PPT1 is a negative regulator of STING signaling in cancer cells and its inhibition reactivates immune surveillance in cold tumors

Abstract

Immunotherapy modalities have revolutionized cancer treatment for a number of metastatic and treatment-refractory tumor types. Still, many malignancies that lack T cell infiltration and are termed immunologically "cold" fail to respond to these modalities. One approach to increase tumor immunogenicity has been to induce stimulator of interferon gene (STING) and downstream interferon signaling that is often dysregulated in cold tumors. Despite some early success of STING agonists in preclinical cancer models, these approaches have not been successful in the clinic due to poor tumor penetrance and systemic toxicities. Here, we performed a genome-wide CRISPR screen to uncover therapeutic targets to activate STING expression in human tumors. We identified the lysosomal hydrolase Palmitoyl Protein Thioesterase1 (PPT1) as a negative regulator of STING highly expressed in cold ovarian and prostate tumors. Genetic or pharmacological PPT1 suppression increased STING protein stability and its downstream activation of interferon and inflammatory cytokine signaling to enhance T cell migration. Treatment of preclinical prostate and ovarian cancer models expressing low levels of STING with the small molecule PPT1 inhibitor GNS561 enhanced STING expression and activation, leading to infiltration and activation of cytotoxic T cells that turned these tumors "hot" and reduced tumor growth, fibrosis, and dissemination without toxicity. Further analysis demonstrated that PPT1 is associated with reduced STING expression, CD8⁺ T cell numbers, overall survival, and immunotherapy outcomes in ovarian and prostate cancer patients. Thus, PPT1 inhibition may be a promising approach to activate STING and potentiate the effects of immunotherapy in cold tumors.

Keywords: Palmitoyl Protein Thioesterase 1; STING; cancer; inflammatory.

Figure 1.. CRISPR knockout screen identifies PPT1 as a negative regulator of STING in tumor cells.

(A) STING (TMEM173) mRNA expression across cancer cell lines analyzed using DepMap. (B) Box and whisker plots displaying STING mRNA levels in tumor or normal tissues from ovarian and prostate cancer patient samples from TCGA and GTEx databases (tumor samples in red, normal tissues in gray) (Ovarian: T=426, N=88 samples; Prostate: T=492, N=152 samples). *, P <0.05. (C) Schematic of the genome-wide CRISPR screening pipeline to identify regulators of STING expression in A1847 cells. (D) Candidates enriched in the STING^hi population were plotted based on B-scores. Y-axis represents the distribution of normalized enrichment scores for each gene. (E) Western blot analysis of STING protein in A1847 and PC3 cells following shRNA-mediated PPT1 knockdown. NS, non-silencing shRNA. (F) Representative histograms of STING levels in A1847 cells harboring non-silencing or PPT1 targeting shRNAs from flow cytometry analysis. (G) Box and whisker plots displaying PPT1 mRNA levels in tumor or normal tissues from ovarian and prostate cancer patient samples from TCGA and GTEx databases (tumor samples in red, normal tissues in gray) (Ovarian: T=426, N=88 samples; Prostate: T=492, N=152 samples). *, P <0.05. (H) Representative immunohistochemical (IHC) analysis of PPT1 expression in healthy and malignant ovaries or prostates of cancer patients. Scale bar, 50 μm. (I) Box and whisker plots displaying PPT1 mRNA levels in ovarian and prostate cancer patient samples from TCGA stratified based on STING expression, with STING^hi samples shown in red and STING^lo samples in gray (n=75 for ovarian and n=30 for prostate cancer samples). Data represent mean ± SEM. P values were calculated by two-tailed, unpaired Student’s t-test.

Figure 2.. PPT1 suppression stabilizes STING and activates its downstream activity.

(A) Western blots of A1847 cells treated with PPT1 inhibitor DC661 at indicated concentrations for 24 hours (hrs). (B) Western blots of A1847 and PC3 cells treated with PPT1 inhibitor GNS561 at indicated concentrations for 24 hrs. (C) Western blots of MycCAP;p53KO murine prostate tumor cells treated with GNS561 at indicated concentrations for 24 hrs. (D) Native PAGE gel of A1847 cells treated with vehicle (DMSO) or 1μM GNS561 for 24 hrs. (E) Western blots of A1847 and PC3 cells stably expressing PPT1 or control non-silencing shRNAs. (F) Western blots of A1847 and MycCAP;p53KO cells treated with GNS561 at indicated concentrations for 24 hrs. (G) Western blots of A1847 cells pre-treated with or without H-151 for 2 hrs, followed by treatment with 1μM GNS561 and/or 1μM Diabzi for 6 hrs. (H) Western blots of A1847 cells treated with cycloheximide (CHX; 50μg/ml) for indicated times. Quantification of STING band intensity relative to GAPDH loading control is shown. (I) Western blots of A1847 cells treated with 50ug/ml CHX, 1μM GNS561, 50nM Bafilomycin, and/or 1μM MG132 for 6 hrs. Quantification of STING band intensity relative to GAPDH loading control is shown. (J) Schematic of proposed mechanism by which PPT1 inhibition stabilizes STING protein and downstream signaling. Created in Biorender.com.

Figure 3.. PPT1 inhibition enhances STING-mediated pro-inflammatory signaling and T cell recruitment.

(A) KEGG pathway analysis of RNA-seq data of A1847 and PC3 cells treated with 5μM GNS561 as compared to control DMSO for 24 hours (n=2–3 samples/condition). (B) Heatmap showing GSEA of HALLMARK genesets in A1847 and PC3 cells treated as in A (n=2–3 samples/condition). (C) Heatmap showing expression of ISGs amongst the differentially expressed genes following GNS561 treatment from RNA-seq data of A1847 and PC3 cells treated as in A (n=2–3 samples/condition). (D) qRT-PCR analysis of pro-inflammatory cytokine gene expression in A1847 cells treated as in A (n=2 samples). A.U., arbitrary units. (E) Transcription factor enrichment analysis showing transcriptional regulators of targets differentially expressed in A1847 and PC3 cells following GNS561 treatment as in A (n= 2–3 samples/condition). (F) Schematic overview of in vitro CD8⁺ T cell migration assay using conditioned media from A1847 and PC3 cells treated with DMSO or 1μM GNS561 for 24 hours. (G) Quantification of number of CD8⁺ T cells that migrated toward conditioned media from DMSO or GNS561-treated A1847 and PC3 cells over 8 hrs (n=3 samples/group). (H) GSEA of antigen presentation and processing genes from RNA-seq data of A1847 and PC3 cells treated as in A (n=3 samples/group). NES, normalized enrichment score. FDR, false discovery rate. (I) Quantification of mean fluorescent intensity (MFI) of MHC-I expression on A1847 and PC3 cells treated with or without GNS561 for 24 hours (n=3 samples/group). Data represent ± SEM. P values were calculated by two-tailed, unpaired Student’s t-test (D,G,I) or weighted Kolmogorov–Smirnov test (H).

Figure 4.. PPT1 inhibition activates STING and CD8⁺ T cell immunity and leads to tumor growth suppression in a prostate cancer mouse model.

(A) Schematic of MycCAP;p53KO orthotopic transplant prostate cancer model and GNS561 treatment schedule in FVB mice. (B) Change in body weight of MycCAP;p53KO prostate tumor-bearing mice treated with vehicle or GNS561 (50mg/kg) for up to 2 weeks (n= 7–10 mice for Vehicle and 8 mice for GNS561). (C) Representative IHC staining and quantification of STING⁺ cells in MycCAP;p53KO prostate tumors from mice treated as in B for 4 days (n= 7 mice for vehicle and 8 mice for GNS561). Scale bar, 50μM. (D) qRT-PCR analysis of pro-inflammatory cytokine gene expression in MycCAP;p53KO prostate tumors from mice treated as in B for 4 days (n=3 samples/group). A.U, arbitrary units. (E) Cytokine array analysis of pro-inflammatory cytokines in MycCAP;p53KO prostate tumors from mice treated as in B for 4 days (n= 7 mice for vehicle and 8 mice for GNS561; Samples outside detectable range were excluded). (F) Flow cytometry analysis of total CD45⁺ immune cells, CD4⁺ and CD8⁺ T cells, and early CD8⁺ T cell activation in MycCAP;p53KO prostate tumors from mice treated as in B for 4 days (n= 7 mice for vehicle and 8 mice for GNS561). (G) Quantification of average distance of CD8⁺ T cells from center of MycCAP;p53KO prostate tumors from mice treated as in B for 4 days (n = 5 mice/group). (H) Flow cytometry analysis of T cell numbers and CD8⁺ T cell and DC activation markers in MycCAP;p53KO prostate tumors from mice treated as in B for 14 days (n= 10 mice for vehicle and 8 mice for GNS561). (I) Representative IHC staining and quantification of Granzyme B (GZMB)⁺ cells in MycCAP;p53KO prostate tumors from mice treated as in B for 14 days (n= 10 mice for vehicle and 8 mice for GNS561). Scale bar, 50 μM. (J) Representative IHC staining and quantification of SMA positivity in MycCAP;p53KO prostate tumors from mice treated as in B for 14 days (n= 10 mice for vehicle and 8 mice for GNS561). Scale bar, 50 μM. (K) Weights of MycCAP;p53KO prostate tumors from mice treated as in B for 14 days (n= 10 mice for vehicle and 8 mice for GNS561). Data represent ± SEM. P values were calculated by two-tailed, unpaired Student’s t-test.

Figure 5.. PPT1 inhibitor GNS561 potentiates STING-mediated anti-tumor immunity in a disseminated ovarian cancer mouse model.

(A) Schematic of BPPNM ovarian cancer transplant model and GNS561 treatment schedule in C57BL/6 mice. (B) Change in body weight over time of BPPNM ovarian cancer-bearing mice treated with vehicle or GNS561 (50 mg/kg) (n= 8 mice for vehicle and 9 mice for GNS561). (C) Representative IHC staining and quantification of STING positivity in BPPNM ovarian tumors from mice treated as in B for 10 days (n= 8 mice for vehicle and 9 mice for GNS561). Scale bar, 50μM. (D-E) Representative IHC staining and quantification of cleaved caspase-3 (CC3) (D) and Ki67 (E) positivity in BPPNM ovarian tumors from mice treated as in B for 10 days (n= 8 mice for vehicle and 9 mice for GNS561). Scale bars, 50 μM. (F-H) Flow cytometry analysis of CD8⁺ T cell (F), NK cell (G), and DC (H) numbers and activation in BPPNM ovarian tumors from mice treated as in B for 10 days (n=7 mice for vehicle and 9 mice for GNS561).(I) Representative IHC staining and quantification of GZMB⁺ cells in BPPNM ovarian tumors from mice treated as in B for 10 days (n= 8 mice for vehicle and 9 mice for GNS561). Scale bar, 50 μM. (J) Weights of BPPNM ovarian tumors after 10 day treatment as in B normalized to radiance from bioluminescent imaging of mice on Day 0 (n= 8 mice for vehicle and 9 mice for GNS561). Data represent ± SEM. P values were calculated by two-tailed, unpaired Student’s t-test.

Figure 6.. High PPT1 expression correlates with reduced CD8⁺ T cell infiltration and poor immunotherapy outcomes in cancer patients.

(A-B) Representative IHC staining of ovarian and prostate tumor patient samples with high or low PPT1 expression. Scale bar, 50 μM. Quantification of CD8⁺ T cell numbers in ovarian (n=16) and prostate (n=10) cancer patient samples stratified by STING or PPT1 expression into high and low groups is shown on right. (C) Kaplan–Meier survival curves of ovarian cancer patients stratified by TMEM173 (STING) and PPT1 mRNA expression. (D) Kaplan–Meier survival curve of pan-cancer patients treated with combined anti-PD-1 and anti-CTLA-4 ICB therapy stratified based on PPT1 mRNA expression. Data represent ± SEM. P values were calculated by two-tailed, unpaired Student’s t-test (A,B) or log-rank test (C,D).