Content uploaded by Fei Wen
Author content
All content in this area was uploaded by Fei Wen on Jul 20, 2022
Content may be subject to copyright.
Nanoparticle toxicity
DOI: 10.1002/smll.200700378
Size-Dependent Cytotoxicity of Gold Nanoparticles
Yu Pan, Sabine Neuss, Annika Leifert, Monika Fischler, Fei Wen,
Ulrich Simon,* Gnter Schmid, Wolfgang Brandau, and Willi Jahnen-Dechent*
Gold nanoparticles are widely used in biomedical imaging and diagnos-
tic tests. Based on their established use in the laboratory and the chemical
stability of Au0, gold nanoparticles were expected to be safe. The recent
literature, however, contains conflicting data regarding the cytotoxicity of
gold nanoparticles. Against this background a systematic study of water-
soluble gold nanoparticles stabilized by triphenylphosphine derivatives
ranging in size from 0.8 to 15 nm is made. The cytotoxicity of these
particles in four cell lines representing major functional cell types with
barrier and phagocyte function are tested. Connective tissue fibroblasts,
epithelial cells, macrophages, and melanoma cells prove most sensitive to
gold particles 1.4 nm in size, which results in IC50 values ranging from 30
to 56 mmdepending on the particular 1.4-nm Au compound–cell line
combination. In contrast, gold particles 15 nm in size and Tauredon
(gold thiomalate) are nontoxic at up to 60-fold and 100-fold higher con-
centrations, respectively. The cellular response is size dependent, in that
1.4-nm particles cause predominantly rapid cell death by necrosis within
12 h while closely related particles 1.2 nm in diameter effect predominant-
ly programmed cell death by apoptosis.
Keywords:
·cell growth
·cytotoxicity
·gold
·nanoparticles
·toxicology
1. Introduction
Nanoscale materials hold great promise for both indus-
trial and biomedical applications. Toxicological studies sug-
gest that nanoparticles may cause adverse health effects, but
the fundamental cause–effect relationships are ill defined.
Thus, the interaction of nanoparticles with biological sys-
tems including living cells has become one of the most
urgent areas of collaborative research in materials science
and biology.[1] The most interesting properties of nanoparti-
cles, that is, the quantum size effect or surface-induced ef-
fects, result from their minute size. Nanoparticles are of sim-
ilar size to typical cellular components and proteins, and
thus may bypass natural mechanical barriers, possibly lead-
ing to adverse tissue reaction. The primary interaction site
of cells and particles smaller than about 100 nm is the peri-
cellular space in and around the microcapillaries. Opsoniza-
tion, the binding of blood components to the particles,
greatly facilitates endocytosis. Opsonins mediate material–
cell receptor interactions and enhance endocytosis. Con-
[*] Y. Pan, Prof. W. Jahnen-Dechent
Biomedical Engineering, Biointerface Laboratory
RWTH Aachen University
Pauwelsstrasse 30, 52074 Aachen (Germany)
Fax: (+49) 241-80-82573
E-mail: willi.jahnen@rwth-aachen.de
Dr. S. Neuss
Pathology
RWTH Aachen University (Germany)
A. Leifert, Dr. M. Fischler, Dr. F. Wen, Prof. U. Simon
Inorganic Chemistry
RWTH Aachen University (Germany)
Fax: (+49) 241-809-9003
E-mail: ulrich.simon@ac.rwth-aachen.de
Prof. G. Schmid
Inorganic Chemistry
University of Duisburg-Essen (Germany)
Prof. W. Brandau
Radiochemistry, University Hospital Essen (Germany)
Supporting information for this article is available on the WWW
under http://www.small-journal.com or from the author.
small 2007,3, No. 11, 1941 – 1949 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1941
Cytotoxicity of Gold Nanoparticles
versely, the surface chemistry of particles can control opso-
nization, thus facilitating or preventing endocytosis. Charged
or hydrophobic particles adsorb serum proteins easily while
particles covered with antifouling polymers, such as polyeth-
ylene glycol, can resist opsonization, effectively creating
“stealth particles” with extended circulation times.
Phagocytosis is prevalent in professional phagocytes of
the monocyte/macrophage lineage including dendritic cells
and osteoclasts. Endothelial cells likewise possess a highly
evolved machinery to endocytose large and small particles.
Once the particles are endocytosed, they may be degraded
in the endolysosomal compartment. Innoxious but nonde-
gradable matter may eventually be excreted in feces or ex-
pectorated by the lung along with the phagocytosing cells.
Failure of degradation or excretion may result in chronic in-
flammation, ultimately leading to severe tissue damage. En-
docytosis of nanomaterials can trigger the binding of nano-
particles to intracellular targets, which causes disturbance of
cellular signaling, motility, and metabolism.
Gold nanoparticles were variously described as nontox-
ic[2] or toxic.[3,4] Gold nanoparticles are used as tracers[5] and
the cellular trajectories change according to the biological
signals added to the bulk material, again suggesting that
gold particles by themselves are nontoxic.[6] Furthermore,
oligonucleotide-modified 13-nm gold particles were applied
to intracellular gene regulation.[7] An entire anti-inflamma-
tory type of therapy, chrysotherapy, actually relies on gold
complexes. A recent report elaborated that goldACHTUNGTRENNUNG(III) salts
attenuate antigen presentation and thus reduce autoimmune
reactions in rheumatoid arthritis,[8] which suggests a molecu-
lar mechanism for the anti-inflammatory/anti-rheumatoid
arthritis activity of gold drugs such as Auranofin or Taure-
don.
Based on in vitro investigations of the interaction of
water-soluble Au55[Ph2PC6H4SO3H(Na)]12Cl6with natural B-
DNA and molecular modeling studies, we found that the
1.4-nm Au55 cluster cores bound to the major groove with
high selectivity and stability. This binding event was accom-
panied by a partial substitution of triphenylphosphine
monoACHTUNGTRENNUNGsulfonate (TPPMS) for negatively charged phosphates
of the DNA, which most likely bound the gold nanoparticles
due to the high electronegativity of the metal.[9] Hypotheti-
cally, this could result in a general blockade of transcription.
This mechanism would require a rather stringent size limita-
tion, because clusters, characteristically smaller or larger
than the 1.4-nm Au55 clusters, would be less likely to interact
with B-DNA in such a way for steric reasons. From these
former experiments it is also known that sulfonated triphe-
nylphosphines do not show visible toxicity under the same
conditions. These results prompted us to start a series of
cell-based experiments, in which the gold-cluster compound
showed lower IC50 values (inhibitory concentrations that ef-
fected 50% growth inhibition) than the well-established cy-
tostatic agent, cisplatin, for different human cancer cells.[10]
From [198Au]gold radioanalysis of cell fractions we deduced
that 20–25% of the gold was associated with DNA-contain-
ing fractions of the nucleus, which is in accordance with the
ex vivo experiments mentioned above.
Gold nanoparticles are readily endocytosed by mamma-
lian cells and the kinetics and saturation depend upon the
physical dimensions of the nanoparticles.[11] Pernodet and
colleagues reported that with the presence of intracellular
gold nanoparticles 131 nm in size, actin stress fibers disap-
peared and major adverse effects on cell viability were in-
duced.[4] Gold nanoparticles 3.50.7 nm in diameter capped
by lysine and poly-l-lysine were biocompatible and nonim-
munogenic.[12] Cysteine- and citrate-capped 4-nm gold nano-
particles, glucose-reduced 12-nm nanoparticles, and citrate-,
biotin-, and cetyltrimethylammonium bromide-capped 18-
nm gold nanoparticles were all endocytosed without signs of
cytotoxicity.[2] Studies from Goodman and co-workers dem-
onstrated that cationic gold nanoparticles were moderately
toxic, whereas anionic gold nanoparticles were nontoxic.[3]
A pilot study from our own laboratory showed that Au55
clusters stabilized with TPPMS of diameter 1.4 nm
(Au1.4MS) were more toxic than cisplatin in a range of cell
lines.[10] Thus, conflicting yet not necessarily contradictive
data exist with regard to the toxicity of gold nanoparticles.
Herein, we studied the influence of particle size on cyto-
toxicity. To this end, we varied the Au cluster size from 0.8
up to 15 nm. We studied the stability of the nanoparticles in
a complex, high-ionic-strength aqueous environment con-
taining macromolecules mimicking blood. Typically this is
achieved by a serum-containing cell-culture medium, which
may affect stability, solubility, and hydrophobicity. We con-
fronted the nanoparticles with various cell types represent-
ing the principal barriers and lining cells of the body (epi-
thelial and endothelial cells), phagocytes (macrophages),
and tissue stromal cells (connective tissue fibroblasts). Cells
are generally most vulnerable during proliferation and tend
to be more stress-tolerant in the quiescent state. Thus, we
tested cytotoxicity in both actively dividing cells in the loga-
rithmic growth phase and quiescent cells in the stationary
phase. In summary, we describe the comparative cytotoxicity
testing in a 96-well plate-based cell assay of gold compounds
including a commercial drug and gold nanoparticles of vary-
ing size.
2. Results and Discussion
2.1. Au Nanoparticle Synthesis and Stability in Media
We varied the size of Au nanoparticles from 0.8 to
15 nm. To exclude the influence on toxicity of various li-
gands protecting the different nanoparticles, we used
Ph2PC6H4SO3Na (TPPMS) throughout, except in one case
where tris-sulfonated triphenylphosphine P(C6H4SO3Na)3
(TPPTS) was used for comparison. The statements of size
used in the following relate to the diameter of the gold core.
We use the abbreviations MS and TS for the mono- and tri-
substituted ligands, respectively, and formulae such as
Au1.4MS for the clusters stabilized by TPPMS.
The synthesis of the cluster Au0.8MS is described in the
Experimental Section. It represents a water-soluble TPPMS
derivative of the fully characterized compound Au9ACHTUNGTRENNUNG(PPh3)8-
ACHTUNGTRENNUNG(NO3)2.[13] Au1.4MS represents the likewise well-defined
1942 www.small-journal.com 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2007,3, No. 11, 1941 – 1949
full papers U. Simon, W. Jahnen-Dechent, et al.
and characterized compound Au55(Ph2PC6H4SO3Na)12Cl6,
which is extensively described in the literature[14, 15] and is a
derivative of the well-known and fully characterized cluster
compound Au55ACHTUNGTRENNUNG(PPh3)12Cl6.[16–18] Au1.4TS was synthesized in
an analogous manner to Au1.4MS (see Experimental Sec-
tion). Au1.2MS and Au1.8MS are available from Strem
Chemicals, Newburyport.[19] Contrary to the stoichiometric
Au0.8MS and Au1.4 clusters, Au1.2MS and Au1.8MS exhibit
size distributions of about 0.2 nm derived from transmis-
sion electron microscopy (TEM) measurements. Thus, a sto-
ichiometric formula cannot be given. However, it can be es-
timated that the average-sized Au1.2MS particles consist of
35 Au atoms and the Au1.8MS particles of 150 Au
atoms. The number of TPPMS ligands corresponds exactly
to the number of Au atoms, that is, the Au/ligand ratio is
1:1, which indicates that the clusters are capped by more
than one monolayer of the ligand molecules. Gold nanopar-
ticles in the size range of 15 nm were prepared by following
the citrate method of Turkevitch.[20] The citrate layer on the
particle surface was replaced by TPPMS as described.[21]
From numerous TEM investigations it is known that the
size distribution of these colloidal species is generally
within151.5 nm.
Scheme 1 shows a model of a triphenylphosphine-
capped gold nanoparticle and Table S1 (Supporting Infor-
mation) lists key features of the nanoparticles used in this
study, as determined in previous studies or as provided by
the manufacturer.
We tested the stability of the materials in the serum-con-
taining cell-culture media formulated for proper growth of
the cell lines used in this study. We reasoned that the
medium composition might affect particle aggregation and
that aggregation of materials would greatly influence the en-
docytic pathway and ultimately the cellular trajectories of
materials precluding or favoring intracellular targets. Fig-
ure S1 (Supporting Information) shows macroscopic and mi-
croscopic aspects of medium–material combinations in the
absence of any cells. Materials were grouped according to
their stability in specific media and over time. A synopsis of
the aggregation behavior of Au compounds is given in
Table S2 (Supporting Information).
2.2. Cytotoxicity Testing of Au Nanoparticles
We added Au compounds to HeLa cervix carcinoma epi-
thelial cells (termed HeLa), SK-Mel-28 melanoma cells
(SK-Mel-28), L929 mouse fibroblasts (L929), and mouse
monocytic/macrophage cells (J774A1) and first studied their
response by optical and electron microscopy. Figure 1 shows
representative microphotographs of HeLa cells that depict
the time course of treatment with Au1.4MS (Figure 1 A–F)
and of J774A1 cells treated with the compounds indicated
(Figure 1G–L). HeLa cells, like SK-Mel-28 and L929 cells
(not shown), formed tight monolayers in the absence of Au
compounds (Figure 1A,B). Figure 2C and D represent opti-
cal and electron microscopic views of HeLa cells, respective-
ly, after 1 h of treatment with Au1.4MS. The cells were swol-
len and started losing substrate contact. TEM showed mem-
brane blebbing and vesicle formation at the plasma mem-
brane, which suggests strong activation of the cells (Fig-
ure 1D). At a treatment time of 12 h (Figure 1E,F) the cells
were strongly swollen, had fragmented nuclei, and many
cells had lost both cell-to-cell and cell-to-substrate contact.
The few cells that remained attached to the culture dish
showed cytoplasmic disorganization, nuclear fragmentation,
and membrane blebbing, thus indicating apoptosis and sec-
ondary necrosis. Figure 1 G–L shows optical micrographs of
J774A1 macrophages treated for 1 h with Au compounds as
indicated. As in HeLa cells, strong membrane blebbing was
observed in treated (Figure 1H–L) but not in untreated
J774A1 cells (Figure 1G). J774A1 cells treated with
Au15MS particles had engorged themselves with the parti-
cles (Figure 1K). Collectively the microscopic data present-
ed in Figure 1 show that the cells had endocytosed Au parti-
cles and showed signs of activation and severe stress, includ-
ing apoptosis and secondary necrosis, after treatment with
Au1.4MS clusters.
We quantified the toxicity of Au compounds in four cell
lines by determining the IC50 values in MTT assays (MTT=
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bro-
mide). We reasoned that cells actively growing and dividing
during logarithmic growth should be more vulnerable to
toxic insults than cells near or at the stationary phase of cell
culture. Thus, we determined growth curves for all cell lines
to estimate the logarithmic and stationary growth phases in
relation to the number of cells seeded into each well of a
96-well cell-culture plate. Figure 2 shows the growth curves
for all four cell lines maintained in the media detailed in the
Experimental Section. This graph served as a reference to
estimate whether cells were in the logarithmic or stationary
phase of cell culture at the start of any given experiment.
2.3. Au Nanoparticles Cause Size-Dependent Cell Death
Having established growth characteristics for the report-
er cell lines, we treated the cells with Au compounds for up
to 48 h to observe the full effect of toxicity. Given the
strong reaction of cells to Au1.4MS (and Au1.4TS) illustrat-
ed in Figure 1C–F, this incubation time should reliably
detect even low or slow-acting toxicity and vice versa. We
performed toxicity tests during both the logarithmic and sta-
tionary phases of cell growth following the time course de-
picted in Figure 2E and F. Invariably cells during the loga-
rithmic growth phase proved 1.5–3.3-fold more sensitive to
Scheme 1. Model of Au cluster with the triphenylphosphine deriva-
tive ligand. Not drawn to size.
small 2007,3, No. 11, 1941 – 1949 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 1943
Cytotoxicity of Gold Nanoparticles
toxic compounds than during the stationary phase (data not
shown). In the following figures, however, we only list IC50
values derived from cells in the logarithmic growth phase
for clarity of presentation.
A typical experiment measuring the cytotoxicity of Au
compounds in logarithmic-phase HeLa cells is illustrated in
Figure 3A. Gold clusters Au1.4MS and Au1.4TS proved
most toxic in this assay with IC50 values of 46 and 30 mm, re-
spectively. Surprisingly, Au clusters of even moderately dif-
ferent size stabilized with the same TPPMS ligand,
Au0.8MS (IC50 250 mm), Au1.2MS (IC50 140 mm), and
Au1.8MS (IC50 230 mm), were four- to sixfold less toxic. De-
spite obvious endocytosis by the cells (see Figure 1K for ex-
ample), the colloidal compound Au15MS was completely
nontoxic up to 6300 mm. Higher concentrations of this com-
pound were not tested because of solubility issues. Likewise
Tauredon (gold thiomalate) scored nontoxic in all cells
tested at concentrations up
to 10 mmexcept in J774A1
cells, which proved more sen-
sitive (IC50 600 mm).
Taken together these re-
sults suggest a stringent size
dependency of the cytotoxici-
ty of nanoscopic gold clus-
ters. The toxicity was inde-
pendent of whether TPPMS
or TPPTS was used to stabi-
lize the Au1.4 cluster. IC50
values of both compounds
were virtually identical in
HeLa cells and very similar
across the four cell lines
tested, ranging from 30 mmin
J774A1 macrophages to
56 mmin L929 fibroblasts
(Figure 3B). We also con-
clude that ligand toxicity did
not contribute to the overall
cluster toxicity, because both
Au1.4MS and Au1.4TS clus-
ters had almost identical IC50
values. Furthermore, the
Au1.4 clusters contained 12
ligand molecules per 55 Au
atoms, which suggests
10 mmligand at the IC50
value of Au1.4MS and
Au1.4TS. Figure 3A shows
that both ligands were non-
toxic at this low concentra-
tion. In addition, Au0.8,
Au1.2, and Au1.8 clusters all
contained equimolar
amounts of Au and ligand
(see Table S1, Supporting In-
formation), but were nontox-
ic at concentrations even
higher than Au1.4 clusters. In
summary, these data strongly suggest that the size of 1.4 nm
and not the ligand chemistry was the chief determinant of
toxicity of the Au clusters.
2.4. Apoptosis Versus Necrosis Caused by Au Clusters
Next we asked what kind of cell death Au clusters cause.
Basically two kinds of cell death are known. Fast-acting
metabolic poisons and strong physical stress, such as freez-
ing, boiling, or shearing, rupture cell membranes and cause
rapid cell necrosis. The contents released by necrotic cells
are highly inflammatory and therefore necrotic cells invaria-
bly cause inflammation in the body. In contrast, a slow-
acting form of cell death called apoptosis does not involve
membrane damage and inflammation. During apoptosis or
programmed cell death, cells undergo an energy-dependent
Figure 1. Microscopic views of HeLa (A–F) and J774A1 cells (G–L) treated with Au compounds as indicat-
ed. A–F) HeLa cells were treated for 0, 1, or 12 h with Au1.4MS clusters. Cells were fixed, flat embed-
ded, turned by 908, and mounted on stubs for semithin and ultrathin sectioning to reveal a cross-sec-
tional view of the monolayer. Sections were stained and viewed with an optical microscope using an
X40 lens (A,C, E) or with an electron microscope (B, D, F). Optical micrographs show a 200-mm sector of
the monolayer; scale bars in the TEM images indicate 0.5 mm(0h)or1mm (1 and 12 h). G–L) J774A1
macrophages were treated for 1 h as indicated and viewed live using an inverted optical microscope
and an X40 lens. The width of each micrograph equals 200 mm. Colloidal Au 15-nm particles stained the
endocytic compartment of the cells black (K) sparing the nucleus. The background hue is due to the
color of dissolved materials.
1944 www.small-journal.com 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2007,3, No. 11, 1941 – 1949
full papers U. Simon, W. Jahnen-Dechent, et al.
sequence of events, ultimately fragmenting nuclei and cyto-
plasmic organelles into small membrane-sealed apoptotic
bodies that can be cleared by phagocytes. Apoptosis is the
bodys default pathway of clearing dead cells or cells
marked for destruction. Membrane blebbing and vesicle for-
mation as observed in Figure 1D are typical of apoptosis. A
salient diagnostic feature of apoptosis is, however, externali-
zation of the membrane lipid phosphatidylserine (PS) to the
outer leaflet of the cell membrane. We double stained cells
with annexin V (aV) for externalized PS as a measure of
apoptosis and with the nuclear stain propidium iodide (PI)
as an indicator of membrane integrity and thus necrosis. A
typical view of healthy cells and of necrotic cells treated
with 110 mmAu1.4MS is shown in Figure 4. Untreated cells
did not expose PS on their external plasma membrane leaf-
let and stained double negative for aV and PI (Fig-
ACHTUNGTRENNUNGure 4A,B). In contrast, necrotic cells stained double posi-
tive, green for aV and red for PI (Figure 4C,D).
To estimate if the cytotoxic Au compounds caused pref-
erentially apoptotic or necrotic cell death, we treated HeLa
cells in the logarithmic growth phase with the pro-apoptotic
compound staurosporine as a positive control substance or
with Au1.2MS or Au1.4MS, two cytotoxic Au cluster com-
pounds varying only slightly in size. The cells were double
stained with aV and PI and subjected to flow cytometry
(Figure 5). Signals were gated for high forward and side-
ward scattering (FSC and
SSC, respectively) to sepa-
rate intact cells from parti-
cles and cell fragments. The
gated signals formed four
groups: aV/PI double nega-
tive, intact live cells; aV posi-
tive/PI negative, apoptotic
cells; aV positive/PI positive,
necrotic cells; and a very low
fraction of aV negative/PI
positive, large nuclear frag-
ments.
Figure 6 shows a compila-
tion of a representative ex-
periment detailing the rela-
tive amounts of live, necrotic,
and apoptotic HeLa cells
after treatment for 6, 12, 18,
and 24 h with buffer only
(untreated), staurosporine,
Au1.2MS, or Au1.4MS. As
expected, untreated cells re-
mained live, nonapoptotic
and nonnecrotic at all time
points. The positive control
staurosporine effected mostly
apopotosis, but a small frac-
tion of aV/PI double positive
cells was also detected,
which indicated secondary
necrosis especially at later
time points. The analysis re-
vealed a striking difference in the cytotoxic capacity of
Au1.2MS and Au1.4MS that was not noted from the IC50
measurements at 48 h. Interestingly, the smaller cluster com-
pound Au1.2MS at 140 mmcaused cell death in about 50 %
of all cells after treatment for 24 h with an almost equal pro-
portion of apoptotic and secondary necrotic cells, thus indi-
cating relatively lower cytotoxicity and slow killing. In con-
trast, 110 mmAu1.4MS caused cell death in 70% after 12 h
and in over 90% after 24 h with a transient population of
apoptotic cells and a steady increase in secondary necrotic
cells. We take this as evidence for a much faster and more
efficient cytotoxic action of Au1.4MS versus Au1.2MS, de-
spite a similar concentration and their close chemical and
physical similarity. Note that this important difference in
action was only revealed in the kinetic analysis, and not in
the traditional endpoint analysis for determining the IC50
values.
We repeated the experiment for Au1.2MS and Au1.4MS
with twice the IC50 concentration previously determined by
endpoint analysis (Figure 3B). Figure S2 (Supporting Infor-
mation) shows an extension of the experiment depicted in
Figure 6C and D. This experimental setup fully confirmed
the previous results, in that Au1.2MS was markedly less cy-
totoxic than Au1.4MS despite a threefold higher concentra-
tion.
Figure 2.
A–D) Cell-growth curves using different cell-plating numbers and the indicated cell lines. The initial cell
number was 1000 (green), 2000 (red), or 4000 cells (blue) seeded into each well of a 96-well tissue-cul-
ture plate. E, F) Protocol of cytotoxicity testing starting during the logarithmic (E) or stationary growth
phase (F). OD=optical density.
small 2007,3, No. 11, 1941 – 1949 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 1945
Cytotoxicity of Gold Nanoparticles
3. Conclusions
We have defined the size range, concentration range,
type of cell culture, and treatment time as important basic
parameters to unravel the exact trajectory and molecular
targets of Au nanoclusters. These clusters constitute a novel
class of tunable nanoscale materials with potential medical
application as cytostatic agents that preferably promote
apoptosis or secondary necrosis depending predominantly
on their size.
The cytotoxicity of TPPMS/TPPTS-modified gold nano-
particles depended primarily on their size and not on ligand
chemistry. Particles 1–2 nm in size were highly toxic and
both smaller gold compounds (Tauredon) and larger 15-nm
gold colloids were comparatively nontoxic, irrespective of
the cell type tested. Differences in the kind of cell death
pathway (apoptosis versus necrosis) were consistently ob-
served. This finding suggests different uptake kinetics and/
or cellular target specificities even for similarly sized gold
nanoparticles. Most likely nanoparticle toxicity follows en-
docytosis, but it is entirely possible that the toxicity may
stem from interactions at the cell membrane, even though
the particles are also endocytosed.
4. Experimental Section
Synthesis and stability of gold nanoparticles: AuPPh3Cl, ben-
zene, BF3·OEt2,CH
2Cl2, diethylene glycol dimethyl ether, ethanol,
HAuCl4·3H2O, H2SO4, NaBH4, PPh3, sodium citrate dihydrate, and
TPPTS were purchased from diverse suppliers. All chemicals
were used as received, and H2O was obtained from a Purelab
Plus water purification system. TPPMS was synthesized as de-
scribed.[22]
Au0.8MS: The Au9ACHTUNGTRENNUNG(PPh3)8ACHTUNGTRENNUNG(NO3)3cluster was synthesized as
published.[23] The single-crystal structure was described previ-
ously.[13] To transfer the cluster into the aqueous phase, a solu-
tion of TPPMS (1.1 mg, 2.8 103mmol) in H2O (0.5 mL) was
added to a solution of Au9ACHTUNGTRENNUNG(PPh3)8ACHTUNGTRENNUNG(NO3)3cluster (1.3 mg, 3.2
104mmol) in CH2Cl2(0.5 mL), and the mixture was stirred at
room temperature until the organic phase was colorless. After
centrifugation, the red supernatant was ready for further use.
The obtained water-soluble cluster solution showed a single res-
onance in 31P{1H} NMR spectra at d=54.5 ppm in D2O, which is
Figure 3. Cytotoxicity of Au compounds during the logarithmic growth
phase of four cell lines. A) HeLa cells were seeded at 2000 cells/well
and grown for 3 days into the logarithmic growth phase. Au com-
pounds were added for 48 h and MTT tests were performed as
detailed in the Experimental Section. The logarithmic curve fits of
tabulated MTT readings are shown. Each data point represents the
meanstandard error (SE) of sample triplicates. B) Note that the IC50
values of Au1.4MS were lowest across all cell lines and that Au com-
pounds of smaller or larger size were progressively less cytotoxic,
which suggests a stringent size dependency of cytotoxicity. All con-
centrations relate to the amount of gold detected by atomic-absorp-
tion spectroscopy (AAS) in the authentic samples after performing
the cytotoxicity test. This procedure ruled out the possibility that
cluster synthesis contaminants or dilution errors may have caused
erroneous results.
Figure 4. Fluorescent staining of HeLa cells with aV for apoptosis
(green fluorescence) and PI for necrosis (red fluorescence). Untreated
cells stained negative. Cells treated with 110 mmAu1.4MS for 30 h
underwent apoptosis and secondary necrosis and thus stained
double positive. Note that the treated cells also rounded up and
detached from the culture dish.
1946 www.small-journal.com 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2007,3, No. 11, 1941 – 1949
full papers U. Simon, W. Jahnen-Dechent, et al.
in good agreement with the data obtained from the CH2Cl2-solu-
ble Au8ACHTUNGTRENNUNG(PPh3)8ACHTUNGTRENNUNG(NO3)2cluster compound.[24] Furthermore, the UV/
Vis spectrum of the cluster in aqueous solution was found to be
consistent with that of the Au8core reported elsewhere.[25]
Au1.4TS: Au1.4TS was synthesized analogously to Au1.4MS
(see Ref. [20]), except that the two-phase system was stirred for
10 days instead of 3 days to achieve full ligand exchange.
Au15MS: Citrate-stabilized gold colloids were synthesized by
following a published protocol.[20] For ligand exchange, TPPMS
(1 mg per 10 mL of a 0.4 mmsolution referring to Au) was added
to the red solution. The mixture was stirred for 5 min and kept at
48C overnight. To increase the concentration and remove excess
ligand, the solution was centrifuged to obtain a dark red solu-
tion. The UV/Vis spectrum showed an absorption maximum at
527 nm.
Au1.2MS and Au1.8MS were kindly provided and character-
ized by Strem Chemicals, Newburyport.[19]
Stability tests were performed in 96-well microtiter plates.
For each test, water-dissolved material (20 mL) was added to the
medium (50 mL). Materials dissolved in culture media were ana-
lyzed microscopically after 5 min and 12 h.
Cell culture and cytotoxicity assays: Four cell lines, SK-Mel-
28 human melanoma, HeLa human cervix carcinoma, L929
mouse fibroblasts, and J774A1 mouse macrophages, were used
in this study. SK-Mel-28 and J774A1 cells were maintained in
high-glucose Dulbecco’s modified Eagle’s medium (DMEM). HeLa
cells were cultured in low-glucose DMEM. Media contained fetal
calf serum (10 %), l-glutamine (2.9 mg mL1), streptomycin
(1 mgmL1), and penicillin (1000 units mL1). L929 cells were
cultured in RPMI 1640 medium with newborn calf serum (5%), l-
glutamine (2.9 mg mL1), streptomycin (1 mg mL1), and penicil-
lin (1000 units mL1). All cells were cultured at 378C in water-sa-
turated air supplemented with 5% CO2. Culture media were
changed every 3 days. Cells were passaged once a week. Cell
numbers were estimated using a cell counter (Schaerfe cell-
counting system, Germany).
Cells were plated in 96-well microtiter plates at initial densi-
ties of 1000, 2000, and 4000 cells per well. The cell-culture
medium was changed every 3 days. Cell growth was tested by
the colorimetric MTT assay, which measures the conversion of
the yellowish water-soluble tetrazolium salt to a water-insoluble
purple formazan product within viable breathing cells as a proxy
of cell number and viability. The water-insoluble formazan was
dissolved in a solvent mixture (100 mL) consisting of isopropanol
(80 mL) with hydrochloric acid (0.04 mm) and 3% sodium dodecyl
sulfate (20 mL). Absorption of the samples was measured with a
spectrophotometer at 584 nm. The amount of formazan pro-
duced is directly proportional to the number of living cells in the
well. MTT assays were performed every day after seeding until
day 8. All experiments were carried out in triplicate.
The cytotoxicity of nanoparticles in various cell types was de-
termined by the MTT assay. Cytotoxicity was measured in both
the logarithmic and stationary phase of cell growth. For cytotox-
icity measurements in the logarithmic phase, each cell line was
incubated for 72 h in 96-well microtiter plates before adding the
nanoparticles. Fresh medium containing increasing concentra-
tions of nanoparticles was added to each well and the cells
were incubated for another 48 h. For cytotoxicity measurements
in the stationary phase, each cell line was incubated for 7 days
before the addition of nanoparticles. Fresh medium containing
various concentrations of nanoparticles was added and cells
were incubated for another 48 h. Phosphate-buffered saline
(PBS, 10 mL ) containing MTT (5 mg mL1) was dispensed into
each well and the plates were incubated for 2 h. Formazan was
solubilized and measured as described under cell growth. A
graphic illustration of the time schedule is given in Figure 2. The
concentrations of materials were rechecked by AAS after comple-
tion of the experiments on the authentic samples.
IC50 values were calculated by using a four-parameter logistic
equation. Data were plotted as a sigmoidal dose–response
Figure 5. Determination of live, apoptotic, and necrotic HeLa cells
treated with test compounds for 6 h. Cells were analyzed by flow
cytometry and gated as shown in the left-hand panels. Gated cells
were scored for aV/PI double staining (right panels) to estimate the
relative amounts of live cells (aV/PI double negative, bottom-left
quadrant), apoptotic cells (aV positive/PI negative, bottom-right
quadrant), and necrotic cells (aV/PI double positive, top-right quad-
rant). The percentages of cells are given for each quadrant. A) HeLa
untreated; B) HeLa, 2 mmstaurosporine, 6 h; C) HeLa, 140 mm
Au1.2MS, 6 h; D) HeLa, 110 mmAu1.4MS, 6 h.
small 2007,3, No. 11, 1941 – 1949 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 1947
Cytotoxicity of Gold Nanoparticles
curve with variable slope using GraphPad PRISM software. For
each material, the IC50 values were determined from triplicate
wells during both the stationary and logarithmic cell growth
phases. IC50 values derived from logarithmic cell growth were
routinely repeated in three independent experiments with
almost identical results.
Quantification of apoptotic and necrotic cells with fluores-
cence-activated cell sorting (FACS): To measure apoptosis, HeLa
cells were seeded into 96-well plates (2000 cells/well). Then
cells were incubated for 72 h at 378C with 5 % CO2before the
addition of nanoparticles. After 72 h Au1.4MS (110 mm) was ap-
plied. After 30 h of incubation, the medium was removed and
the cells were washed twice with binding buffer followed by the
addition of fluorescein isothiocyanate (FITC)-labeled aV (1 mL,
1mgmL
1; Responsif, Erlangen, Germany) per well. The cells
were incubated at room temperature on the shaker (100 rpm) for
15 min. Thereafter, PI stock solution (5 mL, 750 mm) was added
to each well and cells were further incubated for 5 min before
one final wash in binding buffer. For each experiment, untreated
cells served as a negative control and cells incubated for 5 h
with staurosporine (2 mm) served as a positive control.
HeLa cells were incubated in 96-well plates until 80 % con-
fluent. Gold nanoparticles were added and the cells were further
incubated at 378C with 5 % CO2. FACS analysis was performed
after 1, 12, 18, and 24 h. Cells were trypsinized and collected in
FACS analysis tubes. Cells were washed twice with cold PBS and
resuspended in binding buffer in a final volume of 100 mL. Then
aV-FITC (1 mL, 1 mgmL1) was added. The cells were incubated at
room temperature on the shaker (100 rpm) for 15 min in dark-
ness. After this time, PI stock
solution (4 mL, 750 mm) was
added to each tube and the
cells were further incubated for
5 min before a final binding
buffer wash. For each experi-
ment, untreated cells served as
a negative control. Cells incu-
bated with staurosporine
(2 mm) served as a positive
control; 50000 cells were
counted per sample. FACS re-
sults were analyzed by CELL-
Quest software (Becton-Dickin-
son).
TEM: HeLa cells were incu-
bated in Lab-Tek II chamber
slides until 80% confluence.
Gold nanoparticles were added
and the cells were incubated at
378C with 5 % CO2. After 1, 6,
12, 24 h, cells were washed
twice with cold PBS, fixed in
3% glutaraldehyde for 48 h,
and post-fixed in 1 % osmium
tetroxide. After fixation, speci-
mens were rinsed with 8.5%
sucrose and distilled water,
and dehydrated in a graded
series of 30, 50, 70, and 90%
ethanol and three times in 100% ethanol, for 10 min each. Sam-
ples were embedded in a mixture of EPON resin in propylene
oxide polymerized at 378C. For optical microscopy, semithin sec-
tions (1 mm) were prepared and stained with methylene blue. Ul-
trathin sections for TEM were prepared with a diamond knife,
collected on copper grids, and contrasted with uranyl acetate
and lead citrate. Samples were analyzed using an EM400T trans-
mission electron microscope (Philips).
Acknowledgements
Financial support by the German Science Foundation (DFG) is
gratefully acknowledged.
[1] A. Nel, T. Xia, L. Madler, N. Li, Science 2006,311, 622.
[2] E. E. Connor, J. Mwamuka, A. Gole, C. J. Murphy, M. D. Wyatt,
Small 2005,1, 325.
[3] C. M. Goodman, C. D. McCusker, T. Yilmaz, V. M. Rotello, Biocon-
jugate Chem. 2004,15, 897.
[4] N. Pernodet, X. Fang, Y. Sun, A. Bakhtina, A. Ramakrishnan, J.
Sokolov, A. Ulman, M. Rafailovich, Small 2006,2, 766.
[5] H. Yi, J. Leunissen, G. Shi, C. Gutekunst, S. Hersch, J. Histochem.
Cytochem. 2001,49, 279.
[6] A. G. Tkachenko, H. Xie, Y. Liu, D. Coleman, J. Ryan, W. R.
Glomm, M. K. Shipton, S. Franzen, D. L. Feldheim, Bioconjugate
Chem. 2004,15, 482.
Figure 6. Determination of live, apoptotic, and necrotic HeLa cells untreated or treated with the indicated
compounds for 6, 12, 18, and 24 h. Cells were analyzed by aV/PI double staining and flow cytometry.
The proportion of live, apoptotic, and necrotic cells was determined as detailed in Figure 5. Depending
on the material endocytosed, the HeLa cells showed no cell death (untreated, A), predominantly apopto-
sis (staurosporine, B), slow cell death with equal proportions of apoptosis and necrosis (C), or rapid cell
death with transient apoptosis and predominantly necrosis (D).
1948 www.small-journal.com 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2007,3, No. 11, 1941 – 1949
full papers U. Simon, W. Jahnen-Dechent, et al.
[7] N. L. Rosi, D. A. Giljohann, C. S. Thaxton, A. K. Lytton-Jean, M. S.
Han, C. A. Mirkin, Science 2006,312, 1027.
[8] S. L. De Wall, C. Painter, J. D. Stone, R. Bandaranayake, D. C.
Wiley, T. J. Mitchison, L. J. Stern, B. S. DeDecker, Nat. Chem.
Biol. 2006,2, 197.
[9] Y. Liu, W. Meyer-Zaika, S. Franzka, G. Schmid, M. Tsoli, H. Kuhn,
Angew. Chem. 2003,115, 2959; Angew. Chem. Int. Ed. 2003,
42, 2853.
[10] M. Tsoli, H. Kuhn, W. Brandau, H. Esche, G. Schmid, Small
2005,1, 841.
[11] B. D. Chithrani, A. A. Ghazani, W. C. Chan, Nano Lett. 2006,6,
662.
[12] R. Shukla, V. Bansal, M. Chaudhary, A. Basu, R. R. Bhonde, M.
Sastry, Langmuir 2005,21, 10644.
[13] F. Wen, U. Englert, M. Homberger, U. Simon, Z. Anorg. Allg.
Chem. 2006, 2159.
[14] G. Schmid, N. Klein, L. Korste, U. Kreibig, D. Schonauer, Poly-
hedron 1988,7, 605.
[15] G. Schmid, M. Baumle, N. Beyer, Angew. Chem. 2000,112,
187; Angew. Chem. Int. Ed. 2000,39, 181.
[16] G. Schmid, R. Pfeil, R. Boese, F. Bandermann, S. Meyer, G. H. M.
Calis, J. W. A. van der Velden, Chem. Ber. 1981,114, 3634.
[17] G. Schmid, Clusters and Colloids—From Theory to Applications,
Wiley-VCH, Weinheim, Germany, 1994.
[18] G. Schmid, R. Pugin, T. Sawitowski, U. Simon, B. Marler, Chem.
Commun. 1999, 1303.
[19] http://www.strem.com.
[20] J. Turkevitch, P. C. Stevenson, J. Hillier, Discuss. Faraday Soc.
1951,11, 55.
[21] G. Schmid, A. Lehnert, Angew. Chem. 1989,101, 773; Angew.
Chem. Int. Ed. Engl. 1989,28, 780.
[22] F. Jo, J. Kovcs, . Kath, A. Bnyei, T. Decuir, D. Darensbourg,
A. Miedaner, D. Dubois, Inorg. Synth. 1998,32,1.
[23] P. L. Bellon, F. Cariati, M. Manasser, L. Naldini, M. Sansoni J.
Chem. Soc., Chem. Commun. 1971, 1423.
[24] J. W. A. Van der Velden, J. J. Bour, W. P. Bosman, J. H. Noordik,
Inorg. Chem. 1983,22, 1913.
[25] K. P. Hall, D. M. P. Mingos, Prog. Inorg. Chem. 1984,32, 237.
Received: May 29, 2007
Revised: August 27, 2007
small 2007,3, No. 11, 1941 – 1949 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com 1949
Cytotoxicity of Gold Nanoparticles
