|Home | About | Journals | Submit | Contact Us | Français|
The low‐density lipoprotein receptor‐related protein 5 (LRP5) gene, essential for glucose and cholesterol metabolism, may have a role in the aetiology of obesity, an important risk factor for diabetes.
To investigate the association between LRP5 polymorphisms and obesity, 27 single‐nucleotide polymorphisms (SNPs), spacing about 5 kb apart on average and covering the full transcript length of the LRP5 gene, were genotyped in 1873 Caucasian people from 405 nuclear families. Obesity (defined as body mass index (BMI) >30 kg/m2) and three obesity‐related phenotypes (BMI, fat mass and percentage of fat mass (PFM)) were investigated.
Single markers (12 tagging SNPs and 4 untaggable SNPs) and haplotypes (5 blocks) were tested for associations, using family‐based designs. SNP4 (rs4988300) and SNP6 (rs634008) located in block 2 (intron 1) showed significant associations with obesity and BMI after Bonferroni correction (SNP4: p<0.001 and p=0.001, respectively; SNP6: p=0.002 and 0.003, respectively). The common allele A for SNP4 and minor allele G for SNP6 were associated with an increased risk of obesity. Significant associations were also observed between common haplotype A–G–G–G of block 2 with obesity, BMI, fat mass and PFM with global empirical values p<0.001, p<0.001, p=0.003 and p=0.074, respectively. Subsequent sex‐stratified analyses showed that the association in the total sample between block 2 and obesity may be mainly driven by female subjects.
Intronic variants of the LRP5 gene are markedly associated with obesity. We hypothesise that such an association may be due to the role of LRP5 in the WNT signalling pathway or lipid metabolism. Further functional studies are needed to elucidate the exact molecular mechanism underlying our finding.
Low‐density lipoprotein receptor‐related protein 5 (LRP5), consisting of 23 exons and spacing 136.6 kb, is mapped to chromosome 11q13.4 in humans and is part of the low‐density lipoprotein receptor family of cell surface receptors.1,2LRP5 plays an important part in the WNT signalling pathway,3,4,5,6 may regulate bone and eye development,7,8 and is important for glucose and cholesterol metabolism.9
Obesity is a growing healthcare problem and risk factor for common diseases such as type 2 diabetes, heart diseases and hypertension. The prevalence of obesity has increased so rapidly that the obese population in developed countries has more than doubled over the past decade.10 A marked parallel increase in the prevalence of obesity‐related disorders, such as diabetes, is seen worldwide.11
Although a genetic determination of obesity has been established, with estimates of heritability >0.50 for the variation in body mass index (BMI), fat mass and percentage of fat mass (PFM),12,13,14,15,16 the underlying susceptibility genes remain largely unknown. LRP5 polymorphisms have been found important to complex diseases or traits that are related to obesity.1,8,9,17,18,19,20 However, the direct relationship between LRP5 and obesity has never been studied. Given the role of LRP5 in obesity‐related metabolic pathways and diseases, we hypothesise and test in this study that LRP5 is associated with obesity phenotypes.
The study was approved by the Creighton University Institutional Review Board. Signed informed consent was obtained from all study participants before they entered the study. The study participants came from an expanding database created for ongoing studies in the Osteoporosis Research Center of Creighton University, Omaha, Nebraska, USA, to search for genes underlying common human complex traits, including obesity and osteoporosis. The study design and recruitment procedures have been published previously.21 In all, 405 nuclear families (family size ranged from 3 to 12, average 4.62) with 1873 people were analysed in this study, including 740 parents, 389 male children and 744 female children. All of the participants were Midwestern US Caucasians of European origin.
Genomic DNA was extracted from whole blood using a commercial isolation kit (Gentra Systems, Minneapolis, Minnesota, USA) following the procedure detailed in the kit. DNA concentration was assessed by a DU530 UV/VIS Spectrophotometer (Beckman Coulter, Fullerton, California, USA). A total of 41 SNPs within and around LRP5 were selected on the basis of the following criteria: (1) validation status (validated experimentally in human populations), especially in Caucasians, (2) an average density of 1 SNP per 3 kb, (3) degree of heterozygosity—that is, minor allele frequency (MAF) >0.05, (4) functional relevance and importance, and (5) reported to the dbSNP (SNP database) by various sources. Twenty seven SNPs were successfully genotyped using the high‐throughput BeadArray SNP genotyping technology of Illumina (San Diego, California, USA). The average rate of missing genotype data was approximately 0.05%. The average genotyping error rate estimated through blind sample duplication was <0.01%. The 27 SNPs successfully genotyped were spaced about 5 kb apart on average and covered the full length (including upstream and downstream of potential regulatory regions) of the LRP5 gene (fig 11).
BMI was calculated as body weight (kg) divided by the square of height (m). Weight was measured in light indoor clothing, using a calibrated balance beam scale, and height was measured using a calibrated stadiometer. Fat mass was measured by dual‐energy x ray absorptiometry with a Hologic 4500 machine (Hologic INC, Bedford, MA, USA). PFM was calculated as the ratio of fat mass to body weight. The measurement precisions of weight, height and fat mass, as reflected by the coefficient of variation, were 1.2%, 0.9% and 2.2%, respectively. We defined obesity as a dichotomous trait using the World Health Organization criterion (obese, BMI >30 kg/m2; and non‐obese, BMI<30 kg/m2).
PedCheck22 was used to check Mendelian consistency of SNP genotype data, and any inconsistent genotypes were removed. Then the error‐checking option embedded in Merlin23 was run to identify and disregard the genotypes flanking excessive recombinants, further reducing genotyping errors. The Hardy–Weinberg equilibrium for each SNP was tested in parents using the PEDSTATS procedure implemented in Merlin.
Linkage disequilibrium block structure was examined by the program Haploview.24 The D′ values for all pairs of SNPs were calculated and the haplotype blocks estimated using the confidence interval method.25 SNPs with low MAF may inflate estimates of D′ and the use of confidence‐bound estimates for D′ reduces this bias. The default settings were used in these analyses, which invoked a one‐sided upper 95% confidence bound of D′>0.98 and a lower bound of >0.7 to define SNP pairs in strong linkage disequilibrium. A block is identified when at least 95% of SNP pairs in a region meet these criteria for strong linkage disequilibrium. Haplotypes were reconstructed and their frequencies estimated using an accelerated expectation–maximisation algorithm similar to the partition–ligation method26 implemented in Haploview. Haplotype tag SNPs were selected by Haploview on a block‐by‐block basis.
For association analyses, quantitative traits were adjusted for age and sex. Initially, tag SNPs for each block as well as untaggable SNPs (which do not belong to any haplotype block) were tested for single‐marker allelic association, by using family‐based association analysis (FBAT) V.1.55.27 The additive genetic model was applied as it performs well even when the true genetic model is not an additive one.27,28,29 The allelic association examines the transmission of the interested markers from parents to the affected (obese as defined earlier) offspring. The null hypothesis here is no linkage and there is no association between the marker and any obesity susceptibility locus. The ‐o flag was used to minimise the variance of the FBAT statistic. Bonferroni correction was used to adjust for the multiple testing, with the single test significance level established as α=0.05 divided by the number of tests.
The haplotype version of FBAT (HBAT)30 was used to test the association between phenotypes and within‐block haplotypes. For each block, haplotype‐specific analysis (ie, each haplotype was tested against all others) was first carried out to find the highest‐risk haplotype. Similar to FBAT, HBAT is robust to population admixture, phenotype distribution mis‐specification and ascertainment bias. To account for the multiple haplotypes in each block, empirical global (multihaplotype) p values were obtained through 10000 permutations using the program HBAT, which is computed by using Monte Carlo samples from the null distribution of no linkage and no association.30
FBAT and HBAT were also carried out in sex‐stratified (for offspring of the nuclear families) subsamples. In addition, population‐based association analysis was carried out in the parental group (for unrelated people) using the PROC generalised linear model (GLM) section of SAS V.8.0e. Normality tests for the quantitative obesity phenotypes (BMI, fat mass, PFM) were conducted and data normalised before running the GLM procedure.
Table 11 summarises the basic characteristics of the studied sample stratified by sex. The unadjusted mean trait values of BMI, fat mass and PFM between men and women were significantly different, as assessed by the t test (p<0.001). Women had higher mean fat mass and PFM, but men had a higher mean BMI. Among the 1873 participants, 262 women (125 mothers, 137 daughters) and 193 men (119 fathers, 74 sons) were obese. Table 22 shows marker information, including the name, chromosomal position and MAF for SNPs successfully genotyped. SNP17, which has MAF<0.01, was discarded for subsequent analyses according to common practice.31 We found no significant deviation from the Hardy–Weinberg equilibrium for all SNPs (table 22).). Less than 0.02% of the overall 50571 genotypes, namely <11 genotypes, were omitted for LRP5 gene because of the violation of Mendelian Law of Inheritance.
We identified five blocks with high linkage disequilibrium, which ranged in size from 5 to 35 kb (fig 11).). Blocks 1 and 2 mainly spanned from promoter to intron 1. Block 3 ranged from introns 1 to 5. Block 4 extended from introns 5 to 7. Block 5 ranged from introns 7 to 19. Twelve tag SNPs were identified. Four SNPs (SNP8, SNP19, SNP26 and SNP27) had low linkage disequilibrium with any other SNPs and could not be assigned to any blocks (untaggable). Figure 22 shows the haplotype structure, diversity and tag SNPs.
In all, 16 SNPs (12 tag SNPs and 4 independent SNPs) were used for single‐marker association analysis. The significance level for a single test is therefore set as α=0.001 (α=0.05/(16×3)=0.00104; 16 SNPs, 3 related phenotypes). This correction for multiple testing is overly conservative, as the many SNPs are in high linkage disequilibrium and from the same haplotype block, and the three phenotypes are correlated.
Table 33 shows the single‐marker association results. Among the 16 SNPs, SNP4 in block 2 shows the strongest association across all phenotypes. SNP4 (rs4988300) had a preferential transmission of the common allele A to the obese offspring, with p<0.001. In addition, a significant association was found for BMI with p=0.001, which remained significant after Bonferroni adjustment. For fat mass and PFM, the p values were 0.006 and 0.038, respectively, indicating a similar trend, although not reaching a significance level after our overly conservative Bonferroni adjustment. In the same block, a significant association was also detected at SNP6, where the minor allele G was overtransmitted to the obese offspring.
Haplotype analysis carried out in each block confirmed the association at the individual SNP level (table 44).). Specifically, a common haplotype A–G–G–G (frequency 0.376) in block 2 was significantly associated with obesity phenotypes, with the global p values being p<0.001 for obesity, p<0.001 for BMI, p=0.003 for fat mass and p=0.074 for PFM.
FBAT in sex‐specific subgroups showed that the association between block 2 and obesity was mainly driven by female participants, whereas the signals were non‐significant in men (table 55).). GLM analyses of unrelated people further showed that women homozygotic for the overtransmitted allele had larger trait values of BMI, fat mass and PFM, with p values being 0.056, 0.071 and 0.12 for SNP4, and 0.063, 0.098 and 0.13 for SNP6, respectively (fig 33).
To the best of our knowledge, this is the first attempt of FBAT of the relationship between LRP5 polymorphisms and obesity. Our purpose is motivated by the multiple lines of evidence showing that the LRP5 may be associated with the aetiology of obesity.
Firstly, Figueroa et al19 found that LRP5 possibly had a role in type 1 diabetes; Fujino et al9 showed that LRP5 was required for cholesterol and glucose metabolism and is an obvious candidate gene for type 2 diabetes. As diabetes and obesity are highly correlated both genetically and phenotypically,11 genes associated with regulating diabetes may also be candidates for obesity. One good example is the vitamin D receptor gene.32,33 Similarly, LRP5 might exert certain effects on both diabetes and obesity.34
Secondly, the marked genetic correlation between obesity phenotypes such as BMI and osteoporosis phenotyes such as bone mineral density was already established, suggesting the existence of common genetic factors contributing to both obesity and osteoporosis.35,36,37,38 In the study sample, we observed significant genetic correlation between BMI and bone mineral density as well (p<0.05 at spine and hip, data not shown). From the molecular perspective, Pei et al38a found the correlation between bone and fat development, and that both adipogenesis and osteogenesis could be mediated by the same molecular pathway.39 Given the overlapping genetic factors for obesity and osteoporosis as well as the key role of the WNT/LRP5 signalling pathway in switching between adipogenesis and osteoclastogenesis,40,41LRP5 could be one of the pleiotropic genetic factors underlying both phenotypes. Thus, it is logical to consider LRP5, an important candidate gene for osteoporosis,18,42,43,44,45 also as a candidate gene for obesity.
In our study, definite associations with various obesity phenotypes were detected for SNP4, SNP6 and their haplotypes in block 2 of LRP5. The consistent relevant findings from analyses of both dichotomous (defined for obese status) and quantitative traits (BMI, fat mass and PFM) strongly supported that LRP5 is important to obesity. However, the mechanism of how the detected major LRP5 polymorphisms (or functional variants being tagged) influence obesity is unknown, although the effect of the associated markers is highly important. We hypothesise that the major block 2 polymorphisms in intron 1 are in linkage disequilibrium with functional variants regulating mRNA production, which may act in the same way as the Sp1 polymorphism in intron 1 of the collagen type I α1 gene.46,47 Bioinformatic analysis using the VISTA program48 to compare the human and mouse LRP5 genomic sequences (>140 kb from the Celera database) showed that the block 2 region was highly conserved (fig 11),), suggesting the existence of functional variants in this block that may be the “causal” variants influencing obesity phenotypes. Further molecular genetic studies will be helpful to eventually unravel the myth.
The associations observed in block 2 were found to exist mainly in women, suggesting that some sex‐specific factors might be related to the action of LRP5 on obesity phenotypes. Similar findings have recently been reported for the oestrogen receptor α gene, which was associated with obesity only in women.49 These data suggest the existence of some loci conferring risk for obesity in a sex‐specific manner.
In women, the population‐based associations with obesity for block 2 polymorphisms are borderline significant (p<0.10) but with the trends of association across BMI, fat mass and PFM being consistent (fig 33)) with FBAT. Several possible reasons can be given for the only borderline significance in the random sample: potential population stratification may mask the real association50 and, secondly, this might be explained by the lack of statistical power owing to the insufficient sample size of the unrelated group of women.
In summary, we reported the significant association between SNPs and haplotypes in the LRP5 gene with human obesity, suggesting the importance of LRP5 in the pathogenesis of human obesity. Our findings should be stimulating for following up molecular functional studies.
BMI - body mass index
FBAT - family‐based association analysis
GLM - generalised linear model
HBAT - haplotype version of FBAT
LRP5 gene - low‐density lipoprotein receptor‐related protein 5 gene
MAF - minor allele frequencies
PFM - percentage of fat mass
SNP - single‐nucleotide polymorphism
Funding: This work was partially supported by grants from NIH (R01 AR050496, K01 AR02170‐01, R01 AR45349‐01 and R01 GM60402‐01A1) and an LB595 grant from the State of Nebraska. The study also benefited from grants from the National Science Foundation of China, Huo Ying Dong Education Foundation, HuNan Province, Xi'an Jiaotong University and the Ministry of Education of China.
Competing interests: None declared.