Simultaneous Triple Imaging with Two PARASHIFT Probes: Encoding Anatomical, pH and Temperature Information using Magnetic Resonance Shift Imaging

Article abstract of DOI:10.1002/chem.201700447

The chemical shift of paramagnetically shifted resonances in lanthanide(III) complexes encodes information about temperature and has also been made to report pH in parallel, through introduction of a single phosphonate group adjacent to the reporter tert-

Full text of DOI:10.1002/chem.201700447

A Journal of
Accepted Article
Title: Simultaneous triple imaging with two PARASHIFT probes:
encoding anatomical, pH and temperature information using
magnetic resonance shift imaging
Authors: David Parker, Katie-Louise Finney, Alice Harnden, Nicola
Rogers, Kanthi Senanayake, Andrew Blamire, and Dara
O'Hogain
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201700447
Link to VoR: http://dx.doi.org/10.1002/chem.201700447
Supported by

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
Simultaneous triple imaging with two PARASHIFT probes:
encoding anatomical, pH and temperature information
using magnetic resonance shift imaging
[
a]
[a]
[a]
Katie-Louise N. A. Finney , Alice C. Harnden , Nicola J. Rogers , P. Kanthi
[
a]
[b]
[b]
[a]
Senanayake , Andrew M. Blamire , Dara O’Hogain and David Parker *
a) Department of Chemistry, Durham University, South Road, Durham DH1 3LE, UK.
b) Institute of Cellular Medicine & Newcastle MR Centre, Newcastle University,
Newcastle upon Tyne, NE4 5PL, UK.
E-mail: david.parker@dur.ac.uk . Supporting information and ORCID(s) from the author(s) for
this article are available on the www under http://dx.doi.org/10.1002/chem.xxxxxxxxx
Abstract: The chemical shift of paramagnetically shifted resonances in lanthanide(III)
complexes encodes information about temperature and has also been made to report pH in
parallel, through introduction of a single phosphonate group adjacent to the reporter t-butyl
resonance. The enhanced sensitivity of this new probe has allowed the simultaneous triple
imaging of the water signal and the shifted t-butyl signals of thulium and dysprosium
complexes of a common ligand, separated by over 160 ppm. In parallel spectral imaging
experiments, the temperature and pH dependence of the frequency of the Tm and Dy signals
has been deconvoluted, allowing the pH and temperature in the liver, kidney and bladder to
be measured.
Introduction
We introduce a spectroscopic imaging system that allows the spatial distribution of water to
be assessed at the same time as mapping local tissue temperature and pH, using two
paramagnetic complexes of the same ligand.
Proton magnetic resonance spectroscopy in vivo is an inherently insensitive method that has
remained far less exploited than MRI, but offers the opportunity to identify and map
metabolic and physiological processes via the chemically shifted resonance of species other
than water. Both endogenous species and added probes normally need to be present in the
millimolar concentration regime to be detected quickly, over the magnetic field range 1.5 to 7
ꢀ
1ꢀ
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
T. On the other hand, the abundant local water signal resonating around 4.7 ppm can be very
quickly observed, and both its varying local concentration and differing longitudinal
[
1,2]
relaxation time in different tissue types allows valuable image contrast to be obtained.
There are two coupled practical problems inhibiting co-observation of the water signal and
other proton signals in vivo. First, the endogenous or probe signal needs to be enhanced, and
hyperpolarisation techniques have shown considerable promise, notwithstanding their short
[
3]
duration. Another means of enhancing signal intensity per unit time involves paramagnetic
relaxation, wherein a faster relaxing probe signal can be observed with a 15 to 25 times
[
4-6]
increase in intensity versus diamagnetic controls, using fast pulsed NMR techniques.
Second, the observed signal should be shifted well away from the water and fat signals to
allow it to be observed separately, requiring a very large sweep width to be used with very
short repetition times, without the need for water suppression. At a field of 7 T (300 MHz),
the signal should resonate about 15 to 20,000 Hz away from the water/fat signals, i.e. beyond
±
50/60 ppm, (Scheme 1).
Paramagnetic probes based on the fast-relaxing lanthanide(III) ions offer solutions to these
t
problems. To improve sensitivity they can be labelled with homotopic Bu groups that
–
1
t
undergo fast relaxation in the range 80 to 150 s , over the field range 1 to 7 Tesla. The Bu
resonance, when it is located about 6.6 Å from the metal ion in an 8-coordinate complex, can
3
+
[7,8]
be tuned to resonate ±80 ppm from the water signal, by varying the Ln ion.
the lanthanide ion from the series of five ‘fast-relaxing’ and ‘large-shifting’ rare-earth ions
Tb, Dy, Ho, Er and Tm) according to the magnetic field to be used, it is possible to select a
pair of complexes with similarly large, but opposite paramagnetic shifts and similar R and R
By selecting
(
1
2
values. Such features allow the implementation of a common pulse sequence to
simultaneously and directly observe each resonance. Moreover, in addition to a very large
temperature sensitivity of the observed chemical shift, responsiveness to local pH change can
be engineered (or in principle, sensitivity to pM), by introducing an ionisable group that
[
9-11a]
protonates between pH 5 and 8, perturbing the reporter resonance frequency.
We
emphasise here the fundamental difference between our approach using paramagnetic shift
and increased relaxivity to enhance the molecular MR signal from the complex itself, as
opposed to other indirect detection approaches, whereby the interaction between complex and
solvent (tissue water) is detected (e.g. CEST methodology).
Altered extracellular pH in vivo is associated with chronic diseases, such as end stage renal
failure and coronary artery disease, as well as in the tissue ischemia associated with many
cancers. There are very few non-invasive markers used clinically to quantify pH in deep
ꢀ
2ꢀ
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
[
11]
tissue.
calibration
suggested pH values as low as 6.3 in the calyx, whereas the cortex and medulla are believed
There is a degree of uncertainty about the pH of the kidney due to difficulties in
[
11b]
: Gillies using the pH-dependence of Gd complex relaxivity changes,
[
11c]
to be higher. More recent data
also suggests values of 7.3, 7.0 and 6.6 for the cortex,
medulla and calyx respectively, dropping by 0.2 pH units in the ‘acidotic’ disease states that
typify renal failure. Similarly, elevated local temperatures are often associated with
inflammation, notably in hepatitis and in liver and kidney fibrosis. The systems here, i.e.
1
a -
a
based on [Ln.L ] have pK values (50% aq. serum, vide infra) of 6.7: these are well suited to
the clinical need.
1
–
With these issues in mind, we have developed the series of complexes, [HLn.L ] (Scheme 1)
in which the lanthanide ion is Tb, Dy, Ho, Er or Tm. The ligand structure incorporates an
arylphosphonate group, with a pK value in the region of 7, according to the local ionic
a
strength, solvent polarity and temperature. Phenylphosphonic acid itself has been studied by
3
1
[12]
P magnetic resonance as an in vivo pH probe in magnetic resonance spectroscopy.
Scheme 1.
Using two complexes of the same ligand, e.g. Tm(III) and Dy(III), the dual pH/T dependence
of the observed chemical shift can be resolved, provided that the intrinsic pH and T
dependence of the shift for each complex is known, under the conditions found in vivo.
Therefore, by consecutively exciting each well-separated resonance within a brief pulse delay
(
compared to T2), a spectroscopic imaging experiment can be devised, observing the water
t
signal and the two Bu probe resonances by dual excitation, in the same acquisition period.
Pulse sequences that can be used to acquire data on two or more resonances simultaneously
[
13]
–
have been developed some time ago.
Coman has studied the behaviour of [Tm.DOTMA]
5
–
and the pH sensitive complex [Tm.DOTP] , examining multiple frequencies that were very
[
14,15]
well shifted on either side of the water signal.
A process of interleaved acquisitions was
used, representing a reduction in sensitivity, as they were effectively running at a longer
ꢀ
3ꢀ
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
(
(
double) repetition time, T . In that case, the very widely separated 4H reporter signals
R
~26,000 Hz at 500 MHz) had short T and T values of 1 and 0.5 ms respectively, that may
1
2
have made the concept of a dual excitation experiment not feasible, at that time. These
important studies were carried out in nephrectomised animals, using a continuous infusion of
the complex and operated at a field of 11.7 T, in order to get sufficient sensitivity to observe
the rather broad signal. The use of this surgical approach raises and artificially maintains a
high blood concentration of complex, such that in a rat model, complex passed the blood-
brain-barrier and it was possible to measure the distribution of pH from individual voxels
across the entire rat brain. More recently, Huang has extended that work to use a more
efficient dual-band excitation pulse which simultaneously detects the dual frequencies of the
5
–
[
(
Tm.DOTP] complex and evaluated the use of an organic anion transporter inhibitor
[
15b]
probenecid) to slow excretion of the agent
. In the work reported herein, imaging studies
were undertaken using a normal, intact mouse model, with a single tail-vein injection of 0.04
mmoles/kg and with no attempt to alter natural excretion rates. A single pulse train was used
to create a simultaneous triple imaging experiment of agents and water, for the first time.
Results and Discussion
1
-
The synthesis of the target complexes, [HLn.L ] was undertaken using a convergent route in
which the key heterocyclic moiety was introduced late in the scheme by an alkylation reaction
at the fourth ring nitrogen position (Schemes 2 and 3). Nitration of 3-t-butyl-6-methyl
7
pyridine N-oxide gave the 4-nitro derivative, 1, and stepwise reduction, diazotization and
trapping of the aryldiazonium ion with iodide, afforded the p-iodo derivative, 3. Thermolysis
in acetic anhydride led to functionalisation of the proximate benzylic site generating the
[
16]
acetate ester, 4. Palladium catalysis of C-P bond formation using P(OEt)
3
,
followed by
selective base hydrolysis of the carboxylate ester, gave the primary alcohol, 6, from which
the mesylate, 7, was prepared under standard conditions.
NO
2
NH
2
I
tBu
H SO , HNO
2
4
3
tBu
Pd/C, H
2
tBu
4 2
HBF , NaNO , KI
tBu
N
O
74%
N
O
MeOH, 98%
N
O
H
2
O, 31%
N
O
1
2
3
(CH
3
CO)
2
O
8
1%
OEt
OEt
OEt
O
OEt
O
OEt
O
OEt
P
P
P
I
MsCl, NEt
THF
3
,
tBu
t
Bu
cat. NaOEt/ EtOH
t
Bu
tBu
O
O
HO
O
O
9
3%
N
57%
N
P(OEt)
3
6
Pd(dba)
5%
2
,
N
S
N
O
O
O
7
6
5
4
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
Scheme 2
ꢀ
4ꢀ
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
ꢀ
OEt
O
OEt
P
tBu
O
O
OEt
O
O
O
O
OH
O-
S
N
P
OEt
O
P
P
P
OEt
O
K2CO3, MeCN
27%
i) 6 M HCl
7
N
N
N
N
tBu
N
N
N
N
t
N
N
Bu
+
Ln
O
EtO
O
OEt
P
ii) LnCl .6H O,
3
2
P
P
H
H2O, pH 5.5
O
O
N
N
N
P
P
EtO
O
O
O
EtO
N
P
HLn.L1]-
[
1
3
O
P
EtO
O
Ln = Eu, Gd, Tb, Dy, Ho, Er, Tm
1
2
ꢀ
Scheme 3
The triester, 12 (ESI for synthesis) was prepared from mono-BOC-cyclen, and alkylation with
in acetonitrile yielded the penta-ester, 13, which was hydrolysed under acidic conditions.
7
1
-
Complexes with seven different lanthanide ion salts were prepared, [HLn.L ] , (Ln = Eu, Gd,
Tb, Dy, Ho, Er and Tm), each of which was purified by reverse phase HPLC.
Temperature and pH Dependence of the Chemical Shift
Chemical shift data for the paramagnetic complexes (Table 1) shows the very large chemical
shift range for the t-butyl resonance: from +74 ppm for the thulium complex to -78 and -83
ppm for the Dy and Tb complexes respectively. The unusually large shifts for the Tm and Tb
complexes do not follow the expected trend predicted by Bleaney’s theory of magnetic
anisotropy. Such behaviour is in agreement with recent observations that have highlighted
the need to appreciate that this theory is not generally applicable, and is quite imprecise when
the relative magnitude of the components of the magnetic susceptibility tensor varies from
[
7,17]
one lanthanide to another, within a common isostructural series.
t
The chemical shift of the Bu group varied as a function of pH, (Figure 1) allowing pK
a
values
to be determined, associated with protonation of the phosphonate group. Two major isomeric
species were observed in a 5:1 ratio that was more or less invariant with the nature of the
lanthanide ion. The minor species was shifted to higher frequency by ca. 2.5 ppm across the
pH range. Previous work with the parent complex lacking the phosphonate group and its
analogues has suggested that the major species observed in solution derives from the
enantiomeric (RRR-Λ−λλλλ) and (SSS-Δ−δδδδ) complexes, in which the lanthanide ion is 8-
[
8]
coordinate and adopts a twisted square anti-prismatic structure. The minor isomer, by
[
18a]
analogy with the stereoisomerism found for related mono-substituted triphosphinates
is
most likely to be the enantiomeric pair, (SSS-Λ−λλλλ) and (RRR-Δ−δδδδ), as suggested by
recent studies, in which X-ray and NMR structural analysis of an analogous Yb complex,
ꢀ
5ꢀ
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
lacking the 4-phosphonate substituent, has confirmed the Yb(III) ion to be 8-coordinate, in a
[
18b]
twisted square-antiprismatic geometry with an RRR-Λ−(λλλλ) configuration.
1
t
1 –
2
Table 1 H NMR shift data for the Bu reporter resonance in [HLn.L ] (295 K, D O)
a
b
Ln
δH
range of
T sensitivity
ppm/K
pKa (error)
/
ppm
pH sensitivity /ppm
Tb
Dy
Ho
Er
–84.1
–79.1
–33.0
+42.2
+74.8
1.4
1.9
1.1
0.7
1.1
+0.52
+0.41
+0.17
–0.24
–0.44
7.20(05)
7.11(04)
7.12(04)
7.11(04)
7.26(05)
Tm
a
Maximal chemical shift where pD < 5.5; pD values were estimated using the equation: pD
pH (meter reading) + 0.41.
=
b
a a
The reported error in the stated pK value reflects the goodness of the statistical fit; for pK
values in serum and in 0.1 M NaCl, see Figures 3 and 4 : a pH titration in 50:50 murine
1
plasma/H
2
O for [HDy.L ] gave a pK
a
value of 6.80 (±0.08) (295 K).
The pK
a
values averaged 7.16(±0.06) in D
2
O, in agreement with values expected for
[
12]
t
ionization of phenylphosphonic acid,
suggesting that the Bu group does not impose any
steric effect to inhibit hydration of the anionic species involved in the equilibrium. In 0.1 M
NaCl, the pK reduced to 6.71(06), in accord with the expected ionic strength dependence for
a
phosphonate ionization (Figure 2). Surprisingly but usefully, the shift range of pH sensitivity
varied from one lanthanide complex to another, (Figure 3), yet not in a manner that was
consistently proportional to the size of the pseudo-contact shift.
-
2
In contrast, the pseudo-contact shift inherently exhibits a precise T dependence on absolute
-
1
temperature that approximates well to T over the physiological temperature range (Table 1,
Figures 4 and 5), and was observed to be independent of pH and solvent composition. Thus,
given that the Tm and Dy complexes show different pH/shift and temperature dependencies,
the observed shifts for each complex in the region of interest being examined in vivo, at a
given pH and T, allow both the T and the pH to be calculated when each complex is observed
simultaneously using spectroscopic imaging, with a ‘triple imaging’ pulse sequence.
ꢀ
6ꢀ
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
O
O
9.3
O
O
P
P
O
9
.0
8.9
.8
N
N
N
N
N
Dy
O
P
8
O
P
O
O
8.4
7.9
high pH
7
7
7
.8
.6
.3
H+
+ H+
-
O
O
OH
O
O
7.1
.6
5.8
P
P
6
N
N
N
N
Dy
O
N
P
5.2
O
P
O
4
.9
O
low pH
4.6
-75.0 -76.0 -77.0 -78.0 -79.0 -80.0 -81.0 -82.0 -83.0
H
δ /ꢀppm
1
Figure 1. The shift in the major tert-butyl resonance of [Dy.L ] as a function of pD (D
1.7 T, 295 K) and the corresponding protonation equilibrium.
2
O,
1
1
Figure 2. The shift in the major tert-butyl resonance of [Dy.L ] as a function of pH or pD in
an 0.1 M NaCl solution in H O (black), pK = 6.71(0.06), and in pure D O (red), pK
.11(05) (11.7 T, 295 K).
2
a
2
a
=
7
ꢀ
7ꢀ
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
3
1
1 -
Figure 3. (upper left) Variation in the P phosphonate shift for [HDy.L ] ; (remainder):
1
1 -
variation of the H NMR shift for the major tert-butyl resonance of [HLn.L ] (Ln = Dy, Tb,
Ho, Er and Tm) as a function of pD (D O, 11.7 T, 295 K: see Table 1 for pK data).
A pH titration in a 50:50 mixture of murine plasma/H O also gave rise to a +1.8 ppm shift in
the tert-butyl resonance, with a pK of 6.80 (±0.08) (11.7 T, 295 K), suggesting that the
2
a
2
a
presence of protein has relatively little effect on the protonation equilibrium.
ꢀ
8ꢀ
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
-68
-70
-72
-74
-76
-78
-80
-82
pD 6.9
pD 7.9
Murine plasma
2
85 290 295 300 305 310 315 320
T/ K
1
-
Figure 4. Variation in the chemical shift of the major tert-butyl resonance of [Dy.L ] as a
2
function of T: pD 6.9 (blue triangles), gradient = 0.414, R = 0.998; pD 7.9 (blue squares),
2
gradient = 0.414, R = 0.998; and in a 50:50 murine plasma: H
2
O mix at pH 7.1 (red
2
triangles), gradient = 0.431, R = 0.999 (11.7 T).
Field dependence of relaxation rate measurements
Earlier, we and others have shown that the occurrence of fast relaxation allows shorter
acquisition times to be used in gradient echo and related pulse sequences, leading to increases
4
-6
in signal intensity per unit time of the order of 15 to 25 times, versus a diamagnetic control.
The longitudinal relaxation rate, R , of each complex was measured at six fields over the range
to 16.5 Tesla. Relaxation rates were fastest at 1T for the erbium and dysprosium complexes,
1
1
and values were broadly similar in the field range 4.7 to 9.4 T (Table 2). Such behaviour
reflects the additive contributions of dipolar and Curie contributions to relaxation, which are
hypothesized to have squared and fourth power dependences on the mean magnetic
susceptibility, in the most commonly used model used to treat paramagnetic relaxation, based
[
19]
on Bloch-Redfield-Wangsness theory, eq 1.
ꢀ
ꢀ
ꢀ
2
2
2
N
2
eff
2
N
4
eff
2
2
⎛ µ ⎞ γ µ ⎡ 7τ
3τ
⎤ 2 ⎛ µ ⎞ ω µ
3τ
0
R+e
R+e
2
0
R
2
N
R1 =
⎢
+
+
⎥
⎜
⎟
6
2
2
2
⎜
⎟
2
6
1
5 ⎝ 4π ⎠
r
1+ω τ
1+ω τ
5 ⎝ 4π ⎠ 3kT r 1+ω τ
⎣
e
R+e
N
R+e
⎦
(
)
R
(1)
−
1
2
2
e
2
B
2
−1
−1
ˆ
µ
= g µ
S
τR+e = τ +T
(
)
eff
R 1e
0 N
where µ is the vacuum permeability, g is the gyromagnetic ratio of the nucleus, gLn is the
3
+
Landé factor of the fundamental multiplet J of the free Ln ion, τ
time in which isotropic tumbling is assumed, µ is the Bohr magneton (BM), r is the electron-
nuclear distance, ω is the nuclear Larmor frequency, ω is the electron Larmor frequency and
1e is the longitudinal relaxation time of the electron spin. In this approximation, at low
r
is the rotational correlation
B
N
e
T
magnetic fields, relaxation is dominated by electronic relaxation (T1e), while at higher fields
the order of relaxation tends to echo the size of the magnetic susceptibility, µeff. Relaxation
ꢀ
9ꢀ
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
rates were found to vary by less than 2 to 3 % as the pH was varied over the studied range
11.7 T, 310K).
Global minimisation methods were applied to the 5 data sets together and assuming classical
(
r
BRW theory (eq. 1), values of µeff , τ and T1e were estimated (Table 3 and Figure 5). The Ln-
t
Bu hydrogen average distance, was assumed to be 6.6 Å, neglecting the minor variation of
ionic radius associated with the lanthanide contraction (0.07 Å from Tb to Tm). The value of
r has been shown earlier to correspond to the average distance estimated using DFT
[
8]
calculations, based on consideration of closely related X-ray structural analyses. A global
minimum was found, with τ = 341 (9) ps, and the best-fit values of T1e and of µeff are given
r
below (Table 3). The highest value of T1e was found for erbium, in accord with its unique
low field relaxation behaviour. Furthermore, values of µeff deviated significantly from the
commonly accepted free-ion values, notably for Er (+15%) and Tm (+37%), (Table 3), which
fitted least well (Figure 5). It has been suggested that such behaviour reflects the inadequacy
1
7
of this theoretical model, and relaxation needs to consider the anisotropy of the magnetic
[
20]
susceptibility tensor explicitly, in a modified approach first suggested by Fiat and Vega.
The fast and similar relaxation rates of the Tm and Dy complexes in the field range 4.7 to
1.7 T, coupled with their large shift and most favourable pH/T dependencies , encouraged
the assessment of the utility of this pair of complexes for multiple imaging studies.
1
1
1 2-
Table 2. H NMR longitudinal relaxation rate data for [Ln.L ] (295 K, D O,
2
a,b
pD 7.4) as a function of magnetic field (Tesla) .
–
1
R (s )
1
Ln
Tb
Dy
Ho
Er
1 T
4.7 T
73(1)
88(3)
89(5)
117(7)
81(4)
9.4 T
11.7 T
134(1)
179(1)
191(2)
198(3)
160(1)
14.1 T
150(1)
195(1)
209(1)
214(7)
179(1)
16.5 T
164(1)
218(1)
252(1)
239(8)
194(1)
42(3)
54(5)
44(4)
92(10)
25(4)
122(2)
156(1)
160(2)
173(3)
139(1)
Tm
a
At 9.4 T and pD 9, the ratios of R /R were: Tb, 0.30; Dy, 0.50; Ho, 0.46; Er, 0.38; Tm,
1
2
0
a
.58; at the pK , the ratio was 0.20, reflecting the impact of chemical exchange (uncertainty)
broadening.
b
Errors are given in parenthesis.
ꢀ
10
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
1
Table 3. Estimated 'best fit' values for H NMR relaxation parameters following global
a,b
1 2-
fitting with r = 6.6 Å (295 K, D O, data in Table 2 ; fits in Figure 5) for [Ln.L ]
2
Ln
T /ps
µ /BM
1
e
eff
Tb
Dy
Ho
Er
0.44(3)
0.46(5)
0.34(5)
0.79(9)
0.30(4)
9.89(2)
10.65(2)
10.99(2)
10.80(4)
10.46(3)
Tm
a
The five data sets were fitted globally, using r = 6.6 Å, where r is the mean distance
t
between the lanthanide ion and the Bu hydrogen atoms, data minimised to give τ
r
= 345(7)
b
ps; classical values for the magnetic moments of lanthanide(III) ions are: Tb, 9.8; Dy, 10.3;
Ho, 10.4; Er, 9.4; Tm, 7.6 BM.
Tb
Dy
Ho
Er
Tm
1
-1
t
1 2-
Figure 5. H NMR relaxation rates (R
of magnetic field, showing the global fits (line) to the experimental data points (295K D
pD 7.4, fixed r = 6.6 Å, minimizes to τ = 345(9) ps)
1
/s ) for the Bu resonance in [Ln.L ] as a function
2
O,
r
Phantom and in vivo imaging and shift imaging studies
Phantom imaging studies at 300 MHz (7 T) were undertaken to assess the feasibility of
imaging at these very well separated spectral frequencies. Equimolar solutions of both
1
-
1 -
[
Dy.HL ] and [Tm.HL ] (ESI) were prepared in buffered aqueous solution (0.1 M MOPS), at
a range of pH values, inside cut-down 5 mm NMR tubes. The NMR tubes were placed within
a Teflon spacer, (Figure 6) with an NMR tube containing only de-ionised water, occupying
the central position. The experimental array was placed inside the bore of a 30 mm birdcage
coil, typically used for murine imaging. First, an image corresponding to the water protons
ꢀ
11
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
was obtained by imaging with a 3-dimensional gradient echo (3D-GE) MR sequence at the
H
2
O offset. Second, 3DGE images of the tert-butyl reporter resonances were acquired by
t
setting the excitation offset to the resonant frequency of the Bu resonance in each lanthanide
complex at pH 7. The combination of the large offset frequency with selection of a narrow
(
20KHz) spectral bandwidth allowed, the tert-butyl reporter resonance to be acquired,
separately from the water signal.
ꢀ
Figure 6. (Left) Illustration of experimental array, in a Teflon holder, using 4 equimolar
1
-
solutions of buffered [Dy.HL ] at different pH values shown and with de-ionised water in the
centre (MOPS, 1.2 mM, H O). (Middle) Phantom image at the H O offset (intensity scale 0-
000 a.u.). (Right) Phantom image at the [Dy.HL ] tert-butyl offset (-23474 Hz, intensity
scale 0-0.1 a.u.), both: TR/TE=17.2/0.7 ms; averages 4 (H O), 30 (tert-butyl); spectral width,
2
2
1
-
4
2
SW = 8103 Hz; spectral points = 128; matrix 16 x 16; FOV 40 x 40 mm; slice thickness = 4
mm; T = 310 K).
ꢀ
1
-
Figure 7. Spectral analysis from the spectroscopic imaging study of [Dy.HL ] , showing the
frequency detected for each ROI peak for: pH 6.2 (red), 3543 Hz; 6.6 (green), 3644 Hz; 7.1
(
blue), 3773 Hz; and 7.5 (yellow), 3861 Hz. Total shift = 318 Hz = 1.06 ppm; offset
frequency = -23474 Hz. TR/TE = 17.2/0.7 ms, 30 averages, slice thickness = 60 mm, 128
spectral points, SW = 8013 Hz, T = 310K.
Spectroscopic images were then collected using a 2DSI sequence. This method selected a
single imaging slice through the phantom and encoded the spatial location of the signal to
ꢀ
12
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
provide a complete NMR spectrum at each point in the image matrix, allowing spectral
analysis of resonance frequency as a function of pH for each tube of the experimental
phantom (Figure 7). With a total shift of 318 Hz at 7 T (1.06 ppm) between pH 6.2 and 7.5,
the data obtained was in good agreement with results found in vitro, (Table 1). The pH
-
1 -
dependent shift for [Dy.HL] was greater over this range than for [Tm.HL ] (1.06 ppm vs.
.48 ppm). The difference in the size of the shift change is somewhat surprising, but is very
0
useful, as it distinguishes each from the accompanying temperature dependent behaviour,
where the large positive (Tm) or negative (Dy) paramagnetic shift diminishes as temperature
increases.
1
-
The temperature of five solutions of [Tm.HL ] used in the phantom studies was varied, using
an externally controlled heated airflow around the Teflon spacer array, and was monitored by
positioning a MRI compatible thermometer in the middle of the array. Spectroscopic images
were acquired 5 minutes after setting the temperature, in order to minimise any chemical shift
fluctuations, (Figure 8).
7
7
7
7
7
6
6
6
8
6
4
2
0
8
6
4
pH 8.1
pH 7.5
pH 7.1
pH 7 - 11.7 T
pH 6.6
pH 6.2
2
88
298
308
318
T/ K
ꢀ
Figure 8. Variation in the chemical shift of the major tert-butyl resonance of five different
1
pH solutions of [Tm.HL ] as a function of temperature (7 T): pH 6.2 (orange), gradient = -
0
.44; pH 6.6 (blue), gradient = -0.46; pH 7 (black), gradient = -0.44; pH 7.1 (purple), gradient
-0.44; pH 7.5 (green), gradient = 0.46; and pH 8.1 (red), gradient = -0.46. Each line of best-
=
2
fit has R = 0.998.
The tert-butyl chemical shift of each of the five solutions measured from the spectroscopic
imaging data revealed a linear temperature dependence over the range examined, with the
variation in agreement with data obtained in NMR spectroscopy experiments at 11.7 T at pH
7
0
. In each case, the tert-butyl group gave rise to a chemical shift temperature dependence of -
1
-
.45±0.01 ppm/K. With [Dy.HL ] the temperature dependence was found to be +0.41±0.01
ppm/K.
ꢀ
13
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
Preliminary Triple Imaging Phantom and In Vivo Studies
Phantom triple imaging studies were next undertaken using a solution containing equimolar
1
concentrations of both the Dy and Tm complexes (1.2mM each [Ln.HL ], 0.1 M NaCl,
3
10K). A pulse sequence was developed to enable triple imaging, which allows simultaneous
signal acquisition from all 3 signal sources - the normal tissue water signal and each of the
well-separated t-butyl signals. This pulse sequence used a dual, back-to-back RF pulse to
consecutively excite the Dy and Tm complexes. The frequency, shape, duration and power of
both excitation pulses could be separately modified. Fortunately, the relaxation times (T
) of each complex have the same values, within experimental error, at 7 Tesla (T and T
.4 and 7.8 ms) and so the same shape, duration and power could be used for each complex,
1
and
T
2
2
1
=
3
and it was only necessary to change the offset frequency. The encoding gradients were then
applied, and the spectral width (SW) was increased to ~60 kHz, in order to encompass signal
from all three frequency ranges. No explicit excitation pulse was applied to the water
resonance frequency, as there was sufficient excitation within the bandwidth of the main RF
pulses. The dual RF pulse train was applied to both a spectral imaging pulse sequence and a
3
D-gradient echo pulse sequence (Figure 9).
(
a) 2DSI
Figure 9. Pulse sequence diagrams for: (a) a spectral imaging (2DSI) pulse sequence and,
b) a 3D-gradient echo pulse sequence where f1 and f2 are ‘back to back’, Gaussian shaped
excitation pulses applied at the respective transmit frequencies of the tert-butyl resonances
(b) GE3D
(
(
frequencies f1 and f2), each with a duration of 500 μs.
ꢀ
14
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
Dy
Tm
Figure 10. Left: a spectral imaging sequence applied to a phantom set-up, with a TR of 19.5
ms, TE of 0.95 ms TE of 0.45 ms and a spectral width SW of 54 kHz, 32 averages, scan
time = 100 s. Right: a 3D-gradient echo pulse sequence with a TR of 4.28 ms, TE (Dy) of
.87 ms, TE (Tm) of 1.37 ms, SW = 60 kHz, 80 averages, scan time = 354 s. The phantom
1
2
1
0
2
1
1
contained NMR tubes containing 1.2 mM [Tm.HL ], [Dy.HL ] or water and a small vial
containing equimolar mix of the two complexes.
Figure 10 illustrates the in vitro 3DGE and 2DSI results. Using the 3DGE approach with a
triple imaging bandwidth (and triple number of sample points) relative to a normal MRI
acquisition allows sampling across the full MRI frequency range arising from the water, Tm
and Dy agents. The right panel illustrates the full triple imaging field of view which can be
split into 3 entirely independent MRI images of the Dy, water and Tm. The lower vial of the
phantom contains aqueous solutions of a mixture of the Dy and Tm complexes appears in
each of the channels. The central small tube containing only water appears in the central
water image and the other tubes containing only Dy or only Tm appear in their respective
channels. This approach does not degrade the signal to noise ratio of the scan relative to
acquiring individual images of each component with separate excitations, as the noise
bandwidth per pixel is unchanged. The MRSI acquisition generates a matrix of voxels, each
of which can be interrogated to show the full spectral range. Individual peaks from water, and
the Tm and Dy complexes are clearly observed as expected in the example voxel from the
mixed solutions in the lower vial.
These sequences were applied in vivo following tail-vein injection of CD1 healthy mice (n=3)
with 0.05 mmoles/kg of a 1:1 mixture of the Tm/Dy complexes in saline. Illustrative data
from one mouse are shown (Figure 11), where data were collected with the MRSI sequence to
follow the kinetics of the complex through the tissue. Selected spectra from the MRSI data
ꢀ
15
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
grid were extracted to illustrate the large shifts seen in the high and low frequency regions of
the spectrum (Figure 11, stacked spectra), and clearly demonstrate changes in frequency
between tissues, most markedly in the bladder, with the shifts attributable to differences in
tissue pH and temperature. A 3DGE acquisition was also acquired (Figure 11, right). The
central region of the triple image shows the standard water (and fat) MRI, but at the time this
image was taken, 40 minutes post injection, renal clearance of the contrast agent resulted in a
strong signal from the bladder only.
×
Phantom
Liver
Liver
ꢀ
Kidney
×1
Bladder
-21200
-20200
Hz
23500
24500
Hz
Figure 11. In vivo examples of 3DGE and MRSI acquisition in an intact, healthy mouse
following single i.v. injection of a mix of Dy and Tm PARASHIFT contrast agent. Left :
Selected spectra from a time series of 2DSI acquisitions. The stacked spectra ,where the y
axis denotes signal intensity in arbitrary units, show the Tm (left) and Dy (right) spectral
regions from voxels sampling liver (two time points), kidney, bladder and an external
reference phantom. MRSI acquisition parameters were TR 19.5 ms, TE
Tm) of 0.95 ms, SW 54 kHz, 32 averages, scan time 100 s. Right : a 3DGE triple image
using a pulse sequence with TR of 10.66 ms, TE of 1.58 ms, TE of 2.08 ms, SW = 54 kHz,
00 averages, scan time = 1092 s.
1 2
(Dy) 0.45 ms, TE
(
1
2
1
The frequency shift of each resonance was measured in these spectra by peak picking and the
pH and temperature determined by solving the simultaneous equations defining the frequency
versus pH and T dependence for each complex (see ESI). Liver pH was measured to be 7.14,
kidney pH was 6.92 and bladder pH was 6.8, consistent with the expected range of values for
normal wild type animals. Temperature was also significantly lower in the bladder (3.5 C
below core body temperature) and this temperature difference is the cause of the large
frequency shift in the bladder resonances, relative to the other tissues. With the aid of zero-
filling of the raw spectra, (see SI for a discussion) the accuracy of spectral frequency
measurement for in vivo 2DSI acquisitions was assessed as being 13 Hz, which is equivalent
to a temperature accuracy of ±0.1C.
ꢀ
16
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
The accuracy of pH measurement is non-linear given the sigmoidal relationship between
frequency and pH, being greatest (±0.1 pH units) around pH 7.2 (see ESI). The pH accuracy
of our measurements are similar to recent attempts to measure tissue pH using indirect
detection of contrast agent (iopamidol) presence using chemical exchange saturation transfer
[
11d]
(
CEST) methods
, which also suggests an accuracy of ±0.1 pH units, but which is
dependent on concentration. Those CEST methods have been attempted in human subjects to
[
11f]
measure bladder pH and in animal models and human subjects to measure tumour pH
those studies the authors note the sensitivity of their methods to experimental parameters
such as the B field of the CEST saturation) and choice of modelling approaches, which are
In
(
1
consequences of the indirect detection use for CEST imaging, but which are absent from the
PARASHIFT method using the frequency shift of the directly detected complex. Further
challenges were noted due to rapid clearance of the complex which is a common problem for
all molecular MRI tracer experiments.
Other studies using similar paramagnetically shifted resonance detection to measure tissue pH
in rats have demonstrated elegant pH maps with good spatial resolution through the use of
renal ligation or pharmacological manipulation of renal excretion combined with long, steady
[
15]
infusion rates of several hours . In this study, we have constrained our in vivo work to use a
single i.v. bolus of complexes at a clinically equivalent dose level and to follow this in intact
mice with no modification of renal excretion rates. The data demonstrate that this approach is
feasible with these complexes, but the approach necessarily limits the achievable resolution
due to the short scan durations used.
Summary and Conclusions
1
The Dy and Tm complexes of L have several key properties that favour their simultaneous
observation, together with the water signal, in triple imaging experiments at 7 T. They exist
in aqueous solution as one major stereoisomer; thus, one main signal is observed for the t-
1
butyl H NMR reporter group that resonates over 70 ppm from the water signal. Furthermore,
1 2
they possess R and R values that are within 10% of each other, so that Ernst angle pulsing
conditions which achieve optimum imaging sensitivity are achieved using the same shape,
duration and power of each excitation pulse in gradient echo 3D Imaging and 2D spectral
imaging experiments; only the offset frequency needed to be changed. Under the conditions
used, the intense water signal was co-observed without the need for separate excitation. The
shifted t-butyl signals exhibit a large dependence on temperature, +0.41 and -0.44 ppm for the
Dy and Tm complexes respectively. Protonation of the aryl phosphonate group leads to a
1
2-
1 2-
shift of +1.9 ppm, [Dy.L ] and -1.1 ppm for [Tm.L ] ; even at a pH corresponding to the
pK value, uncertainty broadening due to chemical exchange means that the R /R ratio was
a
1
2
not less than 0.2, and so did not inhibit spectral or image acquisition unduly.
ꢀ
17
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
The dual temperature and pH sensitivity of the reporter chemical shift means that both
complexes need to be present in the region of interest examined, in order to observe two shift
values, the simultaneous equation of which can be solved for T and pH. In this manner, using
appropriate phantoms to provide the calibration values, the pH and temperature of the liver
was assessed to be 7.1 and 34.5°C, whilst the values for the bladder were 6.8 and 31°C.
These approaches allow simultaneous spectroscopic imaging of pH and T, and with
appropriate post-processing can be used to generate pH and T maps, for the liver and kidneys
in particular, that may be of benefit to radiologists seeking to identify areas of inflammation
and fibrosis where elevated temperature, for example, may occur.
Experimental
Detailsꢀ ofꢀ NMRꢀ dataꢀ acquisition,ꢀ relaxationꢀ measurementsꢀ andꢀ analysis,ꢀ andꢀ theꢀ
estimationꢀ ofꢀ Tꢀ andꢀ pHꢀ inꢀ PARASHIFTꢀ experimentsꢀ areꢀ givenꢀ inꢀ theꢀ supplementaryꢀ
information.ꢀꢀ
Ligand and complex synthesis
5
-tert-Butyl-2-methyl-4-nitropyridin-1-oxide
5
-tert-Butyl-2-methylpyridine 1-oxide (3.20 g, 19.4 mmol) was dissolved in conc.
o
H
2
SO
4
(98%, 11.0 mL, 209.5 mmol) and stirred for 10 min at 5 C. To this solution
3
was added HNO (70%, 7.0 mL, 161.0 mmol) dropwise and the resulting mixture was
o
heated to 100 C for 18 h. After this time, the solution was dropped onto ice and
stirred for 1 h before the aqueous mixture was extracted with CH
The organic layers were combined, dried over MgSO and the solvent removed under
reduced pressure. The resulting pale green oil was purified by silica gel column
2 2
Cl (3 × 150 mL).
4
chromatography, eluting with a gradient starting from 100% CH
2
Cl
2
to 2% MeOH/
1
CH
2
Cl
2
, to yield a bright yellow oil (3.0 g, 74%). R
f
(10% MeOH/ CH
2
Cl
2
) = 0.47; H
6
3
NMR (400 MHz, CDCl
3
) δ 8.76 (s, 1H, H ), 7.58 (s, 1H, H ), 2.64 (s, 3H, CH
3
), 1.35
1
3
2
5
6
(s, 9H, C(CH
3
)
3
); C NMR (101 MHz, CDCl
3
) δ 152.5 (C ), 151.1 (C ), 142.3 (C ),
4
3
1
39.3 (C ), 122.6 (C ), 35.7 (C(CH
3
)
3
), 29.9 (C(CH
3
)
3
), 17.5 (CH
3
); ESI-LRMS (+)
ꢀ
18
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
+
+
m/z 211.1 [M+H] ; ESI-HRMS (+) calcd for [C10
H
15
N
2
O
3
] 211.1083, found
2
11.1088.
4
-Amino-5-tert-butyl-2-methylpyridin-1-oxide
5
-tert-Butyl-2-methyl-4-nitropyridin-1-oxide (1.9 g, 9.04 mmol) was dissolved in
MeOH (50 mL) and Pd/C was added (Pd content 10 %, 190 mg). The vessel was then
loaded onto a Parr hydrogenator (pressure 40 bar H ) and the reaction mixture was
agitated for 18 h. After this time, the mixture was filtered before the solvent was
evaporated under reduced pressure to yield an orange oil (1.6 g, 98%). R (10%
OD) δ 7.95 (s, 1H, H ), 6.68 (s, 1H,
2
f
1
6
MeOH/ CH
2
Cl
2
) = 0.28; H NMR (400 MHz, CD
3
3
13
H ), 3.32 (s, 3H, CH
3
), 1.37 (s, 9H, C(CH
3
)
3
); C NMR (101 MHz, CD
3
OD) δ 152.2
),
2
5
6
4
3
(
C ), 148.0 (C ), 138.1 (C ), 129.4 (C ), 112.2 (C ), 34.1 (C(CH
3
)
3
), 26.8 (C(CH
3
)
3
+
1
6.9 (CH
O] 181.1341, found 181.1350.
-tert-Butyl-4-iodo-2-methylpyridin-1-oxide
3
); ESI-LRMS (+) m/z 181.1 [M+H] ; ESI-HRMS (+) calcd for
+
10 17 2
[C H N
5
To a cold solution of 4-amino-5-tert-butyl-2-methylpyridin-1-oxide (1.6 g, 8.88
mmol) and HBF (2.3 mL, 17.8 mmol) in H O (40 mL) was added dropwise, a
solution of NaNO (1.2 g, 17.8 mmol) in H O (5 mL). Potassium iodide (2.0 g, 12.4
4
2
2
2
mmol) was slowly added to the suspension, in several portions. The reaction mixture
o
was allowed to stir at rt for 5 min before being heated at 60 C for 30 min. The
mixture was cooled to rt and the pH adjusted to 10 with KOH. The solution was
extracted with CH
mL) before being dried over MgSO
The resulting brown oil was purified by silica gel column chromatography, eluting
with a gradient starting from 100% CH Cl to 2% MeOH/ CH Cl to yield an orange
solid (0.8 g, 31%). R (10 % MeOH/ CH Cl
MHz, CDCl
2
Cl
2
(3 × 100 mL) and washed with NaHCO
3
(sat. aq. solution, 100
4
and the solvent removed under reduced pressure.
2
2
2
2
o
1
f
2
2
) = 0.37; m.p. 153-155 C; H NMR (400
6
3
3
) δ 8.23 (s, 1H, H ), 7.80 (s, 1H, H ), 2.41 (s, 3H, CH
3
), 1.48 (s, 9H,
1
3
2
5
6
C(CH
3
)
3
); C NMR (101 MHz, CDCl
3
) δ 152.5 (C ), 146.5 (C ), 139.8 (C ), 138.4
4
3
(
C ), 122.2 (C ), 35.9 (C(CH
3
)
3
), 29.1 (C(CH
3
)
3
), 16.7 (CH
3
); ESI-LRMS (+) m/z
15 INO] 292.0198, found 292.0208;
15INO: C, 41.26; H, 4.85; N, 4.81, found: C, 41.11; H, 4.78;
+
127
+
2
91.9 [M+H] ; ESI-HRMS (+) calcd for [C10H
Anal Calcd (%) for C10
N, 4.79.
H
(5-tert-Butyl-4-iodopyridin-2-yl)methyl acetate
ꢀ
19
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
5
-tert-Butyl-4-iodo-2-methylpyridin-1-oxide (0.8 g, 2.75 mmol) was dissolved in
acetic anhydride (10 mL) and the solution was heated to 75 °C for 3 h. The solvent
was removed under reduced pressure and the resulting orange oil was purified by
silica gel column chromatography, eluting with a gradient starting from 100% CH
2
Cl
2
to 2% MeOH/ CH
2
Cl
2
to yield an orange oil (0.74 g, 81%). R
f
(10 % MeOH/ CH
2
Cl
2
)
1
6
3
=
0.62; H NMR (400 MHz, CDCl
3
) δ 8.53 (s, 1H, H ), 7.94 (s, 1H, H ), 5.11 (s, 2H,
1
3
CH
2
), 2.16 (s, 3H, CH
3
), 1.54 (s, 9H, C(CH
3
)
3
); C NMR (101 MHz, CDCl
3
) δ 170.7
), 35.9
2
5
6
4
3
(
CO), 158.5 (C ), 147.6 (C ), 145.0 (C ), 136.3 (C ), 107.0 (C ), 65.5 (CH
2
+
(
C(CH
3
)
3
), 29.6 (C(CH
3
)
3
), 21.0 (CH
3
); ESI-LRMS (+) m/z 333.8 [M+H] ; ESI-
] 334.0304, found 334.0307.
5-tert-Butyl-4-(diethoxyphosphoryl)pyridin-2-yl]methyl acetate
1
27
+
HRMS (+) calcd for [C12H17 INO
2
[
O
P
O
O
O
N
O
(5-tert-Butyl-4-iodopyridin-2-yl)methyl acetate (425 mg, 1.28 mmol) was dissolved
2
in triethylphosphite (10 mL, 58.3 mmol). To this solution was added Pd(dba) (73 mg,
o
1
0 mol%, 0.128 mmol) and the resulting mixture was heated at 120 C for 18 h. The
solvent was removed by vacuum distillation and the resulting brown oil was purified
by silica gel column chromatography, eluting with a gradient starting from 100%
CH
2
Cl
2
2
to 1% MeOH/ CH
2
Cl
2
, to afford a yellow oil (284 mg, 65%). R
f
(10% MeOH/
1
4
6
CH
2
Cl
) = 0.56; H NMR (400 MHz, CDCl
3
) δ 8.84 (d, JHP = 8, 1H, H ), 7.80 (d,
3
3
3
J
HP = 16, 1H, H ), 5.21 (s, 2H, CH
2
), 4.05-4.14 (m, J = 7HP, 4H, POCH
2
), 2.16 (s,
4
31
3
H, pyCH
3
), 1.58 (s, 9H, C(CH
3
)
3
), 1.31-1.37 (m, J = 7HP, 6H, POCH
2
CH
3
); P
3
13
NMR (162 MHz, CDCl
3
) δ 16.1-16.5 (m, JPH = 7); C NMR (101 MHz, CDCl
3
) δ
3
6
3
2
2
5
1
1
70.7 (CO), 153.1 (d, JCP = 12, C ), 149.2 (d, JCP = 13, C ), 147.7 (d, JCP = 11, C ),
1
4
2
3
2
2
36.8 (d, JCP = 179, C ), 126.5 (d, JCP = 7, C ), 66.4 (CH O), 63.8 (d, JPH = 6,
3
POCH
ESI-LRMS (+) m/z 344.2 [M+H] ; ESI-HRMS (+) calcd for [C16
44.1627, found 344.1628.
2
), 36.0 (C(CH
3
)
3
), 31.6 (C(CH
3
)), 21.0 (CH
3
), 16.3 (d, JPH = 6, POCH
2
CH
3
);
+
+
H27NO
5
P]
3
Diethyl [5-tert-butyl-2-(hydroxymethyl)pyridin-4-yl]phosphonate
ꢀ
20
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
[5-tert-Butyl-4-(diethoxyphosphoryl)pyridin-2-yl]methyl acetate (300 mg, 0.87
mmol) was dissolved in anhydrous EtOH (4 mL) and purged with argon for 10 min,
before sodium metal (~15 mg) was added. The resulting solution was stirred at rt for 1
h, under an atmosphere of argon. The solvent was removed under reduced pressure
and the residue dissolved in CH
aqueous layer was re-extracted with CH
combined, dried over MgSO and the solvent removed under reduced pressure. The
resulting brown oil was purified by silica gel column chromatography, eluting with a
2
Cl
2
(7 mL) and washed with H
2
O (2 mL), before the
2
Cl (3 × 7 mL). The organic layers were
2
4
gradient starting from 100% CH
2
Cl
2
2
2 2
to 2% MeOH/ CH Cl , to yield a yellow oil (150
1
mg, 57%). R
f
(10% MeOH/ CH
2
Cl
) = 0.48; H NMR (400 MHz, CDCl
3
) δ 8.80 (d,
4
6
3
3
3
J
HP = 8, 1H, H ), 7.79 (d, JHP = 15, 1H, H ), 4.80 (s, 2H, CH
2
), 4.07-4.14 (m, J =
4
7
HP, 4H, POCH
2
), 2.58 (br s, OH), 1.59 (s, 9H, C(CH
3
)
3
), 1.35 (t, J = 7HP, 6H,
3
1
3
13
2 3 3
POCH CH ); P NMR (162 MHz, CDCl ) δ 15.8-16.2 (m, JPH = 7); C NMR (176
3
6
3
2
2
3
MHz, CDCl ) δ 157.3 (d, JCP = 11, C ), 148.1 (d, JCP = 13, C ), 146.7 (d, JCP = 11,
5
1
4
2
3
2
C ), 136.5 (d, JCP = 179, C ), 125.3 (d, JCP = 7, C ), 64.2 (CH
2
O), 62.7 (d, JPH = 6,
3
3
POCH
2
), 35.8 (d, JCP = 3, C(CH
3
)
3
), 31.6 (C(CH
3
)), 16.3 (d, JPH = 6, POCH
2
CH
3
);
+
+
ESI-LRMS (+) m/z 302.2 [M+H] ; ESI-HRMS (+) calcd for [C14
H25NO
4
P]
3
02.1521, found 302.1519.
[5-tert-Butyl-4-(diethoxyphosphoryl)pyridin-2-yl]methyl methane sulfonate
O
O
O
P
O
S
O
N
O
Diethyl [5-tert-butyl-2-(hydroxymethyl)pyridin-4-yl]phosphonate (150 mg, 0.50
o
mmol) was dissolved in anhydrous THF (4 mL) and cooled to 5 C. Triethylamine
(140 µL, 1.00 mmol) and methanesulphonyl chloride (58 µL, 0.75 mmol) were added
dropwise and the resulting solution was stirred at rt for 30 min. The solvent was
removed under reduced pressure before the residue was treated with brine (5 mL) and
2 2
extracted with CH Cl (3 × 10 mL). The organic layers were combined, dried over
MgSO
4
, and the solvent removed under reduced pressure to yield a pale orange oil,
1
which was used immediately (176 mg, 93%). R
f
(10% MeOH/ CH
2
Cl
2
) = 0.56; H
ꢀ
21
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
4
6
3
3
NMR (400 MHz, CDCl
3
) δ 8.78 (d, JHP = 8, 1H, H ), 7.84 (d, JHP = 15, 1H, H ),
3
5
9
.26 (s, 2H, CH
2
), 4.06-4.16 (m, JHP = 6, 4H, POCH
2
3
), 3.05 (s, 3H, CH ), 1.52 (s,
3
4
31
H, C(CH
3
)
3
), 1.29 (dt, JHP = 6, JHP = 1, 6H, POCH
2
CH
3
); P NMR (162 MHz,
3
+
CDCl
3
) δ 15.4-15.8 (m, JPH = 6); ESI-LRMS (+) m/z 380.5 [M+H] ; ESI-HRMS (+)
+
calcd for [C15
H27NO
6
PS] 380.1297, found 380.1302.
1
,4,7-Tri-benzyl 1,4,7,10-tetraazacyclododecane 1,4,7-tricarboxylate
1
2 2
,4,7,10-Tetraazacyclododecane (2.5 g, 14.5 mmol) was dissolved in dry CH Cl (80
mL) and to this was added triethylamine (6.3 mL, 45.0 mmol). The resulting solution
o
was stirred at 5 C for 10 min, before the dropwise addition of a solution of benzyl
2 2
chloroformate (6.4 mL, 45.0 mmol) in dry CH Cl (20 mL). After addition, the
solution was stirred at rt for a further 18 h. The solvent was removed under reduced
pressure and the resulting pale yellow oil was purified by silica gel column
chromatography, eluting with a gradient starting from 100% CH
CH Cl to yield a colourless oil (6.1 g, 73%). R (10% MeOH/ CH
NMR (400 MHz, CDCl ) δ 7.22-7.33 (m, 15H, ArH), 5.00-5.06 (m, 6H, CH
.63-3.67 (m, 2H, cyclen CH ), 3.16-3.55 (m, 12H, cyclen CH ), 2.75-2.92 (m, 2H,
); C NMR (101 MHz, CDCl ) δ 156.1 (CO), 136.4 (Ar), 136.3 (Ar),
O), 66.8
2
Cl
2
to 2% MeOH/
1
2
2
f
2
Cl
2
) = 0.54; H
3
2
O),
3
2
2
1
3
cyclen CH
2
3
1
28.4 (Ar), 128.3 (Ar), 127.7 (Ar), 127.6 (Ar), 127.4 (Ar), 67.2 (CH
2
(
CH
2
O), 53.5 (cyclen CH
2
), 50.6 (cyclen CH
2
), 50.3 (cyclen CH
2
2
), 50.2 (cyclen CH ),
+
5
0.1 (cyclen CH
2
), 45.6 (cyclen CH
2
); ESI-LRMS (+) m/z 575.1 [M+H] ; ESI-HRMS
+
(+) calcd for [C32
H
39
N
4
O
6
] 575.2870, found 575.2867.
1
,4,7-Tri-benzyl
10-tert-butyl
1,4,7,10-tetraazacyclododecane
1,4,7,10-
tetracarboxylate
1
,4,7-Tri-benzyl 1,4,7,10-tetraazacyclododecane 1,4,7-tricarboxylate (1.3 g, 2.26
mmol) was dissolved in dry CH Cl (15 mL) and triethylamine added (0.4 mL, 2.94
2
2
mmol). The resulting solution was stirred at rt for 10 min, before adding di-tert-butyl
dicarbonate (0.64 g, 2.94 mmol). The resulting solution was stirred at rt for 18 h
before the solvent was removed under reduced pressure. The resulting yellow oil was
purified by silica gel column chromatography, eluting with a gradient starting from
1
00% CH
2
Cl
2
to 2% MeOH/ CH
2
Cl
2
to yield a colourless oil (1.3 g, 85%). R
f
(10%
ꢀ
22
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
1
MeOH/ CH
2
Cl
2
) = 0.69; H NMR (400 MHz, CDCl
3
) δ 7.21-7.34 (m, 15H, ArH),
5
9
1
.06 (s, 2H, CH
2
O), 5.03 (s, 4H, CH
2
O), 3.00-3.54 (br m, 16H, cyclen CH
); C NMR (101 MHz, CDCl ) δ 156.8 (CO), 136.4 (Ar), 128.6 (Ar),
28.4 (Ar), 128.3 (Ar), 128.2 (Ar), 82.8 (C(CH
2
), 1.36 (s,
1
3
H, C(CH
3
)
3
3
3
)
3
), 67.3 (CH
2
O), 49.9-50.7 (br,
+
cyclen CH
2
), 28.4 (C(CH
3
)
3
); ESI-LRMS (+) m/z 675.1 [M+H] ; ESI-HRMS (+)
+
calcd for [C37
H
47
N
4
O
8
] 675.3394, found 675.3406.
tert-Butyl-1,4,7,10-tetraazacyclododecane carboxylate
1
,4,7-Tri-benzyl
tetracarboxylate (1.3 g, 1.93 mmol) was dissolved in MeOH (25 mL) and to this was
added Pd(OH) /C (~200 mg). The reaction vessel was loaded onto a Parr
hydrogenator, where it was subject to agitation for 18 h, under a pressure of 40 bar
. The catalyst was filtered through Celite and the solvent was removed under
reduced pressure. The resulting solid was washed with a minimum amount of NaOH
aq. solution, 1 mL), and extracted into CH Cl (3 × 30 mL), before the organic
10-tert-butyl
1,4,7,10-tetraazacyclododecane
1,4,7,10-
2
H
2
(
2
2
layers were combined, dried over MgSO
4
, and the solvent removed under reduced
1
pressure to yield a pale yellow oil (0.52 g, 99%). H NMR (400 MHz, CD
3
OD) δ
), 3.10-
OD) δ
5
3
1
.03 (br s, NH), 3.52-3.60 (m, 4H, cyclen CH
.21 (m, 8H, cyclen CH ), 1.49 (s, 9H, C(CH
57.5 (CO), 82.3 (C(CH ), 48.2 (cyclen CH
2 2
), 3.22-3.27 (m, 4H, cyclen CH
1
3
2
3 3 3
) ); C NMR (101 MHz, CD
3
)
3
2
), 46.9 (cyclen CH
2
), 45.6 (cyclen
+
CH
2
), 44.5 (cyclen CH
2
), 28.7 (C(CH
3
)
3
); ESI-LRMS (+) m/z 273.2 [M+H] ; ESI-
+
HRMS (+) calcd for [C13
H
29
N
4
O
2
] 273.2291, found 273.2296.
tert-Butyl-4,7,10-tri{[ethoxy(methyl)phosphoryl]methyl}-1,4,7,10-
tetraazacyclododecane-1-carboxylate
O
O
P
O
O
N
N
N
O
N
P
O
P
O
O
ꢀ
23
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
tert-Butyl-1,4,7,10-tetraazacyclododecane carboxylate (0.52 g, 1.91 mmol) and
paraformaldehyde (1.15 g, 38.2 mmol) were dissolved in anhydrous THF (30 mL)
o
and the resulting suspension was heated at 80 C for 3 h. Diethyl methylphosphonite
o
(
1.7 mL, 11.5 mmol) was added and the solution stirred at 80 C for a further 18 h.
The solution was cooled and filtered, before the solvent was removed under reduced
pressure. The residual brown oil was purified by alumina gel column
chromatography, eluting with a gradient starting from 100% CH
2
Cl
2
to 2% MeOH/
1
CH
2
Cl
2
to yield a pale yellow oil (0.4 g, 33%). R
f
(10% MeOH/ CH
2
Cl
2
) = 0.76; H
3
NMR (400 MHz, CDCl
PCH
C(CH
3
) δ 3.93 (q, JPH = 7, 6H, POCH
2
), 3.09-3.60 (br m, 6H,
2
N), 2.41-3.00 (br m, 16H, cyclen CH
2
), 1.33-1.47 (m, 9H, PCH
3
), 1.32 (s, 9H,
) δ 52.53,
3
31
3
)
3
), 1.18 (t, JPH = 7, 9H, POCH
2
CH
3
); P NMR (162 MHz, CDCl
3
5
2.31, 52.22, 52.20, 52.18, 52.24, 52.06, 51.95, 51.85, 51.85, 51.59, 51.49 (mixture
1
3
of isomers); C NMR (176 MHz, CDCl
3
) δ 155.5 (CO), 79.5 (C(CH
3
)
3
), 59.9-60.1
N), 45.5
(m, POCH
2
), 52.7-55.1 (m, cyclen CH ), 45.9 (PCH N), 45.6 (PCH
2
2
2
3
(
PCH N), 28.4 (C(CH
2
3
)
3
), 16.6 (t, JCP = 6, POCH
2
CH
3
3
), 13.5-13.9 (m, PCH ); ESI-
+
+
LRMS (+) m/z 633.2 [M+H] ; ESI-HRMS (+) calcd for [C25
H
56
N
4
O
8
P
3
] 633.3311,
found 633.3312.
Ethyl[(4,7-di{[ethoxy(methyl)phosphoryl]methyl}-1,4,7,10-
tetraazacyclododecane-1-yl)methyl](methyl)phosphinate
tert-Butyl-4,7,10-tri{[ethoxy(methyl)phosphoryl]methyl}-1,4,7,10-
tetraazacyclododecane-1-carboxylate (0.4 g, 0.63 mmol) was dissolved in anhydrous
CH Cl (2 mL) and TFA added (2 mL). The resulting solution was stirred at rt for 18
h, before the solvent was removed under reduced pressure. The resulting yellow oil
was dissolved in a minimal amount of H O (3 mL). The pH of the water layer was
adjusted to 12 using solid NaOH before being extracted with CH Cl (4 × 10 mL).
2
2
2
2
2
The CH
2
Cl
2
layers were combined, dried over MgSO
4
, and the solvent removed
1
under reduced pressure to yield a pale yellow oil (0.22 g, 66%). H NMR (400 MHz,
CDCl
3
) δ 3.90-4.08 (m, 6H, POCH
2
), 2.60-3.32 (br m, 22H, cyclen CH
2
, PCH
2
N),
2
3
31
1
.46 (d, JPH = 13, 9H, PCH
3
), 1.25 (t, JPH = 7, 9H, POCH
2 3
CH ); P NMR (162
1
3
MHz, CDCl
3
3
) δ 54.2, 53.8, 53.5, 53.2, 50.4 (mixture of isomers); C NMR (101
MHz, CDCl
) δ 60.1-60.6 (m, POCH
2
), 56.2-56.6 (m, cyclen CH
2
), 55.1-55.5 (m,
3
cyclen CH
2
), 52.5-53.2 (m, cyclen CH
2
), 46.8-47.3 (m, PCH
2
N), 16.7 (t, JCP = 6,
ꢀ
24
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
1
1
POCH
2
CH
3
), 14.2 (t, JCP = 12, PCH
3
3
), 13.3 (t, JCP = 12, PCH ); ESI-LRMS (+) m/z
+
+
5
33.1 [M+H] ; ESI-HRMS (+) calcd for [C20
H
48
N
4
O
6
P
3
] 533.2787, found 533.2811.
Diethyl {5-tert-butyl-2-[(4,7,10-tri{[ethoxy(methyl)phosphoryl]methyl}-1,4,7,10-
tetraazacyclododecan-1-yl)methyl]pyridin-4-yl}phosphonate
O
O
O
O
P
P
O
N
N
N
N
N
O
P
O
P
O
O
Ethyl[(4,7-di{[ethoxy(methyl)phosphoryl]methyl}-1,4,7,10-tetraazacyclododecane-1-
yl)methyl](methyl)phosphinate (540 mg, 1.42 mmol) was dissolved in anhydrous
MeCN (5 mL) and anhydrous K
2 3
CO added (196 mg, 1.42 mmol). [5-tert-Butyl-4-
(diethoxyphosphoryl)pyridin-2-yl]methyl methanesulfonate (500 mg, 0.94 mmol) was
then added dropwise as a solution in anhydrous MeCN (10 mL). The resulting
o
mixture was heated at 70 C for 18 h, under argon. The reaction mixture was cooled
and the inorganic salts were removed by centrifugation. The solvent was removed
under reduced pressure and the resulting reddish oil was purified by alumina gel
column chromatography, eluting with a gradient starting from 100% CH
2
Cl
2
to 2%
Cl ) =
MeOH/ CH
2
Cl
2
to yield a pale orange oil (210 mg, 27%). R
f
(10% MeOH/ CH
2
2
1
4
6
3
0
.65; H NMR (400 MHz, CDCl
3
) δ 8.73 (d, JHP = 8, 1H, H ), 7.88 (d, JHP = 16, 1H,
3
H ), 4.09-4.16 (m, 4H, POCH
2
), 3.93-4.05 (m, 6H, POCH
), 1.39-1.54 (m, 18H, C(CH
); P NMR (162 MHz, CDCl ) δ 53.5, 52.9, 52.5,
2.4, 52.3, 52.2, 52.0, 51.9, 51.7, 51.5 (NCH P, mixture of isomers), 16.6, 15.9, 14.4
2
), 3.50-3.74 (m, 8H,
NCH ), 2.53-3.49 (br m, 16H, cyclen CH
2
2
3
)
3
, PCH ),
3
3
1
1
5
.20-1.32 (m, 15H, POCH
2
CH
3
3
2
1
3
2
5
(
pyP, mixture of isomers); C NMR (101 MHz, CDCl
3
) δ 149.0 (C ), 136.2 (C ),
29.5 (C ), 127.3 (C ), 111.7 (C ), 62.5-62.8 (m, POCH ), 59.7-60.3 (m, POCH ),
), 31.3 (C(CH ), 16.6 (d,
6
4
3
1
2
2
5
3.0-55.0 (m, PCH
2
N, pyCH
2
, cyclen CH
2
), 36.3 (C(CH
3
)
3
3 3
)
3
3
J
CP = 6, POCH
2
CH
3
), 16.2 (d, JCP = 6, POCH
2
CH
3
3
), 13.0-14.9 (m, PCH ); ESI-
+
+
LRMS (+) m/z 816.7 [M+H] ; ESI-HRMS (+) calcd for [C34
H
70
N
5
O
9
P
4
] 816.4124,
found 816.4129.
ꢀ
25
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
Lanthanide(III)
complexes
of
{5-tert-butyl-2-[(4,7,10-
tri{[hydroxy(methyl)phosphoryl]methyl}-1,4,7,10-tetraazacyclododecan-1-
1
yl)methyl]pyridin-4-yl}phosphonic acid, [HLn.L ]
In general, diethyl {5-tert-butyl-2-[(4,7,10-tri{[ethoxy(methyl)phosphoryl]methyl}-
1
,4,7,10-tetraazacyclododecan-1-yl)methyl]pyridin-4-yl}phosphonate was dissolved
o
in aqueous HCl (6 M) and stirred at 70 C for 18 h. Complete ester cleavage was
+
50 5 9 4
confirmed by ESI-MS (+) (LR: m/z 676.6 [M+H]+; HR: m/z calc for [C24H N O P ]
6
76.2559, found 676.2557). The solvent was removed under reduced pressure and the
resulting yellow oil was dissolved in H
2
O (1 mL). The pH of the solution was
.6H O was added. The
adjusted to 5.5 using aqueous NaOH solution and LnCl
3
2
resulting solution was stirred at 70 C for 18 h before the solvent was removed under
reduced pressure. The resulting yellow solid was purified by reverse-phase HPLC
(10-100% MeOH in H
2
O over 10 min) to yield a white solid.
1
-
1
[
HEu.L ] t
R
= 7.3 min; H NMR (400 MHz, pD 7.4, CDCl
3
) δ +3.1 (C(CH
3
)
3
major),
+
+
1.4 (C(CH
3
)
3
minor); ESI-LRMS (+) m/z 825.8 [M+H] ; ESI-HRMS (+) calc. for
1
51
+
[
[
[
C
24
H
47
N
5
O
9
P
4
Eu] 824.1523, found 824.1549.
1
-
+
HGd.L ] t
R
= 7.1 min; ESI-LRMS (+) m/z 830.4 [M+H] ; ESI-HRMS (+) calc. for
1
54
+
-1 -1
C
24
H
47
N
5
O
9
P
4
2
Gd] 827.1533, found 827.1561; r1p = 2.0±0.05 mM s (H O, 60
MHz, 310 K).
1
-
1
[
HTb.L ] t
minor), -83.4 (C(CH
calc. for [C24
R
= 7.3; min; H NMR (400 MHz, pD 7.4, CDCl
3
3 3
) δ -78.3 (C(CH )
+
3
)
3
major); ESI-LRMS (+) m/z 832.2 [M+H] ; ESI-HRMS (+)
1
59
+
H
47
N
5
O
9
P
4
Tb] 832.1578, found 832.1562.
1
-
1
[
HDy.L ] t
R
= 6.9 min; H NMR (400 MHz, pD 7.4, CDCl
3
) δ -77.1 (C(CH
3 3
)
3
1
major), -75.7 (C(CH
3
)
3
minor); P NMR (162 MHz, CDCl
3
) δ -38.2 (pyP); ESI-
+
160
+
LRMS (+) m/z 836.4 [M+H] ; ESI-HRMS (+) calc. for [C24
H
47
N
5
O
9
P
4
Dy]
8
33.1576, found 833.1573.
1
-
1
[
HHo.L ] t
major), -32.2 (C(CH
calc. for [C24
R
= 7.1 min; H NMR (400 MHz, pD 7.4, CDCl
3
3 3
) δ -33.5(C(CH )
+
3
)
3
minor); ESI-LRMS (+) m/z 836.2 [M+H] ; ESI-HRMS (+)
1
65
+
H
47
N
5
O
9
P
4
Ho] 836.1471, found 836.1473.
1
-
1
[
(
[
HEr.L ] t
R
= 7.0 min; H NMR (400 MHz, CDCl
3
) δ +42.5 (C(CH ) major), +44.2
3 3
+
C(CH
3
)
3
minor); ESI-LRMS (+) m/z 840.1 [M+H] ; ESI-HRMS (+) calc. for
1
64
+
C
24
H
47
N
5
O
9
P
4
Er] 837.1616, found 837.1636.
ꢀ
26
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
1
-
1
[
HTm.L ] t
R
= 6.9 min; H NMR (400 MHz, CDCl
3
3 3
) δ +74.6 (C(CH ) major),
+
+
75.5 (C(CH
3
)
3
minor); ESI-LRMS (+) m/z 841.9 [M+H] ; ESI-HRMS (+) calc. for
1
54
+
[C
24
H
47
N
5
O
9
P
4
Tm] 842.1667, found 842.1647.
MRI methodology for PARASHIFT measurements
In vitro and in vivo imaging was performed using a 7T preclinical MRI system (20cm
bore, Varian Direct Drive scanner, Agilent, Palo Alta, CA) equipped with a 30mm i.d.
quadrature birdcage RF coil (Rapid Biomedical GmbH, Germany) for excitation and
detection of the water and PARASHIFT agent signals.
Imaging measurements were developed using the excitation bandwidth and offset
frequencies of the RF pulses together with the receiver spectral width to limit the MRI
detection to individual resonances (water or either of the PARASHIFT probe reported
1
frequencies). The short T times for the tert-butyl resonances allows very short
imaging repetition times (TR) at high flip angle and enables high numbers of signal
averages to be acquired to maximize signal to noise ratio while maintaining practical
total scan durations. Optimal flip angle (Ernst angle) can also be defined a priori
1
since the PARASHIFT R is known and fixed by the intramolecular interaction with
the Ln ion, independent of tissue concentration.
Two types of scan acquisition were used – 3 dimensional gradient echo (3D-GE)
imaging and 2 dimensional spectroscopic imaging (2DSI), which were applied either
for imaging single resonances or for triple frequency imaging. In every case, the
excitation RF pulse was a 500 µs duration Gaussian modulated pulse which has a
-
3
FWHM bandwidth of 2100Hz, falling to a fractional excitation of <10 at 20 kHz.
This pulse specification allows separate excitation of water only, the tert-butyl group
1
-
1 -
of [Dy.HL ] or [Tm.HL ] by varying the transmit frequency.
Single resonance images were acquired with the RF pulse at the appropriate excitation
frequency and with a 20 kHz receiver bandwidth. The 3D-GE method was used for
imaging the basic spatial distribution of the resonances. Other 3D-GE image
acquisition parameters were: TR/TE 4.98/0.57ms, axial FOV 64×64mm, matrix
rd
6
4×64 (1mm in-plane resolutioin), 16 encodes in the 3 direction covering a slab
thickness of 64mm yielding 4mm thick slices. Scans used 30 signal averages for a
total duration of 152s per dataset. Pulse power was calibrated to the PARASHIFT
ꢀ
27
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
o
Ernst angle of 54 . A typical 3D-GE dataset in a multi-tube phantom containing
1
-
[
Dy.HL ] is shown in ESI Figure 1. The 2DSI method was used to acquire full
spectra from a 2D grid of voxels across the object. From these spectra the shifts in
frequency due to changes in T and pH could be measured. Sequence parameters for
2
DSI were TR/TE =10.84/0.65 ms, FOV 64×64mm, matrix16×16 (4mm in-plane
resolution), 512 data points. Scans used 30 signal averages with scan duration of 83s.
The resulting data was 3 dimensional (2 spatial, 1 spectroscopic) with the spatial
dimension localizing a thick axial (XY) plane condensing all detail along the Z
direction of the magnet. Spectra extracted from a typical 2DSI dataset in a multi-tube
1
-
phantom containing [Dy.HL ] are shown in ESI Figure 2.
1
-
Triple imaging (simultaneous but separate imaging of water, [Dy.HL ] and
1
-
[
Tm.HL ] ) was acquired using a novel pulse sequence. 3D-GE and 2DSI sequences
were adapted to include two back-to-back RF pulses whose flip angle, excitation
bandwidth and excitation frequency could be independently adjusted. For the
experiments described here the two pulses were again 500 µs duration Gaussian
1
-
1 -
pulses with equal flip angle (since R
1
values for both the [Dy.HL ] and [Tm.HL ] t-
1
-
Bu resonaces were similar) tuned to the resonance frequencies of the [Dy.HL ] and
1
-
[
Tm.HL ] (approximately -23600 and 19900 Hz respectively). The bandwidth of
these pulses does not extend to the water resonance (0 Hz), however in practice a
small residual off-resonance excitation was present at the water frequency such that
no explicit water excitation was needed for the triple imaging. For 3D-GE triple
imaging signal detection was centred at the water frequency with data matrix trebled
and imaging bandwidth trebled to 60KHz to simultaneous detect all 3 frequencies.
The data matrix can then be equally divided into 3 in the readout (directly detected)
direction, providing independent images of the 3 resonances. Other sequence
parameters were TR/TE1(Dy)/TE2(Tm) = 4.98/0.57/1.27ms, axial FOV 64×64mm,
rd
matrix 64×64 (1mm in-plane resolution), 16 encodes in the 3 direction covering a
slab thickness of 64mm yielding 4mm thick slices. Scans used 30 signal averages for
a total duration of 152s per dataset. Pulse power was calibrated (using the water
o
signal) to the PARASHIFT Ernst angle of 54 .
For triple imaging with 2DSI, spectral width was again set to 60 kHz with the number
of sampled data points set at 512. Other sequence parameters were TR/TE1(Dy)/TE2(Tm)
=
10.84/0.65/1.65ms, axial FOV 64×64mm, matrix 16×16 (4mm in-plane resolution).
ꢀ
28
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201700447
ChemEurJꢀ2017ꢀ
ꢀ
Finney/Rogers/Blamire/Parkerꢀ
Scans used 30 signal averages with scan duration of 83s. The resulting data was 3
dimensional (2 spatial, 1 spectroscopic) with the spatial dimension localizing a thick
axial (XY) plane condensing all detail along the Z direction of the magnet.
In vivo triple imaging acquisitions
Preliminary in vivo imaging studies were conducted in 3 CD1 wild type mice (Charles
River, UK) to demonstrate dynamic triple imaging and non-invasive measurement of
tissue temperature and pH. Mice were mounted in a dedicated bed including a
pneumatic pillow system to measure animal respiration with a fiber-optic
thermometry system for temperature monitoring and control via a warm air system
(SA Instruments Inc., Stony Brook, NY). Mice were anaesthetized using isoflurane
and an i.v. line was inserted into a tail vein to allow injection of contrast agent from
outside of the magnet. Unlike many previous in vivo studies using i.v. injection of
paramagnetically shifted agents, these animals were physiologically intact, i.e.
without renal ligation. All animal experiments were performed complying with the
UK Government Home Office under the animals (scientific procedure) act 1986.
Animals were positioned in the birdcage coil with a phantom placed under the body
of the animal but fastened to a line that allowed it to be withdrawn from the imaging
FOV following calibration to avoid contamination of the in vivo signal from the
phantom. This phantom consisted of a 5 mm diameter NMR tube phantom containing
~100µL of 4mM solution of the two PARASHIFT agents at pH of 7.0, surrounded by
a concentric tube containing ethylene glycol as a NMR temperature reference. To
confirm animal positioning and visualize regional anatomy conventional spin-echo
MRI scans were collected on the water resonance (TR/TE=1000/14.56ms, 40 × 1mm
thick slices, field of view FOV 40 × 40mm, matrix 256 × 256).
Following flip angle calibration, a triple imaging 2DSI data acquisition was then
collected in the axial plane from which the phantom temperature and frequency of
each PARASHIFT resonance could be determined. The phantom was then withdrawn
from the field of view and a further 2DSI dataset collected which showed only the
intrinsic water (and fat) resonance from the animal. A dynamic series of triple
imaging 3DGE and 2DSI acquisitions were then collected commencing with the
injection of 200µl of a mix of the two PARASHIFT agents (0.04 mmol/kg each of
1
-
1 -
[
Dy.HL ] and [Tm.HL ] ), followed by saline flush and continuing for 30 minutes.
ꢀ
29
This article is protected by copyright. All rights reserved.