Correspondence: Richard A. Spritz, M.D., Human Medical Genetics and Genomics Program, University of Colorado School of Medicine, Aurora, CO 80045, USA. Email email@example.com
Vitiligo is a complex disorder in which autoimmune destruction of melanocytes results in white patches of skin and overlying hair. Over the past several years, extensive genetic studies have outlined a biological framework of vitiligo pathobiology that underscores its relationship to other autoimmune diseases. This biological framework offers insight into both vitiligo pathogenesis and perhaps avenues towards more effective approaches to treatment and even disease prevention.
Vitiligo is a complex disorder in which white patches of skin and overlying hair result from autoimmune destruction of melanocytes within the lesions. Vitiligo appears to be of multifactorial causation, involving multiple underlying susceptibility genes and environmental triggers. Over the past several years there has been considerable progress in defining the genetic epidemiology and genetic pathogenesis of vitiligo, and its relationships to other autoimmune diseases. To date, almost all genetic studies have been of generalized or “non-segmental” vitiligo. Recently, advances by these genetic studies have outlined a biological framework that offers real insight into vitiligo pathogenesis and perhaps to more effective approaches to treatment and even disease prevention.
Human pigmentation diseases present the most visually striking of all disorders, and the phenotypes of oculocutaneous albinism, piebaldism and others have been known for thousands of years. Vitiligo itself has had a remarkable history of discovery and re-discovery of key observations overlooked or forgotten over the passage of time. Certainly, patients with various forms of patchy leukoderma were recognized during ancient times; however, vitiligo per se was not described as a specific medical entity until the mid-18th century.
The first clue to vitiligo pathogenesis came from the description and illustration of a patient with concomitant vitiligo, adrenal insufficiency and pernicious anemia, highlighting a link between these three autoimmune diseases. Similar patients with multiple autoimmune diseases were later reported by Schmidt and codified by Neufeld and Blizzard. Further evidence of immune or inflammatory influences in vitiligo came from the recognition that new lesions often occur at sites of skin injury, a phenomenon first described in 1872 by Köbner in the context of psoriasis and which came to bear his name. Perhaps the most important observation regarding vitiligo pathogenesis also came from Kaposi, who illustrated histological lack of cells containing pigment granules within vitiligo skin lesions, later re-described by Hu et al. and yet again by Breathnach et al.
Probably the landmark early study of vitiligo was that of Lerner, who reported on clinical, epidemiological and genetic characteristics of 200 cases, thereby delineating many important clinical and epidemiological aspects of the disease, establishing the clinical terminology and classification scheme that is basically still in use today. This study provided an analytic framework that was adopted by many subsequent epidemiological surveys of vitiligo patients around the world,[12-21] providing key information in formulating the genetic underpinnings of the disorder.
The earliest formal considerations of the genetic basis of vitiligo appear to have been by Stuttgen, Teindel and later by Lerner, all of whom noted familial aggregation of cases not clearly consistent with Mendelian single-gene inheritance. Stuttgen, in particular, was remarkable in suggesting the simultaneous involvement of both recessive and dominant contributory influences, predating the modern genetic concept of “complex inheritance” by decades. Subsequently, Das et al.,[24, 25] Bhatia et al. and Majumder et al. likewise all noted frequent familial clustering of vitiligo cases in a non-Mendelian pattern. Majumder et al. and Nath et al. suggested a multilocus recessive model, Arcos-Burgos et al. postulated multiple forms of vitiligo with different genetic underlying models, whereas most other investigators favored polygenic, multifactorial inheritance,[12, 26, 30-33] which is now generally accepted.
Close relatives of vitiligo patients have elevated risk of vitiligo, as well as other autoimmune diseases, with relative risks for first-degree relatives estimated at approximately 6–18-fold elevated.[12, 24, 27, 33-35] Estimates of vitiligo heritability range from 46% in India to 16% in China, and Alkhateeb et al. found that the concordance for vitiligo in monozygotic twins was 23%, supporting roles for both genetic and non-genetic factors in disease pathogenesis.
Pre-Molecular Genetic Studies
The earliest attempts to identify specific genes that may contribute to vitiligo were genetic association studies of vitiligo using a variety of protein polymorphic markers, including ABO and other blood groups,[15, 25, 36-42] blood group secretor status,[15, 43] and a number of other serum proteins.[25, 44, 45] Essentially, all of these studies proved negative.
Subsequently, there were a number of case–control genetic association studies of histocompatibility antigen (human leukocyte antigen, HLA) types and vitiligo.[46-55] While several of these studies reported weak associations, the results were inconsistent, partly because of inadequate study designs and partly because of likely genetic heterogeneity among the many different populations analyzed. Nevertheless, meta-analysis of data from 11 of these studies subsequently identified association of vitiligo with HLA-A2 (odds ratio, 2.07).
Molecular Genetic Era
Modern genetic studies of vitiligo have principally entailed five scientific approaches: genetic linkage analysis of families with multiple affected relatives (multiplex families), candidate gene association studies comparing relatively small numbers of vitiligo patients (cases) to unaffected controls, genome-wide association studies (GWAS) comparing far larger numbers of cases to controls, DNA sequencing studies and gene expression studies. Genetics researchers give highest credence to disease gene discoveries derived from genome-wide linkage and GWAS, as these approaches are relatively free of biases and are robust to several important sources of potential error. Candidate gene association studies, by contrast, are largely discounted by modern geneticists as an approach to gene discovery, as the vast majority report false-positives resulting from chance fluctuation due to insufficient statistical power, population stratification and the impossibility of adequate correction for multiple testing given publication bias of positive results.[57, 58] DNA sequencing studies largely involve candidate genes previously identified by other means, and thus are subject to the caveats appropriate to the identification of these genes in the first place. Gene expression studies are generally considered invalid as a means of disease gene discovery as they cannot distinguish between primary causal differences versus secondary changes that are the result of pathway dysregulation or disease, but can be useful as a means of confirming the biological relevance of previous genetic findings.
Genetic linkage studies
The first vitiligo genetic linkage studies were of candidate genes. A linkage study of the HLA region in a large family with polyglandular autoimmune disease type II (Schmidt syndrome), including with vitiligo, was negative. Tripathi et al. tested linkage of vitiligo to MITF, again with negative results. The first positive results came from targeted linkage analysis of the major histocompatibility complex (MHC) on chromosome 6p, detecting linkage of vitiligo with microsatellite polymorphic markers in HLA gene regions in families from several different populations.[30, 61, 62]
The first detection of novel vitiligo loci by genome-wide study came from linkage analysis of families with systemic lupus erythematosus (SLE) that included at least one case of vitiligo, detecting the SLEV1 locus on chromosome 17p13. SLEV1 was subsequently confirmed by direct genome-wide study of Caucasian multiplex vitiligo families, and the corresponding causal gene was eventually identified as NLRP1, encoding a key regulator of the innate immune response. Genome-wide analysis of a unique Caucasian family with autosomal dominant vitiligo detected AIS1 on chromosome 1p31.3-32.2, and the corresponding gene was subsequently identified as FOXD3, encoding a developmental regulator of melanoblast differentiation; thus far, additional families with dominant vitiligo and FOXD3 mutations have not been described. Comprehensive genome-wide linkage analyses of other multiplex vitiligo families also detected a number of other vitiligo susceptibility loci, on chromosomes 1, 7, 8, 9, 11, 13, 19 and 21 in Caucasians,[64, 68] and on chromosomes 4, 6 and 22 in Chinese,[61, 69] some of which may correspond to genes detected in subsequent GWAS. Ren et al. studied XBP1 as a candidate gene within the 22q12 linkage region, though this assignment has not yet been confirmed. Xu et al. studied PDGFRA as a candidate gene within the 4q12-q21 linkage region, finding more variation in familial vitiligo cases than controls. However, PDGFRA is not known to play a role in pigment cell biology, and the closely adjacent KIT gene would seem a better candidate for this linkage signal.
Candidate gene association studies
The earliest candidate gene association studies of vitiligo were of loci that previously had been shown to be genetically linked and/or associated with other autoimmune diseases. Though candidate gene association studies are not considered a valid approach to primary discovery of disease genes, this approach is considered acceptable for confirmation of established genes.
The first such study reported genetic association of vitiligo with single nucleotide polymorphisms (SNP) within CTLA4, a locus that had been previously associated with several other autoimmune diseases. Even in that initial report, CTLA4 was only associated with vitiligo in patients who also had concomitant autoimmune diseases, as confirmed in subsequent studies.[74-76] This has been interpreted as perhaps indicating that there are multiple genetically distinct subtypes of vitiligo or that CTLA4 is not actually causal for vitiligo, with apparent genetic association being secondary to epidemiological association of vitiligo with other autoimmune diseases for which CTLA4 is causal. Subsequently, Canton et al. reported association of vitiligo with PTPN22, a gene that also had been previously associated with a number of other autoimmune diseases. This association was subsequently confirmed by other investigators,[79, 80] most importantly by the first GWAS of vitiligo. A number of investigators reported association of vitiligo with loci in the MHC.[82-85] However, because of complex patterns of long-range linkage disequilibrium across the MHC, it has been difficult to assign genetic association to specific genes within the MHC.
Over the following years, a large number of candidate gene studies of vitiligo were published, many based on little or no compelling biological rationale. Birlea et al. published a comprehensive review of 33 claimed vitiligo candidate genes (ACE, AIRE, CAT, CD4, CLEC11A, COMT, CTLA4, C12orf10, DDR1, EDN1, ESR1, FAS, FBXO11, FOXD3, FOXP3, GSTM1, GSTT1, IL1RN, IL10, KITLG, MBL2, NFE2L2, PDGFRA-KIT, PTGS2, STAT4, TAP1-PSMB8, TGFBR2, TNF, TSLP, TXNDC5, UVRAG, VDR, XBP1), finding support for only three (TSLP, XBP1, FOXP3) in analysis of a genome-wide vitiligo GWAS dataset. Additional claimed candidate gene associations with vitiligo include AHR, COX2, GSTP1, IL4, IL19 and IL20RB, INOS, PRO2268, TLR2 and TLR4, while candidate genes reported not to be associated include CD28, ICOS, IL20, IL24e and IL20RA, MTHFR, and SMOC2. Some of these claimed novel candidate gene associations may be valid, though most are likely to represent false-positives.
Genome-wide association studies are currently the “gold standard” of genetic studies of complex diseases. The first vitiligo GWAS was an analysis of patients from a population isolate in Romania with a high prevalence of vitiligo and other autoimmunity, detecting association with SMOC2 located at distal chromosome 6q27. While association with SMOC2 has not been detected in other GWAS, it is located very close to CCR6 in chromosome 6q27, which was detected in GWAS of vitiligo in both Caucasians and Chinese, and it may be that all of three reports represent the same association signal.
Three large GWAS of vitiligo have been reported thus far; two in Caucasians [81, 100, 101] and one in Chinese,[99, 102] while a small gene-centric GWAS of vitiligo has been reported in Indian–Pakistani patients. These studies have detected 30 vitiligo susceptibility loci in Caucasians (Table 1): PTPN22, RERE, IFIH1, CTLA4 (only in patients with other autoimmune diseases), FOXP1, CD80, LPP, CLNK, TSLP, HLA-A, MHC class II (c6orf10-BTNL2-DRB1-DQA1), BACH2, CCR6, TG/SLA, IL2RA, CASP7, CD44, TYR, a gene desert at 11q21, IKZF4, SH2B3, GZMB, OCA2, MC1R, TICAM1, UBASH3A, XBP1, C1QTNF6, TOB2, and FOXP3.[81, 100, 101, 104] The parallel studies of Chinese have detected nine vitiligo susceptibility loci (Table 1): LPP, MHC class I (HLA-B-HLA-C), RNASET2-FGFR1OP-CCR6, IL2RA, ZMIZ1, IKZF4, IL2RB-C1QTNF6, an intergenic interval at 10q22.1 between SLC29A3 and CDH23, and an intergenic interval at 11q23.3 between DDX6 and CXCR5.[99, 102] Thus, vitiligo association with at least LPP, the HLA class I gene region, CCR6, IL2RA, IKZF4 and C1QTNF6 appears to be shared between the Caucasian and Chinese populations, though other genes associated with vitiligo susceptibility or protection may be population-specific, particularly TYR, OCA2 and MC1R. Similarly, the Indian–Pakistani GWAS reported association only with the MHC class II gene region, and formal trans-ethnic analysis indicated that association signals in Caucasians and Indian–Pakistani patients in this genomic region likely share the same ancestral origin.
Table 1. Confirmed and suggestive vitiligo susceptibility loci identified by genome-wide association or linkage studies
Found only in a single family with autosomal dominant vitiligo.[66, 67]
CTLA4 is only associated with vitiligo in patients with other concomitant autoimmune diseases.[74-76] Association of CTLA4 with vitiligo may thus be secondary, driven by primary association with these other diseases.
XBP1, FOXP3 and TSLP are the only vitiligo candidate genes that achieve suggestive association in analysis of genome-wide association study data in Caucasians. XBP1 was identified as a positional candidate based on linkage in Chinese.
SMOC2 is located very close to CCR6 and may represent the same association signal.
TICAM1 achieves highly suggestive association in analysis of GWAS data in Caucasians.
Transcriptional regulator of regulatory T-cell function and development
DNA sequencing studies
Gene-specific DNA sequencing analyses are, in effect, highly detailed candidate gene association studies, typically comparing the frequency of either specific variants or of all observed variation in specific candidate genes in cases versus in controls. Until recently, most such studies have lacked sufficient statistical power or rigor to prove that observed DNA sequence variations are truly causal for the disease versus being rare non-causal polymorphisms.
The first such study was of GCH1, and the claimed association of vitiligo with GCH1 mutations was quickly refuted. DNA sequences of a number of additional vitiligo candidate genes have subsequently been compared in vitiligo cases versus controls, mostly in relatively small numbers. These include ASIP,[107, 108] MC1R,[107-109] MYG1/c12orf10 and POMC. None of these studies showed convincing significant differences in cases versus controls after appropriate correction for multiple testing.
More recently, several vitiligo susceptibility genes detected by GWAS have been subjected to NextGeneration DNA re-sequencing to identify sequence variation that is apparently causal for disease. Jin et al. sequenced both HLA-A and TYR in Caucasian vitiligo patients, finding that the predominant HLA-A vitiligo-associated susceptibility allele is HLA-A*02:01:01:01, while finding that two common non-synonymous substitutions of TYR, S192Y and R402Q, appear to exert both individual and synergistic protective effects. HLA-A2 presents tyrosinase peptide as an autoimmune antigen, whereas the TYR 192Y and 402Q substitutions likely reduce the amount of tyrosinase peptide available for presentation, suggesting that these loci and variants act via a common pathway. Ferrara et al. sequenced GZMB and found that the causal allele involved a common multi-variant haplotype containing three non-synonymous substitutions in strong linkage disequilibrium, Q55R-P94A-Y247H, though the mechanism by which this multi-variant granzyme B affects vitiligo susceptibility is not yet known. Levandowski et al. sequenced NLRP1 and found that the common high-risk haplotype contains three non-synonymous substitutions in strong linkage disequilibrium, L155H-V1059M-M1184V, while a less common but even higher risk haplotype contains nine non-synonymous substitutions, L155H-T246S-T782S-T878M-T995I-M1119V-M1184V-V1241L-R1366C. These authors showed that the common multi-variant high-risk haplotype results in 1.8-fold elevation of processing of the inactive interleukin (IL)-1β precursor to biologically active IL-1β cytokine by the NLRP1 inflammasome, presenting a likely mechanism for disease pathogenesis associated with this haplotype.
Gene expression studies
A number of studies have compared expression of genes in normal versus vitiligo skin or melanocytes, either genome-wide or of selected candidate genes. A serious problem in such studies is the difficulty in distinguishing causal changes from those that result from the disease state or from melanocyte stress or senescence. The first such study identified VIT1, a gene downregulated in vitiligo melanocytes and later re-named FBXO11, the role of which in vitiligo pathogenesis remains uncertain. Analyses comparing genes differentially expressed in vitiligo lesions versus in uninvolved skin principally detect genes that encode melanocyte components, which is not surprising given that a greatly reduced number of melanocytes within vitiligo lesions is the hallmark of the disease. Recently, Yu et al. carried out transcriptome analyses of vitiligo patients and found evidence of upregulated innate immunity in non-lesional skin, perhaps consistent with implication of these pathways in vitiligo pathogenesis by previous genetic studies.
Other investigators have carried out differential expression studies of a very large number of candidate genes, but it is difficult to have confidence that any of the reported differences are in fact causal for vitiligo, or would even remain statistically significant could appropriate multiple-testing corrections be applied across the numerous such reports. Two such studies[89, 92] have correlated altered gene expression with genetic association of specific SNPs in the corresponding genes, and thus provide at least some external validation that the candidate gene in question may be relevant to vitiligo pathogenesis.
To date, approximately 36 loci with convincing or strongly suggestive evidence for a role in vitiligo susceptibility have been identified (Table 1). Most of these loci contain or are in close proximity to plausible biological candidate genes. Approximately 90% of these genes encode immunoregulatory proteins, whereas approximately 10% encode melanocyte proteins that likely serve as autoantigens that both stimulate the melanocyte-specific immune response and act as targets for immune recognition and cell killing (or in the case of HLA class I and perhaps class II molecules, present those autoantigens to the immune system). Together, these proteins constitute a dense immunoregulatory network that highlights systems and pathways that mediate vitiligo susceptibility. Moreover, DNA sequencing and functional analyses have identified apparent causal variation for TYR, HLA-A, NLRP1 and GZMB, yielding deeper insights into the roles played by these proteins in disease pathogenesis.
Several of the genes identified as conferring vitiligo susceptibility, particularly those expressed in melanocytes, have also emerged as genes involved in susceptibility to malignant melanoma, the same SNP having genetically opposite roles in vitiligo versus melanoma susceptibility.[81, 101] This apparent genetically inverse relationship of melanocyte-specific genes to vitiligo versus melanoma led to the suggestion that vitiligo may represent a dysregulated normal process of immune surveillance against malignant melanoma.
These genetic and functional studies have begun to provide a hypothetical framework for understanding the initial triggering, immune propagation and ultimate cytotoxicity of the anti-melanocyte immune response (Fig. 1). A likely common pathway for initial vitiligo triggering may be Köbnerization, with various types of skin damage resulting in localized cell killing and perhaps localized microinfection. Melanocyte peptide antigens are presented to skin-resident dendritic cells (principally Langerhans cells) by HLA class I molecules on the melanocyte surface. At the same time, Langerhans cells take up infection-derived molecules that display “pathogen-associated molecular patterns” (PAMP) and damage-derived molecules that display “damage-associated molecular patterns” (DAMP), which bind to NLRP1 and thereby induce assembly of the NLRP1 inflammasome; other pathways of inflammasome activation and innate immune induction are also possible. NLRP1 inflammasome assembly activates caspases that cleave the IL-1β precursor to biologically active secreted IL-1β. IL-1β is a potent pro-inflammatory cytokine, perhaps facilitating presentation of autoantigens that trigger or provide specificity to the immune response. Within the dendritic cells, melanocyte autoantigens are transferred from HLA class I molecules to HLA class II molecules, which then present these antigens to immature T cells. Immature T cells then express and secrete IL-2, which binds to IL-2 receptor expressed on the cell surface, inducing their maturation to cytotoxic T-cells (cytotoxic T lymphocytes, CTL) that express T-cell receptor molecules specific to the cognate melanocyte autoantigen, as well as cytolytic molecules such as granzyme B. Melanocyte-specific CTL then recognize the cognate melanocyte autoantigens presented on the melanocyte surface by HLA class I molecules, and the target cells are ultimately lysed by granzyme B. This “circuit” of melanocyte triggering, immune propagation, autoantigen programming and target cell killing by immune effector cells thus incorporates many of the genes and proteins that modulate vitiligo susceptibility. While certainly not complete in all details, and perhaps not even substantially correct, this model nevertheless provides for the first time a conceptualization of vitiligo pathogenesis that encompasses advanced biological knowledge of the disease and that may highlight potential new avenues for therapy and even disease prevention in individuals with high genetic susceptibility.
This study was supported in part by grants R01AR045584, R01AR056292 and P30AR057212 from the National Institutes of Health.