Our data show that IL-18 recruits monocytes in vivo
, may be produced early in the acute phase of arthritis, and signals via p38 and ERK½ to recruit PB monocytes to STs. IL-18 is known to function in an autocrine or paracrine fashion, and increased expression of IL-18 in the synovium may play a critical role for development of synovial inflammation, synovial hyperplasia, and articular degradation to which angiogenesis may contribute [37
]. Given the importance of angiogenesis in the pathophysiology of RA, we previously demonstrated a role for IL-18 as an angiogenic mediator [37
]. Supportive of this function was the finding that IL-18 has been shown to stimulate production of angiogenic TNF-α [37
We previously examined the signal-transduction mechanisms by which IL-18 induces vascular cell adhesion molecule-1 (VCAM-1) expression in RA synovial fibroblasts [31
]. In that study, we outlined how IL-18 signals through the IL-18R complex composed of both α and β chains. Concerning the IL-18R complex, the IL-18Rα chain is the extracellular binding domain, whereas the IL-18Rβ is the signal-transducing chain. When bound to the IL-18R, IL-18 induces the formation of an IL-1R-associated kinase (IRAK)/TNF receptor-associated factor-6 (TRAF-6), a multipart structure that has stimulatory activity for nuclear factor κB (NF-κB) in Th1
] and in EL4/6.1 thymoma cells [31
]. From our previous findings, we demonstrated that IL-18 induces VCAM-1 expression through Src kinase, PI3-kinase/Akt, and ERK½ signaling pathways [31
], and outlined the participation of the IRAK/NF-κB pathway in RA synovial fibroblast VCAM-1 expression.
Dinarello and colleagues [65
] showed that distinct differences exist in IL-1 and IL-18 signaling in transfected human epithelial cells, and that IL-1 signaling is primarily through the NF-κB pathway, whereas IL-18
signals via the MAPK p38 pathway. This finding may account for the absence of cyclooxygenase from IL-18
-stimulated human epithelial cells and may explain the inability of IL-18 to induce fever, unlike IL-1 [65
]. These findings also support our current signaling data showing that IL-18 induces p38 and ERK½ pathways in monocytes, confirmed by signaling inhibitory studies, Western blotting, and kinetic analysis showing that p38 is upregulated early in monocytes stimulated by IL-18, with subsequent upregulation of ERK½.
We also investigated a novel function of IL-18 to recruit monocytes in vitro
and in vivo
. Our in vitro
data showed IL-18 chemotaxic activity for monocytes at levels of IL-18 similar to those found in RA SF [5
]. We previously evaluated the role of IL-18 as an angiogenic mediator and showed that HMVECs respond to rhuIL-18 in a modified Boyden chemotaxis system [37
]. For the current study, we purchased the rhuIL-18 from the same vendor with the exact specifics regarding sample purity. Our monocyte chemotaxis findings correlate well with other studies showing IL-18 to be chemotactic for human T cells and dendritic cells [66
]. We also showed that at elevated levels beyond that found in the RA SF, IL-18 appears to be inhibitory for monocyte migration, similar to what we found in previous studies investigating MIP-3α and CXCL16 [35
]. This is likely due to a regulatory feedback loop tempering cytokine function in acute and chronic inflammatory responses.
We then attempted to link the signaling data with in vitro monocyte migration findings by inhibiting monocyte p38 and/or ERK½ with ODNs, and then tested monocyte migratory activity toward IL-18 in a modified Boyden chemotaxis system. We show that disruption of IL-18-induced monocyte signaling using antisense ODNs confirmed our earlier observations of induced monocyte p38 and ERK½ activation by IL-18, resulting in significantly reduced monocyte chemotaxis. Although we did not demonstrate a direct effect of IL-18 by inhibition of downstream kinases, we did show that inhibition of kinases activated by IL-18 can alter monocyte migration toward IL-18 in a dose-dependent manner.
From these in vitro
findings, further examination of the contribution of IL-18 in monocyte chemotaxis in an SCID mouse chimera system was warranted. To do this, SCID mice engrafted with RA ST received intragraft injections of rhuIL-18 with immediate administration of PB monocytes isolated, fluorescently dye tagged, and injected i.v. into chimeric mice, as done previously [61
]. In this setting, IL-18 proved to be a robust monocyte chemotactic agent, directing migration of human monocytes not only to engrafted ST, but also to local (inguinal) murine LNs.
Data from the SCID mouse chimera provided circumstantial evidence that IL-18 may be an effective monocyte recruitment factor in chronic diseases and supported previous findings that IL-18 gene-knockout mice have reduced inflammation in relevant models of RA [1
]. Rodent models of arthritis are indeed useful tools for studying the pathogenic process of RA. Although no model perfectly duplicates the condition of human RA, they are easily reproducible, well defined, and have proven useful for development of new therapies for arthritis, as exemplified by cytokine-blockade therapies. Furthermore, time-course studies consistently found that IL-1β, IL-6, TNF-α and other key pro-inflammatory cytokines and chemokines are functional in a variety of models, including CIA, adjuvant induced arthritis (AIA), SCW, and immune complex arthritis [68
Notably, proinflammatory IL-18 activity has been extensively examined in CIA, an accepted animal model of RA, as it shares many immunologic and pathologic features of human RA [68
]. This model is reproducible in genetically susceptible strains of mice with major histocompatibility haplotypes H-2q
by immunization with heterologous type II collagen in Complete Freund's Adjuvant. Susceptible strains are DBA/1, B10.Q, and B10.RIII [68
]. Drawbacks of this model are that, in some studies, roughly a third of the mice do not develop arthritis, inherent inconsistencies in CIA progression, and that murine CIA can take a substantial time to develop, sometimes as much as 6 to 8 weeks. In addition, many gene-knockout strains are available only on the C57BL/6 background, a strain resistant to development of CIA. Despite the many hurdles, IL-18 has been shown to play a central role in CIA [1
]. When injected into DBA-1 mice immunized with collagen in incomplete Freund's adjuvant, IL-18 increased the erosive and inflammatory component of the condition [1
]. Using mice deficient in IL-18, CIA was less severe compared with Wt controls [1
], and histologic evidence of decreased joint inflammation and destruction also was observed, outlining a direct pathologic role for IL-18.
We chose to use the ZIA model to examine the participation of IL-18 to induce a cytokine cascade by using IL-18 gene-knockout mice. Murine ZIA was first characterized by Keystone in 1977 [71
]. This model is simple and straightforward, with arthritis induction initiated by a single i.a. injection of zymosan. Of note is that ZIA apparently lacks significant lymphocyte involvement and is therefore not well suited for experiments designed for examining T-cell or B-cell function in arthritis development. ZIA was chosen for this study primarily because of the timeliness of the inflammatory response and because IL-18, a monokine, is not known to be highly dependent on lymphocyte activation.
Zymosan is a polysaccharide from the cell wall of Saccharomyces cerevisiae
. Zymosan is composed primarily of glucan and mannan residues [72
]. In vitro
, it has served as a model for the study of innate immune responses, such as macrophage and complement activation [74
]. Zymosan is also recognized and phagocytosed principally by monocytes and macrophages and leads to cellular activation and monokine production [76
], a nice feature when examining the participation of a monokine in vivo
. The subsequent inflammatory response is thought to be mediated by activation of the alternative pathway of complement and the release of lysosomal hydrolases from activated macrophages [77
]. Increasing evidence suggests that Toll-like receptors may also be involved [72
]. The advantages of ZIA include its simplicity and the fact that the resultant inflammation it induces is not strain specific. ZIA also affords the opportunity to investigate cytokines involved in joint inflammation during an acute response that may offer insight into early proinflammatory cytokine release in the initial phases of the inflammatory response. This is often lost by using models such as CIA that normally take weeks to develop [78
Using ZIA, we observed significantly reduced joint inflammation in IL-18 gene-knockout mice in as little as 24 hours after zymosan injection, and this trend continued for up to 48 hours.
We also found many proinflammatory cytokines similarly reduced in the joint homogenates of IL-18 gene-knockout mice, including IL-17, MIP-3α/CCL20, and VEGF. Although, as indicated earlier, ZIA is not known to be highly T-cell dependent, surprisingly, we found significantly reduced IL-17 and MIP-3 CCL20 in IL-18 gene-knockout ZIA joint homogenates, consistent with previous findings of CCR6, the MIP-3α/CCL20 receptor, located on T helper17
) lymphocytes [35
] This interesting finding suggests that during certain acute joint-inflammatory models, T-cell subsets may become activated and express proinflammatory lymphokines. It is tempting to speculate that during an acute inflammatory response, Th17
cell subsets are activated and recruited to the joint, which may explain the increase in IL-17 in the joint homogenates of ZIA mice. This leads to the intriguing possibility that IL-18 may regulate Th17
responses by directly supporting MIP-3α/CCL20 and IL-17 expression in STs.
Also of note were the increased MCP-1/CCL2 levels in joint homogenates from ZIA IL-18 gene-knockout mice. This seemingly paradoxic finding can be explained by noting that IL-18 may induce expression of an unidentified MCP-1/CCL2 inhibiter, much like the association of TNF-α and IL-1-receptor antagonist protein (IL-1Ra). In the latter system, it has been demonstrated in both murine type-1 chronic pulmonary inflammatory models and from treatment of RA patients with a chimeric monoclonal antibody to TNF-α that TNF-α expression supports inhibition of IL-1β by upregulation of IL-1Ra, a natural antagonist of IL-1β [29
]. We envision a similar scenario involving a regulatory loop for IL-18 and MCP-1/CCL2, as disruption of IL-18 significantly increased local expression of MCP-1/CCL2.
Detection of VEGF regulation was somewhat surprising, given the acute nature of ZIA. As ZIA is not normally thought of as a disease dependent on a high degree of vasculature, it should be noted that the inflammatory response observed in ZIA mice does indicate that proinflammatory cells are migrating freely from the peripheral bloodstream into joint tissues, presumably aided by additional vasculature mediated by VEGF. Furthermore, monocytes respond, produce, and migrate toward VEGF, providing further evidence that recruited monocytes may amplify the angiogenic process in acute inflammatory tissues expressing IL-18. It is tempting to speculate that the effects of VEGF on vasculature growth may exacerbate the IL-18-induced pathology seen in murine ZIA.
We found that IL-18 stimulates monocyte migration both in vitro and in an RA ST SCID mouse chimera system. We further show that this is mediated by activation of the p38 and ERK½ signaling pathway. We confirmed the latter finding by use of ODNs designed to disrupt this pathway and, in so doing, significantly reduced IL-18-mediated monocyte chemotaxis. We also showed that IL-18 gene-knockout mice have reduced ZIA, an acute model of RA, and that mice lacking IL-18 have significantly reduced joint homogenate levels of IL-17, MIP-3α/CCL20, and VEGF
Overall, this study indicates that IL-18 is effective very early in acute inflammatory models by inducing proinflammatory cytokine release and monocyte migration to STs, lending support to the notion that IL-18 plays a hierarchic role in the inflammatory cytokine cascade during arthritis development.