Materials and Methods

Simulating and Validating model in COMSOL MultiPhysics®

A 2D computational model was developed through COMSOL Multiphysics (COMSOL Inc, Burlington, MA). Computational domains were generated using COMSOL’s native geometry tools. To model the fluid flow in open and porous domains, the Free and Porous Media Flow physics package was used for porous membranes. Inlet, Outlet and No-Slip Wall conditions were used to define the boundary conditions of the computational domain. A symmetry boundary condition was used to reduce the computational load, and domains were discretized using a combination of swept, tetrahedral and boundary layer elements using native meshing tools. Using both stationary and time-dependent solvers, the 2D model was used to optimize tubing internal diameter (ID) and length (Cole-Parmer, Vernon Hills, IL), and total flow rate from the peristaltic pump (Watson-Marlow, Paramus, NJ). The primary design parameters were achieved a passive fluid flow split at the glomerular filter T-junction (Eldon James, Denver, CO), and obtaining a fluid shear stress of 0.65 dyn-s/cm2 across the top chamber of the proximal tubule device (PCT), which mimics the physiological fluid shear stress in the human proximal tubule.

Fabrication of the Microphysiological System

Construction of the proximal tubule-on-a-chip has been previously described (C. M. Sakolish & Mahler, 2017; C. M. Sakolish et al., 2019). Detailed description of the fabrication method is in the supplementary materials.

Cell Culture

All cell lines were adapted to Endothelial Serum Free Medium (ESFM, Thermo Fisher Scientific, Waltham, MA) supplemented with 10 µg/mL of human plasma fibronectin (Thermo Fisher Scientific, Waltham, MA), 20 ng/mL of human recombinant epidermal growth factor (EGF, Thermo Fisher Scientific, Waltham, MA), and 10 ng/mL of human recombinant basic fibroblast growth factor (bFGF, Thermo Fisher Scientific, Waltham, MA). Conditionally immortalized human podocytes (CIHP-1) purchased from Dr. Saleem Moin’s laboratory at the University of Bristol at passage 16 proliferated at 33°C in 5% CO2 and were passaged every 10-14 days. The human, renal proximal tubule cell line (HK-2) (passage 20-30) obtained from the American Type Culture Collection (ATCC, Manassas, VA) was incubated at 37°C in 5% CO2 and passaged every 5-7 days with 0.05% Trypsin-EDTA (Thermo Fisher Scientific, Waltham, MA). Human umbilical vein endothelial cells (HUVECs) purchased from Lonza Bioscience (Rockland, MD) were maintained at 37°C in 5% CO2 and experiments were completed using passages 3-7.

CIHP-1, HUVECs, and HK-2 adhesion on porous membranes

30 nm pore size polyethersulfone (PES) membranes (Sterlitech, Kent, WA) were coated with glomerular extracellular matrix (GECM), which consist of tri-layer of 8 µg/cm2 rat-tail collagen 1 (BD Biosciences, Franklin Lakes, NJ), 8 µg/cm2 heparin sulfate (Sigma-Aldrich H7640), and 8 µg/cm2 collagen 1 (BD Biosciences, Franklin Lakes, NJ). The selection of the GECM composition was adapted from previous work (C. M. Sakolish & Mahler, 2017; C. M. Sakolish et al., 2019). Detailed description of the adhesion methods is in the supplementary materials.

Assembly of housing unit and microfluidic device

The experimental study was conducted in a sterile, biosafety cabinet cleaned with 70% ethanol. All device components were autoclaved prior to experimentation. The internal and external components of the peristaltic pump were cleaned with 70% ethanol before stringing the Pharmed BPT 0.25 mm ID tubing (Cole-Parmer, Vernon Hills, IL) into the cassette with a conical tube of 1X phosphate buffered saline (PBS). All tubing was subjected to an average flow of 40 µL/min with 1X PBS for 1 hour to eradicate bubble formation. Assembly of the glomerulus stainless steel housing unit (Advantec MFS, Dublin, CA) (C. M. Sakolish et al., 2016; C. M. Sakolish & Mahler, 2017; C. M. Sakolish et al., 2019), consisting of a stainless steel mesh (Advantec MFS, Dublin, CA), pre-cut 14 mm in diameter PES membrane (Sterlitech, Kent, WA), polytetrafluoroethylene (PTFE) gasket (Advantec MFS, Dublin, CA), pre-cut 200 µL pipet tips (VWR International, Radnor, PA), and leur locks (Cole-Parmer, Vernon Hills, IL) was sealed with an autoclavable wrench (SteriTool, Jacksonville, FL). While the assembled glomerulus housing unit was attached to the end of the tubing to fill with 1X PBS, assembly of the T-junction and the CNC machined, polycarbonate PCT device was completed. Within the PCT device, a 0.4 µm pore size polycarbonate membrane and 0.01-in thick silicone polymer gasket (McMaster-Carr, Elmhurst, IL) created anin vitro barrier system with an apical and basolateral sections. The components are assembled in their respective locations after filling with 1X PBS, and placeholder porous membranes were replaced with the cell seeded porous membranes. Afterwards, the inlet fluid was replaced with the supplemented ESFM incubated at 37°C, and two outlet tubing was connected to 50 mL conical tube for filtrate output and bloodstream output. ESFM was changed every 2-3 days for a week.

MPS Cellular Models

Under dynamic conditions, two cellular models (single and tri-cultures) were employed with the glomerulus and PCT MPS components to assess the necessity of a multi-culture model. The single culture contained only proximal tubule cells (HK-2) adhered to the polycarbonate membrane in the PCT device. The single culture MPS was operated with the T-Junction, stainless-steel glomerulus housing unit with blank PES membrane, and seeded HK-2 cells in the PCT device. There was no glomerular or endothelial cells (CIHP-1 and HUVECs) on the PES membranes. The tri-culture housed all three cell-types on their respective membranes. The podocytes (CIHP-1) and endothelial (HUVECs) cells were attached to the PES membrane in the glomerulus unit in addition to the proximal tubule (HK-2) cells attached to the polycarbonate membrane in the PCT device.

Volume Flow Assay

Filtrate output volume was collected daily in 50 mL conical tubes (Corning, Corning, NY) and stored in -20°C for future assessment. In post-experiment testing, the volumes were thawed, weighed, and normalized to an empty conical tube. Based on the density of the volume (1g/mL), total volume was calculated over time in mL/min.

FITC-human serum albumin flow assay

After seven days of fluidic culture, the nephron-on-a-chip system was subjected with 0.1 mg/mL of FITC-HSA (Abcam) diluted with 1X PBS by replacing the inlet conical tube fluid. In two black 96 well plates, samples (n =4) of the inlet and outlet fluid flow were collected at each 15-min intervals for 2 hrs. Fluorescence was assessed with Synergy Plate Reader (Biotek, Winooski, VT) and Gen5 software with an emission/excitation wavelength of 485/528 nm. Results were normalized to the average maximum fluorescent intensity.

Fluorescent staining and image processing

Following the system challenge with HSA, membranes were removed from the devices and placed in a 4-well plate. The cells were fixed with 4% paraformaldehyde (PFA) diluted in 1x PBS for 5 min at room temperature and permeabilized with 0.1% Triton X-100 in 1x PBS for 15 min. The cells were stained with phallodin 568 (Invitrogen) and counterstained with Hoescht 33342 (Thermo Fisher Scientific, Waltham, MA) for 30 mins. Membranes were then washed three times with 1X PBS for each 5 mins. Samples were mounted onto 24 x 40 x 0.13 mm glass slides through ProLong Gold and visualized with Zeiss Leica SP5 X confocal microscope. Images were processed in FIJI ImageJ (Schindelin et al., 2012) for directionality, confluence, and cell count. Z-stacks were taken for the membranes to generate a three-dimensional (3D) profile.

Infinity glucose assay

Glucose concentrations were calculated from the filtrate output volume using the Infinity™ Glucose Hexokinase Liquid Stable Reagent assay (Thermo Scientific, Waltham, MA). The reagent was diluted with filtrate output volume at a ratio of 150:1, incubated in 37°C for three minutes, and plated into a 96 well plate (n=4) for each day. Absorbance was assessed with Synergy Plate Reader (Biotek, Winooski, VT) and Gen5 software with a primary/secondary wavelength of 340/380 nm. Standards were established with both a glucose solution of 1 mg/mL of 0.1% benzoic acid and supplemented ESFM.

Statistical Analysis

Nonparametric one-way ANOVA with post-hoc Dunn’s test, Mann-Whitney test, and regression tests were conducted in GraphPad Prism® (GraphPad, San Diego, CA). All data from single- and tri-cultures were presented with medians and ranges.