Amylin/Calcitonin Receptor–Mediated Signaling in POMC Neurons Influences Energy Balance and Locomotor Activity in Chow-Fed Male Mice

Amylin is a pancreatic gut hormone that is co-released with insulin by β-cells in response to meals (1). The main central effect of amylin is an acute reduction of food intake that occurs after amylin binding to neurons in the area postrema (AP) (2). The amylin receptor consists of a core calcitonin receptor (CTR_A/B), which is coupled to receptor activity–modifying proteins (RAMP1–3) (3,4). These receptor components are expressed in single neurons of the AP (5), but amylin also has other binding sites in the central nervous system that have not been thoroughly studied yet, such as the nucleus of the solitary tract (NTS), the lateral hypothalamic area (LHA), and the ventromedial (VMN) and arcuate (ARC) hypothalamic nucleus (6–8). Recent data of whole brain imaging with fluorescently labeled rat amylin in vivo confirmed its binding in the ARC and AP (9). The ARC has come into focus as a mediator of amylin’s effects on energy expenditure (EE), its interactions with leptin signaling pathways, and its developmental effects on axonal fiber outgrowth (6,10,11). The effects of amylin in adults are of interest because chronic amylin treatment has a sizable effect on weight loss that cannot be solely explained by a reduction in food intake (12) and hypothalamic actions might contribute to the “leptin-sensitizing” effect of amylin (13,14).

Our current hypothetical model of amylin signaling in the ARC includes a direct effect of amylin on POMC neurons through ERK1/2 phosphorylation and an indirect effect on AgRP neurons through microglial IL-6 secretion (15). This study aims to investigate the direct physiological effect of endogenous amylin signaling on POMC neurons with a conditional genetic knockout (KO) model. The depletion of CTR specifically in POMC neurons with tamoxifen (Tx) induction after weaning aims to avoid early developmental effects of disturbed amylin signaling (15) and allows us to test the hypothesis that amylin signaling in POMC neurons is critical for the control of EE.

Animals were kept in a temperature-controlled environment (21 ± 2°C) on a 12:12 h light cycle with lights off at 1000 h. After weaning, they were separated by genotype and group-housed with littermates or single-housed for the Tx injections (16). Tx (cat. no. T5648; Sigma-Aldrich, Merck KGaA, Darmstadt, Germany) was dissolved in pure ethanol and mixed with corn oil (cat. no. C8267; Sigma-Aldrich) for a final concentration of 100 mg/mL. Mice were injected at 4 weeks old with a dose of 150 μg/g or corn oil for 5 days. A first cohort of mice was fed standard chow (65% carbohydrate, 22% protein, and 12.5% fat as percent of total energy content, cat. no. 3430; Provimi Kliba, Kaiseraugst, Switzerland), and two separate cohorts were fed a high-fat diet (HFD) (35% carbohydrate, 20% protein, and 45% fat as percent of total energy content, cat. no. D12451; Research Diets, New Brunswick, NJ) ad libitum. The animals were kept in an enriched environment in wood chip bedding with cardboard houses and tissues as nesting material. The Veterinary Office of the Canton of Zurich, Zurich, Switzerland, approved all animal procedures.

POMC-cre:ER^T2 (C57BL/6J;129X1/SvJ-Tg(Pomc-cre/ERT2)#Jke; MGI:5569339) (16) (kindly provided by Joel Elmquist, UT Southwestern), CTR^fl/fl (Calcr<tm1(fl)>; MGI:5751436) (frozen sperm kindly provided by Drs. Jean-Pierre David and Thorsten Shinke, University Medical Center Hamburg) (17). The crossing resulted in two groups of mice: POMC-WT × CTR^fl/fl (POMC-WT^CTR) and POMC-Cre × CTR^fl/fl (POMC-Cre^CTR). For confirmation of the genetic model, Ai14 reporter mice [B6.Cg-Gt(ROSA)26Sor<tm14(CAG-tdTomato)Hze>/J(#007914)] were bred with CTR^fl/fl POMC-Cre^ERT2 mice. Each mouse was genotyped using previously published primers (17).

Mice were single housed in BioDAQ cages (Research Diets), and following 7 days of acclimation, mice were fasted for 12 h during the light phase. At dark onset mice were injected intraperitoneally (i.p.) with amylin (50, 500 μg/kg) (cat. no. H-9475; Bachem, Bubendorf, Switzerland), salmon calcitonin (sCT) (10 µg/kg) (cat. no. 4033011.0001; Bachem) or saline (NaCl 0.9%) in a crossover design and food was returned. Using the same paradigm described above, a separate cohort of mice was also tested for anorectic response to leptin (5 mg/kg i.p.) (PeproTech, London, U.K.). Food intake was recorded for the following 24 h, and the mice could rest for two more days before the next injection. Baseline food intake was calculated by averaging food intake over a 3-day period prior to injections. Meal pattern criteria were an intermeal interval of 600 s and a minimal meal of 0.02 g (18). Male and female food intake data were pooled, since no differences were observed between sexes (19).

Food was removed 2 h prior to lights off, and mice were gavaged (2 g/kg glucose) or injected (0.5 units/kg insulin) at lights off. Blood glucose (Contour XT; Ascensia Diabetes Care, Basel, Switzerland) at the tail was measured before and 15, 30, 45, 60, 90, and 120 min after gavage or injection. In addition, baseline blood samples after a 2-h fast during the study were obtained by tongue bleeding during a brief 30-s isoflurane anesthesia (2%). Insulin and leptin were measured (Meso Scale Discovery, Rockville, MD).

TA-F10 sensors for body temperature and activity measurements (Data Sciences International, New Brighton, MN) were implanted intraperitoneally under brief isoflurane anesthesia (2%). Treatment with nonsteroidal anti-inflammatory agents (2 mg/kg subcutaneous [s.c.] Metacam) and antibiotics (7.5 mg/kg s.c. Baytril 2.5%) before surgery and during the following 5 days was provided while the mice recovered. Subsequently, mice were single housed in a 16-cage PhenoMaster indirect calorimetry system (TSE Systems, Bad Homburg, Germany) (18). After 1 week of adaptation, data were collected at baseline or following a 12- h fast and injection with saline or sCT (10 µg/kg i.p.). From these values, EE and respiratory exchange ratio (RER) were calculated based on equations of Weir (20). Body mass composition from L1 to L4 was performed using a computed tomography scan (Quantum GX microCT; PerkinElmer, Waltham, MA). Lean and fat mass (FM) were quantified as previously validated (21). Analyze 12.0 software (AnalyzeDirect, Overland Park, KS) was used to quantify visceral and subcutaneous fat volume in the computed tomography images. EE data were corrected for individual lean body mass (LBM) (in grams) and FM (in grams) using the following equation: LBM + 0.2 FM, as recommended by Even and Nadkarni (22).

For assessment of phosphorylated (p)ERK signaling, mice were fasted for 12 h and at dark onset were injected with saline or amylin (50 µg/kg i.p.) as previously described (15). For assessment of pSTAT3 signaling, mice were fasted for 2 h and at dark onset were injected with saline or leptin (5 mg/kg i.p.) (23). The brains were frozen in hexane on dry ice, stored at −80°C, cut in 25 μm sections (Leica Biosystems, Wetzlar, Germany), mounted on Superfrost Plus slides (Thermo Fisher Scientific, Reinach, Switzerland), and stored in cryoprotectant (50% 0.02 mol/L potassium phosphate buffered saline (KPBS), 30% ethylene glycol, 20% glycerol) at −20°C until staining.

For pretreatment, sections were demasked in 0.5% NaOH + 1% H₂O₂ in KPBS and incubated in 0.3% glycine in KPBS. Sections were blocked and incubated with pERK antibody (1:1,000, cat. no. 9101; Cell Signaling Technology) in 2% normal goat serum (NGS)–0.3% Triton–1% BSA in KPBS for 48 h at 4°C. Sections were then placed in Cy3 goat anti-rabbit secondary antibody at 1:100 concentration for 2 h at room temperature. Sections were again blocked before being incubated with primary POMC antibody (1:1,000, cat. no. H-029–30; Phoenix Pharmaceuticals, Karlsruhe, Germany) for 48 h at 4°C followed by Alexa Fluor 488 donkey anti-rabbit (1:100, cat. no. 711-545-152; Jackson ImmunoResearch, Cambridgeshire, U.K.) secondary antibody for 2 h. Sections were counterstained with DAPI (0.5 mg/L) and the slides cover slipped with VECTASHIELD (Vector Laboratories, Burlingame, CA) (15).

Immunohistochemistry (IHC) was performed as previously described (23). Briefly, brain sections were demasked in NaOH and H₂O₂ followed by glycine and SDS. Sections were blocked for 1 h in 4% NGS, 0.4% triton, and 1% BSA in KPBS and were then incubated in rabbit anti-pSTAT3 (1:1,000, cat. no. 9145; Cell Signaling Technology) for 48 h at 4°C. Section were then incubated in Cy3 goat anti-rabbit for 2 h. Sections were blocked again and incubated with rabbit anti-POMC as described above.

For pretreatment, sections were demasked in 0.01 mol/L sodium citrate, pH 6, at 90°C in a steamer for 20 min. After cooling, sections were rinsed, blocked, and incubated with CTR antibody (1:400, cat. no. ab11042; Abcam, Cambridge, U.K.) in 2.5% NGS–0.3% Triton–1% BSA in PBS for 48 h at 4°C. Sections were incubated with a biotinylated goat anti-rabbit (Vector Laboratories) at 1:500 for 2 h followed by streptavidin-conjugated 647 secondary antibody (Invitrogen) at 1:1,000 concentration in PBS–0.3% Triton for 2 h. Sections were counterstained with DAPI (0.5 mg/L) and the slides cover slipped with VECTASHIELD.

Sections were blocked in 0.3% Triton–2% BSA–3% NGS in KPBS before incubation with primary rabbit anti–orexin A (1:1,000; cat. no. H003-30; Phoenix Pharmaceuticals) for 48 h at 4°C followed by Alexa Fluor 488 goat anti-rabbit (1:100; Jackson ImmunoResearch) for 2 h. Sections were then counterstained with DAPI and cover slipped with VECTASHIELD.

IHC was performed as previously described (15), using the same series of sections as for the POMC-pERK staining.

Fresh frozen brain from two POMC-WT × CTR^fl/fl (POMC-WT^CTR) male mice (n = 2) was cut in 14-µm sections onto Superfrost Plus slides. After postfixing in 4% paraformaldehyde for 15 min, dehydration in ethanol, and in situ hybridization target retrieval (30 min of ACD enzyme protease IV [Advanced Cell Diagnostics, Newark, CA]), the mRNA signal of interest was detected using the RNAscope Multiplex Fluorescent Reagent Kit (cat no. 323130; Advanced Cell Diagnostics). Specific probes targeting mouse CTR, POMC, RAMP1, or RAMP3 (Probe-Mm-Calcr, cat no. 494071-C3; Probe-Mm-Pomc-C2, cat no. 314081-C2; Probe-Mm-RAMP1-C1, cat no. 532681-C1, and Probe-Mm-RAMP3-C1, cat no. 497131-C1; Advanced Cell Diagnostics) were used according to the manufacturer’s instructions. The slides were counterstained with DAPI (Advanced Cell Diagnostics) and cover slipped using hard-set fluorescent mounting medium before being scanned using a confocal microscope (Zeiss SP8 confocal system equipped with a ×63/1.40 objective (POMC: HD1 555 laser 10%, RAMP1: HD1 488 laser 20%, RAMP3: HD1 488 laser 10%, CTR: HD2 647 laser 10%, DAPI: HD2 405 laser 5% with 10% gain, Z stack 10 µm, and step size 0.5 µm). The ARC, AP, and NTS were scanned, and for each images, 9–12 tiles of ×63 images were merged.

Interscapular brown adipose tissue (iBAT) was fixed in 4% phosphate buffer-paraformaldehyde for 48 h at 4°C. The tissue was embedded in paraffin and cut into 5-μm sections (Laboratory for Animal Model Pathology, Vetsuisse Faculty, University of Zurich) and processed for hematoxylin-eosin (H-E) (24). These H-E–stained slides were scanned and analyzed with the Visiopharm software (Hoersholm, Denmark) by classifying different areas for vacuoles, membranes, and nuclei and quantifying two regions of interest per scan, with exclusion of vessels or white adipose tissue (24).

Paraffin-embedded tissue was slide cut to 5-μm sections and deparaffinized in an oven at 60°C for 20 min. Slides were then immersed two times in xylene for 10 min and rehydrated with decreasing ethanol steps. After antigen retrieval at 95°C for 20 min in sodium citrate buffer, slides were blocked in PBS with 0.3% Triton–3% normal donkey serum (NDS) and incubated with rabbit anti-UCP1 (1:500, cat. no. ab10983; Abcam) in blocking buffer at 4°C for 48 h. Slides then were placed in secondary antibody (1:100, Alexa Fluor 488 donkey anti-rabbit; Jackson Immunoresearch) in the same buffer and counterstained with DAPI (25).

Cells expressing pERK-POMC and pSTAT3-POMC in the ARC and orexin in the lateral hypothalamus were imaged on a ×20 objective, and three sections of the ARC and lateral hypothalamus per animal were acquired with use of an Axio Imager 2 microscope (Zeiss, Oberkochen, Germany), blinded, and quantified and averaged (15). UCP1 staining was acquired with a ×100 oil-immersed objective, and three regions of interest were quantified per mouse and averaged. Quantitative analysis of CTR-immunopositive POMC:tDTomato neurons was performed with a Zeiss SP8 confocal system equipped with a ×20/0.75 objective (HD1 555, laser 5%, and HD2 647, laser 7%, with 10% gain, Z stack 21 mm, and step of 1.5 mm). Quantitative analysis of α-melanocyte-stimulating hormone (α-MSH) fiber density was performed as described above (HD1 488, laser 2%, with 10% gain, zoom 1, pinhole 1, Z stack 21 mm, and step of 1.5 mm) (15,23).To ensure similar imaging conditions for all images, we used the same microscope setup and acquisition settings to acquire all images within the same experiment. The experimenter was blinded for all quantifications.

Statistical comparisons among variables were made by one- or two-way ANOVA, as appropriate, with Tukey post hoc analysis (GraphPad Prism, La Jolla, CA). When appropriate, the unpaired t test was performed. All data are expressed as means ± SEM.

The data sets generated during or analyzed during the current study are available from the corresponding author upon reasonable request.

The depletion of CTR specifically in POMC neurons was assessed in POMC-WT^CTR and POMC-Cre^CTR using tdTomato reporter mice and CTR IHC. POMC-Cre:ER^T2:tdTomato-CTR floxed mice treated with Tx (i.e., KO) showed that POMC neurons did not colocalize with CTR in the KO mice, which was clearly different from the control oil-treated floxed mice (Fig. 1A and B). It should be noted that ∼20% of POMC neurons coexpressed CTR in the ARC (Fig. 1A).

With use of in situ hybridization, POMC-WT^CTR mice were further characterized to assess the colocalization of POMC and CTR_A with RAMP1 or RAMP3 in the ARC, AP, and NTS (Fig. 1C–H), but as seen at the protein level (Fig. 1A), the colocalization of CTR and RAMP with POMC was not detected in all POMC⁺ cells. We observed that CTR_A is expressed with or without RAMP in the ARC, AP, and NTS (Fig. 1C–H). RAMP3 expression was higher than RAMP1 in both the ARC and hindbrain (Fig. 1C–H). In the AP, RAMP3 was mostly colocalized with CTR_A, while this was not the case for RAMP1 (Fig. 1D and G). In the NTS, RAMP and CTR were found to be less colocalized than in the AP (Fig. 1E and H), suggesting that the AP might be the main site of action for amylin signaling. A few POMC neurons were also detected in the POMC-WT^CTR mice in the rostral part of the NTS, and few colocalized with CTR (Supplementary Fig. 5C and F). Thus, these results suggest that the formation of AMY1 and AMY3 is present in POMC⁺ and POMC⁻ neurons.

Twelve-week-old POMC-Cre^CTR-Tx (KO) mice show no significant increase in pERK-POMC–positive neurons in the ARC after amylin injection versus saline injection—contrary to the three control groups (15) (Fig. 2A and B). The density of α-MSH fibers in the paraventricular nucleus (PVN) (23,26) was similar between KO and control groups (Fig. 2C and D). These results confirm the functional depletion of amylin signaling in the POMC-CTR KO model and that the induced KO at 4 weeks does not affect the early-life development of POMC neuronal projections to other nuclei such as the PVN (15).

Starting at 9 weeks of age, male KO mice showed a 44% increase in body weight gain and 22% increase in cumulative food intake compared with control groups (Fig. 3A–C). Overall, male KO mice tended to have smaller and shorter meals, but this effect was overcompensated by a significant increase in meal number (Supplementary Table 1). Meanwhile, female mice showed a 30% lower body weight gain after Tx treatment independent of genotype (Fig. 3D–F). Body composition analysis on chow diet revealed that the excess body weight of male KO chow-fed mice was fat specific and mostly driven by a 45% increase in visceral FM (Fig. 3G). This effect was not found in chow-fed female mice (Fig. 3H). The increase in visceral fat on chow diet was not sufficient to induce an increase in blood leptin levels, probably because subcutaneous fat, which has a higher expression of leptin (27), was similar between the groups (Table 1).

On an HFD (45% fat), male POMC-Cre^CTR mice had a 20% increase in body weight regardless of Tx or oil treatment, but there was no difference in body weight gain or food intake among all groups (Supplementary Fig. 1A–C). An almost opposite effect was observed in HFD-fed females, where oil- or Tx-treated POMC-Cre^CTR mice showed a 15% decrease in body weight–associated with a 15% decrease in cumulative food intake (Supplementary Fig. 1D–F). Body composition analysis showed no consistent difference in FM in HFD-fed male or female KO versus control mice (Supplementary Fig. 1G and H). After 6 and 12 weeks on HFD, leptin levels were elevated by ∼10- to 20-fold compared with chow-fed mice and were similar across all groups (Table 1).

Eleven-week-old chow-fed male KO mice had the highest glucose levels at 15 and 30 min postgavage compared with control groups, which is reflected by a significantly increased area under the curve (AUC) (+48% vs. Tx control) (Fig. 4A). Similarly, male KO mice displayed an increase of the AUC (+42% vs. Tx control) and peak glucose levels on 45% HFD (Fig. 4E). Interestingly, HFD did not worsen glucose tolerance compared with chow diet–fed KO mice, although baseline fasting glucose levels were higher on HFD (+24%, P = 0.002) (Fig. 4A and E). There was no difference in females among all groups fed chow or HFD except for a lower glucose peak (−20%) in WT controls (Fig. 4C and G). When injected with insulin, male and female mice displayed similar glucose excursions on chow diet, suggesting a similar insulin sensitivity and/or counterregulatory response to hypoglycemia. During HFD feeding, male POMC-Cre^CTR+oil mice showed a reduced insulin tolerance compared with KO mice (Fig. 4F), and overall, all groups fed HFD were less responsive to insulin compared with chow diet–fed male and female mice (Fig. 4B, D, F, and H).

No significant difference in fasting insulin was observed among treatment groups within each diet at any time point (Table 1). Mice that were fed HFD for 6 and 12 weeks showed increased baseline levels of insulin by twofold and fivefold, respectively, compared with chow diet mice.

In a 3-day baseline measurement, male KO mice on chow diet reduced their EE by 11% compared with the Tx control group (Fig. 5A, B, and G), but there was no difference in HFD (Supplementary Fig. 2A–C). Further, in female KO mice, EE was significantly increased by 25% compared with KO males (Fig. 5A), but while it only showed a trend to increase EE on chow (Fig. 5C and G), a significant increase in EE on HFD (Supplementary Fig. 2A–C) compared with their Tx controls was observed. RER was similar between control and KO groups (Fig. 5D–F and H), although female KO mice on chow diet decreased their RER by 5% during the light phase compared with male mice (Fig. 5H).

Relative to Tx controls, male KO mice had a significantly lower number of cells (−42%) and membrane surface (−33%) (Fig. 6A and B). This change in morphology was reflected by a 20% decrease in uncoupling protein 1 (UCP1) immunoreactivity in male KO mice (Fig. 6D and E). On the other hand, no consistent difference between KO and control in iBAT histology and UCP1 immunoreactivity was observed in female mice (Fig. 6B–E).

Male and female mice showed no difference in baseline body temperature (Fig. 6F and G). Male and female KO mice did not have increased body temperature after sCT versus saline injection, while Tx control mice on chow had an average of 0.47°C increase of body temperature during the dark phase (Fig. 6H and I) and 0.49°C on HFD (data not shown).

Male and female KO mice were less active than Tx controls during dark and light phase, which resulted in a 30% reduction in cumulative 3-day activity (Fig. 7A–C). On HFD, the baseline activity was also reduced by 40% in KO mice versus controls (Fig. 7D–F). However, this baseline decrease in locomotor activity did not seem to directly affect EE, since opposite effects on EE were observed in male and female mice, while locomotor activity was decreased in both sexes.

For investigation of potential reasons for the decrease in spontaneous locomotor activity, plasma corticosterone and ACTH levels were measured from 11-week-old chow-fed mice, 2 h after light onset and after dark onset, representing circadian nadir and peak of circulating corticosterone, respectively (28) (Supplementary Table 2). Corticosterone was increased approximately threefold at peak secretion (P < 0.0001) compared with baseline, while ACTH was only increased by 12% on average (P = 0.028). However, there was no significant difference between groups at any time point. Because POMC neurons project to the LHA, a potential involvement of the LHA in altered locomotor activity was investigated, but no difference in the number of orexin A–positive LHA neurons in 12- to 15-week-old control and KO chow-fed mice (Fig. 7G and H) was detected.

Amylin injections resulted in a dose-dependent anorectic effect that persisted for up to 4 h in KO and control mice (Supplementary Fig. 3A–E). Compared with baseline food intake after 12-h overnight fasting and refeeding, amylin had a longer-lasting effect in POMC-Cre^CTR+oil and POMC-Cre^CTR+Tx mice than in POMC-WT^CTR mice (Supplementary Fig. 3F–J). sCT injections resulted in a food intake reduction that lasted up to 12 h (Supplementary Fig. 3K–O). Compared with baseline food intake after a 12-h fast, POMC-WT^CTR+oil showed a weaker response to sCT injection than the other groups (Supplementary Fig. 3P–T). Overall, the depletion of amylin signaling in POMC neurons did not affect the acute, AP-mediated effect of amylin or sCT on food intake.

Leptin and leptin + amylin combination had a significantly stronger effect on food intake 1 h after injection in control mice than in KO mice, although this difference did not persist after 4 h (Supplementary Fig. 4A–C). When the results are expressed as a percent of baseline, control and KO mice responded similarly to the anorectic effect of leptin and leptin + amylin (Supplementary Fig. 4D–F). Furthermore, leptin-induced pSTAT3 in POMC neurons in the ARC was also similar in control and KO mice, suggesting that the depletion of CTR in POMC neurons does not interfere with leptin signaling in these mice (10) (Fig. 8A and B). Leptin-induced pSTAT3 was assessed as a readout of leptin resistance in mice after 9 weeks on HFD. KO mice showed a 30% reduced STAT3 phosphorylation in the ARC compared with Tx controls (P = 0.01), which was not specific for POMC neurons (Fig. 8C and D). Nevertheless, the overall pSTAT3 and POMC immunoreactivity was 80% lower in the HFD-fed cohort compared with chow-fed mice (Fig. 8B and D), which could explain the blunted phenotypic effects of HFD-fed KO mice.

The goal of this study was to identify the physiological mechanisms controlled by amylin signaling in POMC neurons. We have previously shown that amylin selectively activates ERK signaling in POMC neurons to enhance ARC → PVN α-MSH fiber outgrowth (15), but whether amylin influences energy balance through this pathway during adulthood remained to be investigated. For achievement of this aim, Tx-inducible POMC-cre:ER^T2 mice were crossed with CTR-floxed mice. After Tx induction, only male POMC-CTR KO mice had an increase in body weight and FM, which was accompanied by a decrease in EE and UCP1 density in the iBAT. Both sexes were affected by a loss of sCT-sensitivity for body temperature regulation and a remarkable decrease in locomotor activity. Surprisingly, this phenotype was not exacerbated by HFD, which could be due to a general reduction of POMC expression. Further, the depletion of CTR in POMC neurons did not affect amylin’s acute anorectic action.

The selective loss of CTR in POMC neurons through cre:ER^T2 induction allows a normal axonal outgrowth of α-MSH fibers ARC → PVN, which is contrary to our previous KO models (15). We previously showed that amylin specifically activates ERK signaling in 60% of POMC neurons in the ARC and that amylin-induced pERK in POMC neurons is blunted in the ARC of RAMP1 and RAMP3 mice (15). These previous studies pointed out a crucial role of CTR- and amylin-mediated signaling in ARC POMC neurons. POMC neurons are highly pleomorphic in their physiological functions, connections, and transmitter expression (29–31). Thus, it is likely that only a subset of POMC neurons express CTR. In the POMC-CTR KO mice used in this study, ERK phosphorylation was impaired after exogenous amylin application, the density of CTR in the ARC of KO mice was reduced, and CTR was shown to be depleted specifically from POMC neurons in the ARC. The presence of CTR⁺POMC⁺ neurons was also detected in the NTS; however, these neurons are sparse, which is in line with a previous study comparing different POMC reporter mice (32). Some of the CTR⁺POMC⁺ neurons are colocalized with RAMP1 or RAMP3, but whether such a low number of neurons play a metabolic role in our model is questionable. Thus, the metabolic effects seen in the POMC-CTR KO mice could result from the loss of CTR in POMC neurons in the ARC and the NTS. A recent study by Cheng et al. (33) showed that CTR in the NTS is involved in the control energy balance through the parabrachial nucleus and that CTR^NTS neurons also send indirect signals to ARC neurons. In that study, CTR^NTS-depleted mice failed to respond to the anorectic effect of sCT and activation of these neurons acutely decreased food intake. After 5 weeks on chow diet, CTR^NTS-depleted mice displayed higher food intake compared with control similar to what we observed, but while chow diet–fed CTR^NTS-depleted and control mice had a similar FM, POMC-CTR KO mice displayed an increase in fat ratio, suggesting a different role for ARC and NTS CTR-mediated signaling. Since ∼20% of POMC neurons were CTR positive in the ARC, we can hypothesize that the decrease in pERK-POMC ARC signaling and the metabolic alterations seen in the KO mice could also result from indirect signaling via synaptic communication.

The eating response to acute exogenous amylin administration was not affected in our model, suggesting that acute central amylin effect on meal pattern may solely depend on the caudal hindbrain (34,35). Thus, the CTR KO seems to be sufficient to impair amylin signaling in this POMC subpopulation, confirming our hypothesis that amylin exerts a direct and AP-independent (15) effect on POMC neurons through its core receptor component. However, a recent study assessing the binding of fluorescent sCT to ARC neurons did not show binding to POMC neurons but fluorescent sCT was internalized into NPY neurons (9). The binding of fluorescent amylin was also assessed, but we could not see any binding at the cell level in either cell type, while whole brain imaging showed binding in ARC and median eminence (9). This finding was surprising given the fact that exogenous amylin did not activate ERK signaling in NPY neurons (15). Since it has recently been demonstrated that amylin can be produced by ARC neurons (36), we may therefore hypothesize that the effects observed in our study may be caused by brain-produced amylin instead of pancreatic amylin; the neuronal subtype producing amylin is currently under investigation in our laboratory. Further, since we showed that AMY1 is present in the ARC and NTS, albeit at a lower amount, we could also hypothesize that the effects observed in POMC-CTR KO may result from an alteration in the CGRP pathways, since CGRP, which is a widespread neurotransmitter, can bind to its secondary receptor AMY1 in the ARC but not in the NTS (37,38). The in situ hybridization revealed that ARC neurons can be CTR⁺, POMC⁺CTR⁺, CTR⁺RAMP⁺, or POMC⁺CTR⁺RAMP⁺, highlighting the complexity of the study of amylin signaling. Thus, the depletion of CTR in our model can affect either the calcitonin-mediated or the amylin-mediated signaling pathway.

The depletion of CTR in POMC neurons of male mice on chow diet increased weight gain and adiposity, which was characterized by an increased visceral FM and resulted in glucose intolerance. Baseline insulin levels and insulin tolerance were unchanged in this model, and circulating leptin levels corresponded as expected to the stored fat deposits. EE was significantly decreased in male KO mice, which could explain why the development of adiposity preempts a significant change in food intake. Amylin has been shown to increase EE by increasing sympathetic nerve activity, and nestin–human RAMP1–overexpressing transgenic mice also increased their EE (39,40). At least part of this effect could be attributed to amylin increasing POMC mRNA and activating the MC4R pathway that promotes satiety and thermogenesis (12,41), or a thermogenic effect through ERK phosphorylation, as has been shown with respect to leptin (42). The overall effect is relatively small, which makes sense considering that only a small subpopulation of neurons was altered in this mouse model and the hindbrain pathway seems to be unaffected. While baseline body temperature of our KO mice was normal, these mice did not respond to sCT injection by increasing thermogenesis as has previously been shown in intact rats and mice (39,43). Furthermore, iBAT morphology was altered in male KO mice with a decreased membrane surface and cell number. These findings support the idea that reduced activation of iBAT may contribute to the EE phenotype, and this is reflected with lower UCP1 immunoreactivity in iBAT of male KO mice. Conversely, female KO mice did not present any alteration in their EE, iBAT histology, and UCP1 activity and in consequence did not develop adiposity.

Sexual dimorphism has not yet been studied extensively regarding amylin signaling. However, the hypothalamic POMC system has been shown to differ between sexes, which has been attributed to a lower number, lower activity, and/or lower mRNA expression of POMC neurons in the male ARC, a development that is induced by testosterone exposure (44,45). Furthermore, estradiol-deficient DIO rats seem to have a better response to chronic exogenous amylin application regarding body weight and EE (46). Notably, the inactivation of STAT3 signaling in POMC neurons leads to a fat-specific weight gain similar to that seen in this study, albeit in this case only in female mice (47). We did not see any difference in the number of POMC neurons between sexes—contrary to the study by Wang et al. (44) using the same POMC-cre:ER^T2 mice induced at 11 weeks of age. Therefore, we believe that the absolute number of POMC neurons may be less important for the different phenotypes than their properties. A possible confounding factor in this regard is the Tx induction at 4 weeks, which, at least in females, possibly interferes with sexual maturation. Additionally, even low doses of Tx in adult female mice have been linked to substantial adipose tissue browning and increased thermogenesis (48), which may have masked the effects that we attempted to measure.

Both male and female KO mice showed a reduced locomotor activity, which, however, did not seem to be responsible for the changes in EE discussed earlier. For instance, the activity pattern does not match the EE data time-wise in a 24 h cycle, and in females it even contradicts their increased EE. Since POMC neurons project to the LHA (49,50) and since MC3R-expressing orexin neurons in the LHA modulate locomotor activity (51), we hypothesized that the decrease in amylin POMC signaling could affect LHA orexin neurons and, hence, contribute to the decrease in locomotor activity. However, when looking at orexin expression in the LHA, we could not find any differences between groups or sexes, and the activity pattern in metabolic cages also did not point to a fragmented sleep pattern as has been shown with ablation of orexin neurons or orexin receptor KO studies (52,53). Nevertheless, whether the depletion of amylin signaling in POMC neurons decreases orexin neurons’ activity was not assessed. POMC neurons are also prominent in the pituitary gland, but we could not find marked differences in baseline corticosterone and ACTH release. Furthermore, this decrease in locomotor activity did not correspond to decreased food foraging behavior, since male KO mice ate more overall and tended to have smaller and more frequent meals than Tx controls. Therefore, the underlying mechanism of this consistently observed change in locomotor activity is still unclear.

Given the increase in adiposity of chow-fed male POMC-CTR KO mice, we expected an exacerbation of this phenotype under HFD. However, the KO-specific effect was rather lost, which could be due to a marked reduction of POMC expression that prevented clear effects of the defective amylin signaling in POMC neurons on HFD. Indeed, recent pharmacological studies by Li et al. (41) showed that amylin directly uses the melanocortin system to increase thermogenesis. While leptin sensitivity was similar in chow- and HFD-fed KO mice, represented by a comparable pSTAT3 signaling, the HFD cohort had a lower overall pSTAT3 and POMC expression. However, when CTR is depleted in rat ARC and VMN using an AAV shRNA (which targets more cells than specifically POMC neurons), VMN leptin binding and leptin-induced pSTAT3 are decreased (6). We previously demonstrated that amylin activates the secretion of IL-6 by microglia, which then binds to its gp130 neuronal receptor to further enhance leptin-induced pSTAT3 (10,23). These studies suggest that amylin’s enhancing effect on leptin pSTAT3 signaling takes place mostly in NPY neurons, while direct pERK induction in POMC neurons is responsible for alterations in energy balance (10,15,23).

HFD-fed mice gradually become leptin resistant and have higher circulating levels of leptin in their blood. However, reports on the effect of various high-fat diets on POMC expression are ambiguous. A 12-week study with C57BL6 mice showed an upregulation of POMC mRNA (54), but other studies in mice and rats showed a relative downregulation of POMC in the obesity-prone phenotype in response to HFD (55,56). The time course study by Souza et al. (55) also demonstrated that NPY and POMC expressions do not uniformly change but, rather, fluctuate during the first days and weeks of HFD exposure compared with chow diet. They highlight that an early reduction of POMC in response to HFD characterizes an obesity-prone phenotype.

In conclusion, we demonstrated that amylin action in POMC neurons is more involved in energy homeostasis of male than female mice and may promotes EE by affecting iBAT thermogenesis. Notably, this influence persists into adulthood and goes beyond the previously established neurotrophic influence of amylin on fiber outgrowth in the development of a hypothalamic feeding circuit (15,23). However, the direct action of amylin on POMC neurons seems to play a diminished role in a leptin-resistant state of HFD-fed mice and leaves the question of how amylin improves leptin signaling in the hypothalamus. Given that only a fraction of pERK or pSTAT3 expression overlaps with POMC neurons in the ARC, other distinct cell populations are worth exploring. Since we have previously shown the crucial role of microglial amylin signaling during the early postnatal period with regard to the hypothalamic development (10), we hypothesize that these cells could also play a major role in the regulation of energy balance, as microglia seem to play a more prominent role in a state of dietary excess (57) and have been shown to play a crucial role in improving leptin signaling (10). Moreover, the phenotype observed in POMC-CTR KO mice could result from the contribution of ARC and NTS signaling, and it will be worth exploring the contribution of each brain nucleus. Finally, our research underlines the importance of considering both sexes in studies on energy homeostasis, as the therapeutic benefit from interventions in hypothalamic circuits could vary greatly between sexes.

Acknowledgments. The authors acknowledge the technical contributions of Josep Monné Rodríguez (Center for Microscopy and Image Analysis, Vetsuisse Faculty, Institute of Veterinary Pathology, University of Zurich), Petra Seebeck (Zurich Integrative Rodent Physiology), and Joel Elmquist (UT Southwestern, Dallas, TX) and Dr. Thorsten Schinke (University Medical Center Hamburg-Eppendorf, Hamburg, Germany) for providing the POMC-cre:ER^T2 and CTR-floxed mice, respectively. The authors also thank Christina N. Boyle (Vetphysiologie, UZH), Fabienne O. Villars (Vetphysiologie, UZH), Salome Gamakharia (Vetphysiologie, UZH), and Justyna B. Koczwara (Sapienza University of Rome) for technical assistance.

Funding. This work was funded by the Swiss National Science Foundation (grant SNF 31003A_175458 to T.A.L.) and UZH Forschungskredit (grant FK 17-066 to C.L.F.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. C.L.F. and T.A.L. contributed to study conceptualization. C.L.F. designed the experiments. B.C., C.K.-H., and C.L.F. performed the experiments. B.C. wrote the original draft of the manuscript. C.L.F. and T.A.L. reviewed and edited the manuscript. T.A.L. and C.L.F. acquired funding. C.L.F. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 27th Annual Meeting of the Society for the Study of Ingestive Behavior, Utrecht, the Netherlands, 9–13 July 2019.