Unnatural amino acid-substituted (hydroxyethyl)urea peptidomimetics inhibit γ-secretase and promote the neuronal differentiation of neuroblastoma cells

Article abstract of DOI:10.1124/mol.106.024299

γ-Secretase, exhibiting characteristics of aspartyl protease, mediates the intramembranous proteolysis of β-amyloid precursor protein (APP) and Notch, and it is considered to be a prime pharmacological target in the development of therapeutics for Alzheimer's disease (AD). To identify compounds that block γ-secretase-mediated proteolysis, we used a highly sensitive cell-based reporter gene assay for γ-secretase in which Gal4/VP16-tagged C99-APP was expressed as the immediate substrate of γ-secretase, and Gal4/VP16-tagged APP intracellular domain released by the γ-secretase cleavage then activated the expression of the Gal4-driven luciferase reporter gene. Using this reporter assay, we demonstrated that the newly synthesized (hydroxyethyl)urea peptidomimetics, which contain unnatural amino acid moieties at positions P1′ and/or P3′, can effectively inhibit γ-secretase activity and significantly reduce Aβ production. The γ-secretase-dependent S3 cleavage of Notch was also consistently blocked by these (hydroxyethyl) ureas as evidenced by the decreased generation of the Notch intracellular domain, a prerequisite for the activation of Notch signaling. The inhibition of Notch signaling by active Jia compounds efficiently promotes the neuronal differentiation of neuroblastoma cells, intervening in tumorigenesis and the malignancy of neuroblastomas. Our results suggest that (hydroxyethyl) urea peptidomimetics containing unnatural amino acid substitutions could represent a novel class of γ-secretase inhibitors with enhanced stability, providing the basis for the further development of effective therapeutics for AD and neuroblastomas. Copyright

Full text of DOI:10.1124/mol.106.024299

0026-895X/07/7102-588–601$20.00
MOLECULAR PHARMACOLOGY
Copyright © 2007 The American Society for Pharmacology and Experimental Therapeutics
Mol Pharmacol 71:588–601, 2007
Vol. 71, No. 2
24299/3173365
Printed in U.S.A.
Unnatural Amino Acid-Substituted (Hydroxyethyl)urea
Peptidomimetics Inhibit ␥-Secretase and Promote the Neuronal
Differentiation of Neuroblastoma Cells
Yung-Feng Liao, Bo-Jeng Wang, Wen-Ming Hsu, Hsinyu Lee, Chia-Yin Liao,
Shin-Ying Wu, Hui-Ting Cheng, and Ming-Kuan Hu
Laboratory of Molecular Neurobiology, Institute of Cellular and Organismic Biology, Academia Sinica, Taipei, Taiwan (Y.-F.L.,
B.-J.W., S.-Y.W., H.-T.C.); School of Pharmacy, National Defense Medical Center, Taipei, Taiwan (C.-Y.L., M.-K.H.);
Department of Surgery, National Taiwan University Hospital and National Taiwan University, Taipei, Taiwan (W.-M.H.); and
Department of Life Science and Institute of Zoology, National Taiwan University, Taipei, Taiwan (H.L.)
Received March 9, 2006; accepted November 14, 2006
ABSTRACT
␥-Secretase, exhibiting characteristics of aspartyl protease, P3Ј, can effectively inhibit ␥-secretase activity and significantly
mediates the intramembranous proteolysis of ␤-amyloid pre- reduce A␤ production. The ␥-secretase-dependent S3 cleav-
cursor protein (APP) and Notch, and it is considered to be a age of Notch was also consistently blocked by these (hydroxy-
prime pharmacological target in the development of therapeu- ethyl)ureas as evidenced by the decreased generation of the
tics for Alzheimer’s disease (AD). To identify compounds that Notch intracellular domain, a prerequisite for the activation of
block ␥-secretase-mediated proteolysis, we used a highly sen- Notch signaling. The inhibition of Notch signaling by active Jia
sitive cell-based reporter gene assay for ␥-secretase in which compounds efficiently promotes the neuronal differentiation of
Gal4/VP16-tagged C99-APP was expressed as the immediate neuroblastoma cells, intervening in tumorigenesis and the ma-
substrate of ␥-secretase, and Gal4/VP16-tagged APP intracel- lignancy of neuroblastomas. Our results suggest that (hydroxy-
lular domain released by the ␥-secretase cleavage then acti- ethyl)urea peptidomimetics containing unnatural amino acid
vated the expression of the Gal4-driven luciferase reporter substitutions could represent a novel class of ␥-secretase in-
gene. Using this reporter assay, we demonstrated that the hibitors with enhanced stability, providing the basis for the
newly synthesized (hydroxyethyl)urea peptidomimetics, which further development of effective therapeutics for AD and neu-
contain unnatural amino acid moieties at positions P1Ј and/or roblastomas.
␥-Secretase catalyzes the final proteolytic step in the gen- the AD brain), and it has thus been regarded as a prime
eration of A␤ (the principal constituent of senile plaques in
therapeutic target for AD. Mounting evidence from pharma-
cological studies, mutagenesis, affinity labeling, and bio-
chemical isolation strongly suggests that ␥-secretase is an
aspartyl protease and that the active site of ␥-secretase is
This study was supported by National Science Council, Taiwan, grants NSC
93-2320-B-001-037, NSC 94-2320-B-001-004, NSC 94-3112-B-001-001, and
NSC 95-3112-B-001-006 (to Y.-F.L.) and NSC 92-2320-B-016-046 (to M.-K.H.)
and Academia Sinica (to Y.-F.L.). Proteomic mass spectrometry analyses per- located at the interface of the presenilin (PS) heterodimer
formed by the Core Facilities for Proteomics Research located at the Institute
(Wolfe and Haass, 2001). The heterodimeric PS consisting of
of Biological Chemistry, Academia Sinica, were supported by National Science
an ϳ30-kDa N-terminal fragment (NTF) and a ϳ20-kDa
C-terminal fragment (CTF) is thought to be the active form of
PS (Capell et al., 1998). Peptidomimetics that mimic the
Council Grant NSC 91-3112-P-001-009-Y and the Academia Sinica.
Article, publication date, and citation information can be found at
http://molpharm.aspetjournals.org.
doi:10.1124/mol.106.024299.
ABBREVIATIONS: AD, Alzheimer’s disease; PS, presenilin; NTF, N-terminal fragment; CTF, C-terminal fragment; NICD, Notch intracellular
domain; HIV, human immunodeficiency virus; ELISA, enzyme-linked immunosorbent assay; HEK, human embryonic kidney; DMEM, Dulbecco’s
modified Eagle’s medium; FBS, fetal bovine serum; DMSO, dimethyl sulfoxide; APP, amyloid precursor protein; C99, 99-residue C-terminal
fragment of amyloid precursor protein; PBS, phosphate-buffered saline; PLB, passive lysis buffer; DAPT, N-[N-(3,5-difluorophenacetyl-L-alanyl)]-
S-phenylglycine t-butyl ester; GV, Gal4/VP16 DNA binding domain; GAP-43, growth-associated protein 43; PAGE, polyacrylamide gel electro-
phoresis; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; PBST, phosphate-buffer saline/Tween 20; A␤, ␤-amyloid; HRP, horseradish
peroxidase; CHAPSO, 3-[(3-cholamidopropyl)dimethyl-ammonio]-2-hydroxy-1-propanesulfonate; siRNA, small interfering RNA; RNAi, RNA inter-
ference; ESI, electrospray ionization; MS, mass spectrometry, MS/MS, tandem mass spectrometry; LC, liquid chromatography; OD, optical
density; Chy, cyclohexylmethyl; RA, retinoic acid; L-685–458, (2R-(2R*,4R*,5S*))N-(5-(((1,1-dimethylethoxy)carbonyl)amino)-4-hydroxy-1-oxo-6-
phenyl-2-(phenylmethyl)hexyl)-L-leucyl-L-phenylalaninamide.
588

␥-Secretase Inhibitors Induce Neuronal Differentiation
589
transition state of aspartyl protease catalysis have been de- systematically altered at positions P2-P3Ј either with hydro-
rived from sequences around the ␥-cleavage site of APP and phobic L- or D-amino acids are shown to effectively block
are shown to block ␥-secretase activity, indicating that ␥-secretase activity (Bakshi and Wolfe, 2004; Esler et al.,
␥-secretase is an aspartyl protease (Wolfe et al., 1999a). Two 2004). These studies not only substantiate the loose sequence
highly conserved aspartate residues residing within the sixth specificity of ␥-secretase but also raise the possibility that
and seventh transmembrane domains of presenilins are re-
quired for ␥-secretase activity (Wolfe et al., 1999b), suggest-
ing that the active site of this novel aspartyl protease might
be located at the heterodimeric interface. A number of com-
pounds designed to interact with the protease active site
have been shown to bind directly to both PS subunits (Esler
et al., 2000; Li et al., 2000b). Together, these findings
strongly support the notion that PS heterodimers constitute
the active site of ␥-secretase.
Biochemical and genetic analyses further reveal that PSs
are part of a multimeric ␥-secretase complex whose constit-
uents include a heterodimeric PS, a mature glycosylated
nicastrin (NCT), Aph-1, and Pen-2 (Iwatsubo, 2004). Compel-
ling evidence has shown that the full spectrum of ␥-secretase
activity can be reconstituted by the coexpression of human
PS, nicastrin, Aph-1, and Pen-2 in yeast (Edbauer et al.,
2003), providing definitive proof for the minimal required
constituents of a functional ␥-secretase.
␥-secretase could be targeted by transition state analog in-
hibitors encompassing unnatural amino acids that can confer
peptidomimetics extraordinary biological stability along with
enhanced biological activity and proteolytic resistance.
Herein, we report the development of (hydroxyethyl)urea
peptidomimetics containing unnatural amino acid substitu-
tions that can effectively block ␥-secretase activity for lower-
ing A␤ production and attenuating Notch signaling so as to
promote the neuronal differentiation of neuroblastoma cells.
Our studies provide the proof-of-concept for the use of unnat-
ural amino acids as building blocks in the generation of
biologically stable ␥-secretase inhibitors for AD treatments
and the promotion of neuronal differentiation.
Materials and Methods
Reagents. The BCA protein assay reagent kit and SuperSignal
West Pico and SuperSignal West Dura reagents were purchased
from Pierce Chemical (Rockford, IL). The rabbit anti-Notch(Val1744)
antibody was from Cell Signaling Technology Inc. (Danvers, MA).
Horseradish peroxidase-conjugated anti-rabbit IgG was from Santa
Cruz Biotechnology, Inc. (Santa Cruz, CA). Horseradish peroxidase-
conjugated anti-mouse IgG and ECL Western Blotting detection
reagents were from GE Healthcare (Little Chalfont, Buckingham-
shire, UK). The FuGENE6 transfection reagent, Expand long tem-
The ␥-secretase not only mediates the proteolysis of APP
but also is critical for the processing of Notch receptor. The
Notch signaling pathway is essential for cell fate decisions
during development (Mumm and Kopan, 2000). This path-
way is initiated by the binding of the Delta-Serrate-Lag2
ligand family, followed by the shedding of extracellular do-
main (S2 cleavage) mediated by the ADAM family members.
The resultant C-terminal membrane-tethered fragment of plate polymerase chain reaction system, and polymerase chain reac-
tion nucleotide mix were from Roche Applied Science (Indianapolis,
IN). Dual luciferase assay reagents, Steady-Glo luciferase assay
reagents, and the pRL-TK vector were from Promega (Madison, WI).
Human A␤40 and A␤42 colorimetric ELISA kits were from Bio-
Source International (Camarillo, CA). All other reagents were of at
least reagent grade and were obtained from standard suppliers.
Synthesis of (Hydroxyethyl)ureas. The (hydroxyethyl)urea
peptidomimetics containing unnatural amino acid moieties were
synthesized using methods described previously with some modifi-
cations (Getman et al., 1993) (Scheme 1). In brief, a commercially
available Boc-protected amino acid derivative 1, which fits the de-
sired structural feature of the P1 residue, was reduced to ␣-amino-
aldehyde 2 via the Weinreb amide and then transferred to the
corresponding epoxide 4 through alkene 3. The epoxide was ring-
opened by treatment with benzylamine or cyclohexylmethylamine
under heated reflux to give the amino alcohols 7 and 8. These
intermediates were condensed with isocyanates 9a and 9b (in turn
obtained from ␣-amino methyl esters and phosgene) at room tem-
perature to provide the (hydroxyethyl)urea P1-P2Ј isosteres 10a,
10b, and 11. C-Terminal incorporation of P3Ј residues was accom-
plished by hydrolysis of the methyl ester with LiOH and subsequent
coupling of P3Ј residues 14aϳg with the HATU coupling reagent in
DMF to yield the desired (hydroxyethyl)urea isosteres (see Table 1 as
Jia compounds). In this synthesis scheme, the epoxide was treated
with a variety of alkylamines that came up with P1-P1Ј isosteres in
which the P1Ј site was modified with nitrogen at the backbone and
an unnatural cyclohexyl side chain that is not seen in essential
amino acids. Moreover, the selected amines and unnatural amino
acid derivatives 14aϳg were incorporated to the P1-P2Ј fragments to
yield entire P1-P3Ј (hydroxyethyl)ureas. Through this synthesis
scheme, a collection of (hydroxyethyl)urea peptidomimetics that con-
tained unnatural amino acid moieties mainly distributed either at
the P1Ј and/or P3Ј positions was generated. The characterization and
Notch, termed Notch extracellular membrane truncation
(NEXT), is then cleaved within the transmembrane domain
(S3 cleavage) by ␥-secretase to release the Notch intracellu-
lar domain (NICD). Once released from membrane, NICD
bounds to mammalian CBF1/RBPj, Drosophila melanogaster
Suppressor of hairless, and Caenorhabditis elegans Lag-1
DNA-binding proteins and converts them from transcrip-
tional repressors to activators, resulting in the expression of
Notch downstream target genes. The inhibition of ␥-secretase
would likewise block NICD production and reduce Notch
signaling (Wolfe et al., 1999b, 2002).
(Hydroxyethyl)urea peptidomimetics have recently been
identified as a new class of transition state analog inhibitors
that mimic the transition state of aspartyl protease catalysis
and block ␥-secretase much more efficiently than any of the
difluoro ketones or difluoro alcohols (Wolfe et al., 2002).
Highly potent and selective inhibitors containing hydroxy-
ethyl isostere have also been developed and used to block
such aspartyl proteases as renin and HIV protease (Greenlee,
1990; Huff, 1991). Five HIV protease inhibitors currently
available as approved treatments for AIDS all contain the
hydroxyethyl isostere (Flexner, 1998), and the core structure
of (hydroxyethyl)ureas is known to have apparent low toxic-
ity, a prerequisite for clinical applications. (Hydroxyethyl)-
urea peptidomimetics have been developed for the affinity
isolation and characterization of ␥-secretase, and for probing
the active site of this protease (Esler et al., 2004). These
compounds can be generated through systemic replacements
in five positions (P2, P1Ј, P2Ј, P3Ј, and P4Ј) with small,
medium, and large hydrophobic L-amino acids (Wolfe et al.,
2002). Furthermore, (hydroxyethyl)urea peptidomimetics the purity of each compound were confirmed by 1H NMR spectros-

590
Liao et al.
copy, high-resolution mass spectroscopy, and high-pressure liquid ics were included in the medium as specified. Cells incubated with
chromatography. the culture medium containing 1 ␮g/ml tetracycline and 1% DMSO
Cell Culture and Cell Lines. Human embryonic kidney were used to define the basal level of ␥-secretase activity, whereas
(HEK)293 cells were maintained in Dulbecco’s modified Eagle’s me- cells treated with DMSO-containing medium without tetracycline
dium (DMEM) supplemented with 10% fetal bovine serum (FBS) and were used to estimate the nonspecific background emission of the
0.1 mg/ml penicillin and streptomycin. T-REx293 cells were pur- luciferase signal. To terminate the reactions, cells were harvested
chased from Invitrogen and cultured in DMEM supplemented with with PBS containing 20 mM EDTA and lysed in 100 ␮l of 1ϫ passive
10% FBS and 5 ␮g/ml blasticidin. The generation of stably trans- lysis buffer (PLB; Promega). Cell debris was removed by centrifuga-
fected cell lines, T16, T20, N7, and ␥-30 has been described previ- tion at 13,200g for 5 min, and luciferase activity in clarified lysates
ously (Kimberly et al., 2003; Liao et al., 2004; Bakshi et al., 2005).
Cells of the SH-SY5Y human neuroblastoma cell line were grown in
was determined by mixing 20 ␮l of lysates and 20 ␮l of Steady-Glo
luciferase assay reagent in a 96-well LumiNunc microplate (Nalge
DMEM/Ham’s F-12 supplemented with 10% FBS. Cells were incu- Nunc, International, Rochester, NY). After incubation at room tem-
bated in a humidified incubator at 37°C in 5% CO2. perature for 5 min, emitted luminescence in individual microwells
Cell-Based ␥-Secretase Assays. The generation of these stable was determined by an MLX Microplate luminometer (Dynex Tech-
cell lines has been reported previously (Liao et al., 2004). To examine nologies, Chantilly, VA) and subsequently normalized by the protein
the inhibitory effects of the compounds specifically on ␥-secretase, content of the lysates. The protein content of clarified lysates was
T16 or T20 cells were trypsinized and washed with serum-free determined using the BCA protein assay kit (Pierce Chemical) fol-
DMEM before plating onto 12-well microplates in 1 ml/well DMEM lowing the manufacturer’s instructions. The normalized luciferase
supplemented with 10% FBS at 5 ϫ 105 cells/well. After incubation signal emitted by T20 cells in regular culture medium without tet-
at 37°C overnight, cells were treated with 10 ␮M concentrations of racycline was referred as 1-fold of activation. Known ␥-secretase
individual (hydroxyethyl)urea peptidomimetics in culture medium inhibitors, such as compound E and DAPT (Seiffert et al., 2000;
containing 1% DMSO and 1 ␮g/ml tetracycline, the inducer of Dovey et al., 2001), were included as positive controls of ␥-secretase
APP695-Gal4/VP16 and the 99-residue C-terminal fragment of APP- inhibition.
Gal4/VP16 (C99-GV) expressions, and they were incubated at 37°C
To examine compound inhibition of ␥-secretase-dependent S3
for 24 h or various intervals as specified. For the dose-response cleavage of Notch, we generated a HEK293-derived stable line (N7)
studies, various concentrations of (hydroxyethyl)urea peptidomimet- that was constitutively expressing N⌬E (Liao et al., 2004). The N⌬E
Scheme 1. Reagents and conditions: i, EDC, HOBt,
HNMe(OMe) ⅐ HCl, then LAH. ii, MePPh3Br, 35%
KH, HMDS. iii, m-CPBA. iv, isopropanol. v, room
temperature, overnight. vi, LiOH. vii, HATU, DIEA.

␥-Secretase Inhibitors Induce Neuronal Differentiation
591
was a Notch mutant protein lacking its extracellular domain but
␤-Amyloid ELISA. To determine the inhibitory effects of com-
retaining its membrane-spanning region. The recombinant N⌬E was pounds on A␤ production, T20 or ␥-30 cells (5 ϫ 105 cells/well) were
thus expressed as a membrane-tethered protein that can be cleaved seeded onto 12-well tissue culture plates and treated with various
directly by ␥-secretase independently of its ligand activation and amounts of compounds in serum-free DMEM with (for T20) or with-
that served as an alternative substrate for the measurement of
out (for ␥-30) 1 ␮g/ml tetracycline, followed by incubation at 37°C for
␥-secretase activity (Kopan et al., 1996). N7 cells were plated onto 24 h. Conditioned media were harvested, clarified by centrifugation,
12-well microplates in 1 ml/well DMEM supplemented with 10% supplemented with the Complete protease inhibitor cocktail, and
FBS at 5 ϫ 105 cells/well. After incubation at 37°C overnight, cells stored at Ϫ80°C until ready for the assay. Levels of secreted A␤40
were treated with compounds at 10 ␮M as described above and and A␤42 in conditioned media were determined using quantitative
incubated at 37°C for 24 h. Treated cells were harvested using PBS human A␤40 and A␤42 sandwich ELISA kits (BioSource Interna-
containing 20 mM EDTA and dissolved in 50 ␮l of 1ϫ PLB, followed tional) as described in the manufacturer’s instructions. Contents of
by centrifugation at 13,200g for 5 min to remove cell debris. The A␤40 and A␤42 in the conditioned media of T20 and ␥-30 cells
protein concentrations of clarified supernatants were determined treated with regular culture medium alone were defined as the blank
using a BCA protein assay reagent kit, and cell extracts containing for the quantitation of A␤ production.
equivalent amounts of proteins were resolved by SDS-PAGE and
analyzed by Western blotting using an anti-Notch(Val1744) poly- blastoma Cells. SH-SY5Y human neuroblastoma cells were seeded
clonal antibody as described below.
onto six-well microplates (5 ϫ 105 cells/well) and incubated at 37°C
Compound-Induced Neuronal Differentiation of Neuro-
Cell Viability Assay. T20 cells (5 ϫ 104/100 ␮l/well) were seeded overnight. Adherent SH-SY5Y cells were then treated with 10 ␮M
onto the wells of 96-well microplates in culture medium containing respective Jia compounds or vehicle alone (0.1% DMSO) in DMEM/
respective compounds at 10 ␮M and incubated at 37°C for 24 h. Ham’s F-12 medium containing 10% FBS, followed by incubation at
Viable cells were determined using the CellTiter 96 AQueous non- 37°C for 24 h or various intervals as specified. Compound-treated
radioactive cell proliferation assay (Promega) as specified by the SH-SY5Y cells were harvested and lysed by PLB (Promega) contain-
manufacturer’s instructions. In brief, after the addition of the com- ing the Complete protease inhibitor cocktail. Clarified lysates con-
bined 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4- taining equivalent amounts of proteins were resolved by SDS-PAGE.
sulfophenyl)-2H-tetrazolium, inner salt/phenazine methosulfate so- The expression levels of calreticulin and GAP-43, two markers for
lution (20 ␮l/well), microplates were incubated for 3 h at 37°C. The the neuronal differentiation of neuroblastoma cells (Grynfeld et al.,
conversion of 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphe- 2000; Hsu et al., 2005), were visualized by Western blotting using
nyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt into formazan in specific antibodies. The level of GAPDH was also determined as a
viable cells was quantitated by the absorbance at 490 nm using a
Synergy HT ELISA plate reader (Bio-Tek Instruments, Winooski,
protein load control.
To quantify neurite outgrowth of neuroblastoma cells induced by
VT). The number of living cells in culture was directly proportional to ␥-secretase inhibitors, 50 or more SH-SY5Y cells whose dendritic
the absorbance at 490 nm. Viable cells in culture medium containing trees were relatively isolated and did not have discontinuities in
vehicle alone (1% DMSO, control) were referred to as 100% viability. their dendritic trees were chosen for quantification of neurite length.
The background absorbance shown at 0 cells/well was subtracted Image for Windows was the morphometric program used to measure
from these data.
the neurite length.
TABLE 1
Structures of (hydroxyethyl)urea isosteres and their IC50 values for the inhibition of ␥ -secretase
IC50 values for luciferase were determined by the cell-based reporter gene assay for ␥ -secretase; those for A␤40 were determined by the quantitative A␤40 ELISA kit
(BioSource International); and those for NICD were determined by the generation of NICD. DAPT and compound E, two previously identified ␥-secretase inhibitors, were
included as positive controls
IC50
Compound
P1Ј
AA1 (-HN-CH(P2Ј)-CO-)
-HN-P3Ј
Luciferase
A␤40
␮M
NICD
Jia040
Phenyl
Ala
(1R,2S)-AI
N.D.
0.946
5.459
N.D.
3.159
4.513
3.149
0.172
0.991
4.863
0.478
0.013
N.D.
0.704
2.566
N.D.
1.651
2.462
2.485
0.074
0.875
1.177
0.794
0.007
N.D.
3.965
11.188
N.D.
2.018
2.520
4.117
0.235
1.073
3.192
0.039
0.001
Jia046
Phenyl
Leu
Leu
Leu
Leu
Leu
Leu
Leu
Leu
Leu
(L)-Phg-OMe
Jia047
Phenyl
(1R,2R,3R,5S)-IPCA
(L)-Val-OMe
(D)-Phg-OMe
(S)-(Ϫ)-1-PEA
(R)-(ϩ)-1-PEA
(L)-Phg-OMe
Jia097a
Jia101
Phenyl
Phenyl
Jia104
Phenyl
Jia105
Phenyl
Jia138
Cyclohexyl
Cyclohexyl
Cyclohexyl
Jia142
(R)-(ϩ)-1-PEA
(L)-Val-OMe
Jia143
DAPTb
Compound Eb
N.D., not determined; AI, 1-amino-2-indanol; IPCA, isopinocampheylamine; Phg-OMe, phenylglycine methyl ester; 1-PEA: 1-phenylethylamine
a Jia097 was the known inhibitor III-31-C.

592
Liao et al.
SDS-Polyacrylamide Gel Electrophoresis and Western Blot nonspecific cellular events caused by the introduction of the siRNAs
Analysis. Clarified cell extracts containing equivalent amounts of into cells. The levels of calreticulin and NICD in transfected cells
proteins were mixed with equal volumes of 2ϫ SDS sample buffer were examined by Western blotting using anti-Notch1(Val1744) as
(125 mM Tris-HCl, pH 6.8, 4% SDS, 20% glycerol, and 5% ␤-mercap- described above. The efficiency of the Notch1-targeting RNAi knock-
toethanol) and boiled at 100°C for 5 min. Denatured proteins were down was determined by visualizing the endogenous full-length
resolved in Tris-glycine polyacrylamide gels (10 or 12%). Separated Notch1 (ϳ300 kDa) and NICD (ϳ110 kDa) in SH-SY5Y and N7 cells,
proteins were transferred electrophoretically to polyvinylidene diflu- whereas the production of exogenous N⌬E-derived NICD (ϳ50 kDa)
in N7 cells was not affected by RNAi.
oride membranes (Immobilon-P; Millipore Corporation, Billerica,
MA). Membranes were then treated with blocking buffer [5% nonfat
dry milk and 0.1% Tween 20 in PBS (PBST)] at room temperature for
1 h, followed by a brief rinse with PBST. Blocked membranes were
probed with appropriate dilutions of primary antibody in PBST at
room temperature for 1 h. The unbound primary antibody was re-
moved by extensive washes with PBST. Thereafter, washed mem-
branes were incubated with appropriate horseradish peroxidase
(HRP)-conjugated secondary antibody in PBST at room temperature
for 1 h. After extensive washes with PBST, antibody-reacted proteins
were visualized by chemiluminescence using SuperSignal West Dura
and Pico reagents (Pierce Chemical). The antibodies used and their
dilutions were as follows: anti-Notch(Val1744), 1:1000; HRP-conju-
gated anti-mouse IgG (GE Healthcare), 1:10,000; and HRP-conju-
gated anti-rabbit IgG (Santa Cruz Biotechnology, Inc.), 1:1000.
Preparation and Detergent Solubilization of Microsome
Membrane. Freshly harvested ␥-30 cells that overexpressed all four
␥-secretase components (Kimberly et al., 2003) were resuspended
and homogenized in a HEPES buffer (50 mM HEPES, 5 mM MgCl2,
5 mM CaCl2, and 150 mM NaCl). After removal of cell debris and
nuclei, the supernatant solution was centrifuged at 100,000g for 60
min. The microsome-enriched pellets were solubilized in a CHAPSO-
containing buffer (50 mM HEPES, 5 mM MgCl2, 5 mM CaCl2, and
1% CHAPSO) at 4°C for 60 min. The ensuing supernatant solution
was defined as solubilized ␥-secretase.
Mass Spectrometry Analysis. Purified peptidomimetics DAPT
and Jia142 were dissolved in 50% acetonitrile and 0.1% formic acid
to a final concentration of 1 mM and incubated at various tempera-
tures for 2 days. Treated compounds were subjected to one-dimen-
sional LC-nano-ESI-MS/MS analysis performed on an integrated
nano-LC-MS/MS system (Micromass, a division of Waters, Bedford,
MA) composed of a three-pumping Micromass/Waters CapLC system
with an autosampler, a stream select module configured for precol-
umn plus analytical capillary column, and a Micromass Q-Tof Ul-
tima API mass spectrometer fitted with nano-LC sprayer, operated
under MassLynx 4.0 control. Injected samples were first trapped and
desalted isocratically on an LC-Packings PepMap C18 ␮-Precolumn
Cartridge (5 ␮m, 300 ␮m i.d. ϫ 5 mm; Dionex, Sunnyvale, CA) for 2
min with 0.1% formic acid delivered by the auxillary pump at 30
␮l/min after which the peptides were eluted off from the precolumn
and separated on an analytical C18 capillary column (15 cm ϫ 75 ␮m
i.d., packed with 5 ␮m, Zorbax 300 SB C18 particles; Micro-Tech
Scientific, Vista, CA) connected inline to the mass spectrometer, at
300 nl/min using a 40-min fast gradient of 5 to 80% acetonitrile in
0.1% formic acid. Online nano-ESI-MS survey scan was fully auto-
mated and synchronized with the nano-LC runs under the full soft-
ware control of MassLynx 4.0. Before online analysis, the nano-LC
sprayer and Z-spray source parameters were tuned and optimized
with glufibrinopeptide B.
Preparation of Jia138-Conjugated Affinity Resins and Af-
finity Chromatography of ␥-Secretase. The Jia138-conjugated
affinity resins were prepared by a previously described protocol
(Esler et al., 2002). In brief, the methyl esters of Jia138 were hydro-
lyzed by using LiOH in aqueous dioxane. The carboxylic acids were
coupled in DMSO by using 1-ethyl-3-(3-dimethylaminopropyl)carbo-
diimide (3 equivalents) to the primary amine of a six-atom hydro-
philic linker present on the agarose resin Affi-Gel 102 (Bio-Rad
Laboratories, Hercules, CA). The conjugation mixtures were incu-
bated at room temperature overnight with continuous gentle inver-
sion, followed by extensive washes with DMSO. The affinity resins
were subsequently maintained in aqueous buffer.
The solubilized ␥-secretase was incubated with Jia138-conjugated
resins for 2 h at room temperature with gentle rocking, and the resin
was washed twice with a CHAPSO-containing buffer (50 mM
HEPES, pH 7.0, 150 mM NaCl, and 1% CHAPSO). The bound pro-
teins were then released by SDS-PAGE sample buffer and analyzed
by SDS-PAGE and Western blotting as specified above.
Results
Synthesis of (Hydroxyethyl)urea Peptidomimetics
Containing Unnatural Amino Acids. The incorporation of
transition-state mimicking moieties into the substrate-based
substructure has recently been used to develop aspartyl pro-
tease inhibitors (e.g., HIV protease inhibitors; Huff, 1991).
This approach using substrate-based modifications has also
been successfully applied to the generation of potent and
selective ␥-secretase inhibitors and is efficient for probing the
active site of this aspartyl protease. Previous studies showed
that the hydroxyethylene peptidomimetic L-685,458 and the
(hydroxyethyl)urea III-31-C can both inhibit A␤ production
with IC50 values of approximately 10 nM in the cell-free
assays and 200 nM in cell-based systems (Shearman et al.,
2000; Kornilova et al., 2003). It is intriguing that the P1-P1Ј
residues of both inhibitors consisted of Phe-Phe isosteres,
whereas APP substrate-based hydroxyethylene analogs with
Ala-Thr and Val-Ile isosteres showed no inhibitory effects on
the ␥-secretase-mediated formation of A␤42 (Nadin et al.,
2003). These results suggested that inhibitors such as
Transfection of Small Interfering RNAs Targeting Notch1.
The chemically synthesized siRNA duplex oligoribonucleotides
against human Notch1 (siNotch1, accession no. NM_017617) and a
nonspecific scrambled siRNA were purchased from Ambion (Austin,
TX). The target sequences for siNotch1 were 5Ј-CGA CGC AUG CAU
CAG CAA C-3Ј (sense strand) and 5Ј-GUU GCU GAU GCA UGC
GUC G-3Ј (antisense strand). siRNA oligoribonucleotides (50 or 100 L-685,458 and III-31-C were functioning as direct transition
pmol) were transiently transfected into SH-SY5Y or N7 cells using
the Lipofectamine 2000 transfection reagent (Invitrogen) according
to the manufacturer’s instructions. After transfection with siRNAs in
DMEM supplemented with 10% FBS at 37°C for 6 h, transfected cells
were incubated with fresh culture medium at 37°C for an additional
24 h or as specified. Transfected cells were harvested by PBS con-
taining 20 mM EDTA, and clarified lysates were prepared by using
1ϫ PLB containing Complete protease inhibitor cocktail. Protein
contents of cell lysates were determined with a BCA protein assay
reagent kit. A nonspecific oligoribonucleotide that is not homologous
state analogs of the APP ␥-cleavage site and that the Phe-Phe
isosteres at positions P1-P1Ј are critical for their binding and
inhibition of ␥-secretase. We thus sought to determine
whether (hydroxyethyl)ureas isosteres with modifications es-
pecially at positions P1Ј-P3Ј could allow us to delineate the
structure-activity relationships of these (hydroxyethyl)ureas
to the binding pocket of the ␥-secretase active site.
To incorporate unnatural amino acids into newly synthe-
sized (hydroxyethyl)urea peptidomimetics, we introduced a
to any known genes was used as a negative control to rule out cyclohexyl residue into the P1Ј position. The treatment of

␥-Secretase Inhibitors Induce Neuronal Differentiation
593
epoxide 4 with a variety of alkylamines altered the P1-P1Ј
Three additional unnatural amino acid-containing deriva-
isosteres in which the P1Ј site was modified with nitrogen at tives of (hydroxyethyl)urea isosteres (Jia138, Jia142, and
the backbone and unnatural side chains that are not seen in Jia143), whose P1Ј sites were occupied by a cyclohexyl group
essential amino acids (Scheme 1). Alternatively, the selected instead of a phenyl group as in previous Jia compounds, were
amines and unnatural phenylglycine analogs were included subsequently generated and analyzed for their inhibition of
in the P3Ј position of certain (hydroxyethyl)ureas. Therefore, ␥-secretase. We found that the replacement of a phenyl group
two series of (hydroxyethyl)urea peptidomimetics were gen- at the P1Ј position in Jia046 with a cyclohexyl group in
erated. One series of (hydroxyethyl)urea isosteres, including Jia138 further improved the efficacy of the inhibition of
Jia040, Jia046, Jia047, Jia097, and Jia101, carried a phenyl ␥-secretase in a dose-dependent manner (Fig. 3; Table 1).
group at their P1Ј site, whereas the other series, including Nevertheless, the similar replacement in Jia097 and Jia105
Jia138 and Jia143, carried a cyclohexyl group. Subsequently, with a cyclohexyl group in Jia143 and Jia142, respectively,
more lipophilic analogs were prepared by the introduction of also increased their inhibitory potencies. Our data suggested
decarboxylated phenylglycine residues [e.g., (ϩ)- and (Ϫ)-1- that the marginally bulky moiety occupying the P1Ј site of
phenylethylamines] into the P3Ј position (Jia 104, Jia105, these (hydroxyethyl)urea isosteres plays a critical role in
and Jia142) to optimize the relative activities and probe the determining the efficacy of ␥-secretase inhibition by these
stereochemical diversity of the binding sites. The effects of compounds.
these synthesized compounds after purification on the inhi-
bition of ␥-secretase activity were examined using the spe-
cific cell-based ␥-secretase assays as described.
Inhibition of ␥-Secretase-Mediated Cleavage of APP
by (Hydroxyethyl)urea Peptidomimetics Containing
Unnatural Amino Acid Moieties. To determine the effects
of (hydroxyethyl)urea peptidomimetics containing unnatural
amino acid moieties on the inhibition of ␥-secretase-mediated
cleavage of APP, two highly efficient and quantitative cell-
based assays, one assay specifically measuring the ␤/␥-cleav-
ages of APP (AP-GL-T16) as a whole and the other assay the
␥-cleavage of C99 (C99-GL-T20) alone, were used. The gen-
eration and validation of both cell-based assays were de-
scribed previously (Liao et al., 2004; Bakshi et al., 2005). The
effects of a series of (hydroxyethyl)urea pipeptidomimetics
containing unnatural amino acid moieties (Jia compounds)
are shown in Fig. 1. These Jia compounds were structurally
related to two well characterized nontransition state analog
inhibitors of ␥-secretase, compound E and DAPT (Seiffert et
al., 2000; Dovey et al., 2001). Initial screening of these Jia
compounds showed that compounds Jia046, Jia047, Jia101,
Jia104, and Jia105 at 10 ␮M exhibited potent inhibition of
both APP proteolysis and ␥-secretase, comparable with the
levels displayed by compound E and DAPT (Fig. 1, A and B).
The comparable potency of these effective Jia compounds to
those of compound E and DAPT suggested that these Jia
compounds could directly target ␥-secretase. Most of these
Jia compounds, with only the exception of Jia105, presented
no significant cellular toxicity (Fig. 1C). All five effective Jia effects of Jia compounds on the ␥-cleavage of APP using C99-GL-T20
compounds also inhibited ␥-secretase activity in dose-depen-
Fig. 1. Screening of (hydroxyethyl)urea peptidomimetics containing un-
natural amino acids (Jia compounds) on the inhibition of ␥-secretase
activity. A, effects of Jia compounds on the secretase-mediated proteoly-
sis of APP using AP-GL-T16 (T16) cells that were stably cotransfected
with APP-GV and Gal4-driven luciferase reporter gene (Gal4-Luc). B,
(T20) cells that were stably cotransfected with the membrane-tethered
C99 C-terminally tagged with C99-GV and Gal4-Luc. Cells (5 ϫ 105
dent manners (Fig. 2A). The production of secreted A␤40 in
cells/well in 12-well microplates) were treated with DMEM containing
conditioned media of compound-treated cells was consis-
tently reduced in a similar manner (Fig. 2B). The approxi-
mate IC50 values of active Jia compounds obtained from our
cell-based ␥-secretase assay and quantitative A␤40 ELISA
are listed in Table 1. Our data suggested that these active
(hydroxyethyl)urea isosteres containing unnatural amino
acid moieties could target the ␥-secretase-mediated cleavage
of APP and effectively inhibit A␤ production. These inhibi-
tory effects were not due to nonspecific inhibition of lucif-
erase as demonstrated by the relatively unaffected luciferase
signal upon treatments with respective Jia compound in a
control cell line constitutively expressing the luciferase re-
porter gene (data not shown), substantiating the observed
inhibition of ␥-secretase activity by these active Jia com-
pounds.
10% FBS and 1 ␮g/ml tetracycline in the presence or absence of individ-
ual (hydroxyethyl)urea isosteres (10 ␮M Jia compounds) at 37°C for 24 h.
Cells in the control experiments were treated with vehicle alone (1%
DMSO). Treatments with compound E and DAPT, two potent ␥-secretase
inhibitors, were included as positive controls. ␥-Secretase activity in
clarified lysates was determined as described under Materials and Meth-
ods. Background luminescence emitted by T16 or T20 cells treated with
culture medium in the absence of tetracycline is referred to as 1-fold of
activation. C, viability of host cells in the presence of Jia compounds. T20
cells (5 ϫ 104/100 ␮l/well) were seeded onto the wells of 96-well micro-
plates in culture medium containing 10 ␮M respective compounds or 1%
DMSO alone (Control), and they were incubated at 37°C for 24 h. Viable
cells were determined by the CellTiter 96 AQueous nonradioactive cell
proliferation assay kit as described by the manufacturer. Viable cells in
culture medium without compounds (Control) are referred to as 100%
viability. Two known ␥-secretase inhibitors, compound E and DAPT, were
included in these experiments for comparison. Results from a represen-
tative experiment are expressed as the mean Ϯ S.D. of triplicate mea-
surements.

594
Liao et al.
It is noteworthy that four active Jia compounds (Jia047, Jia101, Jia104, Jia105, Jia138, Jia142, and Jia143) were
Jia101, Jia104, and Jia105) at lower concentrations (1 ␮M) shown to significantly block ␥-secretase-mediated S3 cleav-
induced an approximate 3-fold rise in the level of A␤42 pro- age of N⌬E in N7 cells, resulting in complete ablation of
duction (Fig. 4), but they inhibited A␤42 generation at 10 NICD production (Fig. 5A). We found that these effective Jia
␮M, whereas there was no significant elevation in A␤40 compounds consistently exhibited dose-dependent inhibition
production by any of the active Jia compounds at the concen- of ␥-secretase-catalyzed S3 cleavage of N⌬E (Fig. 5B). Our
trations examined (Fig. 2). Similar effects of active Jia com- findings demonstrated that the efficacies of these active Jia
pounds were seen when either T20 cells overexpressing C99 compounds as shown by their IC50 values for blocking the
or ␥-30 cells stably transfected with APPswe, PS1, Aph-1, ␥-secretase-mediated S3 cleavage of N⌬E were comparable
and Pen-2 were treated with these compounds. The elevation with those for the ␥-cleavage of APP (Table 1). The present
of A␤42 at subinhibitory concentrations was previously re- data suggested that these novel (hydroxyethyl)urea isosteres
ported for a set of difluoro ketone peptidomimetic ␥-secretase containing unnatural amino acid moieties can target ␥-secre-
inhibitors (Wolfe et al., 1999a). How these active Jia com- tase independent of substrate selection.
pounds can differentially affect ␥-secretase-dependent pro-
duction of A␤40 and A␤42 remains to be defined.
In Vitro Stability of (Hydroxyethyl)urea Peptidomi-
metics Is Dramatically Enhanced by the Replacement
␥-Secretase-Mediated S3 Cleavage of Notch Is of Phenylethylamine at P3؅ Site. To examine whether the
Blocked by Effective Jia Compounds. We next examined unnatural amino acid substitution in Jia compounds can
whether these ␥-secretase-inhibiting Jia compounds also af- prolong the half-life of these peptidomimetics in culture con-
fect the ␥-secretase-mediated S3 cleavage of Notch. To do so,
a stable cell line (N7) constitutively expressing N⌬E that
acted as the direct substrate of ␥-secretase-catalyzed prote-
olysis independent of ligand activation was treated with var-
ious amounts of respective Jia compounds. The efficiency of
the N7 stable cell line as an alternative cell-based ␥-secretase
assay was described previously (Liao et al., 2004). By using
Western blotting analyses with the anti-Notch(Val1744) an-
tibody, these effective Jia compounds (Jia046, Jia047,
Fig. 3. Inhibition of ␥-secretase by (hydroxyethyl)urea peptidomimetics
containing a cyclohexylmethyl moiety at the P1Ј site. A, T20 cells (5 ϫ 104
cells/well) in 96-well microplates were treated with 10 ␮M respective Jia
compounds (Jia138, Jia142, or Jia143) in DMEM containing 10% FBS
and 1 ␮g/ml tetracycline for 24 h at 37°C. B, T20 cells were treated with
various amounts of Jia138, Jia142, and Jia143 for 24 h at 37°C to
determine the dose-dependent effects on the inhibition of ␥-secretase.
␥-Secretase activities in compound-treated cells were determined by the
Steady-Glo luciferase reporter gene assay as specified by the manufac-
turer. C, secreted A␤40 in the conditioned media of compound-treated
cells was quantitated by an A␤40 sandwich ELISA kit. Basal levels of
␥-secretase activity and A␤40 production in the presence of vehicle alone
(1% DMSO; Control) were also determined. Treatments of DAPT were
included for the comparison of inhibitory potency. Data are shown as the
mean Ϯ S.D. of triplicate measurements from a representative experi-
ment.
Fig. 2. Dose response of Jia compounds on the inhibition of ␥-secretase
activity and A␤ production. T20 cells (5 ϫ 104 cells/well) in 96-well
microplates were treated with various concentrations of respective Jia
compounds in DMEM containing 10% FBS and 1 ␮g/ml tetracycline for
24 h at 37°C. Clarified cell lysates and conditioned media were subjected
to the luciferase reporter gene assay for ␥-secretase (A) and the quanti-
tation of secreted A␤40 (B), respectively. Basal levels of ␥-secretase
activity and A␤40 production in the presence of vehicle alone (1% DMSO;
Control) were determined. Treatments of DAPT were included for the
comparison of inhibitory potency. Data are shown as the mean Ϯ S.D. of
triplicate measurements from a representative experiment.

␥-Secretase Inhibitors Induce Neuronal Differentiation
595
dition, Jia compounds (Jia046, Jia097, Jia138, and Jia142) or 50°C for 2 days (Fig. 6D). Our data revealed that the
were added into culture media of T20 cells whose ␥-secretase abundance of intact Jia142 after incubation at 37 or 50°C for
activity was then monitored daily for a span of 8 days. 2 days is reduced to 86 or 33%, whereas only 66 or 8% of
Whereas Jia046 and Jia138 exhibited an approximate half- DAPT are retained after respective treatments (Fig. 6D, bot-
life of 6 to 8 days that is comparable with the stability of tom). Together, our findings clearly demonstrated that the
DAPT and compound E, the potency of Jia142 can be fully substitution of a cyclohexyl moiety at P1Ј site and a phenyl-
sustained for at least 8 days (Fig. 6A). Using UV-spectrom- ethylamine at P3Ј site could dramatically enhance the in
etry, we found that both Jia142 and DAPT have a character- vitro stability of (hydroxyethyl)urea peptidomimetics.
istic absorbance peak at 244 nm (Fig. 6B). We reasoned that
The Affinity Precipitation of ␥-Secretase by Jia138.
the peak absorbance at 244 nm (OD244) could be a biophysical To establish the mode of action of active Jia compounds in the
property of both compounds and that alteration in OD244 of inhibition of ␥-secretase, we examined whether Jia138 can
these peptidomimetics could be indicative of their conforma- directly target to ␥-secretase. Jia138 was covalently linked to
tional integrity and chemical stability. Our data demon- the free primary amine of a six-atom hydrophilic spacer on an
strated that the incubations at room temperature for 4 and agarose affinity resin (Affi-Gel 102; Bio-Rad). Immobilized
24 h both result in a 25% reduction in the OD244 of Jia142, Jia138 was then used to precipitate PS1 and nicastrin from a
whereas the OD244 of DAPT significantly decreases 50 and CHAPSO-solubilized microsome preparation of ␥-30 cells.
70% after incubations at room temperature for 4 and 24 h, Solubilized ␥-secretase complexes were incubated with the
respectively (Fig. 6C). Using mass spectrometry (MS) analy- Jia138-conjugated affinity resin in batch format for 2 h at
ses, we further confirmed that Jia142 (637.4 Da) is thermally room temperature. Protein-bound resins were washed with a
more stable than DAPT (433.2 Da). The MS spectra of un- CHAPSO-containing buffer, and affinity-precipitated pro-
treated Jia142 (Jia142-Control) exhibited a corresponding
peak intensity of 2.2 ϫ 105 at 637.4 m/z that was diminished
to 2.0 ϫ 105 (Jia142–37°C) or 7.7 ϫ 104 (Jia142–50°C) after
incubation at 37°C or 50°C for 2 days, whereas those of
untreated DAPT (DAPT-Control) displayed a corresponding
peak intensity of 6 ϫ 104 at 433.2 m/z that is reduced to 4 ϫ
104 (DAPT-37°C) or 5000 (DAPT-50°C) after incubation at 37
Fig. 5. Inhibition of ␥-secretase-mediated S3 cleavage of Notch by the
novel (hydroxyethyl)urea peptidomimetics of Jia compounds. Screening
Fig. 4. Effects of active (hydroxyethyl)urea peptidomimetics on A␤42
production. Conditioned media of T20 (A) or ␥-30 (B) cells that were
treated with active Jia compounds for 24 h at 37°C were harvested, and
secreted A␤42 was determined by an A␤42 sandwich ELISA kit. Basal
levels of A␤42 production in the presence of vehicle alone (1% DMSO;
Cont.) were also determined. Treatments of DAPT were performed to
show the complete inhibition of A␤42 production in these cells. Data are
shown as the mean Ϯ S.D. of triplicate measurements from a represen-
tative experiment.
of Jia compounds (A) and dose-dependent response of Jia compounds (B)
for the inhibition of ␥-secretase-mediated S3 cleavage of Notch. HEK293
cells stably transfected with N⌬E (N7) in 12-well microplates (5 ϫ 105
cells/well) were treated with 10 ␮M (A) or various concentrations (B) of
individual compounds in DMEM containing 10% FBS for 24 h at 37°C.
Clarified lysates of treated cells containing equivalent amounts of pro-
teins were resolved by SDS-PAGE and analyzed by Western blotting
using anti-Notch(Val1744). Vehicle alone (1% DMSO) was included as
control. Treatment with DAPT was included for comparison.

596
Liao et al.
teins were analyzed by SDS-PAGE and Western blotting. We differentiation of neuroblastomas through the blocking of
found that PS1 and nicastrin can be effectively retained on Notch activation that is required for maintaining neuroblas-
the Jia138-immobilized matrix, but not on an unconjugated toma cells in an undifferentiated state. Recent evidence has
control matrix (Fig. 7A). Proteins isolated by unconjugated or suggested that the inhibition of Notch signaling leads to
Jia138-conjugated resin visualized by silver staining consis- maturation of neuroblastomas, whereas constitutively active
tently revealed that only Jia138-conjugated resin can pull Notch signaling can impede the induction of differentiation
down ␥-secretase components (Fig. 7B). Using various and neurite formation of neuroblastoma cells (Pahlman et
amounts of solubilized proteins, we further demonstrated al., 2004). We thus reasoned that these effective Jia com-
that ␥-secretase can bind to Jia138 resin in a dose-dependent pounds can block ␥-secretase-dependent Notch signaling and
manner (Fig. 7C). These results clearly suggested that promote neuronal differentiation of neuroblastoma cells. The
Jia138, and probably other active Jia compounds as well, can possible enhancement of neuroblastoma differentiation
exert a specific inhibitor-active site interaction that is re- through the inhibition of ␥-secretase was first explored by
quired for the binding of PS1 and nicastrin. Our findings also treating a human neuroblastoma cell line (SH-SY5Y) with
confirmed that the formation of heterodimeric PS1 (PS1-NTF various ␥-secretase inhibitors. We found that, in 24 h, all
and PS1-CTF) is prerequisite for the constitution of func- seven active Jia compounds examined (Jia046, Jia101,
tional ␥-secretase.
Jia104, Jia105, Jia138, Jia142, and Jia143) could signifi-
␥-Secretase-Blocking (Hydroxyethyl)urea Peptido- cantly induce the expression of GAP-43 and calreticulin, two
mimetics Promote Neuronal Differentiation of Neuro- differentiation markers of neuroblastoma cells, in SH-SY5Y
blastoma Cells. We next sought to determine whether these cells, although to varying degrees (Fig. 8). GAP-43 is the
␥-secretase-inhibitory Jia compounds can promote neuronal most abundant neuron-specific protein in the growth cones
Fig. 6. Stability of unnatural amino acid-substituted (hydroxyethyl)urea peptidomimetics. A, after addition of 10 ␮M respective Jia compounds
(Jia046, Jia097, Jia138, and Jia142) into culture media, ␥-secretase activity of treated T20 cells was monitored daily for 8 days. ␥-Secretase activity
of vehicle-treated (1% DMSO) cells was determined and referred to as 100%. DAPT and compound E were also included for comparison. Data are
shown as the mean Ϯ S.D. of triplicate measurements from a representative experiment. B and C, Jia142 and DAPT (1 mM in DMSO) were incubated
at room temperature, and UV absorbance spectra of both compounds were obtained at 0, 4, and 24 h. The representative spectra of Jia142 and DAPT
at 0 h are shown, and a characteristic peak of absorbance at 244 nm (OD244, arrow) was identified for both compounds (B). The gradual decrease of
OD244 at various intervals was shown to demonstrate the chemical stability of newly synthesized (hydroxyethyl)urea peptidomimetics at room
temperature. The OD244 of Jia142 (solid bar) and DAPT (shaded bar) measured at 0 h is referred to as 100% relative stability (C). D, Jia142 and DAPT
(1 mM in 50% acetonitrile/0.1% formic acid) were incubated at 37 or 50°C for 2 days and subjected to one-dimensional LC-nano-ESI-MS/MS analysis.
The corresponding masses (m/z) of DAPT (433.20 Da) and Jia142 (637.42 Da) were identified in the MS spectra (arrow). The intensity of untreated
DAPT (DAPT-Control, shaded bar) and Jia142 (Jia142-Control, solid bar) is referred to as 100% relative thermal stability (bottom).

␥-Secretase Inhibitors Induce Neuronal Differentiation
597
and has been shown to be critical for neuronal differentiation expression of calreticulin in SH-SY5Y cells that were par-
(Mani et al., 2001; Singh et al., 2003). Calreticulin is essen- tially depleted of Notch1 by RNAi was consistently signifi-
tial for neurite formation during neuronal differentiation, cantly increased, but it was unchanged or slightly reduced in
and its expression is tightly associated with reduced malig- those that were transfected with a Notch1-targeting siRNA
nancy of neuroblastoma (Hsu et al., 2005). The compound- in conjunction with the expression of a constitutively active
induced neurite outgrowth of SH-SY5Y cells was readily N⌬E (Fig. 10A). This Notch1-specific siRNA targeted the
observed after 72 h of treatment (data not shown) and be- exon 7 of endogenous human Notch1 gene that encodes part
came prominent after 5 days (Fig. 9A). Quantitative analyses of its extracellular domain and efficiently down-regulated the
showed that the average neuritic length of compound-treated expression of endogenous Notch1 and its intracellular do-
SH-SY5Y cells is significantly increased, ranging from a 100 main, evidenced by the decreased levels of endogenous full-
to 200% increase (Fig. 9B). To validate the direct link of length Notch1 and NICD (Fig. 10B, top and top middle). The
␥-secretase-mediated cleavage of Notch to the phenotypic expression of the exogenous N⌬E that encoded only the
changes of neuroblastoma cells, we used an RNAi approach transmembrane and intracellular domains of mouse Notch1
to define whether specific down-regulation of endogenous (Kopan et al., 1996) was not affected by the transfection of
Notch signaling is sufficient to drive increased expression of
neuronal differentiation markers in neuroblastoma cells. The
Fig. 7. Affinity precipitation of ␥-secretase by Jia138. A, microsome
preparation of ␥-30 cells was solubilized by a CHAPSO-containing buffer
(50 mM HEPES, pH 7.0, 150 mM NaCl, and 1% CHAPSO). Solubilized
␥-secretase complexes were incubated with Jia138-conjugated (Jia138) or
unconjugated (DMSO) resin for 2 h at room temperature with gentle
rocking, and the resin was washed twice with the CHAPSO-containing
buffer. The bound proteins were then released by SDS-PAGE sample
buffer and analyzed by SDS-PAGE and Western blotting with anti-NCT,
and anti-PS1 (PS1-NTF and PS1-CTF). Each boxed area of the duplicate
blots was then visualized by anti-NCT, anti-PS1-NTF, and anti-PS1-
CTF, respectively. Arrow denotes the antibody-reactive proteins corre-
sponding to NCT, PS1-NTF, or PS1-CTF. Input, total solubilized pro-
teins; ppt, bound proteins; sup, unbound proteins. B, proteins eluted from
Jia138 resin or unconjugated (DMSO) resin were resolved by 12% Tris-
glycine SDS-PAGE and visualized by silver nitrate. Total solubilized
proteins (Input) were also included for comparison. Bands corresponding
to NCT (110 and 150 kDa), PS1-NTF (30 kDa), and PS1-CTF (20 kDa) are
indicated by arrows. C, various amounts of solubilized ␥-secretase com-
plexes (30, 60, and 120 ␮g) were subject to the precipitation by Jia138
resin. NCT and PS1-NTF that were immobilized onto to the resin (ppt)
and those that were remained in the supernatant (sup) were assessed.
Fig. 8. Effects of active (hydroxyethyl)urea peptidomimetics on the neu-
ronal differentiation of neuroblastoma cells. A, human neuroblastoma
SH-SY5Y cells were treated with 10 ␮M active Jia compounds at 37°C for
24 h. Clarified cell lysates were resolved by SDS-PAGE and analyzed by
Western blotting using anti-calreticulin or anti-GAP-43 for the evalua-
tion of induced neuronal differentiation of neuroblastoma cells. GAPDH
was visualized to serve as the load control. The levels of calreticulin,
GAP-43, and GAPDH were determined by densitometry. The relative
expression levels of calreticulin and GAP-43 normalized by those of
GAPDH were shown to demonstrate the neuronal differentiation of neu-
roblastoma cells in the presence of active Jia compounds (B and C).
Lysates generated from cells that were treated with 1 ␮M retinoic acid
(RA), 10 ␮M DAPT, or vehicle alone (0.1% DMSO; Control) were also
included for comparison. Data are shown as the mean Ϯ S.D. of triplicate
measurements from a representative experiment.

598
Liao et al.
Notch1 siRNA (Fig. 10B, bottom middle). Thus, even the moieties retain their potency toward the inhibition of ␥-secre-
endogenous Notch signaling was significantly down-regu-
lated by this Notch1-siRNA, the expression of exogenous
N⌬E can still render continuous activation of Notch signaling
in transfected SH-SY5Y cells. These results demonstrated
that Jia compound-induced neuronal differentiation of neu-
roblastoma cells was intimately associated with the inhibi-
tion of Notch signaling and that the expression of an exoge-
nous N⌬E can rescue the phenotype of undifferentiated
neuroblastoma cells. These data provided direct evidence
strongly suggesting that the attenuation of Notch signaling
by ␥-secretase inhibitors could effectively initiate the neuro-
nal differentiation of neuroblastoma cells. Together, our find-
ings not only identified a novel class of (hydroxyethyl)urea
isosteres that contain unnatural amino acid moieties and
exhibit potent inhibition of ␥-secretase activity but also pro-
vided the direct evidence that the inhibition of Notch signal-
ing by ␥-secretase inhibitors is sufficient to induce the differ-
entiation of neuroblastomas.
tase. This inhibition is verified by using cell lines that over-
express APP695, C99, or N⌬E as substrates of ␥-secretase-
mediated proteolysis. Although the active Jia compounds
might not possess as strong a potency as DAPT and com-
pound E do in terms of IC50 values, they exhibit comparable
levels of potency at submicromolar concentrations to other
reported active (hydroxyethyl)urea peptidomimetics in cell-
based systems (Bakshi and Wolfe, 2004; Esler et al., 2004).
Furthermore, the active (hydroxyethyl)urea peptidomimetics
reported here can efficiently promote the neuronal differen-
tiation of neuroblastoma cells, probably through the inhibi-
tion of Notch signaling. These findings substantiate the no-
tion that the incorporation of unnatural amino acid moieties,
such as phenylethylamine (e.g., Jia142), could stabilize syn-
thetic (hydroxyethyl)urea peptidomimetics without compro-
mising their biological effects and their mode of action. Our
results also suggest that the integration of unnatural pep-
tidomimetics can be a valid methodology for further drug
development.
Discussion
Active Jia compounds are selected for their abilities to
inhibit ␥-secretase-mediated processing of APP695-GV or
C99-GV. We have shown that most of the (hydroxyethyl)u-
reas with a leucyl residue at the P2Ј position (Jia046, Jia047,
Jia101, Jia104, and Jia105) display excellent potencies for
the inhibition of ␥-secretase-dependent processing of APP695
A␤, the principal constituent of senile plaques in the AD
brain, is derived by sequential proteolyses mediated by ␤-
and ␥-secretases. Thus, these proteases have been regarded
as the prime targets for the development of therapeutics
against AD. The present study provides evidence that (hy-
droxyethyl)urea isosteres containing unnatural amino acid and C99, as demonstrated by the reduction in ␥-secretase
Fig. 9. Neuronal phenotypic changes of neuroblastoma
cells induced by active (hydroxyethyl)urea peptidomimet-
ics. A, human neuroblastoma SH-SY5Y cells that were
treated with vehicle alone (0.1% DMSO), 1 ␮M RA, 10 ␮M
DAPT, or 10 ␮M concentrations of respective Jia com-
pounds were incubated at 37°C for 5 days. SH-SY5Y cells
treated with active Jia compounds exhibited significant
neurite outgrowth, exemplifying the neuronal differentia-
tion similar to RA-induced phenotypic changes. Phase-con-
trast images were taken by Sony DSC-W5 digital camera
and processed with Adobe Photoshop software (Adobe Sys-
tems, Mountain View, CA). Scale bar, 20 ␮m. B, neuritic
length of DMSO- and compound-treated SH-SY5Y cells
(n Ͼ 50 for each treatment) were measured and analyzed
by NIH Image as described under Materials and Methods.
Results are expressed as the average (ϮS.D.) neuritic
length and analyzed by Student’s t test. *, p Ͻ 0.001.

␥-Secretase Inhibitors Induce Neuronal Differentiation
599
activity down to the basal level. These results are in agree-
ment with previous reports, in which the S2Ј active site
pocket apparently accommodates moderate isobutyl sub-
stituents (Esler et al., 2004). In addition, alterations in the
P3Ј position reveal that Jia compounds substituted with an
unnatural phenylglycine residue exhibit better potencies
than the compounds containing a valine residue (Jia097).
Consistent with previous studies (Esler et al., 2004), our
results suggest that the S3Ј pocket presents relatively loose
specificity and can accommodate great variation in stereo-
chemistry as shown by the similar potencies between Jia104
and Jia105 or between Jia046 and Jia101. Furthermore,
removal of the carboxyl residue at position P3Ј (such as Jia
046 changed to Jia104 or Jia 101 changed to Jia105) does not
abrogate the inhibitory potency. Therefore, the reduction in
the molecular weight and improvement in the lipophilic prop-
erties of theses (hydroxyethyl)ureas could be optimized with-
out compromising their pharmaceutical profiles.
The modification of dipeptide isosteres at the P1-P1Ј posi-
tion has also been shown to be critical to ␥-secretase inhibi-
tion (Nadin et al., 2003). To test this idea, the P1Ј residues
(Phe) of Jia 097, Jia046, and Jia105 are thus replaced with
an isosteric cyclohexylmethyl (Chy) moiety to generate
Jia138, Jia142, and Jia143. We show that the Chy-substi-
tuted Jia138 exhibits significantly improved efficacy of inhib-
iting ␥-secretase, whereas Jia143 and Jia142 maintained
their inhibitory potencies. It is noteworthy that the Chy-
substituted Jia138 presents dramatic improvement on the
inhibition of A␤40 production over DAPT (an IC50 of 0.074 for
Jia138 versus 0.794 for DAPT), whereas its efficacy of inhib-
iting S3 cleavage of Notch fares less well than DAPT does
(with an IC50 of 0.235 versus 0.039). Both Jia138 and its
Phe-containing analog Jia046 consistently exhibit better po-
tency toward the processing of APP (IC50 values of 0.704 ␮M
for Jia046 and 0.074 ␮M for Jia138 in A␤40 production) than
that of Notch (IC50 values of 3.965 ␮M for Jia046 and 0.235
␮M for Jia138 in NICD production). Moreover, all of these
active Jia compounds have better potency toward the inhibi-
tion of A␤ production than that of NICD production, whereas
DAPT and compound E have better potency in blocking
NICD production. It is thus likely that the incorporation of
selective unnatural amino acids at the P1Ј and P3Ј sites of
(hydroxyethyl)urea peptidomimetics could have greater im-
pact in the ␥-secretase-mediated processing of APP than that
of Notch. The possibility exists that we can technically drive
the transition state analog inhibitors, based on the structural
features of Jia138, to preferentially block the A␤ production
without tampering with the physiological function of Notch
signaling and other ␥-secretase substrates. Together, the
combination of a Chy moiety at the P1Ј site and a phenylg-
lycine analog at the P3Ј position might offer the structural
Fig. 10. Expression of calreticulin in neuroblastoma cells in response to
the modulation of Notch signaling. A, human neuroblastoma SH-SY5Y
cells (5 ϫ 105/well in six-well microplates) were transfected with a siRNA
against Notch1 (siNotch1, 50 or 100 pmol) in the presence (striped bar) or
absence (shaded bar) of N⌬E (1 ␮g/well) using Lipofectamine 2000, fol-
lowed by an additional incubation with DMEM containing 10% FBS for at
37°C for 3 days. Clarified cell lysates were analyzed by SDS-PAGE and
Western blotting using anti-calreticulin for the evaluation of induced
neuronal differentiation of neuroblastoma cells. GAPDH was visualized
to serve as the load control. The levels of calreticulin and GAPDH were
determined by densitometry. The relative expression levels of calreticulin
normalized by those of GAPDH are shown to demonstrate the neuronal
differentiation of neuroblastoma cells by the inhibition of Notch signaling
(lanes 2 and 3) and the reduced neuronal differentiation by activated
Notch signaling (lanes 4 and 5). The relative expression of calreticulin in
cells transfected with a nonspecific siRNA and an empty vector (Mock) is
referred to as 100% (solid bar). Data are shown as the mean Ϯ S.D. of scaffold for the subsequent development of ␥-secretase inhib-
triplicate measurements from a representative experiment. B, down-
regulation of endogenous Notch signaling by a Notch1-targeting siRNA.
itors that selectively block only the pathogenic APP process-
ing.
Our findings also demonstrate that the inhibitory potency
of the Chy-containing Jia142, a C-terminal decarboxylated
analog of Jia046, can be fully sustained for at least 8 days
under cell culture conditions, suggesting that the stability of
(hydroxyethyl)urea peptidomimetics could be dramatically
improved by the substitution of distinctive unnatural amino
acid moiety. Furthermore, these unnatural amino acid-sub-
stituted (hydroxyethyl)urea peptidomimetics, as exemplified
N7 cells (5 ϫ 105/well in six-well microplates) were transfected with
siNotch1 (50 or 100 pmol) using Lipofectamine 2000, followed by an
additional incubation with DMEM containing 10% FBS for at 37°C for 2.5
days. Cells transfected with a nonspecific siRNA (Mock) were also in-
cluded for comparison. Equivalent amounts of protein in clarified cell
lysates were analyzed by SDS-PAGE and Western blotting using anti-
Notch1(Val1744). Full-length endogenous Notch1 (fl-N1, ϳ300 kDa, top)
and NICD generated from endogenous Notch1 (ϳ110 kDa, top middle) or
exogenous N⌬E (ϳ50 kDa, bottom middle) were then visualized. GAPDH
was visualized to serve as the protein load control (bottom). The data are
representative of three independent experiments.

600
Liao et al.
by Jia138, can inhibit ␥-secretase and bind directly to het- pounds on the maturation and differentiation of neuroblas-
erodimeric PSs, resembling the characteristics of previously toma cells.
reported (hydroxyethyl)urea peptidomimetics and hydroxy-
Taken together, we demonstrate the identification of a
ethylene transition state analogs (Li et al., 2000a,b; Esler et novel class of ␥-secretase inhibitors that contain unnatural
al., 2002). The present data are thus in accordance with the amino acid moieties as a scaffold structure using various
conception that ␥-secretase can tolerate unnatural D-amino cell-based ␥-secretase assays. These active (hydroxyethyl)u-
acids as well as natural L-amino acids in transition state rea peptidomimetics effectively reduce A␤ production and
analog inhibitors (Bakshi and Wolfe, 2004; Esler et al., 2004), Notch signaling, and they present beneficial effects on neu-
and they reveal that these unnatural amino acid-substitut- roblastomas beyond their capacity for blocking the A␤-cen-
ing (hydroxyethyl)urea peptidomimetics, like other transi- tered pathogenesis of AD. The present evidence further sug-
tion state analog inhibitors, could specifically target to the gests that the substitution of a peptidic moiety with selective
active site of ␥-secretase. Our findings also substantiate the unnatural amino acids will not compromise the potency for
model that the formation of heterodimeric PS1 (PS1-NTF the inhibition of ␥-secretase and could pave the way for the
and PS1-CTF) is prerequisite for the functional ␥-secretase. development of synthetic peptidomimetics of ␥-secretase in-
These effective Jia compounds significantly blocked the hibitors with great stability in biological systems. The novel
production of A␤40, but they dramatically augmented the structural features of (hydroxyethyl)urea peptidomimetics
production A␤42 at a subinhibitory concentration (1 ␮M). are thus useful molecular tools for further characterization of
This effect has previously been observed for various ␥-secre- this protease, and they could have great implications for the
tase inhibitors (Citron et al., 1996; Wolfe et al., 1999a). The development of anti-AD drugs and therapeutics for neuro-
mechanism underlying the increased production of A␤42 by blastomas.
some active Jia compounds at lower concentrations remains
unclear. Accumulating evidence has suggested that A␤42
production mostly occurs in intracellular compartments, in-
cluding endoplasmic reticulum and Golgi (Cook et al., 1997;
Hartmann et al., 1997; Wild-Bode et al., 1997). Given that
Jia047, Jia104, and Jia105 all cause dramatic increases in
the levels of A␤42 without affecting the levels of A␤40 at a 1
␮M concentration, our findings are in agreement with the
notion that selective inhibition of A␤40 formation can result
in more efficient production of A␤42 from the accumulated
C99 substrate in these intracellular localizations (Wolfe et
al., 1999a). Further development of (hydroxyethyl)urea isos-
teres using the structural scaffold identified in the present
study might allow us to selectively target the A␤42-generat-
ing ␥-secretase that is localized intracellularly without per-
turbing A␤40 production.
The ␥-secretase-dependent S3 cleavage of Notch is a pre-
requisite for its downstream signaling that is essential for a
variety of cell fate determination events during development
and in adults (Mumm and Kopan, 2000). In addition, the
protransforming role of Notch signaling has recently been
recognized and well defined in a variety of cancers (Weng and
Aster, 2004). The constitutive activation of Notch signaling
in neuroblastomas has been shown to block induced differ-
entiation and neurite formation (Grynfeld et al., 2000; Levy
et al., 2002), suggesting the oncogenic effect of Notch in
neuroblastomas. It is thus plausible that the inhibition of
Notch signaling by specific ␥-secretase inhibitors might in-
terrupt its oncogenic effect and promote the differentiation of
neuroblastoma cells, resulting in a decrease in their malig-
nancy. We demonstrated by biochemical and morphological
analyses that all active Jia compounds can block Notch sig-
naling by reducing the generation of NICD and induce the
neuronal differentiation of neuroblastoma cells. These data
are in accordance with the idea that neurite outgrowth is
markedly altered upon the inhibition of ␥-secretase (Figueroa
et al., 2002) and that activation of Notch signaling results in
shrinkage of neuritic processes of cultured neuroblastoma
cells (Ishikura et al., 2005). Our findings thus not only vali-
date the efficacy of unnatural amino acid-containing (hy-
droxyethyl)urea peptidomimetics in biological systems but
also demonstrate the beneficial effects of active Jia com-
Acknowledgments
We thank Dr. Michael Wolfe for providing compound E and DAPT
and Dr. Yu-Min Kuo for providing the SH-SY5Y human neuroblas-
toma cell line. We are also grateful to Drs. Jen-Leih Wu, John Yu,
and to Nin-Nin Chuang for generous support. We thank the Core
Facility of the Institute of Cellular and Organismic Biology, Aca-
demia Sinica, for technical support. We are also indebted to the
National Science Council Regional Instruments Center at National
Taiwan University and National Chung-Kung University for assis-
tance in high-resolution mass spectrometry.
References
Bakshi P, Liao YF, Gao J, Ni J, Stein R, Yeh LA, and Wolfe MS (2005) A high-
throughput screen to identify inhibitors of amyloid {beta}-protein precursor pro-
cessing. J Biomol Screen 10:1–12.
Bakshi P, and Wolfe MS (2004) Stereochemical analysis of (hydroxyethyl)urea pep-
tidomimetic inhibitors of gamma-secretase. J Med Chem 47:6485–6489.
Capell A, Grunberg J, Pesold B, Diehlmann A, Citron M, Nixon R, Beyreuther K,
Selkoe DJ, and Haass C (1998) The proteolytic fragments of the Alzheimer’s
disease-associated presenilin-1 form heterodimers and occur as a 100–150-kDa
molecular mass complex. J Biol Chem 273:3205–3211.
Citron M, Diehl TS, Gordon G, Biere AL, Seubert P, and Selkoe DJ (1996) Evidence
that the 42- and 40-amino acid forms of amyloid beta protein are generated from
the beta-amyloid precursor protein by different protease activities. Proc Natl Acad
Sci USA 93:13170–13175.
Cook DG, Forman MS, Sung JC, Leight S, Kolson DL, Iwatsubo T, Lee VM, and
Doms RW (1997) Alzheimer’s A beta(1-42) is generated in the endoplasmic retic-
ulum/intermediate compartment of NT2N cells. Nat Med 3:1021–1023.
Dovey HF, John V, Anderson JP, Chen LZ, de Saint Andrieu P, Fang LY, Freedman
SB, Folmer B, Goldbach E, Holsztynska EJ, et al. (2001) Functional gamma-
secretase inhibitors reduce beta-amyloid peptide levels in brain. J Neurochem
76:173–181.
Edbauer D, Winkler E, Regula JT, Pesold B, Steiner H, and Haass C (2003) Recon-
stitution of gamma-secretase activity. Nat Cell Biol 5:486–488.
Esler WP, Das C, and Wolfe MS (2004) Probing pockets S2–S4Ј of the gamma-
secretase active site with (hydroxyethyl)urea peptidomimetics. Bioorg Med Chem
Lett 14:1935–1938.
Esler WP, Kimberly WT, Ostaszewski BL, Diehl TS, Moore CL, Tsai JY, Rahmati T,
Xia W, Selkoe DJ, and Wolfe MS (2000) Transition-state analogue inhibitors of
gamma-secretase bind directly to presenilin-1. Nat Cell Biol 2:428–434.
Esler WP, Kimberly WT, Ostaszewski BL, Ye W, Diehl TS, Selkoe DJ, and Wolfe MS
(2002) Activity-dependent isolation of the presenilin-gamma-secretase complex
reveals nicastrin and a gamma substrate. Proc Natl Acad Sci USA 99:2720–2725.
Figueroa DJ, Morris JA, Ma L, Kandpal G, Chen E, Li YM, and Austin CP (2002)
Presenilin-dependent gamma-secretase activity modulates neurite outgrowth.
Neurobiol Dis 9:49–60.
Flexner C (1998) HIV-protease inhibitors. N Engl J Med 338:1281–1292.
Getman DP, DeCrescenzo GA, Heintz RM, Reed KL, Talley JJ, Bryant ML, Clare M,
Houseman KA, Marr JJ, Mueller RA, et al. (1993) Discovery of a novel class of
potent HIV-1 protease inhibitors containing the (R)-(hydroxyethyl)urea isostere.
J Med Chem 36:288–291.
Greenlee WJ (1990) Renin inhibitors. Med Res Rev 10:173–236.
Grynfeld A, Pahlman S, and Axelson H (2000) Induced neuroblastoma cell differen-
tiation, associated with transient HES-1 activity and reduced HASH-1 expression,
is inhibited by Notch1. Int J Cancer 88:401–410.

␥-Secretase Inhibitors Induce Neuronal Differentiation
601
Hartmann T, Bieger SC, Bruhl B, Tienari PJ, Ida N, Allsop D, Roberts GW, Masters
CL, Dotti CG, Unsicker K, et al. (1997) Distinct sites of intracellular production for
Alzheimer’s disease A beta40/42 amyloid peptides. Nat Med 3:1016–1020.
Hsu WM, Hsieh FJ, Jeng YM, Kuo ML, Chen CN, Lai DM, Hsieh LJ, Wang BT, Tsao
PN, Lee H, et al. (2005) Calreticulin expression in neuroblastoma–a novel inde-
pendent prognostic factor. Ann Oncol 16:314–321.
Huff JR (1991) HIV protease: a novel chemotherapeutic target for AIDS. J Med Chem
34:2305–2314.
Ishikura N, Clever JL, Bouzamondo-Bernstein E, Samayoa E, Prusiner SB, Huang
EJ, and DeArmond SJ (2005) Notch-1 activation and dendritic atrophy in prion
disease. Proc Natl Acad Sci USA 102:886–891.
Mumm JS and Kopan R (2000) Notch signaling: from the outside in. Dev Biol
228:151–165.
Nadin A, Owens AP, Castro JL, Harrison T, and Shearman MS (2003) Synthesis and
gamma-secretase activity of APP substrate-based hydroxyethylene dipeptide isos-
teres. Bioorg Med Chem Lett 13:37–41.
Pahlman S, Stockhausen MT, Fredlund E, and Axelson H (2004) Notch signaling in
neuroblastoma. Semin Cancer Biol 14:365–373.
Seiffert D, Bradley JD, Rominger CM, Rominger DH, Yang F, Meredith JE Jr, Wang
Q, Roach AH, Thompson LA, Spitz SM, et al. (2000) Presenilin-1 and -2 are
molecular targets for gamma-secretase inhibitors. J Biol Chem 275:34086–34091.
Shearman MS, Beher D, Clarke EE, Lewis HD, Harrison T, Hunt P, Nadin A, Smith
AL, Stevenson G, and Castro JL (2000) L-685,458, an aspartyl protease transition
state mimic, is a potent inhibitor of amyloid beta-protein precursor gamma-
secretase activity. Biochemistry 39:8698–8704.
Singh US, Pan J, Kao YL, Joshi S, Young KL, and Baker KM (2003) Tissue trans-
glutaminase mediates activation of RhoA and MAP kinase pathways during reti-
noic acid-induced neuronal differentiation of SH-SY5Y cells. J Biol Chem 278:391–
399.
Weng AP and Aster JC (2004) Multiple niches for Notch in cancer: context is
everything. Curr Opin Genet Dev 14:48–54.
Wild-Bode C, Yamazaki T, Capell A, Leimer U, Steiner H, Ihara Y, and Haass C
(1997) Intracellular generation and accumulation of amyloid ␤-peptide terminat-
ing at amino acid 42. J Biol Chem 272:16085–16088.
Wolfe MS, Esler WP, and Das C (2002) Continuing strategies for inhibiting Alzhei-
mer’s gamma-secretase. J Mol Neurosci 19:83–87.
Wolfe MS and Haass C (2001) The Role of presenilins in gamma-secretase activity.
J Biol Chem 276:5413–5416.
Wolfe MS, Xia W, Moore CL, Leatherwood DD, Ostaszewski B, Rahmati T, Donkor
IO, and Selkoe DJ (1999a) Peptidomimetic probes and molecular modeling suggest
that Alzheimer’s gamma-secretase is an intramembrane-cleaving aspartyl pro-
tease. Biochemistry 38:4720–4727.
Wolfe MS, Xia W, Ostaszewski BL, Diehl TS, Kimberly WT, and Selkoe DJ (1999b)
Two transmembrane aspartates in presenilin-1 required for presenilin endoprote-
olysis and gamma-secretase activity. Nature (Lond) 398:513–517.
Iwatsubo T (2004) The gamma-secretase complex: machinery for intramembrane
proteolysis. Curr Opin Neurobiol 14:379–383.
Kimberly WT, LaVoie MJ, Ostaszewski BL, Ye W, Wolfe MS, and Selkoe DJ (2003)
Gamma-secretase is a membrane protein complex comprised of presenilin, nicas-
trin, Aph-1, and Pen-2. Proc Natl Acad Sci USA 100:6382–6387.
Kopan R, Schroeter EH, Weintraub H, and Nye JS (1996) Signal transduction by
activated mNotch: importance of proteolytic processing and its regulation by the
extracellular domain. Proc Natl Acad Sci USA 93:1683–1688.
Kornilova AY, Das C, and Wolfe MS (2003) Differential effects of inhibitors on the
gamma-secretase complex. Mechanistic implications. J Biol Chem 278:16470–
16473.
Levy OA, Lah JJ, and Levey AI (2002) Notch signaling inhibits PC12 cell neurite
outgrowth via RBP-J-dependent and -independent mechanisms. Dev Neurosci
24:79–88.
Li YM, Lai MT, Xu M, Huang Q, DiMuzio-Mower J, Sardana MK, Shi XP, Yin KC,
Shafer JA, and Gardell SJ (2000a) Presenilin 1 is linked with gamma-secretase
activity in the detergent solubilized state. Proc Natl Acad Sci USA 97:6138–6143.
Li YM, Xu M, Lai MT, Huang Q, Castro JL, DiMuzio-Mower J, Harrison T, Lellis C,
Nadin A, Neduvelil JG, et al. (2000b) Photoactivated gamma-secretase inhibitors
directed to the active site covalently label presenilin 1. Nature (Lond) 405:689–
694.
Liao YF, Wang BJ, Cheng HT, Kuo LH, and Wolfe MS (2004) Tumor necrosis
factor-alpha, interleukin-1␤, and interferon-␥ stimulate ␥-secretase-mediated
cleavage of amyloid precursor protein through a JNK-dependent MAPK pathway.
J Biol Chem 279:49523–49532.
Mani S, Shen Y, Schaefer J, and Meiri KF (2001) Failure to express GAP-43 during
neurogenesis affects cell cycle regulation and differentiation of neural precursors
and stimulates apoptosis of neurons. Mol Cell Neurosci 17:54–66.
Address correspondence to: Dr. Yung-Feng Liao, Institute of Cellular and
Organismic Biology, Rm 238, Academia Sinica, 128 Academia Rd. Sec. 2,
Taipei 115, Taiwan. E-mail: yliao@sinica.edu.tw