INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Given the complexity of the "asthmatic phenotype," an approach to unravel genetic factors relevant to bronchial asthma is to focus on associated well-defined phenotypes, thereby reducing the number of genes involved, at best, to a monogenic ideal. Bronchial hyperresponsiveness (BHR), an exaggerated bronchospastic response to a variety of stimuli, is the most common physiologic abnormality found in asthmatic patients (1). BHR with a prevalence of 13% in the general population also occurs in asymptomatic individuals and may represent a risk factor for asthma (2). Linkage of chromosome 5q31-q33 to BHR as well as to atopy, a second phenotype relevant to asthma, has been reported in Dutch families selected through asthmatic index probands (3). This chromosomal region contains a number of candidate genes for both BHR and atopy. For the gene (ADRB2) encoding the ₂-adrenergic receptor (₂AR) several polymorphisms have been described (4). In particular, the amino acid substitutions at positions 16 (Arg or Gly), 27 (Glu or Gln), and 164 (Thr or Ile) alter the receptor function in vitro. Cells expressing the Gly16/Gln27 and Gly16/ Glu27 variants in vitro display significantly more agonist-promoted receptor downregulation than cells expressing the Arg16/Gln27 isoform, whereas the Arg16/Glu27 mutant was found to be completely resistant to downregulation (7, 8). Compared with wildtype ₂AR the Ile164 isoform shows significantly less coupling to adenylate cyclase when expressed in Chinese hamster fibroblast cells in vitro (9). Further evidence for a contribution of ₂AR variants to the asthmatic phenotype is drawn from a number of association studies. The Gly16 allele was found to be associated with steroid-dependent (6) and nocturnal asthma (10) as well as with an increased degree of bronchodilator desensitization (11). An association of Arg16 with enhanced responses to albuterol (12) and of Gln27 with elevated airway hyperresponsiveness among asthmatics has been described (13). In addition, an association of Gln27 with childhood asthma (14) and with high levels of total IgE in asthmatic families (15) has been reported. The in vitro data on the different quantitative and qualitative effects of the three ₂AR polymorphisms on its function led us to analyze the association of ADRB2 haplotypes with BHR.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

As part of the European Community Respiratory Health Survey (ECRHS) in the period from 1990 to 1992, 4,990 subjects age 19 to 65 yr were drawn at random from the offices of population census in Erfurt (Eastern Germany). Full details of this study have been published elsewhere (16, 17). In brief, this cross-sectional study used a two-stage approach. In Stage I, the screening questionnaire (standardized within the ECRHS) was mailed to the selected probands. Of the Stage I responders, a random sample (40.9%) was invited by letter or home visit for Stage II tests. Results reported here are all from Stage II, which comprised a detailed questionnaire, spirometric measurements, methacholine or bronchodilator inhalation tests, skin prick testing, and determination of total and specific IgE. By this protocol, 1,281 probands were finally recruited.

Lung Function Measurement and Bronchial Provocation

Pulmonary function testing comprised baseline spirometry and determination of flow-volume curves (Masterlab; Jäger, Würzburg, Germany). All measurements as well as methacholine challenges were performed strictly adhering to the ECRHS protocol and quality standards (16). After inhalation of saline as diluent, increasing doses of methacholine were administered. According to baseline characteristics of subjects, challenges were started using a short protocol, with fourfold increments, or a long protocol, with twofold increments. The starting dose of methacholine consisted of one deep inhalation of the concentration 0.195 mg/ml for the long protocol and two inhalations for the short protocol. The maximal dose comprised four deep inhalations of 12.5 mg/ml in the long protocol and six inhalations in the short protocol. When subjects showed at least a 10% decrease of FEV₁ relative to the postsaline level, the subsequent challenge was performed according to the long protocol. Provocation was stopped when there was a greater than 20% reduction of FEV₁. For analysis, the concentration of methacholine was weighted by the number of inhalations, and the results were added to give a cumulative dose of methacholine, with a maximal cumulative dose of 2 mg. The method has been described in detail elsewhere (17, 18). Methacholine challenges were performed only if baseline FEV₁ was greater than 1.5 L and 70% of the predicted value (19). Furthermore, challenges were not performed if the subject had had a heart attack within the last 3 mo, any heart disease with concomitant medication, epilepsy with current medication, medication with beta-blocker for any reason, or if she was pregnant or breast-feeding. Postdiluent FEV₁ had to be greater than 90% of the initial FEV₁. A total of 131 subjects stopped the provocation before reaching a 20% reduction of FEV₁ or the maximal cumulative dose and were thus not included in the study. In total, complete data were collected from 1,150 individuals. Of these, all probands (152) who showed a  20% decrease in FEV₁ relative to postsaline values were considered as bronchial hyperreactive.

Probands were investigated between September 1991 and April 1992 when sera and ethylenediaminetetraacetic acid (EDTA)-treated blood samples were taken for storage at 20° C until analysis. Total and specific serum IgE against Dermatophagoides pteronyssinus (house dust mite [HDM]) were measured centrally by Pharmacia (Uppsala, Sweden) using the Pharmacia CAP system.

Molecular Methods

Isolation of DNA. Blood samples stored at 20° C were rapidly thawed and 500 µl were removed for DNA isolation by the QIAamp Blood Kit (Qiagen, Hilden, Germany) according to the manufacturer's recommendation.

Genotyping. The A/G polymorphism at the 46th nucleotide position of the translated ADRB2 sequence (first nucleotide of the triplet [AGC  GGC] encoding amino acid 16) was investigated by indirect restriction fragment length polymorphism (RFLP) technique after the introduction of an NcoI site diagnostic for alleles containing G at nucleic acid 46 (Gly16) using the primer pair ADRB2-NcoI (5'-AGCGCCTTCTTGCTGGCA CCCcAT-3') and ADRB2-197 (5'-AGGCAATGGCATAGGCTTG-3') for amplification. The mismatch introducing the restriction site is indicated by a lower case letter. Each 20-µl polymerase chain reaction (PCR) contained 50 ng of DNA, 1.6 mM (NH₄)₂SO₄, 20 mM TRIS-HCl, pH 8.55, 1.5 mM MgCl₂, 0.2 mM deoxyribonucleoside triphosphate (dNTP), 0.5 U Taq polymerase (AGS, Heidelberg, Germany), and 50 ng of each primer. The reaction conditions were 10 min 94° C, followed by 30 cycles of 94° C for 30 s, 66° C for 30 s, and 72° C for 60 s, and a final extension at 72° C for 10 min. The 586-bp PCR products were digested with NcoI and analyzed on 2.5% agarose gels, stained with ethidium bromide.

ADRB2 haplotypes of the codons for amino acid 16, 27, and 164 were determined by allele-specific amplification of the polymorphism at nucleotide 46 (codon 16) followed by hybridization with sequence-specific oligonucleotides (SSO) for the detection of the polymorphic nucleotides 79 (codon 27) and 491 (codon 164). The amplification refractory mutation system (ARMS) technique was applied using the primer pairs ADRB2-16G (5'-CTTCTTGCTGGCACCCAcTG-3') and ADRB2-197 to specifically amplify alleles encoding Gly at position 16, and ADRB2-16R (5'-CTTCTTGCTGGCACCCAcTA-3') and ADRB2-197 for the amplification of ADRB2 sequences encoding Arg at this position. The A/G polymorphism at nucleotide 46 is shown in bold. A mismatch indicated by a lower case letter was introduced to increase the specificity of the amplification reaction (20). Each 20-µl PCR reaction contained 50 ng of DNA, 1.6 mM (NH₄)₂SO₄, 20 mM TRIS-HCl, pH 8.55, 1.5 mM MgCl₂, 0.1 mM dNTP, 0.5 U Taq polymerase (AGS), and 50 ng of each primer. The reaction conditions were 10 min 94° C, followed by 30 cycles of 94° C for 30 s, 63° C for 30 s, and 72° C for 60 s, and a final extension at 72° C for 10 min. Specificity of the allele-specific amplification was confirmed by segregation analysis in two families and by comparison of the results obtained by the indirect NcoI RFLP typing method described previously. Individual results for codon 16 and/or 27 were confirmed by direct sequencing of amplification products obtained with the primer pairs: ADRB2-Seq-1-5' (5'-CATTGGCCGAAAGTTCCCGTA-3') and ADRB2-Seq-1-3' (5'-CGCACCAGAAGTTGCCAAAAGT-3') using a dye terminator kit (Applied Biosystems, Weiterstadt, Germany) according to the manufacturer's recommendations.

The 581-bp PCR products of the allele-specific amplification contained also the polymorphic sites at both nucleotide 79 (codon 27) and 491 (codon 164). An aliquot of 10 ng of each amplification product was dotted onto Hybond N⁺ nylon membranes (Amersham-Buchler, Braunschweig, Germany). Four filters were prepared in parallel, each hybridized with one of the following SSOs: ADRB2-27Q (5'-GTCACGCAGCAAAGGGACG-3') specific for codon 27 encoding Gln; ADRB2-27E (5'-GTCACGCAGGAAAGGGACG-3') specific for codon 27 encoding Glu; ADRB2-164I (5'-AAGAAGGAGATAAGGCCT G-3') specific for codon 164 encoding Thr; and ADRB2-164T (5'-AAGAAGGAGGTAAGGCCTG-3') specific for codon 164 encoding Ile. The polymorphic nucleotide detected by each SSO is given in bold. Further processing and hybridization of the membranes was performed as described (21, 22). Membranes were hybridized overnight at 53° C with the SSOs 3' end-labeled with digoxigenin by use of the DIG Oligonucleotide 3'-End Labeling Kit (Boehringer Mannheim, Mannheim, Germany) according to the manufacturer's protocol. Two final washings were performed for 10 min in 1× saline sodium citrate (SSC)/0.1% sodium dodecyl sulfate (SDS) at 59° C.

Statistical Methods

First, characteristics of the case and the control sample were compared. For binary variables Pearson's ² test or, for small cell counts, Fisher exact test was used. For quantitative variables Student's t test or, under non-normality, the Mann-Whitney U test was employed. Second, allele/haplotype frequencies were estimated by counting. Deviations from Hardy-Weinberg equilibrium (HWE) were tested for by a ² test of observed and expected cell counts with the degrees of freedom (df) adjusted by the number of independent frequencies estimated. Each allele/haplotype was examined separately for association in a 2 × 2 contingency table by case/control status using a Pearson's ² test. To adjust for additional covariates, the data were reanalyzed by multiple logistic regression with BHR as the outcome variable. For all analyses the significance level was set at  = 0.05. For the descriptive part of comparing case and control sample, no adjustment for multiple comparisons was made. For the association analyses, the Bonferroni correction for multiple testing was applied, i.e., for n independent tests we adjusted the significance level  to _adj = /n. For the haplotype analysis, first an overall test of association was carried out and then, in case of association, each of the haplotypes was tested individually, grouping all other haplotypes together. For an identified association, the odds ratio (OR) (and its 95% confidence interval [95% CI]) was computed as an approximation of the relative risk.

The statistical analyses were done applying the SPSS 6.1 and the SPLUS 4.0 software package.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Description of Case and Control Groups

BHR was analyzed as a qualitative trait and defined as reduction in FEV₁ of 20% up to a cumulative dose of 2 mg of methacholine. Probands who stopped the provocation before reaching a 20% decrease in FEV₁ or the maximal methacholine dose for reasons other than bronchoconstriction were not included in the study because an assignment to either study group was not possible. Adopting this protocol, a total of 152 bronchohyperreactive probands could be identified within a random population sample comprising 1,150 individuals with fully evaluated lung function (Table 1). These were compared with 295 bronchonormoreactive probands (control group) taken at random from the same population sample and matched to the cases with regard to sex and age. Both groups contained similar numbers of individuals with self-reported asthma. The percentage of current smokers was significantly higher in the case group. The mean log[total IgE] of the cases was not significantly elevated compared with that of the control subjects though significantly more BHR⁺ individuals had HDM-specific IgE levels > 0.35 kU/L. Finally, the mean baseline FEV₁ of the cases was significantly reduced as opposed to the control subjects. This impaired airway function might emerge from BHR or could be attributed to other unknown factors.

Association of Alleles

All individuals were genotyped for single nucleotide polymorphisms affecting the codons for amino acids 16 (Gly or Arg), 27 (Glu or Gln), and 164 (Ile or Thr) of ₂AR. A comparison of the allele frequencies for the case and control groups is summarized in Table 2. The third polymorphism changing amino acid position 164 is very rare. Only two DNA samples of the BHR⁺ group were typed heterozygous for the mutant Ile164 variant (genotype frequency of 0.66% for the cases). All other individuals were typed homozygous for the wildtype Thr164 allele. Due to the rarity of Ile164 no association test was carried out for the Ile164/Thr164 polymorphism. For two biallelic polymorphisms, the appropriate adjusted significance level is _adj = /2 = 0.025. The analysis of the individual polymorphisms revealed no significant difference in allele frequencies.

Association of Haplotypes

Because the function of an individual ₂AR molecule would depend on the relative contribution of the different amino acid substitutions in this polypeptide, we investigated the association of the ADRB2 haplotypes (as defined by codons 16, 27, 164) with BHR. Four different haplotypes were present in the studied population: Arg16/Gln27/Thr164, Gly16/Glu27/ Thr164, Gly16/Gln27/Thr164, and Gly16/Gln27/Ile164. The combination Arg16/Glu27 was not detected in our sample. The two BHR⁺ probands positive for Ile164 were typed both Gly16/Gln27 on the chromosome encoding the Ile164 ₂AR allele and Gly16/Glu27/Thr164 on the other chromosome. Because of the rarity of Ile164, these two individuals were discarded from the association analysis. This was also done because the BHR of these two individuals might largely be due to the Ile164 variant, which was shown to display a significant loss in coupling to adenylate cyclase in vitro (9). The overall ² test performed on the remaining 150 BHR⁺ and 295 BHR individuals revealed an association (p = 0.010). Thus, individual tests for each haplotype grouping the other two together were carried out (_adj = /3 = 0.0167). The haplotype Gly16/Gln27/ Thr164 was significantly underrepresented in the BHR⁺ group (Table 3, p = 0.002). The other two haplotypes Arg16/Gln27/ Thr164, Gly16/Glu27/Thr164 did not reveal individually any positive association even when excluding carriers of the haplotype Gly16/Gln27/Thr164 from the sample. Note that neither the case sample nor the control sample showed significant deviations from HWE (² test for three haplotypes with 5  2 = 3 df: cases p = 0.64, controls p = 0.60).

Because elevated HDM-specific IgE serum levels apparently constitute a risk factor for BHR (Table 1), the analyses performed on individuals with HDM-specific IgE  0.35 kU/L also showed the underrepresentation of only the Gly16/Gln27/ Thr164 haplotype in the case group (Table 4).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We analyzed the association of ₂AR variants reported to influence the receptor's function with BHR in a random population-based sample. From 1,150 probands 152 could be extracted as bronchohyperreactive when BHR was defined by means of methacholine provocation as a reduction of FEV₁ of  20% up to a maximal cumulative dose of 2 mg of methacholine. From the remaining subjects a normoreactive, age- and sex-matched control group of 295 individuals was selected at random. All probands were genotyped for biallelic polymorphisms of the ADRB2 gene resulting in amino acid substitutions at positions 16, 27, and 164 of ₂AR. The genotyping protocol applied allowed for the first time the direct determination of ADRB2 haplotypes for each individual. The observed allele and haplotype frequencies are similar to the frequencies reported by Martinez and coworkers (23) and by Weir and coworkers (24) for Caucasian populations. The Ile164 polymorphism was very rare in our subject groups and the two individuals positive for this polymorphism were discarded from further analyses. Whereas at the level of the single polymorphisms no allelic association with BHR could be determined, a significant underrepresentation of the Gly16/ Gln27/Thr164 haplotype frequency was observed in the BHR⁺ group compared with the control group. This result indicates a protective effect of the Gly16/Gln27/Thr164 haplotype with regard to BHR (OR, 0.550; 95% CI, 0.373 to 0.813). No functional characteristic unique to the Gly16/Gln27/Thr164 variant has been described so far.

Functional consequences of the ₂AR polymorphisms with regard to agonist binding, stimulation of adenylate cyclase, and agonist-promoted downregulation of receptor expression have been studied by site-directed mutagenesis and recombinant expression in Chinese hamster fibroblast cells or human smooth muscle cells in vitro (7, 8). Only variants with Arg in position 16 showed an altered degree of ₂-agonist-induced receptor downregulation with the Arg16/Glu27/Thr164 polypeptide being completely resistant (8). It is of interest to note that this haplotype is very rare and not present in the population studied. No functional differences were detected in vitro between the Gly16/Glu27/Thr164 and Gly16/Gln27/Thr164 variants. It is possible that the limitations of the in vitro studies of receptor activity did not allow identification of all differences in the function and regulation of the various ₂AR variants on native cells. It is also conceivable that other yet to be determined amino acid or gene regulatory polymorphisms of ADRB2 relevant to the pathophysiology of BHR are in linkage disequilibrium with this haplotype. In this regard, polymorphisms of an AU-rich sequence element contained within the 3'UT of the ₂AR messenger RNA (mRNA), which promotes an agonist regulation of the mRNA stability (25), might deserve further investigation as well as allelic variants of a 5' leader cistron which encodes a peptide inhibiting the translation of the receptor (26). Finally, the relevant protective gene with regard to BHR might be in linkage disequilibrium with the Gly16/ Gln27/Thr164 ADRB2 haplotype.

From our analysis, we conclude that the ₂AR genotype is indeed a risk-moderating factor for BHR. However, in our study no predisposing haplotype was identified. On the contrary, the haplotype Gly16/Gln27/Thr164 seems to have a protective effect. A previous study on the influence of the ADRB2 alleles on the lower airway reactivity was performed on 65 patients with mild to moderate asthma and reported an association of the Gln27 variant with reduced bronchial reactivity (13). Because asthma not only overlaps with BHR but also with an allergic phenotype, a clear phenotypical distinction may not always be feasible in asthmatic subjects. The differential contribution and mutual interdependence of the phenotypes might explain the conflicting results reported on the association of ADRB2 polymorphisms with asthma and elevated IgE serum levels (14, 15, 27). In our study we focused on BHR in a general population sample of 1,150 individuals. This contained 13.2% (152) bronchohyperreactive probands, which is in good agreement with what has been found, for instance, in a similar study of a Spanish population sample (28). Thus, regarding this aspect a bias as a result of the sampling procedure can be excluded. Cases and control subjects did not show significant differences with regard to two related phenotypes, i.e., with regard to the distribution of total serum IgE or to the frequency of asthmatics, which was 3.3% for the cases and 1.7% for the control subjects. Thus, a confounding effect resulting from these phenotypes is expected to be low.

Sensitization to HDM, smoking, baseline FEV₁, and seasonal variation of BHR examination were further investigated as possible confounders. Adjustment for these confounders within logistic regression or analysis of subgroups confirmed the significant underrepresentation of Gly16/Gln27/Thr164 haplotype carriers within the group of BHR⁺ individuals. In this group significantly more individuals had elevated levels of HDM-specific IgE which might trigger BHR. On the other hand the observed underrepresentation of the Gly16/Gln27/ Thr164 haplotype carriers might also be associated with a sensitization to HDM. However, the finding of a significant underrepresentation of the Gly16/Gln27/Thr164 haplotype among BHR⁺ individuals upon reanalysis of probands negative for HDM-specific IgE supports a direct protective effect of this haplotype on the development of BHR. The contribution of smoking to the prevalence of BHR has been estimated to be relatively low (28). Smoking might rather be a cause for the observed lower mean baseline FEV₁ of the case group compared with the control group. Indeed, the logistic regression analysis with adjustment for smoking still revealed the significant underrepresentation of the Gly16/Gln27/Thr164 haplotype in the BHR⁺ group (p = 0.009). Moreover, the association of the Gly16/Gln27/Thr164 haplotype carrier status with BHR also persisted in a logistic regression analysis when the baseline FEV₁ was introduced as covariate (p = 0.007).

A recent association study demonstrated that the Gly16/ Gln27 haplotype predisposes for the development of BHR persisting for 10 mo in a limited sample of young males ranging in age between 18 and 25 yr (29). Because d'Amato and coworkers (29) did not type the polymorphism encoding the variation of amino acid 164, a direct comparison of their data with our results is not possible. But even if the typing of the polymorphism at amino acid 164 is not considered, our haplotype frequencies differ from those reported by d'Amato and coworkers in their sample (Arg16/Gln27 34.7%, Gly16/Gln27 34.3%, and Gly16/Glu27 31.0% as compared with 36.6%, 44.7%, and 18.7%, respectively, in our sample). Because the study of d'Amato and coworkers only included young males, we reanalyzed the association of the Gly16/Gln27/Thr164 haplotype with BHR separately for males and females (Table 5). The estimated ORs for Gly16/Gln27/Thr164 indicate the possibility that the protective effect of Gly16/Gln27/Thr164 on BHR is more prominent among females than males. In our sample a significant effect was only shown for the larger female subgroup. When age was introduced as covariate into the logistic regression analysis, the underrepresentation of Gly16/ Gln27/Thr164 carriers among bronchohyperreactive individuals remained significant (p = 0.005). Finally, in the report of d'Amato and coworkers persistent BHR coincides with seasonal variation of BHR prevalence. In our study sample no association of BHR with the month of testing by methacholine challenge could be observed (p = 0.811). Introduction of the month when the bronchial provocation was performed as covariate in a logistic regression analysis still revealed the protective effect for Gly16/Gln27/Thr164 carriers (p = 0.005).

Multiple logistic regression using a forward and a backward selection process yielded Gly16/Gln27/Thr164 carrier status, presence or absence of HDM sensitization, age, and sex as important predictors of BHR. However, no notable change of the OR estimates for Gly16/Gln27/Thr164 haplotype carriers as compared with individuals negative for this haplotype was observed when adjusted for the other predictors as covariates. Therefore, only the unadjusted OR for Gly16/Gln27/Thr164 carriers (OR = 0.53; 95% CI, 0.34 to 0.82; p = 0.005) is reported here. Finally, in an attempt to assess the protective role of the Gly16/Gln27/Thr164 haplotype, we determined the OR for heterozygote carriers (OR = 0.560; 95% CI, 0.350 to 0.896; p = 0.016) and for homozygous carriers (OR = 0.361; 95% CI, 0.119 to 1.094; p = 0.072). Because there are only 22 homozygous carriers in the sample, no significant effect for these individuals can be demonstrated. Nevertheless, the data indicate an even larger protective effect for homozygotes than for heterozygotes.

Finally, because we arbitrarily dichotomized the quantitative trait of BHR by setting the cutoff at 20% reduction of FEV₁, there might be a considerable extent of overlap between the BHR⁺ and BHR phenotypes. We therefore reanalyzed our data by separating clearly bronchonormoreactive (FEV₁ reduction < 10%) individuals and bronchohyperreactive (FEV₁ reduction  20%) subjects by a group of probands with an intermediate bronchoreactivity (FEV₁ reduction between 10 and 20%) (Table 6). The overall ² test performed for association of each of the three observed ADRB2 haplotypes with BHR on the three groups revealed only a significant p value for the Gly16/Gln27/Thr164 haplotype (p = 0.006). This was due to the significant underrepresentation of this haplotype in the group of BHR⁺ individuals compared with the BHR probands (OR = 0.517; 95% CI, 0.343 to 0.779; p = 0.001). Again, this significant difference was only observed for female Gly16/Gln27/Thr164 haplotype carriers (9.9% of BHR⁺ compared with 20.5% of BHR; OR = 0.427; 95% CI, 0.237 to 0.767; p = 0.003). Of the other parameters possibly promoting BHR, sensitization for HDM (p = 0.059), current smoking (p = 0.003), and low baseline FEV₁ (p < 0.001) were more frequently within the group of BHR⁺ individuals as compared with the bronchonormoreactive probands. But when these were included as covariates in a logistic regression, the underrepresentation of the Gly16/Gln27/Thr164 haplotype within the group of BHR⁺ individuals as compared with the normoreactive probands still remained significant (p = 0.022).

In conclusion, further analyses aimed at determining the involvement of the ₂AR in bronchial hyperreactivity require not only a careful definition of the BHR phenotype but also the precise determination of ADRB2 alleles.