Skip to main content
NCBI home page
American Journal of Human Genetics logoLink to American Journal of Human Genetics
. 2005 Dec 15;78(2):253–264. doi: 10.1086/499828

The SERPINE2 Gene Is Associated with Chronic Obstructive Pulmonary Disease

Dawn L DeMeo 1,2,4, Thomas J Mariani 2,4, Christoph Lange 5, Sorachai Srisuma 2,4,10, Augusto A Litonjua 1,3,4, Juan C Celedón 1,3,4, Stephen L Lake 1,4, John J Reilly 2,4, Harold A Chapman 11, Brigham H Mecham 2,4, Kathleen J Haley 2,4, Jody S Sylvia 1, David Sparrow 1,2,4,6,7, Avrum E Spira 7,8, Jennifer Beane 8, Victor Pinto-Plata 9, Frank E Speizer 1,4,5, Steven D Shapiro 2,4, Scott T Weiss 1,4,5, Edwin K Silverman 1,2,4
PMCID: PMC1380249  PMID: 16358219

Abstract

Chronic obstructive pulmonary disease (COPD) is a complex human disease likely influenced by multiple genes, cigarette smoking, and gene-by-smoking interactions, but only severe alpha 1-antitrypsin deficiency is a proven genetic risk factor for COPD. Prior linkage analyses in the Boston Early-Onset COPD Study have demonstrated significant linkage to a key intermediate phenotype of COPD on chromosome 2q. We integrated results from murine lung development and human COPD gene–expression microarray studies with human COPD linkage results on chromosome 2q to prioritize candidate-gene selection, thus identifying SERPINE2 as a positional candidate susceptibility gene for COPD. Immunohistochemistry demonstrated expression of serpine2 protein in mouse and human adult lung tissue. In family-based association testing of 127 severe, early-onset COPD pedigrees from the Boston Early-Onset COPD Study, we observed significant association with COPD phenotypes and 18 single-nucleotide polymorphisms (SNPs) in the SERPINE2 gene. Association of five of these SNPs with COPD was replicated in a case-control analysis, with cases from the National Emphysema Treatment Trial and controls from the Normative Aging Study. Family-based and case-control haplotype analyses supported similar regions of association within the SERPINE2 gene. When significantly associated SNPs in these haplotypic regions were included as covariates in linkage models, LOD score attenuation was observed most markedly in a smokers-only linkage model (LOD 4.41, attenuated to 1.74). After the integration of murine and human microarray data to inform candidate-gene selection, we observed significant family-based association and independent replication of association in a case-control study, suggesting that SERPINE2 is a COPD-susceptibility gene and is likely influenced by gene-by-smoking interaction.

Chronic obstructive pulmonary disease (COPD [MIM 606963]) is a complex human disease likely influenced by multiple genes, environmental factors (especially cigarette smoking), and gene-by-smoking interactions. As the worldwide disease burden of COPD continues to increase, better treatments may be developed through understanding the variable susceptibility for developing this devastating lung disease. Cigarette smoking is the most important environmental risk factor for the development of COPD, but individuals vary widely in their susceptibility to the pulmonary effects of tobacco smoke (Burrows et al. 1977).

The only proven genetic risk factor for COPD is severe deficiency of alpha 1-antitrypsin (Laurell and Eriksson 1963). However, this genetic deficiency is present in only 1% of individuals with COPD, suggesting that COPD represents a complex disease with contributions from multiple genes and environmental risk factors. Genomewide linkage analyses in the Boston Early-Onset COPD Study have helped to narrow the scope of candidate-gene investigation for COPD (Silverman et al. 2002a, 2002b; Palmer et al. 2003), but the linkage peaks defined still represent large genomic regions with many biologically plausible candidate genes. Candidate-gene association studies based on the known pathophysiology of COPD have been characterized by inconsistent replication of positive association results (Hersh et al. 2005). Significant linkage of an intermediate phenotype of COPD—the ratio of forced expiratory volume at one second to forced vital capacity (FEV1/FVC)—with chromosome 2q has been observed in extended pedigrees ascertained through probands with severe, early-onset COPD in the Boston Early-Onset COPD Study (Silverman et al. 2002b; Palmer et al. 2003). This region of linkage likely harbors genes that contribute to COPD susceptibility through gene-by-smoking interactions (DeMeo et al. 2004). A similar region on chromosome 2q has been linked to FEV1/FVC in the general population (Malhotra et al. 2003).

We intersected human linkage results on chromosome 2q with data from expression-array analysis of murine and human lung tissues, to prioritize the investigation of positional COPD candidate genes in this region and to identify new candidate genes for COPD. Using this methodology, we identified SERPINE2 for further investigation as a novel candidate COPD-susceptibility gene, potentially relevant in both early-onset and later-onset COPD not related to alpha 1-antitrypsin deficiency.

Methods

Murine Expression-Array Experiments

We evaluated the expression of genes within a genomic region on chromosome 2q bracketed by positions 212,742,577 and 230,264,438 (UCSC Genome Bioinformatics); this region represents the 1.5-LOD unit of support interval for linkage on 2q (Silverman et al. 2002b). We evaluated the expression of genes within this genomic region in a microarray data set of normal mouse lung development, generated using the Affymetrix Mu11K platform (Mariani et al. 2002). Sequence verification of microarray probe sets was performed using the Reference Sequence database with the resources available at the Lung Transcriptome Web site, as described elsewhere (Mecham et al. 2004). Tests for significance of fold change were performed using the variable fold-change threshold method (Mariani et al. 2003).

Human COPD Microarray Experiments

A human microarray data set was available, generated using the Affymetrix human U133A platform; a complete description of case and control selection has been published elsewhere (Spira et al. 2004). This data set included 34 lung tissue samples; 18 were from individuals with moderate-to-severe emphysema who underwent lung volume–reduction surgery (LVRS) for COPD, and 16 were non-LVRS normal-to-minimally emphysematous control human lung samples from individuals undergoing pulmonary nodule resection. Expression values were generated from the raw image files by use of the robust microarray averaging method (Irizarry et al. 2003), including the quantile normalization procedure, as implemented in Bioconductor. Pulmonary function test information was available for these patients, including pre- and postbronchodilator FEV1, FEV1/FVC, total lung capacity (TLC), diffusing capacity for carbon monoxide (DLCO), and intensity of smoking (pack-years [number of packs smoked per day multiplied by the number of smoking years] of cigarettes smoked). The significance analysis of microarray (SAM) method (Tusher et al. 2001) was implemented using the publicly available TIGR Multi-Array Viewer (MEV) software (TM4). Additionally, the Pearson and Spearman rank correlations were calculated for the association of SERPINE2 expression with each of the lung function parameters. The significance of each correlation was determined using the Student’s t test. Raw data from the human microarray experiments have been deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (series reference number GSE1650).

Immunohistochemistry

Immunostaining in normal adult mouse and human lung tissue was performed using a rabbit polyclonal antibody raised against recombinant bovine serpine2. The polyclonal antibodies were produced in rabbits against bovine serpine2, by use of recombinant, purified serpine2 and bovine follicular fluid. The antibodies were validated by immunoblotting. The antibodies used in our mouse and human immunohistochemistry experiments were a generous gift from Dr. J. G. Lussier. The complete method of the serpine2 antibody development has been described elsewhere (Bedard et al. 2003).

Human Populations for Genetic Association Studies

Three separate cohorts were analyzed, including participants in the Boston Early-Onset COPD Study, the National Emphysema Treatment Trial (NETT), and the Normative Aging Study (NAS). The Partners Institutional Review Board (IRB) for human studies approved the study protocols. Anonymized data sets were used for the NAS participants, as approved by the IRBs of both Partners and the Veterans Administration Hospitals.

Boston Early-Onset COPD Study

The recruitment and characteristics of probands and family members enrolled in the Boston Early-Onset COPD Study have been reported elsewhere (Silverman et al. 1998). In brief, ascertainment criteria for probands in the Boston Early-Onset COPD Study included an FEV1 <40% of predicted, age <53 years, and no evidence of severe alpha 1-antitrypsin deficiency. All available first-degree and older second-degree relatives were invited to participate in the study; there were 127 probands and a total of 949 individuals in the family-based association analysis for SERPINE2.

Each study participant completed a modified version of the 1978 American Thoracic Society–Division of Lung Diseases respiratory questionnaire (Ferris 1978). Spirometry was performed as described elsewhere (Silverman et al. 1998). Most individuals completed pre- and postbronchodilator spirometry (performed ∼15 minutes after albuterol administration); we focused on the results from postbronchodilator trials (Palmer et al. 2003). For the analysis of quantitative phenotypes, absolute volume spirometric results with covariate adjustment, as detailed below, were used in all analyses. For the analysis of qualitative phenotypes, percent predicted values for FEV1 and FEV1/FVC were calculated for white adult participants by use of equations formulated by Crapo et al. (1981). For white individuals aged <18 years, predicted values were determined from equations by Hsu et al. (1979), with predicted values for FEV1/FVC calculated from equations by Knudson et al. (1983). For African American participants, predicted values were determined by equations developed by Hankinson et al. (1999). For the qualitative analysis, an FEV1 <60% predicted in the presence of FEV1/FVC <90% predicted defined moderate-to-severe COPD; FEV1 <80% predicted in the presence of FEV1/FVC <90% predicted defined mild-to-severe COPD.

NETT

The NETT is a randomized multicenter treatment trial for the investigation of outcomes in a group of individuals with COPD randomized to conventional medical therapy versus LVRS for the treatment of severe emphysema (NETT Research Group 1999). All participants had an FEV1 of ⩽45% predicted, evidence of hyperinflation on pulmonary-function tests, and emphysema on high-resolution CT scanning of the thorax at study enrollment (NETT Research Group 1999). Of those who were enrolled in the NETT Genetics Ancillary Study and had blood samples available for genotyping, 304 white individuals were included as COPD cases in the present analysis.

NAS

The NAS is a longitudinal study of aging performed by the Veterans Administration (Bell et al. 1966). At the time of study entry, the cohort consisted of 2,280 healthy community-dwelling men from the Boston area who were aged 21–80 years old between 1961 and 1969. Since 1984, participants have completed questionnaires regarding respiratory symptoms and smoking history, and spirometric evaluation has been performed. We selected 441 white control individuals with blood samples available for genotyping; these individuals had no evidence of airflow obstruction (FEV1>80% predicted and FEV1/FVC>90% predicted at last visit) and at least a 10–pack-year smoking history (Celedón et al. 2004).

Genotyping

Initially, seven SERPINE2 SNPs were genotyped in the Boston Early-Onset COPD Study pedigrees, NETT cases, and NAS controls, with use of Taqman 5′ exonuclease assays, with primers obtained from Applied Biosystems (ABI). Major- and minor-allele probes were labeled with different colored fluorophores, and probe fluorescence signal detection was performed using the ABI Prism 7900 Sequence Detector System.

Subsequently, DNA sequencing was performed on all SERPINE2 exons and exon-intron boundaries. Sequencing efforts in 34 COPD-affected probands from the Boston Early-Onset COPD Study and 12 control individuals identified 17 polymorphisms; 4 of these polymorphisms (NCBI dbSNP Build 126 ss numbers 49785623, 49785624, 49785625, and 49785626) had not been previously reported in public databases. For the second round of genotyping, 41 additional SNPs (17 identified during sequencing and 24 more from public databases) were genotyped in both the Boston Early-Onset COPD Study families and NETT-NAS case-control populations, with use of the SEQUENOM platform, for a total of 48 SNPs in both cohorts (with 7 from the first round and 41 from the second round of genotyping).

Statistical Methods for Genetic Association Analysis

Pedigree-Based Association Test (PBAT)

For the family-based single-SNP association analyses and the family-based haplotype analyses of SERPINE2 in the Boston Early-Onset COPD Study, we used PBAT (Lange et al. 2004) (C.L. Web site). We had no biological data supporting a dominant or recessive model, so we used the additive model to minimize multiple statistical testing. The association analysis was adjusted for age, sex, height, smoking history (yes/no), and pack-years of cigarette smoking and included a SNP–by–pack-years (gene-by-environment) interaction term. Because SERPINE2 is located in a genomic region linked to COPD-related phenotypes, the null hypothesis of our PBAT analyses was no association in the presence of linkage. In this setting, PBAT adjusts for linkage by using the empirical variance in the family-based association test (FBAT) statistic. P values ⩽.05 are reported. For the models including the interaction term, a composite P value that includes the association for the main SNP effect and the gene-by-smoking interaction is reported. In general, optimal model fitting is possible in FBAT, by use of the conditional-power approach as described by Lange and Laird (2002). However, model selection that is based on conditional power calculation is computationally feasible only in the setting of small pedigrees, and this approach was not feasible for the extended pedigrees of the Boston Early-Onset COPD Study.

Conditional linkage models were analyzed, including the SNP data and a SNP-by-smoking interaction term by use of a variance component approach, as implemented in the Sequential Oligogenic Linkage Analysis Routines (SOLAR), version 1.7.4 (Almasy and Blangero 1998). Overall and smokers-only linkage models were generated using a panel of 377 STR markers with relevant covariates (including age, sex, height, pack-years of smoking, and SERPINE2 SNP of interest), as described elsewhere (DeMeo et al. 2004).

Case-Control Replication Analysis

Association between SERPINE2 SNPs and severe COPD among NETT cases and NAS controls was investigated with the SAS Genetics program (SAS Institute). Hardy-Weinberg equilibrium was evaluated among the controls for each SNP by use of the goodness-of-fit test, as implemented through SAS Genetics. Association between each of the individual SNPs and COPD was investigated using 2×2 contingency tables and by a trend test to investigate allelic additive effects. To investigate allelic additive effects with the Cochran-Armitage trend test and to investigate dominance effects with the genotype case-control test, 2×3 contingency tables were used. Since the control population was all male, a males-only analysis was performed in addition to an analysis that included all of the NETT case participants.

Haplotype frequencies were estimated, and the association between haplotypes with a frequency >5% and COPD was tested using global and haplotype-specific statistics with the haplo.stats program (Schaid et al. 2002). Sliding windows of 6, 4, and 2 adjacent SNPs were used to test for association across the SERPINE2 gene and to localize the region of most-significant association (Clayton and Jones 1999; Schaid et al. 2002). Simulated P values were also generated for 6-, 4-, and 2-SNP sliding windows by use of 1,000 replicates; the negative log of the P value was graphed against the midpoint SNP of the sliding window to demonstrate the region of most-significant association.

Results

Mouse Expression-Array Experiments

In a microarray data set describing normal mouse lung development generated using the Affymetrix Mu11K platform (Mariani et al. 2002), 34 probe sets in the murine data set that matched 24 independent mouse orthologues for human genes were present within the region of human chromosome 2q linkage. Expression data from the 25 Reference Sequence–verified probe sets, representing 23 genes, are presented in figure 1. The majority of the genes were expressed in the lung at a signal intensity <4,000. Three genes were expressed at a signal intensity >15,000 during at least one time point, including RPL37A (a ribosomal protein), ITM2C (an integral membrane protein highly expressed in the brain), and SERPINE2. SERPINE2 was most highly expressed during murine alveogenesis (postnatal days 4–10) and had the greatest expression change (4.5-fold) across the developmental time series. This fold change was significant with use of the variable fold-change threshold method, suggesting that this gene is regulated during normal lung development.

Figure 1.

Figure 1

Open in a new tab

Gene expression during normal murine lung development. We assessed the expression of genes within the chromosome 2–linked locus, using a microarray data set of normal mouse lung development. Shown are the expression profiles for 25 probe sets, representing 23 mouse orthologues for human genes present within the locus. Signal intensities were determined using MAS 4.0. Most genes were expressed at low levels in the lung (signal intensity <4,000); seven genes were expressed at a signal intensity >4,000. Three of these genes were expressed at a signal intensity >15,000 during at least one time point, including RPL37A (a ribosomal protein), ITM2C (an integral membrane protein highly expressed in brain), and SERPINE2. E = embryonic day; P = postnatal day; AD = adult.

Human Microarray Results

A human lung microarray data set (Spira et al. 2004) from individuals with severe COPD and control subjects was evaluated for disease-associated changes in SERPINE2, ITM3C, and RPL37A expression (in this region, the three most highly expressed genes during murine lung development), by use of pulmonary-function measurements as continuous variables. SERPINE2 was positively correlated with higher TLC (Spearman correlation=0.46; P=.008, Pearson correlation=0.42, P=.020); this suggested a significant correlation with lung hyperinflation. Additionally, SERPINE2 expression was inversely correlated with postbronchodilator FEV1 (Spearman correlation=-0.43; P=.020; Pearson correlation=-0.38; P=.043) and was inversely correlated with DLCO (Spearman correlation=-0.41; P=.025; Pearson correlation=-0.41; P=.023). Higher TLC, reduced postbronchodilator FEV1, and reduced DLCO are all clinical features observed in patients with COPD. ITM2C and RPL37A each had a single Reference Sequence–verified probe set; neither showed significant correlation to any of the pulmonary-function measurements. Examination of these same data with use of SAM further suggested that SERPINE2 expression was higher in individuals with severe emphysema than in controls (fold change of 1.25, with a false-discovery rate of 6.2%). This modest fold change was verified using quantitative RT-PCR. Because of the expression pattern of SERPINE2 across normal lung development and the consistent correlations with COPD-related phenotypes, we selected SERPINE2 for further investigation.

Immunohistochemistry

We performed immunohistochemistry for SERPINE2 in postnatal mouse lungs and demonstrated prominent immunolocalization of SERPINE2 in a cell-associated pattern within conducting airway epithelial cells and in an extracellular matrix-associated pattern in the vascular adventitia (fig. 2). Immunohistochemistry of adult human lungs demonstrated an analogous staining pattern (fig. 3).

Figure 2.

Figure 2

Open in a new tab

SERPINE2 localization in normal adult mouse lung. Immunohistochemistry for SERPINE2 in the normal adult mouse lung was performed using a rabbit polyclonal antibody (PN1). Staining was evident in small airway (AW)–conducting epithelium (EPI) and in the vascular (V) adventitia in an extracellular matrix-associated (ECM) pattern. No staining was observed for control IgG.

Figure 3.

Figure 3

Open in a new tab

SERPINE2 localization in human lung. SERPINE2 immunostaining was observed in normal (A), emphysematous (C), and asthmatic (D) human lung. Staining was evident in small airway epithelial cells (EPI) and in the vascular adventitia extracellular matrix (ECM). No staining was observed for control IgG (B).

PBAT

We next investigated the genetic association of a panel of 48 SERPINE2 SNPs with COPD phenotypes (table 1). Multivariate models were evaluated in 127 pedigrees (949 individuals) in the Boston Early-Onset COPD Study. In additive models that included an interaction term to capture SNP-by-smoking (gene-by-environment) effects, 18 SNPs in SERPINE2 demonstrated significant association with quantitative and/or qualitative spirometric phenotypes (tables 2 and 3).

Table 1.

SERPINE2 SNPs Genotyped and Analyzed in the Family-Based and Case-Control Association Analyses[Note]

SNP ID SNP Alleles Position Location from ATG Start Site
1 rs282253 C/G 5′ Genomic/promoter g.−39863
2 rs1438831 C/T 5′ Genomic/promoter g.−39698
3 rs7579646 A/G Intron 1 g.−34079
4 rs840088 C/T Intron 1 g.−33227
5 rs7562213 A/G Intron 1 g.−31810
6 rs920251 A/G Intron 1 g.−26328
7 rs1371029 A/G Intron 1 g.−21856
8 rs2099601 A/C Intron 1 g.−18689
9 rs2083120 C/T Intron 1 g.−18591
10 rs2083121 C/T Intron 1 g.−18385
11 rs2099602 C/T Intron 1 g.−18298
12 rs4674849 C/T Intron 1 g.−18228
13 rs4574111 A/G Intron 1 g.−17755
14 rs1438829 C/T Intron 1 g.−17086
15 rs6436459 C/T Intron 1 g.−14397
16 rs1371028 C/T Intron 1 g.−12833
17 rs1438828 A/G Intron 1 g.−11684
18 rs1866152 C/T Intron 1 g.−11462
19 rs7588220 A/G Intron 1 g.−6987
20 rs1530020 G/T Intron 1 g.−4753
21 rs1530021 C/T Intron 1 g.−4619
22 rs6436454 C/T Intron 1 g.−4104
23 rs3948261 A/G Intron 1 g.−3275
24 rs2118409 C/G Intron 1 g.−3026
25 rs6742903 C/T Intron 1 g.−2176
26 rs12436 A/C Exon 2 g.41
27 ss49785623 A/G Exon 2 g.190
28 rs7581619 A/G Exon 2 g.230
29 rs3795877 C/T Intron 2 g.440
30 rs1866153 A/G Intron 2 g.2065
31 rs6747096 A/G Exon 3 g.3775
32 rs3795879 C/T Intron 3 exon/intron boundary g.3796
33 ss49785624 A/C Intron 3 g.3917
34 rs6715768 A/G Intron 3 g.4126
35 rs6738983 C/T Intron 3 g.7867
36 rs6721140 A/G Intron 3 g.7974
37 rs2076924 C/T Intron 3 g.9573
38 rs6712954 A/G Exon 4 g.9967
39 rs2099603 A/G Intron 4 g.13936
40 rs7605945 C/T Intron 5 g.18811
41 rs975278 C/T Intron 5 g.18910
42 rs10164837 C/T Intron 6 g.19349
43 rs729631 C/G Intron 7 g.21698
44 ss49785625 A/G Intron 7 g.21753
45 rs7597833 C/T Intron 8 g.24467
46 ss49785626 C/T Intron 8 g.24583
47 rs6734100 C/G Intron 8 g.24622
48 rs1025734 C/T 3′ Genomic g.31248
Open in a new tab

Note.— Position in Human Genome Working Draft (Goldenpath hg17, dbSNP123) from UCSC Genome Bioinformatics.

Table 2.

PBAT of Quantitative Spirometric Phenotypes with SERPINE2 SNPs, Including a SNP-by-Smoking Interaction Term[Note]

P by Spirometric test
SNP Allele FEV1 Postbronchodilator FEV1/FVC Postbronchodilator
rs7579646 G .0009 .0008
rs6436459 C .009 .0005
rs1530020 A .02 .05
rs6436454 T .01 .02
rs3948261 T .02 .05
rs2118409 C .04 NS
rs3795877 A .009 .002
rs6747096 A .02 .001
rs3795879 C NS .007
rs6715768 G .0004 .0007
rs7605945 C .004 .0004
rs975278 T NS .00006
rs729631 C NS .001
ss49785625   A .004 NS
rs7597833 T .007 NS
rs6734100 G .01 .00004
Open in a new tab

Note.— Exact P values are presented unless P>.05; each model was analyzed under the assumption of an additive mode of inheritance and adjusted for age, age2, height, height2, pack-years, pack-years2, sex, and a SNP–by–pack-years interaction term. NS = not significant.

Table 3.

PBAT of Qualitative Spirometric Phenotypes with SERPINE2 SNPs, Including a SNP-by-Smoking Interaction Term[Note]

P by COPD Severity
SNP Allele Moderate to Severe Mild to Severe
rs840088 C .0004 NS
rs6436459 C .0003 .002
rs1530020 A .02 .01
rs975278 T .03 NS
ss49785625 A .02 .01
rs7597833 T .03 .007
rs6734100 G .003 NS
rs1025734 T NS .03
Open in a new tab

Note.— Exact P values are presented unless P>.05; each model was analyzed under the assumption of an additive mode of inheritance and adjusted for age, age2, height, height2, pack-years, pack-years2, sex, and a SNP–by–pack-years interaction term.

Case-Control Association Testing

We attempted to replicate the family-based associations for SERPINE2 in a case-control analysis using 304 NETT individuals with severe smoking-related COPD and 441 NAS controls, all smokers, with normal spirometry (Celedón et al. 2004). As reported elsewhere, there was no significant evidence of population stratification in these cases and controls (Celedón et al. 2004). All SERPINE2 SNPs were in Hardy-Weinberg equilibrium in the controls. In the case-control analysis (table 4), eight SNPs (rs1438831, rs7579646, rs840088, rs7562213, rs920251, rs3795877, rs6747096, and rs3795879) demonstrated significant association (P⩽.05), including five SNPs that were significant in the family-based analysis. In a males-only analysis, the association remained robust.

Table 4.

Replication of Significant Associations of SERPINE2 with COPD in the NETT Cases and NAS Controls[Note]

P for
SNP and Sample Allele and Frequency Genotype and Frequency Allele Test Genotype Test Males-OnlyGenotype Test
rs1438831 (5′ genomic/promoter): C
 T
 CC/CT/TT
 .098 .089 .044
 Cases .70 .30 .48/.41/.09
 Controls .65
 .35
 .41/.49/.10
rs7579646 (intron 1): A
 G
 AA/AG/GG
 .043 .088 .016
 Cases .17 .83 .03/.24/.73
 Controls .21
 .79
 .04/.35/.61
rs840088 (intron 1): C
 T
 CC/CT/TT
 .007 .009 .008
 Cases .70 .30 .51/.37/.11
 Controls .63
 .37
 .38/.50/.13
rs7562213 (intron 1): A
 G
 AA/AG/GG
 .020 .018 .008
 Cases .30 .70 .12/.36/.52
 Controls .36
 .64
 .11/.50/.39
rs920251 (intron 1): A
 G
 AA/AG/GG
 .015 .011 .007
 Cases .30 .69 .11/.39/.50
 Controls .36
 .64
 .12/.49/.39
rs3795877 (intron 2): A
 G
 AA/AG/GG
 .019 .053 .005
 Cases .82 .18 .72/.25/.04
 Controls .77
 .23
 .58/.37/.05
rs6747096 (exon 3): A
 G
 AA/AG/GG
 .002 .005 .0006
 Cases .83 .17 .73/.24/.04
 Controls .76
 .24
 .57/.38/.05
rs3795879 (exon 3/intron 3 boundary): C
 T
 CC/CT/TT
 .009 .014 .0007
 Cases .81 .19 .71/.24/.04
 Controls .75 .25 .55/.40/.05
Open in a new tab

Note.— Only SNPs with significant results (P⩽.05) in any test are shown. Significant results are shown in bold italics.

Haplotype Analysis

Using a sliding-window approach, we analyzed adjacent 6-, 4-, and 2-SNP haplotypes and observed global significance for the 6-, 4-, and 2-SNP haplotypes (fig. 4); this significance was confirmed with empirical P values determined through 1,000 simulations. We used these results to narrow our individual haplotype analyses to these regions. With focus on the most-significant 2-SNP haplotype-specific associations, significance levels for haplotype-specific P values for FEV1/FVC in the Boston Early-Onset COPD pedigrees were similar to associated haplotypes in the NETT-NAS case-control analysis (for 2-SNP haplotype rs1438831-rs7579646, alleles TA, family P=.003, case-control P=.02; for 2-SNP haplotype rs1866153-rs6747096, alleles CG, family P=.01, case-control P=.001; for SNP haplotype rs6747096-rs3795879, alleles GT, family P=.047, case-control P=.006).

Figure 4.

Figure 4

Open in a new tab

Simulated P values for sliding-window haplotype analysis of SNPs in the NETT cases and NAS controls. Graphs of the 6-, 4-, and 2-SNP sliding window–simulated P values (Y-axis) against the midpoint SNP of each sliding window (X-axis), demonstrate the most significant global simulated P value for 2-SNP sliding-window haplotypes in the regions including SNPs rs7579646-rs840088 (P=.02), rs840088-rs7562213 (P=.01), rs7562213-rs920251 (P=.01), rs920251-rs1371029 (P=.02), rs7581619-rs3795877 (P=.02), rs3795877-rs1866153 (P=.04), rs1866153-rs6747096 (P=.004), rs6747096-rs3795879 (P=.02), rs3795879-ss49785624 (P=.02), ss49785624-rs6715768 (P=.02), and rs6715768-rs6738983 (P=.02).

Conditional Linkage Models

Linkage models analyzed in SOLAR, including as covariates those SNPs significant in the single-SNP and haplotype analyses, revealed attenuation of LOD scores for linkage to postbronchodilator FEV1/FVC on chromosome 2q. Specifically, inclusion of SNP rs7579646 in the overall linkage model resulted in attenuation of the LOD score from 4.41 (222 cM) to 2.16 (220 cM); linkage analysis of the smokers-only model resulted in further attenuation of the LOD score to 1.70. In this region of the gene, attenuation of the LOD score was also observed for SNPs rs840088 (LOD attenuation to 2.25 in smokers-only model) and rs7562213 (LOD attenuation to 2.07 smokers-only model). Farther downstream, LOD attenuation was observed with the inclusion of rs3795877 (LOD attenuation to 2.22 in smokers-only model), rs6747096 (LOD attenuation to 3.82 in smokers-only model), and rs3795879 (LOD attenuation to 2.75 in smokers-only model).

Discussion

Collaborative approaches are crucial for identifying candidate genes and uncovering potential new pathways of complex-disease pathogenesis. The phenotypic expression of COPD is likely under the influence of multiple genes, cigarette smoking (the major environmental risk factor for COPD), and gene-by-smoking interactions. In our study, expression-array analyses were utilized for candidate-gene prioritization in a broad linkage region on chromosome 2q; the intersection of genomic expression-array and linkage analyses led to the identification of SERPINE2 as a potential COPD susceptibility gene. Given our current state of understanding of COPD pathophysiology, SERPINE2 is not an obvious COPD candidate gene.

The SERPINE2 gene (GenBank reference sequence NM_006216) consists of nine exons, encoding a 44-kDa cellular and extracellular matrix-associated serine protease inhibitor, mainly involved in coagulation and fibrinolysis (Baker et al. 1980, 1982). The major known role of SERPINE2 is as an inhibitor of thrombin, urokinase, and plasmin (Baker et al. 1980; Scott et al. 1985). Although SERPINE2 has been extensively investigated in the brain and has been observed to hinder neuron apoptosis and injury-mediated cell death (Houenou et al. 1995; Rossignol et al. 2004), there have been no studies investigating SERPINE2 in the human lung. SERPINE2 expression has been observed in the embryonic mouse lung within the conducting-airway epithelium (Mansuy et al. 1993); our observation of expression in postembryonic mouse and human lungs supported further investigation of the role of SERPINE2 in lung disease.

The strengths of our current approach include the identification of a plausible positional candidate gene for COPD by expression analysis and immunohistochemistry and by demonstration of significant association of multiple SNPs in SERPINE2 in extended pedigrees, replication in an independent case-control study, and significance at both the single-SNP and haplotype levels. Although multiple statistical testing may yield false-positive associations, independent replication suggests that these are likely valid associations. LOD score attenuation in overall and smoking-stratified linkage models is further support of SERPINE2 as a COPD candidate gene.

The data from the microarray experiments guided further investigation, hypothesis generation, and candidate-gene selection. One weakness of this approach is that expression changes in the lung are subject to variations in sampling that are unavoidable in a complex and heterogeneous tissue, and gene expression changes detected in end-stage tissue may not be restricted to those involved in COPD mechanisms (and may be secondary or compensatory responses). Although the human control samples may not have had completely normal lungs, the combination of the microarray data and immunohistochemistry supported our further investigation of SERPINE2. Weaknesses of our association studies include the potential heterogeneity of disease phenotypes between our different human populations. However, limiting the potential heterogeneity is the fact that both the Boston Early-Onset COPD probands and the NETT cases represent individuals with very severe COPD. The replicated association of multiple SNPs in SERPINE2 suggests that the next phase of investigation of SERPINE2 in the lung should include association analysis of a comprehensive set of SERPINE2 SNPs and assessment of potential functional variants, as well as the consideration of fibrinolytic pathways in the pathogenesis of COPD. We have not observed complete attenuation of our LOD scores to zero in conditional linkage models; this suggests that other variants in the SERPINE2 gene, or more likely variants in another COPD-susceptibility gene in the region, contribute to the linkage signal in this region on chromosome 2q.

A mechanism through which SERPINE2 may contribute to the development of COPD has yet to be identified. To date, the expression of SERPINE2 has been most extensively studied in the brain and reproductive system; as such, it is a surprising candidate for association with COPD. However, SERPINE2 does belong to the serpin family of proteins, as does alpha 1-antitrypsin, deficiency of which is the only known genetic cause of COPD. Our results suggest that overexpression of SERPINE2, rather than deficiency, is associated with COPD. Nonetheless, the aggregate results of family-based and case-control single-SNP and haplotype association analyses, as well as the conditional linkage results demonstrating LOD score attenuation, suggest the strongest significance of variants in or in linkage disequilibrium with intron 1 and exon 3 of the SERPINE2 gene.

Alterations in coagulation and fibrinolytic pathways have been associated with acute and chronic lung injury and airway hyperresponsiveness, but extensive investigation in the setting of COPD has yet to be performed (Chambers 2003; Idell 2003; Wagers et al. 2004). Although there has been occasional speculation in the past regarding the role of vascular thrombosis and hypercoaguability in the development of emphysema, this has not been widely investigated (Brantigan et al. 1966; Alessandri et al. 1994; Ashitani et al. 2002). SERPINE2 has been demonstrated to be an inhibitor of trypsinlike serine proteases (such as thrombin, trypsin, plasmin, and urokinase); SERPINE2 has not been demonstrated to inhibit neutrophil elastase or chymotrypsinlike proteases (Scott et al. 1985). In addition to proteolytic stress, imbalance in oxidant-antioxidant pathways are potentially important in the development of COPD. To date, we are not aware of any studies of oxidative regulation of SERPINE2 activity.

There exists a complex mechanism of cross talk between serine proteases (and their inhibitors) and metalloproteinases (and their inhibitors), particularly within the lung and in relation to the pathogenesis of emphysema, a major component of COPD. For instance, the serine protease neutrophil elastase is capable of degrading metalloproteinase inhibitors (TIMPs), whereas the metalloproteinase macrophage elastase (MMP-12) is capable of degrading the neutrophil elastase inhibitor alpha 1-antitrypsin (Shapiro et al. 2003). Plasmin, one of the potential substrates for SERPINE2, is also capable of contributing to metalloproteinase activation. Polymorphic variants in the SERPINE2 gene could contribute to the development of COPD through alterations in matrix metalloproteinase pathways. The physiological substrate for SERPINE2 within the lung and the impact of alterations in expression on the complex coordination of proteolytic activities and imbalances in the exposure to cigarette smoke remain to be defined.

In summary, through the intersection of murine and human microarray analyses with linkage information on chromosome 2q, we have demonstrated replicated association of SERPINE2 variants with COPD. SERPINE2 may have a role in defining susceptibility phenotypes of COPD by influencing inflammatory responses to environmental exposures such as cigarette smoke through gene-by-smoking interactions. SERPINE2 has been demonstrated to inhibit extracellular matrix destruction (Bergman et al. 1986), which suggests that polymorphisms in the SERPINE2 gene may influence alterations in repair of smoking-induced lung damage. More research is needed to identify critical variants in the gene as well as to define a functional role for SERPINE2 in COPD.

Acknowledgments

We thank the study participants and their families for their enthusiastic support. We also thank Dr. J. Drazen and Dr. L. Ginns for their role in helping to establish the Boston Early-Onset COPD cohort. The authors acknowledge Dr. J. Brody and Dr. B. Celli, who were instrumental in the human microarray study. We also thank Soumyaroop Bhattacharya and Jake Abernethy for assistance with the microarray data analyses. We also thank Dr. Lussier for his generous gift of the SERPINE2 antibody. This work has been funded by National Institutes of Health (NIH) grant K08 HL072918 (to D.L.D.); an American Lung Association Research Award (to D.L.D.); NIH grants HL071885 (to T.M.), HL61575, HL71393, and HL075478 (to E.K.S.), HL54853 (to S.D.S.), and HL67204 (to H.A.C.); and an American Lung Association Career Investigator Award (to E.K.S.). The NAS is supported by the Cooperative Studies Program/Epidemiology Research and Information Center (ERIC) of the U.S. Department of Veterans Affairs and is a component of the Massachusetts Veterans ERIC. The Human Microarray work was supported by a Doris Duke Charitable Foundation Clinical Scientist Development Award (to A.S.) and NIH grant HL71771 (to J. Brody). NETT was supported by National Heart, Lung, and Blood Institute contracts N01HR76101–N01HR76116, N01HR76118, and N01HR76119; the Centers for Medicare and Medicaid Services; and the Agency for Healthcare Research and Quality.

Web Resources

The accession number and URLs for data presented herein are as follows:

  1. Bioconductor, http://www.bioconductor.org/
  2. C.L. Web site, http://biosun1.harvard.edu/~clange/pbat.htm (for PBAT)
  3. GenBank, http://www.ncbi.nlm.nih.gov/Genbank/ (for SERPINE2 [accession number NM_006216])
  4. Lung Transcriptome, http://lungtranscriptome.bwh.harvard.edu/
  5. Online Mendelian Inheritance in Man (OMIM), http://www.ncbi.nlm.gov/Omim/ (for COPD) [PubMed]
  6. TM4, http://www.tigr.org/software/tm4/mev.html (for TIGR Multi-Array Viewer [MEV] software)
  7. UCSC Genome Bioinformatics, http://genome.ucsc.edu/ (for the June 2002 freeze)

References

  1. Alessandri C, Basili S, Violi F, Ferroni P, Gazzaniga PP, Cordova C, Chronic Obstructive Bronchitis and Haemostasis Group (1994) Hypercoagulability state in patients with chronic obstructive pulmonary disease. Thromb Haemost 72:343–346 [PubMed] [Google Scholar]
  2. Almasy L, Blangero J (1998) Multipoint quantitative-trait linkage analysis in general pedigrees. Am J Hum Genet 62:1198–1211 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Ashitani J, Mukae H, Arimura Y, Matsukura S (2002) Elevated plasma procoagulant and fibrinolytic markers in patients with chronic obstructive pulmonary disease. Intern Med 41:181–185 [DOI] [PubMed] [Google Scholar]
  4. Baker JB, Low DA, Eaton DL, Cunningham DD (1982) Thrombin-mediated mitogenesis: the role of secreted protease nexin. J Cell Physiol 112:291–297 10.1002/jcp.1041120220 [DOI] [PubMed] [Google Scholar]
  5. Baker JB, Low DA, Simmer RL, Cunningham DD (1980) Protease-nexin: a cellular component that links thrombin and plasminogen activator and mediates their binding to cells. Cell 21:37–45 10.1016/0092-8674(80)90112-9 [DOI] [PubMed] [Google Scholar]
  6. Bedard J, Brule S, Price CA, Silversides DW, Lussier JG (2003) Serine protease inhibitor-E2 (SERPINE2) is differentially expressed in granulosa cells of dominant follicle in cattle. Mol Reprod Dev 64:152–165 10.1002/mrd.10239 [DOI] [PubMed] [Google Scholar]
  7. Bell B, Rose CL, Damon A (1966) The Veterans Administration longitudinal study of healthy aging. Gerontologist 6:179–184 [DOI] [PubMed] [Google Scholar]
  8. Bergman BL, Scott RW, Bajpai A, Watts S, Baker JB (1986) Inhibition of tumor-cell-mediated extracellular matrix destruction by a fibroblast proteinase inhibitor, protease nexin I. Proc Natl Acad Sci USA 83:996–1000 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Brantigan OC, Kress MB, Goco RB (1966) Pulmonary emboli: a factor in the etiology and pathogenesis of pulmonary emphysema. Dis Chest 49:491–501 [PubMed] [Google Scholar]
  10. Burrows B, Knudson RJ, Cline MG, Lebowitz MD (1977) Quantitative relationships between cigarette smoking and ventilatory function. Am Rev Respir Dis 115:195–205 [DOI] [PubMed] [Google Scholar]
  11. Celedón JC, Lange C, Raby BA, Litonjua AA, Palmer LJ, DeMeo DL, Reilly JJ, Kwiatkowski DJ, Chapman HA, Laird N, Sylvia JS, Hernandez M, Speizer FE, Weiss ST, Silverman EK (2004) The transforming growth factor-β1 (TGFB1) gene is associated with chronic obstructive pulmonary disease (COPD). Hum Mol Genet 13:1649–1656 10.1093/hmg/ddh171 [DOI] [PubMed] [Google Scholar]
  12. Chambers RC (2003) Role of coagulation cascade proteases in lung repair and fibrosis. Eur Respir J Suppl 44:33S–35S 10.1183/09031936.03.00001003 [DOI] [PubMed] [Google Scholar]
  13. Clayton D, Jones H (1999) Transmission/disequilibrium tests for extended marker haplotypes. Am J Hum Genet 65:1161–1169 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Crapo RO, Morris AH, Gardner RM (1981) Reference spirometric values using techniques and equipment that meet ATS recommendations. Am Rev Respir Dis 123:659–664 [DOI] [PubMed] [Google Scholar]
  15. DeMeo DL, Celedón JC, Lange C, Reilly JJ, Chapman HA, Sylvia JS, Speizer FE, Weiss ST, Silverman EK (2004) Genome-wide linkage of forced mid-expiratory flow in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 170:1294–1301 10.1164/rccm.200404-524OC [DOI] [PubMed] [Google Scholar]
  16. Ferris BG (1978) Epidemiology Standardization Project (American Thoracic Society). Am Rev Respir Dis 118:1–120 [PubMed] [Google Scholar]
  17. Hankinson JL, Odencrantz JR, Fedan KB (1999) Spirometric reference values from a sample of the general U.S. population. Am J Respir Crit Care Med 159:179–187 [DOI] [PubMed] [Google Scholar]
  18. Hersh CP, DeMeo DL, Lange C, Litonjua AA, Reilly JJ, Kwiatkowski D, Laird N, Sylvia JS, Sparrow D, Speizer FE, Weiss ST, Silverman EK (2005) Attempted replication of reported chronic obstructive pulmonary disease candidate gene associations. Am J Respir Cell Mol Biol 33:71–78 10.1165/rcmb.2005-0073OC [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Houenou LJ, Turner PL, Li L, Oppenheim RW, Festoff BW (1995) A serine protease inhibitor, protease nexin I, rescues motoneurons from naturally occurring and axotomy-induced cell death. Proc Natl Acad Sci USA 92:895–899 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hsu KH, Jenkins DE, Hsi BP, Bourhofer E, Thompson V, Tanakawa N, Hsieh GS (1979) Ventilatory functions of normal children and young adults—Mexican-American, white, and black. I. Spirometry. J Pediatr 95:14–23 [DOI] [PubMed] [Google Scholar]
  21. Idell S (2003) Coagulation, fibrinolysis, and fibrin deposition in acute lung injury. Crit Care Med 31:S213–S220 10.1097/01.CCM.0000057846.21303.AB [DOI] [PubMed] [Google Scholar]
  22. Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, Speed TP (2003) Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4:249–264 10.1093/biostatistics/4.2.249 [DOI] [PubMed] [Google Scholar]
  23. Knudson RJ, Lebowitz MD, Holberg CJ, Burrows B (1983) Changes in the normal maximal expiratory flow-volume curve with growth and aging. Am Rev Respir Dis 127:725–734 [DOI] [PubMed] [Google Scholar]
  24. Lange C, DeMeo D, Silverman EK, Weiss ST, Laird NM (2004) PBAT: tools for family-based association studies. Am J Hum Genet 74:367–369 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lange C, Laird NM (2002) Power calculations for a general class of family-based association tests: dichotomous traits. Am J Hum Genet 71:575–584 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Laurell C-B, Eriksson S (1963) The electrophoretic α1-globulin pattern of serum in α1-antitrypsin deficiency. Scand J Clin Lab Invest 15:132–140 [Google Scholar]
  27. Malhotra A, Peiffer AP, Ryujin DT, Elsner T, Kanner RE, Leppert MF, Hasstedt SJ (2003) Further evidence for the role of genes on chromosome 2 and chromosome 5 in the inheritance of pulmonary function. Am J Respir Crit Care Med 168:556–561 10.1164/rccm.200303-410OC [DOI] [PubMed] [Google Scholar]
  28. Mansuy IM, van der Putten H, Schmid P, Meins M, Botteri FM, Monard D (1993) Variable and multiple expression of protease nexin-1 during mouse organogenesis and nervous system development. Development 119:1119–1134 [DOI] [PubMed] [Google Scholar]
  29. Mariani TJ, Budhraja V, Mecham BH, Gu CC, Watson MA, Sadovsky Y (2003) A variable fold change threshold determines significance for expression microarrays. FASEB J 17:321–323 [DOI] [PubMed] [Google Scholar]
  30. Mariani TJ, Reed JJ, Shapiro SD (2002) Expression profiling of the developing mouse lung: insights into the establishment of the extracellular matrix. Am J Respir Cell Mol Biol 26:541–548 [DOI] [PubMed] [Google Scholar]
  31. Mecham BH, Klus GT, Strovel J, Augustus M, Byrne D, Bozso P, Wetmore DZ, Mariani TJ, Kohane IS, Szallasi Z (2004) Sequence-matched probes produce increased cross-platform consistency and more reproducible biological results in microarray-based gene expression measurements. Nucleic Acids Res 32:e74 10.1093/nar/gnh071 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. NETT Research Group (1999) Rationale and design of the National Emphysema Treatment Trial (NETT): a prospective randomized trial of lung volume reduction surgery. J Thorac Cardiovasc Surg 118:518–528 10.1016/S0022-5223(99)70191-1 [DOI] [PubMed] [Google Scholar]
  33. Palmer LJ, Celedón JC, Chapman HA, Speizer FE, Weiss ST, Silverman EK (2003) Genome-wide linkage analysis of bronchodilator responsiveness and post-bronchodilator spirometric phenotypes in chronic obstructive pulmonary disease. Hum Mol Genet 12:1199–1210 10.1093/hmg/ddg125 [DOI] [PubMed] [Google Scholar]
  34. Rossignol P, Ho-Tin-Noe B, Vranckx R, Bouton MC, Meilhac O, Lijnen HR, Guillin MC, Michel JB, Angles-Cano E (2004) Protease nexin-1 inhibits plasminogen activation-induced apoptosis of adherent cells. J Biol Chem 279:10346–10356 10.1074/jbc.M310964200 [DOI] [PubMed] [Google Scholar]
  35. Schaid DJ, Rowland CM, Tines DE, Jacobson RM, Poland GA (2002) Score tests for association between traits and haplotypes when linkage phase is ambiguous. Am J Hum Genet 70:425–434 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Scott RW, Bergman BL, Bajpai A, Hersh RT, Rodriguez H, Jones BN, Barreda C, Watts S, Baker JB (1985) Protease nexin: properties and a modified purification procedure. J Biol Chem 260:7029–7034 [PubMed] [Google Scholar]
  37. Shapiro SD, Goldstein NM, Houghton AM, Kobayashi DK, Kelley D, Belaaouaj A (2003) Neutrophil elastase contributes to cigarette smoke-induced emphysema in mice. Am J Pathol 163:2329–2335 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Silverman EK, Chapman HA, Drazen JM, Weiss ST, Rosner B, Campbell EJ, O’Donnell WJ, Reilly JJ, Ginns L, Mentzer S, Wain J, Speizer FE (1998) Genetic epidemiology of severe, early-onset chronic obstructive pulmonary disease: risk to relatives for airflow obstruction and chronic bronchitis. Am J Respir Crit Care Med 157:1770–1778 [DOI] [PubMed] [Google Scholar]
  39. Silverman EK, Mosley JD, Palmer LJ, Barth M, Senter JM, Brown A, Drazen JM, Kwiatkowski DJ, Chapman HA, Campbell EJ, Province MA, Rao DC, Reilly JJ, Ginns LC, Speizer FE, Weiss ST (2002a) Genome-wide linkage analysis of severe, early-onset chronic obstructive pulmonary disease: airflow obstruction and chronic bronchitis phenotypes. Hum Mol Genet 11:623–632 10.1093/hmg/11.6.623 [DOI] [PubMed] [Google Scholar]
  40. Silverman EK, Palmer LJ, Mosley JD, Barth M, Senter JM, Brown A, Drazen JM, Kwiatkowski DJ, Chapman HA, Campbell EJ, Province MA, Rao DC, Reilly JJ, Ginns LC, Speizer FE, Weiss ST (2002b) Genomewide linkage analysis of quantitative spirometric phenotypes in severe early-onset chronic obstructive pulmonary disease. Am J Hum Genet 70:1229–1239 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Spira A, Beane J, Pinto-Plata V, Kadar A, Liu G, Shah V, Celli B, Brody JS (2004) Gene expression profiling of human lung tissue from smokers with severe emphysema. Am J Respir Cell Mol Biol 31:601–610 10.1165/rcmb.2004-0273OC [DOI] [PubMed] [Google Scholar]
  42. Tusher VG, Tibshirani R, Chu G (2001) Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 98:5116–5121 10.1073/pnas.091062498 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Wagers SS, Norton RJ, Rinaldi LM, Bates JH, Sobel BE, Irvin CG (2004) Extravascular fibrin, plasminogen activator, plasminogen activator inhibitors, and airway hyperresponsiveness. J Clin Invest 114:104–111 10.1172/JCI200419569 [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from American Journal of Human Genetics are provided here courtesy of American Society of Human Genetics

RESOURCES