Discovery of biphenylacetamide-derived inhibitors of BACE1 using de novo structure-based molecular design

Article abstract of DOI:10.1021/jm301127x

β-Secretase (BACE1), the enzyme responsible for the first and rate-limiting step in the production of amyloid-β peptides, is an attractive target for the treatment of Alzheimer's disease. In this study, we report the application of the de novo fragment-based molecular design program SPROUT to the discovery of a series of nonpeptide BACE1 inhibitors based upon a biphenylacetamide scaffold. The binding affinity of molecules based upon this designed molecular scaffold was increased from an initial BACE1 IC₅₀ of 323 μM to 27 μM following the synthesis of a library of optimized ligands whose structures were refined using the recently developed SPROUT-HitOpt software. Although a number of inhibitors were found to exhibit cellular toxicity, one compound in the series was found to have useful BACE1 inhibitory activity in a cellular assay with minimal cellular toxicity. This work demonstrates the power of an in silico fragment-based molecular design approach in the discovery of novel BACE1 inhibitors.

Full text of DOI:10.1021/jm301127x

Article
pubs.acs.org/jmc
Discovery of Biphenylacetamide-Derived Inhibitors of BACE1 Using
de Novo Structure-Based Molecular Design
N. Yi Mok, James Chadwick, Katherine A. B. Kellett, Eva Casas-Arce, Nigel M. Hooper,*
A. Peter Johnson,* and Colin W. G. Fishwick*
School of Chemistry and School of Molecular and Cellular Biology, University of Leeds, Leeds LS2 9JT, U.K.
S
* Supporting Information
ABSTRACT: β-Secretase (BACE1), the enzyme responsible for the first and rate-limiting
step in the production of amyloid-β peptides, is an attractive target for the treatment of
Alzheimer’s disease. In this study, we report the application of the de novo fragment-based
molecular design program SPROUT to the discovery of a series of nonpeptide BACE1
inhibitors based upon a biphenylacetamide scaffold. The binding affinity of molecules
based upon this designed molecular scaffold was increased from an initial BACE1 IC50 of
323 μM to 27 μM following the synthesis of a library of optimized ligands whose structures
were refined using the recently developed SPROUT-HitOpt software. Although a number
of inhibitors were found to exhibit cellular toxicity, one compound in the series was found
to have useful BACE1 inhibitory activity in a cellular assay with minimal cellular toxicity.
This work demonstrates the power of an in silico fragment-based molecular design
approach in the discovery of novel BACE1 inhibitors.
In an analogous fashion to the initial development of HIV
protease inhibitors in the 1990s,10 the development of BACE1
inhibitors has primarily focused on substrate-based polypep-
tides with a noncleavable transition-state isostere replacing the
scissile amide bond. Because of the relatively complex structure
of the substrate and the hydrophobic nature of the BACE1
active site, the majority of the early BACE1 inhibitors were
characterized by high molecular weights and complex motifs
that are lacking in drug-like properties.11 More recently, the
application of high-throughput screening approaches has led to
INTRODUCTION
■
Alzheimer’s disease (AD) is the most common neuro-
degenerative disease which currently affects more than 30
million people worldwide.1 The number of AD patients is
expected to increase significantly with the aging human
population. While current approved treatments using acetyl-
cholinesterase inhibitors and N-methyl-D-aspartate (NMDA)
receptor antagonists provide symptomatic relief for AD
patients, there is a lack of effective therapy targeting the
underlying pathophysiology of AD.2
The accumulation of amyloid-β peptide (Aβ) to form the
extracellular senile or amyloid plaques is widely accepted to
play a central role in the pathogenesis of AD.3 According to the
amyloid cascade hypothesis,4 Aβ is generated through the
sequential proteolytic cleavage of the amyloid precursor protein
(APP) catalyzed by the β- and γ-secretases, respectively.5
Because β-secretase is responsible for the first and rate-limiting
step in the formation of Aβ, it represents an attractive
therapeutic target to halt the progression of this debilitating
disease.
β-Site APP cleaving enzyme 1 (BACE1) is responsible for β-
secretase activity in vivo.6 A type I transmembrane protein,
BACE1 belongs to the family of pepsin-like human aspartyl
proteases, characterized by a single transmembrane domain.
With the catalytic site located on the lumenal side of the
membrane, BACE1 processes the substrate APP at Met671−
Asp672 (APP770 numbering) to release soluble sAPPβ. The
membrane-bound fragment C99 is then processed by γ-
secretase to generate the amyloidogenic species Aβ.3,4,7 Since
the first X-ray crystal structures of BACE1 were reported in
1999,6,8,9 numerous efforts have been made in developing novel
and potent inhibitors of BACE1.
the identification of nonpeptide BACE1 inhibitors.12−14
A
complementary strategy for BACE1 inhibitor discovery is the
application of computer-aided methods to design small-
molecule nonpeptidic inhibitors using X-ray crystallographic
structures of BACE1.
Computational structure-based molecular design is a strong
research tool in medicinal chemistry and chemical biology, and
such methods are gaining popularity both in early stage hit
discovery and in establishing structure−activity relationships
(SAR) during lead optimization stages. The utility of one such
approach, virtual high-throughput screening, in the discovery of
BACE1 inhibitors has recently been demonstrated.15−18
A
powerful alternative ligand identification strategy is the use of in
silico fragment-based molecular design which, to our knowl-
edge, has not been applied to the discovery of BACE1
inhibitors. In this approach, using an X-ray crystal structure of
the target enzyme, new potential ligand structures are
constructed de novo via the precise docking of molecular
fragments into chosen regions of the targeted site. These
Received: August 1, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/jm301127x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 1. De novo design of biphenylacetamide inhibitor 1 in the BACE1 binding pocket (PDB 1M4H). (a) Fragments docked into selected
individual target sites, (b) fragments joined to yield complete ligand scaffold, (c) proposed binding mode of 1 (magenta) with key residues labeled
(orange), (d) proposed binding mode of 1 with surface of the binding pocket.
fragments are then joined together in ways dictated by the user
to yield synthetically approachable ligand scaffolds which are
predicted to show good affinity for the target enzyme.19−21
Here, we present the studies of a new BACE1 inhibitor scaffold
designed using the de novo molecular design program
SPROUT.22,23 A library of structural analogues of the initial
designed skeleton was synthesized and tested against
recombinant BACE1. SAR analysis and ligand optimization of
this library of compounds were achieved using hit optimization
software SPROUT-HitOpt.
RESULTS AND DISCUSSION
■
Figure 2. Schematic showing the proposed interactions of inhibitor 1
(shown in ionic form) with BACE1.
De Novo Ligand Design. We have previously described
the application of the de novo molecular design program
SPROUT to the production of a number of enzyme inhibitors
and receptor antagonists.19−21 To aid the generation of potent
BACE1 inhibitors using SPROUT, the crystal structure of
BACE1 cocrystallized with inhibitor OM00−3 reported by
binding of the substrate APP by blocking its entry into the
binding pocket. During planning for the synthesis of molecules
corresponding to the designed ligand scaffold, we decided, for
simplicity, to place methylene units at the three points linking
the fragments (marked “X”, “Y”, and “Z” in Figure 1b). It
should be noted that other substituents at the “Z” position of
the designed ligand scaffold, including CO (producing an
amide) and SO2 (producing a sulfonamide), were considered as
part of the design process and were predicted to bind with
similar affinity as for the amino variants (“Z”CH2). However,
Tang and co-workers was used (PDB code 1M4H; 2.1 Å).24
A
simple nonpeptidic ligand scaffold was constructed using
SPROUT by joining small fragments docked into selected
individual target sites with spacer templates consisting of six-
membered aromatic rings (Figure 1). The proposed mode of
inhibitory action of the designed scaffold was to obstruct the
B
dx.doi.org/10.1021/jm301127x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
scaffold was synthesized and tested in an in vitro BACE1
activity assay using a quenched fluorescent peptide substrate
analogue based on the Swedish mutant APP sequence
(SEVNL↓DAEFK).25 Recombinant BACE1 (25 nM) (R&D
Systems, Minneapolis, MN, USA) was preincubated with
inhibitor compounds for 30 min before addition of substrate
analogue FAM-SEVNLDAEFK-TAMRA (5 μM) and kinetic
measurement of substrate cleavage by determination of the
increase in relative fluorescence units (RFU) compared to
controls (further details in the Supporting Information).
Compound 1 was found to display a modest affinity against
BACE1, with an IC50 of 323 μM (Table 1).
Table 1. Analogues of Compound 1 Synthesized to Probe
the Essential Features for BACE1 Inhibition
SAR Study. To effectively probe the essential features of
compound 1 required for BACE1 inhibition, we prepared
structural analogues of 1 that possessed different functional
groups and flexible chain lengths in an attempt to establish an
SAR profile and to probe the proposed binding mode of the
inhibitor. As illustrated in Figure 2, the ligand was designed to
make important H-bond contacts with residues Asn37 and
Ile126 and consequently occupying the deep binding pocket
proximal to these residues. The introduction of different flexible
chain lengths in the R1 portion of the inhibitor (Table 1,
compounds 2−4) would alter the geometry of the designed
molecule and was predicted not to bind according to our model
(see Supporting Information for predicted binding scores of
selected compounds). We also anticipated that functional group
alterations would render the designed scaffold inactive (Table
1, compounds 5−6). Biological assay data as listed in Table 1
agreed well with our model predictions and support the
presence of the amine function in compound 1 as being
important for inhibition of BACE1.
a
b
Formulated as the trifluoroacetate salt. Inhibition of BACE1 activity
measured as a percentage of the control activity in an in vitro BACE1
activity assay (see text). Values shown are mean SEM, n ≥ 3. No
inhibition.
c
Table 2. Analogues of Compound 1 with Enhanced
Inhibition of BACE1
Having established one of the important features of 1
required for the inhibition of BACE1, we strived to enhance the
binding affinity of the designed skeleton through the
introduction of additional functional groups by examining our
in silico model. First, we observed that while the central
aromatic ring of the biaryl system was acting as a spacer, its
orientation was in close proximity to the catalytic aspartate
residues, Asp32 and Asp228. Introduction of substituents
bearing H-bond donor groups on this aromatic ring might form
favorable H-bond interactions with the side chains of these
catalytic aspartate residues. Therefore, we synthesized both the
ortho-hydroxy (7) and the meta-hydroxy (8) analogues to test
our hypothesis. However, this proposed alteration did not show
significant differences in the inhibition of BACE1 (Table 2).
Next, we investigated the availability of additional subpockets
located near the ligand binding regions proposed. The S1′
pocket of the enzyme could be filled with substituents
stemming from the α-C position between the biaryl system
and the amide scaffold (“X” position in Figure 1b). Several
analogues with different branch chains at the α-C position were
synthesized and assayed against BACE1. The trend in the
BACE1 IC50 values clearly suggested that the larger the branch
chain was at this α-C position, the higher the affinity was
compound
R1
R2
inhibition at 100 μM (%)
IC50 (μM)
323
>100
>100
1
H
H
H
H
35.6 1.7
42.5 0.6
28.1 3.3
51.7 2.7
55.5 1.6
66.1 2.2
101.5 0.0
5
7
2-OH
3-OH
3-F
8
9
Me
77.1 7.5
85.7 11.8
33.5 4.8
26.9 4.7
10
11
12
3-F
R-Me
iPr
H
H
CH2Ph
in light of the rather hydrophobic nature of the designed
scaffolds, we decided to concentrate on the amines as these
were anticipated to have enhanced aqueous solubility compared
to the amide and sulfonamide variants. This revealed our initial
target as amine 1 (Figure 2, shown bound in BACE1 as the
ammonium form as anticipated to predominate under
physiological conditions and is located near to the surface of
the protein and exposed to solvent). This designed molecular
Scheme 1. SPROUT-HitOpt Core Extension to Optimize the Terminal Aromatic Ring of 1
C
dx.doi.org/10.1021/jm301127x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 3. The optimized candidate structure 14 (pink), designed using SPROUT-HitOpt, overlaid with the predicted binding pose of 1 (magenta).
(a) Proposed binding mode of 14 (pink) with key residues labeled (orange) and (b) proposed binding mode of 14 with surface of the binding
pocket.
Table 3. Analogues of Inhibitor 1 Designed to Increase
Hydrophobicity of the Terminal Phenyl Ring of the Biaryl
Scaffold
structure is achieved by connecting the identified functional
groups to a library of commercially available starting materials,
using connection rules defined in a synthetic chemistry
knowledge base. This program was recently reported to have
successfully optimized inhibitors of the antimalarial target
Plasmodium falciparum dihydroorotate dehydrogenase.20
Because the terminal ring of the biaryl system of inhibitor 1
was docked into a hydrophobic site near Phe108 and Trp115,
we envisaged that more extensive occupancy of this hydro-
phobic site by extending deeper into the pocket, or the
introduction of additional hydrophobic substituents, would
maximize the potential of this binding pocket. Therefore, we
decided to utilize SPROUT-HitOpt to perform a thorough
exploration of the available binding regions proximal to this
hydrophobic site and to improve the predicted binding affinity
of 1. Using 13 as the core structure for SPROUT-HitOpt
extension and an in silico library of commercially available aryl
boronic acids (Scheme 1), compound 14, with a 10-fold
increase in predicted binding affinity compared to inhibitor 1,
was identified as one of the best optimized candidates out of
174 structures generated (Figure 3). Biological assay results
suggested that compound 14 had an increased binding affinity
toward BACE1 compared to inhibitor 1, attaining an IC50 value
of 63.9 μM (Table 3).
Following the improved BACE1 binding affinity of 14
through increasing hydrophobicity of the terminal aromatic
ring, we decided to assess the effects of different hydrophobic
substituents on BACE1 inhibition. Whereas the pentafluor-
ophenyl analogue (15) was found to be slightly less potent than
14, the 3-(trifluoromethyl)phenyl analogue (16), despite
possessing only one CF3 group, exhibited similar inhibition of
BACE1 as compared to 14. Furthermore, we prepared
compound 17 to investigate the importance of hydrophobic
substituents on the terminal phenyl ring for binding to BACE1.
Compound 17 failed to reach an IC50 value below 100 μM, for
which we propose the difference was mainly attributed to the
presence of the polar phenolic moiety (Table 3).
compound
Ar
inhibition at 100 μM (%)
IC50 (μM)
323
1
Ph
35.6 1.7
80.6 2.3
51.8 8.4
60.7 2.2
41.3 3.5
5
14
15
16
17
3,5-(CF3)2C6H3
C6F5
63.9 7.2
101 15.7
58.9 9.0
>100
3-(CF3)C6H4
4-(OH)C6H4
a
Table 4. Relative Inhibition of BACE1 and BACE2
compound
BACE1 IC50 (μM)
323
BACE2 IC50 (μM)
1
5
>1000
9
77.1 7.5
85.7 11.8
33.5 4.8
26.9 4.7
63.9 7.2
101 15.7
58.9 9.0
63.3 9.1
207.2 31.2
193.8 34.1
100.5 4.1
21.0 2.37
33.2 12.0
245.9 34.9
319.3 34.9
>1000
10
11
12
14
15
16
22
a
Values shown are mean SEM, n ≥ 3.
toward BACE1, with compounds containing the isopropyl (11)
and benzyl (12) branch chains achieving a 10- to 12-fold
increase in potency compared to 1, respectively. Compound 12,
in particular, attained a BACE1 IC50 of 26.9 μM (Table 2).
Further Optimization Using SPROUT-HitOpt. Very
recently, we have developed a variant of the de novo molecular
design software SPROUT, SPROUT-HitOpt, to assist ligand
optimization. The aim of this software is to modify the
structure of a known ligand and generate variants predicted to
display enhanced binding affinity toward the target enzyme by
fulfilling additional H-bond target sites and/or hydrophobic
pockets. A particular feature of this software is that the
optimization process is performed using part of the known
ligand as a core structure in which functional groups are
identified as points of extension. Extension from this core
The most potent of our compounds, compound 12, attained
an IC50 of 26.9 μM under our assay conditions. This is relatively
high compared to other published (GSK188909)25 and
commercially available BACE inhibitors (β-secretase inhibitor
IV, Calbiochem, Nottingham, UK) that have IC50s in the nM
range.
D
dx.doi.org/10.1021/jm301127x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
a
Table 5. MTT Compound Toxicity Assay Data
b
compound
Ar
R1
R2
R3
IC50 (μM)
323
cell viability (%)
1
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
H
H
H
H
H
H
CH2NH2
H
5
42
93
82
18
50
11
18
50
15
8
3
1
1
1
7
0
1
4
1
2
5
5
2
3
3
c
5
H
H
na
6
OMe
na
7
2-OH
3-OH
3-F
3-F
H
CH2NH2
CH2NH2
CH2NH2
CH2NH2
CH2NH2
CH2NH2
CH2NH2
CH2NH2
CH2NH2
H
>100
>100
8
9
Me
R-Me
iPr
77.1 7.5
85.7 11.8
33.5 4.8
26.9 4.7
63.9 7.2
101 5.7
58.9 9.0
na
10
11
12
14
15
16
18
19
20
H
CH2Ph
H
3,5-(CF3)2C6H3
C6F5
H
H
H
68
46
76
61
27
3-(CF3)C6H4
3,5-(CF3)2C6H3
C6F5
H
H
H
H
H
H
H
na
3-(CF3)C6H4
H
H
H
na
a
b
c
Values shown are mean SEM, n ≥ 3. Data measured at an inhibitor concentration of 100 μM, percentage inhibition relative to control. Not
active.
Figure 4. (A) Representative Western blot of sAPPα and sAPPβ fragments in conditioned media from HEK 293 cells overexpressing BACE1
incubated for 24 h with compound 15 at 100 μM. (B) Densitometric analysis of (A). Data shown as mean SEM, n = 4, **p < 0.01.
Compound Specificity: BACE2 Inhibition. To test the
specificity of the compounds against other related proteases,
their activity against BACE2 was also measured in the in vitro
assay. BACE2 is also a type I transmembrane aspartyl protease
whose amino acid sequence is 45% identical and 75% similar to
BACE1.26 The assay conditions used were identical to those for
BACE1. The results indicated that all of the compounds tested
showed at least a 2-fold selectivity for BACE1 over BACE2
except for compounds 12 and 14 (Table 4). The specificity of
our compounds for BACE1 over BACE2 suggests that they will
have limited effects on other structurally related proteases.
Compound Toxicity Assay and Cellular BACE1
Inhibition. To assess the potential of this series of de novo
designed inhibitors for further development, we used
mammalian cells to assess compound toxicity and cellular
BACE1 inhibition. Compounds were screened for toxicity using
the cell viability 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetra-
zolium bromide (MTT) assay to determine the percentage of
viable cells remaining following compound incubation at a
concentration of 100 μM over a 24 h period. Compounds that
induced a loss of greater than 50% in cell viability were not
tested further for cellular BACE1 inhibition. Compounds that
retained cell viability of 50% or greater at 100 μM were tested
on human embryonic kidney (HEK 293) cells stably expressing
BACE1 and wild-type human APP that was N-terminally
tagged with alkaline phosphatase (AP-APP) to determine
potential BACE1 inhibitory activity. The release of soluble N-
terminal sAP-APP fragments into the media, was determined by
measuring the resulting alkaline phosphatase activity of
conditioned media samples taken after the incubation.
Inhibition of BACE1 would display a decrease in the alkaline
phosphatase activity of conditioned media compared to control
levels. The alkaline phosphatase activity following compound
incubation was expressed as a percentage of the activity in the
appropriate control sample.
Many of the changes introduced to the middle aromatic ring
of the biaryl scaffold, and also the addition of branch chains at
the α-C position, rendered the compounds very toxic
(compounds 7−12, Table 5). In the collection of inhibitors
which possessed a terminal phenyl ring with hydrophobic
E
dx.doi.org/10.1021/jm301127x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
a
Table 6. Tertiary and Quaternary Amine Analogues of
Inhibitor 1 Designed to Tackle the Toxicity Associated with
This Series of Inhibitors
Scheme 2
a
a
Reaction conditions: (a) (Boc)2O, NEt3, CH2Cl2, 0 °C−rt, 16 h,
95%; (b) LiAlH4, THF, 0 °C−rt, 48 h, 55%.
a
Scheme 3
a
b
a
Reaction conditions: (a) R1-NH2, HOBt, EDAC, NMM, CH2Cl2,
Values shown are mean
SEM, n ≥ 3. Data measured at an
inhibitor concentration of 100 μM, percentage inhibition relative to
control.
0 °C−rt, 20 h, 42−70%.
a
Scheme 4
substituents, compound 14, containing two CF3 groups, was
shown to be highly toxic, whereas the pentafluorophenyl (15)
and 3-(trifluoromethyl)phenyl (16) analogues were less toxic.
The primary amine of this series of inhibitors, which, from our
earlier studies appeared to be important for BACE1 inhibition,
also appeared to contribute to cell toxicity. A comparison
between the percentage cell viability of 1 to that of 5 and 6
revealed that compounds without the primary amine, although
inactive toward BACE1, were much less toxic at 100 μM.
Intrigued by this observation, we decided to further investigate
whether the hydrophobic substituents on the terminal phenyl
ring of the biaryl scaffold were partially responsible for the
toxicity of compounds 14−16. Analogues 18−20, the
corresponding analogues of 14−16 without the primary
amine, were assessed. Consistent with our conclusions from
the SAR study, analogues 18−20 were shown to be inactive
toward BACE1 but compound toxicity was decreased, with the
exception of analogue 20 (we suspected that this discrepancy
might be owing to the low solubility of this compound). It is,
however, evident that the hydrophobic substituents on the
terminal phenyl ring, including the bis-(trifluoromethyl)phenyl
and the pentafluorophenyl scaffolds, were accountable for a
certain level of toxicity in cells, as the percentage cell viability
following incubation with analogues bearing these function-
a
Reaction conditions: (a) 10% aq KOH/MeOH, 90 °C, 3 h, 96−98%;
(b) SOCl2, 60 °C, 3 h; (c) TMSCHN2, NEt3, THF/MeCN, 0 °C,
24 h; (d) PhCO2Ag, NEt3, MeOH, 30 min; (e) 10% aq KOH/MeOH,
90 °C, 3 h, 16−49%; (f) PhB(OH)2, Pd(OAc)2, Na2CO3, TBAB, H2O,
120 °C, microwave, 30−45 min, 64−89%.
alities was comparatively lower than compound 5 containing
only a simple phenyl ring.
The cell viability assay results suggested that only compound
15 could be tested further to assess whether it resulted in
cellular BACE1 inhibition; all other compounds were either
considered too toxic, or showed no significant inhibition of
BACE1 and were therefore not suitable for further cell-based
testing. Incubation of HEK cells expressing AP-APP and
BACE1 with 100 μM of compound 15, resulted in a 39.4
Figure 5. (A) Representative Western blot of sAPPα and sAPPβ fragments in conditioned media from HEK 293 cells overexpressing BACE1
incubated for 24 h with compound 22 at 100 μM. (B) Densitometric analysis of (A). Data shown as mean SEM, n = 3.
F
dx.doi.org/10.1021/jm301127x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
a
cytochrome P450 metabolism at the methylene position to
produce toxic metabolites such as benzaldehydes in cell
media.28,29 As indicated previously, although the amide and
sulfonamide variants of compound 1 (Figure 1b, “Z”CO)
and SO2 were predicted to show favorable binding to BACE1,
we were concerned about the solubility of such variants. As an
alternative, we reasoned that conversion of the primary amine
into a tertiary amine moiety might reduce the toxicity
associated with this inhibitor series. Interestingly, modeling of
the tertiary amine variants of compound 1 within BACE1
indicated that these derivatives may adopt a significantly
different binding pose to that predicted for compound 1. The
tertiary amine moiety (presumably protonated at physiological
pH) would not be involved in H-bonding contacts to the
protein but would instead reorient in order to allow
hydrophobic interaction between the N-alkyl groups and the
side chain of Ile126. Additionally, considerable movement of
the core section of the molecule was predicted to result in the
formation of new H-bond contacts involving Thr231 (see
Supporting Information).
Scheme 5
a
Reaction conditions: (a) MeOH, H2SO4, 65 °C, 16 h, 83%; (b)
KHMDS, THF, 0 °C, 20 min; (c) R2X, THF, 0 °C, 16 h; (d) 10% aq
KOH/MeOH, 90 °C, 3 h, 36−94%.
a
Scheme 6
Four analogues containing tertiary or quaternary amines were
synthesized. All these analogues were shown to be active
against BACE1, with compound 22, in particular, attaining an
IC50 against BACE1 of 63.3 μM (Table 6). Cell viability of
compound 22 was also improved as anticipated (80% cell
viability at 100 μM compound concentration), and in the cell-
based assay, compound 22 (100 μM) resulted in a 21.7 9.9%
inhibition of sAP-APP production. Compound 22 was
therefore assessed for cellular BACE1 inhibition in HEK cells
overexpressing BACE1 by Western blotting (as described for
compound 15). Compound 22 (100 μM) decreased sAPPβ
production by 33% compared to control levels, but this was not
statistically significant (p = 0.064); sAPPα levels were decreased
by 10% (Figure 5).
Chemistry. Preparation of 1 was achieved by first protecting
4-(aminomethyl)benzonitrile 25 by following the procedure
described by Goodyer et al. to give the Boc-protected amine
26,30 which was reduced using LiAlH4 to obtain the
intermediate 27 in 52% yield over two steps (Scheme 2).
The coupling of 4-biphenylacetic acid 28 with 27 delivered 1 in
65% yield using the peptide coupling reagents HOBt and
EDAC. Similarly, compounds 2−6 were readily synthesized by
coupling 4-biphenylacetic acid 28 with the respective amine
(Scheme 3).
a
Reaction conditions: (a) PhCH2NH2 or 27, HOBt, EDAC, NMM,
CH2Cl2, 0 °C−rt, 20 h, 65−81%; (b) R2B(OH)2, Pd(OAc)2, PPh3,
K2CO3, toluene, 120 °C, microwave, 30−45 min, 49−95%; (c) TFA,
CH2Cl2, 0 °C, 2−3 h, 75−95%.
2.7% inhibition of sAP-APP production, indicating a potentially
cell-active compound. To determine the effects of compound
15 on BACE1 activity specifically, we analyzed the effect of the
compound on HEK cells overexpressing BACE1. The cells
were incubated with the compound, then the conditioned
media from these cells was Western blotted using a specific
sAPPβ antibody, 1A9 (see Supporting Information for further
details).27 Compound 15 significantly decreased sAPPβ levels
in these cells, while sAPPα levels were not significantly altered
(Figure 4). These results suggested that compound 15
significantly inhibited cellular BACE1 activity with little effect
on α-secretase activity and indicated that compound 15 was a
cell-active BACE1 inhibitor.
Tackling Compound Toxicity. While the presence of the
primary amine appeared to be an important feature for the
binding of this series of inhibitors within BACE1, the presence
of this functionality resulted in the compounds exhibiting cell
toxicity. We reasoned that this might be owing to the presence
of the benzylamine moiety, which is particularly prone to
The appropriately substituted biaryl acetic acid 33a and 33b,
intermediates for compounds 7 and 8, respectively, were
prepared according to the general scheme shown in Scheme 4.
Methyl 4-bromo-2-methoxybenzoate 29a was first hydrolyzed
to the benzoic acid intermediate 30a, for which the
a
Scheme 7
a
Reaction conditions: (a) 4-(MeO)C6H4B(OH)2, Pd(OAc)2, PPh3, K2CO3, toluene, 120 °C, microwave, 30−45 min, 84%; (b) BBr3, CH2Cl2, rt,
16 h, 66%.
G
dx.doi.org/10.1021/jm301127x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
a
Scheme 8
a
Reaction conditions: (a) 20% aq KOH, 100 °C, 2 h, 87%; (b) C6F5Br, Pd(OAc)2, Na2CO3, TBAB, H2O, 120 °C, microwave, 30−45 min, 39%; (c)
PhCH2NH2, HOBt, EDAC, NMM, CH2Cl2, 0 °C−rt, 20 h, 40%.
a
Scheme 9
a
Reaction conditions: (a) K2CO3, H2O, 120 °C, microwave, 20 min, 34−53%.
Table 7. Synthesis of Compounds 1, 7−12, and 15
b
compound
intermediate
Ar
Ph
R1
R2
deprotection method
yield (%)
1
28
H
H
H
H
1
2
2
1
1
1
1
1
65
83
59
71
79
75
68
77
7
33a
33b
Ph
2-OH
3-OH
3-F
3-F
H
8
Ph
a
9
cs
Ph
Me
a
10
11
12
15
cs
Ph
R-Me
iPr
35
36
43
Ph
Ph
H
CH2Ph
H
C6F5
H
a
b
Commercial source. Reaction conditions: (method 1) TFA, CH2Cl2, 0 °C, 2−3 h; (method 2) BBr3, CH2Cl2, rt, 16 h.
homologation was then achieved following literature proce-
dures described by Aoyama and Cesar to give methyl 4-bromo-
2-methoxyphenylacetate 31a in 62% yield over four steps.31,32
Subsequent hydrolysis of 31a, followed by Suzuki coupling of
the resultant phenylacetic acid 32a with phenylboronic acid
using tetrabutylammonium bromide (TBAB),33 attained the
biaryl scaffold 33a in 50% yield. Similarly, the 3-methoxy-4-
phenyl substituted scaffold 33b was obtained following the
same procedure using methyl 4-iodo-3-methoxybenzoate 29b,
for which the preparation followed the synthetic route
described by Baret et al.34
were subsequently hydrolyzed to give the corresponding
biphenylacetic acid intermediates 35 and 36 (Scheme 5).
Aryl halide intermediates 38 and 39 were prepared by
coupling 4-bromophenylacetic acid 37 with benzylamine and
27, respectively, which the products were coupled to the
appropriately substituted aryl boronic acid under Suzuki
coupling conditions to furnish the corresponding biaryl
scaffolds. Compounds 18 and 20 were obtained in 39−41%
yields, whereas compounds 14 and 16 were obtained in 46−
52% yields after deprotection using TFA (Scheme 6). Similarly,
compound 17 was obtained from the deprotection of 40 using
BBr3 (Scheme 7).
Intermediates 35 and 36 were synthesized from the
conversion of 4-biphenylacetic acid 28 to the methyl ester
34, followed by alkylation at the α-C position using procedures
reported by Robichaud et al.35 The alkylated intermediates
Synthesis of the pentafluorophenyl analogue 19 involved
initially the hydrolysis of 4-cyanomethylphenylboronic acid 41
by following the procedure described by Serafin and Makosza.36
H
dx.doi.org/10.1021/jm301127x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Subsequent Suzuki coupling of 42 with pentafluorophenyl-
bromide furnished 43, which was then coupled with benzyl-
amine to give compound 19 in 14% yield over three steps
(Scheme 8).
Alkylation of 1 using the conditions reported by Ju and
Varma37 afforded compounds 21−24 in moderate yields
(Scheme 9).
Table 7 summarizes the synthesis of compounds 1, 7−12,
and 15 from the respective intermediates described above using
a two-step coupling-deprotection method.
ABBREVIATIONS USED
■
AD, Alzheimer’s disease; APP, amyloid precursor protein; Aβ,
amyloid-β peptide; BACE1, β-site APP cleaving enzyme 1;
SAR, structure−activity relationships
REFERENCES
■
(1) Callaway, E. Gene mutation defends against Alzheimer’s disease.
Nature 2012, 487, 153.
(2) Huang, W. H.; Sheng, R.; Hu, Y. Z. Progress in the development
of nonpeptidomimetic BACE 1 inhibitors for Alzheimer’s disease.
Curr. Med. Chem. 2009, 16, 1806−1820.
(3) Selkoe, D. J. Alzheimer’s disease: genes, proteins, and therapy.
Physiol. Rev. 2001, 81, 741−766.
CONCLUSIONS
■
We have successfully used the structure-based molecular design
program SPROUT to develop a new BACE1 inhibitor scaffold
using a de novo fragment-based ligand design approach. Upon
the basis of a simple biphenylacetamide core structure, an SAR
study around the designed ligand was conducted which
confirmed that the amine functionality was essential for
BACE1 inhibition. The binding affinity of the initial designed
skeleton was enhanced, from an initial IC50 against BACE1 of
323 μM to 26.9 μM for one of the most potent inhibitors, using
a recently developed in silico optimization software SPROUT-
HitOpt. Subsequent cell viability assays revealed significant
toxicity associated with the majority of compounds within this
series of inhibitors. However, we were able to demonstrate that
compound 15, with a relatively low toxicity, was a cell-active
BACE1 inhibitor which selectively inhibited cellular BACE1
activity with little effect on α-secretase activity. Although the
toxicity associated with this series of compounds has rendered
them unsuitable for further studies in in vivo experiments, we
have demonstrated the successful application of SPROUT and
SPROUT-HitOpt in the development of a small-molecule
nonpeptidic inhibitor against an enzyme that is a challenging
therapeutic target.
(4) Hardy, J. A.; Higgins, G. A. Alzheimer’s disease: the amyloid
cascade hypothesis. Science 1992, 256, 184−185.
(5) Hooper, N. M.; Turner, A. J. The search for alpha-secretase and
its potential as a therapeutic approach to Alzheimer’s disease. Curr.
Med. Chem. 2002, 9, 1107−1119.
(6) Vassar, R.; Bennett, B. D.; Babu-Khan, S.; Kahn, S.; Mendiaz, E.
A.; Denis, P.; Teplow, D. B.; Ross, S.; Amarante, P.; Loeloff, R.; Luo,
Y.; Fisher, S.; Fuller, L.; Edenson, S.; Lile, J.; Jarosinski, M. A.; Biere, A.
L.; Curran, E.; Burgess, T.; Louis, J. C.; Collins, F.; Treanor, J.; Rogers,
G.; Citron, M. Beta-secretase cleavage of Alzheimer’s amyloid
precursor protein by the transmembrane aspartic protease BACE.
Science 1999, 286, 735−741.
(7) Mancini, F.; De Simone, A.; Andrisano, V. Beta-secretase as a
target for Alzheimer’s disease drug discovery: an overview of in vitro
methods for characterization of inhibitors. Anal. Bioanal. Chem. 2011,
400, 1979−1996.
(8) Sinha, S.; Anderson, J. P.; Barbour, R.; Basi, G. S.; Caccavello, R.;
Davis, D.; Doan, M.; Dovey, H. F.; Frigon, N.; Hong, J.; Jacobson-
Croak, K.; Jewett, N.; Keim, P.; Knops, J.; Lieberburg, I.; Power, M.;
Tan, H.; Tatsuno, G.; Tung, J.; Schenk, D.; Seubert, P.; Suomensaari,
S. M.; Wang, S. W.; Walker, D.; Zhao, J.; McConlogue, L.; John, V.
Purification and cloning of amyloid precursor protein beta-secretase
from human brain. Nature 1999, 402, 537−540.
(9) Lin, X. L.; Koelsch, C.; Wu, S. L.; Downs, D.; Dashti, A.; Tang, J.
Human aspartic protease memapsin 2 cleaves the beta-secretase site of
beta-amyloid precursor protein. Proc. Natl. Acad. Sci. U. S. A. 2000, 97,
1456−1460.
(10) Eder, J.; Hommel, U.; Cumin, F.; Martoglio, B.; Gerhartz, B.
Aspartic proteases in drug discovery. Curr. Pharm. Des. 2007, 13, 271−
285.
(11) Ghosh, A. K.; Kumaragurubaran, N.; Hong, L.; Koelsh, G.;
Tang, J. Memapsin 2 (beta-secretase) inhibitors: drug development.
Curr. Alzheimer Res. 2008, 5, 121−131.
(12) Cole, D. C.; Manas, E. S.; Stock, J. R.; Condon, J. S.; Jennings,
L. D.; Aulabaugh, A.; Chopra, R.; Cowling, R.; Ellingboe, J. W.; Fan, K.
Y.; Harrison, B. L.; Hu, Y.; Jacobsen, S.; Jin, G. X.; Lin, L.; Lovering, F.
E.; Malamas, M. S.; Stahl, M. L.; Strand, J.; Sukhdeo, M. N.; Svenson,
K.; Turner, M. J.; Wagner, E.; Wu, J. J.; Zhou, P.; Bard, J.
Acylguanidines as small-molecule beta-secretase inhibitors. J. Med.
Chem. 2006, 49, 6158−6161.
ASSOCIATED CONTENT
■
S
* Supporting Information
Tables listing the degree of purity for all target compounds
(area percent and retention time), elemental analyses, details of
the experimental procedures and spectroscopic data for each
compound, details of the biological assays, and modeling data
for selected key compounds. This material is available free of
charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
(13) Baxter, E. W.; Conway, K. A.; Kennis, L.; Bischoff, F.; Mercken,
M. H.; De Winter, H. L.; Reynolds, C. H.; Tounge, B. A.; Luo, C.;
Scott, M. K.; Huang, Y.; Braeken, M.; Pieters, S. M. A.; Berthelot, D. J.
C.; Masure, S.; Bruinzeel, W. D.; Jordan, A. D.; Parker, M. H.; Boyd, R.
E.; Qu, J.; Alexander, R. S.; Brenneman, D. E.; Reitz, A. B. 2-Amino-
3,4-dihydroquinazolines as inhibitors of BACE-1 (beta-site APP
cleaving enzyme): use of structure based design to convert a
micromolar hit into a nanomolar lead. J. Med. Chem. 2007, 50,
4261−4264.
*For N.M.H.: phone, +44 113 3433163; fax, +44 113 3433106;
e-mail, n.m.hooper@leeds.ac.uk. For A.P.J.: phone, +44 113
3436515; fax, +44 113 3436530; e-mail, p.johnson@leeds.ac.uk.
For C.W.G.F.: phone, +44 113 3436510; fax, +44 113 3436530;
e-mail, c.w.g.fishwick@leeds.ac.uk.
Notes
The authors declare no competing financial interest.
(14) Ghosh, A. K.; Brindisi, M.; Tang, J. Developing beta-secretase
inhibitors for treatment of Alzheimer’s disease. J. Neurochem. 2012,
120, 71−83.
(15) Huang, D. Z.; Luthi, U.; Kolb, P.; Edler, K.; Cecchini, M.;
Audetat, S.; Barberis, A.; Caflisch, A. Discovery of cell-permeable non-
peptide inhibitors of beta-secretase by high-throughput docking and
ACKNOWLEDGMENTS
■
We gratefully acknowledge the support of the Hong Kong
Croucher Foundation (scholarship to N.Y.M.) and the
Alzheimer’s Research UK (to J.C., K.A.B.K., and E.C.).
I
dx.doi.org/10.1021/jm301127x | J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
continuum electrostatics calculations. J. Med. Chem. 2005, 48, 5108−
5111.
(16) Huang, D. Z.; Luthi, U.; Kolb, P.; Cecchini, M.; Barberis, A.;
Caflisch, A. In silico discovery of beta-secretase inhibitors. J. Am. Chem.
Soc. 2006, 128, 5436−5443.
(17) Mok, N. Y.; Chadwick, J.; Kellett, K. A. B.; Hooper, N. M.;
Johnson, A. P.; Fishwick, C. W. G. Discovery of novel non-peptide
inhibitors of BACE-1 using virtual high-throughput screening. Bioorg.
Med. Chem. Lett. 2009, 19, 6770−6774.
(18) Xu, W. J.; Chen, G.; Zhu, W. L.; Zuo, Z. L. Molecular docking
and structure−activity relationship studies on benzothiazole based
non-peptidic BACE-1 inhibitors. Bioorg. Med. Chem. Lett. 2010, 20,
6203−6207.
(19) Besong, G. E.; Bostock, J. M.; Stubbings, W.; Chopra, I.; Roper,
D. I.; Lloyd, A. J.; Fishwick, C. W. G.; Johnson, A. P. A de novo
designed inhibitor of D-Ala-D-Ala ligase from E. coli. Angew. Chem., Int.
Ed. 2005, 44, 6403−6406.
(20) Davies, M.; Heikkila, T.; McConkey, G. A.; Fishwick, C. W. G.;
Parsons, M. R.; Johnson, A. P. Structure-based design, synthesis, and
characterization of inhibitors of human and Plasmodium falciparum
dihydroorotate dehydrogenases. J. Med. Chem. 2009, 52, 2683−2693.
(21) Woon, E. C. Y.; Zervosen, A.; Sauvage, E.; Simmons, K. J.;
Zivec, M.; Inglis, S. R.; Fishwick, C. W. G.; Gobec, S.; Charlier, P.;
Luxen, A.; Schofield, C. J. Structure guided development of potent
reversibly binding penicillin binding protein inhibitors. ACS Med.
Chem. Lett. 2011, 2, 219−223.
(33) Bedford, R. B.; Blake, M. E.; Butts, C. P.; Holder, D. The Suzuki
coupling of aryl chlorides in TBAB−water mixtures. Chem. Commun.
2003, 466−467.
(34) Baret, P.; Beaujolais, V.; Beguin, C.; Gaude, D.; Pierre, J. L.;
Serratrice, G. Towards new iron(III) chelators: Synthesis and
complexing ability of a water-soluble tripodal ligand based on 2,2′-
dihydroxybiphenyl subunits. Eur. J. Inorg. Chem. 1998, 613−619.
(35) Robichaud, J.; Oballa, R.; Prasit, P.; Falgueyret, J. P.; Percival, M.
D.; Wesolowski, G.; Rodan, S. B.; Kimmel, D.; Johnson, C.; Bryant, C.;
Venkatraman, S.; Setti, E.; Mendonca, R.; Palmer, J. T. A novel class of
nonpeptidic biaryl inhibitors of human cathepsin K. J. Med. Chem.
2003, 46, 3709−3727.
(36) Serafin, B.; Makosza, M. Arylboronic compoundsVIII: Some
reactions of hydroxy and carboxyphenylboronic acids. Tetrahedron
1963, 19, 821−826.
(37) Ju, Y. H.; Varma, R. S. Aqueous N-heterocyclization of primary
amines and hydrazines with dihalides: microwave-assisted syntheses of
N-azacycloalkanes, isoindole, pyrazole, pyrazolidine, and phthalazine
derivatives. J. Org. Chem. 2006, 71, 135−141.
(22) Gillet, V. J.; Newell, W.; Mata, P.; Myatt, G.; Sike, S.; Zsoldos,
Z.; Johnson, A. P. Sproutrecent developments in the de novo design
of molecules. J. Chem. Inf. Comput. Sci. 1994, 34, 207−217.
(23) Law, J. M. S.; Fung, D. Y. K.; Zsoldos, Z.; Simon, A.; Szabo, Z.;
Csizmadia, I. G.; Johnson, A. P. Validation of the SPROUT de novo
design program. J. Mol. Struct.Theochem 2003, 666, 651−657.
(24) Hong, L.; Turner, R. T.; Koelsch, G.; Shin, D. G.; Ghosh, A. K.;
Tang, J. Crystal structure of memapsin 2 (beta-secretase) in complex
with an inhibitor OM00−3. Biochemistry 2002, 41, 10963−10967.
(25) Hussain, I.; Hawkins, J.; Harrison, D.; Hille, C.; Wayne, G.;
Cutler, L.; Buck, T.; Walter, D.; Demont, E.; Howes, C.; Naylor, A.;
Jeffrey, P.; Gonzalez, M. I.; Dingwall, C.; Michel, A.; Redshaw, S.;
Davis, J. B. Oral administration of a potent and selective non-peptidic
BACE-1 inhibitor decreases beta-cleavage of amyloid precursor protein
and amyloid-beta production in vivo. J. Neurochem. 2007, 100, 802−
809.
(26) Sun, X. L.; He, G. Q.; Song, W. H. BACE2, as a novel APP
theta-secretase, is not responsible for the pathogenesis of Alzheimer’s
disease in Down syndrome. FASEB J. 2006, 20, 1369−1376.
(27) Parvathy, S.; Hussain, I.; Karran, E. H.; Turner, A. J.; Hooper, N.
M. Alzheimer’s amyloid precursor protein alpha-secretase is inhibited
by hydroxamic acid-based zinc metalloprotease inhibitors: Similarities
to the angiotensin converting enzyme secretase. Biochemistry 1998, 37,
1680−1685.
(28) Coutts, R. T.; Su, P.; Baker, G. B. Involvement of Cyp2d6,
Cyp3a4, and other cytochrome-P-450 isozymes in N-dealkylation
reactions. J. Pharmacol. Toxicol. Methods 1994, 31, 177−186.
(29) Guengerich, F. P.; Isin, E. M. Mechanisms of cytochrome P450
reactions. Acta Chim. Slov. 2008, 55, 7−19.
(30) Goodyer, C. L. M.; Chinje, E. C.; Jaffar, M.; Stratford, I. J.;
Threadgill, M. D. Synthesis of N-benzyl- and N-phenyl-2-amino-4,5-
dihydrothiazoles and thioureas and evaluation as modulators of the
isoforms of nitric oxide synthase. Bioorg. Med. Chem. 2003, 11, 4189−
4206.
(31) Aoyama, T.; Shioiri, T. New methods and reagents in organic-
synthesis 0.8. Trimethylsilyldiazomethane, a new, stable, and safe
reagent for the classical Arndt−Eistert synthesis. Tetrahedron Lett.
1980, 21, 4461−4462.
(32) Cesar, J.; Dolenc, M. S. Trimethylsilyldiazomethane in the
preparation of diazoketones via mixed anhydride and coupling reagent
methods: a new approach to the Arndt−Eistert synthesis. Tetrahedron
Lett. 2001, 42, 7099−7102.
J
dx.doi.org/10.1021/jm301127x | J. Med. Chem. XXXX, XXX, XXX−XXX