Fluctuating Roles of Matrix Metalloproteinase-9 in Oral Squamous Cell Carcinoma

One hallmark of cancer is the degradation of the extracellular matrix (ECM), which is caused by proteinases. In oral cancers, matrix metalloproteinases (MMPs), especially MMP-9, are associated with this degradation. MMPs break down the ECM allowing cancer to spread; they also release various factors from their cryptic sites, including cytokines. These factors modulate cell behavior and enhance cancer progression by regulating angiogenesis, migration, proliferation, and invasion. The development of early metastases is typical for oral cancer, and increased MMP-9 expression is associated with a poor disease prognosis. However, many studies fail to relate MMP-9 expression with metastasis formation. Contrary to earlier models, recent studies show that MMP-9 plays a protective role in oral cancers. Therefore, the role of MMP-9 is complicated and may fluctuate throughout the different types and stages of oral cancers.


Introduction
Oral cancer is one of the ten most common cancers worldwide. Nearly 3% of all cancer cases are oral cancers; they are more common in men than in women; and two-thirds of oral cancer cases occur in developing countries [1,2]. One important hallmark of cancer progression is the degradation of the extracellular matrix (ECM), which allows cancer cells to invade the surrounding tissue. Matrix metalloproteinases (MMPs) are zinc-dependent endopeptidases that efficiently degrade the components of the ECM and basement membranes (BM). MMPs also release cytokines, chemokines, and growth factors from their proforms or their cryptic sites [3][4][5][6]. To date, at least 24 distinct MMP genes have been identi�ed in humans. MMPs are classi�ed according to their substrate speci�cities: gelatinases, collagenases, matrilysins, and stromelysins. e structures of all MMPs include an Nterminal signal peptide that directs the protein to either the plasma membrane insertion or to the secretory pathway; its prodomain confers its latency, and its catalytic domain has a zinc atom in its active site. MMPs are either anchored in the membrane or secreted, primarily as latent proforms that require activation before becoming catalytically competent [7][8][9]. Two different soluble gelatinases have been identi�ed: gelatinase A, 72 kDa (MMP-2), and gelatinase B, 92 kDa (MMP-9). Both contain a collagen-binding domain within their catalytic domain, distinguishing them from other MMPs. A more detailed structure of these enzymes is described in a review by Björklund and Koivunen [10]. 2. Activation of MMP-9 Typically, gelatinases are secreted as inactive zymogens that become activated extracellularly. e most relevant natural activators of proMMP-9 are unknown, but proMMP is activated through a few different mechanisms, including proteolytic activation, where the prodomain is cleaved yielding an active enzyme. Latent MMP-9 can be activated by MMP-3, which cleaves proMMP-9 at multiple sites: the �rst cleavage site is Glu 59 -Met 60 ; the second is Arg 106 -Phe 107 [11]. In contrast, MMP-26 activates MMP-9 by cleaving at Ala 93 -Met 94 [12]. Previous studies have also demonstrated that enterokinase, a membrane-bound serine protease, cleaves proMMP-9 at Lys 65 -Ser 66 [13] and that trypsin-2 activates proMMP-9 at very low molar ratios, 1 : 1000. e peptide bond can also be cleaved at Arg 87 -Phe 88 [14]. Other known proteolytic activators are plasmin, chymotrypsinlike proteinase, MMP-2, MMP-7, MMP-10, and MMP-13 [15][16][17][18][19][20]. ere are other identi�ed activation mechanisms for MMP-9: oxidation by reactive oxygen species, Snitrosylation, and allosteric activation, which occurs when proMMP-9 is bound to either a gelatin or type IV collagen [21][22][23]. In an invasive tongue squamous cell carcinoma cell line (HSC-3), MMP-9 is colocalized with trypsin-2 in intracellular vesicles [13]. is intracellular activation may be an alternative activation mechanism for proMMPs in oral cancers. Similar intracellular vesicle transports for MMP-9 are also found in melanoma cells and in ovarian cancer ascites [24,25]. In oral squamous cell carcinoma (OSCC), the activation level of MMP-9 may be associated with a shortened disease-free survival and a high metastatic frequency [26].

Inhibitors of MMP-9
Tissue inhibitors of metalloproteinases (TIMPs) are speci�c endogenous inhibitors of MMPs, which bind MMPs in a 1 : 1 stoichiometry. Four different TIMPs have been identi�ed: TIMP-1, TIMP-2, TIMP-3, and TIMP-4 [27]; they all inhibit MMP-9 in vitro [28][29][30][31]. e role of TIMPs in OSCC is wellstudied and is the focus of a recent review by Garćia et al. [32]. Aer the role of MMPs in cancer invasion and metastasis formation was recognized, researchers began developing synthetic inhibitors. e �rst generation of peptidomimetic MMPs, batimastat (BB94) and ilomastat (GM-6001), mimicked the structure of collagen and reversibly bound the active site of MMPs to inhibit MMP activity [33,34]. Next, Marimastat (BB-2516), a second generation of MMP inhibitors, was developed. However, all of these broad-spectrum inhibitors failed in clinical trials due to their side effects and their lack of efficacy [35][36][37], which led to the development of more selective matrix metalloproteinase inhibitors (MMPIs). We developed the �rst gelatinase-speci�c MMPI CTTH-WGFTLC peptide, which inhibited the invasion of ovarian carcinoma, breast carcinoma, �brosarcoma, Kaposi�s sarcoma, and melanoma cell lines, in vitro; it also increased the survival of human tumor xenogras [38]. Since then, we have developed a cyclic gelatinase-speci�c inhibitor GRENYHGCTTHWGFTLC peptide, which inhibits MMP-9 activation and activity, the growth of human tongue cancer cell xenogras, and angiogenesis in nude mice [39]. e synthetic MMPIs that inhibit MMP-9 are listed in Table 1.

MMP-9 in the Oral Microenvironment
In addition to carcinoma cells, cancers consist of tumor-associated stromal cells, which include �broblasts, endothelial cells, leukocytes, macrophages, nerve cells, and adipocytes. During cancer progression, the cancer cells crosstalk with stromal components and their interactions are partially mediated by transmembrane receptors, which are expressed on cancer cells and stromal cells. Tumor-associated cells promote angiogenesis, in�ammation, invasion, and ECM e Scienti�c World �ournal 3 modeling through cell-cell contact and the production of growth factors, hormones, cytokines, and proteinases such as MMPs [56][57][58]. In OSCC tumors, MMP-9 is expressed in carcinoma and in�ammatory cells around carcinoma islands. Meanwhile, MMP-2 is mainly found in carcinoma-associated �broblasts (CAFs) [59,60]. In an oral squamous cell carcinoma cell line SCC-25, CAFs increase the expression of MMP-9, in vitro, which is thought to occur via a �bronectinintegrin v 6 pathway [61]. In the aggressive human tongue squamous cell carcinoma cell line HSC-3, MMP-2 was only found in its latent form, whereas MMP-9 was found in its active form [13]. MMP-9 ′ s effect during HSC-3 cell line invasion was studied in a human organotypic model based on myoma tissue [62], which contains �broblasts, smooth muscle cells, lymphocytes, macrophages, endothelial cells, and MMP-2, but not MMP-9 [62]. erefore, this model is a better predictor of the in vivo tumor microenvironment compared with the commonly used rat tail-derived type I collagen and/or the mouse EHS sarcoma-derived Matrigel invasion assay. In the myoma organotypic invasion assay, aer inhibiting gelatinase activity in HSC-3 cells using a speci�c gelatinase inhibitor CTTHWGFTLC [38], the tumor cells were surprisingly more invasive than in the control group (unpublished data). Mice bearing HSC-3 xenogra tumors treated with the gelatinase inhibitor CTTHWGFTLC had smaller primary tumors in vivo than the control group [39], but the inhibition of gelatinases did not affect local invasion or metastasis formation [63]. e ability of cancer cells to change their migration under certain circumstances from proteolytic to non-proteolytic, amoeboid type during protease-inhibitor treatment helps to explain the OSCC behaviors we observed. us, these cells change their shape and adapt to squeeze through tissue gaps without degrading the ECM [64]. MMP-9 may not be the only, or even the most important, proteolytic enzyme in the OSCC invasion process, but it may be important for indirect cell signaling by controlling the bioavailability and bioactivity of molecules that target speci�c receptors, which regulate cell growth, migration, in�ammation, and angiogenesis [65][66][67][68][69].

The Role of MMP-9 in OSCC Invasion and Metastasis
MMP-9 is associated with the aggressive nature of many cancers, including OSCC [81][82][83][84], and this aggressive nature was thought to cause type IV collagen degradation, a main component of basement membranes [85]. To date, the spectrum of MMP-9 matrix substrates has signi�cantly increased, and aside from substrates, which originate in the matrix, MMP-9 has other bioactive substrates that independently modulate carcinogenesis, such as the pro-transforming growth factor-1 (TGF-1) and the pro-tumor necrosis factor-(TNF-) [10,86,87]. MMP-9 has traditionally been associated with the aggressive nature of OSCC. However, in spite of increased MMP-9 expression levels, many researchers have presented contradictory results [70][71][72] (Table 2). For example, Guttman et al. [73] did not �nd a correlation between MMP-9 expression and the size of the primary tumor or the neck metastasis in tongue SCC patients. Meanwhile, another study reported that high levels of MMP-9 expression in OSCC patients were correlated with regional lymph node and/or distant metastases and a poor prognosis [74]. In addition, De Vicente et al. [78] showed that MMP-9 expression was not associated with clinical variables, such as tumor stage or recurrence rate. In a study conducted by Ikebe et al., gelatinolytic activity and increased expression of both MMP-2 and MMP-9 in OSCC tumors were related to the invasiveness, but not to the metastatic potential of OSCC tumors [75]. Finally, Kato and co-workers [76] showed that, although MMP-9 expression was high in OSCCs, the activated form:proform ratio was very low, while activated MMP-2s were elevated and associated with advanced stages of disease. ese �ndings suggest that MMP-9 may not be a universal cancer progression promotion factor in OSCCs; instead, it may have �uctuating roles.

MMP-9 in the Modulation of Can�e�-Related In�a��ation
Chronic in�ammation is associated with epithelial cancers, and it differs from normal in�ammation because it is not self-limiting. Cancer cells produce different cytokines that attract innate immune cells, such as mast cells, granulocytes, and macrophages. ese innate immune cells then secrete interleukins, chemokines, reactive oxygen species, and MMPs that modulate angiogenesis, cell proliferation, tumor growth, and invasion [88][89][90]. In OSCCs, the level of a multifunctional cytokine, transforming growth factor-1, is upregulated, which leads to the enhanced expression of snail. Snail is a transcription factor that increases MMP-9 expression and triggers an epithelial-mesenchymal transition (EMT); then, carcinoma cells change their morphology, reduce their intercellular and cell-matrix adhesions, and increase their motility [91,92]. Interestingly, the inactive form of TGF-1 is activated by MMP-9 [93]. Many other cytokines are also substrates for MMP-9, including TNF-, CXCL1, CXCL4, CXCL7, CXCL8, and interleukin-1 [65][66][67]. Interleukin-1 is secreted by tumor cells and induces the expression of lipocalin 2 [94,95]. e plasma levels of lipocalin 2, MMP-9 and the lipocalin 2/MMP-9 complex are associated with more advanced clinical stages and/or tumor sizes in OSCC patients. Interestingly, MMP-9 levels are not correlated with either lymph nodes or distant metastases [77]. Chemokine CXCL8 can induce the release of MMP-9 from tertiary neutrophil granules, and increased CXCL8 expression is associated with OSCC. e CXCL8 expressed in tumor cells is also secreted by OSCC cell lines, and CXCL8 mRNA expression is enhanced by the addition of TNF-and IL-1 . CXCL8 from OSCC cell lines increases cell migration, induces invasion, and increases the expression of MMP-7. However, it does not have an effect on MMP-9 expression. erefore, CXCL8-induced expression of MMP-9 may be cell-type speci�c [96][97][98]. e chemokine receptor, CXCR4, modulates the invasion of OSCCs by regulating MMP-9 expression. In patients, this expression correlates with lymph node metastasis and MMP-9 expression [99,100]. Cyclooxygenase (COX)-2 is an enzyme that converts arachidonic 4 e Scienti�c World �ournal acid into pro-in�ammatory prostanoids. COX-2 has been implicated in carcinogenesis and its mRNA expression is nearly 150-fold greater in head and neck squamous cell carcinomas compared with normal oral mucosa. Using a selective COX-2 inhibitor decreases MMP-9 and MMP-2 expression and suppresses the proliferation and invasion of OSCCs [101,102]. In fact, MMP-9 has many links to the cancer-related in�ammation observed in OSCCs, but MMP-9 ′ s speci�c role in this process remains unclear.

MMP-9 in the Regulation of Angiogenesis
Low oxygen levels, or hypoxia, are typical in solid tumors that grow 1-2 mm 3 without vascularization. Angiogenesis, the formation of new blood vessels, is required to bring nutrition and oxygen to cells, remove metabolic waste, and support larger tumor growth. Not only is angiogenesis associated with tumor growth, it is also related to the development of metastases in OSCCs [103][104][105][106][107]. is process is initiated by vascular endothelial growth factor (VEGF), an angiogenic cytokine. e effect of VEGF is mediated by vascular endothelial growth factor receptors. e hypoxic conditions stabilize hypoxia-inducible factors (HIFs) that bind to the VEGF promoter, which causes upregulation of VEGF and increases the expression of VEGFR-receptor-1 in endothelial cells, cancer cells and tumor-associated cells, such as macrophages [105,[108][109][110][111]. In OSCCs, overexpression of HIF-1 is associated with a poor patient outcome [112]. Increased tumor hypoxia is also associated with increased MMP-2 and MMP-9 activity [113]. e regulation of angiogenesis is a very delicate balance between pro-and antiangiogenic factors, and it seems that MMP-9 plays a dual role in this process. It can act as a proangiogenic factor via VEGF regulation; as Hiratsuka et al. [114] demonstrated, primary tumors in premetastatic lungs induce MMP-9 expression in a VEGFR-1-dependent manner, which enhances the invasion of cancer cells and facilitates metastasis. MMP-9 also triggers the angiogenic switch by releasing VEGF [87]. MMP-9 and VEGF are expressed during invasive OSCCs of the tongue and in metastatic tumors that tend to express higher levels of VEGF and MMP-9, than nonmetastatic tumors [115]. In another study, increased VEGF expression was associated with a poor prognosis in OSCC patients, whereas MMP-9 expression levels had no correlation with patient outcomes [116]. is can be explained by the antiangiogenic role of MMP-9, which causes cleavage of type XVIII collagen, and leads to the release of endostatin, a potent inhibitor of angiogenesis and endothelial cell migration [68,69]. OSCC primary tumors that do not metastasize have high endostatin levels compared with primary tumors that are associated with multiple metastatic lymph nodes. Full-length collagen XVIII expression levels are decreased in aggressive tumors [117]. Based on our study [118], collagen XVIII was expressed in mild oral epithelial dysplasias but was absent in the invasive fronts of OSCCs. Although MMP-9 expression was observed in these same samples, there was no correlation between MMP-9 and the stage of disease. Homer et al. [119] reported that in head and neck squamous cell carcinoma, the plasma levels of endostatin may predict tumor recurrence. Collagen XVIII and VEGF are both expressed at the same time in OSCCs, which further demonstrates that modulation of angiogenesis requires a delicate balance between angiogenic inhibitors and stimulators [109]. Moreover, we have demonstrated that endostatin also directly inhibits the invasion and intravasation of a human tongue SCC cell line and blocks the activity of proMMP-9, suggesting a feedback loop for MMP-9 regulation [120]. When mice e Scienti�c �orld �ournal bearing nasopharyngeal carcinoma tumors were treated with endostar (recombinant endostatin), the treatment led to a signi�cant decrease in MMP-9 and VEGF expression levels and normalized the tumor vasculature [121]. Endostatin was not the only cryptic angiogenesis inhibitor liberated by MMP-9. Angiostatin, a fragment of plasminogen, also inhibits endothelial cell proliferation [86,122]. Pozzi et al. [123] proved that decreased MMP-9 plasma levels in tumor-bearing mice also decreases angiostatin levels, which increases tumor vascularization and growth. Matsumoto et al. [124] observed that angiostatin-overexpressing SCC tumors in mice grew slower than the control tumors. Aside from endostatin and angiostatin production, MMP-9 is also associated with the proteolytic degradation of type IV collagen, which produces an angiogenesis inhibitor, tumstatin [125,126]. ese �ndings suggest that in cancer angiogenesis the variations in spatial and temporal MMP-9 expression may switch between two roles: from a proangiogenic to an antiangiogenic molecule (Figure 1).

Effect of Genetic and Environmental Factors on the Expression of MMP-9
e development and progression of OSCCs are a result of interactions between accumulating genetic alterations and environmental factors, such as alcohol, tobacco, viral infection, or chronic in�ammation [127]. Despite newer cancer treatments, approximately 50% of patients die within 5 years of diagnosis [128]. is can be partially explained by the theory of oral �eld cancerization: an oral mucosa exposed to carcinogens, such as alcohol, causes multiple genetic abnormalities in the entire epithelium and increases the risk of developing several dysplastic lesions [129]. Polymorphisms in the MMP9 gene allele are associated with an increased risk of developing the initial stages of oral cancer among patients without a family history of cancer and high smoking and/or alcohol use [130]. MMP-9 expression did not correlate with age, gender, tumor location, or smoking habits, whereas an association with tumor grade differentiation and alcohol consumption was observed [78]. MMP-9 was not expressed in the normal oral mucosa or dysplasia, whereas in situ carcinomas were weakly detectable. In OSCC, it was expressed in the same areas where collagen (IV) chain loss was observed at the invasive fronts, whereas in other studies, its overexpression was detected in 85% of oral dysplasias and in all of the oral cancer samples. e mRNA levels of MMP-9 were higher in oral dysplasia that progressed to oral SCC [131,132]. Ogbureke et al. 2012 [133] proposed that MMP-9 expression at histologically negative surgical margins could predict OSCC recurrence. Interestingly, MMP-9 was absent from the margins of tumors ≤4 cm and only 10% of tumors without later node metastasis expressed it. MMP-9 expression alternates between different stages of malignant transformations. Different clinicopathological variables in OSCCs can partially be explained by viral infections in epithelial cells. e human papilloma virus (HPV) is associated with OSCCs, especially type 16. HPV16 transgenic mice, HPV16/MMP-9 +/+ and HPV16/MMP-9 +/− develop a similar incidence of SCC aer 12 months of age, whereas HPV16/MMP-9 −/− mice have fewer tumors, but these tumors were more poorly differentiated than those in the other groups. e expression of HPV16 oncoproteins in human keratinocytes induced the upregulation of MMP-9 activity. At the same time, two natural MMP inhibitors were downregulated. One of them was the reversion-inducing cysteine-rich protein with Kazal motifs (RECK), which affects transcription, synthesis, activation, and the activity of MMPs; the other was TIMP-2 [134][135][136]. HPV16 infection may be one mechanism behind the contradictory expression of MMP-9 in OSCC patients; however, further studies are necessary to understand its signi�cance.

Methods for MMP-9 Investigation in Oral Cancer
e most commonly used immunohistological analyses of MMP-9 expression can be misleading because most antibodies do not distinguish between the pro-and active forms of the protein. e total amount of protein expression does not necessarily mean that the enzyme is in an active form [137]. Gelatin zymography or in situ zymography [138,139] are better methods to evaluate the level of gelatinase activity and would have provided more information in the studies referred to here (Table 2) [139]. Many other techniques can investigate the presence, amount and function of MMP-9. For example, in situ hybridization could de�ne the location and number of cells that express MMP-9 mRNA in tissue sections, whereas the polymerase chain reaction (PCR) could detect the presence and amount of mRNA in tissues or cell extracts [140,141]. More speci�c intracellular localization of MMP-9 could easily be achieved using confocal laser scanning microscopy, which enables a 3-dimensional de�nition of protein localization [13]. Enzyme-linked immunosorbent assays (ELISAs) could be used to determine the concentration of MMP-9 protein in serum and plasma samples; based on these studies, MMP-9 is a systemic biomarker that monitors the effectiveness of OSCC treatment [112,141,142]. Different blotting methods, Western, Northern, and Southern, reveal the expression of protein, RNA or DNA, respectively, in various samples [75,143]. MMPIs, transfected cell lines, and knock-out animal models give more functional data about MMP-9 [144,145]. Most likely, the variety of methodologies and the low sample sizes explain the high variation observed among the previous OSCC results. erefore, more studies with larger well-documented clinical materials, using delicate methods, are necessary to determine the impact of MMP-9 in OSCC.

Conclusions
Generally MMP-9 has been associated with aggressive head and neck cancers, but novel studies have shown that it acts as a protective molecule during carcinogenesis and metastasis. For example, in salivary gland myoepithelial carcinoma, MMP-9 expression predicts a better overall survival, and in regional metastases of head and neck cancers, MMP9 gene expression was decreased [79,146]. Similarly, the expression of MMP-9 is associated with a better outcome in breastand colitis-associated carcinomas [147,148]. Additionally, in oral cancer the role of MMP-9 was purely associated with the degradation of the ECM, which led to the enhancement of carcinoma cell invasion. However, the philosophy of oral carcinoma progression has become signi�cantly more complicated; now, MMP-9 is known as a multifunctional modulator that is involved in very complex cell-signaling cascades ( Figure 2). erefore, in the case of MMP-9, one reason for the obvious failures of broad-spectrum and speci�c MMPIs in cancer treatment might be due to its �uctuating role in cancer, which not only affects carcinoma cells but also other cell populations. e tumor microenvironment matrix expresses and sequesters MMP-9. Taken together, our current knowledge of MMP-9 has been extended; it can act as either a carcinoma protector or promoter depending on the speci�c situation, which is related to patient characteristics, including the stage, grade, and location of the tumor.