Sublethal cinobufagin increases ROS levels specifically in cancer cells
To evaluate the in vitro anticancer potency of cinobufagin and determine the dose range for subsequent experiments, we studied the IC50 values of CBG against the human colorectal adenocarcinoma cell lines SW480 and SW1116. After treatment with CBG for 24, 48 or 72 h, the IC50 values were 103.60, 35.47 or 20.51 nM in SW480 cells and 267.50, 60.20 or 33.19 nM in SW1116 cells (Fig. 1A), all were in the low nanomolar range, indicating that CBG was potently cytotoxic in these cancer cells.
Similar to the SW480 and SW1116 colorectal cancer cells, the viability of some diverse human cancer cell lines, including the osteosarcoma MG-63, hepatocellular carcinoma HepG2, melanoma M21, and two other colorectal carcinoma cell lines HCT116 and SW620, was also reduced by 40–50% after 24 h of treatment with 100 nM CBG, a dose close to the IC50 value of 24-h treatment in SW480 cancer cells (Fig. 1B). Similar results were also showed by a 5-day colony formation assay with the pancreatic epithelioid carcinoma PANC-1, cervical squamous cell carcinoma SiHa, hepatocellular carcinoma HepG2, lung epithelial carcinoma A549 and adenocarcinoma HCC827, and the colorectal adenocarcinoma SW480 and SW1116 cells (Additional file 1: Figure S1A). Together, these results demonstrated a similar degree of CBG-induced cytotoxicity in a broad range of cancer cell types. In stark contrast, the viability of three noncancerous human cell lines, including the NCM460 colon and BEAS-2B lung epithelial cell lines, and the L-O2 hepatocyte cell line, was almost not affected by the same treatment (Fig. 1B, Additional file 1: Figure S1A), revealing an interesting difference between cancer and noncancerous cells towards CBG’s cytotoxicity-inducing activity.
Oncogenic transformation is associated with increased generation of ROS and agents that promote generation of additional ROS or weaken the antioxidant systems may push oxidative pressure to toxic levels selectively in cancer cells, resulting in cancer-specific oxidative toxicity. CBG has been shown to markedly increase ROS levels in cancer cells. To understand the mechanisms underlying the differential cytotoxicity of CBG in cancer and noncancerous cells, we measured cellular ROS levels by the oxidant-sensing probe 2’,7’-dichlorodihydrofluorescein diacetate (DCFH-DA). Remarkably, a prominent increase in ROS levels was revealed in the SW480 and SW1116 colorectal cancer cells after 3 h of treatment by 100 nM CBG (Fig. 1C, D); in contrast, similar treatments caused no change in ROS levels in the noncancerous NCM460 colon epithelial cells (Fig. 1C, D). Similarly, a marked increase in ROS levels was induced by the same treatment in the lung epithelial carcinoma A549 and HCC827 and the hepatocarcinoma HepG2 cells but not in the noncancerous BEAS-2B lung epithelial cells and the L-O2 hepatocytes (Additional file 1: Figure S1B–C). Significant ROS increase in the cancer cells was evident as early as 15 min after 100 nM CBG treatment (Fig. 1E), and the ROS inhibitor N-acetyl cysteine effectively blocked the ROS increase in the cancer cells (Fig. 1C, D, Additional file 1: Figure S1B, C). Interestingly, NAC also significantly reduced the cytotoxicity of CBG in cancer cells (Fig. 1F), correlating CBG-induced cytotoxicity with ROS elevation. Thus, a rapid increase in ROS levels was induced specifically in cancer cells by a sublethal dose of CBG, and ROS elevation resulted in cytotoxicity selectively in cancer cells.
Sublethal cinobufagin-induced ROS overload leads to oxidative DNA damage
Elevated ROS can cause oxidative DNA damage that may lead to cell cycle arrest, premature cellular senescence, or programmed cell death if the damage results in intense DNA damage response signaling. To investigate if activation of DDR was responsible for CBG-induced suppression of cancer cell viability, we first checked whether oxidative DNA damage was resulted from CBG-induced ROS elevation.
One of the most common targets of ROS is the nucleobase guanine in both nucleic acid macromolecules (DNA and RNA) and the free nucleotides dGTP and GTP. Oxidation of guanine generates 8-oxoguanine (8-oxoGua) which can be revealed by labeled avidin [39]. Immunofluorescent staining using Alexa 488-conjugated avidin showed that treatment with 100 nM CBG for 3 h markedly increased nuclear 8-oxoGua levels in the SW480 cancer but not in the NCM460 noncancerous colon epithelial cells (Fig. 2A, B). Similarly, a significant increase in 8-oxoGua levels was induced in the A549 lung epithelial and HepG2 hepatocellular carcinoma but not in the noncancerous BEAS-2B lung epithelial cells and the L-O2 hepatocytes (Additional file 2: Figure S2A), demonstrating the presence of CBG-induced, cancer-specific oxidative DNA damage. NAC effectively blocked production of 8-oxoGua in the cancer cells (Fig. 2A, B, Additional file 2: Figure S2A), correlating 8-oxoG generation with CBG-induced ROS overload.
Another major form of oxidative DNA damage is single-strand DNA breaks (SSBs) produced either directly through ROS-mediated oxidization or as intermediates of base excision repair (BER) of oxidized nucleobases. Double-strand DNA breaks (DSBs) may be generated when DNA replication forks collide with SSBs or DNA repair complexes during DNA replication. The alkaline comet assay provides a measure of total DNA strand breaks in single cells because alkaline treatment converts all SSBs into DSBs. The results of alkaline comet assay showed that treatment by 100 nM CBG for 3 h greatly increased the number of total DNA strand breaks in the SW480, A549 and HepG2 cancer but not in the NCM460, BEAS-2B and L-O2 noncancerous cells, and NAC effectively blocked the generation of DNA breaks (Fig. 2C, D, Additional file 2: Figure S2B).
OGG1 is the glycosylase that removes 8-oxoGua and cut the DNA strand to generate an SSB during BER. Pre-incubation with OGG1 significantly increased the number of DNA breaks revealed by alkaline comet assay in the cancer cells (Fig. 2C, D, Additional file 2: Figure S2B), indicating the presence of a large number of 8-oxoGua in the cancer DNA.
To check if DSBs were produced in CBG-treated cancer cells, we stained 53BP1, a protein that concentrates at sites of DSB to yield 53BP1 staining foci, therefore allowing direct visualization and measurement of DSBs. Immunofluorescent staining showed that 3 h of treatment by 100 nM CBG markedly increased the number of SW480, A549 and HepG2 cancer cells with strongly stained nuclear 53BP1 foci (Fig. 2E, Additional file 2: Figure S2C), while no change in 53BP1 staining signal was evident in similarly treated NCM460, BEAS-2B and L-O2 noncancerous cells (Fig. 2E, Additional file 2: Figure S2C). Increase in the number of 53BP1-positive cancer cells became significant 1 h after treatment by 100 nM CBG and continued in a time-dependent manner (Fig. 2F). NAC effectively blocked generation of 53BP1 foci (Fig. 2E-F, Additional file 2: Figure S2C). These results showed that some of the DNA strand breaks detected by the alkaline comet assay were DSBs and confirmed the presence of DSBs, the most toxic form of DNA damage, in CBG-treated cancer cells.
Together, these data demonstrated that sublethal CBG-induced ROS elevation immediately resulted in extensive oxidative DNA damage, including DSBs, specifically in cancer cells.
Replication stress and DDR are resulted from cinobufagin-induced DNA damage
Acute generation of extensive DNA damage may result in collision between moving DNA replication forks and damaged DNA or DNA repair complexes, leading to generation of DSBs and replication stress, both of which may activate the ATM-Chk2 or ATR-Chk1 DDR signaling pathway.
Indeed, in the SW480 and SW1116 colorectal cancer as well as the A549 lung and HepG2 hepatocellular carcinoma cells, treatment by 100 nM CBG produced a rapid and progressive increase in the number of cells with strong pan-nuclear γH2AX staining, which was not seen in the noncancerous NCM460, BEAS-2B and L-O2 cells (Fig. 3A, B, Additional file 3: Figure S3A). A marked increase in the levels of γH2AX was also revealed by Western blot analyses in the cancer but not noncancerous cells (Fig. 3C, Additional file 3: Figure S3B), indicating the presence of intense replication stress. Consistently, a time-dependent increase in phospho-RPA32 (RPA32-pS4, S8) levels was demonstrated by Western blot in SW480 and SW1116 colorectal cancer cells (Additional file 3: Figure S3C). Generation of γH2AX-positive cells and the increase in γH2AX levels in the cancer cells were effectively blocked by NAC (Fig. 3A, B, Additional file 3: Figure S3A–B), correlating CBG-induced replication stress with induction of oxidative DNA damage.
Similar to induction of replication stress, Western blot analyses also showed a time-dependent, significant increase in levels of phosphorylated Chk1 and Chk2, demonstrating strong activation of both ATR-Chk1 and ATM-Chk2 DDR signaling pathways specifically in the cancer but not noncancerous cells (Fig. 3C, Additional file 3: Figure S3B). Again, increase in phosphorylation of Chk1 and Chk2 in the cancer cells was reversed by NAC (Additional file 3: Figure S3B). The induction of increase in the levels of phosphorylated γH2AX, Chk1 and Chk2 in the cancer cells by CBG was similar to that induced by a positive control H2O2 (Additional file 3: Figure S3D).
Interestingly, DSBs, represented by 53BP1 foci, became markedly increased 1 h after CBG-treatment (Fig. 2F), while phosphorylation of Chk2, which is stimulated mainly by DSB, peaked after 3 h of CBG treatment (Fig. 3C), and phosphorylation of Chk1, which is stimulated primarily by replication stress, became significantly increased after 6 h of CBG treatment (Fig. 3C). Taken together, these results showed that CBG-induced oxidative DNA damage resulted in generation of DSBs and intense replication stress, which subsequently caused the activation of the ATM-Chk2 and ATR-Chk1 DDR signaling pathways, respectively.
DDR signaling activates the G2/M cell cycle checkpoint
DDR signaling can cause cell cycle arrest in G1, S or G2 phase via mechanisms including phosphorylation-mediated inactivation of the phosphatase CDC25 and/or upregulation of the p53-p21Cip1/Waf1 axis; p53 can also upregulate pro-apoptotic proteins to induce apoptosis. Cell cycle arrest and apoptosis will both result in suppression of cell viability.
In the SW480 and SW1116 cancer cells, Western blot analyses showed that treatment by 100 nM CBG induced a significant, progressive decrease in the levels of cyclin B and conspicuous accumulation of CDK1-pT15 (Additional file 3: Figure S3C), indicating activation of the G2/M cell cycle checkpoint. Consistently, treatment by 100 nM CBG induced a time-dependent, prominent increase in phosphorylation of CDC25C, the key driver of G2/M transition, and a corresponding decrease in the total protein levels of CDC25C (Fig. 4A), demonstrating that CDC25C was rapidly inactivated by DDR signaling to block G2/M transition, i.e., to activate the G2/M cell cycle checkpoint. The levels of both phosphorylated and total p53 proteins, as well as p21Cip1/Waf1, were similarly increased in the CBG-treated SW480 and SW1116 cancer cells, indicating activation of the p53-p21Cip1/Waf1 axis (Fig. 4A), which could promote activation of the G1, S or G2/M checkpoints.
Consistent with the results of Western blot indicating activation of the G2/M checkpoint, flow cytometry analyses showed that treatment by 100 nM CBG caused a rapid and progressive accumulation of SW480 and SW1116 cancer cells in the 4n group and a fast decrease in the size of the 2n population (Fig. 4B, C), confirming the induction of G2 arrest. The size of the 2n-4n population showed nearly no change (Fig. 4B, C), suggesting little or no induction of cell cycle arrest in S phase, which was consistent with the constant protein levels of CDC25A (Fig. 4A).
Apoptosis is induced after G2 arrest
Cell cycle analyses by flow cytometry showed that treatment by 100 nM CBG resulted in a fast and continuous increase in the size of the subG1 population (cells with < 2n DNA), which reached ~ 33% after treatment by 100 nM CBG for 48 h (Fig. 4B, C), indicating a time-dependent, steady increase of cell death. To evaluate the modes of cell death, we examined the levels of cleaved caspase 3 by Western blot. The results showed that the amount of activated caspase 3 was markedly increased after 24 h of treatment by 100 nM CBG and reached much higher levels after 48 h of treatment (Fig. 5A), suggesting significant induction of caspase-dependent apoptosis. CDC25C was intensely phosphorylated within hours of CBG treatment to implement G2 arrest (Fig. 4A), while caspase 3 became significantly activated after 24 h of CBG treatment (Fig. 5A), suggesting apoptosis was induced after G2 arrest, likely as a consequence of extended G2 arrest and DDR signaling due to persistent stress and/or unresolvable damage.
Consistent with the increase in caspase 3 activation, flow cytometry analyses showed that the number of Annexin V-positive SW480 and SW1116 cancer cells was significantly increased by CBG treatment in a time-dependent manner, reaching ~ 38% after treatment by 100 nM CBG for 48 h (Fig. 5B, C). The pan-caspase inhibitor Z-VAD-FMK blocked the increase of Annexin V-positive cells (Fig. 5C), confirming the induction of caspase-dependent apoptosis by a sublethal dose of CBG.
Moreover, labeling of apoptotic cells by the TUNEL assay revealed that sublethal doses of CBG induced a marked, dose-dependent increase in TUNEL-positive SW480 cancer cells (Fig. 5D), and measurement of MMP by the JC-1 probe demonstrated a dose-dependent induction of MMP dissipation (Fig. 5E). These results further confirmed the activation of the mitochondrial apoptosis pathway by sublethal doses of CBG in the cancer cells, following DNA damage-induced DDR signaling.
Cinobufagin induces regression of tumor xenografts in vivo
To assess the clinical potential of CBG, we evaluated the effects of treatment with CBG on tumor xenografts. Nude mice bearing SW1116 tumor xenografts were treated with either PBS (vehicle control) or CBG (dosed once daily by oral gavage, at 2, 5 or 10 mg/kg), for 20 days. The results showed that all three doses of CBG significantly inhibited the growth of the tumor, with the 10 mg/kg dose induced substantial regression of established tumors (Fig. 6A–C), demonstrating in vivo efficacy for CBG. No significant difference in body weight was observed between control and the drug-treated groups (Fig. 6D), and microscopic examination of hematoxylin–eosin stained tissue sections of liver, heart, and kidney showed normal histological morphology and structure for all groups, suggesting that the drug was well tolerated.