Increased viscosity of concentrated contrast media (CM) in the renal tubules can perturb renal hemodynamics and have a detrimental effect on tubular epithelial cells. However, the effects of viscosity on contrast-induced nephropathy (CIN) remain poorly understood. Conventional in vitro culture studies do not reflect the rheological properties of CM. Therefore, we investigated the effects of CM viscosity on renal tubules using a kidney-on-a-chip and two different types of CM. Renal proximal tubule epithelial cells (RPTEC) were cultured in a three-dimensional microfluidic culture platform under bidirectional fluid shear stress. We treated the RPTEC with two types of CM: low- (LOCM, iopromide) and iso-osmolar contrast media (IOCM, iodixanol). Renal tubular cell injury induced by LOCM and IOCM was examined under different iodine concentrations (50–250 mgI/mL) and shear-stress conditions. LOCM showed a significant dose-dependent cytotoxic effect, which was significantly higher than that of IOCM under static and low-to-moderate shear stress conditions. However, high shear-stress resulted in reduced cell viability in IOCM; no difference between IOCM and LOCM was found under high shear-stress conditions. The cytotoxic effects were pronounced at a mean shear stress of 1 dyn/cm² or higher. The high viscosity of IOCM slowed the fluid flow rate and augmented fluid shear-stress. We suggest an alternative in vitro model of CIN using the three-dimensional kidney-on-a-chip. Our results indicate a vital role of viscosity-induced nephrotoxicity under high shear-stress conditions, contrary to the findings of conventional in vitro studies.

We developed a 3D glomeruli tissue chip for glomerulonephritis (GN) testing, featuring a gravity-driven glomerular filtration barrier (GFB) with human podocytes and endothelial cells with a bidirectional flow in the bottom channel. Using puromycin-induced GN, we observed decreased cell viability, increased albumin permeability, and reduced WT1 and nephrin compared to the normal GFB. Tacrolimus restored cell viability, reduced albumin permeability, and increased WT1 expression. Using serum from five membranous nephropathy (MN) patients, we created MN models using a GFB-mimicking chip. A notable decline in cell viability was observed in the serum-induced MN1 and MN2 models. However, tacrolimus restored it. Albumin permeability was reduced in the MN1, MN2, and MN5 models by tacrolimus treatment. MN1 displayed the best clinical response to tacrolimus, exhibiting increased expression of WT1 in chip-based evaluations after tacrolimus treatment. We successfully evaluated the efficacy of tacrolimus using puromycin-induced and serum-induced GN models on a chip that mimicked the structure and function of the GFB. The GFB-mimicking chip holds promise as a personalized platform for assessing drug efficacy using patient serum samples.