Kinesin Spindle Protein (KSP) inhibitors. 9. Discovery of (2S)-4-(2,5-difluorophenyl)-N-[(3R,4S)-3-fluoro-1-methylpiperidin-4-yl] -2-(hydroxymethyl)-N-methyl-2-phenyl-2,5-dihydro-1H-pyrrole-1-carboxamide (MK-0731) for the treatment of taxane-refractory cancer

Article abstract of DOI:10.1021/jm800386y

Inhibition of kinesin spindle protein (KSP) is a novel mechanism for treatment of cancer with the potential to overcome limitations associated with currently employed cytotoxic agents. Herein, we describe a C2-hydroxymethyl dihydropyrrole KSP inhibitor (11) that circumvents hERG channel binding and poor in vivo potency, issues that limited earlier compounds from our program. However, introduction of the C2-hydroxymethyl group caused 11 to be a substrate for cellular efflux by P-glycoprotein (Pgp). Utilizing knowledge garnered from previous KSP inhibitors, we found that β-fluorination modulated the pK _a of the piperidine nitrogen and reduced Pgp efflux, but the resulting compound (14) generated a toxic metabolite in vivo. Incorporation of fluorine in a strategic, metabolically benign position by synthesis of an N-methyl-3-fluoro-4-(aminomethyl)piperidine urea led to compound 30 that has an optimal in vitro and metabolic profile. Compound 30 (MK-0731) was recently studied in a phase I clinical trial in patients with taxane-refractory solid tumors.

Full text of DOI:10.1021/jm800386y

J. Med. Chem. 2008, 51, 4239–4252
4239
Kinesin Spindle Protein (KSP) Inhibitors. 9. Discovery of (2S)-4-(2,5-Difluorophenyl)-N-[(3R,4S)-
3-fluoro-1-methylpiperidin-4-yl]-2-(hydroxymethyl)-N-methyl-2-phenyl-2,5-dihydro-1H-pyrrole-1-
carboxamide (MK-0731) for the Treatment of Taxane-Refractory Cancer
Christopher D. Cox,*,‡ Paul J. Coleman,‡ Michael J. Breslin,‡ David B. Whitman,‡ Robert M. Garbaccio,‡ Mark E. Fraley,‡
Carolyn A. Buser,§,† Eileen S. Walsh,§ Kelly Hamilton,§ Michael D. Schaber,§ Robert B. Lobell,§ Weikang Tao,§
Joseph P. Davide,§ Ronald E. Diehl,§ Marc T. Abrams,§ Vicki J. South,§ Hans E. Huber,§ Maricel Torrent,|
Thomayant Prueksaritanont, Chunze Li, Donald E. Slaughter, Elizabeth Mahan, Carmen Fernandez-Metzler, Youwei Yan,#
Lawrence C. Kuo,# Nancy E. Kohl,§,† and George D. Hartman‡
Departments of Medicinal Chemistry, Cancer Research, Molecular Systems, Drug Metabolism, and Structural Biology,
Merck Research Laboratories, P.O. Box 4, Sumneytown Pike, West Point, PennsylVania 19486
ReceiVed April 4, 2008
Inhibition of kinesin spindle protein (KSP) is a novel mechanism for treatment of cancer with the potential
to overcome limitations associated with currently employed cytotoxic agents. Herein, we describe a C2-
hydroxymethyl dihydropyrrole KSP inhibitor (11) that circumvents hERG channel binding and poor in vivo
potency, issues that limited earlier compounds from our program. However, introduction of the C2-
hydroxymethyl group caused 11 to be a substrate for cellular efflux by P-glycoprotein (Pgp). Utilizing
knowledge garnered from previous KSP inhibitors, we found that ꢀ-fluorination modulated the pKa of the
piperidine nitrogen and reduced Pgp efflux, but the resulting compound (14) generated a toxic metabolite
in vivo. Incorporation of fluorine in a strategic, metabolically benign position by synthesis of an N-methyl-
3-fluoro-4-(aminomethyl)piperidine urea led to compound 30 that has an optimal in vitro and metabolic
profile. Compound 30 (MK-0731) was recently studied in a phase I clinical trial in patients with taxane-
refractory solid tumors.
Introduction
To avoid side effects attributed to the nonselective mechanism
of action of spindle poisons, recent efforts have been focused
on identification of new targets in the mitotic pathway that play
a role only in dividing cells, or targeted antimitotic therapies.6
A leading candidate to emerge from these efforts is kinesin
spindle protein (KSPa or HsEg5), a member of the kinesin
family of molecular motor proteins. Kinesins bind to tubulin
and utilize the energy generated from the hydrolysis of ATP to
produce a directed force that can move cargo, regulate micro-
tubule dynamics, or control chromosomal attachment.7 KSP is
a mitotic kinesin that is only expressed in dividing cells and
plays a critical role in the formation of a properly organized
bipolar mitotic spindle.8 Inhibition of KSP function results in
mitotic arrest with a characteristic monoaster phenotype and
subsequently leads to apoptosis.9
Advances in molecular biology and a better understanding
of the mechanisms responsible for cancer formation and
metastasis have led to a significant shift in focus toward targeted
therapies for the treatment of neoplastic disease.1 These efforts
have resulted in the recent approval of innovative new drugs
that show promise for extending the length and improving the
quality of life for cancer patients.2 However, cytotoxic agents
have been the mainstay of effective chemotherapeutic ap-
proaches for the past several decades, and they continue to play
an important role in the battle against cancer, both as stand-
alone agents and, more recently, as adjuvants with targeted
therapies.3 It is therefore not surprising that the development
of more effective and better tolerated cytotoxic agents continues
to be an area of focus in the pharmaceutical industry.4
KSP inhibitors have several potential advantages over tradi-
tional spindle poisons as chemotherapeutic agents. Since their
mechanism of action only affects dividing cells, KSP inhibitors
are not expected to cause peripheral neuropathy, a side effect
attributed to disruption of the tubulin network in neurons and
leads to the discontinuation of taxane or Vinca alkaloid treatment
in some patients.6 Additionally, KSP inhibitors can be effective
in cancer cells that are refractory to spindle poisons because of
overexpression of P-glycoprotein (Pgp) or tubulin mutations.10
Finally, a water-soluble KSP inhibitor will circumvent the
onerous dosing regime employed for taxane administration, a
nonaqueous vehicle that requires premedication to prevent
vehicle-associated toxicities.10b These promising characteristics
have encouraged us, as well as others, to initiate clinical trials
Many successful cytotoxic agents currently in clinical use,
such as the taxanes and Vinca alkaloids, operate during mitosis
by interfering with the cellular machinery required for cell
division and diverting the cell down a pathway of programmed
cell death, or apoptosis. As defined by their mechanism of
action, these antimitotic agents are spindle poisons in that they
directly target the mitotic spindle by binding to tubulin, a protein
required not only for mitosis but also for structural integrity
and proper function of terminally differentiated cells.5
* To whom correspondence should be addressed. Phone: +1-215-652-
2411. Fax: +1-215-652-7310. E-mail: chris_cox@merck.com.
‡
Department of Medicinal Chemistry.
Department of Cancer Research.
Current address: Department of Oncology Research, Merck Research
§
†
Laboratories, 33 Avenue Louis Pasteur, Boston, MA 02115.
|
a Abbreviations: KSP, kinesin spindle protein; MDR, multidrug resistant;
Pgp, P-glycoprotein; hERG, human ether-a-go-go related gene; MTD,
maximum-tolerated dose; GI, gastrointestinal.
Department of Molecular Systems.
Department of Drug Metabolism.
Department of Structural Biology.
#
10.1021/jm800386y CCC: $40.75
2008 American Chemical Society
Published on Web 06/25/2008

4240 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14
Chart 1. Previously Reported KSP Inhibitors
Cox et al.
side chains, such as that in 1, was carried out in a straightforward
manner as previously described.13d
Results and Discussion
As discussed in the Introduction, though compound 1
possessed an optimized in vitro profile, it displayed poor activity
in an in vivo assay designed to measure mitotic arrest in a tumor
xenograft (vide infra).14 In an effort concurrent with these
studies, we discovered compounds such as 2 wherein the
solubilizing basic amine is connected to the dihydropyrrole core
though an unsymmetrical urea linkage. These compounds,
though more effective than 1 in vivo, demonstrated an unac-
ceptable in vitro profile characterized by high affinity for the
hERG channel.13b Blockade of this ion channel has been
implicated in drug-induced prolongation of the QT interval of
the EEG, an event associated with a potential fatal polymorphic
ventricular tachycardia, torsade de pointes.15
to evaluate KSP inhibitors for the treatment of taxane-refractory
cancer.4,11,12
We ultimately discovered that it was possible to overcome the
hERG channel binding in compounds related to 2 by reducing the
basicity of the amine and restoring water solubility with a phosphate
prodrug strategy.16 Separately, we found that improved in vivo
efficacy could be obtained in analogues related to 1 by replacing
the dihydropyrrole urea with a dihydropyrazole acetamide.14
However, at the time of the initial work, we reached an impasse
with these two series of compounds and thus began to search for
structural alterations that would generate a new series with the
potential to overcome the shortcomings of 1 and 2.
In a series of recent publications, we disclosed our research
efforts that began with a high-throughput screening (HTS) hit
and culminated in the design of potent and selective KSP
inhibitors 1 and 2 (Chart 1).13 Though these compounds illustrate
that synthesis of water-soluble KSP inhibitors with antimitotic
activity in a Pgp-overexpressing cell line is feasible, 1 had poor
in vivo activity in a mouse xenograft assay,14 whereas 2 and
related compounds demonstrated unacceptable levels of affinity
for the human ether-a-go-go related gene (hERG) channel.13b
Herein, we disclose a new series of C2-hydroxymethyl dihy-
dropyrrole KSP inhibitors derived by hybridization of 1 and 2.
This new series overcame the issues associated with our previous
compounds and led to a clinical candidate for the treatment of
taxane-refractory cancer.
During the synthesis of compounds in the 2,2-disubstituted
dihydropyrrole series, we recognized that incorporation of the
C2-hydroxymethyl group preserved KSP potency, as revealed
by comparison of 9 (KSP IC50 ) 78 nM) to the C2-H analogue
10 (KSP IC50 ) 38 nM) made earlier in our program (Chart
2).13b Though the potency of 9 was not in the range desired for
leading compounds, we were encouraged by the discovery that
a small, polar functionality with the potential to significantly
alter physical properties is tolerated by KSP. We therefore
decided to investigate the effect of C2-hydroxymethylation on
the more potent inhibitor 2 and were pleased to find that the
resulting compound (11) has potency similar to that of 2 in vitro,
as well as superior activity in vivo (Chart 2). Most interesting,
however, was the order of magnitude decrease in affinity for
binding to the hERG channel demonstrated by 11 relative to 2.
Extensive precedent exists for the circumvention of hERG
activity by rational medicinal chemistry design.17 Among the
most successful reported approaches for diminishing binding
to the hERG channel are modulation of lipophilicity, as
measured by log P, and discrete structural modifications that
disrupt the putative π-stacking and hydrophobic interactions
between the drug candidate and the channel cavity. It is therefore
not unexpected that installation of a polar, hydrogen-bond
donating functionality at a central position within the core
architecture of 11 results in a dramatic effect on hERG binding.
The hydroxymethyl group decreased the experimental log P
from 2.5 in 2 to 1.7 in 11, and reduced hERG binding is a
general trend observed for compounds in this series of C2-
hydroxymethyl analogues (vide infra).
Chemistry
Compounds in the 2,2-disubstituted dihydropyrrole series,
including 1, were synthesized by the route described in Scheme
1. Following formation of the benzylimine of racemic phenyl-
glycine methyl ester (3), alkylation with methallyl dichloride
under phase transfer conditions provided the mono C-alkylated
product (not shown or isolated). The benzylimine was hydro-
lyzed with aqueous HCl, and during workup, intramolecular
N-alkylation occurred to furnish pyrrolidine 4. The methyl ester
of 4 was reduced with LAH to afford an amino alcohol that
was treated with 1,1′-carbonyldiimidazole to provide oxazoli-
dinone 5. Ozonolysis of the exocyclic double bond generated
bicyclic ketone 6. This seven-step sequence can easily be carried
out on mole-scale without the need to purify any intermediates
and provides crystalline 6 by trituration from acetone in an
overall yield of 15% from phenylglycine methyl ester.
The triflate of ketone 6 was formed regioselectively with
PhN(Tf)2 and NaHMDS13b and converted to 7 via Suzuki
reaction with commercially available 2,5-difluorophenylboronic
acid in 69% yield over two steps. The oxazolidinone ring was
hydrolyzed with NaOH, and the resulting amino alcohol
protected with tert-butyldimethylsilyl chloride provided 8 in
83% yield. As noted in our previous reports, only the (S)-
antipode of compounds in this class of inhibitors exhibits KSP
activity.13 Chiral HPLC separation was therefore carried out at
this stage to provide optically pure (S)-8. Treatment with
triphosgene, followed by addition of dimethylamine and sub-
sequent deprotection, led to 9 in 78% yield from (S)-8.
Elaboration of the primary alcohol in 9 to functionalized amino
Although introduction of the hydroxymethyl group had a
favorable effect on hERG binding, it negatively impacted the
Pgp profile. For example, whereas 2 is not a Pgp substrate, 11
is efficiently effluxed from cells by Pgp as indicated by its
multidrug resistance (MDR) ratio, a measure of Pgp-mediated
resistance to mitotic arrest that we have recently reported to be
a reliable high-throughput approach to measure Pgp susceptibil-
ity of KSP inhibitors.13d The MDR ratio is calculated by dividing

KSP Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14 4241
Scheme 1a
a Reagents and conditions: (a) ClCH2C(CH2)CH2Cl, Bu4NHSO4, 10 M NaOH, DCM; (b) 1 M HCl; (c) Na2CO2/H2O; (d) LAH, THF, 0°C to room temp;
(e) CDI, TEA, DCM; (f) O3, DCM, -78 °C, then DMS (15% yield from 3); (g) NaHMDS, THF, -78 °C, then PhN(Tf)2; (h) 2,5-difluoroboronic acid,
Na2CO3, LiCl, Pd(PPh3)4, DME (69% for two steps); (i) 10 M NaOH, EtOH, 60°C; (j) TBSCl, imidazole, 5% DMF in DCM (83% for two steps); (k)
i
Chiralpak AD, 1% PrOH in hexanes (0.1% DEA as modifier), first isomer to elute is the desired (S)-isomer; (l) triphosgene, TEA, THF; (m) NHMe2; (n)
HF-TEA, CH3CN (78% for three steps).
Table 1. Effect of N-Alkyl Group on Overall Compound Profile
Chart 2. Relevant Properties of KSP Inhibitors 10 and 11
the IC50 for induction of mitotic arrest in a cell line that highly
overexpresses Pgp by the IC50 in the parental line that does not
express Pgp. An MDR ratio of unity indicates that the KSP
inhibitor is not effluxed by Pgp because it is equipotent in both
cell lines. At the other extreme, paclitaxel [(1S,2S,3R,4S,7R,9S,
10S,12R,15S)-4,12-diacetoxy-15-{[(2R,3S)-3-(benzoylamino)-
2-hydroxy-3-phenylpropanoyl]oxy}-1,9-dihydroxy-10,14,17,17-
tetramethyl-11-oxo-6-oxatetracyclo[11.3.1.0∼3,10∼.0∼4,7∼]
heptadec-13-en-2-yl benzoate] has a ratio of greater than 25 000,
since it is devoid of activity in the overexpressing cell line. We
considered compounds with a ratio of less than 10 to be
acceptable for development, but 11 had a ratio of 21, with an
EC50 of only 150 nM in the Pgp-overexpressing cell line.18
Pgp is encoded in the human MDR1 gene and has been
extensively studied as a major feature of the multidrug resistant
(MDR) phenotype. It plays an important role not only in the
resistance of cancer cells to cytotoxic agents such as paclitaxel
and the Vinca alkaloids but in other aspects of drug design and
delivery as well.19 Extensive efforts in recent years have been
directed toward understanding the mechanisms of substrate
recognition by Pgp, resulting in a number of models that attempt
to explain the SAR of Pgp substrates.20 However, a complete
understanding of the specific molecular elements required for
recognition is complicated by the unusual promiscuity of Pgp
and a lack of generalized assays that allow for direct compari-
sons between studies from different sources. Several successful
efforts to rationally design drug candidates that circumvent Pgp
efflux have been reported.21 However, the exercise remains
mainly an empirical endeavor, albeit one that can be aided by
several guiding principles such as increasing passive perme-
ability, reducing molecular size, and removing or weakening
hydrogen-bond donors and acceptors. The observation that
hydroxymethyl substitution increased Pgp-mediated efflux in
progressing from 2 to 11 is consistent with the recently reported
a Average from at least n ) 3 expermints. b Values were determined
with a Sirius GLpKa titrator, average of n ) 3 determinations. c Average
from at least n ) 2 experiments. n.t. ) not tested.
finding that hydroxyl substitution at C11 in a series of
structurally related glucocorticoids was responsible for enhanced
Pgp affinity and cellular efflux.22
In recent publications, we demonstrated that it is possible to
balance Pgp efflux potential (as measured by the MDR ratio)
with KSP potency by careful modulation of amine basicity for
two related series of KSP inhibitors.13d,14 Specifically, we found
that by constraining the pKa to a range of 6.5-8.0, maximal
KSP inhibitory activity is maintained, and the ability for the
compound to induce mitotic arrest in Pgp-overexpressing cells
is optimized. Interestingly, the pKa of 11 is 8.8, which is
consistent with its high MDR ratio. To determine if a similar
trend holds for modulating basicity of the piperidine nitrogen,
N-alkyl analogues of 11 with groups of varying electron-
withdrawing ability were prepared and evaluated (Table 1).
As we found for the previous series of KSP inhibitors, a good
correlation between optimal profile and pKa is observed: the
inhibitor with pKa < 6.5 (16) has reduced KSP potency, while
those inhibitors with pKa > 8 (11-13) lose appreciable activity
in Pgp-overexpressing cells, as indicated by high MDR ratios.
Notably, all analogues maintained favorable hERG binding
profiles. The two inhibitors with a pKa between 6.5 and 8.0,
ꢀ-monofluorinated analogue 14 and cyclopropylamine 15,
possessed the optimal balance between KSP potency and MDR
ratio and were chosen for further characterization.

4242 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14
Cox et al.
Scheme 2a
Cyclopropyl analogue 15 was found to be a potent time-
dependent inhibitor of CYP3A4, a well-known liability of this
structural motif,23 and was not further pursued. In contrast, 14
demonstrated a promising overall profile, including excellent in
vivo activity (EC90 ) 100 nM), good formulation properties, and
the potential for suitable pharmacokinetic parameters. As part of
our preclinical evaluation of 14, we studied dose proportionality
in the rat with iv bolus doses of 1, 4, and 12 mg/kg. As desired, a
linear increase in exposure with dose was observed, with 14
achieving AUC levels of 0.5, 2.5, and 9.5 µM h-1, respectively;
however, we were surprised to find mortality within 12 h postdose
in 2 of 3 rats in the 12 mg/kg group.
Acute toxicity is not an expected mechanism-based effect for
inhibition of KSP and has not been observed in our preclinical
program with any compound other than 14. The dose-limiting,
mechanism-based toxicity observed for KSP inhibitors both
preclinically and in humans is neutropenia (depletion of white
blood cells) and typically manifests in 4-10 days following
compound adminstration.11 In our search for a compound-
specific source of toxicity, we identified the fluoroethylamine
of 14 as a potential liability. If N-dealkylation of the piperidine
ring were to occur in vivo, a likely byproduct would be
fluoroacetate, a known toxin with acute pharmacology.24 In fact,
we discovered that N-dealkylated analogue 12 is the major
metabolite generated in rat liver microsome and hepatocyte
incubations, suggesting that formation of fluoroacetate is indeed
the source of toxicity in vivo.
a Reagents and conditions: (a) TMSCl, TEA, DMF, 80°C; (b) Selectfluor,
CH3CN, 0°C (90% for two steps); (c) MeNH2, AcOH, NaBH3CN, MeOH
(1.3:1 ratio of 19:20, 62% combined yield); (d)Boc2O, TEA, CH2Cl2; (e)
Chiralpak AD, 15% IPA/hexanes (45% yield of 21 and 49% of 22 from
19); (f) 1,4-cyclohexadiene, 10% Pd/C, EtOH; (g) CH2O, AcOH,
NaBH3CN, MeOH; (h) HCl(g),EtOAc, workup with Na2CO3 (>90% yield
for three steps); (i) Chiralpak AD, 20% IPA/hexanes (45% yield of 23 and
41% of 24 from 20).
Since 14 demonstrated the profile desired in a clinical
candidate, we began a search for isoelectronic replacements of
the fluoroethyl substituent that would maintain the overall
properties of 14 and not impart toxicity in vivo. The most
prudent choice was installation of fluorine directly on the
piperidine ring. This strategy eliminates the possibility of
forming a low molecular weight fluorinated acid as a metabolic
byproduct and also provides the opportunity to carefully tune
the pKa of the piperidine nitrogen by controlling relative
stereochemistry of the fluorine substituent. It has been shown
that an axial 3-fluoro substituent lowers the pKa of a piperidine
nitrogen by approximately 1 log unit, whereas an equatorial
3-fluoro can lower the pKa by nearly 2 units, a tactic that was
employed in an effort to generate 5-HT1D receptor agonists with
improved pharmacokinetic profiles.25 We thus turned our
attention to synthesis of the required stereodefined fluorinated
piperidine analogues to investigate this strategy.
Scheme 3a
a Reagents and conditions: (a) triphosgene, TEA, THF, 0°C (100%); (b)
28, TEA, DMAP, THF; (c) TEA-HF, CH3CN, 37°C (63% yield over two
steps).
Utilizing a variant of the route reported by van Niel,25 we
synthesized the four enantiomerically pure N-methyl-3-fluoro-
4-(aminomethyl)piperidine analogues as shown in Scheme 2.
Formation of the TMS enolate of the protected 4-ketopiperidine
17, followed by electrophilic fluorination with Selectfluor,
afforded R-fluoroketone 18 in 90% yield. Reductive amination
of 18 with methylamine in MeOH provided a separable mixture
of trans and cis isomers 19 and 20 in a combined yield of 62%.
The isomers were processed separately, beginning with Boc
protection of the secondary amine, followed by resolution of
the enatiomers with chiral stationary phase HPLC. Removal of
the CBz group by transfer hydrogenation, reductive amination
with formaldehyde, and Boc removal provided the four enan-
tiomerically pure N-methyl-3-fluoro-4-(aminomethyl)piperidine
analogues 25-28.
quantitative yield and is stable for months when stored in the
freezer. Reaction of 29 with fluoropiperidine 28 provided 30
following removal of the TBS protecting group. Synthesis of
the other three fluoropiperdine analogues was carried out by an
analogous route with 25-27 to provide all four optically and
diastereomerically pure isomers. The more potent cis isomer
(30) and trans isomer (31) are shown in Table 2.26
Because of the larger A-value of the amino group relative to
fluorine, the urea substituent is expected to occupy the equatorial
position of the piperidine ring, placing the fluorine axial in cis
analogue 30 and equatorial in trans analogue 31. This conjecture
was subsequently confirmed by X-ray analysis (vide infra). As
expected from precedent, the equatorial fluorine has a dramatic
effect on pKa, lowering it from 8.8 in nonfluorinated analogue 11
to 6.6 in 31 (Table 2). On the other hand, the axial fluorine results
in a more modest effect on basicity, affording a pKa of 7.6 in 30.
Compound 30 displayed superior potency in both the enzymatic
KSP assay and a cell-based assay relative to 31, and both
compounds displayed little affinity for binding to the hERG
channel. Importantly and consistent with its favorable pKa, 30 also
The four isomers were coupled to the dihydropyrrole core as
illustrated in Scheme 3 using enantiomerically pure cis-
fluoropiperidine 28 as an example. Carbamoyl chloride 29 was
first generated by treatment of (S)-8 with triphosgene and
triethylamine in THF. This intermediate was easily isolated in

KSP Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14 4243
Table 2. Effect of Fluorine Stereochemistry on Compound Profilea
a Average from at least n ) 3 experiments, unless otherwise noted. b Values were determined with a Sirius GLpKa titrator, average of n ) 3 determinations.
c Value is from n ) 1 determination.
has significant activity in Pgp-overexpressing cells, with potencies
of 4.2 and 18.9 nM in the parental and Pgp-overexpressing cell
lines, respectively, resulting in an MDR ratio of 4.5. For compari-
son, the corresponding values for paclitaxel are 0.5 nM and >10
µM. Thus, 30 has the ability to induce a mitotic block with an
IC50 of 19 nM in cells that are refractory to paclitaxel due to Pgp-
overexpression. On the basis of a favorable overall in vitro profile,
30 was selected for in vivo evaluation.
The primary in vivo assay employed during lead optimization
was a pharmacodynamic assay to determine mitotic arrest in
tumors from nude mice bearing an A2780 human ovarian
carcinoma xenograft. The minimum plasma concentration
required to induce the maximum extent of mitotic arrest in the
Figure 1. Induction of mitotic block by 30 in tumors from a mouse
xenograft assay at doses of 2.5, 5, 10, 20, and 40 (mg/kg)/day. Levels
tumor is defined as the EC90 for in vivo activity. We assessed
mitotic arrest by determining the phosphorylation state of histone
H3 on Ser-10, a residue that is phosphorylated exclusively in
mitosis by the Aurora kinases,27 and quantitated the results by
immunohistochemical staining of resected tumors. For com-
pound administration, subcutaneously implanted Alzet osmotic
pumps were used for their ability to mimic a continuous iv
infusion over the 22 h course of the experiment. We envision
employing a continuous iv infusion in the clinic because mitotic
arrest induced by KSP inhibition must be maintained for a
minimum of 16-24 h to induce the apoptotic response.28 This
protocol provides the best opportunity to maintain plasma
exposure for the required length of time without exposing the
patient to high levels of drug following a bolus infusion.
For the in vivo study, aqueous solutions of 30 were
administered to A2780-xenografted mice via minipump at doses
of 2.5, 5, 10, 20, and 40 (mg/kg)/day. As illustrated in Figure
1, 30 exhibited a dose-proportional increase in both exposure
and mitotic arrest in tumors, and the maximal level of mitotic
block (EC90) occurred at a minimum plasma level of 94 nM
from the 10 (mg/kg)/day dose. The degree of maximal mitotic
arrest induced by 30 was similar to that of a historical KSP
inhibitor that served as a positive control. Subsequent studies
in a large cohort of nonxenografted mice revealed that the
maximum-tolerated dose (MTD) of 30 when administered via
minipump is 12 mg/kg over the course of 24 h.29 Thus, 30
provides a maximal pharmacodynamic effect in tumors at its
MTD in our in vivo assay.30
of phosphohistone H3 were measured in the tumor 22 h postdose and
plotted relative to vehicle and a known KSPi positive control. Steady-
state plasma concentrations were determined at the time of sacrifice
by cardiac puncture and also showed a dose dependent increase.
Table 3. Correlation between Cell Potency, Serum Shift, and in Vivo
Efficacya
KSP IC50
(nM)
cell EC50
(nM)
KSP + serum
in vivo
EC90 (nM)
compd
IC50 (nM)b
1
5.2 × 0.8
6.2 ( 2.9
5.0 ( 1.4
2.2 ( 1.2
22.5 ( 5.6
5.6 ( 1.3
4.8 ( 0.8
5.3 ( 2.3
84.2 ( 6.0c
29.1 ( 6.5
38.8 ( 19
1425
125
100
94
11
14
30
29.1 ( 12.2c
a Average from at least n ) 3 experiments, unless otherwise noted. b KSP
assay run in the presence of 30% mouse serum. c Value is from n ) 2
determinations.
increased potency in both cell-based and serum-shifted bio-
chemical KSP assays relative to 1. As illustrated in Table 3,
key compounds in the C2-hydroxymethyl series are approxi-
mately 4-fold more potent in the cell-based assay and 2- to
3-fold more potent in the KSP assay in the presence of 30%
mouse serum. A similar explanation of improved in vivo potency
holds across other series of KSP inhibitors recently reported
from our group.31
On the basis of the ability of 30 to induce mitotic arrest in
tumors in vivo, we were encouraged to test its antitumor efficacy
in xenograft models. In the clinic, cytotoxic chemotherapeutic
agents are typically administered at their MTD on schedules
designed to accommodate the induction and recovery from
mechanism-based hematopoietic and gastrointestinal (GI)
toxicities.4a Typical regimes are once every 3 weeks, once
weekly for 3 weeks, or once daily for 5 consecutive days. Thus,
From a retrospective analysis, we believe that the superior
in vivo activity of compounds in the C2-hydroxymethyl series,
such as 11, 14, and 30, is reflected by the combination of

4244 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14
Cox et al.
Figure 2. Inhibition of A2780 tumor growth in mice. Both paclitaxel
and 30 were dosed at their MTD on the schedule indicated. The data
are plotted as the change in tumor volume relative to its starting volume,
versus days following initial dose.
Figure 3. Inhibition of growth in PTX10 tumors that are paclitaxel-
resistant because of a tubulin mutation. Both paclitaxel and 30 were
dosed at their MTD on the schedule indicated. The data are plotted as
the change in tumor volume relative to its starting volume, versus days
following initial dose.
we wanted to test dosing protocols in our preclinical model that
mimic these likely clinical schedules.
First, studies were undertaken in naive mice to determine the
MTD of 30 in protocols that varied the dose, vehicle, and
frequency of administration, monitoring survival as an end point.
In all cases of lethality following administration of 30, death is
observed at least 6 days after the start of dosing, suggesting
that mechanism-based toxicity in hematopoietic and/or GI tissues
is the causative factor. To provide support for this hypothesis,
peripheral blood counts were determined during these studies
to assess the myelosuppressive activity of 30. Of the different
blood cells, neutrophils are most strikingly and consistently
depleted, with a nadir in counts on day 4 postdose and recovery
to greater than predose levels by day 11 in surviving animals.
The severity of neutrophil depletion is directly related to the
dose of 30 and suggests that neutropenia may be the dose-
limiting toxicity in mice.
Over the course of a large number of experiments, we found
that the schedule for administration of 30 that consistently
provided the best efficacy is a q4d × 3 regime, where the MTD
of 30 is 100 mpk in pH 4 saline given subcutaneously.32 As
indicated in Figure 2, 30 produced statistically significant
inhibition of A2780 tumor growth under these conditions. In
this experiment, the efficacy of 30 appears statistically superior
to that of paclitaxel when given on its optimal schedule of qd
× 5 at 20 mpk ip.
Figure 4. Inhibition of growth in KB-v tumors that are paclitaxel-
resistant because of Pgp overexpression. Both paclitaxel and 30 were
dosed at their MTD on the schedule indicated. The data are plotted as
the change in tumor volume relative to its starting volume, versus days
following initial dose.
inhibits the growth of KB-v tumors that highly overexpress Pgp,
whereas paclitaxel has no effect (Figure 4).34 Thus, 30 provides
significant antitumor efficacy at its MTD in a mouse xenograft
assay and is efficacious in tumors that are resistant to paclitaxel
because of either tubulin mutation or Pgp overexpression.
Studies were carried out with 30 to confirm that no undesired
changes in overall profile or mechanism of action occurred
during our lead optimization process. First, we determined that
inhibition of KSP by 30 is not competitive with either ATP or
microtubules. We also found that 30 is >20000-fold selective
for KSP over a panel of eight structurally and functionally
related kinesins (CENP-E, MKLP-1, Kif3A, Kif1B, uKHC,
nKHC, KIF14, and MCAK; IC50 > 50 µM in every case). These
mechanistic observations suggest that 30 binds in the induced-
fit allosteric site previously defined by our group.35 This mode
of binding nicely explains not only the lack of competition with
microtubules and ATP but also the selectivity of 30 for KSP
over other kinesins, since the binding site is adjacent to the L5
loop that is unique to the motor domain of KSP. Binding in
this allosteric site has subsequently been confirmed for a number
of structurally diverse KSP inhibitors, though several recent
To further our understanding of the potential for 30 to treat
refractory tumors, we developed xenograft models utilizing
paclitaxel-resistant cell lines. For this purpose, we choose the
PTX10 cell line that carries a tubulin mutation at the taxane
binding site,33 as well as the Pgp-overexpressing KB-v cell line
utilized in our assay to determine MDR ratio. In order to provide
a direct comparison of the in vivo efficacy of 30 and paclitaxel
from the same study, we designed the assay as a dual flank
xenograft model, wherein animals were implanted on contralat-
eral flanks with both a taxane-sensitive cell line (A2780 or KB-
3-1) and a taxane-resistant cell line (PTX10 or KB-v).
As pictured in Figure 3, 30 maintains excellent efficacy
against the paclitaxel-resistant PTX10 cell line that carries a
tubulin mutation at the taxane binding site. In contrast, paclitaxel
treatment produces no growth inhibition relative to vehicle in
these tumors. Notably, the tumor growth inhibition pictured in
Figures 2 and 3 was obtained from the same animals in dual
flank mode. In a similar vein, we were able to show that 30

KSP Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14 4245
is >50 mg/mL at pH 4.3, and if kept below room temperature,
it has sufficient stability under these conditions to provide an
acceptable clinical formulation. The metabolism and pharma-
cokinetic properties of 30 were studied in three preclinical
species, and as illustrated in Table 4, moderate-to-high plasma
clearance with a half-life of 1-2 h was observed. The metabolic
turnover of 30 is more rapid in animal liver microsomes and
hepatocytes than in human, and all metabolites detected in
human preparations are found in preclinical species. On the basis
of a metabolic scaling method, 30 is predicted to be a low-to-
moderate clearance compound in humans (1.4-7 (mL/min)/
kg) with a half-life of 2-10 h. These parameters are supportive
of continuous iv infusion in man.
The overall safety profile of 30 is also acceptable for an
antimitotic agent. Unlike 14, no acute toxicity is observed in
the rat with an iv bolus dose of 15 mg/kg (9.6 µM h-1), and 30
demonstrates >3000-fold selectivity for KSP over a panel of
160 enzymes, receptors, and transporters in a Panlabs screen at
MDS Pharma. Gratifyingly, no QT interval prolongation was
seen in preclinical cardiovascular safety studies, consistent with
the optimized hERG profile achieved during the discovery effort.
On the basis of its favorable efficacy, pharmacokinetic, and
safety profile, 30 (MK-0731) was accepted for development and
advanced into clinical trials.38
In a phase I open-label clinical trial to test safety and
pharmacokinetic parameters in patients with taxane-refractory
tumors, 30 was well-tolerated and exhibited dose-proportional
exposure following a 24 h continuous infusion. As predicted,
30 is a low clearance compound (100-250 mL/min on average)
with a half-life of 4-10 h in humans. At the MTD of 17 (mg/
m2)/day every 21 days, 30 was well-tolerated with the antici-
pated dose-limiting toxicity of neutropenia.38
Figure 5. X-ray structure of the active site of the KSP-30-ADP
ternary complex.
reports have identified ATP-competitive inhibitors that bind in
the nucleotide binding site.36
The X-ray structure of the KSP-30-ADP ternary complex
was determined to 2.3 Å resolution and confirms that 30 does
indeed bind in the allosteric site of KSP (Figure 5).37 Also of
note is the conformation of the piperidine ring, wherein the
fluorine is axial and the urea substituent occupies the equatorial
position, as expected.
Three general features present in the KSP-30-ADP ternary
complex have been identified in progenitor compounds:13 (1)
the difluorophenyl moiety of 30 occupies a large, hydrophobic
area of the allosteric binding site (eastern pocket) made up of
lipophilic residues L214, A218, F239, and I136; (2) the
unsubstituted phenyl ring of 30 sits orthogonal to the core in a
second, smaller region (western pocket) sandwiched between
residue P137 (back) and the alkyl side chain of R119 (front);
(3) the polar N-methyl-3-fluoro-4-(aminomethyl)piperidine moi-
ety of 30 extends toward the solvent-exposed region located
above the W127/Y211 π-stacking interaction that seals the
protein cavity. The unique and most interesting interaction is
the 2.5 Å hydrogen bond formed between the C2-hydroxymethyl
group and the backbone carbonyl oxygen of residue G117
located at the roof of the cavity. The G117 carbonyl is the same
functionality responsible for hydrogen-bonding to the side chain
propylamino group of structures such as 1.13c Few polar regions
are present in the allosteric binding site of KSP, and the
hydroxymethyl group in 30 provides an optimal way to utilize
the polarity offered by the G117 carbonyl without compromising
any of the favorable hydrophobic interactions present in the
previously investigated structural series.
Compound 30 also behaves as expected in several assays that,
collectively, strongly suggest that KSP inhibition is the mech-
anism of mitotic arrest and antitumor efficacy. Unlike taxanes
and the epothilones, 30 has no effect on microtubule polymer-
ization in vitro at 20 µM. Additionally, a cell-based assay
measuring caspase-3 activation indicates that 30 induces apo-
ptosis in A2780 cells with an EC50 of 2.7 nM. Finally, monoaster
formation, the phenotype characteristic of KSP inhibition, occurs
in ∼90% of A2780 cells following a 16 h incubation with 50
nM of 30 (Figure 6).
Conclusion
We demonstrated herein that installation of a hydroxymethyl
group at C2 of the dihydropyrrole core provided 11, which had
an optimized hERG profile relative to its progenitor, 2. However,
this change also engendered Pgp susceptibility. By carefully
modulating the pKa of the basic amine in 11, we discovered
fluoroethylated KSPi 14, which had an optimal in vitro profile
but was acutely toxic in vivo. Placement of the fluorine in a
strategic, metabolically benign location on the piperidine ring
allowed us to simultaneously avoid toxic metabolite formation
and tune the pKa of the nitrogen to provide 30 with an optimized
in vitro and in vivo profile. In mouse xenograft assays, 30
induced dose-dependent mitotic arrest in tumors and inhibited
tumor growth comparable to paclitaxel. Compound 30 also
inhibited tumor growth in cell lines that are resistant to paclitaxel
because of either tubulin mutations or Pgp-overexpression. The
overall safety and pharmacokinetic profile supports the use of
30 for the treatment of cancer via continuous iv infusion. In
clinical trials, 30 exhibits low systemic clearance and is well-
tolerated at its MTD, resulting in the expected dose-limiting
toxicity of neutropenia.
Experimental Section
(A) Biology. All animal studies were performed according to
the NIH Guide for the Care and Use of Laboratory Animals, and
experimental protocols were reviewed by the Merck Animal Care
and Use Committee.
KSP ATPase Assay. The primary assay for KSP activity is based
on a colorimetric determination of inorganic phosphate released
from ATP. The KSP motor domain (1-367) was expressed in
bacteria with a His6 tag for purification. Microtubules are polym-
erized from tubulin at 37 °C, in the presence of 1 mM GTP and 10
Importantly, the physical properties and pharmacokinetic
profile of 30 are consistent with that desired for a development
compound. The free-base of 30 is a crystalline solid that is stable
for 4 weeks at 80 °C open to air. The aqueous solubility of 30

4246 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14
Cox et al.
Figure 6. Immunofluorescence pictures of A2780 ovarian carcinoma cells treated with vehicle (left) or 50 nM of 30 (right) for 16 h (blue ) DNA,
green ) microtubules). In the vehicle control image, cells in interphase, metaphase, and anaphase/telophase are visible.
Table 4. Pharmacokinetics of 30 in Preclinical Speciesa
was varied, microtubule concentrations were fixed at 1 µM. When
microtubules were varied, ATP concentration was fixed at 1 mM.
species
iv dose
Cl ((mL/min)/kg)
Vdss (L/kg)
T1/2 (h)
The enzyme concentration tested was 0.5 nM. Reactions for each
condition were initiated in a total volume of 700 µL. Eighty (80)
µL aliquots were quenched by addition of an equal volume of a
solution containing 50 mM EDTA and 1.8 M KCl at 4, 8, 14, 22,
30, 40, 55, and 70 min and held on ice. Phosphate was detected by
addition of 240 µL of the quinaldine red reagent and incubated for
25 min before reading the absorbance at 540 nm. The data were
first analyzed as a function of time to derive linear rates for each
reaction condition. The velocity data were then analyzed with
respect to the Michaelis-Menten or MM fit equation describing
mixed inhibition using a nonlinear least-squares method. The data
were also analyzed using an alternative model, which incorporates
a correction for the reduction in compound concentration in the
assay by the enzyme, as a function of enzyme concentration. This
model gave essentially identical results to the Michaelis-Menten
fit, indicating compound 30 is not competitive with either ATP or
microtubules.
Cell-Based Assay. Mitotic arrest was measured by assessing the
mitosis-specific phosphorylation of nucleolin using an antibody-
coated, bead-based assay. In this assay, total nucleolin is captured
on a streptavidin-coated paramagnetic bead coupled with biotiny-
lated nucleolin monoclonal IgG1 antibody 4E2 (Research Diag-
nostics, Inc.). Nucleolin phosphorylation is detected by an antibody
complex consisting of a phospho-specific nucleolin IgM monoclonal
antibody, TG3 (Applied NeuroSolutions, Inc.) and a goat antimouse
IgM labeled with a ruthenium Tris-bipyridyl complex (BV-TAG
Technology, BioVeris Corp.). A2780 cells were seeded at 10 000
cells/well in 96-well plates and grown overnight. Cells were treated
in triplicate with vehicle (DMSO) or test compound for 16 h. The
final concentration of compound in the assay ranged from 300 to
0.3 nM in a 7-point, half-log dilution series, and the final DMSO
concentration was 0.1% v/v. Cells were incubated with compound
for 16 h, and then plates of cells were centrifuged at 1500 rpm for
5 min to collect floating cells, followed by lysis via shaking at 4
°C for 20 min in RIPA buffer (50 µL/well) that contained a
complete set of protease inhibitors, 1 µL/mL DNase I, 1.0 mM
sodium orthovanadate, and 1.0 µM microcystin. The following
procedures were all done at room temperature. Streptavidin-coated
paramagnetic Dynabeads (10 µg/well) were premixed with bioti-
nylated 4E2 antinucleolin antibody (50 ng/well) in the assay buffer
(pH 7.3 phosphate buffered saline, 1% BSA, and 0.5% Tween-
20). The cell lysates (25 µL/well) were subsequently incubated with
the premixed beads/4E2 antibody (25 µL/well) in 96-well plates
by shaking for 30 min. The antiphosphonucleolin TG3 antibody
was then added (50 ng/well), followed by shaking for another 30
min. Finally, a ruthenylated goat antimouse antibody (prepared
immediately before use) was added (6 ng/well) and mixed by
shaking plates for 2.5 h. The sample plates were read on an IGEN
M-series M8 analyzer.
rat
dog
rhesus
1
0.4
0.4
66.7 ( 12
15.1 ( 6.4
23.1 ( 6.4
3.0 ( 0.6
1.6 ( 0.8
2.3 ( 0.4
1 ( 0.1
2 ( 0.2
1 ( 0.3
a Values are mean ( SD, n ) 3.
µM paclitaxel, and are then separated from unpolymerized tubulin
by centrifugation. Assays were performed in a 96-well plate, with
each well containing 4 µL of a DMSO solution of test compound.
Compounds were tested as an 11-point titration series, with 3-fold
dilutions between each step, ranging from a concentration of 2 µM
(in assay) to 34 pM. The reaction was started by adding 156 µL of
the reaction buffer to each well; final reaction conditions were 80
mM K-HEPES (pH 6.8), 40 mM KCl, 1 mM EGTA, 0.01% BSA,
1 mM DTT, 2 mM MgCl2, 1 mM ATP, 0.5 µM tubulin (as
polymerized microtubules), 2.5% DMSO, and 0.25 nM KSP. After
incubation at room temperature for 3 h, reactions were quenched
by addition of 240 µL/well of a solution containing 65 µg/mL
quinaldine red, 0.093% polyvinyl alcohol, and 4.1 mM ammonium
molybdate in 0.38 M sulfuric acid. After 5 min for color develop-
ment, plates were read for optical absorbance at 540 nm. A serum-
shifted assay was also run on selected compounds and was
performed similarly. Serum was prepared for the assay by extensive
dialysis against the primary reaction buffer, using dialysis mem-
branes with a 12-14 kDa cutoff.
KSP ATPase Selectivity. Other kinesins assayed were isolated
motor domains, cloned from the corresponding human sequences,
and expressed and purified similarly to the procedure used for KSP
(CENP-E:1-340, uKHC:1-337, nKHC:1-340, Kif1B:1-350,
Kif3A:1-350, MKLP-1:1-435, Kif14:342-720). MCAK motor
domain was purchased from Cytoskeleton. Assays were run
similarly to the KSP assay described above except that 1 µL of 30
in DMSO was placed in a well to which 39 µL of reaction buffer
was added. The reaction buffer used here was identical to that
described above except that the tubulin concentration was 1 µM.
After 50 min of incubation at room temperature, reactions were
quenched by the addition of 40 µL of a solution containing 50 mM
EDTA and 1.8 M KCl. Reactions were visualized by the addition
of 120 µL of the quinaldine reagent described above. Compound
30 was tested at 50 and 5 µM. Enzyme concentrations varied
depending on the enzyme tested: CENP-E, 5 nM; uKHC, 1.5 nM;
nKHC, 10 nM; Kif1B, 5 nM; Kif3A, 2 nM; MKLP1, 50 nM; Kif14,
50 nM; MCAK, 50 nM. Compound 30 displayed an IC50 > 50 µM
against each kinesin tested.
Kinetic Competition Studies. Assays used a method similar to
that described for the primary KSP ATPase assay, with the
exception that both ATP and microtubule concentrations varied.
Compound 30 was tested at eight different concentrations, including
a no compound control, from 0 to 12.5 nM. Either ATP or
microtubules were varied independently from the compound over
12 different concentrations; ATP was varied from 10 to 400 µM,
while microtubules were varied from 100 to 4000 nM. When ATP
Tubulin Polymerization Assay. Tubulin polymerization was
monitored using the light-scattering properties of polymerized

KSP Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14 4247
microtubules. The assay was carried out in a 96 well plate, using
tubulin at a final concentration of 2.5 mg/mL, with 0.5 mM GTP,
in a buffer of 80 mM K-PIPES, pH 6.8, 1 mM MgCl2, 1 mM
EGTA. Compound 30 in DMSO (2 µL, final concentration of 20
µM) was added to 198 µL of reaction buffer, and the mixture was
incubated at 37 °C. The optical density was measured at 320 and
340 nm every 30 s for 75 min. Assay plates contained controls for
microtubule stabilization (paclitaxel, 20 µM) and inhibition of
polymerization (colchicine, 20 µM).
were evaluated for their ability to induce mitotic arrest in KB-3-1
and KB-V-1 cells by the phospho-nucleolin/mitotic arrest assay
essentially as described above except that the final concentration
of compound in the assay ranged from 10 000 to 0.1 nM in an
11-point, half-log dilution series. The MDR ratio is calculated by
dividing the IC50 in KB-V-1 cells by the IC50 in KB-3-1 cells.
Coadministration of verapamil, a competitive inhibitor of Pgp,
restores the activity of paclitaxel and KSP inhibitors in the KB-
V-1 cell line to nearly that observed in the parental KB-3-1 line,
confirming that Pgp efflux is responsible for the observed resistance
to drug-mediated mitotic arrest.
Monoaster Formation. The formation of monoastral spindles
was detected by immunofluorescence microscopy in cultured A2780
cells. Cells were grown in microscopy slide chambers, exposed to
30 (5 or 50 nM) or vehicle (0.1% DMSO final concentration) for
16 h, and then permeabilized and fixed with 100 mM Pipes buffer
(pH 6.8) containing 10 mM EGTA, 1 mM MgCl2, 0.2% Triton
X-100, and 4% formaldehyde, for 10 min at room temperature (RT).
The fixed cells were treated with 2% bovine serum albumin (BSA)
in TBST (Tris-buffered saline containing 0.1% Tween-20) for 1 h
to block nonspecific antibody binding. Subsequently, cells were
incubated with an anti-R-tubulin monoclonal antibody (clone DM1,
Sigma) (1:500) for 12 h at 4 °C, followed by an incubation with a
FITC-conjugated donkey antimouse IgG (15 µg/mL) for 1 h at RT
in the dark. Cells were then treated with the DNA stain Hoechst
33342 (10 µg/mL) for 10 min, and slides were mounted with an
antifade kit. Between staining steps, cells were washed 3 times with
TBST. Mitotic spindles were visualized by a fluorescence microscope.
Caspase-3 Assay. A2780 cells were seeded at 20 000 cells /well
in 96-well plates and grown overnight. Cells were treated in
triplicate with vehicle (DMSO) or 30 for 48 h. The final concentra-
tion of compound in the assay ranged from 300 to 0.415 nM in a
7-point, 3-fold dilution series, and the final DMSO concentration
was 0.1% v/v. Cells were lysed and the capase-3 activity in cell
lysates was determined using the ApoAlert caspase fluorescent assay
kit according to the manufacturer’s protocol. For compound 30,
EC50 ) 2.7 ( 0.4 nM (n ) 6).
hERG Binding Assay. The binding assay was performed on
membrane preparations from human embryonic kidney (HEK) cells
constitutively expressing hERG. The membranes were prepared by
homogenization of the HEK cells in Tris-EDTA buffer (50 mM
Tris, 1 mM EDTA, pH 7.5) and centrifugation at 45000g for 20
min at 4 °C. The pellet containing membrane was resuspended in
Tris-EDTA and stored at -70 °C. In the binding assay, hERG
membranes were diluted into assay buffer (60 mM KCl, 71.5 mM
NaCl, 1 mM CaCl2, 10 mM Hepes, pH 7.4) to a final concentration
of 4.16 µg per 400 µL well volume. Radiolabeled ligand (35S-MK-
499, specific activity of ∼1000 Ci/mmol) was then added to a final
concentration of 50 pM, and 400 µL of membrane-ligand mixture
was added per well to 96-well assay blocks containing 4 µL of
100× stocks of tested drugs or 100% DMSO (maximum binding
control) or 100 µM cold MK-499 (nonspecific binding control).
Membranes were incubated at room temperature for 75 min, filtered
through GF/B Unifilters presoaked in 0.1% BSA, and washed 5 ×
500 µL with wash buffer at 4 °C (10 mM Hepes, 131.5 mM NaCl,
1 mM CaCl2, 2 mM MgCl2, pH 7.4). Filters were dried at room
temperature, 50 µL Microscint-20 was added to each well, and
Unifilters were counted for 1 min. Dose-inhibition curves and
inflection points were determined by curve-fitting the equation
response ) ((max - min)/(1 + ([I]/IP)s) + min, where I is the
concentration of inhibitor, max is the maximum response (DMSO
control), min is the minimum response (excess, cold ligand control),
IP is the inflection point, and s is the slope. Generally, the inflection
point is similar to the concentration of drug that inhibits 50% of
radiotracer binding. Compounds were tested as an 8-point titration
series, with 4-fold dilutions between each step, ranging from a
concentration of 100 µM (in assay) to 6 nM.
In Vivo Mitotic Arrest Assay. Subcutaneous (sc) xenograft
tumors were formed by injection of the human A2780 ovarian
carcinoma cell line in the hind flank of 8-12 week-old female
immunodeficient nude mice (Crl:NU/NU-nuBR strain). Typically,
animal weight ranged from 25 to 30 g. To facilitate tumor formation,
cells were injected in media containing Matrigel, a solubilized
basement membrane preparation extracted from the EHS mouse
sarcoma. A2780 cells were grown in cell factories in RPMI medium
+ 10% serum, harvested by trypsinization and centrifugation, and
resuspended in a 60:40 mixture of ice cold Matrigel/serum-free
RPMI. Mice were lightly anesthetized via CO2 inhalation, and 0.5
mL of cells (9 million) was injected sc into their hind flanks. On
day 14 after tumor cell implantation, tumors were calipered and
mice whose tumor volumes ranged typically from 300-900 mm3
were selected for study. Mice were randomized according to tumor
volumes and distributed into treatment groups of 6 mice with
approximately equivalent ranges of tumor volumes between treat-
ment groups. On day 15 after tumor implantation, mice were
implanted with Alzet 2001D osmotic pumps containing vehicle,
positive control (compound 19 from ref 13b), or varying concentra-
tions of 30 solubilized in water acidified to pH 4.0 with HCl. The
dose groups (7 mice per group) of 30 were 40 (mg/kg)/day (5.78
mg/mL), 20 (mg/kg)/day (2.89 mg/mL), 10 (mg/kg)/day (1.45 mg/
mL), 5 (mg/kg)/day (0.72 mg/mL), and 2.5 (mg/kg)/day (0.36 mg/
mL). Pumps were primed for 3 h at 37 °C in 0.9% saline bath
prior to implantation. For pump implantation, mice were anesthe-
tized by intraperitoneal (ip) injection of ∼0.1 mL of 7.7 mg/mL
ketamine/0.46 mg/mL xylazine/0.9% saline mixture per 10 g body
weight (typically, ∼25-28 g). Implantation involved cranial-caudal
incision on the right dorso-lateral flank, and incisions were sutured
with surgical staples. Mice were euthanized 22 h after pump
implantation by CO2 inhalation, tumors were excised for PD
analysis (see below for method), and blood was removed by cardiac
puncture using heparinized 20-gauge needles. Plasma was prepared
from heparinized whole blood by centrifugation at 3000 rpm for
15 min at 4 °C and stored frozen prior to analysis for levels of 30.
Tumors were excised during necropsy and fixed for 24-48 h in
10% formalin (saline buffered). Tissues were paraffin-embedded,
thin-sectioned (5 µm), and subject to antigen retrieval by heating
(8 min, 20% microwave power) in target retrieval solution. Sections
were immunostained with 1:200 diluted rabbit polyclonal anti-
phosphohistone H3-Ser10 antibody followed by incubation with
biotinylated goat antirabbit IgG [H + L] and then avidin-HRP.
Signal was detected by development with HistoMark Black HRP
substrate, followed by counterstaining with light-green counterstain.
Immunostained tumor sections were quantitated by automated image
acquisition and analysis. Images were captured using an inverted
microscope fitted with a motorized stage and a digital camera. The
percent of tumor area positively immunostained with phosphohi-
stone H3-Ser10 was quantitated from the acquired images using
Image Pro Plus software equipped with a custom macro. Necrotic
regions of the tumor and regions with sporadic cellularity consisting
primarily of Matrigel were excluded from the analysis. To eliminate
subjectivity, data analysis using the Image Pro software was
performed in a blinded fashion by two scientists. For each analyst’s
data set, percent phosphohistone H3 stained tumor area in each
drug treated tumor was normalized to the mean percent phospho-
histone H3 stained tumor area in the vehicle control group to
determine the fold induction in mitotic arrest (phosphohistone H3
staining) for each drug treated tumor. Fold-induction values for
MDR Ratio. The human KB-3-1 tumor cell line and a multidrug-
resistant derivative that overexpresses MDR1 (KB-V-1) were used
to evaluate the MDR ratio. The KB-V-1 cell line was originally
derived by culturing KB-3-1 cells in vinblastine and displays cross-
resistance to multiple cancer drugs.39 Relative to KB-3-1, KB-V-1
expresses >500-fold higher levels of MDR1 mRNA. Compounds

4248 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14
Cox et al.
each tumor from the two analysts were averaged, and the mean (
SEM for each drug treatment group was calculated. The minimum
plasma concentration required to induce the maximum amount of
mitotic arrest in the tumor is defined as the EC90 for in vivo activity.
Tumor Growth Inhibition Assay. The mice for the dual flank
xenograft were generated as described above for nude mouse
xenograft formation but were raised by subcutaneous injection of
either 9 million A2780 and 6 million PTX10 cells (method A) or
2.5 million KB-3-1 and 9 million KB-v-1 cells (method B) on
contralateral flanks. On day 13-14 after tumor cell implantation,
tumors were calipered and mice whose tumor volumes ranged
typically from 250 to 500 mm3 were selected. Mice were random-
ized and distributed into treatment groups with approximately
equivalent ranges of tumor volumes. For sc injection, 30 was
formulated as a solution in acidified saline at pH 4.0, 100 µL was
administered for a dose of 100 mpk on the q4d × 3 schedule
(A2780/PTX10 experiment) or as a suspension in 0.5% w/v
methylcellulose, and 100 µL was administered for a dose of 40
mpk on the qd × 1 schedule (KB-3-1/KB-v experiment). Paclitaxel
was solubilized in a 1:1 (v/v) mixture of ethanol and Cremaphor-
EL and then diluted in saline just prior to injection to make a
working stock solution with final vehicle composition being ethanol/
cremaphor/saline (10:10:80). Then 200 µL of this working solution
was injected ip in mice daily for 5 consecutive days (qd × 5),
delivering a dose of 20 mg/kg per day.
For efficacy evaluation, groups of 10-12 mice were dosed as
indicated and their tumors were manually calipered on multiple
days over the course of the study. Tumor volumes were evaluated
for at least 10-11 days from the start of dosing. Tumor volumes
were calculated according to the formula for volume of a prolate
ellipsoid, (W2 × L)/2, where W (width) and L (length) are in
millimeters and L g W. The mean tumor volume change ∆V, for
treatment (T) and control (C) groups was calculated by subtracting
the initial tumor volume (Ti), obtained either the day prior to start
of treatment (d - 1) or on the same day as the start of treatment
but prior to administration of compound, from the tumor volume
on the day of testing, 3-14 days after start of dosing. Percent
change in tumor volume in the treatment group relative to control
group (% T/C) was calculated as follows:
(method B). Both methods provided purities as measured by UV
detection at 215 nm. Proton NMR spectra were recorded on a Varian
INOVA 500 MHz spectrometer, and all chemical shifts are
referenced to an internal standard of tetramethylsilane. High
resolving power accurate mass measurement electrospray (ES) and
atmospheric pressure chemical ionization (APCI) mass spectral data
were acquired by use of a Bruker Daltonics 7T Fourier transform
ion cyclotron resonance mass spectrometer (FT-ICR MS). Optical
rotations were determined at 23 °C on a Perkin-Elmer 241 spec-
trometer in the solvents and concentrations specified.
Methyl 4-Methylene-2-phenylpyrrolidine-2-carboxylate (4).
To a suspension of phenylglycine methyl ester hydrochloride (1.0
kg, 4.96 mol), anhydrous Na2SO4 (1.2 kg, 8.5 mol), and benzal-
dehyde (520 g, 4.91 mol) in 7.0 L of CH2Cl2 in a 20 L jacketed
reactor was added triethylamine (518 g, 5.13 mol). After the mixture
was stirred for 6 days, the solids were filtered off and rinsed with
an additional 4.7 L of CH2Cl2. The filtrates were combined and
used as is for the following reaction. To the solution of crude imine
in a total of 11.7 L of CH2Cl2 was added tetrabutylammonium
hydrogen sulfate (178 g, 0.52 mol, and methallyl dichloride (2.0
kg, 16 mol). Rapid stirring was initiated, and 7.8 L of a 10 M NaOH
solution was added. After the mixture was stirred at room
temperature overnight, the layers were split, and the bottom aqueous
layer was removed. To the reactor was added 6 L of 1 M HCl,
stirring was carried out for 2 h, the layers were separated, and the
aqueous layer was basified to pH 8 with Na2CO3 and extracted
three times with EtOAc. The combined EtOAc extracts were
concentrated to dryness, dissolved in CH2Cl2, washed with saturated
aqueous NaHCO3 and water, dried over Na2SO4, and concentrated
to provide 375 g of crude 4 that was carried into the next step
without further purification.
6-Methylene-7a-phenyltetrahydro-1H-pyrrolo[1,2-c][1,3]oxazol-
3-one (5). A 20 L reactor was charged with 6.8 L of 1 M LAH in
THF (6.8 mol) and cooled to 0 °C. A solution of crude 4 (375 g,
1.73 mol) in 2.0 L of THF was added to the reactor over 40 min,
cooling was removed, and stirring was maintained for 1 h at room
temperature. The reaction was carefully quenched with 257 mL
water, followed by 257 mL of 15% NaOH, and then 770 mL of
water. To the reactor was added 1.2 kg of Na2SO4. Stirring was
maintained overnight, and the solids were filtered and washed with
THF. The combined filtrates were concentrated to provide 320 g
of a crude oil that was used in the next step without purification.
This material was dissolved in 12 L of CH2Cl2, and to the mixture
was added triethylamine (350 mL, 2.5 mol) and 1,1′-carbonyldi-
imidazole (357 g, 2.2 mol). After the mixture was stirred overnight
at room temperature, the reaction was quenched with 1 M HCl,
the layers were split, and the organic was washed with water. The
organic phase was dried over Na2SO4 and concentrated to provide
336 g of 5 as a crude oil that was used in the next step without
purification.
% (T ⁄ C) ) 100 × (∆VT) ⁄ (∆VC) if ∆VT > 0
% (T ⁄ C) ) 100 × (∆VT) ⁄ Ti if ∆VT < 0
The error for each data point was determined by the following
equation, where StdDev is the standard deviation and n is the
number of animals in the given treatment group:
√
error ) StdDev ⁄ n
(B) Chemistry. All solvents used were commercially available
“anhydrous” grade, and reagents were utilized without purification
unless otherwise noted. Nonaqueous reactions were carried out in
oven- or heat-gun-dried glassware under a nitrogen atmosphere.
Magnetic stirring was used to agitate the reactions, and they were
monitored for completion by either TLC (silica gel 60, Merck
KGaA) or LC/MS (Waters 2690 separations module with a YMC
Pro 50 mm × 3 mm i.d. C-18 column interfaced with a Waters
Micromass ZMD spectrometer) utilizing a gradient of 0.05% TFA
in water/acetonitrile with detection at 215 or 254 nm. A Smith-
Creator microwave from Personal Chemistry was used for micro-
wave heating, and a CombiFlash system utilizing RediSep cartridges
by Teledyne Isco was utilized for silica gel chromatography with
fraction collection at 254 nm. Reverse phase HPLC purification
was carried out on a Waters HPLC (XBridge Prep C-18 5 µM 19
mm × 150 mm column) utilizing a gradient of 0.1% TFA in water/
acetonitrile with sample collection triggered by photodiode array
detection. The reported yields are for isolated pure compounds
unless otherwise noted and were not extensively optimized.
Analytical HPLC was carried out on the above-mentioned LCMS
system with a nine minute gradient (method A) and on a Hewlett-
Packard 1100 series HPLC with a Sunfire C-18 5 µM 4.5 mm ×
50 mm column and a gradient of 0.1% H3PO4 in water/acetonitrile
7a-Phenyldihydro-1H-pyrrolo[1,2-c][1,3]oxazole-3,6(5H)-di-
one (6). Crude 5 (336 g, 1.56 mol) was dissolved in CH2Cl2 and
split into two equal portions to process separately. The solution
was cooled to -78 °C and purged with O2, and then ozone was
bubbled through the solution at a rate of 5 L/min until a pale-blue
color persisted (∼4 h). The mixture was purged of excess ozone,
quenched dropwise with excess dimethyl sulfide (500 mL, ∼10
equiv), and allowed to slowly warm to room temperature overnight
with stirring. The two runs were pooled, the solvents were removed
by rotary evaporation, and the solid was triturated with acetone to
provide 160 g of 6 (15% from phenylglycine methyl ester) as a
1
white solid. H NMR (500 MHz, DMSO-d6) δ 7.4 (m, 4H), 7.35
(m, 1H), 4.85 (d, J ) 8.9 Hz, 1H), 4.30 (d, J ) 8.9 Hz, 1H), 4.0
(d, J ) 18.3 Hz, 1H), 3.55 (d, J ) 18.3 Hz, 1H), 3.35 (d, J ) 17.4
Hz), 2.90 (d, J ) 17.4 Hz, 1H) ppm. LC/MS: tR ) 2.35 min (98%
pure). LRMS ) 218.0 (M + H).
6-(2,5-Difluorophenyl)-7a-phenyl-5,7a-dihydro-1H-pyrrolo[1,2-
c][1,3]oxazol-3-one (7). A suspension of 6 (12.5 g, 57.5 mmol) in
500 mL of THF was cooled to -78 °C. With rapid stirring, a 1 M
solution of NaHMDS (69 mL, 69 mmol) was added dropwise and
allowed to age for 30 min before warming to 0 °C for 1 h. The

KSP Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14 4249
solution was cooled back to -78 °C, and solid N-phenylbis(trif-
luoromethanesulphonimide) (24.7 g, 69 mmol) was added in one
portion. The mixture was allowed to warm slowly to room
temperature overnight and quenched with a saturated solution of
NH4Cl. The organic layer was separated, and the aqueous phase
was extracted twice with EtOAc. The combined organic phases
were washed with brine, dried over Na2SO4, and concentrated. This
residue was dissolved in 400 mL of 1,2-dimethoxyethane, and to
this was added 100 mL of water, LiCl (7.3 g, 172 mmol), Na2CO3
(18.3 g, 172 mmol), and 2,5-difluorophenylboronic acid (16 g, 101
mmol). The solution was degassed with N2 for 2 min, charged with
tetrakis(triphenylphosphine)palladium(0) (3.5 g, 3 mmol), and
heated to 90 °C for 3 h. After the mixture was cooled to room
temperature, most of the DME was removed by distillation and
the crude reaction was dumped into a separatory funnel with 500
mL of CH2Cl2 and 200 mL of water. The layers were cut, and the
aqueous layer was extracted twice more with CH2Cl2. The combined
organic phase was washed with water, dried over Na2SO4, and
concentrated by rotary evaporation. The residue was dry loaded
on silica and purified by column chromatography with a gradient
of 0-100% CH2Cl2 in hexanes to provide semipure 7 as a white
solid. This material was triturated with 150 mL of Et2O, stirred
overnight, and filtered to provide 12.5 g (69%) of 7 as a white
(2S)-4-(2,5-Difluorophenyl)-2-(hydroxymethyl)-N,N-dimethyl-
2-phenyl-2,5-dihydro-1H-pyrrole-1-carboxamide (9). To a sus-
pension of triphosgene (3.7 g, 12.5 mmol) in 400 mL of THF at 0
°C was added a solution of (S)-8 (10 g, 24.9 mmol) and
triethylamine (5.2 mL, 37.4 mmol) in THF over ∼2 min. After the
mixture was warmed and stirred for 1 h at room temperature, a 2
M solution of dimethylamine in THF (50 mL, 100 mmol) was added
and stirring was continued for 2 h. The mixture was added to a
separatory funnel containing 100 mL of 1 M HCl, and the mixture
was extracted twice with EtOAc. The combined organic layers were
washed again with HCl and then with brine, dried over Na2SO4,
and concentrated. The residue was dissolved in 300 mL of THF
and cooled to 0 °C, and 4 M HCl in dioxane (25 mL, 100 mmol)
was added dropwise. The mixture was allowed to warm to room
temperature and was stirred for 2 h before being added to a
separatory funnel containing 5% aqueous Na2CO3. The mixture was
extracted twice with EtOAc, washed with saturated brine, dried
over Na2SO4, and concentrated. The residue was purified by column
chromatography with a gradient of 2-100% EtOAc in hexanes to
1
provide 7.0 g (78%) of 9 as a white solid. H NMR (500 MHz,
CDCl3) δ 7.35 (m, 4H), 7.25 (m, 1H), 7.05 (m, 1H), 6.95 (m, 2H),
6.28 (s, 1H), 5.46 (m, 1H), 4.86 (d, J ) 13.8 Hz, 1H), 4.63 (d, J
) 13.8 Hz, 1H), 4.42 (m, 1H), 4.00 (d, J ) 12.2 Hz, 1H), 2.98 (s,
6H) ppm. HPLC purity: method A ) 100%; method B ) 98.5%.
HRMS calcd for C20H20F2N2O2 (M + H): 359.1566. Found:
359.1564. [R]D -66.4 (c 1.0, chloroform).
1
solid. H NMR (500 MHz, CDCl3) δ 7.4 (m, 4H), 7.35 (m, 1H),
7.05 (m, 1H), 6.95 (m, 2H), 6.80 (d, J ) 1.0 Hz, 1H), 4.85 (d, J )
15.0 Hz, 1H), 4.75 (d, J ) 8.7 Hz, 1H), 4.50 (d, J ) 8.7 Hz, 1H),
4.20 (m, 1H) ppm. LC/MS: tR ) 4.56 min (96% pure). LRMS )
313.9 (M + H).
(2S)-4-(2,5-Difluorophenyl)-2-(hydroxymethyl)-N-methyl-N-
(1-methylpiperidin-4-yl)-2-phenyl-2,5-dihydro-1H-pyrrole-1-car-
boxamide (11). To a suspension of triphosgene (982 mg, 9.7 mmol)
in 100 mL of THF at 0 °C was added a solution of (S)-8 (2.6 g, 6.5
mmol) and triethylamine (1.35 mL, 9.7 mmol) in THF over ∼2 min.
After stirring for 1 h at 0 °C, the mixture was warmed to room
temperature and stirred for an additional 2 h. The reaction was
quenched with saturated aqueous brine, added to a separatory funnel
containing EtOAc, and the layers were cut. The aqueous phase was
extracted again with EtOAc and the combined organic layers were
washed with saturated brine, dried over Na2SO4, and concentrated to
provide the intermediate carbamoyl chloride. This material was
dissolved in 100 mL of CH2Cl2, and 4-methylamino-N-methylpiperi-
dine (1.67 g, 13 mmol) and triethylamine (2.7 mL, 19.4 mmol) were
added, followed by a catalytic amount of DMAP. After being stirred
overnight at room temperature, the mixture was added to a separatory
funnel containing saturated aqueous NaHCO3, and the mixture was
extracted twice with CH2Cl2. The combined organic layers were
washed with water, dried over Na2SO4, and concentrated. The residue
was purified by column chromatography with a gradient of 0-50%
20:1:1 EtOH/H2O/NH4OH in EtOAc to provide a colorless oil. This
residue was dissolved in 150 mL of THF. The mixture was cooled to
0 °C, and 4 M HCl in dioxane (6.5 mL, 26 mmol) was added dropwise.
The mixture was allowed to warm to room temperature and was stirred
for 2 h before being added to a separatory funnel containing 5%
aqueous Na2CO3. The mixture was extracted twice with EtOAc,
washed with saturated brine, dried over Na2SO4, and concentrated.
An attempt to purify the residue by column chromatography with a
gradient of 0-50% 20:1:1 EtOH/H2O/NH4OH in EtOAc provided ∼2
g of 11 (70%) as a white solid but with only 87% purity. Analytically
pure material was obtained by reverse phase HPLC purification of a
portion of this material and free-basing the fractions with NaHCO3.
The basic fractions were extracted with EtOAc, dried over Na2SO4,
and concentrated to provide 380 mg of 11 as a white solid. 1H NMR
(500 MHz, CDCl3) δ 7.35 (m, 4H), 7.25 (m, 1H), 7.05 (m, 1H), 6.95
(m, 2H), 6.29 (s, 1H), 5.40 (m, 1H), 4.83 (d, J ) 13.7 Hz, 1H), 4.57
(d, J ) 13.7 Hz, 1H), 4.42 (m, 1H), 4.00 (d, J ) 12.0 Hz, 1H), 3.71
(m, 1H), 2.95 (m, 2H), 2.90 (s, 3H), 2.29 (s, 3H), 2.10 (m, 2H), 2.0 -
1.8 (m, 3H), 1.65 (m, 1H) ppm. HPLC purity: method A ) 100%;
method B ) 100%. HRMS calcd for C25H29F2N3O2 (M + H):
442.2301. Found: 442.2300. [R]D -41.9 (c 1.0, chloroform).
4-(2,5-Difluorophenyl)-2-phenyl-2-[(1,1,2,2-tetramethylpro-
poxy)methyl]-2,5-dihydro-1H-pyrrole (8). To a suspension of 7
(12.2 g, 38.9 mmol) in 120 mL of EtOH was added 3 M NaOH
(77.9 mL, 233 mmol). The mixture was heated to 60 °C for 4 h,
cooled to room temperature, and added to a separatory funnel
containing 250 mL of CHCl3 and saturated brine. The layers were
separated, and the aqueous phase was extracted twice with 2:1
CHCl3/EtOH. The combined organic layers were washed with brine
and concentrated. This residue was suspended in 100 mL of CH2Cl2
and 40 mL of DMF. Imidazole (10.6 g, 156 mmol) and TBSCl
(12.3 g, 82 mmol) were added. The mixture was allowed to stir
overnight, and it was added to a separatory funnel containing water
and CH2Cl2. The organic phase was washed several times with
water, dried over Na2SO4, and concentrated to an amber oil. This
crude material was dissolved in 100 mL of MeOH and treated with
∼100 mL of 2 M methylamine in MeOH for 4 h at room
temperature (to cleave the N-TBS group on the core). Following
concentration by rotary evaporation, the residue was purified by
column chromatography with a gradient of 0-40% EtOAc in
hexanes to provide 12.9 g (83%) of 8 as a white solid. LC/MS: tR
) 4.03 min (100% pure). LRMS ) 402.0 (M + H).
(2S)-4-(2,5-Difluorophenyl)-2-phenyl-2-[(1,1,2,2-tetramethyl-
propoxy)methyl]-2,5-dihydro-1H-pyrrole [(S)-8]. Resolution of
the enantiomers of 8 was carried out chromatographically using a
Chiralpak AD 10 cm × 50 cm column with 1% isopropanol in
hexanes (containing 0.1% diethylamine as a modifier) at 175 mL/
min. Injection of 20 g per run was determined to be optimal, with
retention times of 35-55 min for peak 1 and 60-95 min for peak
2 and recovery of >95%. Analytical HPLC analysis of the eluent,
carried out on a 4 mm × 250 mm Chiralpak AD column with the
same eluent as above at a flow rate of 1 mL/min, indicated that the
first enatiomer to elute (tR ) 7.97 min) was >99% ee and the second
peak to elute (tR ) 10.0 min) was ∼99% ee. The first enatiomer to
elute is the desired (S)-isomer as determined by KSP activity and
X-ray analysis of subsequent analogues (see US2005/0038074 for
small-molecule X-ray analysis of 30). Data for (S)-8: 1H NMR (500
MHz, CDCl3) δ 7.55 (m, 2H), 7.37 (m, 2H), 7.28 (m, 1H), 7.05
(m, 1H), 6.95 (m, 2H), 6.75 (d, J ) 1.6 Hz, 1H), 4.42 (d, J ) 14.0
Hz, 1H), 4.10 (dd, J ) 14.0, 1.6 Hz, 1H), 3.94 (d, J ) 10.2 Hz,
1H), 3.74 (d, J ) 10.2 Hz, 1H), 2.57 (bs, 1H), 0.85 (s, 9H), 0.02
(s, 3H), 0.00 (s, 3H) ppm. LC/MS: tR ) 4.03 min (100% pure).
LRMS ) 402.0 (M + H). [R]D +28.4 (c 1.0, chloroform).
Benzyl 3-Fluoro-4-oxopiperidine-1-carboxylate (18). To a
solution of benzyl-4-oxo-1-piperidinecarboxylate (50 g, 215 mmol)
in 225 mL of DMF was added triethylamine (90 mL, 644 mmol)
and then chlorotrimethylsilane (38.1 mL, 300 mmol). The mixture

4250 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14
Cox et al.
was heated at 80 °C overnight, cooled to room temperature, and
then dumped into a separatory funnel containing hexanes. The
mixture was partitioned with water, washed with brine, dried over
Na2SO4, and concentrated by rotary evaporation. The residue was
dissolved in 1000 mL of CH3CN, cooled to 0 °C, and treated with
Selectfluor (79.8 g, 225 mmol) in three portions over 5 min. After
the mixture was stirred at room temperature overnight, most of the
solvent was removed by rotary evaporation and then partitioned
between EtOAc and water, separated, washed with saturated brine
solution, dried over MgSO4, filtered, and concentrated by rotary
evaporation. The residue was purified by column chromatography
with a gradient of 33-66% EtOAc in hexanes to provide 48.5 g
(90%) of 18 as a white solid. LRMS ) 252 (M + H).
5.13 (bs, 2H), 4.9-4.6 (m, 1H), 4.6-3.9 (m, 3H), 3.1-2.9 (m,
2H), 2.85 (bs, 3H), 2.15 (m, 1H), 1.5 (m, 1H), 1.47 (bs, 9H) ppm.
HPLC purity: method A ) 100%; method B ) 100%. HRMS calcd
for C19H27FN2O4 (M + Na): 389.1847. Found: 389.1847. [R]D
-72.9 (c 1.0, chloroform).
(3R,4S)-3-Fluoro-N,1-dimethylpiperidin-4-amine (28). To a
solution of 24 (4.6 g, 12.5 mmol) in 150 mL of EtOH was added
1,4-cyclohexadiene (29.9 mL, 313 mmol) and 10% Pd on carbon
(∼1 g). After being stirred overnight, the mixture was filtered
through Celite and concentrated by rotary evaporation. The residue
was dissolved in 75 mL of MeOH. Acetic acid (∼2 mL) and 37%
aqueous formaldehyde (30.1 mL, 38 mmol) were added, and the
mixture was stirred for 1 h. At that time, NaCNBH3 (1.58 g, 25.1
mml) in 10 mL of MeOH was added and the mixture was aged for
2 h before being dumped into saturated aqueous NaHCO3. After
extracting three times with CH2Cl2, the organic phase was washed
with water, dried over MgSO4, filtered, and concentrated by rotary
evaporation to provide 3.0 g of a colorless oil. LRMS ) 247.2 (M
+ H). A portion of this material (1.43 g, 5.8 mmol) in 50 mL of
EtOAc was treated with HCl gas until the solution was warm to
the touch. The flask was capped and stirred for 2 h. The volatiles
were removed by rotary evaporation, and the residue was triturated
with Et2O and placed under high vacuum to provide a white solid.
This material was partitioned between 5% aqueous Na2CO3 and
2:1 CHCl3/EtOH. The layers were cut, and the aqueous phase was
extracted twice more with 2:1 CHCl3/EtOH. The combined organic
layers were washed with saturated brine and concentrated by rotary
evaporation and the residue was dissolved in CH2Cl2, dried over
Na2SO4, and concentrated to provide 628 mg (50%) of 28 a
(()-trans-Benzyl 3-Fluoro-4-(methylamino)piperidine-1-car-
boxylate (19) and (()-cis-Benzyl 3-Fluoro-4-(methylamino)pi-
peridine-1-carboxylate (20). To a solution of 18 (10 g, 39.8 mmol)
in 100 mL of MeOH was added acetic acid (∼5 mL) and a 2 M
solution of methylamine in THF (60 mL, 120 mmol). After the
mixture was strirred for 3 h, NaCNBH3 (3.76 g, 60 mmol) was
added. After the mixture was stirred overnight, the reaction was
quenched with saturated aqueous NaHCO3, partitioned with CH2Cl2,
washed with brine, dried over MgSO4, filtered, and concentrated
by rotary evaporation. The residue was loaded onto a silica gel
column and eluted with EtOAc, then 80:10:10 CHCl3/EtOAc/MeOH
to provide 3.7 g (35%) of 19 (100% pure) and 2.9 g (27%) of 20
(100% pure) as colorless oils. Data for 19, first to elute (confirmed
by NOE analysis at -35 °C): 1H NMR (500 MHz, CD2Cl2) δ
7.4-7.3 (m, 5H), 5.1 (m, 2H), 4.4-4.1 (m, 2H), 3.9 (m, 1H),
3.15-3.05 (m, 2H), 2.75 (m, 1H), 2.4 (s, 3H), 2.0 (m, 1H), 1.25
(m, 1H) ppm. Data for 20, second to elute (confirmed by NOE
analysis at -35 °C): 1H NMR (500 MHz, CD2Cl2) δ 7.4-7.2 (m,
5H), 5.1 (m, 2H), 4.9-4.7 (m 1H), 4.4 (m, 1H), 4.15 (m, 1H),
3.1-2.9 (m, 2H), 2.6 (m, 1H), 2.4 (s, 3H), 1.8 (m, 1H), 1.6 (m,
1H) ppm. HRMS calcd for C26H30F3N3O2 (M + H): 267.1504.
Found: 267.1500. By a change of solvents, the cis-isomer can be
obtained as the major product: To a solution of 18 (50.5 g, 201 mmol)
in 700 mL of 1,2-dichloroethane was added a 2 M solution of
methylamine in THF (201 mL, 402 mmol). After the mixture was
stirred for 2 h, Na(OAc)3BH (64 g, 302 mmol) was added. After the
mixture was stirred overnight, the reaction was quenched with saturated
aqueous K2CO3, partitioned with CH2Cl2, separated, and the aqueous
phase was extracted three times with CH2Cl2. The combined organic
extracts were washed with brine, dried over MgSO4, filtered, and
concentrated by rotary evaporation. The residue was loaded onto a
silica gel column and eluted with EtOAc, then 90:5:5 CHCl3/EtOAc/
MeOH, and finally 80:10:10 CHCl3/EtOAc/MeOH. The column was
repeated twice more on mixed fractions to provide 7 g (13%) of 19
and 38.7 g (72%) of 20 (100% pure).
1
colorless oil. HNMR (500 MHz, CDCl3) δ 4.8 (d, J ) 48.9 Hz,
1H), 3.15 (m, 1H), 2.85 (m, 1H), 2.5 (s, 3H), 2.45 (m, 1H), 2.3 (s,
3H), 2.2-2.0 (m, 2H), 1.9-1.7 (m, 2H) ppm. HRMS calcd for
C7H15FN2 (M + H): 147.1292. Found: 147.1300.
(2S)-2-({[tert-Butyl(dimethyl)silyl]oxy}methyl)-4-(2,5-difluo-
rophenyl)-2-phenyl-2,5-dihydro-1H-pyrrole-1-carbonyl Chloride
(29). To a suspension of triphosgene (7.76 g, 26.2 mmol) in 600
mL of THF at 0 °C was added a solution of (S)-8 (21 g, 52.3 mmol)
and triethylamine (11 mL, 79 mmol) over ∼3 min. After being
stirred for 1 h at 0 °C, the mixture was warmed to room temperature
and stirred for an additional 3 h. The reaction was quenched with
300 mL of water, added to a separatory funnel containing EtOAc,
and the layers were cut. The aqueous phase was extracted again
with EtOAc, and the combined organic layers were washed with
saturated brine, dried over Na2SO4, concentrated, resuspended in
CH2Cl2, and concentrated to provide 29 as a thick, colorless oil.
This material is stable for months under nitrogen in a closed
container at 0 °C. LC/MS (3.7 min run): tR ) 3.56 min (100%
pure). LRMS ) 464.1 (M + H).
Benzyl (3S,4R)-4-[(tert-Butoxycarbonyl)(methyl)amino]-3-
fluoropiperidine-1-carboxylate (23) and Benzyl (3R,4S)-4-[(tert-
Butoxycarbonyl)(methyl)amino]-3-fluoropiperidine-1-carboxy-
late (24). To a solution of 20 (2.9 g, 10.9 mmol) in 55 mL of
CH2Cl2 was added triethylamine (4.6 mL, 30.3 mmol) and di-tert-
butyl dicarbonate (3.6 g, 16.4 mmol). After being stirred for 48 h,
the mixture was partitioned between CH2Cl2 and H2O, and the
organic phase was washed with brine, dried over MgSO4, filtered,
and concentrated by rotary evaporation. The residue was loaded
onto a silica gel column in CHCl3 and eluted with a gradient of
0-33% EtOAc in hexanes to provide a white solid. Resolution of
the enantiomers was carried out chromatographically using a
Chiralpak AD 10 cm × 50 cm column with 20% isopropanol in
hexanes (with 0.1% diethylamine) at 150 mL/min to provide 1.78 g
(45%) of the first eluting enatiomer 23 and 1.64 g (41%) of the
second eluting enatiomer 24 (the absolute stereochemistry of 24 is
assigned by small-molecule X-ray analysis of subsequent analogues
(see US2005/0038074 or WO2005/102996). Analytical HPLC
analysis of the eluent on a 4 mm × 250 mm Chiralpak AD column
with 20% isopropanol in hexanes (with 0.1% diethylamine) at 1
mL/min indicated that 23 has tR ) 5.69 min and >98% ee and 24
has tR ) 6.52 min and ee ) 98%. Data for 24: 1H NMR (500
MHz, CDCl3, extensive broadening at 25 °C) δ 7.4-7.3 (m, 5H),
(2S)-4-(2,5-Difluorophenyl)-N-[(3R,4S)-3-fluoro-1-methylpi-
peridin-4-yl]-2-(hydroxymethyl)-N-methyl-2-phenyl-2,5-dihydro-
1H-pyrrole-1-carboxamide (30). To a solution of 29 (1.6 g, 3.45
mmol) in 25 mL of THF was added 28 (630 mg, 4.3 mmol),
triethylamine (1.44 mL, 10.3 mmol), and a catalytic amount of
DMAP. After being stirred for 24 h at room temperature, the
mixture was partitioned between EtOAc and saturated aqueous
NaHCO3, and the organic layer was washed with brine, dried over
Na2SO4, and concentrated by rotary evaporation. The residue was
purified by silica gel chromatography with 0-10% MeOH in
CH2Cl2 (containing ∼0.5% triethylamine) to provide ∼1.7 g of a
pale-yellow taffy. This material was dissolved in 75 mL of CH3CN,
3 mL of triethylamine trihydrofluoride was added, and the mixture
was stirred for 24 h at room temperature. An additional 3 mL of
triethylamine trihydrofluoride was added, and the mixture was
heated at 37 °C for 24 h. The mixture was cooled to room
temperature, dumped into saturated aqueous NaHCO3, extracted
three times with EtOAc, washed with brine, dried over Na2SO4,
and concentrated by rotary evaporation. The residue was purified
by silica gel chromatography with a gradient of 0-50% 20:1:1
EtOH/NH4OH/H2O in EtOAc to provide 1.0 g (63%) of 30 as a
white solid. 1H NMR (500 MHz, CDCl3) δ 7.35 (m, 4H), 7.25 (m,
1H), 7.05 (m, 1H), 6.95 (m, 2H), 6.28 (s, 1H), 5.22 (dd, J ) 8.8,

KSP Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14 4251
Grandinetti, C. A.; Goldspiel, B. R. Sorafenib and sunitinib: novel
targeted therapies for renal cell cancer. Pharmacotherapy 2007, 27,
1125–1144.
3.7 Hz, 1H), 4.92 (d, J ) 13.7 Hz, 1H), 4.80 (d, J ) 50.3 Hz, 1H),
4.62 (d, J ) 13.8 Hz, 1H), 4.44 (m, 1H), 4.1 - 4.0 (m, 2H), 3.15
(m, 1H), 3.11 (s, 3H), 3.0 (m, 1H), 2.4-2.1 (m, 3H), 2.31 (s, 3H),
1.70 (m, 1H) ppm. HPLC purity: method A ) 100%; method B )
100%. HRMS calcd for C25H28F3N3O2 (M + H): 460.2206. Found:
460.2213. [R]D -70.5 (c 1.0, chloroform).
(3) Conlin, A. K.; Seidman, A. D. Taxanes in breast cancer: An update.
Curr. Oncol. Rep. 2007, 9, 22–30.
(4) (a) Miglarese, M. R.; Carlson, R. O. Development of new cancer
therapeutic agents targeting mitosis. Expert Opin. InVest. Drugs 2006,
15, 1411–1425. (b) Schmidt, M.; Bastians, H. Mitotic drug targets
and the development of novel anti-mitotic anticancer drugs. Drug
Resist. Updates 2007, 10, 162–181.
(5) Wood, K. W.; Cornwell, W. D.; Jackson, J. R. Past and future of the
mitotic spindle as an oncology target. Curr. Opin. Pharmacol. 2001,
1, 370–377.
(6) Jackson, J. R.; Patrick, D. R.; Dar, M. M.; Huang, P. S. Targeted
anti-mitotic therapies: can we improve on tubulin agents? Nat. ReV.
Cancer 2007, 7, 107–117.
(7) (a) Goldstein, L. S. B.; Philp, A. V. The road less traveled: emerging
principles of kinesin motor utilization. Annu. ReV. Cell DeV. Biol. 1999,
15, 141–183. (b) Mandelkow, E.; Mandelkow, E.-M. Kinesin motors
and disease. Trends Cell Biol. 2002, 12, 585–591.
(8) (a) Sharp, D. J.; Rogers, G. C.; Scholey, J. M. Microtubule motors in
mitosis. Nature 2000, 407, 41–47. (b) Heald, R. Motor function in
the mitotic spindle. Cell 2000, 102, 399–402. (c) Bergnes, G.; Brejc,
K.; Belmont, L. Mitotic kinesins: prospects for antimitotic drug
discovery. Curr. Top. Med. Chem. 2005, 5, 127–145.
(9) (a) Mayer, T. U.; Kapoor, T. M.; Haggarty, S. J.; King, R. W.;
Schreiber, S. L.; Mitchison, T. J. Small molecule inhibitor of mitotic
spindle bipolarity identified in a phenotype-based screen. Science 1999,
286, 971–974. (b) 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.
Antitumor activity of a kinesin inhibitor. Cancer Res. 2004, 64, 3276–
3280. (c) Jiang, C.; You, Q.; Li, Z.; Guo, Q. Kinesin spindle protein
inhibitors as anticancer agents. Expert Opin. Ther. Pat. 2006, 16, 1517–
1532.
(10) (a) Marcus, A. I.; Peters, U.; Thomas, S. L.; Garrett, S.; Zelnak, A.;
Kapoor, T. M.; Giannakakou, P. Mitotic kinesin inhibitors induce
mitotic arrest and cell death in paclitaxel-resistant and -sensitive cancer
cells. J. Biol. Chem. 2005, 280, 11569–11577. (b) Dutcher, J. P.;
Novik, Y.; O’Boyle, K.; Marcoullis, G.; Secco, C.; Wiernik, P. H.
20th-Century advances in drug therapy in oncology. Part II. J. Clin.
Pharmacol. 2000, 40, 1079–1092.
Benzyl (3S,4S)-4-[(tert-Butoxycarbonyl)(methyl)amino]-3-
fluoropiperidine-1-carboxylate (21) and Benzyl (3R,4R)-4-[(tert-
Butoxycarbonyl)(methyl)amino]-3-fluoropiperidine-1-carboxy-
late (22). To a solution of 19 (3.7 g, 13.9 mmol) in 70 mL of
CH2Cl2 was added triethylamine (5.8 mL, 41.7 mmol) and di-tert-
butyl dicarbonate (4.55 g, 20.9 mmol). After being stirred for 48 h,
the mixture was partitioned between CH2Cl2 and H2O, and the
organic phase was washed with brine, dried over MgSO4, filtered,
and concentrated by rotary evaporation. The residue was loaded
onto a silica gel column in CHCl3 and eluted with a gradient of
0-33% EtOAc in hexanes to provide a white solid. Resolution of
the enantiomers was carried out chromatographically using a
Chiralpak AD 10 cm × 50 cm column with 15% isopropanol in
hexanes (with 0.1% diethylamine) at 150 mL/min to provide 2.3 g
(45%) of the first eluting enatiomer 21 and 2.5 g (49%) of the
second eluting enatiomer 22 (the absolute stereochemistry of 21
and 22 is not known but is tentatively assigned as shown to simplify
visualization of the relative stereochemistry). Analytical HPLC
analysis of the eluent on a 4 mm × 250 mm Chiralpak AD column
with 15% isopropanol in hexanes (with 0.1% diethylamine) at 1
mL/min indicated that 21 has tR ) 7.41 min and 98% ee and 22
has tR ) 11.7 min and ee ) 98%. Data for 21: 1H NMR (500
MHz, CDCl3, extensive broadening at 25 °C) δ 7.4-7.3 (m, 5H),
5.12 (bs, 2H), 4.7-4.35 (m, 2 H), 4.3-3.9 (m, 2H), 2.80 (bs, 5H),
1.8-1.6 (m, 2H), 1.46 (s, 9H) ppm. HPLC purity: method A )
100%; method B ) 100%. HRMS calcd for C19H27FN2O4 (M +
Na): 389.1847. Found: 389.1850. [R]D -9.2 (c 1.0, chloroform).
(2S)-4-(2,5-Difluorophenyl)-N-[(3S,4S)-3-fluoro-1-methylpi-
peridin-4-yl]-2-(hydroxymethyl)-N-methyl-2-phenyl-2,5-dihydro-
1H-pyrrole-1-carboxamide (31). A procedure analogous to that
used to prepare 30 was employed beginning with 21 to provide 31
as a white solid. 1H NMR (500 MHz, CDCl3) δ 7.35 (m, 4H),
7.25 (m, 1H), 7.05 (m, 1H), 6.95 (m, 2H), 6.29 (s, 1H), 4.91 (bs,
1H), 4.85 (dd, J ) 14.2, 1.5 Hz, 1H), 4.8-4.7 (m, 1H), 4.71 (d, J
) 12.8 Hz, 1H), 4.45 (m, 1H), 4.03 (d, J ) 12.0 Hz, 1H), 3.82 (m,
1H), 3.26 (m, 1H), 2.90 (s, 3H), 2.83 (m, 1H), 2.33 (s, 3H), 2.15
(m, 1H), 2.05-1.9 (m, 2H), 1.70 (m, 1H) ppm. HPLC purity:
method A ) 100%; method B ) 100%. HRMS calcd for
C25H28F3N3O2 (M + H): 460.2206. Found: 460.2196. [R]D -49.2
(c 0.5, chloroform).
(11) Sorbera, L. A.; Bolos, J.; Serradell, N.; Baye´s, M. Ispinesib mesilate.
Drugs Future 2006, 31, 778–787.
(12) It is tempting to speculate that antimitotic agents designed to overcome
known mechanisms of resistance, such as Pgp-overexpression and
tubulin mutations, will lead to improved efficacy in the clinic against
refractory tumors. Indeed, some recent studies have provided evidence
in support of this hypothesis. See, for example, the following: (a)
Toppmeyer, D.; Seidman, A. D.; Pollak, M.; Russell, C.; Tkaczuk,
K.; Verma, S.; Overmoyer, B.; Garg, V.; Ette, E.; Harding, M. W.;
Demetri, G. D. Safety and efficacy of the multidrug resistance inhibitor
Incel (Biricodar; VX-710) in combination with paclitaxel for advanced
breast cancer refractory to paclitaxel. Clin. Cancer Res. 2002, 8, 670-
678. (b) Goodin, S.; Kane, M. P.; Rubin, E. H. Epothilones: mechanism
of action and biologic activity. J. Clin. Oncol. 2004, 22, 2015-2025.
However, results from these clinical trials, though positive, have been
less than stellar, and the validity of this rationale is still often
questioned. See the following: (c) Greenberger, L. M.; Sampath, D.
Resistance to Taxanes. In Cancer Drug DiscoVery and DeVelopment:
Cancer Drug Resistance; Teicher, B., Ed.; Humana Press Inc.: Totowa,
NJ, 2006; pp 329-357. (d) Ferlini, C.; Raspaglio, G.; Cicchillitti, L.;
Mozzetti, S.; Prislei, S.; Bartollino, S.; Scambia, G. Looking at drug
resistance mechanisms for microtubule interacting drugs: Does TUBB3
work? Curr. Cancer Drug Targets 2007, 7, 704–712.
Acknowledgment. We thank Rick Woodward for the large-
scale synthesis of intermediate 5, Carl Homnick for determining
chiral separation conditions for 8, Dr. Chuck Ross and Joan
Murphy for HRMS analysis, Dr. Steve Pitzenberger for 2D and
NOE NMR analyses, Matt Zrada for pKa determinations, Dr.
David Dubost for stability and solubility measurements, and
Dr. Tito Fojo of the National Institutes of Health for access to
the PTX10 cell line.
(13) (a) Cox, C. D.; Breslin, M. J.; Mariano, B. J.; Coleman, P. J.; Buser,
C. A.; Walsh, E. S.; Hamilton, K.; Huber, H. E.; Kohl, N. E.; Torrent,
M.; Yan, Y.; Kuo, L. C.; Hartman, G. D. Kinesin spindle protein (KSP)
inhibitors. Part 1: The discovery of 3,5-dihydropyrazoles as potent
and selective inhibitors of the mitotic kinesin KSP. Bioorg. Med. Chem.
Lett. 2005, 15, 2041–2045. (b) Fraley, M. E.; Garbaccio, R. M.;
Arrington, K. L.; Hoffman, W. F.; Tasber, E. S.; Coleman, P. J.; Buser,
C. A.; Walsh, E. S.; Hamilton, K.; Fernandez, C.; Schaber, M. D.;
Lobell, R. B.; Tao, W.; South, V. J.; Yan, Y.; Kuo, L. C.;
Prueksaritanont, T.; Shu, C.; Torrent, M.; Heimbrook, D. C.; Kohl,
N. E.; Huber, H. E.; Hartman, G. D. Kinesin spindle protein (KSP)
inhibitors. Part 2: The design, synthesis, and characterization of 2,4-
diaryl-2,5-dihydropyrrole inhibitors of the mitotic kinesin KSP. Bioorg.
Med. Chem. Lett. 2006, 16, 1775–1779. (c) Cox, C. D.; Torrent, M.;
Breslin, M. J.; Mariano, B. J.; Whitman, D. B.; Coleman, P. J.; Buser,
C. A.; Walsh, E. S.; Hamilton, K.; Schaber, M. D.; Lobell, R. B.;
Tao, W.; South, V. J.; Kohl, N. E.; Yan, Y.; Kuo, L. C.; Prueksari-
tanont, T.; Slaughter, D. E.; Li, C.; Mahan, E.; Lu, B.; Hartman, G. D.
Supporting Information Available: Experimental procedures
for compounds 12-16, potency data for the epimers of 30 and 31,
and proton NMR spectra of all new compounds. This material is
available free of charge via the Internet at http://pubs.acs.org.
References
(1) Chabner, B. A.; Roberts, T. G., Jr. Chemotherapy and the war on
cancer. Nat. ReV. Cancer 2005, 5, 65–72.
(2) (a) Nadal, E.; Olavarria, E. Imatinib mesylate (Gleevec/Glivec) a
molecular-targeted therapy for chronic myeloid leukaemia and other
malignancies. J. Clin. Pract. 2004, 58, 511–516. (b) Hussain, S. A.;
Palmer, D. H.; Spooner, D.; Rea, D. W. Molecularly targeted
therapeutics for breast cancer. BioDrugs 2007, 21, 215–224. (c) Sun,
S.; Schiller, J. H.; Spinola, M.; Minna, J. D. New molecularly targeted
therapies for lung cancer. J. Clin. InVest. 2007, 117, 2740–2750. (d)

4252 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 14
Cox et al.
Kinesin spindle protein (KSP) inhibitors. Part 4: Structure-based design
of 5-alkylamino-3,5-diaryl-4,5-dihydropyrazoles as potent, water-
soluble inhibitors of the miotic kinesin KSP. Bioorg. Med. Chem. Lett.
2006, 16, 3175. (d) Cox, C. D.; Breslin, M. J.; Whitman, D. B.; Coleman,
P. J.; Garbaccio, R. M.; Fraley, M. E.; Zrada, M. M.; Buser, C. A.; Walsh,
E. S.; Hamilton, K.; Lobell, R. B.; Tao, W.; Abrams, M. T.; South, V. J.;
Huber, H. E.; Kohl, N. E.; Hartman, G. D. Kinesin spindle protein (KSP)
inhibitors. Part 5: Discovery of 2-propylamino-2,4-diaryl-2,5-dihydropy-
rroles as potent, water-soluble KSP inhibitors, and modulation of their
basicity by ꢀ-fluorination to overcome cellular efflux by P-glycoprotein.
Bioorg. Med. Chem. Lett. 2007, 17, 2697–2702.
(28) Tao, W.; South, V. J.; Zhang, Y.; Davide, J. P.; Farrell, L.; Kohl,
N. E.; Sepp-Lorenzino, L.; Lobell, R. B. Induction of apoptosis by an
inhibitor of the mitotic kinesin KSP requires both activation of the
spindle assembly checkpoint and mitotic slippage. Cancer Cell 2005,
8, 49–59.
(29) In mouse models, MTD is typically defined as e10% deaths due to
mechanism-based toxicity. In studies with 30 infused at 12 (mg/kg)/
day, there was 1 death out of 34 mice dosed, providing a 3% rate of
death for this cohort.
(30) The sequence of KSP is highly conserved across species, and the IC50
of 30 was found to be 2.2 ( 1.2 nM and 4.8 ( 1.0 nM against human
and mouse KSP, respectively. Therefore, we expect little difference
in therapeutic index across species due to differences in KSP potency.
(31) For instance, compound 24 in ref 14 has cell potency and serum shift
numbers of 3.4 and 6.6 nM, respectively, and its in vivo EC90 is 38 nM.
(14) Coleman, P. J.; Schreier, J. D.; Cox, C. D.; Fraley, M. E.; Garbaccio,
R. M.; Buser, C. A.; Walsh, E. S.; Hamilton, K.; Lobell, R. B.; Rickert,
K.; Tao, W.; Diehl, R. E.; South, V. J.; Davide, J. P.; Kohl, N. E.;
Yan, Y.; Kuo, L.; Prueksaritanont, T.; Li, C.; Mahan, E. A.; Fernandez-
Metzler, C.; Salata, J. J.; Hartman, G. D. Kinesin spindle protein (KSP)
inhibitors. Part 6: Design and synthesis of 3,5-diaryl-4,5-dihydropy-
razole amides as potent inhibitors of the mitotic kinesin KSP. Bioorg.
Med. Chem. Lett. 2007, 17, 5390–5395.
(32) Because mice become neutropenic and recover more quickly than
humans, a once per week for 3 weeks schedule in humans (q7d × 3)
is mimicked by a once every 4 days for 12 days cycle in mouse (q4d
× 3). Minipumps were not used in this protocol to allow for dosing
on multiple days.
(15) Roden, D. M. Drug-induced prolongation of the QT interval. N. Engl.
J. Med. 2004, 350, 1013–1022.
(16) Garbaccio, R. M.; Fraley, M. E.; Tasber, E. S.; Olson, C. M.; Hoffman,
W. F.; Arrington, K. L.; Torrent, M.; Buser, C. A.; Walsh, E. S.;
Hamilton, K.; Schaber, M. D.; Fernandez, C.; Lobell, R. B.; Tao, W.;
South, V. J.; Yan, Y.; Kuo, L. C.; Prueksaritanont, T.; Slaughter, D. E.;
Shu, C.; Heimbrook, D. C.; Kohl, N E.; Huber, H. E.; Hartman, G. D.
Kinesin spindle protein (KSP) inhibitors. Part 3: Synthesis and
evaluation of phenolic 2,4-diaryl-2,5-dihydropyrroles with reduced
hERG binding and employment of a phosphate prodrug strategy for
aqueous solubility. Bioorg. Med. Chem. Lett. 2006, 16, 1780–1783.
(17) Jamieson, C.; Moir, E. M.; Rankovic, Z.; Wishart, G. Medicinal
chemistry of hERG optimizations: highlights and hang-ups. J. Med.
Chem. 2006, 49, 5029–5046.
(33) Giannakakou, P.; Sackett, D. L.; Kang, Y.-K.; Zhan, Z.; Buters,
J. T. M.; Fojo, T.; Poruchynsky, M. S. Paclitaxel-resistant human
ovarian cancer cells have mutant ꢀ-tubulins that exhibit impaired
paclitaxel-driven polymerization. J. Biol. Chem. 1997, 272, 17118–
17125.
(34) The data shown in Figure 4 for efficacy in the KB-v cell line were
generated by dosing 30 on a qd × 1 schedule at 40 mpk as a suspension
in methylcellulose. This represents the MTD for this schedule, and
this dosing paradigm has shown good efficacy in a number of
experiments. The data from the q4d × 3 schedule in the KB-v cell
line are not shown because in this experiment (run only once), the
vehicle control group displayed much less tumor progression than
expected, similar to that of the group treated with 30. However, as
expected, the mice treated with paclitaxel in this study experienced
significant tumor growth. The reason for the low rate of tumor
progression in the vehicle group from this experiment is not known.
(18) By comparison, 2 had an MDR ratio of 4.6 with an EC50 of 26.4 nM
in the overexpressing cell line.
(19) For a review of the relevance of Pgp to drug discovery, see the
following: Lin, J. H.; Yamazaki, M. Clinical relevance of P-
glycoprotein in drug therapy. Drug Metab. ReV. 2003, 35, 417–454.
(20) (a) Seelig, A. How does P-glycoprotein recognize its substrates? Int.
J. Clin. Pharmacol Ther. 1998, 36, 50–54. (b) Stouch, T. R.;
Gudmundsson, O. Progress in understanding the structure-activity
relationships of P-glycoproetin. AdV. Drug DeliVery ReV. 2002, 54,
315–328.
(21) Raub, T. J. P-Glycoprotein recognition of substrates and circumvention
through rational drug design. Mol. Pharm. 2006, 3, 3–25.
(22) Yates, C. R.; Chang, C.; Kearbey, J. D.; Yasuda, K.; Schuetz, E. G.;
Miller, D. D.; Dalton, J. T.; Swaan, P. W. Structural determinants of
P-glycoprotein-mediated transport of glucocorticoids. Pharm. Res.
2003, 20, 1794–1803.
(23) Cerny, M. A.; Hanzlik, R. P. Cyclopropylamine inactivation of
cytochromes P450: role of metabolic intermediate complexes. Arch.
Biochem. Biophys. 2005, 426, 265–275.
(24) Goncharov, N. V.; Jenkins, R. O.; Radilov, A. S. Toxicology of
fluoroacetate: a review, with possible directions for therapy research.
J. Appl. Toxicol. 2006, 26, 148–161.
(25) van Niel, M. B.; Collins, I.; Beer, M. S.; Broughton, H. B.; Cheng,
S. K. F.; Goodacre, S. C.; Heald, A.; Locker, K. L.; MacLeod, A. M.;
Morrison, D.; Moyes, C. R.; O’Connor, D. O.; Pike, A.; Rowley, M.;
Russell, M. G. N.; Sohal, B.; Stanton, J. A.; Thomas, S.; Verrier, H.;
Watt, A. P.; Castro, J. L. Fluorination of 3-(3-(piperidin-1-yl)propy-
l)indoles and 3-(3-(piperazin-1-yl)propyl)indoles gives selective human
5-HT1D receptor ligands with improved pharmacokinetic profiles.
J. Med. Chem. 1999, 42, 2087–2104.
(26) The absolute stereochemistry of 30 was proven by X-ray analysis, whereas
the absolute stereochemistry at C3 and C4 of the piperidine ring in 31 is
not known but is drawn as such to simplify visualization of the relative
stereochemistry. See Experimental Section and Supporting Information
for full details and characterization of all four isomers.
(35) Yan, Y.; Sardana, V.; Xu, B.; Homnick, C.; Halczenko, W.; Buser,
C. A.; Schaber, M.; Hartman, G. D.; Huber, H. E.; Kuo, L. C.
Inhibition of a mitotic motor protein: where, how, and conformational
consequences. J. Mol. Biol. 2004, 335, 547–554.
(36) (a) Parish, C. A.; Adams, N. D.; Auger, K. R.; Burgess, J. L.; Carson,
J. D.; Chaudhari, A. M.; Copeland, R. A.; Diamond, M. A.; Donatelli,
C. A.; Duffy, K. J.; Faucete, L. F.; Finer, J. T.; Huffman, W. F.;
Hugger, E. D.; Jackson, J. R.; Knight, S. D.; Luo, L.; Moore, M. L.;
Newlander, K. A.; Ridgers, L. H.; Sakowicz, R.; Shaw, A. N.; Sung,
C.-M. M.; Sutton, D.; Wood, K. W.; Zhang, S.-Y.; Zimmerman, M. N.;
Dhanak, D. Novel ATP-competitive kinesin spindle protein inhibitors.
J. Med. Chem. 2007, 50, 4939–4952. (b) Rickert, K. W.; Schaber,
M.; Torrent, M.; Neilson, L. A.; Tasber, E. S.; Garbaccio, R. M.;
Coleman, P. J.; Harvey, D.; Zhang, Y.; Yang, Y.; Marshall, G.; Lee,
L.; Walsh, E. S.; Hamilton, K.; Buser, C. A. Discovery and
biochemical characterization of selective ATP competitive inhibitors
of the human mitotic kinesin KSP. Arch. Biochem. Biophys. 2008,
469, 220–231.
(37) The coordinates have been deposited with the RCSB Protein Data
Bank under the accession code 3CJO.
(38) Stein, M. N.; Rubin, E. H.; Taber, K.; Fernandez, R. C.; Agrawal,
N. G. B.; Vandrendreis, E.; Hsu, K.; Walker, A.; Holen, K.; Xue, L.;
Cooley, P. C.; Wilding, G. Phase I clinical and pharmacokinetic (PK)
trial of the kinesin spindle protein (KSP) inhibitor MK-0731 in patients
with solid tumors. J. Clin. Oncol. 2007, 25 (18S), 2548. (ASCO Annual
Meeting Proceedings Part I)
(39) Shen, D. W.; Cardarelli, C.; Hwang, J.; Cornwell, M.; Richert, N.;
Ishii, S.; Pastan, I.; Gottesman, M. M. Multiple drug-resistant human
KB carcinoma cells independently selected for high-level resistance
to colchicine, adriamycin, or vinblastine show changes in expression
of specific proteins. J. Biol. Chem. 1986, 261, 7762–7770.
(27) Crosio, C.; Fimia, G. M.; Loury, R.; Kimura, M.; Okano, Y.; Zhou,
H.; Sen, S.; Allis, C. D.; Sassone-Corsi, P. Mitotic phosphorylation
of histone H3: spatio-temporal regulation by mammalian aurora
kinases. Mol. Cell. Biol. 2002, 22, 874–885.
JM800386Y