Sanguinarine inhibits epithelial–mesenchymal transition via targeting HIF-1α/TGF-β feed-forward loop in hepatocellular carcinoma

Epithelial–mesenchymal transition (EMT) plays a crucial role in hepatocellular carcinoma (HCC) progression. Hypoxia and excessive transforming growth factor-β (TGF-β) have been identified as inducers and target for EMT in HCC. Here, we show hypoxia inducible factor-1α (HIF-1α) and TGF-β form a feed-forward loop to induce EMT in HCC cells. Further mechanistic study indicates under both hypoxia and TGF-β stimulation, Smad and PI3K-AKT pathways are activated. We show sanguinarine, a natural benzophenanthridine alkaloid, impairs the proliferation of nine kinds of HCC cell lines and the colony formation of HCC cells. In hypoxic and TGF-β cell models, sanguinarine inhibits HIF-1α signaling and the expression of EMT markers, translocation of Snail and activation of both Smad and PI3K-AKT pathways. Sanguinarine could also inhibit TGF-β-induced cell migration in HCC cells. In vivo studies reveal that the administration of sanguinarine inhibits tumor growth and HIF-1α signaling, inhibits the expression changes of EMT markers as well as Smad and PI3K-AKT pathway proteins. Our findings suggest that sanguinarine is a promising candidate targeting HIF-1α/TGF-β signaling to improve the treatment for HCC patients.


Introduction
Hepatocellular carcinoma (HCC), the primary liver malignancy, is the fourth leading cause of cancer-related death, and the incidence rate of HCC has substantially increased worldwide 1 . Despite advances in the diagnosis and treatment of HCC, the 5-year survival rate remains only 18%, the second lowest among all cancer 2,3 . In HCC progression, epithelial-mesenchymal transition (EMT) plays an important role in early steps of metastasis when cells lose contacts and acquire motility to spread into surrounding tissues or metastasize to extrahepatic sites 4 .
Although EMT is a multistep process with complex mechanisms, tumor microenvironment features, including hypoxia and excessive production of transforming growth factor-β (TGF-β), undoubtedly contribute to the induction of EMT 5 . HCC is one of the most hypoxic tumors with median oxygen level as low as 0.8% 6 . As one of the major mediators of hypoxic response, hypoxia inducible factor-1 (HIF-1) senses intratumoral oxygen tension and subsequently mediates activation of hypoxia responses involved in cancer progression, including proliferation, metabolism, angiogenesis, invasion, metastasis and therapy resistance, thus representing a potential anti-HCC target. Under normoxia, oxygen-sensitive HIF-1α is hydroxylated by prolyl hydroxylase domain-containing proteins (PHDs), thereby ubiquitinated and degraded by an E3 ligase, the von Hippel-Lindau tumor suppressor protein. Under hypoxia, stabilized HIF-1α heterodimerizes with constitutively expressed HIF-1β, also called aryl hydrocarbon receptor nuclear translocator (ARNT), in the nucleus and binds hypoxia-response element in the promoter region of target genes [7][8][9] . HIF-1α drives the expression of many EMT markers and modulators, such as E-cadherin, N-cadherin, and transcriptional activation of Snail 10 .
TGF-β, one of the most potent EMT inducers, has been identified as a therapeutic target in advanced HCC 11,12 . TGF-β activates TGF-β type I receptors which phosphorylate specific receptor-regulated Smad proteins, Smad2 and Smad3, which assemble into heteromeric complexes with Smad4. Heteromeric Smad complexes translocate into the nucleus, where they regulate the transcription of target genes 13 . In addition, Smadindependent signaling pathways, such as PI3K-AKT and MAP kinase pathways, have been implicated in TGFβ-induced EMT non-Smad pathways in TGF-β signaling 14 .
Recently, a few studies have demonstrated the crosstalk between HIF-1α and TGF-β in tumor progression 15,16 . Independent studies have shown that HIF-1α and TGF-β are able to induce EMT, however, the interaction between HIF-1α and TGF-β feed-forward and its contribution to EMT in HCC still await further investigation. We hypothesized the formation of HIF-1α/TGF-β feed-forward loop might contribute to the induction and development of EMT in HCC cells. Numerous attempts have been made to block EMT process by targeting EMT markers 17,18 , EMT-activating transcription factors 19 or extracelluar inducers 20 . However, regarding the clinical aspect, ongoing trails provided no clinical drugs targeting EMT 21 .
Sanguinarine is a benzophenanthridine alkaloid predominantly isolated from the root of Sanguinaria canadensis, Chelidonium majus and other medicinal poppy Fumaria species. The anticancer potential of sanguinarine has been demonstrated in in vivo and in vitro preclinical studies, including apoptosis inducing, antiproliferative, antiangiogenic, and anti-invasive properties in skin, prostate, cervical, breast, hematological, gastrointestinal, pancreatic, and lung malignancies 22,23 . However, its effects on HIF-1α signaling and TGF-β-mediated EMT in HCC are still unknown.
This study aims to investigate the formation of HIF-1α/ TGF-β feed-forward loop that can contribute to the induction and development of EMT in HCC cells. Further, we establish hypoxia and TGF-β-induced EMT models in HCC cells based on the assessment of EMT extent in different cell lines, and evaluate the antiproliferative and EMT reversing effects of sanguinarine in vitro and in vivo. Our study indicates the potential of sanguinarine in HCC treatment and might bring insights to the application of sanguinarine for research and clinical purposes.

Sanguinarine inhibited the proliferation of epithelial and mesenchymal HCC cells
To determine the EMT extent in HCC cell lines, the expression of E-cadherin, N-cadherin, and Vimentin were analyzed by western blotting (Fig. 3a). While HepG2, Hep3B and Huh-7 were considered to be epithelial based on their expression of E-cadherin, the other six kinds of cell lines (SK-Hep-1, Bel-7402, Bel-7404, SMMC-7721, MHCC-97H, and MHCC-97L) were classified as mesenchymal due to low E-cadherin and high N-cadherin expression, although Vimentin expression varies among the tested cell lines. The effect of sanguinarine (Fig. 3b) on the proliferation of epithelial and mesenchymal HCC cells was analyzed by a MTT assay (Figs. S2 and 3c). The following experiments were performed in HepG2 and SMMC-7721 cells and real-time cell analysis (RTCA) results confirmed the IC 50 of sanguinarine in both cell lines (Fig. 3c, d). A colony formation assay was also conducted, showing a significant suppression of Hep2G and SMMC-7721 cells by sanguinarine (Fig. 3e). Moreover, Annexin V/PI staining and cell cycle analysis demonstrated that sanguinarine induced apoptosis and cell cycle arrest at S phase in HepG2 and SMMC-7721 cells (Fig. 3f, g).

Sanguinarine inhibited HCC cell growth in vivo
We tested the ability of sanguinarine to inhibit xenografts from HepG2 and SMMC-7721 cell lines in nude mice models. As shown in Fig. 4a-h, sanguinarine significantly inhibited the tumor growth by decreasing the volume and weight without influencing on the body and organ weight of the mice (Fig. S3a, b). Moreover, Ki-67 staining in both models indicated that sanguinarine reduced the proliferation of HCC cells (Fig. 4i, left). In both tumor models, the Ki-67 staining indexes in sanguinarine-treated groups were significantly lower than those of the tumors in the control groups ( Fig. 4i, right).
To test whether EMT is induced by TGF-β and hypoxia, E-cadherin and upregulated N-cadherin, Vimentin and Snail levels suggested that TGF-β, CoCl 2 or 1% O 2 treatment induced EMT in HCC cells (Fig. 5b, c).

Sanguinarine blocked TGF-β-induced EMT through inhibiting both Smad and PI3K-AKT pathways in vitro
To investigate whether sanguinarine could inhibit TGFβ-induced EMT, nuclear translocation of EMT transcription factor, Snail, as well as the expression of α-SMA were evaluated by immunofluorescence staining. The results indicated that sanguinarine decreased Snail translocation and expression, as well as α-smooth muscle actin (α-SMA) expression induced by TGF-β in HepG2 cells (Figs. 7a, and S4a). In basal conditions, SMMC-7721 and MHCC-97H cells exert certain mesenchymal characteristics which could be downregulated by sanguinarine (Fig. S4b-e). Moreover, sanguinarine suppressed TGFβ-induced N-cadherin, Vimentin, and Snail expression in SMMC-7721 cells (Fig. 7b). Beside the inhibition of EMT markers, we found sanguinarine could block TGFβ-enhanced HIF-1α and CA9 expression without changing the level of ARNT and PHD2, indicating the inhibition of the TGF-β/HIF-1α feed-forward loop by sanguinarine (Fig. 7c).
To gain insights to the underlying mechanism, Smad and PI3K-AKT signaling pathway proteins were examined. The results showed that sanguinarine blocked p-Smad2/3 translocation to the nucleus (Fig. 7a) and the activation of PI3K catalytic subunits, p110α and p110β, regulatory subunit p85, as well as AKT in both HepG2 and SMMC-7721 cell lines (Fig. 7d, e).
EMT allows tumor cells to acquire migration properties. Therefore, we tested whether sanguinarine could inhibit TGF-β-induced cell migration in a scratch assay and transwell migration assay. TGF-β-induced scratch closure was significantly inhibited in the presence of sanguinarine in HepG2 and SMMC-7721 cells (Fig. S5a). Furthermore, the transwell assay demonstrated that sanguinarine prevented both cell lines from migrating to the lower surface of the chamber (Fig. S5b).

Sanguinarine inhibited TGF-β-induced HCC cell EMT in vivo
In HepG2 and SMMC-7721 xenograft models, immunohistochemistry and western blot analysis of the tumor samples showed that in accordance with in vitro results, sanguinarine could inhibit the expression change of HIF-1α (Fig. 8a), EMT markers, including E-cadherin and Snail in HepG2 (Fig. 8b) so as to N-cadherin, Vimentin, and Snail in SMMC-7721 xenograft models (Fig. 8c). Both Smad and PI3K-AKT pathway proteins were also downregulated in sanguinarine-treated groups (Fig. 8d).

Discussion
Proliferation and EMT are two major pathological changes in hepatocellular carcinoma (HCC) development. Inhibiting the proliferation and EMT of HCC cells can be reasonable strategies for HCC treatment. Hypoxia and an excessive amount of TGF-β are two central modulators in tumor EMT. In this study, we found that HIF-1α and TGF-β form a positive feed-forward loop to induce HCC cell EMT. Sanguinarine could inhibit HCC cell proliferation, HIF-1α/TGF-β feed-forward loop, EMT and migration in vitro and in vivo.
Accumulating evidence has suggested the crosstalk between hypoxia and TGF-β signaling in cancer cells. On one hand, in line with our observations that hypoxia increases TGF-β expression and signaling in HCC cells, hypoxia triggers extracellular matrix-related protein expression in a TGF-β-and PI3K-dependent manner in . ×200 magnification Ki-67 staining index was assessed as the percentage of total cells that were Ki-67 positive from six randomly selected high-power fields in xenografts from three mice of each group (right). Statistic analysis were performed by one-way ANOVA followed by Bonferroni posttest, and data are represented as mean + SEM, n = 6. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. control group.
showed higher HIF1A and TGFB1 gene expression in HCC tissues and its correlation with the overall survival rate, tumor cell proliferation and EMT. Thus, targeting the HIF-1α/TGF-β feed-forward loop could be a feasible therapeutic strategy for HCC patients.
Sanguinarine has potent antiproliferative effects in nine kinds of tested HCC cell lines. Sanguinarine also inhibited the colony formation of HepG2 and SMMC-7721 cells. Furthermore, reduced xenograft tumor volume and weight and Ki-67 expression confirmed that sanguinarine could inhibit the proliferation of HepG2 and SMMC-7721 cells in vivo.
The role of EMT in tumor progression has been intensively studied in the past decade. E-cadherin downregulation and N-cadherin upregulation lead to an increase in cellular motility. Loss of E-cadherin or increase of N-cadherin expression has been implicated in cancer progression and metastasis 27,28 . As a major cytoskeletal component of mesenchymal cells, Vimentin is another marker often used for mesenchymal-derived cells or cells undergoing EMT during metastatic progression 29 . Besides, targeting EMT has been considered as a potential way to overcome drug resistance in cancer treatment 30 . Characterizing hepatoma cell lines as epithelial or mesenchymal by E/N-cadherin and Vimentin expression facilities us better understanding of the cell EMT extent. Studies have shown a strong correlation between EMT extent and the sensitivity of anticancer drugs 31,32 . We chose HepG2, SMMC-7721 and MHCC-97H cells to conduct the experiments and observed that TGF-β-induced changes of EMT markers in these cell types, including decrease of E-cadherin expression, increase of N-cadherin, Vimentin and α-SMA. Moreover, EMT-activating transcription factors, including Snail, Twist and Zeb families, executing EMT and playing important roles in cancer progression from tumor initiation, growth, invasion, dissemination, and metastasis to colonization as well as in drug resistance 33 . Nuclear translocation of EMT transcription factor Snail and elevated Snail expression were found in TGF-β-treated cells.
These findings suggested that TGF-β could induce EMT in epithelial and also mesenchymal cells. In accordance with the knowledge that cells achieved migration ability through EMT, we observed that TGF-β enhanced cell migration in both cell lines.
The prognosis of metastasized HCC is particularly poor due to lack of treatment choices. Higher serum levels of TGF-β in HCC patients were found to be associated with less sensitivity to sorafinib, the first regimen listed in the treatment for advanced HCC 34 . TGF-β signaling should be explored as a therapeutic target to improve treatments for advanced HCC. Indeed, blocking TGF-β has been found to elevate E-cadherin levels and reduce the migration and invasion of HCC cells 35 . Moreover, TGF-β receptor inhibitors are the only compounds interfering with EMT in clinical trial and have shown the ability to inhibit HCC cell migration and invasion 21,36 . This study demonstrated that sanguinarine as a natural compound exerted EMT reversing effects in hypoxia-or TGFβ-treated HCC cells. In vivo studies also confirmed that sanguinarine treatment inhibited the local EMT process which may be the result from the control of both HIF-1α and TGF-β signaling. Although the secretion of TGF-β and VEGF were unaffected by sanguinarine in tumor tissue samples, the involvement of multiple cells in a local area may contribute to the average production. In TGFβ-induced EMT and cell migration models, we demonstrated that sanguinarine could decrease the expression of EMT markers, Snail translocation and cell migration. In addition, sanguinarine also suppressed protein expression of N-cadherin and Vimentin and inhibited SMMC-7721 cell migration without TGF-β stimulation.
Hypoxia and TGF-β activated both Smad and PI3K-AKT pathways in HepG2 and SMMC-7721 cells, indicating these pathways may be the underlying mechanism of induced EMT in HCC cells. To elucidate sanguinarine's mode of action, we found that EMT reversing effects of sanguinarine were resulted from inhibition of these pathways. TGF-β is one of the main drivers of liver fibrosis which is one of the main precancerous conditions for HCC. Inhibition of TGF-(see figure on previous page) Fig. 7 Sanguinarine inhibited TGF-β-induced EMT in vitro. a HepG2 cells were treated with different concentrations of sanguinarine in the absence or presence of 10 ng/mL TGF-β for 24 h. Snail (green), α-SMA (green), p-Smad2/3 (green), DAPI (blue) staining and 2-channel merged images indicated the nuclear translocation of Snail and p-Smad2/3, and expression of α-SMA. The results shown were representative of three independent experiments. Scale bars, 200 μm. b SMMC-7721 cells were treated with different concentrations of sanguinarine in the absence or presence of 10 ng/mL TGF-β. Protein levels of N-cadherin, Vimentin, and Snail were analyzed by western blotting. Quantification plots are shown below. c MHCC-97H cells were treated with different concentrations of sanguinarine in the absence or presence of 10 ng/mL TGF-β. Protein levels of N-cadherin, Snail, HIF-1α, ARNT, PHD2, and CA9 were analyzed by western blotting. GAPDH or β-actin served as controls. Quantification plots are shown below. Western blot analysis of p100α, p110β, p-p85, and p-AKT expression were demonstrated in d HepG2 and e SMMC-7721 cells treated with indicated concentrations of sanguinarine in the absence or presence of 10 ng/mL TGF-β for 48 h. Quantification plots are shown below. Data were expressed as mean ± SEM (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 compared with untreated control cells. The results shown were representative of three independent experiments. # P < 0.05, ## P < 0.01, ### P < 0.001, #### P < 0.0001 one-way ANOVA followed by Bonferroni posttest in comparison with TGF-β-treated samples.
β signaling by sanguinarine may also have potential antifibrosis effects in liver cells preventing the induction of HCC, while TGF-β seems to have various roles depending on tumor type and stage, e.g., TGF-β inhibits cell proliferation in early stages of many malignancies, but promotes cancer progression or late-stage tumors 37 . Further studies regarding the effects of sanguinarine on liver fibrosis and different stages of HCC are still needed.
In summary, the HIF-1α/TGF-β feed-forward loop could be a target for HCC treatment. Sanguinarine could target HIF-1α and TGF-β signaling pathways and inhibit subsequent EMT and cell migration. Given the properties found in this study, sanguinarine can be a candidate as an adjuvant in cancer therapy for HCC patients.

Cell proliferation assay
HepG2, Hep3B, Huh-7, SK-Hep-1, Bel-7402, Bel-7404, SMMC-7721, MHCC-97H, and MHCC-97L cells were seeded into 96-well plates (1 × 10 4 cells/well) and cultured with different concentrations of sanguinarine for 48 h. After treatment, cells were cultured with serum-free medium and MTT (0.5 mg/mL) for 4 h. After the removal of medium,150 μL DMSO was added to each well and the plates were shaken for 15 min. The plates were then analyzed using a microplate reader (Bio-Rad, Hercules, CA, USA) using a wavelength of 490 nm. Three independent experiments, each with triplicate cultures, were performed to determine each data point.

Real-time cell analysis
Real-time cell analysis (RTCA) was conducted using the xCELLigence RTCA S16 (ACEA, San Diego, USA). Briefly, 50 μL of the medium was added to 16-well E-Plates to obtain background readings before adding 100 μL of the cell suspension (4 × 10 4 cells/well). After incubation at room temperature for 30 min, the E-Plates were placed on the reader in an incubator, and the cell index recording continued. After 24 h, the cells were treated with sanguinarine at different concentrations. Cells were monitored every 15 min for 48 h after treatment.

Colony-forming assay
Cells were seeded in a 12-well plate (200 cells/well) and treated with different concentrations of sanguinarine for 48 h. The medium was changed to drug-free medium for additional 10-15 days. Then, the colonies were fixed with methanol and stained with 0.2% crystal violet for 15 min at room temperature. Images were taken using the white light imaging system (Champchemi Professional SG2010084, Sage Creation, Beijing, China) and the inverted fluorescence microscope (DM505, Nikon Co., Ltd., Otawara, Tochigi, Japan).

Flow cytometry for cell cycle and apoptosis analysis
Cells were plated into 6-well plates. After 24 h incubation, cells were treated with the indicated concentrations of sanguinarine for 48 h. Then cells were harvested and subjected to sequential staining with 5 μL AnnexinV-FITC and 10 μL PI (20 μg/mL) was added and incubated for 10 min.
Cells were cultured in 6-well plates and treated with the indicated concentrations of sanguinarine for 48 h after serum starvation for 24 h. Then the cells were harvested, and fixed in 70% ice-cold ethanol at −20°C overnight. After fixation, the cells were incubated with 1 mL RNase (50 μg/mL) for 30 min and stained with 1 mL PI (60 μg/ mL). All stained cells were analyzed by FACS (Becton Dickinson, Mountain View, CA, USA). The obtained data were analyzed with Modfit LT software.

Scratch assay
Cells were seeded into a 12-well plate and allowed to grow to 80% confluence in complete medium. Cell monolayers were scratched by a 200 μL pipette tips and plates were then washed three times with PBS and incubated in medium with different stimulation for 48 h. Cell migration into the scratched surface was photographed under an inverted fluorescence microscope (×200 magnification).

Transwell migration assay
The in vitro migration assay was performed using transwell chambers. Cells were incubated with treatment for 48 h. The medium in the chamber was replaced with serum-free medium and 30% FBS contained medium was added to the 24-well plate as the chemoattractant. After incubation for 24 h, all the non-migrated cells which were located on the upper surface of the chamber were carefully removed with a cotton swab. The migrated cells through the chamber to the lower surface were fixed with 4% paraformaldehyde and then stained with crystal violet for 15 min. The number of migrated cells was photographed under an inverted fluorescence microscope (×200 magnification).

Western blotting
After treatment, cells were lysed in RIPA buffer containing phosphatase and protease inhibitors. Total protein content was quantified with the BCA assay. Secreted protein in the cell supernatant was concentrated by methanol-chloroform precipitation. The protein sample was subjected to SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Millipore, Bedford, MA, USA), which was blocked with 5% non-fat milk in TBST buffer and incubated with the indicated primary antibodies at 4°C overnight. Then the membrane was incubated with species-specific horseradish peroxidase (HRP)conjugated secondary antibodies and antigen-antibody complexes were visualized using the enhanced ECL kit. The images were taken using Tanon5200 imaging system (Tanon, Shanghai, China). β-actin and GAPDH were used as control for normalizing protein loading.

Immunofluorescence
Cells were fixed with 4% paraformaldehyde for 10 min and then blocked with 10% BSA for 30 min at room temperature. After this procedure, the cells were incubated with primary antibodies (1:200) at 37°C for 4 h. Then, cells were incubated with an CoraLite488conjugated anti-rabbit and CoraLite594-conjugated antimouse secondary antibody (1:50) at room temperature for 1 h. The nucleus was stained with DAPI. Fluorescent signals were detected using an inverted fluorescence microscope (DM505, Nikon Co., Ltd., Otawara, Tochigi, Japan).

ELISA
The cell culture supernatant was assayed for TGF-β and VEGF (Sino Biological, Beijing, China) by using commercially available ELISA kits.

Animals and xenograft models
All the experiments were in accordance with the guidelines of the Institutional Animal Care and Use Committee (approval number: 2019-1031). For the induction of xenograft model, 200 μL cell suspension (2 × 10 7 cells/mL) was subcutaneously implanted into the right axillary of the nude mice (male immune-deficient BABL/c nude mice). When the average of the tumor volume reached about 100 mm 3 , mice were randomly divided into control group and treatment groups (n = 6 per group). Mice in the control group were orally administered with 0.5% sodium carboxymethyl cellulose (CMC-Na) solution or different concentrations (1.25, 2.5, 5 mg/kg) of sanguinarine. Animals in every group received ten times of drug administration for consecutive 3 days and every other day during which the body weight and tumor volume were monitored and recorded. On the last day, after the mice were sacrificed, tumors and main organs were removed and weighed. The tumor volume was calculated with the following formula: Tumor volume = A × B × B/2, where A was the longest and B was the shortest dimension of the tumor. Each tumor was cut in half, one was immersed in 0.1% diethyl pyrocarbonate solution and stored at −80°C for western blot analysis, and the other half was fixed in 4% paraformaldehyde for immunohistochemistry study.

Immunohistochemistry assay
Tumor specimens were embedded in paraffin and cut

Statistical analysis
One-way analysis of variance (ANOVA) and further Bonferroni's multiple comparison test was used and P value < 0.05 was considered statistically significant (Prism 6.0, GraphPad, La Jolla, CA, USA). Data are presented as mean + SEM.