Synthesis and biological activity evaluation of 3-(hetero) arylideneindolin-2-ones as potential c-Src inhibitors

Abstract Inhibition of c-Src is considered one of the most studied approaches to cancer treatment, with several heterocyclic compounds approved during the last 15 years as chemotherapeutic agents. Starting from the biological evaluation of an in-house collection of small molecules, indolinone was selected as the most promising scaffold. In this work, several functionalised indolinones were synthesised and their inhibitory potency and cytotoxic activity were assayed. The pharmacological profile of the most active compounds, supported by molecular modelling studies, revealed that the presence of an amino group increased the affinity towards the ATP-binding site of c-Src. At the same time, bulkier derivatizations seemed to improve the interactions within the enzymatic pocket. Overall, these data represent an early stage towards the optimisation of new, easy-to-be functionalised indolinones as potential c-Src inhibitors.


Introduction
c-Src is the first discovered and most studied member of the Src kinase family 1 . It is a cytoplasmic non-receptor protein kinase 2 , highly involved in intracellular signal transmission for the regulation of proliferation, migration, adhesion, invasion, and drug resistance of cancer cells 3,4 . c-Src is a 60-kDa protein anchored to the plasmatic and endosomal membranes, as well as in the nuclear compartment, through its N-terminal tail, in which basic residues are particularly abundant 5 . The protein structure consists of four domains: SH1, the kinase region in which the autophosphorylation site is present; SH2, which hosts the phosphorylated Tyr530 (in the C-terminal tail) and binds to PDGFR; SH3, responsible for the closed and inactive conformation of the protein; SH4, which presents the myristoylation site, essential for the membrane localisation 6 . c-Src is constitutively inactive, with a close conformation in which SH2 and SH3 protect a phosphorylated Tyr530, preventing interactions within the catalytic site. Dephosphorylation of Tyr530 triggers the activation of c-Src, whose open conformation allows Tyr419 phosphorylation and full availability to ligand-protein interactions 7,8 (Figure 1).
Overexpression and/or high kinase activity of c-Src is associated with cancer progression and metastasis processes, mainly because of the reduced cell-cell adhesion 9 and their increased motility, migration, and invasiveness 10 . Moreover, interaction with other overexpressed receptor tyrosine kinases (such as EGFR 11 , PDGFR 12 , ERBB 13 , FGFR 12 , and HGF 14 ) promotes the disruption of the closed conformation and exacerbates the activation of c-Src; such a condition is widely observed in glioblastoma 15 , breast 16 , lung 17 , thyroid 18 , bones 19,20 , colon 21 , pancreas 22 , and prostate cancers 23 .
In this context, many efforts have been made to design and discover new molecules acting as c-Src inhibitors, blocking its downstream pathways and impairing tumour progression [24][25][26][27] . To date, five molecules targeting Src have been approved by FDA (Figure 2), and several others are in clinical trials for the treatment of solid tumours and leukemias, often in association with other cytotoxic agents. In particular, Dasatinib and Bosutinib have been approved for the treatment of chronic myeloid leukaemia (CML) in adult and paediatric patients 28,29 . Saracatinib, characterised by a more selective profile in terms of Src family inhibition 30 , is in clinical trials for the treatment of several solid tumours 24 . Vandetanib and Ponatinib are multi-tyrosine kinase inhibitors, approved by FDA to treat CML, thyroid carcinoma, and Philadelphia chromosome-positive acute lymphoblastic leukaemia in adults 31,32 . Other promising c-Src inhibitors (in phase I or II, (Figure 2) are DGY-06-116 33 , eCF506 34 , Elzovantinib (TPX-0022) 35 , and Tirbanibulin, the latter targeting the Src substrate binding site 36 and recently approved for the topical treatment of actinic keratosis 37 .
As part of an ongoing research program aimed at identifying new chemical entities for anticancer chemotherapy, we focussed our efforts on the search of c-Src inhibitors. For this purpose, we initially screened an in-house collection library of small molecules characterised by great structural diversity. Analogues of the most promising compounds, containing the indolinone scaffold, were synthesised to highlight the structural determinants for the activity. Molecular modelling studies were also performed to propose a binding mode within the c-Src binding site.

Computational details
Docking studies were performed on the three-dimensional structure of the complex between c-Src and AP23464 (entry 2BDJ of the protein data bank, 2.5 Å resolution). The structure of c-Src was prepared using the Protein Preparation Wizard of the Maestro Software Package (Schr€ odinger suite, version 19.2). The software corrects bond orders and partial charges of the ligands and adds hydrogens to all atoms. Water molecules were removed, except those useful for stable binding interactions with the protein or ligand. Prime software was applied to define the correct conformation of missed side chains. Pre-processed protein was optimised with PROPKA and then minimised with an OPSL3 force field (convergence of heavy atoms set to a rmsd 0.3 Å).
The 2 D structures of all synthesised compounds and native ligand AP23464 were designed using the sketcher module of the Maestro suite and prepared with the Ligprep tool, which used the  Epik module to find protonation and tautomerization states of compounds at pH of 7.4 ± 0.5. To optimise the three-dimensional structures of the ligands, the Minimisation tool of the Macro model package was applied, using OPLS3 force field, while the other parameters were set to default values. Glide's receptor grid generation wizard was used to generate a three-dimensional grid with default settings. Then, flexible docking was performed with standard precision docking mode.

Biological assays
N-Terminal His6-tagged Src was purchased from Merck-Millipore (cat. 14-326). The reaction was performed according to the manufacturer's instructions with minor modifications: 500 lM Src-peptide (KVEKIGEGTYGVVYK), 100 lM ATP, 0.00087% NP-40, 10 ng of the enzyme, 10% DMSO in 10 lL at 30 C for 10 min. To avoid peptide adsorption to the plastic surface, protein low-binding tubes were used. ADP-Glo Kinase Assay (Promega) was then used to detect kinase activity accordingly with manufacturer's instruction with minor modifications. In details, reactions were transferred to white 384 well-plates and stopped by adding 10 mL of ADP-Glo Reagent (Promega) for 50 min at rt. 20 lL of Detection Reagent (Promega) was then added for 30 min and the reaction read using GloMax Discover microplate reader (Promega). Data were plotted using GraphPad Prism 5.0. ID 50 values were obtained according to Equation (1 where v is the measured reaction velocity, V is the apparent maximal velocity in the absence of an inhibitor, I is the inhibitor concentration, and ID 50 is the 50% inhibitory dose. Compounds tested were assumed to act as fully ATP-competitive inhibitors. Therefore, K i values were calculated accordingly to Equation (2).
where K i is the affinity of the inhibitor to the enzyme, [S] is the substrate (namely, ATP) concentration, and K m is the affinity of ATP calculated accordingly to Michaelis-Menten equation. Cytotoxicity experiments were carried out using the human MCF7 breast cancer cell line (ATCC HTB-22). Cells were cultured in RPMI1640 plus 10% foetal bovine serum at 37 C and 5% CO 2 . Compound potency was determined by a growth inhibition assay (CellTiter 96 V R Aqueous One Solution Cell Proliferation Assay MTS, Promega). 24 h after seeding in a 96-well plate, cells were exposed to the compounds (concentration range 1-100 mM) and 72 h later, 20 mL of 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium salt (MTS) was added to each well. The absorbance was measured using a FLUOstar OPTIMA plate reader (BMG Labtech GmbH, Offenburg, Germany) at 492 nm after 4 h of incubation at 37 C in 5% CO 2 . The IC 50 is the drug concentration causing 50% cell growth inhibition determined by the dose-response curves. Experiments were performed in triplicate.

Chemistry
All reagents and solvents were reagent grade or were purified by standard methods before use. Melting points were determined in open capillaries by an SMP3 apparatus and are uncorrected. 1 H and 13 C NMR spectra were recorded on Bruker AV600 spectrophotometer at 600 and 150 MHz, respectively. Chemical shifts are reported in ppm (d) relative to TMS. The coupling constants, J are reported in Hertz (Hz). All compounds were routinely checked by thin layer chromatography (TLC) using precoated silica gel 60 F 254 , aluminium foil and the spots were detected under UV light at 254 nm and 365 nm or were revealed by spraying with 10% phosphomolybdic acid in ethanol. Compounds 47 are part of an in-house library. The synthetic procedures relative to the above-mentioned products have been already described and reported in the cited references.

Synthesis and biological activity evaluation
A small collection of molecules containing highly diverse chemical scaffolds was initially screened on c-Src to evaluate their inhibiting activity at a 100 lM concentration (Table 1). Triazolo-triazine 13 showed the best inhibition potency (90%). However, its closely related analogue 14 did not affect the enzyme activity. Indolinone and N-hydroxyl indolinone derivatives 7-9 exhibited a significant activity (percent inhibition ranging from 77 to 84%). A modest inhibition profile was highlighted for azaindole 1 and kynurenic derivative 17 (about 58%), followed by thiazolylbenzoquinone 10 (43%) and dihydroquinolin-2-one 12 (40%), whereas all the other entries of the library showed low to no activity towards c-Src. ID 50 of both triazolo-triazine 13 and indolinone 8 were in the two-digit micromolar concentration (50 and 12.5 mM, respectively). These results prompted us to focus on the indolinone core for further investigations. This skeleton can be considered a very attractive target for biological activity evaluation and SAR studies, offering the possibility of easy chemical accessibility and several points for chemical decoration. We thus prepared and tested a second-generation set of compounds containing the indolinone scaffold substituted at positions 5 and/or 6 and bearing an arylidene side chain appended to position 3. We also prepared 24 to evaluate the role of the heteroaromatic system and both 22 and 23 to investigate the effect of the OH group on the nitrogen atom. All the compounds showed a good dose-dependent activity at 100 lM and 10 lM concentrations, with 26 having the best inhibitory potency of the series (95 and 67% at 100 and 10 lM, respectively) ( Table 2).
Compound 26 was prioritised and its structure was used as the starting point to prepare new analogues, maintaining unaltered the 6-chloro oxindole moiety and introducing decorations on the aromatic ring connected by a methylene bridge to C3. Knoevenagel condensation conditions were exploited to react 6chloro oxindole 31 with substituted aromatic/heteroaromatic aldehydes (Scheme 1), obtaining 15 new variously substituted derivatives.
All the analogues contained a phenyl ring connected by a conjugated vinyl fragment to C3. Various substituents with different stereoelectronic properties were introduced in the para position of the ring: cyano (32), nitro (33), dimethylamino (34), phenolic (35) and carboxylic (36) groups. Commercially available p-benzyloxy benzaldehyde and m-phenoxy benzaldehyde were employed for the preparation of 37 and 38. Additionally, alkylation of 4hydroxy benzaldehyde 39 as in 43-47 gave the opportunity to increase the overall compound length and eventually fill some empty regions of the cavity, either in presence of aromatic or aliphatic moieties, as well as with a terminal acidic or ester portion. Finally, compounds containing heterocyclic substituents were synthesised (48)(49)(50) to investigate their potential role as hydrogen bond acceptors/donors.
The percentage of inhibition at the concentrations of 100 lM and 10 lM, ID 50 of the most active compounds, and their K i were evaluated and reported in Table 3. Compounds 32-34 showed a good inhibitory activity towards the enzyme, with 34 having the best profile of inhibition (85% at 10 mM). Conversely, 36, with a carboxylic group directly linked to the phenyl ring, had very low activity. However, the percentage of inhibition was enhanced by increasing the length of the spacer between the aromatic ring and the acidic group ( cyclopropyl substituent led to an almost inactive compound (45). Moreover, the phenolic derivative 35 and the corresponding benzyl ether 37 maintained high inhibitory activity at the highest dose, which significantly decreased at 10 mM. The shift of a lipophilic aromatic ring to meta position on the phenyl group (38) did not influence the activity. Among compounds containing heterocyclic rings, 50 showed the best activity at both concentrations, whereas the thiophene-and thiazole-containing compounds 48 and 49, respectively, evidenced a drop in activity, in particular at the lowest concentration.
Compound 34 resulted to be the best of the series, showing significant inhibition of the enzyme at 10 mM (85%), with ID 50 and K i in the sub-micromolar range (0.71 and 0.48 mM, respectively). These findings suggest that the presence of an anilino nitrogen could have an important role in the inhibitory activity of the compounds.
With these results in our hands, the cytotoxicity test of selected molecules was carried out on the MCF-7 breast cancer cell line and compared to the c-Src inhibitor Dasatinib used as standard (Table 4). Thiazole 49 showed potency comparable to Dasatinib. Ester 43 and indole derivative 50 were able to exert some relevant but not excellent cytotoxic activity. It is worth noting that 34, the best inhibitor in the cell-free assay, did not show any cytotoxic activity, probably due to pharmacokinetic issues.

Molecular docking studies
In the attempt to find a putative binding mode of the new indolinone derivatives within the ATP-binding site of c-Src, molecular docking simulations were performed using the software Glide. For this purpose, the three-dimensional structure of the co-crystallized complex between c-Src and AP23464 (entry 2BDJ of the protein data bank, 2.50 Å resolution) was used for calculations. Preliminary validation of the docking protocol was done by re-docking AP23464 into the binding site. The resulting rmsd ¼ 0.903 Å, which was calculated on the atomic coordinates of the inhibitor, confirmed the reliability of the docking protocol. Two highly conserved hydrogen bonds were found between N7 and aniline nitrogen of AP23464 with the amide nitrogen and the carbonyl of the backbone of the Met341, respectively (Figure 3). The 3hydroxyphenylethyl group in position N9 on the purine scaffold protruded into the hydrophobic pocket and mediated a series of interactions. Specifically, the hydroxyl substituent formed hydrogen bonds with the carboxyl group of the Glu310 side chain and the backbone NH group of Asp404.
Considering that the new compounds were assayed as E/Z isomers, isomerisation was allowed during calculations. Docking scores showed that for each compound, the E-isomer was preferred to the Z-isomer. Overall, the indolinone derivatives showed a common binding pose with the lactam moiety that mimicked the hydrogen bond acceptor-donor motif represented by the N9 and the anilino NH group of AP23464. They served as hydrogen bond acceptor-donor system to give the classical interactions with Met341 of the hinge region ( Figure 3). Moreover, the chlorophenyl moiety was superimposable to the cyclopentyl ring of AP23464. Finally, the aryl appendage at C3 could mimic the phenol portion of the co-crystallized inhibitor. As an example, the predicted binding mode of 33, which showed single-digit micromolar ID 50 and K i , was able to maintain the two hydrogen bonds with Met341 in its best-scored pose (Figure 3). In addition, one of the oxygen atoms of the nitro group matched the phenol oxygen of AP23464, thus mimicking its ability to make an additional hydrogen bond with Asp404. A very similar binding mode was found for 50, where the NH group of the pendant indole nucleus was able to build a hydrogen bond with the side chain of Glu310, in addition to the classical interactions with Met341 ( Figure S1). On Scheme 1. General synthesis by Knoevenagel condensation starting from 6-chloro oxindole and different aromatic and heteroaromatic compounds. the other hand, molecular docking and scoring were not able to justify the binding affinity of 34, whose peripheral dimethylanilino moiety gave hydrophobic interactions with Ala403, whereas other interactions were not highlighted, in addition to the classical contacts with Met341 ( Figure S2). Increasing the length and bulkiness of the substituent at the pendant phenyl ring forced the terminal molecular edge outside the hydrophobic side pocket where the phenol ring of AP23464 was accommodated. As an example, the benzoic carboxyl group of the long side chain of 47, which retained the two hydrogen bonds with Met341, made two additional hydrogen bonds with Cys277 and Phe278. On the contrary, the docking pose of the corresponding methyl ester 44 lacked the fundamental anchor points on the binding site to avoid steric clashes due to the terminal methyl group of the ester function.

Conclusion
An in-house library of structurally diverse small molecules was screened with the aim of finding new c-Src inhibitors. It emerged that the most active compounds contained an indolinone core, which was selected as a promising scaffold for further investigations. Variously substituted 3-(hetero)arylideneindolin-2-ones were designed and synthesised to highlight the structural determinants for the activity. Most of the compounds showed a percentage of inhibition of the enzyme >70% at 100 lM and some of them also  24 h after seeding, cells were exposed for 72 h to the compounds and cytotoxicity was measured using MTS assay. Experiments were performed in triplicate and data represent mean values ± SD. Dasatinib was used as reference compound.
had a significant activity at 10 lM. Selected molecules were assayed on the human MCF-7 breast cancer cell line, showing moderate antiproliferative activity. Among the tested compounds, 49 demonstrated cytotoxicity comparable to that of the reference compound Dasatinib, but lacked any significant activity in the cell-based assay, indicating the possibility that some off-target effect could take place. Conversely, 34, which showed the best inhibition profile towards the isolated enzyme, resulted ineffective (IC 50 >100 lM) on the cell line. This finding likely depends on pharmacokinetic issues probably related to the difficulty of the compound to reach the intracellular target. Therefore, further studies will be necessary to improve the pharmacological profile of this class of compounds. The results of molecular docking studies, which were performed to propose a binding mode within the c-Src binding site, will guide and complement future activities.