Pharmacoproteomic characterisation of human colon and rectal cancer

Thesis

Colorectal cancer (CRC) is one of the top three most common cancers and among the top four causes of cancer-related deaths worldwide (Torre et al., 2015). CRC patients are well characterised on the transcriptome and proteome level, but proteomics data on representative cell lines as model systems for pre-clinical drug sensitivity studies lag behind. Here, label-free quantitative mass spectrometry was used to characterise the kinomes and full proteomes of 65 CRC cell lines, collectively termed the CRC65 cell line panel. This data was integrated with proteomics data on patient samples, as well as public transcriptome and drug sensitivity datasets, which were reanalysed from raw data in order to unify and streamline the data processing. Protein/mRNA ratios were constant across these datasets, enabling linear prediction of protein abundance from mRNA abundance after appropriate adjustment, which was used for mRNA-guided missing value imputation. An exploration of secondary imputation methods prompted the development of a complementary method for minimum-guided missing value imputation. Combining the proteomics datasets on cell lines and patients led to the discovery of integrated proteomic subtypes of CRC and enabled the identification of representative cell lines for each of them. Modelling publicly available dose-response data generated by four large-scale drug sensitivity studies as a function of kinome/full proteome profiles fuelled the prediction of drug sensitivity for cell lines and patients, allowed the identification of drugs differentially effective between the different integrated proteomic subtypes and revealed MERTK as a predictive biomarker for resistance towards MEK1/2 inhibitors. This predictive role of MERTK was subsequently confirmed using in vitro experiments, while immunohistochemistry of TMAs from 1,074 tumours generated as part of the QUASAR2 clinical trial unveiled that MERTK is also a prognostic biomarker in CRC. This dataset will be made available to the scientific community to facilitate the design of prospective clinical studies.

Authors: Frejno, M

• DOI: 10.5287/ora-xmaaabj6d
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: QUASAR 2; CPTAC; CRC65; proteomics; colorectal cancer; patient stratification; mass spectrometry
• Subjects: Oncology