3D Printed Vascular Networks Enhance Viability in High-Volume Perfusion Bioreactor

Abstract

There is a significant clinical need for engineered bone graft substitutes that can quickly, effectively, and safely repair large segmental bone defects. One emerging field of interest involves the growth of engineered bone tissue in vitro within bioreactors, the most promising of which are perfusion bioreactors. Using bioreactor systems, tissue engineered bone constructs can be fabricated in vitro. However, these engineered constructs lack inherent vasculature and once implanted, quickly develop a necrotic core, where no nutrient exchange occurs. Here, we utilized COMSOL modeling to predict oxygen diffusion gradients throughout aggregated alginate constructs, which allowed for the computer-aided design of printable vascular networks, compatible with any large tissue engineered construct cultured in a perfusion bioreactor. We investigated the effect of 3D printed macroscale vascular networks with various porosities on the viability of human mesenchymal stem cells in vitro, using both gas-permeable, and non-gas permeable bioreactor growth chamber walls. Through the use of 3D printed vascular structures in conjunction with a tubular perfusion system bioreactor, cell viability was found to increase by as much as 50% in the core of these constructs, with in silico modeling predicting construct viability at steady state.

Keywords: 3D printed; Bioreactor; Computer-aided design; Modeling; Perfusion; Vascular; hMSCs.

Figure 1. Bioreactor Setup

a) Reaction chamber schematic. Media is perfused through inlet connector in direction of red arrow. All media travels through the printed network, shown in blue, where it escapes through the network pores and passes through the void space between beads. Media then passes through outlet connector, back to the reservoir, where the cycle begins anew. b) Overall setup (acellular). Reaction chamber is hung vertically (far right) to allow for clearance of any air bubbles. Here, media reservoir is shown stacked on top of bioreactor pump, which can be digitally adjusted to allow for a wide variety of flow rates. Media shown here is simply water with green dye, to demonstrate more clearly the full perfusion achieved by the network. c) 4 cell-seeded reaction chambers in parallel using the same pump, housed in one incubator, hung vertically to remove air bubbles d) Harvest sites are enumerated on a fully-aggregated, acellular construct as follows: 1) Bottom Inner 2) Bottom Outer 3) Middle Inner (Core) 4) Middle Outer 5) Top Inner 6) Top Outer. Once again, construct has been perfused with green dye to demonstrate complete perfusion achieved by the printed network. Here, reaction chamber outer tubing and connectors have been removed to demonstrate structural integrity post aggregation.

Figure 2. Architecture of vascular network

a) Major dimensions. b) Sectioned parallel to flow direction. c) Network arranged vertically, as in experiments. Pores run entire length of network, bead stop disk prevents beads from falling below the lowest network pore. d) Section view shows interior geometry, notably ridges and overhangs designed to ensure even distribution of flow throughout all channels, as well as vertical pores, angled in direction of flow. e) Final product, printed, trimmed, and cleaned.

Figure 3. 2D COMSOL Modeling

All panels share the scale depicted on the far right (mM). a) Middle branches, 1.5 mm pore spacing, PCS tubing b) middle section, 9 mm pore spacing, PCS tubing c) Top/Bottom branches, 1.5 mm pore spacing, FEP tubing d) Top/Bottom branches, 9 mm pore spacing, FEP tubing e) Middle branches, 1.5 mm pore spacing, FEP tubing f) Middle branches, 9 mm pore spacing, FEP tubing g) Single tube, 9 mm pore spacing, FEP tubing h) Static control. Here, the peripheral alginate is directly exposed to media. In all panels, the circles represent individual, cell-laden alginate beads, while the void space is acellular alginate. The rectangular nodes on each channel represent network pores, modeled as constant surface concentrations.

Figure 4. Viability Analyses

a,b) Day 1 images from Middle Inner harvest site of the Silicone group at 2.5 × and 10 × magnifications. c,d) Day 2 images from Static Core at 2.5 × and 10 × magnifications. e,f) Day 2 images from Middle Inner harvest site of the Silicone group at 2.5 × and 10 × magnifications. g,h) Day 2 Dead Control images at 2.5 × and 10 × magnifications. i) Legend to be used for graphs in h–j. In these graphs, the vertical axis displays percentage of live cells at the time point. Groups that do not share a letter are statistically different (p < 0.05). Note there is no statistical difference in any group in panel h. Dead control results are not displayed due to the complete absence of live cells. j) Composite averages across all harvest sites for each group after Day 1. k) Percent live from all Middle Inner harvest sites l) Composite averages across all harvest sites for each group after Day 2. m) COMSOL simulation results, aligned vertically with the percent live graph results for the group that they represent. Units on legend are mM.