Fluorophore labeling of the glycine-rich loop as a method of identifying inhibitors that bind to active and inactive kinase conformations

Article abstract of DOI:10.1021/ja908083e

Targeting protein kinases with small organic molecules is a promising strategy to regulate unwanted kinase activity in both chemical biology and medicinal chemistry research. Traditionally, kinase inhibitors are identified in activity-based screening assays using enzymatlcally active kinase preparations to measure the perturbation of substrate phosphorylation, often resulting in the enrichment of classical ATP competitive (Type I) inhibitors. However, addressing enzymatically incompetent kinase conformations offers new opportunities for targeted therapies and is moving to the forefront of kinase inhibitor research. Here we report the development of a new FLIK (Fluorescent Labels in Kinases) binding assay to detect small molecules that induce changes in the conformation of the glycine-rich loop. Due to cross-talk between the glyclne-rlch loop and the activation loop in kinases, this alternative labeling approach can also detect ligands that stabilize inactive kinase conformations, including slow-binding Type II and Type III kinase inhibitors. Protein X-ray crystallography validated the assay results and identified a novel DFG-out binding mode for a quinazollne-based inhibitor In p38a kinase. We also detected the hlgh-afflnlty binding of a clinically relevant and specific VEGFR2 inhibitor, and we provide structural details of its binding mode In p38a, in which it stabilizes the DFG-out conformation. Last, we demonstrate the power of this new FLIK labeling strategy to detect the binding of Type ligands that induce conformational changes in the glyclnerich loop as a means of gaining affinity for the target kinase. This approach may be a useful alternative to develop direct binding assays for kinases that do not adopt the DFG-out conformation while also avoiding the use of expensive kits, detection reagents, or radioactivity frequently employed with activity-based assays.

Full text of DOI:10.1021/ja908083e

Published on Web 03/04/2010
Fluorophore Labeling of the Glycine-Rich Loop as a Method
of Identifying Inhibitors That Bind to Active and Inactive
Kinase Conformations†
Jeffrey R. Simard, Mattha¨us Getlik, Christian Gru¨tter, Ralf Schneider,
Sabine Wulfert, and Daniel Rauh*
Chemical Genomics Centre of the Max Planck Society, Otto-Hahn-Strasse 15,
D-44227 Dortmund, Germany
Received September 23, 2009; E-mail: daniel.rauh@cgc.mpg.de
Abstract: Targeting protein kinases with small organic molecules is a promising strategy to regulate
unwanted kinase activity in both chemical biology and medicinal chemistry research. Traditionally, kinase
inhibitors are identified in activity-based screening assays using enzymatically active kinase preparations
to measure the perturbation of substrate phosphorylation, often resulting in the enrichment of classical
ATP competitive (Type I) inhibitors. However, addressing enzymatically incompetent kinase conformations
offers new opportunities for targeted therapies and is moving to the forefront of kinase inhibitor research.
Here we report the development of a new FLiK (Fluorescent Labels in Kinases) binding assay to detect
small molecules that induce changes in the conformation of the glycine-rich loop. Due to cross-talk between
the glycine-rich loop and the activation loop in kinases, this alternative labeling approach can also detect
ligands that stabilize inactive kinase conformations, including slow-binding Type II and Type III kinase
inhibitors. Protein X-ray crystallography validated the assay results and identified a novel DFG-out binding
mode for a quinazoline-based inhibitor in p38R kinase. We also detected the high-affinity binding of a
clinically relevant and specific VEGFR2 inhibitor, and we provide structural details of its binding mode in
p38R, in which it stabilizes the DFG-out conformation. Last, we demonstrate the power of this new FLiK
labeling strategy to detect the binding of Type I ligands that induce conformational changes in the glycine-
rich loop as a means of gaining affinity for the target kinase. This approach may be a useful alternative to
develop direct binding assays for kinases that do not adopt the DFG-out conformation while also avoiding
the use of expensive kits, detection reagents, or radioactivity frequently employed with activity-based assays.
1. Introduction
motif, and amino acid residues in the glycine-rich loop located
above the ATP binding cleft.4 Most Type I inhibitors also form
hydrogen bond(s) with the hinge region of the kinase and
thereby compete with the binding of ATP. Given the highly
conserved nature of the ATP binding pocket, poor Type I
inhibitor selectivity and efficacy is a bottleneck in the develop-
ment of effective long-term cancer treatments.5 Modern kinase
inhibitor research is therefore currently shifting focus to the
identification of new chemical entities that bind to and stabilize
the enzymatically incompetent “DFG-out” kinase conformation.2
The DFG-out conformation results from structural changes in
the activation loop induced by a flip of the highly conserved
DFG motif,6,7 an event that also exposes a less-conserved
“allosteric” site adjacent to the ATP binding site. Studies into
the druggability of this site are moving to the forefront of kinase
inhibitor research.8 For example, Type II inhibitors such as
sorafenib (Nexavar)9 and imatinib (Gleevec)10 bind to the hinge
Protein kinases are an important set of enzymes regulating
key cellular processes. The improved understanding of aber-
rantly regulated kinase signaling in cancer biology1 has initiated
the area of targeted tumor therapy and led to the development
of small organic molecules that specifically target unwanted
kinase activities.2 Most small organic molecules modulating
kinase activity are Type I inhibitors such as staurosporine and
dasatinib, which typically inhibit a kinase by binding to the
active “DFG-in” conformation. In this conformation, the regula-
tory activation loop is open and extended, which allows ATP
and substrates to bind.3 The adenine of ATP forms a crucial
hydrogen bond with the hinge region of the kinasesa short,
flexible region located between the N- and C-terminal lobes of
the kinase domainswhile the ꢀ and γ phosphates of ATP are
coordinated by a complex network of ionic and hydrogen-
bonding interactions with several structural elements, including
Mg2+ or Mn2+ ions, the Asp side chain of the conserved DFG
(4) Aimes, R. T.; Hemmer, W.; Taylor, S. S. Biochemistry 2000, 39, 8325–
8332.
† All crystal structures described in this paper were deposited into the
Protein Data Bank (PDB codes: 3HUC, 3HUB, 3L8S).
(1) Gschwind, A.; Fischer, O. M.; Ullrich, A. Nat. ReV. Cancer 2004, 4,
361–370.
(5) Backes, A. C.; Zech, B.; Felber, B.; Klebl, B.; Mu¨ller, G. Expert Opin.
Drug DiscoVery 2008, 3, 1409–1425.
(6) Liu, Y.; Gray, N. S. Nat. Chem. Biol. 2006, 2, 358–364.
(7) Pargellis, C.; Tong, L.; Churchill, L.; Cirillo, P. F.; Gilmore, T.;
Graham, A. G.; Grob, P. M.; Hickey, E. R.; Moss, N.; Pav, S.; Regan,
J. Nat. Struct. Biol. 2002, 9, 268–272.
(2) Zhang, J.; Yang, P. L.; Gray, N. S. Nat. ReV. Cancer 2009, 9, 28–39.
(3) Knighton, D. R.; Zheng, J. H.; Ten Eyck, L. F.; Xuong, N. H.; Taylor,
S. S.; Sowadski, J. M. Science 1991, 253, 414–420.
9
4152 J. AM. CHEM. SOC. 2010, 132, 4152–4160
10.1021/ja908083e  2010 American Chemical Society

Identifying Inhibitors That Bind to Kinase Conformations
A R T I C L E S
region and are ATP-competitive but also extend deep into this
allosteric site, while Type III inhibitors bind exclusively within
the allosteric pocket.7,11
as well as ligands that may bind in the ATP site and induce
other conformational changes in target kinases not regulated
by the DFG-in/out switch, thereby making the FLiK approach
applicable to a larger number of kinases.
Although molecules that utilize this less-conserved site are
believed to have superior selectivity profiles and improved
pharmacological properties and may offer new opportunities for
drug development,6,12 approaches that could discriminate be-
tween different inhibitor types and identify inhibitors that
stabilize the inactive DFG-out conformation were not compatible
with high-throughput screening (HTS) formats used by academia
and industry.13 We previously developed a robust new binding
assay system (FLiK, Fluorescent Labels in Kinases), in which
we tagged the activation loop of cSrc and p38R kinases with
acrylodan,11,14 allowing for the direct measurement of the
dissociation constant (Kd), rate constant of association (kon), and
rate constant of dissociation (koff) of various ligands. More
recently, two additional binding assays based on the displace-
ment of prebound probes from p38R kinase were also reported:
one made use of a fluorophore-labeled inhibitor,15 and the other
employed an enzyme fragment complementation-based ap-
proach.16 In the latter case, a chemiluminescence read-out was
generated by the displacement of a prebound inhibitor-peptide
probe, which then complements and activates ꢀ-galactosidase
to catalyze a chemiluminescence reaction that serves as the assay
read-out. Although these approaches were demonstrated to be
suitable for determining the affinities of displacing ligands using
end point measurements, analysis of kinetic parameters (kon and
koff) is less straightforward since signal detection is rate-limited
by the well-characterized slow dissociation of the chosen
pyrazolourea-based probes from p38R.7
The FLiK approach has so far led to the identification of the
first reported Type III ligands of cSrc,11 identified a new Type
III binding mode for the thiazole-urea scaffold in p38R, and
also led to the identification of several unique Type I ligands
that stabilize the DFG-out conformation of p38R. The sensitivity
in detecting ligands that stabilize the DFG-out conformation is
significantly enhanced by using the FLiK approach to screen
compound libraries since it utilizes the unphosphorylated
inactive form of the kinase. However, manipulation of the
activation loop by mutation and subsequent fluorophore labeling
may not be tolerated by some kinases, may significantly alter
the DFG-in/out equilibrium, and could lead to significant
changes in the affinity of known inhibitors of the target kinase.
Moreover, there is still no method available for predicting
whether a kinase of interest can even adopt the DFG-out
conformation. Therefore, it would be useful to develop alterna-
tive labeling strategies for sensitively detecting DFG-out binders
Here we report on the development of a fluorescent-labeled
kinase assay system that takes advantage of cross-talk between
the activation loop and the glycine-rich loop of the kinase
domain (Figure 1). Aside from the activation loop, the glycine-
rich loop (or P-loop) is another highly conserved flexible
structural element located in the N-terminal lobe of kinases and
contains the canonical Gly-X-Gly-X-X-Gly motif (where X may
be any amino acid). Through the formation of a hydrophobic
interface, it is believed that these two loops interact with one
another (Figure 1a,b) and modulate the conformational equi-
librium of the kinase.9 This cross-talk is also partly mediated
by the nearby helix C in many kinases, which forms the “roof”
of the allosteric site available in the DFG-out conformation.17
Thus, the glycine-rich loop also serves as an important
determinant for inhibitor selectivity and affinity and responds
to movements in the activation loop induced by Type II and
Type III ligand binding within the allosteric site. Additionally,
the glycine-rich loop serves as a regulatory “flap” that controls
the entry of ATP and substrates into the ATP binding site18
and has been shown to change conformation with the binding
of some Type I inhibitors in the ATP binding pocket, thereby
increasing their affinity by shielding them from the surrounding
solvent.19,20
By labeling the glycine-rich loop of the serine-threonine
kinase p38R with acrylodan at a specific position within the
Gly-X-Gly-X-X-Gly motif, we were able to develop a kinase
biosensor that can detect the binding of inhibitors with different
binding modes. Type II and Type III inhibitors, which mediate
a conformational change in the glycine-rich loop via movement
of the activation loop to the DFG-out conformation, were easily
discriminated in HTS formats by monitoring time-dependent
changes in fluorescence signal or Kd over time, or in cuvettes
by measuring kon (<5 s for Type I binders). In addition to
achieving our primary goal of detecting Type II and Type III
inhibitors, this binding approach also sensitively detected Type
I ligands, which gain affinity by interacting directly with the
glycine-rich loop and, in some cases, also stabilize the DFG-
out conformation. We present several crystal structures of
compounds in complex with p38R and report a new binding
mode for quinazoline-based inhibitors identified solely with this
second-generation FLiK approach. Such information may prove
to be useful in further design and synthesis efforts to redirect
known scaffolds to inhibit inactive kinase conformations. We
propose this new FLiK approach as a useful method for
identifying new ligands for kinases that are not known to be
regulated by the DFG-in/-out equilibrium, a required feature
for the success of our original FLiK approach of labeling the
activation loop of kinases.11,14
(8) Backes, A. C.; Zech, B.; Felber, B.; Klebl, B.; Mu¨ller, G. Expert Opin.
Drug DiscoVery 2008, 3, 1427–1449.
(9) Wan, P. T.; Garnett, M. J.; Roe, S. M.; Lee, S.; Niculescu-Duvaz, D.;
Good, V. M.; Jones, C. M.; Marshall, C. J.; Springer, C. J.; Barford,
D.; Marais, R. Cell 2004, 116, 855–867.
(10) Nagar, B.; Bornmann, W. G.; Pellicena, P.; Schindler, T.; Veach, D. R.;
Miller, W. T.; Clarkson, B.; Kuriyan, J. Cancer Res. 2002, 62, 4236–
4243.
(11) Simard, J. R.; Kluter, S.; Grutter, C.; Getlik, M.; Rabiller, M.; Rode,
H. B.; Rauh, D. Nat. Chem. Biol. 2009, 5, 394–396.
(12) Copeland, R. A.; Pompliano, D. L.; Meek, T. D. Nat. ReV. Drug
DiscoVery 2006, 5, 730–739.
(17) Eyers, P. A.; Churchill, M. E.; Maller, J. L. Cell Cycle 2005, 4, 784–
789.
(13) Annis, D. A.; Nazef, N.; Chuang, C. C.; Scott, M. P.; Nash, H. M.
J. Am. Chem. Soc. 2004, 126, 15495–15503.
(18) Tamayo, N.; Liao, L.; Goldberg, M.; Powers, D.; Tudor, Y. Y.; Yu,
V.; Wong, L. M.; Henkle, B.; Middleton, S.; Syed, R.; Harvey, T.;
Jang, G.; Hungate, R.; Dominguez, C. Bioorg. Med. Chem. Lett. 2005,
15, 2409–2413.
(14) Simard, J. R.; Getlik, M.; Grutter, C.; Pawar, V.; Wulfert, S.; Rabiller,
M.; Rauh, D. J. Am. Chem. Soc. 2009, 131, 13286–13296.
(15) Tecle, H.; et al. Chem. Biol. Drug Des. 2009, 74, 547–559.
(16) Klu¨ter, S.; Gru¨tter, C.; Naqvi, T.; Rabiller, M.; Simard, J. R.; Pawar,
V.; Getlik, M.; Rauh, D. J. Med. Chem. 2010, 53, 357–367.
(19) Hanks, S. K.; Hunter, T. FASEB J. 1995, 9, 576–596.
(20) Mapelli, M.; Massimiliano, L.; Crovace, C.; Seeliger, M. A.; Tsai,
L. H.; Meijer, L.; Musacchio, A. J. Med. Chem. 2005, 48, 671–679.
9
J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010 4153

A R T I C L E S
Simard et al.
Figure 1. Conformational changes triggered by ligand binding in p38R. (a) Structurally important regions of the active (DFG-in) kinase domain are highlighted
(helix C, light blue; hinge region, pink). (b) Type II/III inhibitor (mesh surface) binding to the DFG-out conformation causes a conformational change in the
activation loop (white f dark blue) and allows the glycine-rich loop (white f dark red) to adopt a more extended conformation. (c) The glycine-rich loop
(red) was labeled with a fluorophore (colored sphere) to generate a direct binding assay that detects conformational changes in this structural element. Some
Type I inhibitors (blue surface) bind to the active kinase conformation (left panel) and directly interact with the glycine-rich loop. The glycine-rich loop
adopts a more typical extended conformation in the absence of ligand (middle panel). Type II/III inhibitors (green surface) bind to inactive kinase conformations
in which the glycine-rich loop adopts its most extended conformation.
2. Materials and Methods
2.3. Analysis of p38r Labeling by HPLC and Mass Spec-
trometry. Monolabeling of the protein was verified by ESI-MS as
described previously.11,14 Purified unlabeled and acrylodan-labeled
human p38R (∼1.5 mg/mL) were desalted using Protein Desalting
Spin Columns (Pierce) according to the manufacturer’s instructions.
The protein was further prepared for analysis by adding up to 10%
v/v acetonitrile. Aliquots (20 µL) were analyzed by ESI-MS using
a Finnigan LCQ Advantage Max mass spectrometer from Thermo.
Deconvolution and visualization were performed using the Bio-
Works 3.1 software from Thermo Scientific.
To assess irreversible labeling of the protein by the electrophile
of the “hit” compound 6, the unlabeled human p38R (∼1.5 mg/
mL) was incubated with a 3:1 molar ratio of 6 for 20-30 min at
room temperature and then desalted and analyzed using the
procedure described above.
2.4. Fluorescence Characterization of Acrylodan-Labeled
(Glycine-Rich Loop) p38r. Characterization of the fluorescence
response of glycine-rich loop-labeled p38R was carried out using
several types of inhibitors expected to interact with the glycine-
rich loop either directly or indirectly (via movement of the activation
loop). Acrylodan-labeled p38R was excited at 386 nm with
excitation and emission slits set to 3 and 5 nm, respectively.
Emission spectra were recorded every 30 s in the absence and in
the presence of saturating concentrations of the DFG-out binding
ligand, BIRB-796. These measurements revealed the most sensitive
wavelength to monitor fluorescence changes (kinetics) in real-time
(475 nm) and to assess the feasibility of using ratiometric
fluorescence to plot binding curves (Kd). The emission spectra of
glycine-rich loop-labeled p38R (unbound and DFG-out saturated)
2.1. Materials. The fluorophore 6-acryloyl-2-dimethylaminon-
aphthalene (acrylodan) was purchased from Invitrogen GmbH
(Germany). Crystallization plates (EasyXtal Tool; 24-well) were
obtained from Qiagen GmbH (Germany). Cuvettes and mini stir
bars were obtained from Carl Roth GmbH (Germany). Small
volume (20 µL fill volume) black flat bottom 384-well plates were
obtained from Greiner Bio-One GmbH (Solingen, Germany). All
supplies for the HTRF assay kit for p38R were purchased from
CisBio (Bagnols-sur-Ce`ze, France).
2.2. Expression, Purification, and Labeling of p38r Kinase.
The genes for all variants of p38R used in the described FLiK
approach were synthesized containing the necessary mutations for
specific acrylodan labeling on the glycine-rich loop by Geneart
(Regensburg, Germany). The mutated human p38R kinase construct
(Cys119Ser/Cys162Ser/Tyr35Cys) as well as wild-type p38R were
subsequently subcloned into a pOPINF vector and transformed as
an N-terminal His-tag construct with a PreScission Protease
cleavage site into BL21(DE3) Codon+RIL E. coli. Protein expres-
sion, purification, fluorophore labeling, and verification of mono-
labeling of the kinase were performed essentially as described
previously.14
Briefly, fluorophore labeling of both proteins was carried out by
combining mutant p38R and the thiol-reactive fluorophore acrylodan
(dissolved in DMSO) in 50 mM HEPES buffer (pH 7.0) at a molar
ratio of 1:1.5, respectively, and allowed to react in the dark
overnight at 4 °C. The conjugated protein was subsequently
concentrated, washed three times in a 10k-MWCO Centricon using
an appropriate storage buffer, aliquoted and frozen at -80 °C.
9
4154 J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010

Identifying Inhibitors That Bind to Kinase Conformations
A R T I C L E S
were used to calculate the fluorescence parameter ∆Rmax as
described elsewhere21 and revealed that the ratio of emission
intensities at 475 and 512 nm (R ) Iλ512/Iλ475) allowed the most
sensitive detection of ligands that either directly or indirectly
displace the glycine-rich loop upon ligand binding. In the case of
p38R and BIRB-796, we calculated an average ∆Rmax value of 0.303
( 0.083.
2.5. End Point and Kinetic Measurements. To characterize
this new labeling strategy, p38R labeled with acrylodan at the
glycine-rich loop (50 nM) was screened against a small subset
(∼400) of compounds based on scaffolds that are generally known
to be privileged for binding to the DFG-in or DFG-out conformation
of kinases. The kinase was pre-incubated with various concentra-
tions of inhibitor before end point fluorescence measurements were
carried out in either polystyrene cuvettes or 384-well plates to
determine the Kd of each compound. A standard buffer (50 mM
Hepes, 200 mM NaCl, pH 7.45) was used for all experiments. For
cuvette measurements, incubations were carried out overnight in
the dark at 4 °C for p38R. For HTS formats, incubations were
carried out for up to 5 h at room temperature. Long incubation
times are needed to account for the time-dependence of Type II
inhibitor binding to p38R.7
In the cuvette format, a series of cuvettes containing different
amounts of inhibitor were prepared using inhibitor stocks (0.01,
0.1, 1.0, and 10.0 mM in DMSO). All measurements of the cuvettes
were made with a JASCO FP-6500 fluorescence spectrophotometer
(JASCO GmbH, Gross-Umstadt, Germany). A Tecan SafireII (Tecan
Deutschland GmbH, Germany) was used to measure the fluores-
cence read-out in the 384-well plate format. The % v/v DMSO did
not exceed 0.2% in cuvettes and was 5% v/v in 384-well plates. In
the case of p38R, average Z′ factors of 0.67 ( 0.05 (n ) 6) and
0.64 ( 0.10 (n ) 6) were determined for the cuvette and 384-well
formats, respectively, using saturating amounts of BIRB-796 or
sorafenib14 as a positive control for a ligand that induces glycine-
rich loop movement. Vehicle (DMSO) was used as the negative
control.
For acrylodan-labeled p38R, ratiometric fluorescence values
(R ) Iλ512/Iλ475) enabled reliable binding curves of detected
compounds to be plotted, allowing for direct determination of the
Kd of ligand binding. It should be noted that binding modes of
detected molecules can initially be assessed in HTS formats by
measuring plates from primary and/or secondary screens at different
time points. Compounds that change the maximum ratiometric
signal or Kd over time are likely to be Type II/III inhibitors.
Alternatively, kinetics of the association (kon) and dissociation (koff)
of selected compounds can be determined using cuvettes as
described previously.14 To determine kon, a mini stir bar was placed
in the bottom of each cuvette to ensure rapid mixing as inhibitor
was delivered through the injection port located above the cuvette.
Fluorescence changes were monitored at 475 nm for p38R in real-
time using a JASCO FP-6500 fluorescence spectrophotometer
(JASCO GmbH, Groꢀ-Umstadt, Germany). Nearly instantaneous
binding kinetics (<5 s) are characteristic of Type I inhibitors, while
slower kinetics (>10 s) indicate the slow binding of Type II or
Type III inhibitors to the DFG-out conformation. Following binding,
koff was determined by adding a 10-fold excess of unlabeled kinase
to shift the binding equilibrium away from the labeled kinase, as
described previously.14 Addition of excess unlabeled kinase causes
the inhibitor to redistribute and dissociate from acrylodan-labeled
p38R, resulting in a recovery of the fluorescence signal. All binding
and dissociation curves were fit to a single exponential equation:
F(t) ) F(∞) + F(0) exp(-tkobs), where t is time, F(0) is the initial
fluorescence intensity, and F(∞) is the fluorescence at t ) ∞. The
half-time of fluorescence decay (t1/2) was calculated with the
2.6. Co-crystallizations of p38r with Detected Ligands.
Inhibitors were co-crystallized with unlabeled p38R using conditions
previously reported.14,22 Briefly, protein-inhibitor complexes were
prepared by mixing 30 µL of p38R (10 mg/mL) with 0.3 µL of
inhibitor (100 mM in DMSO) and incubating the mixture for 1-2
h on ice. Samples were centrifuged at 13 000 rpm for 5 min to
remove excess inhibitor. Crystals were grown in 24-well crystal-
lization plates using the hanging drop vapor diffusion method and
by mixing 1.5 µL of protein-inhibitor solution with 0.5 µL of
reservoir (100 mM MES pH 5.6-6.2, 20 - 30% PEG4000 and 50
mM n-octyl-ꢀ-D-glucopyranoside).
3. Results
3.1. Selection of the Labeling Position on the Glycine-Rich
Loop. The serine/threonine kinase p38R is an attractive drug
target and has been extensively characterized for drug design.
Additionally, the dynamics of protein-ligand interactions as
well as p38R conformational dynamics have been studied using
both protein X-ray crystallography and NMR approaches and
have provided insights into cross-talk between the activation
loop and the glycine-rich loop.23,24 These data guided us to use
p38R as the model kinase for development and characterization
of an alternative FLiK approach that involves a glycine-rich
loop labeling strategy to detect and characterize the binding of
DFG-out binding Type II and III inhibitors.
We replaced surface-exposed cysteines with serines and
introduced a new Cys residue (Tyr35Cys) via site-directed
mutagenesis into the glycine-rich loop of p38R at the position
that directly precedes the third Gly of the Gly-X-Gly-X-X*-
Gly motif (as indicated by *). Single labeling of the kinase with
the environmentally sensitive fluorophore acrylodan was carried
out and verified by ESI-MS as described previously14 (Figure
S1, Supporting Information). The amino acid residue at the
selected labeling position is conserved as a Tyr or Phe in ∼80%
of human kinases (Figure S2, Supporting Information), sug-
gesting that labeling at this position may make this new FLiK
approach applicable to a larger number of kinases. Furthermore,
replacing a tyrosine with the comparably sized acrylodan-labeled
cysteine should be well tolerated by the kinase. This was
confirmed by using an activity-based assay to measure and
compare the ATP-Km of each phosphorylated active p38R
variant (mutated/unlabeled and acrylodan-labeled) to that of
wild-type p38R (Table S1, Supporting Information). Likewise,
as observed previously for the activation loop-labeled p38R,14
we observed no significant changes in the IC50 values of three
known p38R inhibitors upon mutation or labeling of the glycine-
rich loop, demonstrating that both FLiK assay approaches
ultimately provided similar affinity data for detected ligands.
Nevertheless, ATP-Km data reported here and previously14
suggest that the acrylodan labeling of the glycine-rich loop was
better tolerated by p38R than labeling of the activation loop.
3.2. End Point and Real-Time Measurements: Determina-
tion of Kd, kon, and koff. To compare this approach with our
previously reported method in which we labeled the activation
loop of p38R, and to demonstrate that labeling the glycine-rich
(22) Simard, J. R.; Gru¨tter, C.; Pawar, V.; Aust, B.; Wolf, A.; Rabiller,
M.; Wulfert, S.; Robubi, A.; Klu¨ter, S.; Ottmann, C.; Rauh, D. J. Am.
Chem. Soc. 2009, 131, 18478–18488.
following equation: t1/2 ) ln 2/kobs
.
(23) Honndorf, V. S.; Coudevylle, N.; Laufer, S.; Becker, S.; Griesinger,
C. Angew. Chem., Int. Ed. 2008, 47, 3548–3551.
(24) Vogtherr, M.; Saxena, K.; Hoelder, S.; Grimme, S.; Betz, M.;
Schieborr, U.; Pescatore, B.; Robin, M.; Delarbre, L.; Langer, T.;
Wendt, K. U.; Schwalbe, H. Angew. Chem., Int. Ed. 2006, 45, 993–
997.
(21) de Lorimier, R. M.; Smith, J. J.; Dwyer, M. A.; Looger, L. L.; Sali,
K. M.; Paavola, C. D.; Rizk, S. S.; Sadigov, S.; Conrad, D. W.; Loew,
L.; Hellinga, H. W. Protein Sci. 2002, 11, 2655–2675.
9
J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010 4155

A R T I C L E S
Simard et al.
Figure 2. Fluorescence measurements using glycine-rich loop-labeled p38R incubated with 1 (BIRB-796). (a) Acrylodan emission at 475 nm decreases
significantly relative to the emission at 512 nm (as indicated by arrows) in the presence of increasing amounts of the Type II p38R inhibitor 1. (b) End point
equilibrium measurements of ratiometric fluorescence (R ) Iλ512/Iλ475) were made at several inhibitor concentrations to directly obtain the Kd (∼11 nM for
1). (c) Real-time fluorescence changes at 475 nm resulting from the addition (arrow) of different amounts of 1 were fit to a first-order decay function to
obtain the observed rate constant (kobs) for the binding of each dose of added inhibitor. The fluorescence decrease was dose-dependent. (d) Extraction of
prebound 1 from glycine-rich loop-labeled p38R using an excess of unlabeled p38R allows direct determination of the dissociation rate constant (koff). All
experiments were carried out using 100 nM of the fluorescent-labeled kinase. End point and real-time kinetic data are representative of a typical single
experiment for 1.
loop with acrylodan serves as a reliable alternative approach,
we used the Type II p38R inhibitor 1 (BIRB-796) to characterize
the fluorescence response. End point measurements were carried
out by measuring the emission spectrum of glycine-rich loop-
labeled p38R in the presence of increasing concentrations of 1
(Figure 2a). Subsequently, a ratiometric fluorescence value
(R ) Iλ512/Iλ475) was plotted on a logarithmic scale against the
concentration of ligand to directly determine Kd ) 9.5 ( 2.7
nM for 1 (Figure 2b), which is similar to our previously reported
value of 7.5 ( 2.3 nM obtained with p38R labeled at the
activation loop and demonstrates the reliability of Kd values for
DFG-out binders obtained with this alternative approach.
A significant change of emission at 475 nm also made it
possible to study the kinetics of dissociation and association in
real-time for different concentrations of ligand. Upon addition
of 1, the fluorescence emission at 475 nm decreased in a dose-
dependent manner and was nicely fit to a first-order kinetic
(Figure 2c) to obtain the observed rate constant (kobs) of binding
for each dose of added ligand. These values can subsequently
be plotted and fit linearly to obtain the true kon for a ligand,
which we determined to be ∼4.0 × 104 M-1 s-1 for 1.
Alternatively, addition of a 10-fold molar excess of unlabeled
p38R to glycine-rich loop-labeled p38R prebound with 1 resulted
in a first-order increase in fluorescence as the ligand dissociates
from the labeled kinase, allowing direct determination of koff
(Figure 2d). These measurements demonstrate both the revers-
ibility of the fluorescence response and the changing equilibrium
that exists between the DFG-in and DFG-out conformations.
Determination of kon and koff also allows for an alternative
route for calculating, rather than directly measuring, Kd values
(Kd ) koff/kon) and sheds light on the factors contributing to
different ligand affinities. The values for kon and koff obtained
here are comparable to those published elsewhere for 1, found
using alternative methods.7 Furthermore, the obtained rate
constants can be used to calculate Kd ) 7.1 nM for 1, which is
similar to the calculated Kd previously reported, found by
measuring kon and koff obtained with p38R labeled on the
activation loop (11.9 nM14), again demonstrating that the
glycine-rich loop labeling strategy is a reliable alternative
approach to labeling the activation loop for the detection of
DFG-out binding ligands.
3.3. Adaptation of Acrylodan-p38r to HTS Formats, Hit
Identification, and Characterization. After the fluorescence
response of glycine-rich loop-labeled p38R was characterized,
the assay was adapted to HTS formats aimed at the sensitive
detection of Type II and Type III inhibitors in a small subset of
compounds available in our laboratory16,25-28 comprising vari-
ous scaffolds known to be privileged for binding to the DFG-
in or DFG-out conformation of kinases. After an initial prescreen
(25) Getlik, M.; Gru¨tter, C.; Simard, J. R.; Klu¨ter, S.; Rabiller, M.; Rode,
H. B.; Robubi, A.; Rauh, D. J. Med. Chem. 2009, 52, 3915–3926.
(26) Michalczyk, A.; Klu¨ter, S.; Rode, H. B.; Simard, J. R.; Gru¨tter, C.;
Rabiller, M.; Rauh, D. Bioorg. Med. Chem. 2008, 16, 3482–3488.
(27) Pawar, V. G.; Sos, M. L.; Rode, H. B.; Rabiller, M.; Heynck, S.; van
Otterlo, W. A. L.; Thomas, R. K.; Rauh, D. In press.
(28) Sos, M. L.; et al. Cancer Res. 2009, in press.
9
4156 J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010

Identifying Inhibitors That Bind to Kinase Conformations
A R T I C L E S
Table 1. Characterization of Inhibitor Types Using the Glycine-Rich Loop-Labeled Assay for p38Ra
a Compounds 3 and 4 were not detected (ND). 3 is a sterically bulky derivative of the classical pyrazolourea Type III inhibitors of p38R7,29 that does
not bind in the described allosteric site of p38R14 and serves as a negative control for stabilization of the DFG-out conformation. 4 (dasatinib) is a
known Type I inhibitor of Abl/cSrc and p38R30 that binds to the active DFG-in conformation and does not interact with the glycine-rich loop and thus
served as a negative response control for Type I inhibitors. All Kd determinations are the mean ( SD of at least three independent measurements.
at three fixed inhibitor concentrations (0.5, 5, and 50 µM),
selected compounds, some of which were derivatives of known
p38R inhibitors, were subjected to further studies using con-
centration-dependent direct binding measurements to determine
kon, koff, and Kd (Table 1) (Figure S3, Supporting Information)
and to further compare affinities of compounds detected with
this new FLiK approach and the originally published approach14
as a means of validation.
adopt the classical Type I binding mode and to contact the hinge
region,31,32 were detected using this approach. In the case of 7,
we determined Kd ) 8.2 ( 2.9 nM, which strongly agrees with
the reported IC50 for this compound (∼9 nM)31 and demonstrates
that our alternative FLiK approach is extremely sensitive to this
ligand. Additionally, the VEGFR2 inhibitor 8 (CP-547632) was
also detected and was found to have Kd ) 99 ( 13 nM. This
compound is in clinical trials as an anticancer agent that acts
by inhibiting angiogenesis and tumor growth mediated by
VEGFR2.33 Although no crystal structure for 8 in complex with
a kinase has been published to date, previously reported
pharmacokinetic studies revealed that 8 inhibits VEGFR2 in
an ATP-competitive manner.33 Real-time kinetic measurements
of 6, 7, and 8 show rapid binding (<2 s) of all three compounds
to p38R, which is consistent with the expected Type I binding
mode. However, only 8 could be characterized using p38R
labeled at either the activation loop or the glycine-rich loop;
only the latter detected 6 and 7. This suggests that our new
approach has the added advantage of detecting certain Type I
inhibitors that may induce unique conformational changes, most
likely by making additional contacts to the acrylodan-labeled
glycine-rich loop and altering its conformation.
To determine whether detection of these compounds was the
result of direct interactions with the glycine-rich loop, binding
to the DFG-out conformation, or both, we co-crystallized 6, 7,
and 8 in complex with p38R and solved the structures to
resolutions of 1.8, 2.25, and 2.35 Å, respectively. In the p38R-7
complex (Figure 3A), the inhibitor resides in a typical Type I
binding orientation and forms hydrogen bonds to the hinge
region, analogous to those previously reported for a close
structural analogue (PDB code: 2QD9). The fluorophenyl moiety
of 7 extends beyond the gatekeeper residue and occupies the
hydrophobic subpocket, an interaction that is known to increase
the affinity of compounds for p38R.34 Although the DFG motif
is found in the active “in” conformation, the conformation of
the glycine-rich loop is significantly altered in the case of p38R
As expected, the known DFG-out binder 1 showed a distinct
time dependence (Figure S4, Supporting Information) over a
period of 5 h and was found to have a Kd value similar to that
obtained previously using activation loop-labeled p38R.14 Ad-
ditionally, 1 and 2 were found to differ mainly with respect to
koff rather than kon as the primary determinant for affinity, which
is a well-characterized observation for the binding of pyrazolo-
urea-based compounds to the DFG-out conformation of p38R.7,14
The known Type I p38R inhibitor 5 was also sensitively detected
and had a Kd value comparable to that published previously.14
We previously reported that, despite adopting a Type I binding
mode and contacting the kinase hinge region, 5 binds to p38R
and can stabilize the DFG-out conformation by forming π-π
stacking interactions between the DFG Phe169 and Tyr35 of
the glycine-rich loop,14 thus explaining the sensitive detection
of this compound using both FLiK approaches. The Type I
binding mode of 5 was easily discriminated from Type II/III
binders in a HTS format due to its more rapid binding and
dissociation kinetics and because it did not show the same Kd
time dependence as the DFG-out binder 1 (Figure S4). Thus,
the FLiK assay also allows easy preliminary assessments of
inhibitor binding mode without requiring co-crystallization of
detected ligands with the protein.
Dasatinib (4), a Type I inhibitor that binds to p38R with a
reported Kd ) 27 nM,30 was not detected using glycine-rich
loop-labeled p38R, suggesting that it adopts the expected/
published binding mode observed in the DFG-in conformation
of Abl (PDB code: 2GQG) and cSrc (PDB code: 3G5D).
However, it was surprising to observe that the indole derivative
7 and the 2-phenyl-substituted quinazoline 6, both known to
(31) Murali Dhar, T. G.; et al. Bioorg. Med. Chem. Lett. 2007, 17, 5019–
5024.
(29) Regan, J.; Breitfelder, S.; Cirillo, P.; Gilmore, T.; Graham, A. G.;
Hickey, E.; Klaus, B.; Madwed, J.; Moriak, M.; Moss, N.; Pargellis,
C.; Pav, S.; Proto, A.; Swinamer, A.; Tong, L.; Torcellini, C. J. Med.
Chem. 2002, 45, 2994–3008.
(32) Pierce, A. C.; ter Haar, E.; Binch, H. M.; Kay, D. P.; Patel, S. R.; Li,
P. J. Med. Chem. 2005, 48, 1278–1281.
(33) Beebe, J. S.; et al. Cancer Res. 2003, 63, 7301–7309.
(34) Lafont, V.; Armstrong, A. A.; Ohtaka, H.; Kiso, Y.; Mario Amzel,
L.; Freire, E. Chem. Biol. Drug Des. 2007, 69, 413–422.
(30) Karaman, M. W.; et al. Nat. Biotechnol. 2008, 26, 127–132.
9
J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010 4157

A R T I C L E S
Simard et al.
Figure 3. Crystal structures of wild-type p38R in complex with 6, 7, and 8. Electron density maps (2Fo - Fc) of p38R (gray) and each ligand (red) are
contoured at 1σ. (a) The carbonyl attached to the piperazine ring of 7 forms two hydrogen bonds (dashed red lines) to the hinge region (pink) (backbone NH
of Met109 and Gly110). The glycine-rich loop (green) folds over to directly interact with the inhibitor and shields the indole moiety and piperazine ring from
the solvent. The DFG motif (orange) is in the “in” conformation, with Asp168 pointing into the ATP binding site. (b) Similar to 7, the halogen-substituted
methoxybenzene of 8 bends around the gatekeeper (Thr106) and points into the hydrophobic subpocket. The carboxamide and the urea attached to the
isothiazole ring both form hydrogen bonds to the hinge region (backbone CO of His107, NH and CO of Met109). The pyrrolidine-butane moiety is kinked
by 90° and points away from the solvent into the ATP pocket. The glycine-rich loop is less visible in the electron density, and the DFG motif is clearly in
the “out” conformation. (c) 6 is sandwiched between the side chains of Tyr35 of the glycine-rich loop (green) and Phe169 of the DFG motif (orange) and
does not form any contacts to the hinge region (pink). The electron density as well as ESI-MS measurements proved that the electrophile at the 7-position
of 6 does not react with the protein (see Figure S1, Supporting Information). (d) Structural alignment of the p38R-6 complex and 5 (SB203580) bound to
p38R (PDB code: 3GCP).14 In the p38R-5 complex (green), the pyridine ring of the inhibitor forms an essential hydrogen bond to the hinge region (Met109)
of the kinase. The proximal phenyl substituent of 5 and the quinazoline core of 6 are isostructurally sandwiched between the glycine-rich loop and the DFG
motif to form energetically favorable π-π interactions with the side chains of Tyr35 and Phe169. In both structures, these interactions most likely stabilize
the DFG motif in the “out” conformation.
bound with 7, thus explaining its sensitive detection using only
this new FLiK approach. The glycine-rich loop is folded over
the inhibitor such that the side chain of Tyr35 partly shields
the methyl-substituted piperazine ring and the chloro-substituted
indole ring of 7 from the solvent, thereby stabilizing inhibitor
binding, most likely via hydrophobic interactions. As a conse-
quence, the removal of ordered water molecules from this
surface of the ligand results in the high-affinity binding of 7 to
p38R.
The crystal structure of 8 in complex with p38R (Figure 3b)
revealed that the kinase was stabilized in the DFG-out confor-
mation with the inhibitor bound also to the hinge region (Type
I). The carboxamide attached to the isothiazole of the inhibitor
core forms two parallel hydrogen bonds to the hinge region (CO
of His107 and NH of Met109). To the best of our knowledge,
this represents a new hinge region binding motif. In addition,
the two NH’s of the urea moiety are pointing toward the
backbone CO of Met109 and serve as hydrogen-bonding donors.
The aliphatic linker of the solubilizing pyrrolidine-butane
moiety is surprisingly not pointing toward the solvent but rather
folded inward toward the ATP pocket. The glycine-rich loop is
relatively mobile in this complex and is partially not observed
in the crystal structure. Therefore, the movement of the
activation loop to its inactive conformation and its stabilization
by 8 induces an upward movement of the acrylodan-labeled
glycine-rich loop of p38R (see Figure 1) due to the cross-talk
that exists between these two structural motifs. This explains
the sensitive detection of 8 with this new FLiK approach as
well as with our original FLiK approach in which the activation
loop of p38R was labeled (data not shown).
In the p38R-6 complex (Figure 3c), the quinazoline-based
inhibitor surprisingly adopts a binding mode analogous to that
of trisubstituted imidazoles such as 5 (SB203580).14,35 The
kinase is stabilized in the DFG-out conformation with its
quinazoline core stacked between the glycine-rich loop and the
DFG motif, thus explaining the detection of this compound using
our new FLiK approach. However, unlike 5, compound 6 does
not contact the hinge region, and the attached acrylamide extends
further in the direction of the position used to label the activation
loop in our original FLiK approach.14 Therefore, activation loop
labeling at this position may be incompatible with the binding
mode of 6 described here and might explain why we were not
able to previously detect and report the binding of 6 to p38R
(35) Simard, J. R.; Getlik, M.; Gru¨tter, C.; Pawar, V.; Wulfert, S.; Rabiller,
M.; Rauh, D. J. Am. Chem. Soc. 2009, 131, 13286–13296.
9
4158 J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010

Identifying Inhibitors That Bind to Kinase Conformations
A R T I C L E S
using our original approach of labeling the activation loop. This
unique binding mode differs significantly from those of previ-
ously reported structural analogues of 6 in complex with
GSK3ꢀ32 and calmodulin-dependent protein kinase 1D (PDB
code: 2JC6), where the amino-pyrazole moiety of the inhibitor
strongly interacts with the hinge region via three hydrogen bonds
and the kinase is found in the DFG-in conformation. Our
discovery of a completely new binding mode for quinazoline-
based compounds together with the sensitive detection of the
DFG-in binder 7 (described above) highlights the additional
advantage of our new FLiK approach to detect binders that
induce conformational changes in the glycine-rich loop. These
types of compounds may stimulate the development of new
generations of kinase inhibitors, particularly in kinases that do
not readily adopt the DFG-out conformation.
glycine-rich loop of the kinase domain, allowing the detection
and characterization of the time-dependent binding of Type II
and Type III inhibitors. The binding of these ligands causes
the activation loop to change conformation, which in turn alters
the acrylodan-labeled glycine-rich loop conformation. For BIRB-
796 (1), measured Kd, kon, and koff values obtained with this
new approach were similar to published values determined by
other methods7 and were also comparable to those obtained by
labeling the activation loop,14 clearly suggesting that labeling
kinases at the glycine-rich loop is a reliable alternative screening
tool for detecting DFG-out-binding Type II and Type III ligands
in kinases.
Although the FLiK approach was originally intended to
identify Type II and Type III inhibitors that physically occupy
the allosteric site available in the DFG-out conformation, we
have also been able to sensitively detect several Type I ligands
that contact the hinge region and stabilize the DFG-out
conformation of p38R by labeling the glycine-rich loop with
acrylodan. Details provided by protein X-ray crystallography
(compound 8) revealed that these Type I ligands stabilize the
DFG-out conformation of p38R by occupying the hydrophobic
subpocket beyond the gatekeeper residue as well as by forming
hydrophobic interactions with residues of the DFG motif
(Phe169) or at the N-terminal end of the activation loop, thereby
leaving the allosteric site completely exposed. The identification
of such Type I ligands should not be underestimated since they
might qualify as starting points for further development into
Type II inhibitors that extend in the direction of the less-
conserved allosteric site.6 Such a strategy was successfully
applied to the tyrosine kinase cSrc previously. However, in that
case, Type II inhibitors were developed from Type III hits and
extended in the direction of the hinge region of the kinase.25
The key to being able to detect Type I DFG-out binders using
FLiK approaches is the ability to perform screens using the
unphosphorylated form of the kinase in the absence of both
substrate and ATP. As mentioned above, the unphosphorylated
form of the kinase is more likely to adopt the DFG-out
conformation in which residues in the DFG motif or N-terminal
regions of the activation loop can interact with the ligand and
thus enhance affinity. This is in contrast to classical activity-
based assays that require the phosphorylated kinase, which is
more likely to be found in the DFG-in conformation, thereby
lowering the affinities of DFG-out binders and making it less
likely that such preferred hits are detected.22,36 The established
use of traditional activity-based assays in screening campaigns
desensitizes the detection of DFG-out binders and could explain
the lack of information in the literature about the binding of 8
to active (i.e., phosphorylated) kinases other than VEGFR2. The
reported high specificity of 8 has led to its application as a
VEGFR2 inhibitor in clinical trials to stop tumor growth and
proliferation by inhibiting angiogenesis. Given the submicro-
molar affinities of 8 detected using unphosphorylated p38R with
the FLiK approach, these findings could also stimulate further
studies of this clinically relevant compound or close derivatives
for the treatment of other kinase-associated diseases. Addition-
ally, the structural information provided here for 8 in complex
with p38R (i.e., new type of hinge contact) could stimulate
further medicinal chemistry efforts to build on the affine portions
of this molecule to extend into the adjacent allosteric site and
generate more pharmacologically desirable Type II inhibitors
that bind to inactive kinase conformations.
4. Discussion
Kinase inhibitor drug discovery has progressed significantly
in the past decade and resulted in almost a dozen marketed
drugs.5,8 However, several challenges such as drug resistance
and the poor selectivity of classical ATP-competitive inhibitors
remain prominent issues in the development of new kinase
inhibitors. Since most of the chemical scaffolds that qualify as
ATP-competitive ligands have already been discovered, classical
kinase activity assays, which require the kinase to be phospho-
rylated and in the active conformation, are becoming obsolete
for the discovery of novel hits that serve as lead structures for
future medicinal chemistry and chemical biology research.2
Active kinases are less likely to adopt inactive conformations,
thereby lowering the affinities of DFG-out binders22,36 and
leading to an enrichment of Type I scaffold hits that bind to
the DFG-in conformation. Current medicinal chemistry research
is focused on identifying and developing inhibitors that target
alternative binding sites and/or stabilize inactive kinase
conformations,2,37-40 highlighting the need for innovative new
approaches to detect and characterize such ligands.
We previously reported the development of the FLiK assay
as a method for detecting and discriminating Type II and Type
III DFG-out binders by directly monitoring changes in the
conformation of the activation loop (labeled with acrylodan)
induced by the binding of such ligands. However, not all kinases
can adopt the DFG-out conformation, and there is currently no
method available to predict whether kinases of interest can even
adopt the DFG-out conformation (aside from trial and error).
The identification of such kinases will only be revealed over
time as the activation loop FLiK approach is applied to more
and more kinases. To circumvent some of these issues, we
developed an alternative labeling strategy that may make FLiK
applicable to a larger number of kinases.
Using the serine-threonine kinase p38R as a model system,
we generated a second generation FLiK assay that takes
advantage of the cross-talk between the activation loop and the
(36) Seeliger, M. A.; Nagar, B.; Frank, F.; Cao, X.; Henderson, M. N.;
Kuriyan, J. Structure 2007, 15, 299–311.
(37) Adrian, F. J.; Ding, Q.; Sim, T.; Velentza, A.; Sloan, C.; Liu, Y.;
Zhang, G.; Hur, W.; Ding, S.; Manley, P.; Mestan, J.; Fabbro, D.;
Gray, N. S. Nat. Chem. Biol. 2006, 2, 95–102.
(38) Calleja, V.; Laguerre, M.; Parker, P. J.; Larijani, B. PLoS Biol. 2009,
7, 189–200.
(39) Fischmann, T. O.; Smith, C. K.; Mayhood, T. W.; Myers, J. E., Jr.;
Reichert, P.; Mannarino, A.; Carr, D.; Zhu, H.; Wong, J.; Yang, R.-
S.; Le, H. V.; Madison, V. S. Biochemistry 2009, 48, 2661–2674.
(40) Kirkland, L. O.; McInnes, C. Biochem. Pharmacol. 2009, 77, 1561–
1571.
By labeling the glycine-rich loop, we not only achieve the
goal of identifying DFG-out binders in applicable kinases but
9
J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010 4159

A R T I C L E S
Simard et al.
also allow the detection of Type I ligands that gain affinity for
the DFG-in conformation by directly inducing conformational
changes in the glycine-rich loop. This feature is the most
important advantage of this new approach over our initial FLiK
assay, in which the activation loop was labeled. By using
glycine-rich labeled p38R, we sensitively detected Type I
inhibitors such as 7, which bind to the DFG-in conformation
of p38R. Such compounds may gain affinity for the kinase by
inducing changes in the conformation of the glycine-rich loop
that help shield the ligand from the surrounding solvent.19,20,41
Since the position in the glycine-rich loop responsible for these
interactions is often conserved as an aromatic Tyr or Phe in
more than 80% of kinases, this new labeling strategy extends
the use of FLiK to additional kinases, including many kinases
that are not regulated by a readily inducible DFG-in/out
equilibrium. Detection of Type I inhibitors that interact with
the glycine-rich loop may provide insights for the development
of new scaffolds that take advantage of these interactions while
avoiding the more traditional focus on identifying new types
of hinge region contacts. Changes in glycine-rich loop confor-
mation may also provide additional ways of improving Type I
inhibitor specificities. Of course, the selection of which FLiK
labeling strategy to employ will depend on the types of ligands
that are ultimately desired for a particular kinase of interest and
which labeling strategy the target kinase tolerates best.
loop of the kinase domain. Using protein X-ray crystallography
to validate screening results obtained with this new approach,
we found that this assay detects not only Type I ligands that
stabilize the DFG-out conformation by way of a unique binding
mode but also ligands that directly interact with the glycine-
rich loop, resulting in high-affinity binding. In comparison to
our recently reported assay in which we labeled the activation
loop,11,14 this assay system also utilizes the unphosphorylated
form of the kinase and provides a powerful alternative screening
tool for sensitively detecting changes in kinase conformation
that are correlated with the binding of Type II/III inhibitors.
This approach also avoids manipulation of the activation loop,
which may not be tolerated in some kinases. Moreover, it is
not known whether most kinases can even adopt the DFG-out
conformation, thereby making the binding assay presented here
an attractive alternative approach for detecting and designing
high-affinity Type I compounds that interact directly with the
glycine-rich loop.
Acknowledgment. We thank Willem van Otterlo for helpful
discussions. This work was supported by the German Federal
Ministry for Education and Research through the German National
Genome Research Network-Plus (NGFN-Plus) (Grant No. BMBF
01GS08102). J.R.S. was funded by the Alexander von Humboldt
Foundation. Schering Plough, Bayer-Schering Pharma, Merck-
Serono, and Bayer CropScience are thanked for financial support.
5. Conclusion
Supporting Information Available: Experimental details for
the synthesis of 6; data collection and refinement statistics for
p38R co-crystallized with 6, 7, and 8; ATP Km and IC50 values
for different variants of p38R kinase; additional figures; and
complete refs 15, 28, 30, 31, and 33. This material is available
free of charge via the Internet at http://pubs.acs.org.
We have developed a powerful alternative screening method
for the rapid identification of inhibitors that (i) stabilize inactive
kinase conformations and/or (ii) interact with the glycine-rich
(41) Patel, S. B.; Cameron, P. M.; O’Keefe, S. J.; Frantz-Wattley, B.;
Thompson, J.; O’Neill, E. A.; Tennis, T.; Liu, L.; Becker, J. W.;
Scapin, G. Acta Crystallogr. D: Biol. Crystallogr. 2009, 65, 777–
785.
JA908083E
9
4160 J. AM. CHEM. SOC. VOL. 132, NO. 12, 2010