The effect of shear flow on nanoparticle agglomeration and deposition in in vitro dynamic flow models.

http://informahealthcare.com/nan ISSN: 1743-5390 (print), 1743-5404 (electronic) Nanotoxicology, Early Online: 1–10 DOI: 10.3109/17435390.2015.1018978

ORIGINAL ARTICLE

The effect of shear flow on nanoparticle agglomeration and deposition in in vitro dynamic flow models Christin Grabinski1,2*, Monita Sharma1,3*, Elizabeth Maurer1, Courtney Sulentic3, R. Mohan Sankaran2, and Saber Hussain1,3

Nanotoxicology Downloaded from informahealthcare.com by Nyu Medical Center on 05/11/15 For personal use only.

1

Molecular Bioeffects Branch, Air Force Research Laboratory, Wright Patterson Air Force Base, OH, USA, 2Department of Chemical Engineering, Case Western Reserve University, Cleveland, OH, USA, and 3Department of Pharmacology & Toxicology, Boonshoft School of Medicine, Wright State University, Dayton, OH, USA Abstract

Keywords

Traditional in vitro toxicity experiments typically involve exposure of a mono- or co-culture of cells to nanoparticles (NPs) in static conditions with the assumption of 100% deposition (i.e. dose) of well-dispersed particles. However, cellular dose can be affected by agglomeration and the unique transport kinetics of NPs in biological media. We hypothesize that shear flow can address these issues and achieve more predictable dosage. Here, we compare the behavior of gold NPs with diameters of 5, 10 and 30 nm in static and dynamic in vitro models. We also utilize transport modeling to approximate the shear rate experienced by the cells in dynamic conditions to evaluate physiological relevance. The transport kinetics show that NP behavior is governed by both gravity and diffusion forces in static conditions and only diffusion in dynamic conditions. Our results reveal that dynamic systems are capable of producing a more predictable dose compared to static systems, which has strong implications for improving repeatability in nanotoxicity assessments.

Agglomerate density, dosimetry, exposure methods, nanoparticle transport

Introduction The field of nanotechnology has become a convergence point for many interdisciplinary sciences and applications (Cingolani, 2013). The possibilities of nanoparticle (NP) based tools are innumerable and so are the kinds of NPs that are being developed. This is indicated by the rise in consumer products containing NPs, with 1638 products recorded in 2013, a number that has grown exponentially since 2005 (The Project on Emerging Nanotechnologies, 2014). Cumulative government investment in nanotechnology since 2001 will exceed $20 billion in 2015 (National Nanotechnology Initiative, 2014). It is therefore crucial to evaluate the toxicity of these ‘‘nanocreations’’ to both address safety concerns and support sustainable technology development. There is strong motivation to improve in vitro techniques for assessing NP toxicity (Stokes & Wind, 2010). In their current state, most in vitro models are far from predicting human toxicity. However, animal use is costly and requires extended time lengths for data production, and in vivo results often do not accurately predict human toxicity (Hartung, 2013). A key advantage of in vitro studies is the possibility to improve the rate of throughput to screen the broad range of new materials being developed

*These authors contributed equally to this work. Correspondence: Saber Hussain, PhD, Air Force Research Laboratory, 2729 R Street, Wright Patterson AFB, OH 45433, USA. E-mail: [email protected] This work was authored as part of the Contributor’s official duties as an Employee of the United States Government and is therefore a work of the United States Government. In accordance with 17 U.S.C. 105, no copyright protection is available for such works under U.S. Law.

History Received 9 October 2014 Revised 4 February 2015 Accepted 9 February 2015 Published online 11 May 2015

(Damoiseaux et al., 2011). Therefore, it is critical to address challenges related to repeatability and relevance of in vitro NP toxicity investigations. Key parameters that require deeper understanding include NP characterization and dosimetry (Cho et al., 2013; Editorial, 2011; Hinderliter et al., 2010). It is known that NPs dispersed in cell culture media behave very differently than chemical compounds in terms of stability and transport properties (Teeguarden et al., 2007). NPs form a non-uniform distribution of agglomerates composed of one or more NPs and adsorbed media components, such as proteins (Arvizo et al., 2012; Lundqvist et al., 2008; Mukhopadhyay et al., 2012; Nel et al., 2009; Patil et al., 2007). Parameters used to describe agglomerates, such as hydrodynamic diameter and density, are often not reported (Baalousha & Lead, 2013). Therefore, more accurate characterization of NP agglomerates is critical to describe the exposure characteristics for improved repeatability in NP toxicity investigations (Cho et al., 2013). Traditional in vitro cell models for toxicity analyses typically expose cells grown at the bottom of a dish to NPs under static conditions and assume 100% deposition (Teeguarden et al., 2007). Further, in vitro studies investigating the effect of NP size assume equivalent dose (Chithrani et al., 2006; Coradeghini et al., 2013; Yu et al., 2009). These assumptions are misleading and not useful for hazard assessment in the absence of dosimetry measurements. NPs exposed in a static upright system are deposited onto the cells through a combination of sedimentation (i.e. deposition due to the force of gravity on an object with mass) and Brownian forces (i.e. deposition due to random Brownian motion). The sedimentation force is a function of primary particle size, agglomerate size and density, while the Brownian force only depends on agglomerate size (Hinderliter et al., 2010; Mason & Weaver, 1924).


2

C. Grabinski et al.

Recent studies have validated a volumetric centrifugation method for measuring the density of NP agglomerates (Cohen et al., 2014; DeLoid et al., 2014). However, eliminating sedimentation as a factor in dosimetry models would simplify predictions for particle transport. A recent study comparing cellular orientation (upright versus inverted) in an in vitro culture demonstrated that dosimetry is more predictable in an inverted orientation because NPs deposit onto cells by only Brownian force (Cho et al., 2011). However, some cell lines do not adhere well when inverted, and standard cell culture plates do not provide a facile option for a reproducible inverted setup. Therefore, alternate approaches are required to address issues related to NP deposition. Recent studies have demonstrated that shear flow can reduce particle agglomeration in mixing and nanocomposite applications (Kalra et al., 2010; Scurati et al., 2005). Furthermore, it has been observed that NP interaction with cells was significantly reduced in the presence of flow using a microfluidic-based model (Farokhzad et al., 2005). Toy et al. (2011) demonstrated that NPs with a larger density than the media carry momentum in the direction of flow and do not readily escape flow due to gravity. Therefore, deposition under dynamic conditions is dominated by diffusion. Based on these previous studies, we hypothesize that introducing NPs to cells in a flowing stream of media will reduce both the size of NP agglomerates and the effect of sedimentation during NP exposure, simplifying the dosimetry. In order to test our hypothesis, we carried out a systematic comparison between NP agglomeration and deposition in static versus dynamic exposure systems. Brain endothelial cells and astrocytes were cultured on either side of a porous membrane suspended in a cell culture insert to represent the blood brain barrier. The endothelial cells were treated with three different

Nanotoxicology, Early Online: 1–10

diameters (5, 10 and 30 nm) of gold (Au) NPs. Addition of flowing media reduced both the size of NP agglomerates and sedimentation. Therefore, the dynamic system allowed for the cellular interactions of three sizes to be investigated with improved predictability of dose. Further, the shear rate at the endothelial cell layer was found to be negligible, suggesting that the dynamic treatment approach can be used to improve repeatability in nanotoxicity assessments and is relevant for a broad range of adherent cell types.

Materials and methods In vitro model An in vitro model was developed using 24-well plates containing TranswellÔ inserts with 0.4 micron pore polyester membranes and 0.33 cm2 growing area (Corning Inc., Corning, NY). Two holes (0.2 cm apart) were drilled in the lid of the plate over each well and connectors were inserted into the holes. Two tubes (internal diameter 0.16 cm, outer diameter 0.32 cm) were inserted into the luminal side of the transwell at a height of 0.380 cm from the cell layer. The tubes were connected to a peristaltic pump (Idex Inc., Lake Forest, IL) to circulate flow (Figure 1a and b). Shear stress calculations The shear force experienced at the cell layer caused by flow occurs in the x–y direction (co-planar to the cells). The shear rate at the cell layer ð_ z¼0 Þ is the rate of shearing deformation and is equal to the velocity gradient along the cell layer @u=@zjz¼0 . The units are expressed as reciprocal seconds. The shear stress at the cell layer ðz¼0 Þ depends on the shear rate and the fluid viscosity () through the following relationship: z¼0 ¼ _ z¼0 .

Figure 1. Experimental setup and model geometry. (a) Schematic depicting the transwell insert in one well of a 24-well plate with an inlet and an outlet tube. (b) Image of the 24-well plate with each well connected to the peristaltic pump via an inlet and an outlet tube. (c) Geometry of the setup modeled in multiphysics software (COMSOL), based on dimensions of actual setup (dimensions are in cm). The section of the setup modeled in COMSOL is highlighted by the square with dashed lines in the schematic (a).

Nanoparticle agglomeration and deposition in in vitro dynamic flow models


DOI: 10.3109/17435390.2015.1018978

The viscosity for blood varies with shear rate, but is on the order of 3 mPas at shear rates greater than 10 s1, while the viscosity of media supplemented with 10% serum at 37 C is about 0.71–0.73 mPas (Folger et al., 1978; Hinderliter et al., 2010; Rosencranz et al., 2006). Based on these large differences in values for viscosity, it is unlikely that physiological wall shear stress occurs in an in vitro system. Therefore, we compared values for both shear rate and shear stress in our system versus physiological systems. Physiological shear rates were estimated from reported shear stress measurements in blood vessels using a viscosity of 3 mPas. A three-dimensional geometry of the in vitro model described above was setup in COMSOL (MultiphysicsÕ version 4.2a, Burlington, MA; Figure 1c). The computational fluid dynamics module with no slip condition was implemented and solved using the steady-state solver. A pressure drop from the inlet to outlet was implemented to generate desired flow rate (Table 1). The streamlines through the device were plotted (Figure 2a and c), and the shear rate at the cell layer (z ¼ 0) was plotted and calculated (Figure 2b and d and Table 1). Gold nanoparticles Spherical gold nanoparticles (Au NPs) with a nominal diameter of 30 nm were purchased from National Institute of Standards and Technology (NIST, Gaithersburg, MD) and smaller Au NPs with nominal diameters of 5 and 10 nm were synthesized. For both synthesis procedures, gold (III) chloride trihydrate (HAuCl43H2O) was purchased from MP Biomedicals (Santa Ana, CA). Trisodium citrate dihydrate (Na3C6H5O72H2O) and sodium borohydride (NaBH4) were purchased from SigmaAldrich (St. Louis, MO). Spherical Au NPs with a nominal diameter of 10 nm were synthesized by a modified version of the Turkevich method (Polavarapu & Xu, 2009). Briefly, 100 mL aqueous solution of 1 mM HAuCl4 was brought to boil with vigorous stirring. Next, 10 mL of 58.2 mM Na3C6H5O7 was added. The solution was boiled for 10 min, and then stirred rapidly for an additional 15 min without heating. The solution was then brought to boil again and 5 mL of additional Na3C6H5O7 solution was added, after which the solution was stirred vigorously for an additional 10 min without heating. The final concentration of Na3C6H5O7 was higher than the published method and the second heating/cooling steps were added in an effort to stabilize the particles more quickly, thus reducing their size from 13 to 9–10 nm. Spherical Au NPs with a nominal diameter of 5 nm were synthesized as previously described (Murdock, 2010). Briefly, stock solutions of 0.4% w/v (10 mM) HAuCl43H2O, 1.0% w/v (34 mM) Na3C6H5O7.2H2O and 0.5% w/v (132 mM) NaBH4 were

Table 1. Input and output parameters used in the multiphysics model. Input Flow rate (mL/min) Inlet velocity (cm/s) Pressure drop (Pa) Output Shear rate ð_ z¼0 , s1 Þ Average Maximum Minimum Shear stress ðz¼0 , dynes cm2 Þ Average Maximum Minimum

1.50 1.26 0.65

6.00 5.05 3.50

1.0 2.9 2.0 103

33.0 136 1.2 102

7.1 103 2.1 102 1.4 105

2.3 101 9.7 101 8.5 105

3

freshly prepared and kept on ice. Next, 1.25 mL of 0.4% HAuCl4 was added to 50 mL of de-ionized H2O with vigorous stirring for 1 min. Then, 500 mL of 0.5% NaBH4 was added rapidly to the sidewall of the solution while continuing to stir vigorously for 1 min. Finally, 200 mL of 1.0% Na3C6H5O7 was added to the solution and stirred for another minute. NaBH4 was used in the synthesis of 5 nm, but not 10 nm NPs, because it reduces the Au salt faster than Na3C6H5O7 (Polte et al., 2010). However, NaBH4 is known only to reduce the salt and does not act as a capping agent (Seoudi & Said, 2011). The citrate both reduces and caps the 10 and 30 nm NPs and only caps the 5 nm. Therefore, we assume that most of the NaBH4 was removed during post-synthesis washing steps and did not influence the ensuing experiments. Characterization of Au NPs The primary particle diameters and overall morphology of the Au NPs were characterized using transmission electron microscopy (TEM) on a Hitachi H-7600 (Chiyoda, Tokyo, Japan). NP suspensions were drop cast and dried on formvar/carbon filmcoated copper grids (Electron Microscopy Sciences, Hatfield, PA). The NP diameters were determined by averaging the measurement for 100 particles using Imaging Processing and Analysis in Java software (ImageJ, available from the National Institutes of Health, www.imagej.nih.gov, Bethesda, MD). The optical absorbance of the Au NPs, which shows the surface plasmon resonance band, was obtained with an ultraviolet–visible (UV–vis) absorbance spectrometer (Varian Cary 300 spectrometer, Agilent, Santa Clara, CA). The hydrodynamic diameter (dh), polydispersity index (PdI) and zeta potential (z) were measured using dynamic light scattering (DLS) on a Malvern Zetasizer Nano (Worcestershire, UK). The agglomeration of Au NPs dispersed in biological media was assessed by DLS. Au NPs at each diameter were dispersed in complete culture medium at a concentration of 5 mg/mL and exposed to static, low flow (1.5 mL/min) or high flow (6 mL/min) conditions for 24 h. The dh and PdI were obtained immediately after exposure. The average hydrodynamic diameters (n ¼ 6 for each treatment group) for static and low flow conditions were averaged and represented as the mean ± standard deviation. The peak hydrodynamic diameters (n ¼ 3 for each treatment group) for all three of the conditions were plotted. Functionalization of glass coverslips with aminosilane Glass coverslips were functionalized with aminosilane as described previously (Ungureanu et al., 2010). Briefly, German borosilicate glass coverslips (5 mm diameter, #1 thickness, Warner Instruments, Hamden, CT) were rinsed with 100% ethanol three times, then submerged in 5% solution of (3-aminopropyl) triethoxysilane (APTES, Sigma-Aldrich, St. Louis, MO) and ethanol for 15 min. After functionalization, the coverslips were rinsed five times with 100% ethanol. The coverslips were allowed to air dry completely under sterile conditions (in biohood) before being used for deposition experiments. Cell lines and culture conditions Cell lines of the blood brain barrier (astrocytes and endothelial cells) were chosen as a representative in vitro model for studying the effect of shear flow in a more physiologically relevant environment (Abbott et al., 2006). Nanoparticles exposed via inhalation have been shown to enter blood circulation (Nemmar et al., 2002). Also, Au particles have been shown to cross the blood–brain barrier (Sonavane et al., 2008). Therefore, it is

4

C. Grabinski et al.



Figure 2. Flow simulation for the dynamic flow model. The velocity streamlines (a and c) and shear rate (b and d) at the cell layer, z ¼ 0 are shown for flow rates of 1.5 mL/min (a–b) and 6.0 mL/min (c–d). The units for shear rate are in 1/s and units for the dimensions are in cm.

important to characterize the translocation of nanoparticles across endothelial barriers. Brain astrocytes (C8-D30, CRL-2534, American Type Culture Collection [ATCC], Manassas, VA) and endothelial cells (bEnd.3, CRL-2299, ATCC) were grown according to the supplier’s instructions in Dulbecco’s Modified Eagle’s Medium (ATCC) supplemented with 10% fetal bovine serum (ATCC). Cells were maintained in a humidified cell culture incubator at 37 C and 5% carbon dioxide. For all experiments, the abluminal side of the transwell membrane was coated with MatrigelÔ (BD, San Jose, CA) as previously described (Li et al., 2010), with modifications to obtain a thin coating on the membrane. Briefly, matrigel was diluted in serum-free media to a concentration of 476 mg/mL. The abluminal side of the transwell membrane was coated with 50 mL of the diluted matrix, and the transwells were incubated for 4 h at 37 C. Astrocytes were seeded at 2.8 104 cells/membrane on the matrigel-coated abluminal side of the transwell membrane. The inverted transwells were incubated for 3 h in a humidified cell culture incubator at 37 C and 5% carbon dioxide. The transwells with astrocytes were then put in a 24-well plate and incubated. After 24 h, endothelial cells were seeded on the luminal side at a density of 1.0 105 cells/membrane. The transwells were maintained in the incubator, and the media was changed every other day during culture. After 8 days in culture, the cells were characterized using fluorescence microscopy, TEM, transendothelial electrical resistance (TEER) measurement and permeability assays (Supplemental Figure S1). Cells were also imaged using TEM after a 24 h exposure to Au NPs (Supplemental Figure S5). Conditions for Au NP exposure NPs were dispersed in cell culture media at a concentration of 5 mg/mL and exposed under static or dynamic conditions for 24 h. The concentration was chosen based on a bio-distribution study by Sonavane et al. (2008) where mice were exposed intravenously

to 15 and 50 nm Au NPs. The doses were described in reference to the mass of the mouse. In order to obtain estimates for the mass per volume in the blood, we assumed about 5.5 mL of blood per 100 g mouse (Mouse Genome Database, 2014). The mice were administered about 4000 mg/mL, and at the end of the 24 h exposure, 0.2 mg/mL remained in the blood. NPs are cleared on the order of 30–60 min by the liver, kidney and lung (Niidome et al., 2006). Therefore, we conclude that the 5 mg/mL concentration is within a realistic concentration range for a 24 h exposure. Quantification of Au NP deposition and uptake Functionalized coverslips or co-cultures were exposed to Au NPs of different diameters in static or dynamic conditions for 24 h (n ¼ 3 for each treatment group). For the deposition assay, the coverslips were transferred to a conical tube after the exposure period. For the cell assay, the membrane with the cells was washed three times with warm 1 PBS, and then the membrane was cut out of the transwell and transferred to a conical tube containing 500 mL of deionized water. Since the washing steps may not remove all the surface bound NPs, we refer to the results as NPs interacting with the cells rather than taken up by the cell. However, we did observe a considerable amount of NP uptake via TEM after a 24 h exposure, with all of the NPs observed in the cells (Supplemental Figure S5). The conical tube containing the membranes were frozen at 80 C, then thawed and 75 mL of 1% triton-X 100 (Fisher Scientific, Waltham, MA) was added. For both the deposition and cell interaction assays, aqua regia at a 1:3 v/v ratio of nitric acid (HNO3, Sigma-Aldrich, St. Louis, MO) to hydrochloric acid (HCl, Fisher Scientific) was added to the tubes. Next, 3 mL of internal standard was added (Perkin Elmer, Waltham, MA), and the volume was brought up to 1.5 mL using deionized water. Inductively coupled plasma-mass spectrometry (ICP-MS, Perkin Elmer Nexion 300D, Waltham, MA) was used to



DOI: 10.3109/17435390.2015.1018978

5

Figure 3. Characterization of Au NPs. TEM images (a–c) and respective UV–Vis absorption plots (d–f) for Au NPs with nominal sizes of 5 nm (a and d); 10 nm (b and e); 30 nm (c and f).

Table 2. Characterization of Au NPs. Water

Media

dp (nm)

dh, zave (nm)

PdI

potential (mV)

dh, zave (nm)

PdI

potential (mV)

4.4 ± 0.70 9.0 ± 0.05 27.6 ± 2.10

6.10 ± 0.12 10.79 ± 0.57 28.30 ± 0.36

0.30 ± 0.01 0.13 ± 0.01 0.17 ± 0.01

32.00 ± 1.56 32.90 ± 0.20 40.20 ± 1.80

36.98 ± 0.87 32.37 ± 0.46 31.54 ± 0.21

0.64 ± 0.05 0.57 ± 0.01 0.50 ± 0.004

9.29 ± 1.14 8.32 ± 0.82 8.61 ± 0.77

The data were acquired for NPs in deionized water and immediately after dispersion in media at room temperature. The dp, dh and potential values are reported as the average ± standard deviation for n ¼ 3.

measure the Au content in the samples. Calibration plots were obtained by injecting a series of Au standard solutions (0.002, 0.050, 0.100 and 0.300 mg/mL in 1% HNO3 and 3% HCl), at a flow rate of 1.0 mL/min. Statistical analysis Significant differences at p50.05 were determined by a one-way ANOVA with a Bonferroni’s Multiple Comparison Test using GraphPad Prism software (La Jolla, CA).

Results Model setup and shear rate calculation A dynamic in vitro model was tested by perfusing media through the luminal portion of individual transwells in 24-well plates (Figure 1). The streamlines through the transwell and shear stress at the cell layer were obtained and plotted using COMSOL Multiphysics software (Burlington, MA; Figure 2). The geometry in COMSOL was based on the model setup and simulations were carried out at low flow (1.5 mL/min) and high flow (6.0 mL/min) conditions. The low flow rate condition was the maximum allowable without disrupting the cell monolayers, and the high flow rate condition was the maximum flow rate deliverable by the peristaltic pump. The simulated velocity streamlines within the transwells at both flow rates indicate mixing (Figure 2a and c). Although not

turbulent, it should be noted that vortical mixing can be observed at low Reynolds numbers when there is an obstruction in the flow (Kockmann et al., 2005). Using the simulation, the average shear rate at the cell layer was found to be 1.0 and 33.0 s1 at the low and high flow rates, respectively (Table 1). The maximum shear rate was concentrated around the cells located at the discharge of the inlet for both flow rates (Figure 2b and d). NP characterization Au NPs dispersed in water with nominal diameters of 5, 10 and 30 nm were characterized by TEM, UV–vis absorbance spectroscopy and DLS (Figure 3 and Table 2). The morphology of the primary particles for all three samples was found to be spherical (Figure 3), and the primary particle diameters (dp) were found to be 4.4, 9.0 or 27.6 nm (Table 2). The optical absorbance revealed a peak 520 nm corresponding to the well-known surface plasmon resonance of Au, with a slight shift in the peak towards longer wavelengths as the particle diameter increased, in agreement with previous reports (Hu et al., 2006). The hydrodynamic diameter (dh) characterized using DLS was slightly larger than the primary diameter (dp) determined via TEM due to the presence of citrate and formation of a hydration layer in water (Table 2). The agglomeration of the NPs in cell culture media was assessed by DLS. We observed a relatively large increase in size for the 5 nm Au NPs in media, followed by 10 nm, then 30 nm (Table 2). The polydispersity, as indicated by the increase in PdI,

6

C. Grabinski et al.


Table 3. Characterization data for Au NPs dispersed in media under static versus dynamic conditions. Static

Dynamic (1.5 mL/min)

Nominal diameter (nm)

dh, zave (nm)

PdI

dh, zave (nm)

PdI

Ddh (nm)

DPdI

5

50.17 ± 1.15

0.98 ± 0.04

29.36 ± 0.74

0.71 ± 0.18

10

31.85 ± 2.37

0.61 ± 0.12

25.85 ± 0.53

0.51 ± 0.02

30

37.57 ± 5.07

0.66 ± 0.16

30.29 ± 0.53

0.51 ± 0.01

20.8 ± 0.6 p50.01 6.0 ± 1.0 p50.01 7.3 ± 2.1 p50.01

0.27 ± 0.07 p50.01 0.10 ± 0.05 p ¼ 0.08 0.15 ± 0.06 p ¼ 0.04


The data were acquired for NPs in media after a 24 h static or dynamic exposure. The dh and PdI values are reported as the average ± standard deviation for n ¼ 6.

the peak dh (Figure 4a). A peak was observed for all samples of NPs in media near 9 nm, which we attribute to the presence of serum proteins. To follow trends in agglomeration, we observed the second peak, which represented agglomerate sizes greater than 20 nm. The trend was similar to that for average dh, which was smaller under dynamic (1.5 mL/min) than static conditions. The peak dh continued to decrease when the flow rate was increased from 1.5 to 6.0 mL/min, which is the maximum flow rate allowable in our dynamic setup. NP deposition and association with cells in dynamic versus static conditions cell culture

Figure 4. NP erosion under flow conditions. (a) Peak hydrodynamic diameter data for Au NPs dispersed in complete culture medium at a concentration of 5 mg/mL and exposed to static (S), low flow (LF; 1.5 mL/ min) or high flow (HF; 6.0 mL/min) for 24 h. Samples were analyzed immediately after exposure period using DLS. Results are representative of three experiments. A peak was observed for all samples of NPs in media near 9 nm, which we attribute to the presence of serum proteins. To follow trends in agglomeration, we followed the second larger sized peak. (b) Schematic depicting erosion of NP agglomerates under flow conditions.

also followed this trend, with the 5 nm NP agglomerates being the most polydisperse, followed by 10 nm and then 30 nm (Table 2). The surface charge was characterized by measuring the zeta (z) potential. The z potential for 5, 10 and 30 nm Au NPs exhibited negative values ranging from 40 to 32 mV when dispersed in water and 9 to 8 mV when dispersed in cell culture media. NP agglomeration in static versus dynamic conditions Agglomeration of Au NPs was evaluated in complete media after incubation under static conditions or perfusion across the luminal side of the transwells at low flow dynamic conditions for 24 h. A decrease in the average dh for NP agglomerates was observed under dynamic conditions as compared to static conditions (Table 3). This decrease in size was more evident for 5 nm than 10 nm and 30 nm Au NPs. The results here are unique from the results for NPs dispersed in media at room temperature shown in Table 2. Given limitations of DLS in accurately assessing the average agglomerate size in the presence of polydispersity, we also plotted

Since Au is known to have high affinity for amine groups (-NH2), the deposition of Au NPs was evaluated using aminosilanefunctionalized glass coverslips (Ungureanu et al., 2010). Functionalized coverslips were placed on the luminal side of the transwell membrane and exposed to NP solutions (prepared in media at 5 mg/mL) for 24 h under static or dynamic conditions (1.5 mL/min). Au NP deposition was quantified using ICP-MS and found to be greater under static versus dynamic conditions for the 5 and 10 nm NPs with a statistical significance of p50.05 (Figure 5a). There appeared to be more variability for the 30 nm NP under static versus dynamic conditions, but this difference was not statistically significant. Darkfield images of NPs deposited on the coverslips also indicated more NP deposition under static conditions compared to dynamic conditions for all the Au NP sizes tested (Supplemental Figure S2). Acellular conditions do not directly represent the bio-nano interactions, since NP interaction with cells usually results in energy-dependent internalization processes (Grabinski et al., 2014; Shukla et al., 2005; Untener et al., 2013). Therefore, we treated a co-culture of brain endothelial cells and astrocytes with static or continuous flow of media containing different diameters of Au NPs at 5 mg/mL for 24 h and quantified the NP interaction with cells using ICP-MS (Figure 5b). The cellular interaction followed similar trends as the deposition studies. There was a statistically significant increase in interaction for 5 and 10 nm NPs under static versus dynamic conditions. There appeared to be a greater amount of interaction for 5 and 10 nm versus 30 nm NPs under static conditions, but this was not statistically significant. Association of NPs with cells was also assessed using darkfield imaging, and qualitative analysis revealed that relatively more NPs were associated with cells in static than dynamic conditions (Supplemental Figures S3 and S4). Theoretical NP deposition In order to understand the differences observed between static and dynamic exposures, the deposition of Au NP agglomerates was estimated theoretically. For the static model, the transport process

DOI: 10.3109/17435390.2015.1018978


7

Figure 5. NP deposition and cellular uptake in static versus dynamic (1.5 mL/min) conditions after 24 h exposure as quantified using ICP-MS. (a) Au NP deposition on coverslips; (b) Cellular uptake of Au NPs. The error bars represent standard deviation for n ¼ 3, and the asterisks denote statistical significance (p50.05). The abbreviations S and D represent static and dynamic conditions, respectively.

was modeled by the following equation, which governs steady state transport of small particles (Mason & Weaver, 1924):


@n @2n @n ¼ D 2 vs @t @z @z

ð1Þ

where n is the particle concentration, D is the diffusivity, vs is the sedimentation velocity, t is time and z is distance. This partial differential equation was solved using the theoretical In vitro Sedimentation, Diffusion and Dosimetry (ISDD) model (Hinderliter et al., 2010) Based on Equation (1), it is evident that the transport of particles is a function of the diffusivity and sedimentation velocity. The former is a function of agglomerate size, while the latter is a function of both agglomerate size and density. The density of NP agglomerates can be described theoretically using a packing factor and fractal dimension (Sterling et al., 2005). The value of 0.637 was used for the packing factor, which was previously determined for random cluster packing (Sterling et al., 2005). The fractal dimension is a value between 1 and 3, that characterizes the space-filling factor of an agglomerate. A fractal dimension of three reflects a dense fractal structure with a higher relative agglomerate density and one reflects a porous fractal structure with lower relative agglomerate density. The deposition mass per surface area was calculated and plotted as a function of effective agglomerate density with a horizontal line indicating the value for average deposition measured using ICP-MS at each NP diameter (Figure 6a–c). The agglomerate density was calculated as a function of fractal dimension. It is important to note that recent studies have published results showing that the effective agglomerate density calculated based on the fractal dimension and packing factor often varies significantly from the density measured directly using volumetric centrifugation method (DeLoid et al., 2014). Regardless, dosimetry in the static condition is complicated by the requirement to estimate or measure the agglomerate density. For the dynamic condition, the critical force allowing NPs to escape the flow and deposit onto the cells is diffusion (Toy et al., 2011). Sedimentation is not considered because a particle with sufficient mass experiences momentum due to the flow of media, and gravity does not overcome this force. The flux of NPs was estimated using a simplified mass balance equation assuming that flow velocity occurs only in the x-direction and overall NP flux (j) occurs only in the vertical z-direction (Cussler, 1997). @cM @j @ ¼ þ cM vx @z @x @t

ð2Þ

The transport equation at steady state, assuming that the concentration of particles does not change in the direction of flow, reduces to:

@j @ 2 cM ¼ ¼0 @z @z2

ð3Þ

where cM is the mass concentration of particles per volume. Equation (2) can be solved analytically by applying boundary conditions. Assuming that at the center of the parabolic flow, h, the concentration is equal to that of the bulk, and, at the cell layer, the concentration is zero (i.e. NPs bind to the cells quickly), the appropriate boundary conditions are: At z ¼ 0,

At z ¼ h,

cM ¼ 0

cM ¼ 5

g cm3

ð4Þ

ð5Þ

The flux at the cell layer, jz ¼ 0, or concentration per surface area per time, was calculated using the following equation: jz¼0 ¼

D ðCM, z¼0 CM, z¼h Þ h

ð6Þ

The flux was multiplied by the exposure time for a 24 h exposure for each NP diameter (Figure 6d). A horizontal line is included to note the average value measured using ICP-MS. The analytical results show excellent correlation with the experiments.

Discussion The shear rates at the cell layer in the present study ranged from 2.0 103 to 2.9 s1 at the low flow rate (1.5 mL/min; Figure 2 and Table 1). To test whether the shear in our dynamic system is representative of physiological shear, we compared the shear rate in the dynamic model to physiological conditions. Reported ranges for physiologically relevant shear rates are 50–1600 s1. Specifically, physiological values were reported to be 1.5–2.0 Pa (shear rate 500 to 667 s1) for human arteries, 0.3 Pa (shear rate 50 s1) for higher diameter post-capillary venules, and 9.6 Pa (shear rate 1600 s1) for small capillaries (Koutsiaris et al., 2007; Ku, 1997). Another study reported physiological shear rates from about 587 to 3515 s1 in human arterioles (Koutsiaris et al., 2013). Changes in endothelial cell phenotype in vitro to mimic physiological conditions, including cell alignment with flow and formation of tight junctions, were reported at 0.2–5.0 Pa (shear rate 67–1667 s1; Seebach et al., 2000). Therefore, we conclude that the shear rates achieved using the transwell based dynamic setup were not likely to induce significant biological response from cells in culture. This suggests that a broad range of adherent cell types can be treated using a dynamic exposure condition. Au NPs were chosen for this study as they are relatively stable in a biological environment, i.e. the primary particles retain their morphology and size distribution and do not release ions. The NPs used in this study exhibited spherical morphology and uniform size distribution (Figure 3). The hydrodynamic diameter for NPs diluted in water was slightly larger than the primary size measured using TEM due to the formation of a hydration layer. When dispersed in media, the NP agglomerate size and polydispersity were found to increase with decreasing primary


8

C. Grabinski et al.


Figure 6. Theoretical estimation of NP deposition under static and dynamic conditions. (a–c) Theoretical deposition after 24 h of static exposure for (a) 5 nm, (b) 10 nm and (c) 30 nm Au NPs, as a function of effective density. The measured deposition is represented by a horizontal line for reference. (d) Theoretical deposition after 24 h of dynamic exposure.

particle size (Table 2). The change in z potential observed when the NPs were dispersed in water versus media indicates adsorption of serum proteins on the NP surface and has been previously observed for Au NPs in growth media (Alkilany & Murphy, 2010; Mukhopadhyay et al., 2012). NPs are known to agglomerate in the presence of salts due to a reduction in the electronic double layer around each particle, allowing for adhesion through van der Waals forces (Lyklema et al., 1999). In culture media containing serum proteins, NP agglomerates are stabilized by adsorbing serum proteins, forming a ‘‘protein corona’’ (Arvizo et al., 2012; Lundqvist et al., 2008; Nel et al., 2009). Our first hypothesis was that dynamic flow conditions versus static conditions would affect the size distribution of NP agglomerates. The NP agglomeration was measured using DLS after 24 h in each condition. Our hypothesis was confirmed, and the results indicated reduction in agglomerate size with increasing flow rate (Table 3 and Figure 4a). This can be attributed to the shear forces experienced by NPs under flow, which leads to erosion (depicted in Figure 4b). Erosion of particle agglomerates under flow has been previously observed in mixing applications and is characterized by the detachment of small fragments from the outer surface of the agglomerates (Scurati et al., 2005). Erosion increases with shear until a minimum particle size is achieved. Velocity streamlines within the transwells at both flow rates indicate mixing, which may play a role in the agglomeration behavior of the NPs (Table 1 and Figure 2a and c). At higher shear rates, as reported for physiological conditions, NP agglomeration is likely impacted by erosion to a much greater extent. Thus, erosion may play a role in the dose rate and toxicity of NPs in vivo and should be considered in future in vitro studies investigating the interaction of NPs with endothelial cells. The agglomeration and biomolecule adsorption kinetics will be further

altered due to the difference in formulation of blood versus cell culture media. This complicates the ability to mimic in vivo results using in vitro systems, which is also a topic of interest in future investigations. Our second hypothesis was that introducing NPs to cells in a flowing stream of media would reduce the effect of sedimentation during NP exposure, simplifying the dosimetry. The results confirm our hypothesis, where the deposition and cell interaction of NPs was consistently lower in dynamic versus static conditions (Figure 5). Transport equations were used to elucidate the mechanism by which differences in deposition occurred for the static versus dynamic condition (Figure 6). For the static condition, theoretical predictions are highly dependent on both particle agglomerate size distribution and agglomerate density. The density values are presented to highlight the predicted density values by comparing our experimental results to current theory (Cohen et al., 2014; Hinderliter et al., 2010). For more accurate predictions of agglomerate density, direct measurement by volumetric centrifugation method would be appropriate (DeLoid et al., 2014). Another solution would be to eliminate effective density as a parameter in dosimetry prediction. For the dynamic condition, the theoretical predictions and empirical data were strongly correlated, indicating that the assumptions used to simplify the transport equations were valid. Under flow conditions, deposition occurs almost exclusively by diffusion, which is a function of only particle agglomerate size distribution. Therefore, the deposition in the dynamic condition is more predictable. Further, although the deposition and dose under the dynamic condition appear to be relatively low, the values are likely more predictive of realistic dose rate in an in vivo condition, especially for the case of environmental exposures, where material tends to accumulate slowly over time.


DOI: 10.3109/17435390.2015.1018978


For both the static and dynamic conditions, the effect of polydispersity on NP agglomerate size was ignored. Many assumptions would be required to truly account for polydispersity using DLS data, since the output is in the form of percent intensity, which is qualitative (larger particles scatter more light). Further, in a previous study estimating the sedimentation coefficient for deposition under static conditions, minimal error was found when calculating deposition using averaged hydrodynamic diameter compared to summing calculations from binned particle sizes (DeLoid et al., 2014). Since only one peak distinct from the protein peak was observed in the DLS data in this study (Figure 4), we assume that the error would be similarly minimal here, and that the lower diffusivity of larger agglomerates is cancelled out by the higher diffusivity of smaller agglomerates. Regardless, it is important to note that the use of average diameter is only appropriate for relatively monodisperse size distributions of low polydispersity. Now that the dynamic flow concept has been demonstrated as a useful approach for simplifying Au NP dosimetry, it would be beneficial for follow-up studies to investigate the behavior of a large panel of NPs at a range of particle concentrations. Important NP properties may include agglomerate size, shape and density. For example, it is unclear how flow may affect erosion and deposition of metal oxide nanoparticles that can agglomerate up to several hundreds of nanometers in size (Murdock et al., 2008). Further, unagglomerated rod-shaped nanoparticles adhered to a channel surface more efficiently than spherical particles under flow (Toy et al., 2011). Therefore, different shaped particles would be interesting to test in a dynamic flow model, although it may not always be the case that the shape is maintained in the agglomerated state. Finally, density was previously shown to affect the propensity for particles to escape flow (Toy et al., 2011). However, the velocity was an order of magnitude greater than that used in this study. Since momentum is the product of velocity and mass, it would be valuable to further investigate the effect of agglomerate density in the context of low shear rates used for dosimetry where particle momentum is lower.

Conclusion In conclusion, we showed that treating an in vitro cell culture with NPs under flow conditions reduced the size and polydispersity of NP agglomerates versus static conditions. We also showed that NP deposition and cellular interaction were reduced under flow conditions, due to the elimination of the sedimentation force. Therefore, the dose under flow condition can be estimated from NP agglomerate size, which greatly simplifies the dosimetry. Consequently, at a low shear stress across an adherent cell layer, simplified dosimetry can be achieved, which has significant implications for improving repeatability across nanotoxicity assessments.

Acknowledgements We thank Prof. Harihara Baskaran at Case Western Reserve University for access to the COMSOL Multiphysics software package.

Declaration of interest The authors report no conflicts of interest. This work was funded by the Molecular Bioeffects Branch, Bioeffects Division, Human Effectiveness Directorate, 711 Human Performance Wing, Air Force Research Laboratory under the Student Research Participation Program at the U.S. Air Force Research Laboratory administered by the Oak Ridge Institute for Science and Education (to C.M.G. and M.S.).

9

References Abbott NJ, Ro¨nnba¨ck L, Hansson E. 2006. Astrocyte-endothelial interactions at the blood-brain barrier. Nat Rev Neurosci 7:41–53. Alkilany AM, Murphy CJ. 2010. Toxicity and cellular uptake of gold nanoparticles, what we have learned so far? J Nanopart Res 12: 2313–33. Arvizo RR, Giri K, Moyano D, Miranda OR, Madden B, McCormick DJ, et al. 2012. Identifying new therapeutic targets via modulation of protein corona formation by engineered nanoparticles. PLoS One 7: 1–8. Baalousha M, Lead JR. 2013. Nanoparticle dispersity in toxicology. Nat Nanotechnol 8:308–9. Chithrani BD, Ghazani AA, Chan WCW. 2006. Determining the size and shape dependence of gold nanoparticle uptake into mammalian cells. Nano Lett 6:662–8. Cho EJ, Holback H, Liu KC, Abouelmagd SA, Park J, Yeo Y. 2013. Nanoparticle characterization, state of the art, challenges, and emerging technologies. Mol Pharm 10:2093–110. Cho EC, Zhang Q, Xia Y. 2011. The effect of sedimentation and diffusion on cellular uptake of gold nanoparticles. Nat Nanotechnol 6:385–91. Cingolani R. 2013. The road ahead. Nat Nanotechnol 8:792–3. Cohen JM, Teeguarden JG, Demokritou P. 2014. An integrated approach for the in vitro dosimetry of engineered annomaterials. Part Fibre Toxicol 11:20. Coradeghini R, Gioria S, Garcı´a CP, Nativo P, Franchini F, Gilliland D, et al. 2013. Size-dependent toxicity and cell interaction mechanisms of gold nanoparticles on mouse fibroblasts. Toxicol Lett 217:205–16. Cussler EL. 1997. Diffusion in dilute solutions [chapter 2]. In: Diffusion, Mass Transfer in Fluid System. 2nd ed. New York: Cambridge University Press, 28–31, 45–47. Damoiseaux R, George S, Pokhrel S, Ji Z, France B, Xia T, et al. 2011. No time to lose-high throughput screening to assess nanomaterial safety. Nanoscale 3:1345–60. DeLoid G, Cohen JM, Darrah T, Derk R, Rojanasakul L, Pyrgiotakis G, et al. 2014. Estimating the effective density of engineered nanomaterials for in vitro dosimetry. Nat Commun 5:3514. Editorial. 2011. The dose makes the poison. Nat Nanotechnol 6:329. Farokhzad OC, Khademhosseini A, Jon S, Hermmann A, Cheng J, Chi C, et al. 2005. Microfluidic system for studying the interaction of nanoparticles and microparticles with cells. Anal Chem 77: 5453–9. Folger R, Weiss L, Glaves D, Subjeck JR, Harlos JP. 1978. Translational movements of macrophages through media of different viscosities. J Cell Sci 31:245–57. Grabinski CM, Salaklang J, Garrett CM, Schrand AM, Petri-Fink A, Hofmann H, Hussain S. 2014. Multi-functionalized SPIONs for nuclear targeting: Cell uptake and gene expression. NANO 9:1450009. Hartung T. 2013. Food for thought look back in anger – What clinical studies tell us about preclinical work. ALTEX 30:275. Hinderliter PM, Minard KR, Orr G, Chrisler WB, Thrall BD, Pounds JG, Teeguarden JG. 2010. ISDD: a computational model of particle sedimentation, diffusion and target cell dosimetry for in vitro toxicity studies. Part Fibre Toxicol 7:36. Hu M, Chen J, Li ZY, Au L, Hartland GV, Li X, et al. 2006. Gold nanostructures: engineering their plasmonic properties for biomedical applications. Chem Soc Rev 35:1084–94. Kalra V, Escobedo F, Joo YL. 2010. Effect of shear on nanoparticle dispersion in polymer melts: a coarse-grained molecular dynamics study. J Chem Phys 132:024901. Kockmann N, Engler M, Haller D, Woias P. 2005. Fluid dynamics and transfer processes in bended microchannels. Heat Transfer Eng 26: 71–8. Koutsiaris AG, Tachmitzi SV, Batis N. 2013. Wall shear stress quantification in the human conjunctival pre-capillary arterioles in vivo. Microvasc Res 85:35–9. Koutsiaris AG, Tachmitzi SV, Batis N, Kotoula MG, Karabatsas CH, Tsironi E, Chatzoulis DZ. 2007. Volume flow and wall shear stress quantification in the human conjunctival capillaries and post-capillary venules in vivo. Biorheology 44:375–86. Ku DN. 1997. Blood flow in arteries. Annu Rev Fluid Mech 29:399–434. Li G, Simon MJ, Cancel LM, Shi ZD, Ji X, Tarbell JM, et al. 2010. Permeability of endothelial and astrocyte co-cultures, in vitro blood-brain barrier models for drug delivery studies. Ann Biomed Eng 38: 2499–511.


10

C. Grabinski et al.


Lundqvist M, Stigler J, Ella G, Lynch I, Cedervall T, Dawson KA. 2008. Nanoparticle size and surface properties determine the protein corona with possible implications for biological impacts. Proc Natl Acad Sci USA 105:14265–70. Lyklema J, van Leeuwen HP, Minor M. 1999. DLVO-theory, a dynamic re-interpretation. Adv Colloid Interface Sci 83:33–69. Mason M, Weaver W. 1924. The settling of small particles in a fluid. Phys Rev 23:412–26. Mouse Genome Database (MGD). 2014. Mouse Genome Informatics website, The Jackson Laboratory, Bar Harbor, Maine. Available at: http:www.informatics.jax.org. Accessed on 26 May 2014. Mukhopadhyay A, Grabinski C, Afrooz ARMN, Saleh NB, Hussain SM. 2012. Effect of gold nanosphere surface chemistry on protein adsorption and cell uptake in vitro. Appl Biochem Biotechnol 167: 327–37. Murdock RC. 2010. Radio frequency controlled stimulation of intracellular gold or silver nanoparticle conjugates for use as potential sensors or modulators of biological function. Electronic MS Thesis, Wright State University. Available at: http://rave.ohiolink.edu/etdc/view? acc_num=wright1277753458. Accessed on 24 February 2015. Murdock RC, Braydich-Stolle L, Schrand AM, Schlager JJ, Hussain SM. 2008. Characterization of nanomaterial dispersion in solution prior to in vitro exposure using dynamic light scattering technique. Toxicol Sci 101:239–53. National Nanotechnology Initiative. 2014. NNI Budget. Available at: http//www.nano.gov/about-nni/what/funding. Accessed on 19 Jan 2014. Nel AE, Ma¨dler L, Velegol D, Xia T, Hoek EMV, Somasundaran P, et al. 2009. Understanding biophysicochemical interactions at the nano-bio interface. Nat Mater 8:543–57. Nemmar A, Hoet PH, Vanquickenborne B, Dinsdale D, Thomeer M, Hoylaerts MF, et al. 2002. Passage of inhaled particles into the blood circulation in humans. Circulation 105:411–14. Niidome T, Yamagata M, Okamoto Y, Akiyama Y, Takahashi H, Kawano T, et al. 2006. PEG-modified gold nanorods with a stealth character for in vivo applications. J Control Release 114:343–7. Patil S, Sandberg A, Heckert E, Self W, Seal S. 2007. Protein adsorption and cellular uptake of cerium oxide nanoparticles as a function of zeta potential. Biomaterials 28:4600–7. Polavarapu L, Xu QH. 2009. A simple method for large scale synthesis of highly monodisperse gold nanoparticles at room temperature and their electron relaxation properties. Nanotechnology 20:185606. Polte J, Ahner TT, Delissen F, Sokolov S, Emmerling F, Thu¨nemann AF, Kraehert R. 2010. Mechanism of gold nanoparticel formation in the classical citrate synthesis method derived from coupled in situ XANES and SAXS evaluation. J Am Chem Soc 132:1296–301.

Rosencranz R, Bogen SA. 2006. Clinical laboratory measurement of serum, plasma and blood viscosity. Am J Clin Pathol 125: S78–86. Scurati A, Feke DL, Manas-Zloczower I. 2005. Analysis of the kinetics of agglomerate erosion in simple shear flows. Chem Eng Sci 60: 6564–73. Seebach J, Dieterich P, Luo F, Schillers H, Vestweber D, Oberleithner H, et al. 2000. Endothelial barrier function under laminar fluid shear stress. Lab Invest 90:1819–31. Seoudi R, Said DA. 2011. Studies on the effect of the capping materials on the spherical gold nanoparticles catalytic activity. World J Nano Sci Eng 1:51–61. Shukla R, Bansal V, Chaudhary M, Basu A, Bhonde RR, Sastry M. 2005. Biocompatibility of gold nanoparticles and their endocytic fate inside the cellular compartment: a microscopy overview. Langmuir 21: 10644–54. Sonavane G, Tomoda K, Makino K. 2008. Biodistribution of colloidal gold nanoparticles after intravenous administration, effect of particle size. Colloids Surf B 66:274–80. Sterling MC, Bonner JS, Ernest ANS, Page CA, Autenrieth RL. 2005. Application of fractal flocculation and vertical transport model to aquatic sol-sediment systems. Water Res 39:1818–30. Stokes WS, Wind M. 2010. Recent progress and future directions at NICEATM-ICCVAM, Validation and regulatory acceptance of alternative test methods that reduce, refine, and replace animal use. Alt Anim Ex 27:87–95. Teeguarden JG, Hinderliter PM, Orr G, Thrall BD, Pounds J. 2007. Particokinetics in vitro, dosimetry considerations for in vitro nanoparticle toxicity assessments. Toxicol Sci 95:300–12. The Project on Emerging Nanotechnologies. 2014. Consumer Products Inventory. Available at: http//www.nanotechproject.org/cpi. Accessed on 19 Jan 2014. Toy R, Hayden E, Shoup C, Baskaran H, Karathanasis E. 2011. The effects of particle size, density and shape on margination of nanoparticles in microcirculation. Nanotechnology 22:115101. Ungureanu F, Wasserberg D, Yang N, Verdoold R, Kooyman RPH. 2010. Immunosensing by colorimetric darkfield microscopy of individual gold nanoparticle-conjugates. Sens Actuators B 150:529–36. Untener EM, Comfort K, Maurer EM, Grabinski CM, Comfort D, Hussain SM. 2013. Tannic acid coated gold nanorods demonstrate nonendosomal uptake and unique orientation within cells. ACS Appl Mater Interfaces 5:8366. Yu KO, Grabinski CM, Schrand AM, Murdock RC, Wang W, Gu B, et al. 2009. Toxicity of amorphous silica nanoparticle sin mouse keratinocytes. J Nanopart Res 11:15–24.

Supplementary material available online. Supplemental Figures S1–S5.

Flow dynamic effect of the anastomotic angle: a numerical study of pulsatile flow in vascular graft anastomoses models.

Post-stenotic core flow behavior in pulsatile flow and its effects on wall shear stress.

A master dynamic flow diagram for the shear thickening transition in micellar solutions.

Dynamic association between artery shear flow condition and platelet cytosolic free Ca2+ concentration in human hypertension.

Deposition and fine particle production during dynamic flow in a dry powder inhaler: a CFD approach.

The dynamics of inextensible capsules in shear flow under the effect of the natural state.

Measuring advection and diffusion of colloids in shear flow.

Flow-induced changes in Ca2+ signaling of vascular endothelial cells: effect of shear stress and ATP.

Increasing lifetimes and the growing saddles of shear flow turbulence.

Effect of pulsatile swirling flow on stenosed arterial blood flow.

Effect of hydration repulsion on nanoparticle agglomeration evaluated via a constant number Monte-Carlo simulation.

Prediction of anomalous blood viscosity in confined shear flow.

Applications of a theoretical model for haemolysis in shear flow.

Persistence of Brownian motion in a shear flow.

Strength Deterioration of Nonfractal Particle Aggregates in Simple Shear Flow.

Hydrodynamic interaction of two deformable drops in confined shear flow.

Dynamics of a self-diffusiophoretic particle in shear flow.

Effect of Varying Hemodynamic and Vascular Conditions on Fractional Flow Reserve: An In Vitro Study.

The effect of anchoring on the nematic flow in channels.

Animal models of surgically manipulated flow velocities to study shear stress-induced atherosclerosis.

Effect of TIPS placement on portal and splanchnic arterial blood flow in 4-dimensional flow MRI.

Effect of acute hyperthyroidism on blood flow, muscle oxygenation, and sympathetic nerve activity during dynamic handgrip.

The role of a nanoparticle monolayer on the flow of polymer melts in nanochannels.

Computational modeling of thrombotic microparticle deposition in nonparallel flow regimes.