Identification of an altered peptide ligand based on the endogenously presented, rheumatoid arthritis-associated, human cartilage glycoprotein-39(263–275) epitope: an MHC anchor variant peptide for immune modulation

In rheumatoid arthritis (RA), articular cartilage is destroyed by chronic inflammation that is characterised by activated lymphocytes and major histocompatibility complex (MHC) class II expressing cells in synovial tissue. The latter suggests the ongoing of an antigen-driven response [1]. Although the nature of the antigens responsible for RA pathogenesis is unknown, there is evidence that the disease-associated HLA-DR (B1*0401, 0404, 0405 and 0101) molecules are involved in disease pathogenesis [2]. The clinical observation that cartilage is likely to sustain the inflammatory response suggested a role for cartilage proteins as target autoantigens [3‐7].

The human cartilage glycoprotein-39 (HC gp-39) is a 42 kDa glycoprotein with structural homology to the bacterial chitinase protein family [8‐11]. Although its physiological function is unknown, its expression pattern suggests a role in tissue remodelling [9, 12]. The case for an involvement of HC gp-39 in RA has been well-documented. Serum and synovial fluid HC gp-39 levels are elevated in inflammatory diseases and correlate with disease activity in RA [13, 14]. HC gp-39 mRNA has been detected in synovial specimens and cartilage of RA patients but not in normal cartilage [9, 15]. Also, an increased presence of HC gp-39-expressing monocytic cells in RA synovial tissue is correlated with the degree of joint destruction [16]. Furthermore, HC gp-39-derived peptides with good relative affinity for the RA-associated HLA-DR molecules were identified as dominant T cell epitopes in HC gp-39-immunized, HLA-DR4 transgenic mice [17]. Similarly, these T cell epitopes were recognized by peripheral blood mononuclear cells from RA patients [7, 17, 18]. Interestingly, peripheral blood mononuclear cells from RA patients were found to respond to HC gp-39 in a pro-inflammatory mode whereas cells from healthy donors responded in an anti-inflammatory mode [19]. The demonstration that endogenous presentation of the 263–275 dominant epitope in the context of the RA-associated HLA-DR molecules by synovial dendritic cells is specific for RA pathology adds to the relevance of this epitope for the disease process [20, 21]. Recently, these findings were extended by the demonstration that antigen-presenting cells (APCs) in RA synovial fluid present endogenously processed HC gp-39-derived epitopes to T cells [22, 23]. These observations combined imply that immunotherapeutic strategies based on the endogenous sequence may be beneficial for RA. Given the role of HC gp-39, the tolerizing potential of HC gp-39 in experimental arthritis [7, 24] and the mere fact that a dominant epitope is endogenously expressed at the site of chronic inflammation, we sought to exploit the HC gp-39-derived epitope for antigen-based forms of therapy.

Cognate peptides may be modified at key T-cell receptor (TCR) recognition residues to create altered peptide ligands (APL). TCR interaction with APLs was found to alter the extent of antigen-specific T cell activation by skewing the cytokine profile of responding T cells or by anergy induction [25]. More importantly, in experimental models of autoimmune disease, APLs have been used successfully to deviate or dampen the pathogenic response [26‐29]. The translation of this preclinical work into successful clinical trials, however, has proven difficult. In clinical trials on multiple sclerosis, adverse hypersensitivity reactions were seen with classic APLs modified at key TCR contact sites [30, 31]. Current insights favour the notion that APLs based on MHC anchor substitutions may function as partial TCR agonists on the one hand and prevent unwanted immune reactivity on the other [32, 33]. This approach may thus provide an improved option for APL therapy.

The objectives of the present study were first to identify the MHC and TCR contact residues in the RA-associated epitope using a panel of well-characterized specific T cell hybridomas, second to derive a molecular model to guide APL design, and third, to identify an APL capable of modulating the pro-inflammatory response to the synovial auto-epitope.

We and others have provided a strong case for a role of HC gp-39 as a target of the immune response in RA. The finding that the immunodominant epitope of HC gp-39 is endogenously presented by APCs in RA synovial tissue suggests that immunotherapeutic strategies based on the endogenous sequence may be beneficial for RA. In this study we identified an APL (F265(Dip)) with the capacity to skew the antigen-specific, pro-inflammatory response in HLA-DR4trg mice. The APL identified was not modified at the TCR contact sites but contains a modification at the key MHC anchor position. The latter quality may add to improved safety of APL therapy.

Ideally, the properties of an APL for immunotherapy of RA would blend: affinity for MHC class II binding; capabilities to engage polyclonal TCRs with sufficient avidity to alter pro-inflammatory function; a high resemblance to the naturally processed epitope; expression of the epitope at the site of chronic inflammation for engagement of auto-responsive T cells; and, recently, the prevention of unwanted immune reactivity [30, 31].

In this study, we first characterized a panel of cognate T cell hybridomas. The four hybridomas were found to differ, as shown by different TCRVβ expression and/or fine epitope recognition. The heterogenous panel was used to identify the epitope residues interacting with MHC and/or the cognate TCRs using the single amino acid substitution approach. The core epitope contained one major residue (Phe265 at the p1 anchor position) important for binding to HLA-DR4. Given that we employed a peptide binding motif, imposing certain restrictions for the p1, p4, p6 and p9 positions, for the identification of this T cell epitope within HC gp-39 [7, 41], some more impact of the single site substitutions at p4, p6 and p9 was anticipated. The observation that a dual site substitution (at p1 and p6) gave a similar low binding affinity as the p1 single substitution clearly underlines the importance of the p1 phenylalanine in this sequence as predicted based on the crystal structures [38, 42]. Furthermore, four different amino acid positions within the core region were found to strongly interact with the TCRs since substitution did not affect HLA-DR4 binding but abrogated recognition in all four hybridomas. Notably, the interaction of position 268 with the TCR (A268D) was further substantiated using A268V6 and A268N6 as additional APLs (data not shown). The hybridoma data were confirmed when analysing the same peptides for recognition by polyclonal, cognate T cells from HLA-DR4trg mice, thereby showing that one APL may engage multiple clonalities.

Our analysis was extended to modifications at the MHC anchor residue. This strategy was inspired by the notion that unwanted immune reactivity as seen with APLs in clinical trials is thought to occur more vividly with classic APLs modified at the TCR contact sites, which may be 'more visible' to the immune system and may thus engage and activate novel clonalities [30, 31]. We hypothesised that the p1 pocket, buried in the class II molecule, would allow for accommodation of modified side chains and, thus, selected non-natural amino acids with different side chain modifications not affecting the peptide backbone. Notably, these APLs displayed comparable affinities for binding HLA-DRB1*0401 as the WT peptide. Surprisingly, however, partial TCR agonists were identified using the T hybridoma panel. When assayed using polyclonal T cells from WT peptide-immunized HLA-DR4trg mice, the partial agonistic activity of F265(3Tic) was not confirmed. Instead, this APL showed an increased cytokine production. This heteroclitic response is explained by the notion that a conformational change in the peptide backbone is introduced by substitution with the more rigid bicyclic amino acid 3Tic which may give rise to additional clonalities. In contrast, the partial agonistic activity of the F265(Dip) APL was confirmed using WT peptide-sensitized lymph node cells from HLA-DR4trg mice. A lesser recall of IFNγ production, representative of a Th1 type of response, was observed.

In trials of APL in multiple sclerosis, increased reactivity to APLs and the native myelin basic protein were noted [31]. Thus, we investigated the immune potential of the F265(Dip) APL in the context of HLA-DR4. The results show that this APL does not give rise to increased immune responses to APL or native peptide/protein. Rather, the partial agonistic activity was confirmed by reduced cytokine production while maintaining antigen-specific proliferation. More importantly, neither the WT peptide nor the HC gp-39 protein was capable of recalling pro-inflammatory cytokine production (IFNγ, Mipα and RANTES). The data suggest that the F265(Dip) APL is redirecting the immune response to HC gp-39. Future experiments should address the capacity of this APL to modulate the HC gp-39-specific, HLA-DR4-restricted response in the context of experimental arthritis.

How to explain the immune modulating potential of this APL at the level of TCR recognition, triggering and beyond? The current paradigm is that the recognition of slightly altered ligands by T cells can lead to partial activation by interfering with the stability, clustering or duration of the TCR-mediated signal [43, 44]. For the APLs identified here, it remains to be established whether effects on APL-MHC stability, density or duration of TCR activation can be reconciled with an inhibition of specific cytokine production in an otherwise productive response.

Although a link between RA and HC gp-39 has been established, the therapeutic efficacy of HC gp-39-based forms of therapy still await clinical proof [45, 46].

We have identified an APL resembling the endogenous epitope, which binds the RA-associated HLA-DRB1*0401 and is capable of dampening the polyclonal, pro-inflammatory cytokine response. Moreover, the apparent lack of priming novel reactivities suggests that this APL blends all the properties desired for evaluation in future clinical trials in RA.