Kinesin spindle protein (KSP) inhibitors. Part 1: The discovery of 3,5-diaryl-4,5-dihydropyrazoles as potent and selective inhibitors of the mitotic kinesin KSP

Article abstract of DOI:10.1016/j.bmcl.2005.02.055

Optimization of high-throughput screening (HTS) hits resulted in the discovery of 3,5-diaryl-4,5-dihydropyrazoles as potent and selective inhibitors of KSP. Dihydropyrazole 15 is a potent, cell-active KSP inhibitor that induces apoptosis and generates aberrant mitotic spindles in human ovarian carcinoma cells at low nanomolar concentrations. X-ray crystallographic evidence is presented which demonstrates that these inhibitors bind in an allosteric pocket of KSP distant from the nucleotide and microtubule binding sites.

Full text of DOI:10.1016/j.bmcl.2005.02.055

Bioorganic & Medicinal Chemistry Letters 15 (2005) 2041–2045
Kinesin spindle protein (KSP) inhibitors. Part 1: The discovery
of 3,5-diaryl-4,5-dihydropyrazoles as potent and selective
inhibitors of the mitotic kinesin KSP
Christopher D. Cox,a,* Michael J. Breslin,a Brenda J. Mariano,a Paul J. Coleman,a
Carolyn A. Buser,b Eileen S. Walsh,b Kelly Hamilton,b Hans E. Huber,b Nancy E. Kohl,b
Maricel Torrent,c Youwei Yan,d Laurence C. Kuod and George D. Hartmana
aDepartment of Medicinal Chemistry, Merck Research Laboratories, PO Box 4, Sumneytown Pike, West Point, PA 19486, USA
bDepartment of Cancer Research, Merck Research Laboratories, PO Box 4, Sumneytown Pike, West Point, PA 19486, USA
cDepartment of Molecular Systems, Merck Research Laboratories, PO Box 4, Sumneytown Pike, West Point, PA 19486, USA
dDepartment of Structural Biology, Merck Research Laboratories, PO Box 4, Sumneytown Pike, West Point, PA 19486, USA
Received 25 January 2005; revised 16 February 2005; accepted 17 February 2005
Available online 19 March 2005
Abstract—Optimization of high-throughput screening (HTS) hits resulted in the discovery of 3,5-diaryl-4,5-dihydropyrazoles as
potent and selective inhibitors of KSP. Dihydropyrazole 15 is a potent, cell-active KSP inhibitor that induces apoptosis and gen-
erates aberrant mitotic spindles in human ovarian carcinoma cells at low nanomolar concentrations. X-ray crystallographic evidence
is presented which demonstrates that these inhibitors bind in an allosteric pocket of KSP distant from the nucleotide and micro-
tubule binding sites.
ꢀ 2005 Elsevier Ltd. All rights reserved.
1. Introduction
We describe herein the discovery of potent and selective
3,5-diaryl-4,5-dihydropyrazole inhibitors of KSP that
were identified from optimization of high-throughput
screening (HTS) hits. Dihydropyrazole 15 is a potent,
cell-active KSPi that induces apoptosis and generates
aberrant mitotic spindles in human ovarian carcinoma
cells at low nanomolar concentrations. Additionally,
we show X-ray crystallographic evidence that these
inhibitors bind in an allosteric pocket of KSP distant
from the nucleotide and microtubule binding sites.
Small molecule inhibitors of KSP (kinesin spindle pro-
tein) represent a novel antimitotic approach for the
treatment of cancer. KSP, also known as Hs Eg5, is a
member of the kinesin superfamily of molecular motors
that utilize the energy generated from the hydrolysis
of ATP to transport vesicles, organelles, and micro-
tubules.1 Inhibition of KSP prevents normal bipolar
spindle formation, which leads to mitotic arrest with
a characteristic monoastral phenotype2,3 and subse-
quently to apoptosis in transformed cells. The promise
of designing a small molecule inhibitor of KSP that does
not suffer from the solubility, neurotoxicity, and resis-
tance profile limitations of the currently employed anti-
mitotics led to our interest in developing a KSP inhibitor
(KSPi) for the treatment of cancer.4
2. Chemistry
A medicinal chemistry effort to find small molecule
inhibitors of KSP was initiated following HTS of our
in-house sample collection. The primary screen that we
utilized to determine potency is an in vitro ATPase assay
measuring the test compoundÕs ability to prevent the
hydrolysis of ATP to ADP in the presence of microtu-
bules, thus providing a measure of enzyme inhibition.5
3,5-Diaryl-4,5-dihydropyrazoles 1 and 2 (Fig. 1) stood
out as promising hits from the initial screen. Interest-
ingly, the lack of potency in the unsubstituted analog
Keywords: KSP; Mitotic kinesins; Anti-mitotics; Dihydropyrazoles.
*
Corresponding author. Tel.: +1 215 652 2411; fax: +1 215 652
7310; e-mail: chris_cox@merck.com
0960-894X/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.02.055

2042
C. D. Cox et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2041–2045
Table 1. SAR of the western aryl group of 6a
X
R2
1 X = Cl; Y = H IC50 = 3,600 nM
2 X = H; Y = OH IC50 = 6,900 nM
3 X = Y = H IC50 > 50,000 nM
4 X = Cl; Y = OH IC50 = 450 nM
R1
Y
N
O
N
N
O
CH3
N
CH3
Figure 1. 3,5-Diaryl-4,5-dihydropyrazole inhibitors of KSP.
6
Compound
R1
R2
ATPase (nM)
6a
6b
6c
6d
6e
6f
2-F
2-Br
2-CH3
H
3600
23,300
46,200
>50,000
>50,000
9800
H
O
O
H
2-OCH3
2-CF3
3-Cl
4-Cl
2-F
H
CH3
acid or base
R1
H
R1
+
H
R2
H
6g
6h
6i
H
>50,000
12,700
37,600
10,900
>5000b
94
R2
3-F
4-Cl
6-F
5-F
5-F
2-Cl
2-F
O
6j
6k
61
3-F
2-F
R1
NH2NH2
HOAc,
N
O
N
R2
a All values reported are an average for n = 3 or greater. The standard
deviation limits are generally within 25–50% of the reported values.
b This compound did not provide good titrations, possibly due to
solubility issues; however, all results (n = 15) were >5000 nM.
CH3
5
6
Scheme 1. Synthesis of N-acetyl-3,5-diaryl-4,5-dihydropyrazoles.
genated analogs; however, a breakthrough was uncov-
ered upon dihalogenation of the western ring. Whereas
the 2,3-, 2,4-, 2,6-, and 3,5-dihalogenated analogs 6h–k
were uniformly less active than the best monohalogen-
ated analogs, the 2,5-difluoro derivative 6l, with an
IC50 = 94 nM, displayed an impressive 40-fold boost in
potency relative to 1. At this point we incorporated
the 3-hydroxyl functionality on the eastern ring and
were pleased to observe an additive effect: 7 inhibited
KSP with an IC50 = 51 nM.
3 indicated that substitution on each ring was critical for
activity. We therefore merged the substitution patterns
from the eastern ring of 1 and the western ring of 2 to
provide 4 with a substantial boost in potency, quickly
providing us with a sub-micromolar inhibitor of KSP.
Synthesis of compounds in the 3,5-diaryl-4,5-dihydro-
pyrazole series began with condensation of an aceto-
phenone and an aromatic aldehyde to provide chalcone
5 (Scheme 1). Treatment of 5 with hydrazine in the pres-
ence of acetic acid provided target dihydropyrazole 6,
most likely by a sequence involving hydrazone forma-
tion followed by intramolecular conjugate addition
and then acylation.6 The promising early result with
potent KSPi 4, in conjunction with a facile synthesis
commencing from readily available starting materials,
encouraged us to explore the structure–activity relation-
ship (SAR) of compounds in this series.
We investigated in a parallel fashion the substitution pat-
tern on the eastern aryl ring; unfortunately, this effort
proved to be less fruitful, with no modifications provid-
ing improvement over the 3-hydroxyl substituent identi-
fied earlier. For example, moving the hydroxyl group to
the 2- or 4-position (8a,b), or replacing the 3-hydroxyl
group with 3-methoxy (8c), substantially reduced activ-
ity. Other hydrogen-bond donating and accepting
groups, including those commonly thought of as phenol
replacements,7 provided analogs that were also less active
than the parent (8d–f, 9a,b). Additionally, we investi-
gated heterocyclic and alkyl replacements for the eastern
aryl group; disappointingly, no compounds inhibited
ATPase activity below 10 lM with the exception of thio-
phene analog 10, which had an IC50 = 1.8 lM.
3. In vitro SAR
Utilizing 1 as a starting point, we began the optimiza-
tion process by investigating the substitution pattern
on the western aryl ring (Table 1). Whereas exchange
of chlorine for fluorine was tolerated as demonstrated
by 6a, other modest modifications such as bromo,
methyl, methoxy, and trifluoromethyl (6b–e) resulted
in substantial losses in potency. The 3-chlorophenyl
derivative 6f was somewhat less potent relative to 1,
whereas the 4-chloro analog was inactive (6g).
A third area of the molecule readily probed by a parallel
synthesis approach was the N1-substituent of the
4,5-dihydropyrazole core. To access such structures,
chalcones 11 were treated with hydrazine in dichloro-
methane and the resulting intermediate 12 was trapped
with an electrophilic agent (Scheme 2). A variety of
N1-substituted dihydropyrazoles 13 were prepared in
this manner.8
Attempts to substitute the western aryl ring with polar
substituents, or to replace the ring with a heterocycle,
resulted in significant losses in potency relative to halo-

C. D. Cox et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2041–2045
2043
F
F
8a R = 2-OH
8b R = 4-OH
Cl
8c R = 3-OMe
8d R = 3-CN
8e R = 3-NH2
8f R = 3-NHC(O)CH3
OH
N
O
N
O
R
N
N
CH3
CH3
7 IC50 = 51 nM
H
N
F
F
Cl
9a R =
S
N
O
NH
N
N
R
N
9b R =
CH3
O
CH3
10 IC50 = 1,800 nM
O
Ar
Ar
Ar
NH2NH2
CH2Cl2
electrophile(E+)
R
R
R
N
N
N
E
N
H
11
12
13
Scheme 2. Synthesis of N1-substituted 3,5-diaryl-4,5-dihydropyrazoles 13.
The SAR of N1-substitution is presented in Table 2.
Increasing the size of the acyl group from acetyl to pro-
pionyl or isobutyryl (13a,b) led to analogs with reduced
potency relative to 4, but the original activity was re-
stored in pivaloyl analog 13c. The addition of polar
functionality to a straight-chain alkyl substituent had
a detrimental effect on potency (13d), as did benzoyl
substitution (13e). Replacement of acetyl with methyl
(3f)9 or the addition of an oxygen spacer to provide a
methyl carbamate (13g) each resulted in a loss of mea-
surable activity.
Continuing to investigate the SAR at N1, we changed to
the more potent difluoro-substituted series with 6l as a
benchmark and found that thioacetyl analog 14a pro-
vided a modest loss in potency, whereas the sulfonamide
14b and monomethyl urea 14c resulted in a more drastic
loss. The activity contained in 6l was restored in di-
methyl urea 14d.
4. Mechanistic and X-ray crystallographic studies
Table 2. SAR of the N1-acyl group in 3,5-diaryl-4,5-dihydropyrazoles
13 and 14a
We chose 7 as a tool to further our understanding of the
mechanism of KSP inhibition by compounds in the 3,5-
diaryl-4,5-dihydropyrazole series. Resolution of the
enantiomers of 7 by chiral stationary phase HPLC
revealed that the KSP activity resided only in the (S)-
antipode 15, which had an IC50 = 26 nM.10 Counter-
screening across a panel of eight structurally and func-
tionally related mitotic and transport kinesins indicates
that 15 is >2000-fold selective for KSP.11 Additionally,
inhibition of KSP by 15 is not competitive with either
ATP or microtubules, suggesting an allosteric mode of
action wherein the binding site of the inhibitor is remote
from both the nucleotide and microtubule binding sites.
Finally, we found that 15 does not affect tubulin poly-
merization in vitro at 20 lM,12 distinguishing it mecha-
nistically from the taxanes.
Cl
F
F
OH
N
N
N
E
N
E
13
14
Compound
E
ATPase (nM)
13a
13b
13c
13d
13e
13f
C(O)CH2CH3
C(O)CH(CH3)2
C(O)C(CH3)3
C(O)(CH2)3NH2
C(O)Ph
1200
1100
460
6000
13,500
>50,000
>50,000
250
Me
13g
14a
14b
14c
14d
C(O)OMe
C(S)CH3
SO2Me
1800
3100
˚
We recently described the 1.9 A resolution structure of
the ternary complex of the motor domain of KSP, 16,
and Mg-ADP.13 From this structure, we discovered
C(O)NHMe
C(O)NMe2
84
a See footnote a in Table 1.

2044
C. D. Cox et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2041–2045
moval of either aryl group, or replacement with a
OH
O
F
F
polar heterocycle, led to a reduction in potency due to
the loss of favorable hydrophobic interactions; (4) the
acyl group is close to a solvent exposed area and thus
helps account for the greater structural diversity toler-
ated in this portion of the molecule; (5) the inhibitor
binding site is well-removed from the nucleotide and
microtubule binding sites, and so explains the lack of
competition with ATP and microtubules; and (6) loop
L5 in KSP folds inward to trap the inhibitor in the in-
duced binding site in the ternary structure, whereas L5
is flexible and solvent exposed in the apo-structure.13 Se-
quence alignment of the motor domain of over 100 kine-
sins from nine different subfamilies reveals that the
primary sequence of L5 is unique to the motor domain
of KSP and helps explain the exquisite selectivity of
our inhibitors.
H
OH
N
O
HN
OCH2CH3
CH3
N
S
N
H
CH3
15 IC50 = 26 nM
16 IC50 = 10,000 nM
˚
that the inhibitor occupies an Ôinduced fitÕ pocket 12 A
removed from the nucleotide binding site that is not
present in the apo-structure, and creates widespread
structural changes throughout the protein. We have
now solved a similar structure of KSP-4-ADP at
14
˚
2.5 A.
The active site of the KSP-4-ADP complex is pictured in
Figure 2, overlaid with the KSP-16-ADP active site de-
scribed previously. 4,5-Dihydropyrazole 4 induced very
similar gross structural changes in the protein, and, as
predicted by computer-generated docking models, occu-
pied the same binding site as 16. The inhibitors share
two common features upon binding. First, both possess
a hydrogen-bonding interaction between the phenolic
hydroxyl group and the carbonyl oxygen of the back-
bone amide bond of Glu118 (2.7 A in 16, 2.5 A in 4).
Secondly, each displays a hydrophobic interaction be-
tween the aromatic ring of the phenol and the induced
nonpolar pocket. However, the main difference in the
two structures is the ability of the chlorophenyl group
of 4 to effectively fill a large hydrophobic pocket not uti-
lized by the less potent inhibitor 16. Also of note is the
fact that the N1-acyl group of 4 rests in a solvent ex-
posed area of the binding site.
5. Cell activity
A cell-based assay measuring caspase-3 activation, a
well-known marker of apoptosis, was carried out with
15 to appraise its ability to enter and kill cells.16 In good
agreement with the ATPase results disclosed above, cas-
pase-3 induction occurred in A2780 human ovarian car-
cinoma cells with an IC50 = 15 nM. Additionally, we
found that A2780 cells incubated in the presence of 15
(100 nM) displayed the characteristic monoaster pheno-
type,5 thus providing support for KSP inhibition as the
trigger of apoptosis.
˚
˚
6. Conclusion
We described a novel series of compounds based on the
3,5-diaryl-4,5-dihydropyrazole scaffold that are potent
and selective inhibitors of the mitotic kinesin KSP.
Beginning from a high-throughput screen, we were able
to rapidly identify KSP inhibitors with potency in the
low nanomolar range that possess favorable physical
and biological properties. We used these compounds
to gain functional and structural insight into the mech-
anism of inhibition. Dihydropyrazole 15 has an
IC50 = 26 nM for inhibition of KSP, a logP of 3.1,
and a molecular weight of only 316. It is believed that
further structural modifications to optimize pharmaco-
kinetics and ancillary activities will lead to clinical can-
didates for the treatment of cancer.
These structural features nicely explain the SAR and
mechanistic observations made above: (1) though race-
mic 4 was used for soaking, only the (S)-antipode bound
to the enzyme;15 (2) 3-hydroxy substitution on the east-
ern aryl ring was uniquely potency enhancing; (3) re-
Acknowledgments
We thank Ms. Yi Yang and Ms. Yun Zhang for per-
forming the cell-based analysis of 15.
References and notes
1. (a) Sharp, D. J.; Rogers, G. C.; Scholey, J. M. Nature
2000, 407, 41; (b) Mandelkow, E.; Mandelkow, E.-M.
Trends Cell Biol. 2002, 12, 585; (c) Endow, S. A.; Baker,
D. S. Annu. Rev. Physiol. 2003, 65, 161.
Figure 2. Overlay of the X-ray structures of KSP-4-ADP (yellow) and
KSP-16-ADP (pink). The H-bond from the phenolic hydroxyl group
of each inhibitor to Glu118 is highlighted.

C. D. Cox et al. / Bioorg. Med. Chem. Lett. 15 (2005) 2041–2045
2045
2. (a) Mayer, T. U.; Kapoor, T. M.; Haggarty, S. J.; King,
R. W.; Schreiber, S. L.; Mitchison, T. J. Science 1999, 286,
971; (b) Kapoor, T. M.; Mayer, T. U.; Coughlin, M. L.;
Mitchison, T. J. J. Cell Biol. 2000, 150, 975.
Kif3A, Kif1B, uKHC, nKHC, KIF14, and MCAK) were
greater than 50 lM in ATPase assays similar to that
described for KSP. The selection of these counter-screens
was based on considerations of homology and function.
12. The effect of 15 on the polymerization of tubulin was
measured by turbidity (optical density at 340 nm) in 96-
well plates at 37 ꢁC in a SpectroMax Ultra (modified from
Lopes, N. M.; Miller, H. P.; Young, N. D.; Bhuyan, B. K.
Cancer Chemother. Pharmacol. 1997, 41, 37). KSPi 15 and
control compounds (taxol and colchicine) were resus-
pended in DMSO and diluted to 20 lM in 1· buffer
(80 mM K-PIPES, 1 mM EGTA, 1 mM MgCl2, pH 6.8)
containing 1 mM GTP. The assay was initiated by adding
cold tubulin at 2.5 mg/mL to each well (to a total volume
of 200 lL; 1% final DMSO).
3. Sakowicz, R.; Finer, J. T.; Beraud, C.; Crompton, A.;
Lewis, E.; Fritsch, A.; Lee, Y.; Mak, J.; Moody, R.;
Turincio, R.; Chabala, J. C.; Gonzales, P.; Roth, S.;
Weitman, S.; Wood, K. W. Cancer Res. 2004, 64, 3276.
4. For an overview of the mitotic spindle as target for cancer
therapeutics, see: (a) Wood, K. W.; Cornwell, W. D.;
Jackson, J. R. Curr. Opin. Pharmacol. 2001, 0, 370; For
kinesin-targeted therapeutics, see: (b) Wood, K. W.;
Bergnes, G. Ann. Rep. Med. Chem. 2004, 39, 173.
5. The biological assays utilized herein have been described;
see: (a) Breslin, M. J.; Coleman, P. J.; Cox, C. D.;
Culberson, C. J.; Hartman, G. D.; Mariano, B. J.;
Torrent, M. PCT WO 03/079973 A2, 2003; (b) Ref. 13.
6. For example synthetic procedures, see: (a) Safak, C.;
Tayhan, A.; Sarac, S.; Yulug, N. J. Indian Chem. Soc.
1990, 67, 571; (b) Ref. 5a.
7. Asselin, A. A.; Humber, L. G.; Crosilla, D.; Oshiro, G.;
Wojdan, A.; Grimes, D.; Heaslip, R. J.; Rimele, T. J.;
Shaw, C.-C. J. Med. Chem. 1986, 29, 1009.
8. A similar synthetic approach has previously been taken in
a parallel solution phase library synthesis; see: Bauer, U.;
Egner, B. J.; Nilsson, I.; Berghult, M. Tetrahedron Lett.
2000, 41, 2713.
13. Yan, Y.; Sardana, V.; Xu, B.; Homnick, C.;
Halczenko, W.; Buser, C. A.; Schaber, M.; Hartman, G.
D.; Huber, H. E.; Kuo, L. C. J. Mol. Biol. 2004, 335,
547.
14. Co-crystals of the ternary complex of KSP-ADP(Mg2+)-
monastrol were first formed with the vapor diffusion
method (see Ref. 13). The KSP monastrol ternary crystals
then soaked in the harvest solution (28% PEG3350, 0.2 M
K2HPO4 at pH 8.0) containing 2 mM 4 for 2 days to
replace monastrol at the inhibitor binding site. The X-ray
diffraction data were collected at 100 Kat synchrotron
beamline 17-ID of the Advanced Photon Source at
˚
9. Dihydropyrazole 13f can be easily accessed by heating
chalcone in the presence of methylhydrazine in EtOH at
80 ꢁC for 2 h.
10. Separation of 500 mg of 7 was carried out on a 250 · 5 cm
20 lM Chiralpak AD column with 80% hexanes (modified
with 0.1% TFA) and 20% EtOH at 75 mL/min. The first
Argonne National Laboratory to 2.5 A resolution in the
˚
space group P212121 with cell dimensions of a = 69.3 A,
˚
˚
b = 79.8 A, and c = 159.6 A (Rsym = 0.066 and complete-
ness = 94%). The ternary complex structure of KSP-4-
ADP(Mg2+) was determined by the use of the difference
Fourier method and refined to an R-factor of 0.25. The
coordinates have been deposited with RCSB Protein Data
Bank under the accession code 1YRS.
isomer to elute, the (+) antipode, was ÔinactiveÕ (IC50
=
4.1 lM). The second isomer to elute, the (ꢀ) antipode, was
active (IC50 = 26 nM). Each isomer was P99% ee by
analytical HPLC under the same conditions. Data for 15:
15. We have solved the crystal structure of the ternary
complex of KSP, ADP and over 20 structurally-related
inhibitors. In each case, only the S-antipode was found to
bind in the active site. The absolute stereochemistry of 14
was assigned by analogy.
16. Caspase-3 activity in cell lysates was determined using the
ApoAlert Caspase-3 fluorescent assay kit (Clontech, Palo
Alto, CA), which measures the release of the 7-amino-4-
trifluoromethylcoumarin (AFC) fluorophore from the
substrate DEVD-AFC.
20
½aꢁD ꢀ88.9 (c 3.5, MeOH). NMR (500 MHz, DMSO-d6): d
9.4 (s, 1H), 7.7 (m, 1H), 7.4 (m, 2H), 7.1 (m, 1H), 6.7–6.5
(m, 3H), 5.4 (m, 1H), 3.9 (m, 1H), 3.1 (m, 1H), 2.3 (s, 3H)
ppm. HRMS (ES) calcd M+H for C17H14F2N2O2:
317.1096. Found 317.1105. See Ref. 15 for assignment of
absolute stereochemistry as S.
11. The IC50Õs of KSPi 15 when tested against the motor
domains of eight additional kinesins (CENP-E, MKLP-1,