DOI: 10.1002/cbic.

200700383

Macrocyclic Statine-Based Inhibitors of BACE-1

Alessandra Barazza,[a, d] Marion Gçtz,[a, e] Sergio A. Cadamuro,[a] Peter Goettig,[a]
Michael Willem,[b] Holger Steuber,[c] Tanja Kohler,[c] Anja Jestel,[c]
Peter Reinemer,[c] Christian Renner,[a, f] Wolfram Bode,[a] and Luis Moroder*[a]

Minimal sequence requirements for binding of substrate-derived backbone, including the transition state analogue statine for
statine peptides to the aspartyl enzyme were established on the tight interactions with the two aspartyl residues of the active
basis of the X-ray cocrystal structure of the hydroxyethylene-octa- centre. The conformational preorganization of the inhibitor mole-
peptide OM00-3 in complexation with BACE-1. With this informa- cule was verified by NMR structural analysis and was then con-
tion to hand, macrocyclic compounds that conformationally re- firmed by the crystal structure of the BACE-1/inhibitor complex.
strict and preorganize the peptide backbone for an entropically Detailed insights into the binding mode of this macrocyclic inhib-
favoured binding to the enzyme active site cleft were designed. itor explained its moderate binding affinity in cell-free assays
By means of a side chain-to-side chain ring closure between two (Ki = 2.5 mm) and yielded precious information for possible struc-
aspartyl residues in the P2 and P3’ positions through phenylene- tural optimization in view of the lack of steric clashes of the mac-
1,3-dimethanamine, a 23-membered ring structure was obtained; rocycle with the flap domain of the enzyme.
this structure retained an extended conformation of the peptide

Introduction

Proteolysis of the membrane-anchored amyloid precursor pro- low-molecular-weight inhibitors. As macrocyclization of linear
tein (APP) leads to the Ab peptide, which accumulates in the peptides is a common method for conformational restriction
brain, forming insoluble plaques and neurofibrillary tangles as of peptides to increase bioactivities, cell permeability and pro-
the key pathological features of Alzheimer’s disease.[1] Clear teolytic stability,[8] cyclization has also been applied to restrict
evidence that the aspartyl protease BACE-1 (b-secretase) is conformational properties of synthetic inhibitors.[9] In fact, if a
mainly responsible for the generation of the Ab peptide makes conformational preorganization of the peptide backbone, con-
inhibition of this protease an attractive therapeutic target for trolled by the macrocycle, results in an optimal fit to the active
the treatment and prevention of Alzheimer’s disease.[2] site cleft, the entropic penalty in the binding process will be
The first generation of BACE-1 inhibitors were substrate ana- significantly reduced and binding affinities thus enhanced. We
logues in which the scissile amide bond was replaced with a have therefore selected this strategy for the development of
noncleavable isostere of the statine or hydroxyethylene type.[3] low-molecular-weight inhibitors of BACE-1.[10] In the course of
Examination of the crystal structures of the enzyme in com-
plexation with the pseudo-octapeptide inhibitors OM99-2[3b] [a] Dr. A. Barazza, Dr. M. Gçtz, Dr. S. A. Cadamuro, Dr. P. Goettig, Dr. C. Renner,
and OM00-3,[3c] both containing the Leu-Ala hydroxyethylene Prof. Dr. W. Bode, Prof. Dr. L. Moroder
dipeptide isostere (Leu*Ala), revealed extended conformations Max-Planck-Institut f8r Biochemie
of the peptide backbones and hydrogen bonding of both Am Klopferspitz 18, 82152 Martinsried (Germany)
Fax: (+ 49) 89-8578-2847
active site aspartates to the dipeptide isostere core structures. E-mail: [email protected]
These first X-ray structures have been instrumental in the [b] Dr. M. Willem
design of potent peptidomimetic inhibitors,[4] including macro- Adolf-Butenandt-Institut, Ludwig-Maximilians-UniversitBt
cyclic compounds,[5] that contain in their core structures func- 80336 M8nchen (Germany)
tionalities such as statine, hydroxyethylene, hydroxyethylamine [c] Dr. H. Steuber, Dr. T. Kohler, Dr. A. Jestel, Dr. P. Reinemer
Proteros Biostructures GmbH
or even the reduced amide as transition state motifs for target-
Am Klopferspitz 19, 82152 Martinsried (Germany)
ing interactions with the catalytic aspartates. The binding
[d] Dr. A. Barazza
modes of several of these peptidomimetic compounds within Present address: Mount Sinai School of Medicine
the active site cleft of BACE-1 have been determined by X-ray One Gustave L. Levy Place, Box 1234
analysis, yielding precious information on interactions with the New York, NY 10029 (USA)
enzyme subsites.[3g, 4a–c, h, j–l, 5b, d, 6] [e] Dr. M. Gçtz
Present address: Department of Chemistry
It is well established that an extended b-type conformation
Whitman College, Walla Walla, WA 99362 (USA)
of the substrates is essential for recognition and digestion by
[f] Dr. C. Renner
proteases.[7] Correspondingly, substrate-based inhibitors are Present address: Deutsche Forschungsgemeinschaft
generally used as lead compounds for the development of 53170 Bonn (Germany)

2078 B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2007, 8, 2078 – 2091
Macrocyclic Statine-Based Inhibitors of BACE-1

our work, various macrocyclic inhibitors of this enzyme were

Table 2. Peptoids, peptoid-peptide hybrids and peptides synthesized for
reported.[5] These inhibitors contain the P4–P1 sequence of this study.
peptidic and peptidomimetic compounds in the ring structures
and in the exocyclic position the P1–P1’ dipeptide isosteres as Compound Sequence
the crucial core units that are essential for the specificity of a H-Nglu-Nval-Nasn-Sta-Nval-Nala-Nglu-Nphe-NH2
ACHTUNGREaspartyl protease. Conversely, our approach was to include the b H-Glu-Leu-Asp-Sta-Nval-Nala-Nglu-Nphe-NH2
c H-Glu-Leu-Asp-Sta-Nala-Nval-Nglu-Nphe-NH2
tetrahedral intermediate mimic in the ring structure and to
d Ac-(d)Phe-(d)Glu-(d)Ala-(d)Val-(d)Leu-(d)Asn-(d)Val-(d)Glu-
retain the extended conformation of the peptide backbone NH2
within highly rigid macrocycles. e Ac-(d)Phe-(d)Glu-(d)Val-(d)Ala-(d)Leu-(d)Asp-(d)Leu-(d)Glu-
As starting point for these studies we used the X-ray struc- NH2
f Ac-(d)Phe-(d)Glu-(d)Ala-(d)Asp-(d)Leu-(d)Asn-(d)Val-(d)Glu-
tures of BACE-1 with bound OM99-2[3b] and OM00-3,[3c] which
NH2
showed nearly identical orientations of the inhibitors with re-
spect to residues P4–P2’, with differences being observed only
in the C-terminal P3’–P4’ portion, suggesting weaker interac-
tions of these residues with the protease. From these two pep- parative analysis of the enzyme-bound OM99-2 (1) and OM00-
tidic inhibitors (compounds 1 and 2 in Table 1) in the first in- 3 (2) provided support for a possible trimming of 3 to a crucial
peptide length comprising only the residues P3–P2’ for subse-
quent macrocyclisation experiments. Although statine is one
backbone atom shorter than a dipeptide, there is clear evi-
Table 1. Determination of the minimum peptide size required for inhibi- dence from previous studies that this transition state analogue
tory activity in cell-free enzyme assays. mimics a dipeptide substrate. Taking into account that statine
Compounds Peptide structures Ki [mm] would replace the Leu*Ala isostere in OM00-3, the pseudo-tet-
rapeptide 4, which was expected to encompass the substrate
1 (OM99-2) Glu-Val-Asn-Leu*Ala-Ala-Glu-Phe-OH 9.6 J 10 3 [12a]
2 (OM00-3) H-Glu-Leu-Asp-Leu*Ala-Val-Glu-Phe-OH 0.3 J 10 3 [12b]
portion P3–P2’, was synthesized (Table 1). Additionally, the Leu
3 H-Glu-Val-Asn-Sta-Val-Ala-Glu-Phe-NH2 33 J 10 3 (P3) residue of OM00-3 was replaced with isoleucine. Despite
4 Ac-Ile-Asp-(Phe)Sta-Val-NH2 5.4  1.0 its significantly reduced size, compound 4 retained inhibitory
5 Ac-Ile-Asp-(Phe)Sta-Val-Asp-NH2 5.4  1.1 potency in the low micromolar range.
6 Ac-Asp-(Phe)Sta-Val-Asp-NH2 > 1000
7 Ac-Ile-Asp-(Phe)Sta-NH2 > 1000
In consideration of alternative cyclization strategies, the tet-
8 Ac-Ile-Asp-Sta-Val-Asp-NH2 8.4  2.6 rapeptide 4 was extended C-terminally with an aspartyl resi-
due, in analogy to the Glu residue as P3’ in OM00-3. As expect-
ed, peptide 5 shows micromolar inhibition. However, addition-
al shortening at the N terminus by removal of the P3 residue
stance the peptide 3, in which the P1–P1’ sites were replaced (6) or by C-terminal shortening of 4 by deletion of Val (7) elimi-
by statine, was derived.[11] As the P2 position is occupied by an nates the binding affinities. From these results it was conclud-
asparagine residue, it corresponds to the Swedish mutation of ed that the minimal peptide size necessary for retaining bind-
APP that occurs adjacent to the BACE-1 cleavage site and is ing affinity is that of 4 spanning the parent peptide 3 se-
implicated in familial Alzheimer’s disease. In an initial analysis quence from P3 to P2’. Moreover, a comparison of the binding
of the pseudo-octapeptide inhibitor 3 in terms of its essential affinities of 5 and 8 confirmed that the isobutyl or phenyl
peptidic elements, we used iterative cycles to define the mini- groups act as the P1 residue with practically identical effica-
mal subsite requirements before optimizing the cyclic struc- cies.
tures. This strategy led to a macrocyclic compound of low mi- These experimental data provided support for macrocyclisa-
cromolar inhibitory potency in cell-free enzyme assays, whose tions involving Asp residues in P3’ and P2, the latter, according
binding mode to the enzyme was determined by X-ray analy- to modelling experiments with 5, being solvent-exposed upon
sis. The surprisingly good fit of the macrocycle to the active side chain-to-side chain ring closure with aliphatic or aromatic
site cleft, despite the presence of the flap region of the diamine spacers. This would lead to an exocyclic positioning of
enzyme, yielded interesting new prospects for further develop- the P3 residue, which would be amenable to structural optimi-
ments of this type of structural motif for BACE-1 inhibition. zation. For this purpose, a variety of branched chains or aro-
matic groups were introduced to mimic the Ile (P3) residue
and possibly to improve interactions with the apparently im-
Results and Discussion portant hydrophobic S3 binding pocket of the enzyme. The
Ac-Ile moiety of compound 4 was replaced by 4-methylpenta-
Minimal substrate-based sequence of inhibitory statine-
noyl, 3-methylpentanoyl, phenylacetyl, (4-fluorophenyl)acetyl
peptides
and benzyloxycarbonyl groups (Table 3); all led to inactive
With this selected strategy, conversion of 3 into peptidomimet- compounds. These results contrast with previous findings relat-
ic structures was initially attempted with peptoid/peptide hy- ing to similar replacements of the P3 and even P2 residues in
brids followed by retroinverso peptides to reduce proteolytic peptidomimetic BACE-1 inhibitors containing the hydroxyethy-
susceptibility (Table 2); both designs were unsuccessful. A com- lene Leu*Ala isostere.[4c] This might be attributed to the phe-

ChemBioChem 2007, 8, 2078 – 2091 B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2079
 L. Moroder et al.

in the course of the synthesis, since its O-acylation was found

Table 3. Linear statine-peptides. Numbered compounds are discussed in
the article, while all others are designated alphabetically. to occur only to negligible extents. Attempts to employ two
different orthogonal protecting groups for the b-carboxy func-
Compound Sequence tionalities for selective insertion of the diamine spacer on resin
9 benzyloxycarbonyl-Asp-Sta-Val-Asp-NH2 followed by cyclization in solution proved to be unsatisfactory
g (4-methylpentanoyl)-Asp-(Phe)Sta-Val-NH2 in terms of final yields. Similarly unsuccessful was the direct
h (3-methylpentanoyl)-Asp-(Phe)Sta-Val-NH2
cyclization of the side-chain-deprotected peptide through a
i phenylacetyl-Asp-(Phe)Sta-Val-NH2
j (4-fluorophenyl)acetyl-Asp-(Phe)Sta-Val-NH2 ACHTUNGREdiamine spacer at high dilution. The best synthetic routes, in
k benzyloxycarbonyl-Asp-(Phe)Sta-Val-NH2 which the prefunctionalized Asp residue was incorporated into
the linear precursor, are shown in Scheme 1. Upon acidolytic
deprotection and cleavage from the resin, cyclization was per-
formed in solution at high dilution with HATU/HOAt as cou-
nylstatine, although insertion of the phenyl group into the S1 pling reagents.
pocket should align the P2–P3 portion suitably for interaction In a first series of type A, B and C macrocycles, statine was
with the hydrophobic S3 enzyme subsite. incorporated as a transition state analogue and the flexible
and aromatic benzyloxycarbonyl group as the P3 residue. For
cyclization, phenylene-1,3-dimethanamine and the more rigid
Synthesis and inhibitory potencies of macrocyclic inhibitors
benzene-1,4-diamine were employed as spacers. The inhibitory
Among the various possible macrocycles based on peptides 4 activities of this series of compounds were determined in cell-
and 5 as scaffolds, we initially selected the ring structures free enzyme assays. The potencies of the type A macrocycles
shown in Figure 1 for modelling experiments with the X-ray are listed in Table 4, while for type B and C macrocycles inhibi-
structure of BACE-1 (PDB ID: 1m4h). tion of BACE-1 was not detected with up to 1 mm concentra-
Ring A is based on side-chain-to-side-chain crosslinking of tions. As expected from the results in hand, by replacing the
the b-carboxy groups of the two Asp residues with diamine Ac-Ile moiety in peptide 4 with a benzyloxycarbonyl group,
spacers of different lengths. Ring B (side chain-to-tail cycliza- the linear precursors 9, 10 and 12 were inactive. However,
tion) reflects the fact that the C-terminal Asp residue in pep- ACHTUNGRErestricting the ring size with the rigid phenylene-1,4-diamine
tide 5 does not affect binding affinities. In structure C, ring clo- spacer in compound 13 restored inhibitory activity in the low
sure through diamine spacers involves the b-carboxy group of micromolar range. Unfortunately, shortening of the ring size by
Asp in P2 and the a-carboxy group of the C-terminal Asp, one or four atoms in the macrocycles C and B, respectively,
which would allow the potential formation of a salt bridge be- eliminates inhibition despite the promising modelling experi-
tween the C-terminal Asp residue and Arg128 of the enzyme. ments.
For modelling experiments, various diamines such as propane- In a second generation of type A macrocycles, the N-termi-
1,3-diamine, butane-1,4-diamine, benzene-1,4-diamine and nal benzyloxycarbonyl group was replaced by Ac-Ile and the
phenylene-1,3-dimethanamine were used as spacers. The mac- statine by phenylstatine. In addition to phenylene-1,3-dime-
rocycles were found to be readily accommodated in the active thanamine and benzene-1,4-diamine, various flexible N,N’-alkyl-
site cleft of the enzyme, with the peptide backbone in a large- diamines were introduced as spacers in the macrocycles to
ly extended conformation. Rather unexpected was the obser- modulate the constraints in the ring structures by varying their
vation that the Val residue adjacent to the statine unit pre- size. This series of compounds was expected to yield precious
ferred occupancy of the S2’ enzyme binding pocket in silico, information on the optimal ring size and rigidity for a favoured
whereas the linear precursors tended towards more pro- interaction of the peptide backbone with the enzyme active
nounced interaction of the Val residue with the S1’ subunit. site cleft. The inhibitory potencies of these compounds are
The linear precursors were synthesized by standard Fmoc/ listed in Table 5. The linear precursors with the different dia-
tBu methods with acid-labile resin linkers depending upon the mines linked to the b-carboxy group of the Asp residue show
target C-terminal functionality (that is, amide or free carboxy lower binding affinities than the unmodified peptide 5, except
group). The hydroxy group of Fmoc-Sta-OH was not protected in the cases of those containing the aromatic diamines. How-

Figure 1. Cyclic peptide structures based on peptides 4 and 5. R1: benzyloxycarbonyl or Ac-Ile; R2 : phenyl or isopropyl; L: diaminoalkyl or diaminoaryl
spacers.

2080 www.chembiochem.org B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2007, 8, 2078 – 2091
Macrocyclic Statine-Based Inhibitors of BACE-1

The statine-peptide 2 and the

macrocyclic compound 21 were
compared with the commercial-
ly available BACE-1 inhibitor IV
(Calbiochem) in cell assays.
While this last peptidomimetic
inhibitor has been found to be
active,[4c] both compound 2 and
21 proved to be insufficiently
cell-permeable to inhibit BACE-1
in the in vivo assay.
The structure of the best in-
hibitor of this study, that is,
compound 21, was compared
with selected examples of
known macrocyclic inhibitors of
BACE-1 that contain dipeptide
isosteres (Figure 2). Unlike com-
pound 21, with its endocyclic
phenylstatine as transition state
analogue, the macrocycles D, E
and F contain the dipeptide iso-
steres in exocyclic positions. The
binding mode of macrocycle D
to BACE-1 was derived from the
X-ray structure,[5d] which showed
the expected hydrogen bonding
of the (S)-hydroxy group to the
catalytic aspartates of the
enzyme and the isobutyl and
methyl groups inserted into the
S1 and S1’ subsites, respectively.
Scheme 1. Synthesis of the type A (route A), type B (route B), and type C (route C) macrocycles. The ring structure allows for in-
teraction with the large hydro-
phobic S2–S3 subsites, leading
Table 4. Inhibitory potencies of type A macrocycles and their linear precursors determined in cell-free enzyme to nanomolar binding affinities.
assays. X: Sta; R: benzyloxycarbonyl; L: spacer.
In the macrocycle E the P1 resi-
due is part of the cycle, which
Spacer Ki [mm] Ki [mm]
prevents optimal occupancy of
the S1 pocket as determined by
H 9 > 1000 –
X-ray analysis of the complex
10 > 1000 11 > 1000
with BACE-1, while the overall
binding mode is very similar to
12 > 1000 13 44  18
that of the linear OM99-2 inhibi-
tor in which the Val and pyridine
moieties interact as P2’ and P3’
residues with the respective
ever, cyclization through the aliphatic amines resulted in mac- subsites.[5c] Compound F was derived from an optimized linear
rocyclic peptides displaying 13-, 5.6- and 1.5-fold enhanced in- peptidomimetic inhibitor in which macrocyclization was found
hibitions (15 vs. 14, 17 vs. 16, and 19 vs. 18), despite their flex- to improve potency in cell-based assays, but to reduce it
ibilities. Similarly positive was the result for compound 21, con- slightly in vitro.[5a]
taining the aromatic amine phenylene-1,3-dimethanamine,
which displayed a low micromolar inhibitory potency in cell-
NMR conformational analysis of the macrocyclic inhibitors
free enzyme assays. Unfortunately, cyclization with the even
more rigidifying benzene-1,4-diamine resulted in a macrocycle In comparison with the nanomolar potencies of the macro-
(23) of insufficient solubility for the enzyme assay. ACHTUNGREcycles D, E and F shown in Figure 2, the most potent cyclic in-
hibitors 13 and 21 of the present work exhibited only micro-

ChemBioChem 2007, 8, 2078 – 2091 B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2081
 L. Moroder et al.

molar inhibition; this suggests a

Table 5. Inhibitory potencies of type A macrocycles and their linear precursors determined in cell-free enzyme
assays. R: Ac-Ile and X: (Phe)Sta. possible failure of statine or
phenylstatine to act as effective
Spacer Ki [mm] Ki [mm] transition-state analogues, and/
or stronger deviations of the
H 5 5.4  1.1 – – peptide backbone from the op-
14 135  12 15 10.5  1.5 timal extended conformation.
16 71  8 17 12.5  1.7 Therefore, NMR structural analy-
18 16  1.4 19 10.2  1.6 sis was performed in order to
determine the conformational
20 5.9  0.6 21 2.8  0.6 spaces of the ring structures.
Figure 3 shows the resulting en-
22 7.5  0.8 23 not soluble
semble of the ten lowest-energy
structures of the macrocycle 13.
Highly convergent structures in
the peptide backbone were ob-

Figure 3. Ensemble of the ten lowest-energy structures of the macrocycle

13, as determined by NMR conformational analysis.

tained for this macrocycle, with only the benzyloxycarbonyl

group occupying a relaxed conformational space. The back-
bone exhibits an extended conformation that should fit well
into the active site cleft. Indeed, superimposition of one struc-
ture of this convergent ensemble with inhibitor OM99-2 in the
enzyme-bound state revealed a precise overlap in the peptide
backbone, and the side chain of statine, the N-terminal benzyl-
oxycarbonyl and the valine side chain displayed a similar orien-
tation for occupancy of the S3, S1 and S2’ enzyme subsites, re-
spectively (Figure 4). Most importantly, the hydroxy group of

Figure 4. Superimposition of the macrocycles 13 (orange) and 21 (grey)

onto the central portion of the linear OM99-2 inhibitor in the enzyme-
bound structure (green).

the statine moiety is oriented toward the enzyme active centre

in a manner similar to that of the hydroxyethylamine in OM99-
2. The macrocycle 21 is more flexible because of the two addi-
Figure 2. Structures of known macrocyclic inhibitors (D, E and F) of BACE-1
compared to compound 21; macrocycle D: Ki = 25 nm,[5d] macrocycle E: Ki = tional methylene groups in the spacer, but shows a similarly
65 nm[5c] and macrocycle F: IC50 = 90 nm.[5a] successful overlap with OM99-2.

2082 www.chembiochem.org B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2007, 8, 2078 – 2091
Macrocyclic Statine-Based Inhibitors of BACE-1

Molecular dynamics simulations of the macrocyclic struc-

Table 6. Statistics for data collection and processing.
tures of 13 and 21 bound to BACE-1 did not reveal steric clash-
es, and the P1, P3 and P3’ side chains can optimally interact X-ray source PX (SLS[a])
with the corresponding enzyme subsites. The spacers are wavelength [M] 0.992
detector MARCCD
ACHTUNGREexposed to the bulk water without interfering with the flap
T [K] 100
domain that closes the proteolytic cavity. space group C2
cell: a; b; c [M] 232.29; 100.46; 64.20
a; b; g [8] 90.0; 103.4; 90.0
resolution [M][b] 2.11 (2.20–2.11)
X-ray analysis of the BACE-1/inhibitor 21 complex unique reflections[b] 80 011 (7892)
multiplicity[b] 3.1 (2.8)
In order to validate the predictions from NMR conformational completeness [%][b] 96.9 (81.6)
analysis and docking experiments and to gain detailed insights Rsym [%][b] 7.1 (49.1)
into the binding mode of the macrocyclic inhibitor 21, X-ray Rmeas [%][b] 8.6 (60.7)
mean (I)/s[b] 11.65 (3.78)
analysis was performed. BACE-1 as a complex with the macro-
cycle 21 (Figure 5) crystallized in a new crystal form C2 that [a] Swiss Light Source (SLS, Villigen, Switzerland). [b] Numbers in brackets
correspond to the highest-resolution shell.
contained three molecules per asymmetric unit (Table 6). Of
these three independent BACE-1 molecules, the second mole-

cule was best ordered, with the main chain defined by a con-
tinuous electron density from Gly42p to Asn385 (according to
the numbering proposed by Hong et al.[3b]), except for the
loop segment between Ala157 and Ser169 and the segment
between Glu310 and Asp317. At a few sites (Ser86, His181,
Glu364 and Glu371), the density indicates two alternative side
chain conformations. According to a root mean square devia-
tion of 0.47 M for 365 Ca atoms (with use of a threshold of
2.0 M), the overall fold of our BACE-1 molecule is very similar to
that of BACE-1 in complex with OM00-3,[3c] which is considered
as the reference molecule below. Significant differences were
observed in the N-terminal loop Gly8 to Gln12, also called the
10s loop, which borders the S3 subsite (Figure 6 and below),
and in the adjacent Ser169–Gly172 segment, which follows the
missing Gly158–Ala168 loop. Remarkably similar is the confor-
mation of the b-hairpin “flap” Val67–Glu77, which stacks on
Figure 5. Schematic representation of compound 21 bound to BACE-1. the central part of the inhibitor, and the regions around the

Figure 6. Stereoview of the active-site cleft of BACE-1 (grey ribbon with the Flap, 10s loop and segment 327–332 in pink; both Asp side chains are shown in
full with the oxygen atoms in red) and compound 21 (green carbon atoms), superimposed with the final 2 Fobs Fcalcd electron density that accounted for the
inhibitor, contoured at 1 s (green net). The figure is shown in standard orientation; that is, with peptide substrate/inhibitor backbones running from left to
right.

ChemBioChem 2007, 8, 2078 – 2091 B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2083
 L. Moroder et al.

Figure 7. Stereoview of the active site cleft of BACE-1 in complex with compound 21, superimposed with inhibitor OM00-3 as it binds to BACE-1.[3c] The Con-
nolly surface of BACE-1 is overlaid by the flap domain (bottom) and the loop (top; shown as half-transparent surface in order to allow a direct view of the
subsites S3–S2‘).

two catalytic aspartic acid/aspartate residues Asp32 and the N-terminal part of compound 21 is slightly shifted towards
Asp228, situated at the centre of the active site cleft. the catalytic centre, giving rise to a maximal deviation of 1 M
As in other aspartyl proteinases, the BACE-1 chain (seen in between the hydroxy groups of the two core structures (to the
standard orientation, i.e., with the peptide substrate binding right in Figures 6 and 7), while the C-terminal part is almost
from left to right across the molecular surface; Figures 6 and 7) ACHTUNGREsuperimposable. The lack of one atom in the central statine
folds into a “lower” N-terminal and an “upper” C-terminal group, in relation to peptidic substrates or peptidic inhibitors
domain, which touch each other with formation of the active- containing a hydroxyethylene group, is almost compensated
site cleft. Compound 21 is located between the two lobes. It for by these relatively small shifts and additional slight confor-
has the shape of a handle-containing nail brush (Figure 5), the mational adaptations. The shift of the central methyl hydroxy
brush tightly interdigitates with the subsites of the proteinase, group is paralleled by small shifts of both catalytic Asp side
while the handle does not protrude out of the active-site cleft, chains and by a rotation of the Asp32 carboxylate group,
but is accommodated in the residual part of this cleft (Fig- ACHTUNGREresulting in even stronger hydrogen bonds to the hydroxy
ures 6 and 7). The full peptide inhibitor chain was well defined group. Three of the intermain-chain hydrogen bonds (P3-
by electron density from the N-terminal acetyl group to the C- Ile O···Thr232 N; P2 “Asp” O···Gln73 N; P3’ “Asp” NH···Pro70 O)
terminal carboxamide, and for the P2-“Asp” and the P3’-“Asp” are significantly longer and energetically more unfavourable
side chains, the b-carboxy groups of which are involved in ring than in the reference structure, a fact that might contribute to
closure (Figure 6). Only the phenylene-1,3-dimethanamine the overall reduced affinity. The hydrogen bonds made by the
spacer lacks appropriate electron density, indicating that it is N-terminal P3 acetamido group with Thr232 Og, by the
partially disordered. Modelling and subsequent energy refine- P2 “Asp” side chain carbonyl with the Arg235 side chain, and
ment experiments, however, showed that the defined part of by the C-terminal carboxyamide group with the side chain of
the inhibitor does not allow the spacer to explore a larger con- Arg128 replace similar hydrogen bonds in the OM00-3 com-
formational space. This means that the spacer must reside in plex, while those made by the carboxamido group of the
the cleft with a conformation close to that shown in Figures 6 P3’ “Asp” with Thr72 Og and Tyr198 OH are new.
and 7. The N-acetyl group of compound 21 extends into the essen-
tially hydrophilic S4 pocket, with the carbonyl group directed
into the cleft, while the P4-Glu side chain of OM00-3 is directed
Inhibitor conformation and interaction
away. It is not obvious why the adjacent Gly8–Lys9–Ser10 seg-
The peptidic backbone of compound 21 bound BACE-1 in an ment in the 21 complex is shifted, so that the Ser10 points—
overall extended conformation, which was remarkably similar unlike in the OM00-3 complex—into the “up” position.[13]
to that of OM00-3 (Figure 7). As in the OM00-3 complex, the There it forms a hydrogen bond with Arg307 NH1, through the
inhibitor backbone forms six intermain-chain hydrogen bonds carbonyl group, while disrupting the hydrogen bond with
to the BACE-1 backbone carbonyl and amido functions Thr232 Og. The change in interaction is associated with a con-
(Figure 5) and a hydrogen bond to the hydroxy group of siderable movement of the adjacent Ser169–Gly172 segment,
Tyr198, in addition to the four hydrogen bonds formed be- which follows the fully disordered 158–168 loop. The P3-Ile
tween the two active-site aspartates and the hydroxy group of side chain nestles, apparently quite favourably, in the hydro-
statine. In comparison with the peptidic OM00-3 backbone, phobic S3 pocket, where it is packed between the peptide

2084 www.chembiochem.org B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2007, 8, 2078 – 2091
Macrocyclic Statine-Based Inhibitors of BACE-1

groups of segment Gly11–Gln12–Gly13 and the side chain of Conclusions

Ile110 and the P1-phenyl group. The P2-“Asp” side chain,
which was substituted by the diamine spacer, shifted slightly The crystal structure of BACE-1 in complex with compound 21
along the hydrophilic flat cleft floor towards the catalytic shows that our original design concept of a statine-based
centre (to the right in Figure 7). ACHTUNGREmacrocyclic inhibitor, in which the transition state analogue is
According to positional scanning, Asp residues at P2 confer included in the macrocycle, was a successful one. As predicted
optimal susceptibility to BACE-1 substrates.[14] The P1-phenyl from modelling studies, the peptide backbone binds with its
group extends into the hydrophobic S1 pocket, where it is N-terminal P3–P1 segment in a conformation virtually identical
well packed between the side chains of Tyr71, Phe108 and P3- to that of the linear peptide OM00-3 to compensate for the
Ile and shielded by the flap against the bulk water. Peptide one-atom reduced backbone of the statine group. The P2- and
substrates with Leu at P1 seem to be even slightly better sub- the P3’-derivatized “Asp” side chains can adopt conformations
strates, while the P1 residue of APP is methionine. The hydroxy that allow the apparently unconstrained connectivity of their
group of the transition state-mimicking statine, which corre- carboxyl groups with the two amino groups of the phenylene-
sponds to the solvent molecule activated upon substrate bind- 1,3-dimethanamine spacer, and the even better kinetic data for
ing to initiate nucleophilic attack on the peptide carbonyl the cyclic construct relative to the corresponding linear com-
group, is directed between the carboxylic/carboxylate groups pound show that the ring closure contributes positively to
of Asp32 and Asp228, which are arranged almost coplanar to binding and that the phenylenedimethanamine spacer has no
each other. This hydroxy group is shifted to the right by 1 M in unfavourable effect through steric clashes with the flap
relation to the bound OM00-3, so that the following methylene domain. As shown by the X-ray structure, the still relatively
group is already located on the cleft floor close to the position weak affinity might well be attributed to the statine moiety,
of the Ca atom of the P1’-Ala mimic of OM00-3. This causes which lacks a carbon atom for optimal registration of the back-
the one-atom shorter backbone of compound 21 to span the bone into the active site cleft. Therefore, replacement of this
space occupied in true P1–P1’ dipeptide transition state iso- transition state structure with a more proper P1–P1’ isostere
steres by the exposed intervening aminomethylene group. In might be expected to improve the inhibitory potency signifi-
substrate–BACE-1 complexes, the large central hydrophobic cantly. Because of the strong increase in inhibitory potency ob-
cavity can accommodate P1’ side chains; optimal BACE-1 sub- served upon replacing the flexible benzyloxycarbonyl group as
strates contain a Met at P1’,[14] while APP possesses a P1’-Asp. P3 residue in the inactive macrocycle 12 with Ile in 21, a simi-
Very similar to the OM00-3 complex, the P2’-Val side chain of larly enhanced binding affinity was also expected from this re-
compound 21 fills the hydrophobic S2 pocket suitably, packing placement in the macrocycle 23. Although this last compound
between the side chains of Val69, Pro70 and Tyr71, Ser35 and was too insoluble for the enzyme assays it may still represent a
Ile126 and being covered by the flap. A Val at P2’ confers opti- very promising lead structure for further optimization. Replace-
mal susceptibility to peptide substrates. The P3’ residue is posi- ment of the statine with the hydroxyethylene dipeptide iso-
tioned further towards the molecular surface than in OM00-3; stere in 23 as a more efficient transition state analogue could
due to a 1208 rotation around the N Ca bond, the P3’-“Asp” represent an alternative strategy. This approach would expand
side chain is directed towards the cleft floor, while the C-termi- the ring size by only one atom, which should be well compen-
nal carboxamide group points away from the molecular sur- sated by the rigid benzenediamine spacer.
face.
The phenylenedimethanamine spacer crosslinking the b-car-
boxy groups of the P2 and the P3’ “Asp” side chains, although Experimental Section
not properly defined by electron density, is packed into the
Materials and methods: Solvents and reagents were of the high-
large and essentially hydrophobic S1’ cavity, formed in the est quality commercially available and were used without further
upper rim by the side chains of Thr231, Val332 and Ile226 (cleft purification. Precoated silica gel F254 plates were used for TLC, and
floor), Arg235 and Tyr198 (left and right side), and Arg235, silica gel 60 (40–60 mm) from Merck (Darmstadt, Germany) for silica
Thr329, Asp223 and Lys224 (upper rim). In the productive APP gel chromatography. Analytical HPLC was performed by using
complex, this cavity would accommodate the P1’-Asp side Waters equipment with use of the following systems:
chain, while according to positional scanning experiments the System I: Chromolith Performance RP 18-e, 100 J 4,6 mm C18
Met residue confers optimal cleavage at this position. In spite (Merck); linear gradient from 100 % solvent A (2 % H3PO4/CH3CN,
of some close contacts of the (modelled) phenylene spacer, 95:5) to 100 % solvent B (2 % H3PO4/CH3CN, 90:10) in 6 min, fol-
only the side chain of Val332 seems to be slightly shifted from lowed by 1 min isocratic elution; flow rate: 3 mL min 1.
its position adopted in the OM00-3 complex. We conclude that System II: Xterra-TM-C8 MS 5 mm, 150 J 3.9 mm (Waters); linear
the spacer floats slightly in this primarily hydrophobic cavity, ACHTUNGREgradient from 100 % solvent A (2 % H3PO4/CH3CN, 95:5) to 100 %
but that it essentially adopts a conformation similar to that solvent B (2 % H3PO4/CH3CN, 90:10) in 17 min, followed by 1 min
shown in Figures 6 and 7. isocratic elution; flow rate: 1.5 mL min 1.
System III: ET 125/4 Nucleosil 100–5 C8, 125 J 4 mm (Macherey–
Nagel, DTren, Germany); linear gradient from 100 % solvent A (1 %
TFA in H2O) to 100 % solvent B (0.8 % TFA in CH3CN) in 13 min, fol-
lowed by 1 min isocratic elution; flow rate: 1.5 mL min 1.

ChemBioChem 2007, 8, 2078 – 2091 B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2085
 L. Moroder et al.

System IV: Chromolith Performance RP 18-e, 100 J 4.6 mm C18 Synthesis of retroinverso-peptides: Peptides d, e and f (Table 2)
(Merck); linear gradient from 100 % solvent A (2 % H3PO4/CH3CN, were synthesised by automated solid-phase methodology on an
95:5) to 50 % solvent B (2 % H3PO4/CH3CN, 90:10) in 15 min, fol- ABI-Pioneer Perseptive Biosystems peptide synthesizer by Fmoc/
lowed by 1 min isocratic elution; flow rate: 3 mL min 1. HBTU/DMF chemistry. Peptides d, e and f were synthesized on Ten-
taGel S RAM support (substitution: 0.24 mmol g 1). The protocol in-
System V: Chromolith Performance RP 18-e, 100 J 4.6 mm C18 cluded extended cycles (4 mmol of activated amino acid derivative
(Merck); linear gradient from 100 % solvent A (2 % H3PO4/CH3CN, per cycle). Final acylation was performed with Ac2O/2,6-lutidine/
95:5) to 50 % solvent B (2 % H3PO4/CH3CN, 90:10) in 30 min, fol- DMF (5:6:89). The peptides were deprotected and cleaved from
lowed by 1 min isocratic elution; flow rate: 3 mL min 1. the resin with TFA/TIS/H2O (95:2.5:2.5) at room temperature over
2 h. The crude products were precipitated with cold tert-butyl
Preparative RP-HPLC was carried out with Abimed–Gilson equip-
methyl ether/hexane (2:1) and purified by preparative RP-HPLC.
ment with the following systems:
The purities of the peptides exceeded 97 % as determined by ana-
System 1: VP 250/21 Nucleosil 300–5 C8 PPN (Macherey–Nagel); gra- lytical RP-HPLC and their structural integrities were confirmed by
dient as specified for each compound, with solvents A (1 % TFA in electrospray ionization mass spectrometry.
H2O) and B (0.8 % TFA in CH3CN); flow rate: 10 mL min 1. Compound d: Purification (system 2) by linear gradient from 10 to
System 2: VP 250/21 Nucleosil 100–5 C18 PPN (Macherey–Nagel); 30 % B in 7 min, then 30 to 60 % B in 60 min; anal. HPLC (system I):
gradient as specified for each compound, with solvents A (1 %TFA tR 1.63 min; HPLC (system III): tR 7.83 min; ESI-MS: m/z 960.8
in H2O) and B (0.8 % TFA in CH3CN); flow rate: 10 mL min 1. [M+H] + ; Mr = 960 calcd for C44H68N10O14.

Compound e: Purification (system 2) by linear gradient from 10 to

ESI-MS spectra were recorded on a Perkin–Elmer SCIEX API 165
30 % B in 7 min, then 30 to 60 % B in 60 min; anal. HPLC (system I):
spectrometer equipped with a nebulizer-assisted electrospray
tR 1.72 min; HPLC (system III): tR 7.91 min; ESI-MS: m/z 976.6
source.
[M+H] + ; Mr = 976 calcd for C45H69N9O15.
Peptide synthesis
Compound f: Purification (system 2) by linear gradient from 10 to
Synthesis of peptoids and peptide-peptoid hybrids: Peptoid a 30 % B in 7 min, then 30 to 60 % B in 60 min; anal. HPLC (system I):
and peptide–peptoid hybrids b and c (Table 2) were synthesised tR 1.92 min; HPLC (system III): tR 8.44 min; ESI-MS: m/z 977.8
manually on Rink amide MBHA resin (substitution: 0.64 mmol g 1, [M+H] + ; Mr = 977 calcd for C43H64N10O16.
scale: 0.08 mmol). The syntheses of the peptoidic parts were per-
formed by the submonomer method (in which the oligomer is pre- Synthesis of peptides: The peptides listed in Table 3 were synthes-
pared on resin by repeated coupling of bromoacetic acid followed ised by using standard SPPS methodology with Fmoc/HBTU
by treatment with an appropriate amine that comprises the side chemistry on Rink Amide MBHA resin (loading: 0.64 mmol g 1).
chain functionality),[15] except in the case of Nala, which was intro- Couplings were performed with 2 equiv Fmoc-amino acid/HOBt/
duced as Fmoc-Sar-OH. Acylations with bromoacetic acid HBTU/DIPEA (1:1:1:2) in DMF and Fmoc cleavage was carried out
(0.8 mmol, 112 mg) were performed with N,N’-diisopropylcarbodi- with piperidine in DMF (20 %). Acetylation was performed twice
ACHTUNGREimide (0.8 mmol, 123.8 mL) in DMF (3 mL) in 30 min at room tem- with Ac2O/pyridine/DMF (1:1:8) in 10 min, while acylation with 4-
perature. Acylations of primary amines were monitored by use of methylpentanoic acid, 3-methylpentanoic acid, phenylacetic acid
the Kaiser test,[16] and of secondary amines by use of the chlora- and (4-fluorophenyl)acetic acid was carried out with HOBt/HBTU/
nil[17] or TNBS tests.[18] Bromo displacement was achieved with solu- DIPEA (1:1:2) in DMF. After deprotection/cleavage from the resin
tions (2 m) of the primary amine (0.8 mmol) in DMSO at room tem- with TFA/TIS/H2O (95:2.5:2.5) at room temperature in 2 h, the
perature for 1 h. In the case of isopropylamine the reaction was re- crude products were precipitated with cold tert-butyl methyl
peated a second time. The Nglu side chain was introduced with H- ether/hexane (2:1) and purified by preparative RP-HPLC. The puri-
b-AlaACHTUNGREOtBu·HCl. The statine residue was incorporated after preacti- ties of the peptides exceeded 97 % as determined by analytical RP-
vation of Fmoc-Sta-OH (1.4 mmol, 55.7 mg) with HATU/HOAt/DIEA HPLC and their structural integrities were confirmed by electro-
(1:1:2). Deprotection and cleavage from the resin was carried out spray ionization mass spectrometry.
with TFA/TIS/H2O (95:2.5:2.5) at room temperature over 2 h. The
Compound 3: Purification (system 1), linear gradient from 20 to
crude products were precipitated with cold tert-butyl methyl
70 % B in 60 min; anal. HPLC (system I): tR 3.27 min; HPLC (sys-
ether/hexane (2:1) and purified by preparative RP-HPLC. The puri-
tem II): tR 10.24 min; HPLC (system III): tR 6.39 min; ESI-MS: m/z
ties of the peptides exceeded 97 % as determined by analytical RP-
963.8 [M+H] + ; Mr = 963 calcd for C44H70N10O14.
HPLC, and their structural integrities were confirmed by electro-
spray ionization mass spectrometry. Compound 4: Purification (system 1), linear gradient from 12 to
42 % B in 60 min; anal. HPLC (system I): tR 2.05 min; ESI-MS: m/z
Compound a: Purification (system 1), linear gradient from 15 to
578.4 [M+H] + ; Mr = 577 calcd for C28H43N5O8.
30 % B in 45 min; anal. HPLC (system I): tR 2.77 min; HPLC (sys-
tem II): tR 9.72 min; ESI-MS: m/z 963.6 [M+H] + ; Mr = 963 calcd for Compound 5: Purification (system 1), gradient from 10 to 40 % B in
C44H70N10O14. 60 min; anal. HPLC (system I): tR 1.95 min; ESI-MS: m/z 693.4
[M+H] + ; Mr = 692 calcd for C32H48N6O11.
Compound b: Purification (system 1), linear gradient from 5 to 15 %
B in 5 min, then 15 to 45 % B in 60 min; anal. HPLC (system II): Compound 6: Purification (system 1), linear gradient from 7 to 37 %
tR 8.93 min; ESI-MS: m/z 978.6 [M+H] + ; Mr = 978 calcd for B in 60 min; anal. HPLC (system I): tR 1.33 min; ESI-MS: m/z 580.4
C45H71N9O15. [M+H] + , 1159.6 [2 M+H] + ; Mr = 579 calcd for C26H37N5O10.

Compound c: Purification (system 1), linear gradient from 5 to 30 % Compound 7: Purification (system 1), linear gradient from 7 to 37 %
B in 60 min; anal. HPLC (system II): tR 8.63 min; ESI-MS: m/z 978.6 B in 60 min; anal. HPLC (system I): tR 1.49 min; ESI-MS: m/z 479.2
[M+H] + ; Mr = 978 calcd for C45H71N9O15. [M+H] + , 958.0 [2 M+H] + ; Mr = 478 calcd for C23H34N4O7.

2086 www.chembiochem.org B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2007, 8, 2078 – 2091
Macrocyclic Statine-Based Inhibitors of BACE-1

Compound 8: Purification (system 1), linear gradient from 10 to nomethyl)-benzylamine (326 mg, 1.38 mmol) in DMF (2 mL) was
40 % B in 60 min; anal. HPLC (system I): tR 1.84 min; ESI-MS: m/z added, and the solution was stirred for 12 h at room temperature.
659.4 [M+H] + , 1318.0 [2 M+H] + ; Mr = 658 calcd for C29H50N6O11. The solvent was evaporated, and the oily residue was redissolved
in EtOAc (50 mL). The organic layer was washed with H2O (1 J
Compound 9: Purification (system 1), linear gradient from 20 to
50 mL), NaHCO3 (5 %, 3 J 50 mL), H2O (1 J 50 mL), KHSO4 (5 %, 3 J
50 % B in 60 min; anal. HPLC (system I): tR 2.42 min; ESI-MS: m/z
50 mL) and saturated NaCl (1 J 50 mL), dried over MgSO4 and con-
638.4 [M+H] + , 1275.8 [2 M+H] + ; Mr = 637 calcd for C29H43N5O11.
centrated. The pure product was obtained by precipitation from
Compound g: Purification (system 1), linear gradient from 20 to EtOAc/petroleum ether as a crystalline solid. Yield: 815.4 mg
50 % B in 60 min; anal. HPLC (system I): tR 2.46 min; ESI-MS: m/z (97 %); HPLC (system I): tR 4.52 min; ESI-MS: m/z 614.4 [M+H] + ,
521.4 [M+H] + , 1041.8 [2 M+H] + ; Mr = 520 calcd for C26H40N4O7. 558.4 [M tBu+H] + , 514.4 [M Boc+H] + , 1227.6 [2 M+H] + ; Mr = 613
calcd for C35H39N3O7.
Compound h: Purification (system 1), linear gradient from 20 to
50 % B in 60 min; anal. HPLC (system I): tR 2.42 min; ESI-MS: m/z Fmoc-Asp[4-(Boc-amino)-1-amidobenzene]-OAll: The title compound
521.4 [M+H] + , 1041.8 [2 M+H] + ; Mr = 520 calcd for C26H40N4O7. was prepared and purified as described for Fmoc-Asp[3-(Boc-ami-
nomethyl)-1-amidomethylbenzene]-OAll. Yield: 739 mg (91 %);
Compound i: Purification (system 1), linear gradient from 20 to HPLC (system I): tR 4.51 min; ESI-MS: m/z 586.4 [M+H] + , 530.4
50 % B in 60 min; anal. HPLC (system I): tR 2.36 min; ESI-MS: m/z [M tBu+H] + , 1171.6 [2 M+H] + ; Mr = 585 calcd for C33H35N3O7.
541.2 [M+H] + , 1081.8 [2 M+H] + ; Mr = 540 calcd for C28H36N4O7.
Fmoc-Asp[3-(Boc-amino)-1-amidopropane]-OAll: The title compound
Compound j: Purification (system 1), linear gradient from 20 to was prepared and purified as described for Fmoc-Asp[3-(Boc-ami-
50 % B in 60 min; anal. HPLC (system I): tR 2.43 min; ESI-MS: m/z nomethyl)-1-amidomethylbenzene]-OAll. Yield: 717 mg (94 %);
559.4 [M+H] + , 1117.4 [2 M+H] + ; Mr = 558 calcd for C28H35N4O7:. HPLC (system I): tR 4.11 min; ESI-MS: m/z 552.4 [M+H] + , 496.6
Compound k: Purification (system 1), a linear gradient from 20 to [M tBu+H] + , 452.4 [M Boc+H] + , 1103.6 [2 M+H] + ; Mr = 551 calcd
50 % B in 60 min; anal. HPLC (system I): tR 2.52 min; ESI-MS: m/z for C30H37N3O7: 551.
672.6 [M+H] + , 1343.8 [2 M+H] + ; Mr = 671 calcd for C32H41N5O11. Fmoc-Asp[4-(Boc-amino)-1-amidobutane]-OAll: The title compound
Synthesis of type A macrocycles was prepared and purified as described for Fmoc-Asp[3-(Boc-ami-
3-(Boc-aminomethyl)-benzylamine: (Boc)2O (1.5 g, 6.9 mmol) in diox- nomethyl)-1-amidomethylbenzene]-OAll. Yield: 780 mg (92 %);
ane (40 mL) was added dropwise at room temperature over 3 h HPLC (system I): tR 4.14 min; ESI-MS: m/z 566.4 [M+H] + , 510.6
to phenylene-1,3-dimethanamine (7.5 g, 55.0 mmol) in dioxane [M tBu+H] + , 466.4 [M Boc+H] + , 1131.8 [2 M+H] + ; Mr = 565 calcd
(40 mL). After 22 h, the solvent was removed and water (50 mL) for C31H39N3O7: 565.
was added. Insoluble bis-substituted product was filtered off and Fmoc-Asp[5-(Boc-amino)-1-amidopentane]-OAll: The title compound
the filtrate was extracted with CH2Cl2 (3 J 50 mL). The organic was prepared and purified as described for Fmoc-Asp[3-(Boc-ami-
phase was washed with water (5 J 50 mL), dried (MgSO4) and nomethyl)-1-amidomethylbenzene]-OAll. Yield: 682 mg (85 %);
evaporated. The crude product was purified on a silica gel column HPLC (system I): tR 4.36 min; ESI-MS: m/z 580.4 [M+H] + , 524.6
with a linear gradient from CH2Cl2/MeOH (95:5) to CH2Cl2/MeOH [M tBu+H] + , 480.4 [M Boc+H] + , 1159.8 [2 M+H] + ; Mr = 579 calcd
(8:2). Yield: 1.28 g (79 %); TLC (CH2Cl2/MeOH 6:4): Rf 0.38; HPLC for C32H41N3O7.
(system I): tR 1.58 min; ESI-MS: m/z 237.2 [M+H] + , 473.8 [2 M+H] + ;
Mr = 236 calcd for C13H20N2O2. Fmoc-Asp[3-(Boc-aminomethyl)-1-amidomethylbenzene]-OH: Fmoc-
Asp[3-(Boc-aminomethyl)-1-amidomethylbenzene]-OAll (815 mg,
1-Boc-phenylene-1,4-diamine: The compound was prepared as de- 1.33 mmol) was dried overnight under reduced pressure and
scribed for 3-(Boc-aminomethyl)-benzylamine. After being washed ACHTUNGREdissolved in DMF (10 mL) under argon with PhSiH3 (0.82 mL,
and dried the product was crystallized from petroleum ether. Yield: 6.6 mmol). PdACHTUNGRE(PPh3)4 (31 mg, 27 mmol) was flushed with argon, sus-
1.50 g (78 %); TLC (CH2Cl2/MeOH 95:5): Rf 0.56; HPLC (system I): tR pended in DMF (10 mL) and added to the solution. The mixture
1.44 min; ESI-MS: m/z 209.0 [M+H] + , 417.2 [2 M+H] + ; Mr = 208 was stirred at room temperature under argon for 30 min and the
calcd for C11H16N2O2. progress of the reaction was monitored by HPLC. Upon completion
of the reaction, the solvent was evaporated, the residue was redis-
1-Boc-1,3-diaminopropane: The title compound was prepared and
solved in EtOAc (30 mL), and the organic layer washed with H2O
purified as described for 3-(Boc-aminomethyl)-benzylamine. Yield:
(1 J 30 mL), citric acid (5 %, 5 J 30 mL) and saturated NaCl (1 J
1.93 g (66 %); TLC (CH2Cl2/MeOH 4:1): Rf 0.26; HPLC (system I): tR
30 mL), dried over MgSO4 and concentrated. The product was pre-
5.20 min; ESI-MS: m/z 175.2 [M+H] + , 348.6 [2 M+H] + ; Mr = 174
cipitated from CH2Cl2/petroleum ether as a crystalline solid. Yield:
calcd for C8H18N2O2.
594 mg (78 %); HPLC (system I): tR 3.89 min; ESI-MS: m/z 574.4
1-Boc-1,4-diaminobutane: The compound was prepared and puri- [M+H] + , 518.4 [M tBu+H] + , 474.4 [M Boc+H] + , 1147.8 [2 M+H] + ;
fied as described for 3-(Boc-aminomethyl)-benzylamine. Yield: Mr = 573 calcd for C32H35N3O7.
2.00 g (75 %); TLC (CH2Cl2/MeOH 4:1): Rf 0.24; ESI-MS: m/z 189.0
Fmoc-Asp[4-(Boc-amino)-1-amidobenzene]-OH: This compound was
[M+H] + , 377.6 [2 M+H] + ; Mr = 188 calcd for C9H20N2O2.
obtained as described for Fmoc-Asp[3-(Boc-aminomethyl)-1-amido-
1-Boc-1,5-diaminopentane: The compound was prepared and puri- methylbenzene]-OH. Yield: 505 mg (77 %); HPLC (system I): tR
fied as described for 3-(Boc-aminomethyl)-benzylamine. Yield: 3.96 min; ESI-MS: m/z 546.4 [M+H] + , 490.4 [M tBu+H] + , 1091.6
2.15 g (61 %); TLC (CH2Cl2/MeOH 4:1): Rf 0.24; ESI-MS: m/z 202.8 [2 M+H] + ; Mr = 545 calcd for C30H31N3O7.
[M+H] + , 404.6 [2 M+H] + ; Mr = 202 calcd for C10H22N2O2.
Fmoc-Asp[3-(Boc-amino)-1-amidopropane]-OH: This compound was
Fmoc-Asp[3-(Boc-aminomethyl)-1-amidomethylbenzene]-OAll: Fmoc- obtained as described for Fmoc-Asp[3-(Boc-aminomethyl)-1-amido-
Asp-OAll (600 mg, 1.52 mmol) in DMF (5 mL) was preactivated methylbenzene]-OH. Yield: 652 mg (98 %); HPLC (system I): tR
during 10 min with HOBt (205 mg, 1.52 mmol), HBTU (575 mg, 3.56 min; ESI-MS: m/z 512.4 [M+H] + , 456.4 [M tBu+H] + , 412.2
1.52 mmol) and DIEA (520 mL, 3.03 mmol). A solution of 3-(Boc-ami- [M Boc+H] + , 1023.6 [2 M+H] + ; Mr = 511 calcd for C27H33N3O7.

ChemBioChem 2007, 8, 2078 – 2091 B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2087
 L. Moroder et al.

Fmoc-Asp[4-(Boc-amino)-1-amidobutane]-OH: This compound was Z-AspACHTUNGRE[(4-amino)-1-amidobenzene]-Sta-Val-Asp-NH2 (12): The com-

obtained as described for Fmoc-Asp[3-(Boc-aminomethyl)-1-amido- pound was synthesized as described for 10. Yield: 86 mg (94 %);
methylbenzene]-OH. Yield: 576 mg (87 %); HPLC (system I): tR HPLC (system I): tR 2.07 min (> 90 %); ESI-MS: m/z 728.4 [M+H] + ;
3.61 min; ESI-MS: m/z 526.4 [M+H] + , 470.4 [M tBu+H] + , 426.2 Mr = 727 calcd for C35H49N5O10.
[M Boc+H] + , 1051.6 [2 M+H] + ; Mr = 525 calcd for C28H35N3O7.
Ac-Ile-Asp[3-(Boc-amino)-1-amidopropane]-(Phe)Sta-Val-Asp-NH2 (14):
Fmoc-Asp[5-(Boc-amino)-1-amidopentane]-OH: The compound was The compound was synthesized as described for 10. Final acetyla-
obtained as described for Fmoc-Asp[3-(Boc-aminomethyl)-1-amido- tion was carried out with Ac2O/pyridine/DMF (1:1:8) in 10 min and
methylbenzene]-OH. Yield: 470 mg (74 %); HPLC (system I): tR then repeated a second time. Yield: 93 mg (89 %); HPLC (system I):
3.73 min; ESI-MS: m/z 540.6 [M+H] + , 484.6 [M tBu+H] + , 440.4 tR 1.75 min (> 90 %); ESI-MS: m/z 749.6 [M+H] + ; Mr = 748 calcd for
[M Boc+H] + , 1079.8 [2 M+H] + ; Mr = 539 calcd for C29H37N3O7. C35H56N8O10.

Z-Asp[3-(Boc-aminomethyl)-1-amidomethylbenzene]-OBzl: 3-(Boc- Ac-Ile-Asp[4-(Boc-amino)-1-amidobutane]-(Phe)Sta-Val-Asp-NH2 (16):

aminomethyl)-benzylamine (0.62 mg, 2.64 mmol) was added to a For synthesis see 10 and 14. Yield: 95 mg (88 %); HPLC (system I):
solution of Z-AspACHTUNGRE(OSu)-OBzl (1.00 g, 2.20 mmol) in CH2Cl2 (20 mL), tR 1.71 min (> 90 %); ESI-MS: m/z 763.6 [M+H] + ; Mr = 762 calcd for
and the mixture was stirred for 12 h at room temperature. The sol- C36H58N8O10.
vent was evaporated, and the residue was redissolved in EtOAc
(50 mL). The organic layer was washed with H2O (1 J 50 mL), Ac-Ile-Asp[5-(Boc-amino)-1-amidopentane]-(Phe)Sta-Val-Asp-NH2 (18):
NaHCO3 (5 %, 3 J 50 mL), H2O (1 J 50 mL), KHSO4 (5 %, 3 J 50 mL) The synthesis of this compound was performed as described for
and saturated NaCl (1 J 50 mL), dried over MgSO4 and concentrat- 10 and 14. Yield: 65 mg (67 %); HPLC (system I): tR 1.74 min
ed. The product was obtained by precipitation from EtOAc/petrole- ACHTUNGRE(>90 %); ESI-MS: m/z 777.6 [M+H] + ; Mr = 776 calcd for C37H60N8O10.
um ether as a crystalline solid. Yield: 1.03 g (93 %); TLC (CH2Cl2/
MeOH 9:1): Rf 0.34; HPLC (system I): tR 4.04 min; ESI-MS: m/z 576.4 Ac-Ile-Asp[(3-aminomethyl)-1-amidomethylbenzene]-(Phe)Sta-Val-Asp-
[M+H] + , 520.4 [M tBu+H] + , 476.2 [M Boc+H] + ; Mr = 575 calcd NH2 (20): The synthesis of this compound was performed as de-
for C32H37N3O7. scribed for 10 and 14. Yield: 84 mg (83 %); HPLC (system I): tR
1.85 min; (> 90 %); ESI-MS: m/z 811.4 [M+H] + ; Mr = 810 calcd for
Z-Asp[3-(Boc-aminomethyl)-1-amidomethylbenzene]-OH: Z-Asp[3- C40H58N8O10.
(Boc-aminomethyl)-1-amidomethylbenzene]-OBzl (1.13 g,
1.96 mmol) was saponified in EtOH with NaOH (1 m, 1.1 equiv). The Ac-Ile-AspACHTUNGRE[(4-amino)-1-amidobenzene]-(Phe)Sta-Val-Asp-NH2 (22): The
mixture was stirred for 30 min at room temperature, and the prog- synthesis of this compound was performed as described for 10
ress of the reaction was monitored by HPLC. Upon completion of and 14. Yield: 98 mg (91 %); HPLC (system I): tR 1.75 min (> 90 %);
the reaction, the solution was diluted with water, the pH was ad- ESI-MS: m/z 783.4 [M+H] + ; Mr = 782 calcd for C38H54N8O10.
justed to 2 with HCl (1 m), and the product was extracted with
EtOAc (3 J 50 mL). The organic layer was washed with H2O (3 J Macrocycle 11: HOAt (16.8 mg, 0.12 mmol), HATU (46.8 mg,
50 mL), citric acid 5 % (5 J 30 mL) and saturated NaCl (1 J 30 mL), 0.12 mmol) and DIEA (42 mL, 0.25 mmol) were added to a solution
dried over MgSO4 and concentrated. The product was obtained by of 10 (18.6 mg, 24.6 mmol) in DMF (10 4 m, 250 mL) and the solu-
precipitation from EtOAc/petroleum ether as a crystalline solid. tion was stirred for 12 h at room temperature. The DMF was
Yield: 681 mg (72 %); HPLC (system I): tR 3.44 min; ESI-MS: m/z evaporated, and the residue was purified by preparative HPLC
486.4 [M+H] + , 430.4 [M tBu+H] + , 386.2 [M Boc+H] + , 971.6 (system 2, linear gradient from 0 to 10 % B in 5 min, then from 10
[2 M+H] + ; Mr = 485 calcd for C25H31N3O7. to 40 % B in 60 min). Yield: 2.2 mg (12 %); HPLC (system I): tR
2.65 min; HPLC (system IV): tR 8.26 min; ESI-MS: m/z 738.4 [M+H] + ;
Z-Asp[4-(Boc-amino)-1-amidobenzene]-OBzl: The compound was Mr = 737 calcd for C37H51N7O9.
prepared and isolated as described for Z-Asp[3-(Boc-aminomethyl)-
1-amidomethylbenzene]-OBzl. Yield: 1.02 g (85 %); TLC (CH2Cl2/ Macrocycle 13: The cyclic peptide was obtained from 12 as de-
MeOH 95:5): Rf 0.54; HPLC (system I): tR 4.11 min; ESI-MS: m/z 548.6 scribed for 11. The crude product was purified by preparative
[M+H] + , 492.2 [M tBu+H] + , 448.2 [M Boc+H] + , 1095.9 [2 M+H]+; HPLC (system 2, linear gradient from 10 to 20 % B in 5 min, then
Mr = 547 calcd for C30H33N3O7. from 20 to 50 % B in 60 min). Yield: 2.6 mg (13 %); HPLC (system I):
tR 3.02 min; HPLC (system V): tR 7.61 min; ESI-MS: m/z 710.6
Z-Asp[4-(Boc-amino)-1-amidobenzene]-OH: The title compound was [M+H] + ; Mr = 709 calcd for C35H47N7O9.
prepared and isolated as described for Z-Asp[3-(Boc-aminomethyl)-
1-amidomethylbenzene]-OH. Yield: 728 mg (80 %); HPLC (system I): Macrocycle 15: HOAt (46.6 mg, 0.34 mmol), HATU (130.1 mg,
tR 3.21 min; ESI-MS: m/z 458.4 [M+H] + , 402.4 [M tBu+H] + , 358.2 0.34 mmol) and DIEA (117 mL, 0.68 mmol) were added to a solution
[M Boc+H] + , 915.6 [2 M+H] + ; Mr = 457 calcd for C23H27N3O7. of 14 (51.2 mg, 68.4 mmol) in DMF (10 4 m, 690 mL) and the solu-
tion was stirred for 12 h at room temperature. The DMF was
Z-Asp[(3-aminomethyl)-1-amidomethylbenzene]-Sta-Val-Asp-NH2 (10): evaporated, and the residue was redissolved in EtOAc (5 mL). The
The synthesis was conducted by manual solid-phase methodology organic layer was washed with H2O (1 J 5 mL), NaHCO3 (5 %, 3 J
by use of Fmoc/HBTU chemistry on a Rink amide MBHA resin 5 mL), H2O (1 J 5 mL), KHSO4 (5 %, 3 J 5 mL) and saturated NaCl (1 J
(loading: 0.66 mmol g 1). Couplings were performed with Fmoc- 5 mL) and concentrated. The product was dissolved in acetonitrile/
amino acid/HOBt/HBTU/DIPEA (1:1:1:2) in DMF. Deprotection and water and lyophilized. Yield: 20.0 mg (40 %); HPLC (system I): tR
cleavage from the resin were carried out with TFA/TIS/H2O 2.04 min; ESI-MS: m/z 731.6 [M+H] + ; Mr = 730 calcd for C35H54N8O9.
(95:2.5:2.5) at room temperature in 2 h. The resin was filtered off
and the peptide was precipitated with cold tert-butyl methyl Macrocycle 17: The linear precursor 16 was cyclized, and the prod-
ether/hexane (2:1). Yield: 94.5 mg (85 %); HPLC (system I): tR uct was purified as described for 15. Yield: 19.0 mg (37 %); HPLC
2.15 min (> 90 %); ESI-MS: m/z 756.6 [M+H] + ; Mr = 755 calcd for (system I): tR 2.13 min; ESI-MS: m/z 745.4 [M+H] + ; Mr = 744 calcd
C37H53N7O10. for C36H56N8O9.

2088 www.chembiochem.org B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2007, 8, 2078 – 2091
Macrocyclic Statine-Based Inhibitors of BACE-1

Macrocycle 19: The compound was prepared from 18 and isolated thylbenzene]-OH and the Asp-4 as Fmoc-AspACHTUNGRE(ODmab)-OH. Stan-
as described for 15. Yield: 6.6 mg (21 %); HPLC (system I): tR dard procedures were used for couplings, Fmoc cleavage and
2.25 min; ESI-MS: m/z 759.6 [M+H] + ; Mr = 758 calcd for C37H58N8O9. ACHTUNGREdeprotection/resin cleavage steps. Yield: 117.0 mg (73 %); HPLC
(system I): tR 3.46 min (> 90 %); ESI-MS: m/z: 1068.8 [M+H] + , 535.0
Macrocycle 21: The compound was prepared from 20 and isolated [(M+2 H)/2] + ; Mr = 1067 calcd for C57H77N7O13.
as described for 15. Yield: 9.0 mg (18 %); HPLC (system I): tR
2.55 min; ESI-MS: m/z 793.6 [M+H] + ; Mr = 792 calcd for C40H56N8O9.
Macrocycle p (Table 8): The linear precursor peptide was cyclized as
Macrocycle 23: The compound was prepared from 22 as described described above for compound m. The solvent was evaporated,
for 15. After completion of the reaction, the solvent was evaporat- and the product was dissolved in acetonitrile/H2O (1:1) and lyophi-
ed, and the residue was washed with H2O (1 J 5 mL), NaHCO3 (5 %, lized. The lyophilizate (7.9 mg, 7.5 mmol) was dissolved in DMF
3 J 5 mL), H2O (1 J 5 mL), KHSO4 (5 %, 3 J 5 mL), water and acetone. (980 mL), and hydrazine (20 mL) was added. After 10 min at room
The solid was taken up in MeOH/H2O and lyophilized. Yield: 3.4 mg temperature, the product was precipitated with diethyl ether and
(10 %); HPLC (system I): tR 2.50 min; ESI-MS: m/z 765.4 [M+H] + ; purified by preparative HPLC (system 2, linear gradient from 0 to
Mr = 764 calcd for C38H52N8O9. 25 % B in 10 min, then from 25 to 55 % B in 60 min). Yield: 2.8 mg
(51 %); HPLC (system I): tR 2.92 min; ESI-MS: m/z: 739.4 [M+H] + ;
Synthesis of type B macrocycles
Mr = 738 calcd for C37H50N6O10.
Z-Asp[(3-aminomethyl)-1-amidomethylbenzene]-Sta-Val-OH (l;
Table 7): The synthesis of this compound was performed manually
on 2-chlorotrityl resin (loading: 0.83 mmol g 1) by Fmoc/tBu
chemistry. Couplings were carried out in the standard manner with Table 8. Synthesis of type C macrocycles. X: Sta, R: benzyloxycarbonyl.
HBTU/HOBt/DIPEA in DMF, and Fmoc cleavage with piperidine
(20 %) in DMF. Deprotection and cleavage from the resin was ach-
Spacer
ieved with TFA/TIS/H2O (95:2.5:2.5) at room temperature in 2 h.
The resin was filtered off and the peptide was precipitated with
cold tert-butyl methyl ether/hexane (2:1). Yield: 136 mg (85 %); p
HPLC (system I): tR 1.39 min (> 90 %); ESI-MS: m/z 642.6 [M+H] + ;
Mr = 641 calcd for C33H47N5O8.

Table 7. Synthesis of type B macrocycles. X: Sta, R: benzyloxycarbonyl. NMR conformational analysis and modelling studies: NMR spec-
tra were recorded on a Bruker DRX 500 spectrometer at 1 mm con-
Spacer centrations in [D6]DMSO. TOCSY spectra were recorded with spin
lock periods of 60 ms with use of the MLEV-17 sequence for iso-
l m tropic mixing.[19] For ROESY experiments the standard pulse pro-
gram from Bruker was used. Data processing and assignment was
n o performed with XWINNMR-v3. Distance geometry (DG) and molec-
ular dynamics-simulated annealing (MD-SA) calculations were per-
formed with the INSIGHT II (version 98.0) software package (Accel-
rys, San Diego, CA) on Silicon Graphics O2 R5000 computers (Sili-
con Graphics, Inc., Mountain View, CA). One hundred structures
Z-AspACHTUNGRE[(4-amino)-1-amidobenzene]-Sta-Val-OH (n): The title com-
were generated from the distance-bound matrices. Triangle-bound
pound was obtained as described above for l. Yield: 112 mg (73 %);
smoothing was used. NOE intensities were extracted from spectra
HPLC (system I): tR 2.36 min (> 90 %); ESI-MS: m/z: 614.6 [M+H] + ;
recorded at 30 8C and were converted into interproton distance
Mr = 613 calcd for C31H43N5O8.
constraints with use of the following classification: very strong (vs)
Macrocycle m: HOAt (21.8 mg, 0.16 mmol), HATU (61 mg, 1.7–2.3 M, strong (s) 2.2–2.8 M, medium (m) 2.6–3.4 M, weak (w)
0.16 mmol) and DIEA (55 mL, 0.32 mmol) were added to a solution 3.0–4.0 M, very weak (vw) 3.2–4.8 M, and the distances of pseudo-
of peptide l (20.52 mg, 32.0 mmol) in DMF (10 4 m, 320 mL), and atoms were corrected as described by WTthrich.[20] The structures
the solution was stirred for 12 h at room temperature. The DMF were generated in four dimensions and then reduced to three di-
was evaporated, and the product was purified by preparative HPLC mensions with the EMBED algorithm and optimized with a simulat-
(system 2, linear gradient from 15 to 25 % B in 5 min, then from 25 ed annealing step according to the standard protocol of the DG II
to 55 % B in 60 min). Yield: 2.6 mg (13 %); HPLC (system I): tR package of INSIGHT II. All one hundred structures were refined
3.09 min; (system IV): tR 10.19 min; ESI -MS: m/z: 624.6 [M+H] + ; with a short MD-SA protocol: after an initial minimization, 5 ps at
Mr = 623 calcd for C33H45N5O7. 300 K were simulated, followed by exponential cooling to ~ 0 K
during 10 ps. A time step of 1 fs was used with the CVFF force
Macrocycle o: The title compound was obtained from the peptide field while the solvent H2O was simulated with a dielectric constant
n as described for compound m. Yield: 1.1 mg (6 %); HPLC (sys- of 80.0. The experimentally determined dihedral and distance con-
tem I): tR 2.79 min; HPLC (system IV): tR 8.14 min; ESI-MS: m/z: straints were applied at every stage of the calculation with 30 kcal
596.4 [M+H] + , 1191.8 [2 M+H] + ; Mr = 595 calcd for C31H41N5O7. mol 1·rad 2 and 50 kcal mol 1 M 2, respectively.
Synthesis of type C macrocycles
BACE-1 cell assays
Z-Asp[(3-aminomethyl)-1-amidomethylbenzene]-Sta-Val-AspACHTUNGRE(ODmab)-
OH: The synthesis was conducted manually on 2-chlorotrityl resin Cloning of BACE expression constructs: For the purification of solu-
(loading: 0.78 mmol g 1) by Fmoc/HBTU/DMF chemistry. The Asp-1 ble BACE, a myc-His-tag was fused to the soluble ectodomain of
residue was introduced as Z-Asp[3-(Boc-aminomethyl)-1-amidome- BACE truncated at amino acid 454 and cloned into the pcDN A4

ChemBioChem 2007, 8, 2078 – 2091 B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2089
 L. Moroder et al.

myc/His/A vector. The cDNA was verified for the correct sequence Structure determination and refinement: Coordinates of BACE (PDB-
by automated sequencing.[11b] ID 2B8 L)[23] were used after removal of ligand and water atoms as
search model for initial rigid-body refinement in APRV. Five percent
Cell culture and cell lines: Human embryonic kidney 293 (HEK 293)
of all data were used for Rfree calculation. Amino acid side-chains
cells were maintained in Dulbecco’s modified Eagle’s medium
were fitted to 2 Fo Fc and Fo Fc electron density maps by using
(Sigma) supplemented with foetal bovine serum (10 %) and antibi-
Coot.[24] Subsequent maximum likelihood refinement was per-
otics. Stable transfections of HEK 293 cells were performed with
formed with REFMAC 5.2.[25] After the first refinement cycle, water
Lipofectamine reagent (Invitrogen) as described by the manufac-
molecules and ligand were located in the electron density map
turer. Cell lines expressing BACE constructs with the pcDN A3/Zeo
and added to the model. Refinement at later stages was performed
were selected with Zeocin (Invitrogen, 400 mg mL 1). When trans-
after calculation of TLS parameters.[26] The final model was validat-
fected in human bAPP HEK 293 cells (269) the cells were selected
ed with PROCHECK.[27] Refinement statistics are given in Table 9.
with Geneticin (400 mg mL 1) and Zeocin (Invitrogen, 400 mg mL 1).
Inhibitor treatment and analysis: The inhibitors were applied on
269 cells for 16 h after a 2 h pretreatment. The cellular activity of Table 9. Refinement statistics.[a]
BACE1 in treated cells relative to DMSO-treated control cells was
specifically measured by detecting the N-terminal cleavage prod- resolution [M] 20.0–2.12
uct bAPPs from supernatants of APP expressing cell lines in stan- number of reflections (working/test) 76 009/4001
dard Western blotting procedures with use of the neo-epitope spe- Rcryst [%] 20.3
cific antibody 192 (Elan) for detection. The levels were also moni- Rfree[b] [%] 23.8
total number of atoms
tored for aAPPs with the monoclonal antibody 6E10 (Senetek) and
protein 8851
with the polyclonal antibody 5313 to detect overall secretion of
water 291
total APPs. ligand 57
deviation from ideal geometry[c]
Enzyme assays: Kinetic assays were performed on a 96-well Syner-
bond lengths [M] 0.008
gy HT Bio-Tek plate reader with use of a BACE stock solution in
bond angles [8] 1.17
phosphate-buffered saline containing Triton X-100 (0.1 %) and bonded B’s[d] [M[b]] 1.0
stored at 80 8C in aliquots as described previously.[11b] The con- Ramachandran plot[e]
centration of the enzyme in the stock solution was 1.25 nm, as esti- most favoured regions 88.9
mated by Bradford assay (Bio-Rad). Active enzyme concentrations additional allowed regions 10.7
were obtained by active site titration with the statine-based inhibi- generously allowed regions 0.4
tor 2 and by fitting of data from inhibition experiments approach- disallowed regions 0.0
ing titration conditions to the general equation for tight-binding [a] Values as defined in REFMAC5, without sigma cut-off. [b] Test set con-
inhibitors.[11b] Enzyme inhibition was measured in NaOAc (50 mm, tains 5 % of measured reflections. [c] Root mean square deviations from
pH 4.5) at 37 8C, with use of Mca-(Asn670,Leu671)-amyloidb1 A4 pro- geometric target values. [d] Calculated with MOLEMAN. [e] Calculated
tein precursor770ACHTUNGRE(667–675)-LysACHTUNGRE(Dnp) ammonium salt (Bachem) as with PROCHECK.
the fluorogenic substrate (68 mm final concentration) and 0.14 nm
enzyme concentration at 328 nm excitation and 420 nm emission.
Enzyme and buffer were incubated for 10 min at 37 8C prior to ad-
dition of substrate and inhibitor. Substrate hydrolysis was moni- Acknowledgements
tored over 20 min with inhibitor concentrations ranging from 0 mm
to 77.8 mm with gentle shaking of the samples. Obtaining of rela- The study was supported by the Deutsche Forschungsgemein-
tive fluorescent units (RFU)/min permitted calculation of a Ki value schaft (SFB 469, A2 and SFB 596, B2).
by nonlinear regression analysis (vi/vo = 1/(1+[I]/[Ki]). Compounds
a–p were inactive up to 1 mm concentration. The inhibitory poten-
cies of compounds 1–22 are discussed in the article. Keywords: BACE-1 inhibitors · macrocycles · NMR
conformation · synthesis · X-ray structures
X-ray crystallography
Protein purification and crystallization: The protein was prepared as [1] a) D. J. Selkoe, Nature 1999, 399, A23 – A31; b) T. E. Golde, Brain Pathol.
described previously.[6] Briefly, protein was purified from inclusion 2005, 15, 84 – 87.
bodies and, after refolding and tag cleavage, a buffer exchange to [2] a) H. Potter, D. Dressler, Nature Biotechnol. 2000, 18, 125 – 126; b) D.
TRIS (pH 8.0, 20 mm) containing urea (50 mm), imidazole (25 mm) Beher, S. L. Graham, Expert Opin. Invest. Drugs 2005, 14, 1385 – 1409;
and EDTA (0.5 mm) was performed. Subsequently, the protein solu- c) M. S. Wolfe, Nat. Rev. Drug Discovery 2002, 1, 859 – 866; d) A. K.
tion was adjusted to a concentration of approximately 10 mg mL 1. Ghosh, L. Hong, J. Tang, Curr. Med. Chem. 2002, 9, 1135 – 1144; e) J. N.
Inhibitor complex was prepared by addition of a DMSO stock solu- Cumming, U. Iserloh, M. E. Kennedy, Curr. Opin. Drug Discovery Dev.
tion of the inhibitor to a final concentration of 2 mm and incuba- 2004, 7, 536 – 556; f) L. A. Thompson, J. J. Bronson, F. C. Zusi, Curr.
Pharm. Des. 2005, 11, 3383 – 3404.
tion for several hours prior to crystallization. Crystals of the BACE–
[3] a) S. Sinha, J. P. Anderson, R. Barbour, G. S. Basi, R. Caccavello, D. Davis,
inhibitor complex were obtained by vapour diffusion against a M. Doan, H. F. Dovey, N. Frigon, J. Hong, K. Jacobson-Croak, N. Jewett,
ACHTUNGREreservoir solution of PEG 4000 (10 %) in MES (0.1 m, pH 6.0). P. Keim, J. Knops, I. Lieberburg, M. Power, H. Tan, G. Tatsuno, J. Tung, D.
Schenk, P. Seubert, S. M. Suomensaari, S. W. Wang, D. Walker, J. Zhao, L.
Data collection: Crystals of appropriate size were optimized by use
McConlogue, V. John, Nature 1999, 402, 537 – 540; b) L. Hong, G.
of the Free-Mounting-System[21] and flash-frozen in liquid nitrogen. Koelsch, X. L. Lin, S. L. Wu, S. Terzyan, A. K. Ghosh, X. C. Zhang, J. Tang,
X-ray diffraction data were collected at 100 K at the Swiss Light Science 2000, 290, 150 – 153; c) L. Hong, R. T. Turner, G. Koelsch, D. G.
Source, Villingen, Switzerland. Data processing and scaling were Shin, A. K. Ghosh, J. Tang, Biochemistry 2002, 41, 10963 – 10967; d) R. K.
performed by use of the APRV package.[22] Data collection statistics Hom, L. Y. Fang, S. Mamo, J. S. Tung, A. C. Guinn, D. E. Walker, D. L.
are given in Table 6. Davis, A. F. Gailunas, E. D. Thorsett, S. Sinha, J. E. Knops, N. E. Jewett,

2090 www.chembiochem.org B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 2007, 8, 2078 – 2091
Macrocyclic Statine-Based Inhibitors of BACE-1

J. P. Anderson, V. John, J. Med. Chem. 2003, 46, 1799 – 1802; e) J. S. McGaughey, D. Colussi, M. C. Crouthamel, M. T. Lai, B. Pietrak, E. A. Price,
Tung, D. L. Davis, J. P. Anderson, D. E. Walker, S. Mamo, N. Jewett, R. K. S. Sankaranarayanan, A. J. Simon, G. R. Seabrook, D. J. Hazuda, N. T.
Hom, S. Sinha, E. D. Thorsett, V. John, J. Med. Chem. 2002, 45, 259 – 262; Pudvah, J. H. Hochman, S. L. Graham, J. P. Vacca, P. G. Nantermet, Bioorg.
f) J. Hu, C. L. Cwi, D. L. Smiley, D. Timm, J. A. Erickson, J. E. McGee, H.-C. Med. Chem. Lett. 2007, 17, 4057 – 4061.
Yang, D. Mendel, P. C. May, M. Shapiro, J. R. McCarthy, Bioorg. Med. [6] S. Patel, L. Vuillard, A. Cleasby, C. W. Murray, J. Yon, J. Mol. Biol. 2004,
Chem. Lett. 2003, 13, 4335 – 4339; g) K. G. Bridges, R. Chopra, L. Lin, K. 343, 407 – 416.
Svenson, A. Tam, G. X. Jin, R. Cowling, F. Lovering, T. N. Akopian, E. Di- [7] J. D. A. Tyndall, T. Nall, D. P. Fairlie, Chem. Rev. 2005, 105, 973 – 999.
Blasio-Smith, B. Annis-Freeman, T. H. Marvell, E. R. LaVallie, R. S. Zollner, [8] a) R. Haubner, D. Finsinger, H. Kessler, Angew. Chem. 1997, 109, 1440 –
J. Bard, W. S. Somers, M. L. Stahl, R. W. Kriz, Peptides 2006, 27. 1877 – 1456; Angew. Chem. Int. Ed. Engl. 1997, 36, 1374 – 1389; b) J. Rizo, L. M.
1885. Gierasch, Annu. Rev. Biochem. 1992, 61, 387 – 418.
[4] a) C. A. Coburn, S. J. Stachel, K. G. Jones, T. G. Steele, D. M. Rush, J. Di- [9] J. D. A. Tyndall, D. P. Fairlie, Curr .Med. Chem. 2001, 8, 893 – 907.
Muzio, B. L. Pietrak, M. T. Lai, Q. Huang, J. Lineberger, L. X. Jin, S. [10] a) A. Barazza, C. Renner, M. Willem, L. Moroder, Biopolymers 2005, 80,
Munshi, M. K. Holloway, A. Espeseth, A. Simon, D. Hazuda, S. L. Graham, 554; b) A. Barazza, M. Gçtz, C. Renner, M. Willem, L. Moroder, Under-
J. P. Vacca, Bioorg. Med. Chem. Lett. 2006, 16, 3635 – 3638; b) W. J. Yang, standing Biology Using Peptides (Proceedings of the 19th American Pep-
W. L. Lu, Y. F. Lu, M. Zhong, J. Sun, A. E. Thomas, J. M. Wilkinson, R. V. tide Symposium; Ed.: S. E. Blondelle), Springer, New York, pp. 385 – 386,
Fucini, M. Lam, M. Randal, X. P. Shi, J. W. Jacobs, R. S. McDowell, E. M. 2005.
Gordon, M. D. Ballinger, J. Med. Chem. 2006, 49, 839 – 842; c) S. J. Sta- [11] a) A. Capell, L. Meyn, R. Fluhrer, D. B. Teplow, J. Walter, C. Haass, J. Biol.
chel, C. A. Coburn, T. G. Steele, K. G. Jones, E. F. Loutzenhiser, A. R. Chem. 2002, 277, 5637 – 5643; b) G. G. Westmeyer, M. Willem, S. F.
Gregro, H. A. Rajapakse, M. T. Lai, M. C. Crouthamel, M. Xu, K. Tugu- ACHTUNGRELichtenthaler, G. Lurman, G. Multhaup, I. Assafalg-Machleidt, K. Reiss, P.
ACHTUNGREsheva, J. E. Lineberger, B. L. Pietrak, A. S. Espeseth, X. P. Shi, E. Chen- Saftig, C. Haass, J. Biol. Chem. 2004, 279, 53205 – 53212.
Dodson, M. K. Holloway, S. Munshi, A. J. Simon, L. Kuo, J. P. Vacca, J. [12] a) A. K. Ghosh, D. Shin, D. Downs, G. Koelsch, K. L. Lin, J. Ermolieff, J.
Med. Chem. 2004, 47, 6447 – 6450; d) R. K. Hom, A. F. Gailunas, S. Mamo, Tang, J. Am. Chem. Soc. 2000, 122, 3522 – 3523; b) R. T. Turner, G.
L. Y. Fang, J. S. Tung, D. E. Walker, D. Davis, E. D. Thorsett, N. E. Jewett, Koelsch, L. Hong, P. Castenheira, A. Ghosh, J. Tang, Biochemistry 2001,
J. B. Moon, V. John, J. Med. Chem. 2004, 47, 158 – 164; e) S. F. Brady, S. 40, 10001 – 10006.
Singh, M. C. Crouthamel, M. K. Holloway, C. A. Coburn, V. M. Garsky, M. [13] G. B. McGaughey, D. Colussi, S. L. Graham, M.-T. Lai, S. K. Munshi, P. G.
Bogusky, M. W. Pennington, J. P. Vacca, D. Hazuda, M. T. Lai, Bioorg. Med. Nantermet, B. Pietrak, H. A. Rajapakse, H. G. Selnick, S. R. Stauffer, M. K.
Chem. Lett. 2004, 14, 601 – 604; f) J. Lamar, J. D. Hu, A. B. Bueno, H. C. Holloway, Bioorg. Med. Chem. Lett. 2007, 17, 1117 – 1121.
Yang, D. Q. Guo, J. D. Copp, J. McGee, B. Gitter, D. Timm, P. May, J. Mc- [14] R. T. Turner, 3rd, G. Koelsch, L. Hong, P. Castanheira, A. K. Ghosh, J. Tang,
Carthy, S. H. Chen, Bioorg. Med. Chem. Lett. 2004, 14, 239 – 243; g) S. H. Biochemistry 2001, 40, 10 002 – 10 006.
Chen, J. Lamar, D. Q. Guo, T. Kohn, H. C. Yang, J. McGee, D. Timm, J. [15] R. N. Zuckermann, J. M. Kerr, S. B. H. Kent, W. H. Moos, J. Am. Chem. Soc.
Erickson, Y. Yip, P. May, J. McCarthy, Bioorg. Med. Chem. Lett. 2004, 14, 1992, 114, 10646 – 10647.
245 – 250; h) B. H. Hu, K. Y. Fan, K. Bridges, R. Chopra, F. Lovering, D. [16] E. Kaiser, R. L. Colescot, C. D. Bossinge, P. I. Cook, Anal. Biochem. 1970,
Cole, P. Zhou, J. Ellingboe, G. X. Jin, R. Cowling, J. Bard, Bioorg. Med. 34, 595 – 598.
Chem. Lett. 2004, 14, 3457 – 3460; i) T. Kimura, D. Shuto, Y. Hamada, N. [17] T. Vojkovsky, Pept. Res. 1995, 8, 236 – 237.
Igawa, S. Kasai, P. Liu, K. Hidaka, T. Hamada, Y. Hayashi, Y. Kiso, Bioorg. [18] W. S. Hancock, J. E. Battersby, Anal. Biochem. 1976, 71, 260 – 264.
Med. Chem. Lett. 2005, 15, 211 – 215; j) A. K. Ghosh, N. Kumaragurubar- [19] A. Bax, D. G. Davis, J. Magn. Reson. 1985, 65, 355 – 360.
an, L. Hong, S. S. Kulkarni, X. M. Xu, W. P. Chang, V. Weerasena, R. [20] K. WTthrich, NMR of Proteins and Nucleic Acids, Wiley, New York, 1986.
Turner, G. Koelsch, G. Bilcer, J. Tang, J. Med. Chem. 2007, 50, 2399 – [21] R. Kiefersauer, M. Than, H. Dobbek, L. Gremer, M. Melero, S. Strobl, J. M.
2407; k) M. C. Maillard, R. K. Hom, T. E. Benson, J. B. Moon, S. Mamo, M. Dias, T. Soulimane, R. Huber, Appl. Crystallogr. 2000, 33, 1223 – 1230.
Bienkowski, A. G. Tomasselli, D. D. Woods, D. B. Prince, D. J. Paddock, [22] M. Kroemer, M. K. Dreyer, K. U. Wendt, Acta Crystallogr. Sect. D Biol. Crys-
T. L. Emmons, J. A. Tucker, M. S. Dappen, L. Brogley, E. D. Thorsett, N. tallogr. 2004, 60, 1679 – 1682.
Jewett, S. Sinha, V. John, J. Med. Chem. 2007, 50, 776 – 781; l) J. C. [23] S. J. Stachel, C. A. Coburn, T. G. Steele, M. C. Crouthamel, B. L. Pietrak,
Barrow, K. E. Rittle, P. L. Ngo, H. G. Selnick, S. L. Graham, S. M. Pitzen- M. T. Lai, M. K. Holloway, S. K. Munshi, S. L. Graham, J. P. Vacca, Bioorg.
berger, G. B. McGaughey, D. Colussi, M. T. Lai, Q. Huang, K. Tugusheva, Med. Chem. Lett. 2006, 16, 641 – 644.
A. S. Espeseth, A. J. Simon, S. K. Munshi, J. P. Vacca, ChemMedChem [24] P. Emsley, K. Kowtan, Acta Crystallogr. Sect. D Biol. Crystallogr. 2004, 60,
2007, 2, 995 – 999. 2126 – 2132.
[5] a) S. J. Stachel, C. A. Coburn, S. Sankaranarayanan, E. A. Price, B. L. Pie- [25] G. N. Murshudov, A. A. Vagin, E. J. Dodson, Acta Crystallogr. Sect. D Biol.
trak, Q. Huang, J. Lineberger, A. S. Espeseth, L. X. Jin, J. Ellis, M. K. Hollo- Crystallogr. 1997, 53, 240 – 255.
way, S. Munshi, T. Allison, D. Hazuda, A. J. Simon, S. L. Graham, J. P. [26] M. Winn, M. Isupov, G. N. Murshudov, Acta Crystallogr. Sect. D Biol. Crys-
Vacca, J. Med. Chem. 2006, 49, 6147 – 6150; b) S. Hanessian, G. Q. Yang, tallogr. 2001, 57, 122 – 133.
J. M. Rondeau, U. Neumann, C. Betschart, M. Tintelnot-Blomley, J. Med. [27] R. Laskowski, M. MacArthur, D. Moss, J. Thornton, J. Appl. Crystallogr.
Chem. 2006, 49, 4544 – 4567; c) I. Rojo, J. A. Martin, H. Broughton, D. 1993, 26, 283 – 291.
Timm, J. Erickson, H. C. Yang, J. R. McCarthy, Bioorg. Med. Chem. Lett.
2006, 16, 191 – 195; d) A. K. Ghosh, T. Devasamudram, L. Hong, C. De-
Zutter, X. M. Xu, V. Weerasena, G. Koelsch, G. Bilcer, J. Tang, Bioorg. Med.
Chem. Lett. 2005, 15, 15 – 20; e) S. R. Lindsley, K. P. Moore, H. A. Raja- Received: July 12, 2007
pakse, H. G. Selnick, M. B. Young, H. Zhu, S. Munshi, L. Kuo, G. B. Published online on October 26, 2007

ChemBioChem 2007, 8, 2078 – 2091 B 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chembiochem.org 2091