Profiling human Src homology 2 (SH2) domain proteins and ligand discovery using a peptide-hybrid small molecule microarray

Article abstract of DOI:10.1039/c3cc45413d

A 396-member peptide-hybrid small molecule microarray was fabricated to profile small molecule ligands across 15 SH2 proteins, revealing hits against Lck and Grb2. The Royal Society of Chemistry 2013.

Full text of DOI:10.1039/c3cc45413d

ChemComm
View Article Online
View Journal | View Issue
COMMUNICATION
Profiling human Src homology 2 (SH2) domain
proteins and ligand discovery using a peptide-hybrid
small molecule microarray†
Cite this: Chem. Commun., 2013,
49, 9660
Received 17th July 2013,
Jiaqi Fu,a Zhenkun Na,a Mahesh Uttamchandani*ab and Shao Q. Yao*a
Accepted 21st August 2013
DOI: 10.1039/c3cc45413d
www.rsc.org/chemcomm
A 396-member peptide-hybrid small molecule microarray was
fabricated to profile small molecule ligands across 15 SH2 proteins,
revealing hits against Lck and Grb2.
Src homology 2 (SH2) domains are readers of tyrosine kinase
phosphorylation. By binding phosphorylated target sequences, these
domains relay signals downstream to effect cellular regulation and
gene expression control.1 As a result, they form critical components
of cell signalling networks. Over 110 human proteins contain SH2
domains, including kinases, phosphatases, transcription factors and
adaptor proteins.1b,c These proteins are over-expressed in various
diseases, including cancer, and hence represent key targets for
modulation using therapeutics.1 Through in vitro and in silico
screens, various peptide and small molecule inhibitors have been
identified against SH2 domains.2 Herein, we present a small
molecule microarray (SMM) strategy for rapid fingerprinting
and profiling of SH2 proteins, which enables rapid molecular
characterization and comparative inhibitor profiling.
Various methods have been applied to map the interaction
preference of SH2 domains,3 particularly more recently through the
use of microarrays.4 For example, Huang et al. have mapped the
binding motifs for 76 SH2 domains against phosphopeptides using
SPOT synthesis/macroarrays.4b Using this map, SH2 domains have
been classified into three groups (I, II, III). While the binding
specificity of SH2 domains is attributed largely to the C-terminal
segment of the phosphorylated targets, residues at the N-terminus
also contribute to the overall binding potency.3,4b
Fig. 1 Overall strategy of the screening SH2 domain containing proteins against
a ligand microarray for high-throughput profiling.
class of proteins. Microarrays are particularly advantageous in that
they enable screening of protein–small molecule interactions
en masse, using just minimal quantities of both targets and analytes.
We designed and fabricated a 396-member library of small mole-
cules containing a centrally located phosphotyrosine flanked by 9
N-terminal dipeptides and 44 C-terminal small molecule building
blocks (Fig. 1). The objectives of this strategy were to (1) significantly
expand the specific binding preference of SH2 C-terminal pockets
from traditional peptide-centric approaches to small molecule-
centric strategies, (2) profile the binding spectra across multiple
SH2 domains and identify selective small molecule building blocks
which can bind the C-terminal pocket, and (3) identify complemen-
tary N-terminal sequences that could enhance binding efficacy and
selectivity.
The general strategy is shown in Fig. 1. A panel of SH2 domain
proteins were individually expressed and purified using a one-step
labelling and purification approach. The library was synthesized
using 9 different N-terminal sequences from known SH2 binding
targets4b and 44 commercially available amine building blocks
whose structures cover aromatic, aliphatic, cyclic and bicyclic com-
pounds (X1–44; Fig. 2). A biotin-Gly-Gly linker was added to facilitate
compound immobilization onto avidin-functionalized glass surfaces.
Synthesis was carried out in solid phase using the PL-FMP resin and
Irori Microkant reactors (Fig. 2B and Scheme S1, ESI†). Following
acidic cleavage and ether precipitation, representative compounds
In order to further examine the binding preferences of various
SH2 domains and facilitate inhibitor discovery, we have built a novel
phosphopeptide-hybrid small molecule microarray to investigate this
a Department of Chemistry, National University of Singapore, 3 Science Drive 3,
Singapore 117543. E-mail: chmyaosq@nus.edu.sg; Fax: +65 67791691;
Tel: +65 65162925
b Department of Biological Sciences, National University of Singapore, and Defence
Medical and Environmental Research Institute, DSO National Laboratories,
27 Medical Drive, Singapore 117510. E-mail: mahesh@dso.org.sg;
Fax: +65 64857033; Tel: +65 64857206
† Electronic supplementary information (ESI) available: Synthesis scheme, character-
ization of hits, microarray and thermal shift results. See DOI: 10.1039/c3cc45413d
c
9660 Chem. Commun., 2013, 49, 9660--9662
This journal is The Royal Society of Chemistry 2013

View Article Online
Communication
ChemComm
Fig. 2 Design and synthesis of the 396-member SH2 ligand library. (A) C-terminal substituents comprising of 44 different amine building blocks. Highlighted in red
are the C-terminal structures from the 14 hits identified using the SMM screening results. (B) Scheme of SH2 library synthesis. (a) (i) amine building block/DCE, 2 h;
(ii) 1% HOAc/Na(OAc)3BH, 16 h; (b) (i) Fmoc-Tyr[PO(OBzl)OH]-OH/PyBrOP/DIEA, 12 h; (c) (i) 20% piperidine/DMF 1 h; (ii) Fmoc-AA-OH/HOBt/HBTU/DIEA/DMF, 12 h;
(d) TFA/TIS/H2O (95 : 2.5 : 2.5). (C) Original peptide sequence motifs for the selected 9 SH2 N-terminal substituents (in blue).
were dissolved in DMSO and run on an LC-MS system, to ensure that
purity was >80%, and the MS of the final product was correct.
Microarrays were spotted with the 396-member library in
duplicate. To confirm immobilization, slides were stained with a
phosphospecific binding dye, ProQ Diamonds. All spots displayed a
positive signal, confirming immobilization of the library across
the slides (Fig. S3B, ESI†). Proteins were expressed in E. coli with
hexa-histidine tags, and were labelled on beads with Cy5-NHS (GE
Healthcare), before being eluted with imidazole and size separated
(using a G-25 Nap column, GE Healthcare), facilitating labelling and
purification in a combined step. A total of 15 fluorescently labelled
SH2 proteins were prepared in this manner (Fig. S1, ESI†), and
screened against this small molecule library on microarrays. 10 mM
concentration of the protein was applied to each slide. Following a
washing step by immersion in PBS containing 0.1% Tween 20 (with
shaking), the fluorescence-scanned images were obtained (raw images
are shown in ESI,† Fig. S3C). The microarray data were extracted,
background subtracted, normalized, and displayed in a coloured heat
map (Fig. 3A). We further confirmed that when the arrays were
dephosphorylated through alkaline phosphatase treatment, no
binding was observed. This highlighted that binding to the SMM
was dependent on SH2 domain interactions, as well as on the
presence of a phosphorylated Tyr residue (data not shown).
The complete dataset is shown in Fig. 3A as a coloured heat map.
Each of the 15 SH2 domains exhibited characteristic fingerprints.
Based on these results, several broad conclusions could be drawn.
Fig. 3 Fingerprints of SH2 domains using SMMs. (A) Coloured heat maps
Generally it appeared that positions with N-terminal PP, VV or PI
sequences generated unique hits, more so than any other N-terminal
sequence. Also, the heat map showed that the class II SH2 domain
containing proteins exhibited a binding pattern that was rather
distinctive. As the dataset was extensive, we further simplified the
results for analysis by first averaging the complete microarray dataset
formed by contributions of signals from the respective N-terminal
and C-terminal substituents (Fig. 3B and Fig. S5, ESI†). When this
analysis was performed by grouping the SH2 domains into their
displaying intensities with each of the ligand (scale inset). Src (a), Lck (b), Abl1
(c), Abl2 (d), Sh2d1a (e), Ship2 (f), Crk (g), Brk (h), Bcar3 (j), Bmx (i), Grb2 (k),
Sh3bp2 (l), Shc1 (m), Shb (n), Socs2 (o). (B) Microarray results averaged for class I
(top) and class II (bottom) SH2 proteins. N-terminal substituents shown in blue
bars, while C-terminal substituents shown in red bars. (C) SMM assisted quanti-
tative KD determination of 4 hits against Lck. (D) Structure of the representative
hit, PP-pY-X25. (E) Apparent KD and protein thermal shift assay results.
classes (12 proteins in class I and 3 proteins in class II, Fig. 1), it was
revealed that a common set of C-terminal substituents for class I
c
This journal is The Royal Society of Chemistry 2013
Chem. Commun., 2013, 49, 9660--9662 9661

View Article Online
ChemComm
Communication
proteins, namely 12, 25, 36 and 37, which were long chain ring was proportional to the binding potency of the hits. For instance,
containing compounds, while 1, 5, 6 and 10 which contained cyclic PI-pY-X1 produced a small thermal shift (0.3 1C) while the stronger
rings linked proximately to amine displayed good affinity across the Lck binding hits PP-pY-X5, PP-pY-X6 and PP-pY-X25 contributed to
proteins tested. Compared to those of class I, class II proteins larger shifts (B3 1C). The results indicated that the latter three hits
displayed a distinct preference for binding to aromatic groups with were true binders. Similarly, the thermal shift results of Grb2
short chains such as 16, 17, 22, and 31. This is consistent with a matched that expected from the microarray results. Among the hits
previous study claiming that class II SH2 domains prefer a hydro- tested, PP-pY-X5 and PP-pY-X25 and PI-pY-X1 presented as hits for
phobic residue at the P1 position.4b
Grb2. Taken together, PP-pY-X5 and PP-pY-X25 broadly bound Grb2
In addition, these averaged results also showed that the and Lck, while PI-pY-X1 bound Grb2, but only weakly bound Lck
N-terminal positions also contributed significantly to the binding, as (Fig. S7, ESI†). Future docking or co-crystallization experiments
evidenced by the sizeable error bars (Fig. 3B and Fig. S5, ESI†). Across could confirm the binding pockets and orientation of these hits.
class I SH2 proteins, PP, PI or VV sequences had a clearly higher
In conclusion, we have successfully fabricated a small molecule
contribution to overall binding, compared to the other 6 N-terminal microarray platform which enables, for the first time, convenient
sequences tested. For class II SH2 domains, there was a broader profiling of SH2 domains with small molecule building blocks. In
preference with proline containing P
members contributing the present format, the binding profiles were sufficiently distinct to
À1
towards the strongest relative binding. This result was consistent produce informative fingerprints that clustered the various SH2
with references describing favourable contribution of hydrophobic domains according to their expected functional classes. In addition,
residues on the N-terminus to overall inhibitor potencies.3a,4b
we have identified putative small molecule hits against Grb2 and
To further resolve the groups according to binding prefer- Lck. These peptide-hybrid compounds, though not explicitly
ence, hierarchical clustering was carried out. This, unlike demonstrated herein, might in future be converted into cell-
sequence homology clustering usually performed,1b categorizes permeable small molecule protein–protein interaction (PPI)
SH2 domains according to their binding preferences against inhibitors.7 It will be possible to monitor slight differences in
ligand libraries.5 Our clusters generally grouped class I and class binding using a multi-coloured labelling and application strategy.8
II proteins separately (Fig. S4, ESI†). Class I SH2 domains all Overall this SMM strategy offers an attractive approach for perform-
contained a Tyr or a Phe aromatic residue at bD5, which has ing global selectivity screens of SH2 domains against a much larger
been shown to influence binding at P1 and P3 positions.3a These set of diversity elements.
included namely Src, Abl1 and Lck. Lck, Src, Abl1 and Abl2 from
The authors acknowledge funding support from the Minis-
group I share a common pY-E(L/E)I binding preference,4b which try of Education (R-143-000-517-112). Dr Hao Wu, Ms Madelin
can account for the similarity observed. Shb, Socs2 and Shc1 Teo and Ms Kaijia Peh assisted with the library synthesis and
were grouped most distantly from the majority of other group 1 initial screens.
SH2s, as might have been expected. In particular, Socs2 and
Ship2 formed a sister pair indicating a highly similar binding
Notes and references
profile, as did Grb2 and Sh2d1a, highlighting the clustering 1 (a) M. B. Yaffe, Nat. Rev. Mol. Cell Biol., 2002, 3, 177–184; (b) B. A. Liu,
´
K. Jablonowski, M. Raina, M. Arce, T. Pawson and P. D. Nash,
based on the microarray fingerprints correlated closely with
traditional substrate grouping across these SH2 domains.4b
Mol. Cell, 2006, 22, 851–868; (c) K. Machida and B. J. Mayer, Biochim.
Biophys. Acta, 2005, 1747, 1–25.
In addition to fingerprinting, we also identified several 2 (a) D. Kraskouskaya, E. Duodu, C. C. Arpin and P. T. Gunning, Chem.
Soc. Rev., 2013, 42, 3337–3370; (b) B. A. Liu, B. W. Engelmann and
P. D. Nash, Proteomics, 2012, 12, 1527–1546.
3 (a) Z. Songyang, S. E. Shoelson, M. Chaudhuri, G. Gish, T. Pawson,
strongly binding molecules that appeared to be potential hits.
They showed significant levels of fluorescence on the microarray
screens, across various SH2 proteins. A total of 14 such hits were
identified for follow-up validation and tests. These hits were
resynthesized, purified and LC-MS analysis was conducted, to
confirm purity and identity (Fig. S2, ESI†). These molecules were
re-spotted on smaller sub-grids of avidin slides, and subjected to
concentration-dependent applications of various SH2 proteins.
W. G. Haser, F. King, T. Roberts, S. Ratnofsky and R. J. Lechleider,
et al., Cell, 1993, 72, 767–778; (b) M. J. Eck, S. E. Shoelson and S. C.
Harrison, Nature, 1993, 362, 87–91; (c) Z. Songyang, S. E. Shoelson,
J. Mcglade, P. Olivier, T. Pawson, X. R. Bustelo, M. Barbacid, H. Sabe,
H. Hanafusa, T. Yi, R. Ren, D. Baltimore, S. Ratnofsky, R. A. Feldman
and L. C. Cantley, Mol. Cell. Biol., 1994, 14, 2777–2785; (d) M. L.
Miller, S. Hanke, A. M. Hinsby, C. Friis, S. Brunak, M. Mann and N. Blom,
Mol. Cell. Proteomics, 2008, 7, 181–192.
As a result, apparent KD values were computed for these hits 4 (a) R. B. Jones, A. Gordus, J. A. Krall and G. MacBeath, Nature, 2006,
439, 168–174; (b) H. Huang, L. Li, C. Wu, D. Schibli, K. Colwill, S. Ma,
C. Li, P. Roy, K. Ho, Z. Songyang, T. Pawson, Y. Gao and S. S. C. Li,
Mol. Cell. Proteomics, 2008, 7, 768–784; (c) A. Gordus and
against the various SH2 domains, as shown against Lck (Fig. 3C
and Fig. S6, ESI†). Based on these results, PP-pY-X5 and PP-pY-X25
were identified as hits that bound to both Lck and Grb2 and
PI-pY-X1 appeared to be a potentially selective hit for Grb2.
Further validation was carried out by a protein thermal shift
assay using thermal shift measurements for both Lck and Grb2,
using methods previously described.6 Lck without a ligand acted as
a negative control, with the melt temperature of about 55.4 1C. After
adding the ligand, the melting point shift increased due to the
stability of the protein–ligand complex. Compared to the melting
point of the negative control, all ligand–protein complexes
presented a right shift ranging from 55.7 to 58.6 1C. The shift range
G. MacBeath, J. Am. Chem. Soc., 2006, 128, 13668–13669; (d) For a
recent review, see: H. Wu, J. Ge, M. Uttamchandani and S. Q. Yao,
Chem. Commun., 2011, 47, 5664–5670.
5 M. Uttamchandani, J. Wang, J. Li, M. Hu, H. Sun, K. Y. T. Chen, K. Liu
and S. Q. Yao, J. Am. Chem. Soc., 2007, 129, 7848–7858.
6 N. D. Ambaye, M. J. Gunzburg, R. C. C. Lim, J. T. Price, M. C. J. Wilce
and J. A. Wilce, Bioorg. Med. Chem., 2011, 19, 693–701.
7 H. Wu, J. Ge and S. Q. Yao, Angew. Chem., Int. Ed., 2010, 49,
6528–6532.
8 (a) L. Gao, M. Uttamchandani and S. Q. Yao, Chem. Commun., 2012,
48, 2240–2242; (b) C. H. S. Lu, H. Sun, F. B. Abu Bakar,
M. Uttamchandani, W. Zhou, Y. Liou and S. Q. Yao, Angew. Chem.,
Int. Ed., 2008, 47, 7438–7441.
c
9662 Chem. Commun., 2013, 49, 9660--9662
This journal is The Royal Society of Chemistry 2013