Withaferin A triggers G2/M arrest and intrinsic apoptosis in glioblastoma cells via ATF4‐ATF3‐CHOP axis

Abstract Objective Withaferin A (WA) is a bioactive compound with a remarkable anti‐cancer effect derived from Withania somnifera, commonly known as ashwagandha. However, the anti‐cancer mechanisms of WA in glioblastoma multiforme (GBM) are still unclear. Materials and Methods Cell viability assays and xenografted nude mice were used to evaluate the effects of WA, along with flow cytometry to detect apoptosis and cell cycle of GBM. RNA‐seq analysis, Western blotting, immunofluorescence staining, qRT‐PCR and siRNA gene silencing were carried out to determine the signalling pathways affected by WA. Results Withaferin A significantly inhibited the growth of GBM in vitro and in vivo and triggered the intrinsic apoptosis of GBM cells by up‐regulating expression of Bim and Bad. WA arrested GBM cells at the G2/M phase of the cell cycle through dephosphorylating Thr161 of CDK1 by activating p53‐independent p21 up‐regulation. Knockdown of p21 restored cell cycle progression and cell viability by down‐regulating the expression of Bad rather than Bim. We demonstrated that endoplasmic reticulum (ER) stress induced by WA through the ATF4‐ATF3‐CHOP axis, initiated apoptosis and G2/M arrest in GBM cells. Conclusion We revealed a novel pathway that elucidated WA activation of apoptosis and G2/M arrest in GBM cells through the ATF4‐ATF3‐CHOP axis. This discovery is important for optimization of WA‐based regimens for prevention and/or treatment of GBM.

effective treatment of various ailments in Indian Ayurvedic medicine for centuries. Pharmacological studies of the effects of WA have shown that it is anti-diabetic, 5 protects liver from acetaminopheninduced injury, 6 prevents mammary cancer, 7 inhibits tumour cell invasion and metastasis, 8,9 induces cell cycle arrest 10,11 and triggers apoptosis in breast, prostate, colorectal, non-small cells lung and pancreatic cancer. [12][13][14][15] However, the underlying mechanisms of WA's effects on cancer are still unclear.
The endoplasmic reticulum (ER) stress response can be triggered by numerous conditions that cause imbalances in intracellular homeostasis, which in turn threaten proper cell functioning. In response, the ER stress response activates an adaptive effort to neutralize these threats for restoring homeostasis. When these countermeasures and severe imbalances persist, the ER stress response may initiate a pro-apoptotic program to eliminate the faulty cells for the survival of the whole organism. 16,17 Otherwise, it is also reported that prolonged ER stress caused cell cycle arrest. 18,19 Choi et al demonstrated that WA induced CHOP-mediated apoptosis. 20 However, the relationship among ER stress, cell cycle arrest and apoptosis triggered by WA is still unclear.
In this study, we demonstrated that WA significantly inhibited the growth of U87 and U251 cells by inducing intrinsic apoptosis and arresting the cell cycle at the G2/M stage. WA also suppressed the growth of GBM cells in vivo. Furthermore, we have revealed a new mechanism for WA-induced apoptosis and cell cycle arrest through the ATF4-ATF3-CHOP axis. These results suggested that WA inhibited the growth of GBM by a novel mechanism.

| Cell culture
The human glioblastoma cell lines, U251 and U87, were purchased from the Cancer Institute & Hospital, Chinese Academy of Medical Sciences. The human normal astrocyte cells HA1800 were bought from SCIENCELL. Cells were maintained with DMEM supplemented with 10% FBS, streptomycin (100 μg/mL) and penicillin (100 U/mL).
The cells were cultured at 37°C in an incubator with a humidified atmosphere of 5% CO 2 .

| MTT assay
Cells were seeded into 96-well plates at 5 × 10 3 per well, cultured for 24 hours and treated with WA for the indicated time. After washed with PBS, cells were incubated with serum-free DMEM containing 0.5 mg/mL MTT for 3-4 hours. The supernatants were carefully removed and discarded, and the formazan was dissolved in 100 μL dimethyl sulfoxide, followed by measurement with a SpectraMax M5 plate reader (Molecular Devices) at 570 nm.

| Apoptosis assay
The apoptosis rate of cells was determined by using a Annexin V-FITC kit from Beyotime Biotechnology. After treatment with WA for the indicated time, cells were harvested and washed with PBS.
The cell pellets were resuspended in an annexin v-FITC buffer and incubated on ice for 7 minutes. Then, PI was added into the mixture, and after 3 minutes, cell apoptosis was measured by flow cytometry (FACS Calibur, BD BioSciences). The results were analysed by FlowJo 7.6 software.

| JC-1 assay
The JC-1 dye assay was conducted using JC-1 kits purchased from Beyotime Biotechnology according to manufacturer's instructions.
Briefly, after treated with WA for the indicated time, cells were harvested and washed three times with PBS. The cells were incubated in JC-1 buffer for 20 minutes at 37 °C. After washing once with PBS, the cells were resuspended in PBS and measured by flow cytometry.
Healthy mitochondria show red florescence, while damaged organelles fluoresce green. The primers used are listed in the supplemental materials section (Table S2). GAPDH served as internal control.

| siRNA transfection
siRNA duplexes were obtained from Genepharm and used to transfect cells according to the recommended procedure. 21 Briefly, U251 cells were seeded into 6-well plates and cultured for 24 hours at  (Table S3).

| Immunofluroescence
Immunofluorescence was performed according to a recommended procedure. 22

| Glioblastoma xenograft assay in nude mice
Four-to five-week-old athymic nude mice (

| Statistical analysis
All experiments were repeated three times. The data were expressed as mean ± SD. Statistical analysis was carried out using Origin 9.0 software, and comparisons of each group were made by one-way analysis of variance (ANOVA) using SPSS 19 software. Results were considered statistically significant at P < .05.

| WA-induced apoptosis of U87 and U251 cells via the intrinsic pathway
To confirm the effects of WA ( Figure 1A) on viability of GBM cells and normal astrocytes, an MTT assay was carried out on U87, U251 and HA1800 cells. As shown in Figure 1B, WA inhibited the growth of U87 and U251 more than normal astrocyte cells, HA1800, at the same concentration (1 or 3 μmol/L). The half-maximal inhibitory concentration (IC 50 ) in U87, U251 and HA1800 cells treated with WA for 48 hours was 4.61, 1.37 and 9.13 μmol/L, respectively. Moreover,

WA inhibited the growth of U87 and U251 cells in a concentration-
and time-dependent manner ( Figure 1C). To further investigate the effects of WA on the survival of U87 and U251 cells, the apoptosis rate was measured by Flow cytometry. The number of U87 or U251 cells in the Q2 and Q3 quadrants was augmented with increasing WA concentration and treatment time, indicating that WA-induced apoptosis in a concentration-and time-dependent manner ( Figure 1D).
Once the apoptotic process is initiated, pro-caspases will be cleaved to active forms and the commands of apoptosis, such as cleaving PARP to inhibit its activity to facilitate DNA damage will be executed. 23,24 As shown, cleaved PARP1 and caspase 3/7/9 were remarkably elevated in U87 and U251 cells after 24 hour treatment with 3 μmol/L WA ( Figure 1E and Figure S1A). However, there was no change in cleaved caspase 8, which represents the extrinsic apoptotic pathway. These results suggested that WA induced apoptosis of U87 and U251 cells via the intrinsic pathway. After pre-treatment with 50 μmol/L Emricasan (a pan-caspase inhibitor) for 24 hours, cells were incubated with 1 or 3 μmol/L WA for 48 hours. Results showed that the effects of WA on cell viability and apoptosis were almost completely abrogated by Emricasan ( Figure 1F and 1G, Figure S1B).
Taken together, these results suggested that WA induced apoptosis of U87 and U251 cells via the intrinsic pathway.

| WA-induced apoptosis of U87 and U251 cells partly by up-regulating expression of Bim and Bad
To explore how WA activated the intrinsic apoptotic pathway, JC-1 staining was carried out. As shown in Figure 2A, WA reduced mitochondria membrane potential in a concentration-and time-dependent manner. WA treatment slightly changed the expression of the pro-apoptotic proteins, Bax and Bak, as well as anti-apoptotic proteins, Bcl-2 and Bcl-xL, while expression of the pro-apoptotic proteins, Bim and Bad, was significantly increased after 6 hours ( Figure 2B and Figure S2A). The phospho-Ser 112 level of Bad was decreased after treatment with WA for 24 hours ( Figure 2B and Figure S2A). Thus, Bim and Bad may be the key regulators that initiated the intrinsic apoptotic pathway. To confirm this hypothesis, the expression of Bad and Bim was knocked down by corresponding siRNA in U251 cells. As expected, the viability of WA-treated U251 was increased by siRNAs of Bad or Bim ( Figure 2C). Similarly, the alterations of cleaved caspase 3 and 9 caused by WA in U251 cells were also partly reversed by siRNAs of Bad or Bim ( Figure 2D and Figure S2B). Together, these results suggested that the apoptosis induced by WA was partly mediated by up-regulating the expression of Bad and Bim.

| Cell cycle was arrested at the G2/M phase by WA through p53-independent p21 up-regulation
To determine the effects of WA on cell cycle in U87 and U251 cells, PI was used to stain the cells after WA treatment for 6, 12, 24 and 48 hours. As shown in Figure 3A, the number of cells at  Figure 3B and Figure S3A). The total CDK1 expression was not affected by WA treatment, but the levels of phospho-Thr 161 CDK1 were significantly decreased from 6 hours in both cell types, indicating that the active site of CDK1 was inhibited ( Figure 3B and Figure S3A). The cyclin-dependent kinase inhibitor, p21, suppresses the phosphorylation of CDK1 at Thr 161 , and we thus measured its expression by Western blotting. We found that p21 was up-regulated in both cell types after treatment with 3 μmol/L WA for 6 hours ( Figure 3B and Figure S3A), implying that p21 was and HA1800 cells. *P < .05, **P < .01 and ***P < .01 represented significant differences between U87 and U251 cells vs HA1800 cells. C. Cell viability was measured by MTT assay. *P < .05, **P < .01 and ***P < .001 represented significant differences between the WA-treated group and control group. D, The apoptotic rates were determined by Flow cytometry after staining for Annexin V conjugated with FITC and PI dyes. The Q1, Q2, Q3 and Q4 quadrants represented dead, late-apoptotic, early-apoptotic and normal cells, respectively. *P < .05 and **P < .01 indicated significant differences between the WA-treated group and control group. E, Apoptotic proteins were identified by Western blotting. F, Cell viability treated with 3 μmol/L WA was measured by MTT assay after pre-treated with Emiricasan (50 μmol/L) for 24 h. *P < .05 and **P < .01 represented the significant differences between the Emiricasan-treated group and corresponding non-Emiricasan-treated group. G. Western blot was used to check the alteration of apoptotic proteins after pre-treatment with Emricasan. WA, Withaferin A  Figure 3C). Interestingly, the loss of cell viability caused by WA was significantly restored by p21 siRNA ( Figure 3D). Thus, we further investigated the relationship between p21 and Bad or Bim.
The expression of Bad rather than Bim was remarkably reduced by p21 siRNA in U251 cells ( Figure 3E and Figure S3B). Furthermore, p21 knockdown reduced the levels of cleaved PARP1, an apoptotic marker ( Figure 3E and Figure S3B). Taken together, all these results suggested that WA induced apoptosis by regulating the p21-Bad pathway.

| Expression of HMOX1, DNJB1 and ATF3 was induced by WA at an early stage of treatment
To explore the early responses to WA, the cells were treated with 3 μmol/L WA for 12 hours and used to perform RNA sequence analysis. After analysing the data from RNA-seq, we found 276 up-regulated genes and 264 down-regulated genes in U87 cells, as well as F I G U R E 2 WA-induced apoptosis of U87 and U251 cells partly by up-regulating expression of Bim and Bad. After treatment with WA at the indicated concentrations for the indicated times, cells were separately harvested for Flow cytometry, Western blotting and MTT assay. For siRNA silencing, U251 cells were transfected with siRNA of Bim or Bad for 48 h and treated with 3 μmol/L WA for an additional 24 h. A, Mitochondrial membrane potential was determined by Flow cytometry after staining with JC-1 dyes. The Q2 and Q3 quadrants represented normal cells and cells of decreased mitochondrial membrane potential, respectively. *P < .05, **P < .01 and ***P < .001 represented significant differences between the WA-treated group and control group. B, Proteins regulating the intrinsic apoptotic pathway were detected by Western blotting. C, Cell viability of U251 was determined by MTT assay after transfection with siRNA of Bim or Bad. *P < .05 and **P < .01 represented significant differences between the siRNA-treated group and corresponding non-siRNA-treated group. D, The changes of regulated proteins in intrinsic apoptotic pathway in U251 cells were determined by Western blotting after transfection with siRNA of Bim or Bad. WA, Withaferin A F I G U R E 3 The cell cycle was arrested at the G2/M phase by WA through p53independent p21 up-regulation. Cells were treated with WA at the indicated concentration for the indicated time and then assayed by flow cytometry and Western immunoassay. For siRNA silencing, U251 cells were transfected with siRNA of p21 for 48 h then treated with 3 μmol/L WA for 24 h longer, and finally analysed by MTT, Flow cytometry and Western blotting. A, Cell cycle was determined by Flow cytometry after staining with PI. *P < .05, and **P < .01 represented significant differences of cells at G2/M phase vs control group. B, Proteins regulating the G2/M phase were detected by Western blotting. C, Cell cycle analysis after p21 knockout was performed by Flow cytometry. **P < .01 represented significant differences between the sip21-treated group and NC-treated group. D, Cell viability of U251 was measured by MTT assay after transfection with siRNA of p21. *P < .05 represented significant differences between the siRNA-treated group and the corresponding non-siRNA-treated group. E, The changes of Bim, Bad and cleaved PARP1 proteins in U251 cells were checked by Western blotting after transfection with siRNA of p21. WA, Withaferin A 374 up-regulated genes and 264 down-regulated genes in U251 cells ( Figure 4A). All the differentially expressed genes were uploaded on the DEVID website for GO enrichment to search for involvement of potential cell signalling pathways. The top ten GO enrichments are listed in Figure 4B. These GO enrichments in U87 cells mainly fo-  Figure 4C. Because U251 cells were more sensitive to WA than U87 cells, the genes with higher changes in U251 than U87 were much more likely to be targets for WA. Therefore, HMOX1, PP1R15A, DNAJB1, ATF3 and SH3BGR were tested by perform- ATF3 were significantly elevated after 6 hours treatment with WA in both U87 and U251 cells whereas the expression of DNAJB1 was only increased in U251 cells at an early stage ( Figure 4E and Figure   S4). In addition, PP1R15A was also increased by WA after 12 hours treatment. The results of immunofluorescence confirmed the above results and showed that there was no nuclear translocation of HMOX1, PP1R15A, DNAJB1 and ATF3 occurring in U251 cells ( Figure 4F).

| WA-induced apoptosis and G2/M arrest of GBM cells by ATF4-ATF3-CHOP axis
To investigate the mechanism underlying the effects of WA on GBM cells, RNA-seq was carried out to determine the transcriptional Our results demonstrated that WA triggered up-regulation of ATF4, XBP1 and CHOP, but had little effect on ATF6 and GRP78 at the levels of both mRNA and protein ( Figure 5A and 5B, Figure S5A). WA increased the expression of ATF4, followed by XBP1-s and CHOP ( Figure 5B and Figure S5A). When expressions of HMOX1, DNAJB1, ATF3, ATF4, XBP1 and CHOP were knocked down by corresponding siRNAs, the viability of U251 cells inhibited by 3 μmol/L WA was significantly (P < .01) recovered by ATF3, ATF4 and CHOP siRNAs ( Figure 5C). The induction of apoptosis by WA and its ( Figure 5D) effect on mitochondrial membrane potential ( Figure 5E) and cell cycle ( Figure 5F) in U251 cells were counteracted by separately knocking down ATF3, ATF4 or CHOP. The positive effects of WA in p21, Bad, Bim, cleaved caspase 3/7/9 and cleaved PARP1 were also diminished by all three siRNA duplexes ( Figure 5G and Figure S5B). Those results suggested that ATF3, ATF4 and CHOP may be the pivotal regulators for WA-induced apoptosis and G2/M arrest. The results shown in Figure 5E demonstrated a logical relationship among ATF3, ATF4 and CHOP-ATF4 positively regulated ATF3, subsequently modulating the expression of CHOP.

| WA inhibited the growth of U87 xenograft in nude mice
To check whether WA had anti-tumour effects in vivo, U87 xenografts in nude mice were used. In this study, U87 cells (5 × 10 6 ) in PBS were injected subcutaneously into the right flank. As shown in Figure 6, the weight of the mice was slightly decreased in both experimental and control groups while the volume and weight of tumours in the group injected with 5 mg/Kg WA were significantly smaller than in the control group injected with only saline. These results demonstrated that WA treatment could inhibit the growth of U87 xenografts in mice. A, Genes associated with ER stress were determined by qRT-PCR. *P < .05, **P < .01 and ***P < .001 represented significant differences between U251 cells and U87 cells. B, Proteins associated with ER stress were identified by Western blotting. C, Cell viability of U251 cells treated by 3 μmol/L WA was measured by MTT assay after transfection with siRNAs of HMOX1, DNAJB1, ATF3, ATF4, XBP1 and CHOP, respectively. **P < .01 represented significant differences between the siRNA-treated group and the corresponding negative control group. D, Apoptosis of U251 treated with 3 μmol/L WA was measured by Fow cytometry after transfection with siRNAs of ATF3, ATF4 and CHOP, respectively. *P < .05 and **P < .01 represented significant differences between the siRNA-treated group and the corresponding negative control group. E, Mitochondrial membrane potentials of U251 treated with 3 μmol/L WA were measured by Flow cytometry after transfection with siRNAs of ATF3, ATF4 and CHOP, respectively. *P < .05 and **P < .01 represented significant differences between the siRNA-treated group and the corresponding negative control group. F, The cell cycle stage of U251 treated with 3 μmol/L WA was determined by Flow cytometry after transfection with siRNAs of ATF3, ATF4 and CHOP, respectively. **P < .01 and ***P < .01 represented significant differences between the siRNA-treated group and the corresponding negative control group. G, The changes of p21, Bim, Bad and cleaved-PARP1, caspase 3/7/9 and cleaved caspase 3/7/9 proteins in U251 cells treated with 3 μmol/L WA were determined by Western blotting after transfection with siRNAs of ATF3, ATF4 and CHOP, respectively. H, The overall mechanism of WA activity in GBM cells. WA, Withaferin A inhibition on U251 and U87 cells than on the normal astrocyte line, HA1800 ( Figure 1B), as well as on a U87 xenograft implanted subcutaneously in nude mice (Figure 6), suggesting that WA may be an effective drug for GBM therapy.

| D ISCUSS I ON
Overwhelming  14 Stan et al addressed the mechanism by which WA caused Bim-dependent apoptosis. 26 Our results showed that WA induced apoptosis of GBM cells by the intrinsic rather than the extrinsic pathway ( Figure 1). Furthermore, our data showed that expression of Bax, Bak, Bcl-2 and Bcl-xL was not affected by WA in GBM cells ( Figure 2B and Figure S2A).
Besides Bim which played an important role in intrinsic apoptotic pathway induced by WA, the expressions of Bad and dephosphorylated Ser 112 of Bad were induced to promote apoptosis of GBM cells ( Figure 2C and 2D, Figure S2B), which was first reported by our group.

F I G U R E 5 (Continued)
It is well known that WA can cause cell cycle arrest at the G2/M phase in cancer cells. 10,11,27 Our data confirmed that WA was able to arrest the cell cycle of U87 and U251 cells at the G2/M phase ( Figure 3A). Previous studies reported that the mechanisms involved in effects of WA on cell cycle were accumulation of Tyr 15 phosphorylated (inactive) CDK1 via a decrease in levels of cell division cycle 24B (Cdc24B) and/or Cdc24C proteins, 27 p51-dependent p21 upregulation 28 and down-regulation of cyclins (E2, A and B1). 29 In our study, the results showed that cyclin A and B1 were down-regulated in the later stages of WA treatment (24 and 48 hours). However, significant up-regulation of p21 appeared after 6 hours of WA treatment. Moreover, silencing of p21 with siRNA partly reversed the G2/M phase arrest induced by WA ( Figure 3C). In addition, WA did not change the level of p53 but dephosphorylated CDK1 at Thr 161 ( Figure 3B and Figure S3A). These results suggested that WA induced arrest of G2/M phase by dephosphorylating CDK1 at Thr 161 via a p53-independent p21 up-regulation. More interestingly, p21 knockdown contributed to a decrease in the levels of cleaved PARP1 and Bad rather than Bim, suggesting that WA may induce GBM apoptosis via the p21-Bad axis ( Figure 3D and 3E, Figure S2B).
One important aim of our study was to explore the early protein response to WA. Thus, RNA-seq analysis was carried out and the results implied that ER stress may be a breakthrough point. The ER is an organelle that is responsible for protein folding and assembly, and exquisitely sensitive to alterations in homeostasis. 30 Under an unfolded protein response (UPR), three primary ER-localized protein stress sensors (IRE1α, PERK and ATF6) were activated. 31 and ATF4 form the complex to improve the transcription of CHOP for induction of apoptosis. 41,42 Our work showed that treatment with WA up-regulated expression of ATF4, and then ATF3, CHOP as well as XBP1-s ( Figure 4 and Figure 5). Knockdown of only one of the associated proteins, ATF4, ATF3 and CHOP, could significantly restore the inhibition of U251 cells caused by WA ( Figure 5C). Furthermore, blocking their expression significantly reduced the inhibition of WA in apoptosis ( Figure 5D), mitochondrial membrane potential ( Figure 5E) and cell cycles ( Figure 5F), as well as expression levels of p21, Bad, Bim, cleaved PARP1 and caspase 3/7/9 ( Figure 5G and Figure S5B).
Our work also revealed that treatment with WA triggered the up-regulation of ATF4, which induced expression of ATF3, and then positively modulated CHOP ( Figure 5G and Figure S5B).
In summary, we demonstrated that WA exhibited significant growth inhibitory effect against GBM cells in vitro and in vivo, which is associated with a Bim-and Bad-dependent intrinsic apoptotic pathway and arrest of G2/M phase induced by p21 up-regulation.
The novel upstream regulator of Bim, Bad and p21 discovered in our study is the ATF4-ATF3-CHOP axis. The overall mechanism is illustrated in Figure 5H. Our study provides important mechanistic insights into using WA in future clinical scenarios where it may synergize with current therapeutic strategies.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest that pertain to this work.

AUTH O R CO NTR I B UTI O N S
Qin Tang designed the research methods, performed the experiments, analysed the data and drafted the manuscript. Liwen Ren and Jinyi Liu participated in flow cytometry experiments and data collection.

Xiangjin Zheng and Wan Li collected tumour tissues. Jinhua Wang and
Guanhua Du designed the research and revised the manuscript. All authors have read and approved the final manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.
F I G U R E 6 WA inhibited the growth of subcutaneous U87 xenografts. U87 cells (5 × 10 6 ) were subcutaneously injected into the right flank of nude mice. After the tumour reached a volume of 40-50 mm 3 , the mice were injected with vehicle or 5 mg/kg WA in the tail vein every day for 1 month. A, WA had no effect on the weight of the mice. B, WA reduced the tumour volume. C, WA decreased the tumour weight. **P < .01 represented significant difference between the WA group and the vehicle group. D, The photographs of tumours were collected from different groups of mice at the end of treatment (day 27