BACE-1 inhibitors part 3: Identification of hydroxy ethylamines (HEAs) with nanomolar potency in cells

Article abstract of DOI:10.1016/j.bmcl.2007.12.020

This article is focusing on further optimization of previously described hydroxy ethylamine (HEA) BACE-1 inhibitors obtained from a focused library with the support of X-ray crystallography. Optimization of the non-prime side of our inhibitors and introdu

Full text of DOI:10.1016/j.bmcl.2007.12.020

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 1022–1026
BACE-1 inhibitors part 3: Identification of hydroxy
ethylamines (HEAs) with nanomolar potency in cells
*
Paul Beswick, Nicolas Charrier, Brian Clarke, Emmanuel Demont, Colin Dingwall,
Rachel Dunsdon, Andrew Faller, Robert Gleave, Julie Hawkins, Ishrut Hussain,
Christopher N. Johnson, David MacPherson, Graham Maile, Rosalie Matico,
Peter Milner, Julie Mosley, Alan Naylor, Alistair O’Brien, Sally Redshaw, David Riddell,
Paul Rowland, John Skidmore, Virginie Soleil, Kathrine J. Smith, Steven Stanway,
Geoffrey Stemp, Alistair Stuart, Sharon Sweitzer, Pam Theobald, David Vesey,
Daryl S. Walter, John Ward and Gareth Wayne
Neurology and Gastrointestinal Centre of Excellence for Drug Discovery, GlaxoSmithKline R&D,
New Frontiers Science Park, Third Avenue, Harlow, Essex, CM19 5AW, UK
Received 30 October 2007; revised 7 December 2007; accepted 11 December 2007
Available online 15 December 2007
Abstract—This article is focusing on further optimization of previously described hydroxy ethylamine (HEA) BACE-1 inhibitors
obtained from a focused library with the support of X-ray crystallography. Optimization of the non-prime side of our inhibitors
and introduction of a 6-membered sultam substituent binding to Asn-294 as well as a fluorine in the C-2 position led to derivatives with
nanomolar potency in cell-based assays.
Ó 2007 Elsevier Ltd. All rights reserved.
1
In two previous papers, we outlined how BACE-1
inhibitors with sub-micromolar potency in cells express-
ing APP wild-type (WT) were obtained from a focused
library with the help of X-ray crystallography. To max-
imize the chance of achieving efficacy in vivo, we wanted
to achieve low nanomolar potency in a cell based assay.
We therefore further investigated the SAR of our
inhibitors.
and initial efforts focused on replacing the pyrrolidinone
2
substituent present in compound 1.
It proved possible to replace the lactam with other H-
bond acceptors (HBA) such as ketones 2 and 3 or nitrile
4 with some improvements in potency, albeit with lower
selectivity (Table 1). Much more impressive increases in
potency were seen when sultams were introduced: Inhib-
itors 5 and particularly 6 were significantly more potent
(at least 25-fold) and selective (up to 1000-fold against
Cat-D) than all the previously prepared analogues, with
a non-peptidic prime side, and further work focused on
the 6-membered sultam as HBA to Asn-294.
As the SAR on the prime side of our inhibitors and at
the S pocket had already been extensively explored
3
and the substitution pattern optimized accordingly, it
was felt that the best chance of achieving significant in-
crease in potency lay in further optimization of the sub-
stituents which had been shown to form a H-bond to
Asn-294, or which would bind more tightly in the S1
pocket. The H-bonding interaction was examined first
This increase in potency in the enzyme assay also trans-
lated to increased cellular activity. In fact, the level of
potency achieved in the enzyme assay at this stage led
to a second cell-based assay being introduced to help
in ranking of relative potencies of the compounds. This
additional cell assay was similar to the one already in
use, but used cells expressing the Swedish (SWE) APP
Keywords: Alzheimer; BACE-1; Aspartic protease; Hydroxy
ethylamine.
*
Corresponding author. Tel.: +44 (0) 1279643468; fax: +44 (0)
279622727; e-mail: emmanuel.h.demont@gsk.com
3
substrate in place of the wild-type (WT) substrate.
1
0
960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.12.020

P. Beswick et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1022–1026
Table 1. Substitution of the pyrrolidinone as hydrogen bond acceptor
1023
Somewhat unexpectedly, a significant decrease in po-
tency was observed when comparing the lowering of
amyloid production in the cells expressing SWE APP
to that seen in the cells expressing WT APP (Table 2).
The reason for this reduction in the level of potency
was not entirely clear as the FRET enzyme assay also
used the SWE substrate. Nonetheless, to progress com-
pounds further, we resolved to maximize potency in
both cell lines by further rounds of optimization.
(HBA) interacting with Asn-294
O
1
R
CF3
N
H
N
H
O
O
Due to the similar binding modes, the SAR observed in
the S pocket in both the lactam and sultam series was
similar: O- and N-linked derivatives were more potent
than their carbon analogues (compare 6 and 7 with 8),
and a linear three-atom substituent gave the best com-
promise between activity and selectivity (compare 7
1
R
a,b
Compound
BACE-1
IC50 (nM)
BACE-2
Cat-D
3
IC50 (nM) IC50 (nM)
O
1
380 (2)
26,840
4130
34,325
1750
N
and 9). The N-linked S substituents were also generally
3
O
more potent than their oxygen analogues, particularly in
cells expressing SWE APP (compare 10 and 11 for
example) and consequently the non-prime side of inhib-
itor 7 was selected as the basis for further optimization.
Co-crystallisation confirmed that these derivatives had a
similar binding mode to that seen with the correspond-
2
3
250 (1)
280 (1)
O
4
3080
2940
ing 5-lactam derivatives (Fig. 1).
N
4
5
160 (1)
50 (1)
1050
4680
2020
8510
O
S
O
N
O
S
6
O
6 (1)
1020
5890
N
a
In all tables, IC50s reported are means of the values of n different
experiments, n being reported in bracket and identical for BACE-1,
BACE-2 and Cat-D. Each IC50 is within 3-fold of the mean value.
See Ref. 3 for protocol. FRET-based assay.
b
Figure 1. Superimposition of 6-sultam (light-blue) and 5-lactam
(
magenta) derivatives bound to BACE-1.
Table 2. SAR at the S
3
pocket in the 6-membered sultam series
O
S
N
O
O
2
R
N
H
N
H
OH
1
X
R
1
R
2
R
a
a
a
a
Compound
X
BACE-1
BACE-2
Cat-D
WT Ab40
WT Ab42
SWE Ab40
SWE Ab42
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
6
7
8
9
O
N
C
N
O
N
C
C
C
2
2
2
H
H
H
5
5
5
CF
CF
CF
CF
3
3
3
3
6 (1)
3 (2)
5 (1)
11 (1)
5 (1)
5 (1)
1020
1430
1450
1660
760
5890
3900
27
21
40
58
25
47
21
30
40
62
18
34
715
529
844
910
711
102
205
175
366
414
216
39
4790
CH(CH
3
)
2
25,120
6170
1
0
1
C
C
2
2
H
H
5
OCF
OCF
3
3
1
5
600
2570
a
See Ref. 3 for protocol. IC50 values are means of at least two separate experiments.

1024
P. Beswick et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1022–1026
Table 3. SAR at prime side in the sultam series
O
S
N
O
O
1
R
N
H
N
H
OH
HN
1
R
Compound
BACE-1
BACE-2
Cat-D
WT Ab40
WT Ab42
SWE Ab40
SWE Ab42
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
CF3
7
3 (2)
1430
3900
21
30
529
—
175
—
1
2
3
120 (1)
32,360
2290
26,300
3090
573
565
1
34 (1)
229
154
—
—
a
1
4
18 (1)
12 (2)
2950
6920
1070
36
78
26
72
—
—
O
1
5
N
56,230
3633
1113
N
a
Obtained as a 1:1 mixture of isomers.
Table 4. SAR at the P
1
position
O
S
N
1
R
O
O
OMe
N
H
N
H
O
HN
1
R
Compound
BACE-1
BACE-2
Cat-D
WT Ab40
WT Ab42
SWE Ab40
SWE Ab42
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
16
17
18
8 (2)
4270
630
6760
19
41
928
450
—
299
203
—
F
5 (1)
2400
26
30
F
O
40 (1)
13,215
15,140
129
108
5
5
1
9
0
N
2690 (1)
5750 (1)
>10
>10
—
—
—
—
—
—
—
—
N
5
5
2
>10
>10

P. Beswick et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1022–1026
1025
The SAR of the benzylic prime side was similar to that
1
As shown in Table 4, compounds with substituted aryls
(compound 17) or electron rich heteroaryls (compound
18) were more potent and selective than electron poor
heteroaryls (compound 19) or alkyl substituents (com-
pound 20). However, these compounds offered no
advantage compared to the lead 16. These results, and
the fact that the core of compound 16 could be obtained
from readily available phenylalanine, led us to explore
other ways of improving potency.
previously observed in the lactam series: meta-sub-
stitution was important for activity (compare 7 and
1
2, Table 3); Constraining the benzylic group appeared
beneficial for activity (compare 13 and 12) but meta-
substituted constrained benzylic derivatives were disap-
pointingly not as potent as expected (compare the in-
crease in activity from 13 to 14 versus the difference in
activity between 12 and 7). It was possible to replace
the prime side phenyl ring by heteroaryls (compound
15) but overall, a compound with nanomolar potency
across all assays was not observed.
An inhibitor with nanomolar potency in all of the key
assays appeared elusive but, in another series of
BACE-1 inhibitors we were developing, the introduc-
6
At this stage it was felt that compounds with increased
activity might be obtained by further optimization of
the P substituent and an array of compounds with sat-
1
tion of a fluorine in the 2-position of the benzamide
non-prime side led to an increase in cell activity. It
was hoped that this effect might also be observed in
the sultam series. The synthetic route used to access
the key building blocks that were required to test this
5
urated and unsaturated P substituents was prepared.
1
7
idea is depicted in Figure 2.
F
O
F
O
F
O
O N
H N
Satisfyingly, it was found that the introduction of this
fluoro substituent did indeed prove to be beneficial
a, b
2
c
2
O
O
OH
8
7
2%
8
0%
(compare 16–21, Table 5) and excellent selectivity could
NO2
NO2
be achieved, particularly against Cat-D (compound 22).
From a set of derivatives bearing a meta-substituted
benzyl amine prime side, GSK188909 (compound 23)
was identified as one of the first selective, nanomolar
inhibitors in the cells expressing SWE APP assay.
GSK188909 also had a favourable pharmacokinetic
profile which allowed us to determine whether a
BACE-1 inhibitor would lower amyloid production in
an animal model of Alzheimer’s disease. The positive
O
O
S
N
S
N
O F
O
O F
O
g, h, i
d, e, b, f
7%
OH
O
5
5%
6
HN
NO2
Figure 2. Reagents and conditions: (a) HNO
b) MeOH, H SO
Cl(CH SO Cl, NEt
f) NEt , EtOH, reflux; g) Pd/C, NH
CH CHO, NaHB(OAc) , AcOH, (CH
O, 25 °C.
3
, H
2
SO 10 °C to 95 °C;
4
3
read-out of this study has been reported previously: fol-
(
2
4
, reflux; (c) Fe, AcOH, T < 35 °C (1 isomer); (d)
, CH Cl O, 25 °C;
, 25 °C; (e) NaOH, MeOH/H
COOH, EtOH/H O, reflux; (h)
Cl)
, 25 °C; (i) NaOH, THF/
lowing oral administration (250 mg/kg twice daily for 5
days), decrease of brain Ab40 (18%) and Ab42 (23%)
was observed in TASTPM mice. This lowering can be
significantly enhanced (up to 68% and 55% for Ab40
2
)
4
2
3
2
2
2
(
3
4
2
3
3
2
2
H
2
Table 5. Activity and selectivity of inhibitors bearing a 2-F benzamide non-prime side
O
S
N
1
O R
O
2
R
N
H
N
H
O
NH
1
R
2
R
Compound
BACE-1
BACE-2
Cat-D
WT Ab40
WT Ab42
SWE Ab40
SWE Ab42
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
IC50 (nM)
O
O
16
21
22
23
H
F
F
F
8 (2)
4 (4)
4270
785
6760
6545
19
14
14
5
41
12
12
5
928
283
361
40
299
92
N
6 (2)
2465
177
87,100
2653
176
18
N
CF3
4 (81)

1026
P. Beswick et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1022–1026
and Ab42, respectively) after a single 250 mg/kg oral
dose when GSK188909 is co-administed with a Pgp
inhibitor.
4. See preceding paper for references on BACE-1 construct
used. The X-ray data were collected at ESRF beamline
ID14-4. The PDB deposition code for the BACE-1 complex
˚
crystal structure is 2vij. The structure was refined to 1.6 A
resolution (R = 0.187, Rfree = 0.215). It is difficult to explain
why the sultams appeared more potent than the lactam as
sulfonamides are in general less powerful HBAs than
amides, see: Abraham, M. H.; Duce, P. P.; Prior, D. V.;
Barratt, D. G.; Morris, J. J.; Taylor, P. J. J. Chem. Soc.
Perkin Trans. II. 1989, 10, 1355, This increase in potency
could be explained by a better orientation of the S@O bond
compared to the C@O bond or by a higher similarity
between low energy conformation and enzyme-bound
conformation for the sultam compared to the lactam.
. For the synthesis of the corresponding epoxides, see: Dalla
Croce, P.; La Rosa, C.; Pizzatti, E. Tetrahedron: Asymme-
try 2000, 11, 2635; US Patent App. 2003/0004360(A1).
. Manuscript in preparation.
. For experimental procedures, see: Demont, E.; Faller, A.;
Macpherson, D.; Milner, P.; Naylor, A.; Redshaw, S.;
Stanway, S.; Vesey, D. Walter, D. PCT Int. Appl. (2004),
WO 2004050619.
To conclude, we have shown that further exploration of
the SAR in this HEA series allowed us to achieve the le-
vel of potency in both of our key cellular assays. In par-
ticular, optimization of the HBA binding to Asn-294
and introduction of a fluorine atom in the C-2 position
of our benzamide non-prime side led to a 100-fold in-
crease in potency. These inhibitors represent useful tools
for the further study, in animal models, of the possibil-
ities for treating Alzheimer’s disease via BACE-1 inhibi-
tion. Further optimization of these derivatives towards
more drug-like inhibitors will be reported in due
5
6
course.
6
7
References and notes
1
2
. See the two preceding papers.
. An –OC H5 S3 substituent was used rather than an –
8. This trend is observed regardless of the nature of the group
forming a H-bond with Asn-294 or the group binding into
2
2 5
NHC H substituent in order to simplify the synthesis of
analogues.
the S pocket. The 2-MeO analogue has similar potency
3
3
. Hussain, I.; Hawkins, J.; Harisson, D.; Hille, C.; Wayne,
G.; Cutler, L.; Buck, T.; Walter, D.; Demont, E.; Howes,
C.; Naylor, A.; Jeffrey, P.; Gonzalez, M. I.; Dingwall, C.;
Michel, A.; Redshaw, S.; Davis, J. B. J. Neurochem. 2007,
and selectivity to the 2-H derivative. X-ray structures of
inhibitors bound to BACE-1 having 2-H or 2-F substituents
can be superimposed, hence it is difficult to rationalise why
2-F derivatives are more potent. An inductive effect may be
part of the explanation.
100, 802.