|Home | About | Journals | Submit | Contact Us | Français|
The individual effect of functional single nucleotide polymorphisms within the catalase and myeloperoxidase genes (CAT and MPO) has been studied in relation to asthma; however, their interrelationship with ambient air pollution exposures has yet to be determined. The authors investigated the interrelationships between variants in CAT and MPO, ambient air pollutants, and acute respiratory illness. Health information, air pollution, and incident respiratory-related school absences were ascertained in January–June 1996 for 1,136 Hispanic and non-Hispanic white US elementary schoolchildren as part of the prospective Children's Health Study. Functional and tagging single nucleotide polymorphisms for the CAT and MPO loci were genotyped. The authors found epistasis between functional polymorphisms in the CAT/MPO loci, which differed by levels of oxidant-stress-producing air pollutants. Risk of respiratory-related school absences was elevated for children with the CAT (G/G) and MPO (G/A or A/A) genes (relative risk=1.35, 95% confidence interval: 1.03, 1.77; P-interaction=0.005). The epistatic effect of CAT and MPO variants was most evident in communities exhibiting high ambient ozone levels (P-interaction=0.03). The association of respiratory-illness absences with functional variants in CAT and MPO that differ by air pollution levels illustrates the need to consider genetic epistasis in assessing gene-environment interactions.
Catalase (CAT) and myeloperoxidase (MPO) are enzymes that play a role in the etiology of respiratory conditions related to oxidative stress (1–8). As part of the oxidative stress defense pathway in the airways, both CAT and MPO react with a potent prooxidant (hydrogen peroxide) to form water + oxygen and hypochlorous acid + hydroxide radical, respectively, under conditions of elevated oxidative stress (Figure 1) such as occurs during respiratory infections and exposure to ambient air pollutants. We hypothesized that variation in the level or function of these enzymes would modulate respiratory illness risk, especially under high levels of oxidative stress. There is a paucity of studies that have investigated this hypothesis.
Enzyme activity is recognized as being modulated by a single nucleotide polymorphism (SNP) in the promoter region of the CAT and MPO genes. Functional studies show that in the CAT gene locus, the minor allele (A) (rs1001179: G-330A) increases gene transcription, resulting in increased blood CAT levels (4). For MPO, the minor allele (A) (rs2333227: G-463A) has lower enzyme activity as measured in bronchoalveolar lavage fluid (9). Because these DNA promoter variants affect levels of 2 enzymes that have the same substrate and function in a common biologic pathway, we hypothesized that this biologic interaction would be reflected in chronically increased levels of highly reactive oxidants in children with combinations of MPO and CAT variants and result in a greater risk of respiratory illness. Furthermore, this risk would be expected to be highest under exposures that increase oxidative stress, such as ambient air pollutants. In addition to these 2 functional SNPs, other variations in the DNA coding sequence at these loci have the potential to affect risk. Therefore, we also investigated whether additional SNPs that account for the majority of variation across each locus add information to the functional SNP associations.
The Air Pollution and Absence Study was a substudy of the prospective Children's Health Study (10). It offered us an opportunity to investigate the effects of air pollution, genetic variation, and respiratory-related absences in elementary schoolchildren residing in 12 southern California communities.
The Air Pollution and Absence Study was a population-based, prospective cohort study conducted as part of the Children's Health Study (10, 11). The cohort included 1,935 fourth-grade students from 27 elementary schools across 12 southern California communities and focused on absence data collected between January and June 1996. Parents or guardians of study participants completed written informed consent and baseline questionnaires pertaining to their child's sociodemographics, medical history, exposure history, and household characteristics. Of the 1,935 children, approximately 70% (n=1,351) provided a buccal cell sample. This analysis was restricted a priori to Hispanic and non-Hispanic white children (n=1,136). The University of Southern California's institutional review board for human studies approved the study protocols.
Incident absences were ascertained by using an active surveillance system augmented by telephone interviews with parents or guardians to collect additional information. An incident absence was defined as one that followed attendance on the preceding school day and was regarded as an independent event regardless of whether it occurred for the same subject. Participating schools provided daily absence summary information for study children within 4 weeks of each absence, and each reported absence was categorized as an illness-related or non-illness-related absence. Telephone interviews with parents were conducted within 4 weeks of a reported illness-related school absence or absence that could not be categorized based on existing data; non-illness-related absences were not further characterized (12).
Illness-related school absences were classified into respiratory, gastrointestinal, and other types of absences on the basis of symptoms information collected by telephone interviews with parents. Respiratory absences were defined as absences associated with one or more of the following symptoms: runny nose/sneezing, sore throat, cough (any), earache, wheezing, or asthma attack (12).
Ambient ozone, particulate matter less than 10 μm in aerodynamic diameter (PM10), and nitrogen dioxide were measured continuously, with hourly averaging at central-site monitors in each of the 12 communities (10). We calculated the daily 1-hour maximum ozone, the 24-hour ozone average, and the 10 AM–6 PM ozone average, as well as the 24-hour averages of PM10 and nitrogen dioxide. We focused on the 10 AM–6 PM average of ozone because it is an index of exposure during the temporal peak of ozone and outdoor activity. The 24-hour averages of PM10 and nitrogen dioxide were used because they lack the temporal peak exhibited by ozone. To assess effects of long-term average levels of ozone, PM10, and nitrogen dioxide on acute effects, we divided communities into high and low groups for each pollutant based on their ranking regarding median levels (Web Table 1; this information is described in the first of 4 supplementary tables, each referred to as “Web table” in the text and posted on the Journal’s website (http://aje.oupjournals.org/)).
Genomic DNA was extracted from buccal mucosal cells by using the PUREGENE DNA purification kit (Gentra Systems, Minneapolis, Minnesota). Genotyping for CAT G-330A (rs1001179) and MPO G-463A (rs2333227) was performed by using the TaqMan allelic discrimination assay (Applied Biosystems, Foster City, California). The DNA fragment containing each SNP was amplified by using the primers and probes shown in Web Table 2. The TaqMan genotyping reaction was amplified on a GeneAmp PCR System 9600 (50°C for 2 minutes and 95°C for 10 minutes, followed by 35 cycles of 92°C for 15 seconds and 60°C for 1 minute), and fluorescence was detected on an ABI PRISM 7700 Sequence Detection System (Applied Biosystems). In each run, 10% of the samples were randomly selected and used for quality control. The results from the TaqMan PCR assay were validated by using polymerase chain reaction/restriction fragment length polymorphism methods and automatic sequencing (BigDye version 3.1, 377XL DNA sequencer; Applied Biosystems).
For each gene, we identified a set of SNPs from dbSNP (http://www.ncbi.nlm.nih.gov/projects/SNP/) and other sequencing databases (i.e., the Environmental Genome Project (http://www.niehs.nih.gov/research/supported/programs/egp/)) with an SNP density of 1–3 SNPs/kb over a region 20 kb upstream and 10 kb downstream of each gene. SNPs were selected on the basis of validation status, Illumina design score (Illumina Inc., San Diego, California), and functional potential of the SNPs.
The selected SNPs were first genotyped by using the Illumina GoldenGate Genotyping Assay (Illumina Inc.) in a sample of 71 Hispanic white and 71 non-Hispanic white participants in the Multi-Ethnic Cohort Study (13). These data were used to define ethnic-specific allele frequencies and patterns of linkage disequilibrium. The squared correlation between the true haplotypes (Rh2) and their estimates were then calculated; calculation of Rh2 is described in detail by Stram et al. (14). Haplotype-tagging SNPs were then chosen by using TagSNPs, a software program (written by D. O. Stram of the Keck School of Medicine at the University of Southern California) that implements an expectation maximization algorithm approach by finding the minimum set of SNPs (within a block) that would have Rh2 values of ≥0.85 for all haplotypes with an estimated frequency of ≥5% in either ethnic group. Functional SNPs were forced into the TagSNP selection process irrespective of minor allele frequency. We also examined the pairwise correlation between SNPs and selected additional tagSNPs to provide redundant coverage in the event of assay failure. This SNP list was refined to provide adequate performance on the Illumina BeadArray platform (Illumina Inc.), a high-throughput assay. We excluded poor-performing, monomorphic, and low-allelic-frequency SNPs. We also excluded samples with SNP call rates of <90%.
A Poisson regression model was fitted to estimate relative risks and 95% confidence intervals for associations of genotypes with respiratory-related school absences (15, 16). An offset term, comprising the log-expected value of the dependent variable, was included in the model to normalize the fitted cell means on a per-subject basis. The deviance divided by its degrees of freedom was used as a correction factor on the standard errors to account for overdispersion in all models (17). Inclusion of potential confounders in the models was based on a review of the literature and changes in univariate genetic effect estimates of at least 10% in multivariate analyses. On the basis of these criteria and the study design, the following variables were selected as potential confounders: community, race, gender, age, asthma status, family income, health insurance status, secondhand smoke, in utero smoke, body mass index, and cat and dog ownership. For the single-SNP analyses, the additive genetic model was utilized. Because of low numbers of subjects with homozygous variant genotypes, we used the dominant genetic model for all joint SNP models.
To address potential confounding by population stratification, 4 coefficients of ancestry variables were also included in the model (18, 19). These variables were constructed from 4 principal components derived from a set of 233 unlinked ancestry informative markers selected to differentiate 4 parental populations (African, European, American Indian, and East Asian). Controlling for these ancestry variables provided adjustment for ancestral history beyond adjustment for typical self-reported racial and ethnic categories.
The interactions between SNPs, and the interactions between SNPs and air pollutants, were evaluated by adding the corresponding product term to the model and using a likelihood ratio test to evaluate its significance. All tests assumed a 2-sided alternative hypothesis and a 0.05 significance level. All analyses were conducted by using SAS/STAT software, version 9.1 (SAS Institute, Inc., Cary, North Carolina.
The demographic characteristics of the Hispanic and non-Hispanic white subjects in the Air Pollution and Absence Study are described in Table 1. All study participants were in the fourth grade at the time of data collection, and nearly all were between 9 and 10 years of age. Boys and girls participated in the study equally. Approximately 15% of children had physician-diagnosed asthma in their lifetime, which corresponds to the national asthma prevalence. More than 80% were in the normal range for body mass index, and more than 85% had health insurance. Approximately 17% of children were exposed to secondhand smoke in the home, and 17% were exposed in utero to maternal smoking.
The genotype distribution for each SNP did not deviate significantly from that expected under Hardy-Weinberg equilibrium. SNP2 in the CAT locus and SNP1 in the MPO locus are both functional SNPs located in the promoter region (G-330A and G-463A, respectively). There were no statistically significant associations between individual SNPs and respiratory-related school absences, adjusted for community of residence, age, gender, income, health insurance, secondhand smoke, in utero smoke, body mass index, race/ethnicity, cat and dog ownership, and asthma status (Table 2).
We found evidence that the biologic relation of CAT and MPO in the oxidative stress pathway resulted in an epistatic effect of the known functional CAT and MPO SNPs on respiratory-related school absences (Table 3). The results showed an increased risk of respiratory-related school absences with the following genotype combination: CAT (G/G), MPO (G/A or A/A) (relative risk=1.35, 95% confidence interval: 1.03, 1.77). Subjects with this SNP combination had the highest rate of respiratory absence, equaling approximately 10 per 1,000 subject-days (Web Figure 1; this supplementary figure is also posted on the Journal’s website (http://aje.oupjournals.org/)). Although not statistically significant, the SNP combination with at least one variant in CAT and MPO resulted in the lowest risk of respiratory-related school absences (relative risk=0.81, 95% confidence interval: 0.55, 1.19), which contributed to the highly significant epistatic effect (P-interaction=0.005) (Table 3). These subjects had the lowest rate of respiratory absence (Web Figure 1): 7 per 1,000 subject-days.
Consistent with our hypothesis, the epistatic interaction of CAT and MPO was apparent among children exposed to high levels of ambient air pollutants (Table 4). Among participants exposed to high levels of nitrogen dioxide and ozone, there was strong evidence of CAT/MPO epistasis (P-interaction=0.002 and P-interaction=0.0004, respectively), whereas none was evident in low-pollution communities. In high-ozone communities, the CAT/MPO genotypes that resulted in decreased oxidative stress (i.e., CAT: G/A or A/A and MPO: G/A or A/A) were associated with a decreased risk of respiratory-related school absences compared with the CAT/MPO wild-type genotype (relative risk=0.42, 95% confidence interval: 0.20, 0.89). Among subjects with high-risk genotypes (CAT: G/G, and MPO: G/A or A/A), we found a trend toward an increased risk of school absences in communities with high levels of nitrogen dioxide, ozone, and PM10. Furthermore, relative risk estimates of the effect of CAT and MPO on respiratory-related absences significantly varied by ozone exposure levels (P-interaction=0.03). There were no statistically significant associations in the joint-effect analyses of CAT/MPO variants in areas of low pollution, although there was a suggestion of a CAT/MPO interaction in communities with low levels of PM10 (P-interaction=0.03).
To examine the added effects of additional variants at each locus, we assessed the joint effects of the functional SNPs with multiple tagging SNPs in the CAT (5 SNPs) and MPO (4 SNPs) loci. We did so by performing interaction tests between the known functional SNPs and all other SNPs utilizing Bonferonni correction for multiple testing. We found that no additional interaction achieved statistical significance, suggesting that the known functional CAT/MPO SNPs adequately explained the extent of the CAT/MPO associations with respiratory-related school absences (Web Table 3).
This study provides evidence illustrating the importance of considering genetic epistasis when examining gene-environment associations. We found that 2 functional SNPs within the promoter region of the CAT and MPO loci—genes involved in prooxidant defense by acting on hydrogen peroxide—were jointly associated with acute respiratory illness as measured by respiratory-related school absences. Consistent with our hypothesis, the epistatic effect of CAT/MPO variants was most apparent among subjects exposed to high levels of ambient air pollutants. The common functional CAT G-330A and MPO G-463A polymorphisms, although nonsignificant when considered individually, provide an example of examining interactions between genes known to be involved in clearance of hydrogen peroxide in the lungs, which may be modified by air pollution exposure. Other SNPs within the CAT and MPO loci did not add to the information on susceptibility provided by these 2 functional CAT and MPO SNPs.
Although both CAT and MPO act upon the same substrate, no studies to our knowledge have examined the joint effect of these 2 genes on respiratory illness. Mak et al. (5) found that the CAT -330 variant allele (A) was associated with a decreased risk of asthma for nonsmokers (odds ratio for subjects possessing at least one A allele vs. the wild type=0.35, 95% confidence interval: 0.15, 0.83) and an insignificant decreased risk for ever smokers (odds ratio= 0.28, 95% confidence interval: 0.05, 1.61). However, the CAT genotypes were not associated with erythrocyte CAT enzyme levels. Ghosh et al. (20) performed in vitro studies examining key markers of inflammation due to asthma. The authors found reduced CAT enzyme activity among asthmatics and concluded that CAT inactivity amplifies oxidative stress, contributing to the inflammation inherent in asthmatic airways (20). These reports support a role for the functional CAT variant in diseases associated with increased oxidative stress.
Two recent studies have examined the role of the MPO enzyme among asthmatic patients, with conflicting results (21, 22). Tauber et al. (21) found that serum MPO levels were not associated with childhood asthma, whereas Monteseirín et al. (22) found that patients with allergic asthma, compared with nonallergic controls, had a propensity to release more MPO. Although we are not aware of studies that have examined the association of the functional MPO G-463A SNP with asthma or acute respiratory disease, the inconsistent findings for patients with high levels of oxidant stress may reflect the fact that the CAT variant genotypes were not considered.
Both CAT and MPO are important enzymes in the lungs that act upon hydrogen peroxide, a potent oxidant (Figure 1). The lack of both CAT and MPO enzymes may result in respiratory illness that may be due to an accumulation of prooxidants in the lungs, resulting in an acute inflammatory response or exacerbation of an acute respiratory response. The known functional SNP in CAT (rs100179) produces a G-330A polymorphism in the promoter region of the gene. The A allele is thought to increase CAT enzyme activity, resulting in increased conversion of hydrogen peroxide to water and oxygen (4); this conversion is thought to work optimally during times of severe oxidative stress, whereas glutathione peroxidase is the major antioxidative enzyme in the detoxification of hydrogen peroxide to water and oxygen under normal circumstances (4). This mechanism could underlie our results that show the largest protective effect of CAT/MPO in areas of high pollution, although more work is needed to elucidate these findings. The MPO gene has a functional SNP also located in the promoter region (rs2333227: G-463A) whereby the minor allele (A) is thought to decrease enzyme activity, resulting in slower conversion of hydrogen peroxide to hypochlorous acid (9). MPO is a potent bactericide but is also known to increase production of heme oxygenase, which may lead to acute or chronic respiratory illness (23).
Our results should be interpreted in light of some study limitations. Of the total number of eligible white children, approximately 30% did not provide buccal cell samples for this study, which may give rise to selection bias. However, comparing those children who were and were not genotyped did not reveal marked differences in demographic factors or absence rates, making selection bias an unlikely explanation for our findings. Furthermore, other intrinsic and extrinsic cofactors may potentially confound the relation between CAT/MPO and respiratory illness. Some that have been described in the literature include personal and secondhand smoke (24, 25), pet and plant allergens (26), and asthma and family history of asthma (27, 28). Statistical adjustments for community, secondhand smoke, pets, and asthma status were utilized in all models. Adjustments for family history of asthma and allergens such as pests, mold, and mildew resulted in a negligible change in effect estimates and were thus excluded from the final models.
In association studies involving multiethnic populations, confounding due to population stratification could bias the results. To address this potential bias, we included only white (Hispanic and non-Hispanic) subjects in our analyses, examined risk estimates stratified by ethnicity, and performed sensitivity analyses adjusted for an index of population stratification (q factor) (18, 19). The results did not vary appreciably by ethnic stratification and q-factor adjustment.
In conclusion, the association of respiratory illness absences with functional variants in CAT and MPO that vary by air pollution levels illustrates the need to consider genetic epistasis when assessing gene-environment interactions. In the current approaches to candidate loci and genome-wide association studies, it may be essential to consider epistasis among genes, as defined by this well-studied biologic pathway, when examining susceptible subgroups for the diverse effects of air pollution.
Authors affiliation: Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California.
This work was supported by the National Heart, Lung, and Blood Institute (grants 5R01HL61768 and 5R01HL76647); the Southern California Environmental Health Sciences Center (grant 5P30ES007048) funded by the National Institute of Environmental Health Sciences; the Children's Environmental Health Center (grants 5P01ES009581, R826708-01, and RD831861-01) funded by the National Institute of Environmental Health Sciences and the Environmental Protection Agency; the National Institute of Environmental Health Sciences (grant 5P01ES011627); and the Hastings Foundation.
Conflict of interest: none declared.