Chidamide Suppresses the Growth of Cholangiocarcinoma by Inhibiting HDAC3 and Promoting FOXO1 Acetylation

Inhibitors for histone deacetylases (HDACs) have been identified as epigenetic drug targets to treat a variety of malignancies through several molecular mechanisms. The present study is aimed at investigating the mechanism underlying the possible antitumor effect of the HDAC inhibitor chidamide (CDM) on cholangiocarcinoma (CCA). Microarray-based gene expression profiling was conducted to predict the expression of HDACs in CCA, which was validated in clinical tissue samples from CCA patients. Next, the proliferation, migration, invasion, autophagy, and apoptosis of human CCA QBC939 and SNU308 cells were measured following treatment with CDM at different concentrations. The acetylation level of FOXO1 in the nucleus and cytoplasm of QBC939 and SNU308 cells was determined after overexpression and suppression of HDAC3. A QBC939-implanted xenograft nude mouse model was established for further exploration of CDM roles in vitro. HDAC3 was prominently expressed in CCA tissues and indicated a poor prognosis for patients with CCA. CDM significantly inhibited cell proliferation, migration, and invasion of QBC939 and SNU308 cells, while inducing their autophagy and apoptosis by reducing the expression of HDAC3. CDM promoted FOXO1 acetylation by inhibiting HDAC3, thereby inducing cell autophagy. Additionally, CDM inhibited tumor growth in vivo via HDAC3 downregulation and FOXO1 acetylation induction. Overall, this study reveals that CDM can exhibit antitumor effects against CCA by promoting HDAC3-mediated FOXO1 acetylation, thus identifying a new therapeutic avenue for the treatment of CCA.


Introduction
Cholangiocarcinoma (CCA) is a common biliary malignancy with increased incidence rate in association with suspected risk factors such as obesity and hepatitis C virus infection [1,2]. Due to the present absence of sensitive early biomarkers, diagnosis of CCA is often delayed to an advanced stage of this disease, such that five-year survival is less than 10% [3]. The available treatments include targeted, combined chemotherapy and immunotherapy and personalized therapies, but there is not curative treatment for advanced disease.
Chidamide (CDM), a novel histone deacetylase (HDAC) inhibitor, has been reported to play an antitumor role in T cell tumors through multiple mechanisms [4]. In multiple myeloma cells, CDM induces growth arrest and apoptosis in a caspase-dependent manner [5], but CDM is scantly researched in the context of CCA. Histone deacetylases (HDACs) are a family of enzymes capable of catalyzing the removal of acetyl groups from the acetyl-lysine residues in histone and nonhistone proteins [6]. HDACs exert a vital function in numerous cancers through their regulation of the expression and function of proteins involved in cancer initiation and progression [7]. Selective inhibitors for HDACs have been identified as treatments for epigenetic targets to treat a variety of malignancies through several molecular mechanisms [8,9]. HDAC inhibitors have also demonstrated potent and specific anticancer stem cell activities; CDM can specifically promote apoptosis of leukemia stem cell-like cells and primary acute myeloid leukemia CD34(+) cells in a concentration-and time-dependent manner [10]. In addition, HDAC2 is an essential factor regulating cancer stem cell phenotype as its silencing can promote the stemness of MG63 and Saos2 cells [11]. HDAC inhibitors can inhibit the migration and invasion of CCA cell lines [12], and other research shows that the HDAC inhibitor CG200745 exerts antitumor effects in CCA cell lines via miR-NAs targeting the Hippo pathway [13]. HDAC inhibitors can act on the cellular stress response pathway, reduce angiogenesis through downregulation of angiogenic genes such as VEGF, HIF-1, and eNOS to inhibit the formation of new blood vessels and, in combination with classic chemotherapy drugs, interfere with CCA cell migration [14]. Published research has reported that the HDAC inhibitor CDM synergizes with Rituximab to inhibit the growth of diffuse lager Bcell lymphoma tumors by regulating CD20 [15]. As one of the potential targets of CCA, FOXO1 bears a close relationship to the sensitivity of tumors to drugs [16]. In CCA cells, the interaction between FOXO1 acetylation and Atg7 regulates autophagic flux, promotes cell apoptosis, and exerts an antitumor effect [17]. Metformin enhances autophagy via the AMPK/SIRT1-FOXO1 pathway in diabetic kidney disease, which suggests the involvement of FOXO1 in autophagy [18]. By promoting SIRT1-FOXO1-ATG14 axisdependent autophagy, paeonol prevents palmitic acidinduced dysfunction of lipid metabolism in HepG2 injury [19]. However, there are hitherto no reports of CDM in CCA. In this study, wet set out to explore whether CDM can play an antitumor effect in CCA cells and to test the mechanism associated with HDAC3-mediated deacetylation of FOXO1.

Materials and Methods
2.1. Ethics Statement. The current study was approved by the Ethics Committee of the First People's Hospital of Yongkang, Affiliated to Hangzhou Medical College, and performed in strict accordance with the Declaration of Helsinki. All participants signed informed consent documentation before sample collection. The animal experiments were performed with the approval from the experimental animal institution.
2.2. Cell Screening, Culture, and Transfection. Human CCA cell lines QBC939 and SNU308 purchased from Shunran Biological Technology Co., Ltd. (Shanghai, China) were cultured in Dulbecco's modified Eagle's medium (DMEM) (Gibco Invitrogen Co., USA) with 10% fetal bovine serum (Gibco), 10 μg/mL streptomycin, and 100 μg/mL penicillin in a 5% CO 2 incubator at 37°C. Cells in the logarithmic growth phase were collected after trypsin digestion and seeded in 6-well plates at a density of 1 × 10 5  2.4. Scratch Test. Logarithmic growth phase BC939 and SNU308 cells were evenly inoculated into 2 wells (70 μL/ well) at a density of 4 × 10 5 cells/well in the scratch insert placed in a small Petri dish. A small portion of serum was added to the Petri dish and incubated in the incubator overnight. The next day, when the cells had adhered stably, the scratch plug was carefully removed. The cells were washed gently twice with PBS, 900 μL of serum-free medium was added, and then the cells were photographed immediately (0 hour). Subsequently, 100 μL CDM at concentrations of 25, 50, and 100 μM was added to experimental groups, and 100 μL medium was added to the control dishes. After culture in the incubator for 48 hours, the cells were photographed. The distance between the two holes of the scratch insert was 0.5 mm. The scratch difference value of two photographs was the migration distance of CCA cells. The experiment was repeated three times independently. Ltd., Beijing, China) was added to the sections for incubation at 37°C for 20 minutes. DAB (ST033, Guangzhou Weijia Technology Co., Ltd., Guangzhou, China) was used to develop the sections, which were counterstained with hematoxylin (PT001, Shanghai Bogu Biotechnology Co., Ltd., China, Shanghai) for 1 minute. The sections were blued in 1% ammonia water, dehydrated, and cleared and mounted before observation under a microscope in 5 randomly selected high-power fields from each section, with 100 cells counted in each field. The experiment was repeated three times.

Colony Formation
2.11. Immunoprecipitation (IP). Cells were lysed in lysis buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 10% glycerol, 1 mM EDTA, 0.5% NP-40, and protease inhibitor mixture, and the cell debris was eliminated by centrifugation. The cell lysate was then incubated with 1 μg anti-FLAG antibody (ab125243, Abcam) and 15 μL protein A/G beads (Santa Cruz Biotechnology) for 2 hours. After extensive washing, the beads were placed in a boiling water bath for 5 minutes. The proteins were separated by SDS-PAGE, transferred to nitrocellulose membrane (Millipore, Temecula, CA, USA), and then subjected to Western blotting to evaluate the acetylation level of FOXO1. (c) Western blotting of HDAC3 protein in cancer tissues (n = 26) and normal bile duct tissues (n = 26). * p < 0:05, compared to the normal bile duct tissues. Data were shown as the mean ± standard deviation. An unpaired t -test was employed for data comparison between two groups, while one-way ANOVA with Tukey's post hoc test was used for multigroup data comparison.

HDAC3 Is Highly Expressed in CCA Tissues and Is
Associated with Poor Prognosis of Patients. We first screened the expression of HDAC I (HDAC1, 2, 3, and 8), II (HDAC4, 5, 6, 7, 9, and 10), and IV (HDAC11) in CCA through TCGA database. The results showed higher expression of HDAC3, HDAC7, HDAC10, and HDAC11 in CCA samples compared with the normal samples. Among these, the elevated was associated with lower survival rate of patients with CCA (Figure 1(a)). Moreover, the results of immunohistochemistry displayed that the positive expression of HDAC3 was significantly increased in CCA tissues compared with normal bile duct tissues (Figure 1(b)). Western blotting data further exhibited higher expression of HDAC3 in cancer tissues than in normal bile duct tissues (Figure 1(c)). The above results indicate that HDAC3 is abundantly expressed in CCA tissues and is related to poor prognosis of patients.   mice. * p < 0:05 and * * p < 0:01, compared with control mice. Data were shown as the mean ± standard deviation. One-way ANOVA with Tukey's post hoc test was used for multigroup data comparison and repeated measures ANOVA with Tukey's post hoc test was applied to compare data at different time points. n = 6 for mice in each group. 10 Stem Cells International with CDM-H and oe-HDAC3 led to higher cell colony formation, migration, and invasion than seen with CDM-H treatment alone. Besides, CDM treatment promoted the apoptosis rate of QBC939 and SNU308 cells, which displayed a dose-dependent response. However, the apoptosis rate of QBC939 and SNU308 cells was augmented following combined treatment with CDM-H and oe-HDAC3 relative to CDM-H treatment alone (Figure 2(e)). The results of Western blotting suggested that the expression of autophagy-related proteins LC-3, ULK1, and Beclin-1 was decreased in cancer tissues (Figure 2(f)), while CDM treatment led to a dose-dependent increase in their expression. However, the expression of LC-3, ULK1 and Beclin-1 was reduced in QBC939 and SNU308 cells treated with CDM-H+oe-HDAC3 CDM-H+chloroquine (CQ; the autophagy inhibitor) (Figure 2(g)). Taken together, these data confirm that CDM can repress the proliferation, migration, and invasion of CCA cells while promoting their apoptosis and autophagy by inhibiting HDAC3.

CDM Induces Cell Autophagy by Increasing FOXO1
Acetylation Level. Previous studies have reported that FOXO1 acetylation is involved in cell autophagy [18,21]. Therefore, we speculated that CDM-induced autophagy in CCA cells is related to FOXO1 acetylation. To test this prediction, we first used Western blotting to detect the FOXO1

11
Stem Cells International acetylation level in cancer tissues. The results revealed a lower acetylation level of FOXO1 in cancer tissues than in normal bile duct tissues (Figure 3(a)), demonstrating that CDM-induced autophagy was indeed correlated to FOXO1 acetylation.
Subsequently, we starved cells to induce autophagy. Compared with the control cells, the acetylation level of FOXO1 was higher in the EBSS (starvation treatment) cells (Figure 3(b)). The results of IP also revealed an increase in the acetylation level of FOXO1 in EBSS-and CDM-Htreated cells relative to the control cells (Figure 3(c)). In addition, IP results presented higher FOXO1 acetylation level in the cytoplasm and nucleus of EBSS cells than in control cells (Figure 3(d)). These data demonstrate that CDM can induce cell autophagy by increasing FOXO1 acetylation level.
3.4. CDM Promotes FOXO1 Acetylation by Inhibiting HDAC3. It has been reported that HDAC3 activates the transcription of FOXO1 [22]. We proceeded in this study to investigate whether CDM induced FOXO1 acetylation by inhibiting HDAC3. First, QBC939 and SNU308 cells were treated with sh-HDAC3. The results of Western blotting revealed that, compared with the sh-NC treatment, HDAC3 protein expression was decreased, while the FOXO1 acetylation level was enhanced in the sh-HDAC3-treated cells (Figure 4(a)). Furthermore, after nuclear and cytosolic separation of the cells, the FOXO1 acetylation level in the nucleus and cytoplasm was detected by IP. Here the results showed that the FOXO1 acetylation level was also increased in the nucleus and cytoplasm of cells treated with sh-HDAC3 (Figure 4(b)), thus consistently showing that inhibiting HDAC3 could accelerate the acetylation of FOXO1.
Subsequently, QBC939 and SNU308 cells were transfected with plasmids overexpressing HDAC3 and then treated with CDM. As illustrated in Figure 4(c), the protein expression of HDAC3 was reduced while FOXO1 acetylation level was decreased in cells treated with CDM-H. The effects of CDM-H on HDAC3 protein expression and FOXO1 acetylation level were negated by further overexpression of HDAC3. After nuclear and cytosolic separation of the cells, Ac-lysine was collected to detect the FOXO1 acetylation level. IP results shown in Figure 4(d) indicated that the FOXO1 acetylation level was greater in the nucleus and cytoplasm of CDM-H-treated cells than in the oe-NCtreated cells, while opposite results were noted in the presence of CDM-H+oe-HDAC3. These results serve as new evidence that inhibition of HDAC3 may be responsible for the promotion of FOXO1 acetylation by CDM.  Table 1, CDM inhibited the growth of CCA in a dose-dependent manner. In addition, further analysis of the distribution of CDM in tumors and major organs showed that CDM was mainly enriched in the liver of the nude mice ( Figure 5(c)), indicating that CDM was mainly metabolized by the liver. Besides, H&E staining results illustrated that tumor cells in the tumor tissues of control mice were tightly arranged, with large nuclei, less cytoplasm, and obvious cell atypia. Following treatment with CDM, the cells were relatively reduced in number and replaced by fibrous tissues. More fibrous tissues were present with increasing CDM dose ( Figure 5(d)). At the same time, the results of immunohistochemical staining and Western blotting suggested a decline in the HDAC3 expression in the tumor tissues of CDM-treated mice, while FOXO1 acetylation level was increased in a dose-dependent manner (Figures 5(e) and 5(f)). The above results indicate that CDM can inhibit the growth of CCA in vivo.
Finally, we sought to investigate whether the anti-tumor effect of CDM in vivo was correlated with the regulation of the HDAC3/FOXO1 axis. As shown in Figure 6(a), CDM-H treatment delayed the growth of tumors while further HDAC3 overexpression accelerated the tumor growth. Besides, CDM was still mainly distributed in the liver of mice treated with CDM and oe-HDAC3 (Figure 6(b)). As shown in Figure 6(c), tumor cells in the tumor tissues of oe-NC-treated mice were tightly arranged, with large nuclei, less cytoplasm, and obvious cell atypia. Following treatment with CDM-H, the cells were largely reduced and replaced by fibrous tissues. Conversely, treatment with CDM-H+oe-HDAC3 increased the number of tumor cells and reduced fibrous tissues formation. In addition, the results of Western blotting suggested a decline in HDAC3 expression in the tumor tissues of CDM-H-treated mice, while FOXO1 acetylation level was increased. Further overexpression of HDAC3 led to opposite results ( Figure 6(d)). Overall, these lines of evidence demonstrate that CDM could inhibit HDAC3 expression and increase FOXO1 acetylation, thus preventing the growth of CCA in vivo.

Discussion
CCA is a deadly disease, such that surgery and adjunct treatments are curative in only a few cases [23]. The search to improve the efficacy of CCA treatments calls upon obtaining a better understanding of its molecular pathogenesis of CCA to support the development of rational therapies. In this paper, we studied the antitumor effects and underlying mechanism of CDM in CCA cells. Our experimental results showed that CDM elicited anti-tumor effects by inhibiting HDAC3-mediated deacetylation of FOXO1.
According to previous reports, CDM inhibits the proliferation of glioma cells, lung cancer, pancreatic cancer, and myeloma cells [24][25][26]. In Jurkat and HUT-78 cells, treatment with CDM (2 μM) leads to downregulation of HDAC3 expression, thus inducing necroptosis [27]. Our present research obtained similar results in CCA cells, while our cell migration and colony formation experiments suggested that 12 Stem Cells International CDM inhibited the proliferation of CCA cells in a dosedependent manner, which also promoted cell autophagy. Previous research has indicated that miR-373 inhibits autophagy of CCA cells by targeting ULK1 and promoting apoptosis [28]. Dihydroartemisinin, which was found to have antitumor activity in a variety of human cancers, induces cell apoptosis and autophagy-dependent cell death of CCA cells via the DAPK1-BECLIN1 pathway [29]. Using autophagy as an adaptive mechanism, cancer cells can survive under conditions of extreme stress in the tumor microenvironment and can also promote the invasiveness and resistance of anti-cancer drugs. In the initial stage of CCA development, cell autophagy can be impaired, and the associated accumulation of LC3-II and p62 may reflect defects in the late processing of autophagosomes, rather than an increase in autophagy rate. In preclinical studies, autophagy modulators can promote CCA cell death, reduce invasion, and render CCA cells sensitive to chemotherapy. Inhibition of autophagy may promote oncogenic transformation of bile duct cells, and impaired autophagy caused by inactivation of Beclin1 may promote the malignant phenotype of CCA [30]. Inhibition of autophagy has thus been considered to be a new strategy to prevent the growth of cancer cells. Results of the mechanistic investigation in the current study suggest that CDM exerts an antitumor effect by inhibiting HDAC3 to promote cell autophagy, based on findings of reduced expression of the autophagy-related proteins LC-3, Ulk1, and Beclin1. CDM has been reported to be an HDAC inhibitor [31], which in combination with rituximab can inhibit the growth of diffuse large B-cell lymphoma tumors [15]. Besides, inhibition of HDAC3 expression can decrease the proliferation of myeloma and CCA cells [32,33]. Our present results suggest that HDAC3 has a clear correlation with clinical CCA and is involved in the associated regulation of cell autophagy. Furthermore, the current results in CCA cells showed that CDM treatment increased FOXO1 acetylation level, which led to the inhibition of HDAC3 activity and anti-CCA effects. As a potential therapeutic target, FOXO1 can regulate the autophagy flux of human CCA cells [17]. Moreover, FOXO1 acetylation is closely associated with cell autophagy [18,21]. It is furthermore reported that FOXO1 exerts its anti-tumor action by regulating the autophagy of tumor cells [17]. Thus, FOXO1 deacetylation promotes cell autophagy [34], but inhibiting the autophagy level of cancer cells can enhance their chemotherapeutic sensitivity [35][36][37]. To conclude, our present findings elaborate that CDM ameliorates CCA by inhibiting the expression of HDAC3 and further promoting FOXO1 acetylation, which activates the autophagy of CCA cells.

Conclusions
This study demonstrates that CDM has a significant tumor suppressing effect on CCA and may eventually provide a new treatment approach for CCA. CDM suppresses HDAC3 expression and induces acetylation of FOXO1, thus impeding the proliferation, migration, and invasion of CCA cells, while promoting cell apoptosis and autophagy, ultimately arresting the growth of CCA (Figure 7). We propose that CDM may act synergistically with other chemotherapy drugs in the treatment of CCA. As for FOXO1, further investigations may reveal how FOXO1 could improve the drug sensitivity of CCA tumors and discourage their otherwise inexorable progression.

Data Availability
The datasets generated/analyzed during the current study are available.

Conflicts of Interest
The authors declare no conflict of interest.