Abstract
Objectives: Human pluripotent stem cells (hPSCs) are of great importance in both scientific research and regenerative medicine. The most classic and widely used culture method for hPSCs is co-culture with feeder cells, usually mouse embryonic fibroblasts. However, whether these feeder cell residues can affect the transcriptomic data analysis of hPSCs, especially gene or miRNA expression quantification, is still largely unknown. Methods and results: In this study, reanalysis of published mRNA-Seq and miRNA-Seq data sets revealed the existence of feeder cell-derived reads in the hPSC transcriptomic samples. We identified potentially influenced human genes and miRNAs due to misalignment of sequencing fragments affected by mouse feeder cells. Furthermore, we developed an optimized miRNA analysis pipeline to avoid quantification bias from different miRNA isoforms in the same family. Finally, by comparing the levels of feeder cell residues in hPSC samples isolated by different methods, we found that fluorescence-activated cell sorting and adhesion methods were more effective in feeder cell removal than the gradient centrifugation method. Conclusions: Collectively, our results demonstrate that feeder cell residues affect the transcriptomic data analysis of hPSCs. To minimize the impact of feeder cell contamination in hPSC samples, we provide solutions for both data analysis and sample preparation.