Indole and 2,4-Thiazolidinedione conjugates as potential anticancer modulators

Synthesis of TZD analogues

As reported in Fig. 1, the pharmacophore TZD [1] was prepared by treating chloroacetic acid and thiourea in the presence of the catalytic amount of HCl by refluxing in water. 1 was treated with DMF-DMA at 100 °C for 1 h furnishing 5-dimethylaminomethylene-thiazolidine-2,4-dione [2]. The structure of 2 was assigned on the basis of its IR, ¹H-NMR and LC-MS data. Treatment of 2 with indole [3] in AcOH at 100 °C for 1 h resulted in the formation of 5-(1H-indol-3-ylmethylene)-thiazolidine-2,4-dione AC1 in 95% yield. The obtained compound AC1 was found to be identical in melting point and TLC with the corresponding derivative prepared in the reported method (Tseng, 2017). Using this strategy, various new indolylidine TZD derivatives (AC1-20, Fig. 2) have been synthesized and their structures have been assigned on the basis of spectral data. However, AC1-20 could also be synthesized without isolating 2, using tandem approaches, by adding indole [3] in AcOH to the reaction mixture itself. On the other hand, isatin [4] condensed with TZD [1] in the presence of methanol/KOH under refluxing condition for 30 min, leading to the formation of the TZD derivative AC21. The structure of AC21 was assigned on the basis of its IR, ¹H-NMR and LC-MS data. Melting points are uncorrected and were determined in open capillary tubes in sulphuric acid bath for the crystalline products. TLC was run on silica gel G and visualization was done using iodine or UV light. Infrared (IR) spectra were recorded using Perkin Elmer 1000 instrument in KBr pellets. ¹H NMR spectra were recorded in DMSO-d₆ using TMS as internal standard, by using 400 MHz spectrometer. Mass spectra were recorded on Agilent LC-MS instrument.

General procedure for preparation of 2

A mixture of TZD 1 (10 mM) and DMF-DMA (10 mM) was stirred at 100 °C for 2 h, giving rise to a colourless solid that was separated from the reaction mixture. The solid was collected by filtration, washed with hexane (10 ml) and then dried. The crude product was recrystallized from ethanol solvent to get pure 2.

General procedure for preparation of AC1-AC20

A mixture of 2 (10 mM) and indole 3 (10 mM) was refluxed for 1 h in acetic acid. At the end of this period, a light yellow coloured solid separated out from the reaction mixture, which was collected by filtration. The isolated solid was washed with hot water (10 ml) and dried. The product was recrystallized from a suitable solvent to obtain AC1.

General procedure for preparation of AC21-26

A mixture of TZD 1 (10 mM), isatin 4 (10 mM) and methanol (10 ml), followed by 40% aqueous KOH, was refluxed for 30 min. At the end of this period, solid separated out from the reaction mixture, which was collected by filtration. The isolated solid was washed with methanol (10 ml) and dried. The product was recrystallized from a suitable solvent to obtain AC21.

Spectral data of compounds are detailed in the Appendix S1.

Cell cultures

MCF-7 human breast cancer cells and PC3 human prostate cancer cells were obtained from the American Type Culture Collection (LGC Promochem, Teddington, UK). Cells were maintained in DMEM, supplemented with 2 mmol/liter L-glutamine, 50 IU/ml penicillin, 50 µg/ml streptomycin, and 10% fetal bovine serum (Iiritano et al., 2012; Gruszka, Kunert-Radek & Pawlikowski, 2005), and both viability and cell migration were assessed as described elsewhere (Gruszka, Kunert-Radek & Pawlikowski, 2005) in the presence of either home-made TZD compounds, or commercially available rosiglitazone (rosiglitazone maleate, GSK Pharmaceuticals, Worthing, UK) (Gruszka, Kunert-Radek & Pawlikowski, 2005).

MTT Assay

The MTT assay is a colorimetric assay based on the transformation of tetrazolium salt by mitochondrial succinic dehydrogenases in viable cells. By measuring mitochondrial function and metabolic activity, the MTT test provides a reproducible quantitative indication of cellular viability. In serum-free media, cultured MCF-7 and PC3 cells were tested for viability, using 12 mM MTT reagent (Sigma-aldrich, St. Louis, MO) for 2 h at 37 °C. Formazan crystals were solved with 1 mL (12-wells plate) of an organic acid solution (40% dimethylformamide, 2% glacial acetic acid, 16% sodium dodecyl sulfate, pH 4.7), in order to avoid phenol red interference. Absorbance was read after 24, 48 and 72 h treatment at 540 nm.

Wound healing

The wound healing assay is a simple method to study cell migration and cell interactions in vitro. Wound healing was performed seeding 500,000 cells/well in a 12-wells plate in order to reach confluence after 24 h. Afterwards, scretches were made using sterile 200-µL tips, then, wells were washed three times with PBS and cells were treated with RPMI medium containing the respective TZD compound (5 µM final concentration in DMSO 0.001%). RPMI plus DMSO alone (0.001%) was used as vehicle control. Photos of the wound were taken at 4, 8 and 12 h. Then, snapshots were analyzed with the ImageJ software (Arcidiacono et al., 2015) and percent of wound clousure was calculated using the following equation: ${\displaystyle \text{\%}\text{wound closure}=\frac{\text{Area t}0-\text{Area tx}}{\text{Area t}0}\times 100}$

Protein extract and western blot analysis

Total and nuclear cellular extracts were prepared as previously (Brunetti, Foti & Goldfine, 1993; Brunetti et al., 1996), and Western blots were performed on total and nuclear extracts from PC3 and MCF-7 cells, after 24, 48 and 72 h treatment, as reported elsewhere (Bilotta et al., 2018). Protein extracts were resolved on sodium dodecyl sulfate polyacrylamide gel electrophoresis and electrotrasferred onto a 0.2 µm PVDF membrane (Merck-Millipore). Rabbit polyclonal antibodies anti-BCL-xL (cat. no 2762) and anti-cleaved PARP (C-PARP) (cat. no 9541) were purchased from Cell Signaling Technology. Monoclonal antibody anti-β Actin (cat. no A2228) was from Sigma-Aldrich Inc (Bianconcini et al., 2009). The resulting immunocomplexes were visualized by enhanced chemiluminescence, and densitometric slot blot analysis was performed by using the ImageJ software program.

Statistical analysis

Statistical significance was evaluated by Mann–Whitney test. p < 0.05 was considered significant. All bar graph data shown are mean ± S.E.

Molecular modelling studies

The Schrödinger Suite was adopted for computing all theoretical investigations (Software: Schrödinger, LLC, New York, NY. 2017). The 3D structure of AC-18, AC-20 and AC-22 were built using the Maestro GUI (Software: Maestro Schrödinger, LLC, New York, NY. 2017) and further submitted to LigPrep version 3.9 tool (Software: LigPrep Schrödinger, LLC, New York, NY. 2017) to take into account the most stable protomeric forms at pH 7.4. The Protein Data Bank (https://www.rcsb.org/) crystallographic entry structure 2PRG (Nolte et al., 1998) was prepared by using the Protein Preparation Wizard tool (Schrödinger, LLC, New York, NY, 2017). The molecular recognition evaluation was carried out by means of Glide software ^[g] in combination with Schrödinger Induced Fit docking (IFD) (Glide Schrödinger, LLC, New York, NY, 2017) protocol. The binding site of the target model was defined by means of a regular grid box of about 64,000 Å centered on the co-crystallized ligand. Docking simulations were computed using Glide ligand flexible algorithm version 6.7 at standard precision (SP).

Inhibition of cell viability in MCF-7 and PC3 cells

All the synthetic TZD-based compounds (from AC1 to AC26) were initially screened for their in vitro effect on cell viability, using MTT assays in MCF7 and PC3 cells, that underwent treatment with 5 µM TZD analogues over a period from 24 to 72 h of exposure (Figs. 3A and 3B). As observed both in MCF-7 cells and PC3 cells, not all compounds investigated produced the same effect on cell viability, as compared with untreated control cells. In fact, whereas some of them had no influence on this parameter at all time points of the 72-h period in both cell lines, a decrease of cell viability (with slight difference among MCF-7 and PC3 cells) was observed with AC11, AC12, AC14, AC15, AC17, AC18, AC20, AC21, AC22, AC23, AC24, AC25, and AC26 (Figs. 3A and 3B).

Next, both cell types were incubated with increasing concentrations (1 to 33.3 µM) of each of the twenty-six compounds individually to calculate IC₅₀ values (Table 1). Among all the compounds evaluated, compounds AC18, AC20 and AC22 were found to be the most promising, showing a 50% reduction of cell viability in both PC3 and MCF-7 cells after 48 h exposure to 5 µM concentration of each of the three drugs. However, whereas the inhibition of cell viability in MCF-7 cells was already significant at 48 h with all three TZD analogues, no significant inhibition of cell viability by these compounds was observed before 72 h incubation in PC3 cells (Figs. 3A and 3B), thus suggesting some component of cell specificity amongst these synthetic drugs.

To further evaluate the effect of AC18, AC20 and AC22 on cellular viability, we performed MTT assays, in which the effect of each of these compounds on cell proliferation was compared to that of untreated control cells, or cells treated with either the negative AC1 TZD compound, or 5 µM rosiglitazone, a potent oral antidiabetic drug, whose inhibitory role on cell viability has been reported in vitro, in rat prolactin-secreting pituitary tumor cells (Gruszka, Kunert-Radek & Pawlikowski, 2005). As shown in Figs. 4A and 4B, treatment of both MCF-7 and PC3 cells with 5 µM of either AC18, AC20 or AC22 induced a significant decrease of cell viability as compared with control cells, which, in some instances, was even more evident than that obtained with the same dose of rosiglitazone.

To further support the antiproliferative potential of AC18, AC20 and AC22, Western blot analyses of BCL-xL were carried out in cell lysates from MCF-7 and PC3 cells. BCL-xL is an important anti-apoptotic member of the BCL-2 protein family and a potent regulator of cell death. While overexpression of BCL-xL in PC3 and MCF-7 cells has been reported to contribute to the apoptosis-resistant phenotype of these cells in response to products with antitumor activity (Li et al., 2001; Okamoto, Obeid & Hannun, 2002), the reduction of Bcl-xL levels rendered these cells more sensitive to the drugs (Yu et al., 2014; Lin, Li & Zhang, 2016), thus reducing cell survival. As shown in Figs. 4A and 4B, BCL-xL protein levels were reduced in both MCF-7 and PC3 cells treated with the AC22 TZD compound, and this reduction paralleled the decrease in cell viability (as monitored by the MTT assay). Additionally, as shown in Fig. 4B, treatment of PC3 cells with either AC18 or AC22 showed a better inhibition of cell viability compared with cells treated with rosiglitazone, and also in this case a parallelism was observed between cell viability and BCL-xL protein expression. Therefore, it is tempting to hypothesize that the antiproliferative effect of these TZD compounds, including rosiglitazone, on MCF-7 breast cancer cells and PC3 prostate cancer cells could be mediated, at least in part, by the reduction in BCL-xL expression. These data were in agreement with the analysis of C-PARP, a marker of apoptotic response (Kraus, 2008), indicating that C-PARP expression was higher in MCF-7 and PC3 cells treated with the AC22 TZD compound than in rosiglitazone-treated cells.

Effects of AC18, AC20 and AC22 on cell migration

The effects of AC18, AC20 and AC22 on cell migration were studied as well. To this end, wound healing assays were performed in vitro with both MCF-7 and PC3 cells, either untreated or treated with the synthetic TZD agonists, at 4 h time points (up to 12 h). As shown in Figs. 5A and 5B, a significant inhibition of cell migration was observed in both MCF-7 and PC3 cells, in the presence of compounds AC18, AC20 and AC22, as demonstrated by the decreased migratory response of cells in response to wound healing. Also in this case, as for the MTT test, the effect of all three compounds on cell migration was comparable or even superior than that of rosiglitazone (Figs. 5A and 5B), thereby indicating that these novel compounds may indeed represent a new potential class of antiproliferative agents.

Theoretical binding affinity of compounds AC18, AC20, AC22 in the PPARγ active site in comparison to rosiglitazone

In order to understand the molecular interactions of AC18, AC20 and AC22 into the PPARγ active site, ligand-target recognition studies were performed using the Glide Suite Docking package (Glide Schrödinger, LLC, New York, NY) and refined with the Schrödinger Induced Fit Docking (IFD) protocol (Software: Induced Fit Docking protocol, Schrödinger, LLC, New York, NY) (Table S1, Appendix S1). We selected the rosiglitazone/PPARγ crystal structure (PDB ID: 2PRG) (Nolte et al., 1998) to carry out the docking simulations. We validated our protocol by calculating the Root Mean Square Deviation (RMSD) on the ligand Cα, thus demonstrating that the standard precision (SP) protocol was able to well reproduce the co-crystallized pose (rosiglitazone re-docking RMSD value: 1.38 Å). The receptor grid center was specified from the bound ligand and the energy window for ligand conformational sampling was 2.5 kcal/mol. In order to investigate the binding domain flexibility, the residues within 5.0 Å from the ligand poses were refined using the Prime molecular dynamics module (Prime Schrödinger, LLC, New York, NY, 2017). A maximum of 20 poses was generated. The docking results clearly indicated that all compounds are able to geometrically reproduce rosiglitazone binding mode (Fig. 6). Indeed, by analyzing the docking poses of all compounds, we observed that the ligands were well accomodated into the active site of PPARγ by establishing hydrogen bonds (HBs) between their carbonyl groups of the thiazolidine moiety with H343, Y473 and H449 residues. Furthermore, the two water molecules located into the binding site are involved in an indirect HB network between the compounds and the binding pocket residues. However, it is interesting to note that for AC20 and AC22 the HB with R288 is lost, while it is maintained by AC18. AC18 and AC20 share the same methoxyl group at position 5 of their indole ring, while in AC22 it is substituted with a fluorine atom. Probably the ethyl group on the amine of the thiazolidine ring, absent in AC18 and AC22 compounds, could be responsible for a different geometry of the molecule.