Testing Metal-Oxide Nanomaterials for Human Safety

Full text

Nanomaterials
R. Landsiedel, L. Ma-Hock, A. Kroll,
D. Hahn, J. Schnekenburger, K. Wiench,
W. Wohlleben* ................................xx–xx
a
Testing Metal-Oxide Nanomaterials for
Human Safety
adma.200902658C
a
Final page numbers not assigned
The novel properties of engineered nanomaterials may alter their interaction with the
human body, especially for inhalation of unintentionally released biopersistent material.
We discuss the characterization of nanoparticles in interaction with biological media
and we review animal inhalation and cell culture studies in comparison to original
results. We establish that an intrinsic size-specific toxicity does not exist and identify
material-specific indicators of concern that help to select safe uses.
REVIEW

Testing Metal-Oxide Nanomaterials for Human Safety
By Robert Landsiedel, Lan Ma-Hock, Alexandra Kroll, Daniela Hahn,
Ju
¨rgen Schnekenburger, Karin Wiench, and Wendel Wohlleben*
1. Introduction
The intentional generation and application of nanomaterials
with novel properties is one of the century’s key technology
developments, offering extraordinary opportunities in various
technological fields such as electronics, energy management,
structural materials, functional surfaces, construction, and
information technology, but also in the pharmaceutical and
medical field. Indeed, the appearance of clean-tech, seen as the
capture, storage, and conversion of energy and resource-efficient
materials, depends critically on nanomaterials, whereof the
majority is fabricated by compounding engineered particulate
nanomaterials.
Since the miniaturization of materials down to the nanometer
scale can change physical and chemical properties, nanomaterials
will presumably also influence biological
systems—regardless of a human intention
behind the material’s generation. The
natural nanomaterials and the unintention-
ally man-made nanomaterials by far out-
weigh the engineered nanomaterials, but
the exposure scenarios resemble each
other.
[1]
A systematic risk assessment
requires the separate determination of both
the hazard potential and the actual exposure
levels resulting in a risk characterization
(Scheme 1).
[2]
Typical consumer products
combine low exposure to free nanostruc-
tures and low hazard potential. Materials
with high hazard potential are restricted to
professional handling, where safe levels of
exposure can be enforced by technical
measures.
The scientific community started to
evaluate the potential hazard of nanomater-
ials since 1992,
[3–6]
culminating in the
current developments of regulatory frameworks in the EU, USA,
and Canada.
[7]
Based on the extreme diversity of hazard potential—
from potent to harmless—and diversity of exposure—from
occupational to consumer settings—the regulatory framework
evolves into a case-by-case risk assessment. Hazard potential and
(internal) exposure (Scheme 1) need to be merged into a testing
strategy, e.g., in the REACH Implementation Plan. Current
knowledge is sufficient to shape the first regulation approaches,
but these have to undergo revisions with enhanced risk
assessments.
[8]
The outcome of safety research is also an integral
part of the public awareness and confidence in nanotechnology.
[9,10]
In the present contribution, we focus on the potential adverse
effects of engineered metal oxide nanomaterials, in comparisons
to the frequently discussed toxicity of carbon nanomaterials.
Appropriate toxicity testing requires a thorough understanding of
nanomaterial specific properties with regard to distribution in the
body and possible nano-specific effects on the systemic and
cellular level (Scheme 2). The unique nano-specific properties of
nanomaterials require a careful adaptation of the test methods,
and the OECD recommends that guidelines be newly developed
or revised for sample preparation and dosimetry, degradation and
fate, for inhalation and for the majority of the physicochemical
characterization methods.
[11]
A base set of applicable toxicity
screening systems and characterization tools has been suggested
already by Warheit et al.
[12]
At present, inhalation studies with
animals are the most predictive testing of possible adverse effects
of nanomaterials on humans. But inhalation studies entail the
REVIEW
www.MaterialsViews.com www.advmat.de
1
2
3
4
5
6
7
8
9
10
11
12
13
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
[*] Dr. W. Wohlleben, Dr. R. Landsiedel, Dr. L. Ma-Hock, Dr. K. Wiench
BASF SE 67056 Ludwigshafen (Germany)
E-mail: wendel.wohlleben@basf.com
Dr. A. Kroll, Dr. D. Hahn, Dr. J. Schnekenburger
Gastroenterologische Molekulare Zellbiologie, Medizinische Klinik
und Poliklinik B
Westfa
¨lische Wilhelms-Universita
¨tMu
¨nster
Schloßplatz 2, 48149 Mu
¨nster (Germany)
DOI: 10.1002/adma.200902658
Nanomaterials can display distinct biological effects compared with
bulk materials of the same chemical composition. The physico-chemical
characterization of nanomaterials and their interaction with biological media
are essential for reliable studies and are reviewed here with a focus on widely
used metal oxide and carbon nanomaterials. Available rat inhalation and cell
culture studies compared to original results suggest that hazard potential is
not determined by a single physico-chemical property but instead depends on
a combination of material properties. Reactive oxygen species generation,
fiber shape, size, solubility and crystalline phase are known indicators of
nanomaterials biological impact. According to these properties the sum-
marized hazard potential decreases in the order multi-walled carbon nano-
tubes >> CeO
2
, ZnO >TiO
2
>functionalized SiO
2
>SiO
2
,ZrO
2
, carbon
black. Enhanced understanding of biophysical properties and cellular effects
results in improved testing strategies and enables the selection and
production of safe materials.
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
sacrifice of animals and are quite expensive and time-consuming.
Traditional methods have to be adapted and in vitro methods
[13]
must be improved through better understanding of their
biophysical mechanisms until the in vitro tests achieve predictive
power.
This paper is organized as follows:
Section 2 starts from a wider perspective and discusses
exposure levels and possible routes of internalization in humans.
Sections 3–6 track the physiological effects from biophysical to
cellular to systemic levels.
In Section 3 we review the physicochemical properties
of nanomaterials and their characterization with appropriate
biophysical methods. We focus on the biophysical modification of
the nanomaterial’s surface and state of agglomeration in cell
culture media (Scheme 3).
The in vitro toxicity (cell viability, genotoxicity, inflammation)
of metal oxide and carbon nanomaterials is reviewed in Section 4
and is complemented by original results from different titanium
dioxide (TiO
2
) nanomaterials (Scheme 3).
In Section 5 we give an overview on the limited range of
existing inhalation studies with engineered nanomaterials.
Furthermore, we present original data from our inhalation
studies with six metal oxide materials and two carbon materials.
These results are excellently comparable due to an identical
experiment design (Scheme 4).
Section 6 summarizes the correlations between the in vivo and
in vitro chapters, leading to a ranking of hazard potential for the
materials tested in Section 7. We identify materials properties and
in vitro indicators that should trigger in vivo experiments in a
future testing strategy.
2. Emission versus Exposure
2.1. Emission Quantities and Possible Routes of
Internalization
Potential human exposure to nanomaterials is as manifold as the
potential applications of different nanomaterials. It is beyond the
scope of the present contribution to assess all factors in detail.
The following paragraph introduces four questions to guide a
prioritization:
(i) Emission of nanoparticles from composites or powders?
Touching a composite thermoplastic that was reinforced with
silicon dioxide nanoparticles (Scheme 1) is of less concern
than being exposed to free nanoparticles. Consumer
applications of nanomaterials focus on composite materials
from which only the unintended release of fragments
containing nanoparticles during use, recycling, or disposal
may raise concerns.
[14]
Given typical product lifetimes on the
order of years, the dose of release from composites should
be vanishingly low, even for a hypothetical complete
degradation. First available evidence supports this assump-
tion: Abrasion of acrylate coatings containing ZnO nano-
particles did not lead to significant release of nanoscale
aerosols.
[15]
Not intending to banalize the issue, one should
keep in mind that evolution itself developed most remarkable
nanostructured materials: Human bones and human tooth
enamel are examples of organic–inorganic hybrid nano-
materials with biopersistence.
(ii) If there is emission of free nanoparticles, are these liquid,
soluble or biopersistent? Natural and technological processes
can produce ultrafine droplets or nanoemulsions (e.g., milk).
The ultrafine state may affect the uptake of a substance in
the body, but inside the body the substance will dissolve or
blend in body fluids and only effects different from those
associated with nanometer sizes are expectable. In contrast,
biopersistent nanomaterials could exhibit general nanome-
ter-size-specific effects if internalized. Among the natural
(biogenic, geogenic, or pyrogenic) sources of biopersistent
nanomaterials, black carbon from incomplete biomass
combustion dominates with 50 to 270 megatons per
year,
[16,17]
followed by 16 megatons of inorganic dust from
desert storms.
[18]
But also human activity releases nanoma-
terials as unintended by-products. A typical urban atmo-
sphere contains 10 mgm
3
particulate matter (around
10
5
particles m
3
); a candle or a cigarette release 10 g m
3
particles (around 2 10
11
m
3
).
[18]
Welding fumes consist of
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Robert Landsiedel studied
chemistry, food chemistry, and
toxicology in Kaiserslautern,
Mainz, and Leipzig. After
working for the state, he earned
a PhD with a thesis on the
metabolism and mutagenicity
of benzylic compounds. After a
Postdoc in Potsdam, he joined
BASF in 1999 and worked in
different functions in Ludwig-
shafen, North Carolina, and
Tokyo. Since 2004 he heads a
unit of several toxicological routine and research labs and is
involved in projects on alternative methods and
nanotoxicology.
Wendel Wohlleben studied
physics at Heidelberg Univer-
sity and ENS Paris. He earned
his PhD in 2003 from LMU
Mu
¨nchen with a biophysical
thesis on energy harvesting in
photosynthesis, performed at
the Max-Planck-Institute for
Quantum Optics. He then
developed chemically selective
microscopy in Marburg and
joined BASF polymer physics
research in 2005, acting also as
innovation manager for BASF’s
nanotechnology activities. His research focuses on prepara-
tion, characterization, and self-assembly in complex suspen-
sions, especially with regard to the safety of nanomaterials.
2ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
10
9
m
3
metal nanoparticles.
[19]
These values set a frame of
reference, and they justify the use of CB as reference material
in safety testing. The world production of CB for tires and
printing inks is estimated around 8 megatons per year
(in 1996).
[20]
Emission of CB is relevant in aerosol form, but
its quantity is vanishingly small compared to the background
of black carbon.
[21]
(iii) Is the emission intended or unintended? There are a
limited number of applications of biopersistent nanoma-
terials with intentional (external) exposure of the human
body, especially as sunscreens in cosmetics (Scheme 1).
The global turnover with engineered nanomaterials
[22]
can
be converted into very rough estimates for the quantities
that were actually produced in 2007: metal oxide
nanoparticles: 0.02 megatons (¼20000 tons); metal nano-
particles: 20 tons; carbon nanotubes (CNTs): 100 tons.
Graphene catches up with 15 tons in 2009.
[23]
These values
are worlds apart from CB, but they still outnumber
specialties in the OECD sponsorship program like
quantum dots, dendrimers or fullerenes/C
60
.
[24]
The vastly
dominant applications are technically bound: CNTs and
graphene in polymer nanocomposites, metal nanoparticles
in catalysts, electronics, and antimicrobials. Note that
nanoscale silver ranks high only when the number of
marketed products is counted,
[25]
but not among the
production quantities. Metal oxide nanoparticles find broad
application from coatings and plastics over catalysts to
sunscreens. Specifically nanoscale TiO
2
is
used for coatings and sunscreens, with an
estimated production of 0.005 megatons
per year,
[20]
expected to grow to 0.06
mega-tons per year until 2025.
[26]
Sunscreens represent one of the few
nanomaterial-containing products to
which humans are intentionally exposed.
Preliminary scenarios of coating degrada-
tion
[27]
estimate levels of unintended
emission around 10
2
mgm
3
in air
and around 10 mgL
1
in soil compart-
ments.
[20]
Measure-ments at workplaces
in nanoscale TiO
2
production did not
reveal any significant emission.
[2]
(iv) Is the exposure oral, dermal, or by
inhalation? Nanotechnology in food pro-
cessing focuses on nanostructures for
encapsulation, whose degradation in the
human body is essential to fulfill their
purpose. Migration of particles larger than
1 nm from packaging materials into food
seems to be no concern.
[28]
This may
explain why relatively few investigations on
the absorption and effects of nanoparticles
via the oral route are available. TiO
2
and
ZnO nanoparticles in sun screens were
comprehensively tested on skin and sev-
eral studies demonstrated that the intact
human skin is an effective barrier for those
nanoparticles.
[29,30]
The absence of dermal
penetration minimizes human internal
exposure and hence minimizes the health risks. On the
other hand, there is a wealth of information on the effects of
ultrafine particles in the air
[18,19]
indicating the concerns for
human health arising from the inhalation of ultrafine
particles.
In summary, the highest concern for human health arises from
the unintended exposure to biopersistent nanoparticles. These
may occur mostly in workplaces, and to a much lesser extent in
consumer settings. The existing knowledge of adverse health
effects by inhaled ultrafine particles gives priority to the inves-
tigation of effects caused by inhaled engineered nanomater-
ials,
[31]
and among these, the emission quantities prioritize metal
oxides.
2.2. Approaches to Regulation
The US OSHA Permissible Exposure Limit (PEL) for General
Industry is 5 mg m
3
time-weighted average (TWA) (PEL listed
under Inert or Nuisance Dust). The American Conference of
Governmental Industrial Hygienists (ACGIH) Threshold Limit
Value states the same limit value of 5 mg m
3
TWA for Particles
(insoluble or poorly soluble) Not Otherwise Specified (PNOS). A
distinction between inhalable and respirable dust was changed in
2001 to PNOS.
Whereas in Germany the legally binding Occupational
Exposure Limit (OEL) for inhalable dust is 10 mg m
3
, there is
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
Scheme 1. The generally accepted principle assesses risk as: Risk ¼Hazard Exposure. Risk is
controlled by low hazard or low exposure, ideally both. The examples from present technology
show the laboratory synthesis of nanostructured battery materials (top left) and SiO
2
-reinforced
thermoplastic nanocomposites (bottom left). In contrast to the toxicologically relevant internal
exposure, we show here the external exposure. External exposure does not lead to uptake in all
cases, as demonstrated by the case of nanostructured sun screen pigments
[29,30]
that prevent
skin damage (bottom right).
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 3
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
in addition an OEL for respirable dust (<10 mm) of 3 mg m
3
(excluding soluble particles, ultrafine particles and coarse-
disperse dust). The German MAK commission, a scientific
committee for the investigation of health hazards of chemical
compounds in the workplace, advises public authorities and
established a so-called MAK-value (maximum concentration at
the workplace) for respirable nuisance dust of 1.5 and 4 mg m
3
for inhalable dust.
The regulatory and political framework evolves into a
case-by-case risk assessment that requires knowledge of both
hazard and exposure: California’s Division of Toxic Substances
Control (DTSC) has requested data regarding CNT hazard, but
also regarding uses, life cycle fate, and transport.
[32]
Key milestones for a first approach to regulation
[8]
are a
testing strategy in the REACH Implementation Plan and a
definition of nanomaterials for regulatory purposes. That
definition must complement scientific criteria of size
[33]
or
surface area
[31]
or others with protocols for cheap and
undisputable measurements, e.g., from the
revised OECD guideline of ISO standards.
[11]
3. Review on the In Situ
Biophysics of Nanomaterials
Since the potential nanohazard arises from
the nanosize, the full characterization of the
pristine or as-produced nanomaterial must be
completed with the conditioning effects and
the actual state of dispersion in biological
media (Scheme 3). In the next two subsections,
we discuss the surface and agglomeration
state with their in situ dynamic variability
(Scheme 2) in more detail. These two proper-
ties are decisive for biokinetics, biodistribu-
tion, and clearance of nanomaterials. At the
same time, surface and agglomeration are
rather sensitive to the experimental protocols,
the nanomaterial, and the surrounding med-
ium. In the third subsection we summarize
particle properties influencing measurement
procedures of in vitro toxicity tests and
suggest suitable technologies for particle
characterization.
3.1. Surface Conditioning and Hybridization
The control of fate and biokinetics by surface is
standard practice in biological sensing, deliv-
ery, and imaging of live cells and tissues. Such
issues have been reviewed by De et al.
[34]
in a
previous issue of the same journal, and by
Dobrovolskaia and co-workers
[35]
from a drug
delivery perspective. The nanomaterial’s high
surface to mass ratio enhances any specific
surface characteristics, including the interac-
tion with serum, saliva, mucus, or lung lining
fluid components. The adsorbed molecules
(certainly proteins, presumably also peptides, carbohydrates, and
phospholipids) change the biological identity of the conditioned
nanomaterial. Their differential adsorption
[36]
induces a char-
acteristic ‘‘protein corona’’ around the nanoparticle, a term
coined by Dawson and co-workers.
[37–42]
To date, most work has been devoted to serum interactions.
Blood serum contains about 75 mg mL
1
interface active
components, which will unintentionally, but inevitably interact
with the nanoparticle, and approximately 50 proteins have been
identified in association with various nanoparticles.
[35]
In vitro
assays have reduced buffer compositions, but even these contain
more than 30 components with 11 mg mL
1
salt, 4.5 mg mL
1
glucose, and 3 mg mL
1
proteins (for the case of Dulbecco’s
modified Eagle medium (DMEM) with 10% fetal bovine serum
(FBS)). Using polymer nanoparticles, it is possible to fine-tune
the hydrophilicity by the copolymer composition, e.g., by
changing the ratio of NIPAM and BAM monomers.
[39]
It was
established by different physicochemical methods that a single
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Scheme 2. Fate and effect of nanomaterials in the body. Depending on the source of the
exposure— dusting of a powder, degradation from a nanocomposite matrix or surface— free
nanomaterials may arise as aerosol (dust) or suspension. The irregular shape (the example is
from ZnO in Fig. S1) is characteristic for many metal oxide nanomaterials. In the case of uptake,
the internal exposure does not relate to naked, but to conditioned nanomaterials, indicated as
solid contour. Depending on the distribution and biokinetics, different organs may be reached
where primary and toxic effects may occur. ROS, reactive oxygen species (e.g., radicals); RS,
reactive species (e.g., metal ions).
4ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
layer of albumin is adsorbed to the surface of the largest and
most hydrophobic particle with an adsorption constant around
10
6
mol
1
, whereas a sparser layer is associated with the more
hydrophilic particles.
[38]
Over time, albumin has a residence time
around 100 s
[43]
and is replaced, e.g., by apolipoprotein A-I, a
protein of 30-fold lower abundance, but with higher affinity and
slower kinetics.
[41]
By 1D gel comparison of commercial polymer
nanoparticles with different chemical surface functionalization,
typically 40% of the corona proteins have been found to be
conserved between amine, plain, and carboxyl-modified poly-
styrene nanoparticles, and around 30% of the corona proteins are
specific to a single functionalization.
[42]
Preferential conditioning
by immunoglobulin IgG induces clearance by
MPS cells, whereas dysopsonins (albumin,
IgA) prolong circulation in the bloodstream,
[44]
and bovine serum albumin (BSA) conditioning
decreased resorption into lung tumor cells.
[45]
In summary of the serum interactions,
neutral particles seem to have slower adsorp-
tion kinetics than charged particles, ditto
hydrophobic particles, but these also differ
in protein identity in the protein corona from
hydrophilic particles.
[35]
Polymer nanoparticles have the obvious
methodical advantage of being nicely homo-
geneous, spherical, and well dispersible.
Corona-related mechanisms are known to
much less detail for inorganic nanomaterials.
First results indicate that the corona is selective
also for different naked inorganic surfaces: For
instance, specific pro-inflammatory and anti-
inflammatory precursors displayed an up to 30
times higher affinity to Ni and Al particles than
albumines, as demonstrated by isotope label-
ing (Fig. 1).
[46]
Likewise, diamond nanoparti-
cles showed a high affinity for vitronectin,
which can stimulate tumor necrosis factor a
(TNF-a) release from alveolar macrophages.
However, even with its low relative affinity,
the high concentration of albumin in serum
still represents a significant portion of the
bound protein fraction for all nanoparticles.
[46]
Quantitatively, albumin adsorption onto car-
boxylic-acid functionalized inorganic nano-
particles was anti-cooperative and saturated
at serum level concentrations and one
monolayer.
[43]
Working with semiconductor
(quantum dot) particles, Frangioni and co-
workers
[47,48]
showed in a series of experi-
ments how surface functionalization controls
biodelivery: Particles were filtered by renal
clearance and urinary excretion only for
diameters below 6 nm and with zwitterionic
or neutral organic modification to prevent
protein adsorption. The significant corona
conservation between different polymer parti-
cles
[42]
is reflected by the uniformity of surface
charge of various naked metal oxide nanopar-
ticles when dispersed in serum-containing
media (20 to 10 mV),
[49,50]
attributed to a universal coverage of
the nanoparticle surface (with zeta-potential ranging from 25 to
þ55 mV) by proteins.
[49]
Part of the reduction of zeta-potential
must be attributed to charge-screening in the physiological buffer
with isotonic salt load.
[51]
As expected, smaller particles adsorb
more protein, demonstrated directly by UV and secondary ion
mass spectroscopy (SIMS) detection of the colloidal ZnO
surface
[2]
and indirectly by BCA assay (bicinchoninic acid) of
the protein fraction that did not adsorb onto TiO
2
.
[52]
On
nearly the same series of naked metal oxides as described in
Section 5, protein adsorption was shown to even induce buffer
depletion, but only at completely unphysiological nanoparticle
1
2
3
4
5
6
7
8
9
10
11
12
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
Scheme 3. Workflow of in vitro testing. The nanomaterials are dispersed in a physiological
nutrient medium that contains proteins and other macromolecules (coils) and low-molar-mass
components such as salts (dots). The nanomaterial surface changes by differential adsorption of
some of these components, correlated with changes in the state of agglomeration. Only
afterwards, the interaction with a multitude of cell species is studied by the (typically optical)
readout of a large number of markers and endpoints. Details are shown for the three markers that
are essential for the discussion in Section 4.
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 5
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
concentrations >10 mg mL
1
.
[52]
Interestingly, the adsorption
was blocked by pretreatment of the particles in serum, indicating
longer residence times than on the polymer-functionalized
particles.
[43]
The available results indicate that the protein adsorption and
biokinetics of (stabilized) polymer particles and (polymer
stabilized) inorganic particles follow the same mechanisms.
First evidence emerges also on the interaction between
particles and the lung lining fluid, the first conditioning contact
after inhalative exposure. Apart from proteins, also phospholipids
from lung lining fluid have been shown to adsorb to nanoparticle
surfaces,
[53]
and we demonstrated by antibody staining that the
dominant surfactant protein SP-A does adsorb onto metal oxide
nanoparticles.
[2]
The direct comparison of conditioning CNTs in
either serum-containing medium or in dipalmitoylphosphati-
dylcholine (DPPC)-containing medium showed a significantly
stronger inflammation potential with the DPPC, demonstrating
the direct impact of the corona on cytotoxicity.
[54]
Sometimes, however, minute differences between nanoparti-
cle surfaces strongly change the biodistribution: Surface
functionalization with poly(ethylene glycol) (PEG) of varying
chain length – typically considered an inert molecule – resulted
in major changes in organ/tissue-selective biodistribution and
clearance from the body,
[47]
although 2D gel electrophoresis
showed that immune-competent proteins (IgG, fibrinogen) bind
much more than albumins irrespective of PEG chain length.
[55]
Verma et al.
[56]
demonstrated that of two nanoparticle ‘‘isomers’’
with identical hydrophobic content, one functionalized with
sub-nanometer striations of alternating anionic and hydrophobic
groups, the other with the same moieties in random distribution,
only the striated particles penetrated the plasma membrane
without bilayer disruption. Such phenomena make it difficult to
relate materials properties directly to physio-
logical effects without knowing the biophysical
interactions (compare abstract figure).
We conclude that despite the human risk
being dominated by inhalation exposure and
by metal oxide nanoparticles, most work on the
protein corona has been devoted to polymeric
nanoparticles and serum proteins, often
restricted to albumin. In future, metal oxides
and lung lining fluid interaction with their
higher relevance for human safety should be in
the focus.
3.2. State of Agglomeration
Clearly a correlation between the biological
surface conditioning—controlled by the che-
mical functionalization—and the colloidal
interaction between the thus coated particles
is to be expected.
[24]
In good qualitative
agreement, numerous studies established that
aqueous suspensions of non-functionalized
nanoparticles are stabilized against agglom-
eration by the addition of bovine/human
serum albumin (BSA/HSA) and some other
proteins. The effect has also been exploited in production for the
debundling and dispersion of graphene and CNT material before
chemical compounding (Fig. 2).
[57,58]
Especially albumins in
water or DMEM have dispersed and stabilized a wide variety of
nanomaterials: CNTs,
[57,59–61]
metal nanoparticles,
[62]
metal
carbide nanoparticles,
[63]
and metal oxide nanoparticles.
[51,61,64–67]
It has been shown that a higher protein concentration leads to a
smaller average agglomerate size of the nanoparticles
(Fig. 3b).
[51,66]
Working with a 50-fold excess of serum protein
concentration, which is the relevant range for nanoparticles that
translocated into the human blood stream, Richter and cow-
orkers
[67]
established that agglomeration of TiO
2
and wolfram
carbide is prevented over more than 40 min, compared to 5 min
until complete agglomeration in DMEM. Furthermore, they
showed that BSA alone is sufficient to prevent the agglomeration
process.
[67]
Alternatively, suppression of adsorption and steric
stabilization by PEG functionalization also stabilizes particles in
DMEM/FBS.
[68]
Also natural organic matter such as fulvic or
humic acids can act as wetting and dispersing agents for
nanoparticles and CNTs.
[61,69]
However, contradictory results
demonstrating an increased agglomeration of nanoparticles by
addition of serum proteins or organic acids has also been
reported.
[70,71]
The time course of colloidal stability and the choice
of proper characterization methods may be essential to resolve
this apparent contradiction, see Section 3.3.
There is an essential need for studies investigating the
(de)agglomeration potential of the other ligands that have been
identified in the protein corona of conditioned nanoparticles.
Bronchoalveolar lavage fluid (BALF) was reported to be a vehicle
in which to suspend organic (soot)
[53,72]
and metal oxide
nanoparticles, especially in reduced compositions with only the
most important surfactant protein/phospholipid (phosphatidyl
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
Figure 1. Detailed qualitative characterization of the nanoparticle–protein interaction: Relative
protein abundance in free and bound fraction as estimated from the total peptide score, for the
11 most abundant proteins that were common between Ni and Al nanoparticles. The adsorbed
protein corona is specific and selective for the different pristine nanoparticle surfaces
(figure redrawn with permission from ref. [46).
6ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
choline) constituents
[73]
or mixtures of BSA and DPPC.
[61]
The
use of extensive ultrasonication in these experiments
[61,73,74]
makes it difficult to compare the experimental results to
calculations that find that the interaction energy between
TiO
2
–agglomerates cannot be overcome by the interaction
between the particles and DPPC.
[75]
In a comparison of eight
metal oxide nanomaterials, we showed that the anticipated
interface activity of surfactant proteins is in general not
sufficiently strong to overcome the agglomeration or flocculation
tendency due to other components in complete BALF.
[2]
Strikingly, the two particles that were functionalized with
synthetic polymers evaded near-complete agglomeration and at
the same time differentiated by low overall protein adsorption,
but strong SP-A interaction.
[2]
The physicochemical results are in
good agreement with histological studies of lung slices after
inhalation exposure of rodents, in which the particulate material
that was deposited on the lung surface was found in the form of
agglomerates.
[76]
While the deagglomeration potential by natural macromole-
cules certainly changes transport and biokinetics, a deagglom-
erated nanomaterial is in general not more potent, as demon-
strated by the example of polymer-functionalized BaSO
4
that stays
well-dispersed in a variety of media, but has virtually no in vitro or
in vivo effects.
[2]
What is the mechanism of dispersion by interface-active
proteins? Given the rather low zeta-potential of conditioned
nanoparticles,
[49,51]
the dispersing effect of the protein corona is
not related to electrostatic repulsion. Instead, the stabilization
must be a steric mechanism, whereby the entropy decreases if the
protein coronas of approaching nanoparticles start to overlap.
Electrostatic stabilization collapses in high ionic strength buffers
such as DMEM with 0% FBS, then shifts to a steric stabilization
by the adsorbed proteins in 100% FBS. The steric stabilization by
biopolymers has been exploited industrially for a long time since
protection colloids such as gelatine or starch stabilize organic
composite particles.
[77]
Structural models from X-ray diffraction
seem to suggest that proteins fold into a single well-defined
structure, which would eliminate the entropy stored in the
structural degrees of freedom, hence disabling steric repulsion.
However, most proteins are minimally stabilized mesoscopic
biopolymers whose configuration fluctuates around the time-
average structure under physiological conditions—a field that
was pioneered by Kai Wu
¨thrich (Nobel Prize 2002).
[78]
The quantitative degree of deagglomeration is controversial,
due to (i) the use of different dispersion protocols and (ii) the
disagreement of measurement techniques, which will be
discussed in more detail in the following subsection. The
dispersion protocol defining shear rate, energy input and
duration of conditioning has a drastic influence on the resulting
state of agglomeration as established also in ISO 14887 ‘‘Sample
preparation—Dispersing procedures for powders in liquids’’.
One can either mimic the dispersing action that we assume to be
active in the human body, and since the blood stream is laminar
with rather low shear forces, ultrasonication should be omitted.
Alternatively, one assumes that only the most dispersed fraction
has a profound effect; then one can try to prepare the total
administered dose in the most dispersed state, using wetting
agents, vortexing, and ultrasonication. Since both approaches
have been pursued, biophysical data published so far are hardly
comparable on the quantitative level.
3.3. Characterization of Nanomaterials for Biological Testing
The intrinsic polydispersity and inhomogeneity of nanomaterials
represent major obstacles for a biophysical characterization.
Traditional methods already fail to characterize nanomaterials in
a controlled environment (e.g., distilled water þone surfac-
tant).
[11]
This situation is drastically aggravated under physiolo-
gical conditions since at least 40 components add to the colloidal
domain and interact with each other and with the nanomaterial.
Recent contributions by Hussain, Tiede, Powers, the OECD and
others stress the need for a conscious characterization beyond the
naı
¨ve application of characterization methods that claim to cover
the relevant parameters of nanotoxicology.
[62,64,79–82]
The most
important parameters and appropriate measurement techniques
are summarized in Table 1.
3.3.1. Intrinsic Properties
Impurities are an issue especially for CNTs, with catalysts
(nanoparticulate Co, Fe, Ni, and Mo) and amorphous carbon
being present during their synthesis that may impose additional
toxic effects.
[59]
Such trace elements were the subject of previous
studies on welding fumes.
[83]
The distribution of primary particle
and aggregate sizes of a pristine nanoparticle powder requires
proper statistics of at least 10
6
nanoparticles, corresponding to
more scanning electron microscopy (SEM) images than reported
usually. Some nanoparticles are not stable in aqueous solutions
and can release chemical substrates. If particles are designed to
dissolve in aqueous solutions like water-soluble quantum dots
[84]
or show an intrinsic, size-dependent dissolution in aqueous
media like ZnO,
[85]
particles will release metal ions when
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Figure 2. Qualitative screening of the dispersing action by the protein
corona on CNTs. In varying environments (here: basic, neutral, acidic pH),
the different proteins (lower axis) adsorb effectively to CNTs and ensue a
dispersing action, visualized directly by the black color of dispersed CNTs.
Figure reproduced with permission from ref. [57].
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 7
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
introduced into biological media. Cytotoxicity assays that are
sensitive to metal ions may then rather reflect metal ion toxicity
than particle effects. A high surface/mass ratio of nanoparticulate
materials results in excess surface energy enhancing particle
catalytic activity, depending on the crystalline phase. A variety of
nanoparticles such as metal oxide nanoparticles, fullerenes and
silica (SiO
2
) particles were reported to produce reactive oxygen
species (ROS) in cell free systems.
[86–89]
ROS production was
size dependent, e.g., small 2–4 nm-sized nanoparticles had a
100–1000-fold increased kinetics compared to 100 nm-sized
nanoparticles.
[90]
Redox-active nanoparticles may cause false
positive signals in assays based on substrate oxidation or in assays
detecting cell stress induced ROS production. Few metal oxide
nanoparticles like Fe
2
O
3
are magnetic and may generate local
magnetic fields leading to the production of free radicals that in
turn may interfere with cytotoxicity test methods based on redox
reactions.
[91,92]
ROS measurement by electron spin resonance
(ESR) is a valid, but not widely available technique. However, ROS
inside cells can be tracked as detailed in the materials and
methods.
The characterization of chemical composition and purity,
crystalline phase, morphology, and specific surface can be
regarded as relatively safe and well established—the same does
not hold true for the (last two properties of Table 1) state of
agglomeration and corona conditioning effects.
3.3.2. State of Agglomeration
For inhalation aerosols, the Scanning Mobility Particle Sizer
(SMPS) is widely used to determine size distributions of
submicron aerosols, by balancing the electrostatic force on
particles in an electric field with their aerodynamic drag as they
cross a laminar gas flow.
[93]
Aerosols of nanoparticles can be
generated using a dry powder aerosol generator and by
nebulization of particle suspensions. The mass concentration
of the particles in the inhalation atmosphere can be determined
gravimetrically, and the particle size using a cascade impactor, an
optical particle counter, or the SMPS. Such dispersion techniques
generate fine aerosols with particle size distributions in the
respiratory range, but with no more than a few mass percent of
ultra-fine material (i.e., agglomerates <100 nm).
[93]
Intercom-
parison studies indicate a rather high uncertainty between
different SMPS instruments and manufacturers.
[94,95]
More
appropriate dose metrics need to be developed for a relevant
indication of risk in exposure studies.
[96]
Once the nanoparticles are suspended in any physiological
liquid (lining fluid, blood, or serum), size characterization
constitutes a major difficulty due to the enormous colloidal
polydispersity. Ultrafine particles and agglomerates have to be
quantified in an excess of proteins with diameters that are
comparable to the diameter of potentially present dispersed
nanoparticles (Fig. 3a). For characterization of the state of
agglomeration of samples in biological matrices there are a
number of complementary techniques that rely on different
working principles, such as hydrodynamic/sedimentation,
dynamic light scattering (DLS) and fractionating techniques.
The lower working limit of laser Fraunhoffer diffraction is
exceeded by the potentially present ultrafine components, thus
introducing a strong bias in results from laser diffraction, where
by principle only coarse agglomerates will be detected. The
nominal working range of DLS does cover all components from
proteins to agglomerates; however, evenafter filtering some of the
larger agglomerates, DLS still fails to detect the proteins (BSA at
4–6 nm diameter) that constitute 99 wt % of a typical in vitro
sample (Fig. 3a).
[51,97]
When 99 wt % of the measurable
components (proteins) disappear in DLS measurements, also
the DLS results on the remaining 1 wt % of measurable
components (nanoparticles) are highly questionable.
[98]
First,
the failure of reverse Mie-scattering calculation with the
overwhelming scattering of the >0.1 wt % agglomerates that
are present in nearly every physiological suspension of
nanoparticles should be considered and second the well-known
fact that retrieving a size distribution from the autocorrelation
curve is a mathematically ill-posed problem
[99–101]
that fails
especially for very broad distributions such as the four orders of
magnitude in physiological suspension of nanoparticles (Fig. 3a).
However, with careful sample preparation and elimination of the
very coarse agglomerates by ultrasonication, Hussain et al.
[62]
obtained sub-micrometer average diameters for physiological
suspensions of Cu, Al
2
O
3
, Al, Ag, TiO
2
nanomaterials, and they
confirm also by DLS that serum-containing media reach the
same, often lower, diameters as in water, thereby drastically
reducing the agglomeration compared to serum-free cell culture
media. The majority of published data from scattering techniques
neglects the very loose structure of nanoparticle agglomerates.
The standard Mie routines such as those that retrieve relative
concentrations and distributions in commercial DLS software
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Table 1. Most important properties and the appropriate characterization tools.
Method
Minimal Characterization Needed for Comparability of Studies
Chemical composition and purity (pristine nanoparticles) XRD, ICP-MS
Crystalline phase (pristine nanoparticles) XRD
Morphology, primary particle size (pristine nanoparticles) SEM
Specific surface (pristine nanoparticles) BET
Solubility (in water and after conditioning in the test medium) ICP-MS of supernatant
Surface chemistry (pristine nanoparticles) zeta-Pot., SIMS, XPS
Advanced Characterization for Mechanistic Understanding of Observed Effects
Catalytic activity, esp. ROS generation ESR
Protein corona (in vitro: conditioned nanoparticles) SDS-PAGE, zeta-Pot., SIMS
State of agglomeration and potential of deagglomeration (in vitro: conditioned nanoparticles//inhalation: aerosol) AUC, cryo-TEM//SMPS
8ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
assume a solid spherical shape for the Stokes–Einstein relation.
In reality, diffusion-limited colloidal agglomeration leads to a
fractal morphology, and this has been proposed as the dominating
transport mechanism.
[49]
The fractal dimension can be deter-
mined explicitly by static (light, X-ray, neutron) scattering, and for
many colloids a universal fractal dimension of 2.1 has been
found.
[102]
The fractal shape has been incorporated into DLS
evaluation only by specialist particle labs.
[103]
Concluding the DLS
discussion, dynamic, and static light scattering (DLS, multiangle
laser light scattering(MALLS)) as well as Fraunhoffer diffraction
provide complementary information only if it is known from
other sources that size distributions are narrow.
A mighty tool for the characterization of nanocolloids
(0.5–10 000 nm diameter) is the analytical ultracentrifuge
(AUC)
[104–106]
especially the universal interference optics
Beckman XLI with widespread use in the proteome busi-
ness
[107–109]
and, only to a lesser extent, also the disc centrifuges
(Brookhaven Instruments XDC, CPS Instruments DC24000)
with their rather limited detection optics and lower speeds.
Schlieren, turbidity, interference, UV–vis absorption, and X-ray
absorption detection are published.
[105,110]
The optical AUC
method detects the time- and radius-dependent concentration
of the solutes simultaneously with the sedimentation at
600–60 000 rpm. T hereby, we quantify the amount and the
diameter of each component indepen-
dently.
[104]
At present, AUC is the only method
that detects all components from the agglom-
erates to the dispersed nanoparticles and the
sub-10-nm proteins (Fig. 3a): Note that
interference-AUC retrieves without prior
knowledge the correct molar masses and
correct concen trations of 33 mg mL
1
of
BSA monomer and dimer in serum
(Fig. 3c). In contrast to light scattering, AUC
is a fractionating technique by which a
distribution of sizes is determined with high
resolution. Furthermore, in contrast to trans-
mission electron microscopy (TEM)/SEM, the
AUC integrates over 10
12
–10
14
particles in
approximately 0.5 mL of a te st substance, so
that statistical relevance even of minor frac-
tions is high. If low concentrations of
nanoparticles are present in medium contain-
ing high concentrations of proteins, nanopar-
ticles are easily discerned from sedimenting
proteins (i) due to their much higher density
difference compared to the surrounding med-
ium resulting in faster sedimentation by
several orders of magnitude, and (ii) due to
their different absorption spectrum and
higher refractive index. It is possible, but in
general not mandatory to use X-ray detection
optics that are inherently only sensitive to the
inorganic components with high electron
densities. It is straightforward to take into
account the fractal morphology of nanoparti-
cle agglomerates
[49]
and the hydrodynamic
sedimentation of fractals has already been
derived by Lin et al.
[102]
To complement the ensemble methods, an in situ imaging
technique is desirable, but is not generally available at present.
Any optical microscopy does not cover the relevant structural
sizes. Standard electron microscopy introduces artifacts of
unknown extent by drying and vacuum preparation; cryo-TEM
is a compromise, requiring still a number of preparation steps
(shock-freezing the liquid, then replicating and etching), but
paves the way to a high-resolution image of aqueous structures.
The disagreement between different measurement methods is
exemplified for the case of TiO
2
B nanoparticles in FBS (Fig. 3b,
diamonds). Ensemble techniques (DLS, Fraunhoffer diffraction)
detect only agglomerates and disagree by many orders of
magnitude, while imaging (cryo-TEM) and fractionating techni-
ques (AUC) agree at least within a factor 4.
Hence, it is indispensable to complement and critically
compare measurement techniques of different working princi-
ples, such as hydrodynamic/sedimentation, imaging, and
scattering.
[79–82]
Some complementary techniques may be
field-flow-fractionation (FFF-ICP-MS or FFF-MALLS)
[111]
or
particle tracking. Murdock et al.
[97]
have mentioned these
characterization issues earlier and gave an excellent discussion
of the phenomena, but their preference for the simpler, albeit less
sensitive method of DLS led them to underestimate the amount
of deagglomeration in serum. Hassello
¨v et al.
[80]
published an
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
Figure 3. Colloidal characterization of physiological suspensions of nanomaterials. a) The
suspension of nanoparticles in serum spans four orders of magnitude in diameter. A fractionating
characterization (interference-AUC, dotted blue line; turbidity-AUC, solid blue line) detects all
colloidal components from proteins to agglomerates. The submicrometer fractions are missed by
DLS (magenta line). b) The average diameter of the nanoparticle fraction (X-axis) drops
significantly with increasing protein concentration in the suspension medium (Y-axis). CeO
2
(green triangles), TiO
2
B (black squares), and an organically modified ZrO
2
(red dots) (redrawn
with permission from ref. [51). Diamonds: inter-method comparison, see text Section 3.3. c)
Enlarged sub-10 nm-interval with linear axes in order to facilitate the comparison of the protein
signal to the expected value of the BSA monomer at 66 kDa. d) The metal oxide and carbon
nanomaterials of the present study in DMEM/10% FBS (interference-AUC, this data enters into
Table S1, Phys-bio-chem properties of the test materials).
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 9
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
excellent overview of the pros and cons of measurement
techniques; unfortunately they were unaware of AUC.
3.3.3. Surface Conditioning
Once the nanoparticles are incubated in some physiological
medium, the adsorption of proteins requires a combination of
biochemical (for qualitative identification) with physicochem-
ical methods (for quantification of binding mechanisms). By
centrifugation, harvesting and washing, conditioned nanopar-
ticles can be isolated from the medium and their adsorbed
corona can be assessed by SDS–PAGE (sodium dodecylsulfate
polyacrylamide gel electrophoresis),
[41]
ideally by 2D gels for
full characterization, performed up to now only with polymer
nanoparticles with the exception of an early work on iron
oxide.
[112]
A complementing qualitative method is provided by
SIMS, even if the necessity to prepare samples under
ultra-high vacuum is prone to introduce preparation arti-
facts.
[113]
SIMS records ion fragments from the impact of an
energetic primary ion beam; molecular groups on the surface
are detected with ppm sensitivity.
[113]
A purely elemental
resolution with 10 nm depth integration such as from X-ray
photoelectron spectroscopy (XPS) is of high relevance for the
purity of the pristine nanoparticle surface, but of less value for
the identification of organic matter. XPS bombards the sample
with X-rays that excite characteristic core electrons, and has the
advantage of quantitative information.
[113]
Quantitative infor-
mation can be drawn from fluorescence correlation spectro-
scopy (FCS).
[43]
Unfortunately, FCS is not generally applicable
to industrial metal-oxide nanomaterials due to their lack of
fluorescence and to their quenching of the fluorescence of
adsorbed dyes.
Surface properties like hydrophobicity and surface charge
determine the capacity and kinetics of aqueous solution
dispersion and this in turn modulates particle ability to adsorb
proteins or assay components.
[114]
The zeta-potential is closely
related to the surface charge density, screened by salts, and is
experimentally accessible in many cases. Any changes in
zeta-potential should be observed by step-wise addition of buffer
components, so that charge-screening cannot be misinterpreted
as a surface coating by an organic material.
[51]
While the
zeta-potential records the average surface composition, the
dynamic change of the surface can be assessed by gel filtration
of conditioned nanoparticles.
[38]
The longer a protein is desorbed
on average, the longer is its elution delay. Finally, the adsorption
enthalpy can be determined by isothermal titration (also known as
microcalorimetry).
[38,39]
Microcalorimetry is a particularly sensi-
tive method to measure the heat flow of a sample normally under
isothermal conditions at room temperature or at 37 8C or higher.
Detectable heat flows range from a few to 3000 mW. Due to the
high baseline stability the dynamics of slow reactions can be
studied over minutes up to several days.
3.3.4. Interferences With In Vitro Test Assays
Classic cytotoxicity or genotoxicity assays are often based on the
detection of fluorescence or light absorption of indicators
and chemical or enzymatic reactions. Undesired particulate
interactions interfere with the test mechanism and
detection.
[13,115–117]
If undetected by insufficient in situ char-
acterization, such interferences may lead to a misinterpretation of
results. Especially CNTs and fullerenes
[118]
show quite unex-
pected interactions with the testing systems that induce artifact
signals. It has been reported that endotoxin tests are less
sensitive,
[51]
essential nutrients are adsorbed and hence cells
starved.
[115,119]
Furthermore, carbon nanomaterials have been
shown to interact with indicator substances (methylthiazolyldi-
phenyl-tetrazolium bromide (MTT)).
[117]
When protein concen-
tration or protein activity are read outs of cytotoxicity assays these
parameters can be influenced by particles
[120]
as well as by assay
components used for the detection of cellular activity (e.g.,
substrates, dyes)
[116,117,121]
and proteins (lactate dehydrogenase
(LDH)) may be adsorbed and hence misleadingly low concentra-
tions detected.
Light absorption or fluorescence emission is used to
determine toxicity by most of the in vitro assay systems
(Scheme 3). Optical properties of nanoparticles might therefore
directly interfere with these detection systems. Metallic nano-
particles with light-absorptive or light scattering properties like
sodium titanate or TiO
2
might influence the readout in cell
viability assays.
[122]
Moreover, close proximity of gold nanopar-
ticles and fluorescent dyes, have been shown to result in a
quenching of fluorescence signal intensity.
[123]
We believe that
the characterization methods must be adapted to the in situ
properties of the nanomaterials—instead of modifying the
dispersion procedure until turbidity
[62]
is low enough to apply
widespread methods such as DLS. Since the most common in
vitro assays are pH-dependent and may thus be influenced by
acidic or basic nanoparticles, acidity/alkalinity should be tested
when using nanoparticle concentrations which exceed the buffer
capacity of biological media.
3.4. Correlation of Biophysical Properties
To summarize Section 3, the nanometer-sized entity exposed to
the organism is not identical to the pristine nanomaterial, but
undergoes dynamic changes of both its surface chemistry and its
state of agglomeration. The protein corona is partially conserved,
and partially selective for specific naked metal or metal oxide or
organically functionalized surfaces. Serum tends to decrease the
state of agglomeration, whereas lung lining fluid in general does
not. One must abandon the attractively simple picture of a naked
inorganic nanoparticle in the human body; wehave to take proper
care that the in vitro buffers are nearly identical to human body
fluids, in order to mimic closely the true corona and state of
agglomeration that develops in vivo. Due to the complexity and
polydispersity of a physiological suspension of nanoparticles, a
combination of characterization methods with different physical
measurement principles (imaging, hydrodynamic, scattering) is
mandatory. Similar statements have been stressed most recently
by the ‘‘characterization matters’’ initiative.
[124]
In the present
contribution, we fulfill the criteria (Table 1) of minimal charac-
terization for comparability of studies, and we additionally
provide advanced characterization data that may help to elucidate
the mechanisms underlying nanoparticle-effect relationships
(Supporting Information Table S1).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
10 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
4. In Vitro Studies With Engineered
Nanomaterials
4.1. Critical Aspects of Nanomaterial Test Systems
In comparison to animal models, cytotoxicity testing allows for a
simpler, faster and more cost-efficient determination of defined
toxicity endpoints. Classic cytotoxicity assays were established for
soluble chemicals, not for particles (see Section 3.3.4). Since they
are not sufficient at this time to evaluate toxic nanomaterial
effects in cells, multiple assays have to be employed.
[125]
Nanomaterial specific properties are crucial determinants of
biological effects. Recent in vitro screenings have used a variety of
well-characterized nanomaterials
[126]
or different variants of
two kinds of nanoparticles.
[127]
In most of the earlier studies,
however, nanomaterials were used without prior characterization
regarding their composition and physicochemical properties.
Physicochemical properties of nanomaterials such as surface
charge, size, agglomeration state, and shape have been shown to
heavily influence the outcome.
[31,49,128]
These difficulties might
explain why numerous in vitro studies dealing with nanomaterial
toxicity have provided confounding results with little or no
correlation to in vivo data.
Here, we will provide an overview of in vitro toxicity
experiments of engineered nanomaterials (especially metal-
oxides, and some carbon nanomaterial) and we will focus on
studies that have been performed with well-characterized
materials and multiple test systems. For the complementary
nanomaterial classes of metals (including surface modifications)
and quantum dots the reader is referred to the excellent review by
Lewinski et al.
[129]
4.2. Review on the In Vitro Toxicity of Engineered
Nanomaterials
Cultured cells exposed to toxic agents can respond with various
mechanisms that differ in the level of cell damage. Cellular
reactions range from reversible short term stress responses to
irreversible induction of cell death or the long-term malignant
transformation.
[130]
Engineered nanomaterials have been studied
mainly with established in vitro toxicity assays that analyze major
cellular parameters such as cell viability and response to various
stress factors.
4.2.1. Oxidative Stress
Evidence is accumulating that oxidative stress induced by
nanomaterials is a key route by which these nanomaterials
induce cell damage.
[131]
Oxidative stress is often detected using a
fluorimetric 20,70-dichlorofluorescein (DCF)
[132,133]
or a colori-
metric GSH (reduced L-glutathione) assay
[134]
(Scheme 3). An
approximate 50% increase in DCF fluorescence has been
observed after exposure of cultured human skin fibroblasts to
anatase TiO
2
nanoparticles (UV irradiated).
[135]
However, cells
could be protected against TiO
2
-induced intracellular ROS
formation by encapsulation of particles with NaY zeolites
(TiO
2
@NaY). Sayes et al. reported that the structure of titania
nanoparticles correlates with toxicity (Fig. 4). In their studies with
dermal fibroblasts, rutile TiO
2
particles produced two orders of
magnitude less reactive oxygen species than similarly sized
anatase TiO
2
particles.
[136]
Using P25 TiO
2
(anatase/rutile 80:20),
Xia et al.
[137]
observed TiO
2
to generate ROS in a cell-free system
but not in murine macrophages (RAW 264.7). On the contrary,
SiO
2
nanoparticles doped with 1.6 wt % iron, cobalt, manganese,
and titanium displayed catalytic oxidative effects inside living
cells.
[138]
Human lung epithelial cells (A549) were exposed to
thoroughly characterized particles of the same morphology,
comparable size, shape, and degree of agglomeration to
determine the influence of chemical composition and catalytic
activity on ROS formation. These studies clearly showed that the
chemical composition of nanoparticles is a most decisive factor
influencing ROS formation in lung epithelial cells.
[138]
The role of
particle size, shape, and composition to induce oxidative stress in
primary mouse embryo fibroblasts was also evaluated for SiO
2
,
ZnO, CNTs, and CB.
[139]
Although all four nanomaterials induced
significant ROS generation and GSH depletion in a dose-
dependent manner, the effects were different with ZnO inducing
significantly more oxidative damage than the other nanomater-
ials. Since SiO
2
and ZnO had similar crystal shape and particle
size this further confirms that toxicity diversity can be attributed
to their chemical composition.
[139]
Recently, Park and Park
[140]
observed both, ROS formation and an increased level of nitric
oxide when macrophages (RAW 264.7) were exposed to SiO
2
nanoparticles and ROS formation in these cells may trigger
proinflammatory responses observed in vitro and in vivo. On the
contrary, Diaz et al.
[70]
did not always find a positive correlation
between cytotoxicity of SiO
2
nanoparticles and ROS formation in
human monocytes and mouse peritoneal macrophages. In vitro
toxicity screenings with CeO
2
nanoparticles revealed a dose-
dependent induction of ROS and a decreased level of intracellular
GSH in BEAS-2B as well as in A549 human lung epithelial
cells.
[141,142]
Commercial SWCNTs and MWCNTs (single-walled and
multi-walled CNTs) were found to induce a dose- and time-
dependent increase of intracellular ROSs in rat macrophages
(NR8383) and human lung epithelial cells (A549) that might be
related to metal traces present in manufactured nanotubes.
[143]
4.2.2. Cell Viability
Different endpoints for cell viability have been used in
nanomaterial toxicity testing. Metabolic activity, for instance,
has been widely determined using the colorimetric MTT assay
based on the reduction of a yellow tetrazolium dye (MTT) to a
purple formazan in cells bearing intact mitochondria. Recently,
however, the suitability of MTT for toxicity evaluation of CNTs has
been doubted since SWCNTs have been shown to deplete free
MTT thereby causing false-negative results.
[117]
Moreover,
numerous cytotoxicity studies are based on the detection of
intact lysosomes via neutral red uptake. Neutral red accumulates
in intact lysosomes of viable cells whereas it is excluded from
dead cells. The uptake of neutral red may be detected via
fluorescence or absorption measurement. Cellular necrosis is
another endpoint commonly used in cell viability studies. Upon
necrosis, significant amounts of LDH are released from the
cytosol. This LDH release can be easily detected using INT (a
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 11
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
yellow tetrazolin salt) as a substrate since LDH catalyzes its
oxidation to a red formazan (Scheme 3). Nanomaterial toxicity
leading to apoptosis is commonly assessed with caspase-3 assays.
Caspase-3 is activated in the terminal apoptotic cascade by
cleavage and this step can be detected by measuring the cleavage
of chromogenic or fluorimetric Caspase-3 substrates.
In a comprehensive study, Simon-Deckers et al. determined
the cytotoxicity of well-characterized metal oxide nanoparticles
and CNTs using different cell viability assays. Studying the
response of A549 human lung epithelium cells, Simon-Deckers
et al. found metal oxide nanoparticles (rutile or anatase TiO
2
and
Al
2
O
3
) to be less toxic than CNTs. Although all nanoparticles were
efficiently internalized in A549 cells, their cytotoxicity was
generally low with a maximum cell death rate of 25% for TiO
2
(MTT).
[144]
Since TiO
2
and Al
2
O
3
particles were of similar size
and shape but of different toxicity (with a maximum cell death
rate of 3% for Al
2
O
3
compared to 25% for TiO
2
) this study
revealed again that nanoparticle toxicity can be attributed to their
chemical composition.
[138,139,144]
In line with nanotoxicity data
previously published by Sayes et al.
[136]
Simon-Deckers et al.
[144]
reported that anatase TiO
2
was slightly more toxic than rutile
TiO
2
.
Redox activity in mouse neuroblastom cells has been shown to
decrease significantly when the cells were exposed to ZnO
whereas an exposure to other metal oxide nanoparticles such as
Fe
3
O
4
,TiO
2
,Al
2
O
3
, and CrO
3
had no measurable effect on the
cells.
[145]
Similarly, cell viability assays (MTT, LDH) using TiO
2
and metal nanoparticles (Co, Ni), did not reveal any significant
toxic effect on A549 cells.
[146]
Nanometer-sized and fine-sized
ZnO particles were also found to be more cytotoxic to L2 lung
epithelial cells than SiO
2
particles in LDH assays by Sayes et al.
[64]
However, a comparison of in vivo and in vitro measurements
demonstrated little correlation.
Lin et al. reported that SiO
2
nanoparticles reduce the viability of
human bronchoalveolar carcinoma-derived cells in a dose- and
time-dependent manner. The cytotoxicity of 15- and 46-nm SiO
2
nanoparticles was investigated by using crystalline SiO
2
as a
positive control. Both SiO
2
nanoparticles were more cytotoxic
than the bulk material; however, the cytotoxicities of 15- and
46-nm SiO
2
nanoparticles were not significantly different.
[147]
Moreover, cell viability of A549, endothelial EAHY926 cells, and
J774 monocyte-macrophages in response to SiO
2
particles was
found to be determined by their total mass, number and surface
area as well as by their concentration.
[148]
A time- and
dose-dependent effect of 20 nm-sized CeO
2
particles on cell
viability of A549 cells was reported by Lin et al. In their studies,
cell viability decreased to 53.9% when a CeO
2
concentration of
23.3 mgmL
1
was used.
[141]
Although a variety of cell viability studies using carbon
nanomaterials have been published so far, no coherent picture
has emerged yet. Davoren et al.
[115,149]
found a very low direct
cytotoxicity of SWCNTs in cell viability assays using A549 cells but
the same group reported later that SWCNTs display an indirect
cytotoxicity by depleting cell culture medium components.
Cytoxicity of MWCNTs was significantly higher in studies by
Simon-Deckers et al.
[144]
who observed a maximum cell death rate
of 40% (determined by LDH assays) neither depending on their
length, nor on their Fe impurities. Similarly, a dose- and
time-dependent decrease in cell viability of human epidermal
keratinocytes was found in studies conducted by
Monteiro–Riviere and Imnan
[116]
who used MWCNTs lacking
metal impurities. In contrast, Pulskamp et al.
[143]
did not observe
any acute toxicity on the viability of A549 cells exposed to
SWCNTs or MWCNTs but, as mentioned above, observed a dose-
and time-dependent increase of ROS formation presumably
associated with metal traces found in commercial carbon
nanotubes. These confounding findings may be due to
interference of the nanomaterials with the employed test systems.
Carbon nanomaterials have been reported to distort light
absorption and fluorescence measurements due to their optic
activity
[2]
and to interact with dyes and substrates used in classical
cell viability test systems.
[116,117]
CNTs in particular adsorb and
thereby deplete MTT leading to false negative test results. To
avoid this specific interference MTS was suggested as alternative
substrate for measuring metabolic activity as it did not interact
with CNTs
[125]
The interaction of MTS with other nanomaterials
is still to be tested. Further studies using MTS in addition to
multiple other cytotoxicity assays have to be performed for an
appropriate assessment of carbon nanomaterial toxicity.
4.2.3. Genotoxicity
For a review dedicated entirely to genotoxicity testing of
nanomaterials, the reader is referred to ref. [150]. In the
following, we focus on the most important property–effect
correlations for metal oxide nanomaterials. The classic comet
assay based on gel electrophoresis and the detection of in vitro
mammalian chromosomal aberrations are the most commonly
used test systems to assess genotoxicity. Using comet assays,
Wang et al.
[151]
found genotoxic effects of ultrafine TiO
2
particles
when cells were exposed to high particle concentrations (130 mg
mL
1
). In contrast, Warheit et al.
[12]
reported that ultrafine rutlile
TiO
2
and P25 TiO
2
(anatase/rutile 80:20) particles (of approx.
140 nm size) did not induce chromosome aberrations nor
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 4. Dose–response cellular viability of cultured human cells exposed
to nano-TiO
2
samples for 48 h. While overall the toxicity in culture was low,
different types of nano-TiO
2
did exhibit different levels of toxicity. Nano-
TiO
2
anatase particles were the most cytotoxic to human cells in culture,
while nano-TiO
2
rutile particles were the least cytotoxic, and two mixed
anatase/rutile nano-TiO
2
samples were in between. Figure reproduced with
permission from Sayes et al. [136].
12 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
displayed mutagenicity. Recently, Xu et al.
[152]
demonstrated that
different anatase TiO
2
particles (5 and 40 nm in size, respectively)
and fullerenes increased the mutation rate in mouse primary
embryo fibroblasts (MEF) in a dose-dependent manner. To
elucidate the mechanisms underlying TiO
2
genotoxicity, this
group also conducted studies using a nitric oxide synthase
inhibitor and a chemical inhibitor of cyclooxygenase-2 (COX-2).
Both nanomaterials lead to the formation of peroxinitrite anions
and induced kilobase pair deletion mutations that could be
protected by antioxidants. Furthermore, DNA damage could be
reduced via suppression of COX-2. COX-2 plays an important role
in cellular inflammation and genomic instability, and the particle
induced oxidative stress may activate the COX-2 signaling
pathway.
[152]
In another detailed study of manufactured nano-
particles (ZnO, SiO
2
,TiO
2
, CB, and SWCNTs), SWCNTs were
found to be more genotoxic than ZnO.
[139]
Since ROS production
induced by ZnO was significantly higher than compared to CNTs,
it was assumed that DNA damage induced by carbon nanotubes
can be attributed to mechanical injury rather than to an oxidative
effect. Furthermore, Yang et al.
[139]
provided evidence that DNA
nanoparticle genotoxicity might primarily be due to particle shape
rather than to chemical composition. Comet assays performed
with SiO
2
nanoparticles in two different laboratories using
cultured 3T3-L1 fibroblasts revealed no significant genotoxicity
but showed that in vitro toxicity testing can be quantitatively
reproducible.
[153]
Using comet assays, Jacobsen et al.
[154]
found different carbon
nanomaterials (CB and SWCNTs) to induce significant DNA
damage. However, MWCNTs did not show any mutagenic effects
in chromosome aberration studies using Chinese hamster
lung fibroblasts
[155]
or in bacterial reverse mutation assays.
[156]
Colloidal SiO
2
nanoparticles of different sizes (30, 80, 400 nm)
did not exert any genotoxicity in 3T3-L1 fibroblasts.
[153]
4.2.4. Inflammatory Response
To assess inflammation by nanomaterial immunotoxicity, the
production of inflammatory markers such as the chemokines
Interleukin-8 (IL-8), TNF-a, or IL-6 are usually measured in cell
culture supernatants using enzyme-linked immunosorbant assay
(ELISA). In rat model systems, the production of the inflamma-
tory cytokine MIP-2 (macrophage-inflammatory protein-2)
together with that of TNF-aand/or IL-6 are used as cytotoxicity
endpoints. Comparing the toxicity of rutile and anatase TiO
2
in
A549 cells, Sayes et al.
[136]
demonstrated an overall greater toxicity
of TiO
2
anatase nanoparticles (Fig. 4). Anatase TiO
2
nanoparticles
triggered a dose-dependent release of Il-8 in human dermal
fibroblasts (HDF) and A549 cells that was significantly lower
when the cells were exposed to rutile TiO
2
nanoparticles.
[136]
Ultrafine (P25 rutile/anatase 80:20) but not fine TiO
2
particles
were found to elicite Il-8 release in A549 cells indicating a
size-dependent effect of immunotoxicity. However, TiO
2
ultrafine
particles remained highly aggregated in cell culture as well as
inside the cells.
[157]
Inflammatory properties of TiO
2
particles
therefore appear to be driven by their specific surface area.
[157]
In a comprehensive study aimed to determine the importance
of surface area and surface reactivity of particles to induce
inflammatory responses, Duffin et al. used a variety of
manufactured particles, such as TiO
2
, CB, and metal nanoparticles
(Ni and Co) both for instillation and for treatment of A549 cells.
They observed a correlation between particle surface area dose,
specific surface activity, and the proinflammatory effects in vivo
and in vitro. Their study also demonstrated the utility of in vitro
assays for predicting the ability of nanoparticles to cause
inflammation in vivo on the basis of their surface area and
reactivity.
[146]
Recently, Herzog et al.
[158]
demonstrated that exposure of A549
or normal human bronchial cells to SWCNTs did not induce
inflammatory responses but can lead to the suppression of a
variety of inflammatory mediators including Il-8, Il-6, and MCP-1
(monocyte chemotactic protein-1) in vitro. In contrast, chemically
unmodified MWCNTs caused a dose-dependent Il-8 increase in
HEK cells.
[159]
Since carbon nanomaterials seem to be capable of
adsorbing a variety of substances including cytokines in the
culture medium, classical toxicity assays may not be appropriate
for assessing carbon nanoparticle toxicity.
[125,149,160]
In this context it is important to note that Veranth et al.
[161]
have
observed a significant change of Il-6 response to nanoparticle
treatment, either when different cell types were used or when
the same cell type was grown in different media. Moreover,
inflammatory responses to particles seem to be amplified by
contact-dependent interactions between alveolar macrophages
and epithelial cells.
[162]
Therefore, future studies determining
inflammatory effects of nanoparticles have to be conducted using
co-culture systems with defined cell types and media to generate
comparable data.
4.3. Original Results on the Cytotoxicity of TiO
2
As reviewed above, the available reports of nanomaterial in vitro
testing give a broad overview regarding possible toxicology
effects. However, a valid testing strategy is not available. Moreover
most of the data are not comparable due to a lack of validated test
protocols and a focus on only a few cell lines. Here we report
exemplarily in vitro data from the Nanocare project in vitro
screening strategy highlighting two critical aspects of reliable
nanomaterial in vitro testing: the required number of sensitive
cell lines and the selection of essential assays.
Six different stable cell lines (Supporting Information Table S3)
were exposed to dispersions of two different types of TiO
2
nanoparticles and were tested regarding the formation of ROS,
their metabolic activity, and cell death. The cell lines represented
six different mammalian organs. A549 and RAW264.7, two of the
most commonly used lung derived cell lines in in vitro toxicology,
represent the first line of exposure to inhaled nanoparticles.
While many studies are restricted to these two cell lines, we also
incorporated three cell types representative of other routes of
exposure. CaCo2 cells stem from a human colon carcinoma and
are characteristic for the colon epithelium, while NRK-52E cells
have been widely used as a model for the mammalian kidney
epithelium and have been established from a healthy rat kidney.
Furthermore, the skin is represented by HaCaT, a cell line isolated
from spontaneously transformed human epidermal keratino-
cytes. A sixth cell line, NIH-3T3, represents the fibroblast
phenotype and has been cloned from healthy mouse embryos.
NIH 3T3 is a widely used well-known model for sensitive in vitro
toxicology testing.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 13
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
We focused on standardized cell lines instead of using primary
cells to allow for a sound reproducability of results and high
throughput suitability of the test systems used.
Cells were exposed to 0.1, 1, and 10 mgcm
2
anatase/rutile
TiO
2
nanoparticles that originate from opposed synthesis routes.
TiO
2
A is precipitated in a wet chemical process (sol–gel),
while TiO
2
B is formed in flame pyrolysis of titanate salts.
Concentrations of nanoparticles above 10 mgcm
2
interfered
strongly with the assay systems which were based on optical
detection and were therefore neglected.
Dispersions of TiO
2
A nanoparticles did neither induce a
significant change in any of the three parameters studied nor in
any of the cell lines investigated (Figs. 5a, 6a, and Supporting
Information Fig. S3). TiO
2
B, on the other hand, provoked the
formation of ROS in all cell lines tested in three or more
independent experiments (Fig. 5b). The percent increase in ROS
formation was dependent on the cell line and the concentration of
TiO
2
applied. The mouse fibroblast cell line NIH 3T3 showed the
strongest increase in ROS formation after exposition to 10 mg
cm
2
TiO
2
B. The metabolic activity and the incidence of cell
death remained unaffected by TiO
2
B in all cell lines tested (Fig. 6
Figs. 6b, S3b). The responses described above are summarized in
Table 2.
It has been suggested that inhaled particles excert their adverse
effects primarily by triggering an inflammatory response which is
in turn mostly elicited by the formation of ROS by the particles
themselves and by the cellular stress response. The observation of
ROS formation is therefore a good indication of the inflammatory
potential of a given particle.
[163]
In the present study, TiO
2
B has
been shown to cause the formation of ROS in vitro and may
therefore trigger an inflammatory response in vivo.
It has been shown that the crystal structure of TiO
2
nanoparticles may influence there in vitro toxicity. Rutile TiO
2
nanoparticles triggered two orders of magnitude less ROS than
similarly sized anatase TiO
2
particles in dermal fibroblasts
(Fig. 4).
[136]
However, TiO
2
A, which did not influence any of the
parameters tested in vitro, consists mostly of anatase TiO
2
like
TiO
2
B. Both particles are in a similar size range which leaves the
organic modification detected on TiO
2
A as possible explanation
for the different biological activity of the two types of TiO
2
nanoparticles.
In contrast to the extensive inhalation studies presented in the
following section, this investigation was designed to provide
an exemplary insight into the necessity of using different cell
types and test systems when assessing the in vitro toxicity of
nanoparticles. The six cell types presented here displayed
individual degrees of ROS formation in the presence of TiO
2
B nanoparticles. While some cell lines, such as HaCaT and
RAW264.7, behaved more robust, NIH-3T3 seem to be more
sensitive to the exposition with TiO
2
B. Unpublished results of
our laboratory show that the cell type specific sensitivity is also
dependent on the nanoparticle applied. In line with this, Veranth
et al.
[161]
haven shown that inflammatory
responses to TiO
2
nanoparticles are influenced
by the cell type and culture conditions applied.
Furthermore, our results show that cell types
of the other routes of exposure may also be
affected by nanoparticles. As it has been shown
that inhaled nanoparticles may be translocated
into the body (e.g., ref.
[164
), the need for the
investigation of cell types representing other
organs than the lung becomes evident.
As described, concentrations of nanopar-
ticles above 10 mgcm
2
interfered with the
quantification of the chosen endpoints.
Consequently, in vitro methods and especially
those based on optical detection have to be
adapted with respect to interference with
nanoparticles and are limited regarding the
maximum applicable dose. A comparison to
adverse effects of high doses used in inhala-
tion studies is therefore impossible. In
various studies, higher concentrations of
nanoparticles (e.g., P25 and other TiO
2
particles) than those reported here have been
applied and found to induce strong effects.
However, it remains questionable if the
application of for instance 100 mgcm
2
yields
measurement artifacts or reliable results.
Based on our findings we argue that the
investigation of several parameters at lower
particle concentrations is preferable over the
application of high doses.
Taken together, our results and recently
published data
[125]
demonstrate that it is
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
Figure 5. Oxidative stress (expressed as mean DCF fluorescence [%]) in six different cell lines
upon exposition to dispersions of (10, 1, and 0.1 mgcm
2
) or stirred cell culture medium
(control). Standard deviations are indicated. a) TiO
2
A; b) TiO2 B; *significantly different from the
control at the 0.05 level; N
A549
¼32, N
RAW
¼21, N
other
¼28.
Figure 6. Cell death (measured by LDH activity, expressed as mean INT
reduced
absorption [%]) in
six different cell lines upon exposition to dispersions (10, 1, and 0.1 mgcm
2
) or stirred cell
culture medium (DMEM/10% FBS, control). Standard deviations are indicated. a) TiO
2
A;
b) TiO2 B.
14 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
necessary to use a minimum set of sensitive cell lines and
to consider several test systems as nanoparticles may exert
particle type specific adverse effects which will arise in different
endpoints.
4.4. Correlation of In Vitro Toxicity Data
A number of studies conducted with physicochemical character-
ization and multiple cytotoxicity assays showed that nanoparticle
toxicity can be attributed to size,
[33,147]
chemical composi-
tion,
[138,139]
surface,
[146,165]
and structure.
[136]
Currently, however, sufficient data enabling to predict
adverse effects of nanoparticles based on their physicochemical
properties are still missing. To allow an appropriate interpretation
of test results, it is not sufficient to characterize the intrinsic
properties of nanoparticles only since the interaction of
nanoparticles with physiological media will also influence the
level of toxic effects.
[51,115]
Furthermore, appropriate control
materials and validated protocols for the preparation of particle
suspensions should be used in future studies of nanoparticle
toxicity.
An increasing number of studies designed to analyze the
mechanisms underlying nanoparticle toxicity has been published
recently and evidence is accumulating that many toxic effects
derive from oxidative stress initiated by the formation of ROS.
The ROS generating capacity of nanoparticles seems to correlate
with their potential to induce cellular inflammation and DNA
damage.
[64,136,140,152]
Therefore, measurement of oxidative stress
potential can be regarded as an important and highly sensitive
component of a screening strategy for nanoparticle toxicity
assessment. However, intracellular ROS formation induced by
nanoparticles may not be predictive of all possible cytotoxic
effects. For SiO
2
particles and carbon nano-
materials, for instance, a positive correlation
between cytotoxicity and ROS formation could
not be found.
[70,139]
Multiple tests should
therefore be used in a comparative manner
to enable an appropriate evaluation of nano-
particle cytotoxicity.
Taken together, the presented in vitro
testing strategy may be suitable for predicting
the in vivo effects of nanomaterials. Currently,
however, there is little correlation between
qualitative in vitro data generated in different
laboratories which might result from a lack of
adapted in vitro test systems. Furthermore,
in vitro test systems display a lower complexity
than living organisms and the transfer of doses
applied in vitro to in vivo exposure scenarios is
hardly possible. For an appropriate design of in
vivo experiments, standardized in vitro testing
will be of considerable value.
5. In Vivo Studies With
Engineered Nanomaterials
5.1. Review of Pulmonary Toxicity Studies
With Engineered Nanomaterials
Adverse health effects of air pollution have
been recognized in epidemiological studies.
Part of the pollution is Particulate Matter,
mostly black carbon (see Section 2), and has
been linked with cardiovascular effects and
pulmonary toxicity.
[166–168]
Here we focus on pulmonary toxicity of
engineered nanomaterials, and since study
designs are not standardized yet, we report
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
Table 2. Summary of cellular reactions upon the in vitro exposure to
dispersions of TiO
2
nanoparticles.
TiO
2
ATiO
2
B
Reactive oxygen species None þ
Metabolic activity None None
Cell death None None
Scheme 4. Work flow of in vivo inhalation studies for nanomaterials. Aerosols are generated
from the nanomaterials (Supporting Information Fig. S2) and monitored (Table S2)
[93]
in an
inhalation chamber, typically with head-only exposure. The study design of the short-term
inhalation test developed by the authors
[76]
is shown on the bottom. X, head–nose exposure
to aerosols for 6 h day
1
on five consecutive days; R, post-exposure time; H, histology of selected
organs (especially lungs slices, as shown bottom right) including cell proliferation and apoptosis;
e, examinations of blood and bronchoalveolar lavage fluid (as shown on bottom left).
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 15
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
them together with the results. Three questions need to be
answered in order to assess pulmonary toxicity:
(i) How can nanomaterials get into the lung?
(ii) What effects do nanomaterials cause in the lung?
(iii) Can nanomaterials translocate from the lung to other tissues
and affect them?
Inhalation studies with animals are the best possible
approximation to the exposure of the human respiratory tract
to nanomaterials. But inhalation studies entail the sacrifice of
animals, are quite expensive, consume up to 1 kg of test material
and they require care with regard to aerosol preparation.
Instillation remediates the last two drawbacks and replaces the
aerosol by a suspension; however, nanomaterials in suspension
have different surface properties, different states of agglomera-
tion and deposit differently in the lung compared to nanomater-
ials in aerosols. If these constraints are taken into consideration,
the pulmonary exposure by intratracheal instillation, pharyngeal
or laryngeal aspiration is only acceptable for hazard identification.
Endpoints of concern for pulmonary exposure are organ-specific
markers of inflammation, oxidative stress, cell proliferation, and
histopathology in the lung as well as measurement of damage to
non-pulmonary organs.
5.1.1. Metal-Oxide Nanomaterials
Based on their own intratracheal instillation studies and literature
review, Donaldson et al.
[165,169]
came to the conclusion that
ultrafine particles made of low-solubility, low-
toxicity materials are more inflammogenic in
the rat lung than fine, respirable particles
made from the same material, which is driven
by their surface area. In more details, initial
findings with nanoparticles after inhalation
were published by Heinrich et al. in 1995,
[170]
describing an increased lung tumor incidence
after long-term exposure to ultrafine titanium
dioxide (uf-TiO
2
) P25 and CB. Bermudez
et al.
[171,172]
performed a multispecies, sub-
chronic, inhalation study comparing pulmon-
ary responses to a uf-TiO
2
P25 (average
primary particle size of 21 nm). Female rats,
mice, and hamsters were exposed to aerosol
concentrations of 0.5, 2.0, or 10 mg m
3
TiO
2
nanoparticles for 6 h day
1
, 5 days week
1
, for
13 weeks. Following the exposure period,
animals were held for recovery periods of 4,
13, 26, or 52 weeks (49 weeks for the
uf-TiO2-exposed hamsters) and, at each time
point, TiO
2
burdens in the lung and lymph
nodes and selected lung responses were
examined. Inhalation of 10 mg m
3
TiO
2
nanoparticles for 13 weeks resulted in pul-
monary overload in rats and mice but not in
hamsters in which the lung burdens were
approximately only 23% of lung burdens of
other species. While there were various
responses in mice and rats, hamsters had
very limited responses probably due to the low
lung burdens and rapid clearance of particles in these animals.
Responses in mice were limited to animals exposed to
10 mg m
3
, whereas in rats responses were also observed in
animals exposed to 2 mg m
3
. The magnitude and spectrum of
responses were, in general, equivalent in rats and mice. The
extent and character of the inflammatory responses in rats
differed from that observed in mice; in rats the responses had a
greater neutrophilic component that diminished over time,
whereas in mice significantly increased neutrophil and macro-
phage numbers remained relatively constant. Histopathological
examination of rats and mice uncovered progressive fibro-
proliferative lesions in rats but not in mice. Taken together, the
species-specific differences observed in this study are well in line
with results of previously reported chronic exposure studies with
rats and mice and poorly soluble particulates. They suggest that
susceptibility of the rat to the induction of lung tumors by
pulmonary overloads is related both to dosimetry and biological
response. The authors concluded that the findings of this
multispecies study were consistent with the results of a
companion study using inhaled pigmentary (fine mode)
TiO
2[171]
and demonstrated that the pulmonary responses of
rats exposed to nanoparticle concentrations likely to induce
pulmonary overload are different from similarly exposed mice
and hamsters. Different types of TiO
2
nanomaterials were tested
by Warheit et al.
[12]
by intratracheal instillation in rats examining
pulmonary effects in the BALF and lung tissue up to three
months post-exposure. The TiO
2
nanomaterials had different
crystal structures and surface coatings, showing differential
1
23
4
5
6
7
8
9
10
11
4
56
78
910
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Figure 7. Inhalation of 2–5 nm TiO
2
nanoparticles, dark field micrographs of lung tissue (A,B)
and alvelolar macrophages (C,D). Mice exposed acutely to 0.77 or 7.22 mg m
3
nanoparticles
demonstrated minimal lung toxicity or inflammation. Mice exposed subacutely (8.88 mg m
3
)
and necropsied immediately and at week 1 or 2 post-exposure had higher counts of total cells and
alveolar macrophages in the BAL fluid compared with sentinels, indicating a significant but
moderate inflammatory response. However, mice recovered by week 3 post-exposure. Other
indicators were negative. Figure reproduced with permission from Grassian et al. [173].
16 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
pulmonary effects based on these properties. The rutile and
surface treated TiO
2
showed only transient and fast reversible
pulmonary inflammatory responses, whereas inflammatory
effects, cytotoxicity, and adverse lung effects were described
with the anatase/rutile TiO
2
. Grassain et al.
[173]
performed acute
as well as subacute inhalation studies (whole body exposure) with
TiO
2
nanomaterials (2–5 nm diameter, pure anatase) in mice
(Fig. 7). Mice exposed for 4 h to 0.77 or 7.22 mg m
3
titanium
dioxide nanoparticles demonstrated only minimal lung toxicity or
inflammation. Mice exposed subacutely (4 h day
1
for 10 days) to
8.88 mg m
3
(only one concentration) and examined immediately
or 1 or 2 weeks after the last exposure had higher counts of total
cells and alveolar macrophages in the BAL fluid, no effects were
seen after 3 weeks recovery. No effects were found in lung
histopathology or any other clinical parameter. A recent study by
Sager and Castranova exposed rats via intra-
tracheal instillation to various doses of ultra-
fine and fine CB and TiO
2
, all low toxicity and
low solubility materials. Ultrafine TiO
2
was
more bioactive than ultrafine CB at equivalent
surface area, supporting the hypothesis that
surface area, not mass is a more appropriate
dose metric to assess pulmonary inflamma-
tion.
[174]
Previously, Heinrich et al.
[170]
reported an increased tumor incidence in
inhalation studies using CB and pulmonary
inflammation. Moreover, increased chemo-
kine and mutagenic responses after three
months inhalation exposure were described by
Driscoll et al.
[175]
For instance, they observed
that mutations in the hprt gene of alveolar
epithelial cells encoding the hypoxanthine–
guanine phosphoribosyltransferase occurred
only after CB exposures and resulted in
significant inflammation and epithelial hyper-
plasia. The no observed adverse effect con-
centration (NOAEC) in this study, for this
specific material (particle size 0.016 mm,
220 m
2
g
1
) was 1 mg m
3
.
5.1.1. Carbon-Based Nanomaterials
Similar to the study performed by Bermudez
et al.,
[171,172]
the inhalation toxicity of low
(37 m
2
g
1
) and high surface area (300 m
2
g
1
)
CB was examined in rats, mice, and hamster
after 3-month inhalation exposure.
[176]
Again,
rat was the most sensitive species. The NOAEC
for high surface area CB was 1 mg m
3
for rat.
The low surface area CB did not exert any
toxicity at the 50 mg m
3
.
In 2006, Lam et al.
[177]
compiled the animal
studies available with CNTs. By this date, data
from inhalation studies were not available,
only those from studies on intratracheal
instillation. The animal studies of CNT
pulmonary toxicity showed that CNTs are
capable of inducing inflammation, epithelioid
granulomas, fibrosis, and biochemical toxicity
changes in the lungs that might impair pulmonary functions.
Systematic reduction of the metal content of CNTs did not
eliminate their inflammation potential (Fig. 8).
[177]
Muller et al.
took a similar approach and systematically modified the structural
defects of CNTs. They found that the acute pulmonary toxicity was
reduced upon heating but restored upon grinding, indicating that
the intrinsic toxicity of CNT is mediated by the presence of
defective sites in their carbon framework.
[178,179]
However, the
studies reviewed here were conducted using intratracheal
instillation or modified techniques to administer CNT suspen-
sions to rodents that had been mechanically ultrasonicated. In a
much disputed publication, Poland et al. reported a pilot study
[180]
in which they administered different CNTs by intraperitoneal
application to mice. They reported that CNTs may behave in this
test system comparable to asbestos. However, the relevance of
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Figure 8. Instillation of carbon-based nanomaterials: Lung tissues from mice after a single
intratracheal instillation of unprocessed iron containing HiPco SWCNTs (D: 7 days, E: 90 days
observation), purified CNTs of the same product (F), laser-produced CNTs (B), nickel and yttrium
containing electric-arc CNTs (C), with references CB (A, low toxicity) and quarz (fibrogenic). All
CNT samples tested, regardless of the type and amount of metal impurities they contained,
induce dose-dependent lesions characterized chiefly by interstitial granulomas in the lungs of
mice in the 7-day (C, D) and 90-day group (E, F). Granulomas were not observed in rodents
exposed to CB. The authors concluded that if CNTs reach the lungs, they are much more
toxic than CB and can be more toxic than quarz. Figure reproduced with permission from Lam
et al. [177].
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 17
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
these results for inhalation toxicity is not clear. Even though
intratracheal instillation and modifications of this technique are
common routes of administration used to assess the toxicity of
dust in lungs, the authors concluded that inhalation toxicity
studies are imperative. Only inhalation can demonstrate whether
CNTs can reach the lung to produce those lung lesions that were
observed in the intratracheal instillation studies.
We published in 2009 the first subchronic inhalation study
(OECD TG 413) with MWCNTs.
[181]
Wistar rats were head–nose
exposed for 6 h day
1
, 5 days week
1
, 13 weeks, total 65
exposures, to MWCNT concentrations of 0, 0.1, 0.5 or
2.5 mg m
3
. Highly respirable dust aerosols were produced with
a proprietary brush generator which neither damaged the tube
structure nor increased ROS on the surface. Inhalation exposure
to MWCNTs produced no systemic toxicity. However, increased
lung weights, pronounced multifocal granulomatous inflamma-
tion, diffuse histiocytic and neutrophilic inflammation, and
intra-alveolar lipoproteinosis were observed in lung and
lung-associated lymph nodes at 0.5 and 2.5 mg m
3
. These
effects were accompanied by slight blood neutrophilia at
2.5 mg m
3
. Incidence and severity of the effects were concen-
tration-related. At 0.1 mg m
3
, there was still minimal granulo-
matous inflammation in the lung and in lung-associated lymph
nodes.
After a single acute inhalation exposure of 30 mg m
3
for 6 h,
MWCNTs reached the subpleura in mice.
[182]
CNTs were
embedded in the subpleural wall and within subpleural
macrophages. Mononuclear cell aggregates on the pleural surface
increased in number and size after 1 day and nanotube-
containing macrophages were observed within
these foci. Subpleural fibrosis unique to this
form of CNTs increased after 2 and 6 weeks
following inhalation. None of these effects was
seen in mice that inhaled CB nanoparticles or a
lower dose of CNTs (1 mg m
3
).
[182]
The local
and systemic responses in C57BL/6 mice after
exposure via pharyngeal aspiration to CNTs
(single- and multi-wall) can be evaluated with
combination of blood gene expression and
circulating soluble protein analysis to identify
novel biomarkers.
[183]
In summary, the available evidence sup-
ports that there is a threshold level for effects
by inhalation of CNTs fibers, but the physio-
logical effects are qualitatively different and the
threshold is significantly lower than for
inhalation of metal-oxide particles. Material-
specific differences between different metal
oxides are addressed in Section 5.2.
5.1.3. Other Nanomaterials
Ji et al.
[184]
exposed rats via inhalation to silver
nanoparticles for 6 h day
1
, 5 days week
1
, for
a total of 4 weeks. The atmosphere was
generated by a device that generates silver
nanoparticles by evaporation/condensation
using a small ceramic heater. As such, the
generator was able to distribute the desired
concentrations of silver nanoparticles to
chambers containing experimental animals. The male and
female rats did not show any significant changes in body weight
relative to the concentration of silver nanoparticles during the
28-day experiment. Additionally, there were no significant
changes in the hematology and blood biochemical values in
either the male or the female rats. Therefore, the authors
concluded that the current silver dust limit (100 mgm
3
) fixed by
the American Conference of Governmental Industrial Hygienists
(ACGIH) did not display any significant adverse health effects. At
the highest dose, increased silver concentrations were measured
in the lung, liver, brain, and olfactory bulbus. The same group
performed a 90-day inhalation study.
[185,186]
Although no
statistically significant differences were found in the lavage
parameters, histopathological examinations indicated increases
in lesions related to silver nanoparticle exposure, such as infiltrate
mixed cell and chronic alveolar inflammation, including
thickened alveolar walls and small granulomatous lesions.
Since some of these effects were also observed in the control
animals the outcome of this study is inconclusive.
[185,186]
Different surface functionalizations on the same metal oxide
nanoparticle—a standard procedure in nanocomposite produc-
tion for compatibilization of nanomaterials with an organic
matrix—have received less attention. Initial results by Warheit
et al.
[187]
indicate that the inhalation hazard from TiO
2
particles
does depend on their surface chemistry, but the TiO
2
core þSiO
2
or Al
2
O
3
shell structures studied there are not representative for
the typical inorganic core with polymer functionalization. Much
more on the effects of polymer functionalization could be learnt
from the biotech community. Most pharmaceutical work is
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Figure 9. Nanomedicine controls biodistribution by surface functionalization. Quantum dots
were functionalized with a systematically increasing chain length of PEG and injected intrave-
nously in mice, then quantified by gamma labeling [47]. This has no relation to the exposure by
inhalation, but it exemplifies the importance of surface chemistry for biokinetics. Radioscinti-
graphic gamma images of intact animals 4 h post-injection. a) A 4 unit PEG chain directs the
nanoparticulate contaminations within minutes to the bladder, resulting in efficient clearance.
b) The nanoparticles with 14 PEG units were instead excreted through the liver to the intestines
with little uptake in the kidneys and bladder. Figure redrawn with permission from ref. [47].
18 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
devoted to (bio-)polymer spherical particles and to intravenous
injection.
[34]
This includes the development of quantum dots for
targeted imaging (Fig. 9) and lipid nanoparticles for drug delivery,
where the type and pattern of protein adsorption determines
indeed the organ distribution.
[44]
Intravenous delivery gives no
hints on the uptake and fate from inhalation exposure, and also
the biophysical interactions are different: as one example, the
radius of curvature is well defined for polymer particles and
quantum dots, but the irregular shapes of metal oxide
nanoparticles have many radii of curvature on a single particle,
and an even broader ensemble distribution. It is at present
unclear to what extent the biodistribution mechanisms of
functionalized polymer particles can be transferred to the
industrially relevant metal oxide nanoparticles, but clearly their
surface functionalization is a future way to control their fate in
organisms.
5.2. New Results Generated by a Short-Term Inhalation Study
for Nanomaterials
Previous studies suggest that short-term animal exposures to
synthetic amorphous silicas and crystalline silica can provide
toxicity data comparable to those of 90-day studies.
[177]
In the
frame of in-house and collaborative projects, we developed a
design for a short-term inhalation test in rats with sufficient
power and robustness to allow the prediction of potential adverse
effects as accurately as long-term exposure tests do
[78]
and provide
a benchmark for the development of appropriate in vitro test
systems.
5.2.1. Aerosol Generation
The results of the atmospheric concentrations and the particle
size analysis are presented in Table S5 (for detailed method
description see Section 7.4). Overall, the concentrations for all test
substances were maintained throughout the study. According to
the result of the cascade impactor measurements (Supporting
Information Fig. S2), respirable aerosols were produced for all
tested materials. For MWCNTs, the measured particle size by
optical particle counter (OPC) and SMPS was not reported
because the physicochemical properties (e.g., black, wool-rope
like, conductive) interfere with the measurement principle of
these devices.
5.2.2. Organ Distribution and Biological Effects
The observed biological effects are summarized in Table 3, and
the organ distribution is listed in the Supporting Information
Table S4.
TiO
2
B50mgm
3
nano-TiO
2
B resulted in an overload of the
lung and caused an increase in lung weight. Light microscopic
examination of the respiratory tract revealed effects (e.g.,
histiocytosis, Fig. 10b), which— at a low grade— were considered
a normal and reversible response of macrophages to remove
particulate matter from the lung. One of five animals in the
50 mg m
3
concentration group showed a minimal multifocal
infiltration with neutrophils. The bronchioli and bronchi of the
animals exposed to 50 mg m
3
nano-TiO
2
B displayed a minimal
to mild increase of epithelium thickness which was interpreted as
hypertrophy and/or hyperplasia. Nano-TiO
2
B induced concen-
tration-related inflammation reaction in the lung. Lung inflam-
mation was associated with concentration-dependent increases in
BALF total cell and neutrophil counts, total protein content,
enzyme activities, and levels of a number of cell mediators
(Fig. 10c). The effect was minimal, but significant at 2 mg m
3
nano-TiO
2
B, thus giving a low observed adverse effect
concentration (LOAEC), indicating that the NOAEC is close to
this concentration. We found no indications of systemic effects by
measurement of appropriate clinical pathology parameters. All
effects in the 2 and 10 mg m
3
concentration group were
reversible and partly reversible in the 50 mg m
3
concentration
group within the 16 day recovery period (Fig. 11).
The toxicity of TiO
2
after subchronic (90-day) inhalation
exposure has previously been studied.
[176]
The results of our
short-term study and the 90-day study are overall comparable. The
LOAEC in the 90-day study was an atmospheric concentration of
2mgm
3
determined by cell proliferation rate after the exposure.
The same LOAEC was achieved in the current study determined
by examination of the lavage fluid.
In comparison to the studies with nano-TiO
2
, 5-day inhalation
exposure to 274 mg m
3
pigmentary TiO
2
led to a more than 30%
increase of the lung weight. Again, diffuse histiocytosis was
noted, but without granulocytic infiltration. In three out of six
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Table 3. Toxicological findings after inhalation of aerosols from nanomaterials.
Material Aerosol
concentration
[mg m
3
]
No or low effect
concentration
(NOAEC/LOAEC)
Clin. Path. Pathology Reversibility Translocation
TiO
2
B 2, 10, 50 LOAEC: 2 mg m
3
(only very mild effects)
Inflammation Histocytosis Not complete No indication
ZrO
2
0.5, 2.5, 10 NOAEC: 10 mg m
3
No effects n.d. – n.d.
CeO 0.5, 2.5, 10 LOAEC: 0.5 mg m
3
Inflammation n.d. Not complete n.d.
SiO
2
0.5, 2.5, 10 NOAEC: 10 mg m
3
No effects No effects – No indication
SiO
2
functionalized 0.5, 2.5, 10 NOAEC: 10 mg m
3
No effects No effects – Yes (spleen)
ZnO 0.5, 2.5, 10 LOAEC: 0.5 mg m
3
(only very mild effects)
Early inflammation/
necrosis
Lung: inflammation/
cell death; nose: necrosis
Yes Yes (ions from dissolution)
CB 0.5, 2.5, 10 NOAEC: 10 mg m
3
No effects No effects – No indication
MWCNTs 0.1, 0.5, 2.5 NOAEC: 0.1 mg m
3
Inflammation Inflammation No No indication
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 19
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
animals, the mediastinal lymph nodes were activated, and
pigment-loaded macrophages were found in four out of six
animals. Single animals displayed very few particles on the
surface or intracellularly in the olfactory epithelium of the nasal
cavity. After the recovery period, the numbers of infiltrating
histiocytes (only focal infiltrates present) as well as particle
numbers decreased, which was reflected in a no longer
significantly increased lung weight. The
mediastinal lymph nodes of five out of six
animals showed activation, and in the lymph
nodes of all animals pigment-loaded macro-
phages were observed.
ZrO
2
did not show any effects in the lung or
in other organs up to the highest concentration
tested (10 mg m
3
, NOAEC). No indications of
systemic effects could be found by measure-
ment of appropriate clinical pathology para-
meters and there was also no indication of a
translocation of the inhaled material from the
lung into other organs.
CeO
2
, like TiO
2
, induced a concentration-
related inflammation reaction in the lung,
which was associated with dose-dependent
increases in BALF total cell and neutrophil
counts, total protein content, enzyme activ-
ities, and levels of a number of cell mediators
(Supporting Information Fig. S4). The effect
was still observed at the lowest tested
concentration of 0.5 mg m
3
(LOAEC). No
indications of systemic effects could be found
by measurement of appropriate clinical pathol-
ogy parameters. The observed effects were
only partly reversible within the 16-day
recovery period.
SiO
2
and functionalized SiO
2
—both SiO
2
materials did not show any effects in the
lung up to the highest tested concentration
of 10 mg m
3
(NOAEC). For the non-
functionalized SiO
2
, no indications of systemic
effects could be found and there was also no
indication of a translocation of the inhaled material from the lung
into other organs. The functionalized SiO
2
material was, however,
detected in the spleen (Fig. 12), indicating some translocation of
the material from the lung to this organ. Additionally, the spleen
was significantly enlarged without any other pathological
findings. The enlargement was greater than the mass of the
SiO
2
material deposited in the spleen and probably represented
some reaction of this organ to the SiO
2
material. There were, however, no other
findings in the spleen and no findings in
other organs than the spleen. While crystalline
quartz particles exhibited a strong toxicity,
[188]
the toxicity of non-cristalline SiO
2
particles
seemed to depend on their production process
suggested by a 5-day inhalation study on
three synthetic amorphous SiO
2
nanoparticles.
These three substances were produced either
by precipitation, sol–gel, or by pyrogenic
processes. The target concentrations were 1,
5, and 25 mg m
3
. In the lavage, adverse
effects (increased neutrophiles, increased
activities of LDH, NAG, increased protein
concentrations) were detected in all test groups
at 25 mg m
3
. The NOAEC for the two
substances produced by wet processes was
5mgm
3
. The test substance Car-O-Sil M5,
1
2
3
4
5
6
7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
Figure 11. Time–effect diagram: Effects of 5-day inhalation of TiO
2
B in rats: Changes (fold of
control) of the BALF parameters 3 days after the last exposure. Control values are defined as 1.
The axis is in logarithmic scale.
Figure 10. Effects of 5-day inhalation of TiO
2
B in rats. a) Lung sections, stained with hematoxylin
and eosin, from a control rat and (b) from a rat exposed to 50 mgm
3
TiO
2
B by inhalation for 6 h
day
1
for 5 days. Treatment with 50 mg m
3
TiO
2
B resulted in a minimal to mild diffuse
histiocytosis. The test material was present in the cytoplasm of alveolar macrophages (arrow). c)
Concentration–effect diagram: Changes (fold of control) of the BALF parameters 3 days after the
last exposure. Control values are defined as 1. The axis is in logarithmic scale.
20 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
which is produced by pyrogenic process, seems to have a higher
inflammatory potential, with a NOAEC of 1 mg m
3
. The
toxicological findings were reversible for all three
substances.
[189,190]
ZnO induced a concentration-related inflammation reaction in
the lung, which was associated with dose-dependent increases in
BALF (Supporting Information Fig. S5). In addition to the
inflammation reaction, necrosis was detected in the lung and the
nose. As ZnO is soluble and zinc ions are cytotoxic at higher
concentrations,
[191,192]
the necrosis can be attributed to the zinc
ions dissolved from the ZnO particles. Likewise, elevated zinc
levels were detected in various organs, most likely due to zinc ions
dissolved from the ZnO particles. There was, however, no
indication of systemic effects in other organs than the lung and
the airways.
CB did not show any treatment-related adverse effects at any of
the concentrations tested. Thus, the NOAEC for these materials
was the highest concentration test of 10 mg m
3
. One subchronic
inhalation study with a similar test material was previously
reported.
[171]
In this study, the only tested concentration of
50 mg m
3
was a clear effect level, which was much higher than
the concentrations tested in the current study. Thus, no statement
concerning the predictability of the short-term study can be made
for CB at this time point.
MWCNT exposed animals showed no clinical signs different
from normal. The mean body weight and the mean body weight
change were not significantly different from the control groups.
MWCNTs at the tested high concentration of 2.5 mg m
3
caused
significantly increased absolute (þ11.5%, p<0.01) and relative
(þ11.4%, p<0.01) lung weights. Increased relative lung weight
(þ10.5%, p<0.05) was still determined at the intermediate
concentration of 0.5 mg m
3
. The increases were observed only
directly after the exposure but not after the 3-week recovery
period.
Examination of BALF (Fig. 13c) showed moderate increases of
the polymorphonuclear neutrophils, total protein, and some
enzymes in the 0.5 mg m
3
and in the 2.5 mg m
3
concentration
group. Furthermore, there was a slight but not statistically
significant increase of polymorphonuclear neutrophiles (PMN)
even at the lowest concentration of 0.1 mg m
3
. The total cell
count in BALF was significantly increased in all concentration
groups. These effects were still present, though less pronounced,
after a 3-week recovery period.
During necropsy, all treated animals sacrificed directly after the
last treatment showed black fibrous particles within the alveolar
macrophages (Fig. 13a). In the 0.5 and 2.5 mg m
3
concentration
group, the number of alveolar macrophages was slightly
increased, and the animals of the 2.5 mg m
3
concentration
group showed a diffuse infiltration compared to the other groups
indicating a multifocal distribution pattern. One animal of the
2.5 mg m
3
dose group revealed a minimal granulomatous
inflammation, containing the fibrous particles within the
inflammatory lesion which were interpreted as MWCNTs.
Besides the above mentioned findings, five out of six animals
of the 2.5 mg m
3
dose group and one animal of the 0.5 mg m
3
dose group revealed a minimal granulomatous inflammation
after the 21-day recovery period. Particles were again observed
inside the lesion (Fig. 13b). Based on the changes in the BALF,
the NOAEC was 0.1 mg m
3
for MWCNTs. The effect was not
reversible within the recovery period.
Our results indicate a high inflammatory potency of MWCNTs.
The effects were not reversible but progressive. At 0.1 mg m
3
,
slight effects were still present. Furthermore, the results were in
good correlation with a 90-day inhalation study of MWCNTs
performed in our laboratories.
[181]
We have submitted our results
to the US EPA under TSCA 8e.
[193]
Apart from our 5- and
90-day inhalation studies and the 1-day inhalation study by
Ryman-Rasmussen et al.,
[182]
studies on lung toxicity of MWCNTs
were only performed by intra-tracheal instillation. There were
strong positive correlations between our findings after 5-day
inhalation exposure with 3 weeks post-exposure and those
intra-tracheal instillation studies after 90 days post-exposure
observation.
5.2.3. Correlations Between the Inhalation Results
The aerosol generation from nanomaterials and the short-term
inhalation study for nanomaterials have been established in
previous studies
[76,93]
and the resulting test system was used to
investigate the inhalation toxicity of eight different nanomaterials
in the present study. Selected test concentrations were in the
range of internationally established workplace limit values for
so-called inert or nuisance dust. For each nanomaterial, a LOAEC
or NOAEC was defined along with the characterization of the toxic
effects in the lung and translocation and effects in other organs
(Tables 3 and S4). These results provide essential information for
the safe production and use of these nanomaterials, allowing to
define save exposure levels during production and handling, thus
ensuring save production and use. This may include abandoning
of certain applications or trigger the selection of less toxic
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 12. TEM image of the spleen-section of an animal exposed to
10 mg m
3
SiO
2
acrylate-functionalized nanomaterial. Inorganic particles
are observed within the white pulpa of the spleen (arrows). These particles
could not be detected in control animals.
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 21
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
nanomaterials—and in fact both has happened based on the
results published here.
The results from eight inhalation studies with different
nanomaterials presented here represent the largest dataset on
nanomaterial inhalation toxicity published so far. Although still
limited, this data set allows for some preliminary correlations.
TiO
2
,CeO
2
, ZrO
2
, and SiO
2
are insoluble nanomaterials of
similar particle size and shape, yet the toxic concentrations varies
between NOAEC <0.5 to >10 mg m
3
. Obviously, the chemical
composition influenced the toxicity and not-or not only-the size or
shape of the material.
CB and MWCNTs are both carbon-based nanomaterials,
yet very different in shape and structure. The toxicity in the
short-term inhalation study with rats differs by a factor of
100 with a NOAEC of 0.1 mg m
3
for MWCNTs and 10 mg m
3
for CB.
ZnO and TiO
2
particles were tested as nanometer-sized and as
fine-sized particles. Most of the nanomaterial agglomerated to
particles similar in size to the micrometer-sized material,
resulting in similar particle sizes for both materials. Yet, the
agglomerates of TiO
2
and ZnO nanometer-sized particles
exhibited stronger effects than the solid fine-sized particles of
TiO
2
and ZnO, respectively. Studies with TiO
2
did not reveal any
deagglomeration in the body
[76]
and hence the differences in
toxicity between solid particles and
agglomerates are most likely due to the
fact that the agglomerates are built from
nanoscaled primary particles with a differ-
ent inner structure and higher specific
surface area of the agglomerates compared
to the solid particles.
CeO
2
and ZnO generated aerosols of
similar characteristics. Yet, ZnO is soluble
and releases zinc ions after deposition in
the body, whereas CeO
2
is unsoluble. Both
nanomaterials had the same LOAEC in
the short-term inhalation studies, yet
displayed different effects. The additional
necrosis found with ZnO may be attrib-
uted mainly to the toxicity of released zinc
ions.
Non-functionalized and functionalized
SiO
2
are very similar in shape and size with
the functionalized material having been
derived from the non-functionalized mate-
rial. While both nanomaterials caused no
lung effects in the short-term inhalation
study, the functionalized material, unlike
the non-functionalized one, was translo-
cated to the spleen. The surface modifica-
tion led to striking differences in biological
effects compared to the non-functionalized
SiO
2
nanomaterial.
6. Correlation of In Vitro and
In Vivo Data
The suitability of in vitro test systems to
predict potential adverse health effects of
nanomaterials is still a matter of discussion. To date, only a few
comparative studies have been performed using the same
nanomaterials for toxicity assessment both in vitro and in vivo
and some of these studies provided contradictory results. For
example, Sayes et al.
[64]
found little correlation between in vitro
and in vivo pulmonary toxicity of different fine- and nanome-
ter-sized particles (SiO
2
, ZnO). On the contrary, Donaldson
et al.
[165]
reported a concordance between the in vivo and in vitro
dosimetry of different low-toxicity, low-solubility particles (TiO
2
,
BaSO
4
) based on the surface area dose and discussed in vitro
studies as a valuable complement to animal studies. More
recently, Park and Park
[140]
observed oxidative stress and
proinflammatory responses induced by amorphous SiO
2
nano-
particles both in mice and in mouse macrophage cell lines.
Here, we have collected and analyzed in vivo and in vitro
screening data from the literature as well as from our studies on
the same well-characterized and categorized nanomaterials.
When comparing these data, several correlations of potential
adverse effects induced by nanomaterials in vitro and in vivo
can be found. For instance, the ROS generating capacity of
nanomaterials in vitro seems to correlate with their potential to
induce inflammation in vivo (this study,
[140,152,163]
). Furthermore,
the relationship between crystal structure of TiO
2
nanoparticles
and its biological effects have been demonstrated in several in
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 13. Effects of 5-day inhalation of MWCNTs in rats. a,b) Lung sections of animals exposed to
2.5 mg m
3
MWCNTs, stained with hematoxylin and eosin. a) Alveolar macrophages containing black
particles within the cytoplasm (arrows) were observed in rats immediately after the exposure. b)
Minimal granulomatous inflammation (arrow) with black particles inside the lesion was noted after
the 3-week recovery. c) Concentration–effect diagram: Changes (fold of control) of the BALF
parameters 3 days after the last exposure. Control values are defined as 1. The axis is in logarithmic
scale.
22 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
vitro and in vivo studies with anatase TiO
2
inducing more adverse
effects than rutile TiO
2
.
[12,136,144,173]
Data generated by in vitro
and in vivo studies suggest that inflammatory responses to
nanometer-sized particles are increased in comparison to larger
particles of the same chemical composition and that the total
surface area is the quantity that drives inflammation/inflamma-
tory responses.
[31,146,157,165,169]
Moreover, solubility of ZnO
nanoparticles seems to be responsible for inducing inflammatory
responses and necrosis both in vitro
[139,145]
and in vivo (this
study). Finally, several in vitro studies in correlation with the
inhalation study reported here provide evidence that some
nanoparticle toxicity can be attributed to their chemical
composition.
[138,139,144]
Taken together, in vitro and in vivo studies have identified the
same particle specific properties as crucial determinants of
adverse effects indicating that in vitro tests provide a preliminary
yet relevant assessment of nanoparticle toxicity.
Apparently, in vitro test systems lack the complexity of animal
models or the human body and may therefore not accurately
reflect nanoparticle toxicity in vivo. Moreover, since nanoparticles
may not settle on adherent cultured cells in defined numbers, in
vitro toxicity assays currently do not seem to be suitable for
establishing dose–response relationships. Recommendations on
the handling of the nanomaterial and definition of save exposure
levels will therefore only be reliably determined by in vivo studies.
However, it becomes increasingly clear that in vitro studies based
on standardized testing procedures do not only provide cell
type-specific mechanistic information but may also allow for a
first hazard identification to guide the risk assessment process.
7. Conclusions
The emission quantities prioritize metal oxide nanomaterials
for risk assessment (Section 2.1). The toxic potential of a
nanomaterial cannot be attributed to a single intrinsic property of
the pristine nanomaterial. Several physicochemical properties
together with adsorbed components from the surrounding
environment appear to govern the fate and effect of inhaled
nanomaterials. The protein corona is partially conserved, and
partially selective for specific naked inorganic or functionalized
surfaces (Section 3.1). Serum tends to decrease the state of
agglomeration, whereas lung lining fluid in general does not
(Section 3.2). Due to the complexity and polydispersity of a
physiological suspension of nanoparticles, a combination of
characterization methods with different physical measurement
principles (imaging, centrifugal, mass spectrometry, biochem-
ical) is mandatory (Section 3.3).
For a reliable hazard characterization of nanomaterials in vitro
we suggest to use a minimum set of sensitive cell lines and to
consider several test systems (Section 4.1). These in vitro assays
have to be evaluated (Sections 4.2–4.4) and it must be
demonstrated that the nanomaterials to be tested do not interfere
with the test system (Section 3.3.4).
Inhalation studies have established that some materials show
indeed a stronger toxic potential when dispersed on the nanoscale
(Section 5.1). But the question ‘‘Is nano dangerous?’’ cannot be
answered per se—a general nanospecific toxicity does not exist
(Section 5.2.3). Each material is different, and based on the level
of no adverse effects in inhalation, we find decreasing effects in
the following order: MWCNTs CeO
2
, ZnO >TiO
2
>SiO
2
,
ZrO
2
, CB. Clearly some materials (CB, SiO
2
, and ZrO
2
, in line
with data from the literature) do not show adverse effects at the
highest tested aerosol concentration of 10 mg m
3
, considerably
higher than the existing general thresholds for fine dusts (not
material-specific, Section 2.2). The measured aerosol concentra-
tion with no adverse effects is orders of magnitude beyond the
anticipated degradation scenarios of consumer-relevant nano-
composite products (Section 2.1) Comparing in vivo and in vitro
results (Section 6) we can identify indicators of concern for many
nanomaterials: ROS generation (except for SiO
2
and carbon
materials), fiber shape (for carbon materials), solubility (con:
releasing metal ions; pro: reducing persistence) and the crystal-
line phase (weaker effects for rutile than for anatase). Even with
the presently incomplete mechanistic understanding of inter-
actions of nanomaterials with human cells and body fluids
(Scheme 2), the correlations indicate that biophysical character-
ization and in vitro tests may allow for a testing strategy with
a minimal use of animals. This ensures an appropriate risk
assessment to define safe occupational handling and safe
consumer contact with nanomaterials (Scheme 1).
8. Materials and Methods
8.1. Test Materials for Inhalation
The materials tested were TiO
2
A and TiO
2
B, ZrO
2
,CeO
2
, ZnO,
CB, and a type of MWCNT, as well as two amorphous SiO
2
; the
surface of one SiO
2
was functionalized with polymeric carboxylate
while the other SiO
2
was not modified. TiO
2
A is precipitated in a
wet chemical process (sol–gel), and TiO
2
B is formed in flame
pyrolysis of titanate salts. The physicochemical properties of eight
test materials are presented in the Supporting Information
Table S1, comprising their specific surface area, impurities,
surface chemistry, solubility, crystallinity, state of agglomeration
in DMEM þ10% FBS (Fig. 3), and their monolayer TEM images
(Supporting Information Fig. S1).
The surface-functionalized SiO
2
was produced from the
above-listed SiO
2
by covalent surface modification with an
acrylate carboxypolymer. The solid fraction of the product
suspension was approximately 40%, particle size and BET
surface area of the functionalized material were not analyzed, but
are expected to be similar to the starting material. Some of the
nanoparticles were obtained by partners of the German research
project NanoCare
[2]
(see also www.nanopartikel.info) and char-
acterized as previously described.
[51]
Before entering a cell culture
assay, all nanoparticles were analyzed for Endotoxin contamina-
tion with the Limulus Amebocyte Lysate (LAL) Kinetic-QCL kit
(Lonza, 50-650U).
8.2. Inhalation Study Design
Groups of 14 animals were head–nose exposed to dust aerosols
for 6 h a day for five consecutive days. The respiratory tract was
evaluated by light microscopy in groups of six animals either
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 23
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
immediately after the last exposure or 3 weeks thereafter (study
days 5 and 26), as well as the content of the content of the test
material in the lung and in the mediastinal lymph nodes.
Bronchoalveolar lavage was performed in satellite animals
(five animals per group and time point) 3 days after the exposure
and 3 weeks thereafter (study days 8 and 29). Several biochemical,
cytological parameters as well as a large panel of cytokines/
chemokines were measured in the in BALF. The study design is
summarized in Scheme 4. Details of the parameters examined
are described in the online Supporting Information.
The inhalation studies with nanomaterials were approved by
the competent German authority (Landesuntersuchungsanstalt
Rheinland-Pfalz, http://www.lua.rlp.de), the permission was
issued under no AZ 177-07/053-1. The testing facility at BASF
was certified by the Association for Assessment and Accreditation
of Laboratory Animal Care (AAALAC) in 2007 (http://www.
aaalac.org/accreditedorgs/) and is in accordance with the German
Animal Welfare Act and the European Council Directive 86/609/
EEC.
8.3. For the Dispersion of Nanoparticles in Cell Culture
Medium
Nanoparticles (19.2 mg) were transferred into 10 mL sterile
snap-lid glasses together with a magnetic stir bar (Scheme 3). The
glasses were then exposed to 30 Gy in a Biobeam 8000 gamma
irradiation device (Gamma-Service Medical GmbH) for steriliza-
tion. Following the addition of 6mL of DMEM/10% FBS gold,
dispersions were stirred at 900 rpm for 1 h at room temperature.
Dilutions of this stock dispersion were prepared immediately and
stirred for 24 h at 900 rpm at room temperature. The mode of
dispersion chosen for the presented study was established as
Standard operation procedure in the frame of the German BMBF
funded project NanoCare and was aimed at modeling a real
exposure rather than producing a high fraction of single
dispersed nanoparticles.
8.4. Agglomeration Control by Analytical Ultracentrifugation
(AUC)
The particle size distribution was determined by AUC of 500 mL
of the above (Section 8.3) described dispersion with 0.1 mg mL
1
nanomaterial in DMEM/10% FBS gold. Simultaneous detection
by synchronized interference optics quantified the amount and
the diameter of each fraction independently.
[105,194]
The evalua-
tion of the AUC raw data incorporated the fractal morphology of
nanoparticle aggregates and applied the fractional dimension of
2.1 together with the sedimentation relation as specified in
Equation 6 of ref.
[102
. This value of the fractional dimension has
been shown to be universal for all reaction-limited colloid
aggregates.
[77,102]
The tabulated materials constant of refractive
index allows the interference optics to linearly quantify the
fraction that is dispersed to diameters below 100 nm in the actual
test preparation, as given in the Supporting Information Table S1,
with the full size distributions shown in Figure 3d. The value for
the nanodispersed fraction is regarded as an upper limit, judging
from the comparison of methods in Section 3.3 and Figure 3b.
Additional thorough documentation of the inhalation studies
is available as online Supporting Information and includes:
animals for inhalation; atmosphere generation and monitoring;
biological examinations of inhalation results; histopathology;
bronchoalveolar lavage; cytokines and chemokines in BALF and
lavaged lung tissue; hematology and acute phase proteins in
serum; statistical analysis of inhalation results.
Additional documentation of the cytotoxicity studies com-
prises: chemicals and cell culture components, cell lines,
determination of ROS, metabolic activity, and cell death.
Acknowledgements
This work was partially supported by the Federal German Ministry
of Education and Research BMBF (NanoCare; Fo
¨rderkennzeichen
03X0021C). We thank Sandra Brill for inspiring discussions about
literature. Supporting Information is available online from Wiley
InterScience or from the author.
Received: August 5, 2009
Revised: November 11, 2009
Published online:
[1] J. D. Brain, M. A. Curran, T. Donaghey, R. M. Molina, Nanotoxicology 2009,
3, 174.
[2] The NanoCare Consortium, NanoCare. Health Related Aspects of Nano-
materials, Final Scientific Report, (Eds: T. A. J. Kuhlbusch, H. F. Krug,
K. Nau), Dechema, Frankfurt, Germany 2009, ISBN: 978-3-89746-108-6.
[3] G. Oberdo
¨rster, J. Ferin, R. Gelein, S. C. Soderholm, J. Finkelstein, Environ.
Health Perspect. 1992,97, 193.
[4] G. Oberdo
¨rster, A. Maynard, K. Donaldson, V. Castranova, J. Fitzpatrick,
K. Ausman, J. Carter, B. Karn, W. Kreyling, D. Lai, S. Olin,
N. Monteiro-Riviere, D. Warheit, H. Yang, Part. Fibre Toxicol. 2005,2,8.
[5] A. D. Maynard, R. J. Aitken, T. Butz, V. Colvin, K. Donaldson,
G. Oberdo
¨rster, M. A. Philbert, J. Ryan, A. Seaton, V. Stone,
S. S. Tinkle, L. Tran, N. J. Walker, D. B. Warheit, Nature 2006,444, 267.
[6] D. B. Warheit, P. J. A. Borm, C. Hennes, J. Lademann, Inhal. Toxicol. 2007,
19, 631.
[7] A. D. Maynard, D. Rejeski, Nature 2009,460, 174.
[8] S. Brown, Nat. Nanotechnol. 2009,4, 608.
[9] Special Issue on Public Perceptions of Nanotechnology, Nat. Nano-
technol. 2009,4(2).
[10] N. Pidgeon, B. H. Harthorn, K. Bryant, T. Rogers-Hayden, Nat. Nano-
technol. 2009,4, 95.
[11] OECD, 2009,15. Environment Directorate, OECD. Series on the Safety of
Manufactured Nanomaterials.
[12] D. B. Warheit, R. A. Hoke, C. Finlay, E. M. Donner, K. L. Reed, C. M. Sayes,
Toxicol. Lett. 2007,171, 99.
[13] A. Kroll, M. H. Pillukat, D. Hahn, J. Schnekenburger, Eur. J. Pharm.
Biopharm. 2009,72, 370.
[14] R. Behra, H. Krug, Nat. Nanotechnol. 2008,3, 253.
[15] M. Vorbau, L. Hillemann, M. Stintz, J. Aerosol Sci. 2009,40, 209.
[16] B. Nowack, T. D. Bucheli, Environ. Pollut. 2007,150,5.
[17] T. A. J. Kuhlbusch, P. J. Crutzen, Global Biogeochem. Cycles 1995,9, 491.
[18] C. Buzea, I. I. Pacheco Blandino, K. Robbie, Biointerphases 2008,2,
MR17-MR172.
[19] I. Bru
¨ske-Hohlfield, A. Peters, in Nanotechnology—Environmental Aspects
(Ed.: H. Krug), Wiley-VCH, Weinheim 2008, p. 267.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
24 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
[20] T. A. J. Kuhlbusch, C. Nickel, Fo
¨rderkennzeichen (UFOPLAN) 3708 61 300 1,
2009.
[21] A. A. Koelmans, B. Nowack, M. R. Wiesner, Environ. Pollut. 2009,157,
1110.
[22] Lux Research, 2008,LRNI-SMR-08-01,1.
[23] M. Segal, Nat. Nanotechnol. 2009,4, 611.
[24] OECD, 2009,14. Environment Directorate, OECD. Series on the Safety of
Manufactured Nanomaterials.
[25] Woodrow Wilson Center, http://www.nanotechproject.org/inventories/
consumer/analysis_draft/, 2009.
[26] C. O. Robichaud, A. E. Uyar, M. R. Darby, L. G. Zucker, M. R. Wiesner,
Environ. Sci. Technol. 2009,43, 4227.
[27] R. Kaegi, A. Ulrich, B. Sinnet, R. Vonbank, A. Wichser, S. Zuleeg,
H. Simmler, S. Brunner, H. Vonmont, M. Burkhardt, M. Boller, Environ.
Pollut. 2008,156, 233.
[28] P. Simon, Q. Chaudhry, D. Bakos, J. Food Nutr. Res. 2008,47, 105.
[29] A. O. Gamer, E. Leibold, B. van Ravenzwaay, Toxicol. In Vitro 2006,20,
301.
[30] http://www.uni-leipzig.de/nanoderm/Downloads/Nanoderm_Final_
Report.pdf 2007.
[31] G. Oberdo
¨rster, E. Oberdo
¨rster, J. Oberdo
¨rster, Environ. Health Perspect.
2005,113, 823.
[32] Lux Research, 2009,LRNI-R-09-05,1.
[33] M. Auffan, J. Rose, J. Y. Bottero, G. V. Lowry, J. P. Jolivet, M. R. Wiesner,
Nat. Nanotechnol. 2009,4, 634.
[34] M. De, P. S. Ghosh, V. M. Rotello, Adv. Mater. 2008,20,1.
[35] P. Aggarwal, J. B. Hall, C. B. McLeland, M. A. Dobrovolskaia, S. E. McNeil,
Adv. Drug Deliv. Rev. 2009,61, 428.
[36] G. Oberdo
¨rster, V. Stone, K. Donaldson, Nanotoxicology 2007,1,2.
[37] I. Lynch, K. A. Dawson, S. Linse, Sci. Stke. 2006,2006, e14.
[38] T. Cedervall, I. Lynch, S. Lindman, T. Berggard, E. Thulin, H. Nilsson,
K. A. Dawson, S. Linse, Proc. Nat. Acad. Sci. 2007,104, 2050.
[39] S. Lindman, I. Lynch, E. Thulin, H. Nilsson, K. A. Dawson, S. Linse, Nano
Lett. 2007,7, 914.
[40] I. Lynch, T. Cedervall, M. Lundqvist, C. Cabaleiro-Lago, S. Linse,
K. A. Dawson, Adv. Colloid Interface Sci. 2007,134-135, 167.
[41] T. Cedervall, I. Lynch, M. Foy, T. Berggad, S. C. Donnelly, G. Cagney,
S. Linse, K. A. Dawson, Angew. Chem. Int. Ed. 2007,46, 5754.
[42] M. Lundqvist, J. Stigler, G. Elia, I. Lynch, T. Cedervall, K. A. Dawson, Proc.
Natl. Acad. Sci. USA 2008,105, 14265.
[43] C. Rocker, M. Potzl, F. Zhang, W. J. Parak, G. U. Nienhaus, Nat.
Nanotechnol. 2009,4, 577.
[44] M. D. Joshi, R. H. Mu
¨ller, Eur. J. Pharm. Biopharm. 2009,71, 161.
[45] S. Patil, A. Sanberg, E. Heckert, W. Self, S. Seal, Biomaterials 2007,28,
4600.
[46] S. C. Wasdo, D. S. Barber, N. D. Denslow, K. W. Powers, M. Palazuelos,
M. Stevens, B. M. Moudgil, S. M. Roberts, Int. J. Nanotechnol. 2008,5, 92.
[47] H. S. Choi, B. I. Ipe, P. Misra, J. H. Lee, M. G. Bawendi, J. V. Frangioni,
Nano Lett. 2009,9, 2354.
[48] H. S. Choi, W. Liu, P. Misra, E. Tanaka, J. P. Zimmer, B. I. Ipe,
M. G. Bawendi, J. V. Frangioni, Nat. Biotechnol. 2007,25, 1165.
[49] L. K. Limbach, Y. Li, R. N. Grass, T. J. Brunner, M. A. Hintermann,
M. Muller, D. Gunther, W. J. Stark, Environ. Sci. Technol. 2005,39, 9370.
[50] C. M. Sayes, K. L. Reed, D. B. Warheit, Toxicol. Sci. 2007,97, 163.
[51] C. Schulze, A. Kroll, C.-M. Lehr,U. F. Scha
¨fer, K. Becker, J. Schnekenburger,
C. Schulze-Isfort, R. Landsiedel, W. Wohlleben, Nanotoxicology 2008,2,
51.
[52] M. Horie, K. Nishio, K. Fujita, S. Endoh, A. Miyauchi, Y. Saito,
H. Iwahashi, K. Yamamoto, H. Murayama, H. Nakano, N. Nanashima,
E. Niki, Y. Yoshida, Chem. Res. Toxicol. 2009,22, 543.
[53] W. E. Wallace, M. J. Keane, D. K. Murray, W. P. Chisholm, A. D. Maynard,
T. M. Ong, J. Nanopart. Res. 2007,9, 23.
[54] E. Herzog, G. J. Oostingh, A. G. Lenz, A. Duschl, I. Beck-Speier,
M. Davoren, A. Casey, H. J. Byrne, Proc. Nanotox, Zu
¨rich 2008, p. 39.
[55] A. Gessner, B. R. Paulke, R. H. Muller, T. M.Goppert, Pharmazie 2006,61,
293.
[56] A. Verma, O. Uzun, Y. H. Hu, Y. Hu, H. S. Han, N. Watson, S. L. Chen,
D. J. Irvine, F. Stellacci, Nat. Mater. 2008,7, 588.
[57] D. Nepal, K. E. Geckeler, Small 2007,3, 1259.
[58] D. Li, M. B. Muller, S. Gilje, R. B. Kaner, G. G. Wallace, Nat. Nanotechnol.
2008,3, 101.
[59] L. Lacerda, M. A. Herrero, K. Venner, A. Bianco, M. Prato, K. Kostarelos,
Small 2008,4, 1130.
[60] D. H. Lin, B. S. Xing, Environ. Sci. Technol. 2008,42, 5917.
[61] M. Vippola, G. C. M. Falck, H. K. Lindberg, S. Suhonen, E. Vanhala,
H. Norppa, K. Savolainen, A. Tossavainen, T. Tuomi, Hum. Exp. Toxicol.
2009,28, 377.
[62] S. M. Hussain, L. Braydich-Stolle, A. M. Schrand, R. C. Murdock, K. O. Yu,
D. M. Mattie, J. J. Schlager, M. Terrones, Adv. Mater. 2009,21,1.
[63] S. Bastian, W. Busch, D. Ku
¨hnel, A. Springer, T. Meißner, R. Holke,
S. Scholz, M. Iwe, W. Pompe, M. Gelinsky, A. Potthoff, V. Richter,
C. Ikonomidou, K. Schirmer, Environ. Health Perspect. 2009,117, 530.
[64] C. M. Sayes, K. L. Reed, D. B. Warheit, Toxicol. Sci. 2007,97, 163.
[65] R. A. Yokel, R. Florence, M. Tseng, U. Graham, R. Sultana,
D. A. Butterfield, P. Wu, E. Grulke, Proc. Nanotox., Zu
¨rich 2008, p. 87.
[66] P. Bihari, M. Vippola, S. Schultes, M. Pra
¨tner, A. G. Khandoga,
C. A. Reichel, C. Coester, T. Tuomi, M. Rehberg, F. Krombach, Part. Fibre
Toxicol. 2008,5, 14.
[67] T. Meißner, A. Potthoff, V. Richter, J. Phys.: Conf. Ser. 2009,170, 012012.
[68] C. Fang, N. Bhattarai, C. Sun, M. Zhang, Small 2009,5, 1637.
[69] H. Hyung, J. D. Fortner, J. B. Hughes, J.-H. Kim, Environ. Sci. Technol.
2007,41, 179.
[70] B. Diaz, C. Sanchez-Espinel, M. Arruebo, J. Faro, E. de Miguel,
S. Magadan, C. Yague, R. Fernandez-Pacheco, M. R. Ibarra,
J. Santamaria, A. Gonzalez-Fernandez, Small 2008,4, 2025.
[71] J. M. Pettibone, D. M. Cwiertny, M. Scherer, V. H. Grassian, Langmuir
2008,24, 6659.
[72] M. J. Keane, W. E. Wallace, Inhal. Toxicol. 2005,17,287.
[73] T. M. Sager, D. W. Porter, V. A. Robinson, W. G. Lindsley,
D. E. Schwegler-Berry, V. Castranova, Nanotoxicology 2007,1,118.
[74] D. Porter, K. Sriram, M. Wolfarth, A. Jefferson, D. Schwegler-Berry,
M. E. Andrew, V. Castranova, Nanotoxicology 2008,2, 144.
[75] M. Maier, B. Hannebauer, H. Holldorff, P. Albers, J. Occup. Environ. Med.
2006,48, 1314.
[76] L. Ma-Hock, S. Burkhard, V. Strauss, A. O. Gamer, K. Wiench, B. van
Ravenzwaay, R. Landsiedel, Inhal. Toxicol. 2009,21, 102.
[77] D. F. Evans, H. Wennerstro
¨m, The Colloidal Domain: Where Physics,
Chemistry, Biology and Technology Meet, Wiley–VCH, Weinheim 1994.
[78] K. Wuthrich, Angew. Chem. Int. Ed. 2003,42, 3340.
[79] K. W. Powers, M. Palazuelos, B. M. Moudgil, S. M. Roberts, Nanotox-
icology 2007,1, 42.
[80] M. Hassello
¨v, J. W. Readman, J. F. Ranville, K. Tiede, Ecotoxicology 2008,
17, 344.
[81] D. B. Warheit, Toxicol. Sci. 2008,101, 183.
[82] E. Valsami-Jones, D. Berhanu, A. Dybowska, S. Misra, A. R. Boccaccini,
T. D. Tetley, S. N. Luoma, J. A. Plant, Mineral. Mag. 2008,72, 515.
[83] J. M. Antonini, N. J. Lawryk, G. G. K. Murthy, J. D. Brain, J. Toxicol. Environ.
Health Part A 1999,58, 343.
[84] D. R. Larson, W. R. Zipfel, R. M. Williams, S. W. Clark, M. P. Bruchez,
F. W. Wise, W. W. Webb, Science 2003,300, 1434.
[85] E. A. Meulenkamp, J. Phys. Chem. B 1998,102, 7764.
[86] S. Y. Kim, M. Nishioka, M. Taya, FEMS Microbiol. Lett. 2004,236, 109.
[87] A. Isakovic, Z. Markovic, B. Todorovic-Markovic, N. Nikolic,
S. Vranjes-Djuric, M. Mirkovic, M. Dramicanin, L. Harhaji, N. Raicevic,
Z. Nikolic, V. Trajkovic, Toxicol. Sci. 2006,91, 173.
[88] M. C. Yeber, J. Rodriguez, J. Freer, N. Duran, H. D. Mansilla, Chemosphere
2000,41, 1193.
[89] B. Fubini, A. Hubbard, Free Radic. Biol. Med. 2003,34, 1507.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 25
Final page numbers not assigned

REVIEW
www.advmat.de
www.MaterialsViews.com
[90] A. J. Hoffman, E. R. Carraway, M. R. Hoffmann, Environ. Sci. Technol. 1994,
28, 776.
[91] F. Q. Schafer, S. Y. Qian, G. R. Buettner, Cell. Mol. Biol. (Paris) 2000,46,
657.
[92] J. Dobson, FEBS Lett. 2001,496,1.
[93] L. Ma-Hock, A. O. Gamer, R. Landsiedel, E. Leibold, T. Frechen, B. Sens,
M. Linsenbu
¨hler, B. van Ravenzwaay, Inhal. Toxicol. 2007,19, 833.
[94] C. Asbach, H. Kaminski, H. Fissan, C. Monz, D. Dahmann, S. Mu
¨lhopt,
H. R. Paur, H. J. Kiesling, F. Herrmann, M. Voetz, T. A. J. Kuhlbusch,
Intercomparability of Continuous Particle Number Based Measurement
Techniques for Nanotechnology Workplaces 2007, p. 344.
[95] T. A. J. Kuhlbusch, S. Neumann, H. Fissan, J. Occup. Environ. Hyg. 2004,1,
660.
[96] A. D. Maynard, R. J. Aitken, Nanotoxicology 2007,1, 26.
[97] R. C. Murdock, L. Braydich-Stolle, A. M. Schrand, J. J. Schlager,
S. M. Hussain, Toxicol. Sci. 2008,101, 239.
[98] C. M. Keck, R. H. Muller, Int. J. Pharm. 2008,355, 150.
[99] Macromol. Symp, Vol. 162, Data Evaluation in Light Scattering of Polymers
(Eds: K. Gast, M. Helmstedt), Wiley, Weinheim, Germany 2000, 162.
[100] A. Bardow, Comput. Chem. Eng. 2008,32, 115.
[101] V. P. Maltsev, K. A. Semyanov, Inverse and Ill-Posed Problems Series, VSP,
Netherlands 2004.
[102] M. Y. Lin, H. M. Lindsay, D. A. Weitz, R. C. Ball, R. Klein, P. Meakin, Phys.
Rev. A 1990,41, 2005.
[103] U. Katzel, R. Bedrich, M. Stintz, R. Ketzmerick, T. Gottschalk-Gaudig,
H. Barthel, Part. Part. Syst. Charact. 2008,25,9.
[104] J. A. Jamison, K. M. Krueger, C. T. Yavuz, J. T. Mayo, D. LeCrone,
J. J. Redden, V. L. Colvin, ACS Nano 2008,2, 311.
[105] W. Ma
¨chtle, L. Bo
¨rger, Analytical Ultracentrifugation of Polymers and
Nanoparticles, Springer, Berlin 2006.
[106] H. Co
¨lfen, in Analytical Ultracentrifugation. Techniques and Methods (Eds:
D. J. Scott, S. E. Harding, A. J. Rowe), The Royal Society of Chemistry,
Cambridge 2005, p. 501.
[107] A. Balbo, K. H. Minor, C. A. Velikovsky, R. A. Mariuzza, C. B. Peterson,
P. Schuck, Proc. Natl. Acad. Sci. USA 2005,102, 81.
[108] J. Stoehr, N. Weinmann, H. Wille, T. Kaimann, L. Nagel-Steger,
E. Birkmann, G. Panza, S. B. Prusiner, M. Eigen, D. Riesner, Proc. Natl.
Acad. Sci. USA 2008,105, 2409.
[109] B. Demeler, in Bioinformatics Basics: Application in Biological Science and
Medicine (Eds: L. K. Bu
¨hler, H. H. Rashidi), CRC, Boca Raton, London
2005, p. 226.
[110] H. Strauss, E. Karabudak, S. Bhattacharyya, A. Kretzschmar,
W. Wohlleben, H. Co
¨lfen, Colloid Polym. Sci. 2008,286, 121.
[111] M. Baalousha, F. V. D. Kammer, M. Motelica-Heino, M. Baborowski,
C. Hofmeister, P. Le Coustumer, Environ. Sci. Technol. 2006,40, 2156.
[112] K. Thode, M. Luck, W. Semmler, R. H. Muller, M. Kresse, Pharm. Res.
1999,14, 905.
[113] D. W. Grainger, D. G. Castner, Adv. Mater. 2008,20, 867.
[114] I. Bravo-Osuna, G. Millotti, C. Vauthier, G. Ponchel, Int. J. Pharm. 2007,
338, 284.
[115] A. Casey, E. Herzog, F. M. Lyng, H. J. Byrne, G. Chambers, M. Davoren,
Toxicol. Lett. 2008,179, 78.
[116] N. A. Monteiro-Riviere, A. O. Inman, Carbon 2006,44, 1070.
[117] J.-M. Wo
¨rle-Knirsch, K. Pulskamp, H. F. Krug, Nano Lett. 2006,6, 1261.
[118] C. M. Sayes, A. A. Marchione, K. L. Reed, D. B. Warheit, Nano Lett. 2007,
7, 2399.
[119] L. Guo, A. von dem Bussche, M. Buechner, A. Yan, A. B. Kane, R. H. Hurt,
Small 2008,4, 721.
[120] J. M. Veranth, E. G. Kaser, M. M. Veranth, M. Koch, G. S. Yost, Part. Fibre
Toxicol. 2007,4,2.
[121] A. Casey, E. Herzog, M. Davoren, F. M. Lyng, H. J. Byrne, G. Chambers,
Carbon 2007,45, 1425.
[122] R. R. Davis, P. E. Lockwood, D. T. Hobbs, R. L. W. Messer, R. J. Price,
J. B. Lewis, J. C. Wataha, Biomaterials 2007,83B, 505.
[123] E. Dulkeith, M. Ringler, T. A. Klar, J. Feldmann, A. M. Javier, W. J. Parak,
Nano Lett. 2005,5, 585.
[124] http://characterizationmatters.org, 2008.
[125] N. A. Monteiro-Riviere, A. O. Inman, L. W. Zhang, Toxicol. Appl. Phar-
macol. 2009,234, 222.
[126] S. Lanone, F. Rogerieux, J. Geys, A. Dupont, E. Maillot-Marechal,
J. Boczkowski, G. Lacroix, P. Hoet, Part. Fiber Toxicol. 2009,6, 14.
[127] B. L’Azou, J. Jorly, D. On, E. Sellier, F. Moisan, J. Fleury-Feith, J. Cambar,
P. Brochard, C. Ohayon-Court, Part. Fibre Toxicol. 2008,5, 22.
[128] C. Kirchner, T. Liedl, S. Kudera, T. Pellegrino, A. Munoz Javier, H. E. Gaub,
S. Stolzle, N. Fertig, W. J. Parak, Nano Lett. 2005,5, 331.
[129] N. Lewinski, V. Colvin, R. Drezek, Small 2008,4, 26.
[130] I. Freshney, Cell Biol. Toxicol. 2001,17,213.
[131] J. G. Ayres, P. Borm, F. R. Cassee, V. Castranova, K. Donaldson, A. Ghio,
R. M. Harrison, R. Hider, F. Kelly, I. M. Kooter, F. Marano, R. L. Maynard,
I. Mudway, A. Nel, C. Sioutas, S. Smith, A. Baeza-Squiban, A. Cho,
S. Duggan, J. Froines, Inhal. Toxicol. 2008,20, 75.
[132] S. L. Hempel, G. R. Buettner,Y. Q. O’Malley, D. A. Wessels, D. M. Flaherty,
Free Radic. Biol. Med. 1999,27, 146.
[133] W. Jakubowski, G. Bartosz, Cell Biol. Int. 2000,24, 757.
[134] C. Vandeputte, I. Guizon, I. Genestie-Denis, B. Vannier, G. Lorenzon, Cell
Biol. Toxicol. 1994,10, 415.
[135] B. Shen, J. C. Scaiano, A. M. English, Photochem. Photobiol. 2006,82,5.
[136] C. M. Sayes, R. Wahi, P. A. Kurian, Y. Liu, J. L. West, K. D. Ausman,
D. B. Warheit, V. L. Colvin, Toxicol. Sci. 2006,92, 174.
[137] T. Xia, M. Kovochich, J. Brant, M. Hotze, J. Sempf, T. Oberley, C. Sioutas,
J. I. Yeh, M. R. Wiesner, A. E. Nel, Nano Lett. 2006,6, 1794.
[138] L. K. Limbach, P. Wick, P. Manser, R. N. Grass, A. Bruinink, W. J. Stark,
Environ. Sci. Technol. 2007,41, 4158.
[139] H. Yang, C. Liu, D. Yang, H. Zhang, Z. Xi, J. Appl. Toxicol. 2009,29,69.
[140] E. J. Park, K. Park, Toxicol. Lett. 2009,184,18.
[141] W. Lin, Y. W. Huang, X. D. Zhou, Y. Ma, Int. J. Toxicol. 2006,25, 451.
[142] E. J. Park, J. Choi, Y. K. Park, K. Park, Toxicology 2008,245, 90.
[143] K. Pulskamp, S. Diabate, H. F. Krug, Toxicol. Lett. 2007,168,58.
[144] A. Simon-Deckers, B. Gouget, M. Mayne-L’hermite, N. Herlin-Boime,
C. Reynaud, M. Carriere, Toxicology 2008,253, 137.
[145] H. A. Jeng, J. Swanson, J. Environ. Sci. Health A Tox. Hazard. Subst.
Environ. Eng. 2006,41, 2699.
[146] R. Duffin, L. Tran, D. Brown, V. Stone, K. Donaldson, Inhal. Toxicol. 2007,
19, 849.
[147] W. Lin, Y. W. Huang, X. D. Zhou, Y. Ma, Toxicol. Appl. Pharmacol. 2006,
217, 252.
[148] D. Lison, L. C. Thomassen, V. Rabolli, L. Gonzalez, D. Napierska,
J. W. Seo, M. Kirsch-Volders, P. Hoet, C. E. Kirschhock, J. A. Martens,
Toxicol. Sci. 2008,104, 155.
[149] M. Davoren, E. Herzog, A. Casey, B. Cottineau, G. Chambers, H. J. Byrne,
F. M. Lyng, Toxicol. In Vitro 2007,21, 438.
[150] R. Landsiedel, M. Schulz, M. D. Kapp, F. Oesch, Mutat. Res. 2009,681,
241.
[151] J. J. Wang, B. J. Sanderson, H. Wang, Mutat. Res. 2007,628, 99.
[152] A. Xu, Y. Chai, T. K. Hei, Part. Fibre Toxicol. 2009,6,3.
[153] C. A. Barnes, A. Elsaesser, J. Arkusz, A. Smok, J. Palus, A. Lesniak,
A. Salvati, J. P. Hanrahan, W. H. Jong, E. Dziubaltowska, M. Stepnik,
K. Rydzynski, G. McKerr, I. Lynch, K. A. Dawson, C. V. Howard, Nano Lett.
2008,8, 3069.
[154] N. R. Jacobsen, P. Moller, K. A. Jensen, U. Vogel, O. Ladefoged, S. Loft,
H. Wallin, Part. Fibre Toxicol. 2009,6,2.
[155] U. Wirnitzer, B. Herbold, M. Voetz, J. Ragot, Toxicol. Lett. 2008,186,
160.
[156] A. Di Sotto, M. Chiaretti, G. A. Carru, S. Bellucci, G. Mazzanti, Toxicol.
Lett. 2009,184, 192.
[157] S. Singh, T. Shi, R. Duffin, C. Albrecht, D. van Berlo, D. Hohr, B. Fubini,
G. Martra, I. Fenoglio, P. J. Borm, R. P. Schins, Toxicol. Appl. Pharmacol.
2007,222, 141.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
26 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2010,22, 1–27
Final page numbers not assigned

REVIEW
www.MaterialsViews.com www.advmat.de
[158] E. Herzog, H. J. Byrne, A. Casey, M. Davoren, A. G. Lenz, K. L. Maier,
A. Duschl, G. J. Oostingh, Toxicol. Appl. Pharmacol. 2009,234, 378.
[159] N. A. Monteiro-Riviere, R. J. Nemanich, A. O. Inman, Y. Y. Wang,
J. E. Riviere, Toxicol. Lett. 2005,155, 377.
[160] L. W. Zhang, L. Zeng, A. R. Barron, N. A. Monteiro-Riviere, Int. J. Toxicol.
2007,26, 103.
[161] J. M. Veranth, N. S. Cutler, E. G. Kaser, C. A. Reilly, G. S. Yost, Toxicol. In
Vitro 2008,22, 498.
[162] F. Tao, L. Kobzik, Am. J. Respir. Cell Mol. Biol. 2002,26, 499.
[163] J. G. Ayres, P. Borm, F. R. Cassee, V. Castranova, K. Donaldson, A. Ghio,
R. M. Harrison, R. Hider, F. Kelly, I. M. Kooter, F. Marano, R. L. Maynard,
I. Mudway, A. Nel, C. Sioutas, S. Smith, A. Baeza-Squiban, A. Cho,
S. Duggan, J. Froines, Inhal. Toxicol. 2008,20, 75.
[164] G. Oberdo
¨rster, Z. Sharp, V. Atudorei, A. Elder, R. Gelein, W. Kreyling,
C. Cox, Inhal. Toxicol. 2004,16, 437.
[165] K. Donaldson, P. J. Borm, G. Oberdorster, K. E. Pinkerton, V. Stone,
C. L. Tran, Inhal. Toxicol. 2008,20, 53.
[166] J. A. Araujo, A. E. Nel, Part. Fiber Toxicol. 2009,6, 24.
[167] T. R. Nurkiewicz, D. W. Porter, A. F. Hubbs, J. L. Cumpston, B. T. Chen,
D. G. Frazer, V. Castranova, Part. Fibre Toxicol. 2008,5,1.
[168] K. Donaldson, D. E. Newby, W. MacNee, R. Duffin, A. J. Lucking,
N. L. Mills, in Nanotoxicology: Characterization, Dosing and Health Effects
(Eds: N. A. Monteiro-Riviere, C. L. Tran), Informa Healthcare, New York
2007, p. 267.
[169] K. Donaldson, D. Brown, A. Clouter, R. Duffin, W. MacNee, L. Renwick,
L. Tran, V. Stone, J. Aerosol Med. 2002,15, 213.
[170] U. Heinrich, R. Fuhst, S. Rittinghausen, O. Creutzenberg, B. Bellmann,
M. Koch, K. Levsen, Inhal. Toxicol. 1995,7, 533.
[171] E. Bermudez, J. B. Mangum, B. Asgharian, B. A. Wong, E. E. Reverdy,
D. B. Janszen, P. M. Hext, D. B. Warheit, J. I. Everitt, Toxicol. Sci. 2002,70,
86.
[172] E. Bermudez, J. B. Mangum, B. A. Wong, B. Asgharian, P. M. Hext,
D. B. Warheit, J. I. Everitt, Toxicol. Sci. 2004,77, 347.
[173] V. H. Grassian, P. T. O’Shaughnessy, A. Damcakova-Dodd,
J. M. Pettibone, P. S. Thorne, Environ. Health Perspect. 2007,115, 397.
[174] T. M. Sager, V. Castranova, Part. Fibre Toxicol. 2009,6, 15.
[175] K. E. Driscoll, J. Carter, B. W. Howard, D. G. Hassenbein, W. Pepelko,
R. B. Baggs, G. Oberdo
¨rster, Toxicol. Appl. Pharmacol. 1996,136, 372.
[176] A. Elder, R. Gelein, J. N. Finkelstein, K. E. Driscoll, J. Harkema,
G. Oberdoerster, Toxicol. Sci. 2005,88, 614.
[177] C. W. Lam, J. T. James, R. McCluskey, S. Arepalli, R. L. Hunter, Crit. Rev.
Toxicol. 2006,36, 189.
[178] J. Muller, F. Huaux, A. Fonseca, J. B. Nagy, N. Moreau, M. Delos,
E. Raymundo-Pinero, F. Beguin, M. Kirsch-Volders, I. Fenoglio,
B. Fubini, D. Lison, Chem. Res. Toxicol. 2008,21, 1698.
[179] I. Fenoglio, G. Greco, M. Tornatis, J. Muller, E. Rayrnundo-Pinero,
F. Beguin, A. Fonseca, J. B. Nagy, D. Lison, B. Fubini, Chem. Res. Toxicol.
2008,21, 1690.
[180] C. A. Poland, R. Duffin, I. Kinloch, A. Maynard, W. A. H. Wallace,
A. Seaton, V. Stone, S. Brown, W. MacNee, K. Donaldson, Nat. Nano-
technol. 2008,3, 423.
[181] L. Ma-Hock, S. Treumann, H. Strauss, S. Brill, F. Luizi, M. Mertler,
K. Wiench, A. O. Gamer, B. van Ravenzwaay, R. Landsiedel, Toxicol.
Sci. 2009,112, 468.
[182] J. P. Ryman-Rasmussen, M. F. Cesta, A. R. Brody, J. K. Shipley-
Phillips, J. I. Everitt, E. W. Tewksbury, O. R. Moss, B. A. Wong, D. E.
Dodd, M. E. Andersen, J. C. Bonner, Nat. Nanotechnol. 2009,4,747.
[183] A. Erdely, T. Hulderman, R. Salmen, A. Liston, P. C. Zeidler-Erdely,
D. Schwegler-Berry, V. Castranova, S. Koyama, Y. A. Kim, M. Endo,
P. P. Simeonova, Nano Lett. 2009,9, 36.
[184] J. H. Ji, J. H. Jung, S. S. Kim, J. U. Yoon, J. D. Park, B. S. Choi, Y. H. Chung,
I. H. Kwon, J. Jeong, B. S. Han, J. H. Shin, J. H. Sung, K. S. Song, I. J. Yu,
Inhal. Toxicol. 2007,19, 857.
[185] J. H. Ji, J. H. Jung, I. J. Yu, S. S. Kim, Inhal. Toxicol. 2007,19, 745.
[186] J. H. Sung, J. H. Ji, J. U. Yoon, D. S. Kim, M. Y. Song, J. Jeong, B. S. Han,
J. H. Han, Y. H. Chung, J. Kim, T. S. Kim, H. K. Chang, E. J. Lee, J. H. Lee,
I. J. Yu, Inhal. Toxicol. 2008,20, 567.
[187] D. B. Warheit, W. J. Brock, K. P. Lee, T. R. Webb, K. L. Reed, Toxicol. Sci.
2005,88, 514.
[188] O. Creutzenberg, T. Hansen, H. Ernst, H. Muhle, G. Oberdorster,
R. Hamilton, Inhal. Toxicol. 2008,20, 995.
[189] J. H. E. Arts, H. Muijser, E. Duistermaat, K. Junker, C. F. Kuper, Food Chem.
Toxicol. 2007,45, 1856.
[190] R. Merget, T. Bauer, H. U. Kupper, S. Philippou, H. D. Bauer, R. Breitstadt,
T. Bruening, Arch. Toxicol. 2002,75, 625.
[191] J. M. Fine, T. Gordon, L. C. Chen, P. Kinney, G. Falcone, W. S. Beckett, J.
Occup. Environ. Med. 1997,39, 722.
[192] T. Xia, M. Kovochich, M. Liong, L. Madler, B. Gilbert, H. B. Shi, J. I. Yeh,
J. I. Zink, A. E. Nel, ACS Nano 2008,2, 2121.
[193] http://www.epa.gov/oppt/tsca8e/pubs/8ehq/2008/aug08/
8ehq_0808_17208a.pdf, 2008.
[194] H. Co
¨lfen, Analytical Ultracentrifugation of Nanoparticles, Encyclopedia of
Nanoscience and Nanotechnology, American Scientific Publishers, Valen-
cia, CA, USA 2004, p. 67.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
4444
45
Adv. Mater. 2010,22, 1–27 ß2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 27
Final page numbers not assigned