Synthesis, Anticancer Evaluation and Structure-Activity Analysis of Novel (E)- 5-(2-Arylvinyl)-1,3,4-oxadiazol-2-yl)benzenesulfonamides

To learn more about the structure–activity relationships of (E)-3-(5-styryl-1,3,4-oxadiazol-2-yl)benzenesulfonamide derivatives, which in our previous research displayed promising in vitro anticancer activity, we have synthesized a group of novel (E)-5-[(5-(2-arylvinyl)-1,3,4-oxadiazol-2-yl)]-4-chloro-2-R1-benzenesulfonamides 7–36 as well as (E)-4-[5-styryl1,3,4-oxadiazol-2-yl]benzenesulfonamides 47–50 and (E)-2-(2,4-dichlorophenyl)-5-(2-arylvinyl)-1,3,4-oxadiazols 51–55. All target derivatives were evaluated for their anticancer activity on HeLa, HCT-116, and MCF-7 human tumor cell lines. The obtained results were analyzed in order to explain the influence of a structure of the 2-aryl-vinyl substituent and benzenesulfonamide scaffold on the anti-tumor activity. Compound 31, bearing 5-nitrothiophene moiety, exhibited the most potent anticancer activity against the HCT-116, MCF-7, and HeLa cell lines, with IC50 values of 0.5, 4, and 4.5 µM, respectively. Analysis of structure-activity relationship showed significant differences in activity depending on the substituent in position 3 of the benzenesulfonamide ring and indicated as the optimal meta position of the sulfonamide moiety relative to the oxadizole ring. In the next stage, chemometric analysis was performed basing on a set of computed molecular descriptors. Hierarchical cluster analysis was used to examine the internal structure of the obtained data and the quantitative structure–activity relationship (QSAR) analysis with multiple linear regression (MLR) method allowed for finding statistically significant models for predicting activity towards all three cancer cell lines.


Introduction
Malignant diseases are one of the major threats to global public health. Based on the GLOBOCAN 2018 database, which gathered estimates of incidence and mortality for 36 types of cancer in 185 countries, the number of new cases of cancer in 2018 was estimated to be 18.1 million and the number of deaths from cancer was 9.6 million worldwide [1]. In 2009, the medical and indirect costs of new cancer cases globally were estimated to be 286 billion $ [2]. Despite advanced work on new methods of cancer therapy such as immunotherapy, nanoscale and nanostructure-based therapeutics or gene therapy, the main methods of choice remain surgery, radiotherapy and chemotherapy [3]. Therefore, the search for novel small molecule chemotherapeutic drugs is one of the major challenges of modern medicinal chemistry.
Based on obtained information, we decided to optimize the "hit" structure to better understand the structure-activity relationships in this group of promising anticancer compounds. Therefore, firstly we decided to test the effect of the change in the aryl ring in the vinyl position (Ar) by synthesis derivatives differentiated with substituted benzene ring and 5-or 6-membered heterocycles in that position. Secondly, we checked the significance of the benzylthio substituent at position 2 of the benzenesulfonamide scaffold by comparing compounds containing benzylthio substituent or a chlorine atom. Finally, we decided to check the significance of the sulfonamide moiety in the meta position relative to the oxadiazole ring by synthesizing analogous compounds based on the 4benzenesulfonamide and 2,4-dichlorobenzene scaffold ( Figure 2).
Although standard crystallization of crude product from glacial acetic acid in general yielded in pure E isomer, for several compounds we obtained an E/Z mixture in proportion 1/0.2-1/1 (according to 1 H NMR integration and HPLC analysis). Thus, taking effort to obtain a pure isomer, we observed that refluxing E/Z mixture in acetic acid for 24-48 h results in isomerization to pure E isomer. Isomer assignment was based on the value of 1 H NMR coupling constant between vinyl protons. According to Karplus relation for alkenyl vicinal protons, coupling constant is about 6-12 Hz for Z (cis) and [12][13][14][15][16][17][18] Hz for E (trans) isomer. For target derivatives 7-36 and 47-54, we observed coupling constants about 16 Hz for pure E isomer and in range 12-13 Hz for Z isomer (obtained only in inseparable E/Z mixture). The structures of all compounds were confirmed with spectroscopic methods (IR, 1 H NMR, 13 C NMR) and elemental analyses, as shown in the Materials and Methods section

Anticancer Activity
Compounds 7-36 and 47-54 were tested regarding their effect on growth of three human cancer cell lines: colon cancer HCT-116, breast cancer MCF-7 and cervical cancer HeLa, as well as on noncancerous keratinocyte cell line HaCaT. Cell viability was measured with MTT assay after 72 h of incubation with tested compound in five concentration 1-100 μM. The MTT test enables us to quantify the living cells by measuring the activity of mitochondrial enzymes, enables reducing tetrazolium dye MTT to purple formazan. In this manner, IC50-molar concentration [µM] that inhibits 50% net cell growth-was determined (Tables 1 and 2).

Structure-Activity Relationship
Based on the results of anticancer screening presented in Table 1, several important conclusions on the structure-activity relationship of compounds 7-36 can be made: First of all, it can be easily seen that the human colon cancer cell line HCT-116 is the most susceptible to the tested compounds since 50% of them exhibit IC50 ≤ 25µM.
When comparing compounds 29 and 31, it can be seen that an addition of a nitro group to thiophene ring causes increasing of activity towards HCT-116 over 200 times, towards MCF-7 43 times and towards HeLa 54 times. In contrast, for benzylthio derivatives 30 and 32, the same modification of Ar substituent does not cause a significant change in activity. On the other hand, a high increase in activity and selectivity for HCT-116 is observed when replacing 5-nitrothiophen of 32 to 5-nitrofuran of 35, unfortunately we did not manage to obtain similar 5-nitrofuran derivatives with R 1 = Cl.
The introduction of halogen atoms in the benzene ring of the Ar substituent, for compounds 9, 11, 13, 15, for which R 1 = Cl, has a beneficial effect on activity against all three lines compared to compound 7. However, the activity of these compounds is still very varied and is in a wide range from 16 to 150 µM for all three cancer lines. On the other hand, for compounds 10, 12, 14 and 16, for which R 1 = SBn, the introduction of halogen atoms does not significantly increase activity relative to the unsubstituted Ar ring of compound 8. As a consequence, the activity of these compounds is in a narrow range, namely 11-40 µM.
Analyzing the influence of R 1 substituent, it can be clearly seen that compounds with benzylthio substituent (R 1 = SBn) (8-34 even number, and 35) almost always show higher activity to each cell line than their counterparts with a chlorine atom in the R 1 position (R 1 = Cl) (7-33 odd number, and 36). Consequently, compounds with R 1 = SBn globally show lower average of activity than R 1 = Cl derivatives ( Table 3). On the other hand, the main feature of compounds with the R 1 = Cl substituent is that the small change in the Ar ring causes a much more significant change in activity than for R 1 = SBn analogs. This property of a large variety of activities of R 1 = Cl group is visible in the wider range of IC50 values and higher standard deviation (Table 3). Furthermore, each compound with R 1 = Cl shows greater differentiation in activity between the tested cell lines. Considering the selectivity of the tested compounds, we can see that the most selective is compound 35, which is 22 times more active against HCT-116 than HaCaT, 37 times more than MCF-7 and 17 times more than HeLa. In general, when R 1 = Cl, selectivity index SI (which represents IC50 for HaCaT cell line/IC50 for cancerous cell line) for active derivatives (for which IC50 for cancer line ≤ 25 µM) is in the range from 2 to 16, depending on the cell line. Meanwhile, if R 1 is SBn, it reaches 13 (14) and 22 (35); however, in this group there are also several cases (8,10,18) where activity against HaCaT is at the same level as against cancer lines.
In the next step, to investigate the significance of the sulfonamide moiety we compared ( Figure  3  Similarly, as previously established, compounds build on 2-(benzylthio)-4chlorobenzenesulfonamide scaffold proved to be substantially more active than others.
Comparing compounds built on 2,4-dichlorobenzene (51-54) and 2,4dichlorobenzenesulfonamide scaffold (7, 11, 19 and 21), it can be seen that the presence of sulfonamide moiety always increases the activity against HCT-116, regardless of Ar substituent, while the MCF-7 and HeLa cell lines compounds without sulfonamide display higher, but generally weak, activity only when Ar = Ph (comp 51 compering to 7). When the phenyl ring is substituted with 4-Cl, 4-CN or 4-CF3, the compounds with sulfonamide moiety are always more active (11, 19 and 21).
On the other hand, compounds possessing sulfonamide moiety in the para position to oxadiazole ring (47-50) are less active towards MCF-7 and HCT-116 than compounds with a sulfonamide in the meta position, both with SBn or Cl R 1 substituent (7, 8 11, 12, 19-22). Some exceptions are observed only for HeLa cell lines, especially for compound 48, which, for this cell line, displays the lowest IC50 value of 25 µM.

Quantitative Structure-Activity Relationship Analysis
In order to explain the reasons for the observed variability in the anticancer activity of compounds 7-36, determine the most important parameters controlling pharmacological effects and obtain QSAR equations that will allow us to predict activity of a future compound, we conducted a statistical analysis basing on a set of computed quantitative molecular descriptors. As the values of many descriptors strictly depend on the three-dimensional structure, we performed a two-step procedure to determine the lowest-energy conformation. First, it consists of searching the conformational space of compounds using molecular mechanics. Then, the lowest energy conformations were treated as input conformations for semi-empirical geometry optimization using the PM6 method, which ensures an optimal ratio of precision to computation time. Compounds with optimal conformation were sent to the Molecular Operating Environment software (Chemical Computing Group) to calculate a set of 293 quantitative molecular descriptors.
To check the internal structure of the obtained data, we used hierarchical cluster analysis to group compounds 7-36 based on the received descriptors. Cluster analysis (CA) was used for the natural grouping of samples in clusters that are not known beforehand, with a common property characterized by the values of a set of variables [19]. In QSAR modelling, it is widely used to check out the homogeneity of data, identify some unusual data points, detect patterns, and represent potentially interesting relationships in the data [20,21]. The results are presented in the form of a dendrogram ( Figure 4) which displays the distance level at which there was a combination of objects and clusters-the more on left the clusters merge, the more dissimilarity they show. The most visible and unambiguous observation is the division into compounds with the substituent R 1 = Cl and R 1 = SBn, which are grouped at the end of the dendrite. This indicates a large difference between these two groups in terms of molecular properties, which, along with the differences shown in Table 3, gives us the basis to treat these two groups separately when looking for QSAR equations.
Analyzing the dendrite, we also see that the most active compound 31 (Ar = 5-nitrothiphene) shows significant dissimilarity from other compounds clustering as the latest single compound with 17 (Ar = 4-HOPh) and 21 (Ar = 4-NCPh). This leads us to the conclusion that in terms of molecular parameters, it exhibits features of outliers, which may explain the exceptionally high activity of this compound.
Surprisingly, CA analysis indicates a substantial influence on molecular properties of the substitution pattern of the Ar moiety in the R 1 = Cl group of compounds. Thus, we can see that compound 7 (Ar = Ph) is more similar to compounds with unsubstituted heterocyclic Ar ring 29 and 33 and next to pyridine derivatives 23, 25 and 27 rather than to halogen derivatives 9, 11, 15 and 13. Similar relationships have been observed in the anti-cancer activity of these compounds.
In the next step of analysis, a search for QSAR models was performed with the multiple linear regression (MLR) technique along with stepwise algorithm. As a result, we obtained six statistically significant equations (E1-E6) describing the activity, expressed as pIC50 = −logIC50[M] (to get its normal distribution), separately for group R 1 = Cl and group R 2 = SBn (Table 4). Table 4. The QSAR equations E1-E6 and their statistical parameters of the two groups of derivatives (R 1 = Cl and R 1 = S Bn) against MCF-7 and HCT-116 cell lines.

R 1 = Cl R 1 = SBn Equation Statistics Equation Statistics
HCT-116 The QSAR model must be validated to be considered predictive [22,23]. Due to the small number of observations resulting from dividing the initial set of compounds 7-36 into two groups, we decided to apply five-fold-cross-validation ( Figure 5) in which each set of 15 compounds is randomly divided into five three-member subsets and in turn each of them is treated as a test set, and the other as a training set. The squared correlation coefficient (Q 2 ) obtained in this way are in the range 0.72-0.91 and indicate that our models can be used to confidently predict the activity of newly synthesized compounds. Besides activity prediction, the QSAR equations indicate which properties of the molecule are favorable for obtaining high activity. In Equations E2, E4 and E5, descriptors with the highest influence on activity-with highest regression coefficients-refer to the negatively charged surface of the molecule. These are the fractional charge-weighted negative surface area (FCASA)(E2), fractional negative van der Waals surface area (PEOE_VSA_FNEG)(E4) and the fractional negative water accessible surface area (FASA)(E5). Similarly, in Equations E1 and E3, the descriptors describing the size of the molecule's surface are the most important for activity: SMR_VSA2 = sum of van der Waals surface of atoms with contribution to Molar Refractivity in the range 0.26-0.35 (E1) and vsurf_Wp2 = Polar volume at −0.5 (E3). Only activity towards HeLa cell line in R 1 = SBn group (Equation E6) shows a deviation from this regularity, as it depends mostly on Kier and Hall connectivity indices descriptor chi1v_C = carbon valence connectivity index.

Synthesis
The following instruments and parameters were used: melting points were recorded on the Boetius HMK apparatus; (Franz Kustner Nacht KG, Dresden, Germany) and were uncorrected, IR spectra were measured on Thermo Mattson Satellite FTIR spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) in KBr pellets, and the absorption range was 400-4000 cm −1 . 1 H NMR and 13 C NMR spectra were recorded on a Varian Unity Plus 500 apparatus (Varian, Palo Alto, CA, USA), chemical shifts are expressed at δ values relative to Me4Si (TMS) as an internal standard.
Elemental analyses were performed on PerkinElmer 2400 Series II CHN Elemental Analyzer (PerkinElmer, Inc. Waltham, MA USA).
Methyl esters 39, 40 hydrazides 41, 42 were prepared according to methods described in [25] and their melting points and IR spectra were in accordance with generally available literature data.
Compound 31 proved to be the most active among all tested derivatives, showing IC50 values for tumor cell lines in the range 0.5-4.5 μM. Analysis of the structure-activity relationship showed that the activity of compounds substituted with a chlorine atom in position 2 of benzenesulfonamide scaffold (R 1 = Cl) is much more diverse and more dependent on the structure of Ar substituent than for compounds substituted with benzylthio in position 2 of benzenesulfonamide scaffold (R 1 = SBn) which, however, show lower average activity. Analyzing the influence of different benzenesulfonamide scaffold, we have observed a more favorable effect of the sulfonamide group in the meta position to the oxadiazole ring, than in the para position, or for the total absence of sulfonamide moiety.
Hierarchical cluster analysis of compounds 7-36 with the use of 293 quantitative molecular descriptors obtained by Molecular Operating Environment (MOE) software. Analysis showed a reasonable division of compounds into two clusters (R 1 = Cl and R 1 = SBn), and showed that the most active compound 31 exhibits outlying molecular properties.
Using the MLR method, QSAR equations E1-E6 with good predicting properties were obtained, useful for assessing the activity of new derivatives during further optimization steps. The molecular descriptors used in the QSAR Equations indicate the potentially beneficial effect of increasing the proportion of the surface of negatively charged atoms and increasing the number of atoms with greater polarity or molar refraction.
Supplementary Materials: Supplementary materials can be found at www.mdpi.com/1422-0067/21/6/2235/s1. Author Contributions: K.S. and J.S. created the concept, and designed the study. K.S. performed chemical research and analyzed both the chemical and biological data. K.S. and Ł.T. performed statistical analysis and interpreted the statistical models. A.K. tested the biological activity of the compounds K.S. writing-original draft preparation. J.S. writing-review and editing. All authors read and approved the final version of the article.

Conflicts of Interest:
The authors declare no conflict of interest.