NMR screening applied to the fragment-based generation of inhibitors of creatine kinase exploiting a new interaction proximate to the ATP binding site

Article abstract of DOI:10.1021/jm061460r

Using an in-house fragment NMR library, we identified a set of ligands that bind rabbit muscular creatine kinase, an enzyme involved in key ATP-dependent processes. The ligands docked to the crystal structures of creatine kinase indicated that a phenylfur

Full text of DOI:10.1021/jm061460r

J. Med. Chem. 2007, 50, 1865-1875
1865
NMR Screening Applied to the Fragment-Based Generation of Inhibitors of Creatine Kinase
Exploiting a New Interaction Proximate to the ATP Binding Site
Anne-Sophie Bretonnet,† Anne Jochum,‡ Olivier Walker,† Isabelle Krimm,† Peter Goekjian,‡ Olivier Marcillat,§ and
Jean-Marc Lancelin*,†
Laboratoire de RMN et Spectrome´trie de Masse Biomole´culaires, UniVersite´ Claude Bernard Lyon 1, UMR CNRS 5180 Sciences Analytiques,
ESCPE Lyon, Laboratoire de Chimie Organique II-Glycochimie, UniVersite´ Claude Bernard Lyon 1, UMR CNRS 5181 Me´thodologie de
Synthe`se et Mole´cules Bio-ActiVes, ESCPE Lyon, and Biomembranes et Enzymes Associe´s, UMR CNRS 5246, UniVersite´ Claude Bernard
Lyon 1, 69622 Villeurbanne, France
ReceiVed December 21, 2006
Using an in-house fragment NMR library, we identified a set of ligands that bind rabbit muscular creatine
kinase, an enzyme involved in key ATP-dependent processes. The ligands docked to the crystal structures
of creatine kinase indicated that a phenylfuroic acid could enter into a pocket adjacent to the nucleotide
binding site. This fragment served as an anchor to develop in silico a series of potential inhibitors which
could partly access the nucleotide binding site. The short synthesis of only four derivatives provided entirely
novel hit compounds that reversibly inhibit creatine kinase at micromolar concentrations with a mixed ATP-
competitive/noncompetitive mechanism in agreement with the structural model of the inhibited enzyme.
These initial biologically active compounds are novel and modular and exploit a new interaction proximate
to the ATP binding site.
Scheme 1. Interconversion of Creatine to Phosphocreatine
Catalyzed by Creatine Kinase in the Presence of MgATP
Introduction
Identifying an initial biologically active compound for a new
target protein (hit compound)1 either for drug discovery or as a
pharmacological tool remains a troublesome step. Considerable
progress has been made in the route from hit to clinical
candidate, including early integration of ADMET parameters,1
new computational and chemical methods have already been
successfully tested,2 and a few NMR-based methods have raised
particular interest along this line.3-5 However, widely used
approaches for hit generation, including high-throughput screen-
ing (HTS), natural product leads, structural analogy to existing
inhibitors, or rational design, are not appropriate in a number
of situations.
Small molecular fragments have therefore gained widespread
popularity as a tool for drug design and hit generation over the
past decade. Fragments have simple features, which differ from
those of the more elaborate “leadlike” or “druglike” properties
of combinatorial compounds.6 Fragments are thus more readily
amenable to chemical modifications, and have a higher prob-
ability of binding protein targets.7 Hence, fragment libraries can
be much smaller than HTS libraries while sampling a chemical
space inaccessible to the latter, sometimes disclosing binding
“needles” where HTS would fail to find hits.8
A set of empirical rules have been established that define
the proper features for a fragment to allow tractable optimiza-
tion.9 Those limits, known as the “rule of three” 10 (Mw e 300
g‚mol-1, H bond donors and acceptors <3, rotatable bonds <3,
and polar surface area e60 Å2), usually restrict the affinities of
fragments within the micro- to millimolar range. These affinities
are unsuitable for detection by classical HTS methods, but are
typical of those dealt with by NMR methods11,12 or other
structural techniques. The combined power of both structural
methods and fragment concepts makes the approach particularly
viable. The capacity to generate an initial biologically active
“hit” compound with a minimal number of biological tests is
attractive in cases which are not amenable to broad screening.
Furthermore, some structural information may be obtained which
can be valuable for guiding further inhibitor design. Finally,
the hits generated by fragment-based design are of high
quality: the inhibitors are modular, water-soluble, and simple,
and are thus more readily developed into lead compounds or
pharmacological tools.
Diverse collections of compounds (from a few dozen13 to
hundreds14,15 or even thousands16,17) have previously been
gathered in accordance with various criteria to perform NMR
screening.18 Importantly, known drug scaffolds can be used as
fragment templates to minimize toxicity and untractability issues
in further drug design.19,20 This underpinned a pioneering work
by Fejzo et al. where a 100-compound library called SHAPES
was used to identify several ligands of the p38 MAP kinase
and inosine-5′-monophosphate dehydrogenase.15
Using an approach similar to SHAPES, we developed an in-
house NMR library of just 53 compounds and tested it with
established NMR screening protocols on rabbit muscle creatine
kinase (RMCK; EC 2.7.3.2).21 This enzyme plays multiple roles
in the cellular energy distribution network22 by catalyzing the
reversible transfer of a phosphate group between adenosine
triphosphate (ATP) and creatine (Cr; Scheme 1). It is interesting
as a model system for phosphagen kinases such as brain
creatine kinase and arginine kinase, both potential targets for
new drugs,23,24 because RMCK is commercially available and
compatible with NMR requirements (buffer, temperature, ex-
* To whom correspondence should be addressed. E-mail: lancelin@
hikari.cpe.fr. Phone/fax: 0033 472 431 395.
† Laboratoire de RMN et Spectrome´trie de Masse Biomole´culaires.
‡ Laboratoire de Chimie Organique II-Glycochimie.
§ Biomembranes et Enzymes Associe´s.
10.1021/jm061460r CCC: $37.00 © 2007 American Chemical Society
Published on Web 03/22/2007

1866 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 8
Bretonnet et al.
Table 1. Selected Ligands Detected by NMR and Their STD and
periment time). We describe herein the NMR screening and a
fragment evolution strategy against this enzyme that enabled
us to design and synthesize novel hit inhibitors. These
scaffolds exploit a new binding site in proximity of the nucle-
otide binding site, and could be useful for future design of lead
molecules to refine the current knowledge about RMCK and
related homologues.25
LOGSY Effects
Results and Discussion
Compound Selection and NMR Tests. Initially 61 SHAPES-
like commercial compounds were dissolved in DMSO-d6 stock
1
solutions and tested for aqueous solubility and purity by H
NMR spectra.26 WaterLOGSY (water ligand observed via
gradient spectroscopy) spectra were used to detect possible
aggregation, as large LOGSY effects are characteristic of high
molecular weight compounds in water.27 From this preliminary
study, a fraction of the initial set of compounds were withdrawn
from the library owing to poor aqueous solubility (six com-
pounds) or chemical instability (one compound) or because their
signals were partially suppressed with that of water (one
compound). A few compounds showed limited but still practical
solubility in either DMSO-d6 or H2O and were thus used at
their highest practical concentrations as stock solutions in their
best solvent. The 53 stock solutions were finally tested for
stability at +4 °C, and no degradation was detectable over a
year.
Affinity Screening of Rabbit Muscular Creatine Kinase.
RMCK was first tested with one of its endogenous ligands,
adenosine diphosphate complexed to Mg2+ (MgADP). NMR
saturation transfer difference (STD) spectra confirmed the
affinity of CK for the nucleotide. Using this NMR method, we
found a dissociation constant (KD) in the micromolar range
(KD ) 300 ( 40 µM at 20 °C; see the Experimental Section).
This is slightly higher than the 70 ( 13 µM previously reported
from enzymatic data,21 but still of the same order of magnitude.
It should be noted here that NMR titrations have already
previously been found to somewhat overevaluate dissociation
constants.28
a The ligand concentration was 1 mM and the CK concentration was 5
or 10 µM for STD and WaterLOGSY measurements, respectively. b STD
amplification factors (fSTD) and LOGSY effects were calculated as described
in the Experimental Section. c Not determined.
The NMR screening library was tested against CK. A total
of 19 compounds out of 53 were qualitatively retained as
possible weak-to-good ligands (Table 1) on the basis of their
STD and LOGSY effects (see for instance Figure 1). Among
the NMR hits, phenylfuroic acid (1) displayed the highest STD
amplification factor (see the Experimental Section) and the
highest LOGSY effect and was therefore titrated precisely
against CK. Its dissociation constant was millimolar (KD ) 5
mM ( 1 mM, Figure 2), and no direct competition with MgADP
was evidenced by adding increasing and saturating amounts of
MgADP. This suggests that MgADP and 1 do not share the
same binding site. One may thus be driven to a fragment
evolution strategy in which 1, correctly linked with another
ligand of even moderate affinity for the nucleotide site, would
form a tight-binding ligand owing to the additive strength of
the two moieties.5,29
Structural Insights. Docking of the NMR Hits. Docking
simulations can give valuable insight into the structures of the
CK complexes evidenced by NMR. The program DOCK 5.430
was used to confirm and complement our first data, using
RMCK structural models (see the Experimental Section).
The active site of CK is known to undergo a significant
conformational change when bound to its substrates, creatine
and MgATP.31 This was particularly clear in the recent X-ray
structure of the Torpedo californica dimer,32 where each subunit
was unexpectedly crystallized in different conformations: one
“open”, substrate-free, and one “closed” with a transition-state
analogue of the Cr-ATP adduct. Two flexible loops in
particular (namely, from residues 59-69 and 322-331) moved
20 Å closer to each other upon substrate binding. Thus, the
conformational changes of CK cannot be reasonably anticipated
when fragments enter the large groove of the active site. Hence,
our study was systematically performed on both the open and
closed enzymes as detailed in the Experimental Section.
Fragment 1 is the best fragment in terms of binding energy.
In closed and open structures, 1 was docked in a specific site
located between those of creatine and the nucleotide (MgADP,
MgATP), with the phenyl ring directed toward the nucleotide
pocket (Figure 3a). This result is in accordance with the NMR
experiments since no direct competition could be evidenced
between 1 and MgADP. Addition of diverse substituents to the
phenyl ring of 1, to access the nucleotide binding site, thus
seemed an appropriate follow-up strategy.
In Silico Follow-Up of CK Ligands. A total of 1000 follow-
up compounds of 1, complying with the Lipinski rules for
druglike compounds,33 were randomly generated. They all
consisted of a common moiety, 1, extended from its phenyl ring
using various fragment structures and linker types (see Table
2). The linkers were either meta or para to the furan ring of 1.
Those follow-ups were systematically docked to RMCK. They
were then classified to identify the best follow-up characteristics,
in terms of the positions and types of substituents and linkers.

NMR Screening Applied to CK Inhibitor Generation
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 8 1867
Figure 1. Identification of 1 as a hit fragment among a mixture with two other compounds of the fragment library (middle and right formulas) by
NMR screening. WaterLOGSY and STD experiments both confirm the affinity of 1 for the protein target, CK: the peaks for compound 1 are the
higher LOGSY effect (200%) and the only peaks in the STD spectrum.
nucleotide binding site, this compound might be a good
candidate as a follow-up substituent. As a result, molecules
10a-c were selected, and their syntheses were carried out
according to Scheme 2a. An additional compound without the
amide linker, molecule 11, was also synthesized from another
route detailed in Scheme 2b.
Chemical Syntheses. The substituted phenylfuroic acids were
prepared by Suzuki coupling between a furoic acid derivative
and an appropriate substituted phenylboronic acid (Scheme 2).
The basic conditions of the Suzuki coupling required the
protection of the carboxylic acid with an acid-labile group such
as a tert-butyl ester. Thus, ester 634 reacted with 4-carboxyphe-
nylboronic acid to afford the acid 7 in good yield.35 The
carboxylic acid 7 was activated with oxalyl chloride in the
presence of a catalytic amount of DMF to yield the acyl chloride
8,36 which was reacted with different amines to obtain amides
9a-c. Subsequent hydrolysis of the tert-butyl ester with
trifluoroacetic acid provided the corresponding acids 10a-c
(Scheme 2a).37
Figure 2. Binding curve of compound 1 from STD experiments in
the presence of 10 µM CK at 20 °C. Concentrations are plotted against
STD amplification factors (fSTD) to determine the dissociation constant
KD (5 ( 1 mM).
An analysis of the 250 top-ranked compounds enabled us to
propose the following features: (i) three preferred substituent
structures were represented almost exclusively, fragments 2, 3,
and then 4 and other derivatives of the benzophenone type, such
as 5, in decreasing order (see Table 2); (ii) two-atom linkers
were preferred, and among those, the amide linker was the most
frequent; (iii) meta and para substitutions of the phenyl ring
were equally well represented. Thus, the type of linker and
substitution position seemed to be less important than the
substituent with regard to the calculated binding energy of the
fragment-designed molecules. These substituents interact with
residues in the phosphate and adenosine pockets of the ATP
binding site.
As a consequence, archetypal fragment-based molecules with
potentially optimal affinities could be proposed for synthesis.
From many possible combinations those with benzophenone
substituents were preferred, owing to their commercial avail-
ability and their chemical tractability in a minimal number of
steps. Since DOCK especially predicted ligand 5 in the
Compound 11 was synthesized via a Suzuki coupling between
5-(4-bromophenyl)-2-furoic acid and 4-benzoylphenylboronic
acid35 (Scheme 2b).
NMR Assay of the Candidate Inhibitors. Molecule 10a was
the most amenable to NMR testing owing to its high (millimolar)
solubility in aqueous phosphate buffer. It was thus tested as a
1
first potential inhibitor by NMR, and its H spectrum clearly
showed a significant line broadening upon addition of CK
(Figure 4a). This feature typically indicates a rapid to intermedi-
ate exchange of the molecule between a complexed and a free
form on the NMR time scale. The high affinity of 10a for CK
was confirmed and characterized by STD and WaterLOGSY
spectra in the presence of MgADP. Indeed the affinity of
MgADP was significantly reduced when 10a was added, giving
rise to a sign change in the WaterLOGSY spectrum (not shown)
and to a drop in intensity in the STD spectrum (Figure 4b).
This change can be used to evaluate the KD of 10a for CK,
assuming a direct competition between 10a and MgADP by

1868 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 8
Bretonnet et al.
Figure 3. (a) Binding site of compound 1 (green sticks) modeled by docking calculations on the closed form of CK. ADP (red sticks) is represented
in the vicinity of 1 as well as the major residue side chains involved in catalysis (yellow sticks). (b) Lowest energy model of molecule 10a docked
to the active site of RMCK.
Table 2. In Silico Design of Follow-Upsa
a Different fragment substituents (left column) are appended to the initial ligand (compound 1, top row). Different linkers were used (middle column), and
several restrictions were applied to select a total of 1000 appropriate follow-ups.
MgADP complex (KADDP ) 300 ( 40 µM from NMR), and i is
decrease of the STD signal in the presence of 10a relative to
the same signal without 10a. With an average 38% decrease of
the STD signals of ADP (initially at 500 µM) in the presence
of 10a (at 250 µM) this leads to an approximate dissociation
constant of 150 µM. The significant affinity of 10a for CK,
and its effect on the binding of MgADP, was subsequently
using the formula38
K1D0a ) [10a]
KADDP(1 - i)
i(KADDP + [ADP])
where [10a] and [ADP] are the concentrations of 10a and ADP,
respectively, KADDP is the dissociation constant of the CK-

NMR Screening Applied to CK Inhibitor Generation
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 8 1869
Scheme 2. (a) Five-Step Synthesis of 10a-c and (b) Single-Step Synthesis of 11.
confirmed next by an enzymatic assay on 10a as well as on the
other follow-up compounds.
tration allowed us to calculate the two constants KI ) 25 ( 5
µM and RKI ) 70 ( 4 µM, which respectively describe binding
of 10a on free enzyme and on the enzyme-MgATP complex.
These values are somewhat lower than those extrapolated from
Enzymatic Assays. Residual enzymatic activity in the
presence of increasing 10a concentrations has been measured
in standard pH-stat assay conditions with 4 mM ATP and 40
mM creatine to evaluate the inhibitor concentration giving 50%
inhibition (IC50). The results, plotted as the ratio of the reaction
rate in the presence of 10a versus the rate in its absence, are
shown in Figure 5a, which suggest an IC50 in the 50 µM range.
We noticed that inhibition is progressive (steady state is reached
3 min after inhibitor addition; for all measurements, the enzyme
was thus incubated with 10a and creatine for 3 min, and reaction
was started by addition of ATP). Creatine kinase intrinsic
fluorescence is known to be attenuated by the nucleotide
substrate binding,39 but due to its high UV absorption coefficient,
10a is responsible for a strong inner filter effect which precludes
a quantitative analysis of its binding through fluorescence
measurements.
Reversibility of the inhibition was assessed by measuring the
reaction rate at constant substrates and inhibitor concentrations
but with different enzyme concentrations. A plot of the reaction
rate versus enzyme concentration shows that, with or without
10a, the reaction rate increases linearly with the amount of
enzyme added to catalyze the reaction, the slope decreasing in
the presence of 10a (not shown), as expected for a reversible
inhibitor.
NMR measurements, but this is quite commonly observed.27
A
more complete analysis of CK reaction inhibition by 10a would
also require analysis of reaction rates with variable creatine
concentration; these measurements have not been done here.
All the follow-up molecules were next assayed using a single
concentration of MgATP, and their inhibitory properties were
compared to those from a reference experiment, performed with
no inhibitor. A ranking of all molecules could thus be established
on the basis of their inhibitory properties expressed as a
percentage of the reference activity (Table 3). From this ranking,
two molecules (10c and 11) appear as the best inhibitors, both
inhibiting CK by 63%. On the other hand, molecule 10b only
has a weak 20% inhibitory effect.
Structural Insights. Docking of the Inhibitors. Using the
same protocols as for the first fragments, molecules 10a-c and
11 were docked to the structural models of the closed RMCK.
A ranking of the molecules according to their docking energies
was in good agreement with the experimental results (Table 3)
except for molecule 11, the docking energy of which was
underestimated by DOCK. Figure 3b shows the lowest energy
model for 10a, confirming the competition between this
molecule and MgADP for the same binding site. Indeed,
molecule 10a adopts an elongated conformation in which the
benzophenone moiety makes significant contacts with the
nucleotide binding site. An analysis of ligand-protein contacts
(LPCs) derived with LPC software41 showed that 10a interacts
with Arg129, Arg235, and Arg291, all important and conserved
residues for phosphate chain stabilization, and His190, which
is involved in adenosine stabilization.32 Molecule 10b, having
The kinetics of ATP hydrolysis has been followed, as
described in the Experimental Section, with or without 10a (27.5
and 55 µM) with constant amounts of CK and creatine. Reaction
rates were best fitted to an equation describing a mixed type of
reversible inhibition40 (Figure 5b). The 1/V versus 1/[ATP] plot
(Figure 5c) shows the characteristic pattern expected for this
type of inhibition. Variation of KATP and Vm with 10a concen-

1870 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 8
Bretonnet et al.
Figure 4. (a) Aromatic regions of the 1H spectra of compound 10a at 100 µM in the absence (top trace) and presence (bottom trace) of 10 µM CK.
Line broadening is clearly apparent, showing that 10a is interacting with CK. (b) STD spectra of ADP in the absence (top trace) and presence (other
traces) of increasing amounts of 10a. ADP resonances are marked with an arrow.
a less bulky substituent than other inhibitors, also has less
contact in the nucleotide pocket, and this could explain its lower
inhibitory properties. More importantly, the furoic acid moiety
of 10a-c, analogous to fragment 1, is docked at a position
similar to that of the initial compound 1. Assuming that this
subsite is potentially accessible in CK, either in the presence
or in the absence of MgATP, this could explain the mixed
competitive/noncompetitive inhibition mechanism evidenced
for 10a.
Experimental Section
Library Collection. The compounds (at least 95% pure) were
chosen from Aldrich or Acros online catalogs, using their sub-
structure search facilities combined with SHAPES-like selection
criteria inspired from published lists.15,19,20,43 The chosen compounds
had to fit the rule of three;10 to this effect, partition coefficients
either experimental or calculated were obtained using the KowWin
software (Syracuse Research Corp.).44 Whenever documented,
aqueous solubility was also taken into account since millimolar
concentrations were ideally sought in the final samples.
Stock Solutions of CK for NMR and Enzymatic Assays.
RMCK was purchased from Roche (reference 10 127 566 001) as
a lyophilized, ATPase-free powder. A single stock solution at 26
mg/mL (300 µM dimeric CK) was prepared from each new batch
in phosphate-buffered H2O (50 mM, pH 6.8, and 0.02% NaN3 to
prevent microbial growth). It was dispatched in 550 µL aliquots,
conserved at -30 °C between uses. Enzymatic activity was not
affected by serial cycles of freezing-defreezing over several
months.
Stock Solutions of Library Compounds. Whenever possible,
110 mM stock solutions of the library compounds were prepared
in DMSO-d6. These stocks were conserved at 4 °C without apparent
degradation over at least 4 months. For NMR analyses, appropriate
volumes of stock solutions were diluted to 500 µL samples with
or without protein in phosphate-buffered H2O. A 50 µL portion of
D2O was also added for deuterium lock. Thus, a final concentration
of 1 mM for each screening compound could be obtained from 5
µL of stock solution (0.9% DMSO-d6 in the final sample, with no
Concluding Remarks
We have demonstrated that a minimum fragment library (ca.
50 molecules) could be used in conjunction with NMR to
develop micromolar inhibitors of RMCK on the basis of the
synthesis of only a very limited number of compounds. With
molecular modeling and docking, the hit fragments were readily
extrapolated to fully new, non-nucleotide reversible inhibitors
evidenced to act in a mixed competition/noncompetition mode
against MgATP. The docked models confirm that the inhibitors
interact in part with the ATP/ADP-Mg binding site. The
molecular features of the new hit compounds can now be used
as novel bases for designing more active and selective inhibitors
of creatine kinase, or related isozymes of the guanidinium kinase
family, such as arginine kinase, which is involved in several
infectious deseases.24,42

NMR Screening Applied to CK Inhibitor Generation
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 8 1871
effect on CK reactivity, data not shown). In a few cases, the stock
solutions were further diluted in DMSO-d6 owing to reduced
solubility in either DMSO-d6 or H2O.
NMR Experiments. All spectra were acquired at 293 K with a
Varian Inova 600 MHz NMR spectrometer, equipped with a
standard 5 mm triple-resonance inverse probe with a z-axis field
gradient, actively shielded, and with an autosampler robot. Control
1
1D H spectra preceded all experiments to assess the purity and
state of the analytes. Water suppression was achieved by a double
spin-echo scheme26 with 2.4 ms water-selective 180° square pulses.
Spin locks (3.6 kHz) of 15 ms were also used after the detection
pulse in every experiment to achieve protein suppression. The
acquisition time was 1 s, the sweep width 12 ppm, and the relaxation
delay 1.5 s. As for the processing, the data were zero-filled to 16000
complex points and multiplied by an exponential function (line
broadening 0.3 Hz) prior to Fourier transformation. The same
processing parameters and suppression schemes were also used with
WaterLOGSY27,45 and STD46,47 sequences. In the WaterLOGSY
experiment, the first water-selective 180° Gaussian pulse was 25
ms long, and a weak rectangular pulse field gradient was applied
during the mixing time (1 s). A gradient recovery time of 2 ms
was introduced after the mixing time; gradient recoveries were
otherwise 200 ms long. The relaxation delay was 2 s. In the STD
experiment, selective saturation of the protein at 2046 Hz from the
carrier frequency was carried out by a 2 s pulse train (40 Gaussian
pulses of 50 ms between 1 ms delays). A relaxation delay of 100
ms preceded each new scan. Each NMR screen consisted of 1D
1H, STD, and WaterLOGSY experiments performed on mixtures
of 2-6 compounds. Spectra were processed and analyzed with
MestReC 4.4.1.0 (Mestrelab Research, Santiago de Compostella,
Spain).
For quantitative analyses of STD spectra, the STD amplification
factors fSTD were derived from the equation38
ISTD [L]tot
fSTD
)
I0
[CK]tot
where ISTD and I0 are peak integrals in the STD or 1D experiment,
respectively, and [L]tot and [CK]tot are the total concentrations of
the ligand and CK, respectively. Dissociation constants (KD) were
determined from a plot of [L]tot vs fSTD fitted by nonlinear regression
analysis using the Levenberg-Marquardt algorithm in OriginPro
7 (OriginLab).
Quantitative LOGSY effects48,49 were measured as the relative
percentage differences, [(Ip - If) × 100]/If, between the NMR line
intensities of a ligand in the presence (Ip) or the absence (If) of 10
µM CK. In the absence of binding the two intensities are equal,
leading to a null LOGSY effect (0%).
Structural Models of CK Used for Ligand Docking. Structural
models used for docking calculations were derived from the crystal
structures of rabbit muscular CK available from the Protein Data
Bank. The 2CRK entry50 was chosen as a model for the open,
substrate-free CK after reconstruction of the missing residues 323-
331 with Modeller 851-53 using the conformation of chain A of the
1VRP entry.32 Chain B of the 1U6R coordinates was chosen as a
model for the closed, substrate-bound CK conformation since it is
complexed with creatine, MgADP, and a nitrate ion mimicking a
transition-state conformation.
Virtual Screening and Scoring. Prior to the docking process,
the target protein and the ligands were subjected to slight modifica-
tions using the Sybyl software (Tripos, Inc.). All the hydrogens
and charges in the protein were added, and their positions were
energy minimized according to the Powell method. The different
ligands were built using the Sybyl sketch tool. Several atom types
were modified to account for their protonation states. Charges were
computed according to the Gasteiger-Hu¨ckel method, and the
ligand energy was minimized using the Tripos force field with the
simplex method followed by the conjugate gradient method.
Ligand docking to the target receptor was performed with the
DOCK 5.4 software package.30 After molecular surface calculation,
Figure 5. (a) Inhibition of CK as a function of inhibitor (10a)
concentration, plotted as the ratio of the reaction rate in the
presence of 10a (V′) to the uninhibited reaction rate (V). (b) Reaction
rates versus MgATP concentration in the absence (b) or with 27.5
µM (9) and 55 µM (1) compound 10a. Lines correspond to a fit to a
mixed type of reversible inhibition. (c) Corresponding Lineweaver-
Burk plots.

1872 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 8
Bretonnet et al.
Table 3. Ranking of the Synthesized Follow-Ups Based on Enzymatic Assays (Middle Column)a
a The results of docking calculations are added for comparison (right column) and show a good overall agreement. Inhibition constants of 10a against free
CK (KI) and the complex CK-MgATP (RKI) are indicated (see the text). b c log P values correspond to predicted partition coefficient values, obtained from
the KowWin software (Syracuse Research Corp.).44 Mw is the weight-average molar mass (g/mol).
the potential interaction sites were filled with spheres to map their
locations. Only those spheres within the binding pockets of ADP
and creatine were retained as relevant for further computation. As
an outcome, 170 spheres were selected for the closed conformation
whereas 207 spheres were retained for the open conformation prior
to grid energy computation. The corresponding grid was chosen to
enclose the selected spheres plus an extra margin of 2 Å. We
resorted to the anchor-first search method with an “anchor size” of
6 to account for several anchor fragments, whereas other ligand
flexibility parameters were assigned their standard default values.
A total of 50 configurations per ligand building cycle and a
maximum of 5000 orientations were generated. The docked ligands
were ranked according to their energy scores since this was
previously reported as the most robust method.54,55 Moreover, we
considered here the mean energy score between the runs with open
and closed conformations. Three different values of the seeds were
used as recommended previously.30 The highest deviation in terms
of energy score was 1.5 kcal/mol by considering all the ligands.
The latter value did not alter the final ranking of the complete set
of ligands.
Virtual Follow-Up of the Hits. Follow-up compounds of the
first NMR hit were built in three steps using the Ilib Diverse 1.02
software (Inte:Ligand Software-Entwicklungs-und Consulting GmbH,
Maria Enzersdorf, Austria). On one hand, a few derivatives of 1
with varying linker types and lengths, added to the phenyl ring,
were constructed. The linkers were one, two, or three atoms long
(C or N) and positioned meta or para with regard to the furan ring
(see Table 2). The structure of compound 1 was also used as is.
On the other hand, a list of potential substituents was selected, either
from the list of NMR hits or from a set of the Ilib Diverse default
library (aniline, anisole, benzene, 1-benzofuran, benzophenone,
diphenyl, furane, imidazole, indole, isoquinoline, naphthalene,
pyridine, pyrrole, quinoline). Finally the 1-related motifs and the
precited substituents were randomly paired using the Ilib Diverse
software. A Lipinski-compliant filter was applied to select potential
candidates: Mw e 500 g‚mol-1, a maximum of 10 H bond acceptors
and 5 H bond donors, c log P between -3.5 and 5, and number of
rings e4. A total of 1000 molecules were generated. A 3D
conformation for each compound was readily obtained from the
Ilib Diverse software, which uses CORINA (Molecular Networks
GmbH, Erlangen, Germany). The output was an SDF file, converted
into a mol2 file using OpenBabel for subsequent virtual screening
with DOCK.
Chemistry. General Methods. Thin-layer chromatography
(TLC) was carried out on aluminum sheets coated with silica gel
60 F (Merck 5554). TLC plates were inspected by UV light and
developed by treatment with a mixture of 5% H2SO4 in EtOH
followed by heating. Silica gel column chromatography was
performed with Geduran silica gel Si 60 (40-63 µm) purchased
from Merck (Darmstadt, Germany). 1H and 13C NMR spectra were
recorded using Bruker AC200 and DRX300 spectrometers with the
residual solvent as the internal standard. The following abbreviations
are used to explain the observed multiplicities: s, singlet; d, doublet;
dd, doublet of doublets; ddd, doublet of doublets of doublets; t,
triplet; m, multiplet. NMR solvents were purchased from Euriso-
top (Saint Aubin, France). HRMS (LSIMS, CI) mass spectra were
recorded in the positive mode (unless stated otherwise) using a
Thermo Finnigan Mat 95 XL spectrometer. Elemental analyses were
carried out by the Service de Microanalyses du CNRS, Division
de Vernaison, France.
tert-Butyl 5-Bromofuran-2-carboxylate (6). To a suspension
of MgSO4 (1.8 g, 15 mmol) in CH2Cl2 (28 mL) was added H2SO4
(384 µL, 7.5 mmol). The mixture was stirred at room temperature
for 15 min, and 5-bromo-2-furoic acid (1.37 g, 7.1 mmol) and
tBuOH (3.4 mL, 35 mmol) were introduced. The mixture was stirred
at room temperature for 48 h. The reaction was quenched with
saturated aqueous NaHCO3 and diluted with CH2Cl2. The organic
layer was washed with brine, dried over Na2SO4, filtered, and
concentrated to dryness to afford 6 (1.03 g, 55%) as a yellow oil:
1H NMR (300 MHz, CDCl3) δ 6.98 (d, 1H, J ) 3.5 Hz, Hfuran),
6.38 (d, 1H, J ) 3.5 Hz, Hfuran), 1.54 (s, 9H, tBu); 13C NMR (75
MHz, CDCl3) δ 162.2 (CdO), 147.4 (C2), 126.5 (CBr), 119.0, 113.5
(C3, C4), 82.3 (C(CH3)3), 28.1 (C(CH3)3); MS (EI) m/z 246 [M]+;
HRMS (EI) m/z calcd for C9H11BrO3 245.9892, found 245.9887.
4-[5-(tert-Butoxycarbonyl)-2-furyl]benzoic Acid (7). To a
solution of 6 (1.03 g, 3.9 mmol) in 1,4-dioxane (150 mL) was added
Pd(PPh3)4 (226 mg, 0.19 mmol). The mixture was stirred at room
temperature for 15 min, and 4-carboxyphenylboronic acid (649 mg,
3.9 mmol), dissolved in water (37 mL), and K2CO3 (1.08 g, 7.8
mmol) were introduced. The mixture was stirred at 100 °C for 16
h. The reaction was filtered through a Celite pad, and the solvent
was removed under vacuum. The residue was diluted in EtOAc
and washed with water. The aqueous layer was acidified to pH 6
and extracted with EtOAc. The organic layer was dried over Na2-
SO4, filtered, and concentrated to dryness to afford 7 (826 mg, 74%)
as a white solid: 1H NMR (300 MHz, MeOD) δ 8.05 (d, 2H, J )

NMR Screening Applied to CK Inhibitor Generation
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 8 1873
Data for 5-{4-[(benzylamino)carbonyl]phenyl}-2-furoic acid
8.3 Hz, Harom), 7.85 (d, 2H, J ) 8.3 Hz, Harom), 7.17 (d, 1H, J )
3.5 Hz, Hfuran), 7.02 (d, 1H, J ) 3.5 Hz, Hfuran), 1.57 (s, 9H, tBu);
13C NMR (75 MHz, MeOD) δ 169.1 (COOH), 159.6 (COOtBu),
157.3 (C5), 146.7 (C2), 134.7 (C9), 131.8 (C6), 131.4, 125.4 (C7,
C8, C10, C11), 120.4, 110.0 (C3, C4), 83.4 (C(CH3)3), 28.4 (C(CH3)3);
MS (EI) m/z 288 [M]+; HRMS (EI) m/z calcd for C16H16O5
288.0998, found 288.0994. Anal. Calcd for C16H16O5‚0.4H2O: C,
65.03; H, 5.73. Found: C, 65.09; H, 5.56.
General Procedure for the Synthesis of Amides 9a-c.
Carboxylic acid 7 (1 equiv) was azeotropically evaporated two times
with toluene and dissolved in DMF (1 mL). Oxalyl chloride (2 M
in CH2Cl2, 2 mL) was added to the mixture. The reaction was stirred
at room temperature for 16 h, and the solvent was removed in vacuo
to afford 8. Acyl chloride 8 (1 equiv) was dissolved in pyridine (1
mL), and the appropriate amine (1.02 equiv) was added. The
mixture was stirred at room temperature for 1 h. The reaction was
diluted in CH2Cl2 and washed with water. The organic layer was
dried over Na2SO4, filtered, and concentrated to dryness. Purification
(SiO2, 0-50% gradient of EtOAc in CH2Cl2) afforded 9a-c as
white solids.
1
(10b): yield 46%; H NMR (300 MHz, DMSO-d6) δ 9.14 (t, 1H,
J ) 5.8 Hz, NH), 8.01, 7.91 (2d, 4H, J ) 8.4 Hz, J ) 8.4 Hz, H7,
H8, H10, H11), 7.35-7.23 (m, 7H, H3, H4, H15, H16, H17, H18, H19),
4.50 (d, 2H, J ) 5.9 Hz, CH2); 13C NMR (75 MHz, DMSO-d6) δ
166.3 (CONH), 160.0 (COOH), 156.2 (C5), 145.6 (C2), 140.4 (C14),
135.0, 132.3 (C6, C9), 129.1, 125.0 (C7, C8, C10, C11), 128.9, 128.1,
127.6, 125.0 (C15, C16, C17, C18, C19), 120.6, 110.2 (C3, C4), 43.5
(CH2). Anal. Calcd for C19H15NO4‚0.1H2O: C, 70.62; H, 4.74; N,
4.33. Found: C, 70.43; H, 4.78; N, 4.33.
Data for 5-(4-{[(Biphenyl-4-ylmethyl)amino]carbony}phenyl)-
2-furoic acid (10c): yield 90%; 1H NMR (300 MHz, DMSO-d6) δ
9.18 (t, 1H, J ) 5.8 Hz, NH), 8.03, 7.92 (2d, 4H, J ) 8.3 Hz, J )
8.3 Hz, H7, Η8, Η10, Η11), 7.65-7.62 (m, 4H, Harom), 7.48-7.28
(m, 7H, H3, H4, Harom), 4.54 (d, 2H, J ) 5.3 Hz, CH2); 13C NMR
(75 MHz, DMSO-d6) δ 166.3 (CONH), 160.0 (COOH), 156.2 (C5),
145.5 (C2), 140.8, 139.7, 139.6 (C14, C17, C20), 135.0, 132.3 (C6,
C9), 129.7, 128.9, 128.7, 128.1, 127.5, 127.4, 125.0 (C7, C8, C10,
C11, C15, C16, C18, C19, C21, C22, C23, C24, C25), 120.7, 110.2 (C3,
C4), 43.2 (CH2). Anal. Calcd for C25H19NO4‚1.21H2O: C, 71.63;
H, 5.15; N, 3.34. Found: C, 71.24; H, 4.68; N, 3.22.
Data for tert-butyl 5-(4-{[(benzoylphenyl)amino]carbonyl}-
1
5-(4′-Benzoylbiphenyl-4-yl)-2-furoic Acid (11). To a solution
of 5-(4-bromophenyl)-2-furoic acid (1 equiv) in 1,4-dioxane (8 mL/
0.20 mmol) was added Pd(PPh3)4 (0.05 equiv). The mixture was
stirred at room temperature for 15 min, and 4-benzoylphenylboronic
acid (1 equiv), dissolved in water (2 mL), and K2CO3 (2 equiv)
were introduced. The mixture was stirred at 100 °C for 16 h. The
reaction was filtered through a Celite pad, and the solvent was
removed under vacuum. The residue was diluted in EtOAc and
washed with water. The aqueous layer was acidified to pH 6 and
washed with EtOAc. The organic layer was dried over Na2SO4,
filtered, and concentrated to dryness to afford 11 as a white solid:
yield 10%; 1H NMR (300 MHz, DMSO-d6) δ 7.97-7.68 (m, 11H,
phenyl)-2-furoate (9a): yield 72%; H NMR (300 MHz, CDCl3)
δ 8.96 (br s, 1H, NH), 7.84-7.80 (m, 4H, Harom), 7.73-7.58 (m,
6H, Harom), 7.50-7.45 (m, 1H, Harom), 7.39-7.34 (m, 2H, Harom),
7.02 (d, 1Η, J ) 3.5 Hz, Hfuran), 6.67 (d, 1H, J ) 3.5 Hz, Hfuran),
1.50 (s, 9H, tBu); 13C NMR (75 MHz, CDCl3) δ 196.4 (CO), 166.1
(CONH), 158.6 (COOtBu), 155.9 (C5), 145.9 (C2), 142.8, 138.1,
134.5 (C13, C16, C20), 133.4, 133.0 (C6, C9), 132.7, 131.9, 130.3,
128.7, 128.3, 125.0, 120.0 (C7, C8, C10, C11, C14, C15, C17, C18, C21,
C22, C23, C24, C25), 119.5, 108.8 (C3, C4), 82.7 (C(CH3)3), 28.6
(C(CH3)3). Anal. Calcd for C29H25NO5‚0.05CH2Cl2): C, 73.68; H,
5.38; N, 2.96. Found: C, 73.44; H, 5.39; N, 2.91.
Data for tert-butyl 5-{4-[(benzylamino)carbonyl]phenyl}-2-
furoate (9b): yield 87%; 1H NMR (300 MHz, CDCl3) δ 7.74 (m,
4H, Harom), 7.29-7.25 (m, 5H, Harom), 7.06 (d, 1H, J ) 3.5 Hz,
H
arom), 7.62-7.57 (m, 2H, Harom), 7.35 (d, 1H, J ) 3.6 Hz, Hfuran),
7.25 (d, 1H, J ) 3.6 Hz, Hfuran); 13C NMR (75 MHz, DMSO-d6) δ
196.1 (C18), 160.1 (COOH), 156.5 (C5), 145.3 (C2), 144.0, 138.8,
138.0, 136.8, 133.5 (C6, C9, C12, C15, C19), 133.5, 131.3, 130.4,
129.4, 128.5, 127.5, 125.9 (C7, C8, C10, C11, C13, C14, C16, C17, C20,
C21, C22, C23 , C24), 120.7, 109.4 (C3, C4). Anal. Calcd for C24H16O4‚
1.2H2O: C, 73.91; H, 4.76. Found: C, 73.93; H, 4.67.
H
furan), 6.71 (d, 1H, J ) 3.5 Hz, Hfuran), 6.47 (br s, 1H, NH), 4.57
(d, 2H, J ) 5.6 Hz, CΗ2), 1.52 (s, 9H, tBu); 13C NMR (75 MHz,
CDCl3) δ 167.0 (CONH), 158.4 (COOtBu), 156.0 (C5), 145.9 (C2),
138.4 (C14), 134.2, 132.8 (C6, C9), 129.2, 128.3, 128.0, 127.9, 125.0
(C7, C8, C10, C11, C15, C16, C17 C18, C19), 119.3, 108.5 (C3, C4),
82.5 (C(CH3)3), 44.6 (CH2), 28.6 (C(CH3)3). Anal. Calcd for C23H23-
NO4: C, 73.19; H, 6.14; N, 3.71. Found: C, 72.63; H, 6.18; N,
3.74.
Biological Activity. The activity of CK was measured using the
pH-stat method in the forward reaction MgATP + creatine f
MgADP + phosphocreatine + H+.56 The reaction rate was deduced
from the rate of 10-2 M NaOH addition necessary to maintain a
constant pH of 8.8 at 30 °C. The CK concentration was kept at 10
nM in 2.0 mL of assay medium containing 4 mM MgATP and 40
Data for tert-butyl 5-(4-{[(biphenyl-4-ylmethyl)amino]carbony}-
1
phenyl)-2-furoate (9c): yield 39%; H NMR (300 MHz, CDCl3)
δ 7.85 (q, 4H, J ) 8.6 Hz, Harom), 7.58-7.13 (m, 9H, Harom), 7.06
(d, 1H, J ) 3.5 Hz, Hfuran), 6.71 (d, 1H, J ) 3.5 Hz, Hfuran), 6.47
(br s, 1H, NH), 4.68 (d, 2H, J ) 5.6 Hz, CΗ2), 1.58 (s, 9H, tBu);
13C NMR (75 MHz, CDCl3) δ 162.2 (CONH), 157.4 (COOtBu),
155.6 (C5), 145.5 (C2), 140.6, 140.5 (C17, C20), 137.0 (C14), 133.9,
132.3 (C6, C9), 128.7, 128.3, 127.5, 127.4, 127.3, 127.0, 124.6 (C7,
C8, C10, C11, C15, C16, C18, C19, C21, C22, C23, C24, C25), 118.8, 108.1
(C3, C4), 82.0 (C(CH3)3), 43.8 (CH2), 28.2 (C(CH3)3). Anal. Calcd
for C29H27NO4: C, 76.80; H, 6.00; N, 3.09. Found: C, 77.14; H,
6.24; N, 2.97.
General Procedure for the Synthesis of Acids 10a-c. Com-
pounds 9a-c (1 equiv) were dissolved in CH2Cl2 (1 mL), and TFA
(600 µL) was added. The mixture was stirred at room temperature
for 1 h, and the solvent was removed in vacuo to afford 10a-c as
white solids.
Cr
mM creatine (KM ) 6.1 mM). When variable, MgATP concen-
trations ranged from 0.08 to 4.00 mM, magnesium was added as
magnesium acetate and always kept in a 1 mM excess over the
MgATP complex. A 5 or 10 µL portion of an 11 mM DMSO-d6
stock solution of 10a (27.5 or 55 µM final concentration) was
incubated for 3 min in the assay medium (containing CK and
creatine) before the reaction was started by ATP addition (DMSO
by itself had no effect on the CK reaction rate). The Michaelis
constant KM was obtained by fitting the substrate-dependent reaction
velocities to the rate equation for a rapid equilibrium random
mechanism:
Vm*[S]
V )
[I]
[I]
Data for 5-(4-{[(benzoylphenyl)amino]carbonyl}phenyl)-2-
KM* 1 +
+ [S]* 1 +
(
)
(
)
KI
RKI
1
furoic acid (10a): yield 95%; H NMR (300 MHz, DMSO-d6) δ
10.69 (br s, 1H, NH), 8.10 (d, 2H, J ) 8.6 Hz, Harom), 7.99, 7.96
(2d, 4H, J ) 7.2 Hz, H7, H8, H10, H11), 7.80-7.50 (m, 7H, Harom),
7.35 (d, 1H, J ) 3.6 Hz, Hfuran), 7.32 (d, 1H, J ) 3.5 Hz, Hfuran);
13C NMR (75 MHz, DMSO-d6) δ 195.5 (C19), 166.1 (CONH), 160.0
(COOH), 156.0 (C5), 145.7 (C2), 144.1, 138.3, 135.1 (C13, C16, C20),
133.1, 132.9 (C6, C9), 132.6, 131.8, 130.2, 129.6, 129.3, 125.0,
120.4 (C7, C8, C10, C11, C14, C15, C17, C18, C21, C22, C23, C24, C25),
120.7, 110.6 (C3, C4). Anal. Calcd for C25H17NO5‚0.1H2O: C,
72.67; H, 4.20; N, 3.39. Found: C, 72.35; H, 4.30, N, 3.30.
For the ranking of the different follow-up compounds (10a-c and
11) a single experiment at 400 µM MgATP was performed either
without (reference experiment) or with the tested molecule at 27.5
µM.
Acknowledgment. A.-S.B. is the recipient of a doctoral
fellowship from Re´gion Rhoˆne-Alpes (France), which also
supported this work by a grant in aid 2003-2006 (NTAM

1874 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 8
Bretonnet et al.
(22) Wallimann, T.; Wyss, M.; Brdiczka, D.; Nicolay, K.; Eppenberger,
H. M. Intracellular compartmentation, structure and function of
creatine kinase isoenzymes in tissues with high and fluctuating energy
demands: the ‘phosphocreatine circuit’ for cellular energy homeo-
stasis. Biochem. J. 1992, 281 (Part 1), 21-40.
Projects 04-022-408-01 and 05-01-900-701). Support from the
European Union, Contract No. LSHB-CT-2004-503467, to P.G.
and A.J. is gratefully acknowledged. We are also grateful to
the Institut du De´veloppement et des Ressources en Informatique
Scientifique (IDRIS), CNRS, France, for use of its supercom-
puting facility. We gratefully acknowledge Inte:Ligand Software-
Entwicklungs-und Consulting GmbH, Maria Enzersdorf, Austria,
for providing us with a test copy of the Ilib Diverse 1.02
software. We finally thank Dr. Jonathan Moore for helpful
guidelines regarding the NMR library generation.
(23) Wyss, M.; Kaddurah-Daouk, R. Creatine and creatinine metabolism.
Physiol. ReV. 2000, 80, 1107-1213.
(24) Pereira, C. A.; Alonso, G. D.; Ivaldi, S.; Bouvier, L. A.; Torres, H.
N.; Flawia, M. M. Screening of substrate analogs as potential enzyme
inhibitors for the arginine kinase of Trypanosoma cruzi. J. Eukaryotic
Microbiol. 2003, 50, 132-134.
(25) Min, K. L.; Steghens, J. P.; Henry, R.; Doutheau, A.; Collombel, C.
N-dibenzylphospho-N′-3-(2,6-dichlorophenyl)propyl-guanidine is a
bisubstrate-analog for creatine kinase. Biochim. Biophys. Acta 1997,
1342, 83-89.
(26) Hwang, T. L.; Shaka, A. J. Water suppression that workssExcitation
sculpting using arbitrary wave-forms and pulsed-field gradients. J.
Magn. Reson., A 1995, 112, 275-279.
(27) Dalvit, C.; Fogliatto, G.; Stewart, A.; Veronesi, M.; Stockman, B.
WaterLOGSY as a method for primary NMR screening: practical
aspects and range of applicability. J. Biomol. NMR 2001, 21, 349-
359.
(28) Fielding, L.; Rutherford, S.; Fletcher, D. Determination of protein-
ligand binding affinity by NMR: observations from serum albumin
model systems. Magn. Reson. Chem. 2005, 43, 463-470.
(29) Jencks, W. P. On the attribution and additivity of binding energies.
Proc. Natl. Acad. Sci. U.S.A. 1981, 78, 4046-4050.
(30) Ewing, T. J.; Makino, S.; Skillman, A. G.; Kuntz, I. D. DOCK 4.0:
search strategies for automated molecular docking of flexible
molecule databases. J. Comput.-Aided Mol. Des. 2001, 15, 411-
428.
(31) Forstner, M.; Kriechbaum, M.; Laggner, P.; Wallimann, T. Structural
changes of creatine kinase upon substrate binding. Biophys. J. 1998,
75, 1016-1023.
(32) Lahiri, S. D.; Wang, P. F.; Babbitt, P. C.; McLeish, M. J.; Kenyon,
G. L.; Allen, K. N. The 2.1 A structure of Torpedo californica creatine
kinase complexed with the ADP-Mg(2+)-NO(3)(-)-creatine transi-
tion-state analogue complex. Biochemistry 2002, 41, 13861-13867.
(33) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J.
Experimental and computational approaches to estimate solubility
and permeability in drug discovery and development settings. AdV.
Drug DeliVery ReV. 2001, 46, 3-26.
(34) Wright, S. W.; Hageman, D. L.; Wright, A. S.; McClure, L. D.
Convenient preparations of t-butyl esters and ethers from t-butanol.
Tetrahedron Lett. 1997, 38, 7345-7348.
(35) Jiang, W.; Sui, Z.; Macielag, M. J.; Walsh, S. P.; Fiordeliso, J. J.;
Lanter, J. C.; Guan, J.; Qiu, Y.; Kraft, P.; Bhattacharjee, S.; Craig,
E.; Haynes-Johnson, D.; John, T. M.; Clancy, J. Furoyl and
benzofuroyl pyrroloquinolones as potent and selective PDE5 inhibi-
tors for treatment of erectile dysfunction. J. Med. Chem. 2003, 46,
441-444.
(36) Jung, J. H.; Rim, J. A.; Han, W. S.; Lee, S. J.; Lee, Y. J.; Cho, E. J.;
Kim, J. S.; Ji, Q.; Shimizu, T. Hydrogel behavior of a sugar-based
gelator by introduction of an unsaturated moiety as a hydrophobic
group. Org. Biomol. Chem. 2006, 4, 2033-2038.
(37) Byran, D. B.; Hall, R. F.; Holden, K. G.; Huffman, W. F.; Gleason,
J. G. Nuclear analogues of beta-lactam antibiotics. 2 The total
synthesis of 8-oxo-4-thia-z-azabicyclo[4.2.0]oct-2-ene-2-carboxylic
acids. J. Am. Chem. Soc. 1977, 99, 2353-2355.
(38) Meinecke, R.; Meyer, B. Determination of the binding specificity of
an integral membrane protein by saturation transfer difference
NMR: RGD peptide ligands binding to integrin RIIbâ3. J. Med. Chem.
2001, 44, 3059-3065.
References
(1) MacCoss, M.; Baillie, T. A. Organic chemistry in drug discovery.
Science 2004, 303, 1810-1813.
(2) Rees, D. C.; Congreve, M.; Murray, C. W.; Carr, R. Fragment-based
lead discovery. Nat. ReV. Drug DiscoVery 2004, 3, 660-672.
(3) Lepre, C. A.; Peng, J.; Fejzo, J.; Abdul-Manan, N.; Pocas, J.; Jacobs,
M.; Xie, X.; Moore, J. M. Applications of SHAPES screening in
drug discovery. Comb. Chem. High Throughput Screening 2002, 5,
583-590.
(4) Sem, D. S.; Bertolaet, B.; Baker, B.; Chang, E.; Costache, A. D.;
Coutts, S.; Dong, Q.; Hansen, M.; Hong, V.; Huang, X.; Jack, R.
M.; Kho, R.; Lang, H.; Ma, C. T.; Meininger, D.; Pellecchia, M.;
Pierre, F.; Villar, H.; Yu, L. Systems-based design of bi-ligand
inhibitors of oxidoreductases: filling the chemical proteomic toolbox.
Chem. Biol. 2004, 11, 185-194.
(5) Shuker, S. B.; Hajduk, P. J.; Meadows, R. P.; Fesik, S. W.
Discovering high-affinity ligands for proteins: SAR by NMR. Science
1996, 274, 1531-1534.
(6) Oprea, T. I.; Davis, A. M.; Teague, S. J.; Leeson, P. D. Is there a
difference between leads and drugs? A historical perspective. J. Chem.
Inf. Comput. Sci. 2001, 41, 1308-1315.
(7) Carr, R. A.; Congreve, M.; Murray, C. W.; Rees, D. C. Fragment-
based lead discovery: leads by design. Drug DiscoVery Today 2005,
10, 987-992.
(8) Boehm, H. J.; Boehringer, M.; Bur, D.; Gmuender, H.; Huber, W.;
Klaus, W.; Kostrewa, D.; Kuehne, H.; Luebbers, T.; Meunier-Keller,
N.; Mueller, F. Novel inhibitors of DNA gyrase: 3D structure based
biased needle screening, hit validation by biophysical methods, and
3D guided optimization. A promising alternative to random screening.
J. Med. Chem. 2000, 43, 2664-2674.
(9) Teague, S. J.; Davis, A. M.; Leeson, P. D.; Oprea, T. The Design of
Leadlike Combinatorial Libraries. Angew. Chem., Int. Ed. 1999, 38,
3743-3748.
(10) Congreve, M.; Carr, R.; Murray, C.; Jhoti, H. A ‘rule of 3’ for
fragment-based lead discovery? Drug DiscoVery Today 2003, 8, 876-
877.
(11) Diercks, T.; Coles, M.; Kessler, H. Applications of NMR in drug
discovery. Curr. Opin. Chem. Biol. 2001, 5, 285-291.
(12) Peng, J. W.; Moore, J.; Abdul-Manan, N. NMR experiments for lead
generation in drug discovery. Prog. Nucl. Magn. Reson. Spectrosc.
2004, 44, 225-256.
(13) Fielding, L.; Fletcher, D.; Rutherford, S.; Kaur, J.; Mestres, J.
Exploring the active site of human factor Xa protein by NMR
screening of small molecule probes. Org. Biomol. Chem. 2003, 1,
4235-4241.
(14) Becattini, B.; Sareth, S.; Zhai, D.; Crowell, K. J.; Leone, M.; Reed,
J. C.; Pellecchia, M. Targeting apoptosis via chemical design:
inhibition of bid-induced cell death by small organic molecules.
Chem. Biol. 2004, 11, 1107-1117.
(15) Fejzo, J.; Lepre, C. A.; Peng, J. W.; Bemis, G. W.; Ajay;, Murcko,
M. A.; Moore, J. M. The SHAPES strategy: an NMR-based approach
for lead generation in drug discovery. Chem. Biol. 1999, 6, 755-
769.
(16) Baurin, N.; Aboul-Ela, F.; Barril, X.; Davis, B.; Drysdale, M.;
Dymock, B.; Finch, H.; Fromont, C.; Richardson, C.; Simmonite,
H.; Hubbard, R. E. Design and characterization of libraries of
molecular fragments for use in NMR screening against protein targets.
J. Chem. Inf. Comput. Sci. 2004, 44, 2157-2166.
(39) Hagemann, H.; Marcillat, O.; Buchet, R.; Vial, C. Magnesium-
adenosine diphosphate binding sites in wild-type creatine kinase and
in mutants: Role of aromatic residues probed by Raman and infrared
spectroscopies. Biochemistry 2000, 39, 9251-9256.
(40) Segel, I. H. Enzyme Kinetics: BehaVior and Analysis of Rapid
Equilibrium and Steady-State Enzyme Systems; Wiley-Interscience:
New York, 1975.
(41) Sobolev, V.; Sorokine, A.; Prilusky, J.; Abola, E. E.; Edelman, M.
Automated analysis of interatomic contacts in proteins. Bioinformatics
1999, 15, 327-332.
(42) Pereira, C. A.; Alonso, G. D.; Ivaldi, S.; Silber, A.; Alves, M. J.;
Bouvier, L. A.; Flawia, M. M.; Torres, H. N. Arginine metabolism
in Trypanosoma cruzi is coupled to parasite stage and replication.
FEBS Lett. 2002, 526, 111-114.
(17) Hajduk, P. J.; Bures, M.; Praestgaard, J.; Fesik, S. W. Privileged
molecules for protein binding identified from NMR-based screening.
J. Med. Chem. 2000, 43, 3443-3447.
(18) Lepre, C. A. Library design for NMR-based screening. Drug
DiscoVery Today 2001, 6, 133-140.
(43) Peng, J. W.; Lepre, C. A.; Fejzo, J.; Abdul-Manan, N.; Moore, J. M.
Nuclear magnetic resonance-based approaches for lead generation
in drug discovery. Methods Enzymol. 2001, 338, 202-230.
(44) Meylan, W. M.; Howard, P. H. Atom/fragment contribution method
for estimating octanol-water partition coefficients. J. Pharm. Sci.
1995, 84, 83-92.
(19) Bemis, G. W.; Murcko, M. A. The properties of known drugs. 1.
Molecular frameworks. J. Med. Chem. 1996, 39, 2887-2893.
(20) Bemis, G. W.; Murcko, M. A. Properties of known drugs. 2. Side
chains. J. Med. Chem. 1999, 42, 5095-5099.
(21) McLeish, M. J.; Kenyon, G. L. Relating structure to mechanism in
creatine kinase. Crit. ReV. Biochem. Mol. Biol. 2005, 40, 1-20.

NMR Screening Applied to CK Inhibitor Generation
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 8 1875
(45) 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.
(51) Sali, A.; Blundell, T. L. Comparative protein modelling by satisfaction
of spatial restraints. J. Mol. Biol. 1993, 234, 779-815.
(52) Fiser, A.; Do, R. K.; Sali, A. Modeling of loops in protein structures.
Protein Sci. 2000, 9, 1753-1773.
(53) Marti-Renom, M. A.; Stuart, A. C.; Fiser, A.; Sanchez, R.; Melo,
F.; Sali, A. Comparative protein structure modeling of genes and
genomes. Annu. ReV. Biophys. Biomol. Struct. 2000, 29, 291-325.
(54) Charifson, P. S.; Corkery, J. J.; Murcko, M. A.; Walters, W. P.
Consensus scoring: A method for obtaining improved hit rates from
docking databases of three-dimensional structures into proteins. J.
Med. Chem. 1999, 42, 5100-5109.
(55) Gohlke, H.; Hendlich, M.; Klebe, G. Knowledge-based scoring
function to predict protein-ligand interactions. J. Mol. Biol. 2000,
295, 337-356.
(56) Font, B.; Vial, C.; Goldschmidt, D.; Eichenberger, D.; Gautheron,
D. C. Heart mitochondrial creatine kinase solubilization. Effect of
mitochondrial swelling and SH group reagents. Arch. Biochem.
Biophys. 1981, 212, 195-203.
(46) Mayer, M.; Meyer, B. Characterization of ligand binding by saturation
transfer difference NMR spectroscopy. Angew. Chem., Int. Ed. 1999,
38, 1784-1788.
(47) Mayer, M.; Meyer, B. Group epitope mapping by saturation transfer
difference NMR to identify segments of a ligand in direct contact
with a protein receptor. J. Am. Chem. Soc. 2001, 123, 6108-6117.
(48) Ciulli, A.; Williams, G.; Smith, A. G.; Blundell, T. L.; Abell, C.
Probing hot spots at protein-ligand binding sites: A fragment-based
approach using biophysical methods. J. Med. Chem. 2006, 49, 4992-
5000.
(49) Dalvit, C.; Fasolini, M.; Flocco, M.; Knapp, S.; Pevarello, P.;
Veronesi, M. NMR-based screening with competition water-ligand
observed via gradient spectroscopy experiments: Detection of high-
affinity ligands. J. Med. Chem. 2002, 45, 2610-2614.
(50) Rao, J. K.; Bujacz, G.; Wlodawer, A. Crystal structure of rabbit
muscle creatine kinase. FEBS Lett. 1998, 439, 133-137.
JM061460R