In both RA and CIA joint cartilage is affected by massive inflammation that drives destruction of the tissue. Immune responses against CII, the main synovial component, take part in this process. During arthritis, CII is degraded to small peptides that are presented to T cells. Previously, we identified the immunodominant epitope CII259-273  in which the K264 residue plays a key role for CII-specific T cell recognition in mouse models expressing either Aq or HLA-DR4 MHC II molecules [23, 24]. Furthermore, we have demonstrated that K264 is glycosylated in normal human and rat joint cartilage while in arthritic cartilage the CII259-273 epitope is present in both non-modified and glycosylated form (K264 and GalOK264, respectively) . This finding raised an important question for the role of K264 glycosylation/deglycosylation state in arthritis development. Therefore, the present study aimed to investigate CII-specific T cell responses at different time points of CIA in order to clarify whether there is a shift in T cell recognition of CII259-273 and its posttranslationally modified forms during disease development. We have chosen to focus on Th1 response in this study as CII-specific T cell response in CIA is dominated by Th1 cells , which produce IL-2 and IFN-γ and could thus drive the production of complement-fixing CII-specific IgG2a, an important component in the pathogenesis of CIA . Interestingly, no shift in T cell recognition was observed during the phases investigated. In general, B10.DR4.Ncf1*/* mice displayed dominant T cell reactivity to the non-modified CII259-273 and also to the hydroxylated form of the epitope. The dominant B10.Q T cell response was directed against the glycosylated CII259-273 epitope. An exception to this rule was seen only at the initial stage of CIA in B10.Q mice when a relatively equal T cell reactivity to non-modified CII259-273 and to the glycosylated peptide was detected. One possible reason for this discrepancy could be the CII used for immunization. This protein contains both non-modified and glycosylated CII259-273 epitopes since it is extracted from Swarm chondrosarcoma, which do not uniformly glycosylate the CII molecule . Thus, it is to be expected that the mice would initially react against both glycosylated and non-modified epitope since following immunization both epitopes are picked up from the CII molecule and presented by T cells. Such tendency, although not that strong, is also evident in B10.DR4.Ncf1*/* mice where, in addition to the strong K264 and HOK264 response, we detected certain level of response to the GalOK264 peptide. Moreover, previous studies on T cell responses at an earlier time point following immunization showed even more pronounced response against the glycosylated epitope in DR4 mice (Dzhambazov et al., unpublished data). Therefore, we assume that presentation and recognition of both non-modified and glycosylated CII259-273 following immunization occurs earlier in DR4-transgenic mice compared to B10.Q mice. That would explain also the very high magnitude of response to K264 and HOK264 epitopes detected in B10.DR4.Ncf1*/* mice already at day 15 postimmunization.
The lack of uniform glycosylation level of Swarm-derived rCII could also be the reason for stronger T cell response to non-modified CII259-273 epitope in B10.DR4.Ncf1*/* mice. To test this hypothesis, a group of DR4 mice with human CII was immunized with a protein extracted from normal human cartilage that is known to contain exclusively glycosylated CII259-273 . The results, however, could not clearly confirm our hypothesis. Again, we detected a dominant T cell response to K264 and HOK264 epitopes. The only time point when we identified a significant GalOK264 response was day 50 postimmunization. At this time point, we detected a higher reactivity to the glycosylated epitope also in mice immunized with Swarm chondrosarcoma-derived rCII. During this stage of disease, the stronger response to the GalOK264 peptide could be attributed to epitope spreading in T cell specificity due to the acute inflammation process and developed immunity to the endogenous CII. Similarly, certain response to the other forms of CII259-273 epitope, in addition to the immunodominant one, was detected also in B10.Q mice at later time points of CIA (day 35, day 50, day 65).
The T cell responses in RA patients carrying HLA-DRB*04 alleles show a diverse pattern of CII259-273 epitope recognition, that is, toward both non-modified and glycosylated CII259-273; only few patients exclusively responded to K264 but most of them to GalOK264 [5, 6]. A longitudinal analysis of CII-specific T cell responses showed a consistent reactivity but variable magnitude and specificity of the response over time . Concerning the magnitude of T cell reactivity our results are in line with the RA patients' data since we could not see a correlation between disease visible signs (disease severity) and the level of IFN-γ secreted in response to CII259-273 stimulation in DR4-expressing mice. Also, in concordance with the human patients' data our results demonstrate significantly higher CII responses in arthritic compared to healthy individuals. However, when looking at T cell response specificity the results from RA patients are in contrast with our findings in DR4-transgenic mice where we could demonstrate a very strong response to the K264 epitope. This discrepancy raises an important question - whether the DR4-transgenic mouse model could accurately resemble the human disease setting and whether there is a need for improving the humanized model. Indeed, the standard transgenic technology implies insertion of an unknown number of transgene copies into the mouse genome and also the site/sites of transgene insertion is/are uncertain, which in turn could affect the physiological expression of the transgene itself. Interestingly, the more physiologic Aq mouse model resembles the CII response in RA better than the more artificial but humanized DR4 mouse.
CII-specific immunity in CIA and RA include both T and B cell responses. Thus, we included to our studies evaluation of anti-CII antibody levels during CIA. The longitudinal analysis of total CII-specific titers in B10.DR4.Ncf1*/* showed a trend for correlation with disease severity but in B10.Q mice we detected a different pattern of antibody levels during CIA with day 65 (about the end of the CIA acute phase characterized by slightly decreased disease severity) showing the highest anti-CII IgG concentration. It is likely that the high antibody levels at day 65 is triggered by an endogenous response to cartilage CII and could precede and contribute to another peak of disease severity at a later time point. On the other hand, it has been shown that high anti-CII antibody titers do not exclusively lead to more severe disease . The total CII-specific levels are polyclonal and include clones that do not contribute to arthritis development . Thereby, an epitope-specific antibody response will be expected to show a better tendency for correlation with disease. Three dominant conserved CII epitopes - C1, U1, and the J1 epitope, can be recognized both in CIA and RA [14, 15]. Our experiments included evaluation of epitope-specific antibody responses in B10.Q and B10.DR4.Ncf1*/* mice and indicated a dominant response to the C1 epitope during the different phases of CIA. Anti-C1 antibody levels showed a positive relation to disease clinical score - the highest titers were detected during acute stages of CIA while at day 65 postimmunization C1-specific antibodies decreased in both mouse models. B10.DR4.Ncf1*/* mice showed very low level of recognition of J1 epitope but a sustained response to U1 epitope. Interestingly, anti-U1 antibody levels were relatively constant at all studied CIA time points. In fact, previous studies on U1-specific monoclonal antibody have shown that it binds and destabilizes cartilage independent of inflammation . It is likely that this mechanism of action is the reason for a lack of increase of U1-specific antibody titers at the acute stage of disease. However, additional experiments are needed in order to clarify whether this result could be due to non-optimal physiological setting in the transgenic model. This study was limited to three immunodominant CII epitopes. Despite the lack of epitope shift during the various disease phases investigated one cannot exclude involvement of additional CII epitopes during different stages of CIA, an ongoing study (Lindh et al., unpublished data).