Synthesis and evaluation of a series of 4-azaindole-containing p21-activated kinase-1 inhibitors

Article abstract of DOI:10.1016/j.bmcl.2016.06.031

A series of 4-azaindole-containing p21-activated kinase-1 (PAK1) inhibitors was prepared with the goal of improving physicochemical properties relative to an indole starting point. Indole 1 represented an attractive, non-basic scaffold with good PAK1 affinity and cellular potency but was compromised by high lipophilicity (c log D = 4.4). Azaindole 5 was designed as an indole surrogate with the goal of lowering log D and resulted in equipotent PAK1 inhibition with a 2-fold improvement in cellular potency over 1. Structure–activity relationship studies around 5 identified additional 4-azaindole analogs with superior PAK1 biochemical activity (K_i<10 nM) and up to 24-fold selectivity for group I over group II PAKs. Compounds from this series showed enhanced permeability, improved aqueous solubility, and lower plasma protein binding over indole 1. The improvement in physicochemical properties translated to a 20-fold decrease in unbound clearance in mouse PK studies for azaindole 5 relative to indole 1.

Full text of DOI:10.1016/j.bmcl.2016.06.031

Bioorganic & Medicinal Chemistry Letters xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis and evaluation of a series of 4-azaindole-containing
p21-activated kinase-1 inhibitors
a
a
b
b
c
Wendy Lee a, , James J. Crawford , Ignacio Aliagas , Lesley J. Murray , Suzanne Tay , Weiru Wang ,
Christopher E. Heise d, Klaus P. Hoeflich e,y, Hank La b, Simon Mathieu a,à, Robert Mintzer d,
Sreemathy Ramaswamy d, Lionel Rouge c, Joachim Rudolph a
⇑
a Discovery Chemistry, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
b Drug Metabolism and Pharmacokinetics, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
c Structural Biology, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
d Biochemical and Cellular Pharmacology, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
e Translational Oncology, Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of 4-azaindole-containing p21-activated kinase-1 (PAK1) inhibitors was prepared with the goal
of improving physicochemical properties relative to an indole starting point. Indole 1 represented an
attractive, non-basic scaffold with good PAK1 affinity and cellular potency but was compromised by high
lipophilicity (clogD = 4.4). Azaindole 5 was designed as an indole surrogate with the goal of lowering
logD and resulted in equipotent PAK1 inhibition with a 2-fold improvement in cellular potency over 1.
Structure–activity relationship studies around 5 identified additional 4-azaindole analogs with superior
PAK1 biochemical activity (Ki <10 nM) and up to 24-fold selectivity for group I over group II PAKs.
Compounds from this series showed enhanced permeability, improved aqueous solubility, and lower
plasma protein binding over indole 1. The improvement in physicochemical properties translated to a
20-fold decrease in unbound clearance in mouse PK studies for azaindole 5 relative to indole 1.
Ó 2016 Elsevier Ltd. All rights reserved.
Received 13 May 2016
Revised 10 June 2016
Accepted 11 June 2016
Available online xxxx
Keywords:
PAK1
Kinase
4-Azaindole
Aminopyrazole
Diaminopyrimidine
The p21-activated kinases (PAKs) are serine-threonine protein
kinases in the sterile-20 (STE20) family and play significant roles
in cytoskeletal rearrangement, cellular morphogenesis, and
survival.1 There are six isoforms in the PAK family, divided into
two subfamilies based on sequence homology: PAK1–3 (group I)
and PAK4–6 (group II). Group 1 PAKs are activated by Cdc42 or
Rac1 interaction with GTPase-binding that overlaps with an
auto-inhibitory domain.1,2 The mechanism of activation of group
II PAKs is less clear, and its prominent difference from group I PAKs
is the lack of Rho GTPase-mediated activation.2,3 The most studied
isoform is PAK1, and its involvement in tumorigenesis and
metastasis is well validated and reviewed.1,4,5 PAK1 has emerged
as an attractive target for anticancer therapy and has become the
subject of significant drug discovery efforts in the past few years.6,7
PAK1 signaling is often upregulated in human tumors and PAK1
gene expression is elevated in various human cancers, including
subsets of breast, ovarian, and colorectal carcinoma. In recent years
there have been several studies suggesting therapeutic potential of
small molecule PAK1 inhibitors in breast cancer cell lines8 and in
ovarian cancers that are characterized by PAK1 gene amplifica-
tion.9 To date, no PAK inhibitor has advanced to proof-of-concept
studies in human. A PAK1 selective inhibitor would be a valuable
tool compound to test the efficacy and safety associated with
inhibiting group 1 PAK isoforms.
We have previously reported on an aminopyrazole scaffold-
based series of PAK1 inhibitors that harness the ribose pocket in
order to improve PAK1 affinity and obtain selectivity over Group
II PAKs.10 As part of this effort, indole 1 (Table 1) represented an
attractive, non-basic scaffold with high PAK1 affinity (Ki = 31 nM).
Compound 1 displayed activity in a phospho-S298-MEK1 (pMEK)
EBC-1 cellular assay (IC50 = 120 nM) and was used as a tool mole-
cule for mechanistic studies in the PAK1 pathway. Unfortunately,
this compound was highly lipophilic (clogD = 4.4)11 and was not
progressed due to the likelihood of ADME liabilities.12 Our expec-
tation was that the change from a phenyl to a pyridyl ring would,
as a result of decreasing lipophilicity, improve drug-like properties,
such as solubility and metabolic stability.13 We further postulated
⇑
Corresponding author. Tel.: +1 (650) 225 7704; fax: +1 (650) 742 4943.
E-mail address: wendyle@gene.com (W. Lee).
Present address: Department of Biology, Blueprint Medicines, 215 First Street,
y
Cambridge, MA 02142, USA.
à
Present address: Novartis Institutes for BioMedical Research, 250 Massachusetts
Avenue, Cambridge, MA 02139, USA.
http://dx.doi.org/10.1016/j.bmcl.2016.06.031
0960-894X/Ó 2016 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Lee, W.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.031

2
W. Lee et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
Table 1
SAR of indole isosteres as PAK1 inhibitors
NH
N
N
HN
N
R
d
Compd
R
PAK1a Ki (nM)
PAK4/PAK1 selectivity
1.4
LLEb
3.1
cLogDc
Calcd pKa of indole NH
p-MEKe IC50 (nM)
N
H
1
31
4.4
15.6
14.6
120
N
H
N
H
2
70
1.6
3.7
3.2
—
N
H
N
N
N
H
3
4
120
360
4.2
1.6
3.9
3.2
3.0
3.0
13.7
—
—
—
N
H
N
H
N
N
N
N
H
5
6
32
5
2.9
4.8
3.8
4.8
3.6
3.4
13.1
13.9
51
68
N
H
N
H
N
N
H
a
b
c
PAK1 assay results represent the geometric mean of a minimum of two determinations performed in duplicate.
Ligand-lipophilicity efficiency (LLE) was calculated using the equation: ꢀlog10 (PAK Ki) ꢀ clogD (pH7.4).
The clogD at pH7.4 was calculated using the Moka software (version 1.2.0, Molecular Discovery).
Calculated using MoKa software (version 1.2.0, Molecular Discovery).
d
e
p-MEK assay results represent the geometric mean of a minimum of two determinations.
that the additional nitrogen might improve PAK1 potency by
reducing the pKa of the donor NH of the indole, allowing for a
stronger hydrogen-bonding interaction with the backbone
carbonyl of Asp393 in the ribose pocket.14
the space alongside the (Met) gatekeeper side chain. In the tail
area, the methyl group to the pyrrolopyridinyl amine fits well
a
within the receptor surface and points toward the glycine-rich loop
(P-loop) when in the (S)-configuration (Fig. 1B). In contrast, the
(R)-enantiomer loses 21-fold PAK1 affinity (PAK1 IC50 = 685 nM).
Modeling of this analog in the X-ray structure of 5 suggested the
(R)-methyl moiety would clash with the aminopyrazole head
group and adopt a more strained conformation.
Since 5 had a favorable physicochemical profile with a lower
logD (clogD/logD7.4 = 3.6/2.7) and good cellular potency, we next
set out to evaluate the SAR around the 4-azaindole motif. Table 2
summarizes the SAR of aminopyrazole analogs with various substi-
tuted 4-azaindole tails (7–10). Modeling indicated that extending
We first prepared simple azaindoles 2 and 3 (Table 1). Although
3 was not as potent in our PAK1 biochemical assay, we were
encouraged by the modest gain in ligand-lipophilicity efficiency
(3: LLE = 3.9)15 and the improved selectivity over PAK4 (3: 4.2-fold;
PAK4 Ki = 505 nM) relative to 1. The loss in biochemical potency
with imidazopyridine 4 underscored the importance of the indole
NH as a hydrogen donor moiety in PAK1. Given these factors and
the favorable calculated pKa value of the azaindole NH in 3 relative
to 2, we decided to incorporate the methyl group of 1 with the
4-azaindole analog. Azaindole 5 was found to be equipotent to 1
the (S)-a-methyl vector might be tolerated due to the large degree
in PAK1, with
a
gain in ligand–lipophilicity efficiency (5:
of flexibility in the glycine-rich loop in the PAK1 kinase domain.17
Elongation to an ethyl group as in 7 resulted in retained PAK1
biochemical potency with a small improvement in selectivity over
PAK4 (6.5-fold). Further exploration at this position was not
pursued due to the decrease in LLE as a result of higher clogD
(4.1) coupled with a 3-fold drop in cellular potency. We next inves-
tigated the SAR of 4-azaindole analogs that were substituted with
halogens. The 6-fluoro-azaindole 8 had significantly improved
PAK1 affinity (Ki = 4 nM) with similar potency to 5 in the cell-based
assay. However, the corresponding 6-chloro analog 9 led to a
30-fold drop in biochemical potency (Ki = 119 nM) compared to
8. This likely resulted from an altered conformational preference
due to a steric clash between the chloro substituent and its adja-
cent amino methine linker. Interestingly, the 3-chloro azaindole
10 was equipotent to 8, indicating that incorporation of a halogen
on the pyrrole portion of the azaindole could also improve PAK1
LLE = 3.8, 1: LLE = 3.1) and a 2-fold increase in selectivity for
PAK1 over PAK4 (5: PAK4/PAK1 = 2.9, 1: PAK4/PAK1 = 1.4). Com-
pound 5 also showed greater than two-fold improvement in pMEK
(S298) cellular activity. 6-Azaindole 6 completed the nitrogen walk
around the phenyl ring. Although 6 showed a 6-fold improvement
in biochemical potency relative to 1, the pMEK (S298) cellular
activity for 6 was equipotent to 5 with a greater biochemical-to-
cell shift (13.6-fold) relative to 5 (1.6-fold).
To enable further optimization of this 4-azaindole motif, we
obtained a co-crystal structure of
5 in complex with PAK1
(Fig. 1). The pyrazole amine makes the expected donor–acceptor–
donor hydrogen bond interactions with kinase hinge residues
Glu345 and Leu347, in addition to a favorable hydrogen bond
interaction measuring 2.8 Å from the azaindole NH to the back-
bone carbonyl of Asp393. The cyclopropyl moiety efficiently fills
Please cite this article in press as: Lee, W.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.031

W. Lee et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
3
front pocket alongside the (Met) gatekeeper.10 Fluorine substitu-
tion is a commonly used measure to enhance biological activity
and potentially increase metabolic stability of drug molecules.19
We postulated that the incorporation of a mono-alkyl fluorine
could further improve the physicochemical properties of our
analogs by decreasing lipophilicity.20 All four fluoro-substituted
isomers at the 2-position of the cyclopropyl pyrazole head group
were studied (11, 12, 13, and 14). The trans (1R,2S)-fluoro
cyclopropyl analog 11 showed superior PAK4 selectivity (11-fold)
without any detriment to PAK1 biochemical potency (Ki = 11 nM)
or cellular potency (IC50 = 77 nM). The trans (1S,2R)-fluoro cyclo-
propyl (14) diastereomer was also active in the PAK1 biochemical
(Ki = 25 nM) and p-MEK cellular (IC50 = 92 nM) assays, but less
selective against PAK4 (4.8-fold) relative to 11. Other derivatives
such as the cis-2-fluorocyclopropyls (12, 13) and the difluorocy-
clobutyl analog 15 exhibited decreased PAK1 potencies and offered
no advantages over 5. We also masked the seemingly nonessential
hydrogen at the secondary amino linker to the pyrimidine ring.
While the N-methylated amino linked compound (16) had superior
PAK1 potency (Ki = 18 nM) without compromising selectivity
versus PAK4 (7-fold), it showed decreased cellular potency.
Prior to conducting in vivo mouse pharmacokinetics (PK) with
the azaindole inhibitors, we profiled in vitro ADME properties for
several selected examples. The molecules selected exhibited potent
p-MEK(S298) cellular activity of less than 100 nM (Table 4). Based
on in vitro metabolic stability, we were unable to differentiate
between the indoles and azaindoles since all of the selected com-
pounds were relatively labile in mouse liver microsomes (MLM)
and moderate to labile in human liver microsomes (HLM). On the
other hand, the 4-azaindoles (5, 8, 11, 14) exhibited improved
permeability (MDCK Papp A:B P 3.7) over indole 1 (MDCK Papp A:
B, 2.0) and 6-azaindole 6 (MDCK Papp A:B, 0.9). Mouse plasma
protein binding for the azaindoles ranged from 95.3% to 98.6%
bound (fu = 0.047–0.014), representing a significant decrease over
the unbound fraction of indole 1 (fu = 0.008). Furthermore, with
the exception of compound 8, the azaindoles showed improved
aqueous solubility (73–133 lM) relative to indole 1 (69 lM). Taken
together, inhibitor 5 exhibited the best combination of cellular
potency, MDCK permeability, plasma protein binding, and aqueous
solubility and was progressed to in vivo mouse PK studies.
Table 5 shows the pharmacokinetic (PK) profile of indole 1 and
4-azaindole 5 in mice. Indole 1 exhibited clearance well above
hepatic blood flow (mouse iv CLp 160 mL/min/kg), which is
under-predicted by in vitro liver microsome and hepatocyte stabil-
ity data (MLM CLhep = 76 mL/min/kg, MH CLhep = 56 mL/min/kg),
suggesting that extra-hepatic clearance mechanisms play a role.
In contrast, the in vivo mouse clearance observed for 5 (mouse iv
CLp, 48 mL/min/kg) was comparable to that extrapolated from
in vitro mouse hepatocyte clearance data (MH CLhep = 52 mL/min/
kg). When corrected for protein binding, the unbound mouse
clearance in 5 (CLu = 1067 mL/min/kg) is about 20-fold lower than
1 (CLu = 20,000 mL/min/kg). Inhibitor 5 also showed measurable,
albeit low, oral bioavailability (% F = 4.8).
To aid future compound design, we examined compounds 1 and
5 in additional DMPK assays. Our main intent was to identify the
major routes of elimination and to understand the discrepancy
between in vitro prediction and in vivo observation of 1. For
example, transporter-mediated hepatic uptake could be a rate-
determining step in the hepatic clearance that could cause
under-prediction of clearance extrapolated from in vitro metabolic
stability assays.21 In our studies, no active uptake was observed in
mouse hepatocytes for 1 and 5, indicating that hepatic transporters
were not involved in the elimination process.22 In a cytochrome
P450 (CYP) assay using mouse liver microsomes (MLM) incubated
with and without 1-aminobenzotriazole (ABT, an inactivator of
cytochrome P450 enzymes used to discern CYP- from non
Figure 1. Co-crystal structure of inhibitor 5 (in yellow) in complex with PAK1
displayed from two different orientations.16 (A) Hydrogen bond interactions are
indicated with dotted red lines. The ligand forms hydrogen bond interactions with
kinase hinge residues Glu345 and Leu347, and a crucial interaction between the
azaindole NH to a backbone carbonyl of Asp393 in the ribose pocket. (B) Different
orientation of inhibitor 5 shows the (S)-a-methyl moiety pointing towards the
glycine-rich loop (P-loop).
potency. A possible explanation is that lowering the pKa of the
donor NH relative to 5 increased its H-bond donor strength.14
Compound 10 also exhibited the best PAK1 over PAK4 selectivity
(24-fold) for this sub-series. We attributed this selectivity to
different spatial constraints in the ribose binding pockets of PAK1
and PAK4.18 While Asn294 in PAK1 allows Asp407 to point away
from the ligand, the corresponding PAK4 residue Ser445 in PAK4
orients the matching Asp group (Asp458) towards the ligand.
Molecular modeling studies of 10 in PAK4 showed that the chloro
substituent at the 3-azaindole position could clash with Asp458,
possibly forcing the molecule to adopt
a more strained
conformation and preventing an optimal NH bond interaction with
the backbone carbonyl group of Asp444.
We also combined the azaindole tail in 5 with different
substituents around the aminopyrazole head group to determine
whether potency and PAK1 selectivity could be further increased
(Table 3). Previous SAR studies had demonstrated that small
carbocycles, such as cyclopropyl or cyclobutyl, fit well in the deep
Please cite this article in press as: Lee, W.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.031

4
W. Lee et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
Table 2
Effect of substitutions around the azaindole motif
NH
N
N
HN
N
R
e
Compd
R
PAK1a Ki (nM)
PAK4/PAK1 selectivity
2.9
LLEb
3.8
clogDc
p-MEKd IC50 (nM)
Calc. pKa of indole NH
N
N
H
5
32
3.6
51
13.1
13.1
N
H
N
N
N
N
H
7
23
6.5
3.7
4.1
180
N
H
N
H
8
4
120
6
6
4
4.6
2.4
4.1
3.9
4.4
4.1
57
12.5
12.7
10.8
N
H
F
N
H
9
190
160
N
H
Cl
Cl
N
N
H
10
24
N
H
a
b
c
PAK1 assay results represent the geometric mean of a minimum of two determinations performed in duplicate.
Ligand-lipophilicity efficiency (LLE) was calculated using the equation: ꢀlog10 (PAK Ki) ꢀ clogD (pH7.4).
The clogD at pH7.4 was calculated using the Moka software (version 1.2.0, Molecular Discovery).
p-MEK assay results represent the geometric mean of a minimum of two determinations.
Calculated using MoKa software (version 1.2.0, Molecular Discovery).
d
e
Table 3
Structure–activity relationships of 4-azaindoles
NH
N
N
R1
HN
N
N
N
R2
N
H
Compd
R1
R2
H
PAK1a Ki (nM)
PAK4/PAK1 selectivity
LLEb
cLogDc
3.6
p-MEKd IC50 (nM)
5
32
11
2.9
3.8
3.6
51
77
F
F
F
F
11
H
11.4
3.4
12
13
14
H
H
H
120
84
2.8
5.7
4.8
3.5
3.7
4.2
3.4
3.4
3.4
810
270
92
25
F
15
16
H
68
18
6.6
7
3.1
3.6
4.1
4.1
250
240
F
Me
a
b
c
PAK1 assay results represent the geometric mean of a minimum of two determinations performed in duplicate.
Ligand-lipophilicity efficiency (LLE) was calculated using the equation: ꢀlog10 (PAK Ki) ꢀ clogD (pH 7.4).
The clogD at pH7.4 was calculated using the Moka software (version 1.2.0, Molecular Discovery).
p-MEK assay results represent the geometric mean of a minimum of two determinations.
d
Please cite this article in press as: Lee, W.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.031

W. Lee et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
5
Table 4
In vitro ADME properties of indole 1 and selected azaindole PAK1 inhibitors
a
Compd
pMEK (S298) IC50 (nM)
MLMb/HLMc CLhep (mL/min/kg)
MDCK Papp A:B (10ꢀ6 cm/s)d
Mouse PPB (fu)e
Solubilityf (
lM)
1
5
6
8
11
14
120
51
68
57
72
92
76/13
76/13
77/11
76/12
78/15
78/15
2.0
4.3
0.9
8.3
3.7
4.2
0.008
0.045
0.039
0.014
—
69
133
73
58
104
73
0.047
a
b
c
d
e
f
p-MEK assay results represent the geometric mean of a minimum of two determinations.
Mouse hepatic clearance predicted from mouse liver microsomes intrinsic CL.
Human hepatic clearance predicted from human liver microsomes intrinsic CL.
Permeability in MadinꢀDarby Canine Kidney (MDCK) cells, apical-to-basolateral flux.
Mouse plasma protein binding expressed as fraction unbound.
Kinetic aqueous solubility at pH 7.4 as determined from a single experiment in a high-throughput solubility assay.
Table 5
Comparison of in vivo PK properties of indole 1 and azaindole 5 in nude mice.a
Dosea (mg/kg)
Cmax
(lM)
Tmax (h)
Vss (L/kg)d
CLp/CLu (mL/min/kg)
t1/2 (h)
% Fg
MHh
b
c
e
f
Route
Compd
iv
iv
po
1
5
5
2
2
25
—
—
4.7
—
—
0.58
4.50
1.65
—
160/20,000
48/1067
—
0.32
0.63
—
—
—
4.8
56
52
—
a
b
c
Formulations used: iv 40:60 PEG400/H2O (solution); po 60:40 PEG400/H2O.
Maximum plasma concentration.
Time at which maximum plasma concentration was observed.
Volume of distribution at steady-state.
Clearance unbound = CLp/fu.
Half-life.
Oral bioavailibility.
Mouse hepatic clearance predicted from in vitro mouse hepatocytes data.
d
e
f
g
h
Table 6
first-pass clearance, moderate passive permeability, and P-gp-
mediated efflux that resulted in the low oral bioavailability of 5.
The preparation of the diaminopyrimidines 5 and 6 is depicted in
Scheme 1; this general route was used for the synthesis of
diaminopyrimidines 1–16. Commercially available cyclopropyl-
1H-pyrazol-5-amine 17 was reacted with 2,4-dichloropyrimidine
via an SNAr reaction to yield the corresponding monofunctionalized
Additional in vitro ADME profiling of compounds 1 and 5
CYP-mediated MLM ABTa
% 1 remaining
% 5 remaining
after 30 minb
after 30 minb
MLM ꢀ NADPH
MLM + NADPH
MLM + ABT
81 13
14 0.94
59 5.9
98 8.5
10 2.8
94 13
pyrimidine 18. Additional mono-substituted
pyrimidine
MDCK-MDR1c
intermediates were prepared via this method using mono-fluoro-
cyclopropyl-1H-pyrazol-5-amines24 or commercially available
1
5
(3,3-difluorocyclobutyl)-1H-pyrazol-5-amine.
A
second SNAr
Papp A:B (10ꢀ06 cm/s)
Papp B:A (10ꢀ06 cm/s)
Efflux ratio
0.5
14.6
27.7
0.2
9.8
59
reaction on 18 with bicyclic amines such as 19 afforded the desired
diaminopyrimidine 5. Compounds 1–4 were prepared from
commercially available bicyclic amines. In cases where the
azaindoles were N-protected, acid or base mediated deprotection
was required as a final step. For racemic amines such as 20,25 a chiral
a
CYP-mediated mouse liver microsomes ABT (1-aminobenzotriazole). Com-
pounds were preincubated with NADPH for 20 min before addition of 1 mM of ABT.
Determined by n = 3.
Human P-glycoprotein (P-gp) transfected in Madin Darby canine kidney
(MDCK-MDR1) cell permeability assay.
b
c
separation was required to give the desired (S)-
stereoisomer, as with 6.
a-methyl
For the synthesis of 19, we employed a stereospecific approach to
install the (S)- -methyl group using Ellman’s chiral auxiliary as
a
shown in Scheme 2.26 Stille coupling of tri-n-butyl-1-ethoxyvinyl-
stannane with Boc-protected 5-chloro-1H-pyrrolo[3,2-b]pyridine
22 provided the 1-ethoxyvinyl derivative, which could be readily
hydrolyzed to the corresponding ketone 23a. The N-Boc-protected
ketone 25a was then condensed with (R)-2-methylpropane-2-sulfi-
namide. Reduction of the sulfinamide and acid-mediated removal of
the chiral functionality afforded the desired (S)-1-(1H-pyrrolo[3,2-
b]pyridin-5-yl)ethan-1-amine, 28a, as the HCl salt. Using this
method, the 3-chloro derivative 28b and the one-carbon elongated,
CYP-mediated metabolism), we observed a significant reduction in
disappearance of the parent compounds after 30 min incubation in
the presence of ABT, suggesting that P450-mediated metabolism
was responsible for the in vitro mouse clearance observed for 1
and 5 (Table 6). In a permeability assay with human P-glycoprotein
(P-gp) transfected in Madin Darby canine kidney (MDCK-MDR1)
cells (Table 6), both 1 and 5 exhibited low permeability (Papp A:B
1: 0.5 ꢁ 10ꢀ6 cm/s; 5: 0.2 ꢁ 10ꢀ6 cm/s) with high efflux ratios (B:
A/A:B, 1: 27.7; 5: 59), indicating that these compounds are P-gp
substrates. It is known that for compound with low permeability,
(S)-a-ethyl, analog 28c were prepared from 25b and 25c, respec-
tively. Intermediate 25b was synthesized via chlorination of 23a
with N-chlorosuccinimide followed by Boc protection. A direct pal-
ladium-catalyzed cyanation of 22 afforded 24. Grignard reaction on
24 using ethyl magnesium bromide afforded intermediate 25c
P-gp efflux could play
a role in limiting the absorption of
compounds by driving the drug back into the lumen.23 Thus, it
was likely a combination of the factors outlined above, moderate
Please cite this article in press as: Lee, W.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.031

6
W. Lee et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
N
NH
N
N
H2N
HCl
N
H
HN
N
NH
N
N
19
HN
N
a
b
NH
N
N
N
H
H2N
N
H
Cl
17
18
5
H2N
c
N
N
Boc
20
NH
N
HN
N
N
N
H
N
N
H
6
Scheme 1. Preparation of diaminopyrimidines. Reagents and conditions: (a) 2,4-dichloropyrimidine, DMA, Et3N, 120 °C, 16 h; (b) 19, DIPEA, nBuOH, 115 °C, sealed tube, 18 h,
84% yield (2-steps); (c) (i) 20, DIPEA, nBuOH, 115 °C, sealed tube, 18 h; (ii) 4 N HCl in dioxane, MeOH, 40 °C, chiral separation, 19% yield (2-steps).
O
O
X
X
N
N
Cl
N
Cl
N
a
b
e
c
f
g
R
N
N
Boc
N
H
N
Boc
H
21
22
23a: R = Me, X = H
25a: R = Me, X = H
d
23b
25b
: R = Me, X = Cl
: R = Me, X = Cl
25c: R = Et, X = H
NC
N
N
Boc
24
O
S
R
R
R
X
X
X
N
N
N
h
i
H2N
H2N
HCl
N
N
Boc
N
H
N
Boc
28a: R = Me, X = H
28b: R = Me, X = Cl
27a: R = Me, X = H
27b: R = Me, X = Cl
26a: R = Me, X = H
26b: R = Me, X = Cl
28c
27c
: R = Et, X = H
26c
: R = Et, X = H
: R = Et, X = H
O
NH
N
N
j
N
Boc
N
Boc
29
25a
Scheme 2. Enantioselective synthesis of 28 and analogs thereof. Reagents and conditions: (a) di-tert-butyl dicarbonate, DMAP, ACN, 0 °C to rt, 99%; (b) (i) tributyl(1-
ethoxyvinyl)stannane, Pd(PPh3)4, DMF, 100 °C, sealed tube; (ii) 2 N HCl, THF, rt, 64%; (c) di-tert-butyl dicarbonate, DMAP, ACN, 0 °C to rt, 82–88%; (d) N-chlorosuccinimide,
DMF, 0 °C to rt, 96%; (e) zinc cyanide, zinc powder, Pd2(dba)3, 1,10-bis(diphenylphosphino)ferrocene, N-methylmorpholine, 180 °C, 75%; (f) ethyl magnesium bromide (2 M in
THF), THF, ꢀ20 to 10 °C, 64%; (g) (R)-2-methylpropane-2-sulfinamide, Ti(IV)(OEt)4, THF, 75 °C, 33–54%; (h) L-SelectrideÒ (1 M in THF), THF, 0 °C to rt, 73–81%; (i) 4 N HCl in
dioxane, MeOH, 40 °C, 99%; (j) methylamine hydrochloride, sodium cyanoborohydride, EtOH, 130 °C, microwave, 94%.
following aqueous workup. A final SNAr reaction, as depicted in
Scheme 1, using 18 with 28b or 28c afforded the diaminopyrimidi-
nes 10 or 7, respectively. For compound 16, the N-methylated
azaindole precursor 29 was prepared via a reductive amination of
methyl ketone 25a with methylamine.
protected with a tosyl group, followed by a modified Reissert-
Henze reaction27 to give the cyano products 33a and 33b. A
Grignard reaction on 33a and 33b afforded methyl ketones 34a
and 34b. Installation of the
a-methyl amino group was accom-
plished by condensation of the ketones with hydroxylamine to
form oximes and subsequent reduction to the primary amines gave
the requisite intermediates 35a and 35b. Completion of the
The synthesis of 6-substituted fluoro- and chloro-azaindoles is
outlined in Scheme 3. The indolyl nitrogen of 30a and 30b was
Please cite this article in press as: Lee, W.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.031

W. Lee et al. / Bioorg. Med. Chem. Lett. xxx (2016) xxx–xxx
7
O
N
N
NC
R
N
N
a
b
N
S
N
S
c
N
S
d
R
R
N
H
O
O
O
R
O
O
O
30a: R = F
31a: R = F
32a: R = F
33a: R = F
33b
: R = Cl
30b
31b
32b
: R = Cl
: R = Cl
: R = Cl
O
R
NH2
NH
N
N
N
N
e
f, g, h
HN
N
N
S
N
S
R
O
O
N
O
O
N
H
N
H
R
34a: R = F
34b
35a: R = F
35b
8: R = F
9
: R = Cl
: R = Cl
: R = Cl
Scheme 3. Preparation of 6-halogenated-4-azaindoles. Reagents and conditions:(a) (i) sodiumhydride, THF, 0 °C, 30 min; (ii)p-toluenesulfonylchloride, 0 °Cto rt, 18 h, 84–89%;
(b) mCPBA, DCM, 0 °C to rt, 2 h, 90–95%; (c) dimethylcarbamyl chloride, trimethylsilyl cyanide, DCE, 80 °C, 20 h, 53–56%;(d) methyl magnesiumchloride (1 M in THF), THF, ꢀ10 to
0 °C, 1.5 h, 18–20%; (e) (i) hydroxylamine hydrochloride, sodium acetate, EtOH, rt, 18 h; (ii) zinc powder, ammonium chloride, AcOH, MeOH, reflux, 20 h, 78–81% (2-steps); (f) 18,
2,4-dimethyl-3-pentanol, DIPEA, 140 °C, 18 h; (g) chiral SFC separation; (h) KOH (2 N), MeOH, 100 °C, 3 h, 26–30%.
Haverty, P. M.; Kowanetz, M.; Yan, Y.; Tremayne, J.; Lisle, R.; Harris, A. L.;
Friedman, L. S.; Belvin, M.; Middleton, M. R.; Blackwood, E. M.; Koeppen, H.;
Hoeflich, K. P. J. Natl. Cancer Inst. 2013, 105, 606.
syntheses following the established route of base-mediated SNAr
reaction, chiral separation, and protecting group removal gave
the desired 6-halogenated azaindoles 8 and 9, respectively.
In summary, we were able to improve the physicochemical
properties through lowering lipophilicity of our aminopyrazole
scaffold-based PAK1 inhibitor 1 by replacing the indole moiety with
a 4-azaindole motif. A stereospecific approach to the installation of
6. (a) Crawford, J. J.; Hoeflich, K. P.; Rudolph, J. Expert Opin. Ther. Pat. 2012, 22,
293; (b) Rudolph, J.; Crawford, J. J.; Hoeflich, K. P.; Chernoff, J. Enzymes 2013, 34,
157; (c) Rudolph, J.; Crawford, J. J.; Hoeflich, K. P.; Wang, W. J. Med. Chem. 2015,
58, 111; (d) Xu, Y.; Foulks, J. M.; Clifford, A.; Brenning, B.; Lai, S.; Luo, B.; Parnell,
K. M.; Merx, S.; McCullar, M. V.; Kanner, S. B.; Ho, K. Bioorg. Med. Chem. Lett.
2013, 23, 4072.
7. Ndubaku, C. O.; Crawford, J. J.; Drobnick, J.; Aliagas, I.; Campbell, D.; Dong, P.;
Dornan, L. M.; Duron, S.; Epler, J.; Gazzard, L.; Heise, C. E.; Hoeflich, K. P.;
Jakubiak, D.; La, H.; Lee, W.; Lin, B.; Lyssikatos, J. P.; Maksimoska, J.;
Marmorstein, R.; Murray, L. J.; O’Brien, T.; Oh, A.; Ramaswamy, S.; Wang, W.;
Zhao, X.; Zhong, Y.; Blackwood, E.; Rudolph, J. ACS Med. Chem. Lett. 2015, 6, 1241.
8. Ong, C. C.; Gierke, S.; Pitt, C.; Sagolla, M.; Cheng, C. K.; Zhou, W.; Jubb, A. M.;
Strickland, L.; Schmidt, M.; Duron, S. G.; Campbell, D. A.; Zheng, W.; Dehdashti,
S.; Shen, M.; Yang, N.; Behnke, M. L.; Huang, W.; McKew, J. C.; Chernoff, J.;
Forrest, W. F.; Haverty, P. M.; Chin, S.-F.; Rakha, E. A.; Green, A. R.; Ellis, I. O.;
Caldas, C.; O’Brien, T.; Friedman, L. S.; Koeppen, H.; Rudolph, J.; Hoeflich, K. P.
Breast Cancer Res. 2015, 17, 1.
9. Prudnikova, T. Y.; Villamar-Cruz, O.; Rawat, S. J.; Cai, K. Q.; Chernoff, J. Oncogene
2015, 1, 1.
10. Crawford, J. J.; Lee, W.; Aliagas, I.; Mathieu, S.; Hoeflich, K. P.; Zhou, W.; Wang,
W.; Rouge, L.; Murray, L.; La, H.; Liu, N.; Fan, P. W.; Cheong, J.; Heise, C. E.;
Sreemathy, R.; Mintzer, R.; Liu, Y.; Chao, Q.; Rudolph, J. J. Med. Chem. 2015, 58,
5121.
11. The measured logD for indole 1 is 3.7. We found that our calculated logD was
over-estimated by 0.5–1.0log unit. For design, we used clogD to compare the
compounds in a relative rather than absolute sense.
the (S)-a-methyl group of the azaindole tail was developed that uti-
lized Ellman’s chiral auxiliary. This resulted in the identification of 5,
representing a 2-fold improvement in cellular potency and a slight
enhancement in selectivity against PAK4. Modulation of the pKa of
the aza-indolic NH by incorporating a halogen led to additional
analogs with superior PAK1 biochemical activity (Ki <10 nM) and
increased selectivity versus PAK4 (up to 24-fold). As expected, the
decrease in lipophilicity of azaindole 5 (logD = 2.7) contributed to
improved aqueous solubility and lower plasma protein binding than
indole 1 (logD = 3.7). In mouse PK studies, azaindole 5 displayed
moderate in vivo clearance (48 mL/kg/min) and low oral bioavail-
ability (% F = 4.8). When corrected for protein binding, the unbound
mouse clearance for 5 (CLu = 1067 mL/min/kg) was nearly 20-fold
lower than 1 (CLu = 20,000 mL/min/kg).
12. Arnott, J. A.; Planey, S. L. Expert Opin. Drug Discov. 2012, 7, 863.
13. Blaazer, A. B.; Lange, J. H. M.; Van der Neut, M. A. W.; Mulder, A.; Den Boon, F.
S.; Werkman, T. R.; Kruse, C. G.; Wadman, W. J. Eur. J. Med. Chem. 2011, 46,
5086.
14. Gilli, P.; Pretto, L.; Bertolasi, V.; Gilli, G. Acc. Chem. Res. 2009, 42, 33.
15. Leeson, P. D.; Springthorpe, B. Nat. Rev. Drug Disc. 2007, 6, 881.
16. See Supplementary material for experimental detail describing PAK1 crystal
experiment.
17. (a) Lei, M.; Lu, W.; Meng, W.; Parrini, M.; Eck, M. J.; Mayer, B. J.; Harrison, S. C.
Cell 2000, 102, 387; (b) Meggers, E. Angew. Chem., Int. Ed. 2011, 50, 2442.
18. See Ref. 10 for an explanation of the PAK1 versus PAK4 enzymes and
comparison of crystal structures of a PAK1 inhibitor in PAK1 versus PAK4
enzyme.
Acknowledgement
We thank the purification and analytical groups within
Genentech Small Molecule Drug Discovery for their support.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2016.06.031.
19. (a) Purser, S.; Moore, P. R.; Swallow, S.; Gouverneur, V. Chem. Soc. Rev. 2008, 37,
320; (b) Shah, P.; Westwell, A. D. J. Enzyme Inhib. Med. Chem. 2007, 22, 527.
20. (a) Gillis, E. P.; Eastman, K. J.; Hill, M. D.; Donnelly, D. J.; Meanwell, N. A. J. Med.
Chem. 2015, 58, 8315; (b) Smart, B. E. J. Fluorine Chem. 2001, 109, 3.
21. Chiba, M.; Ishii, Y.; Sugiyama, Y. AAPS J. 2009, 11, 262.
References and notes
1. Radu, M.; Semenova, G.; Kosoff, R.; Chernoff, J. Nat. Rev. Cancer 2014, 14, 13.
2. Baskaran, Y.; Ng, Y. W.; Selamat, W.; Ling, F. T. Y.; Manser, E. EMBO Rep. 2012,
13, 653.
22. See Supplementary material for data.
3. Ha, B. H.; Morse, E. M.; Turk, B. E.; Boggon, T. J. J. Biol. Chem. 2015, 290, 12975.
4. King, H.; Nicholas, N. S.; Wells, C. M. Int. Rev. Cell Mol. Biol. 2014, 309, 347.
5. (a) Huynh, N.; Liu, K. H.; Baldwin, G. S., et al. Biochim. Biophys. Acta 2010, 1803,
1106; (b) Ong, C. C.; Jubb, A. M.; Haverty, P. M.; Zhou, W.; Tran, V.; Truong, T.;
Turley, H.; O’Brien, T.; Vucic, D.; Harris, A. L.; Belvin, M.; Friedman, L. S.;
Blackwood, E. M.; Koeppen, H.; Hoeflich, K. P. Proc. Natl. Acad. Sci. U.S.A. 2011,
108, 7177; (c) Ong, C. C.; Jubb, A. M.; Jakubiak, D.; Zhou, W.; Rudolph, J.;
23. Chan, K. M. S.; Lowes, S.; Hirst, B. H. Eur. J. Pharm. Sci. 2004, 21, 25.
24. See Supplementary material for scheme describing the preparation of mono-
fluoro-cyclopropyl-1H-pyrazol-5-amines.
25. See Supplementary material for scheme describing the synthesis of tert-butyl
5-(1-aminoethyl)-1H-pyrrolo[3,2-b]pyridine-1-carboxylate 20.
26. Tanuwidjaja, J.; Peltier, H. M.; Ellman, J. J. Org. Chem. 2007, 72, 626.
27. Fife, W. K. J. Org. Chem. 1983, 48, 1375.
Please cite this article in press as: Lee, W.; et al. Bioorg. Med. Chem. Lett. (2016), http://dx.doi.org/10.1016/j.bmcl.2016.06.031