In reactive arthritis, microbial exposure has been clearly implicated as an antecedent event in the development of synovial inflammation. The role of pathogens may come into play through interaction of the innate immune system and pattern recognition receptors. Toll-like receptors (TLRs) in humans have been identified as pattern recognition receptors that can recognize pathogen-associated molecular patterns. Microbial products such as lipopolysaccharide (LPS), peptidoglycan, or zymosan function as TLR ligands. Although the TLRs differ from the cytokine receptors for interleukin-1 (IL-1) and IL-18 in terms of their extracellular domain structure, similar cytoplasmic domains (the Toll/IL-1 receptor domain) allow most TLRs to use the same signaling molecules (for review, see ref. 1).
The predominant signaling pathways for all of the TLRs except TLR-3 include myeloid differentiation factor 88 (MyD88), IL-1 receptor (IL-1R)–associated protein kinase (IRAK), and tumor necrosis factor receptor (TNFR)–associated factor 6. Subsequent TLR signal transduction results in the activation of 2 different pathways that involve the JNK and MAPK families and the Rel family transcription factor NF-κB. An MyD88-independent pathway exists, and following signaling through TLR-3 or TLR-4, it leads to the production of type I interferons (Figure 1). Hence, recognition of these ligands by the innate immune system leads to a cascade of events, including the release of cytokines and activation of antigen-presenting cells (for review, see ref. 1).
In this issue of Arthritis & Rheumatism, Kyo et al present the results that they obtained using a murine model of reactive arthritis (2). Those investigators primed an immune response by injecting an extract of enteric bacterial products including Escherichia coli cell wall (ECW) extract into the footpads of mice. Inflammatory arthritis was induced with an intraarticular injection of ECW or LPS. Compared with instilling LPS alone, priming the mice before administration of the intraarticular injection resulted in augmented cellular infiltration and subsequent bone and cartilage damage. Interestingly, the bacterial extract for Propionibacterium acnes, which has been implicated in the development of the SAPHO syndrome (synovitis, acne, pustulosis, hyperostosis, and osteitis), could not be substituted for the E coli preparation. When the investigators expanded their analysis to genetically disrupted mice, they demonstrated that arthritis was markedly attenuated in TLR-4 mutant and Myd88−/− mice, which suggests that the reactivity was secondary to this pathway. Although TNFα and IL-6 were abundant in the joints, Tnf−/− and TnfrI−/− mice developed arthritis after induction. However, Il18−/− and Il18/Il12−/− mice had attenuated arthritis, while Il6−/− mice had minimal disease. In summary, in this model, TLR ligands were significantly involved in the priming of an inflammatory response using bacterial extracts, with the introduction of microbial products into the joint resulting in an inflammatory reaction dependent on MyD88 and IL-6.
Current evidence indicates that inflammatory arthritis is a multicellular process, with 3 phases (for review, see ref. 3). The induction phase involves cells that are resident in the synovium that may “pre-prime” or sensitize the joint to further inflammation prior to the development of clinically apparent synovitis. The second phase ensues when lymphocytes, macrophages, and antigen-presenting cells have been recruited to the joint. During this phase, the innate immune system may be instrumental in creating the priming milieu that defines the type of T cell response to joint-associated antigens. For instance, mice that are deficient in MyD88 fail to mount a typical Th1 response when immunized with antigen and Freund's complete adjuvant (4). In contrast, deficient TLR-4 signaling results in an impaired ability to generate a Th2 response (5). The third and most debilitating phase of disease results in cartilage damage and bone erosions. These destructive mechanisms may be primarily driven by osteoclasts and proteases secreted by pannocytes. In a recent trial of anti–TNFα antibody therapy, an arrest in the progression of radiographic joint destruction occurred independently of improvement in measures of inflammation (6). These data suggest that in established disease there is an uncoupling of antigen-dependent inflammation and bone and cartilage erosions.
Although the dominant role of the innate immune system in sensitizing and propagating joint inflammation is likely to be antigen independent, interactions with the adaptive immune system may play key roles in arthritis. Pathogen-associated molecular patterns trigger dendritic cells (DCs) to release cytokines such as TNFα or IL-12, which are important for the differentiation of T cells into Th1 cells. In addition, these stimuli induce up-regulation of costimulatory molecules on DCs, including CD40, CD80, and CD86, which provide the necessary second signals for T cell priming (for review, see ref. 1). The release of cytokines and chemokines into the local environment after TLR ligation likely influences the priming of the adaptive immune response and may also propagate an inflammatory cascade that is relatively independent of T cell specificity. Hence, innate mechanisms likely play an integral part in the initiation of synovitis as well as in the subsequent phases of arthritis.
Similar to the model system used by Kyo and colleagues, several other rodent models require systemic exposure to TLR ligands in adjuvants for the subsequent development of arthritis. In the rat model of adjuvant-induced arthritis, the presence of mycobacterial DNA is necessary, but not sufficient, for maximal joint inflammation (7). Mycobacterial DNA can be replaced by synthetic oligodeoxynucleotides, which stimulate TLR-9. In treated rats, the mycobacterial DNA did not enter the joints but rather dispersed to the bone marrow and spleen before the onset of systemic joint inflammation. In this model, prolonged activation of macrophages, DCs, and osteoclast precursors in the bone marrow may prime joints for the induction of inflammatory Th1 immune responses to mycobacterial antigens (7). In the collagen-induced arthritis model, the administration of TLR-9 agonists prior to antigen administration results in more aggressive disease, both clinically and histologically (8). These results are similar to the pre-priming effects observed in the E coli extract model, whereby the TLR ligands act as adjuvants in the priming step and then later act as stimulants.
Remission of arthritis typically occurs in mice with antigen-induced arthritis. The introduction of TLR ligands has been used to augment disease in these models. For example, the extent of disease in mice with chronic peripheral arthritis after immunization with type II collagen is markedly enhanced using a protocol in which LPS is administered several days after the antibody injection (9). Similarly, LPS has been used to prolong the disease course. In the setting of previous joint damage, reactivation of synovial inflammation has been induced by systemic exposure to LPS or peptidoglycan–polysaccharide (10–12). Reactivation of disease through the use of TLR ligands systemically, or locally as done by Kyo et al, indicates that innate mechanisms may perpetuate joint inflammation without a specific concomitant antigenic insult.
TLR signaling results in the release of proinflammatory cytokines such as IL-6 and TNFα from a variety of bone marrow–derived cells and barrier tissues, including synovium. The relative dependence of joint swelling and inflammation on IL-6, as seen in the ECW model, has also been noted in other murine models that are influenced by microbial products but are not entirely dependent on TLR signaling. For example, in mice with a defect in T cell signaling, the spontaneous development of arthritis is predominantly dependent on IL-6 (13). In these mice, arthritis is provoked, in a pathogen-free environment, by injection of zymosan, a crude fungal β-glucan, or injection of a TLR-3 agonist (14). Although the activity of zymosan can be attributed to its interaction with dectin 1 (the major β-glucan receptor), in other models, zymosan is a TLR-2 agonist. Indeed, the chronicity of zymosan-induced arthritis has been attributed to the expression of IL-6 and the balance of suppressor of cytokine signaling 1 (SOCS-1) and SOCS-3 expression (15).
The spontaneous arthritis observed in mice that are transgenic for human TNFα and IL-1R antagonist suggests that stromal elements can influence their articular disease (16–18). Furthermore, when DBA/l mice that lack mature lymphocytes are immunized with collagen and adjuvant, synovial hyperplasia and bone erosion develop (19). The TLR ligands in the adjuvant may cause the release of cytokines in joint tissues, which then perpetuate a destructive cascade. These joint-associated cells respond to several TLR ligands that have been directly administered intraarticularly (20–22). Similarly, in vitro treatment of human fibroblast-like synoviocytes with selected TLR ligands has resulted in NF-κB activation and expression of proinflammatory cytokines, chemokines, adhesion molecules, and matrix metalloproteinases (23).
The most significant role of TLR and MyD88 may involve the perpetuation and maintenance of chronic inflammation. In the setting of serum-induced arthritis, in which the priming step of the adaptive immune response is bypassed, Myd88−/− and Il1R−/− mice do not develop arthritis. Stimulation of TLR-4 with LPS, however, can restore the responsiveness of IL-1R–deficient mice to the arthritogenic serum (24). TLR-4 ligands are not microbial products only and include host-derived molecules such as hyaluronic acid, heparan sulfate, fibronectin A, and heat-shock proteins (1). The recognition of microbial products as well as components of the extracellular matrix by TLRs suggests that these receptors may be involved in the recognition of tissue damage (Figure 1). Both fibroblasts and macrophage-like synoviocytes may be directly stimulated by molecules released during the acute phase of tissue damage rather than by direct exposure to pathogens. The expression of chemokines and cytokines may further trigger chronic inflammation, with recruitment of inflammatory cells to the synovium and destruction of cartilage and bone. This mechanism could lead to self-perpetuating loops of cell activation in the latter stages of arthritis that require minimal, if any, adaptive immune mechanisms. Along with other exciting research, the study by Kyo et al places TLRs at center stage in arthritis. Further studies will determine whether TLRs can be targets of therapy.