Synthesis and biological evaluation of a class of mitochondrially-targeted gadolinium(III) agents

Article abstract of DOI:10.1002/chem.201404107

A structure-activity relationship study of a library of novel bifunctional GdIII complexes covalently linked to arylphosphonium cations is reported. Such complexes have been designed for potential application in binary cancer therapies such as neutron cap

Full text of DOI:10.1002/chem.201404107

DOI: 10.1002/chem.201404107
Full Paper
&
Therapeutic Agents
Synthesis and Biological Evaluation of a Class of Mitochondrially-
Targeted Gadolinium(III) Agents
Daniel E. Morrison,[a] Jade B. Aitken,[a] Martin D. de Jonge,[b] Fatiah Issa,[a] Hugh H. Harris,[c]
and Louis M. Rendina*[a]
Abstract: A structure–activity relationship study of a library
of novel bifunctional GdIII complexes covalently linked to
arylphosphonium cations is reported. Such complexes have
been designed for potential application in binary cancer
therapies such as neutron capture therapy and photon acti-
vation therapy. A positive correlation was found between
lipophilicity and cytotoxicity of the complexes. Mitochondria
uptake was determined by means of inductively coupled
plasma mass spectrometry (ICP-MS), and Gd uptake was
determined by means of quantification using synchrotron X-
ray fluorescence (XRF) imaging. A negative correlation be-
tween lipophilicity and tumour selectivity of the GdIII com-
plexes was demonstrated. This study highlights the delicate
balance required to minimise in vitro cytotoxicity and opti-
mise in vitro tumour selectivity and mitochondrial localisa-
tion for this new class of mitochondrially-targeted binary
therapy agents. We also report the highest in vitro tumour
selectivity for any Gd agent reported to date, with a T/N
(tumour/normal cell) ratio of up to 23.5Æ6.6.
Introduction
are the experimental binary therapies known as neutron
capture therapy (NCT)[3] and photon activation therapy (PAT).[4]
NCT combines non-ionising, low-energy (slow) thermal neu-
tron irradiation with suitable nuclides such as the naturally-
occurring and non-radioactive 10B or 157Gd isotopes.[2] Both of
these nuclides possess very high neutron capture cross-sec-
tions, which allow for site-specific neutron capture reactions
leading to the formation of cytotoxic decay products with
a limited range of less than one-cell diameter.[5] In the case of
the 157Gd nucleus, the therapeutically relevant decay products
appear to be the Auger and Coster–Krçnig (ACK) electrons, as
they are high linear energy transfer (LET) particles.[5b–e] In con-
trast, PAT relies on high-energy X-ray photons to instigate the
release of photoelectric species including ACK electrons from
high-Z elements such as Pt or I, which leads to cell death.[6]
Synchrotron stereotactic radiotherapy (SSR) is a subset of PAT
that incorporates X-ray photons sourced from a synchrotron
and tuned at or slightly above the K-edge of the heavy ele-
ment. Rotation of the patient about the target site to increase
the relative radiation dose within the tumour site is another
feature of SSR.[7] GdSSR (GdPAT) has been proposed as a viable
therapy for the treatment of the aggressive and intractable
malignant brain tumour known as glioblastoma multiforme,[7]
but clinically suitable Gd agents are yet to be developed.
GdIII complexes are extensively used as contrast agents for
magnetic resonance imaging (MRI).[1] Recently, there has been
a great interest in the development of Gd agents for therapeu-
tic application, in particular binary therapies for the treatment
of cancers such as those of the brain.[2] The primary focus of
binary therapies is to treat aggressive and intractable cancers
whilst minimising toxic side-effects to improve patient out-
comes. This therapeutic strategy contrasts with traditional
cancer therapies, for example, chemotherapy, which have the
detrimental consequence of causing considerable collateral
damage to a patient’s healthy tissue during the course of treat-
ment. Binary therapies endeavour to address these types of
issues by means of the combination of two relatively innocu-
ous components which, when combined, allow for a powerful
cytotoxic effect to occur in situ. In general, binary therapies
employ a non-specific, “broad area” component such as neu-
trons or high-energy X-ray photons and a tumour-specific
agent with low cytotoxicity. Of particular interest in this regard
[a] Dr. D. E. Morrison, Dr. J. B. Aitken, Dr. F. Issa, Prof. L. M. Rendina
School of Chemistry, The University of Sydney
Sydney, NSW 2006 (Australia)
It has been demonstrated that ACK electrons generated
from DNA-bound Gd3+ ions have the ability to cleave double-
stranded DNA,[8] a direct way of effecting cell death. The lethal
effects of ACK electrons are likely to be due to the subsequent
generation of reactive oxygen species (ROS) from water,
increasing the oxidative stress within the target cell.[5d,7] How-
ever, ACK electrons possess a mean path length of only about
12 nm.[9] This extremely short path length imposes another re-
E-mail: lou.rendina@sydney.edu.au
[b] Dr. M. D. d. de Jonge
X-ray Fluorescence Microscopy Beamline, Australian Synchrotron
Clayton, Victoria 3168 (Australia)
[c] Prof. H. H. Harris
School of Chemistry and Physics, The University of Adelaide
Adelaide, SA 5005 (Australia)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201404107.
Chem. Eur. J. 2014, 20, 16602 – 16612
16602
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
quirement on the design of new Gd agents for use in binary
therapies; that is, the agents must accumulate within critical
sub-cellular organelles such as the nucleus or the mitochon-
dria. Damage to either of these entities can lead to apoptosis.
While there are ever-increasing facilities capable of providing
suitable neutron and X-ray sources, there is a dearth of suita-
ble Gd agents for use in binary therapies such as NCT and PAT.
The design of suitable agents for potential use in these binary
therapies requires that certain key criteria are met, namely the
ability to achieve high tumour selectivity and uptake, accumu-
lation within critical cellular organelles, and low cytotoxicity in
the absence of thermal neutrons or X-ray photons.
mulate within the nuclei of tumour cells in vitro.[10] However,
the inherent cytotoxicity of DNA-targeted agents such as 2 has
hindered further investigation of this class of compounds.
A new avenue for the design of Gd-agents for use in binary
therapies includes mitochondrially-targeted agents.[12] Delocal-
ised lipophilic cations (DLCs) have garnered increasing atten-
tion for their ability to act as tumour-targeted delivery vectors
for a range of tumour imaging and therapeutic applications.[13]
In particular, tetraphenylphosphonium (TPP) and triphenylme-
thylphosphonium (TPMP) cations were shown to selectively ac-
cumulate in brain tumours, with a tumour to normal (T/N)
tissue ratio of up to 48:1 in a canine brain tumour model.[14]
The ability of DLCs to accumulate selectively within the mito-
chondria of cancer cells stems from the fact that cancer cells
typically possess elevated plasma and mitochondrial mem-
brane potentials (Dy).[15] For example, the mitochondrial mem-
brane potential of most cancer cells is elevated by 60 mV
when compared to normal healthy cells. This difference in Dy
promotes the selective uptake of cationic species.
The first class of compounds to be investigated for use in
Gd-based binary therapies were Gd-based contrast agents,
which are already employed for clinical use in MRI.[5d] However,
none of these GdIII complexes displayed a significant ability to
selectively accumulate within tumour sites nor an ability to lo-
calise within important sub-cellular organelles.[5d] DNA-targeted
agents have also been investigated in recent years. Two such
agents include MGd[7] 1 and a PtII–GdIII complex 2.[10] MGd, ini-
tially investigated as a chemo- and radio-sensitiser, had been
heavily investigated in vivo until the FDA ceased further clini-
cal trials with this agent in late 2007.[11] It has been shown that
MGd was able to localise within the nuclei of tumour cells.[7]
The metallointercalator 2 was also shown to selectively accu-
Phosphonium salts have also been implemented extensively
in BNCT applications and as PET imaging agents.[13b–g] Their use
in the therapeutic delivery of Gd to tumour sites represents
a new direction for GdNCT and GdPAT. We recently reported
the tumour cell uptake and selectivity of two phosphonium-
containing GdIII complexes which have set a new benchmark
as potential GdNCT and GdPAT agents.[12] Such GdIII complexes
showed very low in vitro cytotoxicity, high tumour uptake and
selectivity, and an ability to accumulate within the mitochon-
dria of tumour cells. Indeed, the GdIII complexes were able to
achieve comparable tumour cell selectivity to the DNA-target-
ed agents such as 2, but with far greater tumour uptake and
lower cytotoxicity. By means of synchrotron X-ray fluorescence
(XRF) quantification, it was demonstrated that that these com-
plexes were able to accumulate Gd at high levels (ca. 1011 Gd
atoms per tumour cell), a few orders of magnitude higher than
therapeutically relevant thresholds.[12]
Herein we report a comprehensive evaluation of selected
phosphonium bearing GdIII-agents in an effort to aid in their
future design to optimise the key criteria of tumour selectivity,
tumour cell uptake, cytotoxicity, and mitochondrial localisation.
Results and Discussion
A primary challenge in the design of mitochondria-targeted
GdIII complexes is the incorporation of a charged lipophilic spe-
cies within a GdIII chelator whilst not compromising the high
complexation capacity of the chelating agent nor the tumour-
targeting ability of the DLC. Ideally, the complex should be
monocationic at the phosphonium site and the Gd macrocycle
remains uncharged to allow a selective response to the elevat-
ed membrane potentials of cancer cells as well as preserving
its ability to cross cell membranes. 1,4,7,10-tetraazacyclodode-
cane-1,4,7-triacetic acid (DO3A) provides a highly suitable scaf-
fold to achieve this aim because not only is the DO3A-GdIII
moiety neutral at physiological pH (7.4),[16] it also provides the
strong binding affinity associated with the cyclen-derived poly-
acetate class of macrocycles and a secondary amine site for
Chem. Eur. J. 2014, 20, 16602 – 16612
www.chemeurj.org
16603
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
further synthetic functionalisation.[17] A triphenylphosphonium
moiety can easily be generated from the quaternisation of tri-
arylphosphines by using a standard alkylation reaction. Indeed,
coupling the two functional moieties by means of a xylyl linker
allows facile reactions by the use of a dibromoxylene reagent,
which can undergo nucleophilic substitution reactions involv-
ing both triarylphosphines and the secondary amine site of the
DO3A macrocycle.
A small library of phosphonium-containing bifunctional GdIII
complexes was synthesised and characterised in this work. The
library consisted of three distinct series of GdIII complexes, de-
fined by the nature of the phosphonium cation present. Within
each series, the complexes were varied by the substitution pat-
tern of the xylyl group (that is, para-, meta-, and ortho-xylyl)
linking the phosphonium centre to the DO3A-GdIII macrocycle.
The different isomers allowed for a variation in intramolecular
distance between the functional moieties, which may have
a significant effect on the biological activity of the complexes,
in particular their ability to cross cell membranes. The first
series of GdIII complexes were linked to the parent triphenyl-
phosphonium cation. The second series bore an iodo-function-
ality in the form of an iodophenyldiphenylphosphonium
cation, while the third series contained methoxy substituents
in the form of a tris(methoxyphenyl)phosphonium cation.
These latter two classes were synthesised to modulate the lip-
ophilicity of the complexes to allow for structure–activity rela-
tionships (SARs) to be established on the basis of lipophilicity
of the phosphonium cation.
Scheme 1. Synthesis of phosphonium-containing GdIII complexes. TFA=tri-
fluoroacetic acid, DO3A·(tBu)3 =1,4,7,10-tetraazacyclododecane-1,4,7-triacetic
acid tri-tert-butyl ester.
Determination of lipophilicity
The lipophilicity of DLCs largely governs their ability to traverse
and accumulate within tumour-cell mitochondria. Furthermore,
lipophilicity plays an important role in the absorption, distribu-
tion, metabolism, and excretion of therapeutic compounds,
and so the lipophilicity of the prepared complexes was used as
the basis for determining structure–activity relationships (SARs)
of the novel GdIII complexes. LogP values were determined for
the GdIII complexes 3–11, as well as the parent triarylphospho-
nium salts TPP, TPMP, and ITPMP, by using a reverse-phase
HPLC method involving standard reference compounds. A
summary of results is shown in Table 1.
Synthesis
The synthesis of the novel GdIII complexes 3–11 is summarised
in Scheme 1 (see also the Experimental Section). Commencing
from the appropriate triarylphosphine, a,a’-dibromoxylene
(ortho, meta, or para) underwent a nucleophilic substitution re-
action with the triarylphosphine to give a bromoxylene-func-
tionalised arylphosphonium salt. This reaction was performed
in a low-polarity solvent (toluene) to promote the precipitation
of the monosubstituted species from solution, thus precluding
the formation of the undesired disubstituted species. The pres-
ence of the bromoxylyl group within the phosphonium salts
allowed for facile coupling to a tert-butyl protected DO3A
macrocycle under basic conditions. Deprotection under acidic
conditions using TFA yielded the free ligand in high yield. The
ligands were purified by means of reverse-phase HPLC prior to
the complexation of Gd3+ ions. The complexation reaction was
performed by using Gd2O3 as this allowed for any excess (in-
soluble) Gd2O3 to be readily removed by means of filtration.
The GdIII complexes were purified by means of analytical re-
verse-phase HPLC, with a purity of more than 95% achieved
for all of the complexes. The identity of each complex was con-
firmed by means of high-resolution ESI-MS. Through a series of
qualitative experiments, the complexes were found to be
stable in aqueous media at biologically significant pH follow-
ing incubation at 378C for prolonged periods. Both ESI mass
spectrometry and UV/Vis spectroscopy confirmed the robust-
ness of the complexes, even after 24 h in solution.
Table 1. Experimental logP values for GdIII complexes and phosphonium
cations.
Compound
logP[a]
3
4
5
6
7
8
9
10
1.24Æ0.02
1.35Æ0.02
1.56Æ0.02
1.44Æ0.03
1.53Æ0.01
2.09Æ0.03
1.86Æ0.08
1.91Æ0.01
2.73Æ0.01
1.20Æ0.01
0.72Æ0.02
1.47Æ0.01
11
TPP
TPMP
ITPMP
[a] MeanÆSE.
Chem. Eur. J. 2014, 20, 16602 – 16612
www.chemeurj.org
16604
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
The logP values determined for the GdIII complexes were
comparable to those of the simple phosphonium cations, indi-
cating that the inclusion of the DO3A macrocycle does not
greatly impact on the overall lipophilicity of the GdIII com-
plexes. Structural variation of the phosphonium cation resulted
in the observed trend for lipophilicity being trismethoxy bear-
ing> iodo bearing> parent phosphonium complexes. The ex-
perimental logP values were consistent with the observed
aqueous solubility of the complexes, where both the parent
and the iodo parent analogue were soluble at levels>
20 mgmLÀ1, while the methoxy bearing analogue was only
found to be soluble at about 10 mgmLÀ1. Isomeric variations
pertaining to the xylyl linker used within the GdIII complexes
resulted in the observed trend for lipophilicity being ortho>
meta> para. This trend indicates that the greater the spatial
separation of the two functional moieties, the lower the overall
lipophilicity of the complexes. This is likely to be related to the
increase in the overall dipole moment of the complexes in the
order ortho <meta <para.
6 exhibited lower cytotoxicity than complexes 7 and 8. No sig-
nificant difference was observed between complexes 7 and 8.
For the methoxy bearing series, there was no observed differ-
ence between the three complexes 9–11. This gives a general
trend that increased lipophilicity owing to decreasing the
distance between the functional moieties led to either a com-
parable or higher cytotoxicity.
Comparing the different sets of compounds with similar
xylyl isomers, complex 3 was found to be less cytotoxic than 6.
Complex 9 was not found to be significantly different from
either complex 3 or 6. Complex 4 was the least cytotoxic of
the meta-xylyl bearing complexes, followed by 10, with 7
found to be the most cytotoxic. Complexes 5 and 11 did not
show any significant difference between cytotoxicity and were
both less cytotoxic than 8. The effect of the inclusion of the
iodo functionality into the phosphonium cation led to a clear
increase in cytotoxicity in all cases. However, the effect of in-
clusion of methoxy-substituents was less clear. Indeed, due to
the lack of any statistically significant differences, general
trends cannot be drawn from these comparisons.
For the parent series and the iodo bearing series, it was
found that an increase in lipophilicity led to comparable or
higher cytotoxicities. This trend was not observed for the me-
thoxy bearing series and may indicate that these species
behave differently within cells to the parent and iodo bearing
series. Indeed, previously reported synchrotron X-ray fluores-
cence (XRF) elemental mapping has shown a significant differ-
ence between the cellular localisation of the iodo series and
the methoxy series.[12]
In vitro cytotoxicity studies
In vitro cytotoxicities were determined by means of standard
MTT colorimetric assays. Assays were performed up to 500 mm
of GdIII complex. The maximum concentration used in these ex-
periments exceeded the expected dosing concentrations for
subsequent in vitro and in vivo testing. At this maximum con-
centration, cell viability was not reduced below 50% by any of
the GdIII complexes. However the cell viability at the maximum
concentrations were determined and used as the basis for
comparing cytotoxicities (Table 2). Statistical analyses, by
To further probe the in vitro cytotoxicity of the complexes,
higher concentrations of GdIII complex were used (up to a final
concentration of 4 mm). IC50 values are reported in Table 3. The
IC50 values agree with previously obtained trends for cytotoxic-
ity by means of cell viability at 500 mm. (see above).
Table 2. Mean cell viability of T98G cells treated with 500 mm of GdIII
complexes for 72 h.
Compound
% Viability[a]
N
Table 3. IC50 values for complexes 3, 6, and 9 in the T98G cell line.
3
4
5
6
7
8
9
10
11
97Æ3%
101Æ4%
85Æ2%
85Æ3%
60Æ3%
65Æ4%
89Æ4%
90Æ2%
87Æ2%
9
9
9
9
9
9
9
9
9
Compound
IC50 [mm][a]
N
3
6
9
2.55Æ0.33
2.14Æ0.32
1.42Æ0.20
4
4
4
[a] MeanÆSE.
[a] MeanÆSE.
In vitro uptake studies on whole cells
In vitro cellular uptake studies were performed with complexes
3, 6, and 9 in the T98G cell line and primary human carotid
artery endothelial cells (HCtAEC). Cells were incubated in the
presence of 100 mm of GdIII complex for 48 h, prior to cell har-
vesting and analysis by means of ICP-MS. All three of the com-
plexes showed significantly greater uptake into the T98G
tumour cell line compared to the HCtAEC normal cells. Table 4
and Table 5 provide a summary of Gd uptake into the tumour
and normal cell lines expressed in terms of ng of Gd per mg of
protein, respectively.
means of a two-tailed Student’s t-test with an a value of 0.05,
were performed to ascertain which differences in cell viability
as a result from treatment with the GdIII complexes were signif-
icant. The only instances where cell viability did not differ sig-
nificantly from the control cells following treatment were for
complexes 3 and 4. As cell viability correlates inversely with
cytotoxicity, preliminary trends for in vitro cytotoxicity could
thus be determined.
In the parent series, complexes 3 and 4 exhibited a lower cy-
totoxicity than complex 5. For the iodo bearing series, complex
Chem. Eur. J. 2014, 20, 16602 – 16612
www.chemeurj.org
16605
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 4. ICP-MS analysis of Gd in T98G tumour cells treated with 3, 6,
and 9 (100 mm) for 48 h, expressed as ng Gd/mg of protein.
Table 6. ICP-MS analysis of Gd in T98G cells treated with 3, 6, and 9
(100 mm) for 48 h, expressed as a ratio of Gd content in the mitochondrial
fraction compared to the cytosolic fraction.
Control
3
6
9
Control
3
6
9
Mean
SE
N
0.53
0.11
3
4071.6
605.8
3
Y
5355.0
124.7
3
Y
4824.4
1006.8
3
Y
Mean
SE
N
0.5
0.3
4
2.0
0.6
4
3.4
2.3
4
5.4
2.4
4
Significantly different (Y/N)[a]
N/A
[a] Statistical significance was accepted at the 95% confidence interval
with p-values determined using the Mann–Whitney U test.
tivity it is clear that overall lipophilicity is an important tunea-
ble property for maximising tumour uptake and mitochondrial
accumulation of phosphonium bearing Gd-DO3A complexes.
Table 5. ICP-MS analysis of Gd in HCtAEC cells treated with 3, 6, and 9
(100 mm) for 48 h, expressed as ng Gd/mg of protein.
Control
3
6
9
Synchrotron X-ray fluorescence (XRF) studies
Mean
SE
N
0.05
0.05
3
173.3
41.0
4
697.9
144.1
3
803.7
168.1
4
T98G human glioblastoma cell line: Synchrotron X-Ray fluores-
cence (XRF) studies were performed on T98G tumour cells
treated with selected GdIII complexes to allow for visualisation
of elemental distribution within individual cells. Data were col-
lected for control cells and cells treated with 100 mm of com-
plexes 3, 6 and 7 for 24 h. Cells were grown on Si3N4 windows
and fixed using MeOH at À788C.
Significantly different (Y/N)[a]
N/A
Y
Y
Y
[a] Statistical significance was accepted at the 95% confidence interval
with p-values determined using the Mann-Whitney U test.
The ratio of tumour to normal uptake (T/N) provides an indi-
cation of in vitro selectivity of the complexes. Complex 3
showed the highest T/N selectivity at 23.5Æ6.6, while com-
plexes 6 and 7 exhibited T/N selectivities of 7.7Æ1.6 and 6.0Æ
1.8, respectively. The significantly greater selectivity of complex
3 is likely due to its lower relative uptake into the normal cell
line compared to 6 and 9. All three complexes showed com-
parable uptake into tumour cells under the conditions of the
experiment.
Figure 1, Figure 2, and Figure 3 show elemental maps of
a single T98G cell incubated with 100 mm of 3, 6, and 7, re-
spectively, for 24 h. XRF quantitation of intracellular Gd content
The observed uptake and selectivity results for the com-
plexes indicate that a higher lipophilicity is not a necessary
requirement for significant uptake into tumour cells whilst in-
creased lipophilicity can lead to greater uptake into normal
cells leading to a decrease in overall selectivity and a potential-
ly lowered therapeutic effectiveness of the Gd agent.
In vitro mitochondria uptake studies
In vitro mitochondria uptake studies were performed with
complexes 3, 6, and 9 in the T98G cell line. Cells were incubat-
ed in the presence of 100 mm of GdIII complex for 48 h prior to
cell harvesting and then subjected to a reagent-based mito-
chondria isolation kit. The fractions for the pure mitochondria
and the cytosol were collected and analysed by means of ICP-
MS for Gd content. All three complexes showed Gd accumulat-
ed into the mitochondria. Results for mitochondrial selectivity
are reported in terms of the ratio of the mitochondrial to
cytosolic fractions (M/C; Table 6).
Figure 1. Scattered X-ray (SA) and XRF elemental distribution maps showing
P, S, Cl, K, Ca, Fe, Cu, Zn, I, and Gd for the T98G Cell C treated with 100 mm
of 3 for 24 h.
is summarised in Table 7. Statistically significant (single-tailed
Mann–Whitney U test, p <0.05) levels of Gd were present in
the cells after treatment with all of the complexes. For all three
complexes, the synchrotron XRF elemental density maps for
Gd showed a strong correlation to regions of high intensity in
the Fe elemental density maps. This elemental correlation is
consistent with mitochondrial uptake of the complexes owing
to the key role the mitochondria play in cellular Fe regulation,
metabolism (for example heme synthesis and FeÀS cluster as-
Complex 9 showed the highest mitochondrial selectivity
with an M/C ratio of 5.4Æ2.4, while complexes 3 and 6 had M/
C ratios of 2.0Æ0.6 and 3.4Æ2.3, respectively. These results in-
dicate that, as expected, increased lipophilicity allows for
greater mitochondrial accumulation. When the M/C ratios are
considered along with the results for tumour uptake and selec-
Chem. Eur. J. 2014, 20, 16602 – 16612
www.chemeurj.org
16606
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 7. Mean Gd uptake within individual T98G human glioblastoma
cells following treatment with complexes 3, 6, and 7, as determined by
synchrotron X-ray fluorescence (XRF) imaging.
Complex
Incubation time [h]
Gd density/cell[a]
N
3
6
7
24
24
24
0.004Æ0.002
0.06Æ0.04
0.12Æ0.05
5
5
5
[a] MeanÆSD in mgcmÀ2
.
sembly), and storage owing to mitochondrial ferritin (MtFe).[18]
Quantified intracellular content of Gd within the cells following
treatment with 3, 6, and 7 exhibited a positive correlation with
lipophilicity.
A549 human lung cancer cell line: Synchrotron X-ray fluores-
cence (XRF) studies were performed on A549 tumour cells
treated with selected GdIII complexes to allow visualisation of
elemental distribution within individual cells. Data were collect-
ed for control cells and cells treated with 100 mm of complexes
3 and 6 for 24 h. Cells were grown on Si3N4 windows and fixed
using MeOH at À788C. For comparison, additional cells treated
with 6 were fixed by means of flash freezing in a liquid N2/iso-
pentane slurry.
Figure 4 and Figure 5 show elemental maps of a single A549
cell incubated with 100 mm of 3 and 6, respectively, for 24 h
and then fixed in MeOH. Figure 6 shows elemental maps of
Figure 2. Scattered X-ray (SA) and XRF elemental distribution maps showing
P, S, Cl, K, Ca, Fe, Cu, Zn, I and Gd for the T98G Cell A treated with 100 mm
of 6 for 24 h.
Figure 4. Scattered X-ray (SA) and XRF elemental distribution maps showing
P, S, Cl, K, Ca, Fe, Cu, Zn, I and Gd for the A549 Cell A treated with 100 mm
of 3 for 24 h.
a single A549 cell incubated with 6 for 24 h and fixed by flash
freezing. XRF quantitation of intracellular Gd content is sum-
marised in Table 8. Statistically significant (single-tailed Mann–
Whitney U test, p <0.05) levels of Gd were present in the cells
after treatment with all of the complexes. In all three cases,
the synchrotron XRF elemental density maps for Gd showed
a strong correlation to regions of high intensity in the Fe ele-
Figure 3. Scattered X-ray (SA) and XRF elemental distribution maps showing
P, S, Cl, K, Ca, Fe, Cu, Zn, I and Gd for the T98G Cell D treated with 100 mm
of 7 for 24 h.
Chem. Eur. J. 2014, 20, 16602 – 16612
www.chemeurj.org
16607
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Conclusion
A library of new GdIII complexes 3–11 was shown to address
many key criteria for the development of new therapeutic PAT
and NCT agents. All of the complexes were found to be essen-
tially non-cytotoxic at the expected therapeutically relevant
concentrations. They were also shown to selectively accumu-
late within tumour cells more effectively than normal cells and
achieve excellent mitochondrial localisation. Lipophilicity
appears to play a major role in the cytotoxicity, uptake, and se-
lectivity of this class of phosphonium bearing complexes. Of
particular interest, the parent complex 3 exhibited a T/N ratio
of 23.5Æ6.6, a value which was 3–4 times greater than the
substituted derivatives 6 and 9. Indeed, this selectivity value
exceeds those previously determined for clinical agents such
as MGd.
Figure 5. Scattered X-ray (SA) and XRF elemental distribution maps showing
P, S, Cl, K, Ca, Fe, Cu, Zn, I and Gd for the A549 Cell B treated with 100 mm of
6 for 24 h.
The series of complexes containing the tris(methoxyphenyl)-
phosphonium cation exhibited differing behaviour compared
to the parent and iodophenyldiphenylphosphonium series. In
the parent and iodinated series, lipophilicity correlated well
with increased cytotoxicity. However, for the methoxy bearing
series, lipophilicity did not strongly correlate with cytotoxicity.
Coupled with the significant difference in the biological behav-
iour of the methoxy bearing complexes, we hypothesise that
these complexes may undergo a different mechanism of
uptake or mode of action within the cell compared to the
parent or iodo bearing complexes. Further experiments are
required to understand the apparent differences in biological
behaviour between these classes of complexes.
Figure 6. Scattered X-ray (SA) and XRF elemental distribution maps showing
P, S, Cl, K, Ca, Fe, Cu, Zn, I and Gd for the A549 Cell B treated with 100 mm of
6 for 24 h and fixed by flash freezing in a liquid N2/isopentane slurry.
There were key differences found in the observed trends for
Gd uptake correlated to lipophilicity between the ICP-MS stud-
ies and the XRF studies. These differences are likely to arise
from the differences in experimental conditions, for example,
the cell growth density, the incubation times, and the treat-
ment of the cells post incubation; however, further work is re-
quired to unravel the factors involved. Notably, the ICP-MS
studies indicate that lipophilicity may play a role in the ability
of the GdIII complexes to be taken up into normal healthy cells.
This factor alone led to significant differences in the observed
in vitro tumour selectivity of the complexes. Additionally, mito-
chondrial isolation studies indicate that an increased lipophilic-
ity can lead to enhanced mitochondrial localisation of the GdIII
complexes.
Table 8. Mean Gd uptake within individual A549 human lung cancer cells
following treatment with complexes 3 and 6, as determined by synchro-
tron X-ray fluorescence (XRF) imaging.
Complex
Incubation time [h]
Gd density/cell[a]
N
3[b]
6[b]
6[c]
24
24
24
0.010Æ0.009
0.05Æ0.03
0.02Æ0.01
2
5
5
[a] MeanÆSD in mgcmÀ2. [b] Fixed by using MeOH at À788C. [c] Fixed by
flash freezing in a liquid N2/isopentane slurry.
mental density maps. This observation agrees with the obser-
vations made for the XRF experiments performed using the
T98G cell line (see above), supporting the hypothesis of mito-
chondrial localisation. Similarly, the difference in uptake of 3
and 6 were positively correlated to lipophilicity.
The difference in Gd uptake observed for 6, when the
means of cell fixation were varied, highlights the effect of ele-
mental leaching during the fixation process and the impor-
tance of selection of an appropriate procedure to minimise
this effect. Studies that used formaldehyde for fixation showed
higher Gd uptake in T98G cells in comparison to the values re-
ported herein, lending to the hypothesis that less Gd is leach-
ed from the cells during fixation with formaldehyde compared
to MeOH or flash freezing.[12] Alternatively, variations may arise
from differences in uptake between cells in different phases of
the cell cycle.
Taken together, the in vitro cytotoxicity and whole cell
uptake studies demonstrate that an increased lipophilicity of
the GdIII complexes prepared in this work can lead to a some-
what increased cytotoxicity and a decreased tumour cell selec-
tivity. However, this observation must be tempered against the
enhanced mitochondrial localisation that any increased lipo-
philicity might bring. It is clear from this study that a delicate
balance exists between lipophilicity, cytotoxicity, tumour cell
selectivity, and mitochondrial localisation, and all of these
properties must be considered in the design of new phospho-
nium-based Gd-agents for potential use in binary therapies
such as GdNCT and GdPAT. Furthermore, in vivo data are
required to confirm the observed in vitro trends and such
studies will be conducted in due course.
Chem. Eur. J. 2014, 20, 16602 – 16612
www.chemeurj.org
16608
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
The hydration state q of GdIII complexes can be inferred from cor-
relation of the longitudinal relaxivity r1 with similarly structured
GdIII complexes with known hydration states.[23] The relaxivity of
Experimental Section
Materials and methods
complexes 3, 6, and 9 were measured as 5.1, 9.6, and 7.4 mmÀ1 sÀ1
,
All of the reactions were performed under a dry N2 atmosphere
where required. All manipulations were performed by using con-
ventional Schlenk techniques.[19] Distilled water was used for all of
the experiments requiring water. MeCN and PhMe were dried prior
to use according to Armarego and Chai.[20] MeCN was freshly dis-
tilled from CaH2 before use. PhMe was dried over sodium wire and
freshly distilled prior to use. All other solvents were used without
prior preparation. All precursor chemicals used in this study were
commercially available. Triethylenetetraamine was purchased from
Alfa Aeser. All other reagents were purchased from the Sigma Al-
drich Chemical Co. Cyclen was prepared by methods previously
described by Athey et al.[21] DO3A-tBu3·HBr was prepared as pre-
viously described by Moore.[22] Syntheses of precursors and ligand
intermediates are detailed in the Supporting Information.
respectively. Gd-DOTA has a relaxivity of 3.6 mmÀ1 sÀ1 and the N-
substituted DO3A derivative Gadoteridol has relaxivity of
4.1 mmÀ1 À1 [24] Both Gd-DOTA and Gadoteridol possess a hydration
a
s
.
state of 1.[24] The relatively higher relaxivities of 6 and 9 suggest
a hydration state of 2, while the lower relaxivity of 3 suggests a hy-
dration state of 1. The hydration states of the m- and o-isomer
complexes were approximated from the values determined for the
p-isomer complexes.
Stability in aqueous media was observed over a range of pH 6 to
7.4 by two separate methods. In method 1, solutions of the com-
plexes were prepared in acetate buffer (pH 6) in the presence of
bovine serum albumin (BSA) using Xylenol orange as an indicator
and incubated at 378C with periodic recording of UV/Vis spectra.
No change was observed in the spectra over 24 h. In a comparison
experiment using GdCl3, a distinct shift in lmax was observed after
6 h of incubation.
Instrumentation
1
All of the H, 13C{1H}, 19F{1H}, and 31P{1H} NMR spectra were record-
In method 2, solutions of the complexes were prepared in phos-
phate buffered saline (PBS, pH 7.4) and incubated at 378C for 24 h
and observed by mean of ESI-mass spectrometry. The signals with
m/z corresponding to the [MÀCF3CO2]+ of the complexes were ob-
served with no evidence for the presence of free ligand or other
complex species.
ed at 300 K on a Bruker Avance300 spectrometer (1H at 300 MHz,
13C at 75 MHz, 19F at 282 MHz, and 31P at 121 MHz). The NMR sig-
nals (d) are reported in ppm. H and 13C{1H} NMR spectra were ref-
1
erenced according to their solvent residual peaks. 31P{1H} NMR
spectra were referenced to external P(OMe)3 at 140.85 ppm. Relax-
ivity measurements were recorded on a Bruker Minispec mq60
spectrometer at 60 MHz. UV/Vis spectra were recorded on
a Thermo Scientific Nanodrop 2000 UV/Vis spectrophotometer.
Low-resolution ESI-MS were recorded on a Finnigan LCQ mass
spectrometer. High resolution ESI-FT-ICR-MS data was recorded on
a Bruker 7.0T mass spectrometer. Melting points were recorded on
a Barnstead Electrothermal IA9100 digital melting point apparatus.
Values reported are uncorrected.
2,2’,2’’-[10-{4-[(Triphenylphosphonio)methyl]benzyl}-1,4,7,10-tetra-
azacyclododecane-1,4,7-triyl]triacetatogadolinium(III) trifluoroace-
tate (3): The reaction was performed using L1 (79.5 mg, 96.4 mmol)
and Gd2O3 (39.7 mg, 110 mmol). Yield 73.1 mg (77.5%). tR =
11.6 min. Purity> 95% by HPLC. M.p.> 3008C (decomp.). ESI-FT-
ICR-MS for [MÀCF3CO2+H]2+: Calculated m/z 433.6196; Found
433.6196.
2,2’,2’’-[10-{3-[(Triphenylphosphonio)methyl]benzyl}-1,4,7,10-tetra-
azacyclododecane-1,4,7-triyl]triacetatogadolinium(III) trifluoroace-
tate (4): The reaction was performed using L2 (57.8 mg, 70.1 mmol)
and Gd2O3 (27.0 mg, 74.5 mmol). Yield 53.4 mg (76.1%). tR =
13.8 min. Purity> 95% by HPLC. M.p.> 3008C (decomp.). ESI-FT-
ICR-MS for [MÀCF3CO2+H]2+: Calculated m/z 433.6196; Found
433.6196.
HPLC methods
Method 1 (preparative): Performed on a Waters 600 HPLC system
with a Waters 486 tuneable absorbance UV/Vis detector (l=
254 nm) and a Sunfire C18 preparative column (19ꢁ150 mm, 5 mm
pore size). Flow rate 7 mLminÀ1. HPLC was performed under gradi-
ent flow conditions, commencing with 95% solvent A (0.1% tri-
fluoroacetic acid in water) and 5% solvent B (0.1% trifluoroacetic
acid in acetonitrile) and moving to 20% solvent A and 80% sol-
vent B over 30 min.
2,2’,2’’-[10-{2-[(Triphenylphosphonio)methyl]benzyl}-1,4,7,10-tetra-
azacyclododecane-1,4,7-triyl]triacetatogadolinium(III) trifluoroace-
tate (5): The reaction was performed using L3 (46.9 mg, 56.9 mmol)
and Gd2O3 (11.1 mg, 30.7 mmol). Yield 44.2 mg (79.3%). tR =
14.8 min. Purity> 95% by HPLC. M.p.> 2808C (decomp.). ESI-FT-
ICR-MS for [MÀCF3CO2]+: Calculated m/z 866.2322; Found
866.2322.
Method 2 (analytical): Performed on a Waters 2965 separation
module HPLC system with a Waters 2996 photodiode array (PDA)
detector (l=200 to 300 nm) and a Sunfire C18 analytical column
(2.1ꢁ150 mm, 5 mm pore size). Flow rate 0.2 mLminÀ1. HPLC was
performed under gradient flow conditions, commencing with 95%
solvent A (water) and 5% solvent B (acetonitrile) going to 0% sol-
vent A and 100% solvent B over 15 min.
2,2’,2’’-[10-{4-[{(4-Iodophenyl)diphenylphosphonio}methyl]benzyl}-
1,4,7,10-tetraazacyclododecane-1,4,7-triyl]triacetatogadolinium(III)
trifluoroacetate (6): The reaction was performed using L4 (59.0 mg,
62.0 mmol) and Gd2O3 (12.8 mg, 35.3 mmol). Yield 48.1 mg (70.2%).
tR =11.8 min. Purity> 95% by HPLC. M.p.> 2708C (decomp.). ESI-
FT-ICR-MS for [MÀCF3CO2]+: Calculated m/z 992.1285; Found
992.1291.
GdIII complexes
General procedure: The free bifunctional ligands L1–L9 were
stirred as a suspension with Gd2O3 (2 equiv) in H2O at 808C for
12 h (or until completion). The mixture was then centrifuged and
filtered off to remove unreacted and insoluble Gd2O3. The filtrate
was then reduced in vacuo and the product lyophilised to give
a white powder. The purity of the complex was confirmed by
means of analytical reverse-phase HPLC (method 2).
2,2’,2’’-[10-{3-[{(4-Iodophenyl)diphenylphosphonio}methyl]benzyl}-
1,4,7,10-tetraazacyclododecane-1,4,7-triyl]triacetatogadolinium(III)
trifluoroacetate (7): The reaction was performed using L5 (61.9 mg,
65.1 mmol) and Gd2O3 (13.7 mg, 37.7 mmol). Yield 49.0 mg (68.1%).
tR =13.5 min. Purity> 95% by HPLC. M.p.> 2708C (decomp.). ESI-
FT-ICR-MS for [MÀCF3CO2]+: Calculated m/z 992.1285; Found
992.1326.
Chem. Eur. J. 2014, 20, 16602 – 16612
www.chemeurj.org
16609
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
2,2’,2’’-[10-{2-[{(4-Iodophenyl)diphenylphosphonio}methyl]benzyl}-
1,4,7,10-tetraazacyclododecane-1,4,7-triyl]triacetatogadolinium(III)
trifluoroacetate (8): The reaction was performed using L6
(107.6 mg, 113 mmol) and Gd2O3 (43.7 mg, 12.1 mmol). Yield
65.5 mg (52.4%). tR =18.4 min. Purity> 95% by HPLC. M.p.>
2508C (decomp.). ESI-FT-ICR-MS for [MÀCF3CO2]+: Calculated m/z
992.1285; Found 992.1289.
Cytotoxicity assays: The cytotoxicity of complexes was assessed
using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bro-
mide (MTT) assay.[25] Briefly, cells were harvested with trypsin (0.1%
v/v), and cell pellets were isolated by centrifugation. Cells were
then re-suspended to a single cell suspension, cell numbers were
counted using a hemocytometer (Weber), and then cells were
seeded (density 1ꢁ104 cells per well) in growth medium (100 mL)
using 96-well plates and were allowed to adhere overnight at
378C. Cells were incubated at 378C in a humidified atmosphere of
5% CO2 in the presence of complexes 3–11 or the vehicle (control).
Serial dilutions of the Gd complex were added to triplicate or du-
plicate wells. Maximum concentration (MaxC) for the experiments
were either 500 mm (N=9) or 4 mm (N=4). After 72 h, MTT solution
in phosphate-buffered saline (PBS; 30 mL, 0.17% w/v) was added
and the incubation was continued. After a further 4 h, the culture
medium and excess MTT solution were removed and the resulting
MTT–formazan crystals dissolved by addition of 150 mL DMSO. Cell
viability was determined by measuring the absorbance at either
600 nm using a Victor3V microplate reader (PerkinElmer). All read-
ings were corrected for absorbance from wells containing the vehi-
cle alone, and the level of MTT was expressed relative to the corre-
sponding vehicle-treated controls as% viability. Corresponding IC50
values for each of the compounds tested were then determined at
the dose required to induce a 50% decrease in cell viability. All ex-
periments were conducted at least in triplicate and all IC50 values
are reported with standard errors where possible.
2,2’,2’’-[10-{4-[(Tris(4-methoxyphenyl)phosphonio]methyl}benzyl]-
1,4,7,10-tetraazacyclododecane-1,4,7-triyl)triacetatogadolinium(III)
trifluoroacetate (9): The reaction was performed using L7 (98.8 mg,
108 mmol), and Gd2O3 (47.5 mg, 131 mmol). Yield 94.8 mg (81.5%).
tR =14.7 min. Purity> 95% by HPLC. M.p.> 3008C (decomp.). ESI-
FT-ICR-MS for [MÀCF3CO2]+: Calculated m/z 956.2636; Found
956.2646.
2,2’,2’’-[10-{3-[(Tris(4-methoxyphenyl)phosphonio]methyl}benzyl]-
1,4,7,10-tetraazacyclododecane-1,4,7-triyl)triacetatogadolinium(III)
trifluoroacetate (10): The reaction was performed using L8
(46.5 mg, 50.8 mmol), and Gd2O3 (62.77 mg, 173 mmol). Yield
52.2 mg (96.0%). tR =11.5 min. Purity> 95% by HPLC. M.p.>
3008C (decomp.). ESI-FT-ICR-MS for ([MÀCF3CO2]+)3: Calculated m/
z 955.9307; Found 955.9320.
2,2’,2’’-[10-{2-[(Tris(4-methoxyphenyl)phosphonio]methyl}benzyl]-
1,4,7,10-tetraazacyclododecane-1,4,7-triyl)triacetatogadolinium(III)
trifluoroacetate (11): The reaction was performed using L9
(126.5 mg, 138 mmol), and Gd2O3 (234.76 mg, 648 mmol). Yield
144.3 mg (97.7%). tR =11.5 min. Purity> 95% by HPLC. M.p.>
3008C (decomp.). ESI-FT-ICR-MS for [MÀCF3CO2]+: Calculated m/z
956.2639; Found 956.2629.
Cell uptake studies: Stock solutions of 3 and 6 (20 mm in H2O)
and 9 (10 mm in H2O) were prepared. Warm (378C) culture
medium was treated with the stock Gd solutions to a final concen-
tration of 100 mm. The T98G cells were cultured as a monolayer in
25 cm2 flasks to 70–80% confluence and then incubated with the
Gd-containing culture medium at 100 mm for 48 h at 378C in a hu-
midified 5% CO2 atmosphere. The medium was removed and the
cells were washed once with PBS (1 mL). PBS (2 mL) was added to
the culture flask and then cells were harvested with a rubber po-
liceman. Harvested cells were sedimented by centrifugation at
3000 rpm for 5 min, and then the supernatant removed. PBS
(2 mL) was added to the cell pellet, which was then pipetted until
a single cell suspension was achieved. An aliquot (100 mL) was
taken for protein analysis. The remaining cells were sedimented by
centrifugation at 3000 rpm for 5 min, then the supernatant was re-
moved and the cell pellet was analysed for Gd content by means
of ICP-MS.
Determination of LogP values by an HPLC method
The logP values of complexes 3--11 were measured using a stan-
dard reverse-phase HPLC method. Phosphate buffer (0.05m) was
prepared by dissolving KH2PO4 (3.4 g, 25 mmol) in H2O (500 mL)
and the pH was adjusted to ca. 7 with NaOH solution (0.1m). Sam-
ples were analysed using a Waters Sunfire C18 column (2.1ꢁ
150 mm) with a mobile phase of MeOH and phosphate buffer
(65:35 v/v) with a flow rate of 0.2 mLminÀ1, UV absorbance at
254 nm. The lipophilicity of each compound was calculated by
comparison of its retention time to that of known standards. The
standard curve was generated using acetone, 2-butanone, aniline,
toluene, diphenylamine, triphenylamine, and hexachlorobenzene.
Samples were prepared at a concentration of 1 mgmLÀ1 in MeOH.
Retention times were collected in duplicate. Retention times were
adjusted for the T0, and a standard curve plotting logP vs. reten-
tion time was generated (y=0.5368e0.8991x; r2 =0.9949).
The HCtAEC cells were cultured as a monolayer in 6 well plates to
70–80% confluence and then incubated with the Gd-containing
culture medium at 100 mm for 48 h at 378C in a humidified 5%
CO2 atmosphere. The medium was removed and the cells were
washed once with PBS (1 mL). Cells were harvested with trypsin
(1 mL, 0.1% v/v) for 5 min at 378C, and then media (1 mL) was
added. Harvested cells were sedimented by centrifugation at
3000 rpm for 5 min, and then the supernatant removed. PBS
(1 mL) was added to the cell pellet, which was then pipetted until
a single cell suspension was achieved. The cells were re-sediment-
ed by centrifugation. PBS (2 mL) was added to the cell pellet,
which was then pipetted until a single cell suspension was ach-
ieved. An aliquot (100 mL) was taken for protein analysis. The re-
maining cells were sedimented by centrifugation at 3000 rpm for
5 min, then the supernatant was removed and the cell pellet was
analysed for Gd content by means of ICP-MS.
Biological Assays
In vitro cytotoxicity and cell uptake studies: Human glioblastoma
multiforme (T98G) cells were maintained as monolayers in mini-
mum essential medium supplemented with 10% fetal bovine
serum, penicillin (100 units/mL), streptomycin (100 mgmLÀ1), and l-
glutamine (2.5 mm) at 378C in a humidified 5% CO2 atmosphere.
Primary human carotid artery endothelial cells (HCtAEC) were har-
vested from normal carotid arteries and obtained commercially
(Cell Applications, Inc, SD). Where required, HCtAECs were cultured
in complete media comprising MesoEndo Cell Growth Medium
(Cell applications, SD), fetal bovine serum (10% v/v), 2 mmL-gluta-
mine, 100 units/mL penicillin, 100 mgmLÀ1 streptomycin, and
15 mgmLÀ1 endothelial cell growth supplement (Millipore, Sydney,
Australia) at 378C under 5% CO2.
Cell pellets were digested in HNO3 (0.5 mL, 69%) at 658C in
a water bath for 18 h. The digest was diluted to 10 mL with HCl
(0.1m), and then measured for Gd by means of ICP-MS. ICP-MS
was run on a PerkinElmer ELAN 6100 inductively coupled plasma
Chem. Eur. J. 2014, 20, 16602 – 16612
www.chemeurj.org
16610
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
emission mass spectrometer (ICP-MS) at the Solid State and Ele-
mental Analysis Unit (UNSW Analytical Centre). Gd uptake concen-
trations are reported as ng Gd/mg protein.
Synchrotron X-ray fluorescence imaging
T98G human glioblastoma cells were grown directly on 500 nm
thick Si3N4 windows (Silson Pty Ltd) and incubated for 24 h in the
presence of 100 mm of 3, 6, or 7 in MEM buffer (treated cells) or for
24 h with MEM buffer alone (control cells). Following incubation,
the cells were washed with PBS buffer and the cells fixed to the
window by dipping in MeOH at 788C.
Mitochondria isolation studies: Stock solutions of 3 and 6 (20 mm
in H2O) and 9 (10 mm in H2O) were prepared. Warm (378C) culture
medium was treated with the stock Gd solutions to a final concen-
tration of 100 mm. The T98G cells were cultured as a monolayer in
75 cm2 flasks to 70–80% confluence and then incubated with the
Gd-containing culture medium at 100 mm (treated) or with untreat-
ed culture medium (control) for 48 h at 378C in a humidified 5%
CO2 atmosphere. The medium was removed and the cells were
washed once with PBS (1 mL). Cells were harvested with trypsin
(1 mL, 0.1% v/v) for 5 min at 378C, and then media (1 mL) was
added. Harvested cells were sedimented by centrifugation at
3000 rpm for 5 min, and then the supernatant removed. PBS
(1 mL) was added to the cell pellet, which was then pipetted until
a single cell suspension was achieved. The cells were re-sediment-
ed by centrifugation. PBS (2 mL) was added to the cell pellet,
which was then pipetted until a single cell suspension was ach-
ieved. An aliquot (100 mL) was taken for protein analysis. The re-
maining cells were sedimented by centrifugation at 3000 rpm for
5 min.
A549 human lung cancer cells were grown directly on 500 nm
thick Si3N4 windows (Silson, Pty Ltd) and incubated for 24 h in the
presence of 100 mm of 3 or 6 in MEM buffer (treated cells) or for
24 h with MEM buffer alone (control cells). Following incubation,
the cells were fixed to the Si3N4 windows, by either one of two
methods. In the first method, the silicon nitride windows were
washed with PBS buffer and the cells fixed to the window by dip-
ping in MeOH on dry ice. This method was used for sample win-
dows of the control cells and cells treated with 3 and 6. Additional-
ly, a small number of sample windows containing cells treated
with 6 were fixed using the second fixation method. In this
method, the Si3N4 windows were washed with ammonium acetate
solution (0.2m), before flash freezing in a slurry of liquid N2/isopen-
tane. The windows were then freeze-dried overnight.
X-ray fluorescence imaging experiments were performed at the
Australian Synchrotron. A monochromatic 10.2 keV X-ray[27] beam
was focussed (to a spot size of 1 mm) using a zone plate, and the
fluorescence signal was collected using a single-element silicon
drift diode energy dispersive detector (Vortex EM, SII Nanotechnol-
ogy), oriented at 73 degrees to the incident beam for 2 s per spa-
tial point. Individual cells were located using images obtained from
an optical microscope positioned in the beamline downstream
from the sample.[10,28] Data were quantified (elemental area densi-
ties in mg/cm2) using MAPS software[29] by fitting the full fluores-
cence spectrum at every spatial point to modified Gaussians[30] and
comparing it with corresponding measurements on the thin-film
standards NBS-1832 and NBS-1833 from the National Bureau of
Standards (Gaithersburg, MD, USA).
The mitochondria of the cell pellets were isolated using a reagent-
based mitochondria isolation kit (Thermo Scientific product
number 89874). At 48C, the cell pellets were centrifuged at 12
000 rpm for 2 min and the supernatant removed. To the cell pellet
500 mL of reagent A was added, before mixing by means of vortex
and incubation on ice for 2 min. Then 10 mL of reagent B was
added before mixing by means of vortex and further incubation
on ice with intermittent vortex mixing (every minute). 500 mL of
Reagent C was added to the sample prior to centrifugation at
2900 rpm. The supernatant was transferred to a new vial, and the
pellet was retained for analysis. The supernatant was centrifuged
at 12 000 rpm for 15 min. The supernatant was removed and re-
tained as the cytosol fraction. The pellet was washed with a further
500 mL of reagent C, centrifuged at 12 000 rpm for 5 min and the
supernatant discarded and the pellet retained as the mitochondrial
fraction. These cell fractions were analysed for Gd content by
means of ICP-MS.
Acknowledgements
Cell fractions were digested in HNO3 (0.5 mL, 69%) at 658C in
a water bath for 18 h. The digest was diluted to 10 mL with HCl
(0.1m), and then measured for Gd by means of ICP-MS. ICP-MS
was run on a PerkinElmer ELAN 6100 inductively coupled plasma
emission mass spectrometer (ICP-MS) at the Solid State and Ele-
mental Analysis Unit (UNSW Analytical Centre). Gd uptake concen-
trations are reported as ng Gd/mg protein.
We gratefully acknowledge the assistance provided by Dr Ian
Luck with the NMR studies, Dr Keith Fisher and Dr Nick Pro-
schogo for collection of the ESI-MS data, Assoc. Prof. Paul Wit-
ting and Dr Xioaping Cai for providing HCtAEC cells for analy-
sis, Dr Dorothy Yu (UNSW) for ICP-MS analyses, and Dr David
Paterson and Dr Daryl Howard (Australian Synchrotron) for
beamline assistance. We are grateful to the Australian Research
Council (ARC) for funding. This research was in part undertaken
on the XFM beamline at the Australian Synchrotron, Victoria,
Australia.
Protein analyses: The bicinchoninic acid (BCA) protein assay was
used to determine protein concentration, as described previous-
ly.[26] This assay relies on the reduction of alkaline CuII by proteins.
A bovine serum albumin (BSA) protein standard curve was pre-
pared each time the assay was performed. Lysis of cells was ach-
ieved using three snap freeze-thaw cycles and cell debris was sedi-
mented by centrifugation at 13 300 rcf for 5 min. The supernatant
solution was then analysed for protein content by taking repeated
10 or 25 mL samples (N=3 or 4) of the blank, 1 mgmLÀ1 BSA pro-
tein standard (200, 400, 800, and 1000 mgmLÀ1, made up to
volume with MQ water) or Gd-treated protein samples and depos-
iting these into a 96-well plate format. Next, a freshly prepared so-
lution of commercially sourced BCA and CuSO4·5H2O (50:1, 200 mL)
was added to each well and the mixture was incubated at 378C for
60 min. Absorbance was then measured at 600 nm using a Victor3V
microplate reader (PerkinElmer) and protein was determined by
comparison with the BSA standard curve.
Keywords: gadolinium
· mitochondria · phosphonium ·
tumours · X-ray fluorescence
[1] a) J. Reedijk, Curr. Opin. Chem. Biol. 1999, 3, 236–240; b) P. Caravan,
Chem. Soc. Rev. 2006, 35, 512–523; c) Z. Zhang, S. A. Nair, T. J. McMurry,
Curr. Med. Chem. 2005, 12, 751–778.
[2] A. H. Soloway, W. Tjarks, B. A. Barnum, F.-G. Rong, R. F. Barth, I. M. Co-
dogni, J. G. Wilson, Chem. Rev. 1998, 98, 1515–1562.
[3] a) S. Sjçberg, J. Carlsson, H. Ghaneolhosseini, L. Gedda, T. Hartman, J.
Malmquist, C. Naeslund, P. Olsson, W. Tjarks, J. Neuro-Oncol. 1997, 33,
41–52; b) S. C. Mehta, D. R. Lu, Pharm. Res. 1996, 13, 344–351; c) E. L.
Chem. Eur. J. 2014, 20, 16602 – 16612
www.chemeurj.org
16611
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Crossley, E. J. Ziolkowski, J. A. Coderre, L. M. Rendina, Mini-Rev. Med.
Chem. 2007, 7, 303–313.
[4] B. H. Laster, W. C. Thomlinson, R. G. Fairchild, Radiat. Res. 1993, 133,
219–224.
2009, 50, 6457–6461; d) D. E. Morrison, F. Issa, M. Bhadbhade, L. Groe-
bler, P. K. Witting, M. Kassiou, P. J. Rutledge, L. M. Rendina, J. Biol. Inorg.
Chem. 2010, 15, 1305–1318; e) Y.-S. Kim, C.-T. Yang, J. Wang, L. Wang,
Z.-B. Li, X. Chen, S. Liu, J. Med. Chem. 2008, 51, 2971–2984; f) J. Wang,
C.-T. Yang, Y.-S. Kim, S. G. Sreerama, Q. Cao, Z.-B. Li, Z. He, X. Chen, S.
Liu, J. Med. Chem. 2007, 50, 5057–5069; g) Y. Zhou, S. Liu, Bioconjugate
Chem. 2011, 22, 1459–1472.
[5] a) W. Chen, S. C. Mehta, D. R. Lu, Adv. Drug Delivery Rev. 1997, 26, 231–
247; b) D. P. Gierga, J. C. Yanch, R. E. Shefer, Med. Phys. 2000, 27, 1685–
1692; c) J. Stepanek, Med. Phys. 2003, 30, 41–43; d) G. De Stasio, D.
Rajesh, P. Casalbore, M. J. Daniels, R. J. Erhardt, B. H. Frazer, L. M. Wiese,
K. L. Richter, B. R. Sonderegger, B. Gilbert, S. Schaub, R. J. Cannara, J. F.
Crawford, M. K. Gilles, T. Tyliszczak, J. F. Fowler, L. M. Larocca, S. P.
Howard, D. Mercanti, M. P. Mehta, R. Pallini, Neurol. Res. 2005, 27, 387–
398; e) R. C. Greenwood, C. W. Reich, H. A. Baader, H. R. Koch, D. Breitig,
O. W. B. Schult, B. Fogelberg, A. Backlin, W. Mampe, Nucl. Phys. A 1978,
304, 327–428.
[14] a) J. D. Steichen, M. J. Weiss, D. R. Elmaleh, R. L. Martuza, J. Neurosurg.
1991, 74, 116–122; b) I. Madar, J. H. Anderson, Z. Szabo, U. Scheffel, P.-
F. Kao, H. T. Ravert, R. F. Dannals, J. Nucl. Med. 1999, 40, 1180–1185.
[15] J. S. Modica-Napolitano, J. R. Aprille, Cancer Res. 1987, 47, 4361–4365.
[16] V. M. Runge, D. M. Kaufman, M. L. Wood, L. S. Adelman, S. Jacobson, Int.
J. Quantum Chem. Quantum Biol. Symp. Int. J. Rad. Appl. Instrum. B
1989, 16, 561–567.
[6] a) J.-F. Adam, A. Joubert, M.-C. Biston, A.-M. Charvet, M. Peoc’h, J.-F.
Le Bas, J. Balosso, F. Esteve, H. Elleaume, Int. J. Radiat. Oncol. Biol. Phys.
2006, 64, 603–611; b) M.-C. Biston, A. Joubert, J.-F. Adam, H. Elleaume,
S. Bohic, A.-M. Charvet, F. Esteve, N. Foray, J. Balosso, Cancer Res. 2004,
64, 2317–2323; c) J. F. Adam, M. C. Biston, J. Rousseau, C. Boudou, A. M.
Charvet, J. Balosso, F. Esteve, H. Elleaume, Phys. Med. 2008, 24, 92–97.
[7] G. De Stasio, D. Rajesh, J. M. Ford, M. J. Daniels, R. J. Erhardt, B. H.
Frazer, T. Tyliszczak, M. K. Gilles, R. L. Conhaim, S. P. Howard, J. F. Fowler,
F. Esteve, M. P. Mehta, Clin. Cancer Res. 2006, 12, 206–213.
[17] a) E. Brꢂcher, Top. Curr. Chem. 2002, 221, 103–122; b) A. E. Martell, R. J.
Motekaitis, E. T. Clarke, R. Delgado, Y. Sun, R. Ma, Supramol. Chem. 1996,
6, 353–363.
[18] C. Chen, B. H. Paw, Biochim. Biophys. Acta Mol. Cell Res. 2012, 1823,
1459–1467.
[19] D. F. Schriver, The Manipulation of Air-Sensitive Compounds, McGraw-Hill,
New York, 1969.
[20] W. L. F. Armarego, C. Chai, Purification of Laboratory Chemicals, 5th Edi-
tion, 2003.
[8] a) R. F. Martin, G. D’Cunha, M. Pardee, B. J. Allen, Int. J. Radiat. Biol.
1988, 54, 205–208; b) R. F. Martin, G. D’Cunha, M. Pardee, B. J. Allen,
Pigm. Cell Res. 1989, 2, 330–332.
[9] a) T. Goorley, H. Nikjoo, Radiat. Res. 2000, 154, 556–563; b) T. Goorley, R.
Zamenhof, H. Nikjoo, Int. J. Radiat. Biol. 2004, 80, 933–940.
[10] E. L. Crossley, J. B. Aitken, S. Vogt, H. H. Harris, L. M. Rendina, Angew.
Chem. Int. Ed. 2010, 49, 1231–1233; Angew. Chem. 2010, 122, 1253–
1255.
[11] a) J. M. Ford, W. Seiferheld, J. R. Alger, G. Wu, T. J. Endicott, M. Mehta, W.
Curran, S.-C. Phan, Int. J. Radiat. Oncol. Biol. Phys. 2007, 69, 831–838;
b) P. Carde, R. Timmerman, M. P. Mehta, C. D. Koprowski, J. Ford, R. B.
Tishler, D. Miles, R. A. Miller, M. F. Renschler, J. Clin. Oncol. 2001, 19,
2074–2083; c) D. I. Rosenthal, P. Nurenberg, C. R. Becerra, E. P. Frenkel,
D. P. Carbone, B. L. Lum, R. Miller, J. Engel, S. Young, D. Miles, M. F. Re-
nschler, Clin. Cancer Res. 1999, 5, 739–745; d) M. P. Mehta, W. R. Shapiro,
S. C. Phan, R. Gervais, C. Carrie, P. Chabot, R. A. Patchell, M. J. Glantz, L.
Recht, C. Langer, R. K. Sur, W. H. Roa, M. A. Mahe, A. Fortin, C. Nieder,
C. A. Meyers, J. A. Smith, R. A. Miller, M. F. Renschler, Int. J. Radiat. Oncol.
Biol. Phys. 2009, 73, 1069–1076; e) K. A. Bradley, I. F. Pollack, J. M. Reid,
P. C. Adamson, M. M. Ames, G. Vezina, S. Blaney, P. Ivy, T. Zhou, M. Krailo,
G. Reaman, M. P. Mehta, J. Neuro-Oncol. 2008, 10, 752–758.
[21] P. S. Athey, G. E. Kiefer, J. Org. Chem. 2002, 67, 4081–4085.
[22] D. A. Moore, Org. Synth. 2008, 85, 10–14.
[23] P. Caravan, C. T. Farrar, D. Frullano, R. Uppal, Contrast Media Mol. Imag-
ing 2009, 4, 89–100.
[24] H. B. Eldredge, M. Spiller, J. M. Chasse, M. T. Greenwod, P. Caravan,
Invest. Radiol. 2006, 41, 229–243.
[25] T. Mosmann, J. Immunol. Methods 1983, 65, 55–63.
[26] P. K. Smith, R. I. Krohn, G. T. Hermanson, A. K. Mallia, F. H. Gartner, M. D.
Provenzano, E. K. Fujimoto, N. M. Goeke, B. J. Olson, D. C. Klenk, Anal.
Biochem. 1985, 150, 76–85.
[27] D. Paterson, M. D. de Jonge, D. L. Howard, W. Lewis, J. McKinlay, A. Star-
ritt, M. Kusel, C. G. Ryan, R. Kirkham, G. Moorhead, D. P. Siddons, AIP
Conf. Proc. 2011, 1365, 219–222.
[28] a) E. A. Carter, B. S. Rayner, A. I. McLeod, L. E. Wu, C. P. Marshall, A.
Levina, J. B. Aitken, P. K. Witting, B. Lai, Z. Cai, S. Vogt, Y.-C. Lee, C.-I.
Chen, M. J. Tobin, H. H. Harris, P. A. Lay, Mol. Biosyst. 2010, 6, 1316–
1322; b) C. M. Weekley, J. B. Aitken, S. Vogt, L. A. Finney, D. J. Paterson,
M. D. de Jonge, D. L. Howard, P. K. Witting, I. F. Musgrave, H. H. Harris, J.
Am. Chem. Soc. 2011, 133, 18272–18279.
[29] S. Vogt, J. Phys. IV 2003, 104, 635–638.
[30] P. van Espen in Handbook of X-ray spectrometry: Revised and expanded,
(Eds.: R. van Grieken and A. Markowicz), Dekker, New York, 2002.
[12] D. E. Morrison, J. B. Aitken, M. D. de Jonge, J. A. Ioppolo, H. H. Harris,
L. M. Rendina, Chem. Commun. 2014, 50, 2252–2254.
[13] a) J. S. Modica-Napolitano, J. R. Aprille, Adv. Drug Delivery Rev. 2001, 49,
63–70; b) J. A. Ioppolo, J. K. Clegg, L. M. Rendina, Dalton Trans. 2007,
1982–1985; c) J. A. Ioppolo, M. Kassiou, L. M. Rendina, Tetrahedron Lett.
Received: June 24, 2014
Published online on October 16, 2014
Chem. Eur. J. 2014, 20, 16602 – 16612
www.chemeurj.org
16612
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim