A Designed Inhibitor of p53 Aggregation Rescues p53 Tumor Suppression in Ovarian Carcinomas

Abstract

Half of all human cancers lose p53 function by missense mutations, with an unknown fraction of these containing p53 in a self-aggregated amyloid-like state. Here we show that a cell-penetrating peptide, ReACp53, designed to inhibit p53 amyloid formation, rescues p53 function in cancer cell lines and in organoids derived from high-grade serous ovarian carcinomas (HGSOC), an aggressive cancer characterized by ubiquitous p53 mutations. Rescued p53 behaves similarly to its wild-type counterpart in regulating target genes, reducing cell proliferation and increasing cell death. Intraperitoneal administration decreases tumor proliferation and shrinks xenografts in vivo. Our data show the effectiveness of targeting a specific aggregation defect of p53 and its potential applicability to HGSOCs.

Copyright © 2016 Elsevier Inc. All rights reserved.

Figures

Figure 1. p53 aggregation propensity and ReACp53 docking on the p53252-258 amyloid zipper structure

A. ZipperDB (http://services.mbi.ucla.edu/zipperdb/) predicts multiple segments in the p53 DNA-binding domain as aggregation prone. The highest propensity ones are located in the 252-258 region. Colored bars indicate aggregation-prone segments with Rosetta energies below -23 kcal/mol. B. The 252-258 segment (red) is mapped on the p53 DNA-binding domain structure. The segment in yellow (residues 213-217) is the epitope recognized by the PAb240 antibody which binds to partially unfolded p53. Both segments are buried in the p53 structure when the protein is fully folded. DNA is in gold. C. The ReACp53 peptide (ball-and-stick; cyan, blue and red represent carbon, nitrogen and oxygen atoms respectively) is modeled on the p53252-258 amyloid steric zipper structure determined in this study (“PDB: 4RP6”). The arginine in position 3 (in yellow) creates a steric clash with the adjacent β-sheet and additionally impedes incoming molecules from adhering on top while binding to the steric zipper below. Three adjacent β-sheets (in grey and red) of the p53 amyloid spine structure are shown viewed down (left) or nearly perpendicular to the fibril axis (right). See also Tables S1–2 and Figure S1.

Figure 2. ReACp53 inhibits p53 aggregation in primary cells from HGSOC patients, and re-localizes p53 to the nucleus in an active conformation

A. Mutant p53 forms aggregates appearing as puncta in the cytosol of primary cells from two HGSOC patients (see Figure S2A for additional examples). ReACp53 reduced the number of cells with puncta and caused p53 to localize to the nucleus. Scale bar: 20 μm. B. Quantification of number of cells with aggregated p53 and nuclear p53 in three clinical samples. The number of cells with puncta or nuclear p53 counted in 3–5 different fields of view was expressed as % of the total number of cells ± %SD; symbols represent the values for the individual fields of view, bars are average values.C. DO-1, an antibody that recognizes p53 regardless of its conformation, binds to p53 in S1 GODL cells over a range of ReACp53 concentrations. PAb240, a conformation-specific antibody that binds only to mutant-like, inactive p53, recognizes and stains p53 in untreated cells, but not in ReACp53-treated cells, indicating that ReACp53 restores p53 to an active conformation. Scale bars: 50 μm. D. Quantification of PAb240 staining; the number of positively-stained cells in 3-5 different field of views is expressed as % of the total number of cells ± % SD. Symbols represent % calculated for the individual field of views, bars are average values. See also Tables S3-4 and Figure S2.

Figure 3. ReACp53 causes cancer cell death

A. MTS assay shows a ReACp53 concentration-dependent decrease in cell viability in S1 GODL cells. Values are represented as the average of 6 independent experiments (n=3/experiment) ± SEM. Average EC50 values from all experiments and their coefficient of variation (CV) are reported. B. A scrambled control peptide does not exhibit significant effect. ReACp53 is represented as the average of 6 independent experiments (n=3/experiment) ± SEM, Scrambled is presented as average of 3 independent experiments (n=3/experiment) ± SEM. Means were compared with t-tests, ***p<0.0001. C. ReACp53-treated OVCAR3 cells stained with YO-PRO-1 and PI to label apoptotic and necrotic cells. A scrambled peptide control did not elicit significant cell death. Scale bar: 100 μm. D. YO-PRO-1/PI stain of S1 GODL cells treated for 16h as quantified by flow cytometry. Scrambled peptide and staurosporine were included as controls. Symbols represent biological replicates (n=2) for two independent experiments, bars show the average for all experiments ± SD. Statistical significance was calculated by performing a repeated measure ANOVA with Holm-Sidak’s multiple comparison test, **p<0.001, ***p<0.0001. E. Western blot showing Bad cleavage in S1 GODL cell upon treatment with ReACp53 at concentrations of 5 μM and above, indicative of cell death. GAPDH stain was performed on the same membrane after stripping. F. MTS assay for ReACp53/QV-OPh or NEC-1 co-treatments. Triplicates for each concentration were measured, one representative experiment out of n=4 (QV-OPh) or n=3 (NEC-1) is shown. ReACp53-induced cell death could be partially rescued by inhibiting apoptosis with QV-OPh (at low ReACp53 concentrations) or with NEC-1 (at high ReACp53 concentrations). Averaged values normalized to vehicle are reported as % ± SD. Means were compared with unpaired two-tailed t tests. *p<0.005, **p<0.0005. G. Cell cycle distribution of S1 GODL cells treated with vehicle, 5 μM ReACp53 or 5 μM scrambled peptide for 4/5 hours as evaluated by DNA content measured by flow cytometry. Symbols represent biological replicates (n=2) for two independent experiments, bars show the average for all experiments ± SEM. Statistical significance was calculated by performing a repeated measure ANOVA with Holm-Sidak’s multiple comparison test, **p<0.001, ***p<0.0001. H. Schematic of the UtFIB infection experiment. I. Bright field and green fluorescence of cells post-infection show GFP expression. Scale bar: 100 μm. J. Western blot of lysates from GFP- and GFP/R175H p53 infected UtFIB showing p53 expression. GAPDH stain was performed on the same membrane after stripping. K. Immunofluorescence of fixed GFP/R175H p53 infected UtFIB showing p53 distribution in the cells. Scale bars: 50 μm L. Annexin V/PI staining of GFP- and GFP/R175H p53 infected UtFIB grown in 3D treated for 2 days with ReACp53 as measured by flow cytometry. One representative experiment is shown (n=3). Biological replicates (symbols, n=3) are normalized to vehicle and expressed as fold change ± SD. ANOVA with Tukey HSD significance criterion was performed to calculate p-values. ***p<0.0001. See also Figure S3.

Figure 4. ReACp53 causes cell death in organoids generated from HGSOC samples bearing p53 mutations

A. Schematics of the experiments performed in the 3D organoid model system. The blue and green boxes represent the two different types of experiment performed. B. S1 GODL organoids treated with 10 μM ReACp53 undergo a dramatic change in cell morphology and internalization of YO-PRO-1/PI, indicative of cell death. Scale bar: 50 μm. C. Semi-thin sections of the spheroids show the catastrophic effect of ReACp53 on spheroid morphology. D. TEM analysis of the same sample shows several features of apoptotic cells, including condensed mitochondria (a.), an enlarged nuclear envelope (b.) and enlarged ER (c., black arrowheads) and free ribosomes (orange arrowheads). E. ReACp53 affects cell viability of organoids generated from cell lines or HGSOC primary samples bearing p53 mutations. Organoids were treated for 2 days with the indicated ReACp53 concentrations and Annexin V/PI staining was measured by flow cytometry. Symbols are individual replicates (n=3), bars are average ± SEM; one representative experiment shown (n≥2). p-values were calculated by repeated measure ANOVA with Holm-Sidak’s multiple comparison test *p<0.05, **p<0.005, ***p<0.0001. F. Cell viability determined after a week of daily ReACp53 treatments by cell counting of triplicate samples. Values are normalized to vehicle; symbols show the average of triplicates ± SD.G. ReACp53 induces a significant decline in % of Ki67 positive cells relative to vehicle after a one-week treatment course as quantified by intracellular Ki67 levels measured by flow cytometry. Symbols represents individual replicates, bars average ± SD. Statistics calculated as in E. *p<0.05, **p<0.005, ***p<0.0001. See also Figure S4.

Figure 5. RNAseq of ReACp53 treated organoids

A. Scheme of experimental setup and overview of results. B. IPA analysis of molecular functions for the OVCAR3 dataset. C. Heat map of a subset of p53 pathway genes differentially regulated in ReACp53-responsive OVCAR3 organoids. Several canonical p53 targets are present. D. Fold change for a subset of p53 transcriptional targets. Data are shown as log2(Fold Change), bars represent the average of three replicates ± SD.E. OVCAR3 cells have less p73 protein upon ReACp53 exposure as visualized by western blot. F. p73 reduction and thrombospondin increase at the protein level in S1 GODL cells correlates with the mRNA levels. In all cases, GAPDH stain was performed on the same membrane after stripping. All blots were repeated three times on three independent cell lysates. One representative example is shown. See also Table S5 and Figure S5.

Figure 6. ReACp53 causes regression of xenografts bearing an aggregation-prone p53 mutant in vivo

A. Schematics of the experimental design for xenograft models. Both in vivo experiments were performed twice, with n=3 mice/group, one representative experiment is shown. B. Minimal residual disease model. Tumor growth monitored over time showed a reduction in size of OVCAR3 but not MCF7 xenografts in mice treated with ReACp53. Data are shown as average volumes (symbols, n=3) ± SEM. Means of tumor weights (n=3) are shown as averages (bars) ± SD and compared using an ANOVA model and Tukey HSD significance criterion, **p<0.05. Scale bars: 1 cm. C–H. Treatment model. C. Same as in B. *p<0.05, **p<0.005. Scale bars: 1 cm. D. H&E and IHC on residual xenografts. Sections were stained for PanK, p53 and Ki67. Scale bars: 50 μm. E. Total and Ki67-positive cells were quantified on three different fields for each xenograft, and reported as % Ki67 positive cells (symbols). Lines represent the average for each treatment group ± SEM. p-values were calculated by ANOVA using the Tukey HSD significance criterion. **p<0.01. F. QPCR analysis of residual S1 GODL xenografts treated with 30 mg/kg ReACp53 for 9 days. Symbols represents the fold-change normalized to GAPDH for n=3 xenografts, bars are average values ± SEM. *p<0.05. G. H&E and IHC this short-term treated xenografts shows abundant Bax expression, indicative of apoptosis. Scale bar: 100 μm. H. TUNEL assay also showed a significantly higher proportion of death cells in ReACp53-treated grafts. Symbols represent the % of TUNEL-positive cells in five field of view sampling all the tumors while bars show the average values ± SD, ***p<0.0001. Scale bar: 100 μm. See also Table S6 and Figure S6.

Figure 7. ReACp53 causes regression of intraperitonal disseminated tumors in vivo

A. Schematics of the IP disseminated disease model experiment. B. Viability of cancer cells obtained from ascites (discussed in Methods) after 4 daily treatments as measured by flow cytometry of Annexin V/PI stained cells. Symbol represent values from individual mice (n=3/group), bars are average ± SD. Means were compared by t test. ***p<0.001 C. Cell cycle distribution of tumor cells obtained from ascites as measured by flow cytometry. Symbol represent values from individual mice (n=3/group), lines are average ± SD. Means were compared by t test. **p<0.01. D. Analysis of pelleted ascites-derived cancer cells by H&E and IHC. The black boxed area is magnified on the right. Scale bars: 200 μm for H&E; 50 μm for IHC. E. Bright field and IHC of cancer cells obtained from ascites after a three-week treatment and plated to confirm viability. Quadruplicates were pleated for each mouse (n=3 mice/group). The ReACp53-treated samples did not yield any live and proliferating cell. Scale bars: 100 μm. F. Organ implants as visualized by H&E and identified by positive p53 IHC staining. Scale bars: 200 μm for H&E and low magnification IHC; 50 μm for high magnification IHC.G. Quantification of organ implants upon histological examination of p53 stained sections. This conservative analysis does not take into account implant size, which were typically small (3–5 cells) in ReACp53 treated samples. Each symbol represents the average for all sampled sections (n=5) for a given mouse; three mice are reported. Bars represent averages for the three mice ± SEM. A statistically significant difference in the presence of implants in the uterus/ovaries was detected by performing repeated measure ANOVA with Holm-Sidak’s multiple comparison test. **p<0.05. Scale bars: 200 μm for H&E; 50 μm for IHC.

Figure 8. Model for the mechanism of action of ReACp53 (cyan)

Unstable p53 mutants partially unfold, exposing the aggregation prone segment LTIITLE. This segment interacts with the same segment from other p53 molecules, driving p53 into its inactive aggregated state (top). ReACp53 treatment blocks the aggregation pathway, shifting the equilibrium towards functional, soluble p53 (bottom). Functional p53 enters the nucleus and induces cell death and proliferation arrest. Folded p53 interacts with MDM2 and is degraded.