Discovery of a Novel Small-Molecule Inhibitor that Targets PP2A–β-Catenin Signaling and Restricts Tumor Growth and Metastasis

Cancer is now the second major killer of human population as per WHO estimates for 2012, and by 2030, the global cancer burden is expected to rise to 21.7 million new cancer cases and 13 million cancer-related deaths (1). Cancer metastasis is the major reason of cancer-related morbidity and mortality. Loss of epithelial properties as well as acquisition of mesenchymal characteristics is one of the key features of early metastatic process (2). E-Cadherin and β-catenin together play a crucial role in initial dissemination of metastatic cells. Nuclear displacement of active β-catenin and its recruitment to the cell surface along with E-cadherin firms cell-to-cell adhesion and prevents an early release of prometastatic cells (3, 4). In the absence of E-cadherin, β-catenin is phosphorylated in the cytoplasm in a multiprotein destruction complex consisting of adenomatous polyposis coli (APC), Axin, glycogen synthase kinase 3β (GSK3β), casein kinase 1α (CK1α), and protein phosphatase 2A (PP2A), which ultimately leads to proteasomal degradation via E3 ubiquitin ligase SCF^β-TRCP (5–7). Therefore, nonphosphorylated form at Ser^33/37/Thr⁴¹ is an active form of β-catenin, and it regulates transcriptional activation of multiple protumorigenic target genes like c-MYC and CCND1 (8, 9) as well as forms adherence junctions with E-cadherin (10, 11). Kinases like CK1α and GSK3β are well known to control β-catenin phosphorylation, but a recent study by Zhang and colleagues clearly indicated that PR55α, a regulatory subunit of PP2A, specifically modulates PP2A-mediated β-catenin dephosphorylation and thus regulates its oncogenic function (12). So, diminution of active β-catenin (non-phospho form) from nucleus and cytoplasm but its recruitment to the cell surface could be a wonderful therapeutic window to harness its full-blown effect in regulating tumor growth and metastasis.

Novel drugs for controlling cancer metastasis are of great importance in current cancer therapeutics. Molecular hybridization is one of the emerging strategies for rational drug designing, where multiple targets can be engaged with a combination of two or more active groups to tackle both the tumor growth and metastasis (13). In addition, hybridization not only always increases the efficacy of the molecule but also plays a major role in controlling solubility and other drug-like properties of the molecule. In the continuation of our relentless efforts toward the identification of small molecules as promising anticancer therapeutics, previously we described a series of coumarin–monastrol, coumarin–chalcone and carboline–chalcone hybrids as potential antitumor agents (14–16). Recently, by applying the molecular hybridization approach, we discovered a novel dual-targeting MDM2 small-molecule inhibitor, which not only inhibited MDM2–p53 interaction but also promoted MDM2 degradation (17). In our current endeavor, we have synthesized a series of semicarbazone and chalcone hybrid and put forward strong evidence for not only its antiproliferative tumor cell–selective effect but also provided insight into its unique antimetastatic potential via mitigating PP2A–β-catenin signaling pathways.

DAPI, doxorubicin, and anti-β-actin (cat# A3854) antibody, GST protein tag, Crystal violet dye, Hoechst 33342, Meyer hematoxylin solution, and DPX were obtained from Sigma Aldrich. ImmEdge pen (hydrophobic barrier pen), Bloxall blocking solution, DAB Peroxidase Substrate Kit, VECTASTAIN ABC Kit were purchased from Vector Laboratories, Inc. Fluorochrome-conjugated secondary antibodies, Annexin-V Alexa Fluor 488, F(ab')2-goat anti-rabbit IgG (H+L) secondary antibody Qdot 655, and Verso cDNA Synthesis Kit as well as SYBR Green Real-Time PCR Master Mix were purchased from Molecular Probes-Invitrogen. β-Catenin (cat# 8480), active β-catenin (cat# 8814), p-GSK3β^ser-9 (cat# 5558), E-cadherin (cat# 3195), Cyclin D1 (cat# 2978), c-Myc (cat# 5605), DR5 (cat# 8074), cleaved PARP (cat# 5625), caspase-8 (cat# 9746), cleaved caspase-8 (cat# 9496), caspase-9 (cat# 9502), cleaved caspase-9 (cat# 7237), Ki-67 (cat# 12202), and GAPDH (cat# 2118) antibodies, as well as PP2A Sampler Kit (cat# 9780) were procured from Cell Signaling Technology, Inc. Anti-GSK3β total (cat# 612313) antibody was purchased from BD Biosciences. XIAP (cat# sc-55551) and horseradish peroxidase (HRP)-conjugated secondary antibodies and normal rabbit IgG (cat# sc-2027) were obtained from Santa Cruz Biotechnology. Human recombinant GST-tagged PR55α protein (cat# H00005520- P01) was purchased from Abnova. XenoLight D-Luciferin - K⁺ Salt bioluminescent substrate was bought from PerkinElmer. All chemicals and antibodies were obtained from Sigma unless specified otherwise.

Various human cancer cell lines including DLD-1, SW620 (colorectal adenocarcinoma), FaDu (pharynx squamous cell carcinoma), A549, NCI-H446 (lung carcinoma), PLC/PRF/5 (hepatoma), DU145, PC-3 (prostate carcinoma and adenocarcinoma), SKOV3 (ovarian adenocarcinoma), and human breast cancer cell lines MCF-7, MDA-MB-468, ZR-75-1 SK-BR-3, BT-549, mice breast cancer cell line 4T1 (metastatic breast adenocarcinoma), and human immortalized non-transformed breast epithelial cells MCF-10A were obtained from ATCC, USA. 4T1-luc2-GFP cells were purchased from Caliper Life Sciences (PerkinElmer). Early passage cells were resuscitated from liquid nitrogen vapor stocks as needed and cultured according to the manufacturer's instructions. Cells were routinely inspected microscopically for stable phenotype, and all experiments were performed within early passages (within 10) of individual cells. Cell line authentication (STR profiling) was done for SW620, as it was used for most of the important in vitro and especially in vivo experiments.

Rational and detailed synthesis of semicarbazone–chalcone hybrid compounds and their characterization are given in Supplementary Figs. S1 and S2A–S2C, Supplementary Table S1, and Supplementary Methods S1 and S2. Efficacy of these hybrids as anticancer agents on various cancer cell lines was assayed using standard sulforhodamine B (SRB) cytotoxicity assay (18, 19). Absorbance was measured on a plate reader (Epoch Microplate Reader, Biotek) at 510 nm to calculate the percent inhibition in cell growth by using the formula: [100−(absorbance of compound treated cells/absorbance of untreated cells)] × 100.

Clonogenic assay was done following standard procedure. In brief, cells (MCF-7 and MCF-10A) were seeded at 400 cells per well in 6-well plates and allowed to adhere. After 24 hours, cells were treated with either different doses (0, 0.125, 0.5, and 1 μmol/L) of CS-11 or vehicle. After 7 days, media were removed and cells were washed with PBS and fixed with ice-cold methanol, followed by staining with 0.5% crystal violet in methanol for 30 minutes. Excess stain was removed by washing with water thoroughly, and plates were allowed to dry. Representative images were taken to monitor single-cell colony formation efficiency of MCF-7 and MCF-10A cells.

mRNA were isolated from cells by using the Pure Link RNA Mini Kit (Ambion). cDNA was prepared by using Verso cDNA Synthesis Kit (Invitrogen) following the manufacturer's protocols. Each cDNA sample was amplified using SYBR Green in StepOnePlus Real-Time PCR System (Applied Biosystems). The reaction mixture contained 1 μL of cDNA and 0.1 μmol/L of primers in a final volume of 10 μL supermix. The experiments were performed in triplicate. As an internal control, GAPDH mRNA was amplified and analyzed under identical conditions. C_t value (the cycle number at which emitted fluorescence exceeded an automatically determined threshold) for gene of interest was corrected by the C_t value for GAPDH and expressed as ΔC_t. Data were measured as fold changes of mRNA amount, which was calculated as follows: fold changes = 2X (where X = ΔC_t for control group − ΔC_t for experimental group). The oligonucleotide primers used are as follows: for human GAPDH: forward: 5′-GTCAGTGGTGGACCTGACCT-3′, reverse: 5′-AGGGGAGATTCAGTGTGGTG-3′, product size 395 bp; for human Cyclin D1, forward: 5′-GCATGTTCGTGGCCTCTAAG-3′, reverse: 5′-CGTGTTTGCGGATGATCTGT-3′, product size 228 bp; and for human c-Myc: forward: 5′-GATCCAGACTCTGACCTTTTGC-3′, reverse: 5′-CACCAGCAGCGACTCTGA -3′, product size 102 bp.

Treated or untreated cells were lysed in RIPA buffer supplemented with phosphatase and protease inhibitors cocktail. Equal amounts of protein in SDS-PAGE sample buffer were loaded in 12% SDS polyacrylamide gels and transferred to PVDF membranes. After appropriate primary and secondary antibody incubation (20, 21), immunoreactivity was detected by enhanced chemiluminescence solution (Immobilon Western ECL, Millipore) and scanned by gel documentation system (Bio-Rad ChemiDoc XRS+). Densitometric analyses of Western blots were done by normalizing the pixel intensity of desired protein to their respective loading control by using Image Lab Software (Bio-Rad).

For immunoprecipitation (IP) studies, equal amounts (400 μg) of total protein of treated and untreated groups were incubated overnight at 4°C with either anti-human PP2A-B (specific for PR55α) or anti-β-catenin antibody. The immune complexes were precipitated with protein A/G PLUS Agarose beads (Millipore) and subjected to SDS-PAGE along with 10% input control (22), followed by Western blot analysis using primary antibodies and corresponding kappa light chain–specific HRP-conjugated secondary antibodies (to avoid heavy chain IgG bands), Then, blots were developed by enhanced chemiluminescence substrate. Densitometric analysis of IP blots was done by normalizing the pixel intensity of desired proteins to either input control or pull down protein by using Image Lab Software (Bio-Rad). Ratio between normalized desired protein versus pulldown protein was represented in percent change considering vehicle as the standard unit (100%).

Cells were grown on cover slips and treated as designated. After intermediate washes with cold PBS, cells were fixed with 4% paraformaldehyde, permeabilized by 0.1% NP40, followed by blocking with 2% BSA. After overnight incubation with primary antibodies at 4°C, cells were washed and incubated with fluorescent-conjugated appropriate secondary antibodies followed by DAPI staining. After washing, cells were mounted on slides and analyzed under inverted Zeiss confocal laser scanning microscope (Zeiss Meta 510 LSM; Carl Zeiss).

In brief, 4T1 cells were seeded in 6-well plates; next day, when cells were near 90% confluent, they were rinsed with PBS. A scratch was made in each well with the help of a sterile 200-μL pipette tip. Treatment was given and pictures were taken at 0, 12, 24 hours using phase-contrast microscope at ×10 magnification (Primovert; ref. 23). The percentage of open scratch area was quantified by TScratch Software (ETH Zurich).

All animal studies were conducted in accordance with the principles and standard procedures approved by the Institutional Animal Ethics Committee (IAEC) CSIR-Central Drug Research Institute (Lucknow, India). SW620 (human colon cancer) xenograft and orthotopic 4T1-luc2-GFP models have recently been established in the laboratory by following standard protocol. Briefly, early passage 98% viable 2 × 10⁶ SW620 cells or 1 × 10⁶ 4T1-luc2-GFP cells in 100 μL PBS were either subcutaneously injected in the flanks of the right hind leg or mammary fat pad on right flank of each 4- to 6-week-old nude Crl: CD1-Foxn1^nu mice. When tumor volume reached 50 to 100 mm³, mice were randomized and divided into two experimental groups. Tumor-bearing mice were administered intratumoral injection once a week with either vehicle alone (DMSO and serum-free media) or compound (5 mg/kg) dissolved in 50 μL vehicle until sacrifice. In case of SW620 xenograft, tumor size was measured using an electronic digital caliper at regular intervals. Tumor volume was estimated by standard formula V = Π/6 × a² × b, wherein a is the short and b is the long tumor axis (19). In the case of 4T1-luc2-GFP models, at day 25, mice were injected subcutaneously with d-luciferin (150 mg/kg body weight), anesthetized with ketamine, and whole mice were imaged up to 1 hour at various time points using a bioluminescent imaging system (IVIS Spectrum, Caliper). Regions of interest from displayed images were identified on the tumor sites and quantified as photons per second (p/s) using Living Image software.

MCF-7 cells were treated with either vehicle or CS-11 (10 μmol/L) for 24 hours. Cells were harvested and lysed in lysis buffer 17 followed by protein concentration estimation. Apoptosis array analysis was performed using the Proteome Profiler Human Apoptosis Array Kit (ARY009) from R&D Systems as per the manufacturer's instructions (24). Array images were captured by gel documentation system (Bio-Rad ChemiDoc XRS+) and were analyzed using the Image Lab Software (Bio-Rad). Heatmap was prepared by using Plotly software.

To measure intracellular cleaved caspase-8 and 9, cells were harvested by mild treatment of TrypLE (Invitrogen), washed, and first fixed with 4% paraformaldehyde for 10 minutes and permeabilized with 0.1% NP-40 in PBS for 10 minutes. Permeabilized cells were stained with either cleaved caspase-8 or cleaved caspase-9 or isotype control antibody for 30 minutes at 4°C, followed by washing and staining with Qdot 655 IgG secondary antibody for 30 minutess at 4°C. The stained cells were acquired in a FACSCalibur (Becton Dickinson) and analyzed by FlowJo software (Tree Star).

IHC was done following standard procedure established in the lab (19). Briefly, tumor tissues were fixed in 4% paraformaldehyde for 48 hours and embedded in paraffin wax. For staining, tissue sections were deparaffinized, rehydrated, and quenched for endogenous peroxidase. Antigen retrieval was performed by heating of slides in 10 mmol/L sodium citrate buffer (pH 6) for 30 minutes. Slides were then rinsed in PBS, and endogenous peroxide activity was neutralized by incubating with Bloxall (blocking solution) for 25 minutes. Tissue sections were then incubated with primary antibodies in 2% BSA against active β-catenin (1:800), E-cadherin (1:400), Ki-67 (1:400), and BSA only for negative control overnight at 4°C. Next day, slides were rinsed with PBS and incubated with biotinylated secondary antibody at room temperature for 1 hour. Followed by PBS washing and incubation with ABC reagent (Vector Laboratories) at room temperature for 1 hour, slides were incubated with 3′-3′-diamino-benzidine (DAB) and counterstained with hematoxylin. Tissue sections were then dehydrated and mounted by using DPX. Stained sections were observed under ×40 magnification of microscope (Leica).

Detailed methodologies for different pharmacokinetic studies are described in Supplementary Methods S3.

Initially binding pockets were identified on protein surface using Discovery Studio 4.1 (25). Protein–protein docking studies were performed using ZDOCK module of Discovery Studio-4 (26) to identify possible binding residue of β-catenin with PP2A enzyme. Least energy conformation of each cluster out of 2,000 generated poses was selected and further analyzed for its binding affinity and scoring. Structure of compound CS-11 was built and minimized using Marvin Sketch version 6.1.2 from Chem Axon (https://www.chemaxon.com/). Docking studies of CS-11 with PP2A protein were performed on binding pockets identified using Auto Dock 4.2 tool (27), followed by molecular dynamic studies by using Gromacs version 5.04 (28) to check the temporal stability of the compound to its target protein. CHARMM force field was used and parameters for compound were generated using Swiss Param tool (29). Protein–compound complex was solvated in a cubic box with TIP3P water molecules at 10 Å marginal radiuses. At physiologic pH, the structure was found to have a negative charge; thus, to make the system electrically neutral, sodium ions were added into the simulation box. Whole protein–compound system was subjected to energy minimization by conjugant gradient minimization algorithm. Isothermal–isochoric equilibration was performed using PME method for 100,000 steps. Isotropic pressure coupling was performed on the previously equilibrated system using Parrinello–Rahman method for 100,000 steps. Finally, protein–compound complex was subjected to molecular dynamic simulation for 20 nanoseconds. UCSF Chimera (30) was used for the model visualization and image generations.

Fluorescence measurements were performed using spectrofluorometer of PerkinElmer Life Sciences LS 50B as described previously (17). Briefly, the fluorescence spectra were obtained at 25 ± 0.1°C with a 1-cm path length Quartz cell. The intrinsic fluorescence was measured using excitation wavelength for exciting protein at 280 nm, and emission spectra were recorded in the range of 300 to 420 nm. The titration curve was obtained by adding the appropriate compound to the protein PR55α (human recombinant GST-tagged PR55α cat# H00005520- P01 from Abnova) solution (40 μg of protein/mL in 50 mmol/L Tris-HCL, 10 mmol/L reduced glutathione, pH 8.0). The stock compound concentration was 1 mol/L, and the pH was adjusted to 8.0 in the same buffer. Stock solutions were prepared freshly before each titration. For fluorescence quenching experiment, the concentration range of 10 pmol/L to 1 μmol/L of compound was used against PR55α protein, to the final volume. Background GST/buffer fluorescence was observed to be minimal and subtracted from experimental groups. Dissociation equilibrium constant (K_d) value was determined from the data fitted to a single exponential hyperbolic equation, by using the PRISM 3 nonlinear regression tool (GraphPad).

Most of the results are representative of at least three independent experiments. Student t test and two-tailed distributions were used to determine the statistical significance; P values ≤0.05 between groups were considered statistically significant.

Utilizing our molecular hybridization approach, we have synthesized a series of chalcone–semicarbazone hybrids containing 17 compounds (including parent chalcone-8e and semicarbazone-10a), and details of their synthesis and different qualitative and quantitative analyses are provided in Supplementary Figs. S1 and S2A–S2C and Supplementary Methods S1. To test the anticancer potential of our synthesized novel molecules, we performed anticancer screening in seven different cancer cell lines representing seven cancer types at a single dose of 10 μmol/L. Representative cell lines from different cancer types are MCF-7 (breast adenocarcinoma), DLD-1 (colorectal adenocarcinoma), FaDu (pharynx squamous cell carcinoma), DU145 (prostate carcinoma), A549 (lung carcinoma), SKOV-3 (ovarian adenocarcinoma), and PLC/PRF/5 (liver hepatoma). Detailed results of these investigations in terms of cell death (percentage of control) at 10 μmol/L of each compound are given in Supplementary Table S2A. Overall, we observed cytotoxicity of most of the molecules against different cancer types. Interestingly, we observed that our hybridization resulted in the induction of cytotoxic effects in breast and lung cancers compared with active parent chalcone (8e). To further validate this observation, we extended our studies in five different breast cancer cell lines (MDA-MB-468, BT-549, SK-BR-3, ZR-75-1, and 4T1) as well as human immortalized nontransformed/nontumorigenic breast epithelial cell line (MCF-10A) and tested the cytotoxic potential of whole series of molecules against these cells at a 10 μmol/L dose. Thorough analysis of the data represented in Supplementary Table S2B revealed that CS-11 is the molecule in which our hybridization approach worked best, as it showed maximum cytotoxicity to most of the breast cancer cells but less toxic effect on nontumorigenic breast epithelial MCF-10A cells. As in vitro anticancer efficacy of CS-11 as well as its cancer cell selectivity is better than parental chalcone (8e), we tested the comparative dose response of cytotoxic effects of these two in five different breast cancer cell lines as well as in MCF-10A cells and determined the IC₅₀ values for CS-11 and 8e. As demonstrated in Fig. 1A and Supplementary Fig. S3, IC₅₀ values of CS-11 are lower than 8e in all the breast cancer cell lines tested with maximum cytotoxic efficacy in MCF-7 cells. On the other hand, the inverse result was obtained in the case of MCF-10A cells where CS-11 was found to be less cytotoxic than parent chalcones 8e, indicating better tumor cell selectivity of CS-11 over parent chalcone. To compare tumor cell–selective effect of our hybrid molecule with standard marketed breast cancer drug, we next examined the effect of CS-11 and the FDA-approved cancer drug doxorubicin, in multiple doses on nontumorigenic breast epithelial cells MCF-10A versus tumorigenic breast epithelial MCF-7 cells. Most strikingly, we observed (Fig. 1B) that the FDA-approved drug doxorubicin was equally cytotoxic in MCF-10A cells and MCF-7 cells. In contrast, CS-11 was found to be significantly (P ≤ 0.01) more cytotoxic to MCF-7 cells compared with MCF-10A cells. This is a clear indication of better cancer cell selectivity of CS-11 over standard breast cancer drug doxorubicin in terms of inducing in vitro cytotoxic effects. To study the comparative cytotoxic efficacy of CS-11 and parent 8e in terms of inducing apoptosis in transformed tumorigenic cells versus nontumorigenic cells, we treated MCF-7 and MCF-10A cells with CS-11 and 8e at a dose of 10 μmol/L for 24 hours and performed Annexin-V staining. Results shown in Fig. 1C and D clearly indicate that our lead molecule CS-11 has selective apoptotic effect in MCF-7 over MCF-10A cells in comparison with 8e. Apoptotic ability of CS-11 was further validated by cleaved PARP staining of vehicle and CS-11–treated cells, where we observed significant (P ≤ 0.01) induction of cleaved PARP–positive cells in treated cells as compared with vehicle control (Fig. 1E, right and left). In addition, we also found that CS-11 at 0.5 and 1 μmol/L doses significantly (P ≤ 0.01) inhibited clonogenic efficiency in MCF-7 cells but not in MCF-10A cells (Fig. 1F, right and left). Altogether, our results suggest that CS-11 has better tumor cell selectivity and cytotoxic efficacy over standard drug as well as our parent molecule from where it was originally derived.

Furthermore, to gain mechanistic insight of CS-11–mediated breast cancer cell apoptosis, we made use of apoptosis antibody array (ARY009, R&D Systems) to simultaneously assess the expression of 35 apoptosis-related proteins spanning both extrinsic and intrinsic apoptosis pathways. MCF-7 cells were treated with either vehicle or 10 μmol/L of CS-11 for 24 hours, and harvested proteins were incubated with the array as per the manufacturer's protocol. Array immunoblots and/or heatmap image along with spot coordinates of array blots (Figs. 2A–C) clearly indicated the differential expression pattern of multiple apoptotic and anti-apoptotic proteins following CS-11 treatment compared with vehicle control. For validation of array data, we selected two proteins (DR5 and XIAP) that were found to be reversely regulated upon CS-11 treatment as well as members of two different (DR5-extrinsic, XIAP-intrinsic) apoptotic pathways having potent biological significance, although there were some changes in other proteins too. Consistent with array results, CS-11 treatment significantly (P ≤ 0.05) upregulates the expression of proapoptotic DR5 as well as inhibits the expression of antiapoptotic XIAP protein as compared with vehicle-treated cells (Fig. 2D; Supplementary Fig. S4). These upstream death signals ultimately lead to cell death via activation of different prototype extrinsic (caspase-8) and intrinsic (caspase-9) caspases. To check the activation and involvement of caspase-8 and 9 in CS-11–mediated apoptosis, we performed Western blot and FACS analysis of vehicle and treated cells and observed downregulation of pro-caspase-8 and 9 (Fig. 2E) as well as upregulation of active cleaved caspase-8 and caspase-9 following CS-11 treatment (Fig. 2F) as compared with vehicle. Together, the results indicate the involvement of both extrinsic and intrinsic pathways in CS-11 induced apoptosis.

Before going to test the in vivo efficacy of the active molecule (CS-11), we performed initial pharmacokinetic studies. First, we determined the concentration of CS-11 in our cell culture condition where we detected pure CS-11 in the treated cell lysate, and it was found to be enhanced with increasing the dose, indicating its in vitro stability in the media (Fig. 3A). Next, we determined the solubility of CS-11 at various pH to mimic different gastrointestinal conditions and also compared its solubility with parent molecules (8e and 10a). Interestingly, here our hybridization worked better to increase the solubility of CS-11 as compared with phenotypically active parent 8e at a neutral pH, indicating the probability of better absorption of CS-11 in the intestine (Fig. 3B). Stability of CS-11 in simulated gastric fluid (SGF) and simulated intestinal fluid (SIF) is one of the important parameters of any drug-like molecule, and as shown in Fig. 3C, percent drug (CS-11) remaining in SGF and SIF were found to be 93.16 and 87.87, respectively, up to the desired time point, suggesting its drug-like properties. For a drug to be effective, adequate active form must reach the target to obtain the desired effect. Plasma stability of CS-11 at 2 and 10 μg/mL in rat plasma was found to be 82.17% ± 1.52% and 62.24% ± 1.76%, respectively (Fig. 3D). The compound was found to be stable in plasma after 2 hours at room temperature for the duration of typical sample handling and processing. The in vitro microsomal metabolic stability of CS-11 at two different concentrations of 2 and 10 μg/mL was also performed. Percent drug remaining for 2 and 10 μg/mL was found to be 92.53% ± 4.40% and 92.34% ± 0.57%, respectively (Fig. 3E) in rat liver microsomes (RLM). Compound was stable when the experiment was conducted in the absence of NADPH (negative control) and also thermally stable at 37°C (100 mmol/L phosphate buffer). Testosterone (2 μg/mL, positive control) half-life in the RLM was within the acceptable in-house limits. In vitro systems with low metabolic capacity evaluate the cytotoxic effects of parent compound 8e. To reduce misinterpretation of these data, it is important to determine metabolic stability. The plasma protein binding of CS-11 at different concentrations of 2 and 10 μg/mL was found to be 78.5% ± 0.076% and 73.04% ± 1.54%, respectively (Fig. 3F). K_a and K_d at 2 and 10 μg/mL were found to be (61.62 ± 14.35) × 10⁻³, (40.67 ± 15.10) × 10⁻⁴ and (46.60 ± 11. 51) × 10⁻³, (79.02 ± 42.90) × 10⁻⁴/min; per μg/mL, respectively. Increased CS-11 binding with plasma proteins accelerates the rate of elimination (t_1/2 decreases) because it makes more drug available to the site of metabolism. There may be less possibilities of drug–drug interaction due to <90% protein binding.

To evaluate the in vivo efficacy of CS-11, we utilized SW620 subcutaneous xenograft model to study tumor growth, as SW620 has been widely used in in vitro mechanistic studies. To understand its antimetastatic potential, we utilized Luc2-GFP–tagged 4T1 syngeneic orthotopic breast tumor model. Before proceeding to SW620 xenograft model, we first determined the IC₅₀ value of CS-11 and parent 8e in SW620 cells, where we found the IC₅₀ value of CS-11 is 4.06 ± 0.11 μmol/L compared with 10.01 ± 0.66 μmol/L in case of 8e (Supplementary Fig. S5A). In SW620 xenograft model, CS-11 (5 mg/kg/week) intratumor administration showed significant (P ≤ 0.05) inhibition of tumor growth compared with vehicle as observed in Fig. 4A. We also observed significant (P ≤ 0.05) difference in the weight of harvested tumors at the end of the experiment between CS-11–treated and vehicle group (Fig. 4B). Importantly, we did not observe any significant body weight difference between vehicle and treated group during the full-course treatment, suggesting its nontoxic nature in in vivo condition as well (Fig. 4C). Next, we evaluated the effect of CS-11 on breast cancer metastasis through live animal imaging. As observed in Fig. 4D (left and right) compared with vehicle-treated group, CS-11 treatment (5 mg/kg/week) reduced overall spread of breast tumor growth as well as inhibited lung metastasis in 4T1-luc2-GFP orthotopic model. Therefore, CS-11 does not only inhibit tumor growth but also have potential to alleviate metastasis of breast cancer to the secondary organs. After analyzing in vivo antimetastatic property of CS-11, we explored its antimigratory efficacy. Here, we performed an in vitro scratch assay in the presence or absence of CS-11 in a time- and dose-dependent manner and found significantly (P ≤ 0.05) less migration in CS-11–treated 4T1 cells as compared with vehicle-treated cells, which covered the scratch area within 24 hours (Supplementary Fig. S5B).

CS-11–mediated inhibition of lung metastasis inspired us to pursue further its in-depth mechanistic insight. Published literature in patient samples as well as animal data indicate that metastasis of breast cancer, especially to the lungs, largely relies on functional activation of β-catenin signaling (31–34). To test this possibility in our case, we first selected five different metastatic cell lines (4T1, SW620, PC-3, FaDu, and NCI-H446) and treated with two doses (5 and 10 μmol/L) of CS-11for 24 hours and analyzed the expression of active β-catenin and its two predominant target proteins (c-Myc and Cyclin D1). IC₅₀ of CS-11 is around 5 μmol/L at 48 hours, and to observe direct effect of CS-11, we opted for 5 and 10 μmol/L doses at 24 hours to study the mode of action of CS-11. As shown in Fig. 5A (left and right) and Supplementary Fig. S6A, in most of the metastatic cell lines, CS-11 inhibited the expression of β-catenin and its target c-Myc and Cyclin D1. There is a variation of CS-11 dose response among different cell lines with respect to target protein expression, which might be due to their different tissue origin and genetic background of the particular cell line. To check whether the downregulation of c-Myc and Cyclin D1 protein is transcriptionally regulated by the β-catenin in response to CS-11, we performed real-time PCR analysis in SW620 cells treated with vehicle and CS-11. Interestingly, we observed significant (P ≤ 0.05) dose-dependent decrease in the mRNA expression of c-MYC and CCND1 genes following CS-11 treatment compared with vehicle-treated cells (Fig. 5B). As β-catenin signaling is indispensably related to E-cadherin functionality, we next assessed the status of E-cadherin in SW620 and FaDu cells after CS-11 treatment. Strikingly, in both, we observed significant (P ≤ 0.05) upregulation of E-cadherin expression following CS-11 treatment compared with vehicle-treated cells (Fig. 5C; Supplementary Fig. S6B). Association of β-catenin with E-cadherin firms the adherence junction of a cell-to-cell contact, resulting in metastasis inhibition or early dissemination of tumor cells from the primary site (35). To test this, we checked the association of β-catenin with E-cadherin before and after CS-11 treatment by IP experiments. Although the active β-catenin was found to be downregulated in SW620 cells following CS-11 treatment, but to our great surprise, we observed significant (P ≤ 0.05) increase of β-catenin–E-cadherin association after CS-11 treatment compared with vehicle treatment (Fig. 5D). To find out the possible localization of reversely regulated active β-catenin and E-cadherin expression, we performed confocal microscopy in vehicle and CS-11–treated SW620 cells. Interestingly, we observed that cytoplasmic and nuclear active β-catenin expression got decreased after CS-11 treatment compared with control, but the expression of active β-catenin in the cell surface remained same or even more in the treated cells. On the other hand, compared with vehicle treatment, CS-11 treatment robustly increased cell surface E-cadherin expression (Fig. 5E). Because of antibody incompatibility, we were not able to perform dual staining in confocal studies. So, IP results along with confocal analysis clearly indicate that CS-11–induced E-cadherin is strongly associated with cell surface β-catenin and may lead to the inhibition of early release of metastatic cells from the primary tumor site. Furthermore, to validate our major in vitro finding of active β-catenin downregulation and E-cadherin upregulation as well as proliferation inhibition upon CS-11 treatment, we assessed their expression in vivo in harvested tumors via IHC. As observed in Fig. 5F, compared with vehicle control, CS-11 treatment resulted in downregulation of active β-catenin with concomitant increase in E-cadherin, suggesting the in vivo alignment of in vitro findings. Expression of Ki-67 is a hallmark feature to determine tumor cell proliferation in vivo, and here, we observed reduction of nuclear Ki-67 expression in CS-11–treated colon tumors compared with vehicle control.

To address the probable target of CS-11 to destabilize the β-catenin in the cytoplasm and nucleus, we concentrated our focus on key regulators of β-catenin protein. The vast majority of literature suggests that GSK3β is the major regulator of β-catenin signaling (7). GSK3β phosphorylates β-catenin, leading to its degradation, but PP2A prevents its degradation by dephosphorylating β-catenin (12, 36–38). To test the role of GSK3β in CS-11–mediated decrease in active β-catenin expression, we treated 4T1, SW620, and FaDu cells with CS-11 (5 and 10 μmol/L) for 24 hours and assessed the status of total and inactive form of GSK3β (p-GSK3β^ser-9). However, in all three cell lines, we did not observe any significant change of either inactive (p-GSK3β^ser-9) or total form of GSK3β following CS-11 treatment compared with vehicle-treated cells (Fig. 6A; Supplementary Fig. S7), suggesting no involvement of GSK3β in this process. Next, we focus on another potential β-catenin regulator, PP2A, which has three different subunits, namely, PP2A-A (structural), PP2A-B (regulatory), and PP2A-C (catalytic). PR55α, an isoform of PP2A-B, is known to directly interact with β-catenin and promotes its dephosphorylation (12). To test the prospective binding of CS-11 with PP2A, we performed molecular docking. To decipher the binding mode of β-catenin with the PP2A holoenzyme consisting of PR55α B-subunit (PDB id - 3DW8; ref. 39) protein–protein docking studies were carried out. The ZDOCK analysis indicated that the highest ranked pose has a cluster size of 48, ZDOCK score 30.4, and ZRANK score -120.489. The highest ranked pose was bound to the interface of B-subunit and C-subunit of PP2A holoenzyme. Next, we performed docking of CS-11 with individual subunits as well as PP2A holoenzyme. Evidently, we found the same binding pocket of PR55α for β-catenin and CS-11, indicating a probable competitive binding of β-catenin and CS-11 on PR55α.

Docking studies of top three binding grooves of PP2A protein resulted binding energy of -7.12 Kcal/mol with CS-11 having cluster size of 28 conformations. Residues of PR55α, which were found to be important for interaction with CS-11, were Lys95, Asn97, Lys98, Asn181, Asp340, Phe343, Asp344, Lys345, Leu420, and His421. Strong hydrogen bonds of 2.00 Å and 2.26 Å were formed by CS-11 with Phe245 and Asn97, respectively (Fig. 6B). These interactions involve hydrogen bonds, Pi–Pi, Pi–cation, salt bridges within a cut-off radius of 2.5 Å. Molecular dynamic simulation on the CS-11–PP2A complex with lowest energy was preceded to analyze the stability of CS-11 with PP2A holoenzyme. Very low RMSD of 2.5 Å was recorded for the long simulation run of 20 ns, indicating that the CS-11 molecule was highly stable with PP2A (Fig. 6C). To validate in silico results, we performed well-established internal fluorescence quenching measurement assay (17) to determine real biophysical association between PR55α and CS-11. In our binding assay, we found robust association between CS-11 and PR55α as it started to show fluorescence quenching even at 10 pmol/L dose, and finally, we obtained a K_d value of 45 pmol/L (Fig. 6D, left and right). Although fluorescence intensity of PR55α decreased regularly with the addition of increasing concentration of CS-11, it did not affect the basic peak appearance of PR55α, suggesting perfect binding capabilities of CS-11 (Fig. 6D, left). In an additional support of our binding data, we performed PP2A-B (PR55α) pull down experiments, where we analyzed the association of PR55α with β-catenin before and after CS-11 treatment. Interestingly, in our pull down experiments, we observed significantly (P ≤ 0.05) reduced association of PR55α with β-catenin after CS-11 treatment compared with vehicle-treated cells, whereas an association between PP2A-B (PR55α) and PP2A-C remained unchanged. Also, we did not see any change of basal protein expression of PP2A-B and PP2A-C after CS-11 treatment. To avoid the interference of huge IgG band in pull down experiments, we used light chain–specific secondary antibody (Fig. 6E). Together, in silico, physical binding as well as biological data confirmed that CS-11 occupies the same binding pocket of PR55α where β-catenin binds and may promote β-catenin dissociation from PR55α (Fig. 6F).

Targeting simultaneously tumor growth and metastasis with a single chemical entity is a daunting task in cancer drug discovery. Molecular hybridization of different pharmacophores offers the possibility of engagement of different target proteins together to control both tumor growth and metastasis. Our hybridization approach did not remarkably accelerate in vitro cytotoxic activity in multiple cancers, but it resulted in induced efficacy in selective cancers like breast and lung and most strikingly increased tumor cell selective cytotoxic action. Moreover, hybridization has increased solubility of CS-11 at neutral pH, which may result in better pharmacokinetic profile. Albeit, our continuous effort is still going on to have even better molecules around this pharmacophore.

β-Catenin is found to be an unique dual function protein that can control metastasis by regulating cell-to-cell adhesion via interacting with E-cadherin, and as a transcription factor, it regulates the expression of key genes that are indispensable for cell growth and proliferation (40). A series of compelling studies suggest that β-catenin was found to be upregulated/activated in most of the malignant tumors, including breast, colon, lung, prostate, pancreatic, and renal (34, 41–46). Multiple studies have shown that β-catenin signaling dictates cell fate decisions and has the capability to transform normal cells into tumorigenic ones (47, 48). Thus, any disruption of β-catenin signaling represents a great opportunity to develop novel drugs for selective cancer chemoprevention and therapy. However, therapeutic diminution of direct β-catenin transcriptional activation by small-molecule inhibitors would withdraw its antimetastatic effect, as it would result in loosening of adherence junction via less E-cadherin association. Therefore, ideal target engagement situation would be indirect so that we could diminish β-catenin transcriptional activity as well as promote its cell surface recruitment. Our novel chalcone–semicarbazone hybrid molecule CS-11 mimics the ideal situation, where it attenuates transcriptional activity of β-catenin via inhibiting the expression of target genes (CCND1 and c-MYC) by directly interacting with PR55α as well as promotes cell surface association of β-catenin with E-cadherin to enhance cell-to-cell adherence junction. The substrate specificity of PP2A is considered to be determined by its B subunit (49). It has been shown that PR55α, a regulatory subunit of PP2A but not the catalytic subunit, PP2A-C, directly interacts with β-catenin and is essential in PP2A-mediated β-catenin dephosphorylation (12). Our in silico, direct physical binding assay and IP experiments confirmed that CS-11 selectively interacts with PR55α but not with other subunits of PP2A to finally decrease β-catenin nuclear function through its reduced dephosphorylation. Another important aspect is CS-11 mediated strong upregulation of E-cadherin on the cell surface, which helps further recruitment of leftover β-catenin to the cell surface that ultimately leads to the nonavailability of active β-catenin in the nucleus as well as strengthens cellular adherence junction that may prevent early dissemination of metastatic cancer cells. Inhibition of β-catenin transcriptional targets like c-Myc and Cyclin D1 is one of the consequences of CS-11–mediated antiproliferative and proapoptotic effects where death receptor upregulation could be a major player. It has also been reported that overexpressed E-cadherin can also promote apoptosis in cancer cells via death receptor upregulation (50), and in our case, CS-11–induced E-cadherin may also result in death receptor upregulation in treated cells. Together, our synthesized novel small-molecule inhibitor has a unique capability of diminishing transcriptional activation of β-catenin target genes on one hand, but on the other hand, it enhances cell membrane function of β-catenin along with E-cadherin, which may eventually lead to target both tumor growth and metastasis.

No potential conflicts of interest were disclosed.

Part of this work has been filed for provisional Indian patent vide application no. 65NF-DEL2014.

Conception and design: S. Maheshwari, S.R. Avula, L.R. Singh, G.R. Palnati, M.I. Siddiqi, K.V. Sashidhara, D. Datta

Development of methodology: S. Maheshwari, S.R. Avula, A. Singh, L.R. Singh, G.R. Palnati, S. Meena, M. Riyazuddin, M.I. Siddiqi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Maheshwari, A. Singh, R.K. Arya, S.H. Cheruvu, R. Kant, M. Riyazuddin, J.R. Gayen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Maheshwari, S.R. Avula, A. Singh, L.R. Singh, G.R. Palnati, S.H. Cheruvu, S. Shahi, T. Sharma, R. Kant, M. Riyazuddin, H.K. Bora, M.I. Siddiqi, J.R. Gayen, D. Datta

Writing, review, and/or revision of the manuscript: S. Maheshwari, S.R. Avula, A. Singh, L.R. Singh, G.R. Palnati, S.H. Cheruvu, S. Shahi, T. Sharma, A.K. Singh, M. Riyazuddin, M.I. Siddiqi, J.R. Gayen, D. Datta

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Maheshwari, L.R. Singh, R.K. Arya, S. Shahi, J.R. Gayen

Study supervision: K.V. Sashidhara (Chemistry), M.I. Siddiqi (Bioinformatics), J.R. Gayen (Pharmacokinetics), D. Datta (Biology)

Other (designed and performed all in vivo experiments): A. Singh

Other (provided help in in vivo work and confocal and fluorescence microcopy): R.K. Arya

Other (pharmacokinetic study): S. Shahi

Other (anticancer screening and IC₅₀ determination of molecules): S. Meena

Other (chemical synthesis of lead compound): K.V. Sashidhara

We sincerely acknowledge the excellent technical help of Drs. S. P. Singh, Kavita Singh, and Kalyan Mitra of Electron Microscopy unit for Confocal Imaging and A.L. Vishwakarma of SAIF for the Flow Cytometry studies. We are grateful to Dr. Sudhir Kumar Singh, Amit Kumar Tripathi, and Amit Deshmukh for their generous help in performing binding assays. We thank Dr. Tejender S. Thakur of Molecular and Structural Biology Division, CSIR-CDRI, for supervising the X-ray data collection and structure determination of our compound reported in this article. Institutional (CSIR-CDRI) communication number for this article is 9493.

Research of all the authors' laboratories was supported by CSIR-CDRI Institutional Fund, Network Project BSC0106, and Fellowship grants from CSIR and UGC.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.