An important role for cytokines, particularly tumor necrosis factor α (TNFα) and interleukin-1 (IL-1), in the pathogenesis of rheumatoid arthritis (RA) is now well established (1, 2). The dramatic clinical success of treatment with TNF and IL-1 blockade has led to substantial interest in understanding the regulation of TNF and IL-1 production and in therapeutic targeting of signaling pathways and transcription factors utilized by TNF and IL-1. This includes mitogen-activated protein kinase cascades and other pathways that lead to the TNF- and IL-1–induced activation of nuclear factor κB and activator protein 1 (3–7).
The janus kinase/signal transducer and activator of transcription (JAK/STAT) pathway has received less attention among RA researchers. However, several cytokines implicated in RA pathogenesis, including IL-6, IL-15, granulocyte–macrophage colony-stimulating factor, and interferons (IFNs), activate the JAK/STAT pathway, and STAT-3 is constitutively activated in RA (8). Treatment of RA with an anti–IL-6 antibody that is predicted to block JAK/STAT activation appears promising (9). Thus, it is important to evaluate the role of JAKs and STATs in RA pathogenesis and to determine whether these molecules are good therapeutic targets. An important step in this direction has been taken by van der Pouw Kraan and colleagues, who report in this issue of Arthritis & Rheumatism that STAT-1 target genes are expressed in RA synovium (10), thus providing evidence for an active role for this transcription factor in the disease process.
The JAK/STAT signal transduction pathway
The JAK/STAT signal transduction pathway is utilized by many cytokines and growth factors that regulate gene expression and cellular activation, proliferation, and differentiation (for review, see ref.11). The binding of these cytokines to their receptors activates JAKs, protein tyrosine kinases that are physically associated with the receptor (Figure 1). Typically, stimulation by a particular cytokine results in the activation of a distinct pair of 2 of the 4 known JAKs. JAK kinases are required for tyrosine phosphorylation and activation of latent cytoplasmic transcription factors termed signal transducers and activators of transcription (STATs). STATs are rapidly tyrosine-phosphorylated after stimulation with cytokines, and they subsequently dimerize and translocate to the nucleus, where they can activate transcription. Tyrosine phosphorylation of STATs is essential for activation, dimerization, and DNA binding.
To date, 7 distinct but homologous members of the mammalian STAT family have been identified, and designated STAT-1 through STAT-6 (STAT-5A and STAT-5B are encoded by different genes) (11). Most STATs are widely expressed in a variety of cell types. An individual STAT protein may be activated by multiple ligands, but certain cytokines preferentially activate particular STATs. For example, IFNγ preferentially activates STAT-1, IL-6 and IL-10 preferentially activate STAT-3, and IL-4 preferentially activates STAT-6. The interaction of STAT SH2 domains with receptor sequences is an important determinant of the specificity of STAT activation. The JAKs and STATs that are activated by cytokines implicated in RA synovitis are shown in Table 1. Experiments with mutant cell lines and knockout mice show that STATs are critically important for mediating many, but not all, cytokine effects on gene activation and cellular phenotype (11). Activation of effector gene expression and regulation of cell growth, differentiation, and survival help explain the important role that STATs play in immune and inflammatory responses.
Table 1. Activation of the janus kinase/signal transducer and activator of transcription (JAK/STAT) pathway by cytokines expressed in rheumatoid arthritis synovium*
Cytokines that activate JAK/STAT signaling are thought to be primarily proinflammatory, except for interleukin-10 (IL-10), which is a potent deactivator of myeloid cells. LIF = leukemia inhibitory factor; OSM = oncostatin M; GM-CSF = granulocyte–macrophage colony-stimulating factor; IFN = interferon; PDGF = platelet-derived growth factor; EGF = epidermal growth factor.
IL-6 (LIF, OSM, IL-11)
JAK-1, JAK-2, TYK-2
STAT-5 > STAT-3
STAT-1, STAT-2, STAT-3, STAT-4
STAT-3 > STAT-1
STAT activation in RA synovitis
The first reports of JAK/STAT activation in RA described constitutive STAT-3 DNA binding activity in freshly isolated RA synovial fluid (SF) mononuclear cells, and showed that soluble factors contained in RA SFs effectively activated STAT-3 in control human monocytes (12, 13). IL-6 was the major STAT-3–activating factor present in RA SFs. IL-6, and probably STAT-3 activation, were necessary but not sufficient for RA SF activation of expression of the Fcγ receptor type I (FcγRI) and HLA–DR genes that are typically considered to be IFNγ-inducible genes. Thus, IL-6 worked together with other SF factors to activate monocytes in a manner similar to IFNγ. IL-6 present in RA SFs was found to activate STAT-1 in SF cells that consisted primarily of neutrophils (14), consistent with cell type–specific activation of different STATs by the same cytokine (11). Activation of STAT-3 tyrosine phosphorylation in RA synovial tissue in vivo has been detected using immunohistochemistry (15). It is not clear which cytokine activates STAT-3 during RA synovitis in vivo, but the candidates include IL-6 (and related cytokines that share the gp130 receptor subunit with IL-6: leukemia inhibitory factor, oncostatin M, and IL-11), IL-10, IL-15, and IFNα/β (type I IFNs) (Table 1). IL-6 is a good candidate, since it is highly expressed in RA synovium and represents the major STAT-3–activating factor contained in RA SF.
Previous research had established that STAT-3 was activated during RA synovitis, and elevated expression of STAT-1, STAT-4, and STAT-6 had been reported (16–18). However, it was not clear whether these STATs were fully functional during synovitis, defined by the ability to regulate gene expression and synovial cell phenotype. One approach to assess STAT function is to measure the expression of STAT-dependent genes. This approach was taken by van der Pouw Kraan and colleagues, who used microarray technology to generate a comprehensive profile of gene expression in RA synovium (10). This important study extends knowledge gained from previous gene expression profiling experiments in RA (19), and shows that RA tissues segregated into two patterns. One group of RA tissues exhibited gene expression consistent with inflammation and active immunity, with prominent expression of lymphocyte and antigen-presenting cell genes as well as genes encoding activation markers, transcription factors, signaling molecules, chemokines, and chemokine and cytokine receptors. The gene expression profile of the second group of RA tissues was more similar to that of osteoarthritis tissues. These tissues revealed low-level expression of inflammatory and immune genes, and instead expressed genes important in tissue remodeling.
It is not yet clear whether these differences in gene expression patterns reflect different classes of RA or different stages in a disease process characterized by fluctuations in disease activity and a terminal “burned-out” phase characterized by severely damaged joints with diminished inflammation. Germane to this discussion is the fact that the inflammatory group of RA tissues expressed elevated levels of STAT-1 messenger RNA and of putative STAT-1 target genes, including STAT-1 itself, GBP1, ICSBP, IP-10, caspase-1, TAP-1, IRF-1, FcγRI, and HLA class II (the latter indirectly activated by STAT-1 via class II transactivator).
STATs that have been activated by tyrosine phosphorylation can be blocked from binding to DNA or activating transcription (20). Thus, the study by van der Pouw Kraan et al (10), showing expression of STAT-1–dependent genes, represents an important advance, since it presents the strongest evidence to date that STATs in RA synovium are fully functional (i.e., they activate the expression of target genes). Although some of the “STAT-1–dependent” genes detected in this study can be activated by entirely different signaling pathways, it is unlikely that STAT-independent pathways can fully explain the observed pattern of gene expression.
An important question is whether STAT-dependent genes are activated in RA synovium by STAT-1, or whether other STATs contribute to gene expression. The different STATs recognize closely related DNA sequences (21), and previous research (22) has shown that STAT-3 can activate expression of some of the “STAT-1–dependent genes” observed in the study by van der Pouw Kraan et al. The notion of a possible role for STAT-3 in activation of STAT-dependent genes in RA synovium is further supported by the constitutive activation of STAT-3 in these tissues (13, 15), and by evidence that IL-6, together with an unknown SF factor, can activate expression of “STAT-1–dependent genes” in a STAT-3–dependent manner (12). However, the role for STAT-1 in RA synovial gene expression suggested by van der Pouw Kraan et al (10) is supported by evidence that STAT-1 in RA synovium is tyrosine-phosphorylated (Ivashkiv LB, Koch A: unpublished observations). In summary, it is likely that STAT-1 contributes to the pattern of gene expression observed by van der Pouw Kraan et al (10), but it is possible that STAT-3, and possibly other STATs, also regulate synovial gene expression.
Another important question is which cytokine(s) activates STAT-dependent gene expression in RA synovium. The strongest activator of STAT-1 is IFNγ, although IL-6, IL-10, and IFNα/β may also contribute to synovial STAT-1 activation (Table 1). It has been argued that synovial levels of IFNγ in RA are too low to activate cells (2, 23). However, the pattern of gene activation detected (10) is most consistent with a so-called “IFNγ signature,” and a recent study has identified a mechanism whereby concentrations of IFNγ as low as 5 pg/ml (0.1 units/ml) activate robust expression of genes such as IP-10, a chemokine (18). Thus, it is likely that synovial IFNγ contributes to the pattern of gene expression that was detected in the study by van der Pouw Kraan and colleagues. Those investigators also considered the interesting possibility that RA synovial STAT-1 may be activated by the B cell antigen receptor or CCR5. However, the mechanisms by which the latter receptors activate STATs have not been clarified, and it remains to be determined whether these receptors activate STATs in the cell types and under the conditions present in inflammatory RA synovium.
STAT function in RA synovitis: pathogenic or protective?
A pathogenic role for STATs in inflammatory arthritis is suggested by studies demonstrating that suppressor of cytokine signaling proteins (SOCS) attenuate experimental arthritis (15, 24, 25). However, in addition to inhibiting JAK/STAT signaling, SOCS inhibit signaling by lymphocyte antigen receptors and Toll-like receptors, modulate signaling by TNF, and suppress cytokine production (25), and the mechanisms by which SOCS inhibit arthritis have not been clarified.
The role of STATs in synovitis is not known, since STATs are pleiotropic proteins that can have different, and even opposite, activities under different conditions or in different cell types. Studies using cells and mice deficient in STAT-1 have shown that STAT-1 mediates the antiviral and immune/inflammatory effects of IFNs, and that it mediates the induction of immune effector and inflammatory genes, such as HLA, costimulatory molecule, chemokine, complement, IRF-1, inducible nitric oxide synthase, and FcγRI genes (26–29). Hence, one prediction, supported by the data reported by van der Pouw Kraan et al (10), is that STAT-1 plays a role in promoting synovial inflammation by activation of inflammatory gene expression. Alternatively, STAT-1 induces growth arrest and promotes apoptosis in several cell types, including lymphocytes and synovial fibroblasts (30–33). These functions suggest a protective role for STAT-1 in arthritis, and this possibility is supported by the elevated expression of the STAT-1 proapoptotic target gene caspase-1 in RA synovium (10), and by evidence that STAT-1 deficiency results in exacerbation of experimental arthritis (34). Thus, STAT-1 may have both pathogenic and protective roles in RA synovitis, depending on the cell type and possibly on the stage of disease and the inflammatory microenvironment. This notion is consistent with the divergent effects of IFNγ on inflammatory arthritis, which depend on the timing of administration (35).
Strong evidence supports a causal role for STAT-3 in experimental arthritis. A dominant-negative STAT-3 mutant that ablates STAT-3 function attenuated collagen-induced arthritis (15), and an IL-6 receptor knockin mutation that causes hyperactivation of STAT-3 resulted in spontaneous development of inflammatory arthritis (36). One mechanism by which STAT-3 contributes to pathogenesis is suppression of synovial fibroblast apoptosis (33), and STAT-3 also promotes T cell survival and antibody production (37). Interestingly, in myeloid cells, STAT-3 is clearly antiinflammatory and has been shown to play a critical role in suppressing expression of proinflammatory genes and production of cytokines and chemokines, including TNFα (38). Thus, STAT-3, like STAT-1, appears to have divergent effects on arthritis pathogenesis, depending on the cell type and possibly on the stage of disease.
The results of the study by van der Pouw Kraan and colleagues (10) suggest that STAT-1 may drive inflammatory gene expression in a subgroup of RA patients, and thus may represent a novel therapeutic target in RA. The divergent activities of STAT-1 suggest that it may be best to target STAT-1 selectively in nonproliferating cells, such as synovial macrophages, where it activates proinflammatory gene expression, but not in cells in which it suppresses expansion. In contrast, STAT-3 is an attractive therapeutic target in synovial fibroblasts and lymphocytes, where it promotes cell survival and proliferation, but not in macrophages, where it is antiinflammatory. To our knowledge, small molecule inhibitors of STATs have not yet been successfully developed. However, relatively selective inhibitors of JAK kinases already exist, and inhibition of protein tyrosine kinases has been effective in animal models and in the clinic (39–42). It will be interesting to determine the effects of JAK/STAT pathway inhibitors on inflammatory arthritis.
We thank Ioannis Tassiulas for critical review of the manuscript.