Macrocyclic prolinyl acyl guanidines as inhibitors of β-secretase (BACE)

Article abstract of DOI:10.1016/j.bmcl.2015.10.031

The synthesis, evaluation, and structure-activity relationships of a class of acyl guanidines which inhibit the BACE-1 enzyme are presented. The prolinyl acyl guanidine chemotype (7c), unlike compounds of the parent isothiazole chemotype (1), yielded compounds with good agreement between their enzymatic and cellular potency as well as a reduced susceptibility to P-gp efflux. Further improvements in potency and P-gp ratio were realized via a macrocyclization strategy. The in vivo profile in wild-type mice and P-gp effects for the macrocyclic analog 21c is presented.

Full text of DOI:10.1016/j.bmcl.2015.10.031

Bioorganic & Medicinal Chemistry Letters 25 (2015) 5040–5047
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Macrocyclic prolinyl acyl guanidines as inhibitors of b-secretase
(BACE)
⇑
Kenneth M. Boy , Jason M. Guernon, Yong-Jin Wu, Yunhui Zhang, Joe Shi, Weixu Zhai, Shirong Zhu,
Samuel W. Gerritz, Jeremy H. Toyn, Jere E. Meredith, Donna M. Barten, Catherine R. Burton,
Charles F. Albright, Andrew C. Good, James E. Grace, Kimberley A. Lentz, Richard E. Olson, John E. Macor,
Lorin A. Thompson III
Research and Development, Bristol-Myers Squibb Company, 5 Research Parkway, Wallingford, CT 06492, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The synthesis, evaluation, and structure–activity relationships of a class of acyl guanidines which inhibit
the BACE-1 enzyme are presented. The prolinyl acyl guanidine chemotype (7c), unlike compounds of the
parent isothiazole chemotype (1), yielded compounds with good agreement between their enzymatic and
cellular potency as well as a reduced susceptibility to P-gp efflux. Further improvements in potency and
P-gp ratio were realized via a macrocyclization strategy. The in vivo profile in wild-type mice and P-gp
effects for the macrocyclic analog 21c is presented.
Received 30 June 2015
Revised 8 October 2015
Accepted 12 October 2015
Available online 20 October 2015
Keywords:
BACE
Ó 2015 Elsevier Ltd. All rights reserved.
Alzheimer’s disease
Alzheimer’s disease (AD) is a progressive neurodegenerative
disease which currently affects 5.4 million Americans.1 Since age
is the single best predictor of AD risk, the growing elderly
population is expected to cause the ranks of the affected to grow
dramatically to an estimated 11–16 million cases in 2050.1 Current
approved therapies for AD, including acetylcholinesterase
inhibitors and the NMDA antagonist memantine, do not affect
the neurodegeneration seen in AD, but are only palliative in
nature.2 Therapies which would directly slow or halt disease
progression are urgently needed, and would benefit patients, their
families and their caregivers.
BACE-1 is the aspartyl protease which is responsible for the
cleavage of APP to form the N-terminus of the Ab peptides, which
is the first and rate-limiting step in Ab production. Early studies
found that BACE overexpression led to increased Ab production,
and conversely that homozygous BACE knockout mice were both
viable6 and deficient in Ab production. These findings supported
the search for inhibitors of this enzyme as a safe and potentially
disease-modifying treatment for AD. Although studies subse-
quently published have illuminated potential issues with chronic
BACE inhibition,7 this target still remains of high interest.
O
Ab peptides are the central portion of the transmembrane-
bound amyloid precursor protein (APP) liberated by the sequential
action of beta- and gamma-secretase. The latter action of gamma
secretase results in the formation of a heterogeneous mixture of
Ab peptides predominantly 40 and 42 amino acids in length, the
longer of which (Ab42) is particularly prone to aggregation.3
According to the amyloid hypothesis,4 the neurotoxic oligomers5
of Ab42 initiate a series of downstream events, including tau hyper-
phosphorylation and inflammation, which ultimately lead to AD.
The Ab oligomers ultimately deposit as the characteristic Ab pla-
ques, which are observed along with neurofibrillary tangles of
hyperphosphorylated tau in the brains of individuals with AD.
Ki 40 nM (pH 5.0)
NH2
O
Cell EC50 2.0
µM
Cl
N
H
N
N
S
H2N
1
Cl
Historically the first BACE inhibitor chemotypes were pep-
tidomimetic in structure, utilizing known aspartyl protease transi-
tion-state isosteres.8 Guided by crystallography, impressive levels
of potency were achieved. However, the high molecular weight
(MW) and P-glycoprotein (P-gp) efflux at the blood–brain barrier
of those compounds ultimately made this strategy unproductive,
as the compounds did not effectively get to their target in the
brain. In a different approach, a combination of deck screening
and fragment-based design, followed by optimization efforts at
Inhibiting either or both b- and c-secretases should decrease the
pool of available Ab and thus slow or stop disease progression.
⇑
Corresponding author. Tel.: +1 203 677 6278; fax: +1 203 677 7702.
E-mail address: boyk@bms.com (K.M. Boy).
http://dx.doi.org/10.1016/j.bmcl.2015.10.031
0960-894X/Ó 2015 Elsevier Ltd. All rights reserved.

K. M. Boy et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5040–5047
5041
BMS led to the identification of an acyl guanidine (AG) chemotype
(1, for example).9,10 Robust potency in enzyme assays of the BMS
isoxazole and isothiazole AG chemotypes required the incorpora-
tion of an acetamide functionality on the dichloroaniline fragment,
however cellular potency remained low.9 While further elabora-
tion of the acetamide to form glycinamide analogs produced com-
pounds with robust cellular potency, both the acetamides and
glycinamides showed significant P-gp efflux, precluding central
efficacy.9 In this Letter, we describe our efforts to further improve
this chemotype by focusing on limiting P-gp efflux.
The lack of direct translation of enzymatic potency into cellular
potency in the initial BMS series was striking, as detailed by Gerritz
et al.9 This observation was in sharp contrast to our experience
with our diaminopropane isostere chemotype, where the two
activities were in reasonable agreement.11 To further optimize
our AG chemotype we focused on identifying compounds with
similar enzyme and cellular activities, even if they were less potent
overall. While investigating the possible reasons for the potency
differences, we carefully examined our assay conditions. Our initial
in vitro enzyme binding assay12 was conducted at pH 5 where the
enzyme is most active based on pH titrations. This relatively low
pH was reported to be similar to the pH of the endocytic compart-
ment where BACE is catalytically active.13 As we compared data
collected in the pH 5 enzyme assay with cellular potency data, it
was clear that the cellular potency was not well predicted by the
binding potency at pH 5. We then varied the pH in the enzyme
assay, and found that the cellular activity was better correlated
with our in vitro binding assay run at pH 6.4. The pKa of the
original isothiazole acyl guanidines (compound 1)9 was 5.3, imply-
ing that almost equal amounts of the protonated and non-proto-
nated species were present at pH 5. The decrease in potency and
cellular activity at higher pH suggested that the protonated form
of the inhibitor was more important for potent binding. To this
end, we set out to scan more electron-rich saturated and unsatu-
rated heterocycles to afford acyl guanidines with a higher pKa,
where the core would remain protonated at physiological pH.14
Since this work was conducted, others have reported that cellular
activity can be difficult to predict based on low-pH enzyme assays
since the relationship between protonation state of the catalytic
aspartates in BACE as a function of pH can vary with inhibitor
structure.15
Heterocyclic isothiazole replacements were made as shown in
Scheme 1. N-arylation of the appropriate heterocyclic carboxylic
acids using a modified Ullman protocol16 afforded compounds
2a–n. Coupling of the resulting aryl-substituted heterocyclic car-
boxylic acids 2a–n with commercially available N-Boc-2-methyl
isothiourea 3 using HATU provided intermediates 4a–n. These
intermediates were condensed with the functionalized benzy-
lamine 5 to afford the penultimate products 6a–n, which were sub-
sequently deprotected by TFA to furnish the desired acyl
guanidines 7a–n. Note that the alcohol in 7e was carried through
the synthesis as its TBS ether, but this was likely unnecessary
based on the smooth synthesis of the diastereomer 7f with no
protecting group.
Guanyl urethane 11a and ureas 11b–e were also examined
(Scheme 2), and were constructed by combining commercially
Cl
O
Boc
S
Boc
S
O
NH
b
NH
NH
O
a
+
+
NH2
N
Het
HO
N
Het
N
H
N
Cl
5
3
2a-n
4a-n
Boc
NH2
N
O
Cl
NH
O
O
Cl
c
O
N
N
H
N
Het
N
H
N
Het
N
N
H
H
Cl
7a-n
Cl
6a-n
c X=Y=H
Heterocycles:
d enantiomer of c
e X=OH, Y=H, see text
f X=H, Y=OH
N
l
N
N
N
N
N
N
g X=H, Y=OMe
h X=Ph, Y=H
X
c-k
a
b
i X=OPh, Y=H
j X=OBn, Y=H
k X=H, Y=OBn
Y
m
n
Scheme 1. Synthesis of Compounds 7a–n. Reagents and conditions: (a) HATU, NMM, CH2Cl2, rt, 57–84% yield; (b) DIPEA, CH2Cl2, 86–99% yield; (c) TFA, CH2Cl2, rt, 57–72%
yield.
NH2
O
O
Cl
O
NH2
Cl
Bn
R
O
O
N
H
N
O
Boc
b,c
d
NH
N
5
H
N
Cl
Cl
11a
H
N
H
NBoc
O
a
Cl
Cl
P
N
NH
N
N
H
N
9
Cl
Cl
N
H
N
H
Boc
N
H
NBoc
N
H
10b-e
11b-e P = H
8
P = Boc
e
Scheme 2. Synthesis of compounds 11a–e. Reagents and conditions: (a) THF, rt, 87% yield; (b) TFA, CH2Cl2; (c) benzyl chloroformate, CH2Cl2, NaOH, 4% yield from 9; (d):
RNH2, THF, reflux; (d) TFA, CH2Cl2, ꢀ85% yield from 9.

5042
K. M. Boy et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5040–5047
available N,N0-bis-Boc-1H-pyrazole-carboxamidine
8
with the
with our pH hypothesis. Gratifyingly, a saturated ring system such
as the N-aryl pyrrolidine analog 7c simultaneously had a low pH
shift and potency in the sub-micromolar range. The cellular
potency of 7c (EC50 = 680 nM) was in close agreement with the Ki
(340 nM at pH 6.4).
substituted benzylamine 5 in THF to afford intermediate 9. TFA
deprotection followed by coupling with benzylchloroformate
yielded the guanyl urethane 11a. Alternatively, condensing 9 with
an equivalent of an amine using heat, and subsequent TFA
deprotection yielded the desired guanyl ureas 11b–e.
In order to gain further insight into the binding mode of these
new inhibitors, a molecular model was generated based on a
crystal structure of an earlier acyl guanidine lead compound.18
Docking 7c into this model (Fig. 1) revealed many features
common to the parent isoxazole chemotype.9 The catalytic
aspartate residues were each bound to one of the hydrogens of
the protonated acyl guanidine NH2, with an additional contact
between the acyl NH and Asp 32. The benzylic aryl ring occupied
the S1 pocket, and the acetamide was forced into a perpendicular
conformation by the flanking chlorine atoms. The flap of the
enzyme was in the closed conformation, with a H-bond between
Gln 73 and the acyl guanidine carbonyl oxygen. The pyrrolidine
stereocenter induced a bend in the molecule, yielding a global
U-shaped conformation which favorably directed the aryl group
into S2. Consequently, the stereochemistry of the pyrrolidine was
important, and the (S) enantiomer was favored. Specifically,
compound 7c was seven-fold more potent than its enantiomer
7d at pH 6.4 (Table 1). The pyrrolidine ring was located within a
modest-sized pocket, opening the possibility of substitution on this
ring. Also, the aryl ring was proximal to Arg 235, thus motivating
us to make additional analogs to optimize the electronics of the
aryl ring to maximize a predicted pi-cation interaction (Table 2).
Utilizing the same Ullman procedure described for the synthesis
of 2c (Scheme 1), appropriately substituted aryl bromides were
coupled to proline to afford analogs 12a–f. Within this set of aryl
substitutions, the p-methyl and p-fluoro groups slightly dimin-
ished potency (12a and 12c). Somewhat surprisingly, further
manipulation of the electronics of the aryl ring led to very little dif-
ference in potency. With the exception of 12c, these analogs had
small pH shifts, and the pH 6.4 Ki’s and EC50’s were similar to the
parent phenyl analog 7c. The lack of a clear SAR trend based on
electronic factors did not support the hypothesis of a significant
pi-cation interaction, however we cannot exclude the possibility
that additional conformational or other factors may be influencing
the dataset. No other attempts were made to pick up interaction
with Arg235 were made, however the p-methoxy derivative 12e
was equipotent to the parent 7c both in vitro and in the cell-based
assay, and the p-ether linkage was therefore used to append
additional functionality (vide infra).
Table 1 exemplifies some of the isoxazole/isothiazole core
replacements we examined, and the BACE1 binding affinities at
both pH 5 and 6.4 are shown. The ‘pH shift’ [Ki (pH 6.4)/Ki
(pH 5.0)] was calculated, and additionally, the inhibitor’s calcu-
lated pKa17 was determined to provide a crude point of reference.
Serving as controls, the aromatic pyrrole and pyrazole derivatives
7a and 7b had large pH shifts (5- to 24-fold), consistent with hav-
ing pKa values under six. While the guanyl urethane derivative 11a
was not potent at pH 6.4, the analogous alpha-branched guanyl
urea 11b had improved pH 6.4 activity, characterized by a low
pH shift and higher pKa (7.2). Attempts to further improve potency
by cyclization to the indane/tetrahydronaphthalene motifs (11d
and 11e) resulted in an overall loss in potency (Ki’s of ꢀ10
lM),
although the absence of an appreciable pH shift was consistent
Table 1
Alternative core replacements of the acylguanidine chemotype
NH2
Cl
R1
O
N
H
N
N
H
Cl
BACE-1 Ki
Compound R1
BACE-1 Ki
pH
Calcd
(
l
M), pH
(lM), pH
shiftb pKa
5.0a
6.4a
Ph
N
O
7a
7b
7c
0.49
12
24
5.2
6.0
6.4
N
Ph
N
O
O
O
0.71
0.21
4
5.6
1.6
Ph
N
0.34
Ph
N
7d
3.1
7
2.5
>30
2.4
0.8
>4
6.4
n.d.
7.2
Examining our binding model, we predicted substituents were
likely to be tolerated at the pyrrolidine 4-position. Additionally,
4-substituted pyrrolidines could be purchased or made in a few
steps from readily available 4-hydroxyproline, allowing quick
access to a pool of prospective templates. Toward that end, addi-
tional analogs were prepared by the method previously described
in Scheme 1. As we examined these templates (Table 3), it was
clear that large substituents in the 4-(S) position (7i and 7j)
resulted in a 10-fold loss in potency. Hydroxyl substituents at
either diastereomeric position (7e and 7f) were three- to four-fold
less potent than the parent. Ether functionality at the 4(R) position
(7g and 7k) was tolerated. The piperidine analog 7m (as a race-
mate) was essentially equipotent to the pyrrolidine analog. In the
absence of an obvious advantage, additional piperidine analogs
were not pursued. Bicyclic derivatives 7l and 7n also had similar
Ph
O
O
11a
11b
O
Ph
0.99
2.4
Me
N
H
Ph-4-OMe
O
O
11c
11d
2.3
10
n/a
14
n/a
6.6
7.1
N
ꢀ1
N
H
H
H
activity at pH 6.4 (ꢀ0.4
lM) as the other analogs in this cohort.
In an effort to further optimize the pyrrolidine series, we
appended the 4-dimethylglycinamide substituent on the P1 aryl
group (Table 4). In the related isoxazole and isothiazole series, this
modification led to improved cellular activity compared to the
acetamide.9 Surprisingly, the dimethylglycine derivative on
the pyrrolidine template was essentially equipotent with the
11e
O
13
9.1
ꢀ1
7.1
N
H
a
For Ki<1
l
M, n P 2.
b
pH shift = Ki (pH 6.4)/Ki (pH 5.0).

K. M. Boy et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5040–5047
5043
H
N
H
N
Asp228
N
Asp32
NH2+ O
Phe108
Cl
Cl
7c
NHAc
Arg235
Gln73
Lys107
Figure 1. Model of 7c.
Table 3
Table 2
Survey of Aryl Substituents
Pyrrolidine analogs of the acylguanidine chemotype
NH2
R1
Cl
R
O
N
H
N
NH2
O
Cl
O
N
N
H
N
H
N
Cl
N
H
Compound R1
X
Y
BACE-1 Ki
M), pH 5.0a
BACE-1 Ki
(l
M), pH 6.4a
Cl
(
l
a
Compound
R
BACE-1 Ki (
lM),
BACE-1 Ki (
lM),
Cell EC50
M)
7c
7e
7f
H
OH
H
H
H
OH
0.21
0.79
0.61
0.34
Ph
N
pH 5.0a
pH 6.4a
(
l
O
7c
H
p-Me
p-Cl
0.21
1.4
0.34
3.4
0.37
7.2
0.60
0.30
0.63
0.68
7g
7h
7i
7j
7k
H
OMe 0.078
0.39
12a
12b
12c
12d
12e
12f
X
Ph
OPh
OBn
H
H
H
0.53
1.9
0.15
0.45
0.13
0.11
0.26
0.58
Y
p-F
H
1.8
p-CF3
p-OMe
m-OMe
1.7
0.52
0.55
OBn
0.42
Ph
N
O
a
7l
—
—
—
—
0.56
0.23
For Ki’s and EC50’s < 1
l
M, n P 2.
Ph
N
O
O
acetamide, and still several-fold less active than the corresponding
isothiazole. This suggests that the glycine fragment did not serve
the same function on the more basic pyrrolidine scaffold as it did
on the less basic aryl scaffolds. The exact reason for this difference
remains unclear, although one could speculate that having a basic
amine in the molecule (glycinamide or pyrrolidine) confers an
advantage in the acidic environment where BACE is located. What
was also apparent was that the pyrrolidine analogs offered
advantages with respect to P-gp efflux.
7m
0.42
0.36
Ph
N
7n
—
—
0.10
a
For Ki’s < 1 lM, n P 2.
The isoxazole and isothiazole compounds 23–26 had a large
P-gp liability when either the acetamide or glycinamide moieties
were present (Table 4) as measured by monitoring Ab production
in a cell-based assay using KBV cells selected to overexpress
P-gp.19 This system provided a direct readout of our desired bio-
marker, and thus was more useful to us than separate permeability
measures such as the Caco and MDCK cell monolayer systems. In
contrast, the acetamide of the pyrrolidine (12e) had diminished
P-gp efflux (P-gp shift ꢀ6). Furthermore, the P-gp shift of the glyci-
namide pyrrolidine derivative 12g was even lower (P-gp shift = 3).
Despite the lower binding affinity of the pyrrolidines, we viewed
the P-gp profile favorably, and looked for other ways to improve
the potency of these compounds.
From our model of 7c (Fig. 1), the 4-position of the pyrrolidine
N-aryl group of this series of acylguanidine inhibitors was in
close proximity to the glycinamide functionality as a result of the
U-shaped topography of the bound conformation. Additionally,
the SAR previously described indicated that an ether functionality
on the aryl ring was well tolerated. Taken together, these
observations formed the basis for pursuing a macrocyclization
approach.20,10j
Macrocyclic acyl guanidines 20a–c and 21b–c were assembled
as shown in Scheme 3. The preparation of intermediate 15 was
performed in analogous fashion to that previously described for
intermediates 6. Compound 15 was further elaborated by reaction

5044
K. M. Boy et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5040–5047
Table 4
with bromoacetyl bromide, followed by displacement with pri-
The effect of the core group on P-gp mediated efflux
mary 1-aminoalkenes of various lengths. The resulting secondary
amines 17a–c were subsequently protected as N-Boc carbamates
to allow macrocyclization utilizing Grubbs 2nd generation cata-
lyst21 at elevated temperature. The protected macrocycles 19 were
then deprotected using TFA to afford the unsaturated macrocycles
20a–c. Compounds 19 could optionally be hydrogenated using bal-
loon pressure H2 and Pd/C and subsequently deprotected with TFA
to provide the saturated macrocycles 21b–c.
Gratifyingly, the macrocyclization strategy proved effective
(Table 5). While the 24-membered macrocyclic alkene 20a was
not very potent, we observed an order of magnitude improvement
in binding potency for the 25-membered alkyl and alkenyl macro-
cycles 20b (Ki = 20 nM) and 21b (Ki = 43 nM), with the alkenyl
compound 20b having a relatively small P-gp shift (2.1). Further
ring homologation to the 26-membered ring compounds revealed
that the alkenyl variant 20c was essentially equipotent to the 25-
membered analog (Ki = 25 nM), and the saturated macrocycle 21c
was approximately another log unit more potent (Ki = 3 nM).
Importantly, the P-gp shift of compound 21c was also low (2.3).
The cellular potency of the macrocycles 20b, 21b, and 20c were
around 10-fold less active than the binding activity (EC50’s =
OMe
NH2
N
O
Cl
N
N
H
Het
R
N
H
Cl
Compound
Heterocycle
(Het)
R
BACE-1
Cell
P-gp
Ki (
l
M),
EC50
shiftb
pH 5.0a
(l
M)a
23c
24c
Ac
0.013
0.051
0.79
0.18
>3.8
11
C(O)CH2NMe2
N
O
25c
26c
Ac
0.0058
0.020
0.29
0.065
>17
20
C(O)CH2NMe2
N
S
0.2–0.5
(EC50 = 0.14
l
M). Compound 21c was also the most active in cells
12e
12g
Ac
0.11
0.14
0.52
0.42
5.8
3.0
C(O)CH2NMe2
l
M), however the cellular potency did not increase
N
as much as the binding activity for reasons we did not investigate
further. We created a model of 21c (Fig. 2), and the bound confor-
mation showed the macrocycle sitting in a crown-type fashion
over Gln 73 with no adverse interactions arising from the tether.
Excellent selectivity against the aspartyl proteases pepsin and
a
b
c
For Ki’s and EC50’s < 1 lM, n P 2.
Determined in KBV cells tariquidar (P-gp inhibitor), see Ref. 19.
Compounds 23–26 are described in Ref. 9.
O
O
O
O
H
N
a
c
b
Boc
HO
NH
O
O
S
O
N
Cl
N
N
Boc
N
H
N
HO
L-Proline
N
H
N
H2N
13
14
15
Cl
O
N
O
N
e
d
f
Boc
Boc
NH
O
NH
O
Cl
Cl
O
N
H
N
O
N
H
N
n
Br
HN
N
N
17a n=1
H
16
H
17b n=2
17c n=3
Cl
Cl
R
N
Boc
N
O
O
H
N
O
n
O
n
O
n
O
HN
HN
HN
Cl
Cl
Cl
g
h,i
Cl
Cl
Cl
Boc
N
Boc
N
HN
O
HN
O
NH2
O
18a n=1
18b n=2
18c n=3
N
N
N
N
N
H
N
H
N
H
21b n=2
21c n=3
20a n=1, R=H
20b n=2, R=H
20c n=3, R=H
19a n=1, R=Boc
19b n=2, R=Boc
19c n=3, R=Boc
i
Scheme 3. Synthesis of macrocycles 20a–c and 21b–c. Reagents and conditions: (a) allyl p-bromophenyl ether, CuO, K2CO3, 160 °C; (b) 1-N-Boc-2-methyl isothiourea, HATU,
N-methylmorpholine, DMF, 41% yield from L-proline; (c) 4-(aminomethyl)-2,6-dichloroaniline, CH2Cl2, DIPEA, 94% yield; (d) bromoacetyl bromide, triethylamine, CH2Cl2; (e)
1-aminoalkene, Et3N, CH2Cl2, 71–75% yield from 15; (f) Boc2O, NaOH, THF/H2O, ꢀ74% yield; (g) Grubb’s 2nd generation catalyst,20 dichloroethane, rt, ꢀ35% yield; (h) H2, Pd/C,
MeOH, balloon pressure, 84–86% yield; (i) TFA, CH2Cl2, 42–78% yield.

K. M. Boy et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5040–5047
5045
Table 5
Survey of macrocycle size
O
H
N
L
O
N
HN
Cl
Cl
NH2
N
O
N
H
a
Compound
L
Ring Size BACE-1 Ki (
l
M), pH 5.0a BACE-1 Ki (
lM), pH 6.4
pH shift Cell EC50 (l
M)a Cell EC50/Ki P-gp shiftb
HN
20a
24
0.22
1.5
6.8
0.88
1.7
4.0
O
HN
HN
HN
O
O
20b
21b
25
25
0.020
0.043
0.11
5.5
2.1
0.23
0.57
2.1
6.2
2.1
6.2
0.092
20c
21c
26
26
0.025
0.059
n.d.
2.4
0.49
0.14
8.3
8.3
2.3
O
O
HN
0.0032
n.d.
n.d.
a
For Ki’s and EC50’s <1
Determined in KBV cells tariquidar (P-gp inhibitor), see Ref. 19.
l
M, n P 2.
b
Asp228
Phe108
Asp32
Arg235
Gln73
Lys107
Thr329
Figure 2. Model of macrocycle 21c.
cathepsin E was observed for all macrocycles shown, and the
selectivity versus cathepsin D was >40-fold for these compounds
(compound 21c was 700-fold selective vs cathepsin D). Addition-
ally, hERG ion channel blockade as measured in a thallium flux
assay was >13 lM for the series (32 lM for 21c). Having success-
fully created a potent analog with diminished P-gp efflux in KBV
with a peripheral drug concentration of 3.8 lM. The free fraction
of the compound was 5.0% in mouse serum, implying a free drug
concentration of 190 nM. This response is in reasonable agreement
with the cellular potency (140 nM). Measurements of Ab40 in the
brain, however, were at the level of the control group, and levels
of drug were just above the limit of detection. The lack of brain per-
meability observed for 21c suggested that either P-gp shifts of less
than ꢀ2 might be necessary to predict good CNS penetration alter-
natively, the physiochemical properties of these compounds may
need additional refinement for adequate brain penetration. It is
recognized, however, that macrocyclization increases the
propensity of a given molecule to be permeable compared to that
predicted for the related acyclic structure.24 As a follow-up to this
result, permeability was also measured using the Caco-2 cell
monolayer assay. The results verified reasonable passive perme-
ability, but also active efflux. The 8-fold ratio of basolateral to
cells, compound 21c was then further tested in vivo.
Compound 21c was studied in male FVB wild-type mice,22 with
plasma and brain Ab40 lowering (as
a surrogate for Ab42
lowering23) as the measured response. To maximize throughout
the program, sc dosing was chosen to provide robust exposure of
the compound of interest. Samples were collected 5 h after sc
administration of the compound as a solution in 70/30 PEG400/
water, and analyzed in a chemiluminescence ELISA assay utilizing
the rodent-specific 252Q8 monoclonal antibody. At 40 mpk,
plasma Ab was reduced to 26% of control (74% reduction of Ab40),

5046
K. M. Boy et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5040–5047
Bruinzeel, W. D.; Jordan, A. D.; Parker, M. H.; Boyd, R. E.; Qu, J.; Alexander, R. S.;
Brenneman, D. E.; Reitz, A. B. J. Med. Chem. 2007, 50, 4261; (c) Fobare, W. F.;
Solvibile, W. R.; Robichaud, A. J.; Malamas, M. S.; Manas, E.; Turner, J.; Hu, Y.;
Wagner, E.; Chopra, R.; Cowling, R.; Jin, G.; Bard, J. Bioorg. Med. Chem. Lett. 2007,
17, 5353; (d) Cole, D. C.; Stock, J. R.; Chopra, R.; Cowling, R.; Ellingboe, J. W.;
Fan, K. Y.; Harrison, B. L.; Hu, Y.; Jacobsen, S.; Jennings, L. D.; Jin, G.; Lohse, P. A.;
Malamas, M. S.; Manas, E. S.; Moore, W. J.; O’Donnell, M. M.; Olland, A. M.;
Robichaud, A. J.; Svenson, K.; Wu, J.; Wagner, E.; Bard, J. Bioorg. Med. Chem. Lett.
2008, 18, 1063; (e) Barrow, J. C.; Stauffer, S. R.; Rittle, K. E.; Ngo, P. L.; Yang, Z.;
Selnick, H. G.; Graham, S. L.; Munshi, S.; McGaughey, G. B.; Holloway, M. K.;
Simon, A. J.; Price, E. A.; Sankaranarayanan, S.; Colussi, D.; Tugusheva, K.; Lai,
M. T.; Espeseth, A. S.; Xu, M.; Huang, Q.; Wolfe, A.; Pietrak, B.; Zuck, P.; Levorse,
D. A.; Hazuda, D.; Vacca, J. P. J. Med. Chem. 2008, 51, 6259; (f) Malamas, M. S.;
Erdei, J.; Gunawan, I.; Barnes, K.; Johnson, M.; Hui, Y.; Turner, J.; Hu, Y.;
Wagner, E.; Fan, K.; Olland, A.; Bard, J.; Robichaud, A. J. J. Med. Chem. 2009, 52,
6314; (g) Malamas, M. S.; Barnes, K.; Johnson, M.; Hui, Y.; Zhou, P.; Turner, J.;
Hu, Y.; Wagner, E.; Fan, K.; Chopra, R.; Olland, A.; Bard, J.; Pangalos, M.;
Reinhart, P.; Robichaud, A. J. Bioorg. Med. Chem. 2010, 18, 630; (h) Zhu, Z.; Sun,
Z.-Y.; Ye, Y.; Voigt, J.; Strickland, C.; Smith, E. M.; Cumming, J.; Wang, L.; Wong,
J.; Wang, Y.-S.; Wyss, D. F.; Chen, X.; Kuvelkar, R.; Kennedy, M. E.; Favreau, L.;
Parker, E.; McKittrick, B. A.; Stamford, A.; Czarniecki, M.; Greenlee, W.; Hunter,
J. C. J. Med. Chem. 2010, 53, 951; (i) Malamas, M. S.; Erdei, J.; Gunawan, I.;
Turner, J.; Hu, Y.; Wagner, E.; Fan, K.; Chopra, R.; Olland, A.; Bard, J.; Jacobsen,
S.; Magolda, R. L.; Pangalos, M.; Robichaud, A. J. J. Med. Chem. 2010, 53, 1146; (j)
Huang, Y.; Strobel, E. D.; Ho, C. Y.; Reynolds, C. H.; Conway, K. A.; Piesvaux, J. A.;
Brenneman, D. E.; Yohrling, G. J.; Arnold, H. M.; Rosenthal, D.; Alexander, R. S.;
Tounge, B. A.; Mercken, M.; Vandermeeren, M.; Parker, M. H.; Reitz, A. B.;
Baxter, E. W. Bioorg. Med. Chem. Lett. 2010, 20, 3158; (k) Malamas, M. S.;
Robichaud, A.; Erdei, J.; Quagliato, D.; Solvibile, W.; Zhou, P.; Morris, K.; Turner,
J.; Wagner, E.; Fan, K.; Olland, A.; Jacobsen, S.; Reinhart, P.; Riddell, D.;
Pangalos, M. Bioorg. Med. Chem. Lett. 2010, 20, 6597; (l) Tresadern, G.; Delgado,
F.; Delgado, O.; Gijsen, H.; Macdonald, G. J.; Moechars, D.; Rombouts, F.;
Alexander, R.; Spurlino, J.; Van Gool, M.; Vega, J. A.; Trabanco, A. A. Bioorg. Med.
Chem. Lett. 2011, 21, 7255; (m) Cumming, J. N.; Smith, E. M.; Wang, L.;
Misiaszek, J.; Durkin, J.; Pan, J.; Iserloh, U.; Wu, Y.; Zhu, Z.; Strickland, C.; Voigt,
J.; Chen, X.; Kennedy, M. E.; Kuvelkar, R.; Hyde, L. A.; Cox, K.; Favreau, L.;
Czarniecki, M. F.; Greenlee, W. J.; McKittrick, B. A.; Parker, E. M.; Stamford, A.
W. Bioorg. Med. Chem. Lett. 2012, 22, 2444.
apical (380 nm/s) versus apical to basolateral (48 nm/s) permeabil-
ity demonstrated that 21c was an efflux substrate, a result in
greater alignment with the observed brain data.25 It remains a pos-
sibility that a transporter other than P-gp was responsible for the
difference between the observed results in the KBV versus the
Caco-2 assays, but this matter was not pursued further.
In summary, a novel class of potent acyclic and macrocyclic acyl
guanidine inhibitors of BACE were discovered. The introduction of
the pyrrolidine core of these compounds imparted potency at both
low and neutral pH, cellular activity, and a decreased liability with
respect to P-gp efflux over earlier acyl guanidine compounds. A
macrocyclization approach increased potency, both in vitro and
in cellular assays, enabling robust peripheral Ab lowering in vivo.
However, further effort to lower P-gp and optimization of CNS pen-
etration are still required to achieve meaningful central exposure.
The identification of brain-penetrant, non-peptidomimetic BACE
inhibitors will be the subject of future articles.
Acknowledgements
The authors appreciate the contributions of the following indi-
viduals. The original acyl guanidine X-ray crystal structure was
obtained by Jodi K. Muckelbauer and Daniel M. Camac, made pos-
sible by the BACE-1 constructs generated by Paul Morin, Dianlin
Xie, Yaqun Zhang, and Deepa Calambur. The Ab studies and assays
were performed by Nina Hoque, Mark Thompson, Lawrence Iben,
and Cathy Kieras.
References and notes
11. The in vitro and cellular potencies in our isostere series were equivalent for the
most active compounds, and well within 5-fold for most other analogs: (a)
Thompson, L. A.; Shi, J.; Decicco, C. P.; Tebben, A. J.; Olson, R. E.; Boy, K. M.;
Guernon, J. M.; Good, A. C.; Liauw, A.; Zheng, C.; Copeland, R. A.; Combs, A. P.;
Trainor, G. L.; Camac, D. M.; Muckelbauer, J. K. Bioorg. Med. Chem. Lett. 2011, 21,
6909; (b) Boy, K. M.; Guernon, J. M.; Wu, Y.-J.; Zhang, Y.; Gerritz, S. W.; Toyn, J.
H.; Meredith, J. E.; Barten, D. M.; Burton, C. R.; Albright, C. F.; Good, A. C.;
Muckelbauer, J. K.; Camac, D. M.; Grace, J. E.; Lentz, K. A.; Olson, R. E.; Macor, J.
E.; Thompson, L. A. Bioorg. Med. Chem. Lett. 2011, 21, 6916.
1. Thies, W.; Bleiler, L. Alzheimer’s Dementia 2012, 8, 131.
2. Salomone, S.; Caraci, F.; Leggio, G. M.; Fedotova, J.; Drago, F. Br. J. Clin. Pharm.
2012, 73, 504.
3. Brorsson, A.-C.; Bolognesi, B.; Tartaglia, G. G.; Shammas, S. L.; Favrin, G.;
Watson, I.; Lomas, D. A.; Chiti, F.; Vendruscolo, M.; Dobson, C. M.; Crowther, D.
C.; Luheshi, L. M. Biophys. J. 2010, 98, 1677.
4. (a) Hardy, J. J. Neurochem. 2009, 110, 1129; (b) Hardy, J.; Selkoe, D. J. Science
2002, 297, 353.
12. Iben, L. G.; Kopcho, L.; Marcinkeviciene, J.; Zheng, C.; Thompson, L. A.; Albright,
C. F.; Toyn, J. H. Eur. J. Pharmacol. 2008, 593, 10.
5. (a) Walsh, D. M.; Selkoe, D. J. J. Neuorochem. 2007, 101, 1172; (b) Haass, C.;
Selkoe, D. J. Mol. Cell Biol. 2007, 8, 101.
13. Vassar, R.; Bennett, B. D.; Babu-Khan, S.; Kahn, S.; Mendiaz, E. A.; Denis, P.;
Teplow, D. B.; Ross, S.; Amarante, P.; Leoloff, R.; Luo, Y.; Fisher, S.; Fuller, J.;
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. Science 1999, 286, 735.
14. Subsequent to our research, a similar pH dependence was disclosed for a series
of aminoheterocycles: Stachel, S. J.; Coburn, C. A.; Rush, D.; Jones, K. L. G.; Zhu,
H.; Rajapakse, H.; Graham, S. L.; Simon, A.; Holloway, M. K.; Allison, T. J.;
Munshi, S. K.; Espeseth, A. S.; Zuck, P.; Colussi, D.; Wolfe, A.; Pietrak, B. L.; Lai,
M.-T.; Vacca, J. P. Bioorg. Med. Chem. Lett. 2009, 19, 2977.
15. Dominguez, J. L.; Christopeit, T.; Villaverde, M. C.; Gossas, T.; Otero, J. M.;
Nystrom, S.; Baraznenok, V.; Lindstrom, E.; Danielson, U. H.; Sussman, F.
Biochemistry 2010, 49, 7255.
16. For a review of the Ullman reaction, see Ley, S. V.; Thomas, A. W. Angew. Chem.,
Int. Ed. 2003, 42, 5400.
6. (a) Luo, Y.; Bolon, B.; Kahn, S.; Bennett, B. D.; Babu-Khan, S.; Denis, P.; Fan, W.;
Kha, H.; Zhang, J.; Gong, Y.; Martin, L.; Louis, J.-C.; Yan, Q.; Richards, W. G.;
Citron, M.; Vassar, R. Nat. Neurosci. 2001, 4, 231; (b) Roberds, S. L.; Anderson, J.;
Basi, G.; Bienkowski, M. J.; Branstetter, D. G.; Chen, K. S.; Freedman, S. B.;
Frigon, N. L.; Games, D.; Hu, K.; Johnson-Wood, K.; Kappenman, K. E.; Kawaba,
T. T.; Kola, I.; Kuehn, R.; Lee, M.; Liu, W.; Motter, R.; Nichols, N. F.; Power, M.;
Robertson, D. W.; Schenk, D.; Schoor, M.; Shopp, G. M.; Shuck, M. E.; Sinha, S.;
Svensson, K. A.; Tatsuno, G.; Tintrup, H.; Wijsman, J.; Wright, S.; McConlogue, L.
Hum. Mol. Gen. 2001, 10, 1317; (c) Harrison, S. M.; Harper, A. J.; Hawkins, J.;
Duddy, G.; Grau, E.; Pugh, P. L.; Winter, P. H.; Shilliam, C. S.; Hughes, Z. A.;
Dawson, L. A.; Gonzalez, M. I.; Upton, N.; Pangalos, M. N.; Dingwall, C. Mol. Cell.
Neuorosci. 2003, 24, 646.
7. (a) Ohno, M.; Cole, S. L.; Yasvoina, M.; Zhao, J.; Citron, M.; Berry, R.; Disterhoft, J.
F.; Vassar, R. Neurobiol. Dis. 2007, 26, 134; (b) Laird, F. M.; Cai, H.; Savonenko, A.
V.; Farah, M. H.; He, K.; Melnikova, T.; Wen, H.; Chiang, H.-C.; Xu, G.; Koliatsos,
V. E.; Borchelt, D. R.; Price, D. L.; Lee, H.-K.; Wong, P. C. J. Neuorosci. 2005, 25,
11693; (c) Hu, H.; Zhou, X.; He, W.; Yang, J.; Xiong, W.; Wong, P.; Wilson, C. G.;
Yan, R. J. Neurosci. 2010, 30, 8819; (d) Kobayashi, D.; Zeller, M.; Cole, T.; Buttini,
M.; McConlogue, L.; Sinha, S.; Freedman, S.; Morris, R. G. M.; Chen, K. S.
Neurobiol. Aging 2008, 29, 861; (e) Dominguez, D.; Tournoy, J.; Hartmann, D.;
Huth, T.; Cryns, K.; Deforce, S.; Serneels, L.; Camacho, I. E.; Marjaux, E.;
Craessaerts, K.; Roebroek, A. J. M.; Schwake, M.; D’Hooge, R.; Bach, P.; Kalinke,
U.; Moechars, D.; Alzheimer, C.; Reiss, K.; Saftig, P.; De Strooper, B. J. Biol. Chem.
2005, 280, 30797.
17. ACD version 10.
18. The
co-crystal
structure
of
(E)-N-(amino(3,5-dichlorobenzylamino)
methylene)-3-(4-methoxyphenyl)-5-methylisoxazole-4-carboxamide in BACE
was used to guide the selection of potential modifications to improve potency.
The addition of an aniline functionality as found in (E)-N-(amino(4-amino-3,5-
dichlorobenzylamino)methylene)-3-(4-methoxyphenyl)-5-methylisothiazole-
4-carboxamide was hypothesized to form an additional H-bond to the carbonyl
of F108, which was subsequently verified by X-ray diffraction. The model was
constructed in Maestro 7.0 (Schrödinger LLC, New York, NY) and minimized
using Macromodel (version 9.8, Schrödinger LLC, New York, NY).
19. Meredith, J. E.; Thompson, L. A., III; Toyn, J. H.; Marcin, L.; Barten, D. M.;
Marcinkeviciene, J.; Kopcho, L.; Kim, Y.; Lin, A.; Guss, V.; Burton, C.; Iben, L.;
Polson, C.; Cantone, J.; Ford, M.; Drexler, D.; Fiedler, T.; Lentz, K. A.; Grace, J. E.,
Jr.; Kolb, J.; Corsa, J.; Pierdomenico, M.; Jones, K.; Olson, R. E.; Macor, J. E.;
Albright, C. F. J. Pharmacol. Exp. Ther. 2008, 326, 502.
8. Ghosh, A. K.; Osswald, H. Chem. Soc. Rev. 2014, 43, 6765.
9. Gerritz, S. W.; Zhai, W.; Shi, S.; Zhu, S.; Toyn, J. H.; Meredith, J. E., Jr.; Iben, L. G.;
Burton, C. R.; Albright, C. F.; Good, A. C.; Tebben, A. J.; Muckelbauer, J. K.;
Camac, D. M.; Metzler, W.; Cook, L. S.; Padmanabha, R.; Lentz, K. A.; Sofia, M. A.;
Macor, J. E.; Thompson, L. A., III J. Med. Chem. 2012, 55, 9208.
20. For other examples of macrocyclic BACE-1 inhibitors, see (a) Lerchner, A.;
Machauer, R.; Betschart, C.; Veenstra, S.; Rueeger, H.; McCarthy, C.; Tintelnot-
Blomley, M.; Jaton, A.-L.; Rabe, S.; Desreyaud, S.; Enz, A.; Staufenbiel, M.;
Paganetti, P.; Rondeau, J.-M.; Neuman, U. Bioorg. Med. Chem. Lett. 2010, 20, 603;
(b) Sandgren, V.; Agback, T.; Johansson, P.-O.; Lindberg, J.; Kvarnstrom, I.;
Samuelsson, B.; Belda, O.; Dahlgren, A. Bioorg. Med. Chem. 2012, 20, 4377.
21. [1,3-Bis-(2,4,6-trimethylphenyl)-2-imidazolidinylidene]dichloro
(phenylmethylene) (tricyclohexylphosphine)ruthenium.
10. Other acyl-guanidine chemotypes have emerged in the literature. See (a) 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.; 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.; Zhou, P.; Bard,
J. J. Med. Chem. 2006, 49, 6158; (b) Baxter, E. W.; Conway, K. A.; Kennis, L.;
Bischoff, F.; Mercken, M. H.; Winter, H. L.; Reynolds, C. H.; Tounge, B. A.; Luo, C.;
Scott, M. K.; Huang, Y.; Braeken, M.; Pieters, S. M.; Berthelot, D. J.; Masure, S.;

K. M. Boy et al. / Bioorg. Med. Chem. Lett. 25 (2015) 5040–5047
5047
22. Charles River Laboratories http://www.criver.com/EN-US/PRODSERV/BYTYPE/
RESMODOVER/RESMOD/Pages/FVBMouse.aspx (accessed 30 November 2010).
23. Due to matrix effects and the relatively small amount of Ab42 present in the
wild-type mice, the more prevalent Ab40 species was monitored. BACE-1
inhibitors block the production of both species.
24. (a) Yudin, Y. K. Chem. Sci. 2015, 6, 30; (b) Pajouhseh, H.; Lenz, G. R. NeuroRx
2005, 2, 541.
25. The des-methoxy congener of 12g, a non-macrocyclic prolinyl acyl guanidine,
also had a more pronounced efflux ratio when measured in Caco cells (11-fold)
than in the KBV assay (3.9-fold).