Convergent alterations in the tumor microenvironment of MYC-driven human and murine prostate cancer.

The tissue microenvironment in prostate cancer is profoundly altered. While such alterations have been implicated in driving prostate cancer initiation and progression to aggressive disease, how prostate cancer cells and their precursors mediate those changes is unclear, in part due to the inability to longitudinally study the disease evolution in human tissues. To overcome this limitation, we performed extensive single-cell RNA-sequencing (scRNA-seq) and rigorous molecular pathology of the comparative biology between human prostate cancer and key time points in the disease evolution of a genetically engineered mouse model (GEMM) of prostate cancer. Our studies of human tissues, with validation in a large external data set, revealed that cancer cell-intrinsic activation of MYC signaling was the top up-regulated pathway in human cancers, representing a common denominator across the well-known molecular and pathological heterogeneity of human prostate cancer. Likewise, numerous non-malignant cell states in the tumor microenvironment (TME), including non-cancerous epithelial, immune, and fibroblast cell compartments, were conserved across individuals, raising the possibility that these cell types may be a sequelae of the convergent MYC activation in the cancer cells. To test this hypothesis, we employed a GEMM of prostate epithelial cell-specific MYC activation in two mouse strains. Cell communication network and pathway analyses suggested that MYC oncogene-expressing neoplastic cells, directly and indirectly, reprogrammed the TME during carcinogenesis, leading to the emergence of cascading cell state alterations in neighboring epithelial, immune, and fibroblast cell types that paralleled key findings in human prostate cancer. Importantly, among these changes, the progression from a precursor-enriched to invasive-cancer-enriched state was accompanied by a cell-intrinsic switch from pro-immunogenic to immunosuppressive transcriptional programs with coinciding enrichment of immunosuppressive myeloid and Treg cells in the immune microenvironment. These findings implicate activation of MYC signaling in reshaping convergent aspects of the TME of prostate cancer as a common denominator across the otherwise well-documented molecular heterogeneity of human prostate cancer.

Single-cell atlas of epithelial and stromal cell heterogeneity by lobe and strain in the mouse prostate.

BACKGROUND: Evaluating the complex interplay of cell types in the tissue microenvironment is critical to understanding the origin and progression of diseases in the prostate and potential opportunities for intervention. Mouse models are an essential tool to investigate the molecular and cell-type-specific contributions of prostate disease at an organismal level. While there are well-documented differences in the extent, timing, and nature of disease development in various genetically engineered and exposure-based mouse models in different mouse strains and prostate lobes within each mouse strain, the underlying molecular phenotypic differences in cell types across mouse strains and prostate lobes are incompletely understood. METHODS: In this study, we used single-cell RNA-sequencing (scRNA-seq) methods to assess the single-cell transcriptomes of 6-month-old mouse prostates from two commonly used mouse strains, friend virus B/NIH jackson (FVB/NJ) (N = 2) and C57BL/6J (N = 3). For each mouse, the lobes of the prostate were dissected (anterior, dorsal, lateral, and ventral), and individual scRNA-seq libraries were generated. In situ and pathological analyses were used to explore the spatial and anatomical distributions of novel cell types and molecular markers defining these cell types. RESULTS: Data dimensionality reduction and clustering analysis of scRNA-seq data revealed that basal and luminal cells possessed strain-specific transcriptomic differences, with luminal cells also displaying marked lobe-specific differences. Gene set enrichment analysis comparing luminal cells by strain showed enrichment of proto-Oncogene targets in FVB/NJ mice. Additionally, three rare populations of epithelial cells clustered independently of strain and lobe: one population of luminal cells expressing Foxi1 and components of the vacuolar ATPase proton pump (Atp6v0d2 and Atp6v1g3), another population expressing Psca and other stem cell-associated genes (Ly6a/Sca-1, Tacstd2/Trop-2), and a neuroendocrine population expressing Chga, Chgb, and Syp. In contrast, stromal cell clusters, including fibroblasts, smooth muscle cells, endothelial cells, pericytes, and immune cell types, were conserved across strain and lobe, clustering largely by cell type and not by strain or lobe. One notable exception to this was the identification of two distinct fibroblast populations that we term subglandular fibroblasts and interstitial fibroblasts based on their strikingly distinct spatial distribution in the mouse prostate. CONCLUSIONS: Altogether, these data provide a practical reference of the transcriptional profiles of mouse prostate from two commonly used mouse strains and across all four prostate lobes.

Phenotypic characterization of two novel cell line models of castration-resistant prostate cancer.

BACKGROUND: Resistance to androgen deprivation therapies is a major driver of mortality in advanced prostate cancer. Therefore, there is a need to develop new preclinical models that allow the investigation of resistance mechanisms and the assessment of drugs for the treatment of castration-resistant prostate cancer. METHODS: We generated two novel cell line models (LAPC4-CR and VCaP-CR) which were derived by passaging LAPC4 and VCaP cells in vivo and in vitro under castrate conditions. We performed detailed transcriptomic (RNA-seq) and proteomic analyses (SWATH-MS) to delineate expression differences between castration-sensitive and castration-resistant cell lines. Furthermore, we characterized the in vivo and in vitro growth characteristics of these novel cell line models. RESULTS: The two cell line derivatives LAPC4-CR and VCaP-CR showed castration-resistant growth in vitro and in vivo which was only minimally inhibited by AR antagonists, enzalutamide, and bicalutamide. High-dose androgen treatment resulted in significant growth arrest of VCaP-CR but not in LAPC4-CR cells. Both cell lines maintained AR expression, but exhibited distinct expression changes on the mRNA and protein level. Integrated analyses including data from LNCaP and the previously described castration-resistant LNCaP-abl cells revealed an expression signature of castration resistance. CONCLUSIONS: The two novel cell line models LAPC4-CR and VCaP-CR and their comprehensive characterization on the RNA and protein level represent important resources to study the molecular mechanisms of castration resistance.

AIM1 is an actin-binding protein that suppresses cell migration and micrometastatic dissemination.

A defining hallmark of primary and metastatic cancers is the migration and invasion of malignant cells. These invasive properties involve altered dynamics of the cytoskeleton and one of its major structural components ß-actin. Here we identify AIM1 (absent in melanoma 1) as an actin-binding protein that suppresses pro-invasive properties in benign prostate epithelium. Depletion of AIM1 in prostate epithelial cells increases cytoskeletal remodeling, intracellular traction forces, cell migration and invasion, and anchorage-independent growth. In addition, decreased AIM1 expression results in increased metastatic dissemination in vivo. AIM1 strongly associates with the actin cytoskeleton in prostate epithelial cells in normal tissues, but not in prostate cancers. In addition to a mislocalization of AIM1 from the actin cytoskeleton in invasive cancers, advanced prostate cancers often harbor AIM1 deletion and reduced expression. These findings implicate AIM1 as a key suppressor of invasive phenotypes that becomes dysregulated in primary and metastatic prostate cancer.

Humanizing π-class glutathione S-transferase regulation in a mouse model alters liver toxicity in response to acetaminophen overdose.

BACKGROUND: Glutathione S-transferases (GSTs) metabolize drugs and xenobiotics. Yet despite high protein sequence homology, expression of π-class GSTs, the most abundant of the enzymes, varies significantly between species. In mouse liver, hepatocytes exhibit high mGstp expression, while in human liver, hepatocytes contain little or no hGSTP1 mRNA or hGSTP1 protein. π-class GSTs are known to be critical determinants of liver responses to drugs and toxins: when treated with high doses of acetaminophen, mGstp1/2+/+ mice suffer marked liver damage, while mGstp1/2-/- mice escape liver injury. METHODOLOGY/PRINCIPAL FINDINGS: To more faithfully model the contribution of π-class GSTs to human liver toxicology, we introduced hGSTP1, with its exons, introns, and flanking sequences, into the germline of mice carrying disrupted mGstp genes. In the resultant hGSTP1+mGstp1/2-/- strain, π-class GSTs were regulated differently than in wild-type mice. In the liver, enzyme expression was restricted to bile duct cells, Kupffer cells, macrophages, and endothelial cells, reminiscent of human liver, while in the prostate, enzyme production was limited to basal epithelial cells, reminiscent of human prostate. The human patterns of hGSTP1 transgene regulation were accompanied by human patterns of DNA methylation, with bisulfite genomic sequencing revealing establishment of an unmethylated CpG island sequence encompassing the gene promoter. Unlike wild-type or mGstp1/2-/- mice, when hGSTP1+mGstp1/2-/- mice were overdosed with acetaminophen, liver tissues showed limited centrilobular necrosis, suggesting that π-class GSTs may be critical determinants of toxin-induced hepatocyte injury even when not expressed by hepatocytes. CONCLUSIONS: By recapitulating human π-class GST expression, hGSTP1+mGstp1/2-/- mice may better model human drug and xenobiotic toxicology.