Introduction
Ankylosing spondylitis (AS) and related spondyloarthropathies (SpA) are strongly associated with the major histocompatibilty complex (MHC) class I allele human leucocyte antigen (HLA) B27. Several theories have developed to explain the link between HLA-B27 and SpA, the classical example being based on its antigen presentation function and the possibility of molecular mimicry [
1]. However, the absence of a bona fide arthritogenic peptide and transgenic rat studies demonstrating a significant role in disease onset for CD4
+, rather than CD8
+, T cells while not ruling out a role for peptide presentation, suggests that other mechanisms may also be involved [
2,
3].
More recently, theories have emerged based on several non-antigen presentation properties of HLA-B27 [
4]. One area of particular focus has been the demonstration of misfolding of HLA-B27 in the endoplasmic reticulum (ER), which leads to induction of the unfolded protein stress response [
5]. Also, natural killer (NK) receptor recognition of non-canonical conformations of HLA-B27, in the form of heavy-chain homodimers has been reported as a potential contributor to AS development [
6]. B27 homodimers were first discovered during
in vitro MHC class I folding studies [
7], and subsequently reported in cell lines, transgenic animals and patient samples [
8‐
10]. These cell surface HLA-B27 homodimers can be recognised by NK receptors such as KIR3DL2 that do not recognise the monomeric form [
11]. Enhanced numbers of NK cells and CD4+ T cells expressing these receptors have been reported in AS patients [
12]. However, factors influencing the formation of HLA-B27 heavy-chain dimers remain poorly characterised.
Dendritic cells are essential to the initiation of most antigen-specific immune responses, as well as being involved in innate immune responses [
13]. As such they are also pivotal to the understanding of disease and autoimmune phenotypes [
14]. Recent observations into potentially abnormal interactions of HLA-B27 expressing dendritic cells with non-antigen specific T cells have brought dendritic cells into the forefront of AS research [
15].
Here we show, in a human dendritic cell-like cell line and in human monocyte-derived dendritic cells, that the formation of HLA-B27 homodimers follows maturation and activatory stimuli. Our data indicates that heavy-chain dimer formation can be a relatively transitory feature induced by activation, which may impact on dendritic cell behaviour during a critical period of a developing immune response.
Materials and methods
Cells
The human KG-1 cell line (expressing HLA-A30, -A31, -B35 and -Cw4; ECACC, HPA Cultures, Wiltshire, UK) was maintained in Iscove's Modified Dulbecco's Medium (IMDM) (Gibco, Paisley, UK), plus 20% fetal bovine serum ([FBS] Gibco, Paisley, UK) and kanamycin (Gibco, Paisley, UK). Stable transfectants of KG-1 made with cDNA for HLA-B*2705 with and without the C-terminal sv5 epitope tag [
16] were generated using the Amaxa Nucleofector (Amaxa AG., Cologne, Germany). Site-directed mutagenesis to generate mutant B27.H114D.D116Ysv5 (histidine to aspartic acid at position 114, and aspartic acid to tyrosine at position 116) was performed using Stratagene Quickchange (Stratagene, La Jolla, USA) methodology. Transfectants were selected and maintained in 1 mg/ml G418 (Geneticin, Invitrogen, Paisley, UK). KG-1 cells were differentiated/matured with 10 ng/ml phorbyl-12-myristate-13-acetate (PMA) (Sigma, Poole, UK) and 100 ng/ml ionomycin (Sigma, Poole, UK).
In agreement with the local medical school ethics committee, informed written consent was obtained from donors before blood collection. Samples were obtained from two HLA-B27-expressing individuals and two non-HLA-B27-expressing individuals, as determined by flow cytometry with fluorescein isothiocyanate (FITC) labelled-anti-HLA-B27 (VH Bio, Gateshead, UK). For primary monocyte-derived dendritic cells, peripheral blood mononuclear cells were obtained after centrifugation over Histopaque (Sigma, Poole, UK). Monocytes were allowed to adhere for two hours in RPMI-1640 medium supplemented with 10% heat inactivated FBS and kanamycin (Gibco, Paisley, UK). Non-adherent cells were then removed and fresh medium supplemented with 50 ng/ml granulocyte macrophage colony-stimulating factor (GM-CSF) and 50 ng/ml interleukin (IL)-4 was added to the culture. Dendritic cells were allowed to differentiate for four days, before treatment with 50 ng/ml lipopolysaccharide (LPS, Sigma, Poole, UK) for the indicated time periods.
Reagents and antibodies
The following antibodies were used in this study: monoclonal anti-v5 tag (pK); W6/32 recognises folded HLA-A, B and C molecules; ME1 recognises folded HLA-B molecules; HC-10 recognises partially folded HLA-B and -C molecules; 148.3 recognises human transporter associated with antigen processing (TAP)1 [
17]; anti-CD11c (Serotec, Kidlington, UK). Bodipy-ceramide (Molecular Probes, The Netherlands), goat anti-mouse HRP-conjugated secondary antibody (Perbio, Cramlington, UK) and LPS from
Salmonella enteritidis (Sigma, Poole, UK) were also used.
Flow cytometry and immunofluorescence
Cells were resuspended in PFN (phosphate buffered saline [PBS], 2% FBS, 0.1% sodium azide), stained with the indicated antibodies and FITC-labelled second stage. Mouse immunoglobulins were used as negative controls to define background staining. Samples were analysed on a FACScan (BD Biosciences, Oxford, UK) using BD Biosciences CellQuest software. For immunofluorescence microscopy cells were fixed in 2% formaldehyde in PBS, blocked with 1% BSA in PFN containing 0.2% saponin, stained with antibodies in PFN with 0.2% saponin, and stained with FITC-labelled second stage. The Golgi marker Bodipy TR C5-ceramide (Molecular Probes, Leiden, The Netherlands) was used according to the manufacturer's instructions. Stained cells were mounted with Dapi (4',6-diamidino-2-phenylindole) containing Vectashield (Vector Laboratories, Peterborough, UK). Deconvolution microscopy was performed with a DeltaVision Restoration Imaging System (Applied Precision, Issaquah, USA). A z-series of 15 to 45 images at 0.35 μm intervals was captured and processed using constrained interactive deconvolution via SoftWoRx 3.0 software (Applied Precision, Washington, USA). Further image analysis including volume sections were generated using Image J open source software.
Western blotting, biotinylation and immunoprecipitations
Cell lysates were prepared by pre-treating cells with 10 mM N-ethyl maleimide (NEM) in PBS on ice for 10 minutess, and then lysed in 1% NP40 lysis buffer with 10 mM NEM and 1 mM phenylmethanesulphonylfluoride (PMSF). Lysates were centrifuged at 14,000 rpm for 10 minutes. Protein was quantified using Bradford reagents (Sigma, Poole, UK).
For biotinylation, cells were pre-treated with 10 mM NEM on ice for 10 minutes before being labelled with 0.1 mg/ml Sulfo-NHS-biotin (Sigma Poole, UK) for 10 minutes on ice. Free biotin was quenched using Tris buffer (TBS) with 5% FBS, samples were washed twice in TBS, followed by lysis as above. Pre-cleared lysates were incubated with streptavidin-agarose (Sigma, Poole, UK) beads for 25 minutes. Beads were washed three times with lysis buffer, and resuspended in 20 μl non-reducing sample buffer.
Samples were run on 8% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE), transferred to nitrocellulose membrane and probed with the indicated antibodies or streptavidin-HRP (Sigma, Poole, UK) and signals monitored by chemiluminescence (Perbio, Cramlington, UK). Two-dimensional gel analysis was performed as previously described [
18], followed by immunoblotting as above. Pulse-chase analysis was performed by labelling cells with 3.7 MBq Trans-label (MP Biomedicals, Solon, USA), followed by lysis and immunoprecipitation with relevant antibodies and Protein G beads (Sigma, Poole, UK). Immune complexes were digested with 2.5 mU endoglycosidase H (Roche, West Sussex, UK) for one hour at 37°C prior to SDS-PAGE.
Discussion
Most immune responses, both innate and adaptive, involve the activation of multiple immune cell types, as a result of which many up-regulate the expression of components of the antigen presentation pathway, including MHC class I molecules. We therefore set out to determine whether dendritic cells expressed HLA-B27 dimers, and to what extent cell activation could induce HLA-B27 dimers in dendritic cells, which are crucial cells for most immune responses, and have recently been implicated in AS [
15]. In this study we have demonstrated the formation of HLA-B27 heavy-chain dimers in a transfected dendritic cell-like cell line, and in HLA-B27-positive monocyte-derived human dendritic cells.
Significantly, dimers were essentially undetectable in unstimulated cells, but usually appeared within 24 hours of activation. Similarly, in macrophages from HLA-B27-expressing disease-prone transgenic rats, HLA-B27 dimer-like structures are readily detected after stimulation with interferon (IFN)-γ [
5]. Thus, it is possible that these structures may only appear during an active immune response. However, since we detected dimers concomitantly with an increase in overall levels of HLA-B27 heavy-chain dimers, it may be that dimers are normally present at levels below our current detection level, a limitation that we cannot at this stage formally exclude. Nonetheless, it is possible that the immune system, since it relies on many receptor-based interactions which involve the clustering of ligands including MHC class I molecules [
25], may likewise not be able to see very low levels of dimer structures, and may therefore itself rely on cell activation to detect them. This observation could have an impact on the study of HLA-B27-associated arthritis. For example, HLA-B27-associated reactive arthritis usually develops after the significant immunological challenge of a bacterial gut infection. Similarly, the disease-prone HLA-B27 transgenic rat model is essentially disease free when kept in specific pathogen-free conditions, and only succumbs to disease when removed from these conditions [
26]. Both of these could be interpreted as sequelae to the induction of large numbers of HLA-B27 heavy-chain dimers on antigen-presenting cells, although it does not explain why other significant immune challenges, such as viral infections, are not seen to similarly trigger inflammatory arthritis.
In the case of the human monocyte-derived dendritic cells, we also observed a decrease in dimer formation between 48 and 72 hours after activation. Recent
in situ studies in the rat show that dendritic cells from the intestine migrate to mesenteric lymph nodes within 24 to 48 hours [
27]. This could result in the temporal exposure of HLA-B27 dimer structures within the lymph node, where many of the crucial interactions that determine immune responses occur.
The fact that increased expression of MHC class I molecules may predispose to heavy-chain dimer formation may be relevant in other alleles as well as HLA-B27. We detected very low quantities of dimer-sized bands in the HLA-B27-negative dendritic cell cultures (Figure
3d), and are currently investigating the nature of these bands, which may represent a novel MHC class I structure formed in non-HLA-B27-expressing cells (SL and SJP, unpublished observations). We have previously observed transient heavy-chain dimer-like structures in activated peripheral blood lymphocytes [
28]. Furthermore, in studies of the tapasin-deficient .220 cell line, when restored with tapasin and high levels of HLA-B8 we also detect the presence of dimer-like structures (SL and SJP, unpublished observations). How such dimers form in the absence of the unpaired cysteine at position 67 in the peptide-binding groove, which has been shown to be involved in HLA-B27 dimer formation [
7], remains to be determined, although contributions from non-covalent interactions in high molecular weight HLA-B27 structures have been reported [
29].
Although we have not included a significant number of human samples in this present study, nor samples from patients with actively defined AS, our data does suggest it would be of interest to determine the burden of dimer structures that may exist in non-activated and activated cells of different lineages, in both the rat transgenic model and in human cells from HLA-B27 healthy controls and patients with AS.
Conclusion
In summary, our data indicate that detectable HLA-B27 heavy-chain dimer formation may be induced in important cell types such as dendritic cells only on receiving an activatory stimulus. A comprehensive study of heavy-chain dimer formation in HLA-B27-positive individuals, both with and without AS, and its correlation with dendritic cells and other immune cell activation, is currently lacking, but would probably provide further insights into the potential availability of heavy-chain dimers as targets for interaction with T and NK cell receptors.
Acknowledgements
The monoclonal anti-v5 tag (pK) antibody used in this study was a gift from Rick Randall. The 148.3 antibody recognising TAP1 were used with permission from Robert Tampe.
SGS was funded by the Portuguese Foundation for Science and Technology, grant number SFRH/BPD/20964/2004. SL is supported by the University of St Andrews Maitland Ramsay studentship. ANA is funded by the UK Arthritis Research Campaign, grant number 15293.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
SS generated the transfectants and performed experimental work in the study, designed experiments and drafted the manuscript. SL and EC carried out immunoblotting work in the study and drafted the manuscript. AN performed site-directed mutagenesis and was involved in the drafting of the manuscript. SP designed experiments and wrote the manuscript.