Binding or bending: Distinction of allosteric abl kinase agonists from antagonists by an NMR-based conformational assay

Article abstract of DOI:10.1021/ja101837n

Allosteric inhibitors of Bcr-Abl have emerged as a novel therapeutic option for the treatment of CML. Using fragment-based screening, a search for novel Abl inhibitors that bind to the myristate pocket was carried out. Here we show that not all myristate ligands are functional inhibitors, but that the conformational state of C-terminal helix-I is a structural determinant for functional activity. We present an NMR-based conformational assay to monitor the conformation of this crucial helix-I and show that myristate ligands that bend helix-I are functional antagonists, whereas ligands that bind to the myristate pocket but do not induce this conformational change are kinase agonists. Activation of c-Abl by allosteric agonists has been confirmed in a biochemical assay.

Full text of DOI:10.1021/ja101837n

Published on Web 05/07/2010
Binding or Bending: Distinction of Allosteric Abl Kinase
Agonists from Antagonists by an NMR-Based Conformational
Assay
Wolfgang Jahnke,* Robert M. Grotzfeld, Xavier Pelle´, Andre´ Strauss,
Gabriele Fendrich, Sandra W. Cowan-Jacob, Simona Cotesta, Doriano Fabbro,
Pascal Furet, Ju¨rgen Mestan, and Andreas L. Marzinzik
NoVartis Institutes for Biomedical Research, 4002 Basel, Switzerland
Received March 4, 2010; E-mail: wolfgang.jahnke@novartis.com
Abstract: Allosteric inhibitors of Bcr-Abl have emerged as a novel therapeutic option for the treatment of
CML. Using fragment-based screening, a search for novel Abl inhibitors that bind to the myristate pocket
was carried out. Here we show that not all myristate ligands are functional inhibitors, but that the
conformational state of C-terminal helix_I is a structural determinant for functional activity. We present an
NMR-based conformational assay to monitor the conformation of this crucial helix_I and show that myristate
ligands that bend helix_I are functional antagonists, whereas ligands that bind to the myristate pocket but
do not induce this conformational change are kinase agonists. Activation of c-Abl by allosteric agonists
has been confirmed in a biochemical assay.
Introduction
known for its ability to bind to the hinge region in the ATP
pocket of kinases. In order to fully exploit the potential of the
Under normal circumstances, the enzymatic activity of c-Abl
kinase is tightly regulated. This is in part due to myristoyl-
mediated autoinhibition:1 In c-Abl isoform 1b, a myristoyl group
covalently attached to Gly2 at the N-terminus can bind to a
deep pocket within the kinase domain,2 which causes the SH2
and SH3 domains to dock against the kinase domain, thereby
forming an assembled inactive state (Figure 1a,b).2 The onco-
genic fusion protein, Bcr-Abl, lacks the myristoylation site of
c-Abl 1b. It is constitutively active and transforms hematologic
stem cells to cause chronic myelogenous leukemia (CML).3
Although drugs targeting Bcr-Abl, such as imatinib, nilotinib,
or dasatinib, are efficacious treatments of CML, they all bind
within the ATP pocket of the Abl kinase domain, and advanced
patients can relapse due to the emergence of resistance muta-
tions.4
Myristate mimetics capable of stabilizing Bcr-Abl in the
assembled inactive state could function as allosteric inhibitors
of Bcr-Abl. Since they bind at a different pocket, such
compounds could be active against mutant enzymes resistant
to ATP-site inhibitors. Differential cytotoxicity screening has
led to the discovery of GNF-2, which is a highly selective Abl
inhibitor5 and has been shown to bind to the myristate pocket.6
However, GNF-2 is based on an aminopyrimidine scaffold well-
nonconserved myristoyl binding pocket on Abl, we applied
fragment-based screening to identify alternative scaffolds that
inhibit the kinase. In the course of fragment screening,
characterization, and optimization, we gained insights into the
structural requirements and conformational changes that myristate
ligands have to induce in order to be functional inhibitors. We
developed an NMR-based conformational assay to rapidly
monitor the ability of myristate ligands to bend helix_I, which
is a prerequisite for functional activity. Finally, we reason that
myristate ligands that bind but do not bend helix_I should be
allosteric agonists of c-Abl kinase. We prove this hypothesis
in a biochemical c-Abl assay and discuss the potential utility
of c-Abl (not Bcr-Abl) activators.
Experimental Section
Nomenclature. ABL1 is referred to as c-Abl, BCR-ABL1 is
referred to as Bcr-Abl. Residues are numbered according to the
c-Abl 1b isoform.
Protein Expression and Purification. For NMR: Murine c-Abl
(248-534, Abl 1b numbering) was expressed with 15N-valine and
purified as previously described.7 Murine c-Abl and human c-Abl
differ only in three positions (N355 f S355, Q532 f R532, V534
f T534 for human f mouse), neither of which is expected to affect
the myristate pocket. The cost for the 15N-valine isotopes for the
production of 8 mg of 15N-Val-Abl (sufficient for 20 samples) was
below $100. The buffer used for NMR studies was 20 mM Bis-
Tris (pH 6.5), 150 mM NaCl, 1 mM EDTA, 2 mM TCEP. For
X-ray: Murine c-Abl (248-534, Abl 1b numbering) was expressed
in the presence of imatinib (received from the Novartis compound
archive) and purified as previously described.6 The Abl/imatinib
complex (molar ratio 1:1) was concentrated to 20 mg/mL in 20
(1) Hantschel, O.; Nagar, B.; Guettler, S.; Kretzschmar, J.; Dorey, K.;
Kuriyan, J.; Superti-Furga, G. Cell 2003, 112 (6), 845–857.
(2) Nagar, B.; Hantschel, O.; Young, M. A.; Scheffzek, K.; Veach, D.;
Bornmann, W.; Clarkson, B.; Superti-Furga, G.; Kuriyan, J. Cell 2003,
112 (6), 859–871.
(3) Ren, R. Nat. ReV. Cancer 2005, 5 (3), 172–183.
(4) O’Hare, T.; Eide, C. A.; Deininger, M. W. Blood 2007, 110 (7), 2242–
2249.
(5) 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 (2), 95–102.
(7) Strauss, A.; Bitsch, F.; Cutting, B.; Fendrich, G.; Graff, P.; Liebetanz,
(6) Zhang, J.; et al. Nature 2010, 463 (7280), 501–506.
J.; Zurini, M.; Jahnke, W. J. Biomol. NMR 2003, 26 (4), 367–372.
9
10.1021/ja101837n  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 7043–7048 7043

A R T I C L E S
Jahnke et al.
Figure 1. Crystal structures of c-Abl complexes. (a) In the absence of a ligand for the myristate pocket, helix_I (orange) is straight (PDB code 2G2H20).
(b) In the presence of myristate, helix_I is bent and the SH2 and SH3 domains dock against the kinase domain (PDB code 1OPK2). (c,d) Details of the
myristate pocket in complex with fragments 1 and 2, shown above. The position of myristate is superimposed for illustration only. Similarly, helix_I is not
visible in the structures shown in c and d, but for illustration, its position in 1OPK is superimposed. Note the steric clash that would occur if helix_I adopted
this bent conformation. The color code is orange for helix_I, magenta for myristate, green for helix_C and fragments 1 and 2, red for the P-loop, and pink
for the activation loop. The ATP binding site of c-Abl is filled with PD166326 in a and b, and with imatinib in c and d.
mM Tris-HCl, 100 mM NaCl, 1 mM EDTA, pH 7.6 plus 4 mM
DTT before crystallization.
15N-Val Resonance Assignments of Abl/imatinib Complex.
Almost complete backbone resonance assignment for the Abl/
imatinib complex has previously been achieved,10 and the assign-
ments have been deposited with the BioMagResBank (BMRB
submission code 15488). The human Abl construct for resonance
assignment ended at residue Ser519, while the mouse Abl construct
used for this work extended up to Thr534. Val525 is the only valine
residue that is added in the longer construct, and its chemical shift
was thus easily assigned to 7.95 ppm (1H) and 120.7 ppm (15N).
Crystal Structure Determination. For the complex with frag-
ment 1, crystals were obtained by cocrystallization in the presence
of an excess of the fragment using the method of hanging drop
vapor diffusion and a reservoir solution consisting of 0.1 M MES
pH 6.5, 0.2 M magnesium chloride, and 18.4% PEG4000 at 4 °C.
The crystals used for soaking with fragment 2 were obtained by
hanging drop vapor diffusion using a reservoir consisting of 0.1 M
imidazole pH 6.5, 0.2 M magnesium chloride, 2% ethylene glycol,
13.8% PEG4000 at 4 °C. The latter crystals were soaked overnight
at 4 °C in a 2.5 mM solution of fragment 2 in the crystallization
buffer. In both cases, the crystals were then transferred in the same
buffer containing gradually increasing concentrations of ethylene
glycol and flash cooled in liquid nitrogen when the concentration
reached 20%.
NMR Spectroscopy. All NMR experiments were carried out
on a Bruker AV600 spectrometer at a proton frequency of 600 MHz
using a TCI cryoprobe. For the conformational assay, HSQC spectra
were recorded using 15N-Val-Abl at a concentration of 1.4 mg/mL
(44 µM) and a volume of 200 µL. The temperature was 296 K;
160 increments of 80 scans each were recorded, resulting in a
measuring time of 5 h. Shorter measuring times can be used if the
Abl concentration is appropriately raised.
Determination of Dissociation Constants. Dissociation con-
stants (KD) were measured by NMR in one or both of the following
ways.
(1) Direct titration of Abl/imatinib complex with increasing
1
amounts of ligand by H NMR: A methyl group from a residue
lining the myristate binding pocket has a resonance at -0.16 ppm,
1
which is nonoverlapping in a 1D H spectrum and is sensitive to
the presence of a ligand. The chemical shift of this methyl group
was followed as ligand was added. The titration curve follows a
simple exponential and can be interpreted as percentage of bound
protein as a function of ligand concentration. From this curve, the
dissociation constant, KD, is easily extracted by curve fitting using
the program Origin 7.5. Supporting Information Figure 1 shows
sample spectra and sample titration curves for three different
fragments.
Diffraction data were collected from single crystals at beamline
PXII of the Swiss Light Source (wavelength 0.9999 Å, Supporting
Information Table 1). For fragment 1, a complete data set was
collected using a MARCCD 225 after 161° of rotation with 1°
rotation ranges for each image. For fragment 2, 100° of data was
collected in the same way. All data were processed inside APRV11
using XDS.12 The structure of the complex with fragment 1 was
solved using molecular replacement with the program MOLREP.13
The structure with fragment 2 was solved using rigid body
refinement and difference Fourier methods with Refmac514 and the
protein coordinates of PDB entry 1OPJ2 as a starting model. In
both cases, refinement proceeded with cycles of model building
(2) Competition experiments: Binding affinity of a test compound
was measured by its ability to displace a reporter ligand with known
binding affinity, such as compound 1 (refs 8 and 9). Twenty
compounds for which the dissociation constant had been determined
by direct titration were used to experimentally calibrate the
displacement curve. After calibration, dissociation constants de-
termined by direct titration or by competition experiments generally
agreed within a factor of 3.
For selected compounds, the dissociation constants were also
determined by isothermal titration calorimetry (ITC). The ITC-
derived values were identical within a factor of 2 to the values
derived from direct NMR titration.
(10) Vajpai, N.; Strauss, A.; Fendrich, G.; Cowan-Jacob, S. W.; Manley,
P. W.; Jahnke, W.; Grzesiek, S. Biomol NMR Assign 2008, 2 (1), 41–
42.
(11) Kroemer, M.; Dreyer, M. K.; Wendt, K. U. Acta Crystallogr., Sect. D
2004, 60 (9), 1679–1682.
(12) Kabsch, W. J. Appl. Crystallogr. 1993, 26 (6), 795–800.
(13) Vagin, A.; Teplyakov, A. J. Appl. Crystallogr. 1997, 30, 1022–1025.
(14) Murshudov, G. N.; Vagin, A. A.; Dodson, E. J. Acta Crystallogr.,
Sect. D 1997, 53 (3), 240–255.
(8) Dalvit, C.; Flocco, M.; Knapp, S.; Mostardini, M.; Perego, R.;
Stockman, B. J.; Veronesi, M.; Varasi, M. J. Am. Chem. Soc. 2002,
124 (26), 7702–7709.
(9) Jahnke, W.; Floersheim, P.; Ostermeier, C.; Zhang, X.; Hemmig, R.;
Hurth, K.; Uzunov, D. P. Angew. Chem., Int. Ed. 2002, 41 (18), 3420–
3423.
9
7044 J. AM. CHEM. SOC. VOL. 132, NO. 20, 2010

Distinction of Allosteric Abl Kinase Agonists
A R T I C L E S
Figure 2. NMR-based conformational assay for helix_I. (a) Superposition of details around the myristate pocket from structures with an unliganded pocket
(PDB code 2G2H, linear helix_I colored cyan) and with bound myristate (PDB code 1OPK, bent helix_I colored orange). (b-d) 15N-1H HSQC spectra of
Abl/imatinib with unliganded myristate pocket, or complexed with the ligands shown on the top. The resonance of Val525 is indicated by the arrow.10 Also
given is the dissociation constant, KD, as measured by NMR or ITC (see Supporting Information Figure 1), and the IC50 for inhibition of Bcr-Abl
autophosphorylation in Bcr-Abl-transfected Ba/F3 cell lines. Contour plots of Abl/imatinib with unliganded myristate pocket (black) or complexed with
Myr-peptide (red) or 1 (blue) is shown at the bottom.
using the molecular graphics program COOT15 and maximum
GNF-2 showed activity in these assays with an IC50 of 400 and
likelihood based minimization using Refmac5. Imatinib and the
fragments were fitted to clear difference electron density, and then
new water molecules were added using ARP/wARP16 in the final
cycles.
180 nM, respectively, and also the activity of weaker GNF-2
analogues could be detected up to IC50 values of >20 µM.
Why are biophysically validated Abl ligands not active
inhibitors? This is explained by structural considerations: In
order for the SH2 and SH3 domains to dock against the kinase
domain and form the assembled inactive state, a conformational
change is required for the C-terminal helix_I of the kinase
domain, which has to bend toward the myristate pocket (Figure
1a,b). Crystal structures of Abl in complex with several
fragments (Figure 1c,d), however, showed that binding of these
fragments is incompatible with bending of helix_I due to
potential steric clashes (e.g., with Ile521). Building a functional
inhibitor by fragment-based screening thus requires starting
points that allow or even induce helix_I to bend. Therefore, a
rapid method to monitor the state of helix_I in Abl complexed
with various ligands was required and was developed using
NMR spectroscopy. We call this method a conformational assay
because it detects a specific aspect of a protein conformation
with the speed of a binding assay.
Results
Fragment-Based Screening. Fragment-based screening was
carried out by NMR as a biophysical binding assay, using Abl
kinase complexed with imatinib to block the active site. A small
library containing 500 fragments was screened using T1F
relaxation17 and waterLOGSY18 ligand observation experiments.
Several hits were identified from our fragment library, most of
them being non-kinase inhibitor scaffolds such as 1 or 2. While
most fragments were weak ligands, some of them bound
surprisingly tightly for their low molecular weight (KD ) 6 and
22 µM for 1 and 2, respectively; see Supporting Information
Figure 1), yielding high ligand efficiencies and binding ef-
ficiency indices (0.59 and 28.2 for 1, respectively). These hits
were inactive in an Abl kinase biochemical assay, which was
not surprising since only the kinase domain was used in the
assay. However, all hits were also inactive in cell-based assays
measuring either cell proliferation or autophosphorylation in
Bcr-Abl-transfected Ba/F3 cell lines.19 This was surprising since
NMR-Based Conformational Assay. The NMR conforma-
tional assay requires 15N-Val selectively labeled Abl kinase,
which was economically prepared according to previously
established protocols.7 The conformational assay is based on
different mobilities of Val525 in the two different conforma-
tional states (Figure 2a): with the myristate pocket in its
unliganded state, helix_I is linear and ends with Ile521 (PDB
code 1M5221), so that Val525 is unstructured and highly flexible
and therefore leads to an intense NMR signal. With myristate
(15) Emsley, P.; Cowtan, K. Acta Crystallogr., Sect. D 2004, 60 (12), 2126–
2132.
(16) Perrakis, A.; Sixma, T. K.; Wilson, K. S.; Lamzin, V. S. Acta
Crystallogr., Sect. D 1997, 53 (4), 448–455.
(17) Hajduk, P. J.; Olejniczak, E. T.; Fesik, S. W. J. Am. Chem. Soc. 1997,
119 (50), 12257–12261.
(18) Dalvit, C.; Fogliatto, G.; Stewart, A.; Veronesi, M.; Stockman, B.
J. Biomol. NMR 2001, 21 (4), 349–359.
(19) Weisberg, E.; et al. Cancer Cell 2005, 7 (2), 129–141.
(20) Levinson, N. M.; Kuchment, O.; Shen, K.; Young, M. A.; Koldobskiy,
M.; Karplus, M.; Cole, P. A.; Kuriyan, J. PLoS Biol 2006, 4 (5), e144.
(21) Nagar, B.; Bornmann, W. G.; Pellicena, P.; Schindler, T.; Veach, D. R.;
Miller, W. T.; Clarkson, B.; Kuriyan, J. Cancer Res. 2002, 62 (15),
4236–4243.
9
J. AM. CHEM. SOC. VOL. 132, NO. 20, 2010 7045

A R T I C L E S
Jahnke et al.
Figure 3. Structures, cellular activities, and conformational assay results for GNF-2 and close analogues. For contour plots, see Supporting Information
Figure 2.
bound, Val525 is an integral part of helix_I′, is well-structured
and ordered, and therefore has an NMR intensity equal to the
intensities from the rest of the kinase. In this conformational
assay, a strong Val525 NMR signal thus indicates a flexible
and unbent helix_I, while average NMR signals indicate an
ordered and bent helix_I. Intermediate intensities can result from
various degrees of helix I bending and represent the time-
averaged percentage of helix_I being in a linear or bent
conformation.
understand the structure-activity relationships and the mech-
anism of inhibition. As expected, GNF-2 leads to complete
bending of helix_I, consistent with its functional activity (Figure
3). Adrian et al.5 point out the strict requirement for the
4-trifluoromethoxyaniline headgroup for the activity of GNF-
2. We replaced this group by a dichloroaniline headgroup and
confirmed a 20-fold activity decrease of the resulting compound
3 in cellular assays. However, we were surprised to see that
compound 3 still binds to the myristate pocket of Abl with high
affinity (Supporting Information Figure 2). The NMR confor-
mational assay explained this observation by showing that, in
complex with 3, helix_I is not any more fully bent (Figure 3).
Thus, the replacement of the 4-trifluoromethoxyaniline head-
group of GNF-2 does not lead to a compound that lacks binding
affinity to Abl, but to a compound that cannot induce the
conformational changes required for functional activity. Notably,
there is no steric hindrance for helix_I in the bent conformation
even with the dichloroaniline headgroup. Sterically, a bent
helix_I would be possible but is energetically not favored.
Analysis of the Abl/GNF-2 crystal structure6 predicts steric
hindrance of the bent helix_I with a GNF-2 analogue having
the amide in ortho- rather than meta-position. Indeed, compound
4 does not induce full bending of helix_I (Figure 3), and
accordingly, its cellular activity is at least 20-fold reduced.
Combining both changes (compound 5) leads to a complete loss
of helix bending and to inactivity in all cellular assays (Figure
3).
Figure 2 displays NMR HSQC spectra of 15N-Val-labeled
Abl/imatinib with an unliganded myristate pocket (b), bound
to a myristoylated peptide derived from the Abl N-terminus2
(c), and bound to 1, which has a KD of 6 µM but no inhibitory
activity up to 100 µM (d). The flexibility of Val525 in the linear
conformation of helix_I (b) or in presence of a fragment that
does not allow helix_I bending (d) can clearly be seen in Figure
2. In contrast, the reduction in flexibility upon helix_I bending
induced by complexation with the Myr-peptide is shown in
Figure 2c. Each spectrum is recorded within a few hours, does
not require crystallization, and reflects solution conditions
without crystal packing effects.
This conformational assay distinguishes ligands that merely
bind to Abl kinase from ligands that bind and are functional
inhibitors. The assay thus bridges the gap between binding and
inhibition and is an indispensable flowchart assay for a fragment-
based approach using biophysical screens.
Binding and Bending by GNF-2 Analogues. We applied this
assay also to GNF-2 and close analogues in order to better
9
7046 J. AM. CHEM. SOC. VOL. 132, NO. 20, 2010

Distinction of Allosteric Abl Kinase Agonists
A R T I C L E S
inhibition of the kinase. However, our study used a biophysi-
cal screen and showed that ligands for the myristate pocket
are not necessarily functional inhibitors. Using a conforma-
tional assay, we showed that only those myristate ligands
are functional inhibitors that bend helix_I, thus differentiating
“bending” from mere “binding”. Moreover, we showed that
ligands that “bind” without “bending” interfere with the
autoinhibition of Abl and activate the kinase. To our
knowledge, this is the first report of small molecules
influencing the position of a crucial kinase helix that
determines agonism or antagonism and of a biophysical assay
to detect these conformational changes. Previous NMR-based
assays were either binding assays (possibly giving informa-
tion about the binding site) or activity assays,25,26 but the
concept of a conformational assay is novel. It closes the gap
between binding and inhibition for allosteric kinase ligands
and expands the utility of NMR spectroscopy in drug
discovery.27
Pharmaceutical research in the past has focused on kinase
inhibitors. The therapeutic potential of kinase activators has
received little attention, and only few examples for kinase
activators are known,28-30 some of them indirect and
unintended.31-33 This is likely due to the fact that the
discovery of kinase inhibitors is more straightforward than
the discovery of kinase activators: essentially any ligand that
binds to the active site inhibits kinase activity by blocking
the access of substrate. In contrast, kinase activators must
bind outside of the active site to stabilize the active
conformation or interfere with kinase autoinhibition. Suitable
binding sites for such activators may not always be present
and have to be identified individually for each kinase. Since
in contrast to the ATP binding pocket, such allosteric sites
are nonconserved, classical chemical matter targeted for
kinase active sites may not yield potent hits, and alternative
methods such as fragment-based screening may be needed
to identify suitable starting points. Since tool compounds are
rarely available to optimize biochemical or cellular assays
for the detection of kinase activators, biophysical screens can
help to generate such tool compounds, as in the current study.
Due to the nonconserved nature of allosteric binding pockets,
allosteric kinase modulators have potential to reach a high
degree of selectivity. Given the role of Bcr-Abl in CML and
other diseases, and the therapeutic benefit of Bcr-Abl
inhibitors such as imatinib, nilotinib, and dasatinib, an interest
in activators seems paradoxical at first glance. However,
c-Abl (in contrast to Bcr-Abl) does have beneficial effects,
for example, in DNA damage repair, and c-Abl (not Bcr-
Figure 4. Abl inhibitors and activators. Dose-response of an Abl
inhibitor (GNF-2, pink) and an Abl activator (5, blue) in a biochemical
enzyme assay.
Allosteric c-Abl Agonists. A compound that binds to the
myristate pocket but prevents helix_I from adopting the bent
conformation should not be a functional inhibitor, but we
hypothesized that it should still have functional activity. By
competing with endogenous myristoyl groups, or with myristate
mimetics, it should interfere with the autoinhibition of Abl,
destabilize the assembled inactive state, and thus activate the
kinase. Ligands binding to the Abl myristate pocket and inducing
helix_I bending should be Abl antagonists, whereas ligands that
bind but interfere with helix_I bending should be Abl agonists.
Allosteric agonism by myristate ligands destabilizing the
assembled inactive state resembles the G2A Abl mutant that
lacks the myristoylation site and can therefore not form the
assembled inactive state.1 It also resembles the PP mutant where
the assembled inactive state is destabilized by mutation of two
prolines in the SH2-kinase linker.22
In order to test this hypothesis, a biochemical assay was
developed using a c-Abl construct including the SH3, SH2,
and kinase domains. In such an assay, the activity of myristate
pocket ligands can be measured, and GNF-2 has an IC50 of
10 nM. In order to mimic the myristoylation of (unmyris-
toylated) c-Abl, the assay was run with the myristoylated
c-Abl 1b peptide described in ref 1 at a concentration
corresponding to its IC60. Taking this as the intrinsic activity
of naturally myristoylated c-Abl, all functional inhibitors
decrease this activity in a dose-dependent manner. However,
the Abl agonist 5, a close analogue of GNF-2, actually
increases c-Abl activity in a dose-dependent manner, dem-
onstrating the allosteric agonism of myristate ligands that
interfere with the conformational changes required for
allosteric antagonism (Figure 4). Due to the role of c-Abl in
DNA damage repair,23 c-Abl (not Bcr-Abl) activators may
have a therapeutic benefit after radiation therapy. Moreover,
recent data suggest that c-Abl activators can block TGFꢀ-
responsive mammary tumor growth in mice.24
(25) Shuker, S. B.; Hajduk, P. J.; Meadows, R. P.; Fesik, S. W. Science
1996, 274 (5292), 1531-1534.
(26) Dalvit, C. Prog. Nucl. Magn. Reson. Spectrosc. 2007, 51 (4), 243–
271.
(27) Jahnke, W. J. Biomol. NMR 2007, 39 (2), 87–90.
(28) Li, S.; Covino, N. D.; Stein, E. G.; Till, J. H.; Hubbard, S. R. J. Biol.
Chem. 2003, 278 (28), 26007–26014.
Discussion
(29) Hindie, V.; Stroba, A.; Zhang, H.; Lopez-Garcia, L. A.; Idrissova, L.;
Zeuzem, S.; Hirschberg, D.; Schaeffer, F.; Jorgensen, T. J.; Engel,
M.; Alzari, P. M.; Biondi, R. M. Nat. Chem. Biol. 2009, 5 (10), 758–
764.
Recently, Bcr-Abl inhibitors that bind to the myristate
pocket were identified.5 Since the screen was using a cellular
assay, they are functional inhibitors that lead to potent
(30) Zhang, X.; Gureasko, J.; Shen, K.; Cole, P. A.; Kuriyan, J. Cell 2006,
125 (6), 1137–1149.
(31) Heidorn, S. J.; Milagre, C.; Whittaker, S.; Nourry, A.; Niculescu-
Duvas, I.; Dhomen, N.; Hussain, J.; Reis-Filho, J. S.; Springer, C. J.;
Pritchard, C.; Marais, R. Cell 2010, 140 (2), 209–221.
(32) Cameron, A. J.; Escribano, C.; Saurin, A. T.; Kostelecky, B.; Parker,
P. J. Nat. Struct. Mol. Biol. 2009, 16 (6), 624–630.
(33) Okuzumi, T.; Fiedler, D.; Zhang, C.; Gray, D. C.; Aizenstein, B.;
Hoffman, R.; Shokat, K. M. Nat. Chem. Biol. 2009, 5 (7), 484–493.
(22) Barila, D.; Superti-Furga, G. Nat. Genet. 1998, 18 (3), 280–282.
(23) Raina, D.; Pandey, P.; Ahmad, R.; Bharti, A.; Ren, J.; Kharbanda, S.;
Weichselbaum, R.; Kufe, D. J. Biol. Chem. 2005, 280 (12), 11147–
11151.
(24) Allington, T. M.; Galliher-Beckley, A. J.; Schiemann, W. P. FASEB
J. 2009, 23 (12), 4231–4243.
9
J. AM. CHEM. SOC. VOL. 132, NO. 20, 2010 7047

A R T I C L E S
Jahnke et al.
Acknowledgment. We thank Drs. Markus Warmuth and Natha-
nael Gray for helpful discussions, Anke Blechschmidt for ITC
measurements, Janis Liebetanz for the c-Abl clone, Navratna Vajpai
and Stephan Grzesiek for resonance assignment work, and Chryste`le
Henry, Gabriele Rummel, Robert Kastl, and Dominique Kempf for
expert technical assistance.
Abl) activators could have therapeutic benefit after radiation
therapy.23 Recent studies even suggest a direct protective
effect of c-Abl against TGFꢀ-responsive mammary tumor
growth.24 The present publication is one of very few reports
on kinase activators and the first report of a c-Abl activator,
and it will be interesting to see further applications of kinase
activators.
Supporting Information Available: Titration curves for frag-
ments 1 and 2, contour plots for Figure 3, synthesis of
compounds 3-5, crystallographic data, and complete refs 6 and
19. This material is available free of charge via the Internet at
http://pubs.acs.org. All of the crystal structures with fragments
described in this paper have been deposited in the Protein Data
Bank (PDB codes 3MS9 and 3MSS).
A recent report on synergistic effects of a combination of
allosteric and ATP-directed Bcr-Abl inhibitors suggests new
pharmacological opportunities to overcome resistance against
current Bcr-Abl inhibitors,6 but the first inhibitors are rarely
the ones with the best overall profile. Our basic understanding
of allosteric Abl inhibition and the conformational assay to
detect the required structural changes now paves the way for a
rational design of novel allosteric Abl inhibitors with improved
pharmacological properties.
JA101837N
9
7048 J. AM. CHEM. SOC. VOL. 132, NO. 20, 2010