Antitumour potential of BPT: a dual inhibitor of cdk4 and tubulin polymerization

The marine natural product fascaplysin (1) is a potent Cdk4 (cyclin-dependent kinase 4)-specific inhibitor, but is toxic to all cell types possibly because of its DNA-intercalating properties. Through the design and synthesis of numerous fascaplysin analogues, we intended to identify inhibitors of cancer cell growth with good therapeutic window with respect to normal cells. Among various non-planar tryptoline analogues prepared, N-(biphenyl-2-yl) tryptoline (BPT, 6) was identified as a potent inhibitor of cancer cell growth and free from DNA-binding properties owing to its non-planar structure. This compound was tested in over 60 protein kinase assays. It displayed inhibition of Cdk4-cyclin D1 enzyme in vitro far more potently than many other kinases including Cdk family members. Although it blocks growth of cancer cells deficient in the mitotic-spindle checkpoint at the G0/G1 phase of the cell cycle, the block occurs primarily at the G2/M phase. BPT inhibits tubulin polymerization in vitro and acts as an enhancer of tubulin depolymerization of paclitaxel-stabilized tubulin in live cells. Western blot analyses indicated that, in p53-positive cells, BPT upregulates the expression of p53, p21 and p27 proteins, whereas it downregulates the expression of cyclin B1 and Cdk1. BPT selectively kills SV40-transformed mouse embryonic hepatic cells and human fibroblasts rather than untransformed cells. BPT inhibited the growth of several human cancer cells with an IC50 <1 μM. The pharmacokinetic study in BALB/c mice indicated good plasma exposure after intravenous administration. It was found to be efficacious at 1/10th the maximum-tolerated dose (1000 mg/kg) against human tumours derived from HCT-116 (colon) and NCI-H460 (lung) cells in SCID (severe-combined immunodeficient) mice models. BPT is a relatively better anticancer agent than fascaplysin with an unusual ability to block two overlapping yet crucial phases of the cell cycle, mitosis and G0/G1. Its ability to effectively halt tumour growth in human tumour-bearing mice would suggest that BPT has the potential to be a candidate for further clinical development.

A link between development of human cancers and cellular pathways where the retinoblastoma protein (pRb) has a major role is well established. 1,2 One of the frequent events associated with human tumour progression is abnormality in the pathway that links pRb, p16 INK4A , cyclin D1 and Cdk4 (cyclin-dependent kinase 4). 3 Cdk4 along with its activating cyclin partner D1 has a key role in cell cycle control. 4,5 The naturally occurring inhibitor of Cdk4-cyclin D1, p16 INK4a (p16), is a tumour supressor protein. Deletion or inactivating mutations in the p16 gene are observed in many human cancers. 6,7 The catalytic activity of Cdk4 depends on its activation by the protein cyclin D1, which is expressed during the G 0 /G 1 phase of the cell cycle. Many cancers are characterised by abnormal overproduction of cyclin D1. 8,9 As Cdk4 inhibitors target a pathway that links pRb, p16INK4A, cyclin D1 and Cdk4, it makes inhibition of Cdk4-cyclin D1 enzyme a crucially important target for cancer chemotherapy. [10][11][12][13] However, Rb mutations, consistent with loss of Rb function, have been identified in a wide spectrum of tumours including osteosarcomas, small-cell lung carcinomas, breast carcinomas and others, and the Cdk4 inhibitors cannot inhibit such pathway involving Rb-mutated tumours.
A number of potential anticancer agents that selectively modulate the activity of Cdk4-cyclin D1 in vitro have been reported. 14 These molecules also show the genotypic consequences of Cdk4 enzyme inhibition at the cellular level, that is, growth inhibition of cancer cells in vitro, arrest of asynchronous cells at G 0 /G 1 and prevention of pRb phosphorylation at Cdk4-specific serine residues. [15][16][17] Usually, competing with ATP molecules for binding at the protein kinase active site is the normal mechanism by which most small molecules inhibit kinase enzyme activity. Successful attempts to identify selective Cdk4 inhibitors using structure-based chemical design and molecular modelling have been reported. [18][19][20] Furthermore, the success of Cdk4 inhibitors at clinical IIIM Publication Number: IIIM/782/2015. 1 stages [21][22][23][24][25] has indicated it as a promising therapeutic target for anticancer drug discovery. 14 Fascaplysin (1), a natural product originally isolated from a marine sponge, specifically inhibits the Cdk4 enzyme. 26,27 It inhibits Cdk4-cyclin D1 with an IC 50 of~0. 35 μM and blocks growth of cancer cells at the G 0 /G 1 phase of the cell cycle. Similar to cryptolepine and ellipticine, 28 fascaplysin is also a planar structure and thus it intercalates double-stranded (d-s) DNA and shows unusual toxicity at the cellular level. It has been suggested that fascaplysin's planar structure is the possible explanation for its ability to intercalate d-s DNA and also its unusual toxicity at the cellular level. To overcome this unusual toxicity, recently we reported CA224 (2), a non-planar analogue of fascaplysin exhibiting selective Cdk4 inhibition with no DNA-intercalating property. 29 In continuation to these efforts, herein we report identification of tryptoline-based compounds CA198 (3), CA199 (4), CA211 (5) and N-(biphenyl-2-yl)-tryptoline (BPT, 6) as selective Cdk4 inhibitors with no DNA-intercalating property. Based on the molecular modelling design, a number of non-planar analogues of fascaplysin were synthesized. They show specificity towards Cdk4-cyclin D1 enzyme activity and blocks the growth of cancer cells at the G 0 /G 1 phase. Although BPT was also designed using a homology model of Cdk4, based on the X-ray crystallographic structures of Cdk2, Cdk4 and Cdk6, [30][31][32][33] further investigations showed that BPT blocks growth of cells at the G 2 /M phase, in a Cdk-independent manner, through inhibition of tubulin polymerization. BPT shows potent cytotoxicity in a panel of cancer cells and is efficacious against human tumours derived from HCT-116 and NCI-H460 cells in SCID mice models. Here, we present the biological activity of BPT in detail. BPT (6) was synthesized using a one-step procedure by coupling tryptoline with biphenyl 2-carboxylic acid. The chemical synthesis of BPT and chemical structures of 1-6 are shown in Figure 1.
To understand the observed selectivity towards Cdk4-cyclin D1 versus Cdk2-cyclin A, molecular modelling studies were performed. 33 These two Cdks share 45% sequence homology; however, they differ by three peptidic sequences including 94-97 (Glu-His-Val-Asp) Cdk4 /81-84 (Glu-Phe-Leu-His) Cdk2 , 101-102 (Arg-Thr) Cdk4 /88-89 (Lys-Lys) Cdk2 and Glu144 Cdk4 /Gln131 Cdk2 . BPT interacts with ATP-binding pocket of Cdk4-cyclin D1 with 83-fold selectivity with respect to Cdk2-cyclin A because of flexible conformational movement of the BPT amide bond, which allows free rotation of biphenyl ring leading to subsequent gain or loss of major hydrophobic interactions with one or other Cdk. BPT interacts with these Cdks in two different conformational states: (a) in cis conformation (green-coloured ligand in Figure 1c, Ψ = -9.6), it interacts selectively with the side chain of Arg101 residue of Cdk4-cyclin D1 by hydrophobic π-cation interaction, whereas in Cdk2-cyclin A, this interaction is missing as the corresponding Lys88 residue side chain orients away from BPT-binding cavity, and (b) in trans conformation (orange-coloured ligand in Figure 1c; Ψ = 161.1), it interacts with Cdk2-cyclin A.
DNA-binding assay. The ability of BPT to intercalate d-s DNA was important to segregate it from fascaplysin. It was studied by ethidium bromide (EtBr) dispacement assay and topoisomerase-I catalysed DNA relaxation or unwinding assay in vitro. Unlike fascaplysin, BPT failed to displace bound EtBr from DNA, indicating that it does not compete with EtBr-binding sites on DNA (EtBr is known to bind to the minor groove of d-s DNA, and also to DNA double helix and crosslinking sites 34 ) and therefore shows no detectable affinity towards DNA (Table 1 and Supplementary Figure  S5C). The interactions of fascaplysin, actinomycin D and BPT with pBluescript plasmid DNA is depicted in Supplementary Figure S5C. As indicated in Supplementary Figures S5A and C, 100 μM of BPT was incapable of displacing 1.3 μM of EtBr from pBluescript plasmid DNA. In contrast, DNA-intercalating agents actinomycin D and fascaplysin readily dislodges the bound EtBr. Usually, DNA intercalators hinder topoisomerase-catalysed DNA relaxation/unwinding process. Fascaplysin at low dose also hinder topoisomerase-I catalysed DNA relaxation, whereas BPT at much higher concentration than fascaplysin does not hinder DNA relaxation process, as indicated by two different bands on agarose gel (Supplementary Figure S5A). To insure that these results reflected a lack of DNA intercalation rather than an inhibition of topoisomerase-I enzyme activity, a second set of experiment was conducted using relaxed (i.e. negatively supercoiled) pBluescript plasmid DNA as initial substrate. BPT-treated negatively supercoiled DNA remained relaxed even after treatment with as high as 100 μM concentration, whereas fascaplysin-treated negatively supercoiled DNA does not remain intact at low dose (1 μM), clearly indicating the DNA-intercalating nature of the fascaplysin (Supplementary Figure S5B).
Cancer cell growth inhibition. Cancer cell growth inhibition data for BPT (6) and its structural analogues 3-5 is presented in Table 2. These compounds were tested in vitro in 10 cancer cell lines known to be relatively resistant to known chemotherapeutic agents. 35 The inhibitory effects of compounds were quantified using MTT assay. The results of cell proliferation assays indicate that BPT inhibits the growth of cancer cells in vitro at submicromolar concentrations. Among all the analogues, BPT was found to be the most potent compound at the cellular level.
At its IC 50 concentration, BPT arrests asynchronous cancer cells at the G 2 /M phase of the cell cycle (in p53 + A549 and p53 -NCI-H1299 cells). When A549 (p53+) cells were treated with BPT (IC 50 concentration for 24 h), a profound block at the G 2 /M phase was observed with 82% cells appearing to be in the G 2 /M phase (Figure 2b). At the IC 70 concentration, 59% of cells were blocked at the G 2 /M phase and 13% of cells appeared as apoptotic, whereas 15% of cells remained in the G 0 /G 1 phase (Figure 2c). These results indicate that only at higher concentrations, BPT tends to act as a Cdk4 inhibitor blocking cells at the G 0 /G 1 phase. At lower concentrations, greater tendency towards G 2 /M block is observed. Incubation of NCI-H1299 (p53-null) cells with the IC 50 concentration of BPT resulted in a large   Figure 2g), at a stage of the cell cycle where Cdk2-specific inhibitors normally act. When released in the presence of BPT, cells proceed from the G 1 /S phase, confirming that BPT does not inhibit cellular Cdk2. Cells ultimately accumulate at the G 2 /M phase (72%; Figure 2m). Cells released in the fresh medium enter the cell cycle ( Figure 2k). These results again indicate that BPT has an inherent tendency to induce block at the G 2 /M phase.
BPT selectively induces apoptotic cell death in SV40-large T-antigen-transformed normal mouse embryonic liver cells. SV40-large T-antigen inactivates both the tumour suppressor proteins p53 and pRb, and thereby transforms normal cells into tumorigenic cells. The effect of BPT on normal mouse embryonic hepatic (liver) cells BNL CL2 and its SV40-large T-antigen-transformed counterpart BNL SV A.8 was investigated. More than 50% normal cells appeared in the G 2 /M phase of the cell cycle upon 48 h incubation with IC 50 and IC 70 concentrations of BPT; however, only o10% cells were detected in the sub-G 1 phase ( Figure 2o). Interestingly, in the SV40-transformed cell line, significant apoptotic cell death was observed, which is represented by the percentage of cells appearing in the sub-G 1 phase. After 48 h treatment with IC 50 concentration of BPT, 43% cells were detected in sub-G 1 (Figure 2q) phase. The percentage of apoptosis was increased to 51% at the IC 70 concentration. Long-term cell survival assay in mouse embryonic normal hepatic cell line (BNL CL2) and SV40mediated transformed mouse embryonic hepatic cell line (BNL SV A.8) after the treatment with BPT indicated that it selectively reduces the number of colonies in SV40-mediated  Effect of BPT on the levels of cyclin B1, Cdk1, p53, p21 CIP1/WAF1 (p21) and p27 KIP1 (p27) in p53+ cells.
Western blot analyses of p53+ A549 and LS174T cells, after treatment with BPT at the IC 50 concentration for 24 h,

NCI-H358
Hydroxyurea, 250 µM, 18 h Released in BPT, IC 50 , 18 h Antitumour potential of BPT S Mahale et al demonstrated more than 10-fold induction of p53, which corresponds with increase in levels of the Cdk global inhibitor p21 CIP1/WAF1 (p21). The levels of the other pan-Cdk inhibitor p27 KIP1 (p27) were also elevated owing to BPT treatment ( Figure 3). The proteins Cdk1 and cyclin B1 were downregulated in the treated cells as compared with untreated control cells (Figure 3).
Effect of BPT on the levels of cyclin B1, Cdk1, p53, p21 and p27 in mutant p53 cells. Western blot experiments were performed to ascertain the levels of cell cycle regulatory proteins in MIAPaCa cells, which contain p53 gene mutations (data not shown). Upon BPT treatment, p53, p21 and p27 levels remained unchanged, suggesting that the p21 and p27 induction is probably p53-dependent. The MiaPaca cell line has mutated p53 and proteins p21 and p27 are controlled by p53 pathway. The inhibition of cell growth by BPT was independent of the presence or absence of tumour suppressor protein p53, owing to which there was unchanged expression of these proteins after treatment with BPT. Interestingly, the cyclin B1 and Cdk1 levels were elevated and Tyr15 residue of Cdk1 remained dephosphorylated, indicating that Cdk1-cyclin B1 is still active in p53-mutated cells after BPT treatment. In p53+ cancer cells, BPT downregulates cyclin B1 and Cdk1 but upregulates the pan-Cdk inhibitory proteins p21 and p27. This provides a mechanistic insight into the block of cell growth explaining why p53+ cells undergo block at the G 0 /G 1 , G 2 and M phases of the cell division cycle. BPT upregulates cyclin B1 levels both in p53-mutant and p53-null cells (data not shown), indicating that in these cancer cells the major block occurs at a post-G 2 phase of the cell cycle. This is likely to happen in between the G 2 and the G 0 /G 1 phases of the cell cycle, suggesting that BPT somehow affects the functions of mitosis.
Inhibition of tubulin polymerization in vitro by BPT. BPT inhibits the polymerization of tubulin, which is concluded from the dose-dependent decrease in V max (mOD/min) and reduction in final polymer mass (Figure 4a). When tested at four different concentrations, BPT decreased the V max from 19 mOD/min to 12.5, 9.2, 3 and 0.5 mOD/min at 0.5, 1, 2.5 and 5 μM of BPT, respectively. As a consequence of decreased V max , up to 80% reduction in final polymer mass was observed. Interactions of BPT with tubulin are shown in Supplementary Figure S1.
Enhancement of tubulin depolymerization and inhibition of tubulin polymerization in the presence of BPT in live cells. A549 cells were used to explore the interactions between BPT and tubulin protein in live cells. As the mitoticspindle checkpoint in A549 cells is intact, these cells are sensitive to antimicrotubule agents. The assembled (cytoskeletal) and unassembled (cytosolic) forms of tubulin were determined, via western blotting, from their accumulation and disappearance from pellet and supernatant fractions of the cell lysates, after cells were treated with BPT. Paclitaxel treatment of A549 cells for 30 min resulted in the accumulation of cytoskeletal tubulin as a consequence of enhanced tubulin polymerization, whereas in the presence of BPT the paclitaxel-mediated polymerization is inhibited in a dosedependent manner (Figure 4b). More interestingly, when intracellular stabilized tubulin (paclitaxel-treated cells) was subjected to BPT treatment, BPT enhanced the tubulin depolymerization, resulting in the disappearance of cytoskeletal tubulin (pellet) form and accumulation of cytosolic tubulin (supernatant) form ( Figure 4b).
Clonogenic assay. The effects of BPT on colony formation efficiency of p53+ A549 and p53-null Calu-1 cells were investigated. Calu-1 cells, similar to A549, contain a functional pRb protein but, in contrast to A549 cells, have a defective mitotic-spindle checkpoint. 36 The concentration at which BPT prevented colony formation (in a 12-day experiment) was found to be relatively lower than the concentration at which it inhibited cell proliferation (48 h experiment). This could be because very few cells in the total population retain   Supernatant and pellet represent unassembled and assembled tubulin, respectively. Tubulin polymerization is detectable by the increase of tubulin in pellet and its disappearance from supernatant. The western blots show dose-dependent inhibition of tubulin polymerization after the simultaneous treatment of paclitaxel and BPT that resulted in the accumulation of unassembled tubulin in supernatant. BPT also acts as an enhancer for tubulin depolymerization in a dose-dependent manner when paclitaxel-stabilized tubulin was subjected to BPT treatment for 30 min the ability to maintain cell division cycles, which could eventually result in non-detectable colonies of cells. BPT shows significant reduction in the colony formation efficiency of both p53+ and p53-null cancer cells in vitro ( Figure 5).
In vivo experiments in mice: pharmacokinetics and determination of MTD. The pharmacokinetics of BPT was carried out in BALB/c mice at 10 mg/kg (per os) and 1.0 mg/kg (intravenous) dose, which showed AUC 0-t of 17.7 and 170 ng • h/ml, respectively. The PK parameters after intravenous dosing were: t 1/2 = 0.19 h, C max : 413 ng/ml, C 0 : 888 ng/ml, AUC 0-t = 170 ng • h/ml, AUC 0-∞ : 174 ng • h/ml, CL: 95.8 ml/min per kg and V d : 1.56 l/kg. Because of the poor oral bioavailability of BPT, we decided to perform in vivo efficacy via intraperitoneal route. The in vivo study to determine maximum-tolerated dose (MTD) was performed in Swiss-albino mice over 2 weeks. Loss in animal body weight was considered as a measure of overtoxicity for the test compound. The concentration of the compound at which Z10% weight loss was observed was determined and designated as MTD, although usually a weight loss, which is below 20% of the initial weight, is considered harmless as animals can recover once the treatment is stopped. The toxicity results obtained from these studies indicated that for BPT, the MTD in mice was~1000 mpk (milligrams per kilogram of body weight).  (Figures 6b-d).
In both tumour models, treated animals displayed statistically insignificant weight loss (Supplementary Figure S6).

Discussion
BPT, a non-planar analogue of fascaplysin, showed selective inhibition of Cdk4 with an IC 50 of 10 μM, showing no inhibition against other kinases including Cdks at 10 μM. It does not intercalate DNA, which makes it free from unusual toxicity of DNA-intercalating agents. In cellular assays, BPT displays potent cytotoxicity in several cell lines. Considering the p53 and pRb status of the cancer lines tested, it is clear that inhibition of cell growth was p53-and pRb-independent. BPT showed cytotoxicity in p53-null (PC-3, Calu-1, NCI-H1299, NCI-H358) as well as p53-positive cells (LS174T, A549, NCI-H460). Furthermore, it also showed cytotoxicity in pRbpositive as well as pRb-null cells, indicating that inhibition of cell growth was independent of the presence or absence of tumour suppressor proteins p53 and pRb. High potency of BPT in cell lines that lack pRb activity (i.e. NCI-H358 that is pRb-null) suggests that Cdk4 inhibition may not be the only cellular target of BPT. It exhibits G2/M block of NCI-H358 cells, which again indicates that BPT has an inherent tendency to induce block at the G2/M phase. The cell cycle studies indicated that BPT blocks the G 0 /G 1 phase of the cell division cycle as would be expected of a true Cdk4 inhibitor, but only partially. However, it profoundly blocks cells at the G 2 /M phase at comparatively low concentrations. G 2 /M arrest could be the result of cellular stress. In response to cellular stress, induction of the p53 protein may arrest cells at the G 2 /M phase. 37 However, in p53-null NCI-H358 cells, the prometaphase block induced by nocadozole or paclitaxel is maintained by BPT. The G2/M block seems to be p53-independent. The selective cell death induction in cancer cells by BPT is very significant. From the results of cell cycle (Figures 2n-q), percent cell death and apoptosis (Supplementary Figure S2), and colony formation assay (Supplementary Figure S3), it was also observed that BPT selectively induces apoptosis in SV40-large T-antigentransformed cells and not in -untransformed normal cells.
A number of compounds, which are particularly potent Cdk2 inhibitors, have been reported to induce apoptosis selectively in transformed cells. [38][39][40] Relatively low doses of celecoxib has been reported to induce G 2 /M arrest, followed by induction of apoptosis only in transformed cells but not in normal cells. Celecoxib also downregulates cyclin B1 and upregulates p21 expression independent of p53. 41 In addition to manifesting these properties of celecoxib, BPT treatment also increases the level of p53 expression. Western blot studies indicate that repression of cyclin B1 and Cdk1 and elevated levels of p21 and p27 is a possible explanation of the G 2 /M block seen in p53 tumour suppressor-proficient A549 and LS174T cells ( Figure 3). Minor alteration of the microtubule dynamics can arrest the cell cycle progression at mitosis and eventually result in apoptotic cell death. 42 Figure 6 (a) In vivo tumour growth inhibition curve for BPT in the SCID mice-HCT-116 xenograft model. Graphs depict tumour growth inhibition in a group of animals treated with BPTat the concentration 100 mpk, which is compared with the untreated group of animals (shown in the graphs as the control group). Tumour sizes were recorded at 2-5 day intervals. Tumour weight (in mg) was estimated according to the formula for a prolate ellipsoid: (length (mm) × (width (mm) 2 ) × 0.5) assuming specific gravity to be one and π to be three. (b) Tumour growth inhibition curves for BPT in the SCID mice-NCI-H460 xenograft model. Graphs depict tumour growth inhibition in a group of animals treated with BPT compared with the untreated group of animals (shown in the graphs as the control group). Tumour sizes were recorded at 2-5 day intervals. Tumour weight (in mg) was estimated according to the formula for a prolate ellipsoid: (length (mm) × (width (mm) 2 ) × 0.5) assuming specific gravity to be 1 and π to be 3. (c) The pictures of SCID mice showing NCI-H460 tumour growth inhibition followed by treatment with BPT at the concentration 100 mpk. The treatments were continued for 9 consecutive days intraperitoneally when tumour growth had reached about 4-6 mm in diameter after about 6 days following the tumour cell injection. (d) Pictures of NCI-H460 tumour tissues, from SCID mice, exhibiting tumour growth inhibition by BPT resistance of BPT to inhibit growth of p53-null Calu-1 cells, in which the mitotic-spindle checkpoint involving tubulin is known to be abnormal, 44 suggested that BPT may be an antimicrotubule agent. Therefore, the action of BPT on tubulin polymerization in vitro and in live cells was investigated (Figure 4a). BPT was found to inhibit polymerization and enhance depolymerization of tubulin. It is noteworthy to mention that previously discovered tubulin inhibitors are all relatively quite toxic in contrast to what we have discovered. As BPT shows dual mechanism of action at the cellular level, we explored the possibility of its potent cellular activity in reducing in vivo the long-term survival and colony formation efficiency of non-small-cell lung cancer (NSCLC) cells. BPT shows significant reduction in the colony formation efficiency of both p53+ and p53-null cancer cells. In in vivo efficacy studies, BPT showed significant antitumor activity at 1/10th of the MTD in HCT-116 and NCI-H460 xenograft models. In conclusion, a relatively non-toxic compound BPT with dual Cdk4/ tubulin polymerization inhibition activity and promising efficacy in in vivo tumour models has been identified.
EtBr displacement assay. The assays were performed in a 96-well plate with clear bottom (Costar, Washington, DC, USA). The assay involved the addition of 10 μl of 10 × concentrated stock solution of compounds (dissolved in DMSO and further diluted in EtBr displacement assay buffer) to 90 μl of reaction mixture containing 6 μg of purified pBlueScript DNA and 1.3 μM EtBr in EtBr displacement assay buffer with final pH 7.4. Equivalent amounts of DMSO were added to the vehicle controls. In addition to control samples (DNA+EtBr), test samples (DNA +EtBr+test compounds), blank 1 (EtBr only), blank 2 (DNA only), wells containing DNA and compound were also prepared to test any change in the background fluorescence readings. The reduction in relative fluorescence counts was monitored (λ excit = 260 nM; λ emiss = 600 nM) and recorded after 1 min equilibration time. Fascaplysin and actinomycin D, which are known to intercalate d-s DNA, were used as standard compounds in assay. The mean control and test readings were corrected by subtracting blank readings. The percentage fluorescence in the test samples in reaction with control samples was calculated by using following formula: % Fluorescence ¼ corrected mean RFU in control À corrected mean RFU in test ð Þ 100 corrected mean RFU in control Topoisomerase I catalysed DNA relaxation or unwinding assay in vitro. For assay, 5 nM supercoiled pBluescript d-s plasmid DNA and 10 U of topoisomerase-I active enzyme were used in each reaction well. To ensure that the assay determines the DNA-intercalating property of compounds and not topoisomerase I inhibition, in a parallel experiment relaxed plasmid DNA was first prepared by treating with topo I enzyme for 30 min and then used as an initial substrate for the assay. DNA relaxation assays were performed in the presence or absence of compounds in 40 μl of DNA unwinding assay buffer. After 30 min incubation at 37°C, reaction mixtures were treated with 3 μl of 250 mM EDTA and extracted with phenol/chloroform. The DNA was dissolved in tris-EDTA buffer, pH 8.0. The samples (20 μl) were treated with 2 μl of 2.5% SDS, mixed with 2.5 μl agarose gel loading buffer (10x) and subjected to electrophoresis on a 0.8% agarose gel without EtBr (separating the DNA in the presence of EtBr would convert the relaxed DNA into the supercoiled form). After the electrophoretic separation, DNA bands were stained with 1 μg/ml EtBr and visualized using a UV illuminator. The compounds were compared with fascaplysin, which is a known DNAintercalating molecule. Camptothecin, which is a known topoisomerase-I inhibitor, was used to test the activity and inhibition of enzyme.
In vitro cell proliferation assays. All 10 human cancer cell lines were maintained at 37°C in 5% CO 2 in RPMI-1640 medium, supplemented with 10% foetal calf serum and 100 μg/ml normocin. The 10 cancer cell lines used for screening were all obtained from ATCC (Manassas, VA, USA); they were the NSCLC (a form of cancer that is resistant to chemotherapy) lines: NCI-H460 (pRb + , p53 + ), A549 (pRb + , p53 + ), Calu-1 (pRb + , p53-null), NCI-H1299 (pRb + , p53-null), NCI-H358 (pRb-null, p53-null); the colon cancer line LS174T (pRb + , p53 + ); the prostate cancer line PC-3 (pRb + , p53-null); the pancreatic cancer line MiaPaca (pRb + , p53-mutant). The genotypes within brackets indicate the status of the tumour suppressor proteins pRb and p53. The mouse embryonic normal hepatic cell line (BNL CL2) and its SV40 large T-antigen-transformed counterpart cell line (BNL SV A.8) were also purchased from ATCC. The large T-antigen inactivates the tumour suppressor proteins p53 and pRb. The detailed procedure of cell proliferation (MTT) assay and IC 50 determination was described previously. 32 Flow cytometric analysis. The untreated (control) and treated (with test compounds) cells were harvested by trypsinization, washed once with PBS and then fixed in 70% chilled (−20 ºC) ethanol for minimum 1 h. After the fixation step, cells were centrifuged for 5 min at 3000 × g at room temperature, and the pellet was suspended in PBS containing 50 μg/ml propidium iodide (Sigma-Aldrich, Dorset, UK; cat. no. P-4170) and 0.5 mg/ml DNase-free ribonuclease (Sigma-Aldrich; cat. no. R-5503). The cells were stained for 1 h in the dark at 4°C. Cell cycle analysis was performed on the Beckman Coulter (Epics Altra) fluorescenceactivated cell sorter (Beckman Coulter UK Ltd, Buckinghamshire, UK). To gate all the events that represent single cells, and not cell doublets or cell clumps, the following analyses were performed on the samples. Cytograms of propidium iodide fluorescence peak signal versus integrated fluorescence or the linear signal were plotted. All data points on the straight line were isolated in a single gate and the gated data was further used for plotting a histogram that represents a complete cell cycle. The total number of events was not allowed to exceed 200 events per s. Data acquisition was stopped after a minimum of 10 000 events had been collected.   Tris-HCl (pH 6.8)) supplemented with 2 mM phenylmethylsulfonyl fluoride, and a protease inhibitor cocktail (Sigma-Aldrich; cat. no. P8340) was added per well. After a short and vigorous vortex mixing, the cell lysates were incubated at room temperature for 5 min and then centrifuged at 16 000 r.p.m. for 10 min to separate the soluble and polymerized tubulin fractions. Each supernatant and pellet fraction was mixed with 10 × sample buffer, heated for 7 min at 95°C and resolved on 10% SDS-polyacrylamide gels. The resolved proteins were then subjected to western blotting (as described above) with a specific α-tubulin antibody B-7 (Santa Cruz Biotechnology; cat. no. sc-5286).
Cell-free tubulin polymerization assay in vitro. The purified tubulin was obtained commercially (Cytoskeleton Inc., Denver, CO, USA) and the polymerization assays were carried out according to the method described previously. 36 Tubulin polymerization assay is based on the adaptation of the original methods of Lee and Timasheff,46 who demonstrated that light is scattered by microtubules to an extent that is proportional to the concentration of the microtubule polymer. The resulting polymerization curves are representative of the three phases of microtubule polymerization, namely nucleation, growth and steady-state equilibrium. Paclitaxel and nocodazole were used in the assay as a known enhancer and inhibitor of tubulin polymerization, respectively. The ability of BPT to inhibit tubulin polymerization in vitro was determined according to the manufacturer's instructions. Briefly, tubulin protein (3 mg/ml) was polymerized in GTP buffer (80 mM PIPES, pH 6.9, 2 mM MgCl 2 , 0.5 mM EGTA, 10.2% glycerol and 1 mM GTP) in the presence of a range of BPT concentrations at 37°C in a temperature-regulated Biotech spectrophotometer (Potton, Bedfordshire, UK). The absorbance (at 340 nm) kinetics of 61 cycles for each sample was studied and the readings were recorded at an interval of 1 min. MTD finding studies for in vivo experiments. Swiss-albino mice were used to determine the MTD for the compound. BPT was weighed and mixed with 0.5% (w/v) carboxymethylcellulose and triturated with Tween-20 (secundum artum) with gradual addition of water to make up the final concentration. Care was taken not to exceed 40.25% of Tween-20 in the final formulation of the BPT. In this study, six animals per group were administered with BPT at different doses for 5 days (Q1D × 5) via intraperitoneal route. Animals were monitored for weight loss, morbidity symptoms and mortality up to 2 weeks by the end of treatment. Significant weight loss was considered when mean animal weight dropped by ⩾ 10% and was considered highly significant when the drop was ⩾ 20%. When the tumour growth reached about 4-6 mm in diameter (about 6 days), the animals were randomly divided into eight groups, each containing 6 or 7 mice. The treatments were continued for 9 consecutive days intraperitoneally. Flavopiridol (2.5 mpk) was used as a positive control in this study.

Efficacy study in SCID mice
Tumour weight measurements: Tumour size was recorded at 2-5 day intervals. Tumour weight (mg) was estimated according to the formula for a prolate ellipsoid: (length (mm) × width (mm) 2 ) × 0.5) assuming specific gravity to be one and π to be 3. Tumour growth in compound treated animals is calculated as T/C (treated/ control) × 100% and growth inhibition percent (% GI) was (100 − % T/C). [48][49][50] Body weight measurements: The body weights of animals in different treatment and control groups were monitored by taking the measurements daily during the treatment schedule. By considering the body weight at the start of the treatment as 100%, the percent weight loss was calculated on subsequent days of treatments.
Statistical analysis: Data from each experiment was analysed by Microsoft Excel 2000. Statistically significant differences were identified and analysed using Student's t-test for multiple comparisons versus control group. [48][49][50] The experiments were performed by Piramal Life Sciences (Mumbai, India), on a service contract.
Molecular docking and molecular dynamic simulations. The available crystal structures of Cdk4/cyclin D1 are in the apo form and have several missing residues, and thus they cannot be used for molecular modelling. 33 In the present study, we have used a hybrid homology model of Cdk4/cyclin D1 described by Shafiq et al., 51 which was developed from the Cdk4/cyclin D apocrystal structure (PDB: 2W96) by incorporating positions of missing gaps and activation loops from Cdk2/cyclin A (PDB : 1FIN). 52 This hybrid homology model was subjected to protein preparation wizard for H-bond optimization, heterogeneous state generation, protonation and overall minimization. Grid file of docking was constructed using XYZ coordinates of the N atom of Val96 residue with a binding site of 12 Å radius grid box (X = -10.521, Y = 208.683, Z = 107.944). For Cdk2 docking, the Cdk2 apoprotein (PDB ID: 1FIN) was subjected to protein preparation wizard for filling missing loops and side chains (using prime), ionization, H-bond optimization, heterogeneous state generation, protonation and overall minimization. Grid file of docking was constructed using XYZ coordinates of the N atom of Leu83 residue with a binding site of 12 Å radius grid box (X = -10.406, Y = 209.105, Z = 107.576). For tubulin docking, the tubulin-colchicine complex (PDB ID: 1SA0) was retrieved from the protein data bank. 53 In this complex, protein is heterodimeric in nature, consisting of two α-chains (451 residues), two β-chains (452 residues) and the Stathmin-like domain (142 residues). Crystal structure was subjected to protein preparation wizard for filling missing loops and side chains (using prime), ionization, H-bond optimization, heterogeneous state generation, protonation and overall minimization. All other ligands, water and ions were removed except colchicine. Grid file for docking was constructed considering colchicine ligand as centroid of grid box of 10 Å size at interphase of α/β tubulin (C and D chains). All ligands were sketched in Maestro, prepared using ligprep and docked by Glide molecular docking software (Schrodinger LLC, Bangalore, India) in XP mode.
The Cdk4-BPT docked complex obtained from XP docking was subjected to system builder, in which TIP4P-Ew was used as an aqueous solvent model. The cubic box of 12 Å radius was used to define the core and overall complex was neutralized by adding one Cl − counter ion for simulation. Further this complex was minimized by steepest descent method followed by the Broyden-Fletcher-Goldfarb-Shanno algorithm with convergence threshold of 2.0 kcal/mol and overall 1000 iterations. MD simulations were carried out at normal temperature and pressure (300°K and 1.01325 bar, respectively). Thermostat and barostat method opted was langevin with ensemble pathway comprising NVT (constant number of particles, volume and temperature) and isotropic coupling method. Overall model system was relaxed before 10 ns simulation and coulombic interactions were defined by short-range cutoff radius of 9.0 Å and by long-range smooth particle mesh Ewald tolerance to 1e − 09.

Conflict of Interest
The authors declare no conflict of interest.