Pneumatically Actuated Soft Micromold Device for Fabricating Collagen and Matrigel Microparticles

Abstract

Collagen microparticles have recently gained more attention as viable cell confinement blocks in many biomedical research fields. Small volume and high surface area of collagen structure improve cell confinement, viability, and proliferation. Moreover, dense collagen fiber structure can protect cells from immune destruction. The ability to produce collagen microparticles in an accurate and reliable way is of upmost importance to the advancement of many biomedical researches, especially cancer research and tissue engineering. Currently, no such fabrication technique exists due to inherent fragility of collagen. Herein, we report the very first platform, pneumatically actuated soft micromold (PASMO) device, which addresses challenges in collagen microparticle production. Our new platform uses a soft micromold with a pneumatic actuator that can produce arbitrary shapes of collagen microstructures precisely from 100 μm to over 2 mm in range and can encapsulate cells inside without damaging the shape. The duplication accuracy of more than 96% in dimensions and 90% in depth has been demonstrated. The density of collagen fiber distribution is determined to be 86.57%, which is higher than that of collagen microparticles produced by other methods. We have confirmed cell viability in collagen microparticles. We also produce Matrigel™ particles as tool to develop a xenograft cancer model. The results demonstrate that Matrigel particles created by the PASMO device can reduce cell scattering for the xenograft model and the uniformity of tumors developed in mice is 12-fold improved, which can lead to an increased accuracy of cancer metastasis studies and drug screening research. These breakthroughs in the production of modular microparticles will push the boundaries of cancer research in the near future.

Introduction

Collagen is a hydrogel and the most abundant fibrous biomaterial in extracellular matrices and works effectively as a scaffold for the formation of tissue structures. It also contains specific protein binding sites that can maintain normal cell functions, such as differentiation and proliferation.¹ Due to these characteristics, collagen has potential applications in numerous biomedical fields that include regenerative tissue engineering,^2–4 disease study,⁵ cancer research,⁶ drug delivery,⁷ wound dressing,^8,9 biomedical sensor,¹⁰ and soft actuator.¹¹ In particular, miniaturization of collagen has become more important recently as cell confinement blocks for tissue regeneration applications.^12–19 For instance, numerous cells, including HEK293,^12,13 human mesenchymal stem cells,^14–17 oligodendrocyte, B104, PC,¹⁸ and HepG2,¹⁹ have been encapsulated in the centimeter-scale collagen structures.

Currently, several methods to produce hydrogel microparticles have been reported. For instance, a monodispersed microfluidic approach is used to create collagen microspheres.^20–26 While high-throughput production is the main advantage, this approach has not been successful in producing fully gelled collagen hydrogels in the oil phase. The collagen gelling process requires optimal pH environment that can be achieved by exposing collagen to air and is not effective in the microfluidic method. Consequently, collagen microparticles made in oil through the microfluidic approach are fragile, with fiber densities much less than those fully exposed to air during the gelling process. The second harmonic microscope image of collagen microparticles made by the microfluidic approach is shown in Supplementary Figure S1a (Supplementary Data are available online at www.liebertpub.com/soro). The fiber densities of collagen microparticles (62.45%; highest value obtained by ImageJ) produced by microfluidic approaches^20–26 are much lower than those found in real tissue as shown in Supplementary Figure S1b. In addition, the extraction process of collagen particles from the oil phase without any contamination and damage is extremely challenging and significantly reduces the production throughput. These collagen microparticles gelled in the oil phase are also not durable enough for syringe injection processes. Microscopic images of collagen microparticles made by microfluidic approaches²⁷ before and after the injection into agar gel using a syringe are shown in Supplementary Figure S1c. It is seen that collagen microparticles are collapsed and merged because they are not fully gelled. A new technique to chemically crosslink collagen microparticles to enhance their mechanical stability has been investigated²⁷; however, cells cannot be encapsulated inside of microparticles because the crosslinker molecules for collagen can also crosslink collagen in cells, leading to reduced cell viability. Imprint lithography^28–32 is also investigated to pattern microscale hydrogel particles with molding and ultraviolet (UV) exposure processes. Polyethylene (glycol) diacrylate (PEGDA) is dispensed to the patterned surface, and UV exposure initiates the gelling process of PEGDA to form solid microparticles. Although this approach creates controlled shape and dimensions of particles, the application of this approach is limited to UV crosslink material, which excludes collagen. Continuous-flow lithography³³ and stop-flow lithography³⁴ have been recently reported to pattern various shaped PEGDA microparticles. A PEGDA solution is pumped into microfluidic channels with UV light exposure. Exposed volumes in PEGDA solution through masks are crosslinked to form microparticles. However, these approaches have some material limitations and are not appropriate for long-term cell encapsulation. Molding alginate to form microparticles has been reported.³⁵ Specifically, replica molding,³⁶ Particle Replication in Non-wetting Templates (PRINT™),^37,38 is used to produce stable and chemically crosslinked materials, such as alginate. Crosslinked alginate microparticles are extracted from the templates by pushing molds manually. However, alginate is not a good material for the proposed implanting applications due to immune responses. A bottomless molding technique for making millimeter-scale collagen particles³⁶ has been reported. Collagen solution is dispensed into the bottomless mold, and solid collagen is pushed out directly from open bottom areas. This approach is the only technique to produce collagen particles; however, it has very low throughput. Many damaged and deformed collagen microparticles are produced from this platform. The extraction of particles without damage is still challenging due to the inherent softness of collagen. Squeezing the micromold is the most traditional method of extraction for molding approaches; however, the inhomogeneous extraction force easily damages soft collagen microparticles.

For cancer research, the xenograft model is an essential tool to provide real-time monitoring of the progression of tumor. Matrigel™ is derived from reconstituted basement membrane and forms the Engelbreth-Holm-Swarm tumor. Matrigel is constructed mainly from collagen containing proteins and growth factors, such as laminin, and heparin.^39–41

The classic tumor xenograft method in animals is the injection of liquid Matrigel and cancer cells into a vein. This approach normally scatters cancer cells in entire bodies due to the liquid Matrigel, as a result, the tumor xenograft sometimes would not be formed or have a large volume fluctuation. To overcome these issues, injection of gelled Matrigel microparticles confining cancer cells before the injection process is expected to reduce the cell scattering issue that improves the xenograft successful rate. However, there is currently no available tool to generate Matrigel microparticles because of its fragile mechanical property.^41,42 In conclusion, currently, there are neither tools nor methods available to generate collagen and Matrigel microparticles reliably.

In this article, we report a new collagen and Matrigel microstructure production platform—pneumatically actuated soft micromold (PASMO) devices—to overcome the limitations encountered in previous studies. PASMO devices demonstrate highly reproducible generation of collagen microstructures with dimensions as small as 100 μm. To the best of our knowledge, this is the first time that arbitrary shapes of collagen microstructures other than spherical shapes are created with high reproducibility. As a result, we expect that our PASMO devices and resultant collagen and Matrigel microparticles will greatly enhance research and development in the fields of tissue engineering, cancer research, hybrid bio- and microsystems, cancer metastasis study, and anticancer drug screening by providing cellular three-dimensional (3D) microenvironment.

Experiment

PASMO device fabrication process

The PASMO was designed with a main intention to release collagen microparticles at high throughput at relatively small actuating input. As collagen particles are soft, it is also necessary to have the platform made of soft materials to reduce chances of damaging particles due to contact with hard/stiff substrates. For this reason, Ecoflex, a silicone rubber, is considered for PASMO. Due to the soft and elastic nature of Ecoflex, it is possible to design pneumatically actuated devices made of this material, which can be deformed with relatively low air pressure, eliminating the need to use high force for actuating the device. Figure 1a presents the working strategy of PASMO. A small passage for airflow was created and pressurizing it induces large deformations and expands the surface area for releasing the particles. Silicone rubber can also be easily formed into various shapes with low-cost fabrication, and it is biocompatible. One disadvantage of silicone rubber PASMO is that it is quite challenging (if not impossible) to have high-precision motion control such as in electric or magnetic actuator. However, in this particular case, high-precision control is not necessary.

FIG. 1.

(a) Real images and schematic diagram of the working strategy for PASMO devices to produce collagen microstructures. There are five steps for this extraction process. Step 1: Dispensing collagen solution into molds. Step 2: Crosslinking of collagen microparticles. Step 3: Pneumatic deformation process of PASMO devices for the diffusion of buffer solution into gaps between molds and hydrogel particles. Step 4: Releasing particles into culture media. Step 5: Releasing particles completely. (b) Schematic diagram of PASMO, which comprises the surface template layer, pneumatic system layer, and barrier layer. PASMO device is distinguished as five sections. The position, which is closest to air inlet, is #1, while the farthest position from air inlet is #5. (c–f) Schematic diagram of fabrication process for PASMO device. Molding process of (c) surface template layer, (d) pneumatic system layer, and (e) barrier layer. (f) The bonding process of these three layers. (g) Photographic image of PASMO device with Tygon tubing. PASMO, pneumatically actuated soft micromold. Color images available online at www.liebertpub.com/soro

Figure 1a summarizes the process of extracting collagen microstructures from PASMO devices. The collagen or Matrigel microparticle extraction process consists of five steps. Step 1: Dispensing collagen or Matrigel solution containing live cells into molds. Step 2: Physical crosslinking of collagen or Matrigel solution to hydrogel particles. Step 3: Pneumatic deformation process of PASMO devices for diffusion of buffer solution into gaps between molds and hydrogel particles. Step 4: Releasing particles into culture media. Step 5: Collagen or Matrigel particles are completely released from PASMO devices into culture media.

The schematic diagram of the PASMO device is shown in Figure 1b. The PASMO platform consists of three layers: (1) microscale surface template layer for molding liquid collagen, (2) pneumatic system layer to deform microtemplates to extract soft collagen microparticles, and (3) buffer layer with cloth. Due to the relatively stiffer cloth on the bottom buffer layer, the platform deforms asymmetrically with the surface layer expanding the most while the bottom remains intact, which promotes extraction of collagen microstructures from the surface template layer.

The fabrication process of PASMO device is shown in Figure 1c–e. The master mold for the surface template layer is made using the standard photolithographic process with SU8 photoresist, as described previously.⁴³ A PDMS solution is spin-coated on the master mold and cured at 65°C for 2 h, then peeled from the master mold (Fig. 1c). The master molds for the pneumatic system and barrier layers are made by EnvisionTEC 3D printer. Uncured Ecoflex-30 is poured over a pneumatic system mold and cured for 4 h at room temperature and peeled from the mold (Fig. 1d). The width and depth of channel inside the pneumatic system layer are 1.5 and 3.0 mm, respectively. A cleaning cloth with high tensile stress is placed inside the barrier layer mold and encapsulated in this layer; then, uncured Ecoflex-30 is dispensed over the cloth in the mold. Ecoflex-30 is cured at room temperature for 4 h and peeled from the mold (Fig. 1e). Surface template, pneumatic system, and barrier layers are subsequently bonded at 65°C for 2 h with uncured Ecoflex-30 used as glue between layers (Fig. 1f). The image of PASMO device with a tube attached for applying pneumatic pressure is shown in Figure 1g and Supplementary Figure S2 presents the image of PASMO device.

Preparation of collagen structures with and without cells

The concentration of the rat tail collagen type I solution is set to 3.5 mg/mL for the entire set of experiments. This collagen solution is dispensed into the surface templates and gelled under neutral condition (pH = 7.4) at 37°C for 30 min. Pneumatic pressure is applied to the pneumatic channel from a syringe pump, which leads to deformation of PASMO devices and the subsequent release and harvest of collagen microstructures in phosphate saline buffer (PBS). For cell encapsulation inside collagen microstructures, MDA-MB-231 cells overexpressing Green Fluorescent Protein (GFP) are mixed with a 3.5 mg/mL collagen solution as previously described⁴⁴ and dispensed into surface templates. Collagen microstructures with embedded cells are released from the deformed surface template layers of PASMO devices into Dulbecco's minimal Essential medium with 10% fetal bovine serum. Cells inside collagen microstructures are cultured in a humidified incubator (5% CO₂ and 95% air condition) at 37°C for more than 14 days. The PASMO device is autoclaved (131°C) and exposed to UV light for sterilization before encapsulating living cells.

Prepare Matrigel particles within pancreatic cancer cells for Xenograft model and confirmation of tumor size

We encapsulated pancreatic tumor cells that expressed Luciferase enzyme, AsPC 1/Luc, in Matrigel particles. One microliter of chilled liquid Matrigel is blended with AsPC 1/Luc cells (2 × 10⁶/mL concentration) and then dispensed into equilateral triangle-shaped molds on the PASMO devices. The dimensions of this mold are 3 mm width, 3 mm length, and 2 mm thickness. The PASMO devices with Matrigel solution are placed into an incubator at 37°C for gelation. After 30 min, Matrigel particles are extracted to PBS. One Matrigel triangle prism-shaped particle is implanted near pancreas for one nude mouse. The tumor-forming process is monitored by IVIS Spectrum computed tomography for whole mouse scanning. The volume of tumor is measured by Vevo 3100 high-frequency microultrasound system for 3D scanning.

Simulation of PASMO device

The finite element (FE) software ABAQUS 6.12 was used to simulate the expansion of pneumatic actuation of the micromold. The deformation determined from the FE analysis is compared with the experimentally measured deformation of the PASMO devices, shown in Supplementary Figure S3. The displacement of the bottom surface of the PASMO model is a constraint to avoid rigid body motion. Pneumatic pressures are set as the pressure load applied in the PASMO device. The pneumatic pressure is prescribed in the normal direction on pneumatic channel. Since the model is symmetric, only half of the model is generated using 203,331 tetrahedral elements and each element consists of 10-node modified hybrid tetrahedral elements (ABAQUS C3D10MH). The modified tetrahedral elements provide good convergence rate and prevent volumetric locking. Hyperelastic material models are typically used in explaining the nonlinear elastic response of soft materials, such as silicone rubber. Supplementary Figure S3 shows the experimental relationship between stress and strain for Ecoflex under uniaxial tension. With the experiment data fitted and analyzed, the Ecoflex is modeled as an incompressible and isotropic following the Ogden model. The strain energy density function for an incompressible Ogden model (n = 3) is described as \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} U = \mathop \sum \limits_ { i = 1 } ^N { \frac { 2 { \mu _i } } { \alpha _i^2 } } \left( { { { \overline { { \lambda _1 } } } ^ { { \alpha _i } } } + { { \overline { { \lambda _2 } } } ^ { { \alpha _i } } } + { { \overline { { \lambda _3 } } } ^ { { \alpha _i } } } - 3 } \right) \end{align*} \end{document}

where U is the strain energy per unit of reference volume; \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \alpha _i}$$ \end{document} and \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \mu _i}$$ \end{document} are material constants; \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $$\overline {{ \lambda _i}}$$ \end{document} are the deviatoric principal stretches \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $$\overline { { \lambda _i } } = { J^ { - \frac { 1 } { 3 } } } { \lambda _i } $$ \end{document} ; \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \lambda _i}$$ \end{document} are the principal stretches, and J is the determinant of the deformation gradient. For an incompressible material behavior, J = 1. The following material parameters are used for the Ecoflex: n = 3, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \mu _1}$$ \end{document} = 0.0111 MPa, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \mu _2}$$ \end{document} = 1.119e-005 MPa, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \mu _3}$$ \end{document} = 0.0114 MPa, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \alpha _1}$$ \end{document} = 2.393, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \alpha _2} =$$ \end{document} 8.123, and \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} $${ \alpha _3} =$$ \end{document} −2.168.

Results and Discussion

Controlled deformations of PASMO devices are essential for clean extraction of collagen microstructures; therefore, it is important to characterize the deformations of individual micromolds on the surface template layer upon pneumatic actuation. Figure 2a shows the microscopic top view image of microcross molds on the PASMO device with different pressures, 1.1, 1.2, 1.7, and 2.0 atm. The mold starts expanding after injecting air and expansion of the mold is a function of applied pressure.

FIG. 2.

Deformation of micropattern on PASMO with pneumatic actuation. (a) Microscopic images of micromold with pressure application: ^(a)1 atm (no injection of air), ^(b)1.1 atm (0.2 mL injection), ^(c)1.2 atm (0.4 mL air injection), ^(d)1.7 atm (0.8 mL air injection), and ^(e)2.0 atm (1 mL air injection) of pressure are applied. Scale bar is 100 μm for all images. The deformation of soft micromold at (b) x-directional width, (c) y-directional width, and (d) depth of mold at specific locations based on Figure 1b. (e) The expansion for projected area of mold between simulation and experimental data.

The scaled expansion rate is defined as (w₀–w₁)/w₀ (w₀: initial width of the mold, w1: mold width after injecting air) for cross-shaped micromolds in X directional width at positions 1–5 in Figure 1b. The expansion rate of X-directional width of the cross-shaped mold as a function of pressure is shown in Figure 2b. The expansion rates of cross mold for positions 1–5 are 10%, 27%, 103%, 50%, and 68%, respectively, with 2 atm pressure application. The highest expansion rate of micromold is observed at the center position of PASMO device. The expansion rates of cross mold in Y direction for positions 1–5 as a function of pressure application are shown in Figure 2c. They are expanded 8%, 16%, 47%, 24%, and 33% at positions 1–5 in Figure 1b, respectively, with 2 atm pressure applications. The averaged expansion rate of cross mold from positions 1 to 5 in X width is 51.6%, which is higher than one in Y width (25.6%). The differences in expansions in the X- and Y-directions are due to design and position of the air channel; the thickness of Ecoflex from side of air channel is thicker than the one from top of air channel. Therefore, elongation of PASMO from X-direction is larger than in Y-direction. As a result, the deformation in X-direction is more obvious than in Y-direction. In addition to X and Y directional width, the depth of the micromold is investigated as a function of applied pressure that is shown in Figure 2d. The depth of cross molds is reduced linearly from 100 to 56 μm as pressure is increased. This Z directional extrusion from the bottom of micromold increases the successful rate of the particle extraction process.

Figure 2e shows the simulation results on the projected area of cross-shaped micromold on PASMO devices as a function of the pressure from 1 to 2 atm. We use the Ogden model to describe the deformation of hyperelastic material, such as PDMS and Ecoflex. The simulation result demonstrates that Ogden model can describe the deformation of soft micromold as a function of the pressure inside the channel. The simulation result and experimental data show that the expanded area is about twice higher than original shape when the air pressure inside the channel is 2 atm. This expansion generates additional space to allow buffer or culture media to diffuse into the wells for extracting collagen particles.

Figure 3 presents the microscopic images of collagen microstructures of cross column (Fig. 3a), pentagonal prism (Fig. 3c), hollow cylinder (Fig. 3e), triangular prism (Fig. 3g), and square pad (Fig. 3i) produced using PASMO devices. To characterize size reproducibility, dimensions of the structures are compared with those of mold dimensions (Fig. 3b, d, f, h, and j). The results show ∼96% accuracy for each microstructure ranging from 100 to 1000 μm, as shown in Figure 3k. This difference is due to the swelling of collagen by solution. On the other hand, the reproducibility of collagen microstructures smaller than 100 μm is limited because the collagen solution is not completely dispensed into micromolds. The thickness of microstructures is set to 100 μm for all structures and is duplicated with more than 92% accuracy as shown in Supplementary Figure S4.

FIG. 3.

Characterization of collagen microstructures made by PASMO device. Microscopic images of microstructures and molds for (a, b) cross column, (c, d) pentagonal column, (e, f) hollow circular cylinder, (g, h) triangular column, and (i, j) square pad, respectively. The thickness of these microstructures is set to 100 μm. (k) Duplication accuracy as function of dimensions of micromolds. (l) Second harmonic generation microscopic image of collagen fibers in micro cross column. (Red and white scale bars are 500 and 100 μm, respectively. (m) Dimensions of collagen microcubes as function of times after production. Color images available online at www.liebertpub.com/soro

Figure 3l shows the result of second harmonic generation microscopic images of collagen microcross column to characterize their collagen fiber structures. The density of collagen fiber distribution is determined using ImageJ to be 86.57%, which is higher than that of collagen microspheres produced in microfluidic devices (62.45%).²⁶ This matches the density range of collagen fiber in dermis, bone, tendon, and ligament (72–94%).⁴⁵ In addition, the relatively dense collagen microstructures can remain intact when they are injected into an agarose gel without being damaged, unlike their less dense collagen microsphere counterparts, as shown in Figure 4. This demonstrates that collagen microstructures made by PASMO devices are suitable for implantable 3D tissue engineering applications.

FIG. 4.

Collagen microstructures extracted from mold patterns by (a) PASMO device and (b) manual pushing. (c) The graph for microstructure extraction success rate between PASMO device and manual pushing approaches. The scale bar is 500 μm. The percentage is calculated based on the numbers of released particles without damage divided by the total number of molds.

In addition, collagen microstructures produced by PASMO devices maintain their dimensions and shapes, with no observable shrinkage or expansion in size over 50 h (Fig. 3m). Thus, PASMO devices are able to precisely control the dimensions and shapes of collagen microstructures. These collagen microstructures are stored in a 4°C refrigerator for up to 60 days without any shape changes. We have also compared the collagen particle extraction rates between PASMO devices and the manual extraction method. The manual method is simply pushing gelled collagen particles from the micromolds by hand. Microscopic images of collagen microparticles extracted by these two methods are shown in Figure 4a and b. Multiple structural damages are observed among those collagen microparticles extracted manually by hand.

The successful extraction rates of microcross collagen by the PASMO device and manual pushing approach are 95% and 15% (based on 100-particle extraction), respectively, as shown in Figure 4c. The extraction forces in PASMO devices are easily controlled uniformly to collagen particles and can extract fragile collagen structures from molds by applying an incrementally bigger volume of air into the pneumatic channel.

The PASMO device also produces collagen microstructures encapsulating MDA-MB-231 cells that overexpress GFP as long as they are viable. The cellular concentration for all experiments is set to 10⁶ cell/mL. The cellular proliferation and contraction of collagen microstructures are investigated as potential collagen modular microenvironment. Figure 5a shows fluorescent images of MDA-MB-231-GFP cells encapsulated in collagen microcubes for 0, 4, 8, and 12 days. Bright spots on the surface or inside of the collagen structures correspond to MDA-MB-231 cells. The fluorescent intensities in the microcube and normalized microcube volume as a function of days are plotted in Figure 5b. Over time, the fluorescent signal tends to increase in intensity, indicating the proliferation of MDA-MB-231 in collagen microcubes, and remains for more than 14 days. The volumetric shrinkage of collagen microcubes is also significant after the fourth day due to the increase in the number of cells that lead to contraction of collagen fibers. The contractile force in collagen fibers is generated from MDA-MB-231 cells attached on collagen fibers. Cells attached on collagen fibers migrate by contracting cell structures that contract collagen fibers. The increased number of cells increased the total contractile force on collagen fibers, which shrinks the collagen microstructures.⁴⁶

FIG. 5.

MDA-MB-231 cell collagen microparticles (a) Fluorescent microscopic images of microcubes containing MDA-MB-231 cells at days 0, 4, 8, and 12. Scale bar is 500 μm. (b) The graphs for fluorescence intensity and volumetric shrinkage of collagen microcubes as a function of days.

In addition, Supplementary Figure S5 shows the volumetric shrinkage of collagen cubes containing cancer cells as a function of days for three sizes of collagen cubes (1 mm, 500 μm, and 300 μm). There are no significant differences among them because the number of cells per unit volume is set to be constant. Therefore, the contractile force per unit volume is also set as constant and then the volumetric shrinkage of collagen cubes within different sizes maintains the same order.

We have also produced millimeter-scale Matrigel particles containing AsPc1/Luc cells and implanted them in nude mice. The classical orthotopic method, injecting mixture of cancer cells and liquid Matrigel into nude mice, is commonly used for xenograft model. Liquid Matrigel is gelled to confine cancer cells in mice. Figure 6a presents the bioluminescence images of cancer cell distributions immediately after the injections. At day 0, we observed one of three nude mice prepared by the classical orthotopic method has response for bioluminescence, while, the bioluminescent signal can be detected in all nude mice implanted with Matrigel particles made by PASMO devices. This result indicates that Matrigel particles can effectively confine cancer cells at specific regions in mice. After 7 days, bioluminescent signals of cancer tissue prepared by the classical orthotopic method decrease as shown in Figure 6a. On the other hand, bioluminescent signals of cancer tissue prepared by PASMO devices increase as shown in Figure 6a. This growth of cancer cells is due to the communication among cancer cells confined in Matrigel that induces cellular function of cancer cells, especially cell proliferation.^47,48 Classical orthotropic method would cause cell scattering that reduces the level of cell-to-cell communications.

FIG. 6.

Classical orthotopic and Matrigel™ particle methods for xenograft model experiment. (a) The bioluminescence images of cancer tissue prepared by the classical orthotopic method and Matrigel particles. Only one nude mouse prepared by the classical orthotopic method has response at day 0. All three mice prepared by Matrigel particles have response at day zero. After 7 days, the bioluminescent signal of cancer tissue prepared by the classical orthotopic method decreases due to cancer cells spreading in the entire body. On the other hand, three mice prepared by Matrigel particles have strong cancer tissue signals. Classical orthotopic method is not able to hold the cells maintaining the defined volume after injection. Therefore, it might spread closer to skin surface area, which caused the stronger signal. However, it might also move to deeper tissue area or diffuse in circulation, which caused weaker signal. PASMO-Matrigel already predefined the cells within the gel so the signals can be more homogeneous. (b) Ultrasonic scanning image of tumor tissue prepared by the classical orthotopic method and Matrigel particles at day 14. Images of tumor tissue removed from mice prepared by the classical orthotopic method and Matrigel particles are also shown. (c) The average tumor volumes prepared by the classical orthotopic method and Matrigel particles. Tumor tissue prepared by Matrigel particles can effectively reduce the fluctuation of tumor volume. Color images available online at www.liebertpub.com/soro

Figure 6b shows the 3D ultrasonic scanning images developed in nude mice. Volumes of tumors prepared by the classic orthotropic method have large fluctuation ranging between 3.5 and 14.5 mm³. However, volumes of tumor tissues prepared by PASMO devices have uniform size distribution ranging between 12 and 14 mm³. Figure 6c shows tumor volumes prepared by the classic orthotropic method and PASMO devices. The average tumor volumes developed from Matrigel particles is 30% larger than those prepared by the classic orthotropic method. The uniformity of tumor tissue volumes formed from Matrigel particles is 12-fold better than those prepared by the classic orthotropic method. In conclusion, Matrigel particles produced by PASMO devices not only can effectively confine cancer cells at specific locations but also can improve the fluctuation of tumor volumes to improve the successful rate, reliability, and accuracy of animal experiments.

Conclusions

We have demonstrated the fabrication of collagen and Matrigel microstructures by PASMO devices without damaging them. Collagen microstructures maintain their shapes and dimensions for up to 60 days without shrinkage or expansion. The corresponding collagen fiber density has also been found to be comparable with the one in real tissue environment. The high fiber density allows for collagen microstructures to be implanted through syringe and remain intact. MDA-MB-231-GFP cells are encapsulated in these collagen microstructures by PASMO devices as modular microenvironment. The cells proliferate continuously for more than 14 days in this microenvironment. Deformations and contractions of collagen microstructures are simultaneously observed during this period, with cancer cells identified as the primary cause of contraction. Moreover, Matrigel particles produced by PASMO devices can effectively localize the tumor cells at a specific location to develop tumor tissue. Moreover, the tumor volume developed by Matrigel particles is more uniform than those made by the classical orthotopic method. These results indicate that collagen and Matrigel microstructures produced by PASMO devices possess the potential to improve cancer research.

PASMO devices are made of silicone rubber that has elastic responses under large deformations and are durable for multiple cycles of deformations. We used a single PASMO device more than hundred times by pressurizing and depressurizing, we did not notice any permanent deformations or damages in the device. Based on this observation, we can conclude that PASMO devices are reusable for several hundred cycles.

Additionally, soft robotic technology will be more and more in demand in biomedical fields in the future. As proven in this article, soft robotics is superior to handle soft and fragile materials such as cells, tissues, and organs due to soft nature. Handling such fragile biostructures is a critical process in biomedical fields for general surgery, laparoscopic surgery, endoscopy, and implantation applications.

Footnotes

Acknowledgment

This study was supported by a Cancer Prevention Research Institute of Texas Grant (No. DP150052) to Dr. Mien-Chi Hung and Dr. Jun Kameoka.

Author Disclosure Statement

No competing financial interests exist.

References

Heino

. The collagen family members as cell adhesion proteins. BioEssays, 2007; 29:1001–1010.

Parenteau-Bareil

, Gauvin

, Berthod

. Collagen-based biomaterials for tissue engineering applications. Materials, 2010; 3:1863–1887.

Stachowiak

, Irvine

. Inverse opal hydrogel-collagen composite scaffolds as a supportive microenvironment for immune cell migration. J Biomed Mater Res A, 2008; 85:815–828.

Wolf

, Muller

, Borgmann

, Brocker

, Friedl

. Amoeboid shape change and contact guidance: T-lymphocyte crawling through fibrillar collagen is independent of matrix remodeling by MMPs and other proteases. Blood, 2003; 102:3262–3269.

Di Lullo

, Sweeney

, Korkko

, Ala-Kokko

, San Antonio

. Mapping the ligand-binding sites and disease-associated mutations on the most abundant protein in the human, type I collagen. J Biol Chem, 2002; 277:4223–4231.

Yang

, van der Werf

, Fitié

CFC

, Bennink

, Dijkstra

, Feijen

. Mechanical properties of native and cross-linked type I collagen fibrils. Biophys J, 2008; 94:2204–2211.

, Drumright

, Siegwart

, Matyjaszewski

. The development of microgels/nanogels for drug delivery applications. Prog Polym Sci, 2008; 33:448–477.

Doillon

, Silver

. Collagen-based wound dressing: effects of hyaluronic acid and fibronectin on wound healing. Biomaterials, 1986; 7:3–8.

, Xue

, Li

, Jiang

. A thermoresponsive antimicrobial wound dressing hydrogel based on a cationic betaine ester. Adv Funct Mater, 2011; 21:4028–4034.

10.

Heo

, Shibata

, Okitsu

, Kawanishi

, Takeuchi

. Injectable hydrogel microbeads for fluorescence-based in vivo continuous glucose monitoring. PNAS, 2010; 107:17894–17898.

11.

Lee

, Konst

. Novel hydrogel actuator inspired by reversible mussel adhesive protein chemistry. Adv Mater, 2014; 26:3415–3419.

12.

Lee

, Lo

, Cheung

, Wong

, Chan

. GDNF-secreting HEK293 cells in collagen-alginate composite gel and collagen microspheres embedded in alginate gel – New carrier designs for drug delivery applications. Biomaterials, 2009; 30:1214–1221.

13.

Wong

, Wang

, Cheung

, Yao

, Chan

. A 3D collagen microsphere culture system for GDNF-secreting HEK293 cells with enhanced protein productivity. Biomaterials, 2007; 28:5369–5380.

14.

Chan

, Hui

, Yeung

, Li

, Mo

, Chan

GCF

. Self-assembled collagen–human mesenchymal stem cell microspheres for regenerative medicine. Biomaterials, 2007; 28:4652–4666.

15.

, Diao

, Chik

, Chow

, An

, Leung

, et al. Delivering mesenchymal stem cells in collagen microsphere carriers to rabbit degenerative disc: reduced risk of osteophyte formation. Tissue Eng Part A, 2014; 20:1379–1391.

16.

Cheng

H-W

, Luk

KDK

, Cheung

KMC

, Chan

. In vitro generation of an osteochondral interface from mesenchymal stem cell–collagen microspheres. Biomaterials, 2011; 32:1526–1535.

17.

Yeung

, Cheah

, Chan

. Effects of reconstituted collagen matrix on fates of mouse embryonic stem cells before and after induction for chondrogenic differentiation. Tissue Eng Part A, 2009; 15:3071–3085.

18.

Yao

, Phan

, Li

. Collagen microsphere serving as a cell carrier supports oligodendrocyte progenitor cell growth and differentiation for neurite myelination in vitro. Stem Cell Res Ther, 2013; 4:109–116.

19.

Choi

, Young

, Loïc

J-D

, Jürgen

, Gyu

. Fabrication of HepG2 cell laden collagen microspheres using inkjet printing. J Korean Soc Precis Eng, 2014; 31:743–747.

20.

Tana

, Cristini

, Lee

. Monodispersed microfluidic droplet generation by shear focusing microfluidic device. Sen Actuators B, 2006; 114:350–356.

21.

Chen

, Abate

, Lee

, Eugene

, Terentjev

, Weitz

. Microfluidic assembly of magnetic hydrogel particles with uniformly anisotropic structure. Adv Mater, 2009; 21:3201.

22.

Yeh

, Zhao

, Lee

, Lin

. Using a T-junction microfluidic chip for monodisperse calcium alginate microparticles and encapsulation of nanoparticles. Sen Actuators A, 2009; 151:231.

23.

, Cong

, Liu

, Ren

, Wang

, Zhang

, et al. Preparation of monodisperse PEG hydrogel composite microspheres via microfluidic chip with rounded channels. J Micromech Microeng, 2013; 23:095016.

24.

Hwang

, Dendukuri

, Doyle

. Microfluidic-based synthesis of non-spherical magnetic hydrogel microparticles. Lab Chip, 2008; 8:1640.

25.

, Nie

, Seo

, Lewis

, Kumacheva

, Stone

, et al. Generation of monodisperse particles by using microfluidics: Control over size, shape, and composition. Angew Chem Int Ed, 2005; 44:724.

26.

Hong

, Hsu

, Kaunas

, Kameoka

. (2012) Collagen microsphere production on a chip. Lab Chip, 12:3277.

27.

, Black

, Santacana-Laffitte

, Patrick

. Preparation and assessment of glutaraldehyde-crosslinked collagen-chitosan hydrogels for adipose tissue engineering. J Biomed Mater Res A, 2007; 81:59.

28.

Gates

, Xu

, Stewart

, Ryan

, Willson

, Whitesides

. New approaches to nanofabrication: Molding, printing, and other techniques. Chem Rev, 2005; 105:1171.

29.

Glangchai

, Caldorera-Moore

, Shi

, Roy

. Nanoimprint lithography based fabrication of shape-specific, enzymatically-triggered smart nanoparticles. J Control Release, 2008; 125:263.

30.

Takei

, Ogawa

, Deschner

, Jen

, Nihira

, Hanabata

, Willson

. Silicon-containing spin-on underlayer material for step and flash nanoimprint lithography. Jpn J Appl Phys, 2010; 49:075201.

31.

Helgeson

, Chapin

, Doyle

. Hydrogel microparticles from lithographic processes: novel materials for fundamental and applied colloid science. Curr Opin Colloid Interface Sci, 2011; 16:106.

32.

Qiu

, Chen

, Yan

, Wu

. Generation of uniformly sized alginate microparticles encapsulation by using a soft-lithography approach. Adv Mater, 2007; 19:1603.

33.

Dendukuri

, Pregibon

, Collins

, Hatton

, Doyle

. Continuous-flow lithography for high throughput microparticle synthesis. Nat Mater, 2006; 5:365.

34.

Dendukuri

, Gu

, Pregibon

, Hatton

, Doyle

. Stop-flow lithography in a microfluidic device. Lab Chip, 2007; 7:818.

35.

Qiu

, Chen

, Yan

, Wu

. Generation of uniformly sized alginate microparticles for cell encapsulation by using a soft-lithography approach. Adv Mater, 2007; 19:1603–1607.

36.

McGuigan

, Bruzewicz

, Glavan

, Butte

, Whitesides

. Cell encapsulation in sub-mm sized gel modules using replica molding. PLoS One, 2008; 3:e2258.

37.

Rolland

, Maynor

, Euliss

, Exner

, Denison

, DeSimone

. Direct fabrication and harvesting of monodisperse, shape-specific nanobiomaterials. J Am Chem Soc, 2005; 127:10096.

38.

Acharya

, Shin

, McDermott

, Mishra

, Park

, Kwon

, et al. The hydrogel template method for fabrication of homogeneous nano/microparticles. J Control Release, 2010; 141:314.

39.

Asghar

, El Assal

, Shafiee

, Pitteri

, Paulmurugan

, Demirci

. Engineering cancer microenvironments for in vitro 3-D tumor models. Mater Today, 2015; 18:539–553.

40.

Nyga

, Cheema

, Loizidou

. 3D tumour models: novel in vitro approaches to cancer studies, J Cell Commun Signal, 2011; 5:239–248.

41.

Wenger

MPE

, Bozec

, Horton

, Mesquida

. Mechanical properties of collagen fibrils. Biophys J, 2007; 93:1255–1263.

42.

Soofi

, Last

, Liliensiek

, Nealey

, Murphy

. The elastic modulus of Matrigel™ as determined by atomic force microscopy. J Struct Biol, 2009; 167:216–219.

43.

Whitesides

, Ostuni

, Takayama

, Jiang

, Ingber

. Soft lithography in biology and biochemistry. Annu Rev Biomed Eng, 2001; 3:335.

44.

Kang

, Bayless

, Kaunas

. Fluid shear stress modulates endothelial cell invasion into three-dimensional collagen matrices. Am J Physiol Heart Circ Physiol, 2008; 295:H2087.

45.

Eleswarapu

, Responte

, Athanasiou

. Tensile properties, collagen content, and crosslinks in connective tissues of the immature knee Joint. PLoS One, 2011; 6:e26178.

46.

Riching

, Cox

, Salick

, Pehlke

, Riching

, Ponik

, et al. 3D collagen alignment limits protrusions to enhance breast cancer cell persistence. Biophys J, 2014; 107:2546–2558.

47.

Lou

. Intercellular conduits in tumors: The new social network. Trends Cancer, 2016; 2:3–5.

48.

Qiu

, Su

. Development of orthotopic pancreatic tumor mouse models. Methods Mol Biol, 2013; 980:215–223.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.39 MB

0.00 MB