Genetics of complex respiratory diseases: implications for pathophysiology and pharmacology studies

Abstract

There has been a huge influx of data on the genetics and genomics of respiratory diseases in the last few years. Powered by large sample sizes from collaborations worldwide, recent genome-wide association studies have convincingly implicated variants in different regions in the genome for association with complex respiratory traits. These new associations have the potential to offer invaluable insight into the pathophysiology of the normal and diseased respiratory system. The functional mechanisms underlying effects of both identified and novel variants will be the focus of research over the next few years. The identification of these mechanisms will not only increase our understanding of disease but may allow the development of new therapies to alleviate respiratory conditions. The implications of these approaches for studies of asthma and Chronic Obstructive Pulmonary Disease are covered in this review.

LINKED ARTICLES

This article is part of a themed issue on Respiratory Pharmacology. To view the other articles in this issue visit http://dx.doi.org/10.1111/bph.2011.163.issue-1

Background

It is well established that the risk of developing common respiratory diseases such as asthma and Chronic Obstructive Pulmonary Disease (COPD) is influenced by both genetic and environmental factors (Wang et al., 2005). These influences are also important in quantitative traits such as lung function, which are important in such diseases. While the environmental factors underlying the development of asthma and COPD are reasonably well understood, with smoking for example as the main risk factor for COPD, until recently, less has been known about genetic factors underlying these conditions. It is clear from heritability studies that genetic factors play a major role in their development. For example, in asthma, twin studies have shown higher concordance rates in monozygotic than dizygotic twins (Duffy et al., 1990), and there is a fourfold to fivefold increased prevalence in first-degree relatives. Heritability estimates range from 40% to 60% (Bosse and Hudson, 2007). Lung function measures are also highly heritable, with estimates for heritability reaching as high as 77% for forced expiratory volume in 1 second (FEV₁) (Hubert et al., 1982; McClearn et al., 1994). Airflow obstruction (defined by a reduced FEV₁ value and a reduced FEV₁/forced vital capacity (FVC) ratio) is a feature of both asthma and COPD, with fixed (i.e. nonreversible) airflow obstruction being a key diagnostic criterion for COPD (Rabe, Hurd et al., 2007).

Identifying the genes underlying respiratory diseases is of major importance for a number of reasons. First, this will help us to understand more fully the pathophysiology underlying the development of disease and the normal functioning of the respiratory tract. Second, it may facilitate the development of novel treatment strategies based on newly identified drug targets. Third, by identifying a set of risk and safety genetic variants, we may be able to identify better ways to either prevent disease by improving risk assessment, or to make an earlier, more accurate diagnosis. In addition, we may be able to use genetic information to substratify disease into specific phenotypes which may respond differently. Finally, using genetic data, we hopefully will be able to tailor medicines to individuals who are more likely to benefit and less likely to develop adverse events (a subject area known as pharmacogenetics). The aims of this review are to describe recent advances in the genetics of airway disease, focusing on asthma and COPD, and to discuss the implications of these advances for pharmacologists.

Over the last 10 years, there has been a revolution in the ability to identify underlying genetic factors responsible for the development of common complex diseases. This has been driven by the completion of the human genome project and the advent of novel high throughput platforms to aid extensive genotyping studies. Before the completion of the human genome project, two main methods were used for disease susceptibility gene identification: the genome-wide linkage approach and the candidate gene approach.

Linkage and candidate gene approaches

In genome-wide linkage scans, family members are genotyped with evenly spaced genetic markers covering all chromosomes. These are typically microsatellites: polymorphic DNA loci that consist of repeating units of 1–4 base pairs in length. A search is then made for genetic regions containing a higher-than-expected number of shared alleles among the affected individuals. If such a region is discovered, genes in the region become candidates for positional cloning and fine mapping (Baron 2001; Vercelli 2008), in which the region is examined by typing denser collections of single nucleotide polymorphisms (SNPs), the most common genetic variation in humans, in which one nucleotide base is substituted with another) (Bosse and Hudson, 2007). The regions identified by this approach were usually large, and a practical difficulty was that large family cohorts were hard to recruit, especially in late onset diseases like COPD. This approach has also proven generally underpowered to detect linkage where the underlying genetic risk factors were of modest magnitude (Risch and Merikangas 1996; Risch 2000). Nevertheless, there have been some success stories in asthma including the identification of ADAM33 (Van Eerdewegh et al., 2002), DPP10 D (Allen et al., 2003), PHF11 (Zhang et al., 2003), GPRA (Laitinen et al., 2004), HLA-G (Nicolae et al., 2005), CYFIP2 (Noguchi et al., 2005), IRAK-M (Balaci et al., 2007), OPN3(White et al., 2008) and PLAUR (Barton et al., 2009). Full gene names are listed in Table 1.

Table 1.

Gene symbols reported in the review and their full names

Gene symbol	Gene name
ADAM33	ADAM metallopeptidase domain 33
DPP10	dipeptidyl-peptidase 10 (non-functional)
PHF11	PHD finger protein 11
GPRA (NPSR1)	neuropeptide S receptor 1
HLA-G	major histocompatibility complex, class I, G
CYFIP2	cytoplasmic FMR1 interacting protein 2
IRAK-M	interleukin-1 receptor-associated kinase 3
OPN3	opsin 3
PLAUR	plasminogen activator, urokinase receptor
ADRB2	adrenergic, beta-2-, receptor, surface
CCL11	chemokine (C-C motif) ligand 11
CCL5	chemokine (C-C motif) ligand 5
CD14	CD14 molecule
CYSLTR2	cysteinyl leukotriene receptor 2
EDN1	endothelin 1
FCER1B(MS4A2)	membrane-spanning 4-domains, subfamily A, member 2 (Fc fragment of IgE, high affinity I, receptor for; beta polypeptide)
GSTP1	glutathione S-transferase pi 1
IL10	interleukin 10
IL13	interleukin 13
IL4	interleukin 4
IL4 R	interleukin 4 receptor
ITGB3	integrin, beta 3 (platelet glycoprotein IIIa, antigen CD61
LTA	lymphotoxin alpha (TNF superfamily, member 1)
NAT2	N-acetyltransferase 2 (arylamine N-acetyltransferase)
NOD1	nucleotide-binding oligomerization domain containing 1
PAFAH (PAFAH1B1)	platelet-activating factor acetylhydrolase 1b, regulatory subunit 1 (45 kDa)
PTGDR	prostaglandin D2 receptor (DP
TLR9	toll-like receptor 9
TNF	tumor necrosis factor
UGB (SCGB1A1)	secretoglobin, family 2A, member 1
VDR	vitamin D (1,25- dihydroxyvitamin D3) receptor
ORMDL3	ORM1-like 3 (S. cerevisiae)
GSTO2	glutathione S-transferase omega 2
IL6R	interleukin 6 receptor
CHI3L1	chitinase 3-like 1 (cartilage glycoprotein-39)
PDE4D	phosphodiesterase 4D, cAMP-specific
ADRA1B	adrenergic, alpha-1B-, receptor
PRNP	prion protein
TLE4	transducin-like enhancer of split 4 (E(sp1) homolog, Drosophila)
CHCHD9	coiled-coil-helix-coiled-coil-helix domain containing 9
IL1RL1	interleukin 1 receptor-like 1
IKZF2	IKAROS family zinc finger 2 (Helios)
GATA2	GATA binding protein 2
IL5	interleukin 5 (colony-stimulating factor, eosinophil)
SH2B3	SH2B adaptor protein 3
DENND1B	DENN/MADD domain containing 1B
CRB1	crumbs homolog 1 (Drosophila)
RAD50	RAD50 homolog (S. cerevisiae)
HLA-DQB1	major histocompatibility complex, class II, DQ beta 1
HHIP	hedgehog interacting protein
CHRNA 3	cholinergic receptor, nicotinic, alpha 3
CHRNA 5	cholinergic receptor, nicotinic, alpha 5
GSTCD	glutathione S-transferase, C-terminal domain containing
TNS1	tensin 1
HTR4	5-hydroxytryptamine (serotonin) receptor 4
AGER	advanced glycosylation end product-specific recepto
THSD4	thrombospondin, type I, domain containing 4
GPR126	G protein-coupled receptor 126
ADAM19	ADAM metallopeptidase domain 19
FAM13A	family with sequence similarity 13, member A
PTCH1	family with sequence similarity 13, member A
PID1	patched 1
IREB2	iron-responsive element binding protein 2
HLA-DQB1	major histocompatibility complex, class II, DQ beta 1
IL1RL1	interleukin 1 receptor-like 1
IL18R1	interleukin 18 receptor 1
IL33	interleukin 33
SMAD3	SMAD family member 3
IL2RB	interleukin 2 receptor, beta
GSDMA	gasdermin A
GSDMB	gasdermin B
BICD1	bicaudal D homolog 1 (Drosophila)

Nonstandard abbreviations used in the paper: Genes reported in the review are listed along with their full name in Table 1.

The candidate gene approach, on the other hand, mostly utilizes population-based cohorts using a case control design. The underlying principle is to look for a significant difference in the frequency of genetic markers in a gene of interest between the two groups. If association is observed, then it suggests that the marker identified is either causally related to the disease or phenotype of interest, or is in Linkage Disequilibrium (LD) with a causative polymorphism (Rothman et al., 2001). LD occurs when genotypes at two adjacent loci are not independent of each other because of the low probability of recombination events occurring within small genetic distances (Slatkin 2008). The gene choice for these studies is usually based on our knowledge of the function or pathway of that gene and its relevance to disease, and hence does not generally directly lead to discoveries of new biological pathways. Many candidate gene study findings have been hard to replicate (for example, see references Hersh et al., 2005; Smolonska et al., 2009 for reviews of the COPD literature). This is probably a reflection of the modest sample sizes used in many studies which makes them underpowered to detect true associations of modest magnitude. For a comprehensive list of genes identified using this approach in asthma and COPD, the reader is directed to a number of recent reviews (Hersh et al., 2005; Ober and Hoffjan 2006; Vercelli 2008; Zhang et al., 2008; Smolonska et al., 2009; Weiss et al., 2009). Table 2, adapted from reference Weiss et al. (2009), presents a list of candidate genes that have been associated with the asthma phenotype in at least three studies of sample sizes greater than a total of 300 subjects (150 cases and 150 controls).

Table 2.

Susceptibility genes for asthma and related traits using candidate gene approach. The genes that have been associated with the asthma phenotype and reported in at least three independent studies with sample sizes greater than 150 cases and 150 controls, and replication with the same single nucleotide polymorphism (SNP)

Gene	Reference sequence	Total populations showing SNP association with asthma
ADRB2	chr5	5
CCL11	chr17	3
CCL5	chr17	3
CD14	chr5	4
CYSLTR2	chr13	3
EDN1	chr6	3
FCER1B	chr11	9
GSTP1	chr11	8
IL10	chr1	4
IL13	chr5	8
IL4	chr5	11
IL4 R	chr16	7
ITGB3	chr17	3
LTA	chr6	3
NAT2	chr8	3
NOD1	chr7	4
PAFAH	chr6	3
PTGDR	chr14	5
TLR9	chr3	3
TNF	chr6	17
UGB (CC10)	chr11	4
VDR	chr12	3

Adapted from reference (Weiss, Raby et al., 2009).

In October 2010, the Human Genetic Epidemiology Navigator database (HuGE Navigator, Yu et al., 2008) listed 674 and 519 genes reported to be associated with asthma and COPD, respectively, and their related traits. However, as discussed above, there has been a major problem in the replication of many of these findings.

Genome-Wide Association Studies (GWAS) approaches

The completion of the human genome sequencing led to the identification of a vastly expanded list of SNPs and also allowed the documention of the extent of linkage disequilibrium across the human genome in four populations from different ethnic backgrounds (the HapMap project, International HapMap 2005). Using this information, and taking advantage of the technological developments in dense SNP genotyping chips, it became feasible to conduct GWAS (Wellcome Trust Case Control 2007). The GWAS approach relies on the use of a dense set of SNPs giving coverage across the majority of the human genome to survey common genetic variation for a possible role in disease or to identify the heritable quantitative traits that underlie disease (Hirschhorn and Hirschhorn, 2005). By definition, this is a hypothesis-free approach that enables the discovery of novel disease associated genes and molecular pathways.

The era of GWAS in respiratory disease began in 2007 (Table 3), when the first asthma GWAS was published (Moffatt et al., 2007). This study reported association of childhood asthma with ORMDL3, a gene of unknown function at the time. Several studies in different asthmatic populations have followed, mostly in children, replicating the findings (Bouzigon et al., 2008; Galanter et al., 2008; Hirota et al., 2008; Tavendale et al., 2008; Bisgaard et al., 2009). The first GWAS to investigate association with quantitative pulmonary function measures was also reported in 2007. The study proposed a potential role for GSTO2 and IL6R (Wilk et al., 2007). A second GWA study for asthma published in 2008, reported findings of a genetic influence of variants in CHI3L1 on asthma and a chitinase-like protein known as YKL-40 (Ober et al., 2008).

Table 3.

Summery of published genome-wide association study in respiratory diseases and traits

Gene	Phenotype	Sample size and ethnicity	Year	Genome-wide significant findings (Yes/No)	Comments	Ref
ORMDL3	Childhood asthma	Discovery:994 /1,243 * (Caucasians)Replication:2320 subjects from Germany (Caucasians)3301 subjects from the British 1958 Birth Cohort (Caucasians)	2007	Yes	SNPs in the 17q21 region showed a strong association with childhood asthma in both a UK family cohort and German case-control samples. SNPs in this region were also associated with increased ORMDL3 mRNA expression in lymphoblastoid cell lines from asthmatic children.	(Moffatt, Kabesch et al., 2007)
GSTO2 IL6R	Quantitative pulmonary function measures	1097–1222 (depending on phenotype) individuals from the Framingham Heart Study population. (Caucasians)	2007	No	The study utilized data on 71 000 SNPs. Two genes where proposed as potential candidate genes: GSTO2 and IL6R.	(Wilk, Walter et al., 2007)
CHI3L1	Asthma, bronchial hyperresponsiveness and measures of pulmonary function	632 Hutterites(Caucasian)	2008	Yes	Variants associated with elevated serum YKL-40 levels. YKL-40 was previously reported to be increased in the lungs and circulation of patients with severe asthma.	(Ober, Tan et al., 2008)
PDE4D	Asthma	Discovery:359 /846 * (Caucasians)Replication:Ten independent populations with different ethnicities totalling 18 891 individuals (4342 cases)	2009	No	Cases from the Childhood Asthma Management Program (CAMP) and genetically matched controls from the Illumina ICONdb public resource. The strongest region of association seen was on chromosome 5q12 in PDE4D	(Himes, Hunninghake et al., 2009)
ADRA1B PRNP DPP10	Asthma	Discovery:464 /471 * (African American)929 asthmatics and their family members (African Caribbean)Replication:994 / 1243* and 207 families (Caucasians)1456/1973*, 200/200*, 264 /186*, 208/179 * (African Americans)	2009	No	None of the SNPs implicated in the discovery population were replicated in two European cohorts and four additional case-control studies of African Americans.	(Mathias, Grant et al., 2010)
TLE4 CHCHD9	Childhood asthma	Discovery:492 Trios (Mexicans)Replication:177 Trios (Mexicans)	2009	No	Cases were children with asthma, predominantly atopic by skin prick test, and their parents using the Illumina HumanHap 550 K BeadChip.	(Hancock, Romieu et al., 2009)
IL1RL1 IKZF2 GATA2 IL5 SH2B3	Plasma eosinophil count	Discovery:9392 (Icelanders)Replication:12 118 (Europeans)5212 (East Asians)	2009	Yes	Variants in IL1RL1, IKZF2, GATA2, IL5, and SH2B3 showed association with eosinophil count. Three SNPs in IL1RL1, IL33 and WDR36 have also showed association with atopic asthma, with the IL1RL1 SNP showing association with asthma as well.	(Gudbjartsson, Bjornsdottir et al., 2009)
DENND1B CRB1	Asthma	Discovery:793/1988* (Europeans)Replication:917/1546* (Europeans)1667/2045* (African Americans)	2010	Yes	The study also implicated the 17q21 locus harbouring ORMDL3	(Sleiman, Flory et al., 2010)
RAD50 IL13 HLA-DQB1	Asthma	473/1892* (Caucasians)	2010	No	Cases from The Epidemiology and Natural History of Asthma: Outcomes and Treatment Regimens (TENOR) and 1892 Illumina general population controls.	(Li, Howard et al., 2010)
HHIP	FEV1/FVC ratio	Discovery:7691 (Caucasians)Replication:835 (Caucasians)	2009	Yes	Individuals from the Framingham Heart Study. This time, by genotyping more SNPs (550 000 SNPs) in more individuals (7691), the study identified associations of SNPs on chromosome 4q31 near HHIP with the percent predicted FEV1/FVC ratio	(Wilk, Chen et al., 2009)
CHRNA 3 CHRNA 5 HHIP	COPD	Discovery:823/810* (Caucasians)Replication:389/472* (Caucasians)2840 Caucasians family members	2009	Yes	The first GWAS for COPD. The study investigated association with ∼500 000 SNPs. The HHIP locus did not reach genome-wide significance.	(Pillai, Ge et al., 2009)
GSTCD TNS1 HHIP HTR4 AGER THSD4	Lung function measures(FEV1 and FEV1/FVC)	Discovery:20 288 (Caucasians)Replication:32 184 direct genotyping (Caucasians)22 092 in silico replication (Caucasians)54 276 total	2010	Yes	SpiroMeta consortium meta-analysis.>2.5 million genotyped and imputed SNPs tested.mRNA expression analysis showed all variants to be expressed in lung tissue.	(Repapi, Sayers et al., 2010)
GSTCD HHIP HTR4 AGER GPR126 ADAM19 FAM13A PTCH1 PID1	Lung function measures(FEV1 and FEV1/FVC)	Discovery:20 890 (Caucasians)Replication:16 178 in silico (Caucasians)	2010	Yes	CHARGE consortium meta-analysis.	(Hancock, Eijgelsheim et al., 2010)
FAM13A HHIP CHRNA3 CHRNA5 IREB2	COPD	Discovery:2940 /1380* (Caucasians)Replication:502/504* and two family-basedcohorts (n = 3808) (Caucasians)	2010	Yes		(Cho, Boutaoui et al., 2010)
HLA-DQ IL1RL1 IL18R1 IL33 SMAD3 IL2RB GSDMA GSDMB	Asthma	Discovery:10 365/16 110 * (Caucasians)	2010	Yes	The GABRIEL (A Multidisciplinary Study to Identify the Genetic and Environmental Causes of Asthma in the European Community) Consortium.Association with the ORMDL3/GSDMB locus on chromosome 17q21 was specific to childhood-onset disease. Only HLA-DR showed a significant genome wide association with the total serum IgE concentration	(Moffatt, Gut et al., 2010)
BICD1	Emphysema (Both qualitative and quantitative)	Caucasian subjects from three cohorts (n = 1586, 435 and 362)	2010	Yes	BICD1 gene were previously associated with telomere length in leukocytes	(Kong, Cho et al., 2010)

^*Number of cases/number of controls.

A number of additional GWAS papers looking at asthma were published in 2009. One focused on the investigation of association with a specific disease subphenotype, eosinophil counts, in the blood of 9392 Icelanders (Gudbjartsson et al., 2009), and showed associations with variants in five genes: additional analyses were presented looking at asthma in these individuals. A second GWAS in Caucasian subjects reported association with variants in PDE4D (Himes et al., 2009). Two further GWAS papers reported association with asthma in different populations. The first investigated susceptibility for asthma in children from the Mexican population and suggested a contribution of TLE4 and CHCHD9 (Hancock et al., 2009). The second studied two independent populations of African ancestry and suggested association with SNPs in ADRA1B, PRNP and DPP10. (Mathias et al., 2010). Momentum has gathered in 2010. A GWAS for childhood asthma suggested a role for variants in DENND1B. Another GWAS for asthma suggests SNPs in the RAD50-IL13 and HLA-DR/DQ regions were associated with asthma (Li et al., 2010).

The first GWAS for COPD was published in 2009 (Pillai et al., 2009). This study identified risk SNPs in two regions. The first was at the alpha subunit of the nicotinic acetylcholine receptor (CHRNA 3/5) locus, a region previously linked to nicotine dependence and lung cancer (Saccone et al., 2007; Berrettini et al., 2008; Thorgeirsson et al., 2008). The second region contains the gene for hedgehog-interacting protein (HHIP). An accompanying manuscript in the same journal reported the second GWAS for lung function measures in the Framingham Heart Study (Wilk et al., 2009), and identified SNPs near HHIP to be associated with the percent predicted FEV₁/FVC ratio (Wilk et al., 2009).

More recently, a third GWAS investigating associations with COPD has identified variants in FAM13A (Cho et al., 2010). A recent GWAS for emphysema, assessed through high-resolution chest computed tomography in individuals with COPD has also implicated variants in BICD1 (Kong et al., 2010).

From the first wave of GWAS published for asthma and COPD, it became clear that the effect size estimates of the identified variants were typically modest (e.g. odds ratio <1.5). This means that very large sample sizes are needed to identify genetic variants of small magnitude with confidence. Many of these early papers published in the field reported associations which, while of potential interest, were not genome-wide significant using conventional cut offs (multiple testing corrections depending on the number of SNPs on the relevant platforms used for genotyping). This has led to the establishment of consortia comprising multiple independent studies combined to allow pooled analyses to be undertaken (Herbert et al., 2006; Rothman et al., 2006; Zeggini et al., 2008).

Genome-wide meta analyses

The SpiroMeta consortium was established to facilitate large-scale meta-analysis of GWAS of lung function from 14 cohorts of European ancestry (Repapi, Sayers et al., 2010). The Heart and Aging Research in Genomic Epidemiology (CHARGE) consortium also undertook a similar analysis of lung function associations (Hancock, Eijgelsheim et al., 2010). The SpiroMeta consortium analysed associations with FEV₁ and FEV₁/FVC (n = 20 288) using >2.5 million genotyped and imputed SNPs [imputation being the process of predicting genotypes that are not directly assayed in a sample of individuals (Li and Abecasis 2006; Marchini, Howie et al., 2007)], followed by meta-analysis of top signals with data both from direct genotyping (32 184 additional individuals) and in silico summary association data from the CHARGE Consortium (n = 21 209) and the Health 2000 survey (n = 883). SpiroMeta results confirmed the reported locus at 4q31 1 (the HHIP locus) and identified associations with FEV₁ or FEV₁/FVC and common variants at five additional loci: TNS1, GSTCD, HTR4, AGER and THSD4. The CHARGE consortium also analysed associations with FEV₁ and FEV₁/FVC ratio (n = 20 890) and evaluated 30 high signal SNPs in 16 178 SpiroMeta participants. The CHARGE study confirmed the HHIP locus and also identified AGER, HTR4 and GSTCD, and in addition, suggested a potential role of five additional genes: GPR126, ADAM19, FAM13A PTCH1 and PID1.

A similar collaborative approach has also recently been undertaken for asthma. A recent large-scale collaborative GWAS investigated SNPs association in 10 365 persons with physician-diagnosed asthma and 16 110 unaffected persons (the GABRIEL consortium: a multidisciplinary study to identify the genetic and environmental causes of asthma in the European Community) (Moffatt et al., 2010). This study identified IL1RL1/IL18R1, HLA-DQ, IL33, SMAD3 and IL2RB in addition to confirming the previously reported ORMDL3/GSDMB region. The association with the ORMDL3/GSDMB locus on chromosome 17q21 was specific to childhood-onset disease. In this study, only HLA-DR showed a significant genome-wide association with the total serum IgE concentration (Moffatt et al., 2010).

A summary of the key findings from these GWAS approaches is shown in Table 3.

Lessons from genetic studies

Given all the recent papers reporting potentially novel genes important in the development of lung disease and relevant subphenotypes, what have we learned from GWAS approaches so far? Many of the genes/loci identified are novel in that they have not previously been linked to the traits investigated and hence, would have been missed using a traditional candidate gene approach. Interestingly, an evaluation in the SpiroMeta general population sample of genes previously reported in candidate gene studies to alter lung function did not suggest a strong contribution of these genes to FEV₁ or FEV₁/FVC ratio (unpublished data) suggesting that many previously reported candidate gene studies may prove to be false positives.

The statistically convincing novel associations seen in some of these large studies not only provide invaluable insights into the pathophysiology of lung disease but also provide insight into the genetic architecture of complex human diseases. Many of the SNPs reported map to introns or to intergenic regions, with no apparent connection to functionality. In fact, a survey of published GWAS papers found associated SNPs to be 45% intronic and 43% intergenic. Nine percent were nonsynonymous, 2% were in a 5′ or 3′ untranslated region, and 2% were synonymous (Hindorff et al., 2009). This, however, is partly a reflection on the choice of SNPs used for genotyping on genome wide platforms. The important question now is where do we go next?

Resequencing of loci showing strong and reproducible associations, to identify the rare (<1% frequency) and (hopefully) functional variants is the obvious next step (Mardis 2008; Metzker 2010). In addition, other relevant polymorphisms need to be identified and characterized (Weiss et al., 2009). These include copy number variations (Wainr et al., 2009), and other polymorphic variants such as insertions and deletions. Then, we need to investigate interactions (epistasis) between variants, and construct the pathways and networks which may underlie functional effects to better explain the observed genotype phenotype associations and answer the important question of why the identified variants explain only a small proportion of the trait heritability, the appropriately called ‘missing heritability’ issue (Altshuler et al., 2008; Frazer et al., 2009; Goldstein 2009; Manolio et al., 2009).

All of this will only form the first step towards maximizing the biological and clinical applications of these findings. The identified variants need to be functionally characterized using integrated approaches. One example of this is the use of comparisons of genetic expression data and genotyping data to look for SNP associations with expression levels of particular genes of interest [the expression quantitative trait locus (eQTL) approach]. This has been used, for example, to try and determine the major genetic influences on the expression of genes identified to be relevant in asthma such as ORMDL3 (mRNA levels) (Moffatt et al., 2007) and CHI3L1 (YKL-40 protein levels) (Ober et al., 2008).

Nonsynonymous coding SNPs (change amino acid) should be considered a priority for future functional work, as their potential effects are easier to interpret. In addition, in vitro assays need to be developed to evaluate the effect of intronic and intergenic variants on gene expression and to help understand short- and long-range genetic control. Developing animal models to understand how these genes function in complex biological systems will no doubt be valuable (Ober et al., 2010). Finally, integrated databases and bioinformatics tools will be needed to make the best use of data from multiple resources. These considerable challenges should lead to a clearer understanding of how the novel genes and pathways identified contribute to the development of respiratory diseases.

Finally, how relevant will this work prove to the pharmacologist? We believe the findings emerging from genetic approaches will form the basis for pharmacological studies for many years to come, given the probable key role for many of the genes identified in the pathophysiology of lung disease. With this in mind, some of the genes already identified are reasonably well understood: an example would be HTR4, which shows association with lung function in both the SpiroMeta and CHARGE studies described above. Not only has HTR4 signalling and function already been studied (Repapi et al., 2010), but there already exists a range of selective agents active at this receptor which will permit early functional studies to be pursued. However, other genes are much more challenging. For example, GST-CD was identified in both the CHARGE and SpiroMeta studies as being important in determining lung function parameters. Little is known regarding the function of this gene, and no selective agents exist which are known to inhibit or activate protein function. The potential challenges of gaining a full understanding of GST-CD function, and similar poorly characterized genes will be a mainstay of respiratory research over the next few years.

Conflict of interest

We have no conflict of interest to declare.