This manuscript presents an injection molding method to engineer microvessels that recapitulate physiological properties of endothelium. The microfluidic-based process creates patent 3D vascular networks with tailorable conditions, such as flow, cellular composition, geometry, and biochemical gradients. The fabrication process and examples of potential applications are described.
In vitro platforms to study endothelial cells and vascular biology are largely limited to 2D endothelial cell culture, flow chambers with polymer or glass based substrates, and hydrogel-based tube formation assays. These assays, while informative, do not recapitulate lumen geometry, proper extracellular matrix, and multi-cellular proximity, which play key roles in modulating vascular function. This manuscript describes an injection molding method to generate engineered vessels with diameters on the order of 100 µm. Microvessels are fabricated by seeding endothelial cells in a microfluidic channel embedded within a native type I collagen hydrogel. By incorporating parenchymal cells within the collagen matrix prior to channel formation, specific tissue microenvironments can be modeled and studied. Additional modulations of hydrodynamic properties and media composition allow for control of complex vascular function within the desired microenvironment. This platform allows for the study of perivascular cell recruitment, blood-endothelium interactions, flow response, and tissue-microvascular interactions. Engineered microvessels offer the ability to isolate the influence from individual components of a vascular niche and precisely control its chemical, mechanical, and biological properties to study vascular biology in both health and disease.
The microvasculature in each organ helps define the tissue microenvironment, maintain tissue homeostasis and regulate inflammation, permeability, thrombosis, and fibrinolysis 1,2. Microvascular endothelium, in particular, is the interface between blood flow and the surrounding tissue and therefore plays a critical role in modulating vascular and organ function in response to stimuli such as hydrodynamic forces and circulating cytokines and hormones 3–5. Understanding the detailed interactions between the endothelium, blood, and the surrounding tissue microenvironment is important for the study of vascular biology and disease progression. However, progress in studying these interactions has been hindered by limited in vitro tools that do not recapitulate in vivo microvascular structure and function 6,7. As a result, the field and therapeutic advancement has relied heavily on costly and time-consuming animal models that often fail to translate to success in humans 8–10. While in vivo models are invaluable in the study of disease mechanisms and vascular functions, they are complex and often lack precise control of individual cellular, biochemical, and biophysical cues.
Vasculature throughout the body possesses a mature hierarchical structure in conjunction with expansive capillary beds, providing optimized perfusion and nutrient transport simultaneously 11. Initially, vasculature forms as a primitive plexus which reorganizes to a hierarchically branched network during early development 12,13. Although many of the signals involved in these processes are well understood 14–16, it remains elusive how such vascular patterning is determined 15. In turn, recapitulating this process in vitro to engineer organized vascular networks has been difficult. Many existing in vitro platforms to model vasculature, such as two dimensional endothelial cell cultures, lack important characteristics such as multi-cellular proximity, three dimensional luminal geometry, flow, and extracellular matrix. Tube formation assays in 3D hydrogels (collagen or fibrin) 17–19 or invasion assays 20,21 have been used to study endothelial function in 3D and their interactions with other vascular 17,22 or tissue cell types 23. However, assembled lumens in these assays lack interconnectivity, hemodynamic flow, and appropriate perfusion. Furthermore, the propensity for vascular regression in these tube formation assays 24 prevents long term culture and maturation which limits the degree of functional studies that can be performed. Thus, there is a burgeoning need to engineer in vitro platforms of microvascular networks that can appropriately model endothelial characteristics and are capable of long term culture.
A variety of vascular engineering techniques have emerged over the years for medical applications to replace or bypass affected vessels in patients with vascular disease. Large diameter vessels made from synthetic materials such as polyethylene terephthalate (PET), and polytetrafluoroethylene (ePTFE) have had considerable therapeutic success with long term patency (average 95% patency over 5 years) 25. Although small diameter synthetic grafts (< 6 mm) typically face complications such as intimal hyperplasia and thrombopoiesis 26–28, tissue engineered small diameter grafts made with biological material have made significant progress 29,30. Despite advancements of this kind, engineered vessels on the microscale have remained a challenge. To adequately model the microvasculature, it is necessary to generate complex network patterns with sufficient mechanical strength to maintain patency and with a matrix composition that allows for both nutrient permeation for parenchymal cells and cellular remodeling.
This protocol presents a novel artificial perfusable vessel network that mimics a native in vivo setting with a tunable and controllable microenvironment 31–34. The described method generates engineered microvessels with diameters on the order of 100 µm. Engineered microvessels are fabricated by perfusing endothelial cells through a microfluidic channel that is embedded within soft type I collagen hydrogel. This system has the capacity to generate patterned networks with open luminal structure, replicate multi-cellular interactions, modulate extracellular matrix composition, and apply physiologically relevant hemodynamic forces.
1. Microfabrication of Patterned Polydimethylsiloxane (PDMS) with Network Design
2. Housing Devices
3. Microvessel Device Fabrication
4. Device Analysis
The engineered vessel platform creates functional microvasculature embedded within a natural collagen type I matrix and allows for tight control of the cellular, biophysical and biochemical environment in vitro. To fabricate engineered microvessels, human umbilical vein endothelial cells (HUVECs) are perfused through the collagen-embedded microfluidic network where they attach to form a patent lumen and confluent endothelium. As illustrated in Figure 1A-C, the vessel geometry can be specifically designed to answer questions regarding flow rate, tortuosity, and branching angles, among others. Modulating the flow rate through the microvessels provides insight into the shear stress-response of endothelial cells. Previously, fabrication focus has been primarily on vessels in the 100 µm size range, however microvessels up to 500 µm or, in some cases, as small as 50 µm in diameter have been successfully made and cultured. Examples of long term patency are shown as well as the survival of microvessels cultured under gravity driven flow (Figure 2A) and continuous applied flow (Figure 2B). The in vivo-like properties of engineered microvessels can be demonstrated in several ways. Various endothelial functions can be measured using this platform, including barrier function (Figure 3A), cell-cell interactions and signaling (Figure 3B), and angiogenic remodeling (Figure 3C). A critical feature of these microvessels is their ability to respond to inflammatory stimuli in a biologically relevant manner. This is best demonstrated through their interaction with whole blood in Figure 4A-C where non-activated endothelium is quiescent (Figure 4B) and activated endothelium induces thrombus formation (Figure 4C). By culturing endothelial cells of differing origins within this platform, it is possible to understand the functional heterogeneity between endothelial cells from different sources. Figure 5A-C illustrates an example of this heterogeneity with two different human stem cell-derived endothelial cell types that when cultured in the microvessel system display drastically different phenotypes 41. By tuning the flow profiles, cell composition, and culture conditions, engineered microvessels provide a powerful in vitro tool to study endothelial biology and microvasculature in both health and disease.
Figure 1. Schematic of Network Designs and Microchannel Fabrication Protocol. The vascular network geometry can be specifically designed to generate different flow patterns. A. For example, a branched design will have decreased flow rates near the center (an 8 fold reduction in a 3 by 3 grid) resulting in regions of varied shear stress on the endothelium within the same vessel 31. B. A highly branched grid follows the same principle as the previous design but with a larger plexus. With this design, the effects of a 50 fold reduction in flow rate, resulting in a shear rate reduction from 10 to 500 sec-1 can be observed when a pressure drop of 1,000 Pa is applied across the network 32. C. More complex designs such as a tortuous network with sharp corners 32 can be used to answer questions regarding endothelial response to interruptions to laminar flow, which often occur in pathological contexts 42. Microvessels made from these patterns will have a diameter of 100 – 150 µm. D. There are three major steps during microchannel fabrication. First, the top portion of the housing jig is placed on top of the PDMS patterned network mold such that the inlet and outlet are aligned. Collagen I is then injected into the enclosed space through the injection ports (black arrow). Typically, a 0.75% collagen I mixture is used for fabrication. Successful microvessel fabrication can be achieved with as low as 0.6% collagen, however less than 0.6% typically results in insufficient mechanical strength and channel collapse. E. A thin layer of collagen is added to the bottom piece on top of the coverslip, and flattened with another PDMS piece. F. After gelation at 37 °C, the PDMS molds are removed and the top and bottom housing jigs are screwed together to enclose the network. Please click here to view a larger version of this figure.
Figure 2. Microvessels Cultured with Controlled Flow Profiles. A. HUVECs seeded in microvessels cultured under gravity driven flow and B. applied continuous flow at a rate of 3 µl/min. Culture under applied flow is best achieved with the addition of 3.5% Dextran to the media which has been shown to stabilize microfluidic collagen-based microvessels by exerting physical pressures within the lumen that enhance junction formation, although the exact mechanism driving this effect is unclear 39. Microvessels were stained for the endothelial junction protein CD31 or cluster of differentiation 31 (red) and von Willebrand factor (VWF) granules (green). A', B'. In both conditions, HUVECs expressed endothelial markers of cell-cell junctions and VWF granules. A", B". Orthogonal views show patent and rounded lumens after 6 days in culture. Please click here to view a larger version of this figure.
Figure 3. Engineered Microvessels for the Study of Endothelial Function. A. Endothelial barrier function can be assessed after perfusion of FITC (Fluorescein isothiocyanate)-conjugated Dextran through the networks (top) and subsequent measurement of its diffusion into the bulk collagen (bottom). Using custom code, videos of the perfusion can be analyzed to plot pixel intensity over a region of interest within the frame over time to determine the permeability coefficient K (µm/sec) of the FITC-Dextran. Previous publications by the lab have demonstrated that the permeability of these engineered vessels is comparable to that of ex vivo mammalian vessels 31,43. B. Endothelial cell interactions with perivascular or support cells can be analyzed in this platform by modulating the cellular composition of the collagen matrix. Here, an example of these cell-cell interactions can be seen when smooth muscle cells are embedded within the collagen surrounding the vessels. After 14 days in culture, the smooth muscle cells (stained for alpha smooth muscle actin (αSMA) in green) associate with the endothelium and extend processes along the vessel wall and act as perivascular cells as seen in the orthogonal projections 31. Images shown in panel A and B are reproductions from an earlier publication 31. C. Angiogenic sprouting and remodeling can be evaluated in engineered microvessels by modifying the culture media to include angiogenic stimuli. Without angiogenic stimuli, HUVECs show minimal sprouting (left); with angiogenic stimuli (1 µM small molecule inhibitor of GSK-3, 20 ng/ml vascular endothelial growth factor, and 20 ng/ml basic fibroblast growth factor), HUVECs readily sprout into the matrix as seen in top down projections and orthogonal views (right). The sprout length and number is easily quantifiable with ImageJ software 41. Please click here to view a larger version of this figure.
Figure 4. Blood-endothelium Interactions. Engineered microvessels are quiescent and respond appropriately to inflammatory stimuli. A. An unmodified, HUVEC cultured vessel photographed near the inlet shows strong CD31 junctional staining (red) and a round, open lumen. A'. Orthogonal view of the lumen is shown. B. When citrated whole blood with CD41a-labeled platelets (green) is perfused through a similarly quiescent vessel, minimal adhesion of platelets (green) to the endothelium is observed. C. With exposure to inflammatory stimuli such as phorbol-12-myristate-13-acetate (PMA, 50 ng/ml) to the media, perfusion of whole blood results in the formation of large thrombi (CD41a-labeled, green) within the channel and adhesion of leukocytes (CD45+ labeled, white) to the vessel wall 31. Images seen here are reproduced from an earlier publication 31. Please click here to view a larger version of this figure.
Figure 5. Microvessels Generated with Human Stem Cell Derived-ECs. Additional endothelial cell sources can be used to generate engineered microvessels. Earlier this year, Palpant et. al. demonstrated that two distinct subtypes of human embryonic stem cell-derived endothelial cells can be obtained by manipulating wnt/β-catenin signaling during differentiation: hemogenic endothelial cells (characterized by expression of HAND1 and a high hemogenic potential) and endocardial-like endothelial cells (characterized in part by expression of NFATC1 and GATA4) 41. A. When seeded and cultured in the 3D vessel platform, hemogenic ECs undergo some angiogenic sprouting. B. However, the endocardial-like ECs are much more angiogenic and migratory, suggesting functional differences, perhaps in response to flow, between the two subtypes of ECs (this work is presented in more detail in a previous publication 41). Both cell types express CD31 junctional proteins (red) and some VWF (green). Orthogonal views of both show patent lumens. C. A scanning electron microscope image of the endocardial-like ECs within the vessel shows an angiogenic sprout as seen from the luminal side. Images presented in panels A and B are reproductions from Palpant et. al. 41. Please click here to view a larger version of this figure.
Engineered microvessels are an in vitro model where physiological characteristics such as luminal geometry, hydrodynamic forces, and multi-cellular interactions are present and tunable. This type of platform is powerful in that it offers the ability to model and study endothelial behavior in a variety of contexts where the in vitro culture conditions can be matched to that of the microenvironment in question. For example, the mechanisms driving endothelial processes, such as angiogenesis, are known to occur differently in different organs and in different pathological states, such as health and disease 44. For this reason, planar endothelial cell culture is consistently inadequate 6 and as such the field has relied heavily on costly and time consuming animal models.
Animal models, while informative and essential for the progress of biological research, do not always translate well to the clinic. This has been largely attributed to differences between murine and human biology, particularly regarding their response to stimuli 9,45. It is therefore essential to be able to build in vitro models that can provide additional human-specific data to complement information gleaned from animal models. In vitro platforms have the potential to mimic the microenvironment of interest with the additional capacity to tune that environment. Key characteristics of such a system should include three dimensional structure and geometry, extracellular matrix composition (ECM), parenchymal cell proximity and interaction, blood flow, and biochemical stimuli. Engineered microvessels have the capacity to incorporate all these parameters. This can be done simultaneously to create an all-encompassing model, or added in a step wise manner to isolate individual responses and interactions.
A powerful aspect of engineered microvessels is the ability to control flow and manipulate the hydrodynamic forces exerted on the endothelium. As an example, the device can be cultured under gravity driven flow conditions or with continuous perfusion (Figure 2). The magnitude of the shear stress on the vessel wall can be modulated in both conditions through changes in flow rate and the vessel geometry. The main difference between the two conditions is the temporal distribution of shear stress within the vessel. In gravity driven flow conditions, the shear stress is not constant over time. The flow rate and shear stress are highest immediately after a media change when the inlet reservoir is full. As the media drains, the shear stress will decrease. This leads to a range of shear stress over the course of 12 hr. Experiments which require precise control over the hydrodynamic properties (for example, to study the effects of shear stress on the endothelium) should use continuous flow culture conditions.
Engineered microvessels can be tuned to answer a wide range of questions with relatively easy modifications. Different cells types, both parenchymal and endothelial, can be used to model different organ systems. In addition to fabricating microvessels with endothelial cell seeding, epithelial cells could instead be used to line the channel for studies relating to the intestinal lining, lung alveoli, kidney tubule, etc. Once the microenvironment is defined by cell composition, the nutrient solution (i.e., media, blood, growth factors, or other biologically relevant fluid) that is used to perfuse the device can be changed to answer complex questions about cell signaling, quiescence, shear stress, nutrients, drug response, and more. Furthermore, the geometry of the vessel can be specifically designed to investigate endothelial response to shear stress and tortuosity. As an example, a recent publication from Zheng et. al. showed that endothelial cells have increased VWF secretion and fiber assembly in microvessel regions with high shear stress and flow acceleration. Specifically, VWF bundles typically formed at corners and sharp turns within the network. The network geometry can also be designed to fabricate multi-cellular tubules. For example, a design that contains two parallel channels each with their own inlet and outlet could be used to generate a concentration gradient or to model an adjacent tubule environment (e.g., lymphatic vessel with adjacent microvessel). The grid-like structures of current network designs (shown in Figure 1A-C) are advantageous for computational fluid modeling and for structural integrity during fabrication, but are not indicative of vascular structure in the body. In future studies, more physiological network patterns such as ones with hierarchical branching and rounded corners should be used.
While all the steps outlined in this procedure are important, there are several critical steps that are required for successful engineered microvessel fabrication. The collagen gel must be mixed thoroughly to achieve a homogenous solution, and it is essential not to introduce bubbles during this step. This is essential for cell viability and physiological function, particularly for experiments which include parenchymal cells in the collagen matrix. If a uniform mixture of the collagen is difficult to obtain, check the pH of the solution to ensure it is neutral. If the problem persists, it's possible the collagen stock should be replaced. Another critical step is the assembly of the top and bottom housing pieces – it is essential not to use undue force as this can cause the channels to collapse. Several practice rounds may be necessary to obtain a sense of the amount of force needed to apply while rotating screws. If channel collapse or smashing continues to occur even after practice, it is possible the PDMS network has become warped due to absorption of moisture. Starting with freshly made PDMS molds will help solve this problem. Improperly threaded holes in the bottom device can also cause channel collapse. If the screws are not able to rotate smoothly during assembly, extra force must be applied often resulting in structural failure. The last critical step that should be emphasized is the importance of frequent feeding, typically every 12 hr. This is essential for maintaining the endothelium and preventing the device from drying out. In the event that seeded endothelial cells appear unhealthy or detach from the collagen wall, potential ways to improve endothelial health are to feed the vessels more often, use a syringe pump to apply continuous flow, or check the pH of the collagen gel to ensure it is at a physiological level.
Currently, this platform is limited to fabrication with extracellular matrix of appropriate stiffness to maintain structural integrity of the channel during assembly. Dense collagen at or greater than 6 mg/ml is sufficient, but collagen less than 6 mg/ml and other matrices such as decellularized matrix from whole organs are too weak. This can be overcome by using a mixture of collagen with the matrix in question. Engineered microvessels are also limited by their size scale. Vessels can be fabricated to a lower limit of about 100 µm, but substantial modifications would need to be applied to achieve a smaller scale. Although engineered microvessels create a three dimensional lumen, the network itself is planar. To create a truly three dimensional vascular plexus, additional modifications would need to be applied such as creating a multilayered network by stacking multiple engineered vessels.
The representative results presented in this method demonstrate how engineered microvessels can be used to assess barrier function, cell-cell signaling (pericyte recruitment and vascular stabilization), angiogenesis, and thrombosis. Highlights of these studies are presented here with more detailed presentations in previous publications 31,32. These studies show how pericytes migrate and coat the endothelium of engineered microvessels 31, platelets adhere to activated endothelium 31, and fluid shear stress modulates endothelial activation and VWF secretion and assembly 32. Engineered microvessels can further be used to build vascularized tissues and model organ-specific systems, such as the kidney 33 or the heart 34. The ability to replicate these physiological blood-endothelium interactions and tune the microenvironment with respect to shear stress, geometry, ECM, and cellular composition will enable future studies of vascular diseases.
The authors have nothing to disclose.
The authors would like to acknowledge the Lynn and Mike Garvey Imaging Laboratory at the Institute for Stem Cell and Regenerative Medicine as well as the Washington Nanofabrication Facility at the University of Washington. They also acknowledge the financial support of National Institute of Health grants DP2DK102258 (to YZ), and training grants T32EB001650 (to SSK and MAR) and T32HL007312 (to MAR).
Wafer Fabrication | |||
AutoGlow Plasma System | AutoGlow | ||
Headway Spin Coater | Headway Research, Inc | PWM32 Spin Coater | |
ABM Contact Aligner | AB-M | ||
Alpha Step Profilometer | Tencor | Alpha Step 200 | |
SU-8 Developer | Microchem | Y020100 | |
SU-8 Resist | Microchem | SU-8 2000 | |
8" silicon wafer | Wafer World Inc. | ||
Tabletop Micro Pattern Generator | Heidelberg Instruments | μPG 101 | For generation of photomask |
Hot plate | VWR | 97042-646 | |
Ispropyl alcohol | Avantor Performance Materials | 9088 | |
Petri dishes (120 x 120 mm, square) | Sigma-Aldrich | Z617679 | |
Trichloro(3,3,3-trifluoropropyl)silane | Sigma-Aldrich | MKBG3805V | |
Polydimethylsiloxane (PDMS) elastomer base and curing agent | Dow Corning | Sylgard 184 | Mixed at 10:1 (w/w) |
Vacuum desiccator | Sigma-Aldrich | Z119024-1EA | |
Oven | VWR | 9120976 | |
Device Fabrication and Culture | |||
poly(methyl methacrylate) (PMMA) | Plexiglas | ||
Corona Treater | Electro-Technic Products, Inc. | BD-20 | Handheld device for plasma treatment of PMMA devices and PDMS molds |
Soldering Iron | Weller | WTCPS | |
Stainless Steel Truss Head Slotted Machine Screw | McMaster-Carr | 91785A096 | |
Stainless steel dowel pins | McMaster-Carr | 93600A060 | |
Tweezers | Miltex | 24-572 | Any similar tweezers may be used |
Spatula (Micro Spoon) | Electron Microscopy Services | 62410-01 | |
Screw driver | Any flat head screwdriver may be used, autoclaved | ||
Glass coverslips (22 x 22 mm) | Fisher Scientific | 12-542B | |
Bleach | Clorox | 4460030966 | |
Petri dishes (150 X 25mm) | Corning | 430599 | |
Petri dishes (100 X 20 mm) | Corning | 2909 | |
Cotton, cut into 1 cm x 3 cm pieces | Autoclaved | ||
Polyethyleneimine (PEI) | Sigma-Aldrich | P3143 | Dilute to 1% in cell culture grade water |
Glutaraldehyde | Sigma-Aldrich | G6257 | Dilute to 0.1% in cell culture grade water |
Sterile H2O | Autoclaved DI H2O | ||
Type I collagen, dissolved in 0.1% acetic acid | Isolated from rat tails as described in Rajan et. al. 2006 (ref #37) | ||
1 mL syringe | BD | 309659 | |
10 mL syringe | BD | 309604 | |
15 mL conical tubes | Corning | 352097 | |
30 mL conical tubes | Corning | 352098 | |
M199 10X Media | Life Technologies | 11825-015 | |
1N NaOH (sterile) | Sigma-Aldrich | 415413 | Dilute to 1N in cell culture grade water |
HUVECs | Lonza | ||
Endothelial growth media | Lonza | CC-3124 | |
Trypsin | Corning | 25-052-CI | |
Fetal bovine serum (FBS) | Thermofisher Scientific | 10082147 | |
Dextran from Leuconostoc spp. (70kDa) | Sigma-Aldrich | 31390 | |
Phosphate Buffered Saline (PBS) | Corning | 21-031-CV | |
Hemocytometer | Hausser Scientific Co. | 3200 | |
Gel loading tips | VWR | 37001-152 | |
18G Blunt Fill Needle | BD | 305180 | |
20G Stainless Steel Dispensing Needle | McMaster-Carr | 75165A123 | |
Tygon 1/32” ID, 3/32" OD Silicon Tubing | Cole-Parmer | EW-95702-00 | |
1/16" Tube-to-tube Coupling | McMaster-Carr | 5116K165 | |
90° Elbow Connectors, Tube-to-Tube | McMaster-Carr | 5121K901 | |
Luer Lock Coupling (Female, 1/16" ID) | McMaster-Carr | 51525K211 | |
Plastic Forceps, with Jaw Grips | Electron Microscopy Services | 72971 | |
Dual Syringe Pump | Harvard Apparatus | 70-4505 | |
5 mL Polystyrene Round-bottom tube | Fisher Scientific | 14-959-2A | |
Device Analysis | |||
Formaldehyde | Sigma-Aldrich | F8775 | |
Bovine Serum Albumin | Sigma-Aldrich | A8806-5G | |
Triton X-100 | Sigma-Aldrich | T-9284 | |
Rabbit anti-hCD31 | Abcam | ab32457 | 1:25 working dilution |
FITC conjugated anti-von Willebrand Factor antibody | Abcam | ab8822 | 1:100 working dilution |
Goat anti-rabbit 568 secondary antibody | Thermofisher Scientific | A-11011 | 1:100 working dilution |
Hoescht | Thermofisher Scientific | H1399 | Resuspended in DMSO |
Sodium cacodylate | Sigma-Aldrich | C0250 | To make 0.2M cacodylate buffer |
Ethanol | VWR International | BDH1164-4LP | |
40kDa FITC-conjugated Dextran | Sigma-Aldrich | FD40S | |
Additional Culture Reagents | |||
CHIR-99021 | Selleck Chem | S2924 | Small molecule GSK-3 inhibitor |
Human recombinant VEGF | Peprotech | 100-20 | |
Human recombinant bFGF | Peprotech | AF-100-18B |