2 because the phenyl ring in the cis-amide would not occupy the
hydrophobic pocket but clash with the protein.
than 50%. Improvement of selectivity was reflected in an
increase of the Gini coefficient to 0.678.
Structure-guided optimization of 1 was not feasible, because
the crystal structure of FLT3 in the active form has not been
reported, so systematic modification was the only way to evolve
1 into potent and selective lead compound.11 In order to rapidly
evaluate the potential of 1 as a starting compound, we focused on
modification of the aniline ring to obtain initial SAR, because
various aniline building blocks are commercially available, in
contrast to indazole-3-carboxylic acid derivatives.
Compound 26 showed an even better selectivity profile. We
found that converting the benzyl position from sp2 carbon (24) to
sp3 carbon (26) improved not only the potency, but also the
selectivity, presumably due to the three-dimensional character of
the sp3 carbon derivative.15 Inhibition rates of TRKA and HGK
were reduced to 81% and 63%, respectively. The IC50 value for
TRKA was 137 nM (15-fold selectivity vs FLT3), and that for
HGK was 410 nM (45-fold selectivity vs FLT3). The other 47
kinases were inhibited by less than 40%. The Gini coefficient
was improved from 0.678 to 0.757. Thus, we successfully
developed 26 as a potent and selective inhibitor of FLT3 by
simple, systematic structural evolution of fragment 1.
We first set out to identify appropriate positions in the aniline
ring to attach pendant functionalities to improve potency. By
scanning with
a
methyl group, methoxy group, and
methoxycarbonyl group, we found that the 4-position in the
aniline tolerates these modifications (Table 1, compounds 5-13).
Then, we examined 4-amide derivatives, and found that tertiary
amide 15 seemed promising for further modification.
Next, we synthesized various tertiary amide derivatives (Table
2). After comparing the potency of 15 (Table 1) and 16-19 (Table
2), we focused on piperazine amide derivatives. We attached
several moieties to the lateral NH group of the piperazine, and
found that a phenylpiperazine derivative, 24, potently inhibits
FLT3 with an IC50 value of 34 nM, which is a 10-fold
improvement over 1.
Further, we changed the linker moiety between the fragment
core and the pendant phenylpiperazine group. Transformation of
the linker from sp2 carbonyl to sp3 methylene improved the
potency (Table 2, compound 26), indicating that the flexible sp3
methylene linker allows the phenylpiperazine ring to interact
more efficiently with the protein. As expected, removal of the
linker resulted in loss of potency (27).
In addition to potency, we monitored the Ligand Efficiency
(LE) of the derivatives.12 Evolution of 1 into 26 resulted in a
change of the LE from 0.50 to 0.36, which is above the preferred
minimum score of 0.3 for lead compounds.12
Selective inhibition of the targeted kinase is a major concern
in the design of kinase inhibitors, because poor kinase selectivity
may cause unwanted side effects, such as cardiotoxicity.13 Kinase
selectivity data for 1, 24, and 26 are summarized in Fig. 2 and
Table S1 in Supporting Information. The compounds were tested
against a panel of 20 tyrosine kinases and 30 serine/threonine
kinases at a concentration 50-fold higher than the IC50 value in
FLT3 inhibition assay (carried out at Carna Biosciences, Japan).
To find suitable inhibitor concentrations for selectivity study, the
FLT3 IC50 values of 1, 24, and 26 were determined at Carna
Biosciences; they were found to be 350 nM, 23 nM, and 9 nM,
respectively. Therefore, 1, 24, and 26 were used at concentrations
of 17.5 µM, 1.15 µM, and 0.450 µM, respectively, in the kinase
selectivity study.
Figure 2. Kinase selectivity profiles of 1, 24, and 26. Selectivity was
evaluated against a panel of 50 kinases at a concentration 50 times higher
than the IC50 value for FLT3 (1: 17.5 µM, 24: 1.15 µM, 26: 0.450 µM). The
y axis shows % inhibition of each kinase.
To further evaluate the potential of 26, we carried out cell-
based experiments using the FLT3-ITD-harboring human
leukemia cell line MV4-11. In growth inhibition assay, 26
potently inhibited the growth of MV4-11 (IC50 = 164 nM), but
negligibly inhibited the growth of normal human diploid lung
fibroblast cell line WI-38, which was used as a surrogate for
general toxicity (Table 3). Next, we examined the
phosphorylation-inhibitory activity of 26. In MV4-11 cells, FLT3
is reported to auto-phosphorylate Tyr-596. MV4-11 cells were
incubated with 26 for 1.5 h, then lysed, and the phosphorylation
status of FLT3 was evaluated by western blotting. As shown in
Figure 3A, 26 dose-dependently inhibited phosphorylation of
FLT3. Finally, we examined whether or not 26 induces apoptosis
of MV4-11 cells during 48 h incubation, by means of Annexin V
staining (Figure 3B). As expected, 26 induced apoptosis, as has
been reported for the representative FLT3 inhibitor quizartinib.2c,d
These results indicate that 26 is a promising lead compound.
The selectivity profile of 1 shows almost complete inhibition
of two off-target kinases, TRKA and HGK, as well as FLT3
(Figure 2). In addition, 1 potently inhibited ITK, SYK and CDK2
by more than 70%. Other kinases that were inhibited by more
than 50% were JAK3 (52%), CK1ε (51%), DYRK1B (54%), and
GSK3β (51%). Overall selectivity was evaluated in terms of the
Gini coefficient,14 and the calculated score was 0.578.
The selectivity profile of 24 is cleaner than that of 1, as shown
in Fig. 2. Although TRKA and HGK, two major off-targets of 1,
were still potently inhibited by 24 (90% and 91%), the other off-
targets were much less potently inhibited. Besides TRKA and
HGK, ITK (51%) was the only kinase that was inhibited by more
To confirm the potential of 26 for further development, the in
vitro ADME properties were examined (Table 4).16 Among five
major CYP isozymes, CYP2C19 and CYP2D6 were inhibited
about 60% by 26 at 10 µM, and the others were inhibited by less