INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Polymorphisms in several candidate genes have been identified and associated with asthma (1). However, few have been reported to alter the coded protein, and their functional significance is therefore uncertain (2). Clara cell secretory protein (CC16) is a 16-kD protein primarily expressed in the respiratory tract by nonciliated bronchiolar secretory cells (5), accounting for 7% of the total protein content in the bronchoalveolar lavage fluid (BALF) of healthy nonsmokers (6). The immunomodulatory activity of CC16 has been well documented (7) and CC16 messenger RNA (mRNA) levels have been proposed as markers of lung maturation and epithelial differentiation (10). Mice deficient in CC16 expression exhibit a higher susceptibility to oxidant lung injury and an excessive inflammatory response (11, 12).

Studies of the chromosome 11q13 region, in which the CC16 gene is located (13), found linkage to maternal inheritance of atopic IgE responsiveness and bronchial hyperresponsiveness (BHR) (14). Other candidate genes for asthma are located in the 11q13 region, including the lymphocyte surface marker CD20 and high-affinity IgE receptor beta chain (Fc RI-) genes (14, 15). However, despite identification of several polymorphisms in these genes, none had sufficient frequency or caused functional alterations that would explain the genetic linkage to this region (14). In view of its chromosomal location and functional role in respiratory inflammatory processes, abnormalities of the CC16 gene may be involved in the inherited pathogenesis of asthma.

The CC16 gene was previously screened for mutations and a polymorphism (A38G) was identified and associated with an increased risk of physician-diagnosed asthma in a population of Australian children (16), and increased BHR in a population of Australian infants (17). Studies on populations of Japanese and British adults (18) and North American children (19) did not replicate these associations. However, in another study of adults, a moderate risk of asthma was found to be associated with the CC16 38A allele (20). These interstudy differences may be a result of variation in phenotype definitions, ethnicity, and environmental exposures. Therefore, whether the A38G polymorphism had a functional role in the development of asthma or was in linkage disequilibrium with another etiologically important mutation remained to be elucidated. To further assess the relationship of the A38G polymorphism to asthma, plasma levels of CC16 were measured in cohorts of children with and without asthma. The results of these analyses are described in this report with discussion of the evidence suggesting the possible role of alterations in the CC16 gene and an inherited predisposition to asthma.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Phenotypic data and DNA were obtained from two populations of children in this study. The first consisted of children ascertained as part of a long-term follow-up study of a birth cohort recruited randomly from a suburban general hospital antenatal clinic (21) in Perth, Western Australia. The second population included children recruited from a respiratory outpatient clinic at a tertiary children's hospital in Perth, Western Australia. Parental informed written consent was obtained for all subjects. All subjects were white and younger than 18 yr of age.

Phenotypic data were collected as previously described (16). A subject was defined as asthmatic when "yes" and "doctor" were the respective answers to the questions "Has he/she ever had asthma?" and "Who diagnosed the asthma?" in the parental questionnaire. Positive tobacco smoke exposure of a subject was defined as an active or passive smoking history determined from subject and parental questionnaire responses.

Corticosteroids have been shown to upregulate the expression of CC16 (22) and subjects who reported previous or current use of corticosteroids were therefore excluded from the study. Subjects with more severe asthma tend to be those using corticosteroids (23), nonetheless corticosteroid use had the potential to seriously confound the results. Therefore, 100 children (72 from the first population and 28 from the second) were included in this study. The studies were approved by the King Edward Memorial and Princess Margaret Hospitals Ethics Committee.

Genotyping

Genomic DNA was extracted and exon 1 of the CC16 gene was amplified by polymerase chain reaction (PCR) using appropriate primers (16). A38G genotype of each subject was determined by restriction digestion of the PCR product using the restriction enzyme Sau 96 1 (16).

Plasma Assays

Plasma levels of CC16 were evaluated using a sensitive immunoassay that has been validated using an alternate method (24) and is based on the agglutination of latex particles coated with polyclonal anti-CC16 antibodies (25). Plasma creatinine levels were determined by the Jaffé method (26) and used as a marker of renal function.

Statistical Analysis

Conventional multiple regression was used to adjust geometric mean plasma CC16 levels for age and gender. Both conventional multiple and logistic regression were then used to test for association between physician-diagnosed asthma, A38G genotype, and plasma CC16 levels (27). Where appropriate, the effects of age, sex, tobacco smoke exposure, and creatinine levels were investigated for all outcomes.

A test for linear trend of the effect of A38G genotype on phenotype was conducted by entering genotype as a continuous variable in logistic regression analyses. Subjects were coded, homozygous 38A sequence (38A/38A = 2), heterozygotes (38A/38G = 1), and homozygous 38G sequence (38G/38G = 0). Genotype was also analyzed as a 3 level categorical variable generating a 2 degree of freedom test against the null hypothesis that all three genotypes were associated with a similar response. Plasma CC16 and creatinine levels exhibited a skewed distribution and were log_e transformed prior to analysis. Because of the influence of age and gender on the diagnosis of asthma, all analyses were adjusted for these factors (28). However, when these factors were removed from the analysis of plasma CC16 level and A38G genotype, the conclusions remained the same.

Tobacco smoke exposure was significantly associated with the risk of asthma and was included in all logistic regression models where asthma was the outcome. Tobacco smoking has been associated with a 30% decrease in serum CC16 (24, 29). However, there was no significant association between tobacco smoke exposure and plasma CC16 levels. Consequently, these models were not adjusted for tobacco smoke exposure. In this study, tobacco smoke exposure was almost always caused by passive smoking which in many cases may not have been consistently high enough to impact on plasma CC16 levels.

Plasma creatinine was used as a marker of renal function (24), but there was also no association between creatinine levels and either asthma or plasma CC16 levels. Therefore, none of the models was adjusted for creatinine levels. However, because statistical significance is, at best, an indirect pointer to biological relevance, models were also fitted that were adjusted for tobacco smoke exposure and creatinine level. The substantive conclusions of all analyses were unchanged regardless of inclusion or exclusion of these factors. Analyses were conducted on Minitab for Microsoft Windows version 12 software. Statistical significance was defined at the 5% level.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Unadjusted plasma CC16 levels ranged from 3.8 to 21.25 µg/L with a mean of 9.62 µg/L, a standard deviation of 3.75 µg/L, and a geometric mean of 8.91 µg/L. The 38A/38A genotype was present in 12% of the population, with heterozygotes comprising 45% and 38G/38G homozygotes, 43%. There were 36 asthmatic and 64 nonasthmatic subjects and 53% of the population were exposed to tobacco smoke. All subjects had creatinine levels within the normal range with a mean of 0.63 mg/dl, a standard deviation of 0.13 mg/dl, a range of 0.43 to 1.14 mg/dl, and a geometric mean of 0.64 mg/dl.

Conventional multiple regression analysis of log_e plasma CC16 concentration, adjusted for age and sex, showed that nonasthmatic subjects had a geometric mean of 9.98 µg/L (95% confidence interval [CI] = 8.83 to 11.26). Asthmatic subjects had significantly lower levels (p = 0.006) with a geometric mean of 7.96 µg/L (95% CI = 6.79 to 9.31) (Table 1). The geometric mean of plasma CC16 levels adjusted for age and sex, of homozygous 38A/38A subjects was 6.79 µg/L (95% CI = 5.28 to 8.73). This was significantly lower than homozygous 38G/38G subjects (p = 0.003) and heterozygous subjects (p = 0.022) who had adjusted geometric mean plasma CC16 levels of 10.01 µg/L (95% CI = 8.67 to 11.56) and 9.17 µg/L (95% CI = 7.96 to 10.57), respectively (Table 1).

Figure 1 compares the adjusted geometric mean CC16 levels for asthmatic and nonasthmatic subjects with different A38G genotypes. Within the nonasthmatic group shown, conventional multiple regression analysis adjusted for age and sex, found that homozygous 38A/38A subjects had reduced plasma CC16 levels compared with heterozygotes (p = 0.047) and homozygous 38G/38G subjects (p = 0.025). The test for linear trend, investigating the effect of number of 38A alleles on log_e plasma levels of CC16 protein (adjusted for age and gender) showed that each additional 38A allele was estimated to produce a 15% decrease (95% CI = 5% to 24%, p = 0.007) in the geometric mean of the plasma level of CC16 protein.

View larger version (48K): [in this window] [in a new window]   Figure 1.   Plasma CC16 levels for A38G genotype, comparing asthmatic and nonasthmatic subjects. Columns represent geometric mean plasma CC16 levels ± 95% CI (adjusted for age and sex).

Logistic regression analysis adjusted for age, sex, and tobacco smoke exposure showed there was a strong association between log_e plasma CC16 levels and the risk of asthma. A 15% decline in plasma CC16 levels (the estimated fall in plasma level associated with each additional 38A allele) is estimated to be associated with a 63% (95% CI = 21% to 119%, p = 0.001) increase in the risk of asthma. Using logistic regression to adjust for age, sex, and tobacco smoke exposure and taking genotype 38G/38G as a baseline, the odds ratio for the development of asthma in subjects with genotype 38A/38A was 4.78 (95% CI = 1.08 to 21.18, p = 0.04) and for heterozygotes (38A/38G) was 3.91 (95% CI = 1.42 to 10.80, p = 0.008).

The test for linear trend on the risk of developing asthma, adjusted for age, sex, and tobacco smoke exposure, showed that each additional 38A allele increased the odds of asthma by a multiplicative factor of 2.55 (95% CI = 1.27 to 5.13, p = 0.008). When this logistic regression model was extended to simultaneously adjust for the effect of plasma CC16 levels (linear, square, and cubic terms) the estimated multiplicative increase in odds associated with each additional 38A allele fell to 1.83 (95% CI = 0.75 to 4.44, p = 0.18).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study demonstrated a significant association between the A38G genotype, plasma CC16 levels, and asthma in a population of children. The data in Figure 1 show that plasma CC16 levels were reduced in subjects who had the 38A sequence and that individuals with lower plasma CC16 levels were more likely to have asthma. This relationship probably reflected similarly reduced CC16 concentrations in the respiratory tract. Diffusion of CC16 across the epithelium/blood barrier by passive transudation has been shown (30) and serum CC16 levels were highly correlated with expression and secretion in the respiratory tract (30). As the polymorphism occurs in an untranslated region of the CC16 gene, the structure, function, and immunogenicity of the coded protein would be unaffected. Therefore, the reduced CC16 concentrations detected by the immunoassay do not reflect an altered antibody response to a different epitope.

Previous studies showed that plasma and BALF CC16 levels were reduced in asthmatic subjects compared with healthy nonsmokers (31, 32). The investigators postulated that the primary causes of these altered levels were remodeling of the small airways (31) and alteration of CC16 turnover by the respiratory epithelium (32). However, as shown in Figure 1, within the nonasthmatic group, homozygous 38A subjects had reduced plasma CC16 levels compared with heterozygotes (p = 0.047) and homozygous 38G subjects (p = 0.025) and this was similar in the asthmatic group. This indicated that CC16 genotype largely determined plasma CC16 levels and airway remodeling was a less likely cause of reduced CC16 levels in asthmatic patients. However, other genetic and environmental influences may be necessary to create the milieu in which asthma becomes established and is perpetuated by reduced CC16 activity which results in associated airway inflammation.

The results support the hypothesis that the A38G genotype affects CC16 expression levels which contributes to the pathogenesis of asthma. The association between A38G genotype and asthma is qualitatively similar to that found in the previous study (16), although slightly more conservative owing to exclusion of asthmatic subjects who reported current or previous use of corticosteroids. Adjustment for the effect of plasma CC16 levels on this association resulted in a reduced odds ratio, which suggests that a substantial component of the increased risk of asthma associated with the 38A allele is mediated through a mechanism involving altered plasma CC16 levels. However, the fact that the residual multiplicative odds ratio is 1.83 (although not significantly different from 1) indicates that other mechanisms may also be involved.

This analysis is consistent with a simple genotype-phenotype association mediated by serum levels of CC16 protein. However, serum CC16 levels in heterozygotes are very similar to those in 38G homozygotes, whereas the risk of asthma in heterozygotes is raised to almost the same levels as that in 38A homozygotes. These results may simply represent variation by chance, but might also suggest that the true causal mechanism is more complex. Definitive investigations of more complex hypotheses would require much larger populations. While there is adequate power in this study to investigate main effects, the sample size is too small to make conclusive statements about the genetic model (recessive, codominant, or dominant) linking the CC16 gene to asthma. A proper investigation of the effect of any gene-by-environment interactions would also demand a larger study. Nevertheless, the logistic regression modeling and Figure 1 represent convincing evidence that the CC16 gene and its protein product are likely to play some role in the etiology of asthma; this warrants further investigation in larger studies in which more complex hypotheses can be tested rigorously.

It is unlikely that the association between CC16 A38G genotype and asthma is the result of linkage disequilibrium with another disease causing locus in the chromosome 11q13 region. Although associations have been reported between various alleles of other candidate genes in this region, including CD20 and Fc RI- (14, 15), the polymorphisms had low population frequencies and were not shown to alter the function of the coded proteins.

CC16 is the most abundant protein secreted in the airway. Its anti-inflammatory activities include downregulation of interferon gamma and tumor necrosis factor-alpha synthesis and/or biological activity (8), and inhibition of IL-1 (12) and phospholipase A₂ (9). Phospholipase A₂ regulates arachidonic acid metabolism and the cascade of leukotriene inflammatory mediators (33). This supports a role for CC16 in asthma, although the precise functional mechanism is yet to be ascertained. The CC16 gene and its protein product are strongly associated with the risk of developing asthma. There is a logical biological pathway which links the 38A allele with reduced plasma levels of the protein and it is likely that this reflects reduced CC16 concentrations in the respiratory tract. There is also a clear association between CC16 protein levels and the risk of asthma. Therefore, this study supports the concept that sequence variations in the CC16 gene are likely to play a role in the development of asthma.