Author + information
- Received November 1, 2002
- Revision received February 7, 2003
- Accepted March 20, 2003
- Published online June 18, 2003.
- Seyyare Beyzade, BSc*,
- Shaoli Zhang, PhD*,
- Yuk-ki Wong, MD, MRCP†,
- Ian N.M Day, MB, PhD, FRCPath*,
- Per Eriksson, PhD‡ and
- Shu Ye, MD, PhD*,* ()
- ↵*Reprint requests and correspondence:
Dr. Shu Ye, Human Genetics Division, Southampton University Medical School, Duthie Building (808), Southampton General Hospital, Southampton SO16 6YD, United Kingdom.
Objectives The aim of this study was to assess matrix metallloproteinase-3 (MMP3) gene variation in relation to the degree of coronary atherosclerosis and risk of myocardial infarction (MI) in patients with coronary artery disease.
Methods In this study, we systematically screened the promoter and coding regions for sequence variants. All polymorphisms identified were analyzed in 1,240 individuals undergoing coronary angiography. Functional analyses of the polymorphisms were carried out with the use of report assays and electrophoretic mobility shift assays.
Results Six novel polymorphisms were identified. The 6A/6A genotype was associated with greater number of coronary arteries with significant stenosis (odds ratio [OR] 1.52, p = 0.008), whereas the 5A/5A and 5A/6A genotypes were associated with increased risk of MI (OR 2.02 and 1.78, p = 0.016 and 0.032, respectively). A stepwise logistic regression analysis with all polymorphisms taken into account showed that the effect of MI susceptibility was largely attributed to the 5A/6A polymorphism. In a stepwise logistic regression analysis with all haplotypes as independent variables, the most common haplotype (T-5A-A-A-G-A), and two rare haplotypes, all containing the 5A allele, were associated with MI susceptibility. Functional studies showed that the T-5A-A-A-G-A haplotype had a higher promoter activity in macrophages.
Conclusions These data indicate that the effect of MMP3 gene variation is attributable to the 5A/6A polymorphism and that individuals carrying the 6A/6A genotype may be predisposed to developing atherosclerotic plaques with significant stenosis, whereas those carrying the 5A allele may be predisposed to developing unstable plaques.
Matrix proteins (collagen, proteoglycan, and elastin), smooth muscle cells, macrophages, and lipids are the main constituents of atheromas (1). However, the relative proportions of these components vary, giving rise to two types of lesion referred to as fibrotic plaque and lipid-rich plaque, respectively (2,3). The former is rich in matrix proteins and smooth muscle cells, whereas the latter is rich in lipids and macrophages. Fibrotic plaques are associated with high-grade arterial stenosis, whereas lipid-rich plaques are less stenosing but prone to rupture (2,3). Coronary atherosclerotic plaque rupture is the most common cause of myocardial infarction (MI) (4). Pathologic studies have revealed a significant patient-to-patient variability in plaque composition (3). For example, an autopsy study of 54 subjects with coronary artery disease (CAD) showed that in 15% of the subjects, all the plaques present were predominantly fibrous, whereas in another 13% of patients, all plaques were of the lipid type (5).
The amount of matrix proteins in atheromas is determined by the function of matrix protein synthesis over degradation, and the latter is catalyzed by matrix metalloproteinases (MMPs) (2,6). It has been shown that lipid-rich plaques express higher levels of MMPs than fibrotic plaques (7). Among the MMPs that are expressed in atheromas is MMP3 (also known as stromelysin), which has a broad substrate specificity and can activate other enzymes in the MMP family (8,9). Inactivating the MMP3 gene in apolipoprotein E (apoE) knockout mice increases the sizes of atherosclerotic plaques and the amounts of lesional matrix proteins (10). In humans, a naturally occurring sequence variant in the MMP3 gene promoter has been identified. This sequence variant arises from the insertion of an adenine nucleotide at position −1612 relative to the start of transcription, resulting in one allele having a run of five adenine nucleotides (5A) and the other having six adenine nucleotides (6A) (11). Our previous work indicated that the 5A allelic promoter has a higher transcriptional activity than the 6A allelic promoter (11). Several studies have shown that the 6A allele is associated with more rapid progression of coronary atherosclerosis (12–14)and increased carotid intima-media thickness (15–17), suggesting that matrix accumulation is enhanced in individuals carrying this transcriptionally less active allele of the MMP3 gene. In addition, a Japanese study showed that the transcriptionally more active 5A allele was over-represented in a group of patients with acute MI compared with healthy control subjects (18).
In this study, we investigated whether the 5A/6A polymorphism and/or other sequence variants in the MMP3 gene influenced the extent of atherosclerosis and risk of MI in patients with CAD. We scanned the promoter and coding regions of the MMP3 gene for sequence variants and analyzed the variants in a sample of 1,240 Caucasian individuals undergoing coronary angiography. We then performed in vitro experiments to assess the functional significance of the genetic variants.
We recruited 1,240 consecutive Caucasian patients undergoing interventional or diagnostic coronary angiography in the Southampton General Hospital. Among the 1,240 subjects, 943 had significant stenosis (>50%) in at least one major epicardial coronary artery, and the remaining 297 had no significant stenosis (<50%). Coronary angiograms were assessed by one consultant cardiologist. We also recorded demographic and clinical data, including age, gender, weight, height, occupation, smoking habit, hyperlipidemia, hypertension, diabetes mellitus, previous MI, and coronary heart disease in first-degree relatives. The main characteristics of the subjects are summarized in Table 1. The group with MI had a higher percentage of smokers than did the group without MI (p = 0.016). There was no significant difference between patients with and those without MI in terms of age, gender, and prevalence of hypercholesterolemia, hypertension, diabetes, and family history of CAD. Patient recruitment was approved by the local Ethics Committee, and all subjects gave written, informed consent.
Sequence variant scanning
The bi-directional dideoxy fingerprinting (bi-ddF) method (19)was utilized to search for sequence variants in the MMP3 gene, including the promoter (∼2.3 kb), all 10 exons, and the intron–exon junctions. The nature and location of sequence variants identified by bi-ddF were determined by DNA sequencing. The assays were performed on genomic DNA samples from 20 unrelated Caucasian individuals (10 with early-onset MI [<45 years] and 10 healthy subjects). This sample size (40 chromosomes) provided ∼90% power to detect polymorphisms with a minor-allele frequency of >5% (20), on the basis that genetic variants contributing to a common disease in a large proportion of patients are likely to be common in the population, and thus the principal target of this work was common sequence variants.
Determination of genotypes
The subjects previously described were genotyped for the −1986 T>C, −1612 5A>6A, −1346 A>C, −709 A>C, −376 G>C, and +802 A>G polymorphisms. For each polymorphism, a sequence containing the polymorphic site was amplified by the polymerase chain reaction (PCR), and the amplicon was digested with an appropriate restriction endonuclease, which specifically cleaved one of the two alleles. Because of the limited amounts of DNA samples available to us, some of the polymorphisms could not be genotyped for some of the subjects.
The MMP3 gene promoter region (from −2309 bp to +54 bp relative to the transcriptional start site) was amplified by PCR. The amplicon was inserted into a promoterless vector (pGL3-Basic Vector, Promega, Southampton, United Kingdom) containing a firefly luciferase reporter gene. The resultant construct was mixed with a plasmid (pRL-TK, Promega) containing a renillaluciferase gene under the control of a thymidine kinase promoter and transfected into cultured macrophages by electroporation. The transfectants were cultured with or without 1 μmol/l phorbol 12-myristate 13-acetate for 24 h. The cells were then lysed, and the activities of the firefly luciferase and renillaluciferase in the lysates were measured with the use of a Dual-Luciferase assay kit (Promega). The ratio of firefly luciferase level to renillaluciferase level was used as a measurement of the MMP3 gene promoter activity. At least four independent experiments in duplicates were carried out for each construct, and the mean values ± SEM are presented.
Electrophoretic mobility shift assays
With the use of a method by Alksnis et al. (21), nuclear protein extracts were prepared from cultured macrophages differentiated from human monocytic U937 cells with 1 μmol/l phorbol 12-myristate 13-acetate and from cultured human monocytoid MonoMac-6 cells. For each polymorphism, two double-stranded 26-mer oligonucleotides corresponding to the two alleles were used as probes and labeled with [γ32-P]-adenosine triphosphate. The labeled probes were incubated with the aforementioned nuclear protein extracts, followed by nondenaturing polyacrylamide gel electrophoresis and autoradiography. Three independent experiments were carried out for each polymorphism.
The HWE program was used to test whether the observed genotype distributions deviated from the Hardy-Weinberg equilibrium. Linkage disequilibrium between the polymorphisms and the association metric D′ were analyzed with the use of the ASSOCIATE program and according to Devlin and Risch (22). Haplotype frequencies (Table 2) were estimated using the Haplotyper program (23), which employs a bayesian algorithm. To assess the effects of the genetic variants, the MMP3 gene polymorphisms were first examined individually in relation to the number of coronary arteries with >50% stenosis by ordinal logistic regression analysis (Table 3) and in relation to the risk of MI by binary logistic regression analyses (Table 4). Subsequently, stepwise logistic regression analysis with all genotypes and haplotypes, or with all haplotypes but not genotypes, inputted as independent variables, was performed to investigate which polymorphism(s) and haplotype(s) accounted for the association with MI susceptibility (Table 5).
Identification of novel variants in the MMP3 gene
Sequence variant scanning in the promoter region, coding region, and intron–exon junctions of the MMP3 gene identified a total of six novel sequence variants, in addition to the previously reported 5A/6A polymorphism. All newly identified variants were single nucleotide polymorphisms: four located in the promoter region and the remainder in the coding region (Fig. 1). The −1986C, −16126A, −1346C, −709G, −376C, +802G, and +814C alleles were detected in six of the 10 healthy individuals and five of the 10 MI patients subjected to the variant scanning. The frequencies of haplotypes in a group of individuals without significant coronary atherosclerosis are shown in Table 2. There was a substantial linkage disequilibrium between the polymorphisms, with the linkage disequilibrium metric D′ being over 0.9 for all polymorphisms, except for the −709 A>G polymorphism, which had a lower frequency of the minor allele (0.2 vs. >0.4 for the minor allele of the other polymorphisms).
Association of MMP3 gene variation with extent of coronary atherosclerosis
Based on the finding from previous studies that progression of coronary atherosclerosis was more rapid in individuals with the MMP3 gene 6A/6A genotype (12–14), we hypothesized that variation in this gene could influence the extent of coronary atherosclerosis in CAD patients. To test this hypothesis, the polymorphisms in the MMP3 gene were analyzed in a cohort of Caucasian individuals undergoing coronary angiography. In those without a history of MI, the number of coronary arteries with >50% stenosis increased with increasing frequency of the 6A/6A genotype (odds ratio [OR] 1.52, p = 0.008) (Table 3). This association remained significant after adjustment for classic risk factors. Analyses in male and female subjects separately showed a consistent trend. No statistically significant association was detected between the other polymorphisms in the MMP3 gene and the extent of coronary atherosclerosis. There was no significant association between MMP3 genotype and the extent of coronary atherosclerosis in patients with MI.
Association of MMP3 gene variation with risk of MI in CAD patients
We then tested the hypothesis that variation in the MMP3 gene could influence the risk of MI in CAD patients. Logistic regression analyses showed that in patients with >50% stenosis in at least one coronary artery, there was an association between the −1612 5A/6A polymorphism and risk of MI, with the 5A/5A genotype conferring a twofold increase in MI risk (p = 0.016) (Table 4). Individuals who were heterozygous for the 5A/6A polymorphism had an intermediate risk (OR 1.78, p = 0.032). The association between this polymorphism and MI risk remained significant after adjustment for classic risk factors. A similar trend was observed when male and female subjects were analyzed separately.
In addition, there were approaching statistically significant differences in MI risk between the genotypes for the −1986 T>C polymorphism, such that individuals with the T/T genotype had 1.67-fold higher risk of MI compared with those with the C/C genotype (p = 0.089 after adjustment for covariates), whereas heterozygous individuals had an intermediate risk (OR 1.57, p = 0.096). There was no statistically significant association between MI risk and the other polymorphisms in the MMP3 gene (Table 2).
Stepwise logistic regression analysis
To estimate which variants and haplotypes largely accounted for the association between the MMP3 gene and MI susceptibility, we performed a stepwise logistic regression analysis with all genotypes (based on individual polymorphisms) and haplotypes (based on combinations of the alleles of the polymorphisms) inputted as independent variables. In this analysis, only the 5A/6A polymorphism was significantly associated with MI risk and remained in the equation (Table 5). We then carried out a stepwise logistic regression with all haplotypes, but not genotypes, inputted as independent variables. In this analysis without genotype terms, three haplotypes were associated with MI risk and remained in the equation (Table 5). These haplotypes—T-5A-A-A-G-A, T-5A-A-A-C-G, and C-5A-C-G-C-G—contained the 5A allele at the −1612 site but either the major or minor allele at the other polymorphic sites.
Differences in transcriptional activity between haplotypes
Our previous work showed that the 5A allele had a higher promoter activity than the 6A allele (11). In this study, we carried out transient transfection experiments and reporter assays to investigate whether the difference in promoter activity remained when the other polymorphisms were taken into account. In these experiments, the T-5A-A-A-G-A haplotype (which was the most common haplotype and was associated with increased MI risk) consistently showed a higher promoter activity than the C-6A-C-G-C-G haplotype. This difference was detected consistently in two different macrophage cell lines (RAW264.7 and MALU) and became more pronounced when these cells were stimulated with phorbol 12-myristate 13-acetate, a reagent that induces macrophage differentiation (Table 6).
Allele-specific binding of nuclear protein at polymorphic sites of the MMP3 promoter gene
We previously showed that the sequence surrounding the −1612 5A>6A polymorphic site was a transcription factor binding site and that compared with the 5A allele, the 6A allele had a higher affinity with the transcription factor (11). In the present study, we carried out electrophoretic mobility shift assays to investigate whether the other polymorphisms in the MMP3 gene promoter also showed an allele-specific interaction with transcription factors. These assays showed binding of a nuclear protein with a higher affinity to the −1986 T allele than to the −1986 C allele (Fig. 2). No differential binding of transcription factor was detected for the other promoter polymorphisms (−1346 A>C, −709 A>G, or −376 G>C; data not shown).
Analysis of variants in the coding region
Both coding region polymorphisms identified are located in exon 2 of the MMP3 gene. Whereas the +814 A>G polymorphism is a synonymous substitution, the +659 A>G polymorphism results in a change from glutamic acid (Glu) to lysine (Lys) at residue 45, which is located just after the first alpha helix in the propeptide. The three-dimensional structure of the propeptide indicates that the region containing residue 45 is flexible. Thus, the change from glutamic acid (an acidic amino acid) to lysine (a basic amino acid) at this residue could potentially alter its interaction with other amino acids in this region and could have an effect on MMP3 activation. However, clustal alignment revealed that though most residues in this region were conserved across species, the Glu45 residue was not. The amino acid sequences in this region are YDL44E45KDVKQF in humans, YNL44E45KDVKQF in rabbits, YGL44A45KDVKQF in mice, and YGL44E45KDVKQF in rats. This suggests that the Lys45Glu polymorphism is unlikely to have a significant functional impact. We are currently seeking to address this question with in vitro studies.
In this study, we showed that CAD patients carrying the MMP3 gene 5A allele were at a higher risk of MI compared with CAD patients not carrying the 5A allele. A previous Japanese study showed that the 5A/5A and 5A/6A genotypes were over-represented in patients with acute MI compared with healthy control subjects (18). Because the comparison in that Japanese study was between MI patients and healthy control subjects, the difference in MMP3 genotype frequency between the two groups might arise from an association between the MMP3 gene and susceptibility of CAD or from an impact of the MMP3 gene on stability of coronary atherosclerotic plaques. The results of the present study provide an answer to this question. We found that among CAD patients, the 5A/5A and 5A/6A genotypes were over-represented in those with a history of MI compared with those without such a history, indicating an influence of these genotypes on atherosclerotic plaque stability.
In this study, we also showed that individuals with the 6A/6A genotype had a greater extent of coronary atherosclerosis, compared with individuals with other genotypes. This finding is consistent with the finding from several previous studies that the 6A/6A genotype is associated with more rapid progression of coronary atherosclerosis (12–14)and with greater intima-media thickness (15–17).
These findings support the notion that matrix accumulation in the arterial wall is enhanced in individuals carrying the transcriptionally less active 6A allele of the MMP3 gene, whereas matrix degradation and atherosclerotic plaque instability are increased in individuals carrying the transcriptionally more active 5A allele. It is likely that individuals carrying the 6A allele are predisposed to the development of fibrotic plaques, which characteristically cause higher grade stenosis, whereas those carrying the 5A allele are susceptible to the development of lipid-rich plaques, which are typically more prone to rupture, causing MI (3). For both the extent of coronary atherosclerosis and risk of MI, individuals who are heterozygous for the 5A/6A polymorphism have phenotypes similar to those carrying the 5A/5A genotype, suggesting that the effect of the 5A allele is dominant and that of the 6A allele is recessive.
It has been shown that atherosclerotic lesions are significantly smaller and contain significantly less collagen in apoE−/−/MMP3+/+mice than in apoE−/−/MMP3−/−mice (10). In addition, atherosclerotic lesions in apoE−/−/MMP3+/+mice have a higher content of lipids and macrophages than those lesions in apoE−/−/MMP3−/−mice (10). These findings suggest that atherosclerotic plaques in mice with an active MMP3 gene are the lipid-rich type, whereas those in mice lacking the active MMP3 gene are more fibrotic, which is consistent with the findings of the present study of variation in the MMP3 gene in humans.
Among the seven polymorphisms in the MMP3 gene, only the 5A/6A polymorphism was found to have significant effects on the extent of coronary atherosclerosis and MI risk, although the −1986 T>C polymorphism might also exert a moderate influence. We have previously shown that the 5A allelic promoter had a higher transcriptional activity than the 6A allelic promoter (11). The present study showed that this allelic difference in promoter activity remained when the newly identified polymorphisms were taken into account. The functional analyses of the 5A/6A polymorphism in the previous study were carried out in vascular smooth muscle cells and fibroblasts (11). Because the majority of MMP3 in atherosclerotic plaques is produced by macrophages (8,9), this cell type was chosen for the functional analyses in the present study. The results showed that the effect of the MMP3 gene polymorphism on promoter activity also exists in macrophages.
Previous work has shown that the a 26-bp sequence containing the 5A/6A polymorphic site interacted with a transcription factor that had a higher affinity with the 6A allele than the 5A allele (11,24). In the present study, we found that differential nuclear protein binding also occurred on the sequence containing the −1986 T>C polymorphic site, with a nuclear protein interacting more readily with the T allele. Because the two polymorphisms are in almost complete linkage disequilibrium, the differential binding of the nuclear protein to the −1986 T>C polymorphism might also contribute to the difference in promoter activity between the T-5A-A-A-G-A and C-6A-C-G-C-G haplotypes.
The data in this study indicate that variation in the MMP3 gene influences the extent of coronary atherosclerosis and risk of MI in patients with CAD. These influences are largely attributable to the 5A/6A polymorphism, with the 6A/6A genotype being associated with the extent of atherosclerosis and the 5A allele-containing genotypes being associated with the risk of MI. These findings suggest that the 5A/6A polymorphism may contribute to the patient-to-patient variability in atherosclerotic plaque composition (3,5).
We thank Professor Hideaki Nagase for valuable comments on the coding region polymorphisms.
☆ This work was supported by the British Heart Foundation, London (grants PG/98183 and PG/98192), the University of Southampton School of Medicine (PhD studentship to S.B.), the Swedish Medical Research Council, Stockholm (12660), and the King Gustaf V and Queen Victorias Foundation, Stockholm.
- coronary artery disease
- bi-directional dideoxy fingerprinting
- myocardial infarction
- matrix metalloproteinase
- polymerase chain reaction
- Received November 1, 2002.
- Revision received February 7, 2003.
- Accepted March 20, 2003.
- American College of Cardiology Foundation
- Libby P.
- Davies M.J.
- Davies M.J.,
- Thomas A.C.
- Hangartner J.R.,
- Charleston A.J.,
- Davies M.J.,
- Thomas A.C.
- Sukhova G.K.,
- Schonbeck U.,
- Rabkin E.,
- et al.
- Henney A.M.,
- Wakeley P.R.,
- Davies M.J.,
- et al.
- Silence J.,
- Lupu F.,
- Collen D.,
- Lijnen H.R.
- Ye S.,
- Eriksson P.,
- Hamsten A.,
- Kurkinen M.,
- Humphries S.E.,
- Henney A.M.
- Ye S.,
- Watts G.F.,
- Mandalia S.,
- Humphries S.E.,
- Henney A.M.
- Humphries S.E.,
- Luong L.A.,
- Talmud P.J.,
- et al.
- Gnasso A.,
- Motti C.,
- Irace C.,
- et al.
- Rauramaa R.,
- Vaisanen S.B.,
- Luong L.A.,
- et al.
- Rundek T.,
- Elkind M.S.,
- Pittman J.,
- et al.
- Terashima M.,
- Akita H.,
- Kanazawa K.,
- et al.
- Liu Q.,
- Feng J.,
- Sommer S.S.
- Alksnis M.,
- Barkhem T.,
- Stromstedt P.E.,
- et al.