Follow-Up Studies of a Genome-Wide Association Analysis in Pima Indians

National Institute Of Diabetes And Digestive And Kidney Diseases

Investigators

Linked publications & trials

Abstract

Genetic studies in Pima Indians have been hampered by the lack of commercially available genotyping arrays which contain variation enriched or unique to this population, and the absence of publicly available genomic datasets that capture linkage disequilibrium for imputation in this population. To overcome these issues, we obtained whole-genome sequence data on 335 Pima Indians and in collaboration with Affymetrix designed and manufactured a custom Pima Indian genotypic array. This array was then used to genotype 7,701 Pima Indians from the Gila River Indian Community with longitudinal data on T2D, BMI and T2D complications including diabetic nephropathy (DN), of which 550 subjects had additional data on related metabolic traits from non-diabetic inpatient studies in our CRC. Subsequently, 3,209 Urban Indians (all are 50% American Indian of any tribe) with cross-sectional data on T2D and DN have been genotyped with this array. Using genotypic data as a scaffold, we imputed 7,996,530 variants with an info score >0.5 for the 7,701 Pima Indians. These imputed variants allow us to perform comprehensive analyses in these Pima samples, but also enable us to contribute our Pima Indian data for large collaborative meta-analyses. Our GWAS for T2D and insulin secretory function detected 3 independent signals in KCNQ1 and a newly identified signal in TH. The KCNQ1 SNPs tag 39 SNPs in intron 15 of KCNQ1. Crispr/Cas9 gene-editing was used to create a targeted deletion of intron 15 (48kb) in KCNQ1, and RT-PCR showed reduced RNA expression levels of KCNQ1 and neighboring genes KCNQ-OT1 and CDKN1C in these cells. We are pinpointing the actual functional SNPs in this intron by systematically making constructs containing fewer variants and assessing their impact on KCNQ1, KCNQ1-OT1 and CDKN1C promoter activity via in intro luciferase assays. In addition, we are using the Crispr/Cas9 gene-editing approach to generate knockouts in the human beta cell line Endo-BHC1 to better assess the function of these variants in a more biologic cell type. The other GWAS signal in TH, which encodes tyrosine hydroxylase, is also being studied. This variant had a gender specific effect in its association with T2D and insulin secretory function. Our GWAS SNP is in perfect linkage disequilibrium with 2 SNPs in the promoter of TH which appear to be functional. One of these functional SNPs disrupts the binding site for SRY, a factor expressed on the Y chromosome. We propose that gender-specific expression of SRY underlies the gender specific association of these variants with T2D. Our GWAS in Pima Indians also identified a variant within PFKFB2 (rs17258746) that associated with BMI. PFKFB2 encodes 6-phosphofructo-2-kinase/fructose 2,6-bisphosphatase isoform 2, which plays a role in glucose metabolism. To follow-up on the GWAS, tag SNPs across PFKFB2 were genotyped in American Indians who had longitudinal data on BMI (n=6839), T2D (n=7710), diabetic nephropathy (DN; n=2452), % body fat (n=555) and insulin secretion (n=298). Two SNPs were further genotyped in urban American Indians to assess replication for DN (n=864). PFKFB2 expression was measured in 201 adipose biopsies using real-time RT-PCR and 61 kidney biopsies using the Affymetrix U133 array. Two SNPs (rs17258746 and rs11120137), which capture the same signal, were associated with maximum BMI in adulthood, maximum BMI z-score in childhood and % body fat in adulthood. The adiposity-increasing allele correlated with lower PFKFB2 adipose expression. Lower expression of PFKFB2 further correlated with higher % body fat and BMI. This allele was also associated with increased risk for DN in both cohorts of American Indians and similarly correlated with lower PFKFB2 expression in kidney glomeruli. The same allele was also associated with lower insulin secretion assessed by acute insulin response and 30-min plasma insulin concentrations. Therefore, we propose that variation in PFKFB2 appears to reduce PFKFB2 expression in adipose and kidney tissues, and thereby increase risk for adiposity and DN. We also performed a genome-wide association study for energy expenditure (EE) which represents the first ever genome-wide interrogation for variation that affects this predictor of obesity. We identified a variant in GPR158 that associates with two different measures of EE (24-hr EE measured in a metabolic chamber and resting metabolic rate) which is inversely related to body fatness and BMI in Pima Indians. The risk allele for this variant has a frequency of 0.60 in Pima Indians and 0.004 in Caucasians. We have also performed functional studies and have shown that this variant affects gene expression levels. Recently we also conducted a genome-wide association analysis for fasting LDL levels and identified rs12740374 (G/T) as being robustly associated with LDL (P= 1 1022; effect ()= 8% lower per copy of minor allele, T) and total cholesterol (P= 9 1015; = 4% lower). An expression-QTL (eQTL) analysis in the GTEx database showed that rs12740374 associates with the expression of CELSR2 (P= 1034) and neighboring genes PSRC1 (P= 1037) and SORT1 (P= 1054) in liver (N= 208). It is reported that rs12740374 alters the hepatic expression of the SORT1 gene by creating a C/EBP transcription factor binding site. However, the database also showed that rs12740374 is a very strong eQTL for CELSR2 (P= 1076) and PSRC1 (P= 10-12) in skeletal muscle (N= 706). Therefore, we looked for additional phenotypes associated with this variant which may have a physiologic role in muscle and found that rs12740374 associates with resting metabolic rate (RMR) measured by a ventilated hood system (N= 509; P= 0.003; = -0.127). A major determinant of human metabolic rate is muscle mass. We speculated rs12740374, which has a known role for LDL in liver, maybe tagging a functional variant in the muscle that underlies association with energy expenditure, and observed that rs6670347 (T/C) (r2= 0.98) located in CELSR2 intron is similarly associated with LDL and RMR. An eQTL analysis in American Indian skeletal muscle (N= 202) and adipose tissue samples (N= 192) showed that rs6670347 contributes to higher CELSR2 expression in muscle (P= 2 107, = 0.61 SD units per copy of minor allele, C). This eQTL appears to be muscle-specific as no association was detected in studied adipose tissue or other non-muscle tissues reported in the GTEx portal. Further in-silico analysis revealed that rs6670347 forms the core binding motif of the NR3C1 transcription factor, and the C allele enables a stronger binding. NR3C1 encodes a receptor for glucocorticoids, which are key regulators of muscle mass, and their prolonged exposure induces atrophy. Glucocorticoid resistance is shown to decrease CELSR2 expression in human leukemia cells and its inactivation in mice reduces muscle atrophy, which could explain its role in altering RMR. Our findings indicate that different variants in the CELSR2/ PSRC1/SORT1 gene cluster may have tissue-specific effects on diverse metabolic traits.