Publications

Genetic studies have identified thousands of variants associated with complex traits. However, most association studies are limited to populations of European descent and a single phenotype. The Population Architecture using Genomics and Epidemiology (PAGE) Study was initiated in 2008 by the National Human Genome Research Institute to investigate the epidemiologic architecture of well-replicated genetic variants associated with complex diseases in several large, ethnically diverse population-based studies. Combining DNA samples and hundreds of phenotypes from multiple cohorts, PAGE is well-suited to address generalization of associations and variability of effects in diverse populations; identify genetic and environmental modifiers; evaluate disease subtypes, intermediate phenotypes, and biomarkers; and investigate associations with novel phenotypes. PAGE investigators harmonize phenotypes across studies where possible and perform coordinated cohort-specific analyses and meta-analyses. PAGE researchers are genotyping thousands of genetic variants in up to 121,000 DNA samples from African-American, white, Hispanic/Latino, Asian/Pacific Islander, and American Indian participants. Initial analyses will focus on single nucleotide polymorphisms (SNPs) associated with obesity, lipids, cardiovascular disease, type 2 diabetes, inflammation, various cancers, and related biomarkers. PAGE SNPs are also assessed for pleiotropy using the "phenome-wide association study" approach, testing each SNP for associations with hundreds of phenotypes. PAGE data will be deposited into the National Center for Biotechnology Information's Database of Genotypes and Phenotypes and made available via a custom browser.

OBJECTIVES: The thickness of the brain's cortical gray matter (GM) and the fractional anisotropy (FA) of the cerebral white matter (WM) each follow an inverted U-shape trajectory with age. The two measures are positively correlated and may be modulated by common biological mechanisms. We employed four types of genetic analyses to localize individual genes acting pleiotropically upon these phenotypes.

METHODS: Whole-brain and regional GM thickness and FA values were measured from high-resolution anatomical and diffusion tensor MR images collected from 712, Mexican American participants (438 females, age = 47.9 ± 13.2 years) recruited from 73 (9.7 ± 9.3 individuals/family) large families. The significance of the correlation between two traits was estimated using a bivariate genetic correlation analysis. Localization of chromosomal regions that jointly influenced both traits was performed using whole-genome quantitative trait loci (QTL) analysis. Gene localization was performed using SNP genotyping on Illumina 1M chip and correlation with leukocyte-based gene-expression analyses. The gene-expressions were measured using the Illumina BeadChip. These data were available for 371 subjects.

RESULTS: Significant genetic correlation was observed among GM thickness and FA values. Significant logarithm of odds (LOD ≥ 3.0) QTLs were localized within chromosome 15q22-23. More detailed localization reported no significant association (p < 5·10(-5)) for 1565 SNPs located within the QTLs. Post hoc analysis indicated that 40% of the potentially significant (p ≤ 10(-3)) SNPs were localized to the related orphan receptor alpha (RORA) and NARG2 genes. A potentially significant association was observed for the rs2456930 polymorphism reported as a significant GWAS finding in Alzheimer's disease neuroimaging initiative subjects. The expression levels for RORA and ADAM10 genes were significantly (p < 0.05) correlated with both FA and GM thickness. NARG2 expressions were significantly correlated with GM thickness (p < 0.05) but failed to show a significant correlation (p = 0.09) with FA.

DISCUSSION: This study identified a novel, significant QTL at 15q22-23. SNP correlation with gene-expression analyses indicated that RORA, NARG2, and ADAM10 jointly influence GM thickness and WM-FA values.

BACKGROUND: Genome-wide association studies identified several single nucleotide polymorphisms (SNP) associated with prevalent coronary heart disease (CHD), but less is known of associations with incident CHD. The association of 13 published CHD SNPs was examined in 5 ancestry groups of 4 large US prospective cohorts.

METHODS AND RESULTS: The analyses included incident coronary events over an average 9.1 to 15.7 follow-up person-years in up to 26 617 white individuals (6626 events), 8018 black individuals (914 events), 1903 Hispanic individuals (113 events), 3669 American Indian individuals (595 events), and 885 Asian/Pacific Islander individuals (66 events). We used Cox proportional hazards models (with additive mode of inheritance) adjusted for age, sex, and ancestry (as needed). Nine loci were statistically associated with incident CHD events in white participants: 9p21 (rs10757278; P=4.7 × 10(-41)), 16q23.1 (rs2549513; P=0.0004), 6p24.1 (rs499818; P=0.0002), 2q36.3 (rs2943634; P=6.7 × 10(-6)), MTHFD1L (rs6922269, P=5.1 × 10(-10)), APOE (rs429358; P=2.7×10(-18)), ZNF627 (rs4804611; P=5.0 × 10(-8)), CXCL12 (rs501120; P=1.4 × 10(-6)) and LPL (rs268; P=2.7 × 10(-17)). The 9p21 region showed significant between-study heterogeneity, with larger effects in individuals age 55 years or younger and in women. Inclusion of coronary revascularization procedures among the incident CHD events introduced heterogeneity. The SNPs were not associated with CHD in black participants, and associations varied in other US minorities.

CONCLUSIONS: Prospective analyses of white participants replicated several reported cross-sectional CHD-SNP associations.

Although disrupted in schizophrenia 1 (DISC1) has been implicated in many psychiatric disorders, including schizophrenia, bipolar disorder, schizoaffective disorder and major depression, its biological role in these disorders is unclear. To better understand this gene and its role in psychiatric disease, we conducted transcriptional profiling and genome-wide association analysis in 1232 pedigreed Mexican-American individuals for whom we have neuroanatomic images, neurocognitive assessments and neuropsychiatric diagnoses. SOLAR was used to determine heritability, identify gene expression patterns and perform association analyses on 188 quantitative brain-related phenotypes. We found that the DISC1 transcript is highly heritable (h(2)=0.50; P=1.97 × 10(-22)), and that gene expression is strongly cis-regulated (cis-LOD=3.89) but is also influenced by trans-effects. We identified several DISC1 polymorphisms that were associated with cortical gray matter thickness within the parietal, temporal and frontal lobes. Associated regions affiliated with memory included the entorhinal cortex (rs821639, P=4.11 × 10(-5); rs2356606, P=4.71 × 10(-4)), cingulate cortex (rs16856322, P=2.88 × 10(-4)) and parahippocampal gyrus (rs821639, P=4.95 × 10(-4)); those affiliated with executive and other cognitive processing included the transverse temporal gyrus (rs9661837, P=5.21 × 10(-4); rs17773946, P=6.23 × 10(-4)), anterior cingulate cortex (rs2487453, P=4.79 × 10(-4); rs3738401, P=5.43 × 10(-4)) and medial orbitofrontal cortex (rs9661837; P=7.40 × 10(-4)). Cognitive measures of working memory (rs2793094, P=3.38 × 10(-4)), as well as lifetime history of depression (rs4658966, P=4.33 × 10(-4); rs12137417, P=4.93 × 10(-4)) and panic (rs12137417, P=7.41 × 10(-4)) were associated with DISC1 sequence variation. DISC1 has well-defined genetic regulation and clearly influences important phenotypes related to psychiatric disease.

Genes play an important role in lifelong skeletal health. Genes that influence bone building during childhood have the potential to affect bone health not only throughout childhood but also into adulthood. Given that peak bone mass is a significant predictor of adult fracture risk, it is imperative that the genetic underpinnings of the normal pediatric skeleton are uncovered. In a sample of 600 10-year-old children from 144 families in the Fels Longitudinal Study, we examined radiographic cortical bone measures of the second metacarpal. Morphometic measurements included bone width, medial and lateral cortical thicknesses, and the calculated cortical index representing the amount of cortex relative to bone width. We then conducted genome-wide linkage analysis on these traits in 440 genotyped individuals using the SOLAR analytic platform. Significant quantitative trait loci (QTL) were identified for bone traits on three separate chromosomes. A QTL for medial cortical thickness was localized to chromosome 2p25.2. A QTL for lateral cortical thickness was localized to chromosomal region 3p26.1-3p25.3. Finally, a QTL detected for cortical index was localized to the 17q21.2 chromosomal region. Each region contains plausible candidate genes for pediatric skeletal health, some of which confirm findings from studies of adulthood bone, and for others represent novel candidate genes for skeletal health.

OBJECTIVE: Genome-wide association studies that compare the statistical association between thousands of DNA variations and a human trait have detected 958 loci across 127 different diseases and traits. However, these statistical associations only provide evidence for genomic regions likely to harbor a causal gene(s) and do not directly identify such genes. We combined gene variation and expression data in a human cohort to identify causal genes.

RESEARCH DESIGN AND METHODS: Global gene transcription activity was obtained for each individual in a large human cohort (n = 1,240). These quantitative transcript data were tested for correlation with genotype data generated from the same individuals to identify gene expression patterns influenced by the variants.

RESULTS: Variant rs8050136 lies within intron 1 of the FTO gene on chromosome 16 and marks a locus strongly associated with type 2 diabetes and obesity and widely replicated across many populations. We report that genetic variation at this locus does not influence FTO gene expression levels (P = 0.38), but is strongly correlated with expression of RBL2 (P = 2.7 x 10(-5)), approximately 270,000 base pairs distant to FTO.

CONCLUSIONS: These data suggest that variants at FTO influence RBL2 gene expression at large genetic distances. This observation underscores the complexity of human transcriptional regulation and highlights the utility of large human cohorts in which both genetic variation and global gene expression data are available to identify disease genes. Expedient identification of genes mediating the effects of genome-wide association study-identified loci will enable mechanism-of-action studies and accelerate understanding of human disease processes under genetic influence.

Pregnancy is a complex physiological condition, and the growth hormone (GH)-related hormones produced in the placenta, which emerged during the evolution of primates, are thought to play an important metabolic role in pregnancy that is not yet fully understood. The aim of this study was to identify the genes and transcription products of the GH family in baboon (Papio hamadryas) and to assess these in relation to the evolution of this gene family. GH-related transcripts were amplified using total RNA from placental tissue, by reverse transcription coupled to polymerase chain reaction (RT-PCR). Three different GH-related transcripts were identified in baboon placental tissue, with two encoding chorionic somatomammotropins (CSH) and one the placental variant of GH (GH-2). The CSH transcripts showed some minor allelic variation, and a splice variant of CSH-C that retains its in-frame third intron. Gene sequences for GH-1 (probably representing the GH gene expressed primarily in the pituitary gland), GH-2 and the two CSHs were identified in the baboon genomic database, together with a CSH-related pseudogene. Phylogenetic analysis of the baboon GH-related sequences, together with those of a related Old World monkey, macaque, and ape outgroup (human), showed the equivalence of the genes in baboon and macaque, and revealed evidence for several episodes of rapid adaptive evolution. Many of the substitutions seen during the evolution of these placental proteins have occurred in the receptor-binding sites, especially site 2, contrasting with the strong conservation of the hydrophobic core.

BACKGROUND: Melanocortin-4-receptor (MC4R) haploinsufficiency is the most common form of monogenic obesity; however, the frequency of MC4R variants and their functional effects in general populations remain uncertain.

OBJECTIVE: The aim was to identify and characterize the effects of MC4R variants in Hispanic children.

DESIGN: MC4R was resequenced in 376 parents, and the identified single nucleotide polymorphisms (SNPs) were genotyped in 613 parents and 1016 children from the Viva la Familia cohort. Measured genotype analysis (MGA) tested associations between SNPs and phenotypes. Bayesian quantitative trait nucleotide (BQTN) analysis was used to infer the most likely functional polymorphisms influencing obesity-related traits.

RESULTS: Seven rare SNPs in coding and 18 SNPs in flanking regions of MC4R were identified. MGA showed suggestive associations between MC4R variants and body size, adiposity, glucose, insulin, leptin, ghrelin, energy expenditure, physical activity, and food intake. BQTN analysis identified SNP 1704 in a predicted micro-RNA target sequence in the downstream flanking region of MC4R as a strong, probable functional variant influencing total, sedentary, and moderate activities with posterior probabilities of 1.0. SNP 2132 was identified as a variant with a high probability (1.0) of exerting a functional effect on total energy expenditure and sleeping metabolic rate. SNP rs34114122 was selected as having likely functional effects on the appetite hormone ghrelin, with a posterior probability of 0.81.

CONCLUSION: This comprehensive investigation provides strong evidence that MC4R genetic variants are likely to play a functional role in the regulation of weight, not only through energy intake but through energy expenditure.

Chronic kidney disease (CKD) is an important public health problem in American Indian populations. Recent research has identified associations of polymorphisms in the myosin heavy chain type II isoform A (MYH9) gene with hypertensive CKD in African-Americans. Whether these associations are also present among American Indian individuals is unknown. To evaluate the role of genetic polymorphisms in the MYH9 gene on kidney disease in American Indians, we genotyped 25 SNPs in the MYH9 gene region in 1,119 comparatively unrelated individuals. Four SNPs failed, and one SNP was monomorphic We inferred haplotypes using seven SNPs within the region of the previously described E haplotype using Phase v2.1. We studied the association between 20 MYH9 SNPs with kidney function (estimated glomerular filtration rate, eGFR) and CKD (eGFR\60 ml/min/1.73 m(2) or renal replacement therapy or kidney transplant) using age-, sex- and center-adjusted models and measured genotyped within the variance component models. MYH9 SNPs were not significantly associated with kidney traits in additive or recessive genetic adjusted models. MYH9 haplotypes were also not significantly associated with kidney outcomes. In conclusion, common variants in MYH9 polymorphisms may not confer an increased risk of CKD in American Indian populations. Identification of the actual functional genetic variation responsible for the associations seen in African-Americans will likely help to clarify the lack of replication of this gene in our population of American Indians.

CONTEXT: Chemerin is a new adipokine associated with obesity and the metabolic syndrome. Gene expression levels of chemerin were elevated in the adipose depots of obese compared with lean animals and was markedly elevated during differentiation of fibroblasts into mature adipocytes.

OBJECTIVE: The objective of the study was to identify factors that affect the regulation and potential function of chemerin using a genetics approach.

DESIGN, SETTING, PATIENTS, AND INTERVENTION: Plasma chemerin levels were measured in subjects from the San Antonio Family Heart Study, a large family-based genetic epidemiological study including 1354 Mexican-American individuals. Individuals were randomly sampled without regard to phenotype or disease status.

MAIN OUTCOME MEASURES: A genome-wide association analysis using 542,944 single-nucleotide polymorphisms in a subset of 523 of the same subjects was undertaken. The effect of chemerin on angiogenesis was measured using human endothelial cells and interstitial cells in coculture in a specially formulated medium.

RESULTS: Serum chemerin levels were found to be highly heritable (h(2) = 0.25; P = 1.4 x 10(-9)). The single-nucleotide polymorphism showing strongest evidence of association (rs347344; P = 1.4 x 10(-6)) was located within the gene encoding epithelial growth factor-like repeats and discoidin I-like domains 3, which has a known role in angiogenesis. Functional angiogenesis assays in human endothelial cells confirmed that chemerin significantly mediated the formation of blood vessels to a similar extent as vascular endothelial growth factor.

CONCLUSION: Here we demonstrate for the first time that plasma chemerin levels are significantly heritable and identified a novel role for chemerin as a stimulator of angiogenesis.