Benzimidazole inhibitors induce a DFG-out conformation of never in mitosis gene a-related kinase 2 (Nek2) without binding to the back pocket and reveal a nonlinear structure-activity relationship

Article abstract of DOI:10.1021/jm1011726

We describe herein the structure-activity relationship (SAR) and cocrystal structures of a series of Nek2 inhibitors derived from the published polo-like kinase 1 (Plk1) inhibitor (R)-1. Our studies reveal a nonlinear SAR for Nek2 and our cocrystal structures show that compounds in this series bind to a DFG-out conformation of Nek2 without extending into the enlarged back pocket commonly found in this conformation. These observations were further investigated, and structure-based design led to Nek2 inhibitors derived from (R)-1 with more than a hundred-fold selectivity against Plk1.

Full text of DOI:10.1021/jm1011726

ARTICLE
pubs.acs.org/jmc
Benzimidazole Inhibitors Induce a DFG-Out Conformation of Never in
Mitosis Gene A-Related Kinase 2 (Nek2) without Binding to the Back
Pocket and Reveal a Nonlinear Structure-Activity Relationship†
Savade Solanki,^,‡ Paolo Innocenti,^,‡ Corine Mas-Droux,^,§ Kathy Boxall,‡ Caterina Barillari,‡,§
Rob L. M. van Montfort,‡,§ G. Wynne Aherne,‡ Richard Bayliss,*,§ and Swen Hoelder*,‡
‡The Institute of Cancer Research, Cancer Research UK Cancer Therapeutics Unit, 15 Cotswold Road, Sutton, Surrey SM2 5NG,
United Kingdom
§The Institute of Cancer Research, Section of Structural Biology, Chester Beatty Laboratories, 237 Fulham Road, London SW3 6JB,
United Kingdom
S Supporting Information
b
ABSTRACT: We describe herein the structure-activity relations-
hip (SAR) and cocrystal structures of a series of Nek2 inhibitors
derived from the published polo-like kinase 1 (Plk1) inhibitor (R)-
1. Our studies reveal a nonlinear SAR for Nek2 and our cocrystal
structures show that compounds in this series bind to a DFG-out
conformation of Nek2 without extending into the enlarged back
pocket commonly found in this conformation. These observations
were further investigated, and structure-based design led to Nek2
inhibitors derived from (R)-1 with more than a hundred-fold
selectivity against Plk1.
’ INTRODUCTION
decided to investigate if benzimidazole (R)-1 could serve as
starting point for improved compounds with high selectivity over
Plk1. The results of these studies are presented here.
Inducing aberrant mitosis has been explored as a cancer
therapy for many decades.1 While initial drugs predominantly
targeted microtubules, more recently, nonstructural components
of the mitotic machinery have been in the focus of drug discovery
efforts. In particular, kinases such as Plk1 or Aurora kinases have
received great attention and as a result several inhibitors have
reached clinical trials.2
A mitotic kinase that has so far not been exploited is the
serine/threonine kinase Nek2. Nek2 plays a key role in cell
division: it localizes to the centrosome and regulates spindle pole
organization and separation through phosphorylation of sub-
strates including centrosomal Nek2-associated protein 1 (C-
Nap1), rootletin, and ninein-like protein (Nlp).3 In addition to
its centrosomal role, Nek2 has also been implicated in chromatin
condensation through phosphorylation of high mobility group
AT-hook 2 (HMGA2) and spindle checkpoint control through
interaction with, or phosphorylation of, highly expressed in
cancer 1 (Hec1),4 mitotic arrest deficient-like 1 (Mad1), and
mitotic arrest deficient-like 2 (Mad2).5
Chemistry. The thiophene series was prepared according to a
previously described route.7,9,10 Nitration of ester 3 gave nitrothio-
phene 4,11 which could be easily separated from the undesired
regioisomer (Scheme 1). The benzylic ethers 5a and 5b were
prepared by means of a Mitsunobu coupling,12 while methyl ether
5c was obtained by treatment with methyl iodide and potassium
carbonate. Reduction of the nitro group functionality was performed
with iron powder in acetic acid (6a and 6b) or alternatively under
high pressure hydrogenation conditions (6c).
Coupling with the suitable aromatic bromide according to a
modified Buchwald-Hartwig procedure9b gave nitro derivatives 7,
which were easily reduced and cyclized in one pot under hydro-
genation conditions and in the presence of methyl orthoformate to
yield benzimidazoles 8. Protected hydroxyimidazoles 8a and 8b
were further elaborated by treatment with trifluoroacetic acid to
afford alcohols 9a and 9b, which were coupled with 1-Boc-4-
hydroxypiperidine under Mitsunobu conditions to give protected
amines 10a and 10b (Scheme 2). Hydrogenolysis of benzyl
derivative 10b gave thiophene phenol 10c, which was transformed
into ethers 11a and 11b using Mitsunobu conditions or into 11c
employing potassium carbonate and methyl iodide.
Even though published studies exploring Nek2 silencing
through siRNA suggest that Nek2 is a promising target,6 so far
very few inhibitors have been disclosed. Recently published
benzimidazole Plk1 inhibitor (R)-1 has been shown to have
significant activity against Nek2 (Figure 1).7 In addition, we have
recently reported aminopyrazine 2 (Figure 1) as a potent Nek2
inhibitor.8 As part of our efforts to develop Nek2 inhibitors, we
Received: September 9, 2010
Published: March 02, 2011
r
2011 American Chemical Society
1626
dx.doi.org/10.1021/jm1011726 J. Med. Chem. 2011, 54, 1626–1639
|

Journal of Medicinal Chemistry
ARTICLE
hypothesized that trimming the substituents of the benzimidazole
core of rac-1 should eventually lead to a smaller, fragment-like
molecule with reduced potency against Nek2 but significantly
improved LE. This smaller but more ligand efficient molecule
would serve as a starting point for the design of selective Nek2
inhibitors. This hypothesis was based on the published
observations19 that low molecular weight, fragment-type inhibitors
based on hinge binding motifs frequently show better LE than the
fully decorated inhibitors. This should be particularly relevant if the
substitution pattern was optimized for a different kinase.
Removal of the methyl piperidine substituent (13f, Table 1)
led to a sharp decrease in the Nek2 activity and importantly to an
even less ligand efficient inhibitor. One explanation for this result
is that the basic piperidine group engages in a strong interaction
with Nek2 that is reinforced by the observation that related Plk1
inhibitors without a suitably placed basic group did not show
Nek2 cross reactivity.18h Next, the benzyl substituent was
replaced with a methyl group (13d), again resulting in substantial
loss of activity and concomitant drop in LE.
Figure 1. Structures of known Nek2 inhibitors.
Protected amines 10a,b and 11 were transformed into the
corresponding methyl amines using a mixture of formic acid and
formaldehyde to yield thiophenes 12a,b and 12e,f (Scheme 2).
These conditions, though, proved too harsh for benzyl ether 12d,
which was cleanly cleaved leaving thiophene phenol 12c. Despite
this setback, this material was successfully employed to reintro-
duce the ether functionality by means of Mitsunobu coupling
affording 12d. Finally, esters 8c,d, 12a,b, and 12d-f were
submitted to ammonolysis in methanol to yield target thio-
phenes rac-1 and 13 (Scheme 3).
To complete the initial set of experiments, an additional
benzimidazole compound was prepared in which both substituents
were removed (13e). Interestingly, this compound showed activity
comparable to the monosubstituted derivatives 13f and 13d and an
improved LE compared to rac-1, 13f, and 13d. Despite this, the LE
of 13e is still modest (0.29) given its low molecular weight.20
These preliminary results suggest that neither of the two
substituents is capable of improving Nek2 activity of the frag-
ment compound 13e on their own but that they act together in a
cooperative manner to improve the IC50 more than three-
hundred-fold. This is also reflected in the LE trend: undecorated
benzimidazole 13e has the best LE, which substantially drops if
either of the two substituents is introduced (Table 1). Interest-
ingly, LE significantly improves if both substituents are present to
give rac-1 compared with 13f and 13d. These results thus
represent an example of nonlinear SAR around this series of
inhibitors, and it is interesting to speculate whether rac-1 would
have been identified in a classical fragment approach starting
from 13e.
To further understand the SAR around rac-1, we next
removed the substituents of the benzyl group. Again, a non-
linear relationship was observed. Removing either the benzylic
methyl group (13c) or the trifluoromethyl group (13b) caused
a pronounced loss of both activity and LE. When both
substituents were removed, resulting in benzimidazole 13a,
the effect on Nek2 inhibition was again comparable to
removing them individually (13b and 13c). Finally, a compar-
ison between compounds 13a and 13d showed that introduc-
tion of only the phenyl group caused a 12-fold increase in
activity.
The first compound of the phenyl series was assembled by
means of a Mitsunobu coupling of phenol 1413 to afford benzyl
ether 15 (Scheme 4). Reduction of the nitro group with iron
powder in acetic acid gave aniline 16, and subsequent coupling
to 4-(3-bromo-4-nitrophenoxy)-1-methylpiperidine14 using a
modified Buchwald-Hartwig reaction produced nitroderivative
17. A one-pot reduction and cyclization under hydrogenolysis
conditions yielded benzimidazole 18, which was transformed
into the target amide 19 by treatment with ammonia in
methanol.
Compounds 2416 were prepared starting from aniline 16,
which was coupled to known 1-iodo-4-(4-methoxybenzyloxy)-2-
nitrobenzene9b using a modified Buchwald-Hartwig reaction to
produce nitroderivative 20 (Scheme 5). The reduction/cycliza-
tion reaction was performed as usual under hydrogenolytic
conditions, affording benzimidazole 21. The paramethoxybenzyl
(PMB) protective group was removed using trifluoroacetic acid, and
the resulting phenol 22 was coupled under Mitsunobu conditions
with 1-Boc-4-hydroxypiperidine to yield protected amine 23a or
with cyclohexanol to afford ether 23b. Amine 23d was instead
prepared by treatment of phenol 22 with cesium carbonate and
3-dimethylamino-1-propyl chloride hydrochloride. Derivatives 23c
and 23e were also produced starting from compound 23a by
cleavage of the tert-butyl carbamate and subsequent reductive
amination with formaldehyde or (1-ethoxycyclopropoxy)trimethy-
lsilane, respectively. Finally, target amides 24 were prepared by
ammonolysis in methanol (Scheme 5).
’ RESULTS AND DISCUSSION
The benzimidazole-type Plk1 inhibitor (R)-1 (Figure 1) was
reported to inhibit Nek2 with an IC50 of 25 nM.7 In our hands,
the racemic mixture rac-1 showed an IC50 of 160 nM against
Nek2 (Table 1). We ascribed the 6-fold difference compared to
the literature value (25 nM) to differences in the assay conditions
and to the fact that we tested rac-1 as the racemate. The IC50 of
160 nM translated into a modest ligand efficiency17 (LE, 0.24)
due to the relatively high molecular weight (545 Da). The
modest efficiency of benzimidazole rac-1 is not surprising given
that it was optimized for Plk1 and not for Nek2.
As this series of compounds was inspired by Plk1 inhibitors,7
we decided to counter screen compounds rac-1 and 13 against
Plk1 to test whether a similar nonlinear effect was observed. In
agreement with published data7,10 and in contrast to the SAR
observed for Nek2, the basic group was not important for high
potency against Plk1 (compounds rac-1 and 13f, Table 1) even
though the exact IC50 values of rac-1 and 13f could not be
determined as they were beyond the dynamic range of our Plk1
assay. For this reason, it was also difficult to fully assess the
importance of the methyl and the trifluoromethyl substituents.
In agreement with data disclosed for a similar compound,7 the
trifluoromethyl group on the aromatic ring causes a significant
increase (at least 5-fold, compounds rac-1 and 13b) in Plk1
Benzimidazoles and related heteroaromatic motifs have widely
been described as kinase-targeting scaffolds that engage in hydrogen
bonds with the hinge region of the kinase protein.18 We
1627
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
Scheme 1. Synthesis of Thiophene Series Ia
a Reagents and conditions: (a) HNO3, H2SO4, -10 to 0 ꢀC; (b) ROH, di-tert-butyl azodicarboxylate, PPh3, DCM, 0 ꢀC to rt (5a and 5b); (c) MeI,
K2CO3, DMF, rt (5c); (d) Fe, AcOH, 65 ꢀC (6a and 6b); (e) H2, Pt(sulfided)/C 5%, 40 psi, MeOH/H2O (95/5), 35 ꢀC (6c); (f) ArBr, Cs2CO3,
Pd2(dba)3, XANTPHOS, 1,4-dioxane, 50-70 ꢀC; (g) H2, Pt(sulfided)/C 5%, PPTS, CH(OMe)3, EtOAc, rt.
Scheme 2. Synthesis of Thiophene Series IIa
a Reagents and conditions: (a) TFA, DCM, 0 ꢀC to rt; (b) 1-Boc-4-hydroxypiperidine, di-tert-butyl azodicarboxylate, PPh3, DCM, 0 ꢀC to rt; (c) H2, Pd/
C 10%, EtOH/H2O (95/5), rt; (d) ROH, di-tert-butyl azodicarboxylate, PPh3, DCM, 0 ꢀC to rt (11a and 11b); (e) MeI, K2CO3, DMF, rt (11c); (f) aq
HCOH, HCOOH, 85 ꢀC; (g) 1-phenylethanol, di-tert-butyl azodicarboxylate, PPh3, DCM, 0 ꢀC to rt.
activity. On the other hand, as evident from the Plk1 IC50 values
of compounds 13a and 13b, the methyl group on its own did not
significantly enhance the activity in the absence of the trifluoro-
methyl group. However, it is not possible to assess whether the
methyl group further potentiates the effect of the trifluorometh-
yl group (as is does for Nek2) because the activities of
compounds rac-1 and 13c were beyond the sensitivity of our
Plk1 assay. The authors who initially disclosed this series of
compounds encountered the same issue of assay sensitivity but
noticed that the methyl group significantly improved cellular
activity.7
Finally, it is also interesting to compare the Nek2 and Plk1
potencies of 13d and fragment-like 13e. Both molecules bind
significantly more strongly to Plk1. Comparing the Nek2 and
Plk1 activities of this series of compounds thus suggested that
(1) in striking contrast with Plk1, Nek2 requires the presence of
a basic group which acts in cooperative manner with the
substituted phenyl group, (2) when the basic group is present,
the SAR around the phenyl group is roughly parallel for Nek2
and Plk1, and (3) the significantly higher activity of rac-1 for
Plk1 largely stems from more potent binding of the core scaffold
as evident from the IC50 values of 13d and 13e.
1628
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
Scheme 3. Synthesis of Thiophene Series IIIa
a Reagents and conditions: (a) NH3, MeOH, 70-75 ꢀC.
Scheme 4. Synthesis of Compound 19a
a Reagents and conditions: (a) 1-(2-(trifluoromethyl)phenyl)ethanol, di-tert-butyl azodicarboxylate, PPh3, DCM, 0 ꢀC to rt; (b) Fe, AcOH, 65 ꢀC or H2,
Pd/C 10%, MeOH/H2O, rt;15 (c) 4-(3-bromo-4-nitrophenoxy)-1-methylpiperidine,14 Cs2CO3, Pd2(dba)3, XANTPHOS, 1,4-dioxane, 60 ꢀC; (d) H2,
Pt(sulfided)/C 5%, PPTS, CH(OMe)3, EtOAc, rt; e) NH3, MeOH, 70 ꢀC.
To understand the observed SAR and to devise a Nek2
selective inhibitor, the crystal structure of rac-1 bound to Nek2
was solved. The key interactions are shown in Figure 2. As expected
the unsubstituted benzimidazole nitrogen forms a hydrogen bond
with the hinge region (Cys89). The hydrogen atom attached to the
benzimidazole C2 carbon is sufficiently close to Glu87 to form an
aromatic hydrogen bond. The thiophene moiety is positioned
between Phe148 and the gatekeeper Met86. These observations
are in good agreement with the published docking pose for a similar
compound in Plk1.7 The Met86 side chain shows two conforma-
tions. The conformation that points toward the back pocket is
similar to the one observed in the nucleotide-bound Nek2
structures.21 The other conformation closer to the thiophene ring
is only observed in the presence of the compound and might be
induced by an attractive interaction between the sulfur atom of the
gatekeeper and the aromatic ring.22
Because an unphosphorylated form of the kinase was crystal-
lized, a substantial part of the activation loop (residues 167-
175) was disordered. However, Asp159, Phe160, and Gly161 of
the DFG motif are resolved and adopt a DFG-out conformation,
which has not previously been observed in Nek2 structures.
Interestingly, the structure revealed key differences between the
binding mode of known DFG-out inhibitors and rac-1. The
former stabilize the DFG-out conformation by placing hydro-
phobic groups in the enlarged back pocket of the kinase.23 In
addition, two hydrogen bonds, one to the carboxylate side chain
1629
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
Scheme 5. Synthesis of Phenyl Seriesa
a Reagents and conditions: (a) 1-iodo-4-(4-methoxybenzyloxy)-2-nitrobenzene, Cs2CO3, Pd2(dba)3 or Pd(dba)2, XANTPHOS, 1,4-dioxane, 60 ꢀC;
(b) H2, Pt(sulfided)/C 5%, PPTS, CH(OMe)3, EtOAc, rt; (c) TFA, DCM, 0 ꢀC to rt; (d) ROH, di-tert-butyl azodicarboxylate, PPh3, DCM, 0 ꢀC to rt
(23a and 23b); (e) 3-dimethylamino-1-propyl chloride hydrochloride, Cs2CO3, DMF, rt (23d); (f) TFA, DCM, 0 ꢀC then HCOH, AcOH, NaCNBH3,
DCM/MeOH, rt; (g) TFA, DCM, 0 ꢀC then (1-ethoxycyclopropoxy)trimethylsilane, AcOH, NaBH(AcO)3, THF, 55 ꢀC; (h) NH3, MeOH, 70-75 ꢀC.
of the catalytic Glu and one to the main chain nitrogen of the Asp
of the DFG motif, are frequently found.23 Compound rac-1 also
engages in a hydrogen bond with the NH of Asp159 of the DFG
motif but does not form a hydrogen bond to the catalytic Glu,
which is too far away due to a different position of the C-helix.
Instead the amide group of rac-1 forms an additional hydrogen
bond with the backbone carbonyl group of Asp159, thereby likely
stabilizing the DFG-out conformation. Crucially, rac-1 does not
extend into the back pocket like most DFG-out type inhibitors
(Figure 2D). This binding mode of rac-1 in the complex with
Nek2 might be applicable to other kinases, in particular those
featuring a relatively large gatekeeper (Met86 in Nek2) that can
potentially block the entrance to the back pocket.
The substituted benzyl group of rac-1 is well ordered with the
trifluoromethyl group binding to a hydrophobic pocket formed
by part of the glycine-rich loop (Ile14 and Gly15) and Cys22,
which accounts for its contribution to potency (Figure 2C).
Interestingly, one of the fluorine atoms is engaged in an
orthogonal interaction with the carbonyl group of Ile14 (3.6
Å).24 The phenyl ring is relatively solvent exposed, only forming
a few hydrophobic interactions, e.g., with Phe148. The position
of the methyl group clearly reveals the R configuration at the
benzylic stereocenter of rac-1 even though the racemic mixture
rac-1 was used for the soaking process, suggesting that this is the
more potent enantiomer for Nek2.
As mentioned above, the benzylic methyl group increases the
activity of compound rac-1 about 10-fold. Although the literature
suggests that an order of magnitude can be achieved by adding a
methyl group if it is completely buried,25 the crystal structure
indicates that this is clearly not the case for rac-1 (Figure 2C).
The significant contribution of the methyl group can be under-
stood by taking the conformation of rac-1 when bound to Nek2
into account (Figure 2). To avoid the steric interaction between
the methyl group and the trifluoromethyl group, analogous to
A
1,3 strain,26 the benzyl group of rac-1 likely adopts a conforma-
tion in the unbound state similar to the bioactive conformation
observed in the crystal structure. An analogous “cyclopropylic
strain” has recently been reported for histamine H3 receptor
antagonists.27 The contribution of the methyl group can thus at
least partially be attributed to preorganizing the substituted
phenyl group in which the trifluoromethyl group can interact
with the hydrophobic pocket as described above. We also solved
the structure of 13c, the compound lacking the methyl group,
bound to Nek2. The main difference between the cocrystal
1630
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
Table 1. Thiophene Series: Biochemical Data for Nek2 and Plk1 Assaysa
a Results are mean ((standard deviation (SD)) for n g 3, or mean values of two independent determinations with individual determinations in
parentheses or samples run n = 1; compounds are racemic unless otherwise stated.
structure of rac-1 and 13c is that the substituted benzyl group is
not well resolved in the case of 13c and thus adopts different
conformations (Figure 3). This observation supports the hypoth-
esis that the methyl group contributes through conformational
restriction. It is interesting to note that in the Nek2 structure
bound to compound 13c, the activation loop is more disordered
(residues 161-175 are not ordered) and electron density is
observed only for the backbone of Phe160. This suggests that a
more constrained compound “rigidifies” the DFG motif and the
N-terminal end of the activation loop.
1631
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
interacts with Nek2 in a manner similar to rac-1, significantly less
electron density is observed for the benzylic methyl group,
suggesting that it adopts this conformation with lower occupancy
compared to rac-1 (Figure 3C). Like in the case of compound
13c, in the structure of compound 13f bound to Nek2, the DFG
motif is less ordered than with compound rac-1 and only the
backbone of Phe160 was modeled. The DFG-out conformation
is not observed in structures of apo-Nek2 or Nek2 bound to
other ligands, such as ADP and ATPγS, suggesting that the DFG-
out conformation is induced by ligand binding (Figure 3D).
These observations are in line with the hypothesis that the
cooperative SAR is caused by structural tightening. However,
we can not rule out the possibility that the complex of rac-1 with
Nek2 is less flexible simply as a consequence of the increased
potency of rac-1 compared to 13c and 13f. The data (both in
terms of quality and limiting resolution) used to determine the
cocrystal structures of 13f bound to Nek2 was slightly poorer
than that used for the cocrystal structures of rac-1 and 13c (see
Supporting Information, Table S1). The electron density due to
the benzimidazole-thiophene core of the molecule is clear in the
maps of all three structures, indicating that the data was of
sufficient quality to reveal differences between the rigid core of
the ligands and the flexible substituents of compounds 13c and
13f (Figure 3). We attempted to investigate entropy and
enthalpy changes by isothermal titration calorimetry (ITC),
but these experiments unfortunately did not yield interpretable
data (data not shown).
Having established the SAR for Nek2 and Plk1 as well as
having studied the binding mode of rac-1, we next addressed the
Plk1 selectivity issue. Compound rac-1 is a Plk1 inhibitor with
significantly less activity against Nek2. Docking experiments
(data not shown) into the Nek2 crystal structure of rac-1
suggested that the thiophene ring could be replaced by a phenyl
ring without interrupting the key interactions with the protein.
Interestingly, several thiophene derivatives have been reported as
Plk1 inhibitors whereas the corresponding phenyl compounds
have not been explored to the same extent, suggesting that
replacement of the thiophene by a phenyl ring could be beneficial
for selectivity. Compound 19 was thus prepared and showed
Nek2 activity slightly weaker than compound rac-1, but critically
20-fold less activity against Plk1 (Table 2). While this did not
improve potency or LE for Nek2, it was a significant step toward a
selective Nek2 inhibitor. Although the crystal structure of
compound rac-1 in Plk1 was not available, the fact that Plk1
shows a preference for a thiophene-based inhibitor might be
explained by the difference in the gatekeeper residues, Leu in
Plk1 and Met in Nek2. These amino acids strongly interact with
the central aromatic moieties and may be the cause of the
observed difference in recognition.
Figure 2. Binding mode of compound rac-1 in the ATP pocket of Nek2.
(A) Active site viewed from the entrance to show the interactions
between compound rac-1 and the hinge region and the packing of the
thiophene group between the side chains of Met86 gatekeeper and
Phe148. Met86 side chain shows two conformations. (B) Interactions
between compound rac-1 and the DFG motif at Asp159. (C) The
surface of Nek2 is shown to illustrate the packing of the compound into
the active site. The following color scheme is used for atoms: nitrogen,
blue; oxygen, red; sulfur, yellow; fluorine, cyan. Compound rac-1
(carbon atoms are colored magenta) and key residues within the kinase
active site are shown as sticks. The hinge region carbon atoms are
colored purple, and the Gly-rich loop is colored green. All distances are
in Å. (D) rac-1 Does not occupy the back pocket of Nek2. Overlay of
Nek2 bound to rac-1 (red) and kinase insert domain receptor (KDR)
bound to a naphthamide inhibitor (beige, PDB 3B8Q) shows that the
large Met86 gatekeeper in Nek2 hampers the access to the back pocket
compared to the small Thr916 gatekeeper in KDR.
The suggested role of the methyl group and the observation
that the phenyl moiety is relatively solvent exposed also explains
why the latter on its own only modestly adds to activity (12 times
increase, Table 1, compounds 13a and 13d): it primarily acts in
concert with the methyl group to place the trifluoromethyl group
in the correct position to fill the small hydrophobic subpocket
(Figure 2).
The SAR data presented above underline the important role of
the piperidine moiety. The basic group is well resolved in the crystal
structure, but despite its crucial role for Nek2 activity, it does not
seem to engage in any charge-assisted hydrogen bonds. However,
charge-charge interactions can be long ranged and the piperidine
ring is in relatively close proximity to Asp 93 (4.6 Å), potentially
explaining why it contributes to activity (Figure 4).
Despite these extensive studies, neither the crystal structures
nor the biological data offer a conclusive explanation for the
origin of the nonlinear SAR in Nek2 observed for the compounds
described. Examples of positive cooperative effects have recently
been described28 and are frequently explained by structural
tightening when additional interactions are introduced,24 making
the interactions more enthalpically favorable. In this context, it is
noteworthy that we also solved the cocrystal structure of 13f,
which does not have the piperidine group and is much less potent
(Table 1). This structure showed that, while the benzimidazole
Despite the improvement in the ratio between the Nek2 and
Plk1 IC50 values (Table 2), compound 19 still represented an
approximately equipotent inhibitor of these kinases. Comparison
of the crystal structure of rac-1 in Nek2 with a published Plk1 crystal
structure suggested a strategy to further improve selectivity: moving
the basic group from the 6-position of the benzimidazole to the
5-position should lead to a clash with Arg136 of Plk1 while it should
be tolerated in Nek2, which features a glycine (Gly92) at this
position. This assumption was supported by published SAR data
showing that substitution of the 5-position of the benzimidazole
with relatively bulky substituents such as trifluoromethyl or methyl-
sulfonyl groups led a significant drop in Plk1 activity.10 Accordingly,
compound rac-24a was prepared and gratifyingly not only showed
1632
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
Figure 3. Structural biology data supports the structural tightening hypothesis and shows that the conformation of the activation loop of Nek2 is ligand
dependent. (A-C) Upper images: final refined electron density map (2mFo - DFc) at 1 σ around compounds rac-1 (2A), 13c (2B), and 13f (2C).
Lower images: initial electron density maps generated in the absence of ligand and before refinement (gray: 2Fo - Fc at 1 σ, green: Fo - Fc at 2.5 σ). (D)
Superposition of Nek2 protein structures to show the variability in DFG motif conformation dependent on the bound ligand: apo (wheat); compound
rac-1 (gray), ADP (yellow), ATPγS (cyan). For clarity, regions outside the DFG motif and ligand are shown only for Nek2 bound to compound rac-1.
Figure 4. Structural basis of the enhanced selectivity against Plk1 achieved with compound rac-24a. (A) Comparison of the piperidine ring position for
compound rac-1 and rac-24a in Nek2. (B) Overlay of Nek2 structure, compounds rac-1, rac-24a, and two inhibitor-bound Plk1 structures. Residue
Arg136 in Plk1 can adopt different conformations, but both will lead to a steric clash with the piperidine ring of compound rac-24a. Nek2 (carbon atoms
are colored gray and purple for hinge region), Plk1 (carbon atoms are colored beige (3FC2), and green (3DB8). Carbon atoms are colored magenta in
the case of rac-1 and lavender in the case of rac-24a.
equipotent activity on Nek2 but negligible activity on Plk1
(Table 2).
Nek2 and Plk1 inhibition. As expected, (R)-24a was found to be
the most active of the pair with an IC50 of 0.36 μM (rac-24a IC50
= 0.57 μM) while (S)-24a was shown to be a relatively weak
inhibitor of Nek2 with an IC50 of 4.86 μM. The loss of activity of
(S)-24a, which is more than an order of magnitude if compared
with its enantiomer, underlines the importance of the configura-
tion at the benzylic stereocenter.
An interesting difference between the structure of thiophene
rac-1 and rac-24a bound to Nek2 (Figure 4) is that the piperidine
ring was shifted by 4.3 Å toward the back of the catalytic pocket
and was positioned above the Gly91-Gly92 residues of the
hinge region. In both examples no charge-assisted hydrogen
We solved the cocrystal structure of compound rac-24a bound
to Nek2. As expected, the binding modes of rac-1 and rac-24a
have many similarities (Figure 4). It is interesting to point out
that the absolute configuration at the benzylic stereocenter of
rac-24a bound to Nek2 is again R, as observed in the case of
thiophene rac-1 (Figure 4). This observation reinforces the
hypothesis that the R configured compounds in these two series
are responsible for most of the enzyme inhibition.7 To further
investigate this matter, the two enantiomers were prepared
starting from optically pure benzylic alcohols and assayed for
1633
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
Table 2. Phenyl Series: Biochemical Data for Nek2 and Plk1 Assaysa
a Results are mean ((SD) for n g 3, or mean values of two independent determinations with individual determinations in parentheses or samples run n =
1; compounds are racemic unless otherwise stated.
bond can be observed, suggesting that long-range charge-charge
interactions are responsible for the contributions of the piper-
idine ring.
To further investigate the role of this basic residue, and to
elucidate the remarkable impact it has in conjunction with the
presence of the fully substituted benzylic group (see Table 1,
1634
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
While the crystal structures did not conclusively explain the
observed nonlinear SAR, the fact that rac-1 and rac-24a bind to a
DFG-out conformation of Nek2 was intriguing. The DFG-out
conformation is in many cases adopted by the unphosphorylated,
inactive form of the kinase.23 This is consistent with the fact that
we crystallized the unphosphorylated form of the enzyme. Nek2
exists as a dimer and rapidly trans-autophosphorylates on
Thr175 in the presence of ATP (Figure 5). The observation
that the inhibitors described here bind to the inactive conforma-
tion suggests that they should be capable of preventing Nek2
activation through trans-autophosphorylation (Table 3). The
assay used for testing the inhibitors so far employed the nonpho-
sphorylated kinase which is preincubated with the inhibitor
before autophosphorylation and subsequent substrate phosphor-
ylation was triggered by adding ATP (Figure 5). While this assay
was capable of identifying inhibitors of auto- and substrate
phosphorylation, it was interesting to discern to which extent
each of these events was inhibited.
Figure 5. Auto- and substrate phosphorylation cascade: addition of
ATP to the inactive Nek2 dimer initiates trans-autophosphorylation on
Thr175 and activation of the protein. Active Nek2 then phosphorylates
the substrate.
Table 3. Nek2 Autophosphorylation and Substrate Phos-
phorylation Dataa
To answer this question, a specific autophosphorylation assay
was developed employing an antibody specific for Nek2 phos-
phorylated on Thr175. Interestingly, rac-1 and rac-24a showed
IC50 values of 50 and 230 nM, respectively, in this autopho-
sphorylation assay, which is in agreement with the observation
that they bind to the DFG-out conformation. To address the
question of whether rac-1 and rac-24a also inhibit substrate
phosphorylation, the original assay conditions were slightly
altered. The enzyme was first incubated with ATP to ensure
complete autophosphorylation followed by incubation with the
inhibitor and addition of the substrate. Under these conditions,
rac-1 and rac-24a showed an IC50 comparable to the initially
employed conditions (0.20 and 0.91 μM, respectively), suggest-
ing that these compounds can also inhibit the phosphorylated
kinase. The ability of rac-1 and rac-24a to inhibit Nek2 regardless
of its phosphorylation state is highly desiderable.
Our results so far had identified compound 24a as the most
attractive Nek2 inhibitor in this series with good potency and
selectivity versus Plk1, and we decided to further profile it. We
tested rac-24a against a panel of 24 kinases. While the overall
selectivity profile was reasonable (Gini coefficient for rac-24a
was 0.534, see Table S2 in Supporting Information for details), a
few tyrosine kinases like Abelson murine leukemia viral onco-
gene (Abl) or lymphocyte-specific protein tyrosine kinase (Lck)
showed significant inhibition at 2 μM. Next, we tested (R)-24a in
our cell-based assay, which monitors centrosome splitting.30
Disappointingly, benzimidazole (R)-24a failed to have any
significant effect at concentrations below 10 μM while it was
found to have cytotoxic effects at higher concentrations (data not
shown). We attributed the lack of activity of (R)-24a in this
cellular assay below 10 μM at least partially to the significantly
higher ATP concentration in cells compared to the biochemical
assay. The cytotoxic effects at higher concentrations can be
explained by inhibition of other kinases as observed from the
kinase profiling mentioned above. Finally, (R)-24a showed very
good permeability at physiological pH in a parallel artificial
membrane permeability assay (PAMPA) (33.4 ꢀ 10-6 cm/s)
and moderate metabolic degradation (see Table S3 in Support-
ing Information for details).
a Results are mean ((SD) for n g 3, or mean values of two independent
determinations with individual determinations in parentheses; com-
pounds are racemic unless otherwise stated.
compounds rac-1 and 13f), a few derivatives of rac-24a in which
the basic group is altered were prepared (Table 2): the nonbasic
cyclohexyl analogue 24b, the open chain amine 24c, and
cyclopropylamine 24d. While 24b showed very weak inhibition,
24c showed activity comparable to rac-24a, suggesting that the
basic nitrogen is critical while the exact shape of the amine is not
important. The crucial role of the basic nitrogen is further
supported by the fact that the less basic cyclopropylamine is
about 5-fold less active. Literature data29 suggest that the pKa of
piperidine rac-24a is in the order of 9, whereas the pKa of
cyclopropylamine 24d is around 7.5. Compound 24d should
thus be protonated to a significantly lesser degree, possibly
leading to the loss in activity. Taken together, this set of data
suggests that the basicity of the compounds studied here is crucial
to achieve high levels of activity against Nek2 and that a charge
related interaction is responsible for the contribution of the basic
group.
’ CONCLUSIONS
In summary, we designed a potent Nek2 inhibitor character-
ized by more than hundred-fold selectivity against Plk1 (rac-
24a). This and related compounds in the series showed an
1635
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
unusual binding mode to the DFG-out conformation of Nek2. In
addition, both rac-1 and rac-24a displayed potent inhibition of
substrate- and autophosphorylation. Given that very few inhibi-
tors of Nek2 have been published so far, this represents a
significant step forward. However, our data, in particular the lack
of cellular activity, showed that improvement of activity and
selectivity are still required. We believe that the pharmacophore
defined by the cocrystal structures and SAR described here will
be valuable for the identification of such improved inhibitors. A
limitation for the further optimization of compound (R)-24a is
the still relatively moderate LE (Table 2). In this context, it is
noteworthy that 13e, which represents the core scaffold and
hinge binding element of (R)-24a, also shows only modest LE
(Table 1). This observation suggests that exploring the pharma-
cophore described here with more efficient hinge binders will
yield improved inhibitors with better optimization potential.
Studies toward this end are underway.
Furthermore, the body of data discussed herein represents an
interesting example of nonlinear SAR and is particularly relevant
in the context of fragment-based approaches. Compounds 1 and
24a might not have been identified through independent opti-
mization of the substituents of the hinge binding fragment 13e
because the singly substituted inhibitors (compounds 13d and
13f, Table 1) are very weak inhibitors with poor LE. This
underscores the importance of combinatorial exploration, even
though chemically more demanding, in fragment optimization.
The purity of final compounds was determined by HPLC as described
above and is g95% unless specified otherwise.
(()-Methyl 4-Nitro-2-(1-(2-(trifluoromethyl)phenyl)etho-
xy)benzoate (rac-15).
A solution of (()-1-(2-(trifluoromethyl)phenyl)ethanol (1.03 g, 5.43
mmol), methyl 2-hydroxy-4-nitrobenzoate13 14 (765 mg, 3.88 mmol), and
triphenylphosphine (1.43 g, 5.43 mmol) in DCM (30 mL) was cooled at
0 ꢀC and treated with di-tert-butyl azodicarboxylate (1.25 g, 5.43 mmol).
The mixture was allowed to reach room temperature and stirred overnight.
The reaction was quenched with water and extracted with DCM. The
organics were dried (MgSO4), concentrated, and purified by Biotage
column chromatography (0-25% EtOAc/hexane) to provide ether rac-
15 (1.47 g, quant). 1H NMR (500 MHz, CDCl3) δ7.85 (dd, J= 8.1, 5.5 Hz,
2H), 7.79-7.67 (m, 3H), 7.56 (t, J = 7.6 Hz, 1H), 7.40 (t, J = 7.6 Hz, 1H),
5.89 (q, J = 6.2 Hz, 1H), 4.01 (s, 3H), 1.75 (d, J = 6.2 Hz, 3H).
(()-Methyl 4-Amino-2-(1-(2-(trifluoromethyl)phenyl)eth-
oxy)benzoate (rac-16).
’ EXPERIMENTAL SECTION
A solution of nitrobenzene rac-15 (1.47 g, 3.98 mmol) in AcOH (12
mL) was heated at 65 ꢀC for 45 min in the presence of iron powder (1.11
g, 19.88 mmol). The reaction was diluted with DCM (22 mL) and
quenched with 6N NaOH (33 mL) and satd aqueous NaHCO3 (10
mL). The resulting thick solution was filtrated over Celite washing with
DCM and satd NaHCO3. The organic phase was separated, dried
(MgSO4), and concentrated to afford aniline rac-16 (830 mg, 62%).
HRMS (ESI) m/z calcd for C17H17F3NO3 (M þ H) 340.1155, found
340.1153. 1H NMR (500 MHz, CDCl3) δ 7.96 (d, J = 7.9 Hz, 1H), 7.71
(d, J = 8.5 Hz, 1H), 7.67-7.52 (m, 2H), 7.37 (q, J = 7.7 Hz, 1H), 6.17
(dd, J = 8.5, 2.1 Hz, 1H), 6.08 (d, J = 2.1 Hz, 1H), 5.75 (q, J = 6.1 Hz,
1H), 3.89 (s, 3H), 1.69 (d, J = 6.1 Hz, 2H).
General Chemistry Information. Starting materials, reagents,
and solvents for reactions were reagent grade and used as purchased.
Chromatography solvents were HPLC grade and were used without
further purification. Thin layer chromatography (TLC) analysis was
performed using Merck silica gel 60 F-254 thin layer plates. Flash
column chromatography was carried out using columns prepacked with
40-63 μm silica or using standard chromatography techniques when
performed on a large scale. NMR spectra were recorded on a Bruker
Advance 500 MHz spectrometer, and samples were referenced to the
appropriate internal nondeuterated solvent peak. In the case where a 1:1
mixture of MeOD and CDCl3 was used, the samples were referenced to
MeOD. The data is given as follows: chemical shift (δ) in ppm,
multiplicity (where applicable), coupling constants (J) in Hz (where
applicable), and integration (where applicable). LCMS analyses were
performed on a Micromass LCT/Waters Alliance 2795 separations
module HPLC system with a Merck Chromolith SpeedROD RP-18e 50
mm ꢀ 4.6 mm column at a temperature of 22 ꢀC. The following solvent
system, at a flow rate of 2 mL/min, was used: solvent A, methanol;
solvent B, 0.1% formic acid in water. Gradient elution was as follows: 1:9
(A:B) to 9:1 (A:B) over 2.25 min, 9:1 (A:B) for 0.75 min then reversion
back to 1:9 (A:B) over 0.3 min, 1:9 (A:B) for 0.2 min. Detection was
with a Waters 2487 Dual λ absorbance detector (detecting at 254 nm)
and ionization was electrospray (ESI). Some LCMS and all HRMS
analyses were performed on a Agilent 1200 series HPLC system with a
Merck Chromolith SpeedROD RP-18e 50 mm ꢀ 4.6 mm column at a
temperature of 22 ꢀC. The following solvent system, at a flow rate of 2 mL/
min, was used: solvent A, methanol; solvent B, 0.1% formic acid in water.
Gradient elution was as follows: 1:9 (A:B) to 9:1 (A:B) over 2.5 min, 9:1
(A:B) for 1 min then reversion back to 1:9 (A:B) over 0.3 min, 1:9 (A:B) for
0.2 min. This was connected to a Agilent 6200 time of flight (ToF) mass
spectrometer (simultaneous ESI and APCI or ESI only) with detection at
254 nm. The following reference masses were used for HRMS analysis:
caffeine [M þ H]þ = 195.087652, reserpine [M þ H]þ = 609.280657, and
(1H,1H,3H-tetrafluoropentoxy)phosphazene [M þ H]þ = 922.009798.
(()-Methyl 4-(4-(4-Methoxybenzyloxy)-2-nitro-
phenylamino)-2-(1-(2-(trifluoromethyl)phenyl)ethoxy)-
benzoate (rac-20).
A solution of aniline rac-16 (538 mg, 1.59 mmol), 1-iodo-4-(4-meth-
oxybenzyloxy)-2-nitrobenzene9b (582 mg, 1.51 mmol), tris(dibenzyli-
deneacetone)dipalladium(0) (28 mg, 0.027 mmol), XANTPHOS (46
mg, 0.08 mmol), and cesium carbonate (2.50 g, 7.67 mmol) in 1,4-dioxane
(8 mL, degassed) was heated to 60 ꢀC overnight. The mixture was filtered
over Celite washing with Et2O, the solvent removed under reduced
pressure, and the residue purified by Biotage column chromatography
(0-50% EtOAc/cyclohexane) to provide aniline rac-20 (925 mg, quant).
HRMS (ESI) m/z calcd for C31H28F3N2O7 (M þ H) 597.1843, found
597.1852. 1H NMR (500 MHz, CDCl3) δ 9.02 (s, 1H), 7.92 (d, J = 7.9 Hz,
1H), 7.83 (d, J = 8.4 Hz, 1H), 7.70 (t, J = 1.6 Hz, 1H), 7.65 (d, J = 7.9 Hz,
1H), 7.58 (t, J = 7.6 Hz, 1H), 7.42-7.36 (m, 3H), 7.01-6.93 (m, 4H), 6.69
1636
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
(dd, J = 8.5, 2.0 Hz, 1H), 6.60 (d, J = 2.0 Hz, 1H), 5.72 (q, J = 6.1 Hz, 1H),
5.01 (s, 2H), 3.95 (s, 3H), 3.84 (s, 3H), 1.71 (d, J = 6.1 Hz, 3H).
(()-Methyl 4-(5-(4-Methoxybenzyloxy)-1H-benzo[d]imid-
azol-1-yl)-2-(1-(2-(trifluoromethyl)phenyl)ethoxy)benzoate
(rac-21).
mg, 1.28 mmol) in DCM (2.5 mL) was cooled at 0 ꢀC and treated with
di-tert-butyl azodicarboxylate (147 mg, 0.64 mmol). The reaction was
allowed to reach room temperature and stirred overnight. The mixture
was quenched with water and extracted with DCM twice. The organics
were dried (Na2SO4), concentrated, and purified by Biotage column
chromatography (0-60% EtOAc/cyclohexane) to give benzimidazole
rac-23a (205 mg, containing traces of triphenylphosphine oxide, quant).
LCMS (ESI) m/z 640 (M þ H). 1H NMR (500 MHz, CDCl3) δ 7.97-
7.94 (m, 1H), 7.91-7.87 (m, 2H), 7.59-7.55 (m, 1H), 7.53 (dd, J = 2.8,
1.4 Hz, 1H), 7.42-7.38 (m, 1H), 7.31 (d, J = 2.2 Hz, 1H), 7.03 (dd, J =
8.3, 1.9 Hz, 1H), 6.91-6.82 (m, 3H), 5.80 (q, J = 6.1 Hz, 1H), 4.47 (tt, J
= 7.1, 3.5 Hz, 1H), 3.98 (s, 3H), 3.88-3.68 (m, 2H), 3.37-3.27 (m,
2H), 1.98-1.89 (m, 2H), 1.83-1.71 (m, 5H), 1.46 (s, 9H).
(()-Methyl 4-(5-(1-Methylpiperidin-4-yloxy)-1H-benzo-
[d]imidazol-1-yl)-2-(1-(2-(trifluoromethyl)phenyl)ethoxy)-
benzoate (rac-23c).
A solution of amine rac-20 (925 mg, 1.55 mmol) in EtOAc (13 mL)
was treated with platinum sulfided on charcoal (5% w/w, 574 mg, 0.15
mmol), PPTS (78 mg, 0.31 mmol), and methyl orthoformate (1.67 mL,
15.2 mmol). The mixture was stirred in an atmosphere of hydrogen for 20 h.
The mixture was filtered over Celite washing with EtOAc, the solvent
removed under reduced pressure, and the residue purified by Biotage
column chromatography (0-100% EtOAc/cyclohexane) to give benzimi-
dazole rac-21 (630 mg, 71%). HRMS (ESI) m/z calcd for C32H28F3N2O5
(M þ H) 577.1945, found 577.1971. 1H NMR (500 MHz, CDCl3) δ 7.97
(d, J = 8.3 Hz, 1H), 7.93-7.90 (m, 2H), 7.69 (d, J = 7.8 Hz, 1H), 7.59 (t, J =
7.6 Hz, 1H), 7.45-7.38 (m, 3H), 7.36 (d, J = 2.3 Hz, 1H), 7.04 (dd, J = 8.3,
1.9 Hz, 1H), 6.95-6.85 (m, 5H), 5.81 (q, J= 6.1 Hz, 1H), 5.05 (s, 2H), 4.00
(s, 3H), 3.81 (s, 3H), 1.77 (d, J = 6.1 Hz, 3H).
(()-Methyl 4-(5-Hydroxy-1H-benzo[d]imidazol-1-yl)-2-(1-
(2-(trifluoromethyl)phenyl)ethoxy)benzoate (rac-22).
A solution of benzimidazole rac-23a (205 mg, 0.32 mmol) in DCM
(3.4 mL) was treated with trifluoroacetic acid (1.14 mL, 14.7 mmol) at
0 ꢀC. The mixture stirred for 1 h and quenched carefully with aq NaOH
solution (2M, 7 mL). To this was added aq sodium bicarbonate solution,
and the mixture was extracted with DCM and Et2O. The combined
organic layers were concentrated to afford the crude, unprotected amine
rac-25 (170 mg, 98%).
A solution of crude amine rac-25 (87 mg, 0.161 mmol) in DCM (1.8
mL) and MeOH (0.9 mL) was treated with formaldehyde in water (37%
w/w, 26 μL, 0.32 mmol) and AcOH (0.011 mL, 0.193 mmol). To this
mixture was added sodium triacetoxyborohydride (51 mg, 0.242 mmol),
and the reaction mixture stirred for 1 h at room temperature. The
reaction was quenched with satd aqueous sodium bicarbonate and the
mixture extracted with DCM and EtOAc. The organics were dried
(Na2SO4) and concentrated to afford benzimidazole rac-23c (39 mg,
47%). HRMS (ESI) m/z calcd for C30H31F3N3O4 (M þ H) 554.2261,
A solution of benzimidazole rac-21 (630 mg, 1.09 mmol) in DCM (6
mL) was treated with trifluoroacetic acid (803 μL, 10.9 mmol) at 0 ꢀC.
The mixture was allowed to reach room temperature and stirred for 3 h. The
reaction was brought to pH 5-6 with 1 M NaOH and 1 M HCl, and the
aqueous layer was separated and extracted with DCM. The combined
organic layers were concentrated, and the residue was purified by Biotage
column chromatography (0-60% EtOAc/cyclohexane) to afford benzimi-
dazole rac-22 (470 mg, 94%). HRMS (ESI) m/z calcd for C24H20F3N2O4
(M þ H) 457.1370, found 457.1381. 1H NMR (500 MHz, CDCl3) δ 9.13
(br s, 1H), 7.96-7.89 (m, 3H), 7.66 (d, J = 7.8 Hz, 1H), 7.58 (t, J = 7.6 Hz,
1H), 7.42-7.39 (m, 2H), 7.02 (dd, J = 8.3, 1.9 Hz, 1H), 6.91 (d, J = 1.9 Hz,
1H), 6.86 (dd, J = 8.8, 2.3 Hz, 1H), 6.77 (d, J = 1.9 Hz, 1H), 5.81 (q, J = 6.1
Hz, 1H), 4.00 (s, 3H), 1.76 (d, J = 6.1 Hz, 3H).
1
found 554.2272. H NMR (500 MHz, CDCl3) δ 7.96 (d, J = 8.3 Hz,
1H), 7.92-7.89 (m, 2H), 7.70 (d, J = 7.9 Hz, 1H), 7.59 (t, J = 7.6 Hz,
1H), 7.43 (t, J = 7.6 Hz, 1H), 7.31 (d, J = 2.2 Hz, 1H), 7.05 (dd, J = 8.3,
2.2 Hz, 1H), 6.93-6.84 (m, 3H), 5.81 (d, J = 6.2 Hz, 1H), 4.40-4.33
(m, 1H), 4.00 (s, 3H), 2.85-2.72 (m, 2H), 2.39 (d, J = 19.9 Hz, 5H),
2.15-2.06 (m, 2H), 1.97-1.86 (m, 2H), 1.76 (d, J = 6.2 Hz, 3H).
(()-4-(5-(1-Methylpiperidin-4-yloxy)-1H-benzo[d]imidazol-1-
yl)-2-(1-(2-(trifluoromethyl)phenyl)ethoxy)benzamide (rac-
24a).
(()-tert-Butyl 4-(1-(4-(Methoxycarbonyl)-3-(1-(2-(trifluoro-
methyl)phenyl)ethoxy)phenyl)-1H-benzo[d]imidazol-5-yloxy)-
piperidine-1-carboxylate (rac-23a).
A solution of ester rac-23c (20 mg, 0.036 mmol) was treated with
ammonia in methanol (7M, 4 mL) and heated to 70 ꢀC in a closed-cap
vial for 3 days. The mixture was concentrated, and the residue was
A solution of benzimidazole rac-22 (146 mg, 0.32 mmol), 1-Boc-4-
hydroxypiperidine (129 mg, 0.64 mmol), and triphenylphosphine (336
1637
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
purified by Biotage column chromatography (0-100% (DCM/1 M
NH3 in MeOH 9:1)/DCM) to afford amide rac-24a (18 mg 93%).
HRMS (ESI) m/z calcd for C29H30F3N4O3 (M þ H) 539.2265, found
539.2281. 1H NMR (500 MHz, MeOD) δ 8.23 (s, 1H), 8.11 (d, J = 8.3
Hz, 1H), 7.82 (t, J = 7.0 Hz, 2H), 7.70 (t, J = 7.6 Hz, 1H), 7.58 (t, J = 7.6
Hz, 1H), 7.26-7.22 (m, 2H), 6.97 (d, J = 1.9 Hz, 1H), 6.84 (dd, J = 8.9,
2.3 Hz, 1H), 6.66 (d, J = 8.9 Hz, 1H), 5.96 (q, J = 6.1 Hz, 1H), 4.49-4.41
(m, 1H), 2.82-2.74 (m, 2H), 2.49-2.38 (m, 2H), 2.35 (s, 3H), 2.11-
2.02 (m, 2H), 1.90-1.80 (m, 5H).
microsomes; HMGA2, high mobility group AT-hook 2; ITC,
isothermal titration calorimetry; Lck, lymphocyte-specific pro-
tein tyrosine kinase; LE, ligand efficiency; KDR, kinase insert
domain receptor; Mad, mitotic arrest deficient-like; MLM,
mouse liver microsomes; Nek, NIMA (never in mitosis gene
a)-related kinase; Nlp, ninein-like protein; Plk, polo-like kinase;
PAMPA, parallel artificial membrane permeability assay; PMB,
paramethoxybenzyl; PPTS, pyridinium p-toluenesulfonate; SAR,
structure-activity relationship; SD, standard deviation; TFA,
trifluoroacetic acid; THF, tetrahydrofuran
’ ASSOCIATED CONTENT
’ REFERENCES
S
Supporting Information. Experimental procedures, ana-
b
(1) Jackson, J. R.; Patrick, D. R.; Dar, M. M.; Huang, P. S. Targeted
antimitotic therapies: can we improve on tubulin agents?. Nature Rev.
Cancer 2007, 7, 107–117.
(2) (a) Degenhardt, Y.; Lampkin, T. Targeting Polo-like kinase in
cancer therapy. Clin. Cancer Res. 2010, 16, 384–389. (b) Carpinelli, P.;
Moll, J. Is there a future for Aurora kinase inhibitors for anticancer
therapy? Curr Opin. Drug Discovery Dev. 2009, 12, 533–542.
(3) O’Regan, L.; Blot, J.; Fry, A. M. Mitotic regulation by NIMA-
related kinases. Cell Div. 2007, 2, 25–36.
(4) Chen, Y.; Riley, D. J.; Zheng, L.; Chen, P. L.; Lee, W. H.
Phosphorylation of the mitotic regulator protein Hec1 by Nek2 kinase
is essential for faithful chromosome segregation. J. Biol. Chem. 2002, 277,
49408–49416.
lytical data for intermediates and final compounds rac-1, 13a-f,
19, (R)-24a, (S)-24a, 24b-d, a summary of crystallographic
analysis of compounds rac-1, 13f, 13c, and rac-24a, kinase
selectivity data for compound rac-24a, and metabolic data for
rac-24a and (R)-24a. This material is available free of charge via
the Internet at http://pubs.acs.org.
Accession Codes
†Atomic coordinates and structure factors for the crystal struc-
tures of ligand bound Nek2 can be accessed using the following
PDB codes: rac-1 (2XNM), 13f (2XNN), 13c (2XNO), rac-24a
(2XNP).
(5) Lou, Y.; Yao, J.; Zereshki, A.; Dou, Z.; Ahmed, K.; Wang, H.; Hu,
J.; Wang, Y.; Yao, X. NEK2A interacts with MAD1 and possibly
functions as a novel integrator of the spindle checkpoint signaling. J.
Biol. Chem. 2004, 279, 20049–20057.
(6) Hayward, D. G.; Fry, A. M. Nek2 kinase in chromosome
instability and cancer. Cancer Lett. 2006, 237, 155–166.
(7) Emmitte, K. A.; Adjebang, G. M.; Andrews, C. W.; Alberti, J. G.;
Bambal, R.; Chamberlain, S. D.; Davis-Ward, R. G.; Dickson, H. D.;
Hassler, D. F.; Hornberger, K. R.; Jackson, J. R.; Kuntz, K. W.; Lansing,
T. J.; Mook, R. A., Jr.; Nailor, K. E.; Pobanz, M. A.; Smith, S. C.; Sung,
C. M.; Cheung, M. Design of potent thiophene inhibitors of polo-like
kinase 1 with improved solubility and reduced protein binding. Bioorg.
Med. Chem. Lett. 2009, 19, 1694–1697.
’ AUTHOR INFORMATION
Corresponding Author
*For R. B.: phone, þ44 20 71535557; e-mail, richard.bayliss@icr.
ac.uk. For S.H.: phone, þ44 20 87224353; fax, þ44 20
87224047; e-mail, swen.hoelder@icr.ac.uk.
Author Contributions
^These authors contributed equally to this work.
’ ACKNOWLEDGMENT
(8) Whelligan, D. K.; Solanki, S.; Taylor, D.; Thomson, D. W.;
Cheung, K. M. J.; Boxall, K.; Mas-Droux, C.; Barillari, C.; Burns, S.;
Grummitt, C. G.; Collins, I.; van Montfort, R. L. M.; Aherne, G. W.;
Bayliss, R.; Hoelder, S. Aminopyrazine inhibitors binding to an unusual
inactive conformation of the mitotic kinase Nek2: SAR and structural
characterization. J. Med. Chem. 2010, 53, 7682–7698.
(9) (a) Hornberger, K.; Cheung, M.; Pobanz, M. A.; Emmitte, K. A.;
Kuntz, K. W.; Badiang, J. G. Regioselective process for preparing
benzimidazole thiophenes. (Smithkline Beecham Corporation). Patent
WO2007030366, 2007; (b) Hornberger, K. R.; Badiang, J. G.; Salovich,
J. M.; Kuntz, K. W.; Emmitte, K. A.; Cheung, M. Regioselective synthesis
of benzimidazole thiophene inhibitors of polo-like kinase 1. Tetrahedron
Lett. 2008, 49, 6348–6351. (c) Hornberger, K. R.; Adjebang, G. M.;
Dickson, H. D.; Davis-Ward, R. G. A mild, one-pot synthesis of
disubstituted benzimidazoles from 2-nitroanilines. Tetrahedron Lett.
2006, 47, 5359–5361.
(10) Emmitte, K. A.; Andrews, C. W.; Badiang, J. G.; Davis-Ward,
R. G.; Dickson, H. D.; Drewry, D. H.; Emerson, H. K.; Epperly, A. H.;
Hassler, D. F.; Knick, V. B.; Kuntz, K. W.; Lansing, T. J.; Linn, J. A.;
Mook, R. A., Jr.; Nailor, K. E.; Salovich, J. M.; Spehar, G. M.; Cheung, M.
Discovery of thiophene inhibitors of polo-like kinase. Bioorg. Med. Chem.
Lett. 2009, 19, 1018–1021.
(11) (a) Chao, J.; Taveras, A. G.; Aki, C. J. Synthesis of functiona-
lized hydroxy-thiophene motifs as amido- and sulfonamido-phenol
bioisosteres. Tetrahedron Lett. 2009, 50, 5005–5008. (b) Barker, J. M.;
Huddleston, P. R.; Wood, M. L.; Burkitt, S. A. Nitration of methyl-3-
hydroxy- and 5-methyl-3-hydroxy-thiophene-2-carboxylate, and some
chemistry of the products. J. Chem. Res. 2001, (S), 401–402.
We acknowledge NHS funding to the NIHR Biomedical
Research Centre and funding from Cancer Research UK pro-
gramme grant number C309/A8274. R.B. acknowledges support
from Cancer Research UK (grant C24461/A10285 and infra-
structure support for Structural Biology at the ICR), a Royal
Society Research Fellowship and the Career Development
Programme of the ICR. We are thankful to Prof. Julian Blagg,
Prof. Roger Griffin, Dr. Ian Collins, and Prof. Paul Workman for
helpful discussions. We also thank Dr. Amin Mirza, Meirion
Richards, and Dr. Maggie Liu for their help with HPLC, NMR,
and mass spectrometry, Angela Hayes for DMPK analyses, and
Maura Westlake and Dr. Lisa Pickard for cellular assays. We are
indebted to the staff of DIAMOND beamline I03 and ESRF
beamline ID14-2 as well as our many colleagues of the Section of
Structural Biology (ICR in London) for their support during data
collection.
’ ABBREVIATIONS USED
Abl, Abelson murine leukemia viral oncogene; ADP, adenosine
diphosphate; ATP, adenosine triphosphate; Boc, tert-butoxycar-
bonyl; C-Nap1, centrosomal Nek2-associated protein 1; dba,
dibenzylideneacetone; DCM, dichloromethane; DMF, N,N-di-
methylformamide; DFG, Asp-Phe-Gly; Hec1, highly expressed
in cancer 1; DMSO, dimethyl sulfoxide; HLM, human liver
1638
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639

Journal of Medicinal Chemistry
ARTICLE
(12) Mitsunobu, O. The use of diethyl azodicarboxylate and triphe-
nylphosphine in synthesis and transformation of natural-products.
Synthesis 1981, 1–28.
(13) Agback, K. H. Azo-bis-salicylic acid and salt thereof, to treat
inflammatory conditions of the intestine. (Pharmacia AB). U.S. Patent
US4591584, 1986.
(14) Prepared using 1-methylpiperidin-4-ol, 2-bromo-4-fluoro-1-ni-
trobenzene and sodium hydride in THF at room temperature.
(15) Anilines (R)-16 and (S)-16 were obtained by catalytic reduc-
tion under hydrogen atmosphere from the corresponding enantiopure
nitroderivatives (see Supporting Information for details).
(16) Compounds rac-24a, (R)-24a and (S)-24a were prepared
using the same route described in Schemes 4 and 5. The two enantio-
mers were obtained from commercially available enantiopure (S)- or
(R)-1-(2-(trifluoromethyl)phenyl)ethanol respectively (purchased
from Apollo Scientific, Stockport, UK).
Zhong, B.; Mendoza, J.; Collins, E.; Jaramillo, C.; De Diego, J. E.;
Robertson, D.; Spencer, C. D.; Anderson, B. D.; Watkins, S. A.; Zhang,
F.; Brooks, H. B. Structure-based design of a new class of highly selective
aminoimidazo[1,2-a]pyridine-based inhibitors of cyclin dependent ki-
nases. Bioorg. Med. Chem. Lett. 2005, 15, 1943–1947.
(19) (a) Bembenek, S. D.; Tounge, B. A.; Reynolds, C. H. Ligand
efficiency and fragment-based drug discovery. Drug Discovery Today
2009, 14, 278–283. (b) Ferenczy, G. G.; Keseru, G. M. Enthalpic
efficiency of ligand binding. J. Chem. Inf. Model 2010, 50, 1536–1541.
(20) Reynolds, C. H.; Bembenek, S. D.; Tounge, B. A. The role of
molecular size in ligand efficiency. Bioorg. Med. Chem. Lett. 2007, 17,
4258–4261.
(21) Westwood, I.; Cheary, D. M.; Baxter, J. E.; Richards, M. W.; van
Montfort, R. L.; Fry, A. M.; Bayliss, R. Insights into the conformational
variability and regulation of human Nek2 kinase. J. Mol. Biol. 2009, 386,
476–485.
(22) Meyer, E. A.; Castellano, R. K.; Diederich, F. Interactions with
aromatic rings in chemical and biological recognition. Angew. Chem., Int.
Ed. 2003, 42, 1210–1250.
(23) Liao, J. J.-L. Molecular recognition of protein kinase binding
pockets for design of potent and selective kinase inhibitors. J. Med. Chem.
2007, 50, 409–424.
(17) Reynolds, C. H.; Tounge, B. A.; Bembenek, S. D. Ligand
binding efficiency: trends, physical basis, and implications. J. Med. Chem.
2008, 51, 2432-2438 and references therein.
(18) Benzimidazoles: (a) Moriarty, K. J.; Winters, M.; Qiao, L.;
Ryan, D.; DesJarlis, R.; Robinson, D.; Cook, B. N.; Kashem, M. A.;
Kaplita, P. V.; Liu, L. H.; Farrell, T. M.; Khine, H. H.; King, J.; Pullen,
S. S.; Roth, G. P.; Magolda, R.; Takahashi, H. Itk kinase inhibitors: initial
efforts to improve the metabolical stability and the cell activity of the
benzimidazole lead. Bioorg. Med. Chem. Lett. 2008, 18, 5537–5540. (b)
Mader, M.; de Dios, A.; Shih, C.; Bonjouklian, R.; Li, T. C.; White, W.; de
Uralde, B. L.; Sanchez-Martinez, C.; del Prado, M.; Jaramillo, C.; de
Diego, J. E.; Cabrejas, L. M. M.; Dominguez, C.; Montero, C.; Shepherd,
T.; Dally, R.; Toth, J. E.; Chatterjee, A.; Pleite, S.; Blanco-Urgoiti, J.;
Perez, L.; Barberis, M.; Lorite, M. J.; Jambrina, E.; Nevill, C. R.; Lee,
P. A.; Schultz, R. C.; Wolos, J. A.; Li, L. C.; Campbell, R. M.; Anderson,
B. D. Imidazolyl benzimidazoles and imidazo[4,5-b]pyridines as potent
p38 alpha MAP kinase inhibitors with excellent in vivo antiinflammatory
properties. Bioorg. Med. Chem. Lett. 2008, 18, 179–183. (c) Wittman, M.;
Carboni, J.; Attar, R.; Balasubramanian, B.; Balimane, P; Brassil, P.;
Beaulieu, F.; Chang, C.; Clarke, W.; Dell, J.; Eummer, J.; Frennesson, D.;
Gottardis, M.; Greer, A.; Hansel, S.; Hurlburt, W.; Jacobson, B.;
Krishnananthan, S.; Lee, F. Y.; Li, A.; Lin, T. A.; Liu, P.; Ouellet, C.;
Sang, X.; Saulnier, M. G.; Stoffan, K.; Sun, Y.; Velaparthi, U.; Wong, H.;
Yang, Z.; Zimmermann, K.; Zoeckler, M.; Vyas, D. Discovery of a
1H-(benzoimidazol-2-yl)-1H-pyridin-2-one (BMS-536924) inhibitor of
insulin-like growth factor I receptor kinase with in vivo antitumor
activity. J. Med. Chem. 2005, 48, 5639–5643. (d) Arienti, K. L.; Brun-
mark, A.; Axe, F. U.; McClure, K.; Lee, A.; Blevitt, J.; Neff, D. K.; Huang,
L.; Crawford, S.; Pandit, C. R.; Karlsson, L.; Breitenbucher, J. G.
Checkpoint kinase inhibitors SAR and radioprotective properties of a
series of 2-arylbenzimidazoles. J. Med. Chem. 2005, 48, 1873-1885;
purines: (e) Wong, C.; Griffin, R. J.; Hardcastle, I. R.; Northen, J. S.;
Wang, L. Z.; Golding, B. T. Synthesis of sulfonamide-based kinase
inhibitors from sulfonates by exploiting the abrogated SN2 reactivity of
2,2,2-trifluoroethoxysulfonates. Org. Biomol. Chem. 2010, 8, 2457–2464.
(f) Huang, W. S.; Zhu, X.; Wang, Y.; Azam, M.; Wen, D.; Sundara-
moorthi, R.; Thomas, R. M.; Liu, S.; Banda, G.; Lentini, S. P.; Das, S.; Xu,
Q.; Keats, J.; Wang, F.; Wardwell, S.; Ning, Y.; Snodgrass, J. T; Broudy,
M. I.; Russian, K.; Daley, G. Q.; Iuliucci, J.; Dalgarno, D. C.; Clackson,
T.; Sawyer, T, K.; Shakespeare, W. C. 9-(Arenethenyl)purines as dual
Src/Abl kinase inhibitors targeting the inactive conformation: design,
synthesis, and biological evaluation. J. Med. Chem. 2009, 52, 4743–4756.
(g) Oumata, N.; Bettayeb, K.; Ferandin, Y.; Demange, L.; Lopez-Giral,
A.; Goddard, M. L.; Myrianthopoulos, V.; Mikros, E.; Flajolet, M.;
Greengard, P.; Meijer, L.; Galons, H. Roscovitine-derived, dual-specifi-
city inhibitors of cyclin-dependent kinases and casein kinases 1. J. Med.
Chem. 2008, 51, 5229-5242; imidazopyridines: (h) Sato, Y.; Onozaki,
Y.; Sugimoto, T.; Kurihara, H.; Kamijo, K.; Kadowaki, C.; Tsujino, T.;
Watanabe, A.; Otsuki, S.; Mitsuya, M.; Iida, M.; Haze, K.; Machida, T.;
Nakatsuru, Y.; Komatani, H.; Kotani, H.; Iwasawa, Y. Imidazopyridine
derivatives as potent and selective polo-like kinase (PLK) inhibitors.
Bioorg. Med. Chem. Lett. 2009, 19, 4673–4678. (i) Hamdouchi, C.;
(24) Bissantz, C.; Kuhn, B.; Stahl, M. A medicinal chemist’s guide to
molecular interactions. J. Med. Chem. 2010, 53, 5061–5084.
(25) Kuntz, I. D.; Chen, K.; Sharp, K. A.; Kollman, P. A. The maximal
affinity of ligands. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 9997–10002.
(26) Hoffmann, W. W. Allylic 1,3-strain as a controlling factor in
stereoselective transformations. Chem. Rev. 1989, 89, 1841–1860.
(27) Watanabe, M.; Hirokawa, T.; Kobayashi, T.; Yoshida, A.; Ito,
Y.; Yamada, S.; Orimoto, N.; Yamasaki, Y.; Arisawa, M.; Shuto, S.
Investigation of the bioactive conformation of histamine H3 receptor
antagonists by the cyclopropylic strain-based conformational restriction
strategy. J. Med. Chem. 2010, 53, 3585–3593.
(28) Muley, L.; Baum, B.; Smolinski, M.; Freindorf, M.; Heine, A.;
Klebe, G.; Hangauer, D. G. Enhancement of hydrophobic interactions
and hydrogen bond strength by cooperativity: synthesis, modeling, and
molecular dynamics simulations of a congeneric series of thrombin
inhibitors. J. Med. Chem. 2010, 53, 2126–2135.
(29) Cox, C. D.; Coleman, P. J.; Breslin, M. J.; Whitman, D. B.;
Garbaccio, R. M.; Fraley, M. E.; Buser, C. A.; Walsh, E. S.; Hamilton, K.;
Schaber, M. D.; Lobell, R. B.; Tao, W.; Davide, J. P.; Diehl, R. E.; Abrams,
M. T.; South, V. J.; Huber, H. E.; Torrent, M.; Prueksaritanont, T.; Li, C.;
Slaughter, D. E.; Mahan, E.; Fernandez-Metzler, C.; Yan, Y.; Kuo, L. C.;
Kohl, N. E.; Hartman, G. D. Kinesin spindle protein (KSP) inhibitors.
Discovery of (2S)-4-(2,5-Difluorophenyl)-N-[(3R,4S)-3-fluoro-1-
methylpiperidin-4-yl]-2-(hydroxymethyl)-N-methyl-2-phenyl-2,5-dihy-
dro-1H-pyrrole-1-carboxamide (MK-0731) for the treatment of taxane-
refractory cancer. J. Med. Chem. 2008, 51, 4239–4252.
(30) Hayward, D. G.; Newbatt, Y.; Pickard, L.; Byrne, E.; Mao, G.;
Burns, S.; Sahota, N. K.; Workman, P.; Collins, I.; Aherne, W.; Fry, A. M.
Identification by high-throughput screening of viridin analogs as bio-
chemical and cell-based inhibitors of the cell cycle regulated Nek2
kinase. J. Biomol. Screening 2010, 15, 918–927.
1639
dx.doi.org/10.1021/jm1011726 |J. Med. Chem. 2011, 54, 1626–1639