Biocompatible inorganic nanoparticles for [18F]-fluoride binding with applications in PET imaging

Article abstract of DOI:10.1039/c0dt01618g

A wide selection of insoluble nanoparticulate metal salts was screened for avid binding of [¹⁸F]-fluoride. Hydroxyapatite and aluminium hydroxide nanoparticles showed particularly avid and stable binding of [ ¹⁸F]-fluoride in various biological media. The in vivo behaviour of the [¹⁸F]-labelled hydroxyapatite and aluminium hydroxide particles was determined by PET-CT imaging in mice. [¹⁸F]-labelled hydroxyapatite was stable in circulation and when trapped in various tissues (lung embolisation, subcutaneous and intramuscular), but accumulation in liver via reticuloendothelial clearance was followed by gradual degradation and release of [¹⁸F]-fluoride (over a period of 4 h) which accumulated in bone. [¹⁸F]-labelled aluminium hydroxide was also cleared to liver and spleen but degraded slightly even without liver uptake (subcutanenous and intramuscular). Both materials have properties that are an attractive basis for the design of molecular targeted PET imaging agents labelled with ¹⁸F.

Full text of DOI:10.1039/c0dt01618g

View Article Online / Journal Homepage / Table of Contents for this issue
This article is published as part of the Dalton Transactions themed issue entitled:
Radiopharmaceuticals for Imaging and Therapy
Guest Editors Stephen Faulkner (Oxford University) and Nick Long (Imperial College)
Published in issue 23, 2011 of Dalton Transactions
Image reproduced with permission of Martin W. Brechbiel
Articles in the issue include:
PERSPECTIVES:
Multimodal radio- (PET/SPECT) and fluorescence imaging agents based on metallo-radioisotopes:
current applications and prospects for development of new agents
Flora L. Thorp-Greenwood and Michael P. Coogan
Dalton Trans., 2011, DOI: 10.1039/C0DT01398F
Radiometallated peptides for molecular imaging and targeted therapy
João D. G. Correia, António Paulo, Paula D. Raposinho and Isabel Santos
Dalton Trans., 2011, DOI: 10.1039/C0DT01599G
212
Towards translation of Pb as a clinical therapeutic; getting the lead in!
Kwon Yong and Martin W. Brechbiel
Dalton Trans., 2011, DOI: 10.1039/C0DT01387K
ARTICLES:
First dinuclear Re/Tc complex as a potential bimodal Optical/SPECT molecular imaging agent
Alexandre Boulay, Marine Artigau, Yvon Coulais, Claude Picard, Béatrice Mestre-Voegtlé and Eric
Benoist
Dalton Trans., 2011, DOI: 10.1039/C0DT01397H
Synthesis, cytotoxicity and cellular uptake studies of N functionalized Re(CO) thymidine
3
3
complexes
Mark D. Bartholomä, Anthony R. Vortherms, Shawn Hillier, John Joyal, John Babich, Robert P.
Doyle and Jon Zubieta
Dalton Trans., 2011, DOI: 10.1039/C0DT01452D
Visit the Dalton Transactions website for more cutting-edge inorganic, organometallic and
bioinorganic research
www.rsc.org/dalton

Dynamic Article Links
Dalton
Transactions
Cite this: Dalton Trans., 2011, 40, 6226
www.rsc.org/dalton
PAPER
1
8
Biocompatible inorganic nanoparticles for [ F]-fluoride binding with
applications in PET imaging†
a
a
a
a
a
Maite Jauregui-Osoro, Peter A. Williamson, Arnaud Glaria, Kavitha Sunassee, Putthiporn Charoenphun,
b
a
a
Mark A. Green, Gregory E. D. Mullen and Philip J. Blower*
Received 21st November 2010, Accepted 11th February 2011
DOI: 10.1039/c0dt01618g
1
8
A wide selection of insoluble nanoparticulate metal salts was screened for avid binding of [ F]-fluoride.
Hydroxyapatite and aluminium hydroxide nanoparticles showed particularly avid and stable binding of
1
8
18
[
F]-fluoride in various biological media. The in vivo behaviour of the [ F]-labelled hydroxyapatite and
1
8
aluminium hydroxide particles was determined by PET-CT imaging in mice. [ F]-labelled
hydroxyapatite was stable in circulation and when trapped in various tissues (lung embolisation,
subcutaneous and intramuscular), but accumulation in liver via reticuloendothelial clearance was
followed by gradual degradation and release of [ F]-fluoride (over a period of 4 h) which accumulated
in bone. [ F]-labelled aluminium hydroxide was also cleared to liver and spleen but degraded slightly
even without liver uptake (subcutanenous and intramuscular). Both materials have properties that are
1
8
1
8
1
8
an attractive basis for the design of molecular targeted PET imaging agents labelled with F.
2
4–26
Introduction
been a topic of growing interest in recent years.
Methods for
derivatisation of nanoparticles (NPs) such as superparamagnetic
iron oxides (SPIO), drug-carrying micelles and liposomes, and
quantum dots for radiolabelling have been developed, most often
entailing covalent derivatisation of a silica or polymer coating on
Selective binding of fluoride ions to various solid state and soluble
materials has been a topic of growing interest in recent years
for applications including extraction and analysis of fluoride in
environmental samples, rapid trapping of [ F]-fluoride follow-
ing its manufacture in cyclotrons and in purification of [ F]-
1
–3
18
the particle surface, with prosthetic groups such as organic moi-
1
8
18 24–26
eties for covalent labelling with F
or with bifunctional chelat-
4
labelled radiopharmaceuticals for positron emission tomography,
99m
27
64
28
ing agents for metallic radionuclides such as Tc and Cu.
and most recently as inorganic binding sites for incorporation of
We envision a targeted inorganic nanoparticulate formulation in
which the nanoparticle surface is derivatised with both a targeting
moiety such as a protein or peptide, and a detectable probe such
as a positron or gamma emitting radionuclide and/or a beta- or
alpha-emitting therapeutic radionuclide, and in which the surface
derivatisation exploits the intrinsic binding properties of the
inorganic nanoparticle itself, rather thananencapsulating polymer
or membrane. Our purpose here is to identify inorganic materials
that have sufficient intrinsic affinity for the radiolabel that stable,
direct labelling of the inorganic nanoparticle material itself can
be achieved, for radionuclide imaging and therapy applications
in vivo. Such materials would have to be biocompatible and/or
biodegradable and non-toxic. Their surface chemistry should be
readily amenable to derivatisation with targeting molecules such
as proteins, peptides, carbohydrates and nucleic acids, and the
radiolabelling step should be achievable rapidly and efficiently
under mild conditions compatible with sensitive biomolecules and
1
8
5–12
[
F]-fluoride into biomolecules.
Various solid state materials
4
13,14
have been used including anion exchange resins, alumina,
hydroxyapatite,
lar materials containing silicon,
and complexes of aluminium
by the well-known strong adsorption of fluoride ions onto
hydroxyapatite,
competition from, for example, phosphate ions, phosphonates,
polyphosphates (e.g. nucleotides) and hydroxide ions, we are
interested in inorganic nanoparticulate materials that may be used
as targeted carriers of [ F]-fluoride ions for in vivo PET imaging
applications.
Use of nanoparticulate and microparticulate materials for
radionuclide imaging, most recently in combination with other
imaging modalities (especially MRI and optical imaging), has
1
5–17
18
19,20
porous titania, aluminosilicates,
molecu-
and phosphorus,
1
0–12,21
2,8,22
9
boron
5–7
3
and zirconium. Stimulated
15,17,23
which occurs even in vivo in the face of
1
8
1
8
a
with the minimum of chemical manipulation. F was chosen as an
initial radionuclide because of its ubiquitous availability resulting
from the widespread clinical use of [ F]-fluorodeoxyglucose, and
the excellent physical properties of its emissions for PET imaging
half life 110 min and high-yield low-energy positron). It is by far
Division of Imaging Sciences, King’s College London, St. Thomas’ Hospital,
London, UK SE1 7EH. E-mail: Philip.Blower@kcl.ac.uk; Tel: 0207-188-
9
513
18
b
Department of Physics, King’s College London, The Strand, London, WC2R
2
LS
29
(
†
1
Electronic supplementary information (ESI) available. See DOI:
0.1039/c0dt01618g
the most readily available and widely used of the positron emitting
6
226 | Dalton Trans., 2011, 40, 6226–6237
This journal is © The Royal Society of Chemistry 2011

Table 1 Particulate materials used in this study, their origin and size
ogy and fluoride binding affinity. These new formulations are
summarised in Table 2.
a
(
stated by the manufacturer except size and standard deviation measured
by dynamic light scattering, with standard deviation)
Material
Origin
Size
Experimental section
TM
a
Al(OH)
3
-Alhydrogel
Brenntag Biosector
1211 ± 64 nm
Synthesis of nanoparticles
(
Frederikssund, Denmark)
Al
Ag
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
Sigma–Aldrich
< 75 mm
<100 nm
90–210 nm
<160 nm
<200 nm
<50 nm
All chemicals were obtained from commercial sources as analytical
reagents and used without further purification unless other-
wise indicated. MeO-PEG-COOH (a-methoxy-w-carboxylic acid
poly(ethylene glycol), Mw 2000) was purchased from Iris-Biotech
Bi
CaO
Ca (OH)(PO
Co
CuO
Er
Eu
Fe
In
MgO
SnO
TiO
Yb
ZrO
2 3
O
[
5
4 3 x
) ] – HA1
3
O
4
GmbH. Calcium nitrate tetrahydrate (Ca(NO
nium phosphate dibasic ((NH HPO ), cetyltrimethylammonium
bromide (CTAB, C19 42NBr), cetyltrimethylammonium chlo-
ride solution 25% in H O (CTAC, C19 42NCl), triethanolamine
TEA, C ) and tetraethyl orthosilicate (TEOS, C Si)
3
)
2
·4H
2
O), ammo-
<50 nm
4
)
2
4
2
O
3
<100 nm
<150 nm
<50 nm
<100 nm
<50 nm
<100 nm
<100 nm
<100 nm
<100 nm
2
O
3
H
3
O
4
O
H
2
2
3
(
6
H
15NO
3
8
H
20
O
4
were purchased from Sigma–Aldrich. Deionised water (Type I,
18.2 MX·cm) was obtained from an ELGA Purelab Option-Q
system.
2
2
2
O
3
2
Preparation of HA2 (calcined hydroxyapatite). To a solution
of 0.5 M Ca(NO
3
)
2
·4H
2
O in H
2
O (30 ml) was added 0.3 M
(
NH HPO in H
4
)
2
4
2
O (30 ml) drop-wise. The solutions were
radionuclides, and consequently the most likely to be of benefit to
patients.
adjusted to pH 10 with 28% ammonium hydroxide prior to
mixing. The reaction mixture was stirred for 4 h at 60 C and
◦
In this paper we describe the results of an initial survey of the
left to stand overnight. The precipitate was washed four times
with ethanol (20 ml) with centrifugation at 8,000 rpm (7513 g) to
remove the washings. The product was dried at 60 C overnight to
1
8
[
F]-fluoride binding properties of inorganic materials that were
considered likely to have an affinity for fluoride. Those that showed
◦
1
8
the most promise in terms of efficacy of [ F]-fluoride binding and
biocompatibility were subjected to further evaluation in vitro to
determine the affinity, capacity and stability of the interaction
in biological media, and in vivo by PET-CT imaging in mice
to determine the biodistribution and fate of the radiolabel. We
included in our survey materials already known to have a degree
produce “untreated” hydroxyapatite. This material was sintered in
◦
a furnace at 550 C for 6 h yielding a fine white powder.
Preparation of HA3 (hydrothermally treated hydroxyapatite).
“
Untreated” hydroxyapatite was prepared as described above for
-
1
HA2. The HA powder was then suspended in water (15 mg ml ),
placed in an acid digestion bomb (Parr) and heated in an oven at
16
of biocompatibility, including hydroxyapatite (already used for
◦
2
00 C for 24 h followed by freeze drying.
in vivo applications such as bone grafting), aluminium hydroxides
3
0,31
(
(
used safely in vivo as adjuvants in vaccines),
calcium phosphate
Preparation of HA4 (no thermal treatment). In a typical
32
also used as adjuvants ) and silica (for which derivatisation
chemistry is well-established and clinical trials of modified
experiment, 20 ml aqueous solutions of Ca(NO
(NH HPO were freshly prepared with a concentration of
0.75 M. The pH of these solutions was adjusted to 9 by addition of
small amounts of a concentrated NH OH solution (Fluka). Then
in a 50 ml double-neck rounded flask, 5.00 ml of the calcium nitrate
3
)
2
·4H
2
O and
33
4
)
2
4
34
silica nanoparticles have recently been approved). A list of the
particulate and nanoparticulate materials used, and their origin,
is provided in Table 1. Hydroxyapatite showed particular promise
in preliminary studies and for this reason further hydroxyapatite
nanoparticle formulations were synthesised and evaluated to
determined the effect of post-synthesis treatments (calcination,
hydrothermal treatment and surface derivatisation) on morphol-
4
solution were added to 5.52 ml of H
2
O and 0.75 ml of a CTAB
◦
aqueous solution (0.1 M). When the temperature reached 80 C,
2.98 ml (ratio Ca/P was thus 1.68) of the ammonium phosphate
solution were added in one portion, producing a white suspension
Table 2 Hydroxyapatite and silica nanoparticles synthesised in this work. Size and standard deviation was measured by transmission electron microscopy
TEM) and dynamic light scattering (DLS)
(
Designation
Post-synthesis treatment
Size by TEM (nm)
Size by DLS (nm)
Mes-SiO
HA2
HA3
2
none
calcination
hydrothermal
none
Alendronate, 0.1 ml
Alendronate, 0.3 ml
Alendronate, 0.5 ml
MeO-PEG-COOH, 0.1 ml
MeO-PEG-COOH, 0.3 ml
MeO-PEG-COOH, 0.5 ml
70.9 ± 8.6
—
41.6 ± 1.8
2784 ± 158
829 ± 135
2046 ± 470 (main peak)
169 ± 33 (main peak)
107.5 ± 6.2 ¥ 25.8 ± 0.9
132.0 ± 30.3 ¥ 12.6 ± 2.2
HA4
HA4Ale-0.1
HA4Ale-0.3
HA4Ale-0.5
HA4PEG-0.1
HA4PEG-0.3
HA4PEG-0.5
132.0 ± 30.3 ¥ 12.6 ± 2.2
214 ± 68
This journal is © The Royal Society of Chemistry 2011
Dalton Trans., 2011, 40, 6226–6237 | 6227

-
1
TM
which was magnetically stirred for 2 h. The white precipitate was
then collected by centrifugation at 1500 rpm (1409 g) for 5 min and
suspended in water at 1 mg ml for Alhydrogel , HA4, HA4Ale,
-
1
HA4PEG and 0.2 mg ml for HA1, HA2 and HA3.
washed twice with H O removing the washings by centrifugation,
2
to remove CTAB. Finally the solid was washed with ethanol and
dried under vacuum at 80 C.
◦
18
Initial screening for [ F]-fluoride binding
1
8
18
[
F]-fluoride was produced by irradiation of [ O]-water (97
Functionalisation of HA4 with Alendronate (Ale) or
MeO-PEG-COOH (PEG). Alendronate (4-amino-1-hydroxy-1-
phosphonobutyl phosphonic acid, monosodium) was synthesised
atom%, Isochem Ltd., Hook, UK) with 11 MeV protons from
a CTI RDS 112 cyclotron (beam current 30 mA). This solution
1
8
18
in [ O]-water was used without further purification. The F is
35
as described in the literature. Aqueous solutions of the ligands
Ale or MeO-PEG-COOH) were prepared with a concentration
nominally in the form of no-carrier-added fluoride ions, although
(
1
9
typically F is also present in traces. Initial studies were performed
using hydroxyapatite nanoparticles, and focused on testing the
optimal medium in which to carry out the radiolabelling and
of 0.05 M. In the case of MeO-PEG-COOH, the solution was
made in 0.1 M 2-(N-morpholino)ethanesulfonic acid (MES)
buffer (pH 6.15) to avoid the dissolution of the NPs at lower
pH. The previously synthesised NPs were suspended in water at
1
8
optimal incubation times required. [ F]-fluoride solution (100 ml,
-
1
6
MBq) was added to a 1 mg ml suspension of HA in water (1
-
1
a concentration of 10 mg ml and 100 ml of this suspension was
mixed with 100, 300 or 500 ml of the ligand solution. The final
volume of the solution was adjusted to 1 ml by addition of water
so that the final concentration of ligand in these mixtures was
respectively 5, 15 or 25 mM. The mixture was left to react overnight
ml), and the mixture was incubated at room temperature with
continuous shaking for 5, 10, 20 and 60 min. Since labelling
efficiency had plateaued by 5 min in these experiments, a
standard incubation time of 5 min was adopted for screening
of all materials. Each sample was then tested in triplicate, and
nanoparticle-free standards were also tested in order to determine
baseline supernatant activity and non-specific binding to tubes.
To measure binding of radioactivity to particles, the mixture was
then centrifuged (5 min at 10000 rpm, 9391 g) and an aliquot
of the supernatant (100 ml) was collected from each sample.
The radioactivity in this aliquot was measured using a gamma
counter and expressed as the percentage of the total added to the
sample. The % radioactivity associated with the nanoparticles was
calculated using the following equation:
◦
under agitation, and heated for one hour at 80 C before washing
the NPs twice with water and centrifuging (1500 rpm, 1409 g, for
5
min). The resulting materials were named HA4Ale-0.1, -0.3, -0.5
and HA4PEG-0.1, -0.3, -0.5 respectively, according to the volume
of ligand solution added.
Preparation of Mes-SiO
2
nanoparticles. Mesoporous silica
NPs were obtained by a modification of the synthesis described
36
by Bein et al. Briefly 64.0 ml of deionised water was mixed
with 10.5 ml of MeOH (0.259 mol), 10.4 ml of CTAC (7.8
mmol) and 12.37 ml of TEA (93 mmol). Twenty ml of this
Labeling efficiency (%) = [1 - (activity in sample
aliquot/activity in standard aliquot)] ¥ 100
◦
(
1)
solution was stirred for 30 min at 60 C and 1.45 ml of TEOS
(
6.5 mmol) was added dropwise over 3 min. The molar ratios
were TEOS/CTAC/TEA/H O–MeOH: 1/0.24/3/109/5.5. The
2
To check that small quantities of material (e.g. in quantities
that might be injected in vivo) have sufficient capacity for efficient
solution was stirred for a further 2 h before addition of 100 ml
of methanol which induced precipitation of the particles. The
suspension was centrifuged at 2000 rpm (1878 g) for 10 min.
1
8
binding of no-carrier-added [ F]-fluoride, different concentra-
1
8
tions of nanoparticles were labelled with [ F]-fluoride. For both
hydroxyapatite and aluminium hydroxide, these concentrations
◦
The pellet was washed with methanol and dried at 60 C under
-
1
vacuum. The template molecules were then removed by washing
the particles with an acidic methanol solution (21.9 ml of 5 M HCl
in 100 ml of methanol) at room temperature overnight.
varied between 0.025 and 1.0 mg ml .
To provide an objective measure of fluoride binding affinity
allowing semi-quantitative comparison between materials, we
adopted particle concentration at half-maximal % binding after
1
8
5
min incubation of no-carrier-added [ F]-fluoride with suspen-
Characterisation of particulate materials
sions containing different concentrations of particles; this is equiv-
alent to a dissociation constant when expressed in suitable units.
The data were fitted to eqn (2), a sigmoid function from which the
maximum binding (Bmax) and the particle concentration at half-
FT-IR studies were performed using either a Perkin–Elmer
Spectrum RXI FT-IR spectrometer or a Perkin–Elmer Spectrum
1
00 spectrometer equipped with a Universal Attenuated Total
Reflectance (UATR) sampling accessory. Transmission electron
microscopy (TEM) was performed either with an FEI Tecnai T12
instrument at 120 kV or a FEI Tecnai T20 instrument with a
maximal % binding (K ) were obtained, using Solver in Microsoft
Excel.
d
n
n
n
LaB
6
filament operating at 200 kV. The samples were prepared by
Y = Bmax(X )/(Kd + X )
(2)
evaporation of a drop of the aqueous colloidal suspensions onto
a carbon-coated copper grid (Agar Scientific 200 mesh). X-ray
powder diffraction (XRD) was recorded on a PANalytical X’Pert
Pro Multi Purpose diffractometer. Zeta potential measurements
where X is nanoparticle concentration and Y is % bound
radioactivity.
To study the fluoride loading capacity of these materials, a
◦
18
were made 25 C using a Delsa Nano C particle analyser.
radiolabelling study was carried out with F as above but in
-
1
Particles were suspended in water (1 mg ml ) at pH7. Dynamic
light scattering (DLS) measurements were performed with a
the presence of varying concentrations of cold (non-radioactive)
sodium fluoride (X) at different particle concentrations (0.1, 0.5
◦
-1
Beckman Coulter DelsaNano C instrument at 25 C with particles
and 1 mg ml ).
6
228 | Dalton Trans., 2011, 40, 6226–6237
This journal is © The Royal Society of Chemistry 2011

Competition studies
from each sample. The % bound radioactivity was calculated as
before using eqn (1).
One hundred microlitres of solutions containing different con-
centrations of the potential competitor/inhibitor (see ESI Table
Kinetic stability in serum. HA and aluminium hydroxide
-
1
18
-
1
particles suspended in water (1 mg ml ) were labelled with [ F]-
fluoride (100 ml) as described above, in triplicate. The mixture was
centrifuged (5 min at 10000 rpm, 9391 g), the supernatant was
discarded and % radioactivity associated with the nanoparticles
was measured as described above (eqn (3)). The pellets were then
S1 for full list) were added to 1 ml of a 0.5 mg ml (unless
otherwise stated) suspension of nanoparticles in water. Aqueous
1
8
[
F]-fluoride (100 ml) was then added to each sample and the
mixture was incubated at room temperature with continuous
shaking for 5 min. The samples were then centrifuged for 5 min
at 10000 rpm (9391 g), and an aliquot of the supernatant (100 ml)
was collected from each sample. The % bound radioactivity was
determined as described above (eqn (1)).
◦
resuspended in human serum (1 ml) and incubated at 37 C with
continuous shaking for up to 40 min. At different time intervals
the samples were centrifuged and the activity associated with the
nanoparticles was measured, and the pellets resuspended in the
serum for continued incubation.
1
8
Binding in serum. [ F]-fluoride (100 ml) was added to a
1
0
.5 mg ml- suspension of particles in human serum (Sigma
◦
Aldrich), and the mixture was incubated at 37 C with continuous
shaking for 6 h. Each sample was tested in triplicate, and standard
samples without nanoparticles were also tested as a control.
The mixture was centrifuged (5 min at 10000 rpm, 9391 g).
Under these conditions, serum proteins do not sediment while the
nanoparticles do so efficiently. Aliquots of the supernatant (100 ml)
were collected from each sample at different time intervals. The %
radioactivity bound to particles was determined as above (eqn (1)).
To identify possible components of serum that may be responsible
for the reduction or delay in fluoride binding, HA particles were
In vitro macrophage uptake studies
Media and reagents were obtained from PAA (Yeovil, UK).
The J774.1 cell line (macrophage, derived from a tumour in a
female BALB/c mouse, kindly provided by Dr Helen Collins,
King’s College London) was grown in a humidified incubator
◦
at 37 C with 5% CO
2
in Dulbecco’s Modified Eagle Medium
(DMEM) supplemented with 10% foetal calf serum (FCS),
1
0 mM HEPES, 1 mM sodium pyruvate, 2 mM L-glutamine and
penicillin/streptomycin.
1
8
-
1
To measure uptake of [ F]-labelled nanoparticles in
macrophages, J774.1 cells were seeded in a 24 well plate at a
incubated with human serum albumin (4 mg ml , cf. typically
present at 40 mg ml-1 in serum) and with fetuin, a protein that
5
-
1
density of 5 ¥ 10 cells per well, incubated in a humidified
binds strongly to HA and is present in serum (0.1 mg ml , cf.
◦
typically 0.5 mg ml-1 in serum).
incubator at 37 C with 5% CO for 24 h and washed twice in
2
1
8
Hank’s balanced salt solution (HBSS). Next, [ F]-fluoride labelled
particles in HBSS (0.5 ml of a 20 mg ml particle suspension)
were added and cells were incubated at 37 C for up to 3 h.
-
1
Kinetic stability studies
◦
Kinetic stability in water. Each sample was tested in triplicate.
At different time intervals, cells were washed twice with 500 ml
cold HBSS (the washings contain labelled particles and soluble
radioactivity but not cells) and extracted with 500 mL 1 M NaOH
1
8
-1
Aqueous [ F]-fluoride (100 ml, 6 MBq) was added to a 1 mg ml
suspension (1 ml) of nanoparticles in water, and the mixture
was incubated at room temperature with continuous shaking for
(this extract contains all cell-bound radioactivity) for 10 min.
5
min. The mixture was centrifuged (5 min at 10000 rpm, 9391
The cell medium/washings and cell extracts were counted in a
gamma counter (Wallac 1282-001 Compugamma CS). Percent
accumulation in cells was plotted versus incubation time. The
same experiments were carried out in a cell-free 24 well plate,
g), the supernatant was discarded and the percent radioactivity
associated with the nanoparticles was calculated using eqn (3):
Labelling efficiency (%) = [(Activity of nanoparticles)/(input
(
3)
18
activity)] ¥ 100
and uptake into cells of soluble [ F]-fluoride without particles
were also studied as controls.
The particles were then resuspended in water (1 ml), and the
mixture was shaken for 5 min and centrifuged. The supernatant
was discarded and the % particle-bound activity measured again.
This washing step was repeated 3 times.
Phagocytic uptake of hydroxyapatite and aluminium hydroxide
nanoparticles was demonstrated by transmission electron mi-
croscopy as follows. J774.1 cells were seeded in 6 well plates
5
at a density of 5 ¥ 10 cells per well in the above medium
◦
Kinetic stability in competitive media. Each sample was tested
(2 ml) and incubated in a humidified incubator at 37 C with
in triplicate, and compared with standard particle-free solutions.
5% CO
2
for 48 h. HA1 and HA2 or Al(OH)
3
particles (2 ml,
1
8
-1
[
F]-fluoride (100 ml, 6 MBq) was added to a suspension of
100 mg ml ) in phosphate buffered saline (PBS) were added to
the cells and incubated at 37 C, 5% CO
-
1
◦
nanoparticles (1 mg ml ) in water (1 ml) and the mixture was
incubated at room temperature with continuous shaking for 5 min.
The mixture was then centrifuged (5 min at 10000 rpm, 9391 g)
and an aliquot of the supernatant (100 ml) was collected from each
sample. The % bound radioactivity was determined as described
above using eqn (1). The pellet was then resuspended and a 0.01 M
solution of the chosen competitor/inhibitor (100 ml; see ESI† Table
S1for list) was then added to each sample. Samples were then
2
for 2–4 h. Cells were
washed three times with HBSS (2 ml) and detached from the
wells by scraping. The cells were pelleted by centrifugation at
1500 rpm (1409 g) for 5 min and resuspended in piperazine-N,N¢-
bis(2-ethanesulfonic acid) buffer (PIPES) 0.1 M (1 ml). Cells were
then fixed by adding 4% glutaraldehyde (1 ml) and incubating
for 1 h before washing twice and resuspending in 0.1 M PIPES
buffer (1 mL), post fixed in 1% aqueous osmium tetroxide and
dehydrated through a graded series of ethanol and embedded
in TAAB resin. Sections 70 nm thick were cut using a Leica
ultracut E ultramicrotome (Leica, Milton Keynes, UK), mounted
◦
incubated at 25 C with continuous shaking for up to 120 min.
Every 30 min, the mixture was centrifuged (5 min at 10000 rpm,
9
391 g) and an aliquot of the supernatant (100 ml) was collected
This journal is © The Royal Society of Chemistry 2011
Dalton Trans., 2011, 40, 6226–6237 | 6229

Fig. 1 Transmission electron micrographs of hydroxyapatite used in this study: A) commercially available (HA1); B) calcined (HA2); C) hydrothermally
treated (HA3); D) without any thermal treatment (HA4).
-
1
-1
-1
-1
on 200 mesh Cu grids (Agar Scientific, Stansted, UK), and stained
with 1.5% uranyl acetate in 50% ethanol and 0.15% lead citrate
before viewing on a Tecnai T12 electron microscope (FEI, the
Netherlands). Images were captured using a Gatan Bioscan 792
camera (Gatan, Abingdon Oxon, UK).
1030 cm , 963 cm , 603 cm and 563 cm . Additional bands
-
1
-1
2-
at 1737 cm and 1366 cm corresponding to CO
observed on sample HA2 and can be related to carbon dioxide
3
were also
37
insertion during sintering of the NPs under air. The thermally
treated samples (HA2 and HA3) have a smaller size and narrower
size dispersion than HA1 (see Table 2 and Fig. 1). In the case
of HA4, the mild conditions of pressure and temperature led
to thin needles. Zeta potential measurements on samples HA1,
HA2 and HA3 indicated that the particles were negatively charged
In vivo studies
Animal studies were carried out in accordance with UK Research
Councils’ and Medical Research Charities’ guidelines on Respon-
sibility in the Use of Animals in Bioscience Research, under a
UK Home Office licence. Female BALB/c mice (aged 7–12 weeks,
(
-10.7 ± 0.6, -15.8 ± 0.2 and -6.0 ± 0.6 mV respectively). This was
not sufficient to prevent aggregation of the NPs in solution, as
demonstrated by the DLS measurements. Moreover, despite the
use of surfactants (HA4), aggregates were still observable on TEM
images. In order to reduce the interactions between the particles,
functionalisation with a ligand bearing a stronger anchoring group
than the ammonium moiety of the CTAB molecules was used.
Two sets of samples (HA4-Ale and HA4-PEG) were synthe-
sised using bisphosphonate (Alendronate) or carboxyl-terminated
polyethylene glycol (MeO-PEG-COOH). These ligands are likely
to bind to the calcium ions at the surface of the HA NPs via
the phosphonate or the carboxyl groups respectively. The least
aggregation was obtained for the pegylated sample HA4-PEG (see
in vivo experiments in the next section). It has been reported that
the Alendronate can have two different grafting modes due to the
phosphonate and amine functional groups and might thus act as a
20.5 ± 3.4 g) were purchased from Harlan Laboratories, UK. For
HA, mice received i.v. (tail vein) injections of approximately 5 MBq
1
8
-1
of [ F]-fluoride-labelled nanoparticles (1 mg ml ), prepared as
describedabovein50 mlsaline(Fresenius Kabi). For Al(OH) , mice
received either i.m. and s.c injections (in the right and left thigh
3
1
8
respectively), or an i.v. (tail vein) injection of 5 MBq [ F]-fluoride
-
1
labelled particles (1 mg ml ) in saline (50 ml). With the mouse
under isofluorane anaesthesia in a Minerve imaging chamber,
PET/CT scans were acquired from 30 min to 4 h post-injection
(
p.i.) using a NanoPET/CT scanner (Bioscan, Paris, France) with
PET acquisition time 1,800 s, coincidence relation: 1–3; image
reconstruction: ordered subset expectation maximisation (OSEM)
with single-slice rebinning (SSRB) 2-D line of response (LOR),
energy window: 400–600 keV, filter: Ram-Lak cut-off 1, number of
iterations/subsets: 8/6. Animals were culled at 4 h p.i. and tissues
explanted, blotted dry, weighed and counted on a gamma counter
38
linker between theparticles. Infrared analysis after washing of the
-
1
HA4-PEG particles confirms the grafting of the PEG (2887 cm ,
-
1
-1
-1
1
735 cm , 1342 cm and 842 cm in addition to bands originating
(
the carcass activity was measured with an ionisation chamber
from the HA4 mineral; see ESI). No change in the size of the
particles due to the surface modification was observed by TEM.
The mesoporous silica particles were obtained according to the
cross-calibrated with the gamma counter).
Results
36
protocol published by Bein and co-workers. After synthesis of the
NPs the organic template (CTAC) entrapped inside the pores was
removed by acid washing, while monitoring by FT-IR (ESI). These
NPs were mainly spherical but the formation of larger particles due
to the coupling of the smaller ones was also observed, in agreement
Characterisation of particles
Three different samples (HA2, HA3, HA4) have been synthesised
and compared to the commercially available hydroxyapatite NPs
36
with earlier work. The pores can be clearly seen on the TEM
images (not shown). The XRD diffractogram (ESI) only exhibits
(
HA1). Differences in the experimental protocol led to a variation
of both the size and the shape of the particles (see Table 2).
Aggregates of almost spherical nanocrystals (HA2), rod-like
◦
one broad peak centered at 22 confirming that the material is
amorphous or of insufficient density for the diffraction of the
(
HA3) or needle-like (HA4) NPs were obtained (Fig. 1). Charac-
X-rays.
terisation of the different samples confirmed that hydroxyapatite
had been obtained whatever the experimental conditions. Indeed,
all the XRD diffractograms were identical to that of commercial
sample (HA1) and conformed to the ICSD-22059 reference (ESI).
Moreover, the FT-IR spectra of the all the samples exhibited bands
TM
The aluminium hydroxide (Al(OH) , Alhydrogel ) suspension
was used as received. Dilutions in deionised water were made
to allow the use of concentrations from 0.1 to 1 mg mL . The
suspension was characterised using DLS giving a size of 1211 ±
3
-
1
39
3
-
-1
6
4 nm in agreement with the value reported by the manufacturer.
characteristic of hydroxyapatite due to PO
4
ions at 1092 cm ,
6
230 | Dalton Trans., 2011, 40, 6226–6237
This journal is © The Royal Society of Chemistry 2011

18
18
Fig. 2 Top: Initial screening of various inorganic materials for [ F]-fluoride binding. Bottom: comparison of [ F]-fluoride binding efficiency of
hydroxyapatite nanoparticles prepared by different methods (HA1: commercial; HA2: calcined; HA3: hydrothermal; HA4: no treatment) including
increasing surface modification with Alendronate (HA4Ale) and PEG (HA4PEG). The charts show labelling efficiency (percent bound radioactivity)
18
-1
after incubation of [ F]-fluoride solution (100 mL, 6 MBq) with a 1 mg ml suspension of the particles in water (1 ml) at room temperature for 5 min.
1
8
Initial screening for [ F]-fluoride binding
high considering its particle size is greater, and therefore its surface
area lower, than the equivalent mass of the other nanoparticles.
Based on these initial results, hydroxyapatite and aluminium
hydroxide particles, which not only showed the highest binding
but are highly biocompatible, were selected for further evaluation
as fluoride binding materials with a view towards in vivo use.
The effect of the method of preparation and surface derivati-
sation was determined for a range of hydroxyapatite NPs. The
method of preparation (untreated, hydrothermal, calcined) made
little difference to the labelling efficiency but surface derivatisation
with PEG or Alendronate significantly impaired either the rate or
extent (it is not possible to determine which with the present data)
of fluoride binding, in line with the concentration of the respective
ligand used to derivatise the NPs (Fig. 2).
Preliminary binding studies using hydroxyapatite (HA1, HA2
1
8
and HA3) with [ F]-fluoride in water established that binding
equilibrium was generally reached well within 5 min even with
samples containing less than 1 mg particles in 1 ml suspension.
Prolonged incubation for up to 60 min did not increase the percent
of radioactivity associated with the NPs. Similar conditions (5 min
incubation in water with particles suspended at 1 mg ml ) were
therefore adopted for all subsequent screening. The materials in
Table 1 were chosen based on the expected behaviour of fluoride
ions taking into account their known preference for binding to
hard metal ions, with additional materials included for comparison
and contrast. Iron oxide was of interest because of its use as a
MR contrast agent material, and silica because of its ubiquitous
application as a basis for nanoparticles for biological application.
Results of the initial survey are shown in Fig. 2.
-
1
18
Binding efficiency of [ F]-fluoride to HA and Al(OH)
3
in water
Hydroxyapatite showed very high % binding while surprisingly,
other related calcium salts did not. Aluminium hydroxide also
showed very high % binding. Magnesium oxide, aluminium and
erbium oxide showed a high affinity for fluoride under these
conditions, while zirconium oxide showed relatively low affinity
despite the well-known affinity of soluble zirconium complexes for
To provide an objective measure of fluoride binding affinity
allowing semi-quantitative comparison between materials, we
adopted particle concentration at half-maximal % binding after
5 min incubation (it is assumed on the basis of the above data that
the system is close to equilibrium by this time) of no-carrier-added
1
8
[ F]-fluoride with suspensions containing different concentrations
3
fluoride. The binding efficiency of aluminium (particles of which
of particles (HA1, HA2, HA3, Al(OH) ). Examples of the binding
3
are expected to have an aluminium oxide surface) was remarkably
curves are shown in Fig. 3 and the data are summarised in
This journal is © The Royal Society of Chemistry 2011
Dalton Trans., 2011, 40, 6226–6237 | 6231

Table 3 Binding characteristics of HA and Al(OH)
3
nanoparticles for
is the nanoparticle concentration at half-
18
binding to [ F]-fluoride. K
d
maximal % binding; Bmax is the maximal % binding. Al(OH)
3
has the
highest effective affinity as indicated by the low K
d
value
-1
2
Material
B
max (%)
K
d
(mg mL )
n
r
HA2
HA1
HA3
97.90
86.46
97.77
92.96
0.062
0.075
0.075
0.028
1.35
0.9980
0.9728
0.9961
0.9986
1.65
1.92
1.00
Al(OH)
3
18
Fig. 4 Effect of carrier fluoride concentration on binding of [ F]-fluoride
18
to HA1 (top) and Al(OH)
3
(bottom) particles. [ F]-fluoride solution
(
100 mL, 6 MBq) was incubated with a suspension of particles (1.0, 0.5
-1
and 0.1 mg ml ) in aqueous solutions (1 ml) of sodium fluoride of varying
concentrations at room temperature for 5 min. Corresponding data for
HA2 can be found in ESI.†
approximately 10 mmol fluoride/mg NPs and about two orders
of magnitude higher than that of HA NPs. The relatively high
capacity of the Al(OH)
see TEM images in ESI). At these levels of affinity and capacity,
it is clear that for both HA and Al(OH) NPs, at quantities of
3
may be due to its permeable gel-like nature
(
3
NPs (e.g. ꢀ0.1 mg) likely to be used in vivo no-carrier-added
1
8
fluoride binding to nanoparticles will not be limited by saturation
Fig. 3 Binding efficiency of no-carrier-added [ F]-fluoride to hydroxya-
patite HA2 (top) and Al(OH) (bottom) as a function of nanoparticle
concentration. Chart shows labelling efficiency (percent bound radioac-
1
9
or competition from background levels of F fluoride ions.
3
18
18
tivity) after incubation of [ F]-fluoride solution (100 mL, 6 MBq) with a
suspension of HA in water (1 ml) at room temperature for 5 min.
Binding of [ F]-fluoride to HA and Al(OH) in the presence of
3
competitors
Table 3. Efficient binding was observed even at concentrations
The modification and application of the particles in vivo requires
below 1 mg ml- in all cases and Al(OH)
1
had the highest affinity
the presence of solutes other than water, which may inhibit
3
1
8
(
lowest K
d
). The different methods of preparation of HA had little
binding. Therefore the binding of [ F]-fluoride to HA and
Al(OH) was tested in the presence of several potential competitive
effect on these parameters.
3
To study the fluoride loading capacity of these materials, a
study was carried out with varying concentrations of cold (non-
radioactive) fluoride at different particle concentrations (graphs
inhibitors, which included simple anions (chloride, citrate, phos-
phate) and other compounds that may be used in the synthesis
or derivatisation of HA nanoparticles. These included sodium
hexametaphosphate (HMP), cetyltrimethylammonium bromide
(CTAB) and the aminobisphosphonate Alendronate. Among the
substances tested, HMP and citrate had the most marked effect
on fluoride binding; however, surprisingly high concentration
of citrate and HMP was required to reduce fluoride binding
significantly (see ESI). Chloride, alendronate and CTAB did not
show significant inhibition at any of the concentrations employed
below). The Al(OH)
3
particles had markedly higher fluoride
binding capacity than hydroxyapatite (HA1) (Fig. 4). Aluminium
1
8
hydroxide particles showed higher [ F]-fluoride binding than the
-
1
same concentration (mg ml ) of hydroxyapatite at all carrier
fluoride concentrations used. From these data we estimate that the
binding capacity of the HA1 and HA2 particles for fluoride is in
-
1
the order of 0.1 mmol mg . The binding capacity for Al(OH)
3
was
6
232 | Dalton Trans., 2011, 40, 6226–6237
This journal is © The Royal Society of Chemistry 2011

(
up to at least 0.1 M). Likewise phosphate buffered saline and
human serum also resulted in minimal radiolabel loss over a period
of at least 2 h. However, radiolabelled Al(OH) lost radiolabel
steadily to the extent of about 50% by 30 min when incubated
in serum; this may have been accelerated by dissolution of the
particles due to the presence of citrate in the serum.
tris buffer had only a modest inhibitory effect. The lack of
effect of aminobisphosphonate is remarkable considering that
bisphosphonates have very high binding affinity for the surface
3
40
of hydroxyapatite.
If these materials are to be used in vivo, components in the
biological milieu may also have competitive or inhibitory effects
In vitro cell studies
1
8
on fluoride binding. Therefore, we monitored the stability of [ F]-
fluoride-labelled particles added to human serum (see below),
and the labelling process conducted in human serum. Fluoride
binding to hydroxyapatite was markedly slower in human serum
than in water, although given time the % binding levels achieved
are comparable (Fig. 5). It should be noted that the human serum
used in our experiments contains added citrate (10 mM), which is
a known competitor for fluoride binding (see ESI). The effect of
One of the key features of the in vivo behaviour of nanoparticulate
materials is clearance by the reticulendothelial system in which
phagocytic cells — macrophages, Kupffer cells etc. — are the
main players. In some applications this is an advantage while in
others it must be protected against. We have therefore studied
the interaction or the hydroxyapatite and Al(OH)
3
particles with
cultured macrophages (J774.1) in vitro using a combination of
1
8
radiolabelling and electron microscopy. [ F]-labelled HA4 and
serum on labelling of Al(OH) was more marked and % binding
3
1
8
Al(OH)
3
, and [ F]-fluoride as a control, were incubated with
was severely reduced in serum. The tissue culture medium used in
the macrophage experiments had similar effects.
1
8
J774.1 cells. [ F]-fluoride alone was not significantly taken up
by macrophages (<0.2%), whereas the [ F]-fluoride-labelled HA
1
8
showed progressive association with cells reaching over 70%
uptake by 3 h (Fig. 7). Uptake of labelled Al(OH) was much less
3
marked during this period, although given the more rapid release
of radioactivity from these particles (Fig. 6) it is possible that
there is rapid uptake followed by rapid release of free fluoride. The
1
8
Fig. 5 [ F]-fluoride binding of HA4 and Al(OH)
3
in serum as a function
18
of incubation time. [ F]-fluoride (100 ml) was incubated with a suspension
-1
◦
of particles (0.5 mg ml ) in human serum (1 ml) at 37 C with continuous
shaking for 6 h. Binding is delayed compared to labelling in water, which
is complete within 5 min.
Two serum proteins were tested to determine whether they could
play a part in this inhibition of binding: fetuin, because it is
known to bind strongly to calcium salts, and albumin because
of its high concentration in blood. Neither protein caused a
significant reduction in binding (although it should be noted that
the concentration of each protein used in the experiment was about
1
0-fold lower than physiological). In the presence of both albumin
and fetuin simultaneously, each at about half the physiological
1
8
concentration, there was likewise no significant reduction in
binding.
F
1
8
Kinetic stability of [ F]-labelled HA and Al(OH)
3
For the radiolabelled materials to be used in vivo in molecular
imaging, the NP-fluoride bond must resist major dissociation in
biological media for a period compatible with imaging. Repeated
washing of labelled HA and Al(OH) with water did not remove a
3
significant fraction of the radioactivity, and even incubation with
high concentrations of bisphosphonates (Alendronate, known
to have very high affinity for hydroxyapatite) failed to release
significant radiolabel over a 2 h period. Incubation of HA in
18
Fig. 6 Rate of loss of radiolabel from [ F]-fluoride-labelled HA (top)
3
and Al(OH) (bottom) particles during incubation in serum. Charts show
% activity remaining associated with pre-labelled, washed particles after
incubation, with initial activity defined as 100%.
This journal is © The Royal Society of Chemistry 2011
Dalton Trans., 2011, 40, 6226–6237 | 6233

phagocytic uptake of HA and Al(OH)
3
particles was confirmed
In vivo studies
The observation that the association of [ F]-fluoride with hydrox-
by electron microscopy (Fig. 7) which showed the presence of
aggregates of particles within endosomes in the cytoplasm of the
cells.
1
8
yapatite particles (and to a lesser extent Al(OH) ) is kinetically
3
stable in a variety of challenging media including human serum
shows that the in vivo fate of the nanoparticles, and the in vivo
stability of the radiolabel, can be monitored in vivo by PET
1
8
imaging. The latter is possible because [ F]-fluoride released from
nanoparticles in vivo would show a distinctive pattern of uptake
in the joints of the skeleton. Labelled particles were administered
by the sub-cutaneous, intramuscular and intravenous routes. The
1
8
observed biodistribution of [ F]-labelled hydroxyapatite nanopar-
ticles in normal mice was dependant on the route of administration
and the state of aggregation of the particles. PET and PET-CT
scans are shown in Fig. 8 and Fig. 9, and biodistribution data
1
8
are provided in ESI. [ F]-fluoride-labelled HA4 injected by the
subcutaneous and intramuscular route remained at the injection
site and negligible translocation of radioactivity to other sites was
1
8
seen during the 4 h study period (see ESI). [ F]-HA2 particles,
which are largely aggregated, are trapped by embolisation in the
lungs after intravenous injection. A few particles small enough to
escape this biological filtration mechanism are taken up in liver or
excreted renally. There is no significant uptake in the joints of the
skeleton even after 4 h, indicating that particles trapped thus in the
1
8
lung capillary bed are stable towards loss of radiolabel. [ F]-HA4
is a less aggregated form of HA and showed correspondingly less
lung trapping and more uptake in liver and spleen, consistent with
the observed phagocytic uptake shown in Fig. 7. In this instance,
over the four hours of the study, radioactivity was gradually
released and accumulated in the skeleton (Fig. 8), suggesting that
once taken up by the reticuloendothelial system, degradation and
release of fluoride takes place. Particles that are further protected
from aggregation by PEGylation (HA4-PEG-0.1) show even less
trapping in lung and more in liver and spleen at 30 min, which is
largely translocated to the joints subsequently.
1
8
[
F]-labelled Al(OH)
3
particles were injected intramuscularly
(
right thigh, Fig. 9) and subcutaneously (left thigh) in the same
animal. Almost all radioactivity remained at the sites of injection
during the entire scanning period of 4 h. There was evidence of only
slight translocation of radioactivity to the joints by 4 h, indicating
that the radiolabelled particles are moderately stable in vivo over
this time period. When the particles were injected intravenously,
there was no lung trapping and radioactivity was cleared from
blood rapidly by uptake in the liver and spleen and by renal
excretion. Translocation of the liver and spleen activity to the joints
was faster than for i.v. injected HA and faster than for s.c. and i.m.
injected Al(OH) , suggesting that the degradation and release of
3
soluble radioactivity occurs in mainly liver and spleen. The large
amount of radioactivity appearing rapidly in the bladder suggests
that the particles may be broken down into soluble fragments
small enough for renal filtration, but not releasing free fluoride
since uptake in the joints is low at the 30 min scan. Extensive joint
uptake, suggesting release of free fluoride, occurs later than the
renal excretion.
18
18
18
3
Fig. 7 Uptake of [ F]-HA4, [ F]-Al(OH) and [ F]-fluoride (con-
trol) in macrophages (top), and transmission electron micrographs of
macrophages before (centre) and after (bottom) incubation with HA1,
showing HA1 particles sequestered within phagosomes in the macrophages
Discussion and conclusions
(
bottom). Corresponding images showing Al(OH)
3
uptake are shown in
The design of molecular imaging contrast agents based on nano-
particulate carriers requires several contributing components:
ESI.†
6
234 | Dalton Trans., 2011, 40, 6226–6237
This journal is © The Royal Society of Chemistry 2011

Fig. 9 PET and PET-CT scans of mice 30 min (left) and 4 h (right)
18
post injection of [ F]-fluoride-labelled Al(OH)
3
. Top: after intramuscular
(
right thigh) and subcutaneous (left thigh) injection (CT scan omitted
in order to reveal skeletal uptake more clearly); Bottom: after intravenous
injection. Intramuscular and subcutaneous administrations led to minimal
translocation of radioactivity from the injection site and to the joints,
whereas intravenous injection led to early uptake in liver, spleen and
bladder with later translocation from liver and spleen to the joints.
aggregation, control of reticuloendothelial clearance and
molecular targeting), and a means of attaching the imaging
probe (gamma or positron emitting radionuclide, gadolinium
contrast agent, fluorescent tag, etc.). At the present early stage of
development of this field, the nanoparticles reported so far have
been chosen from a very restricted range of materials-typically
silica and gold nanoparticles, cadmium-based quantum dots, and
superparamagnetic iron oxide nanoparticles. In this work we have
begun to extend the choice of nanoparticulate material with the
aim of developing systems that are amenable to use with PET
imaging whilst minimising the complexity of the radiochemical
Fig. 8 PET-CT scans (maximum intensity projection) of mice after i.v.
18
injection of [ F]-fluoride labelled hydroxyapatite, at 30 min (left) and 4 h
18
(
right) post injection. Top: [ F]-labelled HA2 (highly aggregated); Middle:
HA4 (moderately aggregated); Bottom: PEG-modified HA4-PEG-0.1.
There is a trend towards less lung uptake and more liver and spleen uptake
as aggregation is reduced. Higher liver and spleen uptake is associated with
more pronounced transfer of radioactivity to the joints with time.
1
8
synthesis required. By selecting [ F]-fluoride as the radiolabel,
we reduce the radiochemistry to the simplest level possible.
This introduces a requirement for a nanoparticulate material
the nanoparticle itself, a means of derivatising it in a stable
manner to control its biodistribution (for prevention of
This journal is © The Royal Society of Chemistry 2011
Dalton Trans., 2011, 40, 6226–6237 | 6235

1
8
that can be quickly and efficiently labelled with [ F]-fluoride
in mild aqueous conditions and without need for complex
purification steps. We screened a number of likely inorganic
materials for this property, at the same time keeping in mind the
need for a high degree of biocompatibility and low toxicity. Two
materials emerged satisfying both requirements: hydroxyapatite
and aluminium hydroxide. The factors affecting the binding
of fluoride to these particles and the stability of the bond
once formed were studied under biological conditions. The
radiolabelling of hydroxyapatite proved to be robust, resistant
to a variety of potential inhibitors, and adequate for in vivo
imaging applications. The particles are prone to aggregation
but this can be minimised by surface modification with
alendronate or PEG while still retaining adequate capacity for
radiolabelling. The labelling is stable in vivo unless the particles
are taken up by the reticuloendothelial system, whereupon
degradation and release of fluoride takes place over a period of
several hours. It has been shown previously that hydroxyapatite
particles are readily surface-modified by bisphosphonates (as
seen in this case with alendronate), providing a means of
attaching targeting functionality or other radionuclides using
Acknowledgements
We thank Ms Kelly Rausch (NIH, Bethesda, USA) for providing
a sample of Alhydrogel . This work was funded by the Centre
TM
of Excellence in Medical Engineering funded by the Wellcome
Trust and EPSRC under grants (WT 088641/Z/09/Z). MJ-O
was supported by an EPSRC postdoctoral fellowship at the Life
Sciences Interface. PAW was supported by a Harris studentship.
We thank the Wellcome Trust for an equipment grant for purchase
of the PET-CT scanner.
References
1
2
M. Cametti and K. Rissanen, Chem. Commun., 2009, 2809.
T. W. Hudnall, C. W. Chiu and F. P. Gabbai, Acc. Chem. Res., 2009, 42,
3
88.
3 Y. Takahashi, D. A. Pacheco Tanaka, H. Matsunaga and T. M. Suzuki,
J. Chem. Soc. Perkin Trans., 2002, 759.
4
D. J. Schlyer, M. A. Bastos, D. Alexoff and A. P. Wolf, Int. J. Radiat.
Appl. Instrum., Part A, 1990, 41, 531.
5 W. J. McBride, C. A. D’Souza, R. M. Sharkey, H. Karacay, E. A. Rossi,
C.-H. Chang and D. M. Goldenberg, Bioconjugate Chem., 2010, 21,
1
331.
6
7
P. Laverman, W. J. McBride, R. M. Sharkey, A. Eek, L. Joosten, W. J. G.
Oyen, D. M. Goldenberg and O. C. Boerman, J. Nucl. Med., 2010, 51,
454.
W. J. McBride, R. M. Sharkey, H. Karacay, C. A. D’Souza, E. A. Rossi,
P. Laverman, C.-H. Chang, O. C. Boerman and D. M. Goldenberg,
J. Nucl. Med., 2009, 50, 991.
40,41
bisphosphonate conjugates of radionuclides,
proteins and
peptides (unpublished work). We have also shown here that
they can be readily radiolabelled simply by brief incubation
1
8
with [ F]-fluoride in aqueous media, and that the radiolabel
has adequate stability for in vivo imaging. Hydroxyapatite
nanoparticles represent a promising basis for assembly of targeted
8 T. W. Hudnall, T.-P. Lin and F. P. Gabba ´ı , J. Fluorine Chem., 2010, 131,
1182.
9 A. R. Studenov, M. J. Adam, J. S. Wilson and T. J. Ruth, J. Labelled
1
8
PET molecular imaging agents for labelling with [ F]-fluoride.
Aluminium hydroxide gel nanoparticles share the property of
Compd. Radiopharm., 2005, 48, 497.
1
1
0 L. Mu, A. Hohne, P. A. Schubiger, S. M. Ametamey, K. Graham, J. E.
Cyr, L. Dinkelborg, T. Stellfeld, A. Srinivasan, U. Voigtmann and U.
Klar, Angew. Chem., Int. Ed., 2008, 47, 4922.
1
8
highly efficient and convenient radiolabelling with [ F]-fluoride,
and they too are highly biocompatible (the particles used here
are commercially available for use as adjuvants in vaccines).
They also have an extraordinary capacity to sequester a high
payload of fluoride per nanoparticle. Fluoride is known to be
1 P. Bohna, A. Deyine, R. Azzouz, L. Bailly, C. Fiol-Petit, L. Bischoff,
C. Fruit, F. Marsais and P. Vera, Nucl. Med. Biol., 2009, 36,
8
95.
12 R. Schirrmacher, G. Bradtmoller, E. Schirrmacher, O. Thews, J.
Tillmanns, T. Siessmeier, H. G. Buchholz, P. Bartenstein, B. Wangler,
C. M. Niemeyer and K. Jurkschat, Angew. Chem., Int. Ed., 2006, 45,
42
capable of occupying the hydroxide sites in hydroxyapatite and
we speculate that hydroxide ions may be displaced by fluoride
at the crystal surface; alternatively or additionally hydrogen
fluoride may be bound by hydrogen bonding or dipolar bonds
at the crystal surface, since the experimental conditions are not
basic. Similarly, in aluminium hydroxide, fluoride may be bound
directly to aluminium ions by displacement of hydroxide, given
6
047.
13 S. K. Nandy and M. G. R. Rajan, J. Radioanal. Nucl. Chem., 2010, 286,
241.
1
4 M. Jauregui-Osoro, K. Sunassee, A. J. Weeks, D. J. Berry, R. L. Paul,
M. Cleij, J. Banga, M. J. O’Doherty, P. K. Marsden, S. E. M. Clarke,
J. R. Ballinger, I. Szanda, S. Y. Cheng and P. J. Blower, Eur. J. Nucl.
Med. Mol. Imaging, 2010, 37, 2108.
6,7
15 V. E. Badillo-Almaraz, J. Armando Flores, H. Arriola, F. A. L o´ pez and
its known ability to coordinate strongly to aluminium ions.
L. Ruiz-Ramirez, J. Radioanal. Nucl. Chem., 2007, 271, 741.
6 M. Motskin, D. M. Wright, K. Muller, N. Kyle, T. G. Gard, A. E. Porter
However, they are subject to more rapid degradation in the
biological milieu than hydroxyapatite, and this degradation needs
to be curtailed by further modification before use in imaging
applications, except for very short time scale studies. Because of
the simplicity of the radiolabelling, which eliminates the need
for costly robotics for complex automated synthesis which often
1
and J. N. Skepper, Biomaterials, 2009, 30, 3307.
17 J. F. Volker, H. Carpenter Hodge, H. J. Wilson and S. N. van Voorhis,
J. Biol. Chem., 1940, 134, 543.
8 N. L. Ho, T. Ishihara, S. Ueshima, H. Nishiguchi and Y. Takita,
1
J. Colloid Interface Sci., 2004, 272, 399.
19 S. Samatya, U. Yuksel, M. Yuksel and N. Kabay, Sep. Sci. Technol.,
2007, 42, 2033.
1
8
limits the wider application of F-labelled radiopharmaceuticals,
both hydroxyapatite and aluminium hydroxide provide
2
2
2
0 H. M. Kao and Y. C. Liao, J. Phys. Chem. C, 2007, 111,
a
4
495.
promising basis for development of biocompatible, low-toxicity
nanoparticulate contrast agents for multimodality molecular
imaging in vivo. Further work is required to develop approaches
to modify the surface to attach a range of biological targeting
molecules, modifying aggregation, reticulendothelial uptake and
providing molecular targeting. For molecular targeting purposes,
hydroxyapatite nanoparticles can be surface-modified with
bisphosphonate bioconjugates of peptides and proteins; synthesis
of such bioconjugates will be described in future publications.
1 A. R. Bassindale, M. Pourny, P. G. Taylor, M. B. Hursthouse and M. E.
Light, Angew. Chem., Int. Ed., 2003, 42, 3488.
2 R. Ting, C. Harwig, U. auf dem Keller, S. McCormick, P. Austin, C. M.
Overall, M. J. Adam, T. J. Ruth and D. M. Perrin, J. Am. Chem. Soc.,
2
008, 130, 12045.
23 L. E. L. Hammari, A. Laghzizila, P. Barboux, K. Lahlil and A. Saoiabi,
J. Hazard. Mater., 2004, 114, 41.
2
2
4 N. K. Devaraj, E. J. Keliher, G. M. Thurber, M. Nahrendorf and R.
Weissleder, Bioconjugate Chem., 2009, 20, 397.
5 M. J. Welch, C. J. Hawker and K. L. Wooley, J. Nucl. Med., 2009, 50,
1743.
6
236 | Dalton Trans., 2011, 40, 6226–6237
This journal is © The Royal Society of Chemistry 2011

2
2
2
6 J. B. Matson and R. H. Grubbs, J. Am. Chem. Soc., 2008, 130,
731.
7 K. J. Dormer, V. Awasthi, W. Galbraith, R. D. Kopke, K. Chen and R.
Wassel, J. Biomed. Nanotechnol., 2008, 4, 174.
8 B. R. Jarrett, B. Gustafsson, D. L. Kukis and A. Y. Louie, Bioconjugate
Chem., 2008, 19, 1496.
9 D. Le Bars, J. Fluorine Chem., 2006, 127, 1488.
34 A. Burns, P. Sengupta, T. Zedayko, B. Baird and U. Wiesner, Small,
2006, 2, 723.
6
35 US Pat., 7411087, 2008.
36 K. M o¨ ller, J. Kobler and T. Bein, Adv. Funct. Mater., 2007, 17, 605.
37 A. Slosarczyk, Z. Paszkiewicz and C. Paluszkiewicz, J. Mol. Struct.,
2005, 744–747, 657.
38 S. Mukherjee, C. C. Huang, F. Guerra, K. Wang and E. Oldfield, J. Am.
Chem. Soc., 2009, 131, 8374.
2
3
0 V. Aimanianda, J. Haensler, S. Lacroix-Desmazes, S. V. Kaveri and J.
Bayry, Trends Pharmacol. Sci., 2009, 30, 287.
39 E. B. Lindblad, Immunol. Cell Biol., 2004, 82, 497.
40 R. Torres, Martin de Rosales, C. Finucane, J. Foster, S. J. Mather and
P. J. Blower, Bioconjugate Chem., 2010, 21, 811.
41 R. T. M. de Rosales, C. Finucane, S. J. Mather and P. J. Blower, Chem.
Commun., 2009, 4847.
3
3
1 S. L. Hem and H. HogenEsch, Expert Rev. Vaccines, 2007, 6, 685.
2 D. Jiang, G. S. Premachandra, C. Johnston and S. L. Hem, Vaccine,
2
004, 23, 693.
3 Z. C. Xing, Y. M. Chang and I. K. Kang, Sci. Tech. Advanced Mater.,
010, 11, 1.
3
2
42 N. H. de Leeuw, J. Mater. Chem., 2010, 20, 5376.
This journal is © The Royal Society of Chemistry 2011
Dalton Trans., 2011, 40, 6226–6237 | 6237