Inference of tumor cell-specific transcription factor binding from cell-free DNA enables tumor subtype prediction

Deregulation of transcription factors (TFs) is an important driver of tumorigenesis. For TFs to bind DNA, the binding region must be accessible. Hence, TFs and chromatin remodelling complexes shift and position nucleosomes to enhance accessibility. Cell-free DNA (cfDNA) is the product of a digestion process that preferentially degrades DNA that is not protected by proteins such as the histone complex. It follows that cfDNA coverage patterns likely reflect nucleosome positioning and occupancy caused by transcription factors actively binding the genome. These nucleosome occupancy patterns measured through cfDNA may then be used to infer the activity of TFs in the normal and tumor genomes.

Here, we developed and evaluated a minimally invasive method for assessing TF activity based on cfDNA sequencing and nucleosome footprint analysis. We analyzed whole-genome sequencing data of cfDNA samples from patients with prostate (n=11), colon (n=2) and breast cancer (n=2) as well as non-cancer controls (n=24) and pools of low-coverage cfDNA samples (colon cancer: n=100, prostate cancer: n=69, breast cancer: n=60) and inferred accessibility of 504 TFs from the resulting coverage patterns. We observed patient-specific as well as tumor-specific differences in accessibility, highlighting the dynamic nature of TF-mediated chromatin remodelling. These differences include nucleosome changes that are specific for the neuro-endocrine subtype in prostate cancer, which has important clinical implications for the management of patients. We also observed statistically significant different accessibilities between colorectal cancer (n=592) and non-cancer controls (n=177) in a separate set of colorectal cancer samples down to a tumor fraction of 1%.

Measuring TF accessibility unveils a key part of the non-coding genome, adding a powerful new tool to query the tumor from a functional perspective by non-invasive means.