Single-cell RNA-seq of patient-derived cancer-associated fibroblast in vitro cell line models from pancreatic ductal adenocarcinoma

Cancer-associated fibroblasts (CAFs) are sought after as potential therapeutic targets due to their plethora of pro- and antitumorigenic functions. Recent findings attributed this functional heterogeneity to specializations in CAF subtypes. A precise targeting of specific subtypes would thus be required to design therapies that effectively modulate CAF phenotypes, but our knowledge of CAF heterogeneity in solid tumors, particularly pancreatic ductal adenocarcinoma (PDAC), is still lacking. Here, we use single-cell transcriptomics to characterize CAF subtype heterogeneity in in vitro CAF cell lines isolated from PDAC patients and investigate subtype-resolution modulations arising from perturbing potential stromal genes. We demonstrate that co-culturing CAFs with tumor cells can produce primary cell line models that can reliably represent CAF phenotypes observed in tumors, with correlations to immuno-resistant and immuno-modulatory phenotypes. With trajectory analysis, we infer the continuum of state changes that underlie the interconvertibility of CAF subtypes. Finally, we use the immortalized CAF cell lines to perform single-cell CRISPR perturbations of candidate stromal targets, revealing not only the subtype-specific effects of the perturbations, but also the impact of model-type selection on the translatability of insights.