reviews R. Drezek et al.

Cytotoxicity
DOI: 10.1002/smll.200700595

Cytotoxicity of Nanoparticles
Nastassja Lewinski, Vicki Colvin, and Rebekah Drezek*

From the Contents

1. Introduction.............. 27

2. Cytotoxicity Assays... 28

3. Carbon Nanoparticles
.................................. 30

4. Metal Nanoparticles. 34

5. Semiconductor
Nanoparticles............ 39

6. Summary and Outlook

.................................. 45

Keywords:
· biocompatibility
· cytotoxicity
· nanomaterials
· nanoparticles
· toxicology

26 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles

Human exposure to nanoparticles is inevitable as nanoparticles become

more widely used and, as a result, nanotoxicology research is now gaining
attention. However, while the number of nanoparticle types and applications
continues to increase, studies to characterize their effects after exposure and
to address their potential toxicity are few in comparison. In the medical field
in particular, nanoparticles are being utilized in diagnostic and therapeutic
tools to better understand, detect, and treat human diseases. Exposure to
nanoparticles for medical purposes involves intentional contact or adminis-
tration; therefore, understanding the properties of nanoparticles and their
effect on the body is crucial before clinical use can occur. This Review
presents a summary of the in vitro cytotoxicity data currently available on
three classes of nanoparticles. With each of these nanoparticles, different data
has been published about their cytotoxicity due to varying experimental
conditions as well as differing nanoparticle physiochemical properties. For
nanoparticles to move into the clinical arena, it is important that
nanotoxicology research uncovers and understands how these multiple
factors influence the toxicity of nanoparticles so that their undesirable
properties can be avoided.

1. Introduction

The presence of nanoparticles in commercially available
products is becoming more common. Nanoparticles, accord-
ing to the ASTM standard definition, are particles with
lengths that range from 1 to 100 nanometers in two or three
dimensions.[1] It is projected that production of nanoparti-
cles will increase from the estimated 2 300 tons produced
today to 58 000 tons by 2020.[2,3] With this increase in manu-
facturing of nanoparticle-containing merchandise along with
the constant discovery of new applications of nanoparticles,
it is surprising that knowledge on the health effects of nano-
particle exposure is still limited. However, the number of ef-
forts aimed at determining the health risks associated with
nanoparticle exposure continues to grow. This is essential as
public perception of nanotechnology can be jeopardized by
events such as the “nano scare” in 2006 in Germany, involv-
ing the aerosol glass protective Magic-Nano,[4] or the
sunscreen controversy after the United States Environmen-
tal Protection Agency released findings that nanometer-
sized titanium dioxide particles found in sunscreens could
cause brain damage in mice.[5] These two incidents exempli-
fy why the need to verify the safety of nanoparticles is in-
creasingly more pertinent.
In contrast to nanoparticle exposure through use of con-
sumer products, emerging biomedical applications of nano-
particles as drug-delivery agents, biosensors, or imaging con-
trast agents involve deliberate, direct ingestion or injection
of nanoparticles into the body. For biomedical purposes, es-
pecially in vivo applications, toxicity is a critical factor to
consider when evaluating their potential. Nanoparticles for
imaging and drug delivery are often purposely coated with
bioconjugates such as DNA, proteins, and monoclonal anti-
bodies to target specific cells. As these nanoparticles are in-
tentionally engineered to interact with cells, it is important
to ensure that these enhancements are not causing any ad-
verse effects. More significant is whether either naked or
coated nanoparticles will undergo biodegradation in the cel-
lular environment and what cellular responses degraded
nanoparticles induce. For example, biodegraded nanoparti-
cles may accumulate within cells and lead to intracellular
changes such as disruption of organelle integrity or gene al-
terations.
While in vitro nanoparticle applications afford less strin-
gent toxicological characterization, in vivo use of nanoparti-
cles requires thorough understanding of the kinetics and
toxicology of the particles. In vitro cytotoxicity studies of
nanoparticles using different cell lines, incubation times, and
colorimetric assays are increasingly being published. How-
ever, these studies include a wide range of nanoparticle con-
centrations and exposure times, making it difficult to deter-
mine whether the cytotoxicity observed is physiologically
relevant. In addition, different groups choose to use various
[*] N. Lewinski, Prof. R. Drezek
Department of Bioengineering MS-142
Rice University
PO Box 1892, Houston, TX 77251-1892 (USA)
Fax: (+ 1) 713-348-5877
E-mail: drezek@rice.edu
Prof. V. Colvin
Department of Chemistry MS-60
Rice University
PO Box 1892, Houston, TX 77251-1892 (USA)

small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 27
 reviews
cell lines as well as culturing conditions, which makes direct
comparisons between the available studies difficult.
Despite these issues, general trends in the pool of exist-
ing data can be extracted. This Review examines the cyto-
toxicity of several classes of nanoparticles currently being
developed for biomedical applications. The nanoparticles in-
cluded are: 1) Carbon-based nanoparticles, such as ful-
Nastassja Lewinski graduated from Rice
University with a BS in Chemical Engi-
neering in 2006. She is currently a Ph.D.
candidate in the Department of Bioengi-
neering at Rice University under the
guidance of Prof. Rebekah Drezek. Her
research interests include assessing the
use of quantum dots as optical contrast
agents for enhancing in vivo disease
screening and detection, and addressing
public policy concerns of nanoparticle
safety.
Prof. Vicki Colvin was recruited by Rice
University in 1996 to expand its nano-
technology program. Today, she serves
as Professor of Chemistry and Chemical
Engineering at Rice University as well as
Director of the Center for Biological and
Environmental Nanotechnology (CBEN).
CBEN was one of the nation’s first Nano-
science and Engineering Centers funded
by the National Science Foundation.
Prof. Colvin has received numerous acco-
lades for her teaching abilities, including
Phi Beta Kappa’s Teaching Prize for
1998–1999 and the Camille Dreyfus Teacher Scholar Award in 2002.
In 2002, she was also named one of Discover Magazine’s “Top 20
Scientists to Watch” and received an Alfred P. Sloan Fellowship. Dr.
Colvin received her Bachelor’s degree in chemistry and physics from
Stanford University, and obtained her Ph.D. in chemistry from the
University of California, Berkeley. She is a frequent contributor to Ad-
vanced Materials, Physical Review Letters, and other peer-reviewed
journals, and holds patents of four inventions.
Prof. Rebekah Drezek is currently an As-
sociate Professor in the Departments of
Bioengineering and Electrical and Com-
puter Engineering at Rice University. She
has been on the faculty at Rice since
2002 where she conducts basic, ap-
plied, and translational research at the
intersection of medicine, engineering,
and nanotechnology towards the devel-
opment of minimally invasive photonics-
based imaging approaches for detection,
diagnosis, and monitoring of cancer. Her
research has been supported by grants
worth over $ 7M from the Whitaker Foundation, Welch Foundation,
Coulter Foundation, Beckman Foundation, NSF, NIH, DOD CDRMP,
and the Center for Biological and Environmental Nanotechnology.
Prof. Drezek is the recipient of the HSEMB Outstanding Young Scien-
tist Award, the MIT TR100’s Top 100 Young Innovators Award, the
American Association for Medical Instrumentation Career Achieve-
ment Award, and the DOD Era of Hope Scholar Award.
28 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
R. Drezek et al.
lerenes and single- and multi-walled carbon nanotubes;
2) metal-based nanoparticles, such as gold colloids, nano-
shells, nanorods, and superparamagnetic iron oxide nanopar-
ticles, and 3) semiconductor-based nanoparticles such as
quantum dots.
2. Cytotoxicity Assays
The first step towards understanding how an agent will
react in the body often involves cell-culture studies. Com-
pared to animal studies, cellular testing is less ethically am-
biguous, is easier to control and reproduce, and is less ex-
pensive. In the case of cytotoxicity, it is important to recog-
nize that cell cultures are sensitive to changes in their envi-
ronment such as fluctuations in temperature, pH, and nu-
trient and waste concentrations, in addition to the
concentration of the potentially toxic agent being tested.
Therefore, controlling the experimental conditions is crucial
to ensure that the measured cell death corresponds to the
toxicity of the added nanoparticles versus the unstable cul-
turing conditions. In addition, as nanoparticles can adsorb
dyes and be redox active, it is important that the cytotoxici-
ty assay is appropriate. Conducting multiple tests is advanta-
geous to ensure valid conclusions are drawn.
One simple cytotoxicity test involves visual inspection of
the cells with bright-field microscopy for changes in cellular
or nuclear morphology. Fiorito et al. used this technique
when evaluating the cytotoxicity of single-walled carbon
nanotubes (SWNTs).[6] However, the majority of cytotoxici-
ty assays used throughout published nanoparticle studies
measure cell death via colorimetric methods. These colori-
metric methods can be further categorized into tests that
measure plasma membrane integrity and mitochondrial ac-
tivity.
Exposure to certain cytotoxic agents can compromise
the cell membrane, which allows cellular contents to leak
out. Viability tests based on this include the neutral red and
Trypan blue assays. Neutral red, or toluylene red, is a weak
cationic dye that can cross the plasma membrane by diffu-
sion. This dye tends to accumulate in lysosomes within the
cell. If the cell membrane is altered, the uptake of neutral
red is decreased and can leak out, allowing for discernment
between live and dead cells. Cytotoxicity can be quantified
by taking spectrophotonic measurements of the neutral red
uptake under varying exposure conditions.[7] Two studies by
Flahaut et al. and Monterio-Riviere et al. exploring the cy-
totoxicity of carbon nanotubes utilized the neutral red
assay.[8,9] Trypan blue, a diazo dye, is only permeable to cells
with compromised membranes; therefore, dead cells are
stained blue while live cells remain colorless. The amount of
cell death can be determined via light microscopy.[10] This
assay was used by Bottini et al. and Goodman et al. to de-
termine the cytotoxicity of SWNTs and gold nanoparti-
cles.[11,12]
The LIVE/DEAD viability test, which includes two
chemicals calcein acetoxymethyl (calcein AM) and ethidium
homodimer, is another assay measuring the number of dam-
aged cells. This method has been used to test fullerenes and
small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles

gold nanoshells.[13,14,15] Calcein AM, an electrically neutral, an electron acceptor for enzymes such as NADP and
esterfied molecule, can easily enter cells by diffusion. Once FADH during oxygen consumption.[36]
within cells, it is converted to calcein, a green fluorescent As not all disruptive effects result in membrane or meta-
molecule, by intracellular esterases. In contrast, damaged or bolic function defects, more extensive cytotoxicity studies
dead cells are stained by ethidium homodimer, a mem- have attempted to determine the sub-lethal effects of nano-
brane-impermeable molecule, and fluoresce red when the particles. Oxidative stress can be detected using a gluta-
dye binds to nucleic acids. When excited at 495 nm, calcein thione assay. Glutathione (GSH) is a major antioxidant
AM and ethidium homodimer emit distinct fluorescence sig- compound that is oxidized to glutathione disulfide (GSSG)
natures at 515 nm and 635 nm, respectively.[16] in the presence of reactive oxygen species. In order to sus-
A third cytotoxicity assay used in several carbon-nano- tain its protective role against oxidative stress, a high GSH/
particle studies is lactate dehydrogenase (LDH) release GSSG ratio is required, which is maintained by the enzyme
monitoring.[13,17,18] In this assay, LDH released from dam- glutathione reductase. The glutathione assay detects levels
aged cells oxidizes lactate to pyruvate, which promotes con- of glutathione using EllmanFs reagent, 5,5’-dithio-bis-2-nitro-
version of tetrazolium salt INT to formazan, a water-soluble benzoic acid (DTNB), which reacts with the sulfhydryl
molecule with absorbance at 490 nm. The amount of LDH group of GSH to produce a yellow-colored product, 5-thio-
released is proportional to the number of cells damaged or 2-nitrobenzoic acid (TNB). Glutathione reductase also recy-
lysed.[19] cles GSH from the GSH-TNB complex producing more
In addition to distinguishing between live and dead cells TNB. Since the rate of TNB production is directly propor-
by detecting compromised plasma membranes, other colori- tional to the concentration of GSH in the sample, the ab-
metric cytotoxicity assays attempt to determine the mecha- sorbance of TNB can be measured at 405 or 412 nm to de-
nism behind the induced cell death. Mitochondrial activity termine the level of GSH.[37]
can be tested using tetrazolium salts as mitochondrial dehy- Lipid peroxidation of the plasma membrane can be de-
drogenase enzymes cleave the tetrazolium ring. Only active tected using GSH or a variety of other methods including
mitochondria contain these enzymes; therefore, the reaction the widely used thiobarbituric acid (TBA) assay. In the
only occurs in living cells.[20] The most widely used test is TBA assay, malondialdehyde (MDA), a toxic byproduct of
the MTT viability assay.[8,9,13,21–26] MTT, 3-(4,5-dimethylthia- lipid peroxidation, when heated at acidic pH reacts with 2-
zol-2-yl)-2,5-diphenyl tetrazolium bromide, is pale yellow in thiobarbituric acid to form a fluorescent pink chromagen,
solution but produces a dark-blue formazan product within which can be measured colorimetrically when excited at
live cells. A variation of this is the Cell Titer 96 Aqueous 532 nm. Other methods of lipid peroxidation are listed in an
One Solution Cell Proliferation Assay distributed by Prom- extensive review by Halliwell et al.[38]
ega, which has been em-
ployed by several groups.[27–29]
Here MTS, 3-(4,5-dimethyl-
thiazol-2-yl)-5-(3-carboxyme-
thoxyphenyl)-2-(4-sulfophen-
yl)-2H-tetrazolium, and phen-
azine ethosulfate, is used in-
stead. The number of living
cells can be determined simi-
larly by quantifying the pro-
duction of formazan by meas-
uring the absorbance at
492 nm.[30] Another tetrazoli-
um-based assay used to test
the cytotoxicity is the WST
assay.[31] WST-1 or WST-8
converts to a yellow–orange-
colored formazan product,
the concentration of which
can be quantified at
450 nm.[32] Resazurin or
Alamar blue has also been
used to ascertain cytotoxici-
ty.[33–35] This test is also a col-
orimetric assay where the
nonfluorescent alamar blue
dye is reduced to a pink fluo-
rescent dye by cell metabolic Figure 1. Structures and human dermal fibroblast Live/Dead cell viability assay results for C60 and deriva-
activity, mainly by acting as tives. Reprinted with permission from Ref. [13]. Copyright American Chemical Society, 2004.

small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 29
 reviews R. Drezek et al.

Inflammation is also a possible adverse effect of nano-
particle exposure. Commonly tested pro-inflammatory cyto-
kines or protein signals of inflammatory response include
IL-1b, IL-6, and TNF-a plus the chemokine IL-8.[39,40] These
cytokines are detected using enzyme-linked immunosorbant
assay (ELISA) and can be quantified by measuring the ab-
sorbance from either alkaline phosphatase or strepavidin-
horseradish peroxidase labeled antibodies at 405 or 620 nm,
respectively.[41]
More extensive cytotoxicity studies have attempted to
determine the genotoxic potential of nanoparticles by exam-
ining the extent of DNA damage using several methods.
One test that has been used extensively in studying the
effect of carbon-nanoparticle exposure is flow cytometry.[42–
45]
This technique utilizes a laser beam that differentiates
cells based on their size and density. Using DNA intercalat-
ing dyes, the cellular DNA content can also be used to de-
termine the proportion of cells undergoing apoptosis. One
such dye is propidium iodide, a membrane-impermeable red
dye, which undergoes a fluorescence change proportional to
the number of damaged cells. This occurs as binding of the
dye to nucleic acids increases with increased membrane per-
meability.[46] In addition to flow cytometry, the comet assay
has also been used to detect DNA damage in individual
cells using gel electrophoresis. Cells with damaged DNA
appear as “comets” with intact DNA residing in the head
portion and broken DNA pieces migrating away, forming
the tail. A DNA-specific dye such as propidium iodide is
used to read the gel, and the amount of DNA found in the
tail is proportional to the amount of DNA damage.[47] More
recently, to determine which specific genes are up- or down-
regulated due to nanoparticle exposure, some groups have
conducted preliminary DNA microarray studies.[48–51]
For nanoparticles, the major biological effects involve in-
teractions with cellular components such as the plasma
membrane, organelles, or macromolecules. As different
nanoparticles can trigger distinctive biological responses, it
is important that cytotoxicity studies are conducted for each
nanoparticle type. The following sections review the existing
cytotoxicity literature on carbon, metal, and semiconductor
based nanoparticles.
3. Carbon Nanoparticles
Carbon nanoparticles are materials composed mainly of
carbon with one or more dimensions at 100 nm or less.
These include, but are not limited to, carbon dots,[52] ful-
lerenes, nanodiamonds,[53] nanofoam,[54] nanohorns,[55] and
nano-onions.[46] However, as fullerenes are the most estab-
lished of the carbon nanoparticles, the focus is on this type.
Fullerenes, as defined by IUPAC, encompass C60, SWNTs,
and multi-walled (MW) NTs.[56] These three types are the
most widely used and well developed of the carbon nano-
particles. Their unique physiochemical properties (light
weight, high tensile strength, thermal/chemical stability and
conductivity) have generated several applications including
use in biomedical materials and devices such as tissue scaf-
folds, drug-delivery agents, and fluorescent-contrast
agents.[57–59] In terms of cytotoxicity, a major factor influenc-
ing potential toxicity is the carbon nanoparticlesF complexity
and variety in size, shape, charge, methods of production,

Table 1. Cytotoxicity studies on C60. d = diameter.

Cell Line Surface Exposure NP concentration Test Exposure Toxicity Author Year
coating conditions (average size) duration
[h]

Human dermal fibro- COOH, 70% confluency 0.24–2400 ppb MTT, Live/ 48 LD50 = 20 ppb for bare C60 on Sayes 2004
blasts, HDF; human OH, Na (d = 100 nm) Dead, HDF; no cytotoxicity observed [13]
liver carcinoma, HepG2 LDH with C60(OH)24
Guinea pig alveolar pristine 2 < 105 cells mL 1 8.36 < 104 MTT 3 No significant toxicity up to Jia 2005
macrophages in 24-well plates NP mg 1, 226 mg cm 2 [21]
1.41–
226 mg cm 2
Human dermal fibro- COOH, 70% confluency 0.24–2400 ppb MTT, Live/ 48 Nano-C60 is cytotoxic at Sayes 2005
blasts, HDF; human OH, Na (d = 100 nm) Dead, 20 ppb level; after 30 h cells [49]
liver carcinoma, LDH begin to have leaky mem-
HepG2; neuronal branes and lipid oxidation
human astrocytes, NHA
Monocyte-derived pristine 3 < 105 cells mL 1
30 and Nuclear 1, 24, Did not induce damage or Fiorito 2006
macrophages 60 mg mL 1 Morph, PI 48 death of macrophages [6]
Human monocyte pristine 2 < 106 cells well 1 0.16–10 mg mL 1 Neutral 48 No significant toxicity Porter 2006
macrophages for 24-well plate, (d = 60–270 nm) red [62]
0.5–1 < 106 for
96-well plate
Human epidermal N-Boc- 70% confluency, 0.00004– MTT 24, 48 Cytotoxicity at 0.04 and Rouse 2006
1
keratinocytes, HEK Baa 96-well plate 0.4 mg mL 0.4 mg mL 1; IL-8, IL-6, and [63]
IL-1b levels increased
Human umbilical vein C60(OH)24 90% confluency 1–100 mg mL 1 LDH, WST, 24 100 mg mL inhibit cell Yamawaki 2006
endothelial cells, on 6-well plates (d = 7.1  2.4 nm) microarray growth; 10 mg mL 1 inhibit [50]
HUVEC cell attachment

30 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles

chemical compositions, surface chemistry/functionalization, 3.1. C60

and aggregation tendency. A few reviews have been publish-
ed that look at the biocompatibility of carbon nanoparti-
cles.[60,61] This section looks at the studies that have been
conducted to elucidate the safety of these three major types
of carbon nanoparticle. Summaries of the experimental
setup and results on C60, SWNTs, and MWNTs are provided
in Tables 1–3, respectively.
Table 2. Cytotoxicity studies on SWNTs. l = length.
Several groups have studied the effect of C60 exposure
under various experimental conditions with different cell
lines and have yielded different results. The most significant
factor influencing cytotoxicity in this class of carbon nano-
particles seems to be cell type. The groups that reported
noncytotoxic effects studied C60 exposure in macrophage

Cell line Surface Exposure NP Concentration Test Exposure Toxicity Author Year
coating conditions (average size) duration
[h]

3T3 cells FITC N/A 1–10 mm (d = 1 nm, flow cytome- 1 5 mm, 90% viability, Pantarot- 2003
l = 300–1000 nm) try (annexin, 10 mm, 20% viability to [43]
PI)
Immortalized human none 80% confluency 0.06, 0.12, or Alamar blue, 2, 4, 6, viability decreased after Shvedova 2003
epidermal keratino- (w/30% Fe on 96 well 0.24 mg mL 1 GSH 8, 18 4h, 0.24 mg mL 1 ~ 65% [33]
cytes, HaCaT cat) plates or 75-cm2 viability
flasks
Mouse peritoneal pristine 107 cells 0–7.3 mg mL 1 microscopy 4, 8, 12, Cells ingest NT without Cherukuri 2004
macrophage-like (d = 1 nm, l = 1 mm) 18, 24 toxic effects [68]
cells, J774.1A
Human embryonic pristine 24 well plates MTT: 0.7812– MTT, western 24–120 Cytotoxicity dose- and Cui 2004
kidney, HEK293 200 mg mL 1; blot, flow time-dependent; 43.5% [44]
1
others: 25 mg mL cytometry, in G1 cell cycle arrest
(CAS 7782-42-5) microarray after 1 day
Human promyelcytic COOH, 3 < 106 0.05 mg mL 1 PI, flow 1 No significant toxicity Kam 2004
leukemia cells, HL60; biotin, fluo- cells mL 1 (d = 1–5 nm, cytometry for nonstretavidin-modi- [45]
Jurkat T cells rescein, l = 0.1–1 mm) fied SWNTs
streptavidin
Guinea pig alveolar pristine 2 < 105 1.41–226 mg cm 2 MTT 3 Cytotoxic effects seen at Jia 2005
macrophages cells mL 1 in (d = 1.4 nm, 0.38 mg cm 2 ; necrosis [21]
24-well plates l = 1 mm) at 3.06 mg cm 2
Human keratinocytes, pristine 5000 0.1, 0.5, 1, 5, 10, MTT, Live/ 72 Cytotoxic effects seen at Manna 2005
HaCaT; HeLa cells; cells well 1, 20 mg mL 1 Dead 0.5 mg mL 1, NFkB path- [65]
Lung carcinoma 96-well plate way activated by SWNT
(A549, H1299) cells
Monocyte-derived pristine 3 < 105 cells 30 and 60 mg mL 1
nuclear mor- 1, 24, 48 Did not induce damage Fiorito 2006
macrophages mL 1 phology, PI and death of macro- [6]
phages
1
Human dermal fibro- phenyl- 70% confluency 0.2–2000 mg mL MTT, Live/ 24, 48 Cytotoxicity decreased Sayes 2006
blasts, HDF (SO3H, (d = 1 nm, Dead w/ decreased C/phenyl- [66]
SO3Na, or l = 400 nm) SO3H ratio; LD50 could
(COOH)2), not be obtained
pluronic
F108
Human epidermal pristine 5x103 cells 0.8–100 mg mL 1
MTT 24–120 Strongest adverse effect Tian 2006
keratinocytes, HEK well 1 (d = 2 nm, w/ SWNT; 100 mg mL 1 [23]
l = 500 nm) gave 79%, 50% and
31% viability after
1, 3 & 5 days
Lung epithelial-like pristine 25 000 cells 50 mg mL 1 MTT, WST, 24-96 MTT gave different re- Worle- 2006
cells, A549 well 1, (d = 1.4 nm) LDH, MMP sults from WST, LDH Knirsch
96-well plate and MMP [69]
Rat alveolar macro- pristine 105 cells well 1 5–100 mg mL 1
MTT, WST 24–96 Cytotoxicity dose Pulskamp 2007
phage cells, NR8383; in 96-well (d = 1–2 nm, dependent; 100 mg [64]
human alveolar epithe- plates; l = 100 nm) mL 1, 60–80% reduc-
lial cells, A549 2.5 < 104 cells tion
well 1 in 96-well
plates (human)
1
Mesothelioma cells, pristine 3000 cells 7.5, 15, 30 mg mL MTT 72 Cytotoxicity dose depen- Wick 2007
MSTO-211H well 1 in 24- dent; agglomerated [67]
well plates worse than well-dis-
persed

small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 31
 reviews R. Drezek et al.

cell lines. Fiorito et al. found non-toxic responses for pris- toxicity after 24-hour incubation with concentrations of be-
tine C60 in their studies with murine macrophages.[6] They tween 1 and 100 mg mL 1.[50] Morphological changes, in-
reported that C60 had low cellular uptake, did not stimulate creases in LDH release, and growth inhibition were report-
nitric oxide release, and did not induce apoptosis in compar- ed. In addition, the fullerenes were found to aggregate and
ison to graphite and SWNTs. From this, C60 was considered internalize in autophagosomes, suggesting autophagic cell
to be fairly noncytotoxic. C60 was also deemed the least death.
toxic of the carbon nanoparticles by Jia et al.[21] The group
studied alveolar macrophage response to incubation with
3.2. Single-Walled Carbon Nanotubes
C60 and found that, after six hours of exposure to as much
as 226 mg cm 2 of C60, no significant toxicity resulted. Compared to the mixed reports on C60, SWNTs have
Porter et al. studied the effect of C60 in human monocyte typically been labeled as having cytotoxic effects at high
macrophages and also found no significant cytotoxicity.[62] concentrations. In addition, in a few comparison studies
However, looking at the subcellular level, C60 was found to SWNTs were reported as more toxic than the other two
aggregate into hexagonal units along the plasma membrane. major types of carbon nanoparticle.[21,23,64] In a study expos-
In addition, they were found to accumulate intracellularly in ing human embryo kidney cells to SWNTs for one to five
lysosomes, cytoplasm, along the nuclear membrane and days, Cui et al. found dose- and time-dependent decreases
inside the nucleus. In contrast, the groups that studied C60 in cell-adhesion ability, cell proliferation, and increases in
exposure in other cell lines found a dose-dependent cytotox- induction of apoptosis.[44] In addition, flow cytometry analy-
icity relationship. Incubating four different C60 derivatives, sis revealed altered cell-cycle regulation such as G1 phase
as illustrated in Figure 1, for up to 48 hours with human arrest due to SWNT exposure. Testing four different cell
dermal fibroblast and liver carcinoma cells, Sayes et al. lines (human keratinocytes, HeLa cells, and two lung carci-
found, for all four types, the lowest concentration (0.24 ppb) noma lines: A549, H1299), Manna et al. found oxidative
was relatively nontoxic while the highest concentration stress and inhibition of proliferation increased in a dose-
(2400 ppb) was more cytotoxic.[13] and time-dependent manner.[65] They also studied the NFkB
Between the four types, the addition of surface chemis- pathway, which they found was activated by SWNT expo-
tries for water solubility decreased the in vitro cytotoxicity, sure either via MAPK or IKK kinase activation.
with pristine C60 being more cytotoxic while the more hy- While pristine SWNTs were found to exhibit some cyto-
droxylated C60, C60(OH)24, had no apparent cytotoxicity toxic effects, a few groups found these effects were mitigat-
with an LD50 value of > 5 mg mL 1. It has been suggested ed by functionalizing the SWNT surface. Kam et al. provide
that this difference is due to the generation of reactive flow cytometry analysis that revealed no significant toxicity
oxygen species associated with C60. Additionally, particle ag- due to carboxyl-, biotin-, and fluorescein-coated SWNTs in
gregation was also determined to cause some of the cytotox- HL60 and Jurkat T cells after one hour.[45] Sayes et al.
ic effects. A more extensive study by Sayes et al. reported looked at the cytotoxicity of water-soluble, SWNT-treated
that cell apoptosis due to exposure to C60 was caused by human dermal fibroblasts over a range of concentrations
cell-membrane lipid peroxidation from oxygen radicals.[49] (3–30 mg mL 1) for up to 48 hours.[66] As illustrated in
This was confirmed when the addition of an antioxidant, L- Figure 2, cell death was highest in the cultures exposed to
ascorbic acid, prevented
membrane damage resulting
in cell viability comparable to
the control. Rouse et al. stud-
ied the effect of amino acid-
derivatized fullerenes in
human epidermal keratino-
cytes (HEK).[63] The levels of
pro-inflammatory cytokine
cytotoxicity indicators, IL-8,
IL-6, TNF-a, and IL-1b, were
measured. After 24 and 48
hours of incubation, a dose-
dependent decrease in cell vi-
ability was found as well as
higher phagocytosis of parti-
cles observed in cells exposed
to concentrations above
0.004 mg mL 1. Yamawaki
et al. tested the effect of hy-
droxyl fullerene, C60(OH)24,
on human umbilical vein en- Figure 2. Structures and human dermal fibroblast cytotoxicity data for SWNTs and derivatives. Reprinted
dothelial cells and found cyto- with permission from Ref. [66]. Copyright Elsevier, 2007.

32 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles
pristine SWNTs while, of the three functionalized SWNTs,
the SWNT containing the most functional groups yielded
the best cell viability.
Several hypotheses have been postulated to explain the
cytotoxicity observed with SWNTs. One is related to the
mode of production, as the synthesis of SWNTs requires the
use of metal catalysts, which can be toxic themselves. Shve-
dova et al. reported dose- and time-dependent cytotoxicity
in human epidermal keratinocytes exposed to SWNTs.[33] At
higher concentrations and longer incubation times, in-
creased oxidative stress, reduced glutathione levels, and nu-
clear and mitochondrial changes were found. They also
noted that the addition of a metal chelator reduced cytotox-
icity, suggesting that residual iron catalyst in solution may
play a role in the cytotoxicity observed. In addition, Jia
et al. found more than a 20% growth inhibition at the
lowest SWNT dose of 1.41 mg cm 2.[21] This cytotoxicity was
also speculated to be due to the 90% purity of the SWNT
solution as the presence of metallic catalysts could confound
the results.
Particle aggregation has also been suggested to be a
factor in nanoparticle cytotoxicity. Wick et al. aimed to de-
termine how agglomeration influenced SWNT cytotoxicity
and tested four different SWNT solutions: the raw material
involved in SWNT production, the SWNT-agglomerates re-
sulting from the synthesis, and the SWNT bundles and the
SWNT pellets (devoid of nanotubes) produced from centri-
fuging the SWNT agglomerate.[67] Aggregation occurred in
all SWNT fractions except the well-dispersed SWNT bun-
dles. Correspondingly, the SWNT bundles did not induce
adverse cellular effects, and as this was the only solution
where agglomerates were not formed, this corroborates the
hypothesis that SWNT agglomeration leads to cytotoxic ef-
fects. However, an earlier study by Tian et al., testing an un-
refined SWNT solution and a SWNT solution with the
metal catalysts removed, found lower cytotoxicity with the
unrefined SWNTs.[23] The group proposed that the lower cy-
totoxicity of the unrefined SWNTs was a result of their ag-
gregation into larger, and therefore less toxic, particles. This
contradicts the reasoning of Wick et al., who hypothesized
that the agglomerated SWNTs were cytotoxic due to the
stiffness and larger size, making the nanotubes emulate the
effects of asbestos fibers. While the conflicting results may
be due to the use of two different cell lines and asbestos-in-
duced lung-cancer cells versus keratinocytes, the effect of
SWNT aggregation is still questionable.
3.3. Multi-Walled Carbon Nanotubes
Studies on MWNTs have yielded results similar to those
of SWNTs. Monteiro-Riviere et al. reported that cells incu-
bated with higher concentrations of MWNTs for longer ex-
posure times contained more MWNTs.[9] The percentage of
cells with MWNTs inside increased from 59% at 24 hours to
84% after 48 hours. In addition, a dose- and time-dependent
decrease in cell viability was observed, coupled with an in-
crease in release of cytokine IL-8 at the higher MWNT con-
centrations. While the complimentary study by Shvedova
small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
et al. on SWNTs proposed that the cytotoxicity may be due
to trace amounts of catalyst in the solution, the lack of cata-
lyst particles in these MWNT solutions suggests the
MWNTs alone were potentially hazardous. Instead, Mon-
teiro-Riviere et al. hypothesize that the cytotoxicity is due
to MWNT attachment to the cell membrane or MWNT in-
ternalization, as MWNTs were seen in the cytoplasm and
near the nucleus. Sato et al. also found aggregates of
MWNT in cytoplasm in their studies.[70] Bottini et al. also
saw dose- and time-dependent cytotoxicity in T lymphocyte
and Jurkat leukemia cells.[11] In addition, comparing the ef-
fects of different surface coatings, hydrophobic MWNTs
were less toxic than ones coated with hydroxyl or carboxyl
groups. The same conclusions were made by Magrez et al.
after studying MWNT in lung carcinoma cells.[22] A dose-de-
pendent decrease in cell viability was also evident after ex-
posing alveolar macrohages to > 95% purified MWNTs,
conducted by Jia et al.[21] However, as they tested different
diameters of MWNT (10–20 nm) this dose dependence
could be due to particle mass, size, or both. Interestingly,
while these groups found a dose-dependent trend in cyto-
toxicity, Flahaut et al. found a decrease in viability in
human umbilical vein endothelial cells (HUVEC) with dilu-
tion of their MWNT solution.[8] Although they concluded
that the MWNTs were nontoxic as metabolic activity was
maintained above 75%, HUVEC viability seemed to de-
crease with exposure to decreasing concentrations of
MWNTs with large surface areas. The group suspects this is
a result of the aggregation of MWNTs or their enhanced in-
teractions with the cells due to their higher dispersion at
lower concentrations.
Few groups have studied the inflammatory response to
MWNTs. One group, Ding et al., looked at the genetic ef-
fects of MWNTs and found that high concentrations in-
duced immune and inflammatory gene overexpression.[48]
Witzmann et al. considered the protein-expression changes
after human epidermal keratinocyte exposure to MWNTs
and noted upregulation of proteins related to irritation and
cell apoptosis.[71] Muller et al. incubated peritoneal macro-
phages for up to 24 hours containing purified MWNTs and
ground MWNTs at concentrations of 20–100 mg mL 1.[17]
They determined that ground MWNTs had a capacity for
inducing dose-dependent cytotoxicity and up-regulating
TNF-a expression that is similar to that of asbestos and
carbon black. However, the unground MWNT sample ex-
hibited lower effects than the ground sample, which they at-
tributed to the increased agglomeration found in the un-
ground sample preventing cellular uptake. Murr et al. also
found that the cytotoxicity of MWNTs was similar to asbes-
tos.[72] Chlopek et al. investigated the viability and stimula-
tion of fibroblasts and osteoblasts exposed to purified
MWNTs.[27] The group deemed MWNTs to be biocompat-
ible with the tested cell types, as they found unchanged
levels of osteocalcin, cytokine IL-6, and oxygen-free radi-
cals.
www.small-journal.com 33
 reviews R. Drezek et al.

Table 3. Cytotoxicity of MWNTs.

Cell Line Surface Exposure NP concentration (aver- Test Exposure Toxicity Author Year
coating conditions age size) duration
[h]
1
Human epidermal pristine 80% confluency, 0.1, 0.2, 0.4 mg mL Neutral red 1, 4, 8, ~ 73% viability at Monteiro- 2005
keratinocytes, HEK 7000 cells well 1, 12, 24, 0.4 mg mL 1; IL-8 Riviere
96-well plates 48 increases with [9]
MWNT conc.
1
Human skin fibro- pristine 70% confluency, 0.06–0.6 mg L Hoechst 24, 48 Cytotoxicity dose- Ding 2005
blasts, HSF42; 96-well plates 33342, dependent for puri- [48]
human embryonic YO-PRO 1, fied MWNT
lung fibroblasts PI, BrdU,
(IMR-90) microarray
Guinea pig alveolar pristine 2 < 105 cells mL 1
in 1.41–226 mg cm 2
MTT 3 Necrosis seen at Jia 2005
macrophages 24-well plates (d = 10–20 nm, 3.06 mg cm 2 [21]
l = 0.5–40 mm)
Sprague–Dawley rat pristine direct lung 20–100 mg mL 1 LDH 3, 15
LDH doubled from Muller 2005
peritoneal macro- injection 0.5–2 mg rat 1 20 to 100 mg mL 1
days [17]
phages ground nanotubes
Murine alveolar pristine 5 < 105 cells well 1
0.005–10 mg mL 1, MTT, ELISA 48 Cytotoxicity begins Murr 2005
macrophages in 96-well plates (d = 5–30 nm, at 2.5 mg mL 1; [72]
(RAW267.9) l = 0.03–3 mm) similar to asbestos
Human acute mono- pristine 5 < 105 cells well 1
5–500 ng mL 1 0.1 mg HU TNF-a 16 TNF-a production Sato 2005
cytic leukemia cells in 96-well plates (d = 20–40 nm, Flexia histolo- dose dependent, [70]
(THP-1) Wister male l = 0.5–5 mm) gy, microscopy aggregates in sev-
rats eral cell types
T lymphocytes, Jurkat Hydroxyl, 4.4x104 cells mL 1
40–400 mg mL 1 Trypan blue 24–24 Cell death > 80% Bottini 2006
T leukemia cells carboxyl (d = 20–40 nm, in oxidized, < 50% [11]
l = 1–5 mm) in pristine at
400 mg mL 1
Human osteoblastic Poly- 2 cm3 cells/ N/A (d = 10–15 nm) Cell titer 96 24, 48, 7 Small viability de- Chlopek 2006
line hFOB 1.19; sulfone 12 well plate days crease in [27]
Human fibroblastic (PS) PS + MWNTs vs.
line HS-5 pure PS
Human umbilical pristine 6000 cells cm 2 Max: A: 0.5 mg mL 1 MTT, Neutral 24 None were cytotox- Flahaut 2006
vein endothelial in 96-well plates (d = 1.1–3.2 nm), red ic but error bars of [8]
cells, HUVEC B: 0.64 mg mL 1 samples A & B
(d = 1.1–4.3 nm) below threshold
C: 0.9 mg mL 1
(d = 0.7–6.3 nm)
Human lung-tumor Carbonyl N/A 0.002–0.2 mg mL 1 MTT 24–96 Cell viability de- Magrez 2006
cell lines, H596, (CdO), (d  20 nm, aspect creased 33% at [22]
H446, and Calu-1 carboxyl ratio = 80–90 nm) 0.2 mg mL 1; func-
(COOH), tionalized have
hydroxyl lower survival
(OH)
1
Human neonatal pristine 80% confluency, 0.4 mg mL protein array 24, 48 Irritation and cell Witzmann 2006
HEKs 6-well plates apoptosis proteins [71]
upregulated

4. Metal Nanoparticles later section. Gold nanoparticles exhibit an intense color in

4.1. Gold Nanoparticles
Gold colloids, now often referred to as gold nanoparti-
cles, have been used in medical applications during clinical
testing of heavy metals to treat rheumatoid arthritis as early
as the 1920s.[73] This precedence suggests that current appli-
cations of gold nanoparticles should not be limited by their
biocompatibility. While gold nanoparticles refer to particles
spherical in shape, other geometries, such as gold nanorods,
tripods, tetrapods,[74] and nanocages,[75] have been synthe-
sized, with gold nanorods discussed in further detail in a
the visible spectroscopic region and since gold can easily
bind functionalizing or targeting ligands, it has great prom-
ise as a contrast agent for bioimaging.
Due to their small size, gold nanoparticles have been
found to easily enter cells. Early studies with cytotoxicity
data were focused on utilizing this property for nuclear
transfection and targeting. In their work to find nonviral
gene-delivery devices, Thomas et al. found polyethylenimine
(PEI)-modified gold nanoparticles could transfect monkey
kidney (COS-7) cells six times better than PEI alone.[76] Cell
viability was recorded after exposure to PEI–gold nanopar-
ticle complexes, and 80% of the cells were still metabolical-

34 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles
ly active. While PEI–gold nanoparticles with dodecyl–PEI
complexes achieved even better transfection, cell viability
decreased to 70%. This complex was mainly found inside
the cell suggesting that internalization is a factor in cytotox-
icity. However, as the gold nanoparticles were conjugated to
PEI, whether or not the observed decrease in cell viability
was due to the gold nanoparticles is unclear. Another
group, Tkachenko et al., looked first at the nuclear targeting
ability of gold nanoparticles alone, and then at gold nano-
particles with a full-length peptide containing both the re-
ceptor-mediated endocytosis and nuclear localization signal
segments from an adenovirus in HepG2 cells.[77] The group
found that plain gold nanoparticles were readily taken up
into the cytoplasm; however, they did not enter the nucleus.
Experiments conducted at 48C indicated that cell entry was
energy dependent since a decrease in the number of parti-
cles inside the cells was observed. The gold nanoparticles
were determined to be able to enter the cell by receptor-
mediated endocytosis but unable to leave the endosomes,
hindering nuclear targeting. However, the nanoparticle-pep-
tide complex incorporating both transport signals was found
to enter the nucleus. Despite this nuclear exposure, cell via-
bility was greater than 95% after 12 hours of incubation.
Tkachenko et al. conducted another study examining
four different peptide–BSA–gold nanoparticle conjugates in
three cell lines (HeLa, 3T3/NIH, and HepG2).[78] Here they
reported differing effects of the nanoparticles between the
three cell lines. The four peptide–BSA–gold nanoparticles
were able to enter HeLa cells, escape the endosomes, and,
except for the particle with the HIV Tat protein, enter the
nucleus. In contrast, the four peptide–BSA–gold nanoparti-
cles were found clustered together in endosomes within the
3T3/NIH cells. The HepG2 cells did not seem to uptake the
peptide–BSA–gold nanoparticles except for the gold nano-
particle with the integrin-binding domain. The LDH cyto-
toxicity assay also confirmed these cell-line differences.
After three hours of incubation, the peptide–BSA–gold
nanoparticles conjugated with the adenovirus fiber protein
caused 20% cell death in HeLa cells while only 5% in the
3T3/NIH cells. This suggests that the nuclear delivery of the
peptide–BSA–gold nanoparticles influences cell viability
due to particle interactions with cellular DNA. Goodman
et al. also tested the effect of gold nanoparticle exposure in
multiple cell lines.[12] Cationic (ammonium-functionalized)
and anionic (carboxylate-functionalized) gold nanoparticles
with concentrations of 0.38–3 mm were incubated with COS-
1 cells, red blood cells, and Escherichia coli cultures for
24 hours. While the cationic nanoparticles were clearly more
cytotoxic than the anionic, a small variation was observed in
their LC50 values between cell types, showing that different
cell types experience similar toxicity. This contradicts the
findings of Tkachenko et al. as they found a difference be-
tween cell lines; however, this could be due to the use of
different surface coatings.[78] In addition, Goodman et al. [12]
proposed that the nanoparticles interact with the cells pas-
sively rather than by energy-dependent processes, as sug-
gested by Tkachenko et al. [78] since mammalian and bacteri-
al cells exhibited similar nanoparticle uptake. However, re-
duced-temperature incubation studies were not reported.
small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
As the results from Tkachenko et al. and Goodman
et al. show, the type of surface coating can play an impor-
tant role in the cytotoxicity of gold nanoparticles. Connor
et al. studied the effect of size and different surface modifi-
cations on uptake and acute toxicity in human leukemia
(K562) cells.[24] The sizes ranged between 4 and 18 nm with
surface modifiers including biotin, CTAB, cysteine, citrate,
and glucose. After three days of exposure, the largest nano-
particle with citrate and biotin surface modifiers did not
appear to be toxic at concentrations up to 250 mm. In con-
trast, a similar concentration of the gold-salt (AuCl4) solu-
tion was found to be over 90% toxic. Glucose and cysteine
were found to be less effective in rendering the nanoparti-
cles nontoxic. The gold nanoparticle concentration dropped
within the first hour of exposure, suggesting rapid uptake of
nanoparticles into cells. The consumed nanoparticles were
found clustered in endocytic vesicles and maintained their
size after being taken up by the cells.
Cytotoxicity may not be the only adverse effect of nano-
particles; nanoparticles may also affect the immunological
response of cells. Shukla et al. tested the effect of gold
nanoparticles on the proliferation, nitric oxide, and reactive
oxygen species production of RAW264.7 macrophage
cells.[25] After 48 hours of up to 100 mm gold-nanoparticle
treatment, RAW264.7 macrophage cells showed greater
than 90% viability with no increase in pro-inflammatory cy-
tokines TNF-a and IL-1b. Cell viability decreased to 85%
after 72 hours, which was attributed to depletion of media
nutrients since the media was not changed in those 72 hours.
The group found that cells take up gold nanoparticles inter-
nalizing them in lysosomes, which move in a time-depen-
dent manner toward the nucleus but do not enter the nu-
cleus. They also corroborate the findings of Goodman et al.
that gold nanoparticles were not present in cells kept at
cold (48C) temperatures.[25] Fu et al. and Shenoy et al. incu-
bated bare gold nanoparticles and gold nanoparticles func-
tionalized with methoxy-PEG-thiol or coumarin-PEG-thiol
with breast cancer (MDA-MB-231) cells for 24 hours.[28,29]
They found that the functionalized nanoparticles were inter-
nalized, by what they suggest is nonspecific endocytosis,
within the first hour and localize mainly in the cytoplasm
and perinuclear region. This is in agreement with the find-
ings of Shukla et al.[25]
Other noncytotoxic effects of nanoparticles, such as the
influence of nanoparticle exposure on the proliferation,
morphological structure, spreading, migration, and protein
synthesis of human dermal fibroblast cells, were examined
by Pernodet et al.[79] They found that with increasing con-
centration of gold nanoparticles, cell area decreased along
with cell number and density of actin fibers. Although no
cytotoxicity tests were conducted, the decreased number of
cells indicates some cytotoxic effects. The number of va-
cuoles present within the cells increased with time, and the
cells were filled with vacuoles by the sixth day. The gold
nanoparticles accumulated inside the cells, entering not by
endocytosis but rather through diffusion facilitated by their
small size (average size ~ 13 nm). The observed cellular
changes were both dose- and time-dependent. The group
also notes that the gold nanoparticles are not digested in
www.small-journal.com 35
 reviews R. Drezek et al.

Table 4. Cytotoxicity of gold nanoparticles.

Cell line Surface coating Exposure NP concentration Test Exposure Toxicity Author Year
conditions (average size) duration

COS-7 cells PEI2 3x105 cells/ N/A MTT 6h + 70-80% viability after Thomas 2003
well 42 h transfection [76]
Human liver carci- BSA, 4 targeting 85% N/A LDH 12 h Viability slightly compromised Tkachenko 2003
noma, HepG2 peptides confluency (d = 20-25 nm) (< 5%) [77]
COS-1, Red blood NH3, COOH 80% conflu- 0.38, 0.75, 1.5, or MTT, 1, 2.5, LD50 (Cos-1): anionic ~ 1 mM Goodman 2004
cells, E.coli ency, 96- 3 mm Trypan 6, 24 h and cationic > 7.37 mm; similar [12]
well plate blue for other cell types
HeLa, 3T3/NIH, BSA, 4 targeting 75% conflu- 150 pm LDH 3h Cell viability reduced by 20% Tkachenko 2004
HepG2 peptides ency (d = 22 nm) in HeLa cells, but only 5% in [78]
3T3/NIH
Leukemia cell line, citrate, biotin, L- 104 cells/ 0-250 mm Au atoms MTT 3 days No apparent toxicity at 250 mm, Connor 2005
K562 cysteine, glu- well (d = 4, 12, 18 nm) glucose and cysteine modified [24]
cose, CTAB not toxic up to 25 mm
Human breast coumarin-PEG- 105 cells/ 50-200 mg mL 1 Cell 24 h Nanoparticles are internalized Fu/Shenoy 2005
carcinoma xeno- thiol, mPEG- well in (d = 10 nm) Titer but essentially non-toxic up to [28/29]
graft cells, thiol (neg. con- 96-well 96 200 mg mL 1
MDA-MB-231 trol) plates
RAW264.7 lysine, PLL, FITC 105 cells/ 10, 25, 50, and MTT 24, 48, 100 mm - after 72 hr cell viability Shukla 2005
macrophage cells well in 100 mm (d = 3-8 nm) 72 h to decreased to 85% [25]
96-well
plates
Human dermal citrate N/A 0-0.8 mg mL 1 micros- 2-6 Dose-dependent decrease in cell Pernodet 2006
fibroblasts (d = 13 + /-1 nm) copy days area & density; many vacuoles [79]

the lysosomes, and even though the vacuoles accumulate
near the nucleus, no nuclear penetration was seen. A sum-
mary of the experimental setup and results on gold nanopar-
ticles is provided in Table 4.
4.2. Gold Nanoshells
Typically, gold nanoshells are composed of a silica die-
lectric core coated with an ultrathin metallic gold layer. This
core/shell structure allows for the gold nanoshells to be
made by either preferentially absorbing or scattering by
varying the relative core and shell thicknesses. Because of
the “tunability” of their optical properties, nanoshells are
being developed for imaging contrast and photothermal
therapeutic medical applications.[80,81]
A few studies have been published with cytotoxicity re-
sults on gold nanoshells. The first to suggest that gold nano-
shells are nontoxic was Hirsch et al. While the focus of the
study was on the photothermal ablative ability of the nano-
shells, they mention that exposure to nanoshells did not
cause cell death.[14] In later studies by Loo et al., SKBR2
breast-cancer cells exposed for one hour to 8 mg mL 1 or
3 K 109 nanoshells mL 1 of anti-HER2 bioconjugated nano-
shells exhibited no difference in viability compared to con-
trol cells.[15,82] A more recent study by James et al. studied
the biodistribution of gold nanoshells in female albino
mice.[83] A 100 mL nanoshell solution with a 2.4 K
1011 nanoshells mL 1 concentration was injected into the tail
veins of 30 mice. Five mice were sacrificed at several time
points up to 28 days, and the accumulation of nanoshells in
the blood and major organs, such as the liver, kidneys,
spleen, lungs, muscle, brain, and bone was measured. The
nanoshells were found to quickly clear the blood circulation
and predominantly accumulated in the liver and spleen. De-
spite the lack of complete nanoshell clearance from the
body after 28 days, the mice are reported to have shown no
physiological complications from the residual presence of
nanoshells.
Cytotoxicity of a gold/copper nanoshell has also been
studied by Su et al. Au3Cu nanoshell concentrations be-
tween 0.001 and 200 mg mL -1 were incubated with Vero
cells for 6 or 24 hours.[84] Using the WST assay, the group
found cell damage to be dose dependent with cell viability
decreased to 15% at the highest concentration after
24 hours of incubation with the nanoshells. The in vivo ef-
fects were also tested in male BALBc mice and, after
30 days, a dose dependence in viability rates was also found
with 100% viability in the low-dose mice but 67% viability
in the high-dose mice. Urine was collected from the mice
three hours after injection, and the amount of gold and
copper found suggested the nanoshells were being excreted
from the body. The loss of MRI signal after four hours cor-
roborated this finding. A summary of the experimental
setup and results on gold nanoshells is provided in Table 5.
4.3. Gold Nanorods
The advantage of gold nanorods is that they have both a
transverse and longitudinal plasmon. As the optical proper-
ties of materials depend both on the type and shape of the
metal, the unique properties of these rod-shaped particles
can be utilized in several potential applications. While very
few groups have published data on the cytotoxicity of gold
nanorods, the results are similar to those seen for gold nano-

36 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles

Table 5. Cytotoxicity of gold nanoshells.

Cell line Surface Exposure NP concentration Test Exposure Toxicity Author Year
coating conditions (average size) duration

Human breast PEG N/A 4.4x109 nanoshells Live/Dead 1 h No differences in viability Hirsch 2003
carcinoma mL 1 (core = 55 nm, [14]
SK-BR-3 cells shell = 10nm)
HER2-positive antibody- N/A 3x109 nanoshells Live/Dead 1 h No differences in viability Loo 2004-
SKBr3 breast PEG-thiol mL 1 (core = 120 nm, [15,82] 2005
adenocarci- shell = 10nm)
noma cells
Female albino PEG tail vein 2.4x1011 nanoshells observation 4 h - 28 Limit in mice muscle tissue about James 2007
mice injection mL 1 (core = 110 nm, days 70pg, accumulating in the RES [83]
shell = 10 nm) organs, 1–10 ppm levels found in
bone, muscle, kidney, lung
Vero cells, PEI/PAA 4x103 cells/ 0.001-200 mg mL 1 WST 6, 24 h Cell viability decreased 15% at Su 2007
BALBc mice (in mice) well in (core = 48.9  19.1 nm, 200 mg mL 1 [84]
96-well plates shell = 5.8  1.8 nm)

particles. An early study by Salem et al. tested feasibility of
gold/nickel nanorods as gene-delivery agents.[31] A concen-
tration of 44 mg mL 1 was used to test transfection of
human embryonic kidney (HEK293) cell line with Au/Ni
nanorods functionalized with GFP and luciferase reporter
genes. The group notes that this is significantly below the
LD50 value that was determined by the WST assay to be
750 mg mL 1. Transmission electron microscopy (TEM) re-
vealed that the nanorods localized in vesicles or in the cyto-
plasm but not the nucleus. While no biodistribution analysis
was done during their preliminary in vivo studies, no com-
plications due to skin and muscle exposure to nanorods
were reported. While examining the photothermal capabili-
ties of gold nanorods, Takahashi et al. found the viability of
cells incubated with nanorods, but without laser irradiation,
did not decrease significantly.[85]
More recently, other groups have found that the chemi-
cals involved in the synthesis of gold nanorods play a role in
their potential cytotoxicity. Niidome et al. looked at the
effect of poly(ethylene glycol) (PEG)-modified gold nano-
rods on HeLa cells after 24 hours of incubation.[86] Strong
cytotoxicity was associated with a low concentration of
CTAB-stabilized gold nanorods. They proposed that free
CTAB in solution was the source of the cytotoxic effect.
This was corroborated when removal of excess CTAB from
the PEG-modified gold-nanorod solution yielded 90% cell
viability at the highest concentration tested (0.5 mm). Taka-
hashi et al., the same group, published another study that
tested the cytotoxicity of gold nanorods extracted from
CTAB using a phosphatidylcholine (PC)-containing chloro-
form.[26] Concentrations from 0.09 to 0.72 mm exhibited little
cytotoxicity; however, at higher concentrations of 1.45 mm,
cell viability was reduced by approximately 20%. This cell
death was proposed to be due to nanorod aggregation. In
comparison, twice-centrifuged gold-nanorod solutions
showed significant cytotoxicity with the lowest tested con-
centration, 0.09 mm, reducing cell viability by about 15%
after 24 hours of incubation. Cytotoxicity was found to be
dose dependent with almost 0% cell viability at 1.45 mm.
Therefore, the group concluded that extraction process
using PC was a better method than twice centrifugation.
Huff et al. exposed KB cells to gold nanorods to examine
their internalization, whether by endocytosis or by CTAB
interaction on cell membranes.[87] This group found that KB
cells internalized the majority of CTAB-coated nanorods
while mPEG-DTC-coated nanorods were internalized at re-
duced levels. The CTAB-coated nanorods were found local-
ized near the perinuclear region within the KB cells and,
after five days, the cells appeared unaffected by the inter-
nalized nanorods as they grew to confluence over that
period. This study suggests that CTAB promotes nanorod
uptake by cells, which could explain the cytotoxicity ob-
served by Niidome et al. with CTAB stabilized nanorods. A
summary of the experimental setup and results on gold
nanorods is provided in Table 6.
4.4. Super-Paramagnetic Iron Oxide Nanoparticles
Superparamagnetic iron oxide nanoparticles (SPIONs)
are engineered g-Fe2O3 or Fe3O4 particles that exhibit mag-
netic interaction when placed within a magnetic field. In ad-
dition, when encountered by an alternating magnetic field,
the particles heat up, allowing for both imaging and therapy
applications. Specifically, their utilization as an MRI con-
trast agent has been extensively studied.[88–92]
In terms of cytotoxicity, while bare iron oxide nanoparti-
cles exert some toxic effects, coated SPIONs have been
found to be relatively nontoxic. Gupta et al. showed that
PEG-coated nanoparticles were biocompatible as exposed
cells remained more than 99% viable relative to control at
an upper concentration of 1 mg mL 1.[93] On the other hand,
bare iron oxide nanoparticles induced a 25–50% loss in fi-
broblast viability at 250 mg mL 1. In a more extensive study,
Gupta et al. found SPION cytotoxicity to be dose depen-
dent. SPIONs caused a 20% reduction in cell viability at the
lowest concentration tested (0.05 mg mL 1).[94] Further re-
ductions were seen at higher concentrations, with the high-
est concentration tested (2.0 mg mL 1) resulting in about
60% loss of cell viability. However, using a different PEG-
based coating, Yu et al. found PMAO-PEG-coated SPIONs,
illustrated in Figure 3, to be relatively nontoxic, with cell vi-

small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 37
 reviews R. Drezek et al.

Table 6. Cytotoxicity of gold nanorods.

Cell line Surface coating Exposure NP concentration Test Exposure Toxicity Author Year
conditions (average size) duration

Human embry- AEDP, plasmid, 3x105 cells/ 44 mg mL 1 WST 4h LD50 = 750 mg mL 1

Salem 2003
onic kidney, rhodamine, well in 24-well (w = 100 nm, [31]
HEK293 transferrin plates l = 200 nm)
Hela Cells PEG, CTAB 5x103 cells/ 0.01-0.5 mm WST 24 h Cell death: ~ 80% at 0.05 mm Niidome 2006
well in 96-well (w = 11  1 nm, w/CTAB nanorods; only ~ 10% [86]
plates l = 65  5 nm) at 0.5 mm w/PEG nanorods
Hela Cells phosphatidyl- 5x103 cells/ 0.09-1.45 mm MTT 24 h Twice centrifuged more toxic Takahashi 2006
choline well in 96-well (w = 11  1 nm, than PC-NRs; 20% cells died at [26]
plates l = 65  5 nm) 1.45 mm
KB Cells CTAB, mPEG-DTC N/A 100 mL microscopy 5 days Internalized to perinuclear Huff 2007
(l = 50 nm) region w/ CTAB, little uptake [87]
w/mPEG

icity at the concentrations

Figure 3. Example structure and cell viability data for water-soluble iron oxide nanoparticles. Reprinted
with permission from Ref. [95].
ability decreasing by only 9% at the 400 nm exposure
level.[95]
In addition, other groups testing bare iron oxide nano-
particles considered them to be biocompatible. Hussain
et al. had similar findings as the highest dose they tested,
250 mg mL 1, resulted in approximately 30% decrease in
cell viability but this was judged as exhibiting little to no
toxicity.[96] Higher concentrations were tested by Cheng
et al. and at 23.05 mm of nanoparticles the group found no
significant difference between the exposed cells and the
control. However, this may be due to the relatively short ex-
posure time of four hours.[97]
In addition to PEG, several groups have studied the cy-
totoxicity of different surface-coated iron oxide particles
and found little cytotoxicity. Gupta et al. looked at pullulan
(Pn)-coated SPIONs and found no cytotoxic effects, with
1 [94]
the cells remaining more than 92% viable at 2.0 mg mL .
The group attributed the low toxicity of Pn-SPIONs to the
pullulan coating, which prevents the iron oxide core from
interacting with cells. Petri-Fink et al. observed no cytotox-
icity in melanoma after two hours of exposure to amino-
SPION for all polymer/iron ratios tested.[98] After 24 hours,
cytotoxicity became apparent for high polymer concentra-
tions. A similarly coated SPION tested by Cengelli et al.
was found to be nontoxic as N11 microglial cells only took
up aminoPVA-coated SPIONs, and as no nitric oxide was
produced.[99] Wan et al. tested the effects of three surface
coatings on iron oxide cytotoxicity and found MPEG–Asp3-
NH2-coated iron oxide nanoparticles had almost no cytotox-
tested.[100] In comparison,
MPEG–PAA- and PAA-
coated iron oxide nanoparti-
cles significantly reduced cell
viability with only 16% of the
cell remaining at an iron con-
centration of 400 mg mL 1. As
bare iron oxide nanoparticles
adsorbed to the cell surface,
MTS analysis was infeasible;
however, cell counts after in-
cubation indicated that un-
coated iron oxide nanoparti-
cles also significantly reduced cell viability.
The mechanism for SPION cytotoxicity, when it does
occur, has been linked to both cellular uptake and ROS pro-
duction. Hu et al. found PACHTUNGRE(PEGMA)-immobilized nanopar-
ticles were relatively nontoxic, as exposed cells had greater
than 93% viability.[101] However, pristine iron oxide nano-
particles had a viability reduced to 70% in the first two
days, increasing to about 90% by day five. The group sug-
gests that this increase in viability is due to the decrease in
nanoparticle concentration with the increase in cells after
mitosis. This was seen as cell uptake of particles went from
154 pg cell 1 on the first day to 58 pg cell 1 after five days.
PACHTUNGRE(PEGMA) nanoparticles were taken up at 2 pg cell 1 sug-
gesting that their lower toxicity is due to their lack of cell
uptake. Brunner et al. found a cell-specific response to bare
iron oxide nanoparticle exposure.[102] 3T3 cells remained
proliferative with the addition of up to 30 ppm iron oxide;
however, human mesothelioma cells exhibited significant re-
duction in cell viability at only 3.75 ppm iron oxide. The
group attributed the observed toxicity to iron-induced free-
radical production via the Fenton or Haber–Weiss reactions
in addition to internalization of the iron oxide particles. Pi-
sanic et al. showed anionic dimercaptosuccinic acid
(DMSA)-coated iron oxide nanoparticles are readily endo-
cytosed by rat pheochromocytoma cells and are found
either in the cytoplasm, inside endosomes, or accumulated
in the perinuclear region within the cells.[103] Most of the cell
death occurred during the first 48 hours of exposure with cy-
totoxicity and cell detachment being dose dependent.

38 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles

Changes in cell morphology were observed with nanoparti-
cle exposure with the cells assuming a spherical shape with
disruption of the cell cytoskeleton. Muller et al. consistently
found a 20–30% decrease in neutral red uptake in mono-
cyte-macrophages with 10 mg mL 1 Ferumoxtran-10 across
various incubation times.[104] Similar results were found
using the MTT assay. Testing over a longer period showed
cell viability remained about the same between Ferumox-
tran-10-pretreated cells and control cells after two weeks.
The group speculates that the cytotoxicity is due to ROS
production via the Fenton reaction, which can result in lipid
peroxidation, DNA damage, and protein oxidation. Testing
for inflammatory responses, the group found Ferumoxtran-
10 did not induce increases in cytokines IL-1b, TNF-a, IL-6
or IL-12, or superoxide anion production. This led to the
conclusion that the monocyte-macrophages were not acti-
vated by the nanoparticles. A summary of the experimental
setup and results on superparamagnetic iron oxide nanopar-
ticles is provided in Table 7.
5. Semiconductor Nanoparticles
In the case of semiconductor nanocrystals, better known
as quantum dots, the concern of their cytotoxicity is not un-
justified as several are composed of known toxic elements.
However, despite the potential health risks, promising appli-
cations of quantum dots include their use in the medical

Table 7. Cytotoxicity of Fe3O4 nanoparticles.

Cell line Surface Exposure NP concentration Test Exposure Toxicity Author Year
coating conditions (average size) duration

COS-7 cells none 3x104 cells/well, 0.92-23.05 mm MTT 4h No significant difference be- Cheng 2004
24-well plates (d = 9 nm) tween control and exposed [97]
1
Human fibroblasts MA-PEG 10,000 cells mL 0-1000 mg mL 1 MTT, 24 h 250 mg mL 1: 25-50% viability Gupta 2004
in 24-well plates (d = 50nm) Live/ decrease for bare; 1 mg mL 1: [93]
Dead 99% viable for PEG-coated
Melanoma cells PVA N/A 0.25 mg mL 1 MTT 2h No cytotoxicity after 2 h for all Petri- 2004
(amino, (core = ~ 9 nm, polymer/iron ratios; after Fink
caroxyl, d = 19-54 nm) 24 hr, cytotoxicity at high poly- [98]
thiol) mer conc.
Primary human fibro- pullulan 104 cells/well in 0-2 mg mL 1 MTT 24 h Plain SPION showed signifi- Gupta 2005
blasts, hTERT-BJ1 96-well plates (d = 40-50 nm) cant decrease in viability, Pn- [94]
SPION showed no cytotoxicity
w/ > 92% viability
Rat liver cells, BRL none confluent in 6 or 0-250 mg mL 1 LDH, 24 h EC50 > 250 mg mL 1 Hussain 2005
3A 24 well plates (d = 30, 47 nm) MTT, [96]
GSH
Human mesothelio- none N/A 3.75-15 ppm MTT 3, 6 days 3T3 cells viable w/ up to Brunner 2006
ma MSTO-211H, (d = 12-50 nm) 30 ppm; MSTO cell viability [102]
rodent 3T3 fibroblast decrease at 3.75 ppm, free
cells radicals via Fenton rxn
Rat brain-derived en- PVA, ami- 96 or 48-well 2.5 mL NPs mL 1, MTT 48 h Only aminoPVA-SPION upaken Cengelli 2006
dothelial EC219; noPVA, car- plates 11.3 mg iron mL 1 by N11 [99]
murine N9 & N11 oxylPVA, (core = 8-12 nm,
microglial cells ThiolPVA d = 30 nm)
Mouse macrophages, PACHTUNGRE(PEGMA) 105 cells mL 1
0.2 mg mL 1 ratio of 1, 4 days Cytotoxicity dose dependent; Hu 2006
RAW 264.7 (d = 6.2  0.7 nm) treated/ decrease w/time attributed to [101]
control cell division
cells
Human breast carci- PMAO-PEG Confluent 10-400 nm Live/ 1, 24, 91% viability at 400 nm after Yu 2006
noma SK-BR-3 cells, (9.6 nm) Dead 48 h 48 h in HDF cells [95]
human dermal
fibroblasts
Human monocyte- dextran 1–2x106 cells/well 0.0001-10 mg MTT, 24, 48, 1 mg mL 1: not toxic after Muller 2007
macrophages in 24-well plates & mL 1 (d = 30 nm) Neutral 72 h, 4 72 h; 10 mg mL 1: mildly [104]
0.5-1x106 cells/ Red days + toxic; viability similar over
well in 48-well 14 day 2 wks
plates grow
rat pheochromo- DMSA 20,000 cells mL 1 15, 1.5 mm and Live/ 2, 4, 6 SPION exposure reduced PC12 Pisanic 2007
cytoma cell line in 6 or 12 well 150 mm (d = 5- Dead days ability to respond to nerve [103]
PC12M plates 12 nm) growth factors
OCTY mouse cells MPEG– 104 cells/well in 0-400 mg mL 1 Cell 72 h MPEG–Asp3-NH2 almost no Wan 2007
Asp3-NH2, 96-well plates (d = 14 nm) Titer 96 cytotoxicity; MPEG–PAA- and [100]
MPEG-PAA, PAA-coated decrease cell via-
PAA bility

small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 39
 reviews R. Drezek et al.

field as new drug-delivery and biomedical-imaging agents.
This section explores the current research that has been
conducted on quantum-dot toxicity.
Quantum dots (QDs) are nanoscale particles ranging
from 2 to 100 nm in diameter depending on the types of sur-
face coating or functional group added. For biological appli-
cations, QDs typically have a core/shell conjugate structure.
The core of the QD is composed of atoms from groups II–
VI (e.g., CdSe, CdTe, CdS, PbSe, ZnS, and ZnSe) and
groups III–V (e.g., GaAs, GaN, InP, and InAs) on the peri-
odic table.[105] Many of these core metals are known to be
toxic at low concentrations; examples include cadmium, se-
lenium, lead, and arsenic. Therefore, if these QDs are ex-
posed to conditions promoting degradation, such as an oxi-
dative environment, toxicity related to the release of free
metal ions is expected. Thus the crucial factor in QD toxici-
ty is stability. The cytotoxicity of QDs is reduced when their
cores are protected from degradation given that the added
coatings are biocompatible. To prevent core degradation, an
additional shell layer is added, making the QD more bio-
compatible. Additional functionalities or bioconjugates can
be added to the surface to improve bioavailability or intro-
duce bioactivity. Since CdSe/ZnS quantum dots are believed
to be the most versatile for biological applications, most of
the published toxicity studies focus on this type.[106,107] Sum-
maries of the experimental setup and results on CdSe and
CdTe QDs are provided in Tables 8 and 9, respectively.
5.1. Cadmium Selenide Quantum Dots
Historically, QDs were used in animals well before ex-
tensive cytotoxicity studies. These results highlighted key
issues, such as biodistribution and coating integrity, now of

Table 8. Cytotoxicity studies on CdSe quantum dots.

Cell line Surface coating Exposure NP concen- Test Exposure Toxicity Author Year
conditions tration duration
(average
size)

BALB/c nu/nu 3 peptides (GFE, F3, tail vein 100-200 mg histology 5 or Specific tissue targeting achievable, Akerman 2002
mice LyP-1), PEG injection QDs in 20 min accumulate in liver and spleen [109]
0.1-0.2 mL
solution
Xenopus PEG-PE, N/A 2.3 mm microscopy At least At > 5x109 QDs/cell abnormalities Dubertret 2002
embryos phosphatydilcholine 4 days became apparent [108]
Vero cells MUA, SSA N/A 0.24 microscopy 2 h 0.4 mg mL 1: no difference in via- Hanaki 2003
mg mL 1 bility of MUA-QD/SSA complexes [123]
Hela cells, Dic- DHLA N/A 400-600 nm microscopy 45- No differences in viability Jaiswal 2003
tyostelium dis- 60 min [122]
coideum
1
Mice Micelle tail vein 20 nM mmobserva- N/A No abnormal behavior observed Larson 2003
injection tion [110]
BALB/c mice poly(acrylic acid) tail vein 50-500 pmol histology/ 1-3 h @ QDs in endosomes in liver, spleen Ballou 2004
polymer or mPEG- injection QDs in 50- microscopy 1 min and bone; fluorescence at 1 mo. [111]
750 QDs, mPEG- 200 mL saline interval similar to 24 h signal, mPEG-5000
5000 QDs, COOH- QDs have longer circ. time & lower
PEG-3400 QDs accumulation
Hela cells silane, biotin, STV, 300 cells/ 1, 10, colonigenic 2 h On average > 90% cells survived Chen 2004
peptides, NLS 100 mm 100 nm assay [126]
(d = 8-10 nm)
Rat primary MAA, BSA/EDAC, 5x105 cells 0.0625, MTT, 24 h Coating eliminates air oxidation; Derfus 2004
hepatocytes EGF on 35 mm 0.25, 1 mg microscopy however, high conc. with 8 h UV [113]
wells mL 1 exposure still have 95% cell death
4
Human MUA, cysteamine 5x10 0-2 mm Comet, 12 h Crude QDs exhibited decreased cell Hoshino 2004
lymphoblastoid, (NH2), cells/well (d = 9-48 nm) flow cytom- activity, QD fluorescence lost in low [118]
WTK1 thioglycerol (OH) on 96-well etry, MTT pH oxidation, TOPO is cytotoxic
plates
Vero cells, Hela MUA, SSA 3x104 0-4 mg mL 1 MTT, flow 24 h Damaged cells increased sharply at Shiohara 2004
cells, primary cells/well cytometry 0.2 mg mL 1 but slowly at [116]
human hepato- on 96-well 0.1 mg mL 1
cytes plates
B16F10 melano- TOPO, DHLA 5x104 10 mL microscopy 4-6 hr No detectable toxicity Voura 2004
ma cells cells/well [129]
on 6-well
plates
DNA biotin N/A N/A Plasma 0-60 min DNA damage occurs in both Green 2005
nicking in dark environ, UV exposure incr damage [127]
assay or UV

40 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles

Table 8. (Continued)

Cell line Surface coating Exposure NP concen- Test Exposure Toxicity Author Year
conditions tration duration
(average
size)

NRK fibroblasts; MPA, mercapto- 7x104 2-10 mm ratio of 18-48 h NRK cells: poisoning occurs around Kirchner 2005
1
MDA-MB-435S carbonic acid cells mL (core d = 2.4- pre/post 0.65 mm for MPA-CdSe, 5.9 mm for [114]
breast cancer 4.5 nm) adherent CdSe/ZnS; other 3 cells: only larger
cells; Chinese cells, Live/ particles inside cell, polymer better
hampster ovary, Dead than MPA
CHO; RBL cells
COS-7 cells, SiO2, MAA, PA 2x104 cells 0-0.6 mm Alamar 48 h SiO2 Selvan 2005
NIH 3T3 cells, mL 1 in 96- (core blue (cos-7 & coating better than PA, ZnS coating [35]
Human liver well plates d = 5 nm, hepG2), decreases toxicity
carcinoma, d = 25 nm) 72 h
HepG2 (3T3)
Human breast PEG (750-, 22,500 10-150 nM Live/Dead 4h 150 nm - 0%, 70%, 80% cell viabili- Chang 2006
carcinoma 6000-Mw) cells/well ty for bare, 750- & 6000-Mw PEG [124]
SK-BR-3 cells in 96-well coated QDs
plates
Human bone HIV-derived Tat 7x105 cells 1.625 mg WST, flow 24 h Growth curve and cell cycle Hsieh 2006
marrow mesen- peptide in 10 cm cytometry, distribution not affected by QDs [121]
chymal stem dishes Rt-PCR, mi-
cells (hBMSC) croscopy
Sprague– LM, BSA jugular 5 nmol in histology 90 min QDs uptake in Kupffer cells, majori- Fischer 2006
Dawley rats vein injec- 0.2 mL solu- ty uptake in liver with some in [112]
tion tion (d = 25, spleen, lungs, kidneys, lymph
80 nm) nodes, bone marrow
Human breast silica, thiol, PEG N/A 2-10 nm microarray 48 h No adverse effects in lung cells, Zhang 2006
cancer cells, Brca, 8- skin cells showed < 50 gene ex- [128]
Lung (IMR-90), 80 nm lung, pression changes
skin (HSR-42) skin (d = 8-
10 nm)
Human breast NAC, Cys, MPA 105 cells 10 mg mL 1 MTT 24 h MPA capped QDs more toxic than Cho 2007
cancer cells, cm 2 in Cys capped, cells exhibit reactions [130]
MCF-7 24-well of oxidative stress
plates
Hela cells PEG-g-PEI, poly- 60% con- 1 nm (core MTT 2h PEI-coated dots very toxic, toxicity Duan 2007
carboxylate fluency d = 6.5 nm, reduced w/ more PEG added [119]
PEI coated
d = 15.3 nm)
HepG2 cells, PLA, F-68, SDS, 6x104 cells 10-400 ppm MTT 12-72 h > 80% cell viability up to 400 ppm Guo 2007
Wister mice CTAB on 96-well (d = 159- [131]
plates 266 nm)
ctDNA MAA 5 mg mL 1 3.6x10 7 Nucleic 110 min 1.7x10 5 mol L 1 Cd2 + released, Liang 2007
ctDNA mol L 1 acid probe 70% DNA damaged by QDs [132]
(RuACHTUNGRE(bipy)2-
ACHTUNGRE(dppx)2 + )
primary HEKs PEG, PEG-amine, 1.5–2x104 0.2-20 nm MTT 24, 48 h Dose–response significant for PEG- Ryman- 2007
polyacrylic acid cells cm 2, (d = 4.6 nm amine and carboxylic acid but not Rasmussen
40-60% & w = 6 nm, PEG alone [120]
confluency h = 12 nm)

concern in cytotoxicity studies. Dubertret et al. injected mi-
celle-encapsulated CdSe/ZnS QDs into Xenopus embryos.
At 2 K 109 QDs cell 1, the quantum dot injected and control
embryos displayed similar growth patterns with the QDs re-
maining in the injected cells and their progeny. However, at
5 K 109 QDs cell 1, adverse effects were found. The group
speculated that the abnormalities resulted from the QDs af-
fecting the osmotic equilibrium of the cell.[108] The majority
of published in vivo studies were conducted in mouse or rat
models. Akerman et al. injected three different (GFE, F3,
LyP-1) peptide-coated CdSe/ZnS QDs into normal BALBc
mice and studied the tissue distribution after 5 or 20 mi-
nutes of circulation. Each of the peptide-coated QDs were
found to accumulate in the liver and spleen in addition to
the targeted tissue; yet, this nonspecific accumulation was
reduced by adding PEG to the QD surface. The group did
not observe any acute toxicity caused by the QDs after 24
hours of circulation.[109] Similarly, Larson et al. observed no
adverse effects after imaging the mice used in their experi-
ment and hypothesized that CdSe/ZnS QDs clear from the

small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 41
 reviews R. Drezek et al.

Table 9. Cytotoxicity studies on CdTe quantum dots.

Cell line Surface coat- Exposure NP concentra- Test Exposure Toxicity Author Year
ing conditions tion (average duration
size)

Rat pheochro- anionic, cat- 105 cells 0.01-100 mg MTT 24 h 50% decrease in metabolic activity: 50 mg Lovric 2005
mocytoma cells, ionic, BSA cm 2 on mL 1 (d = 2.2- mL 1 (green) 100 mg mL 1 (red) [133]
PC12 24-well 5.7 nm w/o
plates coating)
Human hepato- none 2x105 cells 0, 10 8, 10 7, MTT 24 h ~ 50% viability reduction at 10 5 m; simi- Liu 2006
ma HepG2 cells mL 1 in 10 6, 10 5 m lar cytotoxicity found w/ air and N2 ex- [135]
96-well posed QDs
plates
Human breast NAC, Cys, 105 cells 10 mg mL 1
MTT 24 h MPA capped QDs more toxic than Cys Cho 2007
cancer cells, MPA cm 2 in capped; cys-CdTe QDs more cytotoxic [130]
MCF-7 24-well than cys-CdSe/ZnS QDs; cells exhibit oxi-
plates dative stress
Human neuro- cysteamine, 105 cells 5 mg mL 1
MTT, flow 24 h 52% viability w/ cyc-QDs, 85% viability if Choi 2007
blastoma cells, N-acetylcys- cm 2 in cytometry pretreated w/NAC [136]
SH-SY5Y teine (NAC) 24-well
plates
Human hepato- none 2x105 cells 0-100 mm; 2 MTT 48 h; 24 h IC50: 19.1 mm (red), 4.8 mm (yellow), 3 mm Zhang 2007
ma cells, mL 1 in mm,1 mL kg 1 (0, 0.5, 1, (green); rats: few signs of toxicity, but [134]
HepG2; Spra- 96-well in rats (d = 2- 2, 4 h) - in changes in locomotor activity were ob-
gue-Dawley rats plates 6nm) rats served

body before the protective coating can breakdown. Howev-
er, the group did not report a time line for which the ani-
mals were kept, only mentioning that the animals were
maintained for a long-term toxicity study, which diminishes
their findings.[110]
More recently, the role of the particle surface on distri-
bution and toxicity has been studied. Ballou et al. also ob-
served QD deposition in the liver, spleen, and bone marrow
of BALBc mice depending on the surface coating present.
Of the four coatings tested (poly(acrylic acid), mPEG-750,
mPEG-5000, COOH-PEG-3400), the mPEG-5000 QDs
were found to have the longest circulation time in addition
to reduced nonspecific accumulation. All QDs were found
localized internally in endosomes, primarily within perifol-
licular cells for the spleen, and no visible signs of break-
down were seen via electron microscopy. A long-term stabil-
ity study conducted with mPEG-750 QDs found the QDs
remained in the liver, lymph nodes, and bone marrow for a
month. Although the fluorescence decreased after one
month, the fluorescence distribution was close to what the
group observed 24 hours after injection.[111] More recently,
Fischer et al. injected mercaptoundecanoic acid (MUA),
lysine, and BSA-coated CdSe/ZnS QDs in Sprague–Dawley
rats. A difference in biodistribution was also found as the
liver took up 40% of the lysine-QDs and 99% of BSA-QDs
after 90 minutes. QDs endocytosed by Kupffer cells, similar
to RES processing, were sequestered not excreted. Small
amounts of both QDs appeared in the spleen, kidney, and
bone marrow, but no QDs were detected in the feces or
urine even after ten days. The group also measured the size
of the quantum dots within the vesicles and found the QDs
retained their size, suggesting no degradation after 90 mi-
nutes of exposure.[112] These studies conclude that ZnS
capped CdSe QDs are relatively nontoxic as the animals
were not killed, nor did they exhibit abnormal behavior
after QD injection. However, as several of the group found,
the QDs are internalized and seem to be retained inside the
cells. As clearance from the body is an important aspect of
safety, this suggests possible toxicity could result from the
bioaccumulation.
All of the previously mentioned studies involved ZnS
coated CdSe quantum dots; the ZnS coating provides a
well-terminated surface with few defects and high quantum
yields. A seminal in vitro study conducted by Derfus et al.
found that, when incubated with rat primary hepatocytes,
bare CdSe QDs undergo surface oxidation, resulting in the
release of free cadmium ions. Cadmium is a known toxic
agent that induces cell death via mitochondrial damage and
oxidative stress. When QD surface oxidation was prevented
with surface coatings, the cadmium atoms remained bound
to selenium atoms and the surface-coating molecules ren-
dering them relatively nontoxic. This was demonstrated
with the addition of a ZnS shell; the oxidative degradation
of the CdSe core due to exposure to air was significantly re-
duced resulting in lower cytotoxicity.[113] Several groups
have confirmed the effectiveness of the ZnS shell in reduc-
ing the cytotoxicity of CdSe quantum dots.[34,35,114,115] Chan
et al. proposed a mechanism for the bare CdSe QD-induced
cell death. In addition to determining that apoptosis, not ne-
crosis, occurred in CdSe exposed cells, the group suggested
that QDs induced apoptosis by activating Jun N-terminal
kinase (JNK) in a dose-dependent manner. In addition, mi-
tochondrial-dependent apoptotic processes, involving activa-
tion of caspase 9 and 3, increases in Bax protein and de-
creases in Bcl-2, were also observed. Mitochondrial-mem-
brane potential was reduced with exposure to bare CdSe
QDs resulting in an increase in cytochrome c release.[115]
Researchers have also examined the effect of additional
surface coatings on the cytotoxicity of quantum dots. QDs
must be appropriately encapsulated to prevent cadmium re-

42 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles
lease and subsequent cytotoxicity. At issue for many re-
searchers is the best way to accomplish this. One simple ap-
proach is to replace the original organic coat with water-
soluble ligands. These ligand-exchange reactions yield QDs
with smaller hydrodynamic sizes but generally do not pro-
vide biocompatible materials. Shiohara et al. studied the cy-
totoxicity of three MUA-coated CdSe/ZnS QDs (520-, 570-
and 640-nm emission) in three different cell lines (Vero
cells, Hela cells, and human primary hepatocytes). After in-
cubating the cells with quantum-dot concentrations ranging
from 0 to 0.4 mg mL 1 over 24 hours, a concentration de-
pendence of cytotoxicity was found.[116] It is important to
recognize that, in this case, the toxicity is due to the surface
coating rather than the quantum dots. MUA is a compound
that in previous studies was found to water solubilize CdSe/
ZnS quantum dots.[117] This is important in biological appli-
cations as hydrophobic compounds have poor bioavailabil-
ity. This study reveals that the MUA coating is not appropri-
ate for this purpose as it increases the toxicity of the quan-
tum dots. Hoshino et al. did a larger study incorporating
more surface coatings (MUA, cystamine, thioglycerol) on
CdSe/ZnS QDs. MUA-coated quantum dots were more
toxic than ones without. Of the three coatings, thioglycerol
was found to induce the least genotoxicity and therefore cy-
totoxicity. These results indicate that some hydrophilic sur-
face coatings contribute to the cytotoxicity of QDs. Because
TOPO was also found to be a cytotoxic compound, the com-
plete removal of TOPO from the QD samples is important
in reducing toxicity.[118] Selvan et al. looked at SiO2, mercap-
toacetic acid (MAA), polyanhydride (PA) surface-coated
CdSe/ZnS quantum dots in three different cell lines (human
liver carcinoma (HepG2), NIH T3T cells, COS-7 cells) and
obtained dose-dependent results for all surface coatings in
each cell line. SiO2/CdSe QDs were found to be much less
cytotoxic than MAA or PA coated QDs. SiO2/ZnS-CdSe
QDs were less cytotoxic than SiO2/CdSe QDs, suggesting
that the combination of ZnS capping and SiO2 coating pro-
vided for the optimal protection against CdSe dissolution.[35]
A better option for biocompatible QDs is to use amphi-
philic polymers to encapsulate the inorganic/organic system.
Most commercial sources of QDs prepare their systems in
this way. While larger polymeric coatings increase the hy-
drodynamic size, they yield very bright and stable materials.
Kirchner et al. found PEG to lower cellular uptake of silica-
coated QDs resulting in lower cytotoxicity.[114] Duan et al.
tested PEI-coated QDs in HeLa cells and found they are
endocytosed or macropinocytosed after one to two hours of
incubation. However, since PEI-coated dots are toxic to
cells, PEG was added to reduce this toxicity. The two differ-
ent forms (PEI grafted with two PEG and PEI grafted with
four PEG) of coated QDs exhibited different distribution
patterns inside the cells and different cytotoxicities. The
PEI-g-PEG4 QDs accumulated in the perinuclear region,
which yielded better cell viability while the PEI-g-PEG2
QDs were distributed in the cytoplasm and had significant
cytotoxic effects. The group believes the cytotoxicity is due
to the PEI polymer and not the presence of cadmium
ions.[119] Ryman-Rasmussen et al. tested two different QDs
(565- and 655-nm emission) with three different surface
small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
coatings (PEG, PEG-amine and carboxylic acids). All QDs
were localized intracellularly by 24 hours with the PEG-
coated QDs found in the cytoplasm, perinuclear region, and
for QD 565 within the nucleus. After 24 hours, no cytotoxic-
ity was observed, but by 48 hours toxicity became apparent
at the largest concentration of 20 nm, indicating time-depen-
dent cytotoxicity. Surface coating had an observable effect
on IL-1b, IL-6, and IL-8 pro-inflammatory cytokine release.
Cytokine levels increased after carboxylic acid coated QD
exposure, while there was no increase in cytokine release
with PEG-coated QDs.[120]
In addition to their coatings, size and concentration can
influence the toxicity of quantum dots with smaller sizes
and higher concentrations being more cytotoxic. The ad-
vantage of quantum dots in imaging applications is their
tunability. By changing their size or core diameter, the fluo-
rescence emission peak can be shifted to a wavelength of
choice within a fairly broad range. This is particularly useful
in biological applications as cells contain endogenous fluo-
rophores, which can mask the signal emitted from contrast
agents with similar emission peaks. However, several groups
have found cytotoxicity to be size dependent with smaller
QDs exhibiting larger reductions in cell viability. Kirchner
et al. tested the exposure of CdSe/ZnS QDs in several cell
lines (NRK fibroblasts, MDA-MB-435S breast cancer cells,
CHO cells, RBL cells). After 18 hours exposed to the same
concentration, cytotoxic effects were higher for smaller
QDs, which they suggest could be due to the higher surface-
to-volume ratio of smaller particles.[114] Interestingly, Hsieh
et al. found size-independent internalization of QDs.[121] In
addition, while high concentrations of QDs can be toxic to
cells, the group found that concentration influenced QDs
delivery into cells with a low concentration (15 nm) effec-
tively labeling cells while a 10-fold increase in QD concen-
tration, resulting in poor cellular uptake. However, this was
not the typical finding. As with many chemicals, cytotoxicity
of QDs was also found by many research groups to be dose
dependent with higher concentrations, resulting in signifi-
cantly higher cell death.[113,114,116,120]
In vitro cytotoxicity studies report findings similar to
in vivo studies that QDs are taken up and sequestered intra-
cellularly. Jaiswal et al. demonstrated that targeted CdSe/
ZnS QDs could be internalized by HeLa cells and tracked
in live cells for more than 10 days with no morphological
signs of toxicity.[122] Hanaki et al. studied how long MUA-
coated QDs could stay in Vero cells. QD-containing ve-
containing vesicles.[123] The number of vesicle-containing
cells reduced to half after three days and about 10% of the
cells contained QD vesicles after five days. These findings
correspond to introducing a QD concentration of
0.4 mg mL 1, which was found to have no cytotoxicity. A
more long-term study was conducted by Seleverstov et al.,
who found that QD-labeled cells retained their fluorescent
signal for 52 days in both continuous culture or after cell
passaging.[34] The QDs were internalized and observed
mainly within endosomes near the perinuclear region with
no nuclear involvement. In addition, QD aggregates were
found localized around the mitochondria and after 72 hours
morphological effects included swollen mitochondria and
www.small-journal.com 43
 reviews
enlarged Golgi cisterns. Hsieh et al. confirmed the perinu-
clear localization of QDs and found that cells retained QDs
for at least three weeks.[121] While these studies have shown
that cells can survive long after internalizing QDs, Chang
et al. found with three different QDs (bare, two PEG
coated) that cytotoxicity is similar between the differently
coated QDs when their intracellular concentration is the
same, as seen in Figure 4. Therefore, while extracellular con-
centrations of different QDs suggest dose-dependent cyto-
toxicity, cytotoxicity is dependent on the intracellular QD
content, and biocompatibility can be improved by minimiz-
ing QD uptake.[124,125]
Figure 4. Cytotoxicity results based on intracellular levels of bare
QDs (left image, black bars), and 750- (middle image, gray bars) and
6000- (right image, light gray bars) Mw PEG-substituted QDs.
Reprinted with permission from Ref. [124].
The ability of QDs to be internalized by cells has led
some groups to pursue using CdSe/ZnS QDs for nuclear tar-
geting. Chen et al. revealed that silane-coated CdSe/ZnS
QDs conjugated with the SV40 nuclear localization signal
(NLS) protein only entered the nucleus of 15% of the
cells.[126] Perinuclear accumulation was still observed for the
majority of the NLS-QDs. However, QDs conjugated to a
random peptide did not enter the nucleus and only localized
randomly within the cells. Testing for cytotoxicity on HeLa
cells, the group found most of the transfected cells survived
in all the experiments thus implying negligible toxicity even
with nuclear exposure to QDs.
Since QDs are capable of entering the nucleus, several
groups have suggested QD interaction with nuclear DNA or
proteins to be a factor in their cytotoxicity. Green et al. re-
ported data corroborating this theory as they found biotin-
coated CdSe/ZnS QDs were able to nick DNA in an in vi-
tro, cell-free assay.[127] Hsieh et al. showed that QDs can
alter gene expression in human bone-marrow mesenchymal
stem cells. While flow cytometry analysis showed that the
internalized QDs did not change the cell-cycle distribution
of hBMSCs compared to the control, an inhibited response
of hBMSCs to osteogenesis was found as ALP activity was
significantly suppressed, and mRNA expression of osteo-
pontin and osteocalcin, two osteogenesis specific markers,
44 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
R. Drezek et al.
was also inhibited. These effects could be reproduced using
QDs of various sizes.[121] A DNA microarray study was re-
cently published by Zhang et al., examining the impact of
treating human lung and skin epithelial cells with two doses
of PEG-silane QDs. No adverse effects were found in lung
epithelial cells; however, the skin epithelial cells exhibited
cell-cycle regulator gene repression. Overall, fewer than
50 genes showed significant expression changes after PEG-
silane QD treatment. However, the group did not find in-
volvement of the genes that are associated with heavy metal
exposure. In addition, no pronounced difference in pheno-
typic response was found between low or high QD doses.
Higher QD doses led to more particle uptake made appar-
ent from the stronger measured fluorescent signal.[128]
5.2. Cadmium Telluride Quantum Dots
Besides CdSe/ZnS QDs, one group has published several
studies on CdTe QDs, which also have potential in biomedi-
cal applications such as bioimaging. The initial study pre-
sented by Lovric et al. tested cell exposure to both red
(  5.2 nm) and green (  2.2 nm) CdTe QDs coated with
mercaptopropionic acid (MPA) and cysteamine (Cys).[133] A
range of concentrations (0.01–100 mg mL 1) was used to
test the effect of exposure on metabolic activity. For both
types of QD, a decrease in metabolic activity was found at
concentrations of 10 mg mL 1 or more. Looking at cell mor-
phology, the group found that the rat pheochromocytoma
(PC12) cells took up the QDs at both the low
(3.75 mg mL 1) and high (37.5 mg mL 1) concentrations.
However, at the high concentration, chromatin condensa-
tion and membrane fragmentation were observed, indicative
of apoptosis. In addition, cytotoxicity was more noticeable
with the smaller QDs than with the larger QDs at the same
concentrations. To explain this difference, the group noted
that the red QDs were found primarily in the cell cytoplasm
with none entering the nucleus, while the green QDs were
mainly found in the cell nucleus. This suggests that since the
smaller QDs could access the nucleus they could cause
damage to DNA and induce apoptosis or cell death. As QD
cytotoxicity is believed to be due to free-radical formation
caused by the presence of free Cd2 + from the degradation
of the QD core, the effect of free-radical scavengers N-ace-
tylcysteine (NAC) and Trolox, as well as adding another
protective coating, bovine serum albumin (BSA), was
tested. Both NAC and BSA but not Trolox significantly re-
duced CdTe QD toxicity, suggesting that Cd2 + is a factor in
QD-induced toxicity.[133]
Zhang et al. had similar findings after testing green-
( 2 nm), yellow- (  4 nm), and red- (  6 nm) light emit-
ting, uncoated CdTe quantum dots in human hepatoma cells
(HepG2). A size-dependent difference in toxicity was ob-
served as the IC50 values of the green, yellow, and red
CdTe quantum dots were 3.0, 4.8, and 19.1 mm, respectively,
where IC50 corresponds to the concentration causing a 50%
reduction in MTT activity. Therefore, they confirmed the
findings presented earlier by Lovric et al. that smaller QDs
are significantly more cytotoxic than larger QDs. In addi-
small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles
tion, they looked for the effect of CdTe quantum dots in-
jected into Sprague-Dawley rats. Few signs of morbid toxici-
ty were observed after QD exposure. The group saw no
organ damage, only a small decrease in body weight, and a
temporary decrease in locomotion activity occurring just
after injection. This change was attributed to possible effects
of the quantum dots on neural function, which is plausible
since some nanoparticles can pass the blood–brain barrier.
However, the group recognizes that this change could also
be due to exposure to free cadmium.[134]
The effect of different surface coatings was explored by
Cho et al. using four cadmium QDs (MPA-, Cys-, or NAC-
coated CdTe QDs plus cysteamine-coated CdSe/ZnS QDs)
on human breast-cancer (MCF-7) cells. While minimal re-
duction in metabolic activity occurred after exposure to
Cys-coated CdSe/ZnS QDs, exposure to MPA and Cys-
coated CdTe QDs caused a significant decrease in cellular
metabolic activity with a less distinct decrease in the NAC-
coated CdTe QDs. This finding was confirmed using the
Trypan blue cell viability assay, which revealed significant
cell death with CdTe QDs but not with CdSe/ZnS QDs
after 24 hours of QD exposure.[130] In addition to using sur-
face coatings to reduce cytotoxicity, Liu et al. attempted to
alter the synthesis condition to improve biocompatibility of
CdTe QDs. While the dose-dependent reduction in cell via-
bility after CdTe QD treatment was confirmed, this was
found regardless of either air or nitrogen fabrication condi-
tions.[135]
Recently, Choi et al. revealed a possible signaling path-
way involved in CdTe QD-induced cell death. Activation of
the Fas receptor results in a signaling cascade that culmi-
nates in apoptosis. This study found significant upregulation
of Fas expression on the surface of neuroblastoma (SH-
SY5Y) cells treated with Cys-coated and NAC-conjugated
QDs in comparison to control cells. However, NAC-capped
QDs had little Fas upregulation, and NAC pretreated cells
exposed to Cys-QDs had no Fas upregulation, suggesting
that oxidative stress caused by QD exposure induces Fas ex-
pression.[136]
6. Summary and Outlook
In this paper, cytotoxicity data on carbon-, metal-, and
semiconductor-based nanoparticles have been reviewed. In
general, cells can survive short-term exposure to low con-
centrations (< 10 mg mL 1) of nanoparticles. However at
high doses, several groups have found cytotoxic effects to
emerge in a dose- and time-dependent manner for all of the
nanoparticles reviewed here. While the causes for the in-
crease in cell death observed at higher concentrations and
longer exposure times are material specific, the generation
of reactive oxygen species and the influence of cell internal-
ization of nanoparticles are two common findings through-
out. While this Review attempts to draw parallels between
the research that has currently been conducted and publish-
ed on several classes of nanoparticles, there are still gaps in
knowledge about the interaction of nanoparticles with the
body. Although several studies have been conducted, many
small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
of the earlier experiments were not designed to isolate the
source of the cytotoxicity, allowing the different physio-
chemical properties of the nanoparticles plus experimental-
setup factors to influence and confound the findings. In ad-
dition, a systematic approach to testing has not been estab-
lished.
While much of the function of nanoparticles is due to
their core structure, the surface coating defines much of
their bioactivity. For many nanoparticles to be useful in bio-
logical applications, the addition of some type of surface
coating is required. In the case of quantum dots, surface
coatings serve both to contain the cadmium particles from
leeching and to make the particles water soluble. This addi-
tion of surface coatings confounds the bioactivity and poten-
tial toxicity of the functional groups on the nanoparticle sur-
face with the core nanoparticle making it difficult to inter-
pret the observed changes. For example, many nanoparticles
are not water soluble and therefore require the addition of
a hydrophilic surface coating. However, as seen in MWNT
and QD studies, adding certain hydrophilic molecules re-
sults in lower cell viability as the functional groups them-
selves were toxic.[11,22,114,118] Surface charge also plays a role
in toxicity with cationic surfaces being more toxic than
anionic, and neutral surfaces being most biocompatible.[12]
This may be due to the affinity of cationic particles to the
negatively charged cell membrane. Therefore, adding a coat-
ing that makes the nanoparticle more cationic could make
the nanoparticle appear more toxic than it inherently is.
Traditionally, in vitro toxicity testing focuses on whether
or not exposure to a potentially toxic agent results in cell
death. However, although no cell damage or death may be
apparent after nanoparticle exposure, changes in cellular
function may result. Therefore, it is important to verify that
the end points chosen to signify cytotoxicity are appropriate.
For example, if nanoparticle exposure induces cell senes-
cence but not cell death, this could be considered a toxic
effect as cell proliferation has been disturbed. Looking over
the cytotoxicity assays commonly used in the studies re-
viewed, most either determine membrane damage, metabol-
ic irregularities, or inflammatory response, which may not
materialize with cell senescence. Therefore, sub-lethal cellu-
lar changes should also be taken into account and tested for
when evaluating the effects of nanoparticle exposure on
cells. One way to do this, which a few groups have explored,
is to conduct genomic and proteomic array tests to explore
the cellular signaling alterations behind the toxicity.
In addition, it is important that the assays used to deter-
mine cytotoxicity are valid for the materials being tested.
One example, the neutral red test, has come into question
as it relies on the adsorption of the dye to detect living cells.
Carbon black has been shown to adsorb neutral red dye
molecules giving false positive results.[9] This suggests that
carbon nanomaterials could encounter the same interfer-
ence, and they have been shown to adsorb a similarly struc-
tured chemical, naphthalene.[137,138] The MTT assay has also
come under scrutiny as groups have found discrepancies be-
tween the MTT assay results and those from other assays. In
a study conducted by Pulskamp et al. on SWNTs, the MTT
assay was the only test that revealed a dose-dependent de-
www.small-journal.com 45
 reviews
crease of cell viability. The WST-assay results did not agree
with the MTT assay as it indicated no significant loss in cell
viability except for a small reduction at the highest concen-
tration. PI and annexin stains were also used for validation,
and they confirmed the WST-assay results.[64] To explain the
difference in results, Worle-Knirsch et al. proposed that
SWNTs interact with the MTT-formazan crystals but not
with WST, XTT, or INT reagents. SWNTs attach to the in-
soluble MTT formazan product disrupting the distinguish-
ing, colorimetric reaction. This would account for the find-
ing that very pure SWNTs reduced cell viability below 50%
for MTT assays, but exhibit almost no loss according to the
WST, LDH, and MMP assays.[69]
To date, there is a lack of consensus in the published lit-
erature on nanoparticle toxicity due to variable methods,
materials, and cell lines. Nanotoxicology has emerged re-
cently to apply traditional toxicology methodologies to the
study of nanomaterial toxicity; however, standardization in
experimental set up such as choice of model (cell line,
animal species) and exposure conditions (cell confluency,
exposure duration, nanoparticle-concentration ranges and
dosing increments) is necessary in order for comparisons be-
tween studies conducted by different groups to be effec-
tive.[139] With respect to model choice, both animal- and
human-derived cells have been used. Since the potential
toxicity of nanoparticles in humans is in question, human
cells should be used to better predict human toxicity. In ad-
dition, the cell types tested in cytotoxicity tests should also
be consistently studied. Several groups have tested potential
lung or dermal toxicity; however, in the case of oral or in-
travenous exposure, many internal organ sites can be ex-
posed. While some groups have looked at liver and kidney
exposure, these studies have mainly been conducted using
quantum dots with few to none using fullerenes and gold
and iron oxide nanoparticles. Fewer nanoparticle studies
have been conducted using heart, blood, and brain
cells.[12,49,114,115,136] Detailed recommendations have been out-
lined by a Nanomaterial Toxicity Screening Working
Group.[140]
Consistency in reporting the physiochemical characteris-
tics of nanoparticles would also facilitate re-examination
and cross comparison of nanoparticle toxicity data. Stand-
ardization of materials is more challenging as nanoparticle
characterization can be difficult. Sizing of nanoparticles can
be done using methods such as scanning and transmission
electron microscopy, dynamic light scattering, and size-ex-
clusion chromatography; however, the size values obtained
can vary between these methods. A standard technique for
measuring and reporting the hydrodynamic sizes of nano-
particles would be valuable. Determining the concentration
of nanoparticles in solution is more difficult. Concentration
can be calculated from the optical density using the Beer–
Lambert law given the extinction coefficient of the nanopar-
ticle. However, as Yu et al. point out, the extinction-coeffi-
cient values published for quantum dots differ between
groups by an order of magnitude.[141] Cryogenic TEM was
used as an alternative method of determining concentration.
This method involves direct counting of particles in a rela-
tively fixed volume. As concentration or dose plays a signifi-
46 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
R. Drezek et al.
cant role in biomedical applications of nanoparticles, having
a standard technique of calculating this value is important.
Another important aspect of toxicology is the burden of
multiple dosing of nanoparticles. In the case of bioimaging,
exposure of nanoparticle contrast agent would not occur
once but repeatedly with each screening or diagnostic scan-
ning session. All of the studies reviewed in this paper in-
volved only one administration of various concentrations of
nanoparticles with cytotoxicity tests taken at different time
points. However, this is most likely because most were in vi-
tro versus in vivo studies. As toxicity studies move more
into in vivo nanoparticle evaluation, future experiments
need to incorporate the effect of multiple exposures to
nanoparticles to determine the extent of clearance and bio-
accumulation.
Most of the in vitro studies presented in this paper
assess dosimetry merely by observing the dose-response re-
lationship after external introduction of different concentra-
tions of nanoparticles. Yet, as cells are seen to readily inter-
nalize nanoparticles, the number of internalized nanoparti-
cles correlates to cytotoxicity as Chang et al. revealed.[124]
Future research should measure and record the cellular
dose in addition to the administered dose to better charac-
terize the extent of nanoparticle exposure. Strategies to de-
termine the particokinetics in in vitro systems have been
suggested by Teeguarden et al.[142] In addition, several of the
studies have suggested that after internalization, the groups
observed a persistence of nanoparticles within cells. This se-
questration of nanoparticles could elicit inflammatory re-
sponses, cell-cycle irregularities, and gene-expression altera-
tions. Unfried et al. have reviewed the mechanisms in which
nanoparticles are taken up and processed by cells; however,
knowledge in this area is still limited.[143] Future research
should focus on understanding how cells internalize nano-
particles so that methods to prevent cell opsonization of
nanoparticles can be developed. This potentially could im-
prove the in vivo biocompatibility and clearance of nanopar-
ticles.
One clear result from this analysis is that there is disa-
greement as to what constitutes low toxicity. This may be
due to the lack of a reference nanoparticle system to use as
a benchmark for comparison. Given that some standard in
vitro testing methods will be established, it may be applica-
ble to use gold nanoparticles as a reference nanoparticle for
low toxicity. Gold nanoparticles have been reported to
induce little toxicity, around 15% reduction in cell viability,
at 200 mg mL 1.[28,29,31,84] While higher concentrations could
elicit a cytotoxic effect, many substances become toxic at
high concentrations. Therefore, it may be reasonable to con-
clude that the results from cytotoxicity testing of other
nanoparticle types suggest low toxicity if those results are
similar for gold-nanoparticle solutions containing relatively
the same size particles at the same concentration.
Although nanoparticle-induced cytotoxicity has been re-
ported by several groups, it is important to keep in mind
that in vitro results can differ from what is found in vivo
and are not necessarily clinically relevant. In addition, the
risk of any potentially toxic substance is not only a function
of hazard but also chance of exposure. The nanoparticle
small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles
concentrations needed for biomedical applications have not
been optimized so the levels at which patients may be ex-
posed are not certain. At the current stage in nanoparticle
safety research, it would be premature to conclude, based
on the present studies published, that nanoparticles are in-
herently dangerous. However, now that a basis has been es-
tablished, future research should strive to address the defi-
ciencies in current cytotoxicity testing and exploit the find-
ings to engineer improved nanoparticles ultimately for clini-
cal use.
Acknowledgements
The authors would like to thank Joseph Chang and Ying Hu
for their editorial assistance. This work was supported by the
Center for Biological and Environmental Nanotechnology (NSF
EEC-0118007 and EEC-0647452) and the Howard Hughes
Medical Institute.
[1] ASTM E 2456-06 “Terminology for Nanotechnology.” ASTM In-
ternational, 2006.
[2] http://www.raeng.org/uk/policy/reports/nanoscience.htm
[3] A. D. Maynard, Nanotechnology: A Research Strategy for Ad-
dressing Risk Washington DC, Woodrow Wilson International
Center for Scholars, 2006.
[4] H. Wolinsky, EMBO Reports 2006, 7, 858–861.
[5] T. C. Long, N. Saleh, R. D. Tilton, G. V. Lowry, B. Veronesi, Envi-
ron. Sci. Technol. 2006, 40, 4346–4352.
[6] S. Fiorito, A. Serafino, F. Andreola, P. Bernier, Carbon 2006,
44, 1100–1105.
[7] E. Borenfreund, J. Puerner, Toxicol. Lett. 1985, 24, 119–24.
[8] E. Flahaut, M. Durrieu, M. Remy-Zolghadri, R. Bareille, C.
Baquey, Carbon 2006, 44, 1093–1099.
[9] N. Monteiro-Riviere, A. Inman, Carbon 2006, 44, 1070–1078.
[10] S. Altman, L. Randers, G. Rao, Biotechnol. Prog. 1993, 9, 671–
674.
[11] M. Bottini, S. Bruckner, K. Nika, N. Bottini, S. Bellucci, A. Ma-
grini, A. Bergamaschi, T. Mustelin, Toxicol. Lett. 2006, 160,
121–6.
[12] C. Goodman, C. McCusker, T. Yilmaz, V. Rotello, Bioconjugate.
Chem. 2004, 15, 897–900.
[13] C. Sayes, J. Fortner, W. Guo, D. Lyon, A. Boyd, K. Ausman, Y.
Tao, B. Sitharaman, L. Wilson, J. Hughes, J. West, V. Colvin,
Nano Lett. 2004, 4, 1881–1887.
[14] L. R. Hirsch, R. J. Stafford, J. A. Bankson, S. R. Sershen, B.
Rivera, R. E. Price, J. D. Hazle, N. J. Halas, J. L. West, PNAS
2003, 100, 13 549–13 544.
[15] C. Loo, A. Lin, L. Hirsch, M. H. Lee, J. Barton, N. Halas, J. West,
R. Drezek. Technol. Cancer Res. Treat. 2004, 3, 33–40.
[16] P. Moore, I. MacCoubrey, R. Haugland, J. Cell Biol. 1991, 111,
58A.
[17] J. Muller, F. Huaux, N. Moreau, P. Misson, J. Heilier, M. Delos,
M. Arras, A. Fonseca, J. Nagy, D. Lison, Toxicol. Appl. Pharma-
col. 2005, 207, 221–231.
[18] M. Uo, K. Tamura, Y. Sato, A. Yokoyama, F. Watari, Y. Totsuka,
K. Tohji, Small 2005, 1, 816–819.
[19] G. Haslam, D. Wyatt, P.A. Kitos, Cytotechnology 2000, 32, 63–
75.
[20] T. Mosmann, J. Immunol. Methods 1983, 65, 55–63.
[21] G. Jia, H. Wang, L. Yan, X. Wang, R. Pei, T. Yan, Y. Zhao, X. Guo,
Environ. Sci. Technol. 2005, 39, 1378–1383.
small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
[22] A. Magrez, S. Kasas, V. Salicio, N. Pasquier, J. Seo, M. Celio, S.
Catsicas, B. Schwaller, L. Forro, Nano Lett. 2006, 6, 1121–
1125.
[23] F. R. Tian, D. X. Cui, H. Schwarz, G. G. Estrada, H. Kobayashi,
Toxicol. In Vitro 2006, 20, 1202–1212.
[24] E. Connor, J. Mwamuka, A. Gole, C. Murphy, M. Wyatt, Small
2005, 1, 325–327.
[25] R. Shukla, V. Bansal, M. Chaudhary, A. Basu, R. R. Bhonde, M.
Sastry, Langmuir 2005, 21, 10644–10654.
[26] H. Takahashi, Y. Niidome, T. Niidome, K. Kaneko, H. Kawasaki,
S. Yamada, Langmuir 2006, 22, 2–5.
[27] J. Chlopek, B. Czajkowska, B. Szaraniec, E. Frackowiak, K.
Szostak, F. Beguin, Carbon 2006, 44, 1106–1111.
[28] W. Fu, D. Shenoy, J. Li, C. Crasto, G. Jones, C. Dimarzio, S. Srid-
har, M. Amiji, Mater. Res. Soc. Symp. Proc. 2005, p. 845.
[29] D. Shenoy, W. Fu, J. Li, C. Crasto, G. Jones, C. Dimarzio, S. Srid-
har, M. Amiji, Int. J. Nanomed. 2006, 1, 5–58.
[30] G. Malich, B. Markovic, C. Winder, Toxicology 1997, 124, 179–
192.
[31] A.K. Salem, P. C. Searson, K. W. Leong, Nat. Mater. 2003, 2,
668–71.
[32] H. Tominaga, M. Ishiyama, F. Ohseto, K. Sasamoto, T. Hanamo-
to, K. Suzuki, M. Watanabe, Anal. Commun. 1999, 36, 47–50.
[33] A. Shvedova, V. Castranova, E. Kisin, D. Schwegler-Berry, A.
Murray, V. Gandelsman, A. Maynard, P. Baron, J. Toxicol. Envi-
ron. Health, Part A. 2003, 66, 1909–1926.
[34] O. Selverstov, O. Zabirnyk, M. Zscharnack, L. Bulavina, M. Now-
icki, J. M. Heinrich, M. Yezhelyev, F. Emmrich, R. O’Regan, A.
Bader, Nano Lett. 2006, 6, 2826–32.
[35] S. Selvan, T. Tan, Adv. Mater. 2005, 17, 1620–1625.
[36] J. O’Brien, I. Wilson, T. Orton, F. Pognan, Eur. J. Biochem. 2000,
267, 5421–5426.
[37] C. Vandeputte, I. Guizon, I. Genestie-Denis, B. Vannier, G. Lor-
enzon, Cell Biol. Toxicol. 1994, 10, 415–421.
[38] B. Halliwell, S. Chirico, Am. J. Clin. Nutr. 1993, 57, 715S–
725S.
[39] K. Ley. Physiology of Inflammation Oxford University Press,
New York, 2001.
[40] C. Dinarello, Chest 2000, 188, 503–508.
[41] N. Favre, G. Bordmann, W. Rudin, J. Immuno. Meth. 1997, 204,
57–66.
[42] K. Kostarelos, L. Lacerda, G. Pastorin, W. Wu, S. Kieckowski, J.
Luangsivilay, S. Godefroy, D. Pantarotto, J. Briand, S. Muller,
M. Prato, A. Bianco, Nat. Nanotech. 2007, 2, 108–133.
[43] D. Pantarotto, J. Briand, M. Prato, A. Bianco, Chem. Commun.
2004, 16–17.
[44] D. Cui, F. Tian, C. Ozkan, M. Wang, H. Gao, Toxicol. Lett. 2005,
155, 73–85.
[45] N. Kam, T. Jessop, P. Wender, H. Dai, J. Am. Chem. Soc. 2004,
126, 6850–6851.
[46] M. King, J. Immunol. Meth. 2000, 243, 155–166.
[47] D.W. Fairbairn, P. L. Olive, K. L. O’Neill, Mutat. Res. 1995, 339,
37–59.
[48] L. Ding L, J. Stilwell, T. Zhang, O. Elboudware, H. Jiang, J.
Selgue, P. Cooke, J. Gray, F. Chen, Nano Lett. 2005, 5, 2448–
2464.
[49] C. M. Sayes, A. M. Gobin, K. D. Ausman, J. Mendez, J. L. West,
V. L. Colvin, Biomaterials 2005, 26, 7587–7595.
[50] H. Yamawaki, N. Iwai, Am. J. Physiol. Cell Physiol. 2006, 290,
C1495–502.
[51] F. Chen, D. Gerion, Nano Lett. 2004, 4, 1827–1832.
[52] Y. P. Sun, B. Zhou, Y. Lin, W. Wang, K. Fernando, P. Pathak, M.
Meziani, B. Harruff, X. Wang, H. Wang, P. Luo, H. Yang, M.
Kose, B. Chen, L. Veca, S. Y. Zie, J. Am. Chem. Soc. 2006, 128,
7756–7757.
[53] A. Schrand, H. Huang, C. Carlson, J. Schlager, E. Osawa, S.
Hussain, L. Dai, J. Phys. Chem. B 2007, 111, 2–7.
www.small-journal.com 47
 reviews
[54] R. Service, Science 2004, 304, 42.
[55] S. Bandow, F. Kokai, K. Takahashi, M. Yudasaka, L. C. Qin, S.
Iijima, Chem. Phys. Lett. 2000, 321, 514–519.
[56] E. W. Godly, R. Taylor, Pure Appl. Chem. 1997, 69, 1411–1434.
[57] A. Bianco, K. Kostarelos, C. Partidos, M. Prato, Chem.
Commun. 2005, 571–577.
[58] N. Sinha, J. Yeow, IEEE Trans. Nanobio. 2005, 4, 180–195.
[59] P. Kohli, C. Martin, Curr. Pharm. Biotech. 2005, 6, 35–47.
[60] B. Panessa-Warren, J. Warren, S. Wong, J. Misewich, J. Phys.:
Condens. Matter 2006, 18, S2185–S2201.
[61] S. K. Smart, A. I. Cassady, G. Q. Lu, D. J. Martin, Carbon 2006,
44, 1034–1047.
[62] A. Porter, K. Muller, J. Skepper, P. Midgley, M. Welland, Acta Bi-
omaterialia 2006, 2, 409–419.
[63] J. G. Rouse, J. Z. Yang, A. R. Barron, N. A. Monteiro-Riviere,
Toxicol. In Vitro 2006, 20, 1313–1320.
[64] K. Pulskamp, S. Diabate, H. Krug, Toxicol. Lett. 2007, 168, 58–
74.
[65] S. Manna, S. Sarkar, J. Barr, K. Wise, E. Barrera, O. Jejelowo, A.
Rice-Ficht, G. Ramesh, Nano Lett. 2005, 5, 1676–84.
[66] C. Sayes, F. Liang, J. Hudson, J. Mendez, W. Guo, J. Beach, V.
Moore, C. Doyle, J. West, W. Billups, K. Ausman, V. Colvin, Toxi-
col. Lett. 2006, 161, 135–142.
[67] P. Wick, P. Manser, L. Limbach, U. Dettlaff-Weglikowska, F. Kru-
meich, S. Roth, W. Stark, A. Bruinink, Toxicol. Lett. 2007, 168,
121–131.
[68] P. Cherukuri, S. Bachilo, S. Litovsky, R. Weisman, J. Am. Chem.
Soc. 2004, 126, 15638–15639.
[69] J. Worle-Knirsch, K. Pulskamp, H. Krug, Nano Lett. 2006, 6,
1261–1268.
[70] Y. Sato, A. Yokoyama, K. Shibata, Y. Akimoto, S. Ogino, Y. No-
dasaka, T. Kohgo, K. Tamura, T. Akasaka, M. Uo, K. Motomiya,
B. Jeyadevan, M. Ishiguro, R. Hatakeyama, F. Watari, K. Tohji,
Mol. Biosyst. 2005, 1, 176–82.
[71] F. Witzmann, N. Monteiro-Riviere, Nanomedicine 2006, 2, 158–
168.
[72] L. E. Murr, K. M. Garza, K. F. Soto, A. Carrasco, T. G. Powell,
D. A. Ramirez, P. A. Guerrero, D. A. Lopez, J. Venzor, Int. J. Envi-
ron. Res. Public Health 2005, 2, 31–42.
[73] J. Aaseth, M. Haugen, O. Forre, The Analyst 1998, 123, 3–6.
[74] S. Chen, Z. Wang, J. Ballato, S. Foulger, D. Carroll, J. Am. Chem.
Soc. 2003, 125, 16 186–16 187.
[75] J. Chen, B. Wiley, Z. Li, D. Campbell, F. Saeki, H. Cang, L. Au, J.
Lee, X. Li, Y. Xia, Adv. Mater. 2005, 17, 2255–2261.
[76] M. Thomas, A. Klibanov, PNAS 2003, 100, 9138–9143.
[77] A. Tkachenko, H. Xie, D. Coleman, W. Glomm, J. Ryan, M. An-
derson, S. Franzen, D. Feldheim, J. Am. Chem. Soc. 2003, 125,
4700–4701.
[78] A. Tkachenko, H. Xie, Y. Liu, D. Coleman, J. Ryan, W. Glomm, M.
Shipton, S. Franzen, D. Feldheim, Bioconjugate Chem. 2004,
15, 482–490.
[79] N. Pernodet, X. Fang, Y. Sun, A. Bakhtina, A. Ramakrishnan, J.
Sokolov, A. Ulman, M. Rafailovich, Small 2006, 2, 766–773.
[80] A. Lin, N. Lewinski, J. West, N. Halas, R. Drezek, J. Biomed. Opt.
2005, 10, 0 604 035.
[81] A. Lin, N. Lewinski, M. H. Lee, R. Drezek, J. Nanoparticle Res.
2006, 8, 681–692.
[82] C. Loo, A. Lowery, N. Halas, J. West, R. Drezek, Nano Lett.
2005, 5, 709–711.
[83] W. D. James, L. R. Hirsch, J. L. West, P. D. O’Neal, J. D. Payne, J.
Radioanal. Nucl. Chem. 2007, 271, 455–459.
[84] C. H. Su, H. S. Sheu, C. Y. Lin, C. C. Huang, Y. W. Lo, Y. C. Pu,
J. C. Weng, D. B. Shieh, Y. H. Chen, C. S. Yeh, J. Am. Chem. Soc.
2007, 129, 2139–2146.
[85] H. Takahashi, T. Niidome, A. Nariai, Y. Niidome, S. Yamada,
Nanotechnology 2006, 17, 4431–4435.
48 www.small-journal.com  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
R. Drezek et al.
[86] T. Niidome, M. Yamagata, Y. Okamoto, Y. Akiyama, H. Takaha-
shi, T. Kawano, Y. Katayama, Y. Niidome, J. Controlled Release
2006, 114, 343–347.
[87] T. B. Huff, M. N. Hansen, Y. Zhao, J. X. Cheng, A. Wei, Langmuir
2007, 23, 1596–1599.
[88] J. Bulte, D. Kraitchman, NMR Biomed. 2004, 17, 484–499.
[89] L. LaConte, N. Nitin, G. Bao, Materials Today 2005, 8, 32–38.
[90] C. Corot, P. Robert, J. Idee, M. Port, Adv. Drug Del. Rev. 2006,
58, 1471–1504.
[91] D. Thorek, A. Chen, J. Czupryna, A. Tsourkas, Ann. Biomed.
Eng. 2006, 34, 23–38.
[92] A. K. Gupta, R. Naregalkar, V. Vaidya, M. Gupta, Nanomedicine
2007, 2, 23–39.
[93] A. K. Gupta, S. Wells, IEEE Trans. Nanobio. 2004, 3, 66–73.
[94] A. K. Gupta, M. Gupta, Biomaterials 2005, 26, 1565–1573.
[95] W. W. Yu, E. Chang, C. M. Sayes, R. Drezek, V. L. Colvin, Nano-
technology 2006, 17, 4483–4487.
[96] S. M. Hussain, K. L. Hess, J. M. Gearhart, K. T. Geiss, J. J.
Schlager, Toxicol. In Vitro 2005, 19, 975–83.
[97] F. Y. Cheng, C. H. Su, Y. S. Yang, C. S. Yeh, C. Y. Tsai, C. L. Wu,
M. T. Wu, D. B. Shieh, Biomaterials 2005, 26, 729–38.
[98] A. Petri-Fink, M. Chastellain, L. Juillerat-Jeanneret, A. Ferrari, A.
Hofmann, Biomaterials 2005, 26, 2685–2694.
[99] F. Cengelli, D. Maysinger, F. Tschudi-Monnet, X. Montet, C.
Corot, A. Petri-Fink, H. Hofmann, L. Juillerat-Jeanneret, J. Phar-
macol. Exp. Ther. 2006, 318, 108–116.
[100] S. Wan, J. Huang, M. Guo, H. Zhang, Y. Cao, H. Yan, K. Liu, J.
Biomed. Mat. Res. Part A 2007, 46–954.
[101] F. Hu, K. Neoh, L. Cen, E. Kang, Biomacromolecules 2006, 7,
809–816.
[102] T. Brunner, P. Wick, P. Manser, P. Spohn, R. Grass, L. Limbach,
A. Bruinink, W. Stark, Environ. Sci. Technol. 2006, 40, 4374 –
4381.
[103] T. R. Pisanic, J. D. Blackwell, V. I. Shubayev, R. R. Finones, S.
Jin, Biomaterials 2007, 28, 2572–2581.
[104] K. Muller, J. Skeppera, M. Posfaib, R. Trivedi, S. Howarth, C.
Corot, E. Lancelot, P. Thompson, A. Brown, J. Gillard, Biomateri-
als 2007, 28, 1629–1642.
[105] W. C. Chan, D. J. Maxwell, X. Gao, R. E. Bailey, M. Han, S. Nie,
Curr. Opin. Biotechnol. 2002, 13, 40–46.
[106] N. G. Portney, M. Ozkan, Anal. Bioanal. Chem. 2006, 384,
620–630.
[107] R. Hardman, Environ. Health Perspec. 2006, 114, 165–172.
[108] B. Dubertret, Science 2002, 298, 1759–1762.
[109] M. Akerman, W. Chan, P. Laakkonen, S. Bhatia, E. Ruoslahti,
PNAS 2002, 99, 12617–12621.
[110] D. Larson, W. Zipfel, R. Williams, S. Clark, M. Bruchez, F. Wise,
W. Webb, Science 2003, 330, 1434–1436.
[111] B. Ballou, C. Langerholm, L. Ernst, M. Bruchez, A. Waggoner,
Bioconjugate Chem. 2004, 15, 79–86.
[112] H. Fischer, L. Liu, K. S. Pang, W. Chan, Adv. Funct. Mater. 2006,
16, 1299–1305.
[113] A. Derfus, W. Chan, S. Bhatia, Nano Lett. 2004, 4, 11–18.
[114] 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, 331–338.
[115] W. H. Chan, N. H. Shiao, P. Z. Lu, Toxicol. Lett. 2006, 167, 191–
200.
[116] A. Shiohara, A. Hoshino, K. Hanaki, K. Suzuki, K. Yamamoto,
Microbiol. Immunol. 2004, 48, 669–675.
[117] W. C. Chan, S. Nie, Science 1998, 281, 2016–2018.
[118] A. Hoshino, K. Fujioka, T. Oku, M. Suga, Y. Sasaki, T. Ohta, M.
Yasuhara, K. Suzuki, K. Yamamoto, Nano Lett. 2004, 4, 2163–
2169.
[119] H. Duan, S. M. Nie, J. Am. Chem. Soc. 2007, 129, 3333–3339.
[120] J. P. Ryman-Rasmussen, J. E. Riviere, J. Invest. Dermatol. 2007,
127, 143–153.
small 2008, 4, No. 1, 26 – 49
Cytotoxicity of Nanoparticles
[121] S. C. Hsieh, F. F. Wang, S. C. Lin, Y. J. Chen, S. C. Hung, Y. J.
Wang, Biomaterials 2006, 27, 1656–1664.
[122] J. Jaiswal, H. Mattoussi, J. Mauro, S. Simon, Nat. Biotech.
2003, 21, 47–51.
[123] K. Hanaki, A. Momo, T. Oku, A. Komoto, S. Maenosono, Y. Ya-
maguchi, K. Yamamoto, Biochem. Biophys. Res. Commun.
2003, 302, 496–501.
[124] E. Chang, N. Thekkek, W. W. Yu, V. L. Colvin, R. Drezek, Small
2006, 2, 1412–1417.
[125] W. W. Yu, E. Chang, R. Drezek, V. L. Colvin, Biochem. Biophys.
Res. Commun. 2006, 348, 781–6.
[126] F. Chen, D. Gerion, Nano Lett. 2004, 4, 1827–1832.
[127] M. Green, E. Howman, Chem. Commun. 2005, 121–123.
[128] T. Zhang, J. Stilwell, D. Gerion, L. Ding, O. Elboudwarej, P.
Cooke, J. Gray, A. Alivisatos, F. Chen, Nano Lett. 2006, 6, 800–
808.
[129] E. Voura, J. Jaiswal, H. Mattoussi, S. Simon, Nat. Med. 2004,
10, 993–998.
[130] S. Cho, D. Maysinger, M. Jain, B. Roder, S. Hackbarth, F.
Winnik, Langmuir 2007, 23, 1974–1980.
[131] G. Guo, W. Liu, J. Liang, Z. He, H. Xu, X. Yang, Mater. Lett.
2007, 61, 1641–1644.
[132] J. Liang, Z. He, S. Zhang, S. Huang, X. Ai, H. Yang, H. Han, Ta-
lanta 2007, 71, 1675–1678.
[133] J. Lovric, S. J. Cho, F. M. Winnik, D. Maysinger, Chem. Biol.
2005, 12, 1227–1234.
[134] Y. B. Zhang, W. Chen, J. Zhang, J. Liu, G. P. Chen, C. Pope, J.
Nanosci. Nanotech. 2007, 7, 497–503.
small 2008, 4, No. 1, 26 – 49  2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
[135] Y. Liu, W. Chen, A. Joly, Y. Wang, C. Pope, Y. Zhang, J. Bovin, P.
Sherwood, J. Phys. Chem. B 2006, 110, 16992–17000.
[136] A. Choi, S. Cho, J. Desbarats, J. Lovric, D. Maysinger, J. Nano-
biotechnol. 2007, 5 : 1.
[137] X. Cheng, A. Kan, M. Tomson, J. Chem. Eng. Data 2004, 49,
675–683.
[138] X. Cheng, A. Kan, M. Tomson, J. Nanoparticle Res. 2005, 7,
555–567.
[139] G. Oberdorster, E. Oberdorster, J. Oberdorster, Environ. Health
Perspect. 2005, 113, 823–39.
[140] G. Oberdorster, 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 Toxi-
col. 2005, 2 : 8.
[141] W. Yu, E. Chang, J. Falkner, J. Zhang, A. Al-Somali, C. Sayes, J.
Johns, R. Drezek, V. Colvin, J. Am. Chem. Soc. 2007, 129,
2871–2879.
[142] J. Teeguarden, P. Hinderliter, G. Orr, B. Thrall, J. Pounds, Toxicol.
Sci. 2007, 95, 300–312.
[143] K. Unfried, C. Albrecht, L. Klotz, A. von Mikecz, S. Grether-Beck,
R. Schins, Nanotoxicology 2007, 1, 52–71.
Received: July 25, 2007
Revised: November 14, 2007
Published online on December 28, 2007
www.small-journal.com 49