Survivin Expression Is Differentially Regulated by a Selective Cross-talk between RBM38 and miRNAs let-7b or miR-203a

RNA-binding motif protein 38 (RBM38) and RBM24 constitute a family of RNA-binding proteins with a single RNA-binding motif (1). Functionally, RBM38 binds to mRNA targets, modulating mRNA processing, stability, and/or translation, thereby regulating an array of genes responsible for various cellular functions, including cellular growth and differentiation (2, 3). Importantly, amplified RBM38 expression has been correlated with the progression of colorectal adenoma to carcinoma and poor prognosis in multiple cancers, including breast cancer and lymphoma (2, 4–6). We previously identified that RBM38 suppresses p53 mRNA translation by interacting with eukaryotic translation initiation factor 4E (eIF4E) on p53 mRNA (2). In addition, an eight-amino acid peptide derived from RBM38 (Pep8) increases p53 translation by disrupting the RBM38–eIF4E complex, leading to p53-dependent growth suppression in vitro and in vivo (7). Recently, we showed that RBM38 works in concert with argonaute-2 (AGO2) and miR-203a to regulate p63 expression (8). Moreover, RBM38 can also regulate the accessibility of miRNAs on CDKN1A, LATS2, and SIRT1 transcripts (9). These data suggest that RBM38 alone, or together with various miRNAs, can modulate an array of targets, thereby influencing numerous biological functions.

miRNAs have gained enormous interest given their vast number and robust influence on mRNA stability and translation (10). Aberrant regulation of miRNA expression and function has been linked to many diseases, including diabetes and cancer (11). Compounding the importance of miRNAs, it is estimated that these small noncoding RNAs could influence more than 60% of all protein coding genes (12, 13). A key component of miRNA-guided gene silencing is AGO2, which is also necessary for miRNA maturation. AGO2 is the only member of the AGO family that has a catalytic activity and plays an essential role within the RNA-induced silencing complex (RISC; refs. 14, 15). In a ribonucleoprotein complex, miRNAs bind to target mRNA 3′-untranslated regions (3′-UTR) and then guide the RISC via AGO2 to degrade target mRNAs (16, 17).

Survivin is an essential modulator of cell death as well as a stress responder (18). Of clinical interest, survivin is the fourth most frequently overexpressed transcript in the human cancer transcriptome, and aberrant expression has been associated with chemo-/radioresistance and poor prognosis (19, 20). As RBM38 and survivin are both overexpressed in tumors and promote tumor progression (19–22), we sought to address whether survivin is regulated by RBM38, especially considering that survivin mRNA stability is known to be regulated by multiple miRNAs. Herein, we reported that RBM38 regulates survivin mRNA stability via two distinct mechanisms. RBM38 increases survivin expression by binding to a poly-U element in survivin 3′-UTR hindering the accessibility of let-7b. In contrast, RBM38 decreases survivin expression by interacting with AGO2 and facilitating miR-203a–mediated degradation of survivin mRNA. Due to the abundance of let-7b over miR-203a in the cells tested in this study, RBM38 ultimately enhances survivin expression. We demonstrated that Ser-195 in RBM38 interacts with Glu-73 and/or Glu-76 in AGO2. Furthermore, Pep8 is a potent inducer of survivin by blocking RBM38–AGO2 interaction, preventing miR-203a–mediated degradation of survivin mRNA. Last, we found that combinatorial administration of Pep8 and survivin inhibitor YM155 induces DNA damage in an RBM38-dependent manner, leading to enhanced growth suppression and decreased cell viability. Thus, concomitant administration of Pep8 and YM155 may be an effective therapeutic approach for cancers.

Cell lines HCT116 (ATCC; Cat# CCL-247, RRID:CVCL_0291) and MCF7 (ATCC; Cat# HTB-22, RRID:CVCL_0031) were obtained from the ATCC between 2007 and 2016 and used below passage 25 or within 2 months after thawing. Because cell lines from the ATCC have been thoroughly tested and authenticated, these cell lines were not reauthenticated or tested for Mycoplasma. Cells were cultured in DMEM (Invitrogen) supplemented with 10% FBS (Hyclone) and 100 U/mL penicillin/streptomycin in a humidified incubator with 5% CO₂ at 37°C. Generation of knockout and inducible cell lines is detailed in the Supplementary Materials.

Micro-RNA transfections were performed using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instructions. Detailed methods and primers are listed in the Supplementary Materials and Supplementary Table S1.

HCT116-Tet-On-RBM38 cells were seeded (2 × 10⁵ cells per well) on day 1. The next day, one well was transfected with 5 μg pCDNA3-EGFP-3′-UTR and the other with 5 μg pCDNA3-EGFP-3′-UTR-ΔpolyU. The following morning, each well was split equally into two new wells in a 6-well plate. That evening, one well of each duplicate was treated with 0.5 μg/mL tetracycline to induce RBM38 expression. Eighteen hours later, cells were lysed and subjected to Western blot analysis. All Western blot figures are representative data of at least two independent replicates.

Biotin-based pulldown assay was adapted from Dash and colleagues (23). Detailed methods are listed in the Supplementary Materials.

Total RNA was isolated with Quick-RNA Miniprep Kit (Zymo Research). cDNA was synthesized using a RevertAid First Strand cDNA Synthesis kit (Thermo Fisher Scientific). Detailed methods are listed in the Supplementary Materials and primers in Supplementary Table S1.

Peptides were synthesized by GenScript. Penetratin cell-penetrating peptide was used to facilitate intracellular delivery (24). Peptides used in this study are listed in Supplementary Table S2.

Western blot procedures were as previously described (25). Detailed methods for immunofluorescence and immunoprecipitation (IP)–Western blot analysis are listed in the Supplementary Materials.

For RBM38 RNA-IP assays, RBM38-inducuble MCF7 cells were seeded at 2 × 10⁶ in four 10 cm plates. The next day, two plates were induced to express RBM38 by addition of 100 ng/mL tetracycline for 36 hours followed by RNA chromatin IP (RNA-ChIP) assay as described previously (7). For AGO2 RNA-ChIP with Pep8 treatment, MCF7 cells were treated with 10 μmol/L Pen-Control or Pen-Pep8 for 18 hours. Cell extracts were prepared with IP buffer (10 mmol/L HEPES, pH 7.0, 100 mmol/L KCl, 5 mmol/L MgCl₂, 0.5% Nonidet P-40, and 1 mmol/L DTT) and then incubated with 1 μg of anti-RBM38, anti-AGO2, or an isotype control IgG overnight at 4°C. The RNA–protein immunocomplexes were brought down by magnetic protein A/G beads. qRT-PCR for RBM38 RNA-IP analysis and RT-PCR for AGO2 RNA-IP analysis were carried out to determine the levels of Survivin and HPRT1 transcripts.

Three-dimensional (3D) “mini ring” spheroid culture assay was adapted from Phan and colleagues (26). Detailed methods are listed in the Supplementary Materials.

All data were analyzed using Prism 8 (GraphPad). Experimental values were presented as mean ± SEM. Statistical comparisons between experimental groups were analyzed by a two-tailed Student t test. P values < 0.05 were considered statistically significant.

The authors declare that all data supporting the findings of this study are available within the article and its Supplementary information files. Detailed Materials and Methods are given in the Supplementary Materials. Research reagents will be provided upon a written request.

Utilizing the Xena Browser and The Cancer Genome Atlas Pan-Cancer database (27), we found that the expression level of survivin correlates with that of RBM38 in various cancers (Supplementary Fig. S1A). RBM38 and survivin are also known to be overexpressed in tumors and promote tumor progression (2, 4–6). In addition, survivin 3′-UTR contains a nonconical-binding site for let-7b as well as an adjacent RBM38-binding poly-U element (Fig. 1A; ref. 28). As RBM38 is capable of inhibiting miRNA binding to targets that contain a U-rich element (9), we examined whether survivin expression is regulated by RBM38. We previously reported that RBM38 negatively regulates p53 translation (2), and additionally, survivin and let-7b have been shown to be transcriptionally repressed by p53 (29, 30). Therefore, we examined the effect of RBM38 on survivin expression in both p53-competent and p53-null HCT116 and MCF7 cells. We showed that let-7b inhibited survivin expression in a dose-dependent fashion (Fig. 1B and C), whereas RBM38 enhanced survivin mRNA in both HCT116 and MCF7 cells independent of p53 (Fig. 1D–E; Supplementary Fig. S1B and S1C). Conversely, survivin expression was decreased in RBM38-null cells (Fig. 1F; Supplementary Fig. S1D). Moreover, RNA-ChIP assay showed that the level of survivin mRNA was significantly higher in anti-RBM38 immunoprecipitates from RBM38-expressing cells than that from control cells (Supplementary Fig. S1E). Further, survivin mRNA half-life was decreased by loss of RBM38 by almost half in RBM38-null MCF7 cells (Supplementary Fig. S1F). These data suggest that RBM38 is necessary to maintain survivin mRNA stability.

To determine if RBM38 binding to survivin mRNA is necessary to regulate survivin expression, we generated two GFP reporter constructs in which GFP was coupled to either the full-length survivin 3′-UTR or the one with the poly-U element removed (Fig. 1G). RBM38 increased the level of GFP protein with full-length 3′-UTR but had no effect on the level of GFP from the reporter carrying the poly-U–deleted 3′-UTR (Fig. 1H). As an internal control, the level of endogenous survivin was still increased by RBM38 expression (Fig. 1H). Because the binding sites for RBM38 and let-7b are adjacent (Fig. 1A), we postulated that RBM38 suppresses let-7b from binding to survivin mRNA. To test this, we examined the level of survivin mRNA and protein in p53-null HCT116 cells transfected with RBM38 alone or together with let-7b. Expectedly, let-7b reduced, whereas RBM38 increased, survivin expression (Fig. 1I; Supplementary Fig. S1G), whereas ectopic expression of RBM38 restored the level of survivin mRNA and protein suppressed by let-7b (Fig. 1I; Supplementary Fig. S1G). Although let-7b had similar effects to regulate survivin in RBM38-null cells as compared with isogenic control cells (Fig. 1J), we showed that RBM38 suppressed let-7b degradation of survivin mRNA in a dose-dependent manner (Supplementary Fig. S1H). Furthermore, we measured the effect of a let-7b inhibitor and showed that the let-7b inhibitor was able to increase the level of survivin mRNA, albeit only weakly in RBM38-competent cells (Fig. 1K, compare lane 1 with 2), suggesting that the steady-state level of RBM38 is sufficient to counter endogenous let-7b. In contrast, the let-7b inhibitor rescued survivin expression in RBM38-null cells (Fig. 1K, compare lane 3 with 4). To validate that RBM38 suppresses the binding of let-7b to survivin mRNA, RNA pulldown assay was performed with biotin-tagged locked nucleic acid (LNA) of let-7b mimic. We showed that let-7b ability to interact with survivin mRNA was suppressed by RBM38 in both MCF7 and HCT116 cells (Fig. 1L). Together, these data indicate that RBM38 interacts with the poly-U element in survivin mRNA, thus positively regulating survivin expression by hindering let-7b–mediated mRNA degradation.

miR-203a was found to regulate survivin via binding to a region distal to the poly-U element in survivin 3′-UTR (Fig. 2A; ref. 31). Because RBM38 decreases p63 mRNA stability via miR-203a by directly interacting with AGO2 (8), we postulated that RBM38 may also regulate survivin via miR-203a. First, we confirmed that miR-203a was able to decrease survivin protein expression in WT and p53-null HCT116 cells (Fig. 2B and C). We verified that the levels of survivin mRNA and protein were decreased by miR-203a but increased by RBM38 in p53-null HCT116 cells (Fig. 2D; Supplementary Fig. S2A). However, in the presence of miR-203a, RBM38 was not able to rescue survivin mRNA (Fig. 2D; Supplementary Fig. S2A), suggesting that RBM38 does not suppress miR-203a from targeting survivin mRNA as it does for let-7b. We also showed that RBM38 enhanced miR-203a–mediated degradation of survivin mRNA in a dose-dependent manner (Supplementary Fig. S2B). Further, miR-203a had limited ability to suppress survivin expression in RBM38-null cells as compared with isogenic control cells (Fig. 2E). Conversely, we found that miR-203a inhibitor was not able to enhance survivin expression in RBM38-null cells (Fig. 2F).

The above studies suggest that RBM38 has two opposing effects on survivin expression via let-7b and miR-203a. Because survivin expression is increased by ectopic expression of RBM38 but decreased by knockout of RBM38, we postulated that the relative abundance of these two miRNAs might be responsible for the dominant positive effect of RBM38 on survivin expression. We preformed quantitative RT-PCR to measure the relative levels of miR-203a and let-7b. Indeed, we found that the levels of let-7b were several folds higher than that of miR-203a in both HCT116 and MCF7 cells (Fig. 2G). Together, our data indicate that RBM38 regulates survivin expression via two opposing pathways: (1) the RBM38–AGO2–miR-203a axis decreases the stability of survivin mRNA by promoting the formation of RISC complex; (2) RBM38 enhances survivin mRNA stability by impinging let-7b accessibility.

Previously, we found that RBM38 and AGO2 can be detected by co-IP (8). However, it remains uncertain whether RBM38 and AGO2 interact with each other directly. To test this, recombinant full-length AGO2 protein and several AGO2 fragments were produced and used for IP–Western blot analysis (Fig. 3A; Supplementary Fig. S3A). We showed that RBM38 interacted with the full-length AGO2 and the N-terminal fragment (aa 1–227), but little if any with C-terminal fragments (aa 446–580 and aa 581–859; Fig. 3B). Of note, the AGO2 fragment containing aa 228–445 failed to express, as such was unable to be tested. In addition, we generated two small N-terminal fragments (aa 70–227 and aa 140–227; Fig. 3A). We found that RBM38 was able to interact with fragment aa 70–227 but not aa 140–227 (Fig. 3C), indicating that the region between aa 70 and 140 in AGO2 is necessary for its interaction with RBM38.

We showed previously that Ser-195 phosphorylation inhibits RBM38 ability to interact with eIF4E or AGO2 (8, 32). We also found that Pep8, derived from aa 190–197 in RBM38, including Ser-195, is capable of blocking the interaction between RBM38 and eIF4E (7). Thus, we extrapolated that Pep8 may also block RBM38 from binding to AGO2. To test this, IP–Western blot analysis was performed and showed that the interaction between RBM38 and AGO2 was inhibited by Pep8, supporting the idea that RBM38 interacts with eIF4E or AGO2 via the same domain (Fig. 3D).

To define the critical contact residue(s) in AGO2 for its interaction with RBM38, extensive molecular dynamic simulations were performed to determine the binding site on AGO2 for Pep8, and by extension for RBM38. In the apo state of AGO2, a gating movement of the PAZ domain was observed with the RNA-binding grove completely opened after 4 microsecond simulation (Supplementary Fig. S3B and S3C). The overall root-mean-square deviation (RMSD) of the whole protein reached to 12 Å, mainly due to domain motions (Supplementary Fig. S3D). The distance between the PAZ and MID domains was measured by the Cα distance of Lys-263 in the PAZ domain and Asp-823 in the MID domain. Lys-263 and Asp-823 formed a salt bridge in the closed conformation but were separated by more than 50 Å in the open conformation (Supplementary Fig. S3E). There was a slight variation in the N-terminal domain structure in AGO2 between crystal structures 6n4o and 4f3t (33, 34). In the 6n4o structure, a hydrogen bond was formed between Glu-76 and Tyr-393 (Supplementary Fig. S3B), whereas in the 4f3t structure, Glu-76 and Tyr-393 had no contact, indicating that there is flexibility in the N-terminal domain. In addition, the N-terminal domain still had a relatively large RMSD (3–6 Å) as compared with a rigid protein domain (Supplementary Fig. S3D). The distance between Glu-76 and Tyr-393 varied from a close contact with hydrogen bond formation to a separation of 20 Å.

To screen for possible docking sites for Pep8 around the N-terminal region, the protein docking program MEGADOCK was used (Fig. 3E; ref. 35). We found that a hydrogen bond between Ser-6 in Pep8 and Glu-76 in AGO2 was the predominant binding mode (Fig. 3F, left). We also observed two alternative binding modes with slight variations represented by Ser-6 positioned between Glu-73 and Glu-76 (Fig. 3F, middle and right plots), suggesting that these negatively charged residues in AGO2 may be necessary for the interaction with Pep8.

To experimentally validate the model, we generated three N-terminal AGO2 fragments (aa 1–227) in which one or both negatively charged glutamic acid residues were substituted with a glutamine (AGO2 227-E73Q, AGO2 227-E76Q, and AGO2 227-E73Q/E76Q). IP–Western blot analysis revealed that AGO2 227-E73Q and AGO2 227-E76Q were still capable of interacting with RBM38, albeit at a lower affinity (Fig. 3G). In contrast, AGO2 227-E73Q/E76Q was unable to interact with RBM38 (Fig. 3G). In addition, we generated two Pep8 variants wherein Ser-6 was mutated to a negatively charged aspartic acid (Pep8SD) or a positively charged lysine (Pep8SK). Based on our modeling, mutating Ser-6 to a negatively charged aspartic acid would inhibit, whereas mutating to a positive charged lysine would enhance, the binding of Pep8 to the negatively charged pocket created between Glu-73 and Glu-76 in AGO2. A competitive GST pulldown assay was performed to measure the interaction of recombinant AGO2 with GST-tagged RBM38 in the presence of a control (a scrambled 11-mer peptide: STLWDTAELWQ), Pep8, Pep8SD, or Pep8SK peptide. Indeed, we found that both Pep8 and Pep8SK were able to inhibit the binding between RBM38 and AGO2, whereas Pep8SD had a much weaker effect (Fig. 3H). These data suggest that the negatively charged pocket created between Glu-73 and Glu-76 is likely the binding interface for RBM38.

As an essential component of the RISC complex, AGO2 is recruited by miRNA to serve as a guide to load the RISC complex to target mRNA (16, 17). Therefore, RNA-ChIP assay was performed and showed that Pep8 markedly inhibited AGO2 to interact with survivin mRNA (Fig. 4A). Consistently, we found that Pep8 alone was able to increase survivin expression (Fig. 4B), and further, Pep8 cooperated with RBM38 to markedly increase survivin mRNA (Fig. 4B). As p53 is known to decrease let-7b but increase miR-203 expression (30, 36), we asked if increased expression of p53 by Pep8 (7) plays a role in survivin expression by Pep8. We found that survivin expression was increased by Pep8 in isogenic control and p53-null cells, but not in RBM38-null cells (Supplementary Fig. S4), suggesting that the effect of Pep8 on survivin is RBM38-dependent but p53-independent.

Previously, we showed that AGO2 interacts strongly with RBM38-S195A, but weakly with RBM38-S195D (8). Thus, we postulated that if the disruption of the RBM38–AGO2 complex is necessary for Pep8 to regulate survivin expression, Pep8 would not have any effect on survivin in RBM38-S195D–expressing cells. Thus, we measured the effect of Pep8 on survivin in p53-null HCT116 cells uninduced or induced to express RBM38-S195A or RBM38-S195D. We found that survivin was increased similarly by RBM38-S195A and RBM38-S195D (Fig. 4C and D, compare lane 1 with 2). These results are consistent with the idea that RBM38-S195A and RBM38-S195D, both of which are competent RNA-binding proteins (32), can bind to the poly-U element and prevent let-7b from degrading survivin mRNA. However, we found that survivin expression was further increased by Pep8 in RBM38-S195A–expressing cells but little if any in RBM38-S195D–expressing cells (Fig. 4C and D, compare lane 3 with 4).

Next, we examined whether Pep8 increases survivin expression by preventing miR-203a or let-7b from degrading survivin mRNA. To test this, the effect of Pep8 on survivin was measured in p53-null HCT116 cells cotransfected with either miR-203a or let-7b. Expectedly, survivin mRNA was markedly reduced by miR-203a but increased by Pep8 (Fig. 4E). We showed that Pep8 was able to restore the level of survivin mRNA decreased by miR-203a (Fig. 4E, compare lane 2 with 4), suggesting that Pep8 is highly potent to prevent both exogenous and endogenous miR-203a from targeting survivin mRNA for degradation. Because Pep8 binds to AGO2 but not RBM38, Pep8 should not influence the ability of RBM38 to bind to the poly-U–rich element in survivin mRNA. Indeed, we found that Pep8 had little if any effect to reverse let-7b–mediated degradation of survivin mRNA (Fig. 4F, compare lane 2 with 4). Collectively, our data indicate that Pep8 blocks the interaction between RBM38 and AGO2, thereby enhancing survivin mRNA stability by inhibiting miR-203a activity.

Previously, we demonstrated that Pep8 is a potent growth inhibitor of tumor xenografts in part by enhancing p53 expression (7). As survivin expression has been correlated with chemo-/radioresistance and poor prognosis (20), increased expression of survivin by Pep8 could be detrimental therapeutically. Thus, we hypothesized that by abrogating Pep8-mediated induction of survivin, we may further sensitize tumor cells to Pep8 treatment. Accordingly, we asked if YM155, a potent inhibitor of survivin transcription (37), could abrogate Pep8-mediated enhancement of survivin expression. To that end, WT, RBM38-null, and p53-null MCF7 and HCT116 cells were treated with Pep8, YM155, or in combination. Expectedly, YM155 inhibited survivin expression in all the cells tested (Fig. 5A–C; Supplementary Fig. S5A–S5C). Similarly, Pep8 enhanced survivin expression in both isogenic control and p53-null cells (Fig. 5A and C; Supplementary Fig. S5A and S5C), whereas Pep8 alone was able to enhance p53 expression in isogenic control cells (Fig. 5A; Supplementary Fig. S5A), consistent with our previous study (7). Importantly, YM155 was able to abrogate Pep8-induced survivin expression in both isogenic control and p53-null cells (Fig. 5A and C; Supplementary Fig. S5A and S5C, compare lane 2 with 4). Moreover, we found that Pep8 cooperated with YM155 to further enhance p53 induction (Fig. 5A; Supplementary Fig. S5A). In contrast, Pep8 had no effect on survivin or p53 expression in RBM38-null cells (Fig. 5B; Supplementary Fig. S5B).

We next questioned if concomitant treatment with Pep8 and YM155 could suppress cell growth. To address this, MCF7 3D tumor spheroid cultures were used, which have been shown to more accurately mimic the main features of human solid tumors, thus more precisely recapitulating drug response (38). Pep8 treatment alone resulted in substantial growth suppression and decreased cell viability in isogenic control cells (Fig. 5D and E) but had minimal, if any, effect in cells deficient in RBM38 or p53 (Fig. 5F–I), consistent with our early observations (7). Further, a low dose of YM155 (3 nmol/L) had only a modest effect on MCF7 cells (Fig. 5D and E), which is consistent with a previous report that GI₅₀ of YM155 for MCF7 cells is approximately 29 nmol/L (39). Nevertheless, combined treatment with Pep8 and YM155 significantly decreased tumor spheroid growth and cell viability in isogenic MCF7 cells as compared with Pep8 or YM155 treatment alone (Fig. 5D and E). Intriguingly, although RBM38-null MCF7 cells were not sensitive to combined treatment, p53-null MCF7 cells remained sensitive, albeit to a lesser degree (Fig. 5F–I).

To confirm that Pep8 sensitizes tumor cells to decreased levels of survivin, we measured the levels of cleaved caspase-7 and PARP as well as cell viability in MCF7 cells transfected with survivin siRNA followed by Pep8 treatment. The extent of caspase-7 and PARP cleavage is known to correlate well with the extent of apoptotic response (40). We showed that Pep8 cooperated with knockdown of survivin to induce apoptosis in both WT and p53-null MCF7 cells (Supplementary Fig. S5D and S5E). We also showed that Pep8 cooperated with knockdown of survivin to decrease cell viability in MCF7 spheroids (Supplementary Fig. S5F and S5G).

Increasing reports indicate that YM155 not only inhibits survivin expression but also elicits other cytotoxic effects, including DNA damage (41–43). As Pep8 cooperates with doxorubicin, a topoisomerase II (top-II) inhibitor, to induce DNA damage and growth suppression (7), we speculated that Pep8 may enhance YM155-induced DNA damage and apoptotic cell death. To that end, DNA damage was determined by measuring γH2AX and apoptosis by measuring the levels of cleaved caspase-7 and cleaved PARP in isogenic control, RBM38-null, and p53-null MCF7 cells treated with Pep8 and YM155 alone or in combination. γH2AX was slightly increased in cells treated with Pep8 or YM155 alone (Fig. 6A–C, compare lane 1 with 2–3, respectively). However, concomitant treatment resulted in robust γH2AX accumulation in isogenic control, but not in RBM38-null MCF7 cells (Fig. 6A and B). Further, concomitant treatment resulted in robust γH2AX accumulation in p53-null cells (Fig. 6C), which is consistent with a previous study that YM155 induces DNA damage independent of p53 (43). In addition, concomitant treatment with Pep8 and YM155 resulted in an enhanced apoptotic response in isogenic control and p53-null cells, but not in RBM38-null cells (Fig. 6A–C). Similar results were observed with HCT116 cells treated with Pep8 and YM155 alone or in combination (Supplementary Fig. S6A–S6C).

Although YM155-mediated γH2AX expression can be detected by Western blotting, we questioned whether the assay is sensitive enough to determine subtle or localized increases in γH2AX as previously reported (44). To circumvent this, the number of γH2AX foci was measured by immunofluorescence staining. Indeed, we showed that YM155, but not Pep8, significantly enhanced the number of γH2AX foci in isogenic control and RBM38-null MCF7 cells (Fig. 6D–G). Nevertheless, we showed that the number of γH2AX foci was further increased by concomitant treatment with Pep8 and YM155 in isogenic control but not RBM38-null MCF7 cells (Fig. 6D–G), consistent with the observation that Pep8 cooperates with YM155 to enhance γH2AX and growth suppression in an RBM38-dependent manner. Because the mechanism by which YM155 induces DNA damage remains uncertain (41), we asked if Pep8 enhances etoposide-induced DNA damage, which has a well-defined mode of action as an inhibitor of top-II (45). Concomitant treatment with Pep8 and etoposide resulted in both an increase in the level of γH2AX protein and in the number of γH2AX foci in isogenic control cells (Supplementary Fig. S7A, S7D, and S7E), but to a much lesser extent in RBM38-null MCF7 cells (Supplementary Fig. S7B, S7F, and S7G). Finally, we showed that the level of γH2AX was slightly increased in p53-null MCF7 cells treated etoposide alone and together with Pep8 (Supplementary Fig. S7C), highlighting the necessity of p53 for etoposide-induced, but not YM155-induced, DNA damage.

In this study, we demonstrated that RBM38 not only increases, but also decreases, survivin expression via two distinct pathways (Fig. 7). Mechanistically, RBM38 impedes let-7b binding to survivin 3′-UTR (Fig. 1L) and consequently, suppresses the ability of let-7b to decrease survivin expression (Fig. 1I; Supplementary Fig. S1H). In contrast, RBM38 directly interacts with AGO2 and promotes miR-203a–mediated degradation of survivin mRNA. However, RBM38 ultimately enhances survivin expression by limiting let-7b accessibility, which is expressed at a 7-fold higher level than miR-203a in HCT116 and MCF7 cells (Fig. 2G). Thus, the relative abundance of these miRNAs can dictate the ability of RBM38 to regulate survivin expression: increasing via high let-7b or decreasing via high miR-203a. Let-7b is typically viewed as a potent tumor suppressor because its expression is frequently decreased in cancers and serves as a prognostic marker (46). Interestingly, miR-203a expression is also decreased in a multitude of cancers (47), which would increase survivin. Therefore, in a tumor with decreased expression of let-7b and miR-203a, survivin would be further elevated by overexpression of RBM38, which warrants further investigation. Although the let-7 family encodes 9 mature miRNA, let-7b is the only member known to regulate survivin (48). However, it remains possible that other family members can regulate survivin and/or be regulated by RBM38, which necessitates further investigation.

Pep8 was found to be docked in a pocket in the N-terminal domain of AGO2, where the key contact residues are negatively charged Glu-73 and/or Glu-76 (Fig. 3). Pep8 binds to AGO2 and disrupts the AGO2–RBM38 complex as it does to the eIF4E-RBM38 complex (7). As a result, Pep8 increases survivin expression by preventing miR-203a from degrading survivin mRNA (Fig. 4). Giving the high degree of flexibility of AGO2 N-terminal domain (Supplementary Fig. S3) and the regulation by Pep8, we question whether the AGO2 protein structure and function can be modulated through its interaction with RBM38. Going further, it would be fruitful to determine if the RBM38–AGO2 axis regulates other miRNAs that bind to survivin 3′-UTR.

Pep8 has previously been shown to induce p53 expression and enhance doxorubicin-mediated growth suppression (7). As increased survivin plays a role in chemo-/radioresistance (19, 20), we tested whether Pep8-mediated enhancement of survivin can be thwarted by YM155, a well-defined survivin inhibitor (37). We found that YM155 abrogates induction of survivin by Pep8, enhances apoptosis, and cooperates with Pep8 to decrease the viability of spheroid cultures (Figs. 5 and 6; Supplementary Figs. S5 and S6). Although Pep8 is known to induce growth suppression in RBM38-/p53-dependent manners (7), we showed that p53-null cells remained highly sensitive to the combined treatment with Pep8 and YM155 (Figs. 5 and 6; Supplementary Figs. S5 and S6). These data are consistent with our earlier observations that a combined treatment of Pep8 with doxorubicin is potent to induce growth suppression in p53-null cells (7). Because doxorubicin and YM155 can induce DNA damage, we questioned whether Pep8 sensitizes tumor cells to YM155-induced DNA damage. Indeed, Pep8 enhanced YM155-induced DNA damage in RBM38-dependent but p53-independent manners (Fig. 6; Supplementary Fig. S6). As combined treatment of Pep8 with etoposide was not able to enhance DNA damage in RBM38-null or p53-null cells, our data support a previous study that YM155 induces DNA damage independent of p53 (43). The necessity for p53 in etoposide-induced DNA damage but not for that of YM155 may be due to their different modes of action. For example, etoposide inhibits DNA Top-II blocking DNA religation leading to critical mistakes in DNA synthesis (45). In contrast, YM115 is shown to induce DNA damage via oxidative DNA cleavage upon 2-electron reductive activation (49). Nonetheless, further investigation is warranted to determine the mechanism(s) by which Pep8 enhances YM155-mediated DNA damage. As YM155 exhibit limited antitumor efficacy when used alone or in combination with other chemotherapeutic agents in phase I to II clinical trials (50), Pep8 may be explored to sensitize tumors to YM155 treatment. Collectively, this study provides additional insight into the possibility that Pep8 may be explored as an adjuvant to sensitize tumors to DNA-damaging cancer therapeutic agents.

No disclosures were reported.

The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.

C.A. Lucchesi: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft. J. Zhang: Project administration, writing–review and editing. B. Ma: Data curation, software, writing–review and editing. R. Nussinov: Data curation, software, writing–review and editing. X. Chen: Resources, supervision, funding acquisition, project administration, writing–review and editing.

The authors would like to thank Chrisoula Toupadakis Skouritakis for creating the graphic working model. All simulations reported in this work were performed using the high-performance computational facilities at the NIH Biowulf supercomputer. This project was funded in whole or in part with federal funds from the NCI, NIH, under contract HHSN261200800001E. This research was also supported (in part) by the Intramural Research Program of NIH, NCI, Center for Cancer Research (RO1 CA250338, RO1 CA195828, and T32 CA108459).

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.