RKO, MCF7 and HCT116 cell lines were cultured at 37°C in DMEM (Gibco 11875085, Fisher Scientific) supplemented with 10% fetal bovine serum (Hyclone, Logan, UT, USA) in a humidified incubator with 5% CO₂. Human osteosarcoma cell line SJSA1 and canine osteosarcoma cell lines Gracie and D17 were cultured at 37°C in RMPI 1640 (Dulbecco’s Modified Eagle’s medium, Invitrogen) supplemented with 10% fetal bovine serum (Hyclone, Logan, UT, USA) in a humidified incubator with 5% CO₂. Cell lines were used below passage 25 or within 2 months after thawing. Cells were tested negative for mycoplasma after thawing.

Generation of the RBM24 and RBM38 double knockout cell line was performed by knocking out RBM24 in RBM38-null cell lines as previously described (7). Briefly, RBM24 knockout cell lines were generated by CRISPR-cas9 genome editing method. sgRNAs targeting RBM24 were designed using the CRISPR design tool and cloned into the BbsI sites of CRISPR vector pSpCas9(BB)-2A-Puro. Two specific gRNAs were used: gRNA #1 GTA CAC CAA GAT CTT CGT CG and gRNA #2 CGA GGT CTT CGG CGA GAT CG. The cells were selected with puromycin and each individual clone was confirmed by western blot and sequencing analysis. Generation of HCT116 ΔC17/− was as previously described (9).

GST-RBM38 expression plasmid was generated as previously described (10). pTXB1-eIF4E plasmid was generated by amplifying eIF4E using His-eIF4E expression plasmid as template (7). The amplicon was then cloned into pTXB1 via NdeI and SapI. The primers used to amplify eIF4E were forward primer, 5´- GGT CAT ATG GCG ACT GTC GAA CCG GAA ACC −3´, and reverse primer, 5´- GGT TGC TCT TCC GCA AAC AAA CCT ATT TTT AGT GGT GGA G −3´.

Western blot procedures were as previously described (11). Briefly, cell lysates were resolved in an 8-12% SDS-polyacrylamide gel and then transferred to nitrocellulose membranes. Blots were blocked in PBST containing 3% milk for 1 hour at 20°C. Primary antibodies in PBST containing 3% milk were incubated at 4°C rocking overnight. The following morning, membranes were washed 3x with PBST followed by the addition of secondary antibody in PBST containing 3% milk at 20°C for 2 hours. Membranes were then washed 3x with PBST.

For pTXB1-eIF4E, protein expression and purification was as previously described (12). For GST-RBM38 competitive pull-down assays, pGEX-4T3-RBM38 plasmid was transformed into BL21 (DE3) competent E. coli. One-liter culture was grown at 37°C until OD600 equaled 0.6-0.8 and then was induced with a final concentration of 0.1 mM IPTG for 4 hours. Bacteria were pelleted and placed in -80°C overnight. Pellets were then lysed, sonicated, and centrifuged in 20 mL lysis buffer (50 mM Tris pH 7.5, 150 mM NaCl and 0.1% Triton X-100) supplemented with benzonase, 1 mM DTT and protease inhibitor cocktail. Lysates were then incubated with GST beads rocking at 4°C for 2 hours. Beads were washed 3x with lysis buffer. After brief centrifugation, lysates were carefully removed. Beads were then resuspended in lysis buffer with 0.1% Triton X-100 to make a 50% bead slurry. One-hundred microliter bead slurry was incubated in 830 µL lysis buffer, 10 µM (20 µL) peptide (Ctrl, Pep8-1, Pep7_S2K, Pep7, or Pep8_SK) and 250 µg (50 µL) purified eIF4E in a 1.5 mL tube. Samples were rocked overnight, washed 3x with lysis buffer and eluted with 60 µL of 1x SDS-loading buffer before western blot analysis. For eIF4E-eIF4G competitive pulldowns, two 10 cm plates were seeded (1x10^6 RKO cells) for each control and Pep7 treated groups. The next day the cells were lysed with 1 mL lysis buffer (50 mM Tris pH 7.5, 150 mM NaCl and 0.1% Triton X-100) supplemented with benzonase, 1 mM DTT and protease inhibitor cocktail. After brief sonication (5x, 5 seconds on, 15 seconds off), lysates were centrifuged, and then added to a fresh tube. Magnetic protein A/G beads (30 µL) were added to each tube followed by 1 µg IgG or eIF4E antibody. After 1 hour 25 µM control or Pep7 peptides were added to the corresponding group. After rotating overnight at 4°C, beads were washed 5x with lysis buffer before western blot analysis.

For 2-D cell viability assays, 15,000 cells per well were plated in a triplicate in a 96 well plate. Two hours later, 20 µM peptide was added to each well. Twenty-four hours later, cell viability was measured by CellTiter-Glo 3D according to manufacturer’s guidelines (Promega).

3-D mini ring culture assays were as described previously (13). Briefly, single cell suspensions (15K cells/well) were plated around the rim of the well in 96-well plates in a 4:3 mixture of Matrigel and Mammocult (BD Bioscience CB-40324). After plates were incubated at 37°C with 5% CO₂ for 15 min to solidify the gel, 100 µL of prewarmed Mammocult containing the indicated peptide was added to the corresponding well. Four hours after peptide addition, the irradiated sample group was irradiated with 15 Gy radiation. Twenty-four hours after irradiation (or no irradiation) the media was replaced with 100 µL of prewarmed Mammocult. Seventy-two hrs later, 100 µL pre-warmed PBS was used to wash the cells 3 times. Cells were then released from the Matrigel by incubating at 37°C for 45 min in 50 μL of 5 mg/mL dispase (Life Technologies #17105-041). Images were taken with a 40x objective and then cell viability was measured by CellTiter-Glo 3D according to manufacturer’s guidelines (Promega). The 3-D breast cancer spheroid cultures were as described previously (7). Briefly, MCF7 cells were cultured in MammoCult™ (Stemcell, Cambridge, MA) media per manufacturer’s guidelines. Twenty-thousand cells per well were seeded in 6-well ultra-low adherent plates. The following day the cells were treated with 5 μM Pen-Ctrl or Pen-Pep8. After 7 days, tumor spheres larger than 50 μm were counted.

Cells were seeded in 96-well plates (~7,000 cells per well) and were incubated at 37°C, 5% CO₂ for 24 h. After the incubation period, cells were exposed to a range of drug concentrations in triplicates, and were incubated in the same conditions. After 48 h exposure, cell viability was measured using CellTiter-Glo 3D according to manufacturer’s guidelines (Promega). We estimated LC50 values using a logistic regression (% cell death) model.

The docking program AutoDock Vina (14), within the SAMSON-Connect interface (https://www.samson-connect.net/) was used to model the docking of each peptide to eIF4E. First, Pep-FOLD3 was used to determine the estimated secondary structure of each peptide (15). Next, the experimental X-ray crystal structure of eIF4E (with water removed) was used for the peptide docking assays (PDB code: 1WKW). A grid box was drawn 5 Å around the previously identified binding site for Pep8 (7). Each peptide candidate had missing hydrogens added and where energy minimized. AutoDock Vina was set to an exhaustiveness of 8 and the top 5 modes were used to calculate the average affinity (kcal/mol) and Ki (umol).

Experimental values are presented as mean ± SEM. Statistical comparisons between experimental groups were analyzed by a two-tailed Student’s t-test. P values < 0.05 were considered statistically significant.

Previously, we identified Pep8, an 8-amino acid peptide derived from the carboxyl-terminus of RBM38 that enhances p53 translation by disrupting the RBM38-eIF4E complex leading to tumor growth suppression in vitro and in vivo (7). Further, we determined that Ser:6 of Pep8, which is necessary for its interaction with eIF4E, forms a hydrogen bond with Asp:202 of eIF4E (7). In an effort to enhance the affinity of Pep8 for eIF4E, we performed in-silico docking studies on rationally designed Pep8 derivatives. Based on replica exchange molecular dynamic simulations (REMDS) (7), multiple Pep8 derivatives were predicted to have either enhanced or decreased affinity towards eIF4E (Figure 1A). First, we determined the secondary structures of each peptide using PEP-FOLD3. PEP-FOLD3 is a novel computational framework, which allows for de novo free or biased prediction for linear peptides between 5 and 50 amino acids (15). Second, AutoDock Vina (14), an in silico docking program, was used to determine the affinity of each peptide for eIF4E. Figure 1A summarizes the peptide sequence, binding affinity, and Ki of each peptide tested. Since Ser:6 in Pep8 forms a hydrogen bond with Asp:202 in eIF4E, we hypothesized that substituting Ser:6 with a positive charge (Ser to Lys or Arg, Pep8_S6K and Pep8_S6R) would enhance its affinity towards eIF4E, while substituting to a negative charge (Ser to Asp, Pep8_S6D) should decrease its affinity. In silico docking predicted that Pep8_S6R and Pep8_S6K have higher affinities for eIF4E, whereas, Pep8_S6D was predicted to have a lower affinity for eIF4E than Pep8 (Figure 1A). Furthermore, because prolines are known to modulate or define the structure of a peptide (16), we asked whether both prolines are necessary for Pep8 function. Docking analysis revealed that both Pep8_P2A and Pep8_P7A have less affinity towards eIF4E than Pep8 (Figure 1A), demonstrating that these proline residues may be necessary for Pep8 structure and/or function. Moreover, docking analysis showed that Pep8_Y3W has little affinity towards eIF4E, suggesting that Tyr:3 in Pep8 may also be important for peptide structure and/or binding.

Our in silico docking assays demonstrated two differential docking patterns for Pep8 and Pep8 S6K (Figures 1B, C). Importantly, Lys:6 in Pep8 S6K formed a tighter hydrogen bond with Asp:202 in eIF4E than did Ser:6 in Pep8 (3.0 Å vs 3.7 Å). To confirm our docking results, western blot analysis was performed to measure induction of wild-type p53 in RKO colon cancer cells treated with these peptide derivatives. First, we tested whether Pep8_S6K would enhance p53 induction, whereas Pep8_S6D would be inert. In order to facilitate intracellular delivery, the cell-penetrating peptide Penetratin was fused to these peptides at their N-termini (17). As shown in Figure 1D, Pep8_S6K was more potent than Pep8 to induce p53 expression, whereas Pep8_S6D was inactive. In addition, both prolines in Pep8 were necessary for its function as highlighted by lack of p53 induction in cells treated with either Pep8_P2A or Pep8_P7A (Figure 1E).

Our initial Pep8 derivative docking studies revealed that the first tyrosine in Pep8 was not necessary for Pep8 to bind with eIF4E, therefore, we generated five rationally designed 7-amino acid peptide derivatives (Figure 2A). Our in silico docking results revealed that all Pep7 derivatives, except for Pep7_S5C, had higher affinities for eIF4E than Pep8, whereas Pep7 was shown to have the highest affinity of all the peptides tested (Figure 2A). Further, the docking analysis revealed a differential docking pattern for Pep7 when compared to Pep8 (Figures 2B, C). We postulate that while only 3 out of 8 residues in Pep8 (Tyr:3, Ser:6, and Ala:8) form hydrogen bonds with eIF4E, 6 out of 7 residues in Pep7 (Pro:1, Ser:2, Ala:3, Ala:4. Ser:5, and Pro:6) form hydrogen bonds with eIF4E, which likely contribute to its elevated affinity towards eIF4E.

As Pep7 derivatives were shown to have the highest affinities towards eIF4E (Figure 2A), we questioned whether Pep7 and its derivates were able to modulate the interaction between RBM38 and eIF4E. To that end, we performed competitive pull-down assays with purified GST-tagged RBM38 and purified eIF4E in the presence of control or Pep8-derived peptides. We found that these Pep8 derivatives were able to impede the interaction between RBM38 and eIF4E, with Pep7 exhibiting the strongest activity (Figure 2D).

To confirm the docking results, western blot analysis was performed to measure p53 induction in RKO colon cancer cells treated with these peptide derivatives. We found that these Pep8 derivatives induce p53 expression equal to or greater than Pep8, with Pep7 being the strongest, consistent with our in silico docking results (Figures 2E, F). To determine if Pep7 is more efficacious than Pep8, we calculated the LC50 for both Pep8 and Pep7 in RKO cells. As shown in Figure 2G, Pep7 was more than twice as potent (LC50: 8.1 µM) as Pep8 (LC50: 17.5 µM), further confirming our aforementioned results.

To define the specificity of Pep7 to induce p53 by disruption of the RBM24/RBM38-eIF4E complex, p53 induction was measured in an isogenic control and RBM24/RBM38 double knockout RKO cells treated with Pen-Ctrl, Pen-Pep7, Pen-Pep8, or Pen-Pep8-1. We showed similar to Pep8 and Pep8-1, Pep7 was able to induce p53 expression in wild-type but not double knockout cells (Figures 2H, I). To determine whether Pep7 induces p53 expression through binding to eIF4E, the effect of these Pep8 derivatives on p53 induction was measured in isogenic control and eIF4E-deficient HCT116 colon cancer cells (ΔC17/−). Previously, we showed that the ΔC17/− HCT116 cell line has only one truncated eIF4E allele, in which the last 17 residues, including Asp:202, are deleted, abrogating the ability of RBM38 to interact with eIF4E (9). Similar to Pep8 and Pep8-1, Pep7 was able to induce p53 expression in isogenic control, but not eIF4E-deficient HCT116 cells (Supplementary Figures 1A, B). Likewise, Pep7 was able to decrease tumor cell viability in wild-type but not in ΔC17/− HCT116 cells (Supplementary Figures 1C). To rule out the possibility that Pep7 inhibits eIF4G ability to interact with eIF4E, we performed a competitive pulldown assay with the addition of Pep7. We showed that Pep7 was unable to impede the interaction between eIF4E and eIF4G (Supplementary Figure 1D). In addition, we modeled the binding of Pep7 to eIF4E in the presence of eIF4G or 4E-BP1 (the inhibitor of eIF4G), and determined based on its binding interface, it is unlikely that Pep7 is able to modulate the interaction of eIF4G or 4E-BP1 with eIF4E (Supplementary Figures 1E, F). Collectively, these data indicate that Pep7 is a specific inhibitor of the RBM38-eIF4E complex and is the most potent inducer of p53 expression through eIF4E- and RBM24/RBM38-dependent manners.

Constraining a peptide via cyclization is a frequently used strategy for maintaining the secondary structure of the peptide, and for enhancing conformational stability (18). Therefore, we generated three different cyclic Pep8 peptides using two methods of cyclization. First, disulfide cyclization was utilized by flanking Pep8 with two cystine residues (cYPYAASPAc, referred to as cyclic Pep8). Second, amide bond cyclization was performed by flanking Pep8 with a lysine and glutamic acid (kYPYAASPAe), or by a single glutamic acid on its C-terminus (YPYAASPAe) (Figure 3A). To delineate whether these peptides were able to interact with eIF4E, both linear Pep8 and cyclic peptides were conjugated to TentaGel resins for pulldown assays. We showed that all four peptides interacted with eIF4E, with cyclic Pep8 exhibiting the strongest interaction (Figure 3B). Next, we asked whether cyclic Pep8 is able to induce p53 expression. To facilitate intracellular delivery of cyclic Pep8, we used Pep-1 CPP which is a short amphipathic peptide carrier which releases its cargo after intracellular delivery (19). As shown in Figure 3C, cyclic Pep8 was able to induce p53 expression in a dose-dependent manner (50 nM, 150 nM, and 375 nM). To determine if cyclic-Pep8 is more potent than linear Pep8, we performed a cell viability assay in MCF7 cells treated with Pep-1 delivered control, linear Pep8 and or cyclic Pep8 peptides. We found that cyclic Pep8 was more potent than linear Pep8 to suppress tumor cell growth (Figure 3D). Further, we demonstrated that cyclic Pep8 significantly decreased the formation of MCF7 3D tumor spheres (Figures 3E, F). Together, these results demonstrate that disulfide cyclized Pep8 is a potent inducer of p53 and suppressor of 3D tumor sphere formation.

Even though Penetratin-conjugated control peptide was used as a control, it remains possible that the Penetratin CPP may cooperate with Pep8 to induce p53 expression. To rule out this possibility, Pep8 and its derivatives were delivered into cells by Pep-1 CPP. We showed that p53 expression was induced by Pep8 and Pep8_S6R when delivered by Pep-1, but not Pep8_Y3W and Pep8_S6D (Figures 4A, B), suggesting that Pep8 and its derivatives are responsible for inducing p53 expression.

To improve peptide tumor targeting and intracellular delivery, we sought to determine whether iRGD-conjugated Pep7 (Pep7-iRGD) can be efficiently delivered inside of tumor cells (Figure 4C). In addition, we coupled Pep7 to iRGD using a valine citrulline (Val-Cit) linker which is cleavable by cathepsin B, which is highly expressed in cancer cells to aid in tumor targeting (20). Next, we measured p53 induction in an isogenic control and RBM24/RBM38 double knockout RKO cells treated with Pen-Pep7 or with Pep7-iRGD. We found that Pep7-iRGD was more competent than Pen-Pep7 to induce p53 expression in isogenic control cells (Figure 4D), whereas both Pep7-iRGD and Pen-Pep7 were inert in RBM24/RBM38-null cells (Figure 4E). We would like to mention that the fold change of p53 induction by iRGD-Pep7 (2.7 fold) was higher than that by cyclic Pep8 (from 1.9 to 2.5 fold) (Figure 3C), indicating that iRGD-conjugated Pep7 is the most potent inducer of p53 and allows for a targeted delivery approach. Similarly, Pep7-iRGD was more competent than Pen-Pep7 to inhibit cell survival in isogenic control cells, but not in RBM24/RBM38-null cells (Figures 4F, G). Together, these data support that iRGD is the optimal CPP to deliver Pep7.

While p53 is frequently inactivated in various human and canine tumors, including osteosarcomas (OSA), ~ 60% of OSA contain wild-type p53, thus restoring wild-type p53 function may be an attractive strategy for the treatment of OSA (21). We found that both human and canine RBM38 proteins share ~95% sequence identity, including the region where Pep8 is derived (Figure 5A). Further, human and canine eIF4E proteins share ~99% sequence identity (Figure 5A). Thus, we hypothesized that Pep7 would be able to induce p53 expression in canine cancer cells. Indeed, p53 expression was induced by Pep7 in human SJSA1 and canine Gracie osteosarcoma cells (Figures 5B, C), both of which harbor wild-type p53. Similarly, we found that cell viability was markedly decreased by Pep7 in both cell lines (Figures 5D, E). We would like to mention that iRGD-conjugated Pep7 was more potent than Penetratin-conjugated Pep7 to induce p53 expression and growth suppression (Figures 5B–E). In addition, Pep8 significantly decreased the cell viability of 3D tumor spheroids from another canine osteosarcoma cell line D17, which expresses wild-type p53 (Supplementary Figures 2A, B). These data support our hypotheses that the RBM38-eIF4E pathway is conserved between human and canine and may be of therapeutic relevance.

Resistance to chemo-/radio-therapy is a major challenge for long-term survival of cancer patients (22). As induction of wild-type p53 is considered to re-sensitize tumors to chemo-/radio-therapy (23), we postulated that Pep7 may be further explored as a potential cancer therapeutic agent. Previously, we showed that induction of wild-type p53 by Pep8 can sensitize multiple types of tumor cells to chemotherapeutic agents, such as doxorubicin and etoposide (7). Thus, we determined whether induction of p53 by Pep7 can be used to sensitize radio-resistant human SJSA1 and canine Gracie osteosarcoma cells. To test this, SJSA1 and Gracie cells were treated with Penetratin- or iRGD-conjugated Pep7 for 4 hours followed by 15-Gy radiation for 2 hours. As expected, p53 expression was induced in SJSA1 and Gracie cells by both Penetratin- and iRGD-conjugated Pep7 as compared to their control counterparts (Figures 6A, B, compare lanes 2 and 4 with 1 and 3, respectively). In contrast, p53 expression was not or only slightly induced 2 hours after 15-Gy radiation (Figures 6A, B, compare lane 5 and 7 with 1 and 3 respectively). To confirm that the level of radiation used was sufficient to induce p53 expression, radio-sensitive RKO cells were treated similarly as for SJSA1 and Gracie cells. As expected, p53 expression was induced in RKO cells by Penetratin- and iRGD-conjugated Pep7 as compared to their control counterparts (Supplementary Figure 3A, compare lanes 2 and 4 with 1 and 3, respectively). Additionally, we found that p53 expression was highly induced in RKO cells upon 15-Gy radiation (Supplementary Figure 3A, compare lanes 5 and 7 with 1 and 3, respectively). These observations suggest that the lack of induction of wild-type p53 in SJSA1 and Gracie cells may play a role in their inherent radio-resistance (24, 25).

To determine whether the dose of radiation was high enough to induce DNA damage, we measured γH2AX, a key indicator of DNA damage (26). We showed that 15-Gy radiation led to a marked increase in the level of γH2AX in SJSA1 and Gracie cells as well as in RKO cells (Figures 6A, B and Supplementary Figure 3A, compare lane 5 and 7 with 1 and 3, respectively), suggesting that the DNA damage response pathway is functional in SJSA1 and Gracie cells. We also showed that the level of γH2AX was increased in SJSA1, Gracie, and RKO cells treated with Penetratin- and iRGD-conjugated Pep7 as compared to their control counterparts (Figures 6A, B and Supplementary Figure 3A, compare lanes 6 and 8 with lanes 5 and 7, respectively). These observations are consistent with our previous studies that DNA damage response pathway is activated by Pep8 via induction of wild-type p53 (7, 13).

Next, 3D tumor spheroids were established as a model to determine whether Pep7 can sensitize radio-resistant human SJSA1 and canine Gracie osteosarcoma cells to radiation. Radio-sensitive RKO 3D tumor spheroids were established as a control. We showed that Penetratin- or iRGD-conjugated Pep7 was able to suppress tumor spheroid growth in all three cell lines, whereas 15-Gy radiation was only able to suppress radio-sensitive RKO tumor-spheroid growth (Figures 6C–F and Supplementary Figures 3B, C). However, combined treatment of Pep7-iRGD with radiation made both SJSA1 and Gracie cells highly susceptible to growth suppression and decreased viability (Figures 6C–F). Additionally, iRGD-conjugated Pep7 was able to further enhance the sensitivity of RKO cells to radiation (Supplementary Figures 3B, C). Collectively, these data support that iRGD-conjugated Pep7 may be used as an adjuvant to sensitize radio-resistant cancer cells to radiation.