Synthesis, characterization, and pharmacological evaluation of novel azolo- and azinothiazinones containing 2,4-dihydroxyphenyl substituent as anticancer agents

Abstract We reported the synthesis and characterization of a series of azolo- and azino[1,3]thiazinones containing the 2,4-dihydroxyphenyl substituent. The compounds were prepared by a new one-step reaction of aryl-modified sulfinylbis[(2,4-dihydroxyphenyl)methanethione]s and the corresponding aminoazolo(azino)carboxamides. Their chemical structures were confirmed by IR, NMR: 1H, 13C, HSQC, and EI-MS spectral data. The compounds inhibited proliferation and viability of lung cancer A549, colon cancer HT-29, and glioma C6 cells in a structure- and concentration-dependent manner. The activity of some analogues was below 10 μmol dm−3 (IC50). Glioma C6 cells were the most sensitive to tested compounds. Generally, the derivatives were not toxic for the skin fibroblast HSF culture. Moreover, some of them exerted a protective effect on the treated normal cells. Evaluation of compound properties in silico showed that they possess significant drug-like characteristics and most of them display a low toxicity. Graphical abstract

The studies carried out by our team of two fused of fiveor six-membered rings heterocycles with 2,4-dihydroxyphenyl moiety exhibited an interesting profile of anticancer properties. 4H-3,1-benzothiazin-4-ones show high activity against the human bladder cancer HCV29T, non-small cell lung carcinoma A549, breast cancer T47D, and rectal adenocarcinoma SW707 cells [32]. An antiproliferative effect of some analogues is on the level of cisplatin. Structure-activity elucidation exhibited that the presence of a chlorine atom or alkyl substituent (methyl or ethyl) in position 5 of the resorcinol ring has a beneficial effect on the potency of compounds. A similar spectrum of biological activity and analogous effects of resorcinol Scheme 1 modification on activity were found for 1,3-thiazolo [5,4b]pyridines [33] and 1H-benzimidazoles [34].
In this work to develop new routes for the diversely substituted drug-like heterocyclic scaffolds, we have targeted the 5-6 and 6-6 fused-ring systems of imidazo-, pyrazolo-, and pyridothiazin-4-ones incorporating modified or unmodified 2,4-dihydroxyphenyl substituent. It was assumed that the presence of additional nitrogen atoms compared to 4H-3,1-benzothiazin-4-one can enhance interactions with a potential molecular target and intensify biological activity. Antiproliferative effect of the obtained compounds against the cells of some cancer lines and cytotoxicity against normal cells were described. Additionally, ADMET properties in silico of compounds were evaluated.
The reagents were applied in equimolar proportions. The reactions were performed in MeOH (sometimes with the addition of pyridine) under reflux (2-4 h) with moderate to good yields (70-88 %). STB and its analogues as the initial reagents were obtained from 2,4-dihydroxybenzenecarbodithioic acid or its analogues and SOCl 2 in diethyl ether according to the previously presented method [35].
The contents of C, H, and N were within ±0.4 % of the theoretical values. The mass spectra (EI, 70 eV) of all derivatives exhibited the molecular ion peak [M] ? of the maximal intensity (B = 100 %). The derivatives of group 2 and ethyl analogue 1c showed the peak at m/z [M-15] ? formed after demethylation of molecular ion.
The IR spectra of compounds exhibited a broad strong band in the range of 3462-3100 cm -1 of m(O-H). A band in the 1692-1649 cm -1 range was attributed to C=O group. These spectra confirmed the presence of [C=Nmoiety stretching in the region around 1640-1575 cm -1 . The 13 C NMR spectra of compounds showed signal at 172-165 ppm attributed to carbon atom of C=O group. Proton of OH group was sometimes invisible (compounds 1b, 4c). CH proton of heterocyclic scaffold of all compounds appeared in the range of 8.4-8.0 ppm. NH proton is registered in the wider range of 14.0-12.5 ppm. The crosspeaks of HSQC NMR spectrum of compounds 1b, 1e, and 2c presented in Table 1 further support a formation of new compounds.

Antiproliferative activity
The anticancer effect of compounds was tested in nonsmall cell lung cancer A549, colon carcinoma HT-29, and glioma C6 cells. The cells were exposed to either culture medium as the control culture or the compounds in concentrations of 10, 25, 50, and 100 lmol dm -3 . After 96-h treatment MTT assay was performed and IC 50 (concentration that produced a 50 % viability decrease of cells) values for each cell line were calculated using computerized linear regression analysis of quantal log dose-probit functions [36]. Cisplatin was used as a reference drug.
The tested compounds inhibited proliferation and viability of cells in a structure-and concentration-dependent manner. Glioma C6 cells proved to be the most sensitive to compounds action (Fig. 1). The obtained IC 50 values are presented in Table 2. The most active were compounds 1a and 3c with the IC 50 values 17.2 and 8.2 lmol dm -3 , respectively (cisplatin: 0.4 lmol dm -3 ). The evident effect was also detected for compounds 1b, 1c, 2c, 2d, 3a, and 4c ( Fig. 1; Table 2).
Additionally, the influence of compounds on normal human skin fibroblast primary culture (HSF) was assessed. HSF cells were exposed to compounds (10-100 lmol dm -3 ) for 24 h and the level of lactic dehydrogenase (LDH) released from the damaged cells was measured (LDH method). The obtained results (Fig. 4) demonstrated that the tested derivatives were relatively low toxic for normal cells. Only compound 2c in all tested concentrations and 2b in the concentration 100 lmol dm -3 were highly toxic for skin fibroblasts. What is interesting is the fact that compounds 1a, 1c, 3a, 3b, 3c, and 4a induced a decrease of death cells in the culture. This may indicate protective properties of the compounds.

ADMET properties in silico
The principle goal of the in silico calculation of ADMET properties of compounds is the prediction of their in vivo biokinetics as potential drugs [37,38]. ADMET Predictor Fig. 1 The antiproliferative effect of compounds in rat glioma C6. The cells were grown in the culture medium only (control) and in the presence of tested compounds (10-100 lmol dm -3 ) for 96 h, and the MTT assay was performed. The data represent mean % of control viability ±SEM of six trials and were analyzed by means of linear regression 7.1 was applied to predict ADMET properties of the compounds under consideration [39].
An oral administration is a commonly used route for drugs and a required one for new agents. Absorption of drugs after oral administration is a very complicated process and a number of parameters for its prediction are used. Relatively simple parameters are molecular weight (M), hydrophobicity: log D, log P from different approaches [40], polar surface area (PSA), and a number of hydrogen bonding atoms (HBA, HBD). They are included in the Lipinski's rule of five [41]. The data presented in Table 2 show that all considered descriptors are in the recommended range (the number of unfulfilled rules = 0). Two other parameters included in the Oprea's criteria: the number off rotatable bonds (nRB) and polar surface area (PSA) possess also the recommended values (nRB \ 10; PSA \ 120 Å 2 ) [37]. This shows that the compounds under consideration possess strong drug-like properties.
A more sophisticated model of absorption prediction takes into account the values of the human jejunal effective permeability (Peff) and apparent permeability (Papp) for Madin-Darby Canine Kidney (MDCK) cells. The data collected in Table 2 show that for all compounds medium permeability in the range of 103-250 9 10 -7 cm s -1 is assumed [42]. The estimated Peff parameter is larger than 1.8 9 10 -4 cm s -1 . The values of both descriptors are the highest for the compounds of group 2 and for analogue 5 which are characterized by the highest lipophilicity and the lowest PSA. Good native water solubility (S) for all compounds is estimated ([10 lg cm -3 ) [43][44][45].
The PPB parameter was also calculated ( Table 2). It describes the overall fraction of a drug bound in human blood plasma (in %). The half of the studied compounds indicate strong binding to proteins of plasma (PPB [ 95 %) and for some of them a risk associated with this property (1d, 2d, 2e, 3d, 4c, 5) is predicted. Two compounds show very weak affinity for them (PPB \ 90) (3a, 3b). Generally, a less bounded drug can pass through cell membranes and diffuse more effectively.
Prediction of tissue distribution of a compound is another important task in the drug development. Simple descriptors such as log P and M are in the optimal range for the studied compounds. The more advanced parameters were also taken into account: volume of distribution (V d ) and the descriptors describing the blood-brain barrier (BBB) penetration [46]. The compounds possess V d in the range of 0.23-1.18 dm 3 kg -1 , and the highest one was found for compounds 3a-3c and 5. To predict the BBB penetration, log BB (logarithm of the brain/blood partition coefficient) and BBB filtering (qualitative likelihood high/ low of crossing the blood-brain barrier) were calculated. According to the data collected in Table 2 all studied compounds show low brain penetration (log BB \ -0.07) and they rather do not across BBB [47]. Anticipating metabolism of compounds we have focused on cytochrome P450 (CYP) enzymes, which are the major enzymes involved in drug metabolism [48]. The following enzymes in human were taken into consideration: CYP 1A2, 2B6, 2C9, 2D6, and 3A4. Table 3 shows that all compounds may be substrates of CYP 1A2 in human and some of 2C9. The data presented in Table 3 show that intrinsic clearance (Clint) due to metabolism mediated by CYP 1A2 in human is significantly lower than via 2C9 enzymes.
Toxicity of compounds, which is a major reason for drug candidate failure, is also estimated. It was presented in two ADMET risk models: TOX MUT Risk and TOX Risk developed by Simulation Plus, Inc. (Table 3) [49]. TOX MUT Risk is a summary of the outputs of the ten different TOX MUT models that independently predict the mutagenicity expected for five strains of S. typhimurium with and without microsomal activation. The TOX Risk model consists of seven rules of different toxicities. Table 3 shows that toxicity in Tox Risk model is not predicted for compounds of group 3. A low toxicity for them is anticipated in Tox Mut Risk model. Predicted toxicity for other compounds is also relatively low and may be    connected with carcinogenicity in chronic rat studies (Xm) and hepatotoxicity (Hp) or with mutagenicity in S. typhimurium. The calculated acute rat toxicity values (Tox rat) show that compounds are characterized by medium toxicity in the range of 487-1474 mg kg -1 predicted for rats after oral administration. ADMET Risk (global ADMET risk), a computational filter developed also by Simulations Plus Inc [49], was additionally applied. The results collected in Table 3 show that compounds of group 3 with imidazo[4,5-d] [1,3]thiazin-7(3H)-one skeleton, of the lowest lipophilicity in the studied group of compounds, show the best ADMET properties in the group of the studied compounds. Low risk for compounds 4 was also calculated (\1.6). For comparison, ADMET Risk is larger than 6.5 for about 10 % of the drugs focused WDI [49].

Conclusion
To sum up, we have obtained and characterized a series of azolothiazinones as a new group of heterocyclic compounds possessing the 2,4-dihydroxyphenyl substituent. They were prepared in the one-step novel efficient synthesis procedure. The compounds displayed diverse antiproliferative activities against cancer cell lines. The most sensitive were glioma C6 cells and the most resistant colon carcinoma HT-29. The antiproliferative potency of the most active analogues was below 10 lmol dm -3 . The majority of the tested compounds were not toxic for normal skin fibroblast culture. Moreover, some of them increased fibroblasts viability. Furthermore, the compounds possess strong drug-like properties and good pharmacokinetics as well as low toxicity is predicted for them in silico. In the light of the presented results, compounds 3c seem to be most promising. They also provide an opportunity of laying the foundation for development of more promising molecules of anticancer potency.

Experimental
Melting points were determined using a BÜ CHI B-540 (Flawil, Switzerland) melting point apparatus. The elemental analysis (C, H, N) was performed on Perkin-Elmer 2400. The IR spectra were measured with a Perkin-Elmer FT-IR 1725X spectrophotometer (in KBr) or a Varian 670-IR FT-IR spectrometer (ATR) in the range of 600-4000 cm -1 . NMR spectra were recorded in DMSO-d 6 using a Bruker DRX 500 instrument. Chemical shifts (d/ ppm) were described in relation to tetramethylsilane (TMS). The MS spectra (EI, 70 eV) were recorded using the apparatus AMD-604.

Proliferation assay
Cancer cells were plated on 96-well microplates at a density of 1 9 10 4 (A549), 3 9 10 4 (HT-29), and 0.5 9 10 4 (C6) cells cm -3 . Next day, the culture medium was removed and the cells exposed to serial dilutions of compounds (10,25,50, and 100 lmol dm -3 ) in a fresh medium. Cell proliferation was assessed after 96 h using the MTT method in which the yellow tetrazolium salt (MTT) is metabolized by viable cells to purple formazan crystals. The cancer cells were incubated for 3-4 h with MTT solution (5 mg cm -3 ). Formazan crystals were solubilized overnight in the SDS buffer (10 % SDS in 0.01 mol dm -3 HCl) and the product quantified spectrophotometrically by measuring absorbance at the 570 nm wavelength using a Elx800 microplate reader (BIO-TEK, Highland Park, Winooski, Vermont, USA).

Cytotoxicity assay
Skin fibroblasts HSF were plated on 96-well microplates at a density of 1 9 10 5 cells cm -3 . The following day, the culture medium was removed and the cells exposed to serial dilutions of compounds (10,25,50, and 100 lmol dm -3 ) diluted in a fresh culture medium with a reduced amount of FBS (2 %). Cytotoxicity was detected after 24 h with the use In Vitro Toxicology Assay Kit, Lactic Dehydrogenase based (Sigma). The assay is based on the reduction of NAD by the action of lactic dehydrogenase (LDH) released from damaged cells. The resulting NADH is utilized in the stoichiometric conversion of a tetrazolium dye. The resulting coloured compound is measured spectrophotometrically. The test was carried out according to the kit procedure. The colour product was quantified spectrophotometrically at 450 nm wavelength using an Elx800 microplate reader.

Statistical analysis
Statistical analyses were performed with the use of GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA, USA) and Microsoft Office Excel 2007 computer software.

In silico ADMET evaluation
In silico ADMET evaluation of compounds was performed by ADMET Predictor version 7.1 [39]. Structures of the compounds were saved in the mol format using Chem Office software. Then, mol files of compounds were uploaded into the ADMET predictor software for further evaluation. All descriptors were estimated at pH 7.4.