Genetics of Inflammatory Bowel Disease

[David in Seattle]

Here's a journal article that goes into some depth on genetics and IBD, mentioning the work of Dr. Strober, who was mentioned in the WSJ article posted the other day by garygepner1

From Expert Review of Gastroenterology and Hepatology
Genetics of Inflammatory Bowel Disease: Implications for Disease Pathogenesis and Natural History
Implications for Disease Pathogenesis and Natural History

Charlie W Lees; Jack Satsangi

Authors and Disclosures

Posted: 12/15/2009; Expert Rev Gastroenterol Hepatol. 2009;3(5):513-534. © 2009

Abstract and Introduction
Abstract

Epidemiological data, detailed molecular studies and recent genome-wide association studies strongly suggest that ulcerative colitis (UC) and Crohn's disease (CD) are related polygenic diseases that share some susceptibility loci, but differ at others. To date, there are more than 50 confirmed inflammatory bowel disease genes/loci, a number that is widely anticipated to at least double in the next 2 years. Germline variation in IL23R, IL12B, JAK2 and STAT3 is associated with inflammatory bowel disease susceptibility, consistent with the newly described role for IL23 signaling and Th17 cells in disease pathogenesis. Several genes involved in different aspects of bacterial handling are defective only in CD, including NOD2 and the autophagy genes ATG16L1 and IRGM. IL10 and ECM1 are associated with UC, while inherited variation at the HLA region is related to an inflammatory colonic phenotype. The application of genome-wide association studies to inflammatory bowel disease has been successful in defining the genetic architecture of CD and UC and in delivering genuinely novel and important insights into disease pathogenesis. This has unearthed a plethora of attractive targets for the development of future therapeutics. Insights into the natural history of these complex diseases will follow and may enable appropriate patient selection for early aggressive therapy with the view to modifying the disease course.
Introduction

After birth and during early life, the gut lumen is colonized by an estimated 1014 bacteria. Thereafter, intestinal homeostasis is crucial for efficient energy extraction from foodstuffs and protection from pathogens. The single-layered epithelium separates the intestinal lumen and the mucosal immune system. Renewing itself every 5–7 days, the epithelium consists of stem cells that differentiate into enterocytes, goblet cells, enteroendocrine cells, Paneth cells and microfold cells (Figure 1). Differentiation and maintenance of these epithelial cells is achieved via a complex network of interacting signaling pathways (WNT, Notch, hedgehog, ephrin and BMP genes). Goblet cells produce the mucus layer at the epithelial surface that acts as a first barrier of defence. Paneth cells secrete specialized antimicrobial peptides known as defensins in man (cyptidins in mouse). Microfold cells and lamina propria dendritic cells (DCs) can directly sample intraluminal microbial components. A combination of these ancient innate immune responses, presentation of antigen to naive CD4 T cells in the lamina propria with subsequent conditioning into Th17 or Treg subsets (in draining mesenteric lymph nodes) and appropriate epithelial responses to injury (both via direct epithelial signals and indirect signals back from the lamina propria immune cells), serves to maintain intestinal homeostasis.

In the context of this highly complex environment at the mucosal interface, it is perhaps not surprising that an estimated one in 200 people in western Europe and Northern America develop a form of chronic inflammatory bowel disease (IBD), with disruption of intestinal homeostasis and subsequent chronic inflammation.[1] The precise etiologic and pathogenetic mechanisms underpinning the pathogenesis of Crohn's disease (CD; MIM 26600) and ulcerative colitis (UC; MIM 191390) remain uncertain. However, the presently available data overwhelmingly support a hypothesis centered around a dysregulated host immune response to intestinal bacteria (commensal and/or pathogenic) in genetically susceptible individuals.

Factors known to disturb intestinal homeostasis that probably play a role in IBD pathogenesis include alterations in epithelial barrier function, innate immune cells (including macrophages and DCs), lymphocyte function (imbalance between regulator and effector cell function) and stromally derived factors (e.g., TGF-β).[2] DCs activated by either pathogenic bacteria or disturbed epithelium translocate to the mesenteric lymph node where they instruct naive T cells to adopt a proinflammatory phenotype. Recent data demonstrates the importance of IL17-producing T cells (Th17) in intestinal inflammation. Th17 cells are involved in clearance of pathogens not dealt with by Th1 or Th2 cells and are potent inducers of tissue inflammation.[3]

Genetic Architecture of Inflammatory Bowel Disease

In addition to the importance of environmental factors (e.g., luminal flora and cigarette smoke), there are considerable epidemiological data that implicate genetic susceptibility in the pathogenesis of CD and UC. Most notably, these include the familial prevalence of IBD, concordance rates in twin pairs and ethnic differences in disease susceptibility. It was the studies of twin pairs that provided the strongest impetus towards further investigation of genetic susceptibility in IBD. Three studies have been carried out in Europe, including Tysk's important review of the Swedish Twin Registry in 1988.[4–6] The data from these studies, in Sweden, Denmark and the UK, combine to provide powerful evidence for the role of both genetic and environmental factors in disease susceptibility. The concordance rates for CD in monozygotic and dizygotic twin pairs from these studies are estimated at 37 and 7%, respectively; in UC, the equivalent results are 10 and 3%. The relative risk of developing CD in a first degree relative is 5–35 and for UC 10–15.[7]

Based on these epidemiological data, international teams have been searching for IBD susceptibility genes over the past 15 years (Figure 2). The establishment of a linkage map of the human genome using informative microsatellite markers in the 1990s paved the way for hypothesis-free scanning for loci of association in monogenic and complex genetic disorders. Using this model, nine IBD susceptibility loci (designated IBD1–9) were discovered and replicated to a varying extent. Some of these loci appear to be relatively specific for CD (e.g., IBD1 on 16q [OMIM 266600])[8,9] and UC (e.g., IBD2 on 12q [OMIM 601458]),[10–12] whereas others are associated with IBD as a whole (e.g., IBD3 on 6p [OMIM 604519]).[13–15] Despite much initial promise from these genome-wide linkage studies and the 2001 landmark discovery of NOD2 as a CD susceptibility gene, subsequent progress was largely frustratingly slow, and gene 'discoveries' werer notoriously difficult to consistently replicate.

 

[David in Seattle]

Figure 1.

Inflammatory Bowel Disease Pathogenesis.
*Inflammatory bowel disease susceptibility genes. APC: Antigen-presenting cell; DC: Dendritic cell; E: Enterocyte; G: Goblet Cell; Hh: Hedgehog; MDP: Muramyl dipeptide MØ: Macrophage; P: Paneth cell; RA: Retinoic acid; TLR: Toll-like recptor.

 

[David in Seattle]

Figure 2.

Timeline of Genetic Discoveries in Inflammatory Bowel Disease.
CD: Crohn's disease; GWAS: Genome-wide association studies; IBD: Inflammatory bowel disease; UC: Ulcerative colitis.

 

[David in Seattle]

However, the advent of genome-wide association studies (GWAS) in the past 2–3 years has completely changed the landscape and unparalleled insights into disease pathogenesis have followed ( Box 1 & Box 2). While these insights have been and remain the primary objective of these studies, medium- to long-term benefits for patients will include novel drug discovery and advanced predictive models of disease natural history. An emerging theme of complex disease genetics in the past 2 years is of multiple disease variants, each conferring small effects (odds ratios [OR] of < 1.20 have been consistently and convincingly replicated in several different studies). In this time period, there has been a number of high-profile GWAS in CD and, later, UC ( Table 1 ) that have, to date, yielded over 50 IBD disease genes/loci ( Table 2). A provisional scheme describing the genetic architecture of IBD is emerging, with many genes conferring susceptibility to IBD (CD and UC) and other genes being specific to either CD or UC (Figure 3). The genes involved in bacterial recognition/innate immunity (e.g., NOD2 and the autophagy genes ATG16L1 and IRGM) are specific to CD; the ECM1 barrier gene and IL10 are specific to UC; the IL23 pathway genes (IL23R, IL12B, JAK2 and STAT3) are common to CD and UC; whereas the HLA region probably confers risk to colonic IBD (i.e., UC and CD colitis).[16–25]

Figure 3.

Genetic Architecture of Inflammatory Bowel Disease.
The figure is limited to genes with definitive evidence of association in IBD from Table 2
#CD genes not fully tested in UC.
CD: Crohn's disease; IBD: Inflammatory bowel diease; UC: Ulcerative colitis.

 

[David in Seattle]

Insights into Disease Pathogenesis Arising from Gene Discovery

While the application of IBD susceptibility gene discovery to diagnostic and prognostic systems remains distant, the key findings are already bearing considerable insights into disease pathogenesis. It is not possible to cover all of the genes in great depth; rather, we will discuss several of the major themes emerging: the role of the IL23 pathway/Th17 signaling in IBD, defective bacterial handling in CD, IL10 signaling in UC, developmental pathways disrupted in IBD and the role of tyrosine phosphatases in disease pathogenesis. Brief summaries of other implicated genes are provided in Box 3.
IL23 & Th17 Signaling

Duerr and colleagues performed the first large-scale GWAS in IBD (a much more limited Japanese study was arguably the first GWAS),[26] analyzing over 300,000 single nucleotide polymorphisms (SNPs) on the Illumina® HumanHap300 Genotyping BeadChip (Table 1).[18] After applying Bonferroni correction for multiple testing, only three SNPs remained significant at the 0.05 level. Two of these SNPs were in NOD2 (rs2066843 and rs2076756), effectively providing proof-of-principle for this technique in discovering CD susceptibility genes. The third was a nonsynonymous SNP (rs11209026 Arg381Gln) in the IL23R gene. Several other SNPs within IL23R were subsequently shown to be associated with ileal CD, including some that were independent of the rs11209026 variant. It was proposed that the functional significance of these multiple variants within IL23R may be partly due to differential splicing, as at least six alternatively spliced mRNAs of IL23R are described.[18] Duerr and colleagues demonstrated replication of the IL23R association in an independent case–control association study of non-Jewish and Jewish ileal CD patients and in a family-based association study. Further replication studies for IL23R have been published in adult and pediatric populations in the UK,[27–29] France/Belgium,[23] Quebec Founder Population,[30] Italy,[31,32] Canada,[33] Holland,[34,35] Spain,[36] Germany,[37] Finland,[38] Hungary[39] and Brazil.[40] There was no consistent genotype–phenotype substratification apparent in these studies.[28,41,42] IL23R has also been confirmed as a UC susceptibility gene.[19,20,25]

In addition, the CD meta-analysis (Box 2) has confirmed three additional members of the IL23 signaling pathway as susceptibility genes.[17] IL12B, JAK2 and STAT3 were all shown to be associated with CD, and subsequently with UC,[19,20] confirming that defects in IL23 signaling confer risk to IBD as a whole.

The timing of the initial IL23R gene discovery study in CD was notable, as it coincided with detailed murine studies of bacteria-induced intestinal inflammation, demonstrating a critical role for IL23 in innate immune pathology.[43,44] The proinflammatory cytokine IL12 has long been thought to be a critical element in the development of pathogenic Th1 CD4+ effector cells. The discovery, in 2000, that IL23 is a heterodimer of p19 and the IL12p40 (IL12B) subunit[45] has led to a critical reappraisal of the relative roles that IL12 (a heterodimer of p40 and p35 subunits) and IL23 play in inflammatory disorders.[45] IL23 is secreted by activated DCs, monocytes and macrophages. Transgenic mice that constitutively overexpress IL23p19 develop a fatal multiorgan inflammation.[46] IL23 supports the development of the Th17 subset of CD4+ inflammatory T cells (that produce IL17, IL6 and TNF-α)[47–49] and has potent effects on innate immune cells (monocytes and macrophages) by inducing the production of IL1, IL6 and TNF-α.[49,50] IL23 and IL17 synthesis and secretion is increased in the inflamed murine intestine (Helicobacter hepaticus-infected Rag −/− mice develop a chronic typhlocolitis characterized by the accumulation of innate immune cells), whereas IL12 expression is unchanged.[43] IL23 blockade with an anti-p19 monoclonal antibody significantly decreased H. hepaticus-induced intestinal inflammation as well as decreased proinflammatory cytokines (TNF-α, IFNγ, IL6, IL1β and IL17) in the intestine. Equivalent findings have been demonstrated with anti-IL10R antibody treatment of H. hepaticus-infected T cell-sufficient hosts and CD4+ T-cell transfer into various Rag −/− recipients.[44] Combining these data demonstrate the crucial role for IL23, but not IL12, in intestinal inflammation, via sustained activation of adaptive and/or innate immune mechanisms.

Another related CD susceptibility gene, CCR6, is expressed by Th17 cells in the memory cell compartment of peripheral blood mononuclear cells (PBMCs).[51] CCR6 was identified as the CCL20 receptor in 1997.[52] CCL20 is the only CCR6-triggering chemokine and CCL20 is unable to elicit a biological response through other known chemokine receptors. Together, CCL20 and CCR6 are involved in the maturation of DCs.[53] They play a role in the recruitment of immature DCs and their precursors to sites of potential antigen entry. The Ccr6 −/− mouse lacks any gross abnormalities in any major organ.[54,55] However, it has underdeveloped Peyer's patches, and CD11b+ myeloid DCs, known for their functional CCR6 expression, are absent from the subepithelial dome. Ccr6 −/− mice also have increased subpopulations of T cells in the intestinal mucosa. CCL20 expression, constitutively at low levels in human intestinal epithelium, is induced in inflamed colonic tissue from patients with IBD,[56,57] as well as in PBMCs of patients with active UC compared with healthy controls.[58] CCL20 has been demonstrated to have antibacterial activity comparable to β-defensins.[59] Furthermore, human β-defensins have been reported to be nonchemokine ligands for CCR6, mediating the in vitro chemotactic activity of β-defensins for immature DCs and T cells.

Defects in Bacterial Handling in Crohn's Disease

One of the most consistent themes to arise from a primarily genetic approach is of defective bacterial handling in CD. Gene discovery and replication studies have confirmed that NOD2, ATG16L1 and IRGM are associated with CD but not UC. This has redirected research efforts into innate immunity and autophagy.

NOD2 Fine-mapping of the IBD1 region[9] and two positional candidate gene studies identified NOD2 (formerly CARD15) as a CD susceptibility gene in 2001.[60–62] Structural changes are induced in the leucine-rich repeat region of NOD2 by two SNPs (rs2066845/Gly908Arg and rs2066844/Arg702Trp) and an insertion mutation (rs2066847/3020insC) that leads to a frameshift substitution at Leu1007. The 3020insC mutation results in a premature stop codon and a truncated protein lacking part of the last leucine-rich repeat near the C terminus of the protein. While NOD2 responds to muramyl dipeptide (MDP; the minimal bacterial motif of peptidoglycan, found in the cell wall of Gram-positive and Gram-negative bacteria), no study has yet shown whether direct binding occurs. Unfolding of the NOD2 protein leads to oligomerization and exposure of the CARD domains, providing a platform for RIP2 recruitment and triggering of NF-κB, p38 and Erk MAP kinases.

 

[David in Seattle]

There has been much debate arising out of apparently conflicting experimental data as to whether the NOD2 mutations represent a 'loss-of-function' or 'gain-of-function' phenotype. Two different animal models (NOD2 −/− and NOD2 2939iC) published simultaneously in Science in 2005 added extra insight but failed to resolve this unsettling paradox.[63,64] Neither mutant developed chronic intestinal inflammation spontaneously; indeed, neither model demonstrated any abnormality in intestinal microstructure. However, when stressed with dextran sodium sulfate (DSS), the NOD2 2939iC mice behaved very differently to their wild type (WT) littermates, developing much more severe colitis with increased mortality (35 vs 0%, respectively). However, the Karin group was unable to breed this NOD2 knock-in mouse (NOD2 2939iC) onto a pure Black-6 background.[64] These in vivo data from a mixed background, therefore, need to be interpreted with some caution. NOD2 −/− mice responded less well than WT mice to challenge with oral Listeria monocytogenes,[63] an observation that may be related to decreased Paneth cell antimicrobial activity. The Strober lab has argued a gain-of-function model, whereby CD-associated NOD2 mutations result in increased Toll-like receptor (TLR)2-mediated activation of NF-κB signaling and hence excess secretion of Th1 cytokines.[65,66] The same group has subsequently demonstrated a possible therapeutic extension to this work, as addition of exogenous MDP protected WT mice (but not NOD2 knock-out or frame-shift mutant animals) from trinitrobenzene sulfonic acid and DSS colitis by downregulating multiple TLR responses (not just TLR2).[67]

NOD2 is constitutively expressed by macrophages, DCs and Paneth cells,[68] the latter being specialized epithelial cells located in the small intestine that secrete antimicrobial peptides, including α defensins (HD5 and HD6) into the base of the crypt. Patients with ileal CD have decreased HD5 and HD6 expression, particularly those carrying NOD2 mutations,[69] although this latter observation has not replicated in a subsequent study from Australia.[70] PBMCs from CD patients with NOD2 mutations have defective cytokine production and NF-κB activity in response to MDP,[71] lending further support to the loss-of-function hypothesis.

A recent study has demonstrated decreased production of bacterially-induced IL10 in primary human monocytes from CD patients with the 3020insC mutation.[72] 3020insC was shown to block human IL10 promoter activity. This was not seen in the equivalent murine mutation (2939insC), potentially explaining the lack of a CD phenotype in the NOD2 2939iC mouse. The 3020insC mutation also prevented NOD2 from forming a tri-molecular complex with IL10 and heterogenous nuclear ribonucleoprotein A1 (hnRNP A1).

NLRP3 The NLRP3 gene (1q44) encodes cryopyrin, a protein that controls the inflammation and thus regulates activation of caspase-1 and IL1β. NLRP3, similar to NOD2, is a CATERPILLER gene, in that it has both NOD and leucine-rich repeat domains. Gain-of-function mutations in the NOD domain of NLRP3 cause Muckle–Wells syndrome, familial cold autoinflammatory syndrome and neonatal-onset multisystem inflammatory disease, three hereditary periodic fever syndromes.[73,74] Villani and colleagues demonstrated association of NLRP3 with CD using a tagging SNP candidate gene approach.[75] A total of six SNPs were associated with CD; the lead SNP reaching genome-wide levels of significance.

The effect of the six CD-associated mutations on NLRP3 expression was analyzed in PBMCs.[75] The rs4353135 genotype was significantly associated with altered expression, with homozygosity at the risk allele conferring low expression levels. In a separate assay in cultured monocytes, homozygosity of the risk allele at a different SNP (rs6672995) was associated with decreased levels of IL1β following stimulation with lipopolysaccharide (LPS) (p = 0.0059). Finally, significant elevation of NLRP3 expression was demonstrated in human CD biopsies (fold change: 4.08; p < 0.0028) and acute trinitrobenzene sulfonic acid-induced colitis (fold change: 9.38; p < 0.0009).
Autophagy

Perhaps the most exciting and novel insight to arise from gene discovery in CD is that defective autophagy may play a critical role in disease pathogenesis. Two autophagy genes are now known to be associated with CD: ATG16L1 and IRGM.

Autophagy, the process by which cells digest parts of their own cytoplasm for removal, functions as a homeostatic 'house-keeping' mechanism, a pathogenic process in a variety of cancers and neurodegenerative disorders and as an innate immune mechanism against intracellular bacteria.[76] It is a degradative, membrane-based system by which long-lived intracellular products are removed and turned over via the lysosome. Autophagy is highly conserved; in a unicellular world, it probably evolved as an essential survival mechanism to provide an alternative energy source during periods of starvation, thus ensuring continued minimal cellular function. Most of the upstream regulation of autophagy occurs directly through the mTOR. The autophagy-related proteins control the execution of autophagy, as outlined in Figure 4.

Figure 4.

Mechanics of Autophagy.
The majority of upstream regulation of autophagy is through the mTOR and, subsequently, the autophagy-related proteins. Under starvation conditions, mTOR inhibition leads to activation of the autophage cascade to provide alternative energy source by recycling intracellular organelles. Intracellular organelles or pathogens are enclosed into an autophagosome. This forms from a double-layered membrane called a phagophore that elongates to surround the pathogen before closing in on itself. The formed autophagosome then fuses with a lysosome (forming an autophagolysosome). Lysosomal enzymes first cut through the inner membrane of the autophagosome and then break down the enclosed cellular contents or pathogen, finally releasing amino acids back into the cytoplasm for re-use.

 

[David in Seattle]

The application of autophagic machinery to intracellular bacteria (termed xenophagy) has been demonstrated in vitro for several pathogens, including group A Streptococcus, Mycobacterium tuberculosis, Shigella flexneri and Salmonella enterica, although direct in vivo evidence is presently lacking.

ATG16L1 The ATG16L1 gene was originally identified as a CD gene in a German nonsynonymous SNP GWAS and has subsequently been confirmed in UK, North American and French/Belgian GWASs, as well as in a handful of independent replication cohorts.[16,22–24,77,78] The results of logistic regression and haplotype analysis suggested that the CD risk from ATG16L1 was confined to the G allele of rs2241880 (Thr300Ala). It was noteworthy that the variant confers an amino acid change at the evolutionary conserved position 300 of the N terminus, located in exon 9, which is translated into the same reading frame of all six known splice variants of ATG16L1.

Hampe and colleagues demonstrated expression of ATG16L1 mRNA and protein in the colon, small intestine, intestinal epithelial cells and leukocytes. However, they did not detect any difference in protein expression in intestinal tissue between CD and healthy controls.[22] We have demonstrated downregulation of ATG16L1 mRNA in colonic CD biopsies compared with healthy controls in our large microarray dataset.[79]

Three different murine models of ATG16L1 deficiency have recently been described in two publications in Nature.[80,81] Cadwell and colleagues generated two mouse lines with gene-trap mutations in Atg16l1 (Atg6l1 HM1 and Atg16l1 HM2), where Atg16l1 was expressed at 23–37% of expected levels.[80] Intestinal autophagy was shown to be defective in these mice despite normal intestinal morphology. The transcriptional signature of Atg16l1 deficiency was specific to Paneth cells, with disruption of the granule exocytosis pathway (decreased lysozyme in mucus, aberrant disorganized granules and reduced granule number). Furthermore, CD patients with a homozygous Thr300Ala mutation had a strikingly concordant Paneth cell phenotype, with disorganized granules, reduced granule formation, diffuse cytoplasmic staining of lysozyme and positive staining for leptin.[80]

Saitoh and coworkers generated mice that lack the entire coiled-coil domain of Atg16l1.[81] As these Atg16l1 −/− mice died on the first postnatal day, a chimeric mouse was generated by transplanting Atg16l1 −/− fetal liver cells into lethally irradiated CD45.1+ mice. Under specific pathogen-free conditions, these chimeric mice did not develop spontaneous colitis. However, when stressed with DSS, the 7-day mortality rate was 100% (compared with 0% in WT mice), and large areas of ulceration with severe inflammation were noted in the distal colon. A robust increase in IL1β expression was demonstrated upon stimulation of Atg16l1 −/− macrophages with LPS and noninvasive Gram-negative bacteria (e.g., Escherichia coli) but not with invasive strains (e.g., Salmonella typhimurium).[81] Stimulation with TLR3, TLR4, TLR7 and TLR9 ligands (but not TLR2 or TLR5) also induced IL1β production. Together, these data indicate that reductions in basal autophagy induce IL1β overproduction. The relevance of this was demonstrated in vivo as injection of neutralizing anti-IL1β antibodies significantly improved mortality rate and weight loss in the DSS-stressed chimeric animals.

Two observations suggest that the phenotypes of ATG16L1 and NOD2 mutation are distinct. First, Atg16l1 −/− macrophages have a normal inflammatory cytokine response to MDP.[81] Second, Atg16l1 HM animals have no defect in handling oral Listeria monocytogenes challenge (in contrast to NOD2 −/− animals).[80]

IRGM IRGM was identified as a CD gene in the Wellcome Trust Case–Control Consortium (WTCCC) GWAS (Box 2). Two nonfunctional SNPs (rs13361189 and rs4958847) flanking IRGM on chromosome 5q33.1 were strongly associated with CD. Resequencing of the major IRGM exon failed to identify the causative mutation. Subsequently, a 20-kb deletion was identified immediately upstream of IRGM and in perfect linkage disequilibrium with rs13361189.[82] The frequency of the deletion in a North American reference population was 10%, compared with 15% in CD patients (OR: 1.6; p < 0.01) and 14% in UC (OR: 1.4; p < 0.05). It is intriguing that the deletion was associated with UC in this study, given the absence of association noted for IRGM SNPs in previous studies.[19,20] McCarroll and colleagues went on to demonstrate in vitro that the CD risk (deletion) and protective (reference) haplotypes differentially activate IRGM expression in distinctive cellular contexts.[82]

The immunity-related guanosine triphosphatase (IRG or p47 GTPase) genes are critical to innate immunity against intracellular pathogens.[83] While there are 23 complete IRG genes in mice, only three have been identified in man.[83,84] Of these, IRGM is the human homologue of murine IRGM1 (Lrg-47), a key mediator of IFNγ-induced autophagy.[85] Human IRGM is constitutively expressed and contains no IFN-inducible elements in the promoter, leading to initial reports that humans lacked a p47 resistance system.[84] However, despite being unresponsive to IFNγ, IRGM plays a critical role in IFNγ-induced or conventionally induced autophagy in human macrophages.[85] siRNA knockdown of IRGM in human macrophages infected with mycobacteria leads to defective autophagy, with increased bacterial survival compared with control cells.[85] Lrg-47 knockout mice (Lrg-47 −/−) fail to control Mycobacterium tuberculosis infection (succumbing after several weeks following aerosol or intravenous inoculation), despite the formation of well-organized granulomas replete with infiltrating lympocytes.[86] Lrg-47 −/− macrophages have defective bacterial killing, with impaired maturation of M. tuberculosis-containing phagosomes; Lrg-47 is recruited to these vesicles in WT cells.[85] Similarly, Lrg-47 −/−mice infected with Mycobacterium avium are unable to control bacterial replication; although surviving the acute illness, they succumb 11–16 weeks later.[87] The phenotype of these animals suggests a regulatory role for Lrg-47 on lymphocyte survival, as they have a profound systemic anemia and lymphopenia, with marked lymphocyte deficiency in mycobacterium granulomas.[87]

 

[David in Seattle]

Developmental Genes

Another theme emerging from genetic studies is that signaling pathways critical to normal mammalian gut development are dysregulated in IBD pathogenesis. The key molecular pathways involved in the development of the GI tract are the hedgehog, BMP, Notch, WNT/β-catenin, hox, sox and eph/ephB signaling pathways.[88] Patterned gene expression within the endoderm and surrounding mesoderm regulates the morphogenesis, differentiation and boundaries of the developing gut.[89] This is a highly complex and tightly regulated process with multiple overlapping signaling pathways directing cell fate and tissue patterning. Gene discovery studies have demonstrated a role for NK2 transcription factor-related, locus 3 (NKX2.3) and glioma-associated oncogene homolog 1 (GLI1; a key transcriptional regulator of the hedgehog signaling pathway) in IBD.

NKX2.3 The NKX2.3 gene was identified as a CD susceptibility gene in the WTCCC study,[16,24] with early replication in a German tagging SNP study[20] and subsequent association with UC.[19,20] NKX2.3 is a member of the NKX family of homeodomain-containing transcription factors implicated in cell type specification, and maintenance of differentiation in a number of different tissue types. In mice, NKX2.3 is expressed in the gut mesoderm and branchial arches from E9.5, persisting in the hindgut until adulthood.[90] During later stages of murine development, expression is confined to the inner ring of the gut mesoderm (smooth muscle), immediately underlying the mucosal endodermal epithelium.[91] In the adult, highest levels of expression are noted in the ileum, with lowest levels of expression in the duodenum and rectum. In addition to smooth muscle expression, NKX2.3 is expressed in numerous mesenchymal cells in the intestinal lamina propria and in endothelial cells of small blood vessels.

A significant proportion of NKX2.3 −/− mice die from acute intestinal malabsorption within 2 weeks of birth.[91,92] In the early postnatal period, these animals have steatorrhea and bloody diarrhea, with lipid-staining vesicles predominant in the intestinal mucosa. While TNF-α expression was increased in one study of mutant intestine, no acute inflammation was detected at any postnatal stage.[91] Those mice that survive this period have reduced intestinal length and intestinal distension secondary to hyperplasia (not dilatation), but otherwise appear healthy as adults and are fertile.[91] Meanwhile, heterozygotes are viable and appear to develop normally.[91,92] The homozygous mutant animals have severe defects in both the spleen and intestine. Of note, one in three of these mice are asplenic; the spleens in the remaining animals are approximately tenfold smaller than WT animals and demonstrate severe structural abnormalities.[91,92] They lack a marginal zone, have abnormal segregation of B and T lymphocytes and a paucity of macrophages in the red pulp. They are noted to have a smaller cecum than WT mice, from E15.5 onwards.[91] In the small intestine, villus formation is delayed during development; however, normal differentiation of epithelial cell types is present in surviving adult mutants.[92] Furthermore, these animals have significant abnormalities of gut-associated lymphoid tissue, with fewer and smaller Peyer's patches and lymphoid aggregates than WT mice. Moreover, MadCAM-1 expression is significantly downregulated (~tenfold reduction) in areas that normally express NKX2.3, providing a potential mechanism by which they have deranged lymphocyte homing.[91] The importance of lymphocyte homing via MadCAM-1 in IBD pathogenesis has been clearly demonstrated as testified by the clinical efficacy of the anti-α4 monoclonal antibody, natalizumab, in CD.[93]

GLI1 We have shown that germ-line variation in GLI1 (within IBD2 on 12q13) is associated with IBD.[94] This was demonstrated in three independent northern European populations (Scotland, England and Sweden), but awaits further independent replication. A nonsynonymous SNP (rs2228226) in a highly conserved region of GLI1, next to a known transactivation domain, was associated with IBD and was functionally defective in vitro.[94] GLI1 is a major transcriptional regulator of the hedgehog signaling pathway. Hedgehog, through exclusively paracrine signaling (epithelium to mesenchyme),[95] plays critical roles in GI tract development, homeostasis and disease.[96] It is also centrally involved in chronic injury, inflammation and repair in several different organs including muscle,[97] liver[98,99] and lung.[100,101] Sonic hedgehog is critical for T-lymphocyte development,[102] myeloid cell maturation in the spleen[103] and peripheral human CD4+ T-cell activation,[104,105] and is a direct target of NF-κB.[106]

The GLI1 gene discovery has led to important potential insights into disease pathogenesis. GLI1 mRNA (and by inference hedgehog pathway activity) was shown to increase along the length of the healthy adult colon in man (mirroring the increasing luminal bacterial load in the distal colon) and was reduced in all forms of colonic inflammation studied (UC, colonic CD and non-IBD inflammation).[94] GLI1 −/− mice develop normally, in contrast to lines deficient in all other hedgehog components.[96,107] However, inflammatory challenge (3% DSS) to mice with a 50% reduction in functional GLI1 (GLI1 +/LacZ) led to early, severe colitis with substantial mortality and significant upregulation of IL12, IL17 and IL23 compared with WT littermates.[94] Furthermore, myeloid CD11b- and CD11c-positive cells were identified as direct targets of hedgehog signaling in the lamina propria following inflammatory challenge.[94] Subsequently, Zacharias and colleagues have demonstrated that downregulation of hedgehog signals in the small intestine (overexpression of the pan-hedgehog inhibitor HHIP under the villin promoter) results in robust, spontaneous ileal inflammation and IgA-positive dermatitis.[108] Thus, the body of evidence to date strongly suggests that epithelial hedgehog serves to dampen lamina propria immune responses in the steady state. This may provide options for therapeutic intervention given the intense biotechnology interest in small-molecule hedgehog agonists and antagonists that are in late preclinical and early clinical testing for a variety of different indications.[201]

Intracellular Tyrosine Phosphatases

PTPN2 The PTPN2 gene, identified as a CD susceptibility gene in the WTCCC study,[16,24] encodes the T-cell protein tyrosine phosphatase (TCPTP) that is a key negative regulator of inflammatory responses. TCPTP is located intracellularly, containing no transmembrane domains.[109] It is expressed in all tissues and at all stages of development. There are two splice variants in humans (48 kDa: emdoplasmic reticulum targeted; 45 kDa: nuclear and cytoplasm targeted) but only one is mice (45 kDa). Expression is ubiquitous, but highest in cells of hematopoietic origin; the TCPTP knock-out mouse dies of severe anemia by postnatal weeks 3–5.[110] Interestingly, this mouse, although apparently normal through embryonic development, develops a severe systemic inflammatory/wasting disease characterized by runting, hunched posture, diarrhea, splenomegaly (caused by macrophage infiltrates) and weight loss.[110] The severe immunosuppression is characterized by defective T-cell-dependent B-cell responses. TCPTP is an important regulator of colony stimulating factor (CSF)-1 and mononuclear phagocyte development (Tcptp −/− show increased granulocyte macrophage precursors).[111] Furthermore, the Tcptp −/− mice show widespread lymphocytic infiltrates in nonlymphoid tissues, correlating with increased IFNγ, TNF-α, IL12 and nitric oxide, and increased LPS sensitivity.[112] Increased cytokine production is detected as early as day 3 postpartum, preceding the onset of systemic disease (symptoms develop at 10–14 days postpartum), suggesting that this is a primary abnormality in these animals. IL12/IL23-p40 expression is massively upregulated in the liver and, to a lesser extent, in the spleen at day 3. Intraperitoneal injection of 4 µg LPS to 20-day-old Tcptp −/− mice led to the development of septic shock, not seen in WT littermates, with massive induction of serum IFNγ.

T-cell protein tyrosine phosphatase has been shown to dephosphorylate distinct substrates, including the insulin receptor,[113] EGF receptor (EGFR),[114] JAK1,[115] JAK3,[115] STAT1,[116] STAT3[117] and, most recently, STAT6.[118,119] As a result, it can regulate signaling pathways that are induced by various growth factors (e.g., EGF) and cytokines (e.g., TNF-α). In the case of TNF-α, MAPK signaling is suppressed by TCPTP (it interacts with TRAF2 to dephosphorylate and, hence, inactivate Src tyrosine kinases).[120] TCPTP-deficient cells show enhanced IL6 production in response to TNF-α, an effect blocked with inhibition of ERK1/2 activation.[120] TCPTP dephosphorylates nuclear STAT6 (PTPN1 does the same to cytoplasmic STAT6), inducing the expression of IL4 target genes.[118] The 45-kDa TCPTP recognises mutant EGFR (a common rearrangement with a truncated protein with an in-frame deletion of 267 amino acids that is detected in glioblastoma, breast, lung and prostate cancers) as a cellular substrate and dephosphorylates it, leading to decreased MAPK, ERK2 and PI-3K levels, suppressing the tumorigenicity of mutant EGFR-expressing glioblastoma cells.[121]

 

[David in Seattle]

PTPN22 Evidence for the association of the PTPN22 gene with CD was provided in the meta-analysis by Barrett et al.[17] It is of some considerable interest that the lead SNP (R620W) has been associated with a wide range of conditions with prominent humoral autoimmunity (including Type 1 diabetes mellitus, rheumatoid arthritis, autoimmune thyroid disease, myasthenia gravis, systemic sclerosis, vitiligo, Addison's disease and alopecia areata), yet in CD, this SNP is actually protective.[122] PTPN22 is expressed in many hematopoietic cell types, notably T cells. The 620W autoimmune risk allele behaves as a gain-of-function mutation as it results in a phosphatase with higher catalytic activity and more potent negative regulation of T-cell activation.[123,124] By contrast, knockout mice (Lyp is the mouse ortholog of PTPN22) have increased T-cell activation in combination with an increased production of antibodies.[125] The functional effects have yet to be studied in CD patients.
IL10

As many as ten historic studies (1999–2006) in multiple different populations have analyzed the IL10 promoter (at -592 [rs1800872], -819 [rs1800871] and -1082 [rs1800896]) for evidence of association with UC and CD, with very variable results. A German UC GWAS has recently provided definite evidence of association at IL10, but not with any of these 3′ promoter SNPs.[21] Their lead IL10 SNP was rs3024505, located 1 kb downstream of the 3′ untranslated region. Following an IL10 tagging SNP (n = 22) supplementary study across the various replication panels, two additional intronic SNPs were associated with UC. Although the three associated SNPs were in perfect linkage disequilibrium, regression analysis suggested they might act independently of each other. Of note, rs3024505 was also associated with German CD, albeit weakly. The location of this SNP – 79bp from a highly conserved stretch of DNA at the 3′ end of IL10 – is of some interest, at it contains a putative AP1 binding site.

Il10 −/− mice develop spontaneous colitis in specific pathogen-free conditions owing to defective counter-regulatory anticytokine responses. Indeed, these animals provide one of the oldest (first reported in 1993), most widely used and best characterized genetic animal models of colitis. Physiologically important defects in IL10 signaling in lamina propria mononuclear cells from UC patients were described in 1995,[126] and by 1998, very early clinical trials of subcutaneous recombinant IL10 produced some evidence of benefit in UC patients.[127] For some reason, the Phase II trials of this protein (rHuIL-10) were in CD and when they failed to demonstrate efficacy, this avenue was quietly dropped from further investigation.[128,129] However, given these new genetic data, further exploration of IL10 delivery should be explored as a therapeutic avenue in UC, perhaps utilizing Larry Steidler's ingenious delivery mechanism via recombinant IL10-producing Lactococcus lactis.[130]

Other Considerations/Implications from Gene Discovery
Genotype–Phenotype Associations

The common NOD2 mutations are associated with ileal disease and a fibro-stenosing phenotype.[131,132] The HLA region is associated with colonic IBD, and the DRB1*0103 allele with extensive disease, the need for surgery and extraintestinal manifestations. GWASs have, to date, largely defined association with IBD generally, with CD or UC specifically, or with early-onset IBD. Further than this, there has been a real paucity of consistent genotype–phenotype correlations. There are two possible explanations for this. Either the large GWAS cohorts have been inadequately phenotyped and/or they are underpowered for subanalysis. Certainly a large number of the CD genes/loci identified in the meta-analysis have yet to be subjected to formal genotype–phenotype association. Further collaborative studies designed to specifically address this issue are underway.
Ethnic Variations in IBD Gene Associations

Aside from Yamazaki's initial small-scale GWAS, all the large genetic experiments in recent years have been performed in Caucasian cohorts. As a result, ethnic variation has yet to be studied in much detail. However, the data presently available suggest this to be an avenue ripe for future study as clear differences in some genes have been identified not only between Asian and European populations but also within populations of European descent. NOD2 mutation carriage rates are seen to range widely (0–50%) and are highest in central Europe.[60,133] The contribution of NOD2 to disease susceptibility is relatively lower in northern European (Scottish[134,135] and Scandinavian[136,137]) populations, where the population-attributable risk ranges from 7.9 (early-onset Scottish CD) to 11.4% (Swedish CD). In Japanese,[138] Chinese[139] and South Korean[140] populations, these NOD2 mutations (Gly908Arg, Arg702Trp and Leu1007fsincC) are absent.

TNF-SF15 is the only IBD gene discovery originating in an Asian population to date (Box 3).[26] This was an early GWAS in a small-index cohort, suggesting that larger studies may yield additional disease genes that may be specific to those of Asian descent. The TNF-SF15 association was confirmed in the Caucasian CD meta-analysis[17] and was replicated in other Asian populations.[141] Several other confirmed IBD genes from European populations have been studied in Asian cohorts. Yamazaki and colleagues from Japan have demonstrated associations at NKX2.3, IL12B and ZNF-365, but not at IL23R, ATG16L1, 3p21, 5q23 (IBD5), IRGM or PTPN2.[142,143] The 5p13 gene desert was not polymorphic. The lack of association at IBD5 has been previously reported.[144] However, it is difficult to draw firm conclusions regarding the other loci reported by Yamazaki owing to lack of power (n = 484 CD). PTPN22 has not been directly studied in CD populations other than Caucasians; however, Asians rarely carry the 620W variant that is associated with autoimmune diseases but protective for CD.[122]
Gene–Environment Interactions

Barrett estimates that the currently identified susceptibility genes account for only approximately 20% of the genetic contribution to CD susceptibility.[17] One argument as to why this may be an underestimate is that it takes no account of gene–environment interactions, which probably act synergistically to increase disease risk. The environmental stimuli that lend themselves to study include cigarette smoke and commensal/pathogenic bacteria.

The study of the commensal flora, enteroadherent and intracellular bacteria in CD patients with varying genotypes (e.g., NOD2, ATG16L1 and IRGM) will be of great interest. Not only may this provide specific clues as to disease etiology, but it may serve to direct personalized therapy (e.g., antibiotics) depending on genotype. However, such studies are difficult as they require large cohorts of homogenous patients with specific genotypes early in disease course and preferably off treatment. In the meantime, modeling of gene–environment interactions in genetically modified animals has proved instructive, as best illustrated by the study of Il10 −/− mice and Hlα-B27 transgenic rats.[145] These animals develop colitis in specific pathogen-free conditions, but not when kept completely germ-free. Selective colonization with different bacterial species critically affects aspects of disease phenotype, notably severity and anatomical location. In Il10 −/− mice monoassociated with Enterococcus faecalis or E. coli, a progressive chronic colitis develops, although the regional distribution and kinetics of this colitis varies with the bacteria. Mice monoassociated with Pseudomonas fluorescens or Bacteroides vulgatus do not develop colitis. By stark contrast, Hlα-B27 rats monoassociated with B. vulgatus develop an aggressive colitis, but show no inflammation with E. coli. Clearly, the interaction of both genetic factors (Il10 −/− or Hlα-B27) and environmental factors (E. coli or B. vulgatus) is fundamentally important for the establishment of colitis in these animal models. It is highly probable that a similar phenomenon occurs in human disease, but this awaits formal testing.

 

[David in Seattle]

Expert Commentary & Five-year View

The major evolving paradigm for complex disease genetics is that the biological insights gained (and subsequent abundance of novel therapeutic targets) will turn out to have far greater importance than the relatively small contribution to disease risk conferred by each individual mutation or locus. We speculate that in 5 years' time, the genetic architecture of IBD will be completely described, with around 200 disease genes/variants accounting for the genetic contribution to disease susceptibility. At this stage, an 'IBD chip' will become a realistic possibility that may enable physicians to better diagnose patients, subcategorize (e.g., IBD1, IBD2 and IBD3) and personalize treatment accordingly. However, these data will not be used in isolation; rather, they will be incorporated into existing and future clinical models of disease risk. We will discuss the short- and long-term future of genetic studies and the implications for therapeutics and modeling of disease natural history.
Future of Genetic Studies

There are several explanations as to why the known IBD disease genes/loci only explain a fraction of the observed familial aggregation.[17,146] First, the present GWAS provide only relatively limited surveys of potential sequence variation and provide little or no information on rare alleles. Second, many 'hits' from the index studies are surrogate markers for the true disease causative mutations that may confer greater risk. Third, the risk at some loci will be attributed to multiple, independent mutations. In addition, very few studies into copy number variants have been performed on a genome-wide scale; it remains to be seen whether these will add significantly to the genetic risk. For several or all of these reasons, we are probably currently underestimating the known genetic risk already accounted for. However, the shortfall that remains is (and almost certainly should be) presently limiting the early application of genetics to determining individual disease risk.

The immediate future involves more GWASs in different subphenotypes, meta-analyses of existing individual GWASs (UC and UC plus CD), fine-mapping of loci (a major ongoing phase of the WTCCC follow-up studies, with CD prioritised for deep resequencing of associated loci) and detailed molecular and functional studies to truly understand the biology of various disease-associated mutations (Figure 2).
Implications for Therapeutics

There is an urgent, unmet therapeutic need in IBD, as many of the current therapies are limited by poor efficacy, unacceptable toxicity and poor patient acceptance. Several of the pathogenic insights arising from gene discovery described previously lend themselves to potential therapeutic intervention. For example, the association of multiple IL23-signaling components with IBD strongly suggests that this pathway should be prioritized for drug discovery, not least since proof of concept has already been provided in animal models. These data make the critical re-evaluation of early clinical studies using anti-p40 monoclonal antibody therapy in CD of paramount importance.[147] Anti-p19 monoclonal antibodies that are likely to provide a more specific anti-inflammatory intervention in IBD are currently in development.

Manipulating levels of intestinal autophagy in CD is an intriguing therapeutic option as several drugs targeting mTOR (and therefore increasing autophagy) are already in clinical use. Sirolimus is frequently used as an immunosuppressant post-solid organ transplantation, and sirolimus-eluting coronary artery stents are notable for their low restenosis rates. A recent single case report has described the clinical use of sirolimus in a patient with severe CD, resulting in a marked and sustained improvement in symptoms and endoscopic appearances.[148] In animal models, rapamycin has been demonstrated to be as effective as cyclosporine in reducing experimental chronic colitis induced by DSS.[149] These observations led to the prospect of early clinical trials of these agents to induce autophagy in patients with CD. However, owing to the wide-ranging effects of autophagy in both homeostatic and pathogenic roles in a variety of organs and tissues, not to mention in the GI tract itself, it is likely that more specific targeting of defined autophagy pathways will be required and delivery mechanisms will need to be developed to limit unwanted or dangerous systemic side effects.

In addition, the inclusion of GWAS technology to present IBD therapeutics (e.g., thiopurines and anti-TNF agents) may elucidate genetic factors that predict both clinical response and adverse events. The latter concept has recently been successfully tested in flucloxacillin-induced liver injury, effectively providing proof-of-principle to this approach.[150] Similar success in IBD could rapidly bring pharmacogenetics further into the clinical arena, mirroring the limited success of TPMT genotyping (or measurement of enzymatic activity) presently adopted in many centers prior to commencing azathioprine therapy.[151]
Implications for Natural History

In CD, physicians have lately taken a lead from rheumatologists and have been exploring the use of an aggressive policy of early combined immunosuppression (thiopurine plus an anti-TNF agent) in an attempt to modify the natural history of IBD (i.e., limit progression from inflammatory changes to stricturing or penetrating phenotype).[152] This policy has shown early promise, but adoption into clinical practice will remain limited until we are able to accurately determine which patients will run an aggressive disease course. At least 30% of patients with CD will have a fairly mild disease course; exposing these patients early to toxic therapies is not likely to prove an acceptable option.[153]

Present measures to predict an aggressive disease course are relatively crude and include: early age of disease onset, cigarette smoking and extensive disease. It is widely anticipated that a combination of prospectively and accurately phenotyped patients, DNA, serum, stool and/or intestinal biopsies will allow a detailed accurate predictive model based on clinical data and biomarkers. This reality is at least 5 years away. However, it is in this context that a future 'IBD chip' may revolutionize disease risk modeling and allow IBD genetics to enter clinical practice after over 20 years of promise.

 

[karim]

David-

WOW! I don't have time to read all of this now but my life is full of sleepless nights when I get as much reading in as possible. Thanks for the info and will be sure to peruse it all soon!

Kari

 

[David in Seattle]

David-

WOW! I don't have time to read all of this now but my life is full of sleepless nights when I get as much reading in as possible. Thanks for the info and will be sure to peruse it all soon!

Kari

Took awhile to post, too! :ylol: You're welcome and I hope you find it informative, Kari!

 

[Mark in Seattle]

I wish someone would have given me some guidance when I was young to go into the field of immunology. Such fascinating detective work putting all of these complex pieces together. I so love these complex diagrams - very cool. And thanks David for posting the whole article, diagrams and all!

I couldn't help but email Dr. Strober, since this is the first I've heard of gene therapy being worked on for Crohn's Disease. I inquired with him about a timeline for gene therapy for humans. I read somewhere that he's working on a murine model now. Anyhow, he said he's hoping they'll have gene therapy within one decade or hopefully sooner. I've survived the last 13 years since my last surgery, so if I (and all other Crohn's sufferers) can be so lucky for one more decade, maybe we'll have something to look forward to. I really had no idea that the genetic approach was making such progress of late. I hope it's not just pie in the sky.

 

[Misty-Eyed]

Immunology is such a headache though. It's so hard to understand, for me anyway. It's just kind of ironic that it's all linked in with genetics in Crohn's which is much easier for me to understand. lol

David, did they not publish it in pdf format?

 

[DustyKat]

Well thanks again David!!! I'm a little late to the party but hey I eventually arrived. :ylol:

Over the years I have thought about Roo and her CD and I believed that genetics definitely played a role in it's development but that role was shared by a number of other factors. Now that Matt has also been diagnosed I don't dismiss those other factors but genetics well and truly has the stand out leading role for me, even more so as it would appear that Matt is mirroring Roo.

Dusty.

 

[AndiGirl]

Thank you David. My family has its share of those with CD. I'm sure we have some of those genes described.