Crohn’s disease: Th1, Th17, or both?
Although Crohn’s disease and mouse models that share features with CD were originally attributed to Th1-mediated pathogenesis, the recent emergence of the Th17 lineage has precipitated a reassessment of its immune-based etiology. The discovery of IL-23 and its subsequent link to induction of IL-17 cytokines and the Th17 pathway resulted in a re-examination of the relative roles of the Th1-asssociated IL-12–IFNγ inflammatory axis and the new, Th17-associated IL-23–IL-17 axis in multiple autoimmune and chronic inflammatory diseases, including IBD. Because IL-12 and IL-23 are heterodimeric cytokines that share the IL-12p40 subunit, and the majority of studies pre-dating IL-23 and Th17 that examined the role of IL-12 in IBD pathogenesis targeted the common IL-12p40 subunit, it was unclear whether the observed disease-ameliorating effects were due to IL-12, IL-23, or both. This, coupled with increases of both Th1 and Th17 cells in inflamed colons of both mice and humans, suggested that Th17 cells might also be key players in IBD.
In both the CD45RBhi T cell transfer and IL-10–deficient models of colitis, early studies showed that development of colitis in both models was almost completely abrogated by early treatment with IFNγ-blocking antibody, consistent with a Th1-mediated pathogenesis (Berg et al., 1996
; Powrie et al., 1994
). In one report, disease did not develop in the transfer model if CD4+
T cells were isolated from IFNγ-deficient donors (Ito and Fathman, 1997
). In support of a Th1 pathogenesis, transfer of T cells deficient for the Th1-associated transcription factor, T-bet, failed to induce disease, whereas adoptive transfer of T-bet–transduced T cells resulted in colitis in recipient mice (Neurath et al., 2002
). Further, T cells deficient in the Th1-associated differentiation factor, STAT4, produced substantially reduced disease following transfer (Simpson et al., 1998
). In conflict with a strictly Th1 pathogenesis, however, the transfer of IFNγ-deficient CD4+
CD45RBhi T cells did induce colitis in one study (Simpson et al., 1998
), suggesting the involvement of either an IFNγ-independent Th1-type program or a distinct effector pathway. Consistent with the latter observation, H. hepaticus
-induced colitis developed in mice doubly-deficient for IL-10 and IFNγ comparable to that of IL-10-deficient mice (Kullberg et al., 2001
), demonstrating that IFNγ was dispensable for disease onset.
As in mouse IBD models, CD was initially postulated to result from a dysregulated Th1 response, based on elevated frequencies of IFNγ- and T-bet-expressing CD4+
T cells in diseased intestinal tissues (Fuss et al., 1996
; Matsuoka et al., 2004
; Neurath et al., 2002
). Increased serum IFNγ was also detected in CD patients, but not in UC patients or normal controls (Beltran et al., 2009
). Moreover, expression of both chains of the IL-12 receptor (IL12Rβ1 and IL-12Rβ) and the IL-18R, features of mature Th1 cells, were up-regulated on lamina propria CD4+
T cells of CD patients and enabled enhanced production of IFNγ when the cells were activated ex vivo
with IL-12 and IL-18 (Okazawa et al., 2002
). Finally, although antibody blockade of IFNγ has shown limited effectiveness in IBD patients, consistent with its limited effectiveness in established disease in mice, treatment with anti-human ‘IL-12’ (in fact, an IL-12p40 mAb that neutralizes both IL-12 and IL-23) has shown substantial efficacy in active CD (Mannon et al., 2004
). Although originally offered as evidence in support of a Th1 etiology of CD, recent mouse data demonstrating greater treatment efficacy by interruption of IL-23 rather than IL-12 signaling (see below), has weakened a strict Th1 link in CD.
Although findings implicating factors associated with the Th1 effector pathway in models of CD cannot be ignored, studies that have come in the wake of the discovery of Th17 have increasingly implicated a dominant role for Th17 cells in disease pathogenesis. In addition to the IFNγ-producing cells described in early studies, a sizeable fraction of CD4+
T cells recovered from mucosal compartments in diseased mice of CD45RBhi T cell transfer and IL-10–deficient models are a distinct subset of IL-17-producing CD4+
T cells, as well as some ‘double-producers’ that express both IL-17 and IFNγ (Izcue et al., 2008
; Yen et al., 2006
). Importantly, adoptive transfers of intestinal bacteria-reactive Th17 cells into immunodeficient recipients induced more severe colitis than comparable transfers of Th1 cells, and induced disease at far lower cell doses (Elson et al., 2007
), analogous to earlier findings in a model of EAE (Langrish et al., 2005
). Treatment with an IL-23p19 monoclonal antibody inhibited colitis development when administered at the time of transfer, but also suppressed ongoing disease and resulted in the depletion of the transferred Th17 effectors, indicating that IL-23 is required to sustain the pathogenic Th17 population (Elson et al., 2007
). Further, deletion of the gene encoding IL-23p19, but importantly, not IL-12p35, inhibited spontaneous colitis in IL-10-deficient mice, proving that IL-23, and not IL-12, is necessary for the spontaneous colitis that develops in this model (Yen et al., 2006
). Moreover, treatment with IL-23 accelerated disease onset in RAG-deficient recipients of memory T cells harvested from IL-10-deficient mice. Collectively, these data strongly implicate IL-23-dependent Th17 effector mechanisms in the development of intestinal inflammation.
Other animal studies have recently examined possible contributions of additional components of the Th17 pathway in IBD pathogenesis. Antagonism of IL-6 receptor signaling, or transfer of cells deficient for interferon regulatory factor (IRF)-4 — a transcription factor implicated in Th17 differentiation — blocked induction of colitis in a T cell transfer model. Interestingly, colitis that was indistinguishable from that induced by wild type T cells developed in RAG-deficient recipients of naïve T cells isolated from Il17a−/−
, or Il22−/−
mice (Izcue et al., 2008
; Leppkes et al., 2009
). However, transfer of Il17f−/−
T cells with concomitant neutralization of IL-17A resulted in reduced disease severity, as did transfer of RORγt-deficient naïve T cells, suggesting that RORγt-directed IL-17A and IL-17F redundantly drive inflammation in this model (Leppkes et al., 2009
). This is at apparent odds with earlier experiments showing that T-bet deficiency, IFNγ deficiency, or blockade of IFNγ is sufficient to prevent colitis in this model (Berg et al., 1996
; Neurath et al., 2002
; Powrie et al., 1994
), and with a recent report demonstrating enhanced kinetics of wasting disease in recipients of Il17a−/−
T cells compared to recipients of wild-type T cells (O’Connor et al., 2009
). This latter study suggests a regulatory role for IL-17A early in disease development and the authors propose that this is achieved via suppression of IFNγ production from Th1 effectors, implicating Th1 cells as key culprits in disease onset, consistent with earlier reports. However, none of the histological parameters of experimental colitis were different between recipients of wild type and Il17a−/−
T cells and thus any IL-17-mediated protection appears to be readily overcome by the early Th1 burst.
Pre-dating discovery of the Th17 lineage, elevated expression of IL-17A by CD3+
lymphocytes and CD68+
macrophages or monocytes was found in intestinal tissues of CD and UC patients compared to normal patients, or patients with infectious or ischaemic colitis (Fujino et al., 2003
). Although IL-17A+
lymphocytes were readily detectable in both CD and UC lesions, the frequencies were higher in CD, and correlated with disease activity and serum IL-17A. A different picture emerged when intestinal transcripts of Il17f
, the other IL-17 family member produced by Th17 cells, were measured. Although Il17f
mRNA was elevated in inflamed versus non-inflamed biopsies from the same CD patient, it was substantially higher in inflamed lesions from UC patients in comparison to CD patients (Seiderer et al., 2008
). However, further work will be necessary to examine the contribution of IL-17F in human IBD.
Whether Th1 cells, Th17 cells, or both are required for experimental colitis continues to be contentious as there is experimental evidence supporting the pathogenicity of each population, although this varies between models and in some instances, between laboratories studying the same models. Nevertheless, it is clear that cells competent for expression of both IFNγ and IL-17 are detectable at all stages of disease in T cell-dependent mouse models, as well as in human CD. The fact that early, but not late blockade of IFNγ prevents disease suggests that IFNγ might be more critical for disease onset than for its persistence. Alternatively, it is possible that other T-bet– and STAT4–dependent aspects of the Th1 pathway can compensate for critical functions of IFNγ. T-bet is believed to promote the pathogenic potential of both Th1 and Th17 in other models of autoimmunity, such as EAE (Yang et al., 2009
), and might function in a similar manner in IBD. Finally, recent studies by our group and others have demonstrated developmental plasticity in the Th17 pathway that enables Th17 cells to divert to IFNγ production in the context of colitis (see below) (Annunziato et al., 2007
; Lee et al., 2009b
; Lexberg et al., 2008
), offering the possibility of a pathogenetic mechanism that involves IL-17– and IFNγ–producing T cells that arise from a common developmental pathway.
Ulcerative colitis: a role for Th2 and Th2-like cells?
In contrast to CD, the absence of elevated IFNγ in UC patients fueled early speculation that perhaps the other T effector lineage then known (Th2) might be involved in disease pathogenesis (Niessner and Volk, 1995
). However, direct data in support of a Th2 etiology were lacking. In fact, IL-4-expressing T cells were reduced in IBD patients compared to healthy controls, although IL-5 was shown to be increased in UC patients compared to CD patients and normal controls (Fuss et al., 1996
). It was later found that whereas there were no major differences in Il4
mRNA expression between UC patients with inactive disease and controls, both transcripts were elevated in active UC lamina propria cells. Interestingly, in studies that corroborated the elevated IL-5 and IL-13 secretion by lamina propria cells from UC patients, it was also shown that the major source of the increased IL-13 was a population of CD1d-restricted NKT cells (Fuss et al., 2004
Despite the on-going association of UC with Th2-type responses — or at least Th2 cytokines, mechanisms involved in the pathogenesis of UC have been limited by the paucity of suitable animal models. The oxazolone challenge model was originally characterized as an acute, Th2-dependent model that displayed certain hallmarks of UC, with the caveat that, unlike UC, colitis in this model resolves spontaneously and does not recur (Boirivant et al., 1998
). The early inflammatory infiltrate in this model is characterized by IL-4-, IL-5-, and IL-13-secreting CD4+
T cells, and importantly, systemic administration of anti-IL-4 suppresses disease (Boirivant et al., 1998
). Confirming the importance of effector T cell differentiation in this model, mice deficient for IRF-4 develop minimal disease (Mudter et al., 2008
), although a caveat here is the recent association of IRF-4 with Th17 development in addition to Th2 development (Brustle et al., 2007
). In this regard, disease development is also impeded in the absence of IL-6, a known inducer of Th2 and Th17 cell differentiation (Weigmann et al., 2008
In subsequent studies of a more chronic variant of the oxazalone model, it was found that an IL-4-dependent early phase of intestinal inflammation was superseded by an IL-13-dominated late phase. The major source of IL-13 was identified as a ‘non-classical’ NKT cell that is restricted by the MHC molecule, CD1d, but unresponsive to the invariant NKT ligand, α-galactosylceramide (Fuss et al., 2004
). Accordingly, depletion of NK and NKT cells prior to oxazolone treatment, as well as inhibition of CD1d antigen presentation to NKT cells, prevented the development of disease (Heller et al., 2002
). It is still unknown how IL-13 is induced in this model, except that CD1d-restricted antigen presentation is required. It is also unknown how early induction of IL-4 is linked to disease development and progression.
Although not readily detectable at steady state, Th2 cell responses are rapidly induced in the intestines of mice infected with parasitic worms. Most models involve immune-mediated worm expulsion and resolution of inflammation in contrast to the chronic relapsing and remitting inflammation associated with UC. However, these studies have identified key mechanisms that contribute to the initiation and resolution of type 2 inflammation in intestinal tissues, and although it is still unknown whether any of these mechanisms contribute to the Th2-type responses associated with UC, mechanistic insights from these studies could prove informative.
Worm infections result in activation of intestinal epithelial cells (IEC) to produce thymic stromal lymphopoietin (TSLP), which promotes parasite-specific Th2 responses in mucosal tissues (Zaph et al., 2007
). TSLP-activated DCs upregulate the Th2-attracting chemokines thymus and activation-regulated chemokine (TARC or CCL17) and MDC macrophage-derived chemokine (MDC or CCL22) and induce Th2 differentiation (Soumelis et al., 2002
; Zhou et al., 2005
), suggesting that TSLP directly drives Th2 responses. However, Th2 cells still develop in the absence of TSLP signaling, especially if IFNγ is neutralized (Ramalingam et al., 2009
; Taylor et al., 2009
). Thus, elevated expression of Th1 and Th17 associated cytokines in parasite-infected Tslpr−/−
mice, and the resultant defect in Th2 responses, suggest that TSLP might be more critical for suppression of IL-12- and IL-23-dependent responses in the intestine, thereby allowing for the initiation of type 2 immunity (Taylor et al., 2009
; Zaph et al., 2007
IL-25 is constitutively expressed in the intestines, contingent on the presence of a commensal flora, and is therefore more highly expressed in the large intestines than any other tissue (Zaph et al., 2008
). IL-25 treatment induces heightened expression of the Th2 effector cytokines IL-5, and IL-13 (Fort et al., 2001
), both prominent in human UC, and IL-25 was recently shown to inhibit IL-23-dependent responses in the colon (Zaph et al., 2008
). Thus, IL-25 and TSLP appear to perform redundant and non-redundant roles in type 2 immune responses to parasite infection. In this regard, it was recently reported that TSLP induced by intestinal helminth infection promotes the rapid recruitment of basophils, which act both as the primary APC and source of IL-4 to initiate Th2 responses (Perrigoue et al., 2009
), establishing a unique role for basophils in type 2 intestinal inflammation. Although it is unclear at present whether IL-25 and TSLP contribute to oxazalone-induced colitis, have links to UC pathogenesis, or promote development and function of non-classical NKT cells, or whether basophils are central to immune pathogenesis, these studies suggest novel mechanisms to be explored in the context of Th2-type intestinal inflammation and IBD.
Immunological ‘flip-floppers’: lineage plasticity of effector and regulatory T cells and possible implications for IBD
Several recent studies have unveiled a previously unappreciated feature of CD4+
T cell biology that promises to impact IBD immune pathogenesis — lineage plasticity among effector and regulatory T cell lineages (Lee et al., 2009a
; Zhou et al., 2009
T cells expressing both Foxp3 and RORγt have been detected in the normal intestines of both mice and humans, residing alongside subsets that express one or the other of these factors (Miyara et al., 2009
; Yang et al., 2008a
; Zhou et al., 2008
). It is currently unknown whether these ‘dual-expressors’ represent the earliest precursors of Th17 or Treg cells, each of which have been shown in vitro to express both RORγt and Foxp3 prior to extinction of the opposing factor (Zhou et al., 2008
), whether these cells represent fully differentiated Foxp3+
Treg cells transitioning to Th17 cells in response pro-inflammatory cytokines, or both. However, this finding raises the intriguing possibility that CD4+
T cells that differentiate in or localize to the intestines retain developmental and functional flexibility to facilitate adjustment to this dynamic environment. The transition of Tregs to Th17 progeny might not be unique, as a recent study provides evidence that Tregs might also acquire Th1 effector features in response to Th1 polarizing cytokine conditions (Wei et al., 2009
). It is currently unclear whether similar transitions might occur in the inflamed intestines, and if so, the implications for regulatory control in IBD.
If the developmental plasticity of Tregs in the intestines remains largely speculative, that of Th17 cells is not. Th17 cells defined by an IL-17F reporter transgene could diverge into distinct progeny contingent upon the balance of TGF-β, IL-23 and IL-12 (Lee et al., 2009b
). Importantly, the transition of a subset of Th17 precursors to Th1-type cells was associated with the development of colitis following transfer of Th17 cells into immunodeficient recipients, resulting in a population of intestinal effector T cells with mixed Th17 and Th1 phenotypes similar to that found in CD45RBhi
transfer controls. In a complementary study, it was found that memory Th17 cells isolated from the mesenteric lymph nodes of normal mice were resistant to transition to Th1 cells (Lexberg et al., 2008
), implying that under certain conditions the Th17 program becomes fixed, through mechanisms yet to be defined. These findings might explain the paired coexistence of Th1 and Th17 type cells in intestinal tissues —at homeostasis and in IBD — without invoking independent, parallel development of Th1 and Th17 lineages.
Flexibility in the late Th17 developmental program appears to be shared in man. CD4+
T cells isolated from the intestinal mucosa of CD patients demonstrated distinct subsets of Th1 and Th17 cells, as well as IL-17 and IFNγ double-expressing CD4+
T cells (Annunziato et al., 2007
), analogous to findings in mouse models of CD, and Th17 and ‘Th17-Th1’ cells cloned from intestinal isolates were diverted to a Th1 phenotype in response to IL-12. More recently, the expression of the NKT cell marker CD161 has been linked to Th17 cells present in healthy and chronically-inflamed patients (Cosmi et al., 2008
; Kleinschek et al., 2009
). Notably, CD161 also identifies a sizeable fraction of IFNγ-producers (even larger than the IL-17+
subset). Thus, although initially proposed as a marker for human Th17 cells, CD161 marks precursors that, in response to IL-1 and IL-23 — conditions that favor development of human Th17 cells — also give rise to an important, distinct population of IFNγ+
cells, suggesting that this marker might be common to precursors predisposed to both Th17 and Th1 phenotypes in response to Th17-promoting signals (Cosmi et al., 2008
). Thus, it appears that differentiating and early effector Th17 cells are phenotypically unstable and responsive to multiple stimuli that define their late developmental path. The impact of these phenomena on IBD development and persistence are likely to be important, but remains to be defined.