Genome-wide association studies for discovery of genes involved in asthma

Authors

  • LOUBNA AKHABIR,

    1. University of British Columbia (UBC) James Hogg Research Centre, Providence Heart + Lung Institute, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • ANDREW J. SANDFORD

    1. University of British Columbia (UBC) James Hogg Research Centre, Providence Heart + Lung Institute, Vancouver, British Columbia, Canada
    Search for more papers by this author

  • The Authors: LA is a PhD student in the Respiratory Division, Department of Medicine, UBC whose research interests include the genetic basis of asthma. AJS is an Associate Professor in the Respiratory Division, Department of Medicine, UBC whose research interests include genetics of asthma, COPD and disease severity in cystic fibrosis.

  • SERIES EDITOR: DARRYL KNIGHT

Andrew Sandford, UBC James Hogg Research Centre, Room 166, St. Paul's Hospital, 1081 Burrard Street, Vancouver, BC, Canada V6Z 1Y6. Email: andrew.sandford@hli.ubc.ca

ABSTRACT

Asthma is the result of a complex interaction between environmental factors and genetic variants that confer susceptibility. Studies of the genetics of asthma have previously been conducted using linkage designs and candidate gene association studies. Recently, the association study design has been extended from specific candidate genes to an unbiased genome-wide approach: the genome-wide association study (GWAS). To date, there have been 12 GWAS to look for susceptibility loci for asthma and related traits. The first GWAS for asthma discovered a novel associated locus on chromosome 17q21 encompassing the genes ORMDL3, GSDMB and ZPBP2. None of these genes would have been selected in a candidate association study based on current knowledge of the functions of these genes. Nevertheless, this finding has been consistently replicated in independent populations of European ancestry and also in other ethnic groups. Thus, chromosome 17q21 seems to be a true asthma susceptibility locus. Other genes that were identified in more than one GWAS are IL33, RAD50, IL1RL1 and HLA-DQB1. Additional novel susceptibility genes identified in a single study include DENND1BI and IL2RB. Discovering the causal mechanism behind these associations is likely to yield great insights into the development of asthma. It is likely that further meta-analyses of asthma GWAS data from existing international consortia will uncover more novel susceptibility genes and further increase our understanding of this disease.

INTRODUCTION

Asthma is the result of a complex interaction between environmental factors and genetic variants that confer susceptibility. The genetic basis of allergic sensitization, including asthma, has been long recognized1 with the HLA being the first specific chromosomal region implicated.2 The genetic contribution to asthma can be demonstrated by twin studies where monozygotic twins are more concordant for asthma and other allergic traits than dizygotic twins.3 However, monozygotic twins are not completely concordant for these phenotypes, clearly demonstrating the importance of environmental factors.

Studies of the genetics of asthma have been conducted using family-based designs that detect the co-inheritance (linkage) of genetic variants with the phenotype. The power of this approach is that the genes are detected by virtue of their chromosomal location alone and not on the basis of prior knowledge of their function. Therefore, the entire human genome can be screened in an unbiased fashion. Several novel genes such as ADAM33, DPP10, NPSR1 (GPRA), PHF11 and HLA-G were identified as putative susceptibility loci using this approach.4–8 On the other hand, the results of linkage studies have shown poor reproducibility. In a recent meta-analysis of 20 genome-wide linkage studies there were only two chromosomal regions (2p21–p14 and 6p21) that showed significant evidence for linkage in European families, after adjustment for multiple comparisons.9 Linkage studies have good power to detect rare high-risk disease-causing alleles but are less effective at detecting more common risk alleles with modest effect sizes.10 This may explain at least some of the lack of reproducibility observed in these asthma linkage studies.

An alternative approach is to use hypothesis-driven association studies of specific asthma candidate genes.11,12 This is most commonly performed using a case–control design although cohort and family-based approaches are possible. However, there has also been a general lack of reproducibility in these types of studies.13 There are many possible reasons for such inconsistent results and these include small sample sizes, differences in phenotype definition and lack of adequate matching of study subjects for ethnic background.14 However, it is also possible that different populations have different genetic risk alleles and the lack of replication simply reflects the genetic heterogeneity underlying the asthma phenotype. In addition, environmental factors are clearly important in the pathogenesis of asthma and gene by environment interactions are infrequently included in genetic studies.

A factor that has to be taken into consideration in association studies is the phenomenon of linkage disequilibrium (LD), which is the association between alleles at different sites on a chromosome. LD tends to be lower for polymorphisms that are further apart (due to the effect of recombination), although there is not a simple relationship between LD and genetic distance. The pattern of LD across the genome is not uniform and differs between populations of different ancestry. As a consequence of LD, an association of a genetic polymorphism with a disease outcome does not necessarily imply causality. The polymorphism may be in LD with several nearby variants, any one of which could be the causal locus. On the other hand, the presence of LD reduces the number of polymorphisms that have to be assayed in a given chromosomal region as one variant can act as the surrogate for many others.

Most recently, the association study design has been extended from the examination of a specific candidate gene to an unbiased, genome-wide approach.15 It has been estimated that in order to adequately survey the entire genome, a large number of genetic polymorphisms (250 000 to 1 million) is likely to be required.16–18 However, the number of polymorphisms will vary between studies as different populations have different levels of LD. Genome-wide association studies (GWAS) utilize single nucleotide polymorphisms (SNP) as they are the most efficiently assayed type of genetic variant. In the most commonly used approach to GWAS the genotype frequencies at each SNP are compared between cases and controls. However, such a large number of comparisons requires the use of extremely stringent statistical correction to avoid an overwhelmingly large number of false positive results. For example, in a GWAS consisting of 500 000 SNP, to survive a Bonferroni correction requires a P-value < 1 × 10−7 (0.05/500 000). Therefore, very large sample sizes are needed in such studies if genome-wide statistical significance is to be achieved, for example, several thousand cases and controls. On the other hand, the large amount of genetic data generated in these studies allows efficient correction for differences in ethnic background using a subset of ancestry informative markers (polymorphisms that differ markedly in frequency between different populations).

The SNP that are present on GWAS panels are generally selected because they are reasonably common in a reference population, for example, minor allele frequency >5%. This reflects the hypothesis that the genetic architecture that underlies common disease consists of a limited number of polymorphisms that are common in the population, that is, the common disease—common variant hypothesis.19 While the success of GWAS suggests this hypothesis is true there may also be a contribution of rare alleles to common traits.20

Advances in genotyping technology have made GWAS feasible, although still expensive. As large sample sizes are essential the studies are generally carried out by consortia of investigators who pool their resources. In this review we will summarize the published GWAS for asthma and related traits.

GENOME-WIDE ASSOCIATION STUDIES OF ASTHMA AND RELATED TRAITS

Thus far, there have been 12 GWAS to look for susceptibility loci for asthma and related traits. The significant results from these studies are shown in Table 1 and Figure 1 and more details are given in supplementary table 1. Seven of these studies utilized asthma or childhood asthma as the phenotype. The first GWAS for asthma was published in 2007 by Moffatt et al.22 and it demonstrated that hypothesis-driven approaches have limited power to identify susceptibility genes because that study uncovered a novel associated locus on chromosome 17q21 encompassing the genes ORMDL3, GSDMB and ZPBP2. None of these genes would have been selected in a candidate association study based on current knowledge of the functions of these genes. Nevertheless, this finding has been consistently replicated in independent populations of European ancestry and also in other ethnic groups13,24,27–37 (Table 2) and thus chromosome 17q21 seems to be a true asthma susceptibility locus. Only one study found a lack of association and that was in an African American population.27 In general, these replication studies have demonstrated that the associations were stronger for early onset asthma. Two of the studies29,35 showed an interaction of the polymorphisms in the chromosome 17q21 region with cigarette smoke exposure.

Table 1.  Summary of the most significant associations from genome-wide association studies of asthma and related phenotypes
PhenotypeMarkerChromosomeLocation (Build 37.1)Gene (or closest gene)StudyReplication study
  • All associations shown were significant at the genome-wide level.

  • † 

    Replications were not significant at the genome-wide level and were not necessarily with the same phenotype.

  • ANTXR1, anthrax toxin receptor 1; C14orf180, chromosome 14 open reading frame 180; C1orf53, chromosome 1 open reading frame 53; C21orf94, chromosome 21 open reading frame 94; CRB1, crumbs homologue 1 (drosophila); DENND1B, DENN/MADD domain containing 1B; DNAJC1, DnaJ (Hsp40) homologue, subfamily C, member 1; ERO1LB, ERO1-like beta (Saccharomyces cerevisiae); GATA2, GATA binding protein 2; GSDMB, gasdermin B; HLA-DQB1, major histocompatibility complex, class II, DQ beta 1; HNMT, histamine N-methyltransferase; IKZF2, IKAROS family zinc finger 2 (Helios); IKZF3, IKAROS family zinc finger 3 (Aiolos); IL18R1, interleukin 18 receptor 1; IL1RL1, interleukin 1 receptor-like 1; IL2RB, interleukin 2 receptor, beta; IL33, interleukin 33; IL5, interleukin 5 (colony-stimulating factor, eosinophil); MAGEE1, melanoma antigen family E, 1; NRG1, neuregulin 1; ORMDL3, ORM1-like 3 (S. cerevisiae); PRKG1, protein kinase, cGMP-dependent, type I; PROC, protein C (inactivator of coagulation factors Va and VIIIa); RNGTT, RNA guanylyltransferase and 5'-phosphatase; SH2B3, SH2B adaptor protein 3; SMAD3, SMAD family member 3; ZNF618, Zinc finger protein 618; ZPBP2, zona pellucida binding protein 2.

Childhood asthmars27860981197,325,908CRB1Sleiman et al.21 
Childhood asthmars121344091197,392,067CRB1Sleiman et al.21 
Childhood asthmars21119311197,529,218DENND1BSleiman et al.21 
Childhood asthmars17478151197,698,103DENND1BSleiman et al.21 
Childhood asthmars17754561197,733,055DENND1BSleiman et al.21 
Childhood asthmars19245181197,738,327DENND1BSleiman et al.21 
Childhood asthmars17754441197,740,690DENND1BSleiman et al.21 
Childhood asthmars120261831197,813,032C1orf53Sleiman et al.21 
Childhood asthmars109249931236,459,785ERO1LBMoffatt et al.22 
Childhood asthmars6716266269,245,228ANTXR1Moffatt et al.22 
Eosinophil countrs14201012102,957,716IL1RL1Gudbjartsson et al.23 
Asthmars37711662102,986,222IL18R1Moffatt et al.24 
Childhood asthmars81795212128,150,924PROCMoffatt et al.22 
Childhood asthmars37912442138,731,765HNMTMoffatt et al.22 
Eosinophil countrs126192852213,824,045IKZF2Gudbjartsson et al.23 
Eosinophil countrs48578553128,260,550GATA2Gudbjartsson et al.23 
Eosinophil countrs41438325131,862,977IL5Gudbjartsson et al.23 
Asthmars9273349632,625,869HLA-DQB1Moffatt et al.24Li et al.25
Asthmars13209883689,591,638RNGTTMathias et al.26 
Childhood asthmars4512342832,607,874NRG1Moffatt et al.22 
Asthmars134232696,190,076IL33Moffatt et al.24Gudbjartsson et al.23
Asthmars109819559116,591,472ZNF618Mathias et al.26 
Childhood asthmars26667811022,205,676DNAJC1Moffatt et al.22 
Asthmars169135961054,023,829PRKG1Mathias et al.26 
Eosinophil countrs318450412111,884,608SH2B3Gudbjartsson et al.23 
Asthmars426432514104,979,486C14orf180Mathias et al.26 
Asthmars7449101567,446,785SMAD3Moffatt et al.24 
Childhood asthmars9070921737,922,259IKZF3Moffatt et al.22 
Childhood asthmars93032771737,976,469IKZF3Moffatt et al.22 
Childhood asthmars115574671738,028,634ZPBP2Moffatt et al.22 
Childhood asthmars80673781738,051,348GSDMBMoffatt et al.22Sleiman et al.21
Childhood asthmars23054801738,062,196GSDMBMoffatt et al.24
Childhood asthmars23054801738,062,196GSDMBMoffatt et al.22
Childhood asthmars22904001738,066,240GSDMBMoffatt et al.22
Childhood asthmars72163891738,069,949GSDMBMoffatt et al.22
Childhood asthmars47954051738,088,417ORMDL3Moffatt et al.22 
Childhood asthmars20379862129,476,477C21orf94Moffatt et al.22 
Asthmars22840332237,534,034IL2RBMoffatt et al.24 
Childhood asthmars2311978X75,788,843MAGEE1Moffatt et al.22 
Figure 1.

Summary of genome-wide association study results for asthma and related traits. Associated phenotypes are indicated as follows: asthma in plain font, atopy underlined, total IgE in bold font, eosinophil count in italics, TDI-induced asthma in italics and underlined. Genes indicated with an asterisk were associated in more than one study. Only associations with P-values < 1 × 10−5 are shown.

Table 2.  Replication studies of the association of asthma on chromosome 17q21
PhenotypeEthnic groupPolymorphismGenePositionP-valueReference
  1. GSDMB, Gasdermin B; IKZF3, IKAROS family zinc finger 3 (Aiolos); ORMDL3, ORM1-like 3 (S. cerevisiae).

Childhood asthmaEuropeanrs7216389GSDMB38,069,9499 × 10−11Moffatt et al.22
AsthmaEuropeanrs8067378Intergenic38,051,3484 × 10−3Sleiman et al.27
AsthmaEuropeanrs7216389GSDMB38,069,9492 × 10−12Tavendale et al.28
Childhood asthmaEuropeanrs9303277IKZF337,976,4693 × 10−6Bouzigon et al.29
AsthmaPuerto Ricanrs4378650ORMDL338,080,8658 × 10−2Galanter et al.30
AsthmaAfrican Americanrs4378650ORMDL338,080,8651 × 10−3Galanter et al.30
AsthmaMexicanrs12603332ORMDL338,082,8072 × 10−2Galanter et al.30
Childhood asthmaJapanesers7216389GSDMB38,069,9492 × 10−5Hirota et al.31
Childhood asthmaEuropeanrs2872507Intergenic38,040,7634 × 10−2Rogers et al.13
Childhood asthmaMexicanrs4378650ORMDL338,080,8653 × 10−3Wu et al.32
Childhood asthmaEuropeanrs7216389GSDMB38,069,9491 × 10−2Bisgaard et al.33
Childhood asthmaChinesers7216389GSDMB38,069,9492 × 10−2Leung et al.34
AsthmaEuropeanrs2305480GSDMB38,062,1961 × 10−5Flory et al.35
AsthmaAfrican Americanrs8079416Intergenic38,092,7131 × 10−2Flory et al.35
Childhood asthmaEuropeanrs7216389GSDMB38,069,9491 × 10−9Halapi et al.36
Childhood asthmaKoreanrs7216389GSDMB38,069,9495 × 10−3Halapi et al.36
AsthmaEuropeanrs7219923GSDMB38,074,5185 × 10−3Madore et al.37
Childhood asthmaEuropeanrs2305480GSDMB38,062,1966 × 10−23Moffatt et al.24

It is currently unclear which gene(s) in this chromosomal region are responsible for the association with asthma. One way to investigate this is to determine which polymorphisms in the region have an effect on gene expression. Moffatt et al.22 demonstrated that the SNP associated with asthma were also associated with ORMDL3 expression in lymphoblastoid cell lines (derived from B cells). Halapi and coworkers36 determined gene expression in peripheral blood leukocytes and they found that the polymorphism most strongly associated with asthma in the study by Moffatt et al.22 (rs7216389) was significantly correlated with the expression of both the GSDMB and ORMDL3 genes. Expression of these two genes was highly correlated suggesting that they are coordinately regulated. In order to determine the cause of the chromosome 17q21 locus genetic association, Verlaan et al.38 performed an extensive, elegant functional study, which showed alteration of regulation of genes in the region was due to chromatin remodelling occurring differentially by alleles. The data suggested that more than one causal polymorphism exists in this region and these SNP have effects on the regulation of several genes.

The ORMDL3 (ORM1-like 3) gene posed a challenge to researchers39 as its function is unknown and efforts have since been made to uncover the mechanism by which this gene could be involved in the pathogenesis of asthma. The current understanding of its function is that it regulates endoplasmic reticulum-mediated calcium signalling resulting in the unfolded protein response, which is thought to trigger an inflammatory response.40 Another recent study implicated ORMDL3 in the regulation of sphingolipid metabolism.41 Sphingolipids are a major component of cell membranes and are involved in numerous functions such as cell proliferation, signal transduction and apoptosis.42 Whether these functions provide an explanation for the association with asthma remains to be determined. The GSDMB (gasdermin B) gene is expressed in the epithelial cells of the skin and intestinal tract and appears to be involved in tumourigenesis.43 However, GSDMB is also highly expressed in T cells and at low levels in the fetal lung and bronchial epithelium.44ZPBP2 (zona pellucida binding protein 2) encodes for a protein involved in fertilization and possibly has a structural role in the biogenesis of the acrosome during spermatogenesis.45 Thus, based on function and gene expression patterns ORMDL3 and GSDMB both remain candidates for being the causal locus.

The GWAS conducted by Sleiman et al.21 confirmed that the chromosome 17q21 locus was associated with asthma at a genome-wide significance level. These authors also found several SNP associated with asthma that were from a large region with high levels of LD on chromosome 1q31 containing DENND1B and the 3′ end of CRB1. The 1q31 locus association was originally observed in a population of European descent and validated in an independent European population. The same region was associated with asthma in African Americans although interestingly the alternative allele at each polymorphism was associated in this ethnic group. The association of DENND1BI with asthma in this GWAS was entirely novel and is indicative of the usefulness of GWAS in detecting genes that would not have been the subject of a candidate gene approach. While CRB1 does not seem to be an attractive asthma candidate, as its reported roles lie in photoreception, DENND1BI is of great interest. DENND1BI encodes for the DENN/MADD domain containing 1B protein. It is thought to interact with the tumour necrosis factor receptor type 1 to block its signalling and is expressed by natural killer T lymphocytes and dendritic cells, which play a pivotal role in the inflammatory state characteristic of asthma. No functional studies have been performed to elucidate the biology behind this gene's putative role in asthma.

In 2009, two additional GWAS using the childhood asthma phenotype were published.46,47 In the study performed by Himes et al.47 two regions of the genome (on chromosomes 5 and 8) showed associations that approached genome-wide significance. The chromosome 8p21 polymorphism is in the CHRNA2 (nicotinic cholinergic receptor α2) gene. Nicotinic cholinergic receptor genes have been implicated in lung disease as polymorphisms in them were associated with lung cancer48,49 and COPD50,51 and have been reported to influence smoking behaviour in a GWAS meta-analysis for the number of cigarettes smoked per day and smoking initiation.52 The other noteworthy result of the Himes et al. study was the association of PDE4D SNP on chromosome 5p12. The PDE4D (cAMP-specific phosphodiesterase 4D) gene encodes an enzyme that hydrolyses cyclic AMP, and thus is upstream of many important signalling pathways. Furthermore, PDE4D plays an important role in airway smooth muscle contractility.53 Variants in the PDE4D gene were previously associated with stroke (although the association is controversial54) and other disorders.55–57 Several compounds designed to inhibit PDE4D are in clinical drug development as therapeutics for asthma and COPD.58,59 A comprehensive review of these efforts has recently been published.60 Himes et al.47 found that chromosome 17q21 polymorphisms were associated with asthma and in the same direction as reported by Moffatt et al.22 but the P-values fell short of genome-wide significance (P = 0.0007 to 0.02).

In the second 2009 study, Hancock et al.46 studied childhood asthma in 492 trios of Hispanic asthma cases and their parents. As expected from the sample size, no SNP reached genome-wide statistical significance. However, variants on chromosome 9q21 upstream of TLE4 showed some of the smallest P-values for association with asthma in the initial cohort and one of the associations was replicated in an independent population, although it was not significant after Bonferroni correction. Therefore, these data provide suggestive evidence of involvement of this locus in susceptibility to childhood asthma. The TLE4 (Transducin-like enhancer of split 4 (E (sp1) homologue, Drosophila) is an inhibitor of transcription with no previous association with asthma or related phenotypes. As in the study by Himes and coworkers,47 chromosome 17q21 SNP in the ORMDL3 region were associated with asthma but only at modest significance levels (P = 0.01 to 0.04). Two polymorphisms in the region of the PDE4D gene were also associated with asthma in this population (P = 0.02 and 0.04).

Increased IgE levels and eosinophilia are two of the main characteristics of asthma and are highly correlated with severity of disease.61–64 Results for the first GWAS of total IgE were reported in 2008 by Weidinger et al.65 The GWAS population was from Southern Germany and consisted of 1530 individuals. No single SNP showed genome-wide statistical significance. Efforts to follow up the variants that showed the lowest P-value regardless of genome-wide significance were undertaken using a total of 9769 additional samples from four independent populations. Interestingly, the top findings were for SNP in FCER1A, the gene encoding the α chain of the high affinity receptor for IgE on chromosome 1q23. The high affinity receptor for IgE is composed of three subunits α, β and γ; the α subunit is responsible for IgE binding whereas the β and γ subunits mediate the signalling. The gene encoding the β subunit has also been implicated in atopy and asthma in some studies but not others.11 Recently, it has been shown in vitro that the α subunit could function as a single-chain receptor and was able to bind IgE and subsequently internalizes and shuttles the antigen into the lysosomes.66 These new data suggest that the α chain could be sufficient for IgE signalling. Although this GWAS result is not surprising based on the function of the gene and previous candidate studies of FCER1A,67,68 it is reassuring that this gene was the top hit of an IgE GWAS and can be regarded as a validation of the GWAS approach.

Another interesting finding of the Weidinger et al.65 GWAS was the association with SNP rs12368672. This SNP is located approximately 7 kb from the signal transducer and activator of transcription 6 (STAT6) promoter and is in high LD with an intronic SNP in STAT6 (rs324011) that has been shown to have allele-specific effects on STAT6 promoter activity in vitro as well as STAT6 mRNA levels in vivo.69 STAT6 is a transcription factor that plays an important role in activating genes involved in IgE synthesis70 and polymorphisms in this genes have previously been associated with serum IgE levels.71,72

Both FCER1A and STAT6 are genes with solid biological plausibility to influence IgE levels. However, Weidinger et al.65 also identified a novel gene that had no previous data to show involvement in atopy or asthma. RAD50 is located on chromosome 5q31 near the important cytokine gene cluster containing IL13, IL4, IL5, IL3 and GM-CSF: all of which are involved in the inflammatory state present in the asthmatic lung. RAD50 encodes a protein responsible for DNA double-stranded break repair. However, the 3′ end of RAD50 contains several conserved non-coding sequences and enhancer elements, which act as a locus control region for regulation of the neighbouring IL4 and IL13 genes.73

Interestingly, there was a second report associating RAD50 with asthma in the GWAS performed by Li et al. in a Caucasian population with severe or difficult to treat asthma.25 Most probably due to the small sample size (n = 473 cases), no single SNP was significantly associated with the phenotype at genome-wide threshold. The highest association was seen for an intronic SNP in RAD50; additional SNP spanning the region comprising RAD50 and IL13 were associated as well. It is of note that SNP in the locus control region, regulating IL13 and present in the RAD50 gene, were found associated with asthma. It is uncertain whether the asthma and total IgE levels associations are truly caused by variants in the RAD50 gene, in which case efforts should be put towards uncovering yet unknown functions of this enzyme, or are merely surrogates for variants in IL13. Functional follow up of the region is warranted to fill this knowledge gap.

The other major finding of the study by Li et al.25 is the association of multiple SNP in the HLA-DQB1 and HLA-DRB1 genes, part of the HLA class II region. The role of HLA in antigen presentation and thus in the development of allergy is well established and there are published HLA genetic association studies of atopy and asthma from as early as 1973.2 The HLA-DQB1 and HLA-DRB1 regions in particular have been previously associated with asthma74,75 and atopy.76 Li et al.25 also reported marginal associations of GSDMB and FCER1A SNP that had been reported in previous GWAS.22,65

Choudhry et al. performed an asthma GWAS77 with a small sample (96 cases with moderate to severe asthma and 88 controls) of individuals from Puerto Rico. Expectedly, no single polymorphism survived correction for multiple comparisons and thus no genome-wide significance was observed. The most significant associations were observed for SNP on chromosomes 5q23 and 13q13.

In 2009, Gudbjartsson et al. performed a GWAS with blood eosinophil count as the phenotype using an initial cohort of 9392 Icelandic individuals and followed up the top SNP in different populations.23 The most significant association with eosinophil levels in the initial cohort was with a SNP in SH2B3 (a.k.a. LNK) on chromosome 12q, which codes for an adaptor protein thought to be involved in T-cell signalling. In vivo studies suggest that the SH2B3 adaptor protein is implicated in cytokine signalling and hematopoietic homeostasis.78 The associated SNP (rs3184504) was previously associated with susceptibility to celiac disease79 and type-1 diabetes.80 This suggests that SH2B3 protein might be important in general immune system processes potentially common to all immune diseases. This conclusion is supported by the observation that rs3184504 was associated not just with eosinophils but with the levels of several blood cell types.23 Another SNP that was associated at a genome-wide significance level was in the GATA2 gene on chromosome 3q, which encodes a transcription factor involved in hematopoietic cell development and proliferation and seems particularly critical for eosinophil development.81

The results of Gudbjartsson et al.23 also highlighted genes in an essential pathway for Th2 immunity, namely IL1RL1 and the gene encoding its ligand IL33. SNP in the IL1RL1-IL18R1 region were significantly associated with eosinophil count at the genome-wide level and were also very strongly associated with asthma in different populations.23 This region has been previously implicated in asthma, inflammation and a number of immune disorders.82,83 The association of an IL33 SNP with eosinophil level was strong (P < 10−5) but not genome-wide significant. However, the same SNP was also associated with asthma at a similar level of significance.

Two other loci were found significantly associated with eosinophil count using the genome-wide cut-off. The first was the region containing the IL5 gene in the cytokine gene cluster on chromosome 5q. IL5 is a cytokine that plays important roles in the Th2 type immune response that characterizes asthma; it stimulates B cells to increase IgE production and is a major regulating protein for eosinophils.84 The second was a region near the IKZF2 gene (a.k.a. Helios) a member of the zing finger family that is expressed in the thymus and is thought to have a role in T-cell development.85 Another interesting locus, although one that did not survive the genome-wide statistical significance cut-off is on chromosome 5q and contains the WDR36 and TSLP genes.23 Interestingly, the same region was identified in a GWAS for eosinophilic esophagitis.86 WDR36 is thought to participate in the regulation of T-cell activation.87 TSLP (thymic stromal lymphopoietin) is expressed in the airway epithelium and stimulates dendritic cells to promote the differentiation of naïve CD4+ T cells to Th2 cells.88 Previous candidate gene studies have shown association of TSLP variants with asthma and IgE levels.89–92

A GWAS conducted for asthma by Mathias et al.26 was published in 2010 and was conducted in two populations of African-descent population with a total sample size of 1864 individuals. Although the study yielded a number of nearly genome-wide significant loci, there were inherent limitations due to the difference in SNP allele frequencies and probably LD patterns between the two populations studied (Barbados founders and African Americans), although both are of African descent. In addition, the major findings were not successfully replicated in European cohorts. However, a meta-analysis using less stringent statistical thresholds highlighted three putative loci for asthma susceptibility in the combined populations: DPP10, ADRA1B and PRNP. The DPP10 gene on chromosome 2q14 encodes the inactive dipeptidyl peptidase 10 and although this protein possesses no protease activity it is involved in regulation of potassium channels via direct binding.93 It is noteworthy that DPP10 was one of the genes implicated in the pathogenesis of asthma by positional cloning.5 This association has been relatively well replicated in different populations.94,95ADRA1B codes for the α1b adrenergic receptor, which is a member of the G-protein-coupled receptor family. It is expressed in the lung and been shown to induce proliferation of vascular smooth muscle cells.96

The PRNP gene on chromosome 20p encodes the prion protein; this function of this protein is not well characterized but speculative roles include ion transport and signal transduction. It has been linked to inflammation by regulating phagocytosis97 and apoptosis98 and was shown to be under-expressed in alveolar macrophages of allergic asthmatics versus controls.99

Another GWAS has been performed in a Korean population for the phenotype of toluene-diisocyanate-induced asthma.100 Although no single SNP reached genome-wide statistical significance, multiple SNP in the CTNNA3 gene were the top findings of this study. CTNNA3 gene on chromosome 10q22 encodes for catenin (cadherin-associated protein) α3. This protein is involved in cell–cell adhesion and has been linked to Alzheimer disease in some studies.101,102 A SNP in CTNNA3 was among the top hits in a GWAS of nicotine dependence.103

LARGE-SCALE, CONSORTIUM-BASED GWAS FOR ASTHMA

The most recent GWAS for asthma was a large consortium-based study that included ∼10 000 cases and ∼16 000 controls.24 The subjects were all of European descent and were drawn from 23 separate studies, mostly from the European Community. This is by far the largest meta-analysis of asthma and also included total serum IgE as an outcome variable. Sub-analyses were also performed for asthma patients stratified by age of onset (with 16 years of age as the cut-off for childhood asthma). Each subject was genotyped for ∼580 000 polymorphisms.

The analysis of all subjects yielded HLA-DQ as the strongest associated locus for the phenotype of doctor-diagnosed asthma. SNP in the IL18R1, ORMDL3/GSDMB, IL33, SMAD3 and IL2RB genes were the most significant findings after HLA-DQ. The IL18R1 association may be due to its neighbouring gene IL1RL1 as the SNP is in an LD block that includes a group of amino acid-changing polymorphisms in IL1RL1. Interestingly, IL33 was also among the most significant findings of the same study and this cytokine is the ligand for IL1RL1.

In general, the associations were found to be stronger for childhood onset asthma than later onset asthma, particularly for the ORMDL3/GSDMB locus on chromosome 17. In addition, using less stringent statistical criteria there was a novel association with SNP near the SLC22A5 gene, which is adjacent to IL13. Another novel, suggestive association was found with the RORA gene on chromosome 15. There was an association of a SNP in the TSLP gene with severe asthma in one of the cohorts but the association was not replicated in the other severe asthma cohort. Interestingly, the same SNP (rs1837253) had previously been associated with asthma and airway hyperresponsiveness.89,92

In the analysis of total serum IgE the authors identified a novel locus near HLA-DRB1 and confirmed previous associations with SNP in the FCER1A, IL13, STAT6 and IL4R/IL21R genes. In general, the loci that were identified as modulating IgE levels were not the same as those associated with asthma and therefore the authors concluded that elevation of serum IgE is a downstream effect of asthma rather than being in the causal pathway.

It is interesting that this study highlighted the inflammatory signalling axis IL1RL1/IL33. IL33 and its receptor IL1RL1 have been consistently implicated in a wide range of inflammatory and immune disorders including asthma.104 Most importantly, variants in the IL33 and IL1RL1 genes were among the top findings of the eosinophil count GWAS.23 Although the phenotypes of the two studies are different, it is safe to conclude that the IL1RL1/IL33 pathway is of importance and should be targeted for therapeutic or diagnostics efforts. Together with the association of the TSLP variant, the association of IL33/IL1RL1 strongly indicates that the disruption of the airway epithelium signalling pathways and immune homeostasis is a major susceptibility factor to asthma.

STRUCTURAL GENES IDENTIFIED BY GWAS

A number of genes encoding structural proteins have been identified in the GWAS published thus far. In the eosinophil count GWAS,23TNXB (tenascin XB) on chromosome 6p21 was among the genes that approached genome-wide significance level. Tenascin XB is believed to regulate collagen fibril deposition in the extracellular matrix.105 The gene might be involved in systemic lupus erythematosus via its association with complement four genes.106 In the Moffatt et al. study,22ANTXR1 was among the genome-wide significantly associated genes; it codes for anthrax toxin receptor 1 and is thought to play a role in cell adhesion and migration and binds directly to actin.107 In the same study,22NRG1 was identified; this gene encodes for neuregulin 1, which has been reported to play a role in cell adhesion.108 In the Li et al. study,25 variants in COL5A3 and MKLN1 were associated with asthma although not at the genome-wide significance level. These genes encode a member of the collagen family and muskelin 1, respectively. Collagen is one of the major components of the extracellular matrix and is implicated in the airway remodelling observed in asthma109 as an important component causing thickening of the epithelial basement membrane. Muskelin 1 is a cytoplasmic protein involved in regulation of cytoskeleton arrangement; it also participates in cell spreading by facilitating communication between the cytoplasm and the nucleus.110 Castro-Giner et al.111 also identified variants in two genes coding for other members of the collagen family: COL18A1 and COL6A5.

CONCLUSION

There are many published GWAS for asthma and related traits and this phase of research may be drawing to a close. The most consistent finding from these studies is the association of the 17q21 locus with asthma. Many other association studies, targeted specifically at this region, had added to the weight of evidence in favour of this locus harbouring a true asthma susceptibility gene(s). Discovering the causal mechanism behind this association is likely to yield great insight into the development of asthma. Other genes to be identified in more than one GWAS are IL33, RAD50, IL1RL1 and HLA-DQB1. It is likely that further meta-analyses of asthma GWAS data from existing international consortia will uncover more novel susceptibility genes and further increase our understanding of this disease.

ACKNOWLEDGEMENTS

LA is the recipient of a UBC Four Year Doctoral Fellowship and an AllerGen NCE Inc. Canadian Allergy and Immune Diseases Training Award. AS is the recipient of a MSFHR Senior Scholar Award and a Canada Research Chair.

Ancillary