Menadione : Sodium Orthovanadate Combination Eliminates and Inhibits Migration of Detached Cancer Cells

Exposure of cancer cells to anticancer agents in cultures induces detachment of cells that are usually considered dead. These drug-induced detached cells (D-IDCs) may represent a clinical problem for chemotherapy since they may survive anoikis, enter the circulation, invade other tissues and resume proliferation, creating a metastasis, especially in tissues where the bioavailability of anticancer agents is not enough to eliminate all cancer cells. In this study we evaluated the antiproliferative effect of menadione : sodium orthovanadate (M : SO) combination on A549 lung cancer cells as well as the ability of M : SO to induce cell detachment. In addition, we followed the fate and chemosensitivity of M : SO-induced detached cells. Using transwell chambers, we found that a fraction of the M : SO-induced detached cells were viable and, furthermore, were able to migrate, re-attach, and resume proliferation when re-incubated in drug-free media. The total elimination of A549 detachment-resistant cells and M : SO-induced detached cells were successfully eliminated by equivalent M : SO concentration (17.5 μM : 17.5 μM). Thus, M : SO prevented cell migration. Similar results were obtained on DBTRG.05MG human glioma cells. Our data guarantee further studies to evaluate the in vivo occurrence of D-IDCs, their implications for invasiveness and metastasis and their sensitivity to anticancer drugs.


Introduction
In order to preserve tissue integrity, cells adhere tightly to each other and to the surrounding matrix. By cellular interaction with the environment and adaptation to changes through multiple mechanisms, the tissue structure is maintained for proper function and physiology [1]. Important cellular functions including proliferation, motility, migration, and apoptosis are related to cell adhesion [2]. When cells detach from the tissue, they undergo anoikis, a term that defines cell apoptosis induced by inadequate or inappropriate cell-matrix interactions. Cell detachment has been suggested to be part of apoptosis since paclitaxel-induced detached cells showed a full spectrum of apoptotic changes whereas the attached cells only showed activation of caspase-3-like proteases but not poly (ADP-ribose) polymerase (PARP) cleavage [3].
Interestingly, cancer cells show mechanisms that prevent death by anoikis [4][5][6] making it possible for the detached cells to survive and migrate. This ability of cancer cells to leave home and survive in a foreign tissue is essential for invasion and metastasis. Indeed, suppression of anoikis resulted in rapidly growing tumours in mice that infiltrated lymphatics and blood vessels to colonize distant [7] and also promoted metastasis of esophageal squamous cell carcinoma [8]. Cell detachment has been shown to be the initial step of metastasis [9] and to play an important role for both invasiveness and metastasis [10]. Furthermore, detached cancer cells have been observed to reattach and, under permissive conditions, resume proliferation [9].
Drug-induced cell detachment has been extensively documented and used as a model system to study anoikis [11,12]. However, to our knowledge, there are few recent studies that address the fate and chemosensitivity of drug-induced 2 ISRN Pharmacology detached cells [13,14]. It has been reported that the acquisition of anoikis in human osteosarcoma cells did not change the sensitivity of the cells to several (etoposide, adriamycin, vinblastine, cisplatin, and paclitaxel) anticancer agents [15]. Nevertheless, the chemosensitivity of drug-induced detached cancer cells (D-IDCs) compared to the chemosensitivity of cancer cells that remain attached is largely unknown. D-IDCs may represent a clinical problem for chemotherapy since such cells may survive anoikis, enter the circulation, invade other tissues, and resume proliferation, creating a metastasis, especially in permissive tissues, that is, where the bioavailability of a given anticancer agent is not enough to eliminate all cancer cells. As an example, lung cancer is the most common type of cancer to spread to the brain, [16][17][18] where, due to the presence of the blood brain barrier (BBB), the bioavailability of most anticancer drugs can be very low compared to the systemic bioavailability [17]. In addition, most systemic treatments can transiently weaken the BBB which may facilitate for D-IDCs to seed the CNS. This may in part explain, why despite chemotherapy, brain is still a common site for metastasis for lung as well as other cancers.
The fact that detached cancer cells appear to play a fundamental role in invasion and metastasis gives further support to novel models of cancer biology that propose as ultimate goal for successful treatment of the total elimination and at once of all cancer cells. In gliomas, the "complex system model" [19] and "the stemness phenotype model" [20] predicted that in order to completely eradicate glioma all cancer cell types (stem and nonstem cancer cells) should be eliminated at once since the targeting of only one population, for example, cancer stem cells, will unlikely be a successful strategy [21]. If this prediction is true and a similar situation occurs in other types of cancer, a successful treatment will require the use of pankiller drugs, defined by us as a drug or combination of drugs that eliminates 100% of cancer cells in vitro preventing regrowth when the drug is removed [22,23]. The prefix "pan" comes from the ancient Greek word πανpan-(all, every). To screen for pankiller drugs we introduced the "Regrowth Concentration zero" parameter (RC 0 ) that defines the minimum drug concentration and time exposure needed to kill 100% of cells [24] in order to prevent cancer relapse.
On the other hand, the effect of anticancer drugs on cell proliferation is usually measured by short-term assays, such as the MTT assay [25,26]. In that type of assays, attached cells are exposed for different periods of time (usually between 24 and 72 hours) to anticancer drugs. The number of surviving cells is then estimated by the amount of produced formazan which is directly proportional to the number of viable cells. The medium is removed and replaced with DMSO in order to solubilize the water-insoluble formazan dye. Although in these assays some cells may detach, they are usually not accounted for because they are considered dead. If they are alive, the formazan produced by these cells may precipitate, remain in the well, and contribute to the total activity. The use of soluble tetrazolium salts, for example, the WST-8 (2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt), that does not require the solubilization step [27], may minimize the potential underestimation of cell viability due to the loss of detached but viable cells.
In order to efficiently monitor the ability of drugs to eliminate all cancer cells, according to our experience and despite the extended use of short-term assays, the screening of anticancer drugs may require other type, of proliferation assays such as long-term proliferation assays. For this purpose, we recently introduced a proliferation assay where cells are subjected to prolonged (between 7 and 14 days) drug exposure [22][23][24]. This assay has the potential to detect around 0.01% of resistant cells and is technically easy to perform compared to the clonogenic assay [22] that gives similar information. Using this long-term assay we have reported that combinations of menadione : vitamin C (M : VC) and menadione : sodium orthovanadate (M : SO) have pankiller activity against a panel of glioma cell lines [22,23]. Menadione alone or in combination with vitamin C has shown anticancer activity likely by its ability to generate oxidative stress and ultimately induce cell death [28][29][30]. We previously showed in human glioma cells that the M : SO combination exerts cytotoxic effects likely via the generation of oxidative stress and tyrosine kinase stimulation [22]. Sodium orthovanadate is a tyrosine phosphatase inhibitor [31]. In our studies we observed that M : SO induced cell detachment as early as 24 hours of drug exposure. Although the outcome of those cells was not evaluated at that time, we speculate now that it is possible that a fraction of those detached cells remain viable even after 72 hours.
The aim of this study was (i) to investigate the sensitivity of human non-small-cells A549 lung cancer cell line to M: SO in order to expand our previous finding on a panel of glioma cell lines (ii) to study the outcome of D-IDCs as well as their sensitivity to anticancer agents in comparison to detachment resistant cells and (iii) to design and discuss potential strategies for prevention of drug-induced cell detachment (D-ICD).

Short-Term Antiproliferative
Assay. Cells were plated (∼5000 cells per well) into flat bottom 96-well culture plates (Sigma, Stockholm, Sweden) and allowed to adhere for 24 hours. Cells to be tested were incubated with the indicated concentrations of drug while cells used as control were incubated with the highest concentration of vehicle: DMSO (<1%) or dH 2 O (<10%). After 72 hours cell proliferation was measured by using a cell counting kit (CCK, sigma, Stockholm, Sweden) in a plate reader (Labsystem iEMS reader MF) at a wavelength of 450 nm. Experiments were performed in duplicate and repeated three times in separate occasions.

Long-Term Antiproliferative
Assay. Cells were plated (∼5000 cells per well) into flat bottom 96-well culture plates and allowed to reach semiconfluence during 3 days. Afterwards, cells to be tested were treated with the indicated concentration of drug during different time periods (1-4 weeks). After withdrawing the drug, cells were re-incubated into complete cell culture medium for another 2 weeks. Control cells were, instead, incubated with equal concentration of DMSO (<0.1%) or dH 2 O (<10%) during the same time period as tested cells. In order to retain the potency of the drug and nutrients in the culture medium during the prolonged exposure, complete medium and drugs were changed twice a week. Pankiller activity and RC 0 were routinely evaluated by using inverted microscope as previously described [22][23][24]. Pankiller activity was confirmed by duplicating the experiments three times in different occasions.

Cell Quantification Assay.
Cells were plated (166,500 cells per well, density = 37,000 cells/mL) into flat bottom 6-well culture plates and allowed to reach semiconfluence during 3 days. Cells were, then, incubated with the indicated concentration of drugs or with corresponding vehicle (<0.1% DMSO or <10% dH 2 O) during 24 hours. Surviving cells were harvested in 0.25% trypsin-EDTA (Sigma, Sweden). Trypsin was neutralized with complete cell culture medium. Cell viability was measured by the Trypan blue exclusion test.
2.6. Cell Staining and Photograph. Cells were cultured (166,500 cell per well, density = 37,000 cell/mL) in 8-well chamber slide (VWR, Stockholm, Sweden) during 3-4 days. After incubation, the cells were washed twice with PBS, fixed with 4% formaldehyde for 5 minutes, and washed twice again with PBS. Cells were then stained with Mayer's hematoxylin and visualized by light microscopy.    Figure 3, these experiments clearly showed that each drug alone had low antiproliferative effect but that the M : SO combination almost completely killed all cells already after 4-day exposure.

Short Exposure to Anticancer Drugs Induced Cell
Detachment. Microscopic observation of A549 lung cancer cells exposed overnight (16 hrs) to M : SO 17.5 μM : 17.5 μM showed intensive cell detachment (Figure 4(a)). The detached cells were able to reattach and resume proliferation when reincubated in drug-free media (Figure 4(b)).
For comparison we also tested under the same conditions if other anticancer drugs (2.5 μM aphidicolin, 45 μM etoposide, 250 nM paclitaxel and 10 mM hydroxyurea) also induced cell detachment and if those drugs-induced detached cells were able to reattach and resume proliferation. All the previously mentioned substances induced extensive cell detachment. However, while cells detached by aphidicolin, paclitaxel, and hydroxyurea were able to reattach and resume proliferation, the cells detached by etoposide were unable to resume proliferation (see Supplementary material avalible online at doi:10.5402/2012/307102, Figure S1) although few of them were viable as observed by the trypan blue exclusion test. The concentrations were chosen since they are close to their IC 50 for the A549 cell line, and, therefore, the specific effect on cell detachment is unlikely due to difference in the concentration used for each compound.

Protease Inhibitors Partially Prevented Drug-Induced Cell
Detachment. Since cell detachment requires degradation of extracellular matrix proteins [33], we tested the potential preventive effect of PMSF (serine proteinase inhibitor) and marimastat (matrix metalloproteinase inhibitor) on M : SO-induced A549 lung cancer cell detachment. Based on data obtained by the short (72 hours) proliferation assay we selected 2.5 mM PMSF and 100 and 250 μM marimastat for further experiments (Figure 5(a)). We shall point out that the antiproliferative effect of these protease inhibitors by themselves, at the concentrations used in this experiment, was not significantly different from that observed with DMSO in control cultures ( Figure 5(b)). As described previously, overnight exposure to M : SO combination (17.5 μM : 17.5 μM) produced intensive cell detachment.
Interestingly, the addition of PMSF or marimastat significantly reduced the number of detached cells (Figure 5(c)). The strongest effect of marimastat at lower concentration (100 μM) seen in Figure 5(c) is likely because marimastat at higher concentrations (250 μM) has higher antiproliferative effect when used alone ( Figure 5(a)) or in combination with with M : SO. It is then expected that the number of detached cells (although viable) will be higher in marimastat 250 μM + M : SO treated cultures.  Figure 6(b)). When reincubated with vehicle alone or with M : SO 8 μM : 8 μM, the detached cells were able to reattach, resume proliferation, and migrate ( Figure 6(b)). Similar results were obtained when M : SO-induced detached glioma cells were tested (Figure 6(c)).

Discussion
In this work, we confirmed our previous observation that M : SO combination has stronger antiproliferative effect than any of those two drugs used alone [22]. We found that the M : SO needed to kill all cancer cells (pankiller) induced cell detachment and that these M : SO-induced detached cells were able to migrate, reattach and proliferate. The sensitivity of M : SO detached cells compared to detachment-resistant cells (DRCs) was similar since equivalent concentration of M : SO were able to eliminate them. This indicates that detached cells, although resistant to anoikis, have actually similar sensitivity to those cells that remain attached. In other words, resistance to anoikis does not necessarily translate into increased chemotherapy resistance. In short-term assay (72 hours) the IC 50 of the M : SO combination on A549 cells (human lung adenocarcinoma epithelial cell line) was around 8 μM (Figure 1). The maximum antiproliferative effect for M : SO combination, M alone or SO alone treatment was reached at the approximate concentrations of 25 : 25 μM, 50 μM, and 100 μM, respectively ( Figure 1). In long-term assay, the M : SO combination showed pankiller activity against the A549 human lung cancer cell line after 7-day exposure (M : SO, 17.5 μM : 17.5 μM) (Figure 2). The sensitivity of A549 cell line to M : SO combination was similar to that previously reported for glioma cells which were eliminated after exposure to M : SO (10 μM : 17.5 μM) during 10 days [22].
In the current study we have revealed that M : SO as well as four other anticancer drugs induce extensive cell detachment (Figures 4(a) and S1)). Since cell detachment has been reported to require degradation of extracellular matrix proteins [33] we evaluated the effect of protease inhibitors on M : SO-induced cell detachment. Both, PMSF and marimastat reduced the number of detached cells ( Figure 5(c)) suggesting a role of MMPs in D-IDC, in agreement with results obtained in other systems [34,35].
The fate of the D-IDCs was evaluated by following their ability to migrate and resume proliferation. As seen in Figure 4(b), M : SO detached cells were able to migrate, reattach, and resume proliferation when reincubated in drugfree media. More important, we found that detached cells showed similar sensitivity to the drugs cells showed similar sensitivity to the drugs than detachment resistant cells. The elimination of detachment resistant cells required exposure to pankiller concentrations (17.5 μM : 17.5 μM) of M : SO for 7 days (Figure 2). D-IDCs were not able to resume proliferation (and migrate) when exposed to 13 μM : 13 μM of M : SO during the same time ( Figure 6). This small difference that may suggest that D-IDCs are more sensitive than DRCs may be due to the overnight exposure to 17.5 μM : 17.5 μM of M : SO and is likely clinically irrelevant.
As we observed in this study in vitro, anticancer agents may well in vivo induce cell detachment within a tumor ( Figure S1). This may occur if the anticancer agents do not reach pankiller concentrations (RC 0 required to kill 100% cells) due to, for example, insufficient tissue bioavailability or insufficient administration time. We shall point out that we also observed a small fraction of cells that detached spontaneously ( Figure 5(c) [39]. Similarly, drug-induced detached cancer cells may survive, migrate, reattach, and resume proliferation in permissive tissues [40], that is, where the concentration of the anticancer agent is lower, creating a metastasis. Thus, D-ICD may represent a problem during cancer treatment even in the event that the primary tumor is (apparently) successfully treated.
One proposed strategy to eliminate detached cells has been to find drugs that promote anoikis, [6]. We propose here that in order to avoid metastasis caused by detached cells, at least four additional strategies should be considered.
(1) To prevent D-ICD, for example, by concomitant administration of marimastat, to existing chemotherapy regimens. This approach was shown to be partially successful since marimastat reduced the tumor spread and metastasis in experimental animal models [41,42]. However, since not all cancer types express the MMP, important for cell detachment [33], it is anticipated that use of MMP inhibitor will be useful only in a fraction of patients. In addition, other mechanisms and signaling pathways, such as the PI3K/Akt pathway, [13] are involved in cell detachment. (2) To deliver anticancer agents (after the established chemotherapy) at concentrations enough to eliminate viable D-IDCs in all tissues. Our finding that pankiller concentrations of M : SO not only prevent migration of but also eliminate D-IDC is important regarding metastasis prevention. Pankiller concentration, enough to deplete detached cells, during or immediately after local treatment with anticancer drugs may be necessary and enough to prevent metastasis. If D-IDCs were more resistant than DRCs, the elimination of these cells will require "supra-pankiller" concentrations that will be more difficult to achieve due to toxicity issues. On the other hand, plasmatic pankiller concentrations are more likely to be achieved during systemic treatment with anticancer agents but it will be a difficult task in certain tissues such as brain due to presence of the blood brain barrier. Targeted drug delivered by the use of nanoparticles [43,44] will aid the selective load anticancer drugs at "pankiller" concentrations to cancer cells and likely overcome the blood brain barrier [45] with tolerable toxicity to normal cells. (3) To use drugs that, despite the fact that they induce cell detachment, the detached cells are non-viable (as reported for etoposide (This article, Figure S1) or for other drugs such as safingol [46]). (4) To identify anticancer substances that do not induce cell detachment at all.
In conclusion, since most patients die from the progressive growth of metastases [9], D-ICD may represent an important problem and limitation of chemotherapy.
In the present study we have shown the occurrence of D-ICD and, in addition, that cells that detached were able to migrate, reattach, and resume proliferation in vitro. These results have important implications for metastasis development. The heterogeneous response of cancer cells to anticancer drugs shown in Figure S1 regarding the ability of D-IDCs to reattach and regrow, although in vitro, clearly guarantees further evaluation of the in vivo occurrence of these phenomena as well as the development of strategies to prevent, minimize, and/or eliminate this risk.