Forward genetic screening allows for the identification of any genes important for a particular biological process or phenotype. While the power of this approach is broadly agreed on, the efficacy of currently available tools limits the strength of conclusions drawn from these experiments. This thesis describes a method to optimize molecular tools for high-throughput screening, both for shRNA and sgRNA based reagents. Using large shRNA efficacy datasets, we first designed an algorithm predicting the potency of shRNAs based on sequence determinants. Combined with a novel shRNA backbone that further improves the processing of synthetic shRNAs, we built a library of potent shRNAs to reliably and efficiently knock-down any gene in the human and mouse genomes. We then went on to apply a similar approach to identify sgRNAs with increased activity. We complemented this with conservation and repair prediction to increase the likelihood of generating functional knock-outs. With these tools in hand, we constructed, sequence-verified and validated arrayed shRNA and sgRNA libraries targeting any protein coding gene in the human genome. These resources allow large-scale screens to be performed in a multiplexed or arrayed format in a variety of biological contexts. I have also applied these tools to identify therapeutic targets to circumvent cancer resistance to treatment in two different contexts. To overcome the shortfalls of single target therapy, I have developed multiplexed multidimensional shRNA screening strategy, where two genes are knocked down simultaneously in each cell. This strategy allows the identification of gene pairs that could be targeted in tandem to maximize therapeutic benefits. As a proof of concept, I have used it with a subset of druggable genes in melanoma cell lines. Moreover, we have applied our genome wide shRNA libraries to a different resistance context, stroma-mediated resistance to gemcitabine in PDAC. In this project, we performed screens in a PDAC-CAF coculture setting to try and identify cancer vulnerabilities specifically in the presence of stroma. Overall, the tools developed in this thesis allow for the efficient knockdown or knockout of any gene, both in an individual or combinatorial setting. Apart from providing a resource that will be useful for many fields, we have performed several proof-of-concept studies where we have applied our tools to identify potential cancer drug targets.
Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:744534 |
Date | January 2018 |
Creators | Erard, Nicolas Pascal Jean |
Contributors | Hannon, Gregory J. |
Publisher | University of Cambridge |
Source Sets | Ethos UK |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Source | https://www.repository.cam.ac.uk/handle/1810/271895 |
Page generated in 0.0016 seconds