Paulo Se(´)rgio Gonçalves Nunesa, Gabriel da Silvab, Sofia Nascimentoc,Susimaire Pedersoli Mantoania, Peterson de Andrade a, Emerson Soares Bernardesc,Daniel Fa(´)bio Kawano d, Andreia Machado Leopoldinob, Ivone Carvalhoa,*
Keywords:4-aminoquinazolines;Click Chemistry;ERK kinase
ABSTRACT
ERK1/2 inhibitors have attracted special attention concerning the ability of circumventing cases of innate or log- term acquired resistance to RAF and MEK kinase inhibitors. Based on the 4-aminoquinazoline pharmacophore of kinases, herein we describe the synthesis of 4-aminoquinazoline derivatives bearing a 1,2,3-triazole stable core to bridge different aromatic and heterocyclic rings using copper-catalysed azide-alkyne cycloaddition reaction (CuAAC) as a Click Chemistry strategy. The initial screening of twelve derivatives in tumoral cells (CAL-27, HN13, HGC-27, and BT-20) revealed that the most active in BT-20 cells (25a, IC50 24.6 μM and a SI of 3.25) contains a more polar side chain (sulfone). Furthermore, compound 25a promoted a significant release of lactate dehydrogenase (LDH), suggesting the induction of cell death by necrosis. In addition, this compound induced G0/G1 stalling in BT-20 cells, which was accompanied by a decrease in the S phase. Western blot analysis of the levels of p-STAT3, p-ERK, PARP, p53 and cleaved caspase-3 revealed p-ERK1/2 and p-STA3 were drastically decreased in BT-20 cells under 25a incubation, suggesting the involvement of these two kinases in the mecha- nisms underlying 25a-induced cell cycle arrest, besides loss of proliferation and viability of the breast cancer cell. Molecular docking simulations using the ERK-ulixertinib crystallographic complex showed compound 25a could potentially compete with ATP for binding to ERK in a slightly higher affinity than the reference ERK1/2 in- hibitor. Further in silico analyses showed comparable toxicity and pharmacokinetic profiles for compound 25a in relation to ulixertinib.
1.Introduction
Malignant tumors encompassing uncontrolled cell growth and metastasis is an alarming and devastating disease that affects about 18 million people annually (2019) and accounts for 9.6 million deaths (2018), placing the disease as the second leading cause of death globally, mainly in low- and middle-income countries (70%) [1,2]. Even in this scenario, advances in cancer chemotherapy had a significant contribu- tion to patient survival rates since, in association with an increase in screening campaigns, they will reduce the development of advanced stage tumours [3]. In fact, effective combination therapy has evolved dramatically, although many of these drugs act against common targets found in normal and cancer cells. Cytotoxic drugs, for instance, can induce selective DNA damage based on its higher concentration in cancer cells that replicate and grow faster than normal cells. However, a range of limitations have arisen as these conventional cytotoxic drugs may not reach a dormant tumour besides inducing harmful side effects to patients due to the lack of selectivity towards rapid grow normal cells. To address these issues, a better understanding of genetic and molecular biology of cancer cells has emerged and led to the molecular target therapeutic discovery of highly selective drugs to target a unique or overexpressed protein in cancer cells, with less toxic side effects [4].
Inhibitors of the signalling transduction pathways to target excessive quantities of protein kinases and receptor tyrosine kinases, as well as hamper uncontrolled cell growth and cell division, are already in clinical practise, representing 52 kinases at the end of 2019 [4,5].Gefitinib (certain breast,lung cancers)and imatinib(chronic myelogenous leukaemia, CML) are milestones in anticancer therapy that affect the phosphotransfer kinase cascade and target EGFR and hybrid Bcr-Abl tyrosine kinases, respectively. Subsequently, second and third generations of these drugs were introduced to the market to target a range of therapeutically important tyrosine kinases, promote drug selectivity, and circumvent off-target-mediated toxicity and resistance acquired by the original drugs [5,6].As part of the MAPKs oncogenic cascade that triggers cancer cell proliferation,differentiation,division and prevents apoptosis,the RAS-RAF-MEK-ERK1/2 kinase pathway is unusually activated in more than 30% of human cancers. Although RAS mutations represent most of the MAPK variations (22%) [7], it is usually considered undruggable because of an uncommon affinity for its substrate (GTP) and absence of proper pockets for the development of inhibitors [8]. In this context, the lack of effective RAS inhibitors has pointed to downstream kinases, RAF, MEK andERK, as potential therapeutic targets. Therefore, selective RAF (vemurafenib and dabrafenib) [9] and MEK inhibitors (trametinib, cobimetinib and binimetinib)are FDA approved drugs used in combination with otherkinase (BRAF) inhibitors to reduce the toxicities related to BRAF inhibition and delay the development of drug resistance, though disease recrudescence is usually observed within one year of treatment [10,11].
Therefore, extracellular signal-regulated kinases 1 and 2 (ERK1/2) inhibitors have attracted special attention as they have circumvented cases of innate or long-term acquired resistance to RAF and MEK in- hibitors and simultaneously targeted two nodes of the pathway [12]. Several ERK1/2 drug-like small molecules had an IC50 in nM range and entered preclinical and phase I clinical trial, such as ulixertinib (Ki ERK2 0.04 nM) [13], which is now being assessed as a single agent (not in association with BRAF inhibitors) for the treatment of several cancers. (Clinicaltrials.gov identifiers:NCT04145297,NCT03698994, NCT04488003,NCT03454035,NCT03155620, NCT02465060).In addition, MK-8353 was discovered as a potent experimental ERK1/2 inhibitor (IC5020 and 7 nM, respectively) [14], while its close analogue, SCH772984 (IC50 4 nM and 1 nM, respectively) [15], did not provide satisfactory results inpatients with advanced solid tumors despite good tolerance [16]. Alternatively, 4-[7-(1-benzyl-1H-pyrazol-4-yl)-5H-pyr- rolo[2,3-b]pyrazin-2-yl]pyridine, identified through a fragment-based approach, showed a significant ERK2 inhibition (IC50 37 nM) (Fig. 1A) [17]. Due to structure variations, distinct mechanisms of action were observed and, while the majority acts as reversible inhibitors in an ATP- competitive manner, some are covalent or allosteric ERK1/2 inhibitors [18].
Fig.1.(A). ERK inhibitors under pre-clinical and clinical trials,ulixertinib, MK-8353, SCH772984 and 4-[7-(1-benzyl-1H-pyrazol-4-yl)-5H-pyrrolo[2,3-b]pyrazin-2- yl] pyridine. (B). The recently approved 4-anilino-quinazoline drug, dacomitinib, and the 4-amino-pyrazoloquinazoline dihydrogen phosphate prodrug, barasertib, under clinical trial.Amongst kinase inhibitor drugs, it is worth mentioning the fused pyrimidine core related to 4-anilino-quinazoline: gefitinib, erlotinib, afatinib, lapatinib, canertinib, vandetanib and dacomitinib, the latter displaying a Ki value for EGFR of 0.16 ± 0.01 nM [19]. Additionally, the 4-amino-pyrazoloquinazoline dihydrogen phosphate prodrug,bar- asertib, act as selective Aurora kinase B inhibitor (IC50 0.37 nM) [20], and is under clinical trial (Fig. 1B) [21]. Notably, 4-anilino quinazoline pharmacophore is essential for EGFR tyrosine kinase and HER2/Pan- HER kinase inhibition, while several 4-amino quinazoline derivatives have potential to target checkpoint kinases [22], aurora kinases [23,24], Interleukin-1 receptor associated kinase-4 (IRAK-4) [25,26] and others. Beyond anticancer drugs, quinazolines have broad therapeutic activity and efficacy, such as antimalarial, antimicrobial, antidiabetic, antihy- pertensive, sedative and anti-inflammatory [27].
As many kinase inhibitors tend to cross-interact with the different enzymes from the class because of the conservation of the catalytic cleft [28], we hypothesized the 5H-pyrrolo[2,3-b]pyrazin-2-yl] nuclei at the ERK inhibitor 4-[7-(1-benzyl-1H-pyrazol-4-yl)-5H-pyrrolo[2,3-b]pyr- azin-2-yl]pyridine could interact similarly as the 4-amino-quinazoline moiety from the Aurora kinase B inhibitor barasertib (Fig. 2). Based on the bioisostere concept, we have also proposed the replacement of the pyrazole groups at these inhibitors by 1,4-disubstituted 1,2,3-triazoles since these groups are known to provide a fast exploration of chemical space through the generation of libraries of analogues [29].In fact, superposition of the crystallographic structures of both inhibitors (PDB ID 4QPA and 4C2V) showed a good correspondence,as depicted in Fig. S1.In this context, we designed a small library of eight potential ERK1/2 kinase inhibitors conserving the 4-aminoquinazoline pharmacophore bearing a 1,2,3-triazole stable core to bridge different aromatic and heterocyclic rings using copper-catalysed azide-alkyne cycloaddition reaction (CuAAC) as a Click Chemistry strategy. In addition, four 4-ami- noquinazolines previously prepared by our group were also tested [30,31]. The ability of these compounds to affect the cancer cell growth and division was assessed on oral squamous cell carcinoma (CAL-27 and HN13), gastric (HGC-27) and breast (BT-20) cancer cells. The most active compound in BT-20 cells was subjected to further investigations in order to get some insights about its mode of action. Interestingly, this selected compound (25a) induced cell death and cell cycle arrest with an increase in the subpopulation of cells in G0/G1 phase, as well as syn- chronized decrease in the S phase, which was associated with a remarkable decrease of p-ERK1/2 and p-STAT3, as revealed by Western blot analysis. Further molecular modelling studies confirmed the po- tential interactions of compound 25a toward ATP site of p-ERK1/2 kinase.
2.Results and discussion
2.1.Synthesis
Initially, the conventional Niementoski cyclocondensation reaction between anthranilic acid derivative 1 (R1, R2 = OMe) and formamide under microwave heating was pursued to obtain quinazolin-4(3H)-ones 2b in 70% yields, while the non-substituted 2a was acquired from commercial source [32]. Although intermediate 2b was obtained in reasonable yield, the one-pot intermolecular reductive N-hetero- cyclisation of the corresponding methyl 2-nitrobenzoate ester of 3 with formamide was also tested with indium(III) chloride. However, the use of this Lewis acid at high temperature to induce formamide decompo- sition and form carbon monoxide, for in situ reductive conversion of 2- nitrobenzoate into 2-aminobenzoate derivative, gave the quinazolinone 2b in low yields (35%) (Scheme 1) [33].Intermediate 2c, bearing a benzyloxy group installed at C-6 position of quinazoline core, was obtained through regioselective 5-methoxyl cleavage of 4,5-dimethoxy-2-nitrobenzoic acid (3), taking the advan- tage of the electron withdrawn effect of the p-nitro group, which weakens the Ar-OMe bond and makes it more susceptible to nucleophilic attack (OH− ), in a mechanism similar to SN2Ar [34]. Under strong basic conditions and at high temperature, the structure of the 5-hydroxy-4- methoxy-2-nitrobenzoic acid intermediate (sole isomer obtained in 91% yield) was ascribed based on the NOE-diff autophagosome biogenesis experiments, which showed the selective removal of the methyl group at C-5 position based on the presence of NOE only between H-3 and 4-OMe, Fig. S2 [35,36]. This selective cleavage in the early stages of the synthesis proved to be better than the deprotection of the 6,7-dimethoxy-quinazolin-4(3H)-one (2a) intermediate using methionine, which resulted in an inseparable mixture of both regioisomers on a chromatographic column [37]. After esterification,methyl 5-hydroxy-4-methoxy-nitrobenzoate(4)
Fig. 2. Design of potential 4-amino-quinazoline-based ERK inhibitors based on the structures of the Aurora kinase B inhibitor barasertiband the ERK2 inhibitor 4-[7- (1-benzyl-1H-pyrazol-4-yl)-5H-pyrrolo[2,3-b]pyrazin-2-yl]pyridine.
Scheme 1. General procedures for the synthesis of the target compounds 21-24a,b, 25-27a and 22c,d. Reagents and conditions: i. formamide, 60 W, 150 ◦ C, 40 min; ii. a) MeOH, H2SO4, reflux, 48 h; b) formamide, InCl3, 150 ◦ C; iii. a) NaOH (6 mol.L− 1), 100 ◦ C, 3 h, b) MeOH, H2SO4, reflux, 48 h; iv. BnBr, Cs2CO3, DMF, 24 h; v. formamide, InCl3, 150 ◦ C; vi. PyBOP, DBU, propargylamine, DMF; vii. azide derivatives 7-13 (R3CH2-N3), sodium ascorbate, CuSO4, DMF or 1,10-phenantroline, sodium ascorbate, CuSO4, EtOH:H2O [2:1 (v/v)]; viii. TFA, reflux.
Scheme 2. Synthesis of azides 11-13. Reagents and conditions: (A). i. a) 1-(azidosulfonyl)-1H-imidazol-3-ium hydrogensulfate, CuSO4, NaHCO3, MeOH, H2O; b) BnCl, K2CO3, acetone. (B). i. ethanolamine, triethylamine, 120 ◦ C, 2 h; ii. a) SOCl2, DMF, r.t.; b) NaN3, DMF. (C). i. Boc2O, DCM, 0 ◦ C, r.t.; ii. BnBr, K2CO3, DCM, reflux; iii. a) HCl in dioxane; b) chloroacetyl chloride, DCM, 0 ◦ C, r.t.; iv. NaN3, DMF, 70 ◦ C, 150 W, 15 min.readily protected with benzyl bromide to give 5, which was subjected to one-pot cyclization using the aforementioned formamide and indium chloride to generate the corresponding quinazolinone 2c (Scheme 1).Thus, nucleophilic aromatic substitution (SNAr) was carried out using mild phosphonium-promoted reaction to introduce propargyl- amine group in quinazolinones(2a-c)to give the corresponding N- (prop-2-yn-1-yl)quinazolin-4-amine derivatives 6a (80%), 6b (55%) and 6c (57%) [38].For the cycloaddition reaction with N-(prop-2-yn-1-yl)quinazolin-4- amines 6a-c, azides 7– 13 (R3CH2-N3) were prepared from the corre- spondinghalide derivatives using sodium azide in DMF. The azides 7–10 (SI) were synthesised from commercial halides [39–42], whilst the azides 11– 13 were obtained according to Scheme 2. Thus, the treatment of 4-(aminomethyl)piperidine (14) with 1-(azidosulfonyl)-1H-imidazol- 3-ium hydrogensulfate azide, followed by benzylation of the secondary amine with BnCl and K2CO3 in acetone, afforded azide 11 (Scheme 2A) [30]. On the other hand, addition of ethanolamine at C-4 position of commercial 4,7-dichloroquinoline (15) gave intermediate 16,which was converted to azido 12 under treatment with thionyl chloride, fol- lowed by NaN3 in DMF (Scheme 2B) [43,44]. The route to obtain azide 13 started with mono protection of the commercial piperazine (17) with di-tert-butyl dicarbonate to produce compound 18 with 78% yield [45,46], followed by treatment with BnBr and K2CO3 in DCM under reflux for benzylation of the free secondary amine to obtain compound 19 in reasonable yield (49%) [47]. Removal of the Boc under acidic condition (2.0 mol.L− 1 HCl in dioxane) and subsequent reaction with chloroacetyl chloride gave compound 20 in excellent yield (86%) [48]. Finally, azide 13 was obtained with 84% yield by SN2 reaction in the presence of NaN3 in DMF under microwave irradiation (Scheme 2C).
As for the final step, CuAAC reaction better results were achieved using microwave heating at 100 W, 80 ◦ C for 5 to 10 min with azide precursors 7–13 and N-(prop-2-yn-1-yl)quinazolin-4-amines 6a-c (10% excess) in the presence of sodium ascorbate and CuSO4 in DMF [29,42] to give the final products; except compound 22c, obtained in the pres- ence of 1,10-phenantroline, sodium ascorbate, CuSO4 and EtOH:H2O [2:1 (v/v) at room temperature]. After purification, the small library of 1,2,3-triazole-quinazoline derivatives 21-24a,b, 25-27a and 22c was obtained as the unique 1,4-disubstituted regioisomer from moderate to good yields (42–72%) and characterized by NMR, which revealed the triazole 1H and the corresponding 13C between 7.55–8.15 ppm (H-12) and 122.9– 125.3 ppm (C-12), respectively. The HRMS confirmed the target structures. Compound 22c was further deprotected to afford 22d (55%) to be also biologically assessed.
2.2.Biological evaluation
The antitumoral activity of the products 21-24a,b, 22d and 25-27a was evaluated in cancer cell lineages derived from oral squamous cell carcinoma (CAL-27 and HN13), gastric (HGC-27) and breast (BT-20) cancers. The cytotoxicity assay by resazurin was performed at a fixed concentration (50 μM) (Fig. 3).From the 4-aminoquinazoline series, compounds 21a, 22a, 23b, 24b and 25a displayed the most promising antitumoral activities toward BT- 20 cells. Thus, the five selected compounds were tested at several con- centrations against BT-20 cells and normal fibroblasts (OHMF). The IC50 value of compound 25a at 48 h was 24.6 μM in BT-20 cells and 80 μM in OHMF cells, indicating tumor selectivity over 3-fold (Table 1). In this regard, only compound 25a displayed a selectivity index (SI) above the minimum threshold (SI ≥ 3) making it a potential compound for further optimization (Table 1). [49].For the sake of comparison, the reported IC50 value for the ERK1/2 inhibitor ulixertinib (please refer to the molecular docking section) against BT-20 cells was 5.7 μM [13]. Considering the interlaboratory variability observed for cell-based cytotoxicity assays [50], we therefore assumed 25a as a viable structure for further investigation since it performed similarly as this ERK inhibitor, currently under clinical trials.
We then analysed the release of lactate dehydrogenase (LDH) in cell culture media to understand the cytotoxic effect of the compound 25a in BT-20 cells. The cells exposed to compound 25a (40 and 80 μM) showed high level of LDH release compared to the negative control (0.01% DMSO), suggesting the induction of cell death by necrosis (Fig. 4A). However, we decided to investigate other mechanisms besides cell death under the action of 25a, such as cell cycle arrest. The rationale was that the same concentrations used for cytotoxicity and LDH assays did not cause the same intensity of cellular effects.Given that cell cycle arrest can accompany alterations in the viability assay, we determined the distribution of the BT-20 cells in each cell cycle phase under 25a exposition. The results in Fig. 4B-C showed that 25a dose-dependently induced G0/G1 stalling in BT-20 cells. The in- crease in the subpopulation of cells in G0/G1 was accompanied by a synchronized decrease in the S phase. This has been shown with other ERK inhibitors in melanoma, leukaemia and other human cancers [51–53].Following findings that 25a impacts viability and cell cycle pro- gression in BT20 cells, we analysed proteins related to tumour cell sig- nalling by Western blot (Fig. 4D-E). The results indicated the absence of cleaved caspase-3 and PARP protein and reduced p53 levels, suggesting the absence of apoptosis and p53 activation in BT-20 cells incubated
Fig. 3. Inhibition of cell viability by compounds 21-24a,b, 22d and 25-27a was evaluated by theresazurin assay. The cancer cells CAL-27, HN13, HGC-27 and BT-20 were treated with the compounds and cisplatin at 50 μM for 48 h. Two independent biological experiments were performed in quadruplicate. The values are expressed as mean ± standard deviation of % cell viability inhibition.*SI = IC50_ OHMF/IC50_BT-20. IC = Inhibitory concentration; IC50 = concentration that reduces cell viability in 50%. Each experiment was repeated at least three times in quadruplicate, and the percentage of inhibition of cell viability was used to calculate IC50 values by nonlinear regression with 25a. On the other hand, our results confirmed that p-ERK1/2 and p- STAT3 were drastically decreased in BT-20 cells under 25a incubation. Sustained activation of ERK1/2 is essential for G1 to S phase progres- sion, and it can be considered one of the master regulators to cell cycle phase transition [54,55]. Activation of STAT3 is also associated with cell cycle and tumour progression, including in breast cancer cells [56,57]. These results indicated that inactivation of ERK1/2 and STAT3 pathway signalling are mechanisms underlying 25a-induced cell cycle arrest (20 and 40 μM) accompanied by loss of proliferation and viability in the breast cancer cell.
2.3.Molecular docking
To assess whether 25a could inhibit the ERK enzyme, we outlined molecular docking simulations with the crystallographic structure of this conserved serine/threonine kinase, which can be activated by phosphorylation on both the threonine/tyrosine residues of a conserved motif (Thr-Glu-Tyr).Triggered by the release of growth factors, its activation is particularly essential for cell cycle progression at the G1 phase,in which persistent and elevated ERK activities allow cells to entry the S phase [58].
As more than one-third of the clinically relevant tumors display constitutively activated ERK enzymes [59], MEK/ERK pathway in- hibitors have previously been designed [11]. An important ATP- competitive ERK1/2 inhibitor (ulixertinib – BVD-523) has recently gained attention due to its ability of inducing, alone, tumor regression in xenograft models that are cross-resistant to both BRAF and MEK in- hibitors [11]. Based on the ERK inhibition profile suggested for 25a in the biological assays, we performed molecular docking simulations to compare its most probable binding mode with the corresponding mode predicted for ulixertinib.The kinasesite of serine/threonine kinases lies at the junction of the N- and C-domains, where the adenosine triphosphate (ATP) binds to donate phosphate groups, providing the energy necessary in the re- actions catalyzed by these enzymes. Selective kinase inhibitors can be generally categorized as competitive ATP inhibitors, which can bind to the active and/or inactive kinase forms, or allosteric modulators, where the allosteric site can be contiguous or not to the ATP binding cleft [60]. As suggested by the docking simulations, both ulixertinib (Fig. 5A) and 25a (Fig. 5B) compete with ATP for binding to the ERK with a slightly higher affinity observed for 25a than for the reference ERK1/2 inhibitor: ΔG binding = − 6.6 and − 5.6 kcal/mol, respectively.
Interestingly, 25a seems to interact with the residues from the ATP binding cleft more efficiently than ulixertinib, with an almost perfect superposition between its terminal phenyl ring with the heteroaromatic moiety of ATP (Fig. 5). In fact, when one examines more closely the predicted interactions performed by both compounds (Fig. 6), the hy- drophobic interactions at the terminal phenyl ring of 25a are reinforced by H bonds with the contiguous sulfonyl group. This seems coherent with the results from the biological assays that suggests 25a is the most promising compound we have synthesized, probably due to this chem- ical feature. Furthermore, despite the similar positioning between the pyrrole of ulixertinib (Fig. 6A) and the triazole of 25a (Fig. 6B), which indicates equivalent cation-π (Lys52) and π-π interaction (Tyr34) for both groups, 25a would stand out because of an additional π-π interaction of the benzyl group. This connects the triazole andarylsulfonyl group with (Tyr34) and potentially contributes to a stronger interaction profile for the central moiety of this compound. Finally, despite the similar hy- drophobic interactions performed with Tyr34 and Arg65 in both com- pounds, neither the amide nor the hydroxyl group of the ulixertinib seems to be involved in H-bonding, as observed for the 4-quinazolin- amine moiety of 25a. Once again, the compound we have synthesized seems to explore more efficiently the ERK amino acid residues that are contiguous to the ATP binding cleft than ulixertinib. Accordingly, we decided to compare their potential toxicities and pharmacokinetic pro- files (Table 2) to check if 25a could be (or not) considered a promising lead compound using MetaDrug module of MetaCore platforms from Clarivate Analytics.Except for the predictions of CYP2C9 and CYP3A4 inhibition by ulixertinib, both compounds did not achieve high similarity levels (Tanimoto Prioritization – TP values) with the compounds in the training sets used to compose the ADMETox QSAR models and, accordingly, the results in Table 2 must be viewed with some caution. In this context, both compounds did not display any relevant toxicity, a fact that agrees with the Phase I dose-escalation study published for ulixertinib, as the observed significant side effects were diarrhea, fatigue, nausea, and dermatitis [11].
Fig. 4. Effect of compound 25a on LDH release, cell cycle progression and tumor signaling proteins. (A) LDH activity assay; BT-20 cells were treated with 25a compound at the indicated concentrations for 48 h, and the activity of LDH in the culture medium was measured spectrophotometrically. (B) Cell cycle distribution analysis; the cell population histograms in respect to propidium iodide (PI) fluorescence intensity determined by flow cytometry are representative from 3 inde- pendent experiments performed in triplicate. BT-20 cells treated with 25a at the indicated concentrations for 24 h were collected, fixed in 70% cold ethanol, stained with PI, and analysed by flow cytometry. The percentage of cells in different phases of cell cycle are showed for each condition. (C) Cell cycle distribution analysis; the graph represents the percentages (mean ± SD) of cells distributed in each cell cycle phase of 3 independent experiments performed in triplicate by flow cytometry. (D) Signaling proteins; Total protein from BT-20 cells treated with 25a for 48 h were extracted and subjected to Western blot analysis of the levels of p- STAT3, p-ERK, PARP, p53 and cleaved caspase-3. β-actin was used as a loading control. (E) Protein quantification; the graph represents densitometry of the protein bands of 2 independent experiments. Relative intensity of each target protein band was divided by intensity of the β-actin band, and then the values were compared to the control (0 µM).
Fig.5. Proposed binding modes for ulixertinib (A, in yellow) and 25a (B, in cyan) at the catalytic cleft of ERK (PDB ID 6DCG). The crystallographic pose of the adenosine triphosphate (in green, recovered from the ERK complex PDB ID 4GT3) was used as a reference for the docked ligands in both panels. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Concerning the Cytochrome P450 metabolic stability, ulixertinib displays a slightly superior profile because of the expected stability against the 2D6 isoform and, accordingly, 25a could display relevant drug-drug interactions with known CYP2D6 inhibitors such as bupro- pion, fluoxetine, paroxetine and terbinafine [62]. As depicted in Fig. 7, the first pass metabolism of 25a is restricted to Phase 1 aromatic hy- droxylation reactions and, particularly, to Phase 2 N-conjugations with glucuronide or sulfate. The metabolites of ulixertinib were recently described and involve both N-dealkylation or hydroxylation of the ter- minal carbon of the isopropyl side chain to subsequently yield a car- boxylic acid, the direct conjugation with glucuronide at the free hydroxyl group (both for the parent drug or theN-dealkylated form) and the aromatic hydroxylation at different positions of the phenyl ring [63]. Unlike ulixertinib, 25a is not likely to cross the blood brain barrier,
which hampers the treatment of CNS tumors. However, as both ulix- ertinib and 25a display adequate LogP values, the most interesting optimization achieved for the compound we synthesized is the water solubility. Despite it being below the ideal range (2 < LogWsol_25。C < 4 mg/L) [64], its predicted water solubility is more satisfactory than the one for ulixertinib.
3.Conclusion
We have synthesised a small library of triazolestethered to 4-amino- quinazoline using a Click Chemistry strategy. Initial screening on tumour cells enabled us to identify compound 25a (IC5024.6 μM) as the most potent from the series towards BT-20 cells. Despite the promotion of lactate dehydrogenase (LDH) release linking the mode of action
Fig. 6. Two-dimensional diagrams for the docked ligand–ERK interactions generated from 3-D models using PoseView [61]. (A) Ulixertinib and (B) 25a. Dashed black lines correspond to hydrogen bonds, green dashed lines to π-π or cation-π interactions and the solid green to hydrophobic interactions. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)necrosis, compound 25a was able to induce, in a dose-dependent manner, an increase in the subpopulation of cells in G0/G1 in parallel to the decrease in S phase of BT-20 cells. Taking into account the essential role ofp-ERK1/2 and p-STA3 in the positive regulation of G1 to S phase progression, we incubated BT-20 cells with 25a and analysed proteins related to tumour cell signalling by Western blot assays. The significant p-ERK1/2 and p-STA3 kinases decrease led us to infer their possible inhibition as the main mechanism underlying the 25a-induced cell cycle arrest. Inspired by the importance of p-ERK1/2 as therapeutic target for breast cancer treatment, we performed docking studies to predict the binding mode of compound 25a at ATP binding cleft and concluded that it could bind more efficiently than ulixertinib, a drug
Fig. 7. Main predicted sites for first pass Phase 1 and 2 metabolic reactions in 25a.
under clinical evaluation. In silico pharmacokinetics data strengthens compound 25a as an interesting pharmacological probe for further investigations.
4.Experimental section
4.1.Chemistry
Reagents and solvents were commercially obtained as reagent grade and used without any purification (Merck。and TEDIA。respectively). The reaction progress and separation were monitored by thin layer chromatography and the spots were visualized using ultraviolet light (UV − 254 nm). Flash chromatography equipment Biotage Isolera was used for purification using normal phase cartridges. NMR analysis were obtained on a Bruker Advance spectrometer at 300, 400 or 500 MHz. A Bruker Daltonics micrOTOF-Q II™ ESI-Qq-TOF mass spectrometer was used to perform High Resolution Mass Spectroscopy (HRMS) analysis.Compounds 24a,b,26a and 27a were obtained according to described procedures [30,31] and the novel compounds were obtained via Copper-catalysed Azide-Alkyne Cycloaddition (CuAAC) reactions [29,42]. Non-commercial starting materials, obtained from previously described protocols, are briefly reported in the supplementary information.General procedure A – To a microwave flask containing a magnetic stirring bar, sodium ascorbate (0.2 eq.) diluted with DMF (100 μL) was added followed by CuSO4 (0.05 eq., 1 mol.L− 1 in water). The reaction mixture was stirred for 5 min and then the alkyne derivative (6a-c, 1.1 eq.) and azido derivative (7– 13, 1 eq.), both diluted in DMF, were added. The reaction mixture was microwaved with 100 W, 80 ◦ C for 5–10 min. The crude reaction was poured in EtOAc and extracted with brine. The organic phase was concentrated and purified by flash chro- matography (EtOAc/MeOH 5%).
4.1.1. N-[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]quinazolin-4-amine(21a)
Yield: 67% (53 mg, 0.17 mmol). 1H NMR (300 MHz, DMSOd6) δH (ppm): 8.77 (1H, t, J 5.6 Hz, H-9), 8.44 (1H, s, H-7), 8.20 (1H, d, J 7.8 Hz, H-3), 8.02 (1H, s, H-12), 7.79–7.60 (2H,m,H-2, H-6), 7.47 (1H,ddd, J 8.2 Hz, J 6.9 Hz, J 1.3 Hz, H-1), 7.36–7.21 (5H, m, H-Ar), 5.49 (2H, s, H-13), 4.75 (2H, d, J 5.7 Hz, H-10). 13C NMR (75 MHz, DMSOd6) δC (ppm): 159.2 (C-8), 155.0 (C-7), 149.1 (C-4), 145.2 (C-11), 136.2 (C-14), 132.7 (C-2), 128.7 (C-17, C-18), 128.1 (C-19), 128.0 (C-15, C-16), 127.5 (C-6), 125.8 (C-1), 123.3 (C-12), 122.7 (C-3), 115.0 (C-5), 52.7 (C-13),36.0 (C-10). HRMS(ES+):m/z [M+H]+ calculated for C18H17N6(+): 317.1509, found: 317.1508.
4.1.2. N-[1-(4-nitrobenzyl)-1H-1,2,3-triazol-4-yl] methyl quinazolin-4- amine (22a)
Yield: 42% (38 mg, 0.10 mmol). 1H NMR (500 MHz, DMSOd6) δH (ppm): 8.84 (1H, t, H-9), 8.50 (1H, s, H-7), 8.25 (1H, d, J 8.3 Hz, H-3), 8.21 (2H,d, J 8.7 Hz, H-17, H-18), 8.15 (1H, s, H-12), 7.77 (1H, t, J 7.6 Hz, H-2), 7.70 (1H, d, J 8.3 Hz, H-6), 7.54–7.48 (3H, m, H-1, H-15, H- 16), 5.73 (2H, s, H-13), 4.82 (4H,d, J 5.6 Hz, H-10). 13C NMR (126 MHz, DMSOd6) δC (ppm): 159.1 (C-8), 154.9 (C-7), 149.0 (C-4), 147.2 (C-19), 145.3 (C-11), 143.4 (C-14), 132.5 (C-2), 128.9 (C-15,C-16), 127.4 (C-6), 125.6 (C-1), 123.7 (C-17, C-18), 123.6 (C-12), 122.6, (C-3), 114.9 (C-5), 51.7 (C-13), 35.9 (C-10). HRMS (ES+): m/z [M+H]+ calculated forC18H16N7O2(+): 362.1360, found: 362.1360.
4.1.3.N-[1-(4-methoxybenzyl)-1H-1,2,3-triazol-4-yl]methylquinazolin- 4-amine (23a)
Yield: 53% (17 mg, 0.049 mmol). 1H NMR (400 MHz, CDCl3) δH (ppm): 8.65 (1H, s, H-7), 7.86–7.76 (2H,m,H-3, H-6), 7.72 (1H, t, J 8.3 Hz, H-2), 7.55 (1H, s, H-12), 7.43 (1H, t, J 8.2 Hz, H-1), 7.27–7.21 (2H, m, H-15, H-16), 6.97 (1H, sl, H-9), 6.91–6.87 (2H, m, H-17, H-18), 5.45 (2H, s, H-13), 4.90 (2H,d, J 5.0 Hz, H-10), 3.80 (3H, s, H-20). 13C NMR (101 MHz, CDCl3) δC (ppm): 160.2 (C-8), 159.3 (C-19), 155.0 (C-7), 149.1 (C-4), 144.9 (C-11), 132.9 (C-2), 129.8 (C-15, C-16), 128.3 (C-3), 126.4 (C-1), 126.5 (C-14), 122.3 (C-12), 121.1 (C-6), 115.1 (C-5), 114.7 (C-17, C-18), 55.5 (C-20), 54.0 (C-13), 36.6 (C-10). HRMS (ES+): m/z [M+H]+ calculated for C19H19N6O+: 347.1615, found: 347.1616.
4.1.4.N-[1-(2-((phenylsulfonyl)methyl]benzyl-[(1H-1,2,3-triazol-4-yl) methyl] quinazolin-4-amine (25a)
Yield: 65% (16 mg, 0.034 mmol). 1H NMR (500 MHz, DMSOd6) δH (ppm): 8.81 (1H, t, J 5.6 Hz, H-9), 8.47 (1H, s, H-7), 8.22 (1H, d, J 8.2 Hz, H-3), 8.00 (1H, s, H-12), 7.79 (2H, d, J 8.1 Hz, H-22, H-26), 7.75 (2H, t, J 7.5 Hz, H-23, H-25), 7.69 (1H,d, J 8.3 Hz, H-6), 7.62 (2H, t, J 7.8 Hz, H-2, H-24), 7.50 (1H, t, J 7.6 Hz, H-1), 7.31 (1H, t, J 7.5 Hz, H- 19), 7.23 (1H,d, J 7.6 Hz, H-17), 7.08 (2H,t, J 7.3 Hz, H-16, H-18), 5.60 (2H, s, H-13), 4.91 (2H, s, H-20), 4.78 (2H,d, J 5.6 Hz, H-10). 13C NMR (126 MHz; DMSOd6) δC (ppm): 159.1 (C-8), 154.9 (C-7), 149.0 (C-4), 145.3 (C-11), 138.4 (C-15), 136.5 (C-14), 134.1 (C-24), 133.0 (C-16), 132.6 (C-2), 129.3 (C-23, C-25), 129.2 (C-17), 128.8 (C-18), 128.1 (C- 22, C-26), 128.0 (C-19), 127.4 (C-6), 126.6 (C-21), 125.7 (C-1), 123.4 (C-12), 122.7 (C-3), 114.9 (C-5), 57.8 (C-20), 49.9 (C-13), 35.9 (C-10). HRMS (ES+): m/z [M+H]+ calculated for C25H23N6O2S+: 471.1598,found: 471.1599.
4.1.5. N-[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]-6,7-dimethoxyquinazolin-4-amine (21b)
Yield: 72% (51 mg, 0.13 mmol). 1H NMR (300 MHz, CDCl3) δH (ppm): 8.53 (1H, s, H-7), 7.65 (1H, s, H-12), 7.38 (3H, dd, J 5.0 Hz, J 1.8 Hz, H-Ar), 7.28 (2H,d, J 7.7 Hz, H-Ar), 7.15 (1H, s, H-3), 7.03 (2H, s, H- 6, H-9), 5.50 (2H, s, H-13), 4.89 (2H,d, J 5.4 Hz, H-10), 3.99 (3H, s, H- 20), 3.79 (3H, s, H-21).13C NMR (75 MHz, CDCl3) δC (ppm): 158.1 (C-2), 154.4 (C-8), 153.8 (C-7), 149.0 (C-1), 146.5 (C-4), 134.5 (C-14), 129.3 (C-15, C-16), 129.0 (C-19), 128.2 (C-17, C-18), 122.9 (C-12), 107.5 (C- 3), 100.1 (C-6), 56.3 (C-21), 56.2 (C-20), 54.5 (C-13), 36.3(C-10).HRMS (ES+): m/z [M+H]+calculated for C20H21N6O2(+): 377.1721,found: 377.1719.
4.1.6. 6,7-dimethoxy-N-[1-(4-nitrobenzyl)-1H-1,2,3-triazol-4-yl]methyl quinazolin-4-amine (22b)
Yield: 45% (36 mg, 0.08 mmol). 1H NMR (300 MHz, DMSOd6) δH (ppm): 8.49 (2H, t, J 5.4 Hz, H-7, H-9), 8.22 (2H, d, J 8.5 Hz, H-17, H- 18), 8.14 (1H, s, H-12), 7.63 (1H, s, H-6), 7.51 (2H,d, J 8.6 Hz, H-17, H- 18), 7.12 (1H, s, H-3), 5.73 (2H, s, H-13), 4.79 (2H, d, J 5.4 Hz. H-10), 3.89 (3H, s, H-20), 3.86 (3H, s, H-21).13C NMR (126 MHz, DMSOd6) δC (ppm): 157.9 (C-2), 153.8 (C-8), 153.2 (C-7), 148.3 (C-1), 147.2 (C-19), 145.6 (C-14), BAY 11-7082 concentration 143.4 (C-4), 129.0 (C-17, C-18), 123.8 (C-15, C-16), 123.6 (C-12), 107.0 (C-3), 102.0 (C-6), 55.9 (C-20), 55.6 (C-21), 51.7 (C-13),35.7 (C-10). HRMS (ES+): m/z [M+H]+ calculated for C20H20N7O4(+):422.1577, found: 422.1574.
4.1.7.6-(benzyloxy)-7-methoxy-N-[1-(4-nitrobenzyl)-1H-1,2,3-triazol- 4-yl]methyl quinazolin-4-amine (22c)
A mixture of CuSO4(0.45 mg, 0.003 mmol, 5 mol%), 1,10-phenan- troline monohydrate (0.5 mg, 0.003 mmol, 5 mol%), and sodium ascorbate(11 mg,0.06 mmol) in EtOH:H2O [2:1 (v/v), 1 mL], was stirred for 5 min at room temperature. Subsequently the compounds 4- amino-[N- (propynyl)]-6-benzyloxy-7-methoxy-quinazoline (6c) (19.7 mg, 0.06 mmol) and 1-(azidomethyl)-4-nitro-benzene (8) (10 mg, 0.06 mmol) diluted in EtOH:H2O [2:1 (v/v); 1 mL] were added to the reaction mixture, which was stirred for 18 h at room temperature. After, the reaction mixture was concentrated in vacuo and purified by silica gel column chromatography [EtOAc: MeOH (9.5:0.5)]. The compound 22c was obtained in quantitative yield (28 mg, 0.06 mmol).1H NMR (300 MHz, DMSOd6) δH (ppm): 。8.49 (1H,d, J 5.5 Hz, H-9), 8.39 (1H, s, H-7), 8.22 (2H,d, J 8.8 Hz, H-17, H-18), 8.15 (1H, s, H-12), 7.81 (1H, s, H-6), 7.55–7.47 (4H, m, H-15, H-16, H-22, H-26), 7.46–7.32 (3H, m, H-23, H- 24, H-25), 7.13 (1H, s, H-3), 5.74 (2H, s, H-13), 5.13 (2H, s, H-21), 4.81 (2H,d, J 5.4 Hz, H-10), 3.89 (3H, s, H-20).13C NMR (75 MHz, DMSOd6) δC (ppm): 158.1 (C-8), 154.0 (C-2), 153.5 (C-7), 147.4 (C-1), 147.2 (C- 4), 146.1 (C-19), 145.6 (C-14), 143.6 (C-27), 136.4 (C-11), 129.1 (C-15, C-16), 128.6 (C-23, C-25), 128.3 (C-22, C-26), 128.2 (C-24), 123.9 (C- 17, C-18), 123.8 (C-12), 108.5 (C-5), 107.1 (C-3), 103.3 (C-6), 70.4 (C- 21), 55.8 (C-20), 51.8 (C-13), 35.8 (C-10).HRMS (ES+): m/z [M+H]+calculated for C26H24N7O4(+) 498.1884, found: 498.1881.
4.1.8.7-methoxy-4-[1-(4-nitrobenzyl)-1H-1,2,3-triazol-4-yl] methylaminoquinazolin-6-ol (22d)
A solution of compound 22c (10 mg, 0.02 mmol) in TFA (110 μL) was refluxed for 50 min. After cooling to room temperature the mixture was poured into ice. The solvent was removed in vacuo and the crude product was purified by filtration over Discovery。DSC-18 SPE cartridge (H2O:MeOH). Yield: 55% (4.5 mg, 0.01 mmol). 1H NMR (300 MHz, CD3OD) δH (ppm): 8.64 (1H, s, H-7), 8.23 (2H,d, J 8.8 Hz, H-17, H-18), 8.12 (1H, s, H-12), 7.59 (1H, s, H-3), 7.51 (2H,d, J 8.8 Hz, H-15, Advanced medical care H-16), 7.16 (1H, s, H-6), 5.74 (2H, s, H-13), 5.04 (2H, s, H-10), 4.08 (3H, s, H- 21). 13C NMR (101 MHz, CD3OD) δC (ppm): 161.2 (C-8), 157.6 (C-2), 150.4 (C-1), 149.4 (C-7), 149.3 (C-4), 143.9 (C-19), 134.9 (C-11), 130.0 (C-15, C-16), 125.3 (C-12), 125.0 (C-17, C-18), 109.0 (C-5), 107.3 (C-3),
100.3 (C-6), 57.3 (C-21), 53.9 (C-13), 37.9 (C-10). HRMS (ES+): m/z [M+H]+ calculated for C19H18N7O4(+) 408.1415, found: 408.1412.
4.1.9.6,7-dimethoxy-N-[1-(4-methoxybenzyl)-1H-1,2,3-triazol-4-yl]methylquinazolin-4-amine (23b)
Yield: 55% (42 mg, 0.10 mmol). 1H NMR (300 MHz, DMSOd6) δH (ppm): 9.93 (1H, t, J 4.4 Hz, H-9), 8.59 (1H, s, H-7), 7.88 (1H, s, H-12), 7.60 (1H, s, H-6), 7.03 (2H, d, J 8.6 Hz, H-15, H-16), 6.97 (1H, s, H-3), 6.65 (2H,d, J 8.6 Hz, H-17, H-18), 5.22 (2H, s, H-13), 4.69 (2H,d, J 5.4 Hz, H-10), 3.70 (3H, s, H-21), 3.64 (3H, s, H-22), 3.46 (3H, s, H-20). 13C NMR (101 MHz, DMSOd6) δC (ppm): 159.2 (C-2), 156.1 (C-19), 150.1 (C- 8), 149.1 (C-7), 143.3 (C-1), 134.3 (C-11), 129.8 (C-15, C-16), 127.9 (C- 14), 123.4 (C-12), 114.2 (C-17, C-18), 106.6 (C-5), 103.3 (C-3), 99.6 (C- 6), 56.5 (C-21, C-22), 55.2 (C-20), 52.5 (C-13), 36.8 (C-10). HRMS(ES+): m/z [M+H]+ calculated for C21H23N6O3(+): 407.1826, found: 407.1825.
4.2.Cell culture
The breast carcinoma cell line BT-20 (ATCC。 HTB-19™), tongue squamous cell carcinoma HN13 [65] and Cal-27 (ATCC CRL-2095), gastric carcinoma HGC-27 (BCRJ), and normal fibroblast [66] were grown in Dulbecco ’s Modified Eagle Medium (DMEM, Sigma, St. Louis, MO, USA) supplemented with 10% fetal bovine serum, 100 U/mL penicillin and 100 μg/mL streptomycin. The cultures were maintained in a humidified 5% CO2 atmosphere at 37 ◦ C.
4.3.Cell viability
The resazurin assay was used to determine cell viability. The cells were seeded on 96-well plates (4,000 cells/well) and treated with compounds at different concentrations for 48 h. After treatment, the medium was replaced by DMEM phenol red-free (Sigma-Aldrich) con- tainingresazurin (0.01 mg.mL− 1 dissolved in PBS; Sigma-Aldrich), and the plates incubated in the dark for 4 h at 37 ◦ C. Finally, cell viability was measured by detection of fluorescence in a microplate fluorimeter (excitation 530/25 nm, emission 590/35 nm). Three independent bio- logical experiments were performed in quadruplicate. The data were plotted and the IC50 value was estimated by nonlinear regression. Cell viability was also assessed using the lactate dehydrogenase (LDH) assay (Sigma-Aldrich) to measure the activity of LDH released in the cell culture medium as an indicator of necrosis. The cells were seeded on 24- well plates (50,000 cells/well). After 24 h, cells were treated with compounds at different concentrations in DMEM phenol red-free at 37 ◦ C in 5% CO2 for 48 h. Next, the supernatant was transferred to a new plate and the LDH reaction mix added. After incubation for 30 min at room temp, the LDH activity was quantified by a microplate reader at OD450nm.
4.4. Cell cycle analysis
BT-20 cells were seeded on 6-well plates (3 × 105 cells/well) and then incubated with 25a for 24 h. Next, cells were collected and fixed in 70% cold ethanol for 1 h, treated with RNase (100 μg.mL− 1) for 30 min at 37 ◦ C and stained with propidium iodide (50 μg.mL− 1) for 15 min at 37 ◦ C. Next, cell subpopulations in each cell cycle phase were deter- mined by using FACSCalibur flow cytometry (BD Biosciences – San Jose, CA, USA). The cell cycle distribution was analysed using ModFit LT V3- 0.3.11 software.
4.5.Western blot
BT-20 cells were cultured in 6-well plates (3 × 105 cells/well), fol- lowed by incubation with the compound 25a (20 and 40 µM) in serum- free medium for 48 h. Next, cells were harvested in CelLytic MT buffer (Sigma-Aldrich)supplemented with phosphatases and proteases inhibitors (Sigma-Aldrich), subjected to sonication (three times at 5 s each) and the cell lysate was centrifuged at 13,000g for 15 min at 4 ◦ C. The supernatant was collected, and protein concentration was deter- mined by using Bradford protein assay (BioRad). Total proteins (30 µg) were separated in 10% SDS-PAGE gel and transferred onto nitrocellulose membranes.The protein marker ladder(Spectra multicolour broad range protein ladder,Thermo Scientific) was included in all gels. Membranes were incubated overnight at 4 ◦ C with primary antibodies for p-STAT3 (#9145), p-ERK1/2 (#4377), p53 (#2527), PARP (#9532) and cleaved caspase-3(#9661)from Cell Signalling,and β-Actin (sc47778) from Santa Cruz Biotechnology. Then, membranes were washed and incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature. Immunocomplexes were detected by chemiluminescence using the ECL Western blotting reagent (GE HealthCare)in a ChemiDoc XRS detection system (BioRad). Densitometric analysis of protein bands by Western blot was performed using ImageJsoftware. The intensities of the target proteins bands were divided by β-Actin and the value compared to control (0 µM).
4.6.Statistical analyses
The statistical analysis was carried out with Student ’s t-test or ANOVA. The results were considered statistically significant if * p < 0.05, ** p < 0.01, and *** p < 0.001.
4.7.Molecular modelling
The 3-D ligand structures were built as we previously reported [67], while we have also described the detailed docking procedures elsewhere [68]. Molecular docking simulations were performed using the structure of the mammalian ERK enzyme complexed with the ATP competitive dual ERK1/2 kinase inhibitor MK-8353 (PDB ID 6DCG; complex reso- lution: 1.45 Å), since both the similarity of MK-8353 (Tanimoto coeffi- cient = 0.32) with 25a and the complex resolution were slightly greater than the observed for Ulixertinib (PDB ID 6GDQ) (Tanimoto coefficient = 0.29; Res_6GDQ = 1.86 Å). The choice of the most similar crystallo- graphic compound with the docked ligands tend to improve the accu- racy of the simulation results, probably because of ligand-induced fit effects [69]. Redocking of the crystallographic ligand yielded a heavy atom Root Mean Square Deviation (RMSD) of 0.76 Å, while the docking poses were ranked using the GlideScore fitness function [70]. The top- scored docking poses for the ligands were selected for visual inspec- tion and the subsequent analyses of the protein–ligand interactions. The crystallographic pose of ATP (PDB ID 4GT3) was used as a reference in the visual inspection after superposition of the two ERK complexes. Both ulixertinib and the most promising synthesized, 25a, compound were subjected to subsequent in silico analyses using the MetaDrug module of MetaCore platforms from Clarivate Analytics to predict their potential toxicities and pharmacokinetic profiles.
