Here, we provide a microfluidic chip and an automatically controlled, highly efficient circulation microfluidic system that recapitulates the initial microenvironment of neovascularization, allowing endothelial cells (ECs) to be stimulated by high luminal shear stress, physiological level of transendothelial flow, and various vascular endothelial growth factor (VEGF) distribution simultaneously.
Neovascularization is usually initialized from an existing normal vasculature and the biomechanical microenvironment of endothelial cells (ECs) in the initial stage varies dramatically from the following process of neovascularization. Although there are plenty of models to simulate different stages of neovascularization, an in vitro 3D model that capitulates the initial process of neovascularization under the corresponding stimulations of normal vasculature microenvironments is still lacking. Here, we reconstructed an in vitro 3D model that mimics the initial event of neovascularization (MIEN). The MIEN model contains a microfluidic sprouting chip and an automatic control, highly efficient circulation system. A functional, perfusable microchannel coated with endothelium was formed and the process of sprouting was simulated in the microfluidic sprouting chip. The initially physiological microenvironment of neovascularization was recapitulated with the microfluidic control system, by which ECs would be exposed to high luminal shear stress, physiological transendothelial flow, and various vascular endothelial growth factor (VEGF) distributions simultaneously. The MIEN model can be readily applied to the study of neovascularization mechanism and holds a potential promise as a low-cost platform for drug screening and toxicology applications.
Neovascularization happens in many normal and pathological processes1,2,3,4, which include two major processes in adults, angiogenesis and arteriogenesis5. Besides the best-known growth factors, such as vascular endothelial growth factor (VEGF)6, mechanical stimulations, in particular the blood flow induced shear stress, is important in the regulation of neovascularization7. As we know, the magnitude and forms of shear stress vary dramatically and dynamically in different parts of the vasculature, resulting in important effects on vascular cells8,9,10,11,12. Previous studies have shown that shear stress may affect various aspects of ECs, including cell phenotypic changes, signal transduction, gene expression, and the communication with mural cells13,14,15,16,17,18,19,20; hence, regulate neovascularization21,22,23,24.
Therefore, to better understand neovascularization, it is important to reconstruct the process in natural cellular microenvironment in vitro. Recently, many models have been established to create micro-vessels and provide precise control of microenvironment25,26,27, taking advantage of advances in microfabrication and microfluidic technology. In these models, micro-vessels can be generated by hydrogel28,29, polydimethylsiloxane (PDMS) microfluidic chips30,31,32 or 3D bioprinting33,34. Some aspects of the microenvironment, such as luminal shear stress22,23,35,36, transendothelial flow37,38,39,40, biochemical gradient of angiogenic factors41,42, strain/stretch43,44,45, and co-cultured with other types of cells32,46 have been mimicked and controlled. Usually, a large reservoir or syringe pump was used to provide perfused medium. Transendothelial flow in these models was created by pressure drop between the reservoir and micro-tube22,23,38,40. However, the mechanical microenvironment was hard to maintain constantly in this way. Transendothelial flow would increase and then exceed the physiological level if a high flow rate with high shear stress was used for perfusion. Previous study showed that at the initial period of neovascularization, the velocity of transendothelial flow is very low due to the intact ECs and basement membrane, usually under 0.05 µm/s8. Meanwhile, though luminal shear stress in vascular system varies greatly, it is relatively high with mean values of 5-20 dyn/cm2,11,47. For now, the velocity of transendothelial flow in previous works have been generally kept between 0.5-15 µm/s22,38,39,40, and the luminal shear stress was usually under 10 dyn/cm2 23. It remains a difficult subject to constantly expose ECs to high luminal shear stress and physiological level of transendothelial flow simultaneously.
In the present study, we describe an in vitro 3D model to mimic the initial event of neovascularization (MIEN). We developed a microfluidic chip and an automatic control, highly efficient circulation system to form perfusion micro-tubes and simulate the process of sprouting48. With the MIEN model, the microenvironment of ECs stimulated at the initial period of neovascularization are firstly recapitulated. ECs can be stimulated by high luminal shear stress, physiological level of transendothelial flow and various VEGF distribution simultaneously. We describe the steps of establishing the MIEN model in detail and the key points to be paid attention to, hoping to provide a reference for other researchers.
1. Wafer preparation
NOTE: This protocol is specific for the SU-8 2075 negative photoresist used during this research.
2. Microfluidic sprouting chip fabrication
3. Surface modification and hydrogel injection
4. Cell seeding
5. Measurement of FITC-dextran diffusional permeability
NOTE: To assess barrier function of the micro-vessel, diffusional permeability of the EC culture channel with or without cell lining is assessed.
6. Microfluidic control system setup
NOTE: The microfluidic control system in the present study is consisted of a micro-syringe pump, an electromagnetic pinch valve, a bubble trap chip, a microfluidic chip, a micro-peristaltic pump, and a reservoir. Each part of the system can be replaced by alternatives able to perform the same function.
7. Endothelial sprouting assay
NOTE: A stage top incubator assembled with phase contrast microscope is used in the present study to observe the process of sprouting in real time. The stage top incubator can maintain the temperature, humidity, and CO2 control on microscope stages, being good for live cell imaging. But the equipment is not necessary for the assay. The protocols provided here can also be worked in a basic cell incubator.
8. Data analysis
NOTE: To quantify the sprouts, the normalized area of sprouting, average sprout length, and longest sprout length were calculated. Results represent mean ± SEM obtained from three independent studies. Statistical significance (P < 0.05) is assessed by Student's t-test.
The in vitro 3D model to mimic the initial event of neovascularization (MIEN) presented here consisted of a microfluidic sprouting chip and a microfluidic control system. The microfluidic sprouting chip was optimized from previous publications22,23,37,40,51,52,53. Briefly, it contained three channels and six ports: an endothelial cell culture channel and a liquid channel with four media injection ports, and a central hydrogel channel with two hydrogel injection ports (Figure 3). The microfluidic control system consisted of a micro-syringe pump, an electromagnetic pinch valve, a bubble trap chip, a micro-peristaltic pump, and a culture medium reservoir (Figure 4). A custom program was used to control the micro-syringe pump and electromagnetic pinch valve simultaneously, with which the flow rate and circulation volume can be set up. A compensation volume was introduced to correct the slight change of volume after multiple cycles due to systemic error. To minimize the medium used for perfusion, an electromagnetic pinch valve was introduced for medium recycle. The electromagnetic pinch valve could switch between two states to make the microfluidic system in two phases in a cycle. In the injection phase, culture medium was slowly injected from the micro-syringe pump to the microfluidic sprouting chip. While in the recycle phase, culture medium was very rapidly extracted from the reservoir back to the micro-syringe pump.
The barrier function of the micro-vessel in our MIEN model was assessed by measuring the diffusional permeability coefficient (Pd) of 40 kDa FITC-dextran. As it is shown in Figure 5, the Pd of static cultured chip with cell lining is 0.1 ± 0.3 µm/s, and the Pd of empty channel without cell lining is 5.4 ± 0.7 µm/s. Then, endothelial sprouting assay under static and perfusion was performed in the MIEN model. In static conditions, endothelial sprouting occurred about 4 h after seeding and the sprouts would migrate across the central hydrogel channel to the other side in about 48 h. The sprouts degraded gradually after 48 h. While after 24 h of exposure to 5 or 15 dyn/cm2 shear stress (average 0.07 or 0.2 m/s in cell culture channel), ECs aligned in the flow direction changing from polygonal, cobblestone shape into fusiform, and the degree of sprouting decreased with the increase of shear stress (Figure 6). However, the sprouts under shear conditions were more stable than under static culture conditions. They could maintain over 48 h under perfusion. Quantitative statistics showed that endothelial sprouting decreased significantly in terms of area of sprouting, average sprout length, and longest sprout length (Figure 7).
Figure 1: Endothelial cells coat around the internal surface of the cell culture channel. HUVECs are evenly dispersed in the cell culture channel 2 h after cell seeding. Please click here to view a larger version of this figure.
Figure 2: The custom program used to control the micro-syringe pump and electromagnetic pinch valve simultaneously. Main page (up) and subpage (down) of the custom program with which the flow rate and circulation volume can be set up. A compensation volume was introduced to correct the slight change of volume after multiple cycles due to systemic error. Please click here to view a larger version of this figure.
Figure 3: The schematic of the microfluidic sprouting chip. The chip contains three channels and six ports: an endothelial cell culture channel and a liquid channel with four media injection ports, and a central hydrogel channel with two hydrogel injection ports. Please click here to view a larger version of this figure.
Figure 4: The schematic of the microfluidic control system. The microfluidic control system consists of a micro-syringe pump, an electromagnetic pinch valve, a bubble trap chip, a micro-peristaltic pump, and a culture medium reservoir. Please click here to view a larger version of this figure.
Figure 5: Diffusional permeability (Pd) of 40 kDa FITC-dextran. The Pd of static cultured chip with cell lining is 0.1 ± 0.3 µm/s, and the Pd of empty channel without cell lining is 5.4 ± 0.7 µm/s. **, p < 0.01. Please click here to view a larger version of this figure.
Figure 6: Representative images of endothelial sprouting under different shear conditions. After 24 h of static culturing, HUVECs invade from the cell culture channel into the adjacent hydrogel channel through micro-posts and a large number of sprouts in the hydrogel. While after 24 h of exposure to 5 or 15 dyn/cm2 shear stress, the degree of sprouting decrease obviously. Please click here to view a larger version of this figure.
Figure 7: Quantified area of sprouting, average sprout length, and longest sprout length. After 24 h of static culture (S0) or exposure to 5 (S5) or 15 (S15) dyn/cm2 shear stress, the degree of sprouting decrease with the increase of shear stress. *, p < 0.05; **, p < 0.01. Please click here to view a larger version of this figure.
Supplemental File. Please click here to download this File.
For a long time, real-time observation of neovascularization has been a problem. Several approaches have been developed recently to create perfused vessels lining with ECs and adjacent to extracellular matrix for sprouting22,32,40,46,54, but the mechanical microenvironment is still hard to maintain constantly. It remains a difficult subject to mimic the initial biomechanical microenvironment of ECs, which are subjected to high luminal shear stress and low velocity of transendothelial flow. Here, we presented a MIEN model that firstly simulates the initial event of neovascularization in mimic physiological microenvironment. With the MIEN model, automatic, efficient and bubble-free long-term perfusion is achieved. Luminal shear stress on ECs could be optionally changed at any time during experiments with transendothelial flow staying at physiological level.
The key of the MIEN model is decoupling the effect of luminal flow on ECs from transendothelial flow. To this end, a T-type connector is creatively inserted to the outlet port of the EC culture channel. One end of the connector is connected to the reservoir through the micro-peristaltic pump. The other end of the connector is exposed to the air of the incubator. It keeps pressure inside the endothelial cell culture channel the same as the atmospheric pressure, since there is neither surplus media gathering to increase pressure nor excess media being drawn away to form negative pressure in the channel. In this way, the flow resistance after microfluidic sprouting chip is very small so that the transendothelial flow can be maintained at physiological range even under high luminal shear stress.
One critical step within the protocol is the successful injection of hydrogel in central hydrogel channel. Huang et al. proposed that successful filling of the gels depends on balancing the capillary forces and surface tension within the microfluidic gel channel55. They found three different variables to control the balance: the spacing between micro-posts, the surface properties of the device, and the viscosity of the hydrogel precursor solutions. In the present model, the design of the microfluidic sprouting chip is optimized from previous works22,23,37,40,51,52,53. The geometry and spacing of micro-posts to separate three channels are determined based on previous calculations and numerous experiments. Further, PDL coating and high temperature baking are performed to modify the PDMS surface; hence, adjust the balance between hydrophilicity and hydrophobicity before hydrogel injection.
The other critical step of the protocol is the removal of bubbles. A large amount of air will dissolve into circulation medium during the experiment due to the T-type connector and micro-peristaltic pump, resulting in bubbles in circulation and greatly affecting the function of endothelial cells. To remove bubbles, a bubble trap chip is introduced before microfluidic sprouting chip. It works based on the high gas permeability of the PDMS membrane. When bubbles get into the trap, they are dispersed by the small and dense grid structure and trapped due to the negative pressure generated by the vacuum pump so that they fail to move forward into the microfluidic sprouting chip. Further, the bubbles transport through the PDMS membrane into the negative pressure hole connected to the vacuum pump and disappear eventually. Even so, great attention should be paid to the formation of bubbles during the experiment. As soon as bubbles are found before microfluidic sprouting chip in the pipeline, stop the micro-syringe pump immediately. Gently flinch the pipeline with the finger to let the bubbles move to bubble trap chip and be removed.
Although the initial microenvironment of neovascularization is partially recapitulated, there are some limitations to this MIEN model. The mechanical microenvironments of endothelial cells such as blood induced shear stress and extracellular matrix stiffness are changing during the neovascularization processes. Before neovascularization, the endothelial basement membrane is still intact with complete barrier function, resulting in high luminal shear stress with low velocity of transendothelial flow and interstitial flow8,9. At the onset of angiogenesis and arteriogenesis, a hallmark event is matrix metalloproteases (MMPs) mediation of basement degradation, which will increase ECs permeability and remodel matrix, leading to changes in mechanical microenvironments such as blood induced shear stress and extracellular matrix stiffness. In the present work, we focus on the initiation of neovascularization so the mechanical environments within the microfluidic sprouting chip are designed to keep stable to simulate the physiological conditions. However, changes of mechanical microenvironments will occur with the ECs sprouting, just like in vivo. Besides, similar to many previous studies34,53,56, the present model takes collagen hydrogel as an extracellular matrix. As we focus on the effect of flow induced shear stress, only one stiffness hydrogel is used. However, considering the important effect of matrix stiffness on cell behavior and neovascularization, different stiffnesses of collagen should be studied and can be achieved by regulating the pH of collagen since the mechanical properties of collagen is depending on its pH49. But it is important to note that the hydrogel needs to be thoroughly rinsed with PBS to remove residual acids or bases before cell seeding to prevent damage to the cells. Further, to mimic the complex components of ECM in vivo, various hydrogels such as Matrigel, hyaluronic acid (HA), and fibrinogen etc. should be used in the future to improve the present model. Blood pressure induced cyclic strain is another important hemodynamic force that modulates the morphology and functions of vascular cells57. Previous studies have shown that tensile strain enhanced expression of angiogenic factors in human mesenchymal stem cells58 and induced angiogenesis via degradation of type IV collagen in the vascular endothelial basement membrane59. These results indicated that blood pressure induced strain may affect neovascularization too. The present model focuses on the effect of flow induced shear stress so it isn't applicable to introduce strain as the hydrogel doesn't stick tightly enough to the channel and will fall off when stretched. We will pay attention to overcome this difficulty in future works.
The authors have nothing to disclose.
This work was supported by the National Natural Science Research Foundation of China Grants-in-Aid (grant nos. 11827803, 31971244, 31570947, 11772036, 61533016, U20A20390 and 32071311), National key research and development program of China (grant nos. 2016YFC1101101 and 2016YFC1102202), the 111 Project (B13003), and the Beijing Natural Science Foundation (4194079).
0.25% Trypsin-EDTA | Genview | GP3108 | |
Collagen I, rat tail | Corning | 354236 | |
DAPI | Sigma-Aldrich | D9542 | |
Electromagnetic pinch valve | Wokun Technology | WK02-308-1/3 | |
Endothelial cell medium (ECM) | Sciencell | 1001 | |
Fetal bovine serum (FBS) | Every Green | NA | |
Fibronectin | Corning | 354008 | |
FITC-dextran | Miragen | 60842-46-8 | |
Graphical programming environment | Lab VIEW | NA | |
Image editing software | PhotoShop | NA | |
Image processing program | ImageJ | NA | |
Isopropanol | Sigma-Aldrich | 91237 | |
Lithography equipment | Institute of optics and electronics, Chinese academy of sciences | URE-2000/35 | |
Methanol | Sigma-Aldrich | 82762 | |
Micro-peristaltic pump | Lead Fluid | BT101L | |
Micro-syringe pump | Lead Fluid | TYD01 | |
Oxygen plasma | MING HENG | PDC-MG | |
Paraformaldehyde | Sigma-Aldrich | P6148 | |
PBS (10x) | Beyotime | ST448 | |
Permanent epoxy negative photoresist | Microchem | SU-8 2075 | |
Phenol Red sodium salt | Sigma-Aldrich | P5530 | |
Polydimethylsiloxane (PDMS) | Dow Corning | Sylgard 184 | |
Poly-D-lysine hydrobromide (PDL) | Sigma-Aldrich | P7886 | |
Polytetrafluoroethylene | Teflon | NA | |
Program software | MATLAB | NA | |
Recombinant Human VEGF-165 | StemImmune LLC | HVG-VF5 | |
Sodium hydroxide (NaOH) | Sigma-Aldrich | 1.06498 | |
Stage top incubator | Tokai Hit | NA | |
SU-8 developer | Microchem | NA | |
Trichloro(1H,1H,2H,2H-perfluorooctyl)silane | Sigma-Aldrich | 448931 | |
TRITC Phalloidin | Sigma-Aldrich | P5285 |