Identifying unique cell states in early liver oncogenesis with transcription coupled repair

Abstract

Many cancers, such as the main form of primary liver cancer hepatocellular carcinoma (HCC), typically arise in tissues with a high number of genetic mutations which drive cell growth aberrantly. These mutations appear to precede tumour initiation, as DNA sequencing has shown that cancer driver mutations can be found in otherwise-healthy tissues at high levels. Though an increase in mutational burden increases the risk of cancer, not all cells with driver mutations will form tumours. There is a growing body of evidence that suggests tumour-initiating cells preferentially enter specific transcriptional states that interact with driver mutations to promote tumourigenesis. Characterising the aberrant transcriptional pathways of initiating cells could help elucidate the mechanisms for determining exactly which mutated cells will go on to form cancer.

Here I aim to identify transcriptional cell states which can predispose hepatocytes to tumorigenesis in the diethylnitrosamine (DEN) mouse model of HCC using a novel method of transcriptomic profiling. DEN introduces adducts onto DNA that lead to mutations, and the principal adduct on thymine can only be removed by the DNA repair process transcription coupled repair (TCR). TCR is closely linked to transcription and only transcribed sections of DNA are repaired, resulting in varying levels of mutations across the genome, with highly transcribed sections having low mutation burdens, while the inverse is true for non-transcribed sections.

By assessing the mutational burden across the genomes of established tumours from the DEN model, I infer historical transcriptional activity from tumour initiating cells immediately after DEN treatment. This analysis uncovered multiple genes associated with cilium assembly and cilium organisation preferentially upregulated in response to DEN. Genes associated with cell migration and adhesion were identified as downregulated after DEN treatment. I then validated these findings using bulk RNA sequencing of whole liver tissue post-DEN treatment, demonstrating that mutational foot printing as a result of TCR can be used as a robust technique to infer historical transcriptional changes in cells that form tumours.

To identify specific liver cell populations that are predisposed to tumour formation, I conducted single nuclei RNA sequencing post-DEN treatment and compared the transcriptional profiles of groups and of single nuclei to the mutational data. I show that a subset of DEN-damaged Cyp2e1 expressing hepatocytes has the best expression fit with the mutational data, suggesting these may represent a subset of hepatocytes at the earliest stages of tumorigenesis.

In conclusion, this work identifies a subset of DEN-damaged hepatocytes with a transcriptional profile suspected to convey preference for tumour formation over other damaged and mutated cells and serves as the first description of using mutational data to retroactively define the transcriptional state of cells that form tumours.