Androgen receptor splice variants drive castration-resistant prostate cancer metastasis by activating distinct transcriptional programs.

One critical mechanism through which prostate cancer (PCa) adapts to treatments targeting androgen receptor (AR) signaling is the emergence of ligand-binding domain-truncated and constitutively active AR splice variants, particularly AR-V7. While AR-V7 has been intensively studied, its ability to activate distinct biological functions compared with the full-length AR (AR-FL), and its role in regulating the metastatic progression of castration-resistant PCa (CRPC), remain unclear. Our study found that, under castrated conditions, AR-V7 strongly induced osteoblastic bone lesions, a response not observed with AR-FL overexpression. Through combined ChIP-seq, ATAC-seq, and RNA-seq analyses, we demonstrated that AR-V7 uniquely accesses the androgen-responsive elements in compact chromatin regions, activating a distinct transcription program. This program was highly enriched for genes involved in epithelial-mesenchymal transition and metastasis. Notably, we discovered that SOX9, a critical metastasis driver gene, was a direct target and downstream effector of AR-V7. Its protein expression was dramatically upregulated in AR-V7-induced bone lesions. Moreover, we found that Ser81 phosphorylation enhanced AR-V7's pro-metastasis function by selectively altering its specific transcription program. Blocking this phosphorylation with CDK9 inhibitors impaired the AR-V7-mediated metastasis program. Overall, our study has provided molecular insights into the role of AR splice variants in driving the metastatic progression of CRPC.

Increased AR expression in castration-resistant prostate cancer rapidly induces AR signaling reprogramming with the collaboration of EZH2.

Elevated androgen receptor (AR) expression is a hallmark of castration-resistant prostate cancer (CRPC) and contributes to the restoration of AR signaling under the conditions of androgen deprivation. However, whether overexpressed AR alone with the stimulation of castrate levels of androgens can be sufficient to induce the reprogramming of AR signaling for the adaptation of prostate cancer (PCa) cells remains unclear. In this study, we used a PCa model with inducible overexpression of AR to examine the acute effects of AR overexpression on its cistrome and transcriptome. Our results show that overexpression of AR alone in conjunction with lower androgen levels can rapidly redistribute AR chromatin binding and activates a distinct transcription program that is enriched for DNA damage repair pathways. Moreover, using a recently developed bioinformatic tool, we predicted the involvement of EZH2 in this AR reprogramming and subsequently identified a subset of AR/EZH2 co-targeting genes, which are overexpressed in CRPC and associated with worse patient outcomes. Mechanistically, we found that AR-EZH2 interaction is impaired by the pre-castration level of androgens but can be recovered by the post-castration level of androgens. Overall, our study provides new molecular insights into AR signaling reprogramming with the engagement of specific epigenetic factors.

Deep learning trained on hematoxylin and eosin tumor region of Interest predicts HER2 status and trastuzumab treatment response in HER2+ breast cancer.

The current standard of care for many patients with HER2-positive breast cancer is neoadjuvant chemotherapy in combination with anti-HER2 agents, based on HER2 amplification as detected by in situ hybridization (ISH) or protein immunohistochemistry (IHC). However, hematoxylin & eosin (H&E) tumor stains are more commonly available, and accurate prediction of HER2 status and anti-HER2 treatment response from H&E would reduce costs and increase the speed of treatment selection. Computational algorithms for H&E have been effective in predicting a variety of cancer features and clinical outcomes, including moderate success in predicting HER2 status. In this work, we present a novel convolutional neural network (CNN) approach able to predict HER2 status with increased accuracy over prior methods. We trained a CNN classifier on 188 H&E whole slide images (WSIs) manually annotated for tumor Regions of interest (ROIs) by our pathology team. Our classifier achieved an area under the curve (AUC) of 0.90 in cross-validation of slide-level HER2 status and 0.81 on an independent TCGA test set. Within slides, we observed strong agreement between pathologist annotated ROIs and blinded computational predictions of tumor regions / HER2 status. Moreover, we trained our classifier on pre-treatment samples from 187 HER2+ patients that subsequently received trastuzumab therapy. Our classifier achieved an AUC of 0.80 in a five-fold cross validation. Our work provides an H&E-based algorithm that can predict HER2 status and trastuzumab response in breast cancer at an accuracy that may benefit clinical evaluations.

Susceptibility-Associated Genetic Variation in NEDD9 Contributes to Prostate Cancer Initiation and Progression.

Although American men of European ancestry represent the largest population of patients with prostate cancer, men of African ancestry are disproportionately affected by prostate cancer, with higher prevalence and worse outcomes. These racial disparities in prostate cancer are due to multiple factors, but variations in genomic susceptibility such as SNP may play an important role in determining cancer aggressiveness and treatment outcome. Using public databases, we have identified a prostate cancer susceptibility SNP at an intronic enhancer of the neural precursor expressed, developmentally downregulated 9 (NEDD9) gene, which is strongly associated with increased risk of patients with African ancestry. This genetic variation increased expression of NEDD9 by modulating the chromatin binding of certain transcription factors, including ERG and NANOG. Moreover, NEDD9 displayed oncogenic activity in prostate cancer cells, promoting prostate cancer tumor growth and metastasis in vitro and in vivo. Together, our study provides novel insights into the genetic mechanisms driving prostate cancer racial disparities. SIGNIFICANCE: A prostate cancer susceptibility genetic variation in NEDD9, which is strongly associated with the increased risk of patients with African ancestry, increases NEDD9 expression and promotes initiation and progression of prostate cancer.See related commentary by Mavura and Huang, p. 3764.

Deep learning-based cross-classifications reveal conserved spatial behaviors within tumor histological images.

Histopathological images are a rich but incompletely explored data type for studying cancer. Manual inspection is time consuming, making it challenging to use for image data mining. Here we show that convolutional neural networks (CNNs) can be systematically applied across cancer types, enabling comparisons to reveal shared spatial behaviors. We develop CNN architectures to analyze 27,815 hematoxylin and eosin scanned images from The Cancer Genome Atlas for tumor/normal, cancer subtype, and mutation classification. Our CNNs are able to classify TCGA pathologist-annotated tumor/normal status of whole slide images (WSIs) in 19 cancer types with consistently high AUCs (0.995 ± 0.008), as well as subtypes with lower but significant accuracy (AUC 0.87 ± 0.1). Remarkably, tumor/normal CNNs trained on one tissue are effective in others (AUC 0.88 ± 0.11), with classifier relationships also recapitulating known adenocarcinoma, carcinoma, and developmental biology. Moreover, classifier comparisons reveal intra-slide spatial similarities, with an average tile-level correlation of 0.45 ± 0.16 between classifier pairs. Breast cancers, bladder cancers, and uterine cancers have spatial patterns that are particularly easy to detect, suggesting these cancers can be canonical types for image analysis. Patterns for TP53 mutations can also be detected, with WSI self- and cross-tissue AUCs ranging from 0.65-0.80. Finally, we comparatively evaluate CNNs on 170 breast and colon cancer images with pathologist-annotated nuclei, finding that both cellular and intercellular regions contribute to CNN accuracy. These results demonstrate the power of CNNs not only for histopathological classification, but also for cross-comparisons to reveal conserved spatial behaviors across tumors.

ModEx: A text mining system for extracting mode of regulation of transcription factor-gene regulatory interaction.

BACKGROUND: Transcription factors (TFs) are proteins that are fundamental to transcription and regulation of gene expression. Each TF may regulate multiple genes and each gene may be regulated by multiple TFs. TFs can act as either activator or repressor of gene expression. This complex network of interactions between TFs and genes underlies many developmental and biological processes and is implicated in several human diseases such as cancer. Hence deciphering the network of TF-gene interactions with information on mode of regulation (activation vs. repression) is an important step toward understanding the regulatory pathways that underlie complex traits. There are many experimental, computational, and manually curated databases of TF-gene interactions. In particular, high-throughput ChIP-Seq datasets provide a large-scale map or transcriptional regulatory interactions. However, these interactions are not annotated with information on context and mode of regulation. Such information is crucial to gain a global picture of gene regulatory mechanisms and can aid in developing machine learning models for applications such as biomarker discovery, prediction of response to therapy, and precision medicine. METHODS: In this work, we introduce a text-mining system to annotate ChIP-Seq derived interaction with such meta data through mining PubMed articles. We evaluate the performance of our system using gold standard small scale manually curated databases. RESULTS: Our results show that the method is able to accurately extract mode of regulation with F-score 0.77 on TRRUST curated interaction and F-score 0.96 on intersection of TRUSST and ChIP-network. We provide a HTTP REST API for our code to facilitate usage. Availibility: Source code and datasets are available for download on GitHub: https://github.com/samanfrm/modex.

Stochastic Modeling of Gene Regulation by Noncoding Small RNAs in the Strong Interaction Limit.

Noncoding small RNAs (sRNAs) are known to play a key role in regulating diverse cellular processes, and their dysregulation is linked to various diseases such as cancer. Such diseases are also marked by phenotypic heterogeneity, which is often driven by the intrinsic stochasticity of gene expression. Correspondingly, there is significant interest in developing quantitative models focusing on the interplay between stochastic gene expression and regulation by sRNAs. We consider the canonical model of regulation of stochastic gene expression by sRNAs, wherein interaction between constitutively expressed sRNAs and mRNAs leads to stoichiometric mutual degradation. The exact solution of this model is analytically intractable given the nonlinear interaction term between sRNAs and mRNAs, and theoretical approaches typically invoke the mean-field approximation. However, mean-field results are inaccurate in the limit of strong interactions and low abundances; thus, alternative theoretical approaches are needed. In this work, we obtain analytical results for the canonical model of regulation of stochastic gene expression by sRNAs in the strong interaction limit. We derive analytical results for the steady-state generating function of the joint distribution of mRNAs and sRNAs in the limit of strong interactions and use the results derived to obtain analytical expressions characterizing the corresponding protein steady-state distribution. The results obtained can serve as building blocks for the analysis of genetic circuits involving sRNAs and provide new insights into the role of sRNAs in regulating stochastic gene expression in the limit of strong interactions.

Identification of competing endogenous RNAs of the tumor suppressor gene PTEN: A probabilistic approach.

Regulation by microRNAs (miRNAs) and modulation of miRNA activity are critical components of diverse cellular processes. Recent research has shown that miRNA-based regulation of the tumor suppressor gene PTEN can be modulated by the expression of other miRNA targets acting as competing endogenous RNAs (ceRNAs). However, the key sequence-based features enabling a transcript to act as an effective ceRNA are not well understood and a quantitative model associating statistical significance to such features is currently lacking. To identify and assess features characterizing target recognition by PTEN-regulating miRNAs, we analyze multiple datasets from PAR-CLIP experiments in conjunction with RNA-Seq data. We consider a set of miRNAs known to regulate PTEN and identify high-confidence binding sites for these miRNAs on the 3' UTR of protein coding genes. Based on the number and spatial distribution of these binding sites, we calculate a set of probabilistic features that are used to make predictions for novel ceRNAs of PTEN. Using a series of experiments in human prostate cancer cell lines, we validate the highest ranking prediction (TNRC6B) as a ceRNA of PTEN. The approach developed can be applied to map ceRNA networks of critical cellular regulators and to develop novel insights into crosstalk between different pathways involved in cancer.