When are two waters worse than one? Doubling the hydration number of a Gd-DTPA derivative decreases relaxivity

Article abstract of DOI:10.1002/chem.200500338

The synthesis of a novel ligand, based on N-methyl- diethylenetriaminetetraacetate and containing a diphenylcyclohexyl serum albumin binding group (L1) is described and the coordination chemistry and biophysical properties of its Gd^IIIcomplex G

Full text of DOI:10.1002/chem.200500338

DOI: 10.1002/chem.200500338
When are Two Waters Worse Than One? Doubling the Hydration Number of
a Gd–DTPA Derivative Decreases Relaxivity
Peter Caravan,*[a] John C. Amedio Jr.,[a] Stephen U. Dunham,[a] Matthew T. Greenfield,[a]
Normand J. Cloutier,[a] Sarah A. McDermid,[a] Marga Spiller,[b] Stephan G. Zech,[a]
Richard J. Looby,[a] Arnold M. Raitsimring,[c] Thomas J. McMurry,[a] and
Randall B. Lauffer[a]
Dedicated to Professor AndrØ Merbach on the occasion of his 65th birthday
Abstract: The synthesis of a novel
ligand, based on N-methyl-diethylene-
triaminetetraacetate and containing a
diphenylcyclohexyl serum albumin
binding group (L1) is described and
the coordination chemistry and bio-
physical properties of its GdIII complex
Gd–L1 are reported. The GdIII com-
plex of the diethylenetriaminepenta-
acetate analogue of the ligand de-
scribed here (L2) is the MRI contrast
agent MS-325. The effect of converting
an acetate to a methyl group on metal–
ligand stability, hydration number,
water-exchange rate, relaxivity, and
binding to the protein human serum al-
bumin (HSA) is explored. The complex
Gd–L1 has two coordinated water mol-
ecules in solution, that is, [Gd(L1)-
(H2O)2]2ꢀ as shown by D-band proton
ENDOR spectroscopy and implied by
1H and 17O NMR relaxation rate meas-
rate of the two coordinated waters on
Gd–L1 (kex =4.4”105 sꢀ1) is slowed by
an order of magnitude relative to Gd–
L2. As a result of this slowwater-ex-
change rate, the observed proton relax-
ivity of Gd–L1 is much lower in a solu-
tion of HSA under physiological condi-
tions (ro1bs =22.0 mmꢀ1 sꢀ1 for 0.1 mm
Gd–L1 in 0.67 mm HSA, HEPES
buffer, pH 7.4, 358C at 20 MHz) than
that of Gd–L2 (ro1bs =41.5 mmꢀ1 sꢀ1)
measured under the same conditions.
Despite having two exchangeable
water molecules, slow water exchange
limits the potential efficacy of Gd–L1
as an MRI contrast agent.
ꢀ
urements. The Gd Hwater distance of
the coordinated waters was found to be
identical to that found for Gd–L2,
3.08 Š. Loss of the acetate group de-
stabilizes the GdIII complex by 1.7 log
units (logKML =20.34) relative to the
complex with L2. The affinity of Gd–
L1 for HSA is essentially the same as
that of Gd–L2. The water-exchange
Keywords: gadolinium
·
imaging
agents · magnetic resonance imag-
ing · relaxivity · serum albumin
Introduction
Magnetic resonance imaging (MRI) is a firmly established
clinical technique that provides noninvasive high-resolution
images of body tissues. Clinical MRI measures the NMR
signals of protons, largely those of water. Differences in
signal intensity create contrast in the image and may allow
discrimination between tissue types and disease states.
Tissue contrast is achieved in many ways—by means of dif-
ferences in water content among tissues, by weighting the
imaging sequence to display differences in proton relaxation
rates (1/T1 and 1/T2), differences in chemical shift, differen-
ces in water diffusion, the effect of flowing blood, or by
using magnetization-transfer techniques.[1] In T1-weighted
imaging, a more intense signal is observed in regions in
which the longitudinal relaxation rate is fast (for which T1 is
[a] Dr. P. Caravan, Dr. J. C. Amedio Jr., Dr. S. U. Dunham,
M. T. Greenfield, Dr. N. J. Cloutier, S. A. McDermid, Dr. S. G. Zech,
R. J. Looby, Dr. T. J. McMurry, Dr. R. B. Lauffer
EPIX Pharmaceuticals, Inc., 67 Rogers Street
Cambridge, MA 02142–1118 (USA)
Fax : (+1)617-250-6127
E-mail: pcaravan@epixpharma.com
[b] M. Spiller
Department of Radiology, NewYork Medical College
Valhalla, NY 10595 (USA)
[c] Dr. A. M. Raitsimring
Department of Chemistry, University of Arizona
Tucson, AZ 85721 (USA)
5866
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5866 – 5874

FULL PAPER
short). The longitudinal relaxation rate of water protons can
be further enhanced by the addition of paramagnetic metal
complexes. These complexes, known as MRI contrast
agents,[2] provide enhanced image contrast in regions in
which the complex localizes.
The first generation of clinically approved contrast agents
distribute to plasma and to the extracellular space.[3] These
are low-molecular-weight ternary complexes of gadolini-
um(iii), in which the GdIII ion is complexed by an octaden-
tate ligand and a single coordinated water molecule. The oc-
tadentate ligand is required for safety—to insure that the
somewhat toxic GdIII ion remains sequestered in vivo and
that the complex can be excreted intact. The coordinated
water is required for contrast. The GdIII ion relaxes the coor-
dinated water, which is in fast exchange with bulk water.
This results in a shortened T1 value for the bulk (and MRI
observable) water.
MS-325 (also referred to here as Gd–L2) is a newcon-
trast agent[4] that has recently completed clinical trials for
blood vessel imaging.[5] MS-325 was designed to reversibly
target the protein human serum albumin (HSA) in the
blood plasma.[6] Binding to HSA serves three purposes: 1) it
restricts the distribution of the contrast agent to the intra-
vascular space, 2) reversible binding insures that there is
always an unbound fraction available for renal excretion,
and 3) it enhances the relaxivity of the compound. Relaxivi-
ty (r1) refers to the ability of the complex to enhance the re-
laxation rate of the solvent; see Equation (1) in which D(1/
T1) is the change in relaxation rate of the solvent after addi-
tion of contrast agent of metal concentration [M] in units of
mm.
Dð1=T1Þ
ð1Þ
r1 ¼
½MŠ
At equal concentration, a compound with enhanced relax-
ivity will appear brighter in an image compared to a com-
pound of lower relaxivity; alternately a compound with
higher relaxivity can provide the same contrast as a lowre-
laxivity compound but at a lower dose.
Relaxivity can be factored into a term that accounts for
the relaxation effect due the coordinated inner-sphere water
(rI1S) and an outer-sphere term (rO1 S), which encompasses
contributions of relaxation to the second and outer-sphere
waters [Eq. (2)]. The inner-sphere term is given by Equa-
tion (3), which is derived from the description of two-site
exchange.[7]
r1 ¼ rI1S þ rO1 S
q=½H2OŠ
ð2Þ
ð3Þ
rI1S
¼
T1m þ tm
Here q is the number of coordinated water molecules, the
water concentration is in mM, T1m is the relaxation time of
the coordinated water(s), and tm is the lifetime of the coor-
Chem. Eur. J. 2005, 11, 5866 – 5874
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5867

P. Caravan et al.
dinated water (inverse of the water-exchange rate, kex =1/
tm).
It is apparent from Equation (3) that the relaxivity can be
increased by increasing the number of coordinated water
molecules or by decreasing the denominator. It was recog-
temperature-dependent relaxivities in buffer only and in
ꢀ
buffered solutions of HSA. In addition, the Gd H distance
is estimated from a D-band 1H ENDOR study and com-
pared to that of Gd–L2. Finally the effect of modifying the
DTPA core ligand on the formation constant of the GdIII,
ZnII, and CaII complexes is assessed.
nized early on that the relaxivity at clinical field strengths of
[8]
compounds like [Gd(DTPA)(H2O)]2ꢀ is limited by T1m
.
The reason for this is that the rotational diffusion rate of the
complex is very fast compared to the Larmor frequency of
the hydrogen atom. When MS-325 forms an adduct with
HSA, the rotational diffusion rate is slowed and the relaxa-
tion efficiency of the inner-sphere water is enhanced (T1m is
decreased). This leads to a relaxivity of MS-325 in blood
plasma that is several times the
Results and Discussion
The synthesis of the ligand H5L1 and its gadolinium(iii)
complex Gd–L1 is outlined in Scheme 1. Compound 1,[16]
a
key intermediate in the preparation of MS-325,[15] was treat-
relaxivity obtained in pure
water.[4]
In early mechanistic studies
of GdIII-based contrast agents it
was assumed that the water-ex-
change rate was very fast and
close to the diffusion limit.
However in a series of seminal
papers in the 1990s, the Mer-
bach research group in Lau-
sanne showed that the water-ex-
change rate depended on the
co-ligand and that this rate
could vary over several orders
of magnitude.[9–14] For the GdIII
compounds
used
clinically,
water-exchange rates were 2–3
orders of magnitude slower
than that on the aqua ion. Yet
since relaxation of the coordi-
nated water was not very effi-
Scheme 1. a) BH3–THF, reflux. b) diisopropylethylamine, KI, DMF, BrCH2CO2C(CH3)3. c) i) PCl3, THF; ii) 4,
THF; iii) imidazole; iv) 3; v) NaIO4, aq. HCl; vi) conc. HCl. d) GdCl3, aq. NaOH.
cient due to fast rotation, T1m
@
tm, and the relaxivities of these
compounds are essentially the
same.[2]
ed with a borane–tetrahydrofuran complex under reflux
For protein-targeted or for polymeric gadolinium com-
plexes it is expected that the water-exchange rate may be
the limiting factor for optimizing relaxivity. During the de-
velopment of the synthetic process for MS-325 it was ob-
served that if the borane reduction of a protected amide in-
termediate was carried out under reflux conditions (vide
infra), the protecting group is lost and an N-methyl impurity
forms. Carrying this impurity through the alkylation and
complexation conditions used for MS-325 yields a com-
pound denoted Gd–L1. Compound Gd–L1 is a useful
model for mechanistic studies. The similarity to MS-325 (re-
ferred to hereafter as Gd–L2) suggests that it will bind to al-
bumin at the same site. The albumin-bound relaxivity of
Gd–L1 should differ from Gd–L2, because there are two
coordinated water molecules and the water-exchange rate is
expected to be different as the co-ligand has changed.
conditions to achieve amide and carbamate reduction, yield-
ing N-methyl triamine 2. The triamine compound 2 was tet-
raalkylated by heating with tert-butyl bromoacetate in di-
methylformamide with diisopropylethylamine as an acid
scavenger and potassium iodide as a reaction catalyst. Com-
pounds 3 and 4 were coupled together through a phosphate
diester link to give ligand H5L1. This was accomplished by
reacting compound 4 and phosphorus trichloride together in
tetrahydrofuran
to
generate
4,4-diphenylcyclohexyl
dichlorophosphite in situ. The dichlorophosphite was acti-
vated by reaction with imidazole and the resultant diimida-
zolide was reacted with compound 3 to afford a tetra-tert-
butyl ester of [(diphenylcyclohexyl)phosphinooxymethyl]-N-
methyldiethylenetriamine tetraacetic acid as an intermedi-
ate. In situ oxidation of the phosphorus atom with sodium
periodate followed by deprotection of the tert-butyl ester
groups with concentrated hydrochloric acid afforded H5L1.
This sequence converted alcohols 3 and 4 to H5L1 in one-
This paper reports on the synthesis of Gd–L1, its affinity
for serum albumin, its water-exchange rate, the field- and
5868
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5866 – 5874

Imaging Agents
FULL PAPER
pot, utilizing five synthetic steps. Reaction with a stoichio-
metric amount of GdIII and adjusting the pH to neutral with
sodium hydroxide yields Gd–L1.
The regiochemistry of H5L1 was investigated by means of
heteronuclear 1H-13C correlation spectroscopy. The methyl
group and the two CH groups can be easily identified in a
multiplicity-edited HSQC spectrum, since they showoppo-
site intensity from CH2 groups. The location of the N-
methyl group is confirmed by the observation of a long-
ate groups are more acidic. The protonation constants for
H6L2 determined under the same conditions[26] are also
listed in Table 1 for comparison. The addition of the N-
methyl substituent makes this nitrogen atom of L1 more
basic than that of L2.
Once the protonation constants were determined it was
possible to determine some stability constants with different
metals. The CaII and ZnII stability constants were deter-
mined by direct titration of a 1:1 mixture of metal and
ligand. Figure 1 shows that the CaII complexes (Ca–L1 and
HCa–L1) are rather weak and are only significantly formed
at pH>5. The ZnII complex on the other hand is formed at
a much lower pH. The Gd–L1 stability constant could not
be determined by direct titration because the complex pre-
cipitated (presumably as a protonated form) at the lowpH
(pH<2.5) needed to have a mixture of free ligand and free
metal ion. The stability constant was determined by compe-
tition for GdIII with L1 and the EDTA ligand. The diphenyl-
cyclohexyl moiety on L1 enables it to be retained on a re-
verse phase HPLC column and allows the gadolinium com-
plex to be separated from the free ligand. Quantitative
HPLC-MS of Gd–L1 and HnL1 along with knowledge of
the ligand protonation constants, the pH, and the Gd–
EDTA stability constant enables calculation of the Gd–L1
formation constant.
ꢀ
range J(C H) coupling to the carbon atom at the 1-position
(see graphic above for atom labeling) in the HMBC experi-
ment. Assignment of the proton chemical shifts was ob-
tained from DQF-COSY, TOCSY, and ROESY experiments
that complemented the carbon shifts derived from HSQC
and HMBC experiments.
The ligand H5L1 was isolated from acid in its neutral
form. Titration of H5L1 shows that it has five ionizable pro-
tons (Figure 1) with protonation constants listed in Table 1.
The stability constants for M–L1 are listed in Table 2
along with the stability constants for M–L2 reported previ-
ously under the same conditions (ionic strength, medium,
Table 2. Stability constants and metal complex protonation constants for
GdIII, CaII, and ZnII binding to L1 and comparative values for the M–L2
complexes.[26] Equilibria determined at 258C, 0.1m NaClO4. Values in pa-
rentheses refer to 1 standard deviation.
LogK
L1
L2
[GdL]/[Gd][L]
[CaL]/[Ca][L]
[HCaL]/[H][CaL]
[ZnL]/[Zn][L]
[HZnL]/[H][ZnL]
[H2ZnL]/[H][HZnL]
20.34 (0.04)
10.16 (0.02)
5.70 (0.005)
18.42 (0.03)
4.01 (0.005)
1.83 (0.02)
22.06
10.45
5.66
17.82
5.60
Figure 1. Observed pH versus a (mol OHꢀ/mol ligand) at 1m:1L1 ratios,
258C, m=0.1m NaClO4, [M]=[L]~1.5 mm. Symbols are measured pH
and solid lines are fits using the equilibria in Tables 1 and 2.
2.54
temperature).[26] The stability constant for Gd–L1 is 1.7 log
units lower than that of Gd–L2. A lower stability may be
expected from removing one of the coordinating acetate
groups from L2 and replacing it with a noncoordinating
methyl group. Even with the drop in stability, Gd–L1 still
forms a significantly more stable complex than the commer-
cial contrast agent Gd–DTPA–BMA (logKML =16.85).[27]
The oxophilic calcium(ii) ion also has a slightly lower stabili-
ty constant with L1 relative to that observed with L2. On
the other hand, Zn–L1 is more stable than Zn–L2. This is
probably because of the lower coordination number of ZnII
relative to those of CaII and GdIII. The ZnII ion does not re-
quire the fifth acetate oxygen donor atom. The increased
basicity of the nitrogen donor as a result of the methyl sub-
stitution could explain the increased Zn–L1 stability con-
stant relative to Zn–L2.
Table 1. Protonation constants for L1 and comparison with L2[26] deter-
mined at 258C in 0.1m NaClO4. Numbers in parentheses refer to 1 stan-
dard deviation based on mean of 4 titrations consisting of >100 data
points.
LogK
L1
L2
[HL1]/[H][L1]
9.84 (0.01)
8.73 (0.01)
3.93 (0.03)
2.57 (0.04)
1.92 (0.09)
9.56
8.31
4.41
2.92
2.43
[H2L1]/[H][HL1]
[H3L1]/[H][H2L1]
[H4L1]/[H][H3L1]
[H5L1]/[H][H4L1]
The basicity of the ligand is in line with what might be ex-
pected for a diethylenetriamine substituted with acetate
groups. Two of the nitrogen atoms are considerably basic
(pKa >8.5), while the third nitrogen atom and the carboxyl-
Chem. Eur. J. 2005, 11, 5866 – 5874
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5869

P. Caravan et al.
The Gd–L1 complex used for further characterization was
prepared in situ by mixing stoichiometric amounts of H5L1
and GdCl3 and adjusting the pH to 7 with NaOH. The ab-
sence of excess gadolinium was confirmed by titration with
xylenol orange. The HPLC-MS trace of the solution showed
a single peak with a mass corresponding to the parent ion of
Gd–L1.
It is expected that Gd–L1 would be nine-coordinate with
two bound water molecules (q=2), that is [Gd(L1)-
(H2O)2]2ꢀ. This is analogous to Gd–L2, which was shown to
be nine-coordinate in solution[6] as are other gadolini-
um(iii)–DTPA derivatives.[28–32] It was recently shown that
gadolinium ENDOR spectroscopy is a useful method to de-
termine the distance between the GdIII ion and the coordi-
nated water oxygen atom[24] or protons.[23,25] Figure 2 shows
The affinity of Gd–L1 for HSA is very similar to that re-
ported for Gd–L2 (MS-325). Under the conditions of 0.1mm
Gd–L1 and 4.5% (w/v) HSA in pH 7.4 phosphate-buffered
saline, Gd–L1 was 89.4ꢁ0.4% bound to albumin. Under
identical conditions, Gd–L2 was 88% bound to albumin.[6]
This is not surprising given the common binding group. The
difference in charge on the complex appears to play no role.
There was no measurable effect of temperature on albumin
binding. The binding assay was repeated at 58C and the
Gd–L1 was 89.4% bound at this temperature as well.
Like Gd–L2, Gd–L1 also binds to site II on HSA; this is
binding site on subdomain IIIA at which the anti-inflamma-
tory drugs ibuprofen and naproxen bind. By using a fluores-
cent probe displacement assay described previously, it was
found that Gd–L1 displaces the probe dansylsarcosine from
HSA with an inhibition constant, Ki =100ꢁ10mm (Ki for
Gd–L2=85mm).[6] Using this Ki value, it was calculated that
under the conditions of 0.1mm Gd–L1 and 4.5% HSA
(0.67 mm) there should be 85% of Gd–L1 bound to site II.
The measured value of 89.4% suggests that Gd–L1 is pri-
marily bound to site II. A site I probe was also tested.
Dansyl-l-asparagine is known to bind to site I. Gd–L1
showed only very weak displacement of the site I probe:
Ki =3000ꢁ1000mm (Ki for Gd–L2=1500mm).[6]
The relaxivities of Gd–L1, Gd–L2, and Gd–L3 in
HEPES buffer or in 4.5% HSA solution in HEPES buffer
at 358C (pH 7.4, 20 MHz) are shown graphically in Figure 3.
Figure 2. Normalized (to unity) D-band proton 1H ENDOR spectrum
corresponding to the ꢀ1/2$+1/2 electronic transition. The spectrum for
Gd–L1 is shown by a dotted line and the Gd–L2 spectrum is superim-
posed as a solid line.
the frozen solution D-band proton ENDOR spectrum corre-
sponding to the ꢀ1/2$+1/2 electronic transition. The spec-
trum for Gd–L1 is shown by a dotted line and the Gd–L2
spectrum is superimposed as a solid line. The two spectra
are superimposable except for the shoulders at about 197
and 200 MHz, which are labeled A?. These shoulders corre-
spond to the perpendicular part of the hyperfine coupling
constant to coordinated water molecules. The central fea-
tures of the spectrum arise from protons more distant from
the Gd atom, that is; protons on the co-ligand and protons
from the solvent matrix. The shoulders occur at identical
frequencies indicating that Gd–L1 and Gd–L2 have the
same hyperfine coupling to coordinated water molecules
Figure 3. Observed relaxivities of Gd–L1, Gd–L2, and Gd–L3 in A)
HEPES buffer and B) HEPES buffer
+ 4.5% (w/v) HSA at 358C,
20 MHz, pH 7.4. The estimated second- and outer-sphere relaxivity com-
ponent is shown in black and the estimated inner-sphere component in
gray.
The relaxivity of the q=0 Gd–L3 complex serves as an esti-
mate of the second and outer-sphere contributions to relax-
ivity. The bar graphs showthis second/outer-sphere effect in
black, and the gray part represents the estimate of relaxivity
due to the inner-sphere water. In buffer alone, the relaxivity
of Gd–L1 is higher than that of Gd–L2, consistent with
Gd–L1 having two coordinated water molecules, although
the inner-sphere effect is only 50% greater for Gd–L1.
When HSA is added, the relaxivities increase for all com-
pounds. Now, however, the relaxivity of the q=1 Gd–L2 is
clearly much larger than that of Gd–L1. The relaxivities re-
ported are all “observed relaxivities”, this is the relaxivity
that was calculated based on the measured relaxation rates
and the same Gd Hwater distance.[23,25] This results in a
ꢀ
ꢀ
Gd Hwater distance of 3.08 Š. As discussed previously, the
ꢀ
width of this shoulder indicates that the Gd Hwater distance
is distributed within ꢁ0.1 Š and centered at 3.08 Š. The sig-
nificantly greater intensity of the A? shoulder for Gd–L1
relative to Gd–L2 is qualitative evidence to showthat
Gd–L1 has two bound water molecules in solution. The
ENDOR spectrum of Gd–L1 in the presence of HSA (not
shown) has the same A? shoulder that occurs at the same
frequency and at the same amplitude as for Gd–L1 in
water. This implies that the hydration number does not
change when Gd–L1 is bound to HSA.
5870
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5866 – 5874

Imaging Agents
FULL PAPER
Table 3. Parameters obtained from the simultaneous fit of T1 and T2 re-
laxation rate data for H217O in the presence of Gd–L1 and Gd–L2 (MS-
325).[6] Numbers in parentheses refer to one standard deviation.
and arises from both the albumin-bound and free fractions
of the complexes. As shown above, the albumin affinity is
the same, so this difference in relaxivity does not arise in a
difference in the fraction bound to albumin.
Gd–L1
Gd–L2
(MS-325)[6]
Variable-temperature 17O relaxation-rate studies proved
critical in understanding the proton relaxivity differences.
Figure 4 shows the reduced relaxation-rate data, 1/T1r and
t3m7 [ns]
1160 (600)
69
5.8
53.7
+65
5.0
298
k
[”106 sꢀ1
]
0.44 (0.21)
ex
DHꢀ [kJmolꢀ1
]
40.0 (9.5)
ꢀ3 (30)
DSꢀ [JKꢀ1 molꢀ1
]
37
1/T [”107 sꢀ1
]
]
2.3 (2.6)
ꢀ15 (28)
118 (10)
24.2 (4.4)
1e
DET [kJmolꢀ1
ꢀ7.7
1e
t3R7 [ps]
115
DER [kJmolꢀ1
]
31.5
Figure 4 clearly shows a large difference in water-exchange
rate.
The slowwater-exchange rate has a profound effect on re-
laxivity. For a compound tumbling rapidly at 358C (tR =
115 ps), T1m is on the order of 5.2 ms at 20 MHz. The water
residency time tm is normally on the nanosecond timescale
and the denominator in Equation (3) is dominated by T1m
.
This is certainly true for Gd–L2 in which the water ex-
change has no effect on limiting relaxivity. For Gd–L1 how-
ever, tm =1.3 ms at 358C and nowthe long water residency
time contributes 20% to the denominator in Equation (3)
resulting in a lower than expected relaxivity for Gd–L1
even in HEPES buffer. The long water residency time for
Gd–L1 provides an upper limit on the expected inner-
sphere relaxivity for this compound. If the relaxation time
of the coordinated water, T1m, was made very short, the
inner-sphere relaxivity of Gd–L1 would be at most
27.7mmꢀ1 sꢀ1. The slowwater-exchange result means that al-
though Gd–L1 has two exchangeable water molecules, its
relaxivity is lower in HSA than Gd–L2 with one rapidly ex-
changing water molecule.
The slowwater-exchange kinetics predict that the inner-
sphere relaxation effect should be almost completely shut
down if the solutions are cooled down. This is shown in
Figure 5, in which the NMRD profiles of Gd–L1, Gd–L2,
and Gd–L3 are shown in buffer and in HSA solution at
358C and at 58C. At 58C the water residency times for Gd–
L1 and Gd–L2 are 7.8 and 0.9 ms, respectively, giving upper
limits on inner-sphere relaxivity of 4.6 and 20mmꢀ1 sꢀ1, re-
spectively. Using the relaxivity of Gd–L3 as an estimate of
second/outer-sphere relaxivity, one sees that slowwater ex-
change almost completely limits the relaxivity of Gd–L1 at
lowtemperature. The relaxivity of Gd–L1 in HSA at 58C is
very similar to that of Gd–L3, demonstrating that at this
lowtemperature most of the relaxation enhancement arises
from water protons in the second sphere. Even in buffer
alone, the relaxivity of Gd–L1 is lower than that of Gd–L2,
because of slowexchange of the inner-sphere water.
Figure 4. Reduced 17O relaxation rates (1/T1r, circles and 1/T2r, triangles)
of H217O in the presence of Gd–L1 vs reciprocal temperature. Solid lines
represent the simultaneous fit to the data. The dashed line represents the
1/T2r data for Gd–L2 that was described previously.
1/T2r, for Gd–L1 as a function of reciprocal temperature.
The scatter in the 1/T1r data is greater than that in the 1/T2r
data because fast rotation (vide infra) leads to only a small
paramagnetic enhancement, whereas the paramagnetic
scalar contribution to transverse relaxation is quite large. As
a result, the 17O 1/T1 values determined in the presence of
Gd–L1 are close in magnitude to the those of the solvent
alone, and the difference between the two results in a much
larger relative error for 1/T1r than for 1/T2r.
The data were analyzed as described previously[6] and the
fitted parameters are listed in Table 3 along with the corre-
sponding values for Gd–L2. The hyperfine coupling con-
stant between the Gd and the H217O was fixed at ꢀ3.8”
106 radsꢀ1.[12] The rotational correlation time at 378C deter-
mined by 17O NMR for Gd–L1 was very similar to that ob-
tained for Gd–L2; this result is expected given the close
similarity between the molecules. The striking difference
was in the water-exchange rate. The water-exchange rate for
Gd–L1 is markedly reduced relative to that of Gd–L2,
going from 5.8”106 sꢀ1 for Gd–L2[6] at 298 K to 0.44”106 sꢀ1
for Gd–L1. In Figure 4 the dashed line represents the 1/T2r
data for Gd–L2; this line highlights the large difference in
water exchange between the two complexes. At lower tem-
peratures (right side of graph) 1/T2r approaches kex and
It is not clear why replacing the N-terminal acetate group
with a methyl group has such a dramatic effect on the
water-exchange rate. One possible explanation is that the
charge is reduced in going from Gd–L2 to Gd–L1 making
the GdIII ion more acidic and strengthening the Gd O bond.
ꢀ
Chem. Eur. J. 2005, 11, 5866 – 5874
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5871

P. Caravan et al.
Experimental Section
Materials: Human serum albumin (HSA), product number A-1653 (Frac-
tion V Powder 96–99% albumin, containing fatty acids), and the fluores-
cent probes dansyl-l-asparagine, and dansylsarcosine (piperidinium salt),
were purchased from Sigma Chemical Co. (St. Louis, Mo.). Ultrafiltration
units (UFC3LCC00, regenerated cellulose membrane of 5,000 Dalton
nominal molecular weight cut-off) were obtained from Millipore Corpo-
ration (Bedford, MA). Other reagents were supplied by Aldrich Chemi-
cal Co., and were used without further purification. Solvents (HPLC
grade) were purchased from various commercial suppliers as used as re-
ceived. Column chromatography was conducted using silica gel from EM
Merck. The compounds Gd–L2 (MS-325)[15] and Gd–L3[6] were synthe-
sized as described previously. NMR spectra of synthetic intermediates
were obtained with a Varian Unity 300 or a Bruker Avance 400 spec-
trometers.
Preparation of compound 2: A round-bottomed flask, equipped with an
addition funnel, temperature probe, and a nitrogen purge was charged
with THF (100 mL) and compound 1[16] (10.0 g, 40.4 mmol). The borane–
THF complex (1m solution, 202 mL, 5 equiv, 202 mmol) was added over
a period of 30 min while maintaining an internal temperature of 25–
308C. The reaction mixture was heated to 60–658C (reflux) and the mix-
ture was allowed to stir for 7 days. The reaction mixture was cooled to
room temperature. Then aqueous HCl (2n, 24 mL) was added, followed
by concentrated HCl (16 mL) while maintaining an internal temperature
of 25–308C. The solvent (THF) was removed under vacuum (15 mm Hg,
40–458C water bath) to obtain a mobile oil. The oil was heated to 90–
958C and allowed to stir for 18 h. The mixture was cooled to room tem-
perature and the solids (boric acid) were collected by suction filtration.
The filtrate was concentrated under vacuum (1 mm Hg, 40–458C water
bath) to obtain a pasty solid, which was combined with ethanol (50 mL)
to facilitate precipitation. The solids were collected by suction filtration
and washed with ethanol (3”100 mL), recrystallized with 25% n-butanol
in ethanol and dried to provide 5.5 g of compound 2. 1H NMR: (D2O):
d=2.7 (s, 3H), 3.2–3.4 (m, 6H), 3.5–3.6 (m, 1H), 3.7–3.9 ppm (m, 2H).
Figure 5. NMRD profiles showing observed relaxivities of Gd–L1
(squares), Gd–L2 (triangles), and Gd–L3 (circles) at pH 7.4 in A)
HEPES buffer, 58C, B) HEPES buffer, 358C, C) HEPES buffer + 4.5%
(w/v) HSA, 58C, D) HEPES buffer + 4.5% (w/v) HSA, 358C.
This is not likely. The ENDOR measurement clearly
Preparation of compound 3: A round-bottomed flask, equipped with an
addition funnel, temperature probe, and a nitrogen purge was charged
with DMF (500 mL), compound 2 (15.0 g), potassium iodide (19.4 g,
2.0 equiv) and diisopropylethylamine (152.0 mL, 15.0 equiv). The mixture
was cooled to 0–58C and tert-butyl bromoacetate (60.8 mL, 7 equiv) was
added over a period of 30 min, while maintaining an internal temperature
of 5–108C. The reaction mixture was allowed to warm to room tempera-
ture, and stirring continued for 24 h. The mixture was cooled to 0–58C
and aqueous HCl (3n, 250 mL) was added over a period of 15 min, while
maintaining an internal temperature of 5–108C. Heptane (350 mL) was
added and the mixture was stirred for 20 min. The layers were separated
and the aqueous layer (pH 3) was treated with saturated sodium carbon-
ate until pH 7 was achieved. Heptane (250 mL) was combined with the
neutralized aqueous layer and stirred for 20 min. The layers were sepa-
rated. The organic layer was dried over MgSO4 and filtered, and the sol-
vent was removed under vacuum (15 mm Hg, 40–458C water bath) to
give a crude oil (22.0 g). The oil was subjected to silica gel chromatogra-
phy (25% ethyl acetate/75% hexanes solvent system) to provide purified
compound 3 (14.0 g). 1H NMR 1H NMR (CDCl3): d=1.45 (s, 36H), 2.8
(s, 3H), 2.6–2.9 (m, 3H), 3.1 (dd, 3J=14.7, 7.3 Hz, 1H), 3.30–3.50 (m,
6H), 3.6–3.85 ppm (m, 8H).
ꢀ
showed that there was no change in the Gd Hwater distance.
ꢀ
If the Gd Owater bond was shorter for Gd–L1 than Gd–L2
one would expect the proton distance to change as well. It is
worthwhile to consider a related system. Consider removal
of an acetate group from [Gd(DOTA)(H2O)]ꢀ to give [Gd-
(DO3A)(H2O)2] (DOTA=1,4,7,10-tetraazacyclododecane-
1,4,7,10-tetraacetato,
DO3A=1,4,7,10-tetraazacyclodode-
cane-1,4,7-triacetato). Again the charge is reduced and there
are now two waters bound, but in this case the water-ex-
change rate increases for [Gd(DO3A)(H2O)2].[12,33] It may
simply be that an eight-coordinate transition state (assuming
a dissociative mechanism) is less stable in the case of Gd–
L1 than for Gd–L2. Future work should address obtaining
X-ray crystal structures of Gd–L1 and Gd–L2.
Conclusion
Preparation of H5L1: A round-bottomed flask, equipped with an addition
funnel and temperature probe was charged with phosphorous trichloride
(102 ml) and THF (2 mL). A solution consisting of compound 4 (295 mg)
in THF (3 mL) was added over a period of 35 min, while maintaining an
internal temperature of ꢀ5 to 08C, and the mixture was then stirred for a
further 30 min. A solution consisting of imidazole (400 mg) in THF
(3 mL) was added over a period of 15 min, while maintaining an internal
temperature of 0 to 58C and the mixture was stirred for 20 min. A solu-
tion consisting of compound 3 (706 mg) in a mixture of hexanes (1.5 mL)
and THF (4.0 mL) was added over a period of 15 min, while maintaining
an internal temperature of ꢀ5 to 08C. The mixture was stirred for
20 min. Water (3 mL) was added over a period of 5 min, while maintain-
Removal of one of the acetate oxygen donor atoms from
the MRI contrast agent MS-325 (Gd–L2) and replacement
with a methyl group to give Gd–L1 results in a complex
with two coordinated water molecules that still has relative-
ly high thermodynamic stability. However because of the
slow water-exchange rate of these two water molecules, the
relaxivity of this compound in buffered serum albumin solu-
tion is much lower than may have been expected, and is
considerably worse than MS-325 itself.
5872
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5866 – 5874

Imaging Agents
FULL PAPER
ing an internal temperature 0–58C, and the mixture was stirred for 5 mi-
nutes. Hexanes (9 mL), toluene (1 mL), and aqueous HCl (5n, 3 mL)
were added over 5 min, while maintaining an internal temperature of 5–
108C. Sodium periodate (175 mg) was added over a period of 3 min,
while maintaining an internal temperature of 5–108C. The reaction mix-
ture was warmed to room temperature over 15 min and stirred for an ad-
ditional 30 min. The layers were separated and the organic layer was
washed with 10% aqueous sodium thiosulfate (2”5 mLmL). To the or-
ganic layer was added tetraoctylammonium bromide (63 mg). Concen-
trated HCl (6 mL) was then added over a period of 10 min, while main-
taining an internal temperature of 20–258C. This mixture was stirred for
16 h. The layers were separated and the organic layer was discarded.
Aqueous sodium hydroxide (8m, 10 mL) was added to the aqueous layer
until a pH of 6.5 was reached. The solution was concentrated under re-
duced pressure and then loaded onto a C-18 reverse-phase silica-gel-
packed column for purification. Lyophilization provided compound L1 as
firmed by titration. There are two inflections in the ligand titration curve
and two equivalents of hydroxide were required to span these two inflec-
tions. Perchlorate stock solutions of GdIII, CaII, and ZnII were prepared
by dissolving a known amount of the oxide in a slight excess of perchloric
acid and diluting to a known volume. Because hydrolysis of these metal
ions occurs at pH >5, the excess acid concentration was determined di-
rectly by titration with standard NaOH and analysis by Granꢁs
method.[18] Sodium hydroxide solutions (0.1m) were prepared from dilu-
tion of 50% NaOH with freshly boiled distilled, deionized water that had
been saturated with argon. The base solutions were standardized against
potassium hydrogen phthalate. The amount of carbonate present in the
NaOH solutions was estimated from Gran plots[18] and was always less
than 1%. Acid solutions were standardized against standard NaOH.
The ligand solutions (1–2mm) were titrated with NaOH over a pH range
from 2–11 collecting about 110 data points per titration. The titration
data was fit to a model of a ligand with five ionizable groups by using the
program BEST.[19] The value of pKw was fixed at 13.78 for all analyses.[20]
Equimolar metal–ligand solutions were titrated (110 data points per titra-
tion) over the pH range 2–11 with NaOH for CaII and ZnII, and the sta-
bility constants determined by analysis of the titration curve with
BEST.[19] The CaII data was fit to a model containing two metal–ligand
species: Ca–L1, and HCa–L1. The ZnII data was modeled with three
metal–ligand species: Zn–L1, HZn–L1, and H2Zn–L1. Multinuclear spe-
cies were not included in the models; since the metal/ligand stoichiome-
try was 1:1 and the data was well reproduced using the species described.
For GdIII, aqueous solutions containing 1:1 or 2:1 mixtures of L1 and
GdIII formed precipitates at pH lower than 2.7. Above pH 2.7 only one
species was observed, Gd–L1. The precipition at lowpH made it impos-
sible to work under conditions in which there was a significant fraction of
unchelated GdIII. To circumvent this problem, a competition study was
carried out by using an EDTA competitor ligand and monitoring the
equilibrium by HPLC-MS with a reverse-phase column and eluting with
a pH 6.8 NH4OAc buffer. Six solutions were prepared containing 1 part
L1 to 1 part GdIII to 0.75–1.25 parts EDTA with pH ranging from 3.1 to
3.4. The pH reading stabilized within minutes; however, care was taken
to ensure that there was no slowpH drift due to slowtransmetallation ki-
netics. Under these conditions, [Gd(EDTA)] and EDTA eluted in the
void volume while L1 (3.42 min) and Gd–L1 (2.80 min) were retained
(EDTA=ethylenediaminetetraacetate). This allowed the determination
of the distribution of GdIII in the system, and the formation constant for
Gd–L1 was determined by solving the appropriate mass balance equa-
tions using the protonation constants for L1 and EDTA and the [Gd-
(EDTA)] stability constant.[20]
a
white powder (0.385 g). Elemental analysis calcd (%) for
C32H44N3O12P·H2O: C 54.00, H 6.51, N 5.90, P 4.35, H2O 2.53; found: C
53.88, H 6.75, N 5.91, P 4.56, H2O 2.59; ES+-MS: m/z: 694.3 [M++H].
1H NMR (D2O/NaOD): d=2.88 (H-1), 3.69 (H-2a), 3.83 (H-2b), 2.56 (H-
3a), 2.23 (H-3b), 2.63 (H-4a), 2.52 (H-4b), 2.19 (H-5), 3.26 (H-6a), 2.95
(H-6b), 2.50 (H-7a), 2.21 (H-7b), 4.17 (H-8), 2.56 (H-9a), 2.11 (H-9b),
1.82 (H-10a), 1.58 (H-10b), 7.25 (H-14), 7.33 (H-15), 7.14 ppm (H-16);
13C NMR (D2O/NaOD): d=60.7 (C-1), 62.9 (C-2), 54.5(C-3), 51.7 (C-4),
38.1 (C-5),58.3 (C-6), 51.7 (C-7),74.3 (C-8),32.5 (C-9/9’), 29.5 (C-10/10’),
58.4 (C-11), 58.3 (C-12/13), 128.6 (C-14), 126.1 (C-15),126.8 ppm (C-16).
Preparation of Gd–L1: The concentration of the ligand L1 was deter-
mined by photometric titration with Gd(NO3)3 as described previously.[6]
A solution of GdCl3 (0.447 mL, 150.9mm, 67.5mmol) was added to a solu-
tion of L1 (1.09 mL, 61.5 mm, 67.3 mmol) at pH 6.8, and the pH adjusted
to 6.8 using 1m NaOH. The solution was stirred for 30 min and then
lyophilized affording crude chelate. Inorganic impurities were removed
by elution through a pre-packed and equilibrated C18 column with a gra-
dient of water to 1:1 ethanol/water and conductivity detection. Ethanol
was removed by rotary evaporation and the remaining aqueous solution
was lyophilized to afford purified chelate Gd–L1 as the pentahydrate
disodium salt as a white solid (46.7 mg, 71%). Elemental analysis calcd
(%) for C32H39GdNa2N3O12P·5H2O: C 39.14, H 5.03, N 4.28, Na 4.68, P
3.15; found: C 39.40, H 5.19, N 4.33, Na 4.93, P 3.18. An aqueous solution
of Gd–L1 on a HPLC-MS with UV (254 nm) and +ESI detection with a
gradient of 50mm ammonium formate with 2% (9:1 MeCN/50mm am-
monium formate) rising to 50% (9:1 MeCN/50 mm ammonium formate)
over 5 min (0.8 mLminꢀ1, Kromasil C4, 50”4.6 mm, 3.5 mm) elutes at
3.39 min (97.3% total peak area at 254 nm, positive ion, m/z=849.2 [M+
+2H]). There was no detectable L1 (L1 elutes at 3.16 min. under the
same conditions) or unchelated gadolinium (xylenol orange test).
ENDOR spectroscopy: The pulsed EPR experiments were performed
with frozen (8 K) solutions of 1mm Gd–L1 in 1:1 (v/v) H2O/CD3OH
(methanol added for glassification). In these experiments, which included
the electron spin echo (ESE) field sweep and Mims ENDOR[21] measure-
ments, the D-band (130 GHz) spectrometer[22] of Argonne National Lab-
Determination of protonation and metal–ligand stability constants: Titra-
tion pH measurements of H5L1 in the absence and presence of GdIII
,
1
CaII, and ZnII were performed with a Fisher Accumet 25 pH meter equip-
ped with an Orion Ross combination semimicro electrode. The electrode
was calibrated before each titration by titrating a known amount of
standardized HClO4(aq) with standardized NaOH solution at an ionic
strength of 0.1m using NaClO4 as the inert electrolyte. A plot of mV
(measured) versus pH (calculated) gave a working slope and intercept so
that pH could be read as ꢀlog[H+] directly. In this report, pH refers to
the hydrogen ion concentration and not activity. A Metrohm automatic
buret (Dosimat 665) was used for the NaOH additions and the buret and
pH meter were interfaced to a PC such that each titration was automated
by using the program TITRATE.[17] The temperature of each solution,
maintained in a covered, water-jacketed vessel, was kept constant at
25.0ꢁ0.18C by a Fisher Isotemp 901 circulating bath. The ionic strength
was kept constant at 0.10m NaClO4. Nitrogen, after passage through
30% NaOH, was bubbled through the solutions to exclude carbon diox-
ide.
oratory was used. H ENDOR spectra were acquired at the maximum of
the EPR spectrum (at which all EPR transitions contribute, but the
ꢀ1/2$+1/2 transition dominates) and at 24 mT lower B0 (at which all
EPR transitions contribute except the ꢀ1/2$+1/2 transition). Subtract-
ing the latter spectrum from the former gives, after appropriate normali-
zation, the spectrum associated with solely the ꢀ1/2$+1/2 electron-spin
transition. The data were analyzed as described previously[23–25] to extract
ꢀ
Gd Hwater distance estimates.
Ultrafiltration measurements of binding: Solutions containing 0.1 mm
GdIII chelate and human serum albumin (4.5% w/v) were prepared by
mixing appropriate volumes of Gd–L1 or Gd–L2 stock solution, 6%
HSA and PBS. Two aliquots (400 mL) of each these samples were placed
in 5 kDa ultrafiltration units. Two additional 25 mL aliquots were ana-
lyzed by ICP-MS to determine the total Gd concentration. The samples
were incubated at 378C for 10 min, and then centrifuged at 5800 g for
3.5 min. The filtrates (~30 mL) from these ultrafiltration units were used
to determine the free concentration of complex in each of the samples by
ICP-MS.
Distilled deionized water (Nanopure, Barnstead) was used for all solu-
tions. Solutions of the ligand were prepared by dissolving a weighed
quantity into a known volume of 0.1m NaClO4. The concentration was
calculated based on the molecular weight of the complex and was con-
Relaxivity: Relaxivities were determined at 20 MHz (0.47 T) by using a
Bruker Minispec NMS 120 to determine T1. T1 was measured with an in-
Chem. Eur. J. 2005, 11, 5866 – 5874
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5873

P. Caravan et al.
version recovery pulse sequence and all samples were measured at 378C.
Relaxivity was obtained from the slope of a plot of 1/T1 versus concentra-
tion for 0, 20, 40, and 60mm Gd samples in 660mm HSA. Relaxivities in
HEPES (pH 7.4) buffer were determined using solutions of 0, 100, 150,
and 200mm Gd (HEPES=2-[4-(2-hydroxyethyl)-1-piperazinyl]ethanesul-
fonic acid). The 1H NMRD profiles (5, 358C) were recorded on a field
cycling relaxometer at NY Medical College over the frequency range
0.01 to 50 MHz. For the samples in HSA, the gadolinium concentration
was 100mm and the HSA was 660mm. For the HEPES buffer only solu-
tions, the GdIII concentration was 1mm. Relaxivity was computed by sub-
tracting the relaxation rate of the medium (HSA in HEPES, or HEPES
only) from the relaxation rate of the Gd solution at each field strength
and dividing the difference by the gadolinium in millimoles. All solutions
were assayed for gadolinium concentration by ICP-MS.
17O NMR: H217O transverse relaxation rates were determined for a
HEPES buffer solution in the presence and absence of 9.399 mmolal Gd–
L1 as a function of temperature (ꢀ7 to 958C) on a Varian Unity 300
NMR operating at 40.6 MHz. Probe temperatures were determined from
ethylene glycol or methanol chemical shift calibration curves. T2 was de-
termined by a CPMG pulse sequence. Measurements were repeated after
heating to ensure reproducibility. The variable-temperature relaxation-
rate data was analyzed as described previously[6] to extract water ex-
change and rotational dynamics parameters.
[9] K. Micskei, L. Helm, E. Brucher, A. E. Merbach, Inorg. Chem.
1993, 32, 3844–3850.
[10] K. Micskei, D. H. Powell, L. Helm, E. Brucher, A. E. Merbach,
Magn. Reson. Chem. 1993, 31, 1011–1020.
[11] G. Gonzalez, D. H. Powell, V. Tissieres, A. E. Merbach, J. Phys.
Chem. 1994, 98, 53–59.
[12] D. H. Powell, O. M. Ni Dhubhghaill, D. Pubanz, L. Helm, Y. S. Leb-
edev, W. Schlaepfer, A. E. Merbach, J. Am. Chem. Soc. 1996, 118,
9333–9346.
[13] E. Toth, D. Pubanz, S. Vauthey, L. Helm, A. E. Merbach, Chem.
Eur. J. 1996, 2, 1607–1615.
[14] E. Toth, F. Connac, L. Helm, K. Adzamli, A. E. Merbach, Eur. J.
Inorg. Chem. 1998, 2017–2021.
[15] T. J. McMurry, D. J. Parmelee, H. Sajiki, D. M. Scott, H. S. Ouellet,
R. C. Walovitch, Z. Tyeklar, S. Dumas, P. Bernard, S. Nadler, K.
Midelfort, M. Greenfield, J. Troughton, R. B. Lauffer, J. Med. Chem.
2002, 45, 3465–3474.
[16] J. C. Amedio, Jr., P. J. Bernard, M. Fountain, G. Van Wagenen, Jr.,
Synth. Commun. 1999, 29, 2377–2391.
[17] S. M. Rocklage, S. H. Sheffer, W. P. Cacheris, S. C. Quay, E. F. Hahn,
K. N. Raymond, Inorg. Chem. 1988, 27, 3530–3534.
[18] G. Gran, Acta Chem. Scand. 1950, 4, 559–577.
[19] R. J. Motekaitis, A. E. Martell, Can. J. Chem. 1982, 60, 2403.
[20] A. E. Martell, R. M. Smith, R. J. Motekaitis, NIST Critically Selected
Stability Constants of Metal Complexes: NIST Standard Reference
Data 46, National Institute of Standards and Technology, Washing-
ton, DC, 1997.
[21] W. B. Mims, Proc. R. Soc. London Ser. A 1965, 283, 452.
[22] K. V. Lakshmi, M. J. Reifler, G. W. Brudvig, O. G. Poluektov, A. M.
Wagner, M. C. Thurnauer, J. Phys. Chem. B 2000, 104, 10445–
10448.
[23] A. V. Astashkin, A. M. Raitsimring, P. Caravan, J. Phys. Chem. A
2004, 108, 1900–2001.
[24] A. M. Raitsimring, A. V. Astashkin, D. Baute, D. Goldfarb, P. Cara-
van, J. Phys. Chem. A 2004, 108, 7318–7323.
[25] P. Caravan, A. V. Astashkin, A. M. Raitsimring, Inorg. Chem. 2003,
42, 3972–3974.
Acknowledgements
We would like to acknowledge the many contributions to the field of
MRI contrast agent research and to chemistry as a whole by Professor
Andrꢂ Merbach. One of us (P.C.) is extremely appreciative of having had
Professor Merbach as a mentor. The work done at Argonne National
Laboratory was supported by the U.S. Department of Energy, Office of
Basic Energy Sciences, Division of Chemical Sciences, under Contract W-
31-109-Eng-38.
[26] P. Caravan, C. Comuzzi, W. Crooks, T. J. McMurry, G. R. Choppin,
S. R. Woulfe, Inorg. Chem. 2001, 40, 2170–2176.
[1] R. R. Edelman, J. R. Hesselink, M. B. Zlatkin, MRI: Clinical Mag-
netic Resonance Imaging, Saunders, Philadelphia, 1996.
[2] P. Caravan, J. J. Ellison, T. J. McMurry, R. B. Lauffer, Chem. Rev.
1999, 99, 2293–2352.
[27] W. P. Cacheris, S. C. Quay, S. M. Rocklage, Magn. Reson. Imaging
1990, 8, 467–481.
[28] P. L. Anelli, V. Balzani, L. Prodi, F. Uggeri, Gazz. Chim. Ital. 1991,
121, 359–364.
[3] H. Gries, Top. Curr. Chem. 2002, 221, 1–24.
[29] M. C. Alpoim, A. M. Urbano, C. F. G. C. Geraldes, J. A. Peters, J.
Chem. Soc. Dalton Trans. 1992, 463–467.
[30] C. C. Bryden, C. N. Reilley, Anal. Chem. 1982, 54, 610–615.
[31] C. A. Chang, H. G. Brittain, J. Telser, M. F. Tweedle, Inorg. Chem.
1990, 29, 4468–4473.
[32] C. F. G. C. Geraldes, A. D. Sherry, W. P. Cacheris, K. T. Kuan, R. D.
Brown, III, S. H. Koenig, M. Spiller, Magn. Reson. Med. 1988, 8,
191–199.
[33] S. Aime, M. Botta, S. G. Crich, G. Giovenzana, R. Pagliarin, M.
Sisti, E. Terreno, Magn. Reson. Chem. 1998, 36, S200-S208.
[4] R. B. Lauffer, D. J. Parmelee, S. U. Dunham, H. S. Ouellet, R. P.
Dolan, S. Witte, T. J. McMurry, R. C. Walovitch, Radiology 1998,
207, 529–538.
[5] D. A. Bluemke, A. E. Stillman, K. G. Bis, T. M. Grist, R. A. Baum,
R. DꢁAgostino, E. S. Malden, J. A. Pierro, E. K. Yucel, Radiology
2001, 219, 114–122.
[6] P. Caravan, N. J. Cloutier, M. T. Greenfield, S. A. McDermid, S. U.
Dunham, J. W. Bulte, J. C. Amedio, Jr., R. J. Looby, R. M. Supkow-
ski, W. D. Horrocks, Jr., T. J. McMurry, R. B. Lauffer, J. Am. Chem.
Soc. 2002, 124, 3152–3162.
[7] T. J. Swift, R. E. Connick, J. Chem. Phys. 1962, 37, 307–320.
[8] R. B. Lauffer, Chem. Rev. 1987, 87, 901–927.
Received: March 25, 2005
Published online: July 29, 2005
5874
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 5866 – 5874