The transcription factor NF-κB might be the Holy Grail of therapeutic targets for rheumatoid arthritis (RA). It serves as a “master switch” for an inflammatory cascade that encompasses many key genes involved in synovitis, including cytokines, metalloproteinases, and regulators of small-molecule mediators (1). NF-κB suppression is demonstrably beneficial in a host of animal models of inflammatory disease, and many therapeutic agents, including corticosteroids, leflunomide, sulfasalazine, and aspirin, could mediate at least some of their effects through NF-κB blockade. The creativity of diverse investigators driven to block this pathway is impressive, and has led to experiments using small-molecule inhibitors of regulatory enzymes, gene therapy, decoy oligonucleotides, and various biologics. In this issue of Arthritis & Rheumatism, Blackwell and colleagues (2) describe a novel approach to inhibiting NF-κB activation through the delivery of a fusion protein that includes components of the natural inhibitor of NF-κB, called IκB, and a viral-derived chaperone that escorts proteins through the plasma membrane into the cytoplasm. This novel approach suppresses local cytokine production and the influx of leukocytes into sites of inflammation.
Inflammatory and immune responses, especially after activation of primitive pattern-recognition receptors by pathogens, are largely coordinated by NF-κB (3). For instance, the transcription of many cytokine genes, including interleukin-6 (IL-6), IL-8, tumor necrosis factor α (TNFα), and IL-1β, is initiated by NF-κB activation. Induction of adhesion molecules on endothelial cells (vascular cell adhesion molecule 1 [VCAM-1], E-selectin, and intercellular adhesion molecule 1 [ICAM-1]) with recruitment of inflammatory cells to extravascular sites is also mediated by this transcription factor (4). NF-κB–dependent tissue remodeling and increased vascular permeability through the expression of metalloproteinases, inducible nitric oxide synthase, and cyclooxygenase 2 contribute to local injury. Even antibody production and T cell–dependent delayed-type hypersensitivity use this pathway.
NF-κB is a complex group of heterodimeric and homodimeric transcription factors. Key members of this family include NF-κB1 (p50/p105), NF-κB2 (p52/p100), RelA (p65), RelB, and c-Rel. These molecules are trapped in the cytoplasm as an inactive complex by IκB. Heterodimers containing p65 and either p50 or p52 are among the most frequently observed in various cell lineages and, when freed from IκB, can initiate gene transcription through transactivating domains. Other combinations, such as p50 homodimers, exhibit negative regulatory effects on many genes. Cell activation through cytokine stimulation, engagement of Toll-like receptors (TLRs), or stress initiates a host-defense signaling pathway that can converge on an enzyme complex containing two IκB kinases, known as IKK-1 and IKK-2 (also called IKKα and IKKβ, respectively), as well as a regulatory protein (IKKγ) that is required for IKK activation (5). Upstream kinases, including members of the MAPKKK family and NF-κB–activating kinase (NAK), can phosphorylate the IKK signalsome and initiate the NF-κB cascade (6). IKKs then phosphorylate IκBα at serine 32 and serine 36 (or other IκB isoforms at homologous amino acid residues), which targets the inhibitor for ubiquitination and degradation by the 26S proteasome.
This process, initiated within minutes of surface receptor ligation, releases NF-κB and leads to its nuclear translocation, followed by initiation of gene transcription. The specific genes that are activated depend on the various NF-κB–binding sequences in promoter regions as well as the components of the NF-κB dimers. IKK-2 is an absolute requirement for this pathway in many circumstances, especially those related to innate immunity, while the absence of IKK-1 in knockout mice typically has little effect on NF-κB translocation. However, IKK-1 is an integral part of signaling through RANK, the receptor activator of NF-κB (7). Defective IKK-1 activity has little effect on innate immune responses but is required for lymphoid development and T cell–dependent antibody responses (8). Other kinases, such as NF-κB–inducing kinase, can be activated by lymphotoxin and signals via an alternative pathway that involves only IKK-1, with subsequent processing of NF-κB1 to produce p52 dimers (9).
The rate of IκB degradation and resynthesis depends on a variety of factors, including the particular isoforms of IκB expressed by each cell lineage. IκBα, which is commonly found in synovial macrophages and fibroblast-like synoviocytes, is rapidly degraded and then resynthesized within 1–2 hours. Resynthesis of IκBβ in other cells, however, is often delayed and leads to prolonged NF-κB translocation (10). Degradation of IκBϵ is relatively slow, although this protein can be quickly resynthesized. The binding affinities of different IκB isoforms for NF-κB components also determine gene regulation. For instance, IκBα and IκBβ bind to p65, but not to α c-Rel; however, IκBϵ can associate with both (11). The p50/p65 heterodimer binds to IκBβ with greater avidity than do p50/RelB or p50/c-Rel dimers. These differences are clearly relevant to approaches designed to enhance IκB expression, such as the Tat-srIκBα construct, when the variability in NF-κB dimer components are observed.
NF-κB activation has been implicated in the pathogenesis of RA. As demonstrated by electromobility shift assays, NF-κB binding is significantly higher in RA synovium compared with osteoarthritis synovium (12, 13). Immunohistochemistry studies identify nuclear p50 and p65 translocation in the synovial lining and mononuclear cells of the sublining. In vitro studies confirm a role of NF-κB in the production of cytokines by macrophages, as well as elevated constitutive production of IL-6 by RA synoviocytes (14). Animal models of RA also support the notion that NF-κB participates in synovitis. Time-course studies in both murine collagen-induced arthritis and rat adjuvant-induced arthritis demonstrate NF-κB activation prior to the appearance of clinical disease (15). Selective activation of IKK-2 by intraarticular gene transfer leads to arthritis in rats, thus confirming that IKK activation is sufficient to initiate synovitis (16).
Several commonly used therapeutic agents, most notably, corticosteroids, can suppress NF-κB, but more selective inhibitors are clearly desirable in order to minimize NF-κB–independent toxicity. Blockade of the IKK signal complex has attracted considerable attention based on its pivotal role in NF-κB signaling and potential inhibition by small molecules (Figure 1). IKK-2 is an especially attractive target for therapy because it regulates cytokine production in many cell types, including cultured synoviocytes. For example, a dominant-negative IKK-2 adenovirus construct almost completely abrogates cytokine-induced IL-6, IL-8, and ICAM-1 expression (17). Intraarticular gene therapy with dominant-negative IKK-2 in rats with adjuvant-induced arthritis decreases NF-κB nuclear translocation and suppresses joint swelling.
Although gene transfer has some advantages when considering local therapy, the development of small-molecule inhibitors of IKK-2 has contributed additional proof that this kinase is a potential target. Published data on 2 selective IKK-2 inhibitors demonstrate remarkable clinical efficacy in rat adjuvant-induced arthritis and murine collagen-induced arthritis (18, 19). In addition to decreased inflammation, bone and cartilage destruction is also significantly lower in the animals treated with the IKK-2 inhibitors. Gene-transfer studies suggest that cytokine production by cultured rheumatoid synovial tissue cells is largely dependent on IKK-2 (20). However, production of some cytokines, such as TNFα, is not blocked by dominant-negative IKK-2, indicating a substantial degree of complexity that was not anticipated. Information on the role of IKK-1 in arthritis is less robust, and this kinase plays little role in cytokine production by fibroblast-like synoviocytes. Endogenous peroxisome proliferator–activated receptor γ agonists, such as 5-deoxy-Δ12,14-prostaglandin J2, can also decrease inflammatory responses through inhibition of IKK and could serve as a normal counterregulatory response (21).
While the early focus in arthritis has been on IKK-2, other signaling pathways that feed into NF-κB also might be targeted. Upstream MAPKs, such as MEKK-1 or NAK, have potential, as do the IKK-related kinases inducible IKK (IKK-i) and TANK-binding kinase 1 (TBK-1) (22). The role of the latter two molecules in inflammation is only partially understood at the present time. Although IKK-i can recognize IκB as a substrate, the process is inefficient, because only 1 of the 2 serines required for rapid degradation is phosphorylated. In contrast, the IKK-related kinases also participate in other pathways relevant to RA, including TLR signal transduction and expression of IL-6 and interferon-β (23). Little or no information is available on the pharmacology of IKK-related kinases in inflammation.
Moving downstream from IKK, methods for stabilizing IκB have been evaluated extensively in cell culture. Typically, genetic constructs that overexpress IκB or an engineered protein that lacks the sites for phosphorylation (IκB super-repressor) have been used. These proteins, although very effective in vitro, present major technical hurdles in vivo because they must enter the cells to be effective. In practice, this usually requires gene transfer using viral or nonviral vectors. Blackwell et al have added a very interesting twist to IκB therapy by using a novel construct that allows the therapeutic protein to traverse cell membranes in order to reach the cytoplasm. In a rat model of pleurisy, they demonstrated that this fusion protein does, indeed, penetrate cells and forms a stable complex with NF-κB that blocks activation. Even more striking, systemic administration inhibits leukocyte migration into the pleural space after injection with a proinflammatory stimulus. The mechanism appears to be related to enhanced apoptosis, probably due to loss of NF-κB–mediated protection from cell death. In addition, cytokine and chemokine production in the pleural cavity are decreased and probably contribute to the antiinflammatory effect.
The authors also noted that local administration of the protein has much less benefit than systemic exposure. One possible explanation is that circulating leukocytes are exposed to the construct after intravenous injection, thereby inducing apoptosis and preventing extrapleural cell activation that is required for cell migration. Alternatively, the primary effect could be on the vascular endothelium, where induction of adhesion molecules, such as VCAM-1, ICAM-1, and E-selectin, requires NF-κB (24). Regardless of the mechanism, the results suggest that in this model of acute inflammation, systemic inhibition is necessary for the full benefit. The observation is consistent with the clinical efficacy of selective IKK-2 inhibitors in arthritis, which may be greater with systemic administration than with intraarticular gene therapy techniques. However, it is important to recognize the significant differences between acute inflammation, which is highly dependent on transient influx of neutrophils, and chronic disease, where the prominent infiltration of the synovium with longer-lived mononuclear cells dominates. Furthermore, the pharmacokinetics of the Tat-srIκBα protein after local injection still needs to be determined, and reformulation could improve efficacy.
The Tat-srIκBα approach is a technical tour de force and is unquestionably valuable as a proof-of-concept experiment, but some questions should be answered to determine the utility of the construct as a therapeutic agent. For example, the dosing and frequency of administration need to be assessed for long-term administration. Although it is clear that the construct enters leukocytes, the distribution in other tissues and the toxicity are not defined. Some concerns about the role of NF-κB inhibition in hepatocyte apoptosis need to be addressed, along with effects on other host responses to pathogens. These questions are not unique to Tat-srIκBα, but should be considered with any long-term therapy that inhibits NF-κB.
In addition to enhancing intracellular levels of IκB by providing exogenous protein, efforts to prevent degradation of endogenous IκB have been explored. For instance, inhibitors of E2 ubiquitin–conjugating enzymes could interfere with IκB ubiquitination. Alternatively, one could target the proteasome to prevent degradation of IκB after ubiquitination. One such compound, MG132, increases intracellular IκB levels and, like Tat-srIκBα and other NF-κB–targeted approaches, induces synovial apoptosis in arthritis (25). One concern with this method is the multitude of proteins that are processed by ubiquitin and the proteasome, and toxicity could be a major hurdle. Interfering with NF-κB binding and transactivating effects represents another post-IκB targeting alternative. For instance, decoy oligonucleotides that contain an NF-κB-binding site can be delivered into cells to prevent gene transcription (26). This approach is effective in rat streptococcal cell wall–induced arthritis, and the beneficial effects again appear to be related to induction of synovial apoptosis. Strategies to decrease the expression of specific NF-κB proteins using antisense or RNA interference could selectively alter the gene expression patterns by changing the composition of NF-κB dimers.
Despite the abundant data indicating a key role of NF-κB in inflammatory diseases, significant safety concerns about inhibition of this transcription factor remain. NF-κB sits at the crossroads of host defense, homeostasis, cell survival, and response to stress, and there are major issues related to systemic inhibition of this primordial protective mechanism. Surely, a price will be paid for blocking the innate immune responses that begin with engagement of TLRs and result in NF-κB–driven gene expression. Some anticipated defects in host defense have already been identified, such as increased susceptibility to infections such as Listeria monocytogenes (27).
Equally important, the induction of apoptosis due to NF-κB inhibition has significant potential for collateral damage to normal or physiologically stressed tissues. The most prominent example is florid hepatic apoptosis, which occurs in IKK-2– or RelA-knockout mice (28). This problem can be abrogated by suppressing TNF responses, as shown in TNF and RelA–double-knockout mice (29). TNF-dependent apoptosis is obviously relevant to diseases marked by increased production of TNFα, such as RA, where the risks of NF-κB blockade might be amplified. A similar double-edged sword is apparent in mice lacking IKK-2 in gastrointestinal enterocytes. Although these animals are protected from systemic inflammatory responses after gut ischemia-reperfusion, the local mucosa is damaged by extensive apoptosis related to the lack of NF-κB. Because IKK-2 plays an essential role in B cell development (30), inhibitors could have a significant impact on antibody production, germinal center formation, and lymphoid tissue maturation. The role of NF-κB proteins in other normal immune responses and development also raises concerns. Abnormal antibody production is observed in p50-knockout mice, and absence of RelB interferes with the development of dendritic cells. In addition to B cell defects, osteoclast development has also been noted in p50 and p52–double-knockout mice (31).
The use of small-molecule inhibitors of various kinases and regulatory proteins in the NF-κB pathway also suffers from potential issues related to the specificity of the compound. It is difficult to design truly selective inhibitors because many bind to ATP-binding sites with significant homology across the universe of kinases. Screening for specificity against all possible targets, as well as other ATP- or substrate-binding sites, is problematic at best. Despite this, significant strides have been made in the chemistry of kinase inhibitors, and at least some specificity appears feasible. These issues are, of course, less critical with biologic agents such as Tat-srIκBα.
How are we to approach NF-κB under such circumstances? One goal could be to modulate, rather than completely ablate, NF-κB function. Global inactivation of NF-κB, such as with proteasome inhibitors, enhancement of cellular IκB levels, or decoy oligonucleotides, could have the greatest potential for toxicity, although some selectivity based on cell-specific IκB isoforms is possible. IKK-2 blockade suffers from many constraints as well, but there are clearly alternative pathways that permit some activation of NF-κB. For example, one selective IKK-2 inhibitor, SC-514, suppresses and delays IκB degradation but does not entirely block it (32). Even if IKK-2 were inhibited, IKK-1– and IKK-related kinase-mediated IκB phosphorylation would remain intact, thereby permitting basal or stress-induced NF-κB activation in some tissues. Other selective kinase-targeted approaches could potentially block pathogenic NF-κB activation and inhibit macrophage cytokine production while permitting basal or normal responses in other cell lineages. The use of biologics such as Tat-srIκBα could achieve site and event specificity based on the method of delivery, the NF-κB heterodimers and homodimers expressed, and the relative ability of the construct to enter different cell lineages.
The jury is still out on the relative balance between the benefits and risks of NF-κB blockade; we do not know if NF-κB is the Holy Grail or a tantalizing and dangerous temptation. The recognition that RA is a serious disease with considerable morbidity and mortality helps pave the way for aggressive therapy, and we should not shy away from interventions that might provide a major positive impact on our patients' lives. One recalls that worries about the dangers of anticytokine therapy contributed to considerable trepidation over a decade ago; it turns out that humans are remarkably resilient and adapt more readily than our murine brethren. Proof-of-concept in RA awaits while the multitude of NF-κB–directed approaches move inexorably toward the clinic.