Objectives Cellular expression of non-classical forms of human leukocyte antigen (HLA)-B27 (NC-B27) may be involved in spondyloarthritis (SpA) pathogenesis. We used a novel B27-specific monoclonal antibody, HD6, to ask if B27 transgenic (TG) rat splenocytes express these NC-B27 molecules. We also investigated whether B27-binding peptides could affect the expression and functional immune recognition of HD6-reactive B27 molecules.
Methods Splenocytes from B27-TG, B7-TG and non-transgenic rats, and HLA-B27+ cell lines were stained with monoclonal antibodies recognising classical (ME-1, HLA-ABC-m1) and non-classical (HD6, HC10) B27. Cells were further cultured in the presence of HLA-B27-binding peptides, or subjected to brief low pH treatment prior to mAb staining and/or immunoprecipitation or co-culture with KIR3DL2-CD3ε-expressing Jurkat reporter cells.
Results HD6-reactive molecules were detected in the majority of adult B27-TG rat splenocyte cell subsets, increasing with age and concomitant increased B27 expression. HD6 staining was inhibited by incubation with B27-binding peptides and induced by low pH treatment. HD6 staining correlated with KIR3DL2-CD3ε-expressing Jurkat reporter cell activity. Thus, IL-2 production was decreased when B27-expressing antigen-presenting cells were preincubated with B27-binding peptides, but increased following pretreatment with low pH buffer.
Conclusions Surface expression of HD6-reactive B27 molecules on B27-TG rat splenocytes is consistent with a pathogenic role for NC-B27 in SpA. Interaction of NC-B27 with innate immune receptors could be critical in SpA pathogenesis, and we show that this may be influenced by the availability and composition of the B27-binding peptide pool.
Statistics from Altmetric.com
The spondyloarthropathies (SpA) are a group of closely related chronic inflammatory diseases that share clinical features and a strong genetic association with the human leukocyte antigen (HLA)-B27. Yet the mechanism(s) by which B27 confers disease susceptibility remains unclear.1 The normal immunologic function of B27 is to form heterotrimeric complexes comprising B27 heavy chain (HC) associated with β2-microglobulin and antigenic peptides, which egress to the cell surface for subsequent CD8+ T cell recognition. B27 transgenic (TG) rats spontaneously develop inflammatory disease resembling human SpA, characterised by peripheral arthritis, psoriaform skin lesions, colitis and spondylitis. Disease is dependent on CD4+ T cells and gut flora but independent of CD8+ T cells, suggesting a role for B27 independent of its function as a classical major histocompatibility complex (MHC) class I molecule.2 ,3 Recent theories of pathogenesis have focused on the unusual biochemical properties of B27, including its ability to misfold intracellularly4–6 and to form cell surface β2m-free heavy chain homodimers (B272).7 ,8 Cell surface B272 have been shown to interact with Killer cell Immunoglobulin-like Receptors (KIR) and Leukocyte Immunoglobulin-like Receptors (LILR) in a manner different from B27 heterotrimers.9 ,10 Rodents lack KIR but express Paired-Immunoglobulin-like Receptors (PIR). PIR are structurally related to LILR and also recognise B272. The ‘free HC’ theory proposes that inflammation results from the interaction of these innate immune receptors with non-classical forms of cell surface B27 (NC-B27), including B272 and B27 free HC (FHC). We have recently generated a monoclonal antibody against B272, HD6, and have demonstrated that it specifically binds to in vitro refolded B272 and to .220 and .221 B cell lines transfected with HLA-B27 (known to express cell surface B272).11 Low level HD6 staining was also detected on CD14+ monocytes from B27+ ankylosing spondylitis (AS) patients.11
In order to explore the hypothesis that disease in B27-TG rats is driven by expression of NC-B27, we here characterise determinants of HD6 binding to B27-TG rat splenocytes and B27-expressing cell lines.
Materials and methods
PBMC samples and rat splenocytes
The rats used in this study were produced by backcrossing B27-TG (33-3, Fischer) and B7-TG (120-4, F344) rats with Piebold Virol Glaxo (PVG) rats as detailed in Utriainen et al.12 The 33-3 and 120-4 rat lines were originally produced at the University of Texas Southwestern Medical Centre.13 ,14 The B7 120-4 locus was backcrossed onto an F344 background at Hôpital Cochin, Paris. Animals were maintained under specific pathogen-free conditions at Glasgow University. All procedures were approved by the University of Glasgow Ethical Review Panel and performed under licenses from the UK Home Office. Rat splenocytes were isolated from freshly harvested spleens by pressing through a fine mesh screen and then treating the resulting single-cell suspension with Red Cell Lysis Buffer (Sigma).
Blood samples were obtained from AS patients and healthy individuals with informed consent and appropriate ethical permission (COREC/06/Q1606/139). Peripheral blood mononuclear cells (PBMCs) were isolated as described in Payeli et al.11
Cell lines and antibodies
See online supplementary methods.
Splenocytes, PBMCs and cell lines were blocked in phosphate buffered saline (PBS)/1% fetal calf serum (FCS) supplemented with 10% goat serum and stained with 0.5–2 μg primary and 0.5 μg secondary antibodies. Dead cells were excluded with LIVE/DEAD fixable violet dead cell stain kit (Invitrogen). Cytometric analysis was performed with a CyAn ADP, and data analysed using FlowJo, V.7.2.5 (TreeStar).
About 2×107 cells were incubated in blocking buffer before staining with 20 μg HD6 or HC10. Cells were lysed in buffer (0.5% Nonidet P40, 150 mM NaCl, 20 mM Tris (pH8.0), 2.5 mM iodoacetamide) containing protease inhibitors (Roche), before immunoprecipitation with goat anti-mouse immunoglobulin Dynabeads. Immunoprecipitates were resolved by denaturing sodium dodecyl sulfate–polyacrylamide gel electrophoresis under non-reducing conditions. Western blots were developed with HRP-conjugated HC10 (EZ-Link Plus Activated Peroxidase, Piercenet).
Low pH ‘acid’ treatment
Cell pellets were resuspended in a filter-sterilised low pH glycine buffer (300 mM glycine in PBS/1% bovine serum albumin, pH 2.4) for 2 min. Cell suspensions were neutralised, washed, then stained with mAbs for immunoprecipitation or fluorescence-activated cell sorting (FACS) analysis, or used in co-culture assays.
Exogenous peptide addition
HLA-B27-binding or control peptides were generated by 9-fluorenylmethyloxycarbonyl (FMOC) chemistry (>85% purity), diluted to 1 mg/ml, filter-sterilised and then added to cells at a final concentration of 50–100 μM at 37°C/5% CO2.
Jurkat co-culture assays and IL-2 ELISA
Jurkat T cells transduced with a plasmid-expressing KIR3DL2CD3ε fusion protein were incubated overnight at a 1:1 ratio with untreated or acid pretreated antigen-presenting cells (APCs) in R10 supplemented with 50 μg/ml blocking mAb (HD6, HC10, DX31 or W6/32) and/or with 100 μM B27-binding peptide or control peptide. Supernatants were harvested for IL-2 ELISA (EBiosciences Ltd, UK).
HLA-B27 TG rat splenocytes express HD6-reactive B27 molecules: both HD6 and overall B27 expression levels increase with rat age
Splenocytes isolated from adult B27-TG and age-matched non-transgenic (NT) littermates were stained with the mAbs HD6, HC10 and ME-1. HD6 stained all B27-TG mononuclear cell populations, but not B27-TG granulocytes, although the granulocyte numbers were consistently increased (figure 1A,B). Similarly, HC10 stained mononuclear cells but not granulocytes, while classical B27 molecules were detected in all gates for the B27-TG rat samples (see online supplementary figure S1). Splenocytes taken from HLA-B7-TG rats consistently failed to stain with HD6, despite similar ME-1, HC-10 and W6/32 staining (figure 1C). In contrast to adult B27-TG rats, very little detectable HD6 staining was observed on splenocytes from young B27-TG rats (figure 1D). Expression of classical B27 (figure 1D, right panels, black histograms) and HC10-reactive molecules (data not shown) was also higher in adult rats compared with young B27-TG rats. The increase in cells in the granulocyte gate seen in adult B27-TG rats was not observed in the young rats (data not shown).
Steady-state staining with HD6 is dependent upon high expression of HLA-B27 in both rodent and human cell lines
We next examined whether HD6 staining was dependent upon overall B27 expression levels in a range of B27+ human and rodent cell lines, including the Hmy2.C1R lymphoblastoid B cell line previously shown biochemically to express B272.5 ,7 HD6 stained C1R.B27 (figure 2A), but not several other B27-expressing cell lines, including RBL.B27, U937.B27, T2.B27, CEM.B27 and C58.B27 cells, despite increased HC-10 staining (see online supplementary figure S2A; data not shown). Moreover, HD6 staining was not detected in KG-1.B27 cells, even following phorbol myristate acetate/ionomycin-induced maturation (see online supplementary figure S2B), conditions previously reported to induce HC10-reactive dimers in lysates from these cells.8 However, cell surface HD6-reactive molecules were not limited to bone marrow-derived cell lines, with HD6 staining detectable on HeLa.B27 cells (figure 2B). We next used an HLA-B27 lentivirus construct and transduced .220 and .221 cells with different levels of virus. For both .220.B27 (figure 2C) and .221.B27 (data not shown), only cells transduced to express high surface levels of classical B27 exhibited significant HD6 staining. To further investigate the effect of B27 expression levels on HD6 staining, T2.B27 cells were incubated at 26°C overnight to promote surface expression of class I molecules15 and to induce HC10-reactive dimer formation.4 Lowering the temperature consistently induced HD6 staining of T2.B27 cells (figure 2D).
We next sought to establish whether HD6 could bind to B cell lines (BCLs) expressing endogenous HLA-B27 alongside the normal milieu of human HLA molecules, given that .220, .221 and C1R cells are mutant lymphoblastoid cells lacking the full complement of endogenous HLA alleles. No significant HD6 staining was observed on an AS patient-derived Epstein-Barr virus (EBV)-transformed B27+ BCL, Jesthom cells (homozygous for HLA-B27), or 4 EBV-transformed B27+ BCLs from non-AS individuals, despite the presence of HC-10 reactive molecules and comparable ME-1 reactivity to .221.B27 cells (see online supplementary figure S2C).
Incubation of cells with B27-binding peptides inhibits cell surface HD6 reactivity
Since our data indicate that high B27 levels are required for expression of HD6-reactive molecules, we hypothesised that this might result from exceeding the pool of available peptides with high affinity for B27. Previous studies have shown that the addition of exogenous peptide can stabilise MHC class I molecules at the cell surface.16 ,17 Moreover, our groups have previously demonstrated that incubation of .220.B27 cells with high affinity B27-binding peptides decreases the level of HC10-reactive B272.7 We therefore sought to determine whether the addition of exogenous B27-binding peptides similarly affected the level of HD6-reactive molecules. .220.B27, .220.B27.TPN (with restored functional tapasin expression), .220.B7 and .221B27 cells were incubated overnight in the presence of B27-binding peptides or control peptides. For B27-expressing cells, addition of B27-binding peptides reduced HD6 staining by at least 50%, and was accompanied by a reduction in HC10 but not ME-1 staining (figure 3A and online supplementary figure S3). Similar results were obtained with the C1R.B27 and HeLa.B27 cell lines (data not shown). HD6 levels were reduced 2 h after the addition of exogenous peptide, with maximal reduction detected after 8 h (figure 3B, black line). Addition of further peptide delayed recovery (data not shown), while washing cells 1 h after the addition of peptide accelerated recovery (figure 3B, black dashed line). The ability of B27-binding peptides to reduce HD6 surface levels was confirmed by western blot analysis of immunoprecipitated surface-stained HD6-reactive molecules, with bands detected corresponding to B27 monomeric and B272 species (figure 3C). Two further B27-binding peptides, GRWRGWYTY and GQWRGWYTY, which have been reported to induce a peptide-dependent B27 epitope recognised by the MARB4 antibody, were also investigated.18 Incubation of cells with either of these peptides failed to increase HD6 staining (figure 3D), confirming that the specificity of HD6 is different from that of MARB4. HD6 staining of B27-TG rat splenocytes was also reduced after 4 h incubation with a B27-binding peptide, but not a B7-binding peptide (see online supplementary figure S3).
Low pH iso-osmotic treatment induces HD6 binding
To further assess the contribution of bound peptide to HD6 recognition of NC-B27, iso-osmotic low pH ‘acid’ treatment of cells was carried out. This removes peptide and β2m, denaturing cell surface HLA class I complexes (reducing W6/32 while increasing HC10 reactivity) and allows the discrimination between peptide-free and peptide-containing HC conformations.19 Acid treatment of .221.B27 but not .221.B35 increased HD6 staining (figure 4A). This acid-induced HD6 reactivity was confirmed by western blot analysis of immunoprecipitated HD6-reactive molecules, with bands detected corresponding to B27 monomeric and dimeric species (figure 4B). Similar results were also seen with HeLa.B27 and C1R.B27 cells (data not shown). Acid treatment also consistently induced HD6 staining of EBV-transformed B27 BCLs, as well as U937.B27, RBL.B27 and CEM.B27 transfectants (figure 4C; data not shown). Western blot analysis confirmed that acid treatment increased NC-B27 on B27+ BCLs, with more HD6-reactive material observed for B27 homozygous cells compared with B27 heterozygous cells (figure 4D). Acid-induced HD6 staining was also observed in both young and adult B27-TG rat splenocyte samples (figure 4E and data not shown) but not in B7-TG or NT rats (data not shown), and in both CD3+ and CD3−, PBMCs from a B27+ AS patient but not from a B27− AS patient or a B27− healthy control (figure 4F).
Altered HD6 surface reactivity correlates with the functional interaction of APCs with a KIR3DL2-expressing T-cell line
We next asked whether altering the expression of HD6-reactive molecules could affect cellular immune recognition. We investigated whether acid or peptide treatment of B27+ APCs could affect recognition by KIR3DL2-CD3ε-Jurkat reporter cells that produce IL-2 upon KIR3DL2 ligation. For reporter cells co-cultured with .220.B27.TPN cells, IL-2 production was inhibited by a B27-binding peptide but not by a control peptide (figure 5A). IL-2 production, both with and without acid treatment, could be inhibited by HD6, HC10 and the KIR3DL2-specific antibody DX31. Conversely, IL-2 production was increased by pretreatment of .220.B27.TPN cells with acid (figure 5A). Similarly, for EBV-transformed BCLs co-cultured with reporter cells, the highest IL-2 production was seen with acid-treated B27+ BCLs (figure 5B). Increased IL-2 production, as a result of acid treatment, could be inhibited by HD6 for B27+ but not B27− BCLs (figure 5B), and by the addition of B27-binding peptides (see online supplementary figure S4).
HLA-B27 is strongly associated with the development of SpA; yet, the pathogenic mechanism(s) remain unclear. The ‘free HC hypothesis' proposes that disease results from the interaction of non-classical B27 molecules (B27 FHC and B272) with lymphoid and/or myeloid immune cells bearing innate immune receptors. Here, using a monoclonal antibody (HD6) raised against B272, we demonstrate that multiple B27-TG rat splenocyte cell subsets express cell surface NC-B27. Furthermore, in both TG rat splenocytes and human cell lines, HD6 staining is B27-specific and strongly dependent on the magnitude of B27 expression. We show that HD6 staining can be inhibited by exogenous addition of B27-binding peptides and induced by denaturation of surface molecules with low pH buffer. Altogether these data suggest an important role for B27 stability in HD6 epitope formation. We have shown that B27 expression is necessary, but not alone sufficient for the expression of surface HD6-reactive molecules. We therefore propose a model, shown in figure 6, whereby the expression of aberrant B27 (as detected by HD6) occurs once a threshold level of B27 is exceeded, as it, for example, occurs in aged B27-TG rats or T2.B27 cells after culture at 26°C. This threshold may be significantly modified by factors such as peptide supply. We propose that a minimum number of unstable cell surface B27 molecules are required for HD6-reactive molecules to be generated.
The present study is the first to investigate the surface expression of classical and non-classical B27 on B27-TG rats by flow cytometry in multiple cell populations. We observed that both HD6 and HC10 stained multiple mononuclear cell subsets but did not stain granulocytes (which had markedly lowered HLA-B27 surface expression). Two studies have previously reported intracellular HC10-reactive B272 in adult B27-TG rat samples, with one also reporting that HC10-reactive B272 are expressed at the surface of B27-TG rat DCs.20 ,21 Additionally, Taurog et al have demonstrated that disease susceptibility in B27-TG rats correlates with the B27 transgene copy number, and that, in disease-prone rats, the surface expression of classical B27 increases with age and disease onset.22 We demonstrate here that HD6 staining, though minimal on splenocytes from young B27-TG rats, was abundant on multiple splenocyte subsets from adult B27-TG rats with the inflammatory disease phenotype described in Utriainen et al.12 The presence of increased numbers of HD6-reactive molecules in diseased B27-TG rats is consistent with a role for NC-B27 in SpA-like disease. If HD6-reactive molecules do represent a pathogenic form of B27, these data also raise the interesting question of whether the appearance of HD6-reactive molecules precedes and, in doing so, triggers disease onset. An incidental but consistent finding was that, when compared with age-matched controls, the splenocytes from B27-TG rats had consistently higher proportions of granulocytes, the majority of which were CD11b. An enhanced granulocytic infiltrate has been observed macroscopically in the gastrointestinal (GI) tract of B27-TG rats,23 while a second study has reported that B27-TG rat spleens contain a higher proportion of CD11b cells compared with control littermates.24 It is unclear whether CD11b granulocytes merely represent ongoing inflammation or play a pathogenic role, perhaps through an expression of innate immune receptors, such as PIR, which can interact with NC-B27.
HD6 staining of B27-TG rat splenocytes and of B27-expressing cell lines was inhibited by culture with B27-binding peptides, but increased by acid treatment. The number of ‘acid-induced’ HD6-reactive molecules was dependent on B27 expression levels and comprised both monomeric and dimeric forms of B27. Together, these data strongly suggest that HD6 recognises a population of peptide-free monomeric and multimeric NC-B27 molecules. This is in contrast to a previously characterised B27-specific antibody, MARB4.18 Our findings are consistent with a model, whereby HD6-reactive molecules derive from unstable classical B27, wherein the stability of B27 is central in determining the amount of antibody-reactive molecules present. Polymorphisms of ERAP1 (endoplasmic reticulum-associated aminopeptidase 1) are strongly genetically linked with AS in B27+ individuals,25 and might thus cause disease through effects on B27 egress or stability (see figure 6).
The acid treatment used in our experiments could mimic the effect of long-term exposure of surface B27 to the low pH conditions associated with inflammation within joints and other affected tissues in SpA.26 This is also in agreement with a study that has demonstrated moderate changes in pH values and has the potential to destabilise classical B27 in a subtype-dependent manner.27
We also provide evidence that expression and modulation of aberrant B27 can have a functional effect through immune receptor interactions. Thus, factors affecting HD6 epitope expression also affect IL-2 production by a KIR3DL2-expressing reporter cell line. How might aberrant B27 expression cause inflammation in B27-TG rats? Although there are no direct equivalents to KIRs in rodents, the rodent PIR families are orthologous to human LILRs, and bind to MHC class I ligands in regulating of immune responses, performing a similar function to KIR. As with the human receptors LILRB2 and KIR3DL2, PIR can bind to NC-B27, including B272.9 ,10 We propose that altered PIR signalling is a likely consequence of NC-B27 expression in B27-TG rats, and postulate that PIR–NC-B27 interactions may drive the expansion of pro-inflammatory leukocyte populations. These may include TH17 cells that have been shown to be preferentially induced by DCs from B27-TG rats.28
Our data are consistent with a role for NC-B27 expression in SpA pathogenesis and highlight two potential therapeutic modalities in SpA: (1) monoclonal antibodies, such as HD6, that specifically bind to and block NC-B27; and (2) high-affinity B27-binding peptides or peptide analogues that stabilise the B27-binding groove. In support of the latter approach, a minigene containing a B27 epitope has been shown to ameliorate disease symptoms in the B27-TG rat model.29
In conclusion, we have demonstrated the presence of aberrant and potentially pathogenic NC-B27 at the surface of multiple splenocyte cell subsets isolated from B27-TG rats. Aberrant B27 molecules have the potential to interact with innate immunoreceptors and contribute to inflammation in SpA.
The authors thank Dr Antony Antoniou (University College London, UK) and Dr Simon Powis (Edinburgh University, UK) for provision of B27-transfected cell lines, and Dr Hidde Ploegh (Massachusetts Institute of Technology, MA, USA) and Dr Jo Phillips (DNAX Research Institute, CA, USA) for provision of the HC10 and DX31 antibodies, respectively.
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:
Handling editor Tore K Kvien
Contributors KM designed the experimental plan, acquired, analysed and interpreted the data, and drafted the manuscript. She is a guarantor. OR, SK, JS, LU, MHAM, SP and OM provided key reagents or biological material, and acquired and analysed data. SM, CR, PB were involved in conception and experimental design and manuscript writing and had final approval of the version to be published.
Funding This study was funded by Orthopaedic Research UK, Arthritis Research UK and the Oxford NIHR Biomedical Research Centre. PB thanks the National Institute of Health Research (NIHR) for their funding of the Oxford NIHR biomedical Research Centre and the NIHR Biomedical Research Centre in Musculoskeletal Diseases at Oxford University Hospitals NHS Trust and the University of Oxford.
Disclaimer The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.
Competing interests None.
Patient consent Obtained.
Ethics approval Oxfordshire Research Ethics Committee (COREC/06/Q1606/139).
Provenance and peer review Not commissioned; externally peer reviewed.
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.