A FlAsH-based cross-linker to study protein interactions in living cells

Article abstract of DOI:10.1002/anie.201106404

As you like it: xCrAsH, a dimeric derivative of the arsenical compound FlAsH, enables the highly specific, covalent cross-linking of two proteins containing a 12 amino acid peptide tag. This inducible and (by addition of dithiols) reversible system can be used to detect and manipulate protein-protein interactions both in vitro and in living cells (see picture). Copyright

Full text of DOI:10.1002/anie.201106404

DOI: 10.1002/anie.201106404
Protein Cross-Linking
A FlAsH-Based Cross-Linker to Study Protein Interactions in Living
Cells**
Anna Rutkowska, Christian H. Haering,* and Carsten Schultz*
The dynamics of protein–protein interac-
tions play a central role in various biolog-
ical processes including the function of
multi-protein complexes and the regulation
of signal transduction networks. Methods to
sense or manipulate such interactions are
therefore crucial for studying protein
behavior in living cells. Protein–protein
contacts can be enforced, for example, by
small cell-membrane-permeable molecules
which simultaneously bind to two specific
protein domains.[1] While extremely useful
for a range of applications, the necessity to
fuse the proteins of interest to these bulky
extraneous domains (usually larger than
18 kDa) often affects the proteinsꢀ activities
and/or localization.[2] Moreover, the inter-
actions generated between the additional
domains are frequently irreversible not
allowing the induction of transient linkages.
To circumvent these restrictions, we devel-
Figure 1. The cross-linker xCrAsH. a) Covalent cross-linking with xCrAsH of two interacting
proteins containing internally located or terminal 4cys-tags. The cross-linked complex is
fluorescent and can be released by addition of dithiols. b) Chemical structures of the CrAsH
monomer (5(6)-carboxy-FlAsH) and the xCrAsH cross-linker. In contrast to CrAsH, the
oped a method to perform highly specific
cross-linking of two proteins in living cells,
each containing only a 12 amino acid pep-
tide tag (4cys-tag). The novel cross-linking
xCrAsH molecule is acetylated, which improves its solubility, but leads to significant
approach is based on the formation of
stable, covalent complexes between
dimeric biarsenic derivative of carboxy-
fluorescein (xCrAsH) and two proteins
reduction of its fluorogenic properties in vitro. Acetates are, however, efficiently removed by
intracellular esterases. For analytical purposes, xCrAsH was also synthesized with two 5-
carboxyfluorescein moieties.
a
containing the unique tetracysteine sequence motif
CCPGCC (Figure 1a).[3] In contrast to other methods, the
reaction is easily reversible by addition of membrane-
permeable dithiols. As an additional feature, the protein
pairs become fluorescently labeled upon binding of the cross-
linker, which helps to detect and follow target proteins in real
time by light microscopy.
Biarsenic probes in combination with a split tetracysteine
tag have already been used to connect two parts of a protein
structure or for monitoring the oligomeric state of proteins.[4]
However, application of this approach is limited because the
split motives need to be in very close proximity for sufficient
cross-linking to occur. Thus, to achieve stable dimerization as
well as significant flexibility of the cross-linking approach, we
synthetically connected two biarsenic molecules with a
flexible linker, to give xCrAsH, (Figure 1b) in an overall
yield of 53% starting from carboxyfluorescein (see Support-
ing Information for details). In addition, the cross-linker
molecule was acetylated to increase solubility during syn-
thesis and to ensure permeability in cell applications. The
single biarsenic molecule carboxy-FlAsH (CrAsH, Figure 1b)
was used throughout this work as an internal control. To
maintain high specificity of the cross-linking reaction, an
optimized 12 amino acid sequence (FLNCCPGCCMEP, 4cys-
tag) was fused to the proteins of interest.[5]
[*] Dr. A. Rutkowska, Dr. C. H. Haering, Priv.-Doz. Dr. C. Schultz
Cell Biology & Biophysics Unit, European Molecular Biology
Laboratory
Meyerhofstrasse 1, 69117 Heidelberg (Germany)
E-mail: chaering@embl.de
schultz@embl.de
[**] We thank the Advanced Light Microscopy Facility at EMBL, Heike
Stichnoth for providing cells, Andrea Erlbruch and Alen Piljic for
providing initial PKA, DAPK1, and calmodulin constructs, Birgit
Koch for providing the anti-GFP antibody, Sarah Sternberger for help
during experiments with purified protein, and Stephen R. Adams
(UCSD) for discussions. A.R. is a fellow of the Peter und Traudl
Engelhorn-Stiftung and the EMBL Interdisciplinary Postdoc Pro-
gramme (EIPOD). FlAsH=fluorescein-based arsenical hairpin
binder.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201106404.
Angew. Chem. Int. Ed. 2011, 50, 12655 –12658
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
12655

Communications
As a model protein pair for cross-linking, we first chose
tion, Figure S1). Thus, these results demonstrate that xCrAsH
can specifically, efficiently, and reversibly cross-link two
proteins containing 4cys-tags.
FKBP (FK506-binding protein) and FRB (FKBP-rapamycin
binding domain of mTor), two proteins that only associate
with each other upon binding to the small-molecule rapamy-
cin.[6] Based on the X-ray crystal structure of the FRB–FKBP
complex,[6b] we decided to introduce the 4cys-tags at the
C termini of both proteins.
We initially performed cross-linking with purified His6-
FKBP-4cys and His6-SUMO-FRB-4cys proteins under stan-
dard conditions reported for protein labeling with biarsenic
compounds.[3a] The addition of xCrAsH but not of the CrAsH
monomer led to the predominant formation of a cross-linked
FRB–FKBP complex in the presence of rapamycin (Fig-
ure 2a). In the absence of rapamycin, the FRB–FKBP
We then tested cross-linking in lysates of HeLa Kyoto
cells expressing FLAG3-FKBP-4cys and ECFP-FRB-4cys
(Figure 2b). In this system, we again achieved specific cross-
linking in the presence of xCrAsH, with a yield of up to 70%
as determined by quantitative western blotting (Supporting
Information, Figure S1). Importantly, only the formation of
rapamycin-dependent FKBP–FRB heterodimers, but not of
homodimers, was detected, which demonstrates that a stable
protein–protein interaction is a prerequisite for efficient
xCrAsH cross-linking in cell lysates. Hence, xCrAsH can be
used in cell lysates as a highly specific conditional cross-linker
that does not artificially link non-interacting
proteins under physiological conditions, even if
the proteins contain the tetracysteine
sequence.
We investigated cross-linking of 4cys-
tagged proteins in living cells. Owing to the
fluorogenic properties of biarsenic compounds,
live-cell fluorescence microscopy is a straight-
forward method to test permeability and to
estimate optimal concentrations of these mol-
ecules for a given cell type.[7] After incubation
of HeLa or U2OS cells transiently expressing
mRFP-FKBP-4cys with CrAsH or xCrAsH
according to a standard method for in vivo
labeling with biarsenic derivatives,[7] we
observed specific fluorescent staining of cells
expressing 4cys-tagged proteins, which was
rapidly reverted by the addition of dithiols
(Figure 3a and Supporting Information Fig-
ure S2). Thus, xCrAsH can efficiently cross the
membranes of human cells and specifically
binds to 4cys-tagged proteins.
To probe whether xCrAsH efficiently
cross-links two tagged proteins, living U2OS
(or HeLa) cells transiently expressing FLAG3-
FKBP-4cys and ECFP-FRB-4cys were incu-
bated with micromolar concentrations of
xCrAsH or CrAsH for 2 h in the presence or
absence of rapamycin. We then measured the
efficiency of cross-linking by western blotting
(Figure 3b, some data not shown). As with cell
lysates, the only detectable cross-linked prod-
uct was the FRB–FKBP heterodimer in cells
Figure 2. In vitro cross-linking of a model 4cys-tagged protein pair. a) Purified proteins
(His6-FKBP-4cys and His6-SUMO-FRB-4cys, 15 mm each) preincubated with or without
rapamycin were incubated with CrAsH (30 mm) or xCrAsH (30 mm) at 378C for 1 h.
Samples were analyzed by SDS-PAGE and Coomassie staining or in-gel fluorescence
scan. Bands representing cross-linked products were visible only in the presence of
xCrAsH. Complexes of protein monomers and xCrAsH or CrAsH are indicated by (*).
b) Western blot of a similar cross-linking experiment using lysates from HeLa Kyoto cells
transiently co-expressing FLAG3-FKBP-4cys and ECFP-FRB-4cys; final concentrations:
rapamycin (40 mm), xCrAsH and CrAsH (5 mm each). Cross-linked products were
observed only between interacting proteins and were reverted by addition of 5 mm BAL.
heterodimers as well as FRB–FRB and FKBP–FKBP homo-
dimers were detected in approximately equal stoichiometry.
As expected, the cross-linked complexes were fluorescent,
stable even to denaturing conditions of SDS-PAGE (poly-
acrylamide gel electrophoresis), and sensitive to high con-
centrations of dithiols (5 mm 2,3-dimercapto-1-propanol,
BAL; Figure 2a and Supporting Information Figure S1).
The yield of cross-linking after optimization of the assay
conditions was 80% as judged by quantification using Deep
Purple staining (data not shown). The optimal ratio of cross-
linker to protein was between 1:1 and 2:1, which is in the
range of the expected theoretical value (Supporting Informa-
that had been incubated with rapamycin prior to cross-linker
addition. Cross-linking under these conditions reached a yield
of up to 50% at optimal cross-linker concentrations of 2–5 mm
for U2OS and 10–20 mm for HeLa cells, and was significantly
reduced when cells were subsequently treated with BAL
(data not shown). xCrAsH can therefore be applied as an
inducible, conditional, and reversible cross-linker in living
cells.
To test whether xCrAsH can be used as a general tool to
sense protein–protein interactions and to test their depend-
ence on ligands or stimuli, we applied the cross-linker to
monitor the activity state of two central heterodimeric
12656 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 12655 –12658

Figure 4. Probing protein–protein interactions of kinases in vitro and
in living cells. a) Western blots of in vitro cross-linking experiment
using lysates from U2OS cell transiently co-expressing 4cys-tag fusions
of regulatory and catalytic subunits of PKA (PKAreg-4cys-ECFP and
4cys-PKAcat-mCherry) which were incubated with DMSO, CrAsH
(5 mm), or xCrAsH (5 mm) at 378C for 1 h. In some cases 8-Br-cAMP
(5 mm) was added either 10 min before (B) or 50 min after (A) the
addition of xCrAsH. The holoenzyme complex between PKAcat and
PKAreg was detected only at low cAMP levels. b) U2OS cells incubated
with CrAsH or xCrAsH (2 mm each) for 2 h at 378C. Cells were
afterwards lyzed and analyzed by western blotting. The aberrant gel
running behavior of cross-links containing PKAreg is due to uncom-
pleted denaturation of ECFP (Supporting Information, Figure S3).
Figure 3. Cross-linking in living cells. a) Membrane permeability and
specific labeling with xCrAsH: confocal images of HeLa cells transi-
ently expressing mRFP-FKBP-4cys prior to (left) and after (middle)
labeling with 1 mm xCrAsH for 1 h; CrAsH, excitation 514 nm, emission
530–600 nm. Addition of 5 mm BAL completely reversed binding of
xCrAsH (right). For quantification see Supporting Information. Scale
bar, 20 mm. b) Western blots after in-cell cross-linking: U2OS cells
transiently co-expressing FLAG3-FKBP-4cys and ECFP-FRB-4cys were
preincubated with or without rapamycin (10 mm) and subsequently
incubated with xCrAsH (2 mm, 2 h).
nificant cross-linking of DAPK1-calmodulin complexes in
lysates of U2OS cells transiently expressing DAPK1-4cys-
ECFP (catalytic domain, 334 amino acid residues) and
FLAG3-4cys-calmodulin only in the presence of calcium
ions (Supporting Information, Figure S4). Addition of a
calcium chelating agent (ethylene glycol tetraacetic acid,
EGTA) almost completely prevented protein–protein cross-
linking. Thus, we successfully employed the xCrAsH cross-
linker to validate the protein–protein interaction status of two
kinases in cell lysates. Importantly, future optimization of the
positions of the 4cys-tags may further increase the final yield
of cross-linking between the described protein pairs.
In summary, we established the novel membrane-perme-
able chemical cross-linker xCrAsH as a tool for studying
protein–protein interactions in vitro and in living cells. Three
important features distinguish this technique from other
approaches: reversibility (by addition of dithiols), condition-
ality (no forced dimerization), and the use of only a very small
peptide tag (4cys-tag). In contrast to the larger domains
required for alternative cross-linking methods, the 4cys-tag
mimics a protein loop, which significantly reduces the like-
lihood of disrupting protein function, even when inserted as
an internal fragment.[2c,4d,10] To increase the spectrum of
applications, bifunctional cross-linkers based on the combi-
nation of one biarsenic moiety with another autoreactive
functional group (e.g. benzylguanine[1d] or trimethoprim[1b]
derivatives) could be developed. Such molecules will be
especially important for studying multi-protein complexes,
such as PKA, in which homo- and heterodimers may need to
kinases, namely protein kinase A (PKA) and death-associ-
ated protein kinase 1 (DAPK1). In its non-activated state,
PKA is a homodimer of regulatory subunits (PKAreg)
attached to two catalytic subunits (PKAcat).[8] Upon cooper-
ative binding of four cyclic AMP (cAMP) molecules, the
catalytic subunits are released and thereby activated. We
would hence expect that xCrAsH is able to cross-link the
catalytic subunit to the regulatory subunits before, but not
after, activation of PKA by cAMP. Cross-linking reactions
were performed in lysates of U2OS cells transiently express-
ing PKAreg-4cys-ECFP and 4cys-PKAcat-mCherry as de-
scribed above. Without stimulation, heterodimers of regula-
tory and catalytic subunits as well as both homodimers were
detected (Figure 4a and Supporting Information Figure S3).
The addition of a cAMP analogue (8-Br-cAMP) to lysates
prior to, but not after, the cross-linking reaction significantly
reduced the amount of detected heterodimer, as predicted. In
addition, PKA subunits were also efficiently cross-linked with
xCrAsH in living cells (Figure 4b).
The binding of DAPK1 to its activator calmodulin
depends both on the dephosphorylation of DAPK1 and on
intracellular calcium levels.[9] Accordingly, we detected sig-
Angew. Chem. Int. Ed. 2011, 50, 12655 –12658
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 12657

Communications
[2] a) B. N. Giepmans, S. R. Adams, M. H. Ellisman, R. Y. Tsien,
Science 2006, 312, 217 – 224; b) C. Hoffmann, G. Gaietta, M.
Bunemann, S. R. Adams, S. Oberdorff-Maass, B. Behr, J. P.
Vilardaga, R. Y. Tsien, M. H. Ellisman, M. J. Lohse, Nat.
Methods 2005, 2, 171 – 176; c) S. Chaumont, B. S. Khakh, Proc.
Natl. Acad. Sci. USA 2008, 105, 12063 – 12068.
[3] a) S. R. Adams, R. E. Campbell, L. A. Gross, B. R. Martin, G. K.
Walkup, Y. Yao, J. Llopis, R. Y. Tsien, J. Am. Chem. Soc. 2002,
124, 6063 – 6076; b) B. A. Griffin, S. R. Adams, R. Y. Tsien,
Science 1998, 281, 269 – 272.
[4] a) B. Krishnan, L. M. Gierasch, Chem. Biol. 2008, 15, 1104 –
1115; b) N. W. Luedtke, R. J. Dexter, D. B. Fried, A. Schepartz,
Nat. Chem. Biol. 2007, 3, 779 – 784; c) S. Ray-Saha, A. Schepartz,
ChemBioChem 2010, 11, 2089 – 2091; d) T. M. Webber, A. C.
Allen, W. K. Ma, R. G. Molloy, C. N. Kettelkamp, C. A. Dow,
M. J. Gage, Biochem. Biophys. Res. Commun. 2009, 384, 66 – 70.
[5] B. R. Martin, B. N. Giepmans, S. R. Adams, R. Y. Tsien, Nat.
Biotechnol. 2005, 23, 1308 – 1314.
[6] a) L. A. Banaszynski, C. W. Liu, T. J. Wandless, J. Am. Chem.
Soc. 2005, 127, 4715 – 4721; b) J. Choi, J. Chen, S. L. Schreiber, J.
Clardy, Science 1996, 273, 239 – 242; c) S. N. Ho, S. R. Biggar,
D. M. Spencer, S. L. Schreiber, G. R. Crabtree, Nature 1996, 382,
822 – 826.
be separated from each other. Another beneficial aspect
could be the introduction of two different fluorogenic
biarsenic dyes to enable more precise live cell imaging of
cross-linked proteins. Finally, a photo-activatable cleavage
site may be incorporated into the linker region to allow the
controlled re-opening of the cross-linked interface with
cellular or subcellular spatial resolution. Thus, the cross-
linking technique presented herein is a first step towards the
new generation of chemical inducers of dimerization, which
may be generally applied to probe and manipulate protein–
protein interactions in living cells.
Experimental Section
All samples from cross-linking assays were analyzed on standard
SDS-PAGE using 2 ꢁ SDS sample buffer containing 100 mm tris(2-
carboxyethyl)phosphine. The expression and activity of all constructs
was analyzed by live cell imaging. Representative results from at least
four independent experiments are presented. Detailed experimental
procedures are described in the Supporting Information.
[7] C. Hoffmann, G. Gaietta, A. Zurn, S. R. Adams, S. Terrillon,
M. H. Ellisman, R. Y. Tsien, M. J. Lohse, Nat. Protoc. 2010, 5,
1666 – 1677.
Received: September 9, 2011
Published online: November 16, 2011
[8] a) A. J. Boettcher, J. Wu, C. Kim, J. Yang, J. Bruystens, N.
Cheung, J. K. Pennypacker, D. A. Blumenthal, A. P. Kornev,
S. S. Taylor, Structure 2011, 19, 265 – 276; b) C. Kim, D. Vigil, G.
Anand, S. S. Taylor, Eur. J. Cell Biol. 2006, 85, 651 – 654; c) M.
Zaccolo, F. De Giorgi, C. Y. Cho, L. Feng, T. Knapp, P. A.
Negulescu, S. S. Taylor, R. Y. Tsien, T. Pozzan, Nat. Cell Biol.
2000, 2, 25 – 29.
[9] a) I. de Diego, J. Kuper, N. Bakalova, P. Kursula, M. Wilmanns,
Sci. Signal 2010, 3, ra6; b) D. Gozuacik, A. Kimchi, Autophagy
2006, 2, 74 – 79; c) A. Piljic, I. de Diego, M. Wilmanns, C.
Schultz, ACS Chem. Biol. 2011, 6, 685 – 691.
Keywords: biarsenical compounds · dimerization ·
protein cross-linking · protein–protein interactions ·
small molecules
.
[1] a) E. A. Althoff, V. W. Cornish, Angew. Chem. 2002, 114, 2433 –
2436; Angew. Chem. Int. Ed. 2002, 41, 2327 – 2330; b) J. L.
Czlapinski, M. W. Schelle, L. W. Miller, S. T. Laughlin, J. J.
Kohler, V. W. Cornish, C. R. Bertozzi, J. Am. Chem. Soc. 2008,
130, 13186 – 13187; c) S. Gendreizig, M. Kindermann, K. Johns-
son, J. Am. Chem. Soc. 2003, 125, 14970 – 14971; d) G. Lemercier,
S. Gendreizig, M. Kindermann, K. Johnsson, Angew. Chem.
2007, 119, 4359 – 4362; Angew. Chem. Int. Ed. 2007, 46, 4281 –
4284; e) E. J. Licitra, J. O. Liu, Proc. Natl. Acad. Sci. USA 1996,
93, 12817 – 12821; f) D. M. Spencer, T. J. Wandless, S. L.
Schreiber, G. R. Crabtree, Science 1993, 262, 1019 – 1024.
[10] a) B. A. Griffin, S. R. Adams, J. Jones, R. Y. Tsien, Methods
Enzymol. 2000, 327, 565 – 578; b) S. L. Robia, N. C. Flohr, D. D.
Thomas, Biochemistry 2005, 44, 4302 – 4311; c) X. Y. Zhang,
A. C. Bishop, J. Am. Chem. Soc. 2007, 129, 3812 – 3813.
12658 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 12655 –12658