Macrocyclic peptidomimetic inhibitors of β-secretase (BACE): First X-ray structure of a macrocyclic peptidomimetic-BACE complex

Article abstract of DOI:10.1016/j.bmcl.2005.09.003

The synthesis of novel macrocyclic peptidomimetic inhibitors of the enzyme BACE1 is described. These macrocycles are derived from a hydroxyethylene core structure. Compound 7 was co-crystallized with BACE1 and the X-ray structure of the complex elucidated at 1.6 A resolution. This molecule inhibits the production of the Aβ peptide in HEK293 cells overexpressing APP751sw.

Full text of DOI:10.1016/j.bmcl.2005.09.003

Bioorganic & Medicinal Chemistry Letters 16 (2006) 191–195
Macrocyclic peptidomimetic inhibitors of b-secretase
(BACE): First X-ray structure of a macrocyclic
peptidomimetic-BACE complex
a
a
b
b
Isabel Rojo,a, Jose Alfredo Martın, Howard Broughton, David Timm, Jon Erickson,
*
´
´
Hsiu-Chiung Yangb and James R. McCarthyb
aLilly S.A., Avenida de la Industria 30, 28108 Alcobendas, Spain
bLilly Research Laboratories, Lilly Corporate Center, Indianapolis, IN 46285, USA
Received 16 February 2005; revised 7 September 2005; accepted 8 September 2005
Available online 24 October 2005
Abstract—The synthesis of novel macrocyclic peptidomimetic inhibitors of the enzyme BACE1 is described. These macrocycles are
derived from a hydroxyethylene core structure. Compound 7 was co-crystallized with BACE1 and the X-ray structure of the com-
˚
plex elucidated at 1.6 A resolution. This molecule inhibits the production of the Ab peptide in HEK293 cells overexpressing
APP751sw.
Ó 2005 Elsevier Ltd. All rights reserved.
AlzheimerÕs disease (AD) is a major neurodegenerative
disorder that affects 10% of the population over 65%
and 40% over 80 years.1,2 Symptoms of the disease are
behavioral disturbances and loss of cognitive function.
AD represents an unmet medical need that adds approx-
imately 100 billion dollars to healthcare costs in the
United States alone. In addition, AD also causes great
deal of suffering to the patient and places a burden on
their families.
AD is characterized by the presence of amyloid plaques
that are believed to produce neuronal toxicity and cell
death.3 These plaques are constituted mainly from an
insoluble form of Ab amyloid, a 40–42 amino acid pep-
tide produced in the brain from the amyloid precursor
protein (APP). APP is processed by at least three secre-
tases. a-Secretase cleaves APP to form a carboxy termi-
nal 83 amino acid fragment (C83), which is processed by
c-secretase to produce p3, a non amyloidogenic form of
Ab–peptide. Alternatively, APP can be processed initial-
ly by b-secretase (BACE1) and the resulting C99 mem-
brane bound C-terminal peptide can be hydrolyzed by
c-secretase to form the 40 amino acid Ab-peptide
(Ab40). Shorter and longer forms of the peptide also
are produced, in particular Ab42, which has a high pro-
pensity to aggregate and is the principal Ab species
found in amyloid plaques. Notably, the BACE1 enzyme
is only found in ER/Golgi or endosome compartments
of neurons, whereas several a-secretase candidates and
the c-secretase complex have broad tissue distribution.
Treatments currently available for AD are cholinester-
ase inhibitors and NMDA-receptor antagonists. The
rationale behind the former approach is that these
agents increase cholinergic transmission by interfering
with degradation of the Ach neurotransmitter. Unfortu-
nately, this approach does not stop the progressive loss
of cholinergic neurons, and eventually the treatment
becomes ineffective. Treatment with NMDA-receptor
antagonists appears to be no more effective in terms of
progression of the disease. A better approach would
be to develop agents that interfere with the mechanism
that leads to neurodegeneration.
Conceptually, the inhibition of either b- or c-secretase
by a small molecule would preclude the formation of
Ab and therefore plaque formation. c-Secretase inhibi-
tors have been reported to attenuate the formation of
Ab and plaques in animal models. However, the thera-
peutic index of these inhibitors requires careful monitor-
ing. This is due to the implication of c-secretase in
Keywords: BACE; Macrocycle; X-ray.
*
Corresponding author. Tel.: +34 91 6233503; fax: +34 91 6633411;
e-mail: irojo@lilly.com
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.09.003

192
I. Rojo et al. / Bioorg. Med. Chem. Lett. 16 (2006) 191–195
P2
P1'
P3'
essential functions, including facilitating Notch signal-
ing, which is important for specifying cell fates during
development and for regulating differentiation of self-
renewing cell types in adults.4 In contrast, b-secretase
KO mice are viable and free of the gross phenotypic
changes that are observed, for example with presenilin
knock-outs, except for the complete absence of
Ab in their brains.5
O
O
OH
H
H
H
O
N
N
N
N
N
H
H
N
O
O
O
P2'
P3
N2
P1
1
Beta amyloid cleaving enzyme (BACE1) has recently
been identified as the principal b-secretase in neurons.6
It is a trans-membrane aspartyl protease expressed in
neuronal cells that cleaves APP inside vesicles associated
with ER/Golgi apparatus and endosomes. Tang et al.
reported in 2000 that OM99-2 (Fig. 1), a peptidomimetic
based on the hydroxyethylene transition state isostere,
was a nanomolar inhibitor of BACE. The Oklahoma
group subsequently reported the X-ray structure of the
OM99-2-BACE complex showing the main features of
the enzyme–inhibitor interactions.7
Figure 2.
mined by iterative docking studies13 that the best P1–P3
linkage was a six-atom long saturated chain that forms a
13-membered ring (including the inhibitor backbone).
For the P1–N2 connection, the five-carbon saturated
linker was predicted to be optimal, leading to a 10-mem-
bered ring.
The synthesis of the macrocyclic peptidomimetics was
performed as depicted in Scheme 1. N-Boc serine methyl
ester was converted into the iodoserine derivative by
treatment with iodine and imidazole. Conversion of
the alkyl iodide into the corresponding alkylzinc reagent
followed by alkylation with allyl chloride14 afforded,
after hydrolysis, intermediate 8.
With the goal of designing low molecular weight
BACE1 inhibitors,8 we envisioned that linking two resi-
dues of our small peptidomimetic inhibitors9 would pro-
vide a macrocyclic structure that would be more potent,
and would have improved absorption properties, rela-
tive to the corresponding open chain analog. The link-
age of distant residues should produce a decrease in
the number of possible low energy conformations, and
therefore, if one of the remaining conformations is com-
plementary to the enzyme active site, the binding affinity
should be increased. In addition, it is known that cyclic
peptides are more resistant to gastric tract degradation
adding an additional attractive feature to this class of
inhibitors.10
Compound 8 was transformed into the pivotal lactone 9
by reduction to N-Boc homoallylglycinal,15 alkylation
with the Zn–Cu reagent obtained from 3-iodopropionic
acid ethyl ester,16 thermal lactonization and, for
P10 = Me, stereoselective methylation with LHDMS
and methyl iodide.
The C-terminal fragment 10 was prepared by standard
EDCI coupling of N-Boc alanine and 4-aminomethyl
pyridine or benzylamine followed by cleavage of the
Boc group under acidic conditions. The N-terminal frag-
ment 11, used to prepare compounds 2–5, was prepared
by EDCI coupling of 8 and alanine methyl ester, fol-
lowed by ester hydrolysis.
Examination of TangÕs,7 and our X-ray data9 suggested
P1–P3 and P1–N2 as suitable points of union due to
their proximity, orientation, and localization into the
large P1–P3 lipophillic pocket (Fig. 2).
Linking positions P1–P3 has recently been described in
BACE1 inhibitors for structures derived from an etha-
nolamine core;11 also, other macrocycles derived from
a hydroxyethylene core structure have been synthesized
based on the linkage of positions P2–P3.12
The N-terminal fragment 12 used to build macrocyles 6–
7 was prepared starting with L-alanine. The p-nitro-
benzenesulfonyl group was introduced as an amino pro-
tecting group to facilitate the N–H alkylation with allyl
bromide in the presence of potassium carbonate. Depro-
tection of the N-allylsulfonamide by treatment with
thiophenol and potassium carbonate provided N-allyl-
Ala-(OMe).17 The coupling of this amino acid with
The strategy developed to build the macrocyclic struc-
tures was to modify our lead inhibitor 1 by Grubbs
cyclization of alkene moieties at P1 and P3 or between
P1 and the nitrogen of the P2 amino acid. It was deter-
HO
O
HO
O
NH
2
O
O
O
OH
O
H
N
H
N
H
N
H
N
N
H
OH
H
N
N
H
2
O
O
O
O
Figure 1. OM-99-2.

I. Rojo et al. / Bioorg. Med. Chem. Lett. 16 (2006) 191–195
193
O
O
O
O
O
i,j
H2N
a,b,c
O
N
BocHN
d,e,f,g,h
BocHN
BocHN
OH
H
H2N
OMe
OH
P1'
OH
X
10
9
8
O
O
m,n,o,p,q
BocHN
OH
N
k,l
BocHN
OH
OH
N
P3
O
8
BocHN
H
O
O
11
12
O
OH
O
P1'
O
O
O
r,s,t
OH
O
P1'
O
u
j
H
H
H
H
N
H
BocHN
N
N
BocHN
N
9
N
N
N
H
N
N
P1'
N
H
N
H
H
H
O
O
O
X
O
O
X
Cl
2,3,4
5
13
O
OH
O
P1'
O
O
O
y
H
H
H
N
v,w,x
BocHN
N
N
BocHN
N
N
9
N
P1'
H
O
O
X
P3
P3
O
6,7
14
Scheme 1. Reagents: (a) I2/PPh3/imidazole; (b) Zn/CuCN.LiCl/Allyl Chloride; (c) Cs2CO3; (d) NMM/i-BuOCOCl/MeNHOMe.HCl; (e) LiALH4; (f)
iodopropionic acid ethyl ester/Zn-Cu/DMA/Ti(IV); (g) acetic acid/110 °C; (h) for P1 = Me LHMDS/MeI; (i) 2-aminomethylpyridine or benzylamine/
EDCI/HOBT/TEA/DMAP; (j) HCl; (k) alanine methyl ester/EDCI/HOBT/TEA/DMAP; (l) Cs2CO3; (m) p-nitro-benzenesulfonyl chloride/TEA; (n)
K2CO3/Allyl bromide; (o) thiophenol/K2CO3; (p) BocILE or BocVAL/PyBrop/HOAT/DIPEA; (q) Cs2CO3; (r) 11/EDCI/HOBT/TEA/DMAP; (s)
bis(tricyclohexylphosphine)benzylideneruthenium (IV)dichloride; (t) H2; (u) 2-hydroxypyridine/10; (v) 12/EDCI/HOBT/TEA/DMAP; (w) bis(tri-
cyclohexylphosphine)benzylideneruthenium(IV)dichloride; (x) H2; (y) 2-hydroxypyridine/10.
N-Boc valine or N-Boc isoleucine was not possible under
our standard EDCI coupling conditions. However, the
coupling could be achieved in high yield using PyB-
roP/HOAT to afford 12.17
run at room temperature for 4 h. An increase in the rel-
ative fluorescence of the reaction mixture is used to
determine enzyme activity, with an umbilliferone excita-
tion/emission filter.
Both 11 and 12 were coupled with 9 to produce the diene
intermediates, which were submitted to Grubbs cyclica-
tion. The macrocyclization occurred in 50–70% yield,
and the resulting cycloalkenes were hydrogenated in
quantitative yield to produce 13 and 14, respectively.
Lactones 13 and 14 were opened with 10 by heating a
melted mixture of the two compounds in the presence
of 2-hydroxypyridine18 to produce the BACE inhibitors
2, 3, 4, 6 and 7. BACE inhibitor 5 was prepared by
cleavage of the Boc group of compound 3 under acidic
conditions.
The whole cell assay was performed in order to measure
the cellular permeability of the compounds. Endogenous
BACE1 is expressed in HEK293 cells. This cell line over
expresses the APP751swe protein. The assay is conduct-
ed with 35,000 cells per well in a 96-well plate format in
DMEM containing 10% FBS. Compounds are incubat-
ed with cells for 4 h at 37 °C and 5% CO2. The condi-
tioned media are removed from the culture wells.
The Ab ELISA (using monoclonal antibodies 266 and
biotinylated-3D6 as capture and reporting antibodies,
respectively) is used to measure the amount of Ab (total)
peptide produced during the incubation time.
Compounds 1–7 were tested for the inhibition of
BACE1 both in an in vitro FRET and a cellular assay,9
and the results are shown in Table 1. For the enzymatic
assay, IC50 values were determined using 20 nM purified
recombinant human BACE1:Fc. The assay contains
100 lM ELGY-9 (an aminobenzoate-based FRET pep-
tide containing Swedish mutation at the b-secretase
cleavage site) in 50 mM ammonium acetate, pH 4.6,
1 mg/ml BSA, and 1 mM Triton X-100. The assay is
We started our investigation by preparing the P1–P3
macrocyclic analogs. Compound 2 shows weak enzy-
matic activity and no cellular potency. In order to im-
prove these parameters, we reproduced the previously
reported SAR in the open chain analogs.9 Thus, intro-
ducing a methyl group at P10 (compound 3) improves
35-fold the enzymatic activity but still no cellular

194
I. Rojo et al. / Bioorg. Med. Chem. Lett. 16 (2006) 191–195
Table 1. In vitro values
Compound
P3
P10
X at P30
BACE FRET IC50 (lM)a
HEK293 Ab inhibition IC50 (lM)a
1
2
3
4
5
6
7
CHMe2
—
Me
H
N
H
H
N
H
N
N
0.082 ( 0.023)
9.704 ( 1.413)
0.285 ( 0.016)
4.925 ( 0.169)
2.397 ( 0.089)
0.260 ( 0.002)
0.065 ( 0.039)
2.831 ( 0.129)
>100
>100
—
—
Me
Me
Me
Me
Me
>100
>100
—
CHMe2
(S)-MeCHEt (Ile)
5.305 ( 0.191)
0.880 ( 0.042)
a Values are means of three experiments, standard deviation is given in parentheses.
potency was obtained. It is interesting to compare com-
pound 3 with literature compounds such as BOC-Val-
Met-Leu*Ala-Val-NHBn, an acyclic analog (with the
Leu*Ala hydroxyethylene isostere) described by Ghosh
et al.19, which was reported to have a Ki of 2.5 lM
against recombinant BACE1 enzyme. Part of the poten-
cy of this compound was attributed to occupation of P2
with the Met side chain, yet despite the absence of this
interaction compound 3 shows at least equivalent activ-
ity, even allowing for assay differences, consistent with
the hypothesis that the reduction in degrees of freedom
should produce an increase in inherent affinity.
With the aim of improving cellular penetration and/or
solubility we introduced a basic site either at P30 (com-
pound 4) or at P3 (compound 5). Unfortunately, in con-
trast to our findings in the open chain analog series,
none of these changes improved cellular activity. These
results suggest that this type of P1–P3 cyclization is
not optimal for obtaining whole cell BACE1 inhibitors.
Figure 3. The structure of 7 is shown superimposed with that of
OM99-2 (PDB entry 1FKN). OM99-2 is colored green, while C, N,
and O atoms in 7 are colored yellow, blue, and red, respectively. An
electrostatic surface is shown for the BACE active site, with the surface
around flap residues 71–74 (thin lines) omitted for clarity.
Then we turned our attention to the P1–N2 cyclization.
These analogs could provide advantages in terms of
elimination of one hydrogen bond donor (N2–H), better
occupancy of the S3 site, and reduced flexibility of the
cycle by means of reducing the ring size. In order to test
these hypotheses, we prepared compounds 6 and 7
according to the route shown in Scheme 1. Compound
6 improves enzymatic potency compared with 4 and
shows, for the first time, moderate cellular activity. In
order to further improve of potency, the Val residue at
the P3 position was replaced with Ile9 (compound 7).
This new compound shows a notable improvement in
enzymatic and whole cell potency (4- and 60-fold,
respectively) over compound 6 (see Table 1).
The binding mode for compound 7 reveals a very close
superposition of the main chain peptide groups on
OM99-2 between the P3 and the P30 groups. The side
chains of the two inhibitors also show a very close over-
lap in the BACE1 active site. Thus, the macrocycle
maintains a nearly identical peptide H-bond potential
and side-chain interactions with enzyme subsites S3–
S30; this was the objective that the macrocycle was de-
signed to accomplish. However, there are a number of
subtle but notable differences in binding, including the
loss of the P2 side-chain H-bond potential and the loss
of a water mediated H-bond between the P3/P2 amide
NH and the Gln73 main chain O atom in compound
7. There is also a nonoptimal occupancy of the S1 sub-
site by the macrocycle methylene groups in compound 7
relative to the Phe side chain of compound 1 and the
Leu side chain of OM99-2. This nonoptimal occupancy,
including the absence of the phenyl group, may explain
a major part of the lack of inherent increase in potency
that would be expected from the gains in binding energy
that would be predicted to be associated with the re-
duced degrees of freedom in the macrocyclic compound.
With these results in hand, and in order to gain insight
into the binding mode, an X-ray structure of the com-
plex between BACE1 and compound 7 was obtained.
The crystal structure of BACE1 complexed with 7 has
˚
been refined at 1.6 A resolution to an R-factor of
0.211 (Rfree = 0.227) using data collected at the Ad-
vanced Photon Source beamline 17ID. Crystallographic
methods have been published previously.9
The structure of compound 7 (Fig. 3) superimposed with
that of OM99-2 (PDB entry 1FKN) exhibits a binding
mode in common with previously determined peptide
based inhibitors.9
It is also interesting to note that the P20 Val side chain of
7 penetrates deeper into the S20 subsite than the corre-
sponding Ala residue of OM99-2, and the C-terminal
pyridyl ring occupies a very similar position to the P30

I. Rojo et al. / Bioorg. Med. Chem. Lett. 16 (2006) 191–195
195
8. For a review on this subject, see: John, V.; Beck, J. P.;
Bienkowski, M. J.; Sinha, S.; Heinrikson, R. I. J. Med.
Chem. 2003, 46, 4625.
Glu side chain of OM99-2. These factors appear to be
less important for increasing potency of 7 relative to 1.
9. (a) Lamar, J.; Hu, J.; Bueno, A. B.; Yang, H.; Guo, D.;
Copp, J. D.; McGee, J.; Gitter, B.; Timm, D.; May, P.;
McCarthy, J.; Chen, S. Biorg. Med. Chem. Lett. 2004, 14,
239; (b) Lamar, J.; Hu, J.; Yang, H.; Khon, T. D.; Copp,
J. D.; McGee, J.; Timm, D.; Erikson, J.; May, P.;
McCarthy, J.; Chen, S. Biorg. Med. Chem. Lett. 2004,
14, 245.
10. Tyndall, J. D. A.; Reid, R. C.; Tyssen, D. P.; Jardine, D.
K.; Todd, B.; Passmore, M.; March, D. R.; Pattenden, L.
K.; Bergman, D. A.; Alewood, P. F.; Birch, C. J.; Martin,
J. L.; Fairlie, D. P. J. Med. Chem. 2000, 43, 3495.
11. (a) Cumming, J. N.; Iserloh, U.; Kennedy, E. M. Curr.
Opin. Drug Discov. Dev. 2004, 7(4), 536(b) ELAN
The design and synthesis of 2–7, macrocyclic analogs of
the BACE1 inhibitor 1, provide further insight into the
preferred binding mode for 1 obtained from the crystal
structure of BACE1 complexed with 7. The X-ray struc-
ture showed that compound 7, obtained by cyclization
between N2-P1, was able to maintain the backbone
structure of OM99-2, and at the same time illustrates
the importance of an efficient occupancy of the S1 bind-
ing site in terms of potency. Also, the work demon-
strates that the N2–H in the peptidic structure is not a
required feature for binding to BACE1. Continued work
on the improvement of macrocyclic BACE1 inhibitors
will be published in due course.
PHARMACEUTICALS/PHARMACIA
& UPJOHN.
Macrocycles useful in the treatment of AlzheimerÕs
disease. WO-02100399 (2002).
12. Ghosh, A. K.; Davasamudram, T.; Hong, L.; DeZutter,
C.; Xu, X.; Weerasena, V.; Koelsch, G.; Bilcer, G.; Tang,
J. J. Biorg. Med. Chem. Lett. 2005, 15, 15.
Acknowledgments
13. Docking analysis carried out with in-house algorithm,
CDOCKER, to the BACE/OM99-2 X-ray structure (Ref.
7). See: (a) Vieth, M.; Hirst, J. D.; Kolinski, A.; Brooks, C.
L., III J. Comp. Chem. 1998, 19, 1612; (b) Vieth, M.; Hirst,
J. D.; Dominy, B. N.; Daigler, H.; Brooks, C. L., III J.
Comp. Chem. 1998, 19, 1623; (c) Vieth, M.; Cummins, D.
J. J. Med. Chem. 2000, 43, 3020; (d) Wu, G.; Robertson,
D. H.; Brooks, C. L., III; Vieth, M. J. Comp. Chem. 2003,
24, 1549; (e) Erickson, J. A.; Jalaie, M.; Robertson, D. H.;
Lewis, R. A.; Vieth, M. J. Med. Chem. 2004, 47, 45.
14. Dunn, M. J.; Jackson, R.; Pietruszka, J.; Turner, D.
J. Org. Chem. 1995, 60, 2210.
15. Brady, S. F.; Sisko, J. T.; Stauffer, K. J.; Colton, C. D.;
Qiu, H.; Lewis, S. D.; Assunta, S. Ng.; Shafer, J. A.;
Bogusky, M. J.; Veber, D. F.; Nutt, R. F. Biorg. Med.
Chem. 1995, 3, 1063.
16. Decamp, A. E.; Kawaguchi, A. T.; Volante, R. P.;
Shinkai, I. Tetrahedron Lett. 1991, 32, 1867.
The authors acknowledge Patrick May for his input
and critical review of the manuscript, Richard A. Brier,
Debra Kay Laigle and James E. McGee for their work
in the whole cell assay, and Raquel Torres for her work
in the biochemical assay.
References and notes
1. Janus, C. CNS Drugs 2003, 17, 457.
2. Evans, D. A.; Funkenstein, H. H.; Albert, M. S., et al.
JAMA 1989, 262, 2551.
3. (a) Selkoe, D. J. Physiol. Rev. 2001, 81, 741; (b) Selkoe, D.
J. Nature 1999, 99, A23.
4. Hartmann, D.; Tournoy, J.; Saftig, P.; Annaert, W.; De
Strooper, B. J. Mol. Neurosci. 2001, 17, 171.
5. Luo, Y.; Bolon, B.; Khan, S., et al. Nat. Neuroscience
2001, 4, 231.
6. Vassar, R. et al. Science 1999, 286, 734.
7. Hong, L.; Koelsch, G.; Lin, X.; Wu, S.; Terzyan, S.;
Ghsoh, A. K.; Zhang, X. C.; Tang, J. Science 2000, 290,
150.
17. Reichwein, J. F.; Liskamp, R. M. Tetrahedron Lett. 1998,
39, 1243.
18. Craparo, H. G. et al. Arch. Pharm. Med. Chem. 1996, 329,
273.
19. Ghosh, A. K. et al. J. Med. Chem. 2001, 44, 2865.