Significance
Vitiligo is an autoimmune disease in which spots of white skin and hair result from destruction of melanocytes. Vitiligo is associated with HLA-A*02:01, which presents multiple vitiligo melanocyte autoantigens. We localize vitiligo risk to a SNP haplotype 20 kb downstream of the HLA-A gene, spanning a transcriptional regulatory element. Blood cells from healthy subjects carrying the high-risk haplotype expressed more HLA-A RNA than subjects carrying only nonhigh-risk haplotypes. Vitiligo risk in the MHC class I region thus derives from combined quantitative and qualitative phenomena: an SNP haplotype in a transcriptional regulator that induces elevated expression of HLA-A RNA in vivo, and strong linkage disequilibrium with an HLA-A allele that confers *02:01 specificity. These combine to increase HLA-A2 available to present melanocyte autoantigens.
Keywords: vitiligo, autoimmune disease, HLA, transcription, enhancer
Abstract
HLA-A is a class I major histocompatibility complex receptor that presents peptide antigens on the surface of most cells. Vitiligo, an autoimmune disease in which skin melanocytes are destroyed by cognate T cells, is associated with variation in the HLA-A gene; specifically HLA-A*02:01, which presents multiple vitiligo melanocyte autoantigens. Refined genetic mapping localizes vitiligo risk in the HLA-A region to an SNP haplotype ∼20-kb downstream, spanning an ENCODE element with many characteristics of a transcriptional enhancer. Convergent CTCF insulator sites flanking the HLA-A gene promoter and the predicted transcriptional regulator, with apparent interaction between these sites, suggests this element regulates the HLA-A promoter. Peripheral blood mononuclear cells from healthy subjects homozygous for the high-risk haplotype expressed 39% more HLA-A RNA than cells from subjects carrying nonhigh-risk haplotypes (P = 0.0048). Similarly, RNAseq analysis of 1,000 Genomes Project data showed more HLA-A mRNA expressed in subjects homozygous for the high-risk allele of lead SNP rs60131261 than subjects homozygous for the low-risk allele (P = 0.006). Reporter plasmid transfection and genomic run-on sequence analyses confirm that the HLA-A transcriptional regulator contains multiple bidirectional promoters, with greatest activity on the high-risk haplotype, although it does not behave as a classic enhancer. Vitiligo risk associated with the MHC class I region thus derives from combined quantitative and qualitative phenomena: a SNP haplotype in a transcriptional regulator that induces gain-of-function, elevating expression of HLA-A RNA in vivo, in strong linkage disequilibrium with an HLA-A allele that confers *02:01 specificity.
Autoimmune diseases comprise more than 80 disorders in which the immune system attacks “self” tissues and cells (1), affecting 3–5% of the United States population (2). Many different autoimmune diseases are genetically associated with variation in the major histocompatibility complex (MHC) on chromosome 6p21.3, including class I loci, class II loci, or both. MHC class I molecules present peptide antigens on the surface of almost all cells, providing targets for autoimmune sensitization and targeting by cytotoxic T cells. Extensive polymorphism of the human classic MHC genes produces great diversity in the corresponding polypeptides, enabling both diversity and specificity in the peptide antigens presented. Transcription of the classic MHC genes is also subject to complex regulation (3), which similarly may be subject to genetic variation. However, contributions of the MHC to autoimmunity have thus far largely focused on MHC antigenic specificity.
Vitiligo is associated with MHC class I region SNPs in the vicinity of the HLA-A gene (4), and DNA sequence analysis identified the high-risk allele as HLA-A*02:01:01:01 (5), encoding the canonical HLA-A2 specificity. HLA-A2 can present a diversity of autoantigens, including several derived from melanocyte proteins that include tyrosinase (6), TRP2 (7), OCA2 (8), MC1R (9), gp100 (10), and MART-1/melan-A (11). In the present study we refined genetic mapping, localizing primary vitiligo risk in the MHC class I region to a SNP haplotype 20 kb downstream of the HLA-A gene, in strong linkage disequilibrium with HLA-A*02:01:01:01. This high-risk SNP haplotype is coincident with a predicted transcriptional regulator, which we find drives elevated HLA-A transcription in peripheral blood mononuclear cells. These findings indicate that vitiligo susceptibility in the MHC class I region involves two functional components: a primary quantitative effect of increased HLA-A expression, and a secondary qualitative effect of *02:01:01:01 antigenic specificity as a result of strong linkage disequilibrium through the region. Together, these features likely combine to increase cell-surface presentation of autoimmune target antigens, facilitating recognition of melanocytes by autoreactive cytotoxic T-cells.
Results
Refined Genetic Mapping of Vitiligo Susceptibility in the MHC Class I Region to a Predicted Transcriptional Regulator.
We previously showed that vitiligo is associated with MHC class I region SNP rs12206499 (4), in strong linkage disequilibrium with HLA-A*02:01:01:01 (5). To more precisely localize causal variation in the region, we compared genotypes of 2,853 European-derived Caucasian (EUR) vitiligo cases and 37,412 controls, imputed through the extended MHC (12, 13) using data from the 1,000 Genomes Project. In the MHC class I region, greatest association was with lead SNP rs60131261 (chr6:29937336–29937339; P = 2.15 × 10−50, odds ratio 1.53). Logistic regression analysis conditional on rs60131261 identified 21 additional variants whose effects could not be distinguished from rs60131261, which together thus comprise the primary MHC class I vitiligo-associated haplotype (Table S1). The 22 variants span a 9.6-kb region (chr6:29928838–29938487) that almost precisely encompasses a striking ENCODE (14) transcriptional element (chr6:29,932,250–29,937,500) located ∼20 kb downstream of HLA-A. As shown in Fig. 1, this element was observed in all cell types tested by ENCODE, and has an open hypomethylated chromatin configuration, multiple DNase I hypersensitivity sites, numerous RNA polymerase II and transcription factor binding sites, and prominent H3K4me1, H3K4me3, and H3K27ac marks. Together, these features are suggestive of an active transcriptional promoter and enhancer (16–18).
Table S1.
SNPs comprising the MHC class I vitiligo-associated haplotype
| SNP | Chr6 nt | A1 | A2 | F_A | F_U | CMH P value | CMH OR | Logistic P value | Logistic odds ratio | CADD PHRED |
| rs2523935 | 29928838 | A | G | 0.38 | 0.29 | 4.82E-47 | 1.52 | 7.38E-50 | 1.53 | 0.75 |
| rs72545948 | 29929588 | D | I | 0.38 | 0.28 | 1.22E-45 | 1.52 | 2.99E-48 | 1.52 | 4.29 |
| rs4713269 | 29930496 | T | A | 0.38 | 0.29 | 4.26E-47 | 1.52 | 6.76E-50 | 1.53 | 1.11 |
| rs2394247 | 29930706 | C | A | 0.38 | 0.29 | 4.91E-47 | 1.52 | 7.61E-50 | 1.53 | 0.34 |
| rs6935024 | 29934080 | T | C | 0.38 | 0.29 | 4.50E-47 | 1.52 | 6.88E-50 | 1.53 | 3.17 |
| rs6935053 | 29934163 | C | A | 0.38 | 0.29 | 4.50E-47 | 1.52 | 6.88E-50 | 1.53 | 1.17 |
| rs4713270 | 29934697 | A | G | 0.38 | 0.29 | 4.50E-47 | 1.52 | 6.88E-50 | 1.53 | 0.40 |
| rs34811773 | 29934904 | D | I | 0.38 | 0.29 | 5.31E-47 | 1.52 | 7.43E-50 | 1.53 | 0.30 |
| rs4248141 | 29935843 | A | G | 0.38 | 0.29 | 5.10E-47 | 1.52 | 7.51E-50 | 1.53 | 0.26 |
| rs4248142 | 29935849 | T | C | 0.38 | 0.29 | 5.10E-47 | 1.52 | 7.51E-50 | 1.53 | 0.21 |
| rs4248143 | 29935891 | T | C | 0.38 | 0.29 | 4.50E-47 | 1.52 | 6.88E-50 | 1.53 | 6.93 |
| rs4959038 | 29936351 | A | G | 0.38 | 0.29 | 4.50E-47 | 1.52 | 6.88E-50 | 1.53 | 1.07 |
| rs12175093 | 29936408 | G | A | 0.38 | 0.29 | 7.02E-47 | 1.52 | 9.47E-50 | 1.52 | 1.19 |
| rs73727620 | 29936655 | G | A | 0.38 | 0.29 | 7.02E-47 | 1.52 | 9.47E-50 | 1.52 | 0.54 |
| rs12193100 | 29936914 | A | G | 0.38 | 0.29 | 6.21E-47 | 1.52 | 8.68E-50 | 1.53 | 12.60 |
| rs60131261 | 29937335 | D | I | 0.38 | 0.29 | 6.03E-48 | 1.53 | 2.15E-50 | 1.53 | 1.37 |
| rs6916451 | 29938174 | T | C | 0.38 | 0.29 | 7.59E-47 | 1.52 | 1.21E-49 | 1.52 | 6.82 |
| rs9378141 | 29938368 | C | A | 0.38 | 0.28 | 8.60E-47 | 1.52 | 5.69E-50 | 1.53 | 6.72 |
| rs9378142 | 29938371 | C | A | 0.38 | 0.29 | 1.19E-46 | 1.52 | 9.10E-50 | 1.53 | 8.22 |
| rs9404959 | 29938377 | A | C | 0.38 | 0.28 | 8.99E-47 | 1.52 | 6.30E-50 | 1.53 | 5.54 |
| rs12195260 | 29938455 | A | G | 0.38 | 0.29 | 8.84E-47 | 1.52 | 1.42E-49 | 1.52 | 10.65 |
| rs12202241 | 29938487 | G | T | 0.38 | 0.29 | 6.62E-47 | 1.52 | 8.45E-50 | 1.53 | 1.04 |
Fig. 1.
Vitiligo association in the HLA-A region of human chromosome 6p. Nucleotide positions, HLA-A transcriptional orientation, and the 22 SNPs that define the vitiligo high-risk haplotype are shown. Layered H3K27Ac, H3K4Me1, and H3K4Me3 marks, hidden Markov model chromatin state segmentation (ChromHMM), DNase I hypersensitive site cluster (DNase I Clusters), and transcription factor chromatin immunoprecipitation sequencing (Txn Factor ChIP-seq) data are from ENCODE (14). For layered H3K27Ac, H3K4Me1, H3K4Me3 marks, data are shown for the seven cell lines studied by ENCODE. For ChromHMM, red indicates active promoters, orange indicates strong enhancers, and blue indicates an insulator. Data shown are for GM12878 lymphoblastoid cells. For DNase clusters, darkness indicates relative signal strength in 125 cell types from ENCODE (V3). For Txn factor ChIP-seq, darkness indicates relative signal strength of aggregate binding of 161 transcription factors, and green bars indicate ENCODE Factorbook (15) canonical motifs for specific transcription factors.
The Vitiligo High-Risk MHC Class I Haplotype Is Associated with Elevated HLA-A RNA Expression.
Localization of primary vitiligo risk in the MHC class I region to an apparent transcriptional regulatory element downstream of the HLA-A gene suggested that corresponding vitiligo risk may be mediated by elevated expression of HLA-A RNA. To test this theory, we compared expression of HLA-A RNA in healthy individuals alternatively homozygous for the high-risk MHC class I region haplotype versus for nonhigh-risk haplotypes. We genotyped 81 unrelated EUR individuals without known autoimmune disease for haplotype tagSNP (Table S1) rs12193100, which is in perfect linkage disequilibrium (r2 = 1.0) with the other SNPs that define the high-risk haplotype, and is in near-perfect linkage disequilibrium (r2 = 0.98) with the original high-risk SNP, rs12206499 (4). Among these healthy subjects, we identified 10 homozygous for the high-risk haplotype and 27 homozygous for nonhigh-risk haplotypes.
To quantitate HLA-A RNA, we designed seven different quantitative RT-PCR (qPCR) assays, each agnostic to HLA-A subtype. The corresponding primers avoided all sequence variants in 1,000 Genomes Project data, and all amplicons spanned at least one intervening sequence (Table S2). We prepared peripheral blood cell RNA from three subjects homozygous for the high-risk haplotype and seven homozygous for nonhigh-risk haplotypes. We then used equal amounts of RNA from each subject to assay HLA-A RNA by qPCR using two different primer sets, measuring 18S ribosomal RNA for normalization. As shown in Fig. 2, there was close agreement between the two HLA-A RNA qPCR assays. Strikingly, the average amount of HLA-A RNA was 1.39-fold higher in cells from subjects homozygous for the high-risk haplotype (1.16 ± 0.08; range 1.04–1.30) than in cells from subjects homozygous for various nonhigh-risk haplotypes (0.83 ± 0.05; range 0.72–0.98), with no overlap between groups. This difference was highly significant (P = 0.0048), confirming that the high-risk HLA-A region SNP haplotype is associated with elevated expression of HLA-A RNA.
Table S2.
Quantitative PCR primer sets used to assay HLA-A mRNA
| Name | Amplicon length (bp) | Orientation | Sequence |
| HLA-A primer set 1 | 145 | F | 5′-CTGGAGCTGTGGTCGCTGC-3′ |
| R | 5′-ACAAGGCAGCTGTCTCACACTT-3′ | ||
| HLA-A primer set 2 | 143 | F | 5′-CTGGAGCTGTGGTCGCTGC-3′ |
| R | 5′-AAGGCAGCTGTCTCACACTTTA-3′ |
Fig. 2.
HLA-A RNA in subjects homozygous for the high-risk and nonhigh-risk HLA-A region haplotypes. HLA-A RNA was measured in peripheral blood RNA from subjects homozygous for the high-risk MHC class I haplotype (nos. 1–3) or nonhigh-risk haplotypes (nos. 4–10) using two different qPCR assays (Table S2), and was normalized to 18S rRNA. Black bars, primer set 1; gray bars, primer set 2; each shows the mean of triplicate assays.
In addition, we analyzed HLA-A mRNA expression data for 358 EUR subjects for whom both lymphoblastoid cell line mRNA-seq (18) and whole-genome DNA sequence (19) data were available. Subjects were classified by sequence-based genotypes of lead variant rs60131261, and mRNA-seq data were analyzed for HLA-A (ENSG00000206503). As shown in Fig. 3, the average amount of HLA-A mRNA was significantly higher in subjects homozygous for the high-risk allele of rs60131261 (1067.75 ± 20.78) than in subjects homozygous for the low-risk allele (985.60 ± 17.28) (P = 0.006), with heterozygotes intermediate between the two groups of homozygotes (1012.65 ± 13.56). These results thus confirm that the high-risk allele of lead HLA-A region variant rs60131261 is associated with elevated expression of HLA-A mRNA.
Fig. 3.
Normalized HLA-A mRNA expression data from the 1,000 Genomes Project subjects classified by genotype of lead HLA-A region SNP rs60131261. RNAseq mRNA profiles for 358 EUR subjects of the 1,000 Genomes Project were obtained along with their genotypes for rs60131261 and subjected to ANOVA. RPKM, reads per kilobase of transcript per million mapped reads. The gray box denotes the first through third quartile and the horizontal line in the box denotes the median. Black squares indicate means. Short horizontal lines denote 99% confidence limits. Crosses denote outliers.
Convergent CTCF Sites Define a Contact Domain Between the HLA-A Downstream Regulatory Region and the HLA-A Promoter.
Enhancers and other transcriptional regulatory elements modulate transcription by being brought into close proximity to their cognate promoter (20). Several approaches have been used to detect in vivo long-range enhancer–promoter spatial interactions in chromatin and identify functional domains. We analyzed in situ genome-wide chromosome conformation capture (Hi-C) sequencing data (21) in the HLA-A region of GM12878 lymphoblastoid cells. As shown in Fig. 4, the segment from the 5′ end of the HLA-A gene through the predicted downstream transcriptional regulatory element is marked by convergent CTCF insulator sites, defining an ∼22-kb contact domain (chr6:29,910,000–29,932,000). This configuration is strongly suggestive of a chromatin loop juxtaposing the downstream regulatory region and the HLA-A promoter, itself contained within a larger ∼170-kb chromatin loop.
Fig. 4.
Hi-C analysis of the HLA-A region of chromosome 6p. In situ Hi-C data for the HLA-A region of chromosome 6p of GM12878 lymphoblastoid cells (21) were analyzed by X-Y comparison using Juicebox (www.aidenlab.org/juicebox/). RefSeq genes, CTCF binding sites and orientation, DNase I hypersensitive sites, and H3K27ac, H3K4me1, and H3K4me3 marks are indicated. The box denotes the segment from HLA-A through the predicted downstream transcriptional regulatory element. HLA-A is the only protein coding gene in the region; HLA-H, HCG4B, HCG9, and ZNRD-AS1 are all nonprotein-coding RNAs.
Genomic Run-on Sequence Data Identify Multiple Bidirectional Promoters in the HLA-A Downstream Regulatory Region.
Mammalian transcriptional regulators frequently contain transcriptionally active promoters (20). To assess potential promoter modules in the HLA-A region, we analyzed in vivo genomic run-on sequence (GRO-seq) data generated from the human cell line HCT116 (22). GRO-seq, which provides a more sensitive and quantitative view of ongoing RNA polymerase II transcription than previous nuclear run-on assays (23), showed bidirectional transcription associated with the HLA-A promoter, and also detected bidirectional transcription originating from at least three distinct promoters within the downstream transcriptional regulatory region (Fig. 5). These results confirm that the HLA-A downstream regulatory region is transcriptionally active in vivo.
Fig. 5.
GRO-seq data in the HLA-A region of chromosome 6p. Histogram of reads from HCT116 GRO-seq data shows transcription of both the HLA-A gene and a region 20 kb downstream of the gene coincident with the predicted downstream transcriptional regulatory element. Blue are reads on forward strand and red are reads on reverse strand. Reads from two 1-h replicates were summed from cells treated with control DMSO alone (Gene Expression Omnibus GSE53964). File is a Bedgraph with reads mapped and normalized to millions (22). DNase I clusters track (GM12878) is from ENCODE.
The HLA-A Downstream Regulatory Region on the High-Risk Haplotype Contains Multiple Transcriptional Promoters, but Does Not Act as a Classic Enhancer.
To investigate differential function of the HLA-A downstream transcriptional regulator on high-risk and nonhigh-risk haplotypes, we compared subjects 1 and 10, who expressed the highest versus lowest amounts of HLA-A RNA, respectively (Fig. 2). These two subjects were homozygous for the alternative alleles of all SNPs that defined the high-risk versus nonhigh-risk haplotypes, respectively, as determined by sequencing a 6,020-bp segment of their genomic DNA (chr6:29,932,128–29,938,147) that spanned the downstream regulator (Table S3). We prepared firefly luciferase reporter constructs containing the full-length downstream regulator region from the two high-risk haplotypes (HR1 and HR2) of subject 1 and the two nonhigh-risk haplotypes (NHR1 and NHR2) of subject 10 (Table S4). For each, the element was inserted in either orientation immediately upstream of the luciferase reporter gene (luc2), which lacks a known promoter. Reporter constructs were transiently transfected into HeLa cells. As shown in Fig. 6A, both full-length high-risk haplotypes HR1 and HR2 from subject 1 exhibited significant promoter function, although in opposite orientations. Greater promoter activity was observed from haplotype HR1 than HR2. In contrast, both full-length nonhigh-risk haplotypes from subject 10 had much less promoter activity.
Table S3.
Variations of NHR1, NHR2, HR1, and HR2 haplotypes from genomic reference sequence (GRCh37/hg19)
| Chr6 nt (GRCh37/hg19) | rs number (variation) | Haplotype | Predicted altered transcription factor binding sites | |||
| NHR1 | NHR2 | HR1 | HR2 | |||
| 29932169 | rs4472395 A->T | T | T | T | T | −HMGIY, −Xvent-, +ipf1 |
| 29932292 | rs2523933 G->T | T | T | G | G | +POU6F1, +islet1, +ipf1, −NF-1 |
| 29932330 | rs2517681 T->C | T | T | C | C | +Smad4, +CPBP, −POU2F1 |
| 29932380 | rs2571402 A->T | A | A | T | T | |
| 29932666 | rs9260734 G->A | G | G | A | G | +ZFP105, +RUSH-1α |
| 29932685 | rs9260735 T->A | T | T | A | T | |
| 29932708 | rs9280821 insAG | — | — | insAG | — | |
| 29932738 | rs9393984 G->C | G | G | G | C | −NF-1A, +CPBP, +Muscle initiator |
| 29932865 | rs9357088 G->T | G | G | G | T | −Muscle initiator, −ZF5, +MAFA |
| 29932897 | rs3903160 G->A | G | G | G | A | |
| 29932916 | rs66870184 delT, insCAG | T | T | delT, insCAG | delT, insCAG | +RFX1, vEts, −YY1 |
| 29932923 | rs9260737 G->A | A | A | A | A | −YY1, −Hic1, +MEIS1 |
| 29933000 | rs3873283 A->G | A | A | G | A | -c-Myb |
| 29933159 | rs2523932 A->G | A | A | G | A | +Sox10, +LEF-1 |
| 29933225 | rs2517678 C->T | T | T | C | C | −AP-2αA, −AP-2αA, +RFX1, +TTF-1 |
| 29933236 | rs9260740 T->C | C | C | C | C | +c-Myb, +c-Myb, −RREB-1, −islet1, −ipf1, −YY1, -BRCA1:USF2 |
| 29933261 | rs6457109 T->C | T | T | C | T | +SP100, −YY1 |
| 29933342 | rs34402263 delGTT | delGTT | delGTT | GTT | GTT | −BRCA1:USF2, −FAC1 |
| 29933388 | rs55829038 T->C | T | T | C | T | +MAFA |
| 29933439 | rs2844806 C->T | T | T | T | C | +BRCA1:USF2 |
| 29933549 | rs9260742 T->G | G | G | G | G | rs9260742+rs9260743: −FAC1, +BRCA1:USF2, +RFX1, +RFX, +C/EBPα, +NF-1A |
| 29933550 | rs9260743 T->C | C | C | C | C | |
| 29933881 | rs6457110 T->A | T | T | A | A | −TTF-1 |
| 29933939–29933938 | rs142460960 | — | — | insCACACACACACACACA | * | |
| 29933939–29933938 | — | — | * | insCACACACACACACACAC | ||
| 29933942–29933946 | AAAAAA | AAAAAA | * | delAAAAAA | −FAC1, −FAC1, −FAC1, −FAC1 | |
| 29933942–29933947 | AAAAAAA | AAAAAAA | delAAAAAAA | * | −FAC1, −FAC1 −FAC1, −FAC1, −FAC1, −FAC1 | |
| 29933956–29933957 | rs545870771 delAA | delAA | AA | AA | AA | −FAC1, −FAC1 |
| 29934022 | rs6914699 T->C | T | T | C | C | |
| 29934080 | rs6935024 C->T | C | C | T | T | +Zfx, −ZF5, +P53, −Churchill, +NF-1A |
| 29934109 | rs2571405 A->G | G | G | G | G | +E2A, +E2A, −MAFA, −Pbx, −CDP CR1, +ER-α, +GEN_INI |
| 29934131 | A > TT | A > TT | A > TT | A > TT | +Ets | |
| 29934151 | rs6935189 C->G | C | C | G | G | rs6935189+rs6935198+rs6935053+rs6914900: −ER-α, +ZF5, −CTCF, +BEN, +MZF-1, −Churchill, −Nkx2.5, −AP-1, −Tbx5 |
| 29934154 | rs6935198 C->G | C | C | G | G | |
| 29934163 | rs6935053 A->C | A | A | C | C | |
| 29934170 | rs6914900 T->G | T | T | G | G | |
| 29934348 | rs17186475 A->C | A | A | C | A | −Sox10, -YY1, +AP-2αA, +AP-2αA |
| 29934454 | rs3893259 A->G | G | G | G | G | +Smad4 |
| 29934697 | rs4713270 G->A | G | G | A | A | -HES-1, +Freac-3 |
| 29934905 | rs34811773 delG | G | G | delG | delG | -AP-2αA, −AP-2αA, −ZF5, +Muscle initiator, −HES-1, −CPBP, −Churchill |
| 29934948 | rs4713271 G->A | G | G | A | A | rs4713271+rs4713272: −GKLF, −SP100, +GLI, +CPBP, +HES-1 |
| 29934958 | rs4713272 A->G | A | A | G | G | |
| 29935009 | rs6920709 T->C | T | T | C | C | +RelA-p65, +Churchill |
| 29935017 | rs6941153 C->T | C | C | T | T | −MEF-2, +HMGIY |
| 29935026 | rs200437571 insA | — | — | insA | insA | rs200437571+rs561051433: +FAC1, +p53, +p53, −RUSH-1α |
| 29935033 | rs561051433 G->C | G | G | C | C | |
| 29935049 | rs3893465 T->G | G | G | G | G | +Muscle initiator, −RREB-1, −AML3, −RREB-1, +ING4, +ING4 |
| 29935093 | rs6941600 G->T | G | G | T | T | +islet1, +ZNF333, +ipf1 |
| 29935108 | rs9260757 A->G | A | A | G | G | −TTF-1 |
| 29935166 | rs9378188 G->A | G | G | G | A | −MEIS1, −BEN, +Sox10, +LEF-1 |
| 29935199 | rs369302554 delA | * | * | delA | delA | |
| 29935199–29935200 | rs41271896 delAA | delAA | delAA | * | * | |
| 29935230 | rs6921094 T->C | T | T | C | T | +Nkx2.5, +CP2 |
| 29935250 | rs3893464 G->A | A | A | G | G | −Ebox |
| 29935405 | rs3893463 C->T | T | T | C | C | +C/EBPα, +NF-1, +Hic1, +TTF-1 |
| 29935469 | rs9260761 T->C | C | C | C | C | −RUSH-1α |
| 29935541–29935540 | rs376765655 insAAA | insAAA | * | — | — | +FAC1 (+) |
| 29935541–29935540 | * | insAAAA | — | — | +FAC1 (+), +FAC1 (+) | |
| 29935576 | rs4538750 G->A | G | G | A | A | rs4538750+rs4391295: −CDP CR1, −CDP CR1, +CDX-2, +AP-1, +Gfi1, − RFX1 |
| 29935590 | rs4391295 C->T | C | C | T | T | |
| 29935798 | rs4248140 T->C | T | T | C | C | +RFX1, +CP2, +LRH-1, +SF-1 |
| 29935843 | rs4248141 G->A | G | G | A | A | rs4248141+rs4248142: +Ikaros, +COE1, -RBP-Jκ, −c-Myb, −c- Myb, −SP100, +YY1, −c-Myb |
| 29935849 | rs4248142 C->T | C | C | T | T | |
| 29935850 | rs6415118 G->A | A | A | G | G | -c-Myb, −c-Myb, −SP100 |
| 29935891 | rs4248143 C->T | C | C | T | T | −GLI, +YY1, −GKLF, −CPBP |
| 29936135 | rs4959036 G->T | T | T | T | T | −GKLF, +YY1 |
| 29936269 | rs73416648 C->G | C | C | G | C | −Ikaros, +CREB1, −POU6F1 |
| 29936307 | rs4959037 A->T | A | A | T | T | −SREBP, −GEN_INI |
| 29936351 | rs4959038 G->A | G | G | A | A | −RFX1, +FAC1, −RFX |
| 29936404 | rs6927487 T->C | T | T | C | C | rs6927487+rs12175093: +Muscle initiator, −YY1, +AML3, +MAFA |
| 29936408 | rs12175093 A->G | A | A | G | G | |
| 29936620 | rs9280825 insGTAT | insGTAT | insGTAT | — | — | +TATA, +TATA |
| 29936620–29936619 | rs143960748 insTA | — | — | insTA | insTA | +TATA, +TATA |
| 29936632 | rs531988651 C->T | T | T | C | C | rs531988651+rs2523973: +TATA, +TATA, +TATA, +TATA |
| 29936634 | rs2523973 C->T | T | T | C | C | |
| 29936655 | rs73727620 A->G | A | A | G | G | +POU2F1, −TATA, +POU2F1, +Pit-1 |
| 29936665 | rs567987089 | * | * | delATATATACACACAC | delATATATACACACAC | −TATA, −Freac-3, −TATA |
| 29936672 | delAC | delAC | * | * | ||
| 29936680 | rs200402275 T->C | C | C | T | T | −TATA, −TATA |
| 29936689–29936690 | rs201818644 delAC | AC | AC | delAC | AC | |
| 29936701–29936706 | rs199969293 delACACAT | ACACAT | ACACAT | ACACAT | delACACAT | −TATA, −TATA |
| 29936704 | rs537420750 C->T | T | T | C | C | +TATA, +TATA |
| 29936715–29936718 | rs573907454 delATAC | ATAC | ATAC | delATAC | ATAC | −TATA, −TATA |
| 29936780 | rs62388695 C->T | C | C | T | C | −Pit-1, +GATA, +GATA, −YY1 |
| 29936795 | rs12198625 T->C | T | T | C | C | −Freac-3, −TATA, +GATA, +Pbx, +Pbx, +CDP CR1 |
| 29936796 | — | — | insTATC | * | +CDP CR1, +Pbx, +CDP CR1, +GATA | |
| 29936796 | — | — | * | insTATCTATCTATC | +Pbx, +Pbx, +Pbx, +CDP CR1, +CDP CR1, +CDP CR1, +CDP CR1, +CDP CR1, +CDP CR1, +GATA, +GATA, +GATA | |
| 29936914 | rs12193100 G->A | G | G | A | A | +HMGIY, −Freac-3, −POU2F1, −ATF-2, +HMGIY, +TATA, +CDX-2 |
| 29937104 | rs12193110 C->T | C | C | T | T | −MEIS1, +Hbp1, +GEN_INI, −YY1 |
| 29937127 | rs12206499 A->G | A | A | G | G | +CPBP, −RREB-1, −Xvent-1, −RUSH-1α |
| 29937148–29937149 | rs386698637 TG->CT | TG | TG | CT | CT | −CDP CR1, −Ikaros, −YY1, −Hic1, +GEN_INI |
| 29937262 | rs2517672 A->G | G | G | A | A | +c-Myb |
| 29937336–29937339 | rs60131261 delTTTA | TTTA | TTTA | delTTTA | delTTTA | −ZFP105, −Freac-3, +Cdc5, BBX, −ATF-2, −ATF-2 |
| 29937493 | rs4713274 G->C | G | G | G | C | −ATF-2 |
| 29937541 | rs11752303 G->A | G | G | G | A | +Freac-3, −Nkx2.5, +POU2F1, +Xvent-1 |
| 29937555 | rs1061539 T->C | C | C | C | C | +Xvent-1 |
| 29937567–29937569 | rs66503418 delAAG | AAG | AAG | delAAG | delAAG | −RUSH-1α |
| 29937579 | rs73727624 C->T | C | C | T | C | +TATA, −POU2F1, −C/EBPα, −ATF-2, +CDX-2 |
| 29937580 | rs4713275 A->C | A | A | A | C | −POU2F1, +NF-AT1, −RUSH-1α |
| 29937740 | rs2256539 C->T | T | T | T | T | −Nkx2.5, +POU6F1, +LEF-1 |
| 29937784 | rs2523972 T->C | T | T | C | C | −AP-1 |
| 29937790 | rs6911940 A->G | G | G | G | G | rs6911940: +Nkx2.5 |
| 29937794 | rs4713276 C->G | C | C | G | G | rs6911940+rs4713276: +Nkx2.5, +GKLF. |
| 29937795 | rs1061537 G->A | A | A | G | G | +RUSH-1α |
| 29937826 | rs1061536 T->C | C | C | C | C | rs1061536: −c-Myb, −RFX1, −RFX |
| 29937833 | rs2256543 T->C | T | T | C | C | rs1061536+rs2256543: −RFX, −RFX1, +c-Myb, −FAC, +ipf1, +Pit-1, |
| +ZNF143 | ||||||
| 29937896 | rs3202637 C->T | T | T | T | T | −TEF-1, +TTF-1 |
| 29937924 | rs1061535 T->C | C | C | C | C | |
| 29937977 | rs2517671 A->G | G | G | A | A | −C/EBPα, +GLI, +Sp1, +GKLF |
| 29938110 | rs6916422 C->T | C | C | T | T | +MEF-2, +Xvent-1, +FAC1 |
Different haplotypes have different sequences at same nucleotide position.
Table S4.
Segments used to construct plasmids used in transient transfection assays
| Segment name | Chr6 nt (GRCh37/hg19) | Size (bp) | Primer sequence |
| 6.0 | 29932128–29938147 | 6,020 | 5′-CTGAACCCTCGAGCCCCTCAGTTTAGCAGAACAGCTA-3′ |
| 5′-ACTCGTTCTCGAGTCCCTGTGAGAAAGAAACTCACCCATTC-3′ | |||
| 3.2 | 29932128–29935306 | 3,179 | 5′-CTGAACCCTCGAGCCCCTCAGTTTAGCAGAACAGCTA-3′ |
| 5′-ACTCGTTCTCGAGCTTGTTTAGCCAGGGCATACATTT-3′ | |||
| 1.4 | 29932128–29933533 | 1,406 | 5′-CTGAACCCTCGAGCCCCTCAGTTTAGCAGAACAGCTA-3′ |
| 5′-ACTCGTTCTCGAGCTGTAAGAAGAGCTGAGCAGCC-3′ | |||
| 1.8 | 29933512–29935306 | 1,795 | 5′-ACTCGTTCTCGAGGGCTGCTCAGCTCTTCTTACAG-3′ |
| 5′-ACTCGTTCTCGAGCTTGTTTAGCCAGGGCATACATTT-3′ | |||
| 1.2 | 29933512–29934719 | 1,208 | 5′-ACTCGTTCTCGAGGGCTGCTCAGCTCTTCTTACAG-3′ |
| 5′-ACTCGTTCTCGAGGGCTGCTCAGCTCTTCTTACAG-3′ | |||
| 2.8 | 29935283–29938147 | 2,865 | 5′-ACTCGTTCTCGAGAAATGTATGCCCTGGCTAAACAAG-3′ |
| 5′-ACTCGTTCTCGAGTCCCTGTGAGAAAGAAACTCACCCATTC-3′ |
Fig. 6.
Transient transfection assay of promoter activities in the HLA-A downstream transcriptional regulatory element. Luciferase reporter constructs containing segments of the HLA-A downstream regulatory element inserted immediately upstream of the luc2 gene. (A) Full-length high-risk haplotypes (HR1, HR2) from subject 1 (orange and yellow) or full-length nonhigh-risk haplotypes (NHR1, NHR2) from subject 10. (B) Subfragments of the high-risk HR1 haplotype. (C) Subfragments of the high-risk HR2 haplotype. Arrowheads denote forward (F) and reverse (R) orientations relative to genomic orientation in chromosome 6. Relative light units denote fold-change of transcriptional activity relative to the pGL4.10 backbone plasmid. SEMs are indicated.
To map promoter function within the HLA-A downstream regulatory region, we prepared a series of analogous constructs containing only subfragments of the HR1 or HR2 high-risk haplotypes carried by subject 1 (Table S3). As shown in Fig. 6B, both high-risk haplotypes had multiple segments with promoter activity. Both similarities and differences were observed between the HR1 and HR2 haplotypes in terms of the location and functional orientation of apparent promoters, with no apparent simple pattern of promoter localization. These data are consistent with our GRO-seq findings, indicating the existence of multiple promoters within the HLA-A downstream regulatory region.
Comparison of the nucleotide sequences of the cloned HR1, HR2, NHR1, and NHR2 haplotypes with the human reference sequence (GRCh37/hg19) defined a remarkable pattern (Table S3). All four haplotypes shared 19 nucleotide differences from the reference sequence. Haplotypes NHR1 and NHR2 shared 16 additional differences, with 1 more difference specific to NHR1. In contrast, haplotypes HR1 and HR2 shared 45 additional differences from the reference, plus another 15 specific to HR1 and another 10 specific to HR2. Thus, the two NHR haplotypes are quite similar to each other and are generally similar to the reference, whereas the two HR haplotypes have far more base differences, both compared with the reference sequence and to each other. These sequence differences affect many different predicted transcription factor binding motifs, presumably driving the observed differences in transcriptional function among the different haplotypes.
To assess possible enhancer function of the HLA-A downstream regulatory element, we prepared luciferase reporter constructs containing the full-length downstream regulatory region high-risk haplotypes HR1 and HR2 and nonhigh-risk haplotypes NHR1 and NHR2, as well as corresponding subsegments, inserted in both orientations upstream of a luc2 reporter gene with minimal promoter. In all cases, the results were similar to those obtained using a luc2 reporter with no promoter, with no augmentation of expression (Fig. S1 A–C). Furthermore, reporter constructs containing the full-length HR1, HR2, NHR1, and NHR2 downstream regulatory regions inserted immediately downstream of the luc2 gene yielded essentially no luciferase expression, regardless of the orientation or presence versus absence of a minimal promoter upstream of luc2 (Fig. S1 D and E). Thus, the HLA-A downstream regulatory region does not act as a transcriptional enhancer for this minimal promoter in the context of a conventional assay of circular plasmids in transfected HeLa cells.
Fig. S1.
Transient transfection assay of enhancer activities in the HLA-A downstream regulatory region. (A) Luciferase reporter constructs containing the full-length HLA-A downstream regulatory region inserted immediately upstream of a minimal promoter (green) driving transcription of the luc2 gene. HR1, HR2, high-risk haplotypes from subject 1 (orange and yellow); NHR1, NHR2, nonhigh-risk haplotypes from subject 10 (blue and pale green). Arrowheads denote forward (F) and reverse (R) orientations relative to genomic orientation in chromosome 6. Relative light units denote fold-change of transcriptional activity relative to the pGL4.23 backbone plasmid. SEMs are indicated. (B) Subfragments of the high-risk HR1 haplotype. (C) Subfragments of the high-risk HR2 haplotype. (D) Full-length haplotypes inserted downstream of the promoterless luc2 reporter gene. Relative light units denote fold-change of transcriptional activity relative to the pGL4.10 backbone plasmid. (E) Full-length haplotypes inserted downstream of the minimal promoter:luc2 reporter gene. Relative light units denote fold-change of transcriptional activity relative to the pGL4.23 backbone plasmid.
Discussion
We previously showed that vitiligo is genetically associated with variation in the MHC class I region, in close proximity with HLA-A (4), and specifically with HLA-A*02:01 in both European-derived Caucasians (5) and Japanese (24). Here, we refine genomic localization of this association to an SNP haplotype ∼20 kb downstream of the HLA-A gene itself, spanning a 5-kb ENCODE regulatory element. Primary association of vitiligo is thus with the HLA-A downstream regulatory region, which is secondarily in very strong linkage disequilibrium with HLA-A*02:01:01:01.
The HLA-A promoter and downstream regulatory region are flanked by convergent CTCF sites, with an apparent 22-kb chromatin loop juxtaposing the downstream regulatory region and the HLA-A promoter. This configuration suggests that the downstream regulatory region modulates function of the HLA-A promoter. Consistent with this finding, RT-PCR analysis of HLA-A RNA in peripheral blood cells from normal healthy subjects showed that subjects homozygous for the high-risk SNP haplotype spanning the HLA-A downstream regulatory region express significantly percent more HLA-A RNA than subjects homozygous for nonhigh-risk haplotypes. Similarly, mRNA-seq analysis of lymphoblastoid cells from 1,000 Genomes Project subjects showed that subjects homozygous for the high-risk allele of lead variant rs60131261 express significantly more HLA-A mRNA than subjects homozygous for the low-risk allele.
Nevertheless, the specific function of the HLA-A downstream regulatory region is not yet clear. The downstream regulatory region has an open hypomethylated chromatin configuration in all cell types tested by ENCODE, multiple DNase I hypersensitivity sites, RNA polymerase II and transcription factor binding sites, active bidirectional promoters, and prominent H3K4me1, H3K4me3, and H3K27ac marks, and contains multiple sites of active bidirectional transcription mapped by GRO-seq. These data suggested the presence of multiple promoters, which we confirmed in luciferase reporter assays of transfected cells. These features are all suggestive of an active transcriptional enhancer (16, 17, 20). However, the HLA-A downstream regulatory region did not act as an enhancer in a conventional transfection assay driving a minimal promoter. This finding may reflect specificity for the native HLA-A promoter in a linear chromosome, rather than the minimal promoter in the context of a circular plasmid. Alternatively, bidirectional promoters often serve specialized functions (25), and it may be that the in vivo biological function of the HLA-A downstream regulatory region is more complex, perhaps acting as a superenhancer (26), locus control region (27), or other higher-order transcriptional regulatory element in the context of a locus that is expressed in almost all cell types.
DNA sequence analysis of the HLA-A downstream regulatory region identified a large number of predicted transcription factor binding motifs. Moreover, the DNA sequences of the two nonhigh-risk haplotypes analyzed are generally similar to the GRCh37/hg19 reference sequence, whereas the two high-risk haplotypes analyzed differ far more, both from the reference sequence and from each other. These sequence differences affect many different predicted transcription factor binding motifs, which together presumably account for the differences in transcriptional activity observed among the different haplotypes.
Most studies of HLA autoimmune disease associations have focused on HLA-type specificity, which governs antigen binding and presentation as a result of amino acid sequence differences among alleles. However, genomewide association studies, including those of autoimmune diseases, have implicated transcriptional regulatory elements at many disease loci, accounting for an estimated 79% of total heritability across multiple common complex diseases (28). Our findings show that causal variation underlying genetic association of vitiligo with the HLA-A region affects both HLA-A–type specificity and transcriptional activity, resulting in a combination of qualitative and quantitative consequences. Primary association of vitiligo with the MHC class I region association is with a 9.6-kb SNP haplotype spanning a transcriptional regulatory region downstream of HLA-A. The high-risk haplotype induces gain-of-function, up-regulating expression of HLA-A mRNA in vivo, in strong linkage disequilibrium with HLA-A*02:01:01:01-type specificity. Expression of HLA class I protein molecules corresponds closely with RNA level (3); thus, the vitiligo high-risk haplotype likely causes elevated expression of HLA-A*02:01:01:01 protein. Because HLA-A*02:01 presents a number of melanocyte-derived peptides that constitute vitiligo autoimmune antigens (6–11), its elevated expression would facilitate recognition and immune targeting of melanocytes by cognate autoreactive T cells. Our findings thus highlight the pathogenic importance of quantitative functional effects of variation in the classic MHC genes, beyond just antigenic specificity.
Materials and Methods
Genotypes were imputed through the extended MHC (11, 12) in 2,853 vitiligo patients and 37,412 controls, and we used logistic regression analysis to determine which variants represent the strongest association signal in the MHC class I region. Healthy adult controls were genotyped for SNPs in the MHC class I region, and HLA-A RNA was quantitated by RNA-seq and RT-PCR analyses. Luciferase reporter constructs containing segments of the downstream regulatory region representing high-risk and low-risk haplotypes were transfected into HeLa cells and relative light units were assayed. Full experimental details can be found in SI Materials and Methods. This project was approved by the Colorado Multiple Institutional Review Board (COMIRB), and written informed consent was obtained from all subjects.
SI Materials and Methods
Genotype Imputation.
We imputed genotypes through the extended MHC (11, 12) for a total 2,853 generalized vitiligo patients of non-Hispanic and non-Latino European ancestry (EUR) from North America and Europe [NCBI Database of Genotypes and Phenotypes (dbGaP) accession no. phs000224.v2)] who met strict clinical criteria (29), and 37,412 EUR controls not specifically known to have any autoimmune disease or malignant melanoma (dbGaP; phs000092.v1.p1, phs000125.v1.p1, phs000138.v2.p1, phs000142.v1.p1, phs000168.v1.p1, phs000169.v1.p1, phs000206.v3.p2, phs000237.v1.p1, phs000346.v1.p1, and phs000439.v1.p1; phs000203.v1.p1, and phs000289.v2.p1; phs000196.v2.p1, phs000303.v1.p1, phs000304.v1.p1, phs000368.v1.p1, phs000381.v1.p1, phs000387.v1.p1, phs000389.v1.p1, phs000395.v1.p1, phs000408.v1.p1, phs000421.v1.p1, phs000494.v1.p1, and phs000524.v1.p1). Control datasets were matched to vitiligo case datasets based on platforms used for genotyping.
Quality-control filtering of genome-wide genotype data were carried out using PLINK, v1.9 (pngu.mgh.harvard.edu/∼purcell/plink/), excluding subjects with SNP call rate <98%, sex discordance, duplication, or cryptic relatedness (pi-hat > 0.0625). SNPs were excluded based on genotype missing rate ≥2%, observed minor allele frequency <0.01, or significant (P < 10−4) deviation from Hardy–Weinberg equilibrium.
Genotype imputation was carried out using SHAPEIT version2 (www.shapeit.fr/) to pre-phase genotypes to produce best-guess haplotypes, and then imputed into these estimated haplotypes using IMPUTE2 (https://mathgen.stats.ox.ac.uk/impute/impute_v2.html) using the 1,000 Genomes Project phase I integrated variant set version 3 (March, 2012) (http://www.1000genomes.org/) as reference panel. Only genotypes with imputation INFO > 0.5 were retained, which were combined with prior SNP genotype data.
Statistical Genetic Analyses.
We carried out genetic ancestry matching of patients and controls using GemTools (wpicr.wpic.pitt.edu/WPICCompgen/GemTools/GemTools.htm), and performed a Cochran-Mantel-Haenszel (CMH) analysis to test for association. To determine which variants represent the strongest association signal in the MHC class I region, we then applied logistic regression analysis, comparing the fit of a model containing each variant tested and the most significant variant in the region (rs60131261) to a model containing only rs60131261, assuming a multiplicative genotypic effect for the high-risk allele of each variant. We consider rs60131261 and all variants whose effects could not be distinguished from rs60131261 as representing the strongest association signal in the region. Analyses were performed using PLINK v1.9.
HLA-A RNA Assay.
Eighty-one unrelated healthy adult EUR subjects without known autoimmune disease were genotyped for haplotype tagSNPs rs12193100, rs35066870, and rs12206499 (Table S1). Peripheral blood was collected from three subjects homozygous for the high-risk haplotype and seven homozygous for nonhigh-risk haplotypes and RNA was extracted using the PAXgene Blood RNA Kit (Qiagen). This project was approved by the Colorado Multiple Institutional Review Board (COMIRB), and written informed consent was obtained from all subjects. RNA concentration was measured by NanoDrop Spectrophotometer ND-1000 (NanoDrop Technologies). For each individual, 227 μg RNA was then converted into cDNA using iScript (Bio-Rad).
We developed seven different qPCR assays for HLA-A RNA using primer-pairs that contained no SNPs or other sequence variations identified in 1,000 Genomes Project Phase 1 data. All assays crossed at least one pair of exon boundaries (Table S2). Primer sequences for 18S RNA are from Roche. qPCR assays were carried out on the LightCycler 480 (Roche) using the recommended PCR cycle. Raw data were calibrated using standard curves, and HLA-A RNA was normalized to 18S RNA. All assays were carried out in triplicate and the results were averaged. Differences between individuals homozygous for the high-risk and low-risk haplotypes were assessed by t test.
RNAseq analysis was carried out using data for 358 EUR subjects for whom both whole-genome DNA sequence (www.1000genomes.org/) and lymphoblastoid cell RNAseq (www.geuvadis.org/web/geuvadis/home) data were available. Expression of HLA-A RNA was compared pairwise across the three genotypes of lead variant rs60131261 genotypes by ANOVA.
Cell Culture.
HeLa cells were cultured in DMEM (Sigma-Aldrich) containing 10% (vol/vol) FBS (Invitrogen) supplemented with 2 mM l-glutamine and 1× antibiotics and antimycotic (Gibco-Invitrogen).
Luciferase Reporter Constructs.
For transient transfection, expression analyses were prepared using the pGL4.10 no promoter vector and pGL4.23 MinP vector (Promega). Amplicons (Table S4) containing all (6,020 bp) or portions of the ENCODE element were PCR-amplified from genomic DNA of subject 1, who was homozygous for the high-risk haplotype, and subject 10, who was homozygous for nonhigh-risk haplotypes, and who, respectively, had the highest and lowest HLA-A mRNA expression (Fig. 2). Amplicons representing both alleles from each individual were cloned in both orientations in the XhoI site of the multiple cloning sites just upstream of the luc2 reporter gene in both plasmid vectors. Amplicons containing the corresponding full-length element of both alleles from each subject were also cloned in both orientations into the SalI site in both vectors.
Cell Transfection.
HeLa cells were transiently transfected with each reporter vector using Lipofectamine LTX (Invitrogen). Briefly, equal numbers of cells were seeded in 24-well plates and grown to 80% confluence for 48–72 h. Transfection was performed per the manufacturer’s instructions. Briefly, 500 ng of each reporter construct with 2 μL of lipofectamine LTX and 0.5 μL of Plus reagent was added to 50 μL of OPTI-MEM reduced-serum medium (Gibco-Invitrogen), incubated for 5 min at room temperature and, was applied to cells. After 48 h, cells were harvested using Cell Culture Lysis Reagent (Promega). Luciferase assays were performed using the Luciferase Assay System (Promega). Each transfection experiment was carried out a minimum of six times, with each construct tested and assayed in triplicate in each experiment. Results are expressed as fold-change compared with activity measured for pGL4.10 or pGL4.23, which contained no insert.
Hi-C Analysis.
In situ Hi-C data for GM12878 lymphoblastoid cells (MboI primary + replicate) (21) were compared by X-Y analysis using Juicebox (www.aidenlab.org/juicebox/), with normalization for coverage.
Acknowledgments
We thank the study participants whose contributions made this work possible and Dr. Robin Dowell for assistance with the genomic run-on sequence analysis. This work was funded in part by Grants R01AR045584 and R01AR056292 from the National Institutes of Health. The Janus supercomputer is supported by the National Science Foundation (CNS-0821794), the University of Colorado Boulder, the University of Colorado Denver, and the National Center for Atmospheric Research, and is operated by the University of Colorado, Boulder.
Footnotes
The authors declare no conflict of interest.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1525001113/-/DCSupplemental.
References
- 1.Marrack P, Kappler J, Kotzin BL. Autoimmune disease: Why and where it occurs. Nat Med. 2001;7(8):899–905. doi: 10.1038/90935. [DOI] [PubMed] [Google Scholar]
- 2.Jacobson DL, Gange SJ, Rose NR, Graham NM. Epidemiology and estimated population burden of selected autoimmune diseases in the United States. Clin Immunol Immunopathol. 1997;84(3):223–243. doi: 10.1006/clin.1997.4412. [DOI] [PubMed] [Google Scholar]
- 3.Zachow KR, Orr HT. Regulation of HLA class I transcription in T cells. J Immunol. 1989;143(10):3385–3389. [PubMed] [Google Scholar]
- 4.Jin Y, et al. Variant of TYR and autoimmunity susceptibility loci in generalized vitiligo. N Engl J Med. 2010;362(18):1686–1697. doi: 10.1056/NEJMoa0908547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Jin Y, et al. Next-generation DNA re-sequencing identifies common variants of TYR and HLA-A that modulate the risk of generalized vitiligo via antigen presentation. J Invest Dermatol. 2012;132(6):1730–1733. doi: 10.1038/jid.2012.37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Cox AL, et al. Identification of a peptide recognized by five melanoma-specific human cytotoxic T cell lines. Science. 1994;264(5159):716–719. doi: 10.1126/science.7513441. [DOI] [PubMed] [Google Scholar]
- 7.Parkhurst MR, et al. Identification of a shared HLA-A*0201-restricted T-cell epitope from the melanoma antigen tyrosinase-related protein 2 (TRP2) Cancer Res. 1998;58(21):4895–4901. [PubMed] [Google Scholar]
- 8.Touloukian CE, Leitner WW, Robbins PF, Rosenberg SA, Restifo NP. Mining the melanosome for tumor vaccine targets: P.polypeptide is a novel tumor-associated antigen. Cancer Res. 2001;61(22):8100–8104. [PMC free article] [PubMed] [Google Scholar]
- 9.Salazar-Onfray F, et al. Synthetic peptides derived from the melanocyte-stimulating hormone receptor MC1R can stimulate HLA-A2-restricted cytotoxic T lymphocytes that recognize naturally processed peptides on human melanoma cells. Cancer Res. 1997;57(19):4348–4355. [PubMed] [Google Scholar]
- 10.Bakker AB, et al. Melanocyte lineage-specific antigen gp100 is recognized by melanoma-derived tumor-infiltrating lymphocytes. J Exp Med. 1994;179(3):1005–1009. doi: 10.1084/jem.179.3.1005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kawakami Y, et al. Cloning of the gene coding for a shared human melanoma antigen recognized by autologous T cells infiltrating into tumor. Proc Natl Acad Sci USA. 1994;91(9):3515–3519. doi: 10.1073/pnas.91.9.3515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Horton R, et al. Gene map of the extended human MHC. Nat Rev Genet. 2004;5(12):889–899. doi: 10.1038/nrg1489. [DOI] [PubMed] [Google Scholar]
- 13.Shiina T, Hosomichi K, Inoko H, Kulski JK. The HLA genomic loci map: Expression, interaction, diversity and disease. J Hum Genet. 2009;54(1):15–39. doi: 10.1038/jhg.2008.5. [DOI] [PubMed] [Google Scholar]
- 14.Kellis M, et al. Defining functional DNA elements in the human genome. Proc Natl Acad Sci USA. 2014;111(17):6131–6138. doi: 10.1073/pnas.1318948111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wang J, et al. Sequence features and chromatin structure around the genomic regions bound by 119 human transcription factors. Genome Res. 2012;22(9):1798–1812. doi: 10.1101/gr.139105.112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hon GC, Hawkins RD, Ren B. Predictive chromatin signatures in the mammalian genome. Hum Mol Genet. 2009;18(R2):R195–R201. doi: 10.1093/hmg/ddp409. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Creyghton MP, et al. Histone H3K27ac separates active from poised enhancers and predicts developmental state. Proc Natl Acad Sci USA. 2010;107(50):21931–21936. doi: 10.1073/pnas.1016071107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Lappalainen T, et al. Geuvadis Consortium Transcriptome and genome sequencing uncovers functional variation in humans. Nature. 2013;501(7468):506–511. doi: 10.1038/nature12531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.The 1000 Genomes Project Consortium A map of human genome variation from population-scale sequencing. Nature. 2010;467(7319):1061–1073. doi: 10.1038/nature09534. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Shlyueva D, Stampfel G, Stark A. Transcriptional enhancers: From properties to genome-wide predictions. Nat Rev Genet. 2014;15(4):272–286. doi: 10.1038/nrg3682. [DOI] [PubMed] [Google Scholar]
- 21.Rao SSP, et al. A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell. 2014;159(7):1665–1680. doi: 10.1016/j.cell.2014.11.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Allen MA, et al. Global analysis of p53-regulated transcription identifies its direct targets and unexpected regulatory mechanisms. eLife. 2014;3(3):e02200. doi: 10.7554/eLife.02200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Core LJ, Waterfall JJ, Lis JT. Nascent RNA sequencing reveals widespread pausing and divergent initiation at human promoters. Science. 2008;322(5909):1845–1848. doi: 10.1126/science.1162228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Jin Y, et al. Major association of vitiligo with HLA-A*02:01 in Japanese. Pigment Cell Melanoma Res. 2015;28(3):360–362. doi: 10.1111/pcmr.12356. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Duttke SHC, et al. Human promoters are intrinsically directional. Mol Cell. 2015;57(4):674–684. doi: 10.1016/j.molcel.2014.12.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Hnisz D, et al. Super-enhancers in the control of cell identity and disease. Cell. 2013;155(4):934–947. doi: 10.1016/j.cell.2013.09.053. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Li Q, Peterson KR, Fang X, Stamatoyannopoulos G. Locus control regions. Blood. 2002;100(9):3077–3086. doi: 10.1182/blood-2002-04-1104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Gusev A, et al. Schizophrenia Working Group of the Psychiatric Genomics Consortium SWE-SCZ Consortium Schizophrenia Working Group of the Psychiatric Genomics Consortium SWE-SCZ Consortium Partitioning heritability of regulatory and cell-type-specific variants across 11 common diseases. Am J Hum Genet. 2014;95(5):535–552. doi: 10.1016/j.ajhg.2014.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Ezzedine K, et al. Vitiligo Global Issue Consensus Conference Panelists Revised classification/nomenclature of vitiligo and related issues: The Vitiligo Global Issues Consensus Conference. Pigment Cell Melanoma Res. 2012;25(3):E1–E13. doi: 10.1111/j.1755-148X.2012.00997.x. [DOI] [PMC free article] [PubMed] [Google Scholar]