Key points

Introduction

Genetic studies of ankylosing spondylitis (AS) have, over the past decade, provided major insights into the etiopathogenesis of the disease that have led to major therapeutic innovations. Some of these new treatments have already entered clinical practice, and others are in trials and undergoing development. It is well known that susceptibility to and severity of AS are largely genetically determined. Extensive progress has been made identifying susceptibility alleles in the disease, with 113 established loci identified, contributing roughly 10% of the heritability of AS, over and above the major effect of human leukocyte antigen (HLA)-B27, which determines approximately 20% of the genetic risk. Studies of the genetics of clinical manifestations of AS, such as the extent of bony ankylosis or presence of anterior uveitis, have been more challenging, though some genes have been found to influence uveitis risk beyond their effects on the risk of AS. This article seeks to present the current state of understanding of the genetic influences in AS, focusing on more recent advances and their contribution to understanding mechanisms of disease.

Introduction

Genetic studies of ankylosing spondylitis (AS) have, over the past decade, provided major insights into the etiopathogenesis of the disease that have led to major therapeutic innovations. Some of these new treatments have already entered clinical practice, and others are in trials and undergoing development. It is well known that susceptibility to and severity of AS are largely genetically determined. Extensive progress has been made identifying susceptibility alleles in the disease, with 113 established loci identified, contributing roughly 10% of the heritability of AS, over and above the major effect of human leukocyte antigen (HLA)-B27, which determines approximately 20% of the genetic risk. Studies of the genetics of clinical manifestations of AS, such as the extent of bony ankylosis or presence of anterior uveitis, have been more challenging, though some genes have been found to influence uveitis risk beyond their effects on the risk of AS. This article seeks to present the current state of understanding of the genetic influences in AS, focusing on more recent advances and their contribution to understanding mechanisms of disease.

Major histocompatibility complex and ankylosing spondylitis

Large scale case-control studies of HLA and other major histocompatibility complex (MHC) genes in AS have demonstrated that the genetic associations at this locus are far more complex than initially thought. Since the discovery of the association of HLA-B27 with AS, there have been many studies suggesting the presence of additional MHC-associated variants. With the exception of the association of HLA-B60 with AS, until recently, none of those have been convincingly replicated.

The MHC is under marked genetic selection pressure and HLA frequencies vary substantially between ethnic groups. The development of methods of HLA-typing using imputation from dense single nucleotide polymorphism (SNP) genotyping, together with the availability of large reference sets of subjects genotyped at both HLA loci and MHC SNPs, has enabled analysis of HLA and MHC associations in large case-control cohorts. Another methodologic advance in recent studies is principal components analysis; population stratification can be identified and controlled for, making the findings robust to differences in allele frequencies due to ethnic variation rather than disease affection status. Two such studies have now been published, 1 in subjects of European ancestry and the other in Koreans. Both show convincingly that there are additional HLA-B variants associated with AS, as well as other HLA class I and II variants ( Table 1 ). Although these studies do not exclude the presence of other non-HLA MHC genetic associations, they do indicate that it is unlikely that variants of large effect exist within the MHC once the associations of HLA variants are accounted for.

The additional HLA-associations likely contribute to the known association of AS with other diseases. For example, HLA-B51 , which is also a risk variant for AS, is the major risk allele for Behçet syndrome, a condition that can be complicated by sacroiliitis. HLA-DRB1*0103 is AS-associated and is among the major risk alleles for Crohn disease, which frequently co-occurs with AS. HLA-B7 has a major protective effect on AS and, interestingly, is used as a control allele in transgenic rats. In this model, excess copy numbers of HLA-B27 induce a spondyloarthropathy, whereas rats with similar excess copy numbers of HLA-B7 remain healthy.

Using dense MHC SNP genotyping data, the amino-acid composition of the HLA alleles can be imputed and tested for association with disease, enabling the identification of the key amino-acids involved. This approach has been successfully used to extend the known amino-acids in the rheumatoid arthritis shared epitope, which provides a functional explanation for the association of HLA-DRB1 alleles with that disease. In AS, in the populations studied, the identity of the amino-acid at position 97 in HLA-B was shown to determine the direction of association of the major HLA-B alleles with the disease ( Table 2 ). This does not mean that this amino acid alone is the sole HLA-B determinant of AS risk; rather it indicates that, in the context of the HLA-B alleles involved, this amino acid is a key determinant of disease risk.

Aminopeptidases and ankylosing spondylitis

A key recent discovery has been the demonstration that variants of the M1-aminopeptidase gene, ERAP1 , are associated with AS, and interact genetically with both the AS-associated HLA class I alleles HLA-B27 and HLA-B*4001. Thus, ERAP1 is only associated with AS in HLA-B27 positive cases, or HLA-B27-negative/HLA-B*4001 positive cases. The same ERAP1 haplotypes interact with HLA-Cw6 in psoriasis, and HLA-B51 in Behçet disease. This locus is also strongly associated with the rare ocular uveitis birdshot retinopathy, which is strongly associated with HLA-A29, though in this disease the number of subjects studied is too small to determine if the disease association is with ERAP1 , the neighboring related gene ERAP2 , or both. The ERAP2 association with AS is present in both HLA-B27-positive and HLA-B27-negative disease, suggesting some subtle difference in its functional mechanism in causing AS. For example, ERAP2 may potentially affect peptide handling by other AS-associated HLA class I alleles for which ERAP1 peptide cleavage is less influential.

A recent paper has suggested that haplotypes of ERAP1 variants are more strongly associated with disease and have greater functional effects than individual disease-associated variants. However, the small sample size of this study (19 cases and 17 controls) is too few to distinguish haplotypic from single variant effects, and many of the haplotypic associations reported were not statistically significant. The study also reports the opposite direction of association of the key AS ERAP1 nonsynonymous SNP, rs30187, compared with all other studies, which included more than 1000 times more cases and controls, in multiple ethnic groups. In the absence of further supportive evidence, this study should be considered hypothesis-generating.

Extensive studies of the functional mechanism of the associations of ERAP1 and ERAP2 variants and AS are underway. Both proteins are involved in peptide trimming in the endoplasmic reticulum (ER), changing particularly the length but also the amino acid composition of peptides available for HLA class I presentation. Proposed mechanisms of association include

• •
  

  Effects on the peptidome presented by HLA-B27 and thus leading either to presentation of arthritogenic peptides or failure to present disease-protective peptides

  
  
• •
  

  Effects on HLA-B27 free heavy chain expression and killer immunoglobulin-like receptor (KIR) interactions, in turn influencing activation of interleukin (IL)-17 producing immune cells

  
  
• •
  

  Effects on HLA-B27 folding and ER accumulation leading to ER stress reactions.

It is beyond the scope of this article to discuss these hypotheses and studies in detail. However, a common feature of each model is that variants that are disease-protective in AS exhibit reduced peptide cleavage function. This and the fact that, apart from in Behçet disease, there is no convincing evidence in either humans or animal models that ERAP deficiency increases disease risk, has led to programs targeting these proteins as therapeutics for AS and related diseases.

T cells and ankylosing spondylitis

How genetic variants predisposing to immune-mediated diseases are tied to altered immune system activity is a question of primary importance. Despite being among the first proposed mechanisms of disease development in AS, strongly supported by the robust ERAP1-HLA-B27 epistasis identified, it has not been conclusively demonstrated that auto-reactive T cells recognize a B27-restricted peptide in AS patients to cause disease. T-cell receptors (TCRs) develop by the process of random recombination of numerous encoded gene segments to generate highly variable and cell-specific receptor chains. Unique cell surface TCRs on each T cell have the potential to engage different combinations and conformations of HLA bound peptides ( Fig. 1 ). Clonally expanded T cells that have recognized an antagonistic antigen and proliferated to hone adaptive immune responses are a feature of many immune-mediated diseases and contribute to targeted inflammation. Little evidence of such occurrence in AS patients has been published since the work of Mamedov and colleagues in 2009, which characterized the T-cell populations of 2 patients and found stably expanded clones consistently representing between 5% and 50% of the profiled repertoire over several years. These T cells were found to be cytotoxic and proinflammatory in nature, predominantly CD8+/CD27-/CD28- (regarded as terminally differentiated effector or cytotoxic T cells), and expressed TCRs with homology to several previously reported clones in reactive arthritis, rheumatoid arthritis, and other AS subjects. Earlier work demonstrated that a nonamer of the HLA-B*2705 molecule hypervariable region itself can be recognized by cytotoxic T lymphocytes in AS subjects. Convergence of TCR sequences used by these nonamer-responsive T-cell populations found in the peripheral blood and synovial fluid of numerous AS subjects suggested a mechanism by which self-reactive T cells contribute to inflammation in a B27-dependent fashion. However other studies including in discordant twin pairs found no sharing of CD4+ or CD8+ peripheral T cell receptor Vβ repertoire. TCR involvement in AS has also been looked at from the point of view of germline variable region genes, but neither linkage nor association of the TCRA or TCRB loci have been demonstrated with AS, suggesting that these loci are not important in the familiality of the disease.

KIR gene usage and interacting partners, and their effects on natural killer (NK) cell activation or inhibition.

The future of immunogenetics studies in immune-mediated diseases with suspected autoreactive T-cell involvement will be in the high-resolution profiling of T-cell populations. Although yet to be conducted in an AS cohort, new immunosequencing techniques applied to a diversity of conditions, including juvenile idiopathic arthritis (JIA), has provided insight into the underlying nature of T-cell responses in disease. The JIA study demonstrated a restricted TCRB repertoire in the peripheral blood and synovial fluid regulatory T cell (Treg) population of patients, with patient sharing of expanded clonotypes lacking in healthy children. Results suggested either appropriate but inefficient control of inflammatory processes by Tregs, which are typically an immensely diverse T-cell population, or perhaps pathogenicity of the expanded clones in disease. Clonal restriction of relevant T-cell populations in AS may very well be detected with the profiling of hundreds of thousands of TCRs in this fashion, providing support for the arthritogenic peptide model of disease.