Matrix Metalloproteinase Gene Polymorphisms Are Associated with Breast Cancer in the Caucasian Women of Russia

We conducted this study to explore the association between matrix metalloproteinase (MMP) gene polymorphisms and breast cancer (BC) risk in the Caucasian women of Russia. In total, 358 affected (BC) and 746 unaffected (cancer-free) women were included in this case-control retrospective study. From BC-related genes in previous studies, ten single nucleotide polymorphisms (SNPs) in five MMP genes (MMP1, 2, 3, 8, 9) were genotyped. The BC risk was calculated by logistic regression (to evaluate the SNPs’ independent effects) and model-based multifactor dimensionality reduction (MB-MDR) (to identify SNP–SNP interactions) methods. The allelic variants’ distribution of c.836 A > G (rs17576) and c. 1721 C > G (rs2250889) MMP9 was significantly different between BC and cancer-free women: for G minor alleles, these SNPs manifested disorder protective effects (OR 0.82 and OR 0.67–0.71, respectively, pperm ≤ 0.035). Eleven haplotypes of six SNPs MMP9 were involved in BC risk (nine haplotypes) and protective (two haplotypes) effects. All 10 SNPs of the MMP genes examined were associated with BC within the 13 SNP–SNP interaction simulated models, with a pivotal role of the two-locus (rs17577 × rs3918242) MMP9 epistatic interaction (defined as 1.81% BC entropy within more than 60% of the genetic models). Under in silico bioinformatics, BC susceptibility MMP polymorphic loci are located in functionally active genome regions and impact genes expression and splicing “regulators” in the mammary gland. The biological pathways of BC MMP candidate genes are mainly realized due to metalloendopeptidase activity and extracellular matrix organization (structure, disassembly, metabolic process, etc.). In conclusion, our data show that MMP gene polymorphisms are related to BC susceptibility in the Caucasian women of Russia.


Introduction
BC is a malignant breast tumor with epithelial origin [1]. According to world statistics (material of the International Agency for Research on Cancer), currently, more than 2 million new cases of BC are detected annually among the world population [2] (http://canscreen5 .iarc.fr, accessed on 17 September 2022). Among all cases of cancer (19.3 million new tumor cases are registered annually in the world), BC is the most frequent (11.7%) [2]. Among the female population, the proportion of BC among all oncological diseases is 24.5%, and regarding global female mortality, the proportion of BC is 15.5% [3]. The problem of BC is no less significant for Russia. According to official statistics, the incidence rate is 82.8 per 100,000 women [4]. BC ranks first both in terms of cancer incidence in Russian women (20.9%) and the cause of female death from malignant tumors (16.2%) [5].
Genetic factors have an essential effect on the susceptibility to BC [6][7][8][9][10]. Genome-wide association studies (GWASs) of BC indicate that more than 180 different single nucleotide polymorphic variants have a connection with the disorder [8]. It should be noted that these polymorphic loci explain only 18% of the disease heritability [9], whereas under modern genetic estimates, the contribution of the hereditary component to BC formation reaches up

Results
A comprehensive description of the phenotypic characteristics of the examined BC and cancer-free women is summarized in Table 1. The age range of the study participants was 28-84 years, with a predominance of postmenopausal women (2/3). There was no difference (p > 0.05) between the BC and cancer-free women regarding age, proportion of premenopausal/postmenopausal subjects, age at menarche, age at menopause, and smoker habits. At the same time, affected women had a higher body mass index (BMI) (p = 0.003) and a higher incidence of obesity when compared with the unaffected ones (p = 0.01). The majority of the BC patients were at clinical cancer stage T0-T2 (74%) and were predominantly estrogen receptor (ER) positive (66%), progesterone receptor (PR) positive (59%), and ductal carcinoma (94%) with a well differentiated/moderately differentiated (G1/G2) tumor histological grade (68%) ( Table 1).
All ten investigated SNPs were in agreement with Hardy-Weinberg equilibrium (HWE) in both studied groups, BC and cancer-free woman, taking into account the Bonferroni correction for the number of examined loci (n = 10, p bonf > 0.005 (0.05/10). According to the data of Supplementary Table S1, among the BC subjects, the observed values p HWE were ≥0.015 and among the cancer-free group, the values were ≥0.086. Table 2 shows the association analysis results. Interestingly, the allele frequencies of the c.836 A>G (rs17576) and c. 1721 C>G (rs2250889) MMP9 gene were significantly different between the BC and cancer-free subjects. Women carrying the G alleles of these loci, providing disorder protection, had an 18% (OR 0.82 95%CI 0.68-0.99 p = 0.035 p perm = 0.035) and 29% (OR 0.71 95%CI 0.52-0.97 p = 0.033 p perm = 0.034) lower BC risk than women with A (rs17576) and C (rs2250889) alleles, respectively. It should be noted that in the additive and dominant models, after adjustment for confounding factors, locus c. 1721 C>G (rs2250889) MMP9 was associated with BC and reduced risk of developing BC (OR 0.69 95%CI 0.51-0.95 p = 0.024 p perm = 0.025 and OR 0.67 95%CI 0.47-0.95 p = 0.026 p perm = 0.026, respectively) ( Table 2).  Next, we examined the associations between the MMP9 gene haplotypes and BC. The detailed findings of the association analysis for various haplotypic constructions are presented in Table 3. We found eleven different haplotypes of the MMP9 gene with six loci linked with BC, among which the vast majority of haplotypes determined an increased disorder risk (nine haplotypes) while only two haplotypes were associated with a low disease risk. Of note, five various haplotypic constructions (four-, five-, and six-locus) included different combinations of the allelic variants: rs3918242 (C), rs3918249 (T), rs17576 (A), rs3787268 (A), rs2250889 (C), and rs17577 (G) showed the most pronounced BC risk effect and increased the likelihood of development of the disease by more than three times (Table 3). Importantly, the direction of the effects of the c.836 A > G (rs17576) and c. 1721 C > G (rs2250889) loci in the composition of haplotypes fully corresponded to the nature of their independent association with the disease (minor G alleles of these loci were protective; major A (rs17576) and C (rs2250889) alleles were risky). After demonstrating that individual SNPs and haplotypes in the MMP9 gene may participate in BC susceptibility, we investigated the outcomes of interlocus interactions between all ten studied polymorphisms and five MMP genes. Using MB-MDR software (https://github.com/imbs-hl/mbmdR, accessed on 18 June 2022), the correlation between SNP interactions and BC was computed. According to the obtained simulation results, all 10 examined SNPs within 13 epistatic interaction models (5 (2-locus), 6 (3-locus), 2 (4-locus), and 1 (5-locus)) were associated with BC risk (Table 4). At the same time, it is important to note that, firstly, two polymorphisms: c.2003 G > A (rs17577) and c.-1562 C > T (rs3918242) MMP9, were included in the vast majority (>60%) of the significant models (9 and 8 models, respectively). Secondly, the two-locus combination: c.2003 G > A (rs17577) × c.-1562 C > T (rs3918242), formed the basis of 8 of the 13 considered genetic models, which indicates its key importance in the genetic determination of BC. The maximum value of the Wald indicator (WH = 48.71) was a five-locus model: The carriage of several specific genotype combinations of the MMP9 gene had the most pronounced BC risk effect (beta = 1.54-1.61 and p = 0.0001-0.0004): Table S2). On the contrary, the combination of genotypes 1G1G (rs1799750) MMP1 × TT (rs679620) MMP3 showed a considerable protective effect on the BC relation (beta = −0.79 and p = 0.0004) (Supplementary Table S2). Four-order interaction models (p < 9.93 × 10 −8 ) Five-order interaction models (p = 2.96 × 10 −12 ) Note: NH: number of significant high risk genotypes in the interaction; beta H: regression coefficient for high-risk exposition in the step 2 analysis; WH: Wald statistic for the high-risk category; NL: number of significant low-risk genotypes in the interaction; beta L: regression coefficient for low-risk exposition in the step 2 analysis; WL: Wald statistic for the low-risk category; p perm : permutation p-value for the interaction model (1.000 permutations); the results were obtained using the MB-MDR method with adjustment for covariates.
To visualize the established BC-involved interlocus interactions of MMP genes, a network (entropy graph) was constructed by multifactor dimensionality reduction (MDR) analysis (http://www.multifactordimensionalityreduction.org/, accessed on 18 June 2022). According the data shown in Figure 1, the largest contribution to BC susceptibility is by the synergistic two-locus interaction: c.2003 G > A (rs17577) × c.-1562 C > T (rs3918242) (which defines 1.81% of the BC entropy; in Figure 1, it is represented in red color), whose contribution to the BC predisposition was 4-5 times higher than any of the main effects of the individual loci (each of the studied loci determined no more than 0.27-0.42% of the entropy) ( Figure 1).

Non-Synonymous (nsSNP) and Regulatory (regSNP) Impact
In total, 6 SNPs among 134 analyzed polymorphisms (10 BC-involved loci and 124 SNPs strongly liked with them) were nsSNPs such as c.
In accordance with the HaploReg public epigenetic data (https://pubs.broadinstitute. org/mammals/haploreg/haploreg.php, accessed on 18 June 2022), we predicted a meaningful function of 10 BC-related polymorphic loci of MMP genes and the vast majority of strongly linked polymorphisms (121 loci out of 124 SNPs, 97.58%) in relation to 10 genes located next to them such as MMP9, SLC12A5, MMP8, RP11-465L10.7, MMP3, RP11-465L10.10, MMP2, LOC100288077, MMP1, and ZNF335 (Supplementary Table S4). Interestingly, nine out of the 10 BC-associated loci (with the exception of rs17577) w located in the region of 42 regulatory motifs and determine their sensitivity to 39 TFs (transcription factors) (Supplementary Table S5). SNP c.-1607 2G > 1G (rs1799750) MMP1 is situated in the region of the largest number of DNA motifs (n = 21). In general, 105 loci out of 124 strongly coupled polymorphisms (84.68%) define the DNA motifs' sensitivity to plenty of TFs (Supplementary Table S4 Figure 1, it is represented in red color), whose contribution to the BC predisposition was 4-5 times higher than any of the main effects of the individual loci (each of the studied loci determined no more than 0.27-0.42% of the entropy) ( Figure  1).   Table S3).

Predicted Functional Outputs for BC-Associated SNPs
In accordance with the HaploReg public epigenetic data (https://pubs.broadinstitute.org/mammals/haploreg/haploreg.php accessed on 18 June 2022), we predicted a meaningful function of 10 BC-related polymorphic loci of MMP  Table S4).
Using the Gene Ontology biological resource (http://geneontology.org/, accessed on 18 June 2022), about 50 different biological pathways were identified (Supplementary Table  S11), which are largely related to metalloendopeptidase activity and extracellular matrix organization (structure, disassembly, metabolic process, etc.). Several biological mechanisms demonstrated a Fold Enrichment indicator of more than 100 such as activation of matrix metalloproteinases, extracellular matrix disassembly, collagen metabolic/catabolic process, cellular response to UV-A, and plasminogen-activating cascade (Supplementary Table S11). The interaction network of BC candidate genes (  The biological mechanisms characteristic of the 12 breast-specific genes (we identified more than 60 of these pathways; Supplementary Table S13) are very like those identified for the above-mentioned 32 BC-involved genes (Supplementary Table S11) and are in a large part represented by metalloendopeptidase activity/extracellular matrix organization, etc. The breast-specific gene interactions (Figure 3) are mainly realized due to the overall domains of proteins (45.22%) and positional communications (38.04%) with the maximum weight index (0.52) of the CETP-PLTP genes paired interaction (Supplementary Table S14).

Discussion
In our report, we demonstrated the involvement of MMP gene polymorphisms in BC susceptibility in the Caucasian women of Russia. All examined ten SNPs of the MMP genes were BC-associated: two loci (c.836 A > G (rs17576) and c. 1721 C > G (rs2250889) MMP9) showed an independent disease protective effect, eleven haplotypes of six SNPs MMP9 were mostly involved with BC risk, and thirteen SNP-SNP interaction simulated models are correlated with BC. Under in silico materials, BC susceptibility MMP polymorphic loci are located in functionally active regions of the breast genome, impact on epigenetic processes, and are regulators of gene expression/splicing.
Our results demonstrated a protective impact of the G allele (the C allele is risky) of SNP c. 1721 C > G (rs2250889) MMP9 on BC formation, both independently (OR 0.67-0.71) and as part of eight haplotypes of the MMP9 gene. The relationship of c. 1721 C > G (rs2250889) MMP9 with BC was considered in seven previously published works: four experimental studies (Malaysian, Chinese, and Jordanian populations were studied) [25,27,29,30] and three meta-analyses [20][21][22]. The association of this polymorphism with disease was demonstrated in two of these studies (one experiment performed in a Malaysian population [29] and one meta-analysis [21]). It should be emphasized that the GG genotype (OR 10.84) has a risk value for BC in Malaysian women [29] while in Caucasian women in Russia, the G allele exhibits a protective effect in relation to the disease (our data). The above data make it possible to suggest the presence of interethnic peculiarities (Asian/European) in the association of c. 1721 C > G (rs2250889) MMP9 with BC; however, this assumption needs to be confirmed in further molecular and genetic studies.
Another MMP9 gene polymorphism independently associated with BC in Caucasian women of Russia, c.836 A > G (rs17576) (the G allele is protective, OR 0.82) also contributes to the risk of disorder within nine haplotypes of the MMP9 gene (the A allele is risky). Data from the literature on the association between c.836 A > G (rs17576) MMP9 and the disease are very contradictory. No associations of this locus with the disease were identified in two works [25,27] while opposite results (risk/protective value of polymorphism) were demonstrated in another three studies [29,31,32]. In three meta-analyses, significant associ-ations of c.836 A > G (rs17576) MMP9 with breast cancer were not established [20][21][22]. Our data on the protective effect of the G allele (rs17576) on breast cancer in Caucasian women of Russia are consistent with the results of Resler et al., who studied white women of the Seattle metropolitan area (USA) [31], and contrast with the data obtained in Malaysian [29] and Brazilian [32] populations (according to their data, the G allele is risky).
We obtained interesting data in silico regarding the pronounced cell/organ-specific functional effects of c.836 A > G (rs17576) and c. 1721 C > G (rs2250889) MMP9 in the mammary gland, which is a BC target organ. These BC risk loci (together with coupled SNPs) are involved in amino acid modification of MMP9 protein (p.Q279R and p.574P, respectively), epigenetic control gene activity due to enhancer-and promoter-driven transcriptions in primary epithelial and myoepithelial breast cells, and gene expression/splicing level regulation in BC-related organs (brain, adipose, adrenal gland, thyroid, blood, etc.). The above-stated SNPs' functional actions may be the pathophysiological basis of these loci associations with BC.
The MMP9 protein (gelatinase B), resulting from the transcription of the gene of the same name, is a well-known multifunctional collagenase [33,34]. MMP9, due to its ability to destroy type IV collagen and denatured collagens, resulting in the destruction of basement membranes, is essential for the biology of tumors, including BC [34,35]. Importantly, MMP9 protein is constitutively expressed by breast cancer cell lines (MDA-MB-231) [33]. The tumorigenic effects of MMP9 are realized in lots of ways, such as DNA degradation and accompanying molecular changes, epithelial to mesenchymal transition and cancer stem cell formation/support, matrix degradation, production and activation of matrix-associated and non-matrix cytokines, inflammatory tumor-related activity, cancer immunological monitoring regulation, change in the cyto-skeletal organization during tumorigenesis (tumor-stromal relationship, etc.), proliferation and invasion of cancer cells, effect on angiogenesis (apoptosis) factors, etc., which predetermines the fundamental role of this collagenase in BC progression and metastasis, relapse, and survival of disease-affected patients [34].
There is convincing evidence of an important prognostic value of the MMP9 expression level in BC patients [36,37]. As a result of a meta-analysis (based on data on 2344 women with BC from 15 studies), Song et al. showed a direct link between overexpression of MMP9 and both relapse-free and overall survival in BC patients [36]. The meta-analysis performed by Jiang et al. (the materials of 41 experimental studies, including 6517 BC patients, were used) not only shows the association of increased expression of MMP9 with a shorter overall survival of BC-affected women but also the association of high expression of MMP9 with other BC clinical (clinicopathological) signs such as histological grade, clinical stage, larger tumor sizes, and lymph node metastasis [37]. The literature suggest a link between an increased content of MMP9 protein and negative status of hormonal receptors and a higher tumor grade [35]. As a result of an experimental study of patients with primary early BC, higher expression of MMP9 in tumor cells was found in comparison with tumor-associated stroma and its correlation with a tumor cell proliferation index [38]. A higher level of MMP9 protein in the tumor than in normal tissue and its direct connection with increased tumor size were also reported in the work of Przybylowska et al. [12]. The expression of MMP9 protein concerns important regulators/markers of the cell cycle/ECM remodeling such as cell division control protein 42, epidermal growth factor and cell surface adhesion receptors, cytokeratin 17, and Ki67 proliferation biomarker [35]. The abovementioned data indicate the considerable putative clinical and pathophysiological significance of MMP9 in BC and, consequently, functionally meaningful polymorphisms of this gene may become important prognostic markers of this disease in the future.

Study Subjects
We enrolled 358 affected (BC) and 746 unaffected (cancer-free) women at the Belgorod Regional Oncological Dispensary and Belgorod Regional Clinical Hospital of St. Ioasaph from March 2010 to December 2016. All participants were Caucasian women who were born and living in central Russia [39,40]. BC diagnosis was histopathologically confirmed by certified pathologists. The control group were BC free (confirmed by mammography or breast ultrasound) without malignant tumor history and heavy disorder-related to the vital organs. We obtained ethical approval from the Human Investigation (Ethics) Committee of Belgorod State National Research University.

SNP Selection and Genotyping
We selected the phenol-chloroform method to obtain quality DNA extract from the peripheral venous blood samples (≈4-5 mL) [41]. DNA extraction was carried out according previously described procedure [42]. A Nanodrop 2000 was used to detect the DNA concentration and purity [43].

Statistical Analysis
The SNP genotype frequencies in the BC and cancer-free subjects were analyzed separately to find out the compliance with Hardy-Weinberg equilibrium [53]. Differences in the distribution of polymorphic variant frequencies between the BC and cancer-free groups were evaluated based on four common genetic models (recessive, additive, dominant, and allelic) by logistic regression [54] with age and BMI adjustment. Association links of genetic markers with BC were appreciated based on the indicators' OR (odds ratio) and 95% CIs (confidence intervals for OR) [55]. The individual SNPs and their haplotype association analyses were executed using gPLINK software [56]. The MB-MDR and MDR methods and the corresponding software packages were utilized to evaluate the association of SNP interactions with BC [57] (http://sourceforge.net/projects/mdr, accessed on 18 June 2022). Among the many obtained simulation epistatic models for subsequent permutation procedures, we selected models corresponding to the level of statistical significance chosen by us according to the Bonferroni amendment (taking into account the 10 examined loci combinations), such as p interact < 1.11 × 10 −3 (<0.05/45) (2-locus models), p interact < 4.17 × 10 −4 (<0.05/120) (3-locus), p interact < 2.38 ×10 −4 (<0.05/210) (4-locus), and p interact < 1.98 × 10 −4 (<0.05/252) (5-locus). The association analysis results were corrected for multiple testing by a permutation test [58,59]. As a result, p perm value ≤ 0.05 for individual SNPs (adaptive permutations were carried out [54], their haplotypes, and SNP*SNP interactions (1000 permutations were done [60]) was recognized as the cutoff parameter for statistical significance.

Conclusions
MMP gene polymorphisms located in functionally active genome regions in the mammary gland are related to BC susceptibility in the Caucasian women of Russia.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijms232012638/s1. References  are cited in the supplementary materials.  Informed Consent Statement: Informed consent was obtained from all subjects involved in this study.

Data Availability Statement:
The data generated in the present study are available from the corresponding author upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.