Catalyst-proximity protein chemical labelling on affinity beads targeting endogenous lectins

Article abstract of DOI:10.1039/c9cc05231c

Magnetic affinity beads functionalized with lactose and ruthenium/dcbpy complexes were developed. Using MAUra, a catalyst-proximity labelling reagent, the catalytic labeling of lactose-binding proteins was achieved with high selectivity on the beads. The first unbiased identification of cellular endogenous lectins bound to lactose (galectin-1 and galectin-3) was achieved with chemical labelling on the affinity beads.

Full text of DOI:10.1039/c9cc05231c

ChemComm
View Article Online
View Journal | View Issue
COMMUNICATION
Catalyst-proximity protein chemical labelling on
affinity beads targeting endogenous lectins†
Cite this: Chem. Commun., 2019,
5, 13275
5
ab
a
c
c
Michihiko Tsushima, Sahinichi Sato, * Tatsuya Niwa, Hideki Taguchi and
Hiroyuki Nakamura
Received 8th July 2019,
Accepted 9th October 2019
*
DOI: 10.1039/c9cc05231c
rsc.li/chemcomm
Magnetic affinity beads functionalized with lactose and ruthenium/
Selective chemical labelling of ligand-binding proteins with
dcbpy complexes were developed. Using MAUra, a catalyst-proximity covalent bond formation enables the detection of proteins
labelling reagent, the catalytic labeling of lactose-binding proteins weakly bound to ligands, and photoaffinity probes have been
was achieved with high selectivity on the beads. The first unbiased developed and widely used for detecting target proteins with weak
5,6
identification of cellular endogenous lectins bound to lactose affinity. In the case of lectins, the affinity can be increased by
galectin-1 and galectin-3) was achieved with chemical labelling using carbohydrate dendrimers due to the multivalent binding
(
7
on the affinity beads.
effect between lectin oligomers and carbodydrates. Utilizing
this property, multivalent carbohydrate photoaffinity probes were
8–13
Techniques for identifying proteins bound to ligands are important developed (Fig. 1a).
For example, Pieters and co-workers
for elucidating the mechanisms underlying biological pathways achieved the labelling of targets that were added in a sufficient
and identifying the interactions between proteins and ligands. amount into cell lysate using multivalent carbohydrate photo-
1
4
Most of the studies on protein–ligand interactions focus on high affinity probes. However, the applications of these methods are
À6
affinity interactions (K
D
o 10 M). Because many conventional limited to purified lectin or protein mixture systems containing
À4
approaches fail to apply weak interactions (K
D
4 10 M), informa- artificially added lectins, and the selectivity to lectins is not
tion about weak protein–ligand complexes is still scarce. Weak and sufficient. In fact, labelling of cellular endogenous lectins has
À4
1,2
transient protein–protein interactions (K
D
4 10 M), such as not been reported so far. This limitation is generally caused by
À3
lectin–carbohydrate interactions (K = B10 M), are exploited for the low efficiency of the photoaffinity crosslinking reaction
D
cell differentiation, adhesion, and rapid turnover cell signaling, and
are one of the key factors to understating rapid responses of cellular
3
systems. Although NMR, isothermal titration calorimetry (ITC)
and surface plasmon resonance (SPR) have been applied to
analyze such weakly bound proteins, these methods require
purified target proteins. Therefore it is difficult to utilize them
for the identification of an unknown protein binding to a
4
ligand, and thus, development of technology for the detec-
tion/identification of weak protein–ligand interactions is still
an urgent requirement in molecular biology.
a
Laboratory for Chemistry and Life Science, Institute of Innovative Research,
Tokyo Institute of Technology, 4259-R1-13, Nagatsuta-cho, Midori-ku, Yokohama,
Kanagawa 226-8503, Japan. E-mail: shinichi.sato@res.titech.ac.jp,
hiro@res.titech.ac.jp
b
School of Life Science and Engineering, Tokyo Institute of Technology, Japan
c
Cell Biology Center, Institute of Innovative Research, Tokyo Institute of
Technology, 4259-S2-19, Nagatsuta-cho, Midori-ku, Yokohama,
Kanagawa 226-8503, Japan
Fig. 1 Concept of this work. (a) Photoaffinity probe containing multivalent
carbohydrate ligands. (b) Catalyst-proximity tyrosine residue labelling using
†
Electronic supplementary information (ESI) available. See DOI: 10.1039/
3
a ligand-conjugated Ru(bpy) photocatalyst. (c) This work: lectin labelling on
c9cc05231c
catalyst-functionalized affinity beads.
This journal is ©The Royal Society of Chemistry 2019
Chem. Commun., 2019, 55, 13275--13278 | 13275

View Article Online
Communication
ChemComm
1
5,16
(
less than 10%).
On the other hand, we developed a catalytic
2
+
tyrosine residue labelling technique that uses Ru(bpy)
3
as the
In this method, a tyrosine residue is oxidized
1
7,18
photocatalyst.
by single-electron transfer (SET) of a ruthenium photocatalyst
and a tyrosyl radical is generated to react with a labelling
reagent. Using a ligand-conjugated catalyst, target protein selec-
tive labelling was achieved due to the proximity property of the
SET reaction (Fig. 1b). Recently, we discovered that 1-methyl-4-
arylurazole (MAUra) efficiently labels tyrosine located a few
2
+ 19
nanometers from Ru(bpy)
3
. We also performed target protein
purification and labelling simultaneously using Ru-photocatalyst-
20
functionalized beads. We thought that MAUra could selectively
label bead-binding proteins located in close proximity to the
photocatalyst functionalized on the beads. Using this system,
weak and transient ligand binding proteins, which cannot be
pulled down by conventional affinity chromatography, can be
labelled on the beads. In this communication, we achieved
the first chemical labelling and highly sensitive detection of
endogenous carbohydrate-binding proteins through proximity-
dependent labelling using MAUra on ruthenium/dcbpy complex
functionalized beads (Fig. 1c).
To demonstrate the concept, we chose peanut agglutinin
PNA, K = 770 mM for lactose), a lectin that is derived from a lactose-conjugated photocatalyst or photocatalyst-functionalized affi-
D
Fig. 2 PNA-selective labelling in PNA-containing HeLa cell lysate using
2
1
(
Arachis hypogaea fruit and recognizes the b(1–4)-galactoside nity beads. (a) Structures of the ligand, photocatalyst, and labelling
reagents. (b) PNA-selective labelling using bead 1, bead 2 or lactose-
group, and b-D-lactose as the model target protein and ligand,
respectively. Sakurai and co-workers reported that the binding
conjugated Ru photocatalyst 4. A mixture of PNA (1 mM) and HeLa
À1
cell lysate (1.0 mg mL protein) in lysis buffer (pH 7.4) was photo-
affinity between PNA and b-D-lactose was drastically increased
irradiated (455 nm LED) in the presence of 3 (500 mM) at 0 1C for 5 min.
1
2
by the immobilization of b-D-lactose on nanoparticles. We Azide-labelled proteins were visualized by a copper-free click reaction with
À1
synthesized beads 1 and 2, on which the Ru(bpy)3 complex DBCO-Cy3. * Input: PNA (1 mM) containing HeLa cell lysate (1.0 mg mL
protein). ** PNA control (1 mM).
(bead 1) and Ru/dcbpy complex (bead 2) were immobilized,
respectively, with b-D-lactose (see the ESI,† Section S2–S9, for
details on bead preparation). In order to evaluate the labelling
efficiency and the PNA selectivity for each catalyst or the reaction mixtures more selectively than the ligand-conjugated Ru(bpy)
field on beads, PNA labelling with MAUra-azide (3) was per- complex (Fig. S4, ESI†).
3
formed in PNA-containing HeLa cell lysate using bead 1, bead 2,
or lactose-conjugated Ru photocatalyst 4. When 4 was used as desthiobiotin-conjugated MAUra (MAUra-DTB, 5, Fig. 2a).
As another type of labelling reagent, we previously reported
1
9
the catalyst, much lactose-independent labelling was observed Using the binding property of desthiobiotin to streptavidin,
lane 3, Fig. 2b). This may be due to the weak affinity of the labelled proteins can be enriched with streptavidin beads.
(
ligand and the fact that most of the ligand-conjugated catalysts Adopting the enrichment system, we evaluated the labelling
did not bind to PNA selectively. Improved labelling selectivity efficiency of PNA on bead 2. PNA was labelled by 5 and the
was observed in the case of bead 1 (lane 5, Fig. 2b). We have enriched proteins were quantified by SDS-PAGE and silver
reported that nonspecific protein binding on beads is sup- staining, compared with PNA of known concentrations. As a
pressed by changing the catalyst from the Ru(bpy) to Ru/dcbpy result, 22% labelling efficiency was achieved (Fig. S5, ESI†).
3
2
0
complex. By using bead 2, we succeeded in the selective This value is higher than the general efficiencies of conven-
labelling of PNA in the protein complex mixture (lane 7, tional photoaffinity labelling methods targeting lectins (less
2
2
15,16
Fig. 2b). PNA labelling was hardly observed in the presence than 10%).
of an excess amount of b-D-lactose, suggesting that the PNA- PNA was labelled with MAUra-N
selective labelling was lactose-dependent (lane 8, Fig. 2b and reacted with DBCO-Cy3, in-gel tryptic-digested, and analyzed
Fig. S2, ESI†). Under the conditions where other labelling by LC-MS. In a control experiment using Ru(bpy) Cl , many
In addition, in order to identify the labelled site,
3
(3), and the labelled PNA was
3
2
reagents were used instead of 3, both the labelling efficiency peaks were visible on the fluorescence spectrum because of the
and selectivity were insufficient (Fig. S3, ESI†). These results random site tyrosine labelling on PNA (Fig. S6d, ESI†). On the
suggest that the labelling radius of the radical species of MAUra is other hand, in the case of bead 2, limited peaks were observed
suitable for proximity labelling of bead-binding proteins on bead 2. on the fluorescence spectrum, suggesting that the labelling
Not only when targeting PNA but also in model experiments proceeded site-selectively. We could identify the labelled
targeting carbonic anhydrase, the ligand- and Ru/dcbpy-complex- peptide fragment by MS measurement of the labelled peak
functionalized beads labelled target proteins in protein separated by HPLC. The identified peptide fragment contains
13276 | Chem. Commun., 2019, 55, 13275--13278
This journal is ©The Royal Society of Chemistry 2019

View Article Online
ChemComm
Communication
two tyrosines (Y124 and Y129) in its sequence (Fig. S6b and e, the binding between lactose and galectins is well studied. However,
ESI†). Although we could not identify which tyrosine is labelled to our knowledge, this is the first report of lectin identification
selectively by LC-MS/MS analysis, the topological positions of both from the cellular protein mixture without any bias. Previous
Y124 and Y129 were in close proximity to the lactose binding site lectin-labelling studies were based on model experiments in
23
according to the X-ray structure of PNA (PDB: 2pel, Fig. S7, ESI†).
which purified lectins or lectin-added protein mixtures are used.
This result suggested that PNA bound to beads via lactose binding, In this communication, the combination of the multivalent
and was selectively labelled with tyrosine residue(s) around the effect on the beads and the labelling of bead-binding proteins
binding site. This result also suggested that catalyst-proximity label- enabled us to identify the binding of cellular endogenous lectins
ling on the beads enabled site-selective labelling in the ligand and lactose. Furthermore, integrin, inactive tyrosine protein
binding site of target proteins as well as target protein selective kinase, and heterogeneous nuclear ribonucleoprotein (hnRNP)
labelling in the protein mixture.
were also detected by LC-MS/MS analysis (shown in blue
Then, we performed labelling of an endogenous lactose-binding in Fig. 3b and Fig. S9–11). It was reported that these proteins
protein in A431 cell lysate using bead 2. The bead-binding protein form protein complexes via interactions between galectin-3 and
2
4–26
was labelled by MAUra-DTB (5), and the labelled protein was glycans on the protein surface.
Thus, this result suggests
enriched with streptavidin beads. The enriched protein was that not only proteins that bind to lactose but also protein–
digested with trypsin and identified by LC-MS/MS. To distinguish protein interaction partners were labelled on the beads. The
lactose-independent labelling via non-specific bead binding, we binding affinity between lactose and galectin-1 or galectin-3 is
2
7,28
performed a control experiment. In the presence of an excess weak (K
D
: millimolar order).
According to a previously
2
9
amount of free lactose, only lactose-mediated binding on beads reported procedure, we observed the inhibition of His-tagged
should be inhibited, resulting in the labelling inhibition of lactose- galectin-3 binding to asialofetuin (ASF) on an ELISA plate on
binding proteins (Fig. 3a). We obtained a list of enriched proteins incubation with monomeric lactose or bead 2. In the case of
in both experiments by LC-MS/MS analysis in the absence or monomeric lactose, the IC50 value was 3.87 mM. On the other
presence of an excess amount of free lactose. From this list, we hand, in the case of bead 2, the IC50 value was 0.398 mM,
extracted proteins that were efficiently labelled in the absence of suggesting that the binding affinity was improved by 10000-
free lactose, and whose labelling was suppressed in the presence of fold by lactose functionalization on the beads (Fig. S12, ESI†).
free lactose (see the ESI,† Table S1, for details on extraction Although the binding affinity was improved by the multivalent
conditions). Fig. 3b shows the list of extracted proteins. In this list, binding effect on the beads, the purification by the conventional
we focused on galectin-1 and galectin-3 (shown in red), which are affinity purification method was unsuccessful because the binding
proteins that bind specifically to b-galactoside. The results were affinity is not sufficient to pull down galectin-1 and -3 (Fig. S13,
also confirmed by Western blot analysis of enriched proteins using ESI†). These results indicate that the proximity labelling of bead-
anti-galectin-1 and anti-galectin-3 antibodies (Fig. S8, ESI†). The binding proteins is useful for detection and identification, and
results are reasonable because lactose is a typical b-galactoside and would be applicable to weak affinity targets that cannot be detected
by the conventional affinity purification method.
We also applied this labelling method to two-dimensional
fluorescence difference gel electrophoresis (2D-DIGE). Bead-
binding proteins were labelled by MAUra-azide (3) and the
azide-labelled proteins were visualized by copper-free click
chemistry with dibenzocyclooctyne-conjugated Cy5 (DBCO-Cy5).
To distinguish from lactose-independent labelling, we performed
the same experiment in the presence of an excess amount of free
lactose, and the labelled proteins were visualized by DBCO-Cy3.
The Cy3- and Cy5-labelled samples were mixed, and the proteins
were separated by two-dimensional electrophoresis (Fig. 4). The
use of two fluorophores facilitated the distinction between
lactose-dependent labelling and lactose-independent labelling.
Galectin-1 and galectin-3 could be detected as spots that were
selectively labelled with Cy5.
In conclusion, we have developed proximity-dependent
labelling of affinity-bead-binding proteins. We labelled lectin
in cell lysate using lactose- and ruthenium-photocatalyst-
Fig. 3 Identification of
a bead-binding protein by labelling with 5, functionalized beads. We succeeded in the selective labelling
desthiobiotin affinity enrichment, and LC-MS/MS detection. (a) Scheme
of bead-binding protein labelling in the absence and presence of excess
of PNA in PNA-containing HeLa cell lysate using the beads. The
labelling on the beads showed higher efficiency and selectivity
than when using a lactose-conjugated ruthenium photocatalyst.
We carried out the labelling of endogenous lactose-binding
proteins in A431 cell lysate. The first unbiased labelling and
À1
free ligand. Labelling was performed using beads 2 (5.0 mg mL ) and 5
À1
(
500 mM) in A431 cell lysate (3.0 mg mL ). (b) Data-extracted list of
labelled proteins in the absence of a free ligand. * Peptide spectrum
matches in LC-MS/MS of enriched proteins.
This journal is ©The Royal Society of Chemistry 2019
Chem. Commun., 2019, 55, 13275--13278 | 13277

View Article Online
Communication
ChemComm
Conflicts of interest
There are no conflicts to declare.
Notes and references
1
J. R. Perkins, I. Diboun, B. H. Dessailly, J. G. Lees and C. Orengo,
Structure, 2010, 18, 1233–1243.
2
J. Yoo, T. S. Lee, B. Choi, M. J. Shon and T. Y. Yoon, J. Am. Chem.
Soc., 2016, 138, 14238–14241.
3
4
J. Qin and A. M. Gronenborn, FEBS J., 2014, 281, 1948–1949.
A. D. Thompson, A. Dugan, J. E. Gestwicki and A. K. Mapp, ACS
Chem. Biol., 2012, 7, 1311–1320.
5
(a) G. A. Korshunova, N. V. Sumbatyan, A. N. Topin and M. T. Mtchedlidze,
Mol. Biol., 2000, 34, 823–839; (b) F. Kotzyba-Hibert, I. Kapfer and
M. Goeldner, Angew. Chem., Int. Ed. Engl., 1995, 34, 1296–1312;
(
c) G. Dorm ´a n, H. Nakamura, A. Pulsipher and G. D. Prestwich, Chem.
Rev., 2016, 116, 15284–15398.
6 I. Hamachi, T. Nagase and S. Shinkai, J. Am. Chem. Soc., 2000, 122,
Fig. 4 2D-DIGE analysis of labelled proteins. After labelling with 3
(500 mM), azide-labelled proteins were visualized by the copper-free
click reaction with DBCO-Cy5 or DBCO-Cy3. Cy5-labelled proteins are
shown in red and Cy3-labelled proteins are shown in green. Identified
lactose-binding proteins (galectin-1 and galectin-3) by LC-MS/MS analysis
as indicated by white circles. * Green bands on the left are molecular
weight makers (140 kDa and 35 kDa). ** A spot (MW B 15 kDa, pI B 4)
labelled efficiently with Cy5 could not be identified by trypsin digestion and
LC-MS/MS.
12065–12066.
7
8
S. Cecioni, A. Imberty and S. Vidal, Chem. Rev., 2015, 115, 525–561.
K. G. Rice, O. A. Weisz, T. Barthel, R. T. Lee and Y. C. Lee, J. Biol.
Chem., 1990, 265, 18429–18434.
9
1
G. Lauc, R. T. Lee, J. Dumi ´c and Y. C. Lee, Glycobiology, 2000, 10, 357–364.
0 M. R. Lee, D. W. Jung, D. Williams and I. Shin, Org. Lett., 2005, 7, 5477–5480.
1 L. Ballell, M. Van Scherpenzeel, K. Buchalova, R. M. J. Liskamp and
R. J. Pieters, Org. Biomol. Chem., 2006, 4, 4387–4394.
1
1
1
2 K. Sakurai, Y. Hatai and A. Okada, Chem. Sci., 2016, 7, 702–706.
3 A. Wibowo, E. C. Peters and L. C. Hsieh-Wilson, J. Am. Chem. Soc.,
2
014, 136, 9528–9531.
identification of cellular endogenous lectins were achieved.
Endogenous galectin-1 and -3 (10 M KD values with lactose)
1
4 T. Rungrotmongkol, P. Yotmanee, N. Nunthaboot and S. Hannongbua,
Curr. Pharm. Des., 2011, 17, 1720–1739.
À3
were labelled and identified as cellular endogenous lactose- 15 M. Hashimoto and Y. Hatanaka, Eur. J. Org. Chem., 2008, 2513–2523.
1
1
1
6 K. Sakurai, Asian J. Org. Chem., 2015, 4, 116–126.
7 S. Sato and H. Nakamura, Angew. Chem., Int. Ed., 2013, 52, 8681–8684.
8 S. Sato, K. Morita and H. Nakamura, Bioconjugate Chem., 2015, 26,
250–256.
binding proteins by LC-MS/MS and Western blot. Furthermore,
the protein–protein interaction partners of galectin-3 were also
labelled and identified. These results suggest that the proximity
labelling of bead-binding proteins enables the detection
of ligand-binding proteins with weak affinity. Not only the
direct ligand-binding proteins, but also protein complex con-
taining ligand-binding proteins could be labelled on the bead
surface. This technique can be used to identify ligand-binding
proteins that are difficult to analyze by other conventional
affinity purification methods. Applications are not limited to
the identification of carbohydrate-binding proteins. By chan-
1
2
2
2
9 S. Sato, K. Hatano, M. Tsushima and H. Nakamura, Chem. Commun.,
2018, 54, 5871–5874.
0 M. Tsushima, S. Sato and H. Nakamura, Chem. Commun., 2017, 53,
4838–4841.
1 K. J. Neurohr, D. R. Bundle, N. M. Young and H. H. Mantsch, Eur.
J. Biochem., 1982, 310, 305–310.
2 The amount of immobilized compound on the beads used in Fig. 1,
À1
lane 7 was lactose: 3.8 nmol mg
2
beads, Ru/dcbpy complex:
À1
0.0 nmol mg beads. See Fig. S1 (ESI†).
23 R. Banerjee, K. Das, R. Ravishankar, K. Suguna, A. Surolia and
M. Vijayan, J. Mol. Biol., 1996, 259, 281–296.
ging the ligand part, various protein–ligand interactions can be 24 E. H. Yang, J. Rode, M. A. Howlader, M. Eckermann, J. T. Santos,
D. Hernandez Armada, R. Zheng, C. Zou and C. W. Cairo, PLoS One,
identified. We are now in a position to identify ligand-binding
proteins with weak affinity by this technique.
M. T. gratefully acknowledges the financial support from
JSPS (ID No. 18J21621). This work was partially supported
by Grants-in-Aid for ‘‘Grant-in-Aid for Young Scientists (A)
2017, 12, 1–17.
2
2
2
2
2
5 S. Jaiswal and K. K. Srivastava, Biochem. Biophys. Res. Commun.,
2018, 498, 884–890.
6 K. Fritsch, M. Mernberger, A. Nist, T. Stiewe, A. Brehm and R. Jacob,
BMC Cancer, 2016, 16, 1–10.
7 M. C. Miller, I. V. Nesmelova, D. Platt, A. Klyosov and K. H. Mayo,
Biochem. J., 2009, 421, 211–221.
8 J. Seetharaman, A. Kfanigsberg, R. Slaaby, H. Leffler, S. H. Barondes
and J. M. Rini, J. Biol. Chem., 1998, 273, 13047–13052.
9 D. Laaf, P. Bojarov ´a , H. Pelantov ´a , V. K ˇr en and L. Elling, Bioconjugate
Chem., 2017, 28, 2832–2840.
(15H05490 to S. Sato)’’, ‘‘Homeostatic regulation by various
types of cell death (15H01372 to S. Sato)’’, and ‘‘Chemistry for
Multimolecular Crowding Biosystems (18H04542 to H. Nakamura)’’
from MEXT, Japan.
13278 | Chem. Commun., 2019, 55, 13275--13278
This journal is ©The Royal Society of Chemistry 2019