Adverse effects of titanium dioxide nanoparticles on human dermal fibroblasts and how to protect cells.

Abstract

The effects of exposure of human dermal fibroblasts to rutile and anatase TiO(2) nanoparticles are reported. These particles can impair cell function, with the latter being more potent at producing damage. The exposure to nanoparticles decreases cell area, cell proliferation, mobility, and ability to contract collagen. Individual particles are shown to penetrate easily through the cell membrane in the absence of endocytosis, while some endocytosis is observed for larger particle clusters. Once inside, the particles are sequestered in vesicles, which continue to fill up with increasing incubation time till they rupture. Particles coated with a dense grafted polymer brush are also tested, and, using flow cytometry, are shown to prevent adherence to the cell membrane and hence penetration of the cell, which effectively decreases reactive oxygen species (ROS) formation and protects cells, even in the absence of light exposure. Considering the broad applications of these nanoparticles in personal health care products, the functionalized polymer coating can potentially play an important role in protecting cells and tissue from damage.

Full text

Page 1

Nanoparticle toxicity
Adverse Effects of Titanium Dioxide Nanoparticles on
Human Dermal Fibroblasts and How to Protect Cells
Zhi Pan, Wilson Lee, Lenny Slutsky, Richard A. F. Clark, Nadine Pernodet, and
Miriam H. Rafailovich*
The effects of exposure of human dermal fibroblasts to rutile and anatase
TiO2nanoparticles are reported. These particles can impair cell function,
with the latter being more potent at producing damage. The exposure to
nanoparticles decreases cell area, cell proliferation, mobility, and ability to
contract collagen. Individual particles are shown to penetrate easily through
the cell membrane in the absence of endocytosis, while some endocytosis is
observed for larger particle clusters. Once inside, the particles are
sequestered in vesicles, which continue to fill up with increasing incubation
time till they rupture. Particles coated with a dense grafted polymer brush
are also tested, and, usingflow cytometry, are shown to prevent adherence to
the cell membrane and hence penetration of the cell, which effectively
decreases reactive oxygen species (ROS) formation and protects cells, even
in the absence of light exposure. Considering the broad applications of these
nanoparticles in personal health care products, the functionalized polymer
coating can potentially play an important role in protecting cells and tissue
from damage.
1. Introduction
Due to their unique size-related properties such as high
quantum yield and large surface area to volume ratio,
nanomaterials have shown emerging applications in electro-
nics, biology, and medicine.[1]As a result, the biomedical use
of nanomaterials is being explored in broad areas such as
nanoimaging, drug delivery, biosensors, and cancer photo-
therapy.[2]However, size effects of nanomaterials may also
cause higher toxicity because of their larger surface area,
enhanced chemical reactivity, and easier cell penetration.[3]
Recently, it has been shown that scale alone could have
adverse effects on cell function.[4]Pernodet et al.[5]have
shown that the presence of Au particles, which are considered
an inert material for living cells, was sufficient to induce
aberrant actin formation and interfere with cell migration.
Hence it is to be expected that when the particles are
chemically active, that size effects can also exacerbate the
adverse effects. Actually, for those materials possessing
nanostructure-dependentproperties,
mechanical, electrical, optical, magnetic, and biological
properties, which make them more important for commercial
use, the same properties may potentially lead to adverse
effects that differ from the bulk properties of these materials.
Therefore, the safety issue of nanomaterials, especially those
used for consumer applications, has attracted increasing
attention.[6]
Here, we focused on TiO2, which is a naturally occurring
mineral and known to be biologically inert and nontoxic at the
microscale in both humans and animals.[7]TiO2has been
widely used in industry as an additive for products ranging
from paint, to food colorant, to cosmetics and drugs. The most
common use of TiO2is as an additive in cosmetics where its
high absorbance for UV radiation is used to increase the sun
such aschemical,
Adverse Effects of TiO2 Nanoparticles on Human Dermal Fibroblasts
[?] Prof. M. H. Rafailovich, Z. Pan, Dr. W. Lee, L. Slutsky,
Prof. N. Pernodet
Materials Science and Engineering
State University of New York at Stony Brook
Stony Brook, NY 11794 (USA)
E-mail: miriam.rafailovich@sunysb.edu
Prof. R. A. F. Clark
Biomedical Engineering, Dermatology and Medicine
State University of New York at Stony Brook
Stony Brook, NY 11794 (USA)
DOI: 10.1002/smll.200800798
Keywords:
? cells
? cytotoxicity
? fibroblasts
? nanoparticles
? titanium dioxide
small 2009, 5, No. 4, 511–520
? 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
511

Page 2

protection factors (SPFs) of sunscreens. The same property
also makes TiO2an important component for food packaging
where it is added to prevent damage from UV radiation and
increase shelf life. However, once it absorbs UV light, TiO2, as
a well-known photocatalyst, also catalyzes the generation of
reaction oxygen species (ROS). This property makes TiO2a
favorite component for antibacterial applications,[8]as well a
major component in solar power cells, where sunlight is
converted to electricity.[9]As pointed out by Dunford et al.,[10]
the same property can also make TiO2a potentially hazardous
material since ROS products in turn are known to cause
genetic damage and other adverse effects in living tissues.[11]
The ability of TiO2to photocatalyze the production of ROS
products is further enhanced when the particles are ground to
nanometer dimensions, thereby increasing the surface area
and the electron flux. For example, it was shown that they can
cause inflammatory reactions in both animals and humans
when ground into particles smaller than 20nm.[12]
Many studies reported that cells cultured in the presence of
TiO2nanoparticles showed a dramatic decrease in growth rate
with exposure to concentrations larger than 0.1mg mL?1.[13]
Most of these reports directly correlated the cytotoxicity of TiO2
nanoparticles to induced oxidative damage.[14]But, since most
of the experiments were conducted in the dark, the source of
the oxidative process was not immediately apparent. In order
to understand the mechanism in more detail, and especially to
determine if something could be done to prevent the damage,
wehaveperformed invitroexperimentswithdifferent types of
TiO2nanoparticles. And since most of the previous experi-
ments were performed with transformed cell lines that have
been shown to be more sensitive to
nanoparticle damage than normal primary
cells, we chose to study the phenomena with
human dermal fibroblasts from primary
cultures. The skin is the primary barrier
against contact penetration in cosmetic
products and thus it was important to
understand the effects that are specific to
these cells with different particles that are
commonly encountered in cosmetic formu-
lations. In particular, Lee et al. recently
showed that it was possible to graft a dense
polymercoatingontotheTiO2particlesthat
can trap the photoelectrons and suppress
ROS production.[15]We therefore per-
formed a detailed study in order to
determine how these particles also behaved
in vitro when they came in contact with live
cell cultures.
2. Results and Discussion
2.1. TiO2Nanoparticle
Characterization
In Figure 1 we show scanning electron
microscopy (SEM) and transmission elec-
tron microscopy (TEM) images of the TiO2
particles that we used. The uncoated rutile
particles tend to agglomerate into large spheres several
micrometers in diameter. From Figure 1a it is clear that the
spheres are composed of smaller particles. In order to image
them directly, we dispersed the spheres in ethanol and then
spread a droplet of the solution onto a carbon-coated TEM
grid. The TEM image is shown in Figure 1b, where we see that
the particles are oblate, but of uniform size and shape. The
average width of the particles is 15.0?3.5nm, and they are
elongated with an aspect ratio of ?3.92. In commercial
products, the aggregates are difficult to compound in
formulations, and hence micrometer-sized talc is frequently
added to act as a dispersant. The talc particles are hard and
angular, which allows them to break the aggregates and
disperse the particles more uniformly. This can be seen in the
SEM image of anatase TiO2particles (Figure 1c). Figure 1d is
the TEM image of these anatase TiO2particles with talc,
where it is clearer that they have different shapes from rutile
particles and much larger sizes, 200?13nm. The crystal-
lographic structures of these two different types of TiO2
particles were confirmed by X-ray diffraction (XRD).
These particles, in either rutile or anatase phase, are
commonly used in consumer products. Hence in this paper, we
compared both ultra-fine rutile TiO2 (US Cosmetics)
nanoparticles and dispersed anatase TiO2 (US Cosmetics)
nanoparticles, both of which are regarded as a biocompatible
material in the absence of photoactivation. Although it is
generally thought that smaller nanoparticles penetrate cells
more easily and induce higher damages, we found that phase is
a more crucial factor than size or shape, which is consistent
with previous reports.[13]
full papers
M. H. Rafailovich et al.
Figure 1. Microscopy images of different types of TiO2particles used in this study. a) SEM
images of microscale clusters of rutile TiO2particles. b) TEM images of rutile TiO2nano-
particles and particle size distribution. c) SEM images of clusters of anatase TiO2particles
with talc particles used as dispersant (inset: magnified image of an anatase cluster showing
individual particles). d) TEM images of anatase TiO2nanoparticles and particle size distri-
bution.
512
www.small-journal.com
? 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 4, 511–520

Page 3

In this study, we also noticed that the TiO2particles are
naturally fluorescent with a broad band excitation in the range
from UV to optical. The fluorescence spectrum is also very
broad with no apparent peak in intensity in the observed
region. On one hand, the effect can perturb studies where one
is counting specific florescence from a particular stain, such as
measurements detecting apoptosis, metabolism, and ROS. On
the other hand, if we do not use staining, this can be a useful
tool for detecting nanoparticles associated with cells.
2.2. Rutile TiO2Nanoparticles
In order to determine the effects of nanoparticle exposure
on cell conformation we incubated human dermal fibroblasts
with therutileparticlesshown inFigure 1aandb. Theconfocal
images of cells with stained actin (green) and nuclei (red) are
shown in Figure 2a, from which we can see that after 6 days of
incubation, the introduction of rutile TiO2particles caused a
large decrease in the cell area as compared to the control
sample. Closer examination of the image also indicates that
the morphology of the cells is affected. The control cells are
triangulated and well-spread on the surface, while the cells
exposed to TiO2have become elongated and appear to detach
from the surface. This is also shown graphically in Figure 2b
where we plot the cell area averaged over several hundred
cells that were incubated with 0.4 and 0.8mg mL?1of particles
for 6 days. In Figure 2b we can see that the cell area of the
experimental samples is on average 1/4 and 1/5, respectively,
that of the control sample incubated at the same time without
particles. In Figure 2a, we can also see that in analogy with the
findings of Pernodet et al. on the same cell model,[5]the
decrease in area is also associated with a change in the actin
fibers, where they become thinner and less extended.
Since actin is implicated in many cell functions and
behaviors, such as cell division, we also measured the
proliferation at these particle concentrations. The cell numbers
after 6 days of incubation are plotted in Figure 2b. From the
figure we can see that the cell counts are drastically reduced,
and the cells, although still alive, have nearly stopped dividing.
Most previous toxicity studies have focused on cell death,[4]
but here we show that for low concentration the cells can
survive, but the question of their ability to perform essential
functions arises.
Dermal fibroblasts play a central role in the wound healing
process. In the initial process, they migrate into the wound site
and tissue repair begins when they contract the collagen fibers
so that keratinocytes can differentiate and form the sub-
sequent layers of skin tissue. These functions are composed of
two separate processes: cell migration and collagen contrac-
tion, both of which depend on the ability of the cells to exert
traction forces on the substrate. In order to study whether
nanoparticles influence cell function, we first exposed cells to
0.4mgmL?1TiO2nanoparticlesfor2days,andthencompared
their functions with normal control cells. We first measured
the effect on traction forces exerted by the cells using
techniques described previously.[16]We used a synthetic
hydrogel composed of crosslinked, functio-
nalizedhyaluronicacid(HA)andfibronectin-
functional domains (FNfds) as the flexible
substrateimpregnated
beads.Thedigitalimagespeckle correlation
(DISC) technique combined with the finite
element method (FEM) was then used to
measure the traction stress intensity and
distribution on the surfaces. The results are
shown in Figure 3a for a representative cell.
We can see that in the absence of
nanoparticles, the cells are well-spread on
the hydrogel, and robust traction forces are
exerted along the periphery. The hydrogel
used here was relatively stiff, and good cell
adhesion and triangulation was previously
reported.[17]When the nanoparticles are
added, even though the cells prefer these
surfaces, the cell area is reduced and the
cellsbecomeelongated.Elongationcanalso
be associated with cell migration. This was
previously described by Ghosh et al. when
cells were plated on softer surfaces where
large traction forces only appeared behind
the leading edge and cells migrated fas-
ter.[16]
Here we see that despite the
elongated appearance, only weak traction
forces are exerted.
Cell migration was then measured in a
separate set of experiments where an
agarose droplet cell migration assay was
withfluorescent
Adverse Effects of TiO2Nanoparticles on Human Dermal Fibroblasts
Figure 2. a) Confocal images of human dermal fibroblasts incubated for 6 days with different
concentrations of rutile TiO2nanoparticles. b) Cell area and number as a function of
nanoparticle concentration.
small 2009, 5, No. 4, 511–520
? 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
513

Page 4

done. Here we defined the migration area as the difference
between the total area that cells covered after 18h of
migration and the area of the agarose droplet. The results
are shown in Figure 3b. In the inset phase images, the black
rings delineate the periphery of the migrating circles of cells
and the white rings show the area of agarose droplets, where
we can clearly see that the migration is far smaller for cells
exposed to 0.4mg mL?1TiO2than for the unexposed cells.
Using the migration area to evaluate the cell mobility, we can
see that the TiO2exposure reduced migration by ?59%.
The ability of cells to contract the collagen matrix was also
studied. The same number of normal cells and cells incubated
with 0.4mg mL?1TiO2particles were cultured in collagen gels
for 4h. The optical images of the gels are shown in the inset of
Figure 3c, where we can clearly see that the collagen gels
contacted by the normal cells are significantly smaller than
those contracted by cells exposed to TiO2. The collagen
contraction ratios averaged over three samples are plotted in
Figure 3c. The contraction of cells exposed to TiO2was less
than 80% of the control sample.
Since the TiO2 particles can be excited with 488nm
radiation and their emission spectrum is wide, the particles are
also apparent in the confocal microscope images, especially
those obtained using Hg lamp florescence. Due to the
resolution of microscopy, the fluorescence observed is from
large particle clusters. In order to visualize how the particles
entered the cell and where they were sequestered, we
performed electron microscopy on cells incubated with
0.4mg mL?1rutile TiO2nanoparticles. We first imaged cells
incubated with the TiO2nanoparticles for 2 days, the same
time scale used for migration and contraction studies. A
typical image is shown in Figure 4a. In these images, we can
clearly see the TiO2particles, which are more electron dense
than the surrounding cell environment.
After 2 days, even though the cells seem
tohaveingestedarelativelylargeamountof
particles, we know from the previous
experiments that they have still retained
the ability to perform most functions. Most
of the particles seem to be confined inside
vesicles that are distributed across the
cytoplasm, but which do not cross into the
nucleus. A higher magnification image of
the vesicles was also obtained and is shown
in Figure 4b. Here we see the particles that
comprise the clusters surrounded by the
vesicle membrane. In a previous study,
Pernodet et al. also showed Au nanoparti-
cles capsulated in single vesicles of dermal
fibroblasts.[5]In that case, the vesicles were
only partially filled with particles and
the membranes of the vesicles were intact.
Here it appears that many more particles
are stuffed into each vesicle, thereby
possibly leading to rupture. The boundary
with the nuclear membrane is also visible in
Figure 4b. Here we see that one of vesicles
that is close to the nucleus does not have an
intact enclosure, and the particles have
begun to leak out. Several individual particles shown in the
circle might be able to cross the nuclear membrane. Another
instance is shown in Figure 4c, where we can see a vesicle
containing a very large cluster of particles impinging upon the
nucleus. The vesicle membrane surrounding the cluster
appears ruptured; in the upper corner a segment of particles
has pierced through the membrane, and in the back portion
facing away from the nucleus the membrane is completely
frayed. The nuclear membrane is distorted, but remains
distinct from the vesicle membrane and is intact (dashed line).
A small particle cluster is seen to have become lodged inside
the membrane (indicated by an arrow), but has not penetrated
into the nucleus.
Inordertodeterminehowtheparticlesenteredthecell,we
also performed a series of TEM cross sections corresponding
to a time-sequence incubation (Figure 5a). We can see that
full papers
M. H. Rafailovich et al.
Figure 3. Cellswere incubatedwith0.4mgmL?1rutileTiO2nanoparticlesfor 2days andthen
assayedforthefollowingfunctions:a)tractionstressesexertedbycellsonahydrogelafter6h
of incubation; b) cell migration area out of an agarose droplet after 18h of incubation; and
c) contraction of collagen gels after 4h of incubation.
Figure 4. TEM images of cells incubated with 0.4mg mL?1rutile TiO2
nanoparticles for 2 days. a) A typical cell with nanoparticle clusters
confined in vesicles. b) Vesicles filled with nanoparticles (the red circle
shows individual particles leaking out from the vesicles). c) A vesicle
with a ruptured membrane (the dashed shows the nuclear membrane
and the arrow shows a small particle cluster lodged inside the nuclear
membrane).
514
www.small-journal.com
? 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 4, 511–520

Page 5

after30min, someparticles are attached tothe cell membrane,
but very few have penetrated inside the cells. A closer
examination of the particles on the membrane indicates that
some are lodged on the membrane, but no evidence of
endocytosis is observed for the isolated particles. Larger
clusters can also be seen on the membrane, but no evidence of
penetration as clusters is observed. A closer examination of
the images shows individual particles are
already lodged inside large otherwise
empty vesicles. In Figure 5b we plot the
mean size of both vesicles and clusters as a
function of incubation time. We find that at
30min, the size of the vesicles is fairly
uniform with a mean diameter of ?286nm,
which is much larger than the individual
particle size. After 1h, larger clusters
are seen to form inside the vesicles,
and individual particles are observed at
the membrane. With times up to 5h, the
amount of individual particles at the
membrane increases, and clusters are also
observed with the indented cell membrane,
which is in the process of endocytosis. At
the same time, the size of the clusters in the
vesicleincreasesdramatically.
Figure 5b we can see that the cluster size
approaches ?77% of the mean vesicle size
From
after 5h, and both the vesicle size and the
cluster size do not increase significantly
after that time. We can therefore define an
occupancy factor as the ratio of cluster to
vesicle size. We then plot occupancy factor
versus incubation time in Figure 5c where
weseethatitincreases rapidlyinthefirst5h
and then levels off in the next 43h. In this
case, as more particles enter the cells, more
vesicles are created and then filled. The
clustersizeislimitedbythevesiclesize.Any
further increase in cluster size probably
leads to the rupture of the vesicle. This can
be seen in the TEM images of the particle
clusters shown in Figure 4c, where the size
of the vesicle filled with clusters is almost
ten times the average size.
2.3. Anatase TiO2Nanoparticles
Anatase TiO2is known to be even more
photoactive than rutile. In Figure 6a we
show confocal images of cells incubated
with anatase TiO2 particles at different
concentrations for 3 days. Compared with
Figure 2 we can see that anatase particles
with even lower concentrations and shorter
incubation times can produce far more
damage to the cells than rutile particles. A
closer examination indicated that even the
cells incubated with 0.1mg mL?1of anatase
particles had broken actin fibers and
nanoparticles surrounding the nuclear membrane. Cells
incubated at the higher concentrations had nearly no visible
actin fibers, and the membranes appeared to have ruptured.
To further confirm that the damage to the actin cytoskeleton
was caused by anatase TiO2 nanoparticles rather than
microscale talc particles, another control sample with an
equalamountoftalcwasexamined,wherewecanseethatcells
Adverse Effects of TiO2Nanoparticles on Human Dermal Fibroblasts
Figure 5. A time sequence study of cells incubated with 0.4mg mL?1rutile TiO2nanopar-
ticles. a) TEM images. b) Vesicles size and particle cluster size. c) Occupancy factor as a
function of incubation time.
Figure 6. a) Confocal images of cells incubated with different concentrations of anatase TiO2
nanoparticles for 3 days. b) Cell number as a function of nanoparticle concentration.
small 2009, 5, No. 4, 511–520
? 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
515

Page 6

cultured in media with talc still grew well after 3 days and
showed clear and strong actin fibers. These results are
consistent with the survival cell number after being cultured
for 7 days in the presence of anatase TiO2nanoparticles at
different concentrations. As shown in the bar graph of
Figure 6b, only cells incubated with 0.1mg mL?1or less
survived and were able to proliferate. Cells cultured with a
concentration of 0.3mg mL?1or higher failed to grow. Cell
numbers were significantly decreased after exposure to high
concentrations of anatase TiO2nanoparticles. The dramatic
damage induced by anatase TiO2nanoparticles to actin was
confirmedbyWesternblot.Comparedtotheadverseinfluence
of rutile particles, anatase particles show a dramatic effect on
thecellactinafter2daysofincubation(Figure7),whichwould
severely impair cell function.
Figure 8 shows TEM images of microtomed sections from
cellsexposedto0.4mgmL?1anataseTiO2nanoparticles.From
Figure8awecanseethattheanataseparticlescausedhugeholes
to form in the cell cytoplasm, which may be the remnants of
vesicles that were filled with particles. Although more damage
wasfoundoutsideofthenuclei,wealsonoticedsomevesiculated
nuclei(Figure8b).Thisraisesmoreissues,suchaswhetherthese
nanoparticles could directly harm DNA molecules and induce
furtherdamageatthegenelevel.Previous
studiesaboutthe effectsof TiO2nanopar-
ticles on DNA have already shown that
DNA would be partially decomposed
when exposed to UV radiation.[18]Here,
the TEM images confirmed that this
phenomenon cannot be
because it is possible for nanoparticles to
get into the nuclei of normal tissue cells
under certain conditions.
neglected,
2.4. Coated TiO2Nanoparticles
The first step in the penetration
process appears to be adhesion of the
particles to the cell membrane. Hence we proposed that if the
adhesion mechanism was somehow impeded, the penetration
of the particles into the cells would be greatly reduced. Lee
et al. have recently shown that it was possible to chemically
graft a densely charged polymer layer onto the TiO2particles
using sonochemical methods.[15]This layer was able to trap
electrons emitted by the particles and form a dense polymer
brush with an approximate grafting density of one chain per
0.6nm. It is known that extremely stretched brushes do not
adhere to surfaces, since entropic hindrance prevents further
distortion of the brush, resulting in hard core repulsion. We
therefore postulated that this effect may also prevent the
coated particles from adhering to the cell surface membranes.
In order to probe this hypothesis, we first coated the rutile
TiO2using the technique of Lee et al.[15]The SEM and TEM
micrographs of coated rutile TiO2nanoparticles are shown in
full papers
M. H. Rafailovich et al.
Figure 7. Western blot results of actin from the same number of cells
incubated with 0.4mg mL?1rutile or anatase TiO2nanoparticles for
2 days compared with those exposed to no particles or just talc
particles at the same concentration.
Figure 8. TEMimagesofcellsincubatedwith0.4mgmL?1anataseTiO2
nanoparticles for 2 days. a) Typical cells with huge holes induced by
anatase particles. b) Most damage is distributed in the cytoplasm (left)
and some vesiculated nuclei are also observed (right).
Figure 9. MicroscopyimagesofcoatedrutileTiO2particles. SEMimages(a) andTEM images(b)
of coated rutile TiO2nanoparticles and particle size distribution.
516
www.small-journal.com
? 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 4, 511–520

Page 7

Figure 9. We then added different con-
centrations of the coated rutile TiO2
nanoparticles to cell culture media and
compared the data with that obtained from
samples containing equal concentrations of
thenon-coatedrutile
Figure 2b we have shown that cells
incubated with non-coated rutile TiO2
particles have a much slower proliferation
rate than the control sample. Here, cells
incubated with coated particles are almost
indistinguishable from the control samples
for all times. In Figure 10a, cell counts after
11 days of incubation are still comparable
to the control. Corresponding Hg lamp
fluorescent images of the cells demonstrate
particles in the media, but they do not
appear to be in or on the cells (Figure 10b).
Anothermethod
amount of TiO2particles adhered to the
cells is flow cytometry. Since TiO2particles
have a natural florescence, no additional
staining is required. The results are shown
in Figure 11 for cells incubated with
0.4mg mL?1rutile, anatase, and coated
rutile particles for 2 days. The largest
florescence from all the particles was obtained with an
excitation wavelength of 635nm. Since the cells were washed
many times to remove all the floating particles, the emission
could only come from particles attached to the cells. In
Figure 11, distinct peaks of high fluorescent intensity were
observed with cells incubated with uncoated rutile or anatase
particles, but not from cells incubated with coated TiO2. These
latter cells had an emission spectrum
almost identical to the control, which
consisted of cells that were not incubated
with particles. If we assume that the
average intensity is correlated to the
number of particles associated with the
cells, we can see that the anatase sample
has the largest number. All these are
consistent with optical micrographs of the
samples prior to injection into flow cyto-
metry (inset in Figure 11), where the
particle uptake is significantly larger for
cells incubated with anatase particles than
with rutile ones, while nearly no uptake is
observed for cells incubated with the
coated rutile particles. Since we do not
know the precise difference in intensity
between the two phases of TiO2particles,
we cannot at the moment use this techni-
que to quantify the amount of particles in
the cell. But compared with the control,
which does not contain any particles and
corresponds to the system background, we
can still obtain a clear idea of the presence
of different types of TiO2particles adhered
tothecells.ThisisconsistentwithFigure10
particles.From
tocomparethe
where we do not see coated TiO2particles inside the cells, and
now we can also conclude that very few coated particles were
adhered to the outside of the cells.
Since nanoparticles are also known to induce oxidative
stress, we assayed the ROS production, using H2O2, of cells
incubated under the same conditions as those used in the flow
cytometry test. H2O2produced by the cells incubated with
Adverse Effects of TiO2Nanoparticles on Human Dermal Fibroblasts
Figure 10. a) Cell numbers after being incubated with different concentrations of coated TiO2
nanoparticles for 11 days. b) Hg lamp fluorescent images of cells incubated with coated TiO2
nanoparticles.
Figure 11. Flow cytometry results showing different particle uptake of cells after being
incubated with 0.4mg mL?1rutile, anatase, and coated TiO2nanoparticles for 2 days. The
phase contrast images of cells with attached particles were taken prior to the flow cytometry
analysis.
small 2009, 5, No. 4, 511–520
? 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
517

Page 8

coatedrutileparticleswascomparabletothecontrolsampleas
shown in Figure 12. The H2O2 levels were significantly
elevated by ?24% and ?70% for the cells incubated with
rutile and anatase nanoparticles, respectively, which are
consistent with the particle uptake and the damages induced
by these particles. Since all the studies here are performed
without light exposure, these results indicate that nanoparticle
attachment and penetration may be the reason cells produce
oxidative stress and induce ROS production.
3. Conclusion
In this study, we showed that both rutile and anatase TiO2
nanoparticles impair cell function, with the latter being more
potent at producing damage. Exposure to nanoparticles
decreases cell area, proliferation, mobility, and ability to
contract collagen. These TiO2 nanoparticles can easily
penetrate through cell membranes and are sequestered inside
vesicles. In time vesicles fill up and eventually rupture.
Particles that werecoated with a polymer brushdid not adhere
to cell membranes and hence did not penetrate cells, which
decreased ROS formation and thus allowed normal cell
function. Considering the broad applications of these
nanoparticles in personal health care products, the functio-
nalized polymer coating can potentially play an important role
in protecting cells and tissue from damage.
4. Experimental Section
TiO2particle characterization: Dry TiO2powder was used as a
SEM sample to evaluate the particle size, shape, and suspended
status. In order to image individual particles clearly, the powder
was dispersed in ethanol and spread on a carbon-coated TEM
grid. The crystallographic structures of these two different types of
TiO2 particles were confirmed with XRD. Additionally, different
types of TiO2particles were suspended in phosphate buffer saline
(PBS) and loaded in a 96-well dish prior to the fluorescence
measurement, which was performed with a FLEXstationTMbench-
top scanning fluorometer (Molecular Devices).
Cell culture and function studies: Primary dermal fibroblasts
obtained from a 31-year-old Caucasian female (Clonetics) were
used between passages 5 and 13. The cells were routinely
cultured in full-Dulbecco’s Modified Eagle Medium (DMEM)
supplemented with fetal bovine serum (10%) and penicillin,
streptomycin, and L-glutamine (P/S/G, 1%), in a 37-C, 5% CO2,
95% humidity incubator (Napco Scientific). TiO2 media was
prepared by dispersing the nanoparticles in the same full-DMEM
at different concentrations. Before use, the TiO2 media was
sonicated for ?30min and stirred for at least 5min to achieve a
homogeneous distribution. The experimental samples were
prepared by culturing cells in TiO2media at different concentra-
tions for a certain incubation time, with the media changed every
2–3 days.
Immunofluorescence images of the cells were taken with a
40? oil lens on a Leica TCS SP2 laser scanning confocal
microscope (Leica Microsystems) after the cells were stained by
alexa fluor 488 phalloidin for actin and propidium iodide for
nuclei. With MetaMorph software (Universal Imaging), the area of
the cells on these confocal images can be readily measured.
For cell counting, cells were seeded in 24-well dishes with
2500cells/well at the beginning. After being cultured with full-
DMEM or TiO2media for a certain incubation time, cells were
rinsed three times with PBS to remove any suspended particles
and dead cells. Then they were detached with trypsin and counted
with a hemacytometer. Each experimental group contained at
least three replicates to obtain the average cell number under
each condition.
Quantification of cellular traction forces was achieved with the
DISC technique combined with FEM as previously reported.[16]
After 2 days exposure to the media with or without TiO2
nanoparticles, the experimental and control experimental cells
were reseeded onto HA/FNfd hydrogels with fluorescent beads in
DMEM and kept in a 37-C incubator for 6h. The images before
and after cell detachment were recorded with a 63?, NA 0.9 water
objective on a Leica TCS SP2 laser scanning confocal microscope
(Leica Microsystems). The deformations were quantified with the
DISC technique and the cellular traction stresses were calculated
with FEM.
Cell migration was evaluated with an agarose droplet assay.
The agarose gel was prepared by melting agarose stock solution
(2% w/v) and diluting it with DMEM at a 1:9 ratio. Then the diluted
agarose (0.2% w/v) was added to the cell pellets and the cells
were resuspended to 1.5?107cells mLS1. Droplets (1.25mL) of
such cell suspension were carefully loaded on the prepared HA/
FNfd hydrogel surfaces in a 24-well dish. After that, the whole dish
was placed at 4-C for 20min to allow the agarose droplet to gel
before DMEM (400mL) with platelet derived growth factor (PDGF)
(30ng mLS1) was added into each well. After 18h of incubation at
37-C, the cells were fixed with formaldehyde (3.7%), stained with
crystal violet (0.1%), and rinsed three times with PBS for easy
visualization under a stereomicroscope. To quantify cell migration,
the area of agarose droplets and the area covered by the leading
edges of the out-migrating cells were measured with SPOTCam
software. The cell migration area was defined as the area of the
outward cell migration minus the area of agarose droplets.
Collagen gel contraction is typically used to mimic the process
of wound healing. Here, we measured the extent of gel contraction
full papers
M. H. Rafailovich et al.
Figure 12. H2O2generated and released by the cells incubated with
0.4mg mL?1rutile, anatase, and coated TiO2nanoparticles for 2 days
compared to the control cells exposed to no particles.
518
www.small-journal.com
? 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 4, 511–520

Page 9

to evaluate the ability of cells to contract the collagen matrix. Cells
were suspended in prepared collagen solution composed of
purified collagen (1.8mg mLS1), bovine serum albumin (BSA, 2%),
hFibronectin (30mg mLS1), and PDGF (100ng mLS1) in DMEM with
P/S/G at 3?105cells mLS1. The cell/collagen gels (0.6mL
wellS1) were loaded into a BSA-coated 24-well dish. After
preincubation for 2h to gel the mixture, collagen gels were gently
detached by slight tapping on the wells and DMEM (500mL) with
BSA (2%) and PDGF (100ng mLS1) were added. Then, after
another 2h of incubation, the phase images of the gels were taken
and analyzed by measuring the gel size. The contracted ratio was
defined as the ratio of the contracted area (which is the original
gel size minus the final gel size) to the original gel size.
For Western blotting, whole cell lysate was prepared by
scraping cells in lysis buffer with protease inhibitors and
phenylmethylsulfonyl fluoride (PMSF), and then mixed with
4?SDS-PAGE buffer composed of Tris (0.25 M, pH 6.8), sodium
dodecyl sulfate (SDS, 8%), glycerol (40%), and bromophenol blue
(0.02%) with dithiothreitol (DTT, 80mM) at a volume ratio of 3:1.
After the mixture was boiled at 100-C for 3min, the proteins from
the same number of both control and experimental cells were
separated by SDS-PAGE and electrophoretically transferred to a
nitrocellulose membrane. Blots were probed with the anti-actin
antibody, Act(C-11), goat polyconal IgG (Santa Cruz Biotechnol-
ogy). The primary antibodies were then detected with the
secondary antibodies, rabbit anti-goat IgG(HRL) (ImmunoPure
Antibody, Pierce Biotechnology), followed by enhanced chemilu-
minescence (SuperSignal West Dura Extended Duration Substrate,
Pierce Biotechnology). The picture of the bands was taken with a
ChemiImagerTMReady camera (Alpha Innetech Corporation) and
their densities were quantified with ChemiImager v5.5 (Alpha
Innetech Corporation).
TEM: In order to monitor how TiO2nanoparticles penetrate
cells and where they stay in cells, cells exposed to nanoparticles
with different incubation times were fixed as follows and imaged
with TEM. The fresh samples were rinsed three times with PBS and
fixed for 30min in modified Karnovsky fixative using glutaralde-
hyde (2.5% v/v) and paraformaldehyde (2% w/v) in cacodylate
buffer (0.1 M, pH 7.2–7.4) with calcium chloride (2.5mM). The
samples were then rinsed three times with cacodylate wash buffer
for ?5min. After that, the samples were post-fixed in osmic acid
(2%) for 15min and stained in a saturated aqueous solution of
uranyl acetate for 15min. The dehydration process was performed
in a graded series of 30%, 50%, 70%, 95%, and twice in 100%
ethanol, for 15min each. Then the samples were infiltrated with a
50:50 mixture of 100% ethanol and Spurr’s plastic. After
discarding the mixture 1h later, 100% Spurr’s was added to
infiltrate the cells for another 1h. The samples were kept in a
vacuum oven at 60-C overnight after changing the Spurr’s. Finally,
the samples were cut into light-gold-colored ultrathin sections
with a microtome and stained on the grid with uranyl acetate and
lead oxide, each for 5min. After being washed thoroughly with
distilled water, TEM samples of cells were dried and then imaged
under a Joel (model 1200EX) TEM. We analyzed the images by
characterizing the penetration, distribution, and specific location
of the nanoparticles, and by quantifying the size of vesicles and
particle clusters at different time points for different types of
nanoparticles.
Particle coating: Rutile TiO2 nanoparticles were coated by
chemically grafting dense antioxidant/anionic and hydrophobic
polymer molecules directly onto the surface by a recently reported
technique.[15]Briefly, antioxidant formed from grape seed extracts
(oligomeric proanthocyanidins) and anionic polymer (poly[methyl-
vinylether/maleic acid]) were mixed at a 1:1 ratio and dissolved in
a 22:1water/denatured ethanol solution using a lightening mixer
at 25-C. After the solution became homogeneous, a new mixture
was prepared composed of the antioxidant/anionic polymer
solution (30% w/w), deionized water (22% w/w), TiO2(43% w/
w), and hydrophobic polymer (5% w/w, triethoxysilylethyl poly
(dimethylsiloxyethyl dimethicone), Shin-Etu Chemical Co., Ltd.).
The entire slurry was then sonicated for 30min with medium
intensity at 25-C with an Ultrasonic probe (Sonicor Instrument
Co.) at 20 KHz. In order to precipitate the particles and remove the
excess polymers, the resultant mixture, which was a thick
colloidal suspension, was then centrifuged for 15min at
9000rpm and washed with deionized water. The washing
procedure was repeated three times in order to ensure that all
unattached materials were removed. The product was then dried
at 110-C under vacuum for 16–20h.
Flow cytometry: Both the control and the experimental cells
were carefully rinsed more than three times to remove all the
floating particles in the experimental media and detached with
trypsin-ethylenediaminetetraacetic acid (EDTA). After stopping
trypsin with full-DMEM, the cells were rinsed twice using DMEM
with BSA (0.2%) for good separation. Then the cells were
suspended and fixed in paraformaldehyde (1%) in PBS for
20min. After being rinsed twice using PBS and resuspended in
PBS at the concentration of 1?106cells mLS1, the sample
solutions were ready for flow cytometry, which was performed with
a BD FACSCaliburTMbenchtop flow cytometer.
Measurement of hydrogen peroxide: The cellular level of H2O2
was measured with the aid of the fluorogenic probe, Amplex red,
as described by Eu et al. with certain modifications.[19]Briefly,
both the control and experimental cells exposed to different types
of TiO2 nanoparticles were washed, suspended in ice-cold
distilled water (200 mL), and sonicated for 3min with a 9 s pulse
with 1 s off. Then, in order to avoid the influence of fluorescence
from TiO2nanoparticles, all the samples were spinned and only
clear supernatant (100 mL) was added to the reaction buffer (1mL)
composed of Amplex red (10 mM), Tris-HCl (10mM, pH 7.4), and
horseradish peroxidase (1 unit mLS1), and mixed carefully. The
mixture was transferred into a 96-well dish and incubated at 37-C
for 5min before the fluorescence measurement. Fluorescence of
the reaction buffer and the sample solution was measured with a
FLEXstationTMbenchtop scanning fluorometer (Molecular Devices)
with excitation and emission at 560 and 590nm, respectively.
Acknowledgements
Financial support by National Science Foundation MRSEC
program is gratefully acknowledged.
Adverse Effects of TiO2Nanoparticles on Human Dermal Fibroblasts
small 2009, 5, No. 4, 511–520
? 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
519

Page 10

[1] P. Ball, Nature 2001, 414, 142.
[2] a) L. L. Muldoon, M. Sandor, K. E. Pinkston, E. A. Neuwelt,
Neurosurgery 2005, 57, 785; b) T. Neuberger, B. Schopf, H.
Hofmann, M. Hofmann, B. von Rechenberg, J. Magn. Magn. Mater.
2005, 293, 483; c) J. Kreuter, J. Controlled Release 1991, 16, 169;
d) S. Jin, K. M. Ye, Biotechnol. Prog. 2007, 23, 32; e) N. C. Bellocq,
S. H. Pun, G. S. Jensen, M. E. Davis, Bioconjugate Chem. 2003, 14,
1122; f) J. C. Riboh, A. J. Haes, A. D. McFarland, C. R. Yonzon, R. P.
Van Duyne, J. Phys. Chem. B 2003, 107, 1772; g) J. F. Kukowska-
Latallo, K. A. Candido, Z. Y. Cao, S. S. Nigavekar, I. J. Majoros, T. P.
Thomas,L. P.Balogh,M.K.Khan,J.R.Baker,CancerRes.2005,65,
5317.
[3] a) V. L. Colvin, Nat. Biotechnol. 2003, 21, 1166; b) E. Oberdorster,
Environ. Health Perspect. 2004, 112, 1058; c) S. Foley, C. Crowley,
M. Smaihi, C. Bonfils, B. F. Erlanger, P. Seta, C. Larroque, Biochem.
Biophys. Res. Commun. 2002, 294, 116.
[4] Y. Pan, S. Neuss, A. Leifert, M. Fischler, F. Wen, U. Simon, G.
Schmid, W. Brandau, W. Jahnen-Dechent, Small 2007, 3, 1941.
[5] N. Pernodet, X. H. Fang, Y. Sun, A. Bakhtina, A. Ramakrishnan, J.
Sokolov, A. Ulman, M. Rafailovich, Small 2006, 2, 766.
[6] V. L. Colvin, Nat. Biotechnol. 2003, 21, 1166.
[7] a) B. K. Bernard, M. R. Osheroff, A. Hofmann, J. H. Mennear, J.
Toxicol. Environ. Health 1990, 29, 417; b) J. L. Chen, W. E.
Fayerweather, J. Occup. Environ. Med. 1988, 30, 937; c) G. A.
Hart, T. W. Hesterberg, J. Occup. Environ. Med. 1998, 40, 29.
[8] M. Cho, H. Chung, W. Choi, J. Yoon, Water Res. 2004, 38, 1069.
[9] O. Carp, C. L. Huisman, A. Reller, Prog.Solid State Chem.2004, 32,
33.
[10] R. Dunford, A. Salinaro, L. Z. Cai, N. Serpone, S. Horikoshi, H.
Hidaka, J. Knowland, FEBS Lett. 1997, 418, 87.
[11] T. C. Long, J. Tajuba, P. Sama, N. Saleh, C. Swartz, J. Parker, S.
Hester, G. V. Lowry, B. Veronesi, Environ. Health Perspect. 2007,
115, 1631.
[12] a)E.M.Ophus,L.Rode,B.Gylseth,D.G.Nicholson,K.Saeed,Scand.
J. Work Environ. Health 1979, 5, 290; b) R. C. Lindenschmidt, K. E.
Driscoll, M. A. Perkins, J. M. Higgins, J. K. Maurer, K. A. Belfiore,
Toxicol.Appl.Pharmacol.1990,102,268;c)G.Oberdorster,J.Ferin,
B. E. Lehnert, Environ. Health Perspect. 1994, 102, 173.
[13] C.M.Sayes,R.Wahi,P.A.Kurian,Y.P.Liu,J.L.West,K.D.Ausman,
D. B. Warheit, V. L. Colvin, Toxicol. Sci. 2006, 92, 174.
[14] J.R.Gurr, A.S. S.Wang,C.H.Chen, K.Y. Jan,Toxicology2005, 213,
66.
[15] W. A. Lee, N. Pernodet, B. Q. Li, C. H. Lin, E. Hatchwell, M. H.
Rafailovich, Chem. Commun. 2007, 4815.
[16] K.Ghosh,Z. Pan, E.Guan,S. R.Ge,Y. J.Liu,T. Nakamura,X. D. Ren,
M. Rafailovich, R. A. F. Clark, Biomaterials 2007, 28, 671.
[17] K. Ghosh, X. D. Ren, X. Z. Shu, G. D. Prestwich, R. A. F. Clark, Tissue
Eng. 2006, 12, 601.
[18] H. Hidaka, S. Horikoshi, N. Serpone, J. Knowland, J. Photochem.
Photobiol. A 1997, 111, 205.
[19] J. P. Eu, L. M. Liu, M. Zeng, J. S. Stamler, Biochemistry 2000, 39,
1040.
Received: June 6, 2008
Revised: October 14, 2008
full papers
M. H. Rafailovich et al.
520
www.small-journal.com
? 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 4, 511–520