Sensitizing Immune-Refractory Ovarian Tumors via p53 Mutation-Tailored Immunotherapy

Abstract

High-grade serous ovarian cancer demonstrates limited responsiveness to immune checkpoint inhibitors, owing in part to immunosuppressive environments shaped by nearly universal p53 aberrations. Utilizing an immunocompetent mouse model and individual p53 mutations, we identified a dependence of the p53-R270H mutation (equivalent of human R273H) on regulatory T cells (Tregs) and the PD-1/PD-L1 axis. Analysis of patient datasets associated R273H with elevated levels of two p53 targets, PD-L1 and amphiregulin (AREG), a Tregs growth factor. In contrast to p53-R172H tumors, where there was limited activity, dual antibody therapy targeting AREG and PD-L1 selectively and effectively inhibited R270H tumors. This involved polarization toward M1 macrophages, infiltration of CD8+ T cells, diminished Ly6G+ neutrophils and downregulation of interleukin-4. In patient-derived R273C organoids, the combination treatment reduced the CD4/CD8 ratio. This study is the first to establish a mutation-tailored therapeutic approach that leverages the capacity of p53 to modulate immunosuppressive mechanisms.

Figure 1:. p53-R273 associates with relatively high PD-L1 and shorter survival of patients with cancer; in EOC patients and a mouse model this mutation predicts high aggressiveness relative to p53-R175.

(A) The TCGA dataset was analyzed for survival probabilities of patients with various types of cancer harboring either p53-R175 or p53-R273 (>30 patients per cancer type). The total number of patients was 170 (R175X) and 272 (R273X). (B) RNAseq data from the TCGA database was used to analyze the correlation between p53 mutation status, either p53-R175X or p53-R273X, and transcript levels corresponding to PD-L1 (CD274) using variance stabilizing transformation. The cancer types analyzed were UCEC (uterine corpus endometrial carcinoma), STAD (stomach adenocarcinoma), LGG (lower grade glioma), COAD (colon adenocarcinoma) and BRCA (breast invasive carcinoma). (C) The TCGA-OvCA dataset was separately analyzed for survival probabilities of patients harboring either p53-R175H (n=15) or p53-R273H (n=12). (D) The following derivatives of ID8 cells were examined using immunoblots of whole cell extracts and the indicated antibodies: A p53-deficient stable line (p53−/−), which was used to derive two ID8 isogenic lines harboring either murine p53-R172H or murine R270H (equivalent to human R175H and R273H, respectively). (E) Groups of six C57/Black female mice were intraperitoneally injected with the following isogenic ID8 lines (5×10⁶ cells per animal): (i) p53(−/−), (ii) p53(−/−) cells stably expressing p53-R172H, and (iii) p53(−/−) cells stably expressing murine p53-R270H. indicated isogenic cell lines, including wildtype ID8 cells. Animal survival curves are presented. *, p< 0.05, ***, p< 0.001. (F) The abdominal perimeters of the mice from E were measured once per week and a representative graph for week 5 post inoculation is presented as proxy of tumor volume. (G) The numbers of metastatic nodules corresponding to the experimental arms shown in E were determined till week 10 post tumor cell inoculation. Shown is a graph corresponding to the fifth week’s count of nodules. (H) Ascites fluids were obtained from the mice of E and subjected to ELISA that measured the abundance of secreted AREG in the respective fluids. (I) PD-L1 knock-out clones of ID8 cells were subjected to immunoblotting to determine the protein levels of PD-L1 in the parental and PD-L1 manipulated derivatives of ID8 cells. (J and K) Groups of C57/Black female mice (n=6) were intraperitoneally injected with wildtype ID8 cells (5×10⁶ cells per animal) or derivatives lacking PD-L1 (two clones; panel J). Alternatively, mice were injected with ID8 p53(−/−) cells expressing either wildtype PD-L1 or cells whose PD-L1 was deleted (two clones; panel K). Shown are the corresponding animal survival curves. Indicated are log-rank test scores and the median gain in survival time. ****, p<0.0001.

Figure 2:. Transcriptional profiling uncovers coupling of p53-R270H, but not R172H, to redox regulation.

(A) Bulk RNA sequencing analysis was performed on four ID8 derivatives: WT, p53-KO, p53-R172H and R270H. Shown are differentially expressed genes (DEGs) and their hierarchical clustering. (B) The presented Volcano plot shows all top DEGs that differed between WT and p53-R270H cells. The vertical lines mark log2 fold change of −1 and +1. (C) A comparative PROGENy analysis, which infers pathway activities from transcriptomics, was performed on the data shown in B. Note activation of the JAK/STAT and TGFβ signaling, as well as downregulation of Cdh1 (encoding E-cadherin) and the TNF-related apoptosis-inducing ligand (TRAIL) in ID8 p53-R270H cells. (D and E) The histogram on the left side indicates the top six canonical activated or inhibited pathways (absolute z-scores of 1 and −1, respectively) in ID8 p53-R270H as compared to wildtype ID8 cells. The right panel shows chord plots representing the same set of significant canonical pathways and displaying the relationship between individual pathways and relevant genes. (F) Immunoblotting analysis was performed with the indicated derivatives of ID8 cells to validate differential expression of the indicated EMT markers. GAPDH was used as the loading control. (G) Gene set enrichment analysis (GSEA) was applied on the DEGs of p53-R175H and p53-R273H mutated OvCa patients included in the TCGA dataset with both the KEGG pathways, as well as Hallmark gene sets. Shown is a positive enrichment score of the EMT related genes indicating that EMT related gene sets are altered at the top of the ranked list in the R273H cohort when compared to the R175H cohort. The grey lines are for individual hits, and the blue line represents the overall trend. (H) The indicated derivatives of ID8 cells were tested for ROS levels using either DCFH-DA or DHE, which quantify H₂O₂ and O₂⁻, respectively. Shown are representative photomicrographs for one out of three experiments. Original magnification, 20x. Scale bar: 100-pixel size. The histograms summarize ROS quantification. *, p<0.05, **, p<0.01, ***, p<0.001.

Figure 3:. The two missense p53 mutants differentially control redox potential and Nrf2 activation.

(A) Transcriptional profiles of redox regulatory genes displayed by the indicated ID8 sublines, including derivatives expressing p53-R270H and p53-R172H. The enzymatic antioxidant genes are functionally segregated, and all genes are classified according to their control by Nrf2. (B) The indicated ID8 sub-lines were seeded on coverslips and incubated overnight. Thereafter, the cells were fixed and probed using phalloidin or an anti-Nrf2 antibody, followed by a secondary antibody. DAPI was used to stain nuclei. Shown are representative images corresponding to Nrf2 immunofluorescence (green), phalloidin (red) and DAPI (blue). Bar, 20 μm. Images were captured using a confocal microscope. Nrf2 staining was quantified and normalized using ImageJ. *, p< 0.05, **, p< 0.01, ***, p<0.001. (C) Whole extracts of the indicated ID8 sublines were analyzed using immunoblotting for redox and autophagy markers. Vinculin was used as a gel loading control. (D) (left panel) The indicated ID8 derivatives (3×10³ cells) were seeded in 96-well plates and treated for 72 hours with increasing concentrations of Cisplatin. Cell viability was measured using the MTT (3(4,5 dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide) assay. Averages +S.D. of quadruplicates are shown. This experiment was repeated thrice. (right panel) ID8 derivatives were seeded in 96-well plates (500 cells/well) and treated with Cisplatin (3 μM). Media were refreshed once every three days. After nine days, cells were fixed and stained with crystal violet. The results of one of three experiments are shown. Note that the p53^−/− and the R270H lines gained resistance to CDDP. (E) Shown are the results of a Gene Set Enrichment Analysis (GSEA), which was applied on the differentially expressed genes between p53-R175H and p53-R273H ovarian tumors of the TCGA patient dataset (with both the KEGG pathways and the Hallmark gene sets). Note upregulation of the NF-κB pathway, mediated by TNF-alpha signaling, in the p53-R273H patients.

Figure 4:. Blocking either PD-L1 or AREG extends survival of mice bearing ID8 tumors, especially tumors expressing p53-R270H.

(A) The TCGA ovarian cancer dataset was divided into two equal groups, high and low PD-L1. Shown are the respective survival curves. (B) All p53-positive patients from A were divided according to PD-L1 expression levels and the respective patient survival curves are presented. Note the reciprocal trends of survival curves in A and B. (C) C57/Black female mice were intraperitoneally injected with 5×10⁶ ID8 cells, either WT (left panel) or p53(−/−) cells expressing a p53-R270H allele (right panel). Ten days later, mice were randomized into two groups. One group was treated with saline whereas the other received avelumab (human anti-PD-L1; 0.4 mg per injection), twice weekly starting on day ten and ending on day twenty-five. Animal survival curves are shown along with statistical analysis (**, p<0.01; ***, p<0.001). (D) C57/Black female mice were intraperitoneally injected with 5×10⁶ ID8 cells, either WT (left panel) or p53(−/−) cells expressing an allele of p53-R270H (right panel). Ten days later, mice were randomized into two groups. One group was treated with saline whereas the other was treated with an anti-AREG antibody (AR37; 0.2 mg/injection) from day 10 post tumor cell inoculation till day 40. Shown are the survival curves and median days gained (***, p<0.001; ****, p<0.0001). (E) A schematic model of the combinatorial treatment of p53-R270H ovarian cancer model using an anti-AREG antibody (AR37) in combination with an engineered anti-PD-L1 murine antibody (Avelumab).

Figure 5:. Combining anti-AREG and anti-mPD-L1 antibodies significantly prolongs survival of animals harboring p53-R270H tumors, likely by means of enhanced ADCC.

(A) C57/Black female mice were intraperitoneally injected with p53^−/−-ID8 cells (5×10⁶) expressing p53-R270H. Ten days later, mice were randomized into three treatment groups, each comprising 13 mice. Group 1 was left untreated, Group 2 received a commercially available mouse anti-PD-1 antibody (0.2 mg per injection) and Group 3 was intraperitoneally treated with an anti-PD-L1 antibody (from BioXcell; 0.2mg per injection). All treatments were terminated 40 days later (total, 10 injections). Shown are animal survival curves, along with the median survival gains and log-rank test scores. ****, p<0.0001; n.s., not significant. (B) C57/Black female mice were intraperitoneally injected with p53^−/−-ID8 cells (5×10⁶) expressing p53-R270H. Ten days later, mice were randomized into three treatment groups, each comprising 13 mice. Group 1 was left untreated, Group 2 received a commercially available mouse anti-PD-1 antibody (0.2mg per injection) and Group 3 was intraperitoneally treated with anti-PD-L1 antibody (from BioXcell: 0.2 mg per injection). All treatments terminated 30 days later (total, 10 injections). Shown are animal survival curves, along with the median survival gains and log-rank test scores. ****, p<0.0001. (C and D) Mouse peripheral blood mononuclear cells (PBMC) were isolated from bone marrows of 8 weeks old C57/Black female mice and co-cultured for 8 hours with either ID8 WT (panel C) cells or p53^−/− cells expressing p53-R270H (panel D), in the absence or presence of the indicated antibodies: AR37 and the engineered mouse Avelumab antibody. Note that mAbs tested in isolation were used at 30 μg/ml. When in combination, each mAb was used at 15 μg/ml. To determine ADCC efficacy, we assayed LDH release in triplicates using a kit (from Promega) that measured cell killing. The summary histogram is shown. The experiment was repeated 3 times. (E and F) C57/Black female mice were intraperitoneally injected with ID8 cells (5×10⁶), either WT or p53^−/− cells expressing p53-R270H. Ten days later, mice were randomized into four groups per cell line (13 mice per group). One group from each cell line was left untreated, whereas the other groups received mouse Avelumab (0.2 mg/injection), AR37 (0.2 mg/injection) or the combination of mAvelumab and AR37 (0.1 mg/injection, each). Shown are survival curves and median survival gains with a log-rank test score. *, p<0.05; **, p<0.01; ***, p< 0.001, ****, p<0.0001.

Figure 6:. Treatment of patient-derived OvCA organoids with a combination of anti PD-L1 and anti-AREG antibodies reveals extensive CD8+ T cell recruitment, which is confirmed using a murine co-culture model.

(A) Tabular description of the PDO models selected for the coculture experiment. (B) Cytometry-based estimation of the ratio between CD8+ and CD4+ T cells in co-cultures. Patient-matched PBMC were incubated for 5 days with three different PDOs. The incubations were performed in the presence of indicated antibodies. After the fifth day, the organoid-PBMC cultures were dissociated and subjected to flow cytometry with antibodies specific for CD8+ and CD4+, and the event count for each of these cell types was taken to calculate the ratio. The PBMC only used healthy PBMC. Note that PDO5 and PDO6 express p53-R273C. (C) Shown are results of an imaging-based apoptosis assay depending on detection of caspase-3/7 activation and a fluorogenic substrate. The indicated PDOs were co-cultured with human PBMC (3×10⁵ cells) for 5 days and treated with specific antibodies (15 micrograms per millilitre). (D) Quantification of the caspase activation signals only from the PDO cells is provided in histograms. Note that Avelumab displayed an apoptotic effect toward the chemo-sensitive PDO5. Statistical analysis used two-way ANOVA with a log-rank test score. **, p<0.01, ***, p<0.001 and ****, p<0.0001 is shown. Scale bar, 100 μm. (E) Mouse bone marrow derived cells (BMDCs) were co-cultured for 2 hours with ID8 isogenic cell lines, either wildtype or p53(−/−) R270H cells. Afterwards, cells were stained for surface or nuclear antigens to identify the following T cell subtypes using flow cytometry: naïve and effector memory (cluster 0), exhausted (cluster 1), early differentiated (cluster 2), memory (cluster 3) and effector T cells (cluster 4). (F and G) Boxplots showing the fractions of exhausted CD8+ T cells (F; cluster 1) and effector CD8+ T cells (G; cluster 4) across the indicated untreated or antibody-treated BMDC, which were cultured either alone or together with murine ID8 cells, either wildtype or p53(−/−) R270H cells.

Figure 7:. Concurrent blockade of AREG and PD-L1 drives recruitment of CD8⁺ T cells and M1 macrophage to p53-R270H tumors.

(A) C57/Black female mice intraperitoneally injected with ID8 cells (5×10⁶), either WT or p53^−/− cells overexpressing p53-R273H. Ten days later, mice were randomized into four groups per cell line (5 mice per group). One group was left untreated, whereas the other groups received treatment with either mAvelumab (0.2 mg/injection), AR37 (0.2 mg/injection) or a combination of mAvelumab and AR37 (0.1 mg/injection, each). Ascites fluids were collected on day 75 post cell inoculation and processed for CyTOF analysis. The presented dot plot shows the differential association of the preselected immune cell markers with specific cell populations that were present in the ascites fluid of untreated mice bearing p53-R270H tumors or mice treated with the antibody combination. Dot size refers to the fraction of cells with marker expression above global marker median, and dot color marks the median abundance of the respective marker. Note the variation in markers of T cells, macrophages and antigen presenting cells. (B) viSNE plots of the results of CyTOF analysis of single cells in mouse ascites fluids using an antibody specific to CD274 (PD-L1). Note that wildtype ID8 cells are compared to p53-R270H in terms of the cellular response to two antibodies, AR37 (anti-AREG) and mAvelumab, as well as the combination of the two antibodies. The coloration is proportional to the expression intensity (red = high). Note downregulation of the PD-L1 cell population in the antibody combination group. (C) Shown are analyses of CD3+ T cells in ascites fluids. This analysis used manual gating. Note increased CD8-positive T cells following treatment of the p53-R270H cells with the combination of two antibodies. (D) Ascites fluids from two mice were used to overlay cytokine array membranes (from Proteome Profiler^™ Mouse Cytokine Arrays). Duplicate spots corresponding to 111 cytokines were assayed and the obtained data were normalized to the untreated group. Note antibody-induced changes in the abundance of cytokines involved in CTL recruitment (CCL2, CCL3, CCL5 and IL-17), CTL proliferation (Flt3 ligand, E-selectin, P-selectin, IL7, IL12 and IL15) and CTL activation (IL2, IL4 and IL6). (E) viSNE plots corresponding to the following macrophage markers: CD80, F4/80, CD206 and CD86. Note diminished abundance of M2 macrophage markers and concurrent upregulation of M1 markers upon treatment of the p53-R270H model. (F) Ascites fluids from two mice were used to overlay cytokine array membranes as in D. The obtained data were normalized to the untreated group. Note changes in the abundance of cytokines responsible for M1-like macrophage polarization (left colored box; CCL3 and CCL5), which were upregulated, along with downregulation of M2-promoting cytokines (right colored panel; CCL2, IL4, IL6, IL13 and GCSF). ****, p<0.0001.