Platelet adhesion and aggregation are shear stress dependent phenomena. A repertoire of different receptors promote rolling and adhesion to extracellular matrix proteins within the vascular wall and support homotypic and heterotypic cohesion between platelets and other blood cells [1
]. A useful research tool to study these interactions are microfluidic flow assays, which we refer to here as the suite of microfluidic devices that have been developed to measure platelet function under physiologic and pathologic hemodynamics [2
]. The most common adhesive substrate used in these assays is type I collagen [3
], but a promising alternative is collagen related peptides [4
], which has advantages both in micropatterning and dissecting the different functional roles of collagen on platelet adhesion and activation.
Type I collagen is found in the vascular wall and plays, at least, three roles in supporting platelet adhesion and activation [5
]. First, it binds von Willebrand factor (VWF) in the plasma. VWF is necessary for platelet adhesion at high shear rates (>600 s−1
) by supporting rolling mediated by the glycoprotein Ib-V-IX complex [6
]. Second, it has binding sites for the integrin α2
that supports firm adhesion [7
]. Third, it has binding sites for glycoprotein VI (GPVI), which serves as a key signaling receptor for platelet activation [8
Type I collagen derived from animal and human sources has been the primary collagen substrate for platelet adhesion assays in parallel plate flow chambers. These flow chambers were the primary research tool for platelet and leukocyte adhesion from the 1980s to the 1990s [10
]. In the 2000s, the parallel plate flow chamber was miniaturized with the growing popularity and accessibility of microfluidics [2
]. This transition to smaller channels came with several benefits; lower blood volumes, higher throughput, anatomically inspired geometries, and the ability to micropattern adhesive substrates. Yet as channel sizes became smaller, the ability to homogenously pattern type I collagen emerged as a problem.
Type I collagen fibers are 10–100 µm long and adsorb to substrates in a random pattern [11
]. In parallel flow chambers with channel widths on the order of 10 mm this was not a problem because a large enough representative area could mask the microscale heterogeneity of the fiber distribution. In microfluidic devices, as the channel width approaches the collagen fiber length, it becomes more difficult to achieve homogeneity. One solution is to mechanically degrade the fibers into smaller fibers, for example, by sonication [13
]. Another approach is to deposit the collagen at a high enough concentration to completely coat the surface [14
]. However, very high surface concentrations of collagen can provide such a strong platelet activation signal as to mitigate the importance of other platelet autocrine signaling mediated by adenosine diphosphate (ADP) and thromboxane A2. Moreover, large collagen fibers can extend off the surface into the flowing blood, acting as a net for platelets and other blood cells, in an unpredictable manner [4
Alternatives to native type I collagen fibers include collagen thin films [11
], acid soluble collagen [15
], type III collagen [16
], and collagen peptides [4
]. Among these alternatives, collagen peptides appear to best mimic the platelet response to type I collagen. Pugh et al. reported that a combination of three peptides—VWF binding peptide (VWF-III), α2
binding peptide (GFOGER), and GPVI binding peptide or collagen related peptide (CRP)—supported platelet adhesion over a physiologic range of shear rates and resulted in platelet aggregates with similar volumes and morphologies as those formed on type I collagen [4
]. These peptides, in combination with other adhesive proteins, have been used in microfluidic devices to probe the relative contribution of platelet receptors on adhesion, activation, and aggregation [18
All prior platelet adhesion studies we are aware of adsorb collagen peptides to glass surfaces that serve as the bottom wall of a flow chamber. However, microfluidic patterning devices and flow chamber walls where adhesive proteins are adsorbed are often made of polydimethylsiloxane (PDMS) [15
]. In this study, we examine the adsorption and absorption of collagen peptides to devices in which three or four walls of a channel are made of PDMS. Our findings suggest that collagen peptides are absorbed into PDMS, causing negligible patterning of peptides to an assay surface (e.g., glass) when patterning with a PDMS device or onto a PDMS surface. Modification of PDMS with a fluorinated alkyl silane blocks absorption and promotes collagen peptide adsorption, allowing for platelet adhesion and aggregation under flow.
2. Materials and Methods
Bovine serum albumin (BSA) (A9418), 3,3’-dihexyloxacarbocyanine iodide (DiOC6) (D273), glutaraldehyde (340855) heparin salt (H3393), 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) (H3375), poloxamer 407 (Pluronics® F-127, P2443) were from Sigma–Aldrich (St Louis, MO, USA). 3.2% sodium citrate vacutainers (369714) and 21-gauge Vacutainer® 21 Safety-Lok blood collection sets (367281) were from Becton Dickson (Frankwood Lakes, NJ, USA). 500 µL and 100 uL glass luer lock syringes were from Hamilton (Reno, NV, USA). Tridecafluoro-1,1,2,2-tetrahydrooctyltrichlorosilane (FOTS) (SIT8174.0) was from Gelest (Morrisville, PA, USA). Polydimethylsiloxane and crosslinker were obtained from Krayden (Denver, CO, USA). Collagen related peptides (CRP-XL) [GCO(GPO)10GCOG-amide], 5(6)-carboxyfluorescein (FAM) labeled CRP-XL, GFOGER [GPC(GPP)5GFOGER(GPP)5GPC-amide], and VWF-III [GPCGPP)5GPRGQOGVMGFO(GPP)5GPC-amide] were obtained from Cambcol Laboratories (Cambridgeshire, UK). Glass slides (2” x 3“) were obtained from Fisher Scientific (Lenexa, KS, USA). Pierce Quantitative Fluorometric Peptide Assay (23290) and Texas Red dye (T20175) were from Thermo Fischer (Denver, CO, USA). Phosphate-buffered saline (PBS) was prepared to 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, and pH 7.4.
2.2. Blood Collection and Preparation
Blood was collected from healthy donors by venipuncture into 4.5 mL, 3.2% sodium citrate (final concentration) vacutainers using a 21-gauge needle. The first vacutainer was discarded. DiOC6 was added to the citrated whole blood to a final concentration of 1 μM and incubated at 37°C for 10 min prior to running flow assays. The study and consent process received institutional review board approval from the Colorado Multiple Institutional Review Board in accordance with the Declaration of Helsinki.
2.3. Glass, PDMS and Silicon Substrate Functionalization
Glass slides were immersed in a 50:50 solution of 12 M HCl:methanol for 4 h at room temperature. They were then rinsed with deionized (DI) water, rinsed with isopropyl alcohol, dried with an air brush, and dehydrated in an oven at 60 °C for 4 h. A thin film of PDMS was achieved by pouring a degassed mixture of PDMS base and crosslinker (mixed in 10:1 ratio) into a petri dish. A glass slide was firmly pressed into the petri dish containing PDMS and placed in an oven to cure for 24 h. The slide was removed from the dish and excess PDMS was removed, leaving a thin film of PDMS on one side of the slide. For FOTS functionalization, the PDMS-coated slide was treated with an oxygen plasma for 30 seconds prior to FOTS deposition (Plasma Cleaner, PDC-001, Harrick Plasma, Ithaca, NY, USA). FOTS was vapor deposited onto glass slides, silicon wafers, or slides with PDMS films in a desiccator under vacuum for 12 h.
2.4. Microfluidic Patterning Collagen Peptides
A PDMS microfluidic channel (length (l) = 49 mm, width (w) = 100 µm, height (h) = 50 µm) was either left untreated or was blocked with 2% BSA in PBS or 2% Pluronics F-127 for 45 min and then thoroughly rinsed with deionized water, as indicated in the text. To pattern peptides in a strip, the untreated or blocked channel was laid horizontally across glass, PDMS, FOTS-glass, or FOTS-PDMS surfaces. Collagen peptides were mixed together in 10 mM acetic acid to a final concentration of 250 μg/mL each. The microfluidic channel was filled with the peptide solution and incubated for 2 h at room temperature in a petri dish with wet Kim-Wipes® to maintain a humid environment. The channel was then rinsed with 10 mM acetic acid containing 0.1% w/v Texas Red. The Texas Red was used to locate the micropatterned strip of peptides before running the microfluidic assay. The microfluidic device was then gently removed, and the residual liquid was allowed to air dry.
2.5. Microspot Patterning Collagen Peptides
To pattern peptides in the absence of a PDMS channel, a 0.5 µL drop of peptides was pipetted in 10 mM acetic acid (250 µg/mL of each peptide) on each of the four surfaces described above and incubated for 2 h at room temperature in a petri dish with wet Kim-Wipes® to maintain a humid environment. The surface was then rinsed with 10 mM acetic acid and dried.
2.6. Whole Blood Microfluidic Assays
A 32 channel (w = 300 µm, h = 50 µm) microfluidic device (Figure S1A
) was laid perpendicular to the direction of the microfluidic patterned collagen peptide strip (Figure S2A
). In a second set of experiments, a four channel (w = 500 µm, h = 50 µm) microfluidic device (Figure S1B
) was laid on top, over the microspot patterned collagen peptides (Figure S2B
). Channels were blocked with 2% BSA in PBS for 45 min. Blood was added to reservoirs and perfused through the device at a flow rate to achieve wall shear rates of 300 s−1
and 1500 s−1
on the bottom wall of the channel for 5 min using a syringe pump (Harvard PhD Ultra, Harvard Apparatus, Holliston, MA, USA). Brightfield and epifluorescent images of platelet accumulation were recorded (40X, NA 0.6, Olympus IX83, Hamamatsu Orca Flash 4.0, Hammamatsu Photonics, Hamamatsu City, Japan). PBS with 10 U/mL heparin was perfused for 2 min to rinse the channels. Platelet aggregates were then fixed by perfusing 2% glutaraldehyde in PBS through the channels for 5 min. The channels were rinsed once more with PBS followed by brightfield and epifluorescent imaging of the final platelet aggregates.
2.7. Image Analysis
Custom Python scripts were used to calculate the mean fluorescence intensity (MFI) of each image. Background MFI was defined as the MFI of the first image of the assay where blood had just begun to perfuse through the microfluidic channel, but before significant platelet adhesion. This value was subtracted from each image to correct for background fluorescence. Maximum fluorescence of the assay was used to quantify overall platelet buildup for a given assay.
2.8. Atomic Force Microscopy
The morphology of collagen peptides on silicon and PDMS with or without FOTS treatment was measured, as described above, using atomic force microscopy (AFM) (Asylum, MFP 3D, Asylum Research, Santa Barbara, CA, USA). Collagen peptide solutions were incubated on the surfaces using the microspot patterning, as described above, or in 4 mm diameter PDMS wells. A solution containing 250 µg/mL of each peptide (CRP, GFOGER, and VWF-III) in 10 mM acetic was incubated in these wells for 2 h at room temperature in a humid environment. The wells were rinsed in triplicate using DI water, the PDMS device was removed, and the substrate was dried using an air brush. AFM images of the spots were taken using alternating current (AC) Tapping mode in air. The root mean square roughness (RMS) was calculated for each surface using Gwyddion [23
2.9. Contact Angle Measurement
The contact angle of water on different substrates was measured using an in-house interfacial tensiometer constructed by Aman et al. [24
]. Briefly, the setup consisted of a cell illuminated with a fiber optic lamp in which a camera and substrate holder were both placed on an optical tubular bench. A 20 μL droplet of DI water was placed on top of each substrate and allowed to settle for 5 seconds. The camera then took pictures of the droplets, and an image processing script was used to measure the contact angle of the water droplets on top of the substrate.
2.10. Collagen peptide depletion in PDMS
A PDMS microfluidic device (l = 40 mm, w = 17 mm, h = 50 µm, volume ≈ 35 µL) was laid on top of a PDMS-coated glass slide and filled with a solution of CRP, GFOGER, and VWF-III at a concentration of 250 µg/mL. Peptides were either patterned together (total peptide concentration = 750 µg/mL) on PDMS or FOTS-functionalized PDMS, or they were patterned separately (total peptide concentration = 250 µg/mL) on PDMS. The solution was incubated in the device for 2 h, after which it was recovered from the microfluidic chamber. The concentration of peptides was measured before and after incubation using a peptide-specific fluorometric assay (Quantitative Fluorometric Peptide Assay, Thermo Fischer, Denver, CO, USA).
2.11. Visualization of Absorption of Fluorescent Collagen Peptide into PDMS
A PDMS microfluidic device (l = 55 mm, w = 200 µm, h = 50 µm) was laid on top of a glass coverslip. The device and coverslip were either both clean, untreated surfaces or were both treated with FOTS. The microfluidic channel was filled with FAM-labeled CRP-XL at a concentration of 750 µg/mL in 10 mM acetic acid. After 2 h, the channel was rinsed with ten channel volumes of 10 mM acetic acid, removed from the coverslip it was patterned on, and was then moved to a clean coverslip for imaging. Images of the front of the peptides along the wall of the microfluidic channel were taken at z-sections of the channel (Olympus IX-83 equipped with a disc spinning unit, 60X, NA 1.35, Olympus Life Science, Waltham, MA, USA). Maximum intensity projections, line profiles of fluorescence intensity moving into the PDMS bulk, and mean fluorescence values of the projections were all computed using Fiji image processing software [25
]. Mean fluorescence intensity was calculated by summing the mean fluorescence intensity of each z-slice for each condition and device, and normalizing by the highest average value for any condition.
In this study, we found that previously developed collagen peptides that support VWF capture and platelet adhesion and aggregation do not adsorb to PDMS in a manner that supports platelet adhesion under flow. Rather, our data suggest that PDMS absorbs collagen peptides. Blocking the microfluidic patterning device with Pluronics F-127 or BSA did not show an increase in platelet adhesion in our flow assays, indicating that blocking the surface with common surfactants used in microfluidics does not effectively disrupt the absorption. However, the absorption of collagen peptides to a PDMS patterning device can be blocked by functionalizing with FOTS. Alternatively, FOTS functionalization of the assay surface (glass or PDMS) provides adsorption that is faster than absorption by the PDMS patterning device. This finding has ramifications, both for using PDMS for micropatterning collagen peptides and for using it as a material for microfluidic flow assays.
The absorption of collagen peptides by PDMS is supported by several lines of evidence. First, platelets do not adhere to PDMS surfaces incubated at a concentration of collagen peptides that support adhesion to glass. Second, there was no observable adsorption of peptides on PDMS at the resolution provided by AFM. Third, incubation of collagen peptide solutions in a monolithic PDMS channel results in significant depletion. Finally, significant penetration of a fluorescently labeled collage peptide was observed in untreated PDMS.
Untreated PDMS absorbs small, hydrophobic molecules, such as some fluorophores and sex hormones [28
]. Molecules with a logarithmic octanol-water partition coefficient (log P) > 2.62, such as rhodamine 6G and diazepam, are strongly absorbed into PDMS [30
]. Similarly, the peptide angiotensin II (~1 kDa) is readily absorbed into native PDMS, which can be attenuated by OEGylation [31
]. We are unaware of any prior work showing the absorption into untreated PDMS of triple helix peptides in the molecular weight range (~10 kDa) of those used in this study. It is unknown what properties of peptides might influence their absorption into PDMS, although our data shows that the three peptides used here absorb to different degrees. Whether these observations extends to other peptides, and what the molecular weight cut off is for absorption, requires further research.
The absorption of collagen peptides in PDMS was blocked when the PDMS was functionalized with a fluorinated alkyl silane. This conclusion is based on observations that incubation of collagen peptides on FOTS-PDMS supported platelet adhesion under flow, showed significant topological changes suggesting adsorption, demonstrated negligible depletion of collagen peptide solutions, and showed negligible absorption into the channel wall as visualized with a fluorescently labeled peptide. Whether FOTS blocks absorption by steric hindrance, by promoting hydrophobic interactions, or by some other mechanism requires further investigation. Nevertheless, in practice this simple modification allows PDMS to be used for either micropatterning collagen peptides to glass slides or as a substrate for adsorption itself.
The rate of adsorption of collagen peptides appears to be different for hydrophilic and hydrophobic surfaces. This inference is based on differences observed between untreated and FOTS-functionalized glass and silicon. These findings are supported by molecular dynamics simulations showing that adsorption of a collagen triple helix was more favorable for a hydrophobic surface compared to a hydrophilic surface due to stabilization of the triple helix structure [32
]. For untreated glass or silicon, if PDMS was used for patterning collagen peptides, there was little adsorption observed by AFM and negligible amounts of platelet adhesion under flow. For FOTS-glass or FOTS-silicon, significant changes to the surface topology were observe by AFM, and platelet adhesion was supported under flow. These observations suggest that the rate of adsorption to the FOTS-modified surfaces was faster than absorption by PDMS, while the opposite was found for untreated surfaces.