Osteoarthritis (OA) has been regarded as a multi-factorial disease with multiple risk factors such as age, gender, obesity, hormonal, mechanical, and genetic factors contributing to onset and progression. Twin studies, segregation analyses, linkage analyses, candidate gene association studies and genome-wide association studies (GWAS) have generated important information about inheritance patterns and the location of potentially causative mutations in the genome.1 Associations between genetic susceptibility and risk of development and/or progression of OA are likely to be complex and will include innate and acquired host responses to biomechanical, inflammatory, immunological, and other environmental stimuli.2
Recent studies suggest that the Toll-like receptor (TLR) family is involved in development and progression of OA.3 Ten members of the TLR family are currently recognized.4 These recognize a variety of molecules ranging from viral derived RNA (TLR-3, -7, and -8) and DNA (TLR-9) to bacterial derived lipoproteins (TLR-1, -2, and -6), lipopolysaccharide (LPS; TLR-4), or flagellin (TLR-5). In human OA chondrocytes TLR ligands activate catabolic pathways.5 TLR-induced collagenase expression appears to contribute to cartilage catabolism in OA suggesting that modulation of chondrocyte TLR expression or activation may be a route by which cartilage breakdown can be blocked.6
Single nucleotide polymorphisms (SNPs) occur commonly in TLR genes and the presence of these SNPs is hypothesized to influence the function of these receptors.7 TLR gene polymorphisms have been reported to be associated with a range of inflammatory diseases such as systemic lupus erythematosus (SLE), atherosclerosis, asthma, and multiple sclerosis.8, 9 There is no evidence for major effects of TLR SNPs in rheumatoid arthritis,10, 11 however we have previously demonstrated that the (−1486T/C) promoter polymorphism of the TLR-9 gene is associated with end-stage knee OA.12 TLR-9 is expressed within the endosomal compartments where it binds to microbial DNA rich in CpG motifs. TLR-3, -7, and -8 are also present on the surface of endosomes and respond primarily to nucleic acid based pathogen-associated molecular patterns (PAMPs) from viruses and bacteria.4 Therefore, in this study, we looked specifically at the association of TLR-3, -7, and -8 SNPs with knee OA in a Han Chinese population.
MATERIALS AND METHODS
Two independent case–control studies were carried out. The stage 1 study population, recruited from a rural city (2008–2009), included 385 OA cases and 260 healthy controls. The stage 2 study population, recruited from a central city (2010–2011), included 438 OA cases and 334 healthy controls. Disease severity in the OA populations, who were attending an Orthopaedic hospital, was assessed by Kellgren–Lawrence (K–L) grading. All patients had a K–L score ≥2. Other etiologies of knee joint disease such as inflammatory arthritis, posttraumatic or postseptic arthritis, skeletal dysplasia or developmental dysplasia were excluded. Healthy control subjects had no signs or symptoms of joint disease and standard X-rays of the knee joints confirmed absence of OA. The study was reviewed and approved by the institutional ethical committee of Tri-Service General Hospital. Informed consent was obtained from all study participants.
SNP Selection and Genotyping
SNP genotype information was downloaded from the HapMap database and The National Center for Biotechnology Information dbSNP database. Tag SNPs were selected for TLR-3, -7, and -8 genes with the criteria of minor allele frequency (MAF) >5% presence in regulatory regions or having been reported by other investigators. Genomic DNA was extracted from peripheral blood samples using QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). The TLR-3 (rs3775296, rs3775290, rs3775291, and rs5743312), TLR-7 (rs179008, rs5741880, rs3853839, and rs179010), and TLR-8 (rs3764879, rs3764880, and rs5744080) polymorphisms were screened by polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP). Primer design was based on published sequences13, 14 or designed by primer Z software (S-Table 1). To validate genotyping results at least 10% of samples were randomly selected for repeated genotyping.
In Silico Analysis
To investigate the possibility that the TLR-3 SNP rs3775296 has a direct effect on transcription factor binding, we used the prediction web software transcription element search system (TESS) to search for promoter elements.
Cell Culture and Real-Time PCR Analysis
Human osteoarthritic cartilage samples were obtained from knee joints at total knee replacement (n = 26, mean age, 66 years; range 58–72 years). Articular chondrocytes were extracted by sequential enzymatic digestion at 37°C in 5% CO2 with 0.25% trypsin (Gibco, Carlsbad, CA) for 30 min and then 3 mg/ml blend collagenase type H (Sigma–Aldrich, Inc., St. Louis, MO) for 12 h. Extracted cells were resuspended in 10 ml Dulbecco's modified Eagle's medium (DMEM)/Nutrient Mixture F-12 HAM medium (Gibco) supplemented with 10% FBS (Sigma–Aldrich), 100 IU/ml penicillin (Gibco), and 100 µg/ml streptomycin (Gibco). Chondrocytes were maintained at 37°C in a humidified atmosphere of 5% CO2/95% air until they were confluent and ready to passage for experimentation. 1 × 106 chondrocytes cultured as described above were serum starved and treated with 100 nM dexamethasone for 24 h. Total RNA was extracted using the Trizol RNA isolation protocol (Invitrogen Corp., Carlsbad, CA). The real-time PCR Primer design was based on published sequences.15 Analysis was performed using the Maxima® SYBR green qPCR master mix (Fermentas, Glen Burnie, MD) and an ABI 7500 real-time PCR system (Applied Biosystems, Foster City, CA). Expression of β-actin was used as internal control. TLR-3 gene expression normalized by β-actin was calculated by using the method.
Genotypes and allelic frequencies were compared between OA patients and healthy controls by using the χ2-test or Fisher's exact test when appropriate. The demographics were evaluated by Student's t-test for continuous variables and expressed as mean ± standard deviation (SD). Logistic regression was used to estimate crude and adjusted (age, gender, and body mass index) odd ratios (ORs) and 95% confidence intervals (CIs) as a measure of association with the risk of OA. Linkage disequilibrium (LD) and haplotype analyses were performed using Haploview software and WHAP, respectively.16 An association of the TLR-3 promoter SNP with TLR-3 mRNA expression was tested by Kruskal–Wallis (K–W) test. All statistical analysis was carried out using SPSS 18.0 for Windows (SPSS Inc., Chicago, IL).
The distributions of selected characteristics between the overall 823 OA cases and 594 healthy control subjects are summarized in Table 1. The genotype and allelic distributions of the four SNPs in TLR-3 and their associations with OA risk are shown in Table 2. No deviation from the HWE was observed in the controls (p > 0.05). In the stage 1 cohort, the genotype distributions of rs3775296 and rs3775290 in TLR-3 were significantly different between the OA patients and healthy controls (p < 0.05). The other SNPs (rs3775291 and rs5743312) in TLR-3 demonstrated no significant genotypic and allelic association between the OA cases and healthy controls. To see whether the results could be replicated in a different population we undertook a second stage study for genotyping of all four SNPs. The two promising SNPs (rs3775296 and rs3775290) were again significantly different in their genotype distributions between the OA cases and healthy controls. When the data from the two stages were combined, the association of these two SNPs (rs3775296 and rs3775290) with OA is maintained (95% CI = 1.33–3.33 and 95% CI = 1.32–2.62, respectively; Table 2).
Table 1. Characteristics of OA Cases and Healthy Controls
Age (mean ± SD)
71.66 ± 8.19
68.63 ± 10.04
71.13 ± 8.42
68.83 ± 10.35
Gender (% female)
BMI (mean ± SD)
25.65 ± 3.54
25.00 ± 9.39
26.02 ± 3.48
24.84 ± 9.56
K–L grading (%)
Table 2. The Genotype Distribution of TLR-3 Gene and Their Association with OA Risk
As TLR-7 and -8 genes located on an X chromosome, male and female individuals needed to be analyzed for any disease association separately. In males, the allelic frequencies of TLR-7 rs179010 and TLR-8 rs5744080 polymorphisms were significantly different between OA cases and healthy controls (95% CI = 0.53–0.89 and 95% CI = 1.12–2.15, respectively). However, no associations were found between OA and TLR-7 and TLR-8 polymorphisms among females (S-Table 2).
Stratifying the patients by K–L score identified significant differences in allele and genotype of the TLR-3 rs3775296 (CC/AA, 95% CI = 1.40–4.30; CA/AA, 95% CI = 1.14–3.61), rs3775290 (CT/TT, 95% CI = 1.66–3.78), and rs3775291 (CT/TT, 95% CI = 1.14–2.76) between patients with grade 4 knee OA and controls (S-Table 3).
The haplotype ATCA of the TLR-3 gene was associated with a decreased risk compared to the most frequent haplotype CCTC (95% CI = 0.56–0.87). The haplotypes for the TLR-3 gene (CTCA and CCTA) were associated with an increased susceptibility to OA (95% CI = 1.01–3.18 and 95% CI = 1.23–5.12, respectively) compared to the most frequent haplotype CCTC (Table 3).
Table 3. Haplotype Analysis of the Identified 4 SNPs in TLR-3 between OA Patients and Control Subjects in Stage 1 + Stage 2 Studies
In the order of rs3775296, rs3775290, rs3775291, and rs5743312.
In silico analysis revealed that the T allele of rs3775296 generated several novel binding sites for different transcription factors and the potential regulatory transcription factor-binding motif with the highest score corresponded to a glucocorticoid receptor (GR) response element (S-Fig. 1). We found a significant difference in the expression of TLR-3 following 100 nM dexamethasone stimulation of chondrocytes among the various genotypes (mean ± SE relative mRNA expression 0.96 ± 0.21 [n = 14], 0.73 ± 0.12 [n = 9], and 0.34 ± 0.06 [n = 3]) in CC, CA, and AA, respectively (p = 0.004; Fig. 1).
Following a two-stage study involving the genotyping and assessment of disease severity of 823 OA patients and 594 healthy controls recruited among Han Chinese, we provide, for the first time, evidence for an association between SNPs rs3775296 and rs3775290 of TLR-3 and OA. Functional analysis reveals a significant difference in the regulation of expression of TLR-3 between the various genotypes of rs3775296.
Recently several GWAS have been conducted to identify OA susceptibility markers in Caucasian populations.17–19 These have highlighted chromosome regions that may harbor OA susceptibility genes such as 2q11.2–q13, 16p12.1, 9q22.31, 7q22, and the current total of OA susceptibility loci in European populations is 11. None of these studies have however reported on an association between TLR genes and OA susceptibility. There are a number of possible reasons for the difference in results. Associations with the TLR loci identified in the current study may have been present in previous European GWAS but may not have been reported as these did not reach the stringent p-value threshold applied in GWAS. However ethic differences would appear to be important. A GWAS in Japanese and Han Chinese populations has shown an association with a functionally relevant SNP in the double von Willebrand factor domain A (DVWA) gene and knee OA.20 However subsequent meta-analyses, whilst demonstrating a significant global effect on knee OA did not show an association with DVWA SNPs and knee OA in European populations.21 Differences in disease pathways, given differences in prevalence and joint site distribution between the Asian and Caucasian populations as well as differences in OA selection criteria varying from symptomatic, radiographic OA to total knee replacement are also likely to be involved. The other important observation is that the current study is a candidate gene approach in which we have focused on a restricted number of TLR genes in view of the results from a previous study showing an association of knee OA with a polymorphism in TLR-9 and their potential role in mediating inflammatory signaling in OA.5, 12, 22
Association studies in populations that are genetically heterogeneous can yield large numbers of spurious associations if population subgroups are unequally represented among cases and controls.23 In this study, the data from the two stages were combined. Our pooled population represents a relatively homogeneous group and makes it appropriate to investigate the implication of TLR gene SNPs in different subgroups of OA. We stratified the patients by K–L score or gender to analyze the association between TLR gene polymorphisms and OA. Moreover, we analyzed two different stages of cases and controls and effectively replicated the finding in an independent investigation. This greatly reduces but does not exclude the possibility that the results are due to chance (false positive).
Relatively few studies have investigated the association between TLR polymorphisms and arthritis. Genetic variants in TLRs do not appear to play a major role in susceptibility for developing RA nor in outcomes to anti-TNF treatment in Caucasians with RA.10 A limited study of 100 patients in a Turkish population, while showing no association between a variant in TLR-3 and RA susceptibility did suggest that the rs187084 SNP in TLR9 gene may be a susceptibility factor for RA.24 In the current study TLR-3 SNPs are associated with advanced stages of knee OA as assessed by radiologic grade. The reasons for these associations are not clear as genetic risk factors can influence the risk of OA development and progression at various stages during the course of the disease. How those polymorphisms may functionally influence OA onset or progression remains to be understood but the role of TLRs in the development of OA is supported by several studies. Ligands for TLR-2, -3, -4, and -9 have been shown to induce or exacerbate arthritis in several experimental models25 and play a pivotal role as potent catabolic stimuli in articular chondrocytes.26 TLRs are expressed by chondrocytes in vitro and in vivo and expression is altered in OA lesions.5, 6 As TLR-3 ligands are potent inducers of MMP13 in chondrocytes, increased expression in OA may be important in cartilage degeneration.6
The haplotype of four polymorphisms in TLR-3 was associated with OA. OA patients carrying the ATCA haplotype had a lower risk of OA than those carrying the most common haplotype CCTC. Patients carrying CTCA and CCTA haplotypes also had a higher risk of OA than those carrying the CCTC haplotype. These haplotypes differ from those studied in previous work in patients with asthma where no effects of TLR-3 haplotypes are seen14 and in patients with melanoma where TLR-3 haplotypes are associated with melanoma risk.27 Further work is required to fully understand the role of TLR-3 haplotypes on the genetic susceptibility to OA.
Epidemiological data have shown that the risk of OA differs between males and females whilst linkage and family studies have suggested that there are likely to be both sex-specific and anatomical site-specific genes that influence OA.28 In the present study, we observed a male gender association with TLR-7 and TLR-8 SNPs and OA. This may relate to the fact that TLR-7 and TLR 8 share a common locus on the X-chromosome. Similar sex-specific associations have been identified with TLR-7 and SLE.29
In silico analysis, the TESS results revealed several potential transcription factor binding motifs in both the reference and alternative sequence including GATA-1, GR, HSTF, HST1, Dl, PR A, and PR B. A previous study found that the human TLR-3 promoter contains a functional IFN-stimulated response element/IRF element.30 Glucocorticoids and progesterone have been shown to influence TLR-3 activity in dendritic cells31 however the presence of a glucocorticoid receptor response element in the TLR-3 promoter which may regulate transcriptional activity has not been previously described. Dexamethasone stimulation of chondrocytes appears to alter TLR-3 gene expression depending on the rs3775296 genotype. We hypothesize that the SNP in the promoter region of TLR-3 influences gene expression in response to inflammatory cytokines, glucocorticoids, and progesterone. Interestingly, normally anti-inflammatory glucocorticoids have been shown to synergize with TNFα and IFNγ to stimulate TLR production in lung epithelial cells.32 Altered promoter activity, resulting in an either increased or decreased expression of TLR-3 is likely to have important effects by either enhancing or decreasing chondrocyte catabolic and proteolytic processes in osteoarthritis. Further studies are required to establish the functional effects of the SNPs identified in the current study and their relevance to OA susceptibility or disease progression.
In conclusion, significant associations between the TLR-3 promoter polymorphism (rs3775296) and synonymous SNP (rs3775290) and knee OA have been identified. The functional study has also demonstrated that the promoter polymorphism (rs3775296) may be involved in the development of OA through transcriptional modulation of TLR-3.
This study was supported by grants from the National Science Council, National Defense Medical Center and Tri-Service General Hospital, Taiwan (H.S.L.: TSGH-C100-007-009-10-S03, MAB101-29; C.H.L.: NSC101-2314-B -038-004, NSC100-2314-B-038-041; S.L.S.: NSC99-2314-B-016-001).