Cell lines, source and culture conditions are described in Table 1. All cell lines were cultured in monolayer at 37°C under 5% CO₂.

Sodium 4-carboxymethoxyimino-(4-HPR), synthesized as previously described (Musso et al., 2016), was dissolved at 10 mmol/L in DMSO prior to further dilution in culture medium and stored at -80°C in the dark. Vinblastine (Sigma–Aldrich, St. Louis, MO, USA) was dissolved at 1 mmol/L in water before further dilution in culture medium and stored at 4°C. Vitamin C (Sigma, St. Louis, MO, USA) was added to cells 1 h before Sodium 4-carboxymethoxyimino-(4-HPR) treatment.

Cells were seeded in 96-well tissue culture plates at 7 × 10³ cells/well and were allowed to adhere for 24 h before treatment. Cells were grown in the presence of vehicle or sodium 4-carboxymethoxyimino-(4-HPR) at a final concentration of 0.3, 1, 3, 5, and 10 μM. Cellular growth was assessed after 72 h by sulforhodamine B (SRB) assay, as previously described (Tiberio et al., 2010). The antiproliferative activity of the compound was tested in three independent experiments with four replicate wells for each analysis.

Intracellular production of ROS was assessed by employing the oxidation-sensitive dye 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate (CM-H2DCFDA; Molecular Probes, Inc., Eugene, OR, USA) as described previously (Villani et al., 2006) Briefly, 8 × 10⁵ cells/well were plated in six-well cell culture plates and incubated for 4 h in the presence of 3 and 5 μM sodium 4-carboxymethoxyimino-(4-HPR) and/or 100 μM vitamin C. Medium was discarded and, under low light conditions, replaced with 50 μM CM-H2DCFDA in whole medium for 20 min at 37°C. Cells were harvested, transferred to foil-wrapped tubes and analyzed immediately by flow cytometry. The ROS generation was tested in three independent experiments with three replicates for each sample in each analysis.

Proteins were extracted by using sodium dodecyl sulfate (SDS) sample buffer (62.5 mM Tris–HCl [pH 6.8], 2% SDS) containing 1 mM phenylmethylsulfonyl fluoride, 10 μg/mL pepstatin, 12.5 μg/mL leupeptin, 2 μg/mL aprotinin, 1 mM sodium orthovanadate, and 1 mM sodium molybdate. Proteins were processed for western immunoblotting as previously described (Appierto et al., 2004). The following antibodies used for immunoblotting were purchased from the indicated suppliers and used according the manufacturer testing conditions: PLAB and pJNK from Santa Cruz Biotechnology (Santa Cruz, CA, USA); cleaved caspase-3 and pEIF2α from Cell Signaling (Beverly, MA, USA); and GAPDH, α-tubulin, and actin from Sigma–Aldrich. Quantification of pJNK and pEIF2α band intensities was performed using the ImageJ software freely available at http://rsb.info.nih.gov/ij/. Values were normalized by actin band intensity.

Cells were seeded in 24 mm Petri dishes and the day after were exposed to different concentration of sodium 4-carboxymethoxyimino-(4-HPR) and vinblastine and 24 h later processed for the tubulin polymerization assay as previously described (Cappelletti et al., 2004; Appierto et al., 2009a). Cytosolic and cytoskeletal-associated proteins were separated by SDS–PAGE and tubulin distribution was analyzed by immunoblotting using α-tubulin antibody (Sigma–Aldrich).

Analyses of cell cycle distribution were conducted as previously described by our group (Tiberio et al., 2010) after 24 h of treatment with 3 and 5 μM sodium 4-carboxymethoxyimino-(4-HPR) and/or 100 μM vitamin C. Cell cycle analysis was performed using FACScan flow cytometer (Becton Dickinson, San Jose, CA, USA). The percentage of cells in different phases of cell cycle was determined by ModFit LT cell cycle analysis software (Verity Software House, Topsham, ME, USA), considering only cells with DNA content ≥2n. Apoptotic cells were identified as a sub-G1 population (DNA content <2n).

Cells, grown on glass coverslip slides in 24 mm Petri dishes, were fixed in 100% methanol at -20°C for 7 min, washed with PBS and then blocked at room temperature for 1 h in 3% BSA/ 0.1% (v/v) Triton X-100/PBS. Cells were incubated overnight at 4°C with the primary antibody mouse anti-α-tubulin (Sigma–Aldrich), washed three times with PBS, and then incubated for another hour at room temperature with the secondary antibody anti-mouse Alexa 488 (Molecular Probes, Inc., Eugene, OR, USA), washed thrice with PBS and stained with Hoechst 33342. Slides were mounted with Mowiol (Calbiochem, San Diego, CA, USA), and viewed with a fluorescence microscope [images were recorded with a Spot Insight digital camera (Delta Sistemi, Rome, Italy) equipped with a system of image analysis (IAS 2000; Delta Sistemi)].

All mice were purchased from Charles River (Calco, Italy) and maintained in laminar-flow rooms or individually ventilated cages at constant temperature and humidity, with food and water administered ad libitum. All the in vivo experiments were approved by the Ethics Committee for Animal Experimentation of the Fondazione IRCCS Istituto Nazionale Tumori of Milan according to National and Institutional guidelines.

Three different mouse models have been employed: peritoneal mesothelioma (STO cells), ovarian carcinoma (IGROV-1 in vivo cell line, growing mainly as ascites and maintained by i.p. serial passages of ascitic cells) and breast (MDA-MB-231 cells) cancer. As regards to mesothelioma, nude mice were inoculated s.c. with 10 × 10⁶ STO cells, and treatment with sodium 4-carboxymethoxyimino-(4-HPR) began 1 day after tumor cell inoculation (doses: 30 and 60 mg/kg i.p.; 5 days/week for 5 weeks; end of treatment: day 32). Tumor growth in the control and treated groups were evaluated, and the differences were statistically analyzed.

For the ovarian cancer model, nude mice were inoculated i.p. with 2.5 × 10⁶ IGROV-1 cells, and treatment began 1 day after tumor cell inoculation (doses: 30, 60, and 90 mg/kg i.p.; 5 days/week for 3 weeks; end of treatment: day 18). Mice with ovarian cancer were treated for only 3 weeks because IGROV-1 tumor is characterized by rapid growth and accumulation of ascites. For ethical reasons, mice were sacrificed as ascites became evident (before impending death) and 50% of control mice were sacrificed within day 20. The day of sacrifice was considered as the day of death for calculating and statistically analyzing the survival time of each group. In order to also evaluate ascitic growth (viable tumor cells were counted using the trypan blue exclusion assay) and tumor weight (TW), six additional mice were used: three treated with 60 mg/kg i.p. of sodium 4-carboxymethoxyimino-(4-HPR) (we decided to use an intermediate dosage) and three control mice. Ascitic fluid and tumor tissues were collected during the last week of treatment.

As third cancer model, 5 × 10⁶ human breast cancer cells (MDA-MB-231) were resuspended in PBS and mixed 1:1 (v/v) with Matrigel (BD Biosciences, Bedford, MA, USA) and were inoculated into the right second mammary fat pad (m.f.p.) of NOD/SCID mice under anesthesia (ketamine–xylazine). Treatment with the drug began 1 day after tumor cell inoculation (dose: 90 mg/kg; 4 days/week for 7 weeks; end of treatment: day 45). We decided to adopt for the breast cancer model a different schedule of treatment (i.e., 4 instead of 5 days/week), taking in consideration the different mouse strain employed and a long-term treatment hypothesized for this experiment. Tumor growth in the control and treated groups were evaluated, and the differences were statistically analyzed.

For all the in vivo experiments, the drug was first dissolved in DMSO, diluted in sterile water at final DMSO concentration of 5% (higher dose), and then diluted in sterile water to obtain the other doses. It was administered by i.p. route and delivered at 10 mg/kg of body weight. Control mice were treated with the same solvent as used to dissolve the sodium 4-carboxymethoxyimino-(4-HPR). The animals were examined thrice a week to check tumor growth, body weight and any signs of toxicity. TW was calculated by the formula TW (g) = (d² × D) / 2 where d and D are the shortest and the longest diameters, respectively, measured in centimeters. All the experiments were conducted in duplicate, obtaining comparable results.

Differences between mean values were assessed by two-tailed Student’s t-test and for survival analysis data were compared with the log-rank test. The analyses were performed using GaphPad Prism v.5.02 (GraphPad Software, La Jolla, CA, USA). P-values < 0.05 were considered as statistically significant.

The improvement in solubility obtained through the chemical modification of 4-oxo-4-HPR (Musso et al., 2016), prompted us to investigate the antitumor activity of this new derivative. The growth-inhibitory effect of sodium 4-carboxymethoxyimino-(4-HPR) was tested in a panel of human cancer cell lines of different histotypes. The 50% effective cytotoxic concentrations (EC₅₀) after 72 h of treatment were calculated and reported in Table 2. The compound showed an ability to inhibit the cell growth similar to that of the parent drug 4-oxo-4-HPR (Villani et al., 2006), in ovarian and breast cancer cell lines and in peritoneal mesothelioma and neuroblastoma cells.

Overall, these results indicate that sodium 4-carboxymethoxyimino-(4-HPR) is a retinoid derivative effective in inhibiting proliferation of cancer cells of different histotypes.

We have recently reported that the chemical modification performed on 4-oxo-4-HPR and leading to sodium 4-carboxymethoxyimino-(4-HPR) did not impair the antimitotic activity of the parental drug, in terms of cell cycle arrest and formation of multipolar spindles (Musso et al., 2016). In the present study we deeper dissected the activity of the compound to investigate whether this drug entirely preserved the double mechanism of action previously described for 4-oxo-4-HPR (Appierto et al., 2009a; Tiberio et al., 2010). To this aim, we first analyzed whether sodium 4-carboxymethoxyimino-(4-HPR) induced the generation of ROS in A2780, a human ovarian carcinoma cell line previously used as in vitro model to investigate the mechanism of action of the parental drug (Appierto et al., 2009a; Tiberio et al., 2010). The analysis revealed that the molecule caused the generation of ROS in a dose-dependent manner (Figure 2A). Treatment with both 3 and 5 μM 4-carboxymethoxyimino-(4-HPR) for 24 h caused a twofold increase in JNK phosphorylation, a slight induction of pEIF2α (1.5-fold over control for both drug doses), marked PLAB upregulation and caspase-3 cleavage (Figure 2B), suggesting that sodium 4-carboxymethoxyimino-(4-HPR) was able to induce the activation of the ROS-related apoptotic cascade described for its parental drug (Tiberio et al., 2010). On the other hand, the treatment led to a marked dose-dependent accumulation of cells in the G₂-M phases, associated with abnormal mitotic spindles with loss of normal bipolarity and formation of multipolar spindles, as revealed by immunofluorescence labeling for α-tubulin (Figure 3A).

To investigate whether the sodium 4-carboxymethoxyimino-(4-HPR)-induced mitotic arrest and formation of multipolar spindles were related to the ROS generation, the effects of the antioxidant vitamin C on cell cycle perturbation and formation of aberrant spindles induced by the compound were analyzed. In A2780 cells exposed for 24 h to 5 μM 4-carboxymethoxyimino-(4-HPR), the addition of 100 μM vitamin C abrogated ROS generation (Supplementary Figure S1) and caused a reduction of sub-G1 population but did not prevent the accumulation of G₂-M cell population or the formation of aberrant spindles (Figure 3A). The results indicated that in A2780 cells, the mitotic arrest induced by 4-carboxymethoxyimino-(4-HPR) was independent from ROS generation.

Finally, we investigated the effects of the 4-oxo-4-HPR derivative on the microtubule system of A2780 cells treated for 24 h with two different concentrations of the drug (i.e., 3 and 5 μM). Similarly to 4-oxo-4-HPR (Appierto et al., 2009a), sodium 4-carboxymethoxyimino-(4-HPR) decreased the polymerized portion of tubulin in a dose-dependent manner, as revealed by western blot analysis of free and polymerized tubulin (Figure 3B). Thus, the aberrant formation of mitotic spindles caused by the drug could be ascribed to the inhibition of microtubule polymerization induced by the retinoid in cultured cells, as previously described for 4-oxo-4-HPR (Appierto et al., 2009a). Taken together, our data indicated that sodium 4-carboxymethoxyimino-(4-HPR) entirely maintained the double mechanism of action of its parental drug in A2780 cells (Figure 4).

We next investigated whether the two mechanisms of action of 4-carboxymethoxyimino-(4-HPR) described in A2780 were recapitulated also in other human cancer cells responsive to the retinoid. Specifically, we analyzed the upregulation of PLAB and the cleavage of caspase-3 for the activation of the ROS-induced apoptotic cascade and the accumulation in G₂-M phases of the cell cycle for the antimitotic activity in STO (mesothelioma cells derived from a patient’s surgical specimen), IGROV-1 (ovarian cancer cell line), and MDA-MB-231 (triple negative breast cancer cell line) cells. In the three analyzed cell lines, the treatment with the drug induced both PLAB upregulation and caspase-3 cleavage and a marked G₂-M cell cycle arrest (Figure 5). The results confirmed that the distinctive feature of sodium 4-carboxymethoxyimino-(4-HPR) to have a double mechanism of action in solid tumors was not restricted to A2780 cells but represented a distinctive mode of action of the drug (Figure 4).

Previous in vivo preliminary study demonstrated that, compared to 4-oxo-4-HPR, sodium 4-carboxymethoxyimino-(4-HPR) reached 60-fold higher plasma levels, with extremely low variability and without apparent signs of toxicity for the mice (Musso et al., 2016). This promising evidence led us to investigate the antitumor activity of the compound in mouse models. Specifically, mice xenografted with human mesothelioma (STO), ovarian cancer (IGROV-1) and breast cancer (MDA-MB-231) cells were employed. Mice with mesothelioma xenografts were treated 5 days/week for 5 weeks with 30 or 60 mg/kg i.p. of sodium 4-carboxymethoxyimino-(4-HPR). As shown in Figure 6A, the treatment with the retinoid reduced the tumor growth at both doses, although a statistical significance was not reached. As regards to ovarian model, the mice were treated 5 days/week for 3 weeks with 30, 60, and 90 mg/kg i.p. of retinoid. Survival, TW and the number of ascitic cells were evaluated. At the doses of 60 and 90 mg/kg, the drug significantly increased the survival of the mice inoculated with IGROV-1 cells (p < 0.01; Figure 6B). Consistently, solid tumor mass and the number of ascitic cells were reduced in treated mice compared to control group of 62% (on average g: 0.59 ± 0.13 SEM and 0.22 ± 0.055 SEM, respectively) and 90% (on average cells: 165.5 × 10⁶ ± 68.40 × 10⁶ SEM and 20.13 × 10⁶ ± 15.74 × 10⁶ SEM, respectively).

Based on the results obtained in the previous models, we decided to test a single dose (90 mg/kg) for in vivo experiments on the breast cancer model. Specifically, the mice were subjected for 7 weeks to a 4 days/week treatment with the compound. As shown in Figure 6C, the retinoid significantly reduced the tumor growth of the triple negative breast cancer model (p = 0.01).

In all the experiments the compound administration was well tolerated with maximum body weight loss of 9% restricted to the first 3 days of treatment.

Overall our in vivo data demonstrated that sodium 4-carboxymethoxyimino-(4-HPR) deserves further investigations as a new potential anticancer drug able to exert its activity on different solid tumor types.