Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Nov 18.
Published in final edited form as: Biomed Microdevices. 2019 Nov 18;21(4):99. doi: 10.1007/s10544-019-0436-3

A High-Throughput Microfluidic Method for Fabricating Aligned Collagen Fibrils to Study Keratocyte Behavior

Kevin H Lam 1, Pouriska B Kivanany 2, Kyle Grose 2, Nihan Yonet-Tanyeri 2, Nesreen Alsmadi 1, Victor D Varner 1,3, W Matthew Petroll 2, David W Schmidtke 1,3
PMCID: PMC7228026  NIHMSID: NIHMS1586474  PMID: 31741114

Abstract

In vivo, keratocytes are surrounded by aligned type I collagen fibrils that are organized into lamellae. A growing body of literature suggests that the unique topography of the corneal stroma is an important regulator of keratocyte behavior. In this study we describe a microfluidic method to deposit aligned fibrils of type I collagen onto glass coverslips. This high-throughput method allowed for the simultaneous coating of up to eight substrates with aligned collagen fibrils. When these substrates were integrated into a PDMS microwell culture system they provided a platform for high-resolution imaging of keratocyte behavior. Through the use of wide-field fluorescence and differential interference contrast microscopy, we observed that the density of collagen fibrils deposited was dependent upon both the perfusion shear rate of collagen and the time of perfusion. In contrast, a similar degree of fibril alignment was observed over a range of shear rates. When primary normal rabbit keratocytes (NRK) were seeded on substrates with a high density of aligned collagen fibrils and cultured in the presence of platelet derived growth factor (PDGF) the keratocytes displayed an elongated cell body that was co-aligned with the underlying collagen fibrils. In contrast, when NRK were cultured on substrates with a low density of aligned collagen fibrils, the cells showed no preferential orientation. These results suggest that this simple and inexpensive method can provide a general platform to study how simultaneous exposure to topographical and soluble cues influence cell behavior.

Keywords: collagen, fibrils, keratocytes, microfluidic, PDMS, aligned

1. INTRODUCTION

It is now well appreciated that biochemical, mechanical, and topographical signals emanating from the local extracellular matrix (ECM) are key modulators of cell behavior during both physiological and pathological processes. For example biochemical gradients are important in regulating cell behavior during development(Tabata 2004), inflammation(David and Kubes 2019), and cancer metastasis(Roussos et al. 2011), while mechanical cues are relevant in differentiation(Tse and Engler 2011), tissue growth and migration.(Saez et al. 2007) Likewise topographical signals can modulate cell adhesion(Koh et al. 2010; Teixeira et al. 2003), cell morphology(Kiang et al. 2013; Whited and Rylander 2014), cell proliferation(Chaudhuri et al. 2016; Muhammad et al. 2015), cell differentiation(Dang and Leong 2007; Metavarayuth et al. 2016), and intracellular signaling(Dalby et al. 2003; Lou et al. 2018).

Although both micro and nanoscale topographies have been shown to modulate several aspects of cell behavior, current trends in tissue engineering and biomaterial development favor the development of topographical features that have dimensions and geometries comparable to native ECM structural components. Important structures of the ECM are structural fibers, which are most commonly made from fibrous proteins such as collagen, elastin, fibronectin, and laminin. Of these, type I fibrillar collagen is the most abundant fibrous protein, being present in a variety of connective tissues such as bone, ligaments, tendons, and skin. In addition to providing strength and elasticity to the tissue, the topography of aligned fibers in the ECM has been shown to direct the migration and orientation of cells through contact guidance(Chaubaroux et al. 2015).

Recent studies have demonstrated that topographical cues are important in the behavior of keratocytes(Teixeira et al. 2004; Vrana et al. 2008), fibroblasts(Muthusubramaniam et al. 2012; Phu et al. 2011), epithelial(Karuri et al. 2004; Teixeira et al. 2003), and endothelial cells(Gruschwitz et al. 2010; Koo et al. 2014) in the cornea. For example, it has been reported that culturing rabbit corneal fibroblasts on substrates containing microgroove topography (0.4 – 4.0 μm) could inhibit TGFβ-induced myofibroblastic differentiation(Myrna et al. 2012). In addition, culturing of human corneal fibroblasts on microgrooves(Guillemette et al. 2009) or transwell filters(Guo et al. 2007; Karamichos et al. 2014) stimulates aligned collagen deposition. Likewise, culturing of corneal epithelial cells(Teixeira et al. 2003) or keratocytes(Teixeira et al. 2004) on nanoscale grooves and ridges caused cells to align in the direction of the grooves via contact guidance. Finally culturing of primary human corneal endothelial cells on nanoscale to micron sized pillars not only enhanced proliferation but also enhanced tight-junction expression and cell morphology.(Muhammad et al. 2015)

A potential limitation of these previous studies is that they have generally used topographical substrates of unpolymerized ECM components that do not resemble the natural structure of the aligned type I collagen fibrils that make up the collagen lamella in the corneal stroma. Recent studies have identified a unique pattern of keratocyte alignment and connectivity during stromal repopulation following transcorneal freeze injury or lamellar keratectomy in vivo, which is highly correlated with the structural organization of the lamellae, suggesting contact guidance of intrastromal cell migration(Kivanany et al. 2016; Petroll et al. 2015). Given that contact guidance is frequently driven by both the density of fibers and the degree of fiber alignment(Lara Rodriguez and Schneider 2013), there is a need to study cornea keratocyte behavior on aligned collagen fibrils.

Over the years, several different approaches have been developed to fabricate aligned collagen fibrils. Some of the more common methods of forming aligned collagen fibrils include spin coating(Saeidi et al. 2011), electrospinning(Zhong et al. 2006), electrochemical fabrication(Cheng et al. 2008), magnetic fields(Torbet et al. 2007), and molecular crowding(Saeidi et al. 2012). In addition, several groups have employed shear flow conditions in microfluidic channels to generate Type I aligned collagen fibrils.(Köster et al. 2007; Lanfer et al. 2008; Lee et al. 2006; Saeidi et al. 2009) However, despite the abundance of techniques to fabricate Type I aligned collagen fibrils there has been limited application to the study of corneal keratocytes. Recently we reported on the effects of aligned fibrillar collagen substrates on corneal keratocyte morphology, cytoskeletal organization, and alignment after stimulation with PDGF or TGFβ.(Kivanany et al. 2018) In this study we further extend and characterize this experimental model, and assess the relationships between perfusion shear rate of the collagen solution and the alignment and density of collagen fibrils. We also report on how the degree of keratocyte alignment is dependent upon the substrate parameters.

2. METHODS

2.1. Silicon wafer mold fabrication

A photoresist template with a straight channel geometric pattern was fabricated by negative photolithography as described previously.(Coghill et al. 2013) The channel was designed in AutoCad (AutoDesk; Mill Valley, CA) to have a width of 1,500 μm and a length of 22 mm and a chrome mask was manufactured in house by the University of Texas at Dallas Cleanroom Research Laboratory. KMPR 1050 (Microchem; Westborough, MA) photoresist was spun at 2300 RPM onto a cleaned silicon wafer (University Wafers; Boston, MA) following manufacturers’ protocol to obtain a height of 60 μm. Wafers were then soft-baked on a digital aluminum hot plate (Torrey Pines Scientific; Carlsbad, CA) for 20 minutes. After soft-baking, wafers were cooled for 2 minutes and then inserted into a mask aligner (Karl Suss; Germany) with the chrome mask on top, and exposed to UV light to obtain the desired pattern. Following UV exposure, wafers were soft-baked a second time for 2 minutes, cooled afterwards for 2 more minutes, and then developed in SU-8 developer (Microchem; Westborough, MA) by gentle shaking for 4 minutes. Developed wafers were then rinsed with 40 mL of 2-propanol and dried with a steady stream of nitrogen gas for 30 seconds. Heights of the photoresist templates were measured by profilometry (Veeco Instruments; Oyster Bay, NY), while the template widths were analyzed via optical microscopy. Wafers were then placed in a desiccator under vacuum and exposed to the vapor of 10 μL of Tridecafluoro-1,1,2,2-tetrahydrooctyl)methyldichlorosilane (Gelest Inc; Morrisville, PA) for 4 hours in order to coat them with a nonstick fluorosilane monolayer to facilitate the subsequent release of PDMS molds.

2.2. PDMS Microfluidic Device Fabrication

PDMS devices were fabricated from Sylgard 184 Silicon Elastomer (Dow Corning; Midland, MI) by mixing the elastomer with the curing agent at a 10:1 mass ratio, pouring the PDMS mixture over the photoresist templates, and then degassing for 1 hour in a desiccator. Plastic elbows (Nordson Medical; Loveland, CO) were then placed over the inlet port of each channel following degassing and placed in a 80oC oven to cure for 1 hours. The PDMS devices were then peeled off from the photoresist templates, 3 mm diameter reservoirs cut out with a biopsy punch, and stored in a desiccator until ready to use. Glass coverslips (#1.5, Fisher Scientific; Hampton, NH) were cleaned by rinsing 3 times in 400 mL of Milli-Q water then submerged in a 28.8% nitric acid solution for 1 hour. After acid cleaning, coverslips were rinsed 3 times again in Milli-Q water with a final submersion in 400 mL for 5 min. Glass coverslips were then placed in a vacuum oven to dry completely. Dried coverslips were then made hydrophobic by submerging them in a 1% Aquasil siliconizing solution (Fisher Sci; Hampton, NH) for 15 seconds, and submerged a second time for 15 seconds in Milli-Q water. Coverslips were dried a second time in a vacuum and stored until needed. To reversibly bond the PDMS devices to the glass coverslips, the devices were exposed to an air plasma in a Harrick PlasmFlo chamber (Harrick Plasma; Ithaca, NY) for 1 min on high setting, and placed immediately on the glass substrate. The assembled PDMS devices were then placed on top of a digital aluminum hot plate set at 37°C for 30 minutes (Torrey Pines Scientific; Carlsbad, CA).

2.3. Collagen Solution Preparation

A chilled collagen solution was made with a 8:1:1 volume ratio of Bovine collagen type I (3.0 to 3.2 mg/mL depending on stock) (Advanced BioMatrix; Carlsbad, CA), 0.1M NaOH, and 10X minimal essential media (MEM, Life Tech; Carlsbad, CA). To adjust the final concentration of the collagen solution to 1.6 mg/ml, additional 1X MEM was added to the solution. Reagents were added in the following order: 10X MEM, 0.1 M NaOH, 1X MEM, and ending with the collagen. The pH of the solution was measured following mixing and if needed additional 0.1 M NaOH was added until the solution pH was between 7.2 – 7.6. The collagen solution was prepared immediately before patterning and stored on ice until needed.

2.4. Fabrication of Random Fibrils and Monomeric Collagen

Random collagen fibrils were fabricated by first bonding a PDMS ring to an aquasil-treated glass and placing it on a 37°C hotplate. The neutralized type I collagen solution (1.6 mg/ml) was then deposited onto the glass substrate with a pipette, and the solution within the PDMS well was then gently stirred. The collagen solution was allowed to polymerize for 30 min, washed gently with water, and then dried completely before adding a cell suspension. To fabricate a surface with monomeric collagen, a 50 μg/mL of type I collagen solution was deposited onto an aquasil glass substrate within a PDMS ring and heated on a 37°C hotplate for 30 min. The solution was aspirated off, washed gently with water, and dried completely before adding cells.

2.5. Aligned Collagen Fibril Formation

Aligned collagen fibrils were deposited onto hydrophobic glass coverslips by perfusing chilled collagen solutions at well-defined shear rates through a straight channel microfluidic device placed on a hotplate as illustrated in Figure 1. In specific, 1 mL syringes (BD Biosci; Franklin Lakes, NJ) were fitted with Luer-Lok connecting pieces (Nordson Medical; Loveland, CO), connected with 1.57 mm inner diameter laboratory tubing (Dow Corning; Midland, MI), and filled with the chilled collagen solution. Syringes with the collagen solution were quickly placed on a syringe pump (KD Scientific; Holliston, MA) and connected to the PDMS devices on top of the hot plate. Infusion was initially started at 10 μL/min for a volume of about 40 μL and stopped after the collagen solution reached the reservoir of the device. Flow rates were then changed to the desired shear rate, and the collagen solution was perfused through the microfluidic devices for 30 minutes. The wall shear rates of the collagen solutions perfused through the microfluidic channels were calculated from the solution of the Navier-Stokes equation for laminar flow of a Newtonian fluid between parallel plates: γwall = 6Q/(h2w), where γwall represents the wall shear rate (s−1), Q is the volumetric flow rate (μm3/s), h is the height of the microfluidic channel (60 μm), and w represents the width of the channel (1500 μm). Following perfusion, the devices were disassembled by peeling the PDMS stamps from the collagen fibril coated glass coverslips. The coverslips with aligned collagen fibrils were then rinsed with Milli-Q water for 10 seconds, and allowed to dry completely on the hotplate at 37°C, before being placed into Fluoroware holders (FSI International; Chaska, MN) until needed. As shown in Figure 1B, multiple substrates can be fabricated simultaneously by utilizing a 10-syringe pump.

Figure 1. Schematic of Aligned Collagen Fibril Formation via Microfluidic Deposition.

Figure 1.

Open in a new tab

(A) A PDMS device is bonded to an aquasil-treated glass coverslip. The device is placed on hotplate and infused with collagen solution. As the collagen solution polymerizes, aligned collagen fibrils are formed and deposited on the hydrophobic glass surface. After infusion, the PDMS stamp is peeled off the substrate. (B) Image of four substrates simultaneously being coated with aligned collagen fibrils.

2.6. DTAF Staining and Imaging of Aligned Collagen Fibrils

A 0.5% solution of (5-([4,6-Dichlorotriazin-2-yl]amino)fluorescein hydrochloride) (DTAF, Sigma-Adrich; St. Louis, MO) was prepared by dissolving DTAF in a 0.2 M sodium bicarbonate buffer. The solution was placed in a 37°C water bath for 30 minutes to assist in dissolving the DTAF. Staining of the collagen fibrils was accomplished by covering the collagen fibrils with 50 μL of the 0.5% DTAF solution for 2 minutes. The DTAF solution was aspirated off of the collagen fibrils, and the collagen fibril coated coverslips were submerged 3 times in Milli-Q water for 5 minutes each time. Fluorescent and differential interference contrast (DIC) imaging of the collagen-coated glass coverslips was performed by covering the substrates with Milli-Q water and placing them on either a Zeiss Vert.A1 (Zeiss Microscopy; Germany) inverted microscope equipped with a 63X Apochromat Oil Objective (NA = 1.4) and an ORCA-Flash 4.0 sCMOS camera (Hamamatsu, Bridgewater, NJ) or a Zeiss LSM 800 laser scanning confocal microscope. For quantitative analysis, all images were acquired with the same exposure time, and fluorescent intensity of each DTAF image was calculated via ImagePro (MediaCy; Rockville, MD). For each aligned collagen coated substrate, 5 images were taken at three different positions (e.g. upstream, midstream, downstream, Fig. 4) for a total of 15 images per fabricated substrate. For each perfusion shear rate evaluated, a minimum of at least three separate samples were fabricated and imaged on separate days.

Figure 4. Analysis of Fibril Density Across the Microchannel Length.

Figure 4.

Open in a new tab

(A) Schematic of channel and location of images taken. (B) Average intensity measurements of DTAF labeled collagen fibrils on substrates at the three different image locations for each shear rate, The data represent the mean ± SEM of three independent experiments, each with triplicate observations for each location.

2.7. Fibril Orientation Index Calculations

The degree of collagen fibril alignment was quantified by using an intensity gradient technique to analyze the fluorescent images of the DTAF stained collagen fibrils. In this method, each image is divided into a grid of sub-regions and the dominant fibril angle within each sub-region is determined from pixel intensity gradients using custom-written Matlab code (MathWorks; Natick, MA)(Gjorevski et al. 2015; Karlon et al. 1998). An angle of +/− 90° represents alignment perpendicular to the direction of flow whereas an angle of 0° represents alignment along the direction of flow. Circular histograms (or rose plots) were used to show the distribution of aligned fibrils. To quantify the degree of collagen fibril orientation relative to the direction of flow, an orientation index (OIfibril) was calculated from the angle distributions using the following equation:

OIfibril=2<cos2Φ>−1

where

cos2Φ=∑i=1Ncos2ΦiN

Where Φi represents the angle measured in a given sub-region of the image, and N is the total number of sub-regions. An orientation index of 1.0 corresponds to the case where the fibril distribution is perfectly aligned in the direction of flow, while a random fibril distribution will have a value of 0.

2.8. PDMS Ring – Microwell Culture System

In order to obtain high resolution microscopic images of keratocytes interacting with the aligned collagen fibrils we developed a PDMS ring-glass coverslip culture system that was compatible with high numerical aperture objectives. In the first step a 35 mm petri dish was half filled with a 10:1 mixture of PDMS elastomer:curing agent that had been degassed (Figure S1). Next the cap of a 15 mL conical centrifuge tube is placed in the middle of the PDMS containing petri dish, and the petri dish is covered with its lid and placed in an oven for 1 hour at 80°C. After curing, the PDMS is removed from the Petri dish with an Xacto knife and the centrifuge tube cap is removed leaving behind a PDMS ring with an inner diameter of 15 mm.

To achieve a tight leak-proof seal between the PDMS ring and the aligned collagen fibril coated glass coverslip, both were treated with an air plasma on high RF for 1 min. In order to protect the aligned collagen fibrils from being etched during the plasma treatment, a small slab of PDMS was placed over the collagen fibrils area. Following plasma treatment, the PDMS slab protecting the collagen fibrils was removed and the PDMS ring was immediately brought into contact with the collagen fibril coated glass coverslip to form a permanent seal and creating a well for culturing of the keratocytes.

2.9. Primary Keratocyte Cell Extraction and Cell Culture

Primary rabbit corneal keratocytes (NRKs) were isolated from eyeballs of New Zealand White Rabbits (Pelfreez Arkansas; Rodgers, AR) as previous described (Kivanany et al. 2018; Miron-Mendoza et al. 2017). Following isolation NRKs were cultured in Dulbecco’s modified Eagle’s medium (DMEM) that was supplemented with 100 μM nonessential amino acids (Invitrogen; Carlsbad, CA), 100 μg/mL ascorbic acid, 1% RPMI vitamin mix, and 1% PenStrep (Invitrogen; Carlsbad, CA) to maintain the keratocyte phenotype.(Kivanany et al. 2018) Cells were cultured in basal media for 4 days before seeding onto the aligned collagen fibril substrates.

At the beginning of an adhesion experiment, keratocytes were detached from the culture flasks by adding 2 ml of a 0.25% trypsin/EDTA solution to the flask, placed in a 37°C incubator for 2 minutes, washed with serum-free media, and diluted to 20,000 cells/ml in serum free media supplemented with 50 ng/mL PDGF-BB (Millipore; Burlington, MA). Approximately 2 ml of the keratocyte suspension was added to each PDMS Ring culture system, and four of the PDMS Ring microwells were placed in a 150 mm petri dish (Fig. S2), covered, and placed into an incubator at 37°C and 5% CO2 for 48 hrs.

2.10. Fluorescent imaging of cells

At the end of an experiment, the cells were fixed in 4% paraformaldehyde solution (PFA, Electron Microscopy Sciences; Hatfield, PA) for 10 minutes, washed two times with PBS, permeabilized with a 0.5% Triton X-100 in phosphate buffered saline (PBS) for 15 minutes and washed again with PBS. To label the F-actin, the permeabilized cells were incubated with Alex Fluor 488 Phalloidin (Molecular Probes, Eugene, OR) for 1 hour at 37°C, while the cell nuclei were stained by incubating the fixed cells with a 4’,6-diamidino-2-phenylindole (DAPI) solution (1:1000 dilution, in PBS) for 10 minutes at room temperature and washed with PBS. The phalloidin and DAPI stained keratocytes were imaged with an inverted microscope equipped with DIC and fluorescent capabilities,

The degree of keratocyte alignment was assessed by determining the percent of image content percentage (I(α)) aligned at each radial angle within an image by using ImageJ (NIH; Bethesda, MD) Directionality Fourier components analysis procedure, and then calculating an orientation index (OIkeratocyes). The orientation index of keratocyte alignment was calculated with the following equations:

OIkeratocyes(θ)={2〈cos2θ〉−1}

where

cos2θ=∫−9090I(α){cos2(α−θ)}dα∫−9090I(α)dα

and θ = 0° corresponds to the direction of the collagen fibrils. In this analysis, an OIkeratocyes value of 100% corresponds to the keratocyte being aligned in the direction of the collagen fibrils, while a value of −100% corresponds to the keratocytes having an alignment perpendicular to the fibril alignment. When the keratocyte alignment is random the OIkeratocyes value will be 0

2.11. Statistical Analysis.

Differences in the alignment of the keratocytes and collagen fibrils were determined using a two-way ANOVA with Tukey’s Post-Hoc test in GraphPad Prism. A value of α = 0.05 was used to determine significance.

3. RESULTS AND DISCUSSION

3.1. Effect of Perfusion Time on Collagen Fibril Deposition

Aligned collagen fibrils were deposited on glass coverslips by a microfluidic based technique in a multi-step procedure (Fig. 1) similar to that described by other groups(Lanfer et al. 2009; Lanfer et al. 2008; Saeidi et al. 2009): (a) glass coverslips were made hydrophobic by treatment with a hydrophobic silane (i.e. aquasil coating, trimethoxy(octadecyl)silane); (b) microfluidic chambers were reversibly attached to the aquasil coated slide via plasma bonding; (c) the microfluidic device was placed on a hot plate at 37°C and chilled solutions of type I collagen were perfused through the device at various flow rates and collagen concentrations. A significant and important difference between our procedure and these previous reports is the use of a digital hotplate instead of a CO2-free incubator(Lanfer et al. 2009; Lanfer et al. 2008) or a temperature controlled microscope perfusion chamber(Saeidi et al. 2009). The use of the hotplate not only reduces the cost and effort of experimentation but also provides a portable setup that can be quickly and easily assembled in laboratories where resources and time are limited. In addition, this setup provides a high throughput process that allows for several glass coverslips to be simultaneously modified with aligned collagen fibrils (Figure 1B).

An initial set of experiments was performed to determine the time needed to form a high density of collagen fibrils. Figure 2 shows representative differential interference contrast (DIC) images of the collagen fibrils formed when a 1.6 mg/ml type I collagen concentration was perfused at 150 s−1 for 5, 10, 20, and 30 minutes. At early time points, short, individual fibrils were sparsely deposited on the surface, and the fibrils were generally aligned in the direction of flow. The density of fibrils deposited increased with time, with the surface becoming covered with a dense, mat-like coverage of fibrils at later time points (≥ 15 minutes). Although a majority of the fibrils were short and isolated, clumps of collagen fibrils were sometimes observed. Because the diameters of the fibrils were near or less then the resolution of the optical objective used (63X 1.4 NA) we did not measure the diameter or length of the fibrils formed. The dependence of fibril density on perfusion time that we observed was consistent to the results reported by others.(Lanfer et al. 2008; Saeidi et al. 2009)

Figure 2: Effect of Perfusion Time on Collagen Fibril Deposition.

Figure 2:

Open in a new tab

Representative DIC images of collagen fibrils deposited on glass substrates when chilled collagen solutions are perfused through the microfluidic device at a shear rate of 150 s−1 for different perfusion times of (A) 5 min, (B) 10 min, (C) 15 min, and (D) 30 min. Scale bar = 10 μm

3.2. Density of Collagen Fibrils Deposited Increases with Shear Rate

To examine the effect of shear rate on (i) the density of deposited collagen fibrils and (ii) the degree of fibril alignment, we perfused collagen solutions (1.6 mg/ml) at well-defined shear rates (25, 75, 150, and 225 s−1) over aquasil treated coverslips for 30 minutes, and imaged the substrates by both DIC and fluorescence microscopy. At the lowest shear rate tested (25 s−1), a low density of fibrils were formed on the glass substrates. Although the fibrils were generally aligned in the direction of flow, the fibrils often displayed a wavy morphology (Fig. 3A). As the shear rate was increased to 75 s−1 the length of the fibrils appeared to decrease while the density of fibrils increased (Fig. 3B). In addition, small spherical or punctate globules were also observed on the substrates. At a shear rate of 150 s−1 (Fig. 3C) a high density of fibrils were deposited on the substrates, with many of the fibrils overlapping and/or intertwining with each other. At 225 s−1 (Fig. 3D) a majority of the substrate was covered with fibrils and a large number of the fibrils appeared to be woven together into frayed rope-like structures.

Figure 3: Effect of Shear Rate on Collagen Fibril Deposition.

Figure 3:

Open in a new tab

(A-D) Representative DIC images of collagen fibrils deposited on glass substrates when chilled collagen solutions are perfused through the microfluidic device for 30 minutes at various shear rates. (E-H) Fluorescent images of the same substrates stained with DTAF, scale bar = 10 μm.

In order to quantify the density of the fibrils deposited at the various shear rates, we labeled the deposited collagen fibrils with dicholorotriazynl fluorescein (5-DTAF) and fluorescently imaged the substrates. 5-DTAF has been previously used to label collagen fibrils in collagen matrices(Karamichos et al. 2009) and in vivo(Jester et al. 1995). Figures 3E–3H shows representative fluorescent images of the density of deposited collagen fibrils at the various shear rates. Similar to the results of the DIC imaging, fluorescent images also showed that the density of deposited collagen fibrils increased with shear. To determine whether there were any differences in the density of fibrils deposited along the length of the channel, we took images at upstream, midstream, and downstream locations (Fig. 4A). As shown in Figure 4B, the density of fibrils deposited was fairly uniform throughout the channel. Taken together, this data demonstrates that (i) our method is able to deposit fibrils uniformly throughout the channel and (ii) the density of fibril deposition is dependent upon the shear conditions and time.

At first glance our observation that fibril density increased with shear rate would seem logical, since an increased shear or flow rate would increase the total amount of collagen perfused through the chamber for a given amount of time and thus lead to an increased amount of fibrils deposited. However there are conflicting results in the literature on the effect of flow or shear on the deposition of collagen fibrils. For example Lanfer et al, reported that the amount of adsorbed collagen fibrils increased as the flow rate of collagen was decreased from 11 μL/min to 0.45 μL/min.(Lanfer et al. 2008) In contrast, Saeidi et al, showed that the number of collagen fibril nucleation sites increased as the shear rate increased from 9 – 80 s−1. However further increasing the shear rate to 500 s−1 did not increase the number of fibrils formed. Our observation that the density of fibril deposition increased with shear at low shear rates is similar to those reported by Saeidi et al(Saeidi et al. 2009). We suspect that differences in fibril deposition between these studies are related to differences in the experimental setups. For example it has been shown that the hydrophobic nature of the substrate is an important variable in fibril formation(Lanfer et al. 2008). Likewise the geometry of the microfluidic devices will not only affect the shear profile inside the channel, but will also affect the surface area to volume ratio, total volume, and heat transfer efficiency, and hence the warming rate of the collagen solution(Sung et al. 2009). Given the sensitive nature of the nucleation and growth phase of the collagen polymerization process to temperature, we suspect that small changes in the warming rate of the collagen solution in the different devices can have significant effects on fibril formation

3.3. Effect of Perfusion Shear Rate on Collagen Fibril Alignment

To investigate the relationship between shear rate and the degree of collagen fibril alignment in our system, we analyzed fibril orientation using a custom-written code in Matlab that divides each image into sub-regions and determines the dominant fibril angle within each sub-region from pixel intensity gradients(Gjorevski et al. 2015; Karlon et al. 1998). The distribution of fibril angles within each image was then compared with the direction of flow. Figures 5A–5D shows representative circular histograms (rose plots) of the angles between the collagen fibrils and the direction of flow. In general, a majority of the fibrils had an orientation in the direction of flow (±20°) at all the shear rates tested. The degree to which the fibrils were aligned did not appear to be dependent upon the shear rate.

Figure 5. Effect of Collagen Solution Perfusion Shear Rate on Collagen Fibril Alignment.

Figure 5.

Open in a new tab

Representation rose histograms that show the distribution of the alignment angles of the deposited collagen fibrils relative to the direction of flow in polar coordinates as a function of collagen solution perfusion shear rate: (A) 25 s−1, (B) 75 s−1, (C) 150 s−1, and (D) 225 s−1. Each histogram represents the accumulated data from 9 individual images from a single substrate.

To determine whether the alignment of collagen fibrils was dependent on its position in the microfluidic channel, we calculated a collagen fibril orientation index (OIfibrils). As shown in Figure 6, the OIfibrils values ranged between 0.55 and 0.75. which suggests that there is overall alignment of fibrils in the direction of flow. In addition, for a given shear rate, the OIfibrils values were fairly consistent throughout the channel. To determine whether there were any statistical differences between the OIfibrils values at the different shear rates we performed a two-way ANOVA test and found that there was no significant difference between any of the shear rates or locations. Taken together, these results (Figure 5 & 6) demonstrate that the collagen fibrils deposited consistently aligned in the direction of flow across all shear rates tested.

Figure 6: Effect of Perfusion Shear Rate and Channel Location on Fibril Alignment.

Figure 6:

Open in a new tab

Quantitative assessment of collagen fibril alignment as measured by the orientation index as a function of collagen solution perfusion shear rate from rose plot analysis. Collagen fibril orientation index was measured at three different positions on each substrate, and three different substrates were analyzed. Error bars are S.D.

3.4. Keratocyte Morphology on Random and Aligned Collagen Fibrils

To determine whether the aligned collagen fibrils could serve as a platform for keratocyte cell culture and influence their orientation, primary rabbit keratocytes were seeded onto substrates containing aligned collagen fibrils for 48 hours in serum-free media supplemented with PDGF-BB. As a control, keratocytes were also seeded onto substrates containing randomly aligned collagen fibrils or substrates coated with monomeric collagen in the same media conditions. Figure 7 shows representative DIC images of the overall morphology of keratocytes cultured on (A) aligned collagen fibrils (150 s−1), (B) randomly aligned fibrils, and (C) monomeric collagen in the presence of PDGF BB after 48 hours. Keratocytes cultured on randomly aligned collagen fibrils and monomeric collagen were randomly oriented, elongated, and had multiple extensions that showed no preferential orientation. These observations are consistent with previous reports when keratocytes are cultured in the presence of PDGF BB(Kivanany et al. 2018). In contrast, keratocytes cultured on aligned fibrils were narrow, elongated and aligned in the direction of the collagen fibrils. These results suggest that the aligned collagen fibrils produced by microfluidic deposition support keratocyte culture and cause keratocytes to elongate and migrate in the direction of fibril alignment.

Figure 7: Effect of Collagen Fibril Alignment on Keratocyte Alignment.

Figure 7:

Open in a new tab

DIC and fluorescent images of primary keratocytes cultured for 48 hrs in the presence of PDGF on substrates coated with monomeric collagen, random collagen fibrils, and aligned collagen fibrils. Quantitative analysis of the alignment of the above keratoctye images. Scale = 100 μm.

3.5. Keratocyte Orientation is Influenced by Collagen Fibril Density.

Having shown that the keratocytes would grow and orient on aligned collagen fibrils, we went on to investigate the effect of collagen fibril density on keratocyte orientation. For this set of experiments, we seeded keratocytes onto substrates containing aligned collagen fibrils deposited at the various shear rates. Figure 8 shows representative DIC and fluorescent images of the overall morphology of keratocytes on these substrates after 48 hours in the presence of PDGF-BB. When keratocytes were cultured on substrates containing a low density of aligned collagen fibrils (i.e. deposition shear rates of 25 and 75 s−1), the cells had multiple extensions that showed no preferential orientation similar to the behavior of cells cultured on random collagen fibrils. Similarly the actin cytoskeleton and nucleus of these keratocytes also displayed random orientations (Fig. 8). In contrast, when keratocytes were seeded on substrates with a high density of aligned collagen fibrils (150 or 225 s−1), the keratocytes displayed an elongated morphology that was parallel to the direction of the underlying collagen fibrils, and their extensions, cytoskeletons, and nuclei also appeared aligned with the fibrils.

Figure 8: Influence of Collagen Perfusion Shear Rate on Corneal Keratocyte Alignment.

Figure 8:

Open in a new tab

DIC (A-D) and fluorescent (E-H) images of keratocytes plated on substrates coated with aligned fibrillar fabricated at different collagen perfusion shear rates and cultured for 2 days in serum-free media supplemented with PDGF. Green: phalloidin, blue: DAPI.

To quantify the extent of keratocyte alignment on the substrates of varying fibril density, we analyzed the fluorescent images with an ImageJ Directionality plug-in using Fourier components to generate alignment histograms. An angle of 0° represents the case in which the keratocytes were perfectly aligned in the same direction of the collagen fibrils, while 90° represents the case in which the keratocyte is perpendicular to the aligned fibrils. Figure 9A shows representative histograms of the alignment angle distributions that were measured for the different substrates. Similar to the qualitative observations shown in Figure 8, quantitative histogram analysis demonstrated that on substrates with a low density of aligned fibrils the keratocytes showed no alignment with the fibrils (Figure 9B). However on substrates with a high density of aligned collagen fibrils (225 s−1) the keratocytes showed a high degree of alignment with the fibrils. These results suggest that the density of aligned collagen fibrils is an important parameter to induce keratocyte orientation.

Figure 9. Alignment of Keratocytes Cultured on Substrates of Different Collagen Fibril Density.

Figure 9.

Open in a new tab

(A) Alignment histograms were generated to examine and quantify the alignment of the keratocytes with the collagen fibril direction for the substrates generated by the different collagen perfusion shear rates. These representative histograms were generated from the experiments represented in Figure 8. (B) Plot of the Orientation Index of the keratocytes cultured on substrates fabricated with different collagen perfusion shear rates. The data represent the mean ± SEM of three independent experiments.

4. CONCLUSIONS

In this study we report a method for depositing aligned collagen fibrils onto glass coverslips by perfusing chilled collagen solutions through straight microfluidic channels. Using this technique we were able to generate large areas (1500 μm wide by 22 mm long) of aligned collagen fibrils. Using both fluorescent imaging and differential interference contrast (DIC) microscopy we were able to demonstrate that the density and degree of collagen fibril alignment was dependent upon the perfusion time and the perfusion shear rate. When primary rabbit keratocytes were cultured on substrates with a high density of aligned collagen fibrils in the presence of PDGF the keratocytes preferentially aligned parallel to the fibril orientation. However when keratocytes were cultured on substrates of low fibril density the keratocytes did not align. These results are important and novel because (i) they demonstrate the importance of both topographical and soluble cues in influencing keratocyte behavior; (ii) most microfluidic methods for depositing aligned collagen fibrils require expensive or specialized equipment, in contrast our method of employing a hot plate makes our procedure easy and portable; (iii) the use of a multichannel syringe pump allows for up to eight substrates to be coated with aligned collagen fibrils simultaneously; (iv) to our knowledge this report is the first to investigate the effect of aligned collagen fibril density on keratocyte behavior. We expect that this simple, high-throughput method of fabricating aligned collagen fibrils will provide a powerful platform to better understand the spatial and temporal dynamics of keratocyte behavior in response to simultaneous exposure to topographical cues, soluble cues, and different ECM coatings.

Supplementary Material

Supplementary Material

ACKNOWLEDGEMENTS

This work was supported in part by grants from the National Institutes of Health (R01 EY013322, R01 EY030190), a Trainee Fellowship from the UT-Southwestern Hamon Center for Regenerative Science and Medicine (CRSM) Trainee to KHL, a Pilot and Feasibility grant from the UT Southwestern O’Brien Kidney Research Core Center, a grant from Research to Prevent Blindness, Inc., and funds from the Office of Vice President of Research at the University of Texas at Dallas. The authors would like to thank Somdutta Chakraborty for assistance with some of the confocal imaging. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Footnotes

SUPPLEMENTARY MATERIALS

In the supplementary material, supplementary Figure S1 provides additional information about the steps used to fabricate the PDMS rings. Supplementary Figure S2 provides information about how the PDMS rings and substrates coated with aligned collagen fibrils were integrated and allowed for cell culture.

REFERENCES

  1. Chaubaroux C, Perrin-Schmitt F, Senger B, Vidal L, Voegel J-C, Schaaf P, Haikel Y, Boulmedais F, Lavalle P, and Hemmerlé J, Tissue Eng. Part C Methods 21, 881 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Chaudhuri PK, Pan CQ, Low BC, and Lim CT, Sci. Rep 6, 19672 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Cheng X, Gurkan UA, Dehen CJ, Tate MP, Hillhouse HW, Simpson GJ, and Akkus O, Biomaterials 29, 3278 (2008). [DOI] [PubMed] [Google Scholar]
  4. Coghill PA, Kesselhuth EK, Shimp EA, Khismatullin DB, and Schmidtke DW, Biomed. Microdevices 15, 183 (2013). [DOI] [PubMed] [Google Scholar]
  5. Dalby MJ, Riehle MO, Yarwood SJ, Wilkinson CDW, and Curtis ASG, Exp. Cell Res 284, 272 (2003). [DOI] [PubMed] [Google Scholar]
  6. Dang JM and Leong KW, Adv. Mater 19, 2775 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. David BA and Kubes P, Immunol. Rev 289, 9 (2019). [DOI] [PubMed] [Google Scholar]
  8. Gjorevski N, Piotrowski AS, Varner VD, and Nelson CM, Sci. Rep 5, 11458 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Gruschwitz R, Friedrichs J, Valtink M, Franz CM, Muller DJ, Funk RHW, and Engelmann K, Invest Ophthalmol Vis Sci 51, 6303 (2010). [DOI] [PubMed] [Google Scholar]
  10. Guillemette MD, Cui B, Roy E, Gauvin R, Giasson CJ, Esch MB, Carrier P, Deschambeault A, Dumoulin M, Toner M, Germain L, Veres T, and Auger FA, Integr. Biol 1, 196 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Guo X, Hutcheon AEK, Melotti SA, Zieske JD, Trinkaus-Randall V, and Ruberti JW, Investig. Ophthalmol. Vis. Sci 48, 4050 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Jester JV, Petroll WM, Barry PA, and Cavanagh HD, Invest. Ophthalmol. Vis. Sci 36, 809 (1995). [PubMed] [Google Scholar]
  13. Karamichos D, Funderburgh ML, Hutcheon AEK, Zieske JD, Du Y, Wu J, and Funderburgh JL, PLoS One 9, e86260 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Karamichos D, Lakshman N, and Petroll WM, Cell Motil. Cytoskeleton 66, 1 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Karlon WJ, Covell JW, Mcculloch AD, Hunter JJ, and Omens JH, 625, 612 (1998). [DOI] [PubMed] [Google Scholar]
  16. Karuri NW, Liliensiek S, Teixeira AI, Abrams G, Campbell S, Nealey PF, and Murphy CJ, J Cell Sci 117, 3153 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Kiang JD, Wen JH, Del Álamo JC, and Engler AJ, J. Biomed. Mater. Res. - Part A 101 A, 2313 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Kivanany P, Grose K, Yonet-Tanyeri N, Manohar S, Sunkara Y, Lam K, Schmidtke D, Varner V, and Petroll W, J. Funct. Biomater 9, 54 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kivanany PB, Grose KC, and Petroll WM, Exp. Eye Res 153, 56 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Koh LB, Rodriguez I, and Venkatraman SS, Biomaterials 31, 1533 (2010). [DOI] [PubMed] [Google Scholar]
  21. Koo S, Muhammad R, Peh GSL, Mehta JS, and Yim EKF, Acta Biomater. 10, 1975 (2014). [DOI] [PubMed] [Google Scholar]
  22. Köster S, Leach JB, Struth B, Pfohl T, and Wong JY, Langmuir 23, 357 (2007). [DOI] [PubMed] [Google Scholar]
  23. Lanfer B, Freudenberg U, Zimmermann R, Stamov D, Körber V, and Werner C, Biomaterials 29, 3888 (2008). [DOI] [PubMed] [Google Scholar]
  24. Lanfer B, Seib FP, Freudenberg U, Stamov D, Bley T, Bornhäuser M, and Werner C, Biomaterials 30, 5950 (2009). [DOI] [PubMed] [Google Scholar]
  25. Lara Rodriguez L and Schneider IC, Integr. Biol 5, 1306 (2013). [DOI] [PubMed] [Google Scholar]
  26. Lee P, Lin R, Moon J, and Lee LP, Biomed. Microdevices 8, 35 (2006). [DOI] [PubMed] [Google Scholar]
  27. Lou HY, Zhao W, Zeng Y, and Cui B, Acc. Chem. Res 51, 1046 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Metavarayuth K, Sitasuwan P, Zhao X, Lin Y, and Wang Q, ACS Biomater. Sci. Eng 2, 142 (2016). [DOI] [PubMed] [Google Scholar]
  29. Miron-Mendoza M, Graham E, Manohar S, and Petroll WM, Matrix Biol. 64, 69 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Muhammad R, Peh GSL, Adnan K, Law JBK, Mehta JS, and Yim EKF, Acta Biomater. 19, 138 (2015). [DOI] [PubMed] [Google Scholar]
  31. Muthusubramaniam L, Peng L, Zaitseva T, Paukshto M, Martin GR, and Desai TA, J Biomed Mater Res Part A 100A, 613 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Myrna KE, Mendonsa R, Russell P, Pot SA, Liliensiek SJ, Jester JV, Nealey PF, Brown D, and Murphy CJ, Invest Ophthalmol Vis Sci 53, 811 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Petroll WM, Kivanany PB, Hagenasr D, and Graham EK, Investig. Ophthalmol. Vis. Sci 56, 7352 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Phu D, Wray LS, Warren RV, Haskell RC, and Orwin EJ, Tissue Eng Part A 17, 799 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Roussos ET, Condeelis JS, and Patsialou A, Nat. Rev. Cancer 11, 573 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Saeidi N, Karmelek KP, Paten JA, Zareian R, DiMasi E, and Ruberti JW, Biomaterials 33, 7366 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Saeidi N, Sander EA, and Ruberti JW, Biomaterials 30, 6581 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Saeidi N, Sander EA, Zareian R, and Ruberti JW, Acta Biomater. 7, 2437 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Saez A, Ghibaudo M, Buguin A, Silberzan P, and Ladoux B, Proc. Natl. Acad. Sci 104, 8281 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Sung KE, Su G, Pehlke C, Trier SM, Eliceiri KW, Keely PJ, Friedl A, and Beebe DJ, Biomaterials 30, 4833 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Tabata T, Development 131, 703 (2004). [DOI] [PubMed] [Google Scholar]
  42. Teixeira AI, Abrams GA, Bertics PJ, Murphy CJ, and Nealey PF, J. Cell Sci 116, 1881 (2003). [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Teixeira AI, Nealey PF, and Murphy CJ, J. Biomed. Mater. Res. - Part A 71, 369 (2004). [DOI] [PubMed] [Google Scholar]
  44. Torbet J, Malbouyres M, Builles N, Justin V, Roulet M, Damour O, Oldberg Å, Ruggiero F, and Hulmes DJS, Biomaterials 28, 4268 (2007). [DOI] [PubMed] [Google Scholar]
  45. Tse JR and Engler AJ, PLoS One 6, e15978 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Vrana E, Builles N, Hindie M, Damour O, Aydinli A, and Hasirci V, J Biomed Mater Res Part A 84A, 454 (2008). [DOI] [PubMed] [Google Scholar]
  47. Whited BM and Rylander MN, Biotechnol. Bioeng 111, 184 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Zhong S, Teo WE, Zhu X, Beuerman RW, Ramakrishna S, and Yung LYL, J. Biomed. Mater. Res. - Part A 79, 456 (2006). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
NIHMS1586474-supplement-Supplementary_Material.docx (3.2MB, docx)

RESOURCES