In vitro studies of lanthanide complexes for the treatment of osteoporosis

Article abstract of DOI:10.1039/c2dt32373g

Lanthanide ions, Ln(iii), are of interest in the treatment of bone density disorders because they are found to accumulate preferentially in bone (in vivo), have a stimulatory effect on bone formation, and exhibit an inhibitory effect on bone degradation (in vitro), altering the homeostasis of the bone cycle. In an effort to develop an orally active lanthanide drug, a series of 3-hydroxy-4-pyridinone ligands were synthesized and eight of these ligands (H1 = 3-hydroxy-2-methyl-1-(2-hydroxyethyl)-4-pyridinone, H2 = 3-hydroxy-2-methyl-1- (3-hydroxypropyl)-4-pyridinone, H3 = 3-hydroxy-2-methyl-1-(4-hydroxybutyl)-4- pyridinone, H4 = 3-hydroxy-2-methyl-1-(2-hydroxypropyl)-4-pyridinone, H5 = 3-hydroxy-2-methyl-1-(1-hydroxy-3-methylbutan-2-yl)-4-pyridinone, H6 = 3-hydroxy-2-methyl-1-(1-hydroxybutan-2-yl)-4-pyridinone, H7 = 1-carboxymethyl-3-hydroxy-2-methyl-4-pyridinone, H8 = 1-carboxyethyl-3-hydroxy- 2-methyl-4-pyridinone) were coordinated to Ln³⁺ (Ln = La, Eu, Gd, Lu) forming stable tris-ligand complexes (LnL₃, L = 1^-, 2^-, 3^-, 4^-, 5^-, 6^-, 7^- and 8^-). The dissociation (pK_an) and metal ligand stability constants (log β_n) of the 3-hydroxy-4- pyridinones with La³⁺ and Gd³⁺ were determined by potentiometric titrations, which demonstrated that the 3-hydroxy-4-pyridinones form stable tris-ligand complexes with the lanthanide ions. One phosphinate-EDTA derivative (H₅XT = bis[[bis(carboxymethyl)amino]methyl]phosphinate) was also synthesized and coordinated to Ln³⁺ (Ln = La, Eu, Lu), forming the potassium salt of [Ln(XT)]^2-. Cytotoxicity assays were carried out in MG-63 cells; all the ligands and metal complexes tested were observed to be non-toxic to this cell line. Studies to investigate the toxicity, cellular uptake and apparent permeability (P_app) of the lanthanide ions were conducted in Caco-2 cells where it was observed that [La(XT)] ^2- had the greatest cell uptake. Binding affinities of free lanthanide ions (Ln = La, Gd and Lu), metal complexes and free 3-hydroxy-4-pyridinones with the bone mineral hydroxyapatite (HAP) are high, as well as moderate to strong for the free ligand with the bone mineral depending on the functional group.

Full text of DOI:10.1039/c2dt32373g

Dalton
Transactions
View Article Online
View Journal | View Issue
PAPER
In vitro studies of lanthanide complexes for the
treatment of osteoporosis†
Cite this: Dalton Trans., 2013, 42, 5999
Yasmin Mawani,a Jacqueline F. Cawthray,a Stanley Chang,a Kristina Sachs-Barrable,b
David M. Weekes,a Kishor M. Wasan*b and Chris Orvig*a
Lanthanide ions, Ln(III), are of interest in the treatment of bone density disorders because they are found
to accumulate preferentially in bone (in vivo), have a stimulatory effect on bone formation, and exhibit
an inhibitory effect on bone degradation (in vitro), altering the homeostasis of the bone cycle. In an
effort to develop an orally active lanthanide drug, a series of 3-hydroxy-4-pyridinone ligands were syn-
thesized and eight of these ligands (H1 = 3-hydroxy-2-methyl-1-(2-hydroxyethyl)-4-pyridinone, H2 =
3-hydroxy-2-methyl-1-(3-hydroxypropyl)-4-pyridinone, H3
= 3-hydroxy-2-methyl-1-(4-hydroxybutyl)-4-
pyridinone, H4 = 3-hydroxy-2-methyl-1-(2-hydroxypropyl)-4-pyridinone, H5 = 3-hydroxy-2-methyl-1-
(1-hydroxy-3-methylbutan-2-yl)-4-pyridinone, H6 = 3-hydroxy-2-methyl-1-(1-hydroxybutan-2-yl)-4-pyridi-
none, H7 = 1-carboxymethyl-3-hydroxy-2-methyl-4-pyridinone, H8 = 1-carboxyethyl-3-hydroxy-2-methyl-
4-pyridinone) were coordinated to Ln3+ (Ln = La, Eu, Gd, Lu) forming stable tris-ligand complexes (LnL3, L =
1−, 2−, 3−, 4−, 5−, 6−, 7− and 8−). The dissociation (pKan) and metal ligand stability constants (log βn) of
the 3-hydroxy-4-pyridinones with La3+ and Gd3+ were determined by potentiometric titrations, which
demonstrated that the 3-hydroxy-4-pyridinones form stable tris-ligand complexes with the lanthanide
ions. One phosphinate-EDTA derivative (H5XT = bis[[bis(carboxymethyl)amino]methyl]phosphinate) was
also synthesized and coordinated to Ln3+ (Ln = La, Eu, Lu), forming the potassium salt of [Ln(XT)]2−. Cyto-
toxicity assays were carried out in MG-63 cells; all the ligands and metal complexes tested were observed
to be non-toxic to this cell line. Studies to investigate the toxicity, cellular uptake and apparent per-
meability (Papp) of the lanthanide ions were conducted in Caco-2 cells where it was observed that
[La(XT)]2− had the greatest cell uptake. Binding affinities of free lanthanide ions (Ln = La, Gd and Lu),
metal complexes and free 3-hydroxy-4-pyridinones with the bone mineral hydroxyapatite (HAP) are high,
as well as moderate to strong for the free ligand with the bone mineral depending on the functional
group.
Received 7th October 2012,
Accepted 12th December 2012
DOI: 10.1039/c2dt32373g
www.rsc.org/dalton
fracture occurs, it affects 200 million worldwide, putting an
Introduction
escalating pressure on health care resources.1,2 Primarily com-
Osteoporosis, the most prevalent bone density disorder, is a posed of osseous tissue, bone is responsible for structural
skeletal disease that is characterized by low bone mineral support, protection of organs and storage of minerals in the
mass and a deterioration of bone tissue. Caused by an imbal- body. Bone is a dynamic tissue that constantly undergoes
ance between bone formation and resorption, it leads to an resorption and desorption by a tightly regulated cycle in which
increase in bone fragility and susceptibility of fracture. Often it is destroyed by osteoclast cells and regenerated by osteoblast
known as a “silent” disease because it is asymptomatic until cells, as seen in Fig. 1.3 At the cellular level in osteoporotic
bone, bone loss occurs because of a perturbation between the
activity of osteoclasts and osteoblasts.4,5
aMedicinal Inorganic Chemistry Group, Department of Chemistry, University of
The most widely used class of drugs for the treatment of
British Columbia, 2036 Main Mall, Vancouver, British Columbia, Canada V6T 1Z1.
osteoporosis, bisphosphonates, are stable analogues of pyro-
E-mail: orvig@chem.ubc.ca
bFaculty of Pharmaceutical Sciences, University of British Columbia, 2146 East Mall,
phosphates that have a strong affinity for bone apatite; these
agents inhibit bone resorption by reducing the recruitment
and activity of osteoclasts and increasing apoptosis.6,7 Due to
their poor lipophilicity, they possess very low oral bioavailabil-
ity, typically <1%. As a result, bisphosphonates must be
administered in a very high dosage to compensate for low
Vancouver, BC, Canada V6T 1Z3. E-mail: kwasan@interchange.ubc.ca
†Electronic supplementary information (ESI) available: Detailed experimental
procedures, ligand and metal complex characterization (Tables S.1–S.3), EC50
data (Tables S.4–S.5), stepwise acid dissociation equilibria (Fig. S.1), speciation
diagrams for metal complexes (Fig. S.2–S.3) and hydroxyapatite colourimetric
assay (Fig. S.4). See DOI: 10.1039/c2dt32373g
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 5999–6011 | 5999

View Article Online
Paper
Dalton Transactions
judicious ligand choice ought to positively influence biodistri-
bution towards a modest uptake.
Thus adjustments to the ligand structure have potential
to increase the oral bioavailability of the lanthanide ions.
A library of 3-hydroxy-4-pyridinones and one phosphinate-
EDTA ligand have been synthesized and complexed to lanthanide
ions (Ln = La, Eu, Gd, Lu) under the principle of producing an
orally active anti-osteoporotic agent that can pass through the
gastrointestinal (GI) tract and effectively deliver lanthanides
to bone.
Our previous studies focused on adjusting the ligand struc-
ture around Ln3+ (Ln = La, Eu, Tb, Yb) ions in order to increase
the oral bioavailability of lanthanides, so as to decrease
unwanted side effects.17 The suitability of these first gener-
ation lanthanide complexes for the treatment of bone resorp-
tion disorders was investigated in cellular uptake studies and
bidirectional transport assays in Caco-2 cells, a model of the
gastrointestinal (GI) lining. From these preliminary investi-
gations it was found that one ligand HL1 (3-hydroxy-
1,2-dimethyl-4-pyridinone) increased the lanthanide uptake
Fig. 1 Bone turnover cycle.5,16
bioavailability, resulting in GI upset7,8 and low patient compli- into cells. The complexed form, La(L1)3 (tris(1,2-dimethyl-3-
ance.9,10 Additionally, there is mounting evidence indicating oxy-4-pyridinonato)lanthanum(III)), was found to have low
that bisphosphonates cause osteonecrosis of the jaw, with 5% toxicity in vitro, and to have improved Caco-2 cell uptake com-
of all cases associated with oral bisphosphonates used to treat pared to other compounds investigated, including lanthanum
osteoporosis.11–13 Likewise studies in dogs indicated that carbonate.17 One of the shortfalls of La(L1)3 is its limited solu-
increased bone mineralization due to bisphosphonates leads bility in aqueous media, making it difficult to administer in
to an inability to repair microcracks and microdamage, result- in vitro and in vivo assays.
ing in more fragile bone.12,14,15
Optimizing the lipophilic/hydrophilic balance of the
Due to the limitations of current therapies, our work17 has 3-hydroxy-4-pyridinones by varying the R1-group (Fig. 2) has
focused on the development of 3-hydroxy-4-pyrones and pyridi- the potential to lead to further improvements in the GI uptake
none lanthanide complexes for the treatment of osteoporosis. of the lanthanides. As the binding affinity of the ligand varies
Lanthanide ions, which include the 14 4f-block elements as
well as lanthanum, abbreviated herein as Ln(III), are iso-
morphic replacements of Ca(II) ions,18,19 sharing similar ionic
radii, donor atom preferences, and almost identical coordi-
nation numbers in protein binding sites;20 therefore, Ln(III)
can potentially replace Ca(II) in bone, thus affecting the bone
remodeling cycle.
For decades it has been known that lanthanide ions have a
high affinity for bone,21 and it has been demonstrated that
they can have an inhibitory effect on osteoclasts22 and a proli-
ferative effect on osteoblasts.23 Biodistribution studies in rats
have shown that the heavier Ln(III) ions preferentially accumu-
late in bone rather than the liver or kidney, with 50–70%
depositing in bone, compared to less than 25% for the lighter
Ln.21,24 The elimination of Ln3+ from the bone is also slow,
with a half life of 2.5 years; elimination from soft tissues such
as the liver has a half life of about 15 days.21,24 A lanthanum
carbonate preparation, Fosrenol (La2(CO3)3), is a commercial
agent for treatment of hyperphosphatemia, and also inhibits
osteoclast activity while stimulating osteoblast proliferation.25
However, pharmacokinetic studies in animals have shown that
94–99% of the administered lanthanum carbonate passed
completely through the GI tract.26 This strong elimination
requires that high levels of the compound be administered, an
undesirable protocol for chronic use, and suggests that this study; Ln = La, Eu, Gd and Lu.
Fig. 2 A summary of the ligands and metal complexes, and their syntheses in
6000 | Dalton Trans., 2013, 42, 5999–6011
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
minimally with changes to the R group, the hydrophilicity– primary amine (Scheme 1, Route I). Generally, only the syn-
lipophilicity balance can be tailored, with small or negligible thesis with small alkylamines can be achieved by the direct
changes in their chelating properties, to optimize this property insertion method.29 It is known that the insertion of the
for delivery to the sites where the ligand is required.27–30 These unprotected 3-hydroxy-4-pyrones leads to longer reaction times
functional groups were carefully selected in order to tune the and difficulty isolating the product. As can be seen from
lipophilicity of the ligands, help direct the lanthanides Table S.1,† higher equivalents of the amine in the direct inser-
towards bone, and to increase the water solubility of the com- tion syntheses were required to obtain yields comparable to
plexes. Hydroxyl and carboxylate groups were selected because the benzyl protected route.
of their tendency to form hydrogen bonds and thus increase
The syntheses of the remaining 3-hydroxy-4-pyridinones,
the water solubility of the ligands and their complexes; H2, H3, H4, H5, H6, and H8, were prepared by a three step
hydroxyl groups have also been shown to increase the affinity process (Scheme 1, Route II): first the 3-hydroxy functional
of complexes to bone.31 The alkyl chain length was varied to group of the pyrone was protected with a benzyl-ether generat-
tune the lipophilicity of the ligands.28
ing 3-benzyl-2-methyl-4-pyrone Bnma via a Williamson ether
Additionally, one phosphinate-EDTA derivative (Fig. 2) was synthesis of maltol and benzylchloride.32 The Bnma was then
studied; this ligand was chosen to compare a hexadentate reacted with the appropriate primary amine, affording Bn2,
ligand system to that of the bidentate 3-hydroxy-4-pyridinone Bn3, Bn4, Bn5, Bn6, and Bn8, and finally the protecting
system. It was also selected due to the high affinity in vivo for group was removed to yield the free 3-hydroxy-2-methyl-
phosphonate and phosphinate functional groups for bone.
Herein the synthesis of a library of 3-hydroxy-4-pyridinone
4-pyridinones.
The syntheses of Bn2, B3 and Bn8 have been previously
complexes and one phosphinate-EDTA derivative and their reported by Dobbin et al.33 and were obtained in yields com-
respective lanthanide complexes are explored. The protonation parable to literature reports; Bn4 and Bn6 were all obtained in
constants of the free ligands and formation constants with moderate yields. Bn6 required a higher equivalent of amine
La3+ and Gd3+ are also determined. The lipophilicity of the than did the other syntheses—this is mainly due to the steric
ligands, the toxicity of the ligands and complexes in MG-63 hindrance of the nucleophile compared to the other amines,
cells and Caco-2 cells are explored, along with the cell uptake as can be seen in Table S.1.† Due to poor conversion rates
and bifunctional transport in Caco-2 cells. Finally, the inter- from the pyrone to the pyridinone, the excess primary amine
action of the metal complexes with hydroxyapatite (HAP) are made separation of Bn5 from the starting materials difficult.
discussed. A summary of the ligands and metal complexes can
be seen in Fig. 2.
Once the key desired functional group was inserted into the
pyridinone ring, the 3-benzyloxy pyridinones could be depro-
tected to afford the free 3-hydroxy pyridinones, as shown in
Scheme 1, Route II. Ligands H2, H3, H4, H5 and H6 were
accessed by the palladium on carbon (P/C) catalyzed hydrogen-
ation of Bn2, Bn3, Bn4, Bn5 and Bn6, respectively. For some of
the ligands, in order to increase the aqueous solubility, a
small amount of hydrochloric acid was added. Due to limited
solubility of Bn8 in water, methanol or ethanol under neutral
or acidic conditions, Bn8 was deprotected using HBr in glacial
acetic acid.
Results and discussion
Synthesis of 3-hydroxy-4-pyridinones
The syntheses of 3-hydroxy-2-methyl-4-pyridinones functiona-
lized with carboxy and hydroxyl substituents were achieved by
one of two routes – the direct ammonolysis of maltol, or the
ammonolysis of benzyl-protected maltol, as seen in Scheme 1.
The syntheses of H1 and H7 were accessed by the direct
ammonolysis of the pyrone, maltol, with the appropriate
The ligands were obtained in moderate to good yields, with
the exception of H5. This is likely due to its poor solubility in
Scheme 1 General synthesis of a 3-hydroxy-4-pyridinone. Route I: (a) NaOH, H2O, reflux 24–70 h. Route II: (a) benzyl chloride, NaOH, methanol, reflux 40 h;
(b) NaOH, ethanol/water or methanol, reflux 15–92 h; (c) HBr in acetic acid 33% w/v or H2(g), 10% P/C.
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 5999–6011 | 6001

View Article Online
Paper
Dalton Transactions
water and organic solvents. In the final step of the hydrogen-
ation, the crude mixture must be filtered in order to remove
the hydrogenation catalyst. Due to the limited solubility of H5,
a significant amount of the ligand likely remained in the Pd–C
mixture. Four literature 3-hydroxy-4-pyridinones were syn-
thesized (H1, H2·HCl, H3·HCl and H8), along with three novel
3-benzyloxy-4-pyridinone ligands (H4, H5 and H6).
Synthesis and characterization of tris(pyridinonato)-
lanthanide(III) complexes
All lanthanide pyridinone complexes were synthesized by the
deprotonation of the 3-hydroxy of the pyridinone with sodium
hydroxide in the presence a lanthanide(III) nitrate hexahydrate
salt (Scheme 1). The complexes were synthesized in water,
with the exception of La(5)3; due to the limited solubility of H5
in water, La(5)3 was synthesized in methanol. The ligands H1
and H7, were not soluble in water at room temperature, thus
they were dissolved by the addition of NaOH; the pH was moni-
tored so it was not above 4 to prevent hydrolysis of the metal
ion. Once the ligand was dissolved, the desired lanthanide salt
was added to the solution, and the pH was carefully adjusted
to 9.5–10.5.
Fig. 4 Solid state IR spectra for, top to bottom: H4, La(4)3, Eu(4)3, Gd(4)3 and
Lu(4)3; the spectra are offset for clarity.
the Ln(4)3 complexes can be seen in Fig. 4. A characteristic
four-band IR spectral pattern is observable at 1650–1460 cm−1
in the free ligands; it can be attributed to the pyridinone ring
and CvO stretching frequencies.35,36 Due to strong coupling,
the νring and νCvO stretching modes are indistinguishable
from one another, thus the bands are collectively assigned to
νring and νCvO in Table S.3.† The metal complex spectra were
dominated by peaks corresponding to the ligand (Fig. 4 and
Table S.3†). Upon complexation, the characteristic four-band
IR spectral pattern is preserved with a bathochromic shift
around 30–50 cm−1 and a possible reordering upon formation
of the tris(3-oxy-4-pyridinonato)lanthanide complexes. When
Complexation of the ligands was attempted with La3+, Eu3+,
and Lu3+ and in most cases Gd3+, with the exception of H5 for
which only the lanthanum complex was synthesized. The 1H
NMR spectra of the lanthanum and lutetium complexes were
useful in elucidating the structure of the metal complexes.
As seen in both Fig. 3 and Table S.2† the resonances for the
ring hydrogens, Ha and Hb in the pyridinone metal complexes
are shifted upfield compared to those of the free ligands.
Similar upfield shifts have been previously reported in our
lab.17 In the 1H NMR spectra of the metal complexes, broad
peaks were observed – this is attributed to the coordination of
water in the coordination sphere, and to rapidly interconvert-
ing Δ and Λ stereoisomers and geometric fac and mer isomers
at RT as well as both the Δ and Λ stereoisomers.34
coordinated to
a lanthanide ion, the νO–H stretch at
3300–3000 cm−1 observed in the free ligand is replaced by a
broad peak due to the coordination of waters in the lattice of
the metal complex. The most distinctive characteristics upon
complexation to a lanthanide ion are the new peaks between
400–550 cm−1 assigned to the formation of metal–oxygen
(νM–O) stretching frequencies.37 The IR spectra for the La3+,
Eu3+, Gd3+ and Lu3+ complexes were all nearly superimposable
within each ligand set, indicating that the metal complexes are
isostructural. The <10 cm−1 variation observed between the
different metals is attributed to the size difference between the
lanthanide ions. In the IR spectra of H8 and H9, very broad
bands are observed around 3200 cm−1, attributed to the strong
intermolecular hydrogen bonding of the carboxyl groups.38
The IR spectra of all of the complexes were also helpful in
confirming the formation of the complexes. Table S.3† gives a
summary of the IR stretching frequencies of both the free
ligands and the metal complexes, while IR spectra for H4 and
Synthesis of a hexadentate phosphinate-EDTA derivative
coordinated to lanthanide ions
Lanthanide ions were coordinated to H5XT as previously
reported by our laboratory.39 K2[Ln(XT)] was synthesized as the
phosphinate functionality of the ligand could potentially help
the complex localize towards bone, as in 153Sm3+-EDTMP the
bone seeking agent used in bone palliative care.40,41 Likewise,
it was also synthesized so as to compare one hexadentate
ligand versus three bidentate 3-hydroxy-4-pyridinones co-
ordinated to the lanthanide ions.
Fig. 3 1H NMR (400 MHz, D2O, RT) spectra of H1 (top), La(1)3 (middle) and
Lu(1)3 (bottom).
6002 | Dalton Trans., 2013, 42, 5999–6011
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
evidence of a 4 : 1 [LnL4]− complex was observed with either
Table 1 Acid dissociation constants (pKa1, pKa2, pKa3) for the 3-hydroxy-4-pyri-
dinone ligands HL1, H4 and H8 at 25 °C and 0.16 M NaCl. Numbers in paren-
theses represent SD between replicates
La3+ or Gd3+ ions.
½MLðn3ꢀnÞþꢁ
Constant
HL1
H4
H8
βn ¼
K4 ¼
ð1Þ
ð2Þ
n
½M3þꢁ½Lꢀꢁ
a
pKa1
pKa2
pKa3
3.63(1)
9.71(1)
3.11(3)
9.58(2)
2.86(5)
3.99(4)
9.64(3)
½ML3ðOHÞꢀꢁ
a
a
½ML3ꢁ½OHꢀꢁ
a Where pKan = −log[Kan], n = 1, 2, 3.
Toxicity assay in MG-63 cells
MG-63 cells are derived from a human osteosarcoma47 and
were selected for cytotoxicity studies as they are considered to
behave like osteoblasts.48 Considering the vital role that osteo-
blasts play in the bone cycle (Fig. 1) and in the maintenance of
bone mass, it is essential that compounds used for the poten-
tial treatment of osteoporosis are not toxic to MG-63 cells.
A colourimetric MTT assay was utilized in order to quantify
the toxicity of these compounds.49 Cisplatin was used as a
positive control to induce cell death, verifying the assay.
Cell viability was determined relative to the negative
control, 100% cell medium, or 0.25% medium in DMSO. The
half maximal effective concentration (EC50) was calculated for
cisplatin (Fig. 5a), confirming the assay was effective at deter-
mining toxicity; however, with the exception of La(2)3, none of
the synthesized ligands or Ln(III) complexes tested demon-
strated any significant toxicity with the cell line at the concen-
trations tested. As a result, EC50 could not be calculated for
the compounds tested. Tables S.4 and S.5† summarize the
toxicities of the tested ligands and metal complexes in
Table 2 Log metal–ligand stability constants (βn) and the ML3OH metal–
hydroxide stability constant (log K4) at 25 °C and 0.16 M NaCl. Numbers in
parentheses represent SD between replicates
Constant
Equation
La3+
Gd3+
Log β1
Log β2
7.06(5)
8.15(4)
12.6(1)
17.5(1)
7.94(6)
14.60(6)
20.07(9)
9.1(1)
Log β3
Log K4
Determination of stability constants of 3-hydroxy-4-
pyridinones with lanthanide ions
The determination of the stability constants allows for the pre-
diction of the thermodynamic stability of complexes in vitro
and in vivo at physiologically relevant pHs. The two stepwise
acid dissociation constants (pKan) for HL1 and H4 (pKa1 and
pKa2) and the three stepwise acid dissociation constants for
H8 (pKa1, pKa2 and pKa3), shown in Fig. S.1,† were determined
for the amphoteric 3-hydroxy-4-pyridinones and are reported
in Table 1.
MG-63 cells, respectively. The low toxicity profiles (EC50
>
100 μM for all tested ligands and complexes) of the syn-
thesized compounds makes them of interest for the treatment
of osteoporosis.
Partition coefficients
The acid dissociation constants calculated for HL1 agree
quite well with those reported previously in the literature.42–44
Likewise, the pKa2 of HL1 and H4 and the pKa3 of H8 are
similar as seen in Table 1, indicating that the R group does
not significantly alter the ionizability of the 3-hydroxyl
group.30 It also indicates that the deprotonated carboxyethyl
group of H8 does not significantly affect the electron density
of the pyridinone ring.
As the lipophilicity of a compound is such an important factor
in drug screening,50 the partition coefficients of the free
ligands were determined by the shake-flask method.51
It should be noted that the lipophilicity of the free ligand (HL)
compared to complexed ligand (Ln(L)3) would likely be
The stability constants (log βn) of HL1 with La3+ and Gd3+
reported in Table 2 were determined using eqn (1) and (2). Not
surprisingly, the binding constant of the ligand with Gd3+ is
higher than that of La3+. This can easily be explained by the
larger ionic radius of La3+ (103.2 pm, CN = 6)45 versus that of
Gd3+ (93.8 pm, CN = 6).45 While the 3-hydroxy-4-pyridinones
are known to form stable tris-ligand complexes with other tri-
valent cations such as Ga3+ 42 In3+ 42 and Fe3+,43,44 the larger
,
coordination sphere of the lanthanides, and their preferred
coordination geometry of 8 or 9,46 suggests the possibility of
the formation of 4 : 1 complexes for HL1 with Ln3+ ions. As a
result, excess of ligand was titrated in the presence of the
lanthanides to achieve a 4 : 1 ligand to metal ratio but no
Fig. 5 Sample survival plots of MG-63 cells exposed to varying concentrations
of compound for 48 h, monitored by the MTT assay, n = 3, error bars indicate
SD of the cell viability. The red line indicates the concentration at which 50%
of the cells are no longer viable (EC50); (a) cisplatin (positive control),
(b) [La(XT)]2−
.
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 5999–6011 | 6003

View Article Online
Paper
Dalton Transactions
Table 3 Log Po/w coefficients of 3-hydroxy-4-pyridinones, in order of decreas-
ing lipophilicities
Compound
Functional group (R1)
Log Po/w
H6
H3
HL1
H4
H2
H1
H8
H7
Secbutyl–OH
Butyl–OH
Methyl
Isopropyl–OH
Propyl–OH
Ethyl–OH
−0.42
−0.76
−0.77
−0.82
−0.86
−1.08
−1.87
−1.95
Ethyl–carboxylate
Methyl–carboxylate
markedly different; however, as only the R1 group of the
3-hydroxy-4-pyridinones are varied, a quick screen was accom-
plished by the experimental determination of log Po/w of just
the free ligands (eqn (3), Table 3).
Fig. 6 Cell uptake of lanthanide complexes, La(2)3, La(3)3, La(6)3, [La(8)3]3−
,
[La(9)3]3− and [La(XT)]2− in Caco-2 cells, after 24 h exposure to compounds.
Uptake data are reported as a mean percentage of the concentration of μM of
Ln3+ taken up by the cells per mg of protein SD, n ≥ 3.
!
½soluteꢁoctanol
log Po=w ¼ log
ð3Þ
½soluteꢁwater pH 7:4
There are two ligands, H6 and H3 with a log Po/w greater
than that of HL1, the ligand from the first generation studies
that when complexed to La3+ was found to have the greatest
cell uptake and apparent permeability in Caco-2 cells;17
another two ligands, H4 and H2 have lipophilicities similar to
that of HL1. Thus these ligands could potentially have similar
or better cell uptake and apparent permeabilities than HL1.
permeable membrane filter, the cells form distinct apical (top)
and basolateral (bottom) membranes. It is the apical surface
that possesses the brush border and transport systems that are
characteristic of the small intestine, discussed above.53 Thus
Caco-2 cells, when plated in transwell plates, can be used to
determine the in vitro transepithelial permeability coefficients
or apparent permeability (Papp) of a drug. Several studies have
demonstrated that the Papp (A → B) across Caco-2 monolayers
can be used to estimate the in vivo oral absorption of
compounds.54,55,58
Caco-2 cell studies
For the purposes of this study, first the toxicities of the com-
pounds in Caco-2 cells were evaluated by an MTS assay.52
Triton-X 100 was used as a positive control to induce cell
death. Much like the low toxicities observed for the metal com-
plexes in the MG-63 cell line (Tables S.4 and S.5†), low cyto-
toxicity profiles were also observed in the Caco-2 cell line. EC50
values for La(1)3, La(2)3, La(3)3, La(4)3, La(6)3, Na3[La(7)3],
Na3[La(8)3] and K2[La(XT)] could not be calculated as all the
compounds tested had >80% cell viability at a concentration of
1000 μM.
To evaluate the in vitro absorption of the lanthanide com-
plexes, cell uptake and permeability studies in Caco-2 cell
monolayers were carried out. As the GI tract remains the most
acceptable route for the administration of drug formulations,
it is important to evaluate if these drugs can cross the intesti-
nal barriers. It has been demonstrated that Caco-2 cells,
derived from a human colon adenocarcinoma,53 have similar
in vitro permeability characteristics as human intestinal
tissue.54 Under normal cell culture conditions, these epithelial
cells spontaneously differentiate into enterocytes and form
polarized monolayers.53,55 Several characteristics that mimic
the small intestine are expressed; the Caco-2 cell monolayers
possess well-developed brush border transport systems,
enzymes, ion and nutrient channels.56,57 The cell monolayers
have transepithelial electrical resistance (TEER) values of
approximately 300 Ω cm−2, similar to those found in the
colon.55 When cultured in transwell plates that contain a
In the Caco-2 cell uptake studies, as seen in Fig. 6, it was
observed that one compound, K2[La(XT)] was found to have
the greatest uptake. The uptake of K2[La(XT)] (15.26 0.17%)
is comparable to that determined for the lead compound from
the first generation complexes, La(L1)3 (9.07 2.33%).17 More-
over, K2[La(XT)] and La(3)3 had greater cell uptake than did
the benchmark La2(CO3)3 compound, determined in previ-
ous studies.17 The lanthanide complexes, Na3[La(7)3] and
Na3[La(8)3]3 had the poorest cell uptake, which correlated well
with the log Po/w values determined for the ligands H7 and H8
(Fig. 6). The partition coefficients for H6 and H3 suggested
that La(6)3 would have better cell uptake than La(3)3; however,
as can be seen in Fig. 6, the opposite was found to be true.
It is theorized that the increased uptake of [La(XT)]2−, com-
pared to that of the lanthanide pyridinone complexes, has to
do with the thermodynamic stability of the complex at physio-
logical pH. A look at the speciation plot of La(L1)3 (Fig. S.2†),
shows that at pH = 7.4 the complex is mainly in its 1 : 1 L : La3+
and 2 : 1 L : La3+ forms, with only a small amount in the
3 : 1 L : La3+ form. Conversely, in the speciation plot39 of
[La(XT)]2− (Fig. S.3†) it is observed that at a pH of 7.4
the complex is intact. Thus, the increased stability of the
[La(XT)]2− complex may result in more of the lanthanide being
taken up by the cell, because the entire complex stays intact.
The A → B (apical to basolateral) transport in Caco-2 cells
was also evaluated by introducing the lanthanide complexes to
6004 | Dalton Trans., 2013, 42, 5999–6011
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
Fig. 8 Lanthanide metal complexes hydroxyapatite (HAP) binding studies ana-
lyzed by ICP-MS. Reported as the mean percentage SD of Ln(III) bound to HAP,
n = 3.
Fig. 7 Apparent permeability (Papp), apical to basolateral transport, of
La2(CO3)3, La(1)3, La(4)3, [La(7)3]3− and [La(8)3]3− in Caco-2 cells after 24 h of
exposure to the lanthanide complexes reported as Papp cm s−1 SD, n ≥ 3.
was greater than the amount seen to accumulate in bone
in patients who are treated with lanthanum carbonate for
1 year.60,61
the apical side of the cell monolayer in transwell plates. After
an incubation time of 24 h, samples from the basolateral and
apical side were analyzed for La3+ content by ICP-MS. The
apparent permeability, Papp, was calculated by the determining
the concentration of lanthanide in the basolateral side by
ICP-MS at time point zero and at 24 h. The total lanthanide
concentration added in the donor chamber (apical) was calcu-
lated by the addition of the amount of lanthanide content
found in the apical, basolateral and cell lysate samples.
While it is clear that the lanthanide ions are interacting
with HAP, it is not clear if this is just a surface interaction, or
if the Ln(III) is being incorporated into the HAP crystal lattice.
Thus xylenol orange was used as an indicator for lanthanide
binding. Xylenol orange is a colorimetric indicator of heavy
metals,62 and is well known as an indicator for free lantha-
nides.63,64 In hexamethylenetetramine buffer, the ideal pH
range for this heavy metal indicator is pH 4.5–6.0.65
Apparent permeabilities were calculated for [La(7)3]3−
,
[La(8)3]3−, La(1)3 and La(4)3, and compared to La2(CO3)3. The
compound with the highest Papp value (Fig. 7) was found to be
La(4)3, which had Papp = (0.12 0.02) × 10−6 cm s−1. The Papp
of La(4)3 was found to be an order of magnitude lower, com-
pared to Papp calculated of the lead compound identified from
the first generation complexes, La(L1)3.17 However, it should
be noted that Papp of La(4)3 was still found to be 700 times that
of La2(CO3)3, so a significant increase in bioavailability was
achieved. Likewise, La(4)3 has an apparent permeability three
times that of the anti-osteoporotic drug alendronate.59
Under acidic conditions, xylenol orange is a deep red/
purple colour when bound to free lanthanides. In contrast,
unbound xylenol orange is a yellow-orange colour. Solutions of
Ln-HAP supernatant, acid-digested Ln-HAP, undigested Ln-
HAP, HAP (negative control), La(NO3)3 (positive control) were
buffered to pH 5.0 and a solution of xylenol orange, buffered
to pH 5.0 was introduced into each sample. As seen in
Fig. S.4† the digested HAP is a deep red/purple, while the un-
digested HAP and supernatant samples are pale yellow; this
assay confirms the ICP-MS results, showing a very small
amount of free lanthanide in the supernatant from the un-
digested HAP sample. Likewise the absence of the deep red/
purple colour in the undigested HAP sample, and the appear-
ance of the red/purple upon digestion of the HAP indicates
that the lanthanides are being incorporated into the HAP.
As it is clear that the lanthanides bind to HAP, and since
these were much higher values (180×) compared to what was
observed with the in vivo administration of La2(CO3)3,60,61 it
was important to study if the binding of lanthanides to the
HAP caused structural or physical changes to the material.
This assay was thus carried out on larger scale (0.5 g HAP vs.
20 mg of HAP) so there was enough to analyze HAP samples by
TGA, PXRD, and FTIR.
Lanthanide complex binding studies with hydroxyapatite
Hydroxyapatite (HAP) was used as an in vitro model of bone;
compounds were incubated under physiological conditions for
different time points to evaluate the binding affinity of the
Ln(III). ICP-MS was then used to measure the metal content in
both the supernatant and the HAP pellet. As can be seen in
Fig. 8, all the metals had near quantitative binding of the
lanthanide to hydroxyapatite, with the exception of La(4)3,
with >98% to the HAP after 24 h. The high percentage of
lanthanide bound correlates to our previous work.17 This indi-
cates that the binding kinetics are quite fast, with >90% of the
metal or metal–ligand complex bound to HAP after 5 minutes.
The lanthanide complexes, La(4)3, Na3Eu(7)3, Na3La(8)3 and
K2[La(XT)] were all tested at very low concentration, 2 μM. This
concentration was chosen firstly because it is well below the
respective EC50 values which are reported in Table S.5,†
meaning they are non-toxic at this concentration. Secondly, we
wanted to ensure the amount that could interact with the HAP
As can be seen in Fig. 9a there is not a significant change
in the FTIR of HAP with the metal or free ligand. The PXRD
spectra shown in Fig. 9b also have no observable change; the
peaks are all the same width and the d-spacing has not been
altered by the addition of free ligand or the metal complex.
Fig. 9c shows there is no change in the TGA—the small (∼3%)
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 5999–6011 | 6005

View Article Online
Paper
Dalton Transactions
Fig. 9 Physical spectra of HAP sample (blue), HAP-control (red), HAP-free ligand (green) and HAP-metal (purple). Figure (a) FTIR of HAP samples before TGA,
(b) XRD spectra of samples before TGA, (c) TGA of HAP samples and (d) FTIR of HAP samples after TGA. The FTIR and PXRD spectra are offset for clarity.
weight loss is likely due to the loss of 1 or 2 equivalents of were purchased from Sigma-Aldrich, and the primary amines
water. Lastly Fig. 9d shows there is no change in the FTIR after were purchased either from Sigma-Aldrich or Acros Organics.
the TGA when compared to the FTIR spectrum pre-TGA. With Lanthanum nitrate, europium nitrate, gadolinium nitrate and
no observable change in the FTIR or the PXRD, it is obvious lutetium nitrate were purchased from Sigma-Aldrich and Alfa
that the crystallinity of the HAP is not affected by the addition Aesar as their hexahydrates and used without further purifi-
of the metal complex or free ligand.
cation. Hydrobromic acid (33% w/v in glacial acetic acid) and
sodium hydroxide solution (50%) were obtained from Acros
Organics. For ICP-MS studies the lanthanide standards
(1000 μg mL−1 in 2% HNO3) were purchased from High Purity
Standards (North Charleston, SC, U.S.).
Experimental
Materials
Both Caco-2 and MG-63 cells were obtained from American
All solvents used for the synthesis of the ligands and metal Type Culture Collection (ATCC, Rockville, MD, U.S.). Minimum
complexes were HPLC grade and purchased from Fisher Scien- essential media alpha (MEM-α), fetal bovine serum (FBS),
tific, as were anhydrous sodium sulphate, hydrochloric acid, 0.25% trypsin-EDTA, penicillin–streptomycin–neomycin 100×
potassium hydroxide, sodium chloride, sodium hydroxide and (Pen–Strep), phosphate buffer saline solution (PBS), Dulbec-
formaldehyde (aqueous solution 37% w/w), centrifuge tubes co’s modified Eagle’s medium (DMEM), fetal bovine serum
and Opitma nitric acid (a high purity acid containing the least (FBS), bovine serum albumin (BSA) and Hank’s balanced salt
metal content of any commercially available acid, used in trace solution (HBSS) were purchased from Life Technologies (Bur-
metal analysis). Water was purified using an Elgastat Maxima lington, Ontario, Canada). T-75 culture flasks, transwell
HPLC reverse osmosis and deionization system or a PureLab (12 well) and polycarbonate (12, 24 and 96-well) plates were
Ultra system (Elga, Bucks, England). All water used was type 1, purchased from Corning (Corning, NY, U.S.). Sterile 15 mL and
18.2 MΩ cm, treated with a full spectrum UV to control bac- 50 mL centrifuge tubes were purchased from Fisher Scientific.
terial levels. Maltol, 3-hydroxy-1,2-dimethyl-4-pyridinone (HL1, Barrier pipette tips were purchased from Diamed (Mississauga,
deferiprone), iminodiacetic acid and hypophosphorous acid Ontario, Canada). 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyl-
(50% w/w aqueous solution) and La3+ and Gd3+ atomic absorp- tetrazolium bromide (MTT) was purchased from Alfa Aesar and
tion (AA) standards, hydroxyapatite, xylenol orange, 4-(2-hydro- cis-diamminedichloroplatinum(II) (cisplatin) was obtained
xyethyl)-1-piperazineethanesulfonic acid (HEPES) and 1-octanol from Acros Organics. MTS Cell Titer 96® Aqueous One
6006 | Dalton Trans., 2013, 42, 5999–6011
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
Solution Cell Proliferation Assay and CytoTox96® Non-radio- 1-(2-hydroxypropyl)-4-pyridinone (Bn4), 3-benzyloxy-2-methyl-
active Cytotoxicity Assay were from Promega (Madison, WI, 1-(1-hydroxy-3-methylbutan-2-yl)-4-pyridinone (Bn5), 3-benzyl-
U.S.). Bicinchoninic acid assay (BCA) protein analysis kit was oxy-2-methyl-1-(1-hydroxybutan-2-yl)-4-pyridinone (Bn6) and
provided by Pierce (Rockford, IL, U.S.), while the lysis buffer, 1-carboxyethyl-3-benzyloxy-2-methyl-4-pyridinone (Bn8). The
radioimmunoprecipitation assay (RIPA) was purchased from synthesis of 3-hydroxy-2-methyl-1-(3-hydroxypropyl)-4-pyridi-
Sigma-Aldrich.
none hydrochloride (H2·HCl), 3-hydroxy-2-methyl-1-(4-hydroxy-
butyl)-4-pyridinone hydrochloride (H3·HCl), 3-hydroxy-
2-methyl-1-(2-hydroxypropyl)-4-pyridinone (H4), 3-hydroxy-
Instrumentation
Elemental analyses for C, H, and N were completed on a 2-methyl-1-(1-hydroxy-3-methylbutan-2-yl)-4-pyridinone (H5),
Fisons EA 1108 instrument while high resolution mass spec- 3-hydroxy-2-methyl-1-(1-hydroxybutan-2-yl)-4-pyridinone (H6)
trometry (HRMS) electrospray ionization (ESI) was run on a and 1-carboxyethyl-3-hydroxy-2-methyl-4-pyridinone (H8) were
Micromass LCT. Low resolution mass spectra (MS) were synthesized by the deprotection of Bn2, Bn3, Bn4, Bn5, Bn6
obtained on a Bruker Esquire Ion Trap ESI-MS spectrometer. and Bn8, according to a modification of literature.33
1H and 13C nuclear magnetic resonance (NMR) spectra were
General procedure for the synthesis of tris(pyridinonato)
lanthanide complexes (Scheme 1)
recorded at room temperature using a Bruker AV-300 or AV-400
spectrometer. Fourier transform infrared (FTIR) spectra were
obtained on a Nicolet 6700 FTIR equipped with a Smart Orbit The free 3-hydroxy-4-pyridinone (H1, H2, H3, H4, H5, H6, H7
diamond attenuated total reflectance attachment. Lanthanide or H8; 3.0 equiv.) was suspended in water or methanol
concentration was analyzed on a Perkin Elmer Sciex Elan 6000 (10 mL); Ln(NO3)3·6H2O (1.0 equiv.) was added to the ligand
inductively coupled plasma mass spectrometer (ICP-MS). Ultra- solution and the pH of the mixture was raised slowly over
violet-visible (UV-vis) spectrum measurements were taken on a 10–15 min with the dropwise addition of 1 M NaOH and the
Varian Vary 100 Bio UV-vis. A Savant vacuum concentrator resulting mixture was continuously stirred for 1 h. The solu-
Speedvac Plus, model SC110A, was used at the highest temp- tion was concentrated under reduced pressure, redissolved in
erature setting of 60 °C to concentrate the samples prior to 1–2 mL of water and precipitated out with the addition of
acid digestion for ICP-analysis.
acetone (50 mL). The metal complex was collected by centrifu-
The absorbances of the plates for the MTT studies were gation (4000 RPM); the supernatant was discarded and the
measured at room temperature at 595 nm using a Beckman complex was dried in vacuo. The synthesis of bis[[bis(carboxy-
Coulter DTX 800 Multimode Detector. For the BCA and MTS methyl)amino]methyl]-phosphinate (H5XT·HCl) and its
assays the absorbances were measured at room temperature respective metal complexes, K2[Ln(XT)] were according to lit-
on a Multiskan Ascent Multi-plate reader from Labsystems at erature procedures.39 Detailed synthetic procedures and
wavelengths of 492 nm and 540 nm, respectively. The trans- characterization of all ligands and complexes can be found in
epithelial electrical resistance (TEER) values were monitored the ESI.†
using a Millipore Millicell-ERS (Bedford, MA, U.S.). An Innova
44 incubator shaker (New Brunswick Scientific, Enfield, CT,
Potentiometric titrations
U.S.) was used at 37 °C and 225 RPM to incubate the HAP Potentiometric equilibrium measurements of the 3-hydroxy-
samples. Ultraviolet-visible (UV-vis) spectrum measurements 4-pyridinones HL1, H4 and H9 and stability constant determi-
were acquired on a Varian Cary 100 Bio UV-Vis. A Fisher Scien- nations of HL1 with La3+ and Gd3+ were performed on a
tific Minicentrifuge was used to separate HAP from super- Metrohm Titrando 809 equipped with a Ross combination pH
natant and a Savant vacuum concentrator Speedvac Plus, electrode and a Metrohm Dosino 800 automatic burette. The
model SC110A, was used at the highest temperature setting of titration apparatus consisted of a 10 mL water-jacketed glass
60 °C to concentrate HAP and supernatant samples prior to vessel maintained at 25.0 0.1 °C (Julabo water bath). Nitro-
acid digestion for ICP-MS analysis. Powder X-ray diffraction gen, purified through a 10% NaOH solution to exclude any
samples were run on a Bruker AXS D8 Advance powder X-ray CO2, was passed through the solution prior and during the
diffractometer, wavelength of CuKα (1.54 Å). For thermogravi- titrations. The electrode was calibrated using a standard HCl
metric analysis (TGA), samples were analyzed on a Perkin solution before each potentiometric equilibrium study. Data
Elmer Pyris 6 thermogravimetric analyser.
was collected in triplicate using PC Control (Version 6.0.91,
Metrohm), and were analyzed by standard computer
treatment program (MacCalib) to obtain the calibration
a
Synthesis of ligands and metal complexes (Scheme 1)
3-Hydroxy-2-methyl-1-(2-hydroxyethyl)-4-pyridinone (H1)66 and parameter E0.
1-carboxymethyl-3-hydroxy-2-methyl-4-pyridinone (H7)38 were
A NaOH solution (0.15 M) was prepared from the dilution
synthesized according to modified literature procedures. of 50% NaOH (Acros Organics) with freshly boiled 18.2 MΩ cm
3-Benzyloxy-2-methyl-4-pyrone (Bnma), synthesized according water under a stream of purified N2(g), producing carbonate-
to our literature,32 was used to produce the intermediates free solution. The solution was standardized against freshly
3-benzyloxy-2-methyl-1-(3-hydroxypropyl)-4-pyridinone hydro- recrystallized potassium hydrogen phthalate. Titration of the
chloride (Bn2·HCl), 3-benzyloxy-2-methyl-1-(4-hydroxybutyl)- NaOH solution with a standard HCl solution was performed
4-pyridinone hydrochloride (Bn3·HCl), 3-benzyloxy-2-methyl- regularly to determine the extent of carbonate accumulation in
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 5999–6011 | 6007

View Article Online
Paper
Dalton Transactions
the base solution and determined by plotting (Vo + Vt) × 10−pH To determine the toxicity of La(L1)3, due to its limited aqueous
versus Vt (a Gran plot).67
solubility, a 0.25% DMSO/medium solution was used to test
Determination of the acid dissociation constants was the complex; also as a result of limited solution, the highest
achieved by the titration of base into the ligand in acid. Ligand concentration of La(L1)3 tested was 600 μg mL−1. As a positive
solutions (0.0010 M) were prepared by dissolving the ligands control cisplatin (cis-diamminedichloroplatinum(II)) was incu-
in a known amount of acid with an ionic strength of bated at concentrations of 1, 5, 10, 25, 50, 100, 250, 500 and
0.16 M NaCl. Potentiometric titrations were performed by the 1000 μM in 0.25% DMSO/medium. As negative controls, 100%
automatic addition of base (CO2 free NaOH) into the ligand medium and 0.25% DMSO/medium were incubated with
solutions. While the equilibrium was rapidly established in the cells. For each compound concentration, the assay was
less than 10 min, up to 15 min was permitted for stabilization performed in triplicate.
between each addition of base.
After 48 h of treatment a 50 μL aliquot of MTT (2.5 mg
The metal–ligand stability constants were obtained by the mL−1 in PBS) was added to each well of the plate. The cells
titration of base into variable ratio metal–ligand solutions, were incubated for 3 h with the MTT solution, allowing for the
where the ligand concentration was kept constant and the con- formation of formazan crystals at the bottom of the wells.
centration of either the La(III) or Gd(III) was varied. Metal ion The MTT/media/compound treatment was aspirated, leaving
solutions were prepared from the appropriate atomic absorp- the formazan crystals, which were subsequently dissolved in
tion (AA) standard (Fluka, a subsidiary of Sigma-Aldrich). The 100 μL of DMSO. The absorbance of each well was measured at
exact amount of acid present in the metal standard was deter- 595 nm using a Beckman Coulter DTX 800 Multimode Detec-
mined by titration of an equimolar solution of metal and tor. Cell viability was calculated relative to the negative control
Na2H2EDTA (disodium ethylenediaminetetraacetic acid), and (representative of 100% cell viability). The EC50 value was
determined from the corresponding Gran plot.67 Metal ion solu- determined by finding the concentration at which 50% of the
tions of La (0.00100–0.00029 M) and Gd (0.00100–0.00025 M) cells were viable, relative to the negative control.
were prepared by dilution of the appropriate AA standard with
Determination of the octanol–water partition coefficient (Po/w
metal ratios for the titrations were 1 : 1, 2 : 1, 3 : 1 and 4 : 1 for The shake-flask method51 was used to determine the octanol–
)
HL1 (0.001 M) in a solution of 0.16 M NaCl. The ligand-to-
both HL1–La and HL1–Gd.
water partition coefficients of the free ligands, and were calcu-
The hydrolysis constants for La(III)68 and Gd(III)69 were lated according to eqn (3). A detailed experimental description
taken from Baes and Mesmer,69 and Barnum68 and were can be found in the ESI.†
included in the calculations. The protonation constants of the
Caco-2 cell culture
ligands HL1, H4 and H9, and the metal–ligand stability con-
stants with HL1 were calculated from the titration data using Cells were cultured in complete Dulbecco’s modified Eagle’s
Hyperquad 2008.70 All values and errors represent an average medium (DMEM), supplemented with 10% fetal bovine solu-
of at least three independent titration experiments.
tion (FBS), 292 μg mL−1 L-glutamine, 0.1 mM non-essential
amino acids, 100 μg mL−1 penicillin and 100 μg mL−1 strepto-
mycin in T-75 flasks, 12-well transwell, 12-well, 24-well or
MTT assay in MG-63 cells
The medium used to culture the MG-63 cells was minimum 96-well plates depending on the type of experiment. Cells were
essential medium (MEM-α) supplemented with 10% fetal incubated at 37 °C in a humidified atmosphere of 5% CO2 and
bovine serum (FBS) and 1% penicillin–streptomycin–neomycin 95% O2 and the medium was refreshed every other day. Cells
100X (Pen–Strep). The cells were incubated at 37 °C in a were used for studies once they reached 80% confluency.
humidified atmosphere of 5% CO2 and 95% O2. Cells were cul-
Caco-2 cell MTS assay
tured in T-75 cm3 Corning tissue culture flasks. Medium was
changed every third day and every 5–6 days the cells were sub- Toxicities of the ligand and metal complexes were determined
cultured by trypsination (0.5% trypsin-EDTA) to either new using a modified MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxy-
culture flasks or flat bottom 96-well plates. Cells were either methoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) assay.52 Caco-
passaged to new culture flasks, where 500 000 cells were cul- 2 cells were seeded onto 96-well plates at a density of 40 000
tured in each flask and incubated with 12 mL of medium, or cells cm−2 and the medium was refreshed every 48 h. When
to 96-well plates, where 10 000 cells were plated per well in the cells reached 80% confluency, they were washed 3 times
100 μL of medium.
with Hanks balanced salt solution (HBSS) and the culture
Toxicity of the ligand and metal complexes was determined medium was exchanged for treatment solutions up to 1000 μM
using a modified MTT (2-(4,5-dimethylthiazol-2-yl)-2,5-diphe- for La(1)3, La(2)3, La(3)3, La(4)3, La(6)3, Na3[La(7)3], Na3[La(8)3],
nyltetrazolium bromide) assay.49 MG-63 cells were seeded into K2[La(XT)], 1% Triton X-100 (positive control) and media only
96-well plates at a density of 10 000 cells per well in 100 μL of (negative control). For each compound concentration, the
medium. After 24 h of incubation, the medium was exchanged assay was performed in triplicate. After the cells were incu-
for solutions of either the free ligands or the lanthanide com- bated for 24 h with the lanthanide complexes, MTS (Cell Titer
plexes, affording concentrations of 0.01, 0.1, 10, 50, 100, 200, 96R Aqueous One Solution Cell Proliferation Assay Kit) was
500 and 1000 μg mL−1 of compound dissolved in medium. added to each well. The plate was wrapped in aluminium foil
6008 | Dalton Trans., 2013, 42, 5999–6011
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
and incubated for 3–4 h at 37 °C in a humidified atmosphere in 250 μL of lysis buffer (RIPA) and transferred into a 2 mL
of 5% CO2 and 95% O2, allowing for the formation of soluble Eppendorf tubes. The wells were rinsed with another 250 μL of
formazan crystals. Absorbance was measured at room tempera- lysis buffer (RIPA) and collected in the same Eppendorf tubes;
ture at 492 nm with a Multiskan Ascent Multi-plate reader 60–75 μL from each sample was removed for protein analysis
from Labsystems. Cell viability was calculated relative to the by a Pierce BCA assay to determine the protein concentration.
negative control (representative of 100% cell viability). The The remaining 425–400 μL were reserved for later La3+ analysis
EC50 value was determined by finding the concentration at by ICP-MS. The concentration of Ln3+ accumulated in the
which 50% of the cells were viable, relative to the negative cells, and the apical and basolateral chambers were evaluated
control.
by ICP-MS.
The apparent permeability (Papp, cm s−1) coefficients from
the bidirectional transport of the Ln3+ complexes were calcu-
Caco-2 cell protein concentration assay
After performing the cell uptake and bifunctional transport lated using eqn (4). ΔQ/Δt (nmol s−1) was the flow rate for
assays, cells were lysed (vide infra), and protein concentrations mass transport across monolayers, A was the surface area of
of the cell lysates cells were measured by a bicinchoninic acid the insert membrane (1.13 cm2), and Co was the initial concen-
protein assay (BCA Protein Assay Kit).71 A procedure modified tration (μM) of the compound added in the apical chamber.
from that reported previously by our group was used to study The flow rate was calculated by plotting the rate (nmol) of
the protein concentrations.17 For a detailed experimental compound increase as a function of time (s) in the donor
description, please see the ESI.†
(basolateral) chamber.
ðΔQ=ΔtÞ
ðA ꢂ CoÞ
Caco-2 cell uptake studies
Papp
¼
ð4Þ
Caco-2 cells were seeded at 10 000 cells cm−2 in 12 or 24-well
plates and the medium was refreshed every other day until
80–90% confluency was reached. On the day of the experiment,
the cells were washed 3 times with PBS and the culture
medium was exchanged for treatment solutions of La(2)3,
La(3)3, La(6)3, Na3[La(7)3], Na3[La(8)3], K2[La(XT)]or medium
(negative control) at concentrations up to 1000 μM. The cells
were then incubated with the compounds for 24 h. Treatment
solutions were aspirated and the cells were washed three times
with PBS. After wrapping the plates in aluminium foil, they
were stored at −20 °C until analysis. Cells were lysed with
250 μL of lysis buffer (RIPA); the cell lysate solutions from each
well were transferred into separate 2 mL Eppendorf tubes. The
wells were rinsed with another 250 μL of lysis buffer (RIPA);
60–75 μL from each sample was removed for protein analysis
by a Pierce BCA assay to determine protein concentration. The
remaining 425–440 μL were reserved later for La3+ analysis by
ICP-MS. The concentrations are reported as % uptake of La3+
ion (μM)/mg of protein/well SD.
Lanthanum ion analysis by ICP-MS
All materials used for the acid digestion process were soaked
in a 5% Extran bath overnight, rinsed with 18.2 MΩ cm water,
and washed in a 1% Opitma nitric acid bath overnight.
Materials were then washed with 18.2 MΩ cm water and left to
dry in a dust free environment overnight. A lanthanide cali-
bration curve for ICP-MS analysis was prepared by serial
dilutions of the lanthanum standard (1000 μg mL−1) in 1%
Opitma nitric acid, and analyzed at concentrations ranging
from 0.0001–10 ppm, where 1 part per million (ppm) is equiv-
alent to 1 μg mL−1. The count number was then plotted versus
the concentration of the lanthanum ion in ppm. Concen-
tration of lanthanide in each sample was then determined
against the standard calibration curve.
Acid digestion for ICP-MS analysis
The samples were dissolved in 2 mL of concentrated Optima
nitric acid and transferred into test tubes (16 × 150 mm) with
1–2 Teflon boiling chips and placed in a block heater. The
temperature was raised to 105 °C and maintained overnight.
The solutions were cooled to room temperature and 2 mL of
30% hydrogen peroxide was added to each sample which was
then heated to 140 °C overnight. The temperature was
increased to 150–160 °C until samples were evaporated to
dryness. Samples were then dissolved in exactly 3.00 mL of
10% Optima nitric acid in water (% v/v) and stored in 15 mL
centrifuge tubes at 4 °C until analysis for metal ion content.
Caco-2 cell permeability studies
Caco-2 cells were seeded at 5000 cells cm−2 in polycarbonate
membrane 12-well transwell plates. The medium was
exchanged every other day and uptake studies were conducted
when TEER readings reached 500 Ω cm−2. Prior to the addition
of the lanthanide compounds, the culture medium was
removed, washed with PBS and replaced with fresh medium in
the basolateral (bottom) chamber and fresh medium (negative
control), or medium plus compound (treatment), with concen-
trations up to 1000 μM in the apical (top) chamber. After 24 h
incubation, the integrity of the Caco-2 cell monolayers was
monitored by measuring the TEER value; 0.3–0.4 mL of
Hydroxyapatite in vitro binding study
medium from the apical chamber and 1 mL from the baso- A procedure modified from that reported previously by our
lateral chamber were retained for analysis of lanthanide ion group was used to study the in vitro hydroxyapatite binding of
content by ICP-MS. The cells were washed three times with the metal complexes.17 For a detailed experimental descrip-
PBS, and the polycarbonate membranes were removed, lysed tion, please see the ESI.†
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 5999–6011 | 6009

View Article Online
Paper
Dalton Transactions
Analysis by xylenol orange assay
References
A procedure modified from that reported previously by our
group was used to qualitatively analyze the in vitro hydroxy-
apatite binding of the metal complexes.17 For a detailed exper-
imental description, please see the ESI.†
1 Osteoporosis. Government of Canada (Accessed: May 23,
2012, http://www.biobasics.gc.ca/english/view.asp?x=770).
2 J.-Y. Reginster and N. Burlet, Bone, 2006, 38, 4–9.
3 P. V. Hauschka, A. E. Mavrakos, M. D. Iafrati, S. E. Doleman
and M. Klagsbrun, J. Biol. Chem., 1986, 261, 2665–2674.
4 L. G. Raisz, Clin. Chem., 1999, 45, 1353–1358.
Conclusions
5 P. Sambrook, in Primer on the Rheumatic Diseases, ed.
J. H. Klippel, J. H. Stone, L. J. Crofford and P. H. White,
Springer, New York, 2008, pp. 584–591.
6 H. Skarpos, S. N. Osipov, D. V. Vorob’eva, I. L. Odinets,
E. Lork and G. V. Roschenthaler, Org. Biomol. Chem., 2007,
5, 2361–2367.
7 E. Guenin, M. Monteil, N. Bouchemal, T. Prange and
M. Lecouvey, Eur. J. Org. Chem., 2007, 3380–3391.
8 A. Hoffman, D. Stepensky, A. Ezra, J. M. Van Gelder and
G. Golomb, Int. J. Pharm., 2001, 220, 1–11.
9 J. A. Cramer, M. M. Amonkar, A. Hebborn and R. Altman,
Curr. Med. Res. Opin., 2005, 21, 1453–1460.
10 J. A. Cramer, D. T. Gold, S. L. Silverman and
E. M. Lewiecki, Osteoporosis Int., 2007, 18, 1023–1031.
11 S. Khosla, D. Burr, J. Cauley, D. W. Dempster, P. R. Ebeling,
D. Felsenberg, R. F. Gagel, V. Gilsanz, T. Guise, S. Koka,
L. K. McCauley, J. McGowan, M. D. McKee, S. Mohla,
D. G. Pendrys, L. G. Raisz, S. L. Ruggiero, D. M. Shafer,
L. Shum, S. L. Silverman, C. H. Van Poznak, N. Watts,
S.-B. Woo and E. Shane, J. Bone Miner. Res., 2007, 22,
1479–1491.
A series of 3-hydroxy-4-pyridinone ligands were successfully
synthesized. The 3-hydroxy-4-pyridinones were rationally
designed to incorporate a number of different functional
groups that possess varying lipophilicities and low toxicity pro-
files. The alcohol and carboxy derivatized 3-hydroxy-4-pyridi-
none ligands were coordinated to a variety of lanthanides
(Ln3+ = La, Eu, Gd, Lu). Three ligands (H4, H9 and HL1) were
studied for their acid dissociation constants (pKa), all posses-
sing similar pKas for the 3-hydroxy hydrogen. The stability con-
stants were determined for one ligand, HL1, with La3+ and
Gd3+, confirming the formation of stable tris-ligand complexes
with 3-hydroxy-4-pyridinones. The metal complexes proved to
possess low cytotoxicity profiles in MG-63 cells (EC50
>
100 μM), and a high affinity for HAP (>90%). The most promis-
ing 3-hydroxy-4-pyridinone complex in this study, La(4)3, was
700 times more bioavailable than La2(CO3)3 as determined by
the apparent permeabilities of the compounds in Caco-2 cells,
an assay that mimics the absorption of compounds across the
GI tract.
One phosphinate-EDTA ligand (H5XT) was synthesized, and
coordinated to lanthanide ions (Ln3+ = La, Eu, Lu). [La(XT)]2−
as its potassium salt proved to have the greatest uptake in
Caco-2 cells, and was clearly identified as the lead compound
from these studies. The cell uptake of [La(XT)]2− (15.26
0.17%) in Caco-2 cells was found to be higher than La(L1)3 (9
2%), the lead compound identified from the first generation
studies.17 Likewise the low toxicity of [La(XT)]2− and the high
binding affinity for HAP (>98% after 24 h) warrants further
investigation of this complex. Moreover, the low toxicity profile
of both [Eu(XT)]2− and [Lu(XT)]2− in MG-63 cells also justifies
further investigation into the entire series of lanthanide ions
coordinated by H5XT.
12 D. B. Burr, L. Miller, M. Grynpas, J. L. Li, A. Boyde,
T. Mashiba, T. Hirano and C. C. Johnston, Bone, 2003, 33,
960–969.
13 P. J. Meunier and G. Boivin, Bone, 1997, 21, 373–377.
14 T. Mashiba, C. H. Turner, T. Hirano, M. R. Forwood,
C. C. Johnston and D. B. Burr, Bone, 2001, 28,
524–531.
15 C. H. Turner, Osteoporosis Int., 2002, 13, 97–104.
16 D. J. Hadjidakis and I. I. Androulakis, in Women’s Health
and Disease: Gynecologic, Endocrine, and Reproductive Issues,
ed. G. Creatsas, G. Mastorakos and G. P. Chrousos, Wiley-
Blackwell, Hoboken, 2006, vol. 1092, pp. 385–396.
17 C. A. Barta, K. Sachs-Barrable, J. Jia, K. H. Thompson,
K. M. Wasan and C. Orvig, Dalton Trans., 2007, 5019–5030.
18 S. P. Fricker, Chem. Soc. Rev., 2006, 35, 524–533.
19 C. H. Evans, Trends Biochem. Sci., 1983, 8, 445–449.
20 E. R. Birnbaum, J. E. Gomez and D. W. Darnall, J. Am.
Chem. Soc., 1970, 92, 5287–5288.
Acknowledgements
We thank Dr Elena Polishchuk and Jessie Chen for help and
knowledge about cells, Vivian W.-M. Lai and Dr Robin Stoodley 21 P. W. Durbin, M. H. Williams, M. Gee, R. H. Newman and
for their expertise with ICP-MS and Dr Rizhi Wang for his J. G. Hamilton, Proc. Soc. Exp. Biol. Med., 1956, 91, 78–85.
expertise on hydroxyapatite and providing MG-63 cells. We 22 J. C. Zhang, S. J. Xu, K. Wang and S. F. Yu, Chin. Sci. Bull.,
also acknowledge support of the Natural Sciences and Engi- 2003, 48, 2170–2175.
neering Research Council of Canada (NSERC) and the Canadian 23 J. Zhang, C. Liu, Y. Li, J. Sun, P. Wang, K. Di and Y. Zhao,
Institutes of Health Research (CHIR) for their Collaborative J. Rare Earths, 2010, 28, 138–142.
Health Research Projects (CHRP) funding. C.O. acknowledges 24 T. J. Haley, in Handbook on the Physics and Chemistry of Rare
the Canada Council for the Arts for a Killam Research Fellow-
ship (2011–2013).
Earths, ed. J. K. A. Gschneider and L. Eyring, North-
Holland, Amsterdam, 1979, vol. 4, pp. 553–585.
6010 | Dalton Trans., 2013, 42, 5999–6011
This journal is © The Royal Society of Chemistry 2013

View Article Online
Dalton Transactions
Paper
25 N. D. Atherton, J. W. Totten and M. D. Gaitonde, Treatment 51 OECD, in OECD Guidelines for the Testing of Chemicals,
of Bone Disorders, U.S. Pat., 7,078,059, Jul 18, 2006. Section 1, OECD Publishing, 1995, p. 1.
26 S. J. P. Damment and I. Webster, J. Am. Soc. Nephrol., 2003, 52 T. M. Buttke, J. A. McCubrey and T. C. Owen, J. Immunol.
14, 204A–205A. Methods, 1993, 157, 233–240.
27 J. Burgess, M. Rangel and E. Rudi van, in Advances in In- 53 X.-P. Huang, M. Spino and J. Thiessen, Pharm. Res., 2006,
organic Chemistry, Academic, 2008, vol. 60, pp. 167–243. 23, 280–290.
28 M. A. Santos, M. Gil, L. Gano and S. Chaves, J. Biol. Inorg. 54 W. Rubas and M. E. M. Cromwell, Adv. Drug Delivery Rev.,
Chem., 2005, 10, 564–580. 1997, 23, 157–162.
29 C. Queiros, M. J. Amorim, A. Leite, M. Ferreira, P. Gameiro, 55 P. Artursson and J. Karlsson, Biochem. Biophys. Res.
B. de Castro, K. Biernacki, A. Magalhaes, J. Burgess and
M. Rangel, Eur. J. Inorg. Chem., 2011, 131–140.
30 J. B. Porter, M. Gyparaki, L. C. Burke, E. R. Huehns,
Commun., 1991, 175, 880–885.
56 C.-Y. Huang, C.-Y. Lee, M.-Y. Chen, H.-C. Tsai, H.-C. Hsu
and C.-H. Tang, J. Cell. Physiol., 2010, 224, 475–483.
P. Sarpong, V. Saez and R. C. Hider, Blood, 1988, 72, 57 Z. Yan, J. Sun, Y. Chang, Y. Liu, Q. Fu, Y. Xu, Y. Sun, X. Pu,
1497–1503.
Y. Zhang, Y. Jing, S. Yin, M. Zhu, Y. Wang and Z. He, Mol.
Pharm., 2011, 8, 319–329.
31 R. G. G. Russell, Ann. N. Y. Acad. Sci., 2006, 1068, 367–401.
32 W. O. Nelson, T. B. Karpishin, S. J. Rettig and C. Orvig, 58 M. Yazdanian, S. L. Glynn, J. L. Wright and A. Hawi, Pharm.
Can. J. Chem., 1988, 66, 123–131. Res., 1998, 15, 1490–1494.
33 P. S. Dobbin, R. C. Hider, A. D. Hall, P. D. Taylor, 59 A. Ezra, A. Hoffman, E. Breuer, I. S. Alferiev, J. Mönkkönen,
P. Sarpong, J. B. Porter, G. Xiao and D. van der Helm,
J. Med. Chem., 1993, 36, 2448–2458.
34 W. O. Nelson, T. B. Karpishin, S. J. Rettig and C. Orvig,
Inorg. Chem., 1988, 27, 1045–1051.
35 A. R. Katritzky and R. A. Jones, J. Chem. Soc., 1960,
2947–2953.
36 D. Cook, Can. J. Chem., 1963, 41, 2575–2584.
37 K. S. Mazdiyasni and L. M. Brown, Inorg. Chem., 1970, 9,
2783–2786.
38 Z. Zhang, S. J. Rettig and C. Orvig, Can. J. Chem., 1992, 70,
763–770.
39 L. Xu, S. J. Rettig and C. Orvig, Inorg. Chem., 2001, 40,
3734–3738.
40 P. Anderson, Expert Opin. Pharmacother., 2006, 7,
1475–1486.
41 O. G. Finlay, M. D. Mason and M. Shelley, Lancet Oncol.,
2005, 6, 392–400.
N. El Hanany-Rozen, G. Weiss, D. Stepensky, I. Gati,
H. Cohen, S. Törmälehto, G. L. Amidon and G. Golomb,
J. Med. Chem., 2000, 43, 3641–3652.
60 P. C. D’Haese, G. B. Spasovski, A. Sikole, A. Hutchison,
T. J. Freemont, S. Sulkova, C. Swanepoel, S. Pejanovic,
L. Djukanovic, A. Balducci, G. Coen, W. Sulowicz,
A. Ferreira, A. Torres, S. Curic, M. Popovic, N. Dimkovic
and M. E. De Broe, Kidney Int., 2003, 63, S73–S78.
61 G. B. Spasovski, A. Sikole, S. Gelev, J. Masin-Spasovska,
T. Freemont, I. Webster, M. Gill, C. Jones, M. E. De Broe
and P. C. D’Haese, Nephrol., Dial., Transplant., 2006, 21,
2217–2224.
62 A. Hulanicki, S. Glab and G. Ackermann, Pure Appl. Chem.,
1983, 55, 1137–1230.
63 M. Ali, M. Woods, E. Suh, Z. Kovacs, G. Tircsó, P. Zhao,
V. Kodibagkar and A. Sherry, J. Biol. Inorg. Chem., 2007, 12,
855–865.
42 D. J. Clevette, D. M. Lyster, W. O. Nelson, T. Rihela, 64 E. Gaidamauskas, H. Parker, B. A. Kashemirov,
G. A. Webb and C. Orvig, Inorg. Chem., 1990, 29, 667–672.
43 R. J. Motekaitis and A. E. Martell, Inorg. Chim. Acta, 1991,
183, 71–80.
A. A. Holder, K. Saejueng, C. E. McKenna and D. C. Crans,
J. Inorg. Biochem., 2009, 103, 1652–1657.
65 S. J. Lyle and M. M. Rahman, Talanta, 1963, 10, 1177–
1182.
44 M. A. Kline and C. Orvig, Clin. Chem., 1992, 38, 562–565.
45 R. D. Shannon, Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., 66 N. A. Epstein, J. L. Horton, C. M. Vogels, N. J. Taylor and
Theor. Gen. Cryst., 1976, 32, 751–767. S. A. Westcott, Aust. J. Chem., 2000, 53, 687–691.
46 M. A. Jakupec, P. Unfried and B. K. Keppler, Rev. Physiol., 67 G. Gran, Analyst, 1952, 77, 661–671.
Biochem., Pharmacol., 2005, 153, 101–111.
68 D. W. Barnum, Inorg. Chem., 1983, 22, 2297–2305.
47 Z. Xue, M. Lin, J. Zhu, J. Zhang, Y. Li and Z. Guo, Chem. 69 C. F. Baes and R. E. Mesmer, The Hydrolysis of Cations,
Commun., 2010, 46, 1212–1214.
Wiley, 1976.
48 D. Lajeunesse, Bone Miner., 1994, 24, 1–16.
49 T. Mosmann, J. Immunol. Methods, 1983, 65, 55–63.
70 P. Gans, A. Sabatini and A. Vacca, Talanta, 1996, 43,
1739–1753.
50 C. A. Lipinski, J. Pharmacol. Toxicol. Methods, 2000, 44, 71 J. M. Walker, in The Protein Protocols Handbook, ed.
235–249.
J. M. Walker, Humana, Totawa, 2nd edn, 2002, pp. 11–14.
This journal is © The Royal Society of Chemistry 2013
Dalton Trans., 2013, 42, 5999–6011 | 6011