Tyramine fragment binding to BACE-1

Article abstract of DOI:10.1016/j.bmcl.2008.01.032

Fragment screening revealed that tyramine binds to the active site of the Alzheimer's disease drug target BACE-1. Hit expansion by selection of compounds from the Roche compound library identified tyramine derivatives with improved binding affinities as m

Full text of DOI:10.1016/j.bmcl.2008.01.032

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 1304–1307
Tyramine fragment binding to BACE-1
Andreas Kuglstatter, Martin Stahl, Jens-Uwe Peters, Walter Huber, Martine Stihle,
Daniel Schlatter, Jo¨rg Benz, Armin Ruf, Doris Roth, Thilo Enderle and Michael Hennig*
F. Hoffmann-La Roche, Pharma Research Basel, Grenzacherstr., 4070 Basel, Switzerland
Received 24 November 2007; revised 20 December 2007; accepted 8 January 2008
Available online 11 January 2008
Abstract—Fragment screening revealed that tyramine binds to the active site of the Alzheimer’s disease drug target BACE-1. Hit
expansion by selection of compounds from the Roche compound library identified tyramine derivatives with improved binding affin-
ities as monitored by surface plasmon resonance. X-ray structures show that the amine of the tyramine fragment hydrogen-bonds to
the catalytic water molecule. Structure-guided ligand design led to the synthesis of further low molecular weight compounds that are
starting points for chemical leads.
ꢀ 2008 Elsevier Ltd. All rights reserved.
The most prominent neuropathological finding in Alz-
heimer’s disease brains is the occurrence of cortical pla-
ques containing b-amyloid peptide. Genetic evidence
points to a causal role of b-amyloid during development
of Alzheimer’s disease as it is postulated by the amyloid
cascade hypothesis.1 Several independent approaches
led to the identification of b-site amyloid precursor pro-
tein cleavage enzyme 1 (BACE-1) as the first enzyme of
the amyloid cascade. BACE-1 cleaves the 695 or 751
amino acid amyloid precursor protein to a 99 amino
acid long precursor peptide which is later trimmed by
c-secretase to form the pathological 40 or 42 amino acid
long b-amyloid peptide. BACE-1 is an integral mem-
brane protein with high homology to other aspartic pro-
teases. The enzymatic activity does not depend on the
presence of the transmembrane or intracellular domains.
Given the lack of adverse effects in gene knockout mice
BACE-1 is considered a prime target for the develop-
ment of Alzheimer’s disease therapeutics.2
hairpin loop referred to as the ‘flap’. In apo structures
˚
of BACE-1, the flap backbone is about 12 A away from
the side chains of D32 and D228 which coordinate the
catalytic water molecule,5,6 whereas in complexes with
peptidomimetic inhibitors like OM99-2 (Fig. 1) the flap
˚
moves about 4 A closer to the catalytic center and the
non-cleavable transition state isostere moiety of the
inhibitors replaces the catalytic water molecule.3,7 The
only reported exception is an oxyacetamide inhibitor
that forms a hydrogen bond to the catalytic water rather
than replacing it.8
To exert their pharmacological activity in vivo, BACE-1
inhibitors have to traverse the blood–brain barrier add-
ing another level of complexity for drug discovery. In
order to identify new lead molecules of low molecular
weight, we have applied a fragment screening strategy.9
Using a combination of computational chemistry and
surface plasmon resonance measurements, we selected
48 compounds of molecular weight 100–150 Da for
soaking into BACE-1 crystals and subsequent X-ray
BACE-1 possesses a bilobal structure3 typical for
eukaryotic aspartic proteases4 with the catalytic aspar-
tate residues D32 and D228 located in the substrate
binding cleft between the N-terminal and C-terminal
lobes.3 The active site is partially covered by a flexible
O
O
O
O
N
O
N
N
N
N
O
N
N
O
O
O
O
O
Keywords: BACE-1; Alzheimer; Fragment screening; Surface plasmon
resonance; X-ray structure; Structure-based design; Tyramine.
N
O
O
O
*
Corresponding author. Tel.: +41 61 6886046; e-mail: Michael.
Hennig@roche.com
7, KD = 0.04µM, LE= 0.16
Figure 1. OM99-23 with BACE-1 KD value determined by surface
plasmon resonance and calculated ligand efficiency (LE).
Present address: Roche Palo Alto LLC, 3431 Hillview Ave., Palo
Alto, CA 94304, USA.
0960-894X/$ - see front matter ꢀ 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.01.032

A. Kuglstatter et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1304–1307
1305
˚
structure determination. Compounds for which the
binding mode could be identified were used as starting
point for the design of low molecular weight inhibitors
of BACE-1. This is in contrast to the recently reported
approach where X-ray crystallography was used without
a prior binding assay to identify fragments that bind to
BACE-1.10
end of the S3 pocket (ꢀ5 A). The ligand efficiency as de-
fined by binding energy per non-hydrogen atom11,12 of 1
is 0.37 and therefore much higher than the ligand effi-
ciency of peptidomimetic BACE-1 inhibitors like
OM99-23 which is 0.16 (Fig. 1).
To expand around the primary screening hit tyramine
1, 32 compounds containing the tyramine substructure
were selected from the Roche compound collection for
soaking into BACE-1 crystals and subsequent X-ray
structure determination. Here we present the crystal
structures of the tyramine derivatives 2–4 (Table 1)
bound to the active site of BACE-1 (Fig. 2). Com-
pound 2 was used as a racemic mixture. Analysis of
difference Fourier omit maps shows that its R enantio-
mer is bound to the protein. The positions of the
amine and hydroxyl groups of 1 and 2 and their inter-
actions in the binding pocket are essentially identical
in both structures. However, the connecting phenyl-al-
kyl moieties differ significantly in their positions. In
compound 2, the phenol substructure common to both
compounds is rotated by about 15ꢁ toward Y71 with
the hydroxyl oxygen as pivot point. While the benzylic
methylene of 1 is in contact with the L30 side chain,
the equivalent moiety of 2 points up toward the flap.
The additional methyl group of compound 2 makes
hydrophobic interactions both with G230 and the side
chain of L30.
One of the compounds for which a crystal structure was
determined is the tyrosine metabolite tyramine 1 (Table
1). It binds to the S1 pocket of the BACE-1 active site
(Fig. 2) with KD = 2000 lM. The flap adopts an open
conformation (Fig. 3) as described for the structure of
the BACE-1 apo form.6 The most striking feature of
the tyramine 1 binding mode is that the catalytic water
molecule remains in the position it adopts in the apo
structure. It is situated in an almost symmetric environ-
ment between the two carboxylate side chains of D32
and D228 and forms a hydrogen bond with the primary
amine of tyramine 1 (Fig. 2). This amine group forms
two further hydrogen bonds, a direct one with the car-
boxylate oxygen of D32 closest to the flap and a
water-mediated one with the hydroxyl group of the flap
residue Y71, thus satisfying all three protons on the pro-
tonated amine group. The hydroxyl group of 1 forms a
hydrogen bond with the backbone carbonyl of F108.
The tyramine phenylethyl moiety forms a number of
rather loose lipophilic interactions with aromatic and
aliphatic side chains in the S1/S3 pocket, the strongest
of which is presumably an edge-to-face contact to
F108. Weaker interactions are formed with L30, Y71,
The ortho-substituted tyramine derivative 3 maintains
the position of the amino group and its interactions with
the catalytic water and D228 as observed for 1 and 2,
but its phenylethyl moiety is shifted toward the S3 pock-
et (Fig. 2). The ethyl substituent of 3 reaches into the S3
pocket where it is in contact with the lipophilic side
chain of I110 and the backbone carbonyl of Q12,
slightly increasing the affinity relative to 1.
˚
and W115 (shortest contacts ꢀ4 A). The compound
barely reaches the side chain of I110 forming the distal
Table 1. Tyramine 1, its derivatives 2–6 with BACE KD values
determined by surface plasmon resonance and calculated ligand
efficiency (LE)
Tyramine derivative 4 protrudes even further into the S3
pocket. Its cyclohexyl moiety interacts with backbone
atoms of G11, Q12, G13, and G230, and with the side
chains of I110 and T232. These interactions result in a
nine-fold affinity increase relative to 1. While 1–3 show
ligand efficiencies of 0.36–0.38, 4 exhibits a lower value
because of its increased molecular weight.
Compound
BACE-1 KD (lM)
LE
N
1
2
3
2000
0.37
O
O
N
N
800
660
0.38
0.36
We reasoned that replacing the cyclohexyl ring of 4 with
an aryl ring should lead to compounds of higher affinity.
The aryl ring should be situated in the S3 pocket facing
the terminal methyl group (CD1) of the I110 side chain
O
O
O
˚
at a distance of about 3.7 A. Indeed, we found that the
aryl-substituted compound 5 (KD = 350 lM) retains the
potency of 4.
4
5
220
350
0.31
0.29
N
N
Tyramine derivative 6 was made to further probe possi-
ble interactions with the S3 pocket (see Supplementry
Materials for synthesis methods). 6 was found to be 30
times more potent in the surface plasmon resonance
binding assay than the initial fragment tyramine 1
resulting in similar ligand efficiency. This was unex-
pected, since the modeled binding modes indicated that
a para-substituent would interfere with the rim of the S3
pocket (G11). We must therefore assume that there is a
6
60
0.35
N
O

1306
A. Kuglstatter et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1304–1307
˚
˚
˚
˚
Figure 2. Crystal structures of compounds 1 (2.3 A resolution, PDB Accession code 3BRA), 2 (2.3 A, 3BUF), 3 (2.5 A, 3BUG), and 4 (2.3 A, 3BUH)
bound to the BACE-1 active site.13 The catalytic water is marked by an asterisk. Fo-Fc omit maps contoured at three standard deviations above the
mean density of the map are in green. The binding mode of compound 6 was modeled based on the structural information.
change in binding mode or receptor conformation to
accommodate compound 6. Attempts to determine a
crystal structure of BACE-1 complexed with 6 were
not successful, possibly because of its low solubility in
the buffer used for soaking of BACE-1 crystals.
the molecules show functional activity, an enzyme inhi-
bition analysis was performed to complement the char-
acterization of the compounds. For all but molecule 2
inhibition of enzymatic activity could be detected con-
firming the results by surface plasmon resonance. How-
ever, due to the high ligand concentrations required to
measure weak inhibition (IC50 > 100 lM), IC50 values
could not be calculated with confidence and therefore
no correlation between KD and IC50 values could be
established. This illustrates the superiority of biophysi-
cal binding assays like surface plasmon resonance over
enzymatic assays for the identification of weakly binding
fragments.
The modifications of the initial tyramine fragment lead
to significant improvement of ligand binding affinity
while the ligand efficacy of the compounds was kept
around 0.35. This is remarkable for the drug target
BACE-1 which is considered to exhibit low chemical
tractability and supports the view that the S1 pocket is
a hot spot for binding capability. In order to assess if

A. Kuglstatter et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1304–1307
1307
Figure 3. Surface representation of (a) OM99-23 and (b) tyramine 1 bound to the BACE-1 active site. The flap indicated in green adopts the closed
conformation in (a) and the open conformation in (b).
7. Ghosh, A. K.; Kumaragurubaran, N.; Tang, J. Curr. Top.
Med. Chem. 2005, 5, 1609.
Fragment screening efforts reveal that the tyrosine
metabolite tyramine and derivatives thereof bind to
the active site of BACE-1. Due to their low molecular
weight and high ligand efficiency these compounds serve
as ideal starting points for the design of BACE-1 inhib-
itors with CNS pharmacological activity. In addition,
the obtained structural information can be used for
morphing or scaffold hopping approaches to improve
other chemical lead series.
8. Coburn, C. A.; Stachel, S. J.; Li, Y. M.; Rush, D. M.;
Steele, T. G.; Chen-Dodson, E.; Holloway, M. K.; Xu, M.;
Huang, Q.; Lai, M. T.; DiMuzio, J.; Crouthamel, M. C.;
Shi, X. P.; Sardana, V.; Chen, Z.; Munshi, S.; Kuo, L.;
Makara, G. M.; Annis, D. A.; Tadikonda, P. K.; Nash, H.
M.; Vacca, J. P.; Wang, T. J. Med. Chem. 2004, 47, 6117.
9. Surface Plasmon Resonance measurements were per-
formed on a Biacore S51 instrument. BACE-1 was
immobilized (ꢀ12,000 RU) by standard amine coupling
chemistry on a CM-5 sensor. Binding experiments were
performed using acetate buffer (50 mM, pH 4.6, 150 mM
NaCl, 3 mM EDTA, 0.005% P20, and 4% DMSO) as the
running buffer. Compounds were dissolved in 100 mM
DMSO and subsequently diluted into acetate buffer in
order to adjust the final DMSO content and the concen-
tration of the respective compound. Hits were further
investigated for specific binding to the active site of the
enzyme by competition experiments using a high affinity
(KD = 40 nM) inhibitor of BACE-1 derived from the
substrate (pGlu-Val-Asn-statin-Val-Ala-Glu-Phen-am).
10. Murray, C. W.; Callaghan, O.; Chessari, G.; Cleasby, A.;
Congreve, M.; Frederickson, M.; Hartshorn, M. J.;
McMenamin, R.; Patel, S.; Wallis, N. J. Med. Chem.
2007, 50, 1116.
Acknowledgments
We thank Manfred Brockhaus for discussions and read-
ing of the manuscript and the staff of the synchrotron
facility Swiss Light Source (Paul Scherrer Institut, Villi-
gen, Switzerland) for their assistance and support. We
thank Josef Schneider and Heribert Dollt for compound
analytics.
Supplementary data
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2008.01.032.
11. Kuntz, I. D.; Chen, K.; Sharp, K. A.; Kollman, P. A.
Proc. Natl. Acad. Sci. USA 1999, 96, 9997.
12. Hopkins, A. L.; Groom, C. R.; Alex, A. Drug Discov.
Today 2004, 9, 430.
13. The K246A surface mutant variant of BACE-1 ectodo-
main (20 mg/ml) was crystallized by vapor diffusion in
hanging drops at 20 ꢁC by mixing 1.5 ll protein solution
with 0.5 ll of 2.5 M sodium formate, 100 mM Hepes (pH
7.0). Crystals of space group P6122 with unit cell
References and notes
1. Citron, M.; Oltersdorf, T.; Haass, C.; McConlogue, L.;
Hung, A. Y.; Seubert, P.; Vigo-Pelfrey, C.; Lieberburg, I.;
Selkoe, D. J. Nature 1992, 360, 672.
2. Hills, I. D.; Vacca, J. P. Curr. Opin. Drug Discov. Dev.
2007, 10, 383.
3. Hong, L.; Koelsch, G.; Lin, X.; Wu, S.; Terzyan, S.;
Ghosh, A. K.; Zhang, X. C.; Tang, J. Science 2000, 290,
150.
4. Dunn, B. M. Chem. Rev. 2002, 102, 4431.
5. Hong, L.; Tang, J. Biochemistry 2004, 43, 4689.
6. Patel, S.; Vuillard, L.; Cleasby, A.; Murray, C. W.; Yon, J.
J. Mol. Biol. 2004, 343, 407.
˚
˚
dimensions a = b = 103 A, c = 169 A grew to a maximum
size of 100 · 100 · 400 lm3. Apo crystals were soaked for
8–16 h in a buffer containing 30–100 mM compound,
2.5 M sodium formate, 0.1 M sodium acetate (pH 4.5),
and 10% DMSO. X-ray diffraction data were collected at
beamline X06SA of the Swiss Light Source (Paul Scherrer
Institut, Villigen, Switzerland). The crystal structure of the
K246A variant protein was solved by molecular replace-
ment using coordinates of an unpublished BACE-1
ectodomain structure as search template.