In this study we characterised citrullinated α-enolase as a dominant antigen in citrullinated lysates of differentiated HL-60 cells targeted by a screening panel of serum from patients with RA. The identity of the antigen was established by mass spectrometry, and the sites of nine citrulline residues within the protein were determined by tandem mass spectrometry. Further confirmation was obtained by two-dimensional electrophoresis and Western blotting with a specific anti-enolase antibody. With the use of purified protein, 46% of a larger panel of sera from patients with RA reacted with citrullinated α-enolase by immunoblotting. This suggests that citrullinated α-enolase is at least as immunodominant as citrullinated filaggrin or citrullinated vimentin, because, by immunoblotting, the frequency of antibodies against citrullinated filaggrin has been reported as 41 to 58% [25
] and against citrullinated vimentin as 22 to 40% [19
]. Improved sensitivity and specificity of RA diagnosis may well be obtained by testing RA sera with peptides derived from citrullinated epitopes of α-enolase, as has been demonstrated for citrullinated filaggrin in the first-generation anti-CCP test, in which the sensitivity increased to more than 70%.
α-Enolase, unlike filaggrin, is abundantly expressed in the synovial membrane. Several lines of evidence indicate that it is citrullinated in the joint. First, it was detected in the myeloid-like HL-60s cell line, which expresses PAD and has a similar phenotype to that of cells abundant in the joint. Second, it was detected as a synovial antigen that co-localised with staining for citrullinated proteins. The staining shown in Figure suggests that only a small proportion of the antigen is citrullinated in vivo, which might explain why we were unable to demonstrate citrullination of α-enolase by Western blotting of immunoprecipitates from synovial cells.
Although this is the first report that citrullinated α-enolase is a common target antigen in RA, native α-enolase has previously been observed as an infrequent target antigen for several autoimmune diseases [31
]. For example, Saulot and colleagues [34
] observed that antibodies against (placental) α-enolase occurred in 25% of patients with RA and were predictive of radiological progression. They found that only 8 of the 36 patients reacting with placental α-enolase also reacted with the recombinant protein. This contrasted with 19% of patients with systemic lupus erythematosus and 15% of patients with systemic sclerosis whose serum samples reacted with both forms of α-enolase. They hypothesised that RA sera reacting with placental α-enolase, but not recombinant antigen, were recognising a post-translationally modified form of α-enolase. Although it is tempting to speculate that the modification they predicted is citrullination, the most abundant of the triplet of spots identified as α-enolase in their study migrated in two-dimensional electrophoresis at a pI of 7.0, consistent with native α-enolase. However, it is possible that the two more acidic α-enolase spots, which they attributed to phosphorylation, might in fact be citrullinated. The expression of PAD2 protein in the placenta would account for a degree of deimination either in vivo
or during extraction. It is also consistent with the identification of the Sa antigen, also of placental origin, as citrullinated vimentin [17
]. The higher frequency of anti-citrullinated α-enolase in our study than that of Saulot and colleagues might be due to the fact that our cell lysates were extensively deiminated in vitro
. This is demonstrated by the uniform migration of deiminated enolase at a pI of 5 and by the replacement of arginine by citrulline in all the peptides listed in Table .
In our study, 15% of the control sera reacted with native α-enolase and also with citrullinated α-enolase, whereas reactivity with the citrullinated form alone was restricted to the patients with RA. This is, again, consistent with the results of Saulot and colleagues, assuming that the placental form of the protein was partly deiminated. In turn, this suggests that RA-specific antibodies might be driven by peptides containing one or more of the 17 potential citrulline residues in the sequence of α-enolase. Binding to non-arginine containing regions might account for the 'background', and hence the apparent loss of disease specificity seen when immunoblotting with the normal sera in our study, and the non-RA sera in that of Saulot and colleagues. One way to test this would be to examine reactivity to peptides derived from citrullinated epitopes from α-enolase. Such studies are currently in progress. If both sensitivity and specificity increases for RA it might provide another assay for diagnosis of the disease. More importantly it would provide further data to support the concept that α-enolase is a driving autoantigen in RA.
The properties of citrullinated α-enolase make it an attractive synovial antigen for driving the immune response. α-Enolase is a highly conserved, multifunctional protein that, in addition to its role in glycolysis, binds plasminogen. It is known to be upregulated by hypoxia [35
] and by proinflammatory stimuli [36
], both of which are features of the synovial membrane microenvironment in RA. α-Enolase is expressed during cell differentiation and is used as marker of differentiation in the grading of tumours [37
]. In myeloid cells, the dominant cell type in the inflamed synovium, it is co-expressed with PAD2 and PAD4. In the present study we have shown that its distribution in the subsynovium is similar to that of citrullinated proteins and PAD [38
], but we have not yet demonstrated its citrullination in vivo
There is substantial similarity between human and prokaryotic α-enolases (47% identity with that from Streptococcus pyogenes
, for example), and antibodies raised against streptococcal surface α-enolase also recognise the human enzyme [36
]. Thus the presence of antibodies against uncitrullinated α-enolase, in serum of individuals without RA, might be attributable to cross-reaction with bacterial epitopes. Expression of an enzyme able to citrullinate peptidylarginine has been demonstrated in the oral organism Porphyromonas gingivalis
], which provides a mechanism by which antibacterial antibodies cross-react with endogenous citrullinated proteins and initiate loss of tolerance.