Molecular subtyping of cancer is a critical step towards more individualized therapy and provides important biological insights into cancer heterogeneity. Although gene expression signature-based classification has been widely demonstrated to be an effective approach in the last decade, the widespread implementation has long been limited by platform differences, batch effects, and the difficulty to classify individual patient samples. Here, we describe a novel supervised cancer classification framework, deep cancer subtype classification (DeepCC), based on deep learning of functional spectra quantifying activities of biological pathways. In two case studies about colorectal and breast cancer classification, DeepCC classifiers and DeepCC single sample predictors both achieved overall higher sensitivity, specificity, and accuracy compared with other widely used classification methods such as random forests (RF), support vector machine (SVM), gradient boosting machine (GBM), and multinomial logistic regression algorithms. Simulation analysis based on random subsampling of genes demonstrated the robustness of DeepCC to missing data. Moreover, deep features learned by DeepCC captured biological characteristics associated with distinct molecular subtypes, enabling more compact within-subtype distribution and between-subtype separation of patient samples, and therefore greatly reduce the number of unclassifiable samples previously. In summary, DeepCC provides a novel cancer classification framework that is platform independent, robust to missing data, and can be used for single sample prediction facilitating clinical implementation of cancer molecular subtyping.

During meiotic prophase I, spermatocytes must balance transcriptional activation with homologous recombination and chromosome synapsis, biological processes requiring extensive changes to chromatin state. We explored the interplay between chromatin accessibility and transcription through prophase I of mammalian meiosis by measuring genome-wide patterns of chromatin accessibility, nascent transcription, and processed mRNA. We find that Pol II is loaded on chromatin and maintained in a paused state early during prophase I. In later stages, paused Pol II is released in a coordinated transcriptional burst mediated by the transcription factors A-MYB and BRDT, resulting in ~3-fold increase in transcription. Transcriptional activity is temporally and spatially segregated from key steps of meiotic recombination: double strand breaks show evidence of chromatin accessibility earlier during prophase I and at distinct loci from those undergoing transcriptional activation, despite shared chromatin marks. Our findings reveal mechanisms underlying chromatin specialization in either transcription or recombination in meiotic cells.

Sexual differences have been observed in the onset and prognosis of human cardiovascular diseases, but the underlying mechanisms are not clear. Here, we found that zebrafish heart regeneration is faster in females, can be accelerated by estrogen and is suppressed by the estrogen-antagonist tamoxifen. Injuries to the zebrafish heart, but not other tissues, increased plasma estrogen levels and the expression of estrogen receptors, especially esr2a. The resulting endocrine disruption induces the expression of the female-specific protein vitellogenin in male zebrafish. Transcriptomic analyses suggested heart injuries triggered pronounced immune and inflammatory responses in females. These responses, previously shown to elicit heart regeneration, could be enhanced by estrogen treatment in males and reduced by tamoxifen in females. Furthermore, a prior exposure to estrogen preconditioned the zebrafish heart for an accelerated regeneration. Altogether, this study reveals that heart regeneration is modulated by an estrogen-inducible inflammatory response to cardiac injury. These findings elucidate a previously unknown layer of control in zebrafish heart regeneration and provide a new model system for the study of sexual differences in human cardiac repair.

The recent discovery of the cancer-associated E76K mutation in histone H2B (H2BE76-to-K) in several types of cancers revealed a new class of oncohistone. H2BE76K weakens the stability of histone octamers, alters gene expression, and promotes colony formation. However, the mechanism linking the H2BE76K mutation to cancer development remains largely unknown. In this study, we knock-in the H2BE76K mutation in MDA-MB-231 breast cancer cells using CRISPR/Cas9 and show that the E76K mutant histone H2B preferentially localizes to genic regions. Interestingly, genes upregulated in the H2BE76K mutant cells are enriched for the E76K mutant H2B and are involved in cell adhesion and proliferation pathways. We focused on one H2BE76K target gene, ADAM19 (A Disintegrin And Metalloproteinase domain-containing protein 19), a gene highly expressed in various human cancers including breast invasive carcinoma, and demonstrate that H2BE76K directly promotes ADAM19 transcription by facilitating efficient transcription along the gene body. ADAM19 depletion reduced the colony formation ability of the H2BE76K mutant cells whereas wildtype MDA-MB-231 cells overexpressing ADAM19 mimics the colony formation phenotype of the H2BE76K mutant cells. Collectively, our data demonstrate the mechanism by which H2BE76K deregulates the expression of genes that control oncogenic properties through a combined effect of its specific genomic localization and nucleosome destabilization effect.