Discovery of novel non-peptide inhibitors of BACE-1 using virtual high-throughput screening

Article abstract of DOI:10.1016/j.bmcl.2009.09.103

A novel series of isatin-based inhibitors of β-secretase (BACE-1) have been identified using a virtual high-throughput screening approach. Structure-activity relationship studies revealed structural features important for inhibition. Docking studies suggest these inhibitors may bind within the BACE-1 active site through H-bonding interactions involving the catalytic aspartate residues.

Full text of DOI:10.1016/j.bmcl.2009.09.103

Bioorganic & Medicinal Chemistry Letters 19 (2009) 6770–6774
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of novel non-peptide inhibitors of BACE-1 using virtual
high-throughput screening
a,
N. Yi Mok a, James Chadwick a, Katherine A. B. Kellett b, Nigel M. Hooper b, , A. Peter Johnson
,
*
*
Colin W. G. Fishwick a,
*
a School of Chemistry, University of Leeds, Leeds LS2 9JT, UK
b Institute of Molecular and Cellular Biology, and Leeds Institute of Genetics, Health and Therapeutics, University of Leeds, Leeds LS2 9JT, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 30 June 2009
Revised 22 September 2009
Accepted 24 September 2009
Available online 2 October 2009
A novel series of isatin-based inhibitors of b-secretase (BACE-1) have been identified using a virtual high-
throughput screening approach. Structure–activity relationship studies revealed structural features
important for inhibition. Docking studies suggest these inhibitors may bind within the BACE-1 active site
through H-bonding interactions involving the catalytic aspartate residues.
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
Alzheimer’s disease
b-Secretase
BACE-1
Virtual high-throughput screening
eHiTS
Alzheimer’s disease (AD) accounts for the majority of dementia
diagnosed in patients after the age of 60.1 The formation of insolu-
ble extracellular amyloid plaques by the accumulation of amyloid
b-peptide (Ab) is one of the key pathological features in the brains
of AD sufferers.2 As outlined by the amyloid cascade hypothesis,3
Ab is generated by the proteolytic cleavage of the b-amyloid pre-
cursor protein (APP). b-Secretase (b-site APP cleaving enzyme-1,
or BACE-1) is the first of two enzymes responsible for the sequen-
tial processing of APP, and is considered rate-limiting in the prote-
olytic cleavage process.4 Since BACE-1 knockout mice have been
shown to be viable with only mild phenotypic changes affecting
peripheral nerve development, the inhibition of BACE-1 is consid-
ered a promising therapeutic approach for the treatment of AD.5
BACE-1 belongs to the family of pepsin-like human aspartyl
proteases, characterized by a single transmembrane domain. With
the catalytic site located on the lumenal side of the membrane,
BACE-1 processes the substrate APP at Met671-Asp672 (APP770
numbering) to release soluble sAPPb. The membrane-bound frag-
A number of inhibitors of BACE-1 have been reported.6 In the
case of the earliest reports, the majority of these involved pep-
tide-based structures with relatively high molecular weights and
unfavorable properties in terms of penetration of the blood-brain
barrier (BBB). More recently, however, a range of potent non-pep-
tidic inhibitors of moderate size have been described.6 In light of
the wealth of X-ray crystallographic data available for BACE-1,
we were interested in designing small molecule non-peptide inhib-
itors using structure-based methods.7 Herein, we report the dis-
covery of a novel series of non-peptide inhibitors of BACE-1
using the virtual high-throughput screening software eHiTS.8
eHiTS is an exhaustive flexible docking algorithm with a scoring
function which incorporates both empirical and knowledge-based
features.9,10 Input ligands are divided into rigid fragments and flex-
ible connecting chains.
The rigid fragments are then docked independently within the
active site of BACE-1 and assessed for the best pose combination
before re-connection with the connecting chains to generate the fi-
nal docked conformation. Using the high-throughput screening
mode within the program, three compound libraries containing a
total of 250,000 commercially available compounds11 were
screened against the BACE-1 X-ray crystal structure co-crystallized
with a peptide-based inhibitor (PDB code 1M4H).12 From the top-
ranked structures, six molecules were selected for purchase, based
upon predicted binding affinity and synthetic tractability (see Sup-
plementary data). These compounds were tested in an in vitro
activity assay using recombinant human BACE-1 and a quenched
ment C99 is then cleaved by c-secretase to generate the amyloido-
genic Ab.
* Corresponding authors. Tel.: +44 113 3433163; fax: +44 113 3435638 (N.M.H.);
tel.: +44 113 3436515; fax: +44 113 3436530 (A.P.J.); tel.: +44 113 3436510; fax:
+44 113 3436530 (C.W.G.F.).
E-mail addresses: n.m.hooper@leeds.ac.uk (N.M. Hooper), p.johnson@leeds.ac.uk
(A.P. Johnson), c.w.g.fishwick@leeds.ac.uk (C.W.G. Fishwick).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.09.103

N. Yi Mok et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6770–6774
6771
Table 1
Compound library for SAR studies of a small molecule BACE-1 inhibitor
O
N
O
R3
N
H
N
R2
R1
1-8
Compound
R1
R2
R3
BACE-1 inhibition at 100 l
Ma,
% (IC50 in brackets, l
M)b
O
1
OH
H
NO2
84.5 2.1
(2.4 0.3)
N
H
O
O
O
2
3
4
NO2
NO2
H
28.8 3.6
20.2 0.1
71.1 2.4
N
H
OMe
N
H
OH
N
H
5
6
H
Me
OH
OH
NO2
NO2
35.2 4.9
14.3 4.9
O
7
8
OH
OH
NO2
NO2
92.0 2.5
(4.8 0.1)
N
H
F3C
O
74.6 3.4c
N
H
a
b
c
Inhibition of BACE-1 activity measured as a percentage of the control activity in an in vitro BACE-1 activity assay (see text). Values shown are mean SEM.
IC50 only determined for compounds attaining over 80% BACE-1 inhibition at 100
Precipitation of assayed compound observed during incubation.
lM.
fluorescent peptide substrate based on the Swedish mutant APP se-
quence (SEVNLDAEFK).13 Of the six compounds assayed, com-
pound 1 was identified as a BACE-1 inhibitor with an IC50 value
A series of structural analogs (2–8, Table 1) were synthesized to
probe the structural features required for biological activity. These
molecules were prepared using standard methods involving cou-
pling of various isatins with hydrazides.14 A range of N-substituted
of 2.4 lM (Table 1).
O
O
O
O
O
a
c or d
N
H
N
N
O
O
R1
R4
O
4
1
9
10
12
R
R
R
= OEt
= OH
=
=
=
b
4
N
11
H
O
O
13 R1
N
H
CF3
1
14
R
N
H
Scheme 1. Synthesis of isatin-N-acetamides 12–14. Reagents and conditions: (a) ethyl bromoacetate, K2CO3, DMF, 0 °C, 16 h, 52%; (b) 10% NaOH/EtOH, rt, 10 min,
quantitative; (c) (i) SOCl2, 60 °C, 3 h; (ii) R–NH2, CH2Cl2, 0 °C, 20 h; (d) R–NH2, HOBt, EDAC, CH2Cl2, 0 °C, 20 h, 55–62%.

6772
N. Yi Mok et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6770–6774
O
O
R3
H2N
R3
R5
N
H
c
R2
R2
15a
15b
R
R
2 = OH, R3 = NO2, R5 = OH
a
b
2 = OMe, R3 = NO2, R5 = OMe
R
2 = OMe, R3 = NO2
16b
15c R2 = H, R3 = NO2, R5 = OMe
15d R2 = OH, R3 = H, R5 = OMe
15e R2 = OH, R3 = NO2, R5 = OMe
16c R2 = H, R3 = NO2
16d R2 = OH, R3 = H
16e R2 = OH, R3 = NO2
Scheme 2. Synthesis of hydrazides 16b–e. Reagents and conditions: (a) MeI, KOH, DMSO, rt, 16 h, 65%; (b) fuming HNO3, AcOH, heat, 20 min, 40%; (c) hydrazine
monohydrate, MeOH, reflux, 3 h, 36–85%.
O
O
O
O
H2N
R3
N
O
R3
a or b
N
H
N
H
N
R2
N
R2
R1
R1
9, 12-14
16b-e
2-8
1
17
R
= Me
Scheme 3. Synthesis of analogs 2–8. Reagents and conditions: (a) AcOH, reflux, 3 h; (b) AcOH, microwave, 300 W, 150 °C, 30–60 min, 38–88%.
isatins were prepared by reacting isatin 9 with ethyl bromoacetate
to give ester 10, which was subsequently hydrolyzed to give isatin-
N-acetic acid 11.15
tial role for the acetamide portion of compound 1 in BACE-1
inhibition. By retaining all the structural features of the original
hit compound 1, compounds 7 (p-tolyl moiety of 1 replaced with
a cyclohexyl ring) and 8 (p-trifluoromethyl in place of the p-tolyl
moiety in 1) show comparable levels of BACE-1 inhibition (92%
Isatin-N-acetamides 12–14 were prepared by either forming the
acid chloride intermediate (method c), or direct coupling with the
appropriate amine using HOBt and EDAC (method d) (Scheme 1).
The hydrazides were prepared by reaction of appropriately
substituted methyl benzoates 15b–e with hydrazine to give hydra-
zides 16b–e (Scheme 2).16,17 Condensation of the isatins with the
hydrazides using either conventional reflux in acetic acid or under
microwave irradiation delivered the desired structures in the com-
pound library (Scheme 3).
and 75%, respectively, at 100 lM) to that of compound 1 (Table
1). The selectivity of these inhibitors for BACE-1 was established
following our observation that they do not appear to inhibit the
activity of the closely related enzyme BACE-2 (data not shown).
Additionally, in order to rule out aggregation-related non-specific
inhibition, we observed that the presence or absence of detergent
had no effect upon the inhibitory activities. Furthermore, control
experiments involving fluorescence measurements of the substrate
in the presence of the inhibitors, but in the absence of enzyme,
established that the inhibitors had no effect on the fluorescence
behavior of the substrate.
Firstly, we investigated the effects of altering the phenolic func-
tionality of inhibitor 1 on BACE-1 inhibition. Removal of the pheno-
lic OH (Table 1, as in compound 2) resulted in a significant
reduction in BACE-1 inhibition at 100 lM, indicating that the pres-
ence of the phenolic moiety is important for BACE-1 inhibition. The
methoxy analog compound 3 showed only 20% BACE-1 inhibition
at 100 lM, further supporting the importance of the phenolic func-
tionality in the inhibition of BACE-1.
The role of the nitro group in compound 1 was also explored.
The des-nitro version (compound 4) showed a BACE-1 inhibition
Having established the important molecular features within
these isatin-derived inhibitors required for BACE-1 inhibition, we
wished to probe further the possible binding pose of compound
1 within the BACE-1 active site. It has been previously reported
that isatin 3-benzoylhydrazone 18 (Fig. 1), a substructure of 1,
adopts a near planar structure in solution due to the delocalization
of 71% at 100
l
M, indicating that the nitro function appears not
of the p electrons along the –N–NH–C(@O)– scaffold in addition to
to be important for binding to BACE-1.
an intramolecular H-bond between the isatin carbonyl and the
hydrazone N–H.18 For compound 1, modeling indicated that the
tendency for planarity of the molecule is reinforced by additional
intramolecular H-bonding involving the phenolic O–H to yield
either of the two conformations 19 and 20 (Fig. 2).
Next, we introduced different substituents onto the isatin por-
tion of the inhibitor in order to investigate the importance of the
acetamide portion in 1. BACE-1 inhibition dropped considerably
when R1 was replaced by H or Me (Table 1, compounds 5 and 6,
respectively). The detrimental effects observed indicated an essen-
H
O
O
O
O
N
O
N
O
NO2
N
H
N
H
N
N
N
N
O
H
H
N
H
O
R1
NO2
R1
18
19
20
Figure 1. Structure of isatin 3-benzoylhydrazone 18, appears to adopt a planar
Figure 2. The two important conformations of compound 1 involving the formation
conformation.
of intramolecular H-bonding with the phenolic O–H.

N. Yi Mok et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6770–6774
6773
the inhibitor adopts a planar conformation, which is stabilized by
intramolecular H-bonding from the phenolic moiety in 1. Addition-
ally, binding to BACE-1 appears to involve H-bonding interactions
between the p-tolylamide of 1 and the catalytic residue Asp228.
A recent report20 detailing the discovery of a series of potent small
molecule BACE-1 inhibitors compares the ligand efficiency (LE)21 of
a range of reported inhibitors of BACE-1. In this study, the authors
noted that despite the high potency of the previously reported pep-
tide-based BACE-1 inhibitors such as OM99-2 (Ki = 1.6 nM),22 the
relatively high molecular weights of these systems (e.g., OM99-2
has a molecular weight of 893) often result in them having rela-
tively poor ligand efficiency (e.g., LE = 0.19 for OM99-2). For the
present study, although still somewhat below the preferred mini-
mum value of LE = 0.3,21 compound 1 (molecular weight = 461)
has LE = 0.22 and therefore, is closer to the preferred value than
the potent but considerably larger peptidic inhibitors reported pre-
viously. Although the relatively poor solubility of the present series
of compounds (e.g., clog P values for compounds 1 and 7 are 3.97
and 3.66, respectively), coupled with the presence of the nitro
and phenolic moieties render them challenging candidates for fur-
ther development, we have demonstrated that eHiTS is a powerful
screening tool to identify biologically active compounds quickly
and efficiently.
Arg307
Thr232
Asp32
Asp228
P1
O
Asp228
O
-
N
N
N
O
H
O
N
H
O
Acknowledgments
-
O
H
N
O
H
Arg307
The authors would like to thank the Alzheimer’s Research Trust
(J.C., K.A.B.K.) and the Hong Kong Croucher Foundation (N.Y.M.) for
financial support.
N
H
+
+
H
O
N
P1'
NH2
H
N
P4
P2
Thr232
Supplementary data
Figure 3. (top), binding pose of 1 (corresponding to conformer 20) in the BACE-1
active site generated using AutoDock; (bottom), schematic showing predicted
H-bond contacts and positions of BACE-1 substrate binding pockets.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2009.09.103.
AutoDock19 was used to separately dock the two proposed fa-
vored conformations 19 and 20 of compound 1 into the active site
of BACE-1 (PDB code 1M4H). While the docking of conformer 19
did not give any solutions consistent with the observed biological
activity (data not shown), docking of conformer 20 revealed a
binding pose which was consistent with the observed activities
of compounds 1–8.
References and notes
1. Citron, M. Trends Pharmacol. Sci. 2004, 25, 92.
2. Selkoe, D. J. Phys. Rev. 2001, 81, 741.
3. Hardy, J.; Selkoe, D. J. Science 2002, 297, 353.
4. Johnstone, J. A.; Liu, W. W.; Todd, S. A.; Coulson, D. T. R.; Murphy, S.; Irvine, G.
B.; Passmore, A. P. Biochem. Soc. Trans. 2005, 33, 1096. and references therein.
5. Eder, J.; Hommel, U.; Cumin, F.; Martoglio, B.; Gerhartz, B. Curr. Pharm. Des.
2007, 13, 271.
6. For recent reviews, see: (a) Cole, S. L.; Vassar, R. Curr. Alzheimer Res. 2008, 5,
100; (b) Huang, W. H.; Sheng, R.; Hu, Y. Z. Curr. Med. Chem. 2009, 16, 1806.
7. For a recent example, see: (a) Xu, W.; Chen, G.; Liew, O. W.; Zuo, Z.; Jiang, H.;
Zhu, W. Bioorg. Med. Chem. Lett. 2009, 19, 3188; For a recent review see: (b)
Holloway, M. K.; Hunt, P.; McGaughey, G. B. Drug Dev. Res. 2009, 70, 70.
8. For a useful comparison of eHiTS with other docking algorithms as applied to
the identification of putative BACE-1 inhibitors, see: Polgar, T.; Magyar, C.;
Simon, I.; Keseru, G. M. J. Chem. Inf. Model. 2007, 47, 2366.
9. Zsoldos, Z.; Reid, D.; Simon, A.; Sadjad, B. S.; Johnson, A. P. Curr. Protein Pept. Sci.
2006, 7, 421.
10. Zsoldos, Z.; Reid, D.; Simon, A.; Sadjad, B. S.; Johnson, A. P. J. Mol. Graphics
Modell. 2007, 26, 198.
Figure 3 (top) shows the lowest-energy binding pose identified
for conformer 20 within BACE-1. It is noteworthy that an analo-
gous binding pose was also identified using eHiTS. The acetamide
moiety is predicted to occupy the catalytic site, with the acetamide
N–H acting as an H-bond donor to the catalytic residue Asp228
0
(H-bond length = 1.86 ÅA). This is consistent with the observed bio-
logical activity as the structural analogs lacking the acetamide por-
tion (compounds 5 and 6) displayed very poor activity towards
BACE-1. The phenol unit is predicted to make an H-bond contact
0
11. eHiTS version 5.3 was used. Compounds were obtained for screening from the
collections in the Sigma–Aldrich, SALOR, and Maybridge HitDiscover screening
libraries.
12. Hong, L.; Turner, R. T.; Koelsch, G.; Shin, D. G.; Ghosh, A. K.; Tang, J. Biochemistry
2002, 41, 10963.
with the backbone nitrogen of Thr232 (H-bond length = 2.20 ÅA)
and to partly occupy the P2 substrate pocket. This feature implies
the phenol might be involved in both the formation of the intramo-
lecular H-bonding network, and also in intermolecular H-bonding
interactions with the enzyme. The nitro group in 1 is predicted
to extend into the P4 pocket (Fig. 3, bottom, and see also the Sup-
plementary data), possibly participating in wea0k H-bonding with
the side chain of Arg307 (H-bond length = 2.13 ÅA), consistent with
the slight decrease in the binding affinity exhibited by compound
4.
13. Hussain, I.; Hawkins, J.; Harrison, D.; Hille, C.; Wayne, G.; Cutler, L.; Buck, T.;
Walter, D.; Demont, E.; Howes, C.; Naylor, A.; Jeffrey, P.; Gonzalez, M. I.;
Dingwall, C.; Michel, A.; Redshaw, S.; Davis, J. B. J. Neurochem. 2007, 100, 802.
BACE-1 assay protocol as described herein with the following exceptions;
25 nM recombinant BACE-1 (R&D systems) incubated with inhibitor
compounds for 30 min before kinetic reading over 30 min at 2-min intervals
on a Synergy HT fluorimeter. Rate of relative fluorescence unit (RFU) increase
calculated and compared to DMSO control RFU increase rate to calculate%
inhibition.
14. Typical procedure: To a solution of isatin 9 (3.4 mmol) in DMF (10 mL) was
added potassium carbonate (5.1 mmol). Ethyl bromoacetate (4.1 mmol) was
added dropwise at 0 °C, and the reaction was stirred at room temperature for
16 h. The precipitate was poured into an ice-water mixture, and the orange
In summary, using the virtual high-throughput screening soft-
ware eHiTS, we have discovered a novel non-peptidic inhibitor of
BACE-1 based on an isatin motif. Studies of the biological activity
of structural variants in combination with in silico docking suggest

6774
N. Yi Mok et al. / Bioorg. Med. Chem. Lett. 19 (2009) 6770–6774
precipitate formed filtered, washed with water, and recrystallized in water to
1H), 7.89 (br s, 1H), 7.68 (d, 1H), 7.45 (t, 1H), 7.21–7.16 (m, 2H), 7.00 (d, 1H),
4.39 (s, 2H), 3.59 (m, 1H), 1.77–1.69 (m, 4H), 1.58–1.53 (m, 1H), 1.30–1.17 (m,
5H); 13C NMR (125 MHz, DMSO-d6) d 164.1, 161.7, 159.6, 142.9, 139.7, 136.6,
130.9, 128.4, 127.1, 122.5, 120.1, 119.2, 117.9, 117.5, 109.5, 109.3, ꢁ47.5, 41.9,
31.7, 24.6, 23.9; HRMS (ES-) m/z 464.1578 [MꢁH]ꢁ; C23H22N5O6 requires
464.1576; HPLC (5–95% MeCN/H2O) tR 23.91 min (peak area 100%).
15. Ainley, A. D.; Robinson, R. J. Chem. Soc. 1934, 1508.
give ester 10. 10% NaOH was added to a stirring ethanolic solution of 10, and
the solution stirred at room temperature until the color changed to yellow.
Conc. HCl was added to obtain an orange precipitate, which was filtered and
recrystallized in water to give isatin-N-acetic acid 11. To a solution of 11
(4.9 mmol) and cyclohexylamine (12.25 mmol) in anhydrous DCM was added
HOBt (5.9 mmol) and EDAC (5.9 mmol) at 0 °C, and the reaction was stirred at
room temperature for 20 h. The organic layer was diluted, washed with 2 M aq
HCl (3ꢀ25 mL), satd NaHCO3 (3 ꢀ 25 mL), brine (30 mL), dried with MgSO4,
and the solvent removed in vacuo to give the crude product 13, which was
recrystallized in MeOH/H2O (Scheme 1). Methyl salicylate 15d (10.0 mmol)
was reacted with fuming nitric acid (10.0 mmol, 10% v/v) premixed with glacial
AcOH, and the reaction heated until the color changed to brown. The mixture
was cooled and water was added to give a brown precipitate, which was
filtered and purified using flash chromatography on silica gel with 4:1
petroleum ether (40–60 °C)/EtOAc as eluent to afford methyl 5-
16. Barany, H. C.; Pianka, M. J. Chem. Soc. 1946, 965.
17. Baker, W.; Haksar, C. N.; McOmie, J. F. W. J. Chem. Soc. 1950, 170.
18. Ali, H. M.; Halim, S. N. A.; Ng, S. W. Acta Crystallogr., Sect. E 2005, 61, o3285.
19. Huey, R.; Morris, G. M.; Olson, A. J.; Goodsell, D. S. J. Comput. Chem. 2007, 28,
1145.
20. Edwards, P. D.; Albert, J. S.; Sylvester, M.; Aharony, D.; Andisik, D.; Callaghan,
O.; Campbell, J. B.; Carr, R. A.; Chessari, G.; Congreve, M.; Frederickson, M.;
Folmer, R. H. A.; Geschwindner, S.; Koether, G.; Kolmodin, K.; Krumrine, J.;
Mauger, R. C.; Murray, C. W.; Olsson, L.-L.; Patel, S.; Spear, N.; Tian, G. J. Med.
Chem. 2007, 50, 5912.
nitrosalicylate 15e. Hydrazine monohydrate (30.0 mmol) was added to
a
methanolic solution of 15e (7.5 mmol), and the reaction heated under reflux
for 3 h. The mixture was cooled to 0 °C, the yellow precipitate was then filtered
and recrystallized from MeOH to give 16e (Scheme 2). Amide 13 (0.35 mmol)
was mixed with hydrazide 16e (0.35 mmol) in glacial AcOH, and the reaction
was heated to 150 °C in a CEM microwave reactor at 300 W for 30 min. The
precipitate was filtered and washed with EtOAc to give pure product 7. Mp
>250 °C; 1H NMR (500 MHz, DMSO-d6) d 14.30 (br s, 1H), 8.83 (d, 1H), 8.31 (s,
21. The concept of ‘ligand efficiency’ has been recently introduced as a means of
equating the potency of a compound with its size, and provides a measure of
the free energy of binding per heavy atom; see: Hopkins, A. L.; Groom, C. R.;
Alex, A. Drug Discovery Today 2004, 9, 430; Reynolds, C. H.; Tounge, B. A.;
Bembenek, S. D. J. Med. Chem. 2008, 51, 2432.
22. Hong, L.; Koelsch, G.; Lin, X.; Wu, S.; Terzyan, S.; Ghosh, A. K.; Zhang, X. C.;
Tang, J. Science 2000, 290, 150.