Identification and optimisation of 7-azaindole PAK1 inhibitors with improved potency and kinase selectivity

Article abstract of DOI:10.1039/c4md00280f

A novel series of PAK1 inhibitors was discovered from a kinase directed screen. SAR exploration in the selectivity pocket and solvent tail regions was conducted to understand and optimise PAK1 potency and selectivity against targeted kinases. A liganded PAK1 crystal structure was utilised to guide compound design. Permeability and kinase selectivity impacted the translation of enzyme to cellular PAK1 potency. Compound 36 (AZ-PAK-36) demonstrated improved Gini coefficient, good PAK1 cellular potency and has utility as a tool compound for target validation studies. This journal is

Full text of DOI:10.1039/c4md00280f

MedChemComm
View Article Online
View Journal | View Issue
CONCISE ARTICLE
Identification and optimisation of 7-azaindole PAK1
inhibitors with improved potency and kinase
selectivity†
Cite this: Med. Chem. Commun., 2014,
5, 1533
a
b
William McCoull, Edward J. Hennessy, Kevin Blades,a Matthew R. Box,a
Claudio Chuaqui,b James E. Dowling,b Christopher D. Davies,a Andrew D. Ferguson,b
Frederick W. Goldberg,a Nicholas J. Howe,a Paul D. Kemmitt,a Gillian M. Lamont,a
Katrina Madden,a Claire McWhirter,a Jeffrey G. Varnes,b Richard A. Ward,a
Jason D. Williamsa and Bin Yangb
*
*
A novel series of PAK1 inhibitors was discovered from a kinase directed screen. SAR exploration in the
selectivity pocket and solvent tail regions was conducted to understand and optimise PAK1 potency and
selectivity against targeted kinases. A liganded PAK1 crystal structure was utilised to guide compound
design. Permeability and kinase selectivity impacted the translation of enzyme to cellular PAK1 potency.
Compound 36 (AZ-PAK-36) demonstrated improved Gini coefficient, good PAK1 cellular potency and
has utility as a tool compound for target validation studies.
Received 27th June 2014
Accepted 7th August 2014
DOI: 10.1039/c4md00280f
www.rsc.org/medchemcomm
widely used PAK1 inhibitors to date are PF-3758309 (ref. 11) and
FRAX-597 (ref. 12) (Fig. 1), and they have both shown efficacy in
a KRAS-driven tumour mouse model of skin cancer.13 However
both have sub-optimal kinase selectivity and indeed the former
is an inhibitor of both group I and II PAKs. Thus, to understand
the role of group I PAKs more fully in oncology, small molecules
with improved potency and kinase selectivity are required. In
this paper we report the discovery and optimisation of a novel
7-azaindole series of PAK1 inhibitors with improved selectivity.
Introduction
The p21-activated kinases (PAKs) are a family of six serine/
threonine-specic intracellular protein kinases which are posi-
tioned at the intersection of several signaling pathways of
importance in cancer progression.1,2 The PAK family is divided
into two subgroups based on sequence homology: group I (PAKs
1–3) and group II (PAKs 4–6). Group I PAKs have high sequence
identity in the kinase domain and possess an auto-inhibitory
domain which is relieved by binding of the GTP-binding
proteins Rac or Cdc42, whilst group II PAKs have lower kinase
domain homology and are not activated by Rho GTPases.3
Group I PAKs are overexpressed in a wide variety of cancers and
PAK1 is commonly overexpressed in breast tumours.4 In
tumours characterised by neurobromatosis type 2 (NF2)
inactivation, group I PAKs have been shown to be hyper-
activated.5 Consequently, small molecule inhibitors of group I
PAKs may have therapeutic efficacy in tumours characterised by
PAK activation.
Results and discussion
A kinase focused subset screen (120 000 compounds) of the
AstraZeneca compound collection was conducted, from which
we identied 7-azaindole 1 as a modestly potent inhibitor of
PAK1 with 14-fold selectivity over PAK4 (Table 1). This screening
set consisted of compounds which had previously shown
activity in at least one kinase assay and was ltered by lead-like
properties followed by a diversity-based selection. We did not
expect to obtain selectivity within each of the PAK families, and
so PAK1 was utilised as our primary group I PAK target while
There are relatively few published reports of PAK inhibitors,
including the patent literature.6–10 Potent PAK inhibition with
high kinase selectivity has proven difficult to achieve due to the
large and exible nature of the catalytic pocket. The two most
aAstraZeneca, Alderley Park, Maccleseld, Cheshire, SK10 4TG, UK. E-mail: william.
mccoull@astrazeneca.com; Tel: +44 (0)1625 519444
bAstraZeneca, Gatehouse Park, Waltham, Massachusetts 02451, USA
† Electronic supplementary information (ESI) available: Synthetic details for the
preparation of compounds, kinase selectivity data, and protein
production/purication and X-ray crystallography conditions. See DOI:
10.1039/c4md00280f
Fig. 1 Known PAK inhibitors.
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1533–1539 | 1533

View Article Online
MedChemComm
Concise Article
Table 1 PAK1 and PAK4 potency for compounds 1–4
Cpd
A
PAK1 IC50a (mM)
PAK4 IC50a (mM)
1
2
3
4
CH2
C]O
—
0.083
0.018
0.46
6.3
1.2
0.55
0.17
>10
SO2
a
Geometric mean of at least two experiments. IC50 values were
determined at ATP concentrations within 2-fold of the measured KM
ATP for their respective kinases.
PAK4 was utilised as an anti-target to obtain selectivity against
group II PAKs.14 Gini coefficient15 is commonly used to assess
the kinase selectivity of a compound tested in a broad kinase
panel, where a value between 0 and 1 is returned with more
selective compounds having values closer to 1. An assessment of
1 revealed a low Gini coefficient of 0.29. 7-Azaindoles have
previously been reported as kinase inhibitors, such as the
marketed cancer drug vemurafenib which is an inhibitor of B-
raf kinase (targeting V600E mutant).16 In addition it has been
Fig. 2 Crystal structure of compound 15 in complex with the kinase
domain of human PAK1 at 2.49 A˚ resolution. All residues located within
3.5 A˚ of the inhibitor are shown as green carbon sticks. Compound 15
is shown as yellow carbon sticks. The final 2Fo-Fc electron density
map surrounding the inhibitor is shown as blue mesh (contoured at 1s).
reported that “selectively nonselective” 7-azaindoles were positions. The former led to 111-fold decrease in PAK1 potency
obtained for mixed lineage kinase 3,17 indicating that kinase while the latter was 14-fold less potent against PAK1. Selectivity
selectivity would be a key challenge for this chemotype. against PAK4, KDR and FGFR1 decreased in both cases. Other
Consequently, potency and kinase selectivity became the main groups at the para position were tolerated (12, 13) and we
drivers for our lead optimisation efforts.
decided to explore the combination of ortho-Cl with other
Initial chemistry explored the linker group between the 7- substituents in compounds 14–24. Di-ortho compound 14
azaindole and aryl group (Table 1). The carbonyl linker in maintained potency and selectivity but afforded no clear
compound 2 afforded nearly 5-fold increase in PAK1 potency advantage over 2. Di-substituted 2-chloro, 5-hydroxymethyl (15)
with an increase to 30-fold selectivity over PAK4. Removal of the was a successful attempt to introduce a substituent which
linker altogether in 3 reduced potency 5-fold against PAK1 but would reduce lipophilicity, but isolipophilic 2-chloro, 5-cyano
increased PAK4 potency 7-fold. Sulfone linker 4 signicantly 16 was 10-fold less potent, suggesting the SAR is specic rather
reduced potency against both PAK1 and PAK4. The proposed being driven by lipophicity at this substituent. PAK1 potency
conformation necessary for binding (based on later ligand- could be increased by adding an appropriate substituent at the
PAK1 crystal structure, Fig. 2) is an unfavourable conformation 5-position in combination with the 2-chloro group. Lipophilic
for the sulfone but favourable conformation for the other groups such as halogen (17 and 18) maintained PAK1 potency
linkers. (see ESI†). Broader kinase screening of ketone 2 and kinase selectivity as did the less lipophilic cyano analogue
revealed a signicantly improved kinase selectivity prole with (19), which exhibited the best selectivity prole thus far.
Gini coefficient of 0.46, which encouraged us to maintain the However, more polar groups (20 and 21) eroded PAK1 potency
carbonyl in all subsequent compounds.
and alcohol 22 appeared to give the best potency/lipophilicity
The SAR of substitution around the phenyl ring was subse- balance. Ethers (23 and 24) were tolerated and the larger phenyl
quently explored (Table 2). The chlorine of 2 was replaced with ether 24 was not only more potent against PAK1 but also had
various ortho substituents in compounds 5–9 and all exhibited a signicantly improved selectivity against PAK4. Compounds 22
decrease in PAK1 potency, with only methyl (6), uoro (7) and and 24 were tested in a cellular PAK1 phosphorylation assay19
the original chloro (2) compounds having greater potency than and gave disappointing IC50 values of 6.2 and 1.6 mM respec-
simple phenyl analogue 5. As well as assessing PAK4 selectivity, tively. This larger than expected drop-off from enzyme to cell
we measured KDR and FGFR1 kinase inhibition.18 Our lead potency was attributed to low permeability as 22 and 24 had
compound 2 was not very selective against these additional measured Papp values of 0.14 and 0.11 ꢀ 10ꢁ6 cm sꢁ1 respec-
kinases and neither is FRAX-597, thus we believed this data tively in a Caco-2 permeability assay.
could be used to drive improvements in kinase selectivity. The
We rationalised that the strongly basic piperidine, present in
chloro group was moved to the meta (10) and para (11) all compounds described so far, was contributing to low
1534 | Med. Chem. Commun., 2014, 5, 1533–1539
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
Table 2 PAK1 potency and selectivity vs. PAK4, KDR and FGFR1 for compounds 2 and 5–24
Cpd
R2
R3
R4
R5
R6
PAK1 IC50a (mM)
PAK4 selectivitya,b
KDR selectivitya,b
FGFR1 selectivitya,b
log D7.4
2
5
6
7
8
9
Cl
H
Me
F
CN
MeO
H
H
H
H
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
H
H
H
H
H
H
Cl
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Cl
CH2OH
CH2OPh
H
H
H
F
Cl
CN
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Me
H
H
H
H
H
H
H
H
H
H
0.018
0.4
0.063
0.12
0.74
0.49
2
0.26
0.63
0.084
0.088
0.064
0.65
0.013
0.0075
0.023
0.63
30
3.4
—
6.7
3.5
14
4.5
0.25
2.7
0.68
1.4
0.9
0.055
0.16
0.17
0.42
3.5
3.2
1.5
9.8
4.5
11
0.44
1.5
4.9
2.0
3
1.6
1.4
1.8
1.5
0.57
1.1
2.3
2.2
0.53
3.2
2.3
0.96
1.1
2.1
2.5
1.2
0.5
0.63
0.97
1.7
3.7
0.24
0.56
0.57
0.7
0.13
0.032
0.12
0.79
1.9
1.4
1.2
0.49
4.2
1.4
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
3
6.6c
0.52
90
H
H
H
27
CH2OH
CN
H
H
H
H
H
H
H
4.2
8.3
50
38
36
2.6
>23
31
56
250
7.4
0.52
3.2
13
2.5
CONH2
SO2Me
CH2OH
CH2OMe
CH2OPh
0.44
0.0076
0.021
0.003
H
22
21
a
Geometric mean of at least two experiments unless otherwise stated. IC50 values were determined at ATP concentrations within 2-fold of the
measured KM ATP for their respective kinases. b Selectivity calculated from ratio of IC50 values. c n ¼ 1.
permeability, thus we sought to modulate this part of the Acetylation of the piperidine (26) did reduce potency and
molecule (Table 3). Methylation of the piperidine (25) main- selectivity somewhat but gave a 12-fold increase in permeability,
tained potency and selectivity but did not improve permeability. conrming that variation of this substituent was important to
Table 3 PAK1 potency, selectivity vs. PAK4, KDR and FGFR1 and permeability for compounds 22 and 25–27
PAK1 IC50a (mM) PAK4 selectivitya,b KDR selectivitya,b FGFR1 selectivitya,b Permeability Papp (ꢀ 10ꢁ6 cm sꢁ1
)
log D7.4
c
Cpd R1
22
25
26
0.0076
0.012
0.055
31
24
4
4.9
13
0.14
0.16
1.7
0.97
2.2
4.4
8.3
1.8
0.27
2.5
27
0.037
7.8
0.38
1.1
7.8
3.4
a
Geometric mean of at least two experiments. IC50 values were determined at ATP concentrations within 2-fold of the measured KM ATP for their
respective kinases. b Selectivity calculated from ratio of IC50 values. c Compounds were incubated at 10 mM in cultured human Caco-2 cells at pH 6.5.20
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1533–1539 | 1535

View Article Online
MedChemComm
Concise Article
improving permeability. A further increase in permeability was
A diverse array of compounds with various groups intended
obtained with truncated pyrazole analogue 27, without any to explore the solvent channel were prepared and bicycle 34
further erosion of potency and selectivity relative to 26. This gave encouraging potency, including cellular potency, and
increased permeability correlates with increased lipophilicity selectivity similar to our original piperidine 2 (Table 5).
and removal of the ionisable basic nitrogen.
Expansion to the 6,6-bicyclic systems improved potency in 35
Next we combined the truncated methoxy-pyrazole with and particularly in 36 which afforded the highest selectivity and
potent aryl substituents from our previous SAR exploration cell potency of any compound from this series. The basic
(Table 4). Extended ether 28 was not as potent as earlier nitrogen appeared to be necessary since acetyl analogue 37 had
analogue 24 but still exhibited high selectivity vs. PAK4. 2-Flu- signicantly reduced potency. In 36, we achieved improved
oro (29) and 2-aza (30) benzyl ethers gave similar encouraging selectivity against PAK4, KDR and FGFR1 compared to FRAX-
PAK1 potency and PAK4 selectivity, but 3-aza analogue (31) was 597 (Table 5). In terms of broader kinase selectivity, 36 can be
less potent and selective. Disappointingly, little selectivity was considered equally kinase selective with FRAX-597 given their
obtained for KDR or FGFR1 which correlated with poor overall Gini coefficients of 0.60 and 0.59 (ref. 13) respectively but with
kinase selectivity (Gini coefficient of 0.35 for 30). Consequently differing overall prole. Compound 36 exhibited >95% inhibi-
we abandoned further work with these larger groups at the R4 tion of 11/125 kinases tested at 0.1 mM: Abl1, Fyn, Lck, Lyn,
position.
An earlier compound, 15, was co-crystallised with the kinase
PTK6, Ret, PAK1, Src and Yes1.
To understand the utility of 36 as an in vitro tool for target
domain of human PAK1. The resulting structure conrmed that validation studies, appropriate cellular activity is important.
the 7-azaindole formed a pair of hinge binding interactions with The cell drop off due to increased ATP competition was calcu-
the back bone carbonyl of E345 and amide NH of L347, the lated using the Cheng–Prusoff equation.21 Based on the KM of
pyrazole was positioned toward the solvent, and the chloro-aryl PAK1 for ATP (17 mM) and assuming that the compounds are
ring substituent was directed to the selectivity pocket and in fully competitive with ATP and a cellular ATP concentration of 2
front of the gatekeeper residue M344 (Fig. 2). On the opposing mM, we anticipated a 75-fold drop-off from kinase potency to
side of the aryl ring, the alcohol forms interactions with both cell potency. In agreement with this, an excellent correlation (r2
residues K299 and E315. The length of these electrostatic ¼ 0.992) was obtained for those compounds tested in both ATP
˚
interactions are not ideal given their bond distances of 3.3 A to concentration assays, which included 7-azaindoles and FRAX-
˚
the side chain of residue K299 and 3.5 A to the side chain of 597 (Fig. 3). In the absence of other factors, a similar correlation
residue E315 (Fig. 2). We envisaged that reducing the distance was expected between cell and enzyme potency. However, a
between the alcohol and these catalytic residues would generate larger spread of data was observed (Fig. 4). Compounds which
more favourable electrostatic interactions. Consequently, exhibited higher drop-offs, such as 22 and 24 were characterised
homologated alcohols 32 and 33 were prepared, and the latter by low permeability, while compounds that exhibited lower
demonstrated an 11-fold potency increase relative to 15. drop-offs, such as 28–31 were characterized by poor kinase
Unfortunately, this potency increase was not sufficient to obtain selectivity. Both compound 36 and FRAX-597 are close to the
selectivity against KDR and FGFR1. Thus, we decided to return expected 75-fold drop-off between enzyme and cell potency.
to the ortho-Cl-phenyl selectivity pocket group and explore the Compound 36 has a lower level of plasma protein binding (3.5%
solvent channel as an alternative path to improvements in vs. 0.08% free) which is benecial for use in cellular target
potency and selectivity.
validation experiments run in the presence of serum.
Table 4 PAK1 potency and selectivity vs. PAK4, KDR and FGFR1 for compounds 27–33
Cpd
R4
R5
PAK1 IC50a (mM)
PAK4 selectivitya,b
KDR selectivitya,b
FGFR1 selectivitya,b
log D7.4
27
28
29
30
31
32
33
CH2OH
CH2OPh
H
H
H
H
H
H
0.037
0.043
0.048
0.014
0.13
7.8
0.38
2.1
1.0
2.5
0.71
0.2
1.1
3.9
2.3
10
0.44
0.14
2.4
3.4
>240
>630
100
29
6.4
42
>4.9
>4.1
>4.3
>4.3
3.1
CH2O-2-F-Ph
CH2O-2-pyridyl
CH2O-3-pyridyl
CH2CH2OH
H
0.14
0.0058
CH2CH2OH
1.5
3.2
a
Geometric mean of at least two experiments unless otherwise stated. IC50 values were determined at ATP concentrations within 2-fold of the
measured KM ATP for their respective kinases. b selectivity calculated from ratio of IC50 values.
1536 | Med. Chem. Commun., 2014, 5, 1533–1539
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
Table 5 PAK1 potency and selectivity vs. PAK4, KDR and FGFR1 for compounds 2, 34–37 and FRAX-597
h plasma protein
Cpd R1
PAK1 IC50a (mM) PAK4 selectivitya,b KDR selectivitya,b FGFR1 selectivitya,b pPAK1 IC50a (mM) log D7.4 binding (%free)
2
0.018
0.033
30
63
4.5
4.3
3
0.87
1.1
1.6
11
34
1.3
—
—
35
36
0.0045
0.0010
180
450
18
50
4.3
24
0.65
0.14
4
3.4
3.5
3.3
37
0.27
6.6
0.12
1.2
0.33
1.7
1.3
—
—
FRAX-597
0.0028
260
0.088
>4
0.08
a
Geometric mean of at least two experiments unless otherwise stated. IC50 values were determined at ATP concentrations within 2-fold of the
measured KM ATP for their respective kinases. b selectivity calculated from ratio of IC50 values.
Fig. 3 PAK1 enzyme potency (2 mM ATP against 10 mM ATP) corre-
lation with best fit (bold) and calculated IC50 shift for ATP competitive
compounds (dashed) lines shown.
Fig. 4 PAK1 cell: enzyme correlation for 7-azaindoles and FRAX-597
with 1 : 10/100/1000 lines shown.
hydride and reaction with chloromethyl pivalate provided piv-
aloyloxymethyl (POM) protected 7-azaindole 39 in good yield. A
Suzuki–Miyaura cross coupling with 40 followed by removal of
the POM group with NaOH gave, aer reverse phase purica-
tion, 36 in moderate yield. Intermediate 39 allowed facile
exploration of the 5 position of the 7-azaindole. Variation of the
boronate coupling partner in the Suzuki–Miyuara cross
coupling reaction with 39 afforded the 7-azaindole compounds
in Table 5.
Synthesis
A number of approaches were undertaken to synthesise the
compounds studied, with routes to representative examples
summarised in Schemes 1–3 (full details for all compounds are
available in the ESI†). In the synthesis of compound 36, the aryl
ketone was installed early by a Friedel–Cras acylation between
5-bromo-7-azaindole 38 and 2-chlorobenzoyl chloride (Scheme
1). Subsequent deprotonation of the 7-azaindole with sodium
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1533–1539 | 1537

View Article Online
MedChemComm
Concise Article
Conclusions
In summary, a novel 7-azaindole series of PAK1 inhibitors was
identied. SAR exploitation enabled improvements in enzyme
and cellular potency and kinase selectivity to afford 36 (AZ-PAK-
36) as a useful tool compound for target validation experiments.
Scheme 1 Reagents and conditions: (a) AlCl3, 2-Cl-PhCOCl, DCM,
52%; (b) chloromethyl pivalate, NaH, DMF, 87%; (c) (i) Pd(PPh3)4,
K2CO3, 3 : 1 DME : H2O, 75 ꢂC, (ii) NaOH, MeOH, THF, 25% (2 steps).
Acknowledgements
Martina Fitzek, Renee Garcia-Arenas and Brenda McDermott
are acknowledged for generating kinase inhibition data. Robert
Godin, Mark Roth and Paul Clarkson are acknowledged for
expert technical assistance in interpretation of biological data.
Caroline Rivard and Christopher Jones are acknowledged for
expert DMPK input.
Alternatively, the aryl ketone was installed late in the
synthesis allowing variation of this group as in the synthesis of
compound 5 (Scheme 2). The initial step was a Suzuki–Miyaura
cross coupling between 38 and boronate 41 (ref. 22) which
afforded 42 in excellent yield. Reaction of 42 with benzaldehyde
and KOH in MeOH gave alcohol 43 in moderate yield. Subse-
quent oxidation to the ketone using MnO2 followed by treat-
ment with acid to remove the butoxycarbonyl (Boc) group
afforded 5.
Notes and references
1 M. Radu, G. Semenova, R. Kosoff and J. Chernoff, Nat. Rev.
Cancer, 2014, 14, 13–25.
Further variations to the aryl ring were explored using a third
route (Scheme 3). Key aldehyde intermediate 44 was synthesised
in good yield from 38 in two steps. Aer protection with POM
the aldehyde was installed under Vilsmeier–Haack conditions.
It was possible to successfully react aldehyde 44 with various
aryl lithium or Grignard reagents and ketone 45 was obtained in
excellent yield from the latter aer subsequent MnO2 oxidation.
The pyrazole fragment was installed as previously in good yield.
Compound 9 was then isolated aer removal of POM and Boc
protecting groups.
2 H. King, N. S. Nicholas and C. M. Wells, Int. Rev. Cell Mol.
Biol., 2014, 309, 347–387.
3 R. Kumar, A. E. Gururaj and C. J. Barnes, Nat. Rev. Cancer,
2006, 6, 459–471.
4 C. C. Ong, A. M. Jubb, P. M. Haverty, W. Zhou, V. Tran,
T. Truong, H. Turley, T. O'Brien, D. Vucic, A. L. Harris,
M. Belvin, L. S. Friedman, E. M. Blackwood, H. Koeppen
and K. P. Hoeich, Proc. Natl. Acad. Sci. U. S. A., 2011, 108,
7177–7182.
5 C. W. Menges, E. Sementino, J. Talarchek, J. Xu, J. Chernoff,
J. R. Peterson and J. R. Testa, Mol. Cancer Res., 2012, 10,
1178–1188.
6 C. Yi, J. Maksimoska, R. Marmorstein and J. L. Kissil,
Biochem. Pharmacol., 2010, 80, 683–689.
7 J. J. Crawford, K. P. Hoeich and J. Rudolph, Expert Opin.
Ther. Pat., 2012, 22, 293–310.
8 Y. Xu, J. M. Foulks, A. Clifford, B. Brenning, S. Lai, B. Luo,
K. M. Parnell, S. Merx, M. V. McCullar, S. B. Kanner and
K. Ho, Bioorg. Med. Chem. Lett., 2013, 23, 4072–4075.
9 S. T. Staben, J. A. Feng, K. Lyle, M. Belvin, J. Boggs,
J. D. Burch, C. Chua, H. Cui, A. G. DiPasquale,
L. S. Friedman, C. Heise, H. Koeppen, A. Kotey, R. Mintzer,
A. Oh, D. A. Roberts, L. Rouge, J. Rudolph, C. Tam,
W. Wang, Y. Xiao, A. Young, Y. Zhang and K. P. Hoeich,
J. Med. Chem., 2014, 57, 1033–1045.
Scheme 2 Reagents and conditions: (a) 38, Pd(dppf)Cl2, K2CO3, 10 : 1
DME : H2O, 100 ꢂC, 92%; (b) benzaldehyde, KOH, MeOH, 45%; (c)
MnO2, DCM 87%; (d) HCl, dioxane, 19%.
10 C. Guo, I. McAlpine, J. Zhang, D. D. Knighton, S. Kephart,
M. C. Johnson, H. Li, D. Bouzida, A. Yang, L. Dong,
J. Marakovits, J. Tikhe, P. Richardson, L. C. Guo, R. Kania,
M. P. Edwards, E. Kraynov, J. Christensen, J. Piraino,
J. Lee, E. Dagostino, C. Del-Carmen, Y. Deng, T. Smeal and
B. W. Murray, J. Med. Chem., 2012, 55, 4728–4739.
11 B. W. Murray, C. Guo, J. Piraino, J. K. Westwick, C. Zhang,
J. Lamerdin, E. Dagostino, D. Knighton, C. Loi, M. Zager,
E. Kraynov, I. Popoff, J. G. Christensen, R. Martinez,
S. E. Kephart, J. Marakovits, S. Karlicek, S. Bergqvist and
T. Smeal, Proc. Natl. Acad. Sci. U. S. A., Early Ed., 2010, 1–6.
Scheme 3 Reagents and conditions: (a) chloromethyl pivalate, NaH,
DMF, 0 ꢂC, 58%; (b) POCl3, DMF, 0 to 50 ꢂC, 70%; (c) 2-MeO-PhMgBr,
THF, 97%; (d) MnO2, DCM, 88%; (e) 41, Pd(dppf)Cl2, K2CO3, 10 : 1
DME : H2O, 100 ꢂC, 86%; (f) NaOH, MeOH, THF, 99%; (g) HCl, 1,4-
dioxane, 35%.
1538 | Med. Chem. Commun., 2014, 5, 1533–1539
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
12 S. Licciulli, J. Maksimoska, C. Zhou, S. Troutman, S. Kota, 15 P. P. Graczyk, J. Med. Chem., 2007, 50, 5773–5779.
Q. Liu, S. Duron, D. Campbell, J. Chernoff, J. Field, 16 G. Bollag, J. Tsai, J. Zhang, C. Zhang, P. Ibrahim, K. Nolop
R. Marmorstein and J. L. Kissil, J. Biol. Chem., 2013, 288,
29105–29114.
13 H. Y. Chow, A. M. Jubb, J. N. Koch, Z. M. Jaffer, D. Stepanova,
D. A. Campbell, S. G. Duron, M. O'Farrell, K. Q. Cai,
A. J. P. Klein-Szanto, J. S. Gutkind, K. P. Hoeich and
J. Chernoff, Cancer Res., 2012, 72, 5966–5975.
14 PAK1 and PAK4 kinase assays: the inhibitory activity of
compound was determined by caliper off-chip incubation
mobility shi assays, using a microuidic chip to measure
and P. Hirth, Nat. Rev. Drug Discovery, 2012, 11, 873–886.
17 V. S. Goodfellow, C. J. Loweth, S. B. Ravula, T. Wiemann,
T. Nguyen, Y. Xu, D. E. Todd, D. Sheppard, S. Pollack,
O. Polesskaya, D. F. Marker, S. Dewhurst and
H. A. Gelbard, J. Med. Chem., 2013, 56, 8032–8048.
18 R. A. Norman, A. Schott, D. M. Andrews, J. Breed,
K. M. Foote, A. P. Garner, D. Ogg, J. P. Orme, J. H. Pink,
K. Roberts, D. A. Rudge, A. P. Thomas and A. G. Leach, J.
Med. Chem., 2012, 55, 5003–5012.
the conversion of uorescent labeled peptide to
a
19 PAK1 phosphorylation cellular assay: cryopreserved MCF10A
cells which have been transfected with PAK1 under control
of a doxocycline inducible promoter were used. PAK
expression was induced prior to cryopreservation. Revived
cells were seeded with 3000 cells per well of a black 384
phosphorylated product. Fluorescently tagged peptide
substrates were synthesized by Cambridge Research
Biochemicals (Cleveland, U.K.). 12 point half-log
compound concentration–response curves, with
a top
ꢂ
concentration of 100 mM were generated from 10 mM
stocks of compound solubilised in DMSO using an Echo
555 (Labcyte Inc., Sunnyvale, CA). All assays were
preformed in white Greiner 384-well low volume plates
(Greiner Bio-One, UK), in a total reaction volume of 12 mL
and 1% (v/v) nal DMSO concentration. Enzymes and
substrates were added separately to the compound plates
and incubated at room temperature. The kinase reaction
was then quenched by the addition of 10 mL stop buffer
(100 mM HEPES pH 7.5, 5% v/v DMSO, 44 mM EDTA,
well plate and treated with compound for 2 h at 37 C, 5%
CO2. Aer xing with 4% paraformaldehyde and blocking
of non-specic binding with a solution (1% BSA + 5% goat
serum + 0.3% TritonX100 in PBS) the plates were treated
with a phosphoPAK (Ser144) specic antibody and a
secondary antibody labelled with Alexa Fluor-488 and
Hoechst 33342 Between all reagent steps, the cells were
washed with PBS. The stained cells were imaged on a
Cellomix ArrayScan VTI using Mean Ring spot Average
Intensity analysis method.
0.22% v/v CR-3 0.033% Brij35). Stopped assay plates were 20 B. Over, P. McCarren, P. Artursson, M. Foley, F. Giordanetto,
read using a Caliper LabChip EZ reader (PerkinElmer
Waltham, MA). IC50 values were calculated using Genedata
Screener (Genedata AG, Basel, Switzerland). PAK1 kinase
domain was purchased from Abcam (Cambridge, U.K.).
G. Groenberg, C. Hilgendorf, M. D. Lee, P. Matsson,
G. Muncipinto, M. Pellisson, M. W. D. Perry, R. Svensson,
J. R. Duvall and J. Kihlberg, J. Med. Chem., 2014, 57, 2746–
2754.
The kinase reaction (10 mM/2 mM ATP, 25 pM PAK1, 1 mM 21 A. C. Cheng, J. Eksterowicz, S. Geuns-Meyer and Y. Sun, J.
peptide substrate 5-FAM-KPDRKKRYTVVGNPY-amide) in Med. Chem., 2010, 53, 4502–4510.
phosphorylation buffer (50 mM HEPES pH 7.3, 10 mM 22 A. B. Northrup, M. H. Katcher, M. D. Altman, M. Chenard,
MgCl2, 1 mM DTT 0.01% Tween20, 0.05 mg mLꢁ1 BSA)
was quenched aer a 120 min incubation. PAK4 kinase
domain was purchased from Life Technologies (Thermo
Fisher Scientic, Waltham, MA). The kinase reaction (1.5
mM ATP, 2 nM PAK4, 1 mM peptide substrate 5-FAM-Ahx-
KKRNRRLSVA-amide) in phosphorylation buffer (50 mM
HEPES pH 7.3, 10 mM MgCl2, 1 mM DTT 0.01% Tween20,
0.05 mg mLꢁ1 BSA) was quenched aer a 90 min incubation.
M. H. Daniels, S. V. Deshmukh, D. Falcone, D. J. Guerin,
H. Hatch, C. Li, W. Lu, B. Lutterbach, T. J. Allison,
S. B. Patel, J. F. Reilly, M. Reutershan, K. W. Rickert,
C. Rosenstein, S. M. Soisson, A. A. Szewczak, D. Walker,
K. Wilson, J. R. Young, B. Pan and C. J. Dinsmore, J. Med.
Chem., 2013, 56, 2294–2310.
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1533–1539 | 1539