Inflammatory bowel diseases (IBDs) are multifactorial conditions where multiple environmental and genetic factors come together to promote an exaggerated immune response in the gut wall.1 In Crohn's disease (CD), this immune response appears to be directed against components of the luminal bacterial flora, but in ulcerative colitis (UC) a role for the flora is much less compelling.2 Analysis of immunoinflammatory pathways in the gut of patients with CD or UC has shown that tissue damage is driven by a complex and dynamic crosstalk between immune and nonimmune cells, and that cytokines are key mediators of this interplay.3, 4 It has also been demonstrated that some cytokine-mediated counterregulatory antiinflammatory pathways are defective in IBD, raising the possibility that restoring these antiinflammatory signals may be a therapeutic strategy.5 These advances, together with the many animal models of colitis, have facilitated the development of compounds and biological agents that neutralize cytokines.6 Some of these drugs have already been tested in IBD patients, with others ready to move into the clinic.6, 7 However, given the plethora of immunological manipulations that can prevent colitis in animal models, the recent failures of anti-interferon-gamma (IFN-γ) and anti-interleukin (IL)-17 antibodies in the clinic, and the fact that only 50% of CD patients respond to anti-tumor necrosis factor alpha (TNF-α) therapy, it now becomes extremely difficult to decide which other cytokines should be targeted, and if any will ever beat anti-TNF-α.7 In this article we briefly review the data emerging from studies of cytokine inhibitors in IBD and discuss which cytokine pathway should be targeted in the future.
In the gut of patients with inflammatory bowel disease (IBD), immune and nonimmune cells produce large amounts of cytokines that drive the inflammatory process leading to the tissue damage. Cytokine blockers, such as anti-tumor necrosis factor alpha (TNF-α), have been used with some success in IBD. However, not all patients respond, and the therapeutic effects wane with time, demonstrating the need for more effective and long-lasting antiinflammatory strategies. A key question is whether neutralizing other proinflammatory cytokines such as interleukin (IL)-12, IL-21, IL-27, or IL-33 will lead to a better clinical response than with anti-TNF-α antibodies. Equally, we now know that IBD-related inflammation is marked by defective production/activity of antiinflammatory cytokines, and there are strategies to correct these defects. An alternative approach is to try to target individual therapies to individual patients, to improve clinical efficacy in subsets of patients, but this has proven difficult. Here we try to evaluate the potential of each of these choices. (Inflamm Bowel Dis 2012;)
Why Do We Need More than TNF-α Blockers?
The prototypic example of the ability of an anticytokine agent to effectively change the therapeutic landscape in IBD was provided by the chimeric anti-TNF-α antibody infliximab.6–8 Both inflammatory and fistulizing CD can be treated with infliximab.9–11 Episodic treatment maintains remission, and also reduces the short-term colectomy rate in UC.12–14 Other anti-TNF-α monoclonal antibodies have entered the market (Table 1), namely, adalimumab (a fully humanized monoclonal antibody) and certolizumab (a pegylated humanized Fab2 fragment).15–17 Both these antibodies are effective in inducing and maintaining clinical response in CD patients. Adalimumab also appears to be effective in some patients who do not respond or become unresponsive to infliximab.18 Etanercept, a TNFR p75-IgG fusion protein, is ineffective in CD.19
|Biologic Agent||Target Molecule||Disease||Response Rate vs. Placebo; P value (Week)||Remission Rate vs. Placebo; P value (Week)||Reference|
|Infliximab||TNF-α||CD||65% vs. 17%; P < 0.001 (week 4)||33% vs. 4%; P = 0.005 (week 4)||9, 14, 16, 99, 100, 19, 29, 31, 101, 30, 32, 103, 102|
|UC||69% vs. 37%; P < 0.001 (week 8)||39% vs. 15%; P < 0.001 (week 8)||15|
|Adalimumab||TNF-α||CD||50% vs. 25%; P < 0.05 (week 4)||36% vs. 12%; P = 0.004 (week 4)||16|
|UC||50% vs. 35%; P < 0.001 (week 8)||16% vs. 9%; P = 0.02 (week 8)||99|
|Certolizumab pegol||TNF-α||CD||35% vs. 27%; P = 0.02 (week 6)||22% vs. 17%; P = ns (week 6)||100|
|Etanercept||TNF-α||CD||39% vs. 45%; P = ns (week 4)||9% vs. 20%; P = ns (week 4)||19|
|ABT-874||IL-12, IL-23 (p40 subunit)||CD||75% vs. 25%; P = 0.03 (week 7)||38% vs. 0%; P = ns (week 7)||29|
|Ustekinumab||IL-12, IL-23 (p40 subunit)||CD||49% vs. 40%; P = ns (week 8)||26% vs. 17%; P = ns (week 8)||19|
|Fontolizumab||IFN-γ||CD||44% vs. 33%; P = ns (week 4)||19% vs. 12%; P = ns (week 4)||101|
|73% vs. 60%; P = ns (week 4)||47% vs. 40%; P = ns (week 4)||30|
|38% vs. 31%; P = ns (week 4)||18% vs. 5%; P = ns (week 4)||32|
|Secukinumab||IL-17A||CD||18% vs. 30%; P = ns (week 6)||10% vs. 15%; P = ns (week 6)||Hueber W et al. JCC 2011|
|Tocilizumab||IL-6R||CD||80% vs. 31%; P = 0.02 (week 12)||20% vs. 0%; P = ns (week 12)||102|
Unfortunately, almost half of IBD patients do not respond well to anti-TNF-α antibodies and the response wanes with time.6–8 Anti-TNF treatment optimization with azathioprine can be considered and is effective in maintaining long-term remission in over 50% of patients; however, this strategy must be weighed against the increased risk of serious infections and perhaps lymphoma.20, 21 Moreover, infliximab therapy is also associated with the development or relapse of immune-mediated diseases (e.g., drug induced-lupus, psoriasis).22–24 Taken together, these results underscore the urgent need to develop new biologics, particularly for patients who do not respond, lose responsiveness, or cannot receive TNF-α blockers.
Disappointing Results of Anti-IFN-γ and IL-17A Blockers in CD
The profile of cytokines made by T cells in CD is somewhat different from that seen in UC.1 CD bears the stigmata of a predominant Th1 cell-associated disease, as indicated by the elevated levels of IFN-γ and IL-12, the major Th1-inducing factors in man, and overexpression of Th1-related transcription factors (i.e., Stat4 and T-bet), while in UC it has been suggested that there is high production of Th2-related cytokines, such as IL-5 and IL-13.25–28 The results of clinical studies testing the efficacy of the anti-IFN-γ antibody fontolizumab and two different anti-IL-12/p40 antibodies in patients with active CD, however, were quite disappointing (Table 1), as these antibodies were only slightly superior to placebo in inducing clinical remission.29–33 However, anti-IL-12p40 does seem to be effective in patients who fail anti-TNF-α therapy.31 The fact that anti-IL-12/p40 antibody inhibited production of IFN-γ by intestinal lamina propria mononuclear cells, and fontolizumab reduced intestinal expression of active Stat1, a transcription factor activated by IFN-γ, confirmed the effective neutralization of IL-12 and IFN-γ, suggesting the existence of a non-IL-12-driven IFN-γ-independent pathway of tissue damage in CD.26, 30 Th17 cells were therefore a likely candidate. These are a subset of Th cells with the potential to secrete IL-17A, IL-17F, IL-21, IL-22, and IL-26.34 Th17-cytokines are produced in excess not only in CD but also in UC, even though the majority of Th17-secreting cells are not typical Th17 cells, because they coexpress IFN-γ and FoxP3, the transcription factor needed for the suppressive activity of regulatory T cells (Tregs).34, 35 Moreover, Th17 cells exhibit plasticity and can be converted into Th1-type cells if stimulated with IL-12 or IL-23.36–39 Finally, it is noteworthy that studies in murine models of colitis have shown that IL-17A, the signature cytokine of Th17 responses, has both proinflammatory and tissue-protective properties in the gut, depending on the model (Table 2).40–43 So it is not surprising that blockade of IL-17A with a neutralizing antibody gave no clinical benefit in patients with active IBD. However, given the plethora of cytokines produced in the mucosa in IBD (Fig. 1), some of which may be protective and some of which may be pathogenic, the question must be asked whether it is worth pursuing any others in the hope that when given to all CD patients, it will be more effective than anti-TNF antibodies.
|Genetic Background||Experimental Model||Role of IL-17A||Observation||Reference|
|Rag-KO mice||IL-10 KO naïve T cell transfer||Pathogenic||IL-23-dependent colitis is attenuated by neutralization of IL-6 and IL-17A||41, 43, 40, 42, 44|
|Rag1-null mice||ROR-γ-nullT cell transfer||Pathogenic||Colitis develops only upon IL-17A administration||43|
|Wildtype mice||DSS-induced colitis||Protective||Neutralization of IL-17A aggravates colitis||40|
|Rag1-KO mice||Naïve T cell transfer||Protective||IL-17A deficient T cells induce a more severe colitis than wild-type T cells||42|
|IL-17A directly inhibits Th1 cells|
|Wildtype/IL-17R KO mice||TNBS-induced colitis||? (IL-17R is common to IL-17A and IL-17F)||IL-17R KO mice are protected from colitis||44|
Role of IL-17F and IL-21 in the Control of Gut Inflammation
Studies conducted in mice deficient in IL-17 receptor A, which mediates the functional activities of both IL-17A and IL-17F, have convincingly shown that signals driven by this receptor are essential for driving trinitrobenzenesulfonic acid (TNBS) colitis.44 IL-17A is protective against dextran sodium sulfate (DSS) colitis and in the transfer model of colitis, where T cells are injected into lymphopoenic mice.40 These data suggest that IL-17F, and not IL-17A, is pathogenic in the gut. This is supported by the demonstration that mice deficient in IL-17F are resistant to DSS colitis.45 These data, however, do not exclude the possibility that, under specific circumstances, IL-17A and IL-17F may cooperate in promoting mucosal inflammation. For example, transfer of IL-17A-, IL-17F-, or IL-22-deficient T lymphocytes into RAG1-null mice induces severe colitis indistinguishable from that caused by wildtype cells.43 In contrast, transfer of retinoid-related orphan receptor gamma (RORγ)-null T cells, unable to make IL-17 family cytokines, does not induce colitis.43 Treatment of RAG1 mice that receive IL-17F-null T cells with a neutralizing anti-IL-17A antibody suppresses disease, thus suggesting that simultaneous inhibition of more cytokines is required to suppress Th17-driven pathogenic responses.43
Another Th17-related cytokine that seems to play a pathogenic role in the gut is IL-21.46 IL-21 is produced in excess in the intestine of patients with CD and patients with UC.47, 48 We have previously shown that when mucosal T cells from CD patients are activated in vitro with anti-CD3 in the presence of either a neutralizing anti-IL-21 antibody or an IL-21R-IgG fusion protein, the production of both IL-17A and IFN-γ is reduced.47 These results together with the demonstration that IL-21-deficient mice are resistant against Th1/Th17 cell-driven colitis support the key role of IL-21 in positively regulating Th1 and Th17 cell-associated inflammatory pathways (Fig. 2).49
IL-21 exerts further biological functions that could contribute to its proinflammatory effect in the gut. For example, IL-21 inhibits the peripheral differentiation of Tregs and makes CD4+ T cells resistant to Tregs-mediated immune suppression.50, 51 IL-21 stimulates stromal cells to produce tissue-degrading proteases and enhances secretion of the T-cell chemoattractant macrophage inflammatory protein-3α by intestinal epithelial cells.52, 53 IL-21 also potentates the expression of Th1-related transcription factors and IFN-γ in T and NK cells and the cytolitic activity of NK cells (Fig. 2).54, 55 Taken together, therefore, the multiple pathways by which IL-21 can damage the gut suggest that it may be worthwhile neutralizing in a clinical trial.
IL-27 and IL-35: Newer Members of the IL-12 Family
The IL-12 family is made up of secreted heterodimers with some overlapping usage between family members (Table 3).56 IL-12 (p35/p40) and IL-23 (p19/p40) are the best-known members of the IL-12 family.57 However, other members include IL-27 (EBi3/p28) and IL-35 (EBi3/p35).56, 57 As stated above, neutralizing antibodies to p40 that inhibit IL-12 and IL-23 have already gone into the clinic, so the question is whether an anti-EBi3 antibody which would neutralize both IL-27 and IL-35 is worth pursuing. Like the IL-17 family, however, a critical issue is whether these molecules are pathogenic or protective in the gut.
|Cytokine||Subunits of the Heterodimer||Neutralizing Antibody|
IL-27, produced mostly by myeloid cells, is present at increased concentrations in IBD mucosa. If T cells are taken from mice deficient in the IL-27 receptor (IL-27R) and injected into lymphopoenic mice, they induce significantly less colitis than wildtype T cells, clearly suggesting that IL-27 is pathogenic in this model.58 IL-27R-null mice are also less susceptible to DSS colitis, again suggesting IL-27 is proinflammatory.59 In marked contrast, however, treating mice with established TNBS colitis with IL-27 reduces disease and cytokine production.60 In man, IL-27 reduces proinflammatory cytokine production in macrophages activated with TNF-α.61 Overall, the situation is extremely confusing and there would be significant risk in trying to treat IBD with compounds neutralizing IL-27-driven signals.
An antibody against EBi3 would neutralize both IL-27 and IL-35, which could be problematic if one was protective and the other proinflammatory. Recently, an attempt was made to dissect out the roles of IL-27 and IL-35 in spontaneous colitis (STAT3 deletion in myeloid cells) and the transfer model of colitis.62 EBi3 null mice (no IL-27 or IL-35) developed more severe colitis than IL-27p28 null mice (no IL-27 but IL-35 present), suggesting that IL-35 was protective.62 Recombinant IL-35 also protected against colitis.62 This suggests that IL-27 is pathogenic, but IL-35 is protective, and that IL-27 could be neutralized with an anti-p35 antibody, which of course would also neutralize IL-12 (Table 3).
IL-22: A Cytokine Made by Innate Lymphoid Cells and T Cells
Although essentially all interest in gut inflammation has concentrated on T cells, in the last few years it has been discovered that there is a heterogeneous population of innate lymphoid cells (ILC), particularly present in the gut, which play a role in protection from pathogens and pathology.63
The prototypic ILC is the natural killer (NK) cell, which kills tumor cells, and also kills cells expressing pathogen-associated and stress-associated ligands. NK cells are encoded to secrete cytokines when activated, typically IFN-γ, TNF-α, and granulocyte macrophage colony-stimulating factor (GM-CSF).64 Despite much investigation, however, a clear role for classical NK cells in mucosal health and disease has yet to be identified.
Lymphoid tissue inducer cells (LTi) are a second type of ILC.65 These cells were originally described as being critical for lymphoid tissue organogenesis by activating stromal cells to express adhesion molecules and make the chemokines needed to attract dendritic cells, T and B cells into developing lymph nodes and Peyer's patches.66, 67 LTi depend on RORγt, and, like Th17 cells, LTi also produce IL-17A and IL-22.68, 69
However, matters are complicated by the fact that a third type of ILC has also been identified, which may be an intermediate between NK cells and LTi cells.70 In mice they express NKp46, but in man they express NKp44 and are CD56+.70 These cells produce very little IFN-γ, and are not cytotoxic, but contain abundant IL-22 and have been termed ILC22 cells; they have also been implicated in resistance to Citrobacter infection.71 Other ILCs have been identified in the gut, termed nuocytes, which make Th2-type cytokines.72 A Thy1+ ILC, dependent on IL-23, and which makes IL-17A and IFN-γ appears to be important in the colitis associated with Helicobacter hepaticus infection in lymphopenic mice,73 and CD3− cells making IL-17 has been identified in CD tissue.74 Unraveling the complexity of ILC will be a major challenge in the next few years.
Whether IL-22 is protective or pathogenic in the gut is controversial. In 2005, it was found that IL-22 induced a strong proinflammatory response in human colonic subepithelial fibroblasts.75 Likewise, overexpressing IL-22 in mice induces a strong acute phase response, characteristic of inflammation.76 The IL-22 receptor is expressed on epithelial cells in the gut and IL-22 increases defensin production, proinflammatory cytokines, and increased wound healing, albeit in epithelial cell lines.77, 78 Microinjection of IL-22 into inflamed colon of TCR-α null mice alleviated inflammation, suggesting it is antiinflammatory.79
The role of IL-22 has been examined in mouse infectious disease models. IL-22 null mice rapidly succumb following Citrobacter rodentium infection80; however, RAG2 null mice treated with anti-IL-22 also rapidly succumbed, and the source of innate IL-22 was considered to be dendritic cells.80 Similar results have also been reported when RAG null mice lacking NKp46+ ILC are given Citrobacter, in that mice rapidly succumb to infection.71 In contrast, in the Toxoplasma gondii model of ileitis, IL-22 appears to be pathogenic, since IL-22 null mice show little sign of disease.81
In IBD models (DSS colitis and the T-cell transfer model into lymphopenic recipients), IL-22 appears to be protective.82 However, in the transfer model, whether IL-22 is protective or inflammatory appears to depend on the T cells transferred. After transfer of CD45Rbhi cells, IL-22 is protective, but after transfer of Treg-depleted CD45Rblo cells, IL-22 was pathogenic.83 In mice with a normal immune system, we have recently shown that injection of an agonist to the aryl hydrocarbon receptor protects against TNBS colitis, DSS colitis, and transfer colitis. Protection is associated with increased IL-22, and anti-IL-22 blocks the protective effect.84
Counterregulatory Cytokines as a Strategy to Block Mucosal Inflammation
The detrimental mucosal response occurring in IBD patients is associated with a defective activity of counterregulatory mechanisms. Thus, we can speculate that restoring the balance between inflammatory and antiinflammatory molecules can be useful to dampen the IBD-associated gut inflammation. A first approach to reach this therapeutic goal is to use IL-10, a homodimeric cytokine that limits inflammatory responses in various organs and prevents immune-mediated pathologies.85 Germ-line mutations in IL-10R genes have been documented in some patients with early onset IBD, raising the possibility that defects in IL-10 activity can contribute to sustain the IBD-related pathogenic response.86 These observations collectively have supported the therapeutic use of IL-10 in patients with IBD. However, subcutaneous administration of recombinant human IL-10 to patients with active and steroid-dependent CD was not effective in inducing clinical response/remission, possibly owing to the possibility that the cytokine does not reach therapeutic concentration in the gut following systemic administration.87 To overcome this issue, IL-10-encoding probiotics have been developed and already tested in a Phase 1 study.88
Another cytokine involved in the negative regulation of gut inflammation is transforming growth factor (TGF)-β1.89 In the normal gut TGF-β1 is produced by immune and nonimmune cells and suppresses pathogenic T cell and antigen-presenting cell responses by activating an intracellular signaling pathway, which involves the Smad molecules.90 In both CD and UC, TGF-β1 is produced at a high level, but paradoxically it is not sufficient to stop the mucosal inflammation.91 We have shown that TGF-β1 activity is defective in IBD tissue due to the abundance of Smad7, an inhibitory Smad, that binds the TGF-βR and blocks TGF-β1-driven intracellular signals.92 Consistently, inhibiting Smad7 with a specific antisense oligonucleotide reduces Smad7 expression, restores TGF-β1 signaling and suppresses inflammatory cytokine production (Fig. 3).92 These novel observations led to the development of a Smad7 antisense oligonucleotide-containing pharmaceutical product, which is now being tested in CD patients.
Another counterregulatory molecule is thymic stromal lymphopoetin (TSLP), which is made by epithelial cells and is supposed to play a key role in mucosal healing after insult.93 Indeed, it has been recently shown that TSLP-deficient mice fail to recover from DSS-induced colitis and eventually die.94 TSLP-null mice exhibit increased neutrophil elastase (NE) activity during colitis, which is paralleled by reduced expression of an endogenous inhibitor, the so-called secretory leukocyte peptidase inhibitor (SLPI).94 Pharmacological inhibition of NE or treatment with recombinant SLPI reduces DSS-induced mortality in TSLP-null mice.94 A place in the therapeutic armamentarium of IBD could be perhaps occupied by IL-25, a cytokine made by many cell types, including epithelial cells and macrophages. IL-25 expression is downregulated in the inflamed areas of patients with CD and patients with UC,95 probably as a result of both the negative control exerted by inflammatory cytokines (i.e., TNF-α) and the lack of TGF-β1-mediated positive regulation on IL-25-cell producers.96 Triggering IL-25R on intestinal monocytes/macrophages with recombinant IL-25 results in a marked inhibition of proinflammatory cytokines.95 Consistently, in vivo in mice, administration of IL-25 is both preventive and therapeutic in models of colitis, and this antiinflammatory effect of IL-25 is associated with expansion of alternatively activated macrophages, a subset of regulatory cells.97
In recent years, progress in basic and translational research has led to a better understanding of the role of cytokines in the pathogenesis of IBD. It is now believed that an altered balance between regulatory and inflammatory cytokines can contribute to perpetuate the mucosal inflammation in both CD and UC.1, 2 Since there is evidence that the tissue-damaging immune response is driven by multiple cytokine-driven inflammatory pathways, it is logical to hypothesize that simultaneously targeting two or more of these signals could be more advantageous than inhibiting selectively single pathways. Various approaches for inhibiting inflammatory cytokines have been developed and are now ready to move into the clinic.7 However, in designing clinical interventions around these molecules, it would be useful to take into consideration that inhibition of inflammatory cytokines could be associated with severe side effects, as these molecules are also involved in the control of physiological processes and immune responses against infections and neoplasias.98 Another promising therapeutic strategy would be to restore counterregulatory mechanisms, which are defective in IBD. Since it is conceivable that no drug can work in all patients, further experimentation will be necessary to identify biomarkers that predict responsiveness to the cytokine-based therapy, as well as to ascertain which biological therapy will be effective in an individual patient.