Article Text

This article has a correction. Please see:

Extended report
Endoplasmic reticulum aminopeptidase 1 (ERAP1) exhibits functionally significant interaction with HLA-B27 and relates to subtype specificity in ankylosing spondylitis
  1. Nigil Haroon1,2,
  2. Florence W Tsui2,
  3. Barbara Uchanska-Ziegler3,
  4. Andreas Ziegler3,
  5. Robert D Inman1,2
  1. 1Division of Rheumatology, Toronto Western Hospital, University of Toronto, Toronto, Ontario, Canada
  2. 2Genetics and Development Department, Toronto Western Research Institute and University of Toronto, Toronto, Ontario, Canada
  3. 3Institut für Immungenetik, Charité-Universitätsmedizin Berlin, Freie Universität Berlin, Berlin, Germany
  1. Correspondence to Dr RD Inman, Rheumatology Department, Toronto Western Hospital, 399 Bathurst Street, Toronto, ON, Canada M5T2S8; robert.inman{at}


Objectives The functional interaction of endoplasmic reticulum aminopeptidase 1 (ERAP1) with human leucocyte antigen (HLA)-B*27 could be important in the pathogenesis of ankylosing spondylitis (AS). AS is associated with B*27:04 and B*27:05, but not with B*27:06 and B*27:09. The authors studied the surface expression of peptide-HLA(pHLA)-B27 complexes and HLA class-I free heavy chains (FHCs) on peripheral blood mononuclear cells of patients with AS with different ERAP1 single nucleotide polymorphisms. The effects of ERAP1 suppression on HLA-B*27 subtypes were tested.

Methods Peripheral blood mononuclear cells were collected from Caucasian patients with AS for flow cytometry and were stained for pHLA and FHCs. Genotyping was performed for two ERAP1 single nucleotide polymorphisms (rs27044(C/G) and rs30187(C/T)). C1R cells transfected with different HLA-B27 subtypes (B*27:04, B*27:05, B*27:06 and B*27:09) were subjected to ERAP1 suppression by small interfering RNA and stained using the monoclonal antibody (mAb) MARB4 as well as antibodies for pHLA, FHC, intracellular FHC (IC-FHC). MARB4 has been reported to bind to HLA-B27 with extended peptides.

Results The authors found variations in FHC expression on the monocytes of patients with AS, depending on different ERAP1 variants. Subsequently, using Hmy2.C1R cells in vitro, the authors show that ERAP1 suppression leads to increased IC-FHC and surface pHLA that react with the monoclonal antibody MARB4. The functional interaction between ERAP1 and HLA-B27 molecules appears to be subtype-specific, since ERAP1 suppression leads to changes only in cells expressing B*27:04 or B*27:05, but not B*27:06 or B*27:09.

Conclusions Direct or indirect alterations in the ERAP1-HLA-B27 interaction could be crucial by causing changes in peptide presentation or FHC formation by HLA-B27 molecules, as well as by contributing to differential subtype association in spondyloarthropathies.

View Full Text

Statistics from


Ankylosing spondylitis (AS) is a chronic inflammatory arthritis resulting in significant loss of quality of life and work disability. The pathogenesis of AS remains unresolved. While AS is strongly associated with human leucocyte antigen (HLA)-B27, there is subtype variability, with strong association seen with B*27:04 and B*27:05, but none with B*27:06 and B*27:09.1 It is hypothesised that subtype specificity could be due to differential rates of folding, peptide-binding ability or conformational alterations of the peptide major histocompatibility complex (MHC).2 3 In addition, genome-wide studies have recently identified endoplasmic reticulum aminopeptidase 1 (ERAP1) polymorphisms as being associated with AS.4 5 This novel discovery is appealing, since ERAP1 is an aminopeptidase resident in the endoplasmic reticulum (ER), functioning in N-terminal peptide trimming for pHLA presentation.6 ERAP1-knockout mice have an altered MHC peptide repertoire.7

We have confirmed the ERAP1 association in North American multiplex families with AS8 and have reported a lack of effect of ERAP1 polymorphisms on serum cytokine receptor levels.9 Through its role in peptide processing, ERAP1 could function in close concert with HLA-B27 molecules, and ERAP1 polymorphisms could result in an abnormal peptide-HLA (pHLA)-B27 repertoire that is connected to pathogenic immune responses.10 In addition, abnormal pHLA can be unstable and prone to misfolding.10 Misfolded HLA-B27 molecules can cause ER stress or may even be presented as surface HLA class I free heavy chains (FHCs), leading to abnormal immune interaction with a number of receptors.11,,14

We studied the effects of the two most commonly reported AS-associated ERAP1 polymorphisms on intact HLA-B27 complexes and FHC expression in peripheral blood mononuclear cells (PBMCs) of patients with AS. The ERAP1 rs30187 and rs27044 single nucleotide polymorphisms (SNPs) lead to K528R and Q730E change, respectively, in ERAP1. We then used in vitro assays to study the effect of reduction in ERAP1 levels on surface expression of pHLA, surface FHCs, intracellular FHCs (IC-FHC) and surface pHLA that react with the monoclonal antibody (mAb) MARB4 in Hmy2.C1R (C1R) cells stably transfected with HLA-B*27:05. Furthermore, we tested if there was a difference between C1R cells expressing AS-associated HLA-B27 subtypes B*2704, B*2705 and the non-AS-associated subtypes B*27:06 and B*27:09 with regard to pHLA handling after ERAP1 suppression.



Caucasian patients with AS (Modified New York Criteria) who attended the Spondylitis Clinic at Toronto Western Hospital and who consented to participate were included in the study. Patients who were on anti-tumour necrosis factor treatments or other biological agents were excluded. All patients underwent a comprehensive clinical examination. Patients were assessed for disease activity by Bath AS Disease Activity Index (BASDAI), erythrocyte sedimentation rate and C reactive protein. Functional capacity was evaluated using Bath AS Functional Index (BASFI). Peripheral blood was collected for flow cytometry studies.


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


C1R is a human lymphoid cell line with low expression of its endogenous HLA class I molecules.15 The cells have been previously transfected with different HLA-B27 subtypes (B*27:04, B*27:05, B*27:06 and B*27:09).16 These cells (a kind gift of Dr José López de Castro, Madrid) were cultured in Iscove's Modified Dulbecco's Medium with 10% fetal bovine serum. The BM36.1/B*2705 cell line has been described before.17

ERAP1 suppression

The Stealth RNAi (Invitrogen, Carlsbad, California, USA) technology was used for the small interfering RNA (siRNA) experiments. Pools of three specific RNA duplexes for ERAP1 and a pool of non-targeting duplex as negative control (NC) were used. The NC siRNA (Invitrogen, USA) is a stealth control siRNA that does not induce any stress response and is not homologous to anything in vertebrate transcriptome. The HSS (Human Stealth Select) numbers of the oligonucleotides are HSS122605, HSS122607 and HSS182221. They were delivered to the cells by nucleofection using an Amaxa Cell Line Nucleofector Kit V (Lonza, Walkersville, Maryland, USA). ERAP1 suppression was measured with western blot after 96 h. Five separate experiments were done and were compared to the NC; ERAP1 siRNA achieved a median protein suppression of 74% (range, 69%–79%; n=5). There was no significant difference between the different cell lines in the level of suppression achieved.

Western blot

Cell lysates were preparedfrom the different cell lines before and 96 h after specific ERAP1 or NC siRNA nucleofection. Cells were washed and lysed with a buffered saline containing 1% Nonidet P-40 in the presence of a mixture of protease inhibitors (Sigma-Aldrich, St.Louis, MO). The separated components were electroblotted onto a nitrocellulose membrane at 25 V overnight, following SDS-PAGE (sodium dodecyl sulfate polyacrylamide gel electrophoresis) on an 8% slab gel. The blots were incubated with mouse anti-human monoclonal antibodies against ERAP1 (R&D, Minnesota, USA) and β actin (control). A peroxidase-coupled secondary antibody was used, and the blot was developed using an ECL (enhanced chemilumescent) detection kit. Films were analysed with the Scion Image Analysis software.

Flow cytometry

Flow cytometry analysis was performed with PBMCs that were isolated from patients and with the C1R cells after siRNA treatment. Peripheral blood was drawn, and PBMCs were isolated using Lympholyte cell separation media (Cedarlane, Burlington, Canada) and then stained for flow cytometry. Intact HLA-B27 complexes were identified by the conformation-dependent ME1 antibody18, and MHC class I FHCs, by the HC10 antibody.19 Secondary staining was done using FITC(Fluorescein Isothiocyanate)-labeled goat anti-mouse IgG (Sigma-Aldrich, St.Louis, MO, USA). Since monocytes and B cells are important antigen-presenting cells, they were separately gated and studied using PE anti-CD19 (BD Biosciences, San Jose, California, USA) and antigen-presenting cells anti-CD14 (AbD Serotec, Raleigh, NC, USA) antibodies, respectively. A total of 10 000 PBMCs were acquired, and the mean fluorescence intensities (MFIs) of surface HLA-B27 and FHCs on PBMCs as a whole, and on monocytes and B cells, separately, were measured (supplementary figure 1).

Figure 1

Flow cytometry analysis of intact HLA-B27 complexes (ME1 antibody) and major histocompatibility complex class I FHC surface expression (HC10 antibody) on peripheral blood mononuclear cells of patients (N=54) with AS. Shown are the differences in FHC surface expression on monocytes (A) between patients with different genotypes of the rs27044 polymorphism of ERAP1. A representative histogram plot (B) from patients with median HC10 MFI from the two homozygous genotypic groups of the rs27044 ERAP1 SNP (CC vs GG) shows the difference in FHC on monocytes. The difference remained statistically significant (C) even after correcting for the expression of HLA-B27 complexes (ME1 MFI) on monocytes. (D) Variation in the HLA-B27 surface expression on PBMCs with the ERAP1 SNP rs27044. AS, ankylosing spondylitis; ERAP1, endoplasmic reticulum aminopeptidase 1; FHCs, free heavy chains; HLA, human leucocyte antigen; MFI, mean fluorescence intensities; PBMCs, peripheral blood mononuclear cells.

C1R cell lines (transfected with B*27:04, B*27:05, B*27:06 or B*27:09) were removed from culture 96 h after siRNA treatment. The cells were stained with the following mAb: antibodies: ME1, HC10 and MARB4. The cells were also submitted to fixation and permeabilisation by the BD Cytofix/Cytoperm kit (BD Biosciences) followed by HC10 staining for IC-FHC. Comparison to flow cytometry results on cells treated similarly with control siRNA was made. MARB4 has been reported to bind to HLA B27 with extended longer peptides.17 In addition, since the B*27:03 subtype (His59) is not recognised by MARB4 and differs from B*27:05 (Tyr59) only by a single amino acid exchange, the MARB4 epitope is likely to be located in the vicinity of the molecule's A pocket to which Tyr59 contributes.17 Peptide-binding assays employing the BM36.1/B*27:05 cell line were carried out as described previously.20


In PBMC studies, the MFIs of ME1 and HC10 staining were compared between the genotypic groups of the tested ERAP1 polymorphism using the Kruskal–Wallis Test and the Mann–Whitney U Test for recessive and dominant models, respectively. The Wilcoxon matched-pairs signed ranks test was used to compare the MFIs before and after siRNA use in the B*27:05 cell lines. The Mann–Whitney U Test was used to compare the change in the MFIs of ME1, HC10, MARB4 and IC-HC10 between C1R cells transfected with AS associated and non-associated HLA-B27 subtypes.


A total of 54 (13 women) patients were enrolled in the study with a mean (±SD) age of 41.2 ±13.7 years and disease duration of 10±9.8 years. Standardised disease assessment tools were used. The mean BASDAI, BASFI and Bath AS Metrology Index were 4.2±2.4, 3.0±2.6 and 2.2±1.5, respectively. The number of patients with the different genotypes included in the study were 22, 26 and 6 (CC, CG and GG) for rs27044 and 24, 23 and 7 (CC, CT and TT) for rs30187. There was no significant difference in the age, disease duration, BASDAI, BASFI, Bath AS Metrology Index, erythrocyte sedimentation rate or C reactive protein between the genotypic groups.

PBMCs obtained from the patients were subjected to flow cytometry to determine the surface expression of intact HLA-B27 complexes (mAb ME1) and FHC expression (mAb HC10). Patients with the major allele (C) of ERAP1 rs27044 had significantly higher (92.1 vs 27.9; p<0.01) FHC on monocytes (figure 1A,B). The ME1:HC10 ratio, which corrects for total intact HLA-B27 surface expression, was significantly lower (12.9 vs 26.7; p<0.01) in the same patients (figure 1C). The intact HLA-B27 expression on the whole PBMC population was lower in patients with the ERAP1 rs27044 minor allele (figure 1D). The ERAP1 SNP rs30187 did not affect HLA-B27 or FHC expression on PBMCs (not shown). The ERAP1 SNP rs27044 did not affect the intact HLA-B27 or FHC expression on the B cells (data not shown).

Encouraged by the results above, we tried to answer the question, is the difference in surface expression of FHCs a reflection of decreased ERAP1 activity? This would mean altered peptide trimming, as well as abnormal peptide-MHC binding and FHC formation with retention in the ER. To resolve this, we used an in vitro system to manipulate ERAP1 levels and to look at changes in IC-FHC concentration and abnormal peptide MHCs. Using siRNA, we suppressed ERAP1 expression (mean 74%, range 69–79%) in C1R cells that were stably transfected with B*27:05 (figure 2A). By flow cytometry, we measured intact HLA-B27 and surface FHCs, as well as the level of IC-FHC and a subpopulation of HLA-B27 reacting with the mAb MARB4. There was a significant increase in the IC-HC10 (55.3±24.1 vs 32.9±14.1, p=0.04) and MARB4 staining after ERAP1 silencing (8.2±3.9 vs 6.5±2.7, p=0.04), but no significant difference in the HLA-B27 complexes and surface FHC staining (figure 2B,C).

Figure 2

Following ERAP1 suppression in C1R cells using siRNA, flow cytometry was used to study changes in HLA-B27 complexes, major histocompatibility complex-I FHC, IC-FHC and MARB4-reactive HLA-B27 molecules (MARB4 mAb). Specific ERAP1 siRNA was able to suppress the protein level by 74% compared to control (NC) siRNA. (A) ERAP1 suppression in C1R cells expressing B*27:05 led to significant increase in IC-FHC (B) and MARB4-reactive pHLA (C) The change in IC-FHC was significantly higher (D) in C1R cells expressing the AS-associated subtypes (B*27:04 and B*27:05 vs B*27:06 and B*27:09). Representative histograms (E) show that the increase in IC-FHC is restricted to C1R cells with B*27:04 and B*27:05. The shaded plot and continuous line in blue represent isotype controls. The black line and dashes represent NC siRNA and ERAP1 siRNA results. Representative contour plots (F) show an increase in cells with high IC-FHC when B*27:05 expressing cells were subjected to ERAP1 suppression compared to a decrease in B*27:09 cells. The figures here reflect five separate siRNA experiments. AS, ankylosing spondylitis; C1R, Hmy2.C1R; ERAPI, endoplasmic reticulum aminopeptidase 1; FHC, free heavy chain; HLA, human leucocyte antigen; NC, negative control; siRNA, small interfering RNA.

To further analyse the specificity of MARB4, we carried out binding experiments in BM36.1 cells transfected with B*27:05 (table 1). MARB4 reacted only with B*27:05 presenting peptides with Gly in the N-terminus. Up to six amino acids could be added to the N-terminal end without abolishing MARB4 reactivity, provided that the Gly–Arg motif (Gly and Arg in the A and B pockets, respectively) was present in the peptide. These results show that MARB4 reacts with B*27:05 molecules in a peptide-dependent fashion, which is additionally restricted to peptides that are presented such that a Gly residue occupies the A pocket. Remarkably, we have found a principal structural difference between HLA-B27 molecules displaying such peptides and those with another N-terminus: the presence of Gly leads to a conformational rearrangement of several heavy chain residues at one end of the binding groove (Arg62, Glu163, Trp167) that appear to be part of the MARB4 epitope on HLA-B27 molecules.21

Table 1

Flow cytometry results normalised for background showing the specificity of MARB4 compared to MARB3

We then tested whether ERAP1 interacted differentially with distinct HLA-B27 subtypes. We compared the changes induced by ERAP1 siRNA in C1R cells transfected either with the AS-associated B*27:04 and B*27:05 or with the non-associated B*27:06 and B*27:09 subtypes. At baseline, C1R cells expressing B*27:04, B*27:05, B*27:06 and B*27:09 expressed similar levels of HLA-B27 complexes, surface FHCs and IC-FHC (data not shown). The baseline MARB4 MFI was significantly higher (p<0.01) in the AS-associated HLA-B27 subtypes (7.8±3.4 vs 3.9±1.2). After ERAP1 siRNA treatment, the IC-FHC was significantly increased in C1R cells expressing either B*27:04 (mean 2.4±1.6 fold) or B*27:05 (mean 1.8±0.4 fold), but not in B*27:06 (mean 1.1±0.3 fold) or B*27:09 (mean 1.1±0.1 fold), resulting in a significant difference (p=0.02) between the AS-associated and non-associated B*27 subtypes (figure 2D,E). The percentage of cells expressing high IC-FHC levels was consistently and similarly increased following ERAP1 suppression, but only in the AS-associated subtypes (figure 2F).

In addition, we found an increase (p<0.001) in surface MARB4 staining following ERAP1 siRNA treatment in the C1R cells expressing B*27:04 and B*27:05, but no significant difference in the case of the AS-non-associated subtypes (figure 3A). The increase in MARB4 MFI in the cells expressing the B*27:04 and B*27:05 subtypes was significantly higher (1.25±0.2 vs 0.9±0.1; p<0.01) compared to those with B*27:06 and B*27:09 (figure 3B). Similarly, the proportion of cells with high MARB4 MFI was significantly increased only in the C1R transfectants expressing the AS-associated subtypes (figure 3C). From the mean MFI values (supplementary table 1) representing five separate experiments for isotype control, IC-HC10 and MARB4, it is clear that MARB4 staining was minimal at baseline and after siRNA in the C1R cells expressing the AS-non-associated HLA-B27 subtypes. IC-HC10 staining was seen in all C1R cell lines, and MARB4 staining was present with the AS-associated subtypes.

Figure 3

(A) Following ERAP1 suppression with siRNA, the surface expression of MARB4-reactive HLA-B27 molecules increased in the C1R cells expressing B*27:04 and B*27:05 but not in those expressing B*27:06 and B*27:09. The shaded plot and continuous line in blue represent isotype controls. The black line and dashes represent NC siRNA and ERAP1 siRNA results. (B) There was a significant difference in the change in MARB4 MFI between the cell lines expressing the AS-susceptible subtypes (B*27:04 and B*27:05) compared to those expressing the non-susceptible subtypes (B*27:06 and B*27:09). (C) The number of cells expressing high-MARB4 MFI also increased with ERAP1 suppression only in the cells expressing HLA B*27:04 and B*27:05. The figures here reflect five separate siRNA experiments. AS, ankylosing spondylitis; C1R, Hmy2.C1R; ERAP1, endoplasmic reticulum aminopeptidase 1; HLA, human leucocyte antigen, MFI, mean fluorescence intensities; siRNA, small interfering RNA.


The possibility of an interaction between peptide processing pathways and HLA- B27 has been considered for some time. The recent discovery of a genetic association of ERAP1 with AS raised the hope of finally identifying the missing link. ERAP1 is associated only with HLA-B27 positive AS and not with HLA-B27 negative AS.22 This strengthens the case for a functional interaction between ERAP1 or ERAP1-trimmed peptides and HLA-B27. We have shown that HLA-B27-positive patients with the Q730 variant of ERAP1 exhibit significantly higher levels of FHCs on the surface of monocytes. In HLA-B*27 transfected cells, ERAP1 suppression was associated with an increase in both IC-FHC, as well as surface expression of a MARB4-reactive subpopulation of HLA-B27 molecules. Furthermore, our study is the first to show that ERAP1 polymorphisms can affect HLA-B27 expression in a subtype-dependent fashion and that the functional interaction between ERAP1-trimmed peptides and HLA-B27 molecules is different in cells bearing AS- and non-AS-associated subtypes.

Several polymorphisms in the ERAP1 gene strongly associated with AS have been shown to affect gene expression, possibly resulting in a variation of the enzymatic activity.22 23 The ERAP1 rs30187 SNP has been reported to be associated with hypertension24, and the K528R variant, to have reduced enzymatic activity.25 However, the ERAP1 polymorphisms reported thus far are not in the catalytic site of the molecule.23 It is also possible that these polymorphisms affect the interaction of ERAP1 with peptides. The fact that peptidase activity is affected by internal amino acid sequence variations in candidate peptides and not just the N-terminal end further supports this hypothesis.26

It is now known that the K528R (ERAP1 rs30187) and Q730E (ERAP1 rs27044) variants have decreased peptide trimming effect.27 However, there is significant variability in the enzyme activity, with these variants being more active or less active than the reference allele, depending on the substrate and its concentrations. Hence, the effects of these ERAP1 variants may change with cell type and milieu. In the elegant study by Evnouchidou and colleagues, the Q730E and K528R variants affect HLA-B27 expression on cells, with the Q730E variant having a greater effect.27 Evans et al recently reported decreased trimming by the K528R variant (rs30187) associated with AS.22 However, the Q730E variant was not studied.

Thus, from the earlier studies related to hypertension and the recent functional studies on the peptide trimming properties of ERAP1, the variants associated with AS depict a loss of function. It can be appreciated that gain and loss of function can affect normal antigen processing. What we have demonstrated is that polymorphisms of rs27044 are functionally relevant and are associated with significant changes in the surface HC10 staining population (FHC). The Q730E variant was associated with lower surface FHCs. This by itself shows that the polymorphism has a functional impact on MHC±MHC-peptide expression. FHCs can decrease on the surface as a result of either an increase in function, leading to the generation of more HLA-B27-binding peptides. It could also reflect a loss of ERAP1 function with less degeneration of the HLA-B27-binding peptides. Hence, our results cannot definitively answer the question of whether a loss of function or gain of function underlies the observations.

ERAP1 suppression using siRNA is more extreme than the effects of SNPs. Probably, many pHLAs resulting from ERAP1 suppression are so unstable that they cannot be transported to the surface. Hence, all FHCs present would be retained in the ER. This can explain why the surface FHCs were not higher after siRNA inhibition. The relevance of this finding to AS is further strengthened by the variation observed between AS-associated and non-associated subtypes. There are differences between the conformation and the flexibility of the peptide-binding grooves of these HLA-B27 subtypes.2 Changes in FHC expression have been implicated in the pathogenesis of AS.13 28 FHC dimers can be recognised by leucocyte receptors, such as killer immunoglobulin-like receptors and leucocyte immunoglobulin-like receptors, leading to abnormal immune responses.13 28 FHC accumulation in the ER can also lead to an unfolded protein response and ER stress with production of inflammatory cytokines.29

The mAb MARB4 appears to be reactive, with a distinct subpopulation of pHLA that could be relevant in AS pathogenesis, by affecting the immune response. ERAP1 suppression further exacerbated this difference, indicating a subtype-dependent functional interaction of ERAP1 and HLA-B27 molecules. Peptides with an N-terminal Gly are known to be poor substrates for ERAP1.30 Reducing the activity of ERAP1 could further increase their availability for presentation by HLA-B27 molecules. Hence, by detecting HLA-B27 complexes displaying a peptide with Gly at the N-terminus (bad for ERAP1 activity) and Arg at P2 (required for HLA-B27 binding), MARB4 could be seen as a unique antibody that identifies a pHLA subpopulation resulting from insufficient cleavage by an aminopeptidase which may be of paramount importance in antigen presentation.

A sufficient number of healthy, HLA-B27-positive controls was not available to us for testing the effect of ERAP1-SNPs on FHC expression in PBMCs. It should also be mentioned that the ME1 antibody recognises HLA-B7, -B22 -Bw42 and -B73 with high affinity and is not specific for HLA-B27.31 The patients included in the study did not have a complete HLA typing done, and so, any effects of other ME1 reactive HLA molecules, if present, are unknown. The overwhelming majority (95%) of Caucasian patients are HLA-B*2705-positive, and so, we have not done subtype analysis of the patients included in the study.

We conclude that FHC expression on monocytes is significantly modified by the Q730E variant of ERAP1 in patients with AS. Alteration of ERAP1 levels can affect the conformation and the peptide display by HLA-B27 molecules. Furthermore, these properties appear to be differentially affected in cells expressing AS-associated as opposed to non-AS-associated HLA-B27 subtypes. The ERAP1-peptide-HLA-B27 functional interaction could thus be the missing link in the pathogenesis of AS.


The authors thank Basil Chiu, Hing Wo Tsui, Angelika Zank and Christian Seitz for excellent technical support and Dr Finbar D O'Shea for help with clinical data and sample collection.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

    • Web Only Data - This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
    • Web Only Data - This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Funding Funding was provided by the Canadian Institutes of Health Research.

  • Competing interests None.

  • Ethics approval Ethics approval was obtained from the University Health Network Research Ethics Board.

  • 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.

Linked Articles

  • Miscellaneous
    BMJ Publishing Group Ltd and European League Against Rheumatism