Tumor necrosis factor α (TNFα) plays a central role in the pathogenesis of rheumatoid arthritis (RA) because it is at the apex of inflammatory and destructive processes that operate in the joint. Clinical trials using anti-TNFα blocking biologic agents have demonstrated that TNFα regulates the expression of inflammatory cytokines and chemokines, adhesion molecules, and thus, leukocyte trafficking in the joints, matrix metalloproteinases (MMPs), and joint destruction, as well as vascular endothelial growth factor (VEGF) and angiogenesis (1–5).
Three anti-TNFα blocking agents based on TNFα-neutralizing monoclonal antibodies or soluble TNF receptors (infliximab, adalimumab, and etanercept) have now been approved for human use and many more are under development, but they have some major disadvantages, including their high cost and the inconvenient administration route that requires repeated injections (6–8). Thus, research is currently under way to look for better means to block TNFα expression by designing small molecule inhibitors. This requires the determination of what regulates or is rate-limiting for TNFα expression in human macrophages, the major cell type producing TNFα in the RA joint. TNFα gene expression is under complex control, with the p38 mitogen-activated protein kinase pathway controlling translation possibly by actions on the 3′-untranslated region (9), and with the 5′-untranslated/promoter region containing binding sites for multiple transcription factors, including nuclear factor κB (NF-κB), activator protein 1, nuclear factor for the interleukin-6 (IL-6) gene, and nuclear factor of activated T cells (10–12). NF-κB has recently attracted particular attention because of its ability to regulate macrophage TNFα production induced in response to lipopolysaccharide (LPS), ultraviolet light, phorbol myristate acetate, or contact with cytokine-activated T cells (13, 14).
NF-κB has been detected by immunohistology in human synovial tissue from RA patients during both early and later stages of disease and in both macrophage- and fibroblast-like synoviocytes (15, 16). In particular, macrophage-like synoviocytes that localize in the synovial lining layer and the vascular endothelium have been shown to contain p65 and p50 NF-κB subunits in their nucleus (16). This NF-κB activation is of functional relevance, since we recently showed that it regulates the expression of TNFα as well as other proinflammatory cytokines, such as IL-1β, IL-6, IL-8, MMP-1 (collagenase 1) and MMP-3 (stromelysin 1), without major effects on the expression of antiinflammatory cytokines IL-10, IL-11, and IL-1 receptor antagonist, soluble TNF receptors, or tissue inhibitor of metalloproteinases 1 (17) in ex vivo synovial membrane cultures. In streptococcal cell wall– and pristane-induced arthritis, inhibition of NF-κB through the administration of κB decoy oligonucleotides or proteasome inhibitors has also been shown to be beneficial (18, 19).
However, because NF-κB is involved in normal immune and homeostatic processes, such as the prevention of apoptosis in certain tissues (e.g., liver) (20–22), its prolonged inhibition may have hazards and is unlikely to be a direct therapeutic target. Much more likely and more therapeutically attractive would be the selective inhibition of upstream molecules such as IκB kinase 2 (IKK-2), a kinase previously shown to be essential for TNFα- and IL-1–induced IκBα degradation, NF-κB activation, and cytokine expression (21–28). One study recently demonstrated that IKK-2 is an important kinase required for TNFα- or IL-1–induced NF-κB activation and expression of IL-6, IL-8, intercellular adhesion molecule 1, and MMP-1 in passaged synoviocytes (29), although the significance of these observations in terms of the treatment of RA is not clear. Passaged synoviocytes obtained from patients with RA do not produce TNFα, are not representative of the rheumatoid synovium because the process of growing them selects for the fibroblast-like cells that probably represent ∼10% of the total synovium, and require in vitro stimulation to produce many of the inflammatory mediators seen in vivo.
Another study performed in the rat adjuvant-induced arthritis model demonstrated significant benefit after intraarticular injection of an adenoviral construct encoding a dominant-negative form of IKK-2 (IKK2dn) after disease onset that correlated with a decrease in NF-κB DNA-binding in the nucleus of synovial cells (30). However, that study did not provide the potential mechanisms involved. Moreover, rat adjuvant-induced arthritis is distinguished from RA by its lack of chronicity (31). This suggests differences in disease pathogenesis and makes it difficult to extrapolate observations made in rat adjuvant-induced arthritis to the human situation. Thus, there is a need to further study the role of IKK-2 in systems that more closely resemble human RA.
In this study, we examined the involvement of IKK-2 and IKK-1 in NF-κB activation and TNFα production in primary human macrophages, the main cell type producing TNFα in the RA joint, and in ex vivo RA synovial cultures that consist of the entire population of cells found in vivo (i.e., ∼30% T cells, 30–40% macrophages, and fewer fibroblasts, endothelial cells, dendritic cells, plasma cells, and B lymphocytes) (32) and that, in the absence of extrinsic stimulation, spontaneously produce the same inflammatory mediators seen in vivo. We then extended our studies to other primary human cells, such as RA synovial fibroblasts, normal dermal fibroblasts, and human umbilical vein endothelial cells (HUVECs), as well as other molecules involved in inflammatory, angiogenic, or destructive processes in RA, such as IL-1β, IL-6, IL-8, VEGF, and MMPs 1, 3, 9, and 13 (collagenase 3). Our results are the first to demonstrate that IKK-2 is not universally required for TNFα production in the RA synovium, although it is required for the production of other inflammatory cytokines (IL-1β, IL-6, and IL-8), VEGF, or MMPs. This has important implications for the targeting of this kinase in RA.
- Top of page
- MATERIALS AND METHODS
TNFα blockade is a validated treatment in RA, and extensive research is under way to target molecules that control TNFα expression. Two such molecules are NF-κB and IKK-2, a kinase that activates NF-κB and NF-κB–dependent gene expression in a number of systems. NF-κB has been shown to be activated in RA and to be of functional relevance since it controls the expression of TNFα and other proinflammatory cytokines and MMPs (17, 57). However, no such role for IKK-2 has yet been demonstrated. In this study, we examined the role of IKK-2 in this process and found that IKK-2 is not essential for TNFα production in RA ex vivo synovial membrane cultures from patients undergoing joint replacement surgery. This was unexpected and prompted us to investigate why this is the case.
We found that there is heterogeneity in the requirement of IKK-2 for NF-κB activation and NF-κB–dependent gene expression that depends on the stimulus and cell type used (Figure 6). Thus, in primary human macrophages, the main producers of TNFα in the RA joint, IKK-2 is not required for LPS-induced NF-κB activation and TNFα, IL-6, or IL-8 production, whereas it is still essential for CD40L-, TNFα-, or IL-1–induced NF-κB activation and TNFα, IL-6, or IL-8 production. IKK-1 is not alternatively used instead of IKK-2 to induce NF-κB activation and cytokine production in response to LPS. However, in dermal fibroblasts, RA synovial fibroblasts, and HUVECs, IKK-2 is essential for TNFα-, IL-1–, and LPS-induced NF-κB activation and IL-6 or IL-8 production. This is due to a direct effect of IKK-2 on IκBα degradation and NF-κB activation, and is consistent with other studies that have examined some of these aspects (29, 39, 58).
Figure 6. Cell type– and stimulus-dependent requirement of IκB kinase 2 (IKK-2) for nuclear factor κB (NF-κB) activation and inflammatory cytokine production in humans. SF = synovial fibroblasts; HUVEC = human umbilical vein endothelial cells; TNFα = tumor necrosis factor α; IL-1 = interleukin-1; CD40L = CD40 ligand; LPS = lipopolysaccharide; excl. = excluding; RA = rheumatoid arthritis.
Download figure to PowerPoint
In contrast, our findings in primary human macrophages are inconsistent with those of previous studies of THP-1 human monocytic cells that showed a 50% inhibition of an NF-κB reporter gene by IKK2dn. These studies did not examine endogenous NF-κB–dependent promoters, nor did they look at cytokine gene expression induced by LPS (59, 60). Moreover, IKK1dn could be equally effective as IKK2dn (60), a result that is inconsistent with evidence from mice deficient in these molecules (21, 22, 25–27). In addition, our findings suggest significant differences in LPS signaling between macrophages and endothelial cells or fibroblasts. A possible reason for the discrepancy in these data is that, as for nuclear factor κB–inducing kinase (NIK), there is a different usage of signaling components between transformed cell lines (e.g., THP-1 cells), where cell proliferation is the essential process, and primary cells such as macrophages, which do not divide (42, 61, 62). However, in the case of IKK-2, the situation appears more subtle, since IKK-2 is also involved in LPS signaling in primary cells, such as HUVECs and synovial fibroblasts. The fact that IKK2dn is expressed and is effective at blocking signaling by IL-1, TNFα, and CD40L would discount questions about the effectiveness of AdIKK2dn as an inhibitory construct.
Differential usage of IKK-2 between macrophages and synovial fibroblasts or HUVECs may be due to the involvement of alternative adapter molecules in signaling through the LPS receptor TLR-4. Two such molecules, MyD88 and Mal/TIRAP, have already been reported to play a role in TLR-4–induced TNFα and IL-6, but not IP-10, production (63, 64). It has been suggested that IP-10 production requires the activation of interferon-regulatory factor 3 (65), and this may depend on the adapter molecule TRIF/TICAM-1, another recently identified MyD88 homolog (66, 67). Thus, distinct adapter molecules of TLR-4 are likely to engage alternative signaling pathways and alternative kinases involved in IκBα phosphorylation. In addition to IKK-1 and IKK-2, other kinases have been shown to phosphorylate IκBα, including IKKε/IKKi, p90rsk, PKR (double-stranded RNA–dependent kinase), casein kinase II, and the catalytic subunit of DNA-dependent kinase (68–73), although it remains to be determined whether these kinases specifically phosphorylate both Ser32 and Ser36 of IκBα. The importance and involvement of these kinases in IKK-2–independent pathways of LPS signaling need to be addressed.
Interestingly, in ex vivo RA synovial membrane cells, we found that although IKK-2 is not required for TNFα production, it is still essential for the production of IL-1β, IL-6, and IL-8. This paralleled the results with AdIκBα that confirmed our previous observations that NF-κB is required for IL-1β, IL-6, and IL-8 production (17, 35). In addition, we found that IKK-2 and NF-κB are essential for the expression of the angiogenic cytokine VEGF and the extracellular matrix–degrading enzymes MMPs 1, 3, 9, and 13. Inhibition was not due to increased cell death of AdIKK2dn-infected cells, a finding that was consistent with our previous work in RA synovial membrane cells using AdIκBα (17, 35). On the other hand, IKK-1 was not involved in the release of any of these mediators, with the exception of VEGF. How IKK-1 is involved in VEGF expression in RA synovial membrane cultures is not yet clear, but this may require processing of NF-κB2, since we have evidence that in primary human macrophages that produce NF-κB–dependent VEGF in response to various stimuli (74), NIK is also involved (Andreakos E and Kiriakidis S: unpublished observations).
Because rheumatoid synovial membrane cultures consist of a complex mixture of cells comprising T cells, macrophages, and fibroblasts, these observations may reflect the selective use of IKK-2 for NF-κB activation and NF-κB–dependent gene expression in different cell types and in response to different stimuli. While rheumatoid macrophages are the major source of TNFα, isolated rheumatoid fibroblasts can produce abundant IL-1β, IL-6, IL-8, VEGF, and MMPs (but not TNFα) in response to cytokines in an IKK-2–dependent manner, as our data and those of other investigators (58) demonstrate. Another, but not exclusive, explanation for this observation could be the ability of IKK2dn to block TNFα-mediated NF-κB activation and NF-κB–dependent gene expression. Such a mechanism is supported by our previous finding that TNFα is at the apex of a proinflammatory network or cascade that operates in the rheumatoid synovium, and that blockade of TNFα activity by neutralizing antibodies results in the down-regulation of IL-1β, IL-6, IL-8, and MMP production both in dissociated synovial membrane cultures in vitro (37, 38) and in patients with RA in vivo (3).
In contrast to RA, studies that we performed recently in bronchoalveolar lavage cultures from patients with fibrosing alveolitis showed that IKK-2 is required for the production of TNFα as well as IL-6 and IL-8 (75). The constituent cells of this system are just alveolar T cells and macrophages, and TNFα is mainly produced by macrophages. However, it is likely that the primary stimulus inducing TNFα production in patients with fibrosing alveolitis is different from that in patients with RA, and different or alternative signaling mechanisms are used. In terms of RA therapy, these observations suggest that although IKK-2 is not involved in TNFα production, it still plays a central role in the inflammatory, angiogenic, and destructive processes that operate in the RA synovial joint. This may have the advantage of not compromising the immune system in a major way, since macrophage function is partially maintained. Whether IKK-2 inhibition will be clinically effective can only be resolved by clinical trials, which are eagerly awaited.