Allergic disorders such as asthma, rhinitis and atopic dermatitis constitute a high incidence global health problem that affects an increasing number of individuals who are often faced with a considerable loss in quality of life. When possible, the typical countermeasure is to avoid contact with known allergens. When this is not possible, however, the only option is medical treatment, which currently aims at modulating general immunological/inflammatory responses, often at the price of adverse side effects. A more focused treatment is dependent on a more detailed knowledge of the aetiology of the diseases.

The diseases are complex and depend on multiple poorly defined environmental as well as genetic factors. The genetic component, which is poly- or oligogenic in nature, is considerable, as evidenced by several family and twin studies.1^–5

Several genome scans have been conducted, suggesting a number of susceptibility loci for allergic disease.6 We have obtained evidence of linkage to chromosome Xp22 for atopic phenotypes, including asthma, atopic dermatitis and rhinitis.7^–9 Xp22 harbours two members of the Toll-like receptor family (TLR7 and 8), which play a central role in the immune response. TLRs are key receptors among germline encoded pattern recognising receptors that play a major role in the recognition of microbial pathogens and initiation of the innate and adaptive immune response. They have emerged relatively late on the stage of identified immunological players as their role in pathogen sensing was not really appreciated until 1998, when the gene at the locus responsible for recognising lipopolysaccharide (a cell wall component of gram negative bacteria) was cloned and found to be identical to TLR4.10 Since then, a plethora of microbial derived ligands, collectively known as pathogen associated molecular patterns, as well as a handful of host derived ligands, have been found to trigger TLR signalling through one or more of the 10 TLRs (TLR1–10) so far identified in humans.

TLRs are expressed in various immune cells, including antigen presenting cells and cells that are likely to experience first encounter with incoming pathogens (eg, epithelial cells). Antigen presenting cells and particularly dendritic cells are important in the initial pathogen sensing and initiation of the immune response. Dendritic cells play a critical role in the crossroad between innate and adaptive immunity by controlling signals that direct naive T helper (Th) cells to differentiate into either Th1 or Th2 cells. Whether dendritic cells promote a Th1 or Th2 state, characterised by distinct cytokine profiles, depends to a large extent on which and how TLRs are triggered, which again depends on the type of infection.11 12 In type I allergy and atopy, the Th1/Th2 balance is skewed towards a Th2 state characterised by an IgE hyper-production, and the potential of TLRs for decisive regulation of immune responses has brought considerable focus on these receptors in the search for an aetiological understanding of allergy and atopy.13 In the specific context of asthma and related allergies, attention has been given to TLR2 and TLR4 as receptors of peptidoglycan from gram positive bacteria and lipopolysaccharide, respectively,14 whereas comparably little attention has been given to TLR7 and TLR8.

TLR7 and TLR8 are homologues and have TLR9 as their closest evolutionary relative.15 16 They both have affinity towards single stranded RNA and similar signalling characteristics but differ somewhat in tissue expression patterns.15^–19 TLR7 and TLR8 are not expressed on the cell surface but are found in endosomal compartments (like TLR3 and TLR9) and their activation requires endocytosis of the pathogens. In a study investigating TLR expression in activated monocyte-like THP-1 cells, it was reported that TLR7 and TLR8 had the strongest overall upregulation in expression in response to classic innate immunity cytokines as well as live bacteria.20

We here report the results from a family based association study investigating the positional and functional candidate genes TLR7 and 8 in relation to asthma, rhinitis, atopic dermatitis and increased specific IgE (positive RAST test) in two independently ascertained samples from Denmark.

METHODS

Subjects

Two independently ascertained family samples (A and B) were investigated (table 1). Informed consent was obtained from each participant prior to inclusion in the study. Local ethics committee approval was obtained in all regions where families were recruited.

Sample A included 135 nuclear families with sib pairs who had asthma, who were enrolled from four clinical centres in eastern, western and central parts of Denmark, primarily on the basis of patient records. Criteria for inclusion were recurrent cough, wheezing and dyspnoea as well as a positive metacholine challenge measured using the methods described by Yan et al.21 The affected siblings were between 15 and 45 years of age. Lung function measurements including forced expiratory volume in 1 s and forced vital capacity were performed on all subjects. Blood samples were analysed for allergen specific IgE (CAP phadiatop; Pharmacia Upjohn, Copenhagen, Denmark); a positive test was defined as IgE ⩾0.35 kUA/l for at least one of the allergens. Rhinitis and atopic dermatitis phenotypes were recorded on the basis of questionnaires. Details about the sample have been reported previously.22

Sample B comprised 100 Danish atopic families, including a total of 424 individuals enrolled in the ITA project via four paediatric and one adult outpatient allergy clinic in western Denmark. Recruitment, examination and clinical information have been described previously.7 23 In brief, inclusion criteria were a minimum of two atopic siblings with doctor diagnosed asthma, rhinitis or atopic dermatitis and a reported effect of appropriate medication. All subjects were examined clinically and questionnaire tested by the same doctor who diagnosed the patients according to standard criteria in consensus with a second doctor evaluating the questionnaires only. The most common diagnosis was asthma affecting 70% of the siblings (table 1). Mean age among the offspring was 10.8 years. Blood samples were taken for DNA extraction and serum measurement of specific IgE to 11 common allergens (CAP RAST FEIA; Pharmacia Upjohn); a positive test was defined as above.

Both samples had a male/female ratio not diverging significantly from random distribution.

Sequencing and genotyping

Blood samples were used for preparation of genomic DNA according to standard procedures. In order to increase the chance of identifying causal risk alleles, 10 patients from families showing evidence of linkage to Xp22 in previous studies7^–9 were selected for sequence analysis of TLR7 and 8. Of the 10 patients sequenced, seven were females (ie, 17 X chromosomes were analysed). The TLR genes were screened for sequence variation in the coding sequence and intron/exon boundaries as well as in the promoter region (approximately 1000 bp of the upstream region) by direct sequencing using Big Dye Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, California, USA). Following ethanol precipitation, the sequencing reactions were analysed on an ABI-310 sequenator (Applied Biosystems). Both strands were analysed. The single nucleotide polymorphisms (SNPs) identified were subsequently genotyped in the two family samples, using the SNaP-shot single base extension technology (Applied Biosystems). Standard PCR conditions were used for both sequencing and genotyping. Primers are available on request. The genotypes were scored independently by two experienced investigators and tested for Mendelian inheritance.

Statistical analysis

The program FBAT V.2.0.2c24 25 was used for family based association analysis of single SNPs and haplotypes. In the haplotype analysis, p values were obtained for individual haplotypes as well as for the overall/global distribution of all haplotypes of a specific marker segment. A sliding window approach was applied analysing segments of neighbouring SNPs. Because of prior evidence of linkage observed in both samples,7^–9 the empirical variance (-e option in the FBAT program) was used in all calculations to account for genotype correlation among affected sib pairs that result from linkage, thus testing the null hypothesis “linkage but no association”. The data were analysed using an additive model, which does not assume any specific mode of inheritance and is in general the most sensitive test.24 The additive model is equivalent to the classic transmission disequilibrium test.26 However, power calculations using PBAT27 28 revealed that a recessive model would have the strongest power in our samples. For example, analysis of SNP1 in sample A would have a power of 0.899 under the additive model whereas the recessive model resulted in a power of 1 (assuming a significance level of 0.05, a population prevalence of the disease of 0.1 and a genetic attributable fraction of 0.2). Therefore, we also analysed the data assuming a recessive mode of inheritance. As the majority of tests performed were not independent, the p values reported are not corrected for multiple testing. However, the most significant single SNP and haplotype results survive a Bonferroni correction.

The degree of linkage disequilibrium (LD) between the SNPs in our samples was calculated using the program Haploview, which creates a maximal set of unrelated individuals, using family data only for parent/offspring phasing.29 This program was also used to assess the LD pattern of the X chromosomal region of interest in the CEPH population (CEU) using SNP genotype data from the HapMap project (http://www.hapmap.org/).

In silico analysis

The program MatInspector (www.genomatix.de)30 31</url> was used to analyse potential transcription factor binding sites around SNPs located in the promoter regions. To assess the possible effect of synonymous SNPs, we used FastSNP.32 The SignalP 3.0 server (http://www.cbs.dtu.dk/services/SignalP/)33 was used to predict and analyse changes in the signal sequences for SNPs located in the N terminal end of the protein.

RESULTS

Sequencing of 10 individuals from families showing evidence of linkage to Xp22 revealed seven SNPs in the TLR7 and TLR8 genes (table 2). In addition, two non-synonymous SNPs from the dbSNP database (http://www.ncbi.nlm.nih.gov/projects/SNP/) were added to the study. However, these SNPs were very rare in our samples (table 2). For all SNPs, the genotype distribution was found to be in Hardy–Weinberg equilibrium among the parents in both samples.

Linkage disequilibrium

The nine SNPs investigated spanned a 35 kb region containing the TLR7 and TLR8 genes located head to tail (fig 1). A relatively low degree of LD was found except for one pair (SNP5–SNP6 in the TLR8 gene) displaying r² values of 0.7–0.9 (table 3), which is consistent with the low LD seen across the region in the CEU population (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Linkage disequilibrium (LD) in the Toll-like receptor (TLR)7–TLR8 region based on the CEU population (HapMap data) represented as r². LD blocks are calculated from D′.34 Arrows mark positions of TLR7 and TLR8 single nucleotide polymorphisms (SNPs) used in this study (rs5743781 and rs5744077 are not included as they are not polymorphic in the CEU population).

Association analysis

Association with asthma and related phenotypes was found for SNPs and haplotypes in both independent samples. SNP1 and 9 corresponding to rs179008 (TLR7) and rs2407992 (TLR8), respectively, showed the strongest and most consistent association (table 4 and supplementary table 1 online), both showing association with all tested phenotypes. In each sample, significant associations were observed for SNP9 (TLR8) in asthma whereas SNP1 showed significant associations only in sample B (and merely a trend towards association in sample A). Overall, the most significant results were found for SNP9 (TLR8) in asthma (p = 0.00023) and atopic dermatitis (p = 0.00066) in the two samples combined under a recessive model.

Because TLR7 and TLR8 are located on the X chromosome, we analysed boys and girls separately as well as combined. This revealed some differences in the association pattern. Association with SNP1 was strongest in boys while association with SNP9 seemed more or less the same in boys and girls, although the girls produced the strongest signal in sample A. Evidence of association was also seen for rs5741883 (SNP4 in TLR8) among girls with rhinitis in sample A (p = 0.033, recessive).

The haplotype analysis assessing adjacent SNPs in a sliding window fashion revealed several significant results some of which are shown in supplementary table 2A and 2B (available online). Overall, the results supported the single SNP analysis. Moreover, significant associations were seen in sample A between TLR7 and increased specific IgE (RAST positive) that were not revealed in the single SNP analysis, thereby providing TLR7 association in both samples for this phenotype. Likewise, significant associations were seen in both samples for identical individual haplotypes encompassing SNPs from both TLR7 and 8 (eg, SNP1–4 haplotypes in rhinitis and positive RAST). Specific haplotype analysis of SNP1 and 9 showed association in all phenotypes. The haplotype comprising the two major alleles (TG) was overtransmitted, most significantly in asthma (p = 0.00012, combined sample, additive model).

In conclusion, replication of allelic association for SNPs (TLR8) or individual haplotypes (TLR7/8) was obtained for all four phenotypes investigated in the two samples.

In silico analyses

SNP1 in TLR7 (rs179008) is a non-synonymous change, altering a Gln residue to a Leu residue in the signal peptide sequence of TLR7 at amino acid position 11. SignalP 3.0 identified distinct N, H and C regions along the signal sequence (H for hydrophobic) and the most probable cleavage site between residue 26 and 27. According to this prediction, the Gln 11 constitutes the last amino acid in the N region within the signal sequence. Changing Gln11 to Leu shortens the N region and extends the H region, indicating that this substitution polymorphism may affect the processing of TLR7.

SNP4 in TLR8 (rs5741883) is located in the promoter region where it affects potential transcription factor binding sites according to the MatInspector program. In the presence of the C allele, a binding sequence for the aryl hydrocarbon receptor (AhR)/AhR nuclear translocator (ARNT) heterodimer is formed. The AhR is implicated in mediating responses to various xenobiotics, such as dioxin and polychlorinated biphenyls, which have been reported to alter the cellular and humoral immune responses.35 In the presence of the T allele, a binding sequence for the Atp1a1 regulatory element binding factor 6 is formed. This transcription factor has been shown to regulate the expression of interleukin 2,36 which has recently been associated with allergic diseases and atopy.37

The synonymous TLR8 SNP9 (rs2407992) was analysed using FastSNP. According to this, there is a medium to high risk of rs2407992 affecting the splicing by abolishing a splice enhancer domain.

DISCUSSION

In this study, we have examined the possible role of TLR7 and TLR8 on the X chromosome in the aetiology of asthma and related disorders by conducting a family based association study of two Danish samples involving 540 and 424 individuals, respectively. In the two independently ascertained family samples, replicated association was obtained for SNPs or haplotypes of TLR7 and TLR8, suggesting these genes as novel disease genes for asthma and related disorders.

The general pattern of allelic and haplotypic transmission distortion was the same for the two samples and the four phenotypes investigated. However, some difference regarding the degree/significance of the transmission distortions was evident (table 4 and supplementary table 2A, B online). This could be because of sample heterogeneity originating from demographic and clinical differences between sample A and B. The siblings in sample A were older than 15 years whereas the probands in sample B were primarily children. Other inclusion criteria and diagnostics were not completely the same either. Thus even though the diagnoses are the same, the susceptibility alleles might not be of equal significance in the two samples.

Furthermore, some gender differences were observed. In particular, SNP1 and TLR7 haplotypes showed most significant association in boys, suggesting a predominantly recessive effect of these variants.

The region harbouring TLR7 and 8 is characterised by a very low degree of LD (fig 1, table 3), indicating that the associations observed represent separate independent signals in TLR7 and 8, respectively. For instance, the two SNPs showing the most significant results, SNP1 (TLR7) and SNP9 (TLR8), display an intermarker r² of around 0.01. Furthermore, LD analysis of data provided by the HapMap project shows that there is practically no LD between the TLR7 and 8 genes and flanking genes, making the possibility that the signals observed are caused by LD to neighbouring disease genes highly unlikely.

SNP1 (rs179008, TLR7) is non-synonymous changing a Gln to a Leu in the signal sequence, thereby extending the H region and shortening the N region, as mentioned. In general, there is a high degree of degeneracy in endoplasmic reticulum (ER) signal sequences, which at first glance may reflect a high versatility of the translocation and ER import machinery. However, a number of studies suggest that correct processing of proteins at the ER membrane depends on protein specific sequence specificity within the signal sequences. Thus mutation of the signal sequence has been shown to affect secretion and translocation efficiencies,38 39 post translational processing events such as glycosylation38 and even membrane integration and topology.40 41 Thus even though our theoretical knowledge is not yet sufficient to accurately predict the consequences of the rs179008 variation, it is likely that it somehow affects the processing of TLR7 at the ER membrane.

Likewise, the in silico analyses suggested functional impacts of SNP4 and SNP9 on TLR8 expression and splicing, respectively. These results support the causal relationship and should be scrutinised experimentally. Alternatively, the SNPs might be LD proxies for causal SNPs that were not detected in the sequence screening. However, analysing HapMap SNPs, no (common) SNPs in tight LD with the associated SNPs of the present study seem to be of likely functional significance (data not shown).

Interestingly, two recent studies have found rs179008 (SNP1, TLR7) associated with age related macular degeneration and chronic hepatitis C virus infection, respectively, thus indicating that this variant may be involved in susceptibility to a broad range of inflammatory phenotypes.42 43 Furthermore, an SNP in intron 1 of TLR7 has been reported to protect against advanced inflammation and fibrosis in male patients with chronic HCV infection.44

Our results strongly suggest that TLR7 and TLR8 are novel risk genes in asthma and related atopic disorders. Several non-genetic lines of evidence support the involvement of these two TLRs in immune related diseases and suggest these two TLRs as potential drug targets. Of particular interest, it has recently been shown in murine models that the synthetic TLR7/TLR8 ligand resiquimod (R-848, S-28463) is effective in preventing the development of the asthmatic phenotype as well as inhibiting an inflammatory reaction in response to allergens when the asthmatic phenotype has been established.45^–47 Considering the results of the present study, treatment with TLR7/8 ligands should be causally directed and holds promise for genetically based individualised treatment of these common disorders.