Novel signals and polygenic score for height are associated with pubertal growth traits in Southwestern American Indians.

Most genetic variants associated with adult height have been identified through large genome-wide association studies (GWASs) in European-ancestry cohorts. However, it is unclear how these variants influence linear growth during adolescence. This study uses anthropometric and genotypic data from a longitudinal study conducted in an American Indian community in Arizona between 1965-2007. Growth parameters (i.e. height, velocity, and timing of growth spurt) were derived from the Preece-Baines growth model, a parametric growth curve fitted to longitudinal height data, in 787 participants with height measurements spanning the whole period of growth. Heritability estimates suggested that genetic factors could explain 25% to 71% of the variance of pubertal growth traits. We performed a GWAS of growth parameters, testing their associations with 5 077 595 imputed or directly genotyped variants. Six variants associated with height at peak velocity (P < 5 × 10-8, adjusted for sex, birth year and principal components). Implicated genes include NUDT3, previously associated with adult height, and PACSIN1. Two novel variants associated with duration of growth spurt (P < 5 × 10-8) in LOC105375344, an uncharacterized gene with unknown function. We finally examined the association of growth parameters with a polygenic score for height derived from 9557 single nucleotide polymorphisms (SNPs) identified in the GIANT meta-analysis for which genotypic data were available for the American Indian study population. Height polygenic score was correlated with the magnitude and velocity of height growth that occurred before and at the peak of the adolescent growth spurt, indicating overlapping genetic architecture, with no influence on the timing of adolescent growth.

The utility of a type 2 diabetes polygenic score in addition to clinical variables for prediction of type 2 diabetes incidence in birth, youth and adult cohorts in an Indigenous study population.

AIMS/HYPOTHESIS: There is limited information on how polygenic scores (PSs), based on variants from genome-wide association studies (GWASs) of type 2 diabetes, add to clinical variables in predicting type 2 diabetes incidence, particularly in non-European-ancestry populations. METHODS: For participants in a longitudinal study in an Indigenous population from the Southwestern USA with high type 2 diabetes prevalence, we analysed ten constructions of PS using publicly available GWAS summary statistics. Type 2 diabetes incidence was examined in three cohorts of individuals without diabetes at baseline. The adult cohort, 2333 participants followed from age ≥20 years, had 640 type 2 diabetes cases. The youth cohort included 2229 participants followed from age 5-19 years (228 cases). The birth cohort included 2894 participants followed from birth (438 cases). We assessed contributions of PSs and clinical variables in predicting type 2 diabetes incidence. RESULTS: Of the ten PS constructions, a PS using 293 genome-wide significant variants from a large type 2 diabetes GWAS meta-analysis in European-ancestry populations performed best. In the adult cohort, the AUC of the receiver operating characteristic curve for clinical variables for prediction of incident type 2 diabetes was 0.728; with the PS, 0.735. The PS's HR was 1.27 per SD (p=1.6 × 10-8; 95% CI 1.17, 1.38). In youth, corresponding AUCs were 0.805 and 0.812, with HR 1.49 (p=4.3 × 10-8; 95% CI 1.29, 1.72). In the birth cohort, AUCs were 0.614 and 0.685, with HR 1.48 (p=2.8 × 10-16; 95% CI 1.35, 1.63). To further assess the potential impact of including PS for assessing individual risk, net reclassification improvement (NRI) was calculated: NRI for the PS was 0.270, 0.268 and 0.362 for adult, youth and birth cohorts, respectively. For comparison, NRI for HbA1c was 0.267 and 0.173 for adult and youth cohorts, respectively. In decision curve analyses across all cohorts, the net benefit of including the PS in addition to clinical variables was most pronounced at moderately stringent threshold probability values for instituting a preventive intervention. CONCLUSIONS/INTERPRETATION: This study demonstrates that a European-derived PS contributes significantly to prediction of type 2 diabetes incidence in addition to information provided by clinical variables in this Indigenous study population. Discriminatory power of the PS was similar to that of other commonly measured clinical variables (e.g. HbA1c). Including type 2 diabetes PS in addition to clinical variables may be clinically beneficial for identifying individuals at higher risk for the disease, especially at younger ages.

The power of TOPMed imputation for the discovery of Latino-enriched rare variants associated with type 2 diabetes.

AIMS/HYPOTHESIS: The Latino population has been systematically underrepresented in large-scale genetic analyses, and previous studies have relied on the imputation of ungenotyped variants based on the 1000 Genomes (1000G) imputation panel, which results in suboptimal capture of low-frequency or Latino-enriched variants. The National Heart, Lung, and Blood Institute (NHLBI) Trans-Omics for Precision Medicine (TOPMed) released the largest multi-ancestry genotype reference panel representing a unique opportunity to analyse rare genetic variations in the Latino population. We hypothesise that a more comprehensive analysis of low/rare variation using the TOPMed panel would improve our knowledge of the genetics of type 2 diabetes in the Latino population. METHODS: We evaluated the TOPMed imputation performance using genotyping array and whole-exome sequence data in six Latino cohorts. To evaluate the ability of TOPMed imputation to increase the number of identified loci, we performed a Latino type 2 diabetes genome-wide association study (GWAS) meta-analysis in 8150 individuals with type 2 diabetes and 10,735 control individuals and replicated the results in six additional cohorts including whole-genome sequence data from the All of Us cohort. RESULTS: Compared with imputation with 1000G, the TOPMed panel improved the identification of rare and low-frequency variants. We identified 26 genome-wide significant signals including a novel variant (minor allele frequency 1.7%; OR 1.37, p=3.4 × 10-9). A Latino-tailored polygenic score constructed from our data and GWAS data from East Asian and European populations improved the prediction accuracy in a Latino target dataset, explaining up to 7.6% of the type 2 diabetes risk variance. CONCLUSIONS/INTERPRETATION: Our results demonstrate the utility of TOPMed imputation for identifying low-frequency variants in understudied populations, leading to the discovery of novel disease associations and the improvement of polygenic scores. DATA AVAILABILITY: Full summary statistics are available through the Common Metabolic Diseases Knowledge Portal ( https://t2d.hugeamp.org/downloads.html ) and through the GWAS catalog ( https://www.ebi.ac.uk/gwas/ , accession ID: GCST90255648). Polygenic score (PS) weights for each ancestry are available via the PGS catalog ( https://www.pgscatalog.org , publication ID: PGP000445, scores IDs: PGS003443, PGS003444 and PGS003445).

Characterization of Exome Variants and Their Metabolic Impact in 6,716 American Indians from the Southwest US.

Applying exome sequencing to populations with unique genetic architecture has the potential to reveal novel genes and variants associated with traits and diseases. We sequenced and analyzed the exomes of 6,716 individuals from a Southwestern American Indian (SWAI) population with well-characterized metabolic traits. We found that the SWAI population has distinct allelic architecture compared to populations of European and East Asian ancestry, and there were many predicted loss-of-function (pLOF) and nonsynonymous variants that were highly enriched or private in the SWAI population. We used pLOF and nonsynonymous variants in the SWAI population to evaluate gene-burden associations of candidate genes from European genome-wide association studies (GWASs) for type 2 diabetes, body mass index, and four major plasma lipids. We found 19 significant gene-burden associations for 11 genes, providing additional evidence for prioritizing candidate effector genes of GWAS signals. Interestingly, these associations were mainly driven by pLOF and nonsynonymous variants that are unique or highly enriched in the SWAI population. Particularly, we found four pLOF or nonsynonymous variants in APOB, APOE, PCSK9, and TM6SF2 that are private or enriched in the SWAI population and associated with low-density lipoprotein (LDL) cholesterol levels. Their large estimated effects on LDL cholesterol levels suggest strong impacts on protein function and potential clinical implications of these variants in cardiovascular health. In summary, our study illustrates the utility and potential of exome sequencing in genetically unique populations, such as the SWAI population, to prioritize candidate effector genes within GWAS loci and to find additional variants in known disease genes with potential clinical impact.

Identification and characterization of the long non-coding RNA NFIA-AS2 as a novel locus for body mass index in American Indians.

BACKGROUND: Genome-wide association studies have shown that body mass index (BMI), an estimate of obesity, is highly polygenic. Individual variants typically have small effect sizes, making it challenging to identify unique loci in under-represented ethnic groups which lack statistical power due to their small sample size. Yet obesity is a major health disparity and is particularly prevalent in southwestern American Indians. Here, we identify and characterize a new locus for BMI that was detected by analyzing moderate associations with BMI obtained in a population-based sample of southwestern American Indians together with the well-powered GIANT dataset. METHODS: Genotypes for 10.5 million variants were tested for association with BMI in 5870 American Indians and 2600 variants that showed an association P < 10-3 in the American Indian sample were combined in a meta-analysis with the BMI data reported in GIANT (N = 240,608). The newly identified gene, NFIA-AS2 was functionally characterized, and the impact of its lead associated variant rs1777538 was studied both in-silico and in-vitro. RESULTS: Rs1777538 (T/C; C allele frequency = 0.16 in American Indians and 0.04 in GIANT, meta-analysis P = 5.0 × 10-7) exhibited a large effect in American Indians (1 kg/m2 decrease in BMI per copy of C allele). NFIA-AS2 was found to be a nuclear localized long non-coding RNA expressed in tissues pertinent to human obesity. Analysis of this variant in human brown preadipocytes showed that NFIA-AS2 transcripts carrying the C allele had increased RNA degradation compared to the T allele transcripts (half-lives = 9 h, 13 h respectively). During brown adipogenesis, NFIA-AS2 featured a stage-specific regulation of nearby gene expression where rs1777538 demonstrated an allelic difference in regulation in the mature adipocytes (the strongest difference was observed for L1TD1, P = 0.007). CONCLUSION: Our findings support a role for NFIA-AS2 in regulating pathways that impact BMI.

A missense variant Arg611Cys in LIPE which encodes hormone sensitive lipase decreases lipolysis and increases risk of type 2 diabetes in American Indians.

AIMS: Hormone sensitive lipase (HSL), encoded by the LIPE gene, is involved in lipolysis. Based on prior animal and human studies, LIPE was analysed as a candidate gene for the development of type 2 diabetes (T2D) in a community-based sample of American Indians. MATERIALS AND METHODS: Whole-exome sequence data from 6782 participants with longitudinal clinical measures were used to identify variation in LIPE. RESULTS: Amongst the 16 missense variants identified, an Arg611Cys variant (rs34052647; Cys-allele frequency = 0.087) significantly associated with T2D (OR [95% CI] = 1.38 [1.17-1.64], p = 0.0002, adjusted for age, sex, birth year, and the first five genetic principal components) and an earlier onset age of T2D (HR = 1.22 [1.09-1.36], p = 0.0005). This variant was further analysed for quantitative traits related to T2D. Amongst non-diabetic American Indians, those with the T2D risk Cys-allele had increased insulin levels during an oral glucose tolerance test (0.07 SD per Cys-allele, p = 0.04) and a mixed meal test (0.08 log10 µU/ml per Cys-allele, p = 0.003), and had increased lipid oxidation rates post-absorptively and during insulin infusion (0.07 mg [kg estimated metabolic body size {EMBS}]-1  min-1 per Cys-allele for both, p = 0.01 and 0.009, respectively), compared to individuals with the non-risk Arg-allele. In vitro functional studies showed that cells expressing the Cys-allele had a 17.2% decrease in lipolysis under isoproterenol stimulation (p = 0.03) and a 21.3% decrease in lipase enzyme activity measured by using p-nitrophenyl butyrate as a substrate (p = 0.04) compared to the Arg-allele. CONCLUSION: The Arg611Cys variant causes a modest impairment in lipolysis, thereby affecting glucose homoeostasis and risk of T2D.

Weight tracking in childhood and adolescence and type 2 diabetes risk.

AIMS/HYPOTHESIS: The aim of this work was to examine the associations of average weight and weight velocity in three growth periods from birth through adolescence with type 2 diabetes incidence. METHODS: Child participants were selected from a 43 year longitudinal study of American Indians to represent three growth periods: pre-adolescence (birth to ~8 years); early adolescence (~8 to ~13 years); and late adolescence (~13 to ~18 years). Age-, sex- and height-standardised weight z score mean and weight z score velocity (change/year) were computed for each period. Participants were followed for up to 25 years from the end of each growth period until they developed diabetes. Associations of weight z score mean or weight z score velocity with diabetes incidence were determined with sex-, birth date- and maternal diabetes-adjusted Poisson regression models. RESULTS: Among 2100 participants representing the pre-adolescence growth period, 1558 representing the early adolescence period and 1418 representing the late adolescence period, there were 290, 315 and 380 incident diabetes cases, respectively. During the first 10 years of follow-up, the diabetes incidence rate ratio (95% CI) was 1.72 (1.40, 2.11)/SD of log10 weight z score mean in pre-adolescence, 2.09 (1.68, 2.60)/SD of log10 weight z score mean in early adolescence and 1.85 (1.58, 2.17)/SD of log10 weight z score mean in late adolescence. The diabetes incidence rate ratio (95% CI) was 1.79 (1.49, 2.17)/SD of log10 weight z score velocity in pre-adolescence, 1.13 (0.91, 1.41)/SD of log10 weight z score velocity in early adolescence and 1.29 (1.09, 1.51)/SD of log10 weight z score velocity in late adolescence. There were strong correlations in the weight z score means and weak correlations in the weight z score velocities between successive periods. CONCLUSIONS/INTERPRETATION: Higher weight and accelerated weight gain in all growth periods associate with increased type 2 diabetes risk. Importantly, higher weight and greater weight velocity during pre-adolescence jointly associate with the highest type 2 diabetes risk. Graphical abstract.

Assessment of the potential role of natural selection in type 2 diabetes and related traits across human continental ancestry groups: comparison of phenotypic with genotypic divergence.

AIMS/HYPOTHESIS: Prevalence of type 2 diabetes differs among human ancestry groups, and many hypotheses invoke differential natural selection to account for these differences. We sought to assess the potential role of differential natural selection across major continental ancestry groups for diabetes and related traits, by comparison of genetic and phenotypic differences. METHODS: This was a cross-sectional comparison among 734 individuals from an urban sample (none of whom was more closely related to another than third-degree relatives), including 83 African Americans, 523 American Indians and 128 European Americans. Participants were not recruited based on diabetes status or other traits. BMI was calculated, and diabetes was diagnosed by a 75 g oral glucose tolerance test. In those with normal glucose tolerance (n = 434), fasting insulin and 30 min post-load insulin, adjusted for 30 min glucose, were taken as measures of insulin resistance and secretion, respectively. Whole exome sequencing was performed, resulting in 97,388 common (minor allele frequency ≥ 5%) variants; the coancestry coefficient (FST) was calculated across all markers as a measure of genetic divergence among ancestry groups. The phenotypic divergence index (PST) was also calculated from the phenotypic differences and heritability (which was estimated from genetic relatedness calculated empirically across all markers in 761 American Indian participants prior to the exclusion of close relatives). Under evolutionary neutrality, the expectation is PST = FST, while for traits under differential selection PST is expected to be significantly greater than FST. A bootstrap procedure was used to test the hypothesis PST = FST. RESULTS: With adjustment for age and sex, prevalence of type 2 diabetes was 34.0% in American Indians, 12.4% in African Americans and 10.4% in European Americans (p = 2.9 × 10-10 for difference among groups). Mean BMI was 36.3, 33.4 and 33.0 kg/m2, respectively (p = 1.9 × 10-7). Mean fasting insulin was 63.8, 48.4 and 45.2 pmol/l (p = 9.2 × 10-5), while mean 30 min insulin was 559.8, 553.5 and 358.8 pmol/l, respectively (p = 5.7 × 10-8). FST across all markers was 0.130, while PST for liability to diabetes, adjusted for age and sex, was 0.149 (p = 0.35 for difference with FST). PST was 0.094 for BMI (p = 0.54), 0.095 for fasting insulin (p = 0.54) and 0.216 (p = 0.18) for 30 min insulin. For type 2 diabetes and BMI, the maximum divergence between populations was observed between American Indians and European Americans (PST-MAX = 0.22, p = 0.37, and PST-MAX = 0.14, p = 0.61), which suggests that a relatively modest 22% or 14% of the genetic variance, respectively, can potentially be explained by differential selection (assuming the absence of neutral drift). CONCLUSIONS/INTERPRETATION: These analyses suggest that while type 2 diabetes and related traits differ significantly among continental ancestry groups, the differences are consistent with neutral expectations based on heritability and genetic distances. While these analyses do not exclude a modest role for natural selection, they do not support the hypothesis that differential natural selection is necessary to explain the phenotypic differences among these ancestry groups. Graphical abstract.

Identification and functional validation of genetic variants in potential miRNA target sites of established BMI genes.

OBJECTIVE: MicroRNAs (miRNAs) play an important role in posttranscriptional regulation by binding to target sites in the 3'UTR of protein-coding genes. Genetic variation within target sites may potentially disrupt the binding activity of miRNAs, thereby impacting this regulation. In the current study, we investigated whether any established BMI-associated genetic variants potentially function by altering a miRNA target site. METHODS: The genomic positions of all predicted miRNA target site seed regions were identified, and these positions were queried in the T2D Knowledge Portal for variants that associated with BMI in the GIANT UK Biobank. This in silico analysis identified ten target site variants that associated with BMI with a P value ≤ 5 × 10-8. These ten variants mapped to nine genes, FAIM2, CCDC171, ADPGK, ZNF654, MLXIP, NT5C2, SHISA4, SLC25A22, and CTNNB1. RESULTS: In vitro functional analyses showed that five of these target site variants, rs7132908 (FAIM2), rs4963153 (SLC25A22), rs9460 (ADPGK), rs11191548 (NT5C2), and rs3008747 (CCDC171), disrupted the binding activity of miRNAs to their target in an allele-specific manner. CONCLUSION: In conclusion, our study suggests that some established variants for BMI may function by altering miRNA binding to a 3'UTR target site.

Birthweight and early-onset type 2 diabetes in American Indians: differential effects in adolescents and young adults and additive effects of genotype, BMI and maternal diabetes.

AIMS/HYPOTHESIS: The aim of this work was to estimate the impact of birthweight on early-onset (age <40 years) type 2 diabetes. METHODS: A longitudinal study of American Indians, aged ≥5 years, was conducted from 1965 to 2007. Participants who had a recorded birthweight were followed until they developed diabetes or their last examination before the age of 40 years, whichever came first. Age- and sex-adjusted diabetes incidence rates were computed and Poisson regression was used to model the effect of birthweight on diabetes incidence, adjusted for sex, BMI, a type 2 diabetes susceptibility genetic risk score (GRS) and maternal covariates. RESULTS: Among 3039 participants, there were 652 incident diabetes cases over a median follow-up of 14.3 years. Diabetes incidence increased with age and was greater in the lowest and highest quintiles of birthweight. Adjusted for covariates, the effect of birthweight on diabetes varied over time, with a non-linear effect at 10-19 years (p < 0.001) and a negative linear effect at older age intervals (20-29 years, p < 0.001; 30-39 years, p = 0.003). Higher GRS, greater BMI and maternal diabetes had additive but not interactive effects on the association between birthweight and diabetes incidence. CONCLUSIONS/INTERPRETATION: In this high-risk population, both low and high birthweights were associated with increased type 2 diabetes risk in adolescence (age 10-19 years) but only low birthweight was associated with increased risk in young adulthood (20-39 years). Higher type 2 diabetes GRS, greater BMI and maternal diabetes added to the risk of early-onset diabetes.

Association of CREBRF variants with obesity and diabetes in Pacific Islanders from Guam and Saipan.

AIMS/HYPOTHESIS: Variants in CREBRF (rs12513649 and rs373863828) have been strongly associated with increased BMI and decreased risk of type 2 diabetes in Polynesian populations; the A allele at rs373863828 is common in Polynesians but rare in most other global populations. The aim of the present study was to assess the association of CREBRF variants with obesity and diabetes in Pacific Islander (largely Marianas and Micronesian) populations from Guam and Saipan. METHODS: CREBRF rs12513649 and rs373863828 were genotyped in 2022 participants in a community-based cross-sectional study designed to identify determinants of diabetes and end-stage renal disease (ESRD). Associations were analysed with adjustment for age, sex, ESRD and the first four genetic principal components from a genome-wide association study (to account for population stratification); a genomic control procedure was used to account for residual stratification. RESULTS: The G allele at rs12513649 had an overall frequency of 7.7%, which varied from 2.2% to 20.7% across different Marianas and Micronesian populations; overall frequency of the A allele at rs373863828 was 4.2% (range: 1.1-5.4%). The G allele at rs12513649 was associated with higher BMI (ß = 1.55 kg/m2 per copy; p = 0.0026) as was the A allele at rs373863828 (ß = 1.48 kg/m2, p = 0.033). The same alleles were associated with lower risk of diabetes (OR per copy: 0.63 [p = 0.0063] and 0.49 [p = 0.0022], respectively). Meta-analyses combining the current results with previous results in Polynesians showed a strong association between the A allele at rs373863828 and BMI (ß = 1.38 kg/m2; p = 2.5 × 10-29) and diabetes (OR 0.65, p = 1.5 × 10-13). CONCLUSIONS/INTERPRETATION: These results confirm the associations of CREBRF variants with higher BMI and lower risk of diabetes and, importantly, they suggest that these variants contribute to the risk of obesity and diabetes in Oceanic populations.

Analysis of type 2 diabetes and obesity genetic variants in Mexican Pima Indians: Marked allelic differentiation among Amerindians at HLA.

Prevalence of diabetes and obesity in Mexican Pima Indians is low, while prevalence in US Pima Indians is high. Although lifestyle likely accounts for much of the difference, the role of genetic factors is not well explored. To examine this, we genotyped 359 single nucleotide polymorphisms, including established type 2 diabetes and obesity variants from genome-wide association studies (GWAS) and 96 random markers, in 342 Mexican Pimas. A multimarker risk score of obesity variants was associated with body mass index (BMI; ß = 0.81 kg/m2 per SD, P = 0.0066). The mean value of the score was lower in Mexican Pimas than in US Pimas (P = 4.3 × 10-11 ), and differences in allele frequencies at established loci could account for approximately 7% of the population difference in BMI; however, the difference in risk scores was consistent with evolutionary neutrality given genetic distance. To identify loci potentially under recent natural selection, allele frequencies at 283 variants were compared between US and Mexican Pimas, accounting for genetic distance. The largest differences were seen at HLA markers (e.g., rs9271720, difference = 0.75, P = 8.7 × 10-9 ); genetic distances at HLA were greater than at random markers (P = 1.6 × 10-46 ). Analyses of GWAS data in 937 US Pimas also showed sharing of alleles identical by descent at HLA that exceeds its genomic expectation (P = 7.0 × 10-10 ). These results suggest that, in addition to the widely recognized balancing selection at HLA, recent directional selection may also occur, resulting in marked allelic differentiation between closely related populations.

Identification and functional analysis of a novel G310D variant in the insulin-like growth factor 1 receptor (IGF1R) gene associated with type 2 diabetes in American Indians.

AIMS: Insulin-like growth factor 1 receptor (IGF1R) is involved in cell growth and glucose homeostasis. In the current study, the IGF1R locus was analysed as a candidate gene for type 2 diabetes (T2D) in American Indians. MATERIALS AND METHODS: Whole genome sequence data from 335 American Indians identified 3 novel missense variants in IGF1R. The associations of IGF1R variants with T2D, age of T2D onset and birth weight were analysed in a population-based sample of 7701 American Indians. RESULTS: A novel glycine-to-aspartic acid substitution (G310D) in IGF1R was identified, which associated with T2D in a sex-specific manner (Psex interaction = 0.02). In women, the aspartic acid (D) allele (frequency = 0.034) was associated with increased risk for T2D (n = 4292, P = 2.0 × 10-5 adjusted for age, birth year, and the first 5 genetic principal components; odds ratio [OR] = 2.23 [1.54-3.23] per risk allele) and an earlier age of T2D onset (n = 4292, P = 2 × 10-4 , hazard rate ratio = 1.45 [1.20-1.75], Psex interaction = 0.05). Female carriers of the D-allele also had lower birth weight (n = 1313, ß = -163 g, P = .006, Psex interaction = 0.008). Among 85 siblings discordant for G310D, carriers of the D-allele had shorter stature as compared with carriers of the G-allele (ß = -1.6 cm, P = .001, within family model). The G310D variant was functionally studied in vitro, where the D-allele had a 22% increase (P = .0005) in FOXO1-induced transcriptional activity, due to decreased activation of the PI3K/AKT pathway mediated through reduced IGF1R activity. CONCLUSION: A unique G310D variant in IGF1R, which occurs in 6% American Indians, may impair IGF1R signalling pathways, thereby increasing the risk of T2D.

Admixture Mapping Identifies an Amerindian Ancestry Locus Associated with Albuminuria in Hispanics in the United States.

Increased urine albumin excretion is highly prevalent in Hispanics/Latinos. Previous studies have found an association between urine albumin excretion and Amerindian ancestry in Hispanic/Latino populations. Admixture between racial/ethnic groups creates long-range linkage disequilibrium between variants with different allelic frequencies in the founding populations and it can be used to localize genes. Hispanic/Latino genomes are an admixture of European, African, and Amerindian ancestries. We leveraged this admixture to identify associations between urine albumin excretion (urine albumin-to-creatinine ratio [UACR]) and genomic regions harboring variants with highly differentiated allele frequencies among the ancestral populations. Admixture mapping analysis of 12,212 Hispanic Community Health Study/Study of Latinos participants, using a linear mixed model, identified three novel genome-wide significant signals on chromosomes 2, 11, and 16. The admixture mapping signal identified on chromosome 2, spanning q11.2-14.1 and not previously reported for UACR, is driven by a difference between Amerindian ancestry and the other two ancestries (P<5.7 × 10-5). Within this locus, two common variants located at the proapoptotic BCL2L11 gene associated with UACR: rs116907128 (allele frequency =0.14; P=1.5 × 10-7) and rs586283 (C allele frequency =0.35; P=4.2 × 10-7). In a secondary analysis, rs116907128 accounted for most of the admixture mapping signal observed in the region. The rs116907128 variant is common among full-heritage Pima Indians (A allele frequency =0.54) but is monomorphic in the 1000 Genomes European and African populations. In a replication analysis using a sample of full-heritage Pima Indians, rs116907128 significantly associated with UACR (P=0.01; n=1568). Our findings provide evidence for the presence of Amerindian-specific variants influencing the variation of urine albumin excretion in Hispanics/Latinos.

Differential methylation of genes in individuals exposed to maternal diabetes in utero.

AIMS/HYPOTHESIS: Individuals exposed to maternal diabetes in utero are more likely to develop metabolic and cardiovascular diseases later in life. This may be partially attributable to epigenetic regulation of gene expression. We performed an epigenome-wide association study to examine whether differential DNA methylation, a major source of epigenetic regulation, can be observed in offspring of mothers with type 2 diabetes during the pregnancy (OMD) compared with offspring of mothers with no diabetes during the pregnancy (OMND). METHODS: DNA methylation was measured in peripheral blood using the Illumina HumanMethylation450K BeadChip. A total of 423,311 CpG sites were analysed in 388 Pima Indian individuals, mean age at examination was 13.0 years, 187 of whom were OMD and 201 were OMND. Differences in methylation between OMD and OMND were assessed. RESULTS: Forty-eight differentially methylated CpG sites (with an empirical false discovery rate ≤0.05), mapping to 29 genes and ten intergenic regions, were identified. The gene with the strongest evidence was LHX3, in which six CpG sites were hypermethylated in OMD compared with OMND (p ≤ 1.1 × 10-5). Similarly, a CpG near PRDM16 was hypermethylated in OMD (1.1% higher, p = 5.6 × 10-7), where hypermethylation also predicted future diabetes risk (HR 2.12 per SD methylation increase, p = 9.7 × 10-5). Hypermethylation near AK3 and hypomethylation at PCDHGA4 and STC1 were associated with exposure to diabetes in utero (AK3: 2.5% higher, p = 7.8 × 10-6; PCDHGA4: 2.8% lower, p = 3.0 × 10-5; STC1: 2.9% lower, p = 1.6 × 10-5) and decreased insulin secretory function among offspring with normal glucose tolerance (AK3: 0.088 SD lower per SD of methylation increase, p = 0.02; PCDHGA4: 0.08 lower SD per SD of methylation decrease, p = 0.03; STC1: 0.072 SD lower per SD of methylation decrease, p = 0.05). Seventeen CpG sites were also associated with BMI (p ≤ 0.05). Pathway analysis of the genes with at least one differentially methylated CpG (p < 0.005) showed enrichment for three relevant biological pathways. CONCLUSIONS/INTERPRETATION: Intrauterine exposure to diabetes can affect methylation at multiple genomic sites. Methylation status at some of these sites can impair insulin secretion, increase body weight and increase risk of type 2 diabetes.

A cis-eQTL in PFKFB2 is associated with diabetic nephropathy, adiposity and insulin secretion in American Indians.

A prior genome-wide association study (GWAS) in Pima Indians identified a variant within PFKFB2 (rs17258746) associated with body mass index (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 (AI) who had longitudinal data on BMI (n = 6839), type 2 diabetes (T2D; n = 7710), diabetic nephropathy (DN; n = 2452), % body fat (n = 555) and insulin secretion (n = 298). Two SNPs were further genotyped in urban AI 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 (ß = 1.02 per risk allele, P = 7.3 × 10(-4)), maximum BMI z-score in childhood (ß = 0.079, P = 0.03) and % body fat in adulthood (ß = 3.4%, P = 3 × 10(-7)). The adiposity-increasing allele correlated with lower PFKFB2 adipose expression (ß = 0.81, P = 9.4 × 10(-4)). Lower expression of PFKFB2 further correlated with higher % body fat (r = -0.16, P = 0.02) and BMI (r = -0.17, P = 0.02). This allele was also associated with increased risk for DN in both cohorts of AI [odds ratio = 1.64 (1.32-2.02), P = 5.8 × 10(-6)], and similarly correlated with lower PFKFB2 expression in kidney glomeruli (ß = 0.87, P = 0.03). The same allele was also associated with lower insulin secretion assessed by acute insulin response (ß = 0.78, P = 0.03) and 30-min plasma insulin concentrations (ß = 0.78, P = 1.1 × 10(-4)). Variation in PFKFB2 appears to reduce PFKFB2 expression in adipose and kidney tissues, and thereby increase risk for adiposity and DN.

Assessing variation across 8 established East Asian loci for type 2 diabetes mellitus in American Indians: Suggestive evidence for new sex-specific diabetes signals in GLIS3 and ZFAND3.

BACKGROUND: Eight new loci for type 2 diabetes mellitus (T2DM) were identified in an East Asian genome-wide association study meta-analysis. We assess tag SNPs across these loci for associations with T2DM in American Indians. METHODS: A total of 435 SNPs that tag (R2  ≥ .85) common variation across the 8 loci were analyzed for association with T2DM (n = 7710), early onset T2DM (n = 1060), body mass index (n = 6839), insulin sensitivity (n = 555), and insulin secretion (n = 298). RESULTS: Tag SNPs within FITM2-R3HDML-HNF4A, GLIS3, KCNK16, and ZFAND3 associated with T2DM after accounting for locus-wide multiple testing. The T2DM association in FITM2-R3HDML-HNF4A (rs3212183; P = .0002; OR = 1.19 [1.09-1.30]) was independent from the East Asian lead SNP (rs6017317), which did not associate with T2DM in American Indians. The top signals in GLIS3 (rs7875253; P = .0004; OR = 1.23 [1.10-1.38]) and KCNK16 (rs1544050; P = .002; OR = 1.16 [1.06-1.27]) were attenuated after adjustment for the East Asian lead SNPs (rs7041847 in GLIS3; rs1535500 in KCNK16), both of which also associated with T2DM in American Indians (P = .02; OR = 1.11 [1.01-1.21]; P = .007; OR = 1.19 [1.05-1.36] respectively). The top SNP in ZFAND3 (rs9470794; P = .002; OR = 1.43 [1.14-1.80]) was the identical East Asian lead SNP. Additional SNPs in GLIS3 (rs180867004) and ZFAND3 (rs4714120 and rs9470701) had significant genotype × sex interactions (P ≤ .008). The GLIS3 SNP (rs180867004) associated with T2DM only in men (P = .00006, OR = 1.94 [1.40-2.68]). The ZFAND3 SNPs (rs4714120 and rs9470701) associated with T2DM only in women (P = .0002, OR = 1.35 [1.16-1.59]; P = .0003, OR = 1.37 [1.16-1.63] respectively). CONCLUSIONS: Replication of lead T2DM SNPs in GLIS3, KCNK16, and ZFAND3 was observed in American Indians. Sex-specific T2DM signals in GLIS3 and ZFAND3, which are distinct from the East Asian GWAS signals, were also identified.

Assessment of established HDL-C loci for association with HDL-C levels and type 2 diabetes in Pima Indians.

AIMS/HYPOTHESIS: Epidemiological studies in Pima Indians identified elevated levels of HDL-cholesterol (HDL-C) as a protective factor against type 2 diabetes risk in women. We assessed whether HDL-C-associated single-nucleotide polymorphisms (SNPs) also associate with type 2 diabetes in female Pima Indians. METHODS: Twenty-one SNPs in established HDL-C loci were initially analysed in 2,675 full-heritage Pima Indians. SNPs shown to associate with HDL-C (12 SNPs) were assessed for association with type 2 diabetes in 7,710 Pima Indians (55.6% female sex). The CETP locus provided the strongest evidence for association with HDL-C and was further interrogated by analysing tag SNPs. RESULTS: Twelve of the 21 SNPs analysed had a significant association with HDL-C in Pima Indians; five SNPs representing four loci (CETP, DOCK6, PPP1R3B and ABCA1) reached genome-wide significance. Three SNPs, at CETP, KLF14 and HNF4A, associated with type 2 diabetes only in female participants with the HDL-C-lowering allele increasing diabetes risk (p values: 3.2 × 10(-4) to 7.7 × 10(-5)); the association remained significant even after adjustment for HDL-C. Additional analysis across CETP identified rs6499863 as having the strongest association with type 2 diabetes in female participants (p = 5.0 × 10(-6)) and this association remained independent of the HDL-C association. CONCLUSIONS/INTERPRETATION: SNPs at the CETP, HNF4A and KLF14 locus are associated with HDL-C levels and type 2 diabetes (in female participants). However, since HNF4A and KLF14 are established loci for type 2 diabetes, it is unlikely that HDL-C solely mediates these associations.

MAP2K3 is associated with body mass index in American Indians and Caucasians and may mediate hypothalamic inflammation.

To identify genes that affect body mass index (BMI) in American Indians who are predominately of Pima Indian heritage, we previously completed a genome-wide association study in 1120 American Indians. That study also included follow-up genotyping for 9 SNPs in 2133 additional subjects. A comprehensive follow-up study has subsequently been completed where 292 SNPs were genotyped in 3562 subjects, of which 128 SNPs were assessed for replication in 3238 additional subjects. In the combined subjects (n = 6800), BMI associations for two SNPs, rs12882548 and rs11652094, approached genome-wide significance (P = 6.7 × 10(-7) and 8.1 × 10(-7), respectively). Rs12882548 is located in a gene desert on chromosome 14 and rs11652094 maps near MAP2K3. Several SNPs in the MAP2K3 region including rs11652094 were also associated with BMI in Caucasians from the GIANT consortium (P = 10(-2)-10(-5)), and the combined P-values across both American Indians and Caucasian were P = 10(-4)-10(-9). Follow-up sequencing across MAP2K3 identified several paralogous sequence variants indicating that the region may have been duplicated. MAP2K3 expression levels in adipose tissue biopsies were positively correlated with BMI, although it is unclear if this correlation is a cause or effect. In vitro studies with cloned MAP2K3 promoters suggest that MAP2K3 expression may be up-regulated during adipogenesis. Microarray analyses of mouse hypothalamus cells expressing constitutively active MAP2K3 identified several up-regulated genes involved in immune/inflammatory pathways and a gene, Hap1, thought to play a role in appetite regulation. We conclude that MAP2K3 is a reproducible obesity locus that may affect body weight via complex mechanisms involving appetite regulation and hypothalamic inflammation.

Common genetic variation in the glucokinase gene (GCK) is associated with type 2 diabetes and rates of carbohydrate oxidation and energy expenditure.

AIMS/HYPOTHESIS: Glucokinase (GCK) plays a role in glucose metabolism and glucose-stimulated insulin secretion. Rare mutations in GCK cause MODY. We investigated whether common variation (minor allele frequency ≥0.01) in GCK is associated with metabolic traits and type 2 diabetes. METHODS: Four exonic single-nucleotide polymorphisms (SNPs) and three SNPs predicted to cause loss of promoter function were identified in whole-genome sequence data from 234 Pima Indians. These seven tag SNPs and rs4607517, a type 2 diabetes variant established in other studies, were analysed in 415 full-heritage non-diabetic Pima Indians characterised for metabolic traits, and 7,667 American Indians who had data on type 2 diabetes and BMI. RESULTS: A novel 3' untranslated region (3'UTR) SNP, chr7:44184184-G/A, was associated with the rate of carbohydrate oxidation post-absorptively (ß = 0.22 mg [kg estimated metabolic body size (EMBS)](-1) min(-1), p = 0.005) and during a hyperinsulinaemic-euglycaemic clamp (ß = 0.24 mg [kg EMBS](-1) min(-1), p = 0.0002), the rate of carbohydrate oxidation in a respiratory chamber (ß = 311 kJ/day, p = 0.03) and 24 h energy expenditure, which was attributable to the thermic effect of food (ß = 520 kJ/day, p = 3.39 × 10(-6)). This 3'UTR SNP was also associated with diabetes (OR 1.36, 95% CI 1.11, 1.65, p = 0.002), where the A allele (allele frequency 0.05) was associated with a lower rate of carbohydrate oxidation, lower 24 h energy expenditure and higher risk for diabetes. In a Cox proportional hazards model, a rate of insulin-stimulated carbohydrate oxidation lower than the mean rate at baseline predicted a higher risk for developing diabetes than for those above the mean (hazard rate ratio 2.2, 95% CI 1.3, 3.6, p = 0.002). CONCLUSIONS/INTERPRETATION: Common variation in GCK influences the rate of carbohydrate oxidation, 24 h energy expenditure and diabetes risk in Pima Indians.