The effect of regioisomerism on the coordination chemistry and CEST properties of lanthanide(III) NB-DOTA-tetraamide chelates Topical Issue on Metal-Based MRI Contrast Agents. Guest editor: Valerie C. Pierre

Article abstract of DOI:10.1007/s00775-013-1060-y

Chemical exchange saturation transfer (CEST) offers many advantages as a method of generating contrast in magnetic resonance images. However, many of the exogenous agents currently under investigation suffer from detection limits that are still somewhat short of what can be achieved with more traditional Gd³⁺ agents. To remedy this limitation we have undertaken an investigation of Ln³⁺ DOTA-tetraamide chelates (where DOTA is 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid) that have unusually rigid ligand structures: the nitrobenzyl derivatives of DOTA-tetraamides with (2-phenylethyl)amide substituents. In this report we examine the effect of incorporating hydrophobic amide substituents on water exchange and CEST. The ligand systems chosen afforded a total of three CEST-active isomeric square antiprismatic chelates; each of these chelates was found to have different water exchange and CEST characteristics. The position of a nitrobenzyl substituent on the macrocyclic ring strongly influenced the way in which the chelate and Ln³⁺ coordination cage distorted. These differential distortions were found to affect the rate of water proton exchange in the chelates. But, by far the greatest effect arose from altering the position of the hydrophobic amide substituent, which, when forced upwards around the water binding site, caused a substantial reduction in the rate of water proton exchange. Such slow water proton exchange afforded a chelate that was 4.5 times more effective as a CEST agent than its isomeric counterparts in dry acetonitrile and at low temperatures and very low presaturation powers.

Full text of DOI:10.1007/s00775-013-1060-y

J Biol Inorg Chem (2014) 19:173–189
DOI 10.1007/s00775-013-1060-y
ORIGINAL PAPER
The effect of regioisomerism on the coordination chemistry
and CEST properties of lanthanide(III) NB-DOTA-tetraamide
chelates
Jacqueline R. Slack Mark Woods
•
Received: 13 August 2013 / Accepted: 24 October 2013 / Published online: 28 November 2013
Ó SBIC 2013
3
?
Abstract Chemical exchange saturation transfer (CEST)
offers many advantages as a method of generating contrast
in magnetic resonance images. However, many of the
exogenous agents currently under investigation suffer from
detection limits that are still somewhat short of what can be
influenced the way in which the chelate and Ln coordi-
nation cage distorted. These differential distortions were
found to affect the rate of water proton exchange in the
chelates. But, by far the greatest effect arose from altering
the position of the hydrophobic amide substituent, which,
when forced upwards around the water binding site, caused
a substantial reduction in the rate of water proton
exchange. Such slow water proton exchange afforded a
chelate that was 4.5 times more effective as a CEST agent
than its isomeric counterparts in dry acetonitrile and at low
temperatures and very low presaturation powers.
3
?
achieved with more traditional Gd agents. To remedy
3
this limitation we have undertaken an investigation of Ln
?
DOTA-tetraamide chelates (where DOTA is 1,4,7,10-tet-
raazacyclododecane-1,4,7,10-tetraacetic acid) that have
unusually rigid ligand structures: the nitrobenzyl deriva-
tives of DOTA-tetraamides with (2-phenylethyl)amide
substituents. In this report we examine the effect of
incorporating hydrophobic amide substituents on water
exchange and CEST. The ligand systems chosen afforded a
total of three CEST-active isomeric square antiprismatic
chelates; each of these chelates was found to have different
water exchange and CEST characteristics. The position of a
nitrobenzyl substituent on the macrocyclic ring strongly
Keywords Conformational control ꢀ Lanthanide
chelates ꢀ Macrocyclic ligands ꢀ Paramagnetic
chemical exchange saturation transfer agents ꢀ
Stereoisomerism
Introduction
Interest in the chemistry of macrocyclic lanthanide chelates
has stemmed in large part from the use of gadolinium
chelates as MRI contrast agents. Contrast media based on
Responsible editor: Valerie C. Pierre.
Electronic supplementary material The online version of this
article (doi:10.1007/s00775-013-1060-y) contains supplementary
material, which is available to authorized users.
3
?
the Gd ion are used to shorten the T of the surrounding
1
water, and in T -weighted magnetic resonance images
1
(
short echo and repetition times) this leads to an apparent
J. R. Slack ꢀ M. Woods (&)
Department of Chemistry,
Portland State University,
1719 SW 10th Avenue,
Portland, OR 97201, USA
e-mail: mark.woods@pdx.edu; woodsmar@ohsu.edu
brightening of regions to which the agent has been dis-
tributed. One requirement for a successful T -shortening
1
3?
Gd chelate is that water molecules in the inner coordi-
nation sphere exchange very rapidly with those of the bulk
solvent. This requirement has led to a historical focus on
3
Gd chelates of 1,4,7,10-tetraazacyclododecane-1,4,7,10-
?
M. Woods
Advanced Imaging Research Center,
Oregon Health and Science University,
3181 SW Sam Jackson Park Road, Portland,
OR 97239, USA
tetraacetic acid (DOTA) and its analogues at the expense of
3
?
the related DOTA-tetraamide systems (Fig. 1). The Ln
chelates of DOTA-tetraamide ligands exhibit much slower
123

1
74
J Biol Inorg Chem (2014) 19:173–189
that these slow exchange kinetics could be turned to an
3
advantage. By substituting the Gd ion, which has an
?
3
?
isotropic f shell, for a paramagnetic Ln
ion with an
anisotropic f shell, it is possible to induce very large shifts
in the resonance frequency of coordinated water molecule
protons [2]. These large shifts allow these paramagnetic
chelates to be used as exogenous chemical exchange sat-
uration transfer (CEST) contrast agents of the type pro-
posed by Balaban and coworkers [3]. Balaban proposed
diamagnetic CEST, but paramagnetic CEST (or paraCEST)
offers certain advantages over diamagnetic and endoge-
nous CEST. Firstly, the very large chemical shifts elimi-
nate problems associated with direct off-resonance
saturation of the solvent water that are common with dia-
magnetic agents. Secondly, they permit much faster
exchange kinetics before the slow exchange limit is brea-
ched, potentially allowing for much more effective agents.
In the years since paraCEST contrast agents were first
proposed, a variety of innovative agents have been reported
showcasing some of the many potential advantages of this
approach for generating image contrast. Of particular
importance are the ability to develop ratiometric responsive
probes; the ability to turn off contrast and interleave pre-
contast and postcontrast acquisitions; the ability to perform
‘‘multicolored’’ imaging (imaging more than one agent
simultaneously); and the ability to move beyond lanthanide
chelates and use transition metal chelates. Rather than
attempt to exhaustively cite the many demonstrations of
these advantages, with the inevitable disappointment of
those who are unintentionally overlooked, we direct the
reader to some of the many reviews of this area for a more
comprehensive picture of the field [4–14]. Despite all of
these potential advantages, it is only recently that a con-
certed effort has been made to apply these exogenous
paraCEST agents to in vivo imaging [15–21]. Some of this
delay can be attributed to the relative inefficiency of many
of the paraCEST agents proposed to date. Given the intense
recent activity in this field, work investigating how struc-
tural modification of DOTA-tetraamide chelates might
improve their function as paraCEST agents seems partic-
ularly timely. We have recently reported that the confor-
mation of amide substituents in these chelates can
profoundly alter the CEST properties of the agent [22]. Our
initial idea had been that chiral substituents in the d-posi-
tion of the pendant arms of (S)-2-nitrobenzyl-1,4,7,10-tet-
raazacyclododecane [(S)-NB-cyclen] derived DOTA-
tetraamides (NB-DOTA-tetraamides) would allow the
coordination geometry of the chelate to be controlled in a
manner analogous to that observed for a-substitution [23–
27]. In reality this substitution did not afford control over
the coordination geometry, but the results were more
interesting that we had anticipated. Introducing a d chiral
Fig. 1 The structural formulae of macrocyclic ligands derived from
,4,7,10-tetraazacyclododecane and (S)-2-nitrobenzyl-1,4,7,10-tetra-
azacyclododecane. Ligands 1 and 2 were each produced as two
stereoisomers: S-RRRR and S-SSSS
1
water exchange kinetics than those of DOTA, in large part
because the neutral amide ligands are poorer electron
donors than the anionic acetates of DOTA, rendering the
3
?
Ln ion more electron poor and in greater need of electron
density from water molecules in the inner coordination
sphere. Water molecules in the inner coordination sphere of
DOTA-tetraamide chelates are held more tightly by the
metal and exchange more slowly, rendering these chelates
less effective as T -shortening agents. However, around the
1
-
turn of the last century, Sherry and coworkers [1] realized
carbon, as in Eu1 , afforded some degree of control over
1
23

J Biol Inorg Chem (2014) 19:173–189
175
coordinated water molecule pool, increasing the CEST
effect obtained by saturating the coordinated water mole-
cule pool. In the current report we examine the effect of
replacing the hydrophilic carboxylate with a hydrophobic
3
?
phenyl substituent, Eu2 , our starting hypothesis being
that this should result in a substantial deceleration of the
water exchange rate.
Materials and methods
General remarks
All solvents and reagents were purchased from commercial
sources and used as received unless otherwise stated. The
S and R isomers of 2-bromo-N-(1-phenylethyl)acetamide
(
3) were prepared according to previously described
methods [28]. Purification by high-performance liquid
chromatography (HPLC) was undertaken with a Waters d-
Prep 150 HPLC system using a Phenomenex Luna C-18(2)
reversed-phase (50 mm 9 250 mm) column. Elution was
monitored by the absorption at 270 nm. All NMR spectra
were recorded with a Bruker Advance IIa spectrometer
1
13
operating at 400.13 MHz ( H) and 100.61 MHz ( C)
using a 5-mm broadband probe with the temperature con-
trolled using the installed BVT3200 variable-temperature
controller with a BCU-05 chiller. Samples for two-
dimensional correlation spectroscopy (COSY) were pre-
pared in CD OD (99.8 atom %). Two-dimensional COSY
3
spectra were acquired using 4,096 9 1,024 points using 16
transients per free induction decay at 263 K. Samples for
CEST spectra were prepared in dry CD CN (99.8 atom %),
3
and a known volume of H O was added. The concentration
2
of each sample was determined by comparison of the
1
intensity of the F NMR signal with that of a 1.0 M solution
9
Fig. 2 Description of the nomenclature, and the structural and
conformational parameters related to the chemistry of chiral DOTA-
tetraamide chelates. a Designation of proton nomenclature for protons
on the macrocyclic ring. b Description of the possible conformations
of a chiral amide substituent (shown here with an S configuration)
highlighting how the position of the bulky phenyl substituent changes
depending on the orientation of the pendant arms, either D (right) or
K (left). c Description of the four possible stereoisomeric coordination
geometries of 2. K(dddd) and D(kkkk) are square antiprismatic (SAP)
isomers, whereas K(kkkk) and D(dddd) are twisted square antipris-
matic (TSAP) isomers. The nitrobenzylic substituent on the macro-
cyclic ring has been omitted for clarity, although it may alternatively
be positioned in the equatorial position on either the side or the corner
carbon of one ethylene bridge. The two stereoisomeric structures
rendered inaccessible by substitution of the macrocycle are washed
out. In all structures the configuration at the d-carbon is S
of Cu(OTf) (where OTf is trifluoromethanesulfonate, tri-
2
flate) placed in a concentric insert in the NMR tube. CEST
spectra were acquired at 275 K by measuring the water
signal intensity as a function of the presaturation offset,
in 1-ppm increments. The duration of the presaturation
pulse was 10 s, with a B power of 148, 203, 279, 350, and
1
446 Hz.
0
00 000
(
1R,4R,7R,10R)-d,d ,d ,d -Tetramethyl-[2-(S)-(p-
nitrobenzyl)-1,4,7,10-tetraazacyclododecane]-1,4,7,10-
tetra(2-phenylethyl)acetamide dihydrochloride salt
(
R)-3 (497 mg, 2.05 mmol), (S)-NB-cyclen (150 mg,
the position of the amide substituent; in this case a car-
boxylate. When this substituent was forced upwards into a
pseudoaxial position (see Fig. 2b) towards the coordinated
water molecule, the hydrophilic group was found to
involve a greater number of water protons in the
0.488 mmol), and potassium carbonate (337 mg,
2.44 mmol) were added to acetonitrile (50 mL). The
resulting suspension was heated with stirring for 72 h at
333 K. Following addition of more (R)-3 (83 mg,
0.34 mmol), the reaction mixture was heated for a further
123

1
76
J Biol Inorg Chem (2014) 19:173–189
4
8 h. After the mixture had cooled to room temperature,
143.8, 144.9, 145.2, 146.7, 147.7, 162.0 (C=O), 162.5
(C=O), 168.5 (C=O), 171.4 (C=O). ESI mass spectrometry
the solvents were removed under reduced pressure and the
residue was divided between water (30 mL) and dichlo-
romethane (200 mL). After separation, the aqueous phase
was further extracted with dichloromethane (2 9 50 mL),
the organic phases were combined and dried (Na SO ), and
?
(positive electron ionization) m/z: 952 (100 %, [M ? H] ),
?
974 (46 %, [M ? Na] ).
2
4
General method for the preparation of Ln2(OTf)3
chelates
the solvents removed under reduced pressure. The residue
was taken up into a minimal amount of acetonitrile (6 mL)
and purified by reversed-phase HPLC using a mobile phase
of water (0.37 % HCl) for 5 min followed by a linear
gradient over 20 min to 60 % acetonitrile and 20 % water
The corresponding lanthanide triflate (73 lmol) was dis-
solved in (4 mL) and the mixture was added to a solution
of either S-RRRR-2 or S-SSSS-2 (50 mg, 49 lmol) in 50:50
acetonitrile–water (4 mL). The resulting solution was
heated with stirring for 24 h at 333 K. The solution was
filtered through a 0.45-lm syringe filter and introduced
directly into the HPLC system for purification. Purification
was undertaken using a mobile phase of water (0.1 % tri-
fluoroacetic acid) for 5 min, followed by a linear gradient
to 10 % acetonitrile and 90 % water (0.1 % trifluoroacetic
acid) over 15 min. Each chelate was collected from a
single major peak in the chromatogram, and removal of the
solvents by lyophilization afforded the chelate as the cor-
responding triflate salt as a colorless solid.
(
0.37 % HCl). After removal of the solvent by lyophili-
0
00 000
zation, (1R,4R,7R,10R)-d,d ,d ,d -tetramethyl-[2-(S)-(p-
nitrobenzyl)-1,4,7,10-tetraazacyclododecane]-1,4,7,10-
tetra(2-phenylethyl)acetamide dihydrochloride salt (S-
RRRR-2) was obtained as a colorless solid (1.4 g, 67 %).
1
R = 29.02 min. H NMR (CH CN, 400 MHz) d: 9.52
T
3
3
2H, d, J
3
(
= 8 Hz, p-Ar), 8.04 (2H, d, JH–H = 8 Hz,
H–H
p-Ar), 8.49 (2H, br, CONH), 8.38 (2H, br, CONH), 7.4–6.9
0
(
20H, m, Ar), 4.91 (4H, m, CHCH ), 4.78 [1H, dd (aa
3
3
pattern), J
0
= 7 Hz, CH Ar], 4.76 [1H, dd (aa pattern),
2
H–H
3
JH–H = 7 Hz, CH Ar], 2.5–3.9 (23H, m br, NCH ring
2
2
1
and NCH CO), 1.36 (12H, m, CH ). C NMR (CH CN,
3
2
3
3
1
00 MHz) d: 20.9 (CH ), 21.7 (CH ), 22.0 (CH ), 22.2
3
3
3
S-RRRR-Ce2(OTf)3
(
CH ), 32.6 (CH Ar), 48.6, 48.8, 48.9, 49.0, 49.3, 49.5,
3 2
4
1
9.69, 50.2, 51.1, 51.6, 52.8, 53.3, 53.4, 54.7, 56.5, 56.6,
23.2, 123.5, 125.7, 125.9, 126.3–126.9 (several, overlap-
High-resolution mass spectrometry (HRMS) [ESI time of
3
?
flight (TOF] m/z: [M]
63.8142; found 363.8135; [M ? OTf]
C H F N CeO S 620.1973; found 622.1970. The
calcd for C H N CeO
55 69 9 6
ping), 128.0–128.4 (several, overlapping), 130.2, 143.3,
2
?
3
calcd for
1
1
43.6, 143.9, 144.6, 144.9, 146.4, 146.7, 161.8 (C=O),
62.7 (C=O), 167.9 (C=O), 172.0 (C=O). Electrospray
5
6
69
3
9
9
3
?
appropriate Ce isotope pattern was observed.
ionization (ESI) mass spectrometry (positive electron ion-
?
ization) m/z: 952 (100 %, [M ? H] ).
S-RRRR-Pr2(OTf)3
0 00 000
1S,4S,7S,10S)-d,d ,d ,d -Tetramethyl-[2-(S)-(p-
nitrobenzyl)-1,4,7,10-tetraazacyclododecane]-1,4,7,10-
(
3
?
HRMS (ESI-TOF) m/z: [M]
364.1477; found 364.1472; [M ? OTf]
C H F N PrO S 620.6976; found 620.6977. The appro-
calcd for C H N PrO
55 69 9 6
?
2
tetra(2-phenylethyl)acetamide dihydrochloride salt
calcd for
5
6
69
3
9
9
0
00 000
3?
(
1S,4S,7S,10S)-d,d ,d ,d -Tetramethyl-[2-(S)-(p-nitroben-
priate Pr isotope pattern was observed.
zyl)-1,4,7,10-tetraazacyclododecane]-1,4,7,10-tetra(2-
phenylethyl)acetamide dihydrochloride salt (S-SSSS-2)
was prepared in an similar manner using (S)-3. After
preparative reversed-phase HPLC, S-SSSS-2 was obtained
as a colorless solid (0.8 g, 52 %).
S-RRRR-Nd2(OTf)3
3
?
HRMS (ESI-TOF) m/z: [M] calcd for C H N NdO
6
5
?
5
69
9
2
1
364.4816; found 364.4806; [M ? OTf]
calcd for
R = 29.84 min. H NMR (CD CN, 400 MHz) d: 7.97
T
3
3
2H, d, J
C
56
H
69
F
3
N
9
NdO S 622.1996; found 622.1986. The
9
(
= 8 Hz, p-Ar), 7.44 (20H, m, Ar), 7.31 (2H,
H–H
3?
appropriate Nd isotope pattern was observed.
3
d, J
= 8 Hz, p-Ar), 7.02 (4H, br, CONH), 4.91 (4H,
H–H
m, CHCH ), 4.76 (2H, s br, CH Ar), 2.1–2.7 (23H, m br,
3
2
1
NCH ring and NCH CO), 1.31 (12H, m, CH ). C NMR
3
S-RRRR-Sm2(OTf)3
2
2
3
(
CH CN, 100 MHz) d: 21.2 (CH ), 21.8 (CH ), 22.1
3
3
3
3
?
(
CH ), 22.3 (CH ), 36.4 (CH Ar), 46.9, 47.8, 48.9, 49.0,
3
HRMS (ESI-TOF) m/z: [M] calcd for C H N SmO
55 69 9 6
3
2
2
?
4
5
1
9.4, 49.6, 49.7, 49.8, 51.5, 51.8, 53.0, 53.3, 53.4, 53.7,
6.1, 57.0, 122.8, 125.5–126.9 (several, overlapping),
28.4–128.7 (several, overlapping), 131.4, 142.7, 143.0,
367.6189; found 367.6175; [M ? OTf]
calcd for
C H F N SmO S 626.2044; found 626.2028. The
5
3
appropriate Sm isotope pattern was observed.
6
69
3
9
9
?
1
23

J Biol Inorg Chem (2014) 19:173–189
177
S-RRRR-Eu2(OTf)3
C H F N EuO S 626.7052; found 626.7072. The
56
69
3
9
9
?
3
appropriate Eu isotope pattern was observed.
3
?
HRMS (ESI-TOF) m/z: [M] calcd for C H N EuO
6
5
?
5
69
9
2
3
68.1528; found 368.1538; [M ? OTf]
calcd for
S-SSSS-Tb2(OTf)3
C H F N EuO S 626.7052; found 622.7074. The
9
5
6
69
3
9
3
appropriate Ce isotope pattern was observed.
?
3?
HRMS (ESI-TOF) m/z: [M] calcd for C H N TbO
6
5
?
5
69
9
2
370.1541; found 370.1532; [M ? OTf]
calcd for
S-RRRR-Tb2(OTf)3
C H F N TbO S 629.7072; found 629.7062. The
56
69
3
9
9
3
?
appropriate Tb isotope pattern was observed.
3
?
HRMS (ESI-TOF) m/z: [M] calcd for C H N TbO
6
5
?
5
69
9
2
3
70.1541; found 370.1533; [M ? OTf]
calcd for
S-SSSS-Dy2(OTf)3
C H F N TbO S 629.7072; found 629.7063. The
9
5
6
69
3
9
?
3
appropriate Tb isotope pattern was observed.
3?
HRMS (ESI-TOF) m/z: [M] calcd for C H N DyO
6
5
?
5
69
9
2
371.8221; found 371.8208; [M ? OTf]
calcd for
S-RRRR-Dy2(OTf)3
C H F N DyO S 632.2092; found 632.2075. The
56
69
3
9
9
?
3
appropriate Dy isotope pattern was observed.
3
?
HRMS (ESI-TOF) m/z: [M] calcd for C H N DyO
6
5
?
5
69
9
2
3
71.8221; found 371.8213; [M ? OTf]
calcd for
S-SSSS-Ho2(OTf)3
C H F N DyO S 632.2092; found 632.2083. The
5
6
69
3
9
9
?
3
appropriate Dy isotope pattern was observed.
3?
HRMS (ESI-TOF) m/z: [M] calcd for C H N HoO
6
5
?
5
69
9
2
372.1558; found 372.1547; [M ? OTf]
calcd for
S-RRRR-Er2(OTf)3
C H F N HoO S 632.7097; found 632.7086. The
56
69
3
9
9
3
?
appropriate Ho isotope pattern was observed.
3
?
HRMS (ESI-TOF) m/z: [M] calcd for C H N ErO
6
5
?
5
69
9
2
3
72.4891; found 372.4881; [M ? OTf]
calcd for
S-SSSS-Er2(OTf)3
C H F N ErO S 634.2108; found 634.2098. The appro-
9
5
6
69
3
9
3
priate Er isotope pattern was observed.
?
3?
HRMS (ESI-TOF) m/z: [M] calcd for C H N ErO
6
5
?
5
69
9
2
372.4891; found 372.4886; [M ? OTf]
calcd for
S-RRRR-Yb2(OTf)3
C H F N ErO S 634.2108; found 634.2001. The appro-
9
56
69
3
9
3
priate Er isotope pattern was observed.
?
3
?
HRMS (ESI-TOF) m/z: [M] calcd for C H N YbO
6
5
?
5
69
9
2
3
75.1586; found 375.1589; [M ? OTf]
calcd for
S-SSSS-Yb2(OTf)3
C H F N YbO S 637.2140; found 637.2162. The
5
6
69
3
9
9
?
3
appropriate Yb isotope pattern was observed.
3?
HRMS (ESI-TOF) m/z: [M] calcd for C H N YbO
6
5
?
5
69
9
2
375.1586; found 375.1599; [M ? OTf]
calcd for
S-SSSS-Ce2(OTf)3
C H F N YbO S 637.2140; found 622.2162. The
9
56
69
3
9
3
appropriate Yb isotope pattern was observed.
?
3
?
HRMS (ESI-TOF) m/z: [M] calcd for C H N CeO
6
5
?
5
69
9
2
3
63.8142; found 363.8133; [M ? OTf]
calcd for
C H F N CeO S 620.1973; found 620.1967. The
9
Results and discussion
5
6
69
3
9
3
?
appropriate Ce isotope pattern was observed.
Chelate preparation
S-SSSS-Pr2(OTf)3
The tetraamide derivatives of NB-cyclen pose a greater
synthetic challenge than the corresponding acetate deriva-
tives. In our experience, chelates derived from NB-cyclen
do not crystallize, and this means that there are very few
options for obtaining isomerically pure samples; our pre-
ferred method is to perform preparative reversed-phase
HPLC. This has proven to be very effective for the prepa-
ration of hydrophilic anionic chelates in which the nitro-
benzyl substituent causes substantial retention on a C18
stationary phase, requiring a fairly large proportion of
3
?
HRMS (ESI-TOF) m/z: [M]
64.1477; found 364.1473; [M ? OTf]
C H F N PrO S 620.6976; found 620.6976. The appro-
calcd for C H N PrO
55 69 9 6
?
2
3
calcd for
5
6
69
3
3
9
9
?
priate Pr isotope pattern was observed.
S-SSSS-Eu2(OTf)3
3
?
HRMS (ESI-TOF) m/z: [M] calcd for C H N EuO
6
5
5
69
9
2
?
3
68.1528; found 368.1537; [M ? OTf]
calcd for
123

1
78
J Biol Inorg Chem (2014) 19:173–189
3
?
Scheme 1 The synthesis of Ln2 chelates shown for the S-RRRR
isomer. The reagents and conditions were as follows: i 273 K/K CO
CH Cl ii (S)-2-(p-nitrobenzyl)-1,4,7,10-tetraazacyclododecane/
CO /MeCN/333 K; iii purification by reversed-phase high-
3 2
performance liquid chromatography; iv Ln(OTf) /MeCN/H O (pH
2
3
/
5.5)/333 K; v purification by reversed-phase high-performance liquid
chromatography
2
2
;
K
2
3
acetonitrile in an aqueous solvent to induce elution [24]. A
-
similar situation prevailed for the previously reported Ln1
sample was found to contain two chelates. At the time this
work was undertaken this result was quite unexpected. The
appearance of regioisomeric chelates was first noted by
Ranganathan et al. [29] and is now well documented in
rigid NB-DOTA derived chelates [26, 27], but was not
NB-DOTA-tetraamide chelates—the anionic substituents
giving rise to monoanionic chelates. However, chelates of
neutral NB-DOTA-tetraamide ligands afford trications,
which seem to have higher affinity for the mobile phase and
reduced affinity for the stationary phase (on a C18 column)
and these chelates are much more readily eluted often even
with pure aqueous solvent systems. This previously
unpublished observation caused us to take a slightly dif-
-
previously noted during the preparation of Ln1 chelates
[22]. This may have been because at that time insufficient
attention was paid to the identity of reaction by-products
and the production of a second regioisomer during chela-
tion was simply overlooked. However, it is important to
note that the mechanism of chelate formation for acetate
ligands, the step in which the two regioisomers differen-
tiate, is quite different in tetraamide ligands [30]. Despite
this difference and our previously proposed mechanisms
[26] for the formation of each regioisomer, it is now clear
that both acetate and amide chelation mechanisms can
differentiate regioisomers, presumably dependent on the
direction of approach of the metal ion. Each chelate was
then subjected to a second round of HPLC purification to
try and separate the two regioisomers. However, in all
cases the chelates were eluted as a single major peak,
which subsequent analysis by NMR spectroscopy and mass
spectrometry showed contained two isomeric chelates. In
consequence, only minor impurities and excess metal were
removed during this second round of purification.
3?
ferent approach to the preparation of Ln2 chelates.
Two suitable alkylating agents—(R)-3 and (S)-3—for
the introduction of the pendant arms were prepared using
previously published methods (Scheme 1) [28]. Samples of
(
S)-NB-cyclen were then alkylated with each bromoacet-
amide in acetonitrile using potassium carbonate as a base at
33 K for 3 days. Following removal of the solvents and
3
aqueous workup, the oily residues were taken up into
acetonitrile and purified by reversed-phase HPLC. The aim
here was to remove all diastereoisomeric by-products at the
ligand stage—although these should have been minimal
because no transformation was undertaken at any chiral
carbon—prior to producing the chelate, at which stage
purification would be more challenging. Elution with water
(
0.37 % HCl) and acetonitrile afforded S-RRRR-2 and S-
SSSS-2 as its corresponding dihydrochloride salt in the
form of a colorless solid. Chelates of each ligand were
prepared by dissolving S-RRRR-2 and S-SSSS-2 in a 50:50
mixture of acetonitrile and water and then adding an excess
of the corresponding lanthanide triflate. The pH was
adjusted to 5.5 with sodium hydroxide solution, and the
solutions were heated at 333 K for 24 h. After the solvents
had been removed by lyophilization, each chelate was
A second consideration must be borne in mind when
purifying tricationic NB-DOTA-tetraamide chelates such
as those of 2. The nature of the counterion has been shown
to have a significant effect on the properties of the chelate
[31]. This was not a consideration during the preparation of
-
the Ln1 chelates since the chelate itself was the anion and
the acidic eluent afforded the chelate as the conjugate acid.
In previous studies on tricationic DOTA-tetraamides we
tried to restrict the counterion to the convenient triflate
1
analyzed by H NMR spectroscopy. To our surprise, each
1
23

J Biol Inorg Chem (2014) 19:173–189
179
3
?
anion for reasons of consistency [32–34]. However, triflic
acid is not a good additive to an HPLC eluent because
excess acid is not readily removed under a vacuum. On the
basis of our experience with anionic chelates, it was
expected that retention on the HPLC column would effect
anion exchange with the conjugate base of the eluent acid.
For this reason the hydrochloric acid in the eluent was
replaced with trifluoroacetic acid, the acid most like triflic
acid that is suitable as an eluent additive. To our surprise,
four stereoisomeric structures are possible, and for Ln
chelates such as those of DOTA, all four conformations are
accessible and in dynamic exchange. However, when
substituents are introduced onto the pendant arm or the
macrocyclic ring, the configuration at the resulting chiral
carbon will help define the helicity of that component of
the ligand. A popular substituent is a nitrophenyl, of which
the nitro group is readily converted to the isothiocyanate,
facilitating an easy conjugation to most primary amines.
Nitrophenyl groups have been introduced in different
positions of the ligand carbon skeleton with minimal effect
on the performance of the chelates as contrast media, but
with a significant effect on the coordination chemistry [41–
43]. For instance, we have shown previously that intro-
ducing a nitrobenzyl substituent onto the cyclen ring, with
an S configuration at carbon, freezes the macrocycle into
the dddd conformation, minimizing torsional strain by
placing the nitrobenzyl substituent in an equatorial position
[43]. This conformational constraint is absolute in this case
and the kkkk conformation is effectively inaccessible even
at relatively high temperatures. The K orientation of the
pendant arms is congruent with the configuration at the d-
carbon (R), which means there is no force attempting to
drive this chelate into a TSAP coordination geometry.
However, two macrocyclic structures are still possible—
this is because there are two distinct equatorial positions on
the macrocyclic ring: one on the corner of the [3333]
conformation and one on the side.
3
?
HRMS analysis of each Ln2 chelate revealed the pre-
sence of triflate counterions after purification by reversed-
phase HPLC, but no indication that either chloride or tri-
fluoroacetate counterions were present. The view that each
chelate was isolated as the triflate salt is further supported
1
9
by F NMR analysis, which revealed just a single peak,
indicating that anion exchange with trifluoroacetate did not
occur to any degree during purification of the chelates by
3
?
reversed-phase HPLC. The interaction between Ln2
chelates and the triflate counterions seems to be quite
strong indeed, much stronger than that between the chelate
and either chloride or trifluoroacetate. We may conclude,
therefore, that after a second round of HPLC purification,
3
?
all Ln2 chelates were obtained as a mixture of two iso-
mers in the form of the triflate salt, in which form all
studies were performed.
3
Coordination chemistry of S-RRRR-Ln2 chelates
?
3
?
3?
The Ln chelates of the 1,4,7,10-tetraazacyclododecane
cyclen)-derived analogue of 2 have been extensively
It is of little surprise that all the S-RRRR-Ln2 chelates
studied adopted a SAP coordination geometry. This is
readily determined by examining the lanthanide-induced
S
shifts (LIS) of the most shifted axial ring protons (ax )
(
studied by Dickins and Parker and their coworkers [34–38].
In common with other DOTA-tetraamides bearing bulky
aromatic amide substituents, these chelates exhibit a
marked preference for the square antiprismatic (SAP)
coordination geometry [14, 32, 39]. DOTA-type ligands
(Fig. 3) [44]. Because of the locking effect of the nitro-
benzyl substituent a single stereochemistry is possible for
this SAP coordination geometry: K(dddd). The two che-
3
?
1
will sandwich an Ln
ion between the four coplanar
lates observed in the H NMR spectrum can arise only
nitrogen atoms of the macrocyclic ring and four coplanar
oxygen atoms from the pendant arms (carboxylates,
amides, phosphonates, phosphinates, etc.). This affords a
very stable coordination mode which is usually, but not
always, capped above the four pendant arm oxygen atoms
by a coordinated water molecule. In this coordination mode
from regioisomerism of the nitrobenzyl substituent: either
‘‘side’’ or ‘‘corner.’’ It is tempting to examine these NMR
data to probe the change in regioisomeric distribution
across the lanthanide series; however, this temptation must
be resisted. The production of two regioisomers was not
anticipated prior to preparation of the chelates and so
3
?
3?
(
the only one observed for Ln ions, although others are
reaction conditions from one Ln to another were not
sometimes observed for transition metal ions) [40] both the
macrocyclic ring and the pendant arms are defined by the
helicity of their conformation: either dddd or kkkk in the
ring; and D or K in the arms (Fig. 2c). The relationship
between these two conformations defines the coordination
geometry of the chelate: when the two are opposed—
D(kkkk) and K(dddd)—the chelate adopts a SAP coordi-
nation geometry; when the two are the same—K(kkkk)
and D(dddd)—the chelate adopts a twisted square antipr-
ismatic (TSAP) coordination geometry (Fig. 2c). Thus,
carefully controlled. We have previously shown for NB-
DOTMA chelates that factors such as pH and concentration
have a measurable effect on the distribution of regioi-
somers [26]. One must also consider that these samples
were subjected to preparative reversed-phase HPLC after
the chelates had formed, and it is far from certain that for
each sample the entirety of each peak was collected as it
eluted—if the tail of a peak was cut off for one chelate but
collected for another, this would introduce a sizable dis-
crepancy in the observed isomeric distributions.
123

1
80
J Biol Inorg Chem (2014) 19:173–189
1
Fig. 4 The H– H correlation spectroscopy (COSY) spectrum of S-
1
3
?
S
RRRR-Eu2 expanded to show only the couplings to the four ax
protons of each chelate. The spectrum was acquired in CD OD at
00 MHz and 263 K. The resonances of the proton on the nitroben-
3
4
zyl-substituted carbons are identified by a double dagger (side
isomer) and a dagger (corner isomer). nOe nuclear Overhauser effect
1 3?
Fig. 3 The H NMR spectra of S-RRRR-Ln2 chelates recorded in
CD OD at 400 MHz at 263 K, except for Dy and Ho (298 K).
3
the ring. In contrast, the resonance just downfield of
34 ppm exhibits three couplings, but none are of sufficient
magnitude to be a geminal coupling. Furthermore, one of
these couplings is to a benzylic proton at about 8 ppm—
this is not a J coupling but, as observed in the COSY
spectra of NB-DOTMA [27], is a nuclear Overhauser effect
cross-peak that arises from the proximity of this proton to
the benzylic protons of the nitrobenzyl substituent. This
indicates that this isomer has the nitrobenzyl group located
on this carbon, adjacent to this proton; i.e., the substituent
is located on the side of the macrocyclic ring. One further
observation is worthy of note: the magnitude of the vicinal
3
?
3?
S
The spectra are expanded to show on the region in which only the ax
proton resonances would be expected to appear. The spectra clearly
show that all of the chelates adopt the SAP coordination geometry
exclusively, but that in every case a mixture of side and corner
isomers is observed
1
1
The two regioisomers can be identified through H– H
COSY experiments. The COSY spectrum of S-RRRR-
3
?
S
Eu2 is shown in Fig. 4. For six of the eight ax reso-
nances, three J couplings are evident. In order of
S
decreasing magnitude they are a geminal ax –eq coupling,
S
S
C
S
C
S
C
a vicinal ax –ax coupling, and a vicinal ax –eq coupling,
as described in Fig. 2a. It is notable that the most shifted
resonance is coupled to only two other protons. From the
magnitude of these couplings and the shifts of the protons
to which it is coupled, it is possible to conclude that the
ax –eq coupling is significantly larger on this ethylene
bridge than all the others in the spectrum. Curiously in the
case of NB-DOTMA [27], and even NB-DOTA [43], the
magnitude of this coupling was found to be smaller than
those of the other ethylene bridges, and on occasion so
much smaller that it was not observed in the spectrum. In
light of the Karplus relationship [45], one must conclude
that in each case the ethylene bridge is being distorted by
equatorial proton on the carbon at the corner of the ring
C
eq ) is absent, i.e., the nitrobenzyl substituent is located
(
on this ethylene bridge, but on the carbon at the corner of
1
23

J Biol Inorg Chem (2014) 19:173–189
181
apical position that is of intermediate rate on the NMR
timescale, resulting in broad lines. In fact, in experiments
described later samples of some chelates were prepared in
CD CN and small amounts of water were added. In some
3
cases when insufficient water had been added, and the
NMR spectra were acquired at 263 K, resonances from
both the CH CN-ligated and H O-ligated chelates could be
3
2
clearly resolved (Fig. S1). Because both water and CD OD
3
induced very similar LIS, the ligand exchange reaction can
very easily enter the fast exchange regime, affording
sharper lines. However, to achieve better spectral resolu-
tion through increased LIS, it was often desirable to
acquire NMR data in CD OD at low temperatures. This
3
line of reasoning was supported by the observation that
some of the later lanthanides, which induce much larger
LIS, would often afford broader lines at lower temperatures
in CD OD, presumably because the ligand exchange
3
reaction had been slowed to an intermediate exchange
regime. Nonetheless, there is an alternative explanation for
the observed line broadening; that is, that the two regioi-
somers themselves are in chemical exchange. If that were
the case, then as the temperature of the sample is increased,
the resonances should consistently move towards one
another. In NB-DOTMA chelates we have observed that it
is possible to interconvert the two regioisomers only by
demetallation of the chelate [26], a process that does not
occur on the NMR timescale, but to rule out this possi-
S
Fig. 5 The effect on temperature of the ax protons shifts of S-
RRRR-Eu2 at 400 MHz in CD OD. The resonances arising from
3
the corner isomer are denoted by circles, and those arising from the
side isomer are denoted by diamonds. The protons adjacent to the
nitrobenzyl substituent are denoted by a dagger (corner) and a double
dagger (side)
3?
1
the nitrobenzyl substituent. In an ideal [3333] ring con-
formation each ethylene bridge adopts a gauche confor-
mation, but if the ethylene bridge distorts in such a way as
bility, a variable-temperature H NMR experiment was
3
?
undertaken on S-RRRR-Eu2 (Fig. 5).
S
The shifts of all the ax resonances decrease with
S
C
to make the angle (U) between the ax and eq protons
greater the J coupling between these two protons will
decrease, as observed previously for NB-DOTMA deriva-
tives [27]. Because this vicinal J coupling increases in the
increasing temperature, which is consistent with the inverse
correlation of both the dipolar and the contact contributions
to the LIS with temperature. However, it is notable that
none of the resonances move closer together in a consistent
manner, and significantly the two nitrobenzyl-substituted
peaks remain as far apart at high temperature as they do at
low temperature. We may therefore conclude that, like
their acetate analogues, the regioisomers of NB-DOTA-
tetraamide derivatives are not in chemical exchange and
3
?
side isomer of S-RRRR-Eu2 , the ethylene bridge on
which the nitrobenzyl substituent is located must be dis-
S
C
torting in the opposite direction, making the ax –eq angle
U) smaller (Fig. 2a).
One further point should be addressed: the H NMR data
(
1
shown herein are generally obtained in CD OD and at low
3
the line broadening observed in CD CN solution is the
3
temperatures. The reason for this is that it was frequently
found that the spectral lines of these chelates acquired in
result of ligand exchange reactions on the intermediate
timescale.
CD CN and/or at room temperature were quite broad.
3
3
?
Coordination chemistry of S-SSSS-Ln2 chelates
However, switching the solvent to CD OD had two effects:
3
the lines tended to be sharper but the LIS tended to be
smaller. This observation is consistent with the solvent
3
?
The coordination chemistry of S-SSSS-Ln2 chelates is
richer and more complex than that of the S-RRRR isomers.
Here again the conformation of the macrocyclic ring is
locked into a dddd conformation by the nitrobenzyl sub-
stituent. But in this case the configuration at the d chiral
carbon of each pendant arm (S) is opposed to the helicity
required (K) for the chelate to adopt a SAP coordination
geometry. This means that each chelate has two choices:
3
?
interacting directly with the Ln ion in the apical position
usually occupied with water. Dickins et al. [34] found that
in this position, CD CN would induce larger LIS than
3
CD OD, which induces LIS of a similar order of magnitude
3
as water. This could then be the origin of the line broad-
ening observed in these chelates. In the presence of only
traces of water, the chelates undergo ligand exchange in the
123

1
82
J Biol Inorg Chem (2014) 19:173–189
1
Fig. 7 The H– H COSY spectrum of S-SSSS-Eu2 expanded to
1
3?
S
show only the couplings to the ax protons of each chelate. The
3
spectrum was acquired in CD OD at 400 MHz and 263 K. The
protons adjacent to the nitrobenzyl substituent are denoted by a red
dagger (side) and yellow double dagger (corner)
1 3?
Fig. 6 The H NMR spectra of S-SSSS-Ln2 chelates expanded to
S
focus on the region in which the most shifted ax protons are expected
3
to appear. The spectra were recorded in CD OD at 400 MHz and
-
reported work on S-SSSS-Eu1 [22]. This S-SSSS-Eu1
-
3
?
3?
3?
3?
63 K, except for Tb , Dy , Ho , and Er (298 K)
2
chelate also existed as a mixture of SAP and TSAP iso-
mers; interestingly, we found the proportions of each to be
the same in each of these chelates: 5 % SAP and 95 %
TSAP. Furthermore, the distribution of the coordination
either position the pendant arms in the D orientation pre-
ferred by an S configuration and adopt a TSAP coordina-
tion geometry, or force the arms into a K orientation,
pushing the bulky phenyl substituents into a less favorable
pseudoaxial position above the chelate and adopt a SAP
coordination geometry (Fig. 2). For the early lanthanides
3
?
3?
geometries in either S-SSSS-Eu2 or S-SSSS-Yb2 was
found to be insensitive to either changes in temperature or
solvent (Fig. S2), in marked contrast to the behavior of
some other related chelates [32]. Apparently, differences in
the nature of the amide substituent—smaller and hydro-
3
?
the choice is clear. The larger ionic radius of the early Ln
-
ions is readily accommodated by the more open TSAP
3?
philic for S-SSSS-Eu1 and larger and hydrophobic for S-
3?
SSSS-Eu2 —have little or no effect on the preferred
3
?
geometry [46, 47] and the Ce and Pr chelates of S-
SSSS-2 adopt this coordination geometry exclusively
coordination geometry distribution of the chelate.
3?
The identity of each chelate of S-SSSS-Eu2 obtained
3
Fig. 6). However, as the ionic radius of the Ln ion
?
(
S
was also determined through examination of the ax cou-
decreases, the innate preference for the SAP coordination
1
geometry begins to win out. Thus, we see that in the H
1
1
plings in an H– H COSY spectrum (Fig. 7). Using the
same logic applied in the analysis of the COSY spectrum of
3?
3?
NMR spectrum of S-SSSS-Eu2 small amounts of the SAP
3
?
isomer are present, and as the Ln ionic radius decreases
further, progressively larger and larger proportions of the
SAP isomer are observed (Fig. 6).
S-RRRR-Eu2 (Fig. 4) and the Karplus relationship, one
S
can identify the two TSAP ax protons located on the same
ethylene bridge as the nitrobenzyl group. The most shifted
S
ax resonance (a little downfield of 15 ppm) is found to
3
?
The shift pattern of one of the two S-SSSS-Eu2 reg-
ioisomeric chelates appears to exactly mirror that of the
only regioisomeric chelate isolated in our previously
have three couplings, two vicinal couplings and a nuclear
Overhauser effect cross-peak correlated with a benzylic
1
23

J Biol Inorg Chem (2014) 19:173–189
183
proton. The absence of a geminal coupling and the pre-
sence of the nuclear Overhauser effect cross-peak corre-
lated with a benzylic proton is clear evidence that this
proton is located on the same carbon as the nitrobenzyl
substituent; i.e., this is a ‘‘side isomer’’. One more obser-
S
C
vation of note is that the ax –eq vicinal coupling is
weaker in this case than the same couplings in other eth-
ylene bridges of the same chelate. This is in direct contrast
to the observation made for the side isomer of S-RRRR-
3
?
Eu2 , for which the equivalent coupling was found to be
stronger, but consistent with the observation made for NB-
DOTMA chelates [27]. On the basis of the Karplus rela-
tionship, this suggests that the nitrobenzyl substituent dis-
torts the gauche conformation of the ethylene bridge on
S
C
which it is located in such a way as to increase the ax –eq
angle (U). One other resonance is found to have only two
J couplings, one that is shifted to about 12.5 ppm. In this
S
case, however, the couplings are clearly one geminal ax –
S
S
C
eq coupling and one vicinal ax –ax coupling. In other
words this proton is also located on the same ethylene
bridge as the nitrobenzyl group, but in this case the sub-
stituent is located on the corner carbon. Thus, we may
conclude that here again we observe a mixture of ‘‘side’’
and ‘‘corner’’ isomers. It is worth reiterating our note of
caution against the temptation to examine the distribution
of side and corner isomers across the series, because here
again the experiments were not conducted in such a manner
as to allow such an analysis. On the basis of the shift
pattern, the single SAP isomer observed in the spectrum of
S
Fig. 8 The effect of temperature on the shifts of the ax proton
resonances of S-SSSS-Eu2 in CD OD at 400 MHz. The corner
3
isomer is denoted by the circles and is found to adopt solely a TSAP
coordination geometry. The side isomer is denoted by diamonds and
is observed in both the SAP coordination geometry (filled symbols)
and the TSAP coordination geometry (open symbols)
3?
-
of S-SSSS-Eu1 , in conjunction with population distribu-
-
tion data, reveals that the isomer of S-SSSS-Eu1 reported
previously must also have been a side isomer.
3
?
S-SSSS-Eu2 appears to be a side isomer. Owing to the
low concentrations of this structure only geminal J cou-
plings are observed in the COSY spectrum, and although
one must be circumspect in examining the absence of
cross-peaks, the absence of a geminal coupling to the most
shifted SAP resonance is consistent with the assignment of
this as a side isomer.
These variable-temperature data conclusively demon-
strate that the SAP and TSAP coordination geometries of
3
?
the side isomer of S-SSSS-Eu2 are in chemical exchange
that is slow on the NMR timescale. It is noteworthy that
evidence for this same exchange in the side isomer of S-
-
SSSS-Eu1 was not found, which does not mean that
The assignment of this SAP isomer as a side isomer is
S
confirmed by analysis of the ax chemical shifts of S-SSSS-
exchange is not occurring in this case, just that it was not
observed. In fact, in light of the observed evidence for
3?
3
?
Eu2 as a function of temperature (Fig. 8). Here again we
observe that although the absolute shifts of each TSAP
resonance decrease with increasing temperature, there is no
movement of the side and corner isomers towards one
another. This confirms that the side and corner isomers are
not in chemical exchange with one another. Interestingly,
exchange in the side isomer of S-SSSS-Eu2 , it seems
more likely than not that a SAP $ TSAP chemical equi-
-
librium exists for the side isomer of S-SSSS-Eu1 .
3
As the ionic radius of the Ln ion decreases yet further
?
3?
3?
from Eu to Yb , progressively more and more SAP
geometry is observed for each chelate (Figs. 6, 9). Notably
3?
S
the chemical shifts of four ax protons from one of the
a SAP geometry of the corner isomer, absent for Eu , is
TSAP isomers decrease at a slower rate with increasing
temperature than those of the other four. This appears to be
the result of chemical exchange between this TSAP isomer
and the single SAP isomer that is observed. This affords an
eventually observed for the later lanthanides. The SAP/
TSAP distribution is never the same between the side and
3?
corner isomers for any given Ln ion. Figure 9 shows the
distribution of SAP and TSAP isomers for the corner and
3?
S
unambiguous assignment of these four TSAP ax protons
side isomers of some S-SSSS-Ln2 chelates across the
to the same structure, the side isomer. Examining the shift
patterns of the two coordination geometries of the side
series. In many cases the paramagnetic line broadening of
3
?
the Ln ion precludes an accurate determination of these
ratios, leaving only six data points across the series to
3
?
isomer of S-SSSS-Eu2 and those of the reported isomer
123

1
84
J Biol Inorg Chem (2014) 19:173–189
S
in the aforementioned shift patterns of the ax protons. The
LIS of these protons is almost exclusively the result of a
through-space dipolar shift, which is dependent on the
spatial location of the proton relative to the metal ion and
the magnetic field induced by the metal ion within its
ligand field. It is now evident that chelates based on NB-
cyclen are substantially distorted relative to their cyclen-
based counterparts. However, the uniform differences in
S
the ax shift pattern indicate that the two regioisomeric
chelates distort differently. This distortion may result in a
difference in the spatial positioning of the protons in the
ligand, which, if reflected in the arrangement of the ligating
donor atoms, could lead to a significant nonaxial compo-
nent of the magnetic susceptibility tensor. It seems most
likely both of these occur in these chelates. The resulting
S
ax shift patterns appears to have a good deal in common
S
with the ax shift pattern observed for a bridged DOTA-
Fig. 9 The distribution of SAP [K(dddd)] and TSAP [D(dddd)]
coordination geometries of the side (diamonds) and corner (circles)
3?
tetraamide chelate [48]. The side isomers of Ln2 more
3
isomers of S-SSSS-Ln2 . The dashed lines are a guide for the eye
?
3
?
closely resemble the diagonally distorted bridged Yb
chelates, and the corner isomers resemble the asymmetri-
3
?
illustrate the trend. The assignment of ‘‘side’’ or ‘‘corner’’
S
was made on the basis of the shift pattern of the ax pro-
cally distorted Pr chelates [48]. Such a difference in the
way in which the regioisomeric chelates distort could give
rise to the difference in the size of the coordination cage
and their isomeric distributions. If so, it could also be
expected to have a profound effect on the behavior of the
coordinated water molecule and therefore the CEST
properties of the chelates.
3
?
tons, which appears to be quite consistent over all Ln
ions and SAP and TSAP isomers. The side isomer is
generally characterized by the four resonances appearing in
two pairs, whereas the corner isomer is characterized by
resonances appearing in one pair, with the other resonances
shifted further away, one upfield and one downfield
(
Figs. 3, 5, 6, 8).
Water exchange and saturation transfer experiments
Despite the relatively low number of data points, the
trend in the isomeric ratios is quite clear. As with other
related systems, the TSAP geometry is favored by the
It is now well established that the kinetics of water proton
exchange are substantially reduced in aprotic solvents such
as acetonitrile [33]. Studying the chelates in acetonitrile
solution aids direct observation and assessment of water
3?
larger, early Ln
ions, whereas the SAP geometry is
3
?
increasingly favored by the smaller, later Ln ions. It is
noteworthy that this preference is considerably more pro-
nounced and occurs earlier in the series for the side isomer
3
?
molecules coordinated to the Ln center. To observe the
coordinated water molecules under conditions of minimal
3
?
1
3?
line-broadening, H NMR spectra of S-RRRR-Pr2 , S-
of S-SSSS-Ln2 . However, the situation for chelates of S-
3
SSSS-Ln2 is not directly comparable to that in previously
?
3?
RRRR-Nd2 , S-RRRR-Eu2 , S-SSSS-Pr2 , S-SSSS-
3?
3?
3
?
Eu2 were recorded at 275 K in CD CN with a minimum
published studies of DOTA and DOTA-tetraamide chelates
3
[
43, 47–49]. As discussed earlier, the presence of the chiral
of H O added (Figs. 10, S3). The resulting spectra contain
2
amide substituents renders rotation of the pendant arms
from D to K an energetically unfavorable process. Con-
version of the D(dddd) TSAP isomer to the K(dddd) SAP
isomer can occur only through this arm rotation process.
Thus, a TSAP to SAP conversion will occur only if this
energy penalty is offset by the chelate adopting a SAP
geometry. Clearly this offset is greater for the side isomer,
which more readily adopts a SAP geometry, even with
two resonances corresponding to coordinated water mole-
cules, clearly resolving the resonances from each of the
two regioisomeric chelates present in each sample. In the
3
?
spectrum of S-SSSS-Pr2 (Fig. 10, top) the resonances of
water coordinated to the side and corner isomers of the
TSAP geometry are only barely resolved; the shift differ-
ence is very small. Furthermore, the intensity of each res-
onance is approximately equivalent, consistent with
roughly equal amounts of each regioisomer, rendering it
impossible to accurately identify which peak corresponds
to which diastereoisomer or whether there is a difference in
the exchange rate between the two. In contrast, the two
resonances of water coordinated to the two regioisomers of
3
?
moderately sized Ln ions. This suggests that the coor-
dination ‘‘cage’’ afforded by the side isomer is more suited
3
?
to the accommodation of smaller Ln ions; i.e., it is sig-
nificantly smaller than that afforded by the corner isomer.
A clue to how the coordination cage differs may be found
1
23

J Biol Inorg Chem (2014) 19:173–189
185
1 3?
Fig. 10 The H NMR spectra of S-SSSS-Pr2 (top) and S-RRRR-
3
?
2 3
Pr2 (bottom) in 0.2 % w/v H O in CD CN at 263 K at 400 MHz,
clearly showing the resonances of the water molecules associated
with metal coordination in each regioisomeric chelate
3
?
the SAP isomer in the spectrum of S-RRRR-Pr2 are very
well resolved, with a shift difference somewhat more than
5
ppm. Furthermore, because the two regioisomers are
present in unequal proportions, it is possible to assign each
resonance to its regioisomer: the more shifted water reso-
nance is that coordinated to the side isomer. It is notable
that the resonance from the side isomer is somewhat nar-
rower than that from the corner isomer, and this may be an
indication that exchange of these water protons with those
of the water in the solvent is slower than for those of the
corner isomer. However, one must also take into account
Fig. 11 The chemical exchange saturation transfer (CEST) spectra of
3?
3?
6
.775 mM (Eu ) solutions of S-SSSS-Eu2
(top) and S-RRRR-
CN with
2 1
.1 % w/v H O, using a B presaturation pulse of 10 s at 148 Hz
bound
free
3?
that the H O
2
/H O
2
ratio is different for the two re-
free
Eu2 (bottom) recorded at 263 K and 400 MHz in CD
3
1
gioisomers of this chelate in this sample (the H O
2
(
(
orange lines), 203 Hz (purple lines), 279 Hz (green lines), 350 Hz
cyan lines), and 446 Hz (red lines)
concentration is comparatively small) and this complicates
a direct analysis of these data. By way of comparison, the
1
3?
3?
H NMR spectra of S-RRRR-Nd2 and S-RRRR-Eu2
Fig. S3) also show two resolved coordinated water reso-
(
pulse applied was of sufficiently low power, it was possible
to resolve separate CEST peaks for the side and corner
nances. However, in each of these cases the amounts of
each regioisomer present are more equal and the coordi-
nated water resonances of each regioisomer are quite
3
?
3?
regioisomers of both S-RRRR-Pr2
and S-SSSS-Pr2
(Fig. S4). Similarly, in the CEST spectra of S-RRRR-
3
?
3?
3?
similar in shape. Notably in the spectrum of S-SSSS-Eu2
Nd2 (Fig. S4) and S-RRRR-Eu2 (Fig. 11) one CEST
peak is resolved for water coordinated to each regioisomer
in solution. Given the wealth of information on the CEST
no resonances corresponding to a coordinated water mol-
ecule are observed, suggesting that for both regioisomeric
TSAP chelates the water proton exchange rate increases
3
?
of Eu chelates, for this study we chose to focus our
3?
attention on the two stereoisomeric Eu2 chelates. The
3
CEST spectra of S-RRRR-Eu2 are rather unusual; CEST
3
?
rapidly as the Ln ionic radius decreases.
?
Assessment of water exchange kinetics is more readily
achieved in these types of DOTA-tetraamide chelate by
means of CEST experiments. The effectiveness of a chelate
as a CEST agent is assessed by recording its ‘‘CEST
spectrum’’. This is achieved by measuring the intensity of
the solvent water peak as a function of the frequency offset
of a soft presaturation pulse. CEST spectra were recorded
from two coordinated water molecules is observed. Also
observed at various stages over the selected B power
1
range, are CEST peaks arising from several amide protons
close to the peak arising from direct saturation of water at
0 ppm. These peaks are of note because in general the
CEST peaks from amide protons in SAP isomers of the
EuDOTA-tetraamide chelate are observed slightly upfield
of the direct saturation peak. However, in the spectrum of
under similar conditions—minimal water in CD CN and
3
2
63 K—for the same five chelates. When the presaturation
123

1
86
J Biol Inorg Chem (2014) 19:173–189
3
?
S-RRRR-Eu2 CEST peaks of differing intensities can be
observed both upfield and downfield of the direct saturation
peak. This leads us to conclude that the distortion of the
DOTA framework induced by the nitrobenzyl group, dis-
cussed earlier, has two effects on the amide protons.
Firstly, it alters the electronic distribution in each arm,
causing the amide protons of each pendant arm to exchange
at different rates, hence the differences in intensity on
4.5 times greater than that of either regioisomer of S-
3
?
RRRR-Eu2
.
It is significant that the greatest difference in the
3
?
effectiveness of the side SAP isomers of S-SSSS-2 and S-
3
RRRR-2 is at the lowest B power. To meet specific
?
1
absorption rate restrictions in MRI examinations it is
imperative that the maximum amount of CEST be gener-
ated from the lowest possible B powers. From Woessner’s
1
changing the B power. Secondly, it causes unusual LIS of
1
analysis of the Bloch equations modified for exchange [14,
50], it is clear that to achieve this it is necessary to have
very slow proton exchange kinetics. The fact that S-SSSS-
each of these protons, shifting some upfield and others
downfield.
3
?
In principle CEST is limited by the slow exchange
condition (kex B Dx), but in practice it is often possible to
observe CEST from resonances that exchange slightly
faster than this limit. For example, in the CEST spectrum
Eu2 has higher potency as a CEST agent at the lowest
presaturation power (B power of 148 Hz), of all three
1
SAP isomers is suggestive that our initial hypothesis is
correct—forcing the phenyl substituents into a pseudoaxial
position slows exchange. It is common practice to deter-
mine exchange kinetics by fitting the CEST spectrum to the
Bloch equations modified for exchange. In the present
cases, such an analysis is prohibited by the presence of a
large (11 or 16) number of potentially unique exchanging
pools in each spectrum. Given the large number of vari-
ables involved in each analysis (upwards of 33), a number
that could not reasonably be reduced, this is clearly not a
scientifically robust method for analyzing these data.
Instead we chose to use the Dixon–Sherry ‘‘omega
method,’’ in which the CEST generated is plotted as a
3
?
of S-SSSS-Eu2 , a broad CEST peak is observed arising
from the rapidly exchanging water protons of the TSAP
isomers even though peaks from these resonances are not
1
visible in the H NMR spectrum (Fig. S3). This CEST peak
is centered at about ?25 ppm, consistent with water
3
?
coordinated to Eu in a TSAP coordination geometry but
much too far from the direct saturation peak to arise from
an amide proton of the pendant arm. However, the spec-
trum is most notable for the small CEST peak arising from
the side SAP isomer at 66 ppm. This peak arises from a
species that is present in only very low quantities; consti-
tuting slightly less than 5 % of the side isomer, its overall
concentration in solution is just 0.14 mM. Although it is
present in only small quantities, this chelate is able to
generate a comparatively large CEST effect. The CEST
arising from this chelate on a per millimole basis is much
larger than that arising from either regioisomer of S-RRRR-
function of the presaturation B power to afford the proton
1
residence lifetime as the intercept of the x-axis [51]. The
3
omega plots for both regioisomers of S-RRRR-2 afford
?
straight lines as expected. Each isomer has a different
CEST response to changing the B power; the omega plot
1
reveals that this originates from a difference in the water
proton exchange rate (Fig. 12, Table 1). This result is
rather unexpected; it suggests that even relatively small
distortions to the lanthanide coordination cage can have
subtly different effects on water proton exchange. It is
3
?
Eu2 . CEST is highly sensitive to a number of variables,
3
so unlike Gd chelates there is no single parameter that
?
quantifies the effectiveness of an agent in the manner of
?
relaxivity. However, because the data for each Eu2
3
3
?
chelate were acquired under identical conditions, it is
possible to make comparisons across these data (and only
these data) by calculating the percent CEST generated by
worthy of note that in these Eu chelates it is the more
shifted water that appears to exchange more quickly. In
3
contrast, for the Pr chelates, it appeared that the less
?
the chelate on a per millimolar basis for each B power
1
shifted corner isomer underwent more rapid exchange
used (Table 1). The ‘‘effectiveness’’ of the side SAP iso-
(Fig. 10). No assignment of each water proton resonance to
3?
3
mer of S-SSSS-Eu2 as a CEST agent is between 3.6 and
?
an individual regioisomer of S-RRRR-Eu2
could be
Table 1 Parameters relevant to the chemical exchange saturation transfer (CEST) spectra shown in Fig. 11 for the three square antiprismatic
(
3?
SAP) isomers of Eu2 (for the structure, see Fig. 1)
H
M
Peak (ppm)
Eu2 (mM)
s
(ms)
CEST (%/mM)
1
48 Hz
203 Hz
279 Hz
350 Hz
446 Hz
3
?
S-RRRR-Eu2
63
3.14
3.64
0.14
3.70
10.4
8.6
13.3
11.7
53.4
16.3
14.3
58.7
18.2
16.2
66.2
19.6
17.5
74.2
6
5
3.15
3
?
S-SSSS-Eu2
66
4.68–10.6
47.1
1
23

J Biol Inorg Chem (2014) 19:173–189
187
the exchange process, when exchange is slow, interferes
with the model; or possibly the theory is undermined by the
presence of other abundant exchanging pools. With only
one anomalous data set it is not currently possible to
determine the origin of this nonlinearity. The nonlinearity
of the data means that they cannot be simply fitted using
linear regression analysis to afford the exchange rate.
However, by fitting these data to each extreme (Fig. 12),
one can generate a range within which the exchange rate
must fall (Table 1). This exercise shows that the water
proton residence lifetime lies in the range 4.6–10.4 ms.
Although this is quite a wide range, it is substantially
longer than the water proton residence lifetimes for either
3
?
regioisomer of S-RRRR-Eu2 , which are between 3.1 and
.7 ms. Although these water proton residence lifetimes
3
are all very long in comparison with those usually observed
for the DOTA-tetraamide chelates in aqueous solution, it
must be noted that these data were acquired at very low
temperatures (just 263 K) and with very low water con-
centrations (611 mM compared with 55.6 M in H O).
2
Conclusions
The problems encountered in separating the side and corner
3
?
regioisomers of Ln2 chelates have provided many more
insights into the coordination chemistry of nitrobenzyl-
substituted chelates than were gained from our initial
-
studies into Ln1 chelates alone. Significantly, the pref-
erence for coordination geometry follows the same overall
-
trend that was observed for the Ln1 chelates: the S-RRRR
isomer exhibits a preference for the SAP coordination
geometry across the series and the S-SSSS isomer exhibits a
preference for the TSAP isomer, but with increasing
Fig. 12 Omega plots generated from the CEST data of the three SAP
3
?
3?
isomers of Eu2 shown in Fig. 11; top S-SSSS-Eu2 and bottom S-
3?
3?
RRRR-Eu2 (blue line 63 ppm, red line 65 ppm)
amounts of SAP isomer observed as the Ln ionic radius
decreases. The amount of SAP isomer present differs for
the side and corner regioisomers, hinting at a difference in
the size of the coordination cage between the two regioi-
made; the effectively equal distribution of corner and side
isomers in the sample means there is no basis for dis-
criminating one from the other. As a consequence, it is not
possible to conclude whether the same regioisomer
3?
somers. The idea that the central Ln ion is differentially
crowded by the two regioisomers is supported by the
observation that the water proton exchange kinetics of the
side and corner isomers are not the same. Finally, we find
that placing large amide substituents into a pseudoaxial
position (Fig. 2) can have substantial and unexpected
effects on the water proton pool and its exchange charac-
3
?
exchanges most rapidly for both S-RRRR-Pr2 and S-
3
RRRR-Eu2 or whether it is a different isomer for each
?
3
Ln ion. It is conceivable that the exchange rate differs
?
3
?
between corner and side isomers with differing Ln ions
and the most rapidly exchanging regioisomer may not
consistently be the same.
-
teristics. In the case of S-SSSS-Ln1 , the pseudoaxial
carboxylates of the SAP isomer appear to recruit additional
The omega plot for the SAP side isomer of S-SSSS-
3
?
Eu2 is nonlinear; there is a clear inflection point around
the middle of the data. It is not clear why this should be.
One can speculate that this could arise for a number of
reasons: there is excessive noise or error in the data; the
theory on which this plot is based breaks down when
protons into the ‘‘bound water’’ pool, enhancing CEST. In
3
?
the case of S-SSSS-Ln2 , the pseudoaxial phenyl groups
of the SAP isomer appears to substantially slow the water
exchange kinetics of the chelate. This affords a chelate that
is as much as 4.5 times more effective as a CEST agent
than an isomeric chelate with the phenyl substituents in the
exchange is very, very slow; the involvement of CD CN in
3
123

1
88
J Biol Inorg Chem (2014) 19:173–189
pseudoequatorial position. Clearly, the conditions under
which these chelates have been studied do not approach
those prevailing physiologically. Several modifications
would need to be made to these systems before they could
be useful as CEST agents for MRI. Ideally the agent would
adopt exclusively the conformation of the SAP isomer
19. Liu G, Li Y, Shethy VR, Pagel MD (2012) Mol Imaging
1
1:47–57
0. Sheth VR, Li Y, Chen LQ, Howison CM, Flask CA, Pagel MD
2012) Magn Reson Med 67:760–768
2
(
21. Yoo B, Sheth VR, Howison CM, Douglas MJK, Pineda CT,
Maine EA, Baker AF, Pagel MD (2013) Magn Reson Med.
doi:10.1002/mrm.24763.
2
2. Carney CE, Tran AD, Wang J, Schabel MC, Sherry AD, Woods
M (2011) Chem Eur J 17:10372–10378
3?
observed for S-SSSS-Eu2 . The chelates would need to be
rendered both nontoxic (as these tricationic chelates are
almost not well tolerated in vivo) [52] and made water-
soluble; both of these prerequisites could be achieved
through incorporation of three or four anionic groups, such
as carboxylates, into the chelate structure [52].
23. Woods M, Kovacs Z, Zhang S, Sherry AD (2003) Angew Chem
Int Ed 42:5889–5892
2
2
2
4. Woods M, Botta M, Avedano S, Wang J, Sherry AD (2005)
Dalton Trans 3829–3837
5. Di Bari L, Pescitelli G, Sherry AD, Woods M (2005) Inorg Chem
44:8391–8398
6. Tircso G, Webber BC, Kucera BE, Young VG, Woods M (2011)
Inorg Chem 50:7966–7979
Acknowledgments The authors thank the National Institutes of
Health (EB-11687), the Oregon Nanoscience and Microtechnologies
Institute (N00014-11-1-0193), Portland State University, and the
Oregon Opportunity for Biomedical Research for financial support of
this work.
27. Webber BC, Woods M (2012) Inorg Chem 51:8576–8582
28. Graham B, Loh CT, Swarbrick JD, Ung P, Shin J, Yagi H, Jia X,
Chhabra S, Barlow N, Pintacuda G, Huber T, Otting G (2011)
Bioconjug Chem 22:2118–2125
2
3
3
9. Ranganathan RS, Raju N, Fan H, Zhang X, Tweedle MF, Desreux
JF, Jacques V (2002) Inorg Chem 41:6856–6866
0. Baranyai Z, Banyai I, Brucher E, Kiraly R, Terreno E (2007) Eur
J Inorg Chem 2007:3639–3645
1. Thompson AL, Parker D, Fulton DA, Howard JAK, Pandya SU,
Puschmann H, Senanayake K, Stenson PA, Badari A, Botta M,
Avedano S, Aime S (2006) Dalton Trans 5605–5616
References
1
2
. Zhang S, Winter P, Wu K, Sherry AD (2001) J Am Chem Soc
23:1517–1518
. Sherry AD, Zhang S, Woods M (2005) In: Sessler JL, Doctrow
SR, McMurray TJ, Lippard SJ, (eds) Medicinal inorganic
chemistry. ACS symposium series 903. American Chemical
Society, Washington, pp 151–165
1
32. Miller KJ, Saherwala AA, Webber BC, Wu Y, Sherry AD,
Woods M (2010) Inorg Chem 49:8662–8664
33. Woods M, Pasha A, Zhao P, Tircso G, Chowdhury S, Kiefer G,
Woessner DE, Sherry AD (2011) Dalton Trans 40:6759–6764
34. Dickins RS, Parker D, Bruce JI, Tozer DJ (2003) Dalton Trans
1264–1271
3
. Ward KM, Aletras AH, Balaban RS (2000) J Magn Reson
1
43:79–87
. Bonnet CS, Toth E (2010) Am J Neuroradiol 31:401–409
. Dorazio SJ, Morrow JR (2012) Eur Inorg Chem
012:2006–2014
35. Thompson AL, Parker D, Fulton DA, Howard JAK, Pandya SU,
Puschmann H, Senanayake K, Stenson PA, Badari A, Botta M,
Avedano S, Aime S (2006) Dalton Trans 5605–5616
36. Frias JC, Bobba G, Cann MJ, Hutchison CJ, Parker D (2003) Org
Biomol Chem 1:905–907
37. Parker D, Dickins RS, Puschmann H, Crossland C, Howard JAK
(2002) Chem Rev 102:1977–2010
38. Aime S, Barge A, Batsanov AS, Botta M, Castelli DD, Fedeli F,
Mortillaro A, Parker D, Puschmann H (2002) Chem Commun
1120–1121
4
5
J
2
6
7
8
. Gianolio E, Stefania R, Di Gregorio E, Aime S (2012) Eur J Inorg
Chem 2012:1934–1944
. Hancu I, Dixon WT, Woods M, Vinogradov E, Sherry AD,
Lenkinski RE (2010) Acta Radiol 51:910–923
. Pacheco-Torres J, Calle D, Lizarbe B, Negri V, Ubide C, Fayos
R, Larrubia Pilar L, Ballesteros P, Cerdan S (2011) Curr Top Med
Chem 11:115–130
. Sherry AD, Woods M (2007) In: Modo MMJ, Bulte JWM (eds)
Molecular and cellular MR imaging. CRC, Boca Raton,
pp 101–122
39. Woods M, Zhang S, Von Howard E, Sherry AD (2003) Chem Eur
J 9:4634–4640
40. Meyer M, Dahaoui-Gindrey V, Lecomte C, Guilard L (1998)
Coord Chem Rev 180:1313–1405
9
1
0. Terreno E, Delli Castelli D, Aime S (2010) Contrast Media Mol
Imaging 5:78–98
41. Aime S, Botta M, Ermondi G, Terreno E, Anelli PL, Fedeli F,
Uggeri F (1996) Inorg Chem 35:2726–2736
1
1
1. Tsitovich PB, Morrow JR (2012) Inorg Chim Acta 393:3–11
2. Viswanathan S, Kovacs Z, Green KN, Ratnakar SJ, Sherry AD
42. Adair C, Woods M, Zhao PY, Pasha A, Winters PM, Lanza GM,
Athey P, Sherry AD, Kiefer GE (2007) Contrast Media Mol
Imaging 2:55–58
(2010) Chem Rev 110:2960–3018
1
1
1
3. Winter PM (2012) Wiley Interdiscip Rev Nanomed Nanobio-
technol 4:389–398
4. Woods M, Woessner DE, Sherry AD (2006) Chem Soc Rev
43. Woods M, Kovacs Z, Kiraly R, Br u¨ cher E, Zhang S, Sherry AD
(2004) Inorg Chem 43:2845–2851
44. Marques MPM, Geraldes CFGC, Sherry AD, Merbach AE,
Powell H, Pubanz D, Aime S, Botta M (1995) J Alloys Compd
225:303–307
45. Karplus M (1963) J Am Chem Soc 85:2870–2871
46. Aime S, Botta M, Ermondi G (1992) Inorg Chem 31:4291–4299
47. Aime S, Botta M, Fasano M, Marques MPM, Geraldes CFGC,
Pubanz D, Merbach AE (1997) Inorg Chem 36:2059–2068
48. Vipond J, Woods M, Zhao P, Tircso G, Ren JM, Bott SG, Ogrin
D, Kiefer GE, Kovacs Z, Sherry AD (2007) Inorg Chem
46:2584–2595
3
5:500–511
5. Ali MM, Bhuiyan MPI, Janic B, Varma NRS, Mikkelsen T,
Ewing JR, Knight RA, Pagel MD, Arbab AS (2012) Nanomedi-
cine 7:1827–1837
6. Ali MM, Yoo B, Pagel MD (2009) Mol Pharm 6:1409–1416
7. Ferrauto G, Delli Castelli D, Terreno E, Aime S (2013) Magn
Reson Med 69:1703–1711
1
1
1
8. Li AX, Suchy M, Li C, Gati JS, Meakin S, Hudson RHE, Menon
RS, Bartha R (2011) Magn Reson Med 66:67–72
1
23

J Biol Inorg Chem (2014) 19:173–189
189
4
5
5
9. Aime S, Botta M, Garda Z, Kucera BE, Tircso G, Young VG,
Woods M (2011) Inorg Chem 50:7955–7965
0. Woessner DE, Zhang S, Merritt ME, Sherry AD (2005) Magn
Reson Med 53:790–799
1. Dixon WT, Ren J, Lubag AJM, Ratnakar J, Vinogradov E, Hancu
I, Lenkinski RE, Sherry AD (2010) Magn Reson Med
52. Woods M, Caravan P, Geraldes CFGC, Greenfield MT, Kiefer
GE, Lin M, McMillan K, Prata MIM, Santos AC, Sun X, Wang J,
Zhang S, Zhao P, Sherry AD (2008) Invest Radiol 43:861–870
6
3:625–632
123