Experimental immunity to the G1 domain of the proteoglycan versican induces spondylitis and sacroiliitis, of a kind seen in human spondylarthropathies

Objective. Experimental immunity to the G1 domain of the cartilage proteoglycan (PG) aggrecan (AG1) leads to the development of spondylitis as well as polyarthritis in BALB/c mice. The PG versican contains a structurally similar G1 domain (VG1). This study was conducted to determine whether immunity to VG1 would elicit similar pathology in these mice. Methods. Recombinant natively folded VG1 and AG1 were prepared. BALB/c mice received either a series of 5 injections of human VG1 or AG1, or no protein. Polyarthritis was determined clinically, and spondylitis and sacroiliitis histologically. Immunohistochemistry of rat tissues was used to study the localization of versican. Enzyme-linked immunosorbent assays were employed to study Immoral immunity,to the recombinant proteins as well as to overlapping synthetic peptides covering all these human G1 domains and mouse homologs. Affinity-purified antibodies to human AG1 and VG1 were isolated from sera of hyperimmunized mice. T lymphocyte proliferation assays were performed using recombinant human proteins. T cell lines reactive with specific immunodominant T cell epitopes in human AG1 and VG1 were isolated. Synthetic peptides encoding sequences in these human proteins and in corresponding mouse proteins were used in these analyses. Guanidinium chloride extracts of mouse spines were also used in Western blots to study antibody cross-reactivity. Results. Immunity to recombinant VG1 did not result in clinical polyarthritis. There was, however, clear evidence that VG1, like AG1, could induce spondylitis in the lumbar spine and sacroiliitis. Accumulation of mononuclear cells was observed in spinal ligaments adjacent to the intervertebral disc, in the intervertebral disc, and in the sacroiliac joints, the same sites where versican is localized. In contrast to AG1-immunized mice, in which T cells reactive with human AG1 cross-reacted with mouse AG1, there was no evidence in VG1-immunized mice that T cell immunity to human VG1 was cross-reactive with a mouse synthetic peptide that contained the sequence corresponding to the single immunodominant T cell sequence recognized in human VG1. Antibodies to specific sequences in human VG1 did, however, cross-react with human AG1 and with corresponding peptide sequences in mouse versican and aggrecan and with mouse proteins containing VG1 and AG1, present in mouse spine extracts. Similarly, antibodies to human AG1 cross-reacted with human VG1 and with extracted mouse VG1 and AG1 and synthetic peptides containing mouse sequences that corresponded to the reactive human epitopes in AG1 and VG1. Conclusion. These observations suggest that humoral immunity to human VG1 is involved in the induction of experimental spondylitis and sacroiliitis in BALB/c mice. This humoral immunity is cross-reactive with mouse versican and aggrecan but is not associated with polyarthritis, probably because of the lack of cross-reactive T cell immunity and the absence of detectable versican in articular cartilage limbs. Induction of polyarthritis by bovine or human aggrecan requires the involvement of immunity mediated by T lymphocytes that are cross-reactive to a mouse aggrecan epitope. Together these observations suggest that humoral immunity to versican as well as immunity to aggrecan may be of importance in the development of the spinal pathology characteristic of spondylarthropathies.