Targeting c-Myc with a novel Peptide Nuclear Delivery Device

Biologics such as peptides and antibodies are a well-established class of therapeutics. However, their intracellular delivery remains problematic. In particular, methods to efficiently inhibit intra-nuclear targets are lacking. We previously described that Pseudomonas Exotoxin A reaches the nucleoplasm via the endosomes-to-nucleus trafficking pathway. Here, we show that a non-toxic truncated form of PE can be coupled to peptides and efficiently reach the nucleoplasm. It can be used as a Peptide Nuclear Delivery Device (PNDD) to deliver polypeptidic cargos as large as Glutathione- S-transferase (GST) to the nucleus. PNDD1 is a fusion of PNDD to the c-myc inhibitor peptide H1. PNDD1 is able to inhibit c-Myc dependent transcription at nanomolar concentration. In contrast, H1 fused to various cell-penetrating peptides are active only in the micromolar range. PNDD1 attenuates cell proliferation and induces cell death in various tumor cell lines. In particular, several patient-derived Diffuse Large B-Cell Lymphomas cell lines die after exposure to PNDD1, while normal B-cells survive. Altogether, our data indicate that PNDD is a powerful tool to bring active cargo to the nucleus and PNDD1 could be the basis of a new therapy against lymphoma.

Scientific Reports | (2020) 10:17762 | https://doi.org/10.1038/s41598-020-73998-x www.nature.com/scientificreports/ While c-Myc is an attractive target for cancer treatment, its nuclear localization and the lack of a hydrophobic pocket make it difficult to target by small molecules. c-Myc inhibitors have been developed: for example a small alpha-helix peptide able to inhibit the LZ interactions and called the H1 (S6A, F8A) peptide (hereafter called H1). H1 is a 14 amino acid (NELKRAFAALRDQI) competitive inhibitor derived from the interaction domain with MAX 22 . H1 reduces levels of c-Myc/MAX dimers and therefore inhibits transcription of downstream genes 22 . However, H1 is not membrane permeable, so cell-penetrating peptides (CPP) are employed for H1 intracellular delivery. However, the effective dosages of most CPP-H1 are in the micromolar range 23 , making it impractical for drug development. Another Myc inhibitor is Omomyc, a polypeptide that binds to the E-Box and competes with the c-Myc-Max dimer to inhibit transcription 24,25 .
In this paper, we describe the development of a truncated PE toxoid as a Peptide Nuclear Delivery Device (PNDD). We show that it is able to deliver various cargo to the nucleoplasm, including the H1 peptide. Using this approach, we observed an inhibition of c-Myc activity in various cancer cells, including lymphoma-derived cell lines. In many cell lines, the inhibitor induces a significant reduction in cancer cell proliferation and eventually inducing cell death.

PE domains Ia is sufficient for nuclear delivery.
To determine the PE domains required to reach the nucleus, we generated four truncated toxoids devoid of the cytotoxic catalytic domain III. Here, PE amino acids are numbered from the first amino acid of the peptide signal. PE389 contains domains Ia and II, PE277 contains only domain Ia; PE212 and PE151 are truncated versions of domain Ia (Fig. 1A). We expressed these constructs in E. Coli under an Isopropyl β-D-1-thiogalactopyranoside (IPTG) inducible promoter. As most constructs were not readily soluble, we used urea extraction to obtain the recombinant proteins ( Supplementary Fig. S1A-L). To examine nuclear trafficking of the constructs, we first compared their enrichment in NAE to that of wild-type (wt) PE. To visualise delivery to the NAE, we used MG63, human bone osteosarcoma cells expressing high levels of LRP1. The MG63 cells were incubated with the purified constructs labelled with HiLyte Fluor 488 dye (488 dye for short) together with wt PE labelled with HiLyte Fluor 594 dye (594 dye for short) for an hour. NAE were defined as PE-positive endosomes located above or below the nucleus, in close contact with the nuclear envelope as previously described 2 . We observed colocalization of PE389, PE277 and PE212 with wt PE (Fig. 1B). By contrast, PE151 signal was barely detectable.
We next checked the ability of the constructs to reach the nucleus by cell fractionation. We incubated MG63 cells with PE389 or PE277 for 1 h before fractionation and found both constructs in the nuclear fraction, suggesting that domain II is not essential for nuclear translocation (Fig. 1C). PE212 appeared nearly as efficient as PE277 to be internalised and be enriched in the nuclear fraction (Fig. 1D). By contrast, PE151 was not detected in either total cell lysate nor nuclear fraction (Fig. 1D). In conclusion, PE389, PE277 and PE212 have comparable abilities to enter the nucleus, while PE151 appears to be poorly internalised. These results also indicate that domain II is not required for delivery in the nucleus and thus translocation.
To form an efficient delivery device, the toxoid selected must have good stability and solubility. Since the constructs are solubilised by urea for extraction, they need to be dialysed extensively before use, which can lead to aggregation and loss of material. We thus compared raw eluates, post-dialysis eluate supernatants and pellets by SDS-PAGE followed by Coomassie staining (Fig. 1E). The recovery of PE389 in the soluble dialysed fraction was 95%. By contrast, PE277 and PE212 only had 25% and 60% recovery respectively (Fig. 1E). Therefore, despite  To test the ability of PNDD to deliver protein cargo to the nucleus, we fused the 28 kDa Glutathione S-transferase (GST) in C-terminus of PNDD0. PNDD-GST was coupled with the 488 dye to study its subcellular distribution and found to localise in NAE after 1 h of incubation with MG63 cells (Fig. 1F). PNDD-GST was also detected by both PE and GST antibodies in nuclear fraction after 1 h incubation, suggesting that PNDD could deliver GST effectively to the nucleus (Fig. 1G).
A chimeric cMyc inhibitor, PNDD1, is delivered to the nucleus. We next fused the c-Myc inhibitor H1 at the C-terminus of PNDD0, hereafter called PNDD1 ( Fig. 2A). We found the solubility of PNDD1 to be similar to PNDD0, as seen from comparable levels of recovery after dialysis (Figs. 2B, S2A-B). We verified delivery to the nuclear fraction in MG63 cells after 1 h exposure using cell fractionation ( Supplementary Fig. S2C).
To evaluate the fraction of internalised PNDD1 that reaches the nucleus, we quantified the ratio in the nucleus to total extract ( Supplementary Fig. S2D) and using the nuclear located protein Max as a reference, we estimate www.nature.com/scientificreports/ that around 6% of internalised PNDD1 is found in the nuclear fraction after 1 h of incubation. This seems consistent with the relatively small amount observed in NAE compared to the perinuclear pool of PE (Fig. 1B). We next measured the kinetics of uptake and observed that cellular amounts of PNDD1 appear to saturate after one hour ( Supplementary Fig. S2E). This result suggests that either the LRP1 receptor is saturated and/or degradative mechanisms start to balance out the uptake of additional molecules. To test for intracellular degradation, we pulsed MG63 cells with PNDD1 for 1 h before incubation with normal media. Total cellular PNDD1 was quantified by western blot, leading to an estimated intracellular half-life of around 4 h. A residual intracellular pool of PNDD1 can be detected as late as 30 h post-wash ( Fig. 2C,D, Supplementary Fig. S2F).
Next, we explored how PNDD1 is degraded. Since the toxoid is translocated in the nucleoplasm and the cytosol, its degradation is likely to be in part cytosolic. Treatment with the proteasome inhibitor Bortezomib largely increased PNDD1 stability (Fig. 2E,F, S2G). This suggests that proteasomal degradation is a major pathway and proteasome inhibition a potential way to increase the potency of PNDD1 in vivo. PNDD1 specifically inhibits c-myc transcriptional activity. To analyse whether PNDD1 is able to inhibit c-Myc, we established a Myc activity reporter cell line. We selected the epidermoid carcinoma cell line A431 because of their high c-Myc levels and ease of transfection. We introduced the firefly luciferase under the control of an E-BOX containing promoter. We also introduced a Cytomegalovirus (CMV) promoter-driven Renilla luciferase gene reporter as a control (A431/EBox-Luc/CMV-Ren cells, A431-mrep in short). In this system, inhibition of c-myc should lead to a specific reduction of the firefly luciferase signal (Fig. 3A).
We verified PNDD1 uptake and nuclear location in A431-mrep by cell fractionation after 1 h treatment ( Supplementary Fig. S3A). After a preliminary dose response evaluation, we determined that 25 nM was an approximate EC50 for PNDD1 ( Supplementary Fig. S3B). We next used 50 nM to treat A431-mrep cells for 6 h. In these conditions, Firefly luciferase activity was decreased by ~ 75% while Renilla luciferase activity remained constant ( Supplementary Fig. S3C). Treatment with 50 nM PNDD0 did not have a significant effect on either reporters ( Fig. 3B and S3C).
Since our preliminary data with GST indicated that a larger peptide than H1 could be efficiently delivered by PNDD, we next tested coupling the 10KDa Omomyc to PNDD0 (hereafter referred to as PNDD2) (Fig. 3A). After verifying its cellular internalisation and nuclear enrichment ( Supplementary Fig. S3D), PNDD2 was applied to the A431-mrep cells for 6 h, leading to decreased Firefly luciferase activity in a dose dependent manner (Fig. 3B, S3F). PNDD2 appeared to be about 5 times more potent than PNDD1 in A431-mrep cells (Fig. 3B). We next tested the stability of PNDD2 and found that total cellular PNDD2 in MG63 cells could be detected up to 5 h after wash, suggesting a half-life roughly similar to PNDD1 ( Supplementary Fig. S3E). In sum, delivery by PNDD is adaptable to various, potentially large cargoes.
We next explored the kinetics of PNDD1 effects. A rapid fourfold reduction of Firefly luciferase signal was observed within 6 h hours, further decreasing to near undetectable activity by 24 h (Fig. 3C). By contrast, Renilla luciferase activity was near normal up to 16 h, after which it started to drop as well (Fig. 3C). We interpret this decrease as due to loss of cell fitness.
We next tested how a pulsatile exposure to PNDD1 would impact Myc activity. After 1 h incubation with PNDD1, the supernatant was removed and cells washed. As expected Myc inhibition was less marked but appeared surprisingly lasting, between 24 and 48 h (Fig. 3D, Supplementary Fig. S3G). Firefly luciferase activity rebound at longer time points, 72 h and beyond ( Supplementary Fig. S3G). This suggests that PNDD1′s effect on Myc activity is more prolonged than the half-life of PNDD1. By contrast with continuous exposure, Renilla activity was not significantly impacted by a 1 h exposure (Fig. 3D).
Next, to validate the on-target activity of PNDD1, we measured the changes in mRNA expression of several c-Myc biomarkers by quantitative RT-PCR after exposure at 50 nM for 16 h (Fig. 3E, S3H). Consistently, the expression of Myc targets CDC25A, MAT2A, MYCL, PCNA and SHMT1 were significantly down-regulated after PNDD1 treatment 26,27 . By contrast, mRNA levels of housekeeping genes are not affected (Fig. 3E, S3H).
Altogether, these results show that PNDD is able to deliver the H1 peptide and the Omomyc polypeptide to the nucleus. The fusion does not impair the inhibitory activity of the peptides and inhibition is rapid and lasting. PNDD1 is three thousand times more potent than cell penetrating peptides. In order to compare the efficiency of PNDD with existing cell entry systems, we fused H1 with commonly used cell-penetrating peptides (CPP) such as HIV TAT (TAT), cadherin peptide (CAD) or antennapedia (Int) 28 . We performed dose response assays comparing these CPP-H1 and PNDD1 on A431-mrep after 6 h incubation and at concentrations ranging from 1 nM to 1 uM (Fig. 3F, S3H). H1 peptide alone had no effect, suggesting an inability to penetrate the cell membrane. Among all the CPP-H1 fusion proteins, CAD-H1 was the most efficient with an EC50 of 75uM, followed by Int-H1 (EC50 = 200 uM) and TAT-H1 (EC50 = 500 uM) ( Supplementary Fig. S3H). Some of these EC50 values are higher than previously reported 29 . This may be due to differences in peptide preparation and assay specificity. Nevertheless, in our hands, the EC50 of the most potent CPP fusion, CAD-H1 is ~ 3000 times higher than PNDD1 (Fig. 3F, S3I, S3J).

PNDD1 inhibits proliferation of different tumour cells.
In order to test whether PNDD1 inhibition of c-Myc also affects cell proliferation, we tracked the growth of A431 cells for 8 days using time-lapse microscopy. A431 cell density was evaluated using phase contrast images of the culture and image analysis (Fig. 4A). After seeding at ~ 10% confluency (5000 cells per well in a 96 well plate), the culture reached confluency by day 4 (Fig. 4A). When incubated with 50 nM of PNDD1, the rate of cell proliferation decreased, reaching only about 70% confluency at day 4. In addition, cell confluency started to decrease after 4 to 6 days, suggesting that cells  www.nature.com/scientificreports/ started to die. The effect on cell proliferation was dose dependent, proliferation being slowed more with higher doses until completely blocked with 200 nM PNDD1 (Fig. 4B).
To test the effect of PNDD1 in other cell lines, we treated a range of adherent cell lines, HeLa, HepG2 and MB-MDA231 cells with 50 nM PNDD1 and quantified the effect using the same live proliferation analysis method with PNDD0 and PBS as negative controls. All tested cell lines showed a decrease of cell proliferation, albeit with different sensitivities (Supplementary Fig. S4A-F). To compare the sensitivity of each cell line, confluency was measured over time to estimate a growth rate. At 50 nM PNDD1 (Fig. 4C), the HepG2 cell line showed the highest sensitivity with no cell proliferation at 50 nM PNDD1 and the largest decrease in growth rate at 13.5fold (Fig. 4C, Supplementary Fig. S4A). The effect in HeLa and MB-MDA-231 cells was approximately a twofold decrease in growth rate (Fig. 4C, Supplementary Fig. S4B, S4D).
We next tested PNDD2 on all the cell lines, using 10 nM instead of 50 nM as suggested by results on A431rep. We found that PNDD2 was as or more potent in reducing cell proliferation in A431, Hela and MDA-231. For HepG2, PNDD1 at 50 nM is already blocking completely cell proliferation ( Supplementary Fig. S4A-D).
The observed reduction of growth rate could be due to quiescence or partial cell death. To test whether cell death was induced, DRAQ7, a far-red fluorescent dye that marks dead cells was added while incubating A431 with 100 nM PNDD1. A significant increase of DRAQ7 staining was apparent after 24 h (Fig. 4D,E). The decrease www.nature.com/scientificreports/ of Renilla luciferase reporter signal observed after 24 h of PNDD1 treatment is also signaling with a significant loss of fitness, beyond only quiescence (Fig. 3C,D). Overall, these results indicate that PNDD1 slows cell proliferation in many cancerous cell lines and can induce cell death of A431 cells if c-Myc inhibition is marked and sustained for several hours.

PNDD1 inhibits Diffuse Large Cell Lymphoma proliferation. While c-myc is presumably involved
in many different cancers, its implication in lymphomas has been well established. Diffuse Large B Cell Lymphomas (DLBCL), in particular, typically overexpress c-myc 17,18 . They are also difficult to treat. We obtained from clinicians several patient derived DLBCL lines: HT, OCI-LY3, OCI-LY19 and SUDHL2 and validated PNDD1 uptake and nuclear enrichment by cell fractionation (Fig. 5A, Supplementary Fig. S5A-B).
Next, we incubated the DLBCL cell lines with 50 or 100 nM PNDD1 for 4 days. We observed a decrease in cell count across all the DLBCL cell lines 2 days after treatment in a dose dependent fashion (Fig. 5B,D, Supplementary Fig. S5C, E). The percentage of live cells also decreased in a dose dependent manner, indicating that the treatment induced cell death (Fig. 5C,E, Supplementary Fig. S5D, F). As expected, the effect was specific to H1 as PNDD0 had no similar effect ( Supplementary Fig. S5G-H).
Among the cell lines, we could not observe a clear correlation between the amount of uptake of PNDD1 and reduction of cell proliferation. For instance, PNDD1 uptake is lower in OCI-LY3 than in SUDHL2 yet both cell lines displayed a similar rate of inhibition (Fig. 5C,E). We also did not observe a correlation between c-Myc levels and PNDD1 sensitivity. Similar levels of c-Myc were detected in untreated OCI-LY19 and SUDHL2, but SUDHL2 proliferation was slower than OCI-LY19 under PNDD1 treatment (Supplementary Fig. 5I). This was also observed in the adherent cell lines, where comparable c-Myc levels were detected (Supplementary Fig. 5J). Yet PNDD1 treatment resulted in differing declines in growth rates, with notably HepG2 being very sensitive (Fig. 4C, Supplementary Fig. 5J). It is possible that the sensitivity to PNDD1 depends on a non-linear combination of two or more factors (e.g. amount of PNDD1 translocated and c-Myc levels) instead of a single factor. In such a scenario, detecting a correlation could be challenging.
We did however observe a clear difference between the cancer cell lines and the immortalized B cell line GM22650, which displayed an arrest in proliferation but no cell death (Fig. 5F,G). These results are consistent with previously published work indicating the cancer cell specific addiction to c-Myc 30,31 .
Overall, our results indicate that PNDD1 could be used to treat DLBCL and its toxicity may be limited.

Discussion
In this report, we show that a truncated toxoid form of PE can be used as a nuclear delivery device for biologics. Multiple protein toxins have evolved to efficiently penetrate cells to reach cytosolic or nuclear targets 7 . For this, they bind cell surface receptors that traffic to a compartment where the toxin can interact with a translocation machinery. PE binds to LRP1 and traffics to the ER before PE or its domain III is translocated to the cytosol and reaches its main target, the ribosome 7,8 . In theory therefore, a truncated PE could be used to deliver cargo to the cytoplasm to inhibit protein-protein interaction. In addition, a fraction of LRP1 and thus PE also traffics to the NAE and is translocated in the nucleus 2 . While this trafficking is quantitatively limited and seems to play only a minor role in PE intoxication, we found that it is sufficient for PNDD to deliver cargo directly to the nucleus.
The PNDD approach is amenable to cargo of different sizes as demonstrated by the nuclear delivery of the 2 kDa H1 peptide, the 10 kDa Omomyc polypeptide and the 25 kDa GST protein. Thus, using various peptides, targeting a range of nuclear proteins implicated in cancer and other diseases appears possible. However, the amount of chimeric protein delivered to the nucleus could represent a limitation. Indeed, our data of cell fractionation and immunofluorescence suggests that less than 10% of PNDD1 entering the cell reaches the nucleoplasm. A target that is not too abundant is probably preferable.
The transcription factor c-Myc is a key regulator of cell growth and its expression levels are tightly regulated 32 . This might explain why its inhibition by PNDD1 or 2 is working so efficiently. In fact, nanomolar doses of PNDD1/2 were enough to significantly reduce c-Myc activity and induce cell death in DBCL cell lines. This range is several orders of magnitude lower than the tested CPPs, which require micromolar doses to be effective. A recent study has reported small molecules able to inhibit Myc/Max interaction, which also require low micromolar range 33 . This significant difference could stem from a combination of at least two factors. First, based on previous work, it seems likely that PNDDs translocate through membranes using cellular machinery such as the Sec61 translocon, which would be a directional transport a priori more efficient than physico-chemical diffusion through membranes 2,34 . Second, once in the nucleoplasm, PNDD1/2 are unlikely to diffuse out in the cytosol given their size, by contrast CPPs need to diffuse in the cytosolic fraction in addition to the nucleoplasm.
The reliance on the NAE pathway for delivery may also have a significant impact on the therapeutic index of PNDDs. Previous studies have reported that tumor cells tend to accumulate growth factor receptors in the nucleus, suggesting a more active NAE pathway 3,35 . It suggests that cancer cells may transfer more PNDDs to the nucleus than normal cells, impacting favorably the therapeutic index. Thus, the PNDD approach may be well suited for cancer treatment.
The therapeutic index of a treatment depends on many factors. It appears that tumor cells may be more sensitive than their normal counterpart to c-Myc inhibition, a phenomenon described as oncogene addiction 19 . Still, the specificity of PNDD1 induced cell death for cancer cells will need to be further expanded and verified. The precise mechanism underlying cell death of cancer cells after c-Myc inhibition is not known. We observe a significant delay between decrease in Firefly luciferase activity and the onset of cell death: reduction of luciferase activity occurs within 4 h of continuous incubation and is 80% reduced by 8 h. In contrast, we observe cell death as well as a general inhibition of transcription after 24-48 h in A431 and in DCBL cell lines. The Firefly luciferase Scientific Reports | (2020) 10:17762 | https://doi.org/10.1038/s41598-020-73998-x www.nature.com/scientificreports/ www.nature.com/scientificreports/ assay relies on an mRNA with a 1.5 h half-life and its protein product a 2-3 h half life 36,37 . By contrast, it seems to take about 24 h for Myc target genes to experience significant reduction at the protein level and induce a shutdown of cellular functions. Obviously, the requirement for a lasting inhibition of c-Myc activity may be a challenge for the use of PNDD1/2. Given the 4 h intracellular half-life of PNDD1, a constant delivery over 24 h may be required to irreversibly commit cancer cells to apoptosis. Alternatively, our data shows that it is possible to increase the intracellular half-life of PNDDs by treating cells concomitantly with a proteasome inhibitor such as bortezomib. Since bortezomib is also used sometimes for lymphoma treatment, it seems possible to combine it with PNDD type molecules for therapeutic purposes.
A potential therapeutic indication could be the lymphomas of the DLBCL type. PNDD1 was efficient to slow down cell proliferation and induce cell death in all the patient cell lines we tested. Even in cell lines with comparatively lower PNDD1 uptake, such as OCI-Ly3, the induction of cell death was robust. It might be because these cells are highly dependent on Myc activity, as a study has recently shown that OCI-LY3 contains multiple copies of MYC 38 . By contrast, the immortalised normal B-cells were able to stop proliferation without significant cell death.
In conclusion, PNDD appears to be an efficacious method to deliver peptidic cargo in the nucleus, with potentially many different applications. In the examples we tested, PNDD did not affect the functional properties of its cargos, H1 and Omomyc. It is efficacious at much lower concentrations than Cell Penetrating Peptides that have demonstrated therapeutic abilities 39 . Further characterisation and optimisation of PNDD1/2 might be required to obtain serum and intracellular half-lives values that would be compatible with in vivo use.

Material and methods
Gene synthesis and cloning. The sequences of PNDD0 and PNDD1 were synthesized and cloned in a pET100/D-TOPO vector using gene synthesis (Thermo Fisher Scientific, Carlsbad, CA). H1 peptide sequence is underlined. Firefly luciferase and Renilla luciferase activity reading. 40,000 A431-mrep cells were seeded in a 96-well plate (Falcon) 48 h. Luminescence is detected using the Promega Dual-Glo luciferase assay system, according to the manufacturer's protocol and read using Tecan Infinite M200 microplate reader using 100 ms integration time. www.nature.com/scientificreports/ of an initial denaturation step at 95 °C for 10 min followed by an amplification program for 40 cycles of 15 s at 95 °C, and 60 s at 60 °C with fluorescence acquisition at the end of each extension. The relative expression of each gene between treated and untreated samples is calculated using the comparative ΔΔCT method, using the mock treated sample as calibrator and housekeeping gene as internal control.
DLBCL cell proliferation and cell death assay. 200,000 to 300,000 cells were seeded 24 h before PNDD0 or PNDD1 50 nM treatment. Three samples of the suspension cell culture were taken every 24 h after PNDD0 or PNDD1 50 nM treatment to measure the cell number and the percentage of dead cells. Three aliquots of supernatant were taken and mixed with equal volume of Trypan blue before reading on Countless Cell Automated Cell Counter (Invitrogen).
Disclosure. PNDD1/tPE-H1 method to inhibit c-Myc is filled under the title and patent application: Chimeric Molecule for Targeting c-Myc in Cells. Patent Application No PCT/SG2018/050584.