High-throughput elucidation of Wnt-dependent transcription factor grammar

Szczesnik, Tomasz (2018) High-throughput elucidation of Wnt-dependent transcription factor grammar. PhD thesis, Victor Chang Cardiac Research Institute & St Vincent's Clinical School, Faculty of Medicine, UNSW.

[img] Text
Tomasz Szczesnik 2018 PhD Thesis VCCRI UNSW.pdf
Available under License Creative Commons Attribution No Derivatives.

Download (10MB)


A fundamental limitation in our knowledge of transcription factor binding is that, despite knowing which DNA sequences they interact with, we cannot accurately predict where they bind in the genome. The likely explanation is that we are failing to account for the interactions between transcription factors that promote or inhibit each other's binding -- a process that would also explain cell-type specific binding differences, which occur despite no change in sequence affinity or genome. This should manifest itself as a 'grammar': a logic over how the organisation of transcription factor binding sequences promotes certain interactions and influences binding.In this thesis we develop a new high-throughput assay to discover such a grammar by measuring transcription factor binding to thousands of synthetic DNA sequences that are integrated into a specific genomic location. This circumvents the limitations of existing approaches for studying transcription factor binding. Unlike genomic binding data, we can be sure that any binding differences are due to underlying sequences changes, and not because of more distal features or genomic position. On the other hand, we still maintain the intracellular environment with its numerous protein interactions and chromatin structure, which is lost in /in-vitro/ experiments. For assaying transcription factor binding we use DamID, where the enzyme DNA adenine methyltransferase is fused to a transcription factor and labels sites of interaction. This is a notoriously noisy technique, leading us to a mutational screen of Dam which found several variants that substantially improve its sensitivity and spatial resolution for protein-DNA interactions. We also develop a statistical method for normalising PCR duplicates based on the dropout rate and accounting for sampling variability.We apply these techniques to the Wnt-dependent transcription factor Tcf7l2, showing that the local sequence is indeed responsible for situations where the Tcf7l2 motif does not predict binding. Furthermore, we find that certain cofactor interactions promote Tcf7l2 binding, particularly when they bind within closed chromatin. Finally, by extending a linear model with smooth spatially dependent interactions between motifs we show that these cofactor interactions appear to depend on their relative positioning.

Item Type: Thesis (PhD )
Additional Information: SUPERVISORS: Joshua, Ho, Victor Chang Cardiac Research Institute, Faculty of Medicine, UNSW; Richard, Harvey, Victor Chang Cardiac Research Institute, Faculty of Medicine, UNSW; Richard, Sherwood, Brigham and Women's Hospital.
Subjects: R Medicine > R Medicine (General)
Depositing User: Repository Administrator
Date Deposited: 16 Apr 2019 23:30
Last Modified: 16 Apr 2019 23:31
URI: http://eprints.victorchang.edu.au/id/eprint/828

Actions (login required)

View Item View Item