There is an oligogenic component to the aetiology of ankylosing spondylitis (AS) of which the best known component is the HLA-B27 immune response gene. A meta-analysis of four whole genome linkage scans also shows linkage to several chromosomes, including 19q13 (marker D19S902) adjacent to the killer immunoglobulin-like receptor (KIR) genes.1 These are expressed on natural killer (NK) cells and certain subsets of T lymphocytes. NK cells play an essential part in innate immunity, particularly against tumours and virus-infected cells. The activity of these cells is determined by the balance between activating and inhibitory signals transmitted by a range of receptors, including KIR, leucocyte immunoglobulin-like receptors (LILR) and the CD94/NKG2 family of receptors, expressed on the NK cell surface.

The KIR gene cluster spans approximately 150 kb in the leucocyte receptor complex and has been previously implicated in inflammatory conditions, such as rheumatoid arthritis and psoriatic arthritis.2 3 The KIR region is highly diverse; different KIR haplotypes exhibit differences in both gene content and allelic variation (eg, KIR3DL2). Engagement of HLA class I allotypes by long-tailed KIR (L), with the exception of KIR2DL4, results in the transmission of inhibitory signals while short-tailed KIR (S) transmit activating signals. Downregulation of HLA expression in infected or transformed cells thereby increases vulnerability to NK activity.

The pathological role of HLA-B27 in AS remains unclear. It forms a non-covalent association with β₂-microglobulin (β₂m) but it can also form β₂m-free B27 heavy-chain homodimers.4 These B27 homodimers, bonded through an unpaired cysteine residue at position 67, are expressed on peripheral blood mononuclear cells from patients with spondyloarthropathy.5 KIR3DL1 and KIR3DL2 both bind the HLA-Bw4 public epitope possessed by all HLA-B27 types except HLA-B2708.6 Only B27 homodimers bind to KIR3DL2, while both the heterodimeric B27/β₂m and B27 homodimers are ligands for KIR3DL1.7 Increased numbers of NK cells expressing KIR3DL2, but not KIR3DL1, in the peripheral blood mononuclear cells and synovial fluid of patients with spondyloarthropathy have been observed. Interactions between B27 homodimers and KIR3DL2 appear to protect KIR3DL2-positive NK and T cells from apoptosis by inhibiting interferon-γ production and activation.8

An association between KIR3DL1/KIR3DS1 and AS has recently been described.9 KIR3DL1 and KIR3DS1 represent two distinct but mutually exclusive loci, giving the appearance of alleles segregating at the same locus. We have investigated whether the total KIR gene content or specific KIR3DL2 alleles play a part in AS in British Caucasians.

METHODS

Cases and controls

We genotyped 200 cases with AS for 14 KIR genes and subtyped 368 cases with AS for 12 KIR3DL2 alleles. We also typed an additional 61 B27-positive cases with AS for KIR3DL1 and KIR3DS1 alone. Patients were recruited from outpatients attending the Nuffield Orthopaedic Centre (Oxford, UK), the Royal National Hospital for Rheumatic Disease (Bath, UK) or from an appeal to members of the National Ankylosing Spondylitis Society (UK) following informed consent (MREC project number 98/5/23). All patients were British Caucasians, fulfilling the modified New York criteria for AS. Comparison was made with 554 healthy, ethnically matched blood donors, of whom 405 were typed for all 14 KIR genes, 149 were typed only for KIR3DL1 and KIR3DS1 and 366 were typed for 12 KIR3DL2 alleles. DNA was prepared from peripheral blood leucocytes by standard methods.

HLA-B27 typing

All cases with AS and controls were typed for HLA-B27 using allele-specific polymerase chain reaction as previously described.10

KIR genotyping

KIR genotypes were obtained using a multiplex polymerase chain reaction–single strand conformation polymorphism method for the KIR genes 2DL1, 2DL2, 2DL3, 2DL4, 2DL5, 3DL1, 3DL2, 3DL3, 2DS1, 2DS2, 2DS3, 2DS4n, 2DS4d, 2DS5 and 3DS1 as previously described.11 The frequency of each KIR gene was expressed as the inferred phenotype frequency in the sample population.

KIR3DL2 genotyping

Sequence-specific oligonucleotide probes were used to define the original 12 KIR3DL2 alleles named by the KIR Nomenclature Committee (http://www.ebi.ac.uk/ipd/kir/).12 One minor drawback of this method is the existence of a few ambiguous typing patterns where we were unable to differentiate between three sets of allele pairs. The affected combinations are KIR3DL2*001, *011 producing the same pattern as KIR3DL2*003: *010; KIR3DL2*001, *007 producing the same pattern as KIR3DL2*006, *010; and KIR3DL2*003, *007 producing the same pattern as KIR3DL2*006, *011. Previous studies in UK families have actually demonstrated that in every case the first combination (of those listed above) was correct.12

Statistical analysis

Differences in KIR gene and KIR3DL2 allele frequencies between cases with AS and controls were tested for significance by χ² test. Student t-test was used to compare the number of activating and inhibitory genes present in the KIR genotypes of the cases with AS and the controls.

RESULTS

KIR genotyping

The inferred KIR phenotype frequencies are shown in table 1A. The framework genes KIR2DL4, KIR3DL2 and KIR3DL3 were present in all individuals as expected and every individual also carried either one or both of KIR3DL1/KIR3DS1 and KIR2DL2/KIR2DL3, which segregate as alleles at the same locus. There was no difference in the frequency of any of the 14 KIR genes in cases and controls.

The most frequent genotype in the cases with AS and controls was that corresponding to the homozygous “A” haplotype (2DL1, 2DL3, 2DL4, 3DL1, 3DL2, 3DL3 and 2DS4), which was carried by approximately one-third of all individuals studied (fig 1). Over 80% of individuals were represented by the 10 most frequent genotypes. No specific genotype was over-represented in AS and there was no difference in the average number of activating or inhibitory KIR genes in cases with AS compared with controls.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

The 10 most common KIR genotypes and their frequency in cases with ankylosing spondylitis (AS) and controls. Black box = locus detected.

KIR and HLA

There is some evidence that the HLA and KIR gene regions, while segregating independently on different chromosomes, may have co-evolved. For example, a weak negative correlation between KIR3DS1 and its ligand the Bw4 epitope has been described.13 The original association between AS and KIR3DL1/3DS1 was also reported in B27-positive cases and controls.9 We therefore tested if B27 status affected KIR3DL1 and KIR3DS1 gene frequencies in the UK population (table 2). However, KIR3DS1 and KIR3DL1 gene frequencies were very similar in B27-positive and B27-negative controls (and also the previously published B27-positive controls9 (table 2)). Therefore, it appears unlikely that B27 status alone affects KIR3DL1/3DS1 gene frequencies and it was justified to use a control population, including both B27-positive and B27-negative individuals.

KIR3DL1/KIR3DS1 and ankylosing spondylitis

Finally, as indicated in table 2, the gene frequencies of KIR3DL1 and KIR3DS1 in our cases with AS were not significantly different from the healthy controls (OR = 1.5, 95% CI 0.8 to 3.0 and OR = 1.02, 95% CI 0.2 to 5.3, respectively) or when both groups were stratified for B27 status (OR = 1.5, 95% CI 0.6 to 3.6 and OR = 0.93, 95% CI 0.4 to 2.0, respectively). This study had an 80% power to detect to an OR of 1.6 for the recessively inherited KIR3DL1 based on a risk allele frequency of 0.8 and a disease population prevalence of 0.001 for AS. Similarly for KIR3DS1 there was an 80% power to detect an OR of 2.4 based on recessive inheritance and a risk allele frequency of 0.2.

KIR3DL2 genotyping

The KIR3DL2 allele frequencies are shown in table 1B. The frequencies of each individual KIR3DL2 allele and combined analysis of all alleles were very similar in cases with AS and controls.

DISCUSSION

In contrast to a previous study from Spain and the Azores showing an increase in KIR3DS1 in AS9 our study suggests that the presence or absence of certain KIR genes, specifically KIR3DL1 or KIR3DS1, does not influence susceptibility to AS. We have also shown that KIR3DL2 allelic variation does not have functional consequences that affect susceptibility to AS. The increase in NK cells expressing KIR3DL2 in spondyloarthropathy cannot be explained by KIR3DL2 allelic variation. This result would be consistent with the hypothesis that infection-associated peptide(s) presented by HLA-B27 during infections could promote its misfolding to form B27 homodimers.8 As ligation of KIR3DL2 to B27 homodimers inhibits NK cell apoptosis, overproduction of B27 homodimers could result in increased survival of KIR3DL2-positive NK and T cells.

The KIR gene region is highly polymorphic, rivalling that of the major histocompatibility complex. Before rejecting the possibility that KIR genes are involved in AS pathogenesis, it is important to recognise that the KIR repertoire in a given individual reflects not only the presence or absence of particular genes but also the different alleles encoded at some KIR loci. Expression analysis also reveals that individual NK cells from a given individual express different combinations of KIRs encoded within their genome.14 This phenomenon is also variable between individuals with the frequency of expression of individual KIR genes by NK cells varying by up to 45%.15

A role for KIR in AS remains possible but only a large systematic study of all allelic variants at all the KIR loci and expression analysis could formally exclude the possibility of a genetic association with AS. Our larger UK study has not confirmed the association with KIR3DL1/KIR3DS1 reported in Spanish/Azorean patients.9 Although population differences cannot be entirely excluded it is more likely that the previously reported association is artefactual and needs replication in that population.