Lineage specific core-regulatory circuits determine gene essentiality in cancer cells

Abstract

Cancer cells rely on dysregulated gene expression programs to maintain their malignant phenotype. A cell's transcriptional state is controlled by a small set of interconnected transcription factors that form its core-regulatory circuit (CRC).Previous work in pediatric cancers has shown, that disruption of the CRC by genetic alterations causes tumor cells to become highly dependent on its components creating new opportunities for therapeutic intervention. However, the role of CRCs and the mechanisms by which they are controlled remain largely unknown for most tumor types. Here, we developed a method that infers lineage dependency scores to systematically predict functional CRCs and associated biological processes from context-dependent essentiality data sets. Analysis of genome-scale CRISPR-Cas9 screens in 558 cancer cell lines showed that most tumor types specifically depend on a small number of transcription factors for proliferation. We found that these transcription factors compose the CRCs in these tumor types. Moreover, they are frequently altered in patient tumor samples indicating their oncogenic potential.Finally, we show that biological processes associated with each CRC are revealed by analyzing codependency between lineage-specific essential genes. Our results demonstrate that genetic addiction to lineage-specific core transcriptional mechanisms occurs across a broad range of tumor types. We exploit this phenomenon to systematically infer CRCs from lineage specific gene essentiality. Furthermore, our findings shed light on the selective genetic vulnerabilities that arise as the consequence of transcriptional dysregulation in different tumor types and show how the plasticity of regulatory circuits might influence drug resistance and metastatic potential.

Dynamic transcriptional reprogramming leads to immunotherapeutic vulnerabilities in myeloma

“…Dependency data are based on CERES scores from the Depmap dataset (DepMap Public 19Q1) and lineage-specific dependencies were determined using the R package HDCRC2019 66 . A gaussian mixture model was calculated with 2 components and a cutoff of FDR<0.05 and estimate>1 was chosen for identifying lineage-specific essential genes.…”

“…2,000–5,000 plasma cells were bulk sorted as described above and single cell ATAC-Seq libraries were prepared as described by Chen et al 66 . Libraries were paired-end sequenced using a 75 cycle kit on a NextSeq 500 (Illumina) sequencer with an average sequencing depth of 0.5–1 million reads per cell.…”

Predicting master transcription factors from pan-cancer expression data

“…We present a prioritized collection of candidate MTFs for 34 major tumor types and 140 tumor subtypes, which will be enriched for bone fide MTFs, but it will also likely contain false positive. For users of this resource, we recommend integration of complementary data, where available -such as SE landscapes, dependency data (Rauscher et al, 2019), and motif-based circuitry mapping to inform the design of functional validation experiments. We note that not all cancer MTFs will fulfill all the canonical MTF criteria -for example ZNF217 has lower levels of expression in basal-type breast tumors (expression rank = 232), but high levels of dependency (minimum dependency in basal BRCA cell lines = -1.12).…”