Article Text

Concise report
Association of an ERAP1 ERAP2 haplotype with familial ankylosing spondylitis
  1. Florence W L Tsui1,2,
  2. Nigil Haroon1,
  3. John D Reveille3,
  4. Proton Rahman4,
  5. Basil Chiu1,
  6. Hing Wo Tsui1,
  7. Robert D Inman1,2
  1. 1Genetics and Development Division, Toronto Western Research Institute, Canada
  2. 2University of Toronto, Toronto, Ontario, Canada
  3. 3The University of Texas-Houston Health Science Center, Texas, USA
  4. 4Memorial University, Newfoundland, Canada
  1. Correspondence to Dr Florence W L Tsui, Toronto Western Hospital, Mc14–419, 399 Bathurst Street, Toronto, Ontario M5T 2S8, Canada; ftsui{at}


Objectives To assess whether there is excess transmission of alleles from the ERAP1 ERAP2 locus in families with ankylosing spondylitis (AS).

Methods 199 multiplex families with AS with four non-synonymous single nucleotide polymorphisms (SNPs), three in the endoplasmic reticulum aminopeptidase 1 (ERAP1) gene (rs27044, rs10050860 and rs30187) and one in the endoplasmic reticulum aminopeptidase 2 (ERAP2) gene (rs2549782), were genotyped and family-based association analyses were performed.

Results Family-based association testing (FBAT –e; empirical variance option) analysis showed that ERAP1 rs30187[T] was associated with AS (additive model: p=0.02; dominant model: p=0.007). Haplotype permutation tests (HBAT-p) showed that a haplotype in the ERAP1 and ERAP2 locus (rs27044[G] rs30187[T] rs2549782[T]) was significantly associated with AS (two-sided p value by permutation test 0.009 for additive and 0.008 for dominant model, respectively).

Conclusion This study shows that one ERAP1 SNP and a haplotype in the ERAP1 and ERAP2 locus are associated with familial AS.

Statistics from


The pathogenesis of ankylosing spondylitis (AS) is not completely understood. HLA-B27, the major AS susceptibility gene, has an estimated attributable risk of 16–50%.1 The role of HLA-B27 remains unknown, and non-MHC genes such as interleukin 1A (IL1A)2 and IL23R3 4 are recently identified risk factors.

The ERAP1 and ERAP2 genes reside in the same cluster located on chromosome 5q15 (ref 5, figure 1A). A recent genome-wide non-synonymous single nucleotide polymorphism (nsSNP) association study3 showed that ERAP1 (endoplasmic reticulum aminopeptidase), also known as ARTS-1 (aminopeptidase regulating tumour necrosis factor receptor (TNFR) shedding 1), is the strongest non-MHC gene associated with AS, with an estimated population attributable risk of 26%. Both ERAP1 and ERAP2 are upregulated by interferon γ, form heterodimers and complementarily trim peptides >9 residues.5,,7 Enzymatic analysis showed that ERAP1 preferentially hydrolyses the leucine residues, whereas ERAP2 acts on the basic residues arginine and lysine.5 Mice lacking ERAP1 show reduced surface expression of MHC class I and altered presentation of MHC class I-associated antigens.8 In mice there is no gene coding for ERAP2. Human ERAP1, but not ERAP2, is involved in cytokine receptor shedding (TNFRI,9 IL1RII10 and IL6Rα11).

Figure 1

The ERAP1 and ERAP2 gene loci on human chromosome 5q15. (A) A schematic representation of the localisation and orientations of the ERAP1 and ERAP2 genes. There is some overlap in the 3′ end of both the ERAP1 and CAST (calpastatin) genes. (B) Haploview linkage disequilibrium plots display using the standard colour scheme. The darker the boxes, the stronger the linkage disequilibrium. The numbers in the box represent D′×100.

In the genome-wide nsSNP association study, ERAP2 showed no association with AS. Haplotype analysis within the ERAP1and ERAP2 gene cluster was not evaluated. Unlike the previous case-control study,3 family-based association studies have a lower likelihood of generating false positives as population stratification is not an issue. Here we specifically studied familial AS by assessing whether there is excess transmission of ERAP1 and ERAP2 alleles in Caucasian multiplex families with AS.

Patients and methods

AS families

The transmission study group comprised 199 Caucasian AS multiplex families (a total of 231 nuclear families); 186 of these families had individuals (affected and unaffected) from at least two generations. This study cohort was recruited from the Toronto Western Spondylitis Clinic (16 families) and other sites in the NASC (183 families). All patients met the modified New York criteria for AS.12 Family members aged >40 years with no history of AS signs or symptoms were considered as unaffected individuals. Of the affected and unaffected individuals, 60% and 47% were men, respectively. Almost all the patients with AS in the cohort of AS multiplex families (195 of 199 families) were HLA-B27 positive. However, we do not have information on B27 genotypes in about 0.07% of individuals from these families.


DNA was prepared from peripheral blood lymphocytes using standard techniques. Genotyping was performed using three SNPs in the ERAP1 gene known to be associated with AS (rs27044[C/G], rs10050860[C/T] and rs30187[C/T]) and one in the ERAP2 gene (rs2549782[T/G]). Optimised allelic discrimination assays for SNPs were purchased from Applied Biosystems (Foster City, California, USA).

Statistical analysis

Family-based association analyses

The transmission disequilibrium test (TDT) was used to test for transmission of specific alleles from heterozygous parents to affected offspring. We computed the test statistics using the family-based association testing (FBAT) software Version 1.7.2 (available online at The FBAT empirical variance option (FBAT –e) was used for testing association in an area of known linkage with multiple siblings in a family and using multiple families in a pedigree. Biallelic tests were performed using a dominant genetic model.

The PBAT (power estimation) and Haploview version 4.1 (Pairwise LD) programs were used. Haplotype analyses were performed using the Haplotype-Based Association Testing (HBAT) routine in the FBAT program. The haplotype permutation test (HBAT-p, which uses the full conditional distribution of offspring haplotypes) was used to compute an ‘exact’ p value (Monte Carlo) for each haplotype separately and for the χ2 sum and minimum observed p values among the haplotypes.


Transmission patterns of specific ERAP1 variant and AS

One hundred and ninety-nine multiplex families with AS (890 persons) were genotyped with three missense SNPs in the ERAP1 region: rs27044[C/G], rs10050860[C/T] and rs30187[C/T]). FBAT empirical option (FBAT –e) analysis showed that rs30187[T], but not rs27044[G] and rs10050860[T], was excessively transmitted in these multiplex AS families (table 1). The allele frequency of rs30187[T] was 41.2% (331/804) in the patients with AS and 35.7% (250/700) in the unaffected family members.

Table 1

FBAT –e analysis conducted in 231 ankylosing spondylitis nuclear families (199 pedigrees, 890 persons) (biallelic tests)

Haplotype-based association analyses of SNPs from the ERAP1 ERAP2 locus

We next investigated whether a missense ERAP2 SNP (rs2549782[T/G]) is associated with familial AS. FBAT –e analysis showed that this SNP was not associated with AS (data not shown). Our complete genotyping results were used to examine the linkage disequilibrium (LD) within this gene cluster using the Haploview program. Our genotypes conform with Hardy–Weinberg equilibrium and have no Mendelian inheritance errors. 97.9% of the individuals were successfully genotyped for the markers. Pairwise measures of LD are shown in figure 1B. A haplotype permutation test using the HBAT program (HBAT-p) showed that an ERAP1 ERAP2 haplotype (rs27044[G] rs30187[T] rs2549782[T]) was significantly associated with AS (table 2), using both additive (p_2side=0.009) and dominant (p_2side=0.008) models. The p values of whole marker permutation tests (using χ2 sum or minimal p) were also statistically significant (table 2). In these AS families, 39.4% (159/404) of patients and 31.8% (111/349) of unaffected family members have this critical haplotype.

Table 2

HBAT-p analysis for ERAP1 and ERAP2 variants conducted in 231 ankylosing spondylitis nuclear families†


In this study we have shown excess transmission of the minor allele of an ERAP1 non-synonymous coding SNP (nsSNP: rs30187[T]) using FBAT –e analysis of multiplex AS families. The association of this allele is modest. In the genome-wide nsSNP mapping, the association of the same allele is less robust in cases from the USA than in those from the UK,3 suggesting the likelihood of a differential profile of ERAP1 variants with ethnicity. The minor allele frequencies (MAFs) are higher in Asians than in Caucasians for both rs30187 (48% vs 30%) and rs27044 (40% vs 28%). Pairwise, the haplotypes rs27044[G] rs2549782[T] and rs30187[T] rs2549782[T] are associated with AS. In contrast, the haplotypes rs27044 rs30187 and rs27044 rs10050860 rs30187 showed no significant association with familial AS (see table 1 in online supplement), although these three SNPs are in strong LD and the haplotype rs27044[C] rs10050860[C] rs30187[T] was strongly associated with AS in three Canadian case-control cohorts.13 Owing to the relatively small sample size, we have not carried out linkage analysis of the SNPs genotyped.

Using the PBAT program with parameters close to our cohort of families with AS, the priori power estimation in the 199 multiplex families with AS was ~0.7 (dominant model) and 0.93 (additive model). This is probably an underestimate since some of our families had more than two affected individuals (3–5) and PBAT power estimation takes into account only two affected family members. Nevertheless, we have a novel finding showing that a haplotype in the ERAP1 ERAP2 loci (rs27044[G] rs30187[T] rs2549782[T]) was significantly associated with AS for both additive and dominant models. As mentioned earlier, the MAFs of some of the SNPs vary between ethnic populations. Thus, population stratification, an important issue with case-control studies in this gene cluster, has been overcome by our family-based association study using Caucasian families. Family-based studies are also more powerful for studying gene–environment interactions.14 However, for studying pure genetic associations, the power of family-based studies is lower than for case-control studies.15 The allele frequency of the AS-associated ERAP1 SNP (rs30187[T]) is higher in patients with AS (41.2%) than in unaffected family members (35.7%). From public databases, the frequency of this allele in normal Caucasians is only 30–33%, and thus is lower than we observed in unaffected family members, as expected.15

ERAP2 is mainly localised in the endoplasmic reticulum and the only function known so far is to act together with ERAP1 in N-terminal peptide trimming for MHC class I antigen presentation.6 There are no published reports indicating that ERAP2 has a role in promoting shedding of some cytokine receptors. Thus, our novel finding of an AS-associated ERAP1 ERAP2 haplotype suggests that this haplotype might influence AS susceptibility via its role in peptide trimming for MHC class I antigen presentation. The challenge is to delineate how this haplotype, alone or together with B27, affects AS susceptibility. Further work needs to be done to examine the effect of this susceptible haplotype on the HLA-B27-restricted peptide repertoire, misfolding of HLA B27 and heavy chain dimer formation. There were very few B27-negative individuals with AS (n=18) in our multiplex families to examine the interaction of B27 with this haplotype. In our cohort of multiplex families, 93.6% (131/140) of patients with AS versus 55% (55/100) unaffected family members with the ERAP1 ERAP2 [GTT] haplotype were B27-positive. Overall, 36.8% of patients with AS compared with 17.5% of unaffected family members were B27-positive and had the ERAP1 ERAP2 [GTT] haplotype (OR ~2.1).

In summary, we have identified an AS-associated ERAP1 ERAP2 haplotype which provides a framework for our future studies, especially on its contribution to disease susceptibility and its functional relevance in peptide trimming for antigen presentation.


RDI and JDR represent the investigators of the North American Spondylitis Consortium. The authors thank the Ontario Spondylitis Association and the Spondylitis Association of America for their assistance in recruiting the families included in this study, Drs N Shastri and A Paterson for helpful discussions and Laura Diekman for proficient technical help.


Supplementary materials

  • Web Only Data ard.2008.103804

    Files in this Data Supplement:


  • Funding This study was funded by grants from the Canadian Institutes of Health Research, the Arthritis Center of Excellence and by the NIH (National Institute of Arthritis and Musculoskeletal and Skin Diseases grants P01–052915-01 and R01-AR-46208 to Dr Reveille).

  • Patient consent Obtained.

  • Ethics approval The study was approved by the University Health Network Research Ethics Board and the Committee for the Protection of Human Subjects at the University of Texas Health Science Center, Houston.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.