Microfluidic assay of endothelial cell migration in 3D ... - Springer Link

Biomed Microdevices (2011) 13:717–723 DOI 10.1007/s10544-011-9541-7

Microfluidic assay of endothelial cell migration in 3D interpenetrating polymer semi-network HA-Collagen hydrogel Gi Seok Jeong & Gu Han Kwon & Ah Ran Kang & Bo Young Jung & Yongdoo Park & Seok Chung & Sang-Hoon Lee

Published online: 15 April 2011 # Springer Science+Business Media, LLC 2011

Abstract Cell migration through the extracellular matrix (ECM) is one of the key features for physiological and pathological processes such as angiogenesis, cancer metastasis, and wound healing. In particular, the quantitative assay of endothelial cell migration under the well-defined three dimensional (3D) microenvironment is important to analyze the angiogenesis mechanism. In this study, we report a microfluidic assay of endothelial cell sprouting and migration into an interpenetrating polymer semi-network HA-Collagen (SIPNs CH) hydrogel as ECM providing an enhanced in vivo mimicking 3D microenvironment to cells. The microfluidic chip could provide a well-controlled gradient of growth factor to cells, whereas the hydrogel could mimic a well-defined 3D microenvironment in vivo. Electronic supplementary material The online version of this article (doi:10.1007/s10544-011-9541-7) contains supplementary material, which is available to authorized users. G. S. Jeong : G. H. Kwon : A. R. Kang : S.-H. Lee (*) Department of Biomedical Engineering, College of Health Science, Korea University, 1-boneji San, Jeongneung-dong, Seongbuk-gu, Seoul 136-100, Korea e-mail: [email protected] B. Y. Jung : Y. Park Biomedical Engineering, Brain Korea 21 Project for Biomedical Science, Korea University Medical College, Seoul, Korea S. Chung (*) School of Mechanical Engineering, Korea University, Anam-dong, Seongbuk-gu, Seoul 136-713, Korea e-mail: [email protected]

(In addition/Furthermore, the microfluidic chip gives a well-controlled gradient of growth factor to cells) For this reason, three types of hydrogel, composed of semiinterpenetrating networks of collagen and hyaluronic acid were prepared, and firstly we proved the role of the hydrogel in endothelial cell migration. The diffusion property and swelling ratio of the hydrogel were characterized. It modulated the migration of endothelial cells in quantified manner, also being influenced by additional synthesis of Matrix metalloproteinase(MMP)-sensitive remodeling peptides and Arginine–glycine–lycinee (RGD) cell adhesion peptides. We successfully established a novel cell migration platform by changing major determinants such as ECM material under biochemical synthesis and under growth factor gradients in a microfluidic manner. Keywords Cell migration . 3D matrix . Collagen . Hyaluronic acid . Semi-interpenetrating networks . MMP

1 Introduction In the engineering of 3D tissue, vascularization via the 3D migration of endothelial cells (ECs) into the extracellular matrix (ECM) is one of critical issues (Yamada and Cukierman 2007; Even-Ram and Yamada 2005; Cukierman et al. 2001). The 3D migration strongly depends on various matrix composition (Chung et al. 2009a; Kim et al. 2007, 2008), remodeling activity (Kim et al. 2008; Atkinson and Senior 2003), and mechanical and chemical cues (Decaestecker et al. 2007; Lamalice et al. 2007; Shizukuda et al. 1999). To date, the interplay between these environmental influences and cellular migration, which is referred to as chemotaxis, haptotaxis or mechanotaxis, has been exten-

718

sively studied (Decaestecker et al. 2007; Lamalice et al. 2007; Shizukuda et al. 1999) under the 2D culture environment. In contrast, the migration studies under the 3D environment have been limited due to difficulty in well-organization of 3D environments, while recent advances in microfluidic technology has enabled fine control of 3D chemical microenvironments for cell study (Park et al. 2010; Bauer et al. 2010). Among the achievements, hydrogel incorporating microfluidic cell culture platform developed by the authors has attracted much attention because it provides in vivo mimicking microenvironments and has been actively applied to investigate 3D morphogenesis of various cell types under chemical gradients (Chung et al. 2009a, b; Sudo et al. 2009). In these studies, type 1 collagen was used, a popular 3D ECM abundant in connective tissues, cartilage, ligaments, bone, and skin (Fratzl 2008). Its fibrous structure is advantageous for cell adhesion, but makes it difficult to modify its uniform bulk properties, limiting achievement of structural properties that is similar to architecturally complex native tissues (Park et al. 2002). Therefore, an alternative and biochemically modifiable ECM hydrogel has been highly required for broad biomedical applications. Hyaluronic acid (HA) is one of major ECM components and has been commonly used as scaffold in tissue engineering (Brigham et al. 2008). HA is abundant in the connective tissue of mammals and plays a key role in homeostasis, angiogenesis, inflammation, and cell migration. HA can be chemically modified with diverse biomolecules by covalent bonds (Kim et al. 2007; Park et al. 2003), and has been identified as a tissue remodeling scaffold, i.e. bone, heart, and brain tissue regeneration (Kim et al. 2007, 2008, 2009; Park et al. 2004, 2009). MMP-sensitive polyethylene glycol(PEG)-based hydrogel was used to modulate the function of bovine primary chondrocytes (Park et al. 2004) and mesenchymal stem cells(MSCs) (Kim et al. 2007), and other applications of tissue regeneration (Lutolf et al. 2003). Despite of the several advantages of HA, the cell adhesion and migration on the HA is inferior to collagen due to its hydrophilic properties (Nguyen and West 2002; Hoffman 2001). An optimally characterized semi-interpenetrating network (SIPNs) of collagen and HA (CH) could be the suitable solution to address the limits of HA. Although the composite 3D scaffolds with HA and collagen have been extensively studied for controlling cellular behavior of adhesion (Brigham et al. 2008), viability (Brigham et al. 2008; Suri and Schmidt 2009) and cytotoxicity (Park et al. 2002) of fibroblasts, the quantitative assay of cellular behavior in the SIPNs CH hydrogel embedded into microfluidic channel which regulate the microenvironment to cells has not been reported. In this study, we report a microfluidic assay of endothelial cell sprouting and migration into the SIPNs

Biomed Microdevices (2011) 13:717–723

CH hydrogel as ECM, providing a more in vivo mimicking 3D microenvironment to cells. The microfluidic chip could provide a well-controlled gradient of growth factors to cells, while the hydrogel incorporated inside could provide well-defined 3D microenvironment mimicking in vivo. Three types of SIPNs CH hydrogel with different composition were introduced in the microfluidic platform, and first we characterized their mechanical/chemical properties. Diffusion property and swelling ratio were quantitatively measured providing important information of diffused molecule gradient. The SIPNs CH was successfully injected, positioned and polymerized at intended position forming 3D microenvironment to the cells cultured together. We quantified endothelial cell sprouting and migration into the polymerized SIPNs CH hydrogel, and also found that the polymerized hydrogel could modulate the migration of endothelial cells. We also identify the role of MMPsensitive synthesis of HA in the SIPNs CH hydrogel by adding cell adhesion peptides such as RGDS and/or remodeling peptides by MMPs on the sprouting and migration of the endothelial cells.

2 Materials and methods 2.1 Preparation of HA and collagen The HA (molecular weight (MW): 230 kDa) hydrogel was gelled using acrylated hyaluronic acid and matrix metalloproteinase (MMP)-sensitive and insensitive peptides (Kim et al. 2007). First, two separate solutions were prepared: one containing acrylated hyaluronic acid dissolved in a triethanolamine (TEA)-buffered solution (0.3 M, pH 8), and the other containing a dispersion of MMP-sensitive peptides (GCRDGPQGIWGQDRCG) and MMP-insensitive peptides (GCRDGDQGIAGFDRCG) desolved in in the same solution. Then, the MMP-peptides solution was added as a cross-linker to the hyaluronic acid solution, maintaining a 1:1 molar ratio of acryl and thiol groups. Type I collagen was commercially obtained (BD Biosciences, MA, USA). Gel stiffness varied with the pH of the pre-polymerized collagen solutions over the range pH 7.4–11 (Chung et al. 2009a). In this study, a pH 7.4 collagen solution was utilized for SIPNs CH polymerization, similar pH to that in vivo. The collagen was diluted in sterile water and the pH level was adjusted by the volume of 10× PBS with phenol red and 0.5 N NaOH. 2.2 SIPNs CH polymerization The prepared type 1 collagen solution and HA solution were mixed at ratio of 1:2 (CH12), 1:1 (CH11), and 2:1 (CH21), and MMP-insensitive HA was also mixed at ratio


of 2:1(CH21-insensitive). The CH mixture was premixed just before the injection into the microfluidic cell culture platform (Fig. 1, 1 and 2). To measure swelling properties, the SIPNs CH was placed in phosphate-buffered saline (PBS) at room temperature overnight. The swelling ratio was determined by comparing the wet weight of the hydrogel before and after placing in PBS. 2.3 Hydrogel incorporating microfluidic assay The detailed protocol of microfluidic assay was reported in previous publications (Chung et al. 2009a, b; Sudo et al.

719

2009). Briefly, the microfluidic chip was fabricated by PDMS (poly-dimethyl siloxane, Silgard 184, Dow Chemical, MI, USA) soft lithography process. After autoclaving, PDMS chip and glass coverslide were bonded together by plasma treatment (Femto Science, Korea) to form a closed microfluidic channels. The closed channels were filled with 1 mg/ml of poly-D-lysine solution, and then incubated at 37°C for 6 h. After coating, the channel was aspirated, washed, and dried at 80°C for 24 h to render the channel surface hydrophobic. The premixed CH mixture was injected into the scaffold channel through the gel inlet port, then the chip embedded with the hydrogel was incubated at 37°C for 30 min to allow gelation into SIPNs CH hydrogel (Fig. 1, 3 and 4). 2.4 Endothelial cell seeding and migration into the SIPNs CH scaffold

Fig. 1 Procedure for preparing SIPNs CH hydrogels, injecting into microfluidic chip and application of growth factor: 1 Preparation of 2% HA (MW: 230 kDa) and 0.2% type I collagen solutions. 2 Mixing of HA and collagen solutions. 3 Injection of mixed solution into the microfluidic channel. 4 Incubation of the mixture in an incubator at 37°C for 30 min. At this step, the injected mixture was cross-linked forming SIPNs CH hydrogel. 5 Filling with cell culture medium. 6 Seeding of HUVECs (2×106 cells/ml) in the cell channel. Cells are attached to the SIPNs CH scaffold by gravity while holding the device in a vertical position in an incubator for 1 h. 7 Apply the VEGF gradient. The left channel on the figure was filled with medium supplemented with 20 or 50 ng/ml VEGF

Human umbilical vein endothelial cells (HUVECs) were commercially obtained (LonZa, Basel, Switzerland) and cultured in endothelial cell medium (ECM; Sciencell Research Laboratories Inc., CA, USA) supplemented by 500 ml of basal medium, 5% fetal bovine serum (FBS), 20 ng/ml of VEGF, and 1% penicillin/streptomycin solution. The HUVECs were expanded no more than eight passages. After the SIPNs CH hydrogel had gelled in the microfluidic cell culture platform, one of the microfluidic channel (left channel) was filled with 40–50 μl of HUVEC culture media while the right channel was filled with 40– 50 μl of HUVECs suspension (2×106 cells/ml) (Fig. 1, 5 and 6). Then the device was maintained in a 37°C incubator in a vertical position for 1 h to facilitate attachment of cells to the SIPNs CH scaffold. After cell attachment, medium in the left channel was replaced with medium supplemented by 20 ng/ml (control case) or 50 ng/ml (condition case) VEGF and medium in the right channel containing HUVECs was replaced with medium supplemented only by 20 ng/ml VEGF. The device was kept in an incubator containing 5% CO2 at 37°C and cell culture medium was replaced every day. Cell migration was monitored by phasecontrast microscopy (Zeiss, Oberkochen, Germany), and images for analysis were collected using MetaMorph (Molecular Devices, Inc., CA, USA). The projected area of migrated cells in the scaffold was measured manually with ImageJ (http://rsbweb.nih.gov/ij/) using methods similar to those previously described (Chung et al. 2009a).

3 Results and discussions 3.1 Mechanical properties of the SIPNs CH hydrogel Scanning electron microscopy (SEM) images showed that SIPNs CH scaffolds have porous structure at all mixing

720


ratios (Fig. 2(a–c)) and pore sizes increased with increasing amount of collagen component. We measured the vascular endothelial growth factor (VEGF) diffusivity in SIPNs CH by 40 kDa dextran-FITC molecule having similar molecular weight with VEGF (Lamalice et al. 2007; Cleaver and Krieg 1998), because VEGF is an important physiochem-

ical contributor to several biological cues, influencing cell proliferation, migration, and angiogenesis (Hoffman 2002; Kim et al. 1992). Insets show the normalized intensity of dextran gradients in 2 h after applying dextran (Fig. 2(a–c)). The diffusion coefficient was measured from the gradient obtained in 1~8 h after filling the channel with dextran-

Fig. 2 Characterization of SIPNs CH hydrogel. SEM images of (a) CH12 (300× and 1,500×), (b) CH11 (300×), and (c) CH21, with magnification at 300× (above) and 1,500× (below). Scale bars indicate 100 (300×) and 500 microns (1,500×). (d) Schematic depiction of dextran diffusion through the hydrogel. The gradients was formed 2 h after injecting dextran-FITC (MW: 40 kDa) and the insets of each

SEM image in (a–c) illustrates the dextran gradient. (e) Diffusion coefficients (bar graph) and swelling ratio (black line) of each SIPNs CH hydrogel measured in graphs (f–h) of normalized intensity of dextran-FITC. Measured diffusion coefficients were (f) 1.12×10−13 (m2/s) for SIPNs CH12, (g) 1.78×10−12 for SIPNs CH11, and (h) 2.78×10−12 for SIPNs CH21


FITC (Fig. 2(f–h)); 1.12×10−13 (m2/s) for SIPNs CH12, 1.78×10−12 for SIPNs CH11 and 2.78×10−12 for SIPNs CH21 (Fig. 2(e), bar graph). Apparently CH12 has lower diffusivity than CH11 and CH21, in good agreement with SEM images, showing that porosity and diffusivity of SIPNs CH can be regulated by varying the mixing ratio of collagen and HA. The swelling ratio, expressed as a percentage, was calculated using the equation (Kim et al. iÞ 2008), swelling ratio ð%Þ ¼ 100 ðWwWW , where Ww is the i wet weight of the hydrogel and Wi is the initial weight of hydrogel. The ratios obtained for SIPNs CH12, CH11 and CH21 hydrogels were 215%, 176% and 135%, respectively, lower than that measured in pure HA (245%) (Fig. 2(e), black line). 3.2 HUVEC migration into SIPNs CH scaffolds in response to a VEGF gradient Attachment and migration of HUVECs on and into a scaffold is of importance because vascularization of tissue graft and scaffold ensures supply of nutrients and oxygen, and poor vascularization of engineered tissues is the most critical reason for the unreliable survival of engineered tissue grafts and malfunction of scaffold (Black et al. 1998; Muschler et al. 2004). HA has been considered a good candidate for an implantable hydrogel in tissue engineering and regenerative medicine, due to its biocompatibility and biodegradability. However high molecular weight HA (HMW-HA; in this manuscript, we used HA of 230 kDa MW) was known to have potent anti-adhesive effects on ECs (Pardue et al. 2008), which we also noticed in our experiments with the injected HA into the scaffold channel and seeded HUVECs onto it even modified with RGD and MMP sensitive peptide. Initially, the seeded HUVECs were attached on the HA, following two hours after seeding (day 0), cells detached from the HA, while others attached on the channel surface even near the HA region (Supplementary Figs. 1 and 2a). Over one day of cell seeding (day 1), most cells, even those on the channel surface near the HA, had detached from their initial position, possibly due to both diffusion of anti-adhesion factor from the HA and its swelling. Supplementary figure 1 shows that the diffusion of anti-adhesion factor diffused into the media dominates the cell detachment. We also conclude that RGD and MMP sensitive peptide modification of the HA to enhance cell adhesion and migration could not help the HA adhesive to HUVECs in microfluidic channels. On SIPNs CH12 injected in a microfluidic assay, HUVEC monolayer was also disrupted and detached from their initial position in the presence of 20 ng/ml VEGF gradient, showing similar swelling property. However, in the presence of 50 ng/ml VEGF gradient, HUVECs actively proliferated on the SIPNs CH12 and even migrated into it.

721

It is summarized in the graph in Fig. 3(a) and supplementary Figure 2b, that the anti-adhesion effect of SIPNs CH12 can be overcome. HUVECs on SIPNs CH11 showed similar tendency; under 20 ng/ml VEGF gradient, HUVECs showed limited survival and migration into the SIPNs CH11, while under 50 ng/ml VEGF gradient, they showed good survival and linear increase in migratory area (Fig. 3 (b) and supplementary Figure 1c, white arrow). Synthesized MMP-sensitive and -insensitive SIPNs CH21 showed no anti-adhesive effects on HUVECs, in spite of low VEGF supply, and the migratory area increasing under all conditions (Fig. 3(c) and supplementary Figure 1d). Cells on the MMP-insensitive SIPN CH21 showed good survival but less migration into hydrogel than those on MMP-sensitive SIPN CH21 (Fig. 3(d)), indirectly indicating contribution of collagen fibrous structure mixed with HA on HUVEC adhesion and migration is very similar to that of HA synthesis by MMP sensitive peptides and RGD on HUVEC adhesion and migration. The mean migratory area per day into MMP-sensitive SIPN CH (50 ng/ml VEGF supply) was measured 0.01 mm2 on CH12, 0.02 mm2 on CH11, and 0.03 mm2 on CH21. On the MMP-insensitive SIPN CH21 scaffold, the mean migratory area was similar to that in the SIPNs CH11 (Fig. 4). There is a tendency that cell migration in MMP-insensitive hydrogels showed active migration compared to MMPsensitive hydrogels with 20 ng/ml VEGF. However, we could not see the significant difference between two samples. Driving force for cell migration in this system is not enough in 20 ng/ml VEGF concentrations. Cells with the gradient of 50 ng/ml VEGF could recognize the difference in MMP-sensitive and MMP-insensitive hydrogel systems. In the developed SIPN CH hydrogel, we could induce the initial sprouting (Supplementary Figure 3) and migration (Supplementary Figure 2) of HUVEC, but could not find any perfect tube-like structures. 3D capillary morphogenesis into ECM is very complex process, regulated by a lot of biochemical and physical stimuli, and also by complex interaction of ECM by supporting, degrading and regeneration. We searched for a required factor to perfectly vascularize the SIPNs CH hydrogel, to make it more suitable for tissue engineering applications. Under limited supply of VEGF (20 ng/ml) filled equally in all of the channels, some HUVECs migrated into the SIPNs CH11 and CH21. We can hypothesize a gentle slope of VEGF gradient due to VEGF consumption by HUVECs in the cell channel, resulting in limited but apparent migration of HUVECs. It should be explained with consumption of VEGF by HUVECs in microfluidic channel, raising needs for a deeper investigation on the complex molecular diffusion/convection process and a new light on in vitro stimulatory mechanisms. We also need to

722

Fig. 3 Phase contrast images of HUVEC migration into SIPNs CH hydrogel (day 2): (a) CH12, (b) CH11, (c) CH21, and (d) CH21insensitive. HUVEC migration is dependent on the mixing ratio of collagen and HA, and also MMP and RGD synthesis. The negative


value of migration area in CH12 20 ng/ml is due to the regression of cultured cell outlines by swelling of SIPNs CH12 hydrogel. White arrowhead indicated tip cells and left triangle in each cell figure shows gradient direction of VEGF from the bottom side

develop a solution to acquire enhanced bonding strength between channel surface and SIPN CH hydrogel, to avoid hydrogel contraction and detachment in active vascularization of HUVECs. We found that the HUVECs in the 3D microfluidic cell culture platform survived more than several weeks. However, the SIPN CH hydrogel interacting with HUVEC can only be maintained for 3~5 days because of high contraction force applied by HUVEC monolayer. Poly-D-lysine(PDL) surface coating used to enhance bonding strength between type 1 collagen and channel surface (Chung et al. 2009b) did not show any enhanced effect between channel surface and SIPNs CH hydrogel. Fig. 4 Mean migratory areas per day for CH21 (~0.01 mm2), CH11 (~0.02 mm2), CH21 (~0.04 mm2), and CH21-insensitive (~0.02 mm2) measured under the 50 ng/ml VEGF supply. Total number of experiments was 14 for CH12, 16 for CH11, 10 for CH21, and 8 for CH21-insensitive. Error bars indicate ± standard error (*P