Discovery and optimization of potent and selective imidazopyridine and imidazopyridazine mTOR inhibitors

Article abstract of DOI:10.1016/j.bmcl.2012.06.033

mTOR is a critical regulator of cellular signaling downstream of multiple growth factors. The mTOR/PI3K/AKT pathway is frequently mutated in human cancers and is thus an important oncology target. Herein we report the evolution of our program to discover ATP-competitive mTOR inhibitors that demonstrate improved pharmacokinetic properties and selectivity compared to our previous leads. Through targeted SAR and structure-guided design, new imidazopyridine and imidazopyridazine scaffolds were identified that demonstrated superior inhibition of mTOR in cellular assays, selectivity over the closely related PIKK family and improved in vivo clearance over our previously reported benzimidazole series.

Full text of DOI:10.1016/j.bmcl.2012.06.033

Bioorganic & Medicinal Chemistry Letters 22 (2012) 4967–4974
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery and optimization of potent and selective imidazopyridine
and imidazopyridazine mTOR inhibitors
Emily A. Peterson a, , Alessandro A. Boezio a, , Paul S. Andrews b, Christiane M. Boezio a, Tammy L. Bush d,
Alan C. Cheng e, Deborah Choquette a, James R. Coats a, Adria E. Colletti c, Katrina W. Copeland a,
Michelle DuPont d, Russell Graceffa a, Barbara Grubinska d, Joseph L. Kim f, Richard T. Lewis a,
Jingzhou Liu c, Erin L. Mullady b, Michele H. Potashman a, Karina Romero a, Paul L. Shaffer f,
Mary K. Stanton g, John C. Stellwagen a, Yohannes Teffera e, Shuyan Yi a, Ti Cai d, Daniel S. La a
⇑
⇑
a Medicinal Chemistry, Amgen Inc., 360 Binney St., Cambridge, MA 02142, USA
b Lead Discovery, Amgen Inc., 360 Binney St., Cambridge, MA 02142, USA
c Pharmacokinetics and Drug Metabolism, Amgen Inc., 360 Binney St., Cambridge, MA 02142,USA
d Oncology Research, Amgen Inc., 360 Binney St., Cambridge, MA 02142, USA
e Molecular Structure, Amgen Inc., 1120 Veterans Blvd., South San Francisco, CA 94080, USA
f Molecular Structure, Amgen Inc., 360 Binney St., Cambridge, MA 02142, USA
g Pharmaceutics R&D, Amgen Inc., 360 Binney St., Cambridge, MA 02142, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
mTOR is a critical regulator of cellular signaling downstream of multiple growth factors. The mTOR/PI3K/
AKT pathway is frequently mutated in human cancers and is thus an important oncology target. Herein
we report the evolution of our program to discover ATP-competitive mTOR inhibitors that demonstrate
improved pharmacokinetic properties and selectivity compared to our previous leads. Through targeted
SAR and structure-guided design, new imidazopyridine and imidazopyridazine scaffolds were identified
that demonstrated superior inhibition of mTOR in cellular assays, selectivity over the closely related PIKK
family and improved in vivo clearance over our previously reported benzimidazole series.
Ó 2012 Published by Elsevier Ltd.
Received 15 May 2012
Revised 6 June 2012
Accepted 11 June 2012
Available online 16 June 2012
Keywords:
mTOR
Kinase inhibitor
Oncology
Imidazopyridine
Imidazopyridazine
The mammalian target of rapamycin (mTOR) kinase has been
identified as a critical signaling node in the promotion of cancer cell
growth and survival.1 The mTOR pathway senses the cellular avail-
ability of nutrients and energy, serves as a downstream regulator of
growth factor signaling, and is ultimately responsible for mediating
cell functions such as apoptosis, autophagy, translation, cell cycle
regulation and cytoskeleton reorganization. mTOR functions
through two distinct multi-protein complexes, each defined by
the interaction of mTOR with either Raptor (regulatory-associated
protein of mTOR) as in the case of mTORC1 or Rictor (rapamycin-
insensitive companion of mTOR) as in the case with mTORC2. mTOR
complex 1 (mTORC1) acts as an integral participant in translation
and cell growth by phosphorylation of substrates 4EBP1 and S6K,
whereas mTOR complex 2 (mTORC2) acts to fully activate signaling
through AKT by phosphorylation of S473. Both complexes play crit-
ical roles in the PI3K/AKT cascade, a pathway wherein mutations
within its signaling components are frequently present in cancer
cells.
As an oncology target, mTOR has been validated in humans by
the clinical use of rapamycin analogues ToriselÒ (Temsirolimus)2
and AfinitorÒ (Everolimus),3 both allosteric inhibitors of mTORC1.
While these inhibitors have shown promising results in a limited
number of cancers, it has been reported that selective inhibition
of mTORC1 by rapamycin can increase AKT signaling through elim-
ination of the negative feedback mechanism with IRS1.4 The goal of
our program was to develop an ATP-competitive mTOR inhibitor
that would block the activity of both mTORC1 and mTORC2 and
demonstrate selectivity over kinases within the highly homologous
PIKK family, especially PI3K
the expectation that a selective inhibitor of mTOR could demon-
strate increased tolerability compared to a dual mTOR/PI3K
inhibitor due to the effects of PI3K inhibition on glucose homeo-
stasis.5 A significant amount of research aimed at developing
a. The rationale for this decision was
⇑
Corresponding authors. Tel.: +1 617 444 5027; fax: +1 617 621 3907 (E.A.P.);
a
tel.: +1 617 444 5104; fax: +1 617 621 3908 (A.A.B.).
E-mail addresses: epeterso@amgen.com (E.A. Peterson), aboezio@amgen.com
(A.A. Boezio).
a
0960-894X/$ - see front matter Ó 2012 Published by Elsevier Ltd.
http://dx.doi.org/10.1016/j.bmcl.2012.06.033

4968
E. A. Peterson et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4967–4974
Table 1
introduce an infinitesimal twist in the framework that is enough
to perturb molecular packing enough to improve solubility.9b To
test this hypothesis, the benzimidazole nitrogen was transposed
to provide imidazopyridine 2 (Table 1). This change maintained
the broader planarity between the triazine hinge-binder and the
imidazopyridine ring, which was previously shown to be essential
for maintaining good potency. The calculated global pKa, and the
cLogP were also minimally affected by this change. The pKa was in-
creased for the imidazopyridine relative to the benzimidazole scaf-
fold and the cLogP value was decreased by this subtle structural
change, both factors known to improve aqueous solubility. This
small structural mutation led to a significant improvement with
regard to solubility and bioavailability when compared to benz-
imidazole 1, as evidenced by the improved solubility measured
in SIF and HCl solutions. This critical advancement represented a
major turning point in the program, as it addressed one of the
key PK flaws of the benzimidazole series reported earlier. The
selectivity over related PI3K kinases and the broader kinome was
also maintained, with 2 showing exquisite selectivity over the
402 kinases tested, attenuating concerns for potential confounding
pharmacology in an in vivo setting.10
Comparison of benzimidazole and imidazopyridine
Compound
1
2
Biochemical and cellular assays (nM)
mTORa
PI3Kab
PI3Kbb
p4EBP1c
pAKTd
pS6e
23
12
798 (35X)
875 (38X)
391
55
166
590 (49X)
1130 (94X)
162
19
69
Solubility (
PBS
l
g/mL)f
8.4
8.4
3.7
6.3
27
>200
SIF
0.01 N HCl
Permeability (Human LLC-PK1 cells, lcm/s)
The improvement in bioavailability provided the opportunity to
test 2 in vivo to establish target coverage and tumor growth inhibi-
tion using an orally administered compound.11 To confirm that 2
inhibits mTOR activity in vivo, mice were administered a single dose
of vehicle or 2 at 10, 30 and 100 mg/kg. As shown in Figure 1A,
administration of 2 for 6 h resulted in significant dose-dependent
inhibition of pAKT (at both S473 and T308 sites) and pS612 in HGF-
stimulated mouse liver tissue compared to vehicle-controls (p
P APP A2B
P APP B2A
Efflux Ratio
IV Rat PK (dose = 1.0 mg/kg)g
Cl (L/h/kg)
20.3
35.4
1.7
13.8
33.7
2.4
1.9
1.1
0.72
6
1.8
1.1
0.45
24
Vss (L/kg)
T1/2 (h)
PO (%F)h
Selected calculated properties
cLogPi
1.84
123.2
2.39
1.78
122.7
2.53
60.0368).13 The downstream target of PI3K
a (through PDK1), pAKT
PSA (Å2)
Global pKaj
(T308), was inhibited to a lesser degree with the higher 30 and
100 mg/kg doses, demonstrating desired target selectivity in vivo.
Furthermore, the drug concentrations measured in plasma reflected
the degree of mTOR inhibition (Fig. 1A). After correcting for protein
binding, the measured in vivo IC50 for pAKT (S473) and pS6 were 53
and 71 nM, respectively, which corresponded well with the values
measured in the in vitro cellular assays (Table 1).
a
b
c
n P 2, inhibition of kinase activity (Lantha-Screen).
n P2, inhibition of kinase activity (Alpha-screen).
n P2, cell assay measuring p4EBP1 in U87 cells.
n P2, cell assay measuring pAKT (S473) in U87 cells.
d
e
n P2, cell assay measuring phosphorylation of mTOR substrate S6 (S240/244)
in U87MG cells.
f
Symyx solubility system.
Formulation = 100% DMSO.
1%Tween 80, 2%HPMC, 97% H2O, pH=2.2 (methanesulfonic acid).
Daylight method.
g
A subsequent time-course mouse liver pharmacodynamic study
was conducted with 100 mg/kg at 3, 6, 8 and 24 h to understand
duration of target coverage (data not shown). Oral administration
of 2 resulted in statistically significant inhibition of pAKT (at both
S473 and T308 sites) and pS6 greater than 8 hours with recovery
by 24 h. In anticipation of a long-term efficacy study, we first eval-
uated two high doses, 60 and 100 mg/kg (n = 5 per group), in a 7-day
tolerability study (data not shown). After 3 days, mice dosed with
100 mg/kg daily (QD) resulted in body weight loss (average >20%)
and were removed from the study. In contrast, 60 mg/kg dosed
QD for 7 days was well tolerated, suggesting this dosage could be
used as the maximum starting dose for the xenograft study.
To compare the inhibition of mTOR activity (as measured by the
degree and duration of pAKT and pS6 inhibition) with suppression
of tumor growth in vivo, a tumor xenograft assay was performed.
Mice bearing established U87-MG tumors were orally adminis-
tered vehicle alone or 2 at 10, 30 or 60 mg/kg QD or 30 mg/kg twice
daily (BID) continuously throughout the study. As shown in Figure
1B, continuous administration of 2 resulted in significant tumor
growth inhibition (TGI) that was dose-dependent compared to
vehicle-controls [60 mg/kg (82%) QD or 30 mg/kg (84%) BID; p
60.0001]. Encouraged by the robust in vivo response to treatment
with 2, the focus of our efforts turned to further improvement of
this promising scaffold. Consideration of the target coverage re-
quired to achieve significant TGI, as well as the potency and phar-
macokinetic properties of 2, prompted further investigation into
how the potency, in vivo clearance properties and selectivity of 2
could be improved to provide an mTOR inhibitor with qualities
more suitable for drug development.
h
i
j
ACD method.
ATP-competitive mTOR inhibitors has recently been reported,6
including our own discovery of benzimidazole triazines, which
showed selectivity over related PIKK family members as well as
across the broader kinome.7 While benzimidazole 1 provided
important proof of concept, it showed low, nonlinear exposure
when dosed orally due to poor solubility (Table 1).8 Herein we
report our efforts to address the undesirable physicochemical
properties of this series while seeking enhanced potency and selec-
tivity as well as improved ADME properties.
We attributed the poor bioavailability of 1 to poor aqueous sol-
ubility, since permeability was high and there was minimal efflux
observed. There are several strategies one might employ to im-
prove the aqueous solubility such as lowering cLogP, altering the
pKa, disrupting molecular planarity by disrupting aromaticity or
introducing unsaturation.9 Our attempts to incorporate these strat-
egies were met with varying success. The cLogP for 1 is 1.84, a
value low enough that further reduction by the incorporation of
additional heteroatoms would increase the already high PSA and
the potential for efflux. Alternatively, introducing saturation into
any of the three rings led to significant loss of potency or com-
pound instability. A more subtle approach to disrupting molecular
planarity would be to move one of the benzimidazole nitrogens
into a bridgehead position. It has been previously hypothesized
that a bridgehead nitrogen in a fused aromatic ring system can

E. A. Peterson et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4967–4974
4969
Figure 1. Compound 2 inhibits phosphorylation of mTOR substrates in mouse liver tissue and the growth of U87-MG tumor xenografts. (A) Bar graph representing the
pharmacodynamic-pharmacokinetic assay measuring changes in HGF-induced phosphorylation of pAKT and pS6RP after a single oral dose of vehicle alone (50% PEG400, 1%
Pluronic F68, pH 2.2 with MSA) or 2 at 10, 30 or 100 mg/kg (n = 3 per dose) at 6 h. (B) Mice bearing established U87-MG tumors were orally administered vehicle (j) or 2 at 10
(h), 30 ( ), 60 mg/kg ( ) QD or 30 mg/kg ( ) BID for 2 weeks. Tumor volumes (mm3) are represented as mean SE (n = 10). Asterisk (⁄) indicates statistical significant
inhibition compared to vehicle-control determined by Factorial ANOVA (A, p <0.0368) or repeated measures analysis of variance (RMANOVA) (B, p <0.0001) followed by
Dunnett’s test for multiple comparisons.
In an attempt to decrease the amount of drug required to
achieve robust tumor growth inhibition, efforts were first focused
on improving the potency of 2. An X-ray co-crystal structure of
are likely conserved within the mTOR active site, are between
the triazine amine and the kinase hinge residue VAL882 as well
as the interaction of the catalytic LYS833 with the pyrazole and
the imidazopyridine core. The imidazopyridine core occupies the
space where the ribose sugar moiety of ATP would reside. Exten-
sive SAR had already been performed to optimize the methyl group
214 with structurally related kinase PI3K
c was obtained in order
to understand how 2 was interacting with the kinase active site.
( Fig. 2A).15 The primary hydrogen bonding interactions, which
A. Co-crystal of 2 with PI3Kγ
B. Inhibitor 12 in mTOR homology model
LYS833
ILE697
LYS705
VAL882
TRP779
ARG812
Figure 2. Structural design elements. (A) Co-crystal of 2 with P13Kc. (B) Inhibitor 12 in mTOR homology model.

4970
E. A. Peterson et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4967–4974
Table 2
Exploration of the ribose pocket
IC50 (nM)
Figure 3. Metabolism of 2 in rats.
Compd
R
mTOR
12
P13K
a
4EBP1
162
pAKT
19
2
H
590 (49X)
on the triazine hinge-binder, which was optimal for provision of
broader kinase selectivity. In addition, the pyrazole affinity pocket
group was uniquely effective at providing potency and selectivity
3
4
15
27
648 (43X)
67
64
73
over PI3Ka c co-crystal structure
.7 Further examination of the PI3K
with 2 revealed two areas where additional SAR might be devel-
oped to improve the potency and selectivity for mTOR (Fig. 2A).
The remaining positions that had not been investigated were the
substitution of the amine hinge binder and extension into the
ribose pocket from either the 6 or 7-position of the imidazopyri-
dine ring. Interestingly, both positions also suggested potential
for improving selectivity due to residue differences between the
PI3K family versus mTOR (Fig. 2B, vide infra).16
654 (24X)
306 (55X)
912
5
6
326
26
6
3
119 (36X)
120 (48X)
393 (19X)
73 (47X)
268 (7X)
101
41
12
10
The first area that was explored was substitution of the imi-
dazopyridine ring (Table 2). To preserve the co-planarity of the
triazine and imidazopyridine rings considered essential for po-
tency, substitution of the 5-position was not investigated, since
any substitution would likely disrupt planarity. Further examina-
tion of the structure depicted in Figure 2A revealed that the 6
and 7-positions had the most space available and coincidentally
were the most accessible from a synthetic standpoint. Substitution
at the 7-position was tested and did not result in an improvement
in potency over the parent 2, whereas introduction of aromatic
substituents at the 6-position provided an improvement in po-
tency (Table 2). The data in Table 2 summarize the SAR findings
from substitution of the 6-position. A variety of substituents were
tested at this position including aliphatic, alkoxy, amide, alkenyl,
alkynyl, aromatic and heteroaryl with varying substitutions. In
general, cellular potency was improved with heteroaryl substitu-
ents, giving IC50s < 100 nM for inhibition of the p4EBP1 assay.
As predicted by analysis of the mTOR homology model shown
with inhibitor 12 in Figure 2B, aryl groups containing a nitrogen
at the meta-position gave the best boost in potency, presumably
due to a favorable interaction with LYS705. The proximity of resi-
dues LYS705 and ARG 812 to the two meta-positions of the pyrim-
idine also rationalized why substituents at the 3-position of the aryl
ring did not lead to a potency improvement (4, 11, 14). Introduction
of a methyl group ortho to the ring attachment also provided an in-
crease in mTOR cellular potency, but only in the presence of the
meta-nitrogen group (Table 2, 6, vs 7, vs 8). This combination re-
sulted in a mTOR p4EBP1 cellular potency of 41 nM.17 This addi-
tional potency improvement could be a result of the methyl
group enforcing the twisted conformation shown in Figure 2B, lock-
ing the substituent in a favorable conformation for interaction with
the adjacent LYS705 and avoiding a potential steric clash with adja-
cent ILE697. Five-membered heterocycles were also effective at sig-
nificantly increasing the cellular potency (Table 2, 15–17) The
potency improvement observed by adding an aromatic substituent
at the 6-position was encouraging, but was tempered by the obser-
vation that the binding efficiency (BE) was diminished as was selec-
7
3
8
21
2
652
49
54
13
138
9
10
39
510
11
4
239 (64X)
222
23
12
13
2
3
85 (37X)
68 (24X)
93
57
5
6
14
15
10
5
96 (9X)
115
276
29
23
463 (98X)
16
17
2
3
144 (63X)
177 (71X)
58
72
8
7
tivity over PI3Ka. Furthermore, pharmacokinetics remained a
major hurdle for this series, in particular the high in vivo clearance,

E. A. Peterson et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4967–4974
4971
Table 3
Exploring the imidazopyridazine core and substitution of the linker-binder pyrimidine
IC50 (nM)
Rat PKa
Vss (L/kg)
Sol.
Compd
R
mTOR
p4EBP1
P13K
a
CL (L/h/kg)
T1/2 (h)
Fb (%)
SIF (lg/mL)
2
18
—
—
NH2
—
NHMe
H
12
76
13
65
63
141
3
162
UNc
214
507
706
2170
45
590 (49X)
635 (8X)
1.8
2.5
2.3
4.1e
4.8
1.9
2.6
1.14
1
2
1.2
1.3
0.62
6.3
0.44
0.26
0.85
0.25
0.26
0.23
0.82
24
27
1
150
162
7
NDd
100
2f
19a
20
19b
19c
19d
763 (58X)
5840 (89X)
1140 (18X)
3020 (21X)
47 (13X)
29
ND
46
14
200
NHAc
19e
6
66
122 (21X)
5.8
2.5
0.42
ND
ND
19f
147
24
6390
276
1140 (8X)
173 (7X)
ND
ND
ND
ND
ND
ND
ND
ND
1
19g
ND
19h
35
309
256 (7X)
2.5
3.9
1.59
7g
98
19i
19j
36
3
146
85
105 (3X)
54 (16X)
6
3.2
9.1
0.36
1.18
ND
ND
>200
175
6.2
a
b
c
d
e
f
Rat IV PK, 0.5 mpk, formulation: 100% DMSO.
Rat PO PK, 5.0 mpk, formulation: 1%Tween 80, 2%HPMC, 97% H2O, pH 2.2 (methanesulfonic acid).
UN = Undetermined, failed to achieve proper hill fit to generate IC50 curve.
ND = not determined.
Rat IV PK, 0.3 mpk, formulation: 100% DMSO.
Rat PO PK, 2.0 mpk.
g
Rat PO PK, 4.0 mpk.
which for 2 and the 6-substituted analogs in Table 2 was around 2 L/
h/kg (data not shown) and did not correlate with the low observed
in vitro microsomal clearance in rat and human liver microsomes.
The compounds demonstrated reasonable permeability and were
not substrates for efflux transporters. This lack of in vitro–in vivo
PK correlation and our previous experience with related benzimid-
azole inhibitors 1 suggested that the high in vivo clearance was a
result of phase II metabolism.
In order to understand the mechanism resulting in such high clear-
ance for this series, rat bile duct cannulation studies were per-
formed on parent compound 2. In this experiment,
2 was
administered either PO or IV to male Sprague–Dawley rats and col-
lected urine and bile were analyzed by LCMS-MS to determine the
primary metabolites. A small amount of intact parent was excreted
in both the IV and PO dosed rats, however, glucuronidation was the
primary means of metabolism for 2 for both routes of administra-
tion (Fig. 3). The pyrazole was the primary site of glucuronidation,
although the triazine amine glucuronide was also detected in sig-
The results in Table 2 encouraged us to refocus on improving
the in vivo PK as a means to increase the exposure of our inhibitors.

4972
E. A. Peterson et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4967–4974
nificant quantities. Minor metabolites were formed through oxida-
tion of the imidazopyridine ring followed by subsequent [O] or [N]
glucuronidation.
access a variety of substituents, which permitted further exploration
of the hinge-binder pocket. Combination of the pyrimidine with the
imidazopyridine core was also investigated to test the hypothesis
that the two adjacent hydrogens on the attached pyrimidine and
imidazopyridine rings would create a steric clash that would lead
to a deviation from planarity. Indeed, compound 20 did show dimin-
ished potency, but unexpectedly also displayed improved selectivity
Several strategies could be employed to address the extensive
glucuronidation of 2, such as eliminating or sterically encumbering
the primary site of glucuronidation, or modulating the electronic
properties of the molecule to decrease glucuronidation. The free
pyrazole NH functionality that was the site of glucuronidation
could be capped or replaced by an alternate heteroatom. From pre-
viously reported SAR studies, it was clear that the pyrazole affinity
pocket group was uniquely effective in providing both potency on
over PI3Ka. This was an interesting discovery; however 20 demon-
strated poor bioavailability, cell potency and high in vivo CL.
The results from exploration of the hinge-binder substitution
using the imidazopyridazine core are represented by inhibitors
19b–j (Table 3). The space accessed by these substituents is a nar-
row cleft that opens up toward solvent (see Fig. 2), and thus only
small alkyl or planar groups were tolerated. Acetamide 19d and
pyrazine 19e were initially designed as substituents that would
maintain co-planarity with the pyrimidine hinge-binder. Indeed,
groups lacking a heteroatom adjacent to the point of attachment
suffered a decrease in potency, potentially due to interaction of
the adjacent hydrogens, which could twist the attached ring away
from co-planarity (see 19f). The acetamide 19d, pyrazine 19e and
pyridazine 19j gave the best improvement in mTOR cellular
p4EBP1 potency, with a slight diminishment of selectivity over
mTOR as well as selectivity over structurally related PI3Ka 7
. The
steep SAR in this highly polar region made addressing this PK lia-
bility difficult. Substitution with related heterocycles such as isox-
azole provided compounds with significantly diminished potency
and selectivity. Methylation of the offending pyrazole NH appeared
to be the best path forward, although it led to a marked decrease in
potency. In order to mitigate the loss in the potency, the incorpo-
ration of the methyl pyrazole was paired with substitution at the
6-position of the imidazopyridine, a change previously shown to
boost potency (Table 3). Inhibitor 18 is representative of this SAR
change and revealed that although occupation of the ribose pocket
with a pyrimidine moiety somewhat mitigated the loss of potency
attributed to the methyl pyrazole, the solubility and in vivo PK
PI3Ka. The in vivo CL was not improved by these substitutions,
but this was anticipated, since the pyrazole affinity pocket group
retained its free NH. To understand the potency improvement ob-
tained by these substitutions, both 19d and 19e were co-crystal-
properties were not improved; furthermore 18 failed to inhibit
18
mTOR activity in the cellular p4EBP1 assay.
The high clearance
observed for 18 revealed that blocking only one site of glucuroni-
dation was not sufficient to improve the total clearance of these
inhibitors.
lized in the active site of PI3Kc. These co-crystal structures (only
19e pictured, Fig. 4) revealed that both the acetamide and the pyr-
azine make an additional hydrogen bonding contact with hinge
residue VAL882.19 This contact is best described as a bifurcated
H-bonding interaction with the pyrimidine NH and the acetamide
CH3/pyrazine CH (see Fig. 4).
We speculated that the high in vivo CL of 18 was likely a result of
shifting the primary site of glucuronidation from the pyrazole to the
triazine. To address this hypothesis, substitution of the hinge-binder
moietyto addressglucuronidationofthetriazinewasinvestigated. It
was quickly discovered that the triazine was not particularly amena-
ble to substitution with groups other than alkyl. Attempts to synthe-
size the triazine amine with acyl or aryl substituents led to
chemically unstable compounds. To obviate this problem, a pyrimi-
dine hinge-binder group was investigated with concomitant incor-
poration of an additional nitrogen in the imidazopyridine ring. This
combination of a pyrimidine hinge-binder and an imidazopyrid-
azine core provided 19a, which was intended to maintain overall
structural co-planarity (Table 3). This resulted in a similar potency
profile to 2, but with a greatly increased solubility and bioavailabil-
ity. Furthermore, the pyrimidine hinge-binder provided a means to
Taking advantage of the improved potency provided by the
incorporation of the acetamide-pyrimidine hinge-binder, this sub-
stitution was paired with the methyl pyrazole affinity pocket group
in an attempt to circumvent the glucuronidation at both of these
sites. The results from this effort on both the imidazopyridazine
and imidazopyridine cores are shown in Table 4. For the imidazo-
pyridazine scaffold, capping of both the hinge-binder and the pyr-
azole led to a decrease in the in vivo CL, but when corrected for free
fraction, there was no improvement in the unbound clearance (Clu)
for the imidazopyridazine scaffold 21b.20 The Clu for imidazopyrid-
azines 19a and 21b were 26.1 and 33.3, respectively, whereas the
corresponding Clu for imidazopyridines 20 and 21d were 30.9 and
10.8, respectively. The observed 3-fold improvement in the Clu
encouraged further exploration of the pyrimidine-imidazopyridine
scaffold, which also provided greater selectivity over PI3Ka. To im-
prove the potency on mTOR, which was diminished by the incorpo-
ration of the methyl pyrazole, substitution of the ribose pocket was
explored. Extensive SAR guided by the previous work summarized
in Table 3 revealed that 3-pyridyl substituents and 5-membered
heterocycles provided the best boost in potency. Methylpyridyl-
substituted 21e demonstrated the best combination of cellular po-
ASP964
LYS803
tency on p4EBP1 and selectivity over PI3Ka. Of the ribose-substi-
tuted analogs, 21e also demonstrated the best oral exposure,
solubility, and in vivo clearance, an overall improvement from
our starting point 1.
TRP812
The synthesis described in Scheme 1 for imidazopyridine 2 is
also representative of the procedures used to synthesize imidazo-
pyridazine scaffold 19. Palladium-catalyzed coupling of methyldi-
chlorotriazine 227 with 2-chloroimidazo[1,2-a]pyridine (23)21
provided the imidazopyridine core.22 Subsequent aminolysis of
the remaining triazine chloride with ammonia using microwave
irradiation provided triazine-imidazopyridine 24. The affinity pock-
et aminopyrazole was installed by Buchwald–Hartwig amination of
the chloroimidazopyridine 24 with 3-aminopyrazole catalyzed by
VAL882
Figure 4. X-ray co-crystal structure of 19e in P13Kc.

E. A. Peterson et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4967–4974
4973
Table 4
Capping sites of glucuronidation and introducing a ribose substituent improves PK and potency
IC50 (nM)
Rat PK
Compd
X
R
R1
mTOR
p4EBP1
p13K
a
CLa (L/h/kg)
AUC (
lM h)
Vss (L/kg)
T1/2 (h)
Fb
%
rPPB (%)
1
2
—
—
—
N
N
CH
CH
—
—
—
H
Me
H
Me
—
—
—
H
H
H
H
23
12
13
3
55
9
391
162
210
45
1230
60
798 (35X)
590 (49X)
763 (58X)
47 (13X)
1660 (30X)
112 (12X)
4580 (68X)
1.9
1.8
2.3
2.6
1.1
2.3
0.6
1.7
1.8
1.0
0.6
1.3
0.6
2.3
1.1
1.1
2.0
2.5
0.8
1.2
0.6
0.72
0.44
0.85
0.82
0.39
0.52
0.66
6
24
100
46
34
97.2
97.7
91.2
94.4
96.7
ND
19a
19d (21a)
21b
21c
21d
ND
67
940
4c
94.4
21e
21f
21g
CH
CH
CH
Me
Me
Me
15
16
20
230
694
347
4290 (286X)
>125,000
0.6
0.5
7.0
7.6
1.8
0.2
1.7
2.1
4.1
2.8
4.3
0.7
38c
8c
97.5
99.6
NDd
3170 (161X)
3c
a
b
c
Rat IV PK, 0.5 mpk, formulation: 100% DMSO.
Rat PO PK, 5.0 mpk, formulation: 1%Tween 80, 2%HPMC, 97% H2O, pH = 2.2 (methanesulfonic acid).
Formulation: 30% HPBCD, 70% H2O, pH = 2.2 (methanesulfonic acid).
ND = not determined.
d
Scheme 1. Reagents and conditions: (a) (i) Cs2CO3, Pd(OAc)2, PPh3, dioxane, 130–
150 °C ( W); (ii) NH3 (g), (b) 1H pyrazol-3-amine (or 1-methyl-1H-pyrazol-3-
l
amine), CyPF-t-Bu, Pd2(dba)3, Cs2CO3, t-BuOH/H2O (10:1), 100 °C, 50%.
CyPF-t-Bu and Pd2(dba)3 in the presence of Cs2CO3 to provide the
final product 2 (Scheme 1).23
Once it had been established that the 6-position of the imidazo-
pyridine was the most advantageous for substitution with regard to
potency improvement, the synthesis was modified to enable rapid
diversification of the 6-position (Scheme 2). The most convenient
and versatile handle to carry through the synthesis for late-stage
analog generation was a bromide, which proved problematic during
the initial coupling step between dichloromethyltriazine 22 and 6-
bromo-2-chloroimidazo[1,2-a]pyridine (27). Synthesis of 27
started from 2-amino-5-bromopyridine (25). Alkylation of the pyr-
idyl amine with chloroacetic acid provided amino acid 26. Cycliza-
tion, promoted by phosphorylchloride in refluxing acetonitrile,
gave the imidazopyridine chloride 27, which contained the desired
bromide functionality for later Suzuki coupling. In order to preserve
the bromide functionality, a milder coupling procedure than the
one used for the synthesis of 2 had to be employed. Toward this
end, deprotonation of the 6-bromo-2-chloroimidazo[1,2-a]pyridine
Scheme 2. Reagents and conditions: (a) chloroacetic acid, Et3N, MeCN, reflux 18 h,
61%; (b) POCl3, MeCN, reflux 2 h, 50%; (c) (i) ZnCl(tmp)ÁLiCl, THF, rt; (ii) 4-chloro-6-
methyl-1,3,5-triazin-2-amine, Pd(PPh3)4, rt; (iii) NH3 (g) rt, 50–80%; (d) 5-
pyrimidinylboronic acid, Na2CO3, PdCl2(dppf)ÁCH2Cl2, dioxane/H2O, 80 °C, 40–70%;
(e) 1H-pyrazol-3-amine, CyPF-t-Bu, Pd2(dba)3, Cs2CO3, t-BuOH/H2O (10:1), 100 °C,
39–50%.
27 using Knochel’s LiTMPÁZnCl base,24 followed by a palladium tet-
rakistriphenylphosphine-catalyzed Negishi cross-coupling with
2,4-dichloro-6-methyl-1,3,5-triazine 28 provided the triazine imi-
dazopyridine core with the bromide intact.25 This reaction was con-
veniently telescoped with the amination of the remaining triazine
chloride to provide the aminotriazine 29. A variety of cross-cou-
pling reactions were employed to introduce an assortment of R

4974
E. A. Peterson et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4967–4974
Acknowledgments
The authors would like to acknowledge Wei Hu, Tisha San Mig-
uel, and Leeanne Zalameda for enzyme and cell assay support. We
are also grateful to Loren Berry, Xuhai Be, Liyue Huang, Meghan
Langley, and Jonothan Roberts for PKDM support. Thanks to Pete
Yakowec and Jin Tang for PI3Kc expression and purification as well
as Huilin Zhao and Linda Epstein for mTOR expression and
purification.
References and notes
1. Zoncu, R.; Efeyan, A.; Sabatini, D. M. Nature Rev. 2011, 12, 21.
2. (a) Rini, B. I. Clin. Cancer Res. 2008, 14, 1286; (b) Rini, B.; Kar, S.; Kirkpatrick, P.
Nat. Rev. Drug Disc. 2007, 6, 599.
3. Motzer, R. J.; Escudier, B.; Oudard, S.; Hutson, T. E.; Porta, C.; Bracarda, S.;
Grünwald, V.; Thompson, J. A.; Figlin, R. A.; Hollaender, N.; Kay, A.; Ravaud, A.
Cancer 2010, 116, 4256.
4. Carracedo, A.; Pandolfi, P. P. Oncogene 2008, 27, 5527.
5. Luo, J.; Sobkiw, C. L.; Hirshman, M. F.; Logsdon, M. N.; Li, T. Q.; Goodyear, L. J.;
Cantley, L. C. Cell Metab. 2006, 3, 355.
6. (a) Schenone, S.; Brullo, C.; Musumeci, F.; Radi, M.; Botta, M. Curr. Med. Chem.
2011, 18, 2995–3014; (b) Garcia-Echeverria, C. Bioorg. Med. Chem. Lett. 2010, 20,
4308.
7. Peterson, E. A.; Andrews, P. S.; Be, X.; Boezio, A. A.; Bush, T. L.; Cheng, A. C.;
Coats, J. R.; Colletti, A. E.; Copeland, K. W.; DuPont, M.; Graceffa, R.; Grubinska,
B.; Harmange, J.-C.; Kim, J. L.; Mullady, E. L.; Olivieri, P.; Schenkel, L. B.; Stanton,
M. K.; Teffera, Y.; Whittington, D. A.; Cai, T.; La, D. S. Bioorg. Med. Chem. Lett.
2011, 21, 2064.
Scheme 3. Reagents and conditions: (a) Ar–NH2, CyPF-t-Bu, Pd2(dba)3, Cs2CO3, t-
BuOH/H2O (10:1), 90 °C, 18 h, 50–90%; (b) 3-aminopyrazole, CyPF-t-Bu, Pd2(dba)3,
Cs2CO3, t-BuOH/H2O (10:1), 120 °C, 2 h, 30–50%; (c) (i) Ac2O, pyridine, 80 °C, 20–
60%; (ii) (only for synthesis of R1 = H) DCM/MeOH/H2O (90/10/1), K2CO3, rt, 20%.
groups at the 6-position, however, with aryl and heteroaryl groups
providing the best potency, Suzuki coupling was used most often, as
described for the synthesis of penultimate intermediate 30. Instal-
lation of the affinity-pocket aminopyrazole as previously described
in Scheme 1 resulted in final analog 12. The Knochel coupling pro-
cedure described in Scheme 2 was also amenable to the synthesis of
19a, 20, and compounds 21a–21f in Table 4.
Functionalization of the hinge-binder pyrimidine is outlined in
Scheme 3. For analogs 19e–19j, the heteroaryl group attached to
the pyrimidine amine was installed using a palladium-catalyzed
amination of the dichloride intermediate 31.26 Keeping the tem-
perature at 90 °C for this coupling ensured regioselective reaction
of the pyrimidine chloride, instead of the imidazopyridine chloride.
A second amination was then performed using the identical condi-
tions at 120 °C to couple the affinity pocket pyrazole (Scheme 3,
step b). The acetamide functionality could be installed through
treatment of the pyrimidine amine 32 with acetic anhydride in
pyridine. In the cases where the pyrazole was unsubstituted, a sec-
ond step was required to cleave the extraneous acetate from the
pyrazole to provide inhibitors 19d and 21b.
8. Solubility tested as follows: 1 mg solid compound tested for solubility (0–
200 lg/mL) in 3 media (0.01 HCl = hydrochloric acid, PBS = Phosphate-buffered
saline, SIF = simulated intestinal fluid).
9. (a) Li, Q.; Chu, D. T. W.; Claiborne, A.; Cooper, C. S.; Lee, C. M.; Raye, K.; Berst, K.
B.; Donner, P.; Wang, W.; Hasvold, L.; Fung, A.; Ma, Z.; Tufano, M.; Flamm, R.;
Shen, L. L.; Baranowski, J.; Nilius, A.; Alder, J.; Meulbroek, J.; Marsh, K.; Crowell,
D.; Hui, Y.; Seif, L.; Melcher, L. M.; Henry, R.; Spanton, S.; Faghih, R.; Klein, L. L.;
Tanaka, S. K.; Plattner, J. J. J. Med. Chem. 1996, 39, 3070; (b) Ishikawa, M.;
Hashimoto, Y. J. Med. Chem. 2011, 54, 1539; (c) Gleeson, M. P. J. Med. Chem.
2008, 51, 817.
10. Ambit, kinome scan binding assay measured at
www.kinomescan.com/, last accessed 5–14-2012.
1 lM, http://
11. In our previous report,7 1 was dosed IP due to solubility constraints.
12. S6 instead of p4EBP1 was measured to determine the inhibition of mTORC1
due to the lack of a suitable antibody for p4EBP1 in mouse.
13. pAKT levels in the liver lysate were measured using an MSD assay method.
Measurement of pAKT inhibition in HGF-stimulated liver is a common practice
within the mTOR field, see: Feldman, M. E.; Apsel, B.; Uotila, A.; Loewith, R.;
Knight, Z. A.; Ruggero, D.; Shokat, K. M. PloS Biol. 2009, 7, 1.
14. X-ray coordinates deposited in the Cambridge Crystallographic Data Centre,
PDB code 4FHJ.
15. PI3Kc shares significant homology with the mTOR active site and is commonly
used to understand the binding trajectory of mTOR inhibitors in the ATP active
site.
16. The narrow cleft adjacent to the hinge binder in mTOR is caused by TRP779,
which overlays with PI3Kc and a at the VAL881 residue.
Starting from the previously reported triazine-benzimidazole 1,
which had less desirable PK properties, modifications were made
to the central benzimidazole ring to improve solubility. This effort
resulted in triazine-imidazopyridine 2, which demonstrated im-
proved solubility and bioavailability when compared to 1. This
improvement was sufficient to allow the testing of 2 in multiple
in vivo studies demonstrating up to 84% TGI at a 60 mpk oral dose.
While this outcome was encouraging, the triazine-imidazopyridine
series still suffered from high in vivo clearance that was not pre-
dicted by in vitro liver microsome experiments. Rat BDC metabo-
lism studies indicated that glucuronidation was the primary
means of metabolism and was likely the case for the high observed
in vivo clearance. Subsequent efforts focused on further modifica-
tions to the scaffold to block the sites of glucuronidation while
maintaining potency and selectivity. This resulted in pyrimidine-
imidazopyridine 21e, which demonstrated improved in vivo clear-
ance and selectivity over both 1 and 2, while maintaining sufficient
cellular potency on mTOR substrate p4EBP1.
17. We considered the p4EBP1 cellular assay to be the most important measure of
mTOR potency, since it determined activity on mTOR signaling in the presence
of all the associated proteins that constitute the mTOR complex, whereas the
kinase assay consisted of only the mTOR kinase.
18. UN = Undefined: failed to achieve a hill fit in the assay, in this case, no
significant inhibition was observed.
19. X-ray coordinates deposited in the Cambridge Crystallographic Data Centre,
PDB code 4FHK.
20. (a) Smith, D. A.; Di, L.; Kerns, E. H. Nat. Rev. Drug Disc. 2010, 9, 929; (b)
Pellegatti, M.; Pagliarusco, S.; Solazzo, L.; Colato, D. Expert Opin. Drug Metab.
Toxicol. 2011, 9, 1009.
21. Maxwell, B. D.; Boye, O. G.; Ohta, K. J. Label. Comp. Radiopharm. 2005, 48, 397.
22. Koubachi, J.; Kazzouli, S. E.; Berteina-Raboin, S.; Mouaddib, A.; Guillaumet, G.
Synlett 2006, 3237.
23. For full experimental procedures, please see: Bode, C.; Boezio, A.; Cheng, A. C.;
Choquette, D.; Coats, J. R. Copeland, K. W.; Huang, H.; La, D.; Lewis, R. et al. PCT
Int. Appl. 2010, WO 2010132598.
24. Mosrin, M.; Knochel, P. Org. Lett. 1837, 2009, 11.
25. This procedure was also used to prepare pyrimidine imidazopyridine 20 and
related compounds.
26. This dichloride intermediate could be synthesized using the direct coupling
conditions outlined in Scheme 1, step a, or using the Knochel coupling
described in Scheme 2.