Autophagy induces transforming growth factor‐β‐dependent epithelial‐mesenchymal transition in hepatocarcinoma cells through cAMP response element binding signalling

Abstract Autophagy promotes invasion of hepatocarcinoma cells through transforming growth factor (TGF)‐β‐dependent epithelial‐mesenchymal transition (EMT). This study investigated the mechanism by which autophagy induces TGF‐β‐triggered EMT and invasion of hepatocarcinoma cells. Autophagy was induced in HepG2 and BEL7402 cells by starvation in Hank's balanced salt solution. Induction of autophagy degraded phosphodiesterase (PDE) 4A and increased intracellular cAMP, PKA activity and PKA phosphorylation, resulting in increased cAMP response element binding (CREB) phosphorylation in hepatocarcinoma cells. Autophagy‐induced activation of cAMP/PKA/CREB signalling further enhanced TGF‐β1 expression, downregulated the expression of epithelial markers and upregulated the expression of mesenchymal markers, accelerating invasion of hepatocarcinoma cells. Inhibition of autophagy by Atg3 and Atg7 knockdown or by chloroquine treatment prevented degradation of PDE4A and activation of cAMP/PKA/CREB signalling, suppressing TGF‐β1 expression, EMT and invasion in hepatocarcinoma cells. In addition, inhibition of cAMP/PKA/CREB signalling also blocked autophagy‐induced TGF‐β1 expression and prevented EMT and invasion of hepatocarcinoma cells under starvation. Furthermore, exogenous inhibition of PDE4A or activation of cAMP/PKA/CREB signalling rescued TGF‐β1 expression, EMT and invasion in autophagy‐deficient hepatocarcinoma cells. These findings suggest that autophagy induces TGF‐β1 expression and EMT in hepatocarcinoma cells via cAMP/PKA/CREB signalling, which is activated by autophagy‐dependent PDE4A degradation.

Autophagy has been shown to promote invasion of cancer cells during starvation or hypoxia. 5,6 We reported that autophagy accelerated invasion of hepatocarcinoma cells by inducing epithelial-mesenchymal transition (EMT), which changed the cell phenotype from epithelial into mesenchymal under starvation conditions. 7 Transforming growth factor (TGF)-β is a key cytokine that induces EMT in many types of epithelial cells. 8,9 In our previous study, we showed that TGF-β1 expression and activation of TGF-β signalling played crucial roles in autophagy-induced EMT and invasion of hepatocarcinoma cells. 7 However, the mechanism underlying autophagic induction of TGF-β in EMT and invasion of hepatocarcinoma cells is unclear. cAMP response element binding protein (CREB) is a phosphorylation-dependent transcription factor that is phosphorylated by multiple protein kinases and participates in different protein kinase signal transduction pathways. 10 Activation of the CREB signalling pathway and phosphorylation of CREB were shown to fulfill numerous cellular functions ranging from cell proliferation and the cell cycle to cell differentiation and cytokine production by binding of phosphorylated-CREB to the cAMP response element (CRE) in target genes and promoting their transcription. [11][12][13][14] A CRE site has been identified in the TGF-β gene promoter, [15][16][17] and p-CREB induces TGF-β expression in both normal cells and cancer cells by directly binding to the TGF-β F I G U R E 1 Autophagy contributes to degradation of PDE4A in hepatocarcinoma cells. Starvation of HepG2 and BEL7402 cells in Hank's balanced salt solution (HBSS) for 24 h-induced autophagy, which was shown by the LC3-I to LC3-II conversion and P62 degradation. Induction of autophagy significantly reduced PDE4A expression in hepatocarcinoma cells. Treatment of cells in HBSS with chloroquine (5 μmol/L) for 24 h or combined transfection of cells with siRNA-Atg3 and siRNA-Atg7 inhibited autophagy and preserved expression of PDE4A in hepatocarcinoma cells. A, Representative Western blots and (B) densitometric analysis of LC3-I, LC3-II, P62 and PDE4A normalized to β-actin in HepG2 (left panel) and BEL7402 cells (right panel) cultured in complete medium (CM) or in HBSS with different treatments. Cells that were cultured in CM served as a control. C, PDE4A degradation kinetics at the mRNA level in HepG2 (left panel) and BEL7402 cells (right panel) that were cultured in CM and HBSS for 0, 6, 12, and 24 h were determined by quantitative RT-PCR. mRNA levels of PDE4A were normalized to that of β-actin. D, Representative Western blots and (E) densitometric analysis for PDE4A degradation kinetics normalized to β-actin in HepG2 (left panel) and BEL7402 cells (right panel) that were cultured in CM and in HBSS for 0, 6, 12, and 24 h. Cells that were cultured in CM at the initial time (0 h) served as the control. Data are representative of three independent experiments and are shown as the mean ± SEM, n = 3, *P < 0.05 vs control gene promoter, which contributes to tissue fibrosis or tumour progression. 15,18 The cAMP/PKA cascade is a classical signalling transduction pathway that phosphorylates CREB and regulates target gene expression.
Binding of cAMP to regulatory subunits of PKA releases its catalytic subunits and phosphorylates downstream target molecules. This signalling pathway is regulated by adenylate cyclase and phosphodiesterase (PDE), which controls synthesis and hydrolysis of cAMP. 19 Aberrant activation or inhibition of the cAMP/PKA/CREB pathway has been confirmed to result in cellular dysfunction and participates in tumour progression. 20,21 Autophagy is a lysosome-dependent protein and organelle degradation mechanism that maintains the homoeostasis of the cellular metabolic pool. This process also modulates multiple cellular signalling pathways and physiological functions by degrading redundant proteins or enzymes under stress, such as starvation or hypoxia. 22 Thus, this study investigated the role of autophagy in PDE4 degradation and cAMP/PKA/CREB signalling modulation as a mechanism to induce TGF-β-triggered EMT and invasion of hepatocarcinoma cells under starvation.

| Cell culture and treatment
The human hepatocarcinoma cell line HepG2 was purchased from Inhibition of autophagy by Atg3/7 knockdown or by chloroquine (5 μmol/L) treatment reduced intracellular cAMP concentration (A) and PKA activity, (B) as well as phosphorylation of PKA and CREB (C,D), in hepatocarcinoma cell lines in HBSS compared with cells in CM. Treatment of cells with H 89 2HCl (30 μmol/L), a specific PKA inhibitor, significantly decreased intracellular PKA activity and inhibited PKA and CREB phosphorylation in hepatocarcinoma cell lines both in CM and in HBSS. C, Representative Western blots and (D) densitometric analysis for phosphorylated (p)-PKA normalized to total (t)-PKA and phosphorylated (p)-CREB normalized to total (t)-CREB. Cells that were cultured in CM without treatment served as the control. Data are representative of 3 independent experiments and are shown as the mean ± SEM, n = 3, *P < 0.05 vs control supplemented with 10% fetal bovine serum (HyClone, Logan, UT, USA) and 100 μg/mL each of penicillin and streptomycin (Gibco, Invitrogen, Carlsbad, CA, USA) in 5% CO 2 at 37°C. Cells at passages 3-8 were used for the experiments.
HepG2 and BEL7402 cells were, respectively, cultured in the above complete medium and starved in Hank's balanced salt solution (HBSS; HyClone) for 0, 6, 12 and 24 hours to test the PDE4A degradation kinetics in the absence or presence of autophagy. For inhibition of autophagy, cells were transfected with siRNA-Atg3 and siRNA-Atg7 (as described below) or treated with chloroquine

| siRNA synthesis and transfection
Synthesis and transfection of siRNAs was performed as previously described. 7 Briefly, the cDNA sequences of the Atg3 and Atg7 genes were obtained from GenBank (NM_022488 and NM_006395), and the targeting sequences of three different siRNA were designed using an RNAi algorithm available online https://rnaidesigner.thermofisher.

| cAMP assay
Intracellular cAMP concentrations of HepG2 and BEL7402 cells that were treated as described above were measured using a cAMP ELISA kit (Cayman Chemical, Ann Arbor, MI, USA). Briefly, 1 × 10 5 cells were lysed in 0.1 M HCl for 20 minutes and then were scraped off and dissociated until the suspension was homogeneous.
After centrifugation at 1000 g for 10 minutes, the cAMP concentration of each supernatant was measured according to the manufacturer's instruction. Briefly, 50 μL of each supernatant was added to 50 μL of cAMP AChE Tracer and 50 μL of cAMP ELISA antiserum in each well. After incubation at 4°C for 18 hours, the wells were rinsed, and 200 μL per well of Ellman's reagent was added. After incubation in the dark for 2 hours, the absorbance was measured at OD = 420 nm. The cAMP concentration of each sample was calculated according to the standard curve.

| PKA activity measurement
Intracellular PKA kinase activity of HepG2 and BEL7402 cells with the above treatments was measured using a PKA kinase activity

| Quantitative RT-PCR
Real-time PCR was used to detect the mRNA expression levels of PDE4A in HepG2 and BEL 7402 cells cultured in complete medium and in HBSS for 6, 12 and 24 hours, as well as the mRNA expression levels of TGF-β1 in the above cells with different treatments. In brief, total RNA from these cells was isolated by TRIzol ™ Reagent (Invitrogen) according to the manufacturer's protocol. Total RNA was reverse transcribed into first strand cDNA using an iScript cDNA Synthesis kit (Bio-Rad, München, Germany). RNA expression was analysed by RT-PCR using iQ SYBR Green Supermix in an iCycler Real-Time PCR Detection System (Bio-Rad). The following primer sequences were used: PDE4A: sense 5′-AACTTTCCGCAGACGCCTT-3′, antisense 5′-TCTGAGCGGTACAGGAAGGA-3′, TGF-β1: sense 5′-AACTACTGCTTCAGCTCCAC-3′, antisense 5′-AGGACCTTGCTG-TACTGTGT-3′. 23 Expression was normalized to that of β-actin.

| Western blotting
Western blotting was used to detect the protein expression levels of  Optical density of the bands was quantified using NIH (Bethesda, MD, USA) ImageJ.

| Immunofluorescence
Expression levels of TGF-β1 and epithelial-mesenchymal markers in the above HepG2 and BEL 7402 cells with different treatments were detected by immunofluorescence using an avidin-biotin peroxidase complex method. Briefly, cells were fixed in 4% paraformaldehyde and permeabilized using Triton-X-100. Cells were then treated in 3% hydrogen peroxide to inactivate endogenous peroxidase. Non-specific binding was blocked in PBS containing 10% species-appropriate normal serum for 1 hour at room temperature. The above primary antibodies for TGF-β1 (1:100), E-cadherin (1:200), Cytokeratin18

| ELISA
First, 5 × 10 6 HepG2 and BEL7402 cells that were transfected with siRNA-Atg3 and siRNA-Atg7 or with siRNA-control were cultured in HBSS and complete medium for 24 hours. Then, 5 × 10 6 HepG2 and BEL7402 cells were treated with chloroquine (5 μmol/L) or H 89 2HCl (30 μmol/L) in HBSS and complete medium for 24 hours as described above. Each supernatant was then collected to detect the TGF-β1 concentration by enzyme-linked immunosorbent assays (ELI-SAs) using ELISA kits (Abcam) according to the manufacturer's instructions.

| Invasion assay
The invasiveness of the above HepG2 and BEL 7402 cells with different treatments was analysed using Matrigel-coated invasion chambers containing 8-μm pore filters (BD Biosciences, San Jose, CA, USA). Briefly, each cell line with the above treatments was seeded at a density of 1 × 10 5 cells per well in the upper Matrigel-coated chamber of 24-well plates. HBSS was placed in the upper chamber. Complete medium was placed in the lower chamber as a chemotactic agent. Cells that were seeded in the upper chamber with complete medium served as a control. Invasion assay systems were incubated at 37°C with 5% CO 2 for 6 hours.
Cells that invaded the Matrigel to the bottom of filter were stained with 2 μg/mL DAPI in PBS and counted under a fluorescence microscope.

| Statistical analysis
All data are presented as the mean ± SEM. After demonstration of homogeneity of variance with a Bartlett test, one-way ANOVA, followed by Student-Newman-Keuls test where appropriate, was used to evaluate the statistical significance. Values of P < 0.05 were considered statistically significant. Experiments were performed in triplicate.   Figure 1C). Western blotting showed that the protein level of PDE4A in cells was also not changed by complete medium at different culture time points; however, the protein expression was reduced to 83% and 74%; 47% and 37%; and 18% and 6% in HepG2 and BEL 7402 cells after 6, 12 and 24 hours, respectively, of culture in HBSS compared with that of the initial time ( Figure 1D,E). These results confirmed that PDE4A of hepatocarcinoma cells was degraded by autophagy at the post-transcriptional level.

| cAMP/PKA/CREB signalling contributes to autophagy-induced TGF-β1 expression in hepatocarcinoma cells
p-CREB was reported to induce TGF-β1 expression by binding to the CRE site in the TGF-β1 gene promoter. 15 Since activation of cAMP/ PKA/CREB signalling was significantly promoted by autophagy (Figure 2), and autophagy also induced TGF-β1 expression in hepatocarcinoma cells under starvation, 7 we further investigated the role of We found that HepG2 and BEL7402 cells in complete medium did not express TGF-β1 at both the RNA and protein levels, corresponding to the low basal activation of cAMP/PKA/CREB signalling (Figures 2 and 3). Compared with the cells in complete medium, starvation of these cells in HBSS significantly induced TGF-β1 expression and increased the TGF-β1 concentration in HBSS, consistent with the increased activation of cAMP/PKA/CREB signalling (Figures 2 and 3). However, inhibition of autophagy by Atg knockdown or chloroquine treatment, which inhibited activation of cAMP/PKA/ CREB signalling, significantly reduced starvation-induced TGF-β1 expression in hepatocarcinoma cells (Figures 2 and 3). Furthermore, treatment of cells in HBSS with H 89 2HCl, which inhibited PKA/ CREB signalling activation (Figure 2), also notably reduced autophagy-induced TGF-β1 expression at both the RNA and protein levels ( Figure 3). These results indicated that autophagy-induced TGF-β1 expression in a cAMP/PKA/CREB signalling-dependent manner in hepatocarcinoma cells.

| cAMP/PKA/CREB signalling mediates autophagy-induced EMT and invasion of hepatocarcinoma cells
We previously found that autophagy promoted invasion of hepatocarcinoma cells through TGF-β-dependent EMT 7 Figure 4D). Thus, cAMP/PKA/CREB signalling plays a crucial role in autophagy-induced EMT and invasion of hepatocarcinoma cells.

| Autophagy-induced EMT and invasion of hepatocarcinoma cells is dependent on autophagyinduced TGF-β1
Since cAMP/PKA/CREB signalling plays crucial roles both in autophagy-induced TGF-β1 expression and in autophagy-induced EMT, and TGF-β is a key cytokine that induces EMT in many types of epithelial cells, 8,9 we further investigated the role of the autophagyinduced TGF-β1 in EMT and invasion of hepatocarcinoma cells with autophagy during starvation. We treated HepG2 and BEL7402 cells with SB431542 (10 μmol/L), a TGF-β receptor inhibitor, both in complete medium and HBSS to block the effect of TGF-β1. We found that treatment of these cells with SB431542 inhibited the autophagy-induced downregulation of epithelial marker expression and upregulation of mesenchymal marker expression during starvation compared with complete medium (Figure 5A,B). As expected, SB431542 also reduced the invasive number of HepG2 and BEL7402 cells during starvation when EMT was inhibited by inactivity of the TGF-β receptor ( Figure 5C).

| DISCUSSION
Anti-angiogenesis therapy is a common strategy that promotes cancer cell apoptosis or necrosis through nutrient deprivation or hypoxia. 3,4 However, exposure of cancer cells to starvation or hypoxia leads to cellular autophagy, which promotes cell survival by generating intracellular nutrients, growth factors and energy under stress. 24 Autophagy was shown to promote cancer cell invasion during starvation or hypoxia. 5,6 Hence, elucidating how autophagy mediates invasion of cancer cells under stress would help improve the curative effects and reduce the side effects of anti-tumour therapy. We previously reported that autophagy promoted invasion of hepatocarcinoma cells by inducing TGF-β-dependent EMT. 7 In this study, we describe a further mechanism by which autophagy induces TGF-β1 expression, EMT and invasion in hepatocarcinoma cells. We showed that autophagy degraded PDE4A in hepatocarcinoma cells and Numerous studies have clarified the role of various signalling pathways, ranging from AMPK/mTOR to PI3K/Akt and MAPK signalling, in malignant cell behaviours by regulating cellular autophagy. [29][30][31][32][33] However, few studies have elucidated the mechanisms of how autophagy regulates downstream signalling to mediate cellular activities in cancer. Our previous study showed that autophagy promotes invasion of hepatocarcinoma cells by inducing TGF-β1-dependent EMT. 7 Here, we provided further evidence showing that autophagy-induced TGF-β1 expression in hepatocarcinoma cells by activating cAMP/PKA/CREB signalling, consistent with previous reports demonstrating that p-CREB can induce TGF-β gene expression both in normal cells and in cancer cells by directly binding to the CRE site in the TGF-β gene promoter. [15][16][17] This autophagyinduced activation of cAMP/PKA/CREB signalling was dependent on F I G U R E 6 Inhibition of PDE4 or activation of cAMP/PKA/cAMP response element binding protein (CREB) signalling rescues transforming growth factor (TGF)-β1 expression in autophagy-deficient hepatocarcinoma cells. HepG2 and BEL 7402 cells transfected with siRNA-control or siRNA-Atgs (3 and 7) were, respectively, cultured in complete medium (CM) and Hank's balanced salt solution (HBSS) to induce autophagy deficiency, and these two types of autophagy-deficient cells were then treated with a PDE4 inhibitor, roflumilast (20 nmol/L), or a PKA activator, 8-bromo-cAMP (100 μmol/L), for 24 h. Roflumilast or 8-bromo-cAMP treatment increased intracellular cAMP concentration (A) and PKA activity (B), as well as phosphorylation of PKA and CREB (C,D), which further induced TGF-β1 (C-F) expression in these two types of autophagy-deficient hepatocarcinoma cells. C, Representative Western blots and (D) densitometric analysis of p-PKA, p-CREB and TGF-β1, respectively, normalized to t-PKA, t-CREB and β-actin in HepG2 (left panel) and BEL7402 cells (right panel) transfected with siRNA-control or siRNA-Atgs (3 and 7) in CM or in HBSS. E, Quantitative mRNA level of TGF-β1 normalized to β-actin in HepG2 (left panel) and BEL7402 cells (right panel) transfected with siRNA-control or siRNA-Atgs (3 and 7) in CM or in HBSS. F, Representative immunofluorescence staining for TGF-β1 in HepG2 (upper panel) and BEL 7402 cells (lower panel) transfected with siRNA-control or siRNA-Atgs (3 and 7) in CM or in HBSS (scale bar: 10 μm, magnification ×200). Transfected cells without treatment served as a control. Data are representative images or are expressed as the mean ± SEM (n = 3) from three separate experiments. *P < 0.05 vs control degradation of PDE4A based on the fact that inhibition of autophagy preserved PDE4A expression but inactivated cAMP/PKA/CREB signalling; however, inhibition of PDE4A-activated cAMP/PKA/CREB signalling in autophagy-deficient hepatocarcinoma cells, indicating a negative regulatory mechanism of cAMP/PKA/CREB signalling triggered by autophagy in hepatocarcinoma cells under starvation. The current data also revealed that autophagy-activated cAMP/PKA/ CREB signalling induced EMT and invasion in hepatocarcinoma cells, which was consistent with the findings that CREB plays key roles in TGF-β-mediated EMT and invasion of cells. 34,35 We previously reported that autophagy-induced EMT and invasion of hepatocarcinoma cells depended on TGF-β1 expression and activation of its downstream Smad3 signalling. 34,35 Here, we showed that autophagy-activated cAMP/PKA/CREB signalling plays a critical role in inducing TGF-β1 expression in hepatocarcinoma cells, which finally induces EMT and invasion of hepatocarcinoma cells. Consequently, using TGF-β1 as a pivot, we reported the upstream and downstream mechanisms of how autophagy promotes EMT and invasion of hepatocarcinoma cells. Combined with the results of our previous study, the data indicated that autophagy activates cAMP/PKA/CREB signalling in hepatocarcinoma cells by degrading PDE4A, which then induces TGF-β1 expression, and TGF-β1 further activates downstream Smad3 signalling, resulting in EMT and invasion of hepatocarcinoma cells. These studies suggest that autophagy induces growth factor production and mediates cellular activities, such as EMT and invasion of cells, under stress not only by providing amino acids but also by promoting intracellular initiative signalling. In addition to autophagy and TGF-β signalling, PDE4A and the cAMP/PKA/CREB pathway were identified as promising therapeutic targets to prevent invasion of hepatocarcinoma cells under starvation. Therapeutic approaches that interfere with the above mechanism may reduce the adverse side effects of anti-angiogenesis therapy and improve the prognosis of hepatocarcinoma patients.