Gold(I) biscarbene complexes derived from vascular-disrupting combretastatin A-4 address different targets and show antimetastatic potential

Article abstract of DOI:10.1002/cmdc.201400049

Gold N-heterocyclic carbene (NHC) complexes are an emerging class of anticancer drugs. We present a series of gold(I) biscarbene complexes with NHC ligands derived from the plant metabolite combretastatin A-4 (CA-4) that retain its vascular-disrupting eff

Full text of DOI:10.1002/cmdc.201400049

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201400049
Gold(I) Biscarbene Complexes Derived from Vascular-
Disrupting Combretastatin A-4 Address Different Targets
and Show Antimetastatic Potential
Julienne K. Muenzner,[a] Bernhard Biersack,[a] Hussein Kalie,[a] Ion C. Andronache,[b]
Leonard Kaps,[c] Detlef Schuppan,[c] Florenz Sasse,[d] and Rainer Schobert*[a]
Gold N-heterocyclic carbene (NHC) complexes are an emerging
class of anticancer drugs. We present a series of gold(I) bis-
carbene complexes with NHC ligands derived from the plant
metabolite combretastatin A-4 (CA-4) that retain its vascular-
disrupting effect, yet address different cellular and protein tar-
gets. Unlike CA-4, these complexes did not interfere with tubu-
lin, but with the actin cytoskeleton of endothelial and cancer
cells. For the highly metastatic 518A2 melanoma cell line this
effect was accompanied by a marked accumulation of cells in
the G1 phase of the cell cycle and a suppression of active
prometastatic matrix metalloproteinase-2. Despite these mech-
anistic differences the complexes were as strongly antivascular
as CA-4 both in vitro in tube formation assays with human um-
bilical vein endothelial cells, and in vivo as to blood vessel dis-
ruption in the chorioallantoic membrane of chicken eggs. The
antiproliferative effect of the new gold biscarbene complexes
in a panel of six human cancer cell lines was impressive, with
low sub-micromolar IC50 values (72 h) even against CA-4-refrac-
tory HT-29 colon and multidrug-resistant MCF-7 breast carcino-
ma cells. In preliminary studies with a mouse melanoma xeno-
graft model the complexes led to significant decreases in
tumor volume while being very well tolerated.
Introduction
Tumor blood vessels, which are physiologically different from
normal vasculature, are a promising pharmacological target
that can be addressed in two different ways.[1–3] While antian-
giogenic therapies aim to impede the development of new
blood vessels by interference with angiogenic signaling path-
ways, vascular-disrupting agents (VDA) destroy existing disor-
ganized tumor blood vessels.[4] The cis-stilbene combretastatin
A-4 (CA-4, 1a) is a VDA that was first isolated from the bark of
the South African Cape Bushwillow (Combretum caffrum).[5] It
induces a rapid vascular shutdown in solid tumors, eventually
leading to secondary tumor cell death.[6] The antitumor activity
of 1a originates from two different modes of action. It binds
strongly to the colchicine binding site of tubulin, resulting in
destabilization of the microtubule cytoskeleton and inhibition
of mitosis,[7] and it interferes with the VE-cadherin/b-catenin/
Akt signaling pathway in vascular endothelial cells, which is
crucial for cell migration and adhesion.[8] Fosbretabulin (CA-4P,
1b), a phosphate prodrug of 1a with improved bioavailability,
has demonstrated its selectivity for tumor vasculature in
a number of clinical trials.[7,9] However, due to its insufficient
cytotoxicity, 1b must be administered in combination with
other anticancer drugs such as carboplatin or bevacizumab to
prevent revascularization of persistent cancer cell colonies and
a relapse of the disease.[10,11] One reason for the insufficient an-
ticancer activity of 1a is its propensity to isomerize into a bio-
logically inactive E-alkene isomer.[12,13] This can be prevented
by incorporation of the Z-alkene in heterocycles such as imida-
zoles, triazoles, or oxazoles.[7,14,15] We previously reported new,
chemically stable N-methyl-4,5-diarylimidazole analogues of 1a
that showed a synergism of vascular disruption and enhanced
cytotoxic effects.[16] They also bind tightly to tubulin, interfere
with microtubule formation, and seem to operate by a mecha-
nism akin to that of 1a with the distinction of being chemically
stable, water soluble, and selective for malignant cells, even
those resistant to 1a. Some imidazolium salts also proved effi-
cacious against chemoresistant tumor xenografts in mice.[16]
Imidazoles also happen to be a prominent ligand type in
N-heterocyclic carbene (NHC) complexes of late transition
metals which have been used thus far mainly as robust cata-
lysts for organic synthesis. Recently, imidazol-2-ylidene com-
plexes of silver and gold have been studied as potential anti-
cancer drugs with a focus on metal interactions with DNA and
apoptosis-relevant proteins.[17,18] The aspect of a suitably sub-
stituted imidazole contributing some antivascular effect to the
overall activity profile of such NHC complexes has been largely
[a] J. K. Muenzner, Dr. B. Biersack, H. Kalie, Prof. Dr. R. Schobert
Department of Organic Chemistry, University Bayreuth
Universitꢀtsstraße 30, 95440 Bayreuth (Germany)
E-mail: Rainer.Schobert@uni-bayreuth.de
[b] Dr. I. C. Andronache
Independent Researcher in Fractal Analysis and Medical Geography
Aleea Policlinicii 3, 810262 Braila (Romania)
[c] L. Kaps, Prof. Dr. D. Schuppan
Institute of Translational Immunology
University Medical Center of the Johannes Gutenberg University Mainz
Langenbeckstraße 1, 55131 Mainz (Germany)
[d] Dr. F. Sasse
Department of Chemical Biology, Helmholtz Centre for Infection Research
Inhoffenstraße 7, 38124 Braunschweig (Germany)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201400049.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1195 – 1204 1195

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
neglected. To fill this gap, we synthesized and evaluated
gold(I) carbene complexes derived from imidazoles that had
been optimized for antivascular efficacy. We already reported
on the cytotoxicity and the cellular uptake of a small series of
such new gold(I) NHC complexes.[19,20] Gold complex 2, bearing
an imidazole with the privileged combretastatin substituents,
exhibited both anticancer activity and pronounced antivascular
effects in vitro and in vivo. Surprisingly, and in contrast to 1a
and the underlying imidazolium salts, these effects by complex
2 were not correlated to tubulin binding. Herein we present
an in-depth study of the mechanism of action of a new series
of analogous cationic gold(I) biscarbene complexes 10 and 11.
Results and Discussion
Chemistry
The new gold(I) biscarbene complexes 10 and 11 were pre-
pared by reaction of the imidazolium tetrafluoroborates 8 and
9
with Ag2O to give the corresponding silver carbene
complexes which were subsequently transmetalated with
AuCl(SMe2) (Scheme 1). The salts 8 and 9 were obtained from
reaction of the iodides 6 and 7 with NaBF4, which in turn were
available by N-alkylation of the imidazoles 4 and 5 with methyl
iodide or ethyl iodide, respectively. These imidazoles were ac-
cessible by van Leusen reactions of TosMIC reagents 3[19] with
anisaldehyde and methyl- or ethylamine. We confined this
study to complexes that bear identical residues on the nitro-
gen atoms and to variation in only one meta substituent on
the A-ring, a position that had been found most crucial in pre-
vious tests with similar oxazoles and imidazoles.[16]
Scheme 1. Top box: structures of combretastatin A-4 (1a), fosbretabulin
(1b), and the known complex 2. Scheme: syntheses of complexes 10 and
11. Reagents and conditions: a) anisaldehyde, MeNH2/EtOH or EtNH2/THF,
AcOH, EtOH, reflux, 2 h, then 3, K2CO3, reflux, 3 h; b) MeI or EtI, MeCN, 808C,
16 h; c) NaBF4, acetone, RT, 24 h; d) Ag2O (1 equiv), CH2Cl2, RT, 5 h, then
AuCl(SMe2) (0.5 equiv), CH2Cl2, RT, 16 h, 82% (10a), 81% (10b), 80% (10c),
40% (11 a), 75% (11 b), 77% (11 c).
against the multidrug-resistant MCF-7 breast carcinoma cells.
The bromo compounds 10c and 11 c displayed little cell-line
specificity. All complexes were least active (IC50 >1 mm) against
resistant cervical KB-V1 cells, which overexpress P-glycoprotein
(P-gp) type ABC transporters.[22,23] Interestingly, when these
cells had been pretreated for 24 h with the P-gp substrate ve-
Growth inhibition assay
The inhibitory effect of complexes 10 and 11 on the growth of
cancer cells was evaluated with 3-(4,5-dimethylthiazol-2-yl)-2,5-
diphenyltetrazolium bromide (MTT) to identify viable cells that
reduce it to a violet formazan.[21]
The biscarbene complexes 10
and 11 showed distinct antiproli-
Table 1. Inhibitory concentrations of compounds 1a, 10, and 11 toward various cancer cell lines.
ferative effects against a panel
of six cancer cell lines of five dif-
ferent entities with IC50 values
(72 h) mostly in the low sub-
Cell line[a]
IC50 [mm][b]
1a[16]
10a
10b
10c
11a
11b
11c
518A2
Panc-1
HCT-116
HT-29
MCF-7/Topo
MCF-7/Topo
+fumit.C
KB-V1/Vbl
KB-V1/Vbl
+verapamil
0.02Æ0.01 0.46Æ0.02 0.31Æ0.02 0.37Æ0.02 0.43Æ0.04 0.43Æ0.04 0.23Æ0.06
ND[c]
ND
0.40Æ0.07 0.16Æ0.01 0.26Æ0.00 0.19Æ0.07 0.36Æ0.04 0.18Æ0.01
0.30Æ0.02 0.11Æ0.00 0.11Æ0.01 0.18Æ0.02 0.07Æ0.01 0.12Æ0.00
0.08Æ0.02 0.15Æ0.01 0.13Æ0.01 0.06Æ0.01 0.10Æ0.01 0.16Æ0.01
micromolar
range
(Table 1).
Figure 1 depicts their cell-line
specificities, that is, their IC50
values against individual cell
lines with respect to the mean
IC50 value over all cell lines of
the panel.
3.6Æ0.1
0.50Æ0.20 0.18Æ0.04 0.06Æ0.00 0.15Æ0.06 0.08Æ0.02 0.06Æ0.01 0.10Æ0.02
ND
0.19Æ0.01 0.09Æ0.01 0.12Æ0.00 0.13Æ0.01 0.08Æ0.00 0.10Æ0.01
<0.01
7.7Æ0.3
2.9Æ0.3
2.0Æ0.3
3.9Æ0.5
3.7Æ0.2
1.1Æ0.1
1.5Æ0.0
ND
0.23Æ0.03 0.45Æ0.06 1.3Æ0.1
0.22Æ0.02 0.20Æ0.05
The
all-methoxy-substituted
[b] Human cancer cell lines: 518A2 melanoma, Panc-1 pancreatic ductular adenocarcinoma, HCT-116 and HT-29
colon carcinomas, MCF-7/Topo breast adenocarcinoma (optionally pretreated with 1.2 mm fumitremorgin C for
24 h), KB-V1/Vbl cervix carcinoma (optionally pretreated with 24 mm verapamil·HCl for 24 h). [b] Values are the
means ÆSD of four independent experiments and are derived from dose–response curves obtained by meas-
uring the percentage of viable cells relative to untreated controls after 72 h incubation using the MTT assay.
[c] ND: not determined.
complexes 10a and 11 a showed
their highest activity in HT-29
colon carcinoma cells, whereas
the chloro derivatives 10b and
11 b were most efficacious
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1195 – 1204 1196

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
out concomitant destruction of
the microtubule cytoskeleton.[28]
The untreated control cells of
either type showed the typical
cortical microfilaments with only
few fine filaments traversing the
cells as well as a greater number
of intercellular connections.[28,29]
Cell-cycle analysis
The antimitotic effect of 1a is
based on microtubule disruption
and stress fiber formation lead-
ing to G2/M-phase cell-cycle
arrest.[30,31] Because the new
gold complexes mediated the
formation of stress fibers with-
Figure 1. Cell line specificities of gold biscarbene complexes 10a–c and 11 a–c as deviation of the log(IC50/72 h)
of individual cell lines from the mean over all cell line log(IC50/72 h) values. Negative values indicate higher, posi-
tive values lower than average activities. Mean log(IC50/72 h): À6.31 (10a), À6.71 (10b), À6.55 (10c), À6.54 (11 a),
À6.74 (11 b), À6.70 (11 c).
rapamil,[24,25] the activity of the halo-substituted complexes
10b,c and 11 b,c rose five- to tenfold. In contrast, 10a and 11 a
profited far less from the addition of the competitor substrate
verapamil. Apparently, their methoxy-only substituted ligands
render them better substrates for the P-gp efflux pumps. Cells
of the human breast carcinoma MCF-7/Topo, which overex-
press ABC transporters of the breast cancer resistance protein
(BCRP) type, responded fairly well to all complexes 10 and 11.
Pretreatment of these cells with fumitremorgin C, a selective
BCRP inhibitor,[26] did not alter the IC50 values of the complexes
much, which indicates that they are not substrates of this par-
ticular efflux pump. This fact is important because BCRP is also
thought to be responsible for the drug resistance of certain
cancer stem-like cells.[27]
out affecting microtubules, we investigated whether they also
alter cell-cycle progression in 518A2 melanoma cells (Figure 3).
518A2 cells accumulated in the G1 phase when exposed to
compounds 10 or 11 for 24 h, while the sub-G1 population of
apoptotic cells was barely increased, which means there is
a clear correlation between stress fiber formation and the G1
phase arrest.
Wound healing assay
The effects of complexes 10 and 11 on the motility of cancer
cells were analyzed by introducing an artificial “scratch wound”
to monolayers of 518A2 melanoma cells and monitoring the
closure of this gap over time. While control cells overgrew the
‘wound’ within 48 h (100% of wound healing), it took much
longer for 518A2 cells treated with complexes 10 or 11
(Figure 4; for exemplary light microscopy images see Support-
ing Information, Figure S3). The N-ethyl complexes 11 were
generally more effective than their corresponding N-methyl an-
alogues 10, inhibiting the wound healing to the same extent
as 1a (IC50 =20 nm)[16] after 48 h incubation with equitoxic con-
centrations. Because a functional, dynamic actin cytoskeleton
is essential for cell migration, the antimigratory activity of gold
complexes 10 and 11 is likely to be a consequence of their in-
duction of stress fibers[32,33] and of G1 phase arrest of the
518A2 cell cycle.
Effects on cytoskeletal components
We previously showed that in contrast to 1a and its imidazoli-
um salt derivatives, the gold carbene complex 2 does not in-
terfere with the polymerization of purified tubulin in vitro.[19]
We now subjected the corresponding biscarbene complex 10a
to the same fluorescence-based tubulin polymerization assay
and found that it also does not affect the polymerization of
purified tubulin (Supporting Information, Figure S1). Next, we
looked for effects of compounds 10 and 11 on the cytoskeletal
organization of microtubules and microfilaments (F-actin) in
518A2 melanoma cells and in non-malignant human umbilical
vein endothelial cells (HUVEC). As anticipated, the distribution
of microtubules in both types of cells treated with 10 or 11 for
24 h did not appear different from untreated controls when vi-
sualized by immunofluorescence (Supporting Information, Fig-
ure S2). In contrast, exposure to 1a led to a massive destruc-
tion of the microtubule cytoskeleton in both cell lines (Sup-
porting Information, Figure S2). Exposure of both cell lines to
10 and 11 for 24 h triggered a complete reorganization of the
actin cytoskeleton with formation of stress fibers stretched
across the whole cell body, which is a typical cell response to
chemical or physical stress (Figure 2). In contrast to 1a, the
gold complexes 10 and 11 induced stress fiber formation with-
Effects on matrix metalloproteinases
Tumor angiogenesis and metastasis require degradation of the
basement membrane so that endothelial or tumor cells may
migrate through the extracellular matrix (ECM) and egress
from or invade tissues.[34] This proteolytic degradation of ECM
components is mediated by matrix metalloproteinases (MMPs),
a class of zinc-dependent endopeptidases that are expressed
and secreted at elevated concentrations by inflammatory, fi-
broblastic, and especially vascular endothelial cells and invasive
tumor cells.[35] Because of their essential role in tumor progres-
sion, MMPs were earmarked as promising pharmacological an-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1195 – 1204 1197

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Figure 3. Effects of 1a and complexes 10 and 11 on cell-cycle progression
of 518A2 melanoma cells. Percentages are shown of cells in G1, S, or G2
phases as well as apoptotic cells (sub-G1) as obtained by flow cytometry
after DNA staining with propidium iodide. Cells were treated with 5 nm 1a
and 500 nm each of 10 or 11. Values are the means ÆSD of three independ-
ent experiments.
Figure 4. Inhibitory effects of 1a and gold complexes 10 and 11 on the mi-
gration of 518A2 melanoma cells. Cells were treated with 5 nm 1a or
500 nm 10 or 11. Values are the means ÆSD of three independent experi-
ments.
Figure 2. Effects on the cellular organization of microfilaments. The actin cy-
toskeleton of both cell types was visualized by labeling with an Alexa Fluor
488 phalloidin conjugate. Upper group of four images: 518A2 melanoma
cells treated with vehicle (C, control), 2.5 nm 1a, 250 nm 10a or 500 nm
11 a. Lower group of four images: HUVEC treated with vehicle (C, control),
100 nm each of 1a or 10a and 1 mm 11 a. Scale bars: 50 mm; images are rep-
resentative of two independent experiments.
and depressed the levels of functional MMP-2 significantly,
even at concentrations as low as 100 nm. To clarify whether
complexes 10 and 11 actually do interfere with the expression
or secretion of MMP-2 or rather inhibit metalloproteinase activ-
ity, we subjected pooled zymography protein samples from ex-
periments with or without the most active complex 11 b to
western blot analysis (Supporting Information, Figure S4). Be-
cause the levels of cell-associated and secreted MMP-2 protein
were decreased by complex 11 b, it must have interfered with
the expression or secretion of MMP-2 rather than by acting as
a mere inhibitor of this particular matrix metalloproteinase.
ticancer targets. Notably, activity of the basement membrane
collagen degrading MMP-2 and MMP-9 can be analyzed by zy-
mography.[36] The monocarbene complex 2 was previously
shown by us to only slightly affect the activities of MMP-2 and
-9 in HUVEC, while a related gold(I) N,N-dibenzylimidazol-2-yl
carbene complex significantly decreased the cell-associated
and secreted levels of active MMP-2.[20] In this study we exam-
ined the effects of the biscarbene complexes 10 and 11 on the
intra- and extracellular levels of active MMP-2 and -9 produced
by 518A2 melanoma cells, via gelatin zymography (Figure 5).
The test compounds decreased the levels of active MMP-2,
while the fractions of MMP-9 were left virtually unaltered.
Moreover, the levels of active cell-associated MMP-2 (in cell ly-
sates) were generally decreased to a greater extent than se-
creted MMP-2. Complex 11 b was the most active compound
In vitro and in vivo antivascular activity
The inhibitory effects of the new gold complexes on the ability
of HUVEC to form tubular networks in vitro when seeded on
thin layers of matrigel can be used as a predictor for their anti-
vascular effects in vivo.[37] Hence, the in vitro antivascular activ-
ity of the all-methoxy-substituted gold complexes 10a and
11 a was exemplarily investigated by tube-formation assay
(Supporting information, Figure S5).[38] Control HUVEC devel-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1195 – 1204 1198

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Figure 5. Effects of complexes 10 and 11 at 100 nm to 2.5 mm on the cell-associated and secreted activities of MMP-2 and -9 in 518A2 melanoma cells after
24 h exposure. Conditioned media and cell lysates were subjected to gelatin zymography. C: control (DMF); shown are negative images of the gelatin gels.
oped networks of interconnect-
ed tubes linked by well-defined
branch points after 24 h growth
on matrigel. Exposure of HUVEC
to 10a or 11 a for 24 h com-
pletely prevented the formation
of organized tubular networks,
so that only few sprouting
points and fragmented tubules
could be detected. Complex 11 a
was found to be nearly as effec-
tive as the natural lead com-
pound 1a.
We then used the chorioallan-
toic membrane (CAM) assay as
a model system to identify vas-
cular-disruptive activities of the
complexes in vivo.[39,40] Figure 6
shows such effects as detected
for 1a and the most active com-
plexes, 10b and 11 a. Such vas-
cular-disrupting effects can be
quantified by calculating the de-
Figure 6. Effects of 1a (10 nmol in 10 mL H2O) and the most active complexes 10b and 11 a (each 2.5 nmol in
10 mL H2O) on vascularization of the chorioallantoic membrane of fertilized chicken eggs after 6 and 24 h. C: con-
trol (DMF); images are representative of three independent assays.
crease in the area covered by
blood vessels in CAM micro-
graphs using vessel area analysis
that is derived from fractal analy-
sis.[4142] Figure 7 shows a comparison of all complexes 10 and
11. On average, the chloro-substituted complexes 10b and
11 b were most strongly vascular-disruptive after 6 h exposure.
The all-methoxy-substituted complex 10a was slightly less ef-
fective, while the bromo derivatives 10c and 11 c exhibited the
weakest antivascular activity. The natural lead 1a also showed
strong disruptive effects, yet only when applied at higher con-
centrations than the gold complexes. Unlike in eggs treated
with 1a, the vasculature exposed to complexes 10 and 11 was
able to partially recover and sprout new blood vessels within
24 h. Interestingly, monocarbene complex 2, while leaving
tube formation by HUVEC and migration of 518A2 melanoma
cells virtually unaffected, led to a distinct and irreversible
blood vessel disruption in the CAM assay.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1195 – 1204 1199

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
of 200000 cells the mice were repeatedly administered com-
plex 11 a or 11 c at 15 mg(kg body weight)À1 on two consecu-
tive days. This treatment was tolerated well by the mice,
merely resulting in a slight and reversible decrease in body
weight, but noticeable tumor regression, as depicted in
Figure 8. For both test compounds slight tumor regrowth
could be detected a few days after the first dual-dose applica-
tion. However, the second dual dose of 11 a or 11 c decreased
the tumor volumes further. These preliminary results are indi-
cative of considerable in vivo antitumor activity against this
type of tumor and of excellent tolerance of this compound
class.
Figure 7. Vessel area analysis (derived from fractal analysis) after 6 h incuba-
tion with complexes 10 and 11 (each 2.5 nmol in 10 mL H2O). The area in the
image sections covered by blood vessels before treatment with the com-
plexes (0 h) was set at 100%. Values are the means ÆSD of three vessel area
analysis calculations.
Conclusions
The new gold biscarbene complexes 10 and 11 are strongly cy-
totoxic with low sub-micromolar IC50 values against various
cancer cell lines including multidrug-resistant lines and those
refractory to the lead compound CA-4 (1a). The N-ethyl com-
plexes 11 are generally more active than the N-methyl conge-
ners 10. Notably, Gust and co-workers[43] reported comparably
high cytotoxicities against three cancer lines for analogous
N-ethyl biscarbene gold complexes bearing four indentical
para-methoxy, -hydroxy, or -fluoro substituted phenyl rings.
Mouse xenograft model
Finally, we evaluated the in vivo antitumor activity and toler-
ance of gold complexes 11 a and 11 c in a Balb/c mouse xeno-
graft model of highly metastatic B16-F10 mouse melanoma
cells (Figure 8). Fourteen days after subcutaneous implantation
Figure 8. In vivo effects of 11 a and 11 c in mouse xenografts of the highly metastatic B16-F10 melanoma cell line. Shown are the tumor responses following
single (control) and repeated (complexes 11 a and 11 c) dual-dose applications. Arrows indicate the administration of test compounds or vehicle control
(DMSO).
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1195 – 1204 1200

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
of imidazolium salt 8a (22 mg, 0.048 mmol) in CH2Cl2/MeOH (1:1,
30 mL) was treated with Ag2O (13 mg, 0.056 mmol) and stirred in
the dark at room temperature for 5 h. AuCl(Me2S) (9 mg,
0.03 mmol) was added, and the reaction mixture was stirred for an-
other 20 h. The suspension was filtered, the filtrate was concentrat-
ed in vacuum, and the residue was re-dissolved in CH2Cl2 and fil-
tered over MgSO4/Celite. The filtrate was concentrated in vacuum,
and the residue dried in vacuum. Yield: 20 mg (82%); colorless
solid; mp: 1088C; 1H NMR (300 MHz, CDCl3): d=3.72 (s, 12H, 4ꢁ
Me), 3.79 (s, 12H, 4ꢁMe), 3.82 (s, 6H, 2ꢁMe), 3.88 (s, 6H, 2ꢁMe),
6.40 (s, 4H, 2ꢁC6H2(OMe)3), 6.88 (d, J=8.8 Hz, 4H, 2ꢁ
While unable to pinpoint their mode of action, they could ex-
clude thioredoxin reductase, estrogen receptor and cyclooxy-
genase (COX) as potential targets. In the case of complexes 10
and 11, the residue R1 seems to be decisive for cell line specif-
icity. The all-methoxy substituted complexes 10a and 11 a are
most active against HT-29 colon carcinoma cells, which are tu-
morigenic and when implanted in mice form slowly growing,
poorly vascularized tumors. Therefore, in this case, the greater
cytotoxicity compensates for the less pronounced antivascular
effect of the drug. All complexes 10 and 11 exhibit distinct vas-
cular-disrupting effects in vitro and in vivo, resembling those
effected by lead 1a in magnitude and morphology, yet origi-
nating from a different mode of action. Whereas 1a targets
the microtubules of cancer and endothelial cells causing a
G2/M cell-cycle arrest, complexes 10 and 11 interfere only with
the F-actin dynamics by stress fiber formation, leading to G1
phase arrest. As a consequence, cell motility is severely im-
paired. This contributes to the overall antimetastatic effect of
10 and 11 which is further enhanced by the decrease in the
concentration of prometastatic MMP-2 expressed and secreted
by cancer cells. A similar effect has not been observed for the
lead compound 1a.
Apart from their unique and multimodal mechanism of
action, two more clinically relevant aspects render 10 and 11
superior to 1a and thus as promising candidates for further in-
vestigation and development: first, their strong impact on
mouse xenografts of a highly metastatic melanoma cell line
and the possibility to administer the compounds repeatedly
and at high doses, as they are very well tolerated by the ani-
mals; second, the fact that the chorioallantoic membrane of
hen eggs, after eradication of vasculature by complexes 10
and 11, was able to recover and sprout new blood vessels
might provide a tool for the new concept of “normalization of
tumor vasculature”,[44] that is, the replacement of irregular
tumor blood vessels with normal vessels, allowing a better per-
fusion of the tumor with conventional anticancer drugs. At this
point, we also wish to identify similarities in the modes of
action between our new gold carbene complexes and NAMI-
A,[45] the most advanced antimetastatic metallodrug candidate
that also features an ionic imidazole transition metal complex.
C6Hmeta2Hortho2(OMe)), 7.15 ppm (d, J=8.8 Hz, 4H, 2ꢁC6Hmeta2Hortho
-
2
(OMe)); 13C NMR (75.5 MHz, CDCl3): d=36.4 (NMe), 36.7 (NMe), 55.3
(OMe), 56.2 (OMe), 60.9 (OMe), 107.7, 114.4, 119.4, 122.7, 132.2,
138.7, 153.4, 160.3 ppm (NCCN), 183.6 (AuC); 11B NMR (96.3 MHz,
CDCl3): d=À3.45 ppm; IR (ATR): n˜max =2935, 2836, 1607, 1580,
1517, 1503, 1458, 1410, 1357, 1315, 1292, 1247, 1178, 1123, 1052,
1004, 1021, 909, 837, 818, 789, 765, 732, 670 cmÀ1; MS (EI, 70 eV):
m/z (%): 565 (4), 368 (100), 353 (92), 338 (54), 298 (96), 283 (84),
221 (88), 149 (24); HRMS (ESI): m/z (%): calcd: 935.3205, 934.3171,
933.3138; found: 935.3234 (16), 934.3167 (58), 933.3072 (100)
[C42H48AuN4O8+]; Anal. calcd for C42H48AuBF4N4O8: C 49.43, H 4.74,
N 5.49, found: C 49.58, H 4.69, N 5.39.
Bis[4-(4-methoxyphenyl)-5-(3-chloro-4,5-dimethoxyphenyl)-1,3-
dimethylimidazol-2-ylidene]gold(I) tetrafluoroborate (10b): Anal-
ogously to the synthesis of 10a, complex 10b (42 mg, 81%) was
obtained from 8b (46 mg, 0.101 mmol), Ag2O (27 mg, 0.116 mmol),
and AuCl(Me2S) (19 mg, 0.063 mmol) as an off-white solid; mp:
1308C; 1H NMR (300 MHz, CDCl3): d=3.77 (s, 6H, 2ꢁMe), 3.82 (s,
6H, 2ꢁMe), 3.84 (s, 6H, 2ꢁMe), 3.89 (s, 12H, 4ꢁMe), 6.71 (s, 2H,
2ꢁCHCCl), 6.84 (s, 2H, CHCCCCl), 6.94 (d, J=8.8 Hz, 4H, 2ꢁ
C6Hmeta2Hortho2(OMe)), 7.18 ppm (d, J=8.8 Hz, 4H, 2ꢁC6Hmeta2Hortho
-
2
(OMe)); 13C NMR (75.5 MHz, CDCl3): d=36.4 (NMe), 36.6 (NMe), 55.3
(OMe), 56.2 (OMe), 60.7 (OMe), 113.4, 114.5, 118.9, 123.5, 123.7,
128.4, 130.8, 131.7, 132.5, 146.1, 153.9, 160.4 NCCN), 183.8 ppm
(AuC); 11B NMR (96.3 MHz, CDCl3) d=À1.70 ppm; IR (ATR): n˜max
=
2952, 2031, 1561, 1515, 1492, 1455, 1404, 1293, 1251, 1177, 1032,
915, 834, 758, 710 cmÀ1; MS (EI, 70 eV): m/z (%): 568 (3), 535 (3),
421 (9), 372 (100), 302 (45); HRMS (ESI): m/z (%): calcd: 941.2154,
943.2130, 944.2158; found: 941.2202 (100) [C40H42AuCl2N4O6+],
943.2221 (69), 944.2181 (35).
Bis[4-(4-methoxyphenyl)-5-(3-bromo-4,5-dimethoxyphenyl)-1,3-
dimethylimidazol-2-ylidene]gold(I) tetrafluoroborate (10c): Anal-
ogously to the synthesis of 10a, complex 10c (30 mg, 80%) was
obtained from 8c (34 mg, 0.067 mmol), Ag2O (18 mg, 0.077 mmol),
and AuCl(Me2S) (12 mg, 0.041 mmol) as an off-white solid; mp:
1328C; 1H NMR (300 MHz, CDCl3): d=3.77 (s, 6H, 2ꢁMe), 3.83 (s,
6H, 2ꢁMe), 3.84 (s, 6H, 2ꢁMe), 3.89 (s, 6H, 2ꢁMe), 3.90 (s, 6H, 2ꢁ
Me), 6.76 (s, 2H, 2ꢁCHCCCBr), 6.94 (d, J=8.8 Hz, 4H, 2ꢁ
C6Hmeta2Hortho2(OMe)), 7.00 (s, 2H, 2ꢁCHCBr), 7.19 ppm (d, J=8.8 Hz,
4H, 2ꢁC6Hmeta2Hortho2(OMe)); 13C NMR (75.5 MHz, CDCl3): d=36.4
(NMe), 36.6 (NMe), 55.3 (OMe), 56.2 (OMe), 60.6 (OMe), 114.1, 114.5,
117.7, 119.0, 124.3, 126.2, 130.6, 131.7, 132.5, 153.7, 160.4 (NCCN),
183.8 ppm (AuC); 11B NMR (96.3 MHz, CDCl3) d=À1.29 ppm; IR
(ATR): n˜max =2924, 2853, 1606, 1589, 1554, 1515, 1489, 1461, 1427,
1404, 1292, 1250, 1177, 1020, 914, 830, 808, 788, 754, 737, 702,
665 cmÀ1; MS (EI, 70 eV): m/z (%): 613 (2), 468 (10), 419 (100), 348
(85), 331 (40), 148 (100); HRMS (ESI): m/z (%): calcd: 1029.1143,
1031.1127, 1033.1106; found: 1029.1241 (52), 1031.1227 (100)
[C40H42AuBr2N4O6+], 1033.1233 (64).
Experimental Section
Column chromatography: silica gel 60 (230–400 mesh). Melting
points (uncorrected), Electrothermal 9100; IR spectra, PerkinElmer
Spectrum One FT-IR spectrophotometer with ATR sampling unit;
NMR spectra, Bruker Avance 300 spectrometer, chemical shifts (d)
are given in ppm downfield from tetramethylsilane as internal
standard; Mass spectra, Thermo Finnigan MAT 8500 (EI), Waters
UPLC-Q-TOF (ESI, for HRMS). All tested compounds were >98%
pure, as determined by high-resolution mass spectrometry and
HPLC analysis. All starting compounds were purchased from the
usual retailers and used without further purification.
Chemistry
Bis[4-(4-methoxyphenyl)-5-(3,4,5-trimethoxyphenyl)-1,3-dime-
Bis[4-(4-methoxyphenyl)-5-(3,4,5-trimethoxyphenyl)-1,3-diethyl
thylimidazol-2-ylidene]gold(I) tetrafluoroborate (10a): A solution
imidazol-2-ylidene]gold(I) tetrafluoroborate (11a): Analogously
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1195 – 1204 1201

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
to the synthesis of 10a, complex 11 a (100 mg, 40%) was obtained
from 9a (110 mg, 0.23 mmol), Ag2O (61 mg, 0.26 mmol), and AuCl-
(Me2S) (42 mg, 0.14 mmol) as an off-white solid; mp: 102–1058C;
1H NMR (300 MHz, CDCl3): d=1.37 (t, J=7.2 Hz, 6H, 2ꢁMe), 1.43 (t,
J=7.2 Hz, 6H, 2ꢁMe), 3.72 (s, 12H, 4ꢁOMe), 3.77 (s, 6H, 2ꢁOMe),
3.81 (s, 6H, 2ꢁOMe), 4.17 (q, J=7.2 Hz, 4H, 2ꢁCH2), 4.27 (q, J=
7.2 Hz, 4H, 2ꢁCH2), 6.40 (s, 4H, 2ꢁC6H2(OMe)3), 6.88 (d, J=8.7 Hz,
4H, 2ꢁC6Hmeta2Hortho2(OMe)), 7.16 ppm (d, J=8.7 Hz, 4H, 2ꢁ
C6Hmeta2Hortho2(OMe)); 13C NMR (75.5 MHz, CDCl3): d=17.3 (Me), 17.6
(Me), 44.1 (CH2), 44.3 (CH2), 55.3 (OMe), 56.2 (OMe), 60.8 (OMe),
107.7, 114.4, 119.3, 122.7, 131.5, 131.6, 131.8, 138.7, 153.3, 160.3
(NCCN), 182.2 ppm (AuC); 11B NMR (96.3 MHz, CDCl3) d=
À0.92 ppm; IR (ATR): n˜max =2933, 2837, 1607, 1580, 1514, 1503,
1460, 1412, 1349, 1330, 1291, 1239, 1177, 1123, 1050, 1024, 1006,
886, 838, 810, 745, 731, 667 cmÀ1; HRMS (ESI): m/z (%): calcd:
991.3831, 990.3797, 989.3764; found: 991.3827 (20), 990.3737 (56),
989.3652 (100) [C46H56AuN4O8+]; Anal. calcd for C46H56AuBF4N4O8:
C 51.31, H 5.24, N 5.20, found: C 51.21, H 5.12, N 5.07.
Biology
Cell lines and culture conditions: The human melanoma cell line
518A2 was obtained from the Department of Radiotherapy, Medi-
cal University of Vienna (Austria). The KB-V1/Vbl and MCF-7/Topo
cell lines were obtained from the Institute of Pharmacy at the Uni-
versity Regensburg (Germany), and the HCT-116 colon and Panc-
1 pancreatic carcinoma cell lines from the University Hospital Erlan-
gen (Germany). The human HT-29 colon carcinoma cell line (ACC-
581) was purchased from the German Centre of Biological Materi-
als (DSMZ), Braunschweig, Germany. Human umbilical vein endo-
thelial cells (HUVEC) were provided by the Helmholtz Center for In-
fection Research, Braunschweig. All cell lines were grown at 378C,
5% CO2, 95% humidity in Dulbecco’s modified Eagle’s medium
(DMEM) containing 10% fetal bovine serum (FBS), 1% antibiotic–
antimycotic, and 250 mgmLÀ1 gentamycin (all from Gibco), apart
from HT-29 cells, which were grown in RPMI-1640 medium (10%
FBS, 1% antibiotic–antimycotic, 250 mgmLÀ1 gentamycin) and
HUVEC, which were cultured in EGM-2 medium (Lonza) supple-
mented with 5% FBS. Only mycoplasma-free cultures were used.
Bis[4-(4-methoxyphenyl)-5-(3-chloro-4,5-dimethoxyphenyl)-1,3-
diethylimidazol-2-ylidene]gold(I) tetrafluoroborate (11b): Analo-
gously to the synthesis of 10a, complex 11 b (25 mg, 75%) was ob-
tained from 9b (30 mg, 0.061 mmol), Ag2O (16 mg, 0.071 mmol),
and AuCl(Me2S) (12 mg, 0.04 mmol) as an off-white solid; mp:
948C; 1H NMR (300 MHz, CDCl3): d=1.31–1.47 (m, 12H, 4ꢁMe),
3.73–3.79 (m, 6H, 2ꢁOMe), 3.80–3.84 (m, 6H, 2ꢁOMe), 3.85–3.89
(m, 6H, 2ꢁOMe), 4.13–4.37 (m, 8H, 4ꢁCH2), 6.63–6.77 (m, 2H, 2ꢁ
CHCCCCl), 6.81–6.87 (m, 2H, 2ꢁCHCCl), 6.88–6.99 (m, 4H, 2ꢁ
Determination of tumor cell growth: 3-(4,5-Dimethylthiazol-2-yl)-2,5-
diphenyltetrazolium bromide (MTT; ABCR) was used to identify
viable cells that reduce it to a violet formazan.[21] See the Support-
ing Information for a detailed description.
Fluorescence labeling of actin filaments: Effects of 1a and com-
plexes 10 and 11 on the organization of F-actin in HUVEC and
518A2 melanoma cells were examined by fluorescence microscopy.
Cells were cultured on glass coverslips to 75% confluence. For
HUVEC the glass coverslips were pretreated with 1m HCl/EtOH.
Following treatment with 1a, 10, or 11 (for HUVEC: 100 nm 1a;
100, 500 nm and 1 mm 10 or 11; for 518A2 cells: 2.5 nm 1a; 250
and 500 nm of 10 or 11) or vehicle for 24 h, cells were fixed with
3.5% formalin (in PBS) for 10 min, washed with PBS and permeabi-
lized with 0.1% Triton X-100 (5 min). Then cells were washed and
F-actin was stained (378C, 1 h) using Alexa Fluor 488 Phalloidin (In-
vitrogen) 1:100. Nuclei were counterstained with DAPI (1 mgmLÀ1
in PBS, room temperature, 5 min) before coverslips were washed in
PBS and mounted with ProLong Gold Antifade reagent (Invitro-
gen). Effects were analyzed using an Axioplan fluorescence micro-
scope with a 40ꢁ objective lens (Zeiss, AxioCam MRm).
C6Hmeta2Hortho2(OMe)), 7.13–7.22 ppm (m, 4H, 2ꢁC6Hmeta2Hortho
-
2
(OMe)); 13C NMR (75.5 MHz, CDCl3): d=16.8 (Me), 17.0 (Me), 17.3
(Me), 17.5 (Me), 44.2 (CH2), 44.4 (CH2), 55.3 (OMe), 56.3 (OMe), 60.7
(OMe), 113.5, 114.4, 114.5, 118.9, 123.6, 123.7, 124.6, 128.4, 130.3,
131.7, 131.9, 132.0, 146.0, 146.2, 153.8, 153.9, 160.3, 160.5 (NCCN),
182.5 ppm (AuC); 11B NMR (96.3 MHz, CDCl3) d=À1.28 ppm; IR
(ATR): n˜max =2933, 1606, 1562, 1514, 1491, 1461, 1409, 1345, 1319,
1292, 1250, 1176, 1158, 1110, 1090 1040, 1032, 997, 899, 835, 811,
756 cmÀ1; MS (EI, 70 eV): m/z (%): 597 (3), 416 (11), 372 (100), 357
(75); HRMS (ESI): m/z (%): calcd: 997.2773, 999.2772, 1000.2793;
found: 997.2779 (100) [C44H50AuCl2N4O6+], 999.2802 (77), 1000.2938
(35).
Bis[4-(4-methoxyphenyl)-5-(3-bromo-4,5-dimethoxyphenyl)-1,3-
diethylimidazol-2-ylidene]gold(I) tetrafluoroborate (11c): Analo-
gously to the synthesis of 10a, complex 11 c (35 mg, 77%) was ob-
tained from 9c (41 mg, 0.077 mmol), Ag2O (21 mg, 0.089 mmol),
and AuCl(Me2S) (14 mg, 0.048 mmol) as an off-white solid; mp:
Cell-cycle analysis: 518A2 cells (5ꢁ104 mL) were grown on six-well
plates for 24 h and then treated with 5 nm 1a and 500 of 10 or 11
for 24 h. Solvent controls (DMSO or DMF) were treated identically.
After fixation (70% EtOH, 48C) the cells were incubated with propi-
dium iodide (PI; Roth) staining solution (50 mgmLÀ1 PI, 0.1%
sodium citrate, 50 mgmLÀ1 RNase A in PBS) for 30 min at 378C. The
1
1048C; H NMR (300 MHz, CDCl3): d=1.43 (m, 12H, 4ꢁMe), 3.79 (s,
6H, 2ꢁOMe), 3.83–3.90 (m, 12H, 4ꢁOMe), 4.18–4.34 (m, 8H, 4ꢁ
CH2), 6.79 (s, 2H, 2ꢁCHCCCBr), 6.95 (d, J=8.9 Hz, 4H, 2ꢁ
C6Hmeta2Hortho2(OMe)), 7.03 (s, 2H, 2ꢁCHCBr), 7.21 ppm (d, J=8.9 Hz,
4H, 2ꢁC6Hmeta2Hortho2(OMe)); 13C NMR (75.5 MHz, CDCl3): d=16.8
(Me), 17.1 (Me), 17.3 (Me), 44.2 (CH2), 44.3 (CH2), 44.4 (CH2), 55.3
(OMe), 56.1 (OMe), 56.3 (OMe), 60.5 (OMe), 114.2, 114.5, 117.6,
119.0, 124.4, 126.3, 130.1, 130.2, 131.7, 131.9, 131.9, 153.8, 160.4
(NCCN), 182.5 ppm (AuC); 11B NMR (96.3 MHz, CDCl3) d=
À1.67 ppm; IR (ATR): n˜max =3435, 2981, 2937, 2192, 1606, 1588,
1554, 1515, 1489, 1462, 1418, 1398, 1351, 1293, 1251, 1176, 1150,
1076, 1026, 995, 918, 894, 840, 806, 776, 726 cmÀ1; MS (EI, 70 eV):
m/z (%): 719 (9), 642 (15), 563 (6), 458 (60); HRMS (ESI): m/z (%):
calcd: 1085.1773, 1087.1765, 1089.1730; found: 1085.1842 (55),
1087.2778 (100) [C44H50AuCl2N4O6+], 1089.1682 (70).
fluorescence intensity of 10000 single cells was measured at lem
=
620 nm (lex =488 nm laser source) using a Beckman Coulter Cy-
tomics FC 500 flow cytometer and analyzed (CXP Analysis, Beck-
man Coulter) for the fractions of cells in G1, S, and G2/M phase. The
percentage of apoptotic cells was assessed from sub-G1 peaks.
Wound-healing assay: 518A2 melanoma cells (1ꢁ105 mL) were
seeded on 24-well plates and grown to a sub-confluent monolayer.
A narrow artificial wound was created by scraping off a strip of
cells with a 20–200 mL plastic tip. The medium was replaced before
cells were treated with 5 nm 1a, 500 nm 10 or 11, or vehicle (DMF)
for up to 48 h. The wound-healing process was monitored micro-
scopically (Axiovert 135 with a 10ꢁ objective lens, Zeiss, AxioCam
MRc5) after 24 and 48 h exposure to test compounds. Wound size
was measured at three different positions (top, middle, bottom) of
each microscopy image using Adobe Photoshop CS6 (version
13.01). Thereby, the mean width of the wound was calculated for
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1195 – 1204 1202

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
each documented time point (0, 24, and 48 h), and the percentage
of wound healing over time was determined.
schaft (grant Scho 402/8-3) for financial support, Dr. Thomas Mu-
eller (Department of Internal Medicine IV, Oncology/Hematology,
Martin Luther University, Halle-Wittenberg) for tubulin polymeri-
zation assays, Dr. Holger Schmidt (Department of Plant Physiolo-
gy, University Bayreuth) for high-resolution mass spectra, and the
Elite Study Program “Macromolecular Science”, Elite Network of
Bavaria, for its support (J.K.M.).
Gelatin zymography: 518A2 cells (1ꢁ105 mL) were seeded on six-
well plates and cultured for 24 h. After replacement of the medium
with fresh DMEM supplemented with 0.1% BSA (Roth) and
200 (kIU aprotinin)mLÀ1 (AppliChem), complexes 10 and 11 were
added (dilution series of 10 mm stock solutions in DMF ranging
from 0.1 to 2.5 mm in H2O), and incubation was continued for a fur-
ther 24 h. For solvent controls respective amounts of DMF were
applied. Both the preparation of conditioned media and cell lysates
as well as the gelatin zymography itself were carried out according
to published methods.[46] Samples used for gelatin zymography
were normalized to protein concentration (for each sample 20 mg
total protein were applied to the gels).
Keywords: antimetastatic agents · carbene ligands · gold ·
matrix metalloproteinases · vascular-disrupting agents
[1] G. Taraboletti, B. Margosio, Curr. Opin. Pharmacol. 2006, 1, 378–384.
[2] J. W. Lippert III, Bioorg. Med. Chem. 2007, 15, 605–615.
[3] S. M. Weis, D. A. Cheresh, Nat. Med. 2011, 17, 1359–1370.
[4] D. W. Siemann, Cancer Treat. Rev. 2011, 37, 63–74.
[5] G. R. Pettit, S. B. Singh, E. Hamel, C. M. Lin, D. S. Alberts, D. Garcia-Ken-
dall, Experientia 1989, 45, 209–211.
[6] G. M. Tozer, V. E. Prise, J. Wilson, M. Cemazar, S. Shan, M. W. Dewhirst,
P. R. Barber, B. Vojnovic, D. J. Chaplin, Cancer Res. 2001, 61, 6413–6422.
[7] G. C. Tron, T. Pirali, G. Sorba, F. Pagliai, S. Busacca, A. A. Genazzani, J.
Med. Chem. 2006, 49, 3033–3044.
[8] L. Vincent, P. Kermani, L. M. Young, J. Cheng, F. Zhang, K. Shido, G. Lam,
H. Bompais-Vincent, Z. Zhu, D. J. Hicklin, P. Bohlen, D. J. Chaplin, C. May,
S. Rafii, J. Clin. Invest. 2005, 115, 2992–3006.
[9] C. J. Mooney, G. Nagaiah, P. Fu, J. K. Wasman, M. M. Cooney, P. S. Savvi-
des, J. A. Bokar, A. Dowlati, D. Wang, S. S. Agarwala, S. M. Flick, P. H.
Hartman, J. D. Ortiz, P. N. Lavertu, S. C. Remick, Thyroid 2009, 19, 233–
240.
[10] G. J. Rustin, G. Shreeves, P. D. Nathan, A. Gaja, T. S. Ganesan, D. Wang, J.
Boxall, L. Poupard, D. J. Chaplin, M. R. L. Stratford, J. Balkissoon, M.
Zweifel, Br. J. Cancer 2010, 102, 1355–1360.
[11] P. D. Nathan, M. Zweifel, A. R. Padhani, D. M. Koh, M. Ng, D. J. Collins, A.
Harris, C. Carden, J. Smythe, N. Fisher, N. J. Taylor, J. J. Stirling, S. P. Lu,
M. O. Leach, G. J. Rustin, I. Judson, Clin. Cancer Res. 2012, 18, 3428–
3439.
[12] G. R. Pettit, M. R. Rhodes, D. L. Herald, D. J. Chaplin, M. R. L. Stratford, E.
Hamel, R. K. Pettit, J.-C. Chapuis, D. Oliva, Anti-Cancer Drug Des. 1998,
13, 981–993.
[13] G. R. Pettit, B. E. Toki, D. L. Herald, M. R. Boyd, E. Hamel, R. K. Pettit, J.-C.
Chapuis, J. Med. Chem. 1999, 42, 1459–1465.
Chorioallantoic membrane (CAM) assay: Fertilized, specific patho-
gen-free (SPF) chicken eggs (VALO BioMedia) were bred in an incu-
bator at 378C and 60% relative humidity. On day 5 windows
(Ø 1.5–2 cm) were cut into the eggshell at the more rounded pole.
The cavities were sealed with tape, and incubation was continued
for 12–18 h. Then rings of thin silicon foil (Ø 5 mm) were placed on
the CAM with its developing blood vessels, and 1a (10 nmol), 10
or 11 (2.5 nmol) or respective amounts of DMSO or DMF as solvent
controls (all in a volume of 10 mL H2O) were added. The effects on
the developing vasculature were documented after further incuba-
tion for 6 and 24 h using a light microscope (60ꢁ magnification,
Traveler).[39,40] For quantitative evaluation, image sections were
taken from the light microscopy pictures and the area taken up by
blood vessels (fractal dimension) was determined by fractal analy-
sis[41,42] using the following software: ImageJ 1.48f and Fractal
Analysis System 3.4.7 (see the Supporting Information for a detailed
description). The area in the image sections covered by blood ves-
sels (fractal dimension) before treatment with the test compounds
was set at 100%, and the decrease in this area after exposure to
gold complexes 10 and 11 for 6 h was determined.
Animal studies: All animal work was approved by the committee
on animal care (Government of Rhineland Palatinate). The mice
(Balb/c) were kept under a 12 h light–dark cycle in a controlled
conventional colony room. The mice had free access to sterilized
water and standard rodent diet ad libitum. The mice were sacri-
ficed by cervical dislocation corresponding to the guidelines of the
committee on animal care. To analyze the in vivo antitumor activity
and tolerance of gold complexes 11 a and 11 c, a Balb/c mouse
xenograft model of highly metastatic B16-F10 mouse melanoma
cells (Department of Dermatology, Medical University Mainz, Ger-
many) was used. A suspension of 2ꢁ105 cells in 200 mL PBS was
administered to the flanks of each mouse to generate subcutane-
ous xenograft tumors. For these in vivo studies, the gold com-
plexes 11 a and 11 c were dissolved in DMSO and further diluted
with PBS. The mice (n=2) were injected i.p. with 15 mg(kg body
weight)À1 of complex 11 a or 11 c twice on two consecutive days.
Control mice (n=2) were treated with respective amounts of
DMSO (in PBS) only once on two consecutive days. The size of the
tumors and the weight of the mice were recorded on a daily basis
beginning on the day of the first injection. Tumor volumes were
calculated by the formula a2·b·0.5, with a being the short dimen-
sion and b the long dimension.
[14] L. Wang, K. W. Woods, Q. Li, K. J. Barr, R. W. McCroskey, S. M. Hannick, L.
Gherke, R. B. Credo, Y.-H. Hui, K. Marsh, R. Warner, J. Y. Lee, N. Zielinski-
Mozng, D. Frost, S. H. Rosenberg, H. L. Sham, J. Med. Chem. 2002, 45,
1697–1711.
[15] T. Brown, H. Holt, Jr., M. Lee in Heterocyclic Antitumor Antibiotics (Ed.: M.
Lee), Springer, Berlin/Heidelberg, 2006, pp. 1–51.
[16] R. Schobert, B. Biersack, A. Dietrich, K. Effenberger, S. Knauer, T. Mꢂller,
J. Med. Chem. 2010, 53, 6595–6602.
[17] W. Liu, R. Gust, Chem. Soc. Rev. 2013, 42, 755–773.
[18] L. Oehninger, R. Rubbiani, I. Ott, Dalton Trans. 2013, 42, 3269–3284.
[19] L. Kaps, B. Biersack, H. Mꢂller-Bunz, K. Mahal, J. Muenzner, M. Tacke, T.
Mueller, R. Schobert, J. Inorg. Biochem. 2012, 106, 52–58.
[20] J. Mꢂnzner, B. Biersack, L. Kaps, R. Schobert, F. Sasse, Int. J. Clin. Pharma-
col. Ther. 2013, 51, 44–46.
[21] T. Mosmann, J. Immunol. Methods 1983, 65, 55–63.
[22] D. W. Shen, C. Cardarelli, J. Hwang, M. Cornwell, N. Richert, S. Ishii, I.
Pastan, M. M. Gottesman, J. Biol. Chem. 1986, 261, 7762–7770.
[23] M. M. Cornwell, A. R. Safa, R. L. Felsted, M. M. Gottesman, I. Pastan, Proc.
Natl. Acad. Sci. USA 1986, 83, 3847–3850.
[24] A. Fojo, S. Akiyama, M. M. Gottesman, I. Pastan, Cancer Res. 1985, 45,
3002–3007.
[25] M. C. Willingham, M. M. Cornwell, C. O. Cardarelli, M. M. Gottesman, I.
Pastan, Cancer Res. 1986, 46, 5941–5946.
[26] S. K. Rabindran, D. D. Ross, L. A. Doyle, W. Yang, L. M. Greenberger,
Cancer Res. 2000, 60, 47–50.
Acknowledgements
[27] C. Hirschmann-Jax, A. E. Foster, J. G. Nuchtern, T. W. Jax, U. Gobel, M. A.
Goodell, M. K. Brenner, Proc. Natl. Acad. Sci. USA 2004, 101, 14228–
14233.
We thank the COST Action CM1105 “Functional metal complexes
that bind to biomolecules” and the Deutsche Forschungsgemein-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1195 – 1204 1203

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
[28] C. Kanthou, G. M. Tozer, Blood 2002, 99, 2060–2069.
[29] J. Huot, F. Houle, S. Rousseau, R. G. Deschesnes, G. M. Shah, J. Landry, J.
Cell Biol. 1998, 143, 1361–1373.
[30] H. Y. Yamada, G. J. Gorbsky, Mol. Cancer Ther. 2006, 5, 2963–2969.
[31] H. L. Lin, S. H. Chiou, C. W. Wu, W. B. Lin, L. H. Chen, Y. P. Yang, M. L. Tsai,
Y. H. Uen, J. P. Liou, C. W. Chi, J. Pharmacol. Exp. Ther. 2007, 323, 365–
373.
[32] E. A. Papakonstanti, C. Stournaras, FEBS Lett. 2008, 582, 2120–2127.
[33] S. Pellegrin, H. Mellor, J. Cell Sci. 2007, 120, 3491–3499.
[34] C. Gialeli, A. D. Theocharis, N. K. Karamanos, FEBS J. 2011, 278, 16–27.
[35] U. B. Hofmann, J. R. Westphal, G. N. Van Muijen, D. J. Ruiter, J. Invest.
Dermatol. 2000, 115, 337–344.
[39] K. Norrby, J. Cell. Mol. Med. 2006, 10, 588–612.
[40] B. Nitzsche, C. Gloesenkamp, M. Schrader, M. Ocker, R. Preissner, M.
Lein, A. Zakrzewicz, B. Hoffmann, M. Hçpfner, Br. J. Cancer 2010, 103,
18–28.
[41] L. M. Kirchner, S. P. Schmidt, B. S. Gruber, Microvasc. Res. 1996, 51, 2–14.
[42] B. R. Masters, Annu. Rev. Biomed. Eng. 2004, 6, 427–452.
[43] W. Liu, K. Bensdorf, M. Proetto, A. Hagenbach, U. Abram, R. Gust, J. Med.
Chem. 2012, 55, 3713–3724.
[44] R. K. Jain, Science 2005, 307, 58–62.
[45] A. Bergamo, G. Sava, Dalton Trans. 2011, 40, 7817–7823.
[46] S. Rodrꢃguez-Nieto, M. Gonzꢄlez-Iriarte, R. Carmona, R. MuÇoz-Chꢄpuli,
M. A. Medina, A. R. Quesada, FASEB J. 2001, 15, 261–263.
[36] P. A. Snoek-van Beurden, J. W. Von den Hoff, Biotechniques 2005, 38,
73–83.
[37] M. L. Ponce, Methods Mol. Biol. 2009, 467, 183–188.
[38] Y. Muthukumar, M. Roy, A. Raja, R. E. Taylor, F. Sasse, ChemBioChem
2013, 14, 260–264.
Received: January 21, 2014
Published online on March 19, 2014
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 1195 – 1204 1204