-IR overlapped with Nos1-IR soma in all regions examined (data not shown).

Figure 1

Generation of Nos1 and lack of Nos1 in ARC Pomc and Agrp neurons

(a) Schematic diagram depicting Nos1 and its use to generate Nos1 and Lepr mice. (b) Representative images from the ARC of Pomc mice demonstrating EYFP (green, left panel), Nos1-IR (red, middle panel), and merged images (right panel). (c) Representative images from the ARC of Agrp mice demonstrating EYFP (green, left panel), Nos1-IR (red, middle panel), and merged (right panel). Scale bars = 200 µm.

We examined the leptin-stimulated induction of phosphorylated signal transducer and activator-3 (pStat3; which reveals neurons containing functional LepR-b[27]) and its colocalization with EGFP-immunoreactivity (−IR) in Nos1 mice, as well as evaluating the colocalization of Nos1-IR with EGFP-IR in Lepr mice[12] that express EGFP in LepR-b neurons (Supplemental Figures 1–2). As suggested by NADPH-diaphorase colocalization with leptin-stimulated pStat3-IR[24], our analysis revealed that although Nos1 and LepR-b are each expressed in many brain areas, they are co-expressed predominantly in the hypothalamus; approximately 20% of hypothalamic LepR-b neurons contain Nos1. Most LepR-bNOS1 neurons lie within the ventral premammillary nucleus (PMv); the DMH and ARC also contain a substantial number; other areas contain few LepR-bNOS1 neurons. ARC Nos1-expressing cells are distinct from Pomc and Agrp neurons (Figure 1b, c; Supplemental Figure 3).

To determine the role of LepR-bNOS1 neurons in leptin action, we crossed Nos1 mice with Lepr animals[28]. Since Nos1 is expressed in gametes, germline Lepr excision occurred in the offspring of Nos1Lepr parents, generating Lepr. We therefore interbred Nos1Lepr to Lepr animals to generate Nos1 (Lepr) animals in which LepR-b was ablated specifically from LepR-bNOS1 neurons (Figure 1a), along with Nos1Lepr and Lepr (control) littermates. The persistence of cre in the early embryo of some animals also produced Nos1 littermates with LepR-b inactivated in all tissues (LeprKO).

Analysis of the hypothalamus of control, Lepr, and LeprKO animals (Supplemental Figure 4a, b) demonstrated the absence of pStat3 in LeprKO animals. We also confirmed the ablation of leptin-stimulated pStat3-IR from the PMv in Lepr animals, along with modestly lower pStat3-IR from other areas (ARC and DMH) containing LepR-bNOS1 neurons. Thus, functional LepR-b was ablated from LepR-bNOS1 neurons in Lepr mice. Consistent with the lack of overlap between Pomc and Nos1 neurons (Figure 1b; Supplemental Figure 3), leptin treatment stimulated pStat3-IR in similar proportions of ARC Pomc neurons in control and Lepr mice (Supplemental Figure 4c).

Lepr and LeprKO males exhibited similarly high body weight and food intake relative to controls (Figure 2a, b). VO2 for Lepr mice was lower than controls and not different from LeprKO animals; similar trends were observed for ambulatory activity (Figure 2c, d; Supplemental Figure 5). The excess weight in Lepr and LeprKO mice was largely due to increased adipose mass (Figure 2e), which was reflected by high leptin concentrations relative to controls (Figure 2f). We observed similar trends in Lepr females (Supplemental Figure 6), although the obesity and metabolic dysfunction in Lepr females was less severe relative to LeprKO females. Thus, direct leptin action on LepR-bNOS1 neurons is crucial for the regulation of feeding, activity, and energy expenditure, and therefore for the control of body weight and adiposity.

Figure 2

LepR-bNOS1 neurons regulate energy balance and glucose homeostasis

(a) Body weight and (b) food intake from 4–12 weeks of age for male control (Ctrl), Lepr and LeprKO mice. CLAMS analysis of 12–14 week old male mice determined (c) ambulatory locomotor activity and (d) VO2. Data are shown for dark cycle (Dark), light cycle (Light) and averaged over 24 h (Total). (e) Body composition analysis of 12–14 week old animals. (f) Serum leptin concentrations for 12 week old animals. (g) Biweekly blood glucose concentrations from 4–12 weeks of age for male Ctrl, Lepr and LeprKO mice. (h) Serum insulin concentrations for 8 week old animals. Graphed data represent average values ± SEM; n ≥ 8 for all measurements. ANOVA: *, p < 0.05 vs Ctrl; **, p < 0.01 vs Ctrl; ***, p < 0.001 vs Ctrl; #, p < 0.05; all other comparisons, p = NS.

To determine the potential role for leptin action via LepR-bNOS1 neurons in glucose homeostasis, we examined circulating blood glucose and insulin concentrations in control, Lepr and LeprKO mice (Figure 2g, h). As with body weight and adiposity, male Lepr and LeprKO mice exhibited similarly high blood glucose concentrations relative to controls from an early age, despite high insulin concentrations relative to controls. Thus, leptin action via LepR-bNOS1 neurons is indispensible for the regulation of glucose homeostasis, in addition to energy balance, and LepR-bNOS1 neurons represent crucial regulators of metabolic leptin action.

To interrogate the contribution of LepR-bNOS1 neurons to the control of endocrine systems by leptin, we examined circulating hormone concentrations along with parameters of reproductive function in Lepr mice (Figure 3). While circulating T4 concentrations in Lepr males were lower than for controls, they remained higher than in LeprKO mice (Figure 3a). Furthermore, while circulating corticosterone was higher in LeprKO males than in Lepr and control animals, corticosterone concentrations were not different between Lepr and controls (Figure 3b). The onset of vaginal estrus was delayed relative to controls in singly-housed Lepr females (Figure 3c); upon mating with C57Bl/6 males, all Lepr females delivered litters with similar timing as controls (Control = 23 ± 1 d; Lepr = 24 ± 1 d), and with similar litter sizes (Control = 6.9 ± 1.1; Lepr = 7.1 ± 0.7 pups litter−1), however (Figure 3d). In contrast, the majority of LeprKO females failed to deliver pups within 6 weeks of mating.

Figure 3

LepR-bNOS1 neurons contribute modestly to endocrine functions

Serum T4 (a) and corticosterone (b) concentrations from 7–8 week old male control (Ctrl), Lepr and LeprKO mice. All measurements, n ≥ 5. Graphed data represent average values ± SEM. (c) Analysis of time to first estrus for female Ctrl and Lepr mice; percent exhibiting vaginal estrus is plotted by age. All measurements, n ≥ 9. (d) Analysis of reproductive competence in female Ctrl, Lepr and LeprKO mice. Experimental animals were mated with C57Bl/6 males and the delivery of pups was monitored daily. Percent delivering pups by days post-mating is plotted. All measurements, n ≥ 8. ANOVA: *, p < 0.05 vs Ctrl; **, p < 0.01 vs Ctrl; ***, p < 0.001 vs Ctrl; #, p < 0.05; all other comparisons, p = NS.

Thus, while deletion of LepR-b from LepR-bNOS1 neurons promoted obesity and metabolic dysfunction to an extent similar to that observed in animals with Lepr deleted throughout the body, the control of the thyroid and adrenal axes remained relatively intact, and female reproductive function was partially impacted. Hence, our present findings reveal that leptin action via LepR-bNOS1 neurons contributes modestly to the control of endocrine function, but is indispensible for the control of energy balance and glucose homeostasis.

Relatively few ARC LepR-b neurons express Nos1, and the deletion of LepR-b from LepR-bNOS1 neurons was insufficient to detectably alter ARC Socs3 expression (a surrogate for LepR-b signaling in the ARC[29])(Figure 4a). Altered ARC physiology may contribute to the phenotype of Lepr mice, however; while the expression of Npy and Agrp were unaltered in Lepr mice relative to controls, Pomc expression was lower in Lepr mice than in controls (Figure 4b–d), even though LepR-bNOS1 neurons are distinct from Pomc neurons (Figures 1b; Supplemental Figures 3, 4c). Thus, LepR-bNOS1 cells may indirectly mediate aspects of melanocortin action, consistent with recent findings revealing that leptin action via non-Pomc cells controls important aspects of Pomc neuron function and the regulation of energy balance[5,30]. Given the severity of the metabolic defects exhibited by Lepr mice, LepR-bNOS1 neurons may act by other mechanisms, as well.

Figure 4

ARC gene expression in Lepr animals

(a) Socs3, (b) Agrp, (c) Pomc, and (d) Npy mRNA expression by qPCR of microdissected ARC samples from ad libitum-fed male and female 7–8 week old control (Ctrl), Lepr and LeprKO mice. All measurements n ≥ 14. Graphed data represent average values ± SEM. ANOVA: *, p < 0.05 vs Ctrl; **, p < 0.01 vs Ctrl; ***, p < 0.001 vs Ctrl; all other comparisons, p = NS.

Thus, leptin action via LepR-bNOS1 neurons is crucial for the control of energy balance and metabolism and for the regulation of ARC Pomc neurons. While the frequency of Nos1-mediated excision in gametes dictated that we study mice on the heterozygous Lepr background, Lepr mice express the wild-type receptor on all LepR-b neurons and have no detectable phenotype; it is therefore unlikely that this genetic background contributed substantially to the dramatic phenotype of Lepr mice. While Nos1 expression identifies LepR-bNOS1 neurons important for metabolic control, Nos1 signaling is unlikley to mediate downstream leptin action; although leptin regulates the phosphorylation and presumably the activity of Nos1 in hypothalamic regions containing LepR-bNOS1 neurons[24], NO is a retrograde transmitter[31] and animals null for Nos1 exhibit no obvious primary metabolic phenotype[32]. It is therefore likely that other transmitters in the LepR-bNOS1 neurons play crucial roles.

Recent data suggest that GABAergic (vGat-expressing) LepR-b neurons (which represent many ARC, DMH, and LHA LepR-b neurons- approximately 60–75% of total hypothalamic LepR-b neurons) are important for the control of energy balance. Approximately 20–30% of ARC and DMH LepR-bNOS1 neurons contain Gad1 (which produces GABA), and a similar fraction of ARC/DMH LepR-bNOS1 neurons are activated by leptin (Supplemental Figure 7). By itself, direct leptin action on glutamatergic (vGlut2-expressing) LepR-b neurons (essentially all PMv and VMH LepR-b neurons) contributes only modestly to the control of energy balance by leptin[30]. Leptin activates the vast majority of PMv LepR-b neurons (most of which are LepR-bNOS1 neurons) [25,33], and a variety of data support a role for PMv LepR-bNOS1 neurons (~70% of total LepR-bNOS1 neurons) in fertility, consistent with the delayed estrus observed in Lepr females[34]. Since the PMv represents a sexually-dimorphic nucleus rich in androgen receptor [35], sex-specific differences in LepR-b neurons in this nucleus could contribute to the sexual dimorphism observed in the Lepr phenotype. The apparently minor role played by glutamatergic LepR-b neurons in energy balance suggests that PMv LepR-bNOS1 neurons alone do not mediate the majority of the metabolic phenotype displayed by Lepr mice, suggesting either that the modest number of LepR-bNOS1 neurons elsewhere (e.g., in the DMH and ARC) mediate this dramatic phenotype, or that the PMv LepR-b neurons may reinforce the action of the DMH and ARC LepR-bNOS1 neurons via supporting action on similar neural networks. It will be important to understand the detailed mechanisms of action for LepR-bNOS1 neurons in the future, since this small group of cells controls energy balance and thus represents a potential therapeutic target for obesity and related diseases.

Methods

Materials

Leptin was the generous gift of Amylin Pharmaceuticals, Inc. (San Diego, CA).

Animals

We bred all mice in our colony in the Unit for Laboratory Animal Medicine at the University of Michigan. All animals and procedures used were in accordance with the guidelines and with the approval of the University Committee on the Use and Care of Animals. We provided all animals ad libitum access to food and water. We purchased male C57Bl/6J animals for breeding studies and Gt(ROSA)26Sor (ROSA) mice from Jackson Labs. We produced Pomc and Agrp animals by crossing Pomc and Agrp animals[5,36] onto the ROSA background.

To generate Nos1, PCR amplification produced a 6 kb fragment containing the mouse genomic Nos1 sequence centered on the STOP codon in the final (3’) exon for insertion into pCR2.1. PCR mutagenesis created an AscI site 60 bp 3’ to the STOP codon, for the introduction of the IRES-Cre-Frt-Neo-Frt sequences to generate pCRNos1-IRES-Cre-Frt-Neo-Frt. NotI/NheI digestion excised the entire insert for subcloning into NotI/XbaI-cut pPNT backbone to generate pPNT-Nos1-IRES-Cre for targeting. NotI digestion linearized the vector for electroporation into R1 ES cells. We used Taqman-based qPCR screening to initially identify correctly targeted clones[37], followed by Southern blotting for final confirmation. We injected correctly targeted ES cells into blastocysts to generate chimeras, which we bred to C57Bl/6 animals to establish germline transmission.

We bred Nos1 mice with ROSA mice to generate Nos1;ROSA (Nos1) animals for the analysis of Nos1 expression. We also bred Nos1 animals with Lepr mice. Due to the periodic expression of Nos1 during gametogenesis, we bred Nos1lepr to Lepr mice in order to obtain littermate Nos1;Lepr (Lepr), Nos1; Lepr and Lepr (control) and Nos1 (LeprKO) animals for study. We genotyped the offspring by PCR.

Perfusion and immunolabeling

We anesthetized the mice with an overdose of intraperitoneal (IP) pentobarbital and transcardially perfused them with 10% neutral buffered formalin. We sectioned the brains coronally (30 µm) using a sliding microtome followed by immunofluoresent analysis. We visualized the antigens via immunofluorescence using species-specific AlexaFluor-488 or -568 secondary antibodies (Invitrogen, Cat# A11039 and A10042; 1:200), and processed and imaged the sections as previously described[38]. Antibodies used were GFP (Abcam, chicken, Cat# ab13970; 1:1000) and Nos1 (ImmunoStar, rabbit, Cat# 24287; 1:5000). We purchased normal donkey serum and biotinylated donkey anti-rabbit (Cat# 711-065-152) from Jackson ImmunoResearch. We counted cells using Adobe Photoshop software.

Phenotypic studies

We individually housed mice for study from the time of weaning at 21 d. Beginning at 28 d, we monitored body weight and chow (Purina Lab Diet #5001) intake weekly. We collected blood for serum and measured blood glucose with a glucometer biweekly. We monitored female mice for vaginal opening and then for vaginal estrus by cellular histology until eight weeks of age. We collected all data between 13:00 and 16:00. We analyzed mice for body fat and lean mass between 12–14 weeks of age using an NMR-based analyzer (Minispec LF90II, Bruker Optics). We also analyzed a subset of mice (13–16 weeks old) for oxygen consumption (VO2) and locomotor activity using the Comprehensive Laboratory Animal Monitoring System (CLAMS, Columbus Instruments). We analyzed serum for insulin and leptin using assays from Crystal Chem; we purchased assays for T4 and corticosterone from Siemens and Arbor Assays, respectively.

RNA extraction and qPCR

We prepared RNA from microdissected ARC using Trizol (Invitrogen) and converted 1 µg samples to cDNA using the Superscript First Strand Synthesis System for RT-PCR (Invitrogen). We analyzed these cDNAs in triplicate via quantitative RT-PCR for Gapdh and Socs3 (Applied Biosystems) or Pomc, Npy, and Agrp[39] using an Applied Biosystems 7500. We calculated relative mRNA expression values by the 2−ΔΔCt method, with normalization of each sample ΔCt value to the average ΔCt from Ctrl mice.

Statistics

We used one-way ANOVA followed by Bonferroni post hoc test for multiple comparisons using Graphpad Prism software for PC. Differences were considered significant for p < 0.05.