Article Text

Extended report
Genetics of toll like receptor 9 in ANCA associated vasculitides
  1. C A Husmann1,
  2. J U Holle2,
  3. F Moosig2,
  4. S Mueller1,3,
  5. B Wilde4,
  6. J W Cohen Tervaert4,
  7. L Harper5,
  8. G Assmann6,
  9. W L Gross2,
  10. J T Epplen1,
  11. S Wieczorek1
  1. 1Department of Human Genetics, Ruhr University, Bochum, Germany
  2. 2Department of Rheumatology, University of Luebeck, Vasculitis Center UKSH and Klinikum Bad Bramstedt, Lübeck, Germany
  3. 3Department of Dermatology, University Hospital Duesseldorf, Duesseldorf, Germany
  4. 4Department of Internal Medicine, Division of Clinical and Experimental Immunology and Laboratory of Clinical Immunology, Maastricht University Medical Center, Maastricht, The Netherlands
  5. 5School of Immunity and Infection, Centre for Translational Inflammation Research, University of Birmingham Research Laboratories, Queen Elizabeth Hospital Birmingham, UK
  6. 6Department of Rheumatology, Universitätsklinikum des Saarlandes, Homburg/Saar, Germany
  1. Correspondence to Dr Stefan Wieczorek, Department of Human Genetics, Ruhr University, Bochum 44780, Germany; stefan.wieczorek{at}


Objectives To investigate the contribution of genetic polymorphisms of toll like receptor (TLR) 9 and related genes on the susceptibility and clinical manifestation of anti-neutrophil cytoplasmic antibody (ANCA) associated vasculitides (AAV).

Methods Four single nucleotide polymorphisms (SNPs) in TLR9 were genotyped in 863 German AAV cases and 1344 healthy controls. Significant results were replicated in a cohort of 426 Dutch and British AAV cases. 11 polymorphisms in TLR9 related genes were studied concomitantly.

Results A strong association of TLR9 genotypes and haplotypes with granulomatosis with polyangiitis was observed as well as a contrariwise association with microscopic polyangiitis. The association was confirmed when cases were compared according to ANCA status rather than to clinical entity. This was partly replicated in the second cohort leading to a striking overall difference in TLR9 allele/haplotype frequencies between proteinase 3 (PR3) ANCA+ and myeloperoxidase (MPO) ANCA+ cases (p=0.00000398, pc=0.000016, OR 1.68 (95% CI 1.35 to 2.1) for rs352140; p=0.000011, pc=0.000044, OR 1.64 (95% CI 1.31 to 2.04) for a 3-SNP haplotype). No significant association or epistatic effect was detected for TLR9 related genes: interleukin 6, interleukin 23 receptor, myeloid differentiation primary response 88, TNF receptor-associated factor 6, interleukin-1 receptor-associated kinase 4, discs large homolog 5 and nucleotide-binding oligomerisation domain containing 2.

Conclusions We provide further evidence that PR3-ANCA+ AAV differs genetically from MPO-ANCA+ AAV. TLR9 signalling may be involved in disease pathology, favouring models of infectious agents triggering AAV development.

  • Granulomatosis with polyangiitis
  • Autoantibodies
  • Gene Polymorphism

Statistics from


Anti-neutrophil cytoplasmic antibodies (ANCA)-associated vasculitides (AAV) are commonly divided into three different clinical entities (classified according to Jennette et al1): granulomatosis with polyangiitis (GPA, previously Wegener's granulomatosis), microscopic polyangiitis (MPA) and eosinophilic granulomatosis with polyangiitis (EGPA; also referred to as Churg–Strauss syndrome). They constitute systemic autoimmune disorders that are characterised by necrotising inflammation and destruction of small- to medium-sized blood vessels, strongly linked to the production of different types of ANCA (see Wilde et al2 for a recent review).

GPA, MPA and EGPA show certain similarities (eg, glomerulonephritis and alveolar haemorrhage in GPA and MPA in generalised stage, granulomatous vasculitis in GPA and EGPA), but differ in other clinical aspects (eg, asthmatic symptoms and eosinophilia in EGPA, ‘pure’ vasculitis without granuloma formation in MPA). Therefore, the classification is often difficult and has been controversially discussed in the literature.3

There is a genetic predisposition (eg, HLA-DPB1 in GPA though not in EGPA4), but the aetiology of AAV is complex and not yet understood in detail. Certain environmental factors have been linked to the onset and development of AAV, such as carriage of Staphylococcus aureus, which has been shown to be a risk factor for disease relapses in GPA.5 Frequency and antigen specificities of ANCA differ between the three AAV subtypes. While most GPA cases show cytoplasmic ANCA directed to proteinase 3, perinuclear ANCA to myeloperoxidase are regularly found in MPA. Myeloperoxidase (MPO)-ANCA are also the most frequent ANCA found in EGPA, but overall only about 40% of EGPA patients are typed ANCA+.6 ,7

Recently, it has been demonstrated that toll like receptor 9 (TLR9) may play a role in the pathogenesis of AAV, as TLR9 ligands can trigger the production of ANCA by peripheral blood-derived B lymphocytes from AAV patients in vitro.8 TLR9, a transmembrane protein, plays a fundamental role in pathogen recognition and activation of innate immunity. Members of the TLR family are pattern recognition receptors (PRRs) recognising molecular structures that are broadly shared by pathogens, known as pathogen-associated molecular patterns (PAMPs). Intracellular signals initiated by interaction of TLR and PAMPs result in inflammatory response, such as secretion of inflammatory cytokines, type 1 interferon, chemokines and antimicrobial peptides.9 In addition, signalling pathways of PRRs simultaneously induce maturation of dendritic cells, which are responsible for induction of the second line of host defence, the so-called adaptive immunity. TLR9 appears to respond to a component of (eg, bacterially derived) nucleic acids, the CpG motif. TLR9 is found almost exclusively in intracellular compartments, such as endosomes, and its ligands require internalisation into the endosome before signalling can occur (for a detailed review, see Akira et al10).

In order to investigate the potential role of TLR9 in AAV pathogenesis and ANCA formation on a genetic basis, we genotyped informative polymorphisms of TLR9 and of genes coding for key members in its signalling pathway (interleukin 6 (IL6), IL1 receptor-associated kinase 4 (IRAK4), myeloid differentiation primary response gene 88 (MyD88) and tumour necrosis factor (TNF) receptor-associated factor 6 (TRAF6)). Since epistatic effects have been described between TLR9 and the genes for nucleotide-binding oligomerisation domain containing 2 (NOD2), the interleukin 23 receptor (IL23R) and discs, large homolog 5 (DLG5) genes in inflammatory bowel disease (IBD),11 we also tested for potential genetic influences in AAV. All of these polymorphisms were also stratified according to the clinical disease entities as well as ANCA status.


Cases and controls

A well characterised cohort of 845 unrelated German AAV patients was recruited at the Vasculitis Centre Lübeck/Bad Bramstadt, Germany. Six hundred and forty-six of these patients had been diagnosed with GPA, 164 with EGPA and 53 with MPA according to the Watts algorithm.12 Three separate cohorts of healthy German individuals were used as controls. These cohorts comprised healthy blood donors from Northern, Western and Southern Germany, respectively (1344 samples altogether). The ANCA status of all cases was determined as described earlier.13 Detailed clinical information was available for most of the patients from the University of Lübeck and Klinikum Bad Bramstedt and was evaluated with respect to selected nucleotide polymorphism (SNP) genotypes (see below).

Replication cohorts

Four hundred and twenty-six white European AAV cases from Birmingham (UK) and Maastricht (Netherlands) were recruited for replication of significant results, 273 of which suffered from GPA, 53 from EGPA and 100 from MPA. These cases also fulfilled the diagnostic criteria of the Watts algorithm.12 Five hundred and fifty-four healthy white Europeans from Western Germany were used as controls.

Selection of candidate genes and SNPs

Four SNPs (rs352139, rs352140, rs352162, rs5743836) have been selected to represent the complete genomic region of the TLR9 locus on chromosome 3p21.3 on a tagSNP basis. Additionally, we analysed representative SNPs in genes directly involved in TLR9 regulation. For an initial step, protein-coding genes have been selected which are in close vicinity to TLR9 within its complex cascade; SNPs therein, which have been used in association studies previously and which alter the amino acid sequence of the respective protein, were analysed: rs1800795 in IL6,14–16 rs4251545 (Ala428Thr) in IRAK4, rs7744 in MyD88 (3′ untranslated region, 3′UTR), and rs540386 in TRAF6 (intronic). For evaluation of epistatic effects we also analysed rs2066845, rs2066844 and rs2066847 in NOD2 as well as rs1004819, rs1343151 and rs11209026 in IL23R and rs1248696 in DLG5.


SNPs were genotyped using commercially available TaqMan assays on a StepOnePlus real-time PCR device (Applied Biosystems, Darmstadt, Germany). The NOD2 insertion polymorphism rs2066847 was analysed by fluorescence labelled PCR and capillary electrophoresis on a 3500xL genetic analyser (Applied Biosystems). Representative samples were directly sequenced to confirm genotypes (see online supplementary figure S1; primer sequences available on request).

Data analysis

Genotypes for all polymorphisms were recorded in LINKAGE format. Allele frequencies were compared between cases and controls by χ2 tests on contingency tables as implemented in Haploview 4.1.17 Bonferroni correction was used to adjust for multiple testing. For the initial TLR9 analyses, c=12 was chosen as four SNPs were analysed in three different entities, GPA, MPA and EGPA. The genes investigated subsequently were corrected for the number of SNPs (c=11). Pairwise linkage disequilibrium, haplotype block frequencies and Hardy–Weinberg equilibrium (HWE) were also calculated with Haploview. Power analyses have been performed using Quanto Association of clinical characteristics in patients and genotypes of the associated SNPs was performed using χ2 tests. Age of onset were compared using ANOVA within STATISTICA 10.0 software (Statsoft, Tulsa, Oklahoma, USA). Testing for epistatic effects was performed with GAIA19 using an additive interaction model over and above the main effects of any individual polymorphism.


For all polymorphisms analysed in the three German control cohorts, no significant deviations from HWE were detected, and there was no significant difference in the allele frequencies between the three cohorts for any marker (data not shown). Thus, the control cohorts were combined for all further analyses.

When analysing TLR9 SNPs we found a strong association of all four SNPs for German GPA cases compared to healthy controls. Interestingly, MPA was also significantly associated with these SNPs, yet, each with the contrary allele. No association was detected for EGPA (see table 1A). When haplotypes (defined according to the solid spine of LD method implemented in Haploview, see figure 1) were analysed for rs352162, rs352140 and rs352139, GPA was highly significantly associated with the TCT haplotype, while MPA was associated with the CTC haplotype.

Table 1

Analyses of TLR9 variations in the initial cohort (863 German AAV cases vs 1344 healthy controls)

Figure 1

TLR9 linkage disequilibrium (LD) plot and calculated haplotype block. The LD plot was created with Haploview 4.1. The level of pairwise LD (r2) is intensity-coded, with darker shades of grey corresponding to higher degrees of LD. r2 values are also given in each square representing the pairwise LD between two SNPs. Haplotype architecture was defined according to the solid spine of LD algorithm implemented in Haploview. rs352162, rs352140 and rs352139 are in strong LD while rs5743836 is only weakly linked. Thus, haplotype frequencies have been calculated for the first three SNPs, and rs5743836 has been analysed separately.

We then compared cases according to their ANCA status; again a strong association for PR3-ANCA+ AAV and a reciprocal effect in MPO-ANCA+ AAV was detected for both SNP and haplotype comparisons. No association was found for ANCA AAV cases (see table 1B).

For replication, the same four TLR9 SNPs were analysed in Dutch and British AAV cases. When compared according to clinical entity, neither GPA nor MPA showed allele/haplotype frequencies convincingly deviating from controls (see online supplementary table S1). Yet, very similar differences as in the initial cohort were detectable when PR3+ and MPO+ AAV cases were compared to controls (see table 2). These differences did not reach nominal significance levels, but lack of significance was very likely attributable to a low statistical power, since the ORs were generally comparable to findings in the initial German cohort. For example, the statistical power for the replication cohort was only 43.29% (calculated with the Quanto software using the appropriate parameters; 187 PR3 ANCA+ cases, 554 controls, OR 1.24, MAF 0.43, log-additive model).

Table 2

Analyses of TLR9 variations in the replication cohort stratified according to ANCA status

In addition, we compared certain case cohorts. When PR3-ANCA+ cases from the initial cohort were compared to MPO-ANCA+ cases, an even stronger association was detected for both SNPs and 3-marker haplotypes. The same difference was detected in the replication cohort, where PR3-ANCA+ cases clearly differed from MPO-ANCA+ positive cases, now reaching nominal significance for rs352140 and the calculated 3-SNP haplotypes. Consequently, a striking association was demonstrable when combining all PR3-ANCA+ and MPO-ANCA+ cases (see figure 2).

Figure 2

Forrest plot of case–case comparisons according to ANCA status (PR3-ANCA+ AAV vs MPO-ANCA+ AAV). The plot shows the ORs (vertical lines) and 95% CI (boxes) as revealed from the different comparisons. The upper third of the figure depicts the results from the initial control divided into the TCT haplotype (a), the most significant SNP thereof (rs352140, b) and the separately analysed SNP rs5743836 (c). The central third of the figure shows the respective data from the replication cohort and the lower third from the combined analysis of both cohorts. The main association signal derives from the haplotype (and the tagging SNP rs352140 therein) rather than from the weakly linked SNP rs5743836, which only reaches a borderline significance level in the combined analysis.

Subsequently, we searched for associations of clinical parameters in the German AAV cases and TLR9 SNPs or haplotypes. For some parameters (eg, involvement of central nervous system, pulmonary fibrosis), a nominally significant association was established (see online supplementary table S3). Additionally, both PR3-ANCA+ and MPO-ANCA+ cases carrying the rs352140 CC genotype presented with a earlier age at onset (48.62±1.28 years; mean±SD) than the respective carriers of the TT genotype (52.99±1.16 years; p=0.02; see online supplementary table S4 and figure S2). Yet, none of these associations would have passed correction for multiple testing. Cases from the replication cohort have not been analysed, as these groups were considerably smaller, and further stratification would not have yielded statistically meaningful results.

When analysing SNPs in genes of four members of the TLR9 signalling cascade in German cases and controls, none of the SNPs in the genes for IL6, MYD88, IRAK4 and TRAF6 showed a promising association with an AAV entity or a group stratified according to ANCA status (see online supplementary tables S5 and S6). Consequently, these analyses have not been extended to the replication cohorts.

Finally, we checked for association of polymorphisms in NOD2, IL23R and DLG5 with AAV and for epistatic effects of these with TLR9 SNPs/haplotypes. None of the polymorphisms was significantly associated with AAV or any of its entities a priori (see online supplementary tables S5 and S6). Moreover, there was no significant epistatic association between TLR9 genotypes (or haplotypes) and the investigated genes (data not shown).


There is recent evidence that TLR9 may play a pivotal role in the pathology of AAV.20 TLR9 ligands (ie, CpG motives) can trigger the production of ANCA by peripheral blood-derived B lymphocytes from AAV patients in vitro.8 ,20 A genetic contribution to AAV susceptibility has been shown repeatedly (see Cartin-Ceba et al21 for a recent review) and, interestingly, it has been demonstrated that SNPs reaching genome wide significance were more strongly associated with ANCA status (PR3-ANCA or MPO-ANCA) than with the clinical syndrome (GPA or MPA).22 Hence, evaluation of TLR9 genetics in AAV appears plausible, especially since the TLR9 genomic region is not well covered by the array type used in the recent genome wide association study performed in AAV.22 Yet, the genetic background of multifactorial diseases is often complex and strategies taking into account, for example, phenotypic subgroups and/or gene–gene interactions are increasingly applied. Furthermore, effect sizes of individual polymorphisms are mostly small and may be missed by genome wide approaches, necessitating rigorous correction for multiple testing. Thus, candidate gene approaches as in this study are still contributing to our understanding of complex disorders.

Here, we demonstrate a novel association of TLR9 polymorphisms with AAV in German patients. Additionally, a significant difference between PR3 AAV and MPO AAV was discovered. While the case–control association could not be formally replicated (probably because of the limited size of the Dutch and British replication panel), a significant difference between PR3 AAV and MPO AAV was also demonstrable in the replication cohort.

On the other hand, no association with TLR9 linked genes could be established. Associations of IL23R SNPs have been described for numerous inflammatory diseases and IL23R is also a very interesting candidate gene for AAV.23 Here we did not identify any AAV association of IL23R genotypes, either alone or in combination with TLR9 SNPs, but further analyses in larger cohorts appear necessary for a final evaluation. NOD2 mediates responses to intracellular bacterial lipopolysaccharides (LPS).24 The NOD2 gene was previously shown to be not associated with GPA in a smaller cohort.25 Our data strongly support the lack of association and also argue against a genetic interaction with TLR9 genotypes as described in IBD.11 DLG5 was considered worth being genotyped because of its potential role in epithelial barrier dysfunction discussed in GPA26 and the epistatic effect with TLR9 described in IBD. Yet, no genetic evidence for involvement of DLG5 in AAV was observed here, for either the gene alone, or interaction with TLR9 genotypes.

The fact that we found neither a primary association of the IL6 promotor SNP in AAV nor a significant epistatic effect between the IL6 promotor SNP and any TLR9 SNP does not rule out a potential role of IL6 genotypes in AAV pathology and treatment (eg, with rituximab16). Such mechanisms should be evaluated further.

MYD88, IRAK4 and TRAF6 are members of the TLR9 signalling cascade eventually leading to induction of inflammatory cytokine production and Th1 immune responses.10 As an initial step we have genotyped one representative within each gene, which has a potentially direct functional impact or has been previously linked to inflammatory disorders.27 ,28 Although we did not find any evidence for an association with AAV or epistatic effects with TLR9, genotyping of other polymorphisms of these genes may yield path-breaking results, since the selected SNPs do not fully cover the genetic variability of the respective genes. Moreover, numerous members of the TLR9 cascade have been identified (eg, IRAK1, IRF-7) and should be tested for association with AAV.

Thus, while the genetics of TLR9 related genes need further investigation, the primary association of TLR9 with AAV, which we have presented in this work, yields some exciting insights into AAV pathogenesis. Within the AAV group, the role of infections as triggering factors has predominantly been discussed in GPA. Evidence for such models comes from the detection of S aureus carriage as a risk factor for GPA relapses5 and a certain time periodicity of GPA (but not MPA) incidence rates.3 Since comparable data is widely lacking for MPA and EGPA, it is highly interesting that we found an association of TLR9 with PR3-ANCA+ AAV (mainly GPA), but a completely different allele distribution in MPO-ANCA+ AAV (mainly MPA, EGPA). Although our MPA data have to be taken with caution because of the relatively small sample size, such phenomena of different genetic backgrounds of the respective AAV entities have been reported previously.4 ,22 ,29 These data provide further evidence that AAV should be classified in more detail, and, as a consequence, patients need to be serotyped using the most recent techniques to ensure optimal therapy.30 In this respect it would also be highly interesting to analyse Japanese patients since they often present with a GPA phenotype sparing the ear, nose, throat manifestation and are typed MPO-ANCA+ more often.31

There are preliminary data that the TLR9 polymorphisms mediate functional effects either themselves or due to linkage with one another. In addition to the abovementioned effect of rs5743836 on transcription factor binding,14 certain rs352162 alleles have been associated with an increased TNFα production by peripheral blood leucocytes in response to bacterial DNA stimulation.32 Yet, analysis of the exact functional consequences of the AAV-associated TLR9 polymorphisms appears mandatory before any further clues for AAV pathology can be drawn.

Based on the hypothesis that the associated alleles confer altered TLR9 transcription, one could speculate that TLR9 may have an effect not only on AAV susceptibility but also on certain clinical manifestations. When searching for such relationships within our cohort, only few associations were nominally significant, and these would not have withstood correction for multiple testing. Thus, we cannot provide reliable evidence for the relevance of TLR9 variations on certain AAV manifestations at this point, and further analyses are needed. For instance, it would be quite valuable to correlate our data with clinical information, such as carriage of S aureus, which unfortunately was not available.

Taken together, our data support models of TLR9 playing a pivotal role in AAV pathology and encourage strategies trying to identify bacteria as triggering agents in AAV, especially in (PR3-ANCA+) GPA.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Handling editor Tore K Kvien

  • Contributors SW and CAH designed the study and drafted the manuscript. CAH did most of the lab work and most of the initial data analyses. SM did parts of the lab work and some of the data analyses. JUH, FM, BW, JWCT, WLG, GA and JTE recruited and characterised cases and/or controls, and were involved in final data analyses, interpretation and discussion of the results. All authors critically revised the manuscript and gave approval to the final version.

  • Funding Supported by a grant from the Deutsche Forschungsgemeinschaft [KFO 170].

  • Competing interests None.

  • Ethics approval All subjects gave written informed consent, and ethical principles for medical research involving human subjects according to the Declaration of Helsinki have been followed. The study protocol was approved by the local ethics committees at the Universities of Luebeck (Germany; No. AZ 05-219 and 06-087), Maastricht (The Netherlands; No. 05-158) and Birmingham (UK; No. 0723).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.