Rare variants in the melanocortin 4 receptor gene (MC4R) are associated with abdominal fat and insulin resistance in youth with obesity

Over the past four decades, the prevalence of childhood obesity, and its associated metabolic complications, has been on the rise worldwide [1,2,3]. The leptin-melanocortin system plays a key role in the regulation of energy balance in humans [4,5,6,7]. The process begins with leptin, an adipocyte-derived hormone, binding its receptor in the hypothalamus and inducing the release of alpha-melanocyte stimulating hormone (α-MSH), a post-translational product of pro-opiomelanocortin (POMC) [8]. α-MSH binds the melanocortin 4 receptor (MC4R) in the paraventricular nucleus of the hypothalamus and triggers a signaling cascade that leads to a decrease in food intake and increase in energy expenditure [9]. Yeo et al. and Vaisse et al. were the first to discover that mutations in the MC4R gene can alter this system in humans and result in severe early-onset obesity, usually before the age of 5 years [10, 11]. Later studies in mice revealed how MC4R deficiency impacts ectopic fat accumulation, with Obici et al. demonstrating that MC4R-KO (−/−) mice experience increased visceral fat deposition [12] and Itoh et al. highlighting that MC4R-KO (−/−) mice represent the first novel animal model of non-alcoholic steatosis (NASH) [13]. While it is known that mutations in the MC4R gene are associated with obesity and hyperinsulinemia in children [14], it is unknown whether these mutations affect intra-abdominal fat distribution in humans and insulin kinetics in response to a glucose load. To address this gap in knowledge, we sequenced the coding region of the MC4R gene in 877 youth with obesity who underwent an abdominal MRI to quantify visceral adipose tissue (VAT), abdominal subcutaneous adipose tissue (SAT), and intrahepatic fat content (HFF%), and an oral glucose tolerance test (OGTT) to assess glucose tolerance, insulin sensitivity, and insulin clearance. Interestingly, we observed that pathogenic variants in the MC4R gene are associated with a greater VAT and HFF%, as well as with a greater degree of insulin resistance, independent from the degree of obesity.

Sex as a biological variable

Our study examined male and female children and adolescents.

Study cohort and procedures

Eight hundred seventy-seven male and female children and adolescents with obesity were recruited from the Yale Pediatric Obesity Clinic (New Haven, Connecticut) to participate in this cohort study. Inclusion criteria consisted of age between 6–21 years and a BMI > 85th percentile for age and sex according to the CDC Pediatric Growth Charts. Exclusion criteria consisted of pregnancy in females, significant chronic illness, other than prediabetes or Type 2 diabetes, or the presence of a known genetic condition that is associated with obesity, such as Prader-Willi or Bardel-Biedl syndrome. Participants with Type 1 diabetes, congenital diabetes, or mature onset diabetes of the young (MODY) were excluded. Those with prediabetes or Type 2 diabetes were eligible for study participation. Participants underwent genetic analysis to screen for rare variants (minor allele frequency (MAF) < 0.01) in the coding region of MC4R and were assigned to a Pathogenic Variant or No Pathogenic Variant group based on the presence of a functionally damaging variant in the coding region of MC4R. Participants also completed an oral glucose tolerance test (n = 877) [15] and abdominal MRI (n = 504) to measure HFF%, VAT, and SAT as previously described [16,17,18] and described below. At the OGTT visit, demographic information and anthropometrics were obtained and an 8cc blood sample was collected for the genetic analysis. The OGTT study occurred at the Yale Center for Clinical Investigation (YCCI) and the MRI occurred at the Yale Magnetic Resonance Research Center in New Haven, CT. Data will be available by the investigators upon request.

Sequence analysis of MC4R

Genomic DNA from each subject was extracted from peripheral blood leukocytes using a guanidine hydrochloride DNA extraction protocol. The coding region of MC4R gene was amplified with the polymerase chain reaction (PCR) using the following primers: MC4R-forward = 5’-AGCCTCACAACTTTCAGACAG-3’ MC4R-reverse = 5-AGTACCCTACACGGAAGA-3’. Primers were designed using PrimerQuest (IDT DNA) and synthesized at Yale W.M. Keck Oligo Synthesis Resource Facility. The conditions used for the PCR are as followed: 95 °C for 5 min followed by 35 2-min cycles of denaturing at 94 °C for 30 s, annealing at 65 °C for 30 s, and elongation at 72 °C for 1 min (to account for a larger PCR product). PCR products were verified on a 1% agarose gel and were purified with an ExoSAP kit. PCR products were analyzed by automated sequencing at the Yale W.M. Keck Facility. Variants in MC4R were identified using contig assembly with Sequencher and were compared against the wild-type MC4R gene sequence on Ensembl (ENSG00000166603).

In vitro studies for novel variants

The extended description for in vitro studies is included in Supplementary Methods. Plasmid containing the hMC4R mutation were transiently transfected into HEK293 cells and assayed for their agonist cAMP-based function using the commercially available AlphaScreen^® technology (PerkinElmer Life Sciences, Cat #6760625M) per the manufacturer’s instructions as previously reported [19]. Each experiment was performed in duplicate replicates and in three independent experiments.

Abdominal MRI

Measurement of hepatic fat fraction (HFF%) was obtained from magnetic resonance imaging (MRI) using the 2-point Dixon (2PD) method, as modified by Fishbein et al. [17]. A Siemens Sonata 3.0 Tesla system was used to obtain axial images in T1-weighted and T2-weighted sequences for abdominal MRI. Using MATLAB computer software, five different areas of interest were drawn on the liver image in a single axial slice and the mean pixel signal intensity level was obtained for both in-phase and out-of-phase images. HFF% was quantified from the difference between in-phase and out-of-phase signals and was calculated in duplicate using the formula: [(Sin-Sout)/(2/Sin)] × 100 [20]. Non-alcoholic fatty liver disease was defined as an HFF% greater than 5.5%. An extended description of this protocol has been previously reported [21]. Visceral and abdominal subcutaneous adipose tissue content was quantified from MRI imaging. Five images extending from 15 cm above and 5 cm below the level of L4/L5 were obtained and used to calculate total volumes of visceral and subcutaneous adipose tissue in the abdomen. These images were transferred to MATLAB for analysis. The full protocol has been previously described [20]. Lean mass was not quantified with MRI imaging.

Oral glucose tolerance test

After a 12-h overnight fast, subjects were admitted to the YCCI to undergo a 180-min 10-timepoint OGTT at 8:00 a.m. One peripheral IV cannula was placed in an antecubital vein to obtain plasma samples. Subjects were administered a 1.75 g glucose/kg body mass (max 75 g) orange flavored drink (in 150 mL) at timepoint 0 min of the OGTT. Plasma samples were obtained at −15, 0, 10, 20, 30, 60, 90, 120, 150, 180 for the measurement of glucose, insulin, and C-peptide. An extended description of the OGTT protocol has been previously reported [22].

Biochemical analysis and measures of insulin sensitivity and clearance

Plasma glucose was measured at bedside during the OGTT using the YSI2700-STAT-Analyzer (Yellow Springs Instruments, Yellow Springs, OH), plasma insulin was measured by antibody radioimmunoassay from Millipore Sigma (Billerica, MA), triglyceride (TG) levels were measured using Auto-Analyzer (Model 747-200, Roche Diagnostics, Indianapolis, IN), and alanine aminotransferase (ALT) was measured using standard automated kinetic enzymatic assays. Whole body insulin sensitivity index (WBISI) was calculated from the 180-min OGTT as a measure of insulin sensitivity and was calculated as 10,000/[sqrt(fasting insulin(µU/mL))*(fasting glucose (mg/dL))*(mean glucose (mg/dL))*(mean insulin (µU/mL))] [23, 24]. Insulin secretion rate (ISR) was estimated via deconvolution of C-peptide levels during the OGTT according to Van Cauter [25]. Β-cell function was assessed from the OGTT using a model that describes the relationship between insulin secretion and glucose concentration as the sum of two components [22, 26]. The first component, named β-cell glucose sensitivity (β-GS), represents the slope of the dose-response function between ISR and glucose concentration. This function is modulated by a potentiation factor accounting for insulin stimulatory mechanisms such as prolonged hyperglycemia, incretin hormones, neural modulation, and non-glucose substrates. The potentiation factor values average 1 during the test, indicating relative enhancement or inhibition of ISR. The potentiation factor ratio indicates the ratio between the 160–180 min and the 0–20 min values of potentiation during OGTT. The second component of insulin secretion rate, named β-cell rate sensitivity, represents the dependence of ISR on the rate of change of glucose concentration and is related to early insulin release. Insulin clearance was calculated as the ratio between the areas under the curve (AUC) of ISR and plasma insulin during the OGTT [26, 27].

Bioelectrical impedance analysis

A TANITA digital scale (TBF-400) was used to determine total body mass, fat mass (FM), body fat percentage (bf%), and fat free mass (FFM) by bioelectrical impedance analysis upon admission to the YCCI. As an additional assessment of body composition, load capacity index (LCI) was calculated using the equation FM/FFM [28].

Statistical analysis

Data are reported as mean ± SD or median [25th–75th] for continuous variables (depending on variable distribution, assessed by using Anderson Darling test) and as counts (%) for categorical variables. Mann–Whitney U test, Student’s t test, and Fisher’s exact test were used to compare differences between the Pathogenic Variant and No Pathogenic Variant groups for quantitative and qualitative data, respectively. AUC was calculated according to the trapezoid rule for glucose and insulin levels during an OGTT and was compared between the groups using a Mann–Whitney U test or Student’s t test depending on variable distributions. Statistical significance was established at an alpha of 0.05 and a power analysis was conducted using the GPower software. Analyses were performed using SAS 9.4 (Cary, NC) and GraphPad Prism software 9.0.0 (San Diego, CA).

MC4R gene screening

Screening of MC4R in 877 children and adolescents with obesity revealed 14 subjects who were heterozygous for the Val103Ile (n = 12) or Ile251Leu (n = 2) variants. These variants are considered single nucleotide polymorphisms (SNPs) and have previously been shown to confer no functional change to the protein [29, 30]; therefore, these subjects were included in the No Pathogenic Variant group. Rare variants, defined as a MAF < 0.01, in the MC4R gene were detected in 19 subjects; however, only 13 of these variants have been previously characterized in vitro and reported in the literature. The 6 novel variants found in our cohort—His10Arg, Lys73Asn, Ile104Val, Ala244Gly, Cys277Gly, and Leu325Ile—were functionally characterized with in vitro lab experiments in our study. Between previous studies and our functional studies, 12 of the 19 rare variants were determined to be damaging to the MC4R protein (3 complete loss of function (cLoF) and 9 partial loss of function (pLoF)). These 12 subjects comprised the Pathogenic Variant group. The other 7 subjects with rare variants that were determined to have no loss of function (No LoF) were included in the No Pathogenic Variant group. There were no subjects that were homozygous or compound heterozygotes for any rare variants in the MC4R gene. The functional characteristics of the rare variants are shown in Table 1.

In vitro functional characterization of novel MC4R variants

The half maximal effective concentrations (EC₅₀) for the novel variants characterized in vitro are shown in Supplementary Table 1. When tested for their effect on MC4R signaling, the variants His10Arg, Lys73Asn, Ala244Gly, and Leu325Ile demonstrated wild-type-like activity, producing cAMP concentrations similar to the wild-type (WT-MC4R) when exposed to different endogenous and synthetic ligands (Supplementary Fig. 1A–E). The Ile104Val variant demonstrated reduced activity for α-MSH at the receptor with a 10-fold higher EC₅₀ as compared to WT-MC4R, thus rendering the Ile104Val variant in MC4R a pLoF (Fig. 1A, B). The Cys277Gly variant demonstrated an even greater reduction in activity at the receptor for both agonist ligand potency and efficacy, with an E_max of ~30% (Fig. 1C). Since cAMP was still produced in response to ligand binding, the C277G variant can be considered a pLoF. The considerable reduction in activity of this variant may be attributable to the disrupted disulfide bridge that the cysteine amino acids constitute at positions 271 and 277 (https://www.rcsb.org/3d-view/6W25).

Mean ± SEM of cAMP response curves of A WT MC4R, B I104V, and C C277G to the addition of increasing amounts of ligand on a logarithmic scale. Experiments were performed three independent times in duplicate. α-MSH: naturally occurring POMC-derived agonist for MC4R. NDP-MSH: high affinity synthetic ligand. MTII: superpotent α-MSH analog. γ₂-MSH: Endogenous POMC-derived agonist for MC3R and MC4R. Ac-His-DPhe-Arg-Trp-NH2: Molecular recognition sequence common to the endogenous melanocortin agonists that binds MC4R.

Full size image

Demographic and clinical characteristics

The demographic and clinical characteristics of the study cohort are shown in Table 2. The three main ethnic groups showed a different hepatic fat content and visceral fat (p < 0.001) with African American showing the lowest HFF% and visceral fat and Hispanics the highest values, similar to what reported previously [20]. There were no statistically significant differences between the Pathogenic Variant and No Pathogenic Variant groups for sex (p = 0.77), ethnicity (p = 0.19), BMI z-score (p = 0.13), body fat percentage (p = 0.34), fat free mass percentage (p = 0.66), and waist/hip ratio (p = 0.67); however, there was a difference in the average age for the groups, as the Pathogenic Variant group was younger than the No Pathogenic Variant group (11.4 ± 0.9 years vs. 13.4 ± 0.02 years, respectively) (p = 0.02, power = 0.69) (Table 2).

Impact of MC4R mutation on abdominal fat distribution

Subjects in the Pathogenic Variant group showed a greater amount of VAT (78.9 [62.7–101.0] cm²) as compared to the No Pathogenic Variant group (52.9 [37.6–71.7] cm²) (p = 0.003, power = 0.81) (Fig. 2A). There were no significant differences in the amount of SAT between the two groups (p = 0.64) (Fig. 2B). The VAT to total abdominal adipose tissue ratio (VAT/(VAT + SAT)) was significantly different between the groups, with an average ratio of 0.146 [0.132–0.163] in the Pathogenic Variant group and 0.105 [0.082–0.133] in the No Pathogenic Variant group (p = 0.002, power = 0.84) (Fig. 2C).

A Visceral fat, B subcutaneous fat, and C visceral/(visceral + subcutaneous) fat ratio. Subjects without pathogenic variants in MC4R are depicted in blue and subjects with pathogenic variants in MC4R are depicted in red. p value is from a Mann–Whitney U test.

Full size image

Intrahepatic fat accumulation

The Pathogenic Variant group showed a greater HFF% (21.3 [4.7–29.6]%) compared to the No Pathogenic Variant group 3.7 [0–13.1]% (p = 0.012, power = 0.87) (Fig. 3A). Individual HFF% for each subject with a damaging variant is also compared to average HFF% in subjects without damaging variants (Supplementary Fig. 2). There were no significant differences in fasting levels of ALT, AST, or triglycerides (p > 0.05) between the two groups (Table 2).

A Intrahepatic fat content (HFF%) and B whole-body insulin sensitivity index (WBISI). Subjects without pathogenic variants in MC4R are depicted in blue and subjects with pathogenic variants in MC4R are depicted in red. p value is from a Mann–Whitney U test.

Full size image

Glucose metabolism

Fasting glucose and insulin concentrations were similar between the two groups (p > 0.05) (Table 2). The Pathogenic Variant group showed a lower WBISI as compared to the No Pathogenic Variant group, with an average WBISI of 0.97 [0.59–1.47] and 1.59 [1.08–2.27], respectively (p = 0.0008, power = 0.84) (Fig. 3B). Individual WBISI for each damaging variant in our study cohort was also compared to the average WBISI of the No Pathogenic Variant group (Supplementary Fig. 3).

There was a statistically significant difference in the prevalence of impaired glucose tolerance (IGT) between the Pathogenic Variant and No Pathogenic Variant groups, as 7 (58.3%) subjects had IGT in the Pathogenic Variant group and 183 (21.4%) had IGT in the No Pathogenic Variant group (p = 0.009) (Table 2). The Pathogenic Variant group showed higher glucose levels (AUC_tot: 1.45 [1.35–1.50] mmol/L/min⁻¹) during an OGTT as compared to the No Pathogenic Variant group (AUC_tot: 1.17 [1.01–1.32] mmol/L/min⁻¹) (p = 0.001, power = 0.91) (Fig. 4A). The Pathogenic Variant group showed higher insulin levels (AUC_tot: 40.7 [31.8–72.7] mU/mL/min⁻¹) during an OGTT as compared to the No Pathogenic Variant group (AUC_tot: 27.1 [19.2–39.2] mU/mL/min⁻¹) (p = 0.003, power = 0.81) (Fig. 4B) due to enhanced endogenous insulin secretion (p = 0.008, power = 0.78) (Fig. 4C, D). In the Pathogenetic Variant group, glucose-stimulated hyperinsulinemia was largely secondary to the higher prevailing glucose levels, as it occurred in the absence of substantial group differences in model-derived β-cell function parameters, including β-GS (p = 0.85), β-RS (p = 0.44) and potentiation (p = 0.32) (Fig. 4E–G), and demonstrated only numerically lower insulin clearance (p = 0.11) (Fig. 4H). It is worth noting that the disposition index calculated as the product between WBISI and β-GS, resulted to be significantly lower in the Pathogenetic Variant group compared to the No Pathogenic Variant group (p = 0.042), but with low statistical power (0.48). We concluded that the difference obtained was mostly driven by the insulin resistance state more than beta cell defect.

A Plasma concentrations of glucose, B plasma concentrations of insulin, C plasma concentrations of C-peptide, D insulin secretion rate from C-peptide deconvolution, E model-derived β-cell glucose sensitivity (β-GS), F β-cell rate sensitivity (β-RS), G potentiation factor ratio, and H insulin clearance during an OGTT. Subjects without pathogenic variants in MC4R are depicted in blue and subjects with pathogenic variants in MC4R are depicted in red. In A–D, p value is from a Mann–Whitney U test comparing the total AUC between the two groups. In E–H, p value is from a Mann–Whitney U test. Error bars represent standard deviation in (A–D). Boxplot in E–H reported the median [25th–75th] of the distributions along with the scatter plot for each group.

Full size image

In the present study, we compared the clinical phenotype of children and adolescents with and without rare, damaging variants in the MC4R gene. We found that youth with obesity with damaging variants in MC4R had a significantly greater degree of insulin resistance and lower glucose tolerance, greater intrahepatic fat content, and greater amount of visceral adipose tissue compared to youth without the presence of a damaging variant, despite the two groups having similar BMI z-scores and degrees of abdominal subcutaneous and total adipose tissue.

It is well established that mutations in the MC4R gene lead to severe early-onset obesity and hyperinsulinemia [14], however, it is not known how these mutations affect body fat distribution in humans. Furthermore, it is not fully understood whether the associated hyperinsulinemia is driven by the mutation or by the effect of increased body weight, as in most previous published studies, the study populations included both lean and youth with obesity and a clear difference in BMI between those with and without mutation. By studying youth with obesity only, we have eliminated the confounding effect of BMI on our observations and provide evidence that rare variants in the MC4R gene affect body fat partitioning and insulin levels in youth independent from the effect of BMI or body fat content.

We compared the insulin concentrations in plasma with previous studies reporting insulin concentrations in children and adolescents carrying MC4R mutations [14, 31]. In those studies, average insulin was 27 ± 4.4 uU/ml for totally inactive MC4R and 25 ± 3.8 uU/ml for the partially inactive MC4R mutations [14] and, based on the observation of Fig. 3, about ~160 pmol/L (~35 uU/ml) in youth between 11 and 20 years old [31]. These concentrations tend to be lower than what we observe in our group of participants carrying pathogenic variants (42 uU/ml) and with an average age of 10 years.

An earlier study by Obici et al. showed that intracerebroventricular administration of an MC4R antagonist in mice resulted in significantly higher VAT content and decreased insulin sensitivity, while the administration of alpha-MSH (MC4R agonist) resulted in the opposite effect, when controlling for feeding behavior [12]. This is consistent with our findings of an increased VAT and decreased insulin sensitivity in youth with obesity carrying damaging MC4R variants. In addition, Itoh et al. were the first to show that mc4r-KO (−/−) mice developed a liver condition similar to human NASH when fed a Western diet, ultimately uncovering a novel NASH animal model [13]. The mc4r-KO (−/−) mice exhibited massive hepatic steatosis, in conjunction with increased adipose tissue inflammation, liver antioxidant enzymes, and oxidative metabolite derivative serum concentrations. Moreover, using isotope-based flux analysis, Hasenour et al. showed that mc4r-KO (−/−) mice fed a Western diet resulted in accelerated glucose-producing and oxidative metabolic fluxes in the liver, which enhanced metabolic dysfunction and promoted the progression to NASH [32]. Similarly, in our study, we found that youth with obesity and MC4R deficiency have an increased intrahepatic fat and a higher frequency of impaired glucose tolerance. Liver steatosis associated with MC4R deficiency can explain numerically reduced hepatic insulin clearance in these subjects [33], which may contribute to chronic hyperinsulinemia and β-cell function decline [34]. These studies suggest that damaging variants in the MC4R gene may lead to hyperinsulinemia by worsening whole body insulin resistance. In this context, those with greater VAT content likely experience an enhanced flux of free fatty acids from VAT to the liver due to the increased adipose tissue insulin resistance that accompanies it, thereby, possibly contributing to the development of intrahepatic fat accumulation and a metabolically unfavorable phenotype in this group of patients.

While the importance of MC4R signaling in energy balance is clear, several different mechanisms have been proposed for how MC4R might regulate glucose and insulin homeostasis, including its role in autonomic outflow to peripheral organs [12]. Rossi et al. showed that selective re-expression of MC4Rs in preganglionic sympathetic neurons in mice attenuated hyperglycemia and hyperinsulinemia, improved hepatic insulin action, and suppressed insulin-mediated hepatic glucose production [35]. This suggests that the observed hyperinsulinemia and glucose dysregulation found with MC4R deficiency may be mediated in part by the effect of an altered melanocortin system on sympathetic innervation, consequently resulting in unopposed parasympathetic outflow to the pancreas and liver. Interestingly, a more recent study by Wang et al. revealed a direct link between central MC4R signaling and energy regulation by the liver. The group identified Tsukushi (TSK), an inducible hepatokine that is under inhibitory control of downstream MC4R signaling [36]. It was found that TSK inactivation in the liver serves to restore energy balance, to improve insulin sensitivity, and to decrease food intake in the setting of MC4R deficiency in mice.

While we illustrate the association between MC4R deficiency and intrahepatic fat accumulation, it is important to point out that not every subject in the Pathogenic Variant group presented with an intrahepatic fat fraction above 5.5%, the threshold that is used to define steatosis by MRI assessment. However, it should be noted that two of these three subjects are African American, an ethnic background that has been associated with a relative protection from the accumulation of intrahepatic fat [20]. In addition, the effect of puberty on insulin resistance is an important consideration, as the onset of puberty is associated with a decrease in insulin sensitivity. However, in this study, those in the Pathogenic Variant group showed greater insulin resistance despite being younger in age comparatively to the No Pathogenic Variant group, therefore, it is unlikely that pubertal stage affects these results.

We acknowledge that our study has some limitations, such as the small sample size of the Pathogenic Variant group, the lack of insulin clamp studies to further investigate glucose metabolism and insulin sensitivity, and the lack of broad genetic screening to include genotyping of the entire leptin-melanocortin system. Nonetheless, this study has some important strengths, such as the use of MRI to measure fat distribution and frequency-sampled 180-min OGTT to assess glucose tolerance, insulin sensitivity, model-derived β-cell function, and insulin clearance, as well as the thorough clinical phenotyping of our cohort. Given the young age of our study population, it is unlikely that the subjects have alcoholic fatty liver disease, so our findings more accurately assess MC4R deficiency on metabolic dysfunction.

In conclusion, this is the first study to examine the impact of MC4R variants on abdominal fat distribution measured by MRI in young patients with obesity. Our results suggest that disrupted MC4R signaling may play a major role in the accumulation of visceral adiposity and subsequently influence its associated metabolic consequences, such as hyperinsulinemia, and impaired glucose tolerance. These findings contribute to the understanding of the genetic architecture of fat deposition in humans and may help to expand the application of drugs in treating patients with obesity carrying a disrupted MC4R and leptin-melanocortin system.