Rare Loss-of-Function Variants in NPC1 Predispose to Human Obesity

Obesity is a chronic and complex disease reflecting the interplay between both genetic and environmental determinants. Since the 1990s, several twin studies have estimated the heritability of obesity to be 40–70% (1,2). Recent genome-wide association studies (GWAS) have revealed 97 genetic loci for BMI and 49 for waist-to-hip ratio (WHR) (3,4). Although the causal effectors and specific molecular mechanisms responsible for these GWAS loci in obesity development remain largely undetermined (5), several deep resequencing studies targeting these GWAS risk loci have successfully identified some functional variants in relation to obesity (6–8) and diabetes (9,10). Together, it seems that the so-called “missing heritability” for complex diseases is at least partly attributed to rare or low-frequency variants in candidate genes with moderate to large effect sizes (11,12). Moreover, the ultimate penetrance of these genetic variations/mutations would be affected by environmental factors that also require careful characterization (12,13). Thus, targeted sequencing of candidate genes prioritized for those significant GWAS signals represent a cost-effective strategy to improve the understanding of the genetic basis for complex disorders such as obesity in humans (3,14). As such, the Niemann-Pick type C1 (NPC1) gene offers a specific case in the recent advancements of the field. In 2009, a common single nucleotide polymorphism (SNP) (rs1805081) in NPC1 was first associated with early-onset and morbid obesity in a GWAS of Europeans (15). Subsequent animal studies indicated that heterozygous Npc1 mutant (Npc1^+/−) mice have more fat mass and weight gain than their wild-type littermates (16). Interestingly, genome-wide cis-expression quantitative trait loci analysis displays a significant expression correlational pattern between Npc1 and body fat percentage in >100 inbred strains of mice fed a high-fat diet (HFD) (17).

Niemann-Pick type C (NP-C) disease is a rare, autosomal-recessive disorder resulting from homozygous or compound heterozygous NPC1 mutations and is clinically characterized by progressive neurological deterioration and liver/spleen enlargement due to abnormal sequestration of unesterified cholesterol and glycosphingolipids within the late endosome/lysosome compartments (18). Although the biological parents of NP-C patients carrying one mutant NPC1 allele are considered to be physiologically normal, we and others have recently shown that these apparently healthy carriers of heterozygous NPC1 mutation (NPC1^+/−) have increased plasma bile acid A (a glycine-conjugated 3β,5α,6β-trihydroxycholanic acid) (19) and 7-ketocholesterol (20), an oxidized lipid involved in atherosclerosis progression (21,22), indicating the potential detrimental effects of heterozygous NPC1 mutation on cardiometabolic functions. To date, no studies have directly and systematically investigated NPC1 mutations and their functional consequences in the pathogenesis of human obesity in the available literature.

We therefore recruited 25 Chinese probands with NP-C disease and their parents. Interestingly, we found that male NPC1^+/− carriers had a higher BMI than matched control subjects or the whole population-based male control subjects, suggesting that NPC1 mutation may be an important regulator for body weight. We further performed animal studies with Npc1^+/− mice; targeted sequencing of NPC1 in young, severely obese subjects and matched control subjects; and in vitro functional studies to validate these clinical phenotypes apparently attributable to rare loss-of-function NPC1 variants that were previously unrecognized.

To evaluate BMI of the individuals with NPC1^+/−, we recruited all parents of 25 patients with NP-C disease carrying homozygous or compounded heterozygous NPC1 mutations in three national specialized hospitals (Shanghai Xinhua Hospital, Peking University First Hospital, and Peking Union Medical College Hospital). All 25 patients with NP-C disease received a clinical diagnosis, of whom 21 were genetically confirmed at a certified molecular genetics diagnostic laboratory, and 2 received a pathological diagnosis with the presence of foam cells and sea-blue histiocytes in bone marrow (Supplementary Tables 1 and 2). All the parents of NP-C disease patients were expected to result in NPC1 haploinsufficiency, and some of the patients have been reported on previously (20,23). For each NPC1^+/− carrier, ∼150 control subjects were matched for sex, age (±3 years), and geographic homogeneity (resident in the same province) from an established national cohort database (24) (Supplementary Tables 1 and 2). A total of 3,657 male and 3,486 female subjects were matched for the 25 male and 25 female NPC1^+/− carriers, respectively (mean ± SEM: age 35.64 ± 1.48 vs. 35.84 ± 0.12 years, respectively, for male carriers and control subjects, P = 0.89; 33.84 ± 1.43 vs. 34.85 ± 0.12 years, respectively, for female carriers and control subjects, respectively, P = 0.50). NPC1^+/− carriers from the same province with an age difference of <3 years were matched for control subjects as one unit. Detailed matching information was shown in Supplementary Tables 1 and 2.

In validation stages 1 and 2, we sequenced the whole exons of NPC1 in 471 young, obese patients and 687 normal-weight control subjects of Chinese ancestry from eastern China, whereby it has been established that the population shows high genetic homogeneity to minimize any effects of population stratification (Supplementary Fig. 1). In brief, based on stage 2 of our in-home database, consisting of the exome sequence data of 227 obese patients and 221 control subjects, we used principal component analysis to detect population outliers and stratification (25), and no samples were removed, suggesting that obese patients and control subjects were genetically well matched (Supplementary Fig. 1). All the obese patients were identified and recruited continuously from the specialized obesity outpatient clinic of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, from 2007 to 2011. Unrelated lean control subjects were recruited from among volunteers at the Shanghai Jiao Tong University School of Medicine on the basis of their normal BMI (26,27). Obese patients with secondary or syndromatic obesity and control subjects with diabetes, hypertension, and impaired glucose regulation were excluded. Height and weight of the participants were measured in light clothing without shoes by a height-weight scale (SanYuan). Waist and hip circumstance was measured by tape beneath clothing following standard protocols. All measurements were repeated three times, and mean values were used for the following analysis. Genetic and basic characteristic information were obtained from the Genetics of Obesity in Chinese Youngs (GOCY) study, which was previously established by Ruijin Hospital and registered in ClinicalTrials.gov (Clinical trial reg. no. NCT01084967, clinicaltrials.gov) (26,27).

Genomic DNA was isolated from peripheral blood and Sanger sequencing was performed to identify any variants in NPC1 in stage 1. In validation stage 2, exome capture was performed to collect the CDS region of human genome DNA using NimbleGen EZ 44M array (Roche) guided by the manufacturer protocols. The Illumina HiSEq 2000 platform (Illumina Inc.) was used for sequencing, the generation of pair-end 90–base pair reads for each individual with a sequencing depth of ∼70×. Raw image files were processed by Illumina pipeline (version 1.6) for base calling with default parameters. This information was also detailed in our previous study (28). By filtering on minor allele frequency (MAF) and variant type, we obtained a list of nonsynonymous variants with an MAF <5% for further analysis and the rest nonsynonymous variants with an MAF >5% were shown in Supplementary Table 3. Synonymous coding variants without amino acid change were excluded. Primers used to amplify the entire NPC1 exons in stage 1 are shown in Supplementary Table 4.

We obtained a male BALB/cJ Npc1 heterozygous mouse (BALB/cNctr-Npc1^m1N/J) from The Jackson Laboratory (Bar Harbor, ME). After backcrossing to a C57BL/6 genetic background for six generations, heterozygous Npc1 mutant (Npc1^+/−) mice and their wild-type littermates were genotyped at ∼3 weeks after birth and were housed under a 12-h light/12-h dark cycle at constant temperature (22–24°C) with free access to food and water. Six-week-old mice were fed with a 60% kcal HFD (Research Diets) for 5 months. Body weight was measured by a digital scale (Ohause), and body composition was evaluated by a body composition analyzer (EchoMRI). Fasting plasma triglyceride, total cholesterol, HDL cholesterol (HDL-c), and LDL cholesterol (LDL-c) were measured enzymatically (Rongsheng) following the manufacturer guidance.

The human study of the project was approved by the Institutional Review Board of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine. Written informed consent was provided from each participant prior to inclusion in the study. For the animal study, all procedures were approved by the Animal Care Committee of Shanghai Jiaotong University School of Medicine.

Subcutaneous- and visceral-fat areas were measured by means of computed tomography (CT) scanning at the umbilicus level with a LightSpeed CT350 HD scanner (General Electric Medical Systems). Cross-sectional subcutaneous and intraabdominal fat areas were determined using FatScan software (N2 Systems Inc.) following regular procedures (29), in which the total area indicates the sum of the two parts.

A human cDNA clone of the coding region of NPC1 (Origene; National Center for Biotechnology Information Reference Sequence NM_000271) was subcloned into the pEGFP-N1 vector (COOH-terminal GFP tag) between HindIII and BamHI restriction sites. The nucleotide substitutions were then generated via a site-directed mutagenesis kit (Agilent). We confirmed the full-length coding sequences for all the plasmids via DNA sequencing.

HEK293T and Hela cells were grown in DMEM (Gibco) containing 10% FBS (Gibco), 1 mm penicillin/streptomycin, and 1 mm l-glutamine at 37°C and in 5% CO₂. The CT43 cell line, which is a NPC1-deficient CHO cell line and presents cholesterol accumulation in lysosome (30), was obtained from T.Y. Chang at the Geisel School of Medicine, Dartmouth College (Hanover, NH).

HEK293T cells were transfected with each construct by Lipofectamine 2000 and lysed, and their concentration was determined. Western blotting of wild-type and all NPC1 mutants was performed as previously described (27) using rabbit anti-GFP antibody (Cell Signaling Technology), rabbit anti-HSP90 antibody (Santa Cruz Biotechnology), and horseradish peroxidase–linked goat anti-rabbit secondary antibody (Cell Signaling Technology).

To determine intracellular localization of mutant protein, Hela cells were transfected with plasmids for 48 h followed by fixation and incubation with mouse anti–lysosome-associated membrane protein 2 (LAMP2) antibody (Santa Cruz Biotechnology) and Alexa Fluor 594–conjugated donkey anti-mouse secondary antibody (Molecular Probes). Images were acquired with a confocal laser-scanning microscope (Zeiss). All the representative images were repeated in at least three independent experiments.

Cholesterol-transporting experiments were conducted as previously described (30). In brief, CT43 cells were transfected with NPC1 mutants by FuGENE HD (Promega) and stained with 50 mg/mL filipin for 30 min. Photomicrographs were acquired with a confocal laser-scanning microscope (Zeiss). Complementation of cholesterol transport in CT43 cells after transfection was assessed by filipin staining. Among the GFP-positive cells, those with no filipin-positive organelles were classified as a complete complementation group, whereas those presented with residual filipin were classified as not being a complete complementation group (31). More than 200 cells were analyzed for each construct for three independent experiments.

We conducted a sex-stratified analysis (male and female, separately) to compare BMI between NPC1^+/− carriers and matched control subjects or the whole population-based control subjects. Within each sex, we performed general linear regression analysis to examine the difference in BMI levels between NPC1^+/− carriers and control subjects with adjustment for age and geographic location. The generalized linear model was also used to test the interactions of sex and NPC1 mutations on the difference in BMI levels by including sex, NPC1^+/− mutation (yes/no), and a multiplicative interaction term (sex × NPC1^+/− mutation) in the same multivariable model (32). Differences in categorical variables such as the gene-based frequency distribution of rare NPC1 variants between obese patients and control subjects, and the proportions of filipin-positive cells, were calculated using the χ² test. Continuous variables were first examined for normal distribution with the Shapiro-Wilks test. The differences in normal distributed variables between groups were calculated using the Student t test, and skewed distributed variables were compared using the Wilcoxon rank sum test. The difference in phenotype between Npc1^+/− and wild-type mice was analyzed by Student t test, and the differences in obesity-related characteristics between severely damaged NPC1 variant carriers and mild/no damage variant carriers were analyzed by ANOVA. Differences in obesity traits between the severely damaged NPC1 variant carriers and nonvariant carriers in young obese patients were compared by Wilcoxon rank sum test. Linear regression models were constructed to examine the correlation between rare NPC1 variants and the indicated characteristics by fitting genotypes as the independent variable with the indicated adjustment for potentially confounding variables. Analyses were carried out with SAS version 9.3 (SAS Institute) at a two-tailed α level of 0.05.

Following the classic Mendelian inheritance pattern of autosomal-recessive diseases, both the father and mother of the NP-C patient carry one mutant NPC1 allele, providing a natural model for evaluating the direct effects of NPC1 haploinsufficiency on adiposity, even though homozygous NPC1 mutations of the NP-C patients prevented the observation of body weight–related traits for their early death or developmental defects. We identified and recruited 25 family trios with clinically diagnosed NP-C probands from 17 provinces in China in three clinical centers. After confirming clinically diagnosed NP-C probands via gene sequencing (Supplementary Tables 1 and 2), their parents’ BMI values were compared with those of the sex-, age-, nationality-, and geography-matched control subjects (Supplementary Tables 1 and 2). We found that the BMI in male NPC1 heterozygous mutant (NPC1^+/−) carriers was significantly higher than that in male control subjects (26.25 ± 0.59 vs. 24.34 ± 0.06 kg/m², P = 0.009) (Fig. 1A). In addition, we also compared the BMI values between the NPC1^+/− carriers and the whole population-based male control subjects, which were taken from a nationwide representative population (24). By adjusting age and geographical origin, we found that the BMI values of male NPC1^+/− carriers were still significantly higher than those of male subjects (n = 45,143) in the total population–based cohort (26.25 ± 0.59 vs. 23.77 ± 0.02 kg/m², P = 0.002) (Fig. 1B). These results demonstrated the increased risk of heterozygous NPC1 mutations for human adiposity.

The penetrance or severity of phenotypes for most of the pathogenic genetic defects varies between the two sexes, as shown in mutations of the obesogenic gene MC4R (33,34). This phenomenon was also observed for NPC1 mutations, because no significant differences were observed in BMI values between female carriers of NPC1 mutations and matched control subjects (23.08 ± 0.36 vs. 23.74 ± 0.06 kg/m², P = 0.40) (Fig. 1C). Also, no difference was observed between female carriers and the female subjects (n = 53,515) from the total population–based cohort (23.08 ± 0.36 vs. 23.68 ± 0.02 kg/m², P = 0.41) (Fig. 1D). These results suggested that NPC1 mutations may interact with sex in affecting adiposity risk. A statistical test of interaction showed significance between NPC1^+/− heterozygous status and sex after accounting for covariates in carriers and matched control subjects (P for interaction = 0.01).

Interestingly, we observed that male carriers apparently had more mutations than females in the sterol-sense domain (SSD) and the cysteine-rich luminal loop (CRLL) domain (P = 0.02) (Fig. 1E), which are of the most importance for the NPC1 protein function (35). In addition, three male NPC1^+/− carriers showed higher BMI values than three female carriers possessing the same NPC1 mutation genotypes (P434L, D501Y, and R1077X) (26.60 ± 0.33 vs. 22.43 ± 1.36 kg/m², P = 0.04) (Fig. 1E). We also calculated and obtained a significant sex difference in BMI values in the national population-based cohort (24) (n = 45,143 and 53,515; BMI 23.77 ± 0.02 and 23.68 ± 0.02 kg/m², respectively, for males and females, P = 0.0003). Interestingly, the BMI difference observed in the three male and female carriers showed an increased trend compared with the BMI difference in the population-based cohort of two sexes (ΔBMI 4.17 and 0.09 kg/m², respectively, for the former and latter, P = 0.07).

In order to validate the gene-by-sex interaction observed in humans, we further challenged the wild-type and Npc1^+/− mice with an HFD for 5 months, and, consistent with human data, male Npc1^+/− mice showed 11.2% higher body weight than control subjects (39.6 ± 1.2 vs. 35.6 ± 1.0 g, P = 0.01), but no difference was found in female mice (Fig. 2A and B). These results indicated a sex-specific effect of loss-of-function NPC1 mutation or deficiency for the development of obesity in human and mice. Moreover, fat mass and inguinal subcutaneous fat pad were significantly increased with larger adipocyte size in male Npc1^+/− mice (Fig. 2C–F), suggesting more energy storage in white adipose tissues. Levels of plasma total cholesterol, HDL-c, and LDL-c were also increased in male Npc1^+/− mice (Fig. 2G–J).

Because a subpopulation with extreme phenotypes enriched by genetic defects are ideal for detecting genetic pathogenic factors (36), we thus validate the effects of NPC1 variants on obesity in a two-stage case-control study including 244 young, severely obese patients (mean BMI 36.7 kg/m²) and 468 sex- and geography-matched lean control subjects in the first stage; and 227 young, severely obese patients (mean BMI 39.4 kg/m²) and 219 matched lean control subjects in the second stage (28). Stage 1 identified a total of 14 rare variants (MAF <1%) that cause NPC1 missense/frameshift variants with gene-based prevalence of 4.10% versus 1.07%, respectively, in obese patients and control subjects (odds ratio [OR] 4.0 [95% CI 1.3–11.7], P = 0.0075) (Tables 1 and 2). Stage 2 was from our in-home whole-exome sequencing data, based on which we screened another six carriers with rare NPC1 missense variants (MAF <1%) in the obese group versus zero in the control group (2.64% vs. 0.00%, P = 0.03) (Table 2). These variations were also confirmed using Sanger sequencing (Supplementary Fig. 2). In total, we identified 17 rare NPC1 nonsynonymous/frameshift variants in 3.4% of obese subjects versus 0.73% of control subjects (OR 4.8 [95% CI 1.7–13.2], P = 0.0008) (Tables 1 and 2). Of note, we detected four common NPC1 nonsynonymous variants (MAF >5%) in the coding region in our cohorts (Supplementary Table 3), two of which, the SNPs rs1805081 (p.H215R) and rs1805082 (p.I858V), were identified in the European population associated with obesity (15). We further adjusted these two SNPs for the calculation of the association between rare NPC1 variants and obesity, and found that these rare NPC1 variants were still significantly associated with obesity (OR 7.5 [95% CI 2.1–26.2], P = 0.0017), suggesting that rare NPC1 variants increased human obesity independent of known common obesity-associated NPC1 variants. Interestingly, when stratifying the cohort into males and females, we consistently found that the prevalence of rare NPC1 variants was significantly higher in obese male subjects than male control subjects (4.46% vs. 0.00%, P = 0.0005), whereas no significant association was found in female obese patients and control subjects (2.60% vs. 1.17%, P = 0.23) (Supplementary Table 5). These results suggested a sex-specific association of rare NPC1 variants with obesity. In addition, we did not observe a significant association of rare NPC1 variants with other metabolic parameters such as blood glucose levels, lipid levels, or blood pressure (Supplementary Table 6).

Most of these rare NPC1 variants appeared highly conserved across different species (Supplementary Fig. 3), implicating important biological functions for the NPC1 protein. Similarly, the majority of these rare variants was predicted to be damaged or possibly damaged by PolyPhen2 and SIFT (Table 1). As shown in the schematic diagram of full-length NPC1, 6 of the 17 rare variants were located in the functional domains, including SSD and CRLL domain (Fig. 3A), that determine the functional impact of the NPC1 protein (35), whereas the other rare variants distributed widely covering nearly all of the coding regions. We next cloned these rare variants and examined their expression and cellular location. NPC1 mutants including T743I, Q775K, and V780fs* in the SSD domain exhibited markedly reduced protein levels (Fig. 3B). Moreover, Q775K also showed intracellular dislocation from late lysosome labeled by LAMP2 (Fig. 3C and Supplementary Fig. 4). NPC1 functions to transport free cholesterol out of lysosome, and filipin, which can specifically bind to free cellular (or lysomal) cholesterol, has been used to investigate the cholesterol transporting ability (CTA) of NPC1 protein (31). We found that 8 of 17 mutants showed decreased CTA compared with wild-type NPC1 (Fig. 3D). Five mutants (including four in the SSD and one in the CRLL domain, V1044M) showed >50% reduction in CTA, whereas I1061T, the most prevalent mutation (20% prevalence) in European NP-C patients (31) with severe functional damage as a systematic control showed a nearly 50% reduction in CTA (Fig. 3D and Supplementary Fig. 5). It is worth mentioning that all NPC1 rare variants with functional damage were identified in obese patients, whereas the rare variants found in control subjects showed no functional impact. Furthermore, rare variants in the SSD plus V1044M in the CRLL domain showed more severe damage.

Previous studies (35,37) have shown that mutations in the SSD and CRLL domain resulted in the absence of NPC1 protein expression among patients with infantile onset of “classic or severe” features. And a very recent study (38) illustrating the cryo-EM structure of full-length human NPC1 also suggested the importance of SSD domain for the cholesterol-transporting function of NPC1. In our study, carriers with four rare variants in the SSD along with the V1044M variant in the CRLL domain had both severely damaged NPC1 function and worsened obesity traits, including an increasing trend of BMI, and significantly higher waist circumference and WHR than carriers with mild/no damage to NPC1 function (Fig. 4A–C), indicative of a strong genotype-to-phenotype correlation. To further accurately assess fat content and distribution among rare NPC1 variant carriers, we used high-resolution CT scanning to reveal more total fat accumulation and an increased trend of visceral fat and subcutaneous fat accumulation in severely damaged rare variant carriers (Fig. 4D and E). Moreover, when compared with obese individuals with non-NPC1 rare variants, carriers with severely damaged rare NPC1 mutations had a higher WHR (Supplementary Table 7). These findings were strongly supportive of the notion that rare NPC1 variations (genotype) leading to a >50% loss of function were associated with more severely obese traits (phenotype) in a dose-dependent–like manner among severely obese patients.

We identified multiple functional and novel rare mutations/variants in the NPC1 gene to the development of adiposity in a series of studies incorporating a natural human disease model, a mouse model, and two stages of extreme case-control studies. In both men and male mice, loss-of-function mutations in NPC1/Npc1 clearly and significantly affected the development of adiposity. Rare NPC1 variants were enriched in young, severely obese male, but not female, participants. Furthermore, the rare NPC1 variants that severely damaged its expression and/or cholesterol-transporting function appeared to be associated with worsened obesity traits in carriers.

Clinically, heterozygous NPC1^+/− parents of NP-C patients are usually considered to be healthy carriers because they do not have any observable neurological symptoms or hepatosplenomegaly, which are characteristics of the NP-C disease. We have previously reported that plasma 7-ketocholesterol levels are elevated in NPC1^+/− carriers (20). This coupled with our observation that male NPC1^+/− carriers had higher BMI values than control subjects strongly support the notion that loss of function in the NPC1 gene may play a fundamentally important regulatory role in body weight maintenance. To our knowledge, the increased adiposity in the heterozygous carriers of functionally damaged mutations in the NPC1 gene has not been reported previously in the literature. Using clinical calculation, we identified that an average 7.8-kg body weight variation (mean variation 76.6 vs. 68.8 kg, respectively, for NPC1^+/− carriers and matched control subjects, P = 0.0001) and a BMI increase of 1.91 kg/m² can be attributable to this particular set of NPC1 mutations, the magnitude of which was at least five times greater than the strongest functional variant (rs9939609) in the FTO gene reported to date (a BMI increase of ∼0.36 kg/m²/risk allele) (14,39).

Of note, we did not observe a significant difference in BMI between NPC1^+/− carriers and matched control subjects or whole population–based control subjects in females. Also, we observed that male carriers had more mutations than females in the SSD and CRLL domain, which are important for NPC1 function. Consistently, the missense mutations in the SSD region found in patients with NP-C disease are associated with NPC1 protein instability by accelerating the degradation of misfolded protein and correlated with severe clinical features (31). This might explain part of the sex differences observed in BMI of NPC1^+/− carriers. However, the relatively small sample size of the carriers could not exclude the possibility that this distribution pattern of mutations in the two sexes of carriers were due to chance. More carriers are needed to confirm this finding. Besides, we observed a higher BMI in male NPC1^+/− carriers than female carriers who had the same NPC1 mutations (P434L, D501Y, and R1077X). Although this BMI difference in male and female NPC1^+/− mutation carriers showed an increased trend compared with the general sex differences in BMI of the population-based cohort (24), they were not significant (ΔBMI 4.17 and 0.09 kg/m², respectively, for the former and latter, P = 0.07). A greater number of male and female NPC1 heterozygotes carrying the same mutations would be helpful for illustrating this sex difference.

The sexual dimorphism on adiposity of NPC1^+/− carriers was also replicated in Npc1^+/− mice showing increased body weight in males than wild-type mice in response to an HFD but not in the female group. Jelinek et al. (40) reported increased body weight in both male and female Npc1^+/− mice fed an HFD. This phenotype discrepancy was considered to be potentially attributable to different genetic background, diet content, or the exposure time to an HFD because our study used C57BL/6 background Npc1^+/− mice, whereas Jelinek et al. (40) used mice with a BALB/cJ background. These two strains of mice show a variable response to HFD stimulation in body weight gain (17). Besides, we started to feed mice with an HFD containing 60 kcal% fat at 6 weeks of age, whereas Jelinek et al. (40) fed mice at 4 weeks of age with an HFD containing 45 kcal% fat. Future studies are clearly warranted to investigate these factors affecting the obesity predisposition of NPC1/Npc1 haploinsufficiency, which will be more informative to understand the phenotype discrepancy in male and female subjects even carrying the same NPC1 genotype.

Together, both heterozygous NPC1 mutant humans and mice indicated a biological interaction between NPC1 mutations and sex during fat accumulation, which was consistent with other clinical and epidemiological observations in support of a more general model of sex-specific gene-hormone interactions for the development of complex diseases (11,41,42). In fact, dozens of studies have revealed the effects of gene defects on adiposity depending on sex (4,42,43), especially male sex predisposing mammals to a greater risk for obesity. Although estrogen may have protective effects against obesity and metabolic disorders, as has been reported in several experimental and observational studies (44–46), the direct role of NPC1 in sex hormone metabolism as yet remains unknown.

Because the subpopulation characterized by extreme clinical phenotypes tended to be enriched by rare genetic determinants for some common diseases (36), we further investigated the effect of NPC1 mutations on adiposity by sequencing and comparing the genotypes of severely obese patients with that of their sex- and region-matched control subjects, identifying 17 rare variants that conferred a significantly higher risk for obesity (3.4% prevalence in obese patients vs. 0.73% in control subjects, OR 4.8). Consistent with the heterozygous NPC1 mutant human and mouse models, a sex-specific association of rare NPC1 variants with obesity was also observed in the case-control cohort. Among the detected rare variants, four (V697I, T743I, Q775K, and V780fs*) were located in the SSD of NPC1, whereby NPC1 responds to free lysosomal sterols and transport lipoprotein-derived cholesterol from late lysosomes to other cellular compartments so as to maintain whole-body cholesterol homeostasis (37), including sex hormone production in adrenal, gonadal, and ovary glands. Interestingly, these four rare variants had much reduced protein levels and severely damaged CTA, characteristics that were further significantly associated with worse obese traits. It is very likely that increased degradation of NPC1 protein because of mutations in the SSD region plays a critically important role in the development of obesity. The other eight rare variants identified in the obese group showed a mild biological change and accordingly obesity that did not worsen as much. Five rare variants (P254R, A321V, G1070A, T1176M, and A1187V) identified in control subjects did not appear to have a damaged cholesterol-transporting function. Future studies are clearly warranted to characterize the NPC1^+/− carriers in larger case-control cohorts or other ethnic populations to confirm the genotype-to-phenotype correlation and gene-by-sex interaction observed in our obese Chinese carriers of rare NPC1 mutations.

In summary, we first described human obesity characterized by some rare and functional mutations of the NPC1 gene. These NPC1^+/− carriers appear to be more likely to develop adiposity traits than noncarriers, particularly in men. These data indicate the critical importance of targeting functional rare NPC1 variants and related pathways for the clinical risk stratification and intervention of human obesity.

Acknowledgments. The authors thank Dr. Jian Wei, from the State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, for her contribution to the CT43 cell line culture; Houfeng Zheng, from the Institute of Aging Research, Hangzhou Normal University, for his helpful comments and information on the project; and the staff and participants for their contributions.

Funding. This research was supported by grants from the National Natural Science Foundation of China (81522011, 81570757, 81570758, 81370963, and 81370949), the National International Science Cooperation Foundation (2015DFA30560), the National Basic Research Program of China (2015CB553601), the Foundation for Innovative Research Groups of the National Natural Science Foundation of China (81621061), and the Shanghai Sailing Program (16YF1409800). S.Li. is supported by the Brown University Endowment for Brazil and China Initiatives, the American Heart Association, the National Heart, Lung, and Blood Institute, and the National Institute of Diabetes and Digestive and Kidney Diseases in the U.S.

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

Author Contributions. R.L. carried out most of the experiments and wrote the paper. Y.Zo. carried out most of the experiments, gathered the data, and wrote the paper. J.H. recruited obese and control participants in stages 1 and 2, gathered the data, and contributed valuable comments and advice on the paper. M.Ca. recruited obese and control participants in stages 1 and 2. B.C., M.Ch., T.N., S.Z., W. Liu, W.G., Y.Zh., W. Li, L.M., Y.S., M.Y., R.W., Q.M., M.X., Y.X., T.W., and Y.B. gathered the data. H.Z., H.X., C.W., and Z.Q. provided 25 family trios of NP-C diseases. J.S. recruited obese and control participants in stages 1 and 2, gathered the data, and assisted with the statistical analysis. K.-h.K.C. and H.C. assisted with statistical analysis. X.Z. contributed valuable comments and advice on the paper. L.Q. and S.La. assisted with statistical analysis and contributed valuable comments and advice on the paper. S.D. contributed with Npc1^+/− mice. B.S. assisted with cholesterol-transporting experiments and materials. S.Li. wrote the paper and contributed valuable comments and advice on the paper. W.W. recruited obese and control participants in stages 1 and 2. G.N. conceived the project, designed the experiments, and wrote the paper. J.W. conceived the project, designed and carried out most of the experiments, and wrote the paper. J.W. 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.