Breast cancer is the most frequent malignancy and the leading cause of cancer-related death in women worldwide (1). Based on the hormone receptor (estrogen receptor [ER] or progesterone receptor [PR]) and human epidermal growth factor receptor 2 (HER2) status, breast cancer is classified into four main subtypes; hormone receptor-positive/HER2-negative (or luminal A), hormone receptor-positive/HER2-positive (or luminal B), hormone receptor-negative/HER2-positive (HER2-positive or HER2-enriched), and triple-negative breast cancer (TNBC), and the choice of systemic treatment depend on the cancer subtypes (2, 3). In recent decades, there has been remarkable progress in the development of novel molecular targeted therapies (i.e., cyclin-dependent kinase 4 and 6 inhibitors, poly (ADP-ribose) polymerase inhibitors, and HER2 or trophoblast cell surface antigen-2 antibody-drug conjugates) and immunotherapies (i.e., programmed cell death-1 [PD-1] inhibitors) for patients with breast cancer who failed primary systemic treatment with endocrine therapy or chemotherapy (4). Despite significant improvements in therapies, a considerable number of patients still experience recurrence or metastasis after initial therapy, which is responsible for the high mortality in breast cancer (5–7).

Tumor progression and metastasis are driven by the complex interplay between malignant tumor cells and the surrounding non-malignant stroma, comprising the extracellular matrix (ECM), stromal cells such as endothelial cells (ECs), fibroblasts, and infiltrating immune cells (8, 9). Matrix metalloproteinases (MMPs) are zinc-dependent endopeptidases that can degrade and remodel the ECM (10). Studies have identified 23 MMPs in humans (11), and they are generally categorized as collagenases (MMP-1 [interstitial collagenase], -8 [neutrophil collagenase], -13 [collagenase-3], and -18 [collagenase-4]), gelatinases (MMP-2 [72 kDa gelatinase/type IV collagenase], and -9 [92 kDa gelatinase/type IV collagenase]), stromelysins (MMP-3 [stromelysin-1], -10 [stromelysin-2], and -11 [stromelysin-3]), matrilysins (MMP-7 [matrilysin], and -26 [endometase or matrilysin-2]), membrane-type (MMP-14 [MT1-MMP], -15 [MT2-MMP], -16 [MT3-MMP], -17 [MT4-MMP], -24 [MT5-MMP], and -25 [MT6-MMP]), or other miscellaneous types based on substrate specificity or domain structure (12, 13). MMPs, which are produced by non-malignant stromal cells as well as tumor cells, contribute to tumor progression by regulating tumor growth, invasion, angiogenesis, and metastasis (11, 14–16). Although MMPs were initially thought to promote tumor progression by remodeling the ECM through their proteolytic activity, accumulating evidence has provided a novel understanding of the roles and mechanisms of MMPs in cancer. Some MMPs, such as MMP-3 and -8, exhibit protective roles in cancer, and the same MMP can be pro-tumorigenic or anti-tumorigenic depending on the cell type in which it is expressed or the stage of cancer (12, 17, 18). Moreover, MMPs produced from non-malignant cells within the tumor microenvironment (TME) play crucial roles in cancer progression and metastasis. Stromal MMPs, such as MMP-9 and -12, particularly those produced by infiltrating inflammatory cells, have been shown to play anti-tumorigenic roles (17), highlighting the need for further studies on cell type-specific roles of MMPs. MMPs also contribute to tumor progression through their non-catalytic functions (14, 15). Accumulating evidence supports the involvement of MMPs in additional biological processes, such as cell signaling and immune regulation, independent of proteolytic activity (11, 15). In particular, some MMPs, such as MMP-2 and -9, have been demonstrated to influence cancer progression by modulating immune responses in the TME (19, 20). In addition, their biological roles are mediated by non-ECM substrates such as cell surface receptors, growth factors, cytokines, chemokines, and nuclear proteins, as well as ECM substrates. For example, MMP-1 derived from fibroblasts was found to promote the growth and invasion of breast cancer cells by activating protease-activated receptor 1 (PAR1) (21). Macrophage-derived MMP-12 inactivates Glu-Leu-Arg (ELR)1 CXC chemokines, which are involved in neutrophil recruitment to sites of injury or inflammation, and generates C-C chemokine receptor (CCR) antagonists by inactivating monocyte chemotactic proteins, indicating the role of MMP in the regulation of inflammatory response by proteolysis of chemokines (22). While most MMPs are secreted proteases and are recognized to have extracellular activity, recent findings revealed that some MMPs, including MMP-1, -2, -3, -7, -8, -9, -10, -11, -12, -14, -23, and -26, have intracellular functions, suggesting that intracellular and extracellular MMPs are likely to play roles in cancer progression (23).

A subset of MMPs is frequently overexpressed in breast cancer and is associated with prognosis; thus, they have been considered promising therapeutic targets for breast cancer (10, 24, 25). However, while there is a MMP inhibitor approved by the U.S. Food and Drug Administration for treating periodontitis, no MMP inhibitor has been approved for the treatment of cancer (26–28). This may be partly attributable to the lack of understanding of the cell type-specific roles of individual MMP in breast cancer. In this review, the current understanding of the roles of tumoral and stromal MMPs in breast cancer progression is summarized, and recent efforts or advances in the development of highly selective MMP inhibitors are discussed.

In breast cancer, several MMPs are expressed in various cell types comprising the TME, such as fibroblasts, immune cells, and tumor cells, and their expression at the gene or protein level has been reported to be significantly associated with the prognosis of breast cancer patients ( Table 1 ).

The role of MMPs in breast cancer progression has been investigated based on their overexpression and prognostic significance. Some MMPs have been shown to promote tumor growth through the regulation of cell proliferation, apoptosis, angiogenesis, or metastasis, while some MMPs have been found to exert both pro-tumorigenic and anti-tumorigenic roles depending on the cell types in which they are expressed or the stages of the disease ( Table 2 , Figure 1 ). Furthermore, a few MMPs have exhibited anti-tumorigenic functions in breast cancer ( Table 2 , Figure 1 ).

The proteolytic function of MMPs is regulated by tissue inhibitors of MMPs (TIMPs) (106). The four mammalian TIMPs (TIMP-1, -2, -3, and -4) are endogenous secreted proteins to inhibit MMPs (107). Each of the TIMPs inhibits specific MMPs, and some MMPs such as MMP-2 and -9 interact with several TIMPs (108). Alterations in the expression of TIMPs have been identified in human cancers and their expression correlates with clinical outcome (107, 108). Given their anti-proteolytic function in ECM, TIMPs were initially thought to play protective roles in cancer progression (108). However, TIMPs have been found to be associated with poor prognosis and exert tumor-promoting functions in some human cancers (107, 108). In addition to interaction with MMPs, TIMPs can bind to other interaction partners such as cell surface receptors, and they have multiple biological functions including cell proliferation, apoptosis, migration, invasion and angiogenesis by MMP-dependent or -independent mechanisms (106, 107, 109).

In breast cancer, TIMP-1 expression in primary tumor tissues is an independent poor prognostic factor (32, 110). Moreover, studies have demonstrated that breast cancer patients with high levels of serum TIMP-1 have a significantly shorter survival (51, 111). Consistent with its prognostic significance, TIMP-1 overexpression was shown to promote MDA-MB-231 tumor growth in SCID mice (112). TIMP-1/CD63 signaling enhanced cell motility through the induction of epithelial−mesenchymal transition phenotypes in human breast epithelial cells (113). TIMP-1 knockdown in TNBC cells induced cell cycle arrest, and Akt signaling pathway was associated with the regulation of cyclin D1 expression by TIMP-1, indicating that TIMP-1 increases TNBC cell proliferation by Akt activation (114). This study further showed that blocking TIMP-1 activity using neutralizing antibody inhibits TNBC cell growth in vivo. CAF-derived TIMP-1 was involved in the migration and growth of breast cancer cells though TIMP-1/CD63/ITGB1/STAT3 feedback loop (115). Regarding the prognostic value and role of TIMP-2 and -3 in breast cancer, conflicting results have been reported. High tissue expression levels of TIMP-2 were significantly associated with favorable clinical outcome (116). Several studies have demonstrated tumor suppressive role of TIMP-2 in breast cancer. TIMP-2 adenoviral delivery inhibited tumor growth, angiogenesis and metastasis in MDA-MB-231 breast tumor xenografts (117). In another study, TIMP-2 overexpression delayed growth and angiogenesis of mammary carcinoma in vivo in association with downregulation of vascular endothelial growth factor (VEGF) expression, indicating that reduced VEGF expression plays a role in anti-tumorigenic effects of TIMP-2 (118). A recent study also revealed inhibitory effects of TIMP-2 on tumor growth and metastasis in murine model of TNBC (119). Conversely, the correlation of high TIMP-2 levels with poor prognosis in breast cancer has also been reported (32, 120, 121). Tumor cell-derived TIMP-2 was shown to induce endothelial dysfunction and promote transmigration of breast cancer cells across vascular endothelial monolayers through the activation of endothelial MMP-2 in the presence of active MMP-14, suggesting the role of TIMP-2/MMP-14/MMP-2 pathway during metastasis (122). As for TIMP-3, stromal TIMP-3 expression is a poor prognostic factor in breast cancer (32), while high TIMP-3 mRNA levels are associated with favorable prognosis (123). TIMP-3 overexpression inhibited cell growth and induced apoptosis in breast cancer cells in vitro (124). In contrast, Timp3 loss suppressed tumorigenesis in mouse models of human breast cancer, suggesting the tumor-promoting role of TIMP-3 in vivo (125–127). TIMP-4 expression correlated with poor clinical outcome in early breast cancer (126). Treatment with recombinant TIMP-4 significantly stimulated the growth of MDA-MB-231 cells (127). Further studies to elucidate the mechanistic roles and functions of TIMPs in breast cancer are warranted.

Several MMP inhibitors have been developed and tested for various cancers based on their prognostic significance and promising preclinical data (16, 27). Early clinical trials with broad-spectrum MMP inhibitors (such as batimastat and marimastat) were unsuccessful because of poor bioavailability, lack of survival benefits, and toxicity (16, 27). The main reasons for these failures have been attributed, in part, to a poor understanding of the roles of MMPs, lack of drug specificity, and inappropriate clinical trial design (11, 12, 16). As some MMPs have protective roles in cancer progression, the negative results of early clinical trials with broad-spectrum of MMP inhibitors were thought to be related to the effects of broad-spectrum MMP inhibitors on anti-tumorigenic MMPs (12, 16). Given that MMP inhibitors are likely to be effective in the early or pre-metastatic stages based on their preclinical data, the design of previous clinical trials of MMP inhibitors tested in patients with advanced or metastatic cancer may be inappropriate (16). Moreover, MMP inhibitor-related toxicities such as musculoskeletal toxicity are assumed to be related to the poor specificity of MMP inhibitors (11, 16). Therefore, substantial efforts have been made to develop more specific MMP inhibitors. However, clinical trials for small molecule-based specific MMP inhibitors (such as tanomastat, prinomastat, and rebimastat), which still target multiple MMPs, have also been halted or canceled owing to negative results (16). Since then, attempts have been made to develop highly specific MMP inhibitors that target a single MMP using novel approaches. In a phase III trial of the broad-spectrum MMP inhibitor marimastat in metastatic breast cancer, marimastat did not prolong progression-free survival (128). Phase II small pilot trials of adjuvant marimastat and rebimastat in early-stage breast cancer were performed; however, further clinical trials with these MMP inhibitors in adjuvant settings were not feasible due to failure of chronic administration to reach the target ranges of plasma concentrations and toxicity (129, 130).

Several novel approaches have been attempted for the development of highly selective MMP inhibitors. The non-catalytic domains of MMPs have been targeted. MMPs are composed of various domains, including a signal peptide, propeptide, catalytic domain that binds a Zn residue, and hemopexin (PEX) C-terminal domain, while the domain architecture differs depending on the type of MMP (11, 24). As active sites in the catalytic domain of MMPs are highly conserved (11, 12, 24), MMP inhibitors that target the catalytic domains have the potential to repress multiple MMPs rather than specific MMPs. In contrast, the non-catalytic domains of MMPs, such as the PEX domain, vary between MMPs, and targeting these domains is likely to improve the specificity of MMP inhibitors. In addition, MMPs have been shown to play roles in cancer progression and metastasis through their non-catalytic functions, independent of their proteolytic activity (14, 15). Therefore, targeting the non-catalytic domain of MMPs is a promising approach to develop specific MMP inhibitors. For example, given the pro-tumorigenic role of the PEX domain of MMP-14 through homodimerization or interactions with other molecules, a novel small molecule that specifically inhibits the PEX domain of MMP-14 was identified, which was shown to suppress tumor growth in a xenograft mouse model of breast cancer (131).

Metastatic breast cancer is a major cause of death from breast cancer, and targeting MMPs involved in breast cancer metastasis has also been considered. Breast cancer frequently metastasizes to several organs, including bone, lung, brain, and liver; with bone being the most common metastatic site of breast cancer (97). Accordingly, attempts have been made to develop selective MMP inhibitors that are effective against bone metastatic breast cancer. A MMP-13 selective inhibitor (Cmpd-1), which was identified from a panel of small molecule pyrimidinetrione-based inhibitors (132), was shown to inhibit osteolytic damage and reduce primary tumor growth in a human breast cancer xenograft model of TNBC using MDA-MB-231 cells, suggesting that MMP-13 selective inhibitors reduce tumor-induced bone osteolysis (133). Moreover, bisphosphonic-based MMP-2 inhibitors specifically targeting bones also demonstrated inhibitory effects on tumor growth and tumor-associated bone destruction in bone metastatic mouse models of breast cancer (134)

Currently, MMP inhibitory monoclonal antibodies are considered promising MMP-targeting therapies because they have higher target selectivity and better pharmacokinetic profiles than small-molecule agents (12, 135). Inhibitory monoclonal antibodies targeting a single MMP, mainly MMP-9 or -14, have been developed and demonstrated anti-tumor activity in preclinical models of breast cancer. A monoclonal antibody against murine MMP-9 (AB0046) decreased primary growth in an orthotopic model of HER2-driven breast cancer (HC11-NeuT) in immunocompetent mice (136). This study also showed that combined treatment with an anti-MMP-9 antibody and immune checkpoint inhibitor (ICI) targeting programmed cell death-ligand 1 (PD-L1), improved anti-tumor immune response to anti-PD-L1 by increasing the levels of T-helper cell 1 type cytokines and infiltration of effector/memory T cells into tumors. This suggested that MMP-9 promotes anti-tumor T cell response. In contrast, in the MMTV-PyMT luminal B breast cancer model, blocking active MMP-9 with a monoclonal antibody suppressed lung metastasis without affecting primary tumor growth (137). As MMP-14 is a membrane-type MMP, and active MMP-14 is located on the cell surface, antibody-mediated therapy can block the function of MMP-14 in cancer. An anti-MMP-14 inhibitory antibody (DX-2400) identified using a human Fab displaying phage library demonstrated inhibitory effect on primary tumor growth and incidence of metastasis to the lung and liver in a MDA-MB-231 orthotopic model (103). Based on the anti-angiogenic activity of DX-2400, this study also evaluated the combination of DX-2400 and bevacizumab (an antibody against VEGF) and found that it retarded tumor growth to a greater extent than treatment with either agent individually in the MDA-MB-231 xenograft model, suggesting a promising approach of anti-angiogenic agents in combination with MMP-14 inhibitor for synergistic anti-tumor effects in breast cancer. DX-2400 was further validated to reduce primary tumor growth in orthotopic murine breast cancer models and was found to improve the response to radiotherapy in MMP-14-high-expressing 4T1 tumors (138). Importantly, this study revealed that tumor growth suppression by MMP-14 blockade was associated with reduced immunosuppressive TGF-β and the shift of macrophages to anti-tumor M1 phenotype, while improved vascular function and tissue oxygenation by MMP inhibition-induced increase in tumor inducible nitric oxide synthase led to increased radiation response in MMP-14-high-expressing tumors. Given that dysregulation of ECM genes linked to the activation of immunosuppressive TGF-β signaling in CAFs predicts a failure to PD-1 blockade (139), MMP-14 inhibitory antibody inhibiting TGF-β signaling is likely to improve the response to ICI. Another MMP-14-selective antibody (Fab 3A2) was also identified by screening human Fab fragment libraries carrying long convex-shaped paratopes to overcome limitations in physical access of the antibody to the active region of MMP-14, which is concave-shaped. Fab 3A2 was found to inhibit protease activity of the MMP-14 by accessing the convex pocket of MMP-14 (140). Moreover, a recombinant human IgG specific to MMP-14 (IgG 3A2) demonstrated inhibitory effects on the growth of the primary tumor and lung metastasis in the 4T1 highly metastatic, syngeneic, orthotopic model of breast cancer (141). MMP-14-specific Fabs were also identified using phage-displayed synthetic humanized Fab library against the extracellular domain of MMP-14, and among them, Fab 3369, which inhibits the catalytic domain of MMP-14, demonstrated anti-tumor activities in vitro and in vivo (142). Fab3369 significantly decreased the invasion of TNBC cells in vitro and suppressed tumor growth and metastasis in TNBC mouse models (mammary orthotopic tumor xenografts and syngeneic 4T1 mammary tumors). Importantly, its in vivo anti-tumor effect was associated with the disruption of immunosuppressive TME, including limiting tumor neoangiogenesis and hypoxia, suggesting that MMP-14 inhibitory antibody suppresses tumor progression and metastasis through its effect on TME in TNBC. Importantly, these promising preclinical data of MMP-9 and -14 inhibitory monoclonal antibodies in HER2-positive breast cancer or TNBC mouse models showed that the anti-tumor effects of MMP inhibitory antibodies are associated with disruption of the immunosuppressive TME, indicating that the use of MMP inhibitors in combination with immunotherapy could improve the efficacy of immunotherapy in breast cancer.

The expression and activity of MMPs are regulated by several mechanisms, including TIMPs (107, 108), noncoding RNAs (143, 144), cell signaling pathways (145–147) or E3 ubiquitin ligases (148) in cancer. MMP inhibitors using TIMPs or E3 ubiquitin ligases are emerging as alternative strategies to develop highly specific MMP inhibitors. A recently developed platform identified soluble TIMP-1 variants that are highly selective for the inhibition of MMP-3 over MMP-10, which is the MMP most similar to MMP-3 in sequence, structure, and function, indicating that this platform and screening strategy can be used to develop selective MMP inhibitors (149). Knockdown of the E3 ligase WSB-1 inhibited the metastatic potential of hormone receptor-negative breast cancer in vitro and in vivo in association with decreased MMP activity, suggesting that the activity of pro-tumorigenic MMPs can be regulated by E3 ligase (150). In addition, several microRNAs (miRNAs) have been reported to be involved in cancer progression and metastasis through MMP regulation (144). In breast cancer, miR-106b (39) and miR-429 (151) have been shown to play a role in bone metastasis of breast cancer by regulating MMP-2 and -9, respectively. Another study revealed that miR-509 suppressed the invasion and trans-endothelial migration of brain metastatic TNBC cells by inhibiting Rho-mediated MMP-9 expression, and high MMP-9 gene expression is significantly associated with shorter brain-metastasis-free survival in breast cancer patients, indicating that miR-509/Rho/MMP-9 axis is a potential target of brain metastasis of breast cancer (46). These results suggest that targeting miRNAs that regulate MMP expression is a potential avenue to target MMP activity for breast cancer therapy. MMPs have also been shown to be regulated by upstream signaling pathways in breast cancer. For example, upregulation of MMP-14 was blocked by upstream PI3K-AKT dependent β-catenin accumulation, thereby inhibiting the invasion and migration of breast cancer cells (152). Our recent study also showed that CCL2 produced by MMP-11-overexpressing macrophages activates the MAPK pathway through its receptor CCR2 in breast cancer cells, thereby enhancing the migration of HER2-positive breast cancer cells by increasing the expression of MMP-9, illustrating the regulation of MMP-9 by the MAPK pathway (57).

The failures of early clinical trials evaluating broad-spectrum MMP inhibitors as anti-cancer agents revealed that MMPs can be both drug targets and antitargets depending on the cell type in which they are expressed or the stage of cancer. Despite the disappointing results for early broad-spectrum MMP inhibitors and the anti-tumorigenic roles of some MMPs, selective MMP inhibitors that have been developed recently are still promising strategies for cancer treatment. MMP inhibitory monoclonal antibodies targeting MMP-9 or -14 demonstrated promising anti-tumor activities in preclinical models of breast cancer and may exhibit clinical benefits in cancer patients without significant toxicity in clinical trials. Moreover, their anti-tumor effects were shown to be associated with the modulation of the tumor immune microenvironment, suggesting that a selective MMP inhibitor in combination with immunotherapy could enhance patients’ response to immunotherapy. However, several challenges remain to be resolved. These include: 1) Clinical trials for MMP inhibitors should be appropriately designed considering the stages of cancer in which MMPs are involved. Based on the pro-tumorigenic roles of MMPs in the pre-metastatic stages of cancer, MMP inhibitors are likely to be clinically beneficial in early-stage cancer. 2) Given the broad expression of multiple MMPs in various cell types, including tumor cells, fibroblasts, ECs, and immune cells, local targeted delivery of MMP inhibitors to specific sites may be required. 3) The identification and validation of reliable biomarkers for predicting the efficacy or toxicity of MMP inhibitors are also necessary to accelerate the progress of drug development. 4) Despite the known prognostic significance of tumoral or stromal MMPs in breast cancer, their cell type-specific or stage-specific roles in breast cancer progression and metastasis are not fully understood. Thus, further studies to clarify the mechanistic roles of tumoral or stromal MMPs during cancer progression are required to develop novel MMP inhibitors for breast cancer treatment.

MK wrote, reviewed and edited the manuscript.