Design and combinatorial synthesis of a novel kinase-focused library using click chemistry-based fragment assembly

Article abstract of DOI:10.1016/j.bmcl.2011.10.076

Fragment-based lead discovery is a new approach for lead generation that has emerged in the past decade. Because the initial fragments identified in the fragment screening typically show weak binding affinity, an intensive medicinal chemistry effort would be required to grow initial fragments into a potential lead compound. Here we demonstrate a kinase focused evolved fragment (KFEF) library, constructed by click chemistry-based fragment assembly, that is a valuable source of kinase inhibitors. This combinatorial assembly of two fragments, kinase-privileged alkyne fragments and diversified azide fragments, by two cycloaddition reactions shows a unique potential for the one-step synthesis of structurally diverse evolved fragments. The screening of this triazole-based KFEF library allowed the rapid identification of potent lead candidates for FLT3 and GSK3β kinase.

Full text of DOI:10.1016/j.bmcl.2011.10.076

Bioorganic & Medicinal Chemistry Letters 22 (2012) 591–596
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Design and combinatorial synthesis of a novel kinase-focused library using
click chemistry-based fragment assembly
Takayuki Irie a, , Ikuo Fujii b, Masaaki Sawa a
⇑
a Drug Discovery Research, Carna Biosciences, Inc., 3rd Floor, BMA, 1-5-5 Minatojima-Minamimachi, Chuo-ku, Kobe 650-0047, Japan
b Department of Biological Science, Graduate School of Science, Osaka Prefecture University, 1-1, Gakuen-cho, Naka-ku, Sakai, Osaka 599-8531, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
Fragment-based lead discovery is a new approach for lead generation that has emerged in the past dec-
ade. Because the initial fragments identified in the fragment screening typically show weak binding affin-
ity, an intensive medicinal chemistry effort would be required to grow initial fragments into a potential
lead compound. Here we demonstrate a kinase focused evolved fragment (KFEF) library, constructed by
click chemistry-based fragment assembly, that is a valuable source of kinase inhibitors. This combinato-
rial assembly of two fragments, kinase-privileged alkyne fragments and diversified azide fragments, by
two cycloaddition reactions shows a unique potential for the one-step synthesis of structurally diverse
evolved fragments. The screening of this triazole-based KFEF library allowed the rapid identification of
potent lead candidates for FLT3 and GSK3b kinase.
Received 12 August 2011
Revised 2 October 2011
Accepted 22 October 2011
Available online 30 October 2011
Keywords:
Kinase inhibitors
Click chemistry
Triazoles
Kinase-privileged fragment
Fragment evolution
Ó 2011 Elsevier Ltd. All rights reserved.
Protein kinases play key roles in the signaling pathways that
regulate crucial cellular processes such as proliferation, differenti-
ation and survival.1 Deregulation of protein kinases has been impli-
cated in a variety of diseases, and therefore the inhibition of
protein kinases has emerged as an important target for drug
discovery.2
chemistry’ methodology to readily identify an evolved fragment
compound as a lead candidate. Through a screening campaign of
the constructed KFEF library against a panel of kinases, hit com-
pounds having excellent inhibitory potency against FLT3 and
GSK3b were identified. In addition, the kinase selectivity of hit
compounds was discussed, and the possible binding mode in the
active site of the enzyme was analyzed by computer modeling.
Protein kinases share high structural homologies within their
catalytic domains, and most kinase inhibitors bind to the highly
conserved ATP binding cleft via hydrogen bonds with the ‘hinge’
residues, which are considered as the most important binding
interactions for the recognition of kinases. To incorporate the ki-
nase hinge binding structural features into the KFEF library as ki-
nase-privileged fragments, we extracted chemical structures that
bind to the kinase hinge region from kinase inhibitors reported
in the literature7 (Fig. 1). To construct the KFEF library, we chose
the alkyne-azide cycloaddition reaction known as ‘click reac-
tion’.8,9 The triazole ring formed in this reaction produces planar
structures,10 which would fit to the narrow ATP binding site of ki-
nases. Namely, the KFEF library is constructed from two fragment
parts using click reactions: the kinase-privileged alkyne fragments
and diversified azide fragments which could increase chemical
diversity and allow for modification of drug-like properties, such
as inhibitory potency, target selectivity, and physicochemical
properties.
Recently, fragment-based lead discovery has been used to find
very small drug fragments as medicinal chemistry starting
points,3,4 and successfully delivered novel lead compounds for
some kinase targets.5 However, initial fragments identified in the
fragment screening typically show weak binding affinity. Those
fragments should be converted into hit or lead compounds
(evolved fragments) with sufficient potency keeping smaller
molecular weight and lower lipophilicity compared to traditional
lead compounds,6 and these evolved molecules would make the
subsequent optimization step easier. In order to evolve the initial
fragments into lead compounds efficiently, integration of other
techniques such as X-ray crystallography, or nuclear magnetic res-
onance spectroscopy would be necessary to elucidate the binding
modes of multiple fragments. Even having the structural data, an
intensive medicinal chemistry effort would be still required to
grow initial fragments or assemble multiple fragments into a po-
tential lead compounds.
In this Letter, we describe the design and synthesis of a novel
kinase focused evolved fragment (KFEF) library using the ‘click
With a set of alkyne and azide fragments, we applied two cyclo-
addition reaction conditions: a typical Cu-catalyzed click reaction
to obtain 1,4-disubstituted 1,2,3-triazoles, and a Ru-catalyzed
⇑
Corresponding author. Tel.: +81 78 302 7040; fax: +81 78 302 7086.
E-mail address: takayuki.irie@carnabio.com (T. Irie).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.10.076

592
T. Irie et al. / Bioorg. Med. Chem. Lett. 22 (2012) 591–596
Figure 1. Design concept of KFEF library.
cycloaddition condition11 to construct a 1,5-disubstituted 1,2,3-tri-
azole library.
H2O; RT, 1 day).13 Two 1,5-isomers (5A1 and 5A2) were also
prepared using the Ru catalyzed cycloaddition conditions reported
by Fokin group (cat. (Cp⁄RuCl)4, DMF; microwave heating, 110 °C,
20 min).14 Both 1,4-isomers and 1,5-isomers were successfully ob-
tained in a regioselective manner15 and the structures were con-
firmed by 1H NMR (400 MHz, DMSO-d6)16: the chemical shifts of
H-5 in 1,4-disubstituted triazoles were appeared at lower field
(ꢀ9.6 ppm) than those of H-4 in 1,5-disubstituted triazoles
(ꢀ8.3 ppm).
Before constructing the KFEF library, we performed a prelimin-
ary study to obtain an initial data set to confirm synthetic feasibil-
ity and the potential of a triazole structure to act as a kinase
inhibitor, by screening against a panel of kinases. Based on the re-
port by Graneto et al. that 1,5-disubstituted 5-membered ring ana-
logs are potent MAP kinase inhibitors,12 4-ethynylpyridine A,
possessing a nitrogen atom as a potential hydrogen bond acceptor,
was employed as a privileged fragment.
4-Chlorophenyl azide 1 and 3-chlorophenyl azide 2 were cho-
sen as diversified fragments which can be synthesized from corre-
sponding anilines employing typical diazotization condition
(Scheme 1). With these fragments, two 1,4-disubstituted triazoles
(4A1 and 4A2) were synthesized using a standard Cu catalyzed
click reaction condition (cat. CuSO4, sodium ascorbate, t-BuOH/
To test their inhibitory activity for kinases, the four triazoles
(4A1, 4A2, 5A1, and 5A2) were screened against a panel of 46 ki-
nases at 10 lM concentrations, and the percent inhibition against
each kinase is expressed colorimetrically (Fig. 2). It is noteworthy
that 1,4-disubstituted triazoles demonstrated potent inhibition
primary against
a subset of serine-threonine kinases (STKs)
(DYRK1B, GSK3b, etc.) but were inactive against a majority of tyro-
sine kinases (TKs), while 1,5-disubstituted triazoles preferably
bound to FLT3 (TK) with an excellent selectivity. Different patterns
were observed between 1,4- and 1,5-regio-isomers, suggesting the
KFEF library could span a wide range of the kinome tree with a
small subset of fragments by using two click conditions.
As the initial concept of the KFEF library was validated, we next
designed a small KFEF library composed of 5 alkyne and 8 azide
fragments (a total of 80 compounds, Fig. 3). The library was pre-
pared using the same synthetic procedure described above. Most
of the reactions proceeded very well except azide 5 and 7 that
might have poor reactivity, and the reaction did not produce
4D5, 5C5, 5B7, and 5E7. Thus, a total of 76 compounds successfully
obtained (success rate = 95%) were subjected to screening to iden-
tify promising lead compounds against kinase targets.
FLT3 is a kinase responsible for survival and proliferation of leu-
kemic blast cells. In acute myeloid leukemia (AML), constitutively
activated mutations of FLT3 such as internal tandem duplications
Scheme 1. Regio-selective synthesis of 1,4- and 1,5-disubstituted 1,2,3-triazoles.
Regents and conditions: (a) 1 mol % CuSO4, sodium
L-ascorbate (0.1 equiv), t-BuOH/
H2O (1:1), rt, 1 day; (b) 2.5 mol % [Cp⁄RuCl]4, DMF, microwave, 110 °C, 20 min.

T. Irie et al. / Bioorg. Med. Chem. Lett. 22 (2012) 591–596
593
20 TKs
26 STKs
4A1
4A2
5A1
5A2
% inh.
100%
0%
Figure 2. Protein kinase profiling of triazoles 4A1, 4A2, 5A1, and 5A2. The compounds were tested against a panel of 46 kinases. The color bar represents the percent
inhibition at 10 M concentrations.
l
Figure 3. Click construction of 1,2,3-triazole KFEF library.
are found in up to 30% of the patients, and these FLT3 mutations
are strongly associated with poor prognosis in AML.17 On the other
hand, GSK3b is one of the major kinases responsible for abnormal
hyperphosphorylation of tau protein. Some studies suggested that
this event would promote neurofibrillary tangle formation, which
is involved in the pathogenesis of Alzheimer’s disease (AD).18
Accordingly, inhibitors of FLT3 or GSK3b have received much
attention as potential drugs for AML or AD, respectively. Therefore
FLT3 and GSK3b were selected as target kinases for further study
based on the preliminary panel screening results.
A screening of the KFEF library was conducted against both FLT3
and GSK3b with the results shown in Figure 4. Apparently, there
are clear differences in the hit patterns of the two kinases. For
FLT3, 1,4-isomers incorporating the privileged fragments (A–E)
generally showed weak or moderate inhibitory potencies similar
to the initial compounds, 4A1 and 4A2 (Fig. 4, upper left). In con-
trast, 1,5-isomers exhibited a clear preference for the privileged
fragment A, suggesting the 4-pyridyl group could be an excellent
hinge binder for FLT3 (Fig. 4, upper right). On the other hand, some
1,4-isomers (4A1, 4A2 and 4C5) showed strong inhibition against
GSK3b while most of the remaining compounds showed no inhibi-
tion (Fig. 4, lower left). Interestingly, a small change in the nature
of the para aromatic substituent in the R2 group, ‘Cl’ (4A1) to ‘F’
(4A4) or ‘Me’ (4A3), dramatically diminished inhibitory potency,
indicating a narrow window for activity as observed in some other
kinase inhibitors.19 Contrary to the preliminary screening results,
some hit compounds (5A5, 5A7 and 5D7) for GSK3b were identified
within the 1,5-isomer library (Fig. 4, lower right).
The compounds showing >40% inhibition against each kinase in
the screening were subjected to IC50 determination with the re-
sults shown in Tables 1 and 2. It can be seen from the results (Ta-
ble 1) that the 1,5-isomers exhibited better potency against FLT3
compared with 1,4-isomers overall, as expected from the screening
result at 10 lM concentration. However, 1,4-isomer 4C4 having a

594
T. Irie et al. / Bioorg. Med. Chem. Lett. 22 (2012) 591–596
Figure 4. Inhibitory activity of KFEF library (left: 1,4-isomers; right: 1,5-isomers) against FLT3 (up) and GSK3b (down) at 10 lM concentration. To each cell was allocated a
compound composed of an alkyne (AꢀE) and an azide (1ꢀ8). The percent inhibition is depicted colorimetrically according to the scale shown, while the synthetically
unavailable in gray.
Table 1
Table 2
IC50 values of hit compounds against FLT3
IC50 values of hit compounds against GSK3b
ID
R1
R2
IC50
(lM)
BEIa
ID
R1
R2
IC50 (
l
M)
BEIa
1,4-Disubstituted triazoles
1,4-Disubstituted triazoles
4A1
4A2
4A4
4C1
4C3
4C4
4C5
4D3
4D6
4D7
4D8
4-Pyridyl
4-Pyridyl
4-Pyridyl
2-Pyridyl
2-Pyridyl
2-Pyridyl
2-Pyridyl
6-Quinazolyl
6-Quinazolyl
6-Quinazolyl
6-Quinazolyl
4-Chlorophenyl
3-Chlorophenyl
4-Methylphenyl
4-Chlorophenyl
4-Fluorophenyl
4-Methylphenyl
2-Methylphenyl
4-Fluorophenyl
Benzyl
5.8
10.5
6.6
6.3
6.7
1.1
9.3
12.4
6.3
9.7
20.4
19.4
21.9
20.3
21.5
25.2
21.3
16.8
18.1
15.1
16.0
4A1
4A2
4C4
4C5
4D6
4-Pyridyl
4-Pyridyl
2-Pyridyl
2-Pyridyl
6-Quinazolyl
4-Chlorophenyl
3-Chlorophenyl
4-Methylphenyl
2-Methylphenyl
Benzyl
1.1
1.8
>30
0.49
9.5
23.2
22.4
—
26.7
17.5
1,5-Disubstituted triazoles
5A5
5A7
5D7
4-Pyridyl
4-Pyridyl
6-Quinazolyl
2-methylphenyl
2-adamantyl
2-adamantyl
6.3
2.6
2.8
22.0
19.9
16.8
2-Adamantyl
2-Morpholinoethyl
Staurosporin
0.011
17.1
10.9
a
1,5-Disubstituted triazoles
BEI (binding efficiency index) = ÀlogIC50/MW (KDa).
5A1
5A2
5A3
5A4
5A5
5A7
5C7
5D7
4-Pyridyl
4-Pyridyl
4-Pyridyl
4-Pyridyl
4-Pyridyl
4-Pyridyl
2-Pyridyl
6-Quinazolyl
4-Chlorophenyl
3-Chlorophenyl
4-Fluorophenyl
4-Methylphenyl
2-Methylphenyl
2-Adamantyl
1.7
4.4
1.4
2.3
6.2
4.8
10.2
3.1
22.5
20.9
24.4
23.9
22.0
19.0
17.8
16.6
an increase of the inhibitory activity for GSK3b maintaining FLT3
inhibition, implying an introduction of a bulky fragment might re-
sult in a non-selective kinase inhibitor library.
In order to evaluate the potential of these compounds as lead
candidates, the binding efficiency indices (BEI) were calculated
from their IC50 values and molecular weights (Tables 1 and 2), be-
cause the concept of BEI has been used as a parameter to prioritize
successful fragments for lead optimization.20–22 The most potent
2-Adamantyl
2-Adamantyl
Staurosporin
0.00018
20.9
a
BEI (binding efficiency index) = ÀlogIC50/MW (KDa).
inhibitors (4C4: IC50 = 1.1 lM for FLT3 and 4C5: IC50 = 0.49 lM
4-methylphenyl group at the R2 position showed the best potency
for FLT3 inhibition (IC50 = 1.1 M) contrary to the expectation. In
addition, the compound 4C4 showed no activity against GSK3b
(IC50: >30 M, Table 2). Surprisingly, a shifting of the methyl
groups from the 4-position to the 2-position resulted in a dramatic
change in the inhibitory profile, namely the compound 4C5 exhib-
ited potent inhibitory activity against GSK3b with an IC50 value of
for GSK3b) have excellent binding efficiencies (BEI = 25.2 and
26.7, respectively), which are superior to that of a potent non-spe-
cific kinase inhibitor, staurosporin (BEI = 20.9 for FLT3 and 17.1 for
GSK3b), indicating that a screening of KFEF library would be a
promising approach to identify a lead candidate rapidly.
To predict the binding modes of the hit compounds with GSK3b,
4A1 and 4C5 were computationally docked into GSK3b using Lib-
Dock (Fig. 5).23 The docking study proposed that 4A1 would bind
to GSK3b in a similar position to where ATP binds. The pyridine
nitrogen from the privileged fragment, as expected, forms a
l
l
0.49 lM but showed weak inhibition for FLT3 (IC50 = 9.3 lM) (Ta-
bles 1 and 2). Interestingly, the introduction of a bulky adamantyl
group into the R2 position of 1,5-isomers (5A7 and 5D7) resulted in

T. Irie et al. / Bioorg. Med. Chem. Lett. 22 (2012) 591–596
595
OH
Leu132
HN
O
HO
Hydro-
Phobic
pocket 1
HN
O
O
Adenine
binding
pocket
hinge
N
Phosphate binding
region
H
Val135
O Hydro-
phobic
Sugar
pocket
HN
pocket 2
4A1
4C5
Figure 5. Docking study of hit compounds for GSK3b. (Upper) X-ray crystal structure of GSK3b with ATP–ANP (PDB code 1pyx) and a 2D representation of the ATP binding
site. (Lower) 4A1 and 4C5 were computationally docked into GSK3b using LibDock. The crystal structure used in the docking study was retrieved from PDB (1pyx for 4A1 and
1q3d for 4C5).
5. Erlanson, D. A. In Kinase Inhibitor Drugs; Li, R., Stafford, J. A., Eds.; John Wiley &
Sons, Inc.: Hoboken, NJ, USA, 2009; pp 461–483.
6. Rees, D. C.; Congreve, M.; Murray, C. W.; Carr, R. Nat. Rev. Drug Disc. 2004, 3,
hydrogen bond with the hinge backbone amide. On the other hand,
the binding mode predicted for 4C5 is very different from that of
4A1. The triazole nitrogen can form a hydrogen bond with the
hinge and therefore 4C5 could bind to the hydrophobic pocket 1
and 2.
In summary, 1,4- or 1,5-disubsituted 1,2,3-triazoles having ki-
nase privileged fragments were designed and synthesized as a ki-
nase focused evolved fragment (KFEF) library, and evaluated as
potential kinase inhibitors. We have succeeded in hit generation
660.
7. Ghose, A. K.; Herbertz, T.; Pippin, D. A.; Salvino, J. M.; Mallamo, J. P. J. Med.
Chem. 2008, 51, 5149 (a review) and references therein.
8. Kolb, H. C.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int. Ed. 2001, 40, 2004.
9. Kolb, H. C.; Sharpless, K. B. Drug Discovery Today 2003, 8, 1128.
10. Tron, G. C.; Pirali, T.; Billington, R. A.; Canonico, P. L.; Sorba, G.; Genazzani, A. A.
Med. Res. Rev. 2008, 28, 278.
11. Zhang, L.; Chen, X.; Xue, P.; Sun, H. H. Y.; Williams, I. D.; Sharpless, K. B.; Fokin,
V. V.; Jia, G. J. Am. Chem. Soc. 2005, 127, 15998.
for FLT3 (4C4, IC50 = 1.1 lM) and GSK3b (4C4, IC50 = 0.49 lM) by
12. Graneto, M. J.; Kurumbail, R. G.; Vazquez, M. L.; Shieh, H.-sheng.; Pawlitz, J. L.;
Williams, J. M.; Stallings, W. C.; Geng, L.; Naraian, A. S.; Koszyk, F. J.; Stealey, M.
A.; Xu, X. D.; Weier, R. M.; Hanson, G. J.; Mourey, R. J.; Compton, R. P.; Mnich, S.
J.; Anderson, G. D.; Monahan, J. B.; Devraj, R. J. Med. Chem. 2007, 50, 5712.
13. Pagliai, F.; Pirali, T.; Del Grosso, E.; Di Brisco, R.; Tron, G. C.; Sorba, G.;
Genazzani, A. A. J. Med. Chem. 2006, 49, 467.
14. Rasmussen, L. K.; Boren, B. C.; Fokin, V. V. Org. Lett. 2007, 9, 5337.
15. The reaction was monitored by HPLC-MS analysis (column: Unison US-C18,
4.6 mm i.d. Â 50 mm; eluents: water/methanol with 10 mM HCOOH as
gradient 90/10 to 10/90; flow rate: 2 mL/min; run time: 5 min), which was
able to separate 1,4 and 1,5-disubstitued triazoles (e.g., RT = 2.7 and 2.9 min for
5A2 and 4A2, respectively).
screening the KFEF library, and SAR obtained in this screening
was beneficial for a further lead optimization step. This study re-
veals the potential of the KFEF library to be used in fragment-based
drug design, and also provides a novel concept for design of kinase
focused compound libraries.
Acknowledgments
16. Satisfactory characteristics data were obtained for all new compounds in this
manuscript. Characteristics are given for representative compounds: 4A1,
brown powder, 12% yield, 1H NMR (400 MHz, DMSO-d6) d 9.57 (s, 1H), 8.66–
8.78 (m, 2H), 8.00 (d, J = 9.0 Hz, 2H), 7.83–7.93 (m, 2H), 7.75 (d, J = 8.8 Hz, 2H),
ESI MS 356.8 [M+H]+; 5A1, brown amorphous, 16% yield, 1H NMR (400 MHz,
DMSO-d6) d 8.58–8.67 (m, 2H), 8.35 (s, 1H), 7.62–7.71 (m, 2H), 7.46–7.57 (m,
2H), 7.25–7.35 (m, 2H), ESI MS 356.9 [M+H]+.
17. Kiyoi, H.; Ohno, R.; Ueda, R.; Saito, H.; Naoe, T. Oncogene 2002, 21, 2555.
18. Hooper, C.; Killick, R.; Lovestone, S. J. Neurochem. 2008, 104, 1433.
19. Das, J.; Chen, P.; Norris, D.; Padmanabha, R.; Lin, J.; Moquin, R. V.; Shen, Z.;
Cook, L. S.; Doweyko, A. M.; Pitt, S.; Pang, S.; Shen, D. R.; Fang, Q.; de Fex, H. F.;
McIntyre, K. W.; Shuster, D. J.; Gillooly, K. M.; Behnia, K.; Schieven, G. L.;
Wityak, J.; Barrish, J. C. J. Med. Chem. 2006, 49, 6819.
We thank Mr. Fumio Nakajima for the docking study and the
profiling team of Carna Biosciences, Inc. for the kinase profiling.
References and notes
1. Cohen, P. Eur. J. Biochem./FEBS 2001, 268, 5001.
2. Keri, G.; Orfi, L.; Eros, D.; Hegymegi-Barakonyi, B.; Szantai-Kis, C.; Horvath, Z.;
Waczek, F.; Marosfalvi, J.; Szabadkai, I.; Pato, J.; Greff, Z.; Hafenbradl, D.; Daub,
H.; Muller, G.; Klebl, B.; Ullrich, A. Curr. Signal Transduct. Ther. 2006, 1, 67.
3. Congreve, M.; Chessari, G.; Tisi, D.; Woodhead, A. J. J. Med. Chem. 2008, 51,
3661.
4. de Kloe, G. E.; Bailey, D.; Leurs, R.; de Esch, I. J. P. Drug Discovery Today 2009, 14,
630.
20. Tanaka, D.; Tsuda, Y.; Shiyama, T.; Nishimura, T.; Chiyo, N.; Tominaga, Y.;
Sawada, N.; Mimoto, T.; Kusunose, N. J. Med. Chem. 2011, 54, 851.

596
T. Irie et al. / Bioorg. Med. Chem. Lett. 22 (2012) 591–596
21. Orita, M.; Ohno, K.; Warizaya, M.; Amano, Y.; Niimi, T. Methods Enzymol. 2011,
493, 383.
22. Abad-Zapatero, C.; Metz, J. T. Drug Discovery Today 2005, 10, 464.
23. The docking study was carried out by LibDock module of Discovery Studio 2.5
package (Accelrys, Inc., San Diego, CA). First, we set a docking sphere (radius
9 Å) that included the residues interacting the original ligands in the crystal
structure. Next, ligand conformations generated automatically were docked
within the sphere using LibDock by setting threshold energy as 20 kcal/mol.
Finally, we extracted the docking poses which contain hydrogen bondings with
the hinge reside (Val 135) and depicted them in Figure 5 using Discovery
Studio Visualizer.