Exome Genotyping Identifies Pleiotropic Variants Associated with Red Blood Cell Traits

ARTICLE Exome Genotyping Identifies Pleiotropic Variants Associated with Red Blood Cell Traits Nathalie Chami,1,2,91 Ming-Huei Chen,3,91 Andrew J. Slater,4,5,91 John D. Eicher,3 Evangelos Evangelou,6,7 Salman M. Tajuddin,8 Latisha Love-Gregory,9 Tim Kacprowski,10,11 Ursula M. Schick,12 Akihiro Nomura,13,14,15,16,17 Ayush Giri,18 Samuel Lessard,1,2 Jennifer A. Brody,19 Claudia Schurmann,12,20 Nathan Pankratz,21 Lisa R. Yanek,22 Ani Manichaikul,23 Raha Pazoki,24 Evelin Mihailov,25 W. David Hill,26,27 Laura M. Raffield,28 Amber Burt,29 Traci M. Bartz,30 Diane M. Becker,22 Lewis C. Becker,31 Eric Boerwinkle,32,33 Jette Bork-Jensen,34 Erwin P. Bottinger,12 Michelle L. O’Donoghue,35 David R. Crosslin,36 Simon de Denus,2,37 Marie-Pierre Dubé,1,2 Paul Elliott,6 Gunnar Engström,38,39 Michele K. Evans,8 James S. Floyd,19 Myriam Fornage,40 He Gao,6 Andreas Greinacher,41 Vilmundur Gudnason,42,43 Torben Hansen,34 Tamara B. Harris,44 Caroline Hayward,45 Jussi Hernesniemi,46,47,48 Heather M. Highland,32,49 (Author list continued on next page) Red blood cell (RBC) traits are important heritable clinical biomarkers and modifiers of disease severity. To identify coding genetic variants associated with these traits, we conducted meta-analyses of seven RBC phenotypes in 130,273 multi-ethnic individuals from studies genotyped on an exome array. After conditional analyses and replication in 27,480 independent individuals, we identified 16 new RBC variants. We found low-frequency missense variants in MAP1A (rs55707100, minor allele frequency [MAF] ¼ 3.3%, p ¼ 2 3 10 10 for hemoglobin [HGB]) and HNF4A (rs1800961, MAF ¼ 2.4%, p < 3 3 10 8 for hematocrit [HCT] and HGB). In African Americans, we identified a nonsense variant in CD36 associated with higher RBC distribution width (rs3211938, MAF ¼ 8.7%, p ¼ 7 3 10 11) and showed that it is associated with lower CD36 expression and strong allelic imbalance in ex vivo differentiated human erythroblasts. We also identified a rare missense variant in ALAS2 (rs201062903, MAF ¼ 0.2%) associated with lower mean corpuscular volume and mean corpuscular hemoglobin (p < 8 3 10 9). Mendelian mutations in ALAS2 are a cause of sideroblastic anemia and erythropoietic protoporphyria. Gene-based testing highlighted three rare missense variants in PKLR, a gene mutated in Mendelian non-spherocytic hemolytic anemia, associated with HGB and HCT (SKAT p < 8 3 10 7). These rare, low-frequency, and common RBC variants showed pleiotropy, being also associated with platelet, white blood cell, and lipid traits. Our association results and functional annotation suggest the involvement of new genes in human erythropoiesis. We also confirm that rare and low-frequency variants play a role in the architecture of complex human traits, although their phenotypic effect is generally smaller than originally anticipated. Introduction One in four cells in the human body is a mature enucleated red blood cell (RBC), also called an erythrocyte. RBC mean lifespan in adults is 100–120 days, requiring constant renewal. To that end, we produce on average 2.4 million RBCs per second in the bone marrow. This massive yet well-orchestrated cell proliferation process is necessary to 1 Department of Medicine, Université de Montréal, Montréal, QC H3T 1J4, Canada; 2Montreal Heart Institute, Montréal, QC H1T 1C8, Canada; 3Population Sciences Branch, National Heart, Lung, and Blood Institute, The Framingham Heart Study, Framingham, MA 01702, USA; 4Genetics Target Sciences, GlaxoSmithKline, Research Triangle Park, NC 27709, USA; 5OmicSoft Corporation, Cary, NC 27513, USA; 6Department of Epidemiology and Biostatistics, MRC-PHE Centre for Environment and Health, School of Public Health, Imperial College London, London W2 1PG, UK; 7Department of Hygiene and Epidemiology, University of Ioannina Medical School, Ioannina 45110, Greece; 8Laboratory of Epidemiology and Population Sciences, National Institute on Aging, NIH, Baltimore, MD 21224, USA; 9Department of Medicine, Center of Human Nutrition, Washington University School of Medicine, St Louis, MO 63110, USA; 10Department of Functional Genomics, Interfaculty Institute for Genetics and Functional Genomics, University Medicine, Greifswald and Ernst-Mortiz-Arndt University Greifswald, Greifswald 17475, Germany; 11DZHK (German Centre for Cardiovascular Research), partner site Greifswald, Greifswald QA, Germany; 12The Charles Bronfman Institute for Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10069, USA; 13Center for Human Genetic Research, Massachusetts General Hospital, Boston, MA 02114, USA; 14Program in Medical and Population Genetics, Broad Institute, Cambridge, MA 02142, USA; 15Cardiovascular Research Center, Massachusetts General Hospital, Boston, MA 02114, USA; 16Department of Medicine, Harvard Medical School, Boston, MA 02115, USA; 17Division of Cardiovascular Medicine, Kanazawa University, Graduate School of Medical Science, Kanazawa, Ishikawa 9200942, Japan; 18Division of Epidemiology, Department of Medicine, Institute for Medicine and Public Health, Vanderbilt Genetics Institute, Vanderbilt University, Nashville, TN 37235, USA; 19Department of Medicine, University of Washington, Seattle, WA 98101, USA; 20 The Genetics of Obesity and Related Metabolic Traits Program, Icahn School of Medicine at Mount Sinai, New York, NY 10069, USA; 21Department of Laboratory Medicine and Pathology, University of Minnesota, Minneapolis, MN 55454, USA; 22Department of Medicine/Division of General Internal Medicine, Johns Hopkins University, School of Medicine, Baltimore, MD 21205, USA; 23Center for Public Health Genomics, University of Virginia, Charlottesville, VA 22908, USA; 24Department of Epidemiology, Erasmus, MC Rotterdam 3000, the Netherlands; 25Estonian Genome Center, University of Tartu, Tartu 51010, Estonia; 26Centre for Cognitive Ageing and Cognitive Epidemiology, University of Edinburgh, Edinburgh EH8 9JZ, UK; 27Department of Psychology, University of Edinburgh, Edinburgh EH8 9JZ, UK; 28Department of Genetics, University of North Carolina, Chapel Hill, NC 27514, USA; 29Division of Medical Genetics, Department of Medicine, University of Washington, Seattle, WA 98195, USA; 30Department of Biostatistics, University of Washington, (Affiliations continued on next page) Ó 2016 American Society of Human Genetics. 8 The American Journal of Human Genetics 99, 8–21, July 7, 2016 Joel N. Hirschhorn,14,50 Albert Hofman,24,51 Marguerite R. Irvin,52 Mika Kähönen,53,54 Ethan Lange,55 Lenore J. Launer,44 Terho Lehtimäki,46,47 Jin Li,56 David C.M. Liewald,26,27 Allan Linneberg,57,58,59 Yongmei Liu,60 Yingchang Lu,12,20 Leo-Pekka Lyytikäinen,46,47 Reedik Mägi,25 Rasika A. Mathias,61 Olle Melander,38,39 Andres Metspalu,25 Nina Mononen,46,47 Mike A. Nalls,62 Deborah A. Nickerson,63 Kjell Nikus,48,64 Chris J. O’Donnell,3,65 Marju Orho-Melander,38,39 Oluf Pedersen,34 Astrid Petersmann,66 Linda Polfus,32 Bruce M. Psaty,67,68 Olli T. Raitakari,69,70 Emma Raitoharju,46,47 Melissa Richard,40 Kenneth M. Rice,30 Fernando Rivadeneira,24,71,72 Jerome I. Rotter,73,74 Frank Schmidt,10 Albert Vernon Smith,42,43 John M. Starr,26,75 Kent D. Taylor,73,74 Alexander Teumer,76 Betina H. Thuesen,57 Eric S. Torstenson,18 Russell P. Tracy,77 Ioanna Tzoulaki,6,7 Neil A. Zakai,78 Caterina Vacchi-Suzzi,79 Cornelia M. van Duijn,24 Frank J.A. van Rooij,24 Mary Cushman,78 Ian J. Deary,26,27 Digna R. Velez Edwards,80 Anne-Claire Vergnaud,6 Lars Wallentin,81 Dawn M. Waterworth,82 Harvey D. White,83 James G. Wilson,84 Alan B. Zonderman,8 Sekar Kathiresan,13,14,15,16 Niels Grarup,34 Tõnu Esko,14,25 Ruth J.F. Loos,12,20,85 Leslie A. Lange,28 Nauder Faraday,86 Nada A. Abumrad,9 Todd L. Edwards,18 Santhi K. Ganesh,87,91 Paul L. Auer,88,91 Andrew D. Johnson,3,91 Alexander P. Reiner,89,90,91,* and Guillaume Lettre1,2,91,* accommodate RBCs’ main function: to transport oxygen from the lungs to the peripheral organs, and carbon dioxide from the organs to the lungs. Hemoglobin (HGB), the metalloprotein that constitutes by far the most abundant biomolecule found in mature RBCs, is responsible for oxygen transport. In addition to their critical role in the circulatory system, RBCs also have secondary, often lessappreciated, functions. Within blood vessels, they respond Seattle, WA 98195, USA; 31Department of Medicine/Divisions of Cardiology and General Internal Medicine, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA; 32Human Genetics Center, School of Public Health, University of Texas Health Science Center at Houston, Houston, TX 77030, USA; 33Human Genome Sequencing Center, Baylor College of Medicine, Houston, TX 77030, USA; 34The Novo Nordisk Foundation, Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen 2100, Denmark; 35TIMI Study Group, Cardiovascular Division, Brigham and Women’s Hospital, Boston, MA 02115, USA; 36Department of Biomedical Informatics and Medical Education, University of Washington, Seattle, WA 98195, USA; 37Faculty of Pharmacy, Université de Montréal, Montréal, QC H3T 1J4, Canada; 38Department of Clinical Sciences, Malmö, Lund University, Malmö 221 00, Sweden; 39Skåne University Hospital, Malmö 222 41, Sweden; 40Institute of Molecular Medicine, The University of Texas Health Science Center at Houston, Houston, TX 77030, USA; 41Institute for Immunology and Transfusion Medicine, University Medicine Greifswald, Greifswald 17475, Germany; 42Icelandic Heart Association, 201 Kopavogur, Iceland; 43Faculty of Medicine, University of Iceland, 101 Reykjavik, Iceland; 44 Laboratory of Epidemiology, Demography, and Biometry, National Institute on Aging, Intramural Research Program, NIH, Bethesda, MD 20892, USA; 45 MRC Human Genetics Unit, Institute of Genetics and Molecular Medicine, University of Edinburgh, Edinburgh EH4 2XU, UK; 46Department of Clinical Chemistry, Fimlab Laboratories, Tampere 33520, Finland; 47Department of Clinical Chemistry, University of Tampere School of Medicine, Tampere 33014, Finland; 48University of Tampere, School of Medicine, Tampere 33014, Finland; 49Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, NC 27514, USA; 50Department of Endocrinology, Boston Children’s Hospital, Boston, MA 02115, USA; 51Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA 02115, USA; 52Department of Epidemiology, School of Public Health, University of Alabama at Birmingham, Birmingham, AL 35233, USA; 53Department of Clinical Physiology, Tampere University Hospital, Tampere 33521, Finland; 54Department of Clinical Physiology, University of Tampere School of Medicine, Tampere 33014, Finland; 55Departments of Genetics and Biostatistics, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA; 56Department of Medicine, Division of Cardiovascular Medicine, Stanford University, School of Medicine, Palo Alto, CA 94305, USA; 57Research Centre for Prevention and Health, The Capital Region of Denmark, Copenhagen 2600, Denmark; 58Department of Clinical Experimental Research, Rigshospitalet, Glostrup 2100, Denmark; 59Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen 2200, Denmark; 60Center for Human Genetics, Division of Public Health Sciences, Wake Forest School of Medicine, Winston-Salem, NC 27157, USA; 61Department of Medicine, Divisions of Allergy and Clinical Immunology and General Internal Medicine, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA; 62Laboratory of Neurogenetics, National Institute on Aging, NIH, Bethesda, MD 20892, USA; 63Department of Genome Sciences, University of Washington, Seattle, WA 98105, USA; 64Department of Cardiology, Heart Center, Tampere University Hospital, Tampere 33521, Finland; 65Cardiology Section and Center for Population Genomics, Boston Veteran’s Administration (VA) Healthcare, Boston, MA 02118, USA; 66Institute of Clinical Chemistry and Laboratory Medicine, University Medicine Greifswald, Greifswald 17475, Germany; 67Cardiovascular Health Research Unit, Departments of Medicine Epidemiology and Health Services, University of Washington, Seattle, WA 98101, USA; 68Group Health Research Institute, Group Health Cooperative, Seattle, WA 98101, USA; 69Department of Clinical Physiology and Nuclear Medicine, Turku University Hospital, Turku 20521, Finland; 70Research Centre of Applied and Preventive Cardiovascular Medicine, University of Turku, Turku 20520, Finland; 71Department of Internal Medicine, Erasmus MC, Rotterdam 3000, the Netherlands; 72Netherlands Consortium for Healthy Ageing (NCHA), Rotterdam 3015, the Netherlands; 73Institute for Translational Genomics and Population Sciences, Los Angeles Biomedical Research Institute, Torrance, CA 90502, USA; 74Department of Pediatrics, Harbor-UCLA Medical Center, Torrance, CA 90502, USA; 75Alzheimer Scotland Research Centre, Edinburgh EH8 9JZ, UK; 76Institute for Community Medicine, University Medicine Greifswald, Greifswald 17475, Germany; 77Departments of Pathology and Laboratory Medicine and Biochemistry, University of Vermont College of Medicine, Colchester, VT 05446, USA; 78Departments of Medicine and Pathology, University of Vermont College of Medicine, Burlington, VT 05405, USA; 79Department of Family Population and Preventive Medicine, Stony Brook University, Stony Brook, NY 11794, USA; 80Vanderbilt Epidemiology Center, Department of Obstetrics & Gynecology, Institute for Medicine and Public Health, Vanderbilt Genetics Institute, Vanderbilt University, Nashville, TN 37203, USA; 81Department of Medical Sciences, Cardiology and Uppsala Clinical Research Center, Uppsala University, Uppsala 751 85, Sweden; 82Genetics Target Sciences, GlaxoSmithKline, King of Prussia, PA 19406, USA; 83Green Lane Cardiovascular Service, Auckland City Hospital and University of Auckland, Auckland 1142, New Zealand; 84Department of Physiology and Biophysics, University of Mississippi Medical Center, Jackson, MS 39216, USA; 85The Mindich Child Health and Development Institute, Icahn School of Medicine at Mount Sinai, New York, NY 10069, USA; 86Department of Anesthesiology & Critical Care Medicine, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA; 87Departments of Internal Medicine and Human Genetics, University of Michigan, Ann Arbor, MI 48108, USA; 88 Zilber School of Public Health, University of Wisconsin-Milwaukee, Milwaukee, WI 53205, USA; 89Department of Epidemiology, University of Washington, Seattle, WA 98195, USA; 90Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA 98109, USA 91 These authors contributed equally to this work *Correspondence: apreiner@u.washington.edu (A.P.R.), guillaume.lettre@umontreal.ca (G.L.) http://dx.doi.org/10.1016/j.ajhg.2016.05.007. The American Journal of Human Genetics 99, 8–21, July 7, 2016 9 to shear stress and produce the vasodilator nitric oxide to regulate vascular tonus.1 RBCs participate in antimicrobial strategies to fight hemolytic pathogens2 and in the inflammatory response, acting as a reservoir for multiple chemokines.3 Furthermore, the direct involvement of RBCs in adhering to the vascular endothelium or supporting thrombin generation may help to promote blood coagulation or thrombosis.4,5 Given the paramount importance of RBCs in physiology, it is not surprising that monitoring their features is common practice in medicine to assess the overall health of patients. An excessive number of circulating RBCs (erythrocytosis [MIM: 133100]) can suggest a primary bone marrow disease, a myeloproliferative neoplasm such as polycythemia vera (MIM: 263300), or chronic hypoxemia due to congenital heart defects. Low HGB concentration and hematocrit (HCT) levels (anemia) can indicate inherited HGB or RBC structural gene mutations, malnutrition, or kidney diseases. By considering the volume (mean corpuscular volume [MCV]), hemoglobin content (mean corpuscular hemoglobin [MCH] and mean corpuscular hemoglobin concentration [MCHC]) or the distribution width (RDW) of RBCs, a physician can distinguish between the different causes of anemia (e.g., microcytic/hypochromic due to iron deficiency6). In addition, epidemiological studies have correlated high RDW values with a worse prognosis in heart failure patients.7 RDW is also an independent predictor of overall mortality in healthy individuals, as well as a predictor of mortality in patients with various conditions such as cardiovascular diseases, obesity, malignancies, and chronic kidney disease.8–12 RBC count and indices vary among individuals, and 40%–90% of this phenotypic variation is heritable.13–16 Identifying the genes and biological pathways that contribute to this inter-individual variation in RBC traits could highlight modifiers of severity and/or therapeutic options for several hematological diseases. Already, largescale genome-wide association studies (GWASs) have found dozens of SNPs associated with one or more of these RBC traits.17,18 However, owing to their design, GWASs are largely insensitive to rare (minor allele frequency [MAF] < 1%) and low-frequency (1% % MAF < 5%) genetic variants. Using an exome array, we previously performed an association study for HGB and HCT in 31,340 European-ancestry individuals and identified rare coding or splice site variants in the erythropoietin and b-globin genes.19 Within the framework of the BloodCell Consortium (BCX),20,21 we now report a larger genotyping-based exome survey of seven RBC traits conducted in up to 130,273 individuals, including 23,896 participants of non-European ancestry. With this experiment, our initial goals were to expand the list of rare and lowfrequency coding or splice site variants associated with RBC traits and to explore whether the exome array can complement the GWAS approach to fine map RBC causal genes. 10 The American Journal of Human Genetics 99, 8–21, July 7, 2016 Subjects and Methods Study Participants The Blood-Cell Consortium (BCX) aims to identify novel common and rare variants associated with blood-cell traits using an exome array. BCX is comprised of more than 134,021 participants from 24 discovery cohorts and five ancestries: European, African American, Hispanic, East Asian, and South Asian. Detailed description of the participating cohorts is provided in Table S1. BCX is interested in the genetics of all main hematological measures and is divided into three main working groups: RBC, white blood cell (WBC),21 and platelet (PLT).20 For the RBC working group, we analyzed seven traits available in up to 130,273 individuals: RBC count (31012/L), HGB (g/dL), HCT (%), MCV (fL), MCH (pg), MCHC (g/dL), and RDW (%) (Table S2). The BCX procedures were in accordance with the institutional and national ethical standards of the responsible committees and proper informed consent was obtained. Genotyping and Quality-Control Steps Participants from the different studies were genotyped on one of the following exome chip genotyping arrays: Illumina ExomeChip v.1.0, Illumina ExomeChip v.1.1_A, Illumina ExomeChip12 v.1.1, Affymetrix Axiom Biobank Plus GSKBB1, or Illumina HumanOmniExpressExome Chip. Genotypes were then called either (1) with the Illumina GenomeStudio GENCALL and subsequently recalled using zCALL or (2) by the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) Consortium Exome Chip effort22 (Table S3). The same quality-control steps were followed by each participating study. We excluded variants with low genotyping success rate (<95%, except for WHI that used a cutoff <90%) (Table S3). Samples with call rate < 95% (except for SOLID-TIMI 52 and STABILITY that used 94.5% and 93.5% cutoffs, respectively) after joint or zCALL calling and with outlying heterozygosity rate were also excluded. Other exclusions were deviation from Hardy-Weinberg equilibrium (p < 1 3 10 6) and gender mismatch. We performed principal-component analysis (PCA) or multidimensional scaling (MDS) and excluded sample outliers from the resulting plots through visual inspection, using populations from the 1000 Genomes Project to anchor these analyses. Keeping only autosomal and X chromosome variants for the analysis, we aligned all variants on the forward strand and created a uniform list of reference alleles using the --force alleles command in PLINK.23 Finally, an indexed variant call format file (VCF) was created by each study and checked for allele alignment and any allele or strand flips using the checkVCF package.24 Prior to performing meta-analyses of the association results provided by each participating study, we ran the EasyQC protocol25 to check for population allele frequency deviations and proper trait transformation in each cohort. Phenotype Modeling and Association Analyses When possible, we excluded individuals with blood cancer, leukemia, lymphoma, bone marrow transplant, congenital or hereditary anemia, HIV, end-stage kidney disease, dialysis, splenectomy, or cirrhosis and those with extreme measurements of RBC traits (Table S1). We also excluded individuals on erythropoietin treatment or chemotherapy. Additionally, we excluded pregnant women and individuals with acute medical illness at the time the complete blood count (CBC) was done. For the seven RBC traits, within each study, we adjusted for age, age-squared, gender, the first ten principal components, and, where applicable, other study-specific covariates such as study center via a linear regression model. Within each study, we then applied inverse normal transformation on the residuals and tested the phenotypes for association with the ExomeChip variants using either RVtests (v.20140416)26 or RAREMETALWORKER.0.4.9.27 Discovery Meta-analyses Score files generated by RVtests or RAREMETALWORKER from each participating study were used to carry out meta-analyses of the single variant association results using RareMETALS v.5.9.28 All analyses were performed separately in each of European American (EA) and African American (AA) ancestries. In the multi-ancestry meta-analyses, we combined individuals of European, African American, Hispanic, East-Asian, and South-Asian ancestries (All). We included variants with allele frequency difference between the highest and lowest MAF < 0.3 for EA and AA ancestries and < 0.6 for the combined ancestry meta-analyses. For the gene-based analyses, we used score files and variance-covariance matrices from the study-specific association results and applied the sequence kernel association test (SKAT)29 and variable threshold (VT) algorithms30 in RareMETALS considering only missense, nonsense, and splice site variants with a MAF < 1%. Gene-based analyses were also stratified by ancestry. Significance thresholds were determined using Bonferroni correction assuming ~250,000 independent variants (p < 2 3 10 7 for the singlevariant analyses) and ~17,000 genes tested on the ExomeChip (p < 3 3 10 6 for the gene-based tests). Conditional Analysis and Replication In order to identify independent signals, we performed conditional analyses. In each round of conditional analysis, we conditioned on the most significant single variant in a 1 Mb window. These conditional analyses were performed at the meta-analysis level using RareMETALS. We repeated this step until there were no new signals identified in each region, defined as p < 2 3 10 7. We then checked for linkage disequilibrium (LD) within the list of variants that was retained from the conditional analyses. For variants that were in moderate-to-strong LD (r2 R 0.3), we kept the most significant. We attempted replication of the final list of independent variants in eight additional studies that contributed a total of 27,480 individuals (n ¼ 21,473 for EA and n ¼ 6,007 for AA) (Table S4). The division of discovery and replication samples was dictated by timing because we collected all groups we were aware of for initial discovery and then found others who could participate only much later and hence were used for replication. These studies followed similar analytical procedures and steps as those followed by the discovery analysis (see above). A joint meta-analysis of the discovery and the replication results was carried out using a fixed-effects model and inverse-variance weighting as implemented in METAL.31 We considered as replicated markers those with a nominal preplication < 0.05 and an effect on phenotype in the same direction as in the discovery results. Allelic Imbalance and Expression of CD36 We checked for allelic imbalance (AI) of the rs3211938 variant in CD36 (MIM: 173510) as well as the expression of the gene in 12 samples of fetal liver erythroblasts obtained from anonymous donors. Details on the protocol including RNA extraction and sequencing can be found elsewhere.32 We calculated the difference in the ratio of reads of the reference allele (T) and the alternate allele (G) of rs3211938. In brief, reads overlapping rs3211938 were counted with samtools (v.1.1) mpileup software using genome build hg19. We kept uniquely mapping reads using -q 50 argument (mapping quality > 50) and sites with base quality > 10. Statistical significance of the difference in the ratio of reads between the reference allele and the alternate allele was assessed with a binomial test. For each sample, we summed all reads overlapping all heterozygous SNPs and calculated the expected ratio within each SNP allele combination. Reads that fall in the top 25th coverage percentile were down-sampled so that the highest covered sites do not bias the expected ratio.33 For rs3211938, the expected T:G ratio was 0.507. Expression Quantitative Trait Loci Analysis We cross-referenced our list of RBC novel variants with more than 100 separate expression quantitative trait loci (eQTL) published datasets. Datasets were collected through publications, publically available sources, and private collaborations. A general overview of a subset of >50 eQTL studies has been published,34 with specific citations for >100 datasets included in the current query followed here. A complete list of tissues and studies used can be found in the Supplemental Data. We considered SNPs that are themselves expression SNPs (eSNP) when they meet a p < 0.0001 threshold or when they are in LD (r2 > 0.3) with the best eSNP (p < 0.0001). Results Single-Variant Meta-analyses We meta-analyzed ExomeChip results for seven RBCrelated phenotypes (RBC count, HCT, HGB, MCH, MCHC, MCV, and RDW) available in up to 130,273 individuals from 24 studies and 5 ancestries (Tables S1–S3 and Figure S1). Across these different phenotypes, a total of 226 variants reached exome-wide significance (p < 2 3 10 7) in the combined ancestry analyses (Figures 1 and S2). Given that some of these RBC traits are correlated (Figure S3), these associated variants highlight 71 different loci (defined using a 1 Mb interval). Overall, we observed only modest inflation of the test statistics (lGC ¼ 1.03– 1.05), consistent with little confounding due to technical artifacts, population stratification, or cryptic relatedness. In order to identify independent variants, we performed conditional analyses at the meta-analysis level adjusting for the effect of the most significant variant in a 1 Mb region in a stepwise manner (Subjects and Methods). After this analysis, we obtained a list of 126 independent variants associated with at least one RBC trait at p < 2 3 10 7 (Table S5). Selecting only variants that lie more than 1 Mb away from a known GWAS locus resulted in 23 independent variants located within 20 novel RBC loci, where novel is used to define loci not found in the existing literature (Table 1). We attempted to replicate these 126 variants in 8 independent cohorts totaling 27,480 participants (Table S5). Overall, we observed a strong replication, with 94 of the 126 variants showing consistent direction of effect between the discovery and replication analyses (binomial p ¼ 3 3 10 8; Table S5). Of the 23 novel RBC variants, we replicated 16 at nominal p < 0.05 for at The American Journal of Human Genetics 99, 8–21, July 7, 2016 11 Figure 1. Quantile-Quantile Plots of Single-Variant Association Results in the All Ancestry Meta-analyses for the Seven Red Blood Cell Traits Analyzed (A) Distribution of the single variant results for all variants tested on the exome array. (B) Only markers with a minor allele frequency < 5% are shown here. (C) Variants outside of known RBC GWAS regions. Variants that are within 1 Mb from a previously published RBC GWAS locus were excluded for this QQ plot. Abbreviations are as follows: HCT, hematocrit; HGB, hemoglobin; RBC, red blood cell count; MCV, mean corpuscular volume; MCHC, mean corpuscular hemoglobin concentration; MCH, mean corpuscular hemoglobin; RDW, red blood cell distribution width. least one RBC trait (binomial p ¼ 3 3 10 16; Table 1). Out of these 16 novel and replicated RBC variants, there are five missense variants, including two variants with MAF < 5% in MAP1A (MIM: 600178) and HNF4A (MIM: 600281) and one nonsense variant in CD36 (Table 1). Among the remaining nine novel and replicated RBC variants, there are five intronic, one synonymous, one 50 UTR, and one intergenic marker (Table 1). Prioritization of Candidate Genes and Genetic Variants Our single-variant analyses in EA samples identified one rare missense variant in ALAS2 (MIM: 301300) associated with MCV and MCH (rs201062903, p.Pro507Leu [c.1559C>T], MAF ¼ 0.2%) (Table 1). The association with this variant did not replicate, potentially because of limited statistical power (the replication sample size for this rare marker was 5,044; see also Discussion). ALAS2 encodes 5-aminolevulinate synthase 2, the rate-controlling enzyme of erythroid heme synthesis. Additionally, rare mutations in ALAS2 cause X-linked sideroblastic anemia (MIM: 300751) and erythropoietic protoporphyria (MIM: 300752). Thus, despite the lack of replication, ALAS2 remains an excellent candidate gene to modulate RBC traits. The ALAS2 p.Pro507Leu variant, which is not reported in the ClinVar database, maps between two amino acids (Tyr506 and Thr508) that are important for catalytic activity and known to be mutated in cases of sideroblastic anemia.35 Two low-frequency missense variants identified in our analyses implicate MAP1A and HNF4A in RBC biology (Table 1). MAP1A encodes microtubule-associated protein 1A, a gene highly expressed in the nervous system and mostly studied in the context of neuronal diseases, although it is expressed in many additional tissues, 12 The American Journal of Human Genetics 99, 8–21, July 7, 2016 including hematopoietic cells.36 Deletion of MAP1A in the mouse causes defects in synaptic plasticity.37 This observation is interesting given that inactivation of ANK1 (MIM: 612641), another gene that encodes a cytoskeleton protein and is expressed in neurons and RBCs, is associated with neurological dysfunction in the mouse and spherocytosis and hemolytic anemia in humans (MIM: 182900). Our meta-analyses confirmed two known independent ANK1 variants associated with MCHC: an intronic SNP (rs4737009, MAF ¼ 19.8%, p ¼ 1.3 3 10 8) and a low-frequency missense variant (rs34664882, p.Ala1462Val, MAF ¼ 2.9%, p ¼ 1.7 3 10 16) (Table S5; N.P., U.M.S., J.B.-J., and M.-H.C., unpublished data).17 In the accompanying BCX PLT article,20 we report that the same MAP1A rs55707100 allele (p.Pro2349Leu [c.7046C>T]) associated here with decreased HGB concentration is also associated with increased PLT count. Furthermore, recent studies have identified associations between rs55707100 and HDL-cholesterol and triglyceride levels (S. Mukherjee, 2015, ASHG, conference). Adding to the complexity, the GTEx dataset indicates that rs55707100 is an expression quantitative trait locus (eQTL) for ADAL (peQTL ¼ 9 3 10 11) but not for MAP1A.38 ADAL is a poorly characterized adenosine deaminase-like protein that is highly expressed in human erythroblasts. However, the eQTL association between rs55707100 and ADAL could simply reflect ‘‘LD shadowing’’ from nearby markers that are much stronger eQTL variants for ADAL. Indeed, rs3742971 (a common variant located in ADAL’s 50 UTR) is in partial LD with rs55707100 (r2 ¼ 0.18 in European populations from the 1000 Genomes Project) and strongly associated with ADAL expression levels (peQTL ¼ 6 3 10 49). The second low-frequency missense variant associated with HGB and HCT maps within the coding sequence of Table 1. Association Results of Variants in Novel Loci Associated with Red Blood Cell Traits Marker Info Trait Position Discovery A1/A2 SNP Annotation Gene n Replication AF (A2) Beta (SE) p Value n Combined AF (A2) Beta (SE) p Value RDW-EA 1: 25,768,937 A/G rs10903129* intron TMEM57-RHD 45,573 0.544 0.037 (0.007) 1.19 3 10 7 RDW-All 1: 25,768,937 A/G rs10903129* intron TMEM57-RHD 56,194 0.568 0.034 (0.006) 9.58 3 10 8 24,474 0.600 0.021 (0.01) HCT-All C/T rs4072037* 109,875 0.554 0.025 (0.005) 5.82 3 10 8 25,006 0.563 0.038 (0.009) 5.96 3 10 3,162 1: 155,162,067 synonymous MUC1 Beta (SE) p Value 18,475 0.560 0.023 (0.011) 0.0373 0.033 (0.006) 2.41 3 10 8 0.03 (0.005) 8 0.0252 5 1.32 3 10 0.027 (0.004) 3.47 3 10 11 0.012 (0.026) 0.6410 0.023 (0.044) 1.68 3 10 7 HGB-All 2: 27,741,237 T/C rs780094 intron GCKR 130,273 0.626 0.024 (0.004) 7.14 3 10 8 RBC-All 2: 219,509,618 C/A rs2230115* missense ZNF142 74,488 0.509 0.033 (0.006) 9.74 3 10 9 27,442 0.477 0.024 (0.01) 0.0167 0.031 (0.005) 7.11 3 10 10 HCT-All 3: 56,771,251 A/C rs3772219* missense ARHGEF3 109,875 0.338 0.028 (0.005) 2.38 3 10 9 25,006 0.366 0.021 (0.01) 0.0292 0.027 (0.004) 2.56 3 10 10 HGB-All 3: 56,771,251 A/C rs3772219* missense ARHGEF3 130,273 0.336 0.026 (0.004) 3.76 3 10 9 27,749 0.367 0.02 (0.009) 0.0331 0.025 (0.004) 4.33 3 10 10 HCT-EA 4: 88,008,782 G/A rs236985 intron AFF1 87,444 0.394 0.032 (0.005) 3.89 3 10 10 19,968 0.405 0.02 (0.011) 0.0626 0.03 (0.005) 10 RBC-EA 4: 88,008,782 G/A rs236985* intron AFF1 60,231 0.393 0.034 (0.006) 3.50 3 10 8 21,435 0.405 0.023 (0.011) 0.0273 0.031 (0.005) 4.22 3 10 9 21,743 0.586 0.029 (0.01) 0.0052 0.033 (0.004) 8.23 3 10 15 0.626 1.14 3 10 AFF1 106,377 0.595 0.034 (0.005) 3.97 3 10 13 rs10063647* intron LINC01184SLC12A2 45,573 0.463 0.05 (0.007) 1.72 3 10 13 18,475 0.480 0.033 (0.011) 0.0018 0.045 (0.006) 2.88 3 10 15 A/G rs10063647* intron LINC01184SLC12A2 56,194 0.506 0.044 (0.006) 2.11 3 10 12 24,474 0.545 0.03 (0.01) 0.04 (0.005) 2.37 3 10 14 RDW-EA 5: 127,522,543 C/T rs10089* utr_5p LINC01184SLC12A2 45,573 0.21 0.051 (0.008) 8.45 3 10 10 16,692 0.215 0.058 (0.014) 2.71 3 10 0.053 (0.007) 1.15 3 10 13 RDW-All 5: 127,522,543 C/T rs10089* utr_5p LINC01184SLC12A2 56,194 0.207 0.044 (0.008) 4.08 3 10 9 22,691 0.208 0.045 (0.012) 0.0001 0.044 (0.006) 2.73 3 10 12 HGB-All C/A rs35742417* missense RREB1 130,273 0.174 0.030 (0.005) 1.17 3 10 8 4,074 0.207 0.065 (0.028) 0.0190 0.032 (0.005) 1.50 3 10 9 RDW-AA 7: 80,300,449 T/G rs3211938* nonsense CD36 6,666 0.087 0.174 (0.031) 2.36 3 10 8 5,999 0.086 0.139 (0.032) 1.83 3 10 5 0.161 (0.025) 7.09 3 10 11 RDW-All 7: 80,300,449 T/G rs3211938* nonsense CD36 55,510 0.012 0.171 (0.029) 5.29 3 10 9 22,691 0.023 0.139 (0.032) 1.61 3 10 5 0.157 (0.022) 5.12 3 10 13 16,692 0.466 0.026 (0.011) 0.0210 0.034 (0.006) 1.29 3 10 8 HGB-EA 4: 88,030,261 G/T rs442177* RDW-EA 5: 127,371,588 A/G RDW-All 5: 127,371,588 The American Journal of Human Genetics 99, 8–21, July 7, 2016 13 6: 7,247,344 intron 0.0014 5 RDW-EA 8: 126,490,972 A/T rs2954029* intergenic TRIB1 45,573 0.46 0.036 (0.007) 1.53 3 10 7 RDW-All 8: 126,490,972 A/T rs2954029* intergenic TRIB1 56,194 0.439 0.032 (0.006) 1.83 3 10 7 22,691 0.432 0.021 (0.01) 0.0298 0.029 (0.005) 2.54 3 10 8 MCH-All 10: 105,659,826 T/C rs2487999 missense OBFC1 66,318 0.869 0.047 (0.009) 4.12 3 10 8 26,749 0.861 0.025 (0.013) 0.0601 0.041 (0.007) 1.75 3 10 8 MCH-AA 11: 92,722,761 G/A rs1447352 intergenic MTNR1B 8,273 0.557 0.089 (0.016) 1.85 3 10 8 5,038 0.022 (0.02) 0.07 (0.014) 6 HGB-EA C/T rs55707100* missense MAP1A 106,377 0.033 0.071 (0.013) 1.65 3 10 8 21,743 0.0223 0.099 (0.033) 0.0028 A/G rs2667662* TELO2 10,849 0.099 (0.015) 1.79 3 10 10 5,034 0.724 0.093 (0.022) 3.02 3 10 0.134 (0.025) 7.08 3 10 8 6,002 0.124 0.106 (0.027) 0.0001 15: 43,820,717 MCV-AA 16: 1,551,082 MCV-AA 16: 2,812,939 C/A rs2240140* intron missense SRRM2 8,525 0.725 0.118 0.562 0.2713 5 1.08 3 10 0.075 (0.012) 2.31 3 10 10 0.098 (0.014) 7.32 3 10 12 0.128 (0.022) 5.24 3 10 9 (Continued on next page) 14 The American Journal of Human Genetics 99, 8–21, July 7, 2016 Table 1. Continued Marker Info Trait Position Discovery A1/A2 SNP Annotation Gene n Replication AF (A2) Beta (SE) p Value n Combined AF (A2) Beta (SE) p Value HCT-EA 17: 59,017,025 T/C rs8080784 intron BCAS3-TBX2 79,344 0.158 0.039 (0.007) 2.62 3 10 8 HGB-EA 17: 59,483,766 C/T rs8068318 intron BCAS3-TBX2 106,377 0.722 0.026 (0.005) 1.53 3 10 7 21,743 0.730 0.021 (0.011) 0.0565 MCV-EA 20: 31,140,165 C/T rs4911241* intron NOL4L 61,462 0.04 (0.007) 1.25 3 10 8 21,714 0.252 0.025 (0.012) 0.0302 18,475 0.240 0.241 Beta (SE) p Value 19,968 0.148 0.011 (0.014) 0.4349 0.029 (0.006) 3.39 3 10 6 0.025 (0.005) 2.55 3 10 8 0.036 (0.006) 2.01 3 10 9 0.049 (0.012) 7.44 3 10 5 0.045 (0.007) 2.01 3 10 11 5 0.04 (0.006) 4.60 3 10 11 RDW-EA 20: 31,140,165 C/T rs4911241* intron NOL4L 45,573 0.242 0.043 (0.008) 5.79 3 10 8 RDW-All 20: 31,140,165 C/T rs4911241* intron NOL4L 56,194 0.235 0.038 (0.007) 1.56 3 10 7 24,474 0.222 0.044 (0.011) 6.10 3 10 HCT-EA 20: 43,042,364 C/T rs1800961* missense HNF4A 79,344 0.024 0.083 (0.015) 1.44 3 10 8 19,968 0.033 0.082 (0.028) 0.0037 0.083 (0.013) 1.91 3 10 10 HGB-EA 20: 43,042,364 C/T rs1800961* missense HNF4A 98,277 0.032 0.073 (0.013) 2.53 3 10 8 21,743 0.032 0.062 (0.027) 0.0232 0.071 (0.012) 1.93 3 10 9 HCT-All 20: 43,042,364 C/T rs1800961* missense HNF4A 100,313 0.022 0.077 (0.014) 2.31 3 10 8 25,006 0.027 0.091 (0.028) 0.0010 0.08 (0.012) 11 HGB-All 22: 44,324,727 C/G rs738409 missense PNPLA3 130,273 0.223 0.028 (0.005) 2.24 3 10 8 4,074 0.218 0.053 (0.027) 0.0504 0.029 (0.005) 4.81 3 10 9 5,855 0.001 0.291 (0.235) 0.215 0.323 (0.052) 5.81 3 10 10 9.88 3 10 MCH-EA X: 55,039,960 G/A rs201062903 missense ALAS2 52,758 0.002 0.324 (0.053) 7.32 3 10 10 MCH-All X: 55,039,960 G/A rs201062903 missense ALAS2 65,067 0.002 0.322 (0.051) 3.36 3 10 10 10,893 0.001 0.276 (0.224) 0.218 0.321 (0.051) 2.68 3 10 10 MCV-EA X: 55,039,960 G/A rs201062903 missense ALAS2 60,211 0.002 0.285 (0.049) 7.11 3 10 9 5,044 0.178 (0.248) 0.472 0.282 (0.049) 6.11 3 10 9 0.001 Variants in novel loci with p < 2 3 10 7 and that were retained after conditional analyses are presented here. All variants are >1 Mb apart from a known GWAS signal for RBC traits. Chromosome positions are given on human genome build hg19. Allele frequency and effect size are given for the alternate (A2) allele. Replication was carried out in six cohorts for EA and two cohorts for AA and was performed in RareMetals; meta-analyses of the discovery and replication cohorts are presented under ‘‘Combined’’ and were carried out in METAL. Asterisks (*) indicate variants that replicated with a nominal p < 0.05. Abbreviations are as follows: EA, European American; AA, African American; All, combined ancestry (EA þ AA þ Asians þ Hispanics); A1, reference allele; A2, alternate allele; N, sample size; AF, allele frequency; SE, standard error; HCT, hematocrit; HGB, hemoglobin; RBC, red blood cell count; MCV, mean corpuscular volume; MCHC, mean corpuscular hemoglobin concentration; MCH, mean corpuscular hemoglobin; RDW, red blood cell distribution width. Figure 2. CD36 Expression in Human Erythroblasts (A) In a dataset of 12 human fetal liver erythroblasts, all samples were homozygous at rs3211938 for the reference T-allele with the exception of one heterozygous sample (FL11). FL11 demonstrated strong allelic imbalance: we observed 705 reads for the reference allele (T) and 126 reads for the alternate allele (G) (binomial p ¼ 4.9 3 10 95). (B) FL11 (in green) shows the lowest CD36 expression level when compared to the other 11 samples. Abbreviation is as follows: FPKM, fragments per kilobase of transcript per million mapped reads. the transcription factor HNF4A (Table 1). This marker, rs1800961 (p.Thr117Ile [c.350C>T]), has previously been associated with HDL and total cholesterol, C-reactive protein, fibrinogen, and coagulation factor VII levels.39–42 Mutations in HNF4A cause maturity-onset diabetes of the young (MODY [MIM: 125851]) and a common intronic SNP in HNF4A (rs4812829) has been associated with type 2 diabetes (MIM: 125853) risk.43 The missense rs1800961 associated with HGB and HCT is only in weak LD with rs4812829 (r2 ¼ 0.021 in EA populations from the 1000 Genomes Project). Querying recently released ExomeChip data from Type 2 Diabetes Genetics (Web Resources), we found that rs1800961 is also associated with T2D risk in ~82,000 participants (p ¼ 9.5 3 10 7, odds ratio ¼ 1.16). HNF4A is expressed in the kidney and could influence HGB and HCT through the regulation of erythropoietin production.44 It is also abundantly expressed in the liver, where it could indirectly affect HGB and HCT levels through an effect on blood lipid levels (see Discussion). HNF4A is detectable at low levels in erythroblasts, and the BLUEPRINT Project has found that some HNF4A isoforms may be more highly expressed in this cell type (Figure S4).45 In AA, we identified a nonsense variant (rs3211938, p.Tyr325Ter [c.975T>G], MAF ¼ 8.7%, p ¼ 7.1 3 10 11) in CD36 associated with RDW. This variant displays a wide variation in allele frequency between AA and EA (MAFEA ¼ 0.01%). The association is slightly improved in the ancestry-combined meta-analysis (p ¼ 5.1 3 10 13) because there is also evidence of association in Hispanics (MAF ¼ 1.9%, p ¼ 0.022) (Table 1). We examined a dataset of ex vivo differentiated human erythroblasts to determine whether this CD36 nonsense variant shows allelic imbalance (AI).32 All samples were homozygous at rs3211938 for the reference allele with the exception of one heterozygous sample (FL11). FL11 had the lowest level of CD36 expression among the 12 samples tested and demonstrated strong AI where we observe 705 sequence reads for the reference allele (T) versus 126 for the alternate allele (G) (p ¼ 4.9 3 10 95; Figure 2). To confirm this finding in independent samples, we queried the GTEx dataset, which has compiled RNA-sequencing and genotype information from multiple human tissues.38 GTEx does not include data for human erythroblasts. However, it detected a strong eQTL effect of rs3211938 on CD36 expression in whole blood (peQTL ¼ 1.1 3 10 15), with carriers of the G-allele expressing less CD36 (Figure S5). Furthermore, GTEx reported evidence for moderate AI in multiple tissues for CD36-rs3211938, with the G-allele being under-represented among sequence reads (Figure S5). These results strongly support our observations in human erythroblasts. eQTL Analysis To prioritize additional causal genes at RBC loci that contain non-coding variants, we cross-referenced our list of novel variants with more than 100 published eQTL datasets (Subjects and Methods). Overall, 12 variants were significant eQTLs in a wide variety of tissues (Table S6). The most interesting eQTL finding is the association between rs10903129, a common marker associated with RDW in our analyses and located within an intron of TMEM57 (MIM: 610301), and the expression of RHD (MIM: 111680) in whole blood. RHD is located 112 kb downstream of TMEM57 and encodes the D antigen of the clinically significant Rhesus (Rh) blood group. rs10903129 has also been associated with total cholesterol levels and erythrocyte sedimentation rate (ESR).46,47 The association with ESR is particularly intriguing given that it is considered a non-specific indicator of inflammation. As described above, RDW is also abnormal in chronic diseases, such as atherosclerosis and diabetes, which have an important inflammation component. Gene-Based Association Testing Despite our large sample size, statistical power remains limited for rare variants of weak-to-moderate phenotypic effect. To try to capture these genetic factors, we performed gene-based testing by aggregating coding and splice site variants with MAF < 1% within each gene (Subjects and Methods). The SKAT analyses identified two genes: ALAS2 associated with MCH and PKLR (MIM: 609712) associated with HGB and HCT (Table 2). The ALAS2 signal was driven The American Journal of Human Genetics 99, 8–21, July 7, 2016 15 Table 2. Gene-Based Association Results VT Trait Gene n Number of Variants Analyzed SKAT p Value HGB-EA PKLR 106,377 15 1.92 3 10 HGB-All PKLR 130,273 15 0.00016 p Value 5 Top Variant Top-Variant MAF Top-Variant p Value 7.02 3 10 7 rs116100695 0.003 1.17 3 10 5 6.57 3 10 7 rs116100695 0.003 1.94 3 10 5 7.95 3 10 7 rs116100695 0.003 2.49 3 10 5 7 rs201062903 0.002 7.32 3 10 10 rs202037221 3.0 3 10 HCT-All PKLR 109,875 15 3.96 3 10 5 MCH-EA ALAS2 54,009 11 4.78 3 10 6 5.79 3 10 MCHC-All ALPK3 84,841 28 1.95 3 10 6 0.793 5 0.0005 6 Gene-based results of the VT and SKAT algorithms for genes associated with RBC traits at p < 3 3 10 . We analyzed non-synonymous coding (nonsense, missense) and splice site variants with a minor allele frequency (MAF) < 1%. Abbreviations are as follows: EA, European American; All, combined ancestry (EA þ AA þ Asians þ Hispanics); n, sample size; HCT, hematocrit; HGB, hemoglobin; MCHC, mean corpuscular hemoglobin concentration; MCH, mean corpuscular hemoglobin. by a single rare missense variant (rs201062903) and was described above. PKLR encodes the erythrocyte pyruvate kinase (PK) that catalyzes the last step of glycolysis. PK deficiency, usually caused by recessive mutations, is one of the main causes of non-spherocytic hemolytic anemia (MIM: 266200). In fact, one of the variants identified in our metaanalysis (rs116100695, p.Arg486Trp [c.1456T>G], MAF ¼ 0.3%, betaHGB ¼ 0.242 g/dl, pHGB ¼ 1.2 3 10 5) is a frequent cause of PK deficiency in Italian and Spanish subjects.48,49 This variant was confirmed in the replication cohorts (preplication ¼ 0.039; Table S7). Two additional PKLR rare missense variants contribute to the gene-based association statistic with HGB and HCT: rs61755431 (p.Arg569Gln [c.1706G>A], MAF ¼ 0.2%, betaHGB ¼ 0.179 g/dl, pHGB ¼ 0.006) and rs8177988 (p.Val506Ile [c.1516G>A], MAF ¼ 0.6%, betaHGB ¼ þ0.116 g/dl, pHGB ¼ 0.003). It is noteworthy that the p.Val506Ile substitution is associated with increased HGB concentration given that this amino acid maps to a PKLR structural domain necessary for protein interaction.50 This heterogeneity of effect among the PKLR missense variants might explain why SKAT’s result is more significant than VT’s for this gene (Table 2). A third gene, ALPK3, was identified only in the VT analysis for association with MCHC (Table 2). ALPK3 encodes a kinase previously implicated in cardiomyocyte differentiation.51 We could not test for replication because of the rarity of ALPK3’s coding alleles (Table S7). RBC Variants and Pleiotropic Effects Besides the overlap within the RBC traits themselves, we identified seven novel RBC variants associated with other blood-cell type traits or with lipid levels (Figure 3 and Table 3). To assess whether the genetic associations with RBC traits are independent of lipid levels, we performed additional analyses in a subset of BCX participants from three of our studies (FHS, MHIBB, and WHI) ranging from ~10,000 to 23,000 individuals. We repeated the association analyses for five RBC loci (TMEM57-RHD rs10903129, AFF1 rs442177, TRIB1 rs2954029, MAP1A rs55707100, and HNF4A rs1800961) additionally adjusting for the respective lipid trait and combined the results across the three studies using fixed-effect meta-analysis 16 The American Journal of Human Genetics 99, 8–21, July 7, 2016 (Table S8). There was little or no change in the effect size or p values associated with the five RBC trait loci upon adjustment for the corresponding lipid trait, suggesting that the RBC and lipid associations are independent of one another and thus represent true ‘‘pleiotropic’’ genetic effects. A correlated response to or role in inflammation might explain why some of the RBC variants are also associated with WBC, PLT, or lipid traits. Another plausible explanation for the concomitant association of several markers with RBC, WBC, and PLT phenotypes could be a more general effect of these genes on the proliferation or differentiation of hematopoietic progenitor cells. This is most likely the case for JAK2 (MIM: 147796) and SH2B3 (MIM: 605093), two key regulators of hematopoietic cells (Figure 3). In this category, we also observed two novel findings, AFF1 (MIM: 159557) and NOL4L, which are associated with RBC and WBC phenotypes and have been previously implicated in leukemia.53,54 Finally, we identified a novel missense variant in ARHGEF3 (MIM: 612115) associated with HGB and HCT. In addition to its association with PLT traits, ARHGEF3 plays a role in the regulation of iron uptake and erythroid cell maturation.55 Discussion We present multi-ethnic meta-analyses of seven RBC traits using ExomeChip results of 130,273 individuals. Our statistical thresholds to declare significance at the discovery stage (p < 2 3 10 7 in the single-variant analyses) was adjusted for the approximate number of variants genotyped on the ExomeChip (Bonferroni correction for 250,000 variants), but we decided not to adjust it for the seven RBC phenotypes tested because of the high correlation between some of these traits (Figure S3). Instead, we relied on independent replication to distinguish true from probably false positive associations. Despite the limited size of our replication set (27,480 individuals), it was encouraging to detect a strong replication of direction of effect for known and novel RBC variants, suggesting a low false discovery rate. In total, we identified 23 novel Figure 3. Venn Diagram Summarizing Pleiotropic Effects for Genetic Variants Associated with Red Blood Cell Traits We considered variants only if their association p values with white blood cell (WBC) traits, platelet (PLT) traits, or with lipid levels was p < 1 3 10 4. Results for WBC and PLT are from the accompanying Blood-Cell Consortium articles.20,21 Results for lipids have previously been published (Table 3). Genes highlighted in red are novel RBC trait findings. variants associated with RBC traits in the single-variant analyses and a collection of three rare missense variants in PKLR associated with HGB and HCT in the gene-based analyses. Out of the 23 novel RBC variants, 16 were replicated at p < 0.05 in the independent samples (Table 1). To inform our replication criteria, we conducted a power analysis using a sample size of 20,000 and considering multiple combinations of allele frequencies and effect sizes. Based on allele frequency and effect size, one of our most difficult to replicate variants was rs1800961 (MAF ¼ 0.022, Beta ¼ 0.028). However, we still had approximately 56% power to detect this association in the replication stage. We identified a nonsense variant in CD36 associated with RDW in African Americans. CD36 is a type B scavenger receptor located on the surface of many cell types, including endothelial cells, platelets, monocytes, and erythrocytes. CD36 is a marker of erythroid progenitor differentiation56 and might also be involved in macrophagemediated clearance of red blood cells.57 Furthermore, CD36 plays a role in many biological pathways such as lipid metabolism/transport and atherosclerosis, hemostasis, and inflammation.58 The nonsense CD36 variant identified in our RDW meta-analysis (rs3211938, Table 1) has previously been associated with platelet count, HDL cholesterol, and C-reactive protein levels in African Americans59,60 and malaria resistance in Africans.61,62 The CD36 locus shows a signature of natural selection in AA populations63 and the MAF of rs3211938 varies widely between continents: in the 1000 Genomes Project, the minor allele is absent from European populations but reaches frequency of 24%–29% in some African populations.64 To characterize the molecular mechanism by which rs3211938 can impact RDW, we identified one heterozygous sample among a collection of ex vivo differentiated human erythroblasts.32 In erythroblasts from this donor, we noted a strong allelic imbalance (Figure 2). Importantly, this result was confirmed in independent samples from the GTex dataset (Figure S5). At the molecular level, this CD36 expression phenotype could be explained by nonsensemediated mRNA decay or the regulatory effect of non-coding genetic variants in LD with rs3211938. We observed that many new RBC variants are pleiotropic, being often associated with more than one RBC index as well as with WBC, PLT, and lipid traits (Figure 3). These shared effects could imply that the underlying causal genes at these RBC loci generally controlled blood cell proliferation or modulate inflammatory responses. An additional explanation for the link between RBC traits and lipid variants might be the cholesterol content of RBC membranes. As mentioned earlier, RBC corresponds to a large fraction (~25%) of the cells found in the human body. Genetic variation that modulates RBC count or volume could impact circulating lipid levels. In support of this hypothesis, it has been observed that a thalassemia allele is strongly associated with cholesterol levels in the Sardinian population.65 In total, we found ten loci associated with lipid levels and RBC indices, including four novel RBC variants (AFF1, TMEM57-RHD, TRIB1, HNF4A) (Figure 3). In summary, our multi-ethnic meta-analyses have expanded the genetic knowledge of erythrocyte biology and identified new common, low-frequency, and rare RBC variants. Many of the new RBC variants are pleiotropic, affecting other complex traits such as WBC, PLT, and blood lipid levels. Although our report demonstrates the utility of the ExomeChip for genetic discovery, it also highlights the challenge to attribute gene causality based only on association results. This is particularly evident for loci with common variants, for which coding and non-coding markers are often statistically equivalent. For instance, we found no examples of RBC coding variants that entirely explain RBC GWAS signals among the seven loci that had both a sentinel GWAS variant and ExomeChip coding markers. Although increasing sample sizes will continue to yield additional RBC loci, it has become incredibly clear that only a combination of well-powered genetic studies, transcriptomic and epigenomic surveys, and functional experiments (e.g., using genome editing) will ultimately pinpoint causal variants and genes that control RBC phenotypes. Supplemental Data Supplemental Data include a note on the eQTL analyses, information on supplementary funding, five figures, and eight tables and can be found with this article online at http://dx.doi.org/10.1016/ j.ajhg.2016.05.007. The American Journal of Human Genetics 99, 8–21, July 7, 2016 17 Table 3. Overlap of Red Blood Cell Markers with Other Blood Cell Traits and/or Lipid SNP Position A1/A2 AF (A2) Annotation Gene Trait rs10903129 1: 25,768,937 A/G 0.568 intron TMEM57-RHD RDW 0.037 1.19 3 10 7 TC46 0.061 5.40 3 10 10 PLT 0.021 7.06 3 10 6 HCT* 0.028 2.38 3 10 9 HGB* 0.026 3.76 3 10 9 PLT 0.031 5.93 3 10 10 HGB 0.034 3.97 3 10 13 TG40 0.031 1.00 3 10 18 BASO 0.030 1.99 3 10 5 RDW 0.036 1.53 3 10 7 TG40 0.076 1.00 3 10 7 HGB 0.071 1.65 3 10 8 PLT 0.095 7.03 3 10 14 TG52 0.090 1.40 3 10 17 MCV 0.040 1.25 3 10 8 RDW 0.043 5.79 3 10 8 BASO 0.051 1.35 3 10 10 MONO 0.033 3.57 3 10 5 HCT 0.083 1.44 3 10 8 HGB 0.073 2.53 3 10 8 HDL40 0.127 2.00 3 10 34 rs3772219 rs442177 rs2954029 rs55707100 rs4911241 rs1800961 3: 56,771,251 4: 88,030,261 8: 126,490,972 15: 43,820,717 20: 31,140,165 20: 43,042,364 A/C G/T A/T C/T C/T C/T 0.338 0.595 0.439 0.033 0.241 0.032 missense ARHGEF3 intron AFF1 intergenic missense intron missense TRIB1 MAP1A NOL4L HNF4A Beta p Value Shown here are significant novel variants from the RBC traits association analyses that overlap with other blood-cell traits or with lipids. Results for the white blood cell and platelet traits are from the Blood Cell Consortium, and results for lipids are from the published literature. Results are presented for European-ancestry individuals, except in the presence of an asterisk (*), which stands for result from ‘‘All’’ ancestry. The allele frequency and direction of the effect (beta) is given for the A2 allele. Abbreviations are as follows: A1, reference allele; A2, alternate allele; AF, allele frequency; HCT, hematocrit; HGB, hemoglobin; MCV, mean corpuscular volume; RDW, red blood cell distribution width; TC, total cholesterol; PLT, platelet; TG, triglycerides; WBC, white blood cells; BASO, basophils; MONO, monocytes; HDL, HDL cholesterol. Acknowledgments We thank all participants, staff, and study coordinating centers. We also thank Raymond Doty and Jan Abkowitz for discussion of the ALAS2 finding. We would like to thank Liling Warren for contributions to the genetic analysis of the SOLID-TIMI-52 and STABILITY datasets. Young Finns Study (YFS) acknowledges the expert technical assistance in the statistical analyses by Ville Aalto and Irina Lisinen. Estonian Genome Center, University of Tartu (EGCUT) thanks co-workers at the Estonian Biobank, especially Mr. V. Soo, Mr. S. Smith, and Dr. L. Milani. Airwave thanks Louisa Cavaliero who assisted in data collection and management as well as Peter McFarlane and the Glasgow CARE, Patricia Munroe at Queen Mary University of London, and Joanna Sarnecka and Ania Zawodniak at Northwick Park for their contributions to the study. This work was supported by the Fonds de Recherche du Queébec-Santeé (FRQS, scholarship to N.C.), the Canadian Institute of Health Research (Banting-CIHR, scholarship to S.L. and operating grant MOP#123382 to G.L.), and the Canada Research Chair program (to G.L.). P.L.A. was supported by NHLBI R21 HL121422-02. N.A.A. is funded by NIH DK060022. A.N. was supported by the Yoshida Scholarship Foundation. S.K. was supported by a Research Scholar award from the 18 The American Journal of Human Genetics 99, 8–21, July 7, 2016 Massachusetts General Hospital (MGH), the Howard Goodman Fellowship from MGH, the Donovan Family Foundation, R01HL107816, and a grant from Fondation Leducq. Additional acknowledgments and funding information is provided in the Supplemental Data. Received: February 18, 2016 Accepted: May 3, 2016 Published: June 23, 2016 Web Resources BCX ExomeChip association results, http://www.mhihumangenetics.org/en/resources CheckVCF, https://github.com/zhanxw/checkVCF ClinVar, https://www.ncbi.nlm.nih.gov/clinvar/ OMIM, http://www.omim.org/ RareMETALS, http://genome.sph.umich.edu/wiki/RareMETALS RareMetalWorker, http://genome.sph.umich.edu/wiki/ RAREMETALWORKER RvTests, http://genome.sph.umich.edu/wiki/RvTests Type 2 Diabetes Genetics, http://www.type2diabetesgenetics.org/ References 1. Ulker, P., Sati, L., Celik-Ozenci, C., Meiselman, H.J., and Baskurt, O.K. (2009). Mechanical stimulation of nitric oxide synthesizing mechanisms in erythrocytes. Biorheology 46, 121–132. 2. Jiang, N., Tan, N.S., Ho, B., and Ding, J.L. (2007). Respiratory protein-generated reactive oxygen species as an antimicrobial strategy. Nat. Immunol. 8, 1114–1122. 3. Schnabel, R.B., Baumert, J., Barbalic, M., Dupuis, J., Ellinor, P.T., Durda, P., Dehghan, A., Bis, J.C., Illig, T., Morrison, A.C., et al. (2010). Duffy antigen receptor for chemokines (Darc) polymorphism regulates circulating concentrations of monocyte chemoattractant protein-1 and other inflammatory mediators. Blood 115, 5289–5299. 4. Colin, Y., Le Van Kim, C., and El Nemer, W. (2014). Red cell adhesion in human diseases. Curr. Opin. Hematol. 21, 186–192. 5. Whelihan, M.F., and Mann, K.G. (2013). The role of the red cell membrane in thrombin generation. Thromb. Res. 131, 377–382. 6. Brugnara, C. (2003). Iron deficiency and erythropoiesis: new diagnostic approaches. Clin. Chem. 49, 1573–1578. 7. Huang, Y.L., Hu, Z.D., Liu, S.J., Sun, Y., Qin, Q., Qin, B.D., Zhang, W.W., Zhang, J.R., Zhong, R.Q., and Deng, A.M. (2014). Prognostic value of red blood cell distribution width for patients with heart failure: a systematic review and metaanalysis of cohort studies. PLoS ONE 9, e104861. 8. Nada, A.M. (2015). Red cell distribution width in type 2 diabetic patients. Diabetes Metab. Syndr. Obes. 8, 525–533. 9. Zalawadiya, S.K., Zmily, H., Farah, J., Daifallah, S., Ali, O., and Ghali, J.K. (2011). Red cell distribution width and mortality in predominantly African-American population with decompensated heart failure. J. Card. Fail. 17, 292–298. 10. Zalawadiya, S.K., Veeranna, V., Panaich, S.S., and Afonso, L. (2012). Red cell distribution width and risk of peripheral artery disease: analysis of National Health and Nutrition Examination Survey 1999-2004. Vasc. Med. 17, 155–163. 11. Patel, K.V., Semba, R.D., Ferrucci, L., Newman, A.B., Fried, L.P., Wallace, R.B., Bandinelli, S., Phillips, C.S., Yu, B., Connelly, S., et al. (2010). Red cell distribution width and mortality in older adults: a meta-analysis. J. Gerontol. A Biol. Sci. Med. Sci. 65, 258–265. 12. Patel, H.H., Patel, H.R., and Higgins, J.M. (2015). Modulation of red blood cell population dynamics is a fundamental homeostatic response to disease. Am. J. Hematol. 90, 422–428. 13. Whitfield, J.B., and Martin, N.G. (1985). Genetic and environmental influences on the size and number of cells in the blood. Genet. Epidemiol. 2, 133–144. 14. Pilia, G., Chen, W.M., Scuteri, A., Orrú, M., Albai, G., Dei, M., Lai, S., Usala, G., Lai, M., Loi, P., et al. (2006). Heritability of cardiovascular and personality traits in 6,148 Sardinians. PLoS Genet. 2, e132. 15. Evans, D.M., Frazer, I.H., and Martin, N.G. (1999). Genetic and environmental causes of variation in basal levels of blood cells. Twin Res. 2, 250–257. 16. Lin, J.P., O’Donnell, C.J., Jin, L., Fox, C., Yang, Q., and Cupples, L.A. (2007). Evidence for linkage of red blood cell size and count: genome-wide scans in the Framingham Heart Study. Am. J. Hematol. 82, 605–610. 17. van der Harst, P., Zhang, W., Mateo Leach, I., Rendon, A., Verweij, N., Sehmi, J., Paul, D.S., Elling, U., Allayee, H., Li, X., 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. et al. (2012). Seventy-five genetic loci influencing the human red blood cell. Nature 492, 369–375. Chen, Z., Tang, H., Qayyum, R., Schick, U.M., Nalls, M.A., Handsaker, R., Li, J., Lu, Y., Yanek, L.R., Keating, B., et al.; BioBank Japan Project; CHARGE Consortium (2013). Genome-wide association analysis of red blood cell traits in African Americans: the COGENT Network. Hum. Mol. Genet. 22, 2529–2538. Auer, P.L., Teumer, A., Schick, U., O’Shaughnessy, A., Lo, K.S., Chami, N., Carlson, C., de Denus, S., Dubé, M.P., Haessler, J., et al. (2014). Rare and low-frequency coding variants in CXCR2 and other genes are associated with hematological traits. Nat. Genet. 46, 629–634. Eicher, J.D., Chami, N., Kacprowski, T., Nomura, A., Chen, M.-H., Yanek, L.R., Tajuddin, S.M., Schick, U.M., Slater, A.J., Pankratz, N., et al. (2016). Platelet-related variants identified by exomechip meta-analysis in 157,293 individuals. Am. J. Hum. Genet. 99, this issue, 40–55. Tajuddin, S.M., Schick, U.M., Eicher, J.D., Chami, N., Giri, A., Brody, J.A., Hill, W.D., Kacprowski, T., Li, J., Lyytikäinen, L.-P., et al. (2016). Large-scale exome-wide association analysis identifies loci for white blood cell traits and pleiotropy with immune-mediated diseases. Am. J. Hum. Genet. 99, this issue, 22–39. Grove, M.L., Yu, B., Cochran, B.J., Haritunians, T., Bis, J.C., Taylor, K.D., Hansen, M., Borecki, I.B., Cupples, L.A., Fornage, M., et al. (2013). Best practices and joint calling of the HumanExome BeadChip: the CHARGE Consortium. PLoS ONE 8, e68095. Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M.A., Bender, D., Maller, J., Sklar, P., de Bakker, P.I., Daly, M.J., and Sham, P.C. (2007). PLINK: a tool set for wholegenome association and population-based linkage analyses. Am. J. Hum. Genet. 81, 559–575. Wells, Q.S., Becker, J.R., Su, Y.R., Mosley, J.D., Weeke, P., D’Aoust, L., Ausborn, N.L., Ramirez, A.H., Pfotenhauer, J.P., Naftilan, A.J., et al. (2013). Whole exome sequencing identifies a causal RBM20 mutation in a large pedigree with familial dilated cardiomyopathy. Circ Cardiovasc Genet 6, 317–326. Winkler, T.W., Day, F.R., Croteau-Chonka, D.C., Wood, A.R., Locke, A.E., Mägi, R., Ferreira, T., Fall, T., Graff, M., Justice, A.E., et al.; Genetic Investigation of Anthropometric Traits (GIANT) Consortium (2014). Quality control and conduct of genome-wide association meta-analyses. Nat. Protoc. 9, 1192–1212. Limongelli, G., Elliott, P., Charron, P., Mogensen, J., and McKeown, P.P. (2012). Approaching genetic testing in cardiomyopathies (ESC Council for Cardiology Practice). Olson, T.M., Michels, V.V., Thibodeau, S.N., Tai, Y.S., and Keating, M.T. (1998). Actin mutations in dilated cardiomyopathy, a heritable form of heart failure. Science 280, 750–752. Liu, D.J., Peloso, G.M., Zhan, X., Holmen, O.L., Zawistowski, M., Feng, S., Nikpay, M., Auer, P.L., Goel, A., Zhang, H., et al. (2014). Meta-analysis of gene-level tests for rare variant association. Nat. Genet. 46, 200–204. Wu, M.C., Lee, S., Cai, T., Li, Y., Boehnke, M., and Lin, X. (2011). Rare-variant association testing for sequencing data with the sequence kernel association test. Am. J. Hum. Genet. 89, 82–93. Price, A.L., Kryukov, G.V., de Bakker, P.I., Purcell, S.M., Staples, J., Wei, L.J., and Sunyaev, S.R. (2010). Pooled association tests The American Journal of Human Genetics 99, 8–21, July 7, 2016 19 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. for rare variants in exon-resequencing studies. Am. J. Hum. Genet. 86, 832–838. Willer, C.J., Li, Y., and Abecasis, G.R. (2010). METAL: fast and efficient meta-analysis of genomewide association scans. Bioinformatics 26, 2190–2191. Lessard, S., Beaudoin, M., Benkirane, K., and Lettre, G. (2015). Comparison of DNA methylation profiles in human fetal and adult red blood cell progenitors. Genome Med. 7, 1. Lappalainen, T., Sammeth, M., Friedländer, M.R., ’t Hoen, P.A., Monlong, J., Rivas, M.A., Gonzàlez-Porta, M., Kurbatova, N., Griebel, T., Ferreira, P.G., et al.; Geuvadis Consortium (2013). Transcriptome and genome sequencing uncovers functional variation in humans. Nature 501, 506–511. Zhang, X., Gierman, H.J., Levy, D., Plump, A., Dobrin, R., Goring, H.H., Curran, J.E., Johnson, M.P., Blangero, J., Kim, S.K., et al. (2014). Synthesis of 53 tissue and cell line expression QTL datasets reveals master eQTLs. BMC Genomics 15, 532. Astner, I., Schulze, J.O., van den Heuvel, J., Jahn, D., Schubert, W.D., and Heinz, D.W. (2005). Crystal structure of 5-aminolevulinate synthase, the first enzyme of heme biosynthesis, and its link to XLSA in humans. EMBO J. 24, 3166–3177. Halpain, S., and Dehmelt, L. (2006). The MAP1 family of microtubule-associated proteins. Genome Biol. 7, 224. Takei, Y., Kikkawa, Y.S., Atapour, N., Hensch, T.K., and Hirokawa, N. (2015). Defects in synaptic plasticity, reduced NMDA-receptor transport, and instability of postsynaptic density proteins in mice lacking microtubule-associated protein 1A. J. Neurosci. 35, 15539–15554. GTEx Consortium (2015). Human genomics. The GenotypeTissue Expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science 348, 648–660. Dehghan, A., Dupuis, J., Barbalic, M., Bis, J.C., Eiriksdottir, G., Lu, C., Pellikka, N., Wallaschofski, H., Kettunen, J., Henneman, P., et al. (2011). Meta-analysis of genome-wide association studies in >80 000 subjects identifies multiple loci for C-reactive protein levels. Circulation 123, 731–738. Willer, C.J., Schmidt, E.M., Sengupta, S., Peloso, G.M., Gustafsson, S., Kanoni, S., Ganna, A., Chen, J., Buchkovich, M.L., Mora, S., et al.; Global Lipids Genetics Consortium (2013). Discovery and refinement of loci associated with lipid levels. Nat. Genet. 45, 1274–1283. Taylor, K.C., Lange, L.A., Zabaneh, D., Lange, E., Keating, B.J., Tang, W., Smith, N.L., Delaney, J.A., Kumari, M., Hingorani, A., et al. (2011). A gene-centric association scan for Coagulation Factor VII levels in European and African Americans: the Candidate Gene Association Resource (CARe) Consortium. Hum. Mol. Genet. 20, 3525–3534. de Vries, P.S., Chasman, D.I., Sabater-Lleal, M., Chen, M.H., Huffman, J.E., Steri, M., Tang, W., Teumer, A., Marioni, R.E., Grossmann, V., et al. (2016). A meta-analysis of 120 246 individuals identifies 18 new loci for fibrinogen concentration. Hum. Mol. Genet. 25, 358–370. Kooner, J.S., Saleheen, D., Sim, X., Sehmi, J., Zhang, W., Frossard, P., Been, L.F., Chia, K.S., Dimas, A.S., Hassanali, N., et al.; DIAGRAM; MuTHER (2011). Genome-wide association study in individuals of South Asian ancestry identifies six new type 2 diabetes susceptibility loci. Nat. Genet. 43, 984–989. GTEx Consortium (2013). The Genotype-Tissue Expression (GTEx) project. Nat. Genet. 45, 580–585. Pradel, L.C., Vanhille, L., and Spicuglia, S. (2015). [The European Blueprint project: towards a full epigenome characterization of the immune system]. Med. Sci. (Paris) 31, 236–238. 20 The American Journal of Human Genetics 99, 8–21, July 7, 2016 46. Aulchenko, Y.S., Ripatti, S., Lindqvist, I., Boomsma, D., Heid, I.M., Pramstaller, P.P., Penninx, B.W., Janssens, A.C., Wilson, J.F., Spector, T., et al.; ENGAGE Consortium (2009). Loci influencing lipid levels and coronary heart disease risk in 16 European population cohorts. Nat. Genet. 41, 47–55. 47. Kullo, I.J., Ding, K., Shameer, K., McCarty, C.A., Jarvik, G.P., Denny, J.C., Ritchie, M.D., Ye, Z., Crosslin, D.R., Chisholm, R.L., et al. (2011). Complement receptor 1 gene variants are associated with erythrocyte sedimentation rate. Am. J. Hum. Genet. 89, 131–138. 48. Döbbeling, U. (1997). The effects of cyclosporin A on V(D)J recombination activity. Scand. J. Immunol. 45, 494–498. 49. Zarza, R., Alvarez, R., Pujades, A., Nomdedeu, B., Carrera, A., Estella, J., Remacha, A., Sánchez, J.M., Morey, M., Cortes, T., et al.; Red Cell Pathology Group of the Spanish Society of Haematology (AEHH) (1998). Molecular characterization of the PK-LR gene in pyruvate kinase deficient Spanish patients. Br. J. Haematol. 103, 377–382. 50. Valentini, G., Chiarelli, L.R., Fortin, R., Dolzan, M., Galizzi, A., Abraham, D.J., Wang, C., Bianchi, P., Zanella, A., and Mattevi, A. (2002). Structure and function of human erythrocyte pyruvate kinase. Molecular basis of nonspherocytic hemolytic anemia. J. Biol. Chem. 277, 23807–23814. 51. Van Sligtenhorst, I., Ding, Z.M., Shi, Z.Z., Read, R.W., Hansen, G., and Vogel, P. (2012). Cardiomyopathy in a-kinase 3 (ALPK3)-deficient mice. Vet. Pathol. 49, 131–141. 52. Peloso, G.M., Auer, P.L., Bis, J.C., Voorman, A., Morrison, A.C., Stitziel, N.O., Brody, J.A., Khetarpal, S.A., Crosby, J.R., Fornage, M., et al.; NHLBI GO Exome Sequencing Project (2014). Association of low-frequency and rare codingsequence variants with blood lipids and coronary heart disease in 56,000 whites and blacks. Am. J. Hum. Genet. 94, 223–232. 53. Gu, Y., Nakamura, T., Alder, H., Prasad, R., Canaani, O., Cimino, G., Croce, C.M., and Canaani, E. (1992). The t(4;11) chromosome translocation of human acute leukemias fuses the ALL-1 gene, related to Drosophila trithorax, to the AF-4 gene. Cell 71, 701–708. 54. Guastadisegni, M.C., Lonoce, A., Impera, L., Di Terlizzi, F., Fugazza, G., Aliano, S., Grasso, R., Cluzeau, T., Raynaud, S., Rocchi, M., and Storlazzi, C.T. (2010). CBFA2T2 and C20orf112: two novel fusion partners of RUNX1 in acute myeloid leukemia. Leukemia 24, 1516–1519. 55. Serbanovic-Canic, J., Cvejic, A., Soranzo, N., Stemple, D.L., Ouwehand, W.H., and Freson, K. (2011). Silencing of RhoA nucleotide exchange factor, ARHGEF3, reveals its unexpected role in iron uptake. Blood 118, 4967–4976. 56. Okumura, N., Tsuji, K., and Nakahata, T. (1992). Changes in cell surface antigen expressions during proliferation and differentiation of human erythroid progenitors. Blood 80, 642–650. 57. Kiefer, C.R., and Snyder, L.M. (2000). Oxidation and erythrocyte senescence. Curr. Opin. Hematol. 7, 113–116. 58. Nicholson, A.C., Han, J., Febbraio, M., Silversterin, R.L., and Hajjar, D.P. (2001). Role of CD36, the macrophage class B scavenger receptor, in atherosclerosis. Ann. N Y Acad. Sci. 947, 224–228. 59. Auer, P.L., Johnsen, J.M., Johnson, A.D., Logsdon, B.A., Lange, L.A., Nalls, M.A., Zhang, G., Franceschini, N., Fox, K., Lange, E.M., et al. (2012). Imputation of exome sequence variants into population- based samples and blood-cell-trait-associated loci in African Americans: NHLBI GO Exome Sequencing Project. Am. J. Hum. Genet. 91, 794–808. 60. Elbers, C.C., Guo, Y., Tragante, V., van Iperen, E.P., Lanktree, M.B., Castillo, B.A., Chen, F., Yanek, L.R., Wojczynski, M.K., Li, Y.R., et al. (2012). Gene-centric meta-analysis of lipid traits in African, East Asian and Hispanic populations. PLoS ONE 7, e50198. 61. Ayodo, G., Price, A.L., Keinan, A., Ajwang, A., Otieno, M.F., Orago, A.S., Patterson, N., and Reich, D. (2007). Combining evidence of natural selection with association analysis increases power to detect malaria-resistance variants. Am. J. Hum. Genet. 81, 234–242. 62. Aitman, T.J., Cooper, L.D., Norsworthy, P.J., Wahid, F.N., Gray, J.K., Curtis, B.R., McKeigue, P.M., Kwiatkowski, D., Greenwood, B.M., Snow, R.W., et al. (2000). Malaria susceptibility and CD36 mutation. Nature 405, 1015–1016. 63. Bhatia, G., Patterson, N., Pasaniuc, B., Zaitlen, N., Genovese, G., Pollack, S., Mallick, S., Myers, S., Tandon, A., Spencer, C., et al. (2011). Genome-wide comparison of African-ancestry populations from CARe and other cohorts reveals signals of natural selection. Am. J. Hum. Genet. 89, 368–381. 64. Auton, A., Brooks, L.D., Durbin, R.M., Garrison, E.P., Kang, H.M., Korbel, J.O., Marchini, J.L., McCarthy, S., McVean, G.A., and Abecasis, G.R.; 1000 Genomes Project Consortium (2015). A global reference for human genetic variation. Nature 526, 68–74. 65. Sidore, C., Busonero, F., Maschio, A., Porcu, E., Naitza, S., Zoledziewska, M., Mulas, A., Pistis, G., Steri, M., Danjou, F., et al. (2015). Genome sequencing elucidates Sardinian genetic architecture and augments association analyses for lipid and blood inflammatory markers. Nat. Genet. 47, 1272–1281. The American Journal of Human Genetics 99, 8–21, July 7, 2016 21