Evaluation of a melanocortin-4 receptor (MC4R) agonist (Setmelanotide) in MC4R deficiency

Pro-opiomelanocortin (POMC)-derived peptides act on neurons expressing the Melanocortin 4 receptor (MC4R) to reduce body weight. Setmelanotide is a highly potent MC4R agonist that leads to weight loss in diet-induced obese animals and in obese individuals with complete POMC deficiency. While POMC deficiency is very rare, 1–5% of severely obese individuals harbor heterozygous mutations in MC4R . We sought to assess the efficacy of Setmelanotide in human MC4R deficiency.

In cells, we showed that Setmelanotide is significantly more potent at MC4R than the endogenous ligand alpha-melanocyte stimulating hormone and can disproportionally rescue signaling by a subset of severely impaired MC4R mutants. Wild-type rodents appear more sensitive to Setmelanotide when compared to MC4R heterozygous deficient mice, while MC4R knockout mice fail to respond. In a 28-day Phase 1b clinical trial, Setmelanotide led to weight loss in obese MC4R variant carriers. Patients with POMC defects upstream of MC4R show significantly more weight loss with Setmelanotide than MC4R deficient patients or obese controls.

Neural circuits in the hypothalamus play a critical role in the regulation of energy homeostasis [1] , [2] . Peripheral nutrient-derived signals such as leptin stimulate the expression of Pro-opiomelanocortin (POMC)-derived melanocortin peptides, which reduce food intake and body weight by acting as endogenous ligands for the brain-expressed melanocortin-4 receptor (MC4R) [3] , [4] . MC4R knockout mice are severely obese while loss of one MC4R allele results in an intermediate obesity phenotype demonstrating that weight regulation is sensitive to quantitative variation in MC4R expression [5] . Obesity-associated MC4R mutations confirm a role for MC4R-mediated signaling in human energy homeostasis [6] , [7] , [8] , [9] , [10] . Additionally, MC4R activation may be modulated by accessory proteins such as melanocortin receptor accessory protein 2 (MRAP2) [11] , [12] , [13] . In view of these effects, MC4R, which is a seven-transmembrane domain G-protein coupled receptor (GPCR), has been considered as a potential drug target for the treatment of obesity [14] , [15] , [16] . However, first generation MC4R agonists resulted not only in suppression of food intake and induction of weight loss, but also in a significant increase in blood pressure (BP) and heart rate (HR) in rodents, primates, and humans [16] , [17] , [18] , [19] . For example, treatment of obese volunteers with the agonist LY2112688 at the maximum dose of 1.0 mg/day led to significant increases of systolic (mean 9.3 ± SD 1.9 mmHg) and diastolic (mean 6.6 ± SD 1.1 mmHg) blood pressure after only 24 h of treatment compared with placebo [18] . These adverse effects halted the development of the first generation of MC4R agonists. Setmelanotide (formerly known as RM-493 or BIM-22493) is a synthetic cyclic peptide that binds to human MC4R with high affinity, activates MC4R at nanomolar concentrations (50% effective concentration [EC 50 ] = 0.27 nM), and has been shown to reduce food intake and body weight and restore insulin sensitivity in diet-induced obese (DIO) mice, rats, dogs, and monkeys [20] , [21] . Comparable weight loss effects have been observed in genetic animal models of obesity treated with Setmelanotide, although not in MC4R knockout mice [22] , indicating that the effects of Setmelanotide are indeed due to its action on MC4R. Importantly, in Rhesus monkeys with diet-induced obesity, Setmelanotide infused at 0.50 mg/kg/day for 8 weeks led to significant decreases in food intake (∼40%) and body weight (∼13%) without acute or long-term increases in BP or HR measured by continuous telemetry [20] . Additionally, drug-related changes in heart rate and blood pressure were not observed in a series of Setmelanotide phase 1b/2a or experimental studies in obese individuals [21] . In a recent proof-of-principle study, two rare POMC deficient individuals treated with Setmelanotide experienced significant weight loss (51.0 kg of body weight loss after 42 weeks in patient 1 and 20.5 kg of body weight after 12 weeks in a second patient) without any adverse effects on cardiovascular parameters [23] . These results established that while obesity due to complete POMC deficiency is very rare, it is potentially very responsive to treatment with Setmelanotide. Heterozygous loss of function (LOF) mutations in MC4R are found more frequently with prevalence estimates ranging from 1 to 5% in children or adults with severe obesity [9] , [10] , [24] . This group of individuals are relatively refractory to weight loss through diet and exercise [25] and other treatment options remain limited. Roux-en-Y bypass surgery can be effective, but is not without risks and not suitable for many people [26] . We hypothesized that Setmelanotide might lead to weight loss in MC4R deficiency by rescuing signaling by mutant MC4Rs and/or by increasing signaling through the wild type MC4R allele. We therefore performed a series of studies, which included a detailed interpretation of the functional status of 369 MC4R variants as well as preclinical and clinical studies in a small subset of variants to address the impact of MC4R mutations or Setmelanotide treatment on energy homeostasis in MC4R deficiency.

To assess the functional consequences of all published MC4R mutations, we searched all studies on functional cell-based assays of MC4R variants. We systematically reviewed articles containing the terms “MC4R” or “mc4-r” indexed on MEDLINE from inception to August 31st, 2016 in addition to bibliographies of manually selected articles, mutations published by human genetic databases (1000 Genomes release 16, Exome Variant Server ESP6500SI-V2) and unpublished data from our labs. Because functional consequences of MC4R mutations are assessed and reported in a wide variety of ways, we devised an a priori semi-quantitative scale of the overall residual function. Each non-synonymous mutation could lead to a complete LOF, a partial LOF, or no LOF (wild type like). In addition, contradictory and unclear published data resulted in ambiguous status for many variants. Functional assay data were collected for each published MC4R variant: cell line, cell surface expression, tested ligands (α-MSH, β-MSH, γ1-MSH and artificial agonists such as NDP-α-MSH, melanotan [MT-II]), E max (efficacy, the maximum level of cAMP production in response to increased doses of a given ligand), EC 50 (potency, the concentration of a ligand eliciting half of the maximum response in cAMP production), and in a competitive binding assay B max (the maximal binding capacity to a ligand) and IC 50 (affinity, the concentration of a ligand binding half the receptors). All functional data were normalized to wild type levels, to enable comparison between articles ( Figures S1–S2 ). The assay data were independently abstracted by three authors (T.-H.C., B.D., A.G.) and compared to reach a consensus on the residual function of each variant. Discrepancies were discussed among co-authors to reach a consensus on the residual function of each MC4R variant. Ambiguous variants were re-evaluated in a head-to-head comparison of activity in cell-based assays (see section Functional characterization of additional MC4R variants).

In order to characterize newly identified MC4R variants or to clarify classification of variants for which the systematic review of published data returned ambiguous results, we performed in vitro assays in transiently transfected HEK293 cells. We assessed relative cell surface expression of the mutant receptors in cell-based ELISA and we employed the cAMP response element-binding protein (CREB) trans-reporting system (Stratagene) to examine the impact of these variants on the receptor signaling activity via cAMP dependent pathway. MC4R cDNA constructs barring an N-terminal FLAG tag in pCDNA3.1(+) vector (Invitrogen) were used throughout the study. Site-directed mutagenesis was performed using QuikChange kit (Agilent) according to the manufacturer's protocols. All constructs were verified with Sanger sequencing. HEK293 cells were maintained in high glucose Dulbecco's modified eagle medium (DMEM, Gibco, 31966) supplemented with 10% fetal bovine serum (Gibco, 10270, South America origin), 1% GlutaMAX™ (100X) (Gibco, 35050), 100 units/mL penicillin, and 100 μg/mL streptomycin (Sigma–Aldrich, P0781). Cells were incubated at 37 Celsius degrees in humidified air containing 5% CO 2 and transfections were performed using Lipofectamine 2000™ (Gibco, 11668) in serum-free Opti-MEM I medium (Gibco, 31985) according to the manufacturer's protocols. Briefly, 40,000 cells/well were seeded in a poly- d -lysine coated 96-well plate and transfected the next day with either 30 or 5 ng of MC4R constructs, for cell surface expression ELISA and CREB reporter assay, respectively. 2.4.1 Cell surface expression Cells transfected with FLAG-tagged MC4R constructs were tested for cell surface localization of the receptor with ELISA. Two days after transfection, cells were fixed with 3.7% paraformaldehyde for 15 min at room temperature and washed three times with PBS. Subsequently, non-specific binding sites were blocked with 3% non-fat dry milk in 50 mM Tris-PBS pH 7.4 (blocking buffer) for 1 h at room temperature. Next, cells were incubated with a mouse monoclonal anti-FLAG (M2) antibody (Sigma–Aldrich, F1804) (dilution 1:1000 in blocking buffer) overnight at 4 Celsius degrees followed by triple washing with PBS and incubation with goat anti-mouse IgG(H + L)-HRP conjugate (Bio-Rad Laboratories, 172-1011) (1:1250 in 1.5% non-fat dry milk in 50 mM Tris-PBS pH 7.4) for 2 h at room temperature. Finally, plates were washed three times with PBS and a high performance chromogenic substrate – 3,3′,5,5′ tetramethylbenzidine (TMB CORE+, Bio-Rad Laboratories, BUF062) was used to detect HRP activity. The reaction was terminated with 0.2M H 2 SO 4 and absorbance by the color reaction product at 450 nm was determined using Infinite M1000 PRO microplate reader (Tecan).

For the duration of the Setmelanotide studies, rodents were single-housed in an IACUC approved facility at Vanderbilt University. Wildtype ( Mc4r +/+ ), heterozygous ( Mc4r +/− ) and MC4R null ( Mc4r −/− ) mice were placed on a high fat diet (Research Diets, Inc. #D12451, 45% kcal fat, 35% kcal carb, 20% kcal protein) at 11 weeks of age, which continued for the duration of the experiment. At 13 weeks of age, the animals were placed in a Promethion Sable monitoring system for 3 days to acclimatize and to obtain baseline measurements. After the acclimatization period, animals were weighed and surgically implanted with primed Alzet minipumps (Alzet #1002, 0.25 μL/h infusion rate) delivering saline alone, or 1200 nmol/kg/day of Setmelanotide dissolved in saline. The pumps were placed subcutaneously on each animal's upper back. Animals were monitored for food and water intake, then removed from the Promethion cages, and weighed once more 9 days after the implant. Statistical analyses were performed using Graphpad Prism 5.

3.1 A single system for functional annotation of MC4R mutations We generated a single functional classification system for variants in MC4R as not all variants impact MC4R signaling in cell based assays and this has caused some confusion in the field. Moreover, while some classification schemes have been proposed [14] , [27] , [28] , [29] , these remain difficult to apply uniformly given the discordant and/or incomplete functional data available in different publications. We systematically reviewed and collated data on all published MC4R variants and included unpublished data on 75 variants from our labs ( Figure 1 A, Figures S1–2 ). In total, we identified missense (n = 318, 69.7%), nonsense (n = 24; 5.3%), and frameshift mutations (n = 21; 4.6%) and deletions (n = 6; 1.3%). There were inconsistencies between the reported functional consequences of a subset of variants. To enable reliable classification of these variants, we performed functional assays on 22 MC4R variants with unknown, incomplete or ambiguous functional status for which we had clinical data; 5 resulted in a complete LOF, 14 in a partial LOF and 3 were wild type (WT)-like ( Figure 1 B–D, Table S1 ). We have made the functional classification dataset with MC4R functional assay data, available as a searchable resource ( Table S2 , www.mc4r.org.uk ). Figure 1 Genetic and molecular classification of MC4R mutations. (A) MC4R is a seven transmembrane domain GPCR with 3 intracellular loops (ICL), 3 extracellular loops (ECL), and a C-terminal helix 8 (H8); schematic adapted from GPCRDB Tools [41] . Colors denote amino acids that have been altered by mutations. Functional consequences of these mutations were classified following a systematic review of the literature ( Figure S1 ). Where multiple variants affecting the same amino acid have been reported, the most severe loss of function (LOF) mutation is shown. (B–D) Functional characterization of 22 new MC4R mutations. α-MSH stimulated cAMP generation was assessed in transiently transfected cells using a CREB-luciferase reporter assay ( Table S1 ). Data are mean ± SEM of 4 experiments. (E–G). Genotype-phenotype correlations in 249 children with complete or partial loss of function MC4R mutations (cLOF/pLOF respectively). Data are represented as mean ± SEM. (E) Body mass index (BMI) standard deviation scores (SDS) of children who were homozygous (Hom) for cLOF (n = 16) or pLOF mutations (n = 17) were compared to those who were heterozygous (Het) for cLOF (n = 131) or pLOF (n = 85) MC4R mutations. Homozygotes had a significantly higher BMI SDS than those with heterozygous MC4R mutations (p < 0.001 with and without adjustment for age, sex and ethnicity), while there was no statistically significant difference between those with cLOF or pLOF mutations within the same gene dosage group (both p = 1.00). (F) Ad libitum food intake adjusted for lean mass measured by dual energy X-ray absorptiometry in 57 children (age < 18 years) (Hom cLOF (n = 6); Hom pLOF (n = 3); Het cLOF (n = 29); Het pLOF (n = 19)). (G) Ad libitum food intake adjusted for lean mass in 35 children age < 10 years (Hom cLOF (n = 5); Hom pLOF (n = 2); Het cLOF (n = 17); Het pLOF (n = 11)) (data on obese children have been reported previously [10] . Figure 1 Figure 2 Effects of Setmelanotide in cells expressing different MC4R variants. The cAMP response to α-melanocyte stimulating hormone (α-MSH) and Setmelanotide was studied for wildtype (WT) and 17 mutant forms of MC4R; mean EC50 (concentration required to elicit 50% of the maximal response) of experiments performed in duplicate with 11 point dose titrations. Some variants (R18C, V50M, I102T, G231S, R305W) have been studied previously and values were taken from the relevant reference [42] . The ability of Setmelanotide to rescue signaling by MC4R mutants was calculated (EC50 of α-MSH/EC50 of Setmelanotide) and varied considerably between mutants. The ratio could not be precisely determined for some variants (G55D, E61K, C271Y) because of the high concentrations of α-MSH needed. Figure 2 Using this functional classification, we examined phenotypic data from 224 children and 264 adults carrying MC4R mutations. We found that MC4R variant homozygotes were significantly more obese than heterozygotes (p < 0.001); there was no significant difference in BMI between those carrying heterozygous complete vs partial LOF variants ( Figure 1 E). We did not identify significant genotype–phenotype correlations for most phenotypes after adjusting for age, gender, and BMI (data not shown). Of interest, we found that children heterozygous for complete LOF variants consumed slightly more at an ad libitum test meal (mean 121.1 kJ/kg lean mass ± SEM 12.3) than those with partial LOF mutations (mean 82.2 kJ/kg of lean mass ± SEM 10.4), although this was not statistically significant (p = 0.07, Figure 1 F). This effect was more pronounced when analyzing the data for children under 10 years of age (me

We studied the effect of Setmelanotide administration in mice lacking Mc4r . Wild-type ( Mc4r +/+ ), heterozygous null ( Mc4r +/− ), and a small number of homozygous null ( Mc4r −/− ) mice were placed on a high fat diet (45% kcal fat) and then implanted with subcutaneous osmotic pumps containing either 1200 nmol/kg/day Setmelanotide or vehicle (saline); body weights were monitored for 9 days. All saline treated groups gained weight in response to the high fat diet. Mc4r +/+ mice treated with Setmelanotide exhibited weight loss, while Mc4r −/− mice treated under the same conditions, gained a comparable amount of weight to their vehicle-treated controls. Setmelanotide-mediated weight loss seemed to be mediated by a reduction in food intake, rather than changes in activity or energy expenditure ( Figure S3 ), although the magnitude of the effect was small in these studies. Interestingly, Mc4r +/- mice exhibited an intermediate response to Setmelanotide treatment as evidenced: i) by a reduction in absolute weight gain over the period ( Figure 3 A); ii) when comparing weight gain to baseline in treated mice ( Figure 3 B); and iii) the Setmelanotide-mediated improvement in body weight over saline ( Figure 3 C) (the latter corrects for the weight gain seen in vehicle-treated mice on the high fat diet). Overall, it is apparent that Mc4r +/+ and +/− mice lose weight with Setmelanotide, with wild-type mice being the most sensitive to treatment, while Mc4r −/− mice do not respond (as also shown previously [22] ). Figure 3 Effect of Setmelanotide administration in mice lacking Mc4r. Mc4r +/+ , +/− , and −/− mice were infused with 1200 nmol/kg/day of Setmelanotide, or vehicle (saline), using an implanted subcutaneous minipump. (A) Body weight before (day 0) and after (day 9) among all treatment groups. (B) Cumulative 9-day weight change among all Setmelanotide treated groups. Each symbol represents a single test animal's change in body weight. Statistical significance between groups is calculated using one-way ANOVA with Bonferroni post-test. (C) Setmelanotide mediated 9-day weight improvement over saline treated groups. These values are calculated by subtracting the average body weight change in saline treated groups from the average body weight change in Setmelanotide treated groups (ΔSetmelanotide − ΔSaline). A treatment that causes no change compared to saline would be represented by a weight difference of zero. Values displayed are Mean +/− SD and statistical significance is calculated using a T-test versus saline controls. *P < 0.05; **P < 0.01. Figure 3

Our findings demonstrate that the melanocortin 4 receptor agonist, Setmelanotide can lead to weight loss in obese people with MC4R deficiency (on average 0.6 kg/week). Importantly, Setmelanotide induced weight loss over 4 weeks without eliciting an increase in HR or BP, in contrast to other MC4R agonists tested to date. Additional studies are needed to determine whether the data presented here are effective predictors of weight loss in larger studies enrolling cohorts of patients carrying complete and partial MC4R LOF variants. Such studies may also provide information about the mechanisms by which weight loss occurs. A potential mechanism that could contribute to the efficacy of Setmelanotide in such patients is a chaperone effect, whereby agonists (or antagonists) can rescue mutant forms of the MC4R by restoring their expression at the cell surface. A number of prior studies have also shown that a subset of MC4R variants appear to be rescued in this manner when studied in cells with distinct MC4R agonists [31] , [32] , [33] . In a head-to head comparison of Setmelanotide and α-MSH, we found that some variants were rescued more efficiently than others, especially those with a high mutant/wild type EC 50 ratio. At present the structure of the MC4R has not been determined, although homology modeling on MC1R has been used to suggest possible structural explanations for the molecular mechanisms by which mutations result in a LOF [31] . Interestingly, we find an overrepresentation of non-synonymous variants in transmembrane domain (TM)-I, intracellular loop-1 (ICL-1), TM-II, the intracellular part of TM-VII, and helix 8, domains which form a so-called hydrogen bond network which has been demonstrated to play an important role in receptor activation and signaling specificity [34] , [35] . There are surprisingly few LOF variants in the N-terminal domain and in extracellular segments of the inward looking parts of transmembrane domains which constitute the large binding pocket (TM-III, TM-IV, TM-V, TM-VI, TM-VII). Further studies will be needed to more fully characterize the mutants reported here and the mechanisms by which a subset are rescued by Setmelanotide. Given the large number of possible MC4R variants, we were fortunate to find one patient enrolled who carried the heterozygous C271Y MC4R variant, which was earlier, independently classified as being disproportionally efficiently rescued by Setmelanotide in cells. Indeed, this patient was one among three complete LOF MC4R variant carriers treated with Setmelanotide who showed a profound weight loss response (0.9 kg/week). This could be a basis for future patient stratification whereby certain patients might benefit more readily from Setmelanotide treatment than others. Possible reasons for the cardiovascular profile of Setmelanotide, compared to previous MC4R agonists, might be that some lead compounds tested to date differentially penetrate the CNS [36] and, at these sites, may couple through different G-protein signaling pathways. Recently, Weinstein and colleagues showed that disruption of melanocortin receptor Gα s and Gα q11 signaling in the paraventricular nucleus of the hypothalamus, led to divergent phenotypes in rodents [37] , and several studies have shown that MC4R signaling is more complex than previously thought including evidence of biased signaling [38] and evidence for diverse ligands exhibiting distinct temporal cellular signaling selectivity [39] . We have shown that individuals with MC4R deficiency as well as those expressing the wild type MC4R lose weight following Setmelanotide treatment. Current preclinical and clinical MC4R pathway deficiencies treated with Setmelanotide indeed predict an order of Setmelanotide treatment sensitivity where MC4R pathway upstream mutations (POMC) are more sensitive to Setmelanotide treatment than wild type genotypes [40] , which appear, in turn, more responsive to Setmelanotide than most MC4R LOF variants. We anticipate that stratification of patients harboring mutations in the leptin-melanocortin pathway, which include the pro-opiomelanocortin, proprotein convertase subtilisin/kexin type 1, leptin receptor and defects in MC4R, will help targeting Setmelanotide to the genetic subgroups that can optimally respond to treatment with this promising drug candidate.

LVDP, ISF, and KC designed the study and wrote the paper with THC, AG and BD who led the functional classification and genotype–phenotype correlation analysis. HC, AG, SS, CF, BH, LS, KG, and LVDP designed and performed the clinical trial of Setmelanotide and analyzed the resulting data. JM and EMO performed the functional studies and analyzed the resulting data of UK patient mutations; JMK, EH, RB, SOR, and ISF recruited UK patients to the GOOS cohort, collected and analyzed the clinical data; BD, PT, CPB, JMO, FM, and KC recruited patients to the French cohort, collated and analyzed the clinical data. RA performed the functional analysis of the French mutations and JML, JLB, and DP the genetic screening in the French patients. BP and RC conducted the murine studies.

The following are the supplementary data related to this article: