Cutaneous melanoma is a fatal neoplasm with increasing incidence worldwide

Cutaneous melanoma is a fatal neoplasm with increasing incidence worldwide. Mutated proto-oncogene BRAF is the bona fide therapeutic target for about half of all melanomas. Regrettably, melanoma acquires resistance to BRAF inhibitors, e.g., vemurafenib (PLX4032) casting doubt on this promising melanoma targeted therapy. There is an urgent need for new therapeutic approaches to fight BRAF mutant melanoma. In this study, we explored the potential bioactivity of triterpenoid saponin cumingianoside A (CUMA), isolated from leaves and twigs of Dysoxylum cumingianum against PLX4032-resistant BRAFV600E mutant melanoma A375-R in vitro and in vivo. Our data show that CUMA treatment inhibited A375-R melanoma cell proliferation in a time- and dose-dependent manner. CUMA also suppressed the activity of CDK1/cyclin B1 complex and led to G2/M-phase arrest of A375-R cells. Furthermore, CUMA treatment resulted in induction of apoptosis as shown by the increased activation of caspase 3 and caspase 7, and the proteolytic cleavage of poly(ADP-ribose) polymerase (PARP), and activation of autophagy as shown by the increased expression of autophagy-related genes and increased formation of autophagosomes. Moreover, we found that CUMA treatment induced ER stress response mainly through the IRE1? axis and co-treatment with an ER stress inhibitor (4-PBA) could partially attenuate apoptosis and autophagy induced by CUMA. Importantly, orally administered CUMA as a single agent or in combination with PLX4032 exhibited strong tumor growth inhibition in a PLX4032-resistant A375-R xenograft mouse model, and with little toxicity. This is the first report to explore the anti-tumor activity of CUMA in vitro and in vivo mechanistically, and our results imply that this triterpenoid saponin may be suitable for development into an anti-melanoma agent.
Keywords: BRAF inhibitor-resistant melanoma, triterpenoid saponin, cumingianoside A apoptosis, ER stress, autophagy
Melanoma is a significant public health problem with a rapidly increasing rate of incidence worldwide (Eggermont et al., 2014). It is the most lethal of the skin cancers with high metastatic potential and is notoriously refractory to conventional therapy. Half of all melanoma patients carry activating mutations in proto-oncogene v-Raf murine sarcoma viral oncogene homolog B (BRAF) that cause constitutive mitogen-activated protein kinase (MAPK) signaling and subsequently, unrestricted melanoma growth (Kumar et al., 2004; Eskiocak et al., 2017). Consequently, targeted therapies that specifically inhibit this hyperactive oncogene have revolutionized melanoma treatment. One such example of BRAFV600E targeted therapy, vemurafenib (PLX4032) has shown unprecedented clinical efficacy (Chapman et al., 2011); however, despite its remarkable efficacy, melanoma patients receiving PLX4032 therapy relapse within months. The primary clinical mechanism of acquired melanoma resistance to PLX4032 and other BRAF inhibitors (BRAFi) is the reactivation of MAPK pathway signaling which consequently leads to activation of dysregulated proliferation, aberrant cell cycle progression, and resistance to apoptosis. However, administration of inhibitors of MAPK signaling (e.g., MEK1/2 inhibitors), did not result in any substantial clinical improvement (Duggan et al., 2017; Lim et al., 2017). Therefore, there is an urgent need for new treatment modalities to treat or sensitize PLX4032-resistant melanoma.
The cell cycle is controlled by checkpoints that consist of cyclin-dependent kinase (CDK)-cyclin complexes which orchestrate progression from one phase to another. Various anticancer agents exert their antiproliferative effects by modulation of the cell cycle regulatory units which leads to growth arrest and consequently apoptosis. There are various proapoptotic stimuli that can result in apoptosis or type I programmed cell death, where typical morphological changes can be observed as the result of intracellular proteolytic cascade activation, for example, cell shrinkage and condensation, membrane blebbing and adhesion loss. Autophagy, or type II programmed cell death, is a dynamic catabolic process which sequesters damaged intracellular components into double-membrane vesicles (autophagosomes) and delivers them into the lysosome for degradation. ER stress is a condition triggered by impaired ER structure and function, caused by events such as pharmacological stress leading to disruption of ER homeostasis and accumulation of misfolded or unfolded proteins within the ER lumen. The cell senses ER stress by three ER transmembrane sensor proteins, RNA-activated protein kinase–like ER kinase (PERK), activating transcription factor 6 (ATF6), and inositol-requiring enzyme 1 (IRE1). By activating the unfolded protein response (UPR) and restoring ER homeostasis, ER stress can be cytoprotective, but when stress is sustained or severe, ER stress can become a cytotoxic signal, mainly by activation of the intrinsic apoptotic pathways. In BRAFi-resistant melanomas, activation of the ER stress response can lead to the induction of autophagy. However, the functional relationship between autophagy and ER stress remains controversial, and both phenomena together determine the fate of the cell. Emerging evidence suggests that manipulating autophagy and ER stress response in melanoma with acquired resistance to PLX4032 as a promising therapeutic approach (Cerezo et al., 2016) (Giglio et al., 2015).
Natural products remain an endless frontier for discovery in oncology research due to their novel chemical skeletons, distinct pharmacological activities and low toxicity profile. For instance, plants of the genus Dysoxylum (Meliaceae) have been well documented as a source of structurally diverse chemical constituents with a broad spectrum of pharmacological activities, including antimicrobial, immunomodulatory and anticancer activity add 2-3 refs. In particular, Dysoxylum cumingianum a tree species found in Taiwan, Malaysia and Philippines is a rich source of bioactive triterpenes and triterpenoid saponins (Toshihiro Fujioka, 1997b; a; Yoshiki Kashiwada, 1997; Kurimoto et al., 2011; Shin-ichiro Kurimoto and Kuo-Hsiung Lee, 2011). The triterpenoid saponin cumingianoside A was characterized as one of the major constituents in the leaves of Dysoxylum cumingianum and was shown to possess anti-cancer activities against various human cell lines including human melanoma cells; however, the detailed anti-cancer mechanism remains unexplored (Yoshiki Kashiwada and Lee, 1992) {Toshihiro Fujioka, 1995 #439}{, 1997 #393}.
In this study, we demonstrated the efficacy of CUMA against A375-R, BRAFV600E mutated human melanoma with acquired resistance to PLX4032 in vitro and in vivo. Mechanistically CUMA inhibited A375-R growth by inducing G2/M phase cell cycle arrest, and ER-stress related apoptosis. Orally administered CUMA significantly and dose-dependently reduced tumor growth in A375-R melanoma xenograft mice that had no efficacy to PLX4032. Furthermore, CUMA and PLX4032 combination treatment at reduced administration frequency exhibited comparable melanoma growth inhibition to single, low dose CUMA treatment, indicating that CUMA sensitizes PLX4032 against A375-R melanoma.

Chemicals and Reagents
Dulbecco’s Modified Eagle Medium (DMEM), Minimum Essential Media (MEM), Roswell Park Memorial Institute 1640 (RPMI 1640), fetal bovine serum (FBS), and the mixture of 100 U/ml penicillin and 100 µg/ml streptomycin were purchased from Invitrogen (Carlsbad, CA, USA). Dimethyl sulfoxide (DMSO), crystal violet, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), 4?,6-diamidino-2-phenylindole (DAPI), Bafilomycin A1 (Baf A1), 3-methyladenine (3-MA), Chloroquine (CQ), 4-phenylbutiric acid (4-PBA) and Thapsigargin (Tg) were supplied by Sigma-Aldrich (St. Louis, MO, USA). PLX4032 was purchased from Selleckchem (Houston, TX, USA). Silica gel was purchased from Merck (Darmstadt, Germany). All chemicals and solvents used in the study were of reagent or high-performance liquid chromatography (HPLC) grade.

Plant Materials
A voucher specimen (code No. ??) was deposited in the School of Pharmacy, College of Medicine, National Taiwan University, Taiwan.
Isolation and Identification of CUMA can I put figure in methods
The plant material, Dysoxylum cumingianum was collected from Orchid Island, Taiwan, in April 2012 and identified by one of the authors (Y.C.S.). The CUMA isolation and purification protocols were modified and simplified from previously published studies (Kashiwada, 1992; Kurimoto et al., 2011). The acetone extract from the leaves and twigs of Dysoxylum cumingianum was partitioned to yield EA-fraction which was further subjected to extensive chromatographic separation using Sephadex LH-20 column, silica gel column, and in the final step purified by preparative reverse phase high-performance liquid chromatography (Cosmosil 5C18-AR-II column, Nacali Tesque, Kyoto, Japan) (Figure S1) to obtain pentacyclic triterpene glucoside, cumingianoside A (CUMA, Figure 1A) with ?95% purity as judged by NMR spectrometry (AVII-500 NMR spectrometer, Brüker, Karlsruhe, Germany) (Figure S2B and S2C). The structure was elucidated as 3-O-acetyl-3?,7?,23,24,25-pentahydroxy-14,18-cycloapoeuphanyl 7-O-?-D-(6?-O-acetyl) glucopyranoside on the basis of 1H and 13C NMR and electron spray ionization mass spectrometry (ThermoFinnigan LCQ) (Figure S2) and in comparison to previously published results (Kashiwada, 1992).

Cell Culture
Human melanoma cell lines A375 (ATCC CRL-6475), A2058 (ATCC CRL-11147), SK-MEL-2 (ATCC HTB-68), MeWo (ATCC HTB-65), murine melanoma cell lines B16 (ATCC CRL-6322), B16-F10 (ATCC CRL-6475) and primary epidermal melanocytes (ATCC PCS-200-012) were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). A375, A2058 and B16-F10 cells were cultured in DMEM, MeWo, SK-MEL-2, and melanocytes were cultured in MEM, and B16 was cultured in RPMI 1640, supplemented with 10% heat-inactivated fetal bovine serum, 100 U/mL penicillin and 100 µg/mL streptomycin at 37°C in a humidified 5% CO2 incubator.

Cell Viability
Viability studies were carried out by using MTT-based colorimetric assay which quantitatively measures metabolic activity of the viable cells as published elsewhere (Yang et al., 2018). Briefly, cells (5 ? 103 to 1 ? 104 per well) were seeded in 96-well plates and incubated overnight. Test compounds/inhibitors were dissolved in DMSO and diluted appropriately in a culture media to a final concentration of 0.5% DMSO. Cells were then treated with various concentrations of test compounds/inhibitors and equal volumes of vehicle (0.5% DMSO) for the indicated times, and further incubated for 3 h with media containing 20 µM MTT reagent. Then, the media was replaced by DMSO and absorbance at 570 nm was measured by ELISA reader. A dose-dependent inhibition curve was used to calculate the IC50 (maximal concentration of the tested compound/inhibitor to cause 50% inhibition of the cell viability) values. The data are presented as mean ± SD from four technical repeats and three independent experiments.

Western Blot Analysis
Western blot analyses were performed as described previously (Chiang et al., 2005). Briefly, total cellular proteins were extracted using radio-immunoprecipitation assay (RIPA) lysis buffer (Santa Cruz Biotechnology, Dallas, TX, USA) containing protease and phosphatase inhibitors. Protein concertation was measured using a colorimetric detergent-compatible protein assay kit (Bio-Rad, Hercules, CA USA) according to the manufacturer’s protocol. Proteins were separated by 10% or 15% SDS-PAGE, and transferred onto a polyvinylidene difluoride membrane (Merck Millipore, Burlington, MA, USA). Blots were blocked in washing buffer (Tris-PBS/0.1% v/v Tween 20) containing 5% w/v skimmed milk for 2 h at room temperature and then incubated with specific antibodies for 16 h at 4°C. After washing, blots were probed with appropriate (anti-rabbit, -mouse or -goat) horseradish peroxidase-conjugated secondary antibodies for 3 h at room temperature. Reactive protein bands were detected using enhanced chemiluminescent detection kit (Thermo Fisher Scientific, Waltham, MA, USA) by exposure to chemiluminescence film, Amersham Hyperfilm ECL (GE Healthcare, Chicago, IL, USA) and quantified by using ImageJ software. Primary antibodies against caspase-7 (cat. #9492), cleaved caspase-7 (cat. #9491), Bim (cat. #2933), phospho-ERK1/2 (cat. #9101), MEK1/2 (cat. #4694) and phospho-MEK1/2 (cat. #9121) were purchased from Cell Signaling Technology; caspase 3 (sc-56053), PARP (sc-7150), Bcl-2 (sc-7382), cyclin B1 (sc-594), CDK1 (sc-54), phospho-CDK1 (sc-12341), Cdc25C (sc-327), phospho-Cdc25C (sc-12354), ERK (sc-94), MAP LC3B (sc-376404), IRE1? (sc-20790), ATF-6? (sc-166659), PERK (sc-377400) were purchased from Santa Cruz; and actin (MAB1501) was supplied from Merck Millipore. Three independent experiments were performed to confirm the reproducibility of the data.

Animal Studies in Mice
The animal experiment to assess the efficacy of compound treatment against A375-R PLX4032-resistant melanoma xenograft was performed according to a protocol approved by the Institutional Animal Care and Utilization Committee (IACUC) of Academia Sinica (Taipei, Taiwan). Six-week-old female NOD/SCID mice bred in the Laboratory Animal Core Facility at the Agricultural Biotechnology Research Center, Academia Sinica were given a distilled water and standard laboratory diet ad libitum and kept in a 12 h light/dark cycle at 22 ± 2°C. A375-R cells (3 ? 106) were subcutaneously implanted into the right flank of the mice, except for the sham group, and eight days later when the tumor volume reached 100 mm3 the mice were randomly divided into six groups (five mice in the sham group and six mice in every other group) and orally treated with vehicle (5% DMSO and 0.1% Tween 80 in 0.2 ml of PBS; Tumor control), CUMA (50 and 75 mg/kg body weight; CUMA50 and CUMA75), PLX4032 (50 mg/kg body weight; PLX4032) and CUMA and PLX4032 in combination (CUMA 50 mg/kg body weight and PLX4032 50 mg/kg body weight; CUMA50+PLX4032) once daily in each treatment group except in the combination treatment group in which CUMA and PLX4032 were given alternatively every other day. Tumor dimensions were measured with calipers every three days from the beginning of the treatment and the tumor volumes were calculated by formula V = (Length ? Width2)/2 (Feng et al., 2016). Tumor growth inhibition (TGI) was calculated by the formula TGI (%) = (Vc ? Vt)/(Vc ? V0) ? 100, where Vc and Vt represent the mean group tumor volume of the control and treated groups, respectively at the end of the study (day 29) and V0 at the initiation of the treatments (day 8). Body weights were measured every three days and the percentage of body weight loss was calculated by formula: (BW – BW0)/BW0 ? 100, where BW represents mean body weight of the treated group at day 29 and BW0 at day 8 (DePinto et al., 2006). At the end of the study, the mice were euthanized and tumors and organs (liver and kidney) were excised and prepared for histological analysis.

Histology and Immunohistochemistry
The formalin-fixed and paraffin-embedded tumor and organ (liver and kidney) tissues of the test mice were microtome sectioned (5 µm), heat immobilized, deparaffinized in xylene and rehydrated in a graded series of ethanol to water. Organ samples were subjected for hematoxylin and eosin staining (H&E) and tumor samples were subjected for immunohistochemical staining (IHC). After antigen retrieval, the following antibodies were used for immunohistochemistry analysis: cleaved caspase-3 (cat. #9661), and cleaved PARP (cat. #5625) from Cell Signaling Technology, VEGF (19003-1-AP) from Proteintech, Ki67 (ab15580) and LC3A/B (ab58610) from Abcam. We used histofine polymer detection system (Nichirei Biosciences, Tokyo, Japan) for detection of the primary antibodies, and 3,3?diaminobenzidine tetrahydrochloride reagent (Leica Biosystems, Wetzlar, Germany) for color development. Hematoxylin (Muto Pure Chemicals, Tokyo, Japan) was used to counterstain the nuclei. Images were captured on a Zeiss AxioImager Z1 microscope (Munich, Germany) using a Zeiss AxioCam HRc camera and processed using AxioVision Rel.4.9.1 Software.

Colony Forming Assay
Colony forming assay was performed with modification from previously published method. A375-R cells (3 ? 103 per well) were plated in 24-well plates overnight and then treated with the indicated concentrations of CUMA or an equal volume of vehicle for six days, the amount of time that is needed for A375-R to form a colony comprising of 50 cells. Colonies were fixed with methanol, stained with crystal violet and photographed. Inhibition of the colony formation was quantified by measuring the absorbance of crystal violet at 595 nm of the wells containing compound treated cells and comparing with the wells of vehicle treated cells.

Cell Cycle Analysis
A375-R cells (2 ? 105 per well) were seeded overnight in 6-well plates and then treated with different concentrations of CUMA or an equal volume of the vehicle for 12 h, 24 h, and 48 h, respectively. Cells were trypsinized, washed with PBS and fixed with 70% ethanol overnight at 4°C. Cells were incubated for 30 min at room temperature in a PBS buffer containing RNase (100 µg/ml), 0.1% triton X-100, and propidium iodide (10 µg/ml). Cell cycle distribution was analyzed by flow cytometer Accuri C6 (BD Biosciences, San Jose, CA, USA).
Apoptosis Assay
A375-R cells (2 ? 105 well) were seeded overnight in 6-well plates and then treated with vehicle or CUMA for 48 h and 72 h. After treatment, cells were harvested, washed with PBS and incubated for 15 min at room temperature in 1 ? binding buffer containing propidium iodide and FITC-Annexin V as suggested by the manufacturer (BD Pharmingen, San Diego, CA, USA). Apoptotic cells were analyzed by flow cytometer Accuri C6 (BD Biosciences).
Immunofluorescence Cell Staining
A375-R cells (4 ? 104 per well) were seeded on coverslips overnight in 24-well plates and then treated with CUMA (20 µM) or equal volume of vehicle for 2 h. Cells were fixed by methanol, blocked with PBS containing 3% bovine serum albumin, and stained with primary antibody (LC3B) and FITC conjugated secondary antibody (Jackson ImmunoResearch, West Grove, PA, USA) with 1:200 dilution. The cell nuclei were stained with DAPI.
Statistical Analysis
Quantification of all experimental data are represented as mean ± standard deviation (SD) with the number of experiments indicated in the figure legends. Statistical analysis was conducted by PASW Statistics 18 and significant differences within treatments were determined by one-way ANOVA. P ? 0.05 was considered statistically significant.

CUMA Inhibits Melanoma Cell Proliferation
To assess the anti-melanoma activity of CUMA in vitro, human (A375, A2058, SK-MEL-2, and MeWo) and mouse (B16 and B16F10) melanoma cell lines with different mutational status were treated with CUMA (1 to 25 µM) for 24 h and 48 h and cell viability was determined by MTT assay. CUMA exhibited potent and dose-dependent growth inhibitory effects against all melanoma cell lines harboring BRAF mutation (A375, A2058, B16) or NRAS mutation (SK-MEL-2, B16F10) regardless of whether they were of human or mouse origin as evidenced by the IC50 ranging 15.1 to 22.9 µM and 11.8 to 15.5 µM at 24 h and 48 h treatment, respectively; whereas, for wild-type BRAF and NRAS melanoma cell line (MeWo) IC50 of 24.2 µM was observed only after 48 h treatment (Table 1). Importantly, IC50 of normal human melanocytes was not observed within tested compounds with a concertation ranging up to 25 µM (Table 1).
As BRAF inhibitor PLX4032 induced resistance in melanoma patients bearing BRAFV600E mutation, we also determined the efficacy of CUMA on A375-R, an in-house established BRAFV600E mutant melanoma cell line with acquired resistance to PLX4032 (Feng et al., 2016). Interestingly A375-R cells showed a higher sensitivity to CUMA as compared to the parental A375 cells with IC50 values of 15.8 vs. 22.9 µM at 24 h and 11.8 vs. 15.3 µM at 48 h, respectively (Table 1). We thus focused our investigation on the anti-melanoma activity of CUMA against PLX4032-resistant A375-R melanoma cells, and the underlining mechanisms were examined in the following study.
To analyze the long-term antiproliferative effect of CUMA on the malignant growth of A375-R cells we used a colony-forming assay. As presented in Figure 1B, CUMA treatment for six days dose-dependently inhibited colony formation in A375-R cells and with 88% inhibition at 10 µM CUMA. The antiproliferative effect of CUMA was reflected in the decreased confluence of A375-R compared to the vehicle-treated cells at 24 h, as observed by phase-contrast light microscopy (Figure 1C). After 48 h, CUMA treated cells gained typical apoptotic morphological changes including cytoplasmic shrinkage and membrane blebbing (Figure 1C). Therefore, the effect of CUMA on A375-R cell cycle progression and apoptotic cell death were further examined.

CUMA Induces G2/M Phase Cell Cycle Arrest in A375-R cells
The regulatory activity of CUMA on the cell cycle of A375-R cells was examined using flow cytometry. As presented in Figure 2A, CUMA induced typical dose-dependent and time-dependent G2/M phase arrest. When cells were treated for 48 h with vehicle and increasing concentrations of CUMA (10 µM, 15 µM, 20 µM), the population of cells in G1 phase was decreased from 64% in the vehicle group to 56% and 51% (15 µM and 20 µM CUMA), S phase population was decreased from 12% to 10% and 9% (15 µM and 20 µM CUMA); whereas in the G2/M population it showed an increase from 24% to 34% and 40% (15 µM and 20 µM CUMA) (Figure 2A). At 20 µM CUMA, significant cell cycle arrest was observed earliest at 12 h with 37% of cell population undergoing G2/M phase arrest. The longer treatments of 24 h and 48 h resulted in comparable effects; 34% and 40%, respectively (Figure 2A). The molecular mechanism for CUMA-induced cell cycle arrest was examined by immunoblotting of CDKs and their associated cyclin partners which play critical roles in the G2 to M transition. Treatment with 20 µM CUMA induced a time-dependent decrease in the protein expression of cyclin B1 and its associated partner CDK1 (Figure 2B). CDK1/cyclin B1 activity which is essential for the onset of mitosis was also suppressed by blocking the (activation of) activating phosphorylation of CDK1Thr161 and by reducing the expression of the activating CDC25C protein phosphatase which regulates CDK1 activity by dephosphorylating its inhibitory sites (Figure 2B). These data demonstrate that CUMA inhibited A375-R cell proliferation by inhibition of CDK1/cyclin B1 activity and arrested cells at the G2/M boundary.

CUMA Induces Apoptotic Cell Death in A375-R
To gain insight into the cell-death mechanism, A375-R cells were treated with the indicated concentrations of CUMA for 72 h, and the apoptotic ratio was analyzed by flow cytometry using Annexin V/PI double staining. The percentage of the apoptotic cell population increased from 5.4% in the vehicle group to 9.5%, 36.3%, and 60.7% in the 10 µM, 15 µM and 20 µM CUMA treated cells, respectively (Figure 3A and 3B). CUMA at 20 µM induced comparable apoptotic death to cisplatin (58.7%) which was used as a bona fide control for induction of apoptosis (Figure 3A and 3B).
To characterize the molecular mechanism of the apoptotic effect upon CUMA treatment, apoptosis-related factors were examined by western blot analysis. CUMA treatment for 48 h and 72 h at 20 µM markedly elevated the levels of the cleaved and activated forms of executor caspase 3 and caspase 7 (Figure 3C). Concordantly, the inactive precursor procaspase 3 was decreased, indicating its activation (Figure 3C). DNA repair enzyme poly ADP-ribose polymerase (PARP) showed decreased activity as observed by the elevated PARP cleavage accompanied by a decrease in the original form (Figure 3C). Several Bcl-2 family proteins regulate the intrinsic mitochondria death pathway and we found that CUMA reduced the expression of anti-apoptotic Bcl-2 and induced expression of cytotoxic splice variants of the pro-apoptotic proteins BimEL and BimL but did not change the expression of anti-apoptotic Bax (Figure 3C). Taken together, these results suggest that the mitochondria-dependent pathway might be involved in CUMA-induced A375-R apoptotic cell death.

CUMA Alone or in Combination with PLX4032 Suppresses Tumor Growth in an A375-R Melanoma Orthotropic Xenograft Model with Acquired Resistance to PLX4032
We used a A375-R melanoma xenograft model with acquired resistance to PLX4032 to evaluate the potential therapeutic efficacy of CUMA in vivo against PLX4032-resistant melanoma growth. The anti-melanoma activity of CUMA and PLX4032 in combination was also tested. The detailed experimental design is presented in Figure S3A. Figure S3B shows all tumor tissues excised from test mice. PLX4032 treatment (50 mg/kg daily, 22 doses in total) resulted in only minor tumor growth inhibition (TGI) of 25% and tumor weight reduction by 26% and with no statistically significant difference compared to the tumor control group (Figure 4A and 4B). CUMA low dose (50 mg/kg daily, 22 doses in total) and CUMA high dose (75 mg/kg daily, 22 doses in total) treatment significantly reduced tumor growth by 52% TGI and 67% TGI, and reduced tumor weight by 41% and 59%, respectively, indicating dose-dependent melanoma growth inhibition (Figure 4A and 4B). Notably, treatment with CUMA (50 mg/kg every other day, 11 doses in total) and PLX4032 (50 mg/kg every other day, 11 doses in total) in combination at reduced administration frequency of either compound resulted in similar effects on test animals to CUMA low dose, with 44% TGI and 43% tumor weight reduction, suggesting that CUMA sensitizes PLX4032 against A375-R tumors (Figure 4A and 4B). CUMA50 and CUMA75 induced slight body weight loss (6.5% and 10.2%, respectively), while CUMA50+PLX4032 combination treatment caused negligible mouse body weight loss (1.4%) (Figure 4C) indicating that CUMA and PLX4032 combination treatment is favorable to animal health. Further histopathological data showed no notable changes in the structure and morphology in the liver and kidney among sham, tumor control and treatment groups (Figure 4D), suggesting that CUMA treatment does not exhibit significant animal toxicity.
In vivo, the CUMA anti-tumor effect was observed together with a significant reduction in proliferation as detected by Ki67 positive staining regardless of the dose used and a similar effect was observed in CUMA50+PLX4032 treatment (Figure 4E). PLX4032 treatment showed a higher proliferation rate and Ki67 positive staining similar to the tumor control (Figure 4E). We also observed a marked enhancement in apoptosis as determined by elevated expression of cleaved caspase-3 and increased cleavage of PARP in CUMA-treated animals and higher in CUMA75 compared with the low dose CUMA50 and CUMA50+PLX4032 combination treatment (Figure 4E). Apoptosis markers were slightly increased by PLX4032 and comparable to the tumor control (Figure 4E).
The CUMA treatment significantly impaired angiogenesis in the tumors as detected by decreased VEGF staining, compared to the PLX4032 and tumor control (Figure 4E). These data demonstrated that oral administration of CUMA inhibits the A375-R PLX4032-resistant melanoma growth dose-dependently and sensitizes PLX4032 to A375-R tumors.

CUMA Induces ER-stress Related Apoptosis and Autophagy in A375-R cells
Induction of prolonged ER stress or suppression of ER stress adaptation mechanisms were proposed as alternative strategies to overcome PLX4032 resistance in melanoma cells (Beck et al., 2013; Cerezo et al., 2016). We used immunoblotting to check the expression of PERK, ATF6 and IRE1? in A375-R cells upon CUMA treatment as representative markers of ER stress response. CUMA treatment (20 µM) resulted in the initiation of ER stress, concluded shown by the time-dependent increase in the expression of IRE1?, but with no significant changes in the expression of PERK or ATF6 (Figure 5A). The cleavage form of apoptotic hallmark PARP was increased upon the compound treatment (Figure 5A).
Increasing evidence shows that prolonged ER stress cause not only apoptosis but also another type of programmed cell death, autophagic-cell death (Salazar et al., 2009; Cerezo et al., 2016). Our quantitative real-time polymerase chain reaction (qPCR) data showed that expression of autophagy-related genes including ATG13, ATG12, LC3B, LAMP2 were increased in A375-R cells treated with 20 µM CUMA after 24 h (Figure 5B). To test the effect of CUMA in the induction of autophagic-cell death, we used two different approaches: fluorescence microscopy to visualize the accumulation of LC3B puncta; and immunoblotting to measure the conversion of LC3B-I to LC3B-II as those are indicators of autophagosomes formation (Klionsky et al., 2016; Murugan and Amaravadi, 2016). As expected, CUMA promoted the accumulation of autophagosomes as observed by the increased LC3B puncta and enhanced LC3B fluorescence compared to the vehicle-treated cells (Figure 5C). Accumulation of autophagosomes may result from induction (activation of autophagy) or blockage (inhibition of autophagy) of autophagy flux. As presented in Figure 5D, the level of LC3B-II induced by CUMA can be further enhanced when co-incubated with autophagy inhibitor Bafilomycin A1 (Baf A1), suggesting that CUMA increased the autophagic flux, rather than blockage of its degradation (Figure 5D). The activation of autophagy was observed in vivo as well, as indicated by the increased LC3 staining in the tumors of mice treated with CUMA, PLX4032 and CUMA in combination, but to a much lesser degree in PLX4032 treated mice (Figure 5E).
To further investigate the role of CUMA-induced ER stress in cell death we co-treated A375-R cells with CUMA and a chemical chaperone 4-phenylbutiric acid (4-PBA) which attenuates ER stress by promoting protein folding and protein stabilization (Zhang et al., 2013). Thapsigargin (Tg) was used as a positive control for ER stress (Healy et al., 2009) and autophagy induction. As expected, 4-PBA reduced the expression of IRE1? both basally and in response to CUMA (Figure 5F). Interestingly, CUMA-induced cleavage of PARP was moderately reversed when co-treated with 4-PBA, implying that CUMA-induced ER stress is partially responsible for the A375-R apoptosis (Figure 5F). Furthermore, conversion of LC3B-I to LC3B-II was also reduced suggesting that autophagy may be the downstream effect of CUMA-induced ER stress (Figure 5F). Emerging evidence shows that autophagy plays a dual role in tumor cell death, cytoprotective, to restore the cellular homeostasis, or cell damaging to promote cancer cell death (Scarlatti et al., 2009; Tomic et al., 2011; Liu and Debnath, 2016, Giglio et al., 2015)). To investigate the role autophagy plays in CUMA-induced cell death, we used two autophagy inhibitors, 3-methyladenine (3-MA), which blocks autophagosome formation and chloroquine (CQ), which blocks autophagosome-lysosome fusion (Klionsky et al., 2016). However, when A375-R cells were first pretreated for one hour with either 3-MA (4 mM) or CQ (40 µM) and then additionally treated with CUMA for 24 h, there was no significant alteration in A375-R cell viability compared to CUMA-only treated cells (Figure S4B). The role of autophagy in CUMA-induced cell death and interplay with ER stress requires further evaluation.

PLX4032 shows prominent inhibition of BRAFV600E melanoma cell proliferation by arresting the cells at G1-phase of the cell cycle; however, when melanoma cells acquire resistance to the drug, PLX4032 is not able to control the cell proliferation (ref). A large body of evidence shows that melanomas with acquired resistance to PLX4032 have deregulation at the checkpoints important for orchestrating cell cycle progression, and CDK inhibitors show promising results in preclinical models of melanoma (ref). Here we demonstrated that the triterpene glucoside CUMA significantly suppresses the growth of BRAFV600E mutant melanoma with acquired resistance to PLX4032 in vitro and in vivo. CUMA effectively induced G2/M cell cycle arrest and inhibited proliferation of A375-R cells, in part through inhibition of the CDK1/cyclin B1 complex important for the transition of G2/M, and the level of the CDC25C and its active form essential for promoting the transition to M-phase (Table 1, Figure 2).
It is known that during ER stress, IRE1? acts as a switch between cell survival and cell death. In hepatoma cells, overexpression of IRE1? inhibited cell growth and repression of IRE1? inhibited ER stress-related apoptosis (Li et al., 2012). In our study, we observed that the increased expression of IRE1? was accompanied by increased cleavage of PARP (Figure 5A), suggesting that A375-R cell apoptosis might be the consequence of CUMA-induced ER stress. Further, CUMA treatment increased the levels of the pro-apoptotic molecule Bim which is mainly involved in ER stress-induced apoptosis (Figure 3C). However, further studies, such as knockdown of ER stress sensors (e.g., IRE1?) in A375-R cells are needed to investigate whether the apoptotic cell death caused by CUMA is mainly through the ER stress response pathways.
The role of autophagy in cancer is complex. It is pro-survival by clearing the damaged intracellular components and providing nutrients for facilitating cancer cell growth, or cell damaging when excessive autophagy leads to irreversible cellular function impairment (refs). We found that inhibition of autophagy with the well-known autophagy inhibitors 3-MA and CQ did not alter the antiproliferative effect of CUMA on A375-R (Figure S4B). However, our immunoblotting results showed that the CUMA-induced conversion of LC3B was reduced when co-treated with an ER stress inhibitor (4-PBA) (Figure 5F). These data implied that activation of autophagy in A375-R might be provoked by the CUMA-induced ER stress. On the other hand, prolonged incubation with CUMA led to A375-R apoptosis as observed by activation of apoptotic hallmarks, caspase 3, caspase 7 and PARP; in addition, the anti-apoptotic protein Bcl-2 involved in mitochondria intrinsic cell death was decreased (Figure 3C). We have observed that CUMA treatment induced a 1.5 fold increase in ROS levels (data not shown); however, it is not clear whether this phenomenon will lead to mitochondrial damage and activation of apoptosis in PLX4032 resistant melanoma cells; this point that will need further investigation.
Reactivation of RAF/MEK/ERK signaling is the central mechanism that leads to acquired resistance in BRAFV600E mutant melanoma, and treatment fails in 50% of melanoma patients (ref). In our mechanistic study, we observed that CUMA does not suppress the activity of MEK and ERK (Figure S4A). However, we found that CUMA is effective on A2058 BRAFV600E melanoma cells which are intrinsically resistant to BRAF inhibitors (Table 1), suggesting that CUMA might work by a different mechanism than BRAF and MEK inhibitors to suppress tumor growth.
The significance of CUMA is highlighted by the inhibition of the growth of A375-R tumors with acquired resistance to PLX4032 in animals. CUMA significantly inhibited cell proliferation and angiogenesis in the tumor tissues, and induction of tumor cell apoptosis was also observed (Figure 4E), which is in good agreement with the data obtained from in vitro assays for the CUMA inhibitory effect against PLX4032 resistant A375-R cells. This study is the first to demonstrate the pharmacological activities of CUMA against drug resistant BRAF mutant melanoma. Many natural products with substantial inhibitory activities in cancer cell models display very weak inhibitory activities in vivo as a consequence of their unfavorable pharmacokinetics. We observed that CUMA exhibited much less activity in inhibiting A375-R tumor growth by intraperitoneal injection, but it showed a potent dose-dependent inhibitory effect when administered by oral gavage. It may be worth further elucidating the pharmacokinetic mechanism of CUMA in animals to identify the potential bioactive metabolites derived from CUMA.