BioDiscViz: A visualization support and consensus signature selector for BioDiscML results.

Machine learning (ML) algorithms are powerful tools to find complex patterns and biomarker signatures when conventional statistical methods fail to identify them. While the ML field made significant progress, state of the art methodologies to build efficient and non-overfitting models are not always applied in the literature. To this purpose, automatic programs, such as BioDiscML, were designed to identify biomarker signatures and correlated features while escaping overfitting using multiple evaluation strategies, such as cross validation, bootstrapping and repeated holdout. To further improve BioDiscML and reach a broader audience, better visualization support and flexibility in choosing the best models and signatures are needed. Thus, to provide researchers with an easily accessible and usable tool for in depth investigation of the results from BioDiscML outputs, we developed a visual interaction tool called BioDiscViz. This tool provides summaries, tables and graphics, in the form of Principal Component Analysis (PCA) plots, UMAP, t-SNE, heatmaps and boxplots for the best model and the correlated features. Furthermore, this tool also provides visual support to extract a consensus signature from BioDiscML models using a combination of filters. BioDiscViz will be a great visual support for research using ML, hence new opportunities in this field by opening it to a broader community.

SHP-1 phosphatase acts as a coactivator of PCK1 transcription to control gluconeogenesis.

We previously reported that the protein-tyrosine phosphatase SHP-1 (PTPN6) negatively regulates insulin signaling, but its impact on hepatic glucose metabolism and systemic glucose control remains poorly understood. Here, we use co-immunoprecipitation assays, chromatin immunoprecipitation sequencing, in silico methods, and gluconeogenesis assay, and found a new mechanism whereby SHP-1 acts as a coactivator for transcription of the phosphoenolpyruvate carboxykinase 1 (PCK1) gene to increase liver gluconeogenesis. SHP-1 is recruited to the regulatory regions of the PCK1 gene and interacts with RNA polymerase II. The recruitment of SHP-1 to chromatin is dependent on its association with the transcription factor signal transducer and activator of transcription 5 (STAT5). Loss of SHP-1 as well as STAT5 decrease RNA polymerase II recruitment to the PCK1 promoter and consequently PCK1 mRNA levels leading to blunted gluconeogenesis. This work highlights a novel nuclear role of SHP-1 as a key transcriptional regulator of hepatic gluconeogenesis adding a new mechanism to the repertoire of SHP-1 functions in metabolic control.

Glucocorticoid stimulation induces regionalized gene responses within topologically associating domains.

Transcription-factor binding to cis-regulatory regions regulates the gene expression program of a cell, but occupancy is often a poor predictor of the gene response. Here, we show that glucocorticoid stimulation led to the reorganization of transcriptional coregulators MED1 and BRD4 within topologically associating domains (TADs), resulting in active or repressive gene environments. Indeed, we observed a bias toward the activation or repression of a TAD when their activities were defined by the number of regions gaining and losing MED1 and BRD4 following dexamethasone (Dex) stimulation. Variations in Dex-responsive genes at the RNA levels were consistent with the redistribution of MED1 and BRD4 at the associated cis-regulatory regions. Interestingly, Dex-responsive genes without the differential recruitment of MED1 and BRD4 or binding by the glucocorticoid receptor were found within TADs, which gained or lost MED1 and BRD4, suggesting a role of the surrounding environment in gene regulation. However, the amplitude of the response of Dex-regulated genes was higher when the differential recruitment of the glucocorticoid receptor and transcriptional coregulators was observed, reaffirming the role of transcription factor-driven gene regulation and attributing a lesser role to the TAD environment. These results support a model where a signal-induced transcription factor induces a regionalized effect throughout the TAD, redefining the notion of direct and indirect effects of transcription factors on target genes.

HSP70 mediates a crosstalk between the estrogen and the heat shock response pathways.

Cells respond to multiple signals from the environment simultaneously, which often creates crosstalk between pathways affecting the capacity to adapt to the changing environment. Chaperones are an important component in the cellular integration of multiple responses to environmental signals, often implicated in negative feedback and inactivation mechanisms. These mechanisms include the stabilization of steroid hormone nuclear receptors in the cytoplasm in the absence of their ligand. Here, we show using immunofluorescence, chromatin immunoprecipitation, and nascent transcripts production that the heat shock protein 70 (HSP70) chaperone plays a central role in a new crosstalk mechanism between the steroid and heat shock response pathways. HSP70-dependent feedback mechanisms are required to inactivate the heat shock factor 1 (HSF1) after activation. Interestingly, a steroid stimulation leads to faster accumulation of HSF1 in inactive foci following heat shock. Our results further show that in the presence of estrogen, HSP70 accumulates at HSF1-regulated noncoding regions, leading to deactivation of HSF1 and the abrogation of the heat shock transcriptional response. Using an HSP70 inhibitor, we demonstrate that the crosstalk between both pathways is dependent on the chaperone activity. These results suggest that HSP70 availability is a key determinant in the transcriptional integration of multiple external signals. Overall, these results offer a better understanding of the crosstalk between the heat shock and steroid responses, which are salient in neurodegenerative disorders and cancers.

The gut-liver axis: host microbiota interactions shape hepatocarcinogenesis.

Although their etiologies vary, tumors share a common trait: the control of an oncogenic transcriptional program that is regulated by the interaction of the malignant cells with the stromal and immune cells in the tumor microenvironment (TME). The TME shows high phenotypic and functional heterogeneity that may be modulated by interactions with commensal microbes (the microbiota) both systemically and locally. Unlike host cells, the microbiota adapts after environmental perturbations, impacting host-microbe interactions. In the liver, the bidirectional relationship in the gut and its associated microbiota creates an interdependent environment. Therefore, the gut microbiota and its metabolites modulate liver gene expression directly and indirectly, causing an imbalance in the gut-liver axis, which may result in disease, including carcinogenesis.

Cis-regulatory hubs: a new 3D model of complex disease genetics with an application to schizophrenia.

The 3D conformation of the chromatin creates complex networks of noncoding regulatory regions (distal elements) and promoters impacting gene regulation. Despite the importance of the role of noncoding regions in complex diseases, little is known about their interplay within regulatory hubs and implication in multigenic diseases such as schizophrenia. Here we show that cis-regulatory hubs (CRHs) in neurons highlight functional interactions between distal elements and promoters, providing a model to explain epigenetic mechanisms involved in complex diseases. CRHs represent a new 3D model, where distal elements interact to create a complex network of active genes. In a disease context, CRHs highlighted strong enrichments in schizophrenia-associated genes, schizophrenia-associated SNPs, and schizophrenia heritability compared with equivalent structures. Finally, CRHs exhibit larger proportions of genes differentially expressed in schizophrenia compared with promoter-distal element pairs or TADs. CRHs thus capture causal regulatory processes improving the understanding of complex disease etiology such as schizophrenia. These multiple lines of genetic and statistical evidence support CRHs as 3D models to study dysregulation of gene expression in complex diseases more generally.

Subversion of infiltrating prostate macrophages to a mixed immunosuppressive tumor-associated macrophage phenotype.

Tumor-associated macrophages (TAMs) support tumor progression within the tumor microenvironment (TME). Many questions remain as to the origin, development, and function of TAMs within the prostate TME. Evaluation of TAMs in prostate cancer (PCa) patients identified the immunosuppressive TAM marker CD163 in adjacent normal epithelium as an independent predictor of metastases or PCa death. Flow cytometry analyses identified prostate TAMs as frequently expressing both proinflammatory M1 (CCR7+) and immunosuppressive M2 (CD163+) markers. In vitro, we demonstrate PCa cells similarly subvert human M1 macrophages toward a mixed M1/M2 macrophage phenotype favoring tumor growth. Further the cytokine milieu-induced transition between immunosuppressive M2 to proinflammatory M1 (M2→M1) macrophages is abrogated by the presence of PCa cells. RNA sequencing suggests alterations in chemokine expression in prostate TAMs due to the presence of PCa cells. Together, our results suggest that prostate TAMs originate from inflammatory infiltrating macrophages, which are then reprogrammed mainly by PCa cells, but also the cytokine milieu. A better understanding of this subversion of macrophages within the prostate may lead to novel treatment strategies.

ZNF768 links oncogenic RAS to cellular senescence.

RAS proteins are GTPases that lie upstream of a signaling network impacting cell fate determination. How cells integrate RAS activity to balance proliferation and cellular senescence is still incompletely characterized. Here, we identify ZNF768 as a phosphoprotein destabilized upon RAS activation. We report that ZNF768 depletion impairs proliferation and induces senescence by modulating the expression of key cell cycle effectors and established p53 targets. ZNF768 levels decrease in response to replicative-, stress- and oncogene-induced senescence. Interestingly, ZNF768 overexpression contributes to bypass RAS-induced senescence by repressing the p53 pathway. Furthermore, we show that ZNF768 interacts with and represses p53 phosphorylation and activity. Cancer genomics and immunohistochemical analyses reveal that ZNF768 is often amplified and/or overexpressed in tumors, suggesting that cells could use ZNF768 to bypass senescence, sustain proliferation and promote malignant transformation. Thus, we identify ZNF768 as a protein linking oncogenic signaling to the control of cell fate decision and proliferation.

Modulating HSF1 levels impacts expression of the estrogen receptor α and antiestrogen response.

Master transcription factors control the transcriptional program and are essential to maintain cellular functions. Among them, steroid nuclear receptors, such as the estrogen receptor α (ERα), are central to the etiology of hormone-dependent cancers which are accordingly treated with corresponding endocrine therapies. However, resistance invariably arises. Here, we show that high levels of the stress response master regulator, the heat shock factor 1 (HSF1), are associated with antiestrogen resistance in breast cancer cells. Indeed, overexpression of HSF1 leads to ERα degradation, decreased expression of ERα-activated genes, and antiestrogen resistance. Furthermore, we demonstrate that reducing HSF1 levels reinstates expression of the ERα and restores response to antiestrogens. Last, our results establish a proof of concept that inhibition of HSF1, in combination with antiestrogens, is a valid strategy to tackle resistant breast cancers. Taken together, we are proposing a mechanism where high HSF1 levels interfere with the ERα-dependent transcriptional program leading to endocrine resistance in breast cancer.

Proximity-dependent Mapping of the Androgen Receptor Identifies Kruppel-like Factor 4 as a Functional Partner.

Prostate cancer (PCa) is the most frequently diagnosed cancer in men and the third cause of cancer mortality. PCa initiation and growth are driven by the androgen receptor (AR). The AR is activated by androgens such as testosterone and controls prostatic cell proliferation and survival. Here, we report an AR signaling network generated using BioID proximity labeling proteomics in androgen-dependent LAPC4 cells. We identified 31 AR-associated proteins in nonstimulated cells. Strikingly, the AR signaling network increased to 182 and 200 proteins, upon 24 h or 72 h of androgenic stimulation, respectively, for a total of 267 nonredundant AR-associated candidates. Among the latter group, we identified 213 proteins that were not previously reported in databases. Many of these new AR-associated proteins are involved in DNA metabolism, RNA processing, and RNA polymerase II transcription. Moreover, we identified 44 transcription factors, including the Kru¨ppel-like factor 4 (KLF4), which were found interacting in androgen-stimulated cells. Interestingly, KLF4 repressed the well-characterized AR-dependent transcription of the KLK3 (PSA) gene; AR and KLF4 also colocalized genome-wide. Taken together, our data report an expanded high-confidence proximity network for AR, which will be instrumental to further dissect the molecular mechanisms underlying androgen signaling in PCa cells.

Control of adipogenic commitment by a STAT3-VSTM2A axis.

White adipose tissue (WAT) is a dynamic organ that plays crucial roles in controlling metabolic homeostasis. During development and periods of energy excess, adipose progenitors are recruited and differentiate into adipocytes to promote lipid storage capability. The identity of adipose progenitors and the signals that promote their recruitment are still incompletely characterized. We have recently identified V-set and transmembrane domain-containing protein 2A (VSTM2A) as a novel protein enriched in preadipocytes that amplifies adipogenic commitment. Despite the emerging role of VSTM2A in promoting adipogenesis, the molecular mechanisms regulating Vstm2a expression in preadipocytes are still unknown. To define the molecular mechanisms controlling Vstm2a expression, we have treated preadipocytes with an array of compounds capable of modulating established regulators of adipogenesis. Here, we report that Vstm2a expression is positively regulated by PI3K/mTOR and cAMP-dependent signaling pathways and repressed by the MAPK pathway and the glucocorticoid receptor. By integrating the impact of all the molecules tested, we identified signal transducer and activator of transcription 3 (STAT3) as a novel downstream transcription factor affecting Vstm2a expression. We show that activation of STAT3 increased Vstm2a expression, whereas its inhibition repressed this process. In mice, we found that STAT3 phosphorylation is elevated in the early phases of WAT development, an effect that strongly associates with Vstm2a expression. Our findings identify STAT3 as a key transcription factor regulating Vstm2a expression in preadipocytes.NEW & NOTEWORTHY cAMP-dependent and PI3K-mTOR signaling pathways promote the expression of Vstm2a. STAT3 is a key transcription factor that controls Vstm2a expression in preadipocytes. STAT3 is activated in the early phases of WAT development, an effect that strongly associates with Vstm2a expression.

Defining the Transcriptional Ecosystem.

Fine tuning of the transcriptional program requires the competing action of multiple protein complexes in a well-organized environment. Genome folding creates proximity between genes, leading to accumulation of regulatory factors and formation of local microenvironments. Many roles of this complex organization controlling gene transcription remain to be explored. In this Perspective, we are proposing the existence of a transcriptional ecosystem equilibrium: a mechanism balancing transcriptional regulation between connected genes during environmental disturbances. This model is derived from chromosome architecture studies assigning genes to specific DNA structures and evidence establishing that the transcription machinery and coregulators create dynamic phase separation droplets surrounding active genes. Defining connected genes as ecosystems rather than individuals will cement that transcriptional regulation is a biochemical equilibrium and force a reassessment of direct and indirect responses to environmental disturbances.

Author Correction: Enhancer decommissioning by LSD1 during embryonic stem cell differentiation.

In this Letter, the western blot for LSD1 in the right panel of Fig. 2b ('TCP +') was inadvertently duplicated from the tubulin blot immediately below. The actual tubulin western blot shows the same result, with no significant change to the levels of tubulin (see Fig. 1 of this Amendment). In addition, the western blots for LSD1 and HDAC1 of Fig. 3b and c have been corrected to include vertical black lines to delineate the juxtaposition of lanes that were non-adjacent in the original blotting experiment (see Fig. 2 of this Amendment). Supplementary Figs. 4a, 6b and 9b have also been corrected to delineate non-adjacent lanes with vertical black lines (see Supplementary Information of this Amendment). The complete raw data images from these western blotting experiments can also be found in the Supplementary Information of this Amendment. The original Letter has not been corrected.

Connected Gene Communities Underlie Transcriptional Changes in Cornelia de Lange Syndrome.

Cornelia de Lange syndrome (CdLS) is a complex multisystem developmental disorder caused by mutations in cohesin subunits and regulators. While its precise molecular mechanisms are not well defined, they point toward a global deregulation of the transcriptional gene expression program. Cohesin is associated with the boundaries of chromosome domains and with enhancer and promoter regions connecting the three-dimensional genome organization with transcriptional regulation. Here, we show that connected gene communities, structures emerging from the interactions of noncoding regulatory elements and genes in the three-dimensional chromosomal space, provide a molecular explanation for the pathoetiology of CdLS associated with mutations in the cohesin-loading factor NIPBL and the cohesin subunit SMC1A NIPBL and cohesin are important constituents of connected gene communities that are centrally positioned at noncoding regulatory elements. Accordingly, genes deregulated in CdLS are positioned within reach of NIPBL- and cohesin-occupied regions through promoter-promoter interactions. Our findings suggest a dynamic model where NIPBL loads cohesin to connect genes in communities, offering an explanation for the gene expression deregulation in the CdLS.

FOXA and master transcription factors recruit Mediator and Cohesin to the core transcriptional regulatory circuitry of cancer cells.

Controlling the transcriptional program is essential to maintain the identity and the biological functions of a cell. The Mediator and Cohesin complexes have been established as central cofactors controlling the transcriptional program in normal cells. However, the distribution, recruitment and importance of these complexes in cancer cells have not been fully investigated. Here we show that FOXA and master transcription factors are part of the core transcriptional regulatory circuitry of cancer cells and are essential to recruit M ediator and Cohesin. Indeed, Mediator and Cohesin occupied the enhancer and promoter regions of actively transcribed genes and maintained the proliferation and colony forming potential. Through integration of publically available ChIP-Seq datasets, we predicted the core transcriptional regulatory circuitry of each cancer cell. Unexpectedly, for all cells investigated, the pioneer transcription factors FOXA1 and/or FOXA2 were identified in addition to cell-specific master transcription factors. Loss of both types of transcription factors phenocopied the loss of Mediator and Cohesin. Lastly, the master and pioneer transcription factors were essential to recruit Mediator and Cohesin to regulatory regions of actively transcribed genes. Our study proposes that maintenance of the cancer cell state is dependent on recruitment of Mediator and Cohesin through FOXA and master transcription factors.

metagene Profiles Analyses Reveal Regulatory Element's Factor-Specific Recruitment Patterns.

ChIP-Sequencing (ChIP-Seq) provides a vast amount of information regarding the localization of proteins across the genome. The aggregation of ChIP-Seq enrichment signal in a metagene plot is an approach commonly used to summarize data complexity and to obtain a high level visual representation of the general occupancy pattern of a protein. Here we present the R package metagene, the graphical interface Imetagene and the companion package similaRpeak. Together, they provide a framework to integrate, summarize and compare the ChIP-Seq enrichment signal from complex experimental designs. Those packages identify and quantify similarities or dissimilarities in patterns between large numbers of ChIP-Seq profiles. We used metagene to investigate the differential occupancy of regulatory factors at noncoding regulatory regions (promoters and enhancers) in relation to transcriptional activity in GM12878 B-lymphocytes. The relationships between occupancy patterns and transcriptional activity suggest two different mechanisms of action for transcriptional control: i) a "gradient effect" where the regulatory factor occupancy levels follow transcription and ii) a "threshold effect" where the regulatory factor occupancy levels max out prior to reaching maximal transcription. metagene, Imetagene and similaRpeak are implemented in R under the Artistic license 2.0 and are available on Bioconductor.

Mutant cohesin affects RNA polymerase II regulation in Cornelia de Lange syndrome.

In addition to its role in sister chromatid cohesion, genome stability and integrity, the cohesin complex is involved in gene transcription. Mutations in core cohesin subunits SMC1A, SMC3 and RAD21, or their regulators NIPBL and HDAC8, cause Cornelia de Lange syndrome (CdLS). Recent evidence reveals that gene expression dysregulation could be the underlying mechanism for CdLS. These findings raise intriguing questions regarding the potential role of cohesin-mediated transcriptional control and pathogenesis. Here, we identified numerous dysregulated genes occupied by cohesin by combining the transcriptome of CdLS cell lines carrying mutations in SMC1A gene and ChIP-Seq data. Genome-wide analyses show that genes changing in expression are enriched for cohesin-binding. In addition, our results indicate that mutant cohesin impairs both RNA polymerase II (Pol II) transcription initiation at promoters and elongation in the gene body. These findings highlight the pivotal role of cohesin in transcriptional regulation and provide an explanation for the typical gene dysregulation observed in CdLS patients.

ZFHX4 interacts with the NuRD core member CHD4 and regulates the glioblastoma tumor-initiating cell state.

Glioblastoma (GBM) harbors subpopulations of therapy-resistant tumor-initiating cells (TICs) that are self-renewing and multipotent. To understand the regulation of the TIC state, we performed an image-based screen for genes regulating GBM TIC maintenance and identified ZFHX4, a 397 kDa transcription factor. ZFHX4 is required to maintain TIC-associated and normal human neural precursor cell phenotypes in vitro, suggesting that ZFHX4 regulates differentiation, and its suppression increases glioma-free survival in intracranial xenografts. ZFHX4 interacts with CHD4, a core member of the nucleosome remodeling and deacetylase (NuRD) complex. ZFHX4 and CHD4 bind to overlapping sets of genomic loci and control similar gene expression programs. Using expression data derived from GBM patients, we found that ZFHX4 significantly affects CHD4-mediated gene expression perturbations, which defines ZFHX4 as a master regulator of CHD4. These observations define ZFHX4 as a regulatory factor that links the chromatin-remodeling NuRD complex and the GBM TIC state.

Multiple structural maintenance of chromosome complexes at transcriptional regulatory elements.

Transcription factors control cell-specific gene expression programs by binding regulatory elements and recruiting cofactors and the transcription apparatus to the initiation sites of active genes. One of these cofactors is cohesin, a structural maintenance of chromosomes (SMC) complex that is necessary for proper gene expression. We report that a second SMC complex, condensin II, is also present at transcriptional regulatory elements of active genes during interphase and is necessary for normal gene activity. Both cohesin and condensin II are associated with genes in euchromatin and not heterochromatin. The two SMC complexes and the SMC loading factor NIPBL are particularly enriched at super-enhancers, and the genes associated with these regulatory elements are especially sensitive to reduced levels of these complexes. Thus, in addition to their well-established functions in chromosome maintenance during mitosis, both cohesin and condensin II make important contributions to the functions of the key transcriptional regulatory elements during interphase.