MT3‐MMP down‐regulation promotes tumorigenesis and correlates to poor prognosis in esophageal squamous cell carcinoma

Abstract The membrane‐type matrix metalloproteinases (MT‐MMPs) play an important role in degrading the extracellular matrix (ECM) and facilitating protease‐dependent tumor progression and invasion. Here, we report that unlike MT1‐MMP, MT3‐MMP was down‐regulated in esophageal squamous cell carcinoma (ESCC) as detected by real‐time PCR (qPCR), Western blot analysis, and immunohistochemistry (IHC). Down‐regulation of MT3‐MMP was observed at protein level in 66.3% of ESCC specimens (by IHC, n = 86) for routine pathologic diagnosis, as well as at mRNA level in 63.3% of surgically resected ESCC tumors paired with surrounding nontumor tissues (by qPCR, n = 30). Notably, MT3‐MMP down‐regulation significantly correlated with lymph node metastasis and poor overall survival of patients with ESCC (median 5‐year survival = 50.69 vs. 30.77 months for patients with MT3‐MMP‐negative and ‐positive ESCC, respectively). Mechanistically, MT3‐MMP negatively regulated proliferation, colony formation, and migration of ESCC cells, in association with cell cycle arrest at G1, due to up‐regulation of p21Cip1 and p27Kip1. Together, as a tumor suppressor in ESCC, MT3‐MMP down‐regulation represents an unfavorable factor for prognosis of patients with ESCC.


Introduction
MMPs represents a family of zinc-dependent multidomain endopeptidases, which are responsible for degradation of virtually all structural components of extracellular matrix (ECM), as well as numerous bioactive molecules, thereby playing essential roles in a variety of physiological and pathological events, especially cancer metastasis [6,7]. MT3-MMP or MMP16, as a membrane-anchored matrix metalloproteinase, is a major mediator of pericellular matrix proteolysis involved in the remodeling of extracellular matrix, either directly through execution of proteolysis or indirectly by activating other enzymes [8]. For example, MT3-MMP can activate pro-MMP-2 [9], while MMP-2 is known to promote cancer cell invasiveness [10]. It is also recognized that MT3-MMP has stronger activities in degrading extracellular matrix than MT1-MMP, partially due to activation of MMP-2 [11]. Moreover, MT3-MMP is critical for governing the transition of tumor cells to an invasive phenotype [8,[12][13][14][15].
However, the initial notion that MMPs act as metastasispromoting enzymes has been challenged by most recent findings [16]. Despite the established pro-tumorigenic roles of certain MMPs (e.g., MMP2, MMP9, and MT1-MMP), recent studies have demonstrated that some MMPs such as MMP8 and MMP11 might act against tumor growth and metastasis. For example, MMP8-deficient mice, when challenged with carcinogens, display a markedly increased susceptibility to tumorigenesis, compared to wild-type mice [17]. Further, it has been shown that higher MMP8 levels correlate to lower risk of distant metastasis, as well as better prognosis of patients with breast or oral cancer [16]. Similar antitumor effects or dual (i.e., anti-and pro-tumorigenic in a context-specific manner) functions have also been found in the case of several other MMPs, including MMP11, MMP12, MMP19, and MMP26.
To date, the expression status and clinical role of MT3-MMP in ESCC remains virtually unknown. Here, we report that MT3-MMP was down-regulated in ESCC tumor tissues, which correlates to high metastasis rate and poor survival; therefore, representing a favorable factor for prognosis of patients with ESCC. These findings were further validated mechanistically using ectopic overexpression and shRNA knockdown of MT3-MMP in an in vitro model of human ESCC.

Cell lines
The human ESCC cell lines EC109 and EC9706 were obtained from the Chinese Academy of Medical Science (Beijing, China). Cells were cultured in RPMI-1640 medium (GIBCO, Carlsbad, CA) containing 10% heat-inactivated fetal calf serum (GIBCO, Carlsbad, CA) in a humidified incubator at 37°C in the presence of 5% CO 2 .

Patients and primary tumor specimens
A total of 86 patients diagnosed as ESCC between July 2006 and October 2008 were enrolled into this study at the First Affiliated Hospital of Xiamen University and Xijing Hospital. Patient clinicopathologic characteristics are described in Table 1. The median age was 56 years (ranged from 32 to 73 years). Among them, 78 patients were followed up till the end of this study, that is, till September 2013, while eight patients were lost to follow-up. Disease-specific survival of patients was defined as a period of time from the date of diagnosis to the date of cancerrelated death or the end of the study. Primary tumor specimens for immunohistochemistry (IHC) analysis were obtained from patients undergoing routine procedures of pathologic diagnosis.
Among 86 patients, thirty underwent esophagectomy. None of them had received preoperative chemotherapy. Fresh surgical specimens of primary ESCC tumor and surrounding nontumor (~5 cm away from the tumors) tissues were collected, stored in liquid nitrogen, and later used for qPCR and Western blot analyses. The pathologists evaluated each cancer specimen.
This study was approved by the Ethical Committee of the First Affiliated Hospital of Xiamen University. The informed consents were signed by all patients.

Western blot analysis
Tumor and nontumor tissues obtained from the surgical specimens of 30 ESCC patients, as well as cells of the ESCC cell lines, were lysed by ultrasonication and then incubated with radio-immunoprecipitation assay (RIPA) buffer. After loading the proteins (30 μg) onto an 8% SDS-PAGE, electrophoresis was performed at 20 mA for 60 min under denaturing conditions, and the proteins were then transferred to a nitrocellulose membrane. The membrane was incubated in 5% fat-free milk for 1 h. 30 μg of protein per condition was subjected to SDS-PAGE and then electrotransferred onto nitrocellulose membrane. Anti-MT3-MMP antibody (ZA-0266, 1:100, Boster) was used as primary antibody. Horseradish peroxidase (HRP)-conjugated anti-rabbit or anti-goat antibodies (1:5000; Sigma-Aldrich) were used as secondary antibody. Blots were visualized by an enhanced chemiluminescence (ECL) system (Amersham Pharmacia Biotech, Arlington Heights, IL). Each blot was repeated three times.

Real-time PCR (qPCR)
Total tissue RNA was isolated from 30 surgical specimens using the TRIzol reagent (Invitrogen) as recommended by the manufacturer, and DNase was used to block the contamination of genomic DNA. cDNA was synthesized from 2 μg of total RNA per condition using a PrimeScript Reverse Transcription Reagent Kit and SYBR premix Ex Taq (Takara, Dalian, China). PCR amplification was then carried out as follows: initial denaturation at 95°C for 2 min, followed by 45 cycles of 95°C for 15 sec, 56°C for 20 sec, and 72°C for 15 sec. Reference for quantitation was human GAPDH as housekeeping gene. The primers include: MT3-MMP, forward 5′-TTCGTCGTGAGATGTTTGT-3′, reverse 3′-CCCGCCAGAAGTAAGTAA-5′; GAPDH, forward 5′-GCACCGTCAAGGCTGAGAAC-3′, reverse 3′-TG GTGAAACGCCAGTGGA-5′.

Cell cycle analysis
Cells were seeded onto 60-mm-well plates and incubated overnight in complete medium, followed by incubation for 48 h in serum-free medium to synchronize cells, and then in complete medium to release from synchronized block. At 24 h of release, cells were harvested, washed with ice-cold PBS, fixed with 70% ethanol, and then stained with 20 μg/mL propidium iodide (PI) containing 200 μg/mL RNase A (Sigma). Flow cytometry was then carried out to determine cell cycle distribution.

Colony formation and wound healing assays
Colony formation assays were performed to measure the survival and proliferation of cells. Briefly, one thousand cells were seeded onto 6-cm dishes and cultured for 2 weeks. After fixing with 70% ethanol and stained using Giemsa Stain, colonies were counted under a microscope. Alternatively, to further measure colony formation in anchorageindependent condition, cells were plated in 0.3% soft agar Z. Xue et al.

MT3-MMP Suppress ESCC
over a base of 0.5% agar in complete medium. After culturing for 3 weeks, the number of colonies > 100 μm was counted.
Wound healing assay was conducted to assess cell migration. Briefly, EC109 cells were grown to create a confluent monolayer and then scraped with a p200 pipet tip to create a scratch. Images for three randomly selected fields per condition were captured at the indicated intervals (0-72 h) and then analyzed quantitatively using a computing software.

Statistical analysis
Values represent the means ± SD for at least three independent experiments performed in triplicate. Significance of differences between variables was determined using the chi-square, Fisher's exact, and Wilcoxon rank-sum tests. P < 0.05 was considered statistically significant. Kaplan-Meier survival analysis and multivariate analysis (Cox proportional hazards regression model) were used to analyze overall survival of patients. All data analyses were conducted using the SPSS 21.0 software package (Chicago, IL).

MT3-MMP is down-regulated in primary ESCC tumors
To examine the potential relationship between MT3-MMP expression and disease progression of ESCC, qPCR was first performed to monitor mRNA expression of MT3-MMP in ESCC tumors and their paired surrounding nontumor tissues of 30 surgical specimens. qPCR revealed that MT3-MMP was down-regulated in 19/30 (63.3%) of primary ESCC tumors, compared to their nontumor counterparts (P < 0.05, Fig. 1A). In some representative cases whose MT3-MMP mRNA levels were markedly higher in nontumor than tumor tissues, Western blot analysis confirmed down-regulation of MT3-MMP at protein level in ESCC tumors (Fig. 1B). IHC revealed that MT3-MMP was highly expressed in the membrane and cytoplasm of cells within nonkeratinized stratified squamous epithelium of nontumor esophageal tissues (left panel). However, MT3-MMP positivity was largely seen in cells within the horny pearl in tumor tissues of moderately and welldifferentiated ESCC (middle panel, Fig. 1C). In contrast, no or very low MT3-MMP staining was detected in tumor tissues of poorly differentiated ESCC (right panel, Fig. 1C).  Table 1). Further, down-regulation of MT3-MMP significantly correlated with frequency of either lymph node metastasis (P = 0.039) or resected lymph node metastasis (P = 0.017, Table 1). No correlation was observed between MT3-MMP expression and age, gender, tumor invasion, or clinical stage (P > 0.05, Table 1).

MT3-MMP down-regulation is an independent factor for poor survival of ESCC patients
Moreover, Kaplan-Meier analysis was performed to analyze relation between MT3-MMP expression and disease-specific survival in 78 patients who were followed up till the end of the study. As shown in Figure 1D First, flow cytometry was performed to analyze effects of MT3-MMP expression on cell cycle. As shown in Figure 2A and B, overexpression of MT3-MMP significantly increased the percentage of EC109 cells ectopic in G1 phase (40.5% vs. 31.2% for shRNA control, P < 0.05), while reduced S-phase cells, at 24 h after release from synchronization by serum starvation. Conversely, shRNA knockdown of MT3-MMP in EC9706 cells modestly but clearly decreased the percentage of G1 cells (28% vs. 30.9% for empty vector), while interestingly, it increased the number of cells in G2/M, indicting increased cell division ( Fig. 2A). Moreover, shRNA knockdown of MT3-MMP markedly increased proliferation of EC9706 cells, compared to shRNA control (P < 0.05 for days 3-6, Fig. 2C), while overexpression of MT3-MMP moderately reduced cell growth (days 5-6).

MT3-MMP impairs colony-forming and migration activity of human ESCC cells
Next, colony formation assays with or without soft agar were then carried out in human ESCC cells with either shRNA knockdown or ectopic overexpression of MT3-MMP, as described above. Notably, depletion of MT3-MMP dramatically increased the colony-forming ability of EC9706 cells both on plate (P < 0.05, Fig. 2D) and in soft agar (P < 0.05, Fig. 2E). The latter indicated further that MT3-MMP enhanced colony formation of ESCC cells independently of anchorage. In contrast, overexpression of MT3-MMP clearly suppressed colony formation of EC109 cells (P < 0.05 for either MT3-MMP shRNA vs. shRNA control or MT3-MMP overexpression vs. empty vector, Fig. 2E).
Last, as the clinical findings indicated that downregulation of MT3-MMP was significantly associated with ESCC metastasis (Table 1), a wound healing assay was thus performed to assess the effect of MT3-MMP on migration of human ESCC cells, reflecting metastatic potential of tumor cells in vitro. As shown in Figure 2F, although MT3-MMP-overexpressing EC109 cells displayed an impaired capability of cell migration, MT3-MMP knockdown by shRNA dramatically promoted the speed of wound healing (representative images were shown in Fig. 2F; P < 0.05 for comparison of quantified data).

MT3-MMP up-regulates the endogenous Cdk inhibitors p21 Cip1 and p27 Kip1 in ESCC cells
To further investigate the potential mechanism(s) by which MT3-MMP negatively regulates tumorigenesis and aggressiveness of ESCC, Western blot analysis was conducted to monitor expression of multiple key cell cycle-regulatory proteins, including p21 Cip1 , p27 Kip1 , cyclin A, cyclin D, cyclin B1, cyclin E, MCM, Rb, and PCNA. As shown in Figure 3A, ectopic overexpression of MT3-MMP in EC109 cells resulted in 1.2-and 2.4-fold up-regulation of p21 Cip1 and p27 Kip1 , respectively, accompanied by a modest increase in cyclin E level, while no clear change was observed in protein levels of cyclin A, cyclin D, cyclin B1, cyclin E, MCM, Rb, and PCNA. Conversely, shRNA knockdown of MT3-MMP in either EC109 (shMT3-MMP1,) or EC9706 (shMT3-MMP2) substantially decreased about 70% and 90% in expression MT3-MMP, respectively (Fig. 3B, upper panels). The shMT3-MMP-2 EC9706 cell line and its shRNA control were then used for the functional experiments as described above. Of note, knockdown of MT3-MMP led to marked reductions in expression of p21 Cip1 and p27 Kip1 (Fig. 3B, lower panel). Together, these results indicate that MT3-MMP negatively regulates proliferation and aggressiveness of ESCC cells by blocking progression of cell cycle from G1 to S and then G2/M phase (e.g., through the restriction point of cell cycle), likely in association with up-regulation of the endogenous Cdk inhibitors p21 Cip1 and p27 Kip1 . Importantly, they also provide further evidence supporting the notion that MT3-MMP down-regulation might contribute to tumorigenesis of ESCC, thereby correlating to poor prognosis of patients with ESCC.

Discussion
MMPs belong to a family of zinc-dependent endopeptidases. As proteolytic enzymes that degrade structural components of ECM, MMPs have originally been thought to promote cancer metastasis via disruption of basement membrane [6,7]. To this end, expression of MMPs is generally considered as an adverse factor for prognosis of cancer patients. However, with increasing number of MMPs as well as their functions that have been identified, the initial concept that MMPs positively correlate with cancer metastasis and thereby, poor outcome of patients has been challenged [16]. In contrast to the tumorigenic activity of MMP2, MMP9, and MT1-MMP, some MMPs (e.g., MMP8 and MMP11) might actually execute actions against tumor growth and metastasis [16] . For example, MMP8 expression has been demonstrated as a favorable prognostic factor of patients with breast or oral cancer [16]. We and Tatti, O et al. showed that overexpression of the membrane-type matrix metalloproteinase MMP16 is associated with poor clinical outcome and lymphatic invasion in gastric and melanoma cancer [19,20]. However, in this context, MT3-MMP, a membrane-anchored matrix metalloproteinase, joined this category in patients with ESCC. This study provides first evidence that in contrast to expression of MT3-MMP in nontumor (normal) esophageal tissues, MT3-MMP was down-regulated at both mRNA and protein levels in primary ESCC tumor tissues, particularly those that were poorly differentiated. More importantly, MT3-MMP down-regulation in tumor tissues significantly correlated with higher rate of metastasis and poor survival of patients with ESCC. Therefore, these findings argue that similar to MMP8 in breast or oral cancer, MT3-MMP might represent a favorable factor for prognosis of patients with ESCC.
Unlike MT1-MMP and MT2-MMP that contribute to aggressiveness and metastasis in most cancer types [21], expression of MT3-MMP has been reported only in a few tumor types, including melanoma, gastric, pancreatic, and hepatocellular cancer [12][13][14][15]. In esophageal carcinoma, MT1-MMP is highly expressed (e.g., the tumor/ normal (T/N) ratio = 2.1), which is implicated in tumor aggressiveness and prognosis [22]. Similarly, MT2-MMP expression is found in 85.4% of primary tumors in esophageal cancer, but none or very weak in normal esophageal tissues, and positively correlates to angiogenesis and tumor size, while not to patients' survival [23]. However, there is no report so far, to the best of our knowledge, regarding expression and role of MT3-MMP in human ESCC. In this study, it was found that MT3-MMP was downregulated in both 66.3% of primary ESCC tumors (by IHC, n = 86) and 63.3% of paired fresh surgically resected ESCC (by qPCR, n = 30), compared to nontumor esophageal tissues. Moreover, in sharp contrast to MT1-MMP and MT2-MMP, MT3-MMP expression was significantly adversely correlated with lymph node metastasis and poor survival of patients with ESCC. In addition, although MT3-MMP was expressed in moderately and welldifferentiated ESCC tissues, it was primarily localized within the horny pearl of tumor tissues.
Further in vitro experiments using human ESCC cancer cell line revealed that overexpression of MT3-MMP inhibited tumor cell growth, whereas down-regulation of MT3-MMP by shRNA significantly promotes ESCC cell proliferation. In another sense, MT3-MMP might play a role in inhibition of ESCC cancer cells. Moreover, downregulation of MT3-MMP also enhanced colony-forming and cell migration activity of ESCC cells, while overexpression of MT3-MMP impaired these capabilities of tumor cells. Mechanistically, MT3-MMP arrested ESCC cells at G1 phase by blocking G1/S transition, while down-regulation of MT3-MMP (e.g., by shRNA) drove ESCC cells entering G2/M phase, reflecting active cell division. These results were further supported by downregulation of p21 Cip1 and p27 Kip1 , two key endogenous Cdk inhibitors that cease cell cycle progression by blocking activity of all Cdks, in ESCC cells with MT3-MMP down-regulation. While the exact mechanism(s) for regulation of p21 Cip1 and p27 Kip1 by MT3-MMP remains to be defined, one possibility is that it may up-regulate p21 Cip1 and p27 Kip1 via inactivation of MEK/ERK as MMPs belong to the ADAMTS family that antagonizes the EGFR/MEK/ERK signaling pathway [21]. It is known that ERK activation leads to FOXM1 phosphorylation and nuclear translocation, which inhibits p21 Cip1 expression [24,25]. In summary, the present findings indicate that MT3-MMP is down-regulated in ESCC, which correlates to lymph node metastasis and poor survival of patients with this disease. They also suggest that MT3-MMP might play a tumor-suppressor role in progression of ESCC, probably through arresting tumor cells at G1 to prevent entry of cell cycle by down-regulating p21 Cip1 and p27 Kip1 . Considering MT1-MMP is a primary driving force of tumor progression in most cancer types [26], including esophageal cancer [22]; it has recently been reported that MT3-MMP acts to antagonize MT1-MMP-driven tumor cell invasion [5]. To this end, the present findings raise a possibility that down-regulation of MT3-MMP might release its brake on MT1-MMP that is highly expressed in ESCC, thereby promoting tumor progression and aggressiveness. A better understanding of the tumor-suppressive role of MT3-MMP would significantly improve our knowledge in tumor progression of ESCC and prognosis of patients with this disease.