Design and synthesis of novel amide AKT1 inhibitors with selectivity over CDK2

Article abstract of DOI:10.1016/j.bmcl.2011.07.056

Through the analysis of X-ray crystallographic information and previous SAR studies, a novel series of protein kinase B (PKB/AKT) inhibitors was developed. The compounds showed nanomolar inhibition of AKT1 and were selective against cyclin-dependent kinas

Full text of DOI:10.1016/j.bmcl.2011.07.056

Bioorganic & Medicinal Chemistry Letters 21 (2011) 5191–5196
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Design and synthesis of novel amide AKT1 inhibitors with selectivity over CDK2
Kate S. Ashton a, , David J. St. Jean Jr. a, Steve F. Poon a, Matthew R. Lee a, John G. Allen a, Shiwen Zhang b,
⇑
Julie A. Lofgren b, Xiaoling Zhang b, Christopher Fotsch a, Randall Hungate a
a Chemistry Research and Discovery, Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320, USA
b Oncology Research, Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Through the analysis of X-ray crystallographic information and previous SAR studies, a novel series of
protein kinase B (PKB/AKT) inhibitors was developed. The compounds showed nanomolar inhibition of
AKT1 and were selective against cyclin-dependent kinase 2 (CDK2).
Received 4 June 2011
Revised 11 July 2011
Accepted 13 July 2011
Available online 23 July 2011
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
PKB
AKT
CDK
Cancer
AKT belongs to the serine/threonine family of kinases and has
been the subject of intense anti-cancer research due to its critical
role as a regulator of cell growth, differentiation and survival, pro-
tein synthesis and metabolism.1 AKT is part of a metabolic path-
way that is frequently disregulated in a wide variety of cancers
through an upstream loss of function mutation in PTEN, or a gain
of function mutation in PI3K.2 AKT has three major isoforms
with previous SAR knowledge. The crystal structure of thiadiazole
1 in Figure 2 (residue numbering is based on PKA) is bound to a tri-
ple-mutant construct of PKA’s kinase domain, that mimics the ATP
binding pocket of AKT, and the key interactions are highlighted.11
Analysis of this structure suggested that replacement of the thiadi-
azole with an amide should conserve the Ala123/Glu121 contact,
the amine side chain interactions with Asp184 and Asn171, and
the Met173 floor residue contact.
(AKT1/PKBa, AKT2/PKBb and AKT3/PKBc) that share >85% homol-
ogy in the kinase domain. AKT1 is the most widely expressed
and activated isoform in tumor cells, and knock-out mice display
normal life span phenotype.3 We report herein our efforts thus
far to identify kinase-specific modulators of the AKT-pathway.
Initial internal efforts focused on a series of compounds that con-
tained the thiadiazole/thiazole cores shown in Figure 1.4–7 These
series exhibited excellent potency against AKT18 and PKA enzymes;
however had only a 10–17 fold selectivity over CDK2. CDK2 is a
serine/threonine kinase that regulates cell cycle transition and cell
proliferation,9 and as such activity on this kinase was deemed
undesirable. In addition PKA is highly homologous to AKT1 in the
kinase domain and it was anticipated that selectivity against PKA
would translate into a highly selective AKT1 inhibitor. Hence our
goal was to improve kinase selectivity while maintaining a
favorable off-target profile in a new series of AKT1 inhibitors.
Thus began a search for an analogue that would retain potency
and address kinase selectivity. With a plethora of structural infor-
mation at hand,10 we decided to move away from the scaffold
employed by the previous series and to incorporate rational design
The amide isostere molecule 3 (Fig. 3) retained weak activity for
AKT. The drop in potency could be explained by the failure of the
amide to fully capitalize on the Lys72 hydrogen bond interaction
made by the thiadiazoles/thiazoles. This result however encour-
aged further paring down of the structure.
It was clear from the crystal structure of 1 that several key resi-
dues were being engaged in three distinct areas of the enzyme
N
N
N
NH
NH
S
S
NH2
NH2
HN
O
N
F3C
F3C
2
1
Enzyme IC50 (µM):
AKT1: 0.02 0.006
PKA: 0.17 0.05
CDK2: 0.19 0.01
Enzyme IC50 (µM):
AKT1: 0.003 0.0009
PKA: 0.006 0.02
CDK2: 0.05 0.02
U87MG Cellular (µM):
p-PRAS40: 0.25 0.1
U87MG Cellular (µM):
p-PRAS40: 0.30 0.08
⇑
Corresponding author.
E-mail address: katea@amgen.com (K.S. Ashton).
Figure 1. Thiadiazole/thiazole series of AKT1 inhibitors. Data reported as the
mean SD where n P 3.
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.07.056

5192
K. S. Ashton et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5191–5196
With this encouraging result in hand, we investigated the Phe-
54 region of the protein to try and improve potency and PKA selec-
tivity with a number of substituted aryl groups (Table 1). Space did
not appear to be limited in this region (5 and 9), and a range of sub-
stitution was well tolerated on the phenyl ring (10, 11, 13, and 15).
The CDK2 selectivity remained high throughout these changes,
however no selectivity over PKA could be achieved through varia-
tion in this region of the molecule.
In the next series of molecules we kept the 3,4-dichloro phenyl
group constant and turned our attention to modifying the region of
the molecule that binds Ala 123/Glu 121. Although this had been
previously optimized for the ‘U-shaped’ thiazoles and thiadiazoles,
it was considered that there may have been a slight shift of the
molecule in the binding pocket and that this area should be thor-
oughly re-examined. It had been shown previously that certain
groups in this region could impart some PKA selectivity,7 and with
this in mind a range of heterocyclic bicycles were investigated as
shown in Table 2.
It became evident that the 6,6-bicycles imparted the most po-
tency and that of these the isoquinoline was the most active. Selec-
tivity was achieved over PKA with benzoxazolone 19, however at
the expense of AKT1 activity. Further attempts to increase selectiv-
ity for AKT1 within the active isoquinoline moiety (22–25) were
unsuccessful. A strategically placed fluorine (24) and methoxy
(22) were well tolerated but failed to offer any significant selectiv-
ity for AKT1 over PKA. All of these compounds retained significant
selectivity over CDK2.
Figure 2. Thiadiazole 1 bound to AKT (PKA triple mutant).
O
NH
N
NH2
Finally, the amine portion of the molecule was examined. Fig-
ure 5 shows the binding mode of the amide containing the 3-piper-
idine ring as predicted by modeling. From this representation it
was expected that variation of the amine moiety would be well tol-
erated due to the many polar residues in this region of the enzyme.
The piperidine nitrogen of 27 could engage Glu127/Glu170 as sug-
gested, however the Asp184/Asn171 polar region is also accessible
from this scaffold.
F3C
3
Enzyme IC50
( M):
µ
AKT1: 3.38 2.45
PKA: 1.56
CDK2: >25
Not only did it seem plausible that favorable contacts could
arise from various orientations of the amine it also seemed possi-
ble to reach the floor of the enzyme. Given that AKT1 contains a
methionine (Met173) in this region, whereas PKA contains a leu-
cine in the same position (Leu173), it was thought that by extend-
ing a larger group into this region we could exploit this subtle
difference. Table 3 summarizes the efforts in this area.
As predicted, variation of the ring size was well tolerated with the
4, 5 and 6-membered rings (26–29) being equi-potent. Compound
30 demonstrates that the polar amine group is indeed essential for
potent activity and compounds 31 and 32 make the case for retain-
ing the fully saturated ring and basic amine functionality. Substitu-
tion of the piperidine nitrogen (33, 34, 35 and 36) was reasonably
well tolerated, indeed the methyl substituted azetidine 36 was
equi-potent, but failed to enhance selectivity for AKT1 over PKA.
TheCDK2 selectivity remainedabove 50-fold for all of these changes,
however there was an increase in CDK2 binding activity moving
from the 3- to the 4-piperidine (26 and 27), and with the smaller ring
size (R,S-29 and 36).1 The synthesis of R,S-29 is shown in Scheme 1.
The potent affinity for AKT1 and selectivity against CDK2 of
these compounds, as well as their ease of synthesis, makes them
attractive targets. With regard to the in vivo stability of these com-
pounds, the high in vivo clearance would require improvement to
be of therapeutic use (Table 4).2
Figure 3. Replacement of heterocyclic core. Data reported as the mean SD where
n P 3.
O
N
H
N
NH
4
Enzyme IC50
( M):
µ
AKT1: 0.85
PKA: 0.25
CDK2: 19.7
Figure 4. Rationally designed AKT1 inhibitor. Data reported as the mean SD
where n P 3.
pocketand thatthere wasa fourth, as yetunutilizedspacein thePhe-
54 region that could be explored. Therefore it was decided to forego
the Met173 floor residue interaction, that in previous series had not
imparted significant PKA or CDK2 selectivity, and move away from
the ‘U-shaped’ motif, to incorporate a phenyl ring in the Phe-54 area.
The racemic compound 4 shown (Fig. 4) was initially synthe-
sized based on modeling data that indicated this structure would
successfully access the Ala 123/Glu 121 region as well as the
Phe-54 region of the protein (see Fig. 5 for a model of a similar
structure). The secondary amine was made as literature precedent
suggested that this was well tolerated.12 All of these changes trans-
lated into a compound that was surprisingly active on AKT1 and
PKA; however, the attractive qualities of this molecule were its
structural simplicity and CDK2 selectivity.
1
The higher CDK2 selectivity achieved with the 3-piperidine ring might be
attributed to the amine binding the Glu 127/170 polar region (as is proposed in the
model of 27 in Fig. 5) as opposed to the adjacent Asp184 and Asn171 residues.
2
The discrepancy between the in vitro stability and in vivo clearance was not
investigated, but the compounds showed high efflux ratios and metabolism by non
P450 enzymes was not ruled out by any in vitro assays.

K. S. Ashton et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5191–5196
5193
Figure 5. Model of amide 27 in AKT pocket.
Table 1
Phe-54 region SAR
R
O
N
H
N
NH
Cmpd.
Aryl Group
AKT1 IC50
(
lM)
PKAa IC50
0.26
(l
M)
CDK2a IC50
>25
(lM)
5
0.40 0.02
N
S
6
7
1.57 0.06
1.51 0.26
1.07
0.13
>25
>25
O
F
F
8
9
2.68 0.20
0.20 0.01
0.16
0.22
>25
>25
F
Cl
Cl
10
0.08 0.01
0.16
>25
N
11
12
0.07 .003
0.19 0.02
0.03
0.13
>25
>25
Cl
Cl
Cl
F
13
14
0.04 0.004
0.12 0.003
0.04
0.10
>25
>25
Cl
Cl
Cl
N
15
16
0.04 0.02
0.14 0.01
0.05
0.23
>25
>25
Cl
Cl
Cl
Data reported on the racemates as the mean SD where n P 3 except (a) tested once.

5194
K. S. Ashton et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5191–5196
Table 2
Ala 123/Glu 121 Region SAR
Cl
Cl
O
R
N
H
NH
Cmpd.
Linker Binder
AKT1 IC50
(
l
M)
PKAa IC50
0.05
(lM)
17
0.04 0.02
0.35 0.02
N
O
18
19
20
1.15
13.7
0.29
N
H
O
O
0.46 0.04
N
H
N
0.25 0.10
0.33 0.01
N
H
21b
22
0.85
0.03
N
N
0.039 0.01
6.85
N
OMe
23a
0.67
N
F
Ph(p-OMe)
24
25
0.02 0.003
0.14 0.007
0.09
0.14
N
Cl
N
Data reported on the racemates as the mean SD where nP3 except (a) tested once (b) single enantiomer.
Table 3
Amine SAR
Cl
Cl
O
N
H
N
Cmpd.
Amine
AKT1 IC50
(lM)
PKAa IC50
0.05
(lM)
CDK2a IC50
15.6 2.7
(lM)
26b
0.04 0.02
NH
27c
0.02 0.0009
0.03
0.02
0.01
21
N
H
28a,d
>25
N
H
(R)-2913
(S)-2913
30a,b
0.007 0.007
0.11 0.05
5.80
0.01
0.03
>25
0.69
9.8
NH
NH
>25
N
31b
32b
12.65 1.45
20.44 1.06
3.41
>25
>25
>25
N
N

K. S. Ashton et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5191–5196
5195
Table 3 (continued)
Cmpd.
Amine
AKT1 IC50
2.14
(
l
M)
PKAa IC50
2.01
(lM)
CDK2a IC50
>25
(lM)
33a
34a
35
N
N
0.24
0.39
0.98
0.01
>25
>25
0.36
CF3
2.30 0.23
N
36
0.007 0.003
NMe
Data reported on the single enantiomer as the mean SD where nP3 except (a) tested once (b) racemic (c) cis diastereomers only (d) mixture of
four diastereomers.
Cl
Cl
O
NH2
O
I
Cl
Cl
Cl
Cl
(c),(d)
(b)
(a)
N
H
NBoc
NBoc
N
NH
NBoc
37
R
,S-29
38
39
Scheme 1. Synthesis of 29. Reagents and conditions: (a) activated Zn, 1,2-dibromoethane, THF, 65 °C; CuCN, LiCl, 3,4-dichlorobenzoyl chloride, À10 °C to rt, quant,14; (b)
NH4OAc, NaCNBH4, MeOH, 75 °C, 70%, (c) HATU, Hung’s base, isoquiniline-6-carboxylic acid, DMF, rt, (d) HCl, MeOH, rt, 20% over 2 steps.
Table 4
Selected in vivo and in vitro data
Cmpd.
Structure
AKT1
(IC50
PKAa
(IC50
CDK2a
p-Pras40a
RLMa
(% turnover)
HLMa
(% turnover)
In Vivo Rat CLa
(L/h/kg)
l
M)
l
M)
(IC50
l
M)
(IC50
lM)
Cl
Cl
(R)-29
0.007 0.007
0.02 0.003
0.01
0.03
0.68
0.24
33.4
15.9
10.9
10
10.4
3.5
O
N
H
N
NH
Cl
Cl
24
>25
0.39
O
F
N
H
N
NH
Data reported as the mean SD where n P 3 except (a) tested once.
2. (a) Hennessy, B. T.; Smith, D. L.; Ram, P. T.; Lu, Y.; Mills, G. B. Nat. Rev. Drug
Discovery 2005, 4, 988; (b) Bader, A. G.; Khang, S.; Zhao, L.; Vogt, P. K. Nat. Rev.
Cancer 2005, 5, 921;(c) Testa, J. R.; Bellacosa, A. Proc. Natl. Acad. Sci. U.S.A. 2001, 98,
10983; (d) Cully, M.; You, H.; Levine, A. J.; Mak, T. W. Nat. Rev. Cancer 2006, 6, 184.
3. Dummler, B.; Hemmings, B. A. Biochem. Soc. Trans. 2007, 35, 231.
4. AKT1, PKA, and CDK2 enzymatic activity was measured using a previously
described method. Zhang, X. L.; Zhang, S. W.; Yamane, H.; Wahl, R.; Ali, A.;
Lofgren, J. A.; Kendall, R. L. J. Biol. Chem. 2006, 281, 13949.
In conclusion a novel series of structurally simple amides has
been developed through consideration of previously learned SAR
and structure based design. The compounds display excellent po-
tency on AKT1 and PKA and selectivity against CDK2.
Acknowledgments
5. U87MG cells in 5% FBS media were incubated with inhibitors in threefold serial
dilutions for 1h at 37 °C. The cells were lysed and PRAS40 phosphorylation was
quantified by ELISA assay. Phospho-PRAS40 was normalized to total PRAS40.
Kovacina, K. S.; Park, G. Y.; Bae, S. S.; Guzzetta, A. W.; Schaefer, E.; Birnbaum, M.
J.; Roth, R. A. J. Biol. Chem. 2003, 278, 10189.
The authors are indebted to the PKDM team for in vivo work,
and to Tisha San Miguel and John McCarter for additional in vitro
screening.
6. Zeng, Q.; Bourbeau, M. P.; Wohlhieter, G. E.; Yao, G.; Monenschein, H.; Rider, J.
T.; Lee, M. R.; Zhang, S.; Lofgren, J. A.; Freeman, D. J.; Li, C.; Tominey, E.; Huang,
X.; Hoffman, D.; Yamane, H. K.; Tasker, A. S.; Dominguez, C.; Viswanadha, V. N.;
Hungate, R.; Zhang, X. Bioorg. Med. Chem. Lett. 2010, 20, 1652–1656.
7. Zeng, Q.; Allen, J. G.; Bourbeau, M. P.; Wang, X.; Yao, G.; Tadesse, S.; Rider, J. T.;
Yuan, C. C.; Hong, F.-T.; Lee, M. R.; Zhang, S.; Lofgren, J. A.; Freeman, D. J.; Yang,
S.; Li, C.; Tominey, E.; Huang, X.; Hoffman, D.; Yamane, H. K.; Fotsch, C.;
Dominguez, C.; Hungate, R.; Zhang, X. Bioorg. Med. Chem. Lett. 2010, 20, 1559–
1564.
A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2011.07.056.
References and notes
8. AKT1 was the only AKT isoform routinely tested against as it is thought to be
the major isoform contributing to oncogenic activity. Dummler, B.; Hemmings,
B. A. Biochem. Soc. Trans. 2007, 35, 231. In addition all other isoforms of AKT
potency tracked well with AKT1 upon testing.
1. (a) Lindsley, C. W.; Barnett, S. F.; Layton, M. E.; Bilodeau, M. T. Curr. Cancer Drug
Target 2008, 8, 7; (b) Li, Q. Expert Opin. Ther. Pat. 2007, 17, 1077; (c) Heerding, D.
A.; Safonov, I. G.; Verma, S. K. Annu. Rep. Med. Chem. 2007, 42, 365.

5196
K. S. Ashton et al. / Bioorg. Med. Chem. Lett. 21 (2011) 5191–5196
9. Besson, A.; Dowdy, S. F.; Roberts, J. M. Dev. Cell 2008, 14, 159.
10. Davies, T. D.; Verdonk, M. L.; Graham, B.; Saalau-Bethell, S.; Hamlett, C. C. F.;
McHardy, T.; Collins, I.; Garrett, M. D.; Workman, P.; Woodhead, S. J.; Jhoti, H.;
Ruger, P.; Jucknischke, U.; Schnieder, T.; Huber, R.; Bossemeyer, D.; Engh, R. A. J.
Mol. Biol. 2003, 329, 1021.
12. (a) Saxty, G.; Woodhead, S. J.; Berdini, V.; Davies, T. G.; Verdonk, M. L.; Wyatt, P.
G.; Boyle, R. G.; Barford, D.; Downham, R.; Garrett, M. D.; Carr, R. A. J. Med.
Chem. 2007, 50, 2293–2296; (b) Lippa, B.; Pan, G.; Corbett, M.; Li, C.; Kauffman,
G. S.; Pandit, J.; Robinson, S.; Wei, L.; Kozina, E.; Marr, E. S.; Borzillo, G.; Knauth,
E.; Barbacci-Tobin, E. G.; Vincent, P.; Troutman, M.; Baker, D.; Rajamohan, F.;
Kakar, S.; Clark, T.; Morris, J. Bioorg. Med. Chem. Lett. 2008, 18, 3359–3363.
13. The absolute configuration of the azetidines was determined by Mosher ester
analysis – see Supplementary data.
Barford, D. J. Mol. Biol. 2007, 367, 882; For
a description of a similar
crystallography strategy see: Caldwell, J. J.; Davies, T. G.; Donald, A.;
McHardy, T.; Rowlands, M. G.; Aherne, G. W.; Hunter, L. K.; Taylor, K.;
Ruddle, R.; Raynaud, F. I.; Verdonk, M.; Workman, P.; Garrett, M. D.; Collins, I. J.
Med. Chem. 2008, 51, 2147.
11. Mutations: PKA Val123 was mutated to AKT Ala, and PKA Leu173 was mutated
to AKT Met. Additionally, PKA Lys181 was mutated to AKT Gln, but this residue
faces away from the ATP binding pocket. Gassell, M.; Breitenlechner, C. B.;
14. Billotte, S. Synlett 1998, 4, 379–380.