One- and two-photon live cell imaging using a mutant SNAP-Tag protein and its FRET substrate pairs

Article abstract of DOI:10.1021/ol201430x

A small molecule-assisted protein labeling strategy based on a mutant SNAP-Tag (mSNAP) and its FRET substrate pairs has been developed. Both one- and two-photon fluorescence microscopic experiments were successfully demonstrated in living cells.

Full text of DOI:10.1021/ol201430x

ORGANIC
LETTERS
2011
Vol. 13, No. 16
4160–4163
One- and Two-Photon Live Cell Imaging
Using a Mutant SNAP-Tag Protein and Its
FRET Substrate Pairs
Chong-Jing Zhang,‡,§ Lin Li,‡ Grace Y. J. Chen,†,‡ Qing-Hua Xu,‡ and
Shao Q. Yao*,†,‡,§
Departments of Biological Sciences and Chemistry and NUS MedChem Program of the
Life Sciences Institute, National University of Singapore, Singapore 117543
chmyaosq@nus.edu.sg
Received May 26, 2011
ABSTRACT
A small molecule-assisted protein labeling strategy based on a mutant SNAP-Tag (mSNAP) and its FRET substrate pairs has been developed. Both
one- and two-photon fluorescence microscopic experiments were successfully demonstrated in living cells.
Fluorescence imaging provides an indispensable way to
locate and monitor biological targets within complex and
dynamic intracellular environments.1 The revolutionary
discovery of genetically encoded fluorescent proteins (FPs)
has made it possible to directly visualize proteins and
various biochemical activities, with unprecedented resolu-
tions, in the living system.2 The invention of two-photon
microscopy (TPM), which employs two near-infrared
photons as the excitation source, provides additional
advantages of increased penetration depth (∼500 μm),
localized excitation, and prolonged observation time,
thereby allowing examination of fluorophores present
even in deep living samples.3,4 Unfortunately, because
of the various undesirable photophysical properties of
most FPs, they have not been widely adopted in TPM
applications.4
tractable organic fluorophore/small molecule probe.5À7
They aim to address two intrinsic shortcomings of FP-
based imaging techniques, the size of FPs (>27 kDa) and
their strict confinement to only genetically encoded
fluorophores.8 For example, the FlAsH approach, devel-
oped by Tsien et al., uses a small molecule capable of
binding to a six-amino acid tetracysteine tag fused to the
target protein,5 and the SNAP-tag technology, developed
by Johnsson et al., makes use of a highly efficient enzy-
matic reaction between O6-alkylguanine-DNA-alkyltrans-
ferase (hAGT) and a variety of O6-benzylguanine (BG)-
modified probes.9 Other conceptually similar approaches,
such as HaloTag,10 D4-tag,11 ACP/PCP-tag,12,13 and
others,7 are also available. Few of these methods, with
the exception of FlAsH, make use of fluorogenic probes
for protein labeling (that is, the probe is rendered fluor-
escent only upon protein labeling), which is an important
feature for real-time bioimaging in live cells.5 In fact,
Chemistry-based protein-labeling approaches comple-
mentary to FPs have emerged in recent years, most of
which make use of a highly specific chemical or enzy-
matic reaction between a fusion protein and a chemically
(3) Kim, H. M.; Cho, B. R. Chem. Commun. 2009, 153–164.
(4) Helmchen, F.; Denk, W. Nat. Methods 2005, 2, 932–940.
(5) Griffin, B. A.; Adams, S. R.; Tsien, R. Y. Science 1998, 281, 269–
272.
(6) Johnsson, K. Nat. Chem. Biol. 2009, 5, 63–65.
(7) Chen, I.; Ting, A. Y. Curr. Opin. Biotechnol. 2007, 25, 1483–1487.
(8) Zhang, J.; Campbell, R. E.; Ting, A. Y.; Tsien, R. Y. Nat. Rev.
Mol. Cell Biol. 2002, 3, 906–918.
† Department of Biological Sciences.
‡ Department of Chemistry.
§ NUS MedChem Program of the Life Sciences Institute.
(1) Lichtman, J. W.; Conchello, J.-A. Nat. Methods. 2005, 2, 910–
919.
(2) Tsien, R. Y. Angew. Chem., Int. Ed. 2009, 48, 5612–5626.
r
10.1021/ol201430x
Published on Web 07/25/2011
2011 American Chemical Society

this issue is already being addressed in several recent
reports.14À16 Herein, we report a small molecule-assisted
protein labeling strategy, based on a mutant SNAP-tag
€
(mSNAP) and its Froster Resonance Energy Transfer
(FRET) substrate pairs (Figure 1). Our design principle
was based on the well-known fact that, in the SNAP-tag
technology, in which the SNAP-tag forms a covalent bond
with BG derivatives by nucleophilic attack at the active site
cysteine (Figure 1a), a guanine moiety is released. Further-
more, this reaction is independent of the dye attached to
the BG derivative. An elegant study by Johnsson et al.
provided further evidence that SNAP-tag (an extensively
mutated form of hAGT) and its suitable mutants may also
accept N9-substituded BG derivatives (Figure 1b).17 In our
strategy, quenched probes such as BGQFL-9 and
BGQNP-9 (a two-photon probe; Figure 1c), due to intro-
duction of a fluorescence quencher, Disperse Red 1 (DR1),
would be effectively nonfluorescent. Upon covalent label-
ing with mSNAP, the quencher-containing guanine is
released, resulting in transfer of the dye onto the tag
protein (and fluorescence enhancement). Similar concepts
had been proposed previously,18,19 but successful and
general implementation had not yet been realized in the
literature. It should be noted that while our manuscript
was in preparation, Urano et al. introduced what they call
a fluorescence activation-coupled protein labeling (FAPL)
method, in which BG derivatives modified with a quencher
at the C-8 position were used together with a SNAP-tag.20
Our present study offers a complementary method while
providing the first expansion of the SNAP-tag technology
into the realm of TPM.
Figure 1. (a) SNAP-tag protein labeling with BG derivatives. (b)
Modified strategy based on mSNAP with quenched probes. (c)
Quenched probes (N7- and N9-substituted BG derivatives) used.
Details of probe synthesis are presented in the Support-
ing Information (Scheme S1). In addition to BGQFL-9
and BGQNP-9, we also synthesized N7-substituted probes
BGQFL-7 and BGQNP-7 (Figure 1c) as well as the cell-
permeable BGQAF-9, which is the diacetylated version of
BGQFL-9. The two-photon dye 8-oxoacenaphthopyrrole
(NP) was used because of its desirable photophysical
properties for in vivo imaging.21 Disperse Red 1 was
chosen as the fluorescence quencher since the absorption
spectrum of DR1 overlaps substantially with the emission
spectra of both fluorescein (FL) and NP. All probes were
conveniently synthesized using the highly efficient and
modular click chemistry and fully characterized by LCÀ
MS and NMR (Supporting Information). Optical proper-
ties of the final probes were spectroscopically measured
(Figure S2, Supporting Information).
We carried out fluorescence measurement of the above
compounds (Table S1 and Figure S2 in the Supporting
Information); both BGFL and BGNP (quencher-free ver-
sions of BGQFL-9 and BGQNP-9, respectively) exhibited
excellent one- and two-photon fluorescence properties as
expected. BGQFL-9/-7 and BGQNP-9/-7, on the other
hand, showed almost no fluorescence, demonstrating high
intramolecular FRET efficiency upon the addition of DR1.
We next assessed the labeling efficiency of these probes
toward SNAP-tag and its mutants. SNAP-tag is a signifi-
cantly improved and truncated version of wildtype hAGT
(9) Keppler, A.; Gendreizig, S.; Gronemeyer, T.; Pick, H.; Vogel, H.;
Johnsson, K. Nat. Biotechnol. 2003, 21, 86–89.
(10) www.promega.com
(11) Ojida, A.; Honda, K.; Shinmi, D.; Kiyonaka, S.; Mori, Y.;
Hamachi, I. J. Am. Chem. Soc. 2006, 128, 10452–10459.
(12) George, N.; Pick, H.; Vogel, H.; Johnsson, N.; Johnsson, K. J.
Am. Chem. Soc. 2004, 126, 8896–8897.
(13) Yin, J.; Lin, A. J.; Buckett, P. D.; Wessling-Resnick, M.; Golan,
D. E.; Walsh, C. T. Chem. Biol. 2005, 12, 999–1006.
(14) Blum, G.; Mullins, S. R.; Keren, K.; Fonovic, M.; Jedeszko, C.;
Rice, M. J.; Sloane, B. F.; Bogyo, M. Nat. Chem. Biol. 2005, 1, 203–209.
(15) Komatsu, T.; Kikuchi, K.; Takakusa, H.; Hanaoka, K.; Ueno,
T.; Kamiya, M.; Urano, Y.; Nagano, T. J. Am. Chem. Soc. 2006, 128,
15946–15947.
with key mutations of G131fK131 and G132fT132 17
Pre-
.
(16) Mizukami, S.; Watanabe, S.; Hori, Y.; Kikuchi, K. J. Am. Chem.
Soc. 2009, 131, 5016–5017.
(17) Juillerat, A.; Heinis, C.; Sielaff, I.; Barnikow, J.; Jaccard, H.;
Kunz, B.; Terskikh, A.; Johnsson, K. ChemBioChem 2005, 6, 1263–
1269.
vious structural and mechanistic studies have revealed
that, while bulky residues (such as K131 and T132 in
(18) Jaccard, H.; Johnsson, K.; Kindermann, M. WO Pat, 2005/
085470 A1, 2005.
(19) Stohr, K.; Siegberg, D.; Ehrhard, T.; Lymperopoulos, K.; Oz, S.;
Schulmeister, S.; Pfeifer, A. C.; Bachmann, J.; Klingmuller, U.; Sourjik,
V.; Herten, D. Anal. Chem. 2010, 82, 8186–8193.
(20) Komatsu, T.; Johnsson, K.; Okuno, H.; Bito, H.; Inoue, T.;
Nagano, T.; Urano, Y. J. Am. Chem. Soc. 2011, 133, 6745–6751.
(21) Lee, J. H.; Lim, C. S.; Tian, Y. S.; Han, J. H.; Cho, B. R. J. Am.
Chem. Soc. 2010, 132, 1216–1217.
Org. Lett., Vol. 13, No. 16, 2011
4161

pDEST17 introduces several extra linker residues in GG-
SNAP), His-SNAP/G132-SNAP and GG-SNAP mig-
rated slightly differently on the SDS-PAGE gel (Figure
2a). As expected, BGFL labeled all three proteins with
reasonable efficiency. BGQFL-7, on the other hand, could
not label any of the proteins (Figure S3, Supporting
Information). To our delight, BGQFL-9 was shown to
successfully label all three proteins, with GG-SNAP pro-
ducing the most intense fluorescence band (Figure 2a).
This indicates the K131fG131 and T132fG132 mutations in
SNAP-tag, giving GG-SNAP, have indeed improved this
protein’s reactivity towardourquenchedprobe BGQFL-9.
The labeling efficiency of SNAP-tag by BGQFL-9 de-
creased significantly when compared to the original
BGFL/SNAP labeling (Figure S3, Supporting Information),
indicating the steric bulk of the quencher group in
BGQFL-9 had predictably hindered its enzymatic attach-
ment to SNAP-tag. To test whether the two-photon
quenched probe, BGQNP-9, could also label GG-SNAP
efficiently, we used an indirect competition assay since NP
could not be detected with our fluorescence gel scanner. As
shown in Figure 2b, preincubation of GG-SNAP with
BGQNP-9 effectively blocked the subsequent fluorescence
labeling of BGFL in a time-dependent manner, indicative
of successful labeling between GG-SNAP and BGQNP-9.
We next evaluated whether the enhancement of BGQFL-9
fluorescence upon labeling by GG-SNAP could be mon-
itored spectroscopically (Figure 2c,d); a progressive in-
crease in fluorescence (with max λem = 522 nm) over time
was observed, indicating successful release of the quencher
(Figure 2c). In comparison, both SNAP-tag and G132-
SNAP produced significantly lower fluorescence increases
under identical conditions (Figure 2d). Kinetic data of the
labeling reaction between BGQFL-9 and various SNAP
mutants were obtained, indicating GG-SNAP was indeed
the most efficient partner of our N9-substituted quenched
probe; with a second-order rate constant of 8.16 ( 1.10
Figure 2. (a) In-gel fluorescence scanning of labeling reactions
between different SNAP proteins (1 μM) and BGQFL-9 (1 μM)
after 4-h incubation at room temperature. Fluorescent bands
were quantified and plotted (bottom). (b) Indirect competition
assay of BGQNP-9/GG-SNAP labeling reactions (for 0, 1, 4,
and 9 h), followed by subsequent addition of BGFL (1 μL, 100 μM;
15 min), SDS-PAGE and in-gel fluorescence scanning. Fluor-
escent bands were quantified and plotted (bottom). (c) Time-
dependent emission spectra (λex = 470 nm) of BGQFL-9
(20 μM) in the presence of GG-SNAP (10 μM) in PBS (pH
7.4) at 25 °C. (d) Time-dependent fluorescence intensity of
BGQFL-9 in the presence of different SNAP mutants (λex
=
490 nm, λem = 522 nm). Assay conditions were the same as in
(c). The fitted cuves and corresponding rate constants were
obtained by fitting the data to a first-order reaction model,9
giving rise to the resulting second-order rate constants.
MÀ1 À1, our labeling system is not as efficient as the
s
SNAP-tag) favor BG derivatives, structurally less demand-
ing residues (such as Gly at positions 131 and 132) prefer
N9-substituted BG compounds (see Figure S11, Supporting
Information).17 We therefore generated the corresponding
protein mutants of SNAP-tag (mSNAP)-G132-SNAP
(T132 in SNAP-tag was replaced by G132) and GG-SNAP
(K131 and T132 in SNAP-tag were replaced by G131G132).
The bacterial expression construct G132-SNAP was ob-
tained from the corresponding SNAP-tag template (His-
SNAP)22 using a Quick Change site-directed mutagenesis
kit (Stratagene). The GG-SNAP construct was generated
by Gateway cloning (Invitrogen). The mammalian expres-
sion constructs Flag-GG-SNAP and Flag-H2B-GG-
SNAP, in which a nuclear localization sequence H2B
was fused to the SNAP-tag fusion, were obtained from
the corresponding Flag-SNAP and Flag-H2B-SNAP vec-
tors, respectively.22 All plasmid DNAs were sequence
verified. The recombinant proteins (His-SNAP, G132-
SNAP, and GG-SNAP) were expressed in BL21(Ai) cells
and purified to homogeneity with Ni-NTA beads (Figure
S1, Supporting Information). Because of the difference in
the vector’s backbone (i.e., Gateway destination vector
original BGFL/SNAP combination (k = ∼2 Â 104 MÀ1
sÀ1 17 or the newly reported C8-substituted quenched
)
probe/SNAP pair (k = ∼4 Â 102 MÀ1
s )
À1 20 but should
offer a good starting point for further improvement using
directed protein evolution approaches.17 Finally, the
labeled product of the BGQFL-9/GG-SNAP reaction was
further analyzed by MALDI-TOF MS (Figure S4, Sup-
porting Information); an expected molecular weight in-
crease of 558 Da was observed, further confirming the
success of the labeling reaction.
Lastly, we examined whether the BGQFL-9/GG-SNAP
protein labeling system can be used for bioimaging appli-
cations in live cells. We first applied BGQFL-9 to mam-
malian cell lysates (Figure 3a); a single fluorescent band
was detected only in mammalian cell lysates obtained from
CHO-9 cells transiently transfected with the Flag-GG-
SNAP plasmid, thus demonstrating the successful and
highly specific labeling reaction of our newly developed
(22) Srikum, D.; Albers, A. E.; Nam, C. I.; Iavaron, A. T.; Chang,
C. J. J. Am. Chem. Soc. 2010, 132, 4455–4465.
4162
Org. Lett., Vol. 13, No. 16, 2011

added either BGQFL-9 or BGQNP-9. Following further
incubation, cells were imaged, using one- and two-photon
fluorescence microscopes, respectively (Figure 3b); only
transfected cells exhibited strong fluorescence signals in
their nuclei, indicating successful labeling of GG-SNAP in
live cells.
In conclusion, by modifying the existing SNAP-tag
protein labeling approach, we have successfully developed
both one-and two-photon quenched probes, BGQFL-9
and BGQNP-9, respectively, that could covalently label a
SNAP-tag mutant protein (mSNAP) with moderate effi-
ciency. We showed that, in addition to C-8,20 the N-9
position in BG is also suitable for quencher attachment.
Our results indicate real-time detection of both the labeling
reaction and live cell imaging could be achieved using this
system. To our knowledge, the only other small molecule-
based protein labeling approach for two-photon micro-
scopic applications was done using the DHFR/Mtx non-
covalent strategy.23 Our current system thus represents the
first covalent protein labeling approach using small mole-
cule probes for TPM applications. This, togeher with other
recenlty developed enzyme-detecting TPM probes,24 will
open up new opportunities for bioimaging of important
enzymes and their activities in deep tissues, where conven-
tional fluorescence microscopy has very limited utilities.3,4
Figure 3. (a) Coomassie brilliant blue (CBB)-stained (left),
Western blot (center; with anti-Flag antibody), and fluorescence
(right) gels of Flag-GG-SNAP transfected CHO-9 cell lysates
incubated with BGQFL-9. (b) Live CHO-9 cells transfected with
Flag-H2B-GG-SNAP and then labeled with either BGQFL-9 or
BGQNP-9 (20 μM). Key: (i) nontransfected cells labeled with
BGQNP-9; (ii) transfected cells labeled with BGQFL-9, then
imaged by one-photon microscope (λex = 488 nm; λem = 522 nm);
(iii) Transfected cells labeled with BGQNP-9, then imaged by
one-photon microscope (λex =405 nm; λem = 470 nm); (iv)
same as iii, except the image was acquired by a two-photon
microscope (λex = 800 nm; λem = 470 nm). Scale bar = 10 μm.
Acknowledgment. Funding support was provided by
the Agency for Science, Technology and Research (R-
143-000-391-305), and Ministry of Education (R-143-
000-394-112). We thank Prof. Christopher J. Chang at
the University of California for kindly providing His-
SNAP, Flag-SNAP, and Flag-H2B-SNAP plasmids.
system even in complex biological systems. To label pro-
teins in live cells, we had initially intended to use the cell-
permeable BGQAF-9 (see the Supporting Information for
structure), but it was subsequently discovered that, to our
pleasant surprise, both BGQFL-9 and BGQNP-9 were
equally permeant in CHO-9 cells, probably due to their
overall enhanced hydrophobicity. These two probes were
therefore used to carry out all live cell labeling/imaging
experiments. The plasmid Flag-H2B-GG-SNAP, which
expresses GG-SNAP in the nuclei, was transfected into
CHO-9 cells. After 3 h, to the growth media was directly
Supporting Information Available. Experimental pro-
cedures, characterization of new compounds, and biolo-
gical experiment. This material is available free of charge
via the Internet at http://pubs.acs.org.
(23) Gallagher, S. S.; Jing, C.; Peterka, D. S.; Konate, M.;
Wombacher, R.; Kaufman, L. J.; Yuste, R.; Cornish, V. W. Chem-
BioChem 2010, 11, 782–784.
(24) Hu, M.; Li, L.; Wu, H.; Su, Y.; Yang, P. -Y.; Uttamchandani, M.;
Xu, Q. H.; Yao, S. Q. J. Am. Chem. Soc., 2011, DOI: 10.1021/ja200808y.
Org. Lett., Vol. 13, No. 16, 2011
4163