Matrix Metalloproteinase-2 Polymorphisms and Breast Cancer Susceptibility

As the enzyme capable of cleaving type 4 collagen, the major structural component of the epithelial basement membrane, matrix metalloproteinase-2 (MMP-2, gelatinase A), is well-known to be integral for cancer cell invasion and metastasis (1-3). With many additional extracellular matrix (ECM) and non-ECM substrates, MMP-2 is also involved in a variety of other, potentially pathologic processes, including inflammation, angiogenesis, and cellular proliferation (1-5). Oncogene-mediated cellular transformation was found to induce MMP-2 expression, causing altered cell growth and increased capacity for malignant progression (6, 7). In vitro assays of breast cancer cells stably transfected with MMP-2 showed increased invasive properties, whereas accelerated tumor growth, enhanced metastatic colonization, and increased tumor burden was seen after the transfected cells were injected into mice (8, 9). On the other hand, MMP-2 deficient mice were found to have reduced tumor-induced angiogenesis, significantly slower tumor growth rates, and decreased metastatic colonization of the lung after implantation of either melanoma or lung carcinoma cells (10).

In humans, normal breast tissue and benign breast lesions were rarely found to express MMP-2, whereas expression has been detected in both tumor and surrounding stromal cells (11-15). Furthermore, compared with adjacent breast tissue, a gradual increase in MMP-2 expression was seen from noninvasive to invasive cancers, whereas MMP-2 activity has also been found to be significantly higher in malignant breast tissue compared with other breast tissues (16, 17). In addition, breast cancer patients were found to have significantly higher circulating MMP-2 levels compared with control volunteers (18).

Genetic variation that modulates MMP-2 expression may contribute to individual differences in cancer susceptibility. Two single nucleotide polymorphisms (SNP) in the MMP-2 promoter have been shown to affect expression in vitro; C to T transitions at −1306 (rs243865) and −735 (rs2285053) both result in lower transcriptional activities (19-21). These two SNPs are reported to be in high linkage disequilibrium (LD) and have an interactive effect on MMP-2 transcription (21). Several epidemiologic studies have evaluated these promoter polymorphisms in relation to cancer risk with inconsistent results. To date, only a few studies have evaluated genetic variation in MMP-2 in relation to breast cancer susceptibility, and all included only one polymorphism (rs243865; refs. 22-25). Preliminary results from a small study (89 cases and 100 controls) among Latin American women indicated that there was no association (22), whereas a small study among Mexican women (90 cases and 96 controls) found a significantly increased risk of breast cancer associated with −1306 CC (23). A larger study (462 cases and 509 controls) among Chinese women found a significantly decreased risk of breast cancer for T allele carriers (24), whereas the largest study to date (959 cases and 952 controls), conducted among Swedish women, found no association (25).

As MMP-2 has been shown to contribute not only to cancer invasion and metastasis, but also to cellular transformation and tumor growth, this study was undertaken to comprehensively characterize genetic variation across the MMP-2 gene and evaluate associations of MMP-2 polymorphisms and breast cancer susceptibility.

Study subjects were participants of the Shanghai Breast Cancer Study (SBCS), a population-based case-control study among Chinese women; detailed information on the study design and data collection procedures have been previously described (26). Briefly, phase 1 cases were women diagnosed with breast cancer between August 1996 and March 1998, 25 to 64 y of age, without a previous cancer diagnosis, and alive at the time of interview. Recruitment for phase 2 occurred between April 2002 and February 2005, and eligibility criteria were expanded to include women 20 to 70 y of age (27, 28). All cases were identified via the population-based Shanghai Cancer Registry; diagnoses were confirmed by two senior pathologists. Controls were randomly selected from the general population using the Shanghai Resident Registry, a population registry of adult residents in urban Shanghai; women with previous cancer diagnoses were excluded. Structured questionnaires were administered by trained interviewers, and were used to obtain detailed information on demographic, reproductive, and behavioral factors; height and weight were also measured. Of eligible participants, 1,459 (91.1%) cases and 1,556 (90.3%) controls in phase 1 and 1,989 cases (83.7%) and 1,989 controls (70.4%) in phase 2, completed in-person interviews. In phase 1, 1,193 cases (81.8%) and 1,310 controls (84.2%) donated blood samples. In phase 2, 1,932 (97.1%) cases and 1,857 (93.4%) controls donated either blood or buccal cell samples. Genomic DNA was extracted using Puregene's DNA Purification kits (Gentra Systems) or Qiagen's DNA Purification kits (Qiagen) according to manufacturers' instructions. Laboratory staff was blinded to the case-control status of these subjects for all subsequent genotyping described.

Haplotype-tagging SNPs (htSNPs) were selected by searching Han Chinese data from the HapMap Project (29) using the Tagger program (30). htSNPs were selected to cover polymorphisms with minimum minor allele frequency (MAF) of 0.05 in the MMP-2 gene ± 5 kb with an r² of 0.90 or greater. Selection of htSNPs was completed in December 2005 using HapMap Release 19. Genotyping assays for 19 htSNPs were completed for 2,131 phase 1 participants in 2006 using a Targeted Genotyping System (Affymetrix) based on an advanced Molecular Inversion Probe method (31). Blinded (n = 39) and HapMap samples (n = 12) were also included; consistency rates averaged 99.6%.

Two SNPs with promising results in phase 1 (rs1116195 and rs243865) were selected for additional genotyping among phase 2 participants. Furthermore, a polymorphism reported to be functional that was not genotyped in phase 1 (rs2285053) was also selected. These 3 SNPs were genotyped among 2,932 phase 2 participants using the Sequenom iPLEX MassARRAY platform (Sequenom, Inc.). PCR and extension primers were designed using Sequenom Assay Design software. PCR and extension reactions were done according to manufacturer's instructions as previously described (32). Allele-specific extension products were determined by using matrix assisted laser desorption/ionization time-of-flight mass spectrometry. Blinded duplicate samples and negative controls were included in each 96-well plate; concordance rates between duplicate samples were 100% for all three polymorphisms genotyped by this method.

Recently, we completed genotyping for 4,157 cases and controls (2,213 phase 1 and 1,944 phase 2 participants) using the Affymetrix Genome-Wide Human SNP Array 6.0 (Affymetrix), that includes 906,602 SNPs. From this, data for 19 additional SNPs that are located in MMP-2 ± 10 kb were included in the current study to increase the density of genetic coverage and the statistical power of our analysis.

Hardy-Weinberg equilibrium was tested by comparing the observed and expected genotype frequencies of the controls (χ² test). Characteristics of cases and controls were compared with the χ² test or t test for categorical or continuous variables, respectively. Covariates considered included age at diagnosis, age at menarche, age at first live birth among parous women, age at menopause among postmenopausal women, history of breast fibroadenomas, body mass index, and leisure physical activity in the decade preceding diagnosis. Associations with breast cancer risk were evaluated by computing odds ratios (OR) and corresponding 95% confidence intervals (95% CI) by logistic regression. Additive models of effect were applied to all SNPs; tests for trend were conducted by coding for the number of variant alleles and reporting the P value for the β coefficient. Dominant or recessive effect models and P values were also calculated when suggested to be appropriate for particular SNPs. LD between polymorphisms was assessed by Haploview (33). Associations between haplotypes and breast cancer risk were analyzed with HAPSTAT software (34); additive, dominant, and recessive models of effect were also evaluated. Study phase was adjusted for in analyses that included participants from both phases. All statistical tests were two tailed, and P values of ≤0.05 were interpreted as statistically significant.

A total of 6,066 Chinese women were included in the present study: 2,291 phase 1 participants and 3,775 phase 2 participants. Women in phase 2 were slightly older, and tended to participate in more regular physical activity than phase 1 women, but were generally comparable. As expected, breast cancer cases were found to differ from controls with regard to demographic, reproductive, and other known breast cancer risk factors (Table 1). Cases were more likely to have an earlier age at menarche, older age at first live birth, a history of breast fibroadenomas, higher body mass index and waist-to-hip ratio, and less likely to participate in regular physical activity than controls.

A total of 39 MMP-2 SNPs were genotyped as follows: 19 htSNPs among phase 1 participants, 1 SNP selected from the literature among phase 2 participants, and 19 additional SNPs among participants of both study phases. Of the MMP-2 SNPs genotyped, 3 had MAFs of <1% in our study population, and were excluded from further analyses (rs16955194, rs7189232, and rs11541998). The remaining SNPs were found to have MAFs between 10.7 and 48.9% among genotyped controls; none were found to deviate from Hardy-Weinberg equilibrium. MAFs among SBCS controls were generally comparable with the MAF found among Han Chinese genotyped in HapMap, with 2 exceptions: SBCS controls had more A alleles (44.6%) for rs1005912 than HapMap (38%), and fewer G alleles (40.4%) for rs243867 than HapMap (46%), although neither of these differences was statistically significant. Associations with breast cancer susceptibility were initially assessed using additive models of effect. When the same SNP was genotyped by two methods, the data source with the higher number of genotyped participants was used. One SNP (rs243865) was genotyped by both Sequenom and Affymetrix 6.0 methods for 1,098 samples; the concordance rate between these methods was 100%. Seven SNPs (rs243865, rs1477017, rs865094, rs1053605, rs243847, rs243839, and rs11639960) were genotyped by both Affymetrix Targeted Genotyping and the Affymetrix 6.0 Genome Wide platform for ∼2,000 participants; concordance rates for these samples ranged between 98.96% and 99.90%, and averaged 99.65%.

SNP information, genotyping method, study population genotyped, and associations with breast cancer risk for the 36 polymorphic MMP-2 SNPs are detailed in Table 2. All available genotyped participants are included in the analyses; when study phases 1 and 2 are listed, the estimate of effect is pooled. Two promoter SNPs were found to confer significant reductions in breast cancer risk among rare allele homozygotes. rs11644561 AA participants were ∼40% less likely to be breast cancer cases than those with GG, whereas rs11643630 AA participants were ∼20% less likely to be breast cancer cases than those with the TT genotype. Additionally, one SNP, rs1005912, was associated with a small increased risk among heterozygotes, although rare allele homozygotes did not exhibit a stronger positive association.

Table 3 includes associations with breast cancer stratified by SBCS phase; included are the two SNPs above as well as two polymorphisms that had promising results after our initial targeted phase 1 genotyping. Among phase 1 participants, rs1116195 was significantly associated with an increased risk of breast cancer for minor allele homozygotes (OR, 1.3; 95% CI, 1.0-1.7), whereas minor allele homozygotes for rs243865 tended to have an increased risk (OR, 1.8; 95% CI, 0.8-4.1). Additional genotyping among 1,491 cases and 1,437 controls from phase 2 did not confirm an association between rs1116195 and breast cancer risk. However, results from phase 2 for rs243865 were similar to those from phase 1. In both phases, minor allele homozygotes for rs243865 tended to have a moderate increased risk of breast cancer, although even in combined analysis, this did not reach statistical significance. The two SNPs with significant results among all genotyped participants (rs11644561 and rs11643630) had consistent results when stratified by SBCS phase. Minor allele homozygotes compared with major allele homozygotes were significantly less likely to be breast cancer cases for rs11644561 (OR, 0.6; 95% CI, 0.3-1.0) and for rs11643630 (OR, 0.8; 95% CI, 0.7-1.0).

The LD structure of the 36 polymorphic MMP-2 SNPs was constructed by combining all available genotyping data from 3,027 controls (Fig. 1). The two SNPs with significant risk reductions for homozygotes (rs11644561 and rs11643630) were not found to be in high LD (D = 0.21; r² = 0), and so haplotype analysis was done to assess the effects of these SNPs in concert (Table 4). Compared with the common reference haplotype with both major alleles (H1: GT, 48.5%), the haplotype with minor alleles for both rs11644561 and rs11643630 (H4: AG, 4.4%) was associated with a significantly reduced risk of breast cancer in both additive and dominant models (OR, 0.6; 95% CI, 0.4-0.8). As single SNP analysis resulted in significant risk reductions only for the homozygotes of these variants, and haplotype analysis indicated that it was the two SNPs in combination that best captured a decreased risk of breast cancer, these two SNPS were analyzed further. In logistic regression models that included both rs11644561 and rs11643630, homozygotes for both SNPs were found to have significantly reduced risks; furthermore, an interaction term for the two polymorphisms was not found to reach statistical significance. Finally, haplotype analysis was also conducted for the two MMP-2 SNPs previously reported in the literature; no significant haplotype effects for rs243865 or rs2285053 were observed.

A two-phase case-control study was conducted to first comprehensively evaluate MMP-2 genetic variants in relation to breast cancer risk, and then to validate any promising associations among a second independent sample population. Two MMP-2 SNPs (rs11644561 and rs11643630) were found to have associations with breast cancer risk that were consistent between phase 1 and phase 2 study populations as well as significant in combined analyses. Although the effects of these SNPs were found to be independent, a rare haplotype that included both minor alleles was associated with significant risk reduction. To the best of our knowledge, neither of these two MMP-2 polymorphisms have been previously evaluated for associations with cancer susceptibility.

Two promoter SNPs, rs243865 (−1306) and rs2285053 (−735), have been previously reported to affect MMP-2 transcription in vitro; both C to T transitions result in reduced expression due to the ablation of specificity protein (Sp)1 transcription factor binding sites (19-21). Only one of these SNPs (rs243865) has been previously evaluated in relation to breast cancer risk, and results have been inconsistent. Two studies found no effect (22, 25), whereas two studies (n = 186 and n = 971) found significantly decreased risks of breast cancer associated with the T allele (23, 24). In contrast, in the current study, rs243865 TT homozygotes tended to have an increased risk of breast cancer, although this effect was not significant. Similarly, results from this study for rs2285053 were not compelling; although heterozygotes had a marginally significant increased risk of breast cancer, homozygotes tended to have a diminished risk. Notably, these 2 SNPs were not found to be in high LD in this study population (D = 1; r² = 0.03), and no haplotype effects on breast cancer risk were found.

Although MMP-2 has traditionally been thought of as a mediator of metastasis, a growing body of evidence has connected MMP-2 to earlier aspects of carcinogenesis, including cell growth, inflammation, and angiogenesis (1-5). In addition to a wide range of ECM substrates, including gelatin, elastin, fibronectin, laminin, and collagens, MMP-2 has many non-ECM substrates that include growth factors modulators and cytokines (1, 2, 4). For example, hydrolysis of membrane bound fibroblast growth factor receptor type 1 by MMP-2 releases the soluble ectodomain of the active receptor, thereby influencing the mitogenic and angiogenic activities of FGF (4, 35). Additionally, MMP-2 was shown to contribute to the inflammatory response by being an alternative activator of pro-interleukin 1-β in the absence of the cytokines' favored activator caspace-1 (36). Finally, degradation of ECM substrates may also contribute to cancer development and progression, as MMP-2 cleavage of the proteoglycan decorin releases transforming growth factor-β1 from its extracellular reservoir (37).

Studies of the MMP-2 promoter have identified several putative regulatory regions and transcription factor binding sites within 2 kb of the transcription initiation site, including those for Sp1, p53, S1, S2, activator protein, AP-2, Ets-1, CAAT/enhancer binding protein, cAMP-responsive element binding protein, GCN-His, and Pea3 (38, 39). Some of these elements have been shown to be critical for MMP-2 expression in different cell types or due to different chemical or oncogenic stimuli (38-40). To our knowledge, further upstream MMP-2 promoter sequences do not seem to have been previously characterized. In the current study, rs11644561 and rs11643630 were both found to confer decreased risk for homozygotes; these SNPs are located ∼4 and 2.6 kb upstream of the MMP-2 transcription initiation site, respectively. Using available bioinformatics tools, we tried to evaluate these regions further, but our analysis was uninformative. Therefore, with current evidence, we cannot determine whether these loci represent novel functional SNPs that may affect MMP-2 expression, or else, if together, they best tag another, as yet ungenotyped, variation. As their effects were found to be independent, they may each tag this other loci to different degrees, explaining why the haplotype with both SNPs resulted in capturing the risk reduction more than the recessive effects of either haplotype with only one of the alleles alone.

In summary, we identified two MMP-2 promoter polymorphisms that were associated with modest decreases in breast cancer risk. Homozygotes of the minor alleles for rs11644561 and rs11643630 were 40% and 20% less likely to have breast cancer, respectively. Although a two-phase study design was used to reduce type I error, we cannot rule out the possibility that our findings could be due to chance. Furthermore, neither association remains significant after adjusting for the number of SNPs evaluated. However, this is the largest and most comprehensive analysis of MMP-2 polymorphisms conducted to date, and our results are consistent with both in vitro and in vivo evidence that show a role for MMP-2 in breast cancer development. Therefore, additional studies to evaluate these MMP-2 polymorphisms in population studies are warranted.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the NIH. We thank the participants and research staff of the SBCS for their contributions and commitment to this project and Brandy Venuti for assistance in the preparation of this manuscript. DNA extraction, sample preparation, and genotyping assays using Affymetrix arrays were conducted at the Vanderbilt Survey and Biospecimen and Microarray Shared Resources that are supported in part by the Vanderbilt Ingram Cancer Center (P30 CA68485). Sequenom assays were done by Proactive Genomics.