During disease progression from primary towards metastatic prostate cancer (PCa), and in particular bone metastases, the tumor microenvironment (TME) evolves in parallel with the cancer clones, altering extracellular matrix composition (ECM), vasculature architecture, and recruiting specialized tumor-supporting cells that favor tumor spread and colonization at distant sites. We introduce the clinical profile of advanced metastatic PCa in terms of common genetic alterations. Findings from recently developed models of PCa metastatic spread are discussed, focusing mainly on the role of the TME (mainly matrix and fibroblast cell types), at distinct stages: premetastatic niche orchestrated by the primary tumor towards the metastatic site and bone metastasis. We report evidence of premetastatic niche formation, such as the mechanisms of distant site conditioning by extracellular vesicles, chemokines and other tumor-derived mechanisms, including altered cancer cell-ECM interactions. Furthermore, evidence supporting the similarities of stroma alterations among the primary PCa and bone metastasis, and contribution of TME to androgen deprivation therapy resistance are also discussed. We summarize the available bone metastasis transgenic mouse models of PCa from a perspective of pro-metastatic TME alterations during disease progression and give an update on the current diagnostic and therapeutic radiological strategies for bone metastasis clinical management.

Tumors strongly depend on their surrounding tumor microenvironment (TME) for growth and progression, since stromal elements are required to generate the optimal conditions for cancer cell proliferation, invasion, and possibly metastasis. Prostate cancer (PCa), though easily curable during primary stages, represents a clinical challenge in advanced stages because of the acquisition of resistance to anti-cancer treatments, especially androgen-deprivation therapies (ADT), which possibly lead to uncurable metastases such as those affecting the bone. An increasing number of studies is giving evidence that prostate TME components, especially cancer-associated fibroblasts (CAFs), which are the most abundant cell type, play a causal role in PCa since the very early disease stages, influencing therapy resistance and metastatic progression. This is highlighted by the prognostic value of the analysis of stromal markers, which may predict disease recurrence and metastasis. However, further investigations on the molecular mechanisms of tumor–stroma interactions are still needed to develop novel therapeutic approaches targeting stromal components. In this review, we report the current knowledge of the characteristics and functions of the stroma in prostate tumorigenesis, including relevant discussion of normal prostate homeostasis, chronic inflammatory conditions, pre-neoplastic lesions, and primary and metastatic tumors. Specifically, we focus on the role of CAFs, to point out their prognostic and therapeutic potential in PCa.

Understanding the spatial heterogeneity of tumours and its links to disease initiation and progression is a cornerstone of cancer biology. Presently, histopathology workflows heavily rely on hematoxylin and eosin and serial immunohistochemistry staining, a cumbersome, tissue-exhaustive process that results in non-aligned tissue images. We propose the VirtualMultiplexer, a generative artificial intelligence toolkit that effectively synthesizes multiplexed immunohistochemistry images for several antibody markers (namely AR, NKX3.1, CD44, CD146, p53 and ERG) from only an input hematoxylin and eosin image. The VirtualMultiplexer captures biologically relevant staining patterns across tissue scales without requiring consecutive tissue sections, image registration or extensive expert annotations. Thorough qualitative and quantitative assessment indicates that the VirtualMultiplexer achieves rapid, robust and precise generation of virtually multiplexed imaging datasets of high staining quality that are indistinguishable from the real ones. The VirtualMultiplexer is successfully transferred across tissue scales and patient cohorts with no need for model fine-tuning. Crucially, the virtually multiplexed images enabled training a graph transformer that simultaneously learns from the joint spatial distribution of several proteins to predict clinically relevant endpoints. We observe that this multiplexed learning scheme was able to greatly improve clinical prediction, as corroborated across several downstream tasks, independent patient cohorts and cancer types. Our results showcase the clinical relevance of artificial intelligence-assisted multiplexed tumour imaging, accelerating histopathology workflows and cancer biology.

DHX30 was recently implicated in the translation control of mRNAs involved in p53-dependent apoptosis. Here, we show that DHX30 exhibits a more general function by integrating the activities of its cytoplasmic isoform and of the more abundant mitochondrial one. The depletion of both DHX30 isoforms in HCT116 cells leads to constitutive changes in polysome-associated mRNAs, enhancing the translation of mRNAs coding for cytoplasmic ribosomal proteins while reducing the translational efficiency of the nuclear-encoded mitoribosome mRNAs. Furthermore, the depletion of both DHX30 isoforms leads to higher global translation but slower proliferation and lower mitochondrial energy metabolism. Isoform-specific silencing supports a role for cytoplasmic DHX30 in modulating global translation. The impact on translation and proliferation was confirmed in U2OS and MCF7 cells. Exploiting RIP, eCLIP, and gene expression data, we identified fourteen mitoribosome transcripts we propose as direct DHX30 targets that can be used to explore the prognostic value of this mechanism in cancer. We propose that DHX30 contributes to cell homeostasis by coordinating ribosome biogenesis, global translation, and mitochondrial metabolism. Targeting DHX30 could, thus, expose a vulnerability in cancer cells.

Due to their pivotal roles in tumor progression and therapy resistance, cancer-associated fibroblasts (CAF) are considered key therapeutic targets with loss of stromal androgen receptor (AR) a poorly understood hallmark of aggressive prostate cancer (PCa). A paucity of pre-clinical models however has hampered functional studies of CAF heterogeneity. We demonstrate that our newly-generated CAF biobank contains three FAP ⁺ -fibroblast subtypes, each with unique molecular and functional traits. Cultures with an early-activated phenotype expressed the highest levels of AR and exhibited AR-dependent growth. Consistently, stromal cells expressing early-activation markers co-expressed nuclear AR in clinical specimens and were enriched in pre-neoplastic lesions/low-grade PCa. Conversely, myofibroblastic CAF (myCAF), which expressed low AR levels in vitro and in vivo and were proliferatively-insensitive to AR signaling modulation, constituted the predominant CAF subpopulation in stromogenic high-grade PCa and castration-resistant LACP9 patient-derived xenografts. Exacerbation of the myCAF state upon castration of LAPC9-bearing hosts underscored these findings. Mechanistically, AR loss in myCAF was driven by an NFκB-TGFβ1-YAP1 axis, whose combined targeting synergistically repressed myofibroblastic hallmarks and impaired autophagic flux, effects that were potentiated by enzalutamide resulting in myCAF cell death. Collectively, these findings provide a mechanistic rationale for adjuvant targeting of the YAP1-TGFβ signaling axis to improve patient outcomes.

Understanding the spatial heterogeneity of tumors and its links to disease is a cornerstone of cancer biology. Emerging spatial technologies offer unprecedented capabilities towards this goal, but several limitations hinder their clinical adoption. To date, histopathology workflows still heavily depend on hematoxylin & eosin (H&E) and serial immunohistochemistry (IHC) staining, a cumbersome and tissue-exhaustive process that yields unaligned tissue images. We propose the VirtualMultiplexer, a generative AI toolkit that translates real H&E images to matching IHC images for several markers based on contrastive learning. The VirtualMultiplexer learns from unpaired H&E and IHC images and introduces a novel multi-scale loss to ensure consistent and biologically reliable stainings. The virtually multiplexed images enabled training a Graph Transformer that simultaneously learns from the joint spatial distribution of several markers to predict clinically relevant endpoints. Our results indicate that the VirtualMultiplexer achieves rapid, robust and precise generation of virtually multiplexed imaging datasets of high staining quality that are indistinguishable from the real ones. We successfully employed transfer learning to generate realistic virtual stainings across tissue scales, patient cohorts, and cancer types with no need for model fine-tuning. Crucially, the generated images are not only realistic but also clinically relevant, as they greatly improved the prediction of different clinical endpoints across patient cohorts and cancer types, speeding up histopathology workflows and accelerating spatial biology.