Synthesis and Experimental Validation of New Designed Heterocyclic Compounds with Antiproliferative Activity versus Breast Cancer Cell Lines MCF-7 and MDA-MB-231

Recent drug discovery efforts are highly focused towards identification, design, and synthesis of small molecules as anticancer agents.With this aim, we recently designed and synthesized novel compounds with high efficacy and specificity for the treatment of breast tumors. Based on the obtained results, we constructed aVolsurf+ (VS+)model using a dataset of 59 compounds able to predict the in vitro antitumor activity against MCF-7 cancer cell line for new derivatives. In the present paper, in order to further verify the robustness of this model, we report the results of the projection of more than 150 known molecules and 9 newly synthesized compounds. We predict their activity versus MCF-7 cell line and experimentally verify the in silico results for some promising chosen molecules in two human breast cell lines, MCF-7 and MDA-MB-231.


Introduction
Despite continued research efforts, cancer remains one of the biggest threats to human health.In the United States, it is the second highest cause of death and is likely to soon replace heart disease as the leading cause [1].Breast cancer is the most common cancer in women worldwide, with nearly 1.7 million newly diagnosed cases in 2012 (second most common tumor overall) [2,3].This represents about 12% of all new cancer cases and 25% of all cancers in women.A number of risk factors have been identified, such as a strong family history of breast cancer, weight gain, smoke, menopausal hormone therapy (MHT), and alcohol consumption [4][5][6].However, the causes of the disease are not yet fully understood.Treatment generally involves surgery, radiation, chemotherapy, hormone therapy, and/or targeted therapy.New anticancer agents with unique mechanisms of action have been developed; however, many of them have not been therapeutically useful due to low tumor selectivity or to side effects.As an example, the US Food and Drug Administration (FDA) recently retired bevacizumab, approved for the treatment of metastatic breast cancer, because of potentially adverse side effects and minimal therapeutic benefit [7].
Drug discovery efforts are highly focused towards design and synthesis of small molecules as anticancer agents [8].In parallel, most efforts are devoted towards the "drugrepositioning process" as an alternative drug development strategy [9].This approach gives an opportunity to find new uses of existing drugs or molecules with a tremendous savings of time and money.
During the years, a wide range of small molecules based on heterocyclic ring systems have been studied for the development of novel lead compounds in the drug discovery paradigm.A number of the heterocyclic nuclei can be regarded as "privileged scaffolds" [10] because they are ubiquitous in drug molecules.Since they possess hydrogen bond donors and acceptors in a rigid framework, they can therefore effectively interact with target enzymes and receptors via hydrogen bond interactions.They can enhance binding affinity and improve in vitro potency.Heterocycles can modulate lipophilicity of the drug molecules or improve aqueous solubility of the compounds, thus providing desired pharmacokinetic and pharmaceutical properties [11].These premises prompted us to design and synthesize novel compounds with high efficacy and specificity for the treatment of breast tumors.In 2013, this research group reported the design and synthesis of new heterocyclic compounds [12].To fasten the lead discovery process, a Vol-surf+ (VS+) model was constructed using a dataset of 59 compounds able to predict the in vitro antitumor activity against MCF-7 cancer cell line for the new derivatives.The use of Volsurf+ for in silico models generation is a well established procedure, with proved efficacy [13,14].Novel compounds when tested in vitro showed a significant agreement with in silico predictions.In the present paper, in order to verify further the robustness of this model and to apply the drug-repositioning concept, we report the results of the projection of 176 derivatives.Among them, 167 were previously synthesized for other applications such as  1 adrenergic, 5-HT 1A serotonergic, or endothelin receptor ligands and antiproliferative compounds active in prostate tumor cell lines [15][16][17][18][19][20][21][22][23][24][25].Moreover, 9 compounds were designed and synthesized based on the recent literature [26][27][28] reporting the activity of pyrimidinketones against MCF-7 cell line.The new compounds retain the same heterocyclic core with respect to the literature ones, modified by fusion with thiophene and pyrazole rings with different substituents.Such modifications should improve the antiproliferative activity reported in literature for similar compounds.All the 176 structure formulas were reported in Supporting Information (SI) (in Supplementary Material available online at https://doi.org/10.1155/2017/9729284).The generated dataset is made up of 7 subsets based on different chemical scaffolds, namely, pyrrolopyrimidindione, pyrimidinketone, benzyloxyindole, indole, long chain arylpiperazines (LCAPs), quinoline, and thiophene derivatives (SI).These subsets share the common characteristic of being composed of small molecules possessing a heterocyclic core.The cores were chosen based on the idea of using privileged scaffolds [10].The term "privileged scaffolds" was first coined by Evans et al. in the late 1980s [29], referring to scaffolds seemingly capable of serving as ligands for a diverse array of receptors, for example, indole, quinoline, and arylpiperazines.
All these 176 structures represented our training set that is very different compared to the 59 compounds of the test set because of the lack of the ethylenic bond that is the main feature of the test set.
Herein, we predict the subsets activity versus MCF-7 cell line and experimentally verify the in silico results for promising chosen molecules in two human breast cell lines, MCF-7 and MDA-MB-231.

VS+.
VS+ is an automatic procedure which allows converting information coded into the 3D GRID Molecular Interaction Fields into 128 physicochemically relevant descriptors [31].The obtained descriptors concern molecular size and shape, hydrophilic and hydrophobic properties, hydrogen bonding, amphiphilic moments, and critical packing parameters.Pharmacokinetic descriptors related to solubility, metabolic stability, and cell permeability were also generated.
Surface properties such as shape, electrostatic forces, H-bonds, and hydrophobicity influenced the interaction of molecules with biological membranes.Therefore, potential polar and hydrophobic interaction sites around target molecules by the water (OH 2 ), the hydrophobic (DRY), and the carbonyl oxygen (O) and amide nitrogen (N1) probe [32] were characterized by the GRID force field.
The information contained in the Molecular Interaction Fields (MIF) is converted into a quantitative scale by calculating the volume or the surface of the interaction contours.The VS+ procedure is as follows: (i) First, the interactions of OH 2 , DRY, O, and N1 probes around a target molecule generated the 3D molecular field.
(ii) Then, the descriptors from the 3D maps, obtained in the first step, were calculated.These molecular descriptors, called VS+ descriptors, concern molecular size and shape, hydrophilic and hydrophobic regions and the balance between them, molecular diffusion, log P, log D, "charge state" descriptors, new 3D pharmacophoric descriptors, and descriptors of some appropriate ADME properties.The definition of all 128 VS+ descriptors was reported elsewhere [31].
(iii) Finally, chemometric tools (PCA [33] and PLS [34]) are used to create relationships of the VS+ descriptor matrix with ADME properties.
The scheme of the VS+ programme steps and a detailed definition of VS+ descriptors have recently been reported [35].
The VS+ software was used to perform a Structure-Activity Relationship (QSAR) study using the default settings for protomers and conformers generations.

Treatment with Antitumor Agents and MTT Colorimetric
Assay.Human cancer cell line was plated in 96-well plates "Nunclon TM Microwell TM" (Nunc) and were incubated at 37 ∘ C. MTT assay was performed as reported by Barresi et al. [37].After 24 h, cells were treated with the compounds (final concentration 0.01-100 M).Untreated cells were used as controls.Microplates were incubated at 37 ∘ C in humidified atmosphere of 5% CO 2 , 95% air for 3 days, and then cytotoxicity was measured with colorimetric assay based on the use of tetrazolium salt MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) [38].The results were read on a multiwell scanning spectrophotometer (PlateReader AF2200, Eppendorf, Milan, Italy), using a wavelength of 570 nm.Each value was the average of six to eight wells (standard deviations were less than 15%).The GI 50 value was calculated according to NCI [39]: thus, GI 50 is the concentration of test compound where 100×(−0)/(−0) = 50 ( is the optical density of the test well after a 72 h period of exposure to test drug; 0 is the optical density at time zero;  is the control optical density).The cytotoxicity effect was calculated according to NCI when the optical density of treated cells was lower than the 0 value using the following formula: 100 × ( − 0)/0 < 0.

Results and Discussion
As a first approach we projected all the 176 compounds on the model previously published [12].The PLS model after projection of the test set is shown in Figure 1: the predicted compounds are colored according to the different belonging classes: pyrrolopyrimidindione in green, pyrimidinketones in yellow, benzyloxyindole in light blue, indole in red, LCAPs in purple, quinoline in blu, and thiophene in white.The background of the PLS model is color-coded by the activity values against MCF-7 cell line, using a scale from red (active) to blue (inactive).The 1 and 2 values refer to the coordinates of each single compound within the plot.
Two outliers families, pyrrolopyrimidindione and thiophene derivatives, lying out of the confidence ellipse were discharged, probably due to some solubility and permeability problems.
Moreover, in order to increase the predictivity power, we performed the class model for each single subset discharging the correspondent outliers (SI).By using this strategy, we obtained a test set of 25 compounds that are the best representative molecules for each class (Figure 2).
Starting from the obtained test set, we decided to select 10 out of the 25 derivatives: 5 from active part of the graphic, red area, and 5 from the inactive one, blue area (Table 1).To validate our prediction we tested those compounds in vitro against MCF-7 and MDA-MB-231 cell lines using 5-fluorouracil (5-FU) as standard chemotherapy agent reference (Table 1).The biological characteristics of MCF7 and MDA-MB-231 are dissimilar: the first cell line is a representative model of ductal breast carcinoma, oestrogen receptor-positive, receptor tyrosine-protein kinase (erbB-2, HER2/neu) negative and responding to endocrine therapy,  while the MDA-MB-231 is oestrogen receptor-negative unresponsive to the same treatment.The antiproliferative effects observed in Table 1 for compounds pyrimidinketone 2A, pyrimidinketone 2B, and quinoline 7B are stronger than same similar effect observed with the reference 5-FU in both cell lines.It is noteworthy that these compounds, with very low t1 values in Figure 2, exhibit a very high activity, thus confirming the robustness of the PLS predictions from a quantitative point of view.
In Table 1, we report the experimental activity values, expressed as log GI 50 , for the 10 tested compounds and VS+ predicted values.In particular, in red we report the 5 compounds predicted active (Table 1), while in blue we report the remaining 5 predicted inactive.In silico predictions were confirmed for most of the compounds tested in vitro.Two marked compounds in the table, predicted active, resulted inactive in the experimental test.However, it should be noted that these two compounds showed some solubility problem during the in vitro procedure, even if we added 1% of DMSO; they gave flocculation; thus data relative to these compounds are not reliable.

Conclusion
The predictivity power of the model previously published from our group was confirmed using different type of heterocyclic compounds.Three of these derivatives showed good antiproliferative in vitro activity against MCF-7 and MDA-MB-231 cell lines.The MDA-MB-231 cell line represents a triple-negative breast cancer without oestrogen receptor (ER), progesterone receptor (PR), and epidermal growth factor receptor 2 (HER2), associated with strong aggressiveness, poor prognosis, and unresponsiveness to the usual endocrine therapies.The antiproliferative effects observed on MCF-7 and MDA-MB-231 are worthy of attention to develop molecules that are able to attack the breast tumor oestrogen receptor-positive and/or oestrogen receptor-negative.Studies aimed at the elucidation of the above mechanism pathways are in progress.

Table 1 :
Molecular structures, in vitro activity on MCF-7 and MDA-MB-231 cell lines, and in silico activity values for the ten selected compounds and 5-FU.Activity values expressed as log GI 50 .50 /M ± SD * MDA-MB-231 experimental activity log GI 50 /M ± SD * Each value represents the mean ± SD of six to eight wells.NA, not available.