1,4-Oxazine β-Secretase 1 (BACE1) Inhibitors: From Hit Generation to Orally Bioavailable Brain Penetrant Leads

Article abstract of DOI:10.1021/acs.jmedchem.5b01101

1,4-Oxazines are presented, which show good in vitro inhibition in enzymatic and cellular BACE1 assays. We describe lead optimization focused on reducing the amidine pK_a while optimizing interactions in the BACE1 active site. Our strategy permi

Full text of DOI:10.1021/acs.jmedchem.5b01101

Subscriber access provided by MONASH UNIVERSITY
Article
1,4-Oxazine #-secretase (BACE1) inhibitors: from hit
generation to orally bioavailable brain penetrant leads.
Frederik J.R. Rombouts, Gary Tresadern, Óscar Delgado, Carolina Martinez Lamenca, Michiel Van Gool,
Aránzazu García-Molina, Sergio A. Alonso De Diego, Daniel Oehlrich, Hana Prokopcova, Jose Manuel
Alonso, Nigel Austin, Herman Borghys, Sven Van Brandt, Michel Surkyn, Michel De Cleyn, Ann Vos,
Richard Alexander, Gregor J. Macdonald, Dieder Moechars, Harrie J.M. Gijsen, and Andrés A. Trabanco
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.5b01101 • Publication Date (Web): 17 Sep 2015
Downloaded from http://pubs.acs.org on September 21, 2015
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 123
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
,4-Oxazine β-secretase 1 (BACE1) inhibitors: from
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
hit generation to orally bioavailable brain penetrant
leads.
a,
f
b
a
a
Frederik J. R. Rombouts, * Gary Tresadern, Oscar Delgado, Carolina Martínez-Lamenca,
b
b
b
Michiel Van Gool, Aránzazu García-Molina, Sergio A. Alonso de Diego, Daniel Oehlrich,
a
f
d
d
a
Hana Prokopcova, José Manuel Alonso, Nigel Austin, Herman Borghys, Sven Van Brandt,
a
a
d
e
a
Michel Surkyn, Michel De Cleyn, Ann Vos, Richard Alexander, Gregor Macdonald, Dieder
c
a
b,
Moechars, Harrie Gijsen, Andrés A. Trabanco *
a
Neuroscience Medicinal Chemistry, Janssen Research & Development, Janssen Pharmaceutica
b
N. V., Turnhoutseweg 30, B-2340 Beerse, Belgium. Neuroscience Medicinal Chemistry,
Janssen Research & Development, Janssen-Cilag S. A., C/ Jarama 75A, 45007 Toledo, Spain.
c
Neuroscience-Biology, Janssen Research & Development, Janssen Pharmaceutica N. V.,
d
Turnhoutseweg 30, B-2340 Beerse, Belgium. Discovery Sciences, Janssen Research &
Development, Janssen Pharmaceutica N. V., Turnhoutseweg 30, B-2340 Beerse, Belgium.
e
Biologics Research, Janssen Research & Development, 1400 McKean Rd, Spring House, PA
f
1
9002. Discovery Sciences, Janssen Research & Development, Janssen-Cilag S. A., C/ Jarama
7
5A, 45007 Toledo, Spain.
BACE1 inhibitor, Morpholine, 1,4-Oxazine, Amidine, Alzheimer’s disease
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT: 1,4-Oxazines are presented, which show good in vitro inhibition in enzymatic and
cellular BACE1 assays. We describe lead optimization focused on reducing the amidine pKa
whilst optimizing interactions in the BACE1 active site. Our strategy permitted modulation of
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
properties such as permeation and especially P-glycoprotein efflux. This led to compounds
which were orally bioavailable, centrally active and which demonstrated robust lowering of brain
and CSF Aβ levels respectively in mouse and dog models. The amyloid lowering potential of
these molecules makes them valuable leads in the search for new BACE inhibitors for the
treatment of Alzheimer’s Disease.
INTRODUCTION
Alzheimer´s disease (AD) is a progressive neurodegenerative disorder and the major cause of
dementia in the elderly. According to the World Health Organization (WHO) dementia affects
4
7.5 million people worldwide, a number that is projected to triple by 2050 with the aging
1
,2
population. AD is characterized by progressive deposition of amyloid and misfolded Tau,
3
followed by neurodegeneration and loss of function, leading ultimately to death. Of the many
attempts to target the underlying pathogenesis, amyloid lowering approaches have made the most
4
progress so far, albeit without delivering a therapy to date. Amyloid oligomer and plaque
formation is thought to occur when the balance between non-amyloidogenic (α-secretase
mediated) and the amyloidogenic processing of amyloid precursor protein (APP) is shifted
5
,6
(
familial AD) or clearance of amyloid is impaired (sporadic AD). In the amyloidogenic
pathway, β-secretase 1 (BACE1) cleaves APP producing a 99 amino acid length soluble peptide
fragment called C99, which is the rate-limiting step in Aβ formation. This peptide is further
processed by γ-secretase to 36-43 amino acid length Aβ species, of which the longer isoforms,
ACS Paragon Plus Environment

Page 3 of 123
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
7
especially Aβ42, are the most fibrillogenic and neurotoxic. Consequently both β-secretase and
8
γ-secretase are being pursued as targets to modulate Aβ production. Since its discovery in
9
1
999, BACE1 has been a highly challenging target for drug discovery, and only after years of
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
research medicinal chemists have managed progressing small molecule BACE1 inhibitors in
1
0
clinical trials. A breakthrough in the development of non-peptidomimetic BACE1 inhibitors
was the identification of amidine- and guanidine-containing small molecules. Compared to
previous peptidomimetic and amino-alcohol derived inhibitors, these molecules form a salt
bridge and hydrogen bond interactions with Asp32 and Asp228 in the catalytic site of BACE1 in
1
1
an optimal way (see schematic in Chart 1). The use of a quaternary center alpha to the amidine
or guanidine function permits substituents to enter adjacent binding pockets such as S2’, S1 and
12
S3. The S3 pocket can be efficiently targeted via amide-tethered biaryl systems (Chart 1). In
these amidine prototypes, the central aromatic ring (A) is a direct substituent on the quaternary
center. The A-ring and amide nitrogen occupy the S1 pocket, whereas the distal aromatic ring
(B) extends into the S3 pocket. The B-ring is generally a 2-pyridyl or 2-pyrazinyl ring, allowing
for a quasi-coplanar orientation with the A-ring. The amide NH establishes an interaction with
the backbone carbonyl oxygen of Gly230. The fourth substituent on the quaternary center
(methyl as shown in the examples in Chart 1) is directed towards the S2’ pocket. As such, the
field has progressed towards potent inhibitors with low molecular weight and good ligand
efficiency.
A challenge in the development of amidine and guanidine-based BACE1 inhibitors has been to
modulate the intrinsic high basicity of the amidine function. In our previously reported
13
benzoguanidine series, we found this to be a key factor for obtaining compounds with
favorable pharmacokinetic (PK) properties, as high basicity (pK ~ 10-11) has been associated
a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
14
with unfavorable tissue distribution and P-glycoprotein (P-gp) mediated efflux. Amidine-based
1
5
warheads with pK below 9 have been reported by a number of companies in the form of acyl-
a
1
6
17
18
and sulfonylguanidines, isoureas and isothioureas for which representative examples 1 and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
19,20
2
are shown in Chart 1.
We have previously reported on the de novo design and synthesis of
2
1
piperazinones 3 and 4 as BACE1 inhibitors with moderate enzymatic and cellular activity.
Interestingly, the amide function embedded in the ‘warhead’ of 3 and 4 helped to decrease the
basicity of the amidine leading to compounds with in vivo activity. However, the blood brain
barrier crossing with this piperazinone series was found to be sub-optimal, and concomitantly
subcutaneous administration of high doses was needed to achieve significant in vivo reduction of
Aβ peptides in mice.
S2'
O
O
S
N
S
N
H
+
O
N
H
N
H
F
H 2
N
N
H 2
N
Asp228
F
S1
A
Asp32 H N
Gly230 N
BX
H N
H N
O
N
O
O
N
F
F
S3
1
(LY-2886721, Eli Lilly) 2 (MK-8931, Merck)
Pharmacophore of BACE1 inhibitors
targeting the S3 pocket (X= CH, N)
BACE1 IC50 15.5 nM
BACE1 IC50 3.2 nM
hAβ42 cell IC50 6.6 nM
hAβ42 cell IC50 1.3 nM
1
O
R 1
N
N
O
O
N
N
6
R 2
2
R 3
3
5
F
R 4
H 2
N
H 2
N
H 2
N
N
4
(N)
H N
H N
H N
O
O
O
N
N
N
X
Cl
Cl
R 5
3
(Janssen)
4 (Janssen)
BACE1 IC50 23 nM
hAβ42 cell IC50 28 nM
5, This paper
BACE1 IC50 324 nM
hAβ42 cell IC50 36 nM
ACS Paragon Plus Environment

Page 5 of 123
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Chart 1. Representative overview of reported amidine containing BACE1 inhibitors targeting
the S3 pocket. Internally generated enzymatic and cellular data are shown.
Herein we report the design and synthesis of novel 1,4-oxazines (5) as alternative warheads to
the previously described piperazinones 3 and 4. Lead optimization strategies to modify the pKa
of the amidine function resulted compounds (2R,3R)-7a and (2R,3R)-7d with a robust oral effect
in lowering of Aβ peptides in mouse and dog models.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CHEMISTRY
In order to access 1,4-oxazines 3R-6a-c, acetophenones 13a,b were converted to the
corresponding amino nitriles rac-14a,b via a Strecker reaction (Scheme 1). Acidic hydrolysis of
the cyano group and subsequent esterification of the resulting carboxylic acids rac-15a,b led to
the aminoesters rac-16a,b. Reduction of rac-16a,b with lithium aluminium hydride (LAH)
provided the racemic amino alcohols rac-17a,b. Separation of enantiomers via chiral
supercritical fluid chromatography (SFC) followed by determination of the absolute
stereochemistry by vibrational circular dichroism (VCD) provided the desired enantiomers 2R-
22
1
7a,b. The lactam rings 5R-18a,b were constructed next via a one-pot two step procedure
involving amide formation at -15 °C in the presence of DIPEA, followed by Williamson
etherification with t-BuOK. Subsequently, the corresponding amidine derivatives 5R-20a,b were
obtained by sequential thionation of 5R-18a,b with P S , followed by aminolysis of the resulting
2
5
thioamides 5R-19a,b with aqueous ammonia. Amination reaction on the bromoarene under
Buchwald-type conditions using benzophenone imine as the nitrogen source led to 5R-21a,b.
Finally, the target compounds 3R-6a-c were obtained via HATU or DMTMM mediated coupling
of 5R-21a,b with the corresponding carboxylic acids.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Scheme 1. Synthesis route towards 1,4-oxazines 3R-6a-c.
H O
H O
O
O
N
O
R 1
O
R 1
CN
R 1
R 1
R
1
a
b
c
d
H 2
N
H 2
N
H 2
N
H 2
Br
Br
Br
Br
Br
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
1
3a (R = H)
rac-14a (R = H): 86%
rac-15a (R = H): 71%
rac-16a (R = H): 89%
rac-17a (R = H): 85%
1
1
1
1
1
3b (R = F)
rac-14b (R = F): 60%
rac-15b (R = F): 68%
rac-16b (R = F): 87%
rac-17b (R = F): 96%
1
1
1
1
1
H O
O
O
O
R 1
R 1
R 1
R 1
f
e
h
i
H 2
N
X
N
H
H 2
N
N
H 2
N
N
Br
Br
Br
N H 2
2
R-17a (R = H): 30% 3 steps
5R-18a (R = H, X= O): 88%
5R-20a (R = H): 69%
5R-21a (R = H): 45 %
1
1
1
1
2
R-17b (R = F): 31% 3 steps
5R-18b (R = F, X = O): 89% 5R-20b (R = F): 63% 2 steps 5R-21b (R = F): 64%
1
1
1
1
g
5R-19a (R = H, X = S): 93%
1
5R-19b (R = F, X = S): used as such
1
N
R =
2
O
N
3
3
R-6a: j
R-6b,c: k H 2
Cl
R 1
N
3
3
R-6a (R = H): 41%
1
R-6b (R = F): 78%
1
H N
N
O
R =
2
R 2
O
N
3R-6c (R = F): 70%
1
a
Reagents and conditions: (a) TMSCN, NH Cl, 7 N NH /MeOH, rt, 4 d; (b) 6 N HCl, reflux,
4
3
1
6 h; (c) H SO , MeOH, reflux, 16 h; (d) LiAlH , THF, 0 °C, 1 h; (e) Chiral SFC separation; (f)
i. chloroacetyl chloride, DIPEA, THF, -78 °C, 30 min, ii. t-BuOK, 0 °C, 90 min; (g) P S , THF,
2 4 4
2
5
5
0 °C, 50 min; (h) 33% NH (aq), 60 °C, 4 h; (i) i. benzophenone imine, Pd (dba) , rac-BINAP,
3
2
3
t-BuONa, toluene, 80 °C, 7 h, ii. 1 N HCl, rt, 16 h (j) 5-chloropyridine-2-carboxylic acid, HATU,
N,N-dimethylaniline, DCM, rt, 5 h; (k) 5-methoxypyrazine-2-carboxylic acid, DMTMM, MeOH,
rt, 3 h.
The synthesis routes for target molecules (2S,3R)-7a, (2R,3R)-7a, (2S,3R)-8, (2R,3R)-8,
(
2R,3R)-9 and (2S,3R)-9 bearing electron withdrawing groups (EWG) at the C-2 position start
from the common intermediate 2R-15b (Scheme 2),which was obtained after chiral SFC
separation of the racemate 15b and VCD characterization. Thus, 2R-15b was cyclized to the
morpholinedione 3R-22 upon reaction with chloroacetyl chloride in a two-step procedure: first
the amide bond was formed by treatment with NaOH in 1,4-dioxane, followed by aqueous work-
up and subsequent lactonization by addition of NaHCO in DMF. Compound 3R-22 proved to be
3
ACS Paragon Plus Environment

Page 7 of 123
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a valuable precursor to access all envisioned modifications (I-III, Chart 2). Reduction of the
lactone carbonyl with DIBAL-H to the corresponding hemi-acetal, followed by fluorination with
DAST provided 5R-23 as an inseparable 60:40 mixture of 6R and 6S diastereomer. Addition of
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Ruppert-Prakesh reagent (TMS-CF ) to 3R-22 provided hemi-ketal 5R-24 as an unassigned 3:1
3
mixture of 6S/6R diastereomers. Interestingly 5R-24 also reacted smoothly with DAST providing
the 6-fluoro-6-trifluoromethyl-1,4-oxazine derivatives (5R,6R)-25 and (5R,6S)-25 as inseparable
mixture in a 25:75 diastereomeric ratio. Hemi-ketal 5R-24 was also chlorinated by reaction with
SOCl leading to 5R-26. Treatment of 5R-26 with metallic zinc in acetic acid at 100 °C provided
2
the CF derivative 5R-27, as a 80:20 mixture of 6R and 6S diastereomer respectively.
3
a
Scheme 2. Synthesis of 1,4-oxazine-3-one precursors for BACE1 targeting amidines.
O
O
O
O
F
6
2
H O
R
R
H 2
N
F
F
b, c
F
a
O
N
H
3
O
N
H
5
R
9
6%
66%
Br
Br
Br
2R-15b
3R-22
5R-23 (6R/6S 60:40)
d
98%
CF 3
O
CF 3
O
F
6
O H
R
F
O
N
H
R
F
O
N
H
5
c
Br
9
3%
Br
5
R-24 (6R/6S 3:1)
5R-25 (6R/6S 25:75)
e
41%
CF 3
Cl
O
O
CF 3
R
R
F
O
N
H
F
O
N
H
f
9%
Br
Br
9
5
R-26
5R-27 (6R/6S 80:20)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Reagents and conditions: (a) i. Chloroacetyl chloride, 1 M NaOH , 1,4-dioxane, 2 h, ii. DMF,
NaHCO , 80 °C, 3 h; (b) DIBAL-H, THF, -78 °C, 2 h; (c) DAST, DCM, 0 °C, 40 min (5R-23) to
h (5R-25); (d) TMSCF , TBAT, THF, 0 °C to rt, 20 min; (e) SOCl , DCM, 0 °C, 30 min, then
add pyridine, 0 °C, 30 min; (f) Zn, AcOH, 80 °C, 3 h.
3
2
3
2
As shown in Scheme 3, 6R/6S diasteromeric mixtures 5R-23, 5R-25 and 5R-27 were converted
to the target amidines (5R,6S)-31, (5R,6R)-31, (5R,6S)-32, (5R,6R)-32, (5R,6R)-33 and (5R,6S)-
33 following a similar reaction sequence to that previously described for 3R-6a-c in Scheme 1.
The only modification in the synthesis route was the conversion of the bromoarene to the aniline,
for which we found a copper-catalyzed reaction with sodium azide to provide superior yields to
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
the Buchwald-Hartwig protocol. For occupying the S3 pocket the 5-methoxypyrazin-2-yl group
was consistently used to allow for a comparison of properties between the different warheads in
final compounds (2S,3R)-7a, (2R,3R)-7a, (2S,3R)-8, (2R,3R)-8, (2R,3R)-9 and (2S,3R)-9.
a
Scheme 3. Preparation of 1,4-oxazine-based amidines C-2 substituted with different EWG.
R 1
R 1
R 1
O
R 2
O
R
2
O
N
R
2
F
a
F
b
F
O
N
H
S
N
H
H 2N
Br
Br
Br
(5R,6S)-31 (R = F, R = H): 9%
1 2
5
5
5
R-23 (R = F, R = H)
5R-28 (R = F, R = H): 68%,
1
2
1
2
(5R,6R)-31 (R = H, R = F): 56%
1 2
6
0:40 mix of 6R and 6S
62:38 mix of 6R and 6S
R-25 (R = CF , R = F)
5R-29 (R = CF , R = F): 88%,
(5R,6S)-32 (R = CF , R = F): 21%
1 3 2
1
3
2
1
3
2
c
2
5:75 mix of 6R and 6S
25:75 mix of 6R and 6S
(5R,6R)-32 (R = F, R = CF ): 8%
1 2 3
R-27 (R = CF , R = H)
5R-30 (R = CF , R = H): 92%
1
3
2
1
3
2
(5R,6R)-33 (R = CF , R = H): 59%
1 3 2
8
0:20 mix of 6R and 6S
80:20 mix of 6R and 6S
(5R,6S)-33 (R = H, R = CF ): 13%
1
2
3
R 1
R 1
O
O
N
R
2
R 2
(
(
(
(
(
(
2S,3R)-7a (R = F, R = H): 30%
1 2
2
6
d
F
F
2R,3R)-7a (R = H, R = F): 65%
1
2
H 2
N
N
H 2N
3
5
2S,3R)-8 (R = CF , R = F): 45%
1
3
2
2R,3R)-8 (R = F, R = CF ): 69%
1
2
3
H 2
N
H N
2R,3R)-9 (R = CF , R = H): 65%
1
3
2
O
2S,3R)-9 (R = H, R = CF ): 23%
(5R,6S)-34 (R = F, R = H): 24%
1 2
1
2
3
N
(
5R,6R)-34 (R = H, R = F): 24%
1 2
N
(5R,6S)-35 (R = CF , R = F): 78%
1 3 2
O
(
5R,6R)-35 (R = F, R = CF ): 82%
1 2 3
(
5R,6R)-36 (R = CF , R = H): 90%
1 3 2
(
5R,6S)-36 (R = H, R = CF ): 41%
1 2 3
ACS Paragon Plus Environment

Page 9 of 123
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a
Reagents and conditions: (a) P S , THF, 70 °C, 2-4 h; (b) 33 % NH (aq), 60-80 °C, 24-48 h;
2
5
3
(
c) NaN , CuI, DMEDA, Na CO , DMSO, 110 °C, 1-4 h; (d) 5-methoxypyrazine-2-carboxylic
acid, DMTMM, MeOH, 0 °C, 2-6 h.
3 2 3
A novel stereoselective route using Ellmann’s sulfonamide was developed for the synthesis of
compound (2R,3R)-10 (Scheme 4). Condensation of S-Ellmann’s sulfonamide with
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
phenylglyoxylic ester (37) in the presence of Ti(OiPr) yielded the S-sulfoximine S-38.
4
Conveniently, complete transesterification of the methyl ester with isopropanol occurred during
the condensation step generating the isopropyl ester, which would be less reactive to the addition
of cyclopropyl magnesium bromide in the subsequent reaction step. Similar to the protocol
described in ref. 22, treatment of S-38 with cyclopropylmagnesium bromide led exclusively to
the imine addition product 2R-39 with R-configuration at the quaternary center. Ester hydrolysis
followed by the cleavage of the sulfoxamide S-N bond led to amino acid 3R-40, which was next
cyclized to the lactone 3R-41 following a similar route to the one previously described for 3R-22.
Reduction of 3R-41 with DIBAL-H followed by reaction with DAST provided 5R-42 as a 3:2
mixture of 6R:6S diastereomers. Next, thionation with P S to 5R-43 and subsequent aminolysis,
2
5
yielded the expected (5R,6R)-44 after separation of the minor (5R,6S)-diastereoisomer by
column chromatography. Finally, copper-catalyzed coupling of (5R,6R)-44 with sodium azide to
give aniline (5R,6R)-45, followed by amide formation with 5-methoxypyrazine-2-carboxylate
yielded compound (2R,3R)-10.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Scheme 4. Enantioselective synthesis of C-3 cyclopropyl substituted 1,4-oxazine 9.
S
S
O
N
O
S
N
S
O
O
F
O
O
F
R
F
H O
H 2
N H
O
a
b
c, d
e
O
F
R
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
O
Br
8
9%
97%
72%
99%
Br
Br
Br
2
steps
37
S-38
2R-39
2R-40
j
O
O
O
F
O
N
F
f, g
i
O
N
H
X
N
H
H 2N
F
F
F
92%
2 steps
33%
2 steps
59%
Br
Br
Br
3
R-41
X = O: 5R-42 (6R/6S 3:2)
X = S: 5R-43 (6R/6S 3:2)
(5R,6R)-44
h
O
N
F
O
6
F
2
3
k
5
H 2
N
H 2
N
N
F
F
5
0%
O
N
N
N
H
H 2
N
O
(2R,3R)-10
(
5R,6R)-45
a
Reagents and conditions: (a) S-tBuSONH , Ti(OiPr) , heptane, 80 °C, 24 h; (b) cPrMgBr,
DCM, -78 °C, 30 min; (c) 1 M NaOH, MeOH, reflux, 4 h; (d) 4 N HCl in 1,4-dioxane, rt, 1 h; (e)
i. chloroacetyl chloride, 1 M NaOH, 1,4-dioxane, 1 h, ii. NaHCO , DMF, 80 °C, 2 h; (f) DIBAL-
H, THF, -78 °C to rt, 2 h; (g) DAST, DCM, 0 °C, 40 min; (h) P S , THF, 70 °C, 4 h; (i) 33 %
2
4
3
2
5
NH (aq), 60 °C, 21 h; (j) NaN , CuI, DMEDA, Na CO , DMSO, 110 °C, 1 h; (k) 5-
3
3
2
3
methoxypyrazine-2-carboxylic acid, DMTMM, MeOH, 0 °C, 6 h.
Racemic compounds (2R*,3R*)-10 and (2R*,3R*)-11 were prepared as shown in Scheme 5.
First, ketones 46a,b were converted to the α-aminonitriles rac-47a,b via a Strecker reaction.
Direct acidic hydrolysis of the nitrile group in rac-47a,b to the carboxylic acids rac-49a,b was
not successful in refluxing HCl (aq) due to rapid decarboxylation under the reaction conditions.
This could be circumvented by the hydrolysis of the nitriles in rac-47a,b to the corresponding
amides rac-48a,b with HBr in acetic acid, followed by basic hydrolysis with NaOH to rac-49a,b.
The standard lactonization procedure involving initial N-acylation of rac-49a,b with chloroacetyl
ACS Paragon Plus Environment

Page 11 of 123
Journal of Medicinal Chemistry
1
2
3
chloride led also to decarboxylation. Fortunately, O-alkylation of rac-49a,b with tert-butyl
chloroacetate to rac-50a,b followed by selective TFA-mediated hydrolysis of the tert-butyl ester
yielded the amino acid derivatives rac-51a,b. Intramolecular cyclization of rac-51a,b with
Mukaiyama reagent provided 1,4-oxazine-2,5-diones rac-52a,b. A similar reaction sequence to
those previously described was used for the transformation of rac-52a,b into the target molecules
rac-(2R*,3R*)-11 and rac-(2R*,3R*)-12. However, since the amidine function in rac-(5R*,6R*)-
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
6a,b was found to be more reactive towards a peptidic coupling than the aminopyridine
fragment using DMTMM, the former needed to be selectively protected with a Boc group giving
rac-(2R*,3R*)-57a,b prior to coupling reaction with the 5-methoxypyrazine-2-carboxylic acid.
Subsequent cleavage of the Boc protecting group in rac-(5R*,6R*)-58a,b provided rac-
(2R*,3R*)-11 and rac-(2R*,3R*)-12, respectively.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Scheme 5. Synthesis of 1,4-oxazines rac-(2R*,3R*)-11 and rac-(2R*,3R*)-12 containing a
a
pyridyl as A-ring.
O
W
Y
W
N
O
W
a
b
c
H 2
N
Y
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
X
Y
H 2
N
N
H 2
N
X
4
4
6a (X=N, Y=CH, W=Br)
6b (X=CH, Y=N, W=Cl)
rac-47a (X=N, Y=CH, W=Br): 54%
rac-47b (X=CH, Y=N, W=Cl): 98%
rac-48a (X=N, Y=CH, W=Br): 61%
rac-48b (X=CH, Y=N, W=Cl): 81%
W
W
W
O
O
O
d
e
O
H O
H 2
O
Y
Y
O
Y
H O
N
X
X
H 2
N
X
H 2
N
O
O
rac-49a (X=N, Y=CH, W=Br): 65% rac-50a (X=N, Y=CH, W=Br)
rac-51a (X=N, Y=CH, W=Br): 29% 2 steps
rac-49b (X=CH, Y=N, W=Cl): 100% rac-50b (X=CH, Y=N, W=Cl): 65% rac-51b (X=CH, Y=N, W=Cl): 95%
O
O
O
F
O
F
f
g, h
i
O
N
H
O
N
H
S
N
H
X
X
X
Y
Y
Y
W
W
W
rac-5R*,6R*-54a
X=N, Y=CH, W=Br): 54%
rac-5R*,6R*-54b
X=CH, Y=N, W=Cl)
rac-52a
(X=N, Y=CH, W=Br): 55%
rac-52b
rac-5R*,6R*-53a
(
(X=N, Y=CH, W=Br): 81%
rac-5R*,6R*-53b (6R/6S 65:35)
(X=CH, Y=N, W=Cl): 69%
(
(X=CH, Y=N, W=Cl): 19%
O
F
O
F
O
F
j
k
l
H 2
N
N
H N
BOC
N
H 2
N
N
X
X
X
Y
Y
Y
H 2
N
H 2N
W
rac-5R*,6R*-55a
X=N, Y=CH, W=Br): 85%
rac-5R*,6R*-55b
(X=CH, Y=N, W=Cl): 45% 2 steps
rac-5R*,6R*-56a
(X=N, Y=CH): 64%
rac-5R*,6R*-56b
(X=CH, Y=N): 72%
rac-5R*,6R*-57a
(X=N, Y=CH): 37%
rac-5R*,6R*-57b
(X=CH, Y=N): 64%
(
O
F
O
F
3
6
m
n
5
2
H N
BOC
N
H 2
N
N
X
X
O
O
Y
Y
N
N
H
H
rac-5R*,6R*-58a
(X=N, Y=CH): 47%
rac-5R*,6R*-58b
rac-2R*,3R*-11
(X=N, Y=CH): 51%
rac-2R*,3R*-12
N
N
N
N
O
O
(X=CH, Y=N): 71%
(X=CH, Y=N): 87%
a
Reagents and conditions: (a) TMSCN, NH Cl, NH /MeOH, 12 °C, 4 d; (b) 33 % HBr/AcOH,
4
3
reflux, 12 h; (c) 1 M NaOH, 65 °C, 16 h; (d) ClCH CO tBu, Cs CO , DMF, rt, 3 h; (e) TFA, rt,
2
2
2
3
15 min; (f) 2-chloro-1-methylpyridinium iodide, DIPEA, DCM, reflux, 4 h; (g) DIBAL-H, THF,
-
8
78 °C to rt, 90 min; (h) DAST, DCM, 0 °C to rt; (i) P S , THF, 70 °C, 3 h; (j) 33 % NH (aq),
2 5 3
0 °C, 2 h; (k) NaN , CuI, DMEDA, Na CO , DMSO, 110 °C, 24 h; (l) Boc O, DCM/ACN 1:1,
3 2 3 2
rt, 22 h; (m) 5-methoxypyrazine-2-carboxylic acid, HATU, DIPEA, DMF, 50 °C, 3 h; (n) TFA,
rt, 15 min.
ACS Paragon Plus Environment

Page 13 of 123
Journal of Medicinal Chemistry
1
2
3
RESULTS AND DISCUSSION
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Our early benzoguanidine series provided us with an understanding of the optimal binding
1
3
arrangement of the amidine motifs at the catalytic aspartates. This series suffered from high
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
basicity, examples having measured pK ’s in range of 11 and 12, which we associated with
a
issues of poor permeation, brain penetration and P-gp efflux. We therefore emphasized the need
to find similar amidine and guanidine binding motifs but with reduced basicity, assessed using
calculated pK values. To implicitly capture elements of protein flexibility proposals were
a
docked into multiple BACE1 crystal structures. At this preliminary stage we were using an
21
entirely in silico approach as previously described. A similar rational design concept but with
2
4
different method of execution has been reported by others. Alongside the previously reported
2
1
piperazinones 3 and 4, the morpholine series 5 was prioritized highly by our approach. The
docking of sample 1,4-oxazine molecules showed them to score particularly well, in fact, they
1
2
were the best of all ideas profiled and docked optimally into the protein structure PDB 2VA7.
2
5
With regards to the basicity, the 1,4-oxazine was predicted to be in an ideal range. For instance,
whilst our previous benzoguanidines had a calculated basic pK of 11.1, an oxazine example
a
such as 5-methyl-5-phenyl-5,6-dihydro-2H-1,4-oxazin-3-amine had a calculated basic pK of 6.4
a
2
5
(
Scheme 6). This apparently placed the 1,4-oxazines in a good range of basicity, being
protonated in the acidic endosome environment in which BACE1 is active and therefore binding
at the catalytic aspartates. Such a pK would be expected to provide a balanced degree of
a
26
protonation at higher physiological pH, beneficial for CNS drugs. However, comparison with
the experimental pK reveals inaccuracy in the calculated value. Deeper analysis of this was
a
performed once more examples were synthesised, and as the later discussion reveals, we may
have been fortunate to have prioritised the oxazines via this in silico approach.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The morpholine derivatives 3R-6a-c were synthesized as early prototypes to set the baseline
for potency and physicochemical properties of this class (Table 1). During the course of our
research these molecules also got reported in an extensive warhead exploration published by
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
19,27
Hoffmann-La Roche and Siena Biotech.
Based on previous SAR derived from the
exploration of the piperazinone series 4, 5-chloropyridine-2-carboxamide and 5-methoxy-2-
pyrazinecarboxamide were chosen as B-rings to target the S3 pocket in the morpholine probe
compounds 3R-6a-c. In addition, A-ring variations without (3R-6a) and with (3R-6b) a fluorine
atom in para-position to the aniline on the A ring (see pharmacophore, Chart 1) were synthesized
to assess its influence on pK and potency. Compounds 3R-6a-c had satisfactory inhibitory
a
activity in the BACE1 enzymatic assay, with IC ’s of 44, 22 and 44 nM respectively. This
50
potency nicely translated into an inhibition of Aβ42 production of 9.1 nM (3R-6a), 4.1 nM (3R-
6
b) and 5.4 nM (3R-6c) in a human neuroblastoma cell line. In vivo reduction in Aβ peptides is
assessed in mouse, hence determination of the cellular activity of 3R-6a-c in a mouse
neuroblastoma cell line was also done and found to correlate well with the values obtained with
the human cell line.
ACS Paragon Plus Environment

Page 15 of 123
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a
Table 1. In vitro profile of morpholines 3R-6a-c.
N
N
O
N
3
R-6a R = H, R =
4 5
R 4
Cl
Cl
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
H 2
N
3
3
R-6b R = F, R =
4 5
H N
R 5
N
O
R-6c R = F, R =
4
5
N
O
3R-6a
44
3R-6b 3R-6c
BACE1 IC (nM)
22
44
5
0
hAβ42 cell IC (nM)
9.1
7.4
0
6
2.9
43
4.1
4.5
0
10
3.1
34
5.4
4.3
1
6
10.1
56
32
37
154
4.2
9.0
5
0
mAβ42 cell IC (nM)
5
0
hLM (% met. @ 15 min)
mLM (% met. @ 15 min)
fu (brain, r, %)
fu (plasma, h, %)
fu, (plasma, m, %)
Papp A>B (nm/s)
28
n. d.
21
38
Papp A>B +elacridar (nm/s) n. d.
231
6.0
9.2
Papp ratio
pK
n. d.
9.6
a
a
See SI for assay details.
Further profiling of 3R-6a-c revealed that they had good metabolic stability in human (hLM)
and mouse (mLM) liver microsomes, low binding to human and mouse plasma proteins and
moderate non-specific binding to rat brain homogenate (f in plasma or brain). The experimental
u
pK values were determined for 3R-6a-c and were found to be higher than predicted (3R-6a 9.6;
a
3
R-6b 9.2; 3R-6c 9.0). The effect of a fluorine substituent in the A-ring was modest as seen for
compound 3R-6b which showed only 0.4 Log unit (2.5 fold) reduction in pK when compared
a
with the non-fluorine substituted analogue 3R-6a. In line with previous observations that high
20
pK is associated with increased P-gp efflux, high ratios of apparent permeability (P ) in
a
app
presence or absence of the P-gp inhibitor elacridar were measured for 3R-6b and 3R-6c: 6.0 and
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4
.2 respectively. Subcutaneous (s. c.) administration of a 30 mg/kg dose of 3R-6b to mouse
(Table 2, A) showed modest brain levels (315 ng/g) at 2 h compared to plasma concentration
(2397 ng/mL), resulting in a low brain/plasma ratio (K ) of 0.13. These low concentrations in the
p
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
brain translated into a marginal in vivo reduction of Aβ peptides in the brain of wild type (WT)
CF-1 mice (Table 2, B) 2 h after dosing. When the same experiment was repeated using P-gp
deficient CF-1 mice a more favorable brain/plasma ratio (K ) of 1.1 was observed, which
p
resulted into a much more pronounced reduction of Aβ peptides in brain compared to the one
observed in WT animals. This experiment confirmed the detrimental role of P-gp efflux in the
brain penetration of 3R-6b.
Table 2. Brain levels (A) and brain Aβ reduction (B) of 3R-6b in WT and P-gp deficient CF-1
a
mice.
A
3R-6b levels at 2 h
3
500
000
500
000
500
000
K = 0.13
K = 1.1
p
p
3
2
2
1
1
Plasma ng/mL
Brain ng/g
5
00
0
WT CF1
P-gp def. CF1
ACS Paragon Plus Environment

Page 17 of 123
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
B
3R-6b Aβ (% versus control) at 2 h
1
1
1
40
20
00
Aβ total
Aβ40
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
8
6
4
2
0
0
0
0
0
Aβ42
Vehicle WT Vehicle P-gp def.
6b WT
6b P-gp def.
a
30 mg/kg, s. c. (n=8); 3R-6b was formulated with 20 % SBEβCD at pH 3.5.
Whilst lower than our old benzoguanidine series the measured pK values were higher than
a
1
3
expected from the initial calculations mentioned above. This prompted a deeper inspection of
the accuracy of the calculated pK ’s on these unusual amidine containing ring systems. A
a
selection of 42 compounds containing a cyclic amidine substructure from our compound
collection was subjected to experimental pK measurement (data in SI). The selection set
a
encompassed multiple heterocyclic chemotypes and covering approximately a 10 log unit range
in calculated pK . Molecules with multiple ionizable centers were in general avoided. The pK ’s
a
a
were calculated with a variety of methods, and the correlation with experiment is shown in SI. In
28
contrast to previous studies comparing performance across global datasets, ADMET predictor
2
2
performed better than ACD for these amidine ring systems (R = 0.88 compared to R = 0.75 for
best ACD case). This was especially true in the relevant pK range of 6 to 10. These results
a
prompted us to reassess our 1,4-oxazine series with calculations on specific prototypes (Scheme
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
29
6
) confirming improved performance when using AP. This reiterates the need to calibrate the
performance of pK calculations for novel chemical systems.
a
O
N
F
O
N
CF 3
O
N
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
F
F
H 2
N
H 2
N
H 2
N
H 2
N
H 2N
pKa Experiment
9.8
8.1
3.8
9.7
7.3
8.0
4.6
9.9
6.7
pKa calculated ACDa
6
1
.4
pKa calculated ACDb
0.4
pKa calculated with AP 9.6
Legend
X
X
a
b
Tautomer
Tautomer
X
X
H 2
N
N
H N
N
H
Scheme 6. Comparison of the calculated pK to experiment for prototype 1,4-oxazines using
a
ACD v2014 and AP.
The crystal structure of 3R-6a with BACE1 was solved at 2.6 Å resolution (PDB code 5CLM,
Figure 1). The result confirmed a binding mode similar to that seen for other bisarylamide
17
substituted amidine BACE1 inhibitors. Namely, the amidine group makes the key and strong
network of interactions with the catalytic aspartate dyad (Asp32 and Asp228) involving several
hydrogen bonds and a charge-charge interaction between the protonated amidine and anionic
acid groups. The A-ring (phenyl) occupies the S1 pocket, and the B-ring (5-chloro-2-pyridyl)
enters the S3 pocket. The S1 pocket is formed by amino acids including Phe108 whereas the S3
is bordered by the presence of Thr232 and amino acids from the 10s loop such as Gly11, Gln12
ACS Paragon Plus Environment

Page 19 of 123
Journal of Medicinal Chemistry
1
2
3
and Gly13 (10s loop, blue in Figure 1). Para substituents on the B-ring approach the salt bridge
formed by Arg307 and Glu339. Between the A- and B-ring, the amide N-H forms an H-bond to
the backbone carbonyl of Gly230. This bisarylamide has become a widely used group in
substituted amidine BACE1 inhibitors, likely because of the optimal shape complementarity with
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
5
the S1 and S3 pockets.
Based on the finding that P-gp efflux was influenced by the basicity of the amidine containing
warhead, several options for chemical modification of the 1,4-oxazine ring by the introduction of
electron withdrawing substituents were possible to reduce its high pK . Close inspection of the
a
3
1
,4-oxazine cycle reveals that the two possible sp carbon substitution points have different
environments. The carbon which is between the oxygen and the amidine (C-6) is more exposed
towards solvent. On the other hand, the carbon between the oxygen and the quaternary carbon,
(C-2), would direct its substituents towards the active site flap (green in Figure 1). The flap is a
well-known feature of the BACE1 binding site, it is flexible and more open at lower pH,
3
0
correlating with higher enzyme activity. It is known to adopt different conformations
dependent on the bound inhibitor. Molecules are reported which interact with amino acids such
1
3
30
as Tyr71, (Figure 1) and the adjacent Thr72 for example. Therefore, given that the amidine
warhead was satisfying the interactions at the catalytic aspartates, we set out to target alternative
contacts with the flap. Specifically, C-2 position provided an ideal vector to interact with the
aromatic ring of Tyr71, whilst steric factors would limit the size of the proposed substituents.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Crystal structure of 3R-6a in BACE1 (1-454). Amino acids as labelled. Active site
flap highlighted in green and 10s loop in blue.
An overview of the different modifications explored at C-2 position of the 1,4-oxazine
warhead is shown in Chart 2. A fluorine (I), CF (II) and a combination of both groups (III)
3
were selected as optimal C-2 substituents to further modulate pK of 3R-6c whilst interacting
a
with the active site flap.
F
F
F
F
F
R1
R2
O
F
O
O
O
N
F
2
F
H 2
N
F
I
II
III
Chart 2. Scope of exploration at C-2 position to modulate pKa.
ACS Paragon Plus Environment

Page 21 of 123
Journal of Medicinal Chemistry
1
2
3
4
5
6
BACE1 enzymatic and cellular activity, experimental pK values, P ratios and Log D values
a
app
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
for 3R-6c and fluorinated 1,4-oxazines (2S,3R)-7a, (2R,3R)-7a, (2S,3R)-8, (2R,3R)-8, (2R,3R)-9
31
and (2S,3R)-9 are provided in Table 3. Interestingly, the 2R-fluoro substituent in (2R,3R)-7a
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
was found to be highly effective in reducing the basicity of the amidine: a pK of 7.8 was
a
measured, which is a decrease by 1.2 log units when compared with the C-2 non-substituted
oxazine 3R-6c. This was aligned with our expectations based on the AP calculated values for this
substitution (Scheme 6). The reduction in pK was not detrimental for the primary enzymatic and
a
cellular activity, with the discrepancy between both assays decreased when compared to 3R-6c.
This is a general trend observed for our amidine-based BACE1 inhibitors: the lower the pK , the
a
smaller the discrepancy becomes between enzymatic and cellular activity. A possible explanation
is that highly basic inhibitors accumulate in the acidic endosomes where BACE1 is most active.
As such they display apparently higher activity in cellular than biochemical assays, whereas
lower pK results in cellular activity in closer accordance with the one in the biochemical assay
a
due to a more uniform distribution of the compound in cellular compartments. Furthermore,
compound (2R,3R)-7a showed good permeability in a LLC-MDR cell line (Papp 201 nm/s with
elacridar) and a low Papp ratio of 1.1 providing additional support to the hypothesis that lowering
the amidine pK would have beneficial effects in P ratios. Interestingly, while in the same pKa
a
app
range, the diastereomeric (2S,3R)-7a had a much reduced BACE1 IC of 102 nM, which
50
translated in an equally reduced cellular activity of 93 nM. We hypothesized that this potency
difference could be attributed to the anomeric effect, which favors the 2-fluoro substituent in the
axial position. QM calculations indeed confirmed this effect stabilizes the bioactive
conformation for (2R,3R)-7a (see SI). However in the case of (2S,3R)-7a the conformation with
the 2-fluorine in axial position places the phenyl in the unfavorable equatorial position for
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
32
binding to BACE1. An additional CF group at C-2 position such as 2S-C(F)CF disubstitution
3
3
in (2S,3R)-8 reduces pK by another log unit ((2S,3R)-8, pK = 6.8), while maintaining a potency
a
a
of 20 nM in the BACE1 enzymatic assay. Unfortunately, the 2S-C(F)CF disubstitution had a
3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
detrimental effect on the cellular activity and compound (2S,3R)-8 was ~ 6-fold less potent than
3R-6c. Interestingly, the diastereoisomeric 2R-C(F)CF 1,4-oxazine (2R,3R)-8 displayed poor
3
inhibitory activity in both enzymatic and cellular assays. This drop in potency could be
rationalized after a close inspection of the crystal structure (Figure 1) of 3R-6c, which suggests
that the CF group in (2R,3R)-8 would be directed into the face of Tyr71, producing an
3
unfavorable interaction between the negative electrostatic properties of both the CF and the π
3
cloud of Tyr71. A similar but less pronounced trend in enzymatic activity was also observed for
the 2-C(H)CF diastereoisomeric pair (2R,3R)-9 (IC = 29 nM) and (2S,3R)-9 (IC = 110 nM),
3
50
50
although in this case it did not translate in a difference in cellular activity: (2R,3R)-9 IC = 13
5
0
nM vs (2S,3R)-9 IC = 14 nM. Compound (2R,3R)-9, with a pK of 7.8, illustrated again the
50
a
ability of the electron withdrawing ability of the CF group to reduce the amidine pK and its
3
a
beneficial contribution to a lower Papp ratio of 1.2. Similarly, while the permeability for (2S,3R)-
8 was not determined, close analogues with different B-rings showed low Papp ratios in the 1.2
(
5-chloropyridine-2-carboxamide) to 1.6 (4-cyanopyridyl-2-carboxamide) range.
ACS Paragon Plus Environment

Page 23 of 123
Journal of Medicinal Chemistry
1
2
3
4
5
Table 3. Comparison of BACE1 enzymatic and cellular activities, pK ’s and Papp ratios for
a
a,b
different C-2 substituted 1,4-oxazine warheads.
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
R1
O
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
R2
H 2
N
N
F
H N
O
N
N
O
3
R-6c
(2S,3R)- (2R,3R) (2S,3R)-8 (2R,3R)- (2R,3R)- (2S,3R)-
7
a
-7a
8
9
9
F
F
F
F
F
F
F
F
F
F
F
F
F
F
R1
R2
H
O
O
O
F
O
F
O
O
O
O
H
BACE1 IC50
(nM)
44
102
93
12
20
34
316
324
29
13
110
14
hAβ42 cell IC50 5.4
nM)
Papp A>B (nm/s) 37
9.1
(
n. d.
n. d.
176
201
n. d.
n. d.
n. d.
n. d.
137
160
n. d.
n. d.
Papp A>B 154
+
elacridar
(nm/s)
Papp ratio
4.2
n. d.
9.0
n. d.
n. d.
7.6
1.1
2.04
7.8
n. d.
3.43
6.8
n. d.
n. d.
6.9
1.2
2.52
7.8
n. d.
n. d.
8.0
LogD (pH 7.4)
pK
a
a
b
See SI for assay details. n. d.: not determined
LogD was also measured for the most promising leads (2R,3R)-7a, (2S,3R)-8 and (2R,3R)-9,
with (2R,3R)-7a showing the lowest value (LogD = 2.04). From the data obtained compound
(2R,3R)-7a showed the most balanced profile with good enzymatic and cellular activity, low pKa
and LogD values, and good permeability and Papp ratio. Since sugars with anomeric fluorines are
3
3
known to be hydrolytically unstable, the chemical stability of (2R,3R)-7a was assessed by
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
LCMS-based experiments under a variety of conditions ranging from acidic to basic media and
different temperatures after seven days (see SI). In these experiments, (2R,3R)-7a was found to
be sufficiently stable at room temperature up to one week in buffer at pH 7.4 (92% remaining)
and DMSO (100% remaining) to warrant further exploration.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Analysis of the crystal structure of 3R-6a suggested that space is available around the methyl
substituent at the quaternary center in the S2’ pocket, hence a limited synthetic effort was
devoted to better fill this pocket. Compound (2R,3R)-10 bearing a C-2 fluoro substituent and a
cyclopropyl group at position C-3 was selected as a potential target. BACE1 enzymatic and
cellular activity, metabolic stability data and pK values are given for compounds (2R,3R)-10 and
a
(
2R,3R)-7a in Table 4. An approximately 5-fold decrease in enzymatic activity was measured for
2R,3R)-10 (IC = 66 nM) compared to (2R,3R)-7a (IC = 12 nM), however, a much more
(
5
0
50
significant 66-fold drop in cellular activity was also observed (2R,3R)-10: IC = 603 nM vs
50
(
2R,3R)-7a: IC = 9.1 nM). Compared to (2R,3R)-7a, (2R,3R)-10 showed increased metabolic
50
stability in hLM, but a slightly higher turnover in mLM. Finally, no difference in pK between
a
(2R,3R)-7a and (2R,3R)-10 was measured. We hence decided to maintain the methyl group at the
quaternary center for further optimization of (2R,3R)-7a.
ACS Paragon Plus Environment

Page 25 of 123
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a
Table 4. Modification on the quaternary methyl group.
O
F
R 3
F
H 2
N
N
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
H N
O
N
N
O
(
2R,3R)-7a
(2R,3R)-10
O
N
F
O
F
O
F
R 3
C H 3
H 2
N
H 2
N
N
H 2
N
N
BACE1 IC nM
12
66
5
0
hAβ42 cell IC nM
9.1
7.8
603
7.8
5
0
pK
a
a
See Experimental Section for assay details.
To generate additional SAR on potency and pK variation of the A-ring was studied next.
a
Analogues rac-(2R*,3R*)-11 and rac-(2R*,3R*)-12 in which the A-phenyl ring is replaced by a
2- or 4-pyridyl ring were synthesized. BACE1 enzymatic and cellular activity, metabolic stability
data and pK values are given for compounds (2R,3R)-7a, rac-(2R*,3R*)-11, rac-(2R*,3R*)-12
a
in Table 5. Interestingly the reduced basicity of the amidines was maintained when replacing the
phenyl A-ring in (2R,3R)-7a by a 2-pyridyl as in rac-(2R*,3R*)-11 (pK = 7.8) and the
a
regioisomeric 4-pyridyl analogue rac-(2R*,3R*)-12 (pK = 7.6). In addition, the pK value of the
a
a
pyridyl nitrogen in rac-(2R*,3R*)-11 and rac-(2R*,3R*)-12 was measured to be 2.6 and 2.8
respectively and hence no protonation of this ring is expected to occur under physiological
conditions. Unfortunately, while the pK of rac-(2R*,3R*)-11 and rac-(2R*,3R*)-12 was in the
a
desired range, the presence of the pyridyl ring was found to be detrimental for BACE1 activity,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
with the 2-pyridyl isomer rac-(2R*,3R*)-11 significantly less potent than (2R,3R)-7a and rac-
(2R*,3R*)-12 only active in the µM range.
a
Table 5. Overview of A-ring modifications.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
N
F
H 2
N
X
Y
H N
O
N
N
O
(
2R,3R)-7a rac-(2R*,3R*)-11
rac-(2R*,3R*)-12
F
Y
N
X
N
BACE1 IC nM
12
74
3467
5
0
hAβ42 cell IC nM
9.1
7.8
257
1445
5
0
pK
7.8; 2.6
7.6; 2.8
a
a
See Experimental Section for assay details.
Finally, a limited SAR exploration of the B-ring in (2R,3R)-7a using different pyridine-2-yl
carboxylic acids was performed following a similar synthetic route. A set of representative
compounds (2R,3R)-7a-d along with BACE1 primary activity, data from broader profiling in
ADME-tox assays and physicochemical properties, are given in Table 6. Although in general
excellent enzymatic and cellular activity data were obtained for all compounds having a pyridyl
B-ring ((2R,3R)-7b-d) subtle differences were observed regarding their in vitro ADME
properties. For instance, while (2R,3R)-7b and (2R,3R)-7d retained excellent metabolic stability
ACS Paragon Plus Environment

Page 27 of 123
Journal of Medicinal Chemistry
1
2
3
across tested species, the trisubstituted pyridine (2R,3R)-7c was notably unstable in mLM. All
compounds showed a low potential to inhibit CYP450 isoforms 1A2, 3A4, 2C9 and 2C19
compared to their IC ’s. The strongest inhibition of the 2D6 isoform was observed for
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
5
0
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
compound (2R,3R)-7a with an IC of 600 nM, still leading to a 50-fold margin when compared
5
0
with the enzymatic BACE1 activity (IC = 12 nM). In addition, all compounds displayed high
5
0
free fractions in plasma across species (5.0-60 % free) and a reasonable free fraction in rat brain
tissue (1.0-4.4 % free). Cardiovascular safety of the compounds was first assessed using a hERG
channel patch-clamp assay indicating that the B-ring has a strong effect on hERG inhibition. For
instance, while the 5-methoxypyrazine (2R,3R)-7a showed a relatively weak inhibition of the
hERG channel at 3 µM (33%), the pyridyl derivative (2R,3R)-7b showed a much stronger hERG
inhibition (78%) at the same concentration. This strong interaction could be mitigated by
installing a methyl group on the pyridine as in (2R,3R)-7c (52 % inhibition). Finally the 4-
cyanopyridyl derivative (2R,3R)-7d showed 56 % hERG inhibition at 3 µM. As could be
expected for this 2R-F substituted warhead, Papp ratios for analogues (2R,3R)-7b-d were
generally acceptable, with (2R,3R)-7d being a notable exception with an Papp ratio of 3.6.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 123
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a, b
Table 6. Overview of B-ring modifications.
O
N
F
F
H 2
N
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
H N
O
N
R
X
(
52,63)-7a (2R,3R)-7b (2R,3R)-7c (2R,3R)-7d
N
N
N
N
N
R
N
X
O
Cl
Cl
NC
BACE1 IC nM
12
7.4
5.2
0
17
7
6.9
3.3
16
98
21
7.6
8.1
0
5
0
hAβ42 cell IC nM
9.1
31
3
70
2D6 (0.6)
4.4
50
hLM (% met. @ 15 min)
mLM (% met. @ 15 min)
dLM (% met. @ 15 min)
CYP450 inh. <10 µM
fu (brain, r, %)
2
0
none
1.9
18
none
1.0
11
2D6 (4.3)
3.9
41
30
42
56
51
fu (plasma, h, %)
60
fu (plasma, m, %)
28
14
n. d.
78
5.0
n. d.
52
fu (plasma, d, %)
34
hERG PX (% inh. @ 3 µM)
Papp A>B (nm/s)
33
176
85
93
Papp A>B +elacridar (nm/s)
Papp ratio
201
1.1
7.8
167
2.0
7.9
142
1.5
7.4
183
3.6
7.9
pK
a
a
b
See Experimental Section for assay details. n. d.: not determined
From the exploration of the B-ring, compounds (2R,3R)-7a and (2R,3R)-7d showed the best
balance between potency, hERG liability, and in vitro P-gp efflux liability, and therefore were
progressed to in vivo PK/PD evaluation. The results obtained are summarized in Table 7. In these
experiments, compounds were dosed orally (p. o.) to male Swiss mice at 10 mg/kg. Plasma
ACS Paragon Plus Environment

Page 29 of 123
Journal of Medicinal Chemistry
1
2
3
levels are shown for three time points. For each time point, one hemisphere of the brain was used
to conduct bioanalysis, and the other hemisphere for AlphaLisa quantification of Aβ. In line with
the measured Papp ratios of 1.1 and 3.6, a higher brain uptake relative to plasma was observed for
compound (2R,3R)-7a (K = 0.65) compared to (2R,3R)-7d (K = 0.33). Gratifyingly, a robust
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
p
p
Aβ42 reduction of 62 % for (2R,3R)-7a and 57 % for (2R,3R)-7d was observed in brain
homogenate at 4 h after administration, corresponding to brain levels of 195 ng/g for (2R,3R)-7a
and 98 ng/g for (2R,3R)-7d. This substantial reduction in Aβ42 is in line with their high in vitro
potency (resp. 12 and 7.6 nM) and good free fraction in brain. Moreover, this significant
reduction of Aβ42 in brain is maintained up to 7-8 h.
Table 7. Mouse pharmacokinetic profile of (2R,3R)-7a and (2R,3R)-7d with associated Aβ
a
levels in brain.
(2R,3R)-7a levels
(2R,3R)-7a Aβ (% versus control)
7
6
5
4
00
00
00
00
100
9
0
b
K = 0.65
p
80
70
60
2
4
8
h
h
h
2 h
4 h
8 h
5
4
3
2
1
0
0
0
0
0
0
300
2
1
00
00
0
Plasma ng/mL
Brain ng/g
Abeta42
Abeta Total
ACS Paragon Plus Environment