Metformin Synergizes with BCL-XL/BCL-2 Inhibitor ABT-263 to Induce Apoptosis Specifically in p53-Defective Cancer Cells

p53, an important tumor suppressor, inhibits cancers by inducing cell-cycle arrest, cellular senescence, apoptosis, and DNA damage repair in response to a variety of cellular stresses. TP53 is the most frequently inactivated tumor suppressor gene in cancers due to deletion or mutation. Mutated TP53 is disabled either by a dominant-negative (DN) mechanism or by loss of heterozygosity (LOH; ref. 1). Even in cancer types where TP53 mutations are rare, p53 function can be sufficiently abrogated through aberrant degradation (such as in cervical cancer; ref. 2). Loss of wild-type p53 function by TP53 deletion (null), LOH, DN, or p53 abolishment can be collectively called p53 deficiency (p53-defective). Patients with p53 deficiency commonly show increased chemoresistance to a broad range of anticancer therapeutics and poor prognosis. Although experimental reactivation of p53 can lead to profound tumor regression (2, 3), very few effective targeted strategies have been developed for p53-defective cancers.

Evading apoptosis is a crucial hallmark of cancer. It is well-known that the B-cell lymphoma 2 (BCL-2) family proteins and the inhibitor of apoptosis family proteins (IAPs) play pivotal roles in regulating apoptosis. The proapoptotic and prosurvival BCL-2 family members interact at the mitochondrion to control the release of apoptogenic proteins, whereas the IAPs function as endogenous inhibitors of cleaved caspase-3 and cleaved caspase-7 (4, 5). ABT-263 (orally active analogue of ABT-737), a small-molecular inhibitor of the antiapoptotic BCL-2 family proteins BCL-XL/BCL-2, is undergoing active clinical trials and has shown promising therapeutic potential in various tumor types (5). However, ABT-263/ABT-737 can induce MCL-1, another member of antiapoptotic BCL-2 family, which can attenuate their anticancer effects (5). Importantly, ABT-737 has only minor impact on tumor development in p53-defective mice, and p53 dysfunction alleviates the ABT-737–induced apoptosis (6, 7).

It has been reported that p53 can restrain mTORC1, whereas p53 deficiency can enhance mTORC1 activity (8). mTORC1, an evolutionarily conserved sensor, integrates the signals from nutrients, growth factors, and energy to promote several fundamental biologic processes such as protein and lipid synthesis. Aberrant activation of mTORC1 is closely related with cancer proliferation and survival (9). However, the effect of mTOR inhibitor rapamycin is limited by the feedback derepression of AKT pathway. Although the new-generation mTOR inhibitor PP242 represses mTORC1 more potently than rapamycin and prevents the feedback phosphorylation of AKT, it induces overactivation of ERK (10). Metformin, a widely used antidiabetic drug, has shown remarkable anticancer function partially through suppressing mTORC1 (11). Importantly, metformin abrogates mTORC1 activation without overstimulating AKT phosphorylation and even prevents ERK activation in pancreatic cancer cells (10). Interestingly, the role of p53 in metformin-mediated anticancer effect is controversial. For example, the combination of metformin and 2-deoxyglucose triggers p53-dependent apoptosis in prostate cancer cells (12), whereas another study shows that metformin is toxic to p53-defective cancer cells (13).

It is intriguing and significant to explore whether mTORC1 can determine the incapacity of ABT-263 and whether metformin can synergize with ABT-263 in p53-defective cancer cells. Here, we for the first time demonstrated that metformin and ABT-263 synergistically induced apoptosis in various p53-defective cancer cells. Unexpectedly, this synergistic effect was abolished in p53 wild-type (p53-WT) cancer cells. Our study provides the rationale for developing novel therapeutic strategy selectively targeted p53-defective tumors by using the combination of metformin and ABT-263.

The human cell lines including Hep3B, PC3, HCT-116, NCI-H1299, SW480, SW620, SK-BR-3, HeLa, and HEK-293 were from ATCC in 2013. The HCT-116 p53^−/− cell line was kindly gifted by Dr. Bert Vogelstein (Johns Hopkins University, Baltimore, MD) in 2013. The cell lines were originally authenticated in ATCC or Johns Hopkins Genetic Resources Core Facility (GRCF) Cell Center by short tandem repeat (STR) profiling and passaged less than 6 months in the lab. All the cell lines were cultured in DMEM (Gibco) except for PC3 (grown in 1:1 DMEM/F12), supplemented with 10% FBS (Gibco), penicillin (100 U/mL), and streptomycin (100 μg/mL) at 37°C in a 5% CO₂ humid incubator. The p53 status of each cell line was determined according to the IARC TP53 database. The missense TP53 mutations carried by SW480, SW620, and SK-BR-3 cell lines are c.818G>A, c.818G>A, and c.524G>A, respectively.

ABT-263, AZD8055, PP242, 4EGI-1, MK-2206, selumetinib, and BAY-11-7082 were from Selleck Chemical. Metformin and Z-VAD-FMK were from Beyotime Biotechnology. The structures for ABT-263, AZD8055, PP242, 4EGI-1, MK-2206, selumetinib, BAY-11-7082, and Z-VAD-FMK were separately described in Supplementary References 1–8. Rapamycin was from Cell Signaling Technology (CST). Cycloheximide (CHX) was from Cayman Chemical. DMSO was from Sigma. Neutralizing antibodies against IL1β, IL6, IL8, and CXCL1 were from R&D Systems, and the control IgG was from Beyotime Biotechnology.

Cell viability assay was performed with Cell Counting Kit-8 (CCK-8; Dojindo laboratories) as described (14) and detailed in Supplementary Materials and Methods. To assess the combination effect of metformin and ABT-263, the Bliss Index was calculated according to the observed and expected effects as described (15).

Flow cytometry for apoptosis or cell cycle was performed as described (14, 16). The detailed procedures are in Supplementary Materials and Methods.

Western blotting and co-immunoprecipitation (co-IP) were performed as described (14). The detailed procedures and used antibodies are in Supplementary Materials and Methods.

The cap-binding assay was performed as described (17). The detailed procedures are in Supplementary Materials and Methods.

The siRNAs for MCL-1, XIAP, survivin, BCL-2, BCL-XL, and the control siRNA (siNC) were from GenePharma. The siRNAs for cIAP1 and cIAP2 were from Lab-cell Corporation. The expression plasmids pcDNA3.1 and pcDNA3.1-survivin were from Obio Technology. pCMV-MCL-1 and pEBB-XIAP were from Addgene. After being released from the above two plasmids, the cDNAs of MCL-1 and XIAP were separately subcloned into pcDNA3.1. The detailed procedures of transfection and the sequences of siRNAs are in Supplementary Materials and Methods.

The detailed procedures and sequences of primers for quantitative real-time PCR (qRT-PCR) are in Supplementary Materials and Methods.

The senescence-associated β-galactosidase (SA-β-gal) activity was analyzed to indicate cellular senescence using a SA-β-gal staining kit (CST) according to the manufacturer's instructions.

The animal experiments were performed with male BALB/c nude mice (6- to 8-week-old) from Beijing Vital River Experimental Animals Co. Ltd. Briefly, 5 × 10⁶ HCT-116 p53^+/+ or HCT-116 p53^−/− cells in 150 μL PBS were subcutaneously injected into the axillae of the nude mice. When the average tumor size was about 100 mm³, the mice were randomized into four groups (5 mice/group) and given the treatments including vehicle control, metformin (150 mg/kg/d intraperitoneally), ABT-263 (100 mg/kg/d gavaging), and the metformin/ABT-263 combination. Metformin was dissolved in normal saline (NS), and ABT-263 was dissolved in a mixture vehicle composed of 60% Phosal 50 PG, 30% polyethylene glycol (PEG) 400, and 10% EtOH. The xenograft tumor sizes were recorded once every two days, and the volume was calculated using the following formula: volume = width² × length × 1/2. After two weeks of the treatments, the xenograft tumors were harvested, and the proteins of tumor tissues were extracted for Western blot analyses. All animal experiments were carried out according to the protocol approved by Third Military Medical University Guidelines for Use and Care of Animals.

All data were expressed as mean ± SD unless otherwise stated. Comparisons between two groups were determined using the 2-tailed unpaired Student test (t test). P < 0.05 was considered as statistically significant.

To investigate whether metformin and ABT-263 can synergize to suppress the growth of p53-defective cancer cells, the p53-null hepatocellular carcinoma (HCC) cell line Hep3B and prostate cancer cell line PC3 were treated with various concentrations of metformin, ABT-263, or their combination. As shown in Fig. 1A, cotreatment with metformin and ABT-263 inhibited cell growth more potently than each alone. Bliss Index analysis showed that there was an obvious synergistic effect between metformin and ABT-263 (Fig. 1B). The combination of 10 mmol/L metformin and 5 μmol/L ABT-263 dramatically decreased the cell viability (Fig. 1C) and the corresponding synergistic effect was potent (Bliss Index = 3.28 for Hep3B and 2.33 for PC3 cells), so we chose this combination in the subsequent experiments. Flow cytometry, Western blotting, and CCK-8 assays revealed that the synergism between metformin and ABT-263 was due to the enhancement of apoptosis rather than the induction of cell-cycle arrest (Fig. 1D–F and Supplementary Fig. S1A). Moreover, this synergetic apoptosis induction effect was also verified in p53-defective colorectal cancer cell lines (HCT-116 p53^−/−, SW480, and SW620), non–small cell lung cancer cell line (NCI-H1299), breast cancer cell line (SK-BR-3), and cervical cancer cell line (HeLa; Fig. 1G and H). These data indicated that the synergism between metformin and ABT-263 could be extensively applicable in multiple p53-defective cancer cells. Importantly, metformin or ABT-263 alone or their combination had no significant suppressing effect on the viability of HEK-293 cells (Supplementary Fig. S2), suggesting that this therapeutic paradigm was safe.

To investigate the mechanism by which metformin and ABT-263 synergistically induce apoptosis, the changes of the BCL-2 and IAPs family members were assayed. As shown in Fig. 2A and Supplementary Fig. S1B, ABT-263 significantly elevated the levels of antiapoptotic proteins MCL-1, XIAP, and survivin (but not BCL-2, BCL-XL, cIAP1, and cIAP2), which were dramatically attenuated by metformin. Silencing MCL-1, XIAP, or survivin (but not BCL-2, BCL-XL, cIAP1, or cIAP2) enhanced the killing effect of ABT-263 (Fig. 2B and C and Supplementary Fig. S1C, S3A, and S3B). Furthermore, overexpressing MCL-1, XIAP, or survivin dramatically alleviated the synergistic proapoptotic effect of metformin and ABT-263 (Fig. 2D and E and Supplementary Fig. S3C). These results indicated that metformin sensitized ABT-263 through attenuating ABT-263–induced MCL-1, XIAP, and survivin.

To explore whether metformin/ABT-263 can influence the interactions among BCL-2 family members and the interactions between IAPs and caspases, co-IP assays were performed. As shown in Fig. 2F and G and Supplementary Fig. S4A and S4B, ABT-263 markedly decreased the complex formation between BCL-XL and the crucial apoptosis activator BIM but remarkably increased the complex formation between BIM and MCL-1. Moreover, MCL-1 could also form a complex with BAK (apoptosis effector) or PUMA (apoptosis sensitizer). In addition, ABT-263 modestly increased cleaved caspase-3 and cleaved caspase-7, which were confined by the simultaneously elevated XIAP and survivin (Fig. 2H and Supplementary Fig. S4C and S4D). Furthermore, metformin-unlocked BIM from BIM/MCL-1 complex was subsequently restrained by BCL-XL (Fig. 2G and Supplementary Fig. S4A and S4B). The combination of metformin and ABT-263 not only significantly attenuated the restriction of antiapoptotic BCL-2 family members (MCL-1/BCL-XL) to proapoptotic members (BIM/BAK/PUMA) but also dramatically alleviated the constraint of IAPs (XIAP/survivin) to cleaved caspase-3/cleaved caspase-7, thus overcoming the respective shortcomings of ABT-263 and metformin and leading to more substantial apoptosis (Fig. 2F–H and Supplementary Fig. S4A–S4D).

As shown in Supplementary Fig. S5A, the mRNA level of MCL-1, XIAP, or survivin was not attenuated upon the metformin/ABT-263 cotreatment compared with ABT-263 alone. Moreover, the relative protein reduction of MCL-1, XIAP, or survivin was similar among the groups of control, metformin alone, ABT-263 alone, and their combination in the presence of translational inhibitor CHX (Supplementary Fig. S5B). These results indicated that metformin alleviated ABT-263–induced MCL-1, XIAP, or survivin through neither transcriptional nor posttranslational regulation.

Cap-dependent translation initiation is the rate-limiting procedure of eukaryotic translation. Assembly of the eukaryotic translation initiation factor 4F complex (EIF4F, including EIF4E, EIF4G, and EIF4A) is foremost for cap-dependent translation initiation, and the key mRNA 5′ cap-binding subunit EIF4E strongly promotes cancer progression and chemoresistance (18). mTORC1 dominantly enhances cap-dependent translation through phosphorylating its downstream EIF4E-binding protein 1 (4EBP1) and p70S6K, and the phosphorylated-4EBP1 (p-4EBP1) fails to sequester EIF4E from eIF4G, whereas phosphorylated-p70S6K (p-p70S6K) protects EIF4A from being restrained by PDCD4 (9). Previous reports have shown that the translation of MCL-1 and survivin is regulated in the mTORC1-mediated cap-dependent manner (19, 20). In this study, we found that ABT-263 augmented the phosphorylation of 4EBP1 and p70S6K and increased the complex formation between mTOR and p-4EBP1, which was markedly attenuated by metformin (Fig. 3A and B and Supplementary Fig. S4E and S6), indicating that metformin dramatically alleviated ABT-263–induced mTORC1 activity. Furthermore, cotreatment with metformin and ABT-263 significantly increased the binding of the translational suppressor nonphosphorylated-4EBP1 (non-p-4EBP1) to the EIF4E-mRNA 5′-cap complex compared with ABT-263 alone (Fig. 3C and Supplementary Fig. S4F), indicating that the cotreatment remarkably inhibited cap-dependent translation.

In addition to cap-dependent manner, the internal ribosome entry site (IRES)-dependent mechanism provides a cap-independent way for translation of specific mRNAs (21). Previous study has shown that the translation of XIAP is uniquely IRES-dependent and this process relies on p-EIF4E (22). Here, we found that metformin significantly attenuated ABT-263-elevated p-EIF4E (Fig. 3A and Supplementary Fig. S6), indicating that the downregulation of XIAP was due to suppression of IRES-dependent translation. Moreover, it has been reported that both cap-dependent and IRES-dependent translation require EIF2α, and the phosphorylation of EIF2α inhibits translation initiation (21). As shown in Supplementary Fig. S6, the phosphorylation of EIF2α and its upstream PKR-like ER kinase (PERK) was not elevated upon cotreatment with metformin and ABT-263, indicating that the PERK/EIF2α pathway was not involved in translation inhibition. Taken together, metformin attenuated ABT-263–induced MCL-1, XIAP, and survivin via suppressing cap-dependent translation of MCL-1 and survivin and repressing IRES-dependent translation of XIAP.

Although p-4EBP1 and p-p70S6K are important regulators of cap-dependent translation, they can be functionally distinct (9, 17). Compared with rapamycin, PP242 can more efficiently abrogate 4EBP1 phosphorylation and suppress the cap-dependent translation (17). As shown in Fig. 3D and E, combination of metformin or PP242 with ABT-263 potently decreased p-4EBP1, MCL-1, and survivin compared with ABT-263 alone, leading to remarkable apoptosis and growth inhibition. Cotreatment with rapamycin and ABT-263 only significantly reduced p-p70S6K compared with ABT-263 alone, which caused no augmentation of apoptosis and growth inhibition (Fig. 3D and E). Furthermore, suppression of the EIF4E-EIF4G association by 4EGI-1, which functionally mimics p-4EBP1 inhibition and represses cap-dependent translation, markedly sensitized ABT-263 to inhibit the cells growth (Fig. 3F). In short, inhibition of p-4EBP1 rather than p-p70S6K plays the central role in the synergetic effect of metformin and ABT-263.

To investigate whether p53 restoration further enhances the synergistic proapoptotic effect of metformin and ABT-263, the isogenic colorectal cancer cell pairs HCT-116 p53-WT (p53^+/+) and HCT-116 p53-null (p53^−/−) were used. As shown in Fig. 4A–C and Supplementary Fig. S7A, the synergistic effect of metformin and ABT-263 on growth inhibition and apoptosis induction presented in HCT-116 p53^−/− cells but not HCT-116 p53^+/+ cells. (The average Bliss Index for the different combinations of metformin and ABT-263 was 0.97 in HCT-116 p53^+/+ cells and 1.59 in HCT-116 p53^−/− cells. Specifically, the Bliss Index for the combination of 10 mmol/L metformin and 5 μmol/L ABT-263 was 0.95 in HCT-116 p53^+/+ cells and 1.95 in HCT-116 p53^−/− cells). Moreover, compared with ABT-263 alone, cotreatment with metformin and ABT-263 dramatically decreased p-mTOR, p-4EBP1, p-p70S6K, the mTORC1 upstream kinases (p-AKT, p-ERK), as well as MCL-1 and survivin in HCT-116 p53^−/− cells (but not HCT-116 p53^+/+ cells) (Fig. 4D). Interestingly, the relative changes of p-EIF4E and XIAP in each treatment were similar in the two cell lines (Fig. 4D), indicating that XIAP was not responsible for their different responses to the metformin/ABT-263 combination. In addition, cotreatment with metformin and ABT-263 significantly reduced the interaction between mTOR and p-4EBP1 and notably increased the EIF4E-associated non-p-4EBP1 compared with ABT-263 alone in HCT-116 p53^−/− cells, indicating that the cotreatment presented a robust suppression of mTORC1-mediated cap-dependent translation. Nevertheless, the cotreatment only modestly inhibited cap-dependent translation in HCT-116 p53^+/+ cells (Fig. 4E and F and Supplementary Fig. S4G and S4H).

To uncover whether the above oncogenic kinases conferred resistance to the metformin/ABT-263 combination in HCT-116 p53^+/+ cells, the corresponding kinase inhibitors were used. As shown in Fig. 4G and H, pretreatment with the active site mTOR inhibitor AZD8055 or PP242, AKT inhibitor MK-2206, or MEK/ERK inhibitor selumetinib (but not the allosteric mTOR inhibitor rapamycin) dramatically sensitized the metformin/ABT-263 combination to induce apoptosis and growth inhibition via decreasing p-4EBP1, MCL-1, and survivin. Taken together, the p53-mediated signaling pathway mTORC1–4EBP1–MCL-1/survivin abolishes the synergistic proapoptotic effect between metformin and ABT-263.

To further explore the mechanism by which the metformin/ABT-263 cotreatment does not synergistically induce apoptosis in p53-WT cancer cells, we tried to seek a p53-mediated process which concurrently met the following two requirements. First, the p53-mediated process has the potential to impede the therapeutic effects of anticancer drugs. Second, p53 is able to mediate the activation of oncogenic kinases. These peculiarities make us focus on cellular senescence, which is a stable growth arrest where p53 plays a central role. Generally, cellular senescence suppresses malignancies, but it hampers killing efficiency of chemotherapy in certain circumstances (23, 24). More importantly, senescent cells secrete a set of proinflammatory factors collectively termed senescence-associated secretory phenotype (SASP) which can be either pro- or antitumorigenic (25). In the present study, we found that the metformin/ABT-263 cotreatment induced an obvious cellular senescence in HCT-116 p53^+/+ cells, whereas the cotreatment only slightly evoked the cellular senescence in HCT-116 p53^−/− cells (Fig. 5A). In addition, SASP components such as interleukins (IL1β and IL6), chemokines (IL8 and CXCL1), growth factor (FGFβ), protease (MMP3), interleukin/chemokine receptor (CXCR2), or other inflammatory factors [granulocyte macrophage-colony stimulating factor (GM-CSF) and TNFα] were analyzed. As shown in Fig. 5B and Supplementary Fig. S8A, ABT-263 significantly increased most of them, which was augmented by metformin in HCT-116 p53^+/+ cells. However, these SASP components were unchanged or just mildly increased upon metformin/ABT-263 cotreatment in HCT-116 p53^−/− cells. Pretreatment with the neutralizing antibody against IL1β, IL6, IL8, or CXCL1 sensitized the metformin/ABT-263 combination to elicit apoptosis and suppress cell growth through inhibiting the signaling pathway AKT/ERK–4EBP1–MCL-1/survivin (Fig. 5C and D). These results indicated that the induction of these protumorigenic SASP components was, at least partially, responsible for the resistance to metformin/ABT-263 cotreatment in HCT-116 p53^+/+ cells. Next, we assessed whether NF-κB, the pivotal SASP activator, contributed to the p53-mediated oncogenic signals in response to the cotreatment. As shown in Fig. 5E and F, pretreatment with the NF-κB inhibitor BAY-11-7082 enhanced the cotreatment-induced apoptosis via repressing AKT/ERK–4EBP1–MCL-1/survivin pathway in HCT-116 p53^+/+ cells. In summary, p53-mediated oncogenic SASP components aberrantly activate the AKT/ERK–4EBP1–MCL-1/survivin pathway, which remarkably diminishes the synergistic killing effect of the metformin/ABT-263 combination.

As shown in Fig. 6A, metformin and ABT-263 sensitized each other in nude mice to repress the growth of xenograft tumors derived from HCT-116 p53^−/− (but not HCT-116 p53^+/+) cells. Western blot analyses revealed that the metformin/ABT-263 cotreatment dramatically suppressed 4EBP1–MCL-1/survivin pathway compared with ABT-263 alone in HCT-116 p53^−/− (but not HCT-116 p53^+/+) xenograft tumors. The cotreatment markedly strengthened cleaved PARP in HCT-116 p53^−/− xenograft tumors and only modestly enhanced cleaved PARP in HCT-116 p53^+/+ xenograft tumors (Fig. 6B and Supplementary Fig. S7B). These results indicated that metformin and ABT-263 collaboratively inhibited the growth of p53-defective but not p53-WT cancer cells in vivo.

Herein, we demonstrated that metformin and ABT-263 synergistically elicited apoptosis specifically in p53-defective (but not p53-WT) cancer cells in vitro and in vivo, and the corresponding working model is shown in Fig. 6C. Our findings provide scientific evidence for developing novel therapeutic strategy targeted p53-defective tumors by using the cotreatment with metformin and ABT-263.

The promising BCL-XL/BCL-2 inhibitors ABT-263/ABT-737 can suppress kinds of cancers via unlocking multiple proapoptotic machineries of the BCL-2 family. However, MCL-1 is emerging as a critical refractory determinant in ABT-263/ABT-737 therapy and repressing MCL-1 remarkably sensitizes the killing effect of them (5, 26, 27). In addition to the BCL-2 family, the IAPs also play critical roles in regulating apoptosis in a different hierarchy. But the previous reports about ABT-263 pay little attention on IAPs. In this study, we showed that ABT-263 could elevate XIAP and survivin and silencing each of them sensitized ABT-263 to induce apoptosis comparable to silencing MCL-1 in p53-defective cancer cells, indicating that the IAPs are novel mediators of ABT-263 resistance. Besides, overexpressing MCL-1, XIAP, or survivin partially attenuated the apoptosis elicited by the metformin/ABT-263 combination, indicating that metformin sensitizes ABT-263 through suppressing both MCL-1 and IAPs. Notably, inhibition of the antiapoptotic BCL-2 family members also plays important roles in sensitizing the IAP inhibitors (28). We and others emphasize the interactions between the two family members, and it is probable that a broad suppression of antiapoptotic members of both families can produce a more forceful antitumor effect (29). Overall, metformin or ABT-263 alone shows limited efficacy, but they create a favorable situation mutually where the cancer cells are primed for apoptosis by a second strike, and their combination results in a robust synergistic apoptosis (Fig. 6C). This combinatory strategy is in accord with the “synthetic lethal effect” that although oncogenic kinase inhibitors typically evoke cytostatic function, additional BCL-XL/BCL-2 inhibitors tip the balance to trigger prominent killing responses (30, 31). Interestingly, metformin synergized with ABT-263 to induce remarkable apoptosis in six types of p53-defective carcinoma cells in our study, but it is reported that ABT-263 does not sensitize metformin in p53-defective pediatric glioma cells (32), suggesting that the metformin/ABT-263 combination may not be universally effective in all p53-defective cancers. Indeed, it is reasonable that one therapeutic strategy is not likely to be applicable to all tumor cells and the histologic types of malignances do influence the therapeutic responses to targeted therapies (27, 30, 33).

Although EIF4E is required for cap-dependent translation of all mRNAs, it preferentially stimulates the translation of a subset of prosurvival mRNAs commonly with highly structured 5′-untranslated regions (UTR; ref. 18). It is reasonable that the ABT-263–induced MCL-1 was dramatically attenuated by metformin in our study because MCL-1 mRNA is particularly EIF4E-sensitive and MCL-1 protein highly relies on continuous translation (20). Suppressing MCL-1 translation plays a key role in restraining cancers or sensitizing chemotherapies (17, 26, 34). Survivin mRNA harbors a high GC content and stable secondary structure in its 5′-UTR (19), which makes it vulnerable to metformin-mediated cap-dependent translation inhibition in this study. As the only validated cap-independent translational manner, the IRES mechanism allows for the continued and enhanced translation of specific mRNAs, which delays cell death or confers resistance when cap-dependent translation is suppressed (21). XIAP mRNA is well-known for translating in an IRES-dependent manner due to its unique 5′-UTR and secondary structure (35). Herein, we found that metformin and ABT-263 synergistically attenuated cap-dependent translation, but the translation of XIAP was also strikingly decreased rather than induced. There may be two reasons for this discrepancy. First, not all apoptotic stresses enhance the IRES-dependent translation of XIAP and it prefers to happen in response to acute transient stress conditions (21), which is different from our treatments. Second, metformin significantly inhibits ABT-263–induced p-EIF4E that is crucial for IRES-dependent translation of XIAP.

It is well-known that both p-4EBP1 and p-p70S6K are downstream effectors of mTORC1. In the present study, we found that suppressing p-4EBP1 rather than p-p70S6K was the key nexus that mediated synergism of metformin and ABT-263 in p53-defective cancer cells. These findings were consistent with the previous reports that activating 4EBP1 or disturbing EIF4F assembly plays a central role in strengthening the collaborative targeted cancer therapeutics (15, 36). Notably, metformin sensitized ABT-263 more powerfully than PP242, although both of them vigorously inhibited the 4EBP1–MCL-1/survivin pathway (Fig. 3D and E). The results were in accord with that PP242 may lead to a feedback activation of the antiapoptotic kinase ERK (10). Recently, targeting protein translation has emerged as a novel therapeutic strategy to fight against cancers and overcome intratumor heterogeneity (37). Our study demonstrated that the combination of metformin and ABT-263 orchestrates the translational repression of specific antiapoptotic mRNAs by using cap-dependent and IRES-dependent mechanisms, showing a comprehensive impact to evoke apoptosis by suppressing translation in p53-defective cancer cells.

In this study, we revealed that the metformin/ABT-263 cotreatment dramatically triggered SASP in p53-WT cancer cells, but only slightly induced SASP in p53-defective cancer cells, which was consistent with that ABT-737 stimulates SASP in a p53-dependent manner (38). Interestingly, the cotreatment also upregulated p53 in p53-WT cancer cells (Supplementary Fig. S8B). Besides, previous study has shown that persistent DNA damage can also induce SASP (39). In our study, we showed that the crucial DNA damage protein p-ATM was markedly elevated by ABT-263 alone and the metformin/ABT-263 combination in p53-WT cancer cells, whereas p-ATM was only slightly increased by the metformin/ABT-263 cotreatment in p53-defective cancer cells (Supplementary Fig. S8C). These results suggested that DNA damage response may play an important role in regulating the different SASP levels in p53-WT and p53-defective cancer cells upon treatment with metformin and ABT-263. Interestingly, ABT-263 has been also reported to reduce activity of senescent cells (40). Moreover, the role of metformin on cellular senescence and SASP is also controversial (41, 42). In addition, the components and the overall downstream effects of SASP are considerably diverse (tumor-promoting or tumor-suppressive) depending on the treatment and cellular context (23). In our study, we uncovered that SASP components, such as IL1β, IL6, IL8, or CXCL1, desensitized p53-WT cancer cells to apoptosis upon metformin/ABT-263 cotreatment through potentiating the AKT/ERK–mTORC1–4EBP1 signaling. The results agreed with that these protumorigenic inflammatory factors can activate the oncogenic AKT/ERK–mTORC1 pathway in other tumor models (43–46). Of importance, the heightened mTORC1 function in turn augments the expression of SASP components (47). It could be speculated that once initiated, the aberrant AKT/ERK–mTORC1 signaling and protumorigenic SASP components conspire to make a positive feedback loop which significantly impairs the apoptosis elicited by the metformin/ABT-263 combination in p53-WT cancer cells. Notably, wild type p53 generally represses mTORC1 activation (8), which means the stimulatory effects of p53-meidated SASP overwhelm the inherent inhibitory effects of p53 on mTORC1. The net effects reinforce the paradoxical phenotype that p53 predominantly functions as a tumor suppressor, but in specific conditions, it can render cancer cells resistant to chemotherapy (48, 49).

The association of NF-κB with p53 and their relationships with SASP are complicated. For examples, Lujambio and colleagues report that NF-κB mediates p53-dependent SASP and they cooperate to induce SASP (50), whereas others reveal that p53 can inhibit NF-κB activity and inflammation (51). In our study, we found that NF-κB contributed to the poor response of p53-WT cancer cells to the metformin/ABT-263 combination via activating AKT/ERK–mTORC1–4EBP1 pathway. As an important hallmark of cancer, unwanted and continuous inflammation profoundly exacerbates cancer microenvironment and impedes therapeutic outcome of anticancer therapeutics (52, 53). Here, we also showed that p53-mediated SASP created a tumor-promoting milieu which activated oncogenic kinases and hindered the synthetic lethal interactions between metformin and ABT-263.

In conclusion, we clarified that the IAPs members XIAP and survivin were novel resistant factors that hindered proapoptotic activity of ABT-263 in p53-defective cancer cells. Targeting the cap-dependent translation of MCL-1 and survivin and IRES-dependent translation of XIAP by metformin and the BCL-XL/BCL-2 inhibitor ABT-263 may be an effective and novel therapeutic schedule to synergistically induce apoptosis selectively against p53-defective cancer cells. We expect that the large population of patients suffering from p53-defective cancers would benefit from the therapeutic approach based on the combination of metformin and ABT-263.

No potential conflicts of interest were disclosed.

Conception and design: X. Li, B. Li, F. He

Development of methodology: X. Li, B. Li, P. Zhou, J. He, H. Xiong, Y. Wu, X. Lyu, Y. Zhang, F. He

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Li, B. Li, B. Wang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Li, B. Li, Z. Ni, H. Xiong, F. Yang, F. He

Writing, review, and/or revision of the manuscript: X. Li, B. Li, Z. Ni, P. Zhou, H. Xiong, F. Yang, Y. Wu, X. Lyu, Y. Zeng, J. Lian, F. He

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Li, B. Li, Z. Ni, P. Zhou, J. He, F. Yang, Y. Zeng, J. Lian, F. He

Study supervision: F. He

This work was supported by the National Natural Science Foundation of China (No. 81472291 and No.31671464) to F. He.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.