NPIF Rutherford Fellowship

Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.

Gene regulation is of fundamental biological interest and has immediate medical relevance, as it is essential for normal development, and often disrupted in cancer. Gene regulation takes place in the context of chromatin states and 3D chromatin contacts. This field of study has recently been enriched by the application of machine learning/AI approaches to classify chromatin states and chromatin contacts, and by experimental evidence indicating that concepts of self-organised criticality and phase transitions can be productively applied to transitions between chromatin states and the formation of 3D chromatin contacts. This proposal builds on our recent discovery that cohesin and CTCF are selectively required for the regulation of genes that are dynamically expressed during cellular activation and differentiation. This includes genes programmed for up- or downregulated during mouse T cell differentiation, and genes that respond to inflammatory signals in mouse macrophages. Cohesin mutations are recurrent in human acute myeloid leukaemia (AML), and primary AML cells with cohesin mutations show reduced expression of inducible pro-inflammatory genes. As inflammatory signals promote the differentiation of myeloid cells, our data suggest a mechanism for the selection of cohesin mutations whereby AML cells with reduced expression of inflammatory genes evade differentiation in favour of self-renewal.Focus of this project is to understand the mechanisms that underlie the dependence of developmentally regulated and stimulus-responsive genes on genome organiser proteins. We will use molecular biology and genome informatics tools to assemble comprehensive maps of chromatin state (histone modifications, including numerous ChIP-seq data sets from ENCODE, chromatin accessibility, transcription factor binding, transcription) and chromatin contacts (Hi-C). We imagine that the dynamic regulation of inducible and developmental genes may rely on a spectrum of chromatin states that represent intermediates between full activation and maximal repression. These in-between states may be maintained by specific chromatin features (such as bivalency) and by active formation of chromatin contacts by cohesin and CTCF, which counteracts the stable segregation of active and silent chromatin regions. We will apply machine learning/AI approaches to classify these states and to identify features that predict dependency on genome organiser proteins. The results we will uncover new principles of genome organisation and transcriptional regulation and improve our understanding of cohesin mutations in cancer.