Association of Parkinson Disease with Structural and Regulatory Variants in the HLA Region

• DOI: 10.1016/j.ajhg.2013.10.009
• Publication Date: 2013-10-31T17:16:55Z
• Authors: William T. Wissemann; Erin M. Hill-Burns; Cyrus P. Zabetian; Stewart A. Factor; Nikolaos Patsopoulos; Bryan Hoglund; Cherie Holcomb; Ryan J. Donahue; Glenys Thomson; Henry Erlich; Haydeh Payami
• Keywords: Genome-wide Association Study; Imputation (statistics); Genetic Association; SNP

Historically, association of disease with the major histocompatibility complex (HLA) genes has been tested HLA alleles that encode antigen-binding affinity. The Parkinson (PD), however, was discovered noncoding SNPs in a genome-wide study (GWAS). We show here several HLA-region have since associated PD form two blocks tagged by rs3129882 (p = 9 × 10−11) and rs9268515 and/or rs2395163 3 10−11). investigated whether these SNP-associations were driven HLA-alleles at adjacent loci. imputed class I II for 2000 cases 1986 controls from NeuroGenetics Research Consortium GWAS sequenced subset 194 204 controls. therefore able to assess accuracy imputation algorithms against next-generation-sequencing while taking advantage larger data sets study. Additionally, we 843 856 another replication. risk positively B∗07:02_C∗07:02_DRB5∗01_DRB1∗15:01_DQA1∗01:02_DQB1∗06:02 haplotype negatively C∗03:04, DRB1∗04:04 DQA1∗03:01 alleles. lost significance when conditioned on SNPs, but C∗03:04 (OR 0.72, p 8 10−6) 0.65, 4 10−5) remained significant. Similarly, closely linked significant irrespective are expression quantitative trait loci (eQTLs) HLA-DR HLA-DQ (9 10−5 ≥ PeQTL 2 10−79), suggesting gene might influence PD. Our suggest is both structural regulatory elements HLA. Furthermore, our demonstrates region can be alleles, observed associations sometimes secondary variant. one most regions human genome because it high density genes, extreme polymorphism, complicated patterns linkage disequilibrium (LD) (Figure 1). molecules bind antigens present them T cell receptors, function critical adaptive immune responses. numerous autoimmune, infectious, malignant, neurologic disorders. These established classical i.e., those variations affect However, Parkinson's (PD [MIM 168600]) (GWAS).1Hamza T.H. Zabetian C.P. Tenesa A. Laederach Montimurro J. Yearout D. Kay D.M. Doheny K.F. Paschall Pugh E. et al.Common genetic variation late-onset sporadic disease.Nat. Genet. 2010; 42: 781-785Crossref PubMed Scopus (585) Google Scholar Subsequently, other GWASs SNP-based studies confirmed identified additional strongly associate PD.2Nalls M.A. Plagnol V. Hernandez D.G. Sharma M. Sheerin U.M. Saad Simón-Sánchez Schulte C. Lesage S. Sveinbjörnsdóttir al.International Disease Genomics ConsortiumImputation sequence variants identification risks disease: meta-analysis studies.Lancet. 2011; 377: 641-649Abstract Full Text PDF (714) Scholar, 3Hill-Burns E.M. Factor S.A. Thomson G. Payami H. Evidence more than disease-associated variant within region.PLoS ONE. 6: e27109Crossref (58) 4Pankratz N. Beecham G.W. DeStefano A.L. Dawson T.M. Hamza Hung A.Y. Hyman B.T. Ivinson A.J. al.PD ConsortiumMeta-analysis novel locus, RIT2.Ann. Neurol. 2012; 71: 370-384Crossref (206) 5Ahmed I. Tamouza R. Delord Krishnamoorthy Tzourio Mulot Nacfer Lambert J.C. Beaune P. Laurent-Puig al.Association between HLA-DRB1 locus.Mov. Disord. 27: 1104-1110Crossref (92) 6Simón-Sánchez van Hilten J.J. de Warrenburg B. Post Berendse H.W. Arepalli Bie R.M. Velseboer Scheffer al.Genome-wide confirms extant among Dutch.Eur. Hum. 19: 655-661Crossref (151) original PD-associated SNP,1Hamza rs3129882, intron 1 HLA-DRA (MIM 142860). also map intergenicaly near HLA-DRA, 142857), HLA-DRB5 604776) It possible tag An polymorphisms quite plausible evidence neuroinflammation pathogenesis.7Mosley R.L. Hutter-Saunders J.A. Stone D.K. Gendelman H.E. Inflammation immunity disease.Cold Spring Harb Perspect Med. 2: a009381Crossref (184) few attempted had small sample sizes produced discordant results.5Ahmed 8Saiki Baker Williams-Gray C.H. Foltynie T. Goodman R.S. Taylor C.J. Compston D.A. Barker R.A. Sawcer S.J. Goris Association leucocyte antigen susceptibility disease.J. Neurosurg. Psychiatry. 81: 890-891Crossref (47) unknown specific or represent (eQTLs). aim determine To end, used sets: NGRC,1Hamza which uncovered (2,000 1,986 controls), PROGENI-GenePD9Pankratz Wilk J.B. Latourelle Halter E.W. Gusella J.F. Nichols W.C. Foroud Myers R.H. PSG-PROGENI GenePD Investigators, Coordinators Molecular Genetic LaboratoriesGenomewide contributing familial disease.Hum. 2009; 124: 593-605Crossref (354) (843 controls) replication (NGRC PROGENI-GenePD available dbGaP [search terms CIDR parkinson latter]; see Web Resources). Both composed unrelated European descent described detail elsewhere.1Hamza 9Pankratz In sets, neurologists diagnosed using standard research criteria10Hughes Daniel S.E. Ben-Shlomo Y. Lees diagnosis parkinsonian syndromes specialist movement disorder service.Brain. 2002; 125: 861-870Crossref (968) found late onset (average age 58 ± 11 years NGRC, 62 set), representing common idiopathic disease. website undergone careful quality control, including in-depth investigation population structure.1Hamza A concern principal components ruled out.1Hamza approved institutional review boards participating institutions, proper informed consent obtained. first task generate NGRC sets. Next-generation sequencing (NGS)11Holcomb C.L. Höglund Anderson M.W. Blake L.A. Böhme Egholm Ferriola Gabriel Gelber Goodridge al.A multi-site high-resolution genotyping next generation sequencing.Tissue Antigens. 77: 206-217Crossref (99) gold prohibitively costly large-scale studies. entire cohorts exons NGRC. NGS Two recently developed imputing SNP data: SNP2HLA12Jia X. Han Onengut-Gumuscu Chen W.M. Concannon P.J. Rich S.S. Raychaudhuri Bakker P.I. Imputing amino acid leukocyte antigens.PLoS 2013; 8: e64683Crossref (429) HLA∗IMP.13Dilthey A.T. Moutsianas L. Leslie McVean HLA∗IMP—an integrated framework genotypes.Bioinformatics. 968-972Crossref (124) Because there not yet head-to-head comparison performance methods, applied methods same set (NGRC). use extended impute 4-digit resolution (A 142800], B 142830], C 142840]) (DRB1, DQA1 146880], DQB1 604305], DPB1 142858]) DRB3 612735), DRB4, DRB5, chromosomes, HLA∗IMP digits. effective N 2,000 set. For NGS, DNA whole blood exon DRB1, DRB3, DQA1, an amplicon-sequencing strategy "fusion" primers Roche 454 GS FLX instrument Conexio software as elsewhere.11Holcomb Calls made higher. nine individuals, highly discordant, mix-up plating. After removal assessed comparing generated each algorithm NGS. individual, best-guess compared NGS-called designated matches mismatches. measured sensitivity (SS: probability calling allele being correctly) specificity (SP: absent correctly). Table S1 (in Supplemental Data this article online) shows SNP2HLA HLA∗IMP, side-by-side, generally high, line values reported authors algorithms.12Jia 13Dilthey DQB1, better case papers.12Jia trouble some (e.g., DRB1∗01:03) distinguishing certain DRB1 DRB1∗14 ∗08 alleles) conversion. DPB1∗10:01 DRB1∗14:01), did well poorly. no reason suspect genes; find HLA-C allele, should note papers >97% HLA-C.12Jia size caveat will require scrutiny. later reliable. Next, revealed PD, strong nonfamilial (simplex) PD.1Hamza Therefore, testing, excluded 435 who positive family history. analyzed genotypes 1,565 simplex controls, once dosage probabilities again calls HLA∗IMP. frequency <1%. logistic regression R calculate odds ratios (ORs), errors (SEs), values. All tests adjusted (PC1 PC2).1Hamza Additional covariates presence absence haplotypes, genotypes) added conditional analyses. two-sided tests, without multiple-testing correction. sure miss any explain associations, threshold low, uncorrected < 0.05, minimize false negatives. Alleles 0.05 carried Analyses similar except Bonferroni-corrected 10−3 (for 14 replication). Replication testing PC1 PC2 (and repeated PC4 nominally significant; results unchanged). Lastly, combined Meta-analysis conducted PLINK software.14Purcell Neale Todd-Brown K. Thomas Ferreira Bender Maller Sklar Daly M.J. Sham P.C. PLINK: tool whole-genome population-based analyses.Am. 2007; 559-575Abstract (19634) gave From on, unless noted, data. Of 12 replicated second (Table 1); included C, B, loci, DRB4∗01 DRB5∗01. Six increased six reduced LD, constructed haplotypes method implemented BEAGLE inferring individuals15Browning S.R. Browning B.L. Rapid accurate phasing missing-data inference localized clustering.Am. 1084-1097Abstract (2008) calculated r2 D′, either Haploview v. 4.2,16Barrett Fry Haploview: analysis visualization LD maps.Bioinformatics. 2005; 21: 263-265Crossref (12023) hand. "risk" forming haplotype: S2, ORNGRC 1.29, pNGRC 10−3; ORreplication 1.38, preplication 0.02; ORmeta 1.32 pmeta 10−4). "protective" formed haplotype, B∗40:01_C∗03:04_DRB4∗01_DRB1∗04:04_DQA1∗03:01_DQB1∗03:02 0.60, 0.52, 0.06; 0.58 10−3). protective analysis, where adjustment five three seemingly independent signals detected S3). summarize thus far, four variants: (We SNPs.)Table 1Association AllelesHLA AllelesAllele Frequency NGRCNGRCReplicationMeta-AnalysisPDControlORpORpORpB∗07:020.150.131.208 10−31.310.011.233 10−4B∗40:010.050.060.790.020.710.030.762 10−3C∗03:040.070.090.762 10−30.636 10−40.728 10−6C∗07:020.160.141.223 10−31.250.031.232 10−4DRB1∗04:040.030.050.645 10−40.670.030.654 10−5DRB1∗15:010.160.131.269 10−41.260.021.266 10−5DRB4∗010.260.290.853 10−30.804 10−30.834 10−5DRB5∗010.160.141.258 10−41.250.031.255 10−5DQA1∗01:020.220.201.150.021.220.021.171 10−3DQA1∗03:010.150.190.771 10−40.763 10−30.771 10−6DQB1∗03:020.090.110.753 10−40.737 10−30.747 10−6DQB1∗06:020.160.131.267 10−41.270.021.264 10−5Class 0.01 Fourteen Twelve shown. Tests under additive model 2. (including set) sided multiple testing. 10−3. Open table new tab Class following PD: rs3129882,1Hamza rs75855844,2Nalls rs9268515,3Hill-Burns rs2395163,4Pankratz rs660895,5Ahmed rs42481666Simón-Sánchez 2). rs660895, rs2395163, rs4248166 genotyped arrays into (info_score 0.96) IMPUTE2.2.2 software17Howie B.N. Donnelly Marchini flexible genotype studies.PLoS 5: e1000529Crossref (2793) 1000 Genomes phase (release v3, April 2012) reference panel. unable rs75855844 (this I, removed ImmunoChip technical filtering); best proxies, rs41269955 0.99) rs77539933 0.86) (r2 other, 0.75 Pilot set, pursue further. structure 1) suggests correlated (0.06 ≤ 0.16). Whereas rs9268515, rs41269955, (the latter proxies rs75855844), moderate (0.52 0.77) tagging factor. fact, adjusting abolished rs77539933, rs660895 S4). Thus, elements, rs2395163.Table 2Association RegionPreviously Published PD-Associated SNPsAllele Freqeuncy NGRCPresent StudySNP (Reference)BPLocationGeneOriginal ReportNGRCReplicationMeta-analysisORpPDControlORpORpORprs31298821Hamza Scholar32,409,530intron 1DRA1.313 10−80.470.40original report1.215 10−31.309 10−11rs92685153Hill-Burns Scholar32,379,295intergenicBTNL2-DRA0.804 10−40.150.20original report0.755 10−30.758 10−8rs758558442Nalls Scholar32,480,227intergenicDRA-DRB50.701 10−80.16