Molecular Mimicry of an HLA-B27-derived Ligand of Arthritis-linked Subtypes with Chlamydial Proteins*

HLA-B27 is strongly associated with spondyloarthropathies, including ankylosing spondylitis and reactive arthritis. The latter disease is triggered by various Gram-negative bacteria. A dodecamer derived from the intracytoplasmic tail of HLA-B27 was a natural ligand of three disease-associated subtypes (B*2702, B*2704, and B*2705) but not of two (B*2706 and B*2709), weakly or not associated to spondyloarthropathy. This peptide was strikingly homologous to protein sequences from arthritogenic bacteria, particularly to a region of the DNA primase fromChlamydia trachomatis. A synthetic peptide with this bacterial sequence bound in vitro disease-associated subtypes equally as the natural B27-derived ligand. The chlamydial peptide was generated by the 20 S proteasome from a synthetic 28-mer with the sequence of the corresponding region of the bacterial DNA primase. Molecular modeling suggested that the B27-derived and chlamydial peptides adopt very similar conformations in complex with B*2705. The results demonstrate that an HLA-B27-derived peptide mimicking arthritogenic bacterial sequences is a natural ligand of disease-associated HLA-B27 subtypes and suggest that the homologous chlamydial peptide might be presented by HLA-B27 onChlamydia-infected cells.

HLA-B27 is strongly associated to ankylosing spondylitis (AS), 1 reactive arthritis (ReA), and other spondyloarthropathies (1,2). Although this association is among the strongest of any HLA antigen to a human disease, the pathogenetic mechanism remains unknown. The main function of HLA class I molecules is to present peptide antigens to cytotoxic T lymphocytes (CTL). Thus, without excluding alternative mechanisms (3)(4)(5), it has been proposed that the antigen presenting properties of HLA-B27 may be crucial in the pathogenesis of spon-dyloarthropathies. An external antigen showing molecular mimicry with a self-peptide constitutively presented by HLA-B27 would eventually break tolerance and induce autoimmunity, leading to chronic inflammation (6). This "arthritogenic peptide" hypothesis is supported by much indirect evidence, including the presence of both bacteria-specific and autoreactive CTL in patients with AS and ReA (7,8), the influence of the B27-bound peptide repertoire on development of arthritis in transgenic rats (9), and the differential association of natural HLA-B27 allotypes to AS. Whereas B*2705, B*2702, B*2704, and B*2707 are strongly associated to this disease (10), B*2706 and B*2709 are weakly or not associated to AS (11)(12)(13)(14)(15). These subtypes differ from disease-associated ones in only one (B*2705/B*2709) or two amino acid changes (B*2704/B*2706) located in the peptide-binding site and known to influence peptide specificity and T-cell recognition (16). Thus, differential subtype association to AS suggests that peptide presentation by HLA-B27 may be critical for determining susceptibility to this and related diseases.
Gram-negative bacteria, including species of Chlamydia, Salmonella, Yersinia, Shigella, and Campylobacter are known pathogenetic agents of ReA in humans (17). In some cases, HLA-B27-restricted bacteria-specific epitopes triggering CTL responses in these patients have been identified (18). A bacterial component is also critical in the development of HLA-B27associated arthritis in transgenic rodents (19 -21).
Following the observation that polymorphic regions of HLA-B27 had more homology to bacterial proteins than other HLA class I allotypes (22), and that the amino acid sequence of residues 168 -176 in HLA-B27 had homology to protein sequences from Gram-negative bacteria (23), it was proposed that presentation by HLA-B27 of peptides derived from its own molecule might lead to autoimmunity following bacterial infection through molecular mimicry between bacterial proteins and HLA-B27. Subsequent studies confirmed that an HLA-B27derived peptide, closely related to the predicted one, B27-(169 -179), was a natural ligand of HLA-B27 (24,25). However, this peptide was abundant in the endogenous peptide pools from both disease-associated and non-associated subtypes, which questioned its pathogenetic relevance. More recently, an additional peptide from this region, B27-(169 -181) was identified as a natural HLA-B27 ligand. The subtype distribution of this peptide correlated better, albeit incompletely, with subtype association to AS (26).
Here we report that HLA-B27 constitutively binds in vivo a peptide derived from the cytoplasmic region of its own molecule that has significant homology with proteins from arthritogenic bacteria, specially Chlamydia. The HLA-B27-derived peptide is a natural ligand of three AS-associated subtypes analyzed, but * This work was supported by Grants SAF99/0055 from the Plan Nacional de IϩD, PM99-0098 from the Ministry of Science and Technology and 31-57307.99 from the Swiss National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
not of the two subtypes not associated to this disease. The homologous chlamydial peptide bound in vitro three AS-associated subtypes as efficiently as the natural B27-derived ligand, and was directly produced from a synthetic precursor by the 20 S proteasome.
Isolation of B27-bound Peptides-Isolation of HLA-B27-bound peptides was done as previously described (37). Briefly, about 1 ϫ 10 10 HLA-B27 transfectant cells were lysed at 4°C in 20 mM Tris/HCl buffer, 150 mM NaCl, and 1% Nonidet P-40 (pH 7.5) in the presence of a mixture of protease inhibitors. After centrifugation, cell lysates were subjected to affinity chromatography using the W6/32 mAb. HLA-B27bound peptides were eluted with 0.1% aqueous trifluoroacetic acid at room temperature, filtered through Centricon 3 (Amicon, Beverly, MA), and concentrated to 100 l for HPLC fractionation. This was done in a Waters Alliance system (Waters, Milford, MA), using a Vydac C18 (0.21 ϫ 25 cm) 5-m particle size column (Vydac, Hesperia, CA), at a flow rate of 100 l/min, as follows: isocratic conditions with buffer A (0.08% trifluoroacetic acid in water) for 15 min, followed by a linear gradient of 0 -44% buffer B (80% acetonitrile and 0.075% trifluoroacetic acid in water) for 90 min and a linear gradient of 44 -100% buffer B for another 35 min. Peptide fractionation was simultaneously monitored at 210 and 280 nm. Fractions of 50 l were collected and stored at Ϫ20°C.
Mass Spectrometry Analysis and Sequencing-The peptide composition of HPLC fractions was analyzed by MALDI-TOF MS using a calibrated Kompact Probe instrument (Kratos-Schimadzu) operating in the positive linear mode, as previously described (38). Dried fractions were resuspended in 5 l of methanol/water (1:1) containing 0.1% formic acid, and a 0.5-l aliquot of the sample was deposited onto the stainless steel MALDI probe and allowed to dry at room temperature. Then 0.5 l of matrix solution (saturated ␣-cyano-4-hydroxycinnamic acid in 33% aqueous acetonitrile and 0.1% trifluoroacetic acid) were added and again allowed to dry at room temperature.
Peptide sequencing was carried out by quadrupole ion trap nanoelectrospray MS/MS in an LCQ instrument (Finnigan ThermoQuest, San Jose, CA), exactly as detailed elsewhere (39,40). In some cases, PSD-MALDI-TOF MS sequencing was carried out as previously described (26).
Search for Homologous Peptide Sequences-A first search was made with the B27-(309 -320) sequence with the prokaryotic non-redundant protein data base, using the Smith-Waterman protein searcher application on Bioccelerator in the European Molecular Biology Laboratory server (Bioccelerator: eta.embl-heidelberg.de:8000/), using default settings. A second search was made with the same B27-derived sequence against the individual protein databases of Chlamydia trachomatis, Chlamydia pneumoniae, Campylobacter jejuni, and Salmonella typhimurium, using the BlastP program at the Entrez server of the National Center for Biotechnology Information (www.ncbi.nlm.nih.gov/cgi-bin/ Entrez/genom_table_cgi). The search was made using no filters, an expect value of 10,000, and default settings for other parameters.
Synthetic Peptides-These were obtained using standard Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry, and purified by HPLC. The correct molecular mass of purified peptides was established by MALDI-TOF MS, and their correct composition and quantification by amino acid analysis after hydrolysis in 6 M HCL using a 6300 Amino Acid Analyser (Beckman Coulter, Palo Alto, CA).
Epitope Stabilization Assay-The epitope stabilization assay used to measure peptide binding was performed as described (34), with minor modifications. Briefly, B27 RMA-S transfectants were incubated at 26°C for 22 h in RPMI 1640 medium supplemented with 10% heatinactivated fetal bovine serum. They were then washed three times in AIM-V serum-free medium (Invitrogen), incubated for 1 h at 26°C with various peptide concentrations in the same medium, transferred to 37°C, and collected for flow cytometry after 2 h for B*2702 and B*2704 transfectants or 4 h for B*2705 and B*2706 transfectants. HLA-B27 expression was measured using 50 l of hybridoma culture supernatant containing the mAb ME1. Binding of a natural HLA-B27 ligand, used as reference peptide, was expressed as C 50 , which is the molar concentration of the peptide at 50% of the maximum fluorescence obtained at the Purification of 20 S Proteasome and Digestion of Synthetic Substrates-The 20 S proteasome was purified from B*2705-C1R cell lysates by ion-exchange chromatography and centrifugation in a glycerol gradient as previously described (26). These preparations consisted of a mixture of 20 S proteasome and immunoproteasome, as determined by two-dimensional gel electrophoresis and Western blot analysis (data not shown). Peptide substrates were incubated at 37°C and 125 g/ml with purified 20 S proteasome at an enzyme/substrate ratio of 1:10 (w/w) in 20 mM Hepes buffer, pH 7.6. Digestion was stopped by adding 1/5 volume of 0.4% aqueous trifluoroacetic acid. Digestion mixtures were dried down to 100 l in a SpeedVac and fractionated by HPLC using the same conditions as for HLA-B27-bound peptides. Individual digestion products were identified on the basis of their molecular mass by MALDI-TOF MS and, when necessary for unambiguous assignment, by PSD-MALDI-TOF MS sequencing.
Molecular Modeling-The HLA-B27-derived dodecamer ligand B27-(309 -320) and a homologous peptide from C. trachomatis with the same length, DNA primase (211-222), were modeled in the binding groove of  HLA-B*2705, whose x-ray structure in complex with a model peptide (Protein Data Bank entry 1hsa) had previously been solved at a resolution of 2.1 Å (41). Peptides were built in the peptide-binding site as previously described (42). Briefly backbone coordinates of positions P 1 , P 2 , P 3 , P C-1 , and P C (Pc being the C-terminal peptide residue) as well as both charged termini were first kept constant and identical to that of the 1hsa crystal structure. Rotameric states of side chains at the abovedescribed positions were then assigned by searching an in-house threedimensional data base of 37 x-ray structures of class I MHC-bound peptides. Last, the central loop (P 4 to P C-2 ) bulging out of the binding groove was constructed using a knowledge-based loop search procedure using the LOOPSEARCH module of the SYBYL package (TRIPOS Inc, St. Louis, MO). In this procedure, a set of 1478 high-resolution x-ray structures were searched for a loop of similar length (7 amino acids) and presenting a similar distance between C␣ atoms of the residues delimiting the loop window. The loop showing the highest homology and the lowest root-mean-square deviations was further selected for insertion. After adding all hydrogen atoms and quick steepest descent AMBER5 minimization (43) of the whole complex, the loop was annealed for 50 ps at 1000 K and cooled down to 50 K for another 50 ps. The last simulated annealing conformer was finally relaxed again by 100 steps steepest descent minimization.

RESULTS
HLA-B*2705 Binds a Peptide Derived from its own Cytoplasmic Tail in Vivo-The B*2705-bound peptide pool was isolated from B*2705-C1R transfectant cells and fractionated by HPLC. When analyzed by MALDI-TOF MS, HPLC fraction N. 92 showed a main ion peak at mass/charge (m/z) 1239.63 (Fig. 1A). This peptide was fragmented by nanoelectrospray ion trap MS/MS. The corresponding spectrum (Fig. 1B) was consistent with the sequence of a dodecamer, RRKSSGGKGGSY, corresponding to HLA-B27 residues 309 -320. This assignment was confirmed by showing that the MS/MS spectrum of the synthetic dodecamer was essentially identical to the B27-derived natural ligand (Fig. 1C). The yield of this peptide was estimated on the basis of the intensity of the ion peak in the MALDI-TOF MS spectrum (Fig. 1A), after calibration with serial dilutions of the corresponding synthetic peptide. The value obtained was in the range of 28 -56 pmol/10 10 cell-equivalents or 1700 -3400 molecules/cell.
These results demonstrate that HLA-B*2705 binds a natural ligand derived from the cytoplasmic region of its own molecule. This sequence is conserved among HLA-B molecules, but is different in HLA-A and HLA-C. The sequence in HLA-C differs only by a Y320C change.
Binding of B27-(309 -320) by HLA-B27 Subtypes in Vivo Correlates with Disease Association-The B27-(309 -320) peptide was searched in the endogenous peptide pools from two other HLA-B27 subtypes associated to AS, B*2702 and B*2704, and from the two subtypes not or weakly associated to this disease, B*2706 and B*2709. These are structurally closest to B*2704 and B*2705, respectively. Comparative HPLC analysis of B*2705-and B*2709-bound peptide pools revealed that the prominent absorbance peak at fraction N. 92 lacked a counterpart in B*2709 ( Fig. 2A). Similarly, comparison of B*2704-and B*2706-bound peptide pools revealed a prominent peak at HPLC fraction N.92 from B*2704 and a much smaller peak from B*2706 at the same retention time (Fig. 2B). An absorbance peak was also observed at the corresponding fraction from B*2702 (Fig. 2C). MALDI-TOF MS analysis of these HPLC fractions showed ion peaks at m/z 1239.9, corresponding to B27-(309 -320) in B*2702 and B*2704 (Fig. 3). This was for-  Table II. mally confirmed by postsource decay (PSD)-MALDI-TOF sequencing of the peptide from B*2704 (not shown). The MALDI-TOF MS spectra of the corresponding HPLC fractions from B*2706 and B*2709 (Fig. 3), as well as the two previous and two following ones (not shown), failed to show this ion peak, although in B*2706 unrelated ion peaks were detected. These results indicate that for five HLA-B27 subtypes analyzed constitutive binding of the B27-(309 -320) peptide in vivo correlates with subtype association to AS. B27-(309 -320) Has Homology to Proteins from Arthritogenic Bacteria-Because of the pathogenetic role of various Gramnegative bacteria in ReA, B27-(309 -320) was compared with proteins from arthritogenic micro-organisms, to investigate the possibility of molecular mimicry. Initially, the B27-(309 -320) sequence was screened for homology against prokaryotic protein sequences. In a second step, the same sequence was screened against the protein data bases from C. trachomatis, C. pneumoniae, Chlamydia muridarum, C. jejuni, and S. typhimurium. In these analyses the best matches with sequences containing the canonic HLA-B27-binding motifs Arg2 and basic, aliphatic, or aromatic C-terminal residues, corresponded to 4 sequences from the DNA primase and GTP-binding proteins of C. muridarum, C. trachomatis, and C. pneumoniae, 6 sequences from unrelated proteins of S. typhimurium, and one sequence from C. jejuni (Table I). Thus, B27-(309 -320) has significant homology with protein sequences from several ar-thritogenic bacteria, most strikingly with a region of the DNA primase from Chlamydia.
The DNA Primase (211-222) Peptide from C. trachomatis Binds HLA-B27 in Vitro-B27-(309 -320) and homologous peptides from C. trachomatis and C. pneumoniae (Table I)  The peptides derived from the chlamydial GTP-binding protein bound the B27 subtypes tested with intermediate or low affinity, suggesting that these peptides are less likely to be presented by HLA-B27 in vivo.
The DNA Primase (211-222) and Related Peptides Are Generated by the 20 S Proteasome-Since DNA primase (211-222) from C. trachomatis was a possible natural ligand of HLA-B27, we tested whether this peptide could be generated by proteasomal cleavage in vitro. Thus, a synthetic 28-mer with the sequence of the C. trachomatis DNA primase residues 203-230 was digested by the 20 S proteasome. The digestion mixture was fractionated by HPLC, and individual digestion products were identified by MALDI-TOF MS and/or MS/MS sequencing (Fig. 5). Their yield was estimated on the basis of their absorbance at 210 nm, normalized to take into account peptidic length differences. When various peptides co-eluted, the percentage of each peptide in the absorbance peak was estimated on the basis of their respective ion peak signal intensities in the MALDI-TOF spectra.
Almost all the substrate (99%) was digested after 24 h. Cleavage after Phe-208 and Ser-210 occurred with the highest efficiency. Cleavage after Tyr-222 occurred less efficiently. These three cleavages generated the DNA primase (211-222) peptide and an N-terminally extended precursor spanning residues 209 -222. Each of these peptides accounted for 0.3% of the total digest or 0.8% of the internal fragments obtained by double cleavage of the synthetic substrate (Table III).
Additional cleavage after Gly-207, Ser-209, and, more prominently, Ile-223, resulted in the generation of a 13-mer, DNA primase (211-223) with the Arg2 and C-terminal Ile motifs, appropriate for binding to HLA-B27, as well as 3 N-terminal extensions of this peptide, spanning residues 208-, 209-, and The DNA primase (211-222) sequence is boxed. Thick, medium, and thin lines correspond to peptides recovered at Ͼ5, 1-5, and Ͻ1% yield of the total digest respectively. Only peptides recovered with Ͼ0.1% yield are indicated. Thick, medium, and thin arrows indicate cleavage sites that generated peptides with total yields Ͼ10, 1-10, and Ͻ1% of the total digest respectively. The DNA primase (211-222) peptide and an N-terminally extended precursor, (209 -222), are indicated by asterisks (*).   Table III). This result raises the possibility that, besides DNA primase (211-222), the C-terminally extended peptide (211-223) might be an additional ligand of HLA-B27 in Chlamydia-infected cells.

Molecular Modeling Suggests Similar Bound Conformations of B27-(309 -320) and DNA Primase (211-222)-Modeling both
dodecapeptides in complex with HLA-B*2705 was performed by homology to the existing x-ray structure of HLA-B27 using a previously reported procedure shown to explain and predict peptide binding properties (42,44). Both dodecamers were predicted to adopt a very similar bound conformation (Fig. 6), with positions 4, 6, 7, 8, and 11 bulging out of the peptide binding cleft. These positions would thus be likely to contact an ␣␤ T-cell receptor (TCR) in the ternary TCR-peptide-MHC complex. Interestingly, the apex of the peptide surface accessible to a TCR, located between positions 6 and 8 consisted of similar amino acid sequences (GGK versus GGR), thus suggesting a possible basis for cross-recognition of both dodecapeptides by CTL.

DISCUSSION
The arthritogenic peptide hypothesis (6) predicts that some self-peptide(s) presented by HLA-B27 may show molecular mimicry with bacterial antigens. However, identification of putative arthritogenic peptides has remained elusive. A variant of this hypothesis proposed that presentation by HLA-B27 of peptides derived from its own molecule with homology to bacterial proteins could be responsible for autoimmunity following bacterial infection (23). Although two such peptides, derived from the same region of the ␣2 domain of HLA-B27 have been reported as natural HLA-B27 ligands, their binding in vivo to HLA-B27 subtypes did not fully correlate with subtype association to AS (26).
In contrast to previous findings, our results now show the following novel aspects. First, HLA-B27 binds in vivo a peptide from its own molecule, B27-(309 -320), with strong homology to proteins from C. trachomatis and other arthritogenic bacteria. This homology includes identical or chemically similar residues at both anchor and non-anchor peptide positions. Although this peptide derived from a region that is not polymorphic among HLA-B molecules, its main binding motif, Arg2, is restricted to HLA-B27 and very few other HLA-B allotypes. Second, B27-(309 -320) is a natural ligand of 3 AS-associated subtypes, B*2705, B*2702, and B*2704, but not of the 2 subtypes, B*2706 and B*2709, weakly or not associated to this disease. Thus, among known HLA-B27 ligands derived from the B27 molecule itself, this peptide shows the best correlation with association to AS. Third, the homologous chlamydial peptide DNA primase (211-222) binds in vitro B*2702, B*2704, and B*2705 with the same efficiency as the B27-derived peptide, suggesting that, if generated in vivo, this bacterial peptide could be presented by these three subtypes on Chlamydia-infected cells. The B27 (309 -320) peptide bound B*2706 in vitro more weakly than a natural ligand of this subtype, but still with significant efficiency, despite its absence in the B*2706-bound peptide pool. Since the epitope stabilization assay used in this study is significantly influenced by the association rate, it is possible that the stability of this peptide bound to B*2706 may be insufficient for its presentation in vivo. Absence of B27-(309 -320) in the B*2706-bound peptide pool is unlikely to be explained by highly efficient folding of this subtype in the endoplasmic reticulum, which might impair dislocation and cytosolic processing of the molecule, because another peptide derived from the B*2706 heavy chain, B27-(169 -179), was a very prominent natural ligand of this subtype in the same cell line used in this study (25,26). The chlamydial DNA primase (211-222) peptide bound B*2706 in vitro with moderate efficiency and more weakly than B27-(309 -320), strongly suggesting that it cannot be presented by this subtype in vivo. Fourth, the chlamydial peptide, as well as an N-terminally extended precursor are directly produced in vitro by the 20 S proteasome, suggesting that DNA primase (211-222) might be generated in vivo following chlamydial infection. Direct generation of natural HLA-B27 and other MHC class I ligands by the 20 S proteasome in vitro has been repeatedly demonstrated (38,(45)(46)(47)(48), and in vitro peptide generation by the 20 S proteasome has been used to predict natural ligands (26,49). Involvement of the host proteasome in the generation of chlamydial CTL epitopes was recently suggested (50). Fifth, molecular modeling suggested a similar bound conformation of B27-(309 -320) and DNA primase (211-222) in complex with B*2705. An important feature of this model is that the central region of both peptides (residues 6 -8) that bulges out from the complex, and is therefore likely to provide major contacts with the TCR, has similar amino acid sequences: GGK or GGR. This provides a possible basis for antigenic mimicry and cross-recognition of both dodecapeptides by CTL. Due to the highly flexible nature of the central loop, alternative conformations cannot be excluded. Indeed, the longest peptide ever co-crystallised with a class I MHC protein, a 13-residue peptide in complex with RT1-A a (51) showed two drastically different conformations of the central peptide loop, revealing a significant flexibility of this region which, in both conformations, bulged out considerably from the peptide-binding groove. However, this flexibility would not necessarily abrogate CTL cross-reaction since both the TCR loops and the exposed peptide side chains may adopt distinct conformations to optimize intermolecular contacts in the TCR-peptide-MHC complex (52)(53)(54). In addition, as suggested by Speir et al. (51), the ability of a long peptide to display distinct conformations in complex with the same MHC class I molecule may generate multiple epitopes. This would activate a greater diversity of T cells reactive against a single peptide, increasing the chances for harmful cross-reactivity if tolerance is not developed for the whole set of conformational isomers. Thus, potential CTL cross-reaction between the chlamydial DNA primase and the B27-derived peptides might be favored by both the flexibility and chemical similarity of their central loops.
A critical issue, which was not addressed in our study, is the involvement of DNA primase (211-222) in HLA-B27-restricted Chlamydia-specific CTL responses, especially those taking place in ReA patients, and the possible cross-reactivity of this peptide and B27-(309 -320) at the T-cell level. C. trachomatis has evolved mechanisms to evade CTL recognition and killing of infected cells through degrading the transcription factor RFX5, which is needed for constitutive and IFN-␥-induced MHC class I expression, leading to a severe drop of these molecules from the cell surface 24 h or later after infection (55,56). Despite this, CTL responses can be elicited against C. trachomatis-infected cells (57,58), presumably due to bacterial antigen processing and presentation early after infection. In addition, CD8 ϩ T-cells recognizing HLA-B27-restricted chlamydial peptides have been identified from the synovial fluid of Chlamydia-induced ReA patients and from transgenic mice (59). These peptides, none of which derived from the DNA primase, were selected by searching the genome of C. trachomatis for nonamer peptide sequences with B27-binding motifs that could be generated by the proteasome on the basis of predictive algorithms. Thus, the role of these peptides in eliciting Chlamydia-specific CTL in vivo needs to be confirmed. That study did not rule out the existence of other B27-restricted chlamydial antigens, such as peptides longer than nonamers, in ReA patients.
As mentioned above, CTL responses are presumably elicited against Chlamydia antigens processed and expressed at early times. Chlamydiae initiate transcription and translation very soon after infection (60 -63). Although the temporal expression of DNA primase, an essential enzyme in DNA replication, has not been specifically analyzed, multiple genes involved in chlamydial DNA replication are activated by 4 h after infection (64), suggesting that increased synthesis of DNA primase occurs early. Therefore, peptides produced by degradation of this protein might be available soon for presentation on infected cells. It is also noteworthy that, contrary to other genes such as those involved in cytokinesis, bacterial genes involved in DNA replication are expressed during chlamydial persistence, in which the bacteria modify their cell cycle and survive into the cytoplasmic inclusions of the host cell (65). Persistent Chlamydiae may be the dominant forms in the joints of ReA patients, where they might be a source of chronic antigenic stimulation. Thus, the DNA primase protein might be continuously available to provide peptides, such as that homologous to B27-(309 -320), for presentation to T-cells in HLA-B27-positive Chlamydia-infected patients. If so, chronic stimulation of bacteria-specific CTL in these individuals could lead to autoimmunity directed against the B27-derived peptide. It is worth noting that a chlamydial peptide homologous to a heart muscle-specific protein induced autoimmune inflammatory heart disease in mice, linking autoimmunity and Chlamydia infection through molecular and antigenic mimicry (66).
Finally, it is interesting that B27-(309 -320) also showed significant homology with proteins from other arthritogenic bacteria, such as S. typhimurium. This might provide a basis for the capacity of bacteria other than Chlamydia to trigger HLA-B27-associated ReA.