Antrodia cinnamomea induces autophagic cell death via the CHOP/TRB3/Akt/mTOR pathway in colorectal cancer cells

Antrodia cinnamomea, a well-known traditional medicine used in Taiwan, is a potent anticancer drug for colorectal cancer, but the upstream molecular mechanism of its anticancer effects remains unclear. In this study, A. cinnamomea extracts showed cytotoxicity in HCT116, HT29, SW480, Caco-2 and, Colo205 colorectal cancer cells. Whole-genome expression profiling of A. cinnamomea extracts in HCT116 cells was performed. A. cinnamomea extracts upregulated the expression of the endoplasmic reticulum stress marker CHOP and its downstream gene TRB3. Moreover, dephosphorylation of Akt and mTOR as well as autophagic cell death were observed. Gene expression and autophagic cell death were reversed by the knockdown of CHOP and TRB3. Autophagy inhibition but not apoptosis inhibition reversed A. cinnamomea-induced cell death. Finally, we demonstrated that A. cinnamomea extracts significantly suppressed HCT116 tumour growth in nude mice. Our findings suggest that autophagic cell death via the CHOP/TRB3/Akt/mTOR pathway may represent a new mechanism of anti-colorectal cancer action by A. cinnamomea. A. cinnamomea is a new CHOP activator and potential drug that can be used in colorectal cancer treatment.


Results
A. cinnamomea extract isolation. The A. cinnamomea fruiting bodies used in this study are shown in
ACF2 upregulates CHOP and TRB3 expression. To investigate genetic expression changes between control and ACF2-treated colorectal cancer cells, we analysed the transcriptomic profiles by NGS. Briefly, HCT116 cells were treated with 50 μg/ml ACF2 for 24 h, and RNA was then extracted from control and ACF2-treated cells. Subsequently, whole-genome sequencing was performed by NGS. A posterior probability fold change >2 and p < 0.05 were required for differences to be considered statistically significant. A total of 502 genes exhibited differential expression in ACF2-treated cells compared with that in control cells. Compared with control cells, 201 genes were upregulated in ACF2-treated cells, and 301 genes were downregulated (Supplementary Dataset 1). The 15 top differentially expressed genes are listed in Table 1. ACF2 significantly upregulated the expression of the endoplasmic reticulum stress (ER stress) marker C/EBP homologous protein (CHOP, also known as DNA  genes of CHOP that are involved in apoptosis and autophagy [26][27][28][29][30][31] . After ACF2 treatment, NGS analysis showed no statistically significant difference in these genes except for tribbles pseudokinase 3 (TRIB3, also known as TRB3), which was increased by 6.48-fold after ACF2 treatment (Table 2). To confirm the NGS analysis results, quantitative reverse transcription PCR was used to observe CHOP and TRB3 gene expression. The mRNA expression levels of CHOP and TRB3 were significantly increased by ACF2 treatment, supporting the results of the NGS analysis (Fig. 3). Moreover, ACF2 increased CHOP and TRB3 mRNA expression levels in a dose-dependent manner. To investigate whether increased CHOP and TRB3 expression was associated with protein expression, Western blots were further employed. HCT116 cells were treated with multiple concentrations of ACF2 for 24 h. Western blot analysis revealed that the protein expression of CHOP was increased after treatment with ACF2 at concentrations of 50, 75, 100, and 125 μg/mL, and TRB3 was also increased by treatment with ACF2 at these concentrations. These data suggested that ACF2 markedly increases the RNA and protein expression of CHOP and TRB3 (Tables 1 and 2, Figs 3, 4A).
ACF2 induces autophagic cell death via the CHOP/TRB3/Akt/mTOR pathway. To investigate the anticancer mechanism of ACF2, we analysed molecules downstream of CHOP and TRB3 at the protein level. TRB3 is involved downstream of the CHOP-mediated apoptosis pathway 31 . TRB3 promotes apoptosis via Akt dephosphorylation, which is followed by PUMA (BBC3) induction 32,33 . However, as shown in Table 2, there was no statistically significant difference in BBC3 between control and ACF2-treated cells. We thus converted our attention to the Akt-mammalian target of rapamycin (mTOR) pathway. TRB3 reportedly negatively regulates the Akt-mTOR pathway by directly binding to Akt 34 , and negative regulation of the Akt-mTOR axis activates Ulk1. After the release of ULK1 from inactive mTOR, autophagosome formation is induced 29 . HCT116 cells were treated with increasing concentrations of ACF2 for 24 h, and phospho-Akt, total Akt, phospho-mTOR, and total mTOR were detected by Western blot analysis. As shown in Fig. 4B, at concentrations of 75, 100, and 125 μg/mL, ACF2 obviously decreased the phosphorylation levels of Akt and mTOR. The amounts of total Akt and mTOR proteins remained relatively constant regardless of whether the HCT116 cells were exposed to ACF2. Because previous studies have shown that mTOR is an important regulator of the autophagic process, autophagy was then evaluated. Microtubule-associated protein light chain 3 (LC3) is a soluble protein used to monitor autophagic activity. During the autophagic process, the soluble form of LC3 (LC3-I) is conjugated to phosphatidylethanolamine. The LC3-phosphatidylethanolamine conjugate (LC3-II) is tightly bound to autophagosomal membranes, and the LC3 conversion from LC3I to LC3II is considered one of the important hallmarks of autophagy 35 . We evaluated the autophagic process via detecting LC3 conversion by Western blot analysis. As shown in Fig. 4B, at concentrations of 75, 100, and 125 μg/mL, ACF2 obviously increased the protein expression levels of LC3-II. The accumulation of LC3-II indicated that ACF2 induced autophagy in HCT116 cells. Thus, ACF2 presumably induces autophagy via the CHOP/TRB3/Akt/mTOR pathway (Fig. 4A,B).
To verify our hypothesis that the anticancer mechanism of ACF2 in colorectal cancer is due to CHOP/ TRB3/Akt/mTOR-mediated autophagy, CHOP and TRB3 small interference RNA (siRNA) were employed to suppress CHOP and TRB3 expression in HCT116 cells. HCT116 cells were transfected with siRNA targeting human CHOP or TRB3 or control siRNA. After 48 h of stimulation, HCT116 cells were treated with 75 μg/ mL ACF2 for 24 h. Compared with non-transfected and control siRNA-transfected cells, CHOP siRNA clearly decreased ACF2-induced CHOP protein overexpression (Fig. 4C). TRB3 siRNA decreased ACF2-induced TRB3 protein overexpression (Fig. 4D). Phospho-Akt, total Akt, phospho-mTOR, and total mTOR were also detected by Western blot analysis. As shown in Fig. 4C, the CHOP protein reduction was accompanied by efficient reduction of ACF2-induced TRB3 protein expression in CHOP siRNA-transfected cells. Moreover, CHOP and TRB3 siRNA significantly reversed the ACF2-decreased phosphorylation levels of Akt and mTOR. In CHOP or TRB3 siRNA-transfected cells, ACF2 treatment did not suppress Akt or mTOR phosphorylation. Next, we evaluated LC3-II protein expression in CHOP or TRB3 siRNA-transfected cells, revealing that both CHOP and TRB3 knockdown decreased ACF2-induced LC3-II expression (Fig. 4C,D). Since autophagy has been implicated in cell survival and cell death 29 , we next investigated the role of autophagy in the anticancer mechanism of ACF2. The MTS assay was used to evaluate the viability of HCT116 cells. As shown in Fig. 4E, transfection with control siRNA or TRB3 or CHOP siRNA had no effect on cell viability. Compared with non-transfected and control siRNA-transfected cells treated with ACF2, in the presence of CHOP or TRB3 siRNA, ACF2-induced cell death was significantly diminished. Next, 3-methyladenine (3-MA) and chloroquine (CQ), autophagy inhibitors, were employed to inhibit autophagy. LC3-II protein expression and the cell viability of HCT116 cells were evaluated. As shown in Fig. 4F, ACF2-induced LC3-II protein increase was efficiently reduced by 3-MA but increased by CQ. In the presence of 3-MA or CQ, ACF2-induced cell death was   significantly decreased (Fig. 4G). To exclude apoptotic cell death in this study, cleaved caspase3 protein expression and the cell viability of HCT116 cells were assessed. 5-Fluouracil (5-FU) was used as a positive control to induce apoptosis in HCT116 cells 36 . 5-FU obviously increased the protein expression levels of cleaved caspase3. Cleaved caspase3 was not observed after ACF2 treatment (Fig. 4H). Z-VAD-fmk, a pan-caspase inhibitor, was employed to inhibit apoptosis. As shown in Fig. 4I, in the presence of Z-VAD-fmk, 5-FU-induced apoptotic cell death was significantly diminished. However, Z-VAD-fmk did not rescue ACF2-induced cell death. ACF2 treatment can lead to autophagy-dependent cell death. Taken together, these results suggest that ACF2 induces autophagic cell death via the CHOP/TRB3/Akt/mTOR pathway.

ACF2 inhibits the growth of colorectal cancer cells in vivo.
To investigate whether ACF2 has an anticancer effect on colorectal cancer in vivo, athymic mice xenografted with a human tumour model were utilized. HCT116 cells were inoculated into the nude mice at 3 × 10 6 cells per mouse subcutaneously and treated with ACF2 at doses of 400 mg/kg, 200 mg/kg, and 100 mg/kg (daily gavage) for 15 days. As shown in Fig. 5A, the tumour volume in the control group (Ctrl) increased gradually from days 0 to 15 after administration. From day 6 onwards, the tumour volume in the 400 mg/kg ACF2 treatment group increased slowly, and the differences between the control group and the 400 mg/kg ACF2 treatment group became significant (p < 0.005). On the sixth day after treatment, the tumour volume in the 200 mg/kg ACF2 treatment group was significantly decreased compared with that in the control group (p < 0.05). However, after 9, 12 and 15 days of administration, the tumour volume in the 200 mg/kg ACF2 treatment group increased, and there was no statistically significant difference compared with the control group. Changes in tumour volume in the 100 mg/kg ACF2 treatment group were similar to those in the control group, with the two groups presenting no significant difference at any time point. As shown in Fig. 5B,C, at the end of the 15-day treatment period, the average weight of the tumours in the 400 mg/ kg ACF2 treatment group was significantly smaller than that in the control group (p < 0.005). The body weight of each mouse was observed during the experiments. Importantly, there were no significant decreases in body weight in the ACF2 treatment groups at any of the dosages (Fig. 5D). These results indicate that ACF2 can inhibit the growth of HCT116 colorectal cancer cells in vivo.

Discussion
In this study, we utilized whole-genome expression profiling to discover the molecular mechanisms underlying the anticancer effects of A. cinnamomea on colorectal cancer. In previous studies, A. cinnamomea was determined to exert anticancer effects on colorectal cancer by the induction of apoptotic cell death and cell cycle arrest 10,17,19,20 . However, the upstream molecular mechanisms of these anticancer effects remained unknown. We used NGS to uncover novel mechanisms of A. cinnamomea, revealing that the A. cinnamomea treatment of HCT116 colorectal cancer cells resulted in the upregulation of CHOP, TRB3, and some members of the CXC chemokine family. CHOP, known as growth arrest and DNA damage 153 (GADD153), was originally identified in response to UV-induced DNA damage 37 . However, further studies showed that ER stress plays a critical role in CHOP induction. During ER stress, CHOP was induced by all three ER stress sensors, namely, PERK, IRE1, and ATF6 [28][29][30][31]38 . The numerous inducers of CHOP, such as DNA-alkylating agent methylmethane sulphate, tunicamycin, and thapsigargin, activated CHOP expression via the induction of ER stress [38][39][40] . A. cinnamomea reportedly induces ER stress-mediated apoptosis and autophagy in lung and liver cancer cells 41,42 . In our study, A. cinnamomea treatment induced CHOP-mediated autophagic cell death in colorectal cancer cells.
After CHOP induction, CHOP promotes the transcriptional activation of many pro-apoptotic factors, including TNFRSF10B, ERO1A, TRB3, and members of the BCL2 family. CHOP-mediated apoptosis is involved in many diseases, such as neurodegenerative diseases, metabolic diseases, atherosclerosis, and cancer 31 . CHOP has been shown to trigger apoptotic cell death in lung, colorectal, prostate, and ovarian cancer 43,44 . On the other hand, CHOP also induces the transcriptional upregulation of many autophagic genes, such as SQSTM1, NBR1, and members of the ATG family 27 . In our study, the expression of CHOP and its downstream target TRB3 were upregulated by A. cinnamomea. TRB3 can promote apoptosis via Akt dephosphorylation, followed by BBC3 induction 32,33 . TRB3 can also promote autophagy via Akt and mTOR dephosphorylation 29 . Akt and mTOR dephosphorylation were observed after A. cinnamomea treatment, but the mRNA expression of BBC3 was not increased. Conversion of the autophagy marker LC3 was increased by A. cinnamomea treatment ( Table 2 and Fig. 4B). Furthermore, knockdown of CHOP and TRB3 prevented the effects of A. cinnamomea on TRB3, Akt, mTOR, and LC3 expression (Fig. 4C,D) and diminished A. cinnamomea-induced cell death (Fig. 4E). Autophagy inhibition but not apoptosis inhibition reversed A. cinnamomea-induced cell death (Fig. 4G,I). These findings reveal that autophagic cell death is induced via the CHOP/TRB3/Akt/mTOR pathway in response to A. cinnamomea treatment. Apoptotic cell death was observed in colorectal, lung and liver cancer cells after A. cinnamomea treatment 10,17,42,45 . On the other hand, A. cinnamomea induced autophagic cell death in liver, pancreatic, head and neck cancer cells 41,46,47 . Our study identified that A. cinnamomea also induced autophagic cell death in colorectal cancer.
Chemokines are involved in many cellular processes, such as inflammatory, innate and adaptive responses; tumour growth; and tumour angiogenesis [48][49][50][51] . CXCL1, CXCL2, CXCL3, and CXCL8 are members of CXC chemokine family. CXCL8 has been reported as a central gene that regulates a colon cancer network 52 . High correlation was observed between CXCL8 and other cytokines (CXCL1, CXCL2, and CXCL3) 53 . Previous studies showed that CXCL8 promoted cell proliferation via the activation of Akt and inhibited apoptosis in colorectal cancer [54][55][56] . In response to the chemotherapy drug 5-fluorouracil, HCT116 cells upregulated the expression of CXCL8, CXCR1, and CXCR2; CXCR1 and CXCR2 are receptors of CXCL8 57 . In our study, CXCL1, CXCL2, CXCL3, and CXCL8 were upregulated after A. cinnamomea treatment. However, our results indicated that A. cinnamomea has anticancer effects both in vitro and in vivo (Figs 2, 5). As shown in Fig. 4  different from that of CXCL8. We assumed that upregulation of CXC family members is a cellular response after A. cinnamomea treatment. A. cinnamomea exerted excellent anticancer effects on colorectal cancer. The whole-genome expression profiling of A. cinnamomea in colorectal cancer was established for the first time. We clearly demonstrated the anticancer molecular mechanisms of A. cinnamomea in colorectal cancer, revealing that A. cinnamomea induces autophagic cell death via the CHOP/TRB3/Akt/mTOR pathway. These findings provide a reference for future A. cinnamomea drug development for colorectal cancer treatment. A. cinnamomea fruiting body extracts. A. cinnamomea (Fig. 1A) was cultured on dishes for four months by Yong Teng Biotechnology Co., Ltd. (New Taipei City, Taiwan). The A. cinnamomea strain was confirmed by the PCR fragments of the internally transcribed spacers of the ribosomal RNA genes. The A. cinnamomea fruiting bodies (approximately 100 g) were soaked in 3000 mL of ethanol for 120 h. The samples were extracted three times using filter paper (Advantec No. 1, Tokyo, Japan). The combined filtrates collected from the three extractions were evaporated to dryness under vacuum. The ethanol extract (23 g, AC) from the powdered fruiting bodies of A. cinnamomea (100 g) was applied to a Diaion HP-20 column (460 g) and washed with 50% ethanol, 95% ethanol, and ethyl acetate to yield 3 fractions: ACF1 (14.71 g, yield [w/w] 71.32%), ACF2 (6.11 g, yield [w/w] 29.64%), and ACF3 (0.75 g, yield [w/w] 3.65%). The scheme depicting the methodology is shown in Fig. 1B. Cell culture. Five human colorectal cancer cell lines (HCT116, HT29, SW480, Caco-2, and Colo205) were obtained from the Bioresource Collection and Research Center (Hsinchu, Taiwan). Dulbecco's Modified Eagle's Medium containing 10% foetal bovine serum, 100 U/ml penicillin and 0.1 mg/ml streptomycin served as the cell culture medium throughout all experiments. The cells were maintained at 37 °C in a humidified atmosphere of 5% CO 2 in air and passaged every 2 days to obtain exponential growth. All five cell lines were maintained in culture no more than 15 passages. All culture reagents were sourced from Gibco (Invitrogen, Carlsbad, CA, USA). MTS assay. HCT116, HT29, SW480, Caco-2, and Colo205 cells were seeded in 96-well plates at a density of 5 × 10 4 cells per well and treated with increasing concentrations of AC, ACF1, ACF2 or ACF3 for 48 h. The cytotoxic effects of AC, ACF1, ACF2, and ACF3 on colorectal cancer cells were assessed by the MTS assay (Promega, Fitchburg, WI, USA). The absorbance was measured at 490 nm with a microplate reader. The IC 50 values were calculated using GraphPad Prism 5. HPLC analysis. ACF2 was further separated using a Thermo HPLC system (Thermo Fisher Scientific, Waltham, MA, USA) with an Eclipse XDB-C18 analytical column (Agilent Technologies, Santa Clara, CA, USA). Gradient elution was performed with mobile phases A (0.1% formic acid aqueous solution) and B (acetonitrile). The gradient elution profile was as follows: 0-3 minutes, 70% A and 30% B to 60% A and 40% B (linear gradient); 3-15 minutes, 60% A and 40% B to 42% A and 58% B (linear gradient); 15-21 minutes, 42% A and 58% B (isocratic); 21-26 minutes, 42% A and 58% B to 35% A and 65% B (linear gradient); 26-35 minutes, 35% A and 65% B to 0% A and 100% B (linear gradient); and 35-50 minutes, 0% A and 100% B (isocratic). The flow rate was 0.8 mL/ min, and the wavelength of the photodiode array detector was 254 nm. NGS analysis. HCT116 cells were seeded in 10 cm plates at a density of 2 × 10 6 cells per plate and treated with 50 μg/ml ACF2 for 24 h. Total RNA was extracted using the RNeasy Mini Kit (Qiagen, Hilden, Germany).

Preparation of
An RNA sequencing library was generated using the TruSeq Stranded mRNA Library Prep Kit according to the user's instruction manual (Illumina, San Diego, CA, USA). Briefly, the poly-A containing mRNA was isolated from total RNA using poly-T oligo-attached magnetic beads and fragmented and primed for cDNA synthesis. After double-strand cDNA synthesis, adenylation of the 3′-end, and multiple indexing adapter ligation, DNA purification with magnetic beads and PCR selective amplification were performed. Next, the amplified library was purified, quantified, and applied for template preparation. The NextSeq 500 platform was utilized to generate 75-bp single-end sequencing reads. The initial sequence reads were generated by the Genome Analyser (Illumina, San Diego, CA, USA), and the expression values were calculated using RNA-Seq by Expectation Maximization (RSEM) for each gene (University of California, Santa Cruz (UCSC), Homo.sapiens.hg19.gtf) in individual samples 58 . A total of 20 million reads were randomly selected for quality control with Bowtie 2 and mapped to Genome Reference Consortium Human Build 37/hg19 59 . Differential expression was analysed between different subgroups using Bioconductor EBseq. (version 1.1.6) 60 . A false discovery rate of 0.05 was applied to the resulting P values to correct for multiple hypothesis testing. T-tests were employed to screen genes differentially expressed between control and ACF2-treated cells, with a threshold of P < 0.05 and a posterior probability fold change >2.
Quantitative reverse transcription PCR. HCT116 cells were treated with 25, 50, and 75 μg/ml ACF2 for 24 h. Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany). cDNA was synthesized from the RNA using the SuperScript III First-Strand Synthesis System (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Quantitative real-time PCR was performed using iQ SYBR Green Supermix was analysed by the MTS assay and expressed as cell viability (% control). 5-FU was used as a positive control. Actin protein was used to normalize the results in (A-D), (F), and (H). The groups of images were cropped from different blots. Full-length blots are presented in Supplementary Figures 2 and 3. (E), (G), and (I) All results are expressed as the mean ± standard deviation of three independent experiments. P values of statistical significance are represented as *p < 0.05, **p < 0.005 and ***p < 0.0005. Western blot analysis. HCT116 cells were treated with 25, 50, 75, 100, and 125 μg/ml ACF2 for 24 h. Total cellular protein was extracted using M-PER ™ mammalian protein extraction reagent (Thermo Fisher Scientific, Waltham, MA, USA) and then quantified using the Bradford protein assay (Bio-Rad, Hercules, CA, USA). Denatured protein was loaded onto 10-15% SDS polyacrylamide gels. Following electrophoresis, the proteins on the gel were transferred onto polyvinylidene difluoride membranes using the TE22 mighty small transfer tank (Hoefer, Holliston, MA, USA). The membranes were blocked with 5% skim milk at room temperature for 30 minutes, followed by incubation at 4 °C overnight with the following primary antibodies: CHOP, TRB3, caspase3, actin (Abcam, Milton, Cambridge, UK), total and phosphorylated Akt (ser473), total and phosphorylated mTOR (ser2448), and LC3 (Cell Signaling Technology, Danvers, MA, USA). The membranes were then washed three times for 10 minutes with PBS-Tween buffer and incubated with horseradish peroxidase-conjugated anti-mouse or anti-rabbit secondary antibodies (Abcam, Milton, Cambridge, UK) at room temperature for 1 h. All antibodies were diluted 1:2,000. The antibody binding proteins were detected using ECL chemiluminescence substrates and then captured on the Amersham Imager 600 (GE Healthcare, Chicago, IL, USA). The membranes were reprobed for actin protein expression to show that similar amounts of protein were loaded in each lane.

Scientific
siRNA transfection. For siRNA inhibition, double-stranded RNA duplexes targeting human CHOP (5′-GCC UGG UAU GAG GAC CUG C-3′) or human TRB3 (5′-CGA GCU CGA AGU GGG CCC C-3′) and control siRNA (5′-GCG CGC UUU GUA GGA UUC G-3′) 32 were synthesized by Invitrogen. HCT116 cells were transfected with siRNA using Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. In vivo experiments. Six-week-old male nude nu/nu mice were obtained from the National Laboratory Animal Center (Taipei, Taiwan) and housed in clean specific pathogen-free (SPF) rooms. The mice were inoculated with 3 × 10 6 HCT116 cells suspended in PBS subcutaneously into the right flanks. Twenty-four mice were randomly divided into four groups (6 mice in each group): a control group (treated with 0.2 mL of normal saline) and 100, 200, and 400 mg/kg ACF2 treatment groups. ACF2 treatment was started on the seventh day after tumour inoculation by daily oral gavage for 15 days. The body weights of the mice were recorded every three days. Tumour volume was measured every three days using digital callipers and calculated as length × (width) 2 /2. The experimental protocol was approved by the Institutional Animal Care and Use Committee at National Taiwan University. All methods were carried out in accordance with relevant guidelines and regulations.

Statistical analysis.
All experiments were performed in triplicate, and data are presented as the mean ± standard deviation. Statistical analysis was performed by the ANOVA and LSD tests using SPSS 12.0 software (SPSS, Inc., Chicago, IL, USA). P values of statistical significance are represented as *p < 0.05, **p < 0.005 and ***p < 0.0005.