- Email: [email protected]
In vitro antitumor activity, ADME-Tox and 3D-QSAR of synthesized and selected natural styryl lactones
Computational Biology and Chemistry 83 (2019) 107112
Contents lists available at ScienceDirect
Computational Biology and Chemistry journal homepage: www.elsevier.com/locate/cbac
Research Article
In vitro antitumor activity, ADME-Tox and 3D-QSAR of synthesized and selected natural styryl lactones
T
⁎
Vladimir R. Vukica, , Davor M. Loncara, Dajana V. Vukica, Lidija R. Jevrica, Goran Benedekovicb, Jovana Francuzb, Vesna Kojicc, Milica Z. Karadzic Banjaca, Velimir Popsavinb,d a
University of Novi Sad, Faculty of Technology Novi Sad, Bulevar cara Lazara 1, 21000 Novi Sad, Serbia University of Novi Sad, Faculty of Sciences, Department of Chemistry, Biochemistry and Environmental Protection, Trg Dositeja Obradovića 3, 21000 Novi Sad, Serbia c University of Novi Sad, Faculty of Medicine, Oncology Institute of Vojvodina, Put Dr Goldmana 4, 21204 Sremska Kamenica, Serbia d Serbian Academy of Sciences and Arts, Knez Mihajlova 35, 11000 Belgrade, Serbia b
A R T I C LE I N FO
A B S T R A C T
Keywords: Antitumor activity Styryl lactone ADME-Tox 3D-QSAR analysis PC-3 cells
Prostate cancer is a common cause of death in men and a novel treating methods should be developed. In order to find a new drug for prostate cancer, a series of novel conformationally constrained analogues of (+)-goniofufurone and 7-epi-(+)-goniofufurone, as well as the newly synthesized styryl lactones containing the cinnamic acid ester groups were evaluated for in vitro cytotoxicity against prostate cancer cell (PC-3). Furthermore, prediction of physicochemical characteristics and drugability as well as in silico ADME-Tox tests of investigated compounds were performed. The 3D-QSAR model was established using the comparative molecular field analysis method. According to obtained results, the tricyclic compounds 9 and 10 had the highest potency with IC50 < 20 μM. Evaluation of structural features through 3D-QSAR model identified steric field feature on the cinnamic acid ester groups at C-7 as a crucial for the cytotoxic activity. This research suggests that most of the analysed compounds have desirable properties for drug candidates and high potential in drug development, which recommend them for further research in treatment of prostate cancer. Furthermore, obtained 3D-QSAR model is able to successfully identify styryl lactones that have significant cytotoxic activity and provide information for screening and design of novel inhibitors against PC-3 cell line that could be used as drugs in treatment of the prostate cancer.
1. Introduction According to American Institute for Cancer Research and World Cancer Research Fund prostate cancer is the second most commonly occurring cancer in men and the fourth most commonly occurring cancer overall, with 1.3 million new cases in 2018 (Anon, 2019). As chemotherapy and radiation therapy are largely ineffective, a novel treating methods should be developed (Shin et al., 2012; Shirzad et al., 2015). Medicinal plants are widely used as primary healthcare to cure various diseases. Many tropical plants, such as Goniothalamus species (Family Annonaceae), have been reported to possess interesting biological activities with potential therapeutic application (Harvey, 2000;
Seyed et al., 2014). Main compounds in Goniothalamus are styryl lactones, acetogenins and alkaloids which have cytotoxic, antitumor, insecticidal, antifungal, antimalarial, antitubercular, antibacterial, antiviral and antioxidant activities (Blázquez et al., 1999; de Fátima et al., 2006; Abdullah et al., 2013; Choo et al., 2014; Iqbal et al., 2015). Styryl lactones are bioactive compounds with significant cytotoxicity against several tumour cell lines (Seyed et al., 2014; Choo et al., 2014; Wiart, 2007; Duc et al., 2016). A large body of evidence suggests that the antiproliferative activity of styryl lactones is associated with induction of apoptosis in target cells (Seyed et al., 2014; de Fátima et al., 2006). Due to their antiproliferative properties, much attention was focused on the synthesis of novel styryl lactone derivatives. Many of them exhibited strong cytotoxic activity against human stomach
Abbreviations: BBB, blood-brain barrier; CoMFA, comparative molecular field analysis; CoMSIA, comparative molecular similarity indices analysis; CV, crossvalidation; DILI, drug-induced liver injury; GIA, gastrointestinal absorption; hERG, human ether-à-go-go-related gene; HLM, human liver microsomal stability; MMP, mitochondrial membrane potential; P-gp, permeability glycoprotein; PLS, partial least squares; QSAR, quantitative structure-activity relationship; RMSD, root mean square deviation; SD, standard deviation ⁎ Corresponding author. E-mail address: [email protected] (V.R. Vukic). https://doi.org/10.1016/j.compbiolchem.2019.107112 Received 4 April 2019; Received in revised form 13 August 2019; Accepted 18 August 2019 Available online 23 August 2019 1476-9271/ © 2019 Elsevier Ltd. All rights reserved.
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
topological polar surface area (TPSA) (Ertl et al., 2000). It has been proven as a useful descriptor in many models with regards to biological barrier crossing such as gastrointestinal absorption and brain access (Daina and Zoete, 2016). In the present study, drug-likeness was established from structural or physicochemical inspections of compounds. The Lipinski (Pfizer) filter is the pioneer rule-of-five based on analysis of physicochemical properties of molecules in order to indicate weak permeation/absorption when molecular weight is higher than 500, the number of H-bond acceptors is greater than 10, the number of H-bond donors is greater than 5, and ClogP is greater than 5 (Lipinski et al., 2001). In the present study, modified Lipinski rule of 5 that includes LogD at pH = 5.5 instead of LogP values was applied (Bhal et al., 2007). In order to be accepted for oral use by this rule, the value of LogD should be less or equal to 5. Values of LogD values were calculated using PreADMET online software (PreADMET, 2019). The Ghose (Amgen), Veber (GSK), Egan (Pharmacia) and Muegge (Bayer) methods were also used in order to predict drug-likeness of the compounds (Ghose et al., 1999; Veber et al., 2002; Egan et al., 2000; Muegge et al., 2001). Multiple lipophilicity predictors were used in order to generate consensus estimation represented as Consensus Log P. Liver toxicity (DILI, cytotoxicity), metabolism (HLM, CYP1A2, CYP2C19, CYP2C9, CYP2D6 and CYP3A4 inhibitor), membrane transporters (P-gp inhibitor, P-gp substrate), hERG blocker, MMP and AMES test were performed using vNN-ADMET online server (Schyman et al., 2017). Cytotoxicity is the degree to which a chemical causes damage to cells. In the present study, test compounds with an IC50 of 100 μM or less are considered as cytotoxic.
cancer cells SGC-7901 (Zhou et al., 2005), oestrogen receptor-positive breast adenocarcinoma cells MCF-7 (Kovačević et al., 2016). Raji Burkitt's lymphoma cells (Popsavin et al., 2008, 2010; Francuz et al., 2012; Benedeković et al., 2014a, b), alveolar basal adenocarcinoma cells A549 (Zhou et al., 2005), promyelocytic leukaemia cells HL-60 (Zhou et al., 2005; Popsavin et al., 2008, 2010; Francuz et al., 2012; Benedeković et al., 2014b; Popsavin et al., 2009), T lymphocyte leukaemia Jurkat cells (Francuz et al., 2012; Popsavin et al., 2009; Mohideen et al., 2013), myelogenous leukaemia cells K-562 (Popsavin et al., 2010; Benedeković et al., 2014a), cervix carcinoma cells HeLa (Benedeković et al., 2014a; Popsavin et al., 2012), melanoma cells Hs 294 T (Benedeković et al., 2014a). The in silico prediction of cytotoxic and antiproliferative activity has been identified as a very important step in the synthesis of compounds with desired biological activity (Narayana Moorthy et al., 2012; Scotti et al., 2014; Abdulfatai et al., 2017). In that aim, three-dimensional quantitative structure-activity relationship (3D-QSAR) was successfully used in the early stage of preclinical drug development (Fang et al., 2016; Kumari et al., 2016). Therefore, the present study is a continuation of the previous research with the aim to determine in vitro cytotoxic activity of some newly synthesized and natural styryl lactones, their derivatives, and analogues against prostate cancer cell (PC-3). Furthermore, in silico evaluation of ADME-Tox properties were examined and a predictive 3DQSAR model that will be used for a lead optimization and testing of novel compounds was developed. 2. Materials and methods
2.4. D-QSAR analysis 2.1. Styryl lactones The 3D structures of all compounds are constructed using Chemsketch, a chemically intelligent drawing interface freeware. The geometry optimization of all compounds was carried out using the MMFF94 force field with MMFF94 partial charges. The Powell conjugated gradient algorithm method was implemented with convergence criterion of 0.042 kJ/mol Å, using the maximum iterations set to 1000 (Halgren and Nachbar, 1996). Structure alignment is one of the most critical step in CoMFA (Comparative Molecular Field Analysis) and CoMSIA (Comparative Molecular Similarity Indices Analysis) analysis. So, selection of the template molecule for the alignment was a crucial step in the construction of the QSAR models (Liu et al., 2011). Considering the structural similarities of the studied styryl lactones, furano-furone bicyclic core was selected as a common structural segment for the alignment. The most active compound is used as a template for superimposition, which is assumed to represent the most bioactive conformation. The set of 24 ligands was randomly divided into two subsets: the training set for model building (20 compounds) and the test set (4 compounds) which was used as external validation set (Dixon et al., 2016). The 3D-QSAR analysis was performed on the aligned compounds using the field-based QSAR which is based on CoMFA and CoMSIA (Cramer, 2012; Cramer et al., 1988; Klebe and Abraham, 1999; Klebe et al., 1994). CoMFA and CoMSIA are based on the representation of ligands through molecular fields measured in the space that surrounds them. 3-D contour map of the physicochemical forces that surround a series of aligned compounds was constructed and the interaction energy of an appropriate probe in each of the points in that 3-D contour map was used as structural descriptors to be correlated with biological activity during construction of 3D-QSAR model. CoMFA field-based models are constructed by calculating the value of the electrostatic and steric molecular field, on a rectangular grid that encompasses the molecules in the training set and correlated with bioactivities by use of PLS (Partial Least Squares) using three-factor regression model. CoMSIA is encompassing the steric, electrostatic, hydrogen bonding and hydrophobic effects of ligands fields which are also evaluated at points on a rectangular grid. Fields are calculated by summing the values of
Four naturally occurring styryl lactones (1, 2, 11, 18) and twenty newly synthesized analogues thereof (3, 4‒10, 12‒17, 19‒24) were examined in this work. Synthesis of the compounds used in this research is already published: total syntheses of 7-epi-(+)-goniofufurone (1) and (+)-goniofufurone (2) are in detail described in the work (Benedeković et al., 2014a); (+)-Crassalactones B (11) and C (18), the corresponding 5,7-dicinnamoate 22, a number of new analogues such as 12, 13, 19‒21, the corresponding 7-epimers such as 14, 15, 23, 24, 7-deoxy derivatives such as 16, 17 (Benedeković et al., 2014b), as well as 3 and tricyclic lactones 4‒10 (Benedeković et al., 2014a) have been prepared by semi-synthesis starting from compounds 1 or 2. 2.2. Cytotoxic activity The investigated compounds were tested in vitro against human prostate tumour cell line (PC-3). Cytotoxic activity was evaluated by using the standard colourimetric MTT assay, after exposure of cell line to the tested compounds for 72 h (Popsavin et al., 2006). The commercial antitumor agent doxorubicin (DOX) was used as a positive control. The results are expressed as IC50 and pIC50, the concentration of compound required to inhibit the cell growth by 50% compared to an untreated control. 2.3. ADME-Tox properties and druglikness test The molecular descriptors were calculated on the basis of 2D molecular structures, presented in the Supplementary data (Table S1). ADME test including iLOGP XLOGP3, WLOGP, MLOGP, Silicos-IT Log P, GI absorption, Blood-brain barrier as well as druglikeness including Ghose, Veber, Egan and Muegge violations and Bioavailability Score were performed using SwissADME (Daina et al., 2017). Simple molecular and physicochemical descriptors like molecular weight (MW), molecular refractivity (MR), counts of specific atom types and polar surface area (PSA) are computed with Open Babel, version 2.3.0 (O’Boyle et al., 2011). PSA is calculated using a technique called 2
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
properties of a given atom, weighted by a Gaussian function of the distance between the grid point and the atom. The field-based QSAR models are an implementation of the CoMFA and CoMSIA methods with a specific set of parameters. The CoMFA and CoMSIA descriptor fields were obtained by a 3D cubic lattice with a grid spacing of 1 Å. In the CoMFA model, a hybrid sp3 carbon atom with a positive charge was used as a probe to compute the CoMFA steric and electrostatic fields in which their energy values were truncated at 30 kcal/mol in order to eliminate large repulsive energies that are consequence of close contact between the probe and the molecule (Vilar and Costanzi, 2012; Zhao et al., 2011). The lattice size and probe step size were adjusted automatically. The partial least squares (PLS) analysis is used to construct the best model using linear correlation the CoMFA and CoMSIA fields to the pIC50 (Bush and Nachbar, 1993). Cross-validation analysis was performed using the leave-one-out method (Tetko et al., 2001). The optimum number of components used in the final analysis was identified by the cross-validation method. Correlation and cross-validation coefficients (r2 and q2 respectively) were calculated according to the formula:
r2 = 1 −
Table 1 Cytotoxic activity of examined compounds against PC-3 cell line. IC50 (μM)
Comp. label
IC50 (μM)
Comp. label
IC50 (μM)
1 2 3 4 5 6 7 8
899.90 254.36 487.66 185.35 537.78 334.98 861.37 149.03
9 10 11 12 13 14 15 16
15.64 12.31 25.11 193.43 101.21 54.39 24.35 286.22
17 18 19 20 21 22 23 24
114.32 31.08 191.51 702.22 131.21 401.21 101.32 105.41
(+)-goniofufurone and 7-epi-(+)-goniofufurone (Benedeković et al., 2014a, b) as well as chloro and bromo (+)-goniofufurone mimics (Francuz et al., 2012) showed in most cases stronger cytotoxicity compared to corresponding lead compounds. However, when another electron-withdrawing group (NO2) is introduced in the same position the activity is not increased (compared to 11 and 15). But, insertion of the electron-donating group OCH3 in the para position of cinnamoate aromatic ring resulted in decrease of cytotoxic activity (compared to 11 and 14). Considering the influence of absolute stereochemistry at the C7 position, (7R)-stereoisomer 9 displayed the higher potency than (7S)isomer 8 (Table S1). Similar results were obtained by Benedeković et al. (2014a) after evaluation of their cytotoxicity against a panel of eight human tumour cell lines. Overall, some analogues with cinnamic acid ester group have high cytotoxic activity, which justified the main reason for the preparation of these derivatives in order to combine antitumor potential of styryl lactones and cinnamoates (Zhou et al., 2005; Kovačević et al., 2016; Popsavin et al., 2008, 2010; Francuz et al., 2012; Benedeković et al., 2014a, b; Popsavin et al., 2009; Mohideen et al., 2013; Popsavin et al., 2012; De et al., 2011).
(Yobs − Ypred )2 (Yobs − Ymean )2
q2 = 1 −
Comp. label
(Yobs − YCVpred )2 (Yobs − Ymean )2
Yobs – observed activity Ypred – predicted activity Ymean – average activity of the entire training data set YCVpred – cross-validated predicted activity In order to test “chance correlations” r2 Scramble test was performed by calculating r2 from a series of models using scrambled activities (Yscrambling). The CoMFA/CoMSIA results were graphically interpreted using Maestro software by field contribution maps using field type “stdev*coeff”, and the contour levels set to default values.
3.2. ADME-Tox properties ADMET (absorption, distribution, metabolism, excretion and toxicity) prediction is a useful analysis in medicinal chemistry for the selection of drug-like molecules (Jing et al., 2003). Lipophilicity, characterized here by computed logP and logD values, plays a crucial role in determining several ADMET parameters as well as potency. Also, having a soluble molecule greatly facilitates many drug development activities, primarily the ease of handling and formulation (Ritchie et al., 2013). Water solubility is represented through Silicos-IT values. Results of in silico lipophilicity (iLOGP, XLOGP 3, WLOGP, MLOGP, Consensus LogP and SklogD) and water solubility (Silicos-IT LogS) are presented in the Table 2. According to the obtained results, compounds with high cytotoxic activity (9, 10, 11 and 15) have moderate lipophilicity, which ranks them in a group of drug candidates. Other tested compounds, except compound 22, also have appropriate lipophilicity to be recommended for drug candidates. In silico toxicity predictions of analysed compounds are presented in the Table 3. The knowledge about the interaction of molecules with cytochromes P450 (CYP) is essential as it is a key step in drug elimination through metabolic biotransformation (Testa and Krämer, 2007). In order to maintain an effective concentration in our blood, a drug should not be rapidly metabolized by CYPs. Most potent compounds in our research (9, 10, 15 and 18) should not interact with CYP enzymes.. Drug-induced liver injury (DILI) has been one of the most commonly cited reasons for drug withdrawals from the market (Assis and Navarro, 2009). Therefore, it is essential to eliminate that possibility in early phases of drug development. Several compounds in our research could potentially induce liver injury, but most of the promising drug candidates have negative in silico results. Also, the most potent drug candidates should not interact with potassium ion channels (hERG blocker) or express mutagenic effects. Although most compounds should not inhibit P-gp, some of them could be transported out of the cell by it,
3. Results 3.1. Structure and biological activities The names and 2D structures of studied compounds are presented in the Table S1, supplementary material. All of the examined molecules have a common structural segment, furano-furone bicyclic core, tetrahydro-furo[3,2-b]furan-2-one. The compounds 4‒10 represent conformationally constrained analogues of 1 or 2. Compounds 8-10, 11‒17, 18‒24 represent the cinnamic acid ester derivatives. The main reason for the preparation of these derivatives arises from the fact that certain cinnamoates exhibit significant antitumor activities (De et al., 2011). All of the examined compounds were completely inactive toward normal MRC-5 cells (IC50 > 100 μM), with the exception of dicinnamoate 22 which showed a weak cytotoxicity (IC50 48.64 μM). The antiproliferative activity of the commercial antitumor agent doxorubicin (DOX) was 84.23 μM. The IC50 values of tested molecules showed that PC-3 line is sensitive to compounds 9, 10, 11, 14, 15 and 18, (Table 1). The tricyclic compounds 9 and 10 had IC50 < 20 μM and the compounds 11, 14, 15 and 18 demonstrated cytotoxicity between 20 and 60 μM. However, these compounds have from 1.5 (14) to 7 (10) times higher activity than DOX. The compounds with IC50 values higher than 100 μM (1‒3, 4‒8, 12, 13, 16, 17, 19‒24) are considered to be non-cytotoxic. Abiraterone, which is currently used for the treatment of prostate cancer, has IC50 value 9.32 (Bruno et al., 2008). The analogue 10, which contains electron-withdrawing group F in the para position of cinnamoate aromatic ring, was more potent compared to molecule 9, which doesn’t have F group. Generally, fluoro derivatives of 3
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
According to our results, bioavailability of tested compounds is satisfactory with the bioavailability score of 0.55 for all compounds (Martin, 2005). Druglike properties are most commonly defined using the “rules of five”. However, the properties required of a compound intended to provide leads suitable for further optimization may be rather different (Teague et al., 1999). These include molecular weight, XLOGP and number of rotatable bonds. Our results revealed that compound 9 is suitable as a lead, which could put it in the main focus for further research, although the most potent compound is 10.
Table 2 In silico lipophilicity of analysed compounds. Molecule
iLOGP
XLOGP3
WLOGP
MLOGP
SilicosIT LogS
Consensus LogP
SklogD
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1.57 1.69 2.78 1.83 2.42 2.24 2.1 2.11 2.57 2.3 2.88 3.45 2.8 3.09 2.4 3.05 2.98 2.94 2.29 3.16 3.02 3.75 3.12 3.89
0.43 0.43 1.91 1.31 0.77 0.82 0.82 1.43 1.43 1.53 2.53 2.51 2.63 2.51 2.36 3.65 3.75 2.53 2.36 2.51 2.63 5.19 2.36 2.51
−0.16 −0.16 1.91 1.02 0.83 0.86 0.86 0.92 0.92 1.48 1.99 2 2.55 2 1.9 2.83 3.39 1.99 1.9 2 2.55 4.15 1.9 2
0.18 0.18 1.64 0.76 0.7 0.85 0.85 1.03 1.03 1.42 1.93 1.61 2.3 1.61 1.04 2.74 3.12 1.93 1.04 1.61 2.3 3.4 1.04 1.61
0.79 0.79 2.74 1.09 1.24 1.43 1.43 1.38 1.38 1.8 2.87 2.94 3.29 2.94 0.71 3.65 4.07 2.87 0.71 2.94 3.29 5.04 0.71 2.94
0.56 0.58 2.2 1.2 1.19 1.24 1.21 1.37 1.47 1.71 2.44 2.5 2.71 2.43 1.68 3.18 3.46 2.45 1.66 2.44 2.76 4.31 1.83 2.59
0.10 0.10 2.15 1.42 1.00 0.98 0.98 1.42 1.42 1.56 2.56 2.51 2.70 2.51 2.57 3.34 3.48 2.70 2.70 2.65 2.83 5.16 2.70 2.65
3.4. D-QSAR A tridimensional quantitative structure-activity relationship (3DQSAR) model was established using the comparative molecular field analysis (CoMFA and CoMSIA) method. The robustness and the predictive ability of the generated 3D-QSAR model were examined by using the test set. Statistical evaluation of the two-factor and threefactor QSAR models revealed good correlation between predicted and experimental inhibitor activity values ( r2 = 0.84, 0.93 and q2 = 0.64, 0.86, respectively) indicating a good predictability of the constructed models. Although three-factor model has higher values of r2 and q2, stability of the model was 0.44, indicating over-fitting of the model. Therefore, the two-factor model is used for further analysis. The QSAR statistical analysis is summarized in the Table 4. The QSAR model (two-factor) using steric, electrostatic, hydrophobic and hydrogen bond donor and acceptor fields had high determination coefficient – r2 (0.839) with low standard deviation – SD (0.240), high cross-validation coefficient q2 (0.640) and high Fischer ratio – F (46.9). Furthermore, values of r and r2 for the test set were 0.964 and 0.930, respectively. The measured and the predicted values of activities expressed as pIC50 are presented in the Table 5. These results indicate high applicability of constructed model in testing and design of new compounds (potentially by using compound 9 as a lead) with high cytotoxic activity toward PC-3 cell line. Fig. 1 displays the scatterplot of the observed versus the predicted pIC50 values for the best 3D-QSAR model. The distribution of the data around the regression line (r2 = 0.84) suggests that the model is accurate enough to be used in further analysis.
which could influence druglikness of those compounds. 3.3. Druglikeness of the studied compounds In the present study, druglikeness tests were performed in order to exclude molecules with properties that are incompatible with an acceptable pharmacokinetic profile (Table S2). Obtained results indicate that all tested compounds except for the compound 22 have pass the demands defined by modified Lipinski rule of five, as well as by Ghose, Veber, Egan and Muegge tests, which recommends these compounds as possible oral drugs. High gastrointestinal absorption observed in all compounds is also a desirable characteristic for a potential drug. Table 3 In silico toxicity of analysed compounds. Liver toxicity
Cyp inhibitors
Membrane transporters
Molecule
DILI
Cyto-toxicity
Cyp1A2
Cyp3A4
Cyp2D6
Cyp2C9
Cyp2C19
BBB permeab.
P-gp Inhibitor
P-gp Substrate
hERG Blocker
MMP
AMES
HLM
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
– – – + + + + – – + – – – – – – – – + + – – + +
– – – – – – – – – + + – + – – – – + + + + + + +
– – – – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – + – + – – – – – + – – – +
– – + – – + + – – – – – – – – + + – – – – – – –
+ + + – – – – – – – – – – – – – – – – – – – – –
+ + – – – – – + + + + + + + + + + – + – + – + –
– – + – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – + – – – – – – – –
+ + + + + + + + + + + + + + + + + + + + + + + +
4
Other
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
Table 4 Summary of statistical parameters obtained for the generated 3D-QSAR model. Statistical parameters
QSAR model One-Factor
SD r2 r2 Scramble Stability F p q2 Pearson-r
0.364 0.610 0.188 0.68 29.7 2.93e−005 0.334 0.927 Fraction of field contributions Gaussian Steric 0.481 Gaussian Electrostatic 0.088 Gaussian Hydrophobic 0.227 Gaussian Donor 0.169 Gaussian Acceptor 0.035
Two-Factor
Three-Factor
0.240 0.839 0.223 0.54 46.9 7.28e−008 0.640 0.928
0.168 0.926 0.827 0.44 70.8 8.19e−010 0.856 0.948
0.497 0.092 0.229 0.138 0.043
0.475 0.097 0.233 0.051 0.143
Fig. 1. The plot of predicted versus experimental pIC50 values for 3D-QSAR analysis. The solid line is the regression line for the fitted and predicted bioactivities of training and test set.
Table 5 The measured and predicted activities (pIC50) by the obtained 3D-QSAR model. Ligand name
QSAR Set
Activity
Predicted activity
Prediction error
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
training training training training training training training test training training training training test training training training training training training training test test training training
3.05 3.59 3.31 3.73 3.27 3.48 3.06 3.83 4.80 4.91 4.60 3.71 3.99 4.26 4.61 3.54 3.94 4.51 3.72 3.15 3.88 3.40 3.99 3.98
3.36 3.77 3.66 3.39 3.39 3.47 3.30 3.60 4.90 4.90 4.11 3.86 3.97 3.98 4.04 3.69 3.70 4.09 3.67 3.20 3.79 3.53 4.21 4.24
−0.31 −0.18 −0.34 0.34 −0.12 0.01 −0.23 0.23 −0.09 0.01 0.49 −0.14 0.02 0.29 0.57 −0.15 0.24 0.41 0.04 −0.04 0.09 −0.14 −0.22 −0.26
groups is crucial for cytotoxic activity of examined styryl lactones. Therefore, the compounds which contain the cinnamic acid ester group, showed, in most cases, higher activity compared to compounds without cinnamic acid ester group (Table 1). This is not always consistent, as compounds with different stereochemistry have different activity. For example, compounds 12 and 14 have the same formula, but different stereochemistry, which induced huge differences in binding to the protein and cytotoxic activity. Therefore, to predict the activity of a compound, it is important to include its three-dimensional structure to obtain more reliable results. In addition to the steric feature, Gaussian Hydrophobic groups on/ around cinnamic acid ester groups positively contribute to the compound’s inhibitory activity. Fig. 2C and D show the Gaussian Hydrophobic feature of the QSAR model with the most potent (the best) inhibitor, compound 10, and the least active inhibitor, compound 1. Positive correlation between cytotoxic activity and hydrophobic regions are presented by the yellow contours that show favourable, Gaussian Hydrophobic regions. Negative correlation is presented by the white contour that shows unfavourable region. It is notable that favourable yellow contours are present around the cinnamic acid ester group of compound 10, while compound 1 has yellow contours on the furano-furone bicyclic core. Furthermore, the green contours (positive steric feature) match very well with the yellow contours (positive Gaussian Hydrophobic regions) of the showed ligand. In terms of styryl lactones inhibitory activity, hydrophobic substituents with a larger size are preferential at the C-7 and could enhance the activity of a potent drug. Even more, a hydrophobic substituents are also favourable at the C-5 position. The yellow contour near the phenyl group indicates that hydrophobic substituents in this area could also increase the inhibitory activity of a compound. According to our results, styryl lactones are proven to be potent inhibitors of PC-3 cell line. Furthermore, most of the examined compounds have desirable properties for drug candidates, which recommend them for further research in treatment of prostate cancer. Therefore, our research improved optimization of the compounds with anticancer properties, as the obtained 3D-QSAR model is able to aid screening and provide a set of guidelines for designing compounds with better inhibitory potency.
The great advantage of the 3D-QSAR modelling is the visualization of the results as 3D contour plots through the QSAR features, which are helpful for further understanding and improvement of the receptor-ligand binding regions (Liu et al., 2011). Further improvement of cytotoxic activity of compounds could be performed following the contour plots and using substituents recommended by the obtained 3D-QSAR model. The QSAR features, which include steric field, hydrogen bond donors, hydrogen bond acceptor, hydrophobic/nonpolar interactions, and electrostatic property, were analysed. The most important feature (steric feature) of the QSAR model is visualized through the compound with the highest inhibitor activity ligand 10 and the compound with the lowest activity compound 1 (Fig. 2A and B, respectively). According to the obtained model, positive correlation is observed between cytotoxic activity and steric regions at the cinnamic acid ester groups. This correlation is represented by the green contours that indicate favourable steric regions at the cinnamic acid ester groups, which could be helpful for further improvement of compounds activity. While compound 10 fits into the green contour, compound 1 doesn’t have cinnamic acid ester group and doesn’t occupy the large green contours in the model which resulted in low inhibitory activity. Obtained results suggest that steric field feature (Field fraction 0.497) on the cinnamic acid ester
4. Conclusions Cytotoxicity of synthesized natural styryl lactones (+)-goniofufurone, 7-epi-(+)-goniofufurone, (+)-crassalactones B and C and their twenty newly synthesized analogues were tested in vitro against of 5
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
Fig. 2. Steric (green: favoured, yellow: disfavoured) and hydrophobic (yellow: favoured; white: disfavoured) field contour map: A) steric field contour map compound 10; B) steric field contour map compound 1; C) hydrophobic field contour map compound 10; D) hydrophobic field contour map compound 1.
human prostate cancer cell line (PC-3). According to the obtained results this cell line is sensitive to compounds 9, 10, 11, 14, 15, 18. The tricyclic compounds with cinnamoyl (9) and fluorocinnamoyl functions (10) exhibited the significant inhibitory effect. Most of the compounds have desirable ADMET properties, with emphasis on compounds with high inhibitory activity - 9, 10, 15 and 18. Moreover, compound 9 has leadlikeness properties, which could put it in the main focus for further research. 3D-QSAR model was established using the comparative molecular field analysis (CoMFA and CoMSIA) method. Obtained results suggest that steric field feature on the cinnamic acid ester groups is crucial for the cytotoxic activity. In terms of styryl lactones inhibitory activity, hydrophobic substituents with a larger size are preferential at the C-7 and could enhance the activity of the potent drug. Even more, a hydrophobic substituent is also favourable at the C-5 position. Although compound 22 has cinnamic acid ester groups at favourable positions (atoms C-5 and C-7) its IC50 value is very low and doesn’t have cytotoxic activity, which could be explained through the fact that it doesn’t fit into the Lipinski’s rule of five. Important observation is that compounds 12 and 14, which have different stereochemistry also have different activity. Therefore, it is important to include three-dimensional structures of examined compounds in order to build reliable QSAR model and to predict the activity more accurately. Highly reliable 3D-QSAR model obtained in our research is able to aid screening of novel PC-3 inhibitors and provide a set of guidelines for designing compounds with better cytotoxic activity.
Benedeković, G., Francuz, J., Kovačević, I., Popsavin, M., Srećo Zelenović, B., Kojić, V., Bogdanović, G., Divjaković, V., Popsavin, V., 2014a. Eur. J. Med. Chem. 82, 449–458. Benedeković, G., Popsavin, M., Francuz, J., Kovačević, I., Kojić, V., Bogdanović, G., Divjaković, V., Popsavin, V., 2014b. Eur. J. Med. Chem. 87, 237–247. Bhal, S.K., Kassam, K., Peirson, I.G., Pearl, G.M., 2007. Mol. Pharm. 4, 556–560. Blázquez, M.A., Bermejo, A., Zafra-Polo, M.C., Cortes, D., 1999. Phytochem. Anal. 10, 161–170. Bruno, R.D., Gover, T.D., Burger, A.M., Brodie, A.M., Njar, V.C.O., 2008. Mol. Cancer Ther. 7, 2828–2836. Bush, B.L., Nachbar, R.B., 1993. J. Comput. Aided Mol. Des. 7, 587–619. Choo, C.Y., Abdullah, N., Diederich, M., 2014. Phytochem. Rev. 13, 835–851. Cramer, R.D., 2012. J. Comput. Aided Mol. Des. 26, 35–38. Cramer, R.D., Patterson, D.E., Bunce, J.D., 1988. J. Am. Chem. Soc. 110, 5959–5967. Daina, A., Michielin, O., Zoete, V., 2017. Sci. Rep. 7, 42717. Daina, A., Zoete, V., 2016. ChemMedChem 11, 1117–1121. De, P., Baltas, M., Bedos-Belval, F., 2011. Curr. Med. Chem. 18, 1672–1703. de Fátima, A., Modolo, L.V., Conegero, L.S., Pilli, R.A., Ferreira, C.V., Kohn, L.K., de Carvalho, J.E., 2006. Curr. Med. Chem. 13, 3371–3384. Dixon, S.L., Duan, J., Smith, E., Von Bargen, C.D., Sherman, W., Repasky, M.P., 2016. Future Med. Chem. 8, 1825–1839. Duc, V., Thanh, T.B., Thanh, H.N., Tien, V.N., 2016. J. Appl. Pharm. Sci. 6, 1–5. Egan, W.J., Merz Jr, K.M., Baldwin, J.J., 2000. J. Med. Chem. 43, 3867–3877. Ertl, P., Rohde, B., Selzer, P., 2000. J. Med. Chem. 43, 3714–3717. Fang, Y., Lu, Y., Zang, X., Wu, T., Qi, X., Pan, S., Xu, X., 2016. Sci. Rep. 6 23634-23634. Francuz, J., Srećo, B., Popsavin, M., Benedeković, G., Divjaković, V., Kojić, V., Bogdanović, G., Kapor, A., Popsavin, V., 2012. Tetrahedron Lett. 53, 1819–1822. Ghose, A.K., Viswanadhan, V.N., Wendoloski, J.J., 1999. J. Comb. Chem. 1, 55–68. Halgren, T.A., Nachbar, R.B., 1996. J. Comput. Chem. 17, 587–615. Harvey, A., 2000. Drug Discov. Today 5, 294–300. Iqbal, E., Salim, K.A., Lim, L.B.L., 2015. J. King Saud Univ. - Sci. 27, 224–232. Jing, L., Diana, C.S., Sonia, M.Fd.M., Jinghai, J.X., Robert, J.P., Steven, M.W., 2003. Curr. Top. Med. Chem. 3, 1125–1154. Klebe, G., Abraham, U., 1999. J. Comput. Aided Mol. Des. 13, 1–10. Klebe, G., Abraham, U., Mietzner, T., 1994. J. Med. Chem. 37, 4130–4146. Kovačević, I., Popsavin, M., Benedeković, G., Kojić, V., Jakimov, D., Rodić, M.V., SrdićRajić, T., Bogdanović, G., Divjaković, V., Popsavin, V., 2016. Eur. J. Med. Chem. 108, 594–604. Kumari, M., Chandra, S., Tiwari, N., Subbarao, N., 2016. BMC Struct. Biol. 16 12-12. Lipinski, C.A., Lombardo, F., Dominy, B.W., Feeney, P.J., 2001. Adv. Drug Del. Rev. 46, 3–26. Liu, J., Zhang, H., Xiao, Z., Wang, F., Wang, X., Wang, Y., 2011. Int. J. Mol. Sci. 12, 1807–1835. Martin, Y.C., 2005. J. Med. Chem. 48, 3164–3170. Mohideen, M., Zulkepli, S., Nik-Salleh, N.-S., Zulkefeli, M., Weber, J.-F.F.A., Rahman, A.F.M.M., 2013. Arch. Pharm. Res. 36, 812–831. Muegge, I., Heald, S.L., Brittelli, D., 2001. J. Med. Chem. 44, 1841–1846. Narayana Moorthy, N.S.H., Cerqueira, N.S., Ramos, M.J., Fernandes, P.A., 2012. Med. Chem. Res. 21, 133–144. O’Boyle, N.M., Banck, M., James, C.A., Morley, C., Vandermeersch, T., Hutchison, G.R., 2011. J. Cheminform. 3, 33. Popsavin, V., Benedeković, G., Srećo, B., Popsavin, M., Francuz, J., Kojić, V., Bogdanović, G., 2008. Bioorg. Med. Chem. Lett. 18, 5178–5181. Popsavin, V., Srećo, B., Benedeković, G., Francuz, J., Popsavin, M., Kojić, V., Bogdanović, G., 2010. Eur. J. Med. Chem. 45, 2876–2883. Popsavin, V., Benedeković, G., Srećo, B., Francuz, J., Popsavin, M., Kojić, V., Bogdanović,
Declaration of Competing Interest The authors declare no conflict of interest. Acknowledgement This work was supported by research grants from the Ministry of Education, Science and Technological Development of the Republic of Serbia (Grant No. OI 172006) References Abdulfatai, U., Uzairu, A., Uba, S., 2017. J. Adv. Res. 8, 33–43. Abdullah, N., Sahibul-Anwar, H., Ideris, S., Hasuda, T., Hitotsuyanagi, Y., Takeya, K., Diederich, M., Choo, C., 2013. Fitoterapia 88, 1–6. www.wcrf.org. Assis, D.N., Navarro, V.J., 2009. Expert Opin. Drug Metab. Toxicol. 5, 463–473.
6
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
451–457. Shirzad, M., Beshkar, P., Heidarian, E., 2015. J. Herbmed Pharmacol. 4, 121–124. Teague, S.J., Davis, A.M., Leeson, P.D., Oprea, T., 1999. Angew. Chem. Int. Ed. 38, 3743–3748. Testa, B., Krämer, S.D., 2007. Chem. Biodivers. 4, 2031–2122. Tetko, I.V., Tanchuk, V.Y., Villa, A.E.P., 2001. J. Chem. Inf. Comput. Sci. 41, 1407–1421. Veber, D.F., Johnson, S.R., Cheng, H.-Y., Smith, B.R., Ward, K.W., Kopple, K.D., 2002. J. Med. Chem. 45, 2615–2623. Vilar, S., Costanzi, S., 2012. Methods Mol. Biol. (Clifton, N.J.) 914, 271–284. Wiart, C., 2007. Evid. Complement. Alternat. Med. 4, 299–311. Zhao, X., Chen, M., Huang, B., Ji, H., Yuan, M., 2011. Int. J. Mol. Sci. 12, 7022–7037. Zhou, F.S., Tang, W.D., Mu, Q., Yang, G.X., Wang, Y., Liang, G.L., Lou, L.G., 2005. Chem. Pharm. Bull. (Tokyo) 53, 1387–1391.
G., Divjaković, V., 2009. Tetrahedron 65, 10596–10607. Popsavin, V., Kovacević, I., Benedeković, G., Popsavin, M., Kojić, V., Bogdanović, G., 2012. Org. Lett. 14, 5956–5959. Popsavin, V., Srećo, B., Krstić, I., Popsavin, M., Kojić, V., Bogdanović, G., 2006. Eur. J. Med. Chem. 41, 1217–1222. PreADMET, in. Ritchie, T.J., Macdonald, S.J.F., Peace, S., Pickett, S.D., Luscombe, C.N., 2013. MedChemComm 4, 673–680. Schyman, P., Liu, R., Desai, V., Wallqvist, A., 2017. Front. Pharmacol. 889. Scotti, L., Scotti, M.T., Ishiki, H., Junior, F., dos Santos, P.F., Tavares, J.F., da Silva, M.S., 2014. Nat. Prod. Commun. 9, 609–612. Seyed, M.A., Jantan, I., Bukhari, S.N.A., 2014. Biomed Res. Int. 2014. Shin, S.W., Kim, S.Y., Park, J.W., 2012. Biochim. et Biophysica Acta – Mol. Cell Res. 1823,
7
Contents lists available at ScienceDirect
Computational Biology and Chemistry journal homepage: www.elsevier.com/locate/cbac
Research Article
In vitro antitumor activity, ADME-Tox and 3D-QSAR of synthesized and selected natural styryl lactones
T
⁎
Vladimir R. Vukica, , Davor M. Loncara, Dajana V. Vukica, Lidija R. Jevrica, Goran Benedekovicb, Jovana Francuzb, Vesna Kojicc, Milica Z. Karadzic Banjaca, Velimir Popsavinb,d a
University of Novi Sad, Faculty of Technology Novi Sad, Bulevar cara Lazara 1, 21000 Novi Sad, Serbia University of Novi Sad, Faculty of Sciences, Department of Chemistry, Biochemistry and Environmental Protection, Trg Dositeja Obradovića 3, 21000 Novi Sad, Serbia c University of Novi Sad, Faculty of Medicine, Oncology Institute of Vojvodina, Put Dr Goldmana 4, 21204 Sremska Kamenica, Serbia d Serbian Academy of Sciences and Arts, Knez Mihajlova 35, 11000 Belgrade, Serbia b
A R T I C LE I N FO
A B S T R A C T
Keywords: Antitumor activity Styryl lactone ADME-Tox 3D-QSAR analysis PC-3 cells
Prostate cancer is a common cause of death in men and a novel treating methods should be developed. In order to find a new drug for prostate cancer, a series of novel conformationally constrained analogues of (+)-goniofufurone and 7-epi-(+)-goniofufurone, as well as the newly synthesized styryl lactones containing the cinnamic acid ester groups were evaluated for in vitro cytotoxicity against prostate cancer cell (PC-3). Furthermore, prediction of physicochemical characteristics and drugability as well as in silico ADME-Tox tests of investigated compounds were performed. The 3D-QSAR model was established using the comparative molecular field analysis method. According to obtained results, the tricyclic compounds 9 and 10 had the highest potency with IC50 < 20 μM. Evaluation of structural features through 3D-QSAR model identified steric field feature on the cinnamic acid ester groups at C-7 as a crucial for the cytotoxic activity. This research suggests that most of the analysed compounds have desirable properties for drug candidates and high potential in drug development, which recommend them for further research in treatment of prostate cancer. Furthermore, obtained 3D-QSAR model is able to successfully identify styryl lactones that have significant cytotoxic activity and provide information for screening and design of novel inhibitors against PC-3 cell line that could be used as drugs in treatment of the prostate cancer.
1. Introduction According to American Institute for Cancer Research and World Cancer Research Fund prostate cancer is the second most commonly occurring cancer in men and the fourth most commonly occurring cancer overall, with 1.3 million new cases in 2018 (Anon, 2019). As chemotherapy and radiation therapy are largely ineffective, a novel treating methods should be developed (Shin et al., 2012; Shirzad et al., 2015). Medicinal plants are widely used as primary healthcare to cure various diseases. Many tropical plants, such as Goniothalamus species (Family Annonaceae), have been reported to possess interesting biological activities with potential therapeutic application (Harvey, 2000;
Seyed et al., 2014). Main compounds in Goniothalamus are styryl lactones, acetogenins and alkaloids which have cytotoxic, antitumor, insecticidal, antifungal, antimalarial, antitubercular, antibacterial, antiviral and antioxidant activities (Blázquez et al., 1999; de Fátima et al., 2006; Abdullah et al., 2013; Choo et al., 2014; Iqbal et al., 2015). Styryl lactones are bioactive compounds with significant cytotoxicity against several tumour cell lines (Seyed et al., 2014; Choo et al., 2014; Wiart, 2007; Duc et al., 2016). A large body of evidence suggests that the antiproliferative activity of styryl lactones is associated with induction of apoptosis in target cells (Seyed et al., 2014; de Fátima et al., 2006). Due to their antiproliferative properties, much attention was focused on the synthesis of novel styryl lactone derivatives. Many of them exhibited strong cytotoxic activity against human stomach
Abbreviations: BBB, blood-brain barrier; CoMFA, comparative molecular field analysis; CoMSIA, comparative molecular similarity indices analysis; CV, crossvalidation; DILI, drug-induced liver injury; GIA, gastrointestinal absorption; hERG, human ether-à-go-go-related gene; HLM, human liver microsomal stability; MMP, mitochondrial membrane potential; P-gp, permeability glycoprotein; PLS, partial least squares; QSAR, quantitative structure-activity relationship; RMSD, root mean square deviation; SD, standard deviation ⁎ Corresponding author. E-mail address: [email protected] (V.R. Vukic). https://doi.org/10.1016/j.compbiolchem.2019.107112 Received 4 April 2019; Received in revised form 13 August 2019; Accepted 18 August 2019 Available online 23 August 2019 1476-9271/ © 2019 Elsevier Ltd. All rights reserved.
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
topological polar surface area (TPSA) (Ertl et al., 2000). It has been proven as a useful descriptor in many models with regards to biological barrier crossing such as gastrointestinal absorption and brain access (Daina and Zoete, 2016). In the present study, drug-likeness was established from structural or physicochemical inspections of compounds. The Lipinski (Pfizer) filter is the pioneer rule-of-five based on analysis of physicochemical properties of molecules in order to indicate weak permeation/absorption when molecular weight is higher than 500, the number of H-bond acceptors is greater than 10, the number of H-bond donors is greater than 5, and ClogP is greater than 5 (Lipinski et al., 2001). In the present study, modified Lipinski rule of 5 that includes LogD at pH = 5.5 instead of LogP values was applied (Bhal et al., 2007). In order to be accepted for oral use by this rule, the value of LogD should be less or equal to 5. Values of LogD values were calculated using PreADMET online software (PreADMET, 2019). The Ghose (Amgen), Veber (GSK), Egan (Pharmacia) and Muegge (Bayer) methods were also used in order to predict drug-likeness of the compounds (Ghose et al., 1999; Veber et al., 2002; Egan et al., 2000; Muegge et al., 2001). Multiple lipophilicity predictors were used in order to generate consensus estimation represented as Consensus Log P. Liver toxicity (DILI, cytotoxicity), metabolism (HLM, CYP1A2, CYP2C19, CYP2C9, CYP2D6 and CYP3A4 inhibitor), membrane transporters (P-gp inhibitor, P-gp substrate), hERG blocker, MMP and AMES test were performed using vNN-ADMET online server (Schyman et al., 2017). Cytotoxicity is the degree to which a chemical causes damage to cells. In the present study, test compounds with an IC50 of 100 μM or less are considered as cytotoxic.
cancer cells SGC-7901 (Zhou et al., 2005), oestrogen receptor-positive breast adenocarcinoma cells MCF-7 (Kovačević et al., 2016). Raji Burkitt's lymphoma cells (Popsavin et al., 2008, 2010; Francuz et al., 2012; Benedeković et al., 2014a, b), alveolar basal adenocarcinoma cells A549 (Zhou et al., 2005), promyelocytic leukaemia cells HL-60 (Zhou et al., 2005; Popsavin et al., 2008, 2010; Francuz et al., 2012; Benedeković et al., 2014b; Popsavin et al., 2009), T lymphocyte leukaemia Jurkat cells (Francuz et al., 2012; Popsavin et al., 2009; Mohideen et al., 2013), myelogenous leukaemia cells K-562 (Popsavin et al., 2010; Benedeković et al., 2014a), cervix carcinoma cells HeLa (Benedeković et al., 2014a; Popsavin et al., 2012), melanoma cells Hs 294 T (Benedeković et al., 2014a). The in silico prediction of cytotoxic and antiproliferative activity has been identified as a very important step in the synthesis of compounds with desired biological activity (Narayana Moorthy et al., 2012; Scotti et al., 2014; Abdulfatai et al., 2017). In that aim, three-dimensional quantitative structure-activity relationship (3D-QSAR) was successfully used in the early stage of preclinical drug development (Fang et al., 2016; Kumari et al., 2016). Therefore, the present study is a continuation of the previous research with the aim to determine in vitro cytotoxic activity of some newly synthesized and natural styryl lactones, their derivatives, and analogues against prostate cancer cell (PC-3). Furthermore, in silico evaluation of ADME-Tox properties were examined and a predictive 3DQSAR model that will be used for a lead optimization and testing of novel compounds was developed. 2. Materials and methods
2.4. D-QSAR analysis 2.1. Styryl lactones The 3D structures of all compounds are constructed using Chemsketch, a chemically intelligent drawing interface freeware. The geometry optimization of all compounds was carried out using the MMFF94 force field with MMFF94 partial charges. The Powell conjugated gradient algorithm method was implemented with convergence criterion of 0.042 kJ/mol Å, using the maximum iterations set to 1000 (Halgren and Nachbar, 1996). Structure alignment is one of the most critical step in CoMFA (Comparative Molecular Field Analysis) and CoMSIA (Comparative Molecular Similarity Indices Analysis) analysis. So, selection of the template molecule for the alignment was a crucial step in the construction of the QSAR models (Liu et al., 2011). Considering the structural similarities of the studied styryl lactones, furano-furone bicyclic core was selected as a common structural segment for the alignment. The most active compound is used as a template for superimposition, which is assumed to represent the most bioactive conformation. The set of 24 ligands was randomly divided into two subsets: the training set for model building (20 compounds) and the test set (4 compounds) which was used as external validation set (Dixon et al., 2016). The 3D-QSAR analysis was performed on the aligned compounds using the field-based QSAR which is based on CoMFA and CoMSIA (Cramer, 2012; Cramer et al., 1988; Klebe and Abraham, 1999; Klebe et al., 1994). CoMFA and CoMSIA are based on the representation of ligands through molecular fields measured in the space that surrounds them. 3-D contour map of the physicochemical forces that surround a series of aligned compounds was constructed and the interaction energy of an appropriate probe in each of the points in that 3-D contour map was used as structural descriptors to be correlated with biological activity during construction of 3D-QSAR model. CoMFA field-based models are constructed by calculating the value of the electrostatic and steric molecular field, on a rectangular grid that encompasses the molecules in the training set and correlated with bioactivities by use of PLS (Partial Least Squares) using three-factor regression model. CoMSIA is encompassing the steric, electrostatic, hydrogen bonding and hydrophobic effects of ligands fields which are also evaluated at points on a rectangular grid. Fields are calculated by summing the values of
Four naturally occurring styryl lactones (1, 2, 11, 18) and twenty newly synthesized analogues thereof (3, 4‒10, 12‒17, 19‒24) were examined in this work. Synthesis of the compounds used in this research is already published: total syntheses of 7-epi-(+)-goniofufurone (1) and (+)-goniofufurone (2) are in detail described in the work (Benedeković et al., 2014a); (+)-Crassalactones B (11) and C (18), the corresponding 5,7-dicinnamoate 22, a number of new analogues such as 12, 13, 19‒21, the corresponding 7-epimers such as 14, 15, 23, 24, 7-deoxy derivatives such as 16, 17 (Benedeković et al., 2014b), as well as 3 and tricyclic lactones 4‒10 (Benedeković et al., 2014a) have been prepared by semi-synthesis starting from compounds 1 or 2. 2.2. Cytotoxic activity The investigated compounds were tested in vitro against human prostate tumour cell line (PC-3). Cytotoxic activity was evaluated by using the standard colourimetric MTT assay, after exposure of cell line to the tested compounds for 72 h (Popsavin et al., 2006). The commercial antitumor agent doxorubicin (DOX) was used as a positive control. The results are expressed as IC50 and pIC50, the concentration of compound required to inhibit the cell growth by 50% compared to an untreated control. 2.3. ADME-Tox properties and druglikness test The molecular descriptors were calculated on the basis of 2D molecular structures, presented in the Supplementary data (Table S1). ADME test including iLOGP XLOGP3, WLOGP, MLOGP, Silicos-IT Log P, GI absorption, Blood-brain barrier as well as druglikeness including Ghose, Veber, Egan and Muegge violations and Bioavailability Score were performed using SwissADME (Daina et al., 2017). Simple molecular and physicochemical descriptors like molecular weight (MW), molecular refractivity (MR), counts of specific atom types and polar surface area (PSA) are computed with Open Babel, version 2.3.0 (O’Boyle et al., 2011). PSA is calculated using a technique called 2
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
properties of a given atom, weighted by a Gaussian function of the distance between the grid point and the atom. The field-based QSAR models are an implementation of the CoMFA and CoMSIA methods with a specific set of parameters. The CoMFA and CoMSIA descriptor fields were obtained by a 3D cubic lattice with a grid spacing of 1 Å. In the CoMFA model, a hybrid sp3 carbon atom with a positive charge was used as a probe to compute the CoMFA steric and electrostatic fields in which their energy values were truncated at 30 kcal/mol in order to eliminate large repulsive energies that are consequence of close contact between the probe and the molecule (Vilar and Costanzi, 2012; Zhao et al., 2011). The lattice size and probe step size were adjusted automatically. The partial least squares (PLS) analysis is used to construct the best model using linear correlation the CoMFA and CoMSIA fields to the pIC50 (Bush and Nachbar, 1993). Cross-validation analysis was performed using the leave-one-out method (Tetko et al., 2001). The optimum number of components used in the final analysis was identified by the cross-validation method. Correlation and cross-validation coefficients (r2 and q2 respectively) were calculated according to the formula:
r2 = 1 −
Table 1 Cytotoxic activity of examined compounds against PC-3 cell line. IC50 (μM)
Comp. label
IC50 (μM)
Comp. label
IC50 (μM)
1 2 3 4 5 6 7 8
899.90 254.36 487.66 185.35 537.78 334.98 861.37 149.03
9 10 11 12 13 14 15 16
15.64 12.31 25.11 193.43 101.21 54.39 24.35 286.22
17 18 19 20 21 22 23 24
114.32 31.08 191.51 702.22 131.21 401.21 101.32 105.41
(+)-goniofufurone and 7-epi-(+)-goniofufurone (Benedeković et al., 2014a, b) as well as chloro and bromo (+)-goniofufurone mimics (Francuz et al., 2012) showed in most cases stronger cytotoxicity compared to corresponding lead compounds. However, when another electron-withdrawing group (NO2) is introduced in the same position the activity is not increased (compared to 11 and 15). But, insertion of the electron-donating group OCH3 in the para position of cinnamoate aromatic ring resulted in decrease of cytotoxic activity (compared to 11 and 14). Considering the influence of absolute stereochemistry at the C7 position, (7R)-stereoisomer 9 displayed the higher potency than (7S)isomer 8 (Table S1). Similar results were obtained by Benedeković et al. (2014a) after evaluation of their cytotoxicity against a panel of eight human tumour cell lines. Overall, some analogues with cinnamic acid ester group have high cytotoxic activity, which justified the main reason for the preparation of these derivatives in order to combine antitumor potential of styryl lactones and cinnamoates (Zhou et al., 2005; Kovačević et al., 2016; Popsavin et al., 2008, 2010; Francuz et al., 2012; Benedeković et al., 2014a, b; Popsavin et al., 2009; Mohideen et al., 2013; Popsavin et al., 2012; De et al., 2011).
(Yobs − Ypred )2 (Yobs − Ymean )2
q2 = 1 −
Comp. label
(Yobs − YCVpred )2 (Yobs − Ymean )2
Yobs – observed activity Ypred – predicted activity Ymean – average activity of the entire training data set YCVpred – cross-validated predicted activity In order to test “chance correlations” r2 Scramble test was performed by calculating r2 from a series of models using scrambled activities (Yscrambling). The CoMFA/CoMSIA results were graphically interpreted using Maestro software by field contribution maps using field type “stdev*coeff”, and the contour levels set to default values.
3.2. ADME-Tox properties ADMET (absorption, distribution, metabolism, excretion and toxicity) prediction is a useful analysis in medicinal chemistry for the selection of drug-like molecules (Jing et al., 2003). Lipophilicity, characterized here by computed logP and logD values, plays a crucial role in determining several ADMET parameters as well as potency. Also, having a soluble molecule greatly facilitates many drug development activities, primarily the ease of handling and formulation (Ritchie et al., 2013). Water solubility is represented through Silicos-IT values. Results of in silico lipophilicity (iLOGP, XLOGP 3, WLOGP, MLOGP, Consensus LogP and SklogD) and water solubility (Silicos-IT LogS) are presented in the Table 2. According to the obtained results, compounds with high cytotoxic activity (9, 10, 11 and 15) have moderate lipophilicity, which ranks them in a group of drug candidates. Other tested compounds, except compound 22, also have appropriate lipophilicity to be recommended for drug candidates. In silico toxicity predictions of analysed compounds are presented in the Table 3. The knowledge about the interaction of molecules with cytochromes P450 (CYP) is essential as it is a key step in drug elimination through metabolic biotransformation (Testa and Krämer, 2007). In order to maintain an effective concentration in our blood, a drug should not be rapidly metabolized by CYPs. Most potent compounds in our research (9, 10, 15 and 18) should not interact with CYP enzymes.. Drug-induced liver injury (DILI) has been one of the most commonly cited reasons for drug withdrawals from the market (Assis and Navarro, 2009). Therefore, it is essential to eliminate that possibility in early phases of drug development. Several compounds in our research could potentially induce liver injury, but most of the promising drug candidates have negative in silico results. Also, the most potent drug candidates should not interact with potassium ion channels (hERG blocker) or express mutagenic effects. Although most compounds should not inhibit P-gp, some of them could be transported out of the cell by it,
3. Results 3.1. Structure and biological activities The names and 2D structures of studied compounds are presented in the Table S1, supplementary material. All of the examined molecules have a common structural segment, furano-furone bicyclic core, tetrahydro-furo[3,2-b]furan-2-one. The compounds 4‒10 represent conformationally constrained analogues of 1 or 2. Compounds 8-10, 11‒17, 18‒24 represent the cinnamic acid ester derivatives. The main reason for the preparation of these derivatives arises from the fact that certain cinnamoates exhibit significant antitumor activities (De et al., 2011). All of the examined compounds were completely inactive toward normal MRC-5 cells (IC50 > 100 μM), with the exception of dicinnamoate 22 which showed a weak cytotoxicity (IC50 48.64 μM). The antiproliferative activity of the commercial antitumor agent doxorubicin (DOX) was 84.23 μM. The IC50 values of tested molecules showed that PC-3 line is sensitive to compounds 9, 10, 11, 14, 15 and 18, (Table 1). The tricyclic compounds 9 and 10 had IC50 < 20 μM and the compounds 11, 14, 15 and 18 demonstrated cytotoxicity between 20 and 60 μM. However, these compounds have from 1.5 (14) to 7 (10) times higher activity than DOX. The compounds with IC50 values higher than 100 μM (1‒3, 4‒8, 12, 13, 16, 17, 19‒24) are considered to be non-cytotoxic. Abiraterone, which is currently used for the treatment of prostate cancer, has IC50 value 9.32 (Bruno et al., 2008). The analogue 10, which contains electron-withdrawing group F in the para position of cinnamoate aromatic ring, was more potent compared to molecule 9, which doesn’t have F group. Generally, fluoro derivatives of 3
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
According to our results, bioavailability of tested compounds is satisfactory with the bioavailability score of 0.55 for all compounds (Martin, 2005). Druglike properties are most commonly defined using the “rules of five”. However, the properties required of a compound intended to provide leads suitable for further optimization may be rather different (Teague et al., 1999). These include molecular weight, XLOGP and number of rotatable bonds. Our results revealed that compound 9 is suitable as a lead, which could put it in the main focus for further research, although the most potent compound is 10.
Table 2 In silico lipophilicity of analysed compounds. Molecule
iLOGP
XLOGP3
WLOGP
MLOGP
SilicosIT LogS
Consensus LogP
SklogD
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1.57 1.69 2.78 1.83 2.42 2.24 2.1 2.11 2.57 2.3 2.88 3.45 2.8 3.09 2.4 3.05 2.98 2.94 2.29 3.16 3.02 3.75 3.12 3.89
0.43 0.43 1.91 1.31 0.77 0.82 0.82 1.43 1.43 1.53 2.53 2.51 2.63 2.51 2.36 3.65 3.75 2.53 2.36 2.51 2.63 5.19 2.36 2.51
−0.16 −0.16 1.91 1.02 0.83 0.86 0.86 0.92 0.92 1.48 1.99 2 2.55 2 1.9 2.83 3.39 1.99 1.9 2 2.55 4.15 1.9 2
0.18 0.18 1.64 0.76 0.7 0.85 0.85 1.03 1.03 1.42 1.93 1.61 2.3 1.61 1.04 2.74 3.12 1.93 1.04 1.61 2.3 3.4 1.04 1.61
0.79 0.79 2.74 1.09 1.24 1.43 1.43 1.38 1.38 1.8 2.87 2.94 3.29 2.94 0.71 3.65 4.07 2.87 0.71 2.94 3.29 5.04 0.71 2.94
0.56 0.58 2.2 1.2 1.19 1.24 1.21 1.37 1.47 1.71 2.44 2.5 2.71 2.43 1.68 3.18 3.46 2.45 1.66 2.44 2.76 4.31 1.83 2.59
0.10 0.10 2.15 1.42 1.00 0.98 0.98 1.42 1.42 1.56 2.56 2.51 2.70 2.51 2.57 3.34 3.48 2.70 2.70 2.65 2.83 5.16 2.70 2.65
3.4. D-QSAR A tridimensional quantitative structure-activity relationship (3DQSAR) model was established using the comparative molecular field analysis (CoMFA and CoMSIA) method. The robustness and the predictive ability of the generated 3D-QSAR model were examined by using the test set. Statistical evaluation of the two-factor and threefactor QSAR models revealed good correlation between predicted and experimental inhibitor activity values ( r2 = 0.84, 0.93 and q2 = 0.64, 0.86, respectively) indicating a good predictability of the constructed models. Although three-factor model has higher values of r2 and q2, stability of the model was 0.44, indicating over-fitting of the model. Therefore, the two-factor model is used for further analysis. The QSAR statistical analysis is summarized in the Table 4. The QSAR model (two-factor) using steric, electrostatic, hydrophobic and hydrogen bond donor and acceptor fields had high determination coefficient – r2 (0.839) with low standard deviation – SD (0.240), high cross-validation coefficient q2 (0.640) and high Fischer ratio – F (46.9). Furthermore, values of r and r2 for the test set were 0.964 and 0.930, respectively. The measured and the predicted values of activities expressed as pIC50 are presented in the Table 5. These results indicate high applicability of constructed model in testing and design of new compounds (potentially by using compound 9 as a lead) with high cytotoxic activity toward PC-3 cell line. Fig. 1 displays the scatterplot of the observed versus the predicted pIC50 values for the best 3D-QSAR model. The distribution of the data around the regression line (r2 = 0.84) suggests that the model is accurate enough to be used in further analysis.
which could influence druglikness of those compounds. 3.3. Druglikeness of the studied compounds In the present study, druglikeness tests were performed in order to exclude molecules with properties that are incompatible with an acceptable pharmacokinetic profile (Table S2). Obtained results indicate that all tested compounds except for the compound 22 have pass the demands defined by modified Lipinski rule of five, as well as by Ghose, Veber, Egan and Muegge tests, which recommends these compounds as possible oral drugs. High gastrointestinal absorption observed in all compounds is also a desirable characteristic for a potential drug. Table 3 In silico toxicity of analysed compounds. Liver toxicity
Cyp inhibitors
Membrane transporters
Molecule
DILI
Cyto-toxicity
Cyp1A2
Cyp3A4
Cyp2D6
Cyp2C9
Cyp2C19
BBB permeab.
P-gp Inhibitor
P-gp Substrate
hERG Blocker
MMP
AMES
HLM
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
– – – + + + + – – + – – – – – – – – + + – – + +
– – – – – – – – – + + – + – – – – + + + + + + +
– – – – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – + – + – – – – – + – – – +
– – + – – + + – – – – – – – – + + – – – – – – –
+ + + – – – – – – – – – – – – – – – – – – – – –
+ + – – – – – + + + + + + + + + + – + – + – + –
– – + – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – – – – – – – – – –
– – – – – – – – – – – – – – – + – – – – – – – –
+ + + + + + + + + + + + + + + + + + + + + + + +
4
Other
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
Table 4 Summary of statistical parameters obtained for the generated 3D-QSAR model. Statistical parameters
QSAR model One-Factor
SD r2 r2 Scramble Stability F p q2 Pearson-r
0.364 0.610 0.188 0.68 29.7 2.93e−005 0.334 0.927 Fraction of field contributions Gaussian Steric 0.481 Gaussian Electrostatic 0.088 Gaussian Hydrophobic 0.227 Gaussian Donor 0.169 Gaussian Acceptor 0.035
Two-Factor
Three-Factor
0.240 0.839 0.223 0.54 46.9 7.28e−008 0.640 0.928
0.168 0.926 0.827 0.44 70.8 8.19e−010 0.856 0.948
0.497 0.092 0.229 0.138 0.043
0.475 0.097 0.233 0.051 0.143
Fig. 1. The plot of predicted versus experimental pIC50 values for 3D-QSAR analysis. The solid line is the regression line for the fitted and predicted bioactivities of training and test set.
Table 5 The measured and predicted activities (pIC50) by the obtained 3D-QSAR model. Ligand name
QSAR Set
Activity
Predicted activity
Prediction error
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
training training training training training training training test training training training training test training training training training training training training test test training training
3.05 3.59 3.31 3.73 3.27 3.48 3.06 3.83 4.80 4.91 4.60 3.71 3.99 4.26 4.61 3.54 3.94 4.51 3.72 3.15 3.88 3.40 3.99 3.98
3.36 3.77 3.66 3.39 3.39 3.47 3.30 3.60 4.90 4.90 4.11 3.86 3.97 3.98 4.04 3.69 3.70 4.09 3.67 3.20 3.79 3.53 4.21 4.24
−0.31 −0.18 −0.34 0.34 −0.12 0.01 −0.23 0.23 −0.09 0.01 0.49 −0.14 0.02 0.29 0.57 −0.15 0.24 0.41 0.04 −0.04 0.09 −0.14 −0.22 −0.26
groups is crucial for cytotoxic activity of examined styryl lactones. Therefore, the compounds which contain the cinnamic acid ester group, showed, in most cases, higher activity compared to compounds without cinnamic acid ester group (Table 1). This is not always consistent, as compounds with different stereochemistry have different activity. For example, compounds 12 and 14 have the same formula, but different stereochemistry, which induced huge differences in binding to the protein and cytotoxic activity. Therefore, to predict the activity of a compound, it is important to include its three-dimensional structure to obtain more reliable results. In addition to the steric feature, Gaussian Hydrophobic groups on/ around cinnamic acid ester groups positively contribute to the compound’s inhibitory activity. Fig. 2C and D show the Gaussian Hydrophobic feature of the QSAR model with the most potent (the best) inhibitor, compound 10, and the least active inhibitor, compound 1. Positive correlation between cytotoxic activity and hydrophobic regions are presented by the yellow contours that show favourable, Gaussian Hydrophobic regions. Negative correlation is presented by the white contour that shows unfavourable region. It is notable that favourable yellow contours are present around the cinnamic acid ester group of compound 10, while compound 1 has yellow contours on the furano-furone bicyclic core. Furthermore, the green contours (positive steric feature) match very well with the yellow contours (positive Gaussian Hydrophobic regions) of the showed ligand. In terms of styryl lactones inhibitory activity, hydrophobic substituents with a larger size are preferential at the C-7 and could enhance the activity of a potent drug. Even more, a hydrophobic substituents are also favourable at the C-5 position. The yellow contour near the phenyl group indicates that hydrophobic substituents in this area could also increase the inhibitory activity of a compound. According to our results, styryl lactones are proven to be potent inhibitors of PC-3 cell line. Furthermore, most of the examined compounds have desirable properties for drug candidates, which recommend them for further research in treatment of prostate cancer. Therefore, our research improved optimization of the compounds with anticancer properties, as the obtained 3D-QSAR model is able to aid screening and provide a set of guidelines for designing compounds with better inhibitory potency.
The great advantage of the 3D-QSAR modelling is the visualization of the results as 3D contour plots through the QSAR features, which are helpful for further understanding and improvement of the receptor-ligand binding regions (Liu et al., 2011). Further improvement of cytotoxic activity of compounds could be performed following the contour plots and using substituents recommended by the obtained 3D-QSAR model. The QSAR features, which include steric field, hydrogen bond donors, hydrogen bond acceptor, hydrophobic/nonpolar interactions, and electrostatic property, were analysed. The most important feature (steric feature) of the QSAR model is visualized through the compound with the highest inhibitor activity ligand 10 and the compound with the lowest activity compound 1 (Fig. 2A and B, respectively). According to the obtained model, positive correlation is observed between cytotoxic activity and steric regions at the cinnamic acid ester groups. This correlation is represented by the green contours that indicate favourable steric regions at the cinnamic acid ester groups, which could be helpful for further improvement of compounds activity. While compound 10 fits into the green contour, compound 1 doesn’t have cinnamic acid ester group and doesn’t occupy the large green contours in the model which resulted in low inhibitory activity. Obtained results suggest that steric field feature (Field fraction 0.497) on the cinnamic acid ester
4. Conclusions Cytotoxicity of synthesized natural styryl lactones (+)-goniofufurone, 7-epi-(+)-goniofufurone, (+)-crassalactones B and C and their twenty newly synthesized analogues were tested in vitro against of 5
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
Fig. 2. Steric (green: favoured, yellow: disfavoured) and hydrophobic (yellow: favoured; white: disfavoured) field contour map: A) steric field contour map compound 10; B) steric field contour map compound 1; C) hydrophobic field contour map compound 10; D) hydrophobic field contour map compound 1.
human prostate cancer cell line (PC-3). According to the obtained results this cell line is sensitive to compounds 9, 10, 11, 14, 15, 18. The tricyclic compounds with cinnamoyl (9) and fluorocinnamoyl functions (10) exhibited the significant inhibitory effect. Most of the compounds have desirable ADMET properties, with emphasis on compounds with high inhibitory activity - 9, 10, 15 and 18. Moreover, compound 9 has leadlikeness properties, which could put it in the main focus for further research. 3D-QSAR model was established using the comparative molecular field analysis (CoMFA and CoMSIA) method. Obtained results suggest that steric field feature on the cinnamic acid ester groups is crucial for the cytotoxic activity. In terms of styryl lactones inhibitory activity, hydrophobic substituents with a larger size are preferential at the C-7 and could enhance the activity of the potent drug. Even more, a hydrophobic substituent is also favourable at the C-5 position. Although compound 22 has cinnamic acid ester groups at favourable positions (atoms C-5 and C-7) its IC50 value is very low and doesn’t have cytotoxic activity, which could be explained through the fact that it doesn’t fit into the Lipinski’s rule of five. Important observation is that compounds 12 and 14, which have different stereochemistry also have different activity. Therefore, it is important to include three-dimensional structures of examined compounds in order to build reliable QSAR model and to predict the activity more accurately. Highly reliable 3D-QSAR model obtained in our research is able to aid screening of novel PC-3 inhibitors and provide a set of guidelines for designing compounds with better cytotoxic activity.
Benedeković, G., Francuz, J., Kovačević, I., Popsavin, M., Srećo Zelenović, B., Kojić, V., Bogdanović, G., Divjaković, V., Popsavin, V., 2014a. Eur. J. Med. Chem. 82, 449–458. Benedeković, G., Popsavin, M., Francuz, J., Kovačević, I., Kojić, V., Bogdanović, G., Divjaković, V., Popsavin, V., 2014b. Eur. J. Med. Chem. 87, 237–247. Bhal, S.K., Kassam, K., Peirson, I.G., Pearl, G.M., 2007. Mol. Pharm. 4, 556–560. Blázquez, M.A., Bermejo, A., Zafra-Polo, M.C., Cortes, D., 1999. Phytochem. Anal. 10, 161–170. Bruno, R.D., Gover, T.D., Burger, A.M., Brodie, A.M., Njar, V.C.O., 2008. Mol. Cancer Ther. 7, 2828–2836. Bush, B.L., Nachbar, R.B., 1993. J. Comput. Aided Mol. Des. 7, 587–619. Choo, C.Y., Abdullah, N., Diederich, M., 2014. Phytochem. Rev. 13, 835–851. Cramer, R.D., 2012. J. Comput. Aided Mol. Des. 26, 35–38. Cramer, R.D., Patterson, D.E., Bunce, J.D., 1988. J. Am. Chem. Soc. 110, 5959–5967. Daina, A., Michielin, O., Zoete, V., 2017. Sci. Rep. 7, 42717. Daina, A., Zoete, V., 2016. ChemMedChem 11, 1117–1121. De, P., Baltas, M., Bedos-Belval, F., 2011. Curr. Med. Chem. 18, 1672–1703. de Fátima, A., Modolo, L.V., Conegero, L.S., Pilli, R.A., Ferreira, C.V., Kohn, L.K., de Carvalho, J.E., 2006. Curr. Med. Chem. 13, 3371–3384. Dixon, S.L., Duan, J., Smith, E., Von Bargen, C.D., Sherman, W., Repasky, M.P., 2016. Future Med. Chem. 8, 1825–1839. Duc, V., Thanh, T.B., Thanh, H.N., Tien, V.N., 2016. J. Appl. Pharm. Sci. 6, 1–5. Egan, W.J., Merz Jr, K.M., Baldwin, J.J., 2000. J. Med. Chem. 43, 3867–3877. Ertl, P., Rohde, B., Selzer, P., 2000. J. Med. Chem. 43, 3714–3717. Fang, Y., Lu, Y., Zang, X., Wu, T., Qi, X., Pan, S., Xu, X., 2016. Sci. Rep. 6 23634-23634. Francuz, J., Srećo, B., Popsavin, M., Benedeković, G., Divjaković, V., Kojić, V., Bogdanović, G., Kapor, A., Popsavin, V., 2012. Tetrahedron Lett. 53, 1819–1822. Ghose, A.K., Viswanadhan, V.N., Wendoloski, J.J., 1999. J. Comb. Chem. 1, 55–68. Halgren, T.A., Nachbar, R.B., 1996. J. Comput. Chem. 17, 587–615. Harvey, A., 2000. Drug Discov. Today 5, 294–300. Iqbal, E., Salim, K.A., Lim, L.B.L., 2015. J. King Saud Univ. - Sci. 27, 224–232. Jing, L., Diana, C.S., Sonia, M.Fd.M., Jinghai, J.X., Robert, J.P., Steven, M.W., 2003. Curr. Top. Med. Chem. 3, 1125–1154. Klebe, G., Abraham, U., 1999. J. Comput. Aided Mol. Des. 13, 1–10. Klebe, G., Abraham, U., Mietzner, T., 1994. J. Med. Chem. 37, 4130–4146. Kovačević, I., Popsavin, M., Benedeković, G., Kojić, V., Jakimov, D., Rodić, M.V., SrdićRajić, T., Bogdanović, G., Divjaković, V., Popsavin, V., 2016. Eur. J. Med. Chem. 108, 594–604. Kumari, M., Chandra, S., Tiwari, N., Subbarao, N., 2016. BMC Struct. Biol. 16 12-12. Lipinski, C.A., Lombardo, F., Dominy, B.W., Feeney, P.J., 2001. Adv. Drug Del. Rev. 46, 3–26. Liu, J., Zhang, H., Xiao, Z., Wang, F., Wang, X., Wang, Y., 2011. Int. J. Mol. Sci. 12, 1807–1835. Martin, Y.C., 2005. J. Med. Chem. 48, 3164–3170. Mohideen, M., Zulkepli, S., Nik-Salleh, N.-S., Zulkefeli, M., Weber, J.-F.F.A., Rahman, A.F.M.M., 2013. Arch. Pharm. Res. 36, 812–831. Muegge, I., Heald, S.L., Brittelli, D., 2001. J. Med. Chem. 44, 1841–1846. Narayana Moorthy, N.S.H., Cerqueira, N.S., Ramos, M.J., Fernandes, P.A., 2012. Med. Chem. Res. 21, 133–144. O’Boyle, N.M., Banck, M., James, C.A., Morley, C., Vandermeersch, T., Hutchison, G.R., 2011. J. Cheminform. 3, 33. Popsavin, V., Benedeković, G., Srećo, B., Popsavin, M., Francuz, J., Kojić, V., Bogdanović, G., 2008. Bioorg. Med. Chem. Lett. 18, 5178–5181. Popsavin, V., Srećo, B., Benedeković, G., Francuz, J., Popsavin, M., Kojić, V., Bogdanović, G., 2010. Eur. J. Med. Chem. 45, 2876–2883. Popsavin, V., Benedeković, G., Srećo, B., Francuz, J., Popsavin, M., Kojić, V., Bogdanović,
Declaration of Competing Interest The authors declare no conflict of interest. Acknowledgement This work was supported by research grants from the Ministry of Education, Science and Technological Development of the Republic of Serbia (Grant No. OI 172006) References Abdulfatai, U., Uzairu, A., Uba, S., 2017. J. Adv. Res. 8, 33–43. Abdullah, N., Sahibul-Anwar, H., Ideris, S., Hasuda, T., Hitotsuyanagi, Y., Takeya, K., Diederich, M., Choo, C., 2013. Fitoterapia 88, 1–6. www.wcrf.org. Assis, D.N., Navarro, V.J., 2009. Expert Opin. Drug Metab. Toxicol. 5, 463–473.
6
Computational Biology and Chemistry 83 (2019) 107112
V.R. Vukic, et al.
451–457. Shirzad, M., Beshkar, P., Heidarian, E., 2015. J. Herbmed Pharmacol. 4, 121–124. Teague, S.J., Davis, A.M., Leeson, P.D., Oprea, T., 1999. Angew. Chem. Int. Ed. 38, 3743–3748. Testa, B., Krämer, S.D., 2007. Chem. Biodivers. 4, 2031–2122. Tetko, I.V., Tanchuk, V.Y., Villa, A.E.P., 2001. J. Chem. Inf. Comput. Sci. 41, 1407–1421. Veber, D.F., Johnson, S.R., Cheng, H.-Y., Smith, B.R., Ward, K.W., Kopple, K.D., 2002. J. Med. Chem. 45, 2615–2623. Vilar, S., Costanzi, S., 2012. Methods Mol. Biol. (Clifton, N.J.) 914, 271–284. Wiart, C., 2007. Evid. Complement. Alternat. Med. 4, 299–311. Zhao, X., Chen, M., Huang, B., Ji, H., Yuan, M., 2011. Int. J. Mol. Sci. 12, 7022–7037. Zhou, F.S., Tang, W.D., Mu, Q., Yang, G.X., Wang, Y., Liang, G.L., Lou, L.G., 2005. Chem. Pharm. Bull. (Tokyo) 53, 1387–1391.
G., Divjaković, V., 2009. Tetrahedron 65, 10596–10607. Popsavin, V., Kovacević, I., Benedeković, G., Popsavin, M., Kojić, V., Bogdanović, G., 2012. Org. Lett. 14, 5956–5959. Popsavin, V., Srećo, B., Krstić, I., Popsavin, M., Kojić, V., Bogdanović, G., 2006. Eur. J. Med. Chem. 41, 1217–1222. PreADMET, in. Ritchie, T.J., Macdonald, S.J.F., Peace, S., Pickett, S.D., Luscombe, C.N., 2013. MedChemComm 4, 673–680. Schyman, P., Liu, R., Desai, V., Wallqvist, A., 2017. Front. Pharmacol. 889. Scotti, L., Scotti, M.T., Ishiki, H., Junior, F., dos Santos, P.F., Tavares, J.F., da Silva, M.S., 2014. Nat. Prod. Commun. 9, 609–612. Seyed, M.A., Jantan, I., Bukhari, S.N.A., 2014. Biomed Res. Int. 2014. Shin, S.W., Kim, S.Y., Park, J.W., 2012. Biochim. et Biophysica Acta – Mol. Cell Res. 1823,
7