Aminomethyl-Derived Beta Secretase (BACE1) Inhibitors: Engaging Gly230 without an Anilide Functionality

Article abstract of DOI:10.1021/acs.jmedchem.6b01451

A growing subset of β-secretase (BACE1) inhibitors for the treatment of Alzheimer’s disease (AD) utilizes an anilide chemotype that engages a key residue (Gly230) in the BACE1 binding site. Although the anilide moiety affords excellent potency, it simultaneously introduces a third hydrogen bond donor that limits brain availability and provides a potential metabolic site leading to the formation of an aniline, a structural motif of prospective safety concern. We report herein an alternative aminomethyl linker that delivers similar potency and improved brain penetration relative to the amide moiety. Optimization of this series identified analogues with an excellent balance of ADME properties and potency; however, potential drug-drug interactions (DDI) were predicted based on CYP 2D6 affinities. Generation and analysis of key BACE1 and CYP 2D6 crystal structures identified strategies to obviate the DDI liability, leading to compound 16, which exhibits robust in vivo efficacy as a BACE1 inhibitor.

Full text of DOI:10.1021/acs.jmedchem.6b01451

Subscriber access provided by UNIVERSITY OF SOUTH CAROLINA LIBRARIES
Article
Aminomethyl-Derived Beta Secretase (BACE1) Inhibitors:
Engaging Gly230 without an Anilide Functionality
Christopher R. Butler, Kevin Ogilvie, Luis A Martinez-Alsina, Gabriela Barreiro, Elizabeth M. Beck, Charles
E Nolan, Kevin Atchison, Eric S Benvenuti, Leanne M Buzon, Shawn D Doran, Cathleen Gonzales,
Christopher J. Helal, Xinjun Hou, Mei-Hui Hsu, Eric F. Johnson, Kimberly Lapham, Lorraine F Lanyon,
Kevin Parris, Brian T. O'Neill, David R. Riddell, Ashley Robshaw, Felix F. Vajdos, and Michael A. Brodney
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b01451 • Publication Date (Web): 06 Dec 2016
Downloaded from http://pubs.acs.org on December 7, 2016
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 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Aminomethyl-Derived Beta Secretase (BACE1)
Inhibitors: Engaging Gly230 without an Anilide
Functionality
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
a
c
c,
a
a
Christopher R. Butler* , Kevin Ogilvie , Luis Martinez-Alsina Gabriela Barreiro , Elizabeth M. Beck ,
b
b
e
c
e
b
Charles E. Nolan , Kevin Atchison , Eric Benvenuti , Leanne Buzon , Shawn Doran , Cathleen Gonzales ,
c
a
f
f
e
Christopher J. Helal , Xinjun Hou , Mei-Hui Hsu , Eric F. Johnson , Kimberly Lapham , Lorraine
e
d
c
b
b
d
Lanyon , Kevin Parris , Brian T. O’Neill , David Riddell , Ashley Robshaw , Felix Vajdos , and Michael
a
A. Brodney
a
b
Neuroscience & Pain Medicinal Chemistry, Neuroscience & Pain Research Unit, Pfizer Worldwide
c
Research and Development, 610 Main St, Cambridge, MA, 02139, United States; Neuroscience & Pain
d
e
Medicinal Chemistry, Center of Chemistry Innovation and Excellence, Pharmacokinetics, Dynamics,
and Metabolism, Pfizer Worldwide Research and Development, 445 Eastern Point Road, Groton, CT,
f
0
6340, United States, Molecular and Experimental Medicine, The Scripps Research Institute, 10550
Torrey Pines Road, La Jolla, California, 92024, United States
RECEIVED DATE (to be automatically inserted after your manuscript is accepted if required
ABSTRACT
A growing subset of β-secretase (BACE1) inhibitors for the treatment of Alzheimer’s
disease (AD) utilizes an anilide chemotype that engages a key residue (Gly230) in the BACE1
binding site. Although the anilide moiety affords excellent potency, it simultaneously introduces
a third hydrogen bond donor that limits brain availability and provides a potential metabolic site
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 70
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
leading to the formation of an aniline, a structural motif of prospective safety concern. We
report herein an alternative aminomethyl linker that delivers similar potency and improved brain
penetration relative to the amide moiety. Optimization of this series identified analogs with an
excellent balance of ADME properties and potency; however, potential drug-drug interactions
(DDI) were predicted based on CYP 2D6 affinities. Generation and analysis of key BACE1 and
CYP 2D6 crystal structures identified strategies to obviate the DDI liability, leading to
compound 16, which exhibits robust in vivo efficacy as a BACE1 inhibitor.
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
Introduction
Alzheimer’s disease (AD), a neurological disorder that imparts a slow progression of
cognitive decline, dementia, and ultimately death, has yet to yield to a significant enhancement
in treatment or prevention. Disease progression is marked by the deposition of amyloid β (Aβ)-
derived plaques in the hippocampal and cortical regions of the brain. The amyloid hypothesis
proposes that increased Aβ production or its decreased clearance is responsible for the molecular
cascade that eventually leads to neurodegeneration and AD.1,2 Aβ production is initiated by the
proteolytic cleavage of amyloid precursor protein (APP) by β-site APP cleaving enzyme
3
(
BACE1) within the endosome, to afford a soluble N-terminal ectodomain of APP (sAPPβ) and
the C-terminal fragment C99.4 The membrane-bound C99 is then cleaved by γ-secretase to
release Aβ, including Aβx-40 and Aβx-42 isoforms.5 Recently, an APP “loss of function”
mutation, with protective effects against AD, has been reported to be cleaved more slowly by
BACE1.6 Modulation of the Aβ cascade via safe and effective inhibition of BACE1 has
remained a target of great interest for a number of years.7
Considering the chronic dosing regimen required for a successful AD treatment, an
exquisitely selective and safe profile for a BACE1 inhibitor is paramount. Of particular concern
ACS Paragon Plus Environment

Page 3 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
8
for this target is inhibition of hERG, as well as related aspartyl proteases including Cathepsin D
9
(
CatD), which has confounded early generations of BACE1 inhibitors. The hERG-mediated
cardiovascular liability is traditionally avoided by eliminating basic amine functionality and
lowering lipophilicity.10 This is challenging for BACE1 as the active site is most efficiently
engaged through utilization of such an amine, thus requiring alternate mitigation strategies.
Additionally, the binding sites of CatD and BACE1 have high sequence similarity, and therefore
differentiation requires exploitation of subtle architectural variances in order to maintain affinity
for BACE1 while avoiding CatD inhibition. Compounds that fail to achieve sufficient selectivity
over CatD carry a liability for ocular toxicity, due to the resulting accumulation of fluorescent
material in the retinal pigment epithelium (RPE) layer.9
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
The physiological relevance of BACE2 has emerged in recent years, first as an enzyme
involved in pigmentation processing, specifically acting on PMEL17 in the periphery.11
Improper functioning of BACE2 is believed to result in hypopigmentation.12 BACE2 is also
expressed in the pancreas and plays a role in glucose homeostasis. To our knowledge, there are
limited examples of BACE1 inhibitors possessing significant selectivity over BACE2.
Compounds that lack this selectivity window and exhibit impaired access to the brain will
therefore inherently suffer from significant inhibition of BACE2. In summary, agents developed
for chronic BACE1 inhibition should be designed to minimize activity against related proteases
such as CatD and BACE2.
The amidine-containing BACE1 inhibitors, reported by a number of groups, provide a
suitable scaffold to systematically address the CatD and hERG liabilities.13 A number of these
inhibitors, such as MK-8931 (1), have recently entered clinical studies; two of them are shown
in Figure 1.14 A common construct within this class is an amide moiety connecting two aromatic
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 70
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
rings that ultimately occupy the S1/S3 pockets when bound in the BACE1 active site. The
incorporation of this moiety generally confers potent inhibition of BACE1 in addition to
exquisite selectivity over CatD. Unfortunately, these merits are generally offset by increased P-
gp-mediated efflux, resulting in decreased brain penetration. There is a correlation between the
presence of a third hydrogen bond donor (HBD) and an increased likelihood of efflux transporter
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
15
liabilities. Poor brain penetration inherently increases the body burden required to achieve the
desired brain concentrations, and thus further exacerbates any issues arising from less than
exquisite aspartyl protease selectivity. Moreover, recent reports have shown that there are
relevant peripheral substrates for BACE1 in addition to the targeted central APP processing.16
Figure 1. Selected literature BACE1 inhibitors
Inhibitors bearing the P1/P3 amide motif not only exhibit higher efflux transporter
liability, but also contain a metabolic soft spot associated with amidase activity, which in this
case would reveal anilines upon amide cleavage.17 In addition to this potential metabolic
liability, anilines are themselves a structural alert, known to be a culprit for downstream toxicity
associated either with oxidation of the electron-rich aryl ring and subsequent trapping with
ambient nucleophiles, or oxidation of the nitrogen itself to the N-oxide.18
Design Criteria.
ACS Paragon Plus Environment

Page 5 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Despite a significant and sustained effort, the identification of a potent, safe, selective
BACE1 inhibitor with a balanced ADME profile, including good brain penetration, remains
challenging. Our design criteria were therefore to identify an orally efficacious BACE1 inhibitor
that: a) demonstrates excellent selectivity (>100x) over hERG, CatD and the related aspartyl
proteases; and b) maintains good brain penetration (Cb,u/Cp,u > 0.5 as measured in rodents)
without the use of an anilide functionality.
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
We therefore sought to identify an amide replacement that provided the opportunity for
similar efficiency gains while avoiding the potential for toxicity imparted by the buried aniline
moiety. The recent BACE1 crystal structure of 2, published by Roche, provides a structural
rationale for the high inhibitory efficiency of P1/P3 amides (Figure 2).19 The amide N-H is
positioned at a near-ideal distance and angle to engage the backbone carbonyl of the Gly230
residue in a hydrogen bonding interaction. Additionally, the coplanar conformation of the rings
occupying the S1/S3 pockets serves to efficiently fill the shallow lipophilic binding site.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 70
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
Asp-228
Tyr-71
flap
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
Gly-230
S1
S3
1
0s loop
Figure 2. Interaction between Gly230 and 2 (PDB: 3ZMG, add 61 to residue numbers to match
ZMG’s number)
3
Thus, a suitable amide replacement would need to fulfill these two structural
requirements: effective engagement of Gly230 and reasonable concomitant occupation of the
S1/S3 subpockets. Simple reversal of the amide removes the buried aniline while retaining the
hydrogen bond donor, but rotation around the amide bond is then needed to recapitulate the
optimum hydrogen bond angle, resulting in significant clashes with the edge of the S3 subpocket.
In contrast, a homologated aminomethyl linker (Figure 3) addresses both of the aforementioned
requirements. Among the salient features of this aminomethyl moiety, one key departure from
3
the amide-derived analogs is its non-planarity. The sp nature of the benzylic carbon results in
the two-atom unit preferring a nearly orthogonal orientation of the C-N bond relative to the
ACS Paragon Plus Environment

Page 7 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
fluorophenyl P1 ring, facilitating an optimal geometry for hydrogen bonding to Gly230.
Optimization of the amine substituent in this alternative vector therefore becomes crucial, to
balance the interplay of requisite hydrogen-bond donating ability of the amine with the resultant
ADME characteristics of these compounds.
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
Tyr-71
Asp-228
flap
Gly-230
S3
S1
1
0s loop
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 70
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
Figure 3. Strategy to mitigate challenges associated with amide pharmacophore of literature
BACE inhibitors. Modeled alignment of aminomethyl design (orange) with compound 2
(
yellow).
Results and Discussion
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
As it was not clear what structural attributes would be required to fill the S3 pocket in
this new series, parallel (library) synthesis was employed to broadly evaluate structure-activity
relationships for BACE1 inhibition. Compounds 6a-u were prepared by the three-step protocol
2
0
illustrated in Scheme 1. Formylation of the previously described bromide 3 provided aldehyde
4, which could be converted to the corresponding protected amidines 5 via a reductive amination
with the requisite amines. Removal of the amidine protecting group using standard conditions
provided the analogs of interest.
Scheme 1a
o
Reagents and conditions: (a) MeLi, Et
2
O, n-BuLi, then DMF, -78 C, 92%; (b) amine, Na(OAc)
3
BH,
DCE, rt, 22-97%; (c) HCl, dioxane, rt, 19-90%; (d) benzoic anhydride, TEA, THF/MeOH (2:1), rt, 86%;
e) N ·H O, EtOH, rt, 31%. P = Boc or benzoyl
(
H
2 4
2
Analogs were evaluated in a panel of BACE1, BACE2, and CatD enzymatic assays (cell-
free assay, CFA), as well as a whole-cell assay (WCA) reporting changes in soluble APP
sAPPβ) protein concentrations, indicative of APP processing by BACE1. The data from this
(
assay panel, as well as key ADME parameters, for 6a–6k are provided in Table 1. The direct
replacement of the amide linker with an aminomethylene spacer, as exemplified by 6a, yields
ACS Paragon Plus Environment

Page 9 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
weak activity at BACE1 (CFA IC50 = 73.5 µM), in stark contrast to the low nanomolar potency
observed for many of the corresponding amides reported in the literature. In addition, the
significant ADME challenges for this series are well illustrated by 6a, which exhibits a high
MDR efflux ratio (Er) and significant clearance in human liver microsomes (HLM). Analogs
6b-d, which contain small alkyl substituents, show a modest improvement in BACE1 CFA
potency, in concert with a dramatic increase in WCA potency (~1000x shifted relative to the
CFA) and excellent selectivity over CatD. The improvement is most marked when considering
ligand and lipophilic efficiencies (LE, LipE) for these low molecular weight, polar, dibasic
amine analogs. These compounds have significantly reduced clearance relative to 6a, as
measured by HLM, although they still exhibit significant P-gp transporter liability. Tertiary
amines 6d-e are significantly less active in the BACE1 CFA but still exhibit good potency in the
WCA, albeit an overall decrease in LipE, relative to the secondary amines. Ether-containing
substituents are also well tolerated (6e,f), showing similar potency in the WCA, low microsomal
clearance but with modest to high efflux ratios. Introduction of a branched methyl group (6c vs
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
6
b and 6h vs 6f) offers an enhancement in potency (WCA) and a modest improvement in
selectivity over BACE2. Tying this branching back into a 1,1,1-bicyclopentane (6i) is tolerated
from a potency perspective, but negatively impacts the clearance, potentially due to the increased
overall lipophilicity.
Reduction in the pK
a
of the benzylic amine center (“pK 2”) was attractive due to the
a
potential to simultaneously impact P-gp efflux and hERG liabilities. Indeed, addition of an
electron-withdrawing trifluoromethyl group onto the amine significantly improves both potency
and ADME balance. Although the tertiary amine (6j) is much less potent in the WCA, the
secondary amine (6k) retains good WCA potency and comparable LipE to the more basic
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 70
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
analogs, and exhibits enhanced BACE1 CFA potency and improved selectivity over BACE2
(
5.8x, vs 3.2x for 6f). Although both CF
efflux ratios (1.7 and 1.2), the secondary amine 6k shows greater metabolic stability (HLM CLint
12.0 mL/min/kg) than the methylated version 6j (HLM = 29.4 mL/min/kg)
3
analogs have significantly improved MDR-based
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
=
Table 1. In Vitro Data for 6a-k
Cpd
6a
R =
BACE1
CFA
BACE2 CatD
BACE1
MDR LogD
HLMf
(mL/m LE/LipE
in/
kg)
WCA
e
CFA
IC50
(µM)
CFA
IC50
(µM)
WCA IC50 Er
d
IC50
(µM)
(µM)
a
b
c
7
3.5
>100
>100
1.35
(54x)
8.8
2.4
199
0.34/
5.2
1
1
3.9
5.5
75.7
>94
>100
>100
>100
>100
>100
0.008
(1,737x)
2.0
ND
ND
7.9
3.9
<-1.5
<-1.5
0.3
<8
0.57/
9.9
6
6
b
c
0.002
(7,750x)
ND
<8
0.58/
10.2
50.9
>100
>100
26.8
0.022
(2,313x)
0.54/
7.6
6d
4
8
0.8
.20
0.038
(1,073x)
0.4
17.4
<8
0.44/
7.3
6
e
0.006
(1,366x)
-0.3
0.53/
8.2
6
6
f
g
35.1
4.18
62.9
19.4
>100
>100
0.053
9.0
6.8
-0.5
-0.3
<8
<8
0.45/
8.1
(1,186x)
6
6
h
i
0.004
0.52/
9.0
(1,045x)
2.10
17.5
95.3
0.004
525x)
5.2
1.0
14.8
0.52/
7.7
(
ACS Paragon Plus Environment

Page 11 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
6
6
j
20.2
1.31
>84
7.6
>100
>100
1.013
19.9x)
1.7
1.2
1.2
1.0
29.4
12.0
0.36/
5.1
(
k
0.060
(21.8x)
0.45/
6.7
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
a
b
c
IC50 values obtained from BACE1 CFA. IC50 values obtained from BACE2 CFA. IC50 values
d
obtained from CatD CFA. IC50 values obtained from BACE1 WCA; value in parentheses is the BACE1
CFA/WCA ratio. Ratio from the MS-based quantification of apical/basal and basal/apical transfer rates
of a test compound at 2 μM across contiguous monolayers from MDR1-transfected MDCK cells.
Hepatic clearance predicted from in vitro human microsomal stability study.
e
f
It was clear that the additional basic amine exerts a profound effect on whole-cell
potency, reinforced by comparison to the previously described monocyclic analog devoid of a P3
group (compound 6 in reference 21, BACE1 CFA IC50 = 36 µM, WCA IC50 = 636 nM).21 The
low molecular weight and dibasic nature of the small alkyl exemplars result in highly ligand- and
lipophilic efficient compounds in the whole-cell assay. While a similar rank order trend is
observed for the BACE1 CFA, there is a 20- to 7,750-fold disconnect observed between the two
assay formats. This disconnect can be rationalized by the presence of the second basic center
increasing the propensity for accumulation into the endosome, an acidic intracellular
3
compartment where BACE1 is thought to be localized, in the WCA format. As expected,
decreasing the basicity of the amine by the installation of an electron-withdrawing group
significantly compresses the CFA/WCA disconnect.
It was recognized from the outset that the diversion from an amide moiety would likely
result in a very different BACE1 substituent-derived SAR relative to the amides, as well as
presenting distinct challenges to achieving a balanced ADME profile. A second round of amine
optimization (Table 2) explored utilization of an electron-withdrawing group to balance ADME
properties as well as incorporation of branched amines to further enhance potency. In an effort
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 70
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
to more clearly define the impact of the EWGs on ADME balance, we measured the pK s for the
a
amidine (“pK
a
1”) and benzylic amine (“pK
a
2”) for each of the analogs.22 Removal of one of the
fluorines (6l) increases pK
a
2 relative to analog 6k, therefore increasing MDR Er (1.2 vs 3.1).
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
Introduction of two geminal methyl groups provides a significant improvement in BACE1 CFA
potency (6m vs 6k), albeit with a corresponding increase in MDR Er and a decrease in BACE2
selectivity (2x). Removal of one of the methyl groups (6n) decreases potency both in BACE1
CFA (1.9 µM) and WCA (64 nM), but improves efflux and BACE2 selectivity while reducing
CYP 2D6 inhibition. Increasing the distance between the amine and trifluoromethyl groups (6o)
increases pK 2 and diminishes potency in both assay formats. In an effort to capitalize on the
a
potency increase associated with increased substitution, cyclopropylamine-containing analogs
(6p-q) were prepared. The cyclopropane unit is well tolerated, as exhibited by the excellent
WCA potency, but suffers from a significant degradation in BACE1 CFA potency and MDR Er.
The addition of a difluoromethyl group (6r) improves the BACE1 CFA potency and reduces the
increase in MDR Er, but simultaneously introduces a metabolic liability (HLM = 22 mL/min/kg).
Replacement with a nitrile (6s) enhances BACE2 selectivity, but the MDR Er increases
significantly (MDR Er = 6.0). Use of an oxetane (6t) in place of the cyclopropane improves
WCA potency and decreases LogD, but results in an elevated MDR Er ratio (MDR Er = 3.5)
nearly equivalent to that of the parent cyclopropylamine 6p. In contrast, the CF -cyclopropyl
3
derivative 6u maintains good potency in both assay formats, exhibits a minimal efflux ratio and
low clearance, and maintains excellent selectivity over CatD (470x) and modest selectivity over
BACE2 (5.9x).
Table 2. In Vitro Data for Compounds 6k-6u
ACS Paragon Plus Environment

Page 13 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
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
Cpd R =
BACE1 BACE2 CatD
BACE
1
MDR LogD
Er
HLM CYP pK
a
1/
2
hERG
(µM)
e
g
h
i
CFA
IC50
(µM)
CFA
IC50
(µM)
CFA
IC50
(µM)
(mL/
min
/kg)
2D6
pK
a
WCA
IC50
a
b
c
f
d
(µM)
1.31
7.60
27.3
1.02
>100
>100
>100
0.060
0.058
0.031
1.2
3.1
3.6
1.0
0.3
1.5
12.0
13.0
<8.0
25%
21%
81%
8.9/
3.8
6.1
6
6
k
l
5
.89
8.9/
5.2
14.7
11.2
6m
0.49
8.9/
4
.2
1.97
54.7
20.6
7.13
>100
>100
0.064
0.400
2.2
6.1
0.6
0.5
11.9
9.3
45%
39%
8.8/
3.4
13.9
20.4
6
6
n
o
>100
8.9/
6.0
>100
35.9
>100
>100
0.030
0.014
3.4
6.9
0.5
0.1
<8.0
<8
14%
12%
9.0/
6.2
49.2
33.0
6
6
p
q
1
0
0
5.4
9.3/
6.3
.335
.952
1.01
11.5
65.5
71.8
0.012
0.024
2.5
6.0
0.7
0.1
22.2
<8.0
56%
40%
8.8/
4.2
13.7
28.4
6r
8.8/
1.8
6
6
s
t
0.770
0.069
1.89
>100
32.4
0.017
0.018
3.5
1.5
0.6
1.8
<8
45%
83%
8.8/
NA
16.5
3.5
6
u
0.405
<13
9.0/
2
.9
a
b
c
IC50 values obtained from BACE1 CFA. IC50 values obtained from BACE2 CFA. IC50 values
d
e
obtained from CatD CFA. IC50 values obtained from BACE1 WCA. Ratio from the MS-based
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 70
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
quantification of apical/basal and basal/apical transfer rates of a test compound at 2 μM across contiguous
f
monolayers from MDR1-transfected MDCK cells. Hepatic clearance predicted from in vitro human
g
microsomal stability study. CYP 2D6 % inhibition determined in human microsomes using a probe CYP
h
i
2
D6 substrate (dextromethorphan) and 3 µM of test compound. pK
a
values measured. Measured IC50 in
hERG-expressing CHO cells.
Although the decrease in pK 2 imparted by the introduction of the cyclopropyl group in
a
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
6u (2.9, vs 3.8 for 6k) could contribute to the improved BACE1 CFA potency, it is not sufficient
to explain the 20-fold potency increase observed. In order to better rationalize this improvement,
a co-crystal structure of compound 6u in BACE1 was obtained (PDB ID: 5T1U). As shown in
Figure 4, the thioamidine and difluorophenyl ring systems are oriented in a similar overall
fashion to previously described BACE1 structures, with the difluorophenyl occupying the S1
pocket as expected. As predicted by our initial design hypothesis, the benzylic amine substituent
adopts an orientation orthogonal to the plane of the difluorophenyl P1 group to optimally engage
the carbonyl of Gly230. Interestingly, the trifluoromethyl group orients toward and fills the
entrance to the S3 pocket, while the cyclopropyl substituent fills a small, lipophilic pocket at the
back of the interface between the S1 and S3 pockets. Arguably, the significant improvement in
BACE1 CFA potency observed for 6u is primarily achieved by optimally filling this cleft and
locking the CF into the direction of the S3 pocket.
3
ACS Paragon Plus Environment

Page 15 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Asp-228
Gly-230
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
Tyr-71
flap
S3
S1
10s loop
Figure 4: Crystal structure of 6u in BACE1 binding site. PDB ID: 5T1U
Overall, a liability of the series is P-gp-mediated efflux, likely driven in large part by the
presence of a third hydrogen bond donor and exacerbated by increasing basicity at the benzylic
amine (6o-6q, pK 2 > 6, MDR Er 3.4-6.9). It was recognized that the installation of an electron-
a
withdrawing group at the amine center aids in overcoming this issue, as illustrated by the
difluoromethyl-substituted analog 6r. Although the introduction of a nitrile substituent (6s)
further lowers pK
surface area, a key factor for P-gp liability. In contrast, the relatively non-polar trifluoromethyl
group in 6u decreases pK 2 sufficiently, resulting in a decreased efflux ratio.
a
2, the efflux liability increased, potentially because of the increase in polar
a
Gratifyingly, 6u addresses a number of the key challenges that had emerged throughout
the optimization of this series. As predicted by the in vitro transporter assay, 6u exhibits free
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 70
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
access to the CNS compartment in mice (Cu,b/Cu,p = 1.0), as determined by time-course AUC.
To assess in vivo potency, analog 6u was dosed in wild-type mouse via subcutaneous
administration at two doses, 10 and 100 mg/kg, to measure impact on levels of brain Aβx-42. At
the lower dose, a small but significant decrease is observed at early timepoints, and at 100
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
mg/kg, robust lowering is observed out to 20 hours post-dose, providing a Cu,b Ceff of 31 nM for
5% Aβ lowering using previously described methodology (Figure 5).23
2
Figure 5. Effect of trifluoromethylcyclopropyl 6u on brain Aβx-42 following subcutaneous
administration in wild-type mice.
A goal from the outset had been to mitigate the hERG liability often observed for BACE1
inhibitors. Therefore, hERG IC50 values were generated for this second round of optimization.
All analogs exhibit some affinity for hERG (IC50 3.5-49 µM), although there is no correlation
with increasing logD and decreasing pK . The hERG TI, as defined by comparing the hERG
a
ACS Paragon Plus Environment

Page 17 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
IC50 to the more robust and relevant measured Cu,p
C
eff value for BACE1, provides the most
appropriate selectivity descriptor; for 6u, this corresponds to an in vivo hERG selectivity of
1
13x, satisfying our initial criteria.
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
Whereas 6u had achieved a balance of Aβ-lowering in brain, CNS penetration, and
selectivity over hERG, it was, unfortunately, also characterized by significant inhibition of
cytochrome P450 subtype 2D6 (CYP 2D6). CYP inhibition, in general, is undesirable due to the
potential for drug-drug interactions (DDI) through inhibition of oxidative metabolism, but the
significance is enhanced for subtype 2D6 because of significant polymorphism in 2D6
expression.24 Within this set of analogs, CYP 2D6 inhibition appears to be primarily driven by
two main factors, namely, pK
a
and the size of the benzylamine substituent. Unfortunately, the
, such that analogs with decreased pK at the benzylic
CYP 2D6 values inversely correlate to pK
a
a
amine show greater inhibition – in direct opposition to the SAR utilized to obviate hERG and
efflux transporter liabilities. In addition, increasing the size of the amine substituent appears to
enhance CYP 2D6 inhibition, tracking with BACE1 potency. In an effort to more accurately
gauge the CYP 2D6 inhibition, IC50 curves were generated for a subset of these compounds. In
this assay, the unflanked trifluoroethylamine 6k showed only modest CYP 2D6 inhibition (IC50
=
14 µM), whereas 6u, much more potent at BACE1, significantly inhibited CYP 2D6 (IC50
=
1
57 nM). The intertwined SAR of CYP 2D6 inhibition, BACE1 potency, and transporter
liability suggested that further tuning of the amine was unlikely to balance overall properties.
Several alternative strategies were considered, including: a) disrupting the favorable binding
interactions between this series and CYP 2D6 while maintaining the high BACE1 affinity
observed for 6u; and b) building upon the modest BACE1 potency and absence of CYP 2D6
liability associated with 6k, but utilizing an alternative vector.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 70
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
Disruption of the binding of this series to CYP 2D6 was informed by a recent report of a
pyrazole-containing thioamidine series, which described a similar challenge with CYP 2D6
affinity (Figure 6).25
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 6. Metabolic profile of fused THP series.
In this case, the fused pyrazole 7 is metabolized exclusively by CYP 2D6. Additionally,
the primary metabolism product observed, the demethylated pyrazole, is a potent inhibitor of
CYP 2D6 (IC50 = 0.37 µM). The CYP 2D6 affinity in this series was ultimately conquered
through the installation of a substituent [fluoromethyl (8) or methyl (not shown)] adjacent to
sulfur that disrupts binding to the CYP 2D6 active site. Utilization of this strategy guided the
design of compound 12, where a chiral methyl group has been introduced next to the sulfur of
the thioamidine (Scheme 2). This analog was prepared in a similar fashion to the unsubstituted
thioamidines, starting with the known bromide 9.20
Scheme 2a
ACS Paragon Plus Environment

Page 19 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
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
a
o
Reagents and conditions: (a) MeLi, Et
2
O, n-BuLi, then DMF, -78 C, 89%; (b) amine, Na(OAc)
3
BH,
DCE, rt, 93%; (c) HCl, dioxane, rt, 19%.
Overall, compound 12 maintains potency at BACE1, but exhibits increased clearance and
MDR Er as compared to 6u (Table 3). These general trends were consistent with the pairwise
changes associated with this matched molecular pair, as observed in the fused THP series.
Interestingly, however, there was only modest impact on the CYP 2D6 inhibition (IC50 = 334
nM) for this compound. The lack of translational impact for this methyl group across the two
series is striking, and in an effort to better understand the inconsistent impact on CYP 2D6
inhibition, co-crystal structures of 6u and 12 in CYP 2D6 were generated.
Table 3: In vitro profile of 12
Cpd BACE1
BACE2
CatD
WCA
MDR
Er
LogD HLMf
(mL/min 2D6
CYP
pK
pK
a
1/
2
hERGh
(µM)
e
CFA IC50
CFA IC50
(µM)
CFA IC50
(µM)
IC50
a
a
b
c
d
(µM)
(nM)
/kg)
IC50
µM)g
(
1
2
0.078
0.228
>100
0.024
3.2
2.4
38
0.334
8.8/2.8
2.6
a
b
c
IC50 values obtained from BACE1 CFA Assay. IC50 values obtained from BACE2 CFA Assay. IC50
d
e
values obtained from CatD CFA Assay. IC50 values obtained from BACE1 WCA. Ratio from the MS-
based quantification of apical/basal and basal/apical transfer rates of a test compound at 2 μM across
contiguous monolayers from MDR1-transfected MDCK cells. Hepatic clearance predicted from in vitro
human microsomal stability study. CYP 2D6 inhibition was obtained by measuring inhibition of 5 µM
f
g
h
dextromorphan in pooled HLM (HL-MIX-102). Measured IC50 in hERG-expressing CHO cells.
X-ray crystal structures of CYP 2D6 with 6u and 12 bound were determined as described
2
5
previously for 7 (PDB: 4XRY), by soaking crystals of a CYP 2D6 thioridazine complex with
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 70
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 appropriate compound (6u or 12). Comparison of the resulting structures (Figures 7A–C)
illustrates that each of the thioamidines is anchored through an interaction with Glu-216,
positioning the difluorophenyl ring proximal to Helix I. The aminomethyl substituent in analogs
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
6
u (Figure 7A) and 12 (Figure 7B) is oriented to the lipophilic pocket distal to the heme, whereas
the pyrazole of 7 is oriented towards the heme itself (Figure 7C). The orientations of the
thioamidine rings in 6u and 12 are shifted, relative to that of the fused 7, to facilitate occupation
of the distal pocket by the aminomethyl substituents. This reorientation creates greater distance
between Helix I and the difluorophenyl ring, and this shift is exacerbated slightly by the chiral
methyl group in 12 (Figure 7B), which is easily accommodated. In contrast, the rigid
architecture defined by the ring fusion of the thioamidine and THP restricts the rotation of 7
(Figure 7C), as the compound occupies the cleft between Phe-120 and the residues Glu-244 and
Phe-247 on helix G. The addition of a methyl or fluoromethyl group next to the sulfur of 7 then
creates a direct clash with Helix I, as evidenced by the lack of affinity of 8. Therefore, this
restricted rotation underlies the pronounced substituent impact on CYP 2D6 affinity in the fused
series, whereas the facility for reorientation in the monocyclic series 6u/12 minimizes the
substituent effect.
ACS Paragon Plus Environment

Page 21 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 7. Comparison of the binding of 6u (A) and 12 (B) with the binding of 7 (C) in the active site of
CYP 2D6.
Therefore, an alternative strategy to mitigate CYP 2D6 affinity in the aminomethyl series
would exploit the restricted rotation imparted by ring fusion, precluding occupation of the distal,
lipophilic pocket of CYP 2D6 by an aminomethyl substituent. In addition, from experience in
previous series, the fused THP ring would be expected to enhance BACE1 potency relative to the
monocyclic congeners. Considering the increased lipophilicity and molecular weight imparted
by bolting on the additional THP ring, the unflanked trifluoroethyl amine 6k seemed an
attractive starting template with which to execute this strategy. Fused analog 16 was therefore
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
2
6
designed and prepared as shown in Scheme 3. Starting from bromide 13, the aminomethyl
linker was installed by Pd-mediated cyanation and subsequent reduction using Raney nickel.
Alkylation of the resultant primary amine using trifluoroethyl triflate, followed by cleavage of
the benzamide protecting group, provided the fused analog 16. The structure of 16 was
confirmed by a single crystal X-ray structure carried out on the phosphate salt (See Supporting
Information).
Scheme 3a
aReagents and conditions: (a) Pd
, DMA, 130 C, 85%; (b) Raney Ni, H
o
,
2
(dba)
3
, dppf, zinc, Zn(CN)
2
2
triethylamine, Boc
2
O,
H
2
O, EtOH, rt, then HCl, MeOH, 61%; (c) 2,2,2-trifluoroethyl
o
trifluoromethylsulfonate, trimethylamine, MeCN, 70 C, then DBU, MeOH, 71%.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 70
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
As expected, the bicyclic fusion offered a 17x improvement of BACE1 CFA potency for
1
6 relative to the monocyclic 6k (Table 4), and 16 proved to be equipotent to 6u, reinforcing the
observation that potency can be obtained in either of the two available vectors. Further, 16
maintained excellent CatD selectivity, similar selectivity over BACE2 (3.8x) and a balanced
overall ADME profile, including, gratifyingly, a significantly diminished CYP 2D6 liability
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
(
IC50 = 9.1 µM, 118-fold selectivity over BACE1 CFA) relative to the monocyclic analogs. A
modest increase in efflux was observed relative to 6k, which translated into asymmetry in
brain/plasma ratio (Cu,b/Cu,p = 0.25, AUC ratio). Additionally, 16 exhibited excellent central Aβ-
lowering in mice, with a measured brain Ceff for 25% lowering of Aβx-42 of 31 nM (Figure 8).
The hERG value (IC50 = 4.3 µM) was slightly improved relative to 6k, reflecting a better
nominal selectivity (139x). However, accounting for brain asymmetry, the resultant hERG TI
diminishes to 35x over the requisite plasma concentrations.
Table 4. In Vitro Data for Compound 16
Cpd BACE BACE CatD
CFA 2 CFA CFA
WCA
MDR
Er
LogD
HLMf CYP
pK
pK
a
1/
2
hERG
(µM)
C
C
u,b
/
eff
C ,
(Cu,b,
nM)
e
h
1
IC50
(nM)
(mL/
min
/kg)
29
2D6
IC50
(µM)
a
u,p
d
IC50
IC50
(µM)
0.295
IC50
(µM)
>100
b
c
g
(
µM)a
0.077
1
6
0.006
2.3
1.1
9.1
7.91/
.80
4.3
0.25
31
3
a
b
c
IC50 values obtained from BACE1 CFA. IC50 values obtained from BACE2 CFA. IC50 values
obtained from CatD CFA. IC50 values obtained from BACE1 WCA. Ratio from the MS-based
d
e
quantification of apical/basal and basal/apical transfer rates of a test compound at 2 μM across contiguous
f
monolayers from MDR1-transfected MDCK cells. Hepatic clearance predicted from in vitro human
g
microsomal stability study. CYP 2D6 inhibition was obtained by measuring inhibition of 5 µM
h
dextromorphan in pooled HLM (HL-MIX-102). Measured IC50 in hERG- expressing CHO cells.
ACS Paragon Plus Environment

Page 23 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
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
Figure 8. Effect of fused THP analog 16 on brain Aβx-42 following subcutaneous
administration in wild-type mice.
The observation of such disparate responses in the relative affinities to BACE1 and CYP
2
D6 for 16 versus 6k was striking. The addition of the THP ring resulted in a 17x increase in
potency at BACE1 while essentially having no impact on CYP 2D6 affinity (CYP 2D6 IC50 for
k = 14 µM). A crystal structure of 16 in BACE1 was generated and confirmed that, as
6
expected, reorganization of the S2’ subpocket in BACE1 had occurred, driven by the rotation of
Tyr71 to accommodate the THP ring and fluoromethyl substituent of 16 (Figure 9). Therefore,
selectivity is realized because this subpocket in BACE1 is responsive to increases in lipophilic
bulk in this vector, whereas the same vector in CYP 2D6 directs towards the heme and has no
impact on affinity. In addition, the ring fusion limits the requisite rotation of the thioamidine
ring to optimally accommodate the aminomethyl substituent within the CYP 2D6 binding site.
This discrepant response contrasts with the parallel responsiveness at BACE1 and CYP 2D6 for
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 70
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
small changes in the benzylic amine of the monocyclic series, wherein equivalent increases in
lipophilicity afforded corresponding enhancements in affinity for both BACE1 and CYP 2D6
receptors (6k vs 6u, 20x BACE1, 9x CYP 2D6). Future BACE1 design efforts focused on
avoiding CYP 2D6 liability will therefore target manipulation of the amidine “head group”
architecture, rather than optimization of P3 substituents.
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
Asp-228
S2’
Tyr-71
flap
Gly-230
S3
S1
1
0s loop
Figure 9. Co-crystal structure of 16 in BACE1. PDB ID: 5T1W
Based upon its attractive balance of potency, selectivity, and CNS penetration, 16 was
selected for further safety profiling. The compound showed excellent in vitro selectivity, as
gauged by screening against the broad CEREP bioprint panel (<25% inhibition at all targets at 10
µM). Subsequent advancement to long-term in vivo toxicology studies, including a 4-week arm
ACS Paragon Plus Environment

Page 25 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
at 100 mg/kg (30x AUC over Ceff), revealed no ocular findings, presumably reflecting the
excellent in vitro selectivity observed for 16.
Conclusion
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
The optimization of a series of brain-penetrant BACE1 inhibitors with a novel
aminomethyl linker has been described. This linker mitigates an aniline structural alert prevalent
throughout a large contingent of literature BACE1 series. Careful modulation of the amine
substituent aided in garnering a balanced profile, as the inherent efflux transporter liability and
potency are dependent upon the resultant pK . A number of the more active compounds carried
a
significant CYP 2D6 inhibition that correlated well with the size of the amine substituent.
Unlike the fused THP series, substitution on the monocyclic thioamidine ring did not ameliorate
CYP 2D6 inhibition, which was rationalized through examination of CYP 2D6 crystal structures.
Filling the BACE1 S2’ subpocket by fusing a THP ring onto the monocyclic thioamidine ring
improved BACE1 potency dramatically without impacting CYP 2D6 affinity. One such example
(16) demonstrates good ADME balance, modest brain asymmetry, no DDI potential, and good in
vivo efficacy, illustrating that the aminomethyl linker can serve as a robust replacement to
overcome the significant challenges and structural alerts inherent in the classical P1/P3 amide of
previously described BACE1 inhibitors.
Experimental Section
Biology
In Vitro Pharmacology sAPPβ Whole-Cell Assay (WCA): sAPPβ, the primary cleavage product
of BACE1, was determined in H4 human neuroglioma cells over-expressing the wild-type
human APP695. Cells were treated for 18 h with compound in a final concentration of 1%
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 70
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
DMSO. sAPPβ levels were measured by ELISA with a capture APP N-terminal antibody
(Affinity BioReagents, OMA1-03132), wild-type sAPPβ-specific reporter antibody p192 (Elan),
and tertiary anti-rabbit-HRP (GE Healthcare). The colorimetric reaction was read by an
EnVision (PerkinElmer) plate-reader.
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
BACE1 Enzyme Cell-Free Assay (FP): Beta secretase-1 activity was assessed with soluble
BACE1 and the synthetic APP substrate Biotin-GLTNIKTEEISEISYEVEFR-C[oregon
green]KK-OH in the presence of compounds in a fluorescence polarization (FP) in vitro
assay. Enzyme, substrate and test compounds were incubated in 15 μL of 100 mM sodium
o
acetate pH=4.5 buffer containing 0.001% Tween-20 for 3 hours at 37 C. Following the addition
of saturating immunopure streptavidin, fluorescence polarization was measured with a
PerkinElmer EnVision plate reader (Ex485 nm/ Em530 nm).
In Vivo Experiments All procedures performed on animals in this study were in accordance
with established guidelines and regulations, and were reviewed and approved by the Pfizer (or
other) Institutional Animal Care and Use Committee. Pfizer animal care facilities that supported
this work are fully accredited by AAALAC International.
Acute Treatment in Mice: Male 129/SVE wild-type mice (20–25 g) were in a non-fasted state
prior to subcutaneous dosing with vehicle, or compound 6u or 16, using a dosing volume of
1
1
0 mL/kg in 5:5:90 DMSO:Cremophor:saline vehicle. Mice (n = 5 per group) were sacrificed at
, 3, 5, 7, 14, 20 and 30 h post-dose. Whole blood samples (0.5–1.0 mL) were collected by
cardiac puncture into ethylenediaminetetraacetic acid (EDTA)-containing tubes, and plasma was
separated by centrifugation (1500 × g for 10 min at 4 C). The generated plasma was distributed
into separate tubes on wet ice for exposure measurements (50 L) and A analysis (remainder).
ACS Paragon Plus Environment

Page 27 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
CSF samples (8–12 L) were obtained by cisterna magna puncture using a sterile 25 gauge
needle and collected with a P-20 Eppendorff pipette. CSF samples were distributed into separate
tubes on dry ice for exposure measurements (3 L) and A analysis (remainder). Whole brain
was removed and divided for exposure measurements (cerebellum) and A analysis (left and
right hemispheres), weighed, and frozen on dry ice. All samples were stored at –80 C prior to
assay.
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
Measurement of Rodent Amyloid-β: Frozen mouse hemi-brains were homogenized (10% w/v) in
5 M guanidine HCl, using a Qiagen TissueLyser. Each sample was homogenized with a 5 mm
stainless steel bead, four times, at a shaking rate of 24 times/s for 90 s, then incubated at 25 °C
for 3 h, and ultracentrifuged at 125,000 x g for 1 h at 4 C. The resulting supernatant was
removed and stored in a 96-well polypropylene deep well plate at –80 C. The A peptides were
further purified through solid-phase extraction using Waters Oasis reversed-phase HLB 96-well
column plates (60 mg). Column eluates in ammonium hydroxide from 500-800 L of original
brain supernatant were evaporated to complete dryness and stored at –80 C until assay. For
plasma analysis, 140-175 μL of mouse plasma was treated 1:1 with 5 M guanidine HCl and
incubated overnight with rotation at 4 ºC. The entire volume was then purified through solid-
phase extraction as indicated above.
Samples were analyzed using a Dissociation-Enhanced Lanthanide Fluorescent Immuno-Assay
(DELFIA) platform Enzyme-Linked Immunosorbent Assay (ELISA). Configuration of the
antibodies used in determining the level of Aβx-40 and Aβx-42 utilizes a common detect
antibody (4G8) in combination with specific C-terminal antibodies for the 40 and 42 cleavage
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 70
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
sites. For the Aβx-40 assay, a 384-well black Nunc Maxisorp plate was coated with 15 L/well
(4 g/mL) capture antibody (Rinat 1219) in 0.1 M sodium bicarbonate coating buffer, pH 8.2.
For the Aβx-42 assay, 15 L/well (8 g/mL) capture antibody (Rinat 10G3) was used. The
plates were sealed and incubated at 4 C overnight. Plates were washed with phosphate-buffered
saline containing 0.05% Tween-20 (PBS-T), and blocked with 75 L of blocking buffer (1%
BSA in PBS-T) for 2 h at 25 °C.
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
After washing the plates with PBS-T, rodent Aβx-40 (California Peptide) or Aβx-42 (California
Peptide) standard was serially diluted in blocking buffer and 15 L was applied to the plate in
quadruplicate. Dried brain samples were reconstituted in 120 L of blocking buffer, which
corresponds to a 4.16- to 6.67-fold concentration. Then 15 L of undiluted brain sample was
added to the Aβx-42 assay plate in triplicate, or 15 L of a 1:2 diluted brain sample was added to
the Aβx-40 assay plate in triplicate. Dried plasma samples were reconstituted in 40 μL of
blocking buffer, which corresponds to a 3.5- to 4.38-fold concentration, and 15 μL was added to
the Aβx-40 assay plate in duplicate. CSF samples were diluted 1:8 in blocking buffer, and 15 μL
was added to the Aβx-40 assay plate in duplicate. Plates were incubated with sample or standards
for 2 h at 25 °C. The plates were washed with PBS-T, and 15 L of detecting antibody (4G8-
Biotin, Covance), 200 ng/mL in blocking buffer, was added to each well, incubating for 2 h at 25
°
C. The plates were then washed with PBS-T, and 15 L of europium-labeled Streptavidin
(PerkinElmer), 50 ng/mL in blocking buffer was added for a 1 h incubation in the dark at 25 °C.
The plates were washed with PBS-T, and 15 L of PerkinElmer Enhancement solution was
added to each well with 20 minute incubation at rt. Plates were read on an EnVision plate reader
using DELFIA time-resolved fluorimetry (Exc340/Em615), and samples were extrapolated
ACS Paragon Plus Environment

Page 29 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
against the standard curve using 4-parameter logistics. Measurement of human amyloid- in
plasma and CSF from PS1/APP mice utilizes the same capture and detecting antibodies used for
wild-type mice. Vehicle-treated samples from plasma and CSF were serially diluted to optimize
sample dilution to the linear phase of an A peptide standard curve. A levels were measured
using DELFIA ELISA.
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
Neuropharmacokinetic Studies in Male CD-1 Mice: The in-life and bioanalytical portions of
these studies were conducted at BioDuro, Pharmaceutical Product Development Inc. (Beijing,
China). Male CD-1 mice were obtained from PUMC, China. Mice received a 10 mg/kg
subcutaneous (sc) dose of compounds 6u or 16. The doses were prepared in 5% DMSO/95%
water containing (v/v) 0.5% methylcellulose (w/v) and delivered in a volume of 5 mL/kg.
Animals were sacrificed in a CO chamber. Blood, brain and CSF samples were collected at 1, 4
2
and 7 hours post-dosing. Plasma was isolated after centrifugation. The plasma, brain and CSF
samples were stored at –80 °C prior to analysis.
Measurement of Fractions Unbound in Brain: The unbound fraction of each compound was
determined in brain tissue homogenate using a 96-well equilibrium dialysis method as described
2
7
by Kalvass et al. with the following exceptions. Brain homogenates were prepared from freshly
harvested rat brains following dilution with a 4-fold volume of phosphate buffer and spiked with
1
μM compound. The homogenates were dialyzed against an equal volume (150 μL) of
o
phosphate buffer at 37 C for 6 h. Following the incubation, equal volumes (50 μL) of brain
homogenate and buffer samples were collected and mixed with 50 μL of buffer or control
homogenate, respectively, for preparation of mixed matrix samples. All samples were then
precipitated with internal standard in acetonitrile (200 μL), vortexed, and centrifuged.
ACS Paragon Plus Environment