acs.biomac.8b00196?af=R

Apr 18, 2018 - The formation of a biomolecular corona around engineered particles determines, in large part, their biological behavior in vitro and in...
0 downloads 0 Views 13MB Size
Article Cite This: Biomacromolecules XXXX, XXX, XXX−XXX

pubs.acs.org/Biomac

Microfluidic Examination of the “Hard” Biomolecular Corona Formed on Engineered Particles in Different Biological Milieu Alessia C. G. Weiss,† Kristian Kempe,‡ Stephan Förster,§ and Frank Caruso*,† †

ARC Centre of Excellence in Convergent Bio-Nano Science and Technology, and the Department of Chemical Engineering, The University of Melbourne, Parkville, Victoria 3010, Australia ‡ ARC Centre of Excellence in Convergent Bio-Nano Science and Technology, and Monash Institute of Pharmaceutical Sciences, Monash University, Parkville, Victoria 3052, Australia § Physical Chemistry I, University of Bayreuth, Universitätsstrasse 30, 95447 Bayreuth, Germany S Supporting Information *

ABSTRACT: The formation of a biomolecular corona around engineered particles determines, in large part, their biological behavior in vitro and in vivo. To gain a fundamental understanding of how particle design and the biological milieu influence the formation of the “hard” biomolecular corona, we conduct a series of in vitro studies using microfluidics. This setup allows the generation of a dynamic incubation environment with precise control over the applied flow rate, stream orientation, and channel dimensions, thus allowing accurate control of the fluid flow and the shear applied to the proteins and particles. We used mesoporous silica particles, poly(2-methacryloyloxyethylphosphorylcholine) (PMPC)-coated silica hybrid particles, and PMPC replica particles (obtained by removal of the silica particle templates), representing high-, intermediate-, and low-fouling particle systems, respectively. The protein source used in the experiments was either human serum or human full blood. The effects of flow, particle surface properties, incubation medium, and incubation time on the formation of the biomolecular corona formation are examined. Our data show that protein adhesion on particles is enhanced after incubation in human blood compared to human serum and that dynamic incubation leads to a more complex corona. By varying the incubation time from 2 s to 15 min, we demonstrate that the “hard” biomolecular corona is kinetically subdivided into two phases comprising a tightly bound layer of proteins interacting directly with the particle surface and a loosely associated protein layer. Understanding the influence of particle design parameters and biological factors on the corona composition, as well as its dynamic assembly, may facilitate more accurate prediction of corona formation and therefore assist in the design of advanced drug delivery vehicles.



INTRODUCTION

Particle surface chemistry has a major impact on the composition of the biomolecular corona. Materials that reduce the ability of proteins from adsorbing are classified as low fouling. Poly(ethylene glycol) (PEG) or zwitterionic polymers are commonly used as surface coatings for particles to improve their circulation half-life.10,11 In such systems, a tightly bound water layer, which is formed around the polymer chains, acts as a physical and energetic barrier for protein adsorption.12 Generally, particles with high-fouling surface coatings show significantly greater plasma protein adhesion when compared with low-fouling systems.13−16 However, drawing absolute, quantitative conclusions about the correlation between the type of adsorbed proteins and surface chemistry is challenging.13,17 The biological milieu, including protein concentration,18,19 temperature,20 and incubation medium,8,21−23 is another factor that affects the composition of the biomolecular corona.

One of the main challenges in drug delivery is to overcome particle recognition by macrophages of the immune system, also known as opsonization.1,2 Once delivery systems enter a biological environment, plasma proteins and other biomolecules adsorb onto their surface and form a dynamic biomolecular corona.3 Protein adhesion can trigger biological defense mechanisms, which may lead to reduced circulation half-lives and premature clearance of the particles from the bloodstream.4 The biomolecular corona is generally described as a two-component system comprising an inner tightly bound layer, the so-called “hard” biomolecular corona, and an outer layer of loosely associated proteins, often referred to as the “soft” biomolecular corona.5,6 Most literature studies examine and focus on the “hard” corona, as those proteins are firmly bound to the particle surface and therefore do not detach upon extensive washing.3,4,7 From these studies, a large number of parameters that strongly influence the composition of the biomolecular corona are known, ranging from the physiochemical properties of nanoparticles to the biological milieu.3,7−9 © XXXX American Chemical Society

Received: February 5, 2018 Revised: April 6, 2018

A

DOI: 10.1021/acs.biomac.8b00196 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

tion.31,44,45 Specifically, our findings show that the total amount of adsorbed proteins increases as the fouling degree of the surface increases. Furthermore, overall, more proteins (in terms of type and amount) adsorb from human blood than from human serum. A dynamic incubation environment leads to a more complex corona in terms of absolute mass, which is not observed to such an extent in a static system. Variation of the flow rate confirms the increasing amounts of adsorbed proteins observed under higher flow rates. Studying the influence of incubation time leads to comprehensive data about the kinetics of protein adsorption. We show that a “hard” biomolecular corona is formed as early as 2 s upon particle incubation in a biological medium. Furthermore, “hard” corona formation can be kinetically divided into two different phases: the adsorption of proteins on the particle surface and the adsorption of proteins on proteins that have been previously adsorbed. The kinetics of the first phase are critically dependent on the incubation medium, whereas the kinetics of the second phase are similar in the two incubation media studied.

An aspect that has been little studied is the influence of the dynamic nature of physiological fluids. A dynamic environment creates shear stress and provides a continuous source of new proteins, resulting in a more complex corona.16,24−27 Studies using peristaltic pumps have shown that coronas formed under dynamic flow are more complex and mostly contain low molecular weight proteins.28 Similar results have been obtained from dynamic in vivo studies with subsequent recovery of the particles16,29−32 and in vitro blood flow assays.33 Another crucial factor that affects the corona composition is incubation time. Qualitative studies have shown that based on protein binding rates, an initial, short-lived “soft” corona is formed on short time scales (seconds to minutes) that subsequently develops into a long-lived “hard” corona (over periods of hours).34,35 Quantitatively, as postulated by Vroman et al., proteins present at higher concentrations are first bound to the particle surface and are then gradually replaced with proteins present at lower concentrations that have a higher affinity for the surface.36 Using mass spectrometry, Tenzer and co-workers showed that a “hard” corona is detected after 30 s and changes over time in terms of the amount of bound proteins, but not in composition.37 A minimum incubation and purification time of ∼30 s is typically required using current incubation and purification techniques, thereby limiting our understanding of the kinetics of the “hard” biomolecular corona. Recent studies have highlighted the significance of conducting in vitro studies, wherein the dynamic nature of blood flow is considered,4,26,38 and understanding the correlation between particle design and biological milieu on the biomolecular corona composition.3,4,9,39,40 Generally, a higher quantity of proteins adsorbs on hydrophobic materials compared with hydrophilic materials. Furthermore, protein diversity is considerably broader when the protein corona is formed under fluidic flow. Owing to the complexity of corona formation, a complete, mechanistic picture is yet to be established. Comprehensive studies in which different parameters are varied independently are still needed to provide further insights into the complex process of corona formation. Herein, we apply microfluidics to study protein adsorption onto nanoparticles under flow. As particle systems, mesoporous silica particles were used as a high-fouling material, zwitterionic poly(2-methacryloyloxyethylphosphorylcholine) (PMPC) replica particles (RPs) as a low-fouling material, and silica−PMPC hybrid particles (HPs) as an intermediate fouling material. Defined chip dimensions and precise syringe pumps offer accurate control over the applied flow rate, stream orientation, and channel dimensions.41,42 This in turn enables control of the fluid flow and the shear applied to proteins and particles.43 Furthermore, the use of microfluidics offers the possibility of reducing incubation times to 2 s when high flow rates are applied. Compared with the current minimum incubation and purification time of ∼30 s typically required,37 microfluidics allows examination of protein adsorption on a shorter time scale. The present microfluidic setup was used to conduct a series of comprehensive studies that involved comparing different particle surface chemistries (high fouling vs low fouling), incubation media (human serum vs human blood), incubation conditions (static vs dynamic), and incubation times. By screening the different parameters, we find that the biomolecular corona formed on low-fouling particles after incubation in human blood under flow contains proteins such as different opsonins and clusterin, both of which are often considered to be relevant for improved particle biodistribu-



EXPERIMENTAL SECTION

Materials. All chemicals were used as received without further purification, except for copper(I) bromide (CuBr), which was purified by washing sequentially with glacial acetic acid, absolute ethanol (EtOH), and diethyl ether, and dried under vacuum. Furthermore, all liquid monomers were passed through an aluminum oxide column prior to polymerization to remove the inhibitor. High-purity water (Milli-Q water) with a resistivity of >18.2 MΩ cm was obtained from an inline Millipore RiOs/Origin water purification system (Millipore Corporation, Burlington, MA). Silica particle functionalization was carried out using (3-aminopropyl)triethoxysilane (APTES, 98%), ammonia (NH3, 28−30%), pyridine (anhydrous, 99.8%), tetrahydrofuran (THF) (anhydrous, 99.9%), and α-bromoisobutyryl bromide (98%), which were all purchased from Aldrich. For the surfaceinitiated atom transfer radical polymerization (SI-ATRP), ethylene glycol dimethacrylate (EGDMA, 95%), 2-hydroxyethyl methacrylate (97%), N,N,N′,N″,N″-pentamethyldiethylenetriamine (PMDETA, 99%), 2-methacryloyloxyethylphosphorylcholine (MPC), CuBr (98%), and nitric acid (70%), all from Aldrich, were used. Template removal was performed using hydrofluoric acid (HF) (48%) and ammonium fluoride (NH4F, 98%), both obtained from Aldrich. Piranha cleaning was performed using hydrogen peroxide (H2O2, 30%) and sulfuric acid (H2SO4, 99.9%), both obtained from Aldrich (St. Louis, MO). 1H nuclear magnetic resonance (NMR) spectroscopy measurements were performed using D2O (99.9%). Fluorescence labeling was carried out using N-hydroxysuccinimide (NHS)-activated AF488, purchased from Thermo Fisher (Victoria, Australia), and dimethyl sulfoxide (DMSO) (anhydrous, >99%) was obtained from Aldrich. Device fabrication was carried out using Sylgard184 silicone elastomer and the corresponding curing agent from Dow Corning (Midland, MI). Silicon wafers (diameter 3 in.) were obtained from SiMat Silicon Materials (Germany). Developer mr-DEV 600 and photoresists Nano SU-8 50/SU-8 100 were purchased from MicroChem Corporation (Westborough, MA). Human serum Mixed Pool (nonsterile serum, “off the clot”) was purchased from TCS Biosciences Ltd. (Buckinghamshire, UK). Biomolecular corona characterization was carried out using Dulbecco’s phosphate buffered saline (DPBS), ammonium bicarbonate (NH4 HCO3, 99.0%), acetonitrile (ACN) (99.5%), DL -dithiothreitol (DTT, 99.0%), iodoacetamide (IAA), trypsin from porcine pancreas (proteomics grade), trifluoroacetic acid (TFA, 99.0%), and formic acid (FA, 95.0%), all purchased from Aldrich. BioRad 4−20% Mini-PROTEAN TGX Stain-Free protein Bis-Tris gels and Tris/glycine/sodium dodecyl sulfate (SDS) running buffer were obtained from BioRad (Hercules, CA). LDS sample buffer, NuPAGE sample reducing agent, SeeBlue prestained protein standard, and SimplyBlue safe stain were purchased from Thermo Fisher. B

DOI: 10.1021/acs.biomac.8b00196 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules Mesoporous Silica Particle Modification, SI-ATRP, and Template Removal. Mesoporous silica (MS) particles were synthesized according to a recently published method.46 Modification of the MS particles with ATRP initiator groups was performed following a modified literature method.47 The volume ratio of EtOH/ ammonia/APTES was fixed at 0.5:1:20. Prior to ATRP initiator attachment, the MS particles were functionalized with amine groups. The MS particles (0.7 mg) were first dispersed in EtOH (20 μL). Ammonia (1 μL) and APTES (0.5 μL) were then added to the suspension and stirred overnight at 20 °C. The APTES-modified silica particles were washed via repeated centrifugation (4000g, 8 min) and resuspension steps first in EtOH (twice) and then in Milli-Q water (twice). The washed particles were then reacted with α-bromoisobutyryl bromide for initiator attachment. A volume ratio of SiO2−NH2/ THF/α-bromoisobutyryl bromide of 1:15:3 was used. The particles were dispersed in anhydrous pyridine (100 μL) and THF (anhydrous, 15 μL). After the addition of α-bromoisobutyryl bromide (3 μL), the suspension was stirred overnight at 20 °C. Purification was performed by washing with EtOH (4000g, 8 min, twice) and Milli-Q water (4000g, 8 min, twice). The obtained ATRP-functionalized mesoporous silica particles (denoted as MS@ATRP-Ps) were stored in EtOH in the fridge (4 °C). The HPs were prepared via SI-ATRP under strict oxygen exclusion. A molar ratio of the reactants of MS@ATRP-Ps/monomers/CuBr/ PMDETA of 1:1000:2:2 and a monomer ratio of MPC/hydroxyethyl methacrylate (HEMA)/ethylene glycol dimethacrylate (EGDMA) of 87:3:10 were used. In a Schlenk flask, the monomer mixture (237 mg of 0.8 mmol MPC, 3.9 mg of 0.03 mmol HEMA, 22 mg of 0.09 mmol EGDMA) and MS@ATRP-Ps (10 μL of 100 mg mL−1 stock solution) were mixed in EtOH (1.5 mL). Before the reaction mixture was degassed by three freeze−pump−thaw cycles, PMDETA (0.31 mg, 0.0018 mmol) was injected, and the reaction mixture was sonicated to facilitate infiltration of the monomers into the pores of the particles. CuBr (0.26 mg, 0.0018 mmol) was added, and polymerization was started by heating the reaction mixture up to 50 °C. After 20 h, the reaction was quenched by cooling the mixture to 20 °C and exposing it to air. Purification was performed by repeated centrifugation (6000g, 5 min) and resuspension steps using nitric acid (1 M, once), 10% pyridine/EtOH (once), and EtOH (thrice). To obtain the RPs, the silica cores were removed using NH4Fbuffered (13.3 M, pH 6) HF (5 M); the buffered solution (200 μL) was added to the particle dispersion, and the resulting mixture was left to incubate for 5 min. Caution! HF is highly toxic. Extreme care should be taken when handling HF solution, and only small quantities should be prepared. Purification was performed by repeated centrifugation (10 min, 8000g) and washing steps using Milli-Q water (once) and EtOH (twice). The particles were stored in EtOH in the fridge (∼4 °C). Particle Characterization. Thermogravimetric analysis was conducted on a PerkinElmer Diamond thermogravimetric/differential thermal analysis instrument. The temperature range was set to 50−700 °C, and a heating rate of 10 °C min−1 was used. 1H NMR spectra were recorded using an 800 MHz Bruker Advance-II spectrometer at 25 °C for 20 h. The spin−lattice relaxation times were determined at 24 °C via a conventional inversion recovery pulse sequence under deuterium lock mode. Chemical shifts were reported as parts per million (ppm) downfield from the signal originating from reference tetramethylsilane (δ: 0 ppm). Each spectrum was standardized using the solvent D2O (4.81 ppm). Data evaluation was carried out using the program Spin Works, 2002. Fluorescence microscopy images were taken with an Olympus IX71 inverted fluorescence microscope (10× objective, Olympus) equipped with a DIC slider (U-DICT, Olympus), the corresponding filter sets, and a 100× oil immersion objective (Olympus UPFL20/0.5NA, W.D. 1.6). ImageJ was used for image processing. Dynamic light scattering (DLS) was performed on a Zetasizer Nano ZS from Malvern Instruments Ltd., UK, at a wavelength of 632 nm with a scattering angle of 173° (noninvasive backscatter technology). Zeta-potential measurements of the particles were conducted on a Malvern Zetasizer Nano ZS. Transmission electron microscopy (TEM) experiments were performed using a FEI Tecnai TF20 instrument operating at 120 kV under liquid nitrogen

cooling. For analysis, particle dispersions prepared in Milli-Q water were dropped onto Formvar-coated copper grids (plasma-treated; low oxygen plasma, 30 s) that were then allowed to dry in air. All images were analyzed using ImageJ processing software. Scanning electron microscopy (SEM) characterization was performed using a Philips XL30 scanning electron microscope and a Zeiss Ultra plus fieldemission scanning electron microscope, both operating at 10 kV (high vacuum). SEM samples were prepared on piranha-cleaned (3:7 v/v 30% H2O2/98% H2SO4), plasma-treated (high oxygen plasma, 3 min), nitrogen-dried silicon wafers (Caution! Piranha solution is extremely corrosive and reacts violently with organic materials. It should be handled with great care.) These wafers were then attached onto a sample holder using double-sided adhesive, electrically conductive carbon tape. The wafers were then sputter-coated with gold. Image analysis was performed using ImageJ software. Atomic force microscopy (AFM) images of particles air-dried on piranha-cleaned, plasma-treated (high oxygen plasma, 3 min), nitrogen-dried silicon wafers were captured using a JPK NanoWizard2 Bio-AFM or a Dimension Icon from Bruker equipped with a V controller. Typical scans were recorded in intermittent contact mode with silicon cantilevers (k = 40 N m−1, f = 325 kHz) (MicroMesh, Bulgaria). The AFM instrument was operated on a vibration-isolated table, 512 × 512 pixels were recorded per image, and the scan rate was set to a value of less than 1.0 Hz so that the tip can track the surface profile. The z-limit of the piezo scanner was reduced to increase the resolution of the AFM. All images presented in this study were processed using the JPK image processing software or NanoScope Analysis v1.40r1 (for plain fit). Gwyddion 2.47 was used to adjust the color scale. Particle counting was carried out using flow cytometry (Apogee Flow), and data analysis and gating were undertaken using FlowJo_V10 software. Fabrication of Microfluidic Devices. The microfluidic chips were prepared according to a modified literature method.48 A crossshaped mixer geometry with channel dimensions of 50 μm × 50 μm (height × width) were used. The structure was designed using the AutoCAD 2013 (Autodesk) software and subsequently printed on a photomask foil using a soft photographic emulsion gel. To replicate this design onto a silica wafer, spin-coating cycles of a negative epoxybased photoresist (SU-8) were applied using a mask aligner (Süss MicroTec). An ∼100 μm thick layer was obtained. Following a soft baking (65 °C, 10 min) step, UV patterning was performed by placing the mask onto the wafer and exposing it to UV light. Nonexposed photoresist was removed using developer prior to a final hard baking (95 °C, 30 min) step. Soft lithography was performed by pouring a polydimethylsiloxane (PDMS) base/curing agent mixture (10:1 w/w) onto the silicon master. After degassing for 45 min in a desiccator and subsequent drying for 2 h at 75 °C, the PDMS replicate was removed from the master, and inlet ports for fluids were added. To obtain a 3D channel structure, two PDMS devices were sealed together after plasma activation (air plasma, 5 min) and dried overnight at 35 °C. Microfluidic Setup and Particle Incubation under Flow. Particle incubation under flow was carried out using PDMS microfluidic devices that were connected to gastight Hamilton syringes via Original-Perfusor lines (Type IV-Standard; B. Braun Melsungen AG, Germany) and medical grade polyethylene microtubings (0.38 mm inner diameter × 1.09 mm outer diameter; Scientific Commodities Inc., Lake Havasu City, AZ) and syringe pumps (Nemesy, CETONI GmbH). The incubation time for the static and dynamic measurements was kept constant for a given applied flow rate (Table 1) using a 1 m polyethylene tube for all experiments to collect the protein/particle dispersions. After incubation, the dispersions were immediately frozen using dry ice to stop biomolecular corona formation. For all experiments, 1 × 108 particles were used. The particle dispersion and the protein solution (human serum or human blood) were pumped through the microfluidic device. The flow rate was altered by varying the particle flow rate only (Table 1). To account for the constant protein flow rate, a newly introduced parameter, i.e., relative flow rate, is used. Thus, variations in the particle flow rate correspond to variations in the relative flow rate (i.e., 1200, 600, and 0, which in turn are referred to as “high”, “intermediate”, and “no” relative flow rates, respectively). Human C

DOI: 10.1021/acs.biomac.8b00196 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS-PAGE). The “hard” corona proteins were eluted from the particles by adding NuPAGE LDS sample buffer (10 μL) and NuPAGE sample reducing agent (3 μL) to each sample and heating at 70 °C for 10 min. After dilution of the samples to 100 μL with Milli-Q water, the particles were separated from the eluted proteins by centrifugation (12300g, 10 min). The supernatant was transferred to a LoBIND Eppendorf tube and loaded on the gel (30 μL per sample) to run for 50 min at 150 V. Each gel included one lane of a standard molecular weight ladder. Gel staining was carried out using Coomassie Blue. All incubation studies were repeated at least three times to ensure reproducibility. Data analysis was performed using ImageJ software. The gel images were cropped and presented with their respective standard molecular weight ladder. Uncropped gel images are presented in Figures S1−S4 of the Supporting Information. Mass Spectrometry. Sample preparation was carried out following a standard protocol49 with slight modifications. Bands of interest from Coomassie Blue-stained SDS-PAGE gels were excised and transferred to LoBIND Eppendorf tubes. For destaining, 200 μL of 100 mM NH4HCO3/methanol (1:1 v/v) was added to the gels and incubated for 5 min. This step was performed twice. The supernatant was removed after each step. Following this, 200 μL of 25 mM NH4HCO3/ACN (1:1 v/v) was added to the resulting gels. After incubation for 2 min, the supernatant was discarded, and the gel pieces were incubated in ACN (200 μL) for 30 s. The supernatant was subsequently discarded. For in-gel digestion of the proteins, 50 μL of 25 mM DTT (in 25 mM NH4HCO3) solution was added to the gel pieces and incubated for 20 min at 56 °C. After the gel pieces had cooled to 20 °C, the supernatant was removed, and 50 μL of 55 mM IAA (in 25 mM NH4HCO3) solution was added. Incubation was performed under complete darkness for 20 min. The gel pieces were subsequently washed (thrice) with Milli-Q water (400 μL) for 30 s. To dehydrate the gels, 200 μL of 25 mM NH4HCO3/50% ACN was added and incubated for 5 min at 20 °C. In the final step, ACN (200 μL) was added for 30 s, and the gels were dried using a vacuum centrifuge (∼55 °C, 30 min, 500g). To trypsinize the proteins, 10 μL of trypsin (12.5 ng μL−1 in NH4HCO3, pH 8) was added to the gels that were incubated for 10 min at 20 °C. After adding 10 μL of 25 mM NH4HCO3 solution to each sample, the gel pieces were incubated overnight at 37 °C. The supernatant was removed and transferred to

Table 1. Applied Flow Rates in the Side Channels (Protein Flow) and Main Channel (Particle Flow) and Incubation Times side channel 1 (μL h−1)

main channel (μL h−1)

side channel 2 (μL h−1)

relative flow ratea (μL h−1)

incubation time (s)

1600 1600 1600

400 1000 1600

1600 1600 1600

1200 (high) 600 (intermediate) 0 (no)

341 120 84

Relative flow rates refer to the velocity of the particles with reference to the flow rate of the protein solution. The relative flow rates applied are referred to as high, intermediate, and no. a

whole blood was collected in S-monovettes (1.8 mL 9NC, 3.2% citrate) from a healthy volunteer (single source of blood; however, blood was withdrawn on different days). To remove unbound and loosely bound proteins, the particles were washed by multiple centrifugation (12300g, 10 min) and resuspension steps (Milli-Q water, five times). Note that due to the high number of particles (1 × 108 particles) used for each experiment, two separate pellets, one containing the particles and a second containing the cells, were obtained for incubation in human blood. As the density of the particles is higher than that of the cells, the cell pellet concentrated on top of the particle pellet and thus could be removed. Protein separation was carried out under dry ice cooling to prevent potential reactivation of the incubation process. The obtained biomolecular corona-coated particles were directly used for proteomic analysis. Static Incubation. Particles were incubated in human serum (1 mL) or human blood (1 mL) for different times (341, 120, and 84 s) at 37 °C under constant shaking (Eppendorf Thermomixer Comfort, Germany). To keep the sample preparation identical to that used in the flow experiments, the samples were frozen after incubation and subsequently washed using the same protocol. The different incubation times studied341, 120, and 84 scorrespond to the incubation times used in the dynamic system at high, intermediate, and no relative flow rates examined (i.e., 1200, 600, and 0 μL h−1 (Table 1)), respectively.

Scheme 1. Schematic of the Synthesis of the Three Different Particle Systems Studieda

a

ATRP-functionalized mesoporous silica particles (MS@ATRP-Ps) were used as templates for SI-ATRP to obtain hybrid particles (HPs). Replica particles (RPs) were obtained after template removal. D

DOI: 10.1021/acs.biomac.8b00196 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules an Eppendorf tube, while the remaining gel pieces were incubated in 10 μL of ACN/0.5% TFA (1:1 v/v) solution, and the resulting mixture was sonicated for 20 min. The supernatants from the incubation step with ACN/TFA solution were combined with the supernatants obtained directly after trypsinization. The supernatants were dried in a vacuum centrifuge until completely dry (∼55 °C, 30 min, 500g). Following this, 0.1% FA (20 μL) was added to the dried sample and vortexed for 10 min. The samples were stored at −20 °C until use. Proteins were identified using nanoflow liquid chromatography− electrospray ionization tandem mass spectrometry (nLC-ESI-MSMS) on a Thermo LTQ XL instrument with an online 2D Easy-nLC II running at 400 nL min−1 using a self-packed 12 cm C18 column emitter (New Objective Inc., Woburn, MA). The peptides were eluted from the column over a 30 min gradient consisting of 5% ACN at 0 min to 35% ACN at 30 min. Peaks were selected for MSMS with a charge state of +2 or greater and a minimum threshold of 500 counts. The data were then processed using the Byonic search engine (Protein Metrics, San Carlos, CA) with the following settings: a precursor mass tolerance of 2 Da and a fragment mass tolerance of 0.5 Da; the C terminus of lysine and arginine as cleavage sites with up to two missed cleavages possible; and the possibility of oxidation and carbamylation to M, deamidation to N and Q, dioxidation to W, trioxidation to C, and methylation to E. Searches were made against the Uniprot Human database (accessed 8 November 2016) and successful matches recorded with a false discovery rate of 1%. Proteins were considered identified with three or more peptides with a confidence of >99%.



RESULTS AND DISCUSSION Particle Synthesis and Characterization. To investigate the influence of surface chemistry on protein adsorption, three nanoparticle systems, with different protein affinities (high to low affinity: MS@ATRP, HP, RP), were synthesized (Scheme 1). MS nanoparticles, representing a high-fouling material, were prepared according to a modified literature method.50,46 MS@ ATRP-Ps were obtained upon functionalization of the MS particles with ATRP initiator, of which the content was 0.19 mg per mg of particles, as calculated from thermogravimetric analysis (Figure S5). The HPs were synthesized via SI-ATRP by infiltrating monomers into the pores of the [email protected] Polymerization was conducted using the zwitterionic, thus low-fouling, monomer MPC.51 HEMA and a PEG-based crosslinker (poly(EGDMA) (PEGDMA) were additionally copolymerized. Controlled etching of the silica templates from the HPs resulted in the preparation of RPs. 1H NMR spectroscopy of the RPs in D2O showed the relevant signals of the main component PMPC (Figure S6). The presence of cross-linker PEGDMA was indirectly confirmed by the observation of stable particles after template removal; however, this could not be determined by NMR, as the main signal (3.6 ppm) overlapped with the PMPC signals. Furthermore, the presence of polyHEMA (PHEMA) was indirectly confirmed, as fluorescence labeling was carried out by reacting NHS-activated AF488 with the hydroxyl group of PHEMA (Figure S7). The PHEMA could not be determined by NMR spectroscopy, as the concentration (3 mol %) used during synthesis was below the detection threshold. TEM and SEM analyses revealed the spherical morphology and porous structure of the MS@ATRP-Ps (Figure 1a,b). Following infiltration of the monomers and subsequent polymerization to obtain the HPs, qualitatively the porous structure was still observed, however, to a lesser degree (Figure 1c,d). After template removal, polymeric RPs were obtained (Figure 1e,f). AFM measurements performed in tapping mode in air showed that the RPs collapsed after air drying (Figure S8). Their height decreased from 184 to 32 nm, which

Figure 1. Characterization of (a, b) MS@ATRP-Ps, (c, d) HPs and (e, f) RPs: (a, c, e) TEM and (b, d, f) SEM images.

confirmed successful removal of the silica template. The average sizes of the MS@ATRP-Ps, HPs, and RPs calculated from the electron microscopy and AFM analyses were 164 ± 30, 161 ± 20 nm, and 179 ± 25 nm, respectively (Table S1). DLS analysis revealed swelling of the RPs after template removal (∼30%) (Table S1 and Figure S9). Zeta-potential measurements showed that the MS@ATRP-Ps assessed in DPBS had a zeta potential of ca. 10 mV, indicating successful surface functionalization with amines and ATRP initiators. In comparison, the zeta potential of the bare silica particles was ca. −13 mV. The HPs and RPs exhibited a surface potential of ∼0 mV. A close-to-neutral zeta potential is characteristic of zwitterionic materials. Microfluidic Examination of the “Hard” Biomolecular Corona. To study the corona formation under controlled flow conditions, a microfluidic setup that simulates cardiovascular flow conditions was used.52 A 3D PDMS chip was fabricated using an established soft lithography technique.48 To avoid interactions of the particles with the PDMS channel walls, a cross-shaped mixer geometry was used (Figure 2). Particles were hydrodynamically focused in the central stream, surrounded by the protein solution (Figure 2d). After the particles traveled through the microfluidic device, the protein− particle dispersion was collected and immediately frozen to stop the adsorption process. To subsequently study the “hard” corona, the particles were isolated from unbound or loosely bound proteins by multiple centrifugation and washing steps E

DOI: 10.1021/acs.biomac.8b00196 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

proteins, the “hard” corona proteins were analyzed by 1D gel electrophoresis to determine the effect of particle surface chemistry, incubation medium, and incubation environment (static vs flow). Figure 3 depicts Coomassie blue-stained SDSPAGE gels obtained from proteins present in the “hard” corona of the MS@ATRP-Ps, HPs, and RPs after incubation in human serum or human blood under flow and static conditions. For all three particle types and media studied, complex band patterns were observed, indicating that the “hard” corona is composed of many different proteins. Moreover, the bands and their respective intensities differed for each system with substantial differences observed between the studies conducted under dynamic (Figure 3a, upper row) and static conditions (Figure 3b, bottom row). The studies using MS@ATRP generated more bands with a more pronounced intensity when compared with the studies using polymer-containing particles (HPs and RPs). In general, the results demonstrate that the nature of the particles, the incubation medium, and incubation conditions considerably influence the composition of the biomolecular corona. As the relative amount of protein adsorbed is related to the relative protein band intensities from SDS-PAGE, semiquantitative data were obtained.19 Figure 4 shows the overall relative protein band intensity for each of the conducted studies. The surface chemistry of the particles strongly influenced the composition of the “hard” biomolecular corona. MS@ATRP-Ps adsorbed more proteins, regardless of the incubation medium and conditions used, when compared with the polymer-containing particles (HPs and RPs). A similar behavior was observed by other groups who compared protein adsorption onto PEGylated and non-PEGylated particles.13,14,55

Figure 2. (a) Microfluidic setup for biomolecular corona formation under controlled flow conditions using a 3D PDMS chip with a crossshaped mixer geometry. (b) Human serum or (c) human blood is pumped through the side channels while the particle dispersion is pumped through the main channel. (d) Confocal laser scanning microscopy image of the fluorescently labeled silica particles surrounded by a water-based Rhodamine B solution as a model system to illustrate the flow of the particle stream.

and resuspended in a protein-free medium according to a standard protocol.53 The particle count was determined by flow cytometry and kept constant at 1 × 108 for all subsequent experiments (Figure S10 and Table S2). To determine the influence of particle flow rate, the parameter relative flow rate was used, as this accounts for the constant protein flow rate in the side channels (Table 1). The flow rate of the protein solution (1600 μL h−1) was adjusted to be in the range of average blood velocities in the main blood vessels (arteries, arterioles, capillaries, venules, and veins).54 To compare these flow data to the static incubation data, the incubation times were kept constant within one applied flow rate. All three particle types (MS@ATRP-Ps, HPs, and RPs) were flowed through the microfluidic channel using three different particle flow rates and two different incubation media (human serum and human blood). After removal of loosely bound

Figure 3. “Hard” corona characterization: cropped images of SDS-PAGE gels of the MS@ATRP-Ps, HPs, and RPs incubated in either human serum or human blood under (a) different relative flow rates (upper row: *high relative flow rate (incubation time 341 s); **intermediate relative flow rate (incubation time 120 s); ***no relative flow rate (incubation time 84 s)) and (b) static conditions (bottom row: incubation times +341 s; ++120 s; and +++84 s). A set of reference bands associated with particular molecular weights are displayed on the left of each data set. Uncropped gel images are presented in Figures S1−S4. F

DOI: 10.1021/acs.biomac.8b00196 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 4. Relative amount of adsorbed proteins identified in the “hard” biomolecular corona formed on the MS@ATRP-Ps, HPs, and RPs upon incubation of the particles in human serum and human blood under flow and static conditions. “High”, “intermediate” (Interm.), and “no” correspond to applied relative flow rates of 1200 μL h−1 (341 s), 600 μL h−1 (120 s), and 0 μL h−1 (84 s) in the dynamic incubation studies and incubations times of 341, 120, and 84 s in the static incubation studies.

adsorbed under flow (Table S5). Specifically, in human serum, the overall protein concentration was lower by 28 ± 8% for the MS@ATRP-Ps, 30 ± 6% for the HPs, and 8 ± 1% for the RPs for incubation in a static environment (Table S5). In human blood, decreases of 28 ± 12% for the MS@ATRP-Ps, 43 ± 5% for the HPs, and 50 ± 4% for the RPs were measured as the incubation conditions changed from flow to static (Table S5). The incubation times were kept constant at 341 s for the high relative flow rate, at 120 s for the intermediate flow rate, and at 84 s for the no applied relative flow rate throughout these experiments. Proteins adsorbed in a static environment undergo three successive steps: (1) diffusion to the surface, (2) adsorption onto the surface, and (3) relaxation to reach an equilibrium state.43 However, under flow, new adsorption features occur owing to competition between the actual adsorption and translocation past the surface.43 Enhanced particle diffusion owing to a microscale flow-field generated by movement of the red blood cells56 and margination of particles in blood42,57 influence the protein concentration. Enhanced particle diffusion can explain both the increased protein concentration under flow and after incubation in human blood. The microfluidic setup with attached syringe pumps allowed us to further compare the effect of different applied relative flow rates on the biomolecular corona (Table S6). Under both serum and human blood conditions, lowering the relative flow rate from 1200 μL h−1 (high relative flow rate) to 600 μL h−1 (intermediate relative flow rate) led to a reduction of 4 ± 6% in the amount of adsorbed proteins (average value of adsorption for MS@ATRP-Ps, HPs, and RPs). Applying a lower relative flow rate of 0 μL h−1 (no relative flow rate) led to a further decrease of 12 ± 8%. Under a higher relative flow rate, more proteins flow past the particles per time unit, resulting in more adsorbed proteins on the surface of the particles. The influence of different incubation times can be excluded, as no similar trends were observed in the static system. Semiquantitative analyses of the composition of the obtained biomolecular corona were conducted using SDS-PAGE, and the results are shown in Figure 5. The pie charts show protein

In human serum, the HPs adsorbed on average 36 ± 6% less protein when compared with the MS@ATRP-Ps, and the adsorption of the RPs was lower by 15 ± 13% when compared with the adsorption on the HPs (Table S3). In human blood, the HPs and RPs respectively adsorbed 14 ± 12% and 17 ± 11% less protein when compared with high-fouling MS@ ATRP-Ps. The fact that HPs and RPs adsorbed less protein underscores the low-fouling effect of the zwitterionic PMPC polymer. The incubation medium also plays an important role on corona composition. For all three particle systems, regardless of the incubation conditions (under flow with different relative flow rates or static incubation), more proteins adsorbed from human blood than from human serum (Figure 4 and Table S4). This observation is in accordance with recent literature where lipid bilayer formulations were either intravenously injected into an animal model or incubated in vitro using plasma.29 It is important to note that particles interact with blood cells, thus reducing the total number of particles. Therefore, the number of recovered particles might be higher after incubation in human serum than in human blood. We assume that because of this reduced particle number following blood incubation, the concentration of adsorbed proteins is even higher than that detected by SDS-PAGE. MS@ATRP-Ps under flow adsorbed on average 16 ± 3% more protein from human blood than from human serum (Table S4). Under static conditions, the average difference was 15 ± 11% (Table S4). For the HPs, the protein uptake from human blood increased by 43 ± 1% in a dynamic environment and by 30 ± 13% under static conditions (Table S4). RPs showed a difference of 53 ± 3% (dynamic) and 12 ± 13% (static) from blood to serum (Table S4). Particles in human whole blood experience a different environment when compared with particles in human serum, owing to a different protein content and concentration as well as possible interactions of the particles with cells.23 Regardless of the particle system and incubation medium studied, adsorption conducted under static conditions led to fewer adsorbed proteins when compared with proteins G

DOI: 10.1021/acs.biomac.8b00196 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 5. Protein distributions of the corona, based on their molecular weights (MW), for the MS@ATRP-Ps, HPs, and RPs. The coronas were formed by incubation in human serum and human blood under flow and static conditions. The data represent mean values from all the different applied flow rates (high (1200 μL h−1, 341 s), intermediate (600 μL h−1, 120 s), and no (0 μL h−1, 84 s) relative flow rate) and incubation times studied (341, 120, and 84 s).

pronounced in human blood; similar protein adsorption amounts and protein distributions were observed. The composition of the corona was strongly influenced by the incubation medium (Table S7). In human serum, in a dynamic environment, proteins of intermediate molecular weight (99−31 kDa) mainly adsorbed on all three particle systems (70% for MS@ATRP-Ps, 52% for HPs, and 67% for RPs). The coronas of the polymer-containing particles were additionally composed of high molecular weight proteins (300−100 kDa; 48% for HPs and 33% for RPs). However, the adsorption of low molecular weight proteins (30−4 kDa)

distributions in accordance with their molecular weights for the MS@ATRP-Ps, HPs, and RPs incubated under flow and static conditions. The data represent mean values from all different applied flow rates (high, intermediate, and no relative flow rate) and incubation times (341, 120, and 84 s) studied. Surface chemistry also influenced the semiquantitative composition of the biomolecular corona. From human serum, in general, the MS@ATRP-Ps adsorbed a wider range of proteins when compared with the HPs and RPs. This effect was even more pronounced for adsorption conducted in a dynamic environment. In contrast, the influence of surface chemistry was less H

DOI: 10.1021/acs.biomac.8b00196 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 6. Protein distributions of the corona obtained by mass spectrometry, based on their molecular weights (MWs), for the MS@ATRP-Ps, HP, and RPs. The coronas were formed by incubation in human serum and human blood under flow (no relative flow rate, 0 μL h−1, 84 s) and static conditions (84 s).

The incubation conditions influenced the protein distribution in the respective coronas. A dynamic environment generally led to 27 ± 7% more types of proteins adsorbed when compared with a static environment (Table S8). Flow conditions exert hydrodynamic and shear forces on the particles and proteins that may induce conformational changes in the protein structure.58 New exposed surface-binding functional groups or binding pockets may attract different types of proteins. Furthermore, under static conditions, the concentration of proteins with a higher affinity is reduced over time, which results in proteins adsorbing with a lower affinity.58 Furthermore, regardless of the particle surface chemistry, predominantly high molecular weight proteins adsorbed from

was not observed in these two systems. Under static conditions, an even distribution of proteins was detected (∼33% per molecular weight group). Adsorption from human blood under flow resulted primarily in average molecular weight proteins adhered to the surface, with the second-most abundant species (in all particle systems) being low molecular weight proteins. The percentage of high molecular weight proteins decreased drastically (