A) Diagram shows the single-cell RNA sequencing (scRNA-seq) and analysis workflow. We collected and processed normal fallopian tube tissues from ten cancer patients. All cells (directly sorted or maintained then sorted) were processed by using the Smart-Seq2 protocol. After the initial filtering, there were 3,877 good-quality cells left for downstream analysis. We compared the cells from three conditions to select the optimal condition for scRNA-seq. Cells from cultured or cryopreserved conditions, as well as cells carrying copy number variations and non-epithelial cells, were filtered out, which left 2,132 fresh FTE cells. Next, we used differential expression-based clustering to identify secretory subtypes.

The inter-differentiation between cell states promotes cancer cell survival under stress and fosters non-genetic heterogeneity (NGH). NGH is, therefore, a surrogate of tumor resilience but its quantification is confounded by genetic heterogeneity. Here we show that NGH can be accurately measured when informed by the molecular signatures of the normal cells of origin.We surveyed the transcriptomes of ~ 4000 normal fallopian tube epithelial (FTE) cells, the cells of origin of serous ovarian cancer (SOC), and identified six FTE subtypes. We used subtype signatures to deconvolute SOC expression data and found substantial intra-tumor NGH that was previously unrecognized. Importantly, NGH-based stratification of ~1700 tumors robustly predicted survival. Our findings lay the foundation for accurate prognostic and therapeutic stratification of SOC. Highlights 1. The projection of FTE subtypes refines the molecular classification of serous OC 2. Comprehensive single-cell profiling of FTE cells identifies 6 molecular subtypes 3. Substantial non-genetic heterogeneity of HGSOC identified in 1700 tumors 4. A mesenchymal-high HGSOC subtype is robustly correlated with poor prognosis

No abstract