Clonal deconvolution of transcriptomic signatures and their spatial organisation in a mouse model of breast cancer

Abstract

Intratumour heterogeneity is a phenomenon during cancer progression in which cancer cells diverge and form clonal populations with distinct phenotypic, genetic, or epigenetic states within the same tumour. This intrinsic heterogeneity provides a fuel for cancer evolution enabling tumour cell populations to adapt to selective pressures imposed by the tumour microenvironment or therapeutic interventions. Lineage-tracing approaches shed light into the clonal dynamics of complex populations, but generally lack the ability to directly associate clonal lineage with measurements that infer phenotype such as epigenetics and transcriptomics. In contrast, single-cell sequencing techniques can provide insight into the makeup of complex biological ecosystems, revealing the presence of rare cell populations that are typically masked in bulk analyses, but lack the ability to link these cell populations to clonal lineages.

To address this challenge, we developed the WILDseq platform, a novel approach that allows clonal characterisation at the single-cell transcriptomic level while facilitating the prospective analysis of dynamic regulation of phenotypic heterogeneity under the selective pressure of therapeutic intervention. WILDseq relies on uniquely labelling individual cells with a heritable, expressed DNA barcode coupled with high-throughput single-cell RNA-sequencing. Importantly, this lentiviral-labelling approach can be deployed in any model system that is susceptible to viral transfection. Thus, this platform allows the comprehensive and systematic characterisation of clonal phenotypic di fferences within complex populations. Here we demonstrate how this technology can be used to determine clonal populations which are sensitive or resistant to a particular therapeutic intervention, identify transcriptomic signatures that correlate with these phenotypes and analyse how these cells adapt their transcriptomes to escape therapy. We have applied WILDseq to the study of di fferential clonal responses to chemotherapy in the heterogeneous 4T1 model of breast cancer and validated transcriptomic signatures of therapeutic resistance and sensitivity in primary patient data. We additionally used WILDseq to study the clonal response to the epigenetic regulator JQ1 which revealed intrinsic signatures that primed clones to JQ1 sensitivity. We observed JQ1-dependent depletion of CD8+ cytotoxic T-cells and suggest that this drives changes in clonal distribution. Finally, we are working on developing a high throughput FISH assay to leverage the WILDseq technology for mapping clonal and transcriptional identities spatially.

Collectively, this thesis contributes to the characterisation and understanding of breast cancer heterogeneity and the impact of clonal architecture on tumour progression and response to therapy.