Soyasapogenol-A targets CARF and results in suppression of tumor growth and metastasis in p53 compromised cancer cells

We screened some phytochemicals for cytotoxic activity to human cancer cells and identified Soyasapogenol-A (Snol-A) as a potent candidate anti-cancer compound. Interestingly, Soyasapogenin-I (Snin-I) was ineffective. Viability assays endorsed toxicity of Snol-A to a wide variety of cancer cells. Of note, wild type p53 deficient cancer cells (SKOV-3 and Saos-2) also showed potent growth inhibitory effect. Molecular analyses demonstrated that it targets CARF yielding transcriptional upregulation of p21WAF1 (an inhibitor of cyclin-dependent kinases) and downregulation of its effector proteins, CDK2, CDK-4, Cyclin A and Cyclin D1. Targeting of CARF by Snol-A also caused (i) downregulation of pATR-Chk1 signaling leading to caspase-mediated apoptosis and (ii) inactivation of β-catenin/Vimentin/hnRNPK-mediated EMT signaling resulting in decrease in migration and invasion of cancer cells. In in vivo assays, Snol-A caused suppression of tumor growth in subcutaneous xenograft model and inhibited lung metastasis in tail vein injection model. Taken together, we demonstrate that Snol-A is a natural inhibitor of CARF and may be recruited as a potent anti-tumor and anti-metastasis compound for treatment of p53-deficient aggressive malignancies.


Soyasapogenol-A, but not Soyasaponin-I, caused potent cytotoxicity to cancer cells.
We screened 23 natural compounds for their cytotoxicity in human normal lung fibroblasts (TIG-3) and three cancer cell types, osteosarcoma (U2OS; wild type p53), breast adenocarcinoma (MCF-7; wild type but functionally inactive p53) and HT1080 fibrosarcoma (mutant p53). In comparative cytotoxicity analysis, cells were treated with 5 μM of all the compounds for 24-48 h. We found that Phytochemical (PH)-11 (Soyaspogenol-A; Snol-A) was significantly cytotoxic (50-70%) to U2OS, HT1080 and MCF-7 cells; normal human fibroblast cells showed milder (20-30%) in several independent experiments (Fig. S1). Furthermore, PH-10 (Soyasaponin-I; Snin-I) with similar structure was not toxic to any of the cancer cell types (Fig. S1). In order to confirm such differential effect of Snol-A and Snin-I, we investigated dose-dependent response using more human cancer cells including osteosarcoma (U2OS; wild type p53 and Saos-2; null p53), ovarian adenocarcinoma (SKOV3; null p53), and breast adenocarcinoma (MDA-MB-231; mutant p53). As shown in Fig. 1A,B, whereas Snol-A caused dose-dependent inhibition of cell proliferation in all the four cancer cell types, Snin-I was ineffective. Microscopic observations of U2OS and SKOV-3 cancer cells treated with Snol-A (2-10 μM) for 48 h showed stressed phenotype (irregular and flattened cell shapes) and restricted growth as compared to control cells (Fig. 1C) and was further confirmed by long-term clonogenicity assay (Fig. 1D). Snin-I caused no effect in these assays (Figs. 1B,C and S2A-C). Of note, Snol-A treated cancer cells that lacked wild type p53 function (SKOV-3 and Saos-2) also showed considerable dose-dependent cytotoxicity.
Snin-I structurally owns one hydroxyl group at C-22 and three sugars at C-3, while Snol-A was found to have no sugar chains on the C-3, and possessed two hydroxyl groups at C-21 and C-22 38 (Fig. S3A-C). ADMET predictions on pharmacodynamic activity (distribution, metabolism, excretion and toxicity) revealed better human intestinal absorption (HIA) score for Snol-A than Snin-I (Fig. S3D,E) suggesting that Snol-A will be better absorbed from the intestinal tract upon oral administration. The score for penetration through the Blood-Brain Barrier (BBB) was also higher for Snol-A than Snin-I (Fig. S3D,E). In terms of metabolism, both compounds showed similar characteristics as a substrate for CYP450 enzyme (Fig. S3D,E). Caco-2 permeability, predicts assimilation of drugs into human intestinal 39 , showed better score for Snol-A than Snin-I (Fig. S3D,E). Toxicity predictions (LD 50 ) revealed 40 higher toxicity of Snol-A than Snin-I (Fig. S3D,E). These predictions matched with our in vitro results for several cancer cells lines.
Snol-A caused growth arrest that was mediated by upregulation of p21 WAF1 . In order to investigate the molecular mechanism of Snol-A induced toxicity, we first analyzed the cell cycle profiles in control and treated {sub-toxic (IC 30 ) and moderately toxic doses (IC 50 ) of Snol-A: SKOV-3 (6 μM and 10 μM), and MDA-MB-231 and Saos-2 (10 μM & 20 μM) cells. As shown in Fig. 2A, there was an increase in S phase population, suggesting cell cycle arrest, in all the three cell lines in response to Snol-A treatment. Cells treated with higher dose (10-20 μM) showed a distinct apoptotic sub-population, clearly evident in SKOV-3. Analyses of cell cycle progression key proteins by immunoblotting revealed decrease in the levels of CDK2, Cyclin A, Cyclin D1 and CDK4 in SKOV-3 cells (Fig. 2B). Of note, p21 WAF1 showed a significant increase in Snol-A treated cells (Fig. 2B). Analysis of its transcript levels reaffirmed increased p21 WAF1 levels in Snol-A treated SKOV-3 cells (Fig. 2C). To analyze the effect of Snol-A on p21 WAF1 transcript, we enrolled p21 WAF1 promoter (pWWP)-Luciferase reporter system that affirmed a significant increase in p21 WAF1 promoter activity in Snol-A treated SKOV-3 cells (Fig. 2D). Decreased levels of Cyclin D1 and CDK2 were also confirmed by immunostaining (Fig. 2E) in Snol-A treated SKOV-3 cells. These results suggested that Snol-A induced growth arrest is modulated by the activation of p21 WAF1 in p53-deficient cancer cells. p21 WAF1 activation in Snol-A treated p53-deficient cells was mediated by targeting CARF. We next examined the mechanism of p21 WAF1 activation in p53-null SKOV-3 cells. Based on our earlier report that demonstrated CARF as a transcriptional repressor of p21 WAF1 in p53-deficient cells 36 , we analyzed the status of CARF in control, Snin-I and Snol-A treated SKOV-3 cells. As shown in Fig. 3A, Snol-A, and not Snin-1, (10 μM) treated cells showed a significant decrease in CARF protein levels. Furthermore, Snol-A led CARF suppression was found to be consistent with an increase in p21 WAF1 expression (Fig. 3B), but no such effect was observed with Snin-I. As shown in Fig. 3C,D, Snol-A induced downregulation of CARF and upregulation of p21 WAF1 in SKOV-3 cells was found to be dose and time dependent, showing stronger effects at the higher doses (~6-8 μM) and longer treatment (48 h) time (Fig. 3C,D). Similar results were obtained in p53-deficeint Saos-2 cells (Fig. 2E) and p53 mutant (MDA-MB-231 and H1299) cells (Fig. S4A,B) that required higher IC 50 doses as compared to SKOV3.
Snol-A led CARF-suppression inhibits pATM-Chk1 signaling and promote apoptosis at higher concentration. CARF has earlier been shown to regulate DNA damage response (DDR) in cells.
Overexpression of CARF caused activation of DDR, promoting growth arrest and senescence via activated ATR-Chk1 pathway 34,41 . In light of this information, we examined ATR-Chk1 signaling axis in control and Snol-A treated SKOV-3 cells. As shown in Fig. 4A, dose dependent decrease in CARF levels in Snol-A treated cells were accompanied by decrease in ATR, pATR and Chk-1 expression. Furthermore, Snol-A treated cells also showed a significant decrease in PARP1/2 and corresponding increase in cleaved PARP1/2. Consistent with these altered expression levels and acquired apoptosis phenotype in cells treated with higher Snol-A concentrations, pro-caspase −9 and −3 showed a marked decrease, while an increase in cleaved Caspase-3 was observed in Snol-A treated cells (Fig. 4B). As shown in Fig. 4C, increased expression of cleaved PARP1/2 and its nuclear staining affirmed apoptosis in these cells at higher doses of Snol-A. www.nature.com/scientificreports www.nature.com/scientificreports/ CARF-targeting by Snol-A reduced cancer cell migration and invasion properties. CARF enrichment in cancer cells was shown to promote cell migration and invasion and led to Epithelial-Mesenchymal Transition (EMT) during malignant metastases 37 . In the light of this information, we investigated the effect of Snol-A on cell migration and invasion in cellular and molecular assays. Dose dependent effect of Snol-A was determined by treating the SKOV-3 cells with 0.5, 2 and 4 μM. Although in several independent cell viability assays, 4 μM Snol-A was seen to cause cytotoxicity. In Wound-healing assays wherein the cells are first grown to a monolayer and then treated with Snol-A, only a minor cytotoxicity was observed with 4 μM. On the other hand, as shown in Fig. 5A, significant inhibition of cell migration was observed. Furthermore, Snol-A treated cells showed marked reduction in invasive properties as analyzed by matrigel invasion assay (Fig. 5B). Expression analysis of key proteins associated with cell migration and invasion signaling revealed decrease in CARF, β-catenin, Vimentin, Smad 2/3, heterogeneous nuclear ribonucleoprotein K (hnRNP-K), and matrix metalloproteinase-9 (MMP-9) protein levels (Fig. 5C). CARF upregulation was earlier shown to instigate nuclear enrichment of β-catenin 37 . We, therefore, next examined the effect of Snol-A on β-catenin levels in nucleus. As shown in Fig. 5D, Snol-A treated cells showed a distinct decrease in nuclear β-catenin suggesting Snol-A targeted CARF-inhibition abrogated β-catenin nuclear function, and resulted in reduced migration and invasion capacity of cells. Immunostaining of Vimentin and Fibronectin, the two key mesenchymal markers showed a decrease in their levels in response to Snol-A treatment (Fig. 5D). Furthermore, Snol-A treated cells showed remarkable decrease in hnRNP-K, a key effector protein involved in cell migration.
Overexpression of CARF rescued the cells from Snol-A induced growth arrest, apoptosis, and EMT. In order to check whether overexpression of CARF could rescue the cells from CARF-inhibitory activity of Snol-A, we generated CARF-GFP overexpressing (CARF-OE) SKOV-3 cells and determined the effect of www.nature.com/scientificreports www.nature.com/scientificreports/ Snol-A on their growth arrest, apoptosis, and EMT phenotypes (Fig. 6A). Observation of cell phenotype showed resistance of CARF-OE cells to Snol-A induced growth arrest/apoptosis as compared to the control cells (infected with empty pCX Neo vector) (Fig. 6B). Of note, control, not the CARF-OE, cells showed stressed morphology with high doses (6 and 10 μM) of Snol-A (Fig. 6B). Consistent to these, no significant change in p21 WAF1 expression was observed in Snol-A treated CARF-OE cells (Fig. 6C). Furthermore, immunoblotting for cell cycle (p21 WAF1 , CDK2, Cyclin D1, CDK4) and apoptosis (ATR, Chk1, PARP1 and Procasepase-9) markers showed no significant alteration in their levels in these cells (Figs. 6D and S5A,B). Snol-A treated control (SKOV-3 / pCX Neo) cells, as expected, exhibited decrease in CARF, β-catenin, ATR, PARP1/2, and increase in p21 WAF1 (Fig. S5C). We also analyzed cell migration and invasion characteristics of Snol-A treated CARF-OE cells. As shown in Fig. 6E and consistent with the earlier reports 36,37 , CARF-OE cells showed higher migration as compared to the control. Of note, whereas control cells showed delayed migration when treated with Snol-A, CARF-OE did not show significant effect (Fig. 6E). Similar results were obtained for invasion characteristics of these cells (Fig. 6F). Molecular analyses revealed no significant change in expression level of proteins (β-catenin, vimentin, Smad 2/3, hnRNPK, and MMP9) involved in migration and invasion signaling in CARF-OE Snol-A treated cells (Fig. 6G,H). Taken together, we found that overexpression of CARF rescued the cells from anti-proliferative and anti-migration activity of Snol-A suggesting that CARF is one of its main target proteins.

Snol-A caused suppression of tumor growth and lung metastasis.
We next examined the effect of Snol-A on tumor growth and metastasis in in vivo xenograft model of immunodeficient mice. As shown in Fig. 7A, mice were fed with Snol-A for two weeks (15 mg/Kg of body weight (BW), twice/week) before xenografting of SKOV-3 cells by subcutaneous and intravenous injections. Post-1 week of injections, mice were feed with Snol-A (15 mg/Kg BW, as determined by independent experiment) for next 4 weeks, before sacrifice. As shown in Fig. 7B, Snol-A treated mice exhibited no significant signs of toxicity (monitored by body weight) (Fig. 7B). www.nature.com/scientificreports www.nature.com/scientificreports/ Of note, Snol-A fed mice demonstrated a potent and significant reduction (>55%) in growth of subcutaneous xenografts and lung metastases (>60%) (Fig. 7C,D). On the other hand, heart, liver, stomach, intestine and spleen were not seen to have any tumors either in the control or treated mice group. Furthermore, the fed mice looked as active as the control group. No signs of skin rash or eczema scars or particular slow behavior such as prolonged sleep, inactivity were observed in fed mice.

Snol-A as a natural inhibitor of CARF.
As shown in several earlier reports [29][30][31][32][33][34] , CARF has been demonstrated as a dual regulator of cell proliferation fates. Its upregulation in replicative and stress induced senescence caused activation of p53-p21 WAF1 axis and growth arrest [29][30][31][32][33][34] . Knockdown of CARF caused ATR/Chk1-driven apoptosis 35 , and its super-high levels caused activation of EMT 37 through β-catenin function. The current data showed Snol-A caused downregulation of CARF at the transcript and protein level. In view of the earlier reports that demonstrated that CARF represses p21 WAF1 transcription 36 , leading to pro-proliferation effect. In this context, Snol-A mediated inhibition of CARF was expected to cause upregulation of p21 WAF1 and was actually observed in Snol-A treated cells (Figs. 2, 3 and 8). Consistent to the upregulation of p21 WAF1 , downregulation of CDK/cyclins (essential for cell cycle progression) and growth arrest was observed (Figs. 2 and 8). We found that effect of Snol-A is similar to CARF siRNA and high dose of doxorubicin as reported earlier 32,35 yielding apoptosis through inhibition of ATR/Chk1 signaling (Figs. 4, 8 and S6). Furthermore, we found the decrease in CARF in Snol-A treated cells caused decrease in β-catenin and other proteins involved in cell migration and this was translated into attenuation of cell migration, invasion and EMT signaling (Figs. 5 and 8). Taken together, the data demonstrated that Snol-A could target CARF and bring about effects similar to CARF-compromise obtained by its specific shRNA suggesting that it is a natural inhibitor of CARF (Fig. 8). Consistent to the earlier reports on tumor suppressor effect of CARF shRNA, Snol-A caused delay in tumor growth and inhibited lung metastasis suggested that it may be recruited as a natural inhibitor of CARF for cancer treatment. www.nature.com/scientificreports www.nature.com/scientificreports/ Discussion CARF (Collaborator of ARF) was shown to be an essential nuclear protein that possesses a dose dependent control on cell proliferation 29,34 . CARF-compromised cells showed caspase-dependent apoptosis 35 . It was shown that CARF regulates and stabilizes p53 either directly or indirectly by binding to p14 ARF and HDM2, p53 positive and negative regulators, respectively 31 . In wild type p53 harboring cancer cells, CARF overexpression activated the DNA damage response pathway leading to growth arrest and senescence via activation of p53-p21 WAF1 axis 30 . On the other hand, at excessively high level of expression (super-expression), CARF caused pro-proliferation and malignant transformation of cancer cells by downregulation of p53 and DDR signaling 34,42 . In p53-compromised cancer cells, CARF overexpression was shown to cause caused pro-proliferation effect by transcriptional repression of p21 WAF1 36 . Furthermore, enriched level of CARF was found in a variety of cancer cells and clinical tumor samples 37 suggesting it to be a promising therapeutic target in aggressive malignancies. In the present manuscript, we have identified that Snol-A as a natural inhibitor of CARF.
Natural/herbal compounds have gained much attention for cancer therapy due to their safety and bio-availability as compared to conventional chemotherapeutic drugs that are usually expensive and often exert adverse effects. Soybeans have been reported to possess anti-cancer activity 25,27,28,[43][44][45][46][47][48][49][50] . Natural triterpenoids compounds are well known for their wide range of bioactivities and comprised promising effects against allergy 10 , metabolic syndrome 13 , diabetes 12 , inflammation 9 and cancer 14,15 . Soyasaponins are a group of complex oleanane triterpenoids proven to have diverse biological properties 23 of which the molecular mechanisms remain unclear. The active anti-cancer ingredients of soybeans have been found as different groups of soyasaponins that share a common core structure except the sugar moieties attached to carbon 3 and carbon 22. The compounds with shorter sugar chains have been suggested to be more active due to higher lipophilicity 25 . By comparative assays, we demonstrate that whereas Snin-I (3 sugar moieties) was non-toxic, Snol-A, (no sugar moieties) was toxic to many cancer cell types. Cytotoxicity of soyasponins have earlier been shown to be mediated by caspase-induced apoptosis or disruption of the mitochondrial functions 26 . Based on the cytotoxicity assays in a variety of cancer cells with variable p53 status, we found that Snol-A, but not Snin-1, was cytotoxic to p53 (null) cells as much as www.nature.com/scientificreports www.nature.com/scientificreports/ p53-wild type/ mutant cells. By molecular analysis, we identified that CARF is a specific target of Snol-A. Snol-A led CARF-inhibition attenuated its inhibitory effect on p21 WAF1 protein in p53-null cancer cells and resulted in growth arrest, mediated by decreased levels of p21 WAF1 -effector targets including Cyclin A, CDK2, CDK4 and cyclin D1. Higher doses of Snol-A caused apoptosis, mediated by activation of PARP 1/2, capase-3 and caspase-9. We had earlier reported that the suppression of CARF by siRNA induces cell death, essentially mediated via ATR-Chk1 35 signaling. Snol-A led CARF-inhibition was evidently marked by decrease in ATR and Chk1 levels and treated cells led to apoptosis via activation of Caspase-3 and −9. Snol-A treated cells showed similar expression profile of apoptosis markers and therefore, corroborated the findings (i) CARF-suppression could lead to cancer cell death and (ii) it could be achieved by Snol-A. As shown in Fig. 1, Snol-A treated U2OS cells (harbor wild type p53) showed inhibition of growth equivalent to the p53-null (SKOV-3 and Saos-2) cells. We subjected these cells to molecular analyses and found decrease in CARF endorsing its targeting by Snol-A (Fig. S6A). Consistent to the decrease in CARF, p53 also showed decrease in Snol-A treated cells (Fig. S6A,B). p21 WAF1 , however showed insignificant change. The latter is due to the fact that in wild type p53 harboring cells, p21 WAF1 expression is regulated by transcriptional activation of function of wild type p53 and transcriptional repression function of CARF. The latter also regulates p53 by positive feedback control [30][31][32][33][34] . In view of these, CARF targeting-driven upregulation of p21 WAF1 may be compensated by simultaneous downregulation of p53. Similar www.nature.com/scientificreports www.nature.com/scientificreports/ decrease in CARF in U2OS cells was obtained by high dose of doxorubicin in earlier studies that demonstrated dose-dependent dual control of CARF on cell proliferation 32,41,42 . Furthermore, we found that similar to the high dose of doxorubicin, Snol-A triggered apoptosis signaling in U2OS cells as endorsed by caspase cleavage (hence   [29][30][31][32][33]35,37,51 . Overexpression of CARF causes growth arrest through activation of p21 WAF1 signaling 29-33 , super-expression causes pro-proliferation/malignant transformation 34,36 and activation of EMT signaling 37 . CARF knockdown by siRNA and miRNA 32,35,51 , marked by grey text and lines causes apoptosis through ATR/Chk1 signaling. Current study showed that Snol-A is a natural inhibitor of CARF. Its modulated CARF functions yielding activation of growth arrest/apoptosis and decrease in EMT (shown by grey lines, at right and grey blocks at the bottom).

Scientific RepoRtS |
(2020) 10:6323 | https://doi.org/10.1038/s41598-020-62953-5 www.nature.com/scientificreports www.nature.com/scientificreports/ decrease in procaspase-3 and −9) and decrease in DNA damage regulatory proteins (ATR, pATR and Chk1) (Fig. S6C). Of note, U2OS cells have been shown to possess a low level of CARF expression 37 . Hence, a low dose of Snol-A may be sufficient for inhibition of CARF function yielding apoptosis in these cells. Amongst SKOV3 and Saos-2, both p53-null cells, the former showed higher cytotoxicity suggesting the involvement of factors other than p53 and p21 WAF1 signaling and warrant further investigations. We had recently reported that CARF plays a critical role in EMT process promoting malignant metastasis essentially via activating β-catenin transactivation 37 , while CARF suppression by siRNA attenuated EMT signaling 37 . In view of this, we tested the effect of Snol-A on this function of CARF. We found that Snol-A inhibited cancer cell migration and invasion properties. At the molecular level, Snol-A was found to decrease the malignant mesenchymal markers e.g. Vimentin, Smad2/3, hnRNP-K and N-cadherin in p53-null SKOV-3 cells. Furthermore, overexpression of CARF was found to reverse Snol-A-induced growth arrest, apoptosis, EMT activities. It clearly demonstrated that Snol-A specifically inhibits CARF and modulate diverse cellular activities involving CARF function. Consistent with the in vitro results, Snol-A reduced the growth and lung metastases of p53-null SKOV-3 tumors in nude mice assays. Taken together, in the present study we demonstrate that Snol-A (i) is a natural inhibitor of CARF, (ii) led CARF-suppression activates p21 WAF1, (iii) abrogates nuclear translocation and function of β-Catenin (iv) could be a promising potent natural chemotherapeutic drug for aggressive p53-deficient malignancies.
Immunofluorescence. Cells were harvested and seeded (4 ×10 4 /well) on glass coverslips in a 12-well plate for 24 h. Treatments with indicated Snol-A concentrations were given for 48 h. The cells were fixed in pre-chilled methanol at room temperature for 5 min and then permeabilized with phosphate buffered saline (PBS)-Triton-X-100 (0.1%) for 10 min followed by blocking with 2% bovine serum albumin (BSA) for 20 min. Cells were probed with antibodies as indicated (β-catenin, hnRNP-K, Vimentin, Fibronectin, CARF, Cleaved PARP1/2, CDK2, p21 WAF1 and Cyclin D1) at 4 °C overnight or at room temperature 1 h followed by incubation with Alexa Fluor-conjugated antibodies (Molecular Probes, USA) and then with Hoechst 33258 (Roche) for counterstaining. Immunofluorescence images were acquired on Carl Zeiss Axioplan-2 microscope equipped with Zeiss AxioCam HRc camera.
Wound healing assay. Migration of cells was observed using the Wound-healing assay. Monolayers of SKOV-3 cells were wounded by uniformly scratching the surface with a 20-gauge scrapper tip, followed by PBS washings twice. Cells were fed with fresh reduced-serum medium and were allowed to migrate into the wound for next 24 and 48 h. Images on different time-points were captured using Nikon phase-contrast microscope at 10X objective. Images were processed by ImageJ software. Covered area was calculated using ImageJ after threshold adjustment and selecting RGB stack type (Measurements setting was adjusted to calculate the area, area fraction, label display and limited to the threshold).
www.nature.com/scientificreports www.nature.com/scientificreports/ Matrigel invasion assay. SKOV-3 cells (2.5 × 10 4 ) were seeded into the upper invasion chamber, coated on the surface with 1/10 dilution of Matrigel (BD Biosciences, FL, NJ). Cells were allowed to invade to the lower chamber for next 24 h in control (DMSO) and Snol-A treated cells (0.5 and 2 μM) following the method described earlier 36 . Invaded cells were fixed in chilled methanol and stained in crystal violet. Invaded cells were counted using phase contrast microscopy.
Cell cycle analysis. Cells (SKOV-3, MDA-MB-231 and Saos-2) were seeded in 10-cm dishes and were treated with Snol-A at 60-70% confluence. Control and treated cells were harvested by trypsinization and centrifuged at 2000 rpm for 10 min at RT. Then cell pellet was washed with PBS by centrifugation again. 300 µl cold PBS was added to the pellet and mixed with 700 µl of cold 100% ethanol. The cell pellets were kept at −20 °C for 24 h followed by centrifugation twice (450 g at 4 °C for 5 min). Cells were washed two times with cold PBS with spinning down (450 g at 4 °C for 5 min). RNAse (100 μg/ml, Thermo Fisher Scientific) was added and mixed by slow vortex followed by incubation (37 °C, 1-2 h), centrifugation (450 g at 4 °C for 5 min), and cells were re-suspended in 200 µl Guava Cell Cycle Reagent (Millipore) followed by ~30 min incubation in dark at RT. Finally, cells were diluted in 0.5-1 mL volume and samples were acquired using Guava PCA-96 (Millipore) system. Acquired data was analyzed using ModFit LT (Version 5.0) software to distinguish the cell cycle profiles.
Luciferase reporter assay. SKOV-3 cells were transfected with pWWP-Luc (p21 WAF1 promoter; sequence 2.4 kbp) and control pRL-TK (Renilla Luciferase) control reporter plasmids using Lipofectamine (Thermo Fisher Scientific, USA) transfection reagent, as described earlier 36 . After 24 h of transfection, cells were treated with Snol-A at 6 µM concentration for 48 h, then lysates were prepared from SKOV-3 cells in passive lysis buffer. The luciferase activity was estimated using Dual-Luciferase Reporter Assay System (Promega, WI, USA) by using Infinite M200 PRO (Tecan, Switzerland) luminescent plate reader.
Colony formation assay. 500 cells/well were seeded and cultured for 2 days. On the 3 rd day, cells were treated with Snol-A and Snin-I (2, 4, 6, 8 μM) over 10-14 days. Colonies were rinsed with cold PBS and fixed (methanol: acetic acid (1:1)) at RT for 5-10 min. Next, 0.5% crystal violet stain was added at RT for 2 h, followed by washing and left to dry at RT overnight, colonies were counted using stereomicroscope.
In vivo xenograft assay. Athymic balb/c nude female mice (4-week-old) were purchased from NihonClea, Japan and acclimatized for 2 weeks. In a preliminary experiment, the effect of a range (5-25 mg/kg BW) of Snol-A doses was tested and a reduction in tumor growth in the range of 10-20 mg/kg BW was observed. Based on this preliminary data, we chose 15 mg/ kg BW Snol-A for further in vivo assays. Animals were pre-fed with either vehicle (0.1% carboxymethyl cellulose (CMC)) or Soyasapogenol-A supplemented vehicle (Snol-A 15 mg/ kg BW) in 250 µL suspension twice a week. SKOV3 cells (5 ×10 6 in 200 µL PBS) were injected subcutaneously (for subcutaneous xenograft) over the left and right thigh flanks, and intravenously (for metastases) through tail vein injections. Upon emergence of tumor buds from xenografts, mice were regularly fed thrice a week for 4 weeks. Tumors were regularly monitored and sized with Vernier caliper. Tumor volume (V) was calculated using the formula V = (LxW 2 )/2 with caliper measurements of length (L) and width (W). Mice were sacrificed by post anesthesia cervical dislocation before tumors grew to about 1.5 cm length and examined for tumors in internal organs and lung metastasis. All animals in randomized groups were closely monitored for activity during-and after-treatments, and physiological observations for skin rash or eczema scars was also regularly carried out. This study was carried out in strict accordance with the recommendations from the Animal Experiment Committee, Safety and Environment Management Division, National Institute of Advanced Industrial Science & Technology (AIST), Japan. The experimental protocols were approved by AIST (Experimental plan approval #2017-025).

Statistical analysis.
All experiments were performed in triplicates. The data were expressed as mean ± SEM. Statistical analyses were executed using Student's t-test or nonparametric Mann-Whitney U-test; whichever was applicable. Statistical significance was defined as p-value ≤ 0.05. The p value represents *<0.05, **<0.01, ***<0.001