Antigen presentation of Type II collagen in rats

Abstract

Collagen-induced arthritis (CIA) is a T-cell dependent disease of rats which follows immunization with bovine type II collagen (bCII). Susceptibility to CIA is linked to the genes encoding the major histocompatibility complex (MHC), suggesting that antigen presentation is important in disease pathogenesis. Antigen-presenting cells (APC) (macrophages, dendritic cells (DC) and B cells) were prepared from WA/KIR/KCL rats and presentation of antigen, in the form of native protein (bCII) or synthetic peptide (bCII:184–198), was assessed in T-cell proliferation assays. Whilst macrophages inhibited proliferative responses to bCII, splenic or thymic low density cells, enriched for DC, presented both bCII and bCII(184–198) peptide. However, bone marrow-derived DC, which stimulated T-cell responses to OVA, failed to present bCII, suggesting differences in processing of these two antigens. B-cell depletion from lymph node cells abrogated the proliferative response to bCII and reconstitution of a T-cell population with B cells restored the proliferative response, indicating that B cells are important for stimulating T-cell responses to bCII. B cells play a critical role in CIA by producing pathogenic anti-bCII antibodies, and we propose that B cells are also important APC which present bCII to CD4⁺ T cells.

INTRODUCTION

Collagen-induced arthritis (CIA) is a chronic inflammatory arthropathy, induced in susceptible rodents by immunizing with bovine type II collagen (bCII) [1]. Arthritis is associated with immune complex formation and complement activation within the joint, followed by infiltration of the synovium with inflammatory cells [2]. Susceptibility to CIA is linked to the genes of the major histocompatibility complex (MHC) [3], and CD4⁺ T cells play a critical role in disease pathogenesis [4]. There is an abundance of MHC Class II⁺ cells in arthritic joints [5], and it has been suggested that the interaction between these synovial antigen-presenting cells (APC) and the infiltrating CD4⁺ T cells is responsible for the sustained inflammatory response [6]. The MHC association with disease has led to the proposal that particular MHC Class II molecules present distinct peptides of type II collagen, resulting in activation of a pathogenic subset of CD4⁺ T cells [7].

Dendritic cells (DC), macrophages and B cells can all stimulate CD4⁺ T cells to proliferate and secrete cytokines [8]. It has been proposed that DC are the only population of APC that can activate naïve T cells, whereas macrophages and B cells function to present antigen to memory T cells [9]. However, activated B cells have been shown to present self protein to naïve autoreactive T cells [10], and injection of mice with activated macrophages has been shown to result in autoimmune orchitis, a T-cell-dependent disease [11]. Thus, it is likely that under the appropriate conditions, DC, macrophages and B cells can each induce autoreactive T-cell responses.

Studies in mice have provided evidence that macrophages are the most important APC involved in processing and presenting bCII [12]. Similarly, Manoury-Schwartz et al. [13] argued that the size and fibrillar nature of the native bCII molecule influences antigen uptake and processing, and that macrophages are the only type of APC able to present this antigen. However, rat macrophages are inefficient APC [14] and inhibit T-cell responses rather than stimulating them [15], suggesting that antigen presentation of bCII by macrophages might differ between rats and mice.

Previous work would predict that antigen presentation of bCII, and stimulation of T-cell responses, are important events in the pathogenesis of CIA in rats [2]. In this study, we compared different rat APC populations for their ability to stimulate T-cell responses to bCII.

MATERIALS AND METHODS

Animals

Male rats of the inbred WA/KIR/KCL strain (bred at King's College London) were given food and water ad libitum and kept on sawdust in plastic-bottomed cages in groups of no more than six. Rats were used for experiment at 8–14 weeks of age at 150–250 g bodyweight.

Antigens and immunizations

Soluble native bCII was prepared according to the method of Lee-Own and Anderson [16] and stored lyophilized at − 70°C. bCII was found to contain 7·9 EU/mg protein by the BioWhittaker Kinetic QCL endotoxin assay (BioWhittaker, Wokingham UK). bCII was initially dissolved in 0·1 m acetic acid at a concentration of 2·5 mg/ml. This was used for immunizations, or dialysed against RPMI 1640 medium (Gibco, Paisley, UK) for use in cell culture. Concanavalin A (Con A), ovalbumin (OVA) grade V and incomplete Freund's adjuvant (IFA) were obtained from Sigma (Poole, UK). Synthetic bCII(184–198) peptide [GPEGAQGPRGEPGTP], which includes the dominant T-cell epitope of bCII in WA/KIR/KCL rats, was prepared by Fmoc solid phase synthesis as previously described [17].

Rats were immunized with collagen in a manner which would normally induce arthritis [17]. bCII (1 mg) dissolved in 0·4 ml 0·1 m acetic acid and emulsified in an equal volume of IFA was injected intradermally in the suprascapular region. Non-arthritogenic immunizations were also performed with 0·2 ml of emulsion consisting of equal volumes of IFA and PBS containing OVA (5 mg/ml).

Preparation of cells

Rats were sacrificed 12–14 days after immunization, at the onset of arthritis, and single cell suspensions were prepared from the brachial and axillary lymph nodes (LN). LN cells were washed twice with PBS and resuspended in culture medium. The culture medium used for all experiments consisted of RPMI 1640 supplemented with 2% penicillin/streptomycin (Gibco) 2% l-glutamine (Gibco) and 2% normal rat serum.

T cells were enriched from LN cell suspensions by nylon wool filtration (90–95% TCRαβ; n = 5) [18]. In one experiment, LN cells were depleted of APC as follows. Macrophages and DC were removed by Sephadex G10 filtration [18], then B cells were depleted using an anti-Ig immunocolumn (Cellect rat T, Biotex Laboratories, Edmonton, Canada).

Macrophages were isolated by peritoneal lavage using cold PBS containing 12 mm lignocaine (Sigma). Cells were washed in PBS and resuspended in RPMI culture medium before being used in proliferation assays. Medium was supplemented in selected wells with N-monomethyl-l-arginine (MMA) (Calbiochem, La Jolla, CA). B cells were purified from LN cells by positive selection using BioMag immunomagnetic beads coated with goat anti-rat Ig (PerSeptive Diagnostics, Cambridge, MA). DC were enriched using a modification of the method of Brennan and Puklavec [19] as follows. Single cell suspensions were prepared from spleen or thymus and low density cells (LDC) were isolated by density gradient centrifugation with Nycoprep 1·068 (Gibco). LDC were pulsed with antigen overnight and the non-adherent cells were recovered and used as APC. During the antigen pulse, medium was supplemented in selected wells with inhibitors of antigen processing: chloroquine (Sigma), leupeptin (Sigma) or BB3103 (British Biotech, Oxford, UK).

Macrophages or DC were also generated in vitro from bone-marrow (BM) precursors [20]. BM cells were recovered from rat femurs and resuspended in RPMI culture medium containing 12·5 ng/ml recombinant mouse GM-CSF (R & D Systems, Abingdon, UK). To generate DC, these cultures were further supplemented with 2·5 ng/ml recombinant rat IL-4 (Biosource International, Camarillo, CA). The further incubation of GM-CSF/IL-4-generated DC with TNFα, which is effective in maturing human DC [21], was not used because it induces rat DC to produce nitric oxide that inhibits T-cell proliferation (G. G. MacPherson, University of Oxford, personal communication). Therefore, in these experiments the method of Talmor et al. was followed, which omits this final maturation stage [20].

Flow cytometric analysis was performed using the mouse anti-rat monoclonal antibodies R7·3 (TCRαβ), MRC OX6 (MHC Class II), MRC OX42 (CD11b/c) and MRC OX12 (Ig-kappa chain). Cells were indirectly labelled with a rabbit anti-mouse Ig-FITC conjugate (Dako Ltd, Cambridge, UK) and analysed on a FACSCAN flow cytometer using Lysis II software (Becton-Dickinson, San José, USA). The purity of APC populations was assessed by calculating the proportion of cells with the following characteristics: large, granular CD11b/c⁺ cells were considered to be macrophages, non-adherent, CD11b/c⁺, MHC Class II⁺ cells were considered to be DC, and Ig-kappa chain⁺ cells were considered to be B cells.

Proliferation assays

LN cells or T cells (each at 2 × 10⁵/well) were cultured with the indicated amounts of antigen or antigen-pulsed APC, respectively. Cells were incubated at 37°C in a humid incubator, and proliferation was measured after 96 h (LN-cell assays) or 72 h (T-cell assays) by adding 0·5 µCi ³H-thymidine (25 Ci/mmol) (Amersham Pharmacia Biotech, St Albans, UK) per well for the last 6 h of culture. Results are shown as the mean counts per minute (CPM) of triplicate cultures. The stimulation index is calculated as the mean CPM cells with antigen divided by the mean CPM cells in the absence of antigen.

RESULTS

Macrophages fail to stimulate T-cell proliferative responses to bCII

Rat peritoneal macrophages, co-cultured with bCII-immune LN cells, inhibited proliferation to both bCII and Con A (Fig. 1a). Similar inhibition was observed using bronchoalveolar macrophages, pulmonary interstitial macrophages and splenic macrophages (data not shown). Attempts to block nitric oxide production by peritoneal macrophages using MMA failed to completely reverse the inhibitory effect on LN cell proliferation to bCII (Fig. 1a), suggesting that other mediators or cellular interactions are also involved.

Fig. 1.

Macrophages inhibit mitogen and antigen responses and fail to present bCII or bCII(184–198) peptide. (a) LN cells from rats immunized with bCII were cultured with (◊) bCII (10 µg/ml) or (□) Con A (2·5 µg/ml) in the presence of increasing numbers of peritoneal macrophages (80–88% CD11b/c⁺, 5–11% MHC Class II⁺ n = 3). MMA (0·25 mm) was added to selected cultures; (○) bCII + MMA. (b) Macrophages were generated from BM cultures in medium supplemented in selected wells with () bCII or () bCII(184–198) peptide (each at 25 µg/ml); (▪), no Ag/mitogen; (□), Con A. BM-derived macrophages (97–99% CD11b/c⁺, 19–23% MHC Class II⁺ n = 2) were cultured with bCII-immune T cells. Proliferation was measured by ³H-thymidine incorporation and results are expressed as the mean CPM ± s.e.m. of triplicate cultures.

In the absence of inflammatory cytokines, it has been reported that monocytes differentiate into tissue macrophages of an inhibitory phenotype, whereas monocytes exposed to GM-CSF during differentiation are not inhibitory and can present antigen to T cells [22,23]. To mimic this process, macrophages were generated in vitro by culturing BM cells with recombinant GM-CSF. However, when these BM-derived macrophages were used as APC they also failed to stimulate T-cell responses to bCII or bCII(184–198) peptide (Fig. 1b).

These observations confirm reports that rat macrophages are inefficient APC and can suppress T-cell proliferative responses rather than stimulating them [15]. This contrasts with studies showing that bCII and bCII peptides can be presented by mouse macrophages [12,13].

Antigen presentation of bCII by DC

Splenic and thymic LDC populations, enriched for DC, were both assessed for their ability to present antigen. bCII-immune T cells proliferated when they were cultured with LDC pulsed with bCII or bCII(184–198) peptide (Fig. 2), demonstrating that APC within these populations were able to present antigen. Splenic LDC were pulsed with antigen in the presence of chloroquine (a lysosomotropic agent), BB3103 (a matrix metalloproteinase inhibitor) or leupeptin (a serine and thiol protease inhibitor). Although proliferation was suppressed by chloroquine, suggesting that elevating the lysosomal pH could inhibit bCII presentation, addition of protease inhibitors failed to suppress the T-cell response (Fig. 3). Since collagens are particularly resistant to enzymatic degradation [24,25] and T-cell responses occurred in the presence of selected protease inhibitors, the mechanism of processing of bCII by APC for presentation to T cells remains to be clarified.

Fig. 2.

Splenic and thymic LDC present bCII and bCII(184–198) peptide. (a) Splenic LDC (30–45% CD11b/c⁺, 47–61% MHC Class II⁺ n = 3) or (b) thymic LDC (24–30% CD11b/c⁺, 28–44% MHC Class II⁺ n = 3) were isolated, pulsed with bCII or bCII(184–198) peptide and cultured with bCII-immune T cells. Proliferation was measured after 72 h of culture and results are shown as the mean CPM ± s.e.m. of triplicate cultures. (□), No Ag; (▪), bCII; (), bCII(184–198).

Fig. 3.

Protease inhibitors failed to inhibit presentation of bCII. Splenic LDC were pulsed with bCII (100 µg/ml) in the presence of BB3103 (10 µg/ml), leupeptin (100 µm) or chloroquine (20 µm). Cells were washed and co-cultured with bCII-immune T cells (2 × 10⁵ T cells + 2 × 10⁴ APC per well). Splenic LDC cultured without bCII (no antigen) are shown for comparison. Proliferation was measured after 72 h of culture and results are shown as the mean CPM ± s.e.m. of triplicate cultures.

LDC isolated from solid lymphoid organs are only partially enriched for DC and contain other types of APC. An alternative method of generating rat DC from BM cultures using recombinant GM-CSF and IL-4 was used to increase yield and purity [20]. Whilst OVA-immune T cells proliferated when cultured with OVA-pulsed BM-derived DC, bCII-pulsed BM-derived DC were unable to stimulate T-cell proliferation (Fig. 4), revealing differences in the ability of these cells to present these two antigens.

Fig. 4.

BM-derived DC present OVA but not bCII. DC were generated from BM precursors by culture in medium containing GM-CSF and IL-4. Selected wells were supplemented with OVA or bCII (each at 25 µg/ml) during the culture period. BM-derived DC (94–95% CD11b/c⁺, 67–82% MHC Class II⁺ n = 2) were cultured with T cells isolated from the LN of (a) OVA-immunized rats or (b) bCII-immunized rats. Proliferation was determined after 72 h and results are shown as the mean CPM ± s.e.m. of triplicate cultures. (□), No Ag; (), OVA (a) or bCII (b).

Antigen presentation of bCII by B cells

LN cells, prepared from rats at the onset of arthritis, proliferate to bCII and bCII(184–198) peptide [17]. Addition of monoclonal antibodies during culture in one experiment revealed that LN cell proliferation to bCII could be inhibited with R7.3 (TCRαβ: 89·5% inhibition), W3/25 (CD4: 56% inhibition) and OX6 (MHC II RT1B: 99% inhibition), with only partial inhibition using OX17 (MHC II RT1D: 17·5% inhibition). These findings suggest that proliferation was dependent upon APC:T-cell interaction and that the majority of bCII epitopes were presented by the RT1B MHC Class II molecule.

bCII-immune LN cells, filtered through Sephadex G10 columns to deplete macrophages and DC (< 1% CD11b/c⁺ cells), showed only marginally-reduced proliferative responses compared with the original LN cell population (Fig. 5). In contrast, depletion of B cells using an anti-Ig immunocolumn (< 5% Ig κ-chain⁺ cells, > 95% TCRαβ⁺ cells) abrogated the response of this cell population to both bCII and bCII(184–198) peptide (Fig. 5), implying that B cells were required to stimulate proliferation. This was confirmed by showing that B cells, isolated from the LN of either bCII-immunized or non-immunized rats, could stimulate proliferation of bCII-immune T cells to either bCII or bCII(184–198) peptide (Fig. 6), indicating that rat B cells are able to present these antigens to primed T cells. Although both types of B cell were able to function as APC, B cells from immunized rats were more potent in stimulating responses to bCII than those from non-immune rats (Stimulation index at 10⁴ B cells per well; immune B cells = 3·3, non-immune B cells = 2·26).

Fig. 5.

Depletion of B cells, but not macrophages or dendritic cells, abrogated the LN cell proliferative response to bCII. LN cells from bCII-immune rats were cultured with (▪) bCII (10 µg/ml) or () bCII(184–198) peptide (25 µg/ml) before separation (LNC), after Sephadex G10 filtration (Seph G10) and after anti-immunoglobulin filtration (Anti-Ig). (□), No Ag. Proliferative responses were determined after 96 h of culture and results are expressed as the mean CPM ± s.e.m. of triplicate cultures.

Fig. 6.

Antigen presentation by B cells. B cells (87–92% Igκ⁺, 90–97% MHC Class II⁺ n = 3) were isolated from LN of (a) bCII-immunized or (b) non-immunized rats by immuno-magnetic sorting. Cells were cultured in the presence of (▪) bCII or () bCII(184–198) peptide (each at 100 µg/ml) or without antigen (□) for 18 h and washed before culture with bCII-immune T cells. Proliferation was measured after 72 h of culture and results are shown as the mean ± s.e.m. of triplicate cultures. Proliferation of B cells in the absence of T cells was negligible (< 200 CPM).

DISCUSSION

Macrophages have been proposed as the primary cell type involved in processing and presenting bCII to T cells [12,13]. However, rat peritoneal or BM-derived macrophages failed to stimulate T-cell proliferation to either bCII or the bCII(184–198) peptide. Thus, in contrast to those from mice, rat macrophages are inefficient APC for presentation of bCII.

Rather than stimulating T-cell responses, peritoneal macrophages inhibited proliferation to bCII and Con A, confirming other studies showing that rat macrophages suppress immune responses [15,26]. Nitric oxide is believed to be the major lymphostatic mediator [22], but prostaglandins and regulatory cytokines have also been implicated [27]. LN cell proliferation to bCII was inhibited using cell-free supernatant fluids from cultured peritoneal macrophages (data not shown), implying that this effect was mediated by soluble factors. However, inhibition was only partially reversed by blocking nitric oxide synthesis, suggesting that other soluble mediators are also involved.

DC are potent APC, and presentation of foreign antigens by these cells is likely to be essential for initiating immune responses to pathogens. In contrast, presentation of self antigens by DC usually results in tolerance, but it also has the potential to stimulate autoreactivity [28]. Mouse splenic LDC, rich in DC, have been reported to be unable to present bCII to T cells and only present bCII-peptides [13], inferring that DC are able to present collagen fragments but not native antigen. In contrast, the present study showed that rat splenic or thymic LDC were able to stimulate T-cell proliferative responses to both bCII protein and bCII(184–198) peptide, suggesting that rat DC could be involved in presenting bCII to T cells. However, splenic and thymic LDC are only partially enriched for DC and may contain contaminating B cells or macrophages. Although DC can be further purified from LDC populations by magnetic or flow cytometric sorting for OX62-labelled cells [19], in the present study, a different strategy was pursued to maximize the yield and purity by generating rat DC in vitro from BM precursors.

It is noteworthy that commercial preparations of collagen have been shown to contain high levels of endotoxin which can affect the function of BM-derived DC [29]. Additionally, LPS has been shown to inhibit differentiation of GM-CSF/IL-4-derived DC [30], suggesting that endotoxin contamination could influence antigen presentation of bCII by BM-derived DC. However, the bCII used in the current study was shown to have an endotoxin level of 0·2 EU/ml culture medium, a concentration considered to be below that required for a pyrogenic effect on DC [30].

We demonstrated that rat BM-derived DC were able to stimulate T-cell proliferation to OVA, but not to bCII. The discrepancy in presentation of these two antigens could result from differences in their physico-chemical properties, since bCII, unlike OVA, is a fibrous macromolecule with a repetitive structure which is resistant to enzymatic degradation [24]. It has been suggested that the triple-helical structure of collagen interferes with antigen processing [25], although it has also been reported that similar high molecular weight proteins do not require processing to be presented [31]. Indeed, a requirement for enzymatic degradation of bCII by APC was not substantiated in this study, since splenic LDC were able to present bCII in the presence of protease inhibitors, a similar finding to that reported in mice [13].

Presentation of antigen in the form of heat-denatured bCII, cyanogen-bromide cleavage fragments or synthetic bCII peptides by BM-derived DC, could indicate whether prior degradation of bCII is required for these cells to function as APC. However, since the BM-derived DC used in this study are generated through in vitro manipulation, this would not necessarily predict the mechanism of antigen processing of native bCII by DC which develop in vivo.

The pathogenesis of CIA appears to involve a synergistic mechanism, involving both B-cell and T-cell responses to bCII [32]. Interaction between B and T cells was evident in the present study, since LN cells depleted of B cells failed to proliferate to bCII, and reconstitution of a T-cell population with B cells restored proliferation. Flow cytometric analysis of bCII-immune LN cells has shown that culture with bCII leads to an increased expression of T-cell activation markers, including CD25 and CD134, and a proliferative response dominated by CD4⁺ T cells [33]. The results presented here support a role for B cells as APC in this response.

Since bCII-reactive B cells express surface immunoglobulin specific for this antigen, efficient uptake of bCII by receptor-mediated endocytosis would be expected. Indeed, B cells from bCII-immunized rats were more potent than non-immune B cells in stimulating T-cell responses to bCII. It has been reported that only antigen-specific B cells are capable of acting as APC due to poor uptake of antigen in the absence of specific cell-surface antibody [34]. However, since lymphocytes express β1-integrins [35] and CD26 [36], which bind extracellular matrix proteins including collagen, B cells that do not recognize bCII as an antigen could still efficiently internalize this protein using receptors other than their surface immunoglobulin. The role of such accessory molecules in T-cell responses to bCII was not examined in this study. However, it is clear that integrins and other adhesion molecules are important in the pathogenesis of CIA [37]. In particular, it is noteworthy that interaction between collagens and adhesion molecules on the surface of T cells and APC, can act as a co-stimulatory signal for activation and cytokine production [38].

Following immunization with bCII and drainage of antigen to the LN, we propose that antigen presentation could occur by both bCII-reactive and non-specific B cells. In particular, B cells which recognize bCII in its native triple-helical state could play a critical role in CIA by producing pathogenic autoantibodies, and by acting as APC, which present bCII to pathogenic CD4⁺ T cells.