The Role of the Metzincin Superfamily in Prostate Cancer Progression: A Systematic-Like Review

Prostate cancer remains a leading cause of cancer-related morbidity in men. Potentially important regulators of prostate cancer progression are members of the metzincin superfamily of proteases, principally through their regulation of the extracellular matrix. It is therefore timely to review the role of the metzincin superfamily in prostate cancer and its progression to better understand their involvement in this disease. A systematic-like search strategy was conducted. Articles that investigated the roles of members of the metzincin superfamily and their key regulators in prostate cancer were included. The extracted articles were synthesized and data presented in tabular and narrative forms. Two hundred and five studies met the inclusion criteria. Of these, 138 investigated the role of the Matrix Metalloproteinase (MMP) subgroup, 34 the Membrane-Tethered Matrix Metalloproteinase (MT-MMP) subgroup, 22 the A Disintegrin and Metalloproteinase (ADAM) subgroup, 8 the A Disintegrin and Metalloproteinase with Thrombospondin Motifs (ADAMTS) subgroup and 53 the Tissue Inhibitor of Metalloproteinases (TIMP) family of regulators, noting that several studies investigated multiple family members. There was clear evidence that specific members of the metzincin superfamily are involved in prostate cancer progression, which can be either in a positive or negative manner. However, further understanding of their mechanisms of action and how they may be used as prognostic indicators or molecular targets is required.


Introduction
Prostate cancer (PrCa) is one of the major causes of cancer-related morbidity in men worldwide [1,2]. The early stages of PrCa are androgen-dependent, but during PrCa progression, the tumors become independent of androgens [1,3]. The detection of PrCa is difficult, with symptoms often not being apparent until metastasis has occurred [1]. The use of the Prostate-Specific Antigen (PSA) test is considered a gold standard, yet remains flawed, with a considerable false-positive rate [1,4]. The survival rates for men diagnosed with PrCa have increased, although the treatment options can have significant side effects [1,2]. An increased understanding of the etiology of this disease provides the potential to develop more specific detection methods and/or alternative treatment modalities.
The metzincin superfamily represents a large group of proteases named after a specialized structural component, a zinc ion-binding methionine turn sequence within their catalytic domain [5][6][7]. The superfamily can be divided into families and subgroups on the basis of other structural and functional features ( Figure 1). The Matrixin family consists of the soluble Matrix Metalloproteinase (MMP) and Membrane-Tethered Matrix Metalloproteinase (MT-MMP) subgroups that are principally regulated by the Tissue Inhibitor of Metalloproteinases (TIMP) family, and the Astracin family comprises the BMP1/TLL and Meprin subgroups, whereas the Adamalysin family consists of the A Disintegrin and Metalloproteinase (ADAM) and A Disintegrin and Metalloproteinase with Thrombospondin Motifs (ADAMTS) subgroups [8]. The metzincins are well-known for their roles in development and disease, largely through remodeling of the extracellular matrix (ECM) [9][10][11][12][13]. Members of the metzincin superfamily have been increasingly implicated in cancer progression, including a key role in the metastatic process via their ability to remodel the ECM of tumors [13][14][15]. However, the exact role varies, with some being tumorpromoting, others having an antitumorigenic function and others seemingly not playing a role [13,14,[16][17][18]. The metzincin superfamily is therefore of interest as a potential source of biomarkers and/or targets for therapeutic interventions, although the results of the clinical trials to date have been discouraging [19]. However, given the pressing need for both biomarkers and therapeutics in PrCa, it is timely to conduct this systematic-like review in order to synthesize the role of the metzincin superfamily in this disease.

Results
Extensive database searching was undertaken to identify studies that investigated the role of metzincin superfamily members in PrCa progression, as described in Materials and Methods. This identified 205 articles that are presented in five tables, each covering a specific subgroup of the metzincin superfamily or their regulators-specifically, the Matrixin family subgroups MMPs and MT-MMPs, the TIMPs and the Adamalysin family subgroups ADAMs and ADAMTSs. A number of studies involved more than one of these groups and so are included in more than one table. No articles on the BMP/TLL or Meprin subgroups within the Astracin family were identified.

Soluble Matrix Metalloproteinases (MMPs)
The most extensively studied metzincin superfamily subgroup are the soluble MMPs, with 138 articles included (Table 1). Generally, MMPs have been demonstrated to act in a protumorigenic manner-particularly, MMP-2, MMP-7 and MMP-9, which have been the most widely studied of this subgroup.

Authors
Year
Bonaldi et al. [ De Cicco et al. [39] 2008 MMP-2 and 9 * Yes-Prostate samples Low serum MMP-2 (but not MMP-9) associated with increased risk of disease progression. Expression of MMP-2, 7 and 9 increased with PrCa progression. MMP-2 knockout mice showed reduced tumor burden, prolonged survival, decreased lung metastasis, and decreased blood vessel density. Knockout of MMP-7 or MMP-9 did not impact tumor growth or survival but affected blood vessel formation.
Several publications have also demonstrated increased MMP-7 in PrCa [34,56,73,84,99,104,147,151], including correlations with metastasis [106] and chemoresistance [135]. However, others have identified no change in expression in PrCa [42] or, indeed, a negative correlation with disease [52], including when examining the levels of the active form of this enzyme [29,101]. For MMP-7, gene polymorphisms may be important in terms of the risk of the disease [28] and recurrence [63], while the relative levels of the inhibitors may also influence the impact of MMP-7 on PrCa progression [60]. An enforced expression of MMP-7 in PrCa cells has been shown to mediate an increased invasion [110], while a mouse MMP-7 knockout model exhibited reduced tumor-induced osteolysis [88], indicating the source of this enzyme may not be critical.

Membrane-Tethered Matrix Metalloproteinases (MT-MMPs)
Thirty-four articles were identified detailed the role of membrane-type MMPs in PrCa ( Table 2). The majority of these related to MT1-MMP (formerly MMP-14). There were conflicting reports about whether MT1-MMP was upregulated [38,65,158] or downregulated [78,112] in PrCa, which may be partially explained by studies describing its expression as being variable across the stages of PrCa progression [67,83,101,140], with PrCa cells eliciting altered MT1-MMP expression in surrounding noncancer cells [34,36,37,151]. However, functional studies have consistently shown MT1-MMP to contribute to a more invasive/migratory phenotype [158][159][160][161][162] and, potentially, tumor growth [163,164].  A single study reported that MT2-MMP is downregulated in PrCa [38], but in contrast, MT3-MMP expression was increased and correlated with enhanced aggressiveness/metastatic potential [38,166,167,172]. Likewise, MT6-MMP expression was generally observed to be increased in PrCa [155,165], including one study that indicated a correlation with the PrCa grade [115]. The sole functional study suggested that this MT-MMP also makes a contribution to enhanced invasion [155].

Tissue Inhibitors of Metalloproteinases (TIMPs)
The TIMPs represent direct regulators of the metzincin superfamily-particularly, members of the MMP subgroup (Table 3). Fifty-three studies investigated the role of TIMPs in PrCa progression, which collectively indicated that these proteins typically act to suppress PrCa progression. For TIMP-1, the expression was generally reduced in PrCa [66,82,95,123,146,177], including specifically in the transition from benign to neoplastic disease [25,55,178], and was also decreased in the recurrent [113] and metastatic [179] forms of the disease. However, some studies reported increased expression in more advanced/aggressive/malignant forms [20,38,48,60,68]. This difference may in part be due to its known upregulation by inflammatory cytokines [180] that might independently impact the expression in more advanced PrCa, as well as the mode of analysis, with the protein and mRNA levels not always in correlation [82].   For TIMP-2, the included studies typically reported a reduction in expression in PrCa [55,83,112,182,188], including a negative correlation of the expression to tumor grade [105,146] and metastasis [146], with promoter hypermethylation representing one mechanism by which the expression could be lost [188]. There were also a number of conflicting studies [116,151,185,190]. However, functional investigations have demonstrated that TIMP-2 administration reduced the tumor growth [192], and enforced TIMP-2 expression reduced the tumor invasion [194].
The publications on TIMP-3 provided a similar picture, with most showing a reduced expression in PrCa [20,55,189,196], including a negative correlation with the grade [115], and with promoter hypermethylation again representing a key mechanism [191], although a couple of studies were in disagreement with this interpretation [49,52]. The functional investigations were quite definitive, however, with the ablation of TIMP-3 in mice leading to enhanced tumor growth and invasion [181] and enforced expression decreasing the proliferation, survival, migration and invasion [198], as well as increasing apoptosis and chemosensitivity [183].
Finally, there were only two studies identified on TIMP-4, one of these demonstrating an increased expression in PrCa [20] and the other one indicating a negative correlation with the grade [115].

A Disintegrin and Metalloproteinases (ADAMs)
Twenty-two studies investigated members of the ADAM subgroup in the context of PrCa (Table 4). A number of these provided strong evidence of positive involvement in various aspects of the disease progression. Thus, ADAM-15 expression in PrCa positively correlated with the stage, grade, metastasis and recurrence, with its ablation decreasing both the migration and metastasis [190,199,200]. ADAM-17 expression was also shown to be significantly increased in PrCa and correlated with invasiveness, with ablation decreasing the proliferation and invasiveness [201,202]. ADAM-28 expression was similarly demonstrated to be higher in PrCa, with enforced expression enhancing the proliferation and migration [203].

Authors
Year ADAM

PrCa Platform Role
Arima et al. [204] 2007 ADAM-10 Yes-Prostate cell lines and prostate samples ADAM-10 nuclear localization significantly increased in PrCa compared to benign and correlated with Gleason score. Ablation of ADAM-10 decreased cell proliferation.
Bilgin Dogru et al. [205] 2014 ADAM-12 Yes-Prostate samples/urine Serum and urine ADAM-12 levels significantly higher in PrCa patients compared to healthy controls, but no correlation with stage.
Fritzsche et al. [207] 2008 ADAM-9 Yes-Prostate samples ADAM-9 expression significantly higher in PrCa compared to normal tissue, and associated with shortened PSA relapse-free survival, especially in androgen-ablated patients.
Hoyne et al. [208] 2016 ADAM-19 Yes-Prostate samples and prostate cell lines ADAM-19 expression decreased in PrCa compared to normal tissue, and positively correlated with lower grade and reduced relapse. Over-expression of ADAM-19 reduced proliferation and migration, but increased cell death.

Yes-cells and patients
Expression of ADAM-17 (but not ADAM-9 or ADAM-10) increased in PrCa compared to benign samples.
Kuefer et al. [190] 2006  The results regarding ADAM-9 were more complex, with one study showing no change in expression in PrCa [189] and others showing an increased expression that correlates with malignancy and reduced survival [207,216] but another reporting a decrease in expression in castrate-resistant compared to androgen-sensitive PrCa [210]. However, the ablation of ADAM-9 reduced the proliferation and tumor growth and increased the differentiation, decreasing the metastatic ability while increasing the sensitivity to chemotherapeutic drugs [209]. For ADAM-10, the nuclear localization rather than expression was increased in PrCa, with ablation decreasing the growth [204]. For ADAM-12, the serum levels have been demonstrated to be increased in PrCa, with expression found in stromal tissue, and progression delayed in knockout mice [205,214].
The clear exception in this family was ADAM-19, which was found to be more highly expressed in normal tissue compared to PrCa and negatively correlated to the grade and relapse, with the enforced expression leading to decreased proliferation, metastatic ability and survival [208].

A Disintegrin and Metalloproteinase with Thrombospondin Motifs (ADAMTSs)
Eight studies were identified that related to the ADAMTS subgroup in PrCa progression ( Table 5). The majority focused on ADAMTS-1, providing evidence of a tumorsuppressing function. ADAMTS-1 expression was typically decreased in PrCa samples, patients with metastatic disease, and a PrCa cell line variant with higher metastatic potential but elevated in slower-growing PrCa tumors in mice [65,179,219,220]. This was supported by functional data from the cells in which ADAMTS-1 expression had been modulated, which suggested a role in growth, although this appeared to depend on the cell line used [219]. ADAMTS-15 was also shown to be able to suppress tumor growth and migration, although it augmented survival [221]. Other members of the ADAMTS subgroup have also been shown to be expressed in PrCa cell lines, but their role in PrCa progression remains elusive [222].

Overview
This study used a systematic-like review strategy to identify publications examining the role of metzincins in PrCa progression. While limited to articles in PUBMED and MEDLINE and those written in English, this approach was likely to yield the vast majority of relevant research publications. It is evident from a close examination of the 205 articles identified that the contributions made by members of the metzincin superfamily to PrCa disease progression are complex. For many individual members and, indeed, the entire Astracin family, there is currently no evidence of involvement. However, a significant number of metzincins are positively associated with PrCa, supported by functional evidence in a number of cases, while others were negatively associated with this disease. The positive associations were particularly strong with specific members of the MMP, MT-MMP and ADAM subgroups, while those within the ADAMTS subgroup or the important TIMP family of regulators were more likely to show negative associations.

Positive Associations
The clearest evidence for positive contributions to PrCa and its progression was for MMP-2, MMP-7, MMP-9, MT1-MMP, ADAM-15, ADAM-17 and ADAM-28, with supporting evidence for MMP-1, MT3-MMP, MT6-MMP and ADAM-9 (Tables 1, 2 and 4). This is underpinned by studies that have identified associations between the expression and PrCa, which, in the case of MMP-7 expression [60,63,73,88,99,104,106,110,135,147] and ADAM-15 expression [190,199,200], correlated with the pathological stage and poorer outcomes for patients. This was supported by functional analyses that consistently identified enhancements in the proliferation, invasion/spread and metastasis/migration facilitated by them [56,106,110,200] (Figure 2). This identified these specific metzincins as likely tumor-promoting factors and so represented the obvious candidates as disease biomarkers or as potential targets for therapeutic agents.

Negative Associations
The strongest evidence for negative contribution to PrCa is for TIMP-2 and TIMP-3, as well as ADAM-19, ADAMTS-1 and ADAMTS-15 (Tables 3-5). Such a role for the ADAMTS proteins is somewhat counterintuitive, since these enzymes cleave ECM components like other metzincins [224], including those involved in PrCa disease progression [220,221]. However, the functional evidence points to these enzymes inhibiting key phenotypes, including proliferation and metastasis/migration, although not survival (Figure 2), presumably due to the different specificities for ECM components compared to other metzincins [220,221]. ADAM-19 was also implicated in the proliferation, metastasis/migration and survival (Figure 2), although this could relate to the known nonenzymatic functions for these enzymes. A negative role for the TIMP family was less surprising, given their primary role in the inhibition of MMP enzymes [225], with this impact extending across the entire gamut of relevant cell functions (Figure 2). These molecules can also be considered biomarker candidates to aid in prognosis. Therapeutic approaches targeting these proteins would likely be more limited, however, since they would need to augment, rather than inhibit, their function.

Mixed Associations
For other metzincins, the evidence for their involvement in PrCa was even more variable and contradictory, such as for MMP-11. Indeed, even those metzincins or regulators with consistent positive or negative correlations with PrCa were often reported in some studies to have no correlation or, indeed, the opposite correlation. This suggests a complex interplay between metzincins and PrCa.

Understanding the Complexity
In interpreting the variable and, at times, conflicting data, there are a number of factors that need to be considered. Firstly, different studies have utilized alternate approaches, such as analyzing the expression at the gene versus protein levels, that do not always correlate [24,68,82,83,91,101,140,189] or examining the enzymatic activity, which is not always reflective of metzincin expression [119,226], or, instead, considering the cellular localization [140]. Moreover, different samples have been analyzed in the literature, including plasma/serum, urine and tumor biopsies from PrCa patients, with several studies highlighting the differences between tissues [119], while the exact PrCa stage is also critical [117]. Other studies have employed PrCa cell lines and xenotransplanted tumors in mice, the relevance of which to human disease is assumed but not guaranteed. Secondly, it is clear that the factors controlling the expression of these enzymes and their regulators are complex. Thus, many metzincins have been demonstrated to be regulated by androgens [107,213,217,223,227,228], which can clearly be a complicating factor given the environment in which these cancers develop. In addition, expression is also impacted by oncogenes [85,147], inflammation and inflammatory cytokines [99,229], as well as angiogenic factors such as vascular endothelial factor [216], which are intrinsic features of any cancer. The cellular environment can further influence both expression [127] and activation [143]. Therefore, discerning the direction of causality between the expression and PrCa is not always straightforward.
In most cases, the effects of the metzincin superfamily member (or inhibitor) have been presumed to relate to the primary role for metzincins in regulating components of the ECM, which is known to be a particularly key element of metastasis [9,13]. However, which substrates are important? The cleavage of laminin [159,176], perlecan [56] and beta-4 integrin [230] have all been shown to correlate with the effects of protumorigenic metzincins, particularly on metastasis, whereas versican has been identified as a target of the antitumorigenic ADAMTS-15 [221]. Clearly, more research is required to understand this important aspect of metzincin pathobiology. Moreover, other roles should also be considered, especially given reports suggesting that nuclear localization may be important in some situations [204,213,216], with both ADAMs and ADAMTSs known to have nonenzymatic roles.
There also remains a lack of depth in our understanding of how metzincins are regulated at the protein level, including by other metzincins. TIMPs are clearly important for the negative regulation of MMPs [114]. TIMPs are typically downregulated as cancer progresses and can act as independent correlates of PrCa progression [25,32,95,112,114,116,[177][178][179], especially when combined with MMP expression [83,105,116]. TIMP-2 and TIMP-3 have also been shown to inhibit ADAMTS-1 [185]. Are there equivalents for ADAM and other ADAMTS enzymes? In addition, MT1-MMP has been shown to exert its role at least in part through the direct activation of MMP-2 [140,159]. Is this crosstalk common across metzincins? More research is needed to gain further insight in this area.

Materials and Methods
This study represents a systematic-like review of the role of the metzincin superfamily of proteases in PrCa progression. The search terms were identified through a PCC (population, context and concept) format by the research team with keywords, Boolean operators, truncations and Medical Subject Headings (MeSH) used to develop a database search strategy in collaboration with a specialist health librarian. In reporting the review, the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) was utilized.

Search Strategy
A preliminary search was undertaken using MEDLINE and then a full search run through both the PUBMED and MEDLINE databases.

Inclusion and Exclusion Criteria
All studies were considered based on the inclusion and exclusion criteria shown in Table 6. Search terms for inclusion were "metzinzin", "metalloproteases", "metalloproteinase", "MMP", "TIMP", "ADAM", "ADAMTS", "BMP1" or "meprin" and "neoplasm", "neoplasia", "cancer", "tumor" or "cysts". Reasons for exclusion after the full-text review are detailed in Figure 3. No restrictions were put on the date that articles were published. Table 6. Search terms and inclusion/exclusion criteria used in this systematic-like review.

Inclusion Criteria Exclusion Criteria
Population Men, mice/rats or cells. Involved other animal species or cells not related to PrCa progression.

Concept
All human metzincin superfamily members, including members of the MMP, MT-MMP, ADAM, ADAMTS, BMP1/TLL and Meprin subgroups, as well as the TIMP family of regulators.
Examining proteins other than metzincin superfamily members or TIMPs, or in different cancer types, other diseases, or normal biology.

Context
Studies that investigated the role of the metzincin superfamily in PrCa progression.
Did not specially look at the role of metzincin superfamily members in PrCa cancer progression.

Study Selection and Data Extraction
Searches of the published literature were conducted by M.J.B. in collaboration with a specialist health librarian. Titles and abstracts were retrieved from the search and screened. Full-text article review and data extraction was then conducted, with the reasons for exclusion documented. The reference lists of the included articles were also reviewed to identify further potential articles for inclusion in the review.

Data Analysis
Database searching identified 10,443 publications. After duplicate removal, the titles and abstracts from 8327 were reviewed against the inclusion criteria. Full-text versions of 1248 articles were then further reviewed, identifying 205 articles for inclusion ( Figure 2). The reasons for exclusion were a lack of focus on PrCa (n = 603) or the metzincin superfamily (n = 50) or the role of the metzincin superfamily family in PrCa progression (n = 98) or on the biology of the metzincin superfamily (n = 108) or not peer reviewed (n = 18) or being review articles (n = 173) or articles unable to be accessed or retracted (n = 11) or not in English (n = 3). The 205 included articles covered members of the Matrixin family subgroups MMP and MT-MMPs, the TIMPs and the Adamalysin family subgroups ADAMs and ADAMTSs, but there were none regarding the Astracin family subgroups BMP/TLL or Meprin.