Further Evidence For the Role of ENPP1 in Obesity: Association With Morbid Obesity in Finns

Authors

  • Kaisa Valli-Jaakola,

    Corresponding author
    1. Department of Medicine and Research Program for Molecular Medicine, University of Helsinki, Helsinki, Finland
      (kaisa.valli-jaakola@helsinki.fi)
    Search for more papers by this author
  • Elina Suviolahti,

    1. Department of Molecular Medicine, National Public Health Institute, Helsinki, Finland
    Search for more papers by this author
  • Camilla Schalin-Jäntti,

    1. Department of Medicine and Research Program for Molecular Medicine, University of Helsinki, Helsinki, Finland
    2. Division of Endocrinology, Department of Medicine, Helsinki University Hospital, Helsinki, Finland
    Search for more papers by this author
  • Samuli Ripatti,

    1. Department of Molecular Medicine, National Public Health Institute, Helsinki, Finland
    2. Department of Medical Epidemiology and Biostatistics, Karolinska Insitutet, Stockholm, Sweden
    Search for more papers by this author
  • Kaisa Silander,

    1. Department of Molecular Medicine, National Public Health Institute, Helsinki, Finland
    Search for more papers by this author
  • Laura Oksanen,

    1. Department of Medicine and Research Program for Molecular Medicine, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • Veikko Salomaa,

    1. Department of Epidemiology and Health Promotion, National Public Health Institute, Helsinki, Finland
    Search for more papers by this author
  • Leena Peltonen,

    1. Department of Molecular Medicine, National Public Health Institute, Helsinki, Finland
    2. Department of Medical Genetics, University of Helsinki, Helsinki, Finland
    3. The Broad Institute, Massachusetts Institute of Technology, Boston, Massachusetts, USA
    Search for more papers by this author
  • Kimmo Kontula

    1. Department of Medicine and Research Program for Molecular Medicine, University of Helsinki, Helsinki, Finland
    Search for more papers by this author

(kaisa.valli-jaakola@helsinki.fi)

Abstract

The aim of this study was to investigate a series of single-nucleotide polymorphisms (SNPs) in the genes MC2R, MC3R, MC4R, MC5R, POMC, and ENPP1 for association with obesity. Twenty-five SNPs (2–7 SNPs/gene) were genotyped in 246 Finns with extreme obesity (BMI ≥ 40 kg/m2) and in 481 lean subjects (BMI 20–25 kg/m2). Of the obese subjects, 23% had concomitant type 2 diabetes. SNPs and SNP haplotypes were tested for association with obesity and type 2 diabetes. Allele frequencies differed between obese and lean subjects for two SNPs in the ENPP1 gene, rs1800949 (P = 0.006) and rs943003 (P = 0.0009). These SNPs are part of a haplotype (rs1800949 C-rs943003 A), which was observed more frequently in lean subjects compared to obese subjects (P = 0.0007). Weaker associations were detected between the SNPs rs1541276 in the MC5R, rs1926065 in the MC3R genes and obesity (P = 0.04 and P = 0.03, respectively), and between SNPs rs2236700 in the MC5R, rs2118404 in the POMC, rs943003 in the ENPP1 genes and type 2 diabetes (P = 0.03, P = 0.02 and P = 0.02, respectively); these associations did not, however, remain significant after correction for multiple testing. In conclusion, a previously unexplored ENPP1 haplotype composed of SNPs rs1800949 and rs943003 showed suggestive evidence for association with adult-onset morbid obesity in Finns. In this study, we did not find association between the frequently studied ENPP1 K121Q variant, nor SNPs in the MCR or POMC genes and obesity or type 2 diabetes.

Introduction

The prevalence of obesity, insulin resistance, and type 2 diabetes is increasing worldwide at an alarming rate (1). Identification of the molecular mechanisms predisposing to these conditions is a prerequisite for the development of prevention and treatment strategies.

Genes regulating the central leptin-signaling pathway and the melanocortin system are obvious candidates for obesity (2). Leptin is an important hormone in the physiological system regulating food intake and body weight (3). Leptin is produced by fat tissue and functions as a peripheral satiety signal by reporting nutritional information to the brain. Within the brain, neuropeptides derived from the prohormone proopiomelanocortin (POMC) mediate the signals to a family of melanocortin receptors (4). This signaling cascade plays an important role in the control of energy balance and appetite. Mutations in the melanocortin 4 receptor (MC4R) gene constitute the most common monogenic cause of obesity (2). The MC3R is another candidate gene for human obesity. Three rare MC3R mutations have been described to be associated with obesity (5,6,7,8). Associations between MC3R and obesity-related phenotypes have been reported (9,10) and several linkage results to the genomic region 20q13.2, where MC3R is located, have been identified (11,12,13). Some studies have linked obesity to the chromosomal regions containing the MC2R and MC5R genes (14,15).

The ectonucleotide pyrophosphatase phosphodiesterase 1 (ENPP1) gene has been implicated in common and early onset obesity, insulin resistance, and type 2 diabetes, but its possible role in extreme adult-onset obesity has not been evaluated (16). ENPP1 encodes for a transmembrane glycoprotein that interacts with the insulin receptor by inhibiting its tyrosine kinase activity and subsequent signaling through the receptor (17). This in turn leads to decreased insulin sensitivity. Several studies have demonstrated linkage between the ENPP1 locus and obesity (18,19,20,21). Further evidence for the role of ENPP1 is provided by studies showing an association of the ENPP1 K121Q variant and insulin resistance (22,23,24,25), type 2 diabetes (26), as well as obesity (27,28,29,30). It has been demonstrated that the ENPP1 K121Q variant results in “gain of function” of the insulin receptor, the Q121 variant being a stronger inhibitor of the receptor (31). There are also studies which failed to find evidence for association between ENPP1 and obesity, insulin resistance, or type 2 diabetes (32,33,34,35,36,37). However, meta-analyses show that although the results are somewhat controversial, individuals carrying the Q121 variant have a higher risk of type 2 diabetes (38,39,40). The K121Q variant has received much attention, but little is known about the influence of other ENPP1 polymorphisms on the risk of obesity.

We have previously studied obesity candidate genes involved in the leptin-signaling pathway and the central melanocortin system (10,41,42,43,44,45,46). The aim of this study was to extend our studies by investigating the association between an extreme obesity phenotype and genetic variations in a series of candidate genes MC2R, MC3R, MC4R, MC5R, POMC, and ENPP1. A further aim was to explore possible relationships with insulin resistance and type 2 diabetes. Twenty-five single-nucleotide polymorphisms (SNPs) were genotyped in adult Finnish severely obese and lean subjects. We determined the haplotype blocks within the genes, and tested the SNPs and SNP haplotypes for association with obesity.

Methods and Procedures

Study subjects

The study sample consisted of 246 severely obese subjects (BMI ≥ 40 kg/m2) and 481 lean subjects (BMI 20–25 kg/m2). Clinical characteristics of the obese and lean subjects are presented in Table 1. The obese individuals were recruited from the obesity clinic at the Department of Endocrinology, Helsinki University Central Hospital during years 1989–1995 (47). All subjects were referred to the clinic because of morbid (BMI > 40 kg/m2), treatment-resistant obesity. A detailed history including data on medical history and drug treatment as well as history of weight development was assessed by questionnaire. Blood samples for DNA extraction and serum leptin, lipid, glucose, and insulin were drawn after a 12-h fast. The lean subjects were selected from the national FINRISK97 cohort (48) from the same geographical area as the obese subjects. For lean subjects, serum lipids were measured after a 4-h fast. Data on diabetic status for lean subjects were available from a self-administered questionnaire. In addition, the use of hypoglycemic medication was checked from the nation-wide drug reimbursement register of the Social Insurance Institution of Finland. All subjects were unrelated and represented the same ethnic background, Finnish whites. The study was approved by the local ethics committee and carried out according to the principles of the declaration of Helsinki. All subjects provided written informed consent at participation.

Table 1.  Clinical characteristics of the obese and lean subjects
inline image

SNP selection and genotyping

SNPs rs6127698, rs11575886, and rs3827103 in the MC3R gene, rs11872992 and rs8087522 in the MC4R gene, and rs1044498 in the POMC gene were selected based on the literature. National Center for Biotechnology Information (NCBI; http:www.ncbi.nlm.nih.gov) and Celera Genomics (http:www.celera.com) databases were used to select other SNPs in the MC2R, MC3R, MC4R, MC5R, POMC, and ENPP1 genes. The SNPs were chosen in coding regions of the genes and in noncoding regions that are conserved between human and mouse. The comparison between human and mouse sequences were performed using Pipmaker (49) and Vista (50) software. To confirm that the SNPs with no prior frequency information were polymorphic (had a frequency >10%), 15 control individuals and a DNA-pool of 130 control samples were genotyped. To confirm Mendelian inheritance, 57 nuclear families including father, mother, and a child were genotyped. Reproducibility of genotype data was guaranteed by genotyping 2% of all samples in duplicates. SNPs were excluded from further genotyping if they were in complete linkage disequilibrium (LD) with an adjacent SNP. Assay validity criteria for acceptance of SNPs for genotyping were: (i) no discordant results in duplicates, (ii) allele distributions in Hardy-Weinberg equilibrium, (iii) no Mendelian errors in the nuclear families, and (iv) genotyping success rate >90%. A total of 25 SNPs, including 2–7 SNPs in each gene passed quality control criteria and were selected for further analysis (Table 2). SNPs were genotyped using the homogenous MassEXTEND assay on the MassARRAY system (SEQUENOM, San Diego, CA) according to the manufacturer's instructions. The distance (in kbp), D prime and r2-values between the SNPs genotyped are listed in Supplementary Table S1 online.

Table 2.  Allele frequencies of the SNPs genotyped in obese and lean subjects
inline image

Detection of transcription factor-binding sites

SNPs in noncoding regions reaching significant P values in χ2-tests and logistic regression were initially analyzed using TFSearch (51) (http:www.cbrc.jpresearchdbTFSEARCH.html) and TESS (52) (http:www.cbil.upenn.educgi-bintesstess) software. Further analysis was performed with ConSite (53) (http:mordor.cgb.ki.secgi-binCONSITEconsite) and Con Real (54) (http:conreal.niob.knaw.nl) programs, which enable alignment of two sequences to search for conserved regulatory elements that are affected by the SNP. In the analysis, human sequences were aligned with the genomic sequences of Mus musculus, Rattus norvegicus, Pan troglodytes, and Canis familiaris.

Statistical analysis

Genepop v3.4 Option 2 software (55) (http:genepop.curtin.edu.au) was used to test the LD between adjacent SNPs. Unadjusted differences in allele frequencies between cases and controls were tested using a χ2-test. Haplotype structure and frequencies were estimated using Haploview software (56) (http:www.broad.mit.edumpghaploview) using the algorithm by Gabriel et al. (57). Haplotype specific P values represent the difference in allele frequencies between cases and controls and were calculated separately for each haplotype. The association of each SNP genotype with obesity and type 2 diabetes was assessed using logistic regression analysis in the NCSS 2000 software package (NCSS, Kaysville, UT). The associations between SNPs and type 2 diabetes were analyzed by comparing type 2 diabetic obese subjects with non-type 2 diabetic obese subjects. Logistic regression models were adjusted for age and sex. Type 1 error rate of 0.1 for multiple testing was controlled for by using false discovery rate (58). The false discovery rate corrections were calculated separately for unadjusted χ2, adjusted logistic regression for obesity and adjusted logistic regression for type 2 diabetes tests. We also performed power calculations for the SNPs rs1800949 and rs943003 in the ENPP1 gene, for which allele frequencies between obese and lean subjects were found to differ significantly. Power was also estimated for the K121Q variant (rs1044498) of the ENPP1 gene. Post hoc power for SNPs rs1800949, rs943003, and rs1044498 were 50, 62, and 39%, respectively.

Results

Association between obesity and a haplotype in the ENPP1 gene

Allele frequencies of the SNPs genotyped are presented in Table 2. Allele distributions of the SNPs were in Hardy-Weinberg equilibrium. Three of the SNPs showed differences in allele frequencies between obese subjects and lean controls when tested by a χ2-test (Table 2). Differences in allele frequencies were detected for two adjacent SNPs of the ENPP1 gene: rs1800949 (P = 0.006) located 5′ of exon 1 and rs943003 (P = 0.0009) located in intron 1. For these SNPs, the minor alleles were associated with obesity (allele frequencies in cases vs. controls for rs1800949 and rs943003 were 0.22 vs. 0.16 and 0.48 vs. 0.39, respectively). When corrected by false discovery rate, the association remained significant for both SNPs (rs1800949 and rs943003). These two SNPs of the ENPP1 gene were combined into a haplotype (Figure 1, Table 3). The most common haplotype in block 1 (rs1800949 C-rs943003 A) was observed more frequently in lean than in obese subjects (allele frequencies 0.603 and 0.510 for controls and cases, respectively, P = 0.0007). The minor haplotype (rs1800949 T-rs943003 G) was associated with obesity (frequencies of 0.210 in cases and 0.158 in controls, P = 0.0136).

Figure 1.

Haplotype structure of the ENPP1 gene obtained from the Haploview program. The LD plot shows D′ values. The plot is based on allele frequencies from both obese and lean subjects. D prime values of 1.0 are not shown. The gray shading refers to D prime values, dark gray standing for D′ = 1 and the light gray colors for D′ < 1. White boxes represent an LOD score <2.

Table 3.  Estimated haplotype structure and frequencies of the ENPP1 gene
inline image

Haplotype block structure was estimated for all genes studied, but no other significant results for the haplotypes were detected (results not shown). Association with obesity was also detected with the rs1541276 SNP, which maps to the 3′ region of the MC5R gene (allele frequencies 0.14 and 0.19 in cases vs. controls, P = 0.04, not significant when corrected for multiple testing). When tested by logistic regression and adjusting for age and sex, three SNPs (rs1926065 in the MC3R gene, rs1800949 and rs943003 in the ENPP1 gene) were associated with obesity (Table 2). These associations did not remain significant after correction for multiple testing.

No significant associations between SNPs and type 2 diabetes

When adjusting for age and sex, three SNPs (rs2236700 in the MC5R gene, rs2118404 in the POMC gene, rs943003 in the ENPP1 gene) were associated with type 2 diabetes in the presence of obesity (P = 0.03, P = 0.02, and P = 0.02, respectively). When corrected for multiple testing, none of these associations remained significant.

No conserved regulatory elements at the location of the SNPs rs1800949 and rs943003

ENPP1 SNPs rs1800949 and rs943003, located 5′ to the transcription start point and in intron 1, were initially analyzed using TFSearch and TESS software. Both of these programs predicted the rs1800949 SNP to be located in an upstream stimulatory factor 1-binding site and this site to be destroyed in the minor allele of the SNP. TFSearch predicted the SNP rs943003 to be located in a lymphoid transcription factor 1-binding site, whereas TESS predicted the binding sites for the following transcription factors to be located at this position: E1A-associated protein p300, olfactory neuron-specific transcription factor-1 and Drosophila hunchback. Further analyses were performed with ConSite and Con Real programs permitting analysis of conserved regulatory elements that are possibly affected by the SNP. Using these programs, no conserved regulatory elements at the location of the SNPs rs1800949 or rs943003 were detected when aligning human ENPP1 sequences with the corresponding sequences from the species M. musculus, R. norvegicus, P. troglodytes, or C. familiaris.

Discussion

We examined the association between morbid obesity and genetic variations in MC2R, MC3R, MC4R, MC5R, POMC, and ENPP1 genes. In this study, the common allele of an ENPP1 haplotype, composed of the SNPs rs1800949 and rs943003, was observed more frequently in lean subjects compared to obese subjects. No association of obesity or type 2 diabetes to the previously studied ENPP1 variant K121Q was detected. Less significant associations were detected between the SNPs rs1541276 in the MC5R gene, rs1926065 in the MC3R gene and obesity, and between SNPs rs2236700 in the MC5R gene, rs2118404 in the POMC gene, rs943003 in the ENPP1 gene and type 2 diabetes. These latter associations did not remain significant after correction for multiple testing.

The ENPP1 variant K121Q has frequently been studied for association with insulin resistance, type 2 diabetes, and obesity. In this study, we did not find associations between the K121Q variant and obesity or type 2 diabetes, but other variants of the ENPP1 gene were associated with obesity. Previous results have been inconsistent, some studies finding association to insulin resistance (22,23,24,25), type 2 diabetes (26), and obesity (27,28,29,30), whereas others have failed to find any evidence for association with metabolic traits (32,33,34,35,36,37). The differing results may be explained by factors that vary across different populations and samples studied, such as differences in the obesity phenotype studied, modifying polymorphisms in other genes (59), still unidentified functional polymorphisms that are in LD with the SNPs studied, small sample sizes in various studies, or even different genetic backgrounds of the populations studied (60). Some of the previous positive association results have been detected in studies focusing on early onset obesity (29,30). We studied the possible association to another extreme obesity phenotype, as the subjects were severely obese adults with a BMI ≥40 kg/m2. These reasons may explain the differing association results to some extent. However, we performed post hoc power calculations for the K121Q variant, giving a power of 39%. It is therefore possible that lack of association in this study could be due to low statistical power.

The K121Q variant has received much attention, but little is known about the association of other ENPP1 polymorphisms with obesity, insulin resistance, or type 2 diabetes. In a recent study, Bochenski et al. reported an association between SNP rs997509 and type 2 diabetes, in the presence of obesity, in whites from Poland (61). SNP rs997509 was located in the same haplotype as the K121Q variant. Several studies have investigated an ENPP1 haplotype composed of variants K121Q, IVS20delT-11, and A/G+1044TGA (rs7754561) (29,30,33,34,37). The results regarding the effect of this three-allele risk haplotype on insulin resistance, type 2 diabetes, and obesity have been inconsistent.

To our knowledge, this study is the first report implying suggestive evidence for association between adult morbid obesity and an ENPP1 haplotype composed of SNPs rs1800949 and rs943003. These two ENPP1 SNPs, rs1800949, and rs943003, do not represent functional missense variants affecting the amino acid sequence of the protein. SNP rs1800949 is located 5′ to the transcription start point and rs943003 is located in the first intron of the ENPP1 gene. These regions of a gene usually have important regulatory functions. We were not able to detect conserved regulatory elements at the location of these SNPs using computational in silico analysis. These SNPs may nevertheless serve as regulatory elements of the ENPP1 gene. Alternatively, these sites could be in LD with functional elements of the ENPP1 gene or be in LD with a yet unidentified functional polymorphism of another gene. Comparison of the LD between HapMap SNPs and those SNPs genotyped in this study demonstrated that ∼14% of variation within the ENPP1 gene was covered with the SNPs genotyped. This seemingly low figure is due to the first large intron of the gene showing very low LD.

In this study, SNPs rs1541276 in the MC5R gene (unadjusted χ2-test) and rs1926065 in the MC3R gene (logistic regression adjusted for age and sex) were marginally associated with obesity. Not much is known about the possible role of MC5R in the regulation of body weight. In 1997, Chagnon et al. demonstrated that genetic variants of MC5R were associated with obesity-related traits (14), but no other studies replicating these associations have been published. It has been suggested that MC5R contributes to the regulation and function of exocrine glands and to certain immune responses (62). To date, only three mutations in the MC3R gene (A70T, M134I, and I183N) have been described to be associated with obesity (5,6,7,8). Boucher et al. reported an MC3R +2138InsGAGACC polymorphism to be associated with fat mass, percent body fat, and total abdominal fat (9).

Some evidence for associations were also detected between SNPs rs2236700 in the MC5R gene, rs2118404 in the POMC gene, rs943003 in the ENPP1 gene and type 2 diabetes (obese subjects, logistic regression, adjusted for age and sex). Previously, Santoro et al. demonstrated that an insertional polymorphism of the POMC gene was associated with fasting insulin levels in childhood obesity (63), but no other reports regarding the role of POMC in insulin resistance or type 2 diabetes have been published. The subgroup of patients characterized by both morbid obesity and type 2 diabetes in this study was small (N = 57). In the future, larger study samples are needed to investigate the possible relationship between ENPP1 and type 2 diabetes.

One limitation of this study is the sample size. It should be emphasized, however, that we studied a very unique cohort of severely obese patients in a genetic isolate of Finns, which should improve possibilities to identify an underlying genetic association. In view of the fact that ∼1% of Finns had a BMI >40 kg/m2 at the time of sample collection (64), a replication study of similar nature would represent a difficult task.

In conclusion, we describe here an ENPP1 haplotype showing suggestive association with morbid obesity in adults. Further studies investigating the role of these SNPs in larger cohorts, in different obesity phenotypes and in obesity characterized by concomitant type 2 diabetes are needed.

Supplementary Material

Supplementary material is linked to the online version of the paper at http:www.nature.comoby

Acknowledgment

We are grateful to the volunteers who participated in our study. We thank Ms Siv Knaappila and Ms Minna Suvela for excellent technical assistance. This work was supported by grants from Finska Läkaresällskapet, the Sigrid Juselius Foundation, the Finnish Academy, the Research Funds from the University Central Hospital in Helsinki, the Jalmari and Rauha Ahokas foundation, Helsingin Sanomat Centennial Foundation, Finnish Foundation for Cardiovascular Research, and the Finnish Clinical Chemistry Research Foundation.

Disclosure

The authors declared no conflict of interest.

Ancillary