Adaptable Manufacturing and Biofabrication of Milliscale Organ Chips With Perfusable Vascular Beds

Abstract

Microphysiological systems (MPS) containing perfusable vascular beds unlock the ability to model tissue-scale elements of vascular physiology and disease in vitro. Access to inexpensive stereolithography (SLA) 3D printers now enables benchtop fabrication of polydimethylsiloxane (PDMS) organ chips, eliminating the need for cleanroom access and microfabrication expertise, and can facilitate broader adoption of MPS approaches in preclinical research. Rapid prototyping of organ chip mold designs accelerates the processes of design, testing, and iteration, but geometric distortion and surface roughness of SLA resin prints can impede the development of standardizable manufacturing workflows. This study reports postprocessing procedures for manufacturing SLA-printed molds that produce fully cured, flat, patently bonded, and optically clear polydimethyl siloxane (PDMS) organ chips. Injection loading tests were conducted to identify milliscale membrane-free organ chip (MFOC) designs that allowed reproducible device loading by target end-users, a key requirement for broad nonexpert adoption in preclinical research. The optimized milliscale MFOC design was used to develop tissue engineering protocols for (i) driving bulk tissue vasculogenesis in MFOC, and (ii) seeding the bulk tissue interfaces with a confluent endothelium to stimulate self-assembly of perfusable anastomoses with the internal vasculature. Comparison of rocker- and pump-based protocols for flow-conditioning of anastomosed vascular beds revealed that continuous pump-driven flow is required for reproducible barrier maturation throughout the 3D tissue bulk. Demonstrated applications include nanoparticle perfusion and engineering perfusable tumor vasculature. These easily adaptable methods for designing and fabricating vascularized microphysiological systems can accelerate their adoption in a diverse range of preclinical laboratory settings.

Keywords: 3D printing; endothelial barrier; organ chips; tissue engineering; tumor vasculature; vascularization.

FIGURE 1

Manufacturing milliscale PDMS organ chips using a low‐cost SLA printer. (A) Incomplete PDMS curing in SLA‐printed molds with no postprocessing. Scale bar = 10 mm. (B) Warped 3D printed mold postcuring (top) compared to the same mold after baking and flattening (bottom). Scale bar = 10 mm. (C) Flattened organ chip mold for the top layer of the multilayer device in Panels E, F, and G. Scale bar = 10 mm. (D) CAD render of exploded view of multilayer organ chip. Scale bar = 10 mm. (E) Multilayer organ chip (three PDMS layers, two membrane layers) bonded by PDMS stamping with food dye loading to demonstrate device layering and patent bonding with flattened layers. Scale bar = 10 mm. (F) F‐actin green) and DAPI (blue) labeling of the A549 cell layer and underlying 3D interstitium layer with human lung fibroblasts (HLFs) embedded in collagen Type I hydrogel. Scale bars = 200 µm. (G) 3D reconstruction (1 mm³ volume) of the entire 1 mm height of the OIC containing A549 cells and lung fibroblasts as shown in Panel F. The position of membranes separating the device layers are indicated. HLF stained via F‐actin (green) and nuclei using DAPI (blue) after 7 days in culture. Scale bar = 300 µm. (H) Comparison of uncoated mold (left) and polyurethane (PU) clear coated mold (right). Scale bar = 10 mm. (I) Absorbance of light at varying wavelengths between clear coated and uncoated molds demonstrates the optical transparency of PDMS layers from PU‐coated molds. N = 3 PDMS slabs per group. (J) Example of failed injection defined as liquid breaching over phase guides in device made with uncoated molds (left) and an example of successful injection in a device manufactured using PU‐coated molds (right). Scale bar = 10 mm. (K) Comparison of success rate of loading between devices made using PU clear coated versus uncoated molds; n = 6 users, 10 trials each spread over at least 3 replica devices per group. Compared via unpaired t test with α = 0.05. **** indicates p < 0.0001. DAPI, 4′,6‐diamidino‐2‐phenylindole; PDMS, polydimethylsiloxane; SLA, stereolithography.

FIGURE 2

MFOC design iteration and reduction to practice through user testing. All testing consisted of least n = 6 users performing 10 injection trials per design. (A) CAD render of a prototypical MFOC device. Scale bar = 5 mm. Dimensions of internal features and volumes were varied for user testing of loading success rate. (B, C) CAD render of layers and drawing of cross section of single guide MFOC (B) and double guide MFOC (C). Scale bars = 5 and 1 mm. Key dimensions include guide height (h), tissue chamber height (H), and tissue width (W). A full listing of dimensions for all designs tested in this study are available in the Supplementary Information for this article. General size ranges for single guide MFOC: H = 0.5 mm, l = 12 mm, and W = 1–2 mm; for double guide MFOC: H = 1 mm, l = 12 mm, W = 1.2 mm. Guide width = 0.55 mm for all designs. (D) Loading success rates for single guide and double guide MFOC, N = 3 devices per group. (E) Loading success rates for various h/H values in double guide MFOC, N = 3 devices per group. (F) Loading success rates for tissue volumes ranging from 8 to 48 mm³. H and W were varied with constant l = 12 mm. N = 3 devices per group. (G) Acellular Type I collagen hydrogel containing 40 kDa for visualization FITC‐after injection loading and polymerization in an MFOC. Scale bar = 200 µm. Indicated guide width = 0.55 mm. (H) Proposed free body diagram of forces acting on a liquid (collagen precursor solution for our applications) pinned between the guide structures of a double guide MFOC as shown in Panel F. Fc is the cohesion of the liquid and PDMS surface acting at all points of contact. FN is the normal force acting at all surface contact points. Fst represents the surface tension forces stabilizing the liquid face. Fg is gravity and Fm is the inertial force tending to push the mass of liquid over the guide during injection loading. MFOC, membrane‐free organ chip; PDMS, polydimethylsiloxane.

FIGURE 3

Engineering bulk tissue vasculogenesis in milliscale MFOC. (A) Digital photograph of an MFOC loaded with cell‐laden hydrogel and culture medium. Green lines indicate the position of guide structures that confine the bulk tissue. (B) Differential interference contrast (DIC) micrograph of an MFOC loaded with HUVEC and HLF in a blend of collagen and fibrin hydrogel as described in Section 2. Guide regions are marked by green coloration between the dashed lines. The bulk tissue is visible between the guide structures. Scale bar = 500 µm. (C) 3D laser scanning confocal microscopy (LSCM) stack of a nascent vascular network in the bulk tissue after 3 days of culture. Actin in all cells is labeled with phalloidins (green) and endothelial cells are specifically labeled with UEA‐1 lectin (red). Fibroblasts are green only. Scale bar = 200 µm. (D) Formation of a continuous vascular network throughout the bulk tissue after 7 days of culture. Scale bar = 500 µm. (E) Representative stitches of UEA‐1 staining in 3D LSCM stacks along the bulk tissue length used for image analyses shown in Panel H. Scale bars = 500 µm. (F) Perivascular niche occupied by fibroblasts enrobing endothelial tubules after 7 days of culture. Scale bar = 50 µm. (G) Vascular basement membrane formation after 7 days of culture. Laminin (green), UEA‐1 lectin (red), DAPI (blue). Scale bar = 100 µm. (H) Morphometric analysis of bulk tissue vasculogenesis at Days 1, 3, and 7. N = 3 devices per time point. The number of nonparticipating endothelial cells decreases with the formation of an interconnected vascular network. Mean vessel diameter increases with cell recruitment and lumen formation in the network. Max diffusion distance increases with cell coalescence into the network and network pruning. DAPI, 4′,6‐diamidino‐2‐phenylindole; HLF, human lung fibroblast; HUVEC, human umbilical vein endothelial cell; MFOC, membrane‐free organ chip; UEA‐1, Ulex Europeas agglutinin I.

FIGURE 4

Fabricating milliscale bulk tissues with an anastomosed internal vasculature in MFOC. (A) Workflow for establishing patent and perfusable vasculature tested by FITC‐dextran perfusion as described in Section 2. (B) Differential interference contrast (DIC) image taken at the tissue interface situated on the guide structure immediately after side channel seeding (Step 2 in Panel A, after 48 h of vasculogenesis). Scale bar = 250 µm. (C) 3D LSCM stacks depicting points of vascular anastomosis at the tissue interface after 9 total days of culture. Endothelial cells are labeled with UEA‐1 lectin (red). Actin in all cells is labeled with phalloidins (green). Scale bar = 250 µm. Inset: Open face of a single anastomosed vessel. Scale bar = 20 µm. (D) DIC image taken at the tissue interface situated on the guide structure in a device without side channel seeding after 48 h. Scale bar = 250 µm. (E) 3D LCSM stack of the tissue interface in a device without side channel seeding after 48 h. Scale bar = 250 µm. Inset: Dense layers of cells cover the interface, and internal vascular structures remain inaccessible after 9 days of culture. Scale bar = 20 µm. (F) Perfusion testing with 20 kDa FITC‐dextran after 9 days. LCSM stack depicting a snapshot after 5 min of tissue perfusion. Scale bar = 100 µm. Green fluorescence is contained within vessels and absent in the interstitial spaces. Scale bar = 100 µm. LSCM, laser scanning confocal microscopy; MFOC, membrane‐free organ chip; UEA‐1, Ulex Europeas agglutinin I.

FIGURE 5

Fluid flow conditioning of anastomosed vasculature in MFOC. (A) Time‐lapse images (0, 5, 10 min) of perfusion with 40 kDa FITC‐dextran after 96 h of syringe pump‐driven flow across the bulk tissue at a rate of 20 µL/h. Scale bar = 500 µm. (B) Time‐lapse images (0, 5, 10 min) of perfusion with 40 kDa FITC‐dextran after 96 h of oscillatory rocking at an angle of 15° and a rate of 1 cycle/min. Scale bar = 500 µm. (C) Composite stitched image after 10 min of FITC‐dextran perfusion in a pumped device with patent vascularization that establishes a fluidic connection between the side channels that communicates between liquid channels. Scale bar = 1000 µm. Inset: A region of interest imaged at higher magnification with contrast enhancement reveals perfusion of finer vessels not visible in low magnification composites. Scale bar = 500 µm. (D, E) Cy5‐labeled lignin‐g‐PLGA nanoparticles accumulated in both the intravascular and extravascular spaces. Vessels stained for f‐actin (green), DAPI (blue), and UEA‐1 lectin (red). Scale bar = 200 µm (D) and 100 µm (E). (F) Morphometric analysis of FITC‐dextran perfused vessels in pumped and rocked devices did not reveal significant differences. (G) Mean fluorescence in the intravascular and extravascular spaces of LCSM stacks was calculated to assess the permeability of the perfused vasculature. Recruitment of vessels for perfusion occurs more rapidly in pumped devices compared to rocked devices and there is less diffusion into the extravascular space. Data is shown from a representative perfusion under each condition. Similar results were obtained in at least three independent replicates. DAPI, 4′,6‐diamidino‐2‐phenylindole; LSCM, laser scanning confocal microscopy; MFOC, membrane‐free organ chip; UEA‐1, Ulex Europeas agglutinin I.

FIGURE 6

Engineering patient derived TNBC models using the vascularized MFOC platform. (A) Schematic of PDX expansion, dissociation, and loading in vascular gels within MFOC. (B) Digital photograph of a device loaded with the injectable blend of disaggregated TNBC in collagen‐fibrin hydrogel (left). Scale bar = 1 mm. Differential interference contrast image of TNBC derivative cultured in MFOC without exogenous cells for 7 days (right). Scale bar = 500 µm. (C) Calcein‐AM staining (green) of TNBC derivative in MFOC after 7 days of culture. Scale bar = 200 µm. (D) Ki67 staining (red) of TNBC derivative in MFOC after 7 days of culture. Counterstained actin (green) and nuclei (blue). Scale bar = 200 µm. (E) Ki67 index denotes the percentage of proliferating cells. N = 3 biological replicates. (F) FITC‐dextran perfusion in a control MFOC without TNBC added after 7 days of culture (baseline). Scale bar = 500 µm. (G) FITC‐dextran perfusion in the TNBC model after 7 days of culture. Scale bar = 500 µm. (H) Stitched composite of ICAM‐1 (green), UEA‐1 lectin (red), and DAPI (blue) staining in a control MFOC after 7 days of culture. Scale bar = 500 µm. (I) Stitched composite image of ICAM‐1 (green), lectin (red), and DAPI (blue) staining in the TNBC model after 7 days of culture. Scale bar = 500 µm. (J) ICAM‐1 fluorescence intensity masked to endothelial cells via colocalization with UEA‐1 lectin. N = 3 biological replicates. The statistical comparison was performed using an unpaired t test with Welch's correction. (* Indicates p < 0.05). DAPI, 4′,6‐diamidino‐2‐phenylindole; LSCM, laser scanning confocal microscopy; MFOC, membrane‐free organ chip; TNBC, triple‐negative breast cancer; UEA‐1, Ulex Europeas agglutinin I.