Inhibition of pancreatic cancer stem cell characteristics by α‐Mangostin: Molecular mechanisms involving Sonic hedgehog and Nanog

Abstract The current investigation was intended to elucidate the molecular mechanism of α‐Mangostin in the regulation of pancreatic cancer stem cell (CSC) characteristics. Here, we demonstrate that α‐Mangostin inhibited cell proliferation in pancreatic CSCs and cancer cell lines while it showed no effect on human pancreatic normal ductal epithelial cells. Also, α‐Mangostin inhibited colony formation and induced apoptosis in these cells. Further, α‐Mangostin inhibited the self‐renewal capacity of CSCs isolated from human primary tumours and KrasG12D mice. Furthermore, α‐Mangostin inhibited the invasive and metastatic ability of pancreatic CSCs by suppressing the epithelial‐to‐mesenchymal transition (EMT) via up‐regulation of E‐cadherin and down‐regulation of mesenchymal phenotype by inhibiting N‐cadherin, Snail and Slug expression. Interestingly, the pluripotency maintaining factors and CSC markers were inhibited by α‐Mangostin thus suggesting that α‐Mangostin can target CSCs to inhibit pancreatic cancer effectively. Gli signalling plays a crucial role in the self‐renewal and pluripotency of CSCs. α‐Mangostin inhibited the Gli transcription and the expression of Gli target genes (Nanog, Oct4, c‐Myc, Sox‐2 and KLF4) in CSCs. Using ChIP assay, we demonstrated that Nanog could directly bind to promoters of Cdk2, Cdk6, FGF4, c‐Myc and α‐Mangostin inhibited Nanog binding to these promoters. Conversely, the inhibitory effects of the α‐Mangostin on CSC proliferation and Gli or Nanog transcription and their targets were abrogated by either enforced activation of sonic hedgehog (Shh) or by the overexpression of Nanog. Taken together, our studies suggest that α‐Mangostin may act as Gli inhibitor and establishes the pre‐clinical significance of α‐Mangostin for the prevention and treatment of pancreatic cancer.


| INTRODUC TI ON
Pancreatic cancer is a devastating disease and is the fourth most common cause of cancer-related mortality in the United States. 1 Pancreatic cancer exhibits the poorest prognosis from all other cancers, and the overall 5-year survival rate continues to be less than 6%. 2,3 Pancreatic cancer is characterized by slow growth, late detection and resistance to chemotherapy and radiation and is associated with high mortality rates even after surgery. 3 Unfortunately, by the time the disease is diagnosed most of the pancreatic cancer patients present with unresectable advanced malignancy due to early metastasis. Accumulating evidence supports the role of cancer stem cells (CSCs) in cancer initiation, progression, metastasis and chemotherapy failure. 4 Despite increased advancement in our understanding of the disease progression, diagnosis and therapeutics, available treatment options are limited. Chemo-and radio-therapies have been largely ineffective and associated with enhanced drug toxicity, drug resistance and frequent redevelopment of the metastatic disease.
Further, poor bioavailability of the drug and undesirable side effects are other significant limitations in the effective management of pancreatic cancer. Thus, it is highly desirable to develop an increased understanding of the pathogenesis of the disease, for effective disease management and the development of effective strategies by non-toxic natural agents for the prevention and treatment of pancreatic cancer.
Hedgehog (Hh) signalling pathway is crucially involved in vertebrate development. 5 It is inactive in mature cells of normal adult and found to be aberrantly hyper-activated in pancreatic cancer and various other malignancies. Evidence suggests that Hh signalling can regulate tissue homeostasis by controlling the production of stem or progenitor cells. 6 Deregulation of the Hh signalling pathway is associated with various malignancies. In several of our recent reports, we have demonstrated that for the prevention of pancreatic cancer, a variety of natural products and small molecules displayed antiproliferative properties through targeting the sonic hedgehog (Shh) signalling pathway. [7][8][9][10][11][12] Upon binding of the Shh ligand to transmembrane Patched (Ptch) receptor results in the withdrawal of inhibitory effects of Patched on smoothened. 5 Thus, the pathway is activated via smoothened through Hh protein stimulation or by the loss of patched activity through Ptch mutations. Activation of the Hh pathway via smoothened induces Gli transcriptional activity.
Several reports have demonstrated that Hh pathway activation induces stem cell markers and is involved in the enhancement of epithelial-to-mesenchymal transition (EMT) thus regulating metastasis in various malignancies including pancreatic ductal adenocarcinoma. 5,13 The involvement of Shh increases dramatically from PanIN lesions to PDAC to metastatic tumours. 14 Therefore, targeting the Shh pathway is regarded as a beneficial strategy for the prevention and treatment of pancreatic cancer.
Wnt, Shh and several other extrinsic developmental pathways play significant roles in the maintenance and regulation of pluripotency and self-renewal capacity of progenitor and stem cells. 15 Nanog is a transcription factor which is one of the crucial downstream effectors of these signalling pathways and also a direct transcriptional target of Gli. 16,17 Nanog modulates pluripotency, maintain self-renewal and block differentiation. 18 Nanog is expressed highly in germline stem cells, tumours, carcinomas and seminomas. 19,20 In addition to Gli, various other transcription factors such as Oct4, FoxD3 and P53 can regulate Nanog transcription. 21,22 Interestingly, Nanog can cooperate with Oct4 and Sox2 in maintaining self-renewal capacity and pluripotency in stem cells. 23 As Nanog is a direct target of the Shh pathway, we will examine the regulation of Shh-Nanog pathway by α-Mangostin in pancreatic CSCs.
As natural product-based compounds are non-toxic, can target multiple pathways and can negatively impact the self-renewal capacity of CSCs, they can be developed as an attractive strategy for the prevention and treatment of pancreatic cancer. Mangosteen plant (Garcinia mangostana) grows abundantly in Southeast Asia mainly in the Sunda Islands and the Moluccas of Indonesia and other parts of tropical South America. [24][25][26] Mangosteen is rich in xanthonoid and phytochemicals. α-Mangostin is a xanthonoid derived from Mangosteen which is well tolerated and safe. It possesses antioxidant, anticancer and anti-inflammatory properties that are highly relevant to humans. 24,25,[27][28][29][30] However, the underlying molecular mechanisms of α-Mangostin in the inhibition of pancreatic cancer by targeting CSCs and Shh-Nanog pathway are not well understood.
Therefore, α-Mangostin holds great promise and can be developed as an anticancer agent.
The objective of this study was to delineate the molecular mechanism of α-Mangostin in the regulation of pancreatic CSC characteristics. Our data demonstrate that α-Mangostin inhibits cell proliferation in pancreatic CSCs and cancer cell lines. α-Mangostin is non-toxic to human pancreatic normal ductal epithelial cells. Also, α-Mangostin inhibits the colony formation and induces apoptosis selectively in pancreatic CSCs and cancer cell lines, compared to the pancreatic normal ductal epithelial cells. Further, the self-renewal capacity of CSCs isolated from human primary tumours and Kras G12D mice was inhibited by α-Mangostin.

| Cell culture
Pancreatic cancer cell lines AsPC-1 and PANC-1 were used, and these cells were purchased from American Type Culture Collection (Manassas, Virginia) purchase. Cells were grown and frozen in liquid nitrogen for future use. Cells from second and third passages were used for the experiments. ATCC utilizes short tandem repeat (STR) profiling to authenticate the cell lines. PANC-1 possesses mutations in p53 and K-ras genes in codon 273 and codon 12 respectively. AsPC-1 harbours mutation on codon 12 of K-ras gene. CD133+/CD44+/ CD24+/ESA+human pancreatic CSCs were isolated from primary tumours as described previously. 26 Pancreatic CSCs isolation and characterization from KrasG12D mice were performed as described elsewhere. 26 Pancreatic CSCs were grown in specialized growth medium (Celprogen, Inc, Torrance, California) which contained 1% N2 Supplement (Invitrogen), 2% B27 Supplement (Invitrogen), 20 ng/mL human platelet growth factor (Sigma-Aldrich), 100 ng/mL epidermal growth factor (Invitrogen) and 1% antibiotic-antimycotic (Invitrogen).
Pancreatic CSCs and cancer cell lines were cultured at 37°C in a humidified atmosphere of 95% air and 5% CO 2 .

| Cell proliferation and apoptosis assays
Pancreatic cancer cells and CSCs (1.5 × 10 4 ) in 1 mL of culture medium were incubated for 24 to 48 hours time-points with various concentrations of 0-10 µmol/L of α-Mangostin. Cell viability and apoptosis were measured by trypan blue assay using Countess™ Automated Cell Counter (Invitrogen) and TUNEL assay respectively. 7

| Colony formation assay
For colony formation assay, pancreatic cancer cells and CSCs were seeded at a low density into 6-well plates and then treated with or without α-Mangostin for up to 2 weeks. Cell culture medium containing either α-Mangostin or DMSO was renewed every 3 days.
After fixation of the colonies with cold methanol, 0.5% crystal violet was used to stain them. The colonies were imaged with a microscope, and the number of colonies was counted.

| Spheroid assay
We performed spheroid formation assays as described elsewhere. 7 Briefly, cells at 100-500 cells/mL density were plated in ultra-low attachment plates. The spheroid formation in suspension was evaluated under a microscope after 10 days of culture.

| Motility, transwell migration and invasion assays
Assays for cell motility, transwell migration and invasion have been performed as we described elsewhere. 7,8,26

| Gene expression by quantitative RT-PCR analysis
Total RNA in cells was isolated using TRIzol reagent (Invitrogen).
About 2 μg of the extracted RNA was reverse transcribed into cDNA using High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). qRT-PCR was carried out using fast SYBR Green Master Mix (Thermo Fisher Scientific). Relative mRNA expressions were compared with controls and evaluated using 2 −ΔΔCt methods. protocol. For reuse, the blots were washed in stripping buffer for 30-60 minutes at room temperature.

| Chromatin immunoprecipitation (ChIP)
Pancreatic CSCs were treated with various doses of α-Mangostin, crosslinked and sonicated. The crosslinked sheared chromatin samples were then incubated with 3 μg of Nanog antibody and 5 μL of protein-A and protein-G magnetic beads. The ChIP DNA was further purified, and 1-3 µL elutions were used to measure enrichment using quantitative real-time PCR. ChIP-derived DNA was electrophoresed on 2% agarose gels.

| Gli and Nanog reporter assay
Measurement of Gli and Nanog reporter activities was performed as we described elsewhere. 7,31 In brief, pancreatic cancer cells were stably transduced with lentiviral particles expressing cop-GFP and luciferase genes (pGreen Fire1-4xGli or Nanog-mCMV-EF1-Neo).
The cells for transcription assay were seeded in 96-well plates and treated with or without α-Mangostin (0-10 µmol/L) for various time-points. According to the manufacturer's instructions (Promega Corp., Madison, WI), luciferase reporter activity was measured at the end of incubation period.

| Statistical analysis
The results presented are representative of three independent experiments run in triplicate, unless otherwise indicated. Student t test or ANOVA was used to analyse the differences between groups.
Differences among groups were considered significant at P < 0.05.

| α-Mangostin targets the pancreatic CSCs selfrenewal capacity, stemness and inhibits spheroid and colony formation
A significant characteristic of CSC is its ability to form spheroids in suspension which is reflective of its capability of stemness, reconstitution and propagation of tumours. 8,32 Thus, to examine F I G U R E 2 α-Mangostin inhibits cell viability in spheroids, and colony formation and the expression of stem cell markers and pluripotency maintain factors by CSCs from pancreatic tumours. (A and B), CSCs were isolated from pancreatic tumours of humans and KrasG12D mice and treated with α-Mangostin (0-10 µmol/L) for 7 d to obtain primary spheroids. At the end of the incubation period for a week, the spheroids were collected, reseeded and treated with α-Mangostin for another week to obtain secondary spheroids. Further, secondary spheroids were collected, reseeded and treated with α-Mangostin for another week to obtain tertiary spheroids. Cell viability in the spheroids was measured by trypan blue assay at the end of 7, 14 and 21 d. Data represent mean ± SD. *, &, #, %, ˄ and ** = significantly different from control, P < 0.05. C, In soft agar, human pancreatic CSCs were seeded and treated with α-Mangostin (0-10 μmol/L) for 21 d. At the end of the incubation period, the number of colonies was counted. *, #, % and @ = significantly different from control (n = 4), P < 0.05. D, Human pancreatic CSCs were treated with α-Mangostin (0-10 μmol/L) for 48 h. The expression of CD24, CD44 and CD133 was measured by the Western blot analysis. E, human pancreatic CSCs were treated with α-Mangostin (0-10 μmol/L) for 48 h, and the expression of Nanog, Oct4, Sox2, KLF4 and c-Myc was measured by the Western blot analysis. β-actin was used as a loading control the effect of α-Mangostin on the growth of CSCs isolated from primary pancreatic tumours from human and KrasG12D mice, we measured spheroid formation abilities in the presence and absence of α-Mangostin. As shown in Figure 2A,B, in CSCs isolated from pancreatic tumours of human and KrasG12D mice, the cell viability of primary, secondary and tertiary spheroids was inhibited by α-Mangostin. In the α-Mangostin-treated groups, smaller and fewer spheroids were formed than that in the control group (data not shown). Further, α-Mangostin impaired the colony formation ability of human pancreatic CSCs in a dose-dependent manner ( Figure 2C). We, therefore, examined the underlying mechanism for these inhibitory effects of α-Mangostin on the regenerative and survival capacity of human pancreatic CSCs.

| Inhibitory effects of α-Mangostin on Shh signalling pathway and Gli transcriptional targets
We have demonstrated that Shh signalling pathway is highly activated in pancreatic cancer. 7,11,33 We examined the expression of Shh pathway components to analyse the effects of α-Mangostin in pancreatic CSCs. Gli1, Gli2, Patched1, Patched2 and smoothened protein expression was inhibited by α-Mangostin as measured using Western blot analysis ( Figure 3A). As Gli can regulate its own expression as well as the expression of Patched as they are both its direct transcriptional targets. To examine the effects of α-Mangostin, we next measured the Gli transcriptional activity using luciferase reporter assay. As demonstrated in (Figure 3B F I G U R E 3 α-Mangostin inhibits the components of the sonic hedgehog (Shh) pathway, Gli transcription and markers of cell proliferation and cell cycle. A, Pancreatic CSCs were treated with α-Mangostin (0-10 µmol/L) for 48 h. The expression of Gli1, Gl2, Patched-1, Patched-2 and smoothened was measured by the Western blot analysis. β-actin was used as a loading control. B, Gli-responsive GFP/firefly luciferase viral particles were used to transduce pancreatic CSCs (pGreen Fire1-Gli with EF1, System Biosciences). After transduction, the culture medium was replaced, and CSCs were treated with α-Mangostin (0-10 µmol/L) for 24 h. Gli reporter activity was measured as we described elsewhere. 16 *, #, % and @ = significantly different from control, and each other, P < 0.05. C, α-Mangostin inhibits the expression of Bcl-2 and cyclin D1. Pancreatic CSCs were treated with α-Mangostin (0-10 µmol/L) for 48 h, and the expression of Bcl-2 and cyclin D1 was measured by the Western blot analysis. β-actin was used as a loading control

| α-Mangostin inhibits binding of Nanog to its target genes (Cdk2, Cdk6, FGF4, c-Myc and Oct4) and Nanog transcription
In the maintenance of self-renewal and pluripotency, Nanog is considered to play a critical role. We have demonstrated increased levels of Nanog expression in pancreatic CSCs and cell lines. As Nanog is a transcription factor, the effects of α-Mangostin on Nanog binding to the promoters of its target genes were examined. We performed chromatin immunoassays for investigating the binding of Nanog to promoters of Cdk2, Cdk6, FGF4, c-Myc and Oct4 in the presence and absence of α-Mangostin. As shown by ChIP-PCR assay in Figure 4A, Nanog can bind to Cdk2, Cdk6, FGF4, c-Myc and Oct-4 target gene promoters.
However, the binding of Nanog to these promoters was significantly inhibited by α-Mangostin. We confirmed these ChIP-PCR data with qRT-PCR where α-Mangostin inhibited the binding of Nanog to Cdk2, Cdk6, FGF4, c-Myc and Oct4 genes ( Figure 4B-F).
As Nanog is also a direct transcriptional target of Gli. To investigate the effects of α-Mangostin on Nanog transcription, we measured the luciferase reporter activity. In pancreatic CSCs, AsPC-1 and PANC-1 cell lines, α-Mangostin significantly inhibited the Nanog reporter activity ( Figure 5). Thus, taken together, these results suggest that α-Mangostin regulates pluripotency-, cell survival-and cell cycle-related genes by modulating Cdk2, Cdk6, FGF4, c-Myc and Oct4 expression through Nanog.

| Inhibitory effects of α-Mangostin on cell motility, migration, invasion and markers of epithelialmesenchymal transition
For metastasis to occur, EMT becomes inevitable in which cancer cells acquire genetic changes that equip them to migrate to distant organ sites where they can reestablish and proliferate. 34,35 As CSCs are associated with the metastasis and treatment resistance, F I G U R E 4 α-Mangostin inhibits binding of Nanog to its target genes (Cdk2, Cdk6, FGF4, c-Myc and Oct4). A, Pancreatic CSCs were treated with α-Mangostin (0-10 µmol/L) for 24 h. Cells were harvested, and chromatin immunoprecipitation assays were performed with the anti-Nanog antibody as described in Materials and Methods. PCR was performed to examine the binding of Nanog to Cdk2, Cdk6, FGF4, c-Myc and Oct4 promoters. Lane 1 = input, Lane 2 = immunoprecipitation (IP) with an anti-IgG antibody, Lanes 3-5 = IP with the anti-Nanog antibody of cell lysates from CSCs treated with 0, 5 or 10 µmol/L α-Mangostin respectively. (B-F), Nuclear extracts were prepared, and chromatin immunoprecipitation assays were performed as described above. qRT-PCR was used to examine the binding of Nanog to Cdk2, Cdk6, FGF4, c-Myc and Oct4 promoters. Data represent mean (n = 4) ± SD. *, and # = significantly different from control, and each other, P < 0.05 we further examined the effects of α-Mangostin on acquiring metastatic characteristic namely cell motility, migration, invasion and expression of EMT markers. Figure 6A,B demonstrate that α-Mangostin inhibits cell motility, migration and invasion of pancreatic CSCs. Further as shown in Figure 6C

| D ISCUSS I ON
In the current study, we demonstrate that α-Mangostin inhibits pancreatic CSC characteristics through Shh-Nanog pathway. Pancreatic CSCs and cancer cell lines (AsPC-1 and PANC-1) were transduced with Nanog-responsive GFP/firefly luciferase viral particles (pGreen Fire1-Nanog with EF1, System Biosciences). After transduction, the culture medium was replaced, and cells were treated with α-Mangostin (0-10 µmol/L) for 24 h. Nanog reporter activity was measured as we described elsewhere 16  Patched and HHIP1. We have shown that several natural products can regulate pancreatic CSC characteristics and inhibit tumour growth by suppressing the Shh-Gli pathway. [8][9][10][11]32,[36][37][38][39] In the pres- F I G U R E 7 Enforced activation of the Shh pathway or Nanog overexpression abrogated inhibitory effects of α-Mangostin on cell proliferation and Gli or Nanog transcription and expression respectively. A, Pancreatic CSCs were pretreated with Shh protein (100 nmol/L) for 2 h followed by treatment with α-Mangostin (5 µmol/L) for 48 hrs. Cell proliferation was measured by trypan blue assay. Data represent mean (n = 4) ± SD. *, or % = significantly different from control, P < 0.05. B, Using Gli-responsive GFP/firefly luciferase viral particles (pGreen Fire1-Gli with EF1, System Biosciences), pancreatic CSCs were transduced. After transduction, CSCs were pretreated with Shh protein for 2 h followed by treatment with α-Mangostin for 24 h. Gli transcription was measured by luciferase assay as we described elsewhere. 15 Data represent mean (n = 4) ± SD. *, or % = significantly different from control, P < 0.05. C, Pancreatic CSCs were pretreated with Shh protein (100 nmol/L) for 2 h followed by treatment with α-Mangostin (5 µmol/L) for 48 h. The expression of Gli1 was measured by q-RT-PCR. Data represent mean (n = 4) ± SD. *, or % = significantly different from control, P < 0.05. D, Pancreatic CSCs were transduced with lentiviral particles expressing either empty vector or Nanog cDNA. The expression of Nanog was measured by q-RT-PCR. Data represent mean (n = 4) ± SD. * = significantly different from control, P < 0.05. E, Pancreatic CSCs were transduced with lentiviral particles containing either empty vector or Nanog cDNA and treated with Mang (5 µmol/L) for 48 h. Cell proliferation was measured by trypan blue assay. F, Pancreatic CSCs (CSCs/Empty vector and CSCs/Nanog cDNA) were transduced with Nanog-responsive GFP/firefly luciferase viral particles (pGreen Fire1-Nanog with EF1). After transduction, CSCs were treated with Mang for 24 h. Nanog transcription was measured by luciferase assay as we described elsewhere. 15 Data represent mean (n = 4) ± SD. *, or % = significantly different from control, P < 0.05. G, CSCs/Empty Vector and CSCs/Nanog cDNA cells were treated with Mang (5 μmol/L) for 48 h, and c-Myc expression was measured by q-RT-PCR. Mang = α-Mangostin Inhibition of Shh pathway at the level of Gli transcription will be more effective in treating patients because mutations in smoothened receptors (upstream of Gli) are commonly observed during treatment. Our studies define the new mechanism(s) and novel molecular targets of pancreatic cancer treatment and prevention. α-Mangostin can regulate progression and metastasis through the inhibition of Shh signalling pathway and factors controlling pluripotency and EMT. In the present study, we have used α-Mangostin doses up to 10 µmol/L, which are comparable to other chemodietary agents in the clinic. This dose is low enough to exert clinical impact and may not need any further modification. In another study, we have demonstrated the pre-clinical potential of α-Mangostin-encapsulated PLGA nanoparticles for the management of pancreatic cancer. 40 Thus, α-Mangostin offers excellent potential as a novel preventive and therapeutic agent for pancreatic cancer by targeting CSCs.
Both EMT and CSCs have been increasingly demonstrated to be associated with the development of invasive characteristic and distant metastasis. 41,42 CSCs have also been shown to express genes associated with EMT along with stemness genes. HH promotes EMT by upregulating the expression of Snail, Slug, ZEB1, ZEB2, TWIST2 and FOXC2. Snail inhibits cadherin gene expression and triggers EMT. Snail has been shown to bind to the E-cadherin promoter at the E-box site and thus can repress the transcription of E-cadherin transcription. 43 Further, it has been reported that Snail also stimulates the transcription of mesenchymal genes such as N-cadherin and Vimentin and thus regulate their expressions. 44 In conclusion, we demonstrate here that α-Mangostin inhibits pancreatic CSC characteristics and cancer cell growth through the inhibition of Shh-Gli pathway. α-Mangostin inhibits Gli transcription, and its target genes Nanog, Oct4, c-Myc, Sox2 and KLF4.
α-Mangostin not only inhibited smoothened but also Gli transcription. Therefore, α-Mangostin has an additional clinical advantage, that is, if the pancreatic cancer patients develop resistance to smoothened inhibition due to mutation(s) on the smoothened receptors, they will be able to respond to α-Mangostin because it also inhibits the expression of effector molecule Gli. Therefore, targeting the Shh-Gli pathway through dual inhibition (smoothened and Gli) by α-Mangostin could have enormous clinical significance for pancreatic cancer initiation, progression, metastasis and tumour growth. Overall, our study suggests that α-Mangostin can be used for the treatment and prevention of pancreatic cancer by targeting CSCs.

ACK N OWLED G EM ENTS
The authors acknowledge all the lab members for critical reading, assistance and suggestions during manuscript preparation.

CO N FLI C T O F I NTE R E S T
YM, W.Y and AS have declared that no competing interests exist. SS and RKS have declared intellectual property interest.

AUTH O R CO NTR I B UTI O N S
WY and YM performed the experiments, analysed the data and wrote the manuscript; SS, AS and RKS designed the study and contributed reagents. All authors read and approved the manuscript.