Chemotherapy resistance pathways identified by single cell RNA sequencing

Abstract

<h3>Objectives:</h3> Chemotherapy resistance is a major cause of mortality in recurrent ovarian cancer. The goal of this study is to identify signaling pathways associated with chemotherapy resistance by analyzing gene expression in single cells from ovarian cancer specimens collected from women before and after platinum/taxane based neoadjuvant chemotherapy (NACT). <h3>Methods:</h3> Women were prospectively consented into our study and ovarian cancer tissue samples were collected from six patients with primary epithelial ovarian cancer both before and after administration of NACT. Samples were subjected to single cell RNA sequencing (scRNAseq) using the 10x Genomics platform followed by sequencing on an Illumina NovaSeq. Sequences were mapped to the human transcriptome and gene expression was quantified to produce unique molecular identifier counts for ~85,000 cells from the 12 samples. Unsupervised hierarchical clustering was performed using a shared nearest-neighbor graph based embedding algorithm implemented by the Seurat R package. Multiple annotation methods were used to assign cell types. Cells annotated as epithelial cancer cells for each sample were extracted and re-clustered and average gene expression was calculated for each pre-NACT cluster. To identify the pre-NACT cluster that had the highest similarity to the post-NACT cells, we performed Spearman Rank Correlation testing on each individual cell in the post-NACT sample comparing it to each cluster in the matched pre-NACT sample. The pre-NACT cluster with the highest number of matches to the post-NACT cells were considered the potential chemotherapy-resistant population in the pre-NACT sample, which we refer to as the ‘seeder' cell cluster. We used multiple enrichment tests to identify signaling pathways activated in the seeder cell clusters compared to the other pre-NACT epithelial cell clusters. <h3>Results:</h3> We identified 62 genes that were significantly upregulated in the pre-NACT seeder cell clusters in at least 3 of 6 patient samples analyzed. Pathways enriched in this 62 gene set include 16 hallmark signaling pathways and 36 oncogenic signature pathways. In addition, transcription factor target enrichment identified 36 transcription factors based on the 62 gene set. Expected pathways included p53 and epithelial mesenchymal transition pathways. Novel pathways included TNFa, TGFb, JAK-STAT, interferon, and estrogen. Associated transcription factors included STAT3/5, SMAD4, and TP53. <h3>Conclusions:</h3> ScRNAseq analysis of matched pre- and post-NACT cancer samples identifies potential chemoresistant cells in pre-NACT samples capable of surviving NACT. The signaling pathways enriched in these cells represent potential therapeutic targets for selective elimination of these ‘seeder' cells.