Facile identification of dual FLT3-Aurora A inhibitors: A computer-guided drug design approach

Article abstract of DOI:10.1002/cmdc.201300571

Computer-guided drug design is a powerful tool for drug discovery. Herein we disclose the use of this approach for the discovery of dual FMS-like receptor tyrosine kinase-3 (FLT3)-Aurora A inhibitors against cancer. An Aurora hit compound was selected as a starting point, from which 288 virtual molecules were screened. Subsequently, some of these were synthesized and evaluated for their capacity to inhibit FLT3 and Aurora kinase A. To further enhance FLT3 inhibition, structure-activity relationship studies of the lead compound were conducted through a simplification strategy and bioisosteric replacement, followed by the use of computer-guided drug design to prioritize molecules bearing a variety of different terminal groups in terms of favorable binding energy. Selected compounds were then synthesized, and their bioactivity was evaluated. Of these, one novel inhibitor was found to exhibit excellent inhibition of FLT3 and Aurora kinase A and exert a dramatic antiproliferative effect on MOLM-13 and MV4-11 cells, with an IC₅₀ value of 7 nM. Accordingly, it is considered a highly promising candidate for further development. In silico selection: We selected an Aurora hit compound as a starting point, followed by two consecutive computer-guided strategies to rapidly and efficiently modify the side chain and core. These efforts resulted in the identification of a potential FLT3-Aurora A inhibitor for further development to treat acute myeloid leukemia (AML).

Full text of DOI:10.1002/cmdc.201300571

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201300571
Facile Identification of Dual FLT3–Aurora A Inhibitors: A
Computer-Guided Drug Design Approach
Yung Chang Hsu, Yi-Yu Ke, Hui-Yi Shiao,* Chieh-Chien Lee, Wen-Hsing Lin, Chun-Hwa Chen,
Kuei-Jung Yen, John T.-A. Hsu, Chungming Chang, and Hsing-Pang Hsieh*[a]
Computer-guided drug design is a powerful tool for drug dis-
covery. Herein we disclose the use of this approach for the dis-
covery of dual FMS-like receptor tyrosine kinase-3 (FLT3)–Au-
rora A inhibitors against cancer. An Aurora hit compound was
selected as a starting point, from which 288 virtual molecules
were screened. Subsequently, some of these were synthesized
and evaluated for their capacity to inhibit FLT3 and Aurora
kinase A. To further enhance FLT3 inhibition, structure–activity
relationship studies of the lead compound were conducted
through a simplification strategy and bioisosteric replacement,
followed by the use of computer-guided drug design to priori-
tize molecules bearing a variety of different terminal groups in
terms of favorable binding energy. Selected compounds were
then synthesized, and their bioactivity was evaluated. Of these,
one novel inhibitor was found to exhibit excellent inhibition of
FLT3 and Aurora kinase A and exert a dramatic antiproliferative
effect on MOLM-13 and MV4-11 cells, with an IC50 value of
7 nm. Accordingly, it is considered a highly promising candi-
date for further development.
Introduction
Protein kinases are a class of very important catalysts which
transfer phosphate groups from ATP to specific substrates,
thereby regulating cellular differentiation, growth, and migra-
tion.[1] Abnormal protein kinase activity is associated with cellu-
lar abnormalities such as the uncontrolled and rapid growth
exhibited by cancer cells, and therefore, specific kinases are at-
tractive targets for new chemotherapies.[1] In 2001, imatinib
(Novartis) became the first small molecule kinase inhibitor to
be approved by the US Food and Drug Administration for the
treatment of chronic myeloid leukemia;[2] to date, twenty-five
kinase inhibitors have been approved for the treatment of vari-
ous cancers.[3,4]
cal studies, although it was proven as a critical factor for caus-
ing apoptosis in leukemia cells;[9–11] therefore, development of
a highly efficient and low toxicity therapy has become an
utmost priority.
Aurora kinases A, B, and C are members of the serine/threo-
nine kinase family, which plays a critical role in cell mitosis.[12]
In recent years, Aurora kinase has emerged as a potential mo-
lecular target in studies of solid tumors[12,13] and leukemia.[14]
Notably, several studies demonstrated that dual FLT3–Aurora
inhibition not only benefits patients with mutated FLT3 AML
but is also effective against tumors that have acquired resist-
ance to FLT3-selective inhibitors, due to the additional inhibi-
tion of Aurora kinase.[14,15] Therefore, dual FLT3–Aurora inhibi-
tors may constitute an exciting new generation of treatments
for AML.
Human protein kinases are classified into seven major
groups by their sequence similarity and biochemical function.[5]
Amongst the more than 500 kinases that have been identified,
FMS-like receptor tyrosine kinase-3 (FLT3), a member of the
class III tyrosine kinase receptor family, is expressed by imma-
ture hematopoietic cells and is strongly associated with acute
myeloid leukemia (AML).[6] Overexpression of wild-type FLT3
often occurs in AML patients, and two major classes of activat-
ing FLT3 mutations (FLT3/ITD and FLT3/TKD) have been report-
ed in approximately 30% of AML patients.[7,8] However, FLT3 in-
hibition as an AML treatment showed limited response in clini-
The side chains of kinase inhibitors (though not the core)
have been shown to play a critical role in kinase selectivity. For
example, tandutinib (1),[16] AZD-1152HQPA (2),[17] gefitinib (3),[18]
and lapatinib (4)[19] share the same quinazoline scaffold, which
can form key interactions with the hinge region of kinases, and
yet they target different kinases with varying degrees of selec-
tivity (Figure 1). Similar results were observed by us: our re-
cently discovered kinase inhibitors 5[20] and 6[21] share
a common furanopyrimidine core, which forms hydrogen
bonds with the hinge region of a protein kinase. However, 5,
bearing a urea side chain, was identified as an Aurora kinase A
inhibitor, whereas 6, with a chiral side chain, was found to be
a selective EGFR inhibitor. This success encouraged the devel-
opment of dual FLT3–Aurora A inhibitors, which focused on ex-
ploring the side chain moieties of 5,6-fused pyrimidine cores.
Herein, the aim of this work was first to identify a potential
dual FLT3–Aurora A inhibitor by using computer-guided drug
[a] Dr. Y. Chang Hsu, Dr. Y.-Y. Ke, Dr. H.-Y. Shiao, Dr. C.-C. Lee, Dr. W.-H. Lin,
C.-H. Chen, K.-J. Yen, Dr. J. T.-A. Hsu, Dr. C. Chang, Dr. H.-P. Hsieh
Institute of Biotechnology & Pharmaceutical Research
National Health Research Institutes
35 Keyan Road, Zhunan, Miaoli County 350, Taiwan (ROC)
E-mail: hyshiao@nhri.org.tw
hphsieh@nhri.org.tw
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201300571.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
1
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Figure 1. The quinazoline scaffold (blue) and the furanopyrimidine scaffold (red) form key hydrogen bonds to the kinase hinge region. By attaching different
side chains, inhibitors can target distinct kinases with varying degrees of selectivity.
design to screen potential side chains using our newly built
FLT3 model,[22] based on initial hit 5, which was previously
identified by high-throughput parallel synthesis technology
and structure-based drug design.[20] Furthermore, dual inhibitor
27 was further improved in terms of both Aurora and
FLT3 potency to obtain novel dual inhibitor 39 by bi-
oisosterism and ring variations.
of PHA739358 (moderate FLT3 inhibition, IC50 value of
669 nm)[25] as the minimal threshold (Table S1 and S2, Support-
ing Information). To gain structural diversity and minimize syn-
thetic effort, we classified 154 molecules with higher score into
Results and Discussion
Identification of compound 27 from initial
screening
As described in Scheme 1, using initial hit 5 as a start-
ing point, a set of virtual molecules was designed
and filtered in silico. A small selection of these was
then synthesized, and their bioactivity was evaluated.
Advantages of this strategy over the traditional syn-
thetic medicinal chemistry approach include de-
creased effort and time in synthesis, as well as im-
proved cost-effectiveness.
At first, the diphenylfuranopyrimidine skeleton of
Aurora inhibitor 5 was adopted as the initial core,
which was identified as crucial from our recent dis-
covery.[20,21,23,24] We then collected 288 fragments
(either commercially available or obtained from our
previous kinase projects) as sources of side chain
moieties. All of the identified fragments possessed
common structural features, including a primary or
secondary amino group, which could further act as
the anchor to connect to the core. A virtual library of
potential inhibitors (the diphenylfuranopyrimidine
skeleton combined with each of the 288 fragments)
was created in silico. Next, the FLT3 homology model
established within our group[22] was introduced to
Scheme 1. Flowchart depicting the computer-guided strategy used to identify the dual
the docking study with the docking score (103.154) FLT3–Aurora A inhibitor as our initial lead.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
2
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
steric replacement of the furan
with
a pyrrole, yielded com-
pounds 33–36 (33, 6-phenyl fur-
anopyrimidine; 34, 5-phenyl fur-
anopyrimidine; 35, 6-phenyl pyr-
rolopyrimidine; 36, 5-phenyl pyr-
rolopyrimidine).
Subsequent
FLT3/Aurora kinase A inhibitory
assays were performed, with re-
sults shown in Table 2. Com-
pounds 33–36 possessed similar
Aurora A enzyme inhibition to
that of lead compound 27, sug-
gesting that the position of the
phenyl group was not important
for Aurora kinase A activity. How-
ever, the presence of the phenyl
group at the 6-position (but not
at position 5) was found to be
critical for FLT3 inhibition (IC50
value of 345 nm of 33, compared
with 780 nm for 34; a similar
trend was shown for 35 and 36).
More interestingly, compounds
35 and 36 showed a three- to
sixfold improvement of FLT3 in-
hibition compared with 33 and
34, indicating that the pyrrole
moiety was more appropriate
Scheme 2. Chemical structures of compounds 7–32.
than furan within the hinge
region of FLT3. Docking studies
of 33 and 35 in the FLT3 homol-
26 groups by the DS2.1/Cluster Ligands module. Thus, 26 mol-
ecules representing each group were selected for synthesis
and subjected to in vitro enzyme-based testing of FLT3 and
Aurora kinase A.
ogy model showed common hydrogen bonding between pyri-
midine N1 and the kinase hinge region in a superpositioned
conformation, while additional hydrogen bonding between
the NH of the pyrrole ring and the carbonyl group of the
Cys694 backbone resulted in enhanced inhibition (35;
Figure 2). In addition, compound 37 (with no substituents on
the terminal phenyl moiety) was synthesized to verify the role
of chlorine. The IC50 data for 37 showed that FLT3 inhibition
was not significantly improved over that of 35; however,
Aurora kinase A inhibition increased threefold, suggesting that
removal of the chlorine was essential. To further verify the pre-
dictive ability of our computer simulations, binding energy cal-
culations were also carried out and showed direct correlation
with the FLT3 IC50 values, as shown in Table 2. Therefore, bind-
ing energy calculations using our homology model were vali-
dated as a reliable strategy and used to guide our predictions
of FLT3 activity of molecules in the next step.
Chemical structures of all 26 synthesized molecules depicted
in Scheme 2 and their inhibitory activities at 1.0 mm concentra-
tion against FLT3 and Aurora kinase A are listed in Table 1. Of
these, more than half were found to inhibit Aurora kinase A by
>50%. However, only compound 27 (cluster 21) imparted
>30% inhibition of both kinases (59.0% inhibition of FLT3;
115.9% inhibition of Aurora kinase A) with IC50 values of 322
and 43 nm against FLT3 and Aurora kinase A, respectively
(Table 2). Accordingly, 27 was selected for further optimization.
With 27 in hand, attempts were made to investigate struc-
ture–activity relationships. In our previous report,[27] the impor-
tance of phenyl rings (attached to the furanopyrimidine ring)
for Aurora A inhibition was studied through a simplification
strategy, and we found that the presence of a phenyl ring at
the 6-position led to retention of activity. We thus introduced
the same strategy in hopes of maintaining FLT3 potency while
decreasing the molecular weight of 27 (Mr =567 Da) to achieve
a decrease in the clogP value. Furthermore, the oxygen atom
of the furan does not appear to contribute a significant inter-
action within the hinge region, based on the docking study of
27. Thus, removal of the furan phenyl ring, followed by bioiso-
Sequential screening for dual FLT3–Aurora A inhibitors
We next attempted to replace the phenyl group of the termi-
nal side chain of 37 with a variety of heterocyclic groups using
the above-mentioned computer-guided drug design. A set of
21 molecules were designed, based on bioisosteric utility and
synthetic convenience, and their binding energies were calcu-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
3
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Table 1. Inhibition of FLT3 and Aurora kinase A (AKA) by compounds 7–
32.
Cluster
Representative
Compd
Inhibition [%][a]
FLT3
AKA
1
2
7
8
6.8
1.9
28.3
3.6
3
4
5
9
10
11
12.5
4.8
8.1
91.2
44.9
9.1
6
7
8
9
12
13
14[26]
15
7.1
2.8
6.5
3.8
43.3
70.8
96.3
7.0
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
16
17
18
19
20[26]
21
11.3
3.1
4.6
17.6
12.3
0.3
13.0
13.8
18.3
3.0
10.6
59.0
3.7
0.6
6.5
84.0
97.3
85.8
72.9
81.1
4.8
35.7
28.3
16.7
5.5
Figure 2. Docking studies of 33 (green) and 35 (blue) in the hinge region of
the active site of the FLT3 homology model. Hydrogen bonding interactions
between the protein and the ligand are shown as dotted lines. A common
hydrogen bond was formed (shown on the figure) between the pyrimidine
N1 and the kinase hinge region in a superposition of the binding conforma-
tion of two inhibitors. The NH in the pyrrole ring formed an additional hy-
drogen bond with the carbonyl group of the Cys694 backbone.
22
23[27]
24
25
26
27
28[20]
29[26]
30
80.1
115.9
110.8
95.5
53.7
6.4
31
32
20.4
4.1
70.1
[a] Inhibition was determined at a compound concentration of 1.0 mm.
Data are representative values of single experiment determination (dupli-
cate).
Table 2. Calculated binding energies (Ebind) for FLT3 and inhibitory activi-
ties of compounds 27 and 33–37 against FLT3 and Aurora kinase A
(AKA).
Figure 3. Docking studies of 37 (black) and 38 (green) in the active site of
the FLT3 homology model. Hydrogen bonding interactions between the
protein and the ligand are shown as yellow dotted lines. The terminal heter-
ocyclic urea group of inhibitor 38 binds to FLT3 kinase, using four hydrogen
bonds to Lys644 and Asp829, resulting in a different binding conformation
of 38 compared with that of 37 and a lower binding energy. The extended
functional group at the C5 position of the isoxazole moiety points toward
the unoccupied hydrophobic region (red arrow), and may provide the acces-
sory CH–p interaction with the phenyl group of the Phe621 residue.
Compd
R1
R2
X
Y
Ebind [kcalmolÀ1
]
IC50 [nm][a]
FLT3
AKA
27
33
34
35
36
37
Ph Ph
O
O
O
Cl
Cl
Cl
À20.96
À20.99
À14.14
À24.33
À22.33
À24.95
322Æ30
345Æ33
43Æ10
53Æ2
H
Ph
H
Ph
H
between the pyrrolopyrimidine scaffold and the Cys694 back-
bone, but the position of the urea side chain of 38 was dis-
tinctly different from that of 37: the nitrogen atom of the iso-
xazole and the oxygen atom of the urea from compound 38
formed two critical hydrogen bonds with Lys644. This resulted
in the tail part of the side chain being pulled toward the DFG
loop, such that two additional NH groups from the urea
moiety could interact with Asp829. Furthermore, an unoccu-
pied hydrophobic pocket near the Phe621 residue was ob-
served, extending a suitable functional group at the C5-posi-
tion of the isoxazole for a CH–p interaction with the phenyl
group of phenylalanine. Based on an estimation of the dis-
tance between the extended group and Phe621, as well as the
780Æ108 80Æ19
Ph NH Cl
NH Cl
Ph NH
101Æ11
122Æ10
99Æ17
32 Æ3
42Æ2
9Æ4
Ph
H
H
H
[a] Values represent the mean ÆSD of at least two independent experi-
ments performed in duplicate.
lated (Table S3, Supporting Information). Of these molecules,
isoxazol-3-yl 38 was calculated to exhibit the lowest binding
energy (À26.61 kcalmolÀ1). To understand differences in bind-
ing conformations, 37 and 38 were analyzed within the FLT3
modeling structure (Figure 3). Similar binding was observed
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
4
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
rigidity of the modifying moiety, it was concluded that a tert-
butyl group would be best suited to make this interaction.
Compounds 38 (isoxazol-3-yl moiety) and 39 (5-(tert-butyl)-
isoxazol-3-yl moiety) were synthesized, and their enzymatic ac-
tivity was evaluated (Table 3). They were found to impart
slightly improved FLT3 inhibition and similar Aurora kinase A
of both cell lines (IC50 =7 nm), an unexpected result given its
similar anti-kinase activity to 38. This is probably due to the
improved cellular permeability that results from the adjusted
logP with the added the tert-butyl group.
Due to the high antiproliferative activity of 39, further pro-
tein immunoblot analysis in MV4-11 cells was conducted. Con-
sistent with the enzyme-based
assays, 39 displayed biomarker
inhibition
of
phospho-FLT3
Table 3. Calculated lipophilicity (clogP) of compounds 5, 38, and 39 and their inhibition of FLT3 and Aurora
(Tyr591) and phospho-Aurora
kinase A (Thr288) formation with
an IC50 value of less than 50 nm
(Figure 4), suggesting that the
apoptosis of MV4-11 cells was
caused by 39 through both FLT3
and Aurora kinase A inhibition.
kinase A (AKA) as well as the MOLM-13 and MV4-11 cell lines.
Synthesis
Compd
R
IC50 [nm][a]
AKA
clogP[b]
FLT3
MOLM-13
MV4-11
Pyrimidine chlorides 40a–e were
prepared according to published
procedures.[23,27–29] The pyrimi-
dine chlorides were first treated
with the commercially available
thiazole side chain 41 via nucle-
ophilic aromatic substitution, fol-
lowed by deprotection with tri-
fluoroacetic acid to produce the
5[20]
38
–
H
tBu
>10000
57Æ12
481Æ88
65Æ9
7Æ2
654Æ63
68Æ8
7Æ3
5.62
1.2
2.61
77Æ6
67Æ10
13Æ0.3
15Æ0.04
39
[a] Values represent the mean ÆSD of at least two independent experiments performed in duplicate for
enzyme-based assays and triplicate for cell-based assays. [b] Values were calculated using the Structure Design
Suite from Advanced Chemistry Development, Inc. (Toronto, ON, Canada).
inhibition relative to 37. In vitro testing of 38, 39 (most prom-
ising in terms of enzymatic activities), and 5 (control) was used
to evaluate their cellular activities in two acute monocytic leu-
kemia (AML) cell lines (MOLM-13 and MV4-11). Compound 5
(with potent Aurora kinase A inhibition and no FLT3 inhibition)
caused sub-micromolar antiproliferation, while compound 38
displayed approximately 7- and 10-fold improvement in anti-
proliferative activity in MOLM-13 and MV4-11, respectively.
Compound 39 was found to be a much more potent inhibitor
corresponding amine intermediates 42a–e (Scheme 3). Urea
formation of 42a–e with isocyanates was carried out to afford
urea compounds 27 and 33–37 in 17–28% overall yields. In
addition, phenyl carbamates 43a and b were synthesized fol-
lowing the procedure reported by Keith et al.[30] and further re-
acted with 42d under basic condition in 1,4-dioxane to gener-
ate 38 and 39 in 74–81% yield (Scheme 4).
Conclusions
Computer-guided drug design has been used in conjunction
with chemical synthesis and in vitro testing to identify dual
FLT3–Aurora A inhibitors. The first screening yielded com-
pound 27, which exhibited FLT3 and Aurora kinase A inhibitory
activity, with IC50 values of 322 and 43 nm, respectively. This
discovery necessitated the chemical synthesis of just 26 com-
pounds, emphasizing the highly efficient nature of in silico
drug design. Structure–activity relationship studies and bind-
ing energy analysis of compound 27 and its analogues from
Table 2 were conducted. The bioactivity data of these ana-
logues was found to be highly consistent with the results of
computer simulation, validating our use of binding energy cal-
culations as a reliable strategy for the development of im-
proved inhibitors.
In the second screening, both 38 and 39 were found to
have the most promising dual kinase activities and were sub-
jected to cellular activity tests. Dramatic antiproliferative effects
of 39 against MOLM-13 and MV4-11 were observed, and its
dual FLT3–Aurora A inhibitory mechanism was confirmed by
Figure 4. Western blotting analysis for cellular target modulation by com-
pound 39. Compound 39 inhibited phospho-FLT3 (Tyr591) and phospho-
Aurora kinase A (Thr288) formation in MV4-11 at concentrations of <2 and
~50 nm, respectively.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
5
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Scheme 3. Reagents and conditions: a) 1. Et3N, EtOH or tert-BuOH, reflux, 15–24 h, 2. CF3CO2H, CH2Cl2, RT, 2 h, 42–59% over two steps; b) 3-Cl-PhNCO, or
PhNCO, MeOH, CH2Cl2, RT, 16 h, 41–66%.
0.1% formic acid +10 mmol NH4OAc as mobile phase B.
Elution conditions: 0 min, 10% phase A + 90% phase B;
at 45 min, 90% phase A + 10% phase B; at 50 min, 10%
phase A + 90% phase B; at 60 min, 10% phase A +
90% phase B. The flow rate of the mobile phase was
0.5 mLminÀ1, and the injection volume of the sample
was 5 mL. Peaks were detected at 210 nm. Purity of all
the tested compounds were found to be >95% unless
otherwise stated.
1-(3-Chlorophenyl)-3-(5-(2-(5,6-diphenylfuro[2,3-d]pyri-
midin-4-ylamino)ethyl)thiazol-2-yl)urea (27): A solution
containing 40a[23] (92 mg, 0.30 mmol, 1.0 equiv), tert-
Scheme 4. Reagents and conditions: a) Et3N, 1,4-dioxanes, reflux, 5 h, 74–81%.
butyl 5-(2-aminoethyl)thiazol-2-ylcarbamate (41, 109 mg,
0.448 mmol, 1.5 equiv), and triethylamine (61 mg,
0.60 mmol, 2.0 equiv) in EtOH (3.0 mL) was heated at
reflux for 15 h. After the solvent was evaporated, the residue in
CH2Cl2 (10.0 mL) was treated with trifluoroacetic acid (4.0 mL), and
the mixture was stirred at room temperature for 2 h. The reaction
mixture was concentrated under reduced pressure, and the residue
was purified using gravity column chromatography on silica gel
(MeOH/CH2Cl2/NH4OH, 1:30:0.1 as the eluent) to give 42a (68 mg,
0.16 mmol) in 53% yield as a white solid. 3-Chlorophenyl isocya-
nate (126 mg, 0.820 mmol, 5.1 equiv) was added to a solution of
42a (68 mg, 0.16 mmol, 1.0 equiv) in a mixture of MeOH (1.0 mL)
and CH2Cl2 (7.0 mL). The mixture was stirred at room temperature
for 16 h. The reaction mixture was then concentrated under re-
duced pressure, and the residue was purified using gravity column
chromatography on silica gel (MeOH/CH2Cl2 =1:30 as the eluent)
to give 27 (45 mg, 0.079 mmol) in 49% yield as a white solid:
1H NMR (400 MHz, CD3OD+CDCl3): d=2.96 (t, J=6.0 Hz, 2H), 3.71
(t, J=6.0 Hz, 2H), 6.89 (s, 1H), 7.03–7.06 (m,1H), 7.24–7.49 (m,
12H), 7.68 (t, J=2.0 Hz, 1H), 8.33 ppm (s, 1H); LC–MS (ESI) m/z 567
[M+H]+.
Western blotting analysis. To summarize, we selected Aurora
hit 5 as a starting point, followed by two consecutive comput-
er-guided strategies, to rapidly and efficiently modify the side
chain and the core to identify compound 39 as a potential
anti-AML candidate for further development.
Experimental Section
Chemistry
General methods: All commercial chemicals and solvents were of
reagent grade and used without further purification unless other-
wise stated. All reactions were carried out under an atmosphere of
dry nitrogen and were monitored by TLC, using Merck 60 F254 silica
gel glass-backed plates (5ꢁ10 cm); zones were detected visually
under UV irradiation (254 nm) or by spraying with phosphomolyb-
dic acid reagent (Aldrich), followed by heating at 808C. Flash
column chromatography was carried out using silica gel (Merck
1-(3-Chlorophenyl)-3-(5-(2-(6-phenylfuro[2,3-d]pyrimidin-4-yl-
amino)ethyl)thiazol-2-yl)urea (33): A solution containing 40b[27]
(69 mg, 0.30 mmol, 1.0 equiv), tert-butyl 5-(2-aminoethyl)thiazol-2-
ylcarbamate (41, 109 mg, 0.448 mmol, 1.5 equiv), and triethylamine
(61 mg, 0.60 mmol, 2.0 equiv) in EtOH (3.0 mL) was heated at reflux
for 15 h. After the solvent was evaporated, the residue in CH2Cl2
(10 mL) was treated with trifluoroacetic acid (4.0 mL), and the mix-
ture was stirred at room temperature for 2 h. The reaction mixture
was concentrated under reduced pressure, and the residue was
purified using gravity column chromatography on silica gel
(MeOH/CH2Cl2/NH4OH, 1:30:0.1, as the eluent) to give 42b (45 mg,
0.13 mmol) in 44% yield as a white solid. 3-Chlorophenyl isocya-
nate (126 mg, 0.820 mmol, 6.3 equiv) was added to a solution of
1
Kieselgel 60, no. 9385, 230–400 mesh ASTM). H and 13C NMR spec-
tra were obtained with a Varian Mercury 300 or Varian Mercury 400
spectrometer, and the chemical shifts (d) were recorded in ppm
and reported relative to TMS or the solvent peak. High-resolution
mass spectra (HRMS) were measured with a Finnigan (MAT-95XL)
electron impact (EI) or by using Finnigan/Thermo Quest MAT 95XL
FAB mass spectrometer. Low-resolution mass spectra (LRMS) data
were measured on an Agilent MSD-1100 ESI-MS–MS system. Purity
of the final compounds was determined with an Hitachi 2000
series HPLC system using a C18 column (Agilent ZORBAX Eclipse
XDB-C18 5 mm, 4.6 mmꢁ150 mm), operating at 258C. Elution was
carried out using CH3CN as mobile phase A and water containing
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
6
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
42b (45 mg, 0.13 mmol, 1.0 equiv) in a mixture of MeOH (1.0 mL)
and CH2Cl2 (7.0 mL). The mixture was stirred at room temperature
for 16 h. The reaction mixture was then concentrated under re-
duced pressure and purified using gravity column chromatography
on silica gel (MeOH/CH2Cl2/NH4OH, 1:30:0.1, as the eluent) to give
33 (33 mg, 0.066 mmol) in 51% yield as a white solid: 1H NMR
(300 MHz, [D6]DMSO): d=3.06 (t, J=6.5 Hz, 2H), 3.68–3.80 (m, 2H),
7.02–7.10 (m, 1H), 7.15 (s, 1H), 7.27–7.59 (m, 6H), 7.70 (s, 1H), 7.80
(d, J=7.7 Hz, 2H), 8.19 (bs, 1H, NH), 8.29 (s, 1H), 9.16 ppm (s, 1H,
NH); LC–MS (ESI) m/z 491 [M+H]+; HPLC purity, 94.7%.
40e[29] (24 mg, 0.10 mmol, 1.0 equiv) and tert-butyl 5-(2-amino-
ethyl)thiazol-2-ylcarbamate (41, 80 mg, 0.33 mmol, 3.3 equiv) in
tert-butanol (1.0 mL) was heated at reflux for 24 h. After the sol-
vent was evaporated, the residue in CH2Cl2 (2.0 mL) was treated
with trifluoroacetic acid (1.0 mL), and the mixture was stirred at
room temperature for 4 h. The reaction mixture was concentrated
under reduced pressure, and the residue was purified using gravity
column chromatography on silica gel (MeOH/CH2Cl2/NH4OH,
1:9:0.1, as the eluent) to give 42e (20 mg, 0.059 mmol) in 59%
yield as
a yellow solid. 3-Chlorophenyl isocyanate (70 mg,
0.46 mmol, 7.7 equiv) was added to a solution of 42e (20 mg,
0.059 mmol, 1.0 equiv) in a mixture of MeOH (0.50 mL) and CH2Cl2
(3.5 mL). The mixture was stirred at room temperature for 16 h.
The reaction mixture was then concentrated under reduced pres-
sure and purified using gravity column chromatography on silica
gel (MeOH/CH2Cl2/NH4OH, 1:9:0.1, as the eluent) to give 36 (13 mg,
0.027 mmol) in 45% yield as a white solid: 1H NMR (400 MHz,
CD3OD): d=3.02 (t, J=6.0 Hz, 2H), 3.75 (t, J=6.0 Hz, 2H), 6.89 (s,
1H), 7.01–7.04 (m,1H), 7.07 (s, 1H), 7.24–7.38 (m, 7H), 7.67 (t, J=
2.0 Hz, 1H), 8.20 ppm (s, 1H); LC–MS (ESI) m/z 490 [M+H]+.
1-(3-Chlorophenyl)-3-(5-(2-(5-phenylfuro[2,3-d]pyrimidin-4-yl-
amino)ethyl)thiazol-2-yl)urea (34): A solution containing 40c[27]
(52 mg, 0.23 mmol, 1.0 equiv), tert-butyl 5-(2-aminoethyl)thiazol-2-
ylcarbamate (41, 109 mg, 0.448 mmol, 1.9 equiv), and triethylamine
(67 mg, 0.66 mmol, 2.9 equiv) in EtOH (2.0 mL) was heated at reflux
for 15 h. After the solvent was evaporated, the residue in CH2Cl2
(7.5 mL) was treated with trifluoroacetic acid (3.0 mL), and the mix-
ture was stirred at room temperature for 2 h. The reaction mixture
was concentrated under reduced pressure, and the residue was
purified using gravity column chromatography on silica gel
(MeOH/CH2Cl2, 1:20, as the eluent) to give 42c (34 mg, 0.10 mmol)
in 43% yield as a yellow solid. 3-Chlorophenyl isocyanate (126 mg,
0.820 mmol, 8.2 equiv) was added to a solution of 42c (34 mg,
0.10 mmol, 1.0 equiv) in a mixture of MeOH (1.0 mL) and CH2Cl2
(7.0 mL). The mixture was stirred at room temperature for 16 h.
The reaction mixture was then concentrated under reduced pres-
sure and purified using gravity column chromatography on silica
gel (MeOH/CH2Cl2, 1:20, as the eluent) to give 34 (32 mg,
0.066 mmol) in 66% yield as a white solid: 1H NMR (300 MHz,
CD3OD): d=3.06 (t, J=6.0 Hz, 2H), 3.80 (t, J=6.0 Hz, 2H), 6.99 (s,
1H), 7.06 (d, J=7.5 Hz, 1H), 7.25–7.44 (m, 7H), 7.68–7.69 (m, 2H),
8.35 ppm (s, 1H); LC–MS (ESI) m/z 491 [M+H]+.
1-Phenyl-3-(5-(2-(6-phenyl-7H-pyrrolo[2,3-d]pyrimidin-4-yl-
amino)ethyl)thiazol-2-yl)urea (37): Phenyl isocyanate (178 mg,
1.49 mmol, 8.3 equiv) was added to a solution of 42d (60 mg,
0.18 mmol, 1.0 equiv) in a mixture of MeOH (1.5 mL) and CH2Cl2
(10 mL). The mixture was stirred at room temperature for 16 h. The
reaction mixture was then concentrated under reduced pressure
and purified using gravity column chromatography on silica gel
(MeOH/CH2Cl2/NH4OH, 1:9:0.1, as the eluent) to give 37 (40 mg,
0.088 mmol) in 49% yield as a white solid: 1H NMR (400 MHz,
[D6]DMSO): d=3.04 (t, J=6.2 Hz, 2H), 3.65–3.75 (m, 2H), 6.93 (s,
1H), 7.00 (t, J=7.2 Hz, 1H), 7.12 (s, 1H), 7.24–7.32 (m, 3H), 7.39–
7.47 (m, 4H), 7.64 (bs, 1H), 7.77 (d, J=7.7 Hz, 2H), 8.14 (s, 1H),
8.92 (bs, 1H), 10.29 (bs, 1H), 12.04 ppm (bs, 1H); LC–MS (ESI) m/z
456 [M+H]+.
1-(3-Chlorophenyl)-3-(5-(2-(6-phenyl-7H-pyrrolo[2,3-d]pyrimidin-
4-ylamino)ethyl)thiazol-2-yl)urea (35):
A solution containing
40d[28] (45 mg, 0.24 mmol, 1.0 equiv) and tert-butyl 5-(2-amino-
ethyl)thiazol-2-ylcarbamate (41, 192 mg, 0.790 mmol, 3.3 equiv) in
tert-butanol (2.0 mL) was heated at reflux for 24 h. After the sol-
vent was evaporated, the residue in CH2Cl2 (4.0 mL) was treated
with trifluoroacetic acid (2.0 mL), and the mixture was stirred at
room temperature for 4 h. The reaction mixture was concentrated
under reduced pressure, and the residue was purified using gravity
column chromatography on silica gel (MeOH/CH2Cl2/NH4OH,
1:9:0.1, as the eluent) to give 42d (34 mg, 0.10 mmol) in 42% yield
as a yellow solid: 1H NMR (300 MHz, [D6]DMSO): d=2.90 (t, J=
6.8 Hz, 2H), 3.61 (td, J=6.8, 5.1 Hz, 2H), 6.65 (s, 2H), 6.68 (s, 1H),
6.93 (s, 1H), 7.27 (t, J=7.7 Hz, 1H), 7.42 (t, J=7.7 Hz, 2H), 7.59 (t,
J=5.1 Hz, 1H), 7.76 (d, J=7.7 Hz, 2H), 8.12 (s, 1H), 12.02 ppm (bs,
1H); LC–MS (ESI) m/z 337 [M+H]+. 3-Chlorophenyl isocyanate
(126 mg, 0.820 mmol, 8.2 equiv) was added to a solution of 42d
(34 mg, 0.10 mmol, 1.0 equiv) in a mixture of MeOH (1.0 mL) and
CH2Cl2 (7.0 mL). The mixture was stirred at room temperature for
16 h. The reaction mixture was then concentrated under reduced
pressure and purified using gravity column chromatography on
silica gel (MeOH/CH2Cl2/NH4OH, 1:9:0.1, as the eluent) to give 35
(20 mg, 0.041 mmol) in 41% yield as a white solid: 1H NMR
(400 MHz, [D6]DMSO): d=2.99–3.07 (m, 2H), 3.56–3.74 (m, 2H),
6.92 (s, 1H), 7.00–7.07 (m, 1H), 7.12 (s, 1H), 7.25–7.33 (m, 3H), 7.42
(t, J=7.7 Hz, 2H), 7.62–7.71 (m, 2H), 7.77 (d, J=7.7 Hz, 2H), 8.14 (s,
1H), 9.13 (bs, 1H), 10.44 (bs, 1H), 12.05 ppm (bs, 1H); LC–MS (ESI)
m/z 490 [M+H]+.
1-(1,2-Oxazol-3-yl)-3-(5-{2-[(6-phenyl-7H-pyrrolo[2,3-d]pyrimidin-
4-yl)amino]ethyl}-1,3-thiazol-2-yl)urea (38): A solution containing
42d (50 mg, 0.15 mmol, 1.0 equiv), phenyl 1,2-oxazol-3-ylcarba-
mate[30] (43a, 50 mg, 0.24 mmol, 1.6 equiv), and triethylamine
(45 mg, 0.44 mmol, 3.0 equiv) in 1,4-dioxane (3.5 mL) was heated
at reflux for 5 h. The reaction mixture was concentrated under re-
duced pressure, and the residue was purified using gravity column
chromatography on silica gel (MeOH/CH2Cl2/NH4OH, 1:9:0.1, as the
eluent) to give 38 (54 mg, 0.12 mmol) in 81% yield as a white
solid: 1H NMR (400 MHz, [D6]DMSO): d=3.02–3.09 (m, 2H), 3.64–
3.74 (m, 2H), 6.84 (d, J=1.2 Hz, 1H), 6.92 (d, J=1.8 Hz, 1H), 7.15 (s,
1H), 7.27 (t, J=7.4 Hz, 1H), 7.42 (dd, J=7.5, 7.4 Hz, 2H), 7.64 (t, J=
5.4 Hz, 1H), 7.77 (d, J=7.5 Hz, 2H), 8.14 (s, 1H), 8.75 (d, J=1.2 Hz,
1H), 9.84 (bs, 1H), 10.41 (bs, 1H), 12.04 ppm (bs, 1H); LC–MS (ESI)
m/z 447 [M+H]+.
1-(5-tert-Butyl-1,2-oxazol-3-yl)-3-(5-{2-[(6-phenyl-7H-pyrrolo[2,3-
d]pyrimidin-4-yl)amino]ethyl}-1,3-thiazol-2-yl)urea (39): A solu-
tion containing 42d (50 mg, 0.15 mmol, 1.0 equiv), phenyl (5-tert-
butylisoxazol-3-yl)carbamate[30] (43b, 70 mg, 0.27 mmol, 1.8 equiv),
and triethylamine (45 mg, 0.44 mmol, 3.0 equiv) in 1,4-dioxane
(3.5 mL) was heated at reflux for 5 h. The reaction mixture was
concentrated under reduced pressure, and the residue was purified
using gravity column chromatography on silica gel (MeOH/CH2Cl2/
NH4OH, 1:9:0.1, as the eluent) to give 39 (56 mg, 0.11 mmol) in
1
74% yield as a white solid: H NMR (400 MHz, [D6]DMSO): d=1.26
(s, 9H), 3.02–3.09 (m, 2H), 3.64–3.75 (m, 2H), 6.50 (s, 1H), 6.92 (s,
1H), 7.15 (s, 1H), 7.27 (t, J=7.1 Hz, 1H), 7.42 (dd, J=7.7, 7.1 Hz,
2H), 7.63 (bs, 1H), 7.77 (d, J=7.7 Hz, 2H), 8.14 (s, 1H), 9.73 (bs,
1-(3-Chlorophenyl)-3-(5-(2-(5-phenyl-7H-pyrrolo[2,3-d]pyrimidin-
4-ylamino)ethyl)thiazol-2-yl)urea (36):
A solution containing
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
7
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
1H), 10.37 (bs, 1H), 12.04 ppm (bs, 1H); LC–MS (ESI) m/z 447
[M+H]+.
se A analysis, the cell lysates were obtained from MV4-11 cells incu-
bated for 16 h with 50 ngmLÀ1 nocodazole, followed by drug treat-
ment for 2 h at the indicated concentrations.
Biological methods
Acknowledgements
FLT3 kinase assay: FLT3 assays were conducted as previously de-
scribed.[31] GST-FLT3-KDWT containing the FLT3 kinase catalytic
domain (residues Tyr567–Ser993) were expressed in Sf9 insect
cells transfected with the baculovirus containing pBac-PAK8-GST-
FLT3-KD plasmid. FLT3WT Kinase-Glo assays were carried out in 96-
well plates at 308C for 4 h with tested compounds at a final
volume of 50 mL including the following components: 75 ng GST-
FLT3-KDWT proteins, 25 mm HEPES, pH 7.4, 4 mm MnCl2, 10 mm
MgCl2, 2 mm DTT, 0.02% Triton X-100, 0.1 mgmLÀ1 bovine serum
albumin, 25 mm Her2 peptide substrate, 0.5 mm Na3VO4, and 1 mm
ATP.
Financial support from the Taiwanese National Health Research
Institutes and the National Science Council, Taiwan (NSC-101-
2113-M-400-002-MY4 and NSC-102-2325-B-400-003 to H.-P.H.,
NSC 102-2113-M-400-003-MY3 to H.-Y.S.) is gratefully acknowl-
edged. Y.C.H. is supported by a postdoctoral fellowship from the
National Science Council, Taiwan.
Keywords: Aurora kinase A · drug design · FLT3 · inhibitors ·
leukemia
Aurora kinase A assay: Aurora kinase A assays were conducted as
previously described.[27] The recombinant GST-Aurora kinase A (resi-
dues Ser123–Ser401) containing the kinase domain was expressed
in Sf9 insect cells. The kinase assay was carried out in 96-well
plates with tested compound in a final volume of 50 mL at 378C
for 90 min with the following components: 50 mm Tris-HCl pH 7.4,
10 mm NaCl, 10 mm MgCl2, 0.01% BSA, 5 mm ATP, 1 mm DTT, 15 mm
tetra(-LRRASLG) peptide, and 150 ng recombinant Aurora kinase A.
[1] M. E. M. Noble, J. A. Endicott, L. N. Johnson, Science 2004, 303, 1800–
1805.
[2] J. M. Goldman, J. V. Melo, N. Engl. J. Med. 2003, 349, 1451–1464.
[3] P. Cohen, D. R. Alessi, ACS Chem. Biol. 2013, 8, 96–104.
[4] US Food and Drug Administration (FDA), Drugs@FDA, FDA-Approved
Drug Products: http://www.accessdata.fda.gov/scripts/cder/drugsatfda/
index.cfm.
Following incubation, 50 mL Kinase-Glo Plus Reagent (Promega,
Madison, WI, USA) was added, and the mixture was incubated at
258C for 20 min. A 70 mL aliquot of each reaction mixture was
transferred to a black microliter plate, and luminescence was mea-
sured on a Wallac Vector 1420 multilabel counter (PerkinElmer,
Shelton, CT, USA).
[5] G. Manning, D. B. Whyte, R. Martinez, T. Hunter, S. Sudarsanam, Science
2002, 298, 1912–1934.
[6] A. Markovic, K. L. MacKenzie, R. B. Lock, Int. J. Biochem. Cell Biol. 2005,
37, 1168–1172.
[7] D. G. Gilliland, J. D. Griffin, Blood 2002, 100, 1532–1542.
[8] B. W. Parcells, A. K. Ikeda, T. Simms-Waldrip, T. B. Moore, K. M. Sakamoto,
Stem Cells 2006, 24, 1174–1184.
[9] S. H. Chu, D. Small, Drug Resist. Updates 2009, 12, 8–16.
[10] T. Kindler, D. B. Lipka, T. Fischer, Blood 2010, 116, 5089–5102.
[11] D. W. Sternberg, J. D. Licht, Curr. Opin. Hematol. 2005, 12, 7–13.
[12] C. H. A. Cheung, M. S. Coumar, H.-P. Hsieh, J.-Y. Chang, Expert Opin.
Invest. Drugs 2009, 18, 379–398.
[13] A. A. Mortlock, K. M. Foote, N. M. Heron, F. H. Jung, G. Pasquet, J.-J. M.
Lohmann, N. Warin, F. Renaud, C. De Savi, N. J. Roberts, T. Johnson, C. B.
Dousson, G. B. Hill, D. Perkins, G. Hatter, R. W. Wilkinson, S. R. Wedge,
S. P. Heaton, R. Odedra, N. J. Keen, C. Crafter, E. Brown, K. Thompson, S.
Brightwell, L. Khatri, M. C. Brady, S. Kearney, D. McKillop, S. Rhead, T.
Parry, S. Green, J. Med. Chem. 2007, 50, 2213–2224.
[14] A. S. Moore, A. Faisal, D. Gonzalez de Castro, V. Bavetsias, C. Sun, B.
Atrash, M. Valenti, A. de Haven Brandon, S. Avery, D. Mair, F. Mirabella, J.
Swansbury, A. D. J. Pearson, P. Workman, J. Blagg, F. I. Raynaud, S. A.
Eccles, S. Linardopoulos, Leukemia 2012, 26, 1462–1470.
[15] V. Bavetsias, S. Crumpler, C. Sun, S. Avery, B. Atrash, A. Faisal, A. S.
Moore, M. Kosmopoulou, N. Brown, P. W. Sheldrake, K. Bush, A. Henley,
G. Box, M. Valenti, A. d. H. Brandon, F. I. Raynaud, P. Workman, S. A.
Eccles, R. Bayliss, S. Linardopoulos, J. Blagg, J. Med. Chem. 2012, 55,
8721–8734.
[16] P. P. Zarrinkar, R. N. Gunawardane, M. D. Cramer, M. F. Gardner, D. Brig-
ham, B. Belli, M. W. Karaman, K. W. Pratz, G. Pallares, Q. Chao, K. G.
Sprankle, H. K. Patel, M. Levis, R. C. Armstrong, J. James, S. S. Bhagwat,
Blood 2009, 114, 2984–2992.
[17] D. M. Goldstein, N. S. Gray, P. P. Zarrinkar, Nat. Rev. Drug Discovery 2008,
7, 391–397.
[18] F. Ciardiello, G. Tortora, N. Engl. J. Med. 2008, 358, 1160–1174.
[19] H. A. Burris III, Oncologist 2004, 9, 10–15.
[20] M. S. Coumar, C.-Y. Chu, C.-W. Lin, H.-Y. Shiao, Y.-L. Ho, R. Reddy, W.-H.
Lin, C.-H. Chen, Y.-H. Peng, J.-S. Leou, T.-W. Lien, C.-T. Huang, M.-Y. Fang,
S.-H. Wu, J.-S. Wu, S. K. Chittimalla, J.-S. Song, J. T.-A. Hsu, S.-Y. Wu, C.-C.
Liao, Y.-S. Chao, H.-P. Hsieh, J. Med. Chem. 2010, 53, 4980–4988.
[21] Y.-H. Peng, H.-Y. Shiao, C.-H. Tu, P.-M. Liu, J. T.-A. Hsu, P. K. Amancha, J.-S.
Wu, M. S. Coumar, C.-H. Chen, S.-Y. Wang, W.-H. Lin, H.-Y. Sun, Y.-S. Chao,
P.-C. Lyu, H.-P. Hsieh, S.-Y. Wu, J. Med. Chem. 2013, 56, 3889–3903.
Cell lines and MTS cell viability assay: The MTS cell viability assay
was conducted as previously described.[31] The MOLM-13 human
leukemia cell line was obtained from the German Resource Centre
for Biological Material (DSMZ, Braunschweig, Germany); the MV4-
11 cell line was purchased from the American Type Culture Collec-
tion (ATCC, Manassas, VA, USA). All leukemia cell lines were main-
tained in RPMI 1640 medium supplemented with 10% fetal bovine
serum (FBS), 10 UmLÀ1 penicillin, and 10 gmLÀ1 streptomycin at
378C and 5% CO2. To determine cell viability after drug treatment,
assays were performed by seeding 10000 cells (leukemia cell lines)
per well in a 96-well culture plate. After 16 h, cells were then treat-
ed with vehicle or various concentrations of compound in medium
for 72 h. Viable cells were quantitated using the MTS method
(Promega, Madison, WI, USA) according to the manufacturer’s rec-
ommended protocol. Results were determined by measuring the
absorbance at 490 nm using a plate reader (Victor2; PerkinElmer,
Shelton, CT, USA). The IC50 value was defined as the amount of
compound that caused a 50% reduction in cell viability in compari-
son with DMSO-treated (vehicle) control and was calculated using
Prism version 4 software (GraphPad, San Diego, CA, USA).
Western blotting analysis: Western blot analysis was conducted as
previously described.[32,33] Primary antibodies against phospho-FLT3
(Tyr591) and phospho-Aurora kinase A (Thr288) were purchased
from Cell Signaling Technology. The anti-FLT3 antibody and anti-
Aurora kinase A antibody were purchased from Santa Cruz Biotech-
nology and Upstate, respectively. The anti-b-actin monoclonal anti-
body was purchased from GeneTex. The secondary antibodies
horseradish peroxidase (HRP)-linked goat anti-rabbit IgG were pur-
chased from Jackson Immuno. MV4-11 cells were incubated with
compound 39 for 2 h at the indicated concentrations. Cell lysates
were prepared and analyzed by immunoblotting. For Aurora kina-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
8
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
[22] The details of the homology modeling structure will be published sepa-
rately.
[29] B. C. Bookser, B. G. Ugarkar, M. C. Matelich, R. H. Lemus, M. Allan, M. Tsu-
chiya, M. Nakane, A. Nagahisa, J. B. Wiesner, M. D. Erion, J. Med. Chem.
2005, 48, 7808–7820.
[30] J. M. Keith, R. Apodaca, W. Xiao, M. Seierstad, K. Pattabiraman, J. Wu, M.
Webb, M. J. Karbarz, S. Brown, S. Wilson, B. Scott, C.-S. Tham, L. Luo, J.
Palmer, M. Wennerholm, S. Chaplan, J. G. Breitenbucher, Bioorg. Med.
Chem. Lett. 2008, 18, 4838–4843.
[31] W.-H. Lin, S.-Y. Hsieh, S.-C. Yen, C.-T. Chen, T.-K. Yeh, T. Hsu, C.-T. Lu, C.-P.
Chen, C.-W. Chen, L.-H. Chou, Y.-L. Huang, A.-H. Cheng, Y.-I. Chang, Y.-J.
Tseng, K.-R. Yen, Y.-S. Chao, J. T.-A. Hsu, W.-T. Jiaang, Bioorg. Med. Chem.
2011, 19, 4173–4182.
[32] J. T.-A. Hsu, T.-K. Yeh, S.-C. Yen, C.-T. Chen, S.-Y. Hsieh, T. Hsu, C.-T. Lu, C.-
W. Chen, L.-H. Chou, C.-H. Chiu, Y.-I. Chang, Y.-J. Tseng, K.-R. Yen, Y.-S.
Chao, W.-H. Lin, W.-T. Jiaang, Bioorg. Med. Chem. Lett. 2012, 22, 4654–
4659.
[23] M. S. Coumar, M.-T. Tsai, C.-Y. Chu, B.-J. Uang, W.-H. Lin, C.-Y. Chang, T.-Y.
Chang, J.-S. Leou, C.-H. Teng, J.-S. Wu, M.-Y. Fang, C.-H. Chen, J. T.-A.
Hsu, S.-Y. Wu, Y.-S. Chao, H.-P. Hsieh, ChemMedChem 2010, 5, 255–267.
[24] J.-M. Wu, C.-T. Chen, M. S. Coumar, W.-H. Lin, Z.-J. Chen, J. T.-A. Hsu, Y.-H.
Peng, H.-Y. Shiao, W.-H. Lin, C.-Y. Chu, J.-S. Wu, C.-T. Lin, C.-P. Chen, C.-C.
Hsueh, K.-Y. Chang, L.-P. Kao, C.-Y. F. Huang, Y.-S. Chao, S.-Y. Wu, H.-P.
Hsieh, Y.-H. Chi, Proc. Natl. Acad. Sci. USA 2013, 110, E1779–E1787.
[25] D. Fancelli, J. Moll, M. Varasi, R. Bravo, R. Artico, D. Berta, S. Bindi, A. Ca-
meron, I. Candiani, P. Cappella, P. Carpinelli, W. Croci, B. Forte, M. L. Gior-
gini, J. Klapwijk, A. Marsiglio, E. Pesenti, M. Rocchetti, F. Roletto, D. Se-
verino, C. Soncini, P. Storici, R. Tonani, P. Zugnoni, P. Vianello, J. Med.
Chem. 2006, 49, 7247–7251.
[26] Y.-Y. Ke, H.-Y. Shiao, Y. C. Hsu, C.-Y. Chu, W.-C. Wang, Y.-C. Lee, W.-H. Lin,
C.-H. Chen, J. T. A. Hsu, C.-W. Chang, C.-W. Lin, T.-K. Yeh, Y.-S. Chao, M. S.
Coumar, H.-P. Hsieh, ChemMedChem 2013, 8, 136–148.
[27] H.-Y. Shiao, M. S. Coumar, C.-W. Chang, Y.-Y. Ke, Y.-H. Chi, C.-Y. Chu, H.-Y.
Sun, C.-H. Chen, W.-H. Lin, K.-S. Fung, P.-C. Kuo, C.-T. Huang, K.-Y. Chang,
C.-T. Lu, J. T.-A. Hsu, C.-T. Chen, W.-T. Jiaang, Y.-S. Chao, H.-P. Hsieh, J.
Med. Chem. 2013, 56, 5247–5260.
[33] M. S. Coumar, J.-S. Leou, P. Shukla, J.-S. Wu, A. K. Dixit, W.-H. Lin, C.-Y.
Chang, T.-W. Lien, U.-K. Tan, C.-H. Chen, J. T.-A. Hsu, Y.-S. Chao, S.-Y. Wu,
H.-P. Hsieh, J. Med. Chem. 2009, 52, 1050–1062.
Received: December 30, 2013
[28] S. J. Kaspersen, C. Sørum, V. Willassen, E. Fuglseth, E. Kjøbli, G. Bjørkøy,
E. Sundby, B. H. Hoff, Eur. J. Med. Chem. 2011, 46, 6002–6014.
Published online on && &&, 0000
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
9
&
ÞÞ
These are not the final page numbers!

FULL PAPERS
Y. Chang Hsu, Y.-Y. Ke, H.-Y. Shiao,*
C.-C. Lee, W.-H. Lin, C.-H. Chen, K.-J. Yen,
J. T.-A. Hsu, C. Chang, H.-P. Hsieh*
In silico selection: We selected an
Aurora hit compound as a starting
point, followed by two consecutive
computer-guided strategies to rapidly
and efficiently modify the side chain
and core. These efforts resulted in the
identification of a potential FLT3–Aurora
A inhibitor for further development to
treat acute myeloid leukemia (AML).
&& – &&
Facile Identification of Dual FLT3–
Aurora A Inhibitors: A Computer-
Guided Drug Design Approach
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&10
&
ÞÞ
These are not the final page numbers!