Low tristetraprolin expression activates phenotypic plasticity and primes transition to lethal prostate cancer in mice

Abstract

Phenotypic plasticity is a hallmark of cancer and is increasingly realized as a mechanism of resistance to androgen receptor-targeted (AR-targeted) therapy. Now that many prostate cancer (PCa) patients are treated upfront with AR-targeted agents, it is critical to identify actionable mechanisms that drive phenotypic plasticity, to prevent the emergence of resistance. We showed that loss of tristetraprolin (TTP; gene ZFP36) increased NF-κB activation, and was associated with higher rates of aggressive disease and early recurrence in primary PCa. We also examined the clinical and biological impact of ZFP36 loss with co-loss of PTEN, a known driver of PCa. Analysis of multiple independent primary PCa cohorts demonstrated that PTEN and ZFP36 co-loss was associated with increased recurrence risk. Engineering prostate-specific Zfp36 deletion in vivo induced prostatic intraepithelial neoplasia, and, with Pten codeletion, resulted in rapid progression to castration-resistant adenocarcinoma. Zfp36 loss altered the cell state driven by Pten loss, as demonstrated by enrichment of epithelial-mesenchymal transition (EMT), inflammation, TNF-α/NF-κB, and IL-6-JAK/STAT3 gene sets. Additionally, our work revealed that ZFP36 loss also induced enrichment of multiple gene sets involved in mononuclear cell migration, chemotaxis, and proliferation. Use of the NF-κB inhibitor dimethylaminoparthenolide (DMAPT) induced marked therapeutic responses in tumors with PTEN and ZFP36 co-loss and reversed castration resistance.

Keywords: Cell biology; Mouse models; Oncology; Prostate cancer.

Figure 1. ZFP36/TTP and clinical outcomes.

(A) RNA- and IHC-based forest plots depicting ZFP36/TTP expression related to clinical outcomes (biochemical recurrence and disease-free survival) and risk of lethal PCa (case-control cohorts). TCGA PRAD, The Cancer Genome Atlas Prostate Adenocarcinoma data set; DFCI FIHC, Dana-Farber Cancer Institute Fluorescent IHC; HPHS-PHS, Health Professionals Follow-up Study and Physicians’ Health Study. (B) Upregulated and downregulated genes were identified by differential expression analysis of TCGA PRAD cases divided by lower-quartile expression of ZFP36. (C) Representative images of immunofluorescent (IF) staining for pan-cytokeratin (yellow) and basal (red) markers, as well as TTP (green) in human PCa used for expression analysis. Benign glands (arrowheads) costain for pan-cytokeratin and basal cocktails; tumor cells (arrows) demonstrate absent basal expression. Far right images display diffuse prostate tumor with absent TTP expression. (D) Kaplan-Meier survival analysis demonstrating that TTP deficiency, measured by protein expression (DFCI, refs. 40, 69) and ZFP36 mRNA expression (TCGA PRAD, ref. ; Taylor et al., ref. 34), results in shorter disease-free-survival, and even shorter disease-free survival in combination with PTEN deficiency.

Figure 2. ZFP36/TTP expression in response to enzalutamide.

(A) Representative images and quantification of TTP IHC staining intensity in DARANA patient tissues comparing treatment-naive and post-enzalutamide samples. ****P < 0.0001 (Fisher’s exact test). Scale bars: 100 μm. (B) ZFP36 expression from RNA-Seq before and after enzalutamide in the NCI and DARANA clinical studies. n = 36–52, ****P < 0.0001 (2-tailed paired-samples t test). (C) Correlation of normalized ZFP36 expression versus volume of post-treatment residual cancer burden (RCB) in pre- and post-enzalutamide samples. n = 36 patients (non-parametric Spearman’s correlation). (D) Representative H3K27 acetylation tracks at the ZFP36 locus from 2 DARANA patients, comparing pre- and post-enzalutamide samples. (E) Quantification of H3K27 acetylation signal at the ZFP36 locus before and after enzalutamide treatment (paired-samples t test).

Figure 3. Zfp36 loss accelerates progression of prostate cancer in Pten-null murine tumors.

(A) H&E staining of murine tumors highlighting morphological progression of wild-type (WT), Pten^f/f Zfp36^+/+ (Pten^–/–), Pten^f/f Zfp36^f/+ (Pten^–/– Zfp36^+/–), and Pten^f/f Zfp36^f/+ (Pten^–/– Zfp36^–/–) dorsolateral prostate tissue at 8, 18, and 38 weeks. Scale bars: 100 μm. (B) Comparative weight of dorsolateral and ventral prostate tissue in GEMMs at 18 and 38 weeks. n = 5 mice per genotype, **P < 0.005, ***P < 0.0005, ****P < 0.0001 (1-way ANOVA with Tukey’s post hoc). (C) Kaplan-Meier graphs from GEMM aging studies show that prostate-specific deletion of Zfp36 significantly reduces time to ethical endpoint in PCa driven by loss of Pten n = 10 mice per genotype (Mantel-Cox log-rank test).

Figure 4. Zfp36 loss increases an inflammatory prostate cancer phenotype in Pten-null murine tumors.

(A) GSEA from RNA-Seq of endpoint GEMM PCa tumors comparing Pten^–/– and Pten^–/– Zfp36^–/– GEMMs, highlighting positively and negatively enriched Hallmark pathways. (B) Phospho-p65 IF and Masson’s trichrome staining PCa in Pten^–/–, Pten^–/– Zfp36^+/–, and Pten^–/– Zfp36^–/– GEMM dorsolateral prostate tissue at 38 weeks, with corresponding quantification. Scale bars: 100 μm. n = 5 mice per genotype; each mouse has been assigned a unique symbol for comparison across IHC analyses; *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.0001 (1-way ANOVA with Tukey’s post hoc).

Figure 5. Increased metastatic potential occurs with Zfp36 loss in Pten-null murine tumors.

(A) GSEA from RNA-Seq of endpoint GEMM PCa tumors comparing Pten^–/– and Pten^–/– Zfp36^–/– GEMMs, highlighting significant positively and negatively enriched GOBP pathways. (B) Ki-67 IHC and Krt18 and αSMA IF staining PCa in Pten^–/–, Pten^–/– Zfp36^+/–, and Pten^–/– Zfp36^–/– GEMM dorsolateral prostate tissue at 38 weeks, with corresponding quantification. Increased tumor cell proliferation and basement membrane breakdown are observed with loss of Zfp36. n = 5 mice per genotype, *P < 0.05, **P < 0.005, ***P < 0.0005 (1-way ANOVA with Tukey’s post hoc). Scale bar: 100 μm. (C) Number of mice that displayed PCa cells in distant organs by recombination PCR in Pten^–/–, Pten^–/– Zfp36^+/–, and Pten^–/– Zfp36^–/– GEMMs. (D) Representative images and quantification of the full scan in FFPE sections of tumor-adjacent pelvic lymph nodes (LNs) in Pten^–/– and Pten^–/– Zfp36^–/– mice. LNs were stained by multiplex IHC for Epcam (green), AR (red), and synaptophysin (Syp; white). Positive cells identify epithelial/tumoral cells metastasizing LNs (n = 3 mice per genotype). Scale bar: 50 μm. (E) Representative images and quantification of budding in GEMM-derived organoids highlighting increased invasive and metastatic potential of Pten^–/– Zfp36^–/– organoids. n = 5 unique organoid lines per genotype with experiment repeated once; experiments 1 and 2 are denoted by different symbols (technical replicates are underlaid in gray); ****P < 0.0001 (2-tailed Student’s t test). (F) Scratch assay in GEMM-derived 2D cells, comparing Pten^–/– and Pten^–/– Zfp36^–/– wound healing with that of Pten^–/– Rb1^–/–, a previously described metastatic, neuroendocrine PCa murine cell line (32). Representative images of scratch assay have been overlaid with areas identified as wound infiltrate in black. n = 1 cell line per genotype with experiment repeated twice; experiments 1, 2, and 3 are denoted by unique symbols; *P < 0.05, ***P < 0.001, ****P < 0.0001 (2-way ANOVA with Tukey’s post hoc).

Figure 6. Loss of Zfp36 induces phenotypic plasticity in Pten-null murine prostate tumors.

(A) AR, synaptophysin (Syp), and CD45 IF staining PCa in Pten^–/–, Pten^–/– Zfp36^+/–, and Pten^–/– Zfp36^–/– GEMM dorsolateral prostate tissue at 38 weeks, with corresponding quantification. Scale bars: 100 μm. n = 5 mice per genotype, *P < 0.05 (1-way ANOVA with Tukey’s post hoc). (B) Dual Krt8 and CD45 IF staining in Pten^–/– and Pten^–/– Zfp36^–/– GEMM dorsolateral prostate tissue at 38 weeks. Scale bars: 50 μm.

Figure 7. Zfp36 loss drives castration resistance in Pten-null murine tumors, which is counteracted with the NF-κB inhibitor DMAPT.

(A) Kaplan-Meier graph aged GEMMs (Mantel-Cox log-rank test). Whole prostate weights 12 weeks after castration. n = 5 mice per genotype, *P < 0.05, ***P < 0.0005 (1-way ANOVA with Tukey’s post hoc). (B) GEMM-derived organoid growth with and without enzalutamide (10 μM). n = 1 unique organoid line per genotype, repeated twice; experiments are denoted by unique symbols; *P < 0.05, ***P < 0.0005 (2-tailed multiple t test). (C) Allograft tumor growth in mice treated with DMAPT (100 mg/kg/d) or vehicle with or without surgical castration. n = 5 mice per treatment group. (D) Allograft endpoint tumor volumes. n = 5 mice per treatment group, *P < 0.05, **P < 0.005, ****P < 0.0001 (1-way ANOVA with Tukey’s post hoc). (E and F) Representative images (E) and quantification (F) of cell death (green) in GEMM-derived PCa organoids treated with DMAPT (5 μM), enzalutamide (10 μM), or the combination of both for 72 hours. n = 5 unique organoid lines per genotype, repeated once; experiments 1 and 2 are denoted by unique symbols (technical replicates are underlaid in gray); ****P < 0.0001 (1-way ANOVA with Tukey’s post hoc). Scale bars: 100 μm. (G) Flow cytometry quantification for CD45, synaptophysin, and AR expression in GEMM-derived PCa organoids treated with DMAPT (5 μM) or DMSO vehicle for 72 hours. n = 3 per genotype. *P < 0.05; **P < 0.005; ***P < 0.0005 (2-way ANOVA with Fisher’s least significant difference (LSD) test). (H) Fold change of Fkbp5 in GEMM-derived 2D cell lines treated with DMAPT (5 μM) or DMSO vehicle for 72 hours with or without R1881 (10 nM) stimulation. n = 1 organoid line per genotype, repeated twice; experiments 1, 2, and 3 are denoted by unique symbols; *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.0001 (1-way ANOVA with Tukey’s post hoc). (I) Schematic overview with ZFP36 intact, epithelial cells present with a luminal lineage phenotype and sensitivity to AR inhibition. In response to increased NF-κB expression, ZFP36/TTP increases as part of a negative-feedback loop. Loss of ZFP36 results in an alternative epithelial cell lineage phenotype, with uncontrolled NF-κB activation and reduced response to AR inhibition. DMAPT treatment restores a more luminal epithelial cell type and sensitivity to AR inhibition.