¹Departments of Dermatology and Radiation Oncology, University of Michigan, and Ann Arbor Veterans Affairs Hospital, Ann Arbor, Michigan, USA

Correspondence: Dr James T. Elder, UM Dermatology, 1500 East Medical Center Drive, 3312 CCGC Box 0932, Ann Arbor, Michigan 48109-0932, USA. E-mail: jelder@umich.edu

Abstract

Interleukin-15 (IL-15) is a pro-inflammatory cytokine expressed in several inflammatory disorders. Zhang et al. (2007, this issue) report a significant genetic association between IL-15 and psoriasis, which may involve increased stability of IL-15 mRNA. Among its several activities, IL-15 stimulates the expression of IL-17 by T-cells. Together with recent genetic and therapeutic evidence implicating IL-23, these findings suggest a genetic basis for Th17-mediated inflammation in psoriasis.

In this issue of the Journal of Investigative Dermatology, Zhang and colleagues (2007) report an association between psoriasis and single-nucleotide polymorphisms (SNPs) in the interleukin-15 (IL-15) gene, which is located on chromosomal band 4q31.2. The study follows a genome scan from the same laboratory implicating the 4q28–31 region in Chinese families affected with psoriasis, which was recently confirmed in an independent sample of Chinese families (Yan et al., 2007). This locus, which has been named psoriasis susceptibility 9 (PSORS9), also came to light in a meta-analysis of five genome-wide linkage studies (Sagoo et al., 2004). Other than PSORS1, the major psoriasis susceptibility locus located in the MHC, PSORS9 was the most strongly linked locus identified by the meta-analysis.

To investigate the IL-15 gene in psoriasis, Zhang and colleagues (2007) changed their research strategy from linkage to association. Linkage analysis is the tracing of alleles at marker loci through families, with the precise identity of the alleles involved usually varying from family to family. In contrast, association analysis involves the comparison of the frequencies of specific alleles in cases vs. controls, with the identity of the allele involved being the same, even in unrelated individuals. In effect, the association strategy treats an entire population as one large family that has been subjected to many recombination events over time. Because of this, only markers located very close to a disease gene (50 kilobases) will yield disease associations, whereas markers residing as far as 10 megabases from the disease gene can be used to detect linkage. Although linkage analysis is the method of choice for Mendelian disorders, the association strategy is much more powerful for the so-called "complex genetic disorders." For the most part, these disorders involve common genetic variants, which are common because they arose long ago (Risch and Merikangas, 1996). Such genetic variants do not generally act alone—rather, they act in concert with other common genetic variants to provoke disease, and each individual variant is best thought of as a determinant of disease susceptibility, rather than "the cause" of disease. The search for these common determinants of susceptibility is rapidly gaining momentum across a wide variety of common diseases, aided by the development of a dense map of human haplotypes (the HapMap) and microarray technology for SNP typing. Capable of typing an individual for up to a million SNPs, microarray technology allows an unbiased search of the entire genome for disease associations. The major challenge of these studies is their expense, but genotyping costs continue to decrease and the number of such genome-wide association studies is increasing rapidly.

Sample size is a critical factor for a successful disease gene search via the association strategy. Zhang and colleagues (2007) collected a sample of 632 unrelated psoriatic patients and 485 healthy controls, a sample neither large nor small by today's standards. The power of a given sample to detect associations depends on the disease prevalence, the disease allele frequency, the relative risk posed by possession of the variant, and the genetic model. Utilizing software available on the Web (Power Calculator for Genome Wide Association Studies, http://www.sph.umich.edu/csg/abecasis/CaTS), it can be estimated that this sample has from 29 to 52% power to detect a disease allele conferring the properties reported in this study, depending on disease model (dominant 29%, additive 42%, multiplicative 52%). (These values were calculated for a relative risk of 1.65 at a significance level of 0.00005, which are the values reported for the most strongly associated haplotype identified in this study.) Thus, it could be argued that, despite the positive results of this study, larger sample sizes may be needed to replicate this genetic effect. Certainly, replication is essential for validation of any association study in a complex genetic disorder.

To determine the optimal set of markers for typing, Zhang and colleagues (2007) sequenced the eight known exons of the IL-15 gene and their boundaries, as well as about 2,000 base pairs of the 5 and 3 flanking regions, in 48 cases and 48 controls. This effort identified 19 variants, of which 14 were already known and deposited in the SNP database. This illustrates the considerable power of the SNP database to identify SNPs by database searching, rather than laborious sequencing, as has been necessary in the past. They next determined the patterns of linkage disequilibrium within the IL-15 gene, allowing them to select a subset of 12 markers that best described the differences between individuals across the gene. Their result was very similar to that obtained by querying the HapMap database for the Han Chinese population in Bejing, China, illustrating the power of existing databases to identify optimal markers for genotyping. The most significantly associated SNPs in this study were located in the 3 untranslated region (3-UTR) of the IL-15 gene. The most significant individual marker reached a P value of 0.00006 after correction for testing of 12 markers, with the rarer T allele conferring increased risk. Although the subject of correction for multiple testing always provokes a lively debate, the correction factor used here is reasonable given that there is already linkage evidence for a gene in this region and that the markers used were not in linkage disequilibrium with each other. (A statistical purist might call for more stringent correction, because there are 65 genes in the linkage interval (Yan et al., 2007).) Next, haplotypes were inferred, again with good statistical respect for the multiple testing issue. (Haplotypes are defined as the set of alleles at in a group of linked loci that are present on a single (haploid) chromosome.) Reassuringly, these calculations revealed that the most strongly associated haplotype yielded P values similar to those obtained for the most strongly associated individual marker, and each of the two haplotypes carrying the rare allele (T) at that marker conferred increased risk.

Zhang and colleagues (2007) went on to begin to explore the functional role of the 3-UTR by cloning different variants of the 3-UTR into a luciferase reporter construct and transfecting these constructs into Jurkat cells. Interestingly, the two haplotypes carrying the T allele yielded 34 to 53% higher levels of luciferase activity than did the most common, non-disease-associated haplotype. Although the precise mechanism for this effect remains to be identified, there are numerous precedents for sequences in the 3-UTR to control mRNA levels via elements that regulate RNA stability.

IL-15 is an excellent candidate for involvement in psoriasis. It has been implicated in several other inflammatory disorders, including rheumatoid arthritis, diabetes, and pulmonary inflammation (McInnes and Gracie, 2004). IL-15 has structural similarities with IL-2 and interacts with the - and -chains of the IL-2 receptor, differing from IL-2 in that IL-15 interacts with a unique -chain (IL15R) distinct from the IL2R, which recognizes IL-2. Keratinocytes express both IL-15 and IL15R, suggesting a mechanism for interaction with other immune and/or inflammatory cells via IL-15 and its receptor (McInnes and Gracie, 2004). Two studies have found overexpression of IL-15 in psoriatic lesions, albeit using different methods and producing somewhat different results (McInnes and Gracie, 2004). IL-15 is also a growth factor for CD8+ T-cells, which infiltrate the epidermis during the development of psoriatic lesions. IL-15 also triggers inflammatory cell recruitment, angiogenesis, and production of other cytokines implicated in the pathogenesis of psoriasis, including IFN- and TNF-. A functional role for IL-15 in psoriasis was suggested by studies of psoriatic skin xenografted onto immunocompromised mice. Injection of an antibody targeting IL-15 bound to its receptor led to pronounced reduction in the severity of psoriatic lesions, including reductions in epidermal hyperplasia, parakeratosis, and inflammatory cell infiltration (Villadsen et al., 2003). Thus, based on both genetic and functional data, IL-15 is a prime candidate for involvement in the pathogenesis of psoriasis. Efforts to replicate the genetic findings are clearly in order and, if successful, should drive a major effort to explore the immunologic functions of IL-15 in psoriasis.

After a long period of uncertainty, it would appear that the genetics of psoriasis are coming into focus with the identification of specific susceptibility genes. In addition to the IL-15 findings reported in this issue, HLA-Cw6 was identified as the likely disease allele at PSORS1 (Nair et al., 2006), and two studies identified associations between psoriasis and the SLC12A8 gene at the PSORS5 locus (3q21) (Huffmeier et al., 2005). Another recent study identified a strong association with the p40 subunit of IL-23 (IL12B) and its receptor (IL23R) (Cargill et al., 2007). The study by Cargill et al. was a replication of an association between psoriasis and IL12B reported in 2002 (Tsunemi et al., 2002) and, moreover, identified associations with IL12B in each of three independently ascertained case–control samples, thus providing ample evidence for replication. In further support of an important role for IL-23 in psoriasis, a monoclonal antibody targeting IL-23 was recently shown to be highly effective against psoriasis in a large double-blind, placebo-controlled clinical trial (Krueger et al., 2007). IL-23 has been implicated in supporting the proliferation and survival of the Th17 subset of CD4+ T-cells, which in turn was recently implicated in a variety of autoimmune disorders (Iwakura and Ishigame, 2006). The IL-23 receptor has recently been implicated in Crohn's disease, which may explain the long-standing observation that psoriasis is eight times more common in Crohn's disease patients than in controls. Interestingly, a recent study showed that IL-15 was even more effective than IL-23 in stimulating the production of IL-17 by human T-cell blasts (Hoeve et al., 2006). Assuming that the findings of Zhang et al. (2007) are confirmed in other populations, we can expect lively exploration of the cellular and biochemical pathways relating the genetic defects in IL-15 and IL-23 to the function of Th17 T-cell subset and subsequent tissue injury, not only in psoriasis but also in other inflammatory and autoimmune disorders. As genome-wide association scan technology is increasingly applied to psoriasis, it would not be surprising if other genes residing along these pathways are also implicated in its pathogenesis.