key: cord-0744160-ahsyp92n
authors: Pomarnacka, Elżbieta; Kornicka, Anita; Kuchnio, Anna; Heinrichs, Maike; Grünert, Renate; Gdaniec, Maria; Bednarski, Patrick J.
title: Synthesis, Cytotoxicity Testing, and Structure–Activity Relationships of Novel 6‐Chloro‐7‐(4‐phenylimino‐4H‐3,1‐benzoxazin‐2‐yl)‐3‐(substituted)‐1,4,2‐benzodithiazine 1,1‐dioxides
date: 2011-05-25
journal: Arch Pharm (Weinheim)
DOI: 10.1002/ardp.201000183
sha: 3459256ecb52886484813bf4cae9ec3184cc953b
doc_id: 744160
cord_uid: ahsyp92n

A new series of 16 6‐chloro‐1,1‐dioxo‐7‐{4‐[(4‐R(1)‐phenyl)imino]‐4H‐3,1‐benzoxazin‐2‐yl}‐3‐(substituted amino)‐1,4,2‐benzodithiazines 7–22 was prepared in order to evaluate the cytotoxic activity against six human cancer cell lines. The structures of the new compounds were confirmed by IR, (1)H‐, and (13)C‐NMR, elemental analysis and in the cases of 11 and 31 by X‐ray crystal structure analysis. This analysis showed that contrary to our earlier report the structures contain a benzoxazine ring instead of the proposed quinazolinone ring. The bioassay indicated that the benzodithiazine derivatives 7–22 possess cancer cell growth‐inhibitory properties. Some compounds showed a high level of selectivity for certain cell lines. The most active compounds 11, 12, 16, 19, 21, and 22 exhibited potency higher or comparable to cisplatin. The compounds were particularly effective in LCLC‐103H and MCF‐7 cell lines with IC(50) values of 0.49–1.60 µM. Quantitative structure activity relationships (QSAR) revealed that a chloro substituent R(1) in the phenyl ring as well as the shape of the substituted amino group at R(2) (e.g., unsaturation is beneficial) are important for potency.

Aryl and heteroarylsulfonyl sulfonamides are attracting attention as anticancer agents. Our systematic studies on the synthesis of 1,4,2-benzodithiazine-1,1-dioxides and their subsequent transformation into N-(azolyl or azinyl)-2-mercaptobenzenesulfonamides A (Fig. 1 ) have resulted in promising anticancer agents [1] [2] [3] [4] [5] and potent inhibitors of HIV-1 integrase (MBSAs) [6] . We also found that cyclic sulfonamide derivatives of 2-amino-8-chloro-5,5-dioxo [1, 2, 4] triazolo [2,3-b] [1, 4, 2] benzodithiazine (B, Fig. 1) [7, 8] or of type C-D ( Fig. 1 ) possess interesting in-vitro anticancer properties [9] [10] [11] . In addition, various 1,3-benzoxazine derivatives have been found to show versatile bioactivities such as antimicrobial, antiviral, antifungal [12] , anti-human coronavirus [13] , and anticancer activity [14] .

In the search for more potent and selective agents against cancer, we report the synthesis of 16 new compounds of type E and the results of their in-vitro evaluation for cytotoxic activity. The scaffold of this class of compounds consists of two moieties: 1,1-Dioxo-1,4,2-benzodithiazine and 1,3-benzoxazine. The objective of this drug-design approach was to merge these two moieties to enhance their activity against cancer cells. Our previous studies with this class of compounds showed that electronic character of the benzodithiazine ring system substituent at position-3 was an important factor influencing cytotoxicity [15] . However, in this previous publication we reported that the compounds were quinazolinone derivates. Based on X-ray crystal structure analysis of two representative compounds we now correct the structures to be benzoxazine derivatives. Herein, the effects of further structural modifications on antitumor activity were explored within two structural domains: The benzodithiazine ring (substituent R 2 ) and substituents of benzoxazine moiety (R 1 -phenyl). A correlation between the structures of these derivatives and the potency to inhibit the growth of cancer cells was investigated by using quantitative structure activity relationship (QSAR) methods.

The previously described methods were employed for the synthesis of 6-chloro-3-methylthio-1,1-dioxo-1,4,2-benzodithiazin-7-carbonyl chloride 1 [16, 17] . The reaction of 1 with the appropriate 2-aminobenzanilide was carried out in boiling toluene in the presence of pyridine and afforded the expected N- [2-(phenylcarbamoyl) phenyl]-6-chloro-1,1-dioxo-3methylthio-1,4,2-benzodithiazin-7-carboxamide 2-3 in 80% yield. Treatment of 2-3 with an excess of thionyl chloride under reflux gave rise to the novel 6-chloro-7-{4-[(4-R 1 -phenyl)imino]-4H-3,1-benzoxazin-2-yl}-3-methylthio-1,4,2-benzodithiazine 1,1-dioxides 4, 5, and 6 [15] .

The spectroscopic data do not allow straightforward discrimination between the quinazolone structure and alternative benzoxazine. Therefore, X-ray crystallography was undertaken with a previously described compound 31 [15] as well as with a new one 11 with the goal to establish the more discrete structural features of these compounds. Unexpectedly, X-ray structure analysis showed that compounds are 6-chloro-7-{4-[(4-R 1 -phenyl)imino]-4H-3,1-benzoxazin-2-yl}-3-R 2 -1,4,2-benzodithiazine dioxides (Figs. 2-4), and not the quinazolinone structure we previously reported [15] .

The mechanism of the reaction pathway was not investigated but it can be postulated as follows. First, the reaction of carboxamide 2 or 3 with thionyl chloride involves the initial formation of unstable intermediate A, which with evolution of the hydrogen chloride giving rise to formation intermediate B. In turn, the latter intermediate subsequently undergoes an intramolecular cyclocondensation with evolution molecule of the sulfur dioxide to afford the final products 4-6. Carboxamides 2 or 3 did not undergo the intramolecular cyclization (Scheme 1) in the presence of dehydrating agents such as phosphorous oxychloride or thionyl chloride to obtain quinazolinone C [18] [19] [20] .

Furthermore, nucleophilic displacement of the 3-methylthiol group of 4-6 by the appropriate amine in boiling methanol proceeded with elimination of methyl mercaptane, leading to the target benzodithiazines 7-22 in 27-78% yields (Scheme 2). All final products were characterized by IR and NMR spectroscopy as detailed in the experimental section. Elemental analyses were in accordance with the proposed structures.

Molecules 11 and 31 [15] with their labeling scheme are shown in Figures 2 and 3 . Bonding geometries of the benzoxazin-4-imine and 1,4,2-benzodithiazin-3-amine 1,1-dioxide units are similar to those found in previously reported structures [10, 11] , [21] [22] [23] [24] . Bond lengths indicate a double bond character at C14-N19 [1.274(2) and 1.254(6) Å] and C16-N24 [1.262(3) and 1.257(6) Å], a relevant degree of single bond for C25-N24 [1.417(3) and 1.426(6) Å] and a strong conjugation in the amidine N3-C3-N31 fragment with the C-N bond lengths in the range 1.299(7) to 1.320(8) Å.

The analogous parts for the two molecules have very similar conformations as shown in Fig. 4 , where superposition of the molecular structures of 11 and 31 is presented.

A 1,1-dioxo-1,4,2-dithiazine ring with the lone pairs of electrons on the S atom is known to prefer the boat conformation [25] and this conformation is adopted by this ring in both molecules, leading to their butterfly shape. The central part that comprises a 3,1-benzoxazine system and the fragment of 1,4,2-benzodithiazine unit, consisting of the benzene ring and S substituent atoms, is virtually planar with dihedral angles between the two parts being 3.6 and 6.18 in 11 and 31, respectively.

A microtiter based assay based on the staining of adherent cells with crystal violet was used to quantify the antiprolifer- ative potential of the new compounds on human cancer cell lines. Details of this test have been published elsewhere [26, 27] . Primary screening of compounds 7-22 for antiproliferative activity took place on three human cancer cell lines: A-427, DAN-G and LCLC-103. All compounds showed inhibition of cell growth by more than 50% at 20 mM in one or more of the cell lines. Secondary screening to determine potency was performed on a panel of 6 human cancer cell lines: RT-4 (urinary bladder transitional cell cancer), 5637 (urinary bladder cancer), DAN-G (pancreas cancer), LCLC-103H (large cell lung cancer), A-427 (lung cancer), and MCF-7 (breast cancer). Table 1 lists the average IC 50 values calculated from the doseresponse data obtained from three independent experiments. The IC 50 is the concentration required to inhibit cell growth by 50% compared to the untreated control over a 96 h treatment period [26] .

The present results show that further structural modification of earlier tested benzodithiazine derivatives [15] can lead to further increases in potency. In the series of derivatives 7-22, the compounds differ in a substituent at the position 3 of the benzodithiazine scaffold and in substitution in the position 4 of the phenyl attached to the benzoxazine moiety (chlorine, hydrogen or methyl group).

Based to the substituent at position 3 (R 2 ), compounds can be divided into three groups: 9-11 possessing an aliphatic chain, 12-22 with a pyridine moiety and 7-8 with the morpholine entity. The cytotoxicity of compounds with the morpholine moiety is much lower than of the other benzodithiazines, although their selectivity towards the 5637 cell line is apparent. The most active group appears to be those with pyridine substituent. However, the mean IC 50 value of the compound 11 with allylamino group is lower comparable to pyridine substituted compounds 12, 16, and 21.

A comparison with activity of anticancer-agent cisplatin indicates that benzodithiazines 7-22 possess very good cell Scheme 1. Proposed mechanism of the formation of benzoxazines 4-6. growth inhibitory properties. The values of the IC 50 were taken from a previously published study [26] conducted in analogous conditions to our assay. In general these results show the greatest similarities in activity of tested benzodithiazines to the alkylating agents like cisplatin and DACH-Pt.

The LCLC-103H cell line is the most sensitive of the six cell lines. In this cell line some of the compounds showed similar (16, 19, and 21) or even greater (10-13 and 22) potency compared to cisplatin. Moreover, the RT-4 cell line was susceptible to cell growth inhibition by 12 < 19 < 22 < 11 < 14 < 16, which was similar or greater than cisplatin in the case of compounds 13, 21, and 18.

Quantitative structure activity relationship studies QSAR are frequently used in medicinal chemistry to establish a predictive relationship between structure and potency [28] . In search of possible QSAR with our data, multiple regression analysis was performed with 18 quantitative descriptors for the amines at R 2 and two indicator variables for either a chlorine atom or a methyl group at R 1 (see Table  S1 , Supplementary Material). The dependent variable was the -log of the IC 50 values. The compounds used in the QSAR analysis included the 16 compounds described in this work as well as 8 compounds, 23-30 (Scheme 2) reported in an earlier study [15] . The range in the IC 50 values for the Scheme 2. Synthesis of 6-chloro-7-(4-phenylimino-4H-3,1-benzoxazin-2-yl)-3-(substituted)-1,4,2-benzodithiazine dioxides 7-31. Table S1 ). Significant correlations where also found with three individual cell lines (LCLC-104H, RT-4, and MCF-7). The remaining three cell lines (DAN-G, A-427, and 5637) gave much poorer correlations, which were in many cases no longer significant (data not shown). Cross-validation of the data was performed by the leave-one-out procedure and is reported as the crossvalidated squared correlation coefficient rCV 2 .

Equations (1) and (2) show the results of the multiple regression analysis for two and three best variables for the average IC 50 values of all lines. The single most important parameter contributing to good activity is the presence of a chlorine atom at position R 1 of the phenyl ring, as evidence by the indicator variable I Cl in both equations. The shape of the substituent at R 2 also appears to influence activity; in the two variable equation the unsaturated surface area (USA) was identified as having a positive influence while in the three parameter equation the kappa shape index, order 1 (KSI-1) [29] has a positive effect while too large a molar volume (MV) is detrimental.

In the cases of three individual cell lines where significant correlations where found, the unsaturated surface area of the group at R 2 and the presence of a chlorine atom at R 1 were again important determinates for activity. This was particularly apparent when the data was fitted to the two best parameters, as in the cases of Eqs. (3), (5), and (7). In the cases of the cell lines LCLC-104H and RT-4, inclusion of a second indicator variable for the methyl group at R 1 improved the correlation considerably (see Eqs. (4) and (6)). This was not unexpected because the methyl is bioisoteric with Cl.

In the case of the MCF-7 cell line, the inclusion of the three best parameters in the regression equations still indicated that the shape of the substituent at R 2 is important but now the saturated surface area (SSA) is detrimental while the water accessible surface area (WASA) is beneficial to activity (Eq. (8)). The result that the saturated surface area is detrimental is consistent with the regression equations that showed the unsaturated surface area to be beneficial because the one parameter is the opposite of the other.

ÀlogðIC 50 Þ all lines ¼ 0:024ðAE0:006Þ USA þ 0:306ðAE0:091Þ I Cl þ 4:336ðAE0:249Þ

ÀlogðIC 50 Þ all lines ¼ 0:442ðAE0:094ÞKSI-1 À 0:024ðAE0:008ÞMV þ 0:367ðAE0:074ÞI Cl þ 4:811ðAE0:175Þ n ¼ 24 

The above data show the usefulness of uniting benzoxazine and benzodithiazine moieties to build a scaffold with good cytotoxic activity. It can be concluded that the shape (e.g., unsaturation) of the substituents R 2 of the benzodithiazine ring system as well as a chlorine or methyl group in the phenyl ring at R 1 positively influence the cytotoxicity potency of the compounds. Further structural modification along the lines of the QSAR may help to further increase the potency of this interesting new class of cytotoxic compounds.

Melting points are uncorrected and were determined on a Büchi SMP-20 apparatus (Büchi Labortechnik, Flawil, Switzerland). The IR spectra were recorded on 1600 FTIR Perkin Elmer (Perkin Elmer, Norwalk, CT, USA) spectrometer by using potassium bromide pellets and frequencies were expressed in cm À1 . The 13 Table S2 (Supplementary Material). The starting 6-chloro-3-methylthio-1,1-dioxo-1,4,2-benzodithiazin-7-carboxylic acid [16] , 6-chloro-3methylthio-1,1-dioxo-1,4,2-benzodithiazin-7-carbonyl chloride 1 [17] , N-[2-(phenylcarbamoyl)phenyl]-6-chloro-1,1-dioxo-3-methylthio-1,4,2-benzodithiazin-7-carboxamide [15] , and 6-chloro-1,1dioxo-3-methylthio-7-[4-(phenylimino)-4H-3,1-benzoxazin-2-yl]-1,4,2-benzodithiazine 6 [15] were obtained by the previously described methods.

N-[2-(R 1 -Phenylcarbamoyl)phenyl]-6-chloro-1,1-dioxo-3methylthio-1,4,2-benzodithiazin-7-carboxamides 2, 3

To a suspension of compound 1 (3.42 g, 10 mmol) and 2-amino-N-(p-tolyl or 4-chlorophenyl)benzamide (11 mmol) in anhydrous toluene (120 mL), pyridine (0.8 mL, 10 mmol) in anhydrous toluene (50 mL) was added. The reaction mixture was stirred at reflux for 48 h and then left overnight at room temperature. The precipitate was filtered off, washed with toluene (5 mL), dried and suspended in 0.5% aqueous K 2 CO 3 (200 mL). The mixture was stirred for 1 h, filtered off, washed successively with water (2 Â 50 mL), methanol (20 mL), and dried. In this manner, the following compounds were obtained. the calculations was the negative log(IC 50 ). The independent variables included the descriptors described above as well as two indicator variables, I Cl and I CH3 , which were set to 1 when R 1 was a chlorine or methyl group, respectively, or set to 0 when R 1 was hydrogen. The program was used to identify the QSAR equations with the data best fitted to one, two and three variables from all possibilities. Multiple regression analysis of the best-fitted equations with two and three variables was repeated with MS-Excel (ANOVA analysis). Leave-one-out crossvalidation was performed with the cross-validated squared correlation coefficient rCV 2 , correlation with the Pearson correlation coefficient r, and significance was determined by the F-test.

We wish to thank the EU for Erasmus scholarships to AK.

The authors have declared no conflict of interest.

-methylphenyl)imino]-4H-3,1-benzoxazin-2-yl}-3-[(6-chloro-3-pyridylmethyl)amino]-1,1-dioxo-1

51 g) and 5-(aminomethyl)-2-chloropyridine (0.18 g, 13 mmol); yield: 0.25 g (42%); mp 221-2238C. IR (KBr, cm À1 ) 3225 (NH)

7.11 (d, J ¼ 6.8 Hz, 1H, H-5 pyrid.), 7.13 (d, J ¼ 8.3 Hz, 2H, PhCH 3 )

MHz, DMSO-d 6 ): d 4.64 (s, 2H, NCH 2 ), 7.22 (d, J ¼ 8.7 Hz, 2H, PhCl), 7.28-7.46 (m, 4H, PhCl and H-3,5 pyrid.), 7.56-7.68 (m, 2H, aromat.), 7.81 (dd, J ¼ 6.6 Hz, J ¼ 7.6 Hz, 1H, aromat.), 8.12 (s, 1H, H-5 benzodit.), 8.23 (d, J ¼ 7.4 Hz, 1H, aromat.), 8.39 (s, 1H, H-8 benzodit.), 8.57 (br.s, 2H, H-6 and H-2

Anti-Cancer Drugs

Reviews of Computational Chemistry

CrysAlis Pro Software. vers. 1.171.33, Oxford Diffraction Ltd

The diffraction data for single crystals of compounds 11 and the previous described 31 [15] were collected at 130 K with a KumaCCD diffractometer using graphite monochromated Mo Ka radiation. The intensity data were collected and processed using Oxford Diffraction CrysAlis Software [30] . The structures were solved by direct methods with the program SHELXS-97 [22] and refined by full-matrix least-squares method on F 2 with SHELXL-97 [31] .Crystal data for 11: C 24 H 16 Cl 2 N 4 O 3 S 2 Á C 2 H 6 OS, triclinic, space group P-1, a ¼ 6.7802 (4) 25 149 data were collected up to 2 max ¼ 508 for a crystal with dimensions 0.6 Â 0.4 Â 0.01 mm 3 (R int ¼ 0.0748, R s ¼ 0.0977). Final R indices for 2718 reflections with I > 2s(I) and 372 refined parameters are:The phenylethylene fragment of the molecule is disordered over two orientations with the occupancy factors equal to 0.69 and 0.31. The disordered fragment was refined isotropically with common isotropic displacement parameter for each orientation and restraints imposed on its geometry.Crystallographic data for compounds 11 and 31 have been deposited with the Cambridge Crystallographic Data Centre, with the deposition Nos CCDC 768913 & 768914.

The microtiter plate method used for cytotoxicity testing is based on crystal violet staining of cells and has been described in detail elsewhere [26, 27] . All the cells were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ) (Braunschweig, FRG). Stock solutions of compounds were prepared in DMSO and diluted 1000-fold with cell culture medium (RPMI medium þ 10% fetal calf serum) for testing. For IC 50 determinations, all substances were tested at 5 serially diluted concentrations. The corrected T/C values (T/C corr ) for each concentration were calculated with the following equation: where T is the optical density at l ¼ 570 nm (OD 570 ) of the treated cells at after a 96 h treatment time, C is the OD 570 of the untreated cells after 96 h of growth without substance, C,0 is the OD 570 of the cells on at the time of treatment (i.e. 96 h before T and C). Linear regression analysis of the log concentration versus the T/C corr plots was used to estimate the concentration of substance that caused a T/C corr ¼ 50% (IC 50 ). At least three independent experiments were done to determine the IC 50 values.

Molecular models of the amines R 2 were constructed and minimized with the software PCModel (Serena Software, Bloomington, IN, USA, ver 8.50). The MMX force field of PCModel was used in the calculation of the following descriptors: the saturated surface area (SSA), the unsaturated surface area (USA), the polar surface area (PSA), the molecular volume (molvol) and the molar volume (Mvol) of each fragment. The structures were imported into the program CAChe (Fujitsu Biosystems Group, Beaverton, OR, USA, ver 7.5.0.85) and the following descriptors were calculated with the PM5 semi-empirical method at the PM5 minimum: Dipole moment (DM), HOMO and LUMO energies, molecular refractivity (MR), polarizability (P), heat of formation (HF), water assessable surface area (WASA) and ionization potential (IP). Further descriptors calculated by CAChe were: Molecular weight (MW), logP, and the kappa shape indexes, orders 1, 2 and 3 (KSI-1, KSI-2, KSI-3).

Initial multiple regression analysis was performed with the CAChe ProjectLeader software. The dependent variable used in