Development of 2-aminooxazoline 3-azaxanthene β-amyloid cleaving enzyme (BACE) inhibitors with improved selectivity against Cathepsin D

Article abstract of DOI:10.1039/c7md00106a

As part of an ongoing effort at Amgen to develop a disease-modifying therapy for Alzheimer's disease, we have previously used the aminooxazoline xanthene (AOX) scaffold to generate potent and orally efficacious BACE1 inhibitors. While AOX-BACE1 inhibitors demonstrated acceptable cardiovascular safety margins, a retinal pathological finding in rat toxicological studies demanded further investigation. It has been widely postulated that such retinal toxicity might be related to off-target inhibition of Cathepsin D (CatD), a closely related aspartyl protease. We report the development of AOX-BACE1 inhibitors with improved selectivity against CatD by following a structure- and property-based approach. Our efforts culminated in the discovery of a picolinamide-substituted 3-aza-AOX-BACE1 inhibitor absent of retinal effects in an early screening rat toxicology study.

Full text of DOI:10.1039/c7md00106a

View Article Online
View Journal
MedChemComm
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: J. Low, M.
Bartberger, K. Chen, Y. Cheng, M. Fielden, V. Gore, D. Hickman, Q. Liu, A. Sickmier, H. VArgas, J. Werner,
R. White, D. Whittington, S. Wood and A. E. Minatti, Med. Chem. Commun., 2017, DOI:
10.1039/C7MD00106A.
This is an Accepted Manuscript, which has been through the
Volume 7 Number 1 January 2016 Pages 1–204
Royal Society of Chemistry peer review process and has been
accepted for publication.
MedChemComm
Accepted Manuscripts are published online shortly after
acceptance, before technical editing, formatting and proof reading.
Broadening the field of opportunity for medicinal chemists
www.rsc.org/medchemcomm
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
author guidelines.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
Themed issue: Antibiotic Resistance
ISSN 2040-2503
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
rsc.li/medchemcomm

Please do not adjust margins
MedChemComm
Page 1 of 13
View Article Online
DOI: 10.1039/C7MD00106A
MedChemComm
ARTICLE
Development of 2-Aminooxazoline 3-Azaxanthene β-Amyloid
Cleaving Enzyme (BACE) Inhibitors with Improved Selectivity
against Cathepsin D‡
Received 00th January 20xx,
Accepted 00th January 20xx
Jonathan D. Low,a Michael D. Bartberger,b Kui Chen,c Yuan Cheng,a Mark R. Fielden,d Vijay Gore,†a
Dean Hickman,e Qingyian Liu,a E. Allen Sickmier,f Hugo M. Vargas,d Jonathan Werner,d Ryan D.
White,g Douglas A. Whittington,f Stephen Wood,h Ana E. Minatti,*a
DOI: 10.1039/x0xx00000x
www.rsc.org/
As part of an ongoing effort at Amgen to develop a disease-modifying therapy for Alzheimer’s disease, we have previously
used the aminooxazoline xanthene (AOX) scaffold to generate potent and orally efficacious BACE1 inhibitors. While AOX-
BACE1 inhibitors demonstrated acceptable cardiovascular safety margins, a retinal pathological finding in rat toxicological
studies demanded further investigation. It has been widely postulated that such retinal toxicity might be related to off-
target inhibition of Cathepsin D (CatD), a closely related aspartyl protease. We report the development of AOX-BACE1
inhibitors with improved selectivity against CatD by following a structure- and property-based approach. Our efforts
culminated in the discovery of a picolinamide-substituted 3-aza-AOX-BACE1 inhibitor absent of retinal effects in an early
screening rat toxicology study.
translation of changes in cerebrospinal fluid Aβ levels, as an
effect biomarker for brain A lowering, in response to small
Introduction
β
molecule BACE1 inhibition is well established in rodents and
higher preclinical animal species.7 The first prominent and
A widely held hypothesis for the underlying pathogenesis of
Alzheimer’s disease (AD) proposes aggregation of
peptides (A ) and cerebral deposition of amyloid plaques as
critical events.1 A
is produced via sequential endoproteolytic
cleavage of amyloid precursor protein (APP) by the aspartyl
protease -site APP cleaving enzyme-1 (BACE1) and
secretase.2 Inhibition of BACE1 decreases production of all
major forms of A including the most pathogenic species Aβ42
β-amyloid
long-lasting Aβ reduction in human lumbar cerebrospinal fluid
β
(CSF) was disclosed in 2011.8 Since then, several companies
have advanced additional small molecule BACE1 inhibitors to
human clinical trials for the treatment of AD.9-11 Potential
safety liabilities of BACE1 inhibitors are less well studied. As an
example, the development of the clinical candidate LY2811376
was terminated in phase 1 due to adverse retinal toxicity in a
3-month rat preclinical toxicology study.8 We reported a
similar pathology in a 1-month preclinical toxicology study in
Sprague Dawley rats administered with the lead compound,
AMG-8718.12,13 Retinal toxicity in the rat was characterized by
retinal thinning preceded by the accumulation of
autofluorescent granules (AFG) in the retinal pigment
epithelium (RPE), suggestive of disruption of phagolysosomal
function in the RPE.14 These findings prompted us and others
to evaluate additional BACE1 inhibitors for their potential to
increase AFG in the RPE of preclinical species, as an early
marker of incipient retinal thinning. To date, the results of
several preclinical toxicology studies involving BACE1 inhibitors
as they relate to retinal toxicity have been disclosed (Figure 1):
six BACE1 inhibitors tested “positive” for induction of AFG in
β
β
γ-
β
,
suggesting that BACE1 is a prime therapeutic target for the
development of disease modifying therapies for AD.3-6 The
the RPE (LY2811376 8
and PF-928316
whereas four BACE1 inhibitors tested
“negative” (LY2886721 16, 17 2,18 3 16 and NB-36019). We report
, , , , ,
AMG-871812 115 PF-3195 16 PF-1283 16
)
,
,
herein the discovery of an aminooxazoline xanthene (AOX)-
BACE1 inhibitor (Figure 1, 30) that is devoid of early signs of
retinal pathology.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 1
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 2 of 13
View Article Online
DOI: 10.1039/C7MD00106A
ARTICLE
A current hypothesis is that the retinal toxicity in preclinical
MedChemComm
5Å
of the co-crystallized ligand
4
are identical between the
species is due to off-target inhibition of Cathepsin D (CatD), a two species.24, 25 To the best of our knowledge, this constitutes
structurally related aspartyl protease that plays an important the first time a co-crystal of a BACE1 inhibitor with the CatD
role in phagolysosomal processing in the RPE.20 The scientific protein has been published. The catalytic aspartic acids and
evidence for CatD as a possible mediator of the retinal toxicity the flap tyrosine are conserved in both CatD (Figure 2a) and
is mounting as progress has been made linking the observed in BACE1 (Figure 2b, PDB 4WTU)23 and drive the same alignment
vivo accumulation of AFG in the RPE to enzymatic (biophysical of AOX core in both proteins. The P326 group occupies the S3
or biochemical) and cellular CatD inhibition, together with pocket in CatD as well as in BACE1, but the protein surfaces
lysosomal accumulation.16
reveal an opportunity to extend a vector from the 7 position of
the xanthene core deeper into the S3 pocket of BACE1 and
potentially increase selectivity over CatD (Figure 2 a and b). In
particular, analysis of the aligned S3 regions of BACE1 and
CatD (Figure 2c) reveal a key difference between an alanine
residue (Ala335) in BACE1 and the correspondingly larger and
more polar Asp318 residue in the same binding side region of
CatD. This key steric and polar residue mismatch in a relatively
buried region of the S3 subsite afforded an attractive and
compelling potential selectivity handle to pursue.
S
F
R
H
H2N
H2N
S
O
O
N
N
N
N
N
N
PF-3195 (*) PF-1283 (*)
(R=CH2F) (R=CH2F)
PF-9283 (*)
O
F
F
F
LY2811376 (*)
(R=Me)
H2N
H2N
O
O
O
N
F
N
F
N
N
N
F
N
The challenge was multifaceted regarding the property-based
approach. The importance of choosing an amidine-type BACE1
inhibitor warhead with an optimal pKa has been extensively
discussed in the context of: a) BACE1 cell potency;27-30 b)
central nervous system (CNS) exposure and robust CNS
O
N
O
AMG-8718 (*)
1 (*)
F3C
H2N
S
NH2
F
NC
H
O
N
N
H
N
H
N
N
N
O
O
pharmacodynamic (PD) activity;31, 32 and c) potential lysosomal
O
O
F
F
33-35
NB-360 (**)
LY2886721 (**)
accumulation and polypharmacology.14,
The pKa of the
aminooxazoline warhead is influenced by the core,31 with the
Me
N
H2N
S
H2N
MeO
SO2
H
N
3-aza-xanthene core maintaining the pKa of the 2-
N
H
N
N
N
aminooxazoline (e.g. 1 pKa = 5.7) in a desirable range (5.5 < pKa
O
F
F
F
< 7.5;); i.e., maximizing engagement with the catalytic
aspartate residues within the BACE1 protein while minimizing
the potential for lysosomal accumulation. Additionally, the
overall properties of the molecule as they relate to CNS
penetration/CNS drug-like space were closely monitored
throughout the process of modifying the structure towards the
goal of increasing selectivity over CatD.36-38
F
2 (**)
3 (**)
H2N
O
Cl
N
H
N
N
this publication:
30 (**)
O
O
N
O
Fig 1. BACE1 inhibitors that tested “positive” (*) or “negative” (**) for induction
of AFG in the RPE in preclinical species. Each study design varied with respect to
preclinical species, compound dose and study length.
The aim of this work was to evaluate alkynylpyridines39-47 and
19, 31, 32, 41, 46, 48-53
picolinamides17,
as P3 sub-pocket (P3sp)
groups on the 3-aza-AOX core and test their influence on CatD
selectivity. The strategy was to initially omit the P2’
substituent, in order to offset the increase in molecular weight
(MW) upon extending into the S3sp and maximize lipophilic
efficiency (Figure 3).
Our objective was to improve the selectivity of the AOX-
BACE1 inhibitors against human CatD (CatD/BACE1 selectivity)
by using a structure- and property-based approach. Our
measurement of CatD/BACE1 selectivity was based on an
enzymatic in vitro assay and we aimed for a large in vitro
selectivity window.21 This in vitro data allowed us to quickly
rank-order compounds for subsequent in vivo toxicological
profiling in rats with an emphasis on retinal evaluation. We
focused our efforts on the aminooxazoline 3-aza-xanthene (3-
aza-AOX) core as the retinal toxicity profile of the lead
molecule of this series, (compound 1 in Figure 1) had been
extensively studied in rats and cynomolgus monkeys.15
Fig 3. Strategy to improve selectivity of 3-aza-AOX-BACE1 inhibitors against CatD.
Our structure-based approach was based on exploiting the
residue differences in the S3 sub-pocket (S3sp) – also referred
to as the deep S3 pocket – of the BACE1 protein versus the
CatD protein.18, 22 Figure 2a shows the co-crystal structure of
rat CatD with BACE1 inhibitor 423 (PDB 5UX4) – a closely
related analog of
1. The amino acid sequence of rat CatD
shows 83 % identity with human CatD and the residues within
2 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 3 of 13
View Article Online
DOI: 10.1039/C7MD00106A
MedChemComm
ARTICLE
Fig 2. (a) Co-crystal structure of rat CatD and AOX-BACE1 inhibitor
clarity. (b) Co-crystal structure of human BACE1 and AOX-BACE1 inhibitor
shown in blue and its surface is not depicted for clarity. (c) Overlaid co-crystal structure of human BACE1 (blue) + compound
4
(yellow, PDB 5UX4). The mobile flap region is shown in pink and its surface is not depicted for
(yellow; PDB 4WTU).23 Red spheres represent water molecules. The mobile flap region is
(yellow) overlaid with co-crystal
4
4
structure of rat CatD (pink) + compound
4. Key residues are labelled in yellow for BACE1 and pink for CatD. Key hydrogen bonds are depicted as yellow dashed lines.
The synthesis of deep-P326 alkyne containing 3-aza-AOX
compounds is illustrated in Scheme 2.55 Miyaura borylation of
Results and Discussion
intermediate
boronic ester
5
8
under standard reaction conditions afforded
in 91% yield. Subsequent palladium-catalyzed
Chemistry
Suzuki-cross coupling of
8 with the corresponding 3-bromo-5-
The 3-aza-AOX BACE1 inhibitors profiled in Tables 1-3 were
synthesized via several routes. The synthesis of the truncated
alkyne-pyridines using Pd(AmPhos)2Cl2 provided compounds 9-
12 in moderate yields. The chloro-pyridine of these alkyne-
substituted 3-aza-AOX analogs (9-12) was further converted
into more elaborate P2’ substituents via Suzuki-cross coupling.
Three privileged P2’ substituents from formerly published lead
AOX-BACE1 inhibitors15, 23, 56 were installed by using either the
corresponding boronic acid or pinacol boronic ester affording
compounds 13-18.
BACE1 inhibitors
6
and
7
is shown in Scheme 1. The sequence
was disclosed in a
for the synthesis of key intermediate
5
previous publication15 and additional experimental details for
each step including characterization of the intermediates can
be found in the supporting information of this publication.
Suzuki-Miyaura-coupling of the unprotected enantiomerically
pure intermediate
acid or 5-fluoro-3-pyridine boronic acid yielded compounds
and respectively. The reaction proceeded with high
5 with either 2-fluoro-3-pyridine boronic
All deep-P3 amide-substituted 3-aza-AOX compounds were
synthesized via amide coupling of the corresponding amino-
6
7,
(phenyl)-substituted 3-aza-AOX intermediates (20, 23-26 and
regioselectivity for the bromide (96%
respectively) using (Pd(AmPhos)2Cl2).54
(6) and 58% (7)
31) and 5-chloropicolinic acid but with slight variations, as
depicted in Scheme 3. For ease of discussion “protected” and
“unprotected” refers to the aminooxazoline warhead
functional group. Intermediate
5 was mono-benzoyl protected
to give bromide 19, which was converted to the amino-
substituted-benzoyl-protected 3-aza-AOX 20 via Pd-catalyzed
coupling with LiHMDS as the ammonia surrogate.57
Intermediate 20 was reacted with 5-chloropicolinic acid under
1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDCI) amide
coupling conditions, followed by cleavage of the benzoyl
protecting group using sodium hydroxide, to yield the final
Scheme 1 Reagents: (a) 2-fluoropyridin-3-ylboronic acid or 5-fluoropyridin-3-
ylboronic acid, Pd(AmPhos)2Cl2 (cat.), K3PO4, dioxane/water, 100 °C, 96% ( ) or
58% (
6
7
)
compound 21
.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 3
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 4 of 13
View Article Online
DOI: 10.1039/C7MD00106A
ARTICLE
MedChemComm
R1
H2N
H2N
O
O
O
B
N
N
Cl
N
Cl
O
a
b
5
N
N
O
O
8
R1 = Me (9)
R1 = cyclopropyl (10)
R1 = hydroxymethyl (11)
R1 = 3-methyloxetan-3-yl (12)
Me
R1
H2N
H2N
O
O
O
N
N
N
c
R2
N
N
N
O
O
R1 = cyclopropyl (16)
R1 = hydroxymethyl (17)
R1 = 3-methyloxetan-3-yl (18)
R2 = 3,6-dihydro-2H-pyran-4-yl (13)
R2 = 5,6-dihydro-2H-pyran-3-yl (14)
R2 = 2-fluoropyridin-4-yl (15)
Scheme 2 Reagents: (a) B2Pin2, Pd(dppf)Cl2 (cat.), KOAc, dioxane, 91%; (b) 3-
bromo-5-alkyne-pyridines, Pd(AmPhos)2Cl2 (cat.), K3PO4, dioxane/water, 100 °C,
26-58%;
dioxaborolane (for 13
(c)
2-(5,6-dihydro-2H-pyran-3-yl)-4,4,5,5-tetramethyl-1,3,2-
16 17 and 18), 2-(3,6-dihydro-2H-pyran-4-yl)-4,4,5,5-
,
,
tetramethyl-1,3,2-dioxaborolane (for 14), or (2-fluoropyridin-4-yl)boronic acid
(for 15) with catalyst system A (Pd(PPh3)4 (cat.) , K2CO3, dioxane/water, 120 °C for
synthesis of 13 and 15) or catalyst system B (Pd(AmPhos)2Cl2 (cat.), K3PO4,
dioxane/water, 100 °C for synthesis of 14
,
16
-
18).
Scheme 3 Reagents: (a) (i) NaN3, CuI (cat.), trans-N,N’-dimethylcyclohexane-1,2-
diamine, sodium ascorbate, EtOH/water, 90 °C, (ii) NaBH4, 80%; (b) benzoic
anhydride, 90%; (c) Pd2(dba)3 (cat.), CyJohnPhos, LiHMDS, 69%; (d) (i) EDCI,
HOBt, 5-chloropicolinic acid; (ii) NaOH; 36% overall yield. (e) trimethyl borate
(for 23), cyclopropyl boronic acid (for 24) or corresponding pinacol boranes (for
25 and 26), Pd(AmPhos)2Cl2 (cat.), K3PO4, dioxane/water, 110 °C, 69-99%; (f) 5-
chloropicolinic acid, DMTMM, 26-65%. (g) (i) NaOMe, 55%; (ii) Pd2(dba)3 (cat.),
Alternatively, the unprotected amino-substituted 3-aza-AOX
22 was obtained from the unprotected bromide via a Cu-
catalyzed azide coupling, followed by reduction with sodium
borohydride.58 Precursor 22 was elaborated to intermediates CyJohnPhos, LiHMDS, 71%; (iii) NaOH, 90%;
23-26 through Suzuki coupling without any additional
5
protecting of either the warhead or the aminophenyl group. 4-
(4,6-Dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholinium
Biological Activity
chloride (DMTMM)59,
chloropicolinic acid with intermediates 23-26 giving
compounds 27 30. Analog 32 was synthesized by nucleophilic
promoted the coupling of 5-
60
As previously discussed, our strategy was to initially minimize
MW by reducing the size of the P2’-substituent on the 3-aza-
AOX core while fully exploiting CatD and BACE1 protein
residue differences in the S3 sub-pocket. To this end P2’-Cl-
-
aromatic substitution of chloro-pyridine 19 with sodium
methoxide, introduction of an amino-group via LiHMDS
coupling, warhead deprotection, and amide coupling with 5-
chloropicolinic acid.
substituted compound 6, a truncated version of compound 1,
was chosen as a reference compound to monitor progress
(Table 1). Despite replacing the 2-fluoro-4-pyridinyl substituent
with a chlorine, compound
6 retained high potency in the
BACE1 enzymatic assay (IC50 = 72 nM) and moderate potency
in the cell assay (IC50 = 830 nM). The cellular potency of the
compounds was assessed by measuring the inhibition of Aβ40
production in HEK293 cells expressing APP. The CatD/BACE1
selectivity of
enzymatic CatD potency. Based on the co-crystal structures
shown in Figure 2a, the P3 group (2-fluoropyridin-3-yl) in was
6 was calculated to be 210x based on measured
6
replaced by a structural isomer. It was predicted that a 5-
fluoro-3-pyridyl P3 group would position the fluorine in a
trajectory towards the S3 sub-pocket. The affinity of the 5-
fluoro-3-pyridyl analog
7
towards CatD decreased by 8.5x-fold
, while the affinity towards BACE1
compared to compound
6
dropped only by 4.5x-fold (Table 1). This result suggested that
4 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 5 of 13
View Article Online
DOI: 10.1039/C7MD00106A
MedChemComm
ARTICLE
the fluorine in 5-fluoro-3-pyridyl was a potential vector into We replaced the chlorine in compound
9 with P2’ substituents
the S3 sub-pocket which could be further exploited. As a that were successfully used in former lead AOX-BACE1
23, 56
follow-up, a 5-propynyl-3-pyridyl analog
further probe this vector.39, 40 Compound
9
9
was prepared to inhibitors.15,
Three alkyne analogs 13-15 with these
showed improved privileged P2’ substituents were tested in the BACE1 enzymatic
BACE1 enzyme (13 nM) and cell potency (120 nM), as well as and cellular assays and showed picomolar BACE1 affinity with
increased CatD/BACE1 selectivity ( 5,600x). Additionally, the low cell shift (Table 2). High passive permeability and low P-gp
physicochemical properties of were in an acceptable range efflux ratios were maintained for all three analogs. Compound
∼
9
for CNS penetration (MW = 402 g/mol, clogP = 3.0, PSA = 83 15 was metabolically stable in rat and human liver microsomal
Å2).36
studies, whereas increased microsomal turnover was observed
for the dihydropyrans 13 and 14 Importantly, high
.
CatD/BACE1 selectivity was maintained for all three
Table 1 Different trajectories to access the S3 sub-pocket
compounds, with the 5,6-dihydro-2H-pyran-3-yl analog 13
displaying the highest CatD/BACE1 selectivity (∼19,000x). A x-
R1
=
ray co-crystal structure of compound 13 bound to BACE1 was
obtained to confirm the hypothesized orientation of the
propyne into the S3 sub-pocket (Fig. 4, PDB 5UYU). As
observed in previously reported AOX-BACE1 x-ray co-crystal
structure, the aminooxazoline moiety interacts with the two
catalytic aspartic acids (Asp228, Asp32) and Tyr 71 engages in
R2
=
Cl
Cl
Cl
Cl
Cmd #
BACE1
IC50 [µM]a
1
6
7
9
21
0.00092
0.66
0.072
0.32
0.013
0.021
the
π-stacking interaction with the 3-aza-AOX core. The
nitrogen of the 3-aza-AOX core nitrogen forms a hydrogen
bond interaction with Trp 76 through a water bridge and the
dihydropyran unit resides in the S2’ pocket of the protein
where the oxygen atom is within hydrogen bond distance to
Arg128. As proposed, the propynyl group occupies the S3 sub-
pocket and engages in a lipophilic interaction with Ala335. The
high CatD selectivity could be attributed to the residue
difference in the S3 sub-pocket as highlighted in the CatD
protein overlay in Figure 4b (Ala335 in BACE1 vs. Asp318 in
CatD): whereas the methyl terminus of the propynyl moiety
CatD
15
130
4.7
72
1,100
0.19
IC50 [µM]a
HEK
IC50 [µM]b
Selectivityc
CatD/BACE1
0.021
0.83
0.12
660
443
210
383
400
383
5,600
402
53,000
442
MW [g/mol]
clogPd/LipEe
PSA [Å2]/#HBD
2.6/6.5
95/1
2.5/4.7
83/1
2.5/4.0
83/1
3.0/4.9
83/1
2.8/4.9
112/2
aIC50 values were averaged values determined by at least two independent
experiments. bHuman embryonic kidney cells. cSelectivity ratio [IC50 CatD/IC50
BACE1]. dclogP values were calculated using ACD LogD Suite (version 10). eLipE = -
log10(BACE1 IC50) – cLogP.
engages in a van der Waals interaction (3.8 Å) with the methyl
group of Ala335; structural alignment of the CatD protein
clearly demonstrates the repulsive steric interaction which
would ensue with the Asp318 side chain of CatD.
As an alternative approach, we replaced the 5-substituted-3-
pyridyl P3 group with an elongated picolinamide, which has
been a privileged moiety heavily utilized in several previously
disclosed BACE1 inhibitor designs and initially described by
Shionogi.17, 19, 31, 32, 41, 46, 48-52 The 5-chloropicolinamide was well
tolerated on the 3-aza-AOX core with compound 21 showing
high enzymatic BACE1 activity (21 nM), a 10x cell shift (190
Additional alkyne analogs (Table 2, 16-18) were investigated in
an attempt to improve the microsomal stability of 13 and
explore the impact on CatD/BACE1 selectivity. A larger cell
shift and moderate CatD/BACE1 selectivity was observed for
the cyclopropylethynyl analog 16 and the 3-methyl-3-oxetan-
ethynyl analog 18. In addition, the hydroxymethyl analog 17
demonstrated high P-gp mediated efflux. An attempt to mask
the hydroxy group in form of a 3-methyl-oxetan-3-yl group
nM) and very high CatD/BACE1 selectivity (∼53,000x). Of
particular note, the absolute CatD IC50 value exceeded 1 mM.61
The superior CatD/BACE1 selectivity obtained with the 5-
chloropicolinamide-substituted compound 21 (Table 1) came
(
18) resulted in reduced P-gp efflux.13 However, all three
analogs 16-18 suffered from high in vitro metabolism.
We continued our investigation on the deep-P3 amide-
substituted compound 21 in order to improve cell potency
(Table 1 and 3). Cognizant of the physicochemical properties,
small substituents were chosen as alternative P2’ groups
at the cost of increased MW, PSA (442 g/mol, and 112 Å2
,
respectively) and the addition of another hydrogen bond
donor. In order to assess if CNS penetration, and ultimately in
vivo Aβ40 lowering was attainable for these 3-aza-AOX BACE1
inhibitors, data on permeability, P-gp transporter recognition
and microsomal stability was collected. The results shown in
(Table 3, 27
28) P2’ groups not only improved BACE1 enzyme and cell
potency (52 nM and 13 nM, respectively) but also maintained
the high CatD/BACE1 selectivity relative to chloro analog 21
, 28 and 32). The methoxy (32) and the cyclopropyl
(
Table 2 (entry 1, 9) and Table 3 (entry 1, 21) confirmed that
.
both compounds were highly permeable, had low efflux ratios
and low turnover in human and rat liver microsomes. These
results warranted further work on both types of deep P3
groups with the goal of improving BACE1 potency without
eroding selectivity towards CatD.
Both compounds were highly permeable with favourable P-gp
efflux ratios and were stable in human and rat liver
microsomes. Replacement of the chlorine by a methyl group
(
27) increased P-gp efflux ratios. To further test the limits of
the CNS drug-like space as it relates to the deep-P3 amide-
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 5
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 6 of 13
View Article Online
DOI: 10.1039/C7MD00106A
ARTICLE
MedChemComm
substituted 3-aza-AOX scaffold, the occupancy of S2’ pocket molecular weight (MW = 490 g/mol for 29 and 30), both
was maximized via implementation of dihydropyran P2’ analogs remained highly permeable. The increase in PSA (121
substituents (Table 3, 29 and 30). The additional contact with Å2 for 29 and 30) led only in the case of compound 29 to
the BACE1 protein due to the P2’ substituents led to BACE1 increased recognition by P-gp efflux transporter (h/rER =
affinities in the picomolar range. A low cell shift (2-4 fold) was 6.5/8.5). Elevated turnover in microsomes was measured for
maintained and CatD selectivity values were extremely high, both dihydropyran analogs 29 and 30
ranging from 770,000x to 1,800,000x. Despite the high
.
∼
∼
Table 2 P2’ SAR of deep-P3 alkyne substituted 3-aza-AOX BACE1 inhibitors
Microsomal
f
P-gp efflux
Ratioe
stability Clint
IC50 [µM]a
Selectivityc
[(uL/min)/mg]
Pappd
Cpd
9
R1
R2
BACE1
HEKb
CatD
CatD/BACE1
[nm/s]
rER
hER
RLM
HLM
Me
0.013
0.12
72
5,600
22
18
19
15
13
1.7
2.1
1.6
1.8
1.4
2.8
21
27
13
14
15
16
Me
Me
Me
0.00031
0.00046
0.00043
0.0018
0.0012
0.0022
0.0041
0.087
5.9
2.7
4.5
3.6
19,000
5,900
3.1
1.9
2.4
2.3
63
45
20
23
77
76
40
80
10,000
2,100
17
18
0.00058
0.0055
0.0024
0.10
4.4
12
7,500
2,200
15
18
44
-
50
46
73
84
5.6
420
aIC50 values were averaged values determined by at least two independent experiments. bHuman embryonic kidney cells. cselectivity ratio [IC50 CatD/IC50 BACE1].
dApparent permeability measured in parental LLC-PK1 cells. Values are an average of apical to basolateral (A to B) and basolateral to apical (B to A) velocities and are
f
reported as nm/s. eEfflux measured in LLC-PK1 cells transfected with human MDR1 and are reported as a ratio of (B to A)/(A to B). Rat liver microsomal (RLM) and
human liver microsomal (HLM) clearance reported as apparent intrinsic clearance with no correction for binding and not scaled. Compound concentration = 1 μM.
Microsomal protein concentration = 0.25 mg/mL.
Fig 4. (a) Co-crystal structure of human BACE1 + compound 13 (turquoise, PDB 5UYU). The mobile flap region is shown in orange. Key hydrogen bonds are depicted
as black dashed lines. (b) Co-crystal structure of human BACE1 (blue) + compound 13 (turquoise) overlaid with rat catD (pink). Key residues are labelled in yellow for
BACE1 and pink for CatD.
On the basis of the overall in vitro profiles and lipophilic (SD) rats were administered a single low oral dose (10 mg/kg)
efficiency (LipE) values (Table 3), the PD response of of the corresponding test compound and the Aβ40 levels in CSF
compounds 28 29 and 30 was assessed. Naïve Sprague Dawley and brain as well as exposure levels in plasma and CSF were
,
6 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 7 of 13
View Article Online
DOI: 10.1039/C7MD00106A
MedChemComm
measured after 4 h (Table 4). While all three compounds to IC50 values to generate robust A
ARTICLE
β
40 reduction.15 Compound
showed comparable reduction (> 58%) of Aβ40 in the CSF 30 had the lowest estimated unbound plasma drug
compartment after a 10 mg/kg single oral dose, various concentration corresponding to 50% A 40 reduction in the CSF
β
degrees of Aβ40 reduction in the brain were obtained. The (EC50,unbound = 5 nM) and in the brain (EC50,unbound = 11 nM) and
compounds were rank-ordered using estimated unbound was subsequently evaluated in a time-course PD study in rats
plasma concentration to achieve 50% Aβ40 lowering in the (Figure 5). Aβ40 levels in the CSF and brain as well as drug
brain from the single-point experiment.62 The estimated plasma levels were measured at multiple timepoints in the
unbound plasma drug concentration corresponding to 50% same study. The Aβ40 reduction in the CSF and brain
40 reduction in the brain (EC50,unbound) for compound 29 was compartments of rats after a single 30 mg/kg oral dose
A
β
12-fold shifted relative to the cell IC50 consistent with the reached its maximum after 4 hours and returned to baseline
previous observation that BACE1 inhibitors with large efflux after 24 h.
ratios require higher unbound plasma concentrations relative
Table 3 P2’ SAR of deep-P3 picolinamide substituted 3-aza-AOX BACE1 inhibitors
Microsomal
stability CLint
f
P-gp efflux
Ratioe
IC50 [µM]a
Selectivityc
CatD/
[(uL/min)/mg]
Cpd
21
R
BACE1
HEKb
CatD
BACE1
Pappd [nm/s]
hER
rER
RLM
HLM
LipEg
Cl
0.021
0.19
1,100
53,000
160,000
20,000
28
2.8
2.0
7.8
2.6
6.5
3.4
2.9
<14
<14
<14
<14
79
37
4.9
32
27
28
29
30
OMe
Me
0.0075
0.031
0.052
0.054
1,200
640
25
40
29
27
29
3.1
8.0
2.5
8.5
3.3
<14
16
5.4
5.1
5.5
7.2
7.0
0.0036
0.00057
0.00062
0.013
310
85,000
<14
80
0.0025
0.0014
1,000
480
1,800,000
770,000
84
48
aIC50 values were averaged values determined by at least two independent experiments. bHuman embryonic kidney cells. cselectivity ratio [IC50 CatD/IC50 BACE1].
dApparent permeability measured in parental LLC-PK1 cells. Values are an average of apical to basolateral (A to B) and basolateral to apical (B to A) velocities and are
f
reported as nm/s. eEfflux measured in LLC-PK1 cells transfected with human MDR1 and are reported as a ratio of (B to A)/(A to B). Rat liver microsomal (RLM) and
human liver microsomal (HLM) clearance reported as apparent intrinsic clearance with no correction for binding and not scaled. Compound concentration = 1 μM.
Microsomal protein concentration = 0.25 mg/mL. gLipE = -log10(BACE1 IC50) – cLogP(calculated)
Table 4 Reduction of Aβ40 levels in CSF and brain in wild-type Sprague-Dawley rats and additional profiling
Aβ40 Reduction [%]b
Estimated EC50,unbound [µM]c
Shift
Cpda
1
CSF
74
66
58
62
Brain
65
CSF
Brain
0.017
0.021
0.030
0.011
Cplasma,unbound [µM]d
0.032
CCSF[µM]e
0.06
Brain EC50 /Cell IC50
pKa
5.6f
NAg
NAg
6.1h
0.011
0.012
0.011
0.0049
0.81
1.5
12
28
29
30
52
0.022
0.02
25
0.015
0.005
0.003
41
0.0078
7.9
aCompounds were dosed orally 10 mg/kg as a solution (2% HMPC, 1% Tween80) to male Sprague-Dawley rats. n = 5 animals per group. bCSF and brain Aβ40 measured
c
at t = 4 h. Estimated unbound plasma EC50 for CSF or brain effect from a single dose experiment at 4h timepoint or unbound plasma concentration corresponding to
f
50% Aβ40 in CSF or brain, respectively. dUnbound Plasma drug concentration measured at t = 4 h. eCSF drug concentration measured at t = 4 h. experimental pKa
determined by potentiometry at T = 25 °C. gNA = not available. hexperimental pKa determined by UV spectrophotometry at T = 25 °C.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 7
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 8 of 13
View Article Online
DOI: 10.1039/C7MD00106A
ARTICLE
MedChemComm
An indirect response model was used to model the PK-PD 3-aza-xanthene backbone and the P2’ substituent is conserved
relationship which accounts for the formation and clearance of compared to AOX-BACE1 inhibitors absent of a deep-P3
Aβ
40 in the CSF.63 Using this model in vivo unbound plasma IC50 substituent. Figure 5 highlights the hydrogen-bonding between
values for CSF and brain effect were generated (15 nM and 17 the NH of the amide and the backbone carbonyl of Gly230 as
nM, respectively). These values were in good agreement with well as the near co-planarity of the amide linker between the
the estimated unbound plasma EC50 values obtained in the 3-aza-AOX core and the 2-pyridyl. The picolinamide extends
single dose, single time point PD study (5 nM and 11 nM, into the S3 sub-pocket and the chlorine in para-position
respectively).
engages in a repulsive steric interaction with the Asp318
resulting in the very high CatD selectivity of this compound.
When tested against a panel of other aspartyl proteases,
compound 30 showed high selectivity for BACE1 over
Cathepsin E, pepsin and renin, but no selectivity over
homologue BACE2.
Fig 6. Co-crystal structure of human BACE1 (blue) + compound 30 (green, PDB
5I3W) overlaid with CatD (pink). Key residues are labeled in yellow for BACE1
and pink for CatD.
Fig 5. Time dependent Aβ40 reduction in CSF and brain in rat after 30 mg/kg oral
dose of 30 and total plasma concentrations of compound 30 measured in the
same time-course study.
Compound 30 demonstrated a significant pharmacodynamic
The pharmacokinetic (PK) profile of 30 was evaluated in rat response and an acceptable PK profile in rat and dog. To
and dog (Table 5). Consistent with the in vitro rat microsomal further evaluate the likelihood for retinal toxicity of compound
stability data, 30 showed moderate clearance in rat. Moderate 30, a 4-day repeat dose toxicology study was conducted in
clearance and similar Vdss was also maintained in dog. Taking male Sprague Dawley rats.66 Animals were administered daily
into account the different protein binding of 30 in rat versus oral doses of vehicle, 100, 300 and 900 mg/kg (n=3/dose level)
dog, unbound clearance was lower in dog (Clunbound = 15.6 of compound 30 for 4 days and necropsied 24 hours after the
L/h/kg) compared to rat (Clunbound
= 80 L/h/kg). The last dose. After the last dose, the mean exposure (area under
bioavailability was calculated to be moderate in both species the curve; AUC0-24) was 985 µg•h/mL at 900 mg/kg. No test-
(%F = 30).
article related increases in AFG in the RPE were observed at
any dose level when evaluated by fluorescence microscopy.12
Thus, at high multiples (127X) over the exposure necessary for
50% Aβ40 reduction in the CSF (AUC0-24 = 7.72 µg•h/mL based
on estimated total plasma drug concentration corresponding
to 50% Aβ40 reduction in the CSF EC50,total = 0.658 nM) no
microscopic changes in the retina were observed for 30 after 4
days. In contrast, we observed accumulation of AFG in the RPE
after only 4 days after a daily oral 30 mg/kg dose of compound
Table 5 Pharmacokinetic profiles and plasma protein binding of 30
iva
pob
t1/2
Cl
Vdss
[L/kg]
Cmax
[µM]
%F
PPB
Fu
Species
Rat
[L/h/kg]
[h]
0.61
1.2
1.3
0.7
0.7
2.4
28
30
0.0076
0.023
Dog
0.36
4.5
1
, which corresponded to a 10X multiple.67
aRat: 2 mg/kg dose as a solution in DMSO; Dog: 1 mg/kg dose as a solution in 25%
captisol/75% water at pH = 4; b5 mg/kg dose (rat) or 2 mg/kg dose (dog) as a
solution in 1% Tween80/2% HMPC/97% water at pH = 2 (rat) or pH = 5 (dog).
During the evolution of AOX-BACE1 inhibitors great attention
has been paid to the activity at the hERG cardiac ion
channel.15, 56 Consequently, the cardiovascular (CV) effects of
compound 30 were evaluated. Compound 30 was found to
block hERG-mediated current in vitro (IC50 = 0.9 µM). The risk
A x-ray co-crystal structure of 30 in BACE1 overlaid with CatD
protein is shown in Figure 6 (PDB 5I3W).64, 65 Every interaction
of the BACE1 protein with the amino-oxazoline-warhead, the
8 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 9 of 13
View Article Online
DOI: 10.1039/C7MD00106A
MedChemComm
ARTICLE
of QTc prolongation was further assessed in an anesthetized
dog cardiovascular model. Two male Beagle dogs were
anesthetized and instrumented for cardiovascular monitoring
before and after intravenous infusion of 30 at doses of 3.0, 7.5
and 25 mg/kg (30 min per dose). A blood sample was taken
every 10 min at the end of each 30 min infusion to determine
plasma drug concentration. After each dose of 30, the average
unbound plasma concentration was 0.11 µM, 0.35 µM and
0.85 µM, respectively. These drug levels corresponded to
unbound plasma concentrations of 9x, 25x and 59x fold over
the unbound BACE IC50 in the rat as determined by the indirect
response model. Based on exposure-QTc prolongation
response analysis (18 total PK-QT points; 9 per dog), plasma
Notes and references
1.
E. Karran, M. Mercken and B. De Strooper, Nature
Reviews Drug Discovery, 2011, 10, 698-712.
2.
K. M. Munro, A. Nash, M. Pigoni, S. F. Lichtenthaler and J.
M. Gunnersen, Journal of Molecular Neuroscience, 2016,
60, 305-315.
3.
T. Jonsson, J. K. Atwal, S. Steinberg, J. Snaedal, P. V.
Jonsson, S. Bjornsson, H. Stefansson, P. Sulem, D.
Gudbjartsson, J. Maloney, K. Hoyte, A. Gustafson, Y. Liu, Y.
Lu, T. Bhangale, R. R. Graham, J. Huttenlocher, G.
Bjornsdottir, O. A. Andreassen, E. G. Joensson, A. Palotie,
T. W. Behrens, O. T. Magnusson, A. Kong, U.
Thorsteinsdottir, R. J. Watts and K. Stefansson, Nature
(London, United Kingdom), 2012, 488, 96-99.
drug levels
>
0.3 µM (unbound) caused QTc interval
10% relative to baseline) in a concentration-
4.
J. Yuan, S. Venkatraman, Y. Zheng, B. M. McKeever, L. W.
Dillard and S. B. Singh, Journal of Medicinal Chemistry,
2013, 56, 4156-4180.
prolongation (
≥
dependent manner. We concluded, that the QTc prolongation
observed for 30 in the dog was consistent with its hERG
channel inhibition, and decided not to further pursue
compound 30 due to the low cardiovascular safety margin.
5.
6.
7.
D. Oehlrich, H. Prokopcova and H. J. M. Gijsen, Bioorganic
& Medicinal Chemistry Letters, 2014, 24, 2033-2045.
A. K. Ghosh and H. L. Osswald, Chemical Society Reviews,
2014, 43, 6765-6813.
Y. Lu, D. Riddell, E. Hajos-Korcsok, K. Bales, K. M. Wood, C.
E. Nolan, A. E. Robshaw, L. Zhang, L. Leung, S. L. Becker, E.
Tseng, J. Barricklow, E. H. Miller, S. Osgood, B. T. O'Neill,
M. A. Brodney, D. S. Johnson and M. Pettersson, Journal
of Pharmacology and Experimental Therapeutics, 2012,
342, 366-375.
Conclusions
In summary, a series of highly selective 2-aminooxazoline 3-
aza-xanthene BACE1 inhibitors versus CatD were synthesized
by installing substituents reaching into the S3 sub-pocket of
the BACE1 protein. A combined structure- and property-based
approach led to the identification of picolinamide-substituted
3-aza-AOX BACE1 inhibitors with suitable in vitro profiles,
warranting further in vivo studies. Despite pushing the limits of
physicochemical properties as they relate to CNS drug-
likeness, the picolinamide-substituted 3-aza-AOX BACE1
8.
P. C. May, R. A. Dean, S. L. Lowe, F. Martenyi, S. M.
Sheehan, L. N. Boggs, S. A. Monk, B. M. Mathes, D. J.
Mergott, B. M. Watson, S. L. Stout, D. E. Timm, E. S.
LaBell, C. R. Gonzales, M. Nakano, S. S. Jhee, M. Yen, L.
Ereshefsky, T. D. Lindstrom, D. O. Calligaro, P. J. Cocke, D.
G. Hall, S. Friedrich, M. Citron and J. E. Audia, Journal of
Neuroscience, 2011, 31, 16507-16516.
inhibitors showed robust reduction of brain and CSF Aβ. The
9.
R. Vassar, Alzheimer's Research & Therapy, 2014, 6,
89/81-89/14, 14 pp.
highly BACE1 over CatD selective compound 30 was further
evaluated in a 4-day repeat toxicology study in rats and did not
show any retinal effects, including AFG in the RPE, at doses up
to 900 mg/kg. Despite our efforts culminating in the discovery
10.
11.
C. Sheridan, Nature Biotechnology, 2015, 33, 115-116.
E. Kennedy Matthew, X. Chen, A. Hodgson Robert, A.
Hyde Lynn, R. Kuvelkar, M. Parker Eric, W. Stamford
Andrew, N. Cumming Jared, W. Li, D. Scott Jack, K. Cox, F.
Dockendorf Marissa, J. Kleijn Huub, H. Mei, A. Stone Julie,
M. Egan, L. Ereshefsky, S. Jhee, A. Mattson Britta, J.
Palcza, M. Tanen, D. Troyer Matthew, L. Tseng Jack and S.
Forman Mark, Science translational medicine, 2016, 8,
363ra150.
of a picolinamide-substituted 3-aza-AOX-BACE1 inhibitor (30
)
with significantly greater selectivity over CatD, which appeared
to have reduced the likelihood of retinal toxicity, a low safety
margin for the elongation of the QTc interval in the dog
cardiovascular model prevented the further assessment of
compound 30
.
12.
13.
M. R. Fielden, J. Werner, J. A. Jamison, A. Coppi, D.
Hickman, R. T. Dunn, II, E. Trueblood, L. Zhou, C. A. Afshari
and R. Lightfoot-Dunn, Toxicologic Pathology, 2015, 43,
581-592.
T. A. Dineen, K. Chen, A. C. Cheng, K. Derakhchan, O.
Epstein, J. Esmay, D. Hickman, C. E. Kreiman, I. E. Marx, R.
C. Wahl, P. H. Wen, M. M. Weiss, D. A. Whittington, S.
Wood, R. T. Fremeau, R. D. White and V. F. Patel, Journal
of Medicinal Chemistry, 2014, 57, 9811-9831.
Acknowledgements
The authors thank Jennifer Allen for helpful discussions and
editing of the manuscript, Kyung-Hyun Gahm, Wes Barnhart,
Samuel Thomas, and Larry Miller for chiral separation; Patricia
Lopez for scale-up of compound 30; Alexander Long for
purification of the BACE protein and Opas Nuanmanee for
purification of the CatD protein; Steven Louie, Joel Esmay,
Leeanne Zalameda and Yi Luo for performing in vitro screening
of compounds; Paul H. Wen and Li Zhu for performing in vivo
studies. The Advanced Light Source is supported by the U.S.
Department of Energy under Contract No. DE-AC02-
05CH11231.
14.
15.
S. Khoh-Reiter, S. A. Sokolowski, B. Jessen, M. Evans, D.
Dalvie and S. Lu, Toxicological Sciences, 2015, 145, 383-
395.
J. J. Chen, Q. Liu, C. Yuan, V. Gore, P. Lopez, V. Ma, A.
Amegadzie, W. Qian, T. C. Judd, A. E. Minatti, J. Brown, Y.
Cheng, M. Xue, W. Zhong, T. A. Dineen, O. Epstein, J.
Human, C. Kreiman, I. Marx, M. M. Weiss, S. A. Hitchcock,
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 9
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 10 of 13
View Article Online
DOI: 10.1039/C7MD00106A
ARTICLE
MedChemComm
T. S. Powers, K. Chen, P. H. Wen, D. A. Whittington, A. C.
Cheng, M. D. Bartberger, D. Hickman, J. A. Werner, H. M.
Vargas, N. E. Everds, S. L. Vonderfecht, R. T. Dunn, II, S. 27.
Wood, R. T. Fremeau, Jr., R. D. White and V. F. Patel,
Bioorganic & Medicinal Chemistry Letters, 2015, 25, 767-
774.
A. M. Zuhl, C. E. Nolan, M. A. Brodney, S. Niessen, K.
Atchison, C. Houle, D. A. Karanian, C. Ambroise, J. W.
Brulet, E. M. Beck, S. D. Doran, B. T. O'Neill, C. W. am 28.
Ende, C. Chang, K. F. Geoghegan, G. M. West, J. C. Judkins,
X. Hou, D. R. Riddell and D. S. Johnson, Nature
Communications, 2016, 7, 13042.
P. C. May, B. A. Willis, S. L. Lowe, R. A. Dean, S. A. Monk,
P. J. Cocke, J. E. Audia, L. N. Boggs, A. R. Borders, R. A.
Brier, D. O. Calligaro, T. A. Day, L. Ereshefsky, J. A. Erickso,
H. Gevorkyan, C. R. Gonzales, D. E. James, S. S. Jhee, S. F. 29.
Komjathy, L. Li, T. D. Lindstrom, B. M. Mathe, F. Martenyi,
S. M. Sheehan, S. L. Stout, D. E. Timm, G. M. Vaught, B. M. 30.
Watson, L. L. Winneroski, Z. Yang and D. J. Mergott,
those binding in the S2' pocket are named
P2'substituents.
S. J. Stachel, C. A. Coburn, D. Rush, K. L. G. Jones, H. Zhu,
H. Rajapakse, S. L. Graham, A. Simon, M. Katharine
Holloway, T. J. Allison, S. K. Munshi, A. S. Espeseth, P.
Zuck, D. Colussi, A. Wolfe, B. L. Pietrak, M.-T. Lai and J. P.
Vacca, Bioorganic & Medicinal Chemistry Letters, 2009,
19, 2977-2980.
H. A. Rajapakse, P. G. Nantermet, H. G. Selnick, J. C.
Barrow, G. B. McGaughey, S. Munshi, S. R. Lindsley, M. B.
Young, P. L. Ngo, M. Katherine Holloway, M.-T. Lai, A. S.
Espeseth, X.-P. Shi, D. Colussi, B. Pietrak, M.-C.
Crouthamel, K. Tugusheva, Q. Huang, M. Xu, A. J. Simon,
L. Kuo, D. J. Hazuda, S. Graham and J. P. Vacca, Bioorganic
& Medicinal Chemistry Letters, 2010, 20, 1885-1889.
C. R. Ellis and J. Shen, Journal of the American Chemical
Society, 2015, 137, 9543-9546.
16.
17.
C. R. Ellis, C.-C. Tsai, X. Hou and J. Shen, Journal of Physical
Chemistry Letters, 2016, 7, 944-949.
Journal of Neuroscience, 2015, 35, 1199-1210, 1112 pp.
C. R. Butler, M. A. Brodney, E. M. Beck, G. Barreiro, C. E.
Nolan, F. Pan, F. Vajdos, K. Parris, A. H. Varghese, C. J.
Helal, R. Lira, S. D. Doran, D. R. Riddell, L. M. Buzon, J. K.
Dutra, L. A. Martinez-Alsina, K. Ogilvie, J. C. Murray, J. M.
Young, K. Atchison, A. Robshaw, C. Gonzales, J. Wang, Y.
31.
T. J. Woltering, W. Wostl, H. Hilpert, M. Rogers-Evans, E.
Pinard, A. Mayweg, M. Gobel, D. W. Banner, J. Benz, M.
Travagli, M. Pollastrini, G. Marconi, E. Gabellieri, W. Guba,
H. Mauser, M. Andreini, H. Jacobsen, E. Power and R.
Narquizian, Bioorganic & Medicinal Chemistry Letters,
2013, 23, 4239-4243.
18.
Zhang and B. T. O'Neill, Journal of Medicinal Chemistry, 32.
2015, 58, 2678-2702.
U. Neumann, H. Rueeger, R. Machauer, S. J. Veenstra, R.
M. Lueoend, M. Tintelnot-Blomley, G. Laue, K. Beltz, B.
Vogg, P. Schmid, W. Frieauff, D. R. Shimshek, M.
H. Hilpert, W. Guba, T. J. Woltering, W. Wostl, E. Pinard,
H. Mauser, A. V. Mayweg, M. Rogers-Evans, R. Humm, D.
Krummenacher, T. Muser, C. Schnider, H. Jacobsen, L.
Ozmen, A. Bergadano, D. W. Banner, R. Hochstrasser, A.
Kuglstatter, P. David-Pierson, H. Fischer, A. Polara and R.
Narquizian, Journal of Medicinal Chemistry, 2013, 56,
3980-3995.
19.
20.
Staufenbiel
and
L.
H.
Jacobson,
Molecular
Neurodegeneration, 2015, 10, 44/41-44/15.
R. Steinfeld, K. Reinhardt, K. Schreiber, M. Hillebrand, R. 33.
Kraetzner, W. Brueck, P. Saftig and J. Gaertner, American
N. Anderson and J. Borlak, FEBS Letters, 2006, 580, 5533-
5540.
Journal of Human Genetics, 2006, 78, 988-998.
Standard deviations for the in vitro CatD enzymatic assay
can be found in the supporting information.
34.
D. A. Price, J. Blagg, L. Jones, N. Greene and T. Wager,
Expert Opinion on Drug Metabolism & Toxicology, 2009, 5,
921-931.
J.-U. Peters, J. Hert, C. Bissantz, A. Hillebrecht, G.
Gerebtzoff, S. Bendels, F. Tillier, J. Migeon, H. Fischer, W.
Guba and M. Kansy, Drug discovery today, 2012, 17, 325-
335.
S. A. Hitchcock and L. D. Pennington, Journal of Medicinal
Chemistry, 2006, 49, 7559-7583.
T. T. Wager, R. Y. Chandrasekaran, X. Hou, M. D.
Troutman, P. R. Verhoest, A. Villalobos and Y. Will, ACS
Chemical Neuroscience, 2010, 1, 420-434.
21.
22.
E. T. Baldwin, T. N. Bhat, S. Gulnik, M. V. Hosur, R. C. 35.
Sowder, R. E. Cachau, J. Collins, A. M. Silva and J. W.
Erickson, Proceedings of the National Academy of Sciences
of the United States of America, 1993, 90, 6796-6800.
Y. Cheng, J. Brown, T. C. Judd, P. Lopez, W. Qian, T. S. 36.
Powers, J. J. Chen, M. D. Bartberger, K. Chen, R. T. Dunn,
O. Epstein, R. T. Fremeau, S. Harried, D. Hickman, S. A. 37.
Hitchcock, Y. Luo, A. E. Minatti, V. F. Patel, H. M. Vargas,
R. C. Wahl, M. M. Weiss, P. H. Wen, R. D. White, D. A.
Whittington, X. M. Zheng and S. Wood, ACS Medicinal 38.
Chemistry Letters, 2015, 6, 210-215.
23.
T. T. Wager, X. Hou, P. R. Verhoest and A. Villalobos, ACS
Chemical Neuroscience, 2010, 1, 435-449.
24.
S. Yonezawa, T. Takahashi, X. J. Wang, R. N. S. Wong, J. A. 39.
Hartsuck and J. Tang, Journal of Biological Chemistry,
1988, 263, 16504-16511.
N. P. Birch and Y. P. Loh, Nucleic Acids Research, 1990, 18,
6445-6446.
In this publication, substituents on the aminooxazoline
xanthene core were named in accordance with the space 40.
they occupy in the catalytic domain of the BACE1 enzyme
as it was determined by X-ray cocrystal structure of
previous AOX BACE1 inhibitors with BACE1 protein. For
example, the substituents binding in the S3 region of the
enzyme are named P3 substituents, those binding in the
S3 sp pocket are named P3 sp or deep P3 substituents,
J. N. Cumming, E. M. Smith, L. Wang, J. Misiaszek, J.
Durkin, J. Pan, U. Iserloh, Y. Wu, Z. Zhu, C. Strickland, J.
Voigt, X. Chen, M. E. Kennedy, R. Kuvelkar, L. A. Hyde, K.
Cox, L. Favreau, M. F. Czarniecki, W. J. Greenlee, B. A.
McKittrick, E. M. Parker and A. W. Stamford, Bioorganic &
Medicinal Chemistry Letters, 2012, 22, 2444-2449.
A. W. Stamford, J. D. Scott, S. W. Li, S. Babu, D. Tadesse, R.
Hunter, Y. Wu, J. Misiaszek, J. N. Cumming, E. J. Gilbert, C.
Huang, B. A. McKittrick, L. Hong, T. Guo, Z. Zhu, C.
Strickland, P. Orth, J. H. Voigt, M. E. Kennedy, X. Chen, R.
Kuvelkar, R. Hodgson, L. A. Hyde, K. Cox, L. Favreau, E. M.
Parker and W. J. Greenlee, ACS Medicinal Chemistry
Letters, 2012, 3, 897-902.
25.
26.
10 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 11 of 13
View Article Online
DOI: 10.1039/C7MD00106A
MedChemComm
ARTICLE
41.
B.-M. Swahn, K. Kolmodin, S. Karlstroem, S. von Berg, P. 53.
J. D. Scott, S. W. Li, A. P. J. Brunskill, X. Chen, K. Cox, J. N.
Cumming, M. Forman, E. J. Gilbert, R. A. Hodgson, L. A.
Hyde, Q. Jiang, U. Iserloh, I. Kazakevich, R. Kuvelkar, H.
Mei, J. Meredith, J. Misiaszek, P. Orth, L. M. Rossiter, M.
Slater, J. Stone, C. O. Strickland, J. H. Voigt, G. Wang, H.
Wang, Y. Wu, W. J. Greenlee, E. M. Parker, M. E. Kennedy
and A. W. Stamford, Journal of Medicinal Chemistry, 2016,
59, 10435-10450.
Soederman, J. Holenz, S. Berg, J. Lindstroem, M.
Sundstroem, D. Turek, J. Kihlstroem, C. Slivo, L.
Andersson, D. Pyring, D. Rotticci, L. Oehberg, A. Kers, K.
Bogar, M. Bergh, L.-L. Olsson, J. Janson, S. Eketjaell, B.
Georgievska, F. Jeppsson and J. Faelting, Journal of
Medicinal Chemistry, 2012, 55, 9346-9361.
S. Eketjaell, J. Janson, F. Jeppsson, A. Svanhagen, K.
Kolmodin, S. Gustavsson, A.-C. Radesaeter, K. Eliason, S. 54.
Briem, P. Appelkvist, C. Niva, A.-L. Berg, S. Karlstroem, B.-
M. Swahn and J. Faelting, Journal of Neuroscience, 2013,
42.
A. S. Guram, X. Wang, E. E. Bunel, M. M. Faul, R. D. Larsen
and M. J. Martinelli, Journal of Organic Chemistry, 2007,
72, 5104-5112.
33, 10075-10084.
55.
Y. Chen, R. White, A. Amegadzie, J. Brown, A. C. Cheng, E.
F. Dimauro, T. Dineen, V. K. Gore, J. B. Human, T. Judd, C.
Kreiman, Q. Liu, P. Lopez, V. V. Ma, I. Marx, A. E. Minatti,
H. N. Nguyen, N. A. Paras, V. F. Patel, W. Qian, M. Weiss,
Q. Xue, X. M. Zheng, W. Zhong, Amgen Inc.
WO2011115938, 2011.
43.
44.
S. Eketjaell, J. Janson, K. Kaspersson, A. Bogstedt, F.
Jeppsson, J. Faelting, S. B. Haeberlein, A. R. Kugler, R. C.
Alexander, G. Cebers and P. Ho, Journal of Alzheimer's
Disease, 2016, 50, 1109-1123.
K. W. Hunt, A. W. Cook, R. J. Watts, C. T. Clark, G. Vigers,
D. Smith, A. T. Metcalf, I. W. Gunawardana, M. Burkard, A.
A. Cox, M. K. Geck Do, D. Dutcher, A. A. Thomas, S. Rana,
N. C. Kallan, R. K. DeLisle, J. P. Rizzi, K. Regal, D. Sammond,
R. Groneberg, M. Siu, H. Purkey, J. P. Lyssikatos, A.
Marlow, X. Liu and T. P. Tang, Journal of Medicinal
Chemistry, 2013, 56, 3379-3403.
Y. Gravenfors, J. Viklund, J. Blid, T. Ginman, S. Karlstroem,
J. Kihlstroem, K. Kolmodin, J. Lindstroem, S. von Berg, F.
von Kieseritzky, K. Bogar, C. Slivo, B.-M. Swahn, L.-L.
Olsson, P. Johansson, S. Eketjaell, J. Faelting, F. Jeppsson,
K. Stroemberg, J. Janson and F. Rahm, Journal of
Medicinal Chemistry, 2012, 55, 9297-9311.
Y.-J. Wu, J. Guernon, R. Rajamani, J. H. Toyn, M. K.
Ahlijanian, C. F. Albright, J. Muckelbauer, C. Chang, D.
Camac, J. E. Macor and L. A. Thompson, Bioorganic &
56.
O. Epstein, M. C. Bryan, A. C. Cheng, K. Derakhchan, T. A.
Dineen, D. Hickman, Z. Hua, J. B. Human, C. Kreiman, I. E.
Marx, M. M. Weiss, R. C. Wahl, P. H. Wen, D. A.
Whittington, S. Wood, X. M. Zheng, R. T. Fremeau, Jr., R.
D. White and V. F. Patel, Journal of Medicinal Chemistry,
2014, 57, 9796-9810.
D.-Y. Lee and J. F. Hartwig, Organic Letters, 2005, 7, 1169-
1172.
J. Andersen, U. Madsen, F. Bjorkling and X. Liang, Synlett,
2005, DOI: 10.1055/s-2005-872248, 2209-2213.
Z. J. Kaminski, P. Paneth and J. Rudzinski, Journal of
Organic Chemistry, 1998, 63, 4248-4255.
A. Falchi, G. Giacomelli, A. Porcheddu and M. Taddei,
Synlett, 2000, DOI: 10.1055/s-2000-6490, 275-277.
As suggested in a previous publication (ref 19), attention
should not be limited to achieveing a high numerical
CatD/BACE1 selectivity ratio but also to the absolute CatD
IC50 value.
An estimate of unbound EC50 for Ab lowering was
calculated from a single dose, single time point (4 h)
experiment assuming a standard sigmoid dose response
(Hill slope of 1). This approach was initially validated by
comparing EC50 values from a dose response experiment
to the single dose, single time point estimate of unbound
EC50. Subsequently, estimated EC50s have been shown to
be a good approximation of the IC50 determined from a
time course experiment using an indirect response PK/PD
model.
M. A. Felmlee, M. E. Morris and D. E. Mager, Methods in
Molecular Biology (New York, NY, United States), 2012,
929, 583-600.
A x-ray co-crystal structure of compound 30 in BACE1
protein has previously been published in a different
context (PDB ID 5I3W). .
J. B. Jordan, D. A. Whittington, M. D. Bartberger, E. A.
Sickmier, K. Chen, Ryan D. White, Y. Cheng and T. Judd,
Journal of Medicinal Chemistry, 2016, 59, 3732-3749.
Male Sprague Dawley rats (Rattus norvegicus,
Crl:CD®(SD); Charles River Laboratories; 10 weeks of age)
were group-housed at an AAALAC, International-
accredited facility. Animals were cared for in accordance
with the Guide for the Care and Use of Laboratory
Animals, 8th Edition. All research protocols were
reviewed and approved by the Amgen Institutional
45.
46.
57.
58.
59.
60.
61.
Medicinal
Chemistry
Letters,
2016,
DOI:
10.1016/j.bmcl.2016.10.055, Ahead of Print.
47.
48.
49.
M. Ciordia, L. Perez-Benito, F. Delgado, A. A. Trabanco and
G. Tresadern, Journal of Chemical Information and
Modeling, 2016, 56, 1856-1871.
62.
M. P. Huestis, W. Liu, M. Volgraf, H. E. Purkey, C. Yu, W.
Wang, D. Smith, G. Vigers, D. Dutcher, K. W. Hunt and M.
Siu, Tetrahedron Letters, 2013, 54, 5802-5807.
Y.-J. Wu, J. Guernon, F. Yang, L. Snyder, J. Shi, A. McClure,
R. Rajamani, H. Park, A. Ng, H. Lewis, C. Y. Chang, D.
Camac, J. H. Toyn, M. K. Ahlijanian, C. F. Albright, J. E.
Macor and L. A. Thompson, ACS Medicinal Chemistry
Letters, 2016, 7, 271-276.
F. J. R. Rombouts, G. Tresadern, O. Delgado, C. Martinez-
Lamenca, M. Van Gool, A. Garcia-Molina, S. A. Alonso de
Diego, D. Oehlrich, H. Prokopcova, J. M. Alonso, N. Austin,
H. Borghys, S. Van Brandt, M. Surkyn, M. De Cleyn, A. Vos,
R. Alexander, G. Macdonald, D. Moechars, H. Gijsen and
A. A. Trabanco, Journal of Medicinal Chemistry, 2015, 58,
8216-8235.
G. Tresadern, F. Delgado, O. Delgado, H. Gijsen, G. J.
Macdonald, D. Moechars, F. Rombouts, R. Alexander, J.
Spurlino, M. Van Gool, J. A. Vega and A. A. Trabanco,
Bioorganic & Medicinal Chemistry Letters, 2011, 21, 7255-
7260.
N. Kobayashi, K. Ueda, N. Itoh, S. Suzuki, G. Sakaguchi, A.
Kato, A. Yukimasa, A. Hori, Y. Koriyama, H. Haraguchi, K.
Yasui, Y. Kanda, Shionogi & Co., Ltd. WO2007049532,
2007
50.
63.
64.
65.
66.
51.
52.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 11
Please do not adjust margins

Please do not adjust margins
MedChemComm
Page 12 of 13
View Article Online
DOI: 10.1039/C7MD00106A
ARTICLE
MedChemComm
Animal Care and Use Committee. Rats were housed in an
individual ventilated caging system on an irradiated
corncob bedding (Envigo Teklad 7097). Lighting in animal
holding rooms was maintained on 12:12 hr light:dark
cycle, and the ambient temperature and humidity range
was at 68°F to 79°F and 30% to 70%, respectively. Animals
had ad libitum access to irradiated pelleted feed (Envigo
Teklad Global Rodent Diet; soy protein free extruded
2020X) and reverse-osmosis chlorinated (0.3 to 0.5 ppm)
water via an automatic watering system. Cages were
changed weekly inside an engineered cage changing
station.
67.
unpublished results.
12 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Page 13 of 13
MedChemComm
View Article Online
DOI: 10.1039/C7MD00106A
Development of 2-Aminooxazoline 3-Azaxanthene β-
Amyloid Cleaving Enzyme (BACE) Inhibitors with
Improved Selectivity Against Cathepsin D
Jonathan D. Low, Michael D. Bartberger, Kui Chen, Yuan Cheng, Mark R. Fielden, Vijay Gore, Dean
Hickman, Qingyian Liu, E. Allen Sickmier, Hugo M. Vargas, Jonathan Werner, Ryan D. White, Douglas A.
Whittington, Stephen Wood, Ana E. Minatti,*
We report the discovery of an orally efficacious 3-aza-aminooxazoline-BACE1 inhibitor
with 1,800,000X fold selectivity against CatD and no retinal effects in an early screening
rat toxicology study.