Microarray-assisted high-throughput identification of a cell-permeable small-molecule binder of 14-3-3 proteins

Article abstract of DOI:10.1002/anie.201003257

The way to a 14-3-3 binder: A fragment-based combinatorial small-molecule microarray generates affinity-based fingerprints of the 14-3-3σ protein. One small molecule (see picture; in red box) that disrupts the 14-3-3/protein interaction (green/blue) has b

Full text of DOI:10.1002/anie.201003257

Communications
DOI: 10.1002/anie.201003257
Protein–Protein Interactions
Microarray-Assisted High-Throughput Identification of a Cell-
Permeable Small-Molecule Binder of 14-3-3 Proteins**
Hao Wu, Jingyan Ge, and Shao Q. Yao*
Protein–protein interactions (PPI) control important regula-
tory pathways in virtually all cellular processes.[1] For
example, the phosphorylation-dependent PPI, tightly regu-
lated by kinases, phosphatases, and a large number of proteins
containing phosphopeptide-binding domains, forms the basis
of the highly complex, cellular signal transduction network.
Inappropriate interactions between proteins are known to
cause many human diseases.[2] Inhibitors capable of disrupting
protein–protein interactions, and especially those made of
small, cell-permeable molecules, are therefore attractive leads
for drug discovery.[3] In contrast to mainstream drugs, which
are usually active-site inhibitors of enzymes (such as Gleevec
for Abl kinase), and have historically been the focus of
pharmaceutical research, the development of PPI inhibitors is
perceived to be highly challenging, but at the same time
rewarding, especially for treating diseases of less-common
therapeutic targets, such as transcription factors. Examples
include inhibitors that bind to different Src Homology 2
(SH2) domains, which have been extensively investigated.[4]
SH2 domains are phosphotyrosine (pTyr)-binding domains
present in many important proteins (such as Src and Abl
kinases). Phosphoserine/phosphothreonine (pS/pT)-binding
domains constitute another major class of signaling domains
with prominent roles in regulation of mitosis, DNA damage,
and apoptosis.[5] Among them, 14-3-3 proteins, made up of a
family of acidic, dimeric, a-helical, cup-shaped molecules
present in all eukaryotic cells, have been well-characterized.[6]
Small-molecule-based PPI inhibitors of 14-3-3 proteins, how-
ever, are not currently available.
peptide-binding pocket, most 14-3-3 family members display
significant functional redundancy both in vitro and in vivo.[6b,c]
The elegant work of Yaffe et al. has unequivocally established
that most proteins bind to 14-3-3 proteins through a consensus
hexa- or heptaphosphopeptide binding motif, RXX[pS/
pT]XP or RXXX[pS/pT]XP, respectively (where X represents
any amino acid, but is in most cases an aromatic/hydrophobic
residue).[6b] Recent studies have further identified sequences
such as RFRpSYPP and RLSHpSLPG as optimal peptides
recognized by all 14-3-3 proteins,[6b] as well as LFGpSLLR
and LFGpSLVR that confer some preference toward the
14-3-3s over other isoforms.[7]
The high redundancy of 14-3-3 proteins makes classical
gene knockout approaches ill-suited for the study of their
general functions in cells, as single gene knockout will be
compensated for by other 14-3-3 isoforms, and simultaneous
knockout of all seven genes would be technically impossible.
Fu et al. used the so-called difopein (two 18-amino acid
peptides joined by a linker) as a general 14-3-3 PPI inhibitor.
Upon overexpression in mammalian cells from the corre-
sponding DNA, the compound was shown to disrupt 14-3-3/
ligand interaction, resulting in cell cycle arrest and apopto-
sis.[8a,b] This method, however, depends on transient trans-
fection and is difficult to control (in terms of time, concen-
tration, etc). Yaffe et al. recently showed the use of caged
phosphopeptides for both in vitro and in-cell studies of 14-3-3
functions in a temporally controlled manner.[8c] The peptides
were chemically synthesized and conjugated (to a cell-
penetrating peptide, or CPP) prior to being delivered into
cells, and subsequently “turned on” by UV irradiation.
Although elegant, this approach lacks a “turn-off” mecha-
nism and requires the use of UV, which is harmful to cells.
Furthermore, these peptides lack sufficient cellular perme-
ability (thus the need for CPP conjugation) and hydrolytic
stability (they are readily degraded by endogenous proteases
in cells), making them unsuitable for general chemical biology
applications. Our ongoing interests in
In humans, seven highly homologous 14-3-3 isoforms are
present (b, e, h, g, s, t, and z). Together, they regulate several
hundred proteins, many of which are important pharmaceut-
ical targets, such as Raf, p53, Cdc25, and histone deacetylases
(HDACs).[6a] To date, only 14-3-3s, through its direct inter-
action with p53 (a major tumor suppressor protein), has been
linked to cancer. Inactivation of 14-3-3s has been shown to be
crucial in tumorigenesis. Due to the highly conserved nature
of residues in 14-3-3 isoforms that constitute the pS/pT
14-3-3 proteins and their protein–protein interacting partners
as potential therapeutic targets have prompted us to ask
whether it is possible to develop cell-permeable small
molecules that are capable of binding to all 14-3-3 proteins,
or to a specific isoform, namely 14-3-3s. We envisioned that
such molecules would be highly valuable not only as a
research tool for studies of 14-3-3 biology (especially that of
14-3-3/ligand interaction), but also as potential anti-cancer
agents. Herein, we present one such compound, 2-5
(Figure 1). To the best of our knowledge, this compound is
the first small-molecule PPI inhibitor of 14-3-3 proteins. We
have carried out extensive biochemical assays to confirm that
2-5 directly competes for 14-3-3/ligand binding in a potent and
[*] H. Wu, J. Ge, Prof. Dr. S. Q. Yao
Department of Chemistry, National University of Singapore
3 Science Drive 3, Singapore 117543 (Singapore)
Fax: (+65)6779-1691
E-mail: chmyaosq@nus.edu.sg
Homepage: http://staff.science.nus.edu.sg/~syao
[**] Funding was provided by the Agency for Science, Technology and
Research (R-143-000-391-305) and the Ministry of Education (R-143-
000-394-112).
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201003257.
6528
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 6528 –6532

Angewandte
Chemie
Figure 1. Overview of small-molecule microarray-assisted, fragment-based high-throughput identification of potential peptide–small-molecule
hybrid molecules (steps I and II), which were subsequently reconstituted (step III) to obtain the corresponding cell-permeable, small-molecule
hits, followed by biological validation (step IV). ID=identification, SM=small molecule, PPI=protein–protein interaction.
specific manner in vitro. In mammalian cells, it induces
apoptosis and cell cycle arrest.
Scheme S1), following previously published procedures,[10]
then coupling of the peptide sequence RFRpS. For both
sub-libraries, biotin and a GG linker were introduced at either
the N- or C-terminus of each member to facilitate subsequent
immobilization onto avidin-functionalized glass slides. All the
compounds were characterized by liquid chromatography–
mass spectrometry and most were shown to be of the correct
molecular weight and sufficient purity for direct microarray
applications.[11]
To identify 2-5 and related compounds in high throughput,
we used the recently developed small-molecule microarray
(SMM) technique,[9] and followed the workflow shown in
Figure 1. To convert the original optimal 14-3-3-binding
peptide, RFRpSYPP, into a small-molecule-based 14-3-3
binder, we first synthesized a peptide-small molecule hybrid
library (step I); it is made up of the 243-member N-terminal
and 50-member C-terminal sub-libraries (for synthesis, see
the Supporting Information, Scheme S1). We retained the key
pS residue in the library, and at the same time systematically
replaced the two flanking peptide fragments (RFR and YPP)
with commercially available acid and amine building blocks
(R1 and R2, respectively, in Figure 1, corresponding to the
PÀ1/À2 and P+1/+2 positions in the original peptide). A recently
introduced library design concept termed “fragment-based
combinatorial chemistry” was adopted to ensure that a
sufficient chemical space was covered by a limited number
of building blocks (243 R1 and 50 R2) , and at the same time all
library members retained some degrees of binding affinity
toward 14-3-3 proteins.[7] Aliphatic/aromatic building blocks
were chosen in the library construction, as they may improve
the cell permeability of potential “hits”, and based on X-ray
structures of 14-3-3 proteins, the binding pockets immediately
adjacent to the pS-binding site are mostly hydrophobic.[6b,c]
Briefly, the N-terminal sub-library was made by coupling 243
different acids (R1) to the rink amide resin pre-loaded with
peptide NH2-pSYPP (where R1 is an aliphatic building block)
or NH2-GpSYPP (where R1 is an aromatic building block).
Similarly, the C-terminal sub-library was synthesized by initial
immobilization of 50 different amines (R2) onto the PL-FMP
resin by reductive amination (Supporting Information,
The entire library, together with a positive control
peptide, Biotin-GG-RLSHpSLPG, was spotted (in duplicate)
to generate the corresponding SMM. To ensure uniform
immobilization, spotted slides were stained with Pro-Q
diamond dye (Supporting Information, Figure S1); minimal
slide-to-slide and spot-to-spot variations (r> 0.95) were
observed. We then examined whether this fragment-based,
peptide–small-molecule hybrid SMM could be used to
identify, in high throughput, ligands that bind strongly to
14-3-3 proteins (step II in Figure 1). A fluorescently labeled,
recombinant GST-14-3-3s was used in the SMM screening
(Figure 2); upon data analysis, highly reproducible (as judged
from duplicated spots/slides), concentration-dependent bind-
ing profiles were observed for five of the 293 compounds
(boxed in red in Figure 2a). The positive control peptide
(boxed in yellow in Figure 2a) also showed a robust binding
profile. Microarray-based KD determination was subse-
quently carried out (Figure 2b); besides the control peptide,
which showed an apparent KD of 0.25 mm (compared to the
reported value of about 200 nm[7]), the newly identified
ligands (N111, N180, and N231 from the N-terminal sub-
library, and C021 and C033 from the C-terminal sub-library)
all showed relatively good affinity toward 14-3-3s (0.6–
1.03 mm in KD). Further validation was carried out with a
Angew. Chem. Int. Ed. 2010, 49, 6528 –6532
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6529

Communications
most active site-targeted
enzyme inhibitors (which are
often nanomolar or even
picomolar inhibitors), small-
molecule micromolar PPI
inhibitors are considered rea-
sonably potent inhibitors.[3]
In our case, the most potent
inhibitor, 2-5, with a IC50
value of 2.6 Æ 0.4 mm, is only
about tenfold lower in bind-
ing affinity than its peptide
counterpart,[6,7] thus making
it a good candidate as a
potential
cell-permeable,
small-molecule 14-3-3 PPI
inhibitor. This compound
was therefore chosen for all
subsequent studies.
To investigate whether
compound 2-5 competes
with cellular proteins for
14-3-3 binding in a biological
complex, pull-down experi-
ments were performed with
A549 cell lysates.[8c] Bead-
Figure 2. SMM-assisted identification of potential peptide–small-molecule hybride binders of 14-3-3, and
quantitative KD determination. a) Microarray image showing binding profiles of the 293-member library
against fluorescently labeled GST-14-3-3s. Spots corresponding to the control peptide RLSHpSLPG and
positive “hits” are boxed in yellow and red, respectively. Other unboxed spots were weak binders (see
Supporting Information, Figure S2) that were rejected without further follow-ups. b) Microarray-based KD
determination of the control peptide and five “hits”. The corresponding dose-dependent microarray images,
shown below Figure 2a, were obtained by screening the SMM with varied concentrations of the fluorescently
labeled 14-3-3s (0.1 to 10 mm). RFU=relative fluorescence unit. (c) Summary of the microarray-measured KD
and the corresponding IC50 values obtained from competitive fluorescence polarization (FP) experiments.
immobilized
GST-14-3-3s
was pre-incubated with 2-5
(1-5 and DMSO were used as
negative controls) and then
added to cell lysates. 14-3-3-
bound proteins were sepa-
rated by SDS-PAGE fol-
lowed by immunoblotting
competitive fluorescence polarization (FP) assay (Fig-
ure 2c);[12] the same set of compounds emerged as “hits”
and with IC50 values within the same range as the microarray-
measured KD, indicating that are all likely to be true binders
of 14-3-3s, and the corresponding R1 and R2 building blocks
(boxed in green and blue, respectively, in Figure 1) may be
further reconstituted to obtain the corresponding small
molecule-based 14-3-3 PPI inhibitors (step III in Figure 1).
From the above peptide–small-molecule hybrid library,
we have rapidly identified key building blocks (i.e. three R1
and two R2) capable of substituting peptidyl recognition
fragments in a known 14-3-3-binding phosphopeptide without
significant loss of its original binding property. We then
reconstituted the corresponding phosphoserine-containing
small molecules (2-1 to 2-6 in Figure 1; step III). All six
potential 14-3-3 small molecule binders (from three R1 and
two R2) were synthesized. The non-phosphorylated controls,
1-1 to 1-6, were prepared in a similar fashion (Supporting
Information, Scheme S2).
with a group of different antibodies that recognize either
the 14-3-3 phosphopeptide-binding motif, p53, or Raf-1
substrate proteins. As shown in Figure 3b, the phosphoser-
ine-containing small molecule 2-5 was able to completely
disrupt PPI between 14-3-3s and its substrate proteins. On the
other hand, a same amount of the non-phosphorylated
control 1-5 was unable to disrupt the binding between 14-3-
3s and its endogenous substrates. This result indicates that 2-5
is a good small-molecule PPI inhibitor of 14-3-3 proteins in
vitro. The cell permeability and in-cell biological activities of
2-5 were subsequently assessed. A standard cell permeability
assay using MDCK cells indicated 2-5 indeed possesses
reasonable cell permeability with a Papp value of 554 nmsÀ1.
Previously, it had been shown that expression of 14-3-3/ligand
interaction inhibitors (namely difopein) in mammalian cells
leads to induction of apoptosis and cell death. We therefore
tested the cytotoxicity of 2-5 in A549 cells using the XTT cell
proliferation assay (Figure 3c); results indicate that, at a
concentration of 100 mm, 2-5 caused a large amount of cell
death starting from 4 h to up to 12 h of drug treatment. The
negative control (1-5) on the other hand had a significantly
lower but noticeable degree of cell toxicity, indicating that
although both compounds might have some intrinsic cytotox-
icity, phosphorylation of 1-5 (that is, giving 2-5) had drastically
improved its cell-killing activity, presumably through the
Competitive FP experiments showed potent dose-depen-
dent binding of 2-1 to 2-6, but not the non-phosphorylated
counterparts (1-1 to 1-6), toward 14-3-3s, giving rise to the
corresponding IC50 values of 2.6–3.6 mm, which are only two-
to threefold less potent than the parental peptide–small-
molecule hybrids (Figure 3a). It should be noted that, unlike
6530
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 6528 –6532

Angewandte
Chemie
Figure 3. a) Summary of IC50 obtained from competitive FP experiments for 2-1 to 2-6 and their non-phosphorylated controls 1-1 to 1-6.
(N.A.=IC50 too high to be determined within experimental settings). (b) In vitro competitive binding of 2-5 with cellular proteins for 14-3-3s
binding. 100 mm of compounds 1-5 and 2-5 were pre-incubated with bead-immobilized GST-14-3-3s then added to A549 cell lysates. Bound
proteins were detected by SDS-PAGE and immunoblotting with antibodies against various 14-3-3-binding partners. Controls show proteins bound
by GST-14-3-3s beads in the absence of compounds, or by beads alone. TCL denotes total cell lysate; GST denotes beads alone. c) Time-
dependent (1–24 h) inhibition of A549 cell proliferation by 2-5 (1-5 as control) using XTT assay. Data represent the average standard deviation for
two independent trials. d) Fluorescence microscopic images of A549 cells treated with compound 2-5 or 1-5 (100 mm), respectively. Caspase
activity was monitored by addition of Ac-DEVD-SG1 as previously reported.[13] All images were acquired in the same way. Insets: DIC images.
Scale bars=10 mm. e) Immunoblotting results of the cells from (d), detected with anti-procaspase-3 and anti-caspase-3 (P17) antibodies. Lane 1:
uninduced A549 lysate; lanes 2–4: 2-5 treated with A549 for 2, 8, and 12 h, respectively; lane 5: STS-treated cells (12 h). f) Flow cytometry results
of A549 cells treated with 100 mm of 2-5 for 4, 8, and 12 h (Normal: no 2-5).
increase in its binding affinity toward 14-3-3 proteins. We then
investigated whether 2-5-induced cell death was a direct result
of caspase activation and apoptosis, as previously observed
for other 14-3-3/ligand interaction inhibitors.[8a,b] A549 cells
treated with 2-5 were monitored in a live-cell bioimaging
experiment using the Ac-DEVD-SG1 fluorogenic caspase 3/7
substrate.[13] As shown in Figure 3d, A549 cells treated with
2-5 and staurosporine (STS; a positive control; data not
shown) developed a strong green fluorescence, indicating
apoptosis had occurred. In contrast, no significant increase in
fluorescence was observed for nonapoptotic cells, even after
extended incubation. Addition of a caspase 3/7 inhibitor, Ac-
DEVD-CHO, to the STS and 2-5-treated cells rendered the
fluorescence undetectable (see Supporting Information).
Further analysis of the corresponding cell lysates by SDS-
PAGE and immunoblotting experiments unambiguously
confirmed the time-dependent activation of caspase 3 activity
in both the STS and 2-5 treated cells (Figure 3e). These
results agree well with previous studies that disruption of
14-3-3/ligand interactions leads to caspase-mediated apopto-
sis,[8a,b] and further supports the hypothesis that
14-3-3/ligand interactions are required for critical cellular
processes, including regulating pro-survival cell signaling.[6]
Finally, 14-3-3/ligand interaction inhibitors are known to
regulate cell cycle by causing premature cell cycle entry,
release of G1 cells from interphase arrest, and loss of the S-
phase checkpoint after DNA damage.[8c] We wondered if 2-5
behaves similarly in a flow cytometry-based cell-cycle experi-
ment (Figure 3 f); by measuring the population of cells
containing cleaved DNA upon treatment with 2-5, we
observed a dramatic increase in the fraction of sub-G1 cells,
indicating internucleosomal DNA cleavage (which is a
commonly used marker of apoptosis). 1-5-treated cells, on
the other hand, behaved in a similar fashion to normal cells,
which again validates the critical roles that 14-3-3 proteins
play in cell cycle regulation.[8] Lastly, molecular docking
Angew. Chem. Int. Ed. 2010, 49, 6528 –6532
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6531

Communications
[2] B. T. Seet, I. Dikic, M. M. Zhou, T. Pawson, Nat. Rev. Mol. Cell
Biol. 2006, 7, 473 – 483.
[3] a) T. Berg, Angew. Chem. 2003, 115, 2566 – 2586; Angew. Chem.
Int. Ed. 2003, 42, 2462 – 2481; b) A. J. Wilson, Chem. Soc. Rev.
2009, 38, 3289 – 3300.
studies were carried out to explore the binding mode of 2-5
with 14-3-3s, and results showed that, as might be expected,
that 2-5 binds 14-3-3s well in the phosphopeptide binding site
of the protein (Supporting Information, Figure S5).
In conclusion, we have successfully demonstrated that the
SMM technique offers a highly miniaturized yet powerful
approach for rapid identification of small-molecule PPI
inhibitors. From this study, through the identification of a
cell-permeable small-molecule PPI inhibitor that targets 14-3-
3 proteins, we have successfully developed a versatile
chemical biology approach for studies of 14-3-3 biology.
Compared to other existing methods,[8] our strategy makes
use of the newly discovered compound 2-5, which can be
administered to cells by conveniently adding it to the cell
medium. The method therefore does not require ectopic
delivery of genetic materials or asynchronously delivered
peptides/proteins. It also possesses other reasonable “drug-
like” properties, such as small size and hydrolytic stability (see
Supporting Information, Figure S4).[14] All of these attributes
should make 2-5 a valuable chemical for probing 14-3-3
biology. We note that in a recent work, Waldmann, Ottmann
and co-workers have identified small molecules that selec-
tively stabilize, rather than disrupt, 14-3-3/ligand interac-
tions.[15] We believe these two types of reagents might offer a
powerful synergistic combination for future studies of 14-3-3/
ligand interactions and associated signaling pathways. Our
strategy should be directly applicable to other signaling
modules that are functionally redundant, in which the protein
of interest is part of a large conserved protein family,
including other modular protein domains that recognize
phosphoamino acids.[4,5] Throughout the course of our study,
we have further expanded the utility of SMM beyond some of
its most recent applications.[9,11,16]
[4] a) W. C. Shakespeare, Curr. Opin. Chem. Biol. 2001, 5, 409 – 415;
b) P. L. Toogood, J. Med. Chem. 2002, 45, 1543 – 1558.
[5] M. B. Yaffe, A. E. Elia, Curr. Opin. Cell Biol. 2001, 13, 131 – 138.
[6] a) H. Hermeking, Nat. Rev. Cancer 2003, 3, 931 – 943; b) M. B.
Yaffe, K. Rittinger, S. Volinia, P. R. Carson, A. Aitken, H.
Leffers, S. J. Gamblin, S. J. Smerdon, L. C. Cantley, Cell 1997, 91,
961 – 971; c) X. Yang, W. H. Lee, F. Sobott, E. Papagrigoriou,
C. V. Robinson, J. G. Grossmann, M. Sundstrom, D. A. Doyle,
J. M. Elkins, Proc. Natl. Acad. Sci. USA 2006, 103, 17237 – 17242.
[7] C. H. S. Lu, H. Sun, F. B. A. Bakar, M. Uttamchandani, W. Zhou,
Y.-C. Liou, S. Q. Yao, Angew. Chem. 2008, 120, 7548 – 7551;
Angew. Chem. Int. Ed. 2008, 47, 7438 – 7441.
[8] a) L. Zhang, H. Fu, J. Biol. Chem. 1997, 272, 13717 – 13724;
b) S. C. Masters, H. Fu, J. Biol. Chem. 2001, 276, 45193 – 45200;
c) A. Nguyen, D. M. Rothman, J. Stehn, B. Imperiali, M. B.
Yaffe, Nat. Biotechnol. 2004, 22, 993 – 1000.
[9] a) J. L. Duffner, P. A. Clemons, A. N. Koehler, Curr. Opin.
Chem. Biol. 2007, 11, 74 – 82; b) M. Uttamchandani, J. Wang,
S. Q. Yao, Mol. BioSyst. 2006, 2, 58 – 68.
[10] R. Srinivasan, L. P. Tan, H. Wu, P.-Y. Yang, K. A. Kalesh, S. Q.
Yao, Org. Biomol. Chem. 2009, 7, 1821 – 1828.
[11] a) M. Uttamchandani, W. L. Lee, J. Wang, S. Q. Yao, J. Am.
Chem. Soc. 2007, 129, 13110 – 13117; b) H. Sun, C. H. S. Lu, M.
Uttamchandani, Y. Xia, Y.-C. Liou, S. Q. Yao, Angew. Chem.
2008, 120, 1722 – 1726; Angew. Chem. Int. Ed. 2008, 47, 1698 –
1702.
[12] Y. Du, S. C. Masters, F. R. Khuri, H. Fu, J. Biomol. Screening
2006, 11, 269 – 276.
[13] J. Li, S. Q. Yao, Org. Lett. 2009, 11, 405 – 408.
[14] 2–5 has good stability toward cellular proteases, but showed a
dephosphorylation t1/2 of about 6 h in mammalian cell lysates. Its
intracellular dephosphorylation lifetime will most likely be much
longer.
[15] R. Rose, S. Erdmann, S. Bovens, A. Wolf, M. Rose, S. Hennig, H.
Waldmann, C. Ottmann, Angew. Chem. 2010, 122, 4223 – 4226;
Angew. Chem. Int. Ed. 2010, 49, 4129 – 4132.
Received: May 29, 2010
Published online: August 2, 2010
[16] a) H. Shi, K. Liu, A. Xu, S. Q. Yao, Chem. Commun. 2009, 5030 –
5032; b) J. M. Astle, L. S. Simpson, Y. Huang, M. M. Reddy, R.
Wilson, S. Connell, J. Wilson, T. Kodadek, Chem. Biol. 2010, 17,
38 – 45; c) L. P. Labuda, A. Pushechnikov, M. D. Disney, ACS
Chem. Biol. 2009, 4, 299 – 307.
Keywords: combinatorial chemistry · fragment-based screening ·
microarrays · protein–protein interactions · proteins
.
[1] J. A. Wells, C. L. McLendon, Nature 2007, 450, 1001 – 1009.
6532
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 6528 –6532