Identification of type II and III DDR2 inhibitors

Article abstract of DOI:10.1021/jm500167q

Discoidin domain-containing receptors (DDRs) exhibit a unique mechanism of action among the receptor tyrosine kinases (RTKs) because their catalytic activity is induced by extracellular collagen binding. Moreover, they are essential components in the assimilation of extracellular signals. Recently, DDRs were reported to be significantly linked to tumor progression in breast cancer by facilitating the processes of invasion, migration, and metastasis. Here, we report the successful development of a fluorescence-based, direct binding assay for the detection of type II and III DFG-out binders for DDR2. Using sequence alignments and homology modeling, we designed a DDR2 construct appropriate for fluorescent labeling. Successful assay development was validated by sensitive detection of a reference DFG-out binder. Subsequent downscaling led to convenient application to high-throughput screening formats. Screening of a representative compound library identified high-affinity DDR2 ligands validated by orthogonal activity-based assays, and a subset of identified compounds was further investigated with respect to DDR1 inhibition.

Full text of DOI:10.1021/jm500167q

Article
pubs.acs.org/jmc
Identification of Type II and III DDR2 Inhibitors
Andre Richters,† Hoang D. Nguyen,‡,§ Trang Phan,‡,§ Jeffrey R. Simard,‡,∥ Christian Grutter,†
́
̈
Julian Engel,† and Daniel Rauh*,†,‡
†Department of Chemistry and Chemical Biology, Technical University of Dortmund, Otto-Hahn-Straße 6, 44227 Dortmund,
Germany
‡Chemical Genomics Centre of the Max Planck Society, Otto-Hahn-Straße 15, 44227 Dortmund, Germany
S
* Supporting Information
ABSTRACT: Discoidin domain-containing receptors
(DDRs) exhibit a unique mechanism of action among the
receptor tyrosine kinases (RTKs) because their catalytic
activity is induced by extracellular collagen binding. Moreover,
they are essential components in the assimilation of
extracellular signals. Recently, DDRs were reported to be
significantly linked to tumor progression in breast cancer by
facilitating the processes of invasion, migration, and metastasis. Here, we report the successful development of a fluorescence-
based, direct binding assay for the detection of type II and III DFG-out binders for DDR2. Using sequence alignments and
homology modeling, we designed a DDR2 construct appropriate for fluorescent labeling. Successful assay development was
validated by sensitive detection of a reference DFG-out binder. Subsequent downscaling led to convenient application to high-
throughput screening formats. Screening of a representative compound library identified high-affinity DDR2 ligands validated by
orthogonal activity-based assays, and a subset of identified compounds was further investigated with respect to DDR1 inhibition.
monomers whose dimerization is induced by extracellular
ligand binding, as has been observed for epidermal growth
factor (EGF)/epidermal growth factor receptor (EGFR).10−12
Despite their morphological homology, the extracellular
binding of collagens versus peptidic growth factors leads to
differences in cellular function mediated by the different forms
of DDR. Tumor development and metastasis are critically
influenced by microenvironment interactions. The associated
stroma of primary tumors exhibits increased deposition of
matrix proteins, including fibrillar collagens, which correlate
with the process of metastasis in addition to alterations of cell
types and function in human breast cancer tumors.13,14 Tumor
cells are likewise known to migrate preferentially along linear
collagen fibrils near the tumor extracellular matrix (ECM).15 In
terms of ligand specificity, DDR2 is activated by ECM-
associated collagens I and III, whereas DDR1 was shown to
have a broader collagen specificity, as it is brought into action
by basement membrane (BM) components including collagen
IV.12 Epithelial−mesenchymal transitions likewise contribute to
metastasis by suppressing tumor cell−cell adhesion and by
stimulating tumor cell invasion through the BM as well as
migration through the associated stroma or ECM.16,17 DDR2
was recently shown to foster invasion, migration, and metastasis
of breast cancer cells because it regulates and stabilizes Snail1
protein levels and activity through Src-dependent stimulation of
Erk1 activity.4,18 Snail1 itself is characterized by a very short
half-life because of rapid proteasomal degradation and is
INTRODUCTION
■
The family of discoidin domain-containing receptors, com-
posed of DDR1 and DDR2, is associated with signaling
cascades that regulate cellular transformation processes, such as
epithelial−mesenchymal transition (EMT), as well as cell
adhesion, proliferation, and extracellular remodeling.1 Although
DDR1 is represented by five isoforms (DDR1a−e) generated
by alternative splicing, only a single isoform of DDR2 is
known.2 DDR1 and DDR2 differ in function by tissue-specific
divergence of their expression levels. Epithelial cells, for
instance, exhibit a high abundance of DDR1 whereas
mesenchymal cells are characterized by high levels of
DDR2.3,4 Deregulation ultimately is associated with the
progression and, in turn, metastasis of various tumor types,
including those of breast, ovary, lung, and brain origin.3−8
Structurally, these receptors are composed of an N-terminal
extracellular discoidin domain for ligand binding followed by a
short juxtamembrane (JM) domain and a single trans-
membrane (TM) domain, ensuring structural integrity. In
addition, the regulatory cytosolic JM region is capable of
controlling the catalytic activity of the receptor’s intracellular
kinase domain.
Although phylogenetic analysis links DDRs to the same
family of receptor tyrosine kinases (RTKs) as the ErbB (HER)
receptors, DDRs exhibit a unique mode of activation that
involves the extracellular binding of collagen fibrils to the
extracellular discoidin domain as opposed to the binding of
peptidic growth factors.3,8,9 Their activation kinetics are slow,
and it has been suggested that they exist as preformed dimers in
the absence of extracellular ligand in contrast to RTK
Received: January 29, 2014
Published: April 22, 2014
© 2014 American Chemical Society
4252
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262

Journal of Medicinal Chemistry
Article
Figure 1. Exemplified binding modes of type I−III kinase inhibitors and schematic diagram of the FLiK assay principle. (A) Kinase domain of DDR1
(white) in complex with type II inhibitor imatinib (yellow) representing the inactive DFG-out state (PDB code: 4BKJ). Areas covered by type I
(dasatinib; blue, PDB code: 3G5D) and III (1a; green, PDB code: 3F3U) inhibitors were superposed on the basis of available structural information.
(B) Schematic representation of the FLiK principle. The activation loop of the kinase of interest is labeled with an environmentally sensitive
fluorophore (acrylodan, colored sphere). The active state (DFG-in, left) and inactive state (DFG-out, right) of kinases are characterized by the
orientation of the DFG motif and, accordingly, the activation loop conformation. Rearrangements of the activation loop upon binding of type II/III
inhibitors (lime) can be detected by changes in the resulting fluorescence emission spectra.
predominantly expressed in tumor cells, thereby facilitating
invasive processes and migration. DDR2 significantly increases
Snail1 half-life and therefore represents a critical regulator of
breast cancer metastasis in vivo. Therefore, DDR2 has been
promoted as a potential RTK target for the treatment of breast
cancer metastasis.4 In addition, genetic alterations of DDR2
have been linked to different forms of diseases such as
nonsmall-cell lung carcinoma (NSCLC) and squamous cell
lung carcinoma (SQCLC).19−21
The role of DDR2 in tumor progression indicates the need
for identification of potent DDR2 inhibitors. Although much is
known about DDR2 biology,4,22 there is only limited
understanding of its pharmacological interactions and of the
general principles of DDR2 inhibition. Because only a few
DDR2 inhibitors are known, there is a need to identify ligands
4253
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262

Journal of Medicinal Chemistry
Article
that can serve as starting points for medicinal chemistry
approaches and pharmacological interventions. Recent reports
on classical type II inhibitors possessing inhibitory activity
against DDR2 indicate that the inactive conformation of the
kinase domain, which is characterized by an exposed Asp−
Phe−Gly motif (DFG-out), can be addressed by small
molecules.5
Technologies such as the fluorescent labels in kinases (FLiK)
approach to detect ligands capable of stabilizing inactive kinase
conformations have been successfully developed and have
already been utilized for the identification and characterization
of non-RTK inhibitors and binders.23−30 Herein, we set out to
utilize the FLiK technology to identify type II and III DFG-out
binders of DDR2 in high-throughput screening formats. These
inhibitors take advantage of the structural flexibility of DDR2
and stabilize the inactive DFG-out conformation; therefore,
they may serve as the origin of further medicinal chemistry
approaches.
genetic alterations affecting kinase function.25,36 Therefore,
Ser732 was mutated to Cys for specific reaction with acrylodan.
Additionally, a Cys at the DFG+2 position occurs naturally in at
least 40 kinases.25 Therefore, we predicted that this position
would tolerate modification without having significant effects
on the ability of the activation loop to undergo conformational
changes. To prevent the undesired labeling of multiple sites
with acrylodan,25 we analyzed the distribution of naturally
occurring cysteines in DDR2 by utilizing available structural
information of Abelson kinase (PDB code: 3DK6) to generate
a predictive homology model for DDR2 (Figure S1).19 This
model was used due to the lack of crystal structures of the
DDR2 kinase domain. Using the sequence alignments along
with the predictive homology model, seven intrinsic cysteines
were identified in the catalytic domain of DDR2 (N-terminal
lobe: C580, C644, and C651; C-terminal lobe: C716, C784,
C820, and C831). We anticipated that Cys651 would be
modified by thiol-reactive dyes because it is exposed to the
solvent, and this residue was mutated to Ser. The remaining
cysteines appeared to be buried within the protein structure
and were not likely to be nonspecifically labeled by acrylodan.
Recently, DDR1 was successfully cocrystallized in complex with
ponatinib,37 and its 3D structure was solved by X-ray structure
analysis (PDB code: 3ZOS). Comparative analysis of our
DDR2 homology model and the DDR1 crystal structure
(Figure S1) confirmed the accuracy of the predicted
orientations of the intrinsic cysteines. The accessibility of the
DFG-out conformation, crucial for the binding of small
molecules within the allosteric subpocket, was essential for
utilizing DDR2 for the FLiK approach. Available screening data
of well-known ligands that are representatives for different
types of binding modes (types I, II, II, and IV) generated by
Karaman et al.38 and Davis et al.39 suggested that DDR2 would
be able to adopt the DFG-out conformation because type II
inhibitors such as BIRB-79640 and imatinib41 were detected as
tight binders. Additionally, the protein crystal structures of
DDR1 in complex with imatinib and ponatinib clearly indicated
adoption of the DFG-out conformation upon ligand binding,
suggesting that homologue DDR2 would likewise undergo
similar conformational rearrangements. DDR1 and DDR2 were
also found to be addressed by type II inhibitors in recent
profiling studies of approved tyrosine kinase inhibitors.42,43
Considering these essential requirements, we designed a
construct composed of residues 558−855 of the catalytic
domain of DDR2 including the C651S and S732C mutations to
ensure that the protein could be specifically labeled with
acrylodan for use in the FLiK assay. The construct was
expressed using baculovirus-transfected insect cells and purified
by adapting standard chromatography techniques (see
Experimental Section).
RESULTS
■
Kinase activity is often regulated by switching between active
and inactive conformations, which are commonly referred to as
DFG-in and DFG-out states, respectively. However, some
kinases may not be able to adopt the DFG-out state (e.g.,
EGFR), and their catalytic activity is regulated through
alternative conformational changes. Classical kinase inhibitors
target the active state of the kinase, occupying the ATP binding
site, and thereby are referred to as type I inhibitors, whereas
type II and III inhibitors stabilize the inactive DFG-out state
(Figure 1A).31 The latter inhibitor categories take advantage of
the stabilization of the catalytically inactive states and are
characterized by an extended residence time of the binary
complex composed of protein and ligand, which offers the
opportunity to gain temporal target selectivity.32,33 Type IV
inhibitors are likewise useful for kinase modulation because
they allosterically impair kinase activity by binding at sites well
outside of the ATP binding cleft.23,34 However, type II and
especially type III inhibitors or binders that stabilize catalyti-
cally inactive states are particularly useful in drug development
initiatives.35
Our assay system, based on the FLiK technology, ensures
easy and economical screening for DDR2 DFG-out stabilizers,
which in turn may serve as potential starting points for
subsequent medicinal chemistry approaches. The FLiK
approach presented here relies on the labeling of the flexible
A-loop of the kinase domain with an environmentally sensitive
fluorophore that changes its fluorescent properties upon
conformational rearrangement of the loop segment that occurs
as a result of ligand binding (Figure 1B).
Selection of A-Loop Labeling Position and Construct
Design. To identify type II and III binders for DDR2, we
introduced an artificial cysteine within a flexible motif
(activation loop) of the kinase domain and selectively modified
the thiol with an environmentally sensitive fluorophore
(acrylodan (ac)). On the basis of sequence alignments of the
catalytic domain of human DDR2 and enzymatically active
domains of structurally related DDRs (Caenorhabditis briggsae,
C. vulgaris, and C. elegans), the length of the construct was
defined (residues 558−855) and DFG+2 (Ser732) was chosen
as the attachment point for acrylodan. This position has already
been identified as suitable for tagging in previous studies
because it is influenced by significant conformational changes of
the activation loop and is not classically associated with known
Fluorescent Labeling and Characterization of the ac−
DDR2 Conjugate. To ensure the structural integrity and thus,
the presence of catalytically active protein, we utilized an
activity-based assay system to quantify substrate phosphor-
ylation that confers catalytic activity for DDR2 (data not
shown). The introduced Cys732 was labeled with the thiol-
reactive fluorophore, acrylodan, and selective labeling at DFG
+2 was confirmed by mass spectrometry methods (Figure S2).
The ac−DDR2 conjugate was characterized by comparative
analysis of fluorescence spectra elicited from the DFG-in and
DFG-out conformations using methods of fluorescence spec-
troscopy. The latter state was induced by treatment with
increasing concentrations (0−2 μM) of type III DFG-out
4254
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262

Journal of Medicinal Chemistry
Article
Figure 2. End-point fluorescence measurements using ac−DDR2 labeled on the activation loop. (A) Fluorescence emission spectra of 1a-treated
ac−DDR2. Acrylodan emission at 465 nm decreases upon binding of 1a, and the global maximum is red-shifted to 476 nm. End-point equilibrium
measurements allow for direct Kd determination. (B) Ratiometric fluorescence data (R = Iλ465/Iλ510) were plotted on a logarithmic scale of inhibitor
concentration to obtain the Kd.
binder 1a (Figure 1A). Pyrazolo-urea compound 1a was chosen
as a reference because of previous studies confirming its general
DFG-out binding properties44 along with binding data toward
DDR2 that was derived from Ambit Biosciences (Kd = 15 nM).
The spectrum of the apo form exhibited a global maximum in
fluorescence intensity (FI) at 465 nm as well as a local
maximum at 510 nm (Figure 2A). Treatment with increasing
concentrations of 1a changed the resulting emission spectra.
The global maximum was red-shifted from 465 to 476 nm and
the absolute FI decreased, whereas the FI of the local maximum
remained fairly unchanged at a fixed wavelength (510 nm). We
analyzed the generated spectra using a ratiometric approach,
determining the dual-wavelength ratio of fluorescence
intensities, R = Iλ465/Iλ510, as the readout for end-point
measurements. Notably, this assay system also allowed for
direct determination of dissociation constants for DFG-out
binders (Figure 2B). In contrast, type I binders, capable of
stabilizing the DFG-in conformation, did not induce a shift of
the maximum peak, but the total FIs decreased at both 510 and
465 nm, as was previously observed for the multikinase
inhibitor, staurosporine45 (Figure S3). This type of spectral
change did not change the ratiometric readout between the two
wavelengths as it did for DFG-out binders. Thus, type I binders
were not detected using the presented FLiK approach with
DDR2, as they do not cause conformational changes with
respect to the orientation of the DFG motif.
distribution of 1a was shifted in favor of dissociation from
ac−DDR2 and a time-dependent increase in intensity was
observed as the ac−DDR2 returned to the DFG-in
conformation. This observation substantiates a reversible
binding process as 1a redistributes according to the law of
mass action and associates with the unlabeled protein. The
binding of 1a was rapid and reversible, whereas the dissociation
process was significantly slower, as is prevalent for type III
binders.
Screening Applications of ac−DDR2 in the FLiK Assay.
Previously, our laboratory applied the FLiK technology for
screening purposes using Abl, p38α, and cSrc.23,25,26 Down-
scaling of the assay setup required precise adjustment of all
assay parameters. End-point measurements using 150 nM ac−
DDR2 at saturating concentrations of reference compound 1a
provided the most accurate and strongest signal changes.
Addition of detergent (Triton X-100) to the buffer ensured
homogeneity of the reaction solution and eliminated readout
inconsistencies. To account for the variations in the emission
spectra obtained with apo ac−DDR2 and 1a-treated ac−DDR2
in either microtiter plates or cuvettes, we adjusted the
wavelength ratio for monitoring conformational changes of
the activation loop to R = Iλ452/Iλ520 (Figure S6). These
ratiometric measurements allowed for consistent readouts and
data processing. With 1a, we observed dissociation constants of
15−30 nM, which is in line with data for DDR2 ligand binding
derived from Ambit Biosciences (Kd = 15 nM).
Hit Identification and Activity-Based Validation. We
tested an internal library of 852 compounds to identify
potential DFG-out binders of DDR2. Individual compounds
were screened at a fixed concentration (10 μM) to identify
compounds that stabilize the inactive DFG-out conformation of
ac−DDR2. Background plates were used to correct for intrinsic
compound fluorescence as was done in previous work25 to
eliminate fluorescent substances from the panel of potential
binders. Any compounds that reached a set cutoff (>80% of the
response observed from reference compound 1a) were denoted
as hits; we identified 81 potential binders (hit rate: 9.5%), from
which 32 representative compounds were selected for to
follow-up studies and Kd measurements (Table 1). The quality
of the primary screening data was assessed by taking into
account 16 positive and negative controls on each plate, which
yielded a calculated Z′ factor46 of 0.84 0.02 for the entire
screen. Final validation was completed using an activity-based
assay that quantified phosphorylation of an artificial substrate
Determination of Kd and Reversibility of Binding. We
determined Kd values using ac−DDR2 protein (25 nM) in the
presence of varying concentrations of reference compound 1a
(0−2 μM). Changes in emission intensities at wavelengths
ranging from 410 to 550 nm were monitored after equilibrium
was reached (30 s). Plots of the obtained ratios (Iλ465/Iλ510) as a
function of the logarithm of the compound concentration and
subsequent fitting of the data to a Hill four-parameter equation
afforded a sigmoidal binding curve. Its inflection point
corresponded to a dissociation constant (Kd) of 3.8
0.9
nM for 1a (Figure 2B). These results were reproducible for two
different protein preparations of DDR2 and the corresponding
ac-labeled conjugate (Figure S4). Reversibility of binding was
verified by the time-dependent monitoring of fluorescence
emission of ac−DDR2 (60 nM) at 510 nm (Figure S5).
Subsequent to application of 1a (100 nM), the recorded
fluorescence emission dropped markedly over time, finally
reaching a state of equilibrium. When a 8-fold excess amount of
unlabeled DDR2 (500 nM) was added, the equilibrium
4255
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262

Journal of Medicinal Chemistry
Article
Table 1. Identity of Selected Hits for Further SAR Studies
compd
R1
R2
R3
X
1a
1b
1c
1d
1e
1f
m-NH2
m-NO2
m-Me
4-chlorphenyl
4-chlorphenyl
4-chlorphenyl
phenyl
m-NH2
m-NH2
m-NO2
p-Me
1-naphthyl
1-naphthyl
1-naphthyl
2-thiazole
1g
1h
1i
p-Me
p-CH2CO2H
p-Me
2-thiazole
2a
2b
2c
2d
2e
2f
t-Bu
t-Bu
t-Bu
N
p-Me
CF
p-CH2CO2H
m-CH2CO2H
m-CH2CO2H
p-CH2CO2H
p-CH2CO2H
m-NH2
CH
CH
CH
CH
CF
cyclopentyl
t-Bu
cyclopentyl
t-Bu
2g
3a
3b
3c
4a
4b
7a
7b
m-NH(CH2)2OCH2CH3
m-NHCOCH2CH3
m-4-quinazoline
p-4-quinazoline
H
H
Cl
m/p-ethylacetal
by DDR2 through homogeneous time-resolved FRET measure-
ments (see Experimental Section).
g, 0.178−0.428 μM) and thiazoles (1h−i, 0.702 and >1 μM)
resulted in weaker binding.
Numerous substituted pyrazolo-urea containing compounds
(1a−i, 2a−d, 2g, 4a, and 4b) were identified, and the majority
were shown to significantly bind and stabilize the inactive state,
with Kd values of 0.015−0.424 μM (Table 2). Additionally,
distinct structure−activity relationships (SARs) were revealed.
Tight binding of DDR2 was demonstrated for reference
compound 1a (Kd = 0.015 μM), whereas related derivative
1b, bearing an aromatic nitro instead of an amino group at R1,
exhibited a substantially decreased dissociation constant (0.412
μM). The presence of a methyl substituent (1c) did not affect
the binding properties, affording equally tight binding (0.015
μM). An unsubstituted phenyl at R2 (1d) led to a considerably
weakened interaction (Kd = 9.41 μM), indicating that
substitutions in the para position, as in 1a−c, are beneficial
for the binding process. In contrast, naphthyl substituents (1e−
Activity-based analyses generally reflected the observed
trends in potency. Naphthyl-decorated derivatives (1e−g)
had IC50 values between 0.332 and 0.487 μM, which
corresponded well with the observed Kd values (0.178−0.428
μM). Nevertheless, strong binding does not necessarily
correlate with robust inhibitory effects. Although tight-binding
reference compound 1a (Kd vs IC50 = 0.015 vs 0.558 μM)
represents a weak inhibitor, a methyl substituent at R1 in 1c led
to improved inhibitory activity (IC50 = 0.183 μM). Introducing
a thiazole at R2 was accompanied by a further decrease in
inhibition of DDR2, as demonstrated for 1h and 1i (4.70 μM
and 9.17 μM, respectively). Structurally related scaffolds that
are further extended at the thiazole 4-position (2a−g),27 in
turn, demonstrated effective binding to DDR2. Derivatives 2b−
d and 2g were characterized by dissociation constants of
4256
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262

Journal of Medicinal Chemistry
Article
a
Table 2. Kd and IC50 Determinations of Selected Screening Hits
Kd (μM)
IC50 (μM)
compd
1a
ac−DDR2 conjugate
0.015
DDR2WT
DDR2T654M
0.558 0.109
0.856 0.170
0.183 0.027
9.41 2.10
0.717 0.568
1.233 0.568
0.808 0.176
2.50 0.56
10.8 3.70
≥100
1b
1c
1d
1e
1f
0.412
0.015
>1
0.178
0.487 0.170
0.332 0.047
0.408 0.098
4.70 1.08
0.428
1g
1h
1i
0.252
≥100
0.702
1.15 0.49
0.82 0.17
0.002 0.0003
0.23 0.08
0.67 0.13
0.38 0.13
1.67 0.11
1.84 0.97
0.076 0.023
>1
9.17 1.85
2a
2b
2c
2d
2e
2f
0.063
0.019 0.010
3.04 0.72
0.406
0.382
2.23 0.15
0.385
1.56 0.47
>1
10.7 3.80
>1
2.17 0.48
2g
0.424
1.23 0.27
b
3a
3b
0.020
0.008 0.002
14.4 4.20
0.072 0.010
b
c
>1
b
3c
4a
0.030
0.029 0.007
0.002 0.001
0.003 0.003
0.687 0.033
0.073 0.018
>1
4b
0.093
0.001 0.0003
c
5
0.915
10.9 2.00
c
c
6
>1
7a
>1
80.8 4.00
≥100
c
7b
>1
9.4 2.1
c
c
8
>1
CP547632
BIRB-796
dasatinib
tozasertib
pazopanib
nilotinib
sorafenib
0.422 (N/A)
0.021 (0.033)38
0.018 (0.003)38
>2.5 (0.23)38,39
0.581 (0.098)38,39
0.075 (0.006−0.033)39,56
0.313 (0.007)38,39
0.024 0.002
0.010 0.004
0.0004 0.0003
0.241 0.071
0.050 0.016
0.002 0.001
0.007 0.006
0.450 0.111
0.013 0.005
0.833 0.139
0.018 0.016
0.051 0.021
0.018 0.014
0.036 0.019
a
b
Values in parentheses represent previously reported Kd values with respect to DDR2 wt. N/A: not available. Structurally related to BIRB-796 and
c
therefore included for SAR studies. No inhibition at 100 μM compound concentration.
approximately 0.4 μM, whereas 2a displayed ∼6-fold stronger
binding (Kd = 0.063 μM). This trend was consistent with the
evaluated IC50 data, illustrating strong inhibitory activity for 2a
(IC50 = 0.019 μM) and 65−160-fold less potent inhibition for
2b−d and 2g, respectively (IC50 = 1.23−3.04 μM).
previously.25 Despite recording background fluorescence to
correct for highly fluorescent substances, these compounds
appeared as false positives. Imidazopyrazine 6 illustrated
analogue characteristics, but in validation studies, it exhibited
weak binding (>1 μM) accompanied with a lack of any
inhibitory activity. To address the issue of highly fluorescent
compounds and to minimize the number of false positives, we
recently reported the replacement of acrylodan with other red-
shifted fluorophores for tagging in FLiK methodology.29
Because BIRB-796 has been identified as a DDR2 binder, we
incorporated 3a−c into the validation process due to their
structural similarity (for synthetic procedures, see the
Supporting Information). These compounds, in addition to
1a and 1c, performed potently in both binding and inhibition,
as demonstrated for 3a and 3c (Kd = 0.020 and 0.030 μM,
respectively). Nevertheless, different substituents at R1, as in 3b
which is decorated with an ether, led to remarkably decreased
binding affinity (>1 μM). This trend was clearly underlined by
subsequent IC50 determinations, illustrating significant inhib-
ition by 3a and 3c (IC50 = 0.008 and 0.029 μM, respectively)
but not for 3b (14.4 μM).
Compounds combining classical type I (quinazoline) and III
(pyrazolo-urea) binding motifs, such as 4a and 4b, inhibited
KIT activity in previous work.47 Here, we found that they
equally modulated DDR2 activity. However, important differ-
ences were defined by comparative analysis of both binding and
activity-based assays. Although we observed differences in
binding affinities for 4a and 4b (>1 vs 0.093 μM, respectively),
they were equally potent on DDR2 in activity-based systems
(IC50 = 0.002−0.003 μM). This observation most likely
correlated with the structural features that caused 4a to be
structurally rigid, whereas the flexible central element of 4b
allowed for conformational adaptation to various binding states.
Apart from molecules exhibiting the pyrazolo-urea motif, we
identified CP547632 as a novel binder (Kd = 0.422 μM) and
inhibitor (IC50 = 0.024 μM) of DDR2.
Compounds with high intrinsic fluorescence, such as
flavanones 7a,b and thieno pyridine 8, were identified because
of their interference with acrylodan fluorescence, as shown
Notably, tozasertib48 and dasatinib,49 although representing
classical type I inhibitors, caused significant changes in
4257
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262

Journal of Medicinal Chemistry
Article
Figure 3. Radiometric IC50 measurements of 2a, 4a, and 4b on DDR1 (A) and DDR2 (B).
fluorescence emission upon binding and were likewise detected
with our assay setup. Highly potent type I binders may be
identified in our FLiK assay if they can trigger other
conformational rearrangements of the kinase. Nevertheless,
our findings essentially support this strategy for the enrichment
of type II and III binders for further medicinal chemistry and
chemical biology approaches.
ylation. Our studies revealed inhibitory effects on both
receptors (Figure 3). While 4a and 4b demonstrated inhibitory
activity on DDR1 with IC50 values of 0.235 and 0.039 μM,
respectively, thiazolo urea 2a exhibited similar potency (0.068
μM). As anticipated from our binding and activity-based assay
studies, DDR2 was also potently inhibited, with IC50 values of
0.075, 0.018, and 0.065 μM for 4a, 4b, and 2c, respectively, thus
further validating our successful application of the FLiK
approach.
Because the incidence of drug-resistant mutations remains a
major challenge in targeted cancer therapy and because in vitro
studies on DDR2-dependent lung cancer cell lines exhibited the
potential emergence of gatekeeper mutations upon dasatinib
treatment,22 we explored the inhibitory effect of our identified
DDR2 inhibitors toward the gatekeeper mutant variant DDR2
T654M. Compound 4b is equipped with a flexible 1,4-fused
phenyl ring as a central structural element and was already
shown to be capable of overcoming gatekeeper-associated
resistances in cSrc T338M, Abl T315I, and KIT T670I by
avoiding steric hindrance with bulky gatekeepers through
simple rotation of the central element, as confirmed by protein
X-ray crystallography.47 Likewise, 4b completely retained its
inhibitory activity on DDR2 T654M (IC50 = 0.001 μM), which
indicates a possible future impact of the presented approach for
identifying type II and III inhibitors that may be of benefit for
first- and second-line therapies targeting DDR2. On the
contrary, the analogous rigid 1,3-fusion of the central phenyl
CONCLUSIONS
■
RTKs are central regulators of proliferation, morphological
transitions, migration, and survival. When these interrelated
cellular signaling processes that transduce extracellular stimuli
are deregulated, the onset of various diseases including cancer
can occur, as has been confirmed for various receptors (e.g.,
EGFR and KIT).52 DDRs have been linked to diseases such as
breast cancer and NSCLC even though these receptors are
underrepresented as propagated targets in cancer therapy
because their associated tumor biology remains unclear.4,50,53
Here, we have presented the successful application of the
FLiK technology to DDR2, which was recently shown to be a
key regulator of metastasis and invasion of breast cancer cells.4
Subsequent to appropriate construct design, we produced a
modified variant of the catalytic domain of huDDR2 capable of
being selectively labeled with acrylodan at the DFG+2 position
(A-loop). We confirmed that the ac−DDR2 conjugate
exhibited characteristic changes in fluorescence emission
spectra, which reflected the adoption of the respective DFG-
in and DFG-out conformations. Binding of reference
compound 1a induced the inactive DFG-out state of ac−
DDR2. Furthermore, reversibility of binding was demonstrated
upon the addition of unlabeled DDR2. In addition to the
determination of dissociation constants, this new assay system
could be utilized to screen for new binders of DDR2. Thirty-
two promising binders, stabilizing the enzymatically inactive
conformation, were identified and further investigated for their
inhibitory effects on DDR2. Allosteric inhibitors are one focus
of modern kinase inhibitor research and therefore the portion
of identified type III-like fragments (1a−i), which generally
reveal unique binding characteristics (e.g., slow off-rates), was
of particular interest.35 Optimized type III inhibitors, which
exclusively bind to the allosteric site, allowed us to address less
highly conserved sites adjacent to the active site and therefore
to increase kinase inhibitor selectivity.54,55
ring, as in 4a, led to 35-fold decreased inhibitory effect (IC50
=
0.073 μM) on DDR2 T654M as compared to the wild-type
variant (Table 2). Notably, type III-like compounds exhibiting
sterically demanding naphthyl substituents (e.g., 1e−h)
induced decreased inhibitory activity toward DDR2 T654M,
whereas type II-like inhibitors featuring thiazole extensions
(2a−g) and spanning into the ATP-binding cleft elicited
increased inhibitory activity toward gatekeeper-mutated
compared to wild-type DDR2 (e.g., 2a, 0.002 vs 0.019 μM,
T654M vs wild type).
Follow-up Characterization of the Most Potent Hits
toward DDR1/2. Considering the remarkable homology of
DDRs with respect to their catalytic domains (68% sequence
identity), we hypothesized that inhibitors of DDR2 may
operate equally well on DDR1, which is highly expressed in
various types of cancer cells, including those characterized by
massive invasiveness.50 Thus, although DDR1 and DDR2 are
indicated to be of therapeutic value, the mechanisms by which
they contribute to tumor progression remain unclear. Potent
and selective DDR1 inhibitors have recently been developed
and substantiate the importance and interest in DDR for future
therapies.50,51 For this reason, we sought to test 2a, 4a, and 4b
on DDR1/2 in orthogonal assays (Reaction Biology Corp.)
utilizing radiolabeled ATP to quantify substrate phosphor-
Although tight binding of DDR2 (Kd = 0.015−0.702 μM)
was observed for 1a−c and 1e−h, these DFG-out binders were
found to be moderately potent in activity-based assays (IC50
=
4258
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262

Journal of Medicinal Chemistry
Article
to the instructions of the supplier (BacMagic transfection kit, Merck
Millipore). Virus production was initiated subsequent to cell
attachment and medium (Gibco Sf-900 III SFM, Life Technologies)
removal by adding the transfection mix and incubating at 28 °C in a
humidified container. After 6 h, 2.5 mL of fresh medium supplemented
with 1% FBS was added, and the incubation was continued for 5 days.
The virus-containing supernatant was collected and filtered, and the
titer was determined using the limiting dilution method. Infected Sf21
cells were visualized by fluorescence microscopy, allowing for eGFP
detection of the His−eGFP−huDDR2 fusion protein. Prior to
expression, Sf21 suspensions (700−850 mL) were incubated at 28
°C under constant shaking in Erlenmeyer flasks (5 L) or spinner flasks
(2 L) until they reached a cell density of 1.8−2.5 × 106 cells/mL. Virus
solution corresponding to an MOI of 1 was then added to the cell
suspension, and the mixture was further incubated at 28 °C until
harvest (48−72 h). Cells were collected by centrifugation and
resuspended in chilled lysis buffer (400 mM NaCl, 5% glycerol, 1
mM DTT, 0.1% Triton X-100, 20 μM PMSF, 40 mM imidazole, and
50 mM Tris-HCl, pH 8.0). After repeated homogenization (M-110S
Microfluidizer, Microfluidics, Newton, MA), cell debris and insoluble
components were removed via centrifugation (17 000 rpm, 4 °C), and
the cleared lysate was kept on ice. The lysate was loaded onto a 30 mL
Ni-NTA column (self-packed), washed with 3 column volumes of
washing buffer (400 mM NaCl, 5% glycerol, 1 mM DTT, 40 mM
imidazole, and 40 mM Tris-HCl, pH 8.0), and subsequently eluted
with an increasing gradient of imidazole. The combined protein-
containing fractions were dialyzed (200 mM NaCl, 2% glycerol, 1 mM
DTT, 1 mM EDTA, and 20 mM Tris-HCl, pH 8.0) at 4 °C, and the
protein was subsequently cleaved by PreScission Protease using
dialysis buffer containing 0.05% Triton X-100. After dialysis in washing
buffer to remove remaining EDTA, the resulting fraction was loaded
onto the Ni-NTA column again, and the flow-through, containing
huDDR2, was collected. After concentration (10k MWCO Centricon)
to a volume of ∼2 mL, the remaining sample was passed through a
Sephadex HiLoad 26/60 Superdex 75 column equilibrated with size-
exclusion buffer (100 mM NaCl, 5% glycerol, and 50 mM Tris-HCl,
pH 8.0). The eluted protein was then concentrated and labeled with
acrylodan or aliquoted, flash frozen, and stored at −80 °C.
The labeling reaction was performed by combining protein and
acrylodan (dissolved in DMF) at a ratio of 1:1.5 in labeling buffer (400
mM NaCl, 5% glycerol, and 50 mM HEPES, pH 7.0) and allowing the
reaction to proceed in the dark overnight at 4 °C. Conjugated protein
was concentrated, and simultaneously an excess of acrylodan was
removed by repeated washing in a 10k MWCO Centricon using
storage buffer (200 mM NaCl, 20% glycerol, and 50 mM HEPES, pH
7.0). The sample was aliquoted and stored at −80 °C upon further use.
Monolabeling of the protein was verified by ESI−MS and mass
spectroscopic analysis (see the Supporting Information). Processing of
the raw data was performed with Xcalibur version 1.4 SR1 (Thermo
Electron Corporation), and deconvolution was done using MagTran
(freeware).
Acrylodan-labeled DDR2 conjugate was excited at the known
wavelength (368 nm), and emission spectra were recorded in the
absence and presence of increasing concentrations of DFG-out binder
1a (0−2 μM) to determine how the emission changes as the kinase
shifts from the DFG-in to the DFG-out conformation. These
measurements also revealed the most sensitive wavelength to monitor
fluorescence changes in real-time (kinetics, λ = 510 nm) as well as
assessed the feasibility of using ratiometric fluorescence to plot binding
curves (Kd, R = Iλ465/Iλ510). Excitation and emission slits were set to 3
and 5 nm, respectively, as described elsewhere.25
0.183−4.70 μM). In contrast, type II-like molecules 2a, 4a, and
4b were found to be excellent binders (Kd = 0.020−0.063 μM),
and activity-based studies illustrated their conserved impact on
active DDR2 (IC50 = 0.008−0.029 μM). We predicted these
inhibitors to reveal similar effects on structurally related DDR1,
which was investigated by radiometric studies by quantifying
substrate phosphorylation in the presence of the inhibitors. The
inhibitory effects were characterized by IC50 values ranging
from 0.039 to 0.235 μM, substantially supporting our concept
that the inhibitory features observed for DDR2 were applicable
to DDR1. Inhibitors 4a,b had previously been found to be
effective inhibitors of the wild-type forms of cSrc, Abl, and KIT.
Furthermore, the inhibitory effect of 4b against the gatekeeper
mutations as well as a gain of function mutation was conserved,
which was demonstrated in activity-based assay systems as well
as cellular formats using GIST cells.47 Because the emergence
of drug resistance continues to be a challenge in cancer therapy,
4b may offer a future approach toward adequate inhibitors to
overcome the impact of potential resistant variants of DDR.
These potential future applications are further encouraged by a
recently released study using DDR2-dependent lung cancer cell
lines to model acquired resistance for standard dasatinib
therapy.22 In particular, gatekeeper mutation T654I was
demonstrated to arise during treatment and may represent a
future challenge in anti-DDR2-directed cancer therapy. Because
4b has already been shown to exhibit the ability to overcome
gatekeeper-related drug resistance mechanisms and further-
more has revealed inhibitory activity against gatekeeper mutant
DDR2 (T654M) (IC50 = 0.002 μM), we assume that this type
of inhibitor will be of considerable value in future approaches.
To a large extent, the contribution of DDR1/2 to tumor
progression is unclear, but our presented FLiK approach
provides an easily accessible tool for identifying novel binders
for the receptor tyrosine kinase, DDR2. We anticipate that
these compounds will be useful for medicinal chemistry as well
as for chemical biology approaches such as the development of
probe molecules to decipher the orchestrated network of DDR
signaling and its relevance toward tumor biology.
EXPERIMENTAL SECTION
■
Reagents and Materials. Unless otherwise noted, all reagents and
solvents were purchased from Acros, Fluka, Sigma-Aldrich, or Merck
and were used without further purification. The fluorophore, 6-
acryloyl-2-dimethylaminonaphthalene (acrylodan), was purchased
from Invitrogen GmbH (Karlsruhe, Germany). Cuvettes and mini
stir bars were obtained from Carl Roth GmbH (Karlsruhe, Germany).
Small-volume (20 μL fill volume) black flat-bottom 384-well plates
were obtained from Greiner Bio-One GmbH (Solingen, Germany) or
Corning Inc. (New York, USA). All supplies for the DDR2 HTRF
̀
assay kit were purchased from CisBio (Bagnols-sur-Ceze, France).
Dasatinib, tozasertib, pazopanib, nilotinib, and sorafenib were
purchased from LC Laboratories (Woburn, MA). CP547623 was
provided by Merck Serono, and BIRB-796 was synthetically produced
as stated elsewhere.27 All compounds and screening hits exhibit a
purity of ≥95% confirmed by LC−MS analysis on an LCQ Advantage
MAX (1200 series, Agilent) with an Eclipse XDB-C18-column (5 μM,
150 × 1.6 mm, Phenomenex).
FLiK Assay for End-Point and Kinetic Measurements in
Cuvettes. Fluorescent-labeled DDR2 (25 nM) and various
concentrations of inhibitor in buffer (200 mM NaCl and 50 mM
HEPES, pH 7.45) were incubated in the dark at 4 °C overnight before
end-point fluorescence measurements were carried out in polystyrene
cuvettes to determine the Kd of each compound. The same buffer that
was used to wash and store the kinase was also used for all
measurements. The long incubation times were required to avoid the
Kd time dependence of type II inhibitors binding to DDR2.40 A series
Expression, Purification, Fluorescent Labeling, and Charac-
terization of DDR2. The catalytic domain of human DDR2
(huDDR2, residues 558−855) was expressed using an N-terminal
His−eGFP−DDR2 construct containing the mutations required for
specific labeling of the activation loop (C651S and S732C), in which
the fusion construct contained a cleavage site of PreScission protease
between eGFP and DDR2. Protein expression was performed in
baculovirus-infected Sf21 insect cells using BacMagic DNA according
4259
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262

Journal of Medicinal Chemistry
Article
of cuvettes containing different amounts of inhibitor was prepared
using inhibitor stocks (0.01, 0.1, 1.0, and 10.0 mM prepared in
DMSO), ensuring that the percent v/v DMSO did not exceed 0.2%.
All cuvette measurements were performed with a JASCO FP-6500
fluorescence spectrophotometer (JASCO GmbH, Groß-Umstadt,
Germany). For acrylodan-labeled DDR2, ratiometric fluorescence
values enabled reliable binding curves to be plotted to determine the
Kd of ligand binding to DDR2 directly. Where indicated, ratiometric
fluorescence values obtained for unsaturated and DDR2-saturated in
the DFG-out conformation were given as the percent of DDR2 bound
by the ligand, as was demonstrated before.25 Association and
dissociation of allosteric binders from DDR2 was qualitatively
determined for 1a using the cuvette format, and fluorescence changes
were monitored in real time. A mini stir bar was placed in the bottom
of each cuvette to ensure rapid mixing as inhibitor or protein was
delivered through the injection port located above the cuvette. The
dissociation of allosteric binders was illustrated by adding an ∼8-fold
excess of unlabeled DDR2 to a suspension of fluorophore-labeled
DDR2 (60 nM) complexed with inhibitor 1a (100 nM) while
monitoring the fluorescence intensity at 510 nm in real time. The
addition of excess unlabeled kinase (500 nM) causes the inhibitor to
redistribute, resulting in a net dissociation of inhibitor from labeled
DDR2.
FLiK Assay for Screening Applications. First, a primary screen
was carried out using a single concentration (10 μM) of each library
compound to determine which compounds induce and stabilize the
DFG-out conformation of DDR2. Inhibitor dilutions were prepared by
transferring 0.5 μL of each compound from stock plates (1 compound
per well at 0.2 mM in DMSO) into assay plates containing 9.5 μL of
acrylodan-labeled DDR2 (150 nM) in buffer (50 mM HEPES, pH
7.45, 200 mM NaCl, and 0.002% Triton X-100) or plain buffer for two
independent sets of background measurements (final DMSO content,
5%). All plates also contained 16 wells of negative DMSO control (no
ligand) as well as 16 wells of positive control (1a, 1 μM).
Homogeneity of the reaction solutions was ensured by shaking all
plates in orbital mode using an Eppendorf mixmate (1000 rpm, 30 s).
Subsequently, the plates were covered with adhesive foil and stored at
8 °C for 8 h because DFG-out binders are known to have slow
association rates.25 A Tecan infinite M1000 instrument was used to
measure the fluorescence readout in the 384-well plate format. Data
were first processed by subtracting intrinsic compound fluorescence at
452 and 520 nm from the signal intensity (I) measured in the presence
of acrylodan-labeled DDR2. As described previously, the extent of
binding was then assessed by taking the corrected ratiometric
fluorescence (R = Iλ452/Iλ520).25 Any compound that reached 80% of
the maximal response observed for the positive control was submitted
for further screening. The secondary screen was also carried out in
384-well plates using a range of concentrations of each ligand (0.5 nM
to ∼2.5 μM) to generate binding curves or to identify false hits picked
up because of high intrinsic fluorescence that was not adequately
subtracted by using background plates. Plates were sealed, incubated,
and measured as described for the primary screen. Raw data at 452 and
520 nm as well as background-corrected ratiometric data were used to
eliminate false fluorescent hits. Ratiometric fluorescence values
enabled binding curves to be plotted to determine the Kd of ligand
binding to DDR2 directly.
ATP and substrate peptide. A Tecan infinite M1000 plate reader was
used to measure the fluorescence of the samples at 620 nm (Eu-
labeled antibody) and 665 nm (XL665-labeled streptavidin) 60 μs
after excitation at 317 nm. The quotient of both intensities for
reactions made with eight different inhibitor concentrations was fit to a
Hill four-parameter equation to determine IC50 values. Each reaction
was performed in duplicate, and at least three independent
determinations of each IC50 were made.
Orthogonal activity-based evaluation was carried out for DDR1 and
DDR2 using the kinase profiling service from Reaction Biology
Corporation. Briefly, a mixture of an artificial substrate peptide in base
reaction buffer (20 mM HEPES, pH 7.5, 10 mM MgCl2, 1 mM EGTA,
0.02% Brij35, 0.02 mg/mL BSA, 0.1 mM Na3VO4, 2 mM DTT, and
1% DMSO) and the kinase of interest was prepared and subsequently
treated with the compounds to be investigated. The kinase reaction
was initiated by adding γ-33P-ATP (specific activity, 10 μCi/μL). After
a 120 min incubation at rt, the reactions were spotted onto P81 ion-
exchange paper (Whatman no. 3689-915), which was then extensively
washed in 0.75% phosphoric acid.
ASSOCIATED CONTENT
* Supporting Information
■
S
Detailed synthetic procedures for the preparation of 3b and 3c
(including synthetic Scheme S1), a homology model of DDR2,
MS analytics of labeled and unlabeled DDR2, fluorescence
emission spectra of ac−DDR2 and type I inhibitor
staurosporine, independent Kd determinations of 1a-treated
ac−DDR2, and data on the reversibility of binding for 1a to
ac−DDR2 as well as the miniaturization of the FLiK assay to
microplates. This material is available free of charge via the
Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*Phone: +49 (0) 231-755 7080. Fax: +49 (0) 231-755 7082. E-
mail: daniel.rauh@tu-dortmund.de.
■
Present Addresses
§(H.D.N. and T.P.) Ho Chi Minh City University of Science,
VNU-HCMC, 227 Nguyen Van Cu Street, District 5, Ho Chi
Minh City, Vietnam.
∥(J.R.S.) Amgen, Inc., 360 Binney Street, Cambridge,
Massachusetts 02142, United States.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was cofunded by the German Federal State of North
Rhine−Westphalia (NRW) and the European Union (Euro-
pean Regional Development Fund: Investing In Your Future),
through the German Federal Ministry for Education and
Research (NGFNPlus and e:Med) (grant nos. BMBF
01GS08104 and 01ZX1303C).
Activity-Based Assay for IC50 Determination. IC50 determi-
nations for DDR2 wt and DDR2 T654M (Proqinase, no. 0771-0000-1,
lot 002 and no. 1193-0000-1, lot 001) were performed with the
KinEASE-TK assay from Cisbio according to the manufacturer’s
instructions. A biotinylated substrate peptide was phosphorylated by
the specific kinase of interest. After completion of the reaction, an anti-
phosphotyrosine antibody labeled with europium cryptate and
streptavidin labeled with the fluorophore XL665 were added. FRET
between europium cryptate and XL665 was measured to quantify the
phosphorylation of the substrate peptide. ATP concentrations were
adjusted to 8 μM (DDR2 wt, 2 × Km) and 1 μM (DDR2 T654M, 1 ×
Km) while a substrate concentration of 350 nM (DDR2, 1 × Km) and
700 nM (DDR2 T654M, 1 × Km) was used. Kinase and inhibitor were
preincubated for 30 min before the reaction was started by addition of
ABBREVIATIONS USED
■
Abl, Abelson kinase; KIT, mast/stem cell growth factor
receptor; DDR1/2, discoidin domain-containing receptor 1/2;
ECM, extracellular matrix; EGF, epidermal growth factor;
EGFR, epidermal growth factor receptor; EMT, epithelial−
mesenchymal transition; ErbB/HER, receptor tyrosine-protein
kinase; Erk1, mitogen-activated protein kinase 3; FI,
fluorescence intensity; FLiK, fluorescent labels in kinases;
FRET, Forster-resonance energy transfer; HTRF, homoge-
̈
neous time-resolved FRET; HTS, high-throughput screening;
NSCLC, nonsmall cell lung cancer; p38α, mitogen-activated
4260
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262

Journal of Medicinal Chemistry
Article
Lim, S. M.; Heynck, S.; Peifer, M.; Simard, J. R.; Lawrence, M. S.;
Onofrio, R. C.; Salvesen, H. B.; Seidel, D.; Zander, T.; Heuckmann, J.
M.; Soltermann, A.; Moch, H.; Koker, M.; Leenders, F.; Gabler, F.;
protein kinase 14; RTK, receptor tyrosine kinase; SAR,
structure−activity relationship; Snail1, zinc finger protein
SNAI1; Src, proto-oncogene tyrosine-protein kinase Src;
SQCLC, squamous-cell lung carcinoma; TM, transmembrane
́
Querings, S.; Ansen, S.; Brambilla, E.; Brambilla, C.; Lorimier, P.;
Brustugun, O. T.; Helland, A.; Petersen, I.; Clement, J. H.; Groen, H.;
Timens, W.; Sietsma, H.; Stoelben, E.; Wolf, J.; Beer, D. G.; Tsao, M.
S.; Hanna, M.; Hatton, C.; Eck, M. J.; Janne, P. A.; Johnson, B. E.;
Winckler, W.; Greulich, H.; Bass, A. J.; Cho, J.; Rauh, D.; Gray, N. S.;
Wong, K. K.; Haura, E. B.; Thomas, R. K.; Meyerson, M. Mutations in
the DDR2 kinase gene identify a novel therapeutic target in squamous
cell lung cancer. Cancer Discovery 2011, 1, 78−89.
REFERENCES
■
(1) Walsh, L. A.; Nawshad, A.; Medici, D. Discoidin domain receptor
2 is a critical regulator of epithelial−mesenchymal transition. Matrix
Biol. 2011, 30, 243−247.
(2) Vogel, W. F.; Abdulhussein, R.; Ford, C. E. Sensing extracellular
matrix: an update on discoidin domain receptor function. Cell.
Signalling 2006, 18, 1108−1116.
(20) Davies, H.; Hunter, C.; Smith, R.; Stephens, P.; Greenman, C.;
Bignell, G.; Teague, J.; Butler, A.; Edkins, S.; Stevens, C.; Parker, A.;
O’Meara, S.; Avis, T.; Barthorpe, S.; Brackenbury, L.; Buck, G.;
Clements, J.; Cole, J.; Dicks, E.; Edwards, K.; Forbes, S.; Gorton, M.;
Gray, K.; Halliday, K.; Harrison, R.; Hills, K.; Hinton, J.; Jones, D.;
Kosmidou, V.; Laman, R.; Lugg, R.; Menzies, A.; Perry, J.; Petty, R.;
Raine, K.; Shepherd, R.; Small, A.; Solomon, H.; Stephens, Y.; Tofts,
C.; Varian, J.; Webb, A.; West, S.; Widaa, S.; Yates, A.; Brasseur, F.;
Cooper, C. S.; Flanagan, A. M.; Green, A.; Knowles, M.; Leung, S. Y.;
Looijenga, L. H.; Malkowicz, B.; Pierotti, M. A.; Teh, B. T.; Yuen, S.
T.; Lakhani, S. R.; Easton, D. F.; Weber, B. L.; Goldstraw, P.;
Nicholson, A. G.; Wooster, R.; Stratton, M. R.; Futreal, P. A. Somatic
mutations of the protein kinase gene family in human lung cancer.
Cancer Res. 2005, 65, 7591−7595.
(3) Valiathan, R. R.; Marco, M.; Leitinger, B.; Kleer, C. G.; Fridman,
R. Discoidin domain receptor tyrosine kinases: new players in cancer
progression. Cancer Metastasis Rev. 2012, 31, 295−321.
(4) Zhang, K.; Corsa, C. A.; Ponik, S. M.; Prior, J. L.; Piwnica-
Worms, D.; Eliceiri, K. W.; Keely, P. J.; Longmore, G. D. The collagen
receptor discoidin domain receptor 2 stabilizes SNAIL1 to facilitate
breast cancer metastasis. Nat. Cell Biol. 2013, 15, 677−687.
(5) Day, E.; Waters, B.; Spiegel, K.; Alnadaf, T.; Manley, P. W.;
Buchdunger, E.; Walker, C.; Jarai, G. Inhibition of collagen-induced
discoidin domain receptor 1 and 2 activation by imatinib, nilotinib and
dasatinib. Eur. J. Pharmacol. 2008, 599, 44−53.
(6) Ford, C. E.; Lau, S. K.; Zhu, C. Q.; Andersson, T.; Tsao, M. S.;
Vogel, W. F. Expression and mutation analysis of the discoidin domain
receptors 1 and 2 in non-small cell lung carcinoma. Br. J. Cancer 2007,
96, 808−814.
(7) Goldsmith, E. C.; Zhang, X.; Watson, J.; Hastings, J.; Potts, J. D.
The collagen receptor DDR2 is expressed during early cardiac
development. Anat. Rec. 2010, 293, 762−769.
(21) Drilon, A.; Rekhtman, N.; Ladanyi, M.; Paik, P. Squamous-cell
carcinomas of the lung: emerging biology, controversies, and the
promise of targeted therapy. Lancet Oncol. 2012, 13, e418−e426.
(22) Beauchamp, E. M.; Woods, B. A.; Dulak, A. M.; Tan, L.; Xu, C.;
Gray, N. S.; Bass, A. J.; Wong, K. K.; Meyerson, M.; Hammerman, P.
S. Acquired resistance to dasatinib in lung cancer cell lines conferred
by DDR2 gatekeeper mutation and NF1 loss. Mol. Cancer Ther. 2013,
13, 475−482.
(8) Ren, T.; Zhang, J.; Liu, X.; Yao, L. Increased expression of
discoidin domain receptor 2 (DDR2): a novel independent prognostic
marker of worse outcome in breast cancer patients. Med. Oncol. 2013,
30, 397.
(9) Vogel, W.; Gish, G. D.; Alves, F.; Pawson, T. The discoidin
domain receptor tyrosine kinases are activated by collagen. Mol. Cell
1997, 1, 13−23.
(23) Schneider, R.; Becker, C.; Simard, J. R.; Getlik, M.; Bohlke, N.;
Janning, P.; Rauh, D. Direct binding assay for the detection of type IV
allosteric inhibitors of Abl. J. Am. Chem. Soc. 2012, 134, 9138−9141.
(24) Getlik, M.; Grutter, C.; Simard, J. R.; Kluter, S.; Rabiller, M.;
̈
̈
Rode, H. B.; Robubi, A.; Rauh, D. Hybrid compound design to
overcome the gatekeeper T338M mutation in cSrc. J. Med. Chem.
2009, 52, 3915−3926.
(10) Noordeen, N. A.; Carafoli, F.; Hohenester, E.; Horton, M. A.;
Leitinger, B. A transmembrane leucine zipper is required for activation
of the dimeric receptor tyrosine kinase DDR1. J. Biol. Chem. 2006,
281, 22744−22751.
(25) Simard, J. R.; Grutter, C.; Pawar, V.; Aust, B.; Wolf, A.; Rabiller,
̈
M.; Wulfert, S.; Robubi, A.; Kluter, S.; Ottmann, C.; Rauh, D. High-
̈
(11) Zhang, X.; Gureasko, J.; Shen, K.; Cole, P. A.; Kuriyan, J. An
allosteric mechanism for activation of the kinase domain of epidermal
growth factor receptor. Cell 2006, 125, 1137−1149.
throughput screening to identify inhibitors which stabilize inactive
kinase conformations in p38alpha. J. Am. Chem. Soc. 2009, 131,
18478−18488.
(26) Simard, J. R.; Kluter, S.; Grutter, C.; Getlik, M.; Rabiller, M.;
Rode, H. B.; Rauh, D. A new screening assay for allosteric inhibitors of
(12) Leitinger, B. Transmembrane collagen receptors. Annu. Rev. Cell
Dev. Biol. 2011, 27, 265−290.
̈
̈
(13) Kauppila, S.; Stenback, F.; Risteli, J.; Jukkola, A.; Risteli, L.
̈
cSrc. Nat. Chem. Biol. 2009, 5, 394−396.
Aberrant type I and type III collagen gene expression in human breast
cancer in vivo. J. Pathol. 1998, 186, 262−268.
(27) Getlik, M.; Grutter, C.; Simard, J. R.; Nguyen, H. D.; Robubi,
̈
A.; Aust, B.; van Otterlo, W. A.; Rauh, D. Structure-based design,
synthesis and biological evaluation of N-pyrazole, N′-thiazole urea
inhibitors of MAP kinase p38alpha. Eur. J. Med. Chem. 2012, 48, 1−15.
(14) Provenzano, P. P.; Inman, D. R.; Eliceiri, K. W.; Knittel, J. G.;
Yan, L.; Rueden, C. T.; White, J. G.; Keely, P. J. Collagen density
promotes mammary tumor initiation and progression. BMC Med.
2008, 6, 11-1−11-15.
(28) Getlik, M.; Simard, J. R.; Termathe, M.; Grutter, C.; Rabiller,
̈
M.; van Otterlo, W. A.; Rauh, D. Fluorophore labeled kinase detects
ligands that bind within the MAPK insert of p38alpha kinase. PLoS
One 2012, 7, e39713-1−e39713-10.
(15) Conklin, M. W.; Eickhoff, J. C.; Riching, K. M.; Pehlke, C. A.;
Eliceiri, K. W.; Provenzano, P. P.; Friedl, A.; Keely, P. J. Aligned
collagen is a prognostic signature for survival in human breast
carcinoma. Am. J. Pathol. 2011, 178, 1221−1232.
(16) Perou, C. M. Molecular stratification of triple-negative breast
cancers. Oncologist 2010, 15, 39−48.
(17) Thiery, J. P.; Acloque, H.; Huang, R. Y.; Nieto, M. A.
Epithelial−mesenchymal transitions in development and disease. Cell
2009, 139, 871−890.
(18) Zhang, K.; Rodriguez-Aznar, E.; Yabuta, N.; Owen, R. J.;
Mingot, J. M.; Nojima, H.; Nieto, M. A.; Longmore, G. D. Lats2 kinase
potentiates Snail1 activity by promoting nuclear retention upon
phosphorylation. EMBO J. 2012, 31, 29−43.
(19) Hammerman, P. S.; Sos, M. L.; Ramos, A. H.; Xu, C.; Dutt, A.;
Zhou, W.; Brace, L. E.; Woods, B. A.; Lin, W.; Zhang, J.; Deng, X.;
(29) Schneider, R.; Gohla, A.; Simard, J. R.; Yadav, D. B.; Fang, Z.;
van Otterlo, W. A.; Rauh, D. Overcoming compound fluorescence in
the FLiK screening assay with red-shifted fluorophores. J. Am. Chem.
Soc. 2013, 135, 8400−8408.
(30) Simard, J. R.; Getlik, M.; Grutter, C.; Schneider, R.; Wulfert, S.;
̈
Rauh, D. Fluorophore labeling of the glycine-rich loop as a method of
identifying inhibitors that bind to active and inactive kinase
conformations. J. Am. Chem. Soc. 2010, 132, 4152−4160.
(31) Rabiller, M.; Getlik, M.; Kluter, S.; Richters, A.; Tuckmantel, S.;
̈
̈
Simard, J. R.; Rauh, D. Proteus in the world of proteins:
conformational changes in protein kinases. Arch. Pharm. (Weinheim,
Ger.) 2010, 343, 193−206.
4261
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262

Journal of Medicinal Chemistry
Article
(32) Copeland, R. A. Conformational adaptation in drug-target
interactions and residence time. Future Med. Chem. 2011, 3, 1491−
1501.
kinases, suppresses tumor growth in vivo. Nat. Med. 2004, 10, 262−
267.
(49) Shah, N. P.; Tran, C.; Lee, F. Y.; Chen, P.; Norris, D.; Sawyers,
C. L. Overriding imatinib resistance with a novel ABL kinase inhibitor.
Science 2004, 305, 399−401.
(50) Kim, H. G.; Tan, L.; Weisberg, E. L.; Liu, F.; Canning, P.; Choi,
H. G.; Ezell, S. A.; Wu, H.; Zhao, Z.; Wang, J.; Mandinova, A.; Griffin,
J. D.; Bullock, A. N.; Liu, Q.; Lee, S. W.; Gray, N. S. Discovery of a
potent and selective DDR1 receptor tyrosine kinase inhibitor. ACS
Chem. Biol. 2013, 8, 2145−2150.
(51) Gao, M.; Duan, L.; Luo, J.; Zhang, L.; Lu, X.; Zhang, Y.; Zhang,
Z.; Tu, Z.; Xu, Y.; Ren, X.; Ding, K. Discovery and optimization of 3-
(2-(pyrazolo[1,5-a]pyrimidin-6-yl)ethynyl)benzamides as novel selec-
tive and orally bioavailable discoidin domain receptor 1 (DDR1)
inhibitors. J. Med. Chem. 2013, 56, 3281−3295.
(52) Weinstein, I. B.; Joe, A. K. Mechanisms of disease: oncogene
addiction−a rationale for molecular targeting in cancer therapy. Nat.
Clin. Pract. Oncol. 2006, 3, 448−457.
(53) Miao, L.; Zhu, S.; Wang, Y.; Li, Y.; Ding, J.; Dai, J.; Cai, H.;
Zhang, D.; Song, Y. Discoidin domain receptor 1 is associated with
poor prognosis of non-small cell lung cancer and promotes cell
invasion via epithelial-to-mesenchymal transition. Med. Oncol. 2013,
30, 626.
(54) Ahn, Y. M.; Clare, M.; Ensinger, C. L.; Hood, M. M.; Lord, J.
W.; Lu, W. P.; Miller, D. F.; Patt, W. C.; Smith, B. D.; Vogeti, L.;
Kaufman, M. D.; Petillo, P. A.; Wise, S. C.; Abendroth, J.; Chun, L.;
Clark, R.; Feese, M.; Kim, H.; Stewart, L.; Flynn, D. L. Switch control
pocket inhibitors of p38-MAP kinase. Durable type II inhibitors that
do not require binding into the canonical ATP hinge region. Bioorg.
Med. Chem. Lett. 2010, 20, 5793−5798.
(33) Copeland, R. A.; Pompliano, D. L.; Meek, T. D. Drug-target
residence time and its implications for lead optimization. Nat. Rev.
Drug Discovery 2006, 5, 730−739.
(34) Fang, Z.; Grutter, C.; Rauh, D. Strategies for the selective
̈
regulation of kinases with allosteric modulators: exploiting exclusive
structural features. ACS Chem. Biol. 2013, 8, 58−70.
(35) Over, B.; Wetzel, S.; Grutter, C.; Nakai, Y.; Renner, S.; Rauh, D.;
̈
Waldmann, H. Natural-product-derived fragments for fragment-based
ligand discovery. Nat. Chem. 2013, 5, 21−28.
(36) Torkamani, A.; Kannan, N.; Taylor, S. S.; Schork, N. J.
Congenital disease SNPs target lineage specific structural elements in
protein kinases. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 9011−9016.
(37) O’Hare, T.; Shakespeare, W. C.; Zhu, X.; Eide, C. A.; Rivera, V.
M.; Wang, F.; Adrian, L. T.; Zhou, T.; Huang, W. S.; Xu, Q.; Metcalf,
C. A., 3rd; Tyner, J. W.; Loriaux, M. M.; Corbin, A. S.; Wardwell, S.;
Ning, Y.; Keats, J. A.; Wang, Y.; Sundaramoorthi, R.; Thomas, M.;
Zhou, D.; Snodgrass, J.; Commodore, L.; Sawyer, T. K.; Dalgarno, D.
C.; Deininger, M. W.; Druker, B. J.; Clackson, T. AP24534, a pan-
BCR-ABL inhibitor for chronic myeloid leukemia, potently inhibits the
T315I mutant and overcomes mutation-based resistance. Cancer Cell
2009, 16, 401−412.
(38) Karaman, M. W.; Herrgard, S.; Treiber, D. K.; Gallant, P.;
Atteridge, C. E.; Campbell, B. T.; Chan, K. W.; Ciceri, P.; Davis, M. I.;
Edeen, P. T.; Faraoni, R.; Floyd, M.; Hunt, J. P.; Lockhart, D. J.;
Milanov, Z. V.; Morrison, M. J.; Pallares, G.; Patel, H. K.; Pritchard, S.;
Wodicka, L. M.; Zarrinkar, P. P. A quantitative analysis of kinase
inhibitor selectivity. Nat. Biotechnol. 2008, 26, 127−132.
(39) Davis, M. I.; Hunt, J. P.; Herrgard, S.; Ciceri, P.; Wodicka, L. M.;
Pallares, G.; Hocker, M.; Treiber, D. K.; Zarrinkar, P. P.
Comprehensive analysis of kinase inhibitor selectivity. Nat. Biotechnol.
2011, 29, 1046−1051.
(40) 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.
Inhibition of p38 MAP kinase by utilizing a novel allosteric binding
site. Nat. Struct. Biol. 2002, 9, 268−272.
(41) Schindler, T.; Bornmann, W.; Pellicena, P.; Miller, W. T.;
Clarkson, B.; Kuriyan, J. Structural mechanism for STI-571 inhibition
of abelson tyrosine kinase. Science 2000, 289, 1938−1942.
(55) Swann, S. L.; Merta, P. J.; Kifle, L.; Groebe, D.; Sarris, K.;
Hajduk, P. J.; Sun, C. Biochemical and biophysical characterization of
unique switch pocket inhibitors of p38alpha. Bioorg. Med. Chem. Lett.
2010, 20, 5787−5792.
(56) Duveau, D. Y.; Hu, X.; Walsh, M. J.; Shukla, S.; Skoumbourdis,
A. P.; Boxer, M. B.; Ambudkar, S. V.; Shen, M.; Thomas, C. J.
Synthesis and biological evaluation of analogues of the kinase inhibitor
nilotinib as Abl and Kit inhibitors. Bioorg. Med. Chem. Lett. 2013, 23,
682−686.
(42) Uitdehaag, J. C.; de Roos, J. A.; van Doornmalen, A. M.;
Prinsen, M. B.; de Man, J.; Tanizawa, Y.; Kawase, Y.; Yoshino, K.;
Buijsman, R. C.; Zaman, G. J. Comparison of the cancer gene targeting
and biochemical selectivities of all targeted kinase inhibitors approved
for clinical use. PLoS One 2014, 9, e92146-1−e92146-13.
(43) Kitagawa, D.; Yokota, K.; Gouda, M.; Narumi, Y.; Ohmoto, H.;
Nishiwaki, E.; Akita, K.; Kirii, Y. Activity-based kinase profiling of
approved tyrosine kinase inhibitors. Genes Cells 2013, 18, 110−122.
(44) Simard, J. R.; Getlik, M.; Grutter, C.; Pawar, V.; Wulfert, S.;
̈
Rabiller, M.; Rauh, D. Development of a fluorescent-tagged kinase
assay system for the detection and characterization of allosteric kinase
inhibitors. J. Am. Chem. Soc. 2009, 131, 13286−13296.
(45) Tamaoki, T.; Nomoto, H.; Takahashi, I.; Kato, Y.; Morimoto,
M.; Tomita, F. Staurosporine, a potent inhibitor of phospholipid/Ca++
dependent protein kinase. Biochem. Biophys. Res. Commun. 1986, 135,
397−402.
(46) Zhang, J. H.; Chung, T. D.; Oldenburg, K. R. A simple statistical
parameter for use in evaluation and validation of high throughput
screening assays. J. Biomol. Screening 1999, 4, 67−73.
(47) Richters, A.; Ketzer, J.; Getlik, M.; Grutter, C.; Schneider, R.;
̈
Heuckmann, J. M.; Heynck, S.; Sos, M. L.; Gupta, A.; Unger, A.;
Schultz-Fademrecht, C.; Thomas, R. K.; Bauer, S.; Rauh, D. Targeting
gain of function and resistance mutations in Abl and KIT by hybrid
compound design. J. Med. Chem. 2013, 56, 5757−5772.
(48) Harrington, E. A.; Bebbington, D.; Moore, J.; Rasmussen, R. K.;
Ajose-Adeogun, A. O.; Nakayama, T.; Graham, J. A.; Demur, C.;
Hercend, T.; Diu-Hercend, A.; Su, M.; Golec, J. M.; Miller, K. M. VX-
680, a potent and selective small-molecule inhibitor of the Aurora
4262
dx.doi.org/10.1021/jm500167q | J. Med. Chem. 2014, 57, 4252−4262