Structure-based design: Potent inhibitors of human brain memapsin 2 (β-secretase)

Article abstract of DOI:10.1021/jm0101803

Memapsin 2 (β-secretase) is one of two proteases that cleave the β-amyloid precursor protein (APP) to produce the 40-42 residue amyloid-β peptide (Aβ) in the human brain, a key event in the progression of Alzheimer's disease. On the basis of the X-ray crystal structure of our lead inhibitor (2, OM99-2 with eight residues) bound to memapsin, we have reduced the molecular weight and designed potent memapsin inhibitors. Structure-based design and preliminary structure-activity studies have been presented.

Full text of DOI:10.1021/jm0101803

J . Med. Chem. 2001, 44, 2865-2868
2865
Str u ctu r e-Ba sed Design : P oten t
In h ibitor s of Hu m a n Br a in Mem a p sin 2
(â-Secr eta se)
Arun K. Ghosh,*,† Geoffrey Bilcer,†
Cynthia Harwood,† Reiko Kawahama,†,‡
Dongwoo Shin,† Khaja Azhar Hussain,†,‡ Lin Hong,§
J effrey A. Loy,§ Chan Nguyen,§ Gerald Koelsch,§
J acques Ermolieff,§ and J ordan Tang§,|
F igu r e 1.
Department of Chemistry, University of Illinois at Chicago,
845 West Taylor Street, Chicago, Illinois 60607, Zapaq, Inc.,
Oklahoma City, Oklahoma 73104, Protein Studies Program,
Oklahoma Medical Research Foundation,
of 36 nM). Subsequently, we determined the crystal
structure of the protease domain of human memapsin
2 complexed to inhibitor 2 at 1.9 Å resolution.6 This
crystal structure provided invaluable information re-
garding the specific ligand-binding site interactions in
the active site of memapsin 2. The therapeutic potential
of inhibitor 2 may be limited because of its high
molecular weight (1100 Da) and numerous peptide
bonds. However, the structure of 2 serves as an impor-
tant molecular template for structure-based design of
memapsin inhibitor drugs. Such strategies have been
successfully utilized in the design and synthesis of
inhibitors of other aspartyl proteases including HIV
protease inhibitor drugs.7 The ability of some HIV
inhibitor drugs to cross the blood-brain barrier is
particularly encouraging for development of memapsin
2 inhibitor drugs for treatment of AD.8 Herein, we report
preliminary results of our investigation in which X-ray
structure-based modification of our lead inhibitors
resulted in a series of potent memapsin inhibitors. The
molecular size of these inhibitors is substantially re-
duced while maintaining comparable enzyme inhibitory
potencies to 2.
Resu lts a n d Discu ssion . Our reported X-ray crystal
structure of 2 bound to memapsin 2 provided detailed
information relating to specific interaction and orienta-
tion of the inhibitor in the active site.6 In this structure,
the inhibitor chain turns at P2′ and the side chains of
the P3′-P4′ residues are not involved in any specific
interaction in the S3′-S4′ sites of memapsin 2. The
residues point toward the molecular surface of the
protein, suggesting that the P3′-P4′ residues could be
shortened. Further inspection of the crystal structure
revealed that the P2′-carbonyl is involved in hydrogen
bonding to the hydroxyl of Tyr-198, forming a rare kink
at P2′. The P1-P4 side chains in general are involved in
significant interactions in the S1-S4 subsites. While P1-
Leu and P3-Val fill in the hydrophobic pocket effectively,
the P2-Asn is involved in a hydrogen bond with Arg-
235. The P4-Glu site is hydrogen bonded to P2-Asn,
possibly assisting in the formation of the P2-Asn hydro-
gen bond with Arg-235. Thus, it appeared that the P2-
ligand binding may be further optimized and the P4-
ligand may be eliminated or simplified. With this
molecular insight into the ligand-binding site, we first
elected to delete the P3′-P4′ residues and simplify the
ligand using a benzylamide to preserve the critical P2′-
carbonyl interaction and fill in the hydrophobic pocket
in the S3′ region of the active site. Second, we planned
to probe the new P2-ligand with a view to delete the P3
and P4 residues and further shorten the inhibitor size.
and Department of Biochemistry and Molecular Biology,
University of Oklahoma Health Science Center,
Oklahoma City, Oklahoma 73104
Received April 23, 2001
Abstr a ct: Memapsin 2 (â-secretase) is one of two proteases
that cleave the â-amyloid precursor protein (APP) to produce
the 40-42 residue amyloid-â peptide (Aâ) in the human brain,
a key event in the progression of Alzheimer’s disease. On the
basis of the X-ray crystal structure of our lead inhibitor (2,
OM99-2 with eight residues) bound to memapsin, we have
reduced the molecular weight and designed potent memapsin
inhibitors. Structure-based design and preliminary structure-
activity studies have been presented.
In tr odu ction . The proteolytic enzymes â- and γ-secre-
tases that cleave the â-amyloid precursor protein (APP)
to produce the 40-42 residue amyloid-â peptide (Aâ)
in the human brain are key players in the progression
of Alzheimer’s disease (AD).1 While the γ-secretase has
not been conclusively identified, the recent discovery of
â-secretase as an aspartyl protease marked the begin-
ning of a significant effort for therapeutic intervention
of AD.2 The â-secretase cleaves an easily accessible site
at the lumenal side of APP, and its activity is the rate-
limiting step in Aâ production in vivo.3 Recently, we and
others cloned the human brain aspartyl protease
memapsin 2 (also known as BACE and ASP-2) and
demonstrated it to be the long sought â-secretase.4
Memapsin 2 is an excellent target for inhibitor drugs
since it occupies the initial step in the pathological
cascade of AD. Thus, the in vivo inhibition of memapsin
2 may reduce the production of Aâ and thereby slow or
halt the progression of AD. On the basis of knowledge
of kinetics and specificity of memapsin 2, we initially
designed two seven- and eight-residue peptidomimetic
inhibitors incorporating a nonhydrolyzable Leu-Ala
hydroxyethylene dipeptide isostere as depicted in Figure
1.5 The inhibitor based on the sequence Glu-Val-Asn-
Leu-Ψ-Ala-Ala-Glu-Phe, where Ψ denotes replacement
of CONH by (S)-CH(OH)CH2 (2, OM99-2), has exhibited
a Ki value of 1.6 nM for human memapsin 2, and it is
considerably more potent than inhibitor 1 (OM99-1, Ki
* To whom correspondence should be addressed. Phone: 312-996-
9672. Fax: 312-996-2183. E-mail: arunghos@uic.edu.
† University of Illinois at Chicago.
‡ Zapaq, Inc.
§ Protein Studies Program, Oklahoma Medical Research Foundation.
| University of Oklahoma Health Science Center.
10.1021/jm0101803 CCC: $20.00 © 2001 American Chemical Society
Published on Web 08/03/2001

2866 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 18
Letters
Ta ble 1. Structure and Potencies of Memapsin Inhibitors
Sch em e 1a
a
(a) LiHMDS, MeI, THF, -78 °C, 30 min; (b) aq LiOH, THF-
H2O, 23 °C, 8 h; (c) TBDMSCl, imidazole, DMF, 24 h; (d) EDC,
HOBt, amine 9, iPr2NEt, DMF-CH2Cl2 (1:1), 0-23 °C; (e)
nBu4N+F-, THF, 23 °C; (f) CF3CO2H, CH2Cl2, 23 °C; (g) EDC,
HOBt, acid, iPr2NEt, DMF-CH2Cl2 (1:1), 0-23°C.
F igu r e 2.
The general synthesis of various inhibitors containing
the hydroxyethylene isostere9 is outlined in Scheme 1.
The known10 γ-lactones 3 and 4 were alkylated stereo-
selectively by generation of the corresponding dianion
with lithium hexamethyldisilazide (2.2 equiv) in tet-
rahydrofuran at -78 °C (30 min) followed by addition
of methyl iodide (1.1 equiv) at -78 °C (30 min). The
reaction was quenched with propionic acid (5 equiv)
followed by standard workup and provided the desired
alkylated lactones 5 and 6 (70-76% yield) along with a
small amount (<5%) of the corresponding epimer.11 The
stereochemical assignment of the alkylated lactones was
supported by 1H NMR NOE experiments. Saponification
of these lactones followed by protection of the γ-hydroxyl
group with tert-butyldimethylsilyl chloride in the pres-
ence of imidazole and (dimethylamino)pyridine in di-
methylformamide afforded the corresponding Leu-Ala
7 and Phe-Ala 8 isosteres (85-90% yield). These hy-
droxyethylene isosteres were converted to various in-
hibitors using a standard peptide coupling procedure.
Thus, valine derivative 9 was reacted with dipeptide
isostere 7 in the presence of N-ethyl-N′-(dimethylami-
nopropyl)carbodiimide hydrochloride, diisopropylethy-
lamine, and 1-hydroxybenzotriazole hydrate in a mix-
ture of DMF and CH2Cl2 to afford amide derivative 10.
Removal of the silyl protecting group by treatment with
tetrabutylammonium fluoride in THF afforded memapsin
inhibitor 11. For synthesis of inhibitor 13, Boc-deriva-
tive 11 was first exposed to trifluoroacetic acid (TFA)
in CH2Cl2 to effect removal of the Boc group. Coupling

Letters
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 18 2867
F igu r e 3. Energy minimized model structure of inhibitor 22 (magenta) overlayed with OM99-2 (green) bound X-ray crystal
structure of memapsin 2.
of the resulting amine with Boc-asparagine provided
inhibitor 13. Treatment of 13 with TFA and subsequent
coupling of the resulting amine with Boc-valine afforded
inhibitor 16. The synthesis of inhibitor 19 with P2-
methionine was carried out by exposure of 11 to TFA
followed by coupling of the resulting amine with Boc-
methionine. Removal of the Boc group in 19 and
coupling of the resulting amine with Boc-valine provided
inhibitor 22. Oxidation of 22 with oxone in a mixture
(1:1) of methanol and water at 23 °C for 12 h furnished
inhibitor 23. Other inhibitors in Table 1 were prepared
by analogous procedure as described above. Inhibitor
20 with pyridylmethyl urethane was prepared by re-
moval of the Boc group in 19 followed by coupling of
the resulting amine with pyridylmethylsuccinimidyl
carbonate in CH2Cl2 as described previously (Figure
2).12 By following similar procedures, the corresponding
inhibitor 21 with a phenylalanine side chain at P1 was
prepared from isostere 8.
Structure and inhibitory potencies of various pepti-
domimetic inhibitors against recombinant memapsin 2
are shown in Table 1.13 Our initial inhibitor target had
been inhibitor 12 with Leu-Ala isostere, P2-Asn, and P2′-
Ala-benzylamide; however, it is a weak inhibitor with
a Ki of 22.4 µM against memapsin 2. Because the
2-bound crystal structure revealed that the S2′-site could
possibly accommodate a side chain larger than P2′-Ala,
we have examined the effect of Val at P2′. Indeed, the
resulting inhibitor 13 has exhibited at least a 7-fold
potency enhancement. As mentioned earlier, the P2-Asn
weakly hydrogen bonds to Arg-235 (bonding distance
3.86 Å). We therefore, incorporated a methyl sulfone
derivative at P2, assuming that one of the sulfone
oxygens could conceivably hydrogen bond effectively
with the Arg-235. Such hydrogen bonding capabilities
of sulfone derivatives have been previously demon-
strated.14 The corresponding inhibitor 14 improved its
Ki to 1.1 µM. However, it appeared that the extension
of P3-Val is necessary for further improvement of
inhibitor potency.
Thus, we have appended the P3-Val in 12, and the
resulting inhibitor 15 has shown a marked improvement
in its potency to 61.4 nM (Ki value). Consistent with
earlier observation, the corresponding P2′-Val derivative
16 has exhibited Ki of 5.9 nM, a 10-fold potency
enhancement over inhibitor 15. Inhibitor with P2-methyl
sulfone derivative (inhibitor 18) showed comparable
memapsin inhibitory potency to inhibitor 16, indicating
that one of the sulfone oxygens may be involved in
hydrogen bonding to Arg-235. This speculation is sup-
ported by the fact that the corresponding inhibitor with
P2-methylcysteine is much less potent than 18. Molec-
ular modeling studies support this possible hydrogen
bonding interaction. Interestingly, incorporation of P2-
Met resulted in the potent inhibitor 22 (Ki 2.5 nM) with
comparable inhibitory potency to 2 which has eight
residues. Contrary to inhibitors 17 and 18, oxidation of
the methyl sulfide to sulfone derivative 23 resulted in
a 3-fold attenuation in potency compared to 22. To gain
insight into specific ligand-binding site interactions, an
energy minimized model structure of 22 was created in
the active site of memapsin (the active site based on
the crystal structure of memapsin bound to 2). On the
basis of this model, it seems that the methyl sulfide
moiety of P2-Met hydrogen bonds to Arg-235 in the S2-
region of the active site. This interaction may simply
be a hydrophobic interaction as well. Generation of a
crystal structure of this complex is currently being
attempted in order to further understand this interac-
tion. Furthermore, it appears that the P2′-Val has
substantial hydrophobic contact with the enzyme. The
key hydrogen bonding interaction with Tyr-198 is also
quite apparent as shown in Figure 3.15 Preliminary
structure-activity studies indicated that Leu-Ala isos-
tere is preferred over Leu-Leu isostere (compound 24)
by the enzyme active site. Comparison of inhibitory

2868 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 18
Letters
(4) (a) Lin, X.; Koelsch, G.; Wu, S.; Downs, D.; Dashti, A.; Tang, J .
Human aspartic protease memapsin 2 cleaves the beta-secretase
site of beta-amyloid precursor protein. Proc. Natl. Acad. Sci.
U.S.A. 2000, 97, 1456. (b) Sinha, S.; Lieberburg, I. Cellular
mechanisms of beta-amyloid production and secretion. Proc.
Natl. Acad. Sci. U.S.A. 1999, 96, 11049-11053.
(5) Ghosh, A. K.; Shin, D.; Downs, D.; Koelsch, G.; Lin, X.; Ermolieff,
J .; Tang, J . Design of potent inhibitors for human brain
memapsin 2 (â-secretase). J . Am. Chem. Soc. 2000, 122, 3522-
23.
(6) Hong, L.; Koelsch, G.; Lin, X.; Wu, S.; Terzyan, S.; Ghosh, A.
K.; Zhang, X. C.; Tang, J . Structure of the protease domain of
memapsin 2 (â-secretase) complexed with inhibitor. Science
2000, 290, 150-53.
(7) (a) Flexner, C. HIV-Protease inhibitors. N. Engl. J . Med. 1998,
338, 1281-92. (b) Cihlar, T.; Bischofberger, N. Recent develop-
ments in antiretroviral therapies. Ann. Rep. Med. Chem. 2000,
35, 177-89 and references therein.
properties of compounds 20 (Ki 1.4 mM) and 21 (Ki 2.1
mM) also suggested preference for Leu-Ala over Phe-
Ala isostere by the memapsin active site. Inhibitor 22
also potently inhibited memapsin 1, the closest homolo-
gous protease to memapsin 2,4a with a Ki value of 1.2
nM. Since memapsin 1 is also known to cleave â-amyloid
precursor protein at the â-secretase site, this cross-
inhibition is not entirely surprising.16 Furthermore, it
has been suggested that memapsin 1 may be involved
in the Down’s syndrome form of Alzheimer’s disease.17
If this is confirmed, the cross-inhibition of compound
22 to both enzymes may be of some medicinal impor-
tance.
Con clu sion . Structure-based modification of our lead
inhibitor (2) led to the discovery of a series of potent
and considerably low molecular weight peptidomimetic
memapsin 2 inhibitors. The P3′-, P4′-, and P4-peptidic
ligands in 2 have been deleted, and the P2-ligand has
been replaced with a designed ligand to allow effective
hydrogen bonding with specific residues in the S2-active
site. The size of the modified inhibitors (molecular
weight around 722) are comparable to a number of
approved peptidomimetic HIV protease inhibitor drugs.
Further investigation of peptidomimetic approaches
including specific nonpeptidal ligand design will follow
from these preliminary studies. Our efforts are now
directed toward the incorporation of features in the
inhibitor to address issues relating to cell penetration,
pharmacokinetic properties, and crossing of the blood-
brain barrier.
(8) (a) Sathle, L.; Martin, C.; Svensson, J .; Sonnerborg, A. Indinavir
in Cerebrospinal Fluid of HIV-1 Infected Patients. Lancet 1997,
350, 1823. (b) Brinkman, K.; Kroon, F.; Hugen, P.; Burger, D.
M. Therapeutic Concentrations of Indinavir in Cerebrospinal
Fluid of HIV-1 Infected Patients. AIDS 1998, 12, 537.
(9) For synthesis of hydroxyethylene isostere for renin inhibitor,
see: (a) Rich, D. H.; Salituro, F. G.; Holladay, M. W. Design of
protease inhibitors. Proc. Am. Pept. Symp. 8th 1983, 511-20.
(b) Holladay, M. W.; Rich, D. H. Synthesis of hydroxyethylene
and ketomethylene dipeptide isosteres. Tetrahedron Lett. 1983,
24, 4401.
(10) (a) Fray, A. H.; Kaye, R. L.; Kleinman, E. F. A short, stereose-
lective synthesis of the lactone precursor to 2R,4S,5S hydroxy-
ethylene dipeptide isosteres. J . Org. Chem. 1986, 51, 4828-4833.
(b) Ghosh, A. K.; McKee, S. P.; Thompson, W. J . An efficient
synthesis of hydroxyethylene dipeptide isosteres: the core unit
of potent HIV-1 protease inhibitors. J . Org. Chem. 1991, 56,
6500-02.
(11) Ghosh, A. K.; Fidanze, S. Transition-State Mimetics for HIV
Protease Inhibitors: Stereocontrolled Synthesis of Hydroxyeth-
ylene and Hydroxyethylamine Isosteres by Ester-Derived Tita-
nium Enolate Syn and Anti-Aldol Reactions. J . Org. Chem. 1998,
63, 6146-6154 and references therein.
(12) Ghosh, A. K.; Duong, T. T.; McKee, S. P.; Thompson, W. J . N,N′-
Disuccinimidyl carbonate: a useful reagent for alkoxycarbony-
lation of amines. Tetrahedron Lett. 1992, 33, 2781-84.
(13) Memapsin 2 inhibition was measured using recombinant enzyme
produced from E. coli expression as described in ref 4a. A
fluorogenic substrate Arg-Glu(EDANS)-Glu-Val-Asn-Leu-Asp-
Ala-Glu-Phe-Lys (Dabcyl)-Arg was used with 0.47 µM of the
enzyme in 0.1 M Na-acetate + 5% dimethylsuloxide, pH 4.5 at
37 °C. The excitation wavelength was 350 nm, and the emission
wavelength was 490 nm. Details of this assay will be described
shortly.
Ack n ow led gm en t. Financial support by the Na-
tional Institutes of Health (AG 18933 and GM 53386)
is gratefully acknowledged. G.K. is a Scientist Develop-
ment Awardee of the American Heart Association
(9930115N). J .T. is the holder of the J .G. Puterbaugh
Chair in Biomedical Research at the Oklahoma Medical
Research Foundation.
Su p p or tin g In for m a tion Ava ila ble: Experimental pro-
cedures for compounds 5-11, 13, 16, 19-22 along with purity
(HPLC) and mass spectroscopy data. This material is available
free of charge via the Internet at http://pubs.acs.org.
(14) Ghosh, A. K.; Lee, H. Y.; Thompson, W. J .; Culberson, C.;
Holloway, M. K.; McKee, S. P.; Munson, P. M.; Duong, T. T.;
Smith, A. M.; Darke, P. L.; Zugay, J . A.; Emini, E. A.; Schleif,
W. A.; Huff, J . R.; Anderson, P. S. The development of cyclic
sulfolanes as novel and high affinity P2-ligands for HIV-1
protease inhibitors. J . Med. Chem. 1994, 37, 1177-88 and
references therein.
Refer en ces
(1) Selkoe, D. J . Translating cell biology into therapeutic advances
in Alzheimer’s disease. Nature 1999, 399A, 23-31.
(2) (a) Wolfe, M. S. et al. Two transmembrane aspartates in
presenilin-1 required for presenilin endoproteolysis and γ-secre-
tase activity. Nature 1999, 398, 513. (b) Olson, R. E.; Thompson,
L. A. Secretase Inhibitors as therapeutics for Alzheimer’s
Disease. Ann. Rep. Med. Chem. 2000, 35, 31.
(3) Vassar, R.; Bennett, B. D.; Babu-Khan, S.; Kahn, S.; Mendiaz,
E. A.; Denis, P.; Teplow, D. B.; Ross, S.; Amarante, P.; Loeloff,
R.; Luo, Y.; Fisher, S.; Fuller, J .; Edenson, S.; Lile, J .; J arosinski,
M. A.; Biere, A. L.; Curran, E.; Burgess, T.; Louis, J . C.; Collins,
F.; Treanor, J .; Rogers, G.; Citron, M. Beta-secretase cleavage
of Alzheimer’s amyloid precursor protein by the transmembrane
aspartic protease BACE. Science 1999, 286, 735-41.
(15) To understand the actual interaction in the active site, deter-
mination of crystal structure protein-ligand complex of 22 is
in progress.
(16) Farzan, M., Schnitzler, C. E., Vasilieva, N., Leung, D., Choe, H.
A â-secretase homolog, cleaves at the â-site and within the
amyloid-â-region of the amyloid-â-preursor protein. Proc. Natl.
Acad. Sci. U.S.A. 2000, 97, 9712-17.
(17) Tanzi, R. E. Neuropathology in the Down’s syndrome brain. Nat.
Med. 1996, 2, 31-2.
J M0101803