This study demonstrates the therapeutic potential of an MMP-9/MMP-2i inhibitor in combination with cisplatin for the treatment of recurrent/chemoresistant ovarian cancers overexpressing the MMP-9 gene. Using the novel high content screening technology we document for the first time an additive cytotoxic effect when MMP-9/MMP-2i is combined with cisplatin (Figure 6).

In addition, pretreating cisplatin resistant ovarian cancer cells with an MMP-9/MMP-2i inhibitor prior to treatment with cisplatin resulted in enhanced cytotoxicity at an early point. A recent study has demonstrated that MMP-9 siRNA significantly reduces the invasion and adhesive ability of ovarian cancer cells [23].

We previously identified the MMP-9 gene as upregulated in recurrent ovarian cancer tissue specimens [2]. We also identified a link between this gene and dysregulated miRNAs. The list of predicted mRNA targets for these miRNAs shared common significant pathways including actin and integrin signaling [3]. This is consistent with recent studies that directly relate MMP overexpression/ovarian cancer cisplatin resistance to actin cytoskeleton and integrins [24,25]. Other genes linked to MMP-9 such as the tissue inhibitors of matrix metalloproteinases (TIMPs) and zinc metabolism genes were also dysregulated in this cohort suggesting this gene may play an important role in recurrent/chemoresistant ovarian cancer. This difference was not observed at the protein level when MMP-9 was investigated by immunohistochemistry as both primary and recurrent/chemoresistant cases exhibited strong staining and subtle changes are difficult to interpret by immunohistochemistry. We also used a previously available TMA of primary and recurrent serous ovarian specimens. These were not all platinum-sensitive or platinum-resistant to perfectly mimic the in vitro model, hence no significant difference was observed in the expression of MMP9 levels in the recurrent compared to the primary specimens. This discrepancy is possible as stromal cells also secrete MMP9—which may modulate the invasiveness of ovarian cancer cells—and may exert varying degrees of activity following chemotherapy [13].

Serum MMP-9 showed a pattern of increasing concentration across benign, borderline, malignant and recurrent/chemoresistant serous tumors and one of its inhibitors, TIMP-2 displayed the reciprocal. While this did not reach significance, it does warrant further investigation in a larger cohort. Recent studies have demonstrated the clinical utility of MMP-9 to discriminate between malignant and benign ovarian cancers, however its utility in the recurrent setting has not been investigated [23,26,27].

We observed consistently higher MMP-9 expression in a cisplatin resistant (A2780cis) ovarian cell line compared to its sensitive counterpart (A2780). This was maintained for successive passages and before treatment of the A2780cis cell line with cisplatin to maintain chemoresistance. Over expression of MMP-9 was not observed in the IGROV/IGROV_CDDP paired sensitive/resistant cell lines which is not surprising due to the multifactorial nature of platinum resistance; hence emphasising the need for individual testing of tumors for specific targeted therapy.

In recent years, much attention has been focused on the function of MMP enzymes, given their sophisticated pattern of modulating opposite effects in cancer progression. Their overexpression correlates with increased invasion, metastasis and shortened survival [28]. Their antimetastatic propensity was further investigated by the development of MMP inhibitors (MMPIs) as a new class of anticancer drugs. Despite extensive research, very few inhibitors specific for the gelatinases have been described to date; the most potent also inhibit several other MMP family members. Unfortunately, these MMP intervention strategies have met with limited clinical success and severe toxicities. The mechanism of these toxicities is widely assumed to be due to the poor selectivity of these compounds but this has not been confirmed. In addition, it is now recognized that among MMPs, some possess cancer-promoting activities while others tumor-inhibiting functions underlining the risk of using broad-spectrum MMPIs [29]. A recent study in colorectal cancer has suggested that selective targeting of the tumor rather than stromal cell MMP-9 would be an ideal candidate for antimetastatic therapy [30].

The exact mechanism of their action is not yet completely understood. Apoptosis is induced by many but not all endogenous inhibitors and not necessarily by synthetic MMP inhibitors, thus suggesting that the mechanism may not rely on the inhibition of a MMP [31]. Synthetic MMP inhibitors appear to mimic most of the actions of TIMPs including MMP inhibition but also cell growth and proliferation [32,33]. Our results impose more than one mechanism is taking place as we observe inhibition when MMP-9 is overexpressed but we also observe some cytotoxicity in the cell line that doesn’t overexpress MMP-9 but this effect was not as significant when the cell line was treated with MMP-9i alone. It may well be that the chemotherapy is responsible for the increased expression of MMP-9 as established by a recent study [34]. Further work performing whole genome transriptome array analysis is necessary to evaluate the mechanism of action of this inhibitor.

We used a novel gelatinase inhibitor, derived from N-sulfony-lamino acid, which appears to specifically inhibit MMP-9 and to a lesser extent MMP-2 [35]. Inhibition of gelatinase activity in implanted tumor tissues has been demonstrated by means of film in situ zymography [36]. Such inhibitors selectively target tumors because their overexpression makes them accessible for homing in the tumor vasculature. The cytotoxic effects we observed were often at the lower concentration of the inhibitor where it has a higher specificity for MMP-9. The inhibitor used in this study has been shown to possess a novel proapoptotic function when combined with TNFa, TRAIL or FAS ligands [37]. This alternative mechanism could explain the enhanced apoptosis observed between our inhibitor and cisplatin, although the proposed antineoplastic mechanism has not been tested in an ovarian cancer model. A novel inhibitor of MMP-2 and MMP-9, BAY 12-9566 appears to have a more favorable toxicity profile by perturbation of the cell cycle [38]. Oral treatment with the highly selective MMPI RO 28-2653 decreases the incidence of liver metastases due to reduced MMP-2 and MMP-9 concentration in pancreatic cancer models [39].

We implemented a live-cell multiparametric HCS cytotoxicity assay measuring cell density, nuclear morphology, lysosomal mass and cell membrane integrity; the principle parameters of compromised cell health. This parallel analysis of multiple markers for cytotoxicity allows early reversible and late irreversible effects to be distinguished, therefore it provides a more sensitive interpretation of compound-induced toxicity, offering increased specificity and selectivity for toxic events [40]. We tested critical early measurements of toxicity as assays that target late events in the process of cell injury are more likely to miss toxicities that require chronic exposure or exert adverse but not lethal effects. Preincubating cells with MMP-9i at all doses was significantly cytotoxic at 3 h suggesting an early apoptotic response. Some of the pre-incubation effects followed through to the 6 h incubation and this was probably due to the significant effects observed at 3 h which the cells are trying to adapt to, suggesting it may require continuous or a more extended course of treatment to completely arrest tumor growth. A similar effect was observed using this inhibitor for treatment of melanoma [37]. Our time course results support the notion that if MMP-9 is considered a chemoresistant marker, the ability to acquire drug resistance arises early during the tumorigenesis process, as recently proposed and validated in an ovarian cancer model [41] and is associated with immediate adaptive response to treatment.

Matched sensitive and cisplatin resistant cell lines were investigated for this study. Human ovarian epithelial carcinoma cell lines A2780 and A2780cis (the cisplatin resistant counterpart) were purchased from the European Collection of Cell Cultures (ECACC: Salisbury, UK) and cultured in RPMI-1640 medium supplemented with 10% fetal calf serum, l-glutamine, penicillin and streptomycin (Sigma: Steinheim, Germany) in a humidified 5% CO₂, 95% air atmosphere. IGROV and IGROV_CDDP, were cultured in RPMI supplemented with 10% fetal calf serum.

Following trypsinisation of cultured cells, between 1 and 5 million cells were centrifuged for 12 min at 1200 rpm and supernatant was removed. Cells were lysed using Buffer RLT (Qiagen). Three different passages were used. RNA extraction was performed using the RNeasy^® mini kit (Qiagen: West Sussex, UK) as per manufacturer’s instructions. RNA was quantified using the NanoDrop^® spectrophotometer.

Eleven genes (CLDN16, S100B, BTC, IL27RA, FGF2, NRG2, S100A8, ZNF218, MMP9, TJP3, IL1R2) previously validated as upregulated in recurrent compared to primary ovarian cancers [2] were selected for TaqMan PCR analysis in the cell lines. All reactions were carried out on the ABI Prism 7000 Sequence detection system (Applied Biosystems, Applera UK: Cheshire, UK) using the TaqMan^® Universal PCR master Mix and TaqMan^® Gene Expression Assays (Applied Biosystems). Relative quantitation in A2780cis compared to A2780 cells was carried out using the delta delta cycle time (ΔΔCt) method with 18S ribosomal RNA as an endogenous control. Upregulation was observed for an average of three passages in A2780cis compared to A2780 cell line. Technical replicates were performed (n = 3) for a representative passage for each cell line.

MMP-9/MMP-2 inhibitor, (2R)-2-[(4-Biphenylylsulfonyl)amino]-3 phenylpropionic acid (C₂₁H₁₉NO₄S), was purchased from Calbiochem (Merck, Darmstadt, Germany). It is described as a potent inhibitor of MMP-2 (IC₅₀ = 310 nM) and MMP-9 (IC₅₀ = 240 nM) [35]. Stock solutions of MMP-9i which correspond to the IC50 (1×, 0.2 μM) and a 5× (1.3 μM) and 10× (2.6 μM) concentration were made in DMSO. Cisplatin was obtained from Sigma-Aldrich (Sigma-Aldrich, Ireland Ltd.: Arklow, Ireland, IMB Jena Biocomputing Group, http://www.imb-jena.de). A 10 μM stock solution was prepared in Phosphate Buffer Saline (PBS).

The peak plasma concentration of cisplatin (5 μM) was assessed in addition to a 2× (10 μM) and a 10× (50 μM) concentration. The toxic effect of MMP-9/MMP-2i and cisplatin alone or in combination was determined using the multiparameter cytotoxicity 1 kit. Cells were plated (8 × 10³) in triplicate in 96-well plates overnight and then treated with titrated concentrations of drugs for 3, 6 and 24 h. Only the inner 60 wells of 96-well plates were used due to evaporation-related edge effects in the outside wells. A pre-incubation for 3 h with MMP-9i prior to treatment with cisplatin as well as co-incubation at 3 and 6 h were assessed. DMSO was used as a vehicle control.

A multiparametric cytotoxicity assay was performed using Cellomics^® HCS reagent HitKit™ as per the manufacturer’s instructions (Thermo Fisher Scientific Inc.: Pittsburgh, PA, USA). This kit enables measurements of cell viability, cell membrane permeability and lysosomal mass/pH, which are toxicity-linked cellular markers. Following a toxic compound or nanomaterial insult, cells may respond with changes in nuclear size and/or morphology depending on the cell type and the compound as well as loss of cell membrane integrity [17,18].

Prior to staining, cells were washed with PBS, then 65 μL of a fluorophore mixture (supplied by the manufacturer) containing Hoechst 33342, lysosomal mass/pH indicator and cell membrane permeability dye were made up in 10% supplemented RPMI media. Cells were incubated with 50 μL of the fluorophore mixture for 40 min under tissue culture conditions to enable measurement of cytotoxicity indicators. Dead cells were washed away twice with 100 μL of PBS and 200 μL of wash buffer was added to each well. The experimental layout for the automated microscopic analysis, based on the In Cell analyzer 1000 (GE Healthcare: Uppsala, Sweden), was composed of untreated, MMP-9/MMP-2i and DMSO treated plates. All these were scanned and acquired in a stereology configuration of 6 randomly selected fields. Images were acquired in a stereology configuration of five randomly selected fields at 10× magnification using three detection channels with different excitation filters. The rate of cell viability and proliferation were assessed by the automated analysis of the nuclear count and morphology (DAPI filter); in parallel the fluorescent staining intensities reflecting cell permeability (FITC filter) and lysosomal mass/pH changes (TRITC filter) were also quantified for each individual cell present in the examined microscopic fields (IN Cell Investigator, GE Healthcare: Buckinghamshire, UK).

Immunohistochemical investigation for MMP-9 expression was conducted on 5 μm primary and recurrent/chemoresistant ovarian sections using the avidin-biotin-peroxidase complex detection procedure as previously described [34]. A primary TMA of 20 serous papillary adenocarcinomas was used and 20 recurrent/chemoresistant ovarian cancers of mixed histology were examined. A subset of matched primary and recurrent/chemoresistant specimens from the same patient were also investigated. Stained tissue sections were blindly investigated by a histopathologist using a 1 (weak), 2 (moderate) and 3 (strong) scoring system.

Serum samples from 39 patients were assayed in this study; 10 benign serous cystadenomas, 10 borderline serous cystadenomas, 10 advanced malignant serous papillary adenocarcinomas and 9 recurrent/chemoresistant serous adenocarcinomas. Pre-operative blood samples were taken from patients undergoing surgery for ovarian disease in St James’s Hospital Dublin. This research project had approval of the hospital ethics committee and informed consent was obtained.

The Quantikine TIMP-2 and MMP-9 Immunoassay kits (Catalog No. DTM200 and DMP900 respectively) from R&D systems were used. The ELISA was performed according to manufacturer’s protocols. Serum samples were diluted 1:50 prior to assay.

For cell culture, all treated cells were assayed in triplicate and results were expressed as mean percent surviving cells ± SE. Statistical analysis was carried out by Student’s t-test or two-way ANOVA was carried out using the PRISM^® software. Statistical significance was inferred at p < 0.05. ELISA results were analyzed using SPSS for Windows, Release 18.0.1. 2001. Chicago: SPSS Inc.

An experimental design for comparison of a single anticancer drug dose response relation with that of the same anticancer drug in combination with a fixed concentration of MMP-9/MMP2i was used and the measured responses compared to the Bliss independence reference model of synergy using the CombiTool computer program (version 2.001, IMB Jena Biocomputing Group: Jena, Germany). The drug interaction index (Ix) was calculated according to the method of Chou and Talalay [42] using CombiTool. Ix values are geometric means ± SEM in the presence of a fixed concentration of MMP-9/MMP2i, Ix < 1 = synergy; Ix = 1 additivity and Ix > 1 antagonism. The Bliss independence model is defined by the equation: Exy = Ex + Ey − ExEy for 0 < E < 1 and where Exy is the additive effect of drugs x and y as predicted by their individual effects Ex and Ey. In this case, Exy would be the effect (fractional survival) of the drug used in combination with cisplatin, and Ex and Ey the fractional survival of cells exposed to the drug alone and to the drug in combination with cisplatin, respectively [43]. Effective inhibitory concentrations (EC50) are best-fit values obtained from non-linear regression analysis of the sigmoidal dose-response relation for each drug alone or in combination with MMP9i/MMP2i. The potency ratio and associated 95% CI (confidence interval) were computed by subtracting the log EC50 of MMP9/MMP2i in combination with cispaltin from the log EC50 of drug alone and back-transformation (antilogarithm) of data. GraphPad QuickCalcs, Graphpad Prism (Version 4.03, GraphPad Software: San Diego, CA, USA) were used for data and graphic analysis.