Haplotypes of IL1B, IL1RN, IL1R1, and IL1R2 and the Risk of Venous Thrombosis
Objective— It has been suggested that the overall effect of the major proinflammatory cytokine interleukin-1 (IL-1) on coagulation and fibrinolysis is prothrombotic. The aim of this study was to investigate whether common variations in IL1B, IL1RN, IL1R1, and IL1R2 influence the risk of venous thrombosis.
Methods and Results— In a case–control study on the causes of deep venous thrombosis, the Leiden Thrombophilia Study (LETS), we genotyped 18 single nucleotide polymorphisms (SNPs) in IL1B, IL1RN, IL1R1, and IL1R2, enabling us to tag a total of 25 haplotype groups. Overall testing of the haplotype frequency distribution in patients and controls indicated that a recessive effect was present in IL1RN (P=0.031). Subsequently the risk of venous thrombosis was calculated for each haplotype of IL1RN. Increased thrombotic risk was found for homozygous carriers of haplotype 5 (H5, tagged by SNP 13888T/G, rs2232354) of IL1RN (Odds ratio=3.9; 95% confidence interval: 1.6 to 9.7; P=0.002). No risk was associated with haplotype 3 of IL1RN, which contains the frequently examined allele 2 variant of the intron 2 VNTR.
Conclusions— We found that IL1RN–H5H5 carriership increases the risk of venous thrombosis.
Interleukin-1 (IL-1) is a multifunctional proinflammatory cytokine that can be produced by nearly all cell types, including monocytes, activated macrophages, and endothelial cells.1 IL-1 plays, in synergy with tumor necrosis factor alpha (TNF-α), a key role in autoimmune and inflammatory diseases by activating the expression of genes associated with the innate and adaptive immune response.2 IL-1 synthesis can be induced by bacterial endotoxins, viruses, antigens, and by other cytokines such as TNF-α and the interferons.3 IL-1 can cause fever, inflammation, and tissue damage. The margin between benefit for resistance and toxicity in humans is extremely narrow.3
The IL-1 superfamily comprises the agonists IL-1α and IL-1β (predominant form in humans), and their antagonist IL-1Ra.4 Both IL-1 agonists can bind to IL-1 receptor type 1 (IL-1R1) and the “decoy” receptor IL-1 type 2 (IL-1R2).5 High affinity binding is only established if bound IL-1α or IL-1β is also bound to the IL-1 receptor accessory protein (IL-1R AcP).6 Complex formation of IL-1α or IL-1β with both IL-1R1 and IL-1R AcP is required for IL-1–induced signaling.7 IL-1Ra also functions as ligand for the IL-1R1 receptor, however signal transduction does not occur because IL-1Ra lacks the binding site for IL-1R AcP.4 IL-1α, IL-1β, and IL-1Ra also bind to IL-1R2. However, this receptor is not capable of signal transduction, because it lacks the toll-like region in the cytoplasmic domain.8 By binding IL-1, IL-1R2 controls the amount of IL-1, which is free to bind the IL-1R1 receptor.
Several studies have provided insight in the molecular events that link inflammation to thrombosis.9,10 IL-1 can affect the coagulation system in various ways. Tissue factor expression is upregulated by proinflammatory cytokines like IL-1, TNF-α, and IL-6.11 Because tissue factor plays a central role in the initiation of coagulation, this suggests a strong link between inflammation and hypercoagulability. IL-1 also promotes coagulation by downregulating the expression of thrombomodulin and endothelial cell protein C receptor, two important components of the protein C anticoagulant pathway.9 Furthermore, IL-1 influences fibrinolysis by increasing the production of plasminogen activator inhibitor and decreasing the production of tissue-type plasminogen activator.9,12 Together this suggests an overall prothrombotic effect for IL-1. This would explain the finding that elevated levels of proinflammatory cytokines, including IL-1β, are associated with the risk of venous thrombosis.13 It is also possible, however, that the inflammatory reaction seen in patients with a history of venous thrombosis represents a postthrombotic phenomenon, because no association was observed in a prospective study.14
We hypothesized that common variations in the genes coding for IL-1β, IL-1Ra, IL-1R1, and IL-1R2 (IL1B, IL1RN, IL1R1, and IL1R2) influence the risk of venous thrombosis by modulating the IL-1 pathway. To test this hypothesis we genotyped 18 single nucleotide polymorphisms (SNPs) in these genes, which together tag 25 haplotype groups, in all patients and control subjects of a case–control study on the causes of deep venous thrombosis, the Leiden Thrombophilia Study (LETS).
The design of the Leiden Thrombophilia Study has previously been described in detail.15 We included 474 consecutively diagnosed patients with an objectively confirmed first episode of deep vein thrombosis and 474 controls, frequency matched for sex and age. Individuals with active cancer were excluded. All patients and controls were of Caucasian descent. The mean age for both groups was 45 years (range 15 to 69 for patients, 15 to 72 for controls). Both groups consisted of 272 (57.4%) women and 202 (42.6%) men. Venous blood was collected into 0.1 volume of 0.106 mol/L trisodium citrate. High molecular weight DNA was isolated from leukocytes by standard methods. DNA samples were available from 471 patients and 471 controls. Plasma samples were available from 473 patients and 474 controls.
IL1B, IL1RN, IL1R1, and IL1R2 were resequenced by SeattleSNPs in 23 subjects of European-American descent.16 This resulted in the identification of 23 SNPs in IL1B, 83 in IL1RN, 68 in IL1R1, and 87 in IL1R2. For each gene, haplotypes were constructed using the unphased SNP data from the 46 chromosomes and the software program PHASE 2.17 We identified the most common haplotype groups of these 4 genes and the 18 SNPs needed to tag these 25 haplotype groups (Table 1). All patients and controls were genotyped for these 18 haplotype tagging (ht) SNPs.
Besides the 18 htSNPs, an additional polymorphism in IL1RN (17163C/T, rs4252041) and an 86-bp variable number of tandem repeats (VNTR) in intron 2 of IL1RN18 were genotyped in selected individuals.
The 13888T/G and 17163C/T SNPs in IL1RN were genotyped by polymerase chain reaction (PCR) followed by restriction fragment length polymorphism analysis. The 86-bp VNTR was genotyped by PCR followed by gel electrophoresis. All other polymorphisms were genotyped using a 5′-nuclease/TaqMan assay.19 PCRs with fluorescent allele-specific oligonucleotide probes (Assay-by-Design, Applied Biosystems) were performed on a PTC-225 thermal cycler (Biozym) and fluorescence end point reading for allelic discrimination was done on an ABI 7900 HT (Applied Biosystems).
Fibrinogen and C-Reactive Protein Levels
Plasma levels of the inflammatory biomarkers fibrinogen and C-reactive protein (CRP) were measured as described before.20
In the healthy controls, Hardy–Weinberg equilibrium for each htSNP was tested by the χ2 statistic. To estimate the degree of linkage disequilibrium (LD) in our study population, we calculated D′ and r2 (measures for LD) between SNPs in IL1B and IL1RN and between SNPs in IL1R1 and IL1R2 using Haploview.21 A Pearson χ2 test was performed to detect differences in SNP allele frequency distribution between patients and controls.
TagSNPs (Version 2)22 was used to estimate the frequency of the haplotypes present in the LETS population. R2h values (measure of the uncertainty in the prediction of haplotypes based on the selected htSNPs) were calculated using the SNP genotypes and the program TagSNPs. Haplotypes with R2h >0.95 were considered to be derived without uncertainty. Subsequently, haplotypes (H) were constructed for each individual (Figure, A). When for an individual more than one haplotype combination was possible, haplotypes were only assigned to that individual when the haplotype combination had a probability >95% based on the results of the TagSNPs program; eg, heterozygotes for IL1R1 haplotypes 1 and 2 (H1H2) and heterozygotes for IL1R1 haplotype 3 and 7 (H3H7) have the same genotype (Figure, A), but the TagSNPs results indicated that H1H2 is much more likely (probability=99%).
For further analyses we excluded carriers of haplotypes with a R2h < 0.95 and subjects in whom the haplotype combination could not be assigned with a probability >95%. In addition, all carriers of rare haplotypes were excluded. This resulted in exclusion of 27/471 patients and 44/471 controls for IL1B, 70/471 patients and 74/471 controls for IL1RN and 3/471 patients and 2/471 controls for IL1R2. For IL1R1 no individuals were excluded from the analyses.
A Pearson χ2 test was performed to compare haplotype frequencies between patients and controls. This test will detect additive and dominant effects. Because this test has no power to detect recessive effects, we also performed a Pearson χ2 test on the m+1×2 (patients and controls) table of m categories of homozygous carriers (H1H1, H2H2…., HmHm) and a category containing all heterozygous haplotypes (H1Hx + H2Hx …. + HmHx).
To investigate whether SNPs or haplotypes were associated with venous thrombosis, odds ratios (ORs) and 95% confidence intervals (95%CI) according to Woolf23 were calculated as measure of the relative risk of thrombosis for carriers of the exposure category (eg, H4 carriers) compared with the reference category (eg, non-H4 carriers).
For the risk calculations different reference groups were used for each haplotype. Therefore two additional models were tested to analyze the effect of IL1RN haplotypes on the risk of venous thrombosis. Model 1 was a logistic regression model containing homozygous carriers (excluding H3H3 carriers) and a reference group consisting of H3H3 carriers (H3 is the most common haplotype of IL1RN) and all heterozygous carriers. Model 2 was a logistic regression model containing all 15 IL1RN haplotype combinations (H1H1… …, H1H5, H2H3… …, H5H5) and H3H3 as a reference group.
We have in addition performed an overall recessive test without assigning haplotypes to individuals using the software program Chaplin.24 The effect of the IL1RN haplotypes in a recessive model was also assessed using the program Haplo.stats,25 which also does not assign haplotypes to individuals. H3H3 was used as reference group in this analysis.
For IL1B, IL1R1, and IL1R2, the same haplotype analyses were performed as described above for IL1RN. None of the haplotypes of these three genes were associated with venous thrombosis.
For the analysis of the association of haplotypes with fibrinogen and CRP levels, levels were logarithmically transformed. For each haplotype means with 95%CI were calculated.
Haplotype Tagging SNPs
From the data of SeattleSNPs, we selected 18 SNPs (Table 1) which together tag the 6 most common haplotype groups of IL1B, the 6 most common haplotype groups of IL1RN, the 7 most common haplotype groups of IL1R1, and the 6 most common haplotype groups of IL1R2 (Figure, A). For all htSNPs the distribution of genotypes among control subjects was in Hardy–Weinberg equilibrium, except for SNP 12974C/T (P=0.049). Previous studies indicated LD between SNPs in IL1B and IL1RN.26,27 However, Haploview analysis showed that in our population the degree of linkage disequilibrium (measured as D′) was low between SNPs in IL1B and IL1RN and between SNPs in IL1R1 and IL1R2 (Figure, B). D′ values were high within the genes (Figure, B), indicating that recombination events are rare in these genes. This confirmed the validity of our approach to construct haplotypes over a complete gene. We found low r2 values between the selected SNPs in IL1B (r2 ranging from 0.02 to 0.24), IL1RN (r2 ranging from 0.03 to 0.30), IL1R1 (r2 ranging from 0.02 to 0.30) and IL1R2 (r2 ranging from 0.004 to 0.24), indicating that the SNPs are indeed haplotype specific.
Table 1 shows the allele frequency distribution in patients and controls for all 18 SNPs. For all SNPs, no significant difference in allele frequency between patients and controls was found (data not shown). The risk of venous thrombosis was calculated for all 18 SNPs (supplemental Tables I to IV, available online at http://atvb.ahajournals.org). An increased risk of venous thrombosis was found for homozygous allele A carriers of the IL1B intron 4 SNP 5200G/A (OR=1.4; 95%CI:0.9 to 2.1; P=0.13), homozygous allele A carriers of the IL1R1 3′ flanking SNP 27421T/A (OR=2.1; 95%CI:0.9 to 4.9; P=0.10), and homozygous allele G carriers of the IL1RN intron 2 SNP 13888T/G (OR=2.8; 95%CI:1.3 to 6.1; P=0.007). No effect on venous thrombosis risk was found for heterozygous carriers of these 3 SNPs. Odds ratios less than 1 were found for carriers (heterozygous + homozygous) of the rare alleles of the IL1RN intron 1 SNP 12602G/A (OR=0.8; 95%CI:0.6 to 1.0; P=0.07), of the IL1R1 exon 3 SNP12544 C/G (OR=0.7; 95%CI:0.5 to 1.0; P=0.08), of the IL1R1 intron 3 SNP12974 C/T (OR=0.8; 95%CI:0.6 to 1.0; P=0.09), and of the IL1R2 intron 6 SNP18072 A/G (OR=0.8; 95%CI:0.6 to 1.0; P=0.11).
In total 25 common haplotype groups (Figure, A) were expected on basis of SeattleSNPs data. TagSNPs analysis showed that, in addition to these 25 haplotype groups, 3 rare haplotypes in IL1B (frequency ranging from 0.07% to 1.0%; R2h<0.79), 8 rare haplotypes in IL1RN (frequency ranging from 0.04% to 1.4%; R2h<0.89), and 2 rare haplotypes in IL1R2 (frequency 0.13% and 0.16%; R2h<0.89) were predicted based on the genotypic data. No additional haplotypes in IL1R1 were present in our population. Analysis in the control subjects of LETS showed that haplotype frequencies (Figure, A) differed only slightly from those reported by SeattleSNPs. This can be explained by the relatively small size of the group studied by SeattleSNPs (46 alleles) compared with our group (1884 alleles).
Haplotypes were constructed from genotype data and assigned to each of the patients and control subjects. All common haplotypes, except H6 of both IL1B (R2h=0.81) and IL1RN (R2h=0.92), had a high R2h value (Figure, A), indicating that the assignment of haplotypes to individuals was performed with sufficiently high certainty.
Overall Test of Association of Haplotypes With Thrombosis
Table 2 shows the frequency distribution in patients and controls for the haplotypes of IL1B, IL1RN, IL1R1, and IL1R2. For all 4 genes, 2 global tests were performed to provide an overall test of association. The additive model showed no significant difference in haplotype frequencies between patients and controls for all 4 genes (Table 2). However, for the recessive model, a significant difference between patients and controls was observed for IL1RN. To investigate the cause of this difference, odds ratios were calculated for the most common haplotype groups of IL1RN (Table 3).
An almost 4-fold increased risk of venous thrombosis (OR=3.9; 95%CI:1.6 to 9.7; P=0.002) was found for homozygous carriers of H5 (H5H5; Table 3). No increased risk was found for heterozygous carriers of H5 (H5Hx).
For these risk calculations a different reference group was used for each haplotype. Therefore we analyzed the effect of IL1RN haplotypes on venous thrombosis also with 2 additional models (see Methods section). For both models, only H5H5 carriership showed an effect on the risk of venous thrombosis (model 1: OR=4.0; 95%CI:1.6 to 9.9; P=0.003; model 2: OR=4.4; 95%CI:1.6 to 12.3; P=0.005).
To demonstrate that the results were not biased by the assignment of haplotypes to individuals, we also performed an analysis using software programs not requiring haplotype assignments to individuals. An overall recessive test, using the program Chaplin24 and all genotypic data, showed a significant difference in haplotype distribution between patients and controls for IL1RN (P=0.005). The effect of the IL1RN haplotypes was tested in a recessive model using the program Haplo.stats.25 An increased risk was found for H5H5 carriers (OR=4.0; 95%CI:1.6 to 9.9; P=0.003). This effect is similar to the risk calculated for H5H5 when haplotypes were assigned to individuals (Table 3).
According to SeattleSNPs,16 one prevalent subhaplotype (31%) is present in the H5 group. Because we found an increased risk in the H5H5 carriers of IL1RN, we determined the prevalence of this subhaplotype in all H5H5 carriers by genotyping the 3′ UTR 17163C/T SNP. The rare T allele was found in 3/46 H5 alleles in patients (frequency=0.06) and 2/12 H5 alleles in control subjects (frequency=0.16). Because of its low frequency we did not genotype the entire study population for this polymorphism.
Heterozygous carriers of H2 had a slightly reduced risk of venous thrombosis (OR=0.7; 95%CI:0.5 to 1.0; P=0.043), which was not influenced by stratification for age or sex.
IL1RN Intron 2 VNTR
The 86-bp intron 2 VNTR is a well-known and frequently genotyped polymorphism in IL1RN. The rare allele, allele 2, has been found to be associated with a broad range of inflammatory diseases.28 To identify the IL1RN haplotype(s) in which this allele is located, we genotyped the VNTR in all homozygous carriers of each of the 6 haplotype groups (n=177). Allele 2 was found in 117/120 H3 alleles and in one H5H5 carrier, being heterozygous for the VNTR. Allele 2 was not present in carriers of H1H1, H2H2, H4H4, and H6H6. These results indicate that allele 2 of the VNTR is part of IL1RN H3.
Markers of Inflammation
Fibrinogen and CRP are markers of inflammation that are expected to be increased in subjects with high IL-1 levels. Fibrinogen and CRP levels were slightly higher in patients than in control subjects.20 In the control subjects, none of the haplotypes had an effect on the fibrinogen or CRP levels (data not shown).
IL-1 is a proinflammatory cytokine, which plays an important role in inflammation by activating the expression of acute phase proteins. IL-1 signaling involves the receptors IL-1R1 and IL-1R2, the antagonist IL-1Ra, and the accessory protein IL-1R AcP. IL-1 influences both coagulation and fibrinolysis, suggesting an overall prothrombotic effect. Whereas others studied association of single IL-1 SNPs with disease, we used a haplotype-based approach to investigate whether common variations in IL1B, IL1RN, IL1R1, and IL1R2 influence the risk of venous thrombosis.
Global testing using a recessive model showed a difference in haplotype frequency between patients and controls for IL1RN (P=0.031). Whereas for most haplotypes no or at most marginal effects were observed, homozygous carriers of H5 of IL1RN had an increased risk of venous thrombosis (OR=3.9; 95%CI:1.6 to 9.7). Caution is needed when interpreting these results because the number of H5H5 carriers (23 patients and 6 controls) and 13888GG carriers (tagging SNP of H5; 25 patients and 9 controls) is low. Therefore, subsequent studies will be needed to determine the validity of this finding.
Although we found an increased risk of venous thrombosis for H5H5 carriers, the functional SNP causing this risk still has to be identified. IL1RN H5 is tagged by the combination of 13888G and 19327A (see Figure). It is unlikely that the functional SNP is 19327G/A, because no association between H4 (tagged by 19327A) and thrombosis risk was found. An obvious candidate for being the functional SNP is 13888T/G, which is unique for H5 and is itself also associated with an increased risk of venous thrombosis. The 13888T/G SNP is located in intron 2 of IL1RN in a highly polymorphic region. This region does not contain any obvious regulatory elements which would predict that 13888T/G is a functional variant. It is also possible that the functional SNP is not 13888T/G, but a SNP in linkage disequilibrium with 13888T/G or a SNP forming a subhaplotype of H5. IL1RN H5 contains a number of subhaplotypes, but the frequencies were too low to investigate their effect on the risk of venous thrombosis in LETS. Future resequencing of H5H5 carriers from LETS may also help to identify candidate functional SNPs in IL1RN H5.
Apart from H5 a rare IL1RN haplotype exists (frequency in control subjects=0.35%, R2h=0.81) which is tagged by 13888G (not listed in Figure). This haplotype was too rare to study its effect on venous thrombosis risk in LETS.
H6 carriers of both IL1B and IL1RN and carriers of 13 rare haplotypes were excluded from our haplotype analysis. Inclusion of these haplotypes did not importantly change the global additive and recessive probability values (Table 2) or the haplotype associated thrombotic risk of IL1RN (Table 3).
Although IL-1β levels were previously measured in our study population,13 we did not include these in our analyses because with the assay approach used, only 64 of 942 individuals had detectable IL-1β levels. Instead, we used the inflammatory biomarkers fibrinogen and CRP. Fibrinogen and CRP levels were not associated with any of the haplotypes.
Few studies have been reported on the association of polymorphisms in IL1R1 and IL1R2 with disease, whereas numerous studies report on effects of polymorphisms in IL1B and IL1RN. We genotyped 2 well known SNPs in IL1B, 794C/T (−511C/T in literature29) and 5200G/A (5810G/A in literature30). Although others did observe risks associated with both SNPs in a broad range of inflammatory diseases,31 we only found a slight increase in venous thrombosis risk associated with 5200G/A, whereas no such association was found for 794C/T. Another extensively studied polymorphism is the intron 2 VNTR in IL1RN.18 We found that allele 2 of this VNTR is part of H3 of IL1RN. Although allele 2 of the VNTR has been associated with many different diseases,28 we did not find an association between H3 of IL1RN, which contains allele 2, and venous thrombosis risk. Interestingly, H3 of IL1RN contains apart from allele 2 of the VNTR about 50 haplotype tagging SNPs, which will make it very hard to identify the functional SNP in this haplotype.
Our haplotype-based approach was limited to the most common haplotype groups of the 4 genes (Figure, A). Rare haplotypes found by SeattleSNPs were not tagged by their own haplotype specific SNP in our study, but instead these haplotypes were incorporated into 1 of the 25 haplotype groups listed in Figure, A. Therefore, we cannot exclude a risk associated with one of these rare haplotypes.
Sources of Funding
This study was financially supported by grant 912-02-036 from the Netherlands Organization for Scientific Research (NWO). The LETS study was supported by grant 89-063 from the Netherlands Heart Foundation.
Original received July 11, 2006; final version accepted March 19, 2007.
Mizel SB. The interleukins. FASEB J. 1989; 3: 2379–2388.
Dinarello CA. Induction of interleukin-1 and interleukin-1 receptor antagonist. Semin Oncol. 1997; 24: S9-81–S9-93.
Greenfeder SA, Nunes P, Kwee L, Labow M, Chizzonite RA, Ju G. Molecular cloning and characterization of a second subunit of the interleukin 1 receptor complex. J Biol Chem. 1995; 270: 13757–13765.
Cullinan EB, Kwee L, Nunes P, Shuster DJ, Ju G, McIntyre KW, Chizzonite RA, Labow MA. IL-1 receptor accessory protein is an essential component of the IL-1 receptor. J Immunol. 1998; 161: 5614–5620.
Colotta F, Re F, Muzio M, Bertini R, Polentarutti N, Sironi M, Giri JG, Dower SK, Sims JE, Mantovani A. Interleukin-1 type II receptor: a decoy target for IL-1 that is regulated by IL-4. Science. 1993; 261: 472–475.
Grignani G, Maiolo A. Cytokines and hemostasis. Haematologica. 2000; 85: 967–972.
Orbe J, Chorda C, Montes R, Paramo JA. Changes in the fibrinolytic components of cultured human umbilical vein endothelial cells induced by endotoxin, tumor necrosis factor-alpha and interleukin-1alpha. Haematologica. 1999; 84: 306–311.
Nickerson DA. SeattleSNPs. NHLBI Program for Genomic Applications, UW-FHCRC, Seattle, WA. http://pga.gs.washington.edu. Accessed 22–3–2005.
Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005; 21: 263–265.
Sinnwell JP, Schaid DJ. Haplo.stats: statistical analysis of haplotypes with traits and covariates when linkage phase is ambiguous. R package version 1.2.2. http://mayoresearch.mayo.edu/mayo/research/biostat/schaid.cfm.
Witkin SS, Gerber S, Ledger WJ. Influence of interleukin-1 receptor antagonist gene polymorphism on disease. Clin Infect Dis. 2002; 34: 204–209.
di Giovine FS, Takhsh E, Blakemore AI, Duff GW. Single base polymorphism at -511 in the human interleukin-1 beta gene (IL1 beta). Hum Mol Genet. 1992; 1: 450.