Discovery of a highly selective FLT3 kinase inhibitor from phenotypic cell viability profiling

Article abstract of DOI:10.1039/c2md20169k

We discovered a novel molecular framework 4 containing a heterobiaryl pyrazolopyridine moiety as a selective FLT3 kinase inhibitor from phenotype-based viability profiling. Compound 4g showed outstanding selectivity in cellular cytotoxicity against MV-4-11 leukemic cells via the induction of apoptosis. The hypothesis-driven deconvolution elucidated that compound 4g selectively blocked the phosphorylation of FLT3 and its downstream effectors, such as ERK and STAT5, only in MV-4-11 cells. The inhibitory effect of 4g on in vitro enzyme function and FLT3 phosphorylation in cells proved that FLT3 kinase is a direct molecular target of 4g. Finally, the kinase activity profiling of 4g verified its excellent selectivity toward FLT3 over 40 representative kinases, including the receptor tyrosine kinase (RTK) family. The Royal Society of Chemistry 2013.

Full text of DOI:10.1039/c2md20169k

www.rsc.org/medchemcomm
Volume 4 | Number 1 | January 2013 | Pages 1–296
Themed issue: NEW TALENT
ISSN 2040-2503
CONCISE ARTICLE
Seung Bum Park et al.
Discovery of a highly selective FLT3
kinase inhibitor from phenotypic cell
viability profiling
2040-2503(2013)4:1;1-S

MedChemComm
CONCISE ARTICLE
Discovery of a highly selective FLT3 kinase inhibitor
from phenotypic cell viability profiling†
Cite this: Med. Chem. Commun., 2013,
4, 228
a
a
ab
*
Sanghee Lee, Ala Jo and Seung Bum Park
We discovered a novel molecular framework 4 containing a heterobiaryl pyrazolopyridine moiety as a
selective FLT3 kinase inhibitor from phenotype-based viability profiling. Compound 4g showed
outstanding selectivity in cellular cytotoxicity against MV-4-11 leukemic cells via the induction of
apoptosis. The hypothesis-driven deconvolution elucidated that compound 4g selectively blocked the
phosphorylation of FLT3 and its downstream effectors, such as ERK and STAT5, only in MV-4-11 cells.
The inhibitory effect of 4g on in vitro enzyme function and FLT3 phosphorylation in cells proved that
FLT3 kinase is a direct molecular target of 4g. Finally, the kinase activity profiling of 4g verified its
excellent selectivity toward FLT3 over 40 representative kinases, including the receptor tyrosine kinase
(RTK) family.
Received 27th June 2012
Accepted 9th September 2012
DOI: 10.1039/c2md20169k
www.rsc.org/medchemcomm
limitation of the target-based approach calls forth more interest
in phenotypic screening, which allows the elimination of
Introduction
The current focus in drug discovery research has been shied
from the multi-targeting promiscuous drug to the highly
selective drug. Target-specic small-molecule modulators can
lead to the acquisition of mechanistic clues to understand the
function of biopolymers in a complex biological system. In fact,
a series of studies reported that specic small-molecule
modulators were extremely important as chemical probes,
which can interrogate the function of target proteins, suggest a
new paradigm for complex biological functions,1 and eventually
expand the therapeutic window of those drug candidates.2 In
addition, their selectivity toward certain drug targets ensure
therapeutic activities with minimal adverse effects for various
diseases. Therefore, the discovery of highly specic small-
molecule modulators remains a challenging issue, but a
worthwhile strategy in the eld of drug discovery and chemical
biology.
Under the name of rational drug discovery, the major drug
companies have used the target-based enzymatic assay as a
primary screening platform. High throughput screening with
conventional drug targets has been a major workhorse in the
drug discovery industry and allowed the launching of various
new drugs in the market, however, some of those drugs were
withdrawn from the market, not because of their insufficient
efficacy, but because of their unexpected side effects.3 This
candidate molecules with undesirable side effects in cells or
model animals at the early stage of drug discovery. Therefore,
phenotype-based screening systems recently received huge
attention as one of the major initial screening platforms for
biomedical researchers in academia as well as in the pharma-
ceutical industry. In addition to the reduction in the late-stage
attrition, the phenotypic screening shows a great contribution
to the discovery of rst-in-class drugs with new molecular
entities, which can surpass the conventional target-based
approach.4 For this reason, a large number of drug discovery
centres in the academic environment are much more focused
on the development and application of phenotype-based
screening.5 However, even though the phenotype-based
screening approach might increase the potential to identify
novel molecular frameworks with unique biological activity and
superb selectivity, these bioactive small molecules discovered
by this strategy might be limited in their usage as potential
therapeutics or as chemical biology research tools for the study
of mysterious biological process until the identication or
deconvolution of molecular targets of small-molecule modula-
tors.6 Therefore, the phenotype-based approach might be a
great strategy for the discovery of novel molecular frameworks
with a desired efficacy toward certain disease areas, when it is
tightly combined with the subsequent identication of the
target proteins, which can elucidate the molecular mechanism
of the efficacy and selectivity.
aDepartment of Chemistry, Seoul National University, Seoul, Korea. E-mail: sbpark@
snu.ac.kr; Fax: +82 2 884 4025; Tel: +82 2 880 9090
bDepartment of Biophysics and Chemical Biology/BioMAX institute, Seoul National
Herein, we report the discovery of a novel molecular frame-
work with highly selective cellular cytotoxicity toward an acute
myeloid leukemia (AML) cell line using phenotype-based viability
proling and the subsequent identication of FMS-like tyrosine
kinase-3 (FLT3) as its target protein via hypothesis-driven
University, Seoul 151-747, Korea
† Electronic supplementary information (ESI) available: Cell viability data,
detailed procedures for synthetic and biological experiments, and spectroscopic
data of all new compounds. See DOI: 10.1039/c2md20169k
228 | Med. Chem. Commun., 2013, 4, 228–232
This journal is ª The Royal Society of Chemistry 2013

Concise Article
MedChemComm
deconvolution. In fact, FLT3 has been considered as a therapeutic
target for the treatment of AML patients and it is activated by
FLT3 ligand (FLT3L), which is known to have a crucial role in the
proliferation and apoptosis of hematopoietic cells.7
Results and discussion
Recently, we introduced the concept of the privileged
substructure-based diversity-oriented synthesis (pDOS) strategy
for the efficient construction of skeletally diverse small-mole-
cule collections that maximize molecular diversity with ensured
biological relevance via the incorporation of privileged
substructures into drug-like polyheterocycles.8 With this unique
resource in hand for the discovery of novel bioactive small
molecules, we pursued the identication of novel small mole-
cules with selective cytotoxic activity toward certain cancer cells:
the pDOS-derived small-molecule collection containing over
3000 members with 60 unique molecular frameworks was
subjected to high-throughput cell viability screening (HTS)
against 6 different cells using WST, a water-soluble tetrazolium
salt assay. The extensive HTS screening lead us to the identi-
cation of the pyrazolopyridine-based molecular framework 4 as
an initial hit compound with a specic cellular toxicity pattern
for MV-4-11 cells (see ESI†). To enhance the potency along with
the understanding of their structure–activity relationship, we
then constructed a focused library of 13 compounds using a
synthetic procedure developed in our group.9 In brief, we
prepared the properly functionalized indole moieties 1, which
were transformed into indole-3-carboxaldehyde analogues 2 via
the Vilsmeier–Haack reaction. The resulting dielectrophilic
3-formyl indoles 2 can undergo indole ring opening and
subsequent cyclization with various aminopyrazoles 3 as dinu-
cleophiles to yield the desired heterobiaryl pyrazolopyridines in
a single step (Scheme 1). The substrate scope of this trans-
formation is quite broad, which allows the formation of various
analogues via the combination of R2-substituted 3-amino-
pyrazoles and R1,R3-substituted 3-formyl indole derivatives with
Scheme 1 The synthetic procedure for heterobiaryl pyrazolo[3,4-b]pyridine
derivatives. Reagents and condition: (i) R1,R3-substituted indole, POCl3, DMF, 0 ꢀC,
0.5–2 h; (ii) R2-substituted 3-aminopyrazole, AlCl3, MeOH, 70 ꢀC, 4 h.
various substituents, including halogens, nitro groups, and
esters, at the different positions on the indole rings. This
synthetic route allows the formation of heterobiaryl pyr-
azolopyridine derivatives 4a–4m in moderate to excellent yields
(29–74%). The detailed synthetic procedure and reaction yields
are reported in the ESI.†
As shown in Table 1, a series of heterobiaryl pyrazolopyridine
derivatives 4a–4m was subjected to the cell viability proling
assay against 6 representative cell lines, including two normal
cell lines (NIH3T3 and L6) and four cancer cell lines from
different tissue origins, such as MCF7 and MDA-MB-231 breast
cancer cell lines, MV-4-11 acute myeloid leukemia, and U87-MG
brain glioblastoma cell lines. First of all, we compared the cell
viability pattern upon substituent changes at the R1 position
while keeping 4-methoxyphenyl moiety as the xed R2 substit-
uent (entries 1–8, Table 1). Even though we tested a broad range
of positioning and substitution effects at the R1 position, only a
chlorine substituent at C-6 position 4g showed selective cellular
cytotoxicity toward MV-4-11 AML cells. To determine whether
the 4-methoxyphenyl moiety at the R2 position has an essential
role in its activity, we changed the R2 substituent to either
phenyl 4i or methyl 4j with a 6-chloro substituent at the R1
Table 1 Cell viability profiling results for various kinds of cell linesa
a
Cell viability was measured by WST assay. The series of compounds were treated for 24 h with 1 mM as nal concentration. Each compound was
tested in triplicate. Results are mean value and designated by % which is normalized by DMSO as a control. Data is illustrated by color coding from
dark color for low values to white for high values.
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 228–232 | 229

MedChemComm
Concise Article
Fig. 3 Compound 4g inhibits the phosphorylation of key signalling factors in the
apoptotic process in a dose-dependent manner. Western blot analysis of phos-
phorylated and total ERK and STAT5 upon treatment of 4g at the different
concentrations. Immunoblotting was performed with (a) MV-4-11 cells and (b)
HL60 cells.
Fig. 1 Characterization of the cell-specific cytotoxic agent, 4g. (a) The chemical
structure of compound 4g. (b) Dose-dependent inhibitory activity on cell prolif-
eration and its IC50 toward MV-4-11 AML cells. (c) Cell viability profiling data
against 12 different cell lines with the treatment of 4g at the final concentration
of 1 mM. Individual cell viability data was normalized to DMSO control as 100%.
Each experiment was performed in triplicate; data was figured as mean and SD.
Surprisingly, 4g showed excellent selectivity only toward MV-4-
11 AML cells in cellular cytotoxicity (see Fig. 1c).
With a highly selective cytotoxic agent 4g in hand, we
pursued the in vitro conrmation of 4g for its selective cytotoxic
effect toward MV-4-11 AML cells to explore its mechanism of
action. As shown in Fig. 2, our selective cytotoxic agent 4g
induced apoptosis, which was conrmed by western blot anal-
ysis of cleaved PARP and caspase-3 in MV-4-11 AML cells and
the protein level of cleaved PARP and caspase-3 was also
increased in a dose-dependent manner. These results indicated
that our selective cytotoxic agent 4g inhibited cell proliferation
via the induction of apoptotic cell death in MV-4-11 AML cells.
We then moved forward to study the related protein signalling
pathways in the apoptotic process to access important clues for
the identication of its plausible target proteins. On the basis of
the outstanding selectivity of 4g toward MV-4-11 AML cells over
Fig. 2 Compound 4g induced apoptosis in MV-4-11 cells. Western blot analysis
for cleaved PARP and caspase-3.
position, but we did not observe any cellular cytotoxic effects for other cancer cells, we searched for different phosphorylation
these compounds under the identical condition used for patterns among key players in the apoptotic process upon
compound 4g (entries 9 and 10, Table 1). The further N-modi- treatment of 4g. Aer extensive western blotting analyses on
cation of aniline with methyl 4k, benzyl 4l, and benzyl urea 4m those key factors, we observed the dose-dependent reduction in
did abolish the cytotoxic effects toward all tested cell lines the phosphorylation of ERK and STAT5 on MV-4-11 cells (see
(entries 11–13, Table 1). On the basis of our SAR study, we Fig. 3a). In fact, AML is a type of hematopoietic malignancy,
observed a quite tight SAR pattern of 4g toward selective cellular characterized by the rapid growth of abnormal white blood
cytotoxicity and achieved no improvement of potency and cells, resulting in the interruption of the production of normal
selectivity from our initial hit compound 4g, even though we blood cells. On the basis of a literature survey, we investigated
introduced diverse structural modications on the heterobiaryl what factors might induce the specic cytotoxicity of 4g,
pyrazolopyridine skeleton 4.
observed in MV-4-11 AML cells, and hypothesized that the FLT3
Before probing into the mechanistic study of 4g to under- signalling pathway might be a suitable target because MV-4-11
stand the mode of action, we performed an in vitro evaluation to cells are known to have a high expression level of FLT3 with
conrm the selective cytotoxicity pattern induced by 4g. At rst, internal tandem duplication (ITD) mutation, which is
we evaluated the inhibitory effect of 4g on the cell proliferation commonly found in AML patients and results in constitutive
in a dose-dependent manner. As shown in Fig. 1b, 4g inhibited activation of FLT3.7 FLT3 is one of the class III receptor tyrosine
the cell growth in MV-4-11 cells and resulted in an IC50 value of kinases (RTK) that is activated by FLT3L and has important
592 nM. We also examined the cell viability proling upon roles in the proliferation and apoptosis of hematopoietic cells.
treatment with 4g at 1 mM in a panel of twelve different cell Even though the expression level of FLT3L varies in a wide range
lines, including the original 6 cell lines along with 5 additional of tissues, the expression pattern of FLT3 is quite limited in
cancer cells from different tissue origins (A549 lung cancer, early hematopoietic progenitor cells, which suggested the
HCT116 colon cancer, SW480 colon cancer, DU145 prostate tissue-specic activation of FLT3.10 Therefore, FLT3 has been
cancer, PC3 prostate cancer) and Raw264.7 macrophage cells. recognized as a therapeutic target for the treatment of AML
230 | Med. Chem. Commun., 2013, 4, 228–232
This journal is ª The Royal Society of Chemistry 2013

Concise Article
MedChemComm
Table 2 Kinase activity profiling of 4g against a panel of representative protein
patients. In addition, ERK and STAT5 are the representative
downstream proteins of FLT3 as well as other RTK families.10
We were particularly intrigued by the observation that there
was no signicant difference in the phosphorylation level of
ERK and STAT5 in HL60 AML cells, which is known to have a
quite low or no expression of FLT3,11 upon treatment of 4g
under identical conditions with MV-4-11 AML cells (see Fig. 3b).
In addition, there was no meaningful cellular toxicity in HL60
cells, unlike the high cytotoxicity in MV-4-11 cells (see Fig. S2 in
ESI†). This observation, supported by our hypothesis, led us to
examine whether our selective cytotoxic agent 4g can directly
modulate FLT3 activity by inhibiting its phosphorylation. As
shown in Fig. 4a, 4g inhibited the phosphorylation of FLT3,
conrmed by the immunoblotting of phosphorylated and total
FLT3 in MV-4-11 AML cells, in a dose-dependent manner. We
further validated the inhibition of 4g on FLT3 in vitro activity by
the treatment of 4g using a Kinase Hot Spot assay. Consistent
with our western blotting data, 4g directly inhibited FLT3
activity with an IC50 of 1.3 mM. These results suggested that 4g
blocked the phosphorylation of FLT3 and its kinase activity,
which led to the subsequent blockage of the downstream sig-
nalling pathway, and inhibited the proliferation of MV-4-11
AML cells by triggering apoptosis. Therefore, we concluded that
FLT3 is the direct molecular target of our selective cytotoxic
agent 4g.
kinases, including the class III receptor tyrosine kinase familya
Activity
(%)
Activity
(%)
Kinase
Kinase
Class III receptor tyrosine kinase family
FMS
FLT3
c-KIT
101.5
9.5
99.2
PDGFRa
PDGFRb
93.8
96.3
Representative protein kinases
ABL1
Akt1
ALK
ASK1/MAP3K5
ARAF
Aurora A
CDK1/cyclin A
CDK5/p35
c-MET
c-Src
EGFR
88.9
93.6
96.5
97.8
77.0
87.7
97.4
87.2
118.2
78.7
91.2
95.4
94.0
96.8
90.1
95.0
102.7
101.4
Lck
LYN
LKB1
MEK1
P38a/MAPK14
P38b/MAPK11
p70S6K
PAK1
PKA
PKCa
PLK1
Raf1
RET
STK2
TYK/LTK
VEGFR1/FLT1
VEGFR2/KDR
VEGFR3/FLT4
95.2
86.6
101.1
102.4
110.3
97.8
94.1
98.9
93.0
84.5
107.5
94.9
83.3
85.8
92.9
90.0
98.6
101.2
Erk1
Erk2/MAPK1
FAK/PTK2
FGFR1
IR
JAK1
JNK1
a
Each experiment was performed in duplicate at 10 mM of 4g. The result
Selective inhibitory activity is an essential element for the
development of kinase inhibitors as therapeutic agents or
research tools.2 Therefore, aer conrming the inhibitory
activity of 4g against FLT3 along with its selective cytotoxicity to
MV-4-11 AML cells, we tested the selectivity in its inhibitory
activity against a panel of representative kinases, including the
class III receptor tyrosine kinase (RTK) family, which has a high
structural similarity with FLT3.12 As shown in Table 2, 4g
showed excellent selectivity toward FLT3 over 40 different
was normalized by a DMSO control.
representative kinases. It is worth mentioning that 4g showed
no inhibitory activities toward other class III RTKs, such as FMS,
c-KIT, PDGFRa and PDGFRb, at 10 mM, which is 10-fold higher
than the IC50 value of 4g for FLT3. Therefore, we are condent
that the selective cellular cytotoxicity of 4g to MV-4-11 AML cells
is based on its selective inhibition of the kinase activity of FLT3
over other kinases.
Conclusions
In summary, we identied a novel cell-specic cytotoxic agent
containing the heterobiaryl pyrazolopyridine core skeleton 4
from phenotype-based cell viability proling with over 3000
compounds, synthesized in-house using a pDOS strategy,
against various cell lines. Our hit compound 4g induces
apoptosis with an excellent selectivity in cellular cytotoxicity
with an IC50 of 0.6 mM toward MV-4-11 AML cells over 12
different cell lines. To elucidate the mode of action of 4g, we
pursued the hypothesis-driven deconvolution of protein targets
and identied FLT3 kinase as a plausible target due to its
specic expression in hematopoietic cells. Systematic western
blot analysis revealed that the phosphorylation of ERK and
STAT5, downstream signalling proteins of the FLT3 cascade,
was blocked upon treatment of 4g only in FLT3-positive AML
cells, but not in FLT3-negative HL60 AML cells. The western
blotting and in vitro enzyme assay conrmed that compound 4g
directly inhibits the phosphorylation of FLT3 in a dose-depen-
dent manner with an IC50 of 1.3 mM on FLT3 kinase activity. The
key feature of 4g is an excellent selectivity in the inhibition of
Fig. 4 Compound 4g directly inhibited FLT3 activity. (a) Immunoprecipitation
and western blot analysis for phosphorylated and total FLT3 in MV-4-11 cells. (b) A
dose-dependent inhibitory effect of 4g on the in vitro enzymatic assay of FLT3.
This journal is ª The Royal Society of Chemistry 2013
Med. Chem. Commun., 2013, 4, 228–232 | 231

MedChemComm
Concise Article
FLT3 activity over 40 representative kinases, including class III
receptor tyrosine kinase (RTK) family. Moreover, these results
clearly proved the value of a cell-based phenotype assay as
initial platform for the discovery of highly selective small
molecules when it was combined with hypothesis-driven
deconvolution or target identication. In fact, a continuous
community effort has been devoted to the discovery of FLT3
inhibitors due to the importance of FLT3 as a potential drug
target for the treatment of AML patients.13 Among the reported
FLT3 inhibitors, AC220, developed by Bhagwat and co-work-
ers,13a is one of the advanced FLT3 inhibitors because of its high
selectivity over a panel of kinase and is currently in phase II
clinical trials. However, the build-up resistance in AML relapsed
588; (c) K. R. Brandvold, M. E. Steffey, C. C. Fox and
M. B. Soellner, ACS Chem. Biol., 2012, 7, 1393–1398.
3 F. Sams-Dodd, Drug Discovery Today, 2005, 10, 139–147.
4 D. C. Swinney and J. Anthony, Nat. Rev. Drug Discovery, 2011,
10, 507–519.
5 S. Frye, M. Crosby, T. Edwards and R. Juliano, Nat. Rev. Drug
Discovery, 2011, 10, 409–410.
6 G. C. Terstappen, C. Schlupen, R. Raggiaschi and
G. Gaviraghi, Nat. Rev. Drug Discovery, 2007, 6, 891–903.
7 (a) C. L. Sawvers, Cancer Cell, 2002, 1, 413–415; (b) T. Kindler,
D. B. Lipka and T. Fischer, Blood, 2010, 116, 5089.
8 S. Oh and S. B. Park, Chem. Commun., 2011, 47, 12754–12761.
9 S. Lee and S. B. Park, Org. Lett., 2009, 11, 5214–5217.
patients toward AC220 was also reported in a recent paper.14 10 D. L. Stirewalt and J. P. Radich, Nat. Rev. Cancer, 2003, 3,
Therefore, it might be quite important to identify the new 650–665.
molecular framework 4g as a FLT3 inhibitor with high selec- 11 K. Kojima, M. Konopleva, T. Tsao, M. Andreeff, H. Ishida,
tivity to address this resistance issue, which suggested bene-
cial information about selective anti-cancer therapy for AML.
Y. Shiotsu, L. Jin, Y. Tabe and H. Nakakuma, Leukemia,
2010, 24, 33–43.
12 P. Blume-Jensen and T. Hunter, Nature, 2001, 411, 355–365.
13 (a) Q. Chao, K. G. Sprankle, R. M. Grotzfeld, A. G. Lai,
T. A. Carter, A. M. Velasco, R. N. Gunawardane,
M. D. Cramer, M. F. Gardner, J. James, P. P. Zarrinkar,
H. K. Patel and S. S. Bhagwat, J. Med. Chem., 2009, 52,
7808–7816; (b) B. D. Smith, M. Levis, M. Beran, F. Giles,
H. Kantarjian, K. Berg, K. M. Murphy, T. Dauses,
J. Allebach and D. Small, Blood, 2004, 103, 3669–3676; (c)
A.-M. O. ’Farrell, T. J. Abrams, H. A. Yuen, T. J. Ngai,
S. G. Louie, K. W. H. Yee, L. M. Wong, W. Hong, L. B. Lee,
A. Town, B. D. Smolich, W. C. Manning, L. J. Murray,
M. C. Heinrich and J. M. Cherrington, Blood, 2003, 101,
3597–3605; (d) W.-W. Li, X.-Y. Wang, R.-L. Zheng,
H.-X. Yan, Z.-X. Cao, L. Zhong, Z.-R. Wang, P. Ji,
L.-L. Yang, L.-J. Wang, Y. Xu, J.-J. Liu, J. Yang, C.-H. Zhang,
S. Ma, S. Feng, Q.-Z. Sun, Y.-Q. Wei and S.-Y. Yang, J. Med.
Chem., 2012, 55, 3852–3866.
Acknowledgements
This study was supported by (1) a Chemical Genomics Research
grant (NRF-2011-0000058); (2) a Global Frontier Project grant
(NRF-M1AXA002-0032150); (3) the World Class University
program (R31-2010-000-10032-0) of the National Research Foun-
dation, funded by the Korean Ministry of Education, Science, and
Technology. S.L and A.J are grateful for fellowships awarded by
the BK21 Program. S.L received a Seoul Science Fellowship.
References
1 E. Gavathiotis, D. E. Reyna, J. A. Bellairs, E. S. Leshchiner and
L. D. Walensky, Nat. Chem. Biol., 2012, 8, 639–645.
2 (a) H. Sakamoto, T. Tsukaguchi, S. Hiroshima, T. Kodama,
T. Kobayashi, T. A. Fukami, N. Oikawa, T. Tsukuda,
N. Ishii and Y. Aoki, Cancer Cell, 2011, 19, 679–690; (b) 14 C. C. Smith, Q. Wing, C.-S. Chin, S. Salemo, L. E. Damon,
J. W. Chang, M. J. Niphakis, K. M. Lum, A. B. Cognetta,
C. Wang, M. L. Matthews, S. Niessen, M. W. Buczynski,
L. H. Parsons and B. F. Cravatt, Chem. Biol., 2012, 19, 579–
M. J. Levis, A. E. Perl, K. J. Travers, S. Wang, J. P. Hunt,
P. P. Zarrinkar, E. E. Schadt, A. Kasarskis, J. Kuriyan and
N. P. Shah, Nature, 2012, 485, 260–263.
232 | Med. Chem. Commun., 2013, 4, 228–232
This journal is ª The Royal Society of Chemistry 2013