Abnormal protein phosphorylation is strongly associated with cancer aetiology. Considerable efforts have been undertaken to profile phosphorylation signalling networks acting through serine, threonine and tyrosine residues. Excitingly, we have evidence that other amino acids (histidine, aspartate, glutamate, lysine and arginine) can also be phosphorylated and thus are likely to contribute to cell signalling processes. However, until recently, the study of this so-called ‘non-canonical’ phosphorylation (NCP) has been severely limited due to lack of experimental tools, which we have now overcome. We have developed a method for unbiased phosphorylation site identification by mass spectrometry, discovering widespread human NCP on 1000s of sites. We are now exploring the complete phosphoproteome in key tumour cell models, including breast cancer, and as a function of cell cycle. In addition, there is a wealth of publicly available gene expression data that we can correlate to understand regulatory networks. At present, it is largely unknown how the coherent system signals through canonical and non-canonical kinase functions, and crucially why particular breast cancer sub-types respond poorly to therapy. In this project, we will integrate these data sets, including new data generated in the Eyers’ laboratory and in public repositories, using machine learning to improve predictions of kinases-substrate relationships, and to determine biomarkers for patient stratification. Our ability to generate data on human NCP sites, and correlate with transcriptional information, means that we are in a prime position to drive this new field of phosphorylation-mediated signalling, and understand the implications in disease progression.
1) Collate available datasets (transcriptomic/phosphoproteome) correlating changes across the cell cycle and in different breast cancer sub-types, to predict kinase-substrate relationships using machine learning techniques.
2) Use derived network information to predict key enzymes regulating NCP, which will be targeted for cellular perturbation and pathway validation.
3) Correlate identified sites of NCP with hotspots of cancer mutation with a view to pursuing future avenues for therapeutic intervention
Experimental Approach & Application:
The project will be largely computational, and thus would suit individuals with a background in bioinformatics / computational biology, or from engineering, computing, statistics or related background with a strong interest in biology. Students from a biological background (e.g. BSc in Biochemistry) keen to learn informatics methods are also encouraged to apply, and you will be given training in the relevant methods.
You will develop bioinformatics approaches and machine learning strategies (including e.g. neural networks), to define the relationship between dynamic NCP events and transcript profiles, as well as mutational hotspots in cancer. You will evaluate and validate potential NCP regulatory pathways in cancer models using small molecule inhibitors, siRNA and/or site specific CRIPSR/cas9 knock-in, and phosphoproteomics, either performing the experiments yourself or working in partnership with other team members in lab.
You will join an established collaboration between the two supervisors, Prof. Claire Eyers (Director, Centre for Proteome Research, https://www.liverpool.ac.uk/cpr/
; Twitter: @ClaireEEyers) and Prof. Andy Jones (Director, Computational Biology Facility, https://www.liverpool.ac.uk/computational-biology-facility/
; Twitter: @andy___jones), both at the University of Liverpool. Consequently, they will benefit from extensive support from both ‘wet’ and ‘dry’ scientists generating data on non-canonical phosphorylation and have access to cutting edge mass spectrometry technology. This project will expand considerably our understanding of these new kinase: substrate relationships and allow us to validate experimentally the hypotheses that arise from computational network approaches.