A machine learning and directed network optimization approach to uncover TP53 regulatory patterns

TP53, the Guardian of the Genome, is the most frequently mutated gene in human cancers and the functional characterization of its regulation is fundamental. To address this we employ two strategies: machine learning to predict the mutation status of TP53from transcriptomic data, and directed regulatory networks to reconstruct the effect of mutations on the transcipt levels of TP53 targets. Using data from established databases (Cancer Cell Line Encyclopedia, The Cancer Genome Atlas), machine learning could predict the mutation status, but not resolve different mutations. On the contrary, directed network optimization allowed to infer the TP53 regulatory profile across: (1) mutations, (2) irradiation in lung cancer, and (3) hypoxia in breast cancer, and we could observe differential regulatory profiles dictated by (1) mutation type, (2) deleterious consequences of the mutation, (3) known hotspots, (4) protein changes, (5) stress condition (irradiation/hypoxia). This is an important first step toward using regulatory networks for the characterization of the functional consequences of mutations, and could be extended to other perturbations, with implications for drug design and precision medicine.

Adaptive sampling for nanopore direct RNA-sequencing

Nanopore long-read sequencing enables real-time monitoring and controlling of individual nanopores. This allows us to enrich or deplete specific sequences in DNA sequencing in a process called "adaptive sampling." So far, adaptive sampling (AS) was not applicable to the direct sequencing of RNA. Here, we show that AS is feasible and useful for direct RNA sequencing (DRS), which has its specific technical and biological challenges. Using a well-controlled in vitro transcript-based model system, we identify essential characteristics and parameter settings for AS in DRS, as the superior performance of depletion over enrichment. Here, the efficiency of depletion is close to the theoretical maximum. Additionally, we demonstrate that AS efficiently depletes specific transcripts in transcriptome-wide sequencing applications. Specifically, we applied our AS approach to poly(A)-enriched RNA samples from human-induced pluripotent stem cell-derived cardiomyocytes and mouse whole heart tissue and show efficient 2.5- to 2.8-fold depletion of highly abundant mitochondrial-encoded transcripts. Finally, we characterize depletion and enrichment performance for complex transcriptome subsets, that is, at the level of the entire Chromosome 11, proving the general applicability of direct RNA AS. Our analyses provide evidence that AS is especially useful to enable the detection of lowly expressed transcripts and reduce the sequencing of highly abundant disturbing transcripts.

Characterizing alternative splicing effects on protein interaction networks with LINDA

MOTIVATION

Alternative RNA splicing plays a crucial role in defining protein function. However, despite its relevance, there is a lack of tools that characterize effects of splicing on protein interaction networks in a mechanistic manner (i.e. presence or absence of protein-protein interactions due to RNA splicing). To fill this gap, we present Linear Integer programming for Network reconstruction using transcriptomics and Differential splicing data Analysis (LINDA) as a method that integrates resources of protein-protein and domain-domain interactions, transcription factor targets, and differential splicing/transcript analysis to infer splicing-dependent effects on cellular pathways and regulatory networks.

RESULTS

We have applied LINDA to a panel of 54 shRNA depletion experiments in HepG2 and K562 cells from the ENCORE initiative. Through computational benchmarking, we could show that the integration of splicing effects with LINDA can identify pathway mechanisms contributing to known bioprocesses better than other state of the art methods, which do not account for splicing. Additionally, we have experimentally validated some of the predicted splicing effects that the depletion of HNRNPK in K562 cells has on signalling.

Magnetique: an interactive web application to explore transcriptome signatures of heart failure

Background

Despite a recent increase in the number of RNA-seq datasets investigating heart failure (HF), accessibility and usability remain critical issues for medical researchers. We address the need for an intuitive and interactive web application to explore the transcriptional signatures of heart failure with this work.

Methods

We reanalysed the Myocardial Applied Genomics Network RNA-seq dataset, one of the largest publicly available datasets of left ventricular RNA-seq samples from patients with dilated (DCM) or hypertrophic (HCM) cardiomyopathy, as well as unmatched non-failing hearts (NFD) from organ donors and patient characteristics that allowed us to model confounding factors. We analyse differential gene expression, associated pathway signatures and reconstruct signaling networks based on inferred transcription factor activities through integer linear programming. We additionally focus, for the first time, on differential RNA transcript isoform usage (DTU) changes and predict RNA-binding protein (RBP) to target transcript interactions using a Global test approach. We report results for all pairwise comparisons (DCM, HCM, NFD).

Results

Focusing on the DCM versus HCM contrast (DCMvsHCM), we identified 201 differentially expressed genes, some of which can be clearly associated with changes in ERK1 and ERK2 signaling. Interestingly, the signs of the predicted activity for these two kinases have been inferred to be opposite to each other: In the DCMvsHCM contrast, we predict ERK1 to be consistently less activated in DCM while ERK2 was more activated in DCM. In the DCMvsHCM contrast, we identified 149 differently used transcripts. One of the top candidates is the O-linked N-acetylglucosamine (GlcNAc) transferase (OGT), which catalyzes a common post-translational modification known for its role in heart arrhythmias and heart hypertrophy. Moreover, we reconstruct RBP – target interaction networks and showcase the examples of CPEB1, which is differentially expressed in the DCMvsHCM contrast.

Conclusion

Magnetique (https://shiny.dieterichlab.org/app/magnetique) is the first online application to provide an interactive view of the HF transcriptome at the RNA isoform level and to include transcription factor signaling and RBP:RNA interaction networks. The source code for both the analyses (https://github.com/dieterich-lab/magnetiqueCode2022) and the web application (https://github.com/AnnekathrinSilvia/magnetique) is available to the public. We hope that our application will help users to uncover the molecular basis of heart failure.

Cancer-associated fibroblasts require proline synthesis by PYCR1 for the deposition of pro-tumorigenic extracellular matrix

Elevated production of collagen-rich extracellular matrix is a hallmark of cancer-associated fibroblasts (CAFs) and a central driver of cancer aggressiveness. Here we find that proline, a highly abundant amino acid in collagen proteins, is newly synthesized from glutamine in CAFs to make tumour collagen in breast cancer xenografts. PYCR1 is a key enzyme for proline synthesis and highly expressed in the stroma of breast cancer patients and in CAFs. Reducing PYCR1 levels in CAFs is sufficient to reduce tumour collagen production, tumour growth and metastatic spread in vivo and cancer cell proliferation in vitro. Both collagen and glutamine-derived proline synthesis in CAFs are epigenetically upregulated by increased pyruvate dehydrogenase-derived acetyl-CoA levels. PYCR1 is a cancer cell vulnerability and potential target for therapy; therefore, our work provides evidence that targeting PYCR1 may have the additional benefit of halting the production of a pro-tumorigenic extracellular matrix. Our work unveils new roles for CAF metabolism to support pro-tumorigenic collagen production.

Systems approach reveals distinct and shared signaling networks of the four PGE2 receptors in T cells

[Figure: see text].

PHONEMeS: Efficient Modeling of Signaling Networks Derived from Large-Scale Mass Spectrometry Data

Post-translational modifications of proteins play an important role in the regulation of cellular processes. The mass spectrometry analysis of proteome modifications offers huge potential for the study of how protein inhibitors affect the phosphosignaling mechanisms inside the cells. We have recently proposed PHONEMeS, a method that uses high-content shotgun phosphoproteomic data to build logical network models of signal perturbation flow. However, in its original implementation, PHONEMeS was computationally demanding and was only used to model signaling in a perturbation context. We have reformulated PHONEMeS as an Integer Linear Program (ILP) that is orders of magnitude more efficient than the original one. We have also expanded the scenarios that can be analyzed. PHONEMeS can model data upon perturbation on not only a known target but also deregulated pathways upstream and downstream of any set of deregulated kinases. Finally, PHONEMeS can now analyze data sets with multiple time points, which helps us to obtain better insight into the dynamics of the propagation of signals. We illustrate the value of the new approach on various data sets of medical relevance, where we shed light on signaling mechanisms and drug modes of action.

Converting networks to predictive logic models from perturbation signalling data with CellNOpt

Summary

The molecular changes induced by perturbations such as drugs and ligands are highly informative of the intracellular wiring. Our capacity to generate large datasets is increasing steadily. A useful way to extract mechanistic insight from the data is by integrating them with a prior knowledge network of signalling to obtain dynamic models. CellNOpt is a collection of Bioconductor R packages for building logic models from perturbation data and prior knowledge of signalling networks. We have recently developed new components and refined the existing ones to keep up with the computational demand of increasingly large datasets, including (i) an efficient integer linear programming, (ii) a probabilistic logic implementation for semi-quantitative datasets, (iii) the integration of a stochastic Boolean simulator, (iv) a tool to identify missing links, (v) systematic post-hoc analyses and (vi) an R-Shiny tool to run CellNOpt interactively.

Availability and implementation

R-package(s): https://github.com/saezlab/cellnopt.

Supplementary information

Supplementary data are available at Bioinformatics online.

Causal integration of multi-omics data with prior knowledge to generate mechanistic hypotheses

Multi-omics datasets can provide molecular insights beyond the sum of individual omics. Various tools have been recently developed to integrate such datasets, but there are limited strategies to systematically extract mechanistic hypotheses from them. Here, we present COSMOS (Causal Oriented Search of Multi-Omics Space), a method that integrates phosphoproteomics, transcriptomics, and metabolomics datasets. COSMOS combines extensive prior knowledge of signaling, metabolic, and gene regulatory networks with computational methods to estimate activities of transcription factors and kinases as well as network-level causal reasoning. COSMOS provides mechanistic hypotheses for experimental observations across multi-omics datasets. We applied COSMOS to a dataset comprising transcriptomics, phosphoproteomics, and metabolomics data from healthy and cancerous tissue from eleven clear cell renal cell carcinoma (ccRCC) patients. COSMOS was able to capture relevant crosstalks within and between multiple omics layers, such as known ccRCC drug targets. We expect that our freely available method will be broadly useful to extract mechanistic insights from multi-omics studies.

Abstract B76: Pyruvate dehydrogenase: A key to epigenetic regulation in CAFs

Cancer-associated fibroblasts (CAFs) play fundamental roles in cancer and are emerging as therapeutic target in tumors with extensive stromal regions and in those for which there are limited targeted therapies against the cancer cells, such as ovarian cancer. A unique feature of the CAFs is their ability to secrete abundant collagen-rich extracellular matrix (ECM) that promotes the desmoplastic reaction that accompanies tumor progression and drives tumor growth and metastasis. Altered tumor metabolism is a hallmark of cancer, and understanding whether and how metabolic pathways support protumorigenic and proinvasive CAF functions may identify ways to target these cells to effectively target tumors. Using global phosphoproteomics, we have found that the activity of the pyruvate dehydrogenase complex (PDC), which is the rate-limiting enzyme for the entry of glycolysis-derived metabolites into the TCA cycle by converting pyruvate into acetyl-CoA, is strongly increased in patient-derived CAFs compared to their normal fibroblast counterpart. Consistently, the expression of pyruvate dehydrogenase kinase (PDK), which phosphorylates and inhibits PDC activity, is downregulated in CAFs and in the stroma of tumor patient samples. We found that PDC activity in CAFs leads to increased acetyl-CoA production. Surprisingly, 13C-glucose tracing experiments showed that CAFs do not channel acetyl-CoA into the TCA cycle. Instead, CAFs use acetyl-CoA to activate an epigenetic switch triggered by acetylation of H3K27. H3K27 acetylation is a known marker of gene expression activation. In CAFs, it triggered the expression of several collagen genes. Interestingly, also the expression of enzymes of the proline synthesis pathway was induced following H3K27 acetylation. Collagens have an unusually high content of proline residues, and we show that enhanced proline synthesis is necessary to support the production of collagen-rich ECM in CAFs. Targeting the PDK/PDC pathway or H3K27 acetylation or the proline synthesis pathway was sufficient to inhibit collagen synthesis in CAFs in in vitro experiments. Targeting proline synthesis in the stroma was sufficient to reduce tumor growth in vivo. Our work provides a first evidence that metabolism and epigenetics are tightly intertwined in regulating CAF functions and that targeting the PDK/PDC pathway or the proline synthesis pathway in the stroma could halt the development of a desmoplastic reaction and tumor progression.

Citation Format: Emily Kay, Lisa Neilson, Claudia Boldrini, Juan Hernandez-Fernaud, Enio Gjerga, David Sumpton, Sandeep Dhayade, Grace McGregor, Grigorios Koulouras, Jurre Kamphorst, Karen Blyth, Julio Saez-Rodriguez, Gillian Mackay, Sara Zanivan. Pyruvate dehydrogenase: A key to epigenetic regulation in CAFs [abstract]. In: Proceedings of the AACR Special Conference on Advances in Ovarian Cancer Research; 2019 Sep 13-16, 2019; Atlanta, GA. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(13_Suppl):Abstract nr B76.

Elucidating essential kinases of endothelin signalling by logic modelling of phosphoproteomics data

Endothelins (EDN) are peptide hormones that activate a GPCR signalling system and contribute to several diseases, including hypertension and cancer. Current knowledge about EDN signalling is fragmentary, and no systems level understanding is available. We investigated phosphoproteomic changes caused by endothelin B receptor (ENDRB) activation in the melanoma cell lines UACC257 and A2058 and built an integrated model of EDNRB signalling from the phosphoproteomics data. More than 5,000 unique phosphopeptides were quantified. EDN induced quantitative changes in more than 800 phosphopeptides, which were all strictly dependent on EDNRB. Activated kinases were identified based on high confidence EDN target sites and validated by Western blot. The data were combined with prior knowledge to construct the first comprehensive logic model of EDN signalling. Among the kinases predicted by the signalling model, AKT, JNK, PKC and AMP could be functionally linked to EDN‐induced cell migration. The model contributes to the system‐level understanding of the mechanisms underlying the pleiotropic effects of EDN signalling and supports the rational selection of kinase inhibitors for combination treatments with EDN receptor antagonists.

From expression footprints to causal pathways: contextualizing large signaling networks with CARNIVAL

While gene expression profiling is commonly used to gain an overview of cellular processes, the identification of upstream processes that drive expression changes remains a challenge. To address this issue, we introduce CARNIVAL, a causal network contextualization tool which derives network architectures from gene expression footprints. CARNIVAL (CAusal Reasoning pipeline for Network identification using Integer VALue programming) integrates different sources of prior knowledge including signed and directed protein-protein interactions, transcription factor targets, and pathway signatures. The use of prior knowledge in CARNIVAL enables capturing a broad set of upstream cellular processes and regulators, leading to a higher accuracy when benchmarked against related tools. Implementation as an integer linear programming (ILP) problem guarantees efficient computation. As a case study, we applied CARNIVAL to contextualize signaling networks from gene expression data in IgA nephropathy (IgAN), a condition that can lead to chronic kidney disease. CARNIVAL identified specific signaling pathways and associated mediators dysregulated in IgAN including Wnt and TGF-β, which we subsequently validated experimentally. These results demonstrated how CARNIVAL generates hypotheses on potential upstream alterations that propagate through signaling networks, providing insights into diseases.