Abstract
Objective. Toll-like receptors (TLR) have been implicated in the pathogenesis of arthritis. We investigated the role of functional variants of TLR in the disease phenotype and severity of rheumatoid arthritis (RA).
Methods. All patients from a longterm observational inception cohort (n = 319) were genotyped for 22 single-nucleotide polymorphisms (SNP) in TLR2, 3, 4, 5, 7, 8, and 9 using multiplex assays. Clinical characteristics including sex, age at disease onset, rheumatoid factor (RF), and shared epitope positivity and disease activity score and radiological progression were taken into account. Genotypes were analyzed for association with Disease Activity Scores (DAS28) and joint damage (Rau scores) at 3 and 6 years.
Results. After Bonferroni correction, there was a moderate association between RF positivity and TLR8-rs5741883. No other TLR variant was significantly associated with any RA clinical characteristics.
Conclusion. Using a large inception cohort and strict statistical evaluation, we could not identify an association between functional TLR variants and RA phenotype and disease severity. This suggests the functional TLR variants do not play a major role in RA phenotype and disease severity.
- TOLL-LIKE RECEPTOR
- RHEUMATOID ARTHRITIS
- RHEUMATOID FACTOR
- RADIOLOGICAL JOINT DAMAGE
- DISEASE ACTIVITY SCORE
Rheumatoid arthritis (RA) is an autoimmune disease affecting multiple synovial joints, leading to significant morbidity and shortened life expectancy. Despite longstanding efforts, the precise mechanisms underlying the inflammatory processes remain to be elucidated. Toll-like receptors (TLR) were discovered to be crucial receptors triggering innate immune responses. Currently, 10 TLR subtypes have been described in humans, all thought to have their own specific ligands and cellular localization (as reviewed1). For example TLR1, 2, 4, 5, and 6 are located on the cell surface and scavenge the environment for ligands. In contrast, TLR3, 7, 8, and 9 have an intracellular localization and recognize intracellular ligands, including ligands that are endocytosed2,3. Accumulating evidence suggests a pivotal role for TLR in the recognition of endogenous ligands and, as well, linking innate and adaptive immune responses. Recently, several groups have provided evidence for a role of TLR in arthritis in experimental disease conditions and in humans. It has been demonstrated that various TLR subtypes are expressed at higher levels in synovial tissues from patients with RA compared to those from healthy controls4–6. Ligands for TLR3 [host-derived (RNA) and TLR4 (HSPB8)] are also abundant in the circulation as well as in the synovial joints of patients with RA4,5,7. Additionally, TLR4-mediated stimulation of dendritic cells from patients with RA leads to significantly higher cytokine concentrations compared to similar cells from healthy controls, further supporting a deranged TLR response in RA. On these grounds, we hypothesized that TLR might be involved in the pathogenesis of RA. To test this, we investigated potential associations between TLR variants and RA phenotype and severity.
MATERIALS AND METHODS
Ascertainment of patients
Genotyping was performed in RA patients participating in an early RA inception study started in 1985. Our study includes only those patients who met the American Rheumatism Association (American College of Rheumatology) criteria for RA8, had a disease duration < 1 year, and had not previously been treated with disease-modifying antirheumatic drugs (DMARD) or biological therapies. The local ethics committee approved the study.
Characterization of disease activity and outcome
Patients’ demographic data such as sex, age at disease onset, and the presence of rheumatoid factor (RF), HLA-DR4 and the shared epitope were included in the analysis. We used the Disease Activity Score 28 (DAS28) and the Rau score at baseline and after 3 and 6 years of followup to determine the disease course and radiological joint progression, respectively9,10. The use of DMARD was analyzed using essentially the same protocols as described11,12.
Selection of SNP and description of assay
SNP selected for the assays were primarily functional SNP, and selection was based upon information available at the dbSNP (US National Center for Biotechnology Information; http://www.ncbi.nlm.nih.gov/SNP/), SNPper (Children’s Hospital Informatics Program, Boston, MA, USA; http://snpper.chip.org/bio/), and IIPGA (http://www.innateimmunity.net/) databases (Table 1). Thirteen SNP located in the human TLR2, 4, 5, and 9 genes, and 9 SNP located in the human TLR3, 7, and 8 genes were assessed in 2 multiplexed bead-based assays using a Luminex 100IS flow cytometer (Luminex Corp., Austin, TX, USA). The tests were based on described procedures, with some modifications (a detailed protocol is available from the authors)13–15.
Allele-specific primer extension sequences.
Statistical analysis
Frequencies of the TLR genotypes were tested for Hardy-Weinberg equilibrium using the standard goodness-of-fit test. Similarity of genotype and allele distribution between patients and controls was tested with chi-square tests for 3 × 2 contingency tables. Differences in the disease characteristics between patients were analyzed using Student’s t test or Mann-Whitney U test. For the TLR located on the X-chromosome, genotype and allele frequencies were recalculated after stratification for sex. Correction for multiple testing was performed using the Bonferroni correction. P values ≤ 0.002 were considered statistically significant.
Power calculations were performed using Quanto. Based on a sample size of 177 and 319 individuals the power to detect a locus that explained 5% variation of the continuous trait using a Bonferroni corrected p value of 0.002 was 47% and 83%, respectively.
RESULTS
In total, DAS28 was present for all patients at baseline and after 3 and 6 years. Radiographs were present from 272 patients at baseline, 240 at 3 years, and 177 at 6 years of followup. Demographic variables including age, sex, presence of RF, and age at disease onset were documented for these patients. The SNP in TLR5-rs5744176 (Asp694Gly), TLR7-rs3853839 (Ala448Val), and TLR8-rs5744088 (3’-UTR) were not polymorphic in our RA population.
For the extracellular TLR subtypes (TLR2, 4, and 5) no association was observed between the genotypes and age at onset, sex, or presence of RF (data not shown). Similarly, no association was found between the genotypes of the diverse TLR subtypes and the disease activity and radiological joint damage at baseline (data not shown) and after 3 and 6 years of followup (Table 2).
Genotype distribution of extracellular Toll-like receptors (TLR) and association with disease severity and/or radiological joint damage.
With respect to TLR3, 7, 8, and 9 located intracellularly, no significant associations were observed between the genotypes and sex or age at disease onset. RA patients carrying the TLR8-rs5741883 C allele were significantly more positive for RF, which was clearly correlated with a gene-dose effect (heterozygous for the C allele 76%; p = 0.02, homozygous for the C allele 88%; p = 0.001) compared to patients homozygous for the T allele (67% RF-positive). We found an association between TLR3-rs3775291 and joint score after 6 years (Table 3). In addition, we found an association between TLR8 and DAS28 after 3 years (rs3764879) and 6 years (rs3764879 and rs3764880) (Table 3). After correction for multiple testing and sex (for TLR8, as this gene is located on the X-chromosome), the significant associations between TLR SNP and disease phenotype or severity were lost.
Genotype distribution of intracellular Toll-like receptors (TLR) and association with disease severity and/or radiological joint damage.
DISCUSSION
Much research has focused on the role of TLR in autoimmune diseases, including RA, systemic lupus erythematosus, multiple sclerosis, and inflammatory bowel diseases. A substantial body of evidence points to a role for TLR in RA. One way to investigate the role of TLR themselves is to study genetic variants (e.g., SNP) in the TLR genes that might lead to an altered ligand binding capacity and/or expression leading to an altered TLR-mediated response that might subsequently translate into variations in disease activity and/or severity.
Using a well documented prospective cohort of 319 patients with RA, we were unable to show any significant effect of TLR SNP on RA disease variability and/or severity. Although the total group could be considered large enough, the genotypic distribution of many genes led to very small subgroups, strongly affecting the power of the study and increasing the risk of rejecting clinical associations. Our investigation underscores the need for multicenter studies to evaluate the potential influence of genetic variants on the outcome and behavior of complex diseases such as RA.
Triggering TLR initiates complex cascades of downstream adapter molecules, e.g., MYD88, TRIF, and IRAK, eventually ending in nuclear factor-κB signaling and cell activation. Perhaps genetic, posttranscriptional, and/or posttranslational modifications in these adapter molecules might explain the deranged TLR response observed in RA. Such relationships would not have been detected in our study, and further research focusing on these molecules is therefore warranted.
Our results suggest potential associations between some TLR SNP and RA phenotype, such as the TLR8 SNP and RF positivity. After correction for multiple testing, however, none of these associations reached statistical significance. Multicenter studies are needed to replicate and validate these results.
Acknowledgments
Pia Grothe Meinke is thanked for excellent technical assistance.
Footnotes
-
Supported by the Lundbeck Foundation and the Danish Biotechnology Program. Dr. Radstake is a VIDI Laureate of the Dutch Organization of Research (NWO).
- Accepted for publication December 3, 2009.
