Heat shock proteins, HLA-DR and rheumatoid arthritis

In March 1996, Nature Medicine published an article by Auger et al.¹, accompanied by a News and Views story², making the provocative claim that heat shock protein binding to HLA-DR molecules of certain individuals may profoundly influence susceptibility to rheumatoid arthritis (RA). If true, this would provide a molecular explanation of why Caucasian RA patients are usually HLA-DR4-positive (*0401 or 0404 alleles) or HLA-DR10 positive. Our data do not support this proposal.

The common feature of the disease-associated alleles is that they share a five-amino-acid motif, called the 'shared epitope', at position ß70–74 of the third hypervariable region: (Q/R)(K/R)RAA. Protective alleles contain a more acidic motif, such as DERAA (ref. 3). Auger, et al. suggested that the stretch of residues constituting the shared epitope specifically recruits a molecular chaperone, HSP73, which targets relevant DRb chains to lysozomes, although it is conceptually difficult to imagine where within the cell the cytosolic/nuclear HSP73 chaperone would interact with luminal DRß1 sequence. This work was based upon an earlier observation showing that the bacterial DnaK protein (bacterial homologue of HSP73) could bind peptide affinity columns comprising the shared epitope sequence. These findings were particularly relevant given that autoimmune dysfunction may follow episodes of bacterial infection. Auger and colleagues have also shown, more recently, that bacterial DnaK and DnaJ proteins interact through the QKRAA peptide expressed in the amino terminal J domain of DnaJ⁴. However, the NMR solution structure of the HSP40 J-domain (the mammalian homologue of DnaJ) provides compelling evidence that it is a highly conserved tripeptide, HPD (residues 31–33 of molecular chaperones containing J-domain), that has the essential role in recruiting HSP73 (ref. 5).

We did a series of co-precipitation experiments with arthritis-predisposing and neutral DR alleles and looked for the association of HSP73. Co-precipitation of HSP73 with HLA-DR was observed in extracts of all cell lines studied, with similar intensities, using experimental conditions exactly as described in the paper by Auger et al. In repeated experiments, our data show that the previously reported association of HSP73 with DR is not allele-specific (Fig. 1). Rather, HSP73 promiscuously binds protein in certain detergent lysates. This feature reflects its role as a chaperone in recognizing unfolded protein. Immunoprecipitation experiments under more stringent lysis conditions showed that the nonspecific association of HSP73 with all HLA-DR alleles was completely negated in Triton-X100 and CHAPS lysates in physiologic salt concentrations (not shown).

Our data do not confirm HSP73 DR antigen capture as a molecular mechanism of RA. The initial observation could have been due to minor quantitative variation in non-specific binding, perhaps even post-lysis, or differences in the sensitivity of detection of HSP73 by immunoblotting. HSP73 has dual protein chaperone roles; one is a specific association, targeting protein to lysosomes, and the other is the ATP-dependent stabilization of hydrophobic regions in extended polypeptide sequences. Partial protein denaturation that may occur in certain detergent lysates will reveal hydrophobic patches leading to promiscuous binding of HSP73.

The recently demonstrated crystal structure of DRB1*0401 shows the molecule complexed with a collagen type II peptide in which the amino acid at position 71 in the shared epitope participates in both antigen binding and TCR activation⁶. This structure is consistent with existing models for the role of DR in autoimmunity, which deal with specificity of T-cell stimulation.

HSP73 co-precipitates nonspecifically with HLA class I and II DR molecules. HLA class I and II proteins were immunoprecipitated from the B lymphoblastoid cell lines SWEIG (lanes 1–3), SAVC (lanes 4–6) and HOR (lanes 7–11) using either the anti-class I monoclonal antibody W6/32 (lanes 3, 6 and 10) or the anti-class II monoclonal antibody B8.12.2. (lanes 1, 2, 4, 5, 8 and 9). Co-precipitation of HSP73 was identified by western blotting with the anti-HSP73 monoclonal antibody SPA-815. The western blot shows that HSP73 co-precipitates with either antibody or with a protein G control (lanes 7 and 11). Immunoprecipitates of B8.12.2. were washed at physiologic salt concentration (lanes 2, 5 and 9) or at 300 mM NaCl (lanes 1, 4 and 8). The far left track used 20 μg total lysate as a positive control for HSP73 binding. Methods: Immunoprecipitation conditions were as described (ref. 1 and J.A. Roudier, pers. comm.).