Abstract
It has long been recognized from epidemiological data that inflammatory bowel diseases (IBD), Crohn's disease (CD) and ulcerative colitis (UC), have a strong genetic predisposition, interacting with unknown environmental drivers to render susceptible individuals at risk for relapsing intestinal inflammation. Substantial progress has been made in the last 2 years in characterizing the susceptibility genes involved.
The recent acceleration in understanding has resulted from the use of new technologies of genome-wide association scanning in large panels of cases and controls.
Genome scans have robustly identified 11 susceptibility genes and loci and highlighted a number of new, previously unsuspected pathways as playing an important role in IBD pathogenesis—including the IL23 pathway in IBD overall and specific aspects of innate immunity (particularly NOD2 and the autophagy genes ATG16L1 and IRGM) in CD.
The next challenge is to identify specific causal variants at each of the confirmed susceptibility loci and then characterize their biological impact on gene expression and function of the protein product.
To date, most attention has focused on CD. A recent meta-analysis has increased the number of confirmed susceptibility loci to 32—more than for any other common disease to date. Attention is now turning to the use of the same techniques in UC to identify new, disease-specific genes and understand areas of overlap.
This review explores genetic clues to the pathogenesis of IBD derived from the growing list of confirmed IBD susceptibility genes, and briefly elaborates some of the important themes and overlaps that are becoming evident both within IBD and also with other complex diseases.
Introduction
Inflammatory bowel disease (IBD) is characterized by chronic, relapsing intestinal inflammation producing debilitating symptoms of diarrhoea, abdominal pain and malnutrition. Its main forms are ulcerative colitis (UC), where intestinal inflammation is limited to the colon, and Crohn's disease (CD), which can affect any part of the gastro-intestinal tract—most typically the terminal ileum and colon. As for most auto-inflammatory conditions, it is thought that genetic, immune and environmental factors combine to affect the disease predisposition and its course. Over the last 20 years, most attention has focused on understanding the dysregulated elements within the mucosal immune system that characterize IBD. However, besides implicating an important permissive role for luminal bacteria as drivers of chronic intestinal inflammation, these studies have not identified primary causal factors for IBD. This has changed fundamentally with the recent advances in molecular genetics.
As for other complex diseases, the dramatic progress in understanding of IBD genetics has depended substantially on the elucidation of the basic sequence and structure of the human genome. Of particular importance has been the systematic identification of human sequence variation and patterns of linkage disequilibrium across the genome—particularly from endeavours such as the HapMap project (www.hapmap.org).1 In addition advances in genotyping technology, statistical methodologies and bioinformatics, as well as national and international collaborations using highly powered datasets, have played key roles. These factors, combined in the new technique of genome-wide association (GWA) scanning, have restored momentum to a process that seemed to stall for 5 years following the identification of NOD2 as the first CD susceptibility gene in 2001.2,3 This has now led to the identification of over 30 more confirmed IBD susceptibility genes, opening the door on a new era in IBD research.4 In the current article, we systemically review the progress that has been made, and briefly describe what is known of the function of the genes implicated and the impact of the disease-associated variants.
Genetic epidemiology of IBD
Epidemiological evidence strongly implicates genetic factors in IBD susceptibility. This is reflected convincingly in family studies which identify a much higher incidence of IBD in individuals with a positive family history compared with those without; and in twin studies which show that monozygotic twins have a much higher rate of disease concordance than dizygotic twins.
Among family studies, the highest prevalence of IBD among relatives of IBD patients came from an American study by Farmer et al.5 at the Cleveland Clinic. Among 522 CD patients, 187 cases (35.2%) had a positive family history, in 87 cases (16.7%) this being a first-degree relative. For UC, 93 of 360 patients (29.4%) had a positive family history, with 50 (15.8%) being first-degree relatives.5 Across all studies, the estimate of λs—the risk to siblings of a patient developing the disease compared with a member from the general population—is 20–35 for CD and 8–15 for UC.6 Although confounding factors clearly contribute to the higher rate among family members due to the shared environment, this effect is off-set in twin studies. These demonstrate concordance for CD in 35% of monozygotic twin pairs compared with ∼7% in dizygotics, with the equivalent in UC being 11 versus 3%.7 The consistent pattern emerging from the epidemiological studies therefore is that genetic factors play an important role in IBD susceptibility, and that they have a stronger effect in CD compared with UC.
Studies of multiply affected families have raised another interesting point, that the two clinically similar diseases CD and UC share some genetic susceptibility loci. Evidence for this comes from the fact that an intermediate phenotype—that of indeterminate colitis—accounts for up to 10% of IBD cases, and the recognition that CD and UC co-occur in up to 30% of multiply affected families.8 There are also rare examples of discordant IBD phenotypes in monozygotic twins.9 With the identification of increasing numbers of IBD susceptibility genes, it has been interesting to observe which are common to CD and UC and where they differ.
Molecular genetic techniques used in IBD
Although compelling evidence for the importance of genetic susceptibility to IBD has been available from epidemiological data for decades, identification of the precise genetic determinants has only become possible in recent years.
Broadly, two different genetic research strategies have been applied. One has investigated candidate genes and the other uses hypothesis-free methods of genome-wide scanning. The former is based on specific attributes of the selected candidate gene implicating its role in disease pathogenesis—such as its expression or function, or evidence from sources such as linkage studies, animal models or other related diseases. Overall, candidate gene strategies have not proved to be particularly fruitful in IBD, but one success was the identification of association between the major histocompatibility complex (MHC) region and UC susceptibility—initially in Japanese cohorts (Asakura et al. found HLA-DR2 to be associated, with Sugimura et al. identifying the DRB1*1502 allele as responsible for this10,11) and later in Europeans in whom the DRB1*0103 allele has been implicated in both severe UC and extra-intestinal manifestations of IBD.12,13
Hypothesis-free methods of genome scanning now provide a much more powerful approach. They give an unbiased survey of the whole genome for IBD-associated loci, and have the potential to provide profound and novel insights regarding disease mechanisms/pathways. This method has evolved from the low-resolution linkage analyses based on multiply affected families which were widely applied in the late 1990s. With advances in knowledge of genetic variation, patterns of linkage disequilibrium and genotyping technology, as well as developments in statistics, genome scanning is now based on association analysis and consequently provides much higher levels of resolution for complex disease gene localization.
Association with disease can be tested directly (investigating known coding and functional variants in the hope of detecting the actual disease-causing mutation) or indirectly (testing genetic variants that act as proxies for the mutation by virtue of linkage disequilibrium). Although the former might appear efficient and intuitive, it has become increasingly clear that the latter is much more powerful and indeed that the majority of variants associated with complex disease are non-coding, presumably regulatory polymorphisms.
Development of a dense map of single nucleotide polymorphisms (SNPs) and discovery of the haplotype block structure of the genome have dramatically aided efforts to detect complex disease susceptibility genes by indirect association. The fact that >80% of the genome is made up of 5–100 kb blocks of sequence unbroken by recombination during evolution14 means that all common alleles present in such blocks can be ascertained or ‘tagged’ by genotyping just a small minority of the variants (tag SNPs) present. With this knowledge, a large proportion of common variation in the human genome (including the ∼10 million SNPs as well as more complex insertion/deletion and copy number variants) can be ascertained using the new commercial genotyping chips of 500 000–1 000 000 SNP markers.15 When applied to appropriately large case–control datasets, these provide a highly powered, high resolution means of detecting complex disease susceptibility loci.
‘Historical’ associations
NOD2
The NOD2 gene on chromosome 16q12 was the first susceptibility gene for CD to be successfully identified. It was detected by parallel strategies of positional cloning within a region of linkage and positional candidate gene investigation2,3—and the finding was subsequently widely replicated in European populations. Three common disease-associated mutations have been identified as R702W (Arg702Trp), G908R (Gly908Arg) and 1007fs (Leu1007fs). Interestingly, researchers from China and Japan have not found association between NOD2 variants and CD in their populations, highlighting significant ethnic heterogeneity.16,17 Further evidence for disease heterogeneity came from a meta-analysis which showed that variants within NOD2 mainly predispose to ileal CD, and are not associated with either colonic CD or UC.18
NOD2 encodes an intracellular receptor predominantly expressed in monocytes and Paneth cells. This has been implicated in the innate immune response to muramyldipeptide (MDP), a component of peptidoglycan in bacterial cell walls. The CD-associated variants cluster in the C-terminal LRR part of the NOD2 protein, significantly diminishing responsiveness to MDP.19 On exposure to MDP, oligomerized NOD2 recruits RIP2 (the serine-threonine kinase RICK) and subsequently activates NF-κB transcription factor via NEMO ubiquitination and IκB degradation. This leads to production of cytokines and cryptdins/defensins, hence facilitating clearance of bacteria.19 In addition, there is evidence of cross-talk between NOD2 and toll-like receptor pathways.19 The precise mechanism by which NOD2 mutations lead to increased intestinal inflammation is unknown but it may be that reduced ability to clear bacteria by innate immune mechanisms leads to dysregulation of adaptive immune pathways.
IBD5 and OCTN2
Following fine mapping of the IBD5 linkage region on chromosome 5q1, consistent evidence for association between CD and a haplotype of markers spanning 250 kb has been observed. More recently, the same locus has been associated with UC also.20 However, due to the strong LD across this region, it has been very difficult to identify the causal variant. Peltekova et al.21 reported a two-locus risk haplotype in the region of the organic cation transporter (OCTN) genes and suggested that this accounts for the association. The two-locus haplotype comprises L503F (1672 C–T, missense substitution) and G207C (transversion) in the SLC22A4 (OCTN1) and SLC22A5 (OCTN2) genes, respectively. However, there remains significant debate as to whether these transcripts are truly implicated by the genetic evidence, and a number of other immunoactive candidates remain in the frame—including interferon regulatory factor1 (IRF1) and a number of important cytokine genes (IL-4, IL-13, IL-5, IL-3) located within the 250 kb risk haplotype.22
Many studies have confirmed the association of CD with the IBD5 risk haplotype in Europeans.22 In studies of Japanese populations, the variants present differ; association with CD is seen but to a much more modest extent.23,24
HLA region
Both genome-wide linkage scans and candidate gene studies implicated variants in the MHC in IBD, especially in UC. This region contains multiple immunoregulatory genes, including the human leukocyte antigens and TNF alpha gene. Pinpointing the causal variants is confounded by tight linkage disequilibrium across this region. In addition, there is further ethnic heterogeneity between European and Japanese populations as indicated above.
Satsangi et al.12 reported association between severe UC requiring surgery and the rare HLA DRB1*0103 allele. This finding was replicated but until recently it was unclear whether it reflected indirect association with a distant variant. Recent evidence from GWA scans has at least refined the signal to the 400 kb haplotype block containing DRB1*0103, and shown that this locus is common to UC and the colonic (but not small bowel) subphenotype of CD.25 Further fine mapping and functional analysis is required to clarify whether DRB1*0103 is itself the causal variant at this locus.
Results from GWA scans
IL23R and IL12B
Variants in the interleukin 23 receptor gene (IL23R) on chromosome 1p31 have been unequivocally associated with both CD and UC susceptibility. Duerr et al. performed a GWA scan of 300 000 SNPs in an ileal CD case–control panel of 567 cases and 571 controls and found strongest association at IL23R—as well as the previously identified NOD2 gene. The finding was rapidly replicated by our UK consortium,26 and we also observed highly significant association with UC.25 Perhaps surprisingly a rare variant (Arg381Gln) in the IL23R gene conferred protection against developing IBD, with multiple other IL23R variants showing independent association. Once again association of IL23R was not replicated in a well-powered CD cohort in Japan.
The genetic data for IL23R arrived simultaneously with clear functional data from mouse models of IBD for the importance of the IL23 pathway and Th17 lineage in IBD pathogenesis. Knockout of or antibodies to IL23 prevent the development of intestinal inflammation in such models.27 It is now evident that much of the function previously ascribed to IL12 appears to relate to IL23, both of these cytokines sharing a p40 subunit in their heterodimeric structures. Indeed, recent data from the UK further highlight the importance of this pathway by demonstrating association between variants in IL12B, which encodes the shared p40 subunit, and both CD and UC susceptibility (Fig. 1).25,28
The IL23/IL12 pathway has become the subject of intensive study in the field of immunology as it plays a key role in determining differentiation of naïve T cells into effector Th1 cells (driven by IL12) or Th17 cells (driven by IL23). Begum et al.29 reported that some specific bacterial components such as peptidoglycan can differentially induce antigen presenting cells to produce IL23 rather than IL12—with this early regulatory mechanism potentially leading to distinct patterns of inflammatory response. Th17 cells are particularly interesting for their role in organ-specific inflammation—raising the hope that therapeutic disruption of the IL23 pathway will control such inflammation without impairing systemic immunity. Recent pilot data of anti-p40 therapy—blocking both IL12 and IL23—appeared effective in CD. Specific anti-IL23 antibodies are in development.
ATG16L1
Association between the ATG16L1 (autophagy-related 16-like 1) gene on 2q37.1 and CD was first identified by Hampe et al.30 who studied ∼20 000 non-synonymous SNPs in 735 CD cases and 368 controls. One amino acid changing polymorphism (Ala197Thr, rs2241880) within the ATG16L1 gene was highly associated with CD—a finding replicated in the index report and subsequently widely elsewhere.31,32 Based on haplotype and regression analysis, this variant was found to explain all the signals observed at the ATG16L1 locus. It is not, however, associated with UC.25
The detailed function of this non-synonymous SNP remains unclear, although Ala197Thr occurs in a conserved domain. ATG16L1 itself is known to play a key role in autophagy31—the process by which cells encapsulate cytosolic debris and micro-organisms within a membrane bilayer, which is then fused with a lysosome for degradation of its contents. This is now recognized as a key mechanism of innate immunity. The ATG16L1 protein is widely expressed not only in intestinal epithelial cells, but also in lymphocytes and macrophages. Functional knockout markedly impairs killing of intracellular pathogens such as Salmonella typhimurium.31 It seems likely that ATG16L1 influences CD susceptibility by altered ability of autophagy to eliminate microbes in innate immune cells of the intestine.
IRGM
Further emphasizing the importance of autophagy in CD pathogenesis, our UK group working within the Wellcome Trust Case Control Consortium (WTCCC) identified association between variants in another key autophagy gene—immunity-related guanosine triphosphatase (IRGM)—and CD susceptibility. A panel of 2000 CD cases and 3000 controls was used in the index GWA scan, with replication in an independent panel.28,32 Subsequent resequencing did not identify any coding variants in IRGM, and thus it is thought that the causal variants are likely to exert a regulatory effect on gene expression.
The IRGM protein appears to exist in a number of isoforms. It is small (∼180 amino acids) and binds GTP to induce autophagy and subsequently generate large autolysosomal organelles as a mechanism for eliminating intracellular bacteria.33 A role in control of M. tuberculosis has been demonstrated, and it is possible that defects in the autophagy machinery account for the persistence of subpathogenic intracellular bacteria such as adherent invasive E. coli observed in CD.34 It is widely accepted that commensal intestinal bacteria play an important role in the pathogenesis of CD, but the mechanism to date has remained unclear.
As for NOD2, association between CD and variants in key autophagy components further highlights the importance of defects in innate immunity to CD pathogenesis. This may in part explain how subpathogenic commensal bacteria are able to provoke and sustain an inflammatory response in genetically susceptible individuals—either through persisting low-level infection or failure of bacterial clearance leading to stimulation of adaptive immune/inflammatory pathways.
PTPN2/PTPN22
WTCCC also found a strong association at rs2542151 5.5 kb upstream of PTPN2 on chromosome 18p11—a finding replicated in an independent cohort.28,32 The gene product is T cell protein tyrosine phosphatase (TCPTP), which is a key negative regulator of inflammatory responses. Dysregulated TCPTP is known to be associated with abnormal Jak-Stat signalling, and evidence from mouse knockouts suggests that lack of dephosphorylation within the signalling cascade leads to marked elevation of several pro-inflammatory cytokines including IFN-γ, TNF-α and IL12.35 As with many of the new confirmed susceptibility loci for CD, the key next steps are fine mapping to identify causal variants at this locus, followed by an assessment of their impact on TCPTP expression and function. Other loci are described briefly in Box 1.
ICOSLG—encodes a co-stimulatory molecule on antigen presenting cells; plays a key role in differentiation of Th17 cells from naïve CD4 T lymphocytes4
STAT3 and JAK2—signal transduction molecules critical to many cytokine (and other) pathways including IL23 and IL6 and their stimulation of Th17 cells36
CCR6—Chemokine receptor 6 encodes a G protein-coupled receptor expressed by immature dendritic cells and memory T cells—important for tissue-specific migration of dendritic and T cells during epithelial inflammatory and immunological responses.37 Recently shown to be expressed by Th17 cells in CD38
TNFSF15—encodes tumor necrosis factor superfamily, member 15—strongly associated with CD in Japan, somewhat more modestly so in European populations;39,40 known to be up-regulated in CD41 this molecule has a variety of actions including induction of nuclear factor kappa-B activation, potentiation of IL2 signalling and secretion of gamma interferon by T lymphocytes
NKX2-3—a poorly characterized transcription factor known to be expressed in the intestine; variants associated with CD and UC25,28,32—the former in Japanese as well as Europeans
PTGER4—expression of this gene correlates with polymorphisms mapping to a large gene desert on chromosome 5p13, which show strong association with CD;28,32,42 encodes prostaglandin receptor EP4 strongly implicated in mouse model of IBD43
MST1—(macrophage stimulating 1), encodes a protein which induces phagocytosis by resident peritoneal macrophages—variants associated with CD and UC25,28,32
ITLN1—encodes intelectin which recognizes galactofuranosyl residues found in cell walls of various microorganisms but not in mammals. Expressed throughout intestine, including in Paneth cells, strongly implicating a role in innate immunity44
ECM1—encodes extracellular matrix protein 1, variants of which were recently associated with UC susceptibility in a UK-based study.25 Altered intestinal permeability is the most likely mechanism
Common pathways
GWA scanning presents a hypothesis-free survey of the human genome. Hits are identified purely on the basis of statistical support for association between markers in their vicinity and disease susceptibility. One powerful aspect of this design is its ability to shed new light on biological processes previously unsuspected in the pathogenesis of the disease being studied. It is particularly exciting when more than one gene in the same pathway is highlighted. Such has been the case in CD for the autophagy genes ATG16L1 and IRGM, as discussed above. This whole pathway clearly merits close scrutiny in CD.
Likewise, the identification of strong association between both UC and CD and variants in the IL23 receptor gene as well as IL12B highlights the key role played by IL23 and the Th17 pathway in IBD pathogenesis.27,28,32 Association with IL23 pathway constituents illustrates another important aspect of the overlap seen in GWA scan results—namely the commonality of pathways between different diseases. Some of these were known to be related on epidemiological grounds, while in other cases the overlap has come as a surprise. In the former group is the common association with IL23 pathway variants seen not only in UC and CD, but also in ankylosing spondylitis and psoriasis.45 The role of Th17 pathways in organ-specific auto-inflammatory conditions is becoming more evident, and represents a clear target for therapeutic intervention. Examples of the latter include the common association for CD and type 1 diabetes at PTPN232 and for CD, type 1 diabetes, SLE and rheumatoid arthritis at PTPN2246—and overlapping associations between CD and type 2 diabetes at CDKAL1 and intelectin.47
It has long been postulated that CD and UC would share some susceptibility loci and differ at others. The recent GWA scans have gone a long way to detailing this relationship. It is clear that associations at NOD2 and the autophagy genes ATG16L1 and IRGM are CD-specific, while others such as the IL23 pathway associations, MST1 on chromosome 3p21 and NKX2-3 are generic IBD loci.25 The suggestion is clearly that the pathogenic mechanisms specific to CD relate substantially to microbial processing and handling of bacterial antigens. Another interesting observation is that UC and ‘colon-only’ CD have a common MHC association within a 400 kb haplotype block containing the DRB1*0103 locus—a locus which is not associated with CD which is confined to the small intestine.25 The implication is of shared pathogenic mechanisms for colonic forms of IBD distinct from small bowel inflammation—providing scientific evidence of the need to reappraise clinically based classifications of IBD.
The immediate future
Recent progress in IBD genetics has substantially advanced our understanding of IBD pathogenesis, especially for CD, and the scene is now set for the application of GWA scanning to UC. However, even with our recent meta-analysis identifying more than 30 distinct confirmed CD susceptibility loci, modelling suggests that only one-fifth of the genetic variance is explained by current findings.4 Although this seemingly low estimate might in part reflect uncertainties in the measurement of heritability (based on twin studies which in IBD have been small), nonetheless it is clear that a significant proportion of the genetic contribution to IBD susceptibility awaits discovery. Estimates of the proportion explained will increase as causal variants within the new loci are identified (odds ratios for causal variants will inevitably be higher than the estimates provided by their proxies—the tag SNPs which are the substrate of the GWA scans). However, current methods are poor at detecting rare, highly penetrant susceptibility variants, which might explain a significant proportion of the familial clustering observed in IBD. Furthermore, systematic ascertainment of association between common disease and copy number variants at a genome-wide level is yet to be achieved—but methodologies are evolving rapidly and will be deployed within the next year.
Already much data are available from GWA scans to inform functional assessment of new pathways previously unsuspected in IBD pathogenesis. Identification of causal variants will refine these experiments, and greatly facilitate attempts to understand the major environmental contribution to IBD pathogenesis. While the popular press seizes on the (very limited) crystal-ball gazing opportunities presented by genomics analysis, the true impact of GWA scans will be felt elsewhere. In particular, this will be in the improved understanding of pathogenic mechanisms, delineation of heterogeneity within the disease groups and the potential benefit that both of these will have on the improved therapy for IBD.
Funding
HZ is supported by a scholarship from the government of China; DM by an MRC clinical training fellowship and MT by a Wellcome Trust clinical training fellowship.
