Nuclear magnetic resonance, luminescence and structural studies of lanthanide complexes with octadentate macrocyclic ligands bearing benzylphosphinate groups

Article abstract of DOI:10.1039/a703065g

The solution and solid-state structures of lanthanide complexes of 1,4,7,10-tetraazacyclododecane-1,4,7,10- tetryltetramethylenetetra(benzylphosphinate), L^1a, and of its o-, m- and p-methoxybenzyl analogues (L², L³, L

Full text of DOI:10.1039/a703065g

View Article Online / Journal Homepage / Table of Contents for this issue
Nuclear magnetic resonance, luminescence and structural studies of
lanthanide complexes with octadentate macrocyclic ligands bearing
benzylphosphinate groups
Silvio Aime,a Andrei S. Batsanov,b Mauro Botta,a Rachel S. Dickins,b Stephen Faulkner,b
Clive E. Foster,b Alice Harrison,c Judith A. K. Howard,b Janet M. Moloney,b Timothy J. Norman,b
David Parker,*,b Louise Royleb,c and J. A. Gareth Williamsb
a
Dipartimento di Chimica Inorganica, Chimica Fisica e Chimica di Materiali,
Università di Torino, Via P. Giuria 7-10125 Torino, Italy
b Department of Chemistry, University of Durham, South Road, Durham, UK DH1 3LE
c MRC Radiobiology Unit, Chilton, Didcot, UK OX11 0RD
The solution and solid-state structures of lanthanide complexes of 1,4,7,10-tetraazacyclododecane-1,4,7,10-
tetryltetramethylenetetra(benzylphosphinate), L1a, and of its o-, m- and p-methoxybenzyl analogues (L2, L3, L4)
have been investigated by NMR, relaxometry, crystallographic and photophysical methods. It has been shown that
the number of proximate water molecules is dependent on the size of the bound lanthanide ion: the complex
[LaL1a]Ϫ is nine-co-ordinate and adopts a twisted square-antiprismatic structure with one bound water molecule.
The analogous [YL1a]Ϫ, [YbL1a]Ϫ, [GdL1a]Ϫ and [EuL1a]Ϫ complexes also adopt a twisted square-antiprismatic
geometry, but are eight-co-ordinate, with no metal-bound water. Relaxivity studies with the gadolinium complexes
showed that they are purely ‘outer-sphere’ contrast agents, but they associate strongly with proteins leading to a
pronounced relaxivity enhancement. Detailed biodistribution studies with [GdL1a]Ϫ revealed avid biliary uptake at
low doses and a well defined tendency for the complex to be cleared more slowly from tumour tissue, allowing
tissue differentiation.
O–
The solution complexation chemistry of the lanthanides has
been the subject of renewed interest over the last few years. As a
result of their unique 4fn electronic configurations the lan-
thanide ions exhibit ‘atom-like’ spectroscopic and magnetic
properties in their complexes, rendering them suitable as probes
for time-resolved assays,1–3 for the labelling of DNA,4,5 as
paramagnetic contrast agents for magnetic resonance imaging
(MRI)6–8 and as shift reagents in NMR spectroscopy.9,10
As part of our investigations into kinetically stable lan-
thanide complexes for MRI and as luminescent probes11–16 we
have been engaged in the development of the complexation
chemistry of the octadentate tetrabenzylphosphinate ligand
(L1a) which forms kinetically stable complexes with lanthanide
ions. The complexes of Y, Eu, Gd and Yb have been assumed to
be isostructural, with the lanthanide adopting an eight-co-
ordinate structure, with no proximate water molecules.14,16 In
order to define this phenomenon more closely we have prepared
a wider range of lanthanide complexes of the tetrabenzyl ligand
L1a. We report a variation in complex structure across the series,
associated with the lanthanide contraction from the larger lan-
thanum ion to the smaller, later lanthanides. In addition, the
variation in the 31P NMR chemical shift has been measured
across the lanthanide series and the results correlated with
measurements of related NMR-derived parameters, established
in earlier measurements of variations across the lanthanide
series.17
R
P
O
R
O–
P
L1a R = PhCH2
N
N
N
L1b R = Me
O
O
N
L2 R = 2-MeOC6H4CH2
L3 R = 3-MeOC6H4CH2
L4 R = 4-MeOC6H4CH2
–O
P
R
O P
R
O–
–
CO2
–
CO2
N
N
N
N
L5 (dota)
–O2C
–O2C
co-administering a known inhibitor of the ‘organic anion
transporter’ has been examined, in order to characterise the
selective biliary clearance of this complex.12
Results and Discussion
Synthesis
The absence of a bound water molecule at the binding site of
the tetrabenzyl complexes of Y, Eu, Gd and Yb suggested that a
study be undertaken of the analogous systems incorporating
methoxybenzyl groups in place of the P-benzyl substituents.
It was reasoned that the presence of the methoxy group might
promote the introduction of a hydrogen-bonded water mole-
cule close to the co-ordinated lanthanide ion. The introduction
of such an inner-sphere or ‘second-sphere’ water molecule
could enhance the overall relaxivity of the corresponding gado-
linium complex, particularly when bound to a macromolecule.14
The series of methoxy-substituted ligands, L2–4, was syn-
In order to probe the effect of an aryl substituent on the
structure of the lanthanide complexes, a series of methoxy-
benzylphosphinate derivatives has also been synthesized, and
the properties of selected lanthanide complexes investigated,
including the results of relaxivity measurements on gadolinium
complexes and photophysical measurements on the related
luminescent terbium complexes. Further biodistribution studies
have been undertaken with [GdL1a]Ϫ, in particular examining
the variation of the biodistribution of the complex as a func-
tion of the amount of complex given. In addition, the effect of
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636
3623

View Article Online
Et2O
distinct resonances were observed for the ring carbons, one
of which was coupled to phosphorus (3JCP = 12 Hz). Similar
behaviour, which is consistent with a locally C4 symmetric
twelve-membered ring that is not inverting rapidly on the NMR
RMgCl + ClP(OEt)2
RP(OEt)2
1a R = 2-MeOC6H4CH2
1b R = 3-MeOC6H4CH2
1c R = 4-MeOC6H4CH2
time-scale, has been seen in analogous C3
-symmetric hexaden-
tate triphosphinate complexes (e.g. of Ga3ϩ) based on triaza-
cyclononane. The ring carbon which is coupled to phosphorus
(δC 54.6) possesses a P᎐C᎐N᎐C dihedral angle of close to 180Њ
giving rise to a significant three-bond coupling, whereas the
other carbon (δC 56.7) defines a P᎐C᎐N᎐C dihedral angle of
approximately 90Њ and the coupling constant is too small to be
resolved.18
(CH2O)n, cyclen
thf, 4Å sieves
O–
OEt
R
R
P
O
P
O
R
R
P
O–
P
OH–
OEt
The proton and 31P NMR spectra of the [LaL3]Ϫ complex
were very similar in nature to those of [LaL1a]Ϫ. This pattern of
behaviour is similar to that observed with complexes of [YL1a]Ϫ,
where one dominant diastereoisomer was observed (ratio
>10:1) by NMR spectroscopy which did not undergo signifi-
cant exchange broadening in the temperature range 20 to
70 ЊC.14 The main differences were in the proton NMR spectra.
For example, with the yttrium complex, the chemical shift non-
equivalence of the diastereotopic benzylic CH2P protons was
0.6 ppm, whereas with the lanthanum complex it was less than
0.1 ppm. The non-equivalence in the yttrium complex may be
ascribed to the proximity of the pro-S hydrogen to the aniso-
tropic phosphorus–oxygen double bond: in the lanthanum
complexes the benzylic hydrogens must be oriented differently.
The crystallographic analysis of this yttrium complex had
earlier established an eight-co-ordinate structure with a twisted
square-antiprismatic geometry about the metal centre and a
twist angle of 29Њ for the N4 and O4 planes.14 The similarity of
the 31P shift of the complexes of Y and La (δ 39.4 and 37.6) of
L1 and the observation of one dominant diastereoisomer
(RRRR or SSSS at P) suggested that the two isomers observed
in solution with La were associated with a twisted square-
antiprismatic (major species) and square-antiprismatic geom-
etry about the metal ion.11,14 Interconversion between these two
isomers, observed by NMR on heating, may occur by a co-
operative ring inversion or by a sliding motion of the pendant
nitrogen substituents about the rigid 12-N4 framework. Such
fluxional processes have been identified in explaining the solu-
tion dynamics of lanthanide complexes of the related tetra-
carboxylate ligand 1,4,7,10-tetraazacyclododecane-1,4,7,10-
tetraacetate L5 (dota).19
N
N
N
N
O
O
O
P
O
P
N
N
N
N
–O
EtO
R
O
P
R
O P
R
R
O–
EtO
2a R = 2-MeOC6H4CH2
2b R = 3-MeOC6H4CH2
2c R = 4-MeOC6H4CH2
Scheme
decane
1
thf = Tetrahydrofuran; cyclen = 1,4,7,10-tetraazacyclodo-
thesized following the procedures shown in Scheme 1. The
intermediate dialkoxyphosphines 1a–1c were prepared from
diethyl chlorophosphite and the appropriate Grignard reagent.
The phosphines were used immediately after separation from
the magnesium salts, their purity having been confirmed by a
combination of 31P and H NMR spectroscopy. Co-conden-
1
sation of 1,4,7,10-tetraazacyclododecane with freshly sublimed
paraformaldehyde (CH2O)n in the presence of the appropriate
dialkoxyphosphine yielded the intermediate tetra esters 2a–2c
as a mixture of diastereoisomers. Purification was undertaken
by chromatography on neutral alumina and subsequent base
hydrolysis with aqueous potassium hydroxide was used to
furnish the corresponding phosphinic acid salts.11
Complexation reactions were carried out in aqueous solution
typically by addition of equimolar quantities of the appropriate
lanthanide acetate at pH 6 followed by heating at 70 ЊC for up
to 15 h. The lanthanide complexes of the methoxy-substituted
ligands were much less soluble than those derived from the
parent L1a. Thus while [GdL1a]Ϫ possesses an aqueous solubility
of 40 mmol dmϪ3 at room temperature, none of the correspond-
ing complexes of L2, L3 or L4 exhibited solubilities in excess
of 1 mmol dmϪ3, limiting their potential in any practical
application.
Using single-crystal X-ray diffraction, full crystal structures
were determined for the complexes of Eu and (with lower pre-
cision) La and Yb of L1a. Crystal lattice and space groups for
the gadolinium complex were also determined (Table 5,
Experimental section), proving the similarity of the crystal, and
as far as can be seen, of the molecular structures. The euro-
pium, gadolinium and ytterbium structures were very similar to
those of the yttrium complex which was reported earlier [Figs. 1
and 2(a)].14
Lanthanum complexes
The 31P NMR spectra of the lanthanum complexes of L1a and
L3 recorded at pD 6 revealed two resonances around δ 38 in
relative ratio 8:1. On warming to 80 ЊC the resonances
coalesced (Tc = 328 K, ∆G* = ca. 65 kJ molϪ1) and this
behaviour was reversible over two temperature cycles between
80 and 20 ЊC. Proton NMR spectra for [LaL1a]Ϫ recorded at 298
K revealed that the broad diastereotopic NCH2P resonances
centred at δ 2.52 and 3.61 [assigned with the aid of 31P-
decoupled 1H spectra and 1H᎐1H correlation spectroscopy
(COSY) experiments] coalesced on warming to 353 K. The
benzylic CH2P protons resonated as a broadened multiplet at
298 K, which sharpened on heating to a well defined doublet
(2JH P = 13 Hz) at 353 K. Within each of the five-membered ring
chelates engendered by co-ordination of the La to the four
nitrogens of the macrocyclic ring the ‘axial’ and ‘equatorial’
protons resonated as four sets of resonances consistent with
time-averaged C4 symmetry for this complex. The resonances
due to the ‘equatorial’ protons appeared as broadened doublets
at δ 2.12 and 2.33 while the ‘axial’ protons were observed to
higher frequency as broadened triplets centred at δ 3.5 and 3.6.
In the room-temperature 13C NMR spectrum of [LaL1]Ϫ two
The ligand L1a coordinates the lanthanide ion by four nitro-
gens and four phosphinate oxygen atoms, which occupy the
vertices of a twisted square prism. The twist around the local
four-fold axis [ca. 29Њ in all cases, Fig. 2(a)] is intermediate
between those in an ideal prism (0Њ) and an antiprism (45Њ), but
closer to the latter (such an arrangement has been termed an
‘inverted square antiprism’). 14,15 The antiprism and the
dodecahedron are the two principal polyhedra for co-
ordination number 8; the former can be converted into the
latter by folding both square bases of the antiprism by 29Њ.
There is no evidence for such a conversion in [YL1a]Ϫ and
[EuL1a]Ϫ, where the N4 and O4 bases of the twisted prism are
planar and parallel within experimental error. In [LaL1a(H2O)]Ϫ
the metal co-ordination is completed by an aqua ligand,
capping the O4 base of the twisted prism. This base remains
planar, while the N4 one is folded by 6.5Њ.
Comparison of the co-ordination polyhedra [Fig. 2(b),
Table 1] shows that the size of the N4 square remains practically
3624
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636

View Article Online
the [0 0 z] and [¹ ¹ z] as their axes. In the studied structures these
¯ ¯
² ²
channels are occupied by water (and/or methanol) molecules
of crystallisation, hydrogen bonded to each other and, some of
them, to the outward-looking non-co-ordinated phosphate
oxygen atoms of the anions (no hydrogen bonds with the co-
ordinated water were found). The cations are also located in the
channel. As far as the X-ray diffraction evidence goes, the
number, positions and even nature of the solvent and cation
vary from one complex to another (indeed, may vary for the
same complex, depending on the conditions of crystallisation).
Thus, in the complexes of Y and La, the cation is a hydrated
proton (unobserved in the X-ray experiment), in the europium
complex proton and sodium in a 1:1 ratio, and in the ytterbium
complex probably potassium. Given the width of the channels,
this is not surprising, for even a large moiety (e.g. a hydrated
alkali-metal ion) has room enough to move along the channel,
in accordance with the extreme instability of the crystals out-
side the mother-liquor (even under an inert atmosphere) and
slight but observable variations of unit-cell parameters between
crystals of the same complex. Such structure gives ample possi-
bilities for disorder (both static and dynamic) of the channel
contents, which was really observed in every structure.
Furthermore, with larger counter ions, incommensurate phases
can be expected (which may be the case for Yb). The cumulative
effect of these factors is to reduce substantially the precision
and accuracy of the structure determinations, even more so as
the lattice pseudo-symmetry makes the crystals particularly
prone to twinning.
Biodistribution studies with [GdL1]Ϫ
The lipophilic anionic complex [GdL1a]Ϫ is attractive for
detailed study as a contrast agent for magnetic resonance
imaging. It possesses a high kinetic stability with respect to
proton-catalysed dissociation 12 {kobs at pH 1 is 24 × 10Ϫ6 sϪ1, cf.
1.1 × 10Ϫ3 for [Gd(dtpa)]2Ϫ [H5dtpa = (carboxymethyl)imino-
bis(ethylenenitrilo)tetraacetic acid] and 3.2 × 10Ϫ6 sϪ1 for [Gd-
(dota)]Ϫ}, and exhibits a high degree of selectivity for clearance
via the biliary system, rather than the renal route. It possesses a
water solubility of 40 mmol dmϪ3. Previous reports have
described the biodistribution of the 153Gd-radiolabelled com-
plexes at a low injected dose (0.01 µmol dmϪ3).12 These studies
have been extended to include the effect of varying the injected
dose from this tracer one (i.e. 0.1 µmol dmϪ3) to a clinical
imaging dose (0.1 mmol dmϪ3). In addition, experiments have
been undertaken to confirm that the complex is handled by the
‘organic anion transporter’, which is known to be inhibited by
co-administration of the drug sodium bromosulfophthalein.20
Five minutes after injecting a 0.1 µmol dmϪ3 solution of
[GdL1a]Ϫ over 14% of the dose is in the liver, with 3% in the gall
bladder and 45% in the small intestine. By 2 h (Fig. 3) there was
less than 0.005% of the injected dose left in the blood, and only
0.05% of the dose was in the liver. Most of the complex has
entered the intestine, and by that time has passed into the
caecum and the large intestine. After 24 h most of the complex
had left the animal, with only very small amounts left in the
elimination pathway. When increasing doses of the complex
were administered there were significant differences even after
5 min. The dose dependence was characterised by an increase
in the relative amount of complex that was handled by the
kidneys, a slower rate of blood clearance and a corresponding
decrease in the percentage dose in the gall bladder and small
intestine (Fig. 4). The amount of the complex present in the
liver after 5 min is the sum of that being cleared by the liver and
the blood content of the liver (the liver is 30% perfused by
blood). Thus the total amount of complex found in this organ
does not vary so markedly as the total dose of complex is
increased, dropping from 14 (0.01 µmol dmϪ3) to 9% (0.1 mmol
dmϪ3).
Fig. 1 Structures of the complexes [EuL1a]Ϫ and [LaL1a]Ϫ in the crystal
constant (determined by the favourable square [3.3.3.3] con-
formation of the 12-N4 ring) and the O4 square is of the same
size in [EuL1a]Ϫ and slightly (by ca. 0.1 Å in the side) smaller in
[YL1a]Ϫ. However, the metal atom is not equidistant from the
two bases, but is substantially shifted towards the O4 one, the
Ln᎐O bond being up to 0.4 Å shorter than the Ln᎐N ones.
Additional water co-ordination in [LaL1a(H2O)]Ϫ exacerbates
this shift, while causing the O4 base to expand; thus the latter
complex has wider O᎐Ln᎐O and narrower N᎐Ln᎐N angles
than the eight-co-ordinate ones. The peculiar feature of the
Ln᎐L1a co-ordination is the cis orientation of the four benzyl
groups which form a hydrophobic mantle around the potential
site of water co-ordination, thus making the presence (or
absence) of the aqua ligand practically irrelevant to the overall
shape of the anion and to intermolecular hydrogen bonding.
The overall local symmetry of the complex anion (both with
or without the additional water co-ordination) can be approxi-
mated as C4. The orthorhombic crystal structures of the com-
plexes are characterised by a pseudo-tetragonal lattice with
a ≈ b (in the monoclinic ytterbium structure all this reasoning is
applicable to the pseudo-orthorhombic Niggli cell) and pseudo-
tetragonal packing of the anions. However, the pseudo-four-
fold molecular axes are nearly coincident with the [1 1 0] and
¯
[1 1 0] directions of the lattice, i.e. are perpendicular to the crys-
tal pseudo-tetragonal axis z. Such packing of the anions leaves
the infinite channels in the crystal (ca. 7.5 Å in diameter) having
Although there are only very small amounts of the complex
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636
3625

View Article Online
Fig. 2 (a) View down the C4 axis in the structures of the complexes [LaL1a]Ϫ, [EuL1a]Ϫ and [YL1a]Ϫ. (b) Co-ordination polyhedron about the Ln ion:
a and b are the average sides of the N4 and O4 antiprism bases, c and d the distances (Å) of the lanthanide atom from these bases
Ln
Y
Eu
La
a
b
c
d
2.97(2)
2.88(6)
1.64
2.97(1)
2.99(4)
1.68
2.97(5)
3.27(3)
1.85
0.97
0.97
0.81
remaining in the excretion pathways after 24 h, it is evident
that the larger the dose given the slower is the overall rate of
clearance from the body. Comparison has been made between
the behaviour of [GdL1a]Ϫ, the hydrophilic P-methyl analogue,
[GdL1b]Ϫ, and [Gd(dtpa)]2Ϫ, both in terms of the dose-
dependent biodistribution and with regard to longer-term
retention.12 The comparative data (Figs. 5 and 6) reveal that
5 min after administration of 0.1 mmol dmϪ3 of complex the
amount of [GdL1a]Ϫ in the intestine is still four times greater
than for the other two hydrophilic complexes. The amount of
complex in the blood and kidneys at this time is similar for all
three complexes. After 24 h (Fig. 6) there is still some renal
clearance of [GdL1a]Ϫ at the higher dose, as the percentage dose
in the kidneys was about seven times that found with the lower
dose.
a small degree of hepatobiliary excretion at the higher dose.
The amount of complex retained in the skeleton is 0.07% from
the higher dose after 24 h, and by 7 d this had reduced to 0.03%
dose, consistent with the high stability of the complex in vivo. It
has previously been established that premature decomplexation
of Gd from such complexes is characterised by deposition of
Gd in the skeleton accompanied by retention in the liver.12,21
With [Gd(dtpa)]2Ϫ the biodistribution at 5 min was also more
or less independent of the dose given. The amount of radiolabel
retained in the skeleton and liver at 24 h from the higher dose
was 0.11 and 0.15%, compared to 0.30 and 0.65% respectively
from the lower dose of this complex. This increased percentage
retention in the liver and the skeleton at the lower dose must be
related to the established susceptibility of this complex to
undergo cation- and/or acid-promoted dissociation, but is dif-
ficult to understand.21 At the lower dose the concentration of
complex is low relative to the concentration of cations in the
blood (e.g. Zn2ϩ) and a higher proportion evidently dissociates
than at the higher dose, where the relative concentration of
With the analogous gadolinium P-methyl complex, [GdL1b]Ϫ
the biodistribution at 5 min at both doses is very similar. By
24 h there is ten times more complex in the gut and three
times more in the liver for the 0.1 mmol dmϪ3 dose, i.e. there is
3626
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636

View Article Online
Table 1 Relevant bond distances (Å) and angles (Њ) in the complexes
τ
O
N
β N
N
N
O
N
N
γ
Ln
δ
O
Ln
O
α
O
O
O
φ
ε
N
ψ
OH2
N
O
[YL1a]ϩ
[LaL1a(H2O)]ϩ
[EuL1a]ϩ
[YbL1a]ϩ *
Ln᎐N(1)
Ln᎐N(2)
Ln᎐N(3)
Ln᎐N(4)
average Ln᎐N
Ln᎐O(11)
Ln᎐O(21)
Ln᎐O(31)
Ln᎐O(41)
average Ln᎐O (L1a)
Ln᎐O(w)
2.66(2)
2.64(2)
2.67(2)
2.67(2)
2.667(14)
2.26(2)
2.31(2)
2.20(2)
2.25(2)
2.26(4)
—
2.84(3)
2.69(3)
2.83(2)
2.84(2)
2.80(6)
2.47(2)
2.42(2)
2.50(2)
2.41(2)
2.45(4)
2.66(2)
2.68(1)
2.65(1)
2.70(1)
2.72(1)
2.69(3)
2.349(9)
2.321(8)
2.328(9)
2.304(8)
2.63(1)
2.62(1)
2.62(1)
2.64(1)
2.63(1)
2.22(1)
2.29(1)
2.23(1)
2.26(1)
2.25(3)
—
2.326(16)
—
Average angles
Individual e.s.d.s
N᎐Ln᎐N, α
0.6–0.7
67.7(7)
103.5(7)
69.2(7)
85.5(6)
79.3(1.5)
128.9(6)
—
0.6–0.8
64(2)
97(2)
66(2)
83(1)
84(1)
141(1)
71(2)
0.3
0.2
67.1(5)
102.9(3)
69.0(6)
84.9(1.4)
80(1)
130.8(7)
—
29.5(6)
68.0(3)
104.6(2)
70.4(5)
85.2(4)
78.4(5)
126.6(2)
—
N᎐Ln᎐N, β
N᎐Ln᎐O (L1a), γ
N᎐Ln᎐O (L1a), δ
O (L1a)᎐Ln᎐O (L1a), ε
O (L1a)᎐Ln᎐O (L1a), φ
O (L1a)᎐Ln᎐O(w), ψ
Antiprism twist, τ
29(1)
28.8(7)
30.5(5)
* Averaged over two symmetrically independent molecules.
Fig. 3 Biodistribution of [GdL1a]Ϫ in mice at a dose of 0.1 µmol kgϪ1
[Gd(dtpa)]2Ϫ to available endogenous cations is higher. The
amount of 153Gd in the skeleton after 7 d is 0.15%, which is a
significantly higher figure than that found for the other two
complexes. Unlike the situation with the macrocyclic complexes
where the percentage dose reduces over 7 d, the % dose in the
skeleton remains constant or increases slightly, showing that
[Gd(dtpa)]2Ϫ is less stable in vivo than the other two complexes.
This conclusion accords with the relative rates of acid-catalysed
dissociation for these complexes referred to above.12,21
Fig. 4 Effect of the dose given on the biodistribution of [GdL1a]Ϫ in
mice, 5 min post administration
As the dose of [GdL1a]Ϫ is increased (Fig. 4) saturation of the
transporter sites in the liver cells may be occurring. This would
have the effect of slowing down the rate of clearance of the
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636
3627

View Article Online
Table 2 Biodistribution data for C3H mice bearing a subcutaneous
RIF fibrosarcoma xenograft after administration of [GdL1a]Ϫ (0.1
mmol kgϪ1)*
% injected dose per g tissue
Tissue
1 h
2.23 (1.15)
17.3 (11.7)
13.1 (4.3)
0.55 (0.25)
0.34 (0.13)
0.79 (0.38)
2 h
7 h
Blood
Kidney
Liver
Femur
Muscle
Tumour
0.53 (0.22)
5.61 (2.23)
3.11 (0.73)
0.15 (0.05)
0.09 (0.04)
0.36 (0.07)
0.01 (0.005)
7.42 (2.78)
0.90 (0.28)
0.04 (0.01)
0.013 (0.007)
0.20 (0.03)
* Gadolinium-153 (γ, t₁ = 241 d) was used as the tracer. Mean values for
₂
at least four animals are given with standard deviations in parentheses.
Table 3 Biodistribution data for male nude mice* bearing a human
melanotic melanoma xenograft (HX 118) after administration of
[GdL1a]Ϫ (0.1 mmol kgϪ1
)
% injected dose per g tissue
Tissue
1 h
1.83
11.0
6.20
0.50
0.37
1.29
4 h
Blood
Kidney
Liver
Femur
Muscle
Tumour
0.05 (0.05)
6.98 (5.4)
1.13 (0.72)
0.05 (0.03)
0.02 (0.01)
0.52 (0.21)
* Mean values for five animals are given, with standard deviations in
parentheses. At 1 h, reported values are the means for only three ani-
mals.
Fig. 5 Comparative biodistribution data for [GdL1a]Ϫ (black) com-
Ϫ
pared to [GdL1b
]
(grey) and [Gd(dtpa)]2Ϫ (dotted) at a dose of 0.1
mmol kgϪ1. GB = Gall bladder
Fig. 7 Effect of co-administration of sodium bromosulfophthalein
(BSP) on the biodistribution of [GdL1a]Ϫ, given at a dose of 0.1 mmol
kgϪ1
all tissues of the hepatobiliary route (liver, gall bladder, intes-
tines). As a consequence, the amount of the complex in the
blood is greater and there is an increase in the proportion of
renal clearance.
This pattern of behaviour in vivo highlights the potential util-
ity of [GdL1a]Ϫ or its congeners in MRI studies. It may be
administered at low doses for cholangiography, e.g. at 0.5 to 3
µmol dmϪ3 so that liver receptor sites are not saturated. Further
work was then undertaken examining the behaviour of the
complex in tumour-bearing animals. Using a dose of 0.1 mmol
kgϪ1 (i.e. a typical clinical imaging dose), the biodistribution of
153Gd-labelled complex was examined in mice bearing either a
Rif fibrosarcoma or a xenograft of the human melanotic mela-
noma, HX 118. The former tumour is typically rather dense
and poorly vascularised, whereas the latter is much less dense,
tends to retain a good blood supply and commonly develops
metastases in the liver and lung.
Fig. 6 Comparative biodistribution data at a dose of 0.1 µmol kgϪ1 for
[GdL1a]Ϫ (black) and the related hydrophilic complexes [GdL1b
and [Gd(dtpa)]2Ϫ (dotted)
]
Ϫ (grey)
complex via the hepatobiliary route. A consequence of this is
that the % dose cleared by the renal route increases due to the
higher circulating levels of complex in the blood. The anion
bromosulfophthalein is known to block the receptors in the
liver cells which are responsible for the transport of anionic
lipophilic complexes.20 If [GdL1a]Ϫ is also handled by the
organic anion transporter, then co-administration of sodium
bromosulfophthalein should slow down the rate of passage of
the complex through the biliary route. When 100 µg gϪ1 of
bromosulfophthalein were administered to mice 2 min prior
to injection of 0.1 mmol kgϪ1 of [GdL1a]Ϫ a decrease was
observed (Fig. 7) in the amount of the radiolabelled complex in
Data are presented in Tables 2 and 3, for the amounts of
activity (expressed as the percentage of the injected dose per
gram of tissue) in the given tissue type at periods of up to 7 h
3628
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636

View Article Online
those found with the terbium complex of L1a,14 and clearly
indicate that there is no metal-bound water molecule.
Table 4 Lifetimesa (ms) and quantum yields (pH 6, 298 K) for the
complexes
The ortho- and meta-substituted complexes show very high
emissive quantum yields compared to those found for [TbL1a]Ϫ,
while that for the para-substituted complex is somewhat lower.
The fluorescence quantum yields quoted in Table 4 are normal-
ised relative to that found for the ortho-substituted complex; the
pattern observed is similar. The fluorescence quantum yields
are similar to those of the corresponding gadolinium com-
plexes: the observed emission from Gd is extremely weak, indi-
cating that energy transfer from the ligand singlet state to the
metal may be neglected, consistent with energy transfer from
the aryl triplet to the metal.11
b
Complex τ(D2O) τ(H2O)
q
φ(D2O)
φ(H2O)
φfl
[TbL1a]Ϫ
[TbL2]Ϫ
[TbL3]Ϫ
[TbL4]Ϫ
4.44
4.33
4.42
4.32
4.13
3.97
4.08
4.00
0.07
0.09
0.08
0.08
0.49
0.52
0.46
0.08
0.44
0.44
0.40
0.07
1
0.91
0.54
a
Lifetimes are quoted in milliseconds, for λex = 277 nm and λem = 545
nm. b Fluorescence quantum yields shown are relative to that of
[TbL2]Ϫ, which is arbitrarily set to 1.0.
The very high emissive quantum yields observed for [TbL2]Ϫ
and [TbL3]Ϫ reflect in part the efficient shielding of the metal
centre from vibrational quenching by the solvent and also indi-
cate a limited role for other effective non-radiative decay path-
ways. These are the highest quantum yields reported for a 1:1
complex of terbium in aqueous solution. In addition to the
shielding from energy-matched OH oscillators, the long life-
times and high quantum yields may also be due to the absence
post injection. In each case the rate of clearance of the complex
from skeletal muscular tissue was similar to the rate of clear-
ance from the blood, The rate of clearance of the complex from
the tumour tissue was slower, so that the ratio of the amount
of complex in the tumour to the amount in the surrounding
muscle increased with time. In the fibrosarcoma-bearing ani-
mals, for example, this ratio increased from 2.3 (1 h) to 4 at 2 h
and at 7 h the ratio was 15.3:1. A similar pattern was evident
in the animals bearing the melanoma xenograft. In this case the
tumour :muscle ratio increased from 3.5 (1 h) to 27.4:1 at 4 h.
This differential rate of clearance of lipophilic anionic com-
plexes from tumour tissue has been characterised previously in
studies with manganese complexes of porphyrins.22 It is con-
sistent with the observation from in vivo MRI studies at similar
doses that the site of subcutaneously implanted tumours in
anaesthetised rats may be discerned increasingly easily at time
points 5, 100 and 450 min post injection of 0.1 mmol kgϪ1 of
[GdL1]Ϫ.23
of C᎐O oscillators in these complexes. Higher harmonics of
᎐
C᎐O (ν ca. 1610 cmϪ1) may quench the emissive excited state,
᎐
CO
while P᎐O harmonics (ν ca. 1150 cmϪ1) are less likely to
᎐
PO
quench effectively.
Relaxivity measurements
The value of the measured relaxivity (the increment of the
water proton nuclear magnetic longitudinal relaxation rate per
unit concentration of the paramagnetic complex) and detailed
analysis of the associated NMRD (magnetic field dependence
of the relaxivity) profiles of gadolinium complexes allow a con-
siderable amount of structural information to be obtained. In
the case of [GdL1a]Ϫ the magnitude of the relaxivity and the
nature and form of the NMRD profile have been shown to be
consistent with the behaviour of a complex which does not
possess any contribution from a bound water molecule.14 Such
behaviour accords with the conclusions of photophysical
studies on the corresponding terbium and europium complexes,
which exhibit small q values.13 In addition, in the solid state, the
crystal structures of the complexes of Y, Eu and Gd show no
bound water (Figs. 1 and 2). In these cases, the relaxivity is
simply attributed to the modulation of the nucleus–electron
dipolar interaction by diffusion of the solvent molecules near
the paramagnetic complex.6 Quantitatively, its magnitude and
magnetic field dependence are described by Freed’s equation
(1), in terms of the parameters a [the distance of minimum
Luminescence studies
The luminescence behaviour of the europium and terbium
complexes of L2, L3 and L4 was investigated in water and D2O,
allowing comparison with the information obtained previously
with complexes of L1a. All the complexes showed the character-
istic absorbance spectra of the methoxybenzyl chromophore
(λmax = 274 nm, ε = 3800 dm3 molϪ1 cmϪ1). Using an excitation
wavelength of 277 nm luminescence emission spectra for each
of the six complexes were acquired. In each case the excitation
spectra (λem = 545 nm for the terbium complexes and 619 nm
for the europium complexes) closely resembled the correspond-
ing absorption spectra. Such behaviour is consistent with effi-
cient ligand-to-metal energy transfer, prior to metal-centred
emission.11
The europium complexes gave very low overall quantum
yields for emission, and the luminescence spectra obtained were
rather weak, making precise lifetime determinations difficult.
This behaviour may be ascribed to an efficient photoinduced
electron-transfer process (PET) from the excited singlet state of
the ligand to the metal. Such an electron-transfer process pro-
vides an efficient pathway for quenching of the aryl excited state,
reducing the quantum yield for metal-based emission. Such
behaviour is not at all unprecedented. Indeed similar observa-
tions having been made with the europium complex of a meth-
oxybenzyl derivative of ethylenedinitrilotetraacetate (edta),24
where the favourable free-energy change associated with the
PET process was estimated to be of the order of Ϫ88 kJ molϪ1.
The terbium complexes exhibit much higher quantum yields
and longer excited-state lifetimes, allowing more precise meas-
urement of the luminescence decay. The results of these
experiments, carried out at room temperature in water and
D2O, are summarised in Table 4. Comparison of the life-
times obtained in water and D2O allowed an assessment of the
hydration state of these terbium complexes, estimated using
the ‘conventional’ Horrocks analysis.25,26 The values of q (the
number of bound water molecules) using this analysis were of
the order of 0.07 in all cases. These values are very similar to
2
R1pOS = (32π/405)γH 2g2µB S(S ϩ 1)(Nα/1000)(M/aD) ×
[7J(ωS) ϩ 3J(ωI)] (1)
approach between the gadolinium() ion and the water pro-
tons], D (the relative diffusion coefficient for the solvent and the
gadolinium complex) and the electronic relaxation time τs.
The NMRD profiles of the gadolinium complexes of the
methoxy-substituted ligands L2, L3 and L4 are very similar in
form and nature to those reported previously for [GdL1a]Ϫ (Fig.
8).14 The profiles for the three complexes are nearly super-
imposable in the high magnetic field region, where the relaxivity
is controlled by a2/D, whereas they are slightly different in the
plateau region at lower fields where the contribution of τs
becomes more important. This pattern of behaviour confirms
the absence of proximate water molecules suggested by the
luminescence measurements. Thus the NMRD profiles are fully
consistent with gadolinium complexes of identical molecular
size, co-ordination polyhedron and hydration sphere. The dif-
ferent substitution sites of the methoxy groups are dis-
tinguished by the slightly different τs values for the three com-
plexes. Furthermore, the possibility of a tightly bound water
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636
3629

View Article Online
10). Since a change of the co-ordination number of the metal
ion upon interaction of the complex with albumin can be
excluded, it is possible that this relaxivity increase may be
attributed to the ability of the methoxy groups to promote the
formation of a network of hydrogen-bonded water molecules in
the second co-ordination sphere of the gadolinium() ion, near
to the surface of the protein. This effect is not detected in
aqueous solution because the mean residence lifetime of the
hydrogen-bonded water molecules approaches the solvent
translational diffusion correlation time. On the other hand, the
high structural organisation and the consequent reduced mobil-
ity of the water molecules in the hydration sphere of the protein
allows the observation of the second-sphere interactions which
lead to an increase in the overall relaxivity.
Fig. 8 Experimental NMRD profiles (298 K, pH 6.5) for (from the
bottom to the top) [GdL2]Ϫ, [GdL3]Ϫ and [GdL4]Ϫ. The solid lines
through the data have been calculated with the ‘best-fit’ parameters to
equation (1) for outer-sphere relaxation:14 a = 4.25 Å, D = 2.4 cm2 sϪ1
,
31P NMR Analysis of [LnL1a]؊ complexes
τSO = 102 ([GdL2]Ϫ), 115 ([GdL3]Ϫ), and 146 ps ([GdL4]Ϫ); τV = 16
([GdL2]Ϫ), 10 ([GdL3]Ϫ) and 13 ps ([GdL4]Ϫ) τV is the field-independent
correlation time and τSO is the electronic relaxation time at zero field
Another possible means of probing the changes in structure of
the lanthanide complexes across the series involves a com-
parison of the 31P chemical shifts of the paramagnetic com-
plexes with those of a diamagnetic analogue e.g. [YL1a]Ϫ. In
general, the interaction of a lanthanide cation with an NMR-
active nucleus in the ligand may be quantified in terms of two
components.27,28 The first involves a dipolar (or pseudo-
contact) component arising from an electrostatic interaction
between the metal ion and a Lewis base. The second is a contact
contribution arising from the covalent character of local bond-
ing interactions, where there is some transfer of unpaired
electron-spin density to the ligand molecule. The latter effect
tends to be significant only for nuclei proximate to the metal
centre.
The observed shifts are additive, equation (2) where δdia is the
Fig. 9 Temperature dependence of the longitudinal relaxation rate for
a 2.6 mmol dmϪ3 solution of [GdL3]Ϫ at 20 MHz and pH 6.5
δobs = δdia ϩ δdip ϩ δcon
(2)
shift arising from an analogous diamagnetic ion (La3ϩ, Lu3ϩ or
Y3ϩ), δdip is the pseudo-contact term and δcon is the contact
term. After correcting for the diamagnetic shift, the observed
paramagnetic shift (LISobs) is given by the sum of the pseudo-
contact and contact contributions, i.e. as in equation (3). It
LISobs = LISpc ϩ LISc
(3)
has been shown 29 that when a magnetic field, B, is applied to
the system at a given temperature T, then the thermal average
spin moment, 〈S〉av, for all the J levels is given by equation (4)
〈S〉av = (βB/3kT )gJ(gJ Ϫ 1)J(J ϩ 1)
(4)
Fig. 10 Proton relaxation enhancement, ε*, for [GdL3]Ϫ (0.15 mmol
dmϪ3) at 20 MHz and 298 K, as a function of bovine serum albumin
concentration. The term ε* is defined as the ratio of the paramagnetic
relaxation rates in the presence and absence of macromolecule. The
where gJ = 1 ϩ {[J(J ϩ 1) Ϫ L(L ϩ 1) ϩ S(S ϩ 1)]/2J(J ϩ 1)}.
For a nucleus acted on by a lanthanide the change in effec-
tive magnetic field, ∆B, due to the contact interaction with un-
paired f electrons is related by δcon ∝ 〈S〉a thus δcon ∝ (gJ Ϫ 1).
For the pseudo-contact term Bleaney et al.30 carried out a
theoretical treatment for a bound ligand with polar coordinates
(r, θ), from which equation (5) may be derived, where
solid line represents the best fit for
a 1:1 complexation with
Kd = 1.1 × 10Ϫ4 dm3 molϪ1, as defined in ref. 14
molecule may effectively be ruled out from consideration of the
temperature dependence of the relaxivities (Fig. 9). The tem-
perature dependence of the measured NMRD profiles corres-
ponds to that expected for purely ‘outer-sphere’ complexes that
is simply described (at this magnetic field strength) by the
monoexponential decrease of the diffusion coefficient D with
temperature. In each case then the presence of the aryl methoxy
group has not acted to increase the number of proximate water
molecules, and the relaxivity behaviour remains that of purely
‘outer-sphere’ complexes.6
δdip = CJ(β2/60k2T 2){(3 cos2 θ Ϫ 1)/r3}PЈ
(5)
CJ = g2J(J ϩ 1)(2J Ϫ 1)(2J ϩ 3)〈J|a|J〉, 〈J|a|J〉 is a numerical
coefficient derived from matrix elements of J, and PЈ is a
crystal-field coefficient which should remain constant for a
given series of isostructural complexes. The signs and relative
magnitudes of the values are thus dependent on CJ.
We had found previously that [GdL1a]Ϫ forms a relatively
strong complex with bovine serum albumin (Kd = 2.8 × 10Ϫ4
dm3 molϪ1) leading to a marked relaxivity enhancement.14 An
analogous experiment with [GdL3]Ϫ indicated that with this
methoxy-substituted complex there was both a stronger inter-
action with the protein (Kd = 1.1 × 10Ϫ4 dm3 molϪ1) and a sig-
nificantly higher (ca. 20%) enhancement of the relaxivity (Fig.
Substituting the terms back into equation (2) gives (6) in
LISobs = GCJ ϩ F〈S〉av
(6)
which G is the geometrical term in equation (3) and the term F
contains the hyperfine coupling constant. Equation (6) may be
rearranged to (7) and (8). For isostructural complexes a plot of
3630
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636

View Article Online
Fig. 11 Variation of the 31P NMR shift for the lanthanide complexes
Fig. 14 Analysis of the 31P NMR shifts for [LnL1a]Ϫ, plotted accord-
ing to equation (8)
of L1a (pD 6.5, 293 K), relative to the diamagnetic yttrium complex
suggesting a substantial contribution from the contact term.
This is perhaps not surprising given the proximity of the phos-
phorus atoms in the complex to the bound lanthanide ion (ca.
3.35 Å), and the relatively high degree of spatial penetration by
the phosphorus orbitals.
The plots according to equations (7) and (8) are shown in
Figs. 13 and 14 respectively. This analysis suggests that the lan-
thanides fall into two groups, with the earlier larger lanthanide
complexes (i.e. of Ce3ϩ, Pr3ϩ and Nd3ϩ) behaving similarly,
associated perhaps with a different structure to those of the
later lanthanides (Sm to Yb), which diverge less from a straight
line. It has been suggested that in the analysis according to
equation (8) such an apparent discontinuity may simply be a
consequence of the change in ionic radius across the series, due
to the variation in the polar coordinates of the ion without any
change in overall structure.32 However, there is a clear dis-
continuity here, in both Figs. 13 and 14. When these results are
taken in conjunction with the established structure of the early
and middle [LnL1a]Ϫ complexes, the analysis may be linked to
the occurrence of two closely related types of structure, with
only the larger early lanthanide ions (La to Pr/Nd) able to
accommodate a bound water molecule in their co-ordination
environment.
Fig. 12 Paramagnetic 17O NMR shifts for the lanthanide aqua ions
and H NMR shifts for the complexes [Ln(dpm)3] (Hdpm = dipivaloyl-
methane)
1
Conclusion
For L1a there is a variation in structure of the lanthanide com-
plexes across the series, associated with the decrease in ionic
radius from La to Yb. The measurements outlined above have
demonstrated that the earlier lanthanides and La tend to be
nine-co-ordinate while the middle and later lanthanides are
eight-co-ordinate and lack a bound water molecule. All of
the lanthanide complexes of L1a take up a twisted square-
antiprismatic structure. The exclusion of a water molecule from
the metal co-ordination site reduces the quenching of photo-
excited states, prolonging the excited-state lifetime and possibly
allowing the development of solubilised analogues of such
ligands for time-resolved luminescence microscopy. Although
the lack of an inner-sphere water molecule limits the overall
relaxivity of the gadolinium complexes, their selective biliary
clearance at low concentration of complex, their slower rate of
clearance from tumour tissue combined with the substantial
proton-relaxation enhancement effect observed in the presence
of proteins such as albumin render them attractive complexes
for MRI study.
Fig. 13 Analysis of the 31P NMR shifts for [LnL1a]Ϫ, plotted accord-
ing to equation (7)
LISobs/〈S〉av = G(CJ /〈S〉av) ϩ F
LISobs/CJ = G ϩ F(〈S〉av /CJ)
(7)
(8)
the observed shifts according to equation (7) or (8) should be a
straight line.31
In the case of the lanthanide complexes [LnL1a]Ϫ the vari-
ation in δP with lanthanide relative to that of the yttrium com-
plex (δ 39.4) has been measured (Fig. 11). This correlation may
be compared to that obtained with the oxygen-17 contact
Experimental
1
paramagnetic shifts of the aqua ions and the H shifts for the
Commercial solvents were dried with the appropriate drying
agent according to standard procedures. Water was purified by
the Purite system. Proton and 13C NMR spectra were recorded
on Bruker AC-250 (250.1 and 62.9 MHz respectively) or Varian
series [Ln(dpm)3]17 where only the pseudo-contact contribution
is relevant (Fig. 12). It may be seen that the pattern of
behaviour resembles that of the oxygen-17 shifts more closely,
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636
3631

View Article Online
Table 5 Crystal data for [LnL1a]Ϫ and [LnL1a(H2O)]Ϫ complexes
Y
La
Eu
Gd
Yb
T/K
295
150
150
155
150
Space group
Pbcn
Pbcn
Pbcn
Pbcn
C2/c
Symmetry
a/Å
b/Å
Orthorhombic
22.332(5)
23.052(9)
21.301(8)
Orthorhombic
22.744(1)
23.124(1)
21.198(1)
Orthorhombic
22.842(2)
22.620(2)
21.178(2)
Orthorhombic
22.949(4)
22.718(4)
21.128(4)
Monoclinic
32.135(15)
31.447(8)
21.451(7)
93.98(3)
21 624(13)
16
c/Å
β/Њ
U/Å3
Z
10 965(7)
8
11 149(1)
8
10 942(2)
8
11 015(3)
8
VXR-400 spectrometers, 31P NMR spectra on a Bruker AC-250
spectrometer operating at 101.1 MHz. Proton and carbon shifts
are given to higher frequency of SiMe4, while phosphorus shifts
are referenced to an external sample of 85% aqueous phos-
phoric acid. Proton spectra of the terbium complexes were
obtained on a Bruker AC-250 spectrometer operating with a
sweep width of 500 ppm. Mass spectra were recorded on a VG
7070E and on a VG Platform II for electrospray mass spec-
trometry, high-resolution spectra at the Swansea EPSRC centre.
Visible absorption spectra were acquired using a Uvikon
930 spectrophotometer. The HPLC analysis and separations
were carried out on a Varian 5560 instrument fitted with a
diode-array detector.
Variable-temperature proton–solvent longitudinal relaxation
times were measured at 20 MHz on a Spinmaster spectrometer
[Stelar, Mede(PV), Italy] by means of the inversion-recovery
technique (16 experiments, four scans). The reproducibility in
T1 measurements was within less than 1%. The temperature was
controlled by a Jael airflow heater equipped with a copper con-
stantan thermocouple: the actual temperature in the probehead
was measured with a Fluke 52 k/j digital thermometer with an
uncertainty of 0.5 K.
cated by the peculiar crystal packing. The structures contain
ordered [LnL1a]Ϫ anions (one per asymmetric unit, i.e. Z = 8),
but wide (ca. 7.5 Å in diameter) channels between them, paral-
lel to the crystallographic axis z, are occupied by numerous
molecules of the solvent of crystallisation, most of them dis-
ordered (for [YL1a]Ϫ, 14 independent positions of water mol-
ecules were located). The way in which the crystals were
obtained makes possible the presence, besides water, of alco-
hol of crystallisation as well. Another uncertainty is the
nature of the counter ion, which may be a hydrated proton
(as for [YL1a]Ϫ), or an alkali-metal cation (present in the
course of the syntheses). It must be emphasised that the size
of the cavity suffices to accommodate a cation of substantial
size, e.g. hexaaquametal, without affecting the anion packing
seriously.
For the [EuL1a]Ϫ complex, 48 208 data with 2θ р 52Њ were
collected, of which 9766 were unique (Rint = 0.166); 7102 data
with I > 2σ(I ) were regarded as observed. Slight (ca. 2%) decay
was observed in the course of the data collection (13 h). A
semiempirical absorption correction was applied, based on
Laue equivalents and repeated measurements (at different azi-
muthal angles) of all strong reflections (transmission factors
Tmin = 0.437, Tmax = 0.820). The Eu atom was located by the
Patterson method, other atoms of the anion by a succession of
Fourier syntheses. Within the cavity 20 symmetrically
independent atomic positions were revealed. The strongest peak
of electron density was interpreted as a Naϩ cation with a 50%
The 1/T1 NMRD profiles of water protons were measured
over a continuum of magnetic fields from 0.000 24 to 1.2 T
corresponding to 0.01 to 50 MHz proton Larmor frequency on
the Koenig-Brown field-cycling relaxometer installed at the
University of Florence, Italy. The temperature inside the probe
was controlled by circulation of perfluoroalkanes. Data at
higher frequencies were also obtained from JEOL EX-90 and
EX-400 spectrometers.
occupancy. This cation is disordered over two positions, 0.79 Å
3
–
apart, related via the crystallographic axis 2 [¹ y 4]. Its
¯
²
environment (apart from spurious contacts due to disorder)
consists of five aqua-ligands at distances of 2.31 to 2.66 Å.
Another 19 peaks were treated as oxygen atoms; seven of them
could be interpreted as water molecules in fully occupied posi-
tions (held by hydrogen bonds to the anion), two symmetrically
disordered (50%) in the Naϩ environment. Elsewhere the
electron-density map suggested the presence of disordered
methanol and/or water molecules, in some cases probably shar-
ing the same positions. Finally it was most effectively approxi-
mated by 10 oxygen atoms with occupancies of 2/3 each, linked
pairwise to other such ‘atoms’ or their own symmetrical equiv-
alents at distances of 0.9 to 1.65 Å.
Crystallography
Single-crystal X-ray diffraction experiments were carried out
at T 150 K, for the [GdL1a]Ϫ complex on a Rigaku AFC6S
four-circle diffractometer (ω-scan mode), for other complexes
on a Siemens SMART three-circle diffractometer equipped
with a CCD area detector (scanning a full hemisphere of
reciprocal space by ω in 0.3Њ frames). Graphite-mono-
¯
chromated Mo-Kα radiation (λ= 0.710 73 Å) and a Cryostream
open-flow N2 gas cryostat were used. The structures were solved
by Patterson and Fourier methods and refined by full-matrix
least squares against F 2 for all data using SHELXTL
software.33a
Colourless orthogonal crystals of the complexes of La, Eu
and Gd studied proved isomorphous (see Table 5) with the
yttrium complex reported earlier, crystallising in the ortho-
rhombic space group Pbcn (no. 60), while the C-centred mono-
clinic lattice of the ytterbium complex can be interpreted as a
distortion of the above-mentioned ones, its Niggli cell (in the
corresponding setting) being a = b = 22.539(8), c = 21.490(7) Å,
α = β = 87.15(3), γ = 88.84(3)Њ.
The crystals of all complexes were stable only in the presence
of solvent or its vapour and quickly decomposed on drying.
The structural study was much hindered by the quality of crys-
tals, which exhibited weak high-angle diffraction and were
extremely prone to twinning. The matters were further compli-
Ordered non-H atoms were refined with anisotropic dis-
placement parameters, closely positioned disordered atoms in
isotropic approximation, making a total of 695 refined vari-
ables, converging at R = 0.106, wR = 0.283 (here and below R
refers to |F | for observed data, wR to F 2 for all data), goodness
of fit S = 2.10. All H atoms in the anion were treated as
‘riding’ in calculated positions, solvent H atoms were neglected.
The only conspicuous feature in the residual difference
electron-density distribution are four peaks of 2.2–2.3 e ÅϪ3 at
ca. 1 Å from the Eu atom, and some diffuse electron density
(р0.9 e ÅϪ3) in the channel.
For the [LaL1a(H2O)]Ϫ complex it was not possible to obtain
single crystals; from a deconvoluted twin, 27 104 reflections with
2θ р 38Њ were used (4429 unique, Rint = 0.104, 3720 observed),
the high-angle data being of insufficient quality. All atoms of
the anion were found by Patterson and Fourier techniques. In
3632
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636

View Article Online
the cavity, 13 significant electron-density peaks were revealed.
These were interpreted as molecules of crystallisation water
(some of them disordered between two positions in close prox-
imity) and refined as oxygen atoms (three as fully occupied
positions and 10 with 50% occupancies). Phenyl rings were
refined as rigid groups; La, P and anion O atoms were refined
with anisotropic displacement parameters, other non-H atoms
in isotropic approximation (total of 308 refined variables). The
H atoms of the L1a ligand were treated as ‘riding’; water H
atoms were neglected. The refinement converged at R = 0.199,
desired total concentration in phosphate-buffered saline (pH
7.4). The mixture was vortexed and filtered by centrifugation
through a 0.22 µm Millipore Ultrafree CM filter. Mice were
injected through the tail vein and the procedures used to estab-
lish the tissue biodistribution at the given times post-injection
followed the methods previously described.12,35
Mice (C3H) bearing a subcutaneous implant in the mid-
dorsal region of the murine fibrosarcoma Rif were injected
intravenously with up to 200 µdmϪ3 of the solution of the com-
plex. Typically the tumour weights were in the region 95 to 300
mg (cf. body weight of 25 to 30 g). Experiments involving nude
male mice bearing the HX 118 melanotic melanoma xenograft
followed the methods and practices set out previously.35,36
wR = 0.48, S = 3.31, residual ∆ρmax = 1.38, ∆ρmin = Ϫ1.06 e ÅϪ3
.
For the gadolinium complex the lattice parameters have been
determined and a full diffraction set of 7207 unique data with
2θ р 45Њ was collected, but the reflections were exceedingly
weak [average I/σ(I ) < 4]. The systematic absences and the
heavy-atom positions (from the Patterson map) show clearly
that the crystal is isostructural with those of the complexes of
Y, La and Eu. However, because of insufficient resolution,
neither full structure determination nor refinement of atomic
coordinates from isostructural complexes gave satisfactory
results.
Syntheses
Compound L1a and its complexes were prepared as previously
reported.14
Diethoxy(2-methoxybenzyl)phosphine 1a. To magnesium
turnings (7.5 g) (previously dried at 60 ЊC under high vacuum)
under argon were added sodium-dried diethyl ether (100 cm3)
and a few grains of iodine. The solution was stirred at room
temperature until the purple colour of the iodine disappeared
and the mixture was heated to reflux. To the vigorously stirred
solution was slowly added (over 2 h) o-methoxybenzyl chloride
(4.75 g, 0.030 mol) in ether (20 cm3). The solution was stirred
for 30 min, filtered under argon and completion of Grignard
formation confirmed by addition of D2O to a small sample,
For the ytterbium complex, 87 025 data with 2θ р 55Њ were
collected (24 920 unique, 18 184 observed, Rint = 0.082 after a
SADABS absorption correction,33b Tmin = 0.41, Tmax = 0.70).
The structure was solved by Patterson and Fourier methods,
revealing the entire [YbL1a]Ϫ anion, a [K(H2O)6]ϩ anion and 16
water molecules of crystallisation (some of them disordered) in
the channel. However, the refinement (Yb and P atoms aniso-
tropic, other non-H atoms isotropic, phenyl rings as rigid
groups, H atoms of L1a riding, total of 497 variables) gave
irreducible R = 0.25, wR = 0.48, S = 5.22, the high residual
1
followed by H NMR analysis of the CH2D resonance. To the
solution of the Grignard reagent cooled to Ϫ40 ЊC was slowly
added over 1 h diethyl chlorophosphite (4.27 g, 0.27 mol, 0.9
equivalent). The solution was allowed to warm to room tem-
perature overnight, filtered under argon, and complete conver-
sion into the phosphine (δP ϩ176) confirmed by 31P NMR
analysis. It may be noted that removal of the solvent under
vacuum led to decomposition of the phosphine which necessi-
tated its use in subsequent steps as a standardised ether
solution.
features of electron density (∆ρmax = 9.2, ∆ρmin = Ϫ6.3 e ÅϪ3
)
being in the close vicinity of the Yb atom. The peculiar form
of the diffraction peaks (sharp in two dimensions and showing
‘streaks’ in the l direction) suggest exceedingly anisotropic
mosaicity, disorder or incommensurate structure associated
with the channels (see above).
CCDC reference number 186/659.
Diethoxy(3-methoxybenzyl)phosphine 1b. 3-Methoxybenzyl
chloride (6.42 g, 41 mmol) was added to a stirred suspension of
magnesium turnings (2 g, excess) in tetrahydrofuran (40 cm3),
and a crystal of iodine was added. After an incubation period
of 10 min a vigorous reaction commenced. The reaction mix-
ture was stirred for 1 h, and the solution boiled under its own
exotherm. The quality of the Grignard reagent was assessed by
quenching a sample in D2O. The NMR intensities revealed
almost quantitative conversion. (The reaction has also been
realised in dry ether.) An ether solution of the Grignard reagent
(≈24 mmol) was filtered via a steel cannula, to remove un-
changed magnesium. Diethyl chlorophosphite (3.6 cm3, 3.88 g,
24 mmol) was added at Ϫ20 ЊC. The product was not isolated in
a pure form as the compound exhibits a propensity to
decompose when distilled. δH (CDCl3) 1.23 (6 H, t, CH3), 2.95 (2
H, d, J = 3.9, CH2P), 3.81 (3 H, s, OMe), 3.89 (4 H, dq, J = 14
Hz, OCH2), 6.8 (3 H, m, aryl) and 7.2 (1 H, t, aryl). δC(CDCl3)
16.9 (CH3), 42.7 (CH2P, d, J = 21), 54.8 (OCH3), 62.9 (CH2O,
d, J = 14 Hz), 111.2, 112.4, 114.8, 121.8, 130.0 and 159.1.
δP(CDCl3) 176.8 (s). ν max(film) 2974, 1600, 1490, 1198 and 1163
cmϪ1. Desorption chemical ionisation (DCI) mass spectrum:
m/z 243 (Mϩ ϩ 1).
Luminescence measurements
Luminescence spectra were recorded using a Perkin-Elmer
LS50B spectrofluorimeter operating in time-resolved mode and
equipped with a Hamamatsu R928 photomultiplier tube.
Phosphorescence excitation spectra were obtained by mon-
itoring emission at 619 nm for europium() and 545 nm for
terbium(). Quoted lifetimes, τ, are the average values obtained
from at least five measurements on each complex, made by
monitoring the emission intensity after 20 different delay times
spanning at least two lifetimes. Slit widths of 15 nm and a gate
time of 0.1 ms were used and the phosphorescence decay curves
fitted by an equation of the form (9). In all cases, high
I = I0 exp(Ϫt/τ)
(9)
correlation coefficients were observed, and the lifetimes were
found to be independent of concentration.
Luminescence quantum yields were obtained by the method
of Haas and Stein,34 using the quantum yield of [TbL1a]Ϫ as a
standard (φ = 0.49 in D2O).13,15 The observed phosphorescence
(P) for a cycle time of 20 ms was related to the total phosphor-
escence emission (PT) by expression (10) where td is the delay
time and tg the gate time in ms.
Diethoxy(4-methoxybenzyl)phosphine 1c. This compound
was prepared as described above, and again the product was
used as a 31P NMR standardised solution without being puri-
fied by distillation. δP(CDCl3): 177.1 (s). DCI mass spectrum:
m/z 243 (Mϩ ϩ 1).
1 Ϫ exp(Ϫ20/τ)
(10)
PT /P =
exp(Ϫtd /τ) Ϫ exp[Ϫ(td ϩ tg)/τ]
Biodistribution studies
Batches of [153GdL1a]Ϫ were prepared as described in ref. 12,
and where appropriate were mixed with [157GdL1a]Ϫ to the
Tetraethyl
1,4,7,10-tetraazacyclododecane-1,4,7,10-tetryl-
tetramethylenetetra(2-methoxybenzylphosphinate) 2a. To 1,4,7,
10-tetraazacyclododecane (282 mg, 1.64 mmol) in sodium-
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636
3633

View Article Online
dried thf (80 cm3) at 110 ЊC under argon were added diethoxy-
(2-methoxybenzyl)phosphine 1a (2.38 g, 9.83 mmol) in ether
(25 cm3), at which point the solution turned slightly cloudy,
followed immediately by paraformaldehyde (451 mg, 15.0
mmol). The solution was boiled under reflux overnight with
removal of water using a Soxhlet apparatus containing 4 Å
molecular sieves. A clear yellow solution resulted which, on
removal of the solvent under reduced pressure, yielded a crude
yellow viscous oil. Purification by gradient alumina column
chromatography (0 to 3% methanol in dichloromethane)
yielded the product (762 mg, 44%) as a colourless oil.
δP(CDCl3) 50.0 (s). δH (CDCl3) 1.14 (12 H, t, CH2CH3), 2.8–2.9
(24 H, br, NCH2), 3.27 (8 H, d, J = 15 Hz, PCH2C6H4), 3.80 (12
H, s, CH3), 4.00 (8 H, m, OCH2), 6.88 (8 H, m, CH2CHCOMe)
and 7.23 (8 H, m, CHCOMe, CHCCH2Cl). δC(CDCl3) 16.2 (4
1,4,7,10-Tetraazacyclododecane-1,4,7,10-tetryltetramethyl-
enetetra(2-methoxybenzylphosphinic acid) H4L2. The ester 2a
(283 mg, 0.26 mmol) was dissolved in methanol (1.5 cm3) and
added to a solution of potassium hydroxide (590 mg) in water
(17.5 cm3). The solution was heated at 70 ЊC for 24 h and the
progress of the hydrolysis followed by 31P NMR spectroscopy.
Once reaction was judged to be complete, the pH was lowered
to 6 with aqueous hydrochloric acid solution (6 mol dmϪ3) and
following removal of the water by lyophilisation the product
was purified by HPLC using an AX100 anion-exchange column
(t = 0 min, 20% MeCN, 70% water, 10% 0.1 mol dmϪ3
NH4O2CMe, pH 5.5; t = 20 min, 20% MeCN, 0% water, 80% 0.1
mol dmϪ3 NH4O2CMe, pH 5.5) to yield a colourless solid
(tR = 12.1 min), 128 mg. δP(D2O, pD 14) 37.2. δH (D2O) 2.82,
2.90, 3.01 (32 H, NCH2, PCH2), 3.63 (12 H, s, OCH3), 6.84 [8 H,
m, (CH)2CHCOMe] and 7.11 (8 H, m, CHCOMe, CHCCH2P);
δC(D2O) 35.7 (4 C, d, 1J = 89, PCH2C6H4), 53.9 [8 C, m,
1
C, CH3), 29.0 (4 C, d, J = 81 Hz, PCH2C6H4), 53.8 (4 C, d,
1J = 90 Hz, NCH2P), 53.5 [8 C, (CH2)2N], 55.0 (4 C, OCH3),
60.2 (4 C, OCH2), 110.2, 120.3, 127.6, 131.0 (16 C, CH
aromatic), 127.8 (4 C, CCH2P) and 156.7 (4 C, COMe). DCI
mass spectrum: m/z 1076.45 (Mϩ) and 1077.56 (M ϩ Hϩ)
(C52H80N4O12P4 requires 1076.48).
1
(CH2)2N], 54.8 (8 C, d, J = 84, NCH2P), 58.3 (4 C, OCH3),
114.1 (4 C, CHCHCOMe), 123.7 (4 C, d, J = 2.3, CHCHC-
CH2P), 124.8 (4 C, d, J = 8, CCH2P), 131.0 (4 C, d, J = 3.0,
CHCOMe), 134.2 (4 C, d, J = 5.0, CHCCH2P) and 159.6 (4 C,
d, J = 5.0 Hz, COMe). Positive-ion electrospray mass spectrum:
m/z 987.55 (M ϩ Naϩ) and 965.59 (M ϩ Hϩ).
Tetraethyl
1,4,7,10-tetraazacyclododecane-1,4,7,10-tetryl-
tetramethylenetetra(3-methoxybenzylphosphinate) 2b. To
a
stirred solution of 1,4,7,10-tetraazacyclododecane (475 mg,
2.76 mmol) and paraformaldehyde (414 mg, 13.8 mmol), boil-
ing under reflux over 4 Å molecular sieves in thf and under an
argon atmosphere, was added diethoxy(4-methoxybenzyl)-
phosphine (2.78 g, 11.5 mmol). A very pale orange precipitate
was formed almost immediately, and the remaining solution
became pale orange and slightly turbid, before heating was
stopped after 6 h. Column chromatography (using the condi-
tions defined above for 2a) of the derived residue afforded the
tetraester as a pale yellow glass, 528 mg (18%). HPLC:
Spherisorb 5ODS2 (1.4 cm3 minϪ1); elution A = MeCN–
CF3CO2H; B = water–CF3CO2H from 5% A, 95% B to 95% A,
5% B over 30 min; observed λ = 278 nm; tR = 24.8 min (broad).
δH (CDCl3) 1.71 (12 H, t, CH3), 2.80 (24 H, m, CH2N), 3.20
(8 H, m, C6H4CH2), 3.75 (12 H, s, OMe), 3.8–4.2 (8 H, m,
OCH2), 6.48 (12 H, m, aryl) and 7.17 (4 H, t, aryl). δC(CDCl3)
16.5, 35.8 (CH2P, d, 1J = 81), 52.2 (NCH2P, d, 1J = 106 Hz), 53.5
(CH2N), 55.1 (OCH3), 60.8 (OCH2), 112.1, 115.5, 122.2, 129.4
(aryl C), 132.9 (quaternary C) and 159.5 (COMe). δP(CDCl3)
49.3. ν max(film): 2980, 2835, 1663, 1600, 1498, 1452, 1298, 1268,
1213 and 1151 cmϪ1. Chemical ionisation (CI): m/z 1077
(Mϩ ϩ 1), 215 (100%). FAB mass spectrum (polyethylene glycol
matrix): m/z 1077.480 (C52H81N4O12P4 requires 1077.480).
1,4,7,10-Tetraazacyclododecane-1,4,7,10-tetryltetramethyl-
enetetra(3-methoxybenzylphosphinic acid) H4L3. This com-
pound was prepared as for L2. A small amount of the meta-
substituted ligand was purified by HPLC for characterisation
purposes. HPLC: Synchropak AZ100 (5 cm3 minϪ1); elution
A = MeCN, B = water, C = NH4O2CMe, pH 5.6 from 20% A,
70% B, 10% C to 20% A, 0% B, 80% C in 20 min, tR = 13.1 min.
δH (D2O) 2.8–3.2 [32 H, m (maxima at 2.82, 2.91, 3.09),
CH2N ϩ C6H4CH2], 3.65 (12 H, s, OMe), 6.76 (12 H, br s,
aryl H) and 7.16 (4 H, t, aryl H). δC(D2O, pD 5) 41.2 (C6H4CH2P,
1
d, J = 87), 54.0 (CH2N ring), 54.1 (NCH2P, d, J = 93), 58.1
(OMe), 115.1 (4JCP = 2.8), 118.2 (3JCP = 5.4), 125.5 (3JCP = 5.3),
132.8 (5JCP = 2.6), 137.7 (2J = 8.0) and 161.8 (4JCP = 2.8 Hz).
δP(D2O, pD 3) 31.9 (br). Negative ion electrospray mass spec-
trum: m/z 1003.4 (100%, M ϩ Kϩ).
1,4,7,10-Tetraazacyclododecane-1,4,7,10-tetryltetramethyl-
enetetra(4-methoxybenzylphosphinic acid) H4L4. The ester 2c
(283 mg, 0.26 mmol) was dissolved in methanol (1.5 cm3) and
added to an aqueous solution containing potassium hydroxide
(590 mg) in water (17.5 cm3). The solution was heated at 70 ЊC
for 24 h and the progress of the hydrolysis followed by 31P
NMR spectroscopy. Once complete, the pH was adjusted to 2
and the resulting white precipitate filtered off, dried and
recrystallised from MeOH to yield the product as a white crys-
talline solid (120 mg, 48%), m.p. 146 ЊC (decomp.). δH (D2O)
2.8–3.6 (32 H, m, CH2N ϩ C6H4CH2), 3.82 (12 H, s, OMe),
6.84–6.90 (8 H, m, aryl H), 7.18–7.25 (8 H, m, aryl H). δC(D2O,
Tetraethyl
1,4,7,10-tetraazacyclododecane-1,4,7,10-tetryl-
tetramethylenetetra(4-methoxybenzylphosphinate) 2c. To
a
stirred solution of 1,4,7,10-tetraazacyclododecane (230 mg,
1.34 mmol) and paraformaldehyde (180 mg, 6.1 mmol) reflux-
ing through a Soxhlet apparatus containing 4 Å molecular
sieves in dry tetrahydrofuran (50 cm3) under an argon atmos-
phere was added a solution of diethoxy(4-methoxybenzyl)-
phosphine (1.5 g, 6 mmol) in thf (25 cm3), Upon addition a
vigorous exothermic reaction was observed and a deep yellow
colour developed, after which the reaction was heated under
reflux for 18 h. Solvent was removed under reduced pressure,
and the residue purified by column chromatography on neutral
alumina with gradient elution from 100% CH2Cl2 to 3%
methanol in CH2Cl2. The ester (Rf = 0.6, 5% MeOH–95%
CH2Cl2) was isolated as a viscous pale yellow oil, 340 mg (24%).
δH (CDCl3) 1.16–1.30 (12 H, m, CH3), 2.64–3.22 (32 H, m,
CH2N and C6H4CH2), 3.79 (12 H, s, OMe), 3.90–4.22 (8 H,
m, OCH2), 6.84–6.87 (8 H, m, aryl) and 7.21–7.26 (8 H, m,
aryl). δC(CDCl3) 17.2, 35.7 (CH2P, d, 1J = 82 Hz), 53–55
(m, ring CH2N), 55.6, 61.1, 68.3, 114.4, 123.7, 129.4, 131.3
and 158.9. δP(CDCl3) 49.9. DCI mass spectrum: m/z 1077.51
(M ϩ Hϩ).
1
pD 2) 41.2 (d, J = 87), 51–54 (m, CH2N ring), 53.8 (NCH2P,
d, 1J = 90 Hz), 60.8 (OMe), 114.5, 123.6, 132.3 and 159.0.
δP(D2O, pD 2) 22.1 and 36.0 at pD 8. Positive-ion electrospray
mass spectrum: m/z 987.6 (100%, M ϩ Naϩ).
[LaL1a]؊. The compound H4L1a (210 mg, 0.256 mmol) was
dissolved in water (25 cm3). To this was added lanthanum oxide
(41.6 mg, 0.5 equivalent) and the mixture was stirred for 8 h at
80 ЊC. The pH was modified to 6–7 (addition of aqueous
NaOH), and the reaction left to continue for 3 h. The reaction
mixture was allowed to cool and a colourless solid was filtered
off. The complex was recrystallised from warm water to give
colourless crystals which became opaque on losing water of
crystallisation in vacuo (201 mg, 77%). δH (D2O, pD ≈6) 2.01 (4
H, d, J = 13, CH2), 2.18 (8 H, d, J = 13 Hz, CH2), 3.0–3.6 (20 H,
br m, CH2), 7.14 (4 H, br m, aryl H) and 7.22 (16 H, br m,
aryl H). δC(pD 6) 41.4 (d, JCP = 92, C6H4CH2P), 54.6 (d, ring
carbon anti to C᎐P bond, JCP = 12 Hz), 56.7 (ring CH2N), 58.5
3634
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636

View Article Online
3
(d, JCP = 98, NCH2P), 129.2, 131.5 (aryl), 132.8 (d, JCP = 5,
[H3O][EuL3]: Synchropak AX100 (5 cm3 minϪ1), elution
aryl) and 136.7 (d, 2JCP = 8 Hz, quaternary aryl). δP(pD 6) 37.6
and 38.8 (ratio ≈8:1). ν max(KBr disc) 1654, 1601, 1495, 1453,
1233, 1167, 1124 and 1026 cmϪ1. Negative-ion electrospray
mass spectrum: m/z 979.24 (C40H52LaN4O8P4 requires 979.18).
M.p. >250 ЊC (decomp.) (Found: C, 41.1; H, 6.69; N, 4.48.
C40H53LaN4O8P4ؒ10H2O) requires C, 41.4; H, 6.33; N, 4.83%).
A = MeCN, B = water, C = NH4O2CMe, pH 5.6 from 20% A,
70% B, 10% C to 20% A, 0% B, 80% C in 20 min, tR = 11.9 min;
δH (pD 6) Ϫ18.0 (4 H, NCHP), Ϫ11.4 (4 H, NCHЈP), Ϫ7.4 (4 H,
HaxЈ, CHN ring), Ϫ4.8 (4 H, HeqЈ, CHN ring), Ϫ1.6 (4 H, Heq,
CHN ring), 0.4 [8 H, d (collapses to a singlet on {31P}], 5.3 (12
H, OMe), 8.5 (4 H, p-H), 10.4 (4 H, m-H), 10.6 (4 H, o-H), 14.2,
(4 H, o-H) and 34.3 (4 H, Hax, CHN ring); δP(pD 6) 87.2;
ν max(KBr) 3428, 3148, 3058, 1610, 1488, 1406, 1266, 1229, 1165,
1117 and 1030 cmϪ1; UV (H2O) λmax = 273 and 278 nm
(ε = 3.8 × 103 dm3 molϪ1 cmϪ1); negative-ion electrospray mass
spectrum: m/z 1111.3, 1113.2 and 1114.2 (most abundant ion
C44H60EuN4O12P4 requires 1113.24); m.p. >250 ЊC.
[PrL1a]؊. The tetrabenzylphosphinic acid H4L1a (134 mg,
0.159 mmol) and praseodymium nitrate hexahydrate (69 mg,
0.159 mmol) were stirred for 18 h in water (10 cm3, pH 5) at
80 ЊC. On cooling a colourless solid was filtered off (51 mg,
31%). δH (pD ≈6) Ϫ37.1 (4 H, axial ring CHN), Ϫ3.8 (4 H),
Ϫ1.1 (4 H), 7.8 (4 H), 8.9 (4 H), 9.1 (8 H), 11.1 (8 H), 11.8 (4 H),
14.1 (4 H), 16.9 (4 H) and 32.6 (4 H). δC(pD 6) 14.0 (d,
JCP = 100), 36.0, 52.3, 53.2, 76.7 (d, JCP = 87 Hz), 128.8, 130.7,
132.6 and 139.0. δP(pD 5.5) 34.2. ν max(film) 1656, 1552, 1500,
1456, 1158 and 1032 cmϪ1. Negative-ion electrospray mass
spectrum: m/z 980.21 (C40H52N4O8P4Pr requires 981.18). M.p.
>250 ЊC.
[H3O][GdL3]: HPLC, tR = 13.3 min; NMR spectra could not
be obtained due to line broadening by the paramagnetic broad-
ening induced by the gadolinium ion; ν max(KBr) 3419, 1654,
1601, 1487, 1405, 1266, 1229, 1167, 1117 and 1027 cmϪ1; UV
(H2O) λmax = 273 and 278 nm; negative-ion electrospray mass
spectrum: m/z 1115.3, 1116.3, 1117.3, 1118.39 (most abun-
dant ion C44H60GdN4O12P4 requires 1118.24), 1119.6, 1120.4
and 1121.4 (Found: C, 42.9; H, 5.96; N, 4.37. C44H61GdN4-
O12P4ؒ5H2O requires C, 43.3; H, 5.87; N, 4.63%); m.p. >250 ЊC
(decomp.).
[YbL1a]؊. The compound H4L1a (95 mg, 0.113 mmol) and
ytterbium oxide (22.2 mg, 0.056 mmol) were stirred at 80 ЊC for
18 h. The pH of the solution was raised to 6, by addition of
aqueous potassium hydroxide, and stirring continued for 4 h.
The product, isolated by lyophilisation as a white solid, was
soluble in weakly basic solution (pH > 7.5) and was recrystal-
lised from such a solution (81 mg, 69%). δH (pD ≈10.5) Ϫ68.2,
Ϫ35.4, Ϫ28.4, Ϫ9.4, Ϫ2.8, 2.7 (t), 7.4 (2 H, m), 14.9 (2 H, m),
18.6 (2 H, m), 19.7, 33.0 (2 H) and 104.0. δP(pD 10) Ϫ40.7.
Negative-ion electrospray mass spectrum: m/z 1014.7
(C40H52N4O8P4Yb requires 1014.2) (Found: C, 40.8; H, 6.09;
N, 4.52. C40H53N4O8P4Ybؒ8H2O requires C, 41.1; H, 5.94; N,
4.79%).
[LaL3]Ϫ: prepared in an analogous manner to the europium
complex, ester 2b (508 mg, 0.472 mmol) being subjected to base
hydrolysis, followed by reaction with lanthanum acetate (166
mg, 0.524 mmol); preparative HPLC (tR = 9.2 min) afforded a
white solid (130 mg, 25%); δH (D2O, pD ≈6) 2.14 (4 H, d, J = 15,
CH2), 2.30 (8 H, d, J = 12 Hz, CH2), 3.0–3.6 (20 H, br m, CH2),
3.72 (12 H, s, OMe), 6.8 (12 H, br m, aryl H) and 7.21 (4 H, t,
aryl H); δP(pD 6) 37.5 and 38.8 (ratio ≈8:1) (collapsed to a
singlet on heating); ν max(KBr disc) 3135, 3049, 1762, 1601, 1405,
1261 and 1163 cmϪ1; negative-ion electrospray mass spectrum:
m/z 1099.24 (C44H60LaN4O12P4 requires 1099.22); m.p. >250 ЊC
(decomp.).
General procedures for the complexes of Eu, Tb and Gd. To
approximately 15 mg of ligand in water (2 cm3) were added 1.1
equivalents of either europium, terbium or gadolinium acetate,
and the pH adjusted to 6 (aqueous NaOH solution). The solu-
tions immediately turned opaque, and were heated at 70 ЊC
overnight to ensure complete complexation. The europium
2-methoxybenzyl complex and the 3-methoxybenzyl complexes
were sufficiently soluble (4 mg cmϪ3) to permit preparative
anion-exchange HPLC purification, but the lack of solubility
of the other complexes necessitated removal of the water under
vacuum followed by an aqueous wash of the remaining in-
soluble solid complexes to remove any excess of metal acetate.
All complexes gave molecular ions in electrospray mass spec-
troscopy, which were verified as the metal complexes by com-
parison with the theoretical mass isotope distributions.
[H3O][TbL4]: δP(D2O) 521 (br); negative-ion electrospray
mass spectrum: m/z 1119.3, 1120.3 and 1121.3 (C44H60N4-
O12P4Tb requires 1119.24).
[H3O][EuL4]: δP(D2O) 84.6; negative-ion electrospray mass
spectrum: m/z 1111.3, 1113.2 and 1114.2.
[H3O][GdL4]: negative-ion electrospray mass spectrum: m/z
1115.3, 1116.3, 1117.3, 1118.39, 1119.6, 1120.4 and 1121.4.
Acknowledgements
We thank the MRC, the BBSRC and the EU for support
under the COST programme (D. P., S. A.) and the EPSRC Mass
Spectroscopy Service for support.
[H3O][EuL2]: δP(D2O, pD 5.5) 89.6 (s); δH (D2O, pD 6) Ϫ13.5 (4
H, NCHЈP), Ϫ7.9 (4 H, NCHP), Ϫ4.7 (4 H, HaxЈ, CHN ring),
Ϫ4.3 (4 H, HeqЈ, CHN ring), Ϫ0.6 (4 H, aryl CHP), 1.3 (4 H, aryl
CHЈP), 2.4 (4 H, Heq, CHN), 3.8 (12 H, OMe), 7.85 (4 H, p-H),
8.7 (4 H, m-H), 10.2 (8 H, m-H), 16.5 (4 H, o-H) and 29.5 (4 H,
Hax, CHHN ring) (Found: C, 39.5; H, 6.15; N, 4.30. C44H61-
EuN4O12P4ؒ5H2O requires C, 39.9; H, 5.90; N, 4.65%).
References
1 G. Mathis, Clin. Chem., 1995, 41, 1391; 1993, 39, 1953.
2 E. Soini, L. Hemmila and P. Dhalen, Am. Biol. Clin., 1990, 48, 567;
N. Sabbatini, M. Guardigli and J.-M. Lehn, Coord. Chem. Rev.,
1993, 123, 201.
3 R. A. Evangelista, A. Pollak, B. Allore, E. F. Templeton,
R. C. Moiton and E. P. Diamandis, Clin. Biochem., 1988, 21, 173;
E. F. G. Dickson, A. Pollak and E. P. Diamandis, J. Photochem.
Photobiol. B. Biol., 1995, 27, 3.
4 J. Coates, P. G. Sammes and R. M. West, J. Chem. Soc., Chem.
Commun., 1995, 1107.
5 P. R. Selvin, T. M. Rana and J. E. Hearst, J. Am. Chem. Soc, 1994,
116, 6029.
[H3O][TbL2]: δP(D2O, pD 5.5) 542 (s).
[H3O][TbL3]: purified by HPLC on Synchropak AX100 (1.4
cm3 minϪ1), elution A = MeCN, B = water, C = NH4O2CMe, pH
5.6 from 20% A, 70% B, 10% C to 20% A, 0% B, 80% C in 20
min, tR = 9.7 min; δH (D2O) Ϫ198 (vbr), Ϫ126 (br), Ϫ117 (br),
Ϫ93, Ϫ73 (br), Ϫ57, Ϫ42 (12 H), 35 (vbr), 51 (br), 140 (vbr),
199 (vbr) and 290 (vbr); δP(D2O) 441 (br); ν max(KBr) 3404,
1600, 1487, 1297, 1268, 1231, 1168 and 1119 cmϪ1; UV (H2O)
λmax = 273 and 278 nm (ε = 3.6 × 103 dm3 molϪ1 cmϪ1); τ(H2O) =
4.08, τ(D2O) = 4.42 ms, q = 0.08; φ(H2O) = 0.40, φ(D2O) = 0.46;
negative-ion electrospray mass spectrum: m/z 1119.3, 1120.3
and 1121.3 (C44H60N4O12P4Tb requires 1119.24); m.p. >250 ЊC
(decomp.).
6 R. B. Lauffer, Chem. Rev., 1987, 87, 901.
7 K. Kumar and M. F. Tweedle, Pure Appl. Chem., 1993, 65, 515.
8 D. H. Powell, M. Favre, N. Graeppi, O. M. Ni Dhubghaill,
D. Pubanz and A. E. Merbach, J. Alloys Compounds, 1995, 255,
246.
9 S. Aime, M. Botta, M. Fasano, S. G. Crich and E. Terreno,
J. Bioinorg. Chem., 1996, 1, 312.
10 R. Hazama, K. Umakoski, C. Kabuto and Y. Sasaki, Chem.
Commun., 1996, 15.
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636
3635

View Article Online
11 D. Parker and J. A. G. Williams, J. Chem. Soc., Dalton Trans., 1996,
3613; P. K. Pulukkody, T. J. Norman, D. Parker, L. Royle and
C. J. Broan, J. Chem. Soc., Perkin Trans. 2, 1993, 605.
12 A. Harrison, C. A. Walker, K. A. Pereira, D. Parker, L. Royle,
K. Pulukkody and T. J. Norman, Magn. Reson. Imaging, 1993, 11,
761.
13 S. Aime, M. Botta, D. Parker and J. A. G. Williams, J. Chem.
Soc., Dalton Trans., 1996, 17; S. Aime, M. Botta, D. Parker and
J. A. G. Williams, J. Chem. Soc., Dalton Trans., 1995, 2259.
14 S. Aime, A. S. Batsanov, M. Botta, J. A. K. Howard, D. Parker,
K. Senanayake and J. A. G. Williams, Inorg. Chem., 1994, 33, 4696.
15 D. Parker and J. A. G. Williams, J. Chem. Soc., Perkin Trans. 2,
1996, 1581; A. Beeby, D. Parker and J. A. G. Williams, J. Chem.
Soc., Perkin Trans. 2, 1996, 1565; D. Parker and J. A. G. Williams,
J. Chem. Soc., Perkin Trans. 2, 1995, 1305; M. Murru, D. Parker,
J. A. G. Williams and A. Beeby, J. Chem. Soc., Chem. Commun.,
1993, 1116.
16 A. Beeby, R. S. Dickins, S. Faulkner, D. Parker and J. A. G.
Williams, Chem. Commun., 1997, 1401.
17 W. B. Lewis, J. A. Jackson, J. F. Lemons and H. Taube, J. Chem.
Phys., 1963, 36, 694; H. Donato and R. B. Martin, J. Am. Chem.
Soc., 1972, 94, 4129.
18 E. Cole, R. C. B. Copley, J. A. K. Howard, D. Parker, G. Ferguson,
J. F. Gallagher, B. Kaitner, A. Harrison and L. Royle, J. Chem. Soc.,
Dalton Trans., 1994, 1619.
23 D. Parker, Chem. Br., 1994, 833.
24 A. Abusaleh and C. F. Meares, Photochem. Photobiol. A, 1984, 39,
763.
25 W. DeW. Horrocks, jun. and D. R. Sudnick, Acc. Chem. Res., 1981,
14, 384.
26 R. S. Dickins, D. Parker, A. S. de Sousa and J. A. G. Williams,
Chem. Commun., 1996, 697.
27 I. Bertini, P. Turano and A. J. Vila, Chem. Rev., 1993, 93, 2833.
28 C. N. Reilley, B. W. Good and J. F. Desreux, Anal. Chem., 1975, 47,
2110.
29 B. Bleaney, J. Magn. Reson., 1972, 8, 91.
30 B. Bleaney, C. M. Dobson, B. A. Levine, R. B. Martin, R. J. P.
Williams and A. V. Xavier, J. Chem. Soc., Chem. Commun., 1992,
791.
31 C. N. Reilley, B. W. Good and R. D. Allendoerfer, Anal. Chem.,
1976, 48, 1446; M. D. Kemple, B. D. Ray, K. B. Lipkowitz,
F. G. Prendergast and B. D. N. Rao, J. Am. Chem. Soc., 1988, 110,
8275.
32 J. A. Peters, J. Magn. Reson., 1986, 68, 240.
33 (a) G. M. Sheldrick, SHELXTL, Version 5/VMS, Siemens Analytical
X-ray Instruments, Madison, WI, 1995; (b) SADABS, Empirical
Absorption Corrections Program, University of Göttingen, 1997.
34 Y. Haas and G. Stein, J. Chem. Phys., 1971, 75, 3668; see also
K. Nakamura, Bull. Chem. Soc. Jpn., 1982, 55, 2697; S. R. Meech
and D. Phillips, J. Photochem., 1983, 23, 193.
19 S. Aime, M. Botta and G. Ermondi, Inorg. Chem., 1992, 32, 4296.
20 C. D. Klaassen and J. B. Watkins III, Pharmacol. Rev., 1984, 36, 1.
21 C. J. Broan, J. P. L. Cox, A. S. Craig, R. Kataky, D. Parker,
A. Harrison, A. M. Randall and G. Ferguson, J. Chem. Soc., Perkin
Trans. 2, 1991, 87; P. Wedeking, K. Kumar and M. F. Tweedle,
Magn. Reson. Imaging, 1992, 10, 641.
35 A. Harrison, C. A. Walker, K. A. Pereira, L. Royle, D. Parker,
R. C. Matthews and A. S. Craig, Nucl. Med. Commun., 1992, 13,
667.
36 T. J. Norman, D. Parker, F. C. Smith, A. Harrison, L. Royle and
C. Walker, Supramol. Chem., 1995, 4, 305.
22 P. J. Bohdiewicz, D. K. Lavallee, R. A. Fawwaz, J. H. Neuhouse,
S. F. Oluwofe and P. O. Alderson, Invest. Radiol., 1990, 25, 765 and
refs. therein.
Received 6th May 1997; Paper 7/03065G
3636
J. Chem. Soc., Dalton Trans., 1997, Pages 3623–3636