INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Advances in asthma management are likely to depend on a better understanding of how genetic factors influence susceptibility to and outcome in this disease. Although asthma is polygenic, present knowledge suggests that candidate genes include those that determine the key clinical phenotypes of bronchial hyperresponsiveness (BHR) and atopy. BHR, an exaggerated response to bronchoconstrictor stimuli, is demonstrated by virtually all asthma patients and reflects the presence of bronchial inflammation. In general, its severity parallels that of asthma symptoms, although some individuals with BHR remain asymptomatic (1). Atopy, though important, is not alone sufficient to cause asthma. Indeed, although 30 to 50% of the population is atopic, only 5 to 7% will develop asthma (2). Furthermore, since individuals who are atopic are significantly more likely to have increased airway responsiveness (2), any assessment of genetic risk for asthma must consider the possible interaction of these two components in expression of the asthma phenotype.

Chromosomes 5q and 11q exhibit the most consistent associations with BHR and atopy (3). Candidate genes on these chromosomes include the ₂-adrenoreceptor and interleukin-4 cytokine cluster on chromosome 5q31-33 and the high-affinity IgE receptor (Fc RI) and Clara cell secretory protein (CC16) genes that map to chromosome 11q13 (3, 6, 7). However, although chromosome 11q13 is a "hot spot" for asthma-related genes, data on these candidate genes are not sufficiently convincing to explain the strength of the linkage of asthma to this region (3, 8). For example, there are few functional data on the polymorphisms of these genes, the frequencies of some alleles in white individuals are too low to be clinically significant; many of the studies demonstrate stronger associations with atopy than with BHR; and some linkage markers are too distant from these candidate genes to account for the identified associations (3, 8). Consequently, there is a need to identify further polymorphisms that are common, that are associated with BHR, and that complement existing data from linkage studies.

In the context of new candidate genes for asthma, the presence of inflammation in the airway is an important biochemical feature of asthma. Oxidative stress, with the formation of reactive oxygen species (ROS), is a key component of inflammation (2, 9). Although host antioxidant defenses should detoxify ROS, individuals differ in their ability to deal with an oxidant burden, and such differences are in part genetically determined (9). Inability to detoxify ROS should perpetuate the inflammatory process, activate bronchoconstrictor mechanisms, and precipitate asthma symptoms. Accordingly, we propose that members of the glutathione-S-transferase (GST) supergene family are attractive candidates, for linkage to asthma, as enzymes encoded by members of the mu, theta, and pi class GST gene families are critical in the protection of cells from ROS because they can utilize as substrates a wide variety of products of oxidative stress (10). Thus, the enzymes encoded by these GST gene classes preferentially utilize different ROS products. For example, quinone metabolites of catecholamines (dopachrome) are utilized by mu GST (but not by GSTP1 or GSTT1). GSTT1 utilizes oxidized lipid and DNA, and GSTP1 catalyzes the detoxification of base propenals that arise from DNA oxidation. Mu and theta (but not pi) GST demonstrate activity toward a phospholipid hydroperoxide (11). These GSTs may also influence the synthesis of eicosanoids (critical mediators in the asthmatic response) via modulation of ROS levels (10). Further, the ROS-derived products are essential in the mobilization of arachidonic acid, with subsequent production of proinflammatory eicosanoids. So far, common allelic variants have been identified in two of the five mu-class genes (GSTM1 and GSTM3), one of the theta class genes (GSTT1), and the pi class GSTP1 gene (11).

The biochemical significance of these polymorphisms is indicated by data showing that particular genotypes are associated with an increased susceptibility to and/or poor outcome in several inflammatory pathologies. For example, homozygosity for the deleted GSTM1*0 and GSTT1*0 alleles has generally been linked with an increased risk of disease (11). Thus, the frequencies of these genotypes are increased in patients with photosensitive systemic lupus erythematosus (13). Further, GSTM1 null is also linked with increased radiologic progression (Larsen score) in rheumatoid arthritis (RA), and GSTM1 AB is associated with a reduced risk of this disease (12). The GSTP1 gene, located on chromosome 11q13, encodes the predominant cytosolic GST enzyme in lung epithelium (15). Four alleles of this gene have been identified: GSTP1*A (Ile¹⁰⁵-Ala¹¹⁴), GSTP1*B (Val¹⁰⁵-Ala¹¹⁴), GSTP1*C (Val¹⁰⁵-Val¹¹⁴), and GSTP1*D (Ile¹⁰⁵-Val¹¹⁴) (16). Compared with GSTP1*A, enzymes with Val¹⁰⁵ have a sevenfold greater catalytic efficiency for polycyclic aromatic hydrocarbon diol epoxides but a threefold lower efficiency for 1-chloro-2,4-dinitrobenzene (16). The effect of the Ala¹¹⁴ Val¹¹⁴ substitution is unclear, although it may augment the activity of the Ile¹⁰⁵ Val¹⁰⁵ substitution. We recently reported that homozygosity for the Val¹⁰⁵-containing allele is associated with Larsen score in RA patients (12).

We have examined the hypothesis that polymorphism in the GSTM1, GSTM3, GSTT1, and GSTP1 loci is associated with asthma and related phenotypes. We have determined the prevalence of these genotypes in a population of subjects stratified by atopic status (IgE level and skin prick tests) as well as by airway obstruction/reactivity. Because atopy and airway obstruction interfere with bronchial reactivity (2), we included an additional correction for these potential confounding factors in our analysis of BHR challenge test data.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subject Recruitment

Two hundred-two unrelated Northern European white individuals were recruited from the volunteer and patient database established in the Department of Respiratory Medicine of the North Staffordshire Hospital, by advertisements, chest clinics, and general practice surgeries in the local community. All subjects were residents of North Staffordshire, an area recognized as having a stable patient base with relatively little population movement (21). The subjects recruited were either nonatopic and nonasthmatic, atopic but nonasthmatic, or atopic and asthmatic. Subjects with any other past or current disorder, respiratory or nonrespiratory, were excluded. Individuals were diagnosed as nonatopic and nonasthmatic on the basis of having no past history of allergic or respiratory symptoms, negative skin allergy tests, normal IgE levels, and normal lung function indices. Atopic but nonasthmatic individuals were defined by: (1) a personal history of allergies, seasonal rhinitis, eczema, or allergic conjunctivitis; (2) positivity to skin prick tests (skin reaction with a mean wheal diameter of at least 3 mm more than with a saline control) with a panel of seven common aeroallergens (house dust mite, house dust, grass mix, tree pollen, cat fur, dog fur, feathers); (3) total serum IgE level > 100 IU/ml; and (4) no evidence of airway obstruction. Patients were diagnosed as having atopic asthma by: (1) a history of wheezing, cough, dyspnea, and/or chest tightness; (2) spirometric demonstration of airflow obstruction reversible with a -agonist bronchodilator (> 15% change in FEV₁; and (3) positive atopic status as defined earlier. Spirometric measurements were made on all subjects according to the recommendations of the British Thoracic Society. Bronchial responsiveness was assessed with a methacholine challenge test in all individuals exhibiting an FEV₁ > 80% predicted. A washout of short-acting inhaled bronchodilators (12 h) and antihistamines (1 wk) was allowed before the assessment visit. All subjects were lifelong nonsmokers and had not suffered a viral infection within at least the 6 wk preceding the study. The study protocol was approved by the local ethics committee, and all subjects provided written informed consent.

Methacholine Challenge Testing

Aerosols of methacholine were generated with a jet nebulizer according to the dosimeter method (22). A doubling cumulative dose of methacholine ranging from 0.03 to 32.0 mg/ml was administered. The bolus of the aerosol was delivered from a Nebicheck dosimeter (PK Morgan, Kent, UK) over a period of approximately 25 s, with five such boluses given for each concentration of methacholine (total output 45 µl) without intermittent delay. The doses were given at 5-min intervals, and FEV₁ was measured 30 s and 90 s after each inhalation. The test was stopped when FEV₁ had fallen by  20% below FEV₁ baseline levels (methacholine PC₂₀).

Subject Stratification

According to the degree of airway dysfunction, subjects were stratified into four groups, as follows: Group 1: FEV₁ > 80% predicted and PC₂₀ > 16 mg/ml (BHR-negative); Group 2: FEV₁ > 80% predicted and PC₂₀-positive with a dose of methacholine of 8 to 16 mg/ml (borderline BHR); Group 3: FEV₁ > 80% predicted and PC₂₀-positive with a dose of methacholine of 0.03 to 8 mg/ml (BHR-positive); Group 4; FEV₁  80% predicted (severe airway dysfunction) (Table 1). Subjects who have an FEV₁  80% predicted are likely to demonstrate hyperresponsiveness, but are less likely to be able to perform the test. Accordingly, methacholine challenge tests were not performed on subjects in Group 4 (2).

Identification of GST Genotypes

All genotyping assays were performed by workers who were unaware of the clinical status of individual subjects. Polymerase chain reaction (PCR) assays for identifying GST genotypes included one DNA sample (selected at random) of known genotype for each batch of eight samples of unknown genotype, one negative control (no DNA), and molecular weight markers. About 15% of all patient DNA samples were reassayed on at least one occasion and the genotype assignment was confirmed. All genotype assignments were confirmed by an independent, blinded observer examining the agarose gels used for separating the assay products. Leukocyte DNA was extracted from peripheral blood (5 ml) collected into tubes containing ethylene diamine tetraacetic acid, using standard phenol/chloroform procedures. GSTM1 A, B, A/B, and null genotypes were identified with allele-specific primers to exon 7 (12, 14). GSTT1 (null and expresser) genotypes were also identified with PCR as described previously (12, 14). The GSTM3 AA, AB and BB genotypes were identified with primers to exons 6 and 7 (12, 14). The Ile¹⁰⁵ Val¹⁰⁵ substitution in GSTP1 was examined after PCR amplification through the use of primers to exon 5 (12, 17). The resulting 176-bp fragment was digested with Alw261 to identify the A-G transition at nucleotide +313. PCR products from homozygotes for the Ile¹⁰⁵-encoding allele (GSTP1 Ile¹⁰⁵/Ile¹⁰⁵ genotype) comprised the undigested 176-bp fragment, whereas products from homozygous Val¹⁰⁵-encoding individuals (GSTP1Val¹⁰⁵/Val¹⁰⁵ genotype) comprised fragments of 91 and 85 bp. PCR products from heterozygotes (GSTP1 Ile¹⁰⁵/Val¹⁰⁵ genotype) comprised fragments of 176, 91, and 85 bp.

Statistical Analysis

Chi-square tests were used to assess between-group homogeneity (e.g., skin test-positive versus skin test-negative). Because some allele frequencies were small, the StatXact-Turbo (version 3; Cytel Software Corporation, Cambridge, MA) statistical package was used where appropriate. All other statistical analyses where done with the Stata statistical package (version 5; Stata Corporation, College Station, TX). To correct for imbalances between groups in age and gender, logistic regression was used. The Armitage's trend test was used to examine the relationship between genotypes and ordered categories (e.g., degree of BHR). To correct the trend test analyses for potential confounding factors such as age and gender, we applied ordered logistic regression models. Values of p were corrected for potential multiple testing errors through Holm's procedure (23). For some subjects we were unable to obtain a complete data set because they were unwilling to undertake a methacholine challenge test or did not wish to donate a blood sample. In the latter subjects we isolated DNA from a mouthwash sample, but were unable to determine serum IgE levels. In some cases, we were unable to assign a genotype because of failure to amplify DNA. Loss of the above information occurred in a random manner, and was not concentrated in any of the subgroups described.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The characteristics of the recruited subjects are summarized in Table 1.

Association of GST Genotypes with Atopy

The frequencies of the GSTM1, GSTM3, GSTT1, and GSTP1 genotypes in the total cohorts in which we studied associations between these genotypes and skin test positivity and IgE level were similar to those previously reported in subjects from North Staffordshire (12, 14). For each gene, these frequencies achieved Hardy-Weinberg equilibrium. Table 2 shows GST genotype frequencies in relation to atopic indices, skin test positivity, and IgE level. Homozygosity for GSTP1 Val¹⁰⁵ (GSTP1 Val¹⁰⁵/Val¹⁰⁵) was significantly less common in subjects with positive than in those with negative skin tests (p = 0.010). Similarly, subjects with IgE levels > 100 IU/ml demonstrated significantly reduced GSTP1 Val¹⁰⁵/Val¹⁰⁵ frequencies as compared with those with IgE levels  100 IU/ml (p = 0.026) (Table 2). Indeed, all GSTP1 Val¹⁰⁵ homozygotes had IgE levels  100 IU/ml (mean IgE:46 IU/ml) except for one (IgE:2,000 IU/ml). The relative frequencies of the GSTP1 Ile¹⁰⁵ and GSTP1 Val¹⁰⁵ alleles in skin test-negative (0.616 and 0.384, respectively) and -positive subjects (0.694 and 0.306, respectively) were not significantly different from one another (p = 0.117). Corresponding frequencies for GSTP1 Ile¹⁰⁵ and GSTP1 Val¹⁰⁵ in subjects with IgE levels  100 IU/ml (0.655 and 0.345, respectively) and > 100 IU/ml (0.693 and 0.307, respectively) were similarly not significantly different (p = 0.465). No associations between the GSTM1, GSTM3, and GSTT1 genotypes and atopic status were identified (Table 2).

Association of GST Genotypes with Severity of Airway Obstruction/Reactivity

The frequencies of the GSTM1, GSTM3, GSTT1, and GSTP1 genotypes in the study cohort were similar to those previously reported in subjects from North Staffordshire (12, 14). For each gene, these frequencies achieved Hardy-Weinberg equilibrium. Table 3 shows the GST genotype frequencies according to degree of airway dysfunction. The data were inspected according to the stratification described in Table 1 (Groups 1 through 4) to determine whether GST genotypes were associated with severity of airway dysfunction. Trend test analysis showed that the proportion of subjects with GSTP1 Ile¹⁰⁵/Ile¹⁰⁵ increased in parallel with the degree of airway reactivity/ obstruction, whereas the proportion of Val¹⁰⁵ homozygotes demonstrated an inverse trend (Table 3, Figure 1). These data were also analyzed by inclusion of skin test positivity, IgE level, age, and gender in a multivariate ordered logistic regression model. After correction for the foregoing covariates, comparison of the proportion of subjects with GSTP1 Val¹⁰⁵/ Val¹⁰⁵ with the proportion having GSTP1 Ile¹⁰⁵/Ile¹⁰⁵ demonstrated the same trend in relationship to the severity of airway reactivity/obstruction (p = 0.017). In addition, the frequency of GSTP1 Ile¹⁰⁵/Val¹⁰⁵ (compared with that of GSTP1 Ile¹⁰⁵/ Ile¹⁰⁵) also achieved significance when assessed according to decreasing severity of airway reactivity/obstruction (p = 0.043). The relative frequencies of the GSTP1 Ile¹⁰⁵ and GSTP1 Val¹⁰⁵ alleles in the four subject groups stratified by severity of airway reactivity/obstruction were found to be significantly different according to a trend test (p = 0.004) (Table 3). These data further suggest a direct relationship between GSTP1 polymorphism and airway dysfunction that is independent of the association of GSTP1 polymorphism with atopy.

View larger version (39K): [in this window] [in a new window]   Figure 1.   The association of GSTP1 genotype with BHR. Groups were defined as: Group 1 (first bar in each group): percent predicted > 80%/PC20 > 16 mg/ml; Group 2 (second bar in each group): percent predicted > 80%/PC20 8 to 16 mg/ml; Group 3 (third bar in each group): percent predicted > 80%/PC20 0 to 8 mg/ml; Group 4 (fourth bar in each group): percent predicted  80%.

Association of GST Genotypes with BHR

To eliminate the influence of airway obstruction as a potential confounding factor in studying the association of GST genotypes with BHR, we performed methacholine challenge testing in subjects exhibiting a baseline FEV₁ > 80% (Groups 1 through 3, Table 3). Comparison of genotype frequencies in subjects in Groups 2 and 3 with those of controls (Group 1) showed that the frequency of GSTP1 Val¹⁰⁵/Val¹⁰⁵ was significantly lower in patients with BHR than in controls (p = 0.031, odds ratio [OR]: 0.23; 95% confidence interval [CI]: 0.06 to 0.88). Trend test analysis across the three groups showed a significant association between GSTP1 genotype and degree of BHR alone. GSTP1 Val¹⁰⁵/Val¹⁰⁵ was directly related to increasing PC₂₀ (p = 0.050). This association remained significant after correction for other potential confounding factors (age, gender, skin test positivity, and IgE positivity) through the use of multivariate ordered logistic regression (p = 0.027). No significant associations were identified between GSTM1, GSTT1, and GSTM3 polymorphisms and airway dysfunction (Table 3).

Association of GST Genotypes with Asthma

GSTP1 genotype was also significantly associated with asthma (Groups 2 through 4) when compared with controls (Group 1) (Table 4). Thus, GSTP1 Val¹⁰⁵/Val¹⁰⁵ was associated with a sixfold lower risk of asthma than was GSTP1 Ile¹⁰⁵/Ile¹⁰⁵ (p = 0.003, uncorrected OR: 0.16; 95% CI: 0.05 to 0.55). GSTP1 Ile¹⁰⁵/Val¹⁰⁵, as compared with GSTP1 Ile¹⁰⁵/Ile¹⁰⁵, was associated with an intermediate risk that approached significance (Table 4). Correction of the data for potential confounding factors (skin test positivity, IgE > 100 IU/ml, age, gender), through use of a multivariate logistic regression model, showed that the GSTP1 Val¹⁰⁵/Val¹⁰⁵ genotype was associated with a ninefold lower risk of asthma (corrected OR: 0.11; Table 4).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Previous studies have shown a link between asthma and markers on chromosome 11q13 (3). Although candidate genes for susceptibility to asthma have been identified, they are too distant from these markers to fully account for this linkage (8). We found associations between genotypes of GSTP1, an alternative candidate on 11q13, and both asthma and its related phenotype, BHR.

Age, smoking history, viral infections, atopic status, and in particular baseline airway obstruction all influence airway reactivity (2, 24). We minimized interference from these confounding factors by recruiting lifelong nonsmokers who had not suffered a viral infection within the 6 wk previous to our study. Of the physiologic factors that could influence airway responsiveness, level of lung function or baseline airway caliber (reflected by FEV₁ level) are potentially the most important (2). To minimize the influence of airway caliber on associations between genotypes and BHR, we selectively performed a secondary analysis in subjects exhibiting an FEV₁ > 80% predicted (Groups 1 to 3). There is some disagreement about which cutoff concentration of the test provocative agent defines BHR. Several studies suggest a concentration of 8 mg/ml methacholine as definitive (25, 26). Thus, if Groups 1 and 2 are combined as BHR negative (Table 3), the proportion of individuals with GSTP1 Val¹⁰⁵/Val¹⁰⁵ exceeds that of BHR-positive subjects (Group 3) (p = 0.045; OR: 0.23; 95% CI: 0.05 to 0.97). However, patients in Group 2 exhibited genotype frequencies intermediate between those in Groups 1 and 3 (Table 3), and were therefore considered separately. Trend test analysis across these individual groups showed that as compared with GSTP1 Ile¹⁰⁵/Ile¹⁰⁵, GSTP1 Val¹⁰⁵/Val¹⁰⁵ was associated with a decreasing severity of BHR (p = 0.027). However, the number of individuals with the GSTP1 Val¹⁰⁵/Val¹⁰⁵ genotype in this analysis was small (n = 13); this finding needs confirmation in a larger study sample.

Epidemiologic studies indicate a fundamental role for an IgE-mediated response in the natural history of BHR (2, 3, 24, 27). Because we identified associations between GSTP1 genotype and both BHR and atopy, it was unclear whether these two parameters are independently associated with genotype. The association between GSTP1 Val¹⁰⁵/Val¹⁰⁵ and airway obstruction/reactivity remained significant after correction for atopic status, age, and gender. Although this statistical procedure was undertaken to minimize the influence of skin test positivity and IgE level on the association of GSTP1 genotype with airway reactivity/obstruction, the phenotypes of BHR and atopy are very closely correlated clinically, and we are cautious in interpreting our data as indicating an independent effect. Compared with GSTP1 Ile¹⁰⁵/Ile¹⁰⁵, the heterozygote genotype GSTP1 Ile¹⁰⁵/Val¹⁰⁵ was not associated with either skin test positivity or IgE level, although this genotype was linked with a significantly reduced risk of airway reactivity/obstruction that was intermediate to that with the GSTP1 Ile¹⁰⁵/Ile¹⁰⁵ and GSTP1 Val¹⁰⁵/Val¹⁰⁵ genotypes. These findings are also compatible with our view that the association of GSTP1 genotypes with clinical asthma phenotypes is predominantly with BHR.

BHR reflects inflammation in the asthmatic airway, a key component of which is the generation of ROS (2, 9). BHR may be modulated by ROS levels, possibly through their ability to regulate eicosanoid production via stimulation of arachidonic acid release (28, 29). The GST genes are candidates for having a role in BHR because the enzymes they encode modulate ROS levels (10, 11). We hypothesize that individual ability to detoxify ROS and their products, determined by polymorphism in genes such as those for GST, contributes to the development of BHR and asthma. This view is supported by studies showing that individuals with reduced antioxidant capacity are at increased risk of allergic asthma, and that decreased intake of antioxidants is associated with the expression of asthma-related phenotypes (30).

In contrast to the case with GSTP1, polymorphism in the mu and theta class GST genes that we studied was not associated with asthma and related phenotypes. This finding may reflect differences in the site and level of gene expression, as well as variability in the metabolism of substrates relevant to asthma. Indeed, although adult human lung epithelial cells express various GST gene products, the GSTP1-derived enzyme contributes more than 90% of GST activity (15).

Four GSTP1 alleles, GSTP1*A, GSTP1*B, GSTP1*C, GSTP1*D, have been identified, each defined by amino acid changes at both codons 105 (Ile Val) and 114 (Ala Val) (16). We focused on the Ile¹⁰⁵ Val¹⁰⁵ substitution. GSTP1 Val¹⁰⁵/Val¹⁰⁵ homozygotes therefore include homozygotes for GSTP1*B and GSTP1*C, and GSTP1*B/GSTP1*C heterozygotes. Similarly, patients with GSTP1 Ile¹⁰⁵/Ile¹⁰⁵ include GSTP1*A homozygotes, GSTP1*D homozygotes, and GSTP1*A/GSTP1*D heterozygotes. In unpublished studies, we have also examined the possible influence of the Ala¹¹⁴ Val¹¹⁴ polymorphism on the asthma phenotypes described here. We did not find that this substitution influenced the association of the Ile¹⁰⁵-containing allele with an increased risk of BHR and atopy, or of the Val¹⁰⁵ allele with a reduced risk of BHR and atopy.

In summary, we have identified an association between polymorphism in GSTP1 and asthma and its related phenotypes. Although the presumed biochemical function of GSTP1 and the differences in the catalytic efficiencies of the enzyme products of the Ile¹⁰⁵- and Val¹⁰⁵-containing alleles suggest that the observed association is determined by the GSTP1 genotype, we believe that our data must be considered as preliminary. First, our findings require confirmation in separate cohorts of patients. Second, it is possible that our results reflect linkage disequilibrium between GSTP1 alleles and the true candidate gene, located nearby on chromosome 11q13 (33). Third, disease-marker associations may result from population stratification. Although this possibility cannot be excluded without family-based studies done, for example, with the transmission disequilibrium test, it is noteworthy that the population of North Staffordshire is recognized as particularly homogeneous and stable (21). Further, studies in our laboratory and others have shown that the frequencies of the different GST genotypes are similar in populations throughout Britain and Northern Europe (12, 14, 17, 34). Nonetheless, we propose that our data provide an alternative explanation for linkage of the chromosome 11q13 region to BHR, and suggest that GSTP1 may be important in determining expression of the asthma phenotype.