This protocol demonstrates the use of a microfluidic channel with changing geometry along the fluid flow direction to generate extensional strain (stretching) to align fibers in a 3D collagen hydrogel (<250 µm in thickness). The resulting alignment extends across several millimeters and is influenced by the extensional strain rate.
Aligned collagen I (COL1) fibers guide tumor cell motility, influence endothelial cell morphology, control stem cell differentiation, and are a hallmark of cardiac and musculoskeletal tissues. To study cell response to aligned microenvironments in vitro, several protocols have been developed to generate COL1 matrices with defined fiber alignment, including magnetic, mechanical, cell-based, and microfluidic methods. Of these, microfluidic approaches offer advanced capabilities such as accurate control over fluid flows and the cellular microenvironment. However, the microfluidic approaches to generate aligned COL1 matrices for advanced in vitro culture platforms have been limited to thin “mats” (<40 µm in thickness) of COL1 fibers that extend over distances less than 500 µm and are not conducive to 3D cell culture applications. Here, we present a protocol to fabricate 3D COL1 matrices (130-250 µm in thickness) with millimeter-scale regions of defined fiber alignment in a microfluidic device. This platform provides advanced cell culture capabilities to model structured tissue microenvironments by providing direct access to the micro-engineered matrix for cell culture.
Cells reside in a complex 3D fibrous network called the extracellular matrix (ECM), the bulk of which is composed of the structural protein collagen type I (COL1)1,2. The biophysical properties of the ECM provide guidance cues to cells, and in response, cells remodel the ECM microarchitecture3,4,5. These reciprocal cell-matrix interactions can give rise to aligned COL1 fiber domains6 that promote angiogenesis and cell invasion in the tumor environment7,8,9 and influence cell morphology10,11,12, polarization13, and differentiation14. Aligned collagen fibers also promote wound healing15, play a key role in tissue development16, and contribute to long-range cell communication17,18. Therefore, replicating the native COL1 fiber microarchitecture in vitro is an important step toward developing structured models to study cell responses to aligned microenvironments.
Microfluidic cell culture systems have been established as a preferred technology to develop microphysiological systems (MPS)19,20,21,22,23. Leveraging favorable microscale scaling effects, these systems provide precise control over fluid flows, support the controlled introduction of mechanical forces, and define the biochemical microenvironment within a microchannel21,24,25,26,27. MPS platforms have been used to model tissue-specific microenvironments and study multi-organ interactions28. Simultaneously, hydrogels have been widely explored to recapitulate the 3D mechanics and biological influence of the ECM that are observed in vivo29,30. With a growing emphasis on integrating 3D culture with microfluidic platforms, numerous approaches can combine COL1 hydrogels in microfluidic devices31,32,33. However, the methods to align COL1 hydrogels in microfluidic channels have been limited to thin 2D "mats" (<40 µm in thickness) in channels <1 mm wide, offering limited potential to model cell responses in aligned 3D microenvironments31,34,35,36.
To achieve aligned 3D COL1 hydrogels in a microfluidic system, it has been shown that, when a self-assembling COL1 solution is exposed to local extensional flows (velocity change along the streamwise direction), the resulting COL1 hydrogels display a degree of fiber alignment that is directly proportional to the magnitude of the extensional strain rate they experience37,38. The microchannel design in this protocol is unique in two ways; first, the segmented design introduces local extensional strain to the COL1 solution, and second, its "two-piece" construction allows the user to align COL1 fibers and then disassemble the channel to directly access the aligned fibers in an open format. This approach can further be adopted to develop modular microfluidic platforms that develop microphysiological systems with ordered COL1 matrices. The following protocol describes the process of fabricating segmented microchannels and details the use of the channels to align bovine atelo COL1. This protocol also provides instructions for culturing cells on COL1 in an open well format and discusses adding functionality to the platform using a modular, magnetic base layer.
1. Fabrication of the two-piece channel and modular platform base
NOTE: The microfluidic channel is constructed using two parts — the microfluidic channel "cutout", which is razor cut from a poly dimethyl siloxane (PDMS) sheet of defined thickness, and the channel cover, which reversibly bonds to the cutout and forms the channel. The channel is surrounded by a poly(methyl methacrylate) (PMMA) frame that will acts as a media reservoir (Figure 1). The PMMA frame can also be used to magnetically latch specialized modules for added functionality.
2. Injecting the COL1 solution into the microchannel and removing the cover for cell culture applications
When a self-assembling COL1 solution flows through a channel with decreasing cross-sectional area, the streamwise velocity (vx) of the COL1 solution increases locally by a magnitude, ∂vx, along the length of the constriction between the two segments (∂x), resulting in an extensional strain rate (ε̇) where ε̇ = ∂vx/∂x. The extensional strain rate can be calculated from the fluid velocity, which is measured using particle image velocimetry (PIV), as seen in Figure 2.
Previously, it has been shown that local extensional strain promotes long-range COL1 fiber alignment37,38,40 (Figure 3). The alignment of COl1 is directly influenced by the magnitude of the extensional strain rate in the constriction and can be observed to extend uniformly throughout the 5mm length of the segment. To measure the fiber alignment, confocal reflectance microscopy (CRM) images of fibers were used to create a histogram of fiber alignment. The coefficient of alignment (CoA) is the fraction of fibers aligned within ±15˚ of the mode value in the histogram38,42,43,44. CoA ranges from 0-1 with values >0.5 were considered aligned.
Injecting a 3D hydrogel into traditional, sealed microchannels limits where media and cells can be introduced to the matrix. The two-piece channel in this protocol overcomes the limitation of accessing a hydrogel in a channel by allowing the user to lift off the channel cover and directly access the entire COL1 matrix. Channel lift-off is enabled by functionalizing the glass coverslip with glutaraldehyde to promote COL1 attachment using amine-aldehyde-amine interactions and functionalizing the PDMS cover with BSA to prevent COL1 adsorption on the cover. The COL1 fiber alignment remains unaffected after lifting off the cover, as seen visually in Figure 4.
Aligned COL1 substrates are known to influence cell alignment by directing protrusions9,12,45,46 and, in turn, the organization of stress filaments. Endothelial cells in the unaligned and aligned COL1 segments were imaged to investigate the cellular response to the COL1 matrices developed using this protocol. The cells were fixed and stained 4 h after seeding. Representative images of the HUVECs cultured on the aligned segment of the matrix showed increased actin fiber alignment compared to the HUVECs on the segment with random fibers (Figure 5).
Figure 1: Schematic showing the assembly of the layers. This figure was adapted with permission from Ahmed et al.38. Please click here to view a larger version of this figure.
Figure 2: Stepped channel and extensional strain rates. (A) Schematic of the stepped channel with dimensions. Outset shows the calculation of the extensional strain rate. The extensional strain rate in each constriction is calculated from the velocity measured using particle image velocimetry (PIV). (B) The extensional strain rates in each constriction. This figure was adapted with permission from Ahmed et al.38. Please click here to view a larger version of this figure.
Figure 3: Confocal reflectance images of COL1 fibers in each microchannel segment after COL1 polymerization. The increase in the degree of fiber alignment from segment (a) to segment (e) can be visually observed. Scale bar = 25 µm. This figure was adapted with permission from Ahmed et al.38. Please click here to view a larger version of this figure.
Figure 4: Confocal reflectance images of COL1 fibers showing that the fiber alignment remains unaffected by removing the channel cover. Scale bar = 25 µm. This figure was adapted with permission from Ahmed et al.38. Please click here to view a larger version of this figure.
Figure 5: Representative images of cells and COL1 fibers. Representative images of cells and COL1 fibers in segment (e) of the COL1 matrix and segment (a) of the COL1 matrix. Please click here to view a larger version of this figure.
Supplementary Figure 1: Figure showing laser-cut components with dimensions. (A) Channel, (B) magnetic base, and (C) channel cover mold. All dimensions in millimeters. Please click here to download this File.
Protocols to generate COL1 matrices with aligned fibers have been described using magnetic methods, the direct application of mechanical strain, and microfluidic techniques47. Microfluidic approaches are commonly used to create microphysiological systems because of their well-defined flow and transport characteristics, which enable precise control over the biochemical microenvironment. Since aligned COL1 fibers provide key instructive cues during pathophysiological processes such as wound healing, tumor cell invasion, and tissue development, generating aligned COL1 matrices in microfluidic systems is a step toward developing biologically relevant microphysiological systems.
This protocol provides the ability to fabricate millimeter-scale, aligned COL1 matrices with regions of defined fiber alignment. To induce alignment, a microfluidic channel with stepwise reductions in width is used. As a self-assembling COL1 solution flows through the channel at a constant flow rate, the solution flow velocity increases at the constrictions, resulting in local extensional strain and long-range alignment of the self-assembling COL1 fibers. Compared to shear flow-induced fiber alignment, we observed fiber alignment in channels up to 250 µm in thickness, allowing us to create 3D hydrogels in a microfluidic channel. Since COL1 fiber alignment relies upon the extensional strain rate, users may modify the provided channel or develop custom channel designs to generate extensional flow and create COL1 matrices of the desired size and shape. Additional flexibility is provided through the soft lithography-free fabrication of the channels, allowing for the rapid prototyping of channel designs.
The two-piece channel approach presented in this protocol overcomes the limitations of accessing a 3D hydrogel in microchannels. A two-piece channel provides direct access to the COL1 hydrogel, wherein the channel cover can be lifted off to reveal the COL1 matrix in the channel. Directly accessing the COL1 after polymerization eliminates the need for cells to be mixed into the COL1 precursor solution, allowing a user to manipulate the COL1 precursor solution for extended periods at non-physiological pH and low temperatures. Further, a modular approach can be used to position the cells and provide layer-by-layer biofabrication capabilities, as shown in our earlier work. Modules can be used to introduce membranes for co-culture, introduce continuous perfusion, and also create multi-layered ECM constructs22,38,39. Custom modules can also be fabricated depending on experimental needs. Coupled with magnetic latching, the modular approach is also well suited to experiments that require adding functionality with time.
There are some critical steps that need to be considered in this protocol. All components, including the channel cutout, cover, coverslips, and modular base, should be cleaned appropriately to ensure proper bonding and chemical functionalization. The collagen should be prepared on ice, and all the components must be kept cold to ensure consistency. Once prepared, the collagen solution should be used within 10 min of preparation. The final pH of the collagen solution must be measured using a pH probe to ensure consistency. A clean, moist wipe must be placed in the Petri dish with the collagen to ensure that the collagen does not dry out in the warm incubator.
Some modifications and troubleshooting of the protocol include the following. (i) If media leaks during cell culture, one should ensure that the PMMA frame and coverslips are attached without any gaps in the adhesive layer. (ii) If the magnets are not set flush in the base layer, they will prevent the coverslip from adhering to the base, resulting in gaps. (iii) The glutaraldehyde coating on the coverslip can also be of a lower concentration if needed (down to 0.1%). The optimal concentration of glutaraldehyde to prevent toxicity should be experimentally determined for each cell line to be used in the experiment since glutaraldehyde may present toxicity to cells. (iv) One may include additional materials such as hyaluronic acid, fibronectin, or fluorescent beads along with collagen to customize the matrix further. (v) If needed, the channel design can be modified to have longer or shorter segments to avoid boundary and edge effects48.
This protocol has the following limitations. Atelo bovine collagen is a soft material (G' <100 Pa) at 2.5 mg·mL−1. To increase the gel stiffness, researchers may use higher concentrations of collagen or introduce crosslinking agents. Although this protocol has only been validated for aligning collagen matrices up to 250 µm in thickness, thicker matrices can be developed by increasing the channel thickness and optimizing the flow rates to introduce adequate extensional strain rates.
It is anticipated that this protocol will be of use to researchers seeking to develop tissue-specific ordered microenvironments. It may also be used as a guide for creating modular fluidic platforms to establish microphysiological systems.
The authors have nothing to disclose.
This work was supported in part by the National Institute of Health under award number R21GM143658 and by the National Science Foundation under grant number 2150798. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies.
(3-Aminopropyl)triethoxysilane, 99% (APTES) | Sigma Aldrich | 440140-100ML | |
20 Gauge IT Series Angled Dispensing Tip | Jensen Global | JG-20-1.0-90 | |
3/16" dia. x 1/16" thick Nickel Plated Magnet | KJ Magnetics | D31 | |
3M (TC) 12X12-6-467MP | DigiKey | 3M9726-ND | |
ACETONE ACS REAGENT ≥99.5% | Signa Aldrich | 179124-4L | |
BD-20AC LABORATORY CORONA TREATER | Electro-Technic Products | 12051A | |
Bovine Serum Albumin (BSA), Fraction V, 98%, Reagent Grade, Alfa Aesar | VWR | AAJ64100-09 | |
Clear cast acrylic sheet | McMaster-Carr | 8560K181 | |
Corning 100 mL Trypsin 10x, 2.5% Trypsin in HBSS [-] calcium, magnesium, phenol red, Porcine Parvovirus Tested | VWR | 45000-666 | |
Countess II Automated Cell Counter | Thermo Fisher Scientific | AMQAX1000 | |
CT-FIRE software | LOCI – University of Wisconsin | ||
EGM-2 Endothelial Cell Growth Medium-2 BulletKit, (CC-3156 & CC-4176), Lonza CC-3162, 500 mL | Lonza | CC-3162 | |
Glutaraldehyde 50% in aqueous solution, Reagent Grade, Packaging=HDPE Bottle, Size=100 mL | VWR | VWRV0875-100ML | |
Graphtec CELITE-50 | Graphtec | CE LITE-50 | |
HEPES (1 M) | Thermo Fisher Scientific | 15-630-080 | |
High-Purity Silicone Rubber .010" Thick, 6" X 8" Sheet, 55A Durometer | McMaster-Carr | 87315K62 | |
Human Umbilical Vein Endothelial cells | Thermo Fisher Scientific | C0035C | |
Invitrogen Trypan Blue Stain (0.4%) | Thermo Fisher Scientific | T10282 | |
Isopropanol | Fisher Scientific | A4154 | |
Laser cutter | Full Spectrum | 20×12 H-series | |
Microfluidics Syringe pump | New Era Syringe Pumps | NE-1002X | |
Microman E Single Channel Pipettor, Gilson, Model M1000E | Gilson | FD10006 | |
Molecular Probes Alexa Fluor 488 Phalloidin | Thermo Fisher Scientific | A12379 | |
Molecular Probes Hoechst 33342, Trihydrochloride, Trihydrate | Thermo Fisher Scientific | H3570 | |
Nutragen Bovine Atelo Collagen | Advanced BioMatrix | 5010-50ML | |
Pbs (10x), pH 7.4 | VWR | 70011044.00 | |
PBS pH 7.4 | Thermo Fisher Scientific | 10010049.00 | |
Phosphate-buffered saline (PBS, 10x), with Triton X-100 | Alfa Aesar | J63521 | |
Replacement carrier sheet for graphtec craft ROBO CC330L-20 | USCUTTER | GRPCARSHTN | |
Restek Norm-Ject Plastic Syringe 1 mL Luer Slip | Restek | 22766.00 | |
Silicon wafer | University wafer | 452 | |
Sodium Hydroxide, ACS, Packaging=Poly Bottle, Size=500 g | VWR | BDH9292-500G | |
Sylgard 184 | VWR | 102092-312 | |
Thermo Scientific Pierce 20x PBS Tween 20 | Thermo Fisher Scientific | 28352.00 |