Small molecule microarray-facilitated screening of affinity-based probes (AfBPs) for γ-secretase

Article abstract of DOI:10.1039/b910611a

A small molecule microarray (SMM) platform is developed herein, which enables high-throughput discovery of affinity-based probes (AfBPs) against γ-secretase.

Full text of DOI:10.1039/b910611a

View Article Online / Journal Homepage / Table of Contents for this issue
COMMUNICATION
www.rsc.org/chemcomm | ChemComm
Small molecule microarray-facilitated screening of affinity-based probes
(Af BPs) for c-secretasew
Haibin Shi,ac Kai Liu,b Ashley Xua and Shao Q. Yao*abc
Received (in Cambridge, UK) 1st June 2009, Accepted 25th June 2009
First published as an Advance Article on the web 14th July 2009
DOI: 10.1039/b910611a
A small molecule microarray (SMM) platform is developed
herein, which enables high-throughput discovery of affinity-
based probes (AfBPs) against c-secretase.
To convert an inhibitor into an effective AfBP, the following
two steps are taken: (1) identification of a suitable inhibitor
(or an inhibitor scaffold) against the target enzyme (or the
class of enzymes). This typically limits the AfBP strategy only
to proteins for which inhibitors are known
a
priori.2
Amyloid b-protein (Ab), a central pathogenic feature of
Alzheimer’s disease (AD), is generated through proteolysis
of amyloid precursor protein (APP) by membrane-associated
aspartic proteases including b- and g-secretases.1 b-secretase
has been well characterized; the study on g-secretase, however,
has been much more challenging, because the enzyme is a
multimeric membrane protein complex, comprising presenilin
(PS), nicastrin, anterior pharynx defective 1 (Aph-1) and
presenilin enhancer 2 (Pen-2). The proteolytically active form
of PS is made up of two subunits, termed the N-terminal
fragment (NTF) and C-terminal fragment (CTF). Despite
recent advances, much remains to be understood about what
the g-secretase complex interacts with and how it carries out
the proteolytic process. Due to well-known difficulties
associated with biochemical studies of membrane proteins,
much work on g-secretase has so far relied on the development
of the so-called affinity-based probes (AfBPs), which are
typically generated from known inhibitors of g-secretase by
attaching a photo-crosslinking group (i.e. benzophenone (BP)
or diazirine) and a fluorescent dye (i.e. tetraethylrhodamine
(TER)) to the inhibitors.2 For example, AfBPs based on
L685 458, a potent hydroxylethylene-based g-secretase inhibitor,
were first developed by Li et al.2a Recently, probes based on
DAPT and arylsulfonamide derivatives have also been
reported.2b–d The key advantage of all these probes is their
ability to selectively bind the active subunits of g-secretase
complex (i.e. PS subunits) from the whole cell extract and,
upon UV-photolysis and in gel-analysis, accurately report the
active state of g-secretase without the need of isolating the
enzyme complex. As a result, these probes have provided
invaluable tools for the studies of Alzheimer’s disease, both
in its pathology and potential therapeutics. Similar strategies
are widely used for the development of AfBPs against
other classes of enzymes in the field of activity-based protein
profiling (ABPP) pioneered by Evans and Cravatt3
(2) Introduction of a BP–fluorophore linker into the molecule,
which in many cases may abolish its target-binding ability and,
as a result, leads to inactive probes.2b Consequently, a much
larger number of potential probes (than the final probes
obtained) are usually synthesized chemically and tested
biologically. Clearly, there is an urgent need to develop new
ways for rapid screening and identification of AfBPs against
different protein targets, including both well-known and
less-characterized ones. Herein, we report the first small
molecule microarray (SMM)-facilitated approach for high-
throughput identification of AfBPs, in which g-secretase was
used as our model target.
Small molecule microarrays (SMMs) are becoming
increasingly important for the large-scale determination of
protein–ligand interactions, high-throughput screening of
enzyme substrates/inhibitors and the development of bio-
sensors and biomarkers for disease analysis.4 In our studies, we
realized one of the key steps in most current SMM technologies
requires the site-specific immobilization of small molecules
onto the glass slide through the introduction of a suitable
chemical ‘‘tag’’ (and the connecting linker) into the small
molecules (e.g. biotin in Fig. 1).4b Upon SMM screening, the
‘‘hits’’ identified were in fact the ‘‘tagged’’ small molecules. We
reasoned that the direct conversion of biotin into a linker
containing both BP and a dye should retain the full protein-
binding ability of the ‘‘hits’’, thus making them suitable AfBPs
against the target protein. So by combining concepts in
microarray technologies and activity-based protein profiling,
new AfxBPs may be rapidly identified.
To apply our strategy to the screening of AfBPs against
g-secretase, we developed
a
solid-phase method for
a Department of Chemistry, National University of Singapore,
3 Science Drive 3, Singapore 117543, Singapore
b Department of Biological Sciences, National University of Singapore,
3 Science Drive 3, Singapore 117543, Singapore
c NUS MedChem Program of the Life Sciences Institute, National
University of Singapore, 3 Science Drive 3, Singapore 117543,
Singapore. E-mail: chmyaosq@nus.edu.sg; Fax: +1 (65) 6779169;
Tel: +1 (65) 65162925
w Electronic supplementary information (ESI) available: Detailed
experimental procedures and results. See DOI: 10.1039/b910611a
Fig. 1 Overall strategy of the small molecule microarray (SMM)-
facilitated platform for high-throughput identification of AfBPs.
ꢀc
This journal is The Royal Society of Chemistry 2009
5030 | Chem. Commun., 2009, 5030–5032

View Article Online
Scheme 1 Synthetic strategy of the N- and C-terminal libraries and click assembly for AfBPs.
combinatorial synthesis of hydroxylethylene-based small
molecule inhibitors (Scheme 1). Hydroxylethylenes are
well-known scaffolds as transition state analogs of aspartic
proteases, and have been included in the inhibitor design
of numerous aspartic proteases including g-secretase. In the
123-member N-terminal sublibrary, R1 and R2 positions,
corresponding to the P1 and P2 residues, respectively,
upon binding to the aspartic protease, were varied with
N-terminal compounds. Similarly, 8(a–c), again from the
corresponding three L-amino acids, were attached to DHP
resin, followed by alkylation reaction with i-butylamine,
acylation with commercially available sulfonyl chloride building
blocks (1–25 in Fig. S5w), reduction with SnCl2/PhSH,
attachment of the biotin linker then TFA cleavage, gave the
75-member C-terminal compounds. Finally, the resulting
198 compounds were analyzed by LC-MS, which confirmed
most were of sufficient purity (480%) and were subsequently
used for direct immobilization onto avidin-functionalized slides
to generate the corresponding small molecule microarrays.
To assess whether the SMM could be used to detect aspartic
protease–small molecule interactions, we carried out our
preliminary screening with the recombinantly purified protein,
histo-aspartic protease (HAP), which is one of the aspartic
proteases from the parasite Plasmodium falciparum.6 HAP was
fluorescently labeled with Cy3 before being applied onto the
SMM. As negative controls, three inactive mutant proteins of
HAP (E278K, H34A and S37A mutants), previously shown to
have minimum enzymatic activity, were used. Subsequently,
the slides were scanned using an ArrayWoRx microarray scanner
installed with the relevant filters (Cy3: lex/em = 548/595 nm).
Only HAP showed distinct binding profiles against the SMM,
while none of the three active-site mutants showed any
significant binding (see ESIw). This indicates that our SMM
platform was able to sensitively report activity-based binding
events of aspartic proteases towards their small molecule
inhibitors. Previous SMM strategies have mainly relied on
incubation with a purified protein of interest.4 Recently,
Koehler and coworkers have shown cell lysates expressing a
target protein could be used directly for ligand identification in
SMM as well.7 In our case, since it is notoriously difficult to
isolate g-secretase, we decided to carry out our SMM screening
using the fluorescently labeled membrane fraction of mammalian
cell lysates. The g-30 cell line, which is a g-secretase
overexpressing CHOK1 mammalian cell line,8 was used to
aromatic/aliphatic moieties (R1 = Phe, Leu, Ala; R2
=
aromatic/aliphatic acid (1-55) as shown in Fig. S4 in the
0
0
ESIw). In the 75-member C-terminal sublibrary, R1 and R2
0
positions, corresponding to the P1 and P1/2 positions, respec-
tively, were similarly varied. In both sublibraries, a biotin tag
(and a connecting linker) was introduced in each compound
for subsequent immobilization onto SMM. Since most known
0
aspartic proteases recognize the four key residues (P2, P1, P1
and P2/30) flanking the scissile bond of the substrates, our
198-member combined library could potentially be used for
screening of other aspartic proteases besides g-secretase. As
shown in Scheme 1, the synthetic strategy was so chosen
because (1) it’s solid-phase and amenable to large-scale
synthesis of hydroxylethylene-containing compounds from
readily available Fmoc-protected L-amino acids; (2) the
diversity of the library can be easily introduced throughout
0
the P2–P2 positions; (3) the key azido intermediates in the
synthesis (shaded in Scheme 1) could be easily recovered and
converted to the corresponding AfBPs using ‘‘click chemistry’’.
Briefly, 4(a–c), synthesized from the corresponding commercial
available L-amino acids following previous reported
procedures with some modifications,5 were first immobilized
at the –OH group onto 3,4-dihydro-2H-pyran (DHP) resin
under acid-catalyzed conditions. Subsequently, deprotection
of the Fmoc group followed by standard acylation with
different acid building blocks (1–55 in Fig. S4w) and azide
reduction, gave the corresponding amines. Finally, attachment
of a biotin linker then TFA cleavage gave the 123-member
ꢀc
This journal is The Royal Society of Chemistry 2009
Chem. Commun., 2009, 5030–5032 | 5031

View Article Online
procedures were described in ESI. As shown in Fig. 3a, the
two positive probes identified from SMM, F5 and F24
(corresponding to C-c-5 and C-c-24 in SMM, respectively)
strongly labeled a 26-KDa protein in the total lysate of g-30
mammalian cell line with high specificity. This band was sub-
sequently confirmed as the N-terminal fragment of PS (PS-NTF)
by both western blotting with anti-PS antibody and pull-down
experiments (Fig. 3b). On the contrary, the total lysate of g-30
treated with the negative control A31 (corresponding to N-a-31
in SMM) failed to give any noticeable labeled band. From our
results, it is interesting to note only PS-NTF was labeled by our
probes. Previous work on the g-secretase complex had identified
that both NTF and CTF are needed to form the catalytic core of
PS.1,2 A quick survey of other known AfBPs of g-secretase
revealed that most also labeled only PS-NTF.2b,c This thus
unambiguously confirms our SMM-facilitated screening
approach as a valuable tool for rapid discovery of AfBPs.
In conclusion, a small molecule microarray-facilitated
screening strategy has been developed for high-throughput
identification of affinity-based probes. This approach does not
require known inhibitors of a protein target and minimizes
risks involved in the loss of protein-binding property of these
inhibitors due to linker introduction. We further demonstrated
the utility of this method by successfully identifying highly
specific g-secretase probes from a generic hydroxylethylene
small molecule library.
Fig. 2 SMM of the 198-member library screened against fluorescently
labeled membrane fraction of g-30 cell lysate, with (left) microarray
image showing selected binders and non-binder and (right) their
chemical structures. All compounds were printed in duplicate
(horizontally). See ESI for spotting pattern.w
ensure a sufficient amount of active g-secretase was present in
the lysate. The cells were grown as previously described.8
Subsequently, the membrane fraction was isolated, solubilized
and fluorescently labeled with Cy3 before being applied onto
the SMM. As shown in Fig. 2, highly reproducible and distinct
binding profile of the 198-member library against the
membrane fraction of g-30 lysates could be obtained.
Interestingly, none of the N-terminal sublibrary showed any
significant fluorescent binding. Of the strong binders, most
were identified as members of the C-terminal sublibrary
containing an alanine residue at the P1 position. C-c-5 and
C-c-24 showed the strongest relative fluorescence binding,
and therefore were chosen for further studies. N-a-31, a
non-binder, was also chosen as a negative control.
Funding was provided by MOE (R143-000-394-112),
BMRC (R143-000-391-305) and CRP (R143-000-218-281).
We thank R Yada (University of Guelph, Canada) and
M Wolfe (Harvard) for the kind gifts of HAP (and mutants)
and g-30 cell line, respectively.
Notes and references
To make the corresponding AfBPs of above three compounds,
their azido intermediates (shaded) were taken, and ‘‘click’’
assembled with TER-BP alkyne (for in gel activity-based profiling)
and Biotin-BP alkyne (for pull-down experiments), as shown in
Scheme 1. Previously, the modularity of ‘‘click chemistry’’ had
been successfully explored to facilitate the synthesis of other
classes of activity-based probes.9 To demonstrate the ability of
our SMM-derived AfBPs for UV-initiated, activity-based
profiling of g-secretase in cellular lysates, we carried out in-gel
fluorescence labeling and pull-down experiments. Detailed
1 R. Jakob-Roetne and H. Jacobsen, Angew. Chem., Int. Ed., 2009,
48, 3030–3059.
2 (a) Y.-M. Li, et al., Nature, 2000, 405, 689–694; (b) Y. Takahashi,
Y. Konno, N. Watanabe, H. Miyashita, M. Sasaki, H. Natsugari,
T. Kan, T. Fukuyama, T. Tomita and T. Iwatsubo, ACS Chem.
Biol., 2007, 2, 408–418; (c) J. Chun, Y. I. Yin, G. Yang,
L. Tarassishin and Y.-M. Li, J. Org. Chem., 2004, 69, 7344–7347;
(d) H. Fuwa, K. Hiromotoa, Y. Takahashia, S. Yokoshimaa,
T. Kan, T. Fukuyamaa,
T Iwatsuboa, T. Tomitaa and
H. Natsugari, Bioorg. Med. Chem. Lett., 2006, 16, 4184–4189.
3 M. J. Evans and B. F. Cravatt, Chem. Rev., 2006, 106, 3279–3301.
4 (a) J. L. Duffner, P. A. Clemons and A. N. Koehler, Curr. Opin.
Chem. Biol., 2007, 11, 74–82; (b) M. Uttamchandani, J. Wang and
S. Q. Yao, Mol. BioSyst., 2006, 2, 58–68.
5 (a) M. Chino, M. Wakao and J. A. Ellman, Tetrahedron, 2002, 58,
6305–6310; (b) A. Brik, J. Muldoon, Y.-C. Lin, J. H. Elder,
D. S. Goodsell, A. J. Olson, V. V. Fokin, K. B. Sharpless and
C.-H. Wong, ChemBioChem, 2003, 4, 1246–1248.
6 H. Xiao, T. Tanaka, M. Ogawa and R. Y. Yada, Protein Eng., Des.
Sel., 2007, 20, 625–633.
7 J. E. Bradner, O. M. McPherson, R. Mazitschek, D. Barnes-Seeman,
J. P. Shen, J. Dhaliwal, K. E. Stevenson, J. L. Duffner, S. B. Park,
D. S. Neuberg, P. Nghiem, S. L. Schreiber and A. N. Koehler, Chem.
Biol., 2006, 13, 493–504.
8 W. T. Kimberly, M. J. LaVoie, B. L. Ostaszewski, W. Ye,
M. S. Wolfe and D. J. Selkoe, Proc. Natl. Acad. Sci. U. S. A.,
2003, 100, 6382–6387.
9 K. A. Kalesh, P.-Y. Yang, R. Srinivasan and S. Q. Yao, QSAR
Comb. Sci., 2007, 26, 1135–1144.
Fig. 3 (left) In-gel fluorescence profiling of g-30 lysates using different
Af BPs. The arrowed band was identified as PS-NTF. (right) After
pull-down, western blotting showed the biotinylated protein in g-30
cell lysate labeled by Biotin-F24 as PS-NTF.
ꢀc
This journal is The Royal Society of Chemistry 2009
5032 | Chem. Commun., 2009, 5030–5032