Primary human umbilical vein endothelial cells (HUVECs) were grown to confluence within a microfluidic network device. The endothelial cell junction and F-actin distributions were illustrated and the changes in intracellular calcium concentration and nitric oxide production in response to adenosine triphosphate (ATP) were quantified in real-time at individual cell levels.
Endothelial cells (ECs) lining the blood vessel walls in vivo are constantly exposed to flow, but cultured ECs are often grown under static conditions and exhibit a pro-inflammatory phenotype. Although the development of microfluidic devices has been embraced by engineers over two decades, their biological applications remain limited. A more physiologically relevant in vitro microvessel model validated by biological applications is important to advance the field and bridge the gaps between in vivo and in vitro studies. Here, we present detailed procedures for the development of cultured microvessel network using a microfluidic device with a long-term perfusion capability. We also demonstrate its applications for quantitative measurements of agonist-induced changes in EC [Ca2+]i and nitric oxide (NO) production in real time using confocal and conventional fluorescence microscopy. The formed microvessel network with continuous perfusion showed well-developed junctions between ECs. VE-cadherin distribution was closer to that observed in intact microvessels than statically cultured EC monolayers. ATP-induced transient increases in EC [Ca2+]i and NO production were quantitatively measured at individual cell levels, which validated the functionality of the cultured microvessels. This microfluidic device allows ECs to grow under a well-controlled, physiologically relevant flow, which makes the cell culture environment closer to in vivo than that in the conventional, static 2D cultures. The microchannel network design is highly versatile, and the fabrication process is simple and repeatable. The device can be easily integrated to the confocal or conventional microscopic system enabling high resolution imaging. Most importantly, because the cultured microvessel network can be formed by primary human ECs, this approach will serve as a useful tool to investigate how pathologically altered blood components from patient samples affect human ECs and provide insight into clinical issues. It also can be developed as a platform for drug screening.
Endothelial cells (ECs) lining the blood vessel walls in vivo are constantly exposed to flow, but cultured ECs are often grown under static conditions and exhibit a pro-inflammatory phenotype1,2. The microfluidics technology enables a precisely controlled fluid through a geometrically constrained microscale (sub-millimeter) channels3, which provides the opportunity for cultured cells, especially for vascular ECs, to grow under desired flow conditions. These features make the cell culture conditions closer to in vivo than the conventional, static 2D cell cultures. They are extremely important when the microfluidic devices are used to model different types of vasculatures and to study EC responses to mechanical and/or chemical stimulations.
Despite the advantages demonstrated by the microchannel network over a static cell culture, the adaptation and application of microfluidics in the biomedical field remain limited. Reported by a recent review, the majority of the publications of this field (85%) are still in engineering journals4. The performance of microfluidic devices has not been convincing enough for most biologists to switch from current techniques such as the Transwell assay and the macro-scale culture dish/glass slide to this miniaturized device. Microfluidics is a multidisciplinary field, which requires interdisciplinary collaborations to move this field forward. The objective of this technical article is to reduce the knowledge gaps between disciplines and make the fabrication procedures understandable by biologists, while providing biological application and functional validation of the microfluidic microvessels. The visualized experimental protocols include fabrication of both microfluidic devices and their biological utilities, which represents a close collaboration between engineers and biologists.
We recently reported some biological applications using the in vitro microvessel network with microfluidic device5. In order to appropriately design the dimensions of the microchannel network and apply the desired shear stress, a numerical model was built with computational fluid dynamic software to closely estimate the flow profile. Primary human umbilical vein endothelial cells (HUVECs) that were seeded into the microchannels reached confluence, i.e. covered the entire inner surfaces of the microchannel, in 3-4 days with continuous perfusion. The proper barrier formation was demonstrated by VE-cadherin staining and compared with those formed under static cell culture conditions and in intact microvessels. By applying the experimental protocols developed in individually perfused intact microvessels6-8, we quantitatively measured the changes in EC [Ca2+]i and nitric oxide (NO) production in response to adenosine triphosphate (ATP) with fluorescent indicators and confocal and conventional fluorescence microscopy. The agonist-induced increases in EC [Ca2+]i and NO production have been reported as necessary intracellular signals for inflammatory mediator-induced increases in microvessel permeability6-15. Although some previous studies showed images of DAF-2 DA loaded microfluidic devices16,17, appropriate resolution and data analysis had not yet achieved18. To our knowledge, this study demonstrates the first quantitative measurements of agonist-induced dynamic change in endothelial [Ca2+]i and NO production utilizing microfluidic based system.
Microfabrication techniques have the flexibility to fabricate microchannels down to a few microns and enable the development of complex patterns to mimic the geometries of in vivo microvasculature. Here we presented a typical microchannel network with three levels of branching. This network is fabricated by the combination of photolithography which is performed in a microfabrication cleanroom and soft lithography.
1. Microfluidic Device Fabrication
Supplemental Video 1 (Right click to download).
Supplemental Video 2 (Right click to download).
2. Endothelial Cell Seeding and Culture
3. Immunofluorescent Staining
4. Fluorescence Imaging of Endothelial [Ca2+]i
5. Fluorescence Imaging of Nitric Oxide Production
6. Data Analysis
This section shows some of the results obtained with the cultured microvessel network developed with this protocol. The microchannel pattern is a three level branching network (Figure 1A). In this design, a 159 µm wide mother channel branches into two 126 µm wide channels, and branches again into four 100 µm wide daughter channels. A 3D numerical simulation was performed to estimate the shear stress distributions under the flow rate of 0.35 µl/min (Figure 1B), which indicates three different levels of shear stress within this microchannel network. The laminar flow within the microchannels is predicted to be hydrodynamically stable without the presence of disturbed or secondary flow. After a SU-8 master mold was fabricated with a standard photolithography process (Figure 1C), a PDMS device was fabricated through soft lithography to replicate the microchannel network design into a transparent, air permeable scaffold (Figure 1D–E). After EC seeding (Figure 1F–G) and 3-4 days of continuous perfusion, ECs reached confluence and successfully covered the inner surfaces of the microchannels.
EC F-actin and nuclei were fluorescently labeled within the network and illustrated with confocal images. Under the shear stress of 1.0 dyne/cm2, about 30% of the HUVECs elongated along the flow direction with increased central stress fibers, whereas the rest 70% maintained its cobblestone pattern with dominated peripheral F-actin5. The shear-induced cell shape change appears less prominent than that commonly observed in 2D static cell cultures. Actually under in vivo conditions, the ECs have different cell shapes in different types of vessels, but not easily change cell shape in response to the flow changes. More studies are needed to determine whether this difference was the results of the channel geometry and perfusion environment that are closer to in vivo conditions.
We also successfully measured ATP-induced changes in EC [Ca2+]i and NO production (Figures 4 and 5). ATP induced transient increases in EC [Ca2+]i and NO production. The mean peak FIfluo-4 was 187 ± 22% of the baseline, which occurred at 35 ± 10 sec after the start of ATP perfusion. The NO production rate was increased from a basal level of 0.15 ± 0.05 AU per min to 1.18 ± 0.37 AU per min within the first five minutes of ATP exposure. All the results were reported as the mean ± standard error (SE). They are comparable to the results derived from individually perfused intact venules6,9,14,19.
Figure 1: Microfluidic Device Fabrication and EC loading. (A) The schematic view of the microchannel network. The width and bifurcation angle of the microchannels are indicated at each branching levels. (B) The numerical simulation of shear stress distribution. The flow rate is 0.35 µl/min. The height of the microchannel after development is 80 µm. These two figures have been modified from previous publication with additional details5. (C) The development of the master mold. The master mold of the microchannel network is fabricated with SU-8 photoresist on silicon wafer. (D) The process of soft lithography. PDMS mixture is casted onto the master mold and cured in an oven. (E) Bonding PDMS channel to the substrate. The substrate used for this device is a coverslip spin coated with a thin layer of PDMS. The inlet and outlet are created with a hole-puncher. (F and G) EC loading. About 10 µl of cell solution is placed at the inlet. A gentle flow is then induced by placing a Pasteur pipette at the outlet via capillary flow. Please click here to view a larger version of this figure.
Figure 2: Confocal images of EC F-actin in cultured microchannel network. (A) A schematic view of the network design. The selected regions shown in B–D are indicated in the graph. (B–D) Confocal images of EC F-actin (red) and nuclei (blue) of the in vitro microvessel network. B1, C1 and D1 are the projected images from top half of the microchannel image stacks, and B2, C2, and D2 are the image projections from lower half of the image stacks. E. The cross-sectional view indicated as 1-1', 2-2', and 3-3' in B1–D1, Scale bar = 100 µm. These images are adapted from previous publication5. Please click here to view a larger version of this figure.
Figure 3: Comparisons of VE-cadherin distribution in cultured microvessel network with statically cultured EC monolayer and intact rat mesenteric venule. (A) The schematic view of the network design with selected regions shown in B–D. (B–D) VE-cadherin (green) and cell nuclei (blue) staining of the in vitro microvessel network. D1 and D2 are the top and bottom of the projected images collected from the same bifurcation region, respectively. (E) VE-cadherin staining of statically cultured ECs. (F) VE-cadherin staining of individually perfused intact rat venule. This figure published previously5. Please click here to view a larger version of this figure.
Figure 4: ATP-induced increases in EC [Ca2+]i. Experiments were conducted from 4 devices and 7 to 12 ROIs (individual ECs) per device were selected for data analysis. (A) The time course of ATP (10 µM)-induced changes in EC [Ca2+]i (Fluo 4 FI/FI0×100% ± SE) from one representative experiment. (B) The images of the representative experiment (data is shown in A). This is a previously published figure5. Please click here to view a larger version of this figure.
Figure 5: ATP-induced NO production. Experiments were conducted in 3 devices and 11 to 12 ROIs (individual ECs) per device were selected for data analysis. (A) Time dependent changes of DAF-2 fluorescence intensity (FIDAF ± SE, left Y axis) in ECs under basal conditions and after exposure to ATP. The NO production rate (df/dt, dotted line), was derived from the first differential conversion of the accumulated FIDAF (right Y axis). (B) Representative images from one experiment. The FIDAF further increased after a NO donor, sodium nitroprusside (SNP), was added to the perfusate, indicating a sufficient amount of intracellular DAF-2 to react with NO. This is a previously published figure5. Please click here to view a larger version of this figure.
In this article, we present detailed protocols for the development of cultured microvessel network, the characterization of EC junctions and cytoskeleton F-actin distribution, and the quantitative measurements of EC [Ca2+]i and NO production using a microfluidic device. The perfused microfluidic device provides an in vitro model that allows a close simulation of the in vivo microvascular geometries and shear flow conditions. Since the cultured microvessel network can be formed by primary human ECs, this approach can serve as a useful tool to investigate how pathologically altered blood components from patient samples affect human ECs by direct perfusion of patient blood samples to the cultured microvessels and provide insight into clinical issues. The human ECs developed microvessels can also be used for drug screening to avoid the potential discrepancies in drug responses between human and animal tissues.
The substrate of the microfluidic devices shown in this protocol is a glass cover-slip (150-180 µm) spin-coated with a thin layer of PDMS. This thin layer of PDMS is essential for high resolution and live tissue real time imaging, because the larger numerical aperture objectives usually has short working distance. The laminar flow within the microchannel (submillimeter range) has a very low Reynolds number, and the pressure differential is linearly related to the flow, thus the shear stress distribution is solely dependent on the channel geometry. Therefore, it can be well-controlled by high-accuracy perfusion pump20. In our present study, the shear stress applied to the HUVEC cultured microvessel network was between 1 to 10 dyne/cm2, which is within the range of shear stress measured in the venous system21,22. The macro-fluidic systems such as parallel flow chamber lacks complex flow patterns. In larger scale flow chambers (mm-cm range), ECs are typically only grown in the bottom surface (2D monolayer), lacking the geometric similarities to microvessels in vivo. When the width of the flow chamber is over 2 mm, the homogeneous flow is confined in a region that is usually several millimeters away from the outlet/inlet, and a couple hundred microns away from the side wall23. When the width of the chamber is increased to the range of centimeters, the flow will differentiate into various streamlines and the flow velocities will change across the whole area24. Additionally, the larger scale flow chamber consumes at least 100 times more perfusate to achieve the same shear stress as that in microchannels. The shear stress distribution within the microchannel network can be simulated using a numerical modeling tool, and the stationary incompressible Navier-Stokes equations can be solved by most finite element software.
The microfluidic device can be made with a single mask microfabrication process. The success of cultured microvessel development, however, requires special attentions to the details. After loading the solution into the microchannels, each device needs to be checked for bubble trapping under the microscope before EC seeding. If a bubble occurs, it is still possible to force it out at this stage by applying a hydraulic pressure at the inlet with a closed outlet, as PDMS is an air-permeable material. To maintain a desired perfusion rate, tight fitting without a leak is necessary for all of the needle/tubing connections. The matched hole-puncher and tubing size are critical to achieve a tight fitting. To assure a precise delivery of the flow rate, the syringe on the perfusion pump and the device should be placed at the same level. To change the perfusate, the inserted tubing should be gently disconnected from the device to avoid any stretching. To increase shear stress, step increases (such as 10 steps of increase in 18 hr) are recommended to avoid sudden flow change caused EC peeling.
Previous studies performed on individually perfused intact microvessels indicated that agonist-induced increases in EC [Ca2+]i, and the subsequent endothelial nitric oxide synthase (eNOS) activation and NO production are necessary signaling for increased microvascular permeability under inflammatory conditions6,7,9-13,25. In this protocol, we select ATP as the representative agonist, since it is commonly released by red blood cells, aggregated platelets, and injured tissues or under inflammatory conditions in vivo. A fluorescence calcium indicator, Fluo-4 AM, was used to measure the changes in endothelial [Ca2+]i after exposure to ATP. The EC [Ca2+]i responses in cultured microvessels are similar to those found in intact microvessels7,9. Measuring EC NO production has been technically challenging for biologists. To date, there has been no fluorescent indicator that enables the detection of dynamic changes in NO concentration in a similar manner to the calcium indicators. DAF-2 DA has been widely used to assess NO production. However, the proper usage, data interpretation, and data analysis need to draw users' attention. Since the chemical transformation of DAF-2 to DAF-2T in the presence of NO is irreversible (one way conversion)26, the detected DAF fluorescence intensity profile does not represent the changes in NO concentration. Instead, it indicates the accumulated DAF-2T production with time. Based on this cumulated DAF-2 fluorescence (FIDAF) curve, the NO production rate (df/dt) can be derived by the first differential conversion of the FIDAF7,11. Namely, the slope of the FI curve indicates the NO production rate and the plateau phase indicates no further increased NO production. After the conversion, the data generated from this protocol presented both the basal NO production rate and the changes in NO production in response to ATP with temporal and spatial resolution at individual EC levels within the microvessel network.
Currently PDMS has been widely used for microfluidic applications, as it is an optically transparent, nontoxic, and gas permeable soft elastomer27-30. However, since PDMS is not water permeable, we could not directly measure the permeability of EC monolayer. To overcome this limitation, some applications modified the channel wall structures such as integrating a permeable membrane into the channel31, or creating a permeable "window openings" with micro-pillars or micro-holes32,33, while others have been focused on the modification of device materials such as using the hydrogel devices or hydrogel/PDMS hybrid devices17,34-36. The drawbacks of such devices are their vulnerable nature to dehydration and their relatively weak mechanical properties to stand stress or pressure. Future development of permeable material with mechanical strength will further improve the device and its potential for broader biological applications.
The authors have nothing to disclose.
This work was supported by National Heart, Lung, and Blood Institute grants HL56237, National Institute of Diabetes and Digestive and Kidney Diseases Institute DK97391-03, National Science Foundation (NSF-1227359 and EPS-1003907).
ATP | Sigma-Aldrich | A2383 | |
Acetone | Fisher Scientific | A929 | |
Biopsy punch | Miltex | 33-31 AA | |
Bovine Albumin | MP Biomedicals | 810014 | |
Bovine Brain Extract (BBE) | Lonza | CC-4098 | |
Cover-slip | Fisher Scientific | 12-542C | |
DAF-2 DA | Calbiochem | 251505 | |
Dextran | Sigma-Aldrich | 31390 | |
Donkey anti-Goat IgG (H+L) Secondary Antibody | Life technologies | A-11055 | |
DPBS, no calcium, no magnesium | Gibco | 14190-250 | |
DRAQ5 (nuclei staining) | Cell Signaling Technology | 4084 | |
Endothelial Cell Growth Supplement (ECGS) | Sigma-Aldrich | E2759-15MG | |
Fetal Bovine Serum | Gibco | 16000-044 | |
Fibronectin | Gibco | PHE0023 | |
Fluo-4 AM | Life technologies | F-14201 | |
Gelatin from porcine skin | Sigma-Aldrich | G1890-100G | |
Gentamicin (50 mg/mL) | Gibco | 15750-078 | |
Glass coverslip | Fisher Scientific | 12-548B | |
Glass Pasteur pippette | VWR | 14672 | |
Heparin sodium salt from porcine intestinal mucosa | Sigma-Aldrich | H3393-10KU | |
HEPES Buffered Saline Solution | Lonza | CC-5024 | |
Human umbilical vein endothelial cells (HUVECs) | Lonza | CC-2517 | |
Isopropyl alcohol (IPA) | VWR | 89125 | |
L-Glutamine (200 mM) | Gibco | 25030-081 | |
Mammalian Ringer Solution Ingredient | |||
NaCl (132 mM) | Fisher Scientific | S671-3 | |
KCL (4.6 mM) | Fisher Scientific | P217-500 | |
CaCl2 · 2H2O (2.0 mM) | Fisher Scientific | C79-500 | |
MgSO4 ·7H2O (1.2 mM) | Fisher Scientific | M63-500 | |
Glucose(5.5 mM) | Fisher Scientific | BP350-1 | |
NaHCO3 (5.0 mM) | Fisher Scientific | S233-500 | |
Hepes Salt (9.1 mM) | Research Organics | 6007H | |
Hepes Acid (10.9 mM) | Research Organics | 6003H | |
MCDB 131 Culture Medium | Life technologies | 10372-019 | |
Paraformaldehyde | Electron Microscopy Sciences | 15710 | |
Phalloidin (F-actin staining) | Sigma-Aldrich | P1951 | |
Phosphate Buffered Saline | Life technologies | 14040-133 | |
Polydimethylsiloxane (PDMS) | Dow Corning Corporation | Sylgard 184 | |
Scalpel | Exel Int | 29552 | |
Scotch tape | 3M | 34-8711-3070-3 | |
Silicon wafer | VWR | 14672 | |
SU-8 photoresist | MicroChem | SU-8 2050 Y111072 | |
SU-8 developer | MicroChem | Y020100 | |
tissue culture flasks | Sigma-Aldrich | Z707503-100EA | |
Triton X-100 | Chemical Book | T6878 | |
Trypsin Neutralizer solution | Gibco | R-002-100 | |
Trypsin/EDTA Solution (TE) | Gibco | R-001-100 | |
Tubing | Cole-Parmer | PTFE microbore tubing, 0.012" ID x 0.030" OD | |
VE-cadherin | Santa Cruz Biotechnology | SC-6458 | |
Name of Equipment | Company | Catalog Number | Comments/Description |
Biosafety Laminar hood | NuAire | NU-425 Class II, Type A2 | |
CCD camera | Hamamatsu | ORCA | |
Confocal microscope | Leica | TCS SL | |
Desiccator | Bel-Art | F42022 | |
Hotplate | Wenesco | HP-1212 | |
Incubator | Forma Scientific | 3110 | |
Isotemp oven | Barnstead | 3608-5 | |
Lithography bench | Karl Suss | MA6 Contact Lithography | |
Optical microscope | Nikon | L200 ND & Diaphod 300 | |
Shutter for the CCD camera | Sutter Instrument | Lambda 10-2 | |
Plasma cleaner | PVA TePla/Harrick plasma | M4L/PDC-32G | |
Spin coater | Brewer Science | Cee 200X | |
Syringe pump system | Harvard Apparatus | 703005 | |
Name of Software | Company | Catalog Number | Comments/Description |
CAD software | Autodesk | AutoCAD 2015 | |
CFD simulation software | COMSOL | COMSOL Multiphysics 4.0.0.982 | |
Images acquire and analyse for NO production | Universal Imaging | Metafluor |