SEARCH

SEARCH BY CITATION

Keywords:

  • Toll-like receptor;
  • TLR4;
  • NOD2;
  • inflammatory bowel disease;
  • Crohn's disease

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Background: The first major Crohn's disease (CD) susceptibility gene, NOD2, implicates the innate intestinal immune system and other pattern recognition receptors in the pathogenesis of this chronic, debilitating disorder. These include the Toll-like receptors, specifically TLR4 and TLR5. A variant in the TLR4 gene (A299G) has demonstrated variable association with CD. We aimed to investigate the relationship between TLR4 A299G and TLR5 N392ST, and an Australian inflammatory bowel disease cohort, and to explore the strength of association between TLR4 A299G and CD using global meta-analysis.

Methods: Cases (CD = 619, ulcerative colitis = 300) and controls (n = 360) were genotyped for TLR4 A299G, TLR5 N392ST, and the 4 major NOD2 mutations. Data were interrogated for case-control analysis prior to and after stratification by NOD2 genotype. Genotype–phenotype relationships were also sought. Meta-analysis was conducted via RevMan.

Results: The TLR4 A299G variant allele showed a significant association with CD compared to controls (P = 0.04) and a novel NOD2 haplotype was identified which strengthened this (P = 0.003). Furthermore, we identified that TLR4 A299G was associated with CD limited to the colon (P = 0.02). In the presence of the novel NOD2 haplotype, TLR4 A299G was more strongly associated with colonic disease (P < 0.001) and nonstricturing disease (P = 0.009). A meta-analysis of 11 CD cohorts identified a 1.5-fold increase in risk for the variant TLR4 A299G allele (P < 0.00001).

Conclusions:TLR 4 A299G appears to be a significant risk factor for CD, in particular colonic, nonstricturing disease. Furthermore, we identified a novel NOD2 haplotype that strengthens the relationship between TLR4 A299G and these phenotypes.

(Inflamm Bowel Dis 2007)

The identification of the first major susceptibility gene for Crohn's disease (CD) encoding NOD2 (CARD15)1, 2 confirmed the critical role of enteric bacteria and their interaction with the intestinal innate mucosal immune system in the pathogenesis of this disorder. The mutant leucine rich repeat (LRR) domain of the intracellular NOD2 protein is unable to recognize bacterial muramyl dipeptide and fails to trigger a protective nuclear factor-kappaB (NF-κB) response.3 In intestinal epithelial cells the disease-associated variant is unable to constrain bacterial survival.4

Recognition of the key role of innate mucosal immunity in CD implicates a potential role for other germline encoded pattern recognition receptors (PRRs). The various cell surface Toll-like receptors (TLRs) recognize different pathogen-associated molecular patterns (PAMPs) shared by many pathogens but not expressed by the host.5 TLR4 is a major transducer of bacteria-derived lipopolysaccharide (LPS) specifically binding to LPS-lipid A, and is a first-line receptor in host defense against enteric Gram-negative bacterial invasion.6, 7 In addition to this, TLR4 expression is strongly up regulated in both CD and ulcerative colitis (UC) in inflamed and noninflamed mucosa.7 Franchimont et al6 initially reported an association between the TLR4 Asp299gly (A299G) polymorphism (rs4986790) and CD. Transfection experiments have demonstrated that this mutation affecting the extracellular domain of TLR4 interrupts LPS signaling, blunting the NF-κB response required for inflammatory cytokine production.8 In contrast to the mutations in NOD2, the TLR4 variant was not shown to influence CD phenotype.6

In total at the time of this submission, at least 8 groups have reported allele frequency data for TLR4 A299G in CD cohorts compared to control subjects. All groups were of British or European origin.6, 9–15 Confirmation of the original association has not been universal, creating doubt as to the contribution of the mutation in CD and highlighting the heterogeneous nature of the disease and the complexity of genetic susceptibility in polygenic diseases.

TLR5 recognizes bacterial flagellin, a principal component of bacterial flagella present on many motile bacteria in the gut, e.g., Listeria monocytogenes, Yersinia enterocolitica. TLR5 is expressed in monocytes, immature dendritic cells, and epithelial cells.16 Binding of TLR5 by flagellin again results in potent NF-κB activation and subsequently transcription of other proinflammatory cytokines in intestinal epithelial cells.17 A cytosine-thymidine transition at basepair 1174 (C1174T) in the TLR5 gene converts the arginine at amino acid 392 to a stop codon (TLR5 N392ST) and prematurely truncates the protein in its extracellular domain. Transfection experiments have demonstrated the inability of this variant to respond to flagellin.18

The importance of TLR5 in the pathogenesis of CD has been suggested by the identification of bacterial flagellins as a dominant disease antigen by serological expression cloning.19 Marked reactivity against a family of related flagellins was seen by both T cells and IgG2a in a subset of CD patients. This specific molecular response in patients with CD suggests an important link between the innate and adaptive immune responses in the gut and perhaps an impaired regulatory response to these antigens.19

In this study we sought to replicate the previously published association between TLR4 A299G and CD in a large Australian inflammatory bowel disease (IBD) cohort and to further explore the strength of this association using a global meta-analysis. We also sought to determine the role of TLR4 A299G in the phenotypic expression of CD in our population. In addition, we aimed to identify a novel association between TLR5 N392ST and CD.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Subjects

This candidate gene association study of case control design involved IBD patients (CD = 619, UC = 300) recruited from the IBD clinical and research program at the Royal Brisbane and Women's Hospital (RBWH), Brisbane, Australia. Informed consent was obtained from all patients and the RBWH Human Research Ethics Committee approved the study protocol. The patients are a consecutive series of cases prospectively recruited into a longitudinal IBD research protocol from 1994, from the population of southeast Queensland, Australia. All cases are recorded on an IBD database together with relevant phenotypic information including disease distribution, behavior and duration, surgery, smoking, and histological data such as granulomas. Phenotypic characteristics for CD cases were carefully determined by 2 investigators using clinical, endoscopic, radiological, histological, and surgical data available in the patient's clinical record. The investigators were blinded to the patient's genotype during this process. CD phenotyping was based on the Montreal Classification20 but with disease behavior reclassified as stricturing or nonstricturing independent of the presence of penetrating disease.

Controls (n = 360) were healthy Caucasian subjects recruited from the east coast of Australia over a 4-year period through the Genomics Research Centre (Griffith University, Queensland, Australia).

Genotyping

Genomic DNA was extracted from peripheral blood mononuclear cells using the salting-out technique.21 Both TLR polymorphisms were detected using a polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) technique. The PCR reaction comprised Platinum PCR Supermix (Invitrogen Australia, Mount Waverley, Victoria, Australia) and 2 primers specific for either TLR4 A299G (forward: 5′ GAT TAG CAT ACT TAG ACT ACT ACC TCC ATG 3′; reverse: 5′ GAT CAA CTT CTG AAA AAG CAT TCC CAC 3′)22 or TLR5 N392ST (forward: 5′ GGT AGC CTA CAT TGA TTT GC 3′; reverse: 5′ GAG AAT CTG GAG ATG AGG TAC CCG 3′).18 Following PCR, the products were digested using Nco 1 (TLR4 A299G) or Dde 1 (TLR5 N392ST) (New England Biolabs, Genesearch, Arundel, Queensland, Australia). In both assays the enzyme cuts the PCR product when the variant allele is present, while the wildtype DNA sequence is not cut, allowing the 2 alleles to be identified on an agarose gel (3.0%, TLR4; 2.0%, TLR5) using ethidium bromide for visualization.

Genotyping of the common disease-predisposing NOD2 variants (R702W, G908R, and L1007fsinsC), and a marker of the NOD2 haplotype (IVS8 = 158), P268S, was performed by high-resolution melt curve analysis on a Rotor-Gene 6000 (Corbett Life Science, Sydney, New South Wales, Australia). All samples were amplified in duplicate and each run included DNA from sequence-verified wildtype, heterozygous, and homozygous genotypes for each of the variants. High-resolution melt curve data were analyzed using the manufacturer's software.

Statistical Analysis

All data were in Hardy–Weinberg Equilibrium (as calculated via Haploview). Genotype and allele frequencies were compared between cases and controls using the chi-square test in SPSS (v. 15, Chicago, IL). Following the dominant genetic model, all results are presented for the presence of the TLR4 allele and the allelic relationship with disease. The CD case group was stratified according to NOD2 status and CD distribution to identify NOD2 and location-specific associations with the TLR4 variant. Haplotype identification using an unphased method was conducted via the Haploview software (Broad Institute of MIT and Harvard, Cambridge, MA). Meta-analyses were conducted via Revman (Review Manager [RevMan, computer program] v. 4.2 for Windows, Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2003).

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

TLR 4 A299G and TLR5 N392ST Case-control Analysis

Of 619 unrelated CD cases, 86 (13.9%) carried at least 1 variant TLR4 allele compared to 33/360 (9.2%) controls (P = 0.04). The overall minor allele frequency (MAF) of TLR4 A299G was higher in CD patients compared to controls (0.07 versus 0.05; odds ratio [OR] = 1.53, 95% confidence intervals [CI]: 1.02–2.29; P = 0.04). The prevalence of genotypes containing the minor TLR5 392ST allele was not significantly higher in the CD population compared to controls (0.12 versus 0.10), and the overall MAF was also not significantly different between the groups. Interestingly, when data for both the TLR4 A299G variant and TLR5 N392ST variant were compared, we identified that the TLR4 A299G variant predominantly occurs in a TLR5 N392ST wildtype background (P = 0.04). Case-control comparisons were made for both TLR4 and TLR5 mutations in UC cases. We found no evidence that the variant TLR4 or TLR5 alleles were either over- or underrepresented in the UC group as compared to controls.

Relationship between TLR4 A299G and NOD2

To elucidate the nature of the relationship between the TLR4 variant and the 4 NOD2 variants (P268S, SNP5; R702W, SNP8; G908R, SNP12; L1007fsinsC, SNP13), case-control analysis was conducted, while the CD case group was stratified by NOD2 mutation. We identified that the relationship between the TLR4 mutation and CD was stronger in the presence of the P268S variant allele (P = 0.01) (wildtype: CCC, variant: TCC), and similarly, stronger in the absence of the R702W variant allele (P = 0.02) (wildtype: CGG, variant: TGG). Hence, a NOD2 haplotype was constructed (only those patients with a variant P268S allele and a wildtype R702W allele; TC). In the presence of this novel haplotype, we then compared the TLR4 genotypes between CD patients and controls. This analysis showed a strong association between the TLR4 variant allele in the presence of the NOD2 haplotype and CD (P = 0.003) (Table 1). Similar to R702W, the association between TLR4 A299G and CD was also stronger in the absence of the G908R and L1007fsinsC mutations (P = 0.035 and 0.049, respectively).

Table 1. TLR4 A299G Associations in CD Population Stratified by NOD2
NOD2 Stratified GroupNTLR4 MAFaP-value
  • a

    MAF, minor allele frequency.

  • b

    TC corresponds to P268S T allele and R702W C allele NB: 33 patients were missing NOD2 genotype who had TLR4 genotype.

All participants6190.070.04
 NOD2 P268S WT2820.060.19
 NOD2 P268S variant3090.080.01
 NOD2 R702W WT4920.080.02
 NOD2 R702W variant990.060.50
 NOD2 TCb haplotype2120.090.003

Phenotype–genotype Analysis for TLR4 A299G and CD Cases

Analysis of association between the TLR4 variant and the major clinical (phenotype) characteristics of the Australian CD population (Table 2) identified several interesting relationships. While the TLR4 A299G variant was not significantly associated with age at diagnosis, in the presence of the NOD2 TC haplotype a relationship became apparent between the minor allele and later age of onset (P = 0.02). Furthermore, the presence of the TLR4 A299G variant in CD was associated with colonic localization of the disease (P = 0.04) (Table 3), and this relationship was markedly stronger in the presence of the NOD2 haplotype (P < 0.001). In addition, the association specifically between the TLR4 variant and nonstricturing disease, and among patients not requiring CD surgery, was stronger in the presence of the NOD2 haplotype (P = 0.009 and 0.02, respectively).

Table 2. Phenotypic Characteristics of the IBD Cohort
CharacteristicNumber
(A) Crohn's disease454
Sex: male/female176/278
Age at diagnosis 
 <1529
 17–40353
 >4072
Disease location 
 Ileal207
 Ileal-colon149
 Colon93
 Upper GI5
Disease behavior 
 Stricturing196
 Not stricturing258
Granuloma 
 Yes282
 No172
(B) Ulcerative colitis436
Sex: male/female208/228
Age at diagnosis 
 <1591
 17–40256
 >4089
Disease extent 
 Left-sided220
 Extensive216
Colectomy 
 Yes70
 No366
Table 3. TLR4 A299G Associations with CD Phenotype
Clinical Characteristic LocationTC HaplotypeAll Cases
NaTLR4 MAFbP-valueNTLR4 MAFbP-value
  • a

    Sum totals differ due to missing phenotype data in subgroups.

  • b

    Control MAF used for comparison; 0.05.

 Ileal770.080.132180.060.42
 Ileocolonic610.050.981550.070.15
 Colonic290.17<0.001850.090.02
Stricturing      
 No850.100.0092410.080.06
 Yes650.060.541820.060.5
Surgery      
 No410.110.021270.110.001
 Yes1280.080.123400.050.63

TLR4 A299G Meta-analysis

We reviewed 8 published articles6, 9–15 (1 with 2 separate CD populations6), and 1 unpublished (V. Annese, pers. commun; 2004–2007) studying TLR4 A299G, and compared these with our data (Fig. 1). Six out of 11 CD populations showed a significant increase in risk with an overrepresentation of variant alleles in the CD population. The overall OR showed a significant increase in risk of disease (OR = 1.47, 95% CI: 1.25–1.73; P < 0.00001) in the presence of TLR4 299G.

thumbnail image

Figure 1. Crohn's disease TLR4 A299G meta-analysis.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The key role of the innate mucosal immune response to enteric microorganisms in the pathogenesis of IBD and particularly CD is well recognized. The widely replicated association between 3 major NOD2 variants (SNPs 8, 12, 13) and CD implicates a potential role for other germline-encoded PRRs, such as the Toll-like receptors (TLRs). Evidence has suggested a role for both TLR4 and TLR5 in IBD,6–8, 17–19 with the TLR4 A299G polymorphism implicated in both CD and UC.6 In this study we replicated a previously reported association between the TLR4 A299G polymorphism and CD6 but not UC. We also identified a novel interaction between this variant and a specific NOD2 haplotype in CD patients. However, we could not demonstrate any association between TLR5 N392ST and IBD in our cohort.

Analysis involving TLR4 A299G and CD, stratified by each of the 4 major NOD2 SNPs, identified a significant overrepresentation of the variant TLR4 allele in the CD population, which was strengthened in the presence of NOD2 P268S and in the absence of the 3 major CD variants (R702W, G908R, L1007fsinsC). These results suggest that the effect of TLR4 A299G in CD occurs independently of the major functional NOD2 variants. The presence of P268S in addition to TLR4 299G seems to increase susceptibility to CD. NOD2 P268S has been shown to have no effect on NF-κB expression in HEK293T cells transfected with NOD2 P2685 plasmids in the presence or absence of bacterial components LPS and PGN.23 Hence, the presence of NOD2 P268S alone does not provide a functional explanation for the haplotype results. Its potential linkage to the functional variants R702W, G908R, L1007fsinsC would also not appear to be responsible, as the relationship between TLR4 A299G and CD is only maintained in their absence. Hence, the linkage of NOD2 P268S to another unrecognized functional polymorphism, possibly in another gene in the IBD1 region, may be responsible.

The analysis of the relationship between the TLR4 A299G variant and the phenotypic characteristics of this large CD cohort yielded interesting results. Ouburg et al15 previously reported a relationship between TLR4 A299G and colonic CD in a small population of 112 Dutch CD cases (P = 0.0017; OR = 5.5, 95% CI: 1.9–15.4). Our study confirmed this association and demonstrated that it was strongest in the presence of the novel NOD2 haplotype (Table 3). This finding is also significant in that it represents the first independent confirmation of association between a CD susceptibility gene and colonic disease location.

Studies undertaken in mouse ileal Paneth cells show no expression of TLR4 mRNA, supporting the hypothesis that mouse Paneth cells respond to bacteria by TLR4-independent mechanisms. This was further substantiated by studying Paneth cell secretion in TLR4-deficient mice that also appeared to be independent of TLR4.24 Intestinal epithelial cells from human small intestinal and colonic normal mucosa have been demonstrated to constitutively express TLR5, whereas TLR4 is barely detectable. However, TLR4 is strongly upregulated in both CD and UC in inflamed and noninflamed mucosa.7, 17 These studies support the association between TLR4 A299G and colonic CD, as Paneth cells, key cells in ileal antimicrobial defense, are able to respond to luminal bacteria independent of TLR4 status.

Meta-analysis using studies from Scotland, Belgium, Greece, Hungary, Netherlands, Italy, and Australia found a significant increase in risk of CD from an overrepresentation of variant TLR4 A299G alleles in the CD group. A comparison of sample size from all populations found our cohort to be the largest, contributing the most toward the final CD meta-analysis. Hence, we believe the Australian CD cohort to be a good representation of a global CD population, and that the risk of disease from the TLR4 A299G variant is a true representation of disease risk.

This study demonstrated no association between TLR5 N392ST and CD. TLR5 confers responsiveness to bacterial flagellin. The common stop codon polymorphism renders the truncated ligand-binding domain unable to mediate flagellin signaling. Peripheral blood mononuclear cells from individuals heterozygous for TLR5 N392ST demonstrate significantly decreased IL-6 production when stimulated with flagellin.18 One proposed model implicates an impaired T regulatory response to specific flagellins in the aberrant mucosal immune response and chronic intestinal inflammation in CD.19 This study suggests that the TLR5 N392ST polymorphism is not the mechanism by which this occurs.

It is likely that the dysfunction of TLR4 in CD contributes to the aberrant mucosal immune response to the luminal bacteria. The TLR4 A299G polymorphism renders the protein less responsive to LPS signaling, blunting the NF-κB response. This may limit the hosts' ability to constrain bacterial survival and predispose to chronic intestinal inflammation, as postulated in the case of the NOD2 variants. However, the reduction in inflammatory cytokine production may provide some protection against the more severe complications of chronic intestinal inflammation and explain the lower requirement for surgery in our CD cases carrying the mutant TLR4 A299G allele. In addition, the association of TLR4 A299G with colonic CD and nonstricturing disease may contribute to this finding. These findings are also supported by the lack of association of TLR4 A299G with the 3 major functional NOD2 variants, which are themselves associated with ileal location and stricturing behavior.

This study has presented both an independent contribution to the research community for TLR4 and NOD2 and a collaboration of results from the meta-analysis of the TLR4 data. We recognize that the numbers for the novel NOD2 haplotype were low, and encourage this to be repeated in other populations. As a direct candidate gene association study, specifically testing a candidate SNP that has been previously associated with disease, we chose not to correct for multiple comparisons. We recognize that if our aim was to test multiple common polymorphisms from the population, testing against a more conservative alpha would be appropriate. However, the aim of this study was not to identify new variants that exist in the common population, but to test 2 specific mutations (TLR4 A299G and TLR5 N392ST) and risk of disease in an Australian IBD population, as well as their relationship with the CD susceptibility gene NOD2. In conclusion, this study provided a positive replication of the CD-TLR4 association, and has identified an increased risk of CD for those patients with the novel NOD2 haplotype TC.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors thank all the participants taking part in this research, the Royal Brisbane and Women's Hospital and Mater Adult Hospital nurses, and especially Robert Smith, Susan Mason, Dian Templeton, and Rebecca Cooley for help with sample and data collection. We thank Madeleine Kersting for graphics assistance. The authors declare no conflict of interests.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES