Molecular design, synthesis and in vitro biological evaluation of thienopyrimidine–hydroxamic acids as chimeric kinase HDAC inhibitors: a challenging approach to combat cancer

Abstract A series of thieno[2,3-d]pyrimidine-based hydroxamic acid hybrids was designed and synthesised as multitarget anti-cancer agents, through incorporating the pharmacophore of EGFR, VEGFR2 into the inhibitory functionality of HDAC6. Three compounds (12c, 15b and 20b) were promising hits, whereas (12c) exhibited potent VEGFR2 inhibition (IC50=185 nM), potent EGFR inhibition (IC50=1.14 µM), and mild HDAC6 inhibition (23% inhibition). Moreover, compound (15c) was the most potent dual inhibitor among all the synthesised compounds, as it exhibited potent EGFR and VEGFR2 inhibition (IC50=19 nM) and (IC50=5.58 µM), respectively. While compounds (20d) and (7c) displayed nanomolar selective kinase inhibition with EGFR IC50= 68 nM and VEGFR2 IC50= 191 nM, respectively. All of the synthesised compounds were screened in vitro for their cytotoxic effect on 60 human NCI tumour cell lines. Additionally, molecular docking studies and ADMET studies were carried out to gain further insight into their binding mode and predict the pharmacokinetic properties of all the synthesised inhibitors.


Introduction
The development of multitarget drug therapy has become an important strategy for cancer treatment. Recently, the modulation of receptor tyrosine kinase (RTK) pathways by the inhibition of histone deacetylases (HDACs) has become a promising approach for cancer therapy 1 . HDACs are a family of numerous epigenetic enzymes that are important therapeutic targets for cancer 2 . HDACs have been classified into four distinct classes, in which class I (HDAC1-3 and 8), class II (HDAC4-7, 9 and 10), and class IV (HDAC11) 3 . HDAC6 is considered a therapeutically important target for cancer treatment due to its interaction with proteins involved in cell growth, migration, protein degradation, and apoptosis 4 . It is reported that HDAC inhibitors with large or rigid hydrophobic surface recognition moiety (SRM) and bulky aromatic or short linkers, are more efficient to achieve HDAC6 selectivity, which could achieve a closer approach of the hydroxamate group to Zn 2þ5 . Numerous HDAC6 inhibitors have been developed having large SRM connected to the zinc-binding group (ZBG) with diverse linkers (Figure 1) [5][6][7] .
Despite the efficiency of HDAC inhibition as monotherapy in haematologic malignancies, it did not show significant competency against solid tumours. As a result, multitarget therapy is developed as a challenging approach for increasing the efficacy of HDAC inhibitors 8 . The combination of HDAC and TKIs is considered a promising approach to enhance therapeutic effect and repress resistance 1 . For example, a synergistic effect is observed by a combination of gefitinib (EGFR inhibitor) with vorinostat (HDAC inhibitor) by enhancing tumour growth inhibition and apoptosis compared with monotherapy of selective EGFR inhibitors 9,10 . Another example, combination therapy of pazopanib (VEGFR2 inhibitor) and abexinostat (HDAC inhibitor) showed that HDAC inhibition could improve response and holdback resistance to pazopanib in patients with solid tumour malignancies and renal cell carcinoma 11 . Thus, it is suggested that a single molecule that simultaneously inhibits HDAC and RTK activities targeting multiple biological molecules and multiple signal pathways can not only enhance drug efficacy but also can have additive or even synergistic antitumor effects and can overcome the problems of cancer resistance and relapse. Additionally, it can overcome unfavourable pharmacokinetic properties, drug-drug interactions, poor patient compliance, and high drug cost involved in multicomponent drug cocktails 1,12 . CUDC-101 is a first-in-class multi-targeted hybrid that was designed to simultaneously inhibit HDACs, EGFR, and HER2. It was found that it induces significantly synergistic tumour regression in breast, NSCLC, liver, colon, head, neck, and pancreatic cancers 1 . Series of hybrids of 4-anilinoquinaziline based hydroxamic acid derivatives were synthesised and discovered as dual VEGFR-2/ HDAC inhibitors and EGFR/HDAC inhibitors ( Figure 2) 13,14 .
Moreover, the development of dual-inhibitors targeting EGFR and VEGFR2 has recently become a promising strategy for the treatment of diverse types of cancer to overcome problematic drug resistance and low response to selective tyrosine kinase inhibitors (TKIs) and to produce a synergistic effect 15 . Inhibition of both EGFR and VEGFR2 produces a more significant effect on angiogenesis than selective VEGFR2 inhibitors, where EGFRinduced the production of angiogenic growth factors, which indirectly affects angiogenesis 16,17 . Several EGFR/VEGFR-2 dual inhibitors have been designed and synthesised ( Figure 3) 15,16 .
In this study, six novel series of thieno [2,3-d]pyrimidine-based hydroxamic acid derivatives were rationally designed. The design is based on the strategy of multitarget drug therapy and the clinical advantage of combining HDAC, VEGFR2, or/and EGFR inhibition in one single drug. Investigating the structural requirements of EGFR, VEGFR2, and HDAC inhibitors led us to design several series of novel leads of multitarget or dual-target inhibitors by combining the pharmacophore of EGFR, VEGFR2 inhibitors with that of HDAC inhibitors.

Rationale of the design
Design strategy is based on the study of several reported dual inhibitors such as CUDC101 and compound B, which were reported as dual EGFR/HDAC and VEGFR2/HDAC inhibitors respectively; (bearing 4-anilinoquinaziline fragment) ( Figure 2) 14,18 . Additionally, novel dual EGFR/VEGFR2 inhibitors such as compound D and compound E; (bearing thieno [2,3-d] pyrimidine fragment;) were recently reported (Figure 3) 15 .
Based on these reported dual inhibitors and by investigating the structural requirements for EGFR, VEGFR2, and HDAC inhibition, we designed six series of hybrid molecules (7, 11, 12, 15, 19 and 20) with potent multi acting EGFR/VEGFR/HDAC inhibition in an attempt to synergise inhibitory activity, decrease resistance and relapse caused by single-target therapy and also avoid multicomponent drug cocktails therapy.
The design aimed at replacing the quinazoline fragment in CUDC101 and compound B with thieno[2,3-d]pyrimidine fragment as in compound D and compound E, representing SRM of HDAC inhibitors and also showing suitable interactions in the hinge region of ATP-binding site of EGFR and VEGFR2 ( Figure 4). This bioisostere replacement is also supported by recent reports of novel HDAC inhibitors bearing thieno [2,3-d]pyrimidine fragments representing the SRM that displayed excellent inhibitory activities on HDACs, such as compound A as shown in (Figure 1) 7 . Addition of aniline, 5-aminobenzimidazole, (4-phenylthioureido)phenylamine) and (4-phenylureido)phenylamine) fragments to 4thieno [2,3-d]pyrimidine ring will extend the structure to the hydrophobic pocket of EGFR and VEGFR2 ( Figure 4). Next, a hydroxamic acid functionality was directly attached to the thieno [2,3-d]pyrimidine fragment (SRM) or connected with a short linker through a connecting amide unit as the previously reported HDAC6 inhibitors (Figure 1) 19,20 , in which it served as a bridge to properly display HDAC6 inhibitory pharmacophore features and the solvent-accessible area of EGFR-and VEGFR2-binding sites ( Figure 4).

Chemistry
The route adopted for the synthesis of key intermediates (3) and (5a-d) were prepared according to literature as illustrated in (Schemes 1 and 2), respectively. The synthetic route of the first three series of the target novel hydroxamic acid derivatives with different urea and thiourea moieties (7a-d, 11a-c, 12a-c) is depicted in (Scheme 3). Moreover, the synthesis of another three series of the novel target hydroxamic acid derivatives with different aniline moieties (15a-d, 19 b-d, 20 b-d) is depicted in (Scheme 4).  In (Scheme 1), intermediate (1) was synthesised by applying the Gewald reaction, in which ethyl cyanoacetate, ethyl acetoacetate, and sulphur were dissolved in absolute ethanol in the presence of morpholine 21 . The cyclisation of (1) into (2) was achieved through refluxing (1) in a mixture of formamide and acetic acid for 40 h. The structure was confirmed by its reported melting point. Compound (3) was synthesised through chlorination of (2) by refluxing it with phosphorus oxychloride for 3 h 22 , (Scheme 1).
The synthesis of the intermediates (4a-d) was carried out by reacting the appropriate isocyanate with p-nitroaniline in dry dichloromethane for 48 h at room temperature 23 to yield compounds (4a-d), which are then reduced to their corresponding amino derivatives (5a-d) using 10% Pd-C in methanol 24 , (Scheme 2).
In (Scheme 3), the synthesised key intermediates (5a-d) were then coupled with the previously prepared chloro derivative (3) 25 to yield (6a-d). The first novel series of the target hydroxamic acid derivatives (7a-d) was obtained through the reaction of the ethyl ester group in compounds (6a-d) with freshly prepared hydroxylamine in absolute ethanol to give (7a-d) 26,27 , (Scheme 3). Compounds (7a-d) were confirmed by many spectral data, in which 1 HNMR signals were consistent with protons of the targeted compounds (7a-d). 1 HNMR signals revealed equally integrated signals between d 8.23-9.82 ppm representing D 2 O exchangeable protons of the NH linker, the urea protons, and the NH of hydroxamic acid moiety in all target compounds (7a-d).
Noticeably, 1 HNMR signals revealed the absence of the triplet and the quartette peak around d 1.31 À 1.36 ppm and d 4.30-4.38 ppm, which confirms the absence of the ester group and the entrance of hydroxamic acid moiety.
Moreover, the synthesis of another two novel series of hydroxamic acid derivatives with urea and thiourea moieties (11ac and 12a-c) was depicted in (Scheme 3), through multistep reactions. The synthesis starts with the hydrolysis of the carboxylate ethyl ester of the previously prepared compounds (6a-c) (Scheme 3), through refluxing it with Li(OH) 2 in THF and water to prepare the intermediate carboxylic acid derivatives (8a-c) 28 . Furthermore, the synthesis of novel two series of intermediate ethyl ester derivatives (9a-c, 10a-c) (Scheme 3) was achieved through coupling of the previously prepared carboxylic acid derivatives (8a-c) with the appropriate amino acid ester hydrochloride salt, utilising EDCI with HOBt in dry DMF 29 . This method offered a high yield of novel intermediate compounds (9a-c and 10a-c) confirmed by many spectral data. 1  Finally, the ethyl ester group in the previously prepared novel compounds (9a-c, 10a-c) reacted with freshly prepared hydroxylamine in absolute ethanol to obtain target novel hydroxamic acid derivatives (11a-c, 12a-c) (Scheme 3). The two target novel series (11a-c, 12a-c) were confirmed by many spectral data, where 1 HNMR spectrum revealed the disappearance of the quartette signal of the two protons of OCH 2 -CH 3 around d 4.05-4.25 and the triplet signal of the three protons OCH 2 -CH 3 around d 1.17-1.34 ppm, in which it confirms the absence of the ester group and the entrance of the hydroxamic acid moiety.
In (Scheme 4), the purchased compound (5-nitrobenzimidazole) was reduced to afford the corresponding amino derivative (13a) using 10% Pd-C in methanol. Prepared compound (13a) and purchased aniline derivatives (13b-d) were also coupled with the previously prepared chloro derivative (3) to give the corresponding thieno [2,3-d]pyrimidine derivatives bearing aniline moieties (14a-d) (Scheme 4). Finally, the route adopted for the synthesis of the three target novel series of hydroxamic acid derivatives bearing aniline moiety (15a-d, 19 b-d and 20 b-d) (Scheme 4) is the same route depicted in (Scheme 3) with the same reagents and conditions. The first novel series of the target hydroxamic acid derivatives (15a-d) was obtained through the reaction of the ethyl ester group in compounds (14a-d) with freshly prepared hydroxylamine in absolute ethanol to give (15a-d) 26,27 , (Scheme 4). Then, the hydrolysis of the carboxylate ethyl ester of the previously prepared compounds (14b-d) (Scheme 4), through refluxing it with Li(OH) 2 in THF and water to prepare the intermediate carboxylic acid derivatives (16b-d) 28 . The previously prepared carboxylic acid derivatives (16b-d) are coupled with the appropriate amino acid ester hydrochloride salt, utilising EDCI with HOBt in dry DMF 29 , to obtain a high yield of novel intermediate compounds (17b-d and 18b-d), (Scheme 4). Finally, the ethyl ester group in the previously prepared novel compounds (17b-d, 18bd) reacted with freshly prepared hydroxylamine in absolute ethanol to obtain target novel hydroxamic acid derivatives (19b-d, 20b-d) confirmed by many spectral data, (Scheme 4).

Biological evaluation
The target compounds were evaluated for their in vitro inhibitory activities against EGFR, VEGFR-2, and HDAC6. The EGFR and VEGFR-2 inhibitory assays were performed at Thermo Fischer Scientific, Madison, WI, USA (SelectScreenServices@thermofisher. com). The HDAC inhibitory assays were performed at BPS Bioscience, San Diego, CA, USA (www.bpsbioscience.com). Initially, all of the synthesised target compounds were evaluated for their percent inhibitory activity against EGFR, VEGFR-2, and HDACs at 10 mM single dose concentration, in which their mean percent inhibition is summarised in (Figure 5, Tables 1 and 2). Representative compounds were evaluated for their percent inhibitory activity against EGFR and VEGFR-2 at 6 different concentrations (0.1 nM, 1 nM, 10 nM, 100 nM, 1 mM, 10 mM) to subsequently calculate their IC 50 values (Table 3). Additionally, all of the synthesised target compounds were submitted to the National Cancer Institute "NCI" Bethesda, Maryland, USA (www.dtp.nci.nih. gov), to evaluate theirs in vitro antiproliferative activities against NCI 60-cancer cell line panel.

In vitro EGFR inhibitory assay
By closely investigating the structure activity relationship, thieno [2,3-d]pyrimidine derivatives with aniline moiety showed an excellent inhibitory profile compared to derivatives with urea and benzimidazole moieties (Tables 1 and 2). Moreover, thieno [2,3d]pyrimidine derivatives with aniline moiety, mono-substitutions, or di-substitutions at 3-or 4-positions on its terminal phenyl ring exhibited excellent inhibitory profile, as previously reported 30 .
In general, EGFR inhibition increases by increasing the number of carbons in the linker chain between the hydroxamic acid moiety and the amide group to three carbons or by directly attaching the hydroxamic acid with the thienopyrimidine fragment. So that compounds (20b and 20c) showed 92 and 96% inhibition, compared to inhibition of 76 and 78% for the two carbons linker of (19b and 19c), respectively (Table 2).
Promising candidates, (12c, 15c, 20b and 20d) were chosen as representatives for evaluation of potential enzyme inhibitory activity (IC 50 ) ( Table 3). Compound (15c) demonstrated the most potent EGFR inhibition with IC 50 value of 19 nM since its hydroxyl amine group is directly attached to the thieno[2,3-d]pyrimidine ring and also it has 4-Cl substitution on its terminal phenyl ring. Compound (20d) exhibited more potent inhibition than (20b) with IC 50 value of 68 nM compared to 155 nM. Since both (20d and 20b) have a linker of 3 carbons, thus the (4-F, 3-Cl) substitution at the terminal phenyl ring of compound (20d) is suggested to be responsible for the increase of its inhibitory activity compared to (20b), with no substitution at its terminal phenyl ring.
Noticeably, compound (12c) bearing urea moiety with 3-CF 3 at its terminal phenyl ring exhibited IC 50 value of 1.14 mM, in which it showed the highest EGFR inhibition among all other synthesised derivatives bearing urea moiety.
Promising candidates, (7c, 12c, and 15c) were chosen as representatives for the evaluation of potential enzyme inhibitory activity (IC 50 ) ( Table 3). All of the evaluated target compounds exhibited potent VEGFR-2 inhibitory activity with IC 50 values ranging from (185 nM to 5.58 mM). Compounds (7c and 12c) bearing urea moiety exhibited potent VEGFR-2 inhibition with IC 50 values of 191 and 185 nM, respectively. These results suggest that -CF 3 substitution at the terminal phenyl ring of compounds (7c and 12c) and the urea moiety is responsible for their high inhibitory activity, regardless of having 3 carbons in the linker chain (12c) or without having a linker (7c). Compound (15c) with 4-chloro aniline moiety exhibited significant inhibition to VEGFR-2 with a higher IC 50 value of 5.58 mM than other derivatives bearing urea moiety.

In vitro HDAC inhibitory assay
Initially, all synthesised compounds were screened against HDAC6 at 10 lM concentration. HDAC6 enzyme inhibition assay was selected since our newly synthesised hydroxamic acid compounds have a large capping group suitable for fitting in the large hydrophobic channel of HDAC6, which is unique to the HDAC6 receptor 26 . The newly designed compounds were synthesized either having short hydrocarbon linkers between the zinc-binding group (ZBG) and the connecting amide unit or having no linear hydrocarbon linker between the hydroxamic acid moiety and the thieno[2,3-d]pyrimidine fragment, as the previously reported potent HDAC6 inhibitors 19,20 .   Screening of the synthesised compounds against HDAC6 at 10 lM concentration is shown in (Tables 1 and 2). For derivatives bearing urea moiety, compound (12c) with three carbons in the linker region showed considerable HDAC6 inhibition by 23% (Table 1). For derivatives bearing aniline moiety, compound (20b) exhibited the most significant HDAC6 inhibition by 56% (Table 2). Compound (20b); bearing unsubstituted aniline moiety; is the smallest capping group among all synthesized compounds. The small capping group may allow it to fit in the HDAC6 hydrophobic channel. Additionally, compound (20b) had 3 carbons in the linker region compared to (15b) with no linear hydrocarbon linker and (19b) with 2 carbons in the linker region showing 30% and 10% inhibition, respectively (Table 2). Thus, having a small capping group of unsubstituted aniline and a longer carbon chain in the linker region between the hydroxamic acid moiety and the connecting amide is recommended for optimum HDAC6 inhibition.
Moreover, many of the synthesised novel compounds can be considered as promising leads of EGFR and VEGFR-2 dual inhibitors, (Table 2 and Figure 5). Where, compound (15c) is considered to be the most potent dual inhibitor among all the synthesised compounds, since it exhibits potent EGFR and VEGFR-2 inhibition (97%, IC 50 ¼ 19 nM) and (87%, IC 50 ¼5.58 mM), respectively. Compound (15d) exhibited potent EGFR and moderate VEGFR-2 inhibition (97%) and (54%), respectively. Two selective potent inhibitors were also considered, such as (7c and 20d). Compound (7c) is considered as a potent novel VEGFR-2 inhibitor (100%  . Design strategy of hybrid multitarget (EGFR, VEGFR, and HDAC inhibitors). Colour codes show essential pharmacophore for HDAC inhibitors composed of surface recognition moiety (purple), linker (red), and zinc-binding group (blue). The cyan oval shape represents the area binding to the hinge region of the ATP binding site of VEGFR2 and EGFR and the green square represents the area binding to the hydrophobic pocket of VEGFR2 and EGFR, while the rest of the structure represents the area fitting in the solvent-accessible area of VEGFR2 and EGFR. inhibition, IC 50 ¼ 191 nM), while compound (20d) is considered as a potent novel EGFR inhibitor (96% inhibition, IC 50 ¼ 68 nM).

In vitro antiproliferative activity
All of the synthesised structures were selected and submitted to NCI based on computer modelling techniques and the degree of diversity of structures for the evaluation of their antiproliferative activity. The 20 synthesised compounds were selected with their respective NCI codes [NCS D-820157-NCS D-820176] for the inhibition percent assay at an initial single dose of 10 lM on the full NCI 60 cell panel, where the mean graph of the percent growth for all compounds on the full NCI cell panel is evaluated, compared to the untreated control cells (Supplementary Material). In vitro cell line inhibition % of all synthesised compounds is represented in (Table 1  In series (12a-c) bearing urea moiety with three carbons in the linker chain between the hydroxamic acid moiety and the amide group, compound (12c) with urea moiety and CF 3 at the terminal phenyl ring showed moderate inhibitory activity mostly against CNS cancer [SNB-75 cancer cell line (29.2% inhibition) and SF-539 cell line (28.5% inhibition)]. In series (20 b-d) bearing aniline moiety with three carbons in linker chain between the hydroxamic acid moiety and the amide group, compound (20d) exhibited inhibitory activity mostly against renal cancer UO-31 cancer cell line (23.7% inhibition), breast cancer MDA-MB-468 (16.2% inhibition) and NSCLC NCI-H226 (11.3% inhibition).
In series (11a-c) and (19 b-d) with two carbons in the linker chain between the hydroxamic acid moiety and the amide group, most of the compounds showed lower inhibitory activities against most of the investigated cell lines.
Recent studies revealed that VEGFR-2 is a major contributor to the growth of CNS tumours and its interruption may effectively suppress tumour growth in CNS cancer 31 . It was also reported that VEGFR-2 is a promising platform for anti-angiogenesis cancer therapy for colon cancer cells 32 . Also, EGFR was reported to be Red-coloured data represent potent inhibition 70-100%, green-coloured data represent moderate inhibition 30-70%, blue-coloured data represent mild inhibition 20-30%, and the black-coloured data represent very weak inhibition <20%. Table 1. Mean % inhibition of the investigated compounds on EGFR, VEGFR-2 and HDAC6 at 10 mM concentration.
overexpressed in different cancer cell lines, including NSCLC 33 , renal cancer 34 , and brain cancer 35 . Noticeably, it is observed that most of the compounds showed weak to moderate inhibition towards Renal UO-31 and CNS cancer SNB-75 cell lines (20-46% inhibition). From these findings, it is suggested that the potent in vitro inhibition of compound (7c) towards VEGFR-2 (100% inhibition, IC 50 ¼ 191 nM) contributed to the significant inhibitory activity against CNS [SNB-75 cancer cell line (46% inhibition) and SF-539 cancer cell line (38.6% inhibition)] and colon KM12 cancer cell line (33.7% inhibition). It is also suggested that the potent in vitro inhibition of compound (15c) towards EGFR (97% inhibition, IC 50 ¼ 19 nM) and VEGFR2 (87% inhibition, IC 50 ¼ 5.58 mM) contributed to the significant inhibitory activity against renal, CNS and NSCLC cancer cell lines as discussed. Additionally, it is also observed that (15c) is the only compound that showed considerable inhibition towards NSCLC H522 (25.4% inhibition) and EKVX (20.6% inhibition) cancer cell lines, which is contributed to its potent EGFR inhibition with (IC 50 ¼ 19 nM).
It is obvious that despite most of our newly synthesised compounds exhibited potent EGFR and/or VEGFR-2 in vitro inhibitory activities (90-100% inhibition) but unexpectedly some of the compounds showed moderate in vitro inhibitory activity against NCI cell lines at 10 mM concentration. It is suggested that maybe the increased polarity of our synthesised hydroxamic acids, might contribute to poor cell membrane permeability towards NCI cancer cell lines. Also, VEGFR-2 inhibitors showed less sensitivity towards NCI cancer cell lines than human vascular endothelial cells, which contributed to their moderate antiproliferative activity towards NCI cancer cell lines 30 .

Molecular modelling studies
3.3.1. Docking study A molecular docking study using Discovery Studio 4.5 was performed using the C-DOCKER algorithm. A docking study is performed to analyse binding affinities, binding modes, and orientation of the newly synthesised target compounds into the active binding site of three different receptors such as EGFR, VEGFR-2, and HDAC6. The best docking pose is selected based on its binding mode in comparison to that of the reference ligand. The crystal structure of EGFR, VEGFR-2, and HDAC6 co-crystallised with their reference ligands were downloaded from protein data bank (www.rcsb.org) with PDB codes: 4HJO, 3VHE, and 5G0H, respectively. Re-docking the reference compounds into the X-ray crystal structure of the active site of its receptor was performed for validation.
3.3.1.1. Docking study on EGFR. The first docking study was done on EGFR (PDB code 4HJO) co-crystallised with the reference ligand (erlotinib). The root mean square difference (RMSD) between the re-docked and the co-crystallised conformers of erlotinib is 0.5 Å, which signifies the validity of the C-DOCKER algorithm. Besides, re-docked erlotinib retrieved the reported binding mode into the active binding site of EGFR (-CDOCKER energy ¼ 27.18 Kcal/mol) as depicted in (Figure 6) 36 . Where nitrogen of pyrimidine ring forms a hydrogen bond with NH of MET769, pyrimidine ring forms hydrophobic interactions with ALA719, MET769 and LEU820, phenyl ring forms hydrophobic interaction with LEU694 and LEU820 and the terminal phenyl ring forms hydrophobic interaction with VAL702, ALA719 and LYS721 ( Figure 6). Docking of all synthesised compounds was done on EGFR (PDB code 4HJO), where the binding mode of representative thieno[2,3-d]pyrimidine derivatives were investigated to justify the potent or significant EGFR inhibition (high % inhibition or low IC 50 ) exhibited by some of the synthesised compounds ( Table 2 of Supplementary Material). In general, docking results revealed that thieno[2,3-d]pyrimidine derivatives with aniline moiety (15c, 15d, 19d, 20 b-d) showed higher binding affinities than erlotinib, fulfilled the reported binding mode and showed extra interactions with EGFR active site. Although thieno [2,3-d]pyrimidine derivatives with urea moiety showed higher binding affinities than erlotinib but it did not fulfil all the key interactions with the active site of EGFR. The binding mode and energies of three representatives docked compounds are presented in (Figure 7). Compound (15c) showed the most potent EGFR inhibitory activity (97%, IC 50 ¼ 19 nM) and revealed better binding affinity than erlotinib (-CDOCKER energy ¼ 28.34 Kcal/mol). The binding mode of (15c) was consistent with erlotinib as shown in (Table 2 of Supplementary Material). Additionally, the methyl group of (15c) showed extra hydrophobic interactions with VAL702 and LEU820 and chlorine showed extra hydrophobic interactions with adjacent hydrophobic pockets including LEU764 and LEU834. Compound (20d) exhibited potent inhibitory activity towards EGFR (97%, IC 50 ¼ 68 nM) and revealed better binding affinity than erlotinib (-CDOCKER energy ¼ 42.52 Kcal/mol). The binding mode was consistent with erlotinib as shown in (Table 2 of Supplementary Material), where the oxygen of the hydroxyl group showed 2 additional hydrogen bonds with LYS692 and VAL693, oxygen of hydroxamic carbonyl group and hydrogen of hydroxamic amine showed extra hydrogen bonds with LYS692 and LEU694, respectively. Moreover, the methyl group of (20d) showed additional hydrophobic interaction with VAL702, and chlorine showed additional hydrophobic interaction with ALA719 and LYS721. Compound (12c) bearing urea moiety exhibited significant EGFR inhibitory activity (66%, IC 50 ¼ 1.14 mM), but it showed lower inhibition than other compounds bearing aniline moiety. Compound (12c) revealed a missing hydrogen bond with the essential MET796 although having better binding affinity than erlotinib (-CDOCKER energy ¼ 43.31 Kcal/ mol). Compound (12c) showed five extra hydrogen bonds, where the oxygen of carbonyl urea showed hydrogen bond with THR766, oxygen of hydroxamic carbonyl group showed hydrogen bond with CYS773, oxygen of hydroxyl group showed hydrogen bond with PHE771, F of terminal phenyl ring showed hydrogen bond with PHE832 and hydrogen of urea amine showed hydrogen bond with ASP831. Additionally, the three fluorine atoms of CF 3 showed three halogen interactions with CYS751 and one halogen interaction with ARG752. Pyrimidine ring of (12c) forms hydrophobic interactions with LEU694 and LEU820, thiophene ring forms hydrophobic interaction with LEU694, phenyl ring forms hydrophobic interaction with VAL702 and LEU820 and terminal phenyl ring shows four hydrophobic interactions with extra hydrophobic pocket including LEU753, LEU764, LEU834, and MET742. Moreover, a methyl group and C of CF 3 of (12c) showed additional hydrophobic interactions with LEU694 and CYS751, respectively.  (Table 3 of Supplementary Material). In general, docking results revealed that most of the thieno[2,3-d]pyrimidine derivatives with urea moiety (7c and 12c) showed higher binding affinities than the reference compound and compounds with aniline moiety (15c) which is consistent with VEGFR-2 inhibition assay. In which, compounds bearing urea moiety, such as (12c) with (IC 50 ¼ 185 nM), and (7c) with (IC 50 ¼ 191 nM) exhibited potent VEGFR-2 inhibition compared to compound (15c) bearing aniline moiety with (IC 50 ¼ 5.58 mM). Where (12c) and (7c) retrieved all the essential hydrogen bond and hydrophobic interactions of the reference compound, while compound (15c) missed the essential hydrogen bond interactions with GLU885 and ASP1046 although its pyrimidine ring retrieved the essential hydrogen bond with CYS919 (Table 3 of Supplementary Material). The binding interactions and energies of two representatives docked compounds are presented in ( Figure  9). Additionally, we found that compounds (12c) and (7c); both bearing urea moiety with CF 3 substituent and compound (15c) bearing 4-chloro aniline moiety showed higher binding affinities and made extra hydrophobic bonds than the reference compound. Where CF 3 in compound (12c) made extra hydrophobic bonding with ILE892, LEU1019 and HIS1026, F of CF 3 in compound (7c) made an extra halogen hydrogen bond with HIS1026 and 4-Cl of the terminal phenyl ring in compound (15c) forms additional hydrophobic interaction with LEU889, VAL899 and LYS868. According to previously reported SAR, this area is an allosteric hydrophobic binding pocket revealed for type II VEGFR inhibitors, in which the phenylalanine residue of the DFG loop flips out of its lipophilic pocket to increase the binding affinity of inhibitors to VEGFR receptors and increase residence time 37,38 .
3.3.1.3. Docking study on HDAC6. The final docking study was done on HDAC6 (PDB code: 5G0H) co-crystallised with the reference compound (Trichostatin A; TSA). RMSD is 0.44 Å, which signifies the validity of CDOCKER algorithm. Moreover, the reported binding mode 39 of TSA was retrieved into the X-ray crystal structure of HDAC6 (-CDOCKER energy ¼ 25.49) as shown in ( Figure  10), where the oxygen of hydroxamic hydroxyl group forms a hydrogen bond with HIS573 and oxygen of both hydroxamic carbonyl and hydroxyl groups forms coordinate bonds with zinc metal in a bidentate fashion with distances 2.24 Å and 2.20 Å, respectively. Additionally, the phenyl ring of TSA shows pi-pi stacking with PHE643 and the aliphatic chain of the linker showed hydrophobic interaction with PHE583 and PHE643 ( Figure 10). Docking of all synthesised compounds was performed on HDAC6 (PDB code: 5G0H), where the binding mode of representative thieno [2,3-d]pyrimidine derivatives were explored to justify the significant inhibition of compound (20 b) (56% inhibition) against HDAC6 among other derivatives that showed remarkably lower % inhibition against HDAC6 (Table 4 of Supplementary Material).
The binding interactions and energies of three representatives docked compounds (12c, 19 b and 20 b) are presented in ( Figure  11). Compound (20 b) with unsubstituted aniline moiety and three carbons in the linker region exhibited the most significant HDAC6 inhibition among all the synthesised compounds (56% inhibition) and it revealed higher binding affinity than TSA (-CDOCKER energy ¼ 34.32). This significant inhibitory activity for (20 b) was justified by its consistent binding mode with the reported binding mode of TSA against HDAC6 (PDB: 5G0H), where the oxygen of hydroxamic carbonyl and hydroxyl groups forms coordinate bonds with zinc metal in a bidentate fashion with distances 2.24 Å and 2.36 Å, respectively. The oxygen of the hydroxamic hydroxyl group forms a hydrogen bond with HIS573 and oxygen of hydroxamic and amidic carbonyl group forms two additional hydrogen bonds with TYR745 and HIS614, respectively. Also, the thiophene ring of (20 b) shows pi-pi stacking with PHE643. While, compound (19 b) with unsubstituted aniline moiety and two carbons in the linker region exhibited insignificant inhibitory activity (10%), although it revealed higher binding affinity (-CDOCKER energy ¼ 37.68) than (20 b) (Figure 11). By comparing the binding mode of compounds (20 b) and (19 b), it was found that the insignificant inhibitory activity of the compound (19 b) is certainly accounted for by having a shorter linker between the hydroxamate moiety and the connecting amide group (two carbons) than compound (20 b) (3 carbons) since they both bear the same hydrophobic cap group. The shorter two carbons linker of compound (19 b) contributes to its failure to form a coordinate bond with zinc metal in a bidentate fashion, where the oxygen of hydroxamic carbonyl group forms only one coordinate bond with zinc metal (Figure 11). Whereas compound (12c) with urea moiety exhibited lower HDAC6 inhibition (23%) than (20 b), although it revealed higher binding affinity (-CDOCKER energy ¼ 37.58) than (20 b) and TSA. The decrease in HDAC6 inhibition by compound (12c) is justified by its binding mode as described in (Figure 11), in which the oxygen of hydroxamic carbonyl group forms only one coordinate bond with zinc metal in a monodentate fashion, despite fulfilling all other key interactions with reported essential amino acids (  [2,3-d]pyrimidine derivatives with biphenyl urea or thiourea moiety, in which it has a better binding affinity and showed higher inhibition to HDAC6.

In silico ADMET predictive study
A computer-aided ADMET study was performed using Accelrys Discovery Studio 4.5 software protocols to investigate pharmacokinetics properties of newly synthesised thieno[2,3-d] pyrimidine hydroxamic acid derivatives and their ester derivatives, such as absorption, distribution, metabolism, excretion and toxicity. This study was done to investigate the unexpected moderate in vitro antiproliferative activity of most of the synthesised hydroxamic acid compounds despite their potent in vitro EGFR and VEGFR-2 enzyme inhibition. Aqueous solubility, blood brain barrier (BBB) penetration, CYP2D6 binding, human intestinal absorption (HIA), toxicity and plasma protein binding descriptors are predicted. 2 D ADMET plot for thieno [2,3-d] pyrimidine hydroxamic acid derivatives is drawn between ADMET_PSA_2D versus ADMET_AlogP98 (Figure 1 of Supplementary Material). Two analogous 95% and 99% confidence ellipses are shown in the biplot corresponding to HIA and BBB models 40 . The results of the calculated representative parameters for thieno [2,3-d] pyrimidine hydroxamic acid derivatives are tabulated in (Table 5 of Supplementary Material). There is an inverse relationship between polar surface area (PSA) with percent HIA and thus cell membrane permeability 41 . ALogP is used to indicate the lipophilicity of the investigated compounds, thus calculating ALogP with PSA is used to bring information about hydrogen bond characteristics 42 . Optimum cell membrane permeability is predicted if the compound has (PSA < 140˚A2 and AlogP98 < 5) 42 . In this study, all hydroxamic acid compounds showed ALogP98 < 5, 14 compounds showed high accepted values of PSA: 140˚A2 < PSA > 100˚A2, indicating considerable cell membrane permeability but not highly permeable due to its high value of PSA (>100˚A2), which is inversely proportional with cell membrane permeability. Whereas, the rest of the compounds bearing urea moiety and with linker between the hydroxamic acid moiety and the amide connecting unit (11a-c, 12a-c) showed PSA > 140, which indicated poor cell membrane permeability and absorption.
By comparing the AlogP98 and the PSA of the synthesised thieno [2,3-d] pyrimidine hydroxamic acids and their esters ( Table 6 of Supplementary Material), it was found that almost all of the prepared thieno[2,3-d]pyrimidine esters showed lower PSA (PSA < 140) and higher AlogP than most of the synthesised target thieno[2,3-d]pyrimidine hydroxamic acids. These results coincide with the unsatisfactory poor to moderate in vitro antiproliferative activities of the hydroxamic acid derivatives which could be attributed to their poor cell membrane permeability. Also, it was found that the synthesised hydroxamic acid compounds bearing aniline moiety were predicted to show better intestinal absorption and cell membrane permeability than compounds bearing urea moieties. While compounds (7a-d) bearing urea moiety with no linear hydrocarbon linker between the hydroxamic acid moiety and the thieno[2,3-d]pyrimidine fragment were predicted to have acceptable cell membrane permeability as an exception.

Conclusion
In the current study, series of hybrids of thieno[2,3-d]pyrimidine fragment bearing arylamino and diaryl urea moieties at 4-position linked to hydroxamic acid moiety were designed and synthesised as inhibitors for EGFR/VEGFR-2/HDAC6. Compounds (12c, 15 b and 20 b) may be considered as potential multitarget inhibitors against EGFR/VEGFR-2/HDAC6. Additionally, other synthesised novel hydroxamic acid compounds may be considered as EGFR/ VEGFR-2 dual inhibitors, such as (7a, 15c and 15d), in which the hydroxamic acid moiety is directly attached to the thieno[2,3d]pyrimidine fragment, contributing to maximum inhibitory dual activity. Compound (15c) could be the most potent dual EGFR/ VEGFR-2 inhibitor in this study. Moreover, promising selective VEGFR-2 (7c and 11c) and EGFR (20d) inhibitors were identified. The results of the enzyme inhibitory assays were found to coincide with the molecular docking studies performed on the three enzymes EGFR, VEGFR-2 and HDAC6.
These results supported the rationale of the design strategy, where the thieno[2,3-d]pyrimidine fragment bearing urea moiety or aniline moiety may be considered as a promising capping group. Unexpectedly, it is observed that most of the compounds showed common inhibition (weak to moderate inhibition) towards Renal UO-31 and CNS cancer SNB-75 cell lines (20-29% inhibition) although they showed potent enzyme inhibitory profiles. Results of the ADMET study revealed poor cell membrane permeability, explaining their unsatisfactory in vitro antiproliferative activities. Future structural modifications will be studied by implementing specific structural modifications to optimise the hydrophobicity of the designed compounds for enhancing their cell membrane permeability.

Chemistry
Reagents and starting materials were obtained from Alfa-Aeser organics and Sigma Aldrich without further purification. Melting points were uncorrected and recorded using BUCHI B-540 apparatus. Thin layer chromatography (TLC) is performed on 0.255 mm silica gel plates and is purchased from Merck to monitor reactions under U.V. light (254 nm). Mass spectrum was recorded in Thermo Scientific GCMS model on Direct Inlet part to mass analyser ISQ at the Regional Centre for Mycology and Biotechnology, Al-Azhar University. 1 H and 13 C NMR spectra were recorded on a Bruker Avance III HD FT-high resolution-NMR 400 MHz and scaled as d in ppm at the Centre for Drug Discovery Research and Development, Faculty of Pharmacy, Ain Shams University. Description of coupling patterns is as follows: s, singlet; br s, singlet broad; d, doublet; t, triplet; m, multiplet; and 1H, 2H, 3H, etc. The coupling constants are described as J, which was rounded off to one decimal place. IR spectra were done at the Micro analytical centre, Cairo University, and recorded on a 4100 Jasco spectrophotometer. Elemental analysis was performed at the Regional Centre for Mycology and Biotechnology, AL Azhar University.

General procedure for preparation of compounds (4a-d)
The appropriate isocyanate (7 mmol;1 equiv.) was added to a solution of p-nitroaniline (7 mmol;1 equiv.) in dry methylene chloride (20 ml) and stirred at room temperature for 24 h. The formed solid was filtered, stirred again with dry methylene chloride then collected by filtration and dried. Crystallisation from ethanol yielded the titled compounds (4a-d) 23 .

General procedure for preparation of compounds (5a-d)
A mixture of the appropriate nitro phenyl urea or thiourea derivatives (4a-d) (4 mmol) and Pd-C (0.1 g, 10%) in methanol (100 ml) was stirred under H 2 at 60 bars, at room temperature for 6 h. Then Pd-C was completely removed through filtration over celite. The filtrate was evaporated under vacuum and dried to afford titled compounds (5a-d) which were then recrystallized from its suitable solvent 24 .

5.2.2.2.
HDAC6 inhibitory activity assay protocols. All compounds are dissolved in DMSO. Serial dilutions of the compounds were carried out with 10% DMSO in HDAC assay buffer and 5 ml of the dilution was mixed to a 50 ml reaction so that the final concentration of DMSO is 1% in all of the reactions. Duplicate enzymatic reactions were conducted at 37 C in a 50 ml mixture containing HDAC assay buffer, 5 mg BSA, HDAC substrate 3, HDAC6 (cata-logue# 50006), and a test compound (10 mM) for 30 to 60 min. 50 ll of 2 Â HDAC Developer was added to each well after enzymatic reaction, and the plate was incubated at room temperature for another 15 min. Fluorescence intensity was measured using a Tecan Infinite M1000 microplate reader at an excitation of 360 nm and an emission of 460 nm 5.2.2.3. Data analysis. Duplicates of HDAC activity assays were performed at each concentration. Analysis of the fluorescent intensity data was performed using the Graphpad Prism computer software., the % activity was calculated according to the equation: % activity ¼ (F-F b )/(F t -F b ), where F ¼ the fluorescent intensity in each data set in the presence of the compound, F t ¼ the fluorescent intensity in each data set, defined as 100% activity in the absence of the compound, F b ¼ the fluorescent intensity in each data set, defined as 0% activity in the absence of HDAC.

In vitro anticancer activity
All of the synthesised compounds (20) were submitted to NCI and selected for in vitro anticancer assay against a panel of 60 cancer cell lines. The methodology of assay and calculations is as reported [54,55].