Autophagy and Crohn's Disease: At the Crossroads of Infection, Inflammation, Immunity, and Cancer

I.A. Pathogenesis of Crohn’s Disease

Within the past ten years, epidemiological, clinical and basic studies have highlighted the importance of genetic, immunological and environmental factors in CD pathogenesis. Histologically, this disease is characterized by chronic inflammation with a massive infiltration of leukocytes into the intestinal mucosa. It is thus widely accepted that the development of CD results from an exaggerated inflammatory response to a hitherto undefined lumenal antigen, probably derived from the microbial flora. Given the therapeutic benefits of antibiotic treatment [5], intracellular bacteria appear to be involved in CD inflammation. However, the evidence for a causal pathogen remains elusive. The current consensus is that commensal gut bacteria rather than a particular pathogen may have a pivotal role in the maintenance of chronic inflammation, and that a genetic host defect may initiate the intestinal inflammation.

Indeed, there is a strong genetic basis for CD, as ~20% of people with CD have a family relative with CD, and 36% of monozygotic twins share the disease [4]. Such familial history of the disease, together with the observation that the inflammation can affect any part of the gastrointestinal tract, suggests germline mutations that predispose an individual to CD.

In 2001, the first gene was identified that correlates with susceptibility to CD, NOD2 (Nucleotide-binding Oligomerization Domain containing 2, also known as CARD15 for caspase recruitment domain-containing protein 15 [6–8]), which is an intracellular sensor of bacterial infection that drives via NF-κB the production of proinflammatory cytokines in macrophages and α-defensins in intestinal Paneth cells [9, 10]. However, NOD2 mutations per se are not sufficient to cause the expression of disease, as there are healthy individuals homozygous or compound heterozygous for NOD2 risk alleles. Moreover, the age of diagnosis and the disease aggressiveness differ significantly between family members having the same NOD2 mutation [11].

Since then, IBD have become recognized as complex multigenetic disorders. At least 32 different risk loci have been identified by eleven genome-wide associations (GWA) [12–22]. Some of these susceptibility genes appear to be specific to UC or CD, whereas others are non-specific IBD loci (Wellcome Trust Case Control Consortium-WTCCC [23]). IL23R was the first gene to be significantly associated with CD by GWA [12]. Among the pathways that emerged to be involved in CD is autophagy [24, 25]. In 2007, two major hits were observed in two different autophagy-related genes, ATG16L1 (autophagy-related 16-like 1) and IRGM (Immunity-Related GTPase M) by three independent GWA studies [14, 15, 18]. Soon afterward, more than forty studies confirmed these observations and provided new information on the importance of autophagy in the control of inflammation, immunity and cancer (Tables 1 and 2).

In this review, we discuss how such findings have undoubtedly opened up a new vista on CD pathogenesis. A unifying autophagy model then emerges that may help in understanding the development of CD from bacterial infection, to inflammation and then to cancer. The Pandora's box is now open, releasing a wave of hope for new therapeutic strategies to treat Crohn's disease.

I.B. ATG16L1 and IRGM as New CD-Specific Susceptibility Loci

The strong association of the rs2241880 single-nucleotide polymorphism (Thr300→Ala ATG16L1) first identified by Hampe et al. through GWA [14] was confirmed by twenty-five studies in several northern Caucasian CD cohorts (Table 1). However, no significant association is found in the Japanese, Chinese, and Brazilian populations, and conflicting data are reported in two Italian studies [26–31]. Further genetic studies are thus required to confirm the association of the ATG16L1 rs2241880 variant with ethnically divergent populations. Similarly, several IRGM risk polymorphisms are associated with CD and replicated in independent cohorts of North America, North Europe and New Zealand (Table 2) [12, 14–16, 18, 24, 25, 32–34]. Compelling evidence indicates that the ATG16L1 and IRGM loci do not interact with other susceptibility loci for CD (CARD15; IL23R), and no significant association with specific CD sub-phenotype (inflammatory, stricturing, penetrating disease, etc.) has been identified. Of particular importance, a meta-analysis of all WGA indicates that ATG16L1 and IRGMlike NOD2are involved specifically in ileal CD inflammation [23]. Therefore, at least some types of CD can now be viewed as an autophagy disease similar to infection, neurodegenerative diseases and cancers [35].

III.A. IRGM: A Safeguard Against Bacterial Infection

The human IRGM is a poorly understood regulator of xenophagy. It belongs to the p47 Immunity-Related GTPase (IRG) family, one of the most powerful innate resistance mechanisms against intracellular pathogens (Gram-positive and Gram-negative bacteria, mycobacteria, and protozoans) [52]. It derives its name from the unique presence of a methionine within its GTP-binding motif [53, 54]. The mouse homologs, Irgm1 to Irgm3, are GTPases that upon GTP binding form regulatory homodimers and heterodimers with the other IRG proteins [55–57]. However, the human IRGM has a truncated GTP-binding domain and its GTP-binding properties have not been yet explored.

In uninfected cells, the IRG proteins reside in the endoplasmic reticulum (ER) and cis-Golgi complex [58–60]. Upon infection, the murine Irgm1 proteins translocate within minutes to the plasma membrane at the phagocytic cup as a pathogen (e.g., M. tuberculosis, L. monocytogenes) enters into the cell, and remain associated with the pathogen-containing phagosome as it matures [49, 59, 60]. Consistently, expression of human IRGM confers resistance against bacterial infection [32, 44]. It has been proposed that IRGM may protect the host against intracellular bacteria (a) by driving vesiculation and disruption of the phagosome, releasing the pathogen out of its protective niche into the cytosol (inset (Fig. 1B)) [57, 60, 61]; (b) then it may help drive the engulfment of bacteria into large autophagic vesicles through addition of IRGM-containing vesicles [61]; and (c) finally it may target the pathogen for lysosomal degradation by recruiting the autophagic machinery [44, 49, 60, 61]. At the time of writing, the exact roles of human IRGM, its regulation and its autophagic partners remain to be identified.

III.B. Atg16L1: An Essential Component of the Autophagic Machinery Involved in Host Defense and Paneth Cell Biology

Atg16L1 is a scaffold protein with an N-terminal protein interaction domain, a coiled coil domain and seven C-terminal WD repeats [62–64] (Fig. 1A). It interacts constitutively with Atg12–Atg5 protein conjugates via its N-terminal domain, and Atg16L1 self-assembles via its coiled-coil domain, thus forming a high molecular weight complex (Atg12–Atg5-Atg16) of ~800 kDa. Within this platform, Atg16L1 is absolutely required for the localization of the complex to the phagophore and the formation of autophagosomes via the conjugation of LC3 to phosphatidylethanolamine [65] (Fig. 1B). As a result, the loss of Atg16L1 severely impairs autophagosome formation, degradation of long-lived proteins and the clearance of bacteria within both immune and epithelial cells [15, 32, 40, 43, 44, 66].

Recently, Cadwell et al. (2008) provide the unexpected demonstration that Atg16L1, and likely the entire autophagy process, is important for the biology of the Paneth cell, a specialized epithelial cell that releases antimicrobial peptides (such as lysozyme and α-defensin) into the intestinal lumen [66]. Indeed, ATG16L1–deficient Paneth cells exhibit not only defective granule exocytosis but also unexpected overexpression of two adipocytokines, leptin and adiponectin, known to directly influence intestinal injury responses. Similarly, CD patients homozygous for the ATG16L1 CD risk allele display comparable Paneth cell granule defects and increased leptin levels [66].

III.C. Dosage Effects of Atg16L1 and IRGM Variants in CD Disease

Compelling evidence indicates that a critical threshold of IRGM and Atg16L1 expression needs to be reached to trigger full xenophagy. Indeed, homozygotes for the ATG16L1 T300A variant have a greater risk than heterozygotes, suggesting a potential gene dosage effect [68]. Exposure to pathogens increases the expression of autophagy machinery such as LC3B, Beclin 1, and Irgm1 to upregulate xenophagy. Consistent with this finding, overexpression of human IRGM confers protection against bacterial infection and, conversely, its removal by siRNA renders macrophages and epithelial cells more permissive for microbial replication [32, 44]. Similarly, the bacterial load within autophagic vesicles is reduced by ATG16L1 disruption [15]. A lower rate of xenophagy is correlated with a decrease in both Atg16L1 T300A and Atg5 expression levels [43]. Therefore, both IRGM and Atg16L1 expression levels may serve to regulate the rate of xenophagy. However, the cooperative roles of IRGM and Atg16L1 in xenophagy remain unknown.

Altogether, this points to the attractive pathogenesis model that the germline ATG16L1 and IRGM variations observed in CD patients would impair in vivo innate resistance in all sentinel cells (immune cells, Paneth cells, and enterocytes) and thereby trigger excessive inflammation as a result of increased bacterial load.

IV.A. Bacterial Antigen Presentation on MHC Class II

The role of autophagy in immunity is not limited to pathogen clearance. Autophagy also promotes the second wave of adaptive immune responses by delivering cytosolic antigens to autolysosomes, where they are degraded and loaded onto major histocompatibility complex class II (MHC II) molecules [69–72]. Peptide–MHC II complexes are then presented onto the cell surface and recognized by T cells with their specific T-cell receptor (TCR) [73]. One can therefore assume that the autophagic degradation of intracellular pathogens would provide an important source of bacterial antigens for MHC class II presentation. Such antigen delivery by autophagy might be relevant for professional antigen-presenting cells (dendritic cells, macrophages) and more particularly for epithelial cells with little or no phagocytic capacity.

IV.B. Autophagy as a Regulator of T Cell Life and Death

Inflammatory bowel diseases are associated with sustained T_H1, T_H2 or T_H17 cell responses. As a result, most of the attention in treating IBD has focused on neutralizing the immune response. Normally, homeostasis of the immune response is tightly regulated by a balanced T cell growth and death in which T cell numbers drastically increase in response to their specific antigens and decrease upon antigen clearance.

Of particular importance, autophagy has recently emerged as a key player in the control of T-cell life span. The first indication for a pro-death function of autophagy comes from in vitro studies linking the induction of autophagy to conditions that model T cell homeostasis. Indeed, autophagy is induced upon TCR, IL-2 stimulation as well as growth-factor withdrawal in both T_H1 and T_H2 cells. Interestingly, the induction of autophagy by growth-factor depletion is much more robust and persistent in T_H2 than in T_H1 cells, leading to T_H2 cell death [74]. In contrast, another study demonstrates that activation of autophagy ensures in vivo T cell survival. Autophagy-deficient (Atg5^−/−) mice exhibit multiple defects, including a severe reduction of thymocytes (40% of control) and of peripheral T- and B-lymphocytes (only 10% of control), and an inability to undergo TCR-induced proliferation [75]. Therefore, autophagy may control the duration and strength of adaptive responses by executing both T cell life and death decisions.

A challenge for the future will be to define the consequences of the Atg16L1 and IGRM variants on T cell immune responses. Recently, analyses of IRGM1^−/− mice reveal a critical role for Irgm1 in the survival of mature CD4⁺ T cells [76] and in the renewal of hematopoietic stem cells upon infection [77]. Strikingly, this GTPase was dispensable for haematopoiesis under steady-state conditions. As a result, the IRGM1 null mice have no apparent phenotype, but develop upon infection a profound lymphopenia [45, 60, 78]. Thus, the murine Irgm1 is absolutely required for the expansion of antigen-specific T_H1 populations to combat effectively intracellular infection. From a therapeutic perspective, it will be imperative to find out to what extent a decreased autophagy, due to ATG16L1 and IGRM variants, might affect T cell survival upon infection.

IV.C. Autophagy in Tolerance and Autoimmunity

High levels of autophagy are observed in thymic epithelial cells [79], and consistent with this finding autophagy is involved in the presentation of self-antigens and the maintenance of CD4⁺ T-cell tolerance through thymic T-cell selection [80]. Similar to CD pathogenesis, a defect in autophagy in thymic epithelial cells is sufficient to lead to a breakdown in self-tolerance and the development of autoimmune diseases in the colon [80]. It is therefore tempting to speculate that decreased macroautophagy caused by ATG16L1 or IRGM variants might lead to insufficient tolerance against commensals or self-antigens in the gut; a possible explanation for some of the T cell– driven pathologies observed in Crohn's disease.

V.A. Activation of Autophagy by Inflammatory and Immune Signals

Crohn's disease is characterized by increased production of T_H1 cytokines such as TNF-α and IFN-γ. Interestingly, the relationship between autophagy and immunity is reciprocal: autophagy does enhance adaptive immune responses, and in turn cytokines and receptors involved in innate and adaptive immunity upregulate autophagy. Immune and inflammatory signals that positively regulate autophagy include the bacterial LPS [81], the T_H1 cytokines IFN-γ [44, 49, 61], and members of TNF family (TNF-α, TRAIL and CD40L) [82–87]. Interestingly, both IFN-γ and LPS induce autophagy in infected epithelial cells through upregulation of murine Irgm1 [44, 49, 61, 81, 82]. Such an amplification loop connecting autophagy and immunity might contribute to an effective resolution of infection [73].

Of note, autophagy is negatively regulated by the T_H2-type cytokines, IL-4 and IL-13 both in epithelial [83, 88–90] and immune cells [91], a finding that may help explain the negative role of T_H2-cell response in the control of intracellular pathogens.

V.B. Control of Inflammation by the Autophagy Pathway: Pandora's Box

Autophagy may help to control the intensity and duration of the inflammatory responses not only by eliminating pathogens but also by blocking cell necrosis, protecting cells from oxidative stress and limiting the production of inflammatory cytokines.

VI. AUTOPHAGY: A “TWO-FACED” PATHWAY IN COLORECTAL CANCER

Colon cancer is the second most common cause of tumor-related death worldwide with 900,000 new cases and over 500,000 deaths per year [99, 100]. Of particular interest, it is well recognized that colorectal cancers (CRC) often arise at sites of chronic inflammation, with IBD patients having a six-fold higher risk of developing this cancer. However, there is no specific treatment and no prognostic marker predictive of cancer in this risk population

Interestingly, autophagy seems to be at the heart of tumorigenesis, acting as a potent tumor suppressor pathway (Fig. 2B). During tumorigenesis, autophagy is downregulated by several oncogenes (i.e., PI3K/Akt, Ras) and tumor suppressors (i.e., PTEN and p53) that are frequently mutated in CRC (for review see [101]). Moreover, several ATG genes such as the ultraviolet radiation resistance-associated gene (UVRAG), Atg2, Atg4b and Atg12 are mutated or downregulated in CRC [102–105]. One consensus that therefore emerges is that autophagy suppresses tumor initiation. Conversely, an enhanced autophagy may also promote the progression of established CRC. Enhanced expression of key Atg proteins, LC3 and Beclin 1, is observed in early and advanced stages of CRC [106–108], enabling likely cancer cells to endure hypoxia, and nutrient limitation in the inner area of the tumor [109, 110] and later to survive anoikis during metastasis [111]. Not surprisingly, depending on the conditions, upregulation of autophagy may protect colorectal cancer cells against anticancer treatments such as 5-fluorouracil [112], or may alternatively result in cell death [113–115].

Despite the huge amount of work, there is no marker to predict the course of Crohn's disease, its relapse or recurrence, and the risk of cancer complications. It will be therefore important to determine whether ATG16L1 and IRGM variants may predispose to colorectal cancer. In this regard, evidence of a frequent overexpression of Atg16L1 protein during oral carcinogenesis is of particular interest [116].

Prognosis

CD is a chronic disorder with an unpredictable disease course. Most likely, such clinical heterogeneity results from differences in genetic susceptibility and exposure to environmental factors. Along these lines, we should keep in mind that CD is a complex disorder that involves at least 30 distinct susceptibility loci (meta-analysis of GWA [23]). Of particular importance, all identified polymorphisms are independently associated, and alone only moderately increase the risk for disease development [23]. Individuals carrying each of 7 different risk alleles (including NOD2, IBD5, ATG16L1and IL23R) have 25 x-increased risks for CD development with a more severe disease course (necessity of operations and younger age of onset, <40 years) [123]. In the near future, it might be possible to detect a genetic risk signature that predicts the course of CD and the drug response. Then, clinicians might be able to identify patients with a poor prognosis and to adjust their treatment accordingly.

Effective Therapeutic Strategies

Treatments for CD include corticosteroid in the acute phase. Patients who relapse frequently require immune suppression (usually with azathioprine, 6-mercaptopurine or methotrexate) or anti-TNF antibody treatments such as infliximab. Although the majority of patients can be maintained in remission with the use of such treatments, a significant portion are non-responders at the beginning or lose response after a period on therapy [124]. As such, resectional surgery remains the sole option in up to 80% of CD patients over the long term [125]. However, even surgical procedures carry appreciable morbidity, as the bowel function seldom returns to a pre-disease level.

The recent identification of risk variants in autophagy genes provides the rationale to prioritize this pathway as a new potential target for drug development. Clinically available drugs that upregulate autophagy such as sirolimus and everolimus (two rapamycin analogs) may be helpful for treating CD [126]. This possibility was supported by the efficacy of everolimus in the prevention/treatment of colitis in IL10^−/− mice; a well-characterized mouse model of CD [127]. In 2008, Dumortier et al. and Massey et al. reported two cases of patients with severe refractory CD who were successfully treated with everolimus (4 mg/day – 18 months [128]) or sirolimus (4 mg/day – 6 months [129]). Both studies described a sustained improvement in CD symptoms with normalization of Harvey–Bradshaw index, of serum inflammation markers and of endoscopic appearance (Fig. 3). These encouraging first case reports open the avenue where pharmacological manipulation of autophagy might be used for prevention of CD inflammation. But there remain, undoubtedly, many burning issues to address. In particular, it will be important to evaluate the efficacy, safety and long-term outcomes of upregulating autophagy. While sirolimus and erolimus were effective and well tolerated during the first months [128, 129], long-term systemic autophagy induction might be associated with important side-effects [130, 131]. Of particular concern, such a strategy might exacerbate the progression of established colorectal cancers, as suggested by certain studies [110, 132]. In contrast to these initial reports, a double-blind randomized multicenter study of everolimus (6 mg/day) has failed to demonstrate benefit versus placebo in patients with active CD [131]. Despite these caveats, we guess that pharmacogenetics would rapidly optimize the clinical response of such a strategy.

In conclusion, in less than two years we have witnessed with Crohn's disease the most successful application of GWA from the disease gene association to a therapeutic strategy. The challenge will be now to identify those patients who are more likely to respond and the combination of autophagy inducers that would be most effective to improve/slow down CD development.