MRP1-Collateral Sensitizers as a Novel Therapeutic Approach in Resistant Cancer Therapy: An In Vitro and In Vivo Study in Lung Resistant Tumor

Multidrug resistance (MDR) is the main obstacle to current chemotherapy and it is mainly due to the overexpression of some efflux transporters such as MRP1. One of the most studied strategies to overcome MDR has been the inhibition of MDR pumps through small molecules, but its translation into the clinic unfortunately failed. Recently, a phenomenon called collateral sensitivity (CS) emerged as a new strategy to hamper MDR acting as a synthetic lethality, where the genetic changes developed upon the acquisition of resistance towards a specific agent are followed by the development of hypersensitivity towards a second agent. Among our library of sigma ligands acting as MDR modulators, we identified three compounds, F397, F400, and F421, acting as CS-promoting agents. We deepened their CS mechanisms in the “pure” model of MRP1-expressing cells (MDCK-MRP1) and in MRP1-expressing/drug resistant non-small cell lung cancer cells (A549/DX). The in vitro results demonstrated that (i) the three ligands are highly cytotoxic for MRP1-expressing cells; (ii) their effect is MRP1-mediated; (iii) they increase the cytotoxicity induced by cis-Pt, the therapeutic agent commonly used in the treatment of lung tumors; and (iv) their effect is ROS-mediated. Moreover, a preclinical in vivo study performed in lung tumor xenografts confirms the in vitro findings, making the three CS-promoting agents candidates for a novel therapeutic approach in lung resistant tumors.


Introduction
Drug resistance, either acquired or intrinsic, is the major hurdle in treating cancer by chemotherapy. The simultaneous resistance to structurally unrelated drugs, also named multidrug resistance (MDR), often results from the overexpression of some drug efflux pumps in the cell membrane that reduce the intracellular drug levels to less than therapeutic concentrations [1][2][3][4][5][6][7][8]. In particular, three proteins belonging to the ATP-binding cassette (ABC) transporter superfamily have been identified as involved in MDR: P-glycoprotein (P-gp/ABCB1), multidrug resistance-associated Protein 1 (MRP1/ABCC1), and breast cancer resistance protein (BCRP/ABCG2).

MRP1 Activity and Antiproliferative Activity in MDCK-WT and MDCK-MRP1: Determination of the Selectivity Ratio (SR)
Within the search for novel CS agents acting on MRP1 proteins, the compounds selected among the different series of sigma-2 ligands were investigated for their activity at MRP1, paralleled with their cytotoxicity in the parental Madin-Darby canine kidney (MDCK-wt) cells and in the MDRderived cells (MDCK-MRP1) for the determination of their SR (EC50 parental cells/EC50 MDR cells). The activity at the MRP1 pump was determined measuring the transport inhibition of the profluorescent probe Calcein-AM, an MRP1 substrate, in the MRP1-overexpressing cell line (MDCK-MRP1) by the in vitro biological assay usually performed to study an MRP1 interaction. As depicted in Table 1, compounds F397 and F421, bearing the 1-(4-fluorophenyl)indol-3-yl moiety, as in the sigma-2 reference compound siramesine [43,44], together with compound F400, belonging to the tetralin series (obtained from SAfiR studies on the sigma-2 reference compound PB28 [42]), emerged as the most promising CS agents among the tested ligands with an SR > 2. In detail, as reported in Table 1, the three compounds F397, F400 and F421 showed a moderate activity vs. MRP1 (EC50 = 16.7, 17.6, and 28 M, respectively) performed in the MDCK-MRP1 cells but a high collateral sensitivity action displaying an SR = 3.41, 5.91, and 2.47, respectively. The observed CS effect seems due,

Results and Discussion
2.1. MRP1 Activity and Antiproliferative Activity in MDCK-WT and MDCK-MRP1: Determination of the Selectivity Ratio (SR) Within the search for novel CS agents acting on MRP1 proteins, the compounds selected among the different series of sigma-2 ligands were investigated for their activity at MRP1, paralleled with their cytotoxicity in the parental Madin-Darby canine kidney (MDCK-wt) cells and in the MDR-derived cells (MDCK-MRP1) for the determination of their SR (EC 50 parental cells/EC 50 MDR cells). The activity at the MRP1 pump was determined measuring the transport inhibition of the pro-fluorescent probe Calcein-AM, an MRP1 substrate, in the MRP1-overexpressing cell line (MDCK-MRP1) by the in vitro biological assay usually performed to study an MRP1 interaction. As depicted in Table 1, compounds F397 and F421, bearing the 1-(4-fluorophenyl)indol-3-yl moiety, as in the sigma-2 reference compound siramesine [43,44], together with compound F400, belonging to the tetralin series (obtained from SAfiR studies on the sigma-2 reference compound PB28 [42]), emerged as the most promising CS agents among the tested ligands with an SR > 2. In detail, as reported in Table 1, the three compounds F397, F400 and F421 showed a moderate activity vs. MRP1 (EC 50 = 16.7, 17.6, and 28 µM, respectively) performed in the MDCK-MRP1 cells but a high collateral sensitivity action displaying an SR = 3.41, 5.91, and 2.47, respectively. The observed CS effect seems due, considering the inhibition of calcein transport activity, to the interaction of the three ligands with MRP1. Moreover, these values are comparable to those observed for the CS reference compound verapamil (SR = 4.7). All the other compounds, belonging to our library of sigma-2 ligands, even if displaying a higher MRP1 activity, did not exert CS-action, as they displayed an SR ≤ 1 (data not shown), similarly to the sigma-2 reference compound siramesine (SR = 0.56). The SR was measured as the EC 50 of the MDCK cells/EC 50 of the MDCK-MRP1 cells.

Collatateral Sensitivity Study
In order to confirm the entity of the three compounds F397, F421, and F400 as sensitizers towards MRP1, further experiments were conducted on cancer cell lines that endogenously express this efflux pump. Thus, observing the results drawn from different studies regarding the MRP1 expression in several solid tumors, we screened the pump's expression in a variety of cancer cell lines such as the human colon cancer HT29 cell line and the multidrug-resistant counterpart HT29/DX, human non-small cell lung cancer A549 cell line and the multidrug-resistant counterpart A549/DX, and four different types of human breast cancer cell lines: MCF7, SKBR3, T74D, and MDA-MB-231. The immunoblotting study revealed a high amount of MRP1 in both the multidrug-resistant lung and colon cancer cell lines with the highest prevalence in A549/DX, followed by the respective parental cell lines A549 and HT29, and by the MDA-MB-231 cell line ( Figure 2A).  Therefore, A549/DX cells were selected as a model to investigate the CS properties of our compounds. Firstly, in order to confirm the data observed on the "pure" model MDCK-MRP1, we tested, in the same experimental conditions, the SR of F397, F400, and F421 by measuring their cytotoxicity in the resistant A549/DX cells and their parental counterpart A549, obtaining an SR ≥ 2 for all (SR= 2.5 for derivative F397, SR= 2.1 for F400, and SR = 2 for F421).
Since the doubling time of A549 cells is 22 h and the doubling time of A549/DX cells is 61 h (data not shown), in this first experimental set we used a 24 h time, i.e., the most suitable for A549 cells' cytotoxicity as performed in the "pure" MRP1 model. Indeed, the accurate detection of cell viability in A549 cells was critical in calculating the SR in a reliable manner. The following experiments, all performed in A549/DX cells, were performed after 72 h, i.e., the best timing to detect a cytotoxic effect on the slowly growing resistant population. A549/DX displaying the highest GSH/GSSG ratio. This result was not surprising. Indeed, A549 cells are known for having high levels of GSH, caused mainly by the high levels of anti-oxidant enzymes [46] that prevent the oxidation of GSH. Moreover, we previously demonstrated that cells with acquired resistance to doxorubicin increase the metabolic flux through the pentose phosphate pathways (PPP) that produce NADPH, a critical metabolite to regenerate the reduced form of glutathione. This phenotype is peculiar for drug-resistant MRP1-expressing cells [47], as A549/DX are. The high levels of GSH may protect cells from the oxidative damage induced by cis-Pt [48] and activate the glutathione S-transferase (GST) enzymes that conjugate GSH to cis-Pt and promote its efflux via MRP1 [49].
Therefore, A549/DX cells were selected as a model to investigate the CS properties of our compounds. Firstly, in order to confirm the data observed on the "pure" model MDCK-MRP1, we tested, in the same experimental conditions, the SR of F397, F400, and F421 by measuring their cytotoxicity in the resistant A549/DX cells and their parental counterpart A549, obtaining an SR ≥ 2 for all (SR = 2.5 for derivative F397, SR = 2.1 for F400, and SR = 2 for F421).
Since the doubling time of A549 cells is 22 h and the doubling time of A549/DX cells is 61 h (data not shown), in this first experimental set we used a 24 h time, i.e., the most suitable for A549 cells' cytotoxicity as performed in the "pure" MRP1 model. Indeed, the accurate detection of cell viability in A549 cells was critical in calculating the SR in a reliable manner. The following experiments, all performed in A549/DX cells, were performed after 72 h, i.e., the best timing to detect a cytotoxic effect on the slowly growing resistant population.
First, we wondered if there was a correlation between the inhibition of the MRP1 protein and the CS properties of the compounds. To this aim, we adopted a concentration similar to their EC50, i.e., 10 µM. In this experimental condition, the three compounds resulted as cytotoxic, confirming their behavior as CS-promoting agents in A549/DX cells ( Figure 3). their resistant counterparts that overexpress MRP1 but not P-gp), that by contrast, is devoid of P-gpmediated CS properties (results from MCF7 cells). We next observed the antiproliferative effect of these three compounds co-administrated with the elective drug usually chosen for the treatment of the non-small cell lung cancer, cis-platinum (cis-Pt). Cis-Pt is effective against cancer cells because it creates cross-links between purine bases, alters the double helix conformation, and induces DNA strand breaks, impairing the DNA repairing machinery as well. These DNA damages trigger apoptosis in sensitive cells [50]. For this broad spectrum of activities, cis-Pt is the first line of treatment in several cancer types, including lung cancer [50]. Unfortunately, since cis-Pt is a substrate of MRP1 [1], which effluxes cis-Pt, limiting its intracellular accumulation, MRP1-expressing cancer cells are usually resistant to the cytotoxic effect of the drug [51,52]. This is the case of A459/DX cells, rich with MRP1, a prototype of lung cancer cells resistant to cis-Pt.
At first, the viability of each compound on A549/DX after 72 h treatment, alone or in combination with the MRP1 inhibitor MK571 or the antineoplastic drug cis-Pt both at 25 M has been evaluated. The concentration of cis-Pt at 25 M was chosen as A549/DX cells are platinum-resistant [53], but cells are re-sensitized when the anticancer drug is co-administered with MK571 at the same concentration. As illustrated in Figure 4, MK571 alone was not cytotoxic to A549/DX cells. The sensitivity to cis-Pt was restored when co-administrated with MK571, suggesting that MRP1 plays a key role in inducing the resistance to cis-Pt in our model. In order to confirm that the observed effect is due to the interaction of the ligands with MRP1, the cellular viability at 72 h has been measured, testing the three ligands in co-administration with the MRP1 inhibitor, MK571, at 25 µM, a concentration that fully inhibited the MRP1 ATPase activity on A549/DX cells (ATPase activity in untreated A549/DX cells: 4.32 ± 0.58 nmoles Pi/min/mg prot; ATPase activity in MK571-treated A549/DX cells: 0.40 ± 0.12 nmoles Pi/min/mg prot). As depicted in Figure 3, the cytotoxicity of the three compounds was reverted by the MRP1 inhibitor MK571, confirming that MRP1 is the target of the F397, F400, and F421 compounds in A549/DX.
While F400 is devoid of sigma-2 receptor affinity [42], F421 and F397 respectively display a moderate (K i = 169 nM) [41] to a high affinity (K i = 5.34 nM) [28] towards sigma-2 receptors. Nevertheless, the involvement of a sigma-2 receptor-mediated effect was ruled out by the detection of a low density of sigma-2 receptors in A549 cells (data not shown). Additionally, besides the reversal of the activity of these ligands by the MRP1 inhibitor MK571, worthy of note is the result which was obtained with the three sensitizers in the MCF7 cell line, having a high level of sigma-2 and low level of MRP1 (as depicted in Figure 1). While F400 and F421 were not cytotoxic in these cells (EC 50 > 100 µM, data not shown), F397 showed an EC 50 = 17.8 µM [28]. Importantly, cytotoxic activity of F397, which was shown to be a potent P-gp inhibitor in the MCF7 cell line (EC 50 = 0.21 µM), was lower in the corresponding resistant MCF7/DX overexpressing P-gp (EC 50 = 21.8 µM) [28]. These data strongly support the MRP1-mediated CS action of F397 (results from A549 and MDCK cells and their resistant counterparts that overexpress MRP1 but not P-gp), that by contrast, is devoid of P-gp-mediated CS properties (results from MCF7 cells).
We next observed the antiproliferative effect of these three compounds co-administrated with the elective drug usually chosen for the treatment of the non-small cell lung cancer, cis-platinum (cis-Pt). Cis-Pt is effective against cancer cells because it creates cross-links between purine bases, alters the double helix conformation, and induces DNA strand breaks, impairing the DNA repairing machinery as well. These DNA damages trigger apoptosis in sensitive cells [50]. For this broad spectrum of activities, cis-Pt is the first line of treatment in several cancer types, including lung cancer [50]. Unfortunately, since cis-Pt is a substrate of MRP1 [1], which effluxes cis-Pt, limiting its intracellular accumulation, MRP1-expressing cancer cells are usually resistant to the cytotoxic effect of the drug [51,52]. This is the case of A459/DX cells, rich with MRP1, a prototype of lung cancer cells resistant to cis-Pt.
At first, the viability of each compound on A549/DX after 72 h treatment, alone or in combination with the MRP1 inhibitor MK571 or the antineoplastic drug cis-Pt both at 25 µM has been evaluated. The concentration of cis-Pt at 25 µM was chosen as A549/DX cells are platinum-resistant [53], but cells are re-sensitized when the anticancer drug is co-administered with MK571 at the same concentration. As illustrated in Figure 4, MK571 alone was not cytotoxic to A549/DX cells. The sensitivity to cis-Pt was restored when co-administrated with MK571, suggesting that MRP1 plays a key role in inducing the resistance to cis-Pt in our model.  One-way analysis of variance (ANOVA) analysis: **** p < 0.0001 vs. control.
One of the possible reasons of this sensitization is the ability of these three compounds to inhibit the efflux of cis-Pt, which is MRP1-mediated. Interestingly, the cytotoxic effect of the three compounds combined with cis-Pt was slightly more marked than the effect of MK571 in terms of cell viability. This could be associated with a second reason, considering one of the putative mechanisms of CS consisting of an increased production of ROS, capable of triggering the apoptosis of MDR cells. To investigate whether this mechanism is involved in the sensitizing effects elicited by F397, F400, and F421, we measured the intracellular ROS production in A549/DX after 24 h treatment with these three molecules at 10 M, with and without 25 M cis-Pt ( Figure 5A). We noticed that these three compounds alone increased the amount of intracellular ROS. This was further enhanced when they were co-administrated with the antineoplastic drug, thus allowing its antitumor activity. As reported in Figure 5B, this effect was reverted by the ROS scavenger Tempol, tested at 10 mM. Moreover, to verify whether the increase in ROS induced the reduction in cell viability elicited by the compounds, Using MK571 as a reference, a comparison was made between the cytotoxic effect of the compounds F397, F400, and F421 alone at 10 µM and in association with cis-Pt at 25 µM (Figure 4). All the three compounds have a cytotoxic effect on their own, that drastically increases upon co-administration with cis-Pt.
One of the possible reasons of this sensitization is the ability of these three compounds to inhibit the efflux of cis-Pt, which is MRP1-mediated. Interestingly, the cytotoxic effect of the three compounds combined with cis-Pt was slightly more marked than the effect of MK571 in terms of cell viability.
This could be associated with a second reason, considering one of the putative mechanisms of CS consisting of an increased production of ROS, capable of triggering the apoptosis of MDR cells. To investigate whether this mechanism is involved in the sensitizing effects elicited by F397, F400, and F421, we measured the intracellular ROS production in A549/DX after 24 h treatment with these three molecules at 10 µM, with and without 25 µM cis-Pt ( Figure 5A). We noticed that these three compounds alone increased the amount of intracellular ROS. This was further enhanced when they were co-administrated with the antineoplastic drug, thus allowing its antitumor activity. As reported in Figure 5B, this effect was reverted by the ROS scavenger Tempol, tested at 10 mM. Moreover, to verify whether the increase in ROS induced the reduction in cell viability elicited by the compounds, we evaluated the antiproliferative effect at 72 h of these three sensitizers alone and with 25 µM cis-Pt, adding 10 mM Tempol at time zero and after 48 h from the incubation. As shown in Figure 5B, the data obtained are in line with the ROS production hypothesis: higher were the ROS levels, lower was cell viability, and vice-versa.  An increase in ROS of mitochondrial origin is particularly effective in killing multidrug-resistant cells [53]. To investigate whether the ROS were of mithocondrial origin, the same experiment was set up evaluating ROS production at 24 h with Mitotempol, the scavenger of mithocondrial ROS. As illustrated in Figure 6A, compounds F327, F400, and F421 increased the intramithocondrial ROS. This effect was enforced by the concurrent administration of cis-Pt, but reversed by Mitotempol. Evaluating the relative cellular vitality on A549/DX at 72 h, the antiproliferative effect of cis-Pt, negligible when the drug was used alone, was enhanced by the CS-promoting agents and reverted by Mitotempol ( Figure 6B)

In Vivo Tumor Growth
In order to translate in vivo the capability of F397, F400, and F421 to act as "sensitizers", A549/DX cells were implanted in immune-deficient BALB/C mice, and once the tumor reached the volume of 50 mm 3 , the mice were randomized and treated three times weekly with the vehicle (saline solution) and/or cis-Pt. As depicted in Figure 7, the tumor volume was monitored daily, confirming the resistance of A549/DX against the standard chemotherapy regimen.
The animals were treated for three times weekly with a single dose of the indole-based ligands F397 or F421, while F400 could not be dissolved in solvents suitable for in vivo studies and could not

In Vivo Tumor Growth
In order to translate in vivo the capability of F397, F400, and F421 to act as "sensitizers", A549/DX cells were implanted in immune-deficient BALB/C mice, and once the tumor reached the volume of 50 mm 3 , the mice were randomized and treated three times weekly with the vehicle (saline solution) and/or cis-Pt. As depicted in Figure 7, the tumor volume was monitored daily, confirming the resistance of A549/DX against the standard chemotherapy regimen. viability assays performed in vitro and could lead to an underestimation of the compounds' potency. Further, the repeated administration of F397 and F421 followed in vivo instead of the single dosage used in vitro can be an additional explanation. Indeed, repeated and lower doses of anti-tumor agents have been reported to be more effective than one single higher dosage against chemoresistant tumors [54]. This could also be the case of F397 and F421 against A549/DX tumors. Notably, the reduction of cell proliferation in vitro (i.e., 50%) is similar to the extent of tumor decreases observed in vivo, indicating a good correlation between the two experimental settings. It is also worth noting that the treatment was not toxic for liver, heart, and kidney compared with the control, as shown from the hemato-chemical parameters of the animals ( Table 2).  The animals were treated for three times weekly with a single dose of the indole-based ligands F397 or F421, while F400 could not be dissolved in solvents suitable for in vivo studies and could not be tested in vivo. The administration of F397 or F421 as single agents determined a low growth reduction ( Figure 7A), but their co-administration with cis-Pt greatly reduced tumor growth ( Figure 7B). As for the in vitro experiments, co-administration of the MRP1 inhibitor MK571 with these sensitizers abated the cytotoxic effect, supporting that the effect of F397 and F421 is MRP1-mediated ( Figure 7C). Of note, the dose of F397 and F421 effective as an anti-tumor agent in vivo was lower than the dosage effective as an anti-proliferative agent in vitro. One possible reason could be that part of the anti-tumor effect of the compounds was mediated by their activity on the tumor microenvironment, not necessarily only on tumor cells. This aspect cannot be evaluated in the viability assays performed in vitro and could lead to an underestimation of the compounds' potency. Further, the repeated administration of F397 and F421 followed in vivo instead of the single dosage used in vitro can be an additional explanation. Indeed, repeated and lower doses of anti-tumor agents have been reported to be more effective than one single higher dosage against chemoresistant tumors [54]. This could also be the case of F397 and F421 against A549/DX tumors. Notably, the reduction of cell proliferation in vitro (i.e., 50%) is similar to the extent of tumor decreases observed in vivo, indicating a good correlation between the two experimental settings. It is also worth noting that the treatment was not toxic for liver, heart, and kidney compared with the control, as shown from the hemato-chemical parameters of the animals ( Table 2). Balb/C mice (n = 8 animals/group) were treated as described. Blood was collected immediately after euthanasia and analyzed for lactate dehydrogenase (LDH), aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (AP), creatinine, and creatine phosphokinase (CPK). Data are presented as means ± SD. * p < 0.05: vs. ctrl group.

Cell Cultures
MDCK and MDCK-MRP1 cells were a gift of Prof. P. Borst, NKI-AVL Institute, Amsterdam, The Netherlands. The MDCK cells were grown in DMEM high glucose supplemented with a 10% fetal bovine serum, 2 mM glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin in a humidified incubator at 37 • C with a 5% CO 2 atmosphere.

Calcein-AM Experiments
These experiments were carried out as described by Riganti et al., with minor modifications [55]. Each cell line (30,000 cells per well) was seeded into a black CulturePlate 96/well plate with a 100 µL medium and allowed to become confluent overnight. The 100 µL of test compounds was solubilized in a culture medium and added to monolayers, with final concentrations ranging from 0.1 to 100 µM. The 96/well plates were incubated at 37 • C for 30 min. Calcein-AM was added in 100 µL of phosphate buffered saline (PBS) to yield a final concentration of 2.5 µM, and then the plates were incubated for 30 min. Each well was washed 3 times with ice-cold PBS. Saline buffer was added to each well and the plates were read with the Victor3 plate reader (PerkinElmer) at excitation and emission wavelengths of 485 and 535 nm, respectively. In these experimental conditions, the Calcein cell accumulation in the absence and in the presence of the tested compounds was evaluated and the fluorescence basal level was estimated with the untreated cells. In treated wells, the increase in fluorescence with respect to the basal level was measured. EC 50 values were determined by fitting the fluorescence increase percentage vs. log(dose).

Antiproliferative Assay
The determination of cell growth was performed using an MTT assay at 24 or 72 h [55]. On day 1, 10,000 cells/well were seeded into 96-well plates in a volume of 100 µL. On day 2, the drugs concentrations (0.1, 1, 10, 25 µM) were added. In all the experiments, the various drug solvents (ethanol, DMSO) were added in each control to evaluate a possible solvent cytotoxicity. After the established incubation time with drugs, 10 µL MTT (0.5 mg/mL) was added to each well, and after 3 h incubation at 37 • C, the supernatant was removed. The formazan crystals were solubilized using 100 µL of DMSO and the absorbance values at 570 and 630 nm were determined on the microplate reader Victor 3 from PerkinElmer Life Sciences. The absorbance of the untreated cells was considered equal to 100% cell viability; the viability of cells measured in each experimental condition was expressed as a percentage of viable cells in the considered condition vs. the viability of the untreated cells.

Glutathione Measurement
Cells were washed with PBS and 600 µL 0.01 N HCl was added. After gentle scraping, cells were frozen/thawed twice and proteins were precipitated by adding 120 µL of 6.5% w/v 5-sulfosalicylic acid to 480 µL of lysate. Each sample was placed in ice for 1 h and centrifuged for 15 min at 13,000× g (4 • C). The protein content was measured with a BAC Kit (Sigma Chemicals. Co), as per the manufacturer's instruction. Total glutathione was measured in 20 µL of the cell lysate or supernatant with the following reaction mix: 20 µL of stock buffer (143 mM NaH 2 PO 4 and 63 mM EDTA, pH 7.4), 200 µL of daily reagent (10 mM 5,5'dithiobis-2-nitrobenzoic acid and 2 mM NADPH in stock buffer), and 40 µL of glutathione reductase (8.5 U/mL). The content of the oxidized glutathione (GSSG) was obtained after the derivatization of GSH with 2-vinylpyridine (2VP): 10 µL of 2VP was added to 200 µL of cell lysate or culture supernatant and the mixture was shaken at room temperature for 1 h. Glutathione was then measured in 40 µL of the sample as described. The reaction was followed kinetically for 5 min using a Synergy HT Multi-Mode Microplate Reader (Bio-Tek Instruments, Winooski, VT, USA), measuring the absorbance at 415 nm. Each measurement was made in triplicate and results were expressed as nmol of glutathione/min/mg cellular protein, according to a titration curve set up with serial dilutions (10 µM-0.1 nM) of 1:1 GSH/GSSG mix. For each sample, GSH was obtained by subtracting the GSSG from the total glutathione.

Total Reactive Oxygen Species (ROS) Measurement
Here, 1 × 10 6 whole cells were re-suspended in a final volume of 0.5 mL PBS and incubated for 30 min at 37 • C with 5 µM of the fluorescent probe 5-(and-6)-chloromethyl-2',7'dichlorodihydro-fluorescein diacetate-acetoxymethyl ester (DCFDA-AM, Sigma Chemicals Co.) in the dark. Cells were then centrifuged at 13,000× g at 37 • C and re-suspended in 0.5 mL PBS. The fluorescence of each sample, considered as the index of ROS levels, was read at 492 (λ excitation) and 517 nm (λ emission) using a Synergy HT Multi-Mode Microplate Reader (Bio-Tek Instruments). The results were expressed as nmol total ROS/mg cell proteins. A preliminary titration curve was set up by incubating the cells for 1 h with the cells treated with serial dilutions (10-0.01 nM) of the pro-oxidant agent menadione.

Mitochondrial ROS Measurement
Here, 1 × 10 6 cells were re-suspended in a final volume of 0.5 mL PBS and incubated for 30 min at 37 • C with 5 µM of the fluorescent probe MitoSOX Red (Invitrogen Life Technology, Milano, Italy) in the dark. Cells were then centrifuged at 13,000× g at 37 • C and re-suspended in 0.5 mL PBS. The fluorescence of each sample, considered as the index of ROS levels, was read at 510 (λ excitation) and 580 nm (λ emission) using a Synergy HT Multi-Mode Microplate Reader (Bio-Tek Instruments). The results were expressed as nmol mitochondrial ROS/mg cell proteins. A preliminary titration curve was set up by incubating the cells for 1 h with the cells treated with serial dilutions (10-0.01 nM) of the pro-oxidant agent menadione.
The animal care and experimental procedures were approved by the Bio-Ethical Committee of the Italian Ministry of Health (#122/2015-PR).

Statistical Analysis
All data in the text and figures are provided as means ± SD. The results were analyzed by a Student's t-test and ANOVA test, using Graph-Pad Prism (Graph-Pad software, San Diego, CA, USA). p < 0.05 was considered significant. The investigators responsible for the data analysis were unaware of the experimental conditions analyzed.

Conclusions
CS, that is, a sort of synthetic lethality according to which resistant tumor cells are selectively killed (rather than their wild type counterparts), appears as a promising therapeutic approach for the treatment of resistant tumors. In this context, MDR pump modulators may be endowed with CS properties. Besides the deeply explored P-gp-mediated CS, the less explored MRP1-mediated CS has an important role. With the aim of identifying promising MRP1 modulators as CS inducers, we screened a library of sigma-2 receptor ligands which were previously identified as P-gp-mediated CS inducers. Three compounds, F397, F400 and F421, that modulated MRP1 and displayed cytotoxicity in a number of cell lines, exerted a more potent cytotoxicity in the MRP1 overexpressing cells (MDCK/MRP1 and A549/DX) compared with the wild type counterparts, thus showing important CS properties. All the three compounds were found to alter the GSH/GSSG ratio in the cell lines studied, and to increase the mitochondrial ROS. The three compounds, that were able to exert cytotoxicity by themselves, potently synergized with cis-Pt, re-activating this "traditional" antitumor drug in cells refractory to the drug because of the MRP1 expression. These in vitro results were translated in the corresponding in vivo model of cis-Pt-resistant A549/DX xenografts, where the co-administration of the MRP1-mediated sensitizers F397 and F421 with cis-Pt greatly reduced tumor growth, with no signs of toxicity. Importantly, both in vitro and in vivo, the activity of these sensitizers was demonstrated to be mediated by the interaction with MRP1. Overall, we demonstrated that MRP1-mediated CS is a promising approach for the treatment of resistant tumors such as the non-small cell lung tumor, whose bad prognosis urgently requires novel therapeutic approaches. The results herein obtained make these classes of compounds worthy to be further investigated for other MRP1 overexpressing resistant tumors, while F397, F400, and F421 may be considered as first in class for the development of novel MRP1-mediated collateral sensitizers.

Funding:
The work was financially supported by the Italian Association for Cancer Research (IG21408 to CR).

Conflicts of Interest:
The authors declare no conflict of interest.