Phase 1: genetic screening. The EPR Consortium Project is a large translational research project initiated at the NIEHS by a multidisciplinary team of basic scientists, geneticists, toxicologists, clinicians, and biostatisticians (the EPR Consortium). Consortium members have selected 87 genes for study that, based on cell culture, animal, and/or human studies, are candidate genes for asthma, atherosclerosis, cancer, autoimmune disease, aging, and other conditions. Most are environmental response genes that work in concert with environmental exposures to elicit a phenotype. Examples include cytochromes P450 (CYP2J2, CYP2C8, CYP2C9), which are involved in xenobiotic or drug metabolism, and AhR (aryl hydrocarbon receptor), ARNT (AhR nuclear translocator), and AhRR (AhR repressor), which mediate the effects of polycyclic aromatic hydrocarbons and other endocrine- and immune-disrupting xenobiotics.
From the 87 genes, we identified 717 SNPs that were predicted to alter protein sequence and/or function, are in evolutionarily conserved regulatory regions, or tag European and African ancestral haplotypes. About 70% of these SNPs can be found in dbSNP [National Center for Biotechnology Information (NCBI) 2010b] and HapMap databases (NCBI 2010a) and were selected using SNPselector (Xu et al. 2005
) or based on a priori
functional significance. The other 30% of SNPs are novel and potentially significant based on research conducted by individual consortium members. An additional 51 sex and ancestral informative markers (AIMs) are being genotyped and will be compared with self-reported sex and race for quality control. The AIMs also will be used to measure admixture in the population and to adjust for population stratification.
In this first round of genotyping, 4,000 subjects are being screened using custom high-throughput 384-plex Illumina arrays (Illumina, Inc., San Diego, CA). The 4,000 consists of approximately 500 subjects of Hispanic or Latino ethnicity and equal numbers of males and females of African and European ancestry. Important aims of phase 1 genotyping are to assess genotype frequency in the EPR population stratified by race, ethnicity, and sex, inform phase 2 study design, and identify appropriate subjects for follow-up studies. We expect that most (but not all) genotyping studies will lead to follow-up phenotyping studies, and in these situations we screen only active subjects who can be readily recontacted. Nonactive EPR subjects can be used in genotyping projects where follow-up is not important, for example, in simple assessments of SNP prevalence rates or in approximation of haplotypes using statistical methods (Stephens et al. 2001
Phase 2: phenotyping. Based on phase 1 results, subgroups of EPR subjects with shared genotypes are invited to participate in various phase 2 studies. The studies vary in hypotheses, and their design depends on minor allele frequencies, population stratification, and gene penetrance. Cell phenotyping studies have been the most common type of follow-up study proposed to date, as we expected in early-phase EPR research. Subjects with genotypes of interest are invited to donate viable tissue for basic laboratory experiments aimed at characterizing some molecular or functional attribute of the genotype. Here statistical power depends on the allele frequency and magnitude of the biochemical or molecular effect, and small numbers of subjects are usually adequate. Higher levels of follow-up studies have been proposed and include observational or interventional clinical trials, epidemiological surveys, and cohort studies of disease risk. Four follow-up studies are described below to illustrate the usefulness of the EPR in translational research. The first three have been approved by the NIEHS IRB and are under way; the fourth is under review.
As described above, cell phenotyping studies are a common use of EPR resources. In the first example, we screened subjects for SNPs in p53 response elements of p53
downstream genes (FLT1
). During a follow-up study, these subjects were asked to donate blood for viable lymphocytes to test the potential of the SNPs to alter cell function, p53 promoter occupancy, and transactivation of downstream genes by p53 tumor protein (Bond et al. 2004
; Menendez et al. 2006
; Murphy 2006
; Tomso et al. 2005
). We treated the cells ex vivo
to induce p53-mediated stress and DNA damage and examined them for gene expression by microarray technology.
In the second example, viable mononuclear cells were isolated from the blood of subjects with SNPs in ApoE
, and other genes that regulate cholesterol trafficking and immune response (Mahley and Rall 2000
; Singaraja et al. 2003
); these cells were then used to test the potential of SNPs to alter inflammatory response following ex vivo
bacterial lipopolysaccharide challenge. In both the first and second examples, we also recruited appropriate genetic controls from the EPR and matched these subjects for sex, race, and ethnicity to subjects with the minor alleles.
The EPR can provide adequate numbers of subjects for highly powered cell phenotyping studies such as those described above. To test the null hypothesis (no differences between genotypes) using t-tests, in the first example 10 subjects were needed with each genotype to detect a 1.4-fold change in gene expression (90% power). In the second example, 9 subjects with each genotype were needed to detect a 1.3-fold change in cytokine (tumor necrosis factor-α) levels (85% power). These calculations assume a significance level of 0.05.
In the third example, we used EPR resources at different levels of research, starting with basic cell phenotyping experiments that led to comprehensive clinical observational studies. First, we isolated viable mononuclear cells from EPR subjects harboring potentially functional SNPs in hGR
(human glucocorticoid receptor) (Jewell and Cidlowski 2007
; Schaaf and Cidlowski 2002
). The cells were exposed ex vivo
to glucocorticoids and examined for immune function and gene expression. At the next level, subjects with impaired cellular immune function underwent modified dexamethasone suppression tests to examine the potential of SNPs to alter steroid responsiveness. We also examined subjects for risk factors (body mass index, hip:waist ratio, and blood levels of cortisol, lipids, glucose, insulin, and other metabolites), family history with emphasis on inflammatory and metabolic disease (Manenschijn et al. 2009
), and stress. The goals were to examine how cells, organs, and humans respond to physiological and environmental stressors and how polymorphisms in hGR
affect those responses.
Glucocorticoids that elicit responses through hGR
regulate numerous homeostatic functions (glucose homeostasis, protein and lipid metabolism, skeletal growth, connective tissue metabolism, respiratory function, immune surveillance, and behavior) (Ren and Cidlowski 2005
). Glucocorticoids are also among the most prescribed drugs in the world and are a primary treatment for inflammatory and immune disease (asthma, arthritis, inflammatory bowel disease). Chronic elevation of glucocorticoids from prolonged stress and/or chronic therapeutic administration can have detrimental effects on human health. Therefore, by identifying significant hGR
polymorphisms and understanding how they affect glucocorticoid responsiveness, we can identify populations at risk for these conditions and/or predict how others might respond to glucocorticoid therapy. Next levels in this line of research might include a cohort study or personalized medicine trial, both possible using EPR resources.
The fourth example is a clinical toxicity study designed to examine gene × environment interactions. EPR subjects with functional SNPs in CD44
(receptor for hyaluronic acid) and I
(inter-α-inhibitor) will be exposed to ozone via inhalation and examined for bronchoconstrictive responses to inhaled methacholine. In addition, associations between the SNPs and inflammatory and immune markers will be examined in peripheral white blood cells and alveolar macrophages collected from the same subjects. In animals and humans, both genes have been shown to have roles in ozone-induced airway hyperreactivity (Garantziotis et al. 2009
) and/or inflammatory responses in alveolar macrophages (McKee et al. 1996
). The ultimate goal is to identify populations at risk for lung inflammation after ozone exposures.