Lanthanide containing compounds for therapeutic care in bone resorption disorders

Article abstract of DOI:10.1039/b705123a

Lanthanide ions, Ln(iii), are known functional mimics of Ca(ii) ions and have been shown to affect the bone remodeling cycle. Exploiting this disruption to the bone remodeling cycle has potential for the treatment of bone density disorders, such as osteoporosis. In an effort to find new orally active agents for these disorders, a series of Ln(iii) containing complexes incorporating small, non-toxic, bidentate pyrone and pyridinone ligands have been synthesized and characterized (LnL₃, Ln = La, Eu, Gd, Tb, Yb, L = 3-oxy-2-methyl-4-pyrone (ma^-), 3-oxy-2-ethyl-4-pyrone (ema ^-), 3-oxy-1,2-dimethyl-4-pyridinone (dpp^-) and 3-oxy-2-methyl-4(1H)-pyridinone (mpp^-)). Preliminary biological analysis included cytotoxicity, cell uptake and bidirectional transport studies in Caco-2 cells and in vitro hydroxyapatite (HA) binding studies. The proportion of intact compounds bound to HA was calculated based on determination of Ln(iii) concentration by ICP-MS and by UV-vis spectrophotometric assay of the proligand in solution. The LnL₃ species were found to have IC ₅₀ values at least 6 times greater than that of cisplatin, ≥ 98% HA-binding capacity, and permeability coefficients in the moderate range. La(dpp)₃ was ascertained to be the lead compound for the treatment of bone density disorders with the highest percentage cell uptake of 9.07 ± 2.33% and the highest preliminary P_app value of 3.54 ± 2.86 × 10^-6 cm s^-1 compared to the other LnL₃ complexes tested. This journal is The Royal Society of Chemistry.

Full text of DOI:10.1039/b705123a

View Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/dalton | Dalton Transactions
Lanthanide containing compounds for therapeutic care in bone resorption
disorders†‡§
Cheri A. Barta,a Kristina Sachs-Barrable,b Jessica Jia,b Katherine H. Thompson,a Kishor M. Wasan*b and
Chris Orvig*a
Received 4th April 2007, Accepted 31st May 2007
First published as an Advance Article on the web 11th September 2007
DOI: 10.1039/b705123a
Lanthanide ions, Ln(III), are known functional mimics of Ca(II) ions and have been shown to affect the
bone remodeling cycle. Exploiting this disruption to the bone remodeling cycle has potential for the
treatment of bone density disorders, such as osteoporosis. In an effort to find new orally active agents
for these disorders, a series of Ln(III) containing complexes incorporating small, non-toxic, bidentate
pyrone and pyridinone ligands have been synthesized and characterized (LnL3, Ln = La, Eu, Gd, Tb,
Yb, L = 3-oxy-2-methyl-4-pyrone (ma−), 3-oxy-2-ethyl-4-pyrone (ema−), 3-oxy-1,2-dimethyl-
4-pyridinone (dpp−) and 3-oxy-2-methyl-4(1H)-pyridinone (mpp−)). Preliminary biological analysis
included cytotoxicity, cell uptake and bidirectional transport studies in Caco-2 cells and in vitro
hydroxyapatite (HA) binding studies. The proportion of intact compounds bound to HA was
calculated based on determination of Ln(III) concentration by ICP-MS and by UV-vis spectrophoto-
metric assay of the proligand in solution. The LnL3 species were found to have IC50 values at least
6 times greater than that of cisplatin, ≥ 98% HA-binding capacity, and permeability coefficients in the
moderate range. La(dpp)3 was ascertained to be the lead compound for the treatment of bone density
disorders with the highest percentage cell uptake of 9.07 2.33% and the highest preliminary Papp value
of 3.54 2.86 × 10−6 cm s−1 compared to the other LnL3 complexes tested.
the existing treatments is low due to adverse gastrointestinal side
Introduction
effects and, in the case of calcitonin, high costs.2,4,5 A new class
Bone density disorders, including osteoporosis, affect 1 in 4 women
of osteoporosis drugs, including oral strontium ranelate, stimulate
and 1 in 8 men over age 50 in North America.1 As the population
ages, these diseases are incurring substantial annual health care
costs escalating into billions of dollars. Osteoporosis is charac-
terized by low bone mineral density that leads to enhanced bone
fragility and a consequent risk of low-impact bone fractures.2,3
The low bone mineral density is a result of an imbalance between
bone resorption and bone formation. Normally, building and
absorption of bone is a tightly regulated cycle wherein the
bone matrix is manufactured by osteoblast cells and removed
by osteoclast cells. Either increased activity of osteoclasts or
decreased bone formation by osteoblasts leads to microarchitec-
tural deterioration of bone tissue. Many contributing factors are
known to influence the pathogenesis of the disease with the most
prominent being inadequate calcium uptake.2 Few therapeutic
agents exist currently, either for prevention or for amelioration
of these serious diseases.2,4,5 In addition, patient compliance with
osteoblast proliferation and inhibit osteoclast activity,6,7 however,
uncertainty regarding the potential toxicity of chronic strontium
accumulation in bone may limit the utility of this product.8
Low doses of Ln(III) have been shown to act similarly to
strontium ranelate.9–11 Lanthanides (Ln), the fourteen elements
from cerium to lutetium (Ce–Lu, Z = 58–71), are known
for their therapeutic and diagnostic applications as agents for
magnetic resonance imaging,12 cancer,13 luminescent labeling of
biomolecules,14 and radiotherapy.15–17 Lanthanum (La, Z = 57),
although technically not a Ln, has many of the same character-
istics of lanthanides and will be considered herein as a Ln. In
vivo, Ln(III), a functional mimic of Ca(II),15 has been found to
exchange with Ca(II) in bone9 and modify the bone remodeling
cycle by stimulating osteoblast proliferation10,11 and impeding
bone resorption by inhibiting osteoclast differentiation.11 Based
on this evidence, lanthanum carbonate (La2(CO3)3) has been
proposed as a potential preventative measure for post-menopausal
osteoporosis, however, gastrointestinal upset is a known negative
side effect of this treatment.9,11
La2(CO3)3 is currently being used to treat hyperphosphatemia
under the trade name of FosrenolTM. Unfortunately, the extremely
low bioavailability (<0.0007%) of FosrenolTM requires that high
doses of elemental La(III) be administered to control phosphate
levels. The high doses of La(III) lead to adverse gastrointestinal
(GI) tract side effects, with consequent poor patient compliance.9
Adjustments to the ligand structure around the Ln(III) ions
have the potential to increase the oral bioavailability of Ln(III)
aMedicinal Inorganic Chemistry Group, Department of Chemistry, Univer-
sity of British Columbia, 2036 Main Mall, Vancouver, British Columbia,
V6T 1Z1, Canada. E-mail: orvig@chem.ubc.ca
bFaculty of Pharmaceutical Sciences, University of British Columbia,
2146 East Mall, Vancouver, BC, V6T 1Z3, Canada. E-mail: kwasan@
interchange.ubc.ca
† Based on the presentation given at Dalton Discussion No. 10, 3rd–5th
September 2007, University of Durham, Durham, UK.
‡ The HTML version of this article has been enhanced with colour images.
§ Electronic supplementary information (ESI) available: Table of IR data
1
and H NMR example of metal binding (Hma and La(ma)3 spectrum).
See DOI: 10.1039/b705123a
This journal is
The Royal Society of Chemistry 2007
Dalton Trans., 2007, 5019–5030 | 5019
©

View Online
for the treatment of bone density disorders while decreasing
unwanted side effects. Hydroxypyrones and hydroxypyridinones,
related classes of small, bidentate, O,O proligands, exhibit high
bioavailability and favorable toxicity profiles. Consisting of a 6-
membered O or N heterocycle, respectively, these proligands form
thermodynamically stable five membered chelate rings using the
hard hydroxyl O atom and the neighboring ketone functionality.
Upon deprotonation within the physiological pH range, both
hydroxypyrones and hydroxypyridinones form neutral, stable
(mostly tris) complexes with trivalent cations including Fe3+,
Al3+, Ga3+, In3+,18 and Ln(III),19–21 as well as the actinides22 and
several divalent ions including Zn2+, Ru2+, [VO]2+, and [MoO2]2+.23
A handful of these metal complexes and their corresponding
hydroxypyrone and hydroxypyridinone proligands have been
extensively investigated for a myriad of potential biological
applications including iron overload disorders,24,25 aluminium
toxicity,26,27 orally active cancer treatments,28 radiotherapeutic
agents and use in SPECT imaging,29–31 restoration of iron balance
in anemia patients,32–34 insulin-enhancing agents,35 and Gd MRI
contrast agents.23,36
Specific proligands of interest include maltol (Hma, 3-hydroxy-
2-methyl-4-pyrone), its analogue ethylmaltol (Hema, 3-hydroxy-
2-ethyl-4-pyrone), deferiprone (Hdpp, 3-hydroxy-1,2-dimethyl-4-
pyridinone), and 3-hydroxy-2-methyl-4(1H)-pyridinone (Hmpp)
(Scheme 1).37 Hmpp, the only non-commercially available pro-
ligand used in this study, is synthesized in moderate yields
by ammonolysis of a benzyl-protected maltol followed by acid
deprotection.38 Hydroxypyridinones are known to be significantly
stronger chelators than are their hydroxypyrone analogues allow-
ing the stability of the complexes to be tuned by switching the
ligand or the metal to obtain desired properties.25
cinoma cells with intestinal cell-like properties (Caco-2 cells),41–44
and hydroxyapatite binding assays.
Results and discussion
Metal complexes
The neutral LnL3 where L = ma−, ema−, mpp−, or dpp− complexes
were synthesized using simple procedures (Scheme 1).39–41,45–48
Due to the tendency of Ln(III) ions to hydrolyse easily with
increasing pH, extreme care was taken when adjusting the pH
of the solution. The complexes precipitated from the respective
reaction mixtures to afford white or off-white air stable powders
in moderate to high yields, with the exception of the Eu(III)
complexes which precipitated as yellow powders. Attempts to
recrystallize the complexes (in hot MeOH or EtOH) resulted,
unfortunately, in decomposition of the product and hydrolysis
of the Ln(III) ions. Column chromatography was useless for
purification of the compounds due to their low solubilities.
However, washing the precipitated solid thoroughly with cold
solvents to remove unreacted ligand, additional Ln(NO3)3 salts or
other side products yielded complexes pure by elemental analysis
(Table 1). The compounds were thoroughly characterized by
elemental analysis (EA), positive ion electrospray ionization-mass
spectrometry (+ESI-MS), IR, and NMR (LaL3). IR, EA and
+ESI-MS results are all consistent with the formation of neutral
LnL3 species.
The IR spectra for all the complexes were dominated by peaks
corresponding to the ligand (see Table S1 in ESI§);39,49,50 however,
the IR spectra for LnL3 were nearly superimposable within each
ligand set (<10 cm−1 variations between Ln(III) ions with the same
ligand, attributed to mass differences amongst the metal ions),
indicating that the Ln(ma)3, Ln(ema)3, Ln(mpp)3, and Ln(dpp)3,
regardless of the Ln(III) ion, are isostructural in the solid state.
The mC=O stretching bands were shifted 40–50 cm−1 to lower energy
with respect to the proligand, evincing normal coordination of the
metal via the ketonic oxygen and the oxygen of the deprotonated
phenol group. The broad band for mOH disappeared upon complex-
ation indicating complete deprotonation of the phenol OH group.
New bands at 3400 and 3200 cm−1 appeared for the pyrone and
pyridinone complexes, respectively, due to lattice and coordinated
water molecules. Resonance of the coordinated aromatic bidentate
ligands resulted in a hyperchromic shift observed for mC–O due
to the increasing double bond character of the C–O linkage.
Scheme 1 3-Hydroxy-4-pyrones and 3-hydroxy-4-pyridinones of interest
and their Ln complexes (Ln = La, Eu, Gd, Tb, Yb).
Previously, a handful of Ln(III) complexes using the same pyrone
and pyridinone ligands have been investigated, including Ln(ma)3
Ln = Pr, Nd, Sm, Gd, Dy, Yb, Gd(ema)3 and Gd(dpp)3;39–41
however, these complexes have neither been fully characterized,
nor have their potential utilities as bone density agents been
investigated.
Herein, the synthesis and complete characterization of the
neutral Ln(III) tris(bidentate ligand) complexes using Hma, Hema,
Hmpp, and Hdpp as the ligands is presented. La(III), Gd(III),
and Yb(III) ions were investigated, with selection of specific
lanthanides based on their known medicinal applications and for
size comparison. The luminescent properties of the Eu(III) and
Dy(III) complexes were also studied. In addition, the potential
of these compounds as therapeutic agents for the treatment of
bone resorption disorders was assessed in cytotoxicity studies,
comparative bifunctional transport studies in human colon car-
In addition, the mC= vibrational modes were bathochromically
C
shifted indicating a decrease in the double bond character of the
aromatic ring due to donation of p electrons to the chelate ring.
This trend is similarly seen in Ln(ONO) systems.51 The absence
of mNO3 (∼1382) in the IR spectra indicated the absence of any
counterions in the complexes and the appearance of new peaks
between 400–500 cm−1 for mM–O stretching frequencies confirmed
the complexation of ligands to their respective metal ions.52,53
The elemental analyses for all the lanthanide complexes con-
firmed the empirical formula LnL3·xH2O. Hydrated species have
been observed with the previously synthesized Ln(ma)3 and
Gd(ema)3 systems, in which at least one water molecule was
associated with the metal compounds (Table 1).39,40 In addition, the
N–H functionality of the mpp− ligand is a potential hydrogen bond
donor that may increase hydration numbers for its complexes.
5020 | Dalton Trans., 2007, 5019–5030
This journal is
The Royal Society of Chemistry 2007
©

View Online
Table 1 Elemental analysis, % yield and +ESI-MS results for all prepared LnL3 complexes
%C
found (calc)
%H
found (calc)
%N
found (calc)
Complex
% yield
[NaLnL3]+
La(ma)3
42.04 (42.40)
39.65 (39.57)
39.27 (39.28)
39.15 (38.87)
39.43 (39.43)
45.34 (45.74)
41.02 (41.05)
43.89 (44.18)
42.44 (42.61)
42.72 (42.32)
36.33 (36.06)
38.58 (38.63)
40.82 (40.49)
38.11 (37.80)
42.93 (42.78)
45.58 (45.95)
43.16 (43.18)
42.77 (42.76)
41.39 (41.67)
39.32 (39.25)
2.94 (3.20)
3.14 (3.32)
3.11 (3.11)
3.10 (3.04)
2.76 (3.10)
3.80 (4.14)
3.94 (4.27)
3.68 (3.97)
3.90 (3.91)
3.58 (3.89)
4.94 (4.71)
3.96 (3.68)
3.43 (3.82)
3.91 (3.71)
4.12 (3.89)
4.37 (4.74)
4.48 (4.82)
4.44 (4.67)
4.63 (4.88)
4.71 (4.54)
40
62
59
54
36
68
64
53
56
85
52
46
38
56
64
40
38
76
54
68
537
551
556
557
571
579
593
598
599
614
534
547
553
554
568
576
590
595
596
610
Eu(ma)3·H2O
Gd(ma)3·H2O
Tb(ma)3·H2O
Yb(ma)3
La(ema)3
Eu(ema)3·2.5H2O
Gd(ema)3
Tb(ema)3·H2O
Yb(ema)3
La(mpp)3·2H2O
Eu(mpp)3·2H2O
Gd(mpp)3
6.05 (6.12)
7.50 (7.87)
7.93 (8.06)
7.41 (7.62)
7.15 (6.92)
7.59 (7.82)
7.19 (7.18)
7.13 (7.04)
6.90 (7.03)
6.55 (6.82)
Tb(mpp)3·2H2O
Yb(mpp)3
La(dpp)3
Eu(dpp)3·H2O
Gd(dpp)3·H2O
Tb(dpp)3·2H2O
Yb(dpp)3·3H2O
Electrospray ionization mass spectrometry in the positive ion
mode gave diagnostic mass spectra with the expected isotopic
patterns. The most intense peak corresponded to the sodium
adduct of the parent ion peak [NaLnL3]+ (Table 1). Peaks
corresponding to [LnL2]+ species generated by the gas-phase loss
of one ligand were observed in addition to peaks with the formula
[Ln2L5]+, as commonly reported for other trivalent metal-tris(O,O
bidentate ligand) analogues.47,54
The literature crystal structure of Pr(ma)3 revealed three
bidentate deprotonated coordinated ligands and two bound water
molecules arranged in an irregular geometry around the Pr(III)
ion,41 matching perfectly the analytical data obtained in this study.
In addition, the C–O bond distances of the donor oxygens were
not significantly different, suggesting that the negative charge
is delocalized upon coordination, as would be expected in the
reported complexes.
The NMR spectra of LaL3 were also useful in elucidating
the structure of the metal complexes. All diamagnetic LaL3
1
were studied in solution by H NMR (Table 2 and Fig. 1S in
ESI§). Only La(ma)3 and La(ema)3 were analyzed by 13C NMR
due to limited solubility of the pyridinone complexes (Table 3).
The absence of ligand peaks and the relative shifts in both
1
the H and 13C NMR signals conclusively indicated binding of
the ligand to the La3+ ion. Both ring hydrogens, Ha and Hb,
for the La(III) pyrone and pyridinone metal complexes shifted
upfield. Similar upfield shifts were reported in the corresponding
Ln(III) complexes with the pyridinethiones and pyronethiones as
ligands.46
Table 2 1H NMR data for La(III) complexes (300 MHz, RT, DMSO-d6, X = O for Hma, Hema, La(ma)3 and La(ema)3, NH for Hmpp, La(mpp)3 and
NCH3 for Hdpp, La(dpp)3). Coupling constants are listed in parentheses.
Compound
Ra
Ha(d)
Hb(d)
OH
8.80
8.79
Hma
La(ma)3
Hema
2.23
2.20
2.61
8.00 (3J = 5.4)
7.93 (3J = 5.0)
8.01 (3J = 4.6)
7.90 (3J = 5.0)
6.33 (3J = 5.4)
6.23 (3J = 3.1)
6.33 (3J = 5.4)
6.16 (3J = 4.6)
1.09 (3J = 7.3)
2.57
La(ema)3
Hmpp
1.00 (3J = 7.3)
2.17
7.39 (3J = 6.9)
6.09 (3J = 6.9)
8.31
8.31
b
La(mpp)3
2.15
7.26 (br = 0.3 ppm)
7.54 (3J = 7.31)
6.07 (br = 0.3 ppm)
6.07 (3J = 6.85)
Hdpp
2.25
3.62c
b
La(dpp)3
2.26
7.47 (br = 0.35 ppm)
6.03 (br = 0.35 ppm)
3.63c
a R = CH3 (s) for Hma, Hmpp, Hdpp, La(ma)3, La(mpp)3, La(dpp)3; R = CH2 (q) CH3 (t) for Hema and La(ema)3 b The coupling constants for the
aromatic protons could not be calculated due to broad peaks instead of defined doublets seen for the proligands. c d of the –NCH3 moiety of Hdpp,
La(dpp)3; abbreviations: s = singlet, d = doublet, t = triplet, q = quartet.
This journal is
The Royal Society of Chemistry 2007
Dalton Trans., 2007, 5019–5030 | 5021
©

View Online
Fig. 1 Cell uptake of LnL3 in Caco-2 cells determined by ICP-MS. Values are presented as a mean percentage of Ln(III) added g protein−1 well−1 SD,
n = 6. La(dpp)3 exhibits significantly higher cell uptake (P < 0.001).
Table 3 13C NMR data for the La(III) pyrone complexes (300 MHz, RT,
geometric (facial and meridional) isomers at RT, or due to the
presence of water molecules in the coordination sphere.54
In the 13C NMR spectra of La(ma)3 and La(ema)3, the most
DMSO-d6).
=
significant chemical shifts were observed for the C O (Dd ∼7.54–
8.10 ppm) and C–O–La (Dd ∼4.92–14.87 ppm) consistent with the
La(III) ions being bound to the ligand through the O,O moiety.
Both these signals were shifted downfield due to donation of
electron density from the binding moieties to the metal ion. Peaks
representing the other carbons in the ring, C1, C4, and C5 shifted
upfield, with the biggest shifts seen for the C4 and C5 carbons (Dd
∼2.08–9.28 ppm). The 13C NMR signals from the methyl group
of La(ma)3 or from the ethyl group of La(ema)3 did not shift
significantly (Dd ∼0.24–0.55 ppm). The broadening of the carbon
signals also suggested the presence of isomers at RT.
Cmpd
C1
C2
C3
C4
C5
C6a
C7
Hma
La(ma)3
Hema
149.97 149.23 172.56 113.56 154.62 13.94
149.27 154.15 180.10 111.16 152.54 14.30
154.81 153.39 172.90 113.52 142.39 21.09 10.88
La(ema)3 153.46 168.26 181.00 110.86 152.07 20.54 11.12
a C6 is CH3 group for Hma and Ln(ma)3
The unique electronic properties of lanthanide cations, partic-
ularly Tb(III) and Eu(III), (long luminescent lifetimes, sharp emis-
sion bands, and large energy gap between absorbance and emission
band of complexes) render these metal centers attractive for
luminescence studies.55,56 Unfortunately, the extinction coefficients
for lanthanide ions tend to be small, however, this can be easily
corrected by utilizing sensitized luminescence with radiationless
energy transfer from an aryl chromophore to the metal ion.57,58 By
using this ‘antenna effect’, the Ln(III) ionluminesces with increased
intensity in the visible region.
The pyrone and pyridinone ligands can photosensitize Ln(III)
luminescence due to their UV absorption and appreciable molar
absorptivity coefficients.51 Luminescent excited states of the Tb(III)
and Eu(III) complexes were generated via excitation of the pyrone
and pyridinone moieties using UV wavelengths ranging from 315–
324 nm. Energy was transferred to the 5D4 or 5D0 excited states of
the Tb(III) or Eu(III), respectively (Table 4). Their fluorescence
A larger upfield shift was seen for Hb (Dd ∼0.10–0.17 ppm)
than for Ha (Dd ∼0.07–0.11 ppm) on the La(III) pyrone ligands,
whereas a larger shift for Ha (Dd ∼0.13–0.07 ppm) compared
to Hb (Dd ∼0.02–0.04 ppm) was seen in the La(III) pyridinone
ligands. This difference in the shifts of the La(III) pyrone and
pyridinone ligands can be attributed to the increased electron
density in the pyridinone ligands donated from the ring N atom.
The nitrogen atom stabilizes the pyridinone complex by adding
electron density to the ring current of the ligand from its lone pair
of electrons. The chemical shifts of the methyl and ethyl groups
on the ligands remain almost unchanged. In addition, there is
only one set of broad peaks assigned to the ring hydrogens on the
ligands, suggesting that their chemical shifts are either averaged
values from the rapidly interconverting optical (D and K) and
Table 4 Fluorescence data for the Eu(III) and Tb(III) complexes (k in nm). The strongest emissions for the Eu(III) and Tb(III) complexes are from 5D0 →
7
7F2 and 5D4 → F5, respectively. All other emission peaks seen in the fluorescence spectra are listed
7
7
7
7
7
7
Complex
kexc/nm
5D0 → F2
5D0 → F0
5D0 → F1
Complex
kexc/nm
5D4 → F5
5D4 → F6
5D4 → F4
Eu(ma)3
Eu(ema)3
Eu(mpp)3
Eu(dpp)3
319
320
322
294
642
639
647
616
618
617
704
703
713
682
Tb(ma)3
Tb(ema)3
Tb(mpp)3
Tb(dpp)3
315
317
324
309
641
637
651
620
545
545
705
701
721
684
5022 | Dalton Trans., 2007, 5019–5030
This journal is
The Royal Society of Chemistry 2007
©

View Online
Table 5 Cytotoxicity data (MTT assay) for the LnL3 complexes, n ≥ 6
was red-shifted ca. 300 nm from the maximum absorbance
7
( SD)
through possible emission transition states Fj j = 6–0 with the
5
7
5
7
most intense peaks resulting from D4 → F5 and D0 → F2
transitions, respectively.59,60 Most of the other transition states
were not observed as they were weak or outside the fluorimeter
detection wavelengths. The emission bands of the complexes
were also red-shifted slightly from the spectra of the Eu(NO3)3
and Tb(NO3)3 starting materials, reflecting the substitution of
water and nitrate ligands with the coordinating O atoms of
the pyrones and pyridinones. Ligand emission was observed,
suggesting incomplete ligand-to-lanthanide energy transfer. The
Complex
IC50/lM
cisplatin
Hma
La(ma)3
Gd(ma)3
Yb(ma)3
Hema
La(ema)3
Gd(ema)3
Yb(ema)3
Hmpp
La(mpp)3
Gd(mpp)3
Yb(mpp)3
Hdpp
8.67 (0.26)
>800
155.58 (8.58)
118.37 (9.16)
107.92 (4.46)
>714
138.95 (18.05)
107.75 (4.42)
144.13 (25.11)
451.7 (4.56)
142.30 (6.18)
136.1 (1.15)
136.5 (4.67)
719
7
intensity ratio for the hypersensitive 5D0 → F5 transition and the
7
magnetic dipole transition 5D0 → F1 with a value of ∼25, indicates
the absence of imposed symmetry on the complex observed in the
NMR data. Complexes with 7F2/7F1 intensity ratio lower than 0.7
have centrosymmetric coordination spheres, whereas an intensity
ratio higher than 8 is indicative of low symmetry environments.61,62
La(dpp)3
Gd(dpp)3
Yb(dpp)3
161.72 (2.08)
120.78 (16.05)
70.47 (4.42)
Cellular toxicity and uptake studies
stages of drug screening, determines cell viability by measuring
the mitochondrial activity of in vitro cell lines after the addition of
selected compounds.70 The NADPH or NADH produced by mi-
tochondrial dehydrogenase enzymes reduces the tetrazolium salts
to colored formazan crystals that can be measured spectrophoto-
metrically. A low IC50 implies cytoxity or antiproliferation of the
cells at low drug concentrations.
The metal complexes were considerably more toxic than the
analogous proligands after incubation with the cells for 72 h, but
less toxic than cisplatin, a known chemotherapeutic which has an
IC50 = 8.67 0.26 lM in Caco-2 cells.71 All IC50 values were in the
same range, independent of the particular bidentate ligand, with
the lowest IC50 = 70.47 4.42 lM for Yb(dpp)3 and the highest
being 161.72 2.08 lM for La(dpp)3. A lack of differential toxicity
for the various complexes was determined.
All compounds, including La2(CO3)3 exhibited similar cell
uptake independent of the particular ligand (0.26–0.65% uptake),
with the exception of La(dpp)3 that showed cell uptake of 9.07
2.33% (P < 0.001, Fig. 1). Hdpp is the most lipophilic of the
ligands used and lipophilicity may increase cell passage. Masking
of the Ln(III) metal with the dpp− ligand resulted in a considerable
increase in cellular uptake of elemental Ln(III) compared to
La2(CO3)3.
One of the limitations of Ln(III) ions is their inability to cross
cell membranes, particularly intestinal barriers, as exemplified by
the FDA approved La2(CO3)3.15,63 Masking the Ln(III) ion with
ligands designed to form small, lipophilic, low molecular weight
compounds, we aim to lower the dose and minimize unwanted
side effects by increasing the selective bioavailability of the Ln(III)
ions.
A Caco-2 cell assay was used as a model for the absorption and
secretion of molecules crossing the intestinal membrane. Caco-2
cells, originally derived from a human colorectal carcinoma line,
differentiate at confluency into highly functional epithelial barriers
with morphological and biochemical similarities to the small-
intestinal columnar epithelium.42–44,49 These cells have been used
to evaluate the relative bioavailability and transport mechanisms
of diverse drugs in vitro.64–68 When grown as a monolayer on a
semipermeable membrane, the passage of drugs across the Caco-2
cell barrier can be monitored by analyzing the concentration of
the desired compounds transported to either the basolateral (B,
bottom) side through the mimicked intestinal environment, or to
the apical (A, top) side through secretory transport. In addition,
the concentrations of compounds present inside the cells can be
measured by digesting the cell membrane to release the cytosol
contents.
Inductively coupled plasma mass spectrometry, ICP-MS, was
used to measure the bifunctional transport of the LnL3 across
the Caco-2 cell barrier, and the metal ion concentrations of
the apical, basolateral, and cell membrane. Regarded as one
of the best analytical tools for the quantitative determination
of lanthanides, ICP-MS can detect Ln(III) ion concentrations
≤ 1 ppt.69 Unfortunately, this study does not allow us to determine
the composition of the compound being transported, in the
biological media, or in the cell. However, one can qualitatively
determine if the metal complexes increase the bioavailability
of Ln(III) ions across an intestinal barrier, compared to the
benchmark compound, La2(CO3)3.
No reduction in the transepithelial electrical resistance (TEER)
was observed during the bifunctional uptake studies indicating
that the integrity of the monolayer of the Caco-2 cells was
not disturbed at the concentrations of LnL3 and time points
tested. Slight differences in apparent permeability (Papp, cm s−1)
coefficients from A-to-B vs. B-to-A transport were found between
compounds (Fig. 2). Yb(ma)3 exhibited a lower Papp value for both
A–B and B–A compared to the other LnL3 species. In contrast,
La(dpp)3, presumably the most stable of the complexes, showed a
higher A–B uptake. However, the B-to-A Papp value for La(dpp)3
was not significantly different than the values for the other LaL3
tested. When chelated to the Ln(III), the Papp (× 10−6 cm s−1)
values were significantly lowered compared to the proligand,
namely Hddp with previously reported Papp values of 22.3 1.2 ×
MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium
bromide) assays were used to evaluate the cytotoxicity of Ln(ma)3,
Ln(ema)3, Ln(mpp)3 and Ln(dpp)3 complexes using Caco-2 cells
(Table 5). This colorimetric assay, commonly used for the initial
10−6 cm s−1and 18.5
0.8 × 10−6 cm s−1 for the A-to-B and
B-to-A transport, respectively.72 In any case, the LnL3 species
tested exhibit similar partition coefficients as other small
This journal is
The Royal Society of Chemistry 2007
Dalton Trans., 2007, 5019–5030 | 5023
©

View Online
Fig. 2 Apparent permeability (Papp
to apical (B → A) and apical to basolateral (A → B), were measured by ICP-MS after 4 h of exposure to LnL3.
)
SD of La(ma)3, La(ema)3, La(dpp)3 and Yb(ma)3, across a Caco-2 membrane, n = 6. Both data sets, basolateral
bioavailable drugs such as dopamine, a sympathetic nervous
system agent (9.33 3.48 cm s−1), mannitol, used to reduce in-
tracranial pressure (0.38 0.12 cm s−1), acebutalol hydrochloride,
a cardiovascular agent (0.51 0.02 cm s−1), and Zidovudine, an
antiretroviral drug for the treatment of HIV (6.93 0.17 cm s−1)
as reported by other investigators.65 In addition, one of our
complexes has significantly greater bioavailability than does the
benchmark La2(CO3)3 based on preliminary results.
may explain the slower uptake of La(dpp)3 (95.8 1.7% at 3 h)
compared to the other LaL3 complexes tested (99.1 0.4% for
La(ma)3 and 99.81 0.01% for La(ema)3 at 3 h). Pyridinones are
known to be stronger binders than are their pyrone analogs due to
the ring N donating more electron density than its electronegative
O counterpart.23,73
Fluorescence studies were used to determine quantitatively the
amount of the Ln(III) ions unbound to HA by measuring the
fluorescence of the supernatant from the Tb(ma)3 and Eu(ma)3
samples. The fluorescence results confirmed the ICP-MS data
showing > 98% Ln–HA binding after 24 h (P > 0.05 compared to
the LnL3 species analyzed by ICP-MS, Fig. 3).
A colorimetric assay, with xylenol orange (sodium salt) dye
indicative of free Ln(III) ions, was also used to confirm the ICP-
MS results.74–76 In acidic buffered solutions, xylenol orange reddens
in the presence of free Ln(III) ions. In contrast, appearance of a
yellow–orange colour indicates the absence of unbound Ln(III)
ions. Solutions of proligand and HA were used as negative
controls, solutions of Ln(NO3)3 served as positive controls. As
expected, the proligands, HA and the LnL3 solutions turned
weakly yellowish–orange suggesting the absence of unbound
Ln(III) in solution; however, the Ln(NO3)3 solution turned a dark
red–purple.
In vitro hydroxyapatite binding studies
La(ma)3, Tb(ma)3, Gd(ma)3, Eu(ma)3, Yb(ma)3, La(ema)3,
La(dpp)3, and La2(CO3)3 (tested at concentrations of 1 lM, well
below their respective IC50 values) showed high HA binding with
a quantitative uptake of > 98% of the Ln(III) onto HA after 24 h
incubation. Minimal variability was observed between lanthanides
(Fig. 3). In addition, > 95% Ln(III) ions were adsorbed onto
the HA in 5 min or less. La(dpp)3, Gd(ma)3 and Yb(ma)3 had
slightly slower binding kinetics, possibly due to the stability of the
complexes. Yb(III) and Gd(III) ions are stronger Lewis acids than is
the slightly larger La(III) ion. Thus, these ions bind more tightly to
the hard O,O donors of the pyrone and pyridinone ligands further
competing with the HA binding moieties. Stability of the complex
Fig. 3 Percentage SD of Ln(III) bound to HA samples, n = 6. Tb(III) and Eu(III) concentrations were determined by measuring the fluorescence of the
Tb(ma)3 or Eu(ma)3 species in the supernatant, respectively, and subtracting the amount found from the total amount of LnL3 added. All other Ln(III)
concentrations were determined by ICP-MS analysis (P > 0.05 between samples at 24 h).
5024 | Dalton Trans., 2007, 5019–5030
This journal is
The Royal Society of Chemistry 2007
©

View Online
Solid samples from the metal–HA binding studies as well as
the supernatant from the same experiment were tested for the
presence of free Ln(III) ions. The addition of acid was presumed
to digest the hydroxyapatite of the solid sample thereby releasing
any bound Ln(III) ions into solution. In concordance with the
ICP-MS data, the supernatant turned a slightly darker orange–
yellow than did the controls, suggesting a low concentration of free
Ln(III). The digested Ln–HA samples turned dark red indicating
an abundance of free Ln(III) ions. These results confirm the ICP-
MS data indicating a large proportion of the Ln(III) ions present
on the HA matrix.
To further confirm these results, xylenol orange was added to
an undigested Ln–HA sample in the buffer solution. The solution
remained yellow–orange, but the particles turned a dark red–
purple suggesting the presence of Ln(III) ions on the surface of
the HA solid samples.
Fig. 5 IR spectra from HA binding studies.
Spectrophotometric determination (UV-vis assay) of ligand
concentration in supernatants from La(ma)3, Eu(ma)3, Gd(ma)3,
Tb(ma)3, Yb(ma)3, La(ema)3, and La(dpp)3–HA binding studies
indicated that at the 24 h time point, > 97% of the ligand remained
in solution independent of the metal complex tested (Fig. 4). High
concentration of the ligand in solution indicated disassociation
of the LnL3 complex suggesting that the bidentate ligands on the
Ln(III) ion were almost completely replaced by the phosphates and
carbonate of the HA matrix.
IR spectra of the solid HA samples exposed to the Ln(III)
complexes were compared to spectra of pure HA, the proligand,
and the HA-bound LnL3 complexes (Fig. 5). Irrespective of the
metal ion, the complex or the exposure time of the LnL3 species
to the solid HA, the IR spectra were nearly superimposable with
that of pure HA suggesting that the HA composition had not been
modified upon complexation of the Ln(III) salt.77–80 Although not
prominent, the hydroxyl stretching (–OH) and bending bands were
observed at 3569 and 630 cm−1 for the HA–Ln samples. Broad
bands at 3450 and 1640 cm−1 were attributed to adsorbed water.
The v3 and v1 PO43−vibrations were assigned to broad bands at
1043 and 969 cm−1, respectively. Bands at 602 and 567 cm−1 were
from the v4 vibrations. The band observed at 877 cm−1 corresponds
to HPO42− vibrations. The v3 bands of carbonate ions observed at
1460 and 1425 cm−1 are generally representative of substitution of
phosphate groups for carbonate species. A new peak observed at
∼474 cm−1 corresponds to mLn–O vibrations showing that the Ln(III)
is bound to the HA upon coordinating to either the phosphate
or the carbonate oxygens of the hydroxyapatite. In addition,
vibrations from the proligand in the Ln–HA samples were not
observed, consistently suggesting that the Ln(III) dissociates upon
complexation to the HA.
The lack of IR shifts arising from the pure HA versus the Ln(III)–
HA are not surprising due to the similarities of the Ca(II) and
Ln(III) ions. Despite the slightly larger unit-cell volume found
for Ln(III)–HA compared to Ca(II)–HA, Ln(III) are essentially
functional mimics of Ca(II) ions sharing similar ionic radii, donor
atom preferences, and almost identical coordination numbers in
protein binding sites.9,81
Summary
Twenty LnL3 were successfully synthesized using Hma, Hema,
Hmpp, or Hdpp as proligands. La(III), Gd(III), and Yb(III)
complexes were chosen, in order to evaluate size affects and assess
their medicinal applications, Tb(III) and Eu(III) complexes were of
interest due to their characteristic fluorescence properties. Toxicity,
bifunctional transport across a model of an intestinal barrier, and
binding coefficients to HA were evaluated for several of the LnL3
species to assess their potential therapeutic use for bone resorption
disorders. By comparison with FosrenolTM, an FDA approved
lanthanum-containing phosphate binder, the LnL3 compounds
Fig. 4 Percent ligand concentration SD remaining in the supernatant of the Ln(III)–HA binding studies determined by UV-vis spectroscopy (P > 0.05
between complexes at 24 h), n = 6.
This journal is
The Royal Society of Chemistry 2007
Dalton Trans., 2007, 5019–5030 | 5025
©

View Online
investigated showed similar HA-binding (> 98%) and one of the
compounds had greater absorption across an intestinal barrier,
compared to lanthanum carbonate. The LnL3 compounds were
significantly less toxic than cisplatin and did not appear to disturb
the HA structure upon binding. These novel LnL3 thus have
potential as orally available bone resorption inhibition drugs, with
minimized side effects that may also serve as potential preventative
supplements for osteoporosis. La(dpp)3 was identified as the lead
compound with high HA-binding (98.45 1.45%) and a moderate
Papp value. This compound will be further tested in appropriate
animal models of bone disorders for both bioavailability and anti-
resorptive activity.
of additional water was added and the product was extracted into
methylene chloride (2 × 20 mL). The water layer was discarded
and the CH2Cl2 extracts were washed with 5% NaOH (3 ×
15 mL), then water (1 × 20 mL). The CH2Cl2 layer was dried
over MgSO4, filtered, and the solvent was removed by rotary
evaporation affording a viscous yellow oil. The oil was dissolved
in hot ethanol and maintained at −5 ◦C for 48 h, after which time
white cubic crystals formed. These were filtered out, washed with
cold ethanol (3 × 5 mL), and vacuum dried (6.8 g, 31.5 mmol, 79%
1
yield). H NMR (300 MHz, RT, CDCl3) d = 2.04 (s, 3H, CH3),
3
5.10 (s, 2H, Bn–CH2), 6.33 (d, 1H, Hb, Jb,a = 5.1 Hz), 7.29 (m,
5H, Bn C6H5), 7.54 (d, 1H, Ha, 3Ja,b = 5.6)
3-Benzyloxy-2-methyl-4(1H)-pyridinone, Bnmpp. Bnma (6.8 g,
31.5 mmol) in 80.5 mL of concentrated NH4OH and 30 mL of
ethanol was stirred in a round bottomed flask at room temperature
for 3 days. After each day, an additional 10 mL of ethanol
was added. Ethanol and the excess ammonia were removed by
rotary evaporation and the residue washed with 10 mL of acetone.
The 3-benzyloxy-2-methyl-4(1H)-pyridone, Bnmpp, product was
crystallized from ethanol as colorless crystals, filtered, and dried
under vacuum overnight (4.3 g, 20.1 mmol, 64% yield). 1H NMR
(300 MHz, RT, CD3OD) d = 2.05 (s, 3H, CH3), 5.01 (s, 2H, Bn-
CH2), 6.34 (d, 1H, Hb, 3Jb,a = 5.8 Hz), 7.31 (m, 5H, Bn C6H5), 7.86
(d, 1H, Ha, 3Ja,b = 5.6)
Experimental
Chemicals
NaOH, 3-hydroxy-1,2-methyl-4-pyridinone (Hdpp), hydrated lan-
thanum carbonate and hydrated lanthanide nitrate salts were
obtained from Sigma-Aldrich, Fisher Scientific and Alfa Aesar
and used without further purification. All solvents were reagent
grade from Fisher Scientific. Maltol (Hma) and ethylmaltol
(Hema) were obtained from Pfizer or Cultor Food Science. Water
was deionized (Barnstead D8902 and D8904 Cartridges) and
distilled (Hytrex II GX50-9-7/8 and GX100-9-7/8) before use.
Water for HA studies and ICP-MS analysis was collected from
Elgastat Prima 2 reverse osmosis and ultra pure water system
(High Wycombe, England). NMR solvents were purchased from
Aldrich and used as received. Yields for the analytically pure
compounds were calculated based on the respective metal ion
starting material.
3-Hydroxy-2-methyl-4(1H)-pyridinone, Hmpp. Bnmpp (4.3 g,
20.1 mmol) was debenzylated by treatment with 10 mL of 45%
w/v hydrogen bromide–acetic acid solution for 30 min on a steam
bath. Toluene by-product and excess solvent were removed by
rotary evaporation. The precipitate was filtered out, redissolved
in 10 mL of water and brought to pH 7–8 with the addition of
6 M NaOH. The solvent was removed by rotary evaporation.
3-Hydroxy-2-methyl-4(1H)-pyridinone (Hmpp) crystallized from
ethanol as light pink crystals after 24 h in 4 ◦C (1.1 g, 8.64 mmol,
Instruments
1H NMR spectra (300 MHz) and 13C NMR spectra (75 MHz)
were recorded on a Bruker Varian XL-300 spectrometer and
referenced internally to residual solvent protons. Infrared spectra
were recorded as Nujol mulls (KBr windows) in the range 4000–
500 cm−1 on a Mattson Galaxy Series FTIR-5000 spectropho-
tometer and were referenced to polystyrene (1601 cm−1). Analyses
of C, H, and N were performed by Mr M. Lakha at the University
of British Columbia. UV-visible spectra were recorded in water
or 0.1% saline solution using a Hewlett-Packard 8543 diode array
spectrophotometer and a 1 cm cuvette. Mass spectra were obtained
on a Bruker Esquire Ion Trap (electrospray ionization mass
spectrometry, ESIMS) spectrophotometer. Fluorescence spectra
were obtained in 0.1% saline solution or methanol on a PTI
QuantaMaster fluorimeter using a 1 cm quartz cuvette. A high
resolution ICP-MS (Thermo Finnigan ELEMENT 2) was used
for determination of metal ion concentration, calibrated against
Delta Scientific Ltd. Standards.
1
43%). H NMR (300 MHz, RT, CD3OD) d = 2.17 (s, 3H, CH3),
6.09 (d, 1H, Hb, 3Jb,a = 6.9 Hz), 7.39 (d, 1H, Ha, 3Ja,b = 6.9), 8.31
(s, 1H, OH). MS (+ESI-MS, MeOH): m/z 126 [M + H]+
.
General procedure for Ln(ma)3·nH2O, Ln = La, Eu, Gd, Tb, Yb.
Ln(NO3)3·6H2O (0.433–0.467 g, 1.0 mmol) was added to a solution
of Hma (0.378 g, 3.0 mmol) in 10 mL of deionized, distilled water
and the solution was heated and stirred until the ligand completely
dissolved (∼10 min). The pH of the clear solution was raised slowly
over 10 min with 1 M NaOH (Ln = La, pH 8.03; Eu, pH 8.00;
Gd, pH 8.26; Tb, pH 8.5; Yb, pH 8.62) and the resulting reaction
mixture was continuously stirred for 16 h. A white, or slightly off-
white precipitate formed (except in the case of Eu(ma)3 for which
a yellow solid formed) and was collected by vacuum filtration,
washed with 2 × 5 mL of cold water and 2 × 5 mL of cold
MeOH. All compounds were dried in vacuo to yield 36–59% of
product.
3-Benzyloxy-2-methyl-4-pyrone, Bnma. This intermediate was
prepared similarly to the previously published procedure
with some modifications.38 3-Hydroxy-2-methyl-4-pyrone (5 g,
40 mmol) was dissolved in 200 mL of methanol and a solution
of sodium hydroxide (3 g, 75 mmol in 30 mL of water) was added.
After addition of benzyl chloride (7 mL, 61 mmol), the mixture
was refluxed overnight and cooled in an ice bath. MeOH was
removed from the yellow solution by rotary evaporation, 20 mL
General procedure for Ln(ema)3·nH2O, Ln = La, Eu, Gd, Tb,
Yb. Ln(NO3)3·6H2O (0.433–0.467 g, 1.0 mmol) was added to
a solution of Hema (0.420, 3.0 mmol) in 10 mL of ethanol and
the solution was heated and stirred until the ligand completely
dissolved (∼5 min). Triethylamine (0.303 g, 3 mmol) was added
slowly over 10 min to the clear solution. The resulting reaction
mixture was continuously stirred for 18–24 h. A white or slightly
off-white precipitate formed (except in the case of Eu(ema)3 for
5026 | Dalton Trans., 2007, 5019–5030
This journal is
The Royal Society of Chemistry 2007
©

View Online
which a yellow solid formed) and was collected via vacuum
filtration, washed with 2 × 5 mL of cold water and 2 × 5 mL
of cold MeOH. All compounds were dried in vacuo to yield 53–
85% of product.
100 lg mL−1 penicillin and 100 lg mL−1 streptomycin at 37 ◦C
in humidified air containing 5% CO2 in either T75 flasks, 6-
well, 12-well, 96-well or Transwell plates depending on the type
of experiment. Medium was changed every other day. Cells
used for experiments reached 80 to 90% confluency (in the case
of transfer studies when the transepithelial electrical resistance
(TEER) values were above 400 X cm2, measured using Millicell-
ERS from Millipore).
Data from cell permeability, cytotoxicity and hydroxyapatite
binding (vide infra) studies were analyzed by ANOVA and
Newman–Keuls post-analysis, where appropriate, and expressed
as means SD. Studies were compared against La2(CO3)3 unless
otherwise stated. Statistical significance was accepted at P < 0.05.
General procedure for Ln(dpp)3·nH2O, Ln = La, Eu, Gd, Tb,
Yb. Ln(NO3)3·6H2O (0.433–0.467 g, 1.0 mmol) was added to
a solution of Hdpp (0.417, 3.0 mmol) in 10 mL of deionized
and distilled water and the solution was heated and stirred until
the ligand completely dissolved (∼30 min). The pH of the clear
solution was raised slowly over 10 min with 6 M NaOH (Ln =
La, pH 10.02; Eu, pH 10.20; Gd, pH 10.42; Tb, pH 11.00; Yb,
pH 11.13) and the resulting reaction mixture was continuously
stirred for 16 h. A white, or slightly off-white precipitate formed
(except in the case of Eu(dpp)3 for which a yellow solid formed)
and was collected by vacuum filtration, washed with 2 × 5 mL of
cold water and 2 × 5 mL of cold MeOH. All compounds were
dried in vacuo to yield 40–76% of product.
MTT assay
A modified procedure was used for the MTT assay (MTT =
(3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide).70
Caco-2 cells were seeded onto 96-well plates at a density of 1 ×
104 cells per well. On the day of the experiment, the culture
medium was exchanged for treatment solution of 1 lg mL−1,
10 lg mL1, 50 lg mL−1, 100 lg mL−1, 200 lg mL−1, 500 lg mL−1
equivalent to 2–1092 lM) of Hma, Hema, Hdpp, Hmpp, Ln(ma)3,
Ln(ema)3, Ln(mpp)3, and Ln(dpp)3, Ln = La, Gd, Yb (dissolved
in 1% DMSO/media solution, highest concentration of DMSO
100 lg mL−1). After an incubation of 72 h, a 50 lL aliquot
of MTT (2.5 mg mL−1 in PBS) was added to each well of the
plate. The plate was further incubated for 3 h at 37 ◦C during
which time purple formazan crystals formed at the bottom of the
wells. The solution of each well was carefully aspirated leaving
the formazan crystals which were dissolved in DMSO (150 lL).
The absorbance of each well was measured at 570 nm using a
microplate reader (Spectra Max 190, Molecular Devices). Cell
viability was calculated relative to the 100% control (DMSO plus
media). The IC50 was determined by plotting the percent cell
viability vs. drug concentration (logarithmic scale) and finding
the concentration at which 50% of the cells were viable relative to
the control. Values are presented as means SD, based on ≥ 6
independent trials.
General procedure for Ln(mpp)3·nH2O, Ln = La, Eu, Gd, Tb,
Yb. Ln(NO3)3·6H2O (0.144–0.154 g, 0.33 mmol) was added to a
solution of Hmpp (0.125 g, 1.0 mmol) in 10 mL of deionized,
distilled water and the solution was heated and stirred until
the ligand completely dissolved (∼30 min). The pH of the clear
solution was raised slowly over 10 min with 6 M NaOH (Ln = La,
pH 8.10; Eu, pH 8.20; Gd, pH 8.20; Tb, pH 8.30; Yb, pH 8.30) and
the resulting reaction mixture was continuously stirred for 16 h. A
white or slightly off-white precipitate formed (except in the case of
Eu(mpp)3 for which a yellow solid formed) and was collected by
vacuum filtration, washed with 2 × 5 mL of cold water and 2 ×
5 mL of cold MeOH. All compounds were dried in vacuo to yield
38–64% of product.
A complete list of compounds prepared and analytical data is
presented in Table 1.
Caco-2 cell culture
Caco-2 cells were purchased from American Type Culture Col-
lection (ATCC, Rockville, MD and Manassas, VA, USA). Cell
culture media were purchased from Gibco BRL (Grand Island,
NY, USA). Sterile SteritopTM 0.22 lm Express membrane bottle
top filters were purchased from Millipore (Bedford, MA, USA).
Sterile 50 ml centrifuge tubes, disposable 10 and 25 mL serological
pipettes were purchased from Starstedt (Montreal, PQ, Canada).
Culture flasks, Transwell and polycarbonate plates were obtained
from Corning-Costar (Cambridge, MA, USA). Triton X-100 was
purchased from Sigma.
Caco-2 cells for MTT studies were cultured in 75 cm2 Falcon
tissue culture flasks in Eagle’s Minimum Essential Medium
(EMEM) supplemented with Eagle’s BSS and 2 mM glutamine
modified with 1.0 mM sodium pyruvate, 0.1 mM nonessential
amino acids, 1.5 g L−1 sodium bicarbonate and supplemented
with 20% fetal bovine solution. The cells were incubated at 37 ◦C
in a humidified atmosphere of 5% CO2. Medium was changed
every other day and the cell stock was routinely subcultured by
trypsinization (0.25% trypsin with 0.53 mM EDTA) after reaching
80% confluency.
Protein concentration
Total protein content was determined in lysed cells using the
bicinchoninic acid protein assay (BCATM Protein Assay Kit) from
Pierce Biotechnology, Inc. (Rockford, IL, USA). After performing
the cell uptake and bifunctional transport studies, the assay media
was aspirated, cells were washed 3 times with PBS and were lysed.
A BSA standard curve was constructed in the range of 25 lg mL−1
to 2 mg mL−1. Exactly 10 lL of both standards and cell samples
(in triplicates) were aliquoted into a 96-well microtiter plate and
200 lL of Reagent (Mix A : B = 50 : 1) was added to each well.
Absorbance was measured after 1 h at room temperature at 540 nm
with a Multiscan Ascent Multi-plate reader from Labsystems.
Protein concentration of each sample was determined against the
standard curve.
Caco-2 cell uptake studies
Caco-2 cells for cell uptake and transport studies were cultured
in EMEM supplemented with 10% fetal bovine serum (FBS),
292 lg mL−1 glutamine, 0.1 mM non-essential amino acids,
Caco-2 cells, passage number 25 to 30, wer◦e seeded at 5000
cells cm−2 into 6-well plates and incubated in 37 C in a humidified
This journal is
The Royal Society of Chemistry 2007
Dalton Trans., 2007, 5019–5030 | 5027
©

View Online
atmosphere of 5% CO2. The medium was changed every other day.
After reaching 80% confluency, the cells were washed with
PBS. The medium was removed and replaced with either fresh
medium, or medium plus compound (La(ma)3, Gd(ma)3, Yb(ma)3,
La(ema)3, La(dpp)3, La2(CO3)3) at concentrations of 100 lg mL−1
(equivalent to 218–179 lM). The plate was incubated at 37 ◦C in
a humidified atmosphere of 5% CO2 for 4 h. Treatment solutions
were aspirated and cells were washed three times with ice-cold PBS
to remove any unbound LnL3. Cells were lysed with 1 mL of lysis
buffer (50 mM Hepes, 150 mM NaCl, 2 mM EDTA, 0.5% Na-
deoxycholate, 1% NP-40) plus protease inhibitor cocktail (1 : 1000
dilution). The lysed cell solution was transferred into separate
1.5 mL Eppendorf tubes. Culture plate wells were rinsed with
another 0.3 mL of lysis buffer, the combined cell/media samples
were kept for later analysis by ICP-MS. The protein content was
determined by a Pierce BCA assay using 10 lL of the lysed cells
(vortexed thoroughly and pelleted by centrifugation at 12 000 rpm
for 3 min). The concentrations are reported as % uptake of lM
where dQ/dt (nmol s−1) is the flux rate of mass transport across
the monolayers, A is the surface area of the insert membrane
(1.13 cm2), and Co is the initial concentration (lM) of the
compound in the donor chamber. The flux rates were determined
by plotting the concentration of Ln(III) ions determined by ICP-
MS as a function of time, with the slope calculated via linear
regression.
Hydroxyapatite in vitro binding study
Samples containing exactly 20 mg of dried hydroxyapatite (HA,
Sigma, St. Louis, MO, USA) were suspended in 2 mL of
physiological 0.9% saline solution, pH 7.4 in 13 × 100 mm test
tubes and incubated overnight at 37 ◦C to allow for equilibration
of the samples. The metal complexes (La(ma)3, Eu(ma)3, Gd(ma)3,
Tb(ma)3, Yb(ma)3, La(ema)3, La(dpp)3, and La2(CO3)3) were
dissolved in DMSO, final concentration 5 mg mL−1. Appropriate
volumes of the DMSO solutions were added in order to achieve
1 lM concentration for all compounds tested (maximum concen-
tration of DMSO in each sample ≤ 0.0005%) and the mixture
was shaken gently. After incubation (5, 15 min, 3, and 24 h, 6
vials for each time point), the solution was centrifuged at 3200
rpm for 5 min. Supernatants were carefully removed and placed
in 10 mL scintillation vials. Each pellet was washed twice with
1 mL of saline solution to remove any unbound metal ion from
the pellet. The washings were collected after centrifugation and
the supernatant was filtered through a 0.22 lm Millipore filter.
Exactly1 mL of the supernatant samples were vacuum centrifuged,
dried overnight (excluding Eu(ma)3 and Tb(ma)3 samples) and
resuspended in 2 mL of concentrated Seastar nitric acid for
digestion (see later for procedure) and diluted accordingly for ICP-
MS analysis. The remaining solution was used in determination
of the free ligand content by UV-vis of appropriate dilutions. The
Eu(ma)3 and Tb(ma)3 samples were used after filtering and diluted
appropriately for fluorescence studies. The HA samples were
dissolved in 2 mL of concentrated Seastar nitric acid, acid digested
and diluted accordingly for ICP-MS analysis. The total amount
of Ln(III) for each HA-binding experiment was determined by
adding the Ln(III) concentration found in the supernatant with the
concentration found in the HA samples. Percent HA-binding
SD was determined by dividing the Ln(III) concentration from
the HA sample by the total Ln(III) concentration found (HA and
supernatant) for each sample. Group values were compared by
ANOVA with Newman–Keuls post analysis. A P value < 0.05 was
considered to be statistically significant.
Ln(III) mg−1 protein
SD. Group values were compared by
ANOVA with a Newman–Keuls post analysis. A P value < 0.05
was considered to be statistically significant.
Bifunctional transport assay
Caco-2 cells, passage number 30, were seeded at 5000 cells cm−2
in polycarbonate membrane 12-well Transwell plates (Corning,
Costar Corp.) and incubated at 37 ◦C in a humidified atmosphere
of 5% CO2. The media was changed every other day. Uptake stud-
ies were conducted when the TEER readings reached 400 X cm−2.
Prior to permeability studies, the cells were washed with PBS
and the culture media was replaced with fresh media (control)
or media plus compound (concentration of 100 lg mL−1 for all
tested compounds, equivalent to 218–179 lM, including La(ma)3,
Gd(ma)3, Yb(ma)3, La(ema)3, La(dpp)3, La2(CO3)3) on the apical
or basolateral side. The integrity of the Caco-2 cell monolayer was
monitored by measuring the TEER values at each time point. The
amount of permeated solute was determined by removing 100 lL
aliquots from either the basolateral or apical side, opposite to
the side on which the compound was added, at 10, 30, 60, 120,
180 and 240 min and replaced with fresh media. The plate was
incubated at 37 ◦C in a humidified atmosphere of 5% CO2 in
between sample removal periods. All media was aspirated after 4 h,
vacuum dried, and acid digested for further analysis. The cells were
washed 3 times with PBS, the membrane was removed and lysed
in 1 mL of 0.1 N NaOH. Exactly 10 lL of the lysed cells (vortexed
thoroughly and pelleted by centrifugation at 12 000 rpm for 3 min)
were analyzed for protein content using the Pierce BCA assay. The
amount of Ln(III) ions accumulated in the cells, in addition to the
concentration of Ln(III) ions in the apical or basolateral chambers
were evaluated by ICP-MS. The % compound transported SD
was determined by dividing the Ln(III) concentration found in
either the basolateral or apical chamber by the total concentration
of Ln(III) found in the study (concentration of Ln(III) in the apical
and basolateral chambers, and in the cell membrane).
Ln(III) analysis by ICP-MS
High resolution ICP-MS Thermo Finnigan Element2 was used to
analyze the Ln(III) content of the cell lysates, solutions from both
the apical and basolateral transport studies, and solutions and
solid HA samples from the HA binding experiments. High purity
standards (In2O3, La2O3, Gd2O3, and Yb2O3 1000 lg mL−1 in 2%
HNO3) were purchased from Delta Scientific Laboratory Products
Ltd. (Spring City, PA). All materials used for the digestions of
samples and for ICP-MS including glassware, Teflon boiling chips,
sample holders, pipettes, and pipette tips were soaked in 2% Extran
solution for 24 h, rinsed with deionized, distilled water, soaked
overnight in 1% Seastar nitric acid, rinsed with deionized distilled
The apparent permeability (Papp, cm s−1) coefficient from the
bidirectional transport of the Ln(III) complexes were calculated
using eqn (1):
Papp = (dQ/dt)/(ACo)
(1)
5028 | Dalton Trans., 2007, 5019–5030
This journal is
The Royal Society of Chemistry 2007
©

View Online
water, and finally left to dry in a dust-free environment prior to
use. All samples were vacuum centrifuged to complete dryness
and digested in acid using the following procedure before analysis.
The redissolved digests were diluted accordingly with 1% nitric
acid and their lanthanide contents analyzed by monitoring m/z
139, 158, and 174 for 139La, 158Gd, 174Yb, respectively. Samples
were analyzed in the ‘peak jump’ mode and all signals corrected
according to the internal 115In standard. Interference corrections
for 158Gd were made by monitoring Dy at m/z 161.
appropriately with an aqueous 20% hexamethylenetetramine
buffer. Presence of free Ln(III) ions were confirmed by the
immediate appearance of a red colour in the solution upon the
addition of xylenol orange sodium salt.
Acknowledgements
We thank Dr Elena Polishchuk and Candice Martins for help with
MTT assays, Drs M. J. Abrams and G. J. Bridger for discussions,
Dr W. R. Cullen and Vivian W.-M. Lai for their expertise in acid
digestion technique and for the use of their equipment, and Bert
Mueller for help with the ICP-MS. We acknowledge support from
theNatural Sciences and Engineering Research Councilof Canada
and a University Graduate Fellowship (CAB) at the University of
British Columbia.
Acid digestion for ICP-MS analysis
Dried samples (vacuum centrifugation) were dissolved in 2 mL of
Seastar concentrated nitric acid. Samples were transferred to test
tubes with two or three Teflon boiling chips and placed in a block
heater (Standard Heatblock, VWR Scientific Products). The tem-
perature was slowly raised to 105 ◦C over 1 h and then maintained
at 105 ◦C for 24 h. Approximately 2 mL of 30% hydrogen peroxide
(Fisher Scientific) was added to each sample before heating at
140 ◦C for 24 h. The temperature was increased to 150 ◦C and
samples were evaporated to dryness. Subsequently, samples were
redissolved in 3 mL of 10% nitric acid and stored in 4 mL high
density polyethylene bottles (Wheaton, Milville, NJ) at 4 ◦C.
References
1 Government of Canada; http://www.biobasics.gc.ca/english/View.
asp?x = 770, (accessed Jan. 11, 2007).
2 K. F. Mauck and B. L. Clarke, Mayo Clin. Proc., 2006, 81, 662.
3 D. W. Dempster and R. Lindsay, Lancet, 1993, 341, 797.
4 M. T. McDermott, C. Zapalowski and P. D. Miller, Osteoporosis,
Hanley & Belfus, Toronto, ON, 2004, p 1-168.
5 J. N. Fordham, Osteoporosis, Churchill Livingstone, Toronto, 2004,
p 223.
Ligand analysis by UV-Vis Assay
6 M. A. Karsdal and C. Christiansen, Future Rheum., 2006, 1, 683.
7 P. J. Marie, Bone, 2006, 38, S10.
A calibration curve consisting of appropriate concentrations
of dissolved ligand in 0.9% physiological saline solution was
constructed for Hma, Hema and Hdpp at kmax = 273, 279, 275 nm,
respectively. The undigested supernatants from La(ma)3, Eu(ma)3,
Gd(ma)3, Tb(ma)3, Yb(ma)3 La(ema)3, and La(dpp)3–HA binding
studies for all four time points were diluted 100 fold with 0.9%
physiological saline solution. The concentration of ligand in
solution was calculated by comparing the absorbances collected
from the supernatant solutions to the previously constructed
calibration curves. High concentration of the ligand in solution
indicates dissociation of the LnL3 complex upon HA binding.
Percent ligand in solution was determined by dividing the ligand
concentration determined in the supernatant by the total ligand
concentration added.
8 S. C. Verberckmoes, M. E. D. Broe and P. C. D’Haese, Kidney Int.,
2003, 64, 534.
9 E. Pidcock and G. R. Moore, J. Biol. Inorg. Chem., 2001, 6, 479.
10 T. J. Webster, E. A. Massa-Schlueter, J. L. Smith and E. B. Slamovich,
Biomaterials, 2004, 25, 2111.
11 D. Nigel, W. Joseph and D. Michael, World Patent, WO 02/00227 A2,
2002.
12 P. Caravan, J. J. Ellison, T. J. McMurry and R. B. Lauffer, Chem. Rev.,
1999, 99, 2293.
13 I. Kostova, Curr. Med. Chem., 2005, 5, 591.
14 N. Sabbatini, M. Guardigli and J.-M. Lehn, Coord. Chem. Rev., 1993,
123, 201.
15 S. P. Fricker, Chem. Soc. Rev., 2006, 35, 524.
16 D. Nayak and S. Lahiri, J. Radioanal. Nucl. Chem., 1999, 242, 423.
17 W. A. Volkert and T. J. Hoffman, Chem. Rev., 1999, 99, 2269.
18 M. A. Santos, M. Gil, L. Gano and S. Chaves, J. Biol. Inorg. Chem.,
2005, 10, 564.
19 M. A. Santos, S. Gama, L. Gano, G. Cantinho and E. Farkas, Dalton
Trans., 2004, 3772.
20 D. M. J. Doble, M. Melchior, B. O’Sullivan, C. Sierring, J. Xu, V. C.
Pierre and K. N. Raymond, Inorg. Chem., 2003, 42, 4930.
21 M. M. Jones, J. J. Molenda, T. P. Hanusa and M. W. Voehler, J. Inorg.
Biochem., 1996, 62, 127.
Eu(ma)3 and Tb(ma)3 analysis by fluorescence assay
A calibration curve over a range of concentrations was constructed
by integrating the area under the most intense fluorescence
peak (kem = 641 for Tb(ma)3 and 642 nm for the Eu(ma)3)
and plotted against the known concentration. Concentrations of
the complexes in solution were determined by comparing their
integrated area under the most intense fluorescence peak against
the calibration curve. Percent Ln(III) bound was determined
by subtracting the amount of Tb(III) or Eu(III) found in the
supernatant from the total amount of Ln(III) added to the sample.
This value was then divided by the total ligand concentration.
22 A. E. V. Gordon, J. Xu, K. N. Raymond and P. W. Durbin, Chem. Rev.,
2003, 103, 4207.
23 K. H. Thompson, C. A. Barta and C. Orvig, Chem. Soc. Rev., 2006, 35,
545.
24 R. C. Hider and T. Zhou, Ann. N. Y. Acad. Sci., 2005, 1054, 141.
25 Z. D. Liu and R. C. Hider, Med. Res. Rev., 2002, 22, 26.
26 R. A. Yokel, Coord. Chem. Rev., 2002, 228, 97.
27 M. A. Santos, Coord. Chem. Rev., 2002, 228, 187.
28 Titan Pharmaceuticals Inc., http://www.titanpharm.com/pt-gallium.
html, (accessed Jan 2006).
29 D. J. Clevette and C. Orvig, Polyhedron, 1990, 9, 151.
30 Z. Zhang, D. M. Lyster, G. A. Webb and C. Orvig, Nucl. Med. Biol.,
1992, 19, 327.
31 D. E. Green, C. L. Ferreira, R. V. Stick, B. O. Patrick, M. J. Adam and
C. Orvig, Bioconjugate Chem., 2005, 16, 1597.
32 M. A. Barrand, B. A. Callingham and R. C. Hider, J. Pharm.
Pharmacol., 1987, 39, 203.
Xylenol orange assay
Approximately 20 mg of proligand, HA, LnL3, Ln(NO3)3, solid
samples from the metal–HA binding studies as well as the
supernatant from the same experiment were dissolved in minimal
amounts of 6 M HCl. The pH was adjusted to 5 and diluted
33 M. A. Barrand, R. C. Hider and B. A. Callingham, J. Pharm.
Pharmacol., 1990, 42, 279.
This journal is
The Royal Society of Chemistry 2007
Dalton Trans., 2007, 5019–5030 | 5029
©

View Online
34 D. G. Maxton, R. P. H. Thompson and R. C. Hider, J. Nutr., 1994, 71,
203.
60 L. Tei, A. J. Blake, M. W. George, J. A. Weinstein, C. Wilson and M.
Schroder, Dalton Trans., 2003, 1693.
61 A. F. Kirby, D. Foster and F. S. Richardson, Chem. Phys. Lett., 1983,
95, 507.
35 K. H. Thompson, J. H. McNeill and C. Orvig, Chem. Rev., 1999, 99,
2561.
36 K. N. Raymond and V. C. Pierre, Bioconjugate Chem., 2005, 16, 3.
37 L. F. Cavalieri, Chem. Rev., 1947, 4, 525.
38 R. L. N. Harris, Aust. J. Chem., 1976, 29, 1329.
39 N. K. Dutt and U. U. M. Sarma, J. Inorg. Nucl. Chem., 1975, 37, 1801.
40 S. Ahmed, J. Burgess, B. Hurman and S. A. Parsons, Polyhedron, 1994,
13, 23.
62 A. F. Kirby and F. S. Richardson, J. Phys. Chem., 1983, 87, 2544.
63 S. J. P. Damment and M. Gill, J. Am. Soc. Nephrol., 2003, 14, FP0653.
64 A. W. K. Ng, T. Lukic, H. P. Pritchard and K. M. Wasan, Cardiovasc.
Drug Rev., 2003, 21, 151.
65 M. Yazdanian, S. L. Glynn, J. L. Wright and A. Hawi, Pharmacol. Res.,
1998, 15, 1490.
41 D. Fregona, G. Faraglia, R. Graziani, U. Casellato and S. Sitran, Gazz.
Chim. Ital., 1994, 124, 153.
66 H. Okhee, M. L. Failla, D. Hill, E. R. Morris and J. J. C. Smith, J. Nutr.,
1994, 124, 580.
67 V. D. Makhey, A. Guo, D. A. Norris, P. Hu, J. Yan and P. J. Sinko,
Pharmacol. Res., 1998, 15, 1160.
68 V. Risovic, K. Sachs-Barrable, M. Boyd and K. M. Wasan, Drug Dev.
Ind. Pharm., 2004, 30, 767.
69 M. Ochsenkuhn-Petropoulou, K. Ochsenkuhn and J. Luck, Spec-
trochim. Acta, Part B, 1991, 46, 51.
42 J. Fogh, J. M. Fogh and T. Orfeo, J. Natl. Cancer Inst., 1977, 59, 221.
43 M. N. Garcia-Casal, I. Leets and M. Layrisse, J. Nutr., 2000, 130, 5.
44 J. F. Vanchon and P. Beaulieu, Gastroenterology, 1992, 103, 414.
45 Z. Zhang, S. J. Rettig and C. Orvig, Inorg. Chem., 1991, 30, 509.
46 V. Monga, B. O. Patrick and C. Orvig, Inorg. Chem., 2004, 44, 2666.
47 M. M. Finnegan, T. G. Lutz, W. O. Nelson, A. Smith and C. Orvig,
Inorg. Chem., 1987, 26, 2171.
70 T. Mosmann, J. Immunol. Methods, 1983, 65, 55.
71 E. van Niekerk, J. F. O’Sullivan, G. K. Joone and C. E. J. van Rensburg,
Invest. New Drugs, 2001, 19, 211.
48 W. O. Nelson, T. B. Karpishin, S. J. Rettig and C. Orvig, Can. J. Chem.,
1987, 66, 123.
49 G. Faraglia, D. Fregona and S. Sitran, Main Group Met. Chem., 1994,
17, 649.
50 D. J. Clevette, D. M. Lyster, W. O. Nelson, T. Rihela, G. A. Webb and
C. Orvig, Inorg. Chem., 1990, 29, 667.
72 X.-P. Huang, M. Spino and J. J. Thiessen, Pharmacol. Res., 2006, 23,
280.
73 M. A. Santos, S. Gama, L. Gano, G. Cantinho and E. Farkas, Coord.
Chem. Rev., 2002, 228, 187.
51 C. Tedeshi, J. Azema, H. Gornitzka, P. Tisnes and C. Picard, Dalton
Trans., 2003, 1738.
74 F. C. Alves, P. Donato, A. D. Sherry, A. Zaheer, S. Zhang, A. J. M.
Lubag, M. E. Merritt, R. E. Lenkinski, J. V. Frangioni, M. Neves,
M. I. M. Prata, A. C. Santos, J. J. P. de Lima and C. F. G. C. Geraldes,
Invest. Radiol., 2003, 38, 750.
52 E. C. Alyea, A. Malek and A. E. Vougioukas, Can. J. Chem., 1981, 60,
667.
53 L. M. Brown and K. S. Mazdiuasni, Inorg. Chem., 1970, 9, 2783.
54 W. O. Nelson, T. B. Karpishin, S. J. Rettig and C. Orvig, Inorg. Chem.,
1988, 27, 1045.
55 N. Sabbatini and M. Guardigli, Coord. Chem. Rev., 1993, 123, 201.
56 G. R. Choppin and D. R. Peterman, Coord. Chem. Rev., 1998, 174, 283.
57 S. I. Weissman, J. Chem. Phys., 1942, 10, 214.
58 X.-P. Yang, B.-S. Kang, W.-K. Wong, C.-Y. Su and H.-Q. Liu, Inorg.
Chem., 2003, 42, 169.
75 L. Yu, D. Chen, J. Li and P. G. Wang, J. Org. Chem., 1997, 62, 3575.
76 J. Rudovsky, M. Botta, P. Hermann, A. Koridze and S. Aime, Dalton
Trans., 2006, 2323.
77 M. Wakamura, K. Kandori and T. Ishikawa, Colloids Surf., A, 2000,
164, 297.
78 A. Lebugle, F. Pelle, C. Charvillat, I. Rousselot and J. Y. Chane-Ching,
Chem. Commun., 2006, 606.
79 D. Choi and P. Kumta, J. Am. Ceram. Soc., 2006, 89, 444.
80 M. C. Chang and J. Tanaka, Biomaterials, 2002, 23, 4811.
81 M. E. Fleet and X. Liu, J. Solid State Chem., 2000, 149, 391.
59 W. D. Kim, G. E. Kiefer, F. Maton, K. McMillan, R. N. Muller and
A. D. Sherry, Inorg. Chem., 1995, 34, 2233.
5030 | Dalton Trans., 2007, 5019–5030
This journal is
The Royal Society of Chemistry 2007
©