A Cell-Penetrating Peptide Targeting AAC-11 Specifically Induces Cancer Cells Death

AAC-11 (antiapoptosis clone 11), also called Api5 or FIF, is an antiapoptotic protein whose expression prevents apoptosis following growth factor deprivation (1). Many cancer cells exhibit elevated levels of AAC-11, which were found to correlate with poor prognosis in patients with non–small cell lung cancer and cervical cancer and contribute to tumor invasion and metastases (2–10). Interestingly, AAC-11 appears to be involved in anticancer drugs-induced apoptosis of tumor cells. Indeed, we have showed that AAC-11 gene silencing remarkably decreased chemoresistance, whereas its expression interfered with drug-induced apoptosis (11). Increased expression of AAC-11 was also found to protect from UV-induced apoptosis whilst its depletion is cancer cells lethal under condition of low-serum stress (10, 12). Although the precise mechanisms by which it suppresses apoptosis remain unclear, AAC-11 is now known to interfere with both E2F1-mediated apoptosis and Acinus-dependent apoptotic DNA fragmentation (11, 12). Finally, AAC-11 has been recently demonstrated to confer tumor immune resistance to antigene-specific T cells (13). These observations make AAC-11 a significant player in cancer cell progression and survival.

AAC-11 interacts with several apoptosis-related proteins and this complex-forming ability, probably favored by its elongated 3D structure (14), appears to be essential for AAC-11 to fulfill its antiapoptotic functions (1, 9, 11, 15, 16). Therefore, inhibitors of AAC-11 protein–protein interactions could prove to be of clinical benefit. We have developed a cell-penetrating peptide spanning the heptad leucine repeat region domain of AAC-11 (residues 363–399) fused at the N-terminus to the transmembrane-penetrating sequence penetratin. This peptide was able to disrupt the endogenous AAC-11–Acinus complex and drastically increase drug-induced cytotoxicity in a variety of cancer cells (11). Here, we characterized the anticancer properties of this AAC-11 targeting peptide, herein named RT53. We find that RT53 induced plasma membrane permeabilization of cancer cells while sparing nonmalignant cells. Interestingly, RT53 was able to inhibit tumor growth in vivo in human melanoma xenograft models, BRAF wild-type and V600E mutant, without systemic toxicity. Moreover, RT53-treated mouse melanoma cells mediated anticancer effects in a tumor vaccination model. These findings provide support for the use of AAC-11-inactivating peptides as a novel treatment strategy for melanoma and possibly other cancers.

Peptides were synthesized by Proteogenix (Strasbourg, France) and were >95% pure as determined by HPLC and mass spectrographic analysis.

A375, C8161, COLO 792, MEWO, Lu1205, and SK-Mel-28 cells were provided by Dr. N. Dumaz (INSERM U976) and genotyped to verify their authenticity. MRC-5, B16F10 and H1299 cells were provided by Drs. M. Dutreix (CNRS UMR3347, INSERM U1021) and R. Fåhraeus (INSERM U1162), and were purchased from ATCC. HaCat cells were provided by Prof. N. Basset-Seguin (AP-HP, Hôpital Saint-Louis, Paris, France) and their characteristics were described elsewhere (17). A549, MCF7, HeLa, THP-1, and KARPAS 299 were purchased from The European Collection of Cell Cultures. SU-DHL-5 cells were purchased from ATCC. Cells were cultivated either in DMEM or RPMI 1640 (Life Technologies), supplemented with 10% FBS and 1% penicillin/streptomycin. Apoptosis, immunoprecipitations, Western blot, and cytochemistry analysis were performed as previously described (18, 19).

All chemicals were purchased from Sigma. Antibodies were from Santa Cruz Biotechnology (mouse anti-Acinus, rabbit anti-AAC-11), Cell Signaling Technology (rabbit anti-Acinus, rabbit anti-Cox IV, rabbit anti-cytochrome c, mouse anti-Smac/Diablo, and rabbit anti-HMGB1), and Thermofisher (mouse anti-AIF).

Cells survival was assessed with the CellTiter 96 Aqueous One Solution Cell Proliferation Assay kit (Promega). For ΔΨ_m Assessment, cells were stained with 100 nmol/L DiOC6(3) and analyzed by flow cytometry. Release of lactate dehydrogenase (LDH) and ATP in the culture medium were assessed with the CytoTox 96 Non-Radioactive Cytotoxicity Assay and Enliten ATP Assay, respectively (Promega). Caspase-3/7 and caspase-9 activities measured using the Caspase-Glo 3/7 and Caspase-Glo 9 assay systems (Promega), respectively. Data are means ± SEM (n = 3). For HMGB1 release, cells supernatants were collected at the indicated times and concentrated by Centricon (Millipore) 10-kDa filter. Cells were harvested and lysed in Laemmli sample buffer. The lysates or equal volumes of the concentrated supernatants were then analyzed by Western blotting.

HeLa cells were resuspended in buffer A (250 mmol/L sucrose, 20 mmol/L HEPES, 10 mmol/L KCl, 1.5 mmol/L MgCl2, 0.5 mmol/L EGTA, and pH 7.5) and homogenized in a Dounce homogenizer. The homogenate was centrifuged twice 10 minutes at 750 × g. Supernatants were centrifuged 15 minutes at 10.000 × g and the resulting mitochondrial pellets resuspended in buffer A. The supernatants were further centrifuged at 100,000 × g for 1 hour and the resulting supernatants (designed as S100) frozen at −80°C.

HeLa cells mitochondria (10 μg) were incubated with RT53 or RT53M in a final volume of 25 μL of buffer A for 1 hour at 30°C. The samples were then centrifuged (10,000 × g, 10 minutes) to pellet the mitochondria. The resulting supernatant and pellets were fractionated by SDS-PAGE followed by immunoblotting.

HeLa cells mitochondria (10 μg) were incubated with 5 μmol/L of Rhodamine-labeled RT53 or RT53M for 1 hour at 30°C. Uptake was stopped by centrifugation (10,000 × g, 10 minutes), the mitochondrial pellet was washed twice in buffer A and resuspended in PBS containing 0.2% Triton X-100. The solution containing the mitochondria (100 μL) was transferred in a 96-well black plate and the fluorescence of rhodamine measured at 555 nm (excitation, 580 nm).

Lentiviral particles were produced in HEK293 with the helper plasmids pLvVSVg and pLvPack (Sigma) plus a lentiviral plasmid. shRNAs against AAC-11 originate from lentiviral plasmids MISSION pLKO.1-puro (Sigma; clone A: NM_006595.2-278s1c1, containing the sequence CCGGGCAGCTCAATTTATTCCGAAACTCGAGTTTCGGAATAAATTGAGCTGCTTTTTG and clone B: NM_006595.2-224s1c1, containing the sequence CCGGGCCTATCAAGTGATATTGGATCTCGAGATCCAATATCACTTGATAGGCTTTTTG). shSCR (SHC016-1EA, Sigma) contains the sequence CCGGCAACAAGATGAAGAGCACCAACTCGAGTTGGTGCTCTTCATCTTGTTGTTTTTG. Lentiviral infection of target cells was done as previously described (19).

Secondary structure predictions were performed with PSIPRED (20) and JPred4 (21) while the meta server MESSA was used for in-depth sequence analysis (22). Three-dimensional structure predictions were carried out with the PEP-FOLD server (23), and predictions of orientation of the peptide in the membrane were done with the PPM server (24). Figures were generated with PyMOL (http://www.schrodinger.com).

Animal experiments were approved by The University Board Ethics Committee for Experimental Animal Studies (#2303.01). Two groups of 3 nude mice were exposed to single (acute toxicity) or daily intraperitoneal (i.p.) administration for 5 consecutive days (subacute toxicity) of increasing doses of RT53 in normal saline or normal saline alone. For each treatment schedule, weight change, signs of toxicity were monitored up to 14 days after the last administration. Blood was collected from the saphenous vein before injections started and 2 hour following the last injection and the samples analyzed for cell count using an automated hematology analyzer (MedoniCA620, Stockholm, Sweden). All mice were euthanized by cervical dislocation under anesthesia. At the moment of euthanasia, the organs were harvested and underwent extensive macroscopic and microscopic examination.

Male FVB/N mice were immunized with 125 μg of RT53 daily for 5 weeks. Blood samples were collected before and 5 weeks after the first immunization, and antibodies were detected using an IgG mouse ELISA Kit (Abcam) assay against mouse whole IgG.

Human melanoma xenograft tumors were obtained by subcutaneous (s.c.) injection of 4 × 10⁶ of SK-Mel-28 or 2 × 10⁶ of C8161 cells (100 μL) in the right flank of 6-week-old nude mice (Centre-Elevage-Janvier). Mice were maintained in specific pathogen-free animal housing (IUH). Treatment started after randomization when tumors were visible and consisted of daily i.p. injection of normal saline solution or RT53 or RT53M in normal saline solution (n = 7 per group). Tumor growth was monitored by a digital caliper and volume was calculated using the formula: Length x Width²/2. Animals were euthanized after 21 days of treatment or when tumor size reached the ethical endpoint.

Nude mice bearing subcutaneous C8161 human melanoma xenografts were injected i.p. with either 5 mk/kg of rhodamine-labeled RT53 in normal saline or normal saline alone. The mice were euthanized 1 hour after injection, and the organs and tumors were isolated and the fluorescence observed by an IVIS spectrum in vivo imaging system (Caliper).

Tumors were fixed in 4% neutral buffered formalin and embedded in paraffin. Sections (4 μm) were stained with hematoxylin-eosin (H&E) and subjected to microscopic analysis. Anti-Ki-67 antibody (Abcam) was used as marker for tumor cell proliferation. The incidence of apoptotic tumor cells was assessed by TUNEL assay with the In-Situ Cell Death Detection Kit (Roche). Histological analysis was performed at the HistIM facility of Cochin Institute (Paris, France). Slides were imaged using a Lamina multilabel slide scanner (Perkin Elmer).

B16F10 cells were exposed to 30 μmol/L RT53 for 20 hour for cell death induction and inoculated subcutaneously (2 × 10⁶ cells) into the left flanks of C57BL/6 mice (n = 8 per group). Seven days later, the mice were challenged subcutaneously on the right flank with 1 × 10⁶ live B16F10 cells. Tumor growth on the challenge site was evaluated using a digital caliper. Animals were euthanized when tumor size reached the ethical end point or were necrotic.

The Student t test was used to test for statistical significance of the differences between the different group parameters. P values of less than 0.05 were considered statistically significant.

We previously demonstrated that preincubation of cancer cells with a low dose of RT53 (see Fig. 1A) increases drug-induced cell death and prevents AAC-11 interaction with the apoptotic factor Acinus (11). In line with our previous data (11), a mutant peptide (RT53M, see Fig. 1A) in which leucines 384 and 391 (AAC-11 numbering) were substituted by glycines did not sensitize cancer cells to etoposide-mediated cell death nor it prevented AAC-11 interaction with Acinus (Supplementary Fig. S1). We assessed the viability of melanoma SK-MEL-28 cells or nonmalignant HaCaT cells following exposure to increasing concentration of RT53. As shown in Fig. 1B, RT53 exhibited dose-dependent cell death activity in SK-MEL-28 cells, whereas very low toxicity was observed in HaCat cells. However, RT53M as well as the penetratin domain alone (see Supplementary Table S1) did not decrease cell viability of the cancer cells (Fig. 1B). The effect of RT53 on cellular viability was further assessed on various solid and hematological tumor cell lines, as well as normal cells. As shown in Fig. 1C, RT53 exhibited cell death activity in all cancer cells tested above 15 to 20 μmol/L in both BRAF^V600E (SK-Mel-28, Lu1205, and A375) and BRAF^WT (C8161, MEWO, and COLO 792) melanoma cells. Interestingly, RT53 was essentially not toxic to the normal cells tested (Fig. 1C). Here again neither RT53M nor the penetratin domain alone induced cell toxicity (data not shown). We next investigated the effect of the internalization sequence upon cancer cell toxicity by conjugating the AAC-11 (363-399) domain to the transactivator of transcription (TAT) cell-penetrating sequence (25). The resulting TAT-(363-399) peptide (Supplementary Table S1) was at least as effective as RT53 for decreasing SK-MEL-28 cancer cells viability (Fig. 1D). Finally, attaching the penetratin sequence on the C-terminal had no influence on the cytotoxic properties of the AAC-11 (363-399) domain as the resulting (363–399)-Pen peptide (Supplementary Table S1) elicited similar cytotoxicity as RT53 toward cancerous cells (Fig. 1D). Combined, these data indicate that the cytotoxic effect of RT53 is not a nonspecific effect of the leader sequence.

We next investigated RT53 mechanisms of cancer cell death. As shown in Fig. 2A, nontoxic concentrations of cycloheximide did not alter RT53 cytotoxic action, suggesting that RT53-induced cell death does not depend on de novo protein synthesis. Caspase activity assays indicated that RT53, but not RT53M, treatment of A549 cells promoted caspase-9 as well as caspase-3/7 activation (Fig. 2B), albeit at relatively low levels compared to etoposide-induced cell death (not shown). However, treatment of cells with the pan-caspase inhibitor zVAD-fmk failed to prevent RT53-induced cell death (Fig. 2C), suggesting that the observed cytotoxicity of RT53 is not dependent on caspases.

Inhibition of receptor-interacting protein kinase 1 (RIPK1) by necrostatin-1 did not inhibit RT53 cytotoxic activity, thereby excluding necroptosis (Fig. 2D). Finally, 3-methyladenine failed to inhibit RT53-cytotoxic effect, thus excluding autophagic cell death (Fig. 2E). Combined, our results indicate that RT53-mediated cell death mechanism is independent of caspases and necroptosis, as well as autophagy.

To determine whether AAC-11 availability was a necessary prerequisite for the cytotoxic effect of RT53, AAC-11 expression was knocked down in SK-MEL-28 and C8161 cells by using two different shRNAs, introduced through lentiviral particles. As shown in Fig. 2F, silencing of AAC-11 did not have any significant effect on RT53 cytotoxicity in SK-MEL-28 or C8161 cells, indicating that the cytotoxic effect of RT53 does not depend on AAC-11 expression levels.

We then evaluated necrotic cell death of SK-MEL-28 cells treated with RT53 by propidium iodide (PI) uptake assay. As shown in Fig. 3A, SK-MEL-28 cells, but not HaCat cells, displayed an increasing loss of plasma membrane integrity rapidly after exposure to RT53, whereas RT53M treatment did not induce plasma membrane alteration. Similar results were obtained using A549, C8161, MCF-7, Karpas-299, and SU-DHL-5 cell lines (data not shown).

To monitor the effect of RT53 at the morphological level, we performed time-lapse fluorescence imaging of C8161 cells exposed to RT53 or RT53M in the presence of PI. Incubation with RT53, but not RT53M, induced massive cell blebbing and swelling, together with PI incorporation and accumulation of cellular debris after 3 hours of incubation (Fig. 3B).

Necrotic cells release endogenous molecules, such as LDH, high-mobility group box 1 (HMGB1), or ATP (26, 27). As shown in Fig. 3C, RT53 exposure resulted in a drastic increase in LDH release into SK-MEL-28 cell supernatants, indicating membrane damage, whereas RT53M-incubated cells or nonmalignant HaCat cells showed little LDH release. Release of LDH was concomitant with PI permeability (Fig. 3B). RT53, but not RT53M, treatment of SK-Mel-28 cells increased levels of extracellular HMGB1 (Fig. 3D) and ATP (Fig. 3E). Combined, these data indicate that RT53 induces rapid necrosis, via membranolysis, of cancerous cells.

To investigate whether RT53 targets the mitochondria for cell death induction, we assessed mitochondrial transmembrane potential (ΔΨ_m) by using the fluorescent potentiometric probe DiOC6(3), by flow cytometry. Significant loss of ΔΨ_m was observed following RT53, but not RT53M, treatment, as indicated by a decrease in DiOC6(3) intensity, indicating a breakdown of mitochondrial membrane integrity (Fig. 3F). Necrosis is accompanied by mitochondrial swelling and loss of mitochondrial membrane potential (28). Therefore, the observed mitochondrial damage following RT53 incubation could either constitute a subsequent event of RT53 intracellular action or be a side effect caused by membranolysis-induced necrosis. To examine whether RT53 induces mitochondrial release of apoptogenic factors, purified mitochondria from HeLa cells were incubated with increasing amounts of RT53 or RT53M with or without cytosolic extracts. As shown in Supplementary Fig. S2A, neither peptide induced any detectable cytochrome c, AIF, and Smac release, as opposed to the control atractyloside. We then assessed RT53 mitochondrial uptake by incubating rhodamine-labeled RT53 or RT53M with isolated mitochondria. No fluorescence was detected from mitochondria incubated with RT53 or RT53M, in contrast with the Rhodamine 123 control (Supplementary Fig. S2B), indicating lack of uptake of RT53 by isolated mitochondria. Combined, these results suggest that the oncolytic effect of RT53 mostly involves plasma membrane perturbation, the observed mitochondrial alterations likely being the consequence of the resulting necrotic death.

Because of its membrane lytic activity, we studied whether RT53 can bind to the plasma membrane. C8161 or HaCat cells were incubated with rhodamine-labeled RT53 or RT53M and the peptides respective fluorescent patterns analyzed. To avoid toxicity, we used a 5 μmol/L sublethal concentration of RT53 (see Fig. 1C). Strikingly, RT53 localized in a punctate pattern at the plasma membrane of the cancerous C8161 cells together with a diffuse intracellular distribution (Fig. 4A). This punctate pattern suggests a compartmentalized accumulation for RT53 in membrane microdomains rather than nonspecific binding. Interestingly, RT53 distributed rather uniformly throughout the untransformed HaCa cells, without accumulation at the cell membrane. Finally, RT53M peptide exhibited a homogeneous pattern in both transformed and untransformed cells (Fig. 4A). Overall, these observations suggest that the discrete, punctate pattern observed for RT53 in cancer cells membranes could be indicative of binding of RT53 to cancer cells' membrane specific target(s). Because RT53 is only cytotoxic to cancer cells, our data imply that the resulting membrane sequestration of RT53 could be necessary for its cancer cells membranolysis properties.

Although membrane-active peptides (MAP) exhibit significant variability in conformations, a number of MAPs possess a linear α-helical structure (29). To obtain structural information about RT53, we used two well-established secondary structure prediction servers: PSIPRED (20) and JPred4 (21) as well as the MESSA meta server for in-depth sequence analysis and search for homologous structural templates (22). The secondary structure prediction methods indicate that RT53 should essentially adopt an α-helical structure (not shown), in agreement with the known 3D structures of the two parts of the peptide, the penetratin segment (30) and the AAC-11 protein (14), yet with a possible, but less probable, break around the RT53 peptide residues FARGL. Three-dimensional structure predictions carried out with the PEP-FOLD server (23) also suggested an essentially helical structure for RT53 (Fig. 4B). We further explored the possible orientation of the RT53 peptide with a cell membrane with the PPM web server (24). Spatial positions prediction analysis indicated that RT53 should belong to the peripheral protein type as its predicted transfer energy from the solvent to the membrane would be around −8.6 kcal/mol. Therefore, the peptide could have a surface orientation where its long axis would be parallel to the membrane due to the partial amphipatic nature of the molecule (Fig. 4C, left). Yet, an alternative, probable transmembrane orientation is possible due to the presence of a relatively long, mainly hydrophobic segment (residues FARGLQVYIRQL; Fig. 4C, right). Overall, these data suggest that RT53 appears to possess a membrane active conformation, which likely explains its pore-forming ability when retained in the membrane of cancer cells.

We next sought to explore if RT53 peptide might represent a possible therapeutic strategy to suppress tumor growth in vivo. We chose to focus our study on melanoma because the clinical management of this highly aggressive skin cancer remains challenging (31). Preliminary toxicity analysis in nude mice indicated that neither repetitive i.p. administration of RT53 at doses up to 60 mg/kg nor single i.p. administration at doses up to 100 mg/kg affected mouse behavior or growth (Fig. 5A). No noticeable changes in complete blood counts or organ toxicity (macroscopic or microscopic) were noted with either treatment schedule (data not shown). Measure of the antibody response using immunocompetent mice did not reveal antibody production against RT53 after 5 weeks of daily injections (Fig. 5B). Combined, these observations indicate that RT53 exhibits limited or nonexistent immunogenicity and toxicity in mice.

We next evaluated the efficacy of RT53 in both BRAF^V600E (SK-Mel-28) and BRAF^WT (C8161) xenograft models of melanoma. Interestingly, treatment with RT53, but not RT53M, induced significant tumor growth inhibition in both C8161 (approximate tumor growth reduction of 63%, P < 0.005; Fig. 5C) and SK-Mel-28 (approximate tumor growth reduction of 80%, P < 0.005; Fig. 5D) xenograft models as compared with saline-injected mice. These data demonstrated that RT53 was able to reduce the growth of melanoma tumors at distant site as single agent and regardless of their BRAF mutational status.

To better understand the in vivo performance of RT53, we studied the biodistribution of a rhodamine-labeled peptide administrated i.p. in mice-bearing subcutaneous C8161 tumors using an IVIS imaging system (Xenogen). RT53 was detectable in the liver, lungs, spleen, and kidneys, but not in the brain, suggesting that it may not cross the blood–brain barrier (Fig. 6A). Very interestingly, accumulation of the fluorescent conjugate in the tumor was clearly evident, indicating that RT53 is able to reach and accumulate in tumors in vivo.

We next investigated the mechanism by which RT53 inhibits tumor growth by examining necrosis, cell proliferation, and apoptosis in time- and size-matched xenograft tumors generated by RT53-treated or control groups. As shown in Fig. 6B (top), hematoxylin–eosin staining of tumor sections revealed larger percentages of necrotic regions in RT53-treated tumors compared with tumors from control groups. In non-necrotic regions, no obvious differences in proliferation (Ki-67 staining) were seen between the different tumors (Fig. 6B, middle). Interestingly, an increased number of apoptotic cells were observed in RT53-treated tumors compared with control tumors (Fig. 6B, bottom). Combined, these data indicate that RT53 inhibited melanoma tumor growth by inducing necrosis and triggering apoptosis.

Our results indicate that RT53 can induce the release of HMGB1, cytochrome c, and ATP from cancer cells, which are known to be able to function as damage-associated molecular patterns (DAMP) that induce host antitumor response (32). We therefore investigated the ability of RT53 treated tumor cells to activate the adaptive immune system using a well-established vaccination assay in immunocompetent C57BL/6 mice (33). In vitro experiments demonstrated that RT53 efficiently killed B16F10 murine melanoma cells (Fig. 7A) and, as described for human melanoma cell lines, RT53-treated B16F10 cells released both HMGB1 and ATP (not shown). Very interestingly, vaccination of C57BL/6 mice with RT53-killed B16F10 cells induced tumor growth inhibition at the challenge site compared with control mice (Fig. 7B), strongly suggesting that RT53-treated cells can activate the adaptive immune system. Notably, tumor growth was absent in 25% of the animals 30 days after challenge with live cells, while all the control animals developed tumors at the challenge site within 7 days after inoculation (Fig. 7C). Combined, these data indicate that RT53 elicits immunogenic cell death and that RT53 treatment of B16F10 melanoma cells mediates anticancer effect in a prophylactic tumor vaccination model.

In this study, we conducted proof-of-principle experiments for testing the therapeutic value of using RT53 as a novel cancer treatment agent. RT53 selectively killed multiple cancer cell lines in vitro, while sparing nonmalignant cells, through membranolysis, leading to release of DAMPs. Interestingly, RT53 can inhibit tumor growth in vivo in xenotransplant melanoma models, BRAF wild-type and mutant, without off-target toxicity, and RT53-treated mouse melanoma cells mediated anticancer effects in a prophylactic tumor vaccination model.

RT53 inserts in the membrane of cancer cells, but not normal cells, suggesting binding with a membrane partners that is absent or minimally present in the membranes of untransformed cells. This binding could allow the peptide to accumulate in the cancer cells' membranes where it could undergo pore formation and induce necrosis, as described for other MAPs (34). The inability of RT53 to alter normal cells as well as mitochondrial membranes further supports the idea that RT53 cytotoxicity toward cancer cells is not due to an unspecific detergent-like effect on the plasma membrane. Therefore, our hypothesis is that the cytotoxic effect of RT53 results from a two-step mechanism: first RT53 binds, through its AAC-11 sequence, to cancer cell–specific targets on the cell membrane. Second, after a critical threshold concentration of the peptide is reached, a marked membrane depolarization occurs, caused by pore formation, leading to cell death. This is in line with our observations that RT53 induces necrotic tumor cell death only when used at a certain concentration. Interestingly, an RT53 proposed mechanism of action is very reminiscent to that of another anticancer, MAP called PNC-27. PNC-27, which contains an HDM-2-binding domain derived from p53 fused to the penetratin sequence, induces membrane leakage of cancer cells, but not normal cells, through binding to HDM-2 in the transformed cell plasma membranes (35–37). Structure prediction analysis suggests that like PNC-27 (37), RT53 possesses a membrane active conformation. At this moment, the identity of RT53 membrane partner(s) is still unknown. RT53 is still cytotoxic against cancer cells following AAC-11 knockdown. Even known, we cannot rule out an off-target effect of the peptide that could result in cancer cell death in the absence of AAC-11, our data strongly suggest that AAC-11 is not a membrane partner of RT53. Preliminary experiments indicate that among the currently known AAC-11 interactors, FGF2, Acinus, ALC1 (11, 15, 16), and ALC1 do not appear to be involved in RT53 retention in cancer cells' membrane (not shown). AAC-11, which is overexpressed in most cancer cells, upregulates FGF2 signaling (13). As FGF2 is known to be present at the plasma membrane (38), we are investigating whether RT53 might bind to FGF2, which could mediate RT53 membrane retention. Interestingly, FGF2 stimulation is known to induce plasma membrane translocation of the scaffold protein c-Jun-NH2-terminal kinase (JNK)/stress-activated protein kinase-associated protein-1 (JSAP1; ref. 39). It is therefore possible that AAC-11-mediated sustained FGF-2 signaling in cancer cells might induce membrane relocalization of RT53-binding proteins. We are currently using genome-scale knockout screenings as well as peptide pulldown-based strategies to gain insight into the identity of the molecules that mediates RT53 membrane retention in cancer cells.

Necrosis induces intracellular potassium effluxes. Potassium efflux has been shown to play an important role in cell death and caspase activation, and depletion of potassium with depolarizing drugs was shown to cause caspase activation (40–42). Therefore, caspase activation by RT53 probably represents a side effect caused by its pore-forming properties, as witnessed with Staphylococcus aureus α-toxin, which induces cell death in a necrotic-like manner, through insertion into the plasma membrane and subsequent pore formation, despite caspase activation (43). This explains why caspase inhibition using a pan-caspase inhibitor did not prevent RT53-mediated cell death.

Importantly, RT53 anticancer effects in vitro translated well in s.c. mouse xenografts with melanoma cells. Indeed, RT53 alone substantially reduced tumor growth of both BRAF wild-type and mutant models, when injected systemically. Histological analysis of RT53-treated tumors indicated both increased tumor cell apoptosis and necrotic cytotoxicity, compared with controls. AAC-11 deregulation is known to make cancer cells more vulnerable to additional stress, such as nutritional stress (12) and, when used at sub-cytotoxic doses, RT53 decreases the prosurvival functions of AAC-11 and sensitize cancer cell to various factors in vitro, such as chemotherapeutic drugs and serum deprivation (11). It is therefore likely that RT53 makes cancer cells more vulnerable to environmental factors within the tumor, such as decreased oxygen tension, pH, and nutrient availability, hence increasing apoptosis in tumors from RT53-treated mice. Based on these observations, we hypothesize that RT53 could behave as a chemosensitizer in clinical settings, hence potentiating the efficacy of chemotherapeutic agents, and we are now evaluating this hypothesis. Cancer cells treated with RT53 in vitro released DAMPs, such as HMGB1 and ATP, suggesting that RT53-induced cell death is immunogenic. Interestingly, RT53-killed B16F10 cells mediated anticancer effect in syngeneic C57BL/6 mice in a tumor vaccination assay, indicating potent immune response in vivo. Experiments are currently under way to evaluate cross-priming and cytokine release of cytotoxic T lymphocytes induced by RT53-treated cells, to further examine the adaptive immune response to the antitumor vaccination achieved with RT53-treated cells.

RT53 possesses favorable drug-like properties. It is not toxic, even at doses 20-fold higher than the efficacious dose, not immunogenic, exhibits efficient biological effect and it reaches distant tissues and organs, including subcutaneous tumors where it accumulates. Therefore, our results demonstrate that peptide-based targeting of AAC-11 can constitute a promising new approach in the treatment of melanoma and possibly a significant fraction of human cancers.

No potential conflicts of interest were disclosed.

Conception and design: L. Jagot-Lacoussiere, B.O. Villoutreix, H. Bruzzoni-Giovanelli, J.-L. Poyet

Development of methodology: L. Jagot-Lacoussiere, E. Kotula, J.-L. Poyet

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Jagot-Lacoussiere, E. Kotula, B.O. Villoutreix

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Jagot-Lacoussiere, E. Kotula, B.O. Villoutreix, H. Bruzzoni-Giovanelli, J.-L. Poyet

Writing, review, and/or revision of the manuscript: L. Jagot-Lacoussiere, B.O. Villoutreix, H. Bruzzoni-Giovanelli, J.-L. Poyet

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Kotula, J.-L. Poyet

Study supervision: J.-L. Poyet

We thank Prof. Nicole Basset-Seguin and Drs. Marie Dutreix, Robin Fåhraeus, and Nicolas Dumaz for providing cell lines. We gratefully thank the coworkers of the Animal Experimental Facilities and the Imagery Department of the IUH.

This work was supported by the INSERM and a grant from INSERM Transfert and SATT IDF Innov.

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.