Inhibiting citrullination in rheumatoid arthritis: taking fuel from the fire

Citrullination is a post-translational modification catalysed by peptidylarginine deiminase and is a common feature of inflammation. The presence of anti-citrullinated protein/peptide antibodies (ACPA), however, is unique to rheumatoid arthritis. Several lines of evidence suggest that ACPA are important in the pathogenesis of rheumatoid arthritis. A popular hypothesis for this pathogenesis is a two-hit model. The first hit gives rise to ACPA, and the second hit, an unrelated episode of synovial inflammation accompanied by citrullination, is perpetuated by the pre-existing antibodies. This model suggests that reducing citrullination might ameliorate disease. Recent findings indicate that citrullination closely correlates with inflammation, and that glucocorticoids decrease peptidylarginine deiminase expression independent of their other anti-inflammatory effects.

In this issue, Makrygiannakis and colleagues study the eff ect on synovial citrullination of treatment with two commonly used drugs in the treatment of rheumatoid arthritis (RA) [1]. Th ey found by immuno histo chemistry that intracellular citrullination, as deter mined by F95 antibody staining, as well as peptidyl arginine deiminase (PAD) expression were correlated with measures of synovial infl ammation. Intra-articular injection of glucocorticoid, but not oral methotrexate, was associated with a reduction in synovial infl ammation, intracellular citrulli nation, and PAD expression. Based on cultures of synovial fl uid mononuclear cells and synovial explants, they also make the interesting proposal that gluco corticoids may suppress citrullination independent of infl ammation by inhibiting PAD enzyme expression. Th ere are a number of caveats to this proposition, including the specifi city of anti-PAD antibodies. In our laboratory we have found a number of these antibodies to be crossreactive with other proteins, which could confound immuno histochemistry fi ndings unless specifi city is confi rmed or the results are corroborated by other techniques. Neverthe less, given the dearth of information on the regulation of citrullination in RA, their paper is welcome.
Th e presence of anti-citrullinated protein/peptide antibodies (ACPA) defi nes a major subset of RA that is associated with distinctive genetic and environmental risk factors and with a more severe clinical phenotype [2]. Frequently cited evidence to support the importance of ACPA in pathogenesis includes their appearance years before clinical diagnosis, their production within the joint, the ability of ACPA immune complexes to activate macrophages, and some animal model data [2]. Th e actual role of ACPA in pathogenesis is still a matter for investigation. A widely-held hypothesis for this pathogenesis comprises two hits [2]. Th e fi rst hit follows accelerated citrulli nation of proteins in an extra-articular site -due to smoking or infection, for example -which in the context of a permissive HLA type gives rise to ACPA. Th e second hit, which may occur years later, would be an unrelated episode of otherwise self-limiting synovial infl ammation. Since citrullination of proteins is a feature of infl am matory tissue, the presence of preexisting ACPA would exacerbate and perpetuate the synovitis. If this was to be the case, one might predict that inhibiting citrullination would ameliorate diseasethe fi ndings of Makry giannakis and colleagues suggest this may be a previously unappreciated mechanism of action of glucocorticoids [1].
Such a hypothesis argues for investigation of specifi c PAD inhibitors. Th e chemotherapeutic drug paclitaxel inhibits PAD enzymes and is effi cacious in a rat collageninduced arthritis model [3]. Th is agent has other notable eff ects relevant to RA, however, including inhibition of microtubule formation and angiogenesis. More recently, Willis and colleagues reported that the pan-PAD

Abstract
Citrullination is a post-translational modifi cation catalysed by peptidylarginine deiminase and is a common feature of infl ammation. The presence of anti-citrullinated protein/peptide antibodies (ACPA), however, is unique to rheumatoid arthritis. Several lines of evidence suggest that ACPA are important in the pathogenesis of rheumatoid arthritis. A popular hypothesis for this pathogenesis is a two-hit model. The fi rst hit gives rise to ACPA, and the second hit, an unrelated episode of synovial infl ammation accompanied by citrullination, is perpetuated by the pre-existing antibodies. This model suggests that reducing citrullination might ameliorate disease. Recent fi ndings indicate that citrullination closely correlates with infl ammation, and that glucocorticoids decrease peptidylarginine deiminase expression independent of their other anti-infl ammatory eff ects.
irrever sible inhibitor Cl-amidine partially inhibited arthritis in a mouse collagen-induced arthritis model [4]. Th ey ob served a reduction in antibody levels to native, but not bovine, collagen, and to a limited number of other candi date autoantigens that they studied by microarray. Th e only reduction in ACPA reactivity was to fi laggrin. Inter est ingly, despite a reduction in clinical and histological arthritis scores, synovial infi ltration by immune cells was unaff ected. Given that ACPA autoimmunity is unlikely to be of major importance in collagen-induced arthritis, the role of citrullination in this model, the mechanism of action of Cl-amidine, and the relevance of the fi ndings to RA are all unclear.
Th e safety of PAD inhibition is a major consideration. Th e physiological role of the diff erent PAD enzymes is incompletely understood but there are proteins in the stratum corneum and myelin sheath that are constitutively deiminated. Citrullination also appears to play a role in apoptosis, formation of neutrophil extracellular traps, altering chemokine function and, in the case of PAD4, regulating DNA transcription [5]. Only PAD2 and PAD4 have been identifi ed in the synovium [6] and most attention in RA has focused on PAD4, which has a more limited tissue distribution being mainly expressed in leucocytes. PADI4 polymorphisms are associated with RA in East Asian populations [2], and autocitrullinating PAD4 is itself an autoantigen in RA [7]. Whilst PAD2 can be found in the synovium of both osteoarthritis and RA, Foulquier and colleagues could only identify PAD4 in infl ammatory synovitis [6]. It is therefore of interest that Makrygiannakis and colleagues found glucocorticoids to reduce PAD4 consistently, but not PAD2 [1].
PAD2 levels, however, are known to increase as monocytes diff erentiate into macrophages, which make up a large proportion of lining cells in RA synovium [8]. Studies of syno vial fl uid in our laboratory showed that PAD4 could be found in both osteoarthritis and infl ammatory arthritis, whereas PAD2 was limited to infl ammatory disease [9]. Notably, Makrygiannakis and colleagues found synovial PAD2 levels to correlate with infl ammation rather better than PAD4 [1].
Only a small number of potentially citrullinated proteins are antigenic targets in RA, and evidence is emerging that PAD2 and PAD4 diff er in their substrate specifi city [10]. Further studies of these enzymes are needed to better understand their regulation, their relative contribution to citrullination in RA and, in particular, their substrate specifi city, to guide therapeutic development. Vossenaar and van Venrooij have described citrullinated proteins as 'sparks that may ignite the fi re of RA' [11]. Th e fi ndings of Makrygiannakis and colleagues suggest that PAD enzymes may also provide the fuel that keeps the fi re burning, and that their inhibition may be a key target for novel therapy.