User-Friendly Microfluidic System Reveals Native-Like Morphological and Transcriptomic Phenotypes Induced by Shear Stress in Proximal Tubule Epithelium

Drug-induced nephrotoxicity is a leading cause of drug attrition, partly due to the limited relevance of pre-clinical models of the proximal tubule. Culturing proximal tubule epithelial cells (PTECs) under fluid flow to mimic physiological shear stress has been shown to improve select phenotypes, but existing flow systems are expensive and difficult to implement by non-experts in microfluidics. Here, we designed and fabricated an accessible and modular flow system for culturing PTECs under physiological shear stress, which induced native-like cuboidal morphology, downregulated pathways associated with hypoxia, stress, and injury, and upregulated xenobiotic metabolism pathways. We also compared the expression profiles of shear-dependent genes in our in vitro PTEC tissues to that of ex vivo proximal tubules and observed stronger clustering between ex vivo proximal tubules and PTECs under physiological shear stress relative to PTECs under negligible shear stress. Together, these data illustrate the utility of our user-friendly flow system and highlight the role of shear stress in promoting native-like morphological and transcriptomic phenotypes in PTECs in vitro, which is critical for developing more relevant pre-clinical models of the proximal tubule for drug screening or disease modeling. Overall design: Human hTERT-immortalized proximal tubule epithelial cells purchased from ATCC were cultured as monolayers on ibidi channel slides and subsequently exposed to physiological (0.35 dynes/cm2) or near-zero levels of fluidic shear stress using a user-friendly microfluidic flow system for 3 days. Cells were subsequently lysed and RNA was sequenced. 8 samples (GSM4982338, GSM4982384, GSM4982350, GSM4982357, GSM4982391, GSM4982364, GSM4982344, GSM4982370) were reanalyzed from GSE163603. They were first sub-sampled to normalize the sequencing depth (by a factor of 2, 2, 5, 1, 1, 1, 1, and 5, respectively). Together with our data, sequences were trimmed from both ends based on a Phred quality score of 20 and a minimum read length of 25, aligned to the hg38 genome build using STAR (2.7.8a), and quantified using Gencode 38 (V2) annotation in Partek Flow Genomics Analysis software. Read counts per gene were normalized to fragments per kilobase of transcript per million mapped reads (FPKM) and then analyzed for differential gene expression using Partek Gene Specific Analysis method with fold change and FDR cut-offs of 1.5 and 0.05, respectively.