Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Biomaterials 32 (2011) 195e205

Contents lists available at ScienceDirect

Biomaterials
journal homepage: www.elsevier.com/locate/biomaterials

Cytotoxic effects of iron oxide nanoparticles and implications for safety in


cell labelling
Stefaan J.H. Soenen a,1, Uwe Himmelreich b, Nele Nuytten a, Marcel De Cuyper a, *
a
Lab of BioNanoColloids, Interdisciplinary Research Centre, Katholieke Universiteit Leuven, Campus Kortrijk, B8500 Kortrijk, Belgium
b
Biomedical NMR Unit/MoSAIC, Faculty of Medicine, Katholieke Universiteit Leuven, Campus Gasthuisberg, B3000 Leuven, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: The in vitro labelling of cultured cells with iron oxide nanoparticles (NPs) is a frequent practice in
Received 11 August 2010 biomedical research. To date, the potential cytotoxicity of these particles remains an issue of debate. In
Accepted 27 August 2010 the present study, 4 different NP types (dextran-coated Endorem, carboxydextran-coated Resovist, lipid-
Available online 22 September 2010
coated magnetoliposomes (MLs) and citrate-coated very small iron oxide particles (VSOP)) are tested on
a variety of cell types, being C17.2 neural progenitor cells, PC12 rat pheochromocytoma cells and human
Keywords:
blood outgrowth endothelial cells. Using different NP concentrations, the effect of the NPs on cell
(Iron oxide) nanoparticle
morphology, cytoskeleton, proliferation, reactive oxygen species, functionality, viability and cellular
Magnetoliposome
Cytotoxicity
homeostasis is investigated. Through a systematic study, the safe concentrations for every particle type
Cell labelling are determined, showing that MLs can lead up to 67.37  5.98 pg Fe/cell whereas VSOP are the most toxic
Biomedical materials particles and only reach 18.65  2.07 pg Fe/cell. Using these concentrations, it is shown that for MRI up to
500 cells/ml labelled with VSOP are required to efficiently visualize in an agar phantom in contrast to only
50 cells/ml for MLs and 200 cells/ml for Endorem and Resovist. These results highlight the importance of
in-depth cytotoxic evaluation of cell labelling studies as at non-toxic concentrations, some particles
appear to be less suitable for the MR visualization of labelled cells.
Ó 2010 Published by Elsevier Ltd.

1. Introduction homeostasis. As FDA-approved, dextran-coated iron oxide cores


(Endorem) are intrinsically not well suited for cell labelling due to
In the past two decades, labelling of cultured cells with iron their inefficient uptake, the low affinity-binding and possible
oxide nanoparticles (IONPs) has become a frequently employed desorption of the dextran molecules, and their possible effects on
method in biomedical research. These magnetic IONPs can be used stem cell differentiation [6e10], many research groups have tried to
to non-invasively visualize labelled cells by magnetic resonance develop IONPs specifically suited for optimal cell labelling.
imaging (MRI) after transplantation, enhance the efficiency of drug However, the lack of appropriate control particles, standard
or gene delivery or be used as tools for magnetic hyperthermia procedures for cell labelling and the wide variety in types of
cancer treatment [1,2]. Recently, clinical trials have been set up labelled cells have made it impossible to compare different IONPs in
where cells are labelled in vitro with Endorem and subsequently terms of uptake efficiency and cytotoxic effects [6,11]. The disparate
injected in human patients in order to track the migration and data obtained from several studies have thus hardened making any
distribution of these cells [3]. As the use of IONPs for cell labelling is general conclusion regarding the safety of IONPs in cell labelling.
clearly increasing, the lack of information regarding the interaction Prior to evaluating the clinical potential of these particles in cell
of these particles with cells also becomes more apparent [4]. transplantation studies, the effects of IONPs on cultured cells
Recently, it has been shown that despite the initial belief in the should be carefully evaluated.
non-cytotoxic properties of IONPs, the physico-chemical properties Previously, it was shown by several groups that high intracel-
of nanoparticles [5] and the high intracellular concentrations of lular concentrations of IONPs diminished the proliferative capacity
IONPs required for efficient MRI poses serious threats on cell of cultured cells [12,13]. These effects were recently ascribed by our
group to be the result of cytoskeletal changes induced by the high
IONP concentrations which impeded actin-mediated signalling
[14]. Furthermore, when present in the acidic endosomal
* Corresponding author. Tel.: þ32 56 246221; fax: þ32 56 246997.
E-mail address: Marcel.DeCuyper@kuleuven-kortrijk.be (M. De Cuyper).
compartments of the cell, several types of particles have been
1
Current address: Lab of General Biochemistry and Physical Pharmacy, Faculty of shown to be degraded, rendering them near useless in terms of MR
Pharmaceutical Sciences, University of Gent, B9000 Gent, Belgium. contrast generation [15,16]. The degradation of the particles also

0142-9612/$ e see front matter Ó 2010 Published by Elsevier Ltd.


doi:10.1016/j.biomaterials.2010.08.075
196 S.J.H. Soenen et al. / Biomaterials 32 (2011) 195e205

affected cell functionality and homeostasis, indicating that the cells/well in 12 well plates containing glass coverslips at 0, 3 and 6 days post IONP
labelling procedure, if not tightly-controlled, could have grave incubation and allowed to settle for 24 h. Then, cells were fixed (2% para-
formaldehyde (PFA) for 15 min), permeabilized (1% Triton X-100 for 15 min) and
effects on the outcome of any cell transplantation study [17]. blocked for 30 min in PBS containing 10% goat serum (Gibco, Invitrogen, Belgium)
In the present study, four different types of IONPs (Endorem: and 2% bovine serum albumin (BSA). Cells were then incubated with primary
dextran-coat; Resovist: carboxydextran-coat; Magnetoliposomes antibody in blocking solution: anti-a-tubulin mouse monoclonal (no. 322500,
(MLs): lipid-coat; very small iron oxide particles (VSOP): citrate- 2.5 mg/ml; Molecular Probes, Leiden, Netherlands) or anti-vinculin mouse
monoclonal (no. ab18058, 1:200; Abcam, Cambridge, UK) for 2 h at ambient
coat) which are frequently employed in cell labelling studies are
temperature followed by 1 h incubation at ambient temperature with secondary
used to label C17.2 neural progenitor cells (NPCs), PC12 rat Alexa Fluor 488-conjugated goat anti-mouse antibody (1:250; Molecular Probes,
pheochromocytoma cells and human blood outgrowth endothelial Leiden, Netherlands) and Alexa Fluor 546-conjugated phalloidin (Molecular Probes,
cells (hBOECs). Using a systematic approach, the safe concentra- Leiden, Netherlands). Subsequently, cells were washed three times with blocking
tions of the IONPs were assessed by evaluating their effect on cell solution and mounted on microscope slides prior to being analysed by confocal
laser scanning microscopy (LSM510, Zeiss, Germany). To get quantitative data,
morphology and cytoskeleton, cell proliferation, induction of images were also collected at a 20 magnification by epifluorescence microscopy
reactive oxygen species (ROS), cell homeostasis, viability and using an Olympus BX51TF (Olympus, Tokyo, Japan) equipped with an Infinity 2
functionality. The use of the different cell types, each with their camera (BFi Optilas, Alphen aan den Rijn, Netherlands). Cell areas were calculated
specific traits allowed to efficiently analyze the various parame- using ImageJ (NIH, USA) for at least 150 cells per sample. For analysis of focal
adhesion areas, confocal images displaying vinculin were background-corrected,
ters. hBOECs are large, well-spread cells with a clear cytoskeleton,
thresholded, focal adhesions were identified and the total areas per cell were
C17.2 NPCs are very rapidly proliferating stem cells and were calculated for 20 cells per condition.
previously used to evaluate ROS induction and TfR1 expression
and PC12 cells have been introduced as a model cell type for the 2.5. Detection of reactive oxygen species
evaluation of cell functionality [18]. Upon establishing safe
Induction of ROS was quantitated using nitroblue tetrazolium salt (NBT; Sig-
concentrations of IONPs for cell labelling, the efficiency of the
maeAldrich, Bornem, Belgium). C17.2 cells were seeded in 96 well plates and after
different particles on MR visualization of cells labelled with the incubation with the IONPs in the absence or presence of 1 mM desferrioxamine
IONPs at non-toxic concentrations was studied and compared for (SigmaeAldrich, Bornem, Belgium), media were removed, cells washed twice with
the various IONPs. PBS (150 ml/well) and fresh medium containing NBT (1 mg/ml) was given, followed
by 6 h incubation at 37  C and 5% CO2. Media were then removed, cells washed once
2. Materials and methods with PBS after which 95 ml lysis solution (0.04 N HCl in isopropanol) was added to
each well. Plates were shaken at 1200 rpm for 30 min at ambient temperature. Then,
2.1. Nanoparticles 105 ml 10 N KOH was added, and plates were again shaken at 1200 rpm for 30 min
prior to measuring absorbance at 620 nm (FluoStar Optima plate reader, BMG
Resovist was purchased from Schering (Berlin, Germany), Endorem from Guer- Labtech, Isogen Life Sciences, Sint-Pieters-Leeuw, Belgium).
bet (Villepinte, France), VSOP C200 from Ferropharm (Teltow, Germany) and cationic
MLs (3.33% distearoyltrimethylammonium propane (DSTAP)-containing) were 2.6. Cell viability, proliferation and cellular iron content
produced and characterized by transmission electron microscopy, dynamic light
scattering, gas-liquid chromatography and electrophoretic mobility measurements Cell viability was assessed by means of a lactate dehydrogenase assay (CytoTox
as described previously [13,19]. Physico-chemical data on all the particles used can 96 non-radioactive cytotoxicity assay, Pierce, Rockford, US) and intracellular iron
be found in Supplementary Table S1. content was determined spectrophotometrically using Tiron (SigmaeAldrich, Bor-
nem, Belgium) as described previously [14]. Cell proliferation was assessed by
2.2. Cell culture conditions manual counting using a Bürker Chamber. To this end, C17.2 cells were seeded at
1  106 cells/culture flask and labelled with the IONPs for 24 h. After labelling, cells
C17.2 neural progenitor cells and PC12 rat pheochromocytoma cells were were kept in culture and were re-seeded at 1  106 cells/culture flask every 48 h. The
cultured in high glucose containing Dulbecco’s modified Eagle’s medium (DMEM), average cell division time for three independent tests is given.
supplemented with 10% fetal calf serum, 5% horse serum, 1 mM sodium pyruvate,
2 mM L-Glutamine and 1% penicillin/streptomycin (Gibco, Invitrogen, Belgium). C17.2 2.7. PC12 neurite formation
cells were passaged every 48 h and split 1/5. PC12 cells were passaged when
reaching near 70% confluency and split 1/5 in tissue culture dishes (Greiner Bio-One PC12 cells were seeded at 2  104 cells per well in 12 well plates containing
BA/BV, Wemmel, Belgium) which were coated with synthetic laminin peptide laminin-coated glass coverslips. After IONP incubation, cells were kept in culture in
(Synthetic laminin peptide for Rat Neural Stem Cells, Millipore SA/NV, Brussels, IONP-free medium overnight. Then, cells were given nerve growth factor (NGF)
Belgium). induction medium, consisting of high glucose DMEM, supplemented with 1% fetal
Human blood outgrowth endothelial cells were kindly donated by Dr. Simon De calf serum, 5% horse serum, 1 mM sodium pyruvate, 2 mM L-Glutamine, 1% penicillin/
Meyer (KULeuven Campus Kortrijk, Lab of Thrombosis Research). For culture, the streptomycin and 100 ng/ml NGF (Sigma Aldrich, Bornem, Belgium). Cells were kept
cells were maintained in endothelial basal/growth culture medium (EBM-2/EGM-2, in this medium for 48 h where after 24 h, half the medium was replaced by fresh
Clonetics, San Diego, CA) with medium changes every 48 h. Cells were passaged induction medium. Samples were then fixed, permeabilized and stained for F-actin
when reaching near 80% confluency by lifting the cells with 0.05% trypsin (Gibco) and a-tubulin and images were collected as described in Section 2.4. Image pro-
and were plated (1/5) onto tissue-culture flasks coated with collagen. cessing occurred using ImageJ software and using the epifluorescence images, the
number of neurites e defined as having twice the length of the cell body e could be
2.3. Cell exposure to IONPs calculated using an ImageJ plug-in, specifically developed for the analysis of neurites
[20]. Neurite length was also calculated and data are expressed as the average
For cell exposure to IONPs, all cell types were usually seeded at 5  104 cells/well number of neurites of a certain length per cell. At least 150 cells were analysed per
unless indicated otherwise. Cells were then allowed to settle overnight in a humid sample to allow statistical analysis.
atmosphere at 37  C and 5% CO2. Prior to cell labelling, Resovist, Endorem, VSOPs
and MLs were diluted in 0.5 ml phosphate buffered saline (PBS; Gibco, Invitrogen, 2.8. Transferrin receptor-1 expression
Belgium) each, apart from the already cationic MLs, supplemented with 25 ml Lip-
ofectamine 2000 and incubated for 30 min at ambient temperature with light To analyze expression of transferrin receptor-1 (TfR1), C17.2 cells were seeded at
shaking to allow complex formation. Next, 4.5 ml cell culture media was added and 2  105 cells/dish in 6 cm diameter tissue culture dishes. After IONP labelling (4
the appropriate volume (0.2 ml per well in 96 well plates and 1 ml per well in 12 well dishes per type of IONP), cells were kept in culture for 24 h in IONP-free medium.
plates) could be given to the cells. As controls, cells were exposed to media not Cells were then re-seeded at 2  105 cells/dish and kept in culture for another 72 h.
containing any particles but having 10% PBS as was the case for the IONP suspen- This step was repeated once, meaning that in total, the cells had been further
sions also. Cells were incubated for 24 h at 37  C and 5% CO2 prior to being further cultivated for 1 week. Cells were then lifted by scraping and counted using a Bürker
analysed as described in the respective paragraphs below. chamber followed by repeated centrifugation and washing with PBS (2 repeats).
Cells were resuspended in 1 ml PBS and equal portions (0.5 ml) of every tube was
2.4. Staining of F-actin, a-tubulin and vinculin divided over 2 tubes followed by centrifugation. One tube was incubated for 30 min
with FITC-tagged rat anti-mouse TfR1 antibody (BD Biosciences Pharmingen,
To analyze the effects on cell morphology, hBOECs were labelled with the Erembodegem, Belgium; 1 mg/ml in FACS buffer: PBS containing BSA (1%) and fetal
different IONPs, kept in culture in IONP-free medium and then re-seeded at 2  104 calf serum (0.1%)); the other tube was treated with FITC-tagged rat anti-mouse
S.J.H. Soenen et al. / Biomaterials 32 (2011) 195e205 197

Fig. 1. High intracellular nanoparticle concentrations affect cell spreading. Confocal micrographs of hBOECs incubated for 24 h either untreated (A) or incubated with (B) VSOP, (C)
Endorem, (E) MLs or (G) Resovist at 600 mg Fe/ml for 24 h, (D) Endorem at 400 mg Fe/ml, (F) MLs or (H) Resovist at 300 mg Fe/ml for 24 h. Cells were kept in culture for an additional 3
days. F-actin is coloured red, the nucleus is coloured blue by DAPI staining and a-tubulin is indicated in green. Green and red channels are shown separately for every image. Scale
bars: 50 mm.

antibody (BD Biosciences Pharmingen, Erembodegem, Belgium; 1 mg/ml) and served 8 h. Cells were then lifted by trypsin, fixed by 2% PFA and stock suspensions were
as an isotype control. Cells were centrifuged and washed with FACS buffer repeat- made of all the cells in PBS at a density of 2  105 cells/ml. Cell samples are trans-
edly (3 repeats) and then resuspended in 0.4 ml FACS buffer and analysed using ferred to eppendorf tubes, reaching densities of 2000, 500, 200, 50, 20 cells/ml. For all
fluorescence activated cell sorting (EPICSÒ XL-MCL, Coulter, Analis, Gent, Belgium). the cells, a total volume of 150 ml was taken, which was then mixed and solidified in
a 1.5% agarose gel (Invitrogen, Merelbeke, Belgium). After solidifying, the eppendorfs
2.9. Magnetic resonance imaging and sample preparation were completely filled with 1.5% agarose gel. For control samples, unlabelled cells at
a density of 2000 cells/ml were taken. For every type of IONP, the samples were put
To prepare samples for MRI, C17.2 cells at 70% confluency in culture flasks were together in a plastic cylinder which was filled with 1.5% agarose gel. Upon solidi-
labelled with the IONPs at the non-toxic concentrations indicated in the main text. fying, these agar cups were analysed in a Bruker Biospec 9.4 T small animal MR
After labelling, cells were kept in culture with IONP-free medium for an additional scanner (Bruker Biospin, Ettlingen, Germany; horizontal bore, 20 cm) equipped with
198 S.J.H. Soenen et al. / Biomaterials 32 (2011) 195e205

Fig. 2. High intracellular nanoparticle concentrations affect focal adhesion formation and maturation. Confocal micrographs of hBOECs incubated for 24 h either untreated (A) or
incubated with (B) VSOP, (C) Endorem, (E) MLs or (G) Resovist at 600 mg Fe/ml for 24 h, (D) Endorem at 400 mg Fe/ml, (F) MLs or (H) Resovist at 300 mg Fe/ml for 24 h. Cells were kept
in culture for an additional 3 days. F-actin is coloured red, the nucleus is coloured blue by DAPI staining and vinculin is indicated in green. Green and red channels are shown
separately for every image. Scale bars: 50 mm.

actively shielded gradients (600 mT m1). A quadrature RF resonator (transmit/ loaded areas in the agar phantoms. Data were expressed as mean  standard
receive; inner diameter 7 cm, Bruker Biospin) was used. 2D multi-slice-multi-echo deviation.
(MSME) experiments were acquired for the calculation of T2-maps (TR ¼ 3000 ms
and 16 TE increments of 10 ms, 2562 matrix, 275  275 mm in plane resolution, 2.10. Statistical analysis
0.35 mm slice thickness). T2* maps were acquired similarly to MSME experiments
using a gradient echo pulse sequence and 16 TE increments (first TE ¼ 4.5 ms with All data are expressed as mean  s.e.m. unless indicated otherwise and analysed
increments of 6.7 ms). Three-dimensional, high-resolution T2*-weighted MR using one-way analysis of variance (ANOVA). For cellular iron content, multiple
images were acquired using a gradient echo sequence (FLASH, TR ¼ 200 ms, comparisons were analysed using the Tukey-HSD post-hoc method. When
TE ¼ 15 ms, flip angle 30 ). The resolution was usually 75  75  75 mm. comparing the different NPs to the same control group (reference), the Dunnett
Images were processed using Paravision 5.0 (Bruker Biospin). Relaxation times post-hoc analysis method was used. In all cases, the degree of significance is indi-
(T2/T2*) were determined as mean values of homogeneous sections of the cell cated when appropriate (*: p < 0.05; **: p < 0.01; ***: p < 0.001).
S.J.H. Soenen et al. / Biomaterials 32 (2011) 195e205 199

Day 1 Day 3
A 35 B 35
**
** **
* **
Cell doubling time (h)

Cell doubling time (h)


28 * 28

21 21

14 14

7 7

0 0

)
)

)
)
)

)
)

)
)

)
)
)

ol
ol

00

00
00

00
00
00

00
00

00
00

00
00
00

00

VS ntr
VS ntr

(6

(3
(6

(3
(4
(6

(6
(6

(3
(6

(3
(4
(6

(6

o
o

o
do

L
do
P

L
C
o

o
do

L
do
P

L
C

es

es
M
M
es

es
M

O
M
O

En

En
En

En

R
R

C 35 Day 6
Cell doubling time (h)

28
* *
21

14

0
)

)
)

)
)
)

)
l

00

00
00

00
00
00

00
tro

(6

(3
(6

(3
(4
(6

(6
on

o
do

L
do
P

L
C

es

es
M
M
O

En

En
VS

Fig. 3. High intracellular nanoparticle concentrations impede cell cycle progression. Cell doubling times for C17.2 NPCs, either as control cells or after incubation with the particles
at the indicated concentrations for 24 h. Cell numbers were counted after (A) 1, (B) 3 or (C) 6 days after incubation with the nanoparticles. Values are represented as
mean  standard error to the means (n ¼ 3). Where appropriate, statistical significance is indicated: *: p < 0.05; **: p < 0.01.

3. Results Resovist and MLs at 600 mg Fe/ml clearly displayed a diminished


cell spreading, whereas Endorem-treated cells showed slight
3.1. Physico-chemical properties of the IONPs effects but VSOP-treated cells were not affected. The effects seen
were only transient (data not shown) and after 6 days, no signifi-
In the present work, four highly differing IONPs were used, cant changes in cell surface area between control cells and IONP-
providing a broad scala in different physico-chemical properties. An treated cells could be observed. Reducing the concentration of
overview of the most important features (core size, hydrodynamic Endorem to 400 mg Fe/ml and Resovist and MLs to 300 mg Fe/ml
diameter, surface charge) can be found in the supplementary abolished any effects on cell cytoskeleton as can be seen in Fig. 1D,
information (see Supporting Table S1). The MLs used throughout this F, H. Histograms representing the global distributions of the
work consist of 14 nm diameter iron oxide cores each individually respective cell surface areas for the different conditions can be
surrounded by a lipid bilayer (96.67% dimyristoylphosphatidylcho- found in Supporting Figs. S1 and S2.
line and 3.33% distearoyltrimethylammonium propane). As MRI
requires high doses of contrast agent due to the intrinsic low
3.3. Effects of high intracellular IONP concentrations on focal
sensitivity of the method, uptake of all particles here apart from the
adhesions and proliferation
already cationic MLs, was enhanced by combining the IONPs with
Lipofectamine 2000 (5 ml/ml total incubation medium), a frequently
Apart from the cell surface area, the total area occupied by focal
used transfection agent in cell labelling procedures [21].
adhesions is also affected when hBOECs are incubated with
Resovist and MLs at 600 mg Fe/ml, but not at 300 mg Fe/ml as can be
3.2. Effects of high intracellular IONP concentrations on cell seen in Fig. 2. In line with the minor (Endorem) or lack (VSOP) of
cytoskeleton any effect on the actin cytoskeleton and microtubule network,
Endorem led to slight effects whereas VSOP did not have any effects
In line with previous findings [13,14], cellular cytoskeleton and on focal adhesion complexes (see Supporting Fig. S3 for quantita-
morphology were affected at high intracellular concentrations of all tive data). As a diminished signalling through focal adhesion
types of IONPs. As can be seen in Fig. 1, hBOECs incubated with complexes along actin fibers was previously found to result in
200 S.J.H. Soenen et al. / Biomaterials 32 (2011) 195e205

A 900%
particles displayed a pronounced increased level of TfR1, indicative
of an altered iron metabolism. For MLs, the increase in ROS levels
*** Normal and the augmentation in TfR1 expression levels were only minimal,
750% +Desf suggesting the least intracellular degradation.
Relative ROS levels

3.5. Effects of intracellular IONPs on cell functionality


600%
To verify whether the effects on cell homeostasis have any
450% consequences for cell functionality, PC12 cells were first incubated
*** with the different IONPs and subsequently exposed to nerve
300% growth factor (NGF) after which the responsiveness of the cells to
their biological cue was assessed by means of analysing neurite
outgrowth (Fig. 6). The results indicate a clear effect of all particles
150% on the functionality of PC12 cells at the concentrations used. Using
a concentration-series for all IONPs, PC12 cell functionality was
0% found to be unaffected when incubating the cells with MLs at
VSOP (600) Endo (400) ML (300) Reso (300) 250 mg Fe/ml; Resovist: 150 mg Fe/ml; Endorem: 200 mg Fe/ml and
VSOP: 200 mg Fe/ml. The respective length and number of neurites
found in PC12 cells incubated with the different particles at the
B 900% concentrations used can be found in Fig. 6KeO.
Normal
750% +Desf 3.6. Efficacy of labelling cells with IONPs at non-toxic
concentrations
Relative ROS levels

600% ***
*** To verify the lack of any toxic effects of the IONPs at the
*** concentrations described in the paragraph above, a lactate dehy-
450% drogenase assay was performed on C17.2 cells as shown in Fig. 7A.
The results indicate that IONP labelling at the non-toxic concen-
trations indeed did not cause any acute toxicity. However, as for
300% ** ** most biomedical applications concerning the use of IONPs, not the
** **
** administrated dose of IONPs, but rather the intracellular concen-
150% tration of these particles is an essential parameter in determining
the efficacy of the envisioned application. To this end, the intra-
cellular IONP concentration in C17.2 cells was determined as shown
0%
VSOP (600) Endo (400) ML (300) Reso (300) in Fig. 7B. The results show that although the IONP concentrations
in the incubation media do not vary much, the intracellular IONP
Fig. 4. Nanoparticle internalization induces ROS. Relative levels of ROS in C17.2 NPCs levels are significantly different. The VSOPs are only poorly inter-
incubated with the various particles at the concentrations indicated for (A) 4 h or (B) nalized, reaching 18.65  2.07 pg Fe/cell, whereas Resovist and
24 h, in the absence (light grey) or presence of 1 mM desferrioxamine. Values are
Endorem reach levels of 31.99  2.99 and 46.59  4.70 mg Fe/cell.
represented as mean  standard deviation (n ¼ 20). Where appropriate, statistical
significance is indicated: *: p < 0.05; **: p < 0.01; ***: p < 0.001. Cationic MLs display the greatest cellular internalization at non-
toxic levels and finally lead to 67.37  5.98 pg Fe/cell.
a slowing down of cell cycle progression, cell proliferation was
3.7. Effect of safe IONP concentrations on MR contrast
assessed for C17.2 NPCs and found to transiently decrease when
incubated with Resovist and MLs at 600 mg Fe/ml but not at 300 mg
As the amount of intracellular iron was quite variable for the
Fe/ml (Fig. 3).
different IONPs, their effect on MR contrast generation of labelled
cells was investigated, as shown in Fig. 8. The data clearly indicate
3.4. Effects of intracellular IONPs on cell homeostasis that at least 500 cells/ml loaded with VSOPs can are needed for clear
MR detection, whereas Endorem and Resovist require 200 cells/ml
Based on the above-mentioned findings, VSOP was further used and MLs only 50 cells/ml.
at 600 mg Fe/ml, Endorem at 400 mg Fe/ml, whereas MLs and
Resovist were used at 300 mg Fe/ml. To further evaluate the effect of 4. Discussion
the IONPs at these concentrations, the level of ROS produced by the
particles was measured by means of an NBT assay after 4 and 24 h From a historical perspective, IONPs were considered safe, due
(Fig. 4). Citrate-coated VSOP particles led to the highest induction of to the high levels of Fe-ions which could be tolerated (up to 4 mM)
ROS and also reached the fastest maximum (4 h incubation), which and the high LD50 of the original dextran-coated IONPs in first
could lead to toxic effects. The generation of free radicals hints at animal trials [22,23]. Based on the assumption of non-toxic IONPs,
the possible degradation of the particles and the release of ferric these particles were widely used in cell labelling strategies as well,
ions from the IONPs into the acidic endolysosomal compartments. without paying much attention to the possible effects of IONPs on
This could be validated by the use of desferrioxamine, an iron cell homeostasis [24]. Later on, it became more apparent that
chelator, which significantly reduced the induction of ROS levels in IONPs and NPs in general do possess several specific traits which
all setups apart from MLs (Fig. 4). To further verify this, the makes them different from bulk material or free ions, in terms of
expression of transferrin receptor-1 (TfR1), involved in iron surface reactivity and toxic effects [5,25]. Due to these insights, the
metabolism, was investigated as shown in Fig. 5. In line with the number of studies focussing on IONP toxicity has steadily increased
induced generation of ROS, C17.2 NPCs incubated with the different over the last 5 years [24], but due to the wide variability in types of
S.J.H. Soenen et al. / Biomaterials 32 (2011) 195e205 201

A B C
VSOP Endo

256
256

256
Control
(600) (400)

Events
Events

Events
0
0
0

100 101 102 103 104 100 101 102 103 104 100 101 102 103 104
FL1 LOG FL1 LOG FL1 LOG

D E

256
256

ML Reso
(300) (300)
Events

Events
0

0
100 101 102 103 104 100 101 102 103 104
FL1 LOG FL1 LOG

Fig. 5. Iron oxide nanoparticle internalization and degradation affects cellular iron metabolism. Flow cytometric analysis of TfR1 expression in (A) untreated control cells or cells
incubated with (B) VSOPs, (C) Endorem, (D) MLs or (E) Resovist at the indicated concentrations for 24 h. After 1 week of continuous culture post NP-exposure, where cell densities
between the different groups were kept identical, samples were collected and analysed by flow cytometry. Expression levels were normalized using isotype control antibody. Data
are given for 10,000 events per condition.

IONPs or cells used and differences in incubation conditions (IONP The low proliferation rate of VSOP-labelled cells at day one seems,
concentrations varying from a few mg/ml over 1 mg/ml), many at first sight, somewhat contradictory with the above-mentioned
confusing data have been obtained and the need for more infor- reasoning that the effects are all caused by high intracellular
mation is primordial to advance the use of NPs in biomedical concentrations of IONPs. However, unlike the effects seen for
applications [4]. By combining different assays to assess multiple Resovist, MLs and Endorem, where proliferation was minimal after
parameters important for determining IONP safety, the present three days, the effect of VSOPs was limited to the first time point
study aims to determine the safe concentrations of some of the only. This could be ascribed to the acute toxicity of VSOP particles at
most widely used IONPs for cell labelling purposes. the high concentration used [14] and is therefore not the result of
Recently, our group showed that high intracellular concentra- an affected cell signalling.
tions of IONPs affected the actin cytoskeleton which resulted in Here, the high ROS levels and the increases in TfR1 expression for
a diminished cell proliferation [14]. The important role of actin all IONP types were significant, but in terms of toxicity of IONPs it is
fibers in maintaining cell morphology and their effect on signalling yet unclear what the precise effect of these factors is on cell viability
pathways monitoring several parameters such as cell death, as all cells have defence mechanisms which allows them to deal
migration and differentiation is well known [26]. Recently, the with certain levels of oxidative stress [28,29]. Regarding the use of
effect of NPs on focal adhesion-mediated cell signalling pathways IONPs, the term “biocompatibility” is often employed, as being
has also been confirmed by the work of Miller et al. [27] who found a necessary asset of any IONP to be used in a biological context.
that depending on the nature of the NP coating, the NPs could Quite often, IONPs are named biocompatible when no acute cyto-
interact more or less with the extracellular domain of integrins and toxic effect is apparent. In its full definition “biocompatible” should
thereby trigger their activation and intracellular signalling cascade. however signify a particle which, when taken up by the cells, has no
The results obtained in the present study are in line with the effect whatsoever on cell homeostasis meaning that no difference
intracellular concentration-dependent effects as MLs and Resovist should be observable in any cellular parameter before or after
elicited the gravest effects, but were previously found to lead to the labelling. In this context, the induction of high ROS levels, although
highest uptake efficiencies. The results presented here show that perhaps not toxic to the cells as such, could have long-term effects
Resovist and MLs significantly affect cellular morphology and cell and should be considered as a hazard. The induction of ROS, which
cytoskeleton when incubated at 600 mg Fe/ml (Fig. 1). Lowering the could be diminished by the use of an iron chelator, and the alter-
concentration of the IONPs diminished its effect, leading to the ations in TfR1 expression indicate the intracellular degradation of
absence of any effect when incubated at 300 mg Fe/ml. The finding the IONPs, where free ferric ions are generated in endolysosomal
that only MLs and Resovist significantly affected cell morphology structures and shuttled towards the cytoplasmic iron pool. This
was likely due to the fact that these particles are internalized most intracellular degradation of IONPs and the generation of free ferric
avidly and leads to the highest amount of cell-associated iron [17]. ions have recently been described by Huang et al. [30] who noticed
In line with our previous study [14], the effect on actin cytoskeleton an altered expression of protein regulators of the cell cycle which
and tubulin network also resulted in a diminution of focal adhesion was ascribed to Resovist degradation.
formation and maturation as can be seen in Fig. 2. This further The intracellular degradation of IONPs and the associated release
resulted in a transiently decreased proliferative capacity (Fig. 3). of free iron can have profound effects on cell homeostasis, as has
202 S.J.H. Soenen et al. / Biomaterials 32 (2011) 195e205

Fig. 6. Iron oxide nanoparticle internalization and degradation affects cell functionality. (AeJ) Representative micrographs of PC12 cells at 2 days of NGF-exposure, showing
a-tubulin (green) and G-actin (red) staining. Scale bars ¼ 10 mm. Controls cells (A) and non-NGF-exposed control cells (B) are shown as well as cells incubated with VSOP (C,D),
Endorem (E,F), ML (G,H) or Resovist (I,J) at the concentrations indicated for 24 h. Figures KeO show the number of neurites of a certain length per cell for control cells (K), VSOP (L),
Endorem (M), ML (N) or Resovist (O)-treated cells. Data are expressed as mean  standard error to the means. When appropriate, the degree of significance when compared with
untreated controls cells is indicated (*: p < 0.05; **: p < 0.01; ***: p < 0.001).

recently been illustrated in the study by Chen and colleagues [31]. endolysosomal structures of the cell [14]. This degradation had clear
There, it was shown that labelling of human mesenchymal stem cells effects on cell functionality as could be shown by the PC12 cell model
by Resovist resulted in a dose-dependent inhibition of osteogenic system [17]. The use of the PC12 cells as a model system to assess cell
differentiation and altered cell migration. All these effects were functionality offers many advantages as this is a very sensitive cell
suppressed when desferrioxamine, an iron chelator was used, which system which allows to easily compare the effects of different IONPs
emphasized the importance of free iron on the observed effects. In and the data can easily be quantified [18]. In terms of IONP uptake,
a different study, it was shown that various types of IONPs were PC12 cells internalize IONPs less avidly than C17.2 NPCs or hBOECs
prone to degradation at pH-levels which can be found in [17]. This effect could possibly be due to the small size of the cells as
S.J.H. Soenen et al. / Biomaterials 32 (2011) 195e205 203

A 120
4h
it is well known that NP uptake between different cell types can lead
to large variations, but when NP uptake is expressed in terms of cell
24 h surface area available for NP binding, internalization levels become
100
more comparable. This rule does, however, not apply to all cells, as
Relative viability (%)

the C17.2 NPCs internalized similar amounts of IONPs than hBOECs,


80 although the latter are much larger.
The cationic MLs used in this work appear to be the least toxic,
60 despite the inherent toxicity of cationic lipids [32] and resulted in
the highest amount of cell-associated iron. This is in line with
previous data showing a highly reproducible production of these
40
MLs and an optimized liposomal coating suited for cell labelling
[20]. This lipid coat also provides a good protection of the iron oxide
20 core from endolysosomal degradation in contrast with (carboxy)
dextran or citrate coatings [17]. The VSOPs were found to have the
0 gravest effects on cell functionality and viability and could not be
VSOP Endo ML Reso Triton used to reach high intracellular levels of iron oxide. Next to the
possible degradation of the citrate-coat, the small size of the VSOPs
B 90 4h
could also play a very important role in their higher toxicity. As the
radius of the iron oxide core of VSOPs is approximately three times
24 h smaller than the radius of the ML core, it follows that the volume
75 and thus the mass of iron oxide for a single VSOP will be nearly 27
times lower than for a single ML. Although the amount of iron
internalized is 3.6 times less than is the case for MLs, this means
Iron per cell (pg)

60
that in total, still more VSOPs (11.85 times, see Supporting
Information for a calculation of the number of MLs and VSOPs
45 per cell) are internalized. When the size of an NP decreases, the
surface area over volume ratio greatly increases accordingly [33].
30 This suggests that the total surface area of VSOPs available for
interaction with the cellular structures could be larger than for MLs,
but the relative amount of iron oxide present at the surface and
15
thus prone to degradation and cellular interactions over total iron
oxide present in the particles is vastly larger for VSOPs. This could
0 greatly influence toxic effects which have been described to be
VSOP Endo ML Reso dominated by NP surface properties [25]. For Endorem and
Resovist, this comparison cannot be made as these NPs are beads,
Fig. 7. Effects of iron oxide nanoparticles on cell viability and internalization efficiency.
The results of (A) an LDH assay or (B) cellular iron determination for C17.2 NPCs
meaning that several small iron oxide cores are embedded within
incubated with VSOP (200 mg Fe/ml), Endorem (200 mg Fe/ml), ML (250 mg Fe/ml) or a larger matrix. The number of cores and the size of the matrices are
Resovist (150 mg Fe/ml) after 4 and 24 h incubation. For (A), data are normalized quite heterogeneous [34], which hardens the analysis and renders
relative to the values obtained for untreated control cells (¼ 100% viability). For the it impossible to view any effects in terms of NP numbers.
LDH assay, cells treated with 1% Triton X-100 for 15 min prior to the LDH assay were
Of further interest for MRI is the actual size of the iron oxide
used as negative controls. Data are expressed as mean  standard error to the means
(n ¼ 10). core. For a similar amount of iron, larger cores will have a bigger
effect on negative contrast enhancement than smaller ones. For

Fig. 8. The effect of safe IONP concentrations on MR contrast generation. (A) Colour-coded T2*-maps of C17.2 cells labelled with VSOP (200 mg Fe/ml), Endorem (200 mg Fe/ml), ML
(250 mg Fe/ml) or Resovist (150 mg Fe/ml) for 24 h. MR images were acquired for cells placed in an agar tube at a density of 0, 20, 50, 200, 500 or 2000 cells/ml, respectively. (B)
Calculated T2* relaxation times for C17.2 cells incubated with VSOP (200 mg Fe/ml), Endorem (200 mg Fe/ml), ML (250 mg Fe/ml) or Resovist (150 mg Fe/ml) for 24 h. Data are
expressed as mean  standard deviation (n ¼ 3). Where appropriate, the degree of significance with respect to control samples (no cells or particles) is indicated: *: p < 0.05; **:
p < 0.01; ***: p < 0.001.
204 S.J.H. Soenen et al. / Biomaterials 32 (2011) 195e205

maghemite particles with core sizes of 3.5 and 12 nm, the satura- [6] Soenen SJ, Hodenius M, De Cuyper M. Magnetoliposomes: versatile innovative
nanocolloids for use in biotechnology and biomedicine. Nanomedicine
tion magnetization values were determined to be 41 and 83 emu/g
2009;4:177e91.
Fe, respectively [35]. For MLs, with a core size of approximately [7] Wilhelm C, Gazeau F. Universal cell labelling with anionic magnetic nano-
14 nm compared with 4 nm for Endorem, Resovist and VSOP particles. Biomaterials 2008;29:3161e74.
particles, this, next to a higher intracellular iron oxide levels, will [8] Santra S, Kaittanis C, Grimm J, Perez JM. Drug/dye-loaded, multifunctional iron
oxide nanoparticles for combined targeted cancer therapy and dual optical/
further enhance the MR contrast. The beneficial MR properties of magnetic resonance imaging. Small 2009;5:1862e8.
MLs for cell labelling compared with common dextran-coated iron [9] Kostura L, Kraitchman DL, Mackay AM, Pittenger MF, Bulte JW. Feridex
oxide NPs have recently been shown to be enhanced by intracel- labeling of mesenchymal stem cells inhibits chondrogenesis but not adipo-
genesis or osteogenesis. NMR Biomed 2004;17:513e7.
lular clustering on the MLs [17]. Upon exposure to the degradative [10] Arbab AS, Yocum GT, Rad AM, Khakoo AY, Fellowes V, Read EJ, et al. Labeling
lysosomal environment, the outer leaflet of the lipid bilayer is easily of cells with ferumoxides-protamine sulfate complexes does not inhibit
degraded, leaving a monolayer-coated ML which has a hydrophobic function or differentiation capacity of hematopoietic or mesenchymal stem
cells. NMR Biomed 2005;18:553e9.
surface and tends to cluster together, further enhancing transversal [11] Soenen SJ, De Cuyper M. Assessing cytotoxicity of (iron oxide-based) nano-
MR contrast [17]. particles: an overview of different methods exemplified with cationic
magnetoliposomes. Contrast Media Mol Imaging 2009;4:207e19.
[12] Hu F, Neoh K, Cen L, Kang E. Cellular response to magnetic nanoparticles
5. Conclusion “PEGylated” via surface-initiated atom transfer radical polymerization. Bio-
macromolecules 2006;7:809e16.
Assessing the effects of IONPs on cell homeostasis requires [13] Soenen SJ, Illyes E, Vercauteren D, Braeckmans K, Majer Z, De Smedt SC, et al.
The role of nanoparticle concentration-dependent induction of cellular stress
a multidisciplinary approach where many parameters should be in the internalization of non-toxic cationic magnetoliposomes. Biomaterials
analysed as IONPs can lead to a wide variety of toxicological effects. 2009;30:6803e13.
Furthermore, apart from toxicological aspects, the final usefulness [14] Soenen SJ, Nuytten N, De Meyer SF, De Smedt SC, De Cuyper M. High
intracellular iron oxide nanoparticle concentrations affect cellular cyto-
of the IONPs in terms of MR contrast generation of labelled cells is skeleton and focal adhesion kinase-mediated signaling. Small 2010;6:
also evaluated. The results obtained highlight the importance of 832e42.
toxicity assessments in evaluating nanoparticle efficiency as at non- [15] Idée JM, Port M, Raynal I, Schaefer M, Bonnemain B, Prigent P, et al. Super-
paramagnetic nanoparticles of iron oxides for magnetic resonance imaging
toxic concentrations of IONPs, several of them might not be well
applications. In: Kumar C, editor. Nanotechnologies for the life sciences.
suited for their original purpose. The data shown here could Weinheim: Wiley-VCH; 2007. p. 51e84.
provide a good reference point for other researchers who would [16] Arbab AS, Wilson LB, Ashari P, Jordan EK, Lewis BK, Frank JA. A model of
lysosomal metabolism of dextran coated superparamagnetic iron oxide (SPIO)
want to safely label mammalian cells with IONPs or, alternatively,
nanoparticles: implications for cellular magnetic resonance imaging. NMR
could be used as a model system for the further toxicological Biomed 2005;19:383e9.
analysis of IONPs. [17] Soenen SJ, Himmelreich U, Nuytten N, Pisanic 2nd T, Ferrari A, De Cuyper M.
Intracellular nanoparticle coating stability determines nanoparticle
diagnostics efficacy and cell functionality. Small; 2010; doi:10.1002/
Acknowledgements smll.201000763.
[18] Pisanic 2nd TR, Blackwell JD, Shubayev VI, Fiñones RR, Jin S. Nanotoxicity of
iron oxide nanoparticle internalization in growing neurons. Biomaterials
SJHS is a post-doctoral fellow from the FWO-Vlaanderen. M.D.C.
2007;28:2572e81.
and U.H. are recipients of an IWT grant sponsoring project [19] De Cuyper M, Soenen SJ. Cationic magnetoliposomes. Methods Mol Biol
SBO80017. U.H. received financial support by the European 2010;605:97e111.
Commission for EC-FP6 network DiMI (LSHB-CT-2005-512146), [20] Meijering E, Jacob M, Sarria J-C, Steiner P, Hirling H, Unser M. Design and
validation of a tool for neurite tracing and analysis in fluorescence microscopy
EC-FP7 network ENCITE (2008-201842), and the K.U. Leuven images. Cytometry Part A 2004;58:167e76.
Centers of Excellence ‘MoSAIC’. [21] Soenen SJH, Vercauteren D, Braeckmans K, Noppe W, De Smedt S, De
Cuyper M. Stable long-term intracellular labelling with fluorescently tagged
cationic magnetoliposomes. Chembiochem 2009;10:257e67.
Appendix. Supplementary data [22] Küstermann E, Himmelreich U, Kandal K, Geelen T, Ketkar A, Wiedermann D,
et al. Efficient stem cell labeling for MRI studies. Contrast Media Mol Imaging
Supplementary data associated with this article can be found in 2008;3:27e37.
[23] Weissleder R, Stark DD, Engelstad BL, Bacon BR, Compton CC, White DL, et al.
the on-line version, at doi:10.1016/j.biomaterials.2010.08.075. Superparamagnetic iron oxide e pharmacokinetics and toxicity. Am J Roent-
genol 1989;149:1161e5.
[24] Pisanic 2nd TR, Jin S, Shubayev VI. Iron oxide magnetic nanoparticle
Appendix
nanotoxicity: incidence and mechanisms. In: Sahu S, Casciano D, editors.
Nanotoxicity: in vivo and in vitro models to health risk. Chichester: John
Figure with essential color discrimination. Most of the figures in Wiley & Sons, Ltd; 2009. p. 397e425.
this article are difficult to interpret in black and white. The full color [25] Colvin V. Correlating the physical and chemical properties of nanoparticles to
their biological activity. Environ Mol Mutagen 2007;48:533.
images can be found in the on-line version, at doi:10.1016/j. [26] Neuhuber B, Gallo G, Howard L, Kostura L, Mackay A, Fischer I. Reevaluation of
biomaterials.2010.08.075. in vitro differentiation protocols for bone marrow stromal cells: disruption of
actin cytoskeleton induces rapid morphological changes and mimics neuronal
phenotype. J Neurosci Res 2004;77:192e204.
References [27] Miller IS, Lynch I, Dowling D, Dawson KA, Gallagher WM. Surface-induced cell
signaling events control actin rearrangements and motility. J Biomed Mater
[1] Duguet E, Vasseur S, Mornet S, Devoisselle JM. Magnetic nanoparticles and Res A 2010;93:493e504.
their applications in medicine. Nanomedicine 2006;1:157e68. [28] Díaz B, Sanchez-Espinel C, Arruebo M, Faro J, de Miguel E, de Magadán S, et al.
[2] Gupta AK, Naregalkar RR, Vaidya VD, Gupta M. Recent advances on surface Assessing methods for blood cell cytotoxic responses to inorganic nano-
engineering of magnetic iron oxide nanoparticles and their biomedical particles and nanoparticle aggregates. Small 2008;4:2025e34.
applications. Nanomedicine 2007;2:23e39. [29] Pawelczyk E, Arbab AS, Pandit S, Hu E, Frank JA. Expression of transferrin
[3] US National Institutes of Health, University of Edinburgh. Cell tracking using receptor and ferritin following ferumoxides-protamine sulfate labeling of
superparamagnetic particles of iron oxide (SPIO) and magnetic resonance cells: implications for cellular magnetic resonance imaging. NMR Biomed
imaging (MRI) e a pilot study. Available from URL:. UK: University of 2006;19:581e92.
Edinburgh http://clinicaltrials.gov/archive/NCT00972946/2009_09_08; 2009. [30] Huang D-M, Hsiao J-K, Chen Y-C, Chien L-Y, Yao M, Chen Y-K, et al.
Online. The promotion of human mesenchymal stem cell proliferation by
[4] L’azou B, Jorly J, On D, Sellier E, Moisan F, Fleury-Feith J, et al. In vitro effects of superparamagnetic iron oxide nanoparticles. Biomaterials 2009;30:
nanoparticles on renal cells. Part Fibre Toxicol 2008;5:22. 3645e51.
[5] Verma A, Stellacci F. Effect of surface properties on nanoparticle-cell inter- [31] Chen YC, Hsiao JK, Liu HM, Lai IY, Yao M, Hsu SC, et al. The inhibitory
actions. Small 2010;6:12e21. effect of superparamagnetic iron oxide nanoparticle (ferucarbotran) on
S.J.H. Soenen et al. / Biomaterials 32 (2011) 195e205 205

osteogenic differentiation and its signaling mechanism in human mesen- [34] Cengelli F, Maysinger D, Tschudi-Monnet F, Montet X, Corot C, Petri-Fink A,
chymal stem cells. Toxicol Appl Pharmacol 2010;245:272e9. et al. Interaction of functionalized superparamagnetic iron oxide nano-
[32] Soenen SJH, Brisson A, De Cuyper M. Addressing the problem of cationic particles with brain structures. J Pharmacol Exp Ther 2006;318:108e16.
lipid-mediated toxicity: the magnetoliposome model. Biomaterials 2009;30: [35] Miguel OB, Gossuin Y, Morales MP, Gillis P, Muller RN, Veintemillas-
3691e701. Verdaguer S. Comparative analysis of the 1H NMR relaxation enhancement
[33] Chan AT, Lewis JA. Size ratio effects on interparticle interactions and phase produced by iron oxide and core-shell iron-iron oxide nanoparticles. Magn
behavior of microsphere-nanoparticle mixtures. Langmuir 2008;24:11399e405. Reson Imaging 2007;25:1437e41.

You might also like