The Role of Melanocortin 3 Receptor Gene in Childhood Obesity

The MC3R gene was analyzed in 201 unrelated children and adolescents with early-onset severe obesity, recruited from the general clinics of the nationwide School Health Service. Inclusion criteria were percentage ideal weight for height (WFH) of >140% and onset of obesity before 6 years of age. There were 128 boys (63.7%); the ethnic distribution was as follows: Chinese 105 (52.2%), Malay 69 (34.3%), Indian 21 (10.4%), and others 6 (3%). The age (mean ± SD) was 11.1 ± 3.0 years, WFH 170.5 ± 22.7%, BMI 31.9 ± 5.5 kg/m², and body fat 40.7 ± 5.2% by dual-energy X-ray absorptiometry (DEXA) or 45.7 ± 9.1% by bipedal bioimpedance analysis (BIA) for body fat composition. This research was approved by research and ethics committee of our hospital, and informed written consent was obtained from all subjects and parents.

Blood samples were obtained in the fasted state. Leptin was measured using DSL-10-23100 Human Leptin Enzyme-Linked Immunosorbent kit (Diagnostic Systems Laboratories, Webster, TX). Insulin resistance was calculated using the homeostasis model assessment (HOMA) (14), and insulin sensitivity was calculated by quantitative insulin sensitivity check index (QUICKI) (15). Bone density and body fat composition were assessed using DEXA (model XR-36; Norland, Fort Atkinson, WI). BIA was performed using the portable Tanita Body Composition Analyzer (TBF-300GS; Tanita, Tokyo).

Qualitative food-intake evaluation was performed using the Three-Factor Eating Questionnaire, a psychometric instrument to study eating behavior, measuring three dimensions: restraint, disinhibition, and hunger (16). Restrained eating is defined as the tendency to restrict food intake to control body weight. Disinhibition is the inability to resist emotional and social eating cues. Hunger is the subjective feeling of hunger. Numeric scores for each of the three factors were derived for each subject. Data concerning estimation of their physical activities were also obtained by using a physical activity questionnaire modified from the Modified Activity Questionnaire for Adolescents and validated in our local population (17,18), and each subject was categorized depending on physical activity for the past month: 1, inactive; 2, relatively inactive; 3, light physical activity; 4, moderately active; and 5, vigorous physical activity.

Genomic DNA was extracted from peripheral leukocytes. The single exon was amplified by PCR and sequenced as previously described (13).

Three novel mutations, Ile183Asn, Ala70Thr, and Met134Ile, were found during the screening process. Genomic DNA samples (n = 188) from 99 healthy children with normal height and weight (age [mean ± SD] 7.1 ± 4.6 years, BMI 16.8 ± 3.4 kg/m², 63 males) and 89 healthy adults (mean BMI <25 kg/m², 30 males) were analyzed as normal control subjects. The three novel mutations were not found in these 188 normal control subjects. Allele-specific PCR for Ile183Asn was performed as previously described (13). The other methods used are restriction enzyme digest using HhaI for Ala70Thr (G208A) and allele-specific PCR for Met134Ile.

The two common variants Thr6Lys and Val81Ile were always found together, and we confirmed that they resided on the same allele (haplotype) in the heterozygotes. Allele-specific PCR of Thr6 and Lys6 alleles were performed using selective forward primers, and amplified DNA was then incubated with BseDI. Only amplicons with Val81 (triplet GTT) would be cut.

Wild-type, variant (Thr6Lys and Val81Ile), and mutant (Ala70Thr, Met134Ile, and Ile183Asn) MC3R were directly amplified from individual genomic DNA using Pfx DNA Polymerase (Invitrogen Life Technologies, Carlsbad, CA). Forward and reverse primers were designed with HindIII and XbaI enzyme restriction sequences at the 5′ end. The Renilla luciferase reporter gene was amplified from the phRG-TK reporter vector (Promega, Madison, WI) with a pair of primers containing the KpnI and XhoI restriction sites at the 5′ end. All generated amplicons were TA cloned into the pDrive vector using the QIAGEN PCR Cloning Plus kit (Qiagen, Hilden, Germany).

All MC3R inserts were excised from the pDrive/MC3R constructs using HindIII and XbaI and ligated into the pBudCE4 mammalian dual-expression vector (Invitrogen Life Technologies) under the control of the cytomegalovirus promoter while the Renilla luciferase reporter gene was ligated into the KpnI and XhoI site of the vector under the control of the elongation factor-1α promoter. For creating stable cell lines expressing the receptors, amplified MC3R gene was TA cloned into the pcDNA5/FRT/V5-His vector (Invitrogen Life Technologies).

One day before transfection, subconfluent Griptite human embryonic kidney cells (Invitrogen Life Technologies) were trypsinized and seeded in 96-well plates at a density of 1 × 10⁴ cells/well. Transient transfection was performed using the Effectene transfection reagent (Qiagen) with 90 ng pCRE-Luc (Stratagene, La Jolla, CA) and 10 ng pBudCE4/MC3R/Renilla construct per well.

Stably expressed MC3R cells were created with the Flp-In System (Invitrogen Life Technologies) as recommended by the manufacturer. Briefly, pcDNA5/FRT/V5-His/MC3R and pOG44 vectors were transfected into Flp-In 293 cells at a ratio of 9:1. After 48 h, the cells were trypsinized and seeded into T75 flasks containing complete Dulbecco's modified Eagle's medium (DMEM) and 100 μg/ml hygromycin B for selection.

Twenty-four hours after transient transfection, cells were washed with PBS and stimulated with α-MSH (Sigma-Aldrich, St. Louis, MO) with concentrations ranging from 1 pmol/l to 1 μmol/l in DMEM containing 0.2% BSA for 6 h. Cells were then washed and lysed with 1× passive lysis buffer and then assayed for luciferase activity using the Dual Luciferase Reporter Assay System (Promega) in the TD-20/20 luminometer (Turner Biosystems, Sunnyvale, CA). cAMP-induced firefly luciferase readings from pCRE-Luc expression were normalized with Renilla luciferase readings expressed from the pBudCE4/MC3R/Renilla plasmid construct. Normalized luciferase readings were then expressed as fold activity over basal (nonstimulated transfected cells). Data points were fitted by nonlinear regression analysis using GraphPad Prism software (version 4.00 for Windows; GraphPad, San Diego, CA).

To investigate effects of receptor dimerization between wild-type and mutant receptors, stable cells expressing wild-type receptors were transfected with pBudCE4/mutant constructs, pCRE-Luc, and phRG-TK reporter vector. Receptor stimulation and luciferase assay were then performed similarly as above.

Stable cell lines expressing receptors were grown in selective media until subconfluent. Cells were washed with ice-cold PBS and scraped off, pelleted, and then resuspended in 2 ml ice-cold homogenization buffer (20 mmol/l HEPES, pH 7.0, 1 mmol/l MgCl₂, 1 mmol/l EDTA, and 1 mmol/l phenylmethylsulfonyl fluoride). The cell suspension was homogenized in Dounce glass homogenizer on ice. The homogenate was centrifuged at 1,700g for 10 min at 4°C, and the supernatant was then collected and centrifuged at 40,000g for 40 min at 4°C. The pellet containing crude cell membranes was then resuspended in homogenization buffer, aliquoted, and stored at −80°C immediately. Protein concentration of crude membranes was determined using Bio-Rad Protein Assay reagent (Bio-Rad, Hercules, CA) with BSA as the standard.

Binding studies were performed using time-resolved fluorometry (TRF) technology with europium-labeled ligand. Twenty-four hours after transient transfection in 96-well plates, cells were washed once in PBS and incubated for 2 h with increasing concentrations of europium-labeled [Nle4, D-Phe7] (NDP)-αMSH (PerkinElmer, Boston, MA) in binding buffer (25 mmol/l HEPES, pH 7.0, 1.5 mmol/l CaCl₂, 1 mmol/l MgSO₄, 100 mmol/l NaCl, 25 μmol/l EDTA, and 0.2% BSA). Nonspecific binding was determined in the presence of 3 μmol/l unlabeled NDP-αMSH. After incubation, cells were washed four times with wash buffer (50 mmol/l Tris-Cl, pH 7.5, and 5 mmol/l MgCl₂) and incubated with 100 μl/well DELFIA Enhancement Solution (PerkinElmer) at room temperature for 15 min. Fluorescence were measured in the Wallac Victor²-V 1420 multilabel HTS counter (PerkinElmer) fitted with filters for europium TRF measurements.

Saturation studies using crude cell membranes were performed similarly in 100 μl HEPES binding buffer using 2 μg membrane protein, in Acrowell filter plates (Pall, East Hills, NY). Reactions were terminated by rapid filtration using a vacuum manifold with four washes of wash buffer before the addition of the DELFIA enhancement solution.

Competitive ligand binding studies were performed in 100 μl HEPES binding buffer containing 2 μg crude membranes, europium-labeled ligand at a concentration close to the kilodalton values, and competitor ligands (NDP-αMSH and α-MSH). Incubation conditions were similar as for saturation studies, and plates were processed as described above.

Griptite 293 cells grown on glass coverslips were transiently transfected with 200 ng pBudCE4/MC3R/Renilla using the Effectene reagent. After 48 h of incubation, the cells were washed once in PBS and fixed with 3.7% paraformaldehyde for 15 min at room temperature. The cells were washed three times with PBS and then incubated with blocking buffer (5% BSA and 2% fetal bovine serum) for 30 min at room temperature. After rinsing in PBS, cells were incubated with polyclonal rabbit anti-MC3R IgG (Santa Cruz Biotechnology, Santa Cruz, CA) in 1% BSA for 2 h at room temperature. Cells were washed four times with PBS and then incubated with goat anti-rabbit IgG conjugated with Alexa fluor 488 dye (Molecular Probes, Eugene, OR) in 1% BSA for 1 h at room temperature. Cells were then subjected to five washes with PBS, and coverslips were mounted with ProLong Gold Antifade reagent (Molecular Probes). Fluorescence was visualized with the Carl Zeiss Axioskop microscope (×100).

Comparison of parameters was performed with Student's t test; general linear model with covariates analyses to adjust for age, sex, and ethnicity; one-way ANOVA trend analysis; Mann-Whitney U test; Fisher's exact test; and χ² test where appropriate (SPSS, Chicago, IL). Graph fit curves and half-maximal effective concentration (EC₅₀)/half-maximal inhibitory concentration (IC₅₀) were obtained and performed using GraphPad Prism 4 for Windows (v. 4.02 GraphPad software), and curves were fitted using the logistical equation. K_i values were determined from IC₅₀ values using the equation of Cheng and Prusoff (19).

We found three rare variants or potentially novel missense mutations, Ile183Asn (T548A), Ala70Thr (G208A), and Met134Ile (G402A), in the heterozygous state in three unrelated subjects (Fig. 1A–C) and two common variants, Thr6Lys and Val81Ile. The three potentially novel mutations Ile183Asn, Ala70Thr, and Met134Ile were not found in 188 nonobese normal control subjects.

We confirmed the Thr6Lys (17C>A) and Val81Ile (241G>A) variants were in complete linkage disequilibrium and located on the same allele (haplotype) in our cohort and normal control subjects. Of 198 subjects (excluding the three with novel variants), 121 subjects were homozygous for the wild-type allele, 70 subjects were heterozygous, and 7 subjects were homozygous for the allele (haplotype) with common variants. Eighty-eight DNA samples from the 99 nonobese children (control subjects) were randomly selected for genotyping for the common variants, and five DNA samples failed to yield satisfactory PCR products or sequence readings. Among these 83 nonobese children, 31 were found to be heterozygous and 2 homozygous for the 6Lys/81Ile haplotype (henceforth identified as 6K/81I). No significant difference in genotypic frequencies was detected when compared with the obese group (P = 0.843).

Obese subjects with 6K/81I had significantly higher leptin levels, percentage body fat, and insulin sensitivity index QUICKI, with lower insulin resistance index HOMA. The causative role of the 6K/81I variants is further supported by the presence of an additive effect, where the heterozygotes had an intermediate phenotype compared with homozygotes (Table 1). In addition, the obese subjects heterozygous and homozygous for 6K/81I had significantly lower triglyceride levels (1.19 ± 0.50 vs. 1.41 ± 0.73 mmol/l, P = 0.012) and lower fasting glucose levels (4.4 ± 0.5 vs. 4.8 ± 0.8 mmol/l, P = 0.024) compared with obese subjects with wild-type MC3R alleles. We did not detect any significant differences in other parameters, including insulin and C-peptide levels, insulin-to-glucose ratios, frequency of diabetes/impaired glucose tolerance, physical activity pattern, or feeding behavior scores. Our clinical findings are supported by in vitro studies, which revealed a modest but significant decrease (P < 0.01) in maximal activity of the MC3 receptor with variants 6K/81I compared with wild-type MC3 receptor (Fig. 2A), without any significant difference in binding affinity to α-MSH or NDP-MSH (Fig. 2E). Based on the MC3R knockout mouse model, reduced MC3 receptor activity is expected to lead to increased body fat (feed efficiency) and leptin levels (1,2).

This novel MC3R mutation was found in an Indian family, and only part of the pedigree has been described previously (11). We have further examined the extended pedigree in an attempt to better define the effects of this mutation and also completed our in vitro studies of the Ile183Asn mutant MC3 receptor. A 13-year-old Indian girl (proband) with early-onset severe obesity was heterozygous for this mutation (Fig. 3A). Four years later, at 17 years of age, the proband had developed morbid obesity with high percentage ideal body WFH of 216% and percent body fat of 71%. Her heterozygous father's WFH has increased to 167%, and percent body fat was 48%. The proband was the most overweight of the family with the highest percent body fat. The eldest sibling (not reported previously) was recently found to be heterozygous for this mutation. At 23 years old, he was not obese, with BMI of 23.5 kg/m². However, he was overweight as a child. At 7 years old, his weight was documented as 27 kg with WFH of 125% (>120% defined as mild obesity). He only started to lose weight in his teenage years as he was physically active and underwent army training for 2 years. He has since maintained a very active lifestyle with regular exercise and sports and is careful with his diet. The paternal uncles and aunt had late onset obesity or overweight phenotypes, and genotyping revealed wild-type MC3R alleles. An interesting observation was the cosegregation of early-onset childhood obesity with the mutant MC3R allele (Fig. 3A). Among the four siblings (paternal uncles and aunt), the proband's father was the only one who was obese since early childhood. His documented weight and BMI at 16 years were 86 kg and 33 kg/m². The other three sibs were overweight only after 30 years of age: the eldest uncle was 64 kg at 18 years of age, and the second uncle was 60 kg at 20 years of age.

The Ala70Thr mutation was found in an 11-year-old boy and his mother (Fig. 3B). The proband was overtly obese, whereas his heterozygous mother had mild obesity (WFH 121%). His mother was overweight as a child, but documented weight and height measurements were unavailable. She has been careful with her diet and exercises regularly, because she is very conscious of her weight and body image.

The heterozygous Met134Ile mutation was found in a 12-year-old Indian girl and her obese mother, and both mother and child have type 2 diabetes (Fig. 3C). The proband was asymptomatic but was diagnosed by abnormal oral glucose tolerance test.

The three probands were compared with groups of obese control subjects of similar age and same sex as shown in Table 2. The control subjects selected were 2–3 years younger or older than the subjects with rare variants. This age range was chosen arbitrarily to include a sizeable number of control subjects for this comparison. All subjects who fulfilled the criteria are included in the analysis. The probands had significantly higher leptin levels (P = 0.018) and appeared to have higher BMI, WFH, and percent body fat, although these did not reach statistical significance. We also observed that heterozygotes had significantly lower Hunger scores (less subjective feeling of hunger) compared with that of the obese control subjects (P = 0.026) (Table 3). The scores for restrained eating and disinhibition were not significantly different between the heterozygotes and the obese control subjects without MC3R mutations (P > 0.05 for each factor analyzed). This is reminiscent of the MC3R knockout mice, which were hypophagic with high leptin levels and body fat mass (1–2). We did not detect any significant difference in HOMA, insulin-to-glucose ratio, or physical activity levels.

Because wild-type MC3R and MC3R with common variants 6K/81I (haplotype) showed significant differences in response to α-MSH stimulation (11), we compared the signaling properties of the mutant MC3Rs with the corresponding MC3R with or without 6K/81I. This is also in consideration of possible conformational changes induced by interaction of these variants. The Ile183Asn and Met134Ile mutant alleles of our two subjects carried the 6K/81I (haplotype) concomitantly and were thus compared with MC3R with 6K/81I (haplotype) in our transfection studies. This was determined by cloning each of the two MC3R alleles from the subjects' DNA samples into pDrive vector followed by direct sequencing and also by allele-specific PCR for Lys6 followed by direct sequencing as described above.

Both transient and stable transfection studies revealed severely impaired signaling response of the Ile183Asn mutant receptor to α-MSH. The cells transiently expressing the mutant MC3R Ile183Asn showed negligible response to increasing α-MSH concentration (Fig. 2B and C). Mutant receptor Met134Ile demonstrated a modest but significantly higher EC₅₀ compared with MC3R with 6K/81I (Fig. 2B). The mutant Ala70Thr receptor also demonstrated significantly reduced cAMP response to MSH stimulation (Fig. 2D). Our competitive ligand binding studies did not find any significant differences in IC₅₀ between the wild-type, 6K/81I, Ala70Thr, Met134Ile, and Ile183Asn MC3Rs (Fig. 2E). The three mutant MC3 receptors did not exert any dimerization effect on wild-type or 6K/81I MC3 receptors (Fig. 4). The findings were similar using NDP-MSH instead of α-MSH (data not shown). Immunofluorescence staining of HEK cells transiently transfected with these variant MC3Rs revealed good level of cell surface expression comparable with wild-type MC3R qualitatively (Fig. 5).

Common obesity is a polygenic trait resulting from interaction of multiple genetic loci with the environment. Sequence variants in a large set of genes implicated in energy regulation could predispose an individual to excessive weight gain in a given environment. Although MC3R mutations are unlikely to result in an autosomal dominant form of monogenic obesity, this study provides evidence that MC3R can be one of the predisposing genes that contributes to increased adiposity and that the wide variation in the adiposity of the individuals with common and rare variants may be due to other modifying genetic and environmental factors.

MSH, a derivative of proopiomelanocortin (POMC), acts on MC4R and MC3R to reduce feeding and feed efficiency (1–3). The Mc3r^−/− mice demonstrated mild obesity but increased body fat and leptin, whereas the phenotype of the heterozygous Mc3r^+/− mice did not appear to differ significantly from wild-type mice (1,2). In comparison, the Pomc^−/− mice also exhibit an obese phenotype, whereas the heterozygous Pomc^+/− mice appeared to have a similar phenotype to wild-type mice on standard chow (20,21). However, Pomc^+/− mice developed obesity when put on a high-fat diet (21), exhibiting an intermediate obese phenotype between wild-type and Pomc^−/− mice, demonstrating that a single functional copy of the POMC gene is not sufficient for maintaining normal energy homeostasis under certain environmental conditions and that haploinsufficiency can interact with dietary factors to increase body weight. There is evidence to suggest that loss of one POMC allele, and even genetic variants with subtle effects on POMC function, could influence susceptibility to obesity in humans in our modern “obesogenic” environment. Heterozygous parents of obese children homozygous for POMC-null mutations had high normal to high BMI (22), and other family members heterozygous for POMC-null allele were more overweight than those with wild-type alleles (23). Heterozygous partially inactivating POMC mutations Arg236Gly and Tyr221Cys were also found more frequently in obese children, and the phenotype is reminiscent of MC4R deficiency (24,25). Therefore, partially inactivating genetic variants of MC3R may likewise exert a significant effect on the phenotype even in the heterozygous state, in the “obesogenic” environment. This notion was supported by the linkage of a locus encoding MC3R on human chromosome 20q13.2 to the regulation of BMI, subcutaneous fat mass, and fasting insulin (26). We report common and rare novel variants at the MC3R locus, which result in partially reduced activity of the MC3R in response to MSH and demonstrate that the common variants, and possibly the rare variants, are associated with increased body fat and leptin levels (with additive effect) and perhaps decreased hunger, in human subjects, congruous with the phenotype of the Mc3r^−/− mice. Of note is the presence of childhood or early-onset obesity in all the carriers of the rare mutations (Fig. 3A–C), although the significance of this is uncertain.

There were varying reports of increased insulin-to-glucose ratio, HOMA, leptin, BMI, and body fat in humans with 6K/81I (8,11,12). Our report reinforced the evidence that these two variants contribute significantly to the human phenotype by demonstrating a significant additive effect of 6K/81I on adiposity and leptin. Feng et al. reported that the MC3R with 6K/81I resulted in reduced receptor activity, binding affinity, and protein expression compared with wild-type MC3R (12). Tao and Segaloff (27), however, reported that there was no difference in receptor activity and ligand binding affinity for these two variants. Our report has furthered this debate; we found that MC3R with 6K/81I had reduced receptor activity in response to MSH as reported by Feng et al. (12), but we did not detect any difference in ligand binding, and there was qualitative evidence that the variant MC3R was well expressed on the cell surface.

The three rare variants were not found in other similar studies (8–12). Isoleucine residue 183 of MC3R is located in the second intracytoplasmic loop of this G-protein–coupled receptor (4). Isoleucine at codon 183 is a highly conserved residue, present in the MC3R sequence of many other species, from teleost fish through to mammals (accession nos.: zebrafish, NP_851303; chicken, BAA32555; mouse, AAI03670; rat, NP_001020441; and human, NP063941). This hydrophobic isoleucine was substituted by hydrophilic asparagine in Ile183Asn MC3R. The transmembrane domain directly interacts with G-proteins and control subjects cAMP production; this mutation may cause partial reduction in receptor function resulting from abnormal G-protein interaction. After our first communication of this mutation (13), Tao and Segaloff (27) and Rached et al. (28) subsequently reported that the Ile183Asn MC3R had totally abolished cAMP response to MSH, and Rached et al. (28) also reported a dominant-negative effect on wild-type MC3R and complete intracellular retention with no cell surface expression. Our studies demonstrated near total but not complete loss of cAMP generation in response to MSH stimulation, with normal cell surface expression, normal ligand binding affinity, and absence of dominant-negative activity on wild-type MC3R.

The Ala70Thr mutation affects the extracellular domain (4), within the critical region for binding activity (29), where the hydrophobic alanine residue was changed to hydrophilic threonine. The hydrophobic alanine residue at codon 70 is replaced by another hydrophobic residue glycine in the MC3R of chicken, mouse, rat, and spiny dogfish (accession no. AAS66720); thus it may be possible that the substitution by a hydrophilic residue (threonine) may result in a significant change in conformation and function of the receptor. The Met134Ile mutation is located at the second transmembrane region of the seventh transmembrane domain (4). The hydrophobic methionine is conserved in chicken, mouse, and rat.

The three novel rare mutations were not found in the population sample nor reported elsewhere, and the clinical characteristics of our subjects coupled with the in vitro studies support their pathogenic role. We believe that our report has shed light on the human MC3R mutation phenotype, and further studies with bigger numbers will continue to unravel the phenotype and support the role of MC3R in human weight regulation and pathogenesis of obesity.

This study has been funded by the National Medical Research Council (Singapore) and the Singapore Pediatric Society Research Award.

We are grateful for the valuable contribution by the staff of the School Health Service, Health Promotion Board. We also acknowledge the contribution of Evelyn Ng and Angeline Ling for the in vitro studies, Dr. Goh Siok Ying and Dr. Natalie Ong for the recruitment of subjects, and Dr. Heng Chew Kiat for his assistance in the statistical analyses.