Characterization of a Common Susceptibility Locus for Asthma-Related Traits

See allHide authors and affiliations

Science  09 Apr 2004:
Vol. 304, Issue 5668, pp. 300-304
DOI: 10.1126/science.1090010


Susceptibility to asthma depends on variation at an unknown number of genetic loci. To identify susceptibility genes on chromosome 7p, we adopted a hierarchical genotyping design, leading to the identification of a 133-kilobase risk-conferring segment containing two genes. One of these coded for an orphan G protein–coupled receptor named GPRA (G protein–coupled receptor for asthma susceptibility), which showed distinct distribution of protein isoforms between bronchial biopsies from healthy and asthmatic individuals. In three cohorts from Finland and Canada, single nucleotide polymorphism–tagged haplotypes associated with high serum immunoglobulin E or asthma. The murine ortholog of GPRA was up-regulated in a mouse model of ovalbumin-induced inflammation. Together, these data implicate GPRA in the pathogenesis of atopy and asthma.

Asthma is a complex phenotype with a proven genetic component, and several projects to map susceptibility genes for asthma and related traits have been undertaken (1). The first published genome-wide scan in asthma suggested six tentative genetic loci, among them chromosome 7p, which was then strongly implicated in a study of Finnish and Canadian families and confirmed in West Australian families (24). To positionally clone asthma candidate genes on chromosome 7p, we studied the Finnish Kainuu subpopulation and considered three alternative hypotheses. First, that only one copy of the susceptibility allele may have survived in this population, with long-conserved haplotypes being observed, as would be consistent with some previous findings (57). Second, there might exist a founder effect, common to many European populations. This would be consistent with the common disease/common variant hypothesis, as observed for example in psoriasis (8, 9). In this case, the carrier frequency should be higher than a few percent, with only a short conserved haplotype (<200 kb) detected (8). Third, numerous mutations might exist in the putative gene, in which case only weak or absent haplotype associations might be detectable.

To distinguish between these hypotheses, we adopted a genotyping scheme whereby we increased the density of markers used, with intermediate analyses to guide further genotyping (Fig. 1A). More specifically, if genetic association analysis suggested that a haplotype occurred in patients more often than in controls, additional markers were genotyped to either exclude or support the identity-by-descent of the haplotypes observed in unrelated patients. For the haplotypes to be identical by descent, all newly typed markers would have to be identically shared between them. The genotyping was done on 86 original genome scan families and an additional 103 trios (all together, 874 subjects) (10). Successive rounds of genotyping and analysis by the haplotype pattern mining (HPM) algorithm (11) suggested the strong association of a conserved haplotype pattern spanning between NM51 and SNP563704, separated by 46 kb (Fig. 1A). The HPM algorithm searches for allele patterns shared between several haplotypes among large sets of unrelated haplotypes.

Fig. 1.

(A) Hierarchical gene mapping strategy. The linkage region of 20 cM implicated by our genome scan (3) was refined by genotyping 76 microsatellite markers in families from Kainuu. We used the HPM algorithm (11) for finding haplotypes associated with high serum IgE. Haplotype patterns spanning 12 microsatellite markers within 3.5 cM were found associated by a permutation test implemented in HPM. At the next round of fine mapping, 10 additional microsatellites implicated a 301-kb haplotype pattern (5 markers yielded the highest associations). A further five microsatellites and 13 SNPs were genotyped next, implicating a 47-kb haplotype pattern (10 markers) between NM51 and SNP563704. All together, a 133-kb region was sequenced around this segment from a homozygous patient with asthma. Eighty polymorphisms were identified by comparison to the public genomic sequence. (B) Plot displaying pairwise linkage disequilibrium (LD) values between 51 markers genotyped in Kainuu trios. The markers are arranged in linear order along both axes starting from lower left corner. The scale from blue to dark red indicates LD ranging from 0 to 1, respectively. LD values shown for patients and controls are essentially identical. A segment of very strong LD between 41 markers is detected. In the lower panel, the same markers were evaluated for association to high serum IgE by a permutation test. (C) Plot displaying LD in French Canadian trios between 22 markers spanning the same segment as in (B). A segment of strong LD has the same boundaries as in Kainuu. In the lower panel, permutation test indicated significant association to asthma in French Canadian trios. (D) Phylogenetic analysis of haplotypes H1 to H7 within a 77-kb segment in Kainuu, North Karelia, and Quebec. The same seven haplotypes occur in all three populations at frequencies >2%. H4 and H5 are the most common risk-associated haplotypes in Kainuu, H7 in North Karelia, and H2 among French Canadians. H1, H3, and H6 are nonrisk haplotypes in all three populations.

To fully explore the genetic variation in associated haplotypes, we sequenced nonrepetitive DNA segments in this interval (from position 506,401 to 638,799 in the public sequence NT_000380; all positions are given with reference to this sequence) in one patient homozygous for the susceptibility haplotype and one control subject homozygous for the most common (nonrisk) haplotype. These sequences were then compared to the public sequence (NT_000380). Two observations emerged from these analyses: first, the patient did not reveal a single instance of heterozygosity, confirming the identity-by-descent of this chromosome segment. Second, comparison of the susceptibility sequence to the public sequence [and the control subject's sequence that differed from NT_000380 only by two single nucleotide polymorphisms (SNPs)] revealed 72 previously unknown SNPs and 8 deletion or insertion polymorphisms (DIPs) specifying the susceptibility haplotype (table S1).

To determine the limits of the critical region, we genotyped additional SNPs in 131 trios [304 high immunoglobulin E (IgE)–associated and 220 control chromosomes], yielding a total of 51 SNPs between 490,331 and 691,245 base pairs (bp) and an average marker density of 4 kb (tables S2 and S3). Analysis of the data by HPM revealed strong association of a conserved 133-kb pattern between 514,743 and 647,327 bp (Fig. 1A). Within this segment, there was strong linkage disequilibrium between the markers (Fig. 1B). A permutation test for association showed P ≤ 0.01 for all 43 markers in this 133-kb segment (10,000 permutations). For comparison, the nominal P value by χ2 association test was 0.00001 for the best associated haplotype pattern.

We subsequently genotyped two additional population samples with subjects that had been diagnosed either for asthma (from Northeastern Quebec, Canada) or high IgE (from North Karelia, Finland) and corresponding family-based controls. We used 22 SNPs among Quebec families (514 asthma-associated and 258 control chromosomes) and 29 SNPs among North Karelian families (75 high IgE–associated and 49 control chromosomes). A haplotype pattern with the same limits as in Kainuu was also identified in Quebec (Fig. 1C) and North Karelia (12). Most of the SNPs were shared, although some were distinct from those in Kainuu. In the three populations, 13 SNPs across the most conserved 77 kb formed seven alternative haplotypes with frequencies >2%. We next sequenced the 133-kb segment from six additional individuals (including an asthmatic subject from Quebec and a North Karelian with high IgE), each homozygous for a different haplotype. This confirmed that each of the haplotypes was different from each other in SNP composition. To assess the relationships of the haplotypes, we considered a total of 40 SNPs, and a phylogenetic analysis confirmed that the risk haplotypes were closely related and distinct from the nonrisk haplotypes in all three populations (Fig. 1D).

We next tested the hypothesis that the related haplotypes (identified either on the basis of high IgE in Kainuu or North Karelia or asthma in Quebec) together conferred risk in all three populations. The risk haplotypes could be tagged by SNP522363 (allele C, table S1) and indeed associated with significant risk (P = 0.004 for all data combined, all three populations contributing), consistent with the common disease/common variant hypothesis (8). The relative risk for high serum IgE among H4 or H5 carriers in Kainuu was 1.4 (95% confidence interval 1.1 to 1.9, P = 0.01), and for asthma among homozygous H2 carriers in Quebec, 2.5 (95% confidence interval 2.0 to 3.1, P = 0.0009). Corresponding transmission disequilibrium test yielded P = 0.05 for Kainuu families (n = 86 trios). To assess whether genetic linkage to chromosome 7p could be explained by these haplotypes, we considered parent-offspring transmissions and sibling-pair sharing of high IgE in Kainuu families (3). One of the risk haplotypes cosegregated in 26 of 51 transmissions (51%) and was shared in 26 of 40 sibling pairs (65%), suggesting that a majority of the linkage signal was because of the observed risk haplotypes.

These results strongly implicated the 133-kb genomic segment as a susceptibility locus for asthma-related phenotypes. To understand more precisely how the observed genetic variation might influence susceptibility, we examined the DNA segment for specific genes (10). Two genes were identified, one with exons 3 to 5 and the other with exons 3 to 10 lying within the susceptibility haplotype. Structures of both indicated complex alternative splicing of the mRNAs (Fig. 2 and tables S4 and S5), suggesting translation to varying protein isoforms. One of the genes (Fig. 2A) was predicted to belong to the G protein–coupled receptor family, and we named this gene GPRA (for G protein–coupled receptor for asthma susceptibility). The two main transcripts of GPRA (A and B) had alternative 3′ exons encoding proteins of 371 and 377 amino acids, respectively (GenBank AY310326 and AY310327). The sequences of all predicted isoforms of the other gene, named AAA1 (for asthma-associated alternatively spliced gene 1, GenBank AY312365 to AY312373) (Fig. 2B) showed only weak homologies to any previously identified proteins. Transcripts for both genes were detected by Northern blot hybridizations (Fig. 2C). Both genes displayed coding polymorphisms in the asthma susceptibility haplotype. In GPRA, SNP591694 changed an amino acid (Asn107 → Ile107) in the first exoloop lining the putative ligand-binding pocket. Even if the position was functionally important, it was unclear how the altered protein would affect cells and tissues.

Fig. 2.

Gene content around the conserved 133-kb haplotype segment (gray box). (A) The 133-kb segment spans from intron 2 to intron 5 of GPRA. GPRA undergoes alternative splicing with multiple variants; the three longest variants are shown (thin lines joining exons marked E1 to E9b). Exon 2 donor site may join to alternative exon 3 acceptor sites, separated by 33 bp in the same reading frame, and there are two alternative 3′ exons, 9a and 9b. Further splice variants may skip exon 3 or 4 or both, suggesting an involvement of the associated polymorphisms in regulation of splicing and protein isoform production. (B) In the opposite DNA strand, there is a previously unknown gene, AAA1, with at least 18 exons (numbered 1 to 18) with complex alternative splicing. AAA1 spans a total of 500 kb of genomic sequence. Eight exons of GPRA (E1 to E8) are shown for orientation. (C) Northern blot hybridization with a 1285-bp full-length GPRA-A cDNA probe (left) and a mixed splice variant probe for AAA1 (right). A 2.4-kb transcript is visible in all nine lanes (upper arrow) and a 1.8-kb transcript (lower arrow) in four tissues for GPRA. Several alternative transcripts are seen for AAA1 (arrows).

Polyclonal antibodies were raised against the different carboxyl termini of the A and B isoforms of the predicted GPRA protein (10). Immunohistochemical staining of bronchus, gut, and skin sections showed that the A isoform is predominantly expressed by smooth muscle cells, whereas the B isoform was predominantly detected in epithelial cells (Fig. 3). In bronchial biopsies, the isoform patterns were distinct between asthma patients and control samples (Fig. 4A). Most clearly, strong expression of the B isoform in smooth muscle cells in asthmatic airways compared with an absence of such staining in control samples. Staining for the B isoform in epithelial cells varied between healthy individuals (Figs. 3 and 4A) but was consistently stronger in the asthma samples than in controls. The A isoform showed no consistent differences. These results suggested also that one or more SNPs or DIPs in the risk haplotypes might critically alter the balance between the isoforms.

Fig. 3.

Expression of the A (left) and B (right) isoforms of GPRA as assessed by immunohistochemistry. The A isoform is found in bronchial smooth muscle cells, basally in colon epithelium, and in occasional basal keratinocytes. The B isoform stains apical epithelial cells in bronchus and gut and all layers of the epidermis. Negative controls are shown with preimmune sera.

Fig. 4.

(A) Expression of GPRA isoform B in bronchial biopsies from a healthy control (left) and an asthma patient (right). E, epithelium; BM, basement membrane; LP, lamina propria; SM, smooth muscle. (Top) The airway epithelium in the control sample shows only faint staining. In the asthma patient, the epithelium shows Goblet cell hyperplasia and basement membrane thickening typical of asthma and positive immunostaining for GPRA throughout the ciliated cells but not Goblet cells. (Bottom) The asthmatic smooth muscle stains strongly positive for GPRA isoform B, in contrast to the negative finding in control. Results are typical of 8 asthmatic and 10 control biopsies studied. (B) Relative expression levels of Gpra mRNA in lungs from sensitized (n = 7) and control (n = 8) mice after inhaled ovalbumin challenge. Gpra was significantly up-regulated in sensitized compared with control mice. (C) Variable alternative splicing for AAA1 depending on genotype. Reverse transcription polymerase chain reaction spanning exons 6 to 10b of AAA1 was performed on lymphoblast RNA samples genotyped for the susceptibility haplotype. Only noncarriers process normal amount of the exon 6-10b transcript, whereas homozygote and heterozygotes show either absent transcript or smaller splice variants. Beta-actin was used as control in parallel amplifications.

We examined next the potential role of the mouse ortholog of GPRA (Gpra) in a mouse model of ovalbumin-induced lung inflammation (10). In general agreement with the results in human asthma, Gpra mRNA was significantly up-regulated in mouse lung after ovalbumin tests in sensitized compared with nonsensitized mice (Fig. 4B). No ortholog for AAA1 was found in mouse and could thus not be assayed. These data support a role for GPRA in the pathogenesis of asthma and provide an animal model that may be useful to further assess the functions of this protein.

In contrast to the GPRA results, several lines of evidence suggested that AAA1 may not represent a protein-coding gene, although its expression was modified by the haplotype (Fig. 4C). Its longest open reading frame comprised only 74 potential amino acids, and in vitro translation failed to yield a stable polypeptide. Transiently transfected cells did not produce recombinant protein. Polyclonal peptide antibodies detected the antigen but no proteins in Western blots or immunohistochemistry (12).

Recently, a small number of candidate genes influencing susceptibility to asthma have been identified by positional cloning. However, the biochemical mechanisms linking the candidate genes to pathogenetic processes in asthma remain poorly understood (1, 1315). The properties of GPRA make it a strong candidate for involvement in the pathogenesis of asthma and other IgE-mediated diseases, as well as a possible drug target. GPRA might act as a receptor for an unidentified ligand. The putative ligand, isoforms of GPRA, and their putative downstream signaling molecules may define a new pathway critically altered in asthma. GPRA encodes isoforms that are produced in distinct patterns by bronchial epithelial cells and smooth muscle cells in asthmatic and healthy individuals. In addition, it is expressed by gut epithelia and keratinocytes of the skin, suggesting a potential role in a wider spectrum of allergic diseases.

Supporting Online Material

Materials and Methods

Tables S1 to S6

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article