Introduction

Abstract Introduction References

Constitutive endothelial nitric oxide synthase (NOS3) oxidatively deaminates L-arginine to L-citrulline with the release of nitric oxide (1). NOS3 may regulate platelet aggregation, leukocyte adhesion to the endothelium, monocyte chemotaxis, and the tone and mitotic activity of vascular smooth-muscle cells (1). NOS3 is a particulate enzyme that has been detected in the Golgi complex (2) and caveolae, the site of other signaling proteins (3). In target tissues, nitric oxide activates soluble guanylate cyclase, thus stimulating regulatory pathways involving cGMP-dependent protein kinases (1).

Participation of NOS3 polymorphic variants in the pathogenesis of disease has been studied in coronary artery disease (6) and hypertension (7). Nitric oxide, including that produced by NOS3, may regulate vascular and airway tone in the lung and influence various aspects of airway homeostasis (8, 9). In view of the importance of these functions to the lung, we hypothesized that abnormalities in NOS may contribute to the pathogenesis of pulmonary disease. To date, however, there is little information on the possible role of genetic variants of NOS3 in pulmonary diseases.

Chronic emphysema resulting from a deficiency in 1-antitrypsin (1AT), an inhibitor of neutrophil elastase, is an autosomal recessive disorder caused by mutant variants of the human PI locus on chromosome 14q 32.1 (10). The degree of lung pathology and the rate of decline of pulmonary function in individuals with 1AT deficiency varies markedly (11), suggesting that other factors, including hereditary background, may play a role in the clinical presentation. The genetic profile of an individual including loci with relevant roles in pulmonary function may constitute a background that affects the individual's susceptibility to lung destruction (12). Genetic variants of NOS3 may be involved in such a background. In this study, we surveyed genetic polymorphisms within the NOS3 gene in relation to the severity of lung disease in patients with 1AT deficiency. A higher incidence of 774T and 894T alleles was observed in the most severely affected individuals, suggesting that NOS3 allelic variants contribute to the pathogenesis of the disease.

	Materials and Methods

Subjects

Genomic DNA samples from 345 individuals with 1AT deficiency were used. All patients were either PiZ homozygotes or compound heterozygotes of PiZ with other "deficiency" alleles. Demographic information, 1AT plasma levels, pheno- and genotypes of 1AT, results of pulmonary function testing, and clinical and smoking histories are available elsewhere (11, 13). The patient sample was stratified according to the severity of lung disease based on pulmonary function testing. The group with more advanced disease was defined prior to data analysis as those individuals with a forced expiratory volume in 1 s (FEV₁) less than 35% of predicted values. Within this group, the individuals with a diffusion capacity for carbon monoxide (DL_CO) below expected were considered the most affected subgroup. Estimation of the expected DL_CO was based on the linear regression DL_CO = 0.6896 × FEV₁ + 27.36 (Figure 1). The control group contained 93 asymptomatic whites of both sexes aged from 18 to 55 yr who participated in the National Heart, Lung, and Blood Institute clinical protocols as normal volunteers.

View larger version (23K): [in this window] [in a new window]   Figure 1.   Severity of lung disease in individuals with 1AT deficiency. In severe emphysema, reduction in the FEV1 correlates with a decrease in the DLCO. The linear regression DLCO = 0.6896 × FEV1+ 27.36 connects these two parameters of pulmonary function. The individuals with an FEV1 less than 35% of predicted values (to the left of the vertical line) were considered a more affected group. Within the latter, subjects with a DLCO lower than expected from the regression (shaded sector) are the most severely affected, with a pattern of lung disease consistent with an advanced emphysema.

Polymerase Chain Reaction

Amplification of the polymerase chain reaction (PCR) fragments was performed in a 10-µl volume using the model 9600 GeneAmp PCR system (Perkin-Elmer, Norwalk, CT). PCR primers were synthesized on the model 380B DNA/RNA synthesizer (Perkin-Elmer Applied Biosystems Division, Foster City, CA). Primer design was based on available NOS3 genomic and cDNA sequences (GenBank accession numbers M95296, M93718, X76303- X76316, L23210, L10693-L10709, D26607, and U24214). The reaction mixture contained 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl₂, 0.2 mM each of deoxynucleotide triphosphates, 0.4 µM of each primer, and 0.025 U/µl Taq polymerase (Promega, Madison, WI). PCR was performed with 28 to 30 cycles of 20 s at 95°C, 25 s at 56°C to 70°C, and 30 s at 72°C, followed by final incubation for 5 min at 72°C.

Detection of NOS3 Polymorphisms

Search for variable sites within the NOS3 gene was performed by single-strand conformational polymorphism (SSCP) (14) and denaturing gradient gel electrophoresis (DGGE) (15). For the SSCP analysis, the PCR fragments of the NOS3 gene were denatured by heating at 95°C with an equal volume of formamide, followed by electrophoresis in nondenaturing 6% polyacrylamide gel with 1× Tris-borate-ethylenediaminetetraacetic acid (EDTA) buffer at room temperature. DGGE was performed with the same PCR products. Taq polymerase in the PCR samples was inactivated by addition of 0.1 volume of 125 mM EDTA; samples were then denatured by heating 5 min at 95°C and renatured by cooling to room temperature. Presence of heteroduplexes was assessed by electrophoresis in gels containing 0 to 7 M urea and 5 to 10% acrylamide gradients. To enhance resolution, a 4.5% acrylamide gel containing 0.75× Tris-glycine (TG) buffer was used as a "stacking" gel, and 0.75× TG was used as a cathode buffer. In both SSCP and DGGE gels, the DNA bands were visualized by staining with ethidium bromide or SYBR Green II (FMC, Rockland, ME).

Genotyping of NOS3 Polymorphisms

When the products of different alleles varied in length, as with the intron 4 variable number of tandem repeats (VNTR), the genotypes were determined directly by electrophoresis of the PCR fragments in polyacrylamide gels (Figure 2D). Genotyping by SSCP was used for the 1998C/ G site, where variants produced unambiguous SSCP patterns (Figure 2G). For other nucleotide substitutions, PCR-restriction fragment length polymorphism (PCR-RFLP), was used (Figures 2A-2C, 2E, and 2F). Primer sequences and procedures are given in Table 1. The restriction endonucleases were obtained from New England Biolabs, Beverly, MA.

View larger version (49K): [in this window] [in a new window]   Figure 2.   Genotyping of NOS3 polymorphisms (also see Table 1). DNA fragments were PCR amplified from genomic DNA. PCR products (7 to 10 µl) were used for further genotype determination using polyacrylamide gel electrophoresis followed by ethidium bromide staining. (A, B, C, E, F ) Electrophoretic patterns of the 924A/G, 788C/T, 691C/T, 774C/T, and 894G/T polymorphic sites genotyped using PCR-RFLP and resolved in 6% polyacrylamide gels. (D) PCR fragments containing the intron 4 VNTR, resulting in different, allele-specific fragment size were directly typed in 6% polyacrylamide gels. (G) Polymorphic site 1998C/G produced easily interpretable SSCP patterns that were directly used for genotyping.

DNA Sequencing

PCR fragments for sequencing were amplified from genomic DNA or the primary PCR product isolated from polyacrylamide gels by water elution of excised bands. The PCR fragments were gel-purified, and approximately 100 fmol were bidirectionally sequenced with the dsDNA Cycle Sequencing System (Life Technologies, Gaithersburg, MD). Primers were 5'-end-labeled using ³³P- adenosine triphosphate (NEN, Boston, MA).

Data Analysis

Genotypes were directly counted, and general population genetic statistical analysis was performed, including analysis of genotype distribution, calculation of allele frequencies, and testing of Hardy-Weinberg equilibrium (16). Genotype distributions in different groups were compared by testing the hypothesis of homogeneity using chi-square criteria. Linkage disequilibrium was evaluated by Lewontin's measure of linkage disequilibrium (D') (17). Calculation of relative risk and corresponding 95% confidence intervals (CI) was performed as described (18). In the comparison of allele frequencies, a P value less than 0.025 (0.05/2) rather than 0.05 was regarded as statistically significant to compensate for the multiple testing of the two groups of uncorrelated polymorphic sites (i.e., 924A/G, 788C/T, 691C/T versus 774C/T, 894G/T, and 1998C/G). Also, independence was assumed between the two alleles for each person at each site during analysis.

	Results

Polymorphism of the NOS3 Gene

Sites 774C/T and 894G/T, counted from the start codon of human NOS3, are polymorphic (19). These polymorphisms are in a strong linkage disequilibrium; the 774T allele appears to be in a cis-phase with 894T (Table 2). Four additional polymorphic sites were identified: 924A/G, 788C/T, and 691C/T in the NOS3 promoter, and 1998C/ G in exon 16. In two individuals, a rare allele, NOS3 c, with six repeats in a locus within intron 4 VNTR, was found. Both of them had the genotype NOS3 a/c (Figure 2D). Measures of linkage D' and pairwise correlation coefficients between markers (r) (17) are in a moderate correspondence with genomic distances (Table 2, Figure 3).

View larger version (18K): [in this window] [in a new window]   Figure 3.   Human endothelial nitric oxide synthase gene polymorphism. Positions of the six detected polymorphisms are indicated above the scheme of the genomic organization of NOS3 (Reference 19 and this study). The polymorphic sites are numbered from the translation start. The numbered boxes on the scheme indicate the NOS3 exons. Measures of linkage disequilibrium (D') between the adjacent sites do not completely correlate with the genomic distances (see also Table 2), suggesting that other genetic polymorphisms linked to the 774C/T and 894G/T sites exist in the wide genomic area of the NOS3 gene.

Distribution of the 774C/T and 894G/T Polymorphisms in Patients and Control Subjects

Tables 3 and 4 summarize a distribution of the 774C/T and 894G/T genotypes and allele frequencies in the control group, the entire patient group, and in subgroups with differences in severity of lung pathology, based on the results of pulmonary function testing of 177 patients. In the total patient population studied, there were 21 related patients representing nine families. None of the patients with an FEV₁ < 35% were related and none were related in the control population. Individuals with FEV₁ less than 35% of predicted values were considered as having more advanced pulmonary disease. The frequency of the 774T allele in that group, 0.417, significantly exceeds that in the control group, 0.269 (² = 8.11, P = 0.018), and in the rest of the patient group, 0.288 (² = 6.79, P = 0.034). The relative risk for homo- and heterozygous carriers of the 774T allele with FEV₁ less than 35% versus healthy control individuals is 1.427 (95% CI, 1.116 to 1.825) (Table 5).

Frequency of the 894T allele in the individuals with FEV₁ < 35% is 0.427 versus 0.280 in the control group (² = 7.48, P = 0.024) (Table 4). In the rest of the patient group, this allele has an intermediate frequency of 0.344, which is slightly higher than in the control group. The relative risk for homo- and heterozygous carriers of the 894T allele with FEV₁ of less than 35% versus healthy control individuals is 1.439 (95% CI, 1.196 to 1.836) (Table 5).

The FEV₁ and DL_CO in the studied group of patients with 1AT deficiency are correlated (r = 0.69). In patients with severe emphysema, reduction in the FEV₁ usually is accompanied by a dramatic decrease in the DL_CO (11). For the purpose of further patient classification according to the severity of emphysema, we calculated the linear regression, DL_CO = 0.6896 × FEV₁ + 27.36. The individuals with an FEV₁ less than 35% of predicted values and a DL_CO lower than expected from the regression equation, were considered the most severely affected group (Figure 1). This subgroup of patients has an especially high frequency of the 774T allele, 0.519 (n = 26), which is significantly different from the control group (² = 11.45, P = 0.003) and from the remainder of the patients (² = 10.63, P = 0.005) (Table 3). The 894T allele frequency in this subgroup is 0.517 (n = 29), which also is significantly higher than that in the control group (² = 11.77, P = 0.003) and in the remainder of the patients (² = 7.21, P = 0.027) (Table 4).

Analysis for Confounders

To investigate whether the interesting significant findings in the FEV₁ < 35% patients were possibly affected by confounding from a host of factors, we compared this group of 55 patients to a subset of age-, sex-, and race-matched controls in an additional analysis. Age was matched within 10 yr. The same trends in the 774T and 894T frequencies in the age-adjusted comparison were found as in the age-unadjusted control comparison. In patients with an FEV₁ < 35% versus their matched controls, the frequency of the 774T allele was 0.417 versus 0.278 (P = 0.032), respectively, whereas the frequency of the 894T allele was 0.427 versus 0.311 (P = 0.078), respectively. These frequencies changed very little in the new analysis, and the observed change in the P values is due mostly to the smaller sample size. We conclude that it is not likely that the factors of age, sex, and race confound the original findings.

We also focused on United States regional variation as a possible confounder in FEV₁ < 35% patients and matched control subjects. Geographic data were not available on four patients and 10 control subjects because these data were not collected routinely at the beginning of the study. Specifically, the percentages of patients in the Northeast, Southeast, Northwest, and Southwest were 35%, 33%, 8%, and 24%, respectively; the percentages of control subjects were 55%, 41%, 2%, and 2%, respectively. The numbers were too small to permit comparisons across regions.

We also attempted to compare the effect of smoking status in the subset of 55 patients with FEV₁ < 35% and their matched controls. Almost all of our patients and control subjects were ex-smokers or nonsmokers by self-report, so it is not possible to investigate the effect of current smoking. Regarding the ex-smokers, we do not have detailed information as to when they last smoked or their smoking pattern before quitting. The smoking status of four patients and one control subject was unknown. We found the smoking distribution between the patients and control subjects to be comparable. In patients, the percentages of smokers, ex-smokers, and nonsmokers were 6%, 43%, and 51%, respectively; in control subjects the percentages were 8%, 30%, and 62%, respectively (P = 0.390). In subsequent analyses, the few smokers were grouped together with the ex-smokers.

The comparison of patients and control subjects by smoking category on the frequency of 774T and 894T alleles shows a trend toward significance in the group of smokers plus ex-smokers. Specifically, in the group of smokers plus ex-smokers, the 774T allele frequency in the patients versus controls was 0.460 versus 0.275 (P = 0.072), respectively, whereas the 894T allele frequency was 0.480 versus 0.300 (P = 0.083), respectively. In the nonsmokers, we find that the 774T allele frequency in the patients versus controls was 0.365 versus 0.278 (P = 0.371), respectively, and the 894T allele frequency was 0.385 versus 0.328 (P = 0.527), respectively. In summary, we find a trend toward significance in the group of smokers and ex-smokers, but it is not well established with the current data, which captured too little detail on smoking history.

	Discussion

The 774C/T polymorphism, which is located in exon 6, is "silent," that is, the C to T transition alters the third base of the GAC codon, changing it to GAT, both encoding an Asp. The 894G/T alleles, located in exon 7, probably are not functionally different because the resulting amino acid difference at position 298 in the oxygenase domain is a conservative one (Glu versus Asp). In fact, NOS3 variants containing either glutamate or aspartate, when expressed in COS7 cells, exhibited catalytic activity (P. Rogliani and colleagues, unpublished data). Thus, neither glutamate nor aspartate at position 298 is necessary for function. Therefore, both 774C/T and 894G/T polymorphisms may be neutral genetic markers of a yet unidentified functional NOS3 polymorphism, which is in linkage disequilibrium with them. The alleles 774T and 894T must be in a cis-phase with the proposed NOS3 allele with altered function. It is reasonable to assume that the possible location of the proposed polymorphism will be confined to the area of the NOS3 gene that includes the 774C/T and 894G/T sites, and other nearby polymorphisms, which are in linkage disequilibrium.

Because disequilibrium between the 1998C/G and 774C/T to 894G/T is high, the region downstream from the 774C/T to 894G/T site may contain the proposed mutation. However, identification of more polymorphic sites here, in addition to the 1998C/G, would be useful for a more accurate conclusion. In general, relatively poor correspondence of the genomic distance between NOS3 polymorphisms and parameter D' of linkage disequilibrium (Table 2, Figure 3) makes it difficult to define accurately the area where the NOS3 functional polymorphism may be located. Thus, a relatively wide area of the NOS3 gene with the epicenter at the 774C/T to 894G/T locations must be examined.

Age, sex, and race do not change the frequency of the T allele as shown in the analyses with age-, sex-, and race-matched controls. Because of the trend toward significance in the frequency of the 774T and 894T alleles in smokers and ex-smokers, future studies should be performed with a more detailed smoking history. Future studies need to be performed on regional variation, as our study was performed as a post hoc finding in the presence of multiple comparisons.

Wang and colleagues (6) recently reported an association between the intron 4 VNTR and coronary artery disease (CAD). This genetic marker is located 1.34 kilobases (kb) from the 774C/T and 1.73 kb from the 894G/T site (Figure 3). Values of D' between them are 0.845 and 0.926, respectively, although the correlation coefficients are low (Table 2). This discrepancy between the two measures of interloci correlation may be explained by the fact that the less frequent allele, NOS3 a, of the intron 4 VNTR, is in a coupling phase with the more frequent allele "C" of the 774 locus. The majority of 774T/T homozygotes fall into a category of NOS3 b/b homozygotes. Conceivably, a putative functional alteration, caused by the alleged NOS3 polymorphism in a vicinity of the 774C/T and the intron 4 VNTR markers, may be associated with an elevated risk of lung pathology but lower risk of CAD and vice versa.

In conclusion, we have found a statistically significant association between the polymorphic sites 774C/T and 894G/T of human NOS3 gene and severity of lung disease in individuals with 1AT deficiency. The observed accumulation of hereditary isoforms of NOS3 in subjects with more severe lung pathology may be related to other NOS3 mutations with functional significance. Further investigation is needed to identify these variants, which may be located in a wide genomic area, including but not limited to the location of the polymorphic sites 774C/T and 894G/T.