Characterization of Diabetic and Non-Diabetic Foot Ulcers Using Single-Cell RNA-Sequencing

Micromachines (Basel). 2020 Aug 28;11(9):815. doi: 10.3390/mi11090815.


Background: Recent advances in high-throughput single-cell sequencing technologies have led to their increasingly widespread adoption for clinical applications. However, challenges associated with tissue viability, cell yield, and delayed time-to-capture have created unique obstacles for data processing. Chronic wounds, in particular, represent some of the most difficult target specimens, due to the significant amount of fibrinous debris, extracellular matrix components, and non-viable cells inherent in tissue routinely obtained from debridement. Methods: Here, we examined the feasibility of single cell RNA sequencing (scRNA-seq) analysis to evaluate human chronic wound samples acquired in the clinic, subjected to prolonged cold ischemia time, and processed without FACS sorting. Wound tissue from human diabetic and non-diabetic plantar foot ulcers were evaluated using an optimized 10X Genomics scRNA-seq platform and analyzed using a modified data pipeline designed for low-yield specimens. Cell subtypes were identified informatically and their distributions and transcriptional programs were compared between diabetic and non-diabetic tissue. Results: 139,000 diabetic and non-diabetic wound cells were delivered for 10X capture after either 90 or 180 min of cold ischemia time. cDNA library concentrations were 858.7 and 364.7 pg/µL, respectively, prior to sequencing. Among all barcoded fragments, we found that 83.5% successfully aligned to the human transcriptome and 68% met the minimum cell viability threshold. The average mitochondrial mRNA fraction was 8.5% for diabetic cells and 6.6% for non-diabetic cells, correlating with differences in cold ischemia time. A total of 384 individual cells were of sufficient quality for subsequent analyses; from this cell pool, we identified transcriptionally-distinct cell clusters whose gene expression profiles corresponded to fibroblasts, keratinocytes, neutrophils, monocytes, and endothelial cells. Fibroblast subpopulations with differing fibrotic potentials were identified, and their distributions were found to be altered in diabetic vs. non-diabetic cells. Conclusions: scRNA-seq of clinical wound samples can be achieved using minor modifications to standard processing protocols and data analysis methods. This simple approach can capture widespread transcriptional differences between diabetic and non-diabetic tissue obtained from matched wound locations.

Keywords: cellular ecology; diabetes; fibrosis; single cell RNA sequencing; tissue repair; transcriptomics; wound healing.