Kinetics from indirectly detected hyperpolarized NMR spectroscopy by using spatially selective coherence transfers

Article abstract of DOI:10.1002/chem.201002151

An important recent development in NMR spectroscopy is the advent of ex situ dynamic nuclear polarization (DNP) approaches, which are capable of yielding liquid-state sensitivities that exceed considerably those afforded by the highest-field spectrometers. This increase in sensitivity has triggered new research avenues, particularly concerning the in vivo monitoring of metabolism and disease by NMR spectroscopy. So far such gains have mainly materialized for experiments that focus on nonprotonated, low-γ nuclei; targets favored by relatively long relaxation times T₁, which enable them to withstand the transfer from the cryogenic hyperpolarizer to the reacting centers of interest. Recent studies have also shown that transferring this hyperpolarization to protons by indirectly detected methods could successfully give rise to ¹H NMR spectra of hyperpolarized compounds with a high sensitivity. The present study demonstrates that, when merged with spatially encoded methods, indirectly detected ¹H NMR spectroscopy can also be exploited as time-resolved hyperpolarized spectroscopy. A methodology is thus introduced that can successfully deliver a series of hyperpolarized ¹H NMR spectra over a minutes-long timescale. The principles and opportunities presented by this approach are exemplified by following the in vitro phosphorylation of choline by choline kinase, a potential metabolic marker of cancer; and by tracking acetylcholine's hydrolysis by acetylcholine esterase, an important enzyme partaking in synaptic transmission and neuronal degradation.

Full text of DOI:10.1002/chem.201002151

FULL PAPER
DOI: 10.1002/chem.201002151
Kinetics from Indirectly Detected Hyperpolarized NMR Spectroscopy by
Using Spatially Selective Coherence Transfers
Talia Harris,[a] Patrick Giraudeau,[a, b] and Lucio Frydman*[a]
Abstract: An important recent devel-
opment in NMR spectroscopy is the
advent of ex situ dynamic nuclear po-
larization (DNP) approaches, which
are capable of yielding liquid-state sen-
sitivities that exceed considerably those
afforded by the highest-field spectrom-
eters. This increase in sensitivity has
triggered new research avenues, partic-
ularly concerning the in vivo monitor-
ing of metabolism and disease by
NMR spectroscopy. So far such gains
have mainly materialized for experi-
ments that focus on nonprotonated,
low-g nuclei; targets favored by rela-
tively long relaxation times T1, which
enable them to withstand the transfer
from the cryogenic hyperpolarizer to
the reacting centers of interest. Recent
studies have also shown that transfer-
ring this hyperpolarization to protons
by indirectly detected methods could
also be exploited as time-resolved hy-
perpolarized spectroscopy. A method-
ology is thus introduced that can suc-
cessfully deliver a series of hyperpolar-
1
ized H NMR spectra over a minutes-
1
successfully give rise to H NMR spec-
long timescale. The principles and op-
portunities presented by this approach
are exemplified by following the in
vitro phosphorylation of choline by
choline kinase, a potential metabolic
marker of cancer; and by tracking ace-
tylcholineꢀs hydrolysis by acetylcholine
esterase, an important enzyme partak-
ing in synaptic transmission and neuro-
nal degradation.
tra of hyperpolarized compounds with
a high sensitivity. The present study
demonstrates that, when merged with
spatially encoded methods, indirectly
detected 1H NMR spectroscopy can
Keywords: dynamic nuclear polari-
zation · enzyme catalysis · NMR
spectroscopy · spatial encoding ·
time-resolved spectroscopy
Introduction
chemical shifts, coupled with the spatial localization capabil-
ities that arise upon the employment of magnetic field gradi-
ents, have been key ingredients in enabling these applica-
tions. Further progress in this usage of NMR spectroscopy
and of MRI hinges on improving the information available
from these techniques. Particularly pressing is the need to
enhance the sensitivity of NMR spectroscopy, given the
weak signals associated with these experiments and the low
natural concentrations of typical metabolic markers. Recent
years have witnessed what promises to become a major
thrust of these efforts, with emerging alternatives capable of
preparing nuclei in “hyperpolarized” liquid-phase states.
These are metastable spin arrangements in which the bulk
nuclear polarizations depart from their usual (ca. 10ꢀ5)
Boltzmann distributions, and approach unity values.[3] Once
prepared and transferred into an NMR/MRI scanner, the
“super-signals” that can be elicited from these spin arrange-
ments can be exploited within a timescale dependent on the
longitudinal relaxation time T1 of the hyperpolarized nuclei.
Methods proposed and demonstrated for producing and ex-
ploiting these extreme spin-polarization levels under in vivo
compatible conditions include the use of parahydrogen,[4,5]
Continuous efforts are being invested towards developing
new approaches with improved abilities to characterize dy-
namic biochemical processes in general, and in vivo metabo-
lism in particular. These tools can find use in biological re-
search, as well as in clinical applications that assess disease
and its response to therapeutic treatments. For several de-
cades, NMR spectroscopy and magnetic resonance imaging
(MRI) have served as central methods to enable such stud-
ies; both in vitro, as well as in animal and human metabolic
settings.[1,2] The atomic site specificities carried by NMRꢀs
[a] T. Harris, Dr. P. Giraudeau, Prof. L. Frydman
Department of Chemical Physics
Weizmann Institute
Rehovot, 76100 (ISRAEL)
Fax : (+972)-8-934-4123
E-mail: lucio.frydman@weizmann.ac.il
[b] Dr. P. Giraudeau
CEISAM
UMR CNRS 6230
University of Nantes (France)
Chem. Eur. J. 2011, 17, 697 – 703
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
697

the optical pumping of noble gases,[6,7] and microwave-
driven dynamic nuclear polarization (DNP) from unpaired
electrons.[8–10] Of these methods, the last option has the
appeal of providing high enhancements while making few
demands in terms of chemical substrate or sample prepara-
tion. Several in situ or ex situ experimental approaches have
been proposed to perform DNP prior to NMR spectroscopic
acquisition.[11] Particularly promising have been the metabol-
ic results afforded by the ex situ DNP approach proposed
by Golman et al.,[12] whereby the sample to be analyzed is
comixed with a free radical, frozen in a glass at cryogenic
temperatures, and hyperpolarized by irradiating the radical
close to its unpaired electronꢀs Larmor frequency for rela-
tively long periods of time. Rapid melting and shuttling of
the nuclear spins from this cryogenic environment into a
scanner enables an otherwise routine solution-state NMR
spectroscopic or MRI observation.[12,13] Yet, upon taking
into account the effects of the cryogenic cooling and of the
DNP process, these “routine” observations will take place
on nuclei the polarization of which has been enhanced by a
factor of around (gelectron/gnucleus)ꢂ(BDNPTNMR/BNMRTDNP), or
the equivalent of around 104 ꢂ magnification.
zation to neighboring protons by using spatially selective
principles. This maximizes the efficiency of the transfer
while providing the temporal dimension needed for carrying
out metabolic characterizations. The potential offered by
these new spectroscopic approaches is illustrated by hyper-
polarized NMR spectroscopy measurements of two choline-
related enzymatic reactions involved in cancer metabolism
and in neuronal transmission.
Results and Discussion
Scheme 1 illustrates the kind of problems this study at-
tempts to tackle and the methods adopted to address them.
These are all water-soluble metabolites with a number of
Scheme 1. Choline derivatives analyzed in this study, and the 1H and 15N
NMR chemical shifts associated with each structure.
Despite its outstanding promise, the time needed for ex
situ DNP to transfer the sample from the polarizer to the
NMR/MRI scanner, combined with the delays that may be
subsequently needed for monitoring the ensuing metabolic
processes, implies that only sites with relatively long liquid-
state T1s will be capable of retaining a significant hyperpola-
rization. For this reason, hyperpolarized metabolic studies
have focused principally on nonprotonated low-g nuclei,
such as the 13C of carbonyls or the 15N of quaternary
amines.[14–27] While bearing an exciting potential for improv-
ing high-resolution liquid-state NMR spectroscopy sensitivi-
ty in general, and in vivo spectroscopy in particular, these
procedures appear to leave aside the important realm of
1H NMR spectroscopy. This is a considerable drawback
given that this spectroscopy 1) is associated with the highest-
sensitivity target in all NMR spectroscopy, 2) utilizes the
most common hardware available in MRI scanners, and
3) may be characterized by better spectral and much superi-
or spatial resolutions than those associated with measure-
ments on low-g counterparts. The first of these features, in
particular, has made so-called inverse methods of detections
(whereby NMR information initially encoded on low-g
distinct biological roles: choline plays an integral part in
methyl-transfer metabolism; its transformation to phospho-
choline is an essential step in the formation of lipids making
up cell membranes,[34] and in its acetylated form it is a prin-
cipal brain neurotransmitter.[35] Moreover, an aberrant me-
tabolism of choline is associated with a number of illnesses:
the choline$phosphocholine transformation has been
shown to be upregulated in cancerous cells,[36,37] and altera-
tions in the choline$acetylcholine interconversion have
been found in Alzheimerꢀs and Parkinsonꢀs diseases.[38,39]
When viewed as a potential target for ex situ DNP hyperpo-
larization, choline shows great promise: it possesses a non-
protonated nitrogen that has been shown to undergo effi-
cient hyperpolarization, and is endowed with very long T1
relaxation times ranging from about 150 s in the moleculesꢀ
protonated forms to around twice this time upon perdeuter-
ation.[19] Figure 1 gives an idea of the sensitivity enhance-
ments and long probing times that can be necessary for this
site, for the Hypersense and 11.7 T Varian DNP NMR setup
used in this study. The interest of detecting this polarization
enhancement for in vivo studies has been very recently high-
lighted.[40]
1
nuclei is transferred to H nuclei for final detection) a main-
stream approach in analytical and biomolecular studies.[28–30]
This promising alternative has also been recently explored
within the contexts of DNP- and para-hydrogen-enhanced
hyperpolarized NMR spectroscopy,[31,32] in which advantages
could result from storing the hyperpolarization in low-g
nuclei capable of undergoing an efficient enhancement, and
retaining it for the long timescale compatible with metabolic
processes, and then transferring it to nearby protons on
which the NMR measurements are actually made.[33] Herein,
we further explore the usefulness of these approaches, par-
ticularly in combination with new methods capable of re-
peatedly transferring aliquots of the low-g spin hyperpolari-
In spite of this potential, cholineꢀs 15N site is ill-positioned
for following the acetylation or phosphorylation reactions
that one might be interested to quantify in metabolic set-
tings. Both of these processes take place on the moleculeꢀs
hydroxyl site, and result in the minor 15N NMR shift effects
summarized in Scheme 1. In contrast, much larger shift
changes affect the protons bonded to the C2 sites of these
molecules, because they are closer to the hydroxyl epicenter
of the metabolic transformations. This chemical-shift data
highlights the convenience of exploiting cholineꢀs slowly re-
698
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 697 – 703

Kinetics from Indirectly Detected Hyperpolarized NMR Spectroscopy
FULL PAPER
background metabolite signals that could potentially ob-
scure the relevant peaks. Figure 2A illustrates the results of
applying this sequence to hyperpolarized choline in D2O
and to acetylcholine dissolved in a buffered H2O solution.
Relative to conventional 1H NMR spectra of the same
sample, this sequence gives rise to a site-specific polarization
enhancement of about 300. Although high, this enhance-
ment still amounts to only around 50% of the maximum
theoretical enhancement that one could expect from a full
1
15N! H transfer. At the same time, however, the DNP-en-
1
hanced H spectrum reveals an almost complete elimination
of background signals, including the strong water peak that
Figure 1. A) Comparison between conventional (bottom) and hyperpolar-
ized (top) 15N NMR spectra arising from choline and acetylcholine for
the indicated concentrations and number of signal averages. On the basis
of the signal to noise ratios measured in each trace, polarization enhance-
ment factors of 5500ꢂ and 6000ꢂ could be inferred for each of these
compounds. B) T1-driven decay of the 15N hyperpolarization of choline
and acetylcholine, as measured by a train of single-pulse 1D NMR spec-
troscopy experiments applied after sample injection. Hyperpolarization
arose from cofreezing the analytes with tris{8-carboxyl-2,2,6,6-tetra[2-(1-
hydroxyethyl)]benzo(1,2-d:4,5-d’)bisACTHNUTRGNEUGN(1,3)dithiole-4-yl}methyl sodium salt
(trityl) in a 1:1 solution of D2O/[D6]DMSO, followed by irradiation for
about 3 h at 94.188 GHz with 100 mW microwave power, and quickly dis-
solving them in either D2O (left) or H2O/tris(hydroxymethyl)aminome-
thane (tris) buffer (right) to the indicated final concentrations.
laxing hyperpolarized 15N site as a source of enhanced NMR
signal, while following the actual reaction process in an indi-
rect detection mode that exploits the larger chemical-shift
changes that the oxygen-bonded methylene protons will un-
dergo upon reacting. As recently discussed elsewhere,[32,33]
such information can be transferred through the small but
Figure 2. A) 1H NMR spectra of choline and acetylcholine recorded at
thermal equilibrium (top), and counterparts arising from an INEPT se-
quence [see Eq. (1)], starting from hyperpolarized 15N, and using an opti-
mized transfer of the spin order to the C2 methylene protons. Minor dif-
ferences in the chemical shifts of the two spectra might reflect small dif-
ferences in the samplesꢀ temperatures. B) The pulse sequence employed
to transform the single-shot transfer used in A) into an experiment that
makes multiple transfers of the 15N hyperpolarization to these protons at
various times: replacing the hard (p/2)N pulses of Equation (1) by selec-
tive ones applied at shifted frequencies {fi}i=1,N in the presence of suitable
field gradients opens the possibility of “drawing” magnetization from dif-
ferent sample positions over the long timescales supported by the 15N T1.
Additional sequence components include a hard (p)N pulse that leaves
the 15N magnetization of all nonaddressed slices along ꢁz, spectrally se-
lective (p)H 1H pulses that solely address the targeted methylene reso-
nances (solid lines), a reliance on gradients that further purify the desired
coherence transfer pathway, and a delay (t=32 ms) optimized to transfer
the hyperpolarization to the methylene protons. Immediate repetition of
the sequence yielded a reference background, from which residual water
signals were subtracted when necessary. C) Results obtained upon loop-
ing the pulse sequence in B) to extract 15N-choline hyperpolarization at
seven different times over lapses that exceed those supported by the nat-
ural T1s of the targeted protons by orders of magnitude. The illustrated
trace arose from applying 0.71 ms selective (p/2)N sinc pulses on a 5 mm
hyperpolarized sample dissolved in D2O; each of these was spaced by ap-
proximately 3 kHz frequency increments and applied in the presence of
35 G/cm gradients. The effective decay constant of the resulting data is
T1*=125 s.
ACHTUNGTRENNUNG
non-negligible 3J(N,H) couplings between the 15N and the
C2-bearing protons, by using either single-quantum INEPT-
based or multiple-quantum DEPT-like processes. We chose
to rely on the former, by exploring a spectrally selective ver-
sion of the classical Freeman–Morris three (p/2) sequence
[Eq. (1)]:[41]
N
N
H
N ðhypÞ ðp=2Þ
ƒƒƒƒƒƒƒ!
N
2 H N
z x
0:25=J,ðpÞ =ðpÞ ,0:25=J
x
sel
ƒƒƒ! ƒƒƒƒƒƒƒƒƒƒƒƒ!
z
y
N
H
ð1Þ
ðp=2Þ ðp=2Þ
1
y
2H N ꢀAcqð HÞ
x
z
in which superscript indices denote the targeted species, sub-
script indices are relevant pulse phases, and 3 Hz is the ap-
proximate coupling measured between the 15N and the perti-
1
nent Hs. The sequence in Equation (1) begins from the hy-
perpolarized 15N state, and relies on a selective H p pulse
1
applied on-resonance with the targeted methylenes for im-
proving a transfer which, if broadband, would share a sub-
stantial part of the 15N hyperpolarization with the more nu-
merous (but otherwise uninformative) methyl protons. The
heteronuclear transfer process in Equation (1) also provides
an opportunity to use gradient-based coherence selection
pulses for reducing the intense 1H water peak and other
Chem. Eur. J. 2011, 17, 697 – 703
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
699

L. Frydman et al.
resonates in the neighborhood of acetylcholineꢀs diagnostic
methylene 2-CH2 peak.
geted sample. This approach presents some similarities with
the strategy recently proposed by Panek et al. to include
slice selection in single-scan experiments on hyperpolarized
This promise notwithstanding, the methylene enhance-
ment observed in Figure 2A is ill-suited for acquiring the
multiple scans that are needed for extracting metabolic in-
formation from hyperpolarized molecules. Indeed, for
simple experiments, this can be accomplished by monitoring
1D spectra after applying a series of low-excitation-angle
pulses on the hyperpolarized state. The sequence in Equa-
tion (1), on the other hand, relies on the use of two (p/2)N
pulses and an optimized interpulse delay, which is meant to
transfer the full polarization from the low-g nucleus to the
1
samples.[50] Figure 2C shows H NMR results obtained upon
executing this sequence on a sample of hyperpolarized chol-
ine: as can be appreciated, a clean series of informative
methylene 1H NMR spectra with good sensitivity is ob-
served; these are characterized by the long lifetimes sup-
ported by the extreme T1 of the 15N site acting as their
source of hyperpolarization.
Note that there are characteristic losses that (beyond the
T1 decay) may arise with this kind of approach. Diffusion, in
particular, is a potential source of loss that could affect ex-
periments of this kind through two main mechanisms: one
involves spatial displacements of the spins before the mag-
netization windings that derive from the application of gra-
dients have been refocused; the other involves molecular
displacements of the spins from their ideal positions to be
targeted by the selective pulses to unintended regions that
are not meant to be addressed. The first kind of phenomena,
which are usually the main source of signal loss in gradient-
based NMR spectroscopy measurements, are essentially
absent in these experiments because of the immediate refo-
cusing gradients applied in conjunction with the excitation/
storage pulses that are used to select the spatial slices (Fig-
ure 2B). The second mechanism would become active if
1) spins are displaced in between the spatially selective 15N
excitation and storage pulses that address a given slice, or
2) or spins undergo considerable displacement between the
consecutive slice-selective scans that follow the common hy-
perpolarization process. To estimate how much such dis-
placement effects would influence the experimentꢀs sensitivi-
ty, we consider a fast water-like diffusion coefficient (D
1
targeted H nucleus. All the ex situ hyperpolarization is thus
expended in a single shot. Options that could help in con-
serving the 15N hyperpolarization, and that are compatible
with indirect detection conditions, include reducing the
pulse angles involved in the nuclear excitation, or choosing
an interpulse delay that does not fully transfer the single-
spin 15Nz hyperpolarization into the single-quantum anti-
1
phase states involved in the subsequent H detection. Both
of these approaches, however, entail signal losses: reliance
on 15N pulses that are not p/2 rotations will dissipate por-
tions of the low-g hyperpolarization in a process that rapidly
accumulates; keeping the use of the two (p/2)N pulses but
manipulating their interpulse delay will transfer aliquots of
magnetization, but will create, in the process, longitudinal
2NzHz-like spin-order states, the decay of which will be do-
minated by the protonꢀs T1 and therefore be incompatible
with the reaction timescales normally addressed by metabo-
lism.
To bypass these constraints, we explored the integration
of the DNP-enhanced indirect detection sequence sketched
in Equation (1) with spatial-encoding techniques. Spatial en-
coding is an approach typically used to replace the serial in-
crementation of a temporal variable (for instance, the t1 in-
direct domain typically changed by constant steps within
2D NMR sequences) with a spin coordinate position within
the sample.[42,43] This can be implemented with the aid of
imaging-like magnetic field gradients acting in combination
with frequency-swept (or stepped) manipulations of the
spins. Besides enabling the recording of arbitrary multidi-
mensional NMR spectra within a single scan (including the
indirectly detected acquisition of 2D NMR correlations on
DNP-hyperpolarized samples[44–46]), spatially encoded meth-
ods can deliver in a single shot an array of inversion–recov-
ery T1 measurements, or multiple-exchange 2D NMR spec-
tra as a function of their intervening mixing periods.[47–49] In
a similar spirit to these latter studies, the sequence in Fig-
ure 2B explores the opportunities that spatially selective
manipulations open for measuring a kinetic series, starting
from a singly hyperpolarized injected sample. This experi-
ment involves a simple modification of the sequence in
Equation (1), whereby the two (p/2)N manipulations have
been made spatially selective. The main consequence of in-
troducing this modification is to enable the acquisition of a
series of experiments like the one illustrated in Figure 2A,
each of these arising from a different location within the tar-
ꢂ10ꢀ5 cm2sꢀ1), and rely on Fickꢀs Law,
NACHTUNGTRENNUNG(z,t)=
p
N
(z,0)*
E
(z/
(4Dt))], to evaluate particle dispersion
for the relevant time- and length-scales, in which t was as-
sumed to be around 0.05 s for the RF interpulse delay, 100 s
for the interscan delay, and Dz around 2.25 mm for the slab
thickness (which was the smallest value used throughout our
experiments; larger slabs should be less affected by these
kind of “edge effects”). Under these assumptions it follows
that the interpulse diffusion effects will be <0.2%, which is
fully negligible; the interscan effects in contrast are much
larger and should result in signal losses of around 9% rela-
tive to the full, integrated ideal signal. These losses stem
from the fact that, in the aforementioned interscan delay, a
portion of the hyperpolarized sample will migrate away
from the targeted slab, and be replaced with spins that arise
from an expended slab entering with negligible polarization.
Effects of such magnitude might be visible, and may be
partly responsible for the slightly different effective T1s that
could be derived from small-angle measurements like those
in Figure 1B (ca. 170 s) as opposed to those arising from the
spatially selective measurements in Figure 2C (ca. 125 s).
This difference, however, probably also reflects the presence
of other nonidealities (including pulse shaping and p-inver-
sion-related losses in the latter sequences). Moreover, even
700
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 697 – 703

Kinetics from Indirectly Detected Hyperpolarized NMR Spectroscopy
FULL PAPER
1
if quantitatively measurable, these diffusion effects should
not bring about any noticeable distortions in the kinetics de-
rived from these hyperpolarized NMR measurements.
Indeed, this information will arise from measuring ratios be-
tween metabolic reactants and products; assuming that the
D coefficients for these molecules are similar, these ratios
should be, for all practical purposes, independent of all the
diffusion effects considered in this paragraph.
With these considerations as background, we set out to
explore the usefulness of this approach for monitoring meta-
bolic processes. The first reaction targeted was the phos-
phorylation of choline by choline kinase. This same reaction
has been utilized with hyperpolarized 15N NMR spectrosco-
py,[19] and presents a good opportunity to follow the kinetics
through both direct detection of the low-g nucleus using a
small-pulse-angle train, as well as through the indirectly de-
tected 1H NMR sequence introduced in Figure 2. Figure 3
summarizes results obtained for this system, and shows the
expected decrease in the initial choline peak and a concomi-
tant increase in the phosphocholine peak. This behavior is
evidenced by both the 15N direct-excitation hyperpolarized
experiments, the spectra of which show the reactant and
product peaks separated by 0.2 ppm (ca. 10 Hz at 11.7 T); as
well as by the hyperpolarized H methylene spectra, which
show peaks separated by 0.12 ppm (60 Hz at 11.7 T). Good
agreement can be seen between the kinetic results obtained
by these directly and indirectly detected set of spectra,
based on an integration of the resolved H and 15N resonan-
1
ces (see the data points in Figure 3C).
A second reaction that we targeted involved monitoring
the hydrolysis of acetylcholine by an esterase. As the 15N
peaks of acetylcholine and choline are, in this instance, sepa-
rated by ꢃ0.05 ppm (2.5 Hz at 11.7 T), it would be difficult
or impossible to resolve these two species by direct 15N de-
tection. In contrast, transferring the hyperpolarization to the
methylene protons adjacent to the reacting hydroxyl
group—separated in the two compounds by over 0.4 ppm—
opens a new route to following these reaction kinetics. Cru-
cial in this endeavor is the feasibility of probing the reaction
in the presence of an efficient gradient-driven suppression
of the water resonance, which falls less than 0.25 ppm away
from the product peak. The outcome of four consecutive re-
action time points is illustrated in Figure 4B. Although
direct comparison with the 15N NMR results is not possible,
there is good agreement between this data and longer-term
1
(ꢄ300 s) observations based on the direct H NMR detec-
tion of the reaction products.
Figure 3. NMR spectral changes revealed by a 5 mm solution of hyperpo-
larized choline upon undergoing phosphorylation by 0.5 units of choline
kinase. A) Emergence of the new phosphocholine resonance shown by
directly detected single-pulse 15N NMR spectroscopy experiments.
B) Emergence of the H NMR resonance associated with the methylenes
in the C2-position of phosphocholine, as revealed by the indirect detec-
Figure 4. A) Spectral changes seen for hyperpolarized acetylcholine upon
undergoing deacetylation into choline, under the action of 0.6 units of
acetylcholine esterase. The 15N-enhanced 1H spectra only show resonan-
ces associated with the methylenes in the C2 position, as revealed by the
indirect detection sequence introduced in Figure 2B. B) Comparison be-
tween the enzymatic kinetic results afforded by the indirectly detected
1
tion sequence introduced in Figure 2B. C) Comparison between the ex-
15
&
pected enzyme kinetics of kinase with results afforded by the N- ( )
1
^
and H-detected ( ) hyperpolarized experiments, as derived from the rel-
hyperpolarized experiments in A), and the enzymatic activity rate ob-
1
*
ative peak ratios of the NMR peaks in A) and B). The straight line illus-
trates the best fit of the combined set of data points, and corresponds to
an initial phosphorylation rate of 0.3 mmolminꢀ1 under these conditions.
served by conventional H NMR ( ) spectroscopy at longer reaction/
signal-averaging times (ꢄ200 s). Under the assayed conditions, these data
(straight line) indicate a deacetylation rate of 0.12 mmolminꢀ1
.
Chem. Eur. J. 2011, 17, 697 – 703
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
701

L. Frydman et al.
100 mm KCl, 50 mm MgCl2, and 10 mm ATP. For the esterase experi-
ments, acetylcholine esterase (Sigma) was also prepared in a buffered
aqueous solution (pH 7) to give, after hyperpolarization, final solution
concentrations of 4 mm acetylcholine, 40 mm tris and 0.5% Triton X-100.
Conclusion
The data summarized in Figures 2–4 may open up valuable
new opportunities in the field of metabolic NMR spectros-
copy, since they show that the unique sensitivity enhance-
ment afforded by ex situ DNP can be combined with the
long lifetimes typical of low-g heteronuclei, as well as with
the wealth of information and additional sensitivity that
characterizes indirect detection. Neither the sequences nor
the principles involved in these new experiments are partic-
ularly complex, but the resulting data is unique. Although
exemplified for the case of a quaternary 15N site, similar ad-
vantages may also be brought to bear for other sites, includ-
ing hyperpolarized carbonyl 13C groups adjacent to methyl-
ene or methyl groups (e.g., acetates) or 15N sites in aromatic
bases.
A particularly encouraging feature of our approach is
that, by transferring the observation of the signal to 1H
nuclei, these experiments become compatible with the re-
ceivers that are normally available in MRI scanners. This
hardware, which has been highly optimized over the years,
would still have to be complemented with low-g irradiation
facilities for performing the INEPT-derived sequences, but
this can often be carried out in an ad hoc, less demanding
fashion. Moreover, 1H-based detection would enable opti-
mal use of a scannerꢀs field gradients, both for spatial locali-
zation as well as for suppression of background resonances
through coherence-selective gradients. The spatial nature of
the features exploited in this study has its own challenges,
which include complications that may arise in vivo if the
spatial homogeneity present in the in vitro setup analyzed in
this work is absent. Still, a built-in way to compensate for
such spatial heterogeneity effects arises from the reliance of
this approach on measuring the ratio between the signal in-
tensities of the reactant and product to extract its kinetic in-
formation. Assuming that these metabolites share similar
diffusion and/or flow characteristics, this normalization
would also account to a large extent for interpulse and even
interscan molecular displacements. These hypotheses and
further uses of the presented approach are currently being
explored.
Spectroscopic methods: DNP-enhanced NMR spectroscopy experiments
were performed by using an OIMBL Hypersense polarizer operating at
1.5 K and a nominal electron Larmor frequency of 94 GHz. Hyperpolari-
zation was implemented on 1.5 and 0.56m solutions of 15N-labeled choline
and acetylcholine, respectively, which were prepared in
a 1/1 D2O/
[D6]DMSO solvent together with 20 mm of OX063 trityl radical.[51] Typi-
cal experiments utilized 5–20 mL aliquots of these solutions. Following
DNP hyperpolarization the samples were transferred with 3 mL of pres-
surized water vapor (10 bar) into 5 mm NMR tubes, to give approximate-
ly 2 mL sample volumes (final concentrations indicated throughout the
paper). A 500 MHz Varian Inova spectrometer equipped with an inverse
HCN triple-resonance triple-axis gradient probe was employed in these
NMR spectroscopy experiments. In a typical run, about 2.5 s would
elapse between triggering the sampleꢀs melting/transfer process, and fill-
ing the NMR sample tube. Given the much longer timescales supported
by the hyperpolarized 15N T1s, an additional delay of about 10 s would be
usually taken for homogenizing the sample outside the NMR magnet,
and another 30 s would be devoted to locking/shimming the spectrometer
before starting the NMR pulse sequences. Various other parameters re-
quired for executing the experiments introduced in this work (radiofre-
quency offsets and shapes, gradient strengths, etc.) were calibrated prior
to the injections using a reference 0.5m 15N-choline sample and standard
procedures.
Acknowledgements
We are grateful to Prof. Hadassa Degani (Weizmann Institute, Biological
Regulation) for the generous gift of choline kinase, to Dr. Veronica Fryd-
man (Weizmann Institute, Research Infrastructures) for the preparation
of the 15N-labeled acetylcholine, and to Dr. Rachel Katz-Brull (Hebrew
University - Hadassa Medical Center) and Prof. Geoffrey Bodenhausen
(EPFL, Lausanne) for valuable discussions. This research was supported
by the Israel Science Foundation (ISF 447/09), by a Helen and Martin
Kimmel Award for Innovative Investigation, and by the generosity of the
Perlman Family Foundation. P.G. acknowledges the Weizmann Feinberg
Graduate School and the French Ministry of Foreign and European Af-
fairs (Lavoisier program) for fellowships.
[1] R. A. de Graaf, In Vivo NMR Spectroscopy: Principles and Tech-
niques, Wiley, New York, 1998.
[2] R. G. Shulman, D. L. Rothman, Metabolism by In Vivo NMR, Wiley,
New York, 2004.
[3] W. Kockenberger, T. Prisner, Appl. Magn. Reson. 2008, 34, 213.
[4] C. R. Bowers, D. P. Weitekamp, Phys. Rev. Lett. 1986, 57, 2645.
[5] T. C. Eisenschmid, R. U. Kirss, P. P. Deutsch, S. I. Hommeltoft, R.
Eisenberg, J. Bargon, R. G. Lawler, A. L. Balch, J. Am. Chem. Soc.
1987, 109, 8089.
[6] M. S. Albert, G. D. Cates, B. Driehuys, W. Happer, B. Saam, C. S.
Springer, A. Wishnia, Nature 1994, 370, 199.
[7] G. Navon, Y.-Q. Song, T. Rꢃꢃm, S. Appelt, R. E. Taylor, A. Pines,
Science 1996, 271.
Experimental Section
Materials: This study focused on 15N-labeled choline (Sigma Aldrich) and
on 15N-labeled acetylcholine. The latter was prepared by coevaporating
15N-choline chloride with absolute ethanol and further drying under
vacuum at 458C. The resulting salt (40 mg) was suspended in anhydrous
Cl2CH2 (2 mL) under nitrogen, and acetylated with CH3COCl (0.3 mL)
overnight. To ensure full acetylation, this treatment was repeated twice.
Solvents were then removed at reduced pressure and the residue was
once again coevaporated with ethanol and dried under vacuum to yield
15N-acetylcholine chloride as a white crystalline solid. Hyperpolarized
samples of 15N-labeled choline and acetylcholine were then studied on
being subjected to enzymatic reactions by kinases and esterases, respec-
tively. The choline kinase (Sigma) experiments were carried out in buf-
fered solutions in H2O (pH 8) that contained, once inside the NMR spec-
trometer, concentrations of 5 mm hyperpolarized choline, 100 mm tris,
[8] A. Abragam, M. Goldman, Nuclear Magnetism: Order and Disorder,
Oxford University Press, Oxford, 1982.
[9] T. R. Carver, C. P. Slichter, Phys. Rev. 1953, 92, 212.
[10] K. H. Hausser, D. Stehlik, Adv. Magn. Reson. 1968, 3, 79.
[11] R. G. Griffin, T. Prisner, Phys. Chem. Chem. Phys. 2010, 12, 5737.
[12] J. H. Ardenkjaer-Larsen, B. Fridlund, A. Gram, G. Hansson, L.
Hansson, M. H. Lerche, R. Servin, M. Thaning, K. Golman, Proc.
Natl. Acad. Sci. USA 2003, 100, 10158.
[13] J. Wolber, F. Ellner, B. Fridlund, A. Gram, H. Johannesson, G.
Hansson, L. H. Hansson, M. H. Lerche, S. Mansson, R. Servin, M.
702
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 697 – 703

Kinetics from Indirectly Detected Hyperpolarized NMR Spectroscopy
FULL PAPER
Thaning, K. Golman, J. H. Ardenkjaer-Larsen, Nucl. Instr. and
Meth. in Phys. A 2004, 526, 173.
[14] M. J. Albers, R. Bok, A. P. Chen, C. H. Cunningham, M. L. Zierhut,
V. Y. Zhang, S. J. Kohler, J. Tropp, R. E. Hurd, Y. F. Yen, Cancer
Res. 2008, 68, 8607.
[15] P. Bhattacharya, E. Y. Chekmenev, W. H. Perman, K. C. Harris,
A. P. Lin, V. A. Norton, C. T. Tan, B. D. Ross, D. P. Weitekamp, J.
Magn. Reson. 2007, 186, 150.
[30] L. Mueller, J. Am. Chem. Soc. 1979, 101, 4481.
[31] E. Y. Chekmenev, V. A. Norton, D. P. Weitekamp, P. Bhattacharya,
J. Am. Chem. Soc. 2009, 131, 3164.
[32] R. Sarkar, A. Comment, P. R. Vasos, S. Jannin, R. Gruetter, G. Bod-
enhausen, H. Hall, D. Kirz, V. P. Denisov, J. Am. Chem. Soc. 2009,
131, 16014.
[33] J. A. Pfeilsticker, J. E. Ollerenshaw, V. A. Norton, D. P. Weitekamp,
J. Magn. Reson. 2010, 205, 125.
[16] A. P. Chen, M. J. Albers, C. H. Cunnigham, S. J. Kohler, Y.-F. Yen,
R. E. Hurd, J. Tropp, R. Bok, J. M. Pauly, S. J. Nelson, J. Kurhane-
wicz, D. B. Vigneron, Magn. Reson. Med. 2007, 58, 1099.
[17] A. P. Chen, J. Kurhanewicz, R. Bok, D. Xu, D. Joun, V. Zhang, S. J.
Nelson, R. E. Hurd, D. B. Vigneron, Magn. Reson. Imaging 2008, 26,
721.
[18] S. E. Day, M. I. Kettunen, F. A. Gallagher, D.-E. Hu, M. Lerche, J.
Wolber, K. Golman, J. H. Ardenkjaer-Larsen, K. M. Brindle, Nat.
Med. 2007, 13, 1382.
[34] S. H. Zeisel, Annu. Rev. Nutr. 2006, 26, 229.
[35] M. Sarter, V. Parikh, Nat. Rev. Neurosci. 2005, 6, 48.
[36] G. Eliyahu, T. Kreizman, H. Degani, Int. J. Cancer 2007, 120, 1721.
[37] K. Glunde, M. A. Jacobs, Z. M. Bhujwalla, Expert Rev. Molec. Diag-
nostics 2006, 6, 821.
[38] K. Herholz, Eur. J. Nucl. Med. Molec. Imag. 2008, 35, S25.
[39] H. Shimada, S. Hirano, H. Shinotoh, A. Aotsuka, K. Sato, N.
Tanaka, T. Ota, M. Asahina, K. Fukushi, S. Kuwabara, Neurology
2009, 73, 273.
[19] C. Gabellieri, S. Reynolds, A. Lavie, G. S. Payne, M. O. Leach, T. R.
Eykyn, J. Am. Chem. Soc. 2008, 130, 4598.
[20] F. A. Gallagher, M. I. Kettunen, S. E. Day, M. Lerche, K. M. Brin-
dle, Magn. Reson. Med. 2008, 60, 253.
[21] K. Golman, J. H. Ardenkjaer-Larsen, J. S. Petersson, S. Mansson, I.
Leunbach, Proc. Natl. Acad. Sci. USA 2003, 100, 10435.
[22] K. Golman, J. S. Petersson, P. Magnusson, E. Johansson, P. Akeson,
C. M. Chai, G. Hansson, S. Mansson, Magn. Reson. Med. 2008, 59,
1005.
[40] C. Cudalbu, A. Comment, F. Kurdzesau, R. B. van Heeswijk, K.
Uffman, S. Jannin, V. P. Denisov, D. Kirik, R. Gruetter, Phys. Chem.
Chem. Phys. 2010, 12, 5818.
[41] G. A. Morris, R. Freeman, J. Am. Chem. Soc. 1979, 101, 760.
[42] L. Frydman, A. Lupulescu, T. Scherf, J. Am. Chem. Soc. 2003, 125,
9204.
[43] L. Frydman, T. Scherf, A. Lupulescu, Proc. Natl. Acad. Sci. USA
2002, 99, 15858.
[44] L. Frydman, D. Blazina, Nat. Phys. 2007, 3, 415.
[45] P. Giraudeau, Y. Shrot, L. Frydman, J. Am. Chem. Soc. 2009, 131,
13902.
[23] K. Golman, R. i. t. Zandt, M. Lerche, R. Pehrson, J. H. Ardenkjaer-
Larsen, Cancer Res. 2006, 66, 10855.
[24] T. Harris, G. Eliyahu, L. Frydman, H. Degani, Proc. Natl. Acad. Sci.
USA 2009, 106, 18131.
[25] S. J. Kohler, Y. Yen, J. Wolber, A. P. Chen, M. J. Albers, R. Bok, V.
Zhang, J. Tropp, S. Nelson, D. B. Vigneron, Magn. Reson. Med.
2007, 58, 65.
[26] M. E. Merritt, C. Harrison, C. Storey, F. M. Jeffrey, A. D. Sherry,
C. R. Malloy, Proc. Natl. Acad. Sci. USA 2007, 104, 19773.
[27] M. A. Schroeder, L. E. Cochlin, L. C. Heather, K. Clarke, G. K.
Radda, D. J. Tyler, Proc. Natl. Acad. Sci. USA 2008, 105, 12051.
[28] G. Bodenhausen, D. J. Ruben, Chem. Phys. Lett. 1980, 69, 185.
[29] J. Cavanagh, W. J. Fairbrother, A. G. Palmer, N. J. Skelton, M.
Rance, Protein NMR spectroscopy: Principles and Practice, Elsevier,
Amsterdam, 2006.
[46] M. Mishkovsky, L. Frydman, ChemPhysChem 2008, 9, 2340.
[47] R. Bhattacharyya, A. Kumar, Chem. Phys. Lett. 2004, 383, 99.
[48] N. M. Loening, M. J. Thrippleton, J. Keeler, R. G. Griffin, J. Magn.
Reson. 2003, 164, 321.
[49] B. Shapira, L. Frydman, J. Magn. Reson. 2003, 165, 320.
[50] R. Panek, J. Granwehr, J. Leggett, W. Kockenberger, Phys. Chem.
Chem. Phys. 2010, 12, 5771.
[51] M. Thaning, U.S. Patent 6,013,810, 2000.
Received: September 24, 2010
Published online: November 9, 2010
Chem. Eur. J. 2011, 17, 697 – 703
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
703