Novel synthesised flavone derivatives provide significant insight into the structural features required for enhanced anti-proliferative activity

With many cancers showing resistance to current chemotherapies, the search for novel anti-cancer agents is attracting considerable attention. Natural flavonoids have been identified as useful leads in such programmes. However, since an in-depth understanding of the structural requirements for optimum activity is generally lacking, further research is required before the full potential of flavonoids as anti-proliferative agents can be realised. Herein a broad library of 76 methoxy and hydroxy flavones, and their 4-thio analogues, was constructed and their structure–activity relationships for anti-proliferative activity against the breast cancer cell lines MCF-7 (ER +ve), MCF-7/DX (ER +ve, anthracycline resistant) and MDA-MB-231 (ER −ve) were probed. Within this library, 42 compounds were novel, and all compounds were afforded in good yields and >95% purity. The most promising lead compounds, specifically the novel hydroxy 4-thioflavones 15f and 16f, were further evaluated for their anti-proliferative activities against a broader range of cancer cell lines by the National Cancer Institute (NCI), USA and displayed significant growth inhibition profiles (e.g. compound-15f: MCF-7 (GI50 = 0.18 μM), T-47D (GI50 = 0.03 μM) and MDA-MB-468 (GI50 = 0.47 μM) and compound-16f: MCF-7 (GI50 = 1.46 μM), T-47D (GI50 = 1.27 μM) and MDA-MB-231 (GI50 = 1.81 μM)). Overall, 15f and 16f exhibited 7–46 fold greater anti-proliferative potency than the natural flavone chrysin (2d). A systematic structure–activity relationship study against the breast cancer cell lines highlighted that free hydroxyl groups and the B-ring phenyl groups were essential for enhanced anti-proliferative activities. Substitution of the 4-CO functionality with a 4-CS functionality, and incorporation of electron withdrawing groups at C-4′ of the B-ring phenyl, also enhanced activity. Molecular docking and mechanistic studies suggest that the anti-proliferative effects of flavones 15f and 16f are mediated via ER-independent cleavage of PARP and downregulation of GSK-3β for MCF-7 and MCF-7/DX cell lines. For the MDA-MB-231 cell line, restoration of the wild-type p53 DNA binding activity of mutant p53 tumour suppressor gene was indicated.


Introduction
Natural products and their derivatives provide a signicant number of new chemical leads for drug discovery programmes within a wide range of clinical areas. 1 For example, analysis of plant-based compounds has highlighted many potential clinical applications for avonoids. 2 Within the area of avonoids as anticancer agents, [3][4][5][6][7][8][9][10][11] the success of such approaches has been exemplied through the development of avone derivatives such as avopiridol (for chronic lymphocytic leukemia), [12][13][14] silibinin (for prostate cancer), 15 and quercetin 16 and its derivative QC12 17 (for ovarian cancer), which are at various stages of clinical application. Although the natural avonoids demonstrate anti-proliferative activities and can themselves act as useful leads, these activities have oen been reported at pharmacologically non-achievable high micromolar concentrations (50-100 mM). [18][19][20][21] Therefore, medicinal chemistry strategies are essential to optimise the pharmacokinetic proles of these compounds, in order to ensure that the full clinical applications for avonoids are realised. In this study, we have proposed methylation of hydroxyls, and conversion of 4-C]O to 4-C]S, as a promising strategy to improve the activities of avonoids, since these structural motifs could lead to avonoids with desirable physiochemical properties such as increased lipophilicity and metabolic stability. Therefore, the aim of this study was to elucidate the effects of methylation of free hydroxyls and 4-C]S substitution on the anti-proliferative properties of the avonoids by undertaking the systematic design, synthesis and biological analysis of a well-dened library of hydroxy avone analogues. In addition, synthesis of this library of compounds would allow structure-activity relationships (SARs) to be determined. This is of particular importance for avonoids as in-depth understanding of the SARs for anticancer properties of avones is still limited. 22,23 Specically, the signicance of some of their core structural features for activity, such as hydroxyl substitutions at different positions on the avonoid rings, and the effect of methylation of the free hydroxyls on activity, are poorly dened. Also, the impact of modifying the 4-C]O substituent on the anticancer activities of avones has not been extensively studied.

Design and synthesis
To realise the aims of this study, structurally related hydroxy and methoxy avones, and their 4-thio derivatives, were synthesised. The synthesis of the methoxy and hydroxy avone scaffolds was achieved using a synthetic strategy involving the Baker-Venkataraman rearrangement as a key step (Scheme 1). 24,25 The 4-thio analogues were obtained according to Scheme 2 26 in 20-30% overall yield (see Experimental section in ESI † material for full details).
First, the effects of methylation and 4-thio modications on avones with different numbers of hydroxyl groups and different positions of hydroxyls on ring-A, were investigated. Thus, 20 avones comprising of (i) dihydroxy avones with hydroxyl substitution on ring-A at the C-7,8 or C-5,7 positions, (ii) tetrahydroxy avones with dihydroxyl substitution on ring-A (at the C-7,8 or C-5,7 positions) and on ring-B (at the C-3 0 and C-4 0 positions), and (iii) pentahydroxy avones with dihydroxyl substitution on ring-A (at the C-5,7 position only) and on ring-B (at the C-3 0 and C-4 0 positions) as well as a hydroxyl group at the C-3 position on ring-C were designed and synthesised. The analogous methoxy and 4-thio derivatives were also prepared (compounds-1c-f to 5c-f, series-1, Fig. 1).
The pentahydroxy avone (quercetin) derivatives (5c-f) were obtained starting from pentahydroxy avone (purchased) according to Scheme 3 27,28 in 36% overall yield and with 99% purity by HPLC. This represents an improvement on the recently reported synthesis of 5f, both in terms of overall yield and purity 29 (22% overall yield and 95% purity). The panel of avones studied herein also contained the well-known natural avones chrysin (2d), luteolin (4d) and quercetin (5d) to allow comparisons to be made between the novel analogues prepared in our laboratory and some standard avonoids. The design of further series of avones was guided by the anti-proliferative activities of series-1 compounds. Therefore, compound-2f, a dihydroxy 4-thio avone that was identied to possess promising anti-proliferative activity in in vitro screening (discussed below), was considered as a candidate for further optimisation through the bioisostere approach and Topliss scheme. 30 This resulted in the synthesis of a focused set of dihydroxy avone analogues (bearing C-7,8 and C-5,7 hydroxyls on ring-A) with phenyl bioisosteres (series-2, Fig. 1), and phenyl groups with electron-withdrawing substituents at the C-4 0 position (based on the Topliss scheme for the optimisation of Scheme   phenyl ring substitution in drug design 30 ) (series-3, Fig. 1) as ring-B. The series-2 bioisosteric analogues of 2f comprised of dihydroxy and dimethoxy avones, as well as their 4-thioavone derivatives, with a ve-membered 2-thienyl and 2-furanyl group, and a six-membered 3-pyridyl group [compounds 6c-f to 11c-f]. Series-3 involved 7,8 and 5,7 dihydroxy, dimethoxy avones, and their 4-thio avone analogues with electron-withdrawing groups (EWGs) such as uoro (-F), chloro (-Cl), bromo (-Br) and cyano (-CN) at the C-4 0 position of the phenyl ring (ring-B) [compounds 12c-f to 19c-f]. It is noteworthy that for the synthesis of novel avones containing nitrogen atoms (e.g. the 3-pyridyl and 4-cyano phenyl groups), during the cyclisation step (step (iii-a)) in the presence of concentrated sulfuric acid in glacial acetic acid, the 1,3-diketone intermediates (compound 10b, 11b and 19b) were found to undergo hydrolysis instead of cyclisation and dehydration, to yield the corresponding methoxy avones. Therefore, the cyclisation of such 1,3-diketones was instead accomplished in high yields (70-80%) by the addition of concentrated sulfuric acid into the solution of the 1,3-diketone in CHCl 3 at 0 C 31 (step (iii-b)).

Anti-proliferative evaluation
As avonoids have been reported to show activity against breast cancer cell lines, 32-34 all of the synthesised compounds (1c-f to 19c-f) were evaluated for their anti-proliferative activity against the estrogen-responsive breast cancer cell lines MCF-7, and the clinically relevant anthracycline-resistant MCF-7/DX (which mimics the multi-drug resistant scenario in patients), as well as the estrogen-independent, triple negative breast cancer cell line MDA-MB-231, using the MTT assay. The initial series of compounds (series 1, i.e. 1c-f to 5c-f) was rst evaluated against MCF-7 and MCF-7/DX cell lines to investigate the effects of methylation and 4-thio substitution on avones with differing numbers of hydroxyl groups as well as with different positions on the A-ring. The IC 50 values (half maximal inhibitory concentration) of these compounds were determined as a measure of their respective cytotoxicity and these values are tabulated in Table 1. For this series, compound-2f with the dihydroxy groups at the C-5,7 positions on ring-A, and 4-thio substitution, showed the highest anti-proliferative potency with IC 50 values of 7.1 AE 0.51 mM and 34.93 AE 5.75 mM against MCF-7 and MCF-7/DX cell lines, respectively. Compounds 1f, 3e, 3f, 4d and 5d showed moderate cytotoxicity with IC 50 values of 25.6 AE 1.26 mM, 23.5 AE 4.27 mM and 11.8 AE 1.79 mM, respectively however, these compounds were weakly anti-proliferative against the MCF-7/DX cell line with IC 50 values of 176.3 AE 1.61 mM, >250 mM and >250 mM, respectively. Compounds 1c, 1d, 1e, 9c, 3c, 3d, 4c, 4e, 5c, 5e and 5f were found to be less active against both MCF-7 and MCF-7/DX cell lines with IC 50 values >30 mM and >85 mM, respectively. The IC 50 values for the wellknown natural avones in this series, 2d (chrysin), 4d (luteolin) and 5d (quercetin) were found to be 25.6 AE 1.26 mM, 21.6 AE 0.81 mM and 13.7 AE 0.61 mM, respectively, against the MCF-7 cell line and the obtained values are in the range of their literature reported values. 35,36 In general, all compounds exhibited a 3-10 fold lower anti-proliferative activity against MCF-7/DX, which could be because these compounds might act as substrates of the efflux pump P-gp, which is generally over expressed in the MCF-7/DX cell line. 37 Next, series 1 compounds (1c-f to 5c-f) were assessed for their anti-proliferative activities at 10 mM concentration against an ER Àve hormone-independent breast cancer cell line, MDA-MB-231. The 4-thio derivatives, and in particular the hydroxy compounds, exhibited greater anti-proliferative activity than the methoxy and hydroxy avone analogues. However, no derivatives reached the IC 50 at 10 mM (see ESI, Fig. S1 †).
As compound-2f was identied as a promising lead compound, further novel analogues of compound-2f were prepared to discover more active avone derivatives. Thus a series of bioisosteric analogues of the B-ring phenyl group of compound-2f (series-2, compounds 6c-f to 11c-f), and a series of compound-2f analogues with electron-withdrawing substituents at the C4 0 position (series-3, compounds 12c-f to 19c-f), along with their methoxy (both 4-C]O and 4-C]S) and their 7,8-regioisomeric analogues (both 4-C]O and 4-C]S), were synthesised and evaluated for their anti-proliferative activities. Since compound-2f was identied to be active below 10 mM and 50 mM against MCF-7 and MCF-7/DX, respectively, these subsequently synthesised compounds were initially assessed for their anti-proliferative activities at 10 mM concentration against the MCF-7 and MDA-MB-231 cell lines, and at 50 mM against the MCF-7/DX cell line. Compounds that reduced the cell viability below 50% in the initial screening were arbitrarily classied as 'active' and they were then evaluated at a range of concentrations (between 0.1 and 100 mM) against the respective cell line to determine their IC 50 values.
In the initial screening, series-2 compounds with bioisosteric replacements showed a decreased/complete loss of activity with cell viability > 50% (see ESI, Fig. S2 †). Among series-3, the novel compounds 14f and 16f were found to be active against all three cell lines, and novel compounds 13f, 15f, 17f and 19f were found to be active against both the MCF-7 and MCF-7/DX cell lines. Novel compounds 12f and 18f were found to be active only against the MCF-7 cell line (Fig. 2). These active compounds were further tested against the respective cell lines at different concentrations ranging from 0.1 to 100 mM to determine their IC 50 values. The determined IC 50 values are listed in Table 2. Among these, novel compounds 15f (IC 50 ¼ 1.0 AE 0.1 mM and 9.1 AE 0.1 mM against MCF-7 and MCF-7/DX cell lines respectively) and 16f (IC 50 ¼ 4.9 AE 0.7 mM, 6.5 AE 0.4 mM and 8.9 AE 0.9 mM against MCF-7, MCF-7/DX and MDA-MB-231 cell lines respectively) were identied as active compounds.
Novel insights into structure-activity relationships (SARs). Comparison of the anti-proliferative activities of methoxy avones (both 4-C]O and 4-C]S) with their corresponding hydroxy avones illustrates that the methoxy avones were less anti-proliferative than their hydroxy analogues with the exception of compound-3e [7,8,3 0 ,4 0 -tetramethoxy 4-thioavone, IC 50 ¼ 11.8 AE 1.79 mM against the MCF-7 cell line]. For example, hydroxy avone 4d possesses an IC 50 value of 21.6 AE 0.8 mM, whereas the IC 50 value of its methoxy avone derivative-4c was found to be >250 mM. This highlights that the free hydroxyls are important for anti-proliferative activities (presumably via hydrogen bonding interactions with the potential biological cellular target).
The importance of free hydroxyls for anti-proliferative activity previously has been debated in the literature, with contradicting results. Thus, the results from our programme support ndings 38,39 that indicate that polymethoxylation decreases the anti-proliferative activity, but contradict earlier ndings that suggested higher anti-proliferative activities for polymethoxylated avones against MCF-7, 20,40 human leukemic HL-60 41 and melanoma cell lines. 19 Further, from the comparison of anti-proliferative activities between the hydroxy avones (free -OH and 4-C]O) and their 4-thio analogues (free -OH and 4-C]S), it was apparent that the substitution of 4-C]S for 4-C]O enhances the antiproliferative activity [compound-2f (free -OH and 4-C]S), IC 50 (a) Inuence of number and position of hydroxyls. By comparing the anti-proliferative activities of avone derivatives (free -OH and 4-C]O) with different numbers of hydroxy groups, it was observed that the anti-proliferative activities against the MCF-7 cell line were in the order pentahydroxy avone (5d) [IC 50 However, an opposite correlation was observed in the case of the hydroxy 4-thioavone derivatives (free -OH and 4-C]S), where the order of anti-proliferative activities was found  . This indicates that the number of hydroxyls is not a determining factor for the anti-proliferative activities, in the case of hydroxy avones (free -OH and 4-C]O). In complete contrast, the number of hydroxyl groups in hydroxy 4-thioavones plays an important role in their anti-proliferative activities and the dihydroxy 4-thioavones were found to be the most active avones.
In terms of the position of the hydroxyls, avones with 5,7-hydroxyls on the A-ring were more active than their corresponding 7,8-hydroxy avones. For example, 5,7,3 0 ,4 0 -tetrahydroxy avone (4d) [IC 50 (b) Effect of bioisosteric replacements. Replacement of the Bring phenyl group with a bioisostere (e.g. either a vemembered 2-thienyl, a 2-furanyl group, or a six-membered 3pyridyl group) resulted in a signicant loss of anti-proliferative activity. This highlighted that the presence of a phenyl group as ring-B is favourable for anti-proliferative activity.
(c) Inuence of electron-withdrawing groups. Interestingly, introduction of EWGs at the C-4 0 position of the phenyl group, mainly with halogens such as F (12f and 13f), Cl (14f and 15f) and Br (16f and 17f) was found to enhance anti-proliferative activities in comparison to the avones with an unsubstituted phenyl B-ring (compounds 1f and 2f). In particular, the introduction of Cl and Br groups increased the anti-proliferative potencies 2-7 fold (14f, 15f, 16f and 17f) against MCF-7 and MCF-7/DX cell lines, and also their introduction afforded 7,8dihydroxy 4-thioavones with an increased anti-proliferative activity towards MDA-MB-231 cells (compound-1f: cell viability at 10 mM was 79%, whereas, compound-14f (containing -Cl) and compound-16f (containing -Br) showed 49% and 20% cell viability at 10 mM against MDA-MB-231). The pronounced activity observed for the -F, -Cl and -Br substituted 4-thio-avones could be attributed to their increased lipophilicity, as can be observed by the decrease in IC 50 values (more active) for increasing theoretical log P values ( Table 2). A summary of SARs for the anti-proliferative activities of avones analysed herein is depicted in Fig. 3.
Molecular mechanism of action. Next, the probable molecular mechanism of the anti-proliferative activities of the avonoids was explored. Two plausible molecular mechanisms were postulated. (a) As the majority of avones evaluated herein exhibited a strong and selective anti-proliferative activity against the estrogen-responsive breast cancer cell line that expresses high levels of estrogen receptor-a (ERa), 42 it was anticipated that these compounds may act as estrogen-receptor antagonists by interacting with ERa. (b) As these compounds have been reported to interact with cell signalling proteins 4,43,44 it was hypothesised that these compounds could modulate the cell signalling proteins that are involved in cell survival and cell death.
(a) Interaction with ERa. The interaction of avones 1c-19f with the estrogen receptor-a (ERa) was rst investigated using a molecular docking approach. For this, compounds 1c-19f were docked sequentially into the binding site of the prepared X-ray crystallographic structure of human ERa complexed with an antagonist 2-phenyl-1-[4-(2-piperidin-1-yl-ethoxy)-phenyl]-1,2,3,4-tetrahydro-isoquinolin-6-ol [native ligand] (PDB code 1UOM, 2.28Å resolution) 45 using a validated docking procedure (see Experimental section). In general, no correlation was observed between the order of anti-proliferative activity and their binding affinities with ERa (as determined by their respective docking scores). One notable exception was compound-3e which showed higher binding affinity towards the ERa receptor [(docking score of 10.02, versus the native ligand docking score of 13.0), see ESI (Fig. S3 †) for details]. This suggests that the majority of the avonoid derivatives mediate their anti-proliferative effects via estrogen-independent mechanisms. A similar observation has been reported for 5-amino-avones, which showed remarkable anti-proliferative activity against the ER +ve MCF-7 cell line without ER competition, as determined by the ER binding assay. 46 To conrm this nding, an ERa antagonism assay was carried out using the human estrogen receptor alpha assay kit which consists of ERa reporter cells in which an ERa-responsive promoter is functionally linked to the luciferase reporter gene, and allows quantication of changes in the ERa activity via measurement of the changes in luciferase expression. In order to evaluate whether the avone derivatives mediate their antiproliferative effects via antagonism of ERa, compound-15f (IC 50 ¼ 1 mM) compound-16f (IC 50 ¼ 4.9 mM) and compound-3e (IC 50 ¼ 11.8 AE 1.79 mM), that were potent against MCF-7 cells, were assessed for their antagonist activity at four-point concentrations (10-0.1 mM) in the presence of a constant concentration of 17b-estradiol (3.2 nM, EC 75 ). The known antagonist fulvestrant was used as a positive control. Compounds 15f and 16f did not show any decrease in the luciferase activity (induced by 17bestradiol), which was in contrast to fulvestrant. However, compound-3e showed a dose-dependent decrease in the luciferase activity [see ESI (Fig. S4 †) for details]. These results support the ndings from the in silico study that the majority of these  compounds (except compound-3e) most likely act through an estrogen-independent mechanism. (b) Interactions with cell signalling proteins. Next, the signals mediated by compounds 15f and 16f in ER +ve MCF-7 and MCF-7/DX cells and ER Àve MDA-MB-231 cells were examined using the PathScan® Intracellular Signalling Array Kit, which allows simultaneous detection of 18 signicant and well-characterised signalling molecules that undergo covalent post-translational modications, such as phosphorylation and proteolysis (cleavage). Protein samples obtained from the treatment of MCF-7, MCF-7/DX and MDA-MB-231 cells with compounds 15f (IC 50 < 10 mM against MCF-7 and MCF-7/DX) and 16f (IC 50 < 10 mM against MCF-7, MCF-7/DX and MDA-MB-231 cells) were analysed (Fig. 4). The results showed that the anti-proliferative effects were mediated by signicant inhibition of the Akt associated downstream target GSK-3b, which is a key signalling molecule for cell survival and proliferation. In addition, compound-15f, which showed greater potency towards MCF-7 cells (IC 50 ¼ 1 mM for 15f vs. IC 50 ¼ 4.9 mM for 16f), was found to trigger cell death through PARP cleavage, which is a hallmark of the cell death pathway. 47 This explains the greater potency of compound-15f towards MCF-7 cells. It is noteworthy that no caspase-3 cleavage was observed; this suggests that the observed PARP cleavage is caspase-3 independent and the activation of PARP cleavage in MCF-7 cells might have been executed via other caspases such as caspases-6/9. 48 In the case of MCF-7/DX cells, compound-16f, that has greater activity towards MCF-7/DX cells than compound-15f, mediated its antiproliferative effects through a signicant reduction in the phosphorylation of several cell survival proteins such as mTOR, GSK-3b and Stat3, along with induction of cell death through caspase-3 and PARP cleavage. However, compound-15f was found to elicit its anti-proliferative effects through a reduction in the phosphorylation of GSK-3b as well as through induction of cell death via PARP cleavage. Thus, the higher potency of compound-16f could be related to its ability to downregulate multiple cell survival proteins (mTOR, GSK-3b and Stat3) along with its ability to induce cell death via caspase-3 and PARP cleavage.
Interestingly, treatment of MDA-MB-231 cells with compound-16f showed a 2-fold increase in the levels of p53 compared to the control. MDA-MB-231 cells have a mutant p53 gene 49,50 that is more stable and oncogenic unlike the wild type p53 in MCF-7 cells. 51 The presence of mutant p53 drives the invasiveness of the tumour. 52 which probably leads to the restoration of wild-type p53 DNA binding activity and induction of cell death. A similar observation has been reported for resveratrol (a chalcone) wherein cell death was induced by restoring the wild-type p53 DNA binding activity via the phosphorylation of a mutant p53 at serine-15 in prostate cancer DU145 cells. [54][55][56] This might explain the higher activity of compound-16f against MDA-MB-231 cells (IC 50 ¼ 8.9 mM for 16f vs. IC 50 > 10 mM for 15f).
Overall, these results correlate well with the in vitro ndings, and suggest that for the MCF-7 and MCF-7/DX cell lines, the anti-proliferative effects are mediated via ER-independent cleavage of PARP and downregulation of GSK-3b. For the MDA-MB-231 cell line, activity could be due to restoration of the wild-type p53 DNA binding activity of the mutant p53 tumour suppressor gene.
Apoptosis determination. As compounds 15f and 16f were found to act via induction of the apoptotic signalling pathway (PARP, caspase-3 and p53), these compounds were further assessed for their ability to induce apoptosis. For this, MCF-7 and MDA-MB-231 cells were treated (at their IC 50 concentration for 24 h) with compounds 15f (1 mM and 10 mM against MCF-7 and MDA-MB-231, respectively) or 16f (5 mM and 9 mM against MCF-7 and MDA-MB-231, respectively) and were Statistical significance was estimated with respect to the control (untreated sample) by one-way ANOVA, followed by Bonferroni's post hoc test (ns ¼ nonsignificant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). analysed for phosphatidylserine externalization (apoptosis marker) 57,58 using an annexin V-FITC/propidium iodide (PI) apoptosis detection assay and ow cytometry. As shown in with the cytotoxicity data/protein array data. Therefore, from the molecular mechanism of action studies, and the apoptotic assay, for compounds 15f and 16f it can be concluded that these compounds induce apoptotic cell death via PARP cleavage in MCF-7 cells, and 16f triggers apoptosis by p53 induction in the MDA-MB-231 cell line. NCI 60 cell line screening. Finally, to further probe the promising and signicant anti-proliferative activities of the novel avonoids 15f and 16f, these compounds were screened for in vitro cytotoxicity by the National Cancer Institute Developmental Therapeutic Program, 59 USA (NCI/DTP, USA). Initial evaluation was performed at a single dose of 10 mM against 60 different human tumour cell lines using the sulforhodamine B cytotoxic assay. 60 Both compounds showed potential growth inhibition against a wide range of cancer cell lines (see ESI †). As a result, these compounds were selected for further 5-dose assay screening against the 60 cell lines (0.01-100 mM). Three end points, specically the GI 50 (concentration that causes 50% growth inhibition), TGI (total growth inhibition) and LC 50 (the concentration of the drug at which the original cell number is reduced by 50%) were determined. The mean inhibitory doses (MID) against 60 cell lines for compound-15f and compound-16f were 2.81 mM and 2.39 mM respectively. The GI 50 values determined for these two compounds against 60 cell lines are presented in Table 3. Interestingly, both compounds 15f and 16f exhibited nanomolar range activity various cell lines. Specically, compound-15f showed the highest cytotoxicity against the breast cancer cell lines MCF-7 (GI 50 ¼ 0.18 mM), T-47D (GI 50 ¼ 0.03 mM) and MDA-MB-468 (GI 50 ¼ 0.47 mM), HCT-15 (colon cancer, GI 50 ¼ 0.48 mM), OVCAR-4 (ovarian, GI 50 ¼ 0.57 mM). Also, compound-16f, showed signicant cytotoxicity against NCI-H460 (NSCLC, GI 50 ¼ 0.97 mM) and HCT-15 (colon cancer, GI 50 ¼ 0.85 mM). The cytotoxicity prole of the wellknown avone chrysin (2d) [NSC407436] was retrieved from the NCI database (mean dose graph provided in ESI †) for comparison. Overall, compounds-15f and 16f were found to be 7-8 fold more potent than chrysin (MID-20.10 mM, against 51 cell lines). In particular, compounds-15f and 16f were found to be 46 fold and 13 fold more active than chrysin against an ovarian OVCAR-4 cell line, 43 fold and 24 fold more active than chrysin against colon HCT-15 and 5 fold and 15 fold more active than chrysin against the NSCLC NCI-H460 cell line, respectively (Fig. 6).

Conclusions
A library of 76 compounds containing structurally related methoxy and hydroxy avones, and their 4-thio analogues, has been designed, synthesised and evaluated for anti-proliferative activity against the breast cancer cell lines MCF-7 (ER +ve), MCF-7/DX (ER +ve, anthracycline resistant) and MDA-MB-231 (ER Àve). Within this library, 42 compounds were prepared and characterised for the rst time within our laboratory. The study provided signicant insight into the structural features required for enhancing the anti-proliferative proles of avones, and identied two novel hydroxy 4-thioavones 15f and 16f, as lead anti-proliferative agents (15f (IC 50 ¼ 1.0 AE 0.1 mM and 9 AE 0.1 mM against MCF-7 and MCF-7/DX cell lines, respectively) and 16f (IC 50 ¼ 4.9 AE 0.7 mM, 6.5 AE 0.4 mM and 8.9 AE 0.8 mM against MCF-7, MCF-7/DX and MDA-MB-231 cell lines, respectively)). A systematic SAR study highlighted the presence of free hydroxyl groups, and the B-ring phenyl groups, as essential for enhanced anti-proliferative activities. Replacement of the 4-C]O functional group with the 4-C]S functional group also enhanced the anti-proliferative activities. Incorporation of lipophilic electron withdrawing groups at C-4 0 of the Bring phenyl was found to be favourable, and the increased  GI 50 values presented here involve a correction for the cell count at time zero. Therefore the GI 50 value for a test drug corresponds to the concentration [(T À T 0 )/(C À T 0 )] Â 100 ¼ 50, where T ¼ absorbance of the test well aer 48 h exposure to the test compound, C ¼ absorbance of control aer 48 h and T 0 ¼ absorbance at time zero.

Molecular docking
Docking validation. To validate the accuracy of the docking procedure to be used, the original ligands were extracted from the coordinate les (taken from the Protein Data Bank PDB), and then docked again into the corresponding crystal structure of the proteins, using the automated docking procedure in the program Surex-Dock (SFXC), 62,63 as provided by SYBYL-X-2.1. The resulting ligand conformation from the docking procedure was compared with the ligand conformation as found in the actual crystal structure of the complex. Comparative structural orientation of the ligand was calculated as the root mean square deviation (RSMD) between the docked ligand and the ligand as found in the crystal structure, using the programme LSQKAB, as provided in the CCP4 64 suite. If the root mean square deviation (RMSD) value between the real and the bestscored conformations is equal to or less than 2.0Å (representing the grid spacing used for the docking procedure), then the docking process was considered successful. 65 In this case the docking procedure was rst validated by re-docking the extracted co-crystallised ligand of human estrogen receptora (ERa) (2-phenyl-1-[4-(2-piperidin-1-yl-ethoxy)-phenyl]-1,2,3,4tetrahydro-isoquinolin-6-ol) into the prepared target protein to be used for docking. The RMSD between the docked conformation, as generated by the docking algorithm and the native co-crystallised ligand conformation was found to be 0.12Å, which was well within the 2Å grid spacing used in the docking procedure, indicating that the docking method to be used was reliable and valid. Furthermore, the interactions between the docked ligand and the prepared target receptor mimicked those observed in the crystal structure of the same.
Docking procedure. Docking studies were performed using the programme Surex-Dock (SFXC) 62 as provided by Sybyl-X 2.1. The X-ray crystallographic structures of human estrogen receptor-a (ERa) complexed with an antagonist 2-phenyl-1-[4-(2piperidin-1-yl-ethoxy)-phenyl]-1,2,3,4-tetrahydro-isoquinolin-6-ol (PDB code-1UOM, 2.28Å resolution) 45 was retrieved from the Protein Data Bank. The protein structure was prepared for docking using the Biopolymer Structure Preparation Tool with the implemented default settings provided in the SYBYL programme suite. Hydrogens were added to the protein structures in idealised geometries and an overall energy minimisation of each protein was performed using the MMFF94 force eld, employing a conjugate gradient algorithm 66 with a convergence criterion of This journal is © The Royal Society of Chemistry 2016 RSC Adv., 2016, 6, 64544-64556 | 64553 0.5 kcal mol À1 A and up to 5000 iterations. Finally, before the docking run, all water molecules were removed and the ligand 2-phenyl-1-[4-(2-piperidin-1-yl-ethoxy)-phenyl]-1,2,3,4-tetrahydroisoquinolin-6-ol was extracted from the coordinate le of the ERa (PDB-1UOM). The protomol, representing the ligand binding groove, was generated using a ligand directed method, which allows the docking of ligands into predened sites, as dened by occupancy of any co-crystallised ligand at the site of interest.
The Surex-X docking algorithm docks a given ligand to a receptor using a exible ligand and a semi-exible receptor; in this case the peptides were allowed to be fully exible while the receptor was semi-exible. This approach allows for optimisation of potentially favourable molecular interactions, such as those dened by hydrogen bond and van der Waal forces. The docking results yield a docking score, which takes into consideration entropic, polar, hydrophobic, repulsive and desolvation factors. Here, the docking scores were expressed in Àlog 10 (K d ) units to represent binding affinities, where K d is the dissociation constant. The free energy of binding of the ligand to the protein was extrapolated from eqn (1).

Free energy of binding
The docking results were visualised using the programme PyMOL 67,68 and the molecular interactions of the docked ligands were analysed by the programme CONTACTS, as provided in the CCP4 suite of programmes. 64,69,70 Potential hydrogen bonds were assigned if the distance between two electronegative atoms was less than 3.3Å, whereas any separation greater than 3.3Å, but less than 4.5Å, was considered a van der Waal interaction.
ERa antagonist assay. ERa reporter cells consisting of an ERa-responsive promoter gene functionally linked to the luciferase gene were defrosted and seeded into a 96-well plate and these cells were immediately dosed with the test compounds at different concentrations (10-0.1 mM) and with 17b estradiol (3.2 nM, EC 75 concentration) according to the manufacturer's protocol. Aer 24 h incubation in the presence of the test compound or solvent (DMSO), the cell viability of these treated/untreated reporter cells was measured to eliminate false positives using the uorescence-based live cell multiplex (LCM) assay. The uorescence from the live cells was measured using the plate reader with the lter combination of [485 nm ex|535 nm em]. Following this, fold reduction in the luciferase intensity, which is the measure of the antagonist activity was measured by using a luminometer (TECAN) according to the manufacturer's protocol. The data are expressed as the fold reduction as compared to the control. The mean of three experiments and the standard error is reported.
PathScan sandwich immunoassay. The PathScan Intracellular Signalling array kit was used for the simultaneous detection of 18 signicant and well-characterised cellular proteins and signalling nodes that were phosphorylated or cleaved at the specic residues.
(a) Preparation of cell lysate. MCF-7 and MCF-7/DX cells (8 Â 10 4 cell per mL) and MDA-MB-231 cells (4 Â 10 4 cells per mL) were seeded into 24 well plates (2 mL per well) and incubated for 24 h. Aer 24 h, the cells were treated with the test compounds at 10 mM concentration for 24 h. Following this 24 h exposure, the cells were washed with ice-cold 1Â phosphatebuffered saline and lysed in 1Â cell lysis buffer provided (phosphotase and protease inhibitors added). These lysates were quantied using the BCA protein assay.
(b) Assay procedure. The array blocking buffer was added to each well of the glass slide provided and incubated for 15 min at room temperature. Subsequently, the cell lysate, diluted to 0.3 mg mL À1 in array diluents, was added to each well and incubated for 2 h at room temperature. Subsequent to washing, the detection antibody cocktail was added to each well and incubated for 1 h at room temperature. Horseradish peroxidase (HRP)-linked streptavidin was added to each well and incubated for 30 min at room temperature. The slide was then covered with LumiGLO/peroxide reagent (Cell Signaling Technology) and exposed to lm for 2-30 s. The image was captured by a digital imaging system, ImageQuant LAS 4000 (GE Healthcare).
Apoptosis detection assay. Apoptosis in MCF-7 and MDA-MB-231 cells was assayed by annexin V and propidium iodide (PI) costaining using an Annexin-V-FITC staining kit according to the manufacturer's instruction. 1 Â 10 6 MCF-7 and MDA-MB-231 cells were plated into a 6 well plate (3.4 Â 10 5 cells per mL, 3 mL per well). Aer 24 h incubation, cells were treated without and with test compounds (at their IC 50 concentration) for 24 h. Cells were harvested using Accutase™ cell detachment solution (1 mL per well) for 5 min at 37 C. Accutase was inactivated by addition of complete medium (2 mL). Cells were collected by centrifugation at 100 Â g and the pellet was washed twice with cold PBS and then resuspended in 1 mL of annexin-binding buffer. 100 mL of the cell suspension was transferred to 1.5 mL Eppendorf tube and 5 mL of FITC-conjugated annexin V and 5 mL of PI was added and incubated in dark for 15 min at room temperature. Aer 15 min incubation 400 mL of annexin binding buffer was added to the cells and the uorescence was measured using BD accuri™ instrument ow cytometry. The instrument was set for FL1 (annexin V-FITC) vs. FL3 (PI) bivariant analysis. Data from 10 000 cells per sample was collected and dot plots of FL1 vs. FL3 were generated. The quadrants were set based on the population of healthy, unstained cells in untreated samples compared to cells treated with a known apoptotic inducer camptothecin (5 mM) for 24 h. BD CSampler™ soware was used to calculate the percentage of the cells in the respective quadrants. The experiment was performed in triplicate.