Department of Developmental Biology and Cancer Research, Institute for Medical Research Israel-Canada, Faculty of Medicine, The Hebrew University of Jerusalem
Auto-generated profile, based on publicly available information
Irene Unterman
Department of Developmental Biology and Cancer Research, Institute for Medical Research Israel-Canada, Faculty of Medicine, The Hebrew University of Jerusalem
Cancer is a predominant disease across animals. We applied a comparative genomics approach to systematically characterize genes whose conservation levels correlate positively (PC) or negatively (NC) with cancer resistance estimates across 193 vertebrates. Pathway analysis reveals that NC genes are enriched for metabolic functions and PC genes in cell cycle regulation, DNA repair, and immune response, pointing to their corresponding roles in mediating cancer risk. We find that PC genes are less tolerant to loss-of-function (LoF) mutations, are enriched in cancer driver genes, and are associated with germline mutations that increase human cancer risk. Their relevance to cancer risk is further supported via the analysis of mouse functional genomics and cancer mortality of zoo mammals’ data. In sum, our study describes a cross-species genomic analysis pointing to candidate genes that may mediate human cancer risk.
Inactivating mutations in the Methyl-CpG Binding Protein 2 (MECP2) gene are the main cause of Rett syndrome (RTT). Despite extensive research into MECP2 function, no treatments for RTT are currently available. Here, we used an evolutionary genomics approach to construct an unbiased MECP2 gene network, using 1028 eukaryotic genomes to prioritize proteins with strong co-evolutionary signatures with MECP2. Focusing on proteins targeted by FDA-approved drugs led to three promising targets, two of which were previously linked to MECP2 function (IRAK, KEAP1) and one that was not (EPOR). The drugs targeting these three proteins (Pacritinib, DMF, and EPO) were able to rescue different phenotypes of MECP2 inactivation in cultured human neural cell types, and appeared to converge on Nuclear Factor Kappa B (NF-κB) signaling in inflammation. This study highlights the potential of comparative genomics to accelerate drug discovery, and yields potential new avenues for the treatment of RTT.
Global loss of DNA methylation in mammalian genomes occurs during aging and cancer, primarily in heterochromatin-associated Partially Methylated Domains (PMDs). It has previously been shown that local sequence context (100bp) has a strong influence on the rate of demethylation of individual CpG dinucleotides within PMDs. Here, we train a deep learning model to capture this sequence dependence, finding that methylation loss in healthy tissues and cancer can be predicted with high accuracy based on the 150bp surrounding a CpG. We use a published whole-genome map of the re-methylation rate of newly synthesized DNA during mitosis to show that CpGs with a “slow-loss” sequence context are efficiently re-methylated, while CpGs with a “fast-loss” sequence context are inefficiently re-methylated. Intriguingly, we find that the 10% most slow-loss CpGs lose almost no DNA methylation in healthy cell types, but lose significant DNA methylation in many cancer types. This finding suggests that loss of DNA methylation at slow-loss CpGs could underlie some cancer-specific transcriptional deregulation that has been linked to DNA hypomethylation, including the derepression of cancer antigens and transposable elements.
Background: Deconvolution is used to estimate the proportion of mixed cell types from tissue or blood samples based on genomic profiling. DNA methylation is commonly used because specific CpG positions reflect cell type identity and can be accurately measured at either the population or single-molecule level. Methylation sequencing techniques can profile multiple individual CpGs on a single DNA molecule, but few deconvolution models have been developed to exploit these single-molecule methylation haplotypes for cell type deconvolution. Results and Conclusions: We used simulated whole-genome methylation data and in silico mixtures of real data to compare existing deconvolution tools with two new models developed here. We found that adapting an existing model CelFiE to incorporate methylation haplotype information improved deconvolution accuracy by ∼30% over other tools, including the original CelFiE. In addition to overall higher accuracy, our new tool CelFiE Integrated Single-molecule Haplotypes (or CelFiE-ISH ) outperformed others in detecting rare cell types present at 0.1% and below. Detection of rare cell types is important for the analysis of circulating DNA, which we demonstrate using a patient-derived plasma sequencing dataset.Finally,we show that marker selection strategy has a strong effect on deconvolution accuracy, concluding that haplotype-aware deconvolution can take advantage of markers tailored for that purpose.
Liquid biopsy using ultra-low-pass whole-genome sequencing (ULP-WGS, ∼0.25x coverage) is a promising tool to detect circulating tumor DNA (ctDNA) for cancer management, and the use of the native Oxford Nanopore (ONT) sequencing platform adds DNA methylation to the set of detectable features. Here, we test the performance of methylation-based cell-type deconvolution in ULP-WGS samples from diverse epithelial malignancies and investigate several new computational strategies using our CelFiE-ISH deconvolution framework. We find that incorporating larger numbers of markers restricted to the epithelial cell lineage can reduce the cancer fraction limit of detection down to 1.7-3.1%, matching or exceeding the 3% floor of established copy-number alteration (CNA) benchmarks. Our study provides a useful strategy for analysis of ULP-WGS ONT data and indicates that marker selection remains a key challenge for analyzing methylation-based cancer datasets.
Inactivating mutations in the Methyl-CpG Binding Protein 2 (MECP2) gene are the main cause of Rett syndrome (RTT). Despite extensive research into MECP2 function, no treatments for RTT are currently available. Here we use an evolutionary genomics approach to construct an unbiased MECP2 gene network, using 1,028 eukaryotic genomes to prioritize proteins with strong co-evolutionary signatures with MECP2. Focusing on proteins targeted by FDA approved drugs led to three promising candidates, two of which were previously linked to MECP2 function (IRAK, KEAP1) and one that was not (EPOR). We show that each of these compounds has the ability to rescue different phenotypes of MECP2 inactivation in cultured human neural cell types, and appear to act on Nuclear Factor Kappa B (NF-κB) signaling in inflammation. This study highlights the potential of comparative genomics to accelerate drug discovery, and yields potential new avenues for the treatment of RTT. Abstract Figure: