Autoantigenic HCgp39 Epitopes Are Presented by the HLA-DM-Dependent Presentation Pathway in Human B Cells¹

Autoimmune disease represents a failure of development of tolerance to self-proteins. Studies in a variety of systems have identified several mechanisms for the generation and maintenance of self-tolerance (reviewed in Ref. 1). These mechanisms can be divided into those that operate centrally, in the thymus or bone marrow, to purge the developing repertoire of potentially self-reactive lymphocytes, and those that operate peripherally to control self-reactive responses in the periphery. Ag presentation to T cells is a key event in several of these cases. For example, in the thymus, selection on low avidity MHC ligands may allow escape of T cells with relatively high affinity for self-peptide/MHC complexes (2). Alternatively, in the periphery, self-Ags may be immunogenic if they are novel (not presented in the thymus) or presented in an altered form via a nontraditional presentation pathway (reviewed in Ref. 3). These scenarios may be particularly relevant for autoimmune diseases that are associated with MHC class II alleles (4, 5, 6).

The traditional MHC class II Ag presentation pathway selects for stable MHC/peptide complexes. Class II molecules are synthesized in the endoplasmic reticulum, assembled onto invariant chain (Ii)⁴ and transported to endosomes, where Ii is degraded to terminal class II-bound Ii peptides (CLIP)⁴ (reviewed in Refs. 7 and 8). CLIP release from class II molecules is catalyzed by HLA-DM. HLA-DM also selects for stable MHC-peptide complexes, apparently by releasing DM-unstable peptides until an optimal ligand is found (9, 10, 11).

In contrast to epitopes selected in the traditional pathway, autoantigenic peptides in several disease models bind weakly to their cognate MHC (12). For instance, encephalitogenic peptide epitopes derived from myelin basic protein, the immunizing Ag in murine autoimmune encephalitis, bind poorly to class II molecules (12, 13, 14). Low affinity epitopes have also been described for adjuvant arthritis, experimental autoimmune uveoretinitis, and experimental myasthenia gravis (12). However, some high affinity autoantigenic epitopes have also been described for autoimmune encephalitis and diabetes (15, 16, 17, 18, 19).

The low to moderate affinity of some disease-associated autoantigenic epitopes suggests that these epitopes may have been generated via an alternate pathway. Peptide epitopes of the acetylcholine receptor may also be generated via nontraditional pathways because acetylcholine receptor-reactive T cell clones obtained from healthy and diseased subjects, respond only to peptide Ag, but not to intact Ag presented by PBMC (20).

Rheumatoid arthritis (RA) is a class II-associated autoimmune disease in which certain HLA-DRB1 alleles confer both predisposition to RA and an increased risk of severity of disease (21, 22, 23). The RA-associated DRB1 alleles, including DRB1*0401 contain a short consensus sequence, called the “shared epitope” (SE) that consists of residues DRB 67, 70, 71, and 74 including positively charged residues at pocket 4 (24, 25). The SE is not present in closely related non-RA-associated alleles, such as DR*0402, implicating this region in disease pathogenesis. The crystal structure of DR*0401 with a collagen peptide shows that the SE residues contribute to peptide-binding specificity and to TCR contact (26). The SE region of RA-associated alleles may influence interaction with CLIP and/or HLA-DM. Allelic variations in interactions of MHC class II molecules with Ii and with CLIP, and their varying degrees of dependence on HLA-DM, have been described (27, 28). These allelic differences could affect the peptides bound to the class II molecules and possibly facilitate a surface display of MHC molecules bearing lower affinity autoantigenic ligands.

To assess the influence of HLA-DM on the presentation of autoantigens, we selected human cartilage glycoprotein 39 (HCgp39). HCgp39 is a major secretory product of articular chondrocytes in culture (29, 30). It is detected in inflamed synovial joints, including joints in active RA, but rarely in normal articular joints (31, 32). T cells from HLA-DR4-positive RA patients and some HLA-DR4-matched healthy individuals show a proliferative response to HCgp39 peptides (33, 34). The HCgp39 peptide epitopes used in this study have IC₅₀ values ranging from 75 to 220 nM for soluble DR*0401 (34). These affinities are similar to the affinity of the DM-susceptible CLIP peptide for DR*0401, which has an IC₅₀ value of 141 nM (27). In this study we assess whether, like CLIP, the HCgp39 peptides are susceptible to DM editing.

To determine whether presentation of HCgp39 autoantigenic epitopes requires the presence of intracellular HLA-DM, we used human B lymphoblastoid cells with and without intracellular HLA-DM as model APCs. These B cell lines lacked HCgp39-specific B cell receptors, hence Ag uptake is presumed to be via endocytosis. We examined responses of T cell hybridomas generated from either DR*0401- or DR*0402-transgenic mice immunized with the HCgp39 protein. We used DR*0401-restricted T cell hybridomas that respond to the same HCgp39 peptides that induce a proliferative T cell response in RA patients (34).

B lymphoblastoid cell lines (B-LCL) were maintained in RPMI 1640 with 15% bovine calf serum and 2 mM l-glutamine. T hybridoma cells were cultured in hypoxanthine/aminopterin/thymidine media (RPMI 1640 supplemented with 10% FCS, 2 mM l-glutamine, 10 mM HEPES, 1 mM sodium pyruvate, 50 μM 2-ME, 100 μM hypoxanthine, 0.4 μM aminopterin, and 0.016 mM thymidine). Media and supplements were purchased from Life Technologies (Gaithersburg, MD).

The B-LCL cells used to generate HLA-DR4 transfectants were 8.1.6 and its HLA-DM-null derivative, 9.5.3. These cells are hemizygous in the DR/DQ region of the MHC and express HLA-DR3, DQ2, and DP4 (35). 9.5.3 cells do not express DMB (36). DR molecules from 9.5.3 are predominantly associated with CLIP peptides and have fewer DR3 non-CLIP-peptide complexes on the cell surface than HLA-DM-positive 8.1.6 cells (37, 38). Transfection of HLA-DMB into 9.5.3 cells restores the wild-type phenotype (36). Both 8.1.6 and 9.5.3 cells have been extensively characterized and are capable of presenting exogenously supplied peptides to specific T cells (39).

8.1.6 and 9.5.3 cells were electroporated with either DRA plus DRB1*0401 or DRA plus DRB1*0402 cDNAs on a pSVneo-plasmid containing the neomycin resistance gene (34, 40). Control 8.1.6 and 9.5.3 cells were electroporated with the DRA gene alone and will be referred to as DR4-null APC. Transfectants were selected by growth in 1 mg/ml G418. Single-cell clones were derived by limiting dilution.

Clones were characterized by flow cytometry using a panel of Abs including a DRB1*04-specific mAb, NFLD.D1 (41); a SE-specific mAb, NFLD.D2 (42); a DR dimer-specific mAb, L243 (43); a DRB1*04-specific mAb that recognizes an HLA-DM-dependent epitope NFLD.D11 (44); a DRB1*0401-specific mAb that recognizes an epitope on DM-null cells but not on DM-positive cells, NFLD.D13 (S. Drover, unpublished data); an irrelevant IgM Ab NS ‘M’ (TerraNova Biotechnology, Newfoundland, Canada); a CLIP-specific mAb, CerCLIP.1 (45); and a DRB1*0301-CLIP-specific mAb, 5-41 (E.D. Mellins, unpublished data). For FACS analysis, cells were incubated with primary Ab and washed twice with medium (RPMI 1640, 2 mM l-glutamine, 25 mM HEPES, 0.1% sodium azide, and 5% FCS, adjusted to pH 8). Bound Ab was detected using fluorescein-labeled goat anti-mouse IgG from Life Technologies or PE-labeled goat anti-Mouse IgM, μ-chain specific (Jackson ImmunoResearch, West Grove, PA). Cells were analyzed on a Becton Dickinson (Mountain View, CA) FACScan using CellQuest software. HLA-DM-positive and HLA-DM-null cells expressing comparable amounts of HLA-DR4 were selected on the basis of NFLD.D1 staining.

Long chain biotinylated hemagglutinin peptide 307–319 (PKYVKQNTLKLAT) was obtained from Dr. Anand Gautam (M & E Biotech, Denmark) and was verified by mass spectrometry. Surface peptide binding was measured as described (46).

Triple-transgenic mice, expressing HLA-DR4 (DRA *0101 and DRB1*0401 or DRB1*0402) and human CD4 on an I-Aβ^−/− background were generated as described (34). These mice were immunized with HCgp39 and T cell hybridomas were generated as described (34). Three immunodominant HCgp39 epitopes, which accounted for > 80% of the hybridomas generated, were identified from DR*0401-transgenic mice, and two distinct immunodominant epitopes accounting for 86% of the T cell hybridomas were identified from the DR*0402 mice (Ref. 34 ; see Table I). Peptides 22–37 and 298–313 were immunodominant in the context of DR*0402, but nonimmunogenic in DR*0401 mice, which were immunized with intact HCgp-39. Immunization of DR*0401 transgenic with these epitopes in free peptide form, resulted in the generation of T cell hybridomas that could respond to free peptide but not whole protein. Representative T hybridomas were selected from each category.

The capacity of T cell hybridomas to respond to Ag was evaluated using either live, irradiated, or glutaraldehyde-fixed APCs. Live cells were irradiated (12,000 rad) with a cesium source. For fixed APC, cells were washed twice in Dulbecco’s PBS (DPBS) set at 10⁶ cells/ml and fixed with 0.1% glutaraldehyde in DPBS for 15 s. The fixation process was stopped by adding an equal volume of 0.2 M l-lysine followed by two additional washes in DPBS. Adequate fixation was assessed by addition of 10 μl of 5 mg/ml MTT (Sigma, St. Louis, MO) to 100 μl of cells followed by incubation at 37°C for 4 h (47). Live cells convert the yellow tetrazolium salt into blue formazan crystals by the action of dehydrogenase enzymes while fixed cells do not. Irradiated and appropriately fixed cells were plated at 3–5 × 10⁵ cells/well, preincubated for 30–60 min, with either HCgp39 protein or specific synthetic peptide (Research Genetics, Huntsville, AL) over a dose range of 0.1–5 μM. HCgp39 protein was purified from Chinese hamster ovary cells expressing HCgp39 cDNA from human articular chondrocytes (Organon, Oss, The Netherlands). B-LCL cells were cocultured with T hybridomas in the presence of Ag for 36 h at 37°C. T cell stimulation was measured using an IL-2 immunoassay (48). Briefly, IL-2 from the supernatant was captured with anti-IL-2 Ab (PharMingen, San Diego, CA), washed, and incubated with a biotin-conjugated anti-IL-2 Ab (PharMingen) in a sandwich immunoassay. Biotinylated IL-2 Ab was detected using a streptavidin europium read-out. The Ab was quantitated against a standard curve of recombinant IL-2 and europium fluorescence was measured on an LKB Wallac (Gaithersburg, MD) fluorometric plate reader. Results are expressed as the mean IL-2 amount of duplicate samples. Duplicates were generally within 1–5% of each other.

To obtain cells that express HLA-DR4 alleles in the presence or absence of intracellular DM, we transfected HLA-DR4 cDNAs into B-LCL 8.1.6 (DR3, DQ2, DP4) and its HLA-DM-null derivative, 9.5.3. This approach ensured that, with the exception of HLA-DM, the cellular components of the Ag presentation machinery were the same in all the transfectants.

Bulk transfectant lines were generated by electroporation with plasmids encoding either DRA plus DRB*0401 or DRA plus DRB*0402, and single-cell clones were isolated by limiting dilution. The identities of the HLA-DR4-transfected cells were confirmed using Ab NFLD.D2, which is specific for the SE and recognizes DR*0401, but not DR*0402. NFLD.D2 recognizes an epitope that is poorly expressed in DM-null cells (42). Thus, 8.1.6 DR*0401 cells bound greater amounts of NFLD.D2 than 9.5.3 DR*0401 cells (Fig. 1). Levels of surface HLA-DR4 expression were determined by flow cytometry using the anti-DR4 Ab NFLD.D1 (Fig. 1). This Ab recognizes HLA-DR4 alleles in an HLA-DM-independent manner (41). We selected HLA-DM-positive (8.1.6) and HLA-DM-null (9.5.3) clonal transfectants with closely matched levels of DR*0401 (or DR*0402) surface expression to ensure that differences in MHC density would not contribute to differences in Ag presentation between these cells.

In the absence of HLA-DM, CLIP peptide release from class II molecules is based on the intrinsic dissociation rate of CLIP/MHC class II complexes (27). We were interested in assessing the level of DR4/CLIP complexes on HLA-DM-null 9.5.3 cells expressing DR*0401 or DR*0402. HLA-DR4 expression on these cells was confirmed by binding of the NFLD.D1 Ab (Fig. 1). To detect DR-CLIP complexes, we used two DR-CLIP Abs, CerCLIP.1 and 5-41; CerCLIP.1 Ab binds to the N terminus of the CLIP fragment that extends beyond the peptide-binding groove (45). Thus CerCLIP.1 can detect both DR3-CLIP and DR4-CLIP complexes on the HLA-DM-null cells. However, the 5-41 Ab binds in an allele-specific manner and detects DR3-CLIP complexes but not DR4-CLIP complexes (E.D. Mellins, unpublished data). We observed equivalent or enhanced binding of CerCLIP.1 Ab to DR*04-transfected 9.5.3 cells as compared with untransfected 9.5.3 cells (Fig. 2,A) indicating total numbers of DR-CLIP complexes on DR*04 cells were either equal to or greater than DR-CLIP complexes on untransfected cells. However, using the 5-41 Ab we detected a significant decrease in 5-41 binding to DR*04-transfected 9.5.3 cells over untransfected 9.5.3 cells (Fig. 2 A) showing a reduction in DR3-CLIP complexes in the DR*04-transfected 9.5.3 cells. One likely explanation for this reduction is that the introduced β-chains compete with DR3β-chains for binding to DRα, which is hemizygous in these cells and may be limiting. Together, these data indicate that a significant proportion of the DR-CLIP complexes on the surface of DR*04-transfected 9.5.3 cells are DR4-CLIP complexes.

To verify that the HLA-DM-null DR*0401 cells were capable of presenting HLA-DM-independent epitopes, we examined these cells for the expression of HLA-DM-independent Ab epitopes and HLA-DM-inhibited Ab epitopes.

The HLA-DM-dependent Ab NFLD.D11 recognizes DR*0401-restricted epitopes in the presence of HLA-DM but not in its absence. We observed strong binding of NFLD.D11 to 8.1.6 DR*0401 cells and no recognition of 9.5.3 DR*0401 cells (Fig. 2,B). Conversely, HLA-DM-independent Ab NFLD.D13, preferentially recognizes epitopes in the absence of HLA-DM but not in its presence. For NFLD.D13, we observed positive staining of 9.5.3 DR*0401 cells and no staining of 8.1.6 DR*0401 cells (Fig. 2 B). Thus, the DM-null cells express epitopes that are suppressed in the presence of HLA-DM, and HLA-DM-positive cells express DM-dependent epitopes that the DM-null cells do not.

To demonstrate that both the HLA-DM-positive and the HLA-DM-null DR*0401 APC were capable of stimulating the DR*0401-restricted T hybridomas, the APCs were incubated with synthetic peptides corresponding to the immunodominant epitopes, HCgp39 100–115, 262–277, and 322–337 (Fig. 3,A). Peptides bind DR molecules predominantly at the cell surface (49), independent of intracellular HLA-DM. Both the HLA-DM-positive and the HLA-DM-null APC stimulated the T cells when pulsed with the appropriate peptides (Fig. 3 A). No stimulation was observed in the absence of Ag or by DR4-null cells with Ag.

The HLA-DM-null DR*0401 cells were 6- to 8-fold more effective at peptide presentation than the HLA-DM-positive cells for the 262–277 and 322–337 epitopes, despite both cells having similar surface HLA-DR4 levels (Fig. 3,A). We hypothesized that greater peptide presentation reflected greater surface peptide exchange on the DM-null DR*0401 cells. To test this hypothesis, we assessed the surface binding of a DR4-restricted peptide to DM-positive DR*0401 8.1.6 cells, DM-null DR*0401 9.5.3 cells, and to untransfected 8.1.6 and 9.5.3 cells. We observed over 10-fold higher binding of a stable DR*0401 binding peptide (hemagglutinin peptide 307–319) to DM-null DR*0401 cells as compared with DM-positive DR*0401 cells and little binding to untransfected cells (Fig. 3 C).

For the 100–115 epitope, T cell stimulation was the about the same for both DM-null and DM-positive cells. It is likely that the 100–115 peptide binds better on DM-null cells than DM-positive cells, akin to other DR*0401 binding peptides. However, the 100–115-specific T hybridoma may require fewer complexes for maximal activation because it responds strongly at low peptide doses (Fig. 3 B, inset).

To test whether efficient generation of the HCgp39 epitopes requires HLA-DM, we conducted Ag presentation assays using intact HCgp39 protein as the Ag. As shown in Fig. 3 B, presentation of immunodominant epitopes of HCgp39 (100–115, 262–277, and 322–337) was significantly enhanced by intracellular HLA-DM. Under the conditions used, HLA-DM-null APC presented the DR*0401 epitope, 100–115, at much lower levels than DM-positive APC and failed to present epitopes 262–277 and 322–337 altogether. These findings do not reflect the unique properties of the particular APC clones used in these assays, because we obtained similar results using bulk DR*0401-transfected APC with and without intracellular HLA-DM (data not shown).

To examine whether cellular processing is required for presentation of the whole protein by the HLA-DM-positive APCs, we fixed the APC before the presentation assay. Fixation abolished protein presentation but not peptide presentation (Fig. 4) arguing that intracellular processing is required.

We observed that fixed cells presented peptides more efficiently than live DR*0401 cells (Fig. 4). Peptide loading occurs mainly on surface class II molecules in live 8.1.6 cells and though some peptide is endocytosed, it is rapidly degraded (49). Fixing the cells probably obliterates this phenomenon, hence equimolar amounts of peptide may be better presented by fixed cells.

Taken together, the data from the DR*0401-restricted T hybridomas indicated that the presentation of all three immunodominant HCgp39 epitopes was dependent or greatly augmented by expression of HLA-DM by the APC.

To assess whether HLA-DM dependence of the immunodominant epitopes was unique to the RA-associated DR*0401 allele, we tested the closely related HLA-DR4 allele, DRB1*0402, (not associated with RA) in a similar experiment using HCgp39-specific T cell hybridomas from DR*0402-transgenic mice. Both the HLA-DM-positive and HLA-DM-null DR*0402-expressing APC stimulated the T hybridomas when pulsed with the relevant synthetic peptides (Fig. 5,A). No T cell response was observed with the DR*0402 APC in absence of Ag or with the control DR4-null cells with Ag (Fig. 5 A). Unlike the results observed with DR*0401 APC, peptide presentation by HLA-DM-positive DR*0402-expressing cells was comparable to HLA-DM-null DR*0402-expressing cells. The lack of enhanced peptide presentation by live DM-null DR*0402 cells may reflect reduced surface peptide exchange at the surface of these cells as a result of the significantly higher affinity of DR*0402 for the CLIP peptide⁵.

In assays using intact HCgp39 protein as Ag, we observed efficient presentation of HCgp39 DR*0402 immunodominant epitopes (22–37 and 298–313) by the HLA-DM-positive APC (Fig. 5,B). The HLA-DM-null APC presented the 298–313 epitope at much lower efficiency than the HLA-DM-positive cells and presented 22–37 very poorly, if at all. Fixation of the APCs abolished protein presentation but not peptide presentation as expected (Fig. 4). These results were reproduced with bulk DR*0402-transfected APC with and without intracellular HLA-DM (data not shown). Thus, similar to DR*0401 cells, HLA-DM-null DR*0402 APC were defective in presentation of intact protein.

Protein immunization of DR*0401 mice failed to produce any T cell hybridomas specific for HCgp39 epitope 22–37, which is an immunodominant epitope in DR*0402 mice. However, immunization of DR*0401 mice with synthetic peptides with amino acid sequences corresponding to HCgp39 22–37 produced T cell hybridomas (45G9 and 44E10) that recognized the peptide epitope but not whole protein. These hybridomas responded to DM-expressing APC incubated with peptide but not to cells incubated with intact HCgp39 protein as the Ag source. We hypothesized that complexes of these epitopes with DR*0401 may be susceptible to editing by HLA-DM. If so, then HLA-DM-null cells might express these epitopes after incubation with intact HCgp39 protein.

To test whether DM-null cells could present intact protein to peptide-only hybridomas, we incubated DM-null and DM-positive APC with protein and assayed stimulation of T hybridomas 45G9 and 44E10 (Fig. 6). No response to intact protein Ag was observed from either DM-positive or DM-null DR*0401 cells. However, these hybridomas did respond to specific peptide Ag incubated with either DM-positive or DM-null DR*0401 cells (Fig. 6), suggesting that these DR-peptide complexes are not generated from intact protein in DR*0401 cells.

Autoimmunity may be associated with the presentation of previously “cryptic epitopes” of self-proteins to CD4⁺ T cells (50). We hypothesized that presentation of self-Ags by cells lacking functional HLA-DM might provide a mechanism for revealing new T cell epitopes to the immune system. This hypothesis is based on the observations that HLA-DM edits the repertoire of peptides displayed at the cell surface in favor of stable MHC/peptide complexes (9, 10, 51, 52). To test this hypothesis, we examined the role of DM in presentation to a panel of T cell hybridomas that recognize epitopes from a human cartilage autoantigen, HCgp39. These hybridomas were generated from transgenic mice, but are specific for determinants recognized by human T cells from DR4+ individuals. The epitopes recognized by the hybridomas were presented by HLA-DM-expressing APC and poorly, if at all by HLA-DM-null cells, after incubation with intact HCgp39. The requirement for HLA-DM expression is similar to that observed for presentation of various foreign Ags, such as tetanus toxoid (39). Thus, the lack of T cell tolerance to this autoantigen does not appear to be the consequence of its presentation by alternative pathways that generate previously cryptic determinants.

However, one caveat to our conclusion is that while the B-LCLs used in this study demonstrate a requirement for DM, they may not be representative of available APCs in vivo. For instance, B cells expressing Ag-specific receptors have been shown to be able to overcome the requirement for DM (53). Therefore, it remains a possibility that these HCgp39 epitopes maybe generated by DM-independent pathways in vivo.

Lack of exogenous protein presentation to T cells by DM-null APC has been associated with inefficient removal of CLIP from the class II molecules (36). However, peptide elution from surface DR*0401 complexes expressed in DM-null T2 cells reveals other self-peptides in addition to CLIP, indicating that some spontaneous dissociation of CLIP from DR*0401 occurs (54). Thus, we were surprised by the lack of presentation of the HCgp39 epitopes by HLA-DM-null DRB1*0401 cells incubated with whole protein. These peptides are probably generated in the DM-null cells because they are presented by the parental HLA-DM-positive 8.1.6 cells. We believe the lack of presentation may reflect a requirement of DR*0401 molecules for chaperoning by DM. A high proportion of MHC class II molecules may be irreversibly inactivated after spontaneous CLIP release in the late endosomal compartments (55, 56, 57). It is also possible that the HCgp39 epitopes are generated at low levels and do not compete well for presentation in the absence of HLA-DM editing of low affinity ligands.

The HCgp39 100–115 epitope was presented from intact protein by the HLA-DM-null DR*0401 cells, albeit at lower levels than by DM-positive cells. Because this epitope has roughly the same affinity for DR*0401 as the other two epitopes tested, the differences may be due to the sensitivity of the 100–115-specific T hybridoma, which may be activated by fewer complexes on the cell surface. Alternatively, this epitope may be generated in higher abundance, allowing it to compete well for binding sites made available after CLIP release. Another possibility is that this epitope is generated in the early endosomal compartments where the half-life of empty DR*0401 molecules may be longer because the environment is less acidic.

Peptide presentation for two of three HCgp39 epitopes by DM-null, DR*0401 cells was significantly more effective than peptide presentation by DM-positive DR*0401 cells. Our direct peptide binding results showed enhanced surface peptide exchange on DM-null compared with DM-positive DR*0401 cells using a labeled DR4-binding peptide that forms a long-lived complex. We believe this enhanced peptide binding and presentation by DR*0401 is attributable to the instability of the DR*0401-CLIP complex that reaches the cell surface in the absence of HLA-DM, thus resulting in generation of more empty, peptide-receptive class II molecules on the cell surface⁵. These peptide binding data are consistent with the hypothesis that some autoantigenic RA epitopes may be presented as a result of extracellular proteolysis and surface loading of class II molecules in the acidic synovial joint environment. However, this pathway need not be invoked to explain the activation of HCgp39 T cells in RA patients.

In this study we also examined the presentation pathway used by epitopes recognized by peptide-only HCgp39 T hybridomas. These hybridomas are similar to type-B T cells defined by Viner et al. (58) using the hen egg lysozyme Ag. The type-B T cells respond to APC incubated with synthetic peptide or even peptide eluted from MHC-peptide complexes generated by intracellular processing of intact HEL protein but not to APC incubated with intact protein (58). It is possible that these T cells recognize a specific peptide-MHC conformation generated by surface binding of peptide and this conformation is different from that generated by intracellular processing (58, 59). Intracellular processing in late endosomal compartments may be influenced by several factors including lower pH and presence of accessory molecule HLA-DM. Using peptide-only HCgp39-specific T hybridomas, we tested whether the lack of presentation of these specific epitopes from intact protein was a result of these complexes being edited out by HLA-DM. Our results show that absence of intracellular HLA-DM did not restore presentation of these epitopes from HCgp39 protein.

The HCgp39 epitopes were originally generated and presented in DR*04-transgenic mice that are H-2 M positive. The requirement for intracellular HLA-DM in human B cells for presentation of these epitopes implies that both HLA-DM and H-2 M interact with the HLA-DR4 molecules and favor the same MHC-peptide complexes for surface presentation. H-2 M and HLA-DM molecules have been recently crystallized (60, 61). The crystal structures indicate that though overall structures are conserved between the two molecules, there are several regions that are considerably different between H-2 M and HLA-DM (61). Our results suggest that despite these differences, HLA-DM and H-2 M may be functionally very similar.

Our data suggest that the traditional HLA-DM-dependent presentation pathway is involved in the presentation of autoantigenic HCgp39 epitopes. HCgp39 is a secreted protein and may be endocytosed for presentation by professional APCs present in the joint. However, in some autoimmune diseases nonprofessional APC have been postulated to be involved such as the cytokine-activated thyrocytes in thyroid disease (62). Recent work has shown that IFN-γ-treated thyrocytes express HLA-DMβ, class II molecules, Ii and CIITA (62). Thus in some circumstances, the requirement for HLA-DM may be fulfilled by nonprofessional APCs that have been activated to express the machinery of the traditional Ag presentation pathway.

We thank Dr. Robert Busch for helpful discussions and critical reading of the manuscript. We also thank Kevin Visconti for technical assistance in growing the T cell hybridomas.