Synthesis, Molecular Structure, Metabolic Stability and QSAR Studies of a Novel Series of Anticancer N-Acylbenzenesulfonamides

A series of novel N-acyl-4-chloro-5-methyl-2-(R1-methylthio)benzenesulfonamides 18–47 have been synthesized by the reaction of N-[4-chloro-5-methyl-2-(R1-methylthio)benzenesulfonyl]cyanamide potassium salts with appropriate carboxylic acids. Some of them showed anticancer activity toward the human cancer cell lines MCF-7, HCT-116 and HeLa, with the growth percentages (GPs) in the range from 7% to 46%. Quantitative structure-activity relationship (QSAR) studies on the cytotoxic activity of N-acylsulfonamides toward MCF-7, HCT-116 and HeLa were performed by using topological, ring and charge descriptors based on the stepwise multiple linear regression technique (MLR). The QSAR studies revealed three predictive and statistically significant models for the investigated compounds. The results obtained with these models indicated that the anticancer activity of N-acylsulfonamides depends on topological distances, number of ring system, maximum positive charge and number of atom-centered fragments. The metabolic stability of the selected compounds had been evaluated on pooled human liver microsomes and NADPH, both R1 and R2 substituents of the N-acylsulfonamides simultaneously affected them.


Introduction
Chemotherapeutics are the most important tool in the fight against cancer. For decades scientists have discovered many anticancer drugs by designing low-molecular weight compounds with the ability to bind to a preselected protein target or by screening small molecules for their ability to modulate a biological pathway in cells or organisms, or without regard for any particular protein target [1,2]. However, the question remains whether these drugs will offer lasting survival advantages to the patients in the light of the problem with of tumor unpredictability and drug resistance. A promising solution that is a focus of contemporary medicinal chemistry is the development of potent and selective anticancer drugs. Many of these new agents are being tested in combination with therapies currently used to treat specific cancers and usually include cytotoxic drugs. On the other hand, these newer agents are affording novel ways to mechanistically attack cancer, even if one cannot realize efficacy without toxicity [2].
N-Acylsulfonamides constitute a class of organic compounds with various types of pharmacological activities such as antibacterial inhibitors for tRNA synthetase [3], antagonists for angiotensin II [4], agents for the treatment of Alzheimer's desease [5], and osteoporosis [6]. In the last decade, particular attention has been focused on N-acylsulfonamides as cyclooxygenase [7], tubulin [8] and cyclin-dependent kinases [9] inhibitors with antiproliferative activity.
Our systematic studies on aryl-and heteroarylsulfonamides have resulted in promising anticancer agents [19][20][21][22][23][24], among which we have reported some N-acylsulfonamides (I, Figure 2) with antiproliferative activity against a broad spectrum of cancer cell lines [23]. Searching for innovative low-molecular chemotherapeutics we designed a series of new N-acylsulfonamides II ( Figure 2) with potential anticancer activity.   To explain how structural features influence the biological activities the quantitative structure-activity relationship (QSAR) method was applied [25]. Metabolic stability studies, one of the parameters characterizing the pharmacokinetic properties of a molecule, was performed using pooled human liver microsomes and NADPH.

Chemistry
The starting substrates, 3-aminobenzodithiazine 1, dipotassium salt 2, as well as potassium salts 3-9, were obtained according to the reported procedures for preparation of N-(benzenesulfonyl)cyanamide potassium salts [22,26,27]. Novel substrates 10-17 were synthesized analogously by nucleophilic substitution of arylmethyl chloride with dipotassium salt 2 as was shown in Scheme 1. The final N-acylbenzenesulfonamides 18-47 were achieved in the one-step reaction of N-[4-chloro-5-methyl-2-(R¹ -methylthio)benzenesulfonyl]cyanamide potassium salts with appropriate carboxylic acids under reflux. For solid acid reagents (i.e., trans-cinnamic and benzoic acids) reactions were ran in water or toluene, whereas for liquid reactants (acetic, propionic, isobutyric, cyclohexylpropionic acids) syntheses were carried out without solvent. The proposed mechanism for the formation of final N-acylsulfonamides 18-47 is outlined in Scheme 2. The initial step we believe is the protonation of the nitrile nitrogen atom by the carboxylic acid with simultaneous acceptance of the lone pair of electrons from the carboxylate anion (Step 1). Then, the nitrogen atom attached to the sulfonyl group attacks the carbonyl carbon atom causing C-O bond cleavage of the carboxylate group (Step 2). In the final step the proton from the =N-H group is transferred to the deprotonated sulfonamide nitrogen atom with simultaneous potassium cyanate elimination (Step 3).
Many methods are known for the acylation of sulfonamides. N-Acylbeznenesulfonamides can be prepared, for instance, by the reaction of a sulfonamide with a carboxylic acid in the presence of coupling reagents such as carbodiimides (DCC, EDCI), in which the by-product must be removed from the reaction mixture [8,28]. The N-acylation of sulfonamides has also based on the reaction with mixed anhydrides [9], acyl chlorides [29] or N-acylbenzotriazoles [30] in basic reaction media as well as under acidic medium with carboxylic acids or anhydrides in the presence of Lewis acid as a catalyst [31].
In the present study we utilized a new and facile method for the synthesis of N-acylated sulfonamides using N-(benzenesulfonyl)cyanamide potassium salts which can be applied as a convenient alternative to the known methods. Basic advantages of described approach include low cost, absence of coupling reagents and harmful substances, and easy isolation of the products.
The structures of the final compounds 18-47 were confirmed by IR, 1 H-NMR and 13 C-NMR spectroscopy. IR spectra showed the absence of the nitrile (C≡N) group and the presence of bands in the 1683-1734 cm −1 range corresponding to the stretching vibration of the carbonyl group (C=O). This functional group was also identified in the 13 C-NMR spectra where the N-acylbeznenesulfonamide moiety displays one carbon signal at 163. .90 ppm coming from the carbonyl carbon atom. The appearance of NH signals at 12.20-12.82 ppm in the 1 H-NMR spectra proved the presence of the amide group and also conformed the proposed structure of the final compounds. Moreover, X-ray analysis was done to confirm proposed structures on the representative compound 22.
Details on data collection, structure solution and refinement are given in Table 2. Compound 22 crystallized in the monoclinic system, the space group P21/c, with one molecule in the asymmetric unit. Atom numbering scheme was presented in Figure 3. The trifluoromethyl group was refined as disordered over two positions with probabilities 0.746(16)/0.254 (16). The molecule is bent at the -CH2-Sbridge in such a way that the acyl group is placed over the Ph-CF3 ring. The valence angle at S2, ca. 103° is typical for a thioether group. Notably, the S-C bond length to aliphatic carbon is longer than the bond to aromatic ring, which may indicate to some mesomeric effect. Since the sulfonamide group is acylated only one N-H hydrogen-bond donor is available. The intermolecular hydrogen bonding, of the N-H…O type, forms infinite chains (graph set symbol C(4), [32]) spreading along the b axis ( Figure 4, Table 3). The chain propagates exactly along the crystal screw axis 21 so it is described by the 1121 (R9) rod group symmetry [33].   Oxygen atom O(1), involved in the interaction, is a little more distant from S(1) than the other oxygen atom O3 (1.438(2) vs. 1.430(2) Å), which is expected and confirms the hydrogen bonding. The chains are packed in the crystal in such a way that allows ring stacking interactions between ring C1-C6 and its symmetry equivalent related by inversion centre at (½ ½ 1) and ring C9-C14 and its equivalent related by inversion centre at (1 1 1) (see Figure 5).

Cytotoxic Activity
Compounds 19-47 were evaluated in vitro for their effects on the viability of three human cancer cell lines: MCF-7 (breast cancer), HCT-116 (colon cancer) and HeLa (cervical cancer). Cytotoxic evaluations were performed in a 5-dose assay and data is reported as the percent growth of the treated cells. Results in Table 4 show the growth percent (GP) at two representative doses, i.e., 10 and 100 µM. The most active compound 27 belonged to the 1-naphthyl series (R 1 = 1-naphthyl) and showed outstanding anticancer activity at low and high concentrations against all tested cell lines, especially towards to the HeLa cell line (GP = 7% at 100 µM, GP = 22% at 10 µM ).  It was noted that changing R 1 = 1-naphthyl in active compound 27 to the corresponding 2-naphthyl isomer (compound 28) markedly decreased this activity. Another 1-naphthyl derivative 47 also stood out from the other compounds and demonstrated dilution-dependent good cytotoxic activity toward the three cell lines (9% ≤ GP ≤ 46% at 100 µM ). The significance of the 1-naphthyl substituent for the anticancer activity emphasize the fact that, replacement of this group (R 1 = 1-naphthyl, 47) by phenyl reduced activity of compound (R 1 = phenyl, 46, 79% ≤ GP ≤ 84% at 100 µM ).
For the most active compounds 27, 44, 45 and 47, the concentration required for 50% inhibition of cell viability IC50 was calculated and compared with the reference drug cisplatin, and the results are given in Table 5. Despite our observations concerning mentioned above correlations between structure and activity, we decided to apply more objective methods of establishing the structure-activity relationships, such as QSAR methodology.

Quantitative Structure-Activity Relationships (QSARs) of Cytotoxic Activity
The exploration of the relationship between structure and cytotoxic activity can be carried out by QSAR study. Recently, pharmacophore-based approaches are considered as the most advanced and successful, however if the molecular target of a group of compounds is not known, traditional regression-based techniques can be applied.
Multiple linear regression technique (MLR) along with stepwise algorithm usually leads to a statistically significant equation, that can be interpret based on a values of selected descriptors. Many pitfalls and mistakes can be done during development of a QSAR model. Among them, overfitting is one of the most common, and may results in false interpretation of a QSAR model. The second crucial issue is to develop a model in a way that enables to interpret quantitative relationship in the view of chemical structure. In order to achieve informative and statistically significant relationship we carefully designed several QSAR models using three-dimensional structures developed in Gaussian software and activity data taken from three distinct cell lines. Only selected descriptors available in DRAGON package were used (Table 6). MLR gave three statistically significant equations, one for each cell line ( Table 7). The comparison between experimentally observed and predicted GP values is visualized in Figure 6. For the MCF-7 cell line, three descriptors were pointed out as significant for compounds' activity. First of all the number of ring systems (NRS), with its positive coefficient shows that additional rings generally decrease the biological activity. CATS2D_06_DL shows a preferable topological distance (6 bonds) between lipophilic and H-bond donor centers and F07[C-C] supplements its performance indicating a preferred distance between C atoms ( Figure S1, see Supplementary Data). QSAR model for HeLa cell line contains four molecular descriptors. It shows two beneficial features: frequency of C-S atom pair at topological distance 8 and positive charge at the S atom of the sulfonamide moiety. Also two features are shown to decrease activity: topological distance between H-bond donor and acceptor (seven bonds) and the longest geometric distance from the atom to any other atom in the molecule (molecule width) ( Figure S1, see Supplementary Data).   The HCT-116 QSAR model showed the best performance (R 2 = 0.89; Q 2 = 0.83) for both the training and test sets of compounds ( Figure 6C). This points out the increase of biological activity with the increased number of lipophilic centers at topological distance 3 and the frequency of C-S atom pair at distance 5. On the other hand, C(Ar)-CH-C(Ar) fragment as well as presence of C-S at topological distance 10 were showed to be unfavorable ( Figure S1, see Supplementary Data).

Metabolic Stability
Most small-molecule anticancer drugs are cytotoxic agents with a narrow therapeutic window. Moreover their activity can be influenced by significant inter-and intra-subject variability, which can lead to life-threatening consequences. The probability to develop a successful drug can be increased by assessments of drugs' pharmacokinetic characteristics, i.e., absorption, distribution, metabolism and excretion properties, as early as possible during the preclinical stage. Among them metabolism is very important because it involves conversion of a drug candidate into a set of metabolites that differ in both activity and toxicity, leading to complex interactions within the human body. Metabolic stability can be assessed by incubation of a potent drug in a presence of liver microsomes and NADPH to give an insight to metabolic properties [34][35][36].
Selected compounds were submitted to in vitro metabolic stability study in the presence of human liver microsomes and NADPH ( Table 8). Determination of metabolic half-time (t1/2) was performed by LC-MS analysis. Metabolic stability in this set of compounds, expressed as t1/2, depends on a type of substituent in R 1 and R 2 positions. The most liable derivatives were 42 and 44 with t1/2 below 5 min. Among the most potent compounds 27 and 47, the first one was more stable with t1/2 24.23 min., which is a moderate value in a studied set of compounds. Two compounds substituted with ethynyl (compound 19) and amide (compound 20) moieties were characterized by a high metabolic stability, exceeding 40 min. Interestingly, big difference in t1/2 value was observed between 28 and 27. The increased metabolic stability results from the shift of naphthyl substituent from 1-to 2-naphthyl position. In order to describe this phenomenon in more detailed way, we decided to use in silico approach. Recently, Zaretzki and co-workers developed on-line accessible tool for accurate prediction of xenobiotic sites of metabolism, called XenoSite [37]. Among available models, one is able to predict stability measured in a presence of human liver microsomes. Taking into account in silico results illustrated in Figure 7, the relation with experimental determination of metabolic stability is clear. Compound 27 has two more sites vulnerable for metabolic biotransformation than 28, one in naphthalene ring and in methylene linker.

Figure 7.
Sites of metabolism predicted for 27 and 28 by XenoSite software [37]. Green color indicates more vulnerability to biotransformation than blue. Some significant differences are additionally pointed out.
Among the most potent compounds, 44 bears an ethylcyclohexyl moiety, which is probably the reason for its low stability. This indicates that this substituent should be excluded from further synthesis in this series of derivatives. On the other hand, although 2-naphtyl substitution is preferable for metabolic stability in comparison with a 1-naphtyl moiety, the high biological response observed for the latter one predisposes it to be a lead compound, as it shows a good balance between stability and cytotoxic properties.

X-ray Structure Determination
Experimental diffraction data were collected on a KM4 CCD kappa-geometry diffractometer (Oxford Diffraction, Yarnton, Oxfordshire, UK), equipped with a Sapphire2 CCD detector. An enhanced X-ray Mo Ka radiation source with a graphite monochromator was used. Determination of the unit cell and diffraction data collection were carried out at room temperature (298 K). All calculations (data reduction, structure solution, and refinement) were carried out using CrysAlisPro package [40].
The structure was solved by direct methods, and all nonhydrogen atoms were refined with anisotropic thermal parameters by full-matrix least squares procedure based on F 2 . Final refinements were carried out using the SHELX-97 package [41], run under control of WinGX program [42]. Scattering power of all the crystals tested was low, so in spite of long frame exposure time (240 s), the ratio of observed to unique reflections is only 40%.
Crystallographic data for structure of 22 reported in this paper have been deposited with the Cambridge Crystallographic Data Center as supplementary publication No. CCDC 1400372. Copies of the data can be obtained free of charge on application to CCDC, 12 Union Road, Cambridge CB2 1EZ, UK (Fax: +44-1223-336-033; Email: deposit@ccdc.cam.ac.uk).

Cell Culture and Cell Viability Assay
All chemicals, if not stated otherwise, were obtained from Sigma-Aldrich (St. Louis, MO, USA). The MCF-7 cell line was purchased from Cell Lines Services (Eppelheim, Germany), the HeLa and HCT-116 cell lines were obtained from the Department of Microbiology, Tumor and Cell Biology, Karolinska Institute (Stockholm, Sweden). Cells were cultured in in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 2 mM glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin. Cultures were maintained in a humidified atmosphere with 5% CO2 at 37 °C in an incubator (HeraCell, Heraeus, Langenselbold, Germany).
Cell viability was determined using the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide) assay. Cells were seeded in 96-well plates at a density of 5 ×10 3 cells/well and treated for 72 h with the examined compounds in the concentration range 1-100 μM (1, 10, 25, 50 and 100 μM). Following treatment, MTT (0.5 mg/mL) was added to the medium and cells were further incubated for 2 h at 37 °C. Cells were lysed with DMSO and the absorbance of the formazan solution was measured at 550 nm with a plate reader (1420 multilabel counter, Victor, Jügesheim, Germany). The optical density of the formazan solution was measured at 550 nm with a plate reader (1420 multilabel counter, Victor, Germany). The experiment was performed in triplicate. Values are expressed as the mean ± SD of at least three independent experiments.

Molecular Modeling Methodology/Calculations
The two-dimensional structure of the 29 studied compounds were built using the graphical user interface of Gaussian software (Gaussian 03W, v 6.0, Gaussian Inc., Wallingford, CT, USA), before starting molecular modeling algorithms, in order to obtain the lowest energy geometry of a molecule. The resulting structures were optimized by means of a semi-empirical method (with the use of the AM1 Hamiltonian). The developed three-dimensional structures were submitted to descriptor calculations using Dragon software. Among over four thousand descriptors available, only selected blocks were calculated: Constitutional Descriptors, Ring Descriptors, Charge descriptors, Topological indices, Geometrical descriptors, Functional group counts, Atom centered fragments, Atom-type E-state indices, CASTS2D descriptors, 2D atom pairs, Molecular properties and subsequently used in the QSAR study.

QSAR Study
Multiple linear regression with stepwise algorithm was performed and evaluated in Statistica software (Statsoft, Tulsa, OK, USA). 473 descriptors were used as a candidate for independent variables and percent of growth (GP) was used as a dependent variable. Dataset was randomly divided into training (n = 22, 75% of compounds) and test set (n = 7, 25% of compounds).

Metabolic Stability
Stock solutions of studied compounds were prepared at concentration of 10 mM in DMSO. Working solutions were prepared by dilution of stock with reaction buffer or acetonitrile, final concentration of organic solvent did not exceed 1%. Incubation mixture contained 10 μM of a studied derivative, 1 mM of NADPH (Sigma-Aldrich) and 0.5 mg/mL of human liver microsomes (HLM, Sigma-Aldrich) in potassium phosphate buffer (0.1 M, pH 7.4). Incubation was carried out in thermostat at 37 °C and started by addition of compound of interest. 50 μL samples were taken after 5, 15, 30, 45 and 60 min. Enzymatic reaction was terminated by the addition of the equal volume of ice-cold acetonitrile containing 10 μM of internal standard (IS). Compound 47 served as IS for compound 19. Compound 19 served as IS for the remaining compounds. Control incubations were performed without NADPH to assess chemical instability. After collection, samples were immediately centrifuged (10 min, 10,000 rpm) and resulted supernatant was directly analyzed or kept in −80 °C until LC-MS analysis. Natural logarithm of a compound over IS peak area ratio was plotted vs. incubation time. Metabolic half-time (t1/2) was calculated from the slope of the linear regression, as reported in [43].
LC-MS analysis was performed on an Agilent 1260 system coupled to SingleQuad 6120 mass spectrometer (Agilent Technologies, Santa Clara, CA, USA). Cortex C18+ column (2.1 × 75 mm, 2.7 μm, Waters, Milford, MA, USA) was used in reversed-phase mode with gradient elution starting with 30% of phase A (10 mM ammonium formate in water) and 70% of phase B (10 mM ammonium formate in acetonitrile-water mixture, 95:5 v/v). The amount of phase B was linearly increased to 100% in 12 min. Total analysis time was 20 min at 40 °C, flow rate was 0.25 mL/min and the injection volume was 10 μL. The mass spectrometer was equipped with electrospray ion source and operated in positive ionization. Mass analyzer was set individually to each compound to detect [M + H] + pseudomolecular ions. MSD parameters of the ESI source were as follows: nebulizer pressure 50 psig (N2), drying gas 13 mL/min (N2), drying gas temperature 300 °C, capillary voltage 3.5 kV, fragmentor voltage 100 V.

Conclusions
We have developed a new and facile method for the synthesis of N-acylated sulfonamides using N-(benzenesulfonyl)cyanamide potassium salts and carboxylic acids. The molecular structures of novel compounds were confirmed by NMR, IR, and also by single crystal X-ray structure analysis for representative compound 22. The newly synthesized N-acylsulfonamides were tested for their in vitro cytotoxic activity against the MCF-7, HCT-116 and HeLa cell lines. The most active compounds belonged to a series of 1-naphthyl derivatives. Compound 27 showed outstanding anticancer activity at low and high concentrations against all tested cell lines, especially with regard to the HeLa cell line (GP = 7% at 100 µM, GP = 22% at 10 µM ). Compound 47 demonstrated dependent on dilution good cytotoxic activity toward three cell lines (9% ≤ GP ≤ 46% at 100 µM ). Performed QSAR analysis demonstrate the importance of some molecular descriptors in the cytotoxic activities against the investigated cell lines. The MCF-7 and HCT-116 QSAR models showed that antitumor activity of N-acylsulfonamides depends on distinct topological descriptors and number of fragments C(Ar)-CH-C(Ar). On the other hand, QSAR analysis indicated the significance of topological descriptors as well as positive charge at sulfonamide S atom in the prediction of cytotoxic activity against HeLa cell line. Finally, the QSAR studies revealed three predictive and statistically significant models for the investigated compounds, with squared correlation coefficients for training set R 2 = 0.76-0.89 and squared correlation coefficients for test set Q 2 = 0.59-0.83. In vitro tests for metabolic stability in the presence of human liver microsomes and NADPH run on selected compounds showed that chemical nature of both R 1 and R 2 substituents of N-acylsulfonamides affected their stability. In the studied series of N-acylbenzenesulfonamides MLR-based considerations on structure-activity relationships served as a supporting approach beneficial for explanation of biological response. The usage of easily interpretable molecular descriptors is a base to plan further, rational synthesis of potentially more potent derivatives.