Article
pubs.acs.org/jmc
Ligand−Protein Affinity Studies Using Long-Lived States of Fluorine19 Nuclei
Roberto Buratto,*,† Daniele Mammoli,† Estel Canet,†,‡,§,∥ and Geoffrey Bodenhausen†,‡,§,∥
†
Institut des Sciences et Ingénierie Chimiques, Ecole Polytechnique Fédérale de Lausanne, 1015 Lausanne, Switzerland
Département de Chimie, Ecole Normale Supérieure−PSL Research University, 24 Rue Lhomond, 75231 Paris Cedex 05, France
§
Sorbonne Université, UPMC Univ Paris 06, 4 place Jussieu, 75005 Paris, France
∥
CNRS, UMR 7203 LBM, 75005 Paris, France
‡
ABSTRACT: The lifetimes TLLS of long-lived states or TLLC of
long-lived coherences can be used for the accurate determination of
dissociation constants of weak protein−ligand complexes. The
remarkable contrast between signals derived from LLS or LLC in
free and bound ligands can be exploited to search for weak binders
with large dissociation constants KD > 1 mM that are important for
fragment-based drug discovery but may escape detection by other
screening techniques. Alternatively, the high sensitivity of the
proposed method can be exploited to work with large ligand-toprotein ratios, with an evident advantage of reduced consumption
of precious proteins. The detection of 19F−19F long-lived states in
suitably designed fluorinated spy molecules allows one to perform
competition binding experiments with high sensitivity while
avoiding signal overlap that tends to hamper the interpretation of
proton spectra of mixtures.
■
INTRODUCTION
Fragment-based drug discovery (FBDD) has emerged as a
fruitful approach to develop new drugs.1 Initially, one must
identify small molecular fragments that bind weakly to a
macromolecular target with dissociation constants KD on the
order of 10 μM to 10 mM or greater.2 These ligands can be
subsequently developed through medicinal chemistry in order
to optimize features such as absorption, distribution, metabolism, excretion, and toxicological (ADMET) properties.
Nuclear magnetic resonance (NMR) seems particularly suitable
to reveal such weak interactions. A plethora of NMR
experiments has been developed for screening fragment
libraries, such as WaterLOGSY,3 saturation transfer difference
(STD),4 and relaxation-edited experiments. The latter can
exploit the contrast of longitudinal or transverse relaxation
upon binding.5 Experiments that monitor the relaxation of
long-lived states (LLS)6,7 have been demonstrated to be
particularly sensitive to binding phenomena.8,9 The immunity
of LLS to dipolar relaxation between the two participating
nuclei explains the dramatic contrast
CLLS =
obs
free
RLLS
− RLLS
obs
RLLS
easier it is to screen weakly binding fragments and to determine
their dissociation constants.10
Experiments based on the direct observation of ligands suffer
from a number of limitations: nonspecific binders may give
similar effects as specific ones, ligands are difficult to detect if
their solubility is low, and strong ligands in slow exchange are
easily mistaken for nonbinders. To overcome these drawbacks,
Dalvit and co-workers11 introduced so-called competition
experiments for ligand screening. In this approach, a weakaffinity ligand is used as a spy molecule; a stronger binder
displaces the spy molecule, and the latter’s expulsion affects the
relaxation rates of nuclei on the spy. The concentration of a
competitor that is required to displace a spy molecule is
inversely proportional to the former’s affinity for the macromolecular target: the higher the affinity, the lower the
concentration needed. The study of weakly binding fragments
turns out to be challenging since high concentrations and
therefore high solubility are required. Moreover, if mixtures of
potential competitors are tested, the risk of signal overlap must
be sidestepped by a careful choice of the cocktail of molecules.
This work demonstrates that the excitation of LLS involving
pairs of 19F nuclei in spy ligands that have been designed to
feature a favorable contrast CLLS between free and bound forms
provides a very effective tool to study weak protein−ligand
interactions. Such experiments benefit from a good sensitivity
100[%]
(1)
that is obtained if the observed relaxation rate Robs
LLS is compared
with the rate Rfree
LLS of the ligand in its free state. The contrast
CLLS is often larger for LLS than for other relaxation rates such
as T1selective, T2, and T1ρ. The more favorable the contrast, the
© 2016 American Chemical Society
Received: October 9, 2015
Published: January 22, 2016
1960
DOI: 10.1021/acs.jmedchem.5b01583
J. Med. Chem. 2016, 59, 1960−1966
Journal of Medicinal Chemistry
Article
since 19F has a high gyromagnetic ratio and 100% natural
abundance, and since 19F spectra do not suffer from signal
overlap.
Long-lived states6,12−14 decay with a time constant TLLS that
can be much longer than the longitudinal relaxation time
constant T1 under suitable conditions. Likewise, long-lived
coherences (LLC)15 relax with a time constant TLLC that can be
longer than the transverse relaxation time constant T2. In
systems with two coupled spins-1/2, singlet states |S0⟩
correspond to antisymmetric linear combinations of the two
product states |αβ⟩ and |βα⟩. If a “triplet−singlet population
imbalance”, or TSI,16 is created, it can only be dissipated in the
presence of chemical shift anisotropy (CSA), dipolar
interactions with external spins, or dipolar interactions to
paramagnetic species.
To perform an LLS experiment, it is convenient to start with
a weakly coupled two-spin AX system, temporarily suppress the
chemical shift difference by radio frequency (RF) irradiation to
obtain an A2-like spin system, and eventually revert to the AX
system by interrupting the RF irradiation. In the experimental
procedure that we commonly use,7 (i) one converts the
Boltzmann equilibrium populations of an AX system into an
LLS, which amounts to setting up a triplet−singlet imbalance
(Figure 1 a-b);6,7 (ii) one sustains the LLS in an interval τm by
Figure 2. Pulse sequence for excitation, locking, and observation of
long-lived coherences (LLC) involving a pair of 19F nuclei. As in
Figure 1, proton decoupling can be used during signal observation to
remove J(1H, 19F) splittings if desired.
with the carrier νRF halfway between the chemical shifts of the
two spins to obtain two magnetically equivalent spins A2 in the
interval τm (Figure 2b,c). Finally, after interrupting the RF
irradiation, the system reverts to a magnetically inequivalent AX
configuration, and the signal is acquired after point c in Figure
2. At this point, the density operator can be expressed as a
function of the duration τm of the RF field:15
σ(τm) = [(Ix − Sx) cos(2πJIS τm) + (2IySz − 2IzSy)
⎛ τ ⎞
sin(2πJIS τm)] exp⎜ − m ⎟
⎝ TLLC ⎠
(2)
Thus, the terms Ix − Sx give oscillatory signals that slowly
decay as a function of the interval τm.
The J-modulation can be suppressed by inserting a π/2 pulse
at the midpoint of a double spin echo in the interval τm, so that
the LLC decay can be fitted to a simple exponential function.19
In either case, the LLC contrast is
CLLC =
Figure 1. Pulse sequence for excitation, locking, and observation of
long-lived states (LLS) involving a pair of 19F nuclei. Note that it is
sufficient to apply proton decoupling during signal observation in
order to remove J(1H, 19F) splittings if desired.
■
obs
free
RLLC
− RLLC
obs
RLLC
100[%]
(3)
RESULTS AND DISCUSSION
To the best of our knowledge, long-lived states have so far only
been observed in systems comprising pairs of 1H, 13C or 15N
nuclei.20 Here we explore the possibility of observing LLS and
LLC using pairs of 19F nuclei. We have synthesized a molecule
that contains a pair of diastereotopic aliphatic fluorine atoms:
1,1-difluoro-1-phenylacetyl-Gly-Arg, abbreviated as DFPA-GR
(Figure 3). The presence of an arginine residue gives rise to a
applying a resonant RF field with the carrier (νRF) placed
halfway between the two chemical shifts6,14,17,18 (Figure 1b,c);
(iii) finally, after turning off the RF field, one applies a suitable
pulse sequence to convert the remaining triplet−singlet
imbalance TSI (which is equivalent to an LLS) back into
observable magnetization (Figure 1c,d). The lifetime TLLS of a
long-lived state can be determined by fitting the signal intensity
as a function of the locking interval τm (Figure 1b,c).
LLC15 are closely related to LLS. Their lifetimes TLLC can be
longer than transverse relaxation times T2 and are also strongly
affected by interactions with proteins. Therefore, TLLC can also
be a useful observable for protein−ligand binding studies. LLC
oscillate in the interval τm with the scalar coupling constant
J(19FI, 19FS). The Fourier transform of this oscillatory decay
yields J-spectra with narrow line widths (ΔνLLC = 1/(πTLLC)).
In the singlet−triplet basis set, an LLC can be defined as |S0⟩
⟨T0| + |T0⟩⟨S0|. This is equivalent to a linear combination of Ix
− Sx and 2IySz − 2IzSy in the Cartesian product operator basis.
An LLC experiment comprises the same three intervals for
excitation, locking, and detection. First, the Boltzmann
equilibrium population of a magnetically inequivalent AX spin
system is excited to yield a spin density operator corresponding
to Ix − Sx (Figure 2a). Then, the chemical shift difference
between the two spins is temporarily suppressed by a RF field,
Figure 3. Structure of the custom-synthesized spy ligand 1,1-difluoro1-phenylacetyl-Gly-Arg (DFPA-GR) containing a pair of aliphatic
diastereotopic fluorine atoms.
weak affinity for the active site of trypsin, thus allowing one to
explore the behavior of LLS and LLC of pairs of 19F nuclei
upon binding. Aliphatic fluorine atoms have been preferred to
aromatic ones, in order to minimize the relaxation of LLS due
to chemical shift anisotropy (CSA.) The CSAs of 19F nuclei in
1961
DOI: 10.1021/acs.jmedchem.5b01583
J. Med. Chem. 2016, 59, 1960−1966
Journal of Medicinal Chemistry
Article
LLC relaxation rate is RLLC = 2.30 s−1, while R2 = 2.61 and 2.65
s−1, respectively, for the two diastereotopic fluorine nuclei, so
that the ratios are R2/RLLC = 1.13 and 1.15 for these two nuclei,
i.e., less favorable than the ratio R1/RLLS > 4. Furthermore, the
rates RLLC are sensitive to binding. If the experiment is
performed in a solution of 370 μM DFPA-GR and 2 μM
trypsin (185-fold excess), RLLC increases to 4.55 s−1, leading to
a respectable contrast CLLC = 49%, which is however far below
the remarkable contrast CLLS = 87% under the same conditions.
In order to determine the dissociation constant KD of the
DFPA-GR/trypsin complex, a titration experiment was
performed to monitor TLLS = 1/RLLS, while stepwise increasing
the ligand concentration [Ltot] with an initial concentration of 2
μM trypsin. The addition of small aliquots of a concentrated
solution of DFPA-GR allows one to increase [Ltot], while the
protein concentration is hardly affected. The titration curve was
fitted to the following equation:
aliphatic positions in amino acids are smaller than those of 19F
nuclei in aromatic rings.21,22
No less than seven bonds separate the fluorine nuclei from
the closest chiral center (Cα of the arginine residue, indicated
by a star in Figure 3). The chemical shift difference ΔυI,S
between the two diastereotopic fluorine nuclei is only 0.8 ppm,
or 300 Hz, at B0 = 9.4 T (400 and 376 MHz for 1H and 19F,
respectively), while the scalar coupling constant is 2JI,S = 252
Hz. Since an RF field amplitude υ1 = 5ΔυI,S is generally
sufficient to sustain LLS,8 a weak field υ1 = 1.5 kHz allows one
to achieve TLLS = 2.63 s in DFPA-GR at B0 = 9.40 T.
In general, ΔυIS can be reduced by increasing the distance
between the fluorine nuclei and the chiral center. Conversely,
when the glycine residue in DFPA-GR is deleted, one observes
an increase of the chemical shift difference between the two
diastereotopic fluorine nuclei to ΔυIS = 2.05 ppm, thus
requiring an increased RF field amplitude of about υ1 = 3.5
kHz to lock the LLS. The resulting TLLS = 2.06 s is slightly
shorter than the one in DFPA-GR. On the other hand, the
insertion of an additional Gly residue would lead to a further
reduction of the chemical shift difference ΔυIS.
The excitation of LLS for the pair of fluorine nuclei of DFPAGR was achieved with the pulse sequence shown in Figure 1. In
the absence of protein, the LLS relaxation rate of the ligand was
found to be RLLS = 0.38 s−1, much smaller than the longitudinal
relaxation rate R1 = 1.64 s−1, leading to a favorable ratio R1/RLLS
> 4. This shows that it is possible to obtain LLS with
sufficiently long lifetimes for pairs of fluorine nuclei. Attempts
to use pairs of aromatic 19F nuclei lead to smaller ratios R1/RLLS
that are less favorable for drug screening.
Of all parameters that can be exploited in 19F NMR, the
transverse relaxation rate R2(19F) = 1/T2(19F) is widely
considered to be one of the most sensitive to binding
phenomena because R2(19F) is strongly affected by exchange
broadening.23 The rates R2(19F) of our spy ligand DFPA-GR
have been determined in the presence or absence of trypsin.
The resulting R2(19F) contrast lies in the range 32 < C2 < 40%
for the two diastereotopic 19F nuclei for 370 μM DFPA-GR
with 2 μM trypsin (i.e., for a 185-fold excess). However, if we
switch our attention to LLS, the contrast CLLS increases to 87%
under the same conditions (Table 1). This confirms that RLLS is
extremely sensitive to binding.
LLC can also be readily excited and sustained in DFPA-GR.
Using the pulse sequence introduced by Singh and Kurur,19 the
LLC decay could be fitted to a monoexponential function. The
obs
RLLS
=
Rbound
LLS
R2
(s−1)
RLLC (s−1)
RLLS (s−1)
19 pro‑R
F
19 pro‑S
F
370 μM DFPA-GR
+ 2 μM trypsin
contrast
2.61
2.65
2.30
0.38
3.85
4.39
4.55
2.85
C2 = 32%
C2 = 40%
CLLC = 49%
CLLS = 87%
(4)
Rfree
LLS
where
and
are the LLS relaxation rates of the ligand
in its fully bound and free forms, respectively. The molar
fraction [PL]/[L]tot of the bound form can be calculated from
[Ptot ] + [Ltot ] + KD −
[PL]
=
[Ltot ]
([Ptot ] + [Ltot ] + KD)2 − 4[Ptot ][Ltot ]
2[Ltot ]
(5)
The dissociation constant of the DFPA-GR/trypsin complex
that can be derived from the titration curve (Figure 4) is KD =
Table 1. Transverse Relaxation Rates R2(19F) of the Two
Diastereotopic 19F Nuclei, Relaxation Rate RLLC(19F) of the
Long-Lived Coherences and Relaxation Rate RLLS(19F) of the
Long-Lived States of the Spy Molecule DFPA-GR of Figure
3, and Contrast Observed in the Presence or Absence of
Trypsina
370 μM
DFPA-GR
[PL] bound
free
free
(RLLS − RLLS
) + RLLS
[L]tot
Figure 4. Lifetimes TLLS of the long-lived state LLS associated with the
two 19F nuclei of the spy ligand DFPA-GR in the presence of 2 μM
trypsin as a function of the concentration [DFPA-GR] at 298 K and
9.4 T. The curve shows a fit of the experimental data to eq 4.
106 ± 26 μM. The thermodynamic constant KD must be equal
to the ratio of the kinetic dissociation and association rate
constants, KD = koff/kon. Since the latter is usually assumed to be
limited by diffusion (107 < kon< 109 M−1 s−1),24 koff must be in
the range 103−105 s−1. Thus, the system easily fulfills the fast
exchange condition, i.e., kex = (kon[P] + koff) ≈ kof f ≫ Δω,
where [P] is the concentration of the free protein and Δω =
2πΔν is the difference in 19F chemical shifts between the bound
and free forms of the ligand. Even if the assumption that kon is
a
The selective longitudinal relaxation rates R1,s (19F) could not be
determined because of the difficulty of selectively inverting only one of
the two spins.
1962
DOI: 10.1021/acs.jmedchem.5b01583
J. Med. Chem. 2016, 59, 1960−1966
Journal of Medicinal Chemistry
Article
Figure 5. (Top) Signals derived from 19F−19F LLS of 500 μM DPFA-GR without protein after τm = 0.7 s. (Center) The same in the presence of 2
μM trypsin. (Bottom) The same in the presence of 2 μM trypsin and 485 μM morin as competitor, which partly displaces DPFA-GR from the
binding site of the protein, leading to a partial restoration of its signals. A total of 128 scans were recorded for each spectrum, with acquisition and
repetition times of 0.7 and 3 s.
Figure 6. (Left, black dots) LLS decays of 495 μM DFPA-GR in the absence of protein and competitor; (blue dashed line) in the presence of 2 μM
trypsin and 500 μM morin; (green continuous line) in the presence of 2 μM trypsin and 500 μM BT-GGR; and (red dashed-dotted line) in the
presence of 2 μM trypsin only. (Right) 19F spectra derived from 19F−19F LLS of the spy DFPA-GR in the four solutions described above, after
locking for τm = 0.7 s.
rates of the spy molecule. The competing ligand leads to a
partial expulsion of the spy molecule and hence to a change in
the molar fraction Xbound
spy . The larger the contrast CLLS, the
that can be detected and the
smaller the changes of Xbound
spy
higher the sensitivity to competitive binding.
Figure 5 shows the LLS spectrum of 500 μM DFPA-GR in
the absence of protein (top), in the presence of 2 μM trypsin
(center) and with 2 μM trypsin plus 485 μM morin, a known
trypsin inhibitor25 (bottom). In the absence of protein the
signal is intense because the LLS relaxation rate RLLS is small,
while in the presence of protein RLLS increases so that the
signals are attenuated. In the presence of both the protein and a
competing molecule, the competitor hinders the interaction
between the spy molecule and the protein, thus leading to a
decrease of RLLS. Nevertheless, since the inhibitor never
completely displaces the spy molecule from the active site of
the protein, a small fraction of the spy molecule remains in
exchange with the protein, so that the signal is not completely
restored. Note that the relative intensities of the four signals are
determined by diffusion were violated, additional experimental
evidence demonstrates that the DFPA-GR/trypsin system is in
fast exchange. For example, the relaxation rates RLLS in Figure 4
show a trend typical of a weighted average of the rates of free
and bound states. In the presence of a system in slow exchange,
the observed relaxation rate would not change during a titration
since only the signals of the ligand in the free state would be
observable (the small fraction of ligand in the bound state
would not contribute to the average rate).
In practice, it may be challenging to identify a spy ligand that
contains a pair of 19F nuclei and fulfills these conditions. In
most cases, one should begin by selecting a spy molecule with
suitable exchange rates, and then insert a 19F spin pair by
synthesis, as we have demonstrated for spin-pair labels
containing two protons.9
Once a suitable spy molecule such as DFPA-GR that fulfills
the conditions of fast exchange has been identified, libraries of
potential binders can be screened by competition binding
experiments,11 by observing changes in the LLS or LLC decay
1963
DOI: 10.1021/acs.jmedchem.5b01583
J. Med. Chem. 2016, 59, 1960−1966
Journal of Medicinal Chemistry
Article
Figure 7. (Top) Proton spectrum of 495 μM DFPA-GR (spy). (Bottom) Spectrum of 495 μM DFPA-GR mixed with 500 μM BT-GGR. The
rectangle shows the range where the signals of the α-protons of the glycine residues appear.
If one would observe proton signals, such experiments would
need a careful choice of the mixtures in order to avoid overlap
of signals of the ligands with those of the spy molecule. For
example, Figure 7 shows a comparison between the 1H spectra
of the spy molecule DFPA-GR alone and mixed with BT-GGR,
a competitor with a similar molecular structure. The overlaps
between the proton signals of the spy molecule and those of the
competitor are too severe to allow one to perform reliable
competition experiments.
The problem can become even more severe when looking for
weak binders. Indeed, the larger the dissociation constants, the
higher the required concentrations of the competitors. In this
context, 19F NMR has no rivals. It allows one to perform
experiments with high sensitivity while avoiding problems of
overlap with protonated buffers and signals of mixtures. The
combination of the high sensitivity to binding phenomena
offered by the LLS method and the lack of overlap in 19F NMR
can be put to good use for fragment-based drug discovery.
Since competition experiments are based on indirect detection
of competing ligands through changes in the properties of the
spy molecule, a deconvolution step may be needed, where
mixtures containing fewer or single compounds are tested
separately.
different in the presence or absence of the protein. This effect is
likely due to cross-correlated relaxation interference amplified
by the slow motion regime of the ligand in the bound state.
Once a competitor has been identified, its dissociation
constant needs to be determined. To do so, the spy ligand
DFPA-GR can be titrated in the presence of a fixed
concentration of competitor or vice versa.11 The curve of TLLS
= (1/RLLS) vs [spy], respectively, vs [competitor] can be fitted
using eq 4 to determine an apparent dissociation constant
spy
competitor
KD,
can be
app . The true dissociation constant KD
calculated from the relationship:
K Dcompetitor =
[Lcompetitor ]KDspy, true
KDspy, app − KDspy, true
(6)
where Kspy
D, true = 106 μM is the true dissociation constant of the
spy ligand DFPA-GR and [Lcompetitor] is the total concentration
of the competitor. A quick estimate of Kcompetitor
can be obtained
D
from a single titration point: knowing Robs
LLS, the mole fraction of
the spy molecule in its bound form can be estimated using eq 4.
At this point, Kspy
D, app can be calculated by rearranging eq 5.
Following this approach, we have estimated Kcompetitor
= 28 μM
D
for morin and Kcompetitor
= 250 μM for BT-GGR, in reasonable
D
agreement with values reported in the literature (30 and 200
μM).9,25 This shows that estimates of the affinities of
competitors can be obtained quickly. A more accurate
determination of Kcompetitor
requires a full titration experiment.
D
The results demonstrate the applicability of the method over
a broad range of competitor affinities. Morin can be considered
as a typical medium-affinity binder, while BT-GGR can be
considered as a representative of binding fragments, with a
weak affinity for trypsin and a low molecular weight
(MWBT−GGR = 385 Da) (Figure 6).
Often, screening is performed by testing mixtures (also
known as “cocktails”) of 3−10 putative competitors in order to
reduce experimental time and minimize protein consumption.
■
CONCLUSIONS
The lifetimes TLLS of long-lived states and, to a lesser extent,
the lifetimes TLLC of long-lived coherences can be used very
effectively to characterize protein−ligand interactions.8 In the
context of fragment-based drug discovery, the exquisite
sensitivity to ligand−protein binding, as evidenced by a
favorable contrast CLLS or CLLC, can be exploited to search
for weak binders with large dissociation constants Kcompetitor
>1
D
mM, which may escape detection by other screening
techniques.10 Alternatively, the high sensitivity of the proposed
method can be exploited to work with large ligand-to-protein
ratios, with an evident advantage in terms of protein saving. By
1964
DOI: 10.1021/acs.jmedchem.5b01583
J. Med. Chem. 2016, 59, 1960−1966
Journal of Medicinal Chemistry
Article
TBTU (4 equiv, 4 mmol, 1.29 g), and DIPEA (4 equiv, 4 mmol, 0.52
g) were dissolved in DMF (10 mL). The solution was added to the Ndeprotected resin. The reaction mixture was shaken at rt for 3 h at 125
rpm to give Fmoc-Gly-Arg(Pbf)-O-resin (3), which was washed with
DMF (4 × 10 mL).
DFPA-Gly-Arg(Pbf)-O-resin (4). Fmoc-Gly-Arg(Pbf)-O-resin (3)
was suspended in a solution of 20% piperidine in DMF (30 mL) for 30
min and shaken at 125 rpm to give the N-deprotected resin. The resin
was washed with DMF (4 × 10 mL) and DCM (4 × 10 mL).
Difluoro-phenyl-acetic acid (4 equiv, 4 mmol, 0.69 g), HOBt (4 equiv,
4 mmol, 0.54 g), TBTU (4 equiv, 4 mmol, 1.29 g), and DIPEA (4
equiv, 4 mmol, 0.52 g) were dissolved in DMF (30 mL). The solution
was added to the N-deprotected resin. The reaction mixture was
shaken at rt for 3 h at 125 rpm to give DFPA-Gly-Arg(Pbf)-O-resin
(4), which was washed with DMF (4 × 10 mL) and DCM (4 × 10
mL).
DFPA-Gly-Arg (5). Cleavage from the resin and deprotection of
the side chain group was carried out with 10 mL of TFA/H2O/TIS
(95:2.5:2.5) for 3 h. TFA was then removed by evaporation, and the
final product (5) was obtained after lyophilization.
combining LLS experiments with dissolution DNP so that the
concentrations of both ligands and proteins can be drastically
reduced, one can alleviate solubility problems that may arise
when high concentrations of competitors are required to
displace a spy molecule from its binding site.9
We have shown for the first time that long-lived states and
coherences can be observed using pairs of 19F nuclei.
Fluorinated spy molecules allow one to avoid signal overlap
with resonances of buffers and competitors, so that one can
increase the number of putative ligands that can be tested
simultaneously, thus speeding up the entire screening process.
Competition experiments allow one to detect the binding of
arbitrary molecules that do not contain any fluorine nuclei. The
affinity of competing ligands can be estimated quickly without
performing a full titration. The new method should facilitate
the identification of lead compounds that can be optimized in
later stages of drug design.
■
■
EXPERIMENTAL SECTION
All solutions for screening and titration experiments were prepared
with 20 mM TRIS buffer, 150 mM NaCl, and 5 mM MgCl2. Stock
solutions were prepared in d6-DMSO containing 200 mM BT-GGR,9
200 mM morin (Sigma-Aldrich), and 200 mM DFPA-R or 200 mM
DFPA-GR. All experiments were performed using type IX-S trypsin
from porcine pancreas (Sigma-Aldrich). During the titrations, 2 μL
aliquots of 150 mM DFPA-GR were added to 500 μL of the starting
solution. The ligand concentrations were measured by the PULCON
technique.26 The rates RLLS and RLLC were obtained by fitting the
decays with monoexponentially decaying functions, using 10 delays τm
ranging from 10 ms to 5 times the expected relaxation time TLLS. The
R2 measurements were recorded using a CPMG sequence with a π/2
pulse at the top of an echo to suppress J-modulations in the manner of
PROJECT,27 combined with continuous-wave decoupling of the
protons during acquisition, but not during the echo sequence, to avoid
interference. The spin−echo de- and refocusing delays were τ = 20 ms.
Titration curves were fitted to eq 2. In the presence of a competitor,
the apparent dissociation constant of the spy molecule was used as
input to eq 4 to calculate the true dissociation constant of the
competitor. All NMR measurements were performed at 25 °C and B0
= 9.4 T where 19F and 1H nuclei resonate near 376.38 and 400 MHz,
respectively.
The synthesis of 1,1-difluoro-1-phenylacetyl-Gly-Arg (DFPA-GR)
was performed by solid-phase peptide synthesis (SPPS) using 2chlorotrityl chloride resin and Fmoc-protected amino acids. The first
step is a SN1 substitution of Fmoc-Arg(Pbf)-OH on the resin. All
remaining reactive 2-chlorotrityl groups were then capped with
MeOH. The product was obtained by coupling of Fmoc-protected
Gly in the presence of HOBt and TBTU, followed by deprotection of
the N-terminus of the dipeptide. Finally, difluoro-phenyl-acetic acid
was conjugated at the N-terminus of the dipeptide. Cleavage from the
resin, followed by deprotection of the arginine side chain, afforded
DFPA-GR.
N-Fmoc-Arg(Pbf)-O-resin (2). After swelling with dry DCM (80
mL) for 5 min, the 2-chlorotrityl chloride resin (0.83 mmol·g−1, 1
equiv, 1 mmol, 1.2 g) was treated with a solution of Fmoc-Arg(Pbf)OH (1) (1.2 equiv, 1.2 mmol, 0.78 g) in dry DCM (10 mL) and
DIPEA (2.5 equiv, 6.23 mmol, 0.81 g) shaken at 125 rpm at rt for 2 h.
The reaction was performed in 4 × 10 mL filtration tubes with a
polyethylene frit. MeOH (10 mL) was added to cap the free sites, and
the reaction mixture was shaken for 1 h at 125 rpm. The resin was
washed with DCM (3 × 12 mL), DCM/MeOH 1:1 (3 × 12 mL),
MeOH (3 × 12 mL), and diethyl ether (3 × 12 mL) and dried for 12 h
in vacuo to give the N-Fmoc-Arg(Pbf)-O-resin (2).
N-Fmoc-Gly-Arg(Pbf)-O-resin (3). N-Fmoc-Arg(Pbf)-O-resin (2)
was suspended in a solution of 20% piperidine in DMF (10 mL) for 30
min and shaken at 125 rpm to give the N-deprotected resin. The resin
was washed with DMF (4 × 10 mL) and DCM (4 × 10 mL). FmocGly-OH (4 equiv, 4 mmol, 1.19 g), HOBt (4 equiv, 4 mmol, 0.54 g),
AUTHOR INFORMATION
Corresponding Author
*Tel: (+41) 21 69 39428. E-mail: [email protected].
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
The authors are grateful to Dr. Ranganath Gopalakrishnan and
Prof. Christian Heinis for helping in the synthesis of 19Flabelled molecules, Dr. Shivdas Sangram for helpful discussions,
and Dr. Pascal Miéville for valuable assistance. This work was
supported by the Swiss National Science Foundation (SNSF),
the Swiss Commission for Technology and Innovation (CTI),
the EPFL, the CNRS, and the European Research Council
(ERC project “Dilute para-water”).
■
ABBREVIATIONS USED
ADMET, absorption, distribution, metabolism, excretion, and
toxicological properties; CPMG, Car−Purcell−Meiboom−Gill
spin echo sequence; CSA, chemical shift anisotropy; DFPA-GR,
1,1-difluoro-1-phenylacetyl-Gly-Arg; FBDD, fragment-based
drug discovery; LLC, long-lived coherences; LLS, long-lived
states; PULCON, pulse length based concentration determination; STD, saturation transfer difference; TSI, triplet-singlet
imbalance; Water-LOGSY, water-ligand observed via gradient
spectroscopy
■
REFERENCES
(1) Baker, M. Fragment-based lead discovery grows up. Nat. Rev.
Drug Discovery 2013, 12, 5−10.
(2) Hubbard, R. E. Structure-Based Drug Discovery: An Overview;
Royal Society of Chemistry: Cambridge, 2006.
(3) Dalvit, C.; Pevarello, P.; Tato, M.; Veronesi, M.; Vulpetti, A.;
Sundstrom, M. Identification of compounds with binding affinity to
proteins via magnetization transfer from bulk water. J. Biomol. NMR
2000, 18, 65−68.
(4) Mayer, M.; Meyer, B. Characterization of ligand binding by
saturation transfer difference NMR spectroscopy. Angew. Chem., Int.
Ed. 1999, 38, 1784−1788.
(5) Hajduk, P. J.; Olejniczak, E. T.; Fesik, S. W. One-dimensional
relaxation- and diffusion-edited NMR methods for screening
compounds that bind to macromolecules. J. Am. Chem. Soc. 1997,
119, 12257−12261.
(6) Carravetta, M.; Levitt, M. H. Long-lived nuclear spin states in
high-field solution NMR. J. Am. Chem. Soc. 2004, 126, 6228−6229.
1965
DOI: 10.1021/acs.jmedchem.5b01583
J. Med. Chem. 2016, 59, 1960−1966
Journal of Medicinal Chemistry
Article
(7) Sarkar, R.; Vasos, P. R.; Bodenhausen, G. Singlet-state exchange
NMR spectroscopy for the study of very slow dynamic processes. J.
Am. Chem. Soc. 2007, 129, 328−334.
(8) Salvi, N.; Buratto, R.; Bornet, A.; Ulzega, S.; Rebollo, I. R.;
Angelini, A.; Heinis, C.; Bodenhausen, G. Boosting the sensitivity of
ligand-protein screening by NMR of long-lived states. J. Am. Chem. Soc.
2012, 134, 11076−11079.
(9) Buratto, R.; Bornet, A.; Milani, J.; Mammoli, D.; Vuichoud, B.;
Salvi, N.; Singh, M.; Laguerre, A.; Passemard, S.; Gerber-Lemaire, S.;
Jannin, S.; Bodenhausen, G. Drug screening boosted by hyperpolarized
long-lived states in NMR. ChemMedChem 2014, 9, 2509−2515.
(10) Buratto, R.; Mammoli, D.; Chiarparin, E.; Williams, G.;
Bodenhausen, G. Exploring weak ligand-protein interactions by longlived NMR states: improved contrast in fragment-based drug
screening. Angew. Chem., Int. Ed. 2014, 53, 11376−11380.
(11) Dalvit, C.; Flocco, M.; Knapp, S.; Mostardini, M.; Perego, R.;
Stockman, B. J.; Veronesi, M.; Varasi, M. High-throughput NMRbased screening with competition binding experiments. J. Am. Chem.
Soc. 2002, 124, 7702−7709.
(12) Levitt, M. H. Singlet and other states with extended lifetimes.
eMagRes. 2010, DOI: 10.1002/9780470034590.emrstm1036.
(13) Pileio, G. Relaxation theory of nuclear singlet states in two spin1/2 systems. Prog. Nucl. Magn. Reson. Spectrosc. 2010, 56, 217−231.
(14) Pileio, G.; Levitt, M. H. Theory of long-lived nuclear spin states
in solution nuclear magnetic resonance. II. Singlet spin locking. J.
Chem. Phys. 2009, 130, 214501.
(15) Sarkar, R.; Ahuja, P.; Vasos, P. R.; Bornet, A.; Wagnieres, O.;
Bodenhausen, G. Long-lived coherences for line-narrowing in highfield NMR. Prog. Nucl. Magn. Reson. Spectrosc. 2011, 59, 83−90.
(16) Meier, B.; Dumez, J. N.; Stevanato, G.; Hill-Cousins, J. T.; Roy,
S. S.; Hakansson, P.; Mamone, S.; Brown, R. C. D.; Pileio, G.; Levitt,
M. H. Long-lived nuclear spin states in methyl groups and quantumrotor-induced polarization. J. Am. Chem. Soc. 2013, 135, 18746−
18749.
(17) Gopalakrishnan, K.; Bodenhausen, G. Lifetimes of the singletstates under coherent off-resonance irradiation in NMR spectroscopy.
J. Magn. Reson. 2006, 182, 254−259.
(18) Sarkar, R.; Ahuia, P.; Moskau, D.; Vasos, P. R.; Bodenhausen, G.
Extending the scope of singlet-state spectroscopy. ChemPhysChem
2007, 8, 2652−2656.
(19) Singh, M.; Kurur, N. D. An improved method for the
measurement of lifetimes of long-lived coherences in NMR. RSC
Adv. 2015, 5, 8236−8238.
(20) Pileio, G.; Carravetta, M.; Levitt, M. H. Storage of nuclear
magnetization as long-lived singlet order in low magnetic field. Proc.
Natl. Acad. Sci. U. S. A. 2010, 107, 17135−17139.
(21) Durr, U. H. N.; Grage, S. L.; Witter, R.; Ulrich, A. S. Solid state
F-19 NMR parameters of fluorine-labeled amino acids. Part I:
Aromatic substituents. J. Magn. Reson. 2008, 191, 7−15.
(22) Grage, S. L.; Durr, U. H. N.; Afonin, S.; Mikhailiuk, P. K.;
Komarov, I. V.; Ulrich, A. S. Solid state F-19 NMR parameters of
fluorine-labeled amino acids. Part II: Aliphatic substituents. J. Magn.
Reson. 2008, 191, 16−23.
(23) Dalvit, C. Ligand- and substrate-based 19F NMR screening:
Principles and applications to drug discovery. Prog. Nucl. Magn. Reson.
Spectrosc. 2007, 51, 243−271.
(24) Lepre, C. A.; Moore, J. M.; Peng, J. W. Theory and applications
of NMR-based screening in pharmaceutical research. Chem. Rev. 2004,
104, 3641−3675.
(25) Checa, A.; Ortiz, A. R.; de Pascual-Teresa, B.; Gago, F.
Assessment of solvation effects on calculated binding affinity
differences: Trypsin inhibition by flavonoids as a model system for
congeneric series. J. Med. Chem. 1997, 40, 4136−4145.
(26) Wider, G.; Dreier, L. Measuring protein concentrations by
NMR spectroscopy. J. Am. Chem. Soc. 2006, 128, 2571−2576.
(27) Aguilar, J. A.; Nilsson, M.; Bodenhausen, G.; Morris, G. A. Spin
echo NMR spectra without J modulation. Chem. Commun. 2012, 48,
811−813.
1966
DOI: 10.1021/acs.jmedchem.5b01583
J. Med. Chem. 2016, 59, 1960−1966