We analysed six HLA-B27⁺ and three HLA-B27^- patients with ReA after infection with Chlamydia trachomatis (Table 1). We diagnosed ReA if patients had a prior urogenital infection, which was confirmed by the detection of Chlamydia trachomatis in the morning urine by polymerase chain reaction. An additional criterion was the detection of Chlamydia-specific antibodies 6 at the beginning of the disease or highest synovial T cell proliferation against Chlamydia trachomatis 22 in proliferation assays with whole Chlamydia antigen. The results were compared with tetramer staining in six HLA-B27⁺ healthy blood donors. We also examined synovial T cells from three HLA-B27⁺ patients with ReA after gastroenteritis and having highest synovial proliferation against enterobacteria. We also tested the synovial fluid of three patients with rheumatoid arthritis. In addition we used HLA-B27⁺ and HLA-A2⁺ blood donors with previous EBV infection for experiments comparing HLA-B27 and HLA-A2 tetramers.

The ethical committee of the Benjamin Franklin Medical Centre gave ethical approval for this study.

The quantification of HLA-B27 binding affinity was conducted with two different programs that analyse HLA-peptide binding motifs, one called SYFPEITHI described by Rammensee and colleagues 23 and the other called BioInformatics and Molecular Analysis Section (BIMAS; http://bimas.dcrt.nih.gov/molbio/hla_bind/).

Nonamer peptides were synthesized by standard 9-fluorenyl-methyloxy-carbonyl solid-phase synthesis methods on a Syro-Synthesizer (MultiSyn Tech, Witten, Germany), purified by high-performance liquid chromatography (Shimadzu LC-10; Shimadzu Scientific Instruments, Duisburg, Germany) and identified by mass spectroscopy (LCQ, ion trap; Thermoquest, Eberbach, Germany). The purity of the peptides was more than 95%. Peptides were dissolved in dimethyl sulphoxide. For T cell stimulation and fluorescence-activated cell sorting (FACS) analysis of intracellular cytokine staining, the peptides were further diluted with serum-free medium at a concentration of 5 mg/ml and frozen at -80°C.

HLA-B27 tetramers were generated as described previously 19, with some modifications. The expression vector pLM1-HLA-B27 was modified by tagging with the BirA recognition sequence as described previously and by mutating the cysteine residue at position 67 to serine. After being refolded, the recombinant protein was concentrated and centrifuged at 13,000 rpm (16,060g; Haereus Biofuge Pico; Kendro Laboratories, Langenselbold, Germany) followed by biotinylation and gel filtration with a Superose 12 column (Pharmacia) on an Äkta Basic system (Pharmacia). Correct folding and biotinylation were analysed by gel filtration (Äkta Basic, Pharmacia) and gel electrophoresis (Bio-Rad). Tetramers were generated by adding phycoerythrin (PE)-labelled streptavidin (Molecular Probes) at a ratio of 1.5:1. We generated HLA-B27 tetramers with the EBV EBNA peptide (residues 258–266) 24. For the detection of Chlamydia-peptide-specific CD8⁺ T cells we used the previously described immunodominant peptides 8, 68, 80, 131, 133, 138, 144, 145, 146, 194, 195 and 196 10 (Table 2). Peptides 144 and 194 caused heavy aggregation during refolding procedure and were excluded from tetramer staining; peptide 146 was excluded because of high background staining in more than 50% of the patients.

HLA-A2 monomers with the EBV peptide 21 were generated with an HLA-A2 heavy chain (gift from Dr KH Lee, Berlin, Germany) with the same protocol.

For FACS analysis, frozen mononuclear cells (MNCs) from synovial fluid or peripheral blood were incubated with tetramer and PerCP-labelled anti-human CD8 antibody (BD Pharmingen, San Diego, USA) in parallel for 30 min at room temperature (20°C) followed by washing twice with phosphate-buffered saline (PBS)/2% bovine serum albumin (BSA) and incubation with Cy5-labelled anti-human CD3 antibody for 30 min at room temperature. Cells were washed twice in PBS/2% BSA and resuspended in Annexin V buffer (Molecular Probes) and 2.5 μl of Alexa 488-labelled Annexin V (Molecular Probes) was added. CD8⁺ and tetramer-positive T cells were analysed after gates were set on CD3⁺ and Annexin V-negative cells. Depending on the availability of additional synovial lymphocytes we repeated the staining experiments, which was true for the synovial fluid of patient no. 6.

Peripheral MNCs were incubated for 30 min with Cy5-labelled anti-CD8 antibody (BD) and 5 μg/ml PE-labelled HLA-B27/EBV EBNA (258–266) tetramer at room temperature. Cells were washed twice and incubated for 15 min at 4°C with anti-PE-labelled MACS beads (Miltenyi) at a ratio of 20 μl of beads to 80 μl of cell suspension. Labelled cells were loaded on an LS MACS column (Miltenyi) and eluted after the column had been washed three times with washing buffer including PBS, EDTA and BSA. MACS-sorted tetramer-positive and CD8⁺ T cells were further separated by FACS sorting. Sorted cells (1000) were incubated with 500,000 autologous antigen-presenting cells in the presence of 20 U/ml interleukin (IL)-2, 10 ng/ml IL-7 and 10 ng/ml IL-15 added every 3–4 days.

The refolding rate of recombinant HLA-B27 monomer was analysed by gel filtration and by determining the relative amount of soluble HLA-B27 monomer eluted at 13.7 ml in comparison with precipitated protein eluted earlier in a Superose 12 column (Pharmacia). An Akta basic system (Pharmacia) was used. The elution profile was analysed by using Unicorn (version 4) software (Pharmacia). Refolding was defined as ++ when more than 75% of proteins loaded on the gel filtration column after refolding, biotinylation and sharp centrifugation was soluble HLA-B27 monomer molecule; + for more than 50% soluble HLA-B27 monomer, (+) for more than 10% soluble HLA-B27 monomer, and - for less than 10% soluble HLA-B27 monomer (Table 2).

Intracellular cytokine staining was used after antigen-specific T cell stimulation. Synovial MNCs and peripheral MNCs were stimulated for 6 hours in 1 ml of culture medium with anti-CD28 antibody (1 μg/ml) plus single peptides (10 μg/ml) or without antigenic peptide as a negative control. Brefeldin A was added after 2 hours to stop the stimulation, and cells were harvested after a further 4 hours and then stained with 5 μg/ml anti-CD69-PE antibody (BD Pharmingen) and 1 μg/ml anti-CD8-PerCP (BD Pharmingen). Cells were then fixed in 2% formalin and resuspended in saponin buffer, followed by incubation with 1 μg/ml Cy5-conjugated anti-human interferon-γ antibody (IFN-γ; BD). Gated CD8⁺ T cells that were positive for early activation marker CD69 and for intracellular IFN-γ were counted as antigen-specific. Analysis was performed with a BD Biosciences FACScan flow cytometer with CellQuest software.

CD14⁺ cells from peripheral blood were incubated for 1 hour with anti-CD14-conjugated magnetic beads (Miltenyi) and sorted by MACS. The purity of separated cells was confirmed by FACS analysis. Cells (500,000) were cultured for 7 days in 24-well plates at 37°C at 5% CO₂ in 1 ml of RPMI culture medium supplemented with 10% fetal calf serum, 2 mM L-glutamine and 50 ng/ml granulocyte/macrophage colony-stimulating factor and 10 ng/ml IL-4 to induce transformation to dendritic cells (DCs). Cells were washed and harvested and incubated for 24 hours with infectious elementary bodies of Chlamydia trachomatis at a ratio of 1:50. DCs were analysed by FACS with the use of anti-CD80, anti-CD86, anti-HLA-DR, anti-CD14 (BD Pharmingen) and anti-Chlamydia trachomatis lipopolysaccharide antibodies (Dako) before and after infection with viable Chlamydia trachomatis.

We stimulated CD8⁺ T cells from peripheral blood with Chlamydia trachomatis-infected peripheral-blood-derived DCs at a ratio of 50:1 in RPMI culture medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin. Recombinant IL-7 (10 ng/ml) and IL-15 (10 ng/ml) were added on both days 2 and 7. T cells were analysed by FACS on day 14.

To determine optimal conditions for the refolding procedure of soluble HLA-B27 monomers we used HLA-A2 tetramers and HLA-B27 tetramers with well-described HLA-B27 and HLA-A2 restricted viral epitopes from EBV. The binding scores of the two immunodominant EBV peptides to the HLA-A2 21 (score 29) and HLA-B27 24 (score 28) receptor were almost identical in the SYFPEITHI program. By generating the HLA-B27 tetramer (Fig. 1a,1c; lanes 1 and 2) and the HLA-A2 tetramer (Fig. 1b,1c; lanes 3 and 4) the percentages of protein aggregates eluted between 7 and 13 ml in gel filtration, and the refolded monomer eluted at 13.7 ml in gel filtration, were also similar. The large peak at 16 ml most probably contained reagents from the biotinylation reaction because we did not detect any proteins with a molecular mass of more than 5 kDa by SDS-PAGE. This peak was excluded when the relative amount of soluble HLA-B27 monomers was estimated. In FACS analysis, tetramer-positive antigen-specific T cells could be detected with both tetramers (Fig. 1d), although HLA-A2 tetramers stained with greater intensity (log 0.8 more) than the HLA-B27. On the basis of these results we generated HLA-B27/Chlamydia peptide tetramers.

To determine whether tetramer-binding CD8⁺ T cells could be sorted and further cultured we stained peripheral MNCs with HLA-B27/EBV EBNA (258–266) tetramer. Before MACS, 0.22% of peripheral MNCs were CD8⁺ and tetramer-positive. After MACS, EBV EBNA (258–266)-specific T cells were enriched to 41.4%. MACS-sorted cells were further separated by FACS sorting and cultured for 4 weeks in the presence of IL-2. The purity of antigen-specific CD8⁺ was increased to 95.0%, as shown by tetramer staining (Fig. 2a). In parallel we performed intracellular cytokine staining after peptide-specific stimulation of peripheral MNCs and of the tetramer-sorted T cell line after 4 weeks of culture. In comparison with HLA-B27 tetramer staining, only 68.3% of these antigen-specific T cells were detected by intracellular cytokine staining of IFN-γ (Fig. 2b).

The generation of HLA-B27 tetramers with Chlamydia-derived peptides strongly indicated that the yield of refolded and soluble HLA-B27/Chlamydia peptide monomers depended on the binding affinity of the peptide for HLA-B27. Gel-filtration analysis showed that Chlamydia peptide 133 (Table 2; binding score 25 in 23) (Fig. 3a) induced significantly more protein aggregation, seen by protein elution between 7 and 13 ml, than Chlamydia peptide 8 (Table 2; binding score 26 in 23 but 10,000 in BIMAS) (Fig. 3b). In SDS-PAGE analysis the large quantity of aggregated proteins is also shown by numerous bands of higher molecular mass (Fig. 3a). After the addition of streptavidin, the major band with biotinylated HLA-B27 molecule could be captured to become a tetramer (Fig. 3a,3b; SDS-PAGE). This phenomenon of protein aggregation depending on the affinity between peptide and HLA-B27 could also be observed with the other Chlamydia-derived peptides. The refolding rate of all HLA-B27 tetramers used in this manuscript are summarized in Table 2.

On the basis of our recently identified Chlamydia-derived immunodominant peptides in Ct-ReA 10 we successfully synthesized nine HLA-B27 Chlamydia peptide tetramers and used them to stain MNCs from the synovial fluid of nine patients (six HLA-B27⁺, three HLA-B27^-) with Ct-ReA. Four of the six HLA-B27⁺ patients had a specific T cell binding to at least one HLA-B27/Chlamydia peptide tetramer.

The results of tetramer staining in all patients are summarized in Table 3; HLA-B27/Chlamydia peptide 195 tetramer bound to the synovial T cells of three (patient nos 2, 3 and 5) of these four patients. Two patients (nos 5 and 6) showed a T cell response to Chlamydia peptide 133 as detected by tetramer staining, and one (patient no. 3) had a T cell response to Chlamydia peptide 68.

The results of three patients are illustrated in Figs 4 and 5. Figure 4a shows that T cells specific for Chlamydia peptides 195 and 68 were detected with HLA-B27/Chlamydia peptide tetramers in patient no. 3: 0.09% of CD8⁺ T cells were positive for peptide 195 and 0.06% were positive for peptide 68. All other HLA-B27 tetramers with Chlamydia-derived peptides such as peptide 138 were negative (data not shown). In patient no. 2 we detected 0.06% HLA-B27/Chlamydia peptide 195 tetramer-positive T cells (Fig. 4b). All other HLA-B27 tetramers such as HLA-B27/Chlamydia peptide 138 were negative (data not shown). We did not analyse the cytokine secretion profile of CD8⁺ T cells in response to chlamydial peptides in these two patients. The example of patient no. 6 is shown in Fig. 5a, with 0.02% of HLA-B27/Chlamydia peptide 133 tetramer binding to CD8⁺ T cells but no binding to any of the other HLA-B27 Chlamydia peptide tetramers. In patient no. 6 we were able to repeat this experiment and obtained a similar result, with 0.02% of tetramer binding to CD8⁺ T cells.

To confirm the specificity of the T cell response to peptide 133, two further experiments were performed in this patient. First, synovial T cells were expanded by Chlamydia peptide 133-specific T cell stimulation for 1 week ex vivo, which revealed 0.22% tetramer-positive CD8⁺ T cells (Fig. 5a). Second, when FACS analysis of IFN-γ secretion after peptide-specific stimulation was done in the same patient, only peptide 133 induced this cytokine secretion (Fig. 5b), again confirming the specificity of this response.

We also analysed CD8⁺ T cells from peripheral blood of patient nos 2, 3, 5 and 6, who were responders when synovial fluid was tested for HLA-B27/Chlamydia peptide binding, but we could not detect any specific binding (data not shown). The HLA-B27^- patients with Ct-ReA and all six HLA-B27⁺ healthy controls had no HLA-B27-restricted, Chlamydia-peptide-specific T cell response (data not shown). Tetramer staining of synovial T cells from three HLA-B27⁺ patients with enterobacteria-triggered ReA and from three patients with rheumatoid arthritis revealed no specific staining of CD8⁺ T cells with 0–0.01% tetramer binding to CD8⁺ T cells.

Because the frequency of Chlamydia-specific CD8⁺ T cells in these patients is low in synovial fluid and absent in peripheral blood with both methods (tetramer staining and intracellular cytokine staining), we investigated whether enrichment of these cells could be achieved by short-term stimulation with autologous Chlamydia-infected DCs. By doing this we intended to obtain a higher frequency of tetramer-positive CD8⁺ T cells, to underline the specificity of tetramer staining.

We generated DCs from CD14⁺ monocytes from peripheral blood of patient no. 5; they were separated by MACS first. After 7 days of cultivation in vitro, the cells turned into DCs, as indicated by the loss of CD14 receptors and the upregulation of HLA-DR, CD80 and CD86 receptors (data not shown). We infected these DCs with viable Chlamydia trachomatis and confirmed infection by using an anti-Chlamydia trachomatis lipopolysaccharide antibody and by quantification of Chlamydia-positive cells by FACS analysis (data not shown). We revealed at least 41.3% Chlamydia-infected DCs.

Peripheral MNCs from the same patient were stimulated with these Chlamydia-infected DCs for 2 weeks in the presence of IL-7 and IL-15. Subsequently, FACS analysis for intracellular cytokine staining for IFN-γ performed after restimulation of this cell line with Chlamydia-infected DCs revealed 0.11% IFN-γ-secreting CD8⁺ T cells, and stimulation with different peptide pools including the nine relevant peptides revealed between 0.07% and 0.21% antigen-specific IFN-γ-secreting CD8⁺ T cells (data not shown). When the cell line was analysed with HLA-B27/Chlamydia peptide tetramers we found a similar quantity of expanded CD8⁺ T cells with significant tetramer staining of CD8⁺ T cells specific for Chlamydia peptides 8 (0.09%), 68 (0.10%), 133 (0.17%), 138 (0.08%), 195 (0.23%) and 196 (0.06%) (Fig. 6) and a weaker response to the other Chlamydia-derived peptides. HLA-B27 tetramer staining with peptides 133 and 195 showed some unusual bright staining, which was also frequently observed with the HLA-B27/EBV EBNA (258–266) tetramer and might have been caused by aggregated tetramers. Staining of untreated peripheral MNCs from the same patient did not reveal any tetramer binding (data not shown); staining with an HLA-B27/EBV peptide tetramer was performed as a positive control. We repeated this procedure in an HLA-B27^- patient with Ct-ReA (Fig. 7) and in an HLA-B27⁺ healthy blood donor (data not shown). In neither case could we observe staining with any of the HLA-B27/Chlamydia peptide tetramers even after stimulation with Chlamydia-infected DCs (patient no. 9; Fig. 7).