Identification of the physical-chemical properties that modulate the nanoparticles aggregation in blood

Abstract

Inorganic materials are receiving significant interest in medicine given their usefulness for therapeutic applications such as targeted drug delivery, carriers of active pharmaceutical and medical imaging.
However, the poor knowledge of the side effects related to their use is an obstacle to their clinical translation. For the molecular drug development, safe-by-design has become as a novel pharmaceutical strategy that allows a reduction of the costs and an acceleration of the translation of research to market. In the case of materials, the application of such approaches is hampered by a poor knowledge of how the physical and chemical properties of the material trigger biological response.
Hemocompatibility is a crucial factor for those materials that are intended for medical applications. In particular, the formation of agglomerates is a serious side effect that may induce occlusion of blood vessels and thrombotic events. Additionally, nanoparticles can interfere with the coagulation cascades where they have been reported to induce both pro- and anti-coagulant properties where their properties like size, shape and surface charge have been see to be critical parameters.
Here, we developed two sets of tailored carbon and silica nano/submicron-particles with three different sizes (100-500 nm) with the purpose of investigating the role of surface curvature and chemistry on platelet aggregation, activation and adhesion.
We show that that large carbon nanoparticles, but not small carbon nanoparticles or silica nanoparticles, have a strong tendency to form aggregates both in plasma and blood, as a consequence of the formation of a protein corona and not of platelets activation. Substantial differences were found in the composition of the protein corona depending upon the chemical nature of the nanoparticles, while the surface curvature plays a minor role. On the other hand, coagulation proteins were abundant in the corona of both silica and carbon nanoparticles.
The results presented herein suggest that vessel occlusion and formation of thrombi in vivo may occur through independent mode of action (MoA), differently affected by the physico-chemical properties of the materials.

Full text

License and Terms: This document is copyright 2019 the Author(s); licensee Beilstein-Institut.
This is an open access publication under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0). Please note that the reuse,
redistribution and reproduction in particular requires that the author(s) and source are credited.
The license is subject to the Beilstein Archives terms and conditions: https://www.beilstein-archives.org/xiv/terms.
The definitive version of this work can be found at: doi: https://doi.org/10.3762/bxiv.2019.112.v1
This open access document is published as a preprint in the Beilstein Archives with doi: 10.3762/bxiv.2019.112.v1 and is
considered to be an early communication for feedback before peer review. Before citing this document, please check if a final,
peer-reviewed version has been published in the Beilstein Journal of Nanotechnology.
This document is not formatted, has not undergone copyediting or typesetting, and may contain errors, unsubstantiated scientific
claims or preliminary data.
Preprint Title Identification of the physical-chemical properties that modulate the
nanoparticles aggregation in blood
Authors Ludovica Soddu, Duong N. Trinh, Eimear Dunne, Dermot Kenny,
Giorgia Bernardini, Ida Kokalari, Arianna Marucco, Marco P.
Monopoli and Ivana Fenoglio
Publication Date 01 Okt 2019
Article Type Full Research Paper
Supporting Information File 1 SI.docx; 1.3 MB
ORCID® iDs Ludovica Soddu - https://orcid.org/0000-0001-6675-2318; Dermot
Kenny - https://orcid.org/0000-0001-5548-3263; Ida Kokalari -
https://orcid.org/0000-0003-0859-1354; Marco P. Monopoli -
https://orcid.org/0000-0002-2035-6894; Ivana Fenoglio -
https://orcid.org/0000-0002-6946-3105

1
Identification of the physical-chemical properties that modulate the
nanoparticles aggregation in blood
Ludovica Soddua,b, Duong N. Trinh c, Eimear Dunneb, Dermot Kennyb, Giorgia Bernardinia,c, Ida
Kokalaria, Arianna Maruccoa, Marco P Monopolic * and Ivana Fenoglioa*.
aDepartment of Chemistry, University of Torino, 10125 Torino, Italy
bMolecular and Cellular Therapeutics, Royal College of Surgeons in Ireland, Dublin, Ireland.
cDepartment of Chemistry, Royal College of Surgeons in Ireland, Dublin, Ireland
Correspondent Authors:
Ivana Fenoglio; ivana.fenoglio@unito.it
Marco Monopoli; marcomonopoli@rcsi.ie
Abstract
Inorganic materials are receiving significant interest in medicine given their usefulness for therapeutic
applications such as targeted drug delivery, carriers of active pharmaceutical and medical imaging.
However, the poor knowledge of the side effects related to their use is an obstacle to their clinical
translation. For the molecular drug development, safe-by-design has become as a novel
pharmaceutical strategy that allows a reduction of the costs and an acceleration of the translation of
research to market. In the case of materials, the application of such approaches is hampered by a poor
knowledge of how the physical and chemical properties of the material trigger biological response.
Hemocompatibility is a crucial factor for those materials that are intended for medical applications.
In particular, the formation of agglomerates is a serious side effect that may induce occlusion of blood
vessels and thrombotic events. Additionally, nanoparticles can interfere with the coagulation cascades
where they have been reported to induce both pro- and anti-coagulant properties where their
properties like size, shape and surface charge have been see to be critical parameters.

2
Here, we developed two sets of tailored carbon and silica nano/submicron-particles with three
different sizes (100-500 nm) with the purpose of investigating the role of surface curvature and
chemistry on platelet aggregation, activation and adhesion.
We show that that large carbon nanoparticles, but not small carbon nanoparticles or silica
nanoparticles, have a strong tendency to form aggregates both in plasma and blood, as a consequence
of the formation of a protein corona and not of platelets activation. Substantial differences were found
in the composition of the protein corona depending upon the chemical nature of the nanoparticles,
while the surface curvature plays a minor role. On the other hand, coagulation proteins were abundant
in the corona of both silica and carbon nanoparticles.
The results presented herein suggest that vessel occlusion and formation of thrombi in vivo may occur
through independent mode of action (MoA), differently affected by the physico-chemical properties
of the materials.
Keywords: platelets aggregation, nanoparticles, aggregation, size, surface chemistry
1. Introduction
Nanomedicine is probably one of the most exciting field of research in the branch of nanotechnology
as it has the potential to solve practical and effective solution on several diseases where traditional
drugs currently fail. However, a tremendous gap exists between the number of formulations proposed
and those approved for clinics [1]. Among the several factors that hampers the translation process,
the most critical ones are the batch-to-batch variability and the poor knowledge of the properties
modulating nanomaterial efficacy and safety [2].
Safe-by-design (SbD) approach has a great potential in accelerating the entrance of medicines into
market [3], reducing pre-clinical research and costs. However, a deep knowledge of the processes
leading to the adverse effects and of the physico-chemical properties governing such processes are
mandatory to build structure-activity relationships (SARs) that in turn enable the SbD approaches.

3
To the latter aspect, knowledge may only be derived by collecting a substantial amount of data from
libraries of nanomaterials with definite synthetic properties.
Therefore, understanding the processes that occur in the bloodstream are particularly relevant, not
only for nanoformulations administered by intravenous injection, but also for any material introduced
into the body.
Nanoparticles have been found to be able of promoting both haemorrhage or thrombosis [4,5], by
interfering with the coagulation system through different pathways. Depletion of soluble coagulation
factors (e.g fibrinogen, XII factor) may occur following adsorption of the factors at the surface of the
nanoparticles. On the other hand, activation of some factors by surface-driven exposure of cryptic
domains following adsorption was reported in some studies. Other studies reported the ability of
nanoparticles to damage or activate platelets, endothelial cells or monocytes [5].
Physico-chemical properties were found to be critical in determining the ability of the materials to
induce adverse effects. Several studies focused on the effect of surface charge and particle size. Albeit
such properties were shown to clearly affect the pro-anticoagulant activity of nanoparticles, the
direction of the effect varies, depending on the kind of material. For example, positive charged
dendritic nanoparticles were found to be more thrombogenic than negatively charged while for
polystyrene nanoparticles both positive and negative charged surfaces were found to induce platelet
activation [6,7]. These differences are likely due to the coexistence of different molecular processes,
but also to the combined effect of different physico-chemical properties.
Due to their promising properties, carbon and silica nanoparticles are among the most studied
inorganic materials for medical applications. Their ability to induce thrombus formation, along with
their mechanisms of action, is still under debate.
Both single-walled carbon nanotubes (SWCNTs) and multiwalled carbon nanotubes (MWCNTs)
were found to induce platelet activation [8,9] by perturbing Ca2+ homeostasis, an effect that was
hypothesized to be caused by the interaction of CNT with plasma and dense tubular system
membranes likely related to the fibrous shape [10]. On the other hand, contrasting data have been
reported on the potential of isometric carbon nanoparticles like carbon black, fullerenes and diesel

4
exhaust to induce platelet activation and aggregation [8,9,11]. Systemic administration of carbon
black in mice resulted in fibrinogen and platelet deposition in post capillary venules in the liver and
heart [12,13], suggesting a role of this protein in nanoparticle mediated platelet aggregation.
Silica nanoparticles of different sizes were found to activate glycoprotein IIb/IIIa and to induce the
expression of P-selectin [14] in platelets. On the other hand, silica nanoparticles are known to induce
oxidative stress in several cell lines including endothelial cells [15] and leucocytes [16,17], a process
that in vivo may indirectly induce platelet aggregation.
The interference with the coagulation system is not the only possible mechanism that may induce
vascular occlusion. Nanoparticles have a strong tendency to agglomerate in water. The degree and
kinetics of agglomeration are controlled by size, shape and surface chemistry of the particles. Strong
repulsive electrostatic charges and steric hindrance may stabilize the nanoparticles and prevent
agglomeration. In the bloodstream, agglomeration is related to the formation of a biocorona that
modifies the electrostatic and steric repulsion among particles [18]. Finally, protein–protein
interaction may lead to bridging among particles thus promoting agglomeration [19].
In the present study, a sample set of six silica and carbon nanoparticles with definite size and
morphology was used to explore the effect of size and surface properties on the protein corona
composition, agglomeration in plasma, platelet aggregation, activation and VWF-mediated adhesion.
2. Materials and methods
2.1 Reagents.
Sodium polyacrylate, D-(+)-glucose, thionine acetate salt, phosphate buffered saline powder, EDTA,
glutathione reduced and 5,5′-dithiobis(2-nitrobenzoic acid), tetraethyl orthosilicate (TEOS) were
obtained from Sigma-Aldrich (Germany). 5,5-dimethyl-1-pyrroline-N-oxide(DMPO) was obtained
from Cayman chemicals (USA). Ultrapure water was obtained from a Milli Q Plus system (Millipore,
Bedford, MA, USA) and was always used freshly prepared. All other chemicals and solvents used

5
were at least of analytical grade. When not otherwise specified, other reagents were purchased from
Sigma-Aldrich.
2.2 Synthesis of carbon nanoparticles.
Carbon nanoparticles (CNP) were produced starting from glucose using a one-step hydrothermal
process as described in Kokalari et al. 2019 [20]. Briefly, glucose was dissolved in 50 ml of ultrapure
water followed by the addition of 15 mg of sodium polyacrilate. The solution was introduced in a
pressure reactor system (Büchi AG) and heated at 190°C for 3 or 8 h. Synthesis parameters are
described in details in table 1. CNP were then purified using ultrapure water by filtration for large
carbon nanoparticles (CNP-L) or tangential flow ultrafiltration (Vivaflow 50R, MW 30 kDa) for the
medium and small carbon nanoparticles (CNP-M and CNP-S).
Glucose (g)
Surfactant
(mg)
Time
(h)
Temperature,
(°C)
CNP-S
2
15
3
190
CNP-M
2
15
8
190
CNP-L
5
15
8
190
Table 1. Synthesis parameters used for CNP
2.2 Synthesis of silica nanoparticles.
The silica nanoparticles (SNP) were prepared by hydrolysis and condensation of TEOS in the
presence of ammonia as a catalyst, based on the Stöber process [21]. Briefly, a defined amount of
TEOS was added dropwise, under magnetic stirring, at room temperature, to a solution of ethanol,
ammonia (33%) and ultrapure water. The detailed protocols are summarized in table 2.
The suspensions were subjected to repeated centrifugation steps at 11.000 rpm, for 15 min, in
ultrapure water and ethanol in order to isolate and wash the nanoparticles. The purified nanoparticles
were suspended in ultrapure water and stored at 4°C.

6
TEOS
(ml)
Ethanol
(ml)
NH3
(ml)
H2O
(ml)
Time
(min)
SNP-S
0.756
20
0.852
0.831
40
SNP-M
0.756
20
1.704
1.067
40
SNP-L
0.330
44
18
-
30
Table 2. Synthesis parameters for silica nanoparticles
2.3 Scanning Electron Microscopy (SEM).
Nanoparticle morphology was characterized using scanning electron microscopy, SEM Zeiss Evo
50XVP (Assing). Carbon and silica nanoparticles suspensions were diluted up to 0.05 mg/ml in
ultrapure water. A volume of 20 μl of the diluted suspensions was mounted on aluminium stubs, using
double-sided adhesive carbon tape and silicon wafers. Samples were dried overnight at room
temperature. In the case of silica nanoparticles, the samples were sputter-coated with thick gold film
(~ 17 nm) under argon atmosphere to improve secondary electron emission during SEM imaging.
Nanoparticle morphology was observed at an acceleration voltage of 20 kV.
2.4 Dynamic Light scattering (DLS).
The mean diameter and polydispersity index (PDI) of nanoparticles were obtained using a Zetasizer
(Nano ZS Malvern Instruments, UK) based on the dynamic light scattering (DLS) technique.
Measurements were performed on purified nanoparticles by analysing 0.5 ml of suspension in
ultrapure water, placed in a square polystyrene cuvette, at 25 °C. PBS 0.01 M, pH 7.4, Sigma Aldrich,
was used as a diluent in the case of the evaluation of the size after the protein corona formation.
2.5 Nanoparticle Tracking Analysis (NTA).
Analysis of the size distribution and concentration of CNP and SNP were performed by NTA using
a Nanosight NS300 (Malvern, UK) equipped with a blue laser (488 nm) and a quartz chamber for
sample injection, equipped with an O-ring top plate. For nanoparticles-hard protein corona

7
complexes, the samples were diluted in PBS, 0.01 M, pH 7.4. Dilution factor was chosen in order to
obtain 30 particles per frame, as suggested by the manufacturer’s recommendations. The
measurement duration was set at 60s.
2.6 ζ-potential.
Zeta-potential measurements were performed based on the Electrophoretic Light Scattering (ELS)
technique, using Zetasizer (Nano ZS Malvern Instruments, UK) as a function of the pH in the range
from 2-9. The nanoparticles suspensions were diluted in ultrapure water at a final concentration of
0.5 mg/ml. The pH of the suspensions was adjusted by adding diluted NaOH or HCl solutions and
the samples were introduced in a disposable folded capillary cuvettes (Malvern Panalytical).
2.7 Protein corona characterization.
2.7.1 Procurement and preparation of plasma.
The human plasma was obtained from healthy donors kindly provided by the Irish Blood Transfusion
Service (IBTS), St. James Hospital, Dublin. The plasma from eight different donors was mixed,
aliquoted and stored -80°C until use. Before using, the plasma was thawed and centrifuged at 16.000
rpm, 3 min, in order to discard any precipitated protein.
2.7.2 Methods.
Carbon and silica nanoparticles were incubated with different concentrations of human plasma diluted
in PBS 0.01M, pH 7.4 (Sigma Aldrich) at 37°C under agitation (150 rpm). The number of
nanoparticles in all samples was optimized in order to ensure a total surface area of 1.0 x 10-2 m2.
After the incubation, in order to obtain nanoparticles-hard protein corona complexes, the samples
were washed three times by centrifugation as previously described [22]. After the last washing step,
the pelleted samples were suspended in PBS, 0.01 M, pH 7.4, and the denaturizing solution, prepared
with dithiothreitol and 3x blue loading buffer that contained DTT in a ratio of 1:10 (New England

8
biolabs) and following the manufacturer instructions. The samples were sonicated for 5 minutes in an
ultrasonic bath and then incubated for 5 minutes at 95 °C to complete the protein denaturation. The
samples were loaded in a 10-wells 12% stacking gel and the electrophoretic analysis was conducted
at 130 V. After the electrophoretic separation, the gels were stained in Imperial Protein stain, (Thermo
Scientific) for 1 h and distained overnight (in ultrapure water). The densitometry analysis was
performed using the software ImageJ (NIH).
2.7.3 Mass spectrometry analysis
Samples were run on SDS-PAGE for 10 min before the protein bands were excised from the gel. The
proteins in the gel pieces were reduced with dithiothreitol, alkylated with iodoacetamide and digested
with trypsin (Promega Corporation) overnight at 37°C. The peptides were then extracted from the gel
matrix and prepared for MS analysis by using Pierce C18 Tips (Thermo Fisher) following the
manufacture procedure.
Peptide samples were analyzed on a quadrupole Orbitrap (Q-Exactive, Thermo Scientific) mass
spectrometer equipped with a reversed-phase NanoLC UltiMate 3000 HPLC system (Thermo
Scientific). Samples were loaded onto C18 reversed phase columns (10cm length, 75µm inner
diameter) and eluted with a linear gradient from 2 to 27% acetonitrile containing 0.5% acetic acid in
58 min at a flow rate of 250nL/min. The injection volume was 5μl. The mass spectrometer was
operated in data dependent mode, automatically switching between MS and MS2 acquisition. Survey
full scan MS spectra (m/z 300 – 1600) were acquired in the Orbitrap with a resolution of 70,000. MS2
spectra had a resolution of 17,500. The twelve most intense ions were sequentially isolated and
fragmented by higher-energy C-trap dissociation.
MS raw files were processed with MaxQuant software (version 1.6.2). The peak lists were searched
against the human FASTA database. The search included the modifications of cysteine
carbamidomethylation, methionine oxidation and protein N-terminal acetylation. A maximum of two
missed trypsin cleavages were allowed in the database search. The false discovery rate for both

9
peptides and proteins was set at 1%. After that, the ProteinGroup file from Maxquant was processed,
filtered and analysed with Perseus software to generate the top abundance table, hierarchical
clustering graph and numeric Venn diagrams.
2.8 Preparation of blood and PRP.
2.8.1 Ethics statement
Blood collection for this study was approved by the Royal College of Surgeons in Ireland and
Beaumont Hospital ethics committees. Written informed consent was obtained from all donors prior
to phlebotomy. All blood samples were taken in accordance with the declaration of Helsinki.
The blood plasma used for the corona studies was obtained from the Irish Blood Transfusion service
(IBTS) St Vincent hospital and the use is covered by the RCSI REC 1246b
2.8.2 Study participants
25 healthy donors were recruited for this study. All donors had no previous history of any major
disease and were free from any medication such as statins, antihypertensive medication, antiplatelet
agents such as aspirin, or anti-inflammatory medications such ibuprofen, for at least 12 days prior to
blood draw.
2.8.3 Preparation of blood and PRP
Venous blood was drawn from the antecubital vein using a 19-gauge butterfly needle connected
to a sterile polypropylene syringe. Blood was drawn into 3.2% (w/v) trisodium citrate
anticoagulant (1:9 volume of citrate to blood, final citrate concentration of 0.32%). Blood samples
were kept at room temperature with gentle rocking and used within 1 hour of phlebotomy. Whole
blood cell counts were recorded for each donor, using a Sysmex-KX21N haematology analyser
(Kobe, Japan). Blood samples were centrifuged at 170 g for 10 min to obtain platelet-rich plasma
(PRP) (Centrifuge 5417R, BIOTOOLS, CA).

10
2.9 Light Transmission Aggregometry (LTA).
Platelet aggregation was monitored by light transmission aggregometry (LTA) in a Chronolog-490D
aggregometer (CHRONO- LOG® Corporation, Havertown, PA). Adenosine diphosphate ADP and
collagen were used as activators of platelet aggregation.
The nanoparticles suspended in phosphate buffered saline (PBS) were added to 250 µl PRP and
incubated for 1 minute. The concentration of nanoparticles was calculated on the base of mean
diameter and nanoparticles concentration in number (NTA) to have a total of 8.6x10-4 m2/ml of
exposed surface area, equal to 1 mg/ml for CNP-S. In the case of CNP-L the concentration was lower
to avoid interference (7.25 x 10-5 m2/ml). 2.5 μl of 10 M ADP or 12.5 μl of 10 M collagen was
then added and aggregation monitored for 5 minutes with the suspension continuously stirred. Optical
density was also measured for nanoparticles suspended in PRP in the absence of agonists for up to 40
minutes of incubation. PRP without nanoparticles was used as control. The data are expressed as mean
of three independent experiments.
2.10 Flow Cytometry Fluorescence activated cell sorting (FACS).
Flow cytometry fluorescence-activated cell sorting (FACS) was used to evaluate platelet activation.
Nanoparticles were suspended in PRP at the same concentration used for LTA. 2.5 µl of the
fluorophore-conjugated antibody (CD62P) (1.5 µg/ml) (Becton Dickinson, Oxford, UK), which binds
P-Selectin and 87.5 µl of PBS were added to 10 µl of the suspension and incubated for 10 minutes.
Reactions were stopped by the addition of PBS to a final volume of 1ml. Mean fluorescent intensity
(MFI) was read on a Beckman Coulter Cytomics FC500 flow cytometer. The experiments were
repeated by adding 5 µl of ADP (10µM). The data are expressed as mean of three independent
experiments.
2.11 Dynamic Platelet Function Assay (DPFA).

11
The DPFA is a well-characterised real-time assay of platelet interaction with von Willebrand Factor
(VWF) under conditions of arterial shear [23-25].
The initial phases of platelet aggregation were assayed using the DPFA as previously described [23-
24]. Briefly, custom parallel plate perfusion chambers were coated overnight with 100 g/ml VWF,
washed with phosphate-buffered saline and blocked with 1% bovine serum albumin for 1 hour prior
to use. Whole blood was labelled with 1M DiOC6 (Sigma Aldrich, Ireland) for 5 minutes at 37 ºC
prior to perfusion through the chamber at an arterial rate of shear (1500 s-1). Platelet translocation
behaviour was recorded using real-time video microscopy at a rate of 19 frames per second. Image
stacks were analysed by a custom-designed and validated software package [23]. The assay
measurements obtained from this analysis include the number of platelets that interacted with the
VWF surface (Platelet tracks), the number of platelets that translocate over VWF (Translocating
platelets), the average speed at which platelet translocation occurred (Platelet translocation speed),
the distance a platelet translocated along the VWF surface (Platelet translocation distance), the
number of platelets that stably adhered to the VWF-coated surface (Static platelets), and the percent
surface coverage on the final frame (Percentage of platelet surface coverage). For this study we just
looked at the platelet adhesion parameters.
3. Results
3.1 Preparation of the libraries of silica and carbon nanoparticles.
In this study we synthetized two matching set of nanoparticles of silica and carbon that had similar
mean hydrodynamic diameters. DLS and NTA mean hydrodynamic diameters (table 3) confirmed a
similar size distribution across the two materials. The low polydispersity indexes indicate high
colloidal stability and narrow size range distribution.

12
DLS
NTA
Hydrodynamic
diameters
(nm)
Standard
Dev.
PDI
Hydrodynamic
diameters
(nm)
Standard
Dev.
SNP-S
114.1
±0.351
0.081
115.0
±1.52
SNP-M
235.1
±4.754
0.012
217.3
±4.16
SNP-L
488.1
±5.387
0.031
333.3
±14.5
CNP-S
179.5
±3.482
0.074
128.3
±2.52
CNP-M
259.7
±2.193
0.010
232.0
±6.08
CNP-L
485.2
±2.452
0.123
349.0
±4.36
Table 3. Mean of hydrodynamic diameters, PDI and standard deviation of each sample measured after
purification obtained with DLS (white background), compared with the mean of hydrodynamic
diameters and standard deviation obtained using NTA (grey background).
The hydrodynamic diameters measured by NTA were similar to those measured by DLS, except for
the large samples, for which the NTA value was lower. In fact, NTA, but not DLS, detected four
populations of particles having a different size (SI) that, in DLS contribute to the mean diameter
observed. This is in agreement with the PDI higher than the others.
In figure 1 the representative SEM images of the samples are presented. All particles appear perfectly
spherical with a uniform size, the latter confirming the DLS analysis.
Figure 1. SEM micrograph of silica and carbon nanoparticles

13
The ζ- potential of the samples was measured by ELS in the pH range from 2 to 9 (figure 2a)
Figure 2. a) Zeta potential versus pH curves for carbon nanoparticles (CNP-M and CNP-S) and silica nanoparticles
(SNP-L, SNP-M and SNP-S) suspended in water. b) Density of acid groups exposed at the surface of carbon
nanoparticles.
As expected, both silica and carbon samples exhibited a negative zeta potential across the whole pH
range. It gradually increased with increasing of the pH of the suspension never reaching positive
values indicating the presence at the surface of weakly acidic groups. In the case of carbon
nanoparticles, acidic carboxylic or phenolic groups formed during the synthesis are expected, while
silica exposes at the surface hydroxyl groups. At physiological pH (7.4) all particles exhibit a high ζ-
potential in the range of -40/-70 mV. Carbon nanoparticles exhibit a ζ- potential more negative than
the correspondent silica nanoparticles. Note that the ζ- potential curve of CNP-L is not reported since
this samples rapidly agglomerate by lowering the pH value, making unfeasible the measurement.
The presence of acidic groups at the surface of carbon nanoparticles was quantified by titration using
the dye thionine acetate [26]. In Figure 2b the density of acidic groups for carbon nanoparticles is
reported. The three samples differ slightly in terms of the density of acidic surface groups, with the

14
small NPs having the lowest density, in agreement with the observed less negative ζ- potential value.
The density of the acidic groups for silica was not determined since widely reported in literature [27,
28].
3.2 Physico chemical and proteomics characterisation of the NP-Hard corona
We then evaluated how the NP physico chemical properties would affect the biomolecular corona
formation. For this purpose, we exposed the same surface area of three-sized NP of silica and carbon
to increasing concentrations of human plasma, 10% to 80%, to mimic the in vitro / in vivo conditions
respectively. After 1 hour incubation at 37°C, the NP-corona complexes were separated from
unbound proteins and washed three times by centrifugation. SDS-PAGE and MS were used to obtain
information on the biomolecular corona composition in each sample. In figure 3 the SDS-PAGE gels
for the small silica and carbon nanoparticles are shown. The corona composition between the two
materials has some similarities when incubated at 10% of plasma, however it became highly specific
to the NP surface properties at higher concentration as we detected a significantly difference in the
corona composition.
Figure 3. SDS-PAGE gel of hard protein corona formed after 1h of incubation in human plasma. The gel bands
correspondent to fibrinogen and apolipoprotein A1 are shown.
In particular, at 10% of plasma, both NPs preferentially adsorb three gel bands of 72, 60 and 50kDa,
later identified as fibrinogen alpha, beta and gamma chain respectively, but significant differences

15
were also observed at lower molecular weight where a gel band of 25kDa was detected in the silica
corona while in the carbon NP corona was formed by 3 gel bands in the region of 36-30kDa and a
less pronounced band of a MW lower than 28kDa. At higher plasma concentration, the corona
composition of the SNP changed significantly where the fibrinogen gel bands were displaced by three
predominant bands of 90kDa and a duplet of 50kDa, later identified with histidine rich glycoprotein.
These findings were in agreement with a previous study where a similar effect was detected for 200nm
silica NP as previously reported [22]. Notably, the bands of correspondent to fibrinogen become more
attenuated at higher percentage of plasma in silica, but remain abundant in carbon corona (figure 4).
Small molecular weight bands also became displaced at higher plasma concentration.
Little differences in the protein corona composition were found depending upon the size of the
nanoparticles (SI) suggesting that the surface curvature, in the range considered, plays a minor role.
Figure 4. Abundance of fibrinogen and apolipoprotein A1 in silica and carbon corona at different plasma
concentration.
Label-free mass spectrometry was used to obtain the semi quantitative protein abundance of the
corona of small silica and carbon NP across three different plasma concentrations. A total of 118
proteins were found in the biomolecular corona of small carbon and silica NPs after incubation with

16
human plasma 10, 40 and 80%. Table 4 contains the top 20 proteins detected in each condition by
MS.
Table 4: Top 20 most abundant proteins in small silica (SNP-S) and carbon (CNP-S) hard corona samples at three
different plasma concentration (10, 40, 80%) based on the LFQ intensity.

17
Figure 5: Venn diagrams showing the number of proteins shared by small silica (Black) and carbon NP (Grey) protein
corona formed at different plasma concentrations.
Venn diagrams (figure 5) highlighted that the majority of the proteins were detected both in the SNP
and CNP at the higher plasma concentrations, while a minor overlap occurred at 10%. However, a
pronounced difference was observed when we compared the protein abundance by means of the label
free quantification (LFQ) across all conditions (table 4 and figure 6).
The presence of fibrinogen decreased firmly with increasing plasma concentration (80%) in silica
nano-corona, and was displaced by less abundant but with higher affinity proteins, like histidine rich
glycoprotein, kallikrein B and plasminogen confirming the SDS-PAGE findings and the literature.
Apolipoprotein A1, a major protein that form the HDL, has shown to have a preferential affinity
towards silica NP where it was detected across all conditions. The findings were also in agreement
with the SDS-PAGE where a gel band of 28kDa was detected only for silica NP. Other HDL
apolipopoproteins including apoA2 and A4 are also more abundant in silica corona than carbon one,
which might be attributed to the surface chemistry.
Interestingly, fibrinogen was strongly bound to the CNP also at higher concentration of plasma,
although with a less amount. Similarly, ApoB100, ApoB100, a protein that forms the VLDL/LDL,
and histidine rich glycoprotein were enriched at 10% of plasma but they become displaced by other
proteins, such as vitronectin and ITIH4 at higher concentration of plasma in CNP. Interestingly,

18
albumin (66 kDa), the most abundant protein in human plasma, is outside the top 10 proteins
identified with MS in all samples confirming that the composition of the protein corona, is
independent to the protein original abundance.
Figure 6: Relative concentration of relevant proteins on small silica (SNP) and carbon (CNP) coronas at plasma
concentrations of 10, 40 and 80%. The percentages were calculated based on the total LFQ intensity in each sample.
Protein grouping (Figure 7) confirmed that the coagulation factors are highly enriched in the corona
across all conditions although with different percentage (55-75%)

19
Figure 7: Classification of the human plasma corona proteins identified on small silica (SNP-S) and carbon (CNP-S) NPs
according to their biological functions. The LFQ intensity is used to calculate the percentages of protein groups.
In terms of molecular weights, most proteins found in the corona of both nanomaterials are between
20-60 kDa in weight, which accounts for about 70% of proteins (SI). Around 8% of the total corona
proteins have high molecular weights (>150 kDa).
3.3 Effect of hard corona on agglomeration
The effect of the protein corona on the tendency of nanoparticles to agglomerate was evaluated by
preparing the protein corona using the protocol described above, re-suspending the particles in PBS
and measuring the size distribution by NTA. The mean hydrodynamic diameters of the particles
with the protein corona generated at three different plasma concentrations (10, 40 and 80%) are
compared in Figure 8 with the hydrodynamic diameter of the pristine nanoparticles.

20
Figure 8. Effect of hard corona formed at different plasma concentrations on nanoparticles agglomeration in water.
The presence of the hard protein corona induced a substantial agglomeration in all carbon
nanoparticles. This effect was evident for the protein corona formed in plasma at 10% concentration.
By increasing the plasma concentration, the mean hydrodynamic size of the small carbon
nanoparticles becomes similar to the nanoparticles without the protein corona. Conversely, large and
mean size carbon nanoparticles remain highly agglomerated with a mean size higher than 1 µm. In
the case of silica nanoparticles no agglomeration was observed at all concentrations of plasma tested.
3.4 Platelet aggregation
pristine
10%
40%
80%
pristine
10%
40%
80%
pristine
10%
40%
80%
pristine
10%
40%
80%
pristine
10%
40%
80%
pristine
10%
40%
80%
0
200
400
1000
1200
1400
1600
1800
2000
2200
2400
CNP-L
CNP-M
mean dH
CNP-S SNP-S SNP-M SNP-L

21
The effect of the nanoparticles on platelet aggregation was measured using light transmission
aggregometry (LTA). NPs were added to platelet rich plasma (PRP) in the absence or presence of
two platelet activators, collagen and ADP.
In figure 9a,b the percent aggregation of platelets in after 5 minutes from the addition of ADP or
collagen induced by silica and carbon samples is reported. When platelets were activated by collagen,
a slight but significant increase in aggregation was observed for all samples. A similar trend was also
observed in the presence of ADP as activator, albeit the values did not significantly differ from the
control. In the case of medium and large CNP, a progressive decrease in optical density was observed
upon the addition of the nanoparticles (SI). To monitor this process, the optical density was measured
up to 40 minutes after addition of the nanoparticles in the absence of the activators (Figure 9c, SI).
Significant aggregation was detected for both silica and carbon medium and large nanoparticles, with
a major effect observed for CNP-L. This effect was previously reported by Bihiari and co-workers
[9] for SWCNTs and was attributed to the formation of nanoparticle-platelet aggregates. Note that
for large and medium size carbon nanoparticles, black aggregates were clearly visible at the end of
the experiments (SI).

22
Figure 9: Effect of nanoparticles on platelets aggregation. A) platelets activated with collagen; b) platelets activated
with ADP; c) platelets not activated after 5 (grey bars) and 40 (dark bars) minutes of incubation. ** p < 0,01 , * p <
0,05.
3.5 Platelet activation
The activation of platelets by the silica and carbon samples was evaluated by flow cytometry.
Activation was evaluated by using a specific antibody, which binds the antigen CD62P (P-Selectin)

23
present on the surface of activated platelets (Figure 10). No significant activation was detected in the
absence of activators. When platelets were activated with ADP, an increase of activation was
observed for both silica and carbon nanoparticles of mean size only. This increase was evident, but
not statistically significant due to the high variability of the response form one donor to the other.
The intensity of forward (FS) and side (SS) scattered light were also measured to evaluate the size of
platelets and the granularity respectively. A slight increase of platelet size was observed in the
presence of the silica and carbon NPs of medium size. The analysis of the platelet activation by flow
cytometry is particularly critical in the presence of particles due to the possible interference in the
intensity of scattered light. However, here this is not the case, since particles are clearly visible in the
SC-FC plot (SI) as separate population having a size smaller than platelets, and therefore excluded
by the measurement. However, possible interference may derive form aggregates.
Figure 10. Effect of the nanoparticles on platelet activation measured as secretion of P-Selectin

24
3.6 Platelet adhesion
Activated platelets are programmed to adhere to the endothelial wall of damaged blood vessels. Von
Willebrand factor protein (VWF) anchored to damaged endothelial cells plays a major role in this
process, encouraging platelets to tether, roll and finally adhere at the site of damage. To investigate
the possible interference of the nanoparticles on this process the Dynamic Platelet Function Assay
(DPFA) was used. This well-characterised assay monitors shear-mediated dynamic platelet
interactions with surface-immobilized VWF. Adhesion was measured as total number of platelets
adhering to the substrate (figure 11) in the presence or absence of the nanoparticles.

25
Figure 11. Effect of the nanoparticles on platelets adhesion. a) total number of platelets adhering to the substrate;
b) and c) representative images of the substrate during the measurements.
A size dependent decrease in the number of platelets adhering to the substrate was detected (Figure
11a). Figure 11b shows two representative images of the VWF coated microfluidic channel captured
during the flow run. For CNP-L, large aggregates were observed at the surface of the substrate
(circles). These aggregates were not visible for the other samples.
4. Discussion.
The identification of the correlations existing among physical and chemical properties and biological
effects is a laborious but necessary process allowing the design of more efficacious and safe
substances for medical applications. In the case of (nano)biomaterials this process is more challenging
than for molecular substances, due to the higher number of parameters to be controlled. A library of
nanomaterials that differ by one single property at time and accurate testing strategies are necessary.
This is not always straightforward due to the interdependence between the various chemical and
physical properties.
In the present study, two sets of nanoparticles were prepared with the aim to specifically investigate
the effect of the surface curvature and surface chemistry on platelet-dependent and independent
aggregation, platelet activation and adhesion. Silica and carbon nanoparticles were chosen since both
are highly studied for medical applications. Furthermore, being produced by wet methods, the
selected nanoparticles have both hydrophilic surfaces and are negatively charged. Their comparison,
therefore excludes surface charge and hydrophilicity as variables to be investigated. In figure 12 the
strategy used to unravel possible SARs is reported.

26
Figure 12. Strategy used to unravel possible structure-activity relationships.
This strategy allowed us to identify surface chemistry as key factor in the protein corona composition
while both surface chemistry and size modulate a platelet-independent aggregation potential of
particles in blood.
Platelet aggregation is a complex process modulated by several chemical and physical parameters.
Ordinarily platelets circulate in blood in a quiescent state near the endothelial cells lining the blood
vessels without forming stable adhesions. After infringement of the vasculature proteins like von
Willebrand factor (VWF), collagen and fibronectin are exposed on the sub endothelial matrix and act
as ligands for platelet surface receptors, glycoproteins like GPVI and GPIbα [29-31]. These receptor-
ligand interactions initiate a cascade of intracellular responses resulting in amplification of platelet
activation through the secretion of soluble agonists including thromboxane A2 (TXA2) and ADP.
TXA2 and ADP act jointly with the engaged platelet receptors to mobilize intracellular Ca2+, which
instigates platelet shape change, degranulation, and up-regulation of the adhesive function of another
platelet surface receptor, integrin αIIbβ3 [29]. The active conformation of αIIbβ3 integrin can then
bind fibrinogen (FG), VWF and fibronectin with high affinity allowing haemostatic platelets to

27
aggregate and thrombus formation [32]. FG plays a key role in platelet aggregation, forming bridges
between platelets and acting as an aggregation glue. On the other hand, FG also has a key role in
surface-driven aggregation. FG has a high affinity for surfaces [33], and it is commonly present in
the protein layer/corona of several materials [22] [34]. On hydrophilic surfaces, this protein tends to
be displaced by other proteins by a mechanism known as Vroman’s effect [33]. However, in some
cases adsorbed FG undergoes conformational changes thus exposing cryptic domains. Platelets may
adhere to fibrinogen immobilized onto biomaterials through integrins, a mechanism that may lead to
thrombotic events. Furthermore, in the case of NMs, this protein may act as glue in a similar way to
that observed with platelets [17], inducing nanoparticle aggregation. However, in this case the effect
is not due to the interaction with integrin, but it is a non-specific process due to the tendency of FG
to form fibrils similar to fibrin. This tendency is a consequence of specific arrangement of the FG
molecule onto surfaces, modulated by the surface properties [19].
Starting from the hypothesis of a possible effect of FG on nanoparticles platelet-dependent or
independent aggregation, we firstly investigate whether FG was present on the hard protein corona
of the NPs.
Fibrinogen was found in the hard protein corona of both silica and carbon nanoparticles regardless
their size, albeit it was relatively more abundant in the protein corona of carbon nanoparticles at high
serum concentration when compared with silica. In fact, the amount of FG decreases by increasing
the concentration of plasma due to the displacement of FG by proteins having higher affinity for the
surface, an effect more evident for silica nanoparticles and already described by some of us elsewhere
[22].
Unexpectedly, the presence of the protein corona clearly induces agglomeration of carbon
nanoparticles but not of silica. Notably, aggregation was observed for medium CNP and large CNP
at all plasma concentrations tested, while in the case of small CNP aggregation was observed only
with 10% of plasma, correspondent to the condition used in in vitro tests, but not in vivo.

28
The differences observed among silica and carbon nanoparticles might be a consequence of the
different composition of the protein corona, or to a different arrangement of protein molecules at the
surface, as a consequence of a different ability of the surfaces to interact with the proteins [18]. The
affinity of a protein for a certain surface and the mode of interaction rise form the interplay of
electrostatic interactions, hydrogen bonding, and hydrophobic forces [28]. Silica and carbon
nanoparticles are both negative at physiological pH. However, silica nanoparticles exhibit a less
negative ζ- potential when compared with carbon nanoparticles of similar size, likely due to a lower
mean Brönsted acidity of the surface, being the density of acidic carboxyl/ phenolic groups at the
carbon surface similar to the expected density of silanols (Si-OH) at a fully hydroxylated surface, i.e.
of 4-5 groups/nm2 [27-28]. Both surfaces exhibit surface sites able to form hydrogen bonds or
hydrophobic interaction with proteins. However, such tendency may be different since hydrogen
bonds formation obeys to geometrical constraints due to the directional character of this bond. On the
other hand, both silica and carbon surfaces exhibit hydrophobic patches, i.e. siloxane bridges and
carbon-carbon bonds respectively. These moieties have a different nature having the first a higher
dipolar character.
Previous studies reported fibrinogen-induced aggregation for silica nanoparticles [19,34]. However,
in these studies pyrogenic silica samples were used. These materials are very different to silica
produced by sol-gel methods, being formed by large aggregates and having a surface with a lower
degree of hydrophilicity [16]. A different arrangement of the fibrinogen molecule at the surface of
silica is therefore likely. In fact, we previously reported that the tendency of fibrinogen to self-
assemble to form fibrin-like fibrils increased by decreasing the hydrophilic character of silica [19].
When incubated in the presence of platelets, silica nanoparticles and small carbon nanoparticles
induce only mild aggregation (figure 9), in agreement with what previously found on silica
nanoparticles [14]. Aggregation was observed for prolonged time of incubation with large carbon
nanoparticles only. This process does not involve platelet activation and appears related to the ability
of particles to act as bridges among platelets, similarly to that observed by other authors with other

29
carbon nanomaterials [9,11]. This was confirmed for large nanoparticles (figure 10), while for
medium size particles platelet activation cannot be excluded.
In light of these evidences, the observed reduction of VWF-mediated adhesion of platelets to
endothelial wall induced by all NPs should be regarded as a consequence of the sequestration of
platelets by particles. In fact, this effect is more evident for large particles.
The different aggregation potential of CNP depending upon their size may explain the contrasting
data found in literature on isometric carbon nanoparticles. In fact, secretion of P-Selectin in vitro was
observed for carbon black [11] but not diesel exhaust particles [9], while platelet aggregation was
observed for amorphous carbon nanoparticles but not for the small-sized fullerenes [8]. Note however
that limited information relating to the physico-chemical properties of these materials was given by
these studies making it difficult to form a critical analysis of the results. Note moreover that while
CNT was reported to induce platelet aggregation [8,9] CNP did not. This supports the hypothesis by
De Paoli Lacerda and co-workers that an elongated shape is necessary for this process [10].
Conclusions
In conclusion, the present study suggest that NP-NP and NP-platelet aggregates may be generated in
blood following administration of nanoparticles that are stable and monodisperse in water also
without platelets activation. This process should be regarded as concern, since aggregates might
induce vessels occlusion in vivo. CNP have an aggregation potential higher than silica nanoparticles.
However, the reduction of size to less than 100 nm appears to improve the stability these materials
and, possibly, their biocompatibility. Further investigations will be necessary to confirm in vivo this
hypothesis.
Supporting Information
Supporting Information File 1:
File Name: SI.pdf

30
File Format: PDF
Acknowledgements
We are grateful for the financial supports from European Union’s Horizon 2020 (BIORIMA, No
760928) and Irish Research Council (Enterprise Partnership Scheme project EPSPG/2019/511).

31
References
1. Coty, J. B.; Vauthier C. J. Control. Release 2018, 275, 254–268.
2. Khorasani A. A.; Weaver J. L.; Salvador-Morales C. Intern. J. Nanomed. 2014, 9, 5729–5751.
3. Fadeel B. J. Internal Med. 2013, 274 (6), 578-580.
4. Ilinskaya A. N.; Dobrovolskaia M. A. Nanomedicine 2013, 8(5), 773-784. doi:
10.2217/NNM.13.48
5. Ilinskaya A. N.; Dobrovolskaia M. A. Nanomedicine 2013, 8(6), 969-81 doi:
10.2217/nnm.13.49
6. Dobrovolskaia M. A.; Patri A. K., Simak J.; Hall J. B.; Semberova J.; De Paoli Lacerda S. H.;
McNeil S. E. Mol. Pharmaceutics 2012, 9(3), 382–393. doi: 10.1021/mp200463e
7. Jones C. F.; Campbell R. A.; Franks Z.; Gibson C. C.; Thiagarajan G.; Vieira-de-Abreu A.;
Sukavaneshvar S.; Mohammad S. F.; Li D. Y.; Ghandehari H.; Weyrich A. S.; Brooks B. D.;
Grainger D. W. Mol Pharm. 2012, 9(6), 1599–1611.
8. Semberova J.; De Paoli Lacerda S. H.; Simakova O.; Holada K.; Gelderman M. P.; Simak J.
Nano Lett. 2009, 9, 3312–3317.
9. Bihari P.; Holzer M.; Praetner M.; Fent J.; Lerchenberger M.; Reichel C. A.; Rehberg M.;
Lakatos S.; Krombach F. Toxicology 2010, 10; 269(2-3);148-54.
doi:10.1016/j.tox.2009.08.011
10. De Paoli Lacerda S. H.; Semberova J.; Holada K.; Simakova O.; Hudson S. D.; Simak J. ACS
Nano 2011, 5(7), 5808–5813. doi:10.1021/nn2015369
11. Holzer M.; Bihari P.; Praetner M.; Uhl B.; Reichel C.; Fent J.; Vippola M.; Lakatos S.;
Krombach F. J Appl. Toxicol. 2014, 34(11), 1167-76. doi 10.1002/jat.2996
12. Khandoga A.; Stoeger T.; Khandoga A. G.; Bihari P.; Karg E.; Ettehadieh D.; Lakatos S.;
Fent J.; Schulz H.; Krombach F. J. Thrombosis Haemostasis 2010, 8, 1632–1640. doi:
10.1111/j.1538-7836.2010.03904

32
13. Khandoga A.; Stampfl A.; Takenaka S.; Schulz H.; Radykewicz R.; Kreyling W.; Krombach
F.; Circulation 2004, 109, 1320–1325. doi:10.1161/01.CIR.0000118524.62298.E8
14. Corbalan J. J.; Medina C.; Jacoby A.; Malinski T.; Radomski M. W. Int. J. Nanomed. 2012, 7,
631–639.
15. Napierska D.; Thomassen L. C.; Rabolli V.; Lison D.; Gonzalez L.; Kirsch‐Volders M.;
Martens J. A.; Hoet P. H. Small. 2009, 5(7), 846-853. doi:10.1002/smll.200800461
16. Gazzano E.; Ghiazza M.; Polimeni M.; Bolis V.; Fenoglio I.; Attanasio A.; Mazzucco G.;
Fubini B.; Ghigo G. Toxicol. Sci. 2012, 128, 158-170.
17. Marucco A.; Gazzano E.; Ghigo D.; Enrico E.; Fenoglio I. Nanotoxicology 2016, 10(1), 1-9.
18. Fenoglio I.; Fubini B.; Ghibaudi E.; Turci F. Adv. Drug Deliver. Rev. 2011, 63, 1186-1209.
19. Marucco A.; Turci F.; O’ Neill L.; Byrne H. J.; Fubini B.; Fenoglio I. J. Coll. Interf. Sci.
2014, 419, 86-94.
20. Kokalari I.; Gassino R.; Giovannozzi A. M.; Croin L.; Gazzano E.; Bergamaschi E.; Rossi
An. M.; Perrone G.; Riganti C.; Ponti J. Fenoglio I. Free Radic. Biol. Med. 2019, 134, 165-
176.
21. Stöber W.; Fink A.; Bohn E. J. Colloid Interf. Sci. 1968, 26(1), 62-69.
22. Monopoli M.; Walczyk D.; Campbell A.; Elia G.; Lynch I.; Baldelli Bombelli F.; J. Am.
Chem. Soc. 2011, 133 (8) 2525-2534.
23. Ralph A.; Somers M.; Cowman J.; Voisin B.; Hogan E.; Dunne H.; Dunne E.; Byrne B.; Kent
N.; Ricco A. J.; Kenny D.; Wong S. Cardiovasc Eng. Tech. 2016, 7(4), 389–405. doi:
10.1007/s13239-016-0282-x
24. Cowman J.; Richter L.; Walsh R.; Keegan N.; Tinago W.; Ricco A. J.; Hennessy B. T.;
Kenny D.; Dunne E. Platelets 2019, 30(6), 737-742
25. Dunne E.; Qi Q. M.; Shaqfeh E. S.; O'Sullivan J. M.; Schoen I.; Ricco A. J.; O'Donnell J. S.;
Kenny D. Blood 2019, 133(12), 1371-1377.

33
26. Pietroiusti A.; Massimiani M.; Fenoglio I.; Colonna M.; Valentini F.; Palleschi G.; Camaioni
A.; Magrini A.; Siracusa G.; Bergamaschi A.; Sgambato A.; Campagnolo L. ACS Nano 2011,
5, 4624–4633. doi:https://doi.org/10.1021/nn200372g.
27. Iler R.K., The Chemistry of Silica, Wiley and Sons, New York, 1979.
28. Rimola A.; Costa D.; Sodupe M.; Lambert J. F.; Ugliengo P. Chem. Rev. 2013, 113,
4216−4313. doi: dx.doi.org/10.1021/cr3003054 |
29. McFadyen J.D.; Kaplan Z.S. Transfusion Med. Rev. 2015, 29(2),110-9. doi:
10.1016/j.tmrv.2014.11.006
30. Ruggeri Z.M. Microcirculation 2009, 16(1), 58-83. doi: 10.1080/10739680802651477
31. Jackson S.P. Blood J. 2007, 109(12), 5087-95. doi:10.1182/blood-2006-12-027698
32. McFadyen J.D.; Jackson S.P. Thrombosis Haemostasis 2013, 110(5), 859-67. doi:
10.1160/TH13-05-0379.
33. Vroman L.; Adams A. L.; Fischer G.C.; Munoz P. C. Blood 1980, 55, 156-159.
34. Kendall M.; Ding P.; Kendall K. Nanotoxicology 2011, 5(1), 55-65.
doi:10.3109/17435390.2010.489724