Second-generation covalent TMP-tag for live cell imaging

Article abstract of DOI:10.1021/ja303374p

Chemical tags are now viable alternatives to fluorescent proteins for labeling proteins in living cells with organic fluorophores that have improved brightness and other specialized properties. Recently, we successfully rendered our TMP-tag covalent with a proximity-induced reaction between the protein tag and the ligand-fluorophore label. This initial design, however, suffered from slow in vitro labeling kinetics and limited live cell protein labeling. Thus, here we report a second-generation covalent TMP-tag that has a fast labeling half-life and can readily label a variety of intracellular proteins in living cells. Specifically, we designed an acrylamide-trimethoprim-fluorophore (A-TMP-fluorophore v2.0) electrophile with an optimized linker for fast reaction with a cysteine (Cys) nucleophile engineered just outside the TMP-binding pocket of Escherichia coli dihydrofolate reductase (eDHFR) and developed an efficient chemical synthesis for routine production of a variety of A-TMP-probe v2.0 labels. We then screened a panel of eDHFR:Cys variants and identified eDHFR:L28C as having an 8-min half-life for reaction with A-TMP-biotin v2.0 in vitro. Finally, we demonstrated live cell imaging of various cellular protein targets with A-TMP-fluorescein, A-TMP-Dapoxyl, and A-TMP-Atto655. With its robustness, this second-generation covalent TMP-tag adds to the limited number of chemical tags that can be used to covalently label intracellular proteins efficiently in living cells. Moreover, the success of this second-generation design further validates proximity-induced reactivity and organic chemistry as tools not only for chemical tag engineering but also more broadly for synthetic biology.

Full text of DOI:10.1021/ja303374p

Article
pubs.acs.org/JACS
Second-Generation Covalent TMP-Tag for Live Cell Imaging
Zhixing Chen,† Chaoran Jing,† Sarah S. Gallagher,†,§ Michael P. Sheetz,‡ and Virginia W. Cornish*,†
‡
†Department of Chemistry and Biological Sciences, Columbia University, New York, New York 10027, United States
S
* Supporting Information
ABSTRACT: Chemical tags are now viable alternatives to
fluorescent proteins for labeling proteins in living cells with
organic fluorophores that have improved brightness and other
specialized properties. Recently, we successfully rendered our
TMP-tag covalent with a proximity-induced reaction between
the protein tag and the ligand-fluorophore label. This initial
design, however, suffered from slow in vitro labeling kinetics
and limited live cell protein labeling. Thus, here we report a
second-generation covalent TMP-tag that has a fast labeling
half-life and can readily label a variety of intracellular proteins in living cells. Specifically, we designed an acrylamide-
trimethoprim-fluorophore (A-TMP-fluorophore v2.0) electrophile with an optimized linker for fast reaction with a cysteine
(Cys) nucleophile engineered just outside the TMP-binding pocket of Escherichia coli dihydrofolate reductase (eDHFR) and
developed an efficient chemical synthesis for routine production of a variety of A-TMP-probe v2.0 labels. We then screened a
panel of eDHFR:Cys variants and identified eDHFR:L28C as having an 8-min half-life for reaction with A-TMP-biotin v2.0 in
vitro. Finally, we demonstrated live cell imaging of various cellular protein targets with A-TMP-fluorescein, A-TMP-Dapoxyl, and
A-TMP-Atto655. With its robustness, this second-generation covalent TMP-tag adds to the limited number of chemical tags that
can be used to covalently label intracellular proteins efficiently in living cells. Moreover, the success of this second-generation
design further validates proximity-induced reactivity and organic chemistry as tools not only for chemical tag engineering but also
more broadly for synthetic biology.
trimethoprim-fluorophore (TMP-fluorophore) conjugate.
eDHFR is an attractive protein tag because it is 18 kD
(about two-thirds the size of GFP) and monomeric and thus
minimally disrupts biological function of the tagged protein and
pathway. With low nanomolar affinities for eDHFR, the TMP-
fluorophore conjugates can be used at near stoichimetric
quantities to efficiently label tagged cellular proteins, which
average a ∼1 μM concentration in the cell. At the same time,
with >1000-fold selectivity for E. coli over mammalian DHFRs,
TMP-tag shows minimal background labeling of endogenous
proteins and no apparent cellular toxicity in mammalian cell
lines. TMP is commercially available and can be readily
modified without disrupting binding to eDHFR, facilitating the
preparation of a wide variety of TMP analogues.15 Finally, there
is a wealth of biochemical and structural knowledge of the
interaction between TMP and eDHFR, which facilitates further
engineering of the TMP-tag.16 Fine tuning of the fluorophore
hydrophobicity and linker structure have produced optimized
versions of TMP-tag for lower, unspecific, background staining
and better cell permeability.17 On the strength of its robustness,
TMP-tag labels have been developed to enable super-resolution
imaging14 and two-photon imaging of cellular proteins,18
chromophore-assisted laser inactivation (CALI) of components
of the focal adhesion complex,19 and single-molecule imaging of
spliceosome assembly.11
INTRODUCTION
■
Chemical tags are emerging from the proof-of-principle stage to
viable reagents for labeling proteins in living cells with
fluorophores with high photon outputs and other specialized
properties1−3 to enable experiments difficult or not possible
with the fluorescent proteins (FPs).4 With chemical tags, rather
than tagging the target protein with an FP, the target protein is
tagged with a polypeptide, which is subsequently labeled with a
cell-permeable fluorophore ligand or substrate. Thus, chemical
tags combine the advantage of specificity through genetic
encoding with a modular organic fluorophore. Chemical tags
now in use include the seminal peptide chelator-based FlAsH/
ReAsH system,5 the enzyme suicide substrate-based SNAP/
CLIP-tags6,7 and Halo-tag,8 the small-molecule inhibitor-based
TMP-tag,9 and the enzyme-mediated polypeptide labeling-
based lipoic acid ligase tag.10 Exciting recent applications of the
chemical tags include single-molecule imaging of spliceosome
function in yeast cell extracts,11 magnetically modulating
mammalian cells using decorated iron oxide nanoparticles,12
imaging LDL receptor oligomerization during endocytosis,13
and super-resolution imaging of cellular proteins.14 While new
chemical tags are regularly being introduced in the literature,1−3
our TMP-tag still stands out as one of the few chemical tags
able to label intracellular, as opposed to cell-surface, proteins
with high selectivity.
With TMP-tag, the target protein is tagged with Escherichia
coli dihydrofolate reductase (eDHFR) through standard genetic
encoding and then labeled by binding to a cell-permeable
Received: April 8, 2012
Published: August 9, 2012
© 2012 American Chemical Society
13692
dx.doi.org/10.1021/ja303374p | J. Am. Chem. Soc. 2012, 134, 13692−13699

Journal of the American Chemical Society
Article
To provide a more permanent label for advanced
applications such as single-molecule tracking or pulse−chase
labeling, we recently reported a covalent variant of the TMP-tag
based on a proximity-induced reaction between the eDHFR tag
and the TMP-fluorophore label.20 Briefly, a unique Cys
nucleophile was engineered just outside the TMP-binding
pocket of eDHFR (eDHFR:L28C) in position to react with an
acrylamide electrophile installed on the TMP-fluorophore label
(acrylamide-TMP-fluorophore, or A-TMP-fluorophore). This
design was based on the long-standing use of proximity-induced
reactivity21,22 for the design of covalent inhibitors23,24 and more
recent application to chemical biology tools.25 Work of Belshaw
and co-workers26,27 led us to believe that the acrylamide
electrophile would have the right balance in reactivity, being a
sufficiently mild electrophile to minimize nonspecific, back-
ground labeling of cellular components, but being reactive
enough to undergo a rapid Michael addition upon TMP
binding to eDHFR. Our initial design was successful, and we
demonstrated that A-TMP-biotin reacted with eDHFR:L28C
with a half-life of ∼50 min in vitro and that A-TMP-fluorophore
could covalently label a nuclear-localized eDHFR fusion protein
in live cells with minimal background labeling of other cellular
proteins. However, this first-generation covalent TMP-tag was
unable to label cytoplasmic proteins tagged with eDHFR:L28C,
limiting its utility. We hypothesized that this limited reactivity
resulted from the slow half-life with which A-TMP reacted with
eDHFR:L28C.
Thus, we sought to design a second-generation covalent
TMP-tag with a rapid labeling half-life that would improve its
utility for live cell imaging. Previous reports in the literature
have shown that the half-life of both covalent inhibitors24 and
chemical biology tools26 can be improved to a few minutes with
optimization of the reaction geometry between the protein
nucleophile and the organic electrophile. We present the design
and synthesis of an optimized v2.0 A-TMP-probe in
conjunction with the rational design and screening of a panel
of eDHFR:Cys variants to generate a v2.0 covalent TMP-tag
with a rapid reaction half-life. Finally, we challenge the
robustness of this v2.0 covalent TMP-tag with live-cell, confocal
fluorescence imaging of multiple intracellular proteins in
different mammalian cell lines.
Figure 1. Schematic representation of the covalent TMP-tag design.
Previously, we demonstrated that the noncovalent TMP-tag, which
exploits the high-affinity, selective interaction between trimethoprim
(TMP) and E. coli dihydrofolate reductase (eDHFR), could be
rendered covalent by proximity-induced reaction between a Cys
residue engineered on the eDHFR surface and a mild acrylamide
electrophile installed on the TMP-fluorophore probe. Here, by
optimizing the positioning of the Cys nucelophile and the acrylamide
electrophile, we achieve rapid covalent labeling of the eDHFR tag by
the TMP-fluorophore probe, rendering the covalent acrylamide TMP-
tag (A-TMP-tag) a robust reagent for live cell imaging.
long linker length, it was not surprising that the initial covalent
TMP-tag had an in vitro labeling reaction half-life of almost 1 h.
We chose to use rational design in combination with
screening of a small number of variants to create a covalent
TMP-tag with the minimum necessary distance between the
Cys nucleophile and the acrylamide electrophile to achieve the
desired reduction in reaction half-life. First, molecular modeling
was applied to the high-resolution structure of eDHFR16,28 to
identify residues that had solvent-accessible side chains in
which the side chain faced the binding pocket to ensure the
engineered Cys residue would be accessible to react with the
acrylamide electrophile. A model of TMP bound to eDHFR
was created by structurally aligning a high-resolution structure
of E. coli DHFR28 to a high-resolution structure of TMP bound
to Lactobacillus casei DHFR.29 Only residues in close proximity
to the binding pocket were selected, since precedent has shown
that the closer the residue to the binding pocket, the more
rapid the rate of alkylation.26 Four residues were selected that
met this criterion: Ala19, Asn23, Leu28, and Arg52. An
approximation was made of the minimum linker length
between TMP and the electrophile that would allow
proximity-induced covalent labeling to occur upon binding to
a mutant eDHFR containing each of these Cys mutants (Table
S1). According to the model described above, we envisioned
that a 10-atom spacer between the 4′-OH group of TMP and
the β-carbon of the acrylamide would enable the electrophile to
reach all the four engineered Cys nucleophiles (Figure 2).
Synthesis of the A-TMP-Probe Heterotrimer. Guided
by molecular modeling, we designed an A-TMP-probe v2.0
heterotrimer with a 10-atom spacer between the 4′-OH group
of TMP and the β-carbon of the acrylamide. We chose aspartic
acid as the trifunctional core because amino acid derivatives
could provide a convenient protection strategy for the
RESULTS
■
Design of the Second-Generation Covalent TMP-Tag.
On the basis of the success of our initial covalent TMP-tag, a
second-generation covalent TMP-tag was also built around the
Cys nucleophile and acrylamide electrophile (Figure 1), while
optimizing the positioning of the nucleophile and electrophile
to improve the reaction half-life. The Cys nucleophile and
acrylamide electrophile had exceeded our expectations for
minimal background labeling of endogenous cellular compo-
nents and minimal cellular toxicity and yet still were able to
undergo a fairly rapid binding-induced Michael addition.
Previous literature on the design of covalent inhibitors and
chemical biology reagents suggested that we could achieve a
reaction half-life of a few minutes simply by optimizing the
positioning of the Cys side-chain and acrylamide group
undergoing the Michael addition.26 For our initial covalent
TMP-tag,20 we chose a conservative design containing a 21-
atom linker between the 4′−OH group of TMP and the
reactive β-carbon of the acrylamide functional group to ensure
that the acrylamide would be available to react with the Cys
nucleophile installed on the surface of eDHFR. Because of this
13693
dx.doi.org/10.1021/ja303374p | J. Am. Chem. Soc. 2012, 134, 13692−13699

Journal of the American Chemical Society
Article
diacetate conjugate as DMF stock solution. Addition of the
probe in the final step could be beneficial to the generality of
the tag, facilitating the preparation of a variety of A-TMP
derivatives with different probe molecules.
The synthetic route of the target molecule is summarized in
Scheme 1. H-Asp(OBu-t)-OH (2), a commercially available
aspartic acid derivative, was treated with acryloyl chloride to
yield carboxylic acid 3. Amine 7 was prepared by O-alkylation
of TMP phenol 420 with a three-carbon Boc-amino iodide (5)
followed by TFA deprotection of the Boc group. Coupling of
carboxylic acid 3 and amine 7 with EDCI led to tert-butyl ester
8, which was subjected to TFA deprotection to yield carboxylic
acid 9, a key intermediate toward the A-TMP-probe
heterotrimer. A PEG linker was incorporated to carboxylic
acid 9 by EDCI-mediated coupling with a monoprotected PEG
bis-amine (10) followed by TFA deprotection of the Boc
group. The product, amine 12, was coupled with protected
fluorescein NHS-ester (13) and purified by HPLC to yield the
final heterotrimer, compound 1, in pure form. The heterotrimer
was prepared from aspartic acid derivative 2 in 3.4% overall
yield with the longest linear sequence consisting of seven steps.
This modular synthetic plan would allow us to prepare a variety
of v2.0 A-TMP-probe molecules. For an illustrative example, A-
TMP-biotin v2.0 (S2) was also prepared by a similar synthetic
plan (Figure S2).
In Vitro Screening of A-TMP v2.0 with eDHFR:Cys
Variants. After we obtained the A-TMP-probe v2.0 molecule,
we moved to in vitro labeling studies to determine the best
eDHFR:Cys variant to pair with A-TMP (Figure 3A). We
found that the most rapid reaction occurred between
eDHFR:L28C and A-TMP-biotin v2.0 among the tested
eDHFR variants. Under the tested labeling conditions, the
half-life of the labeling reaction was determined to be 8 min in
the presence of NADPH.
Figure 2. Design of the optimized, second-generation covalent A-
TMP-tag. The acrylamide elecrophile on the TMP-fluorophore probe
and the Cys nucleophile on the eDHFR surface were redesigned to
bring the two in close proximity to achieve a rapid reaction half-life.
Depicted is a cartoon of a second-generation A-TMP molecule (stick
representation, electrophile highlighted in orange) with a 10-atom
spacer between the TMP ligand bound in the active site of eDHFR
(green ribbon diagram) and the acrylamide electrophile with the four
residues chosen for mutation to Cys highlighted (stick representation,
α carbon highlighted in purple). Since there is no reported high-
resolution structure of TMP bound to eDHFR, this model was created
by structurally aligning a high-resolution structure of E. coli DHFR28 to
a high-resolution structure of TMP bound to L. casei DHFR.29 The
acrylamide-TMP structure was built in Maestro30 and then super-
imposed on TMP in the eDHFR model. The graphic was prepared
using PyMOL.31.
sequential addition of the three different groups. In the proof-
of-principle demonstration we chose fluorescein diisobutyrate
as the probe, as cell behavior of fluorescein derivatives have
been well-studied.17,32 We chose fluorescein diisobutyrate
conjugate due to its higher chemical stability over fluorescein
Evaluation of the reactions between A-TMP V2.0 and
eDHFR variants was conducted using purified proteins. The
Scheme 1. Synthesis of Optimized Acrylamide-TMP-Fluorescein Heterotrimer (A-TMP-fluorescein v2.0, compound 1)
13694
dx.doi.org/10.1021/ja303374p | J. Am. Chem. Soc. 2012, 134, 13692−13699

Journal of the American Chemical Society
Article
that all of the eDHFR variants (Figure S5) reacted with A-
TMP-biotin, but with significantly different reaction half-lives
(Table 1). Intriguingly, NADPH promoted the reaction of A-
Table 1. Reaction Half-Lives [min] of A-TMP-Biotin v2.0
a
with eDHFR:Cys variants
eDHFR variants
L28C
N23C
A19C
R52C
NADPH
NADPH
−
+
17
8
35
20
330
100
130
220
a
Purified eDHFR:Cys variant was labeled under the same conditions
as in Figure 3B.
Figure 3. Determination of the rate of covalent labeling between A-
TMP-biotin v2.0 and eDHFR:L28C in vitro. (A) Illustrative reaction
scheme of the proximity-induced Michael addition of the thiol
nucleophile of L28C to the acrylamide electrophile of A-TMP-biotin
v2.0. (B) Analysis of the labeling reaction between A-TMP-biotin v2.0
and eDHFR:L28C by SDS-PAGE. Purified eDHFR:L28C (5 μM) was
incubated with A-TMP-biotin v2.0 (10 μM) in PBS buffer (pH 7.4)
with reduced glutathione (1 mM) at 37 °C, with or without NADPH
(50 μM). At proper time points, aliquots (20 μL) were removed from
the reaction mixture, quenched with 6× SDS and boiled for 5 min.
The time point aliquots were then analyzed by SDS-PAGE and
Coomasie staining. Conveniently, covalent modification of
eDHFR:L28C gave rise to a gel shift such that the reaction progress
could be readily measured by densitometry analysis of Coomassie
stained gels using Image-J. The labeling half-life was determined by
linear regression, applying the pseudo-first-order model onto the ratio
of the substrate and product. eDHFR:L28C was found to react with A-
TMP-biotin v2.0 with a half-life of 8 min at these physiologically
relevant conditions.
TMP-biotin v2.0 with eDHFR:L28C, N23C and A19C, but
slowed the reaction with eDHFR:R52C, perhaps indicative of
complex conformational effects of NADPH binding. Overall,
eDHFR:L28C was chosen as the fastest target for the designed
second-generation A-TMP-probe molecule, especially in the
presence of NADPH. This system would be particularly
suitable for intracellular targets because of the abundance of
NADPH in the reducing cellular environment.
Protein Labeling in Live Cells with the Second-
Generation Covalent TMP-tag. Encouraged by the rapid
in vitro labeling reaction between eDHFR:L28C and A-TMP-
probe v2.0, the selected pair was next evaluated by labeling of
cellular proteins. eDHFR:L28C was genetically fused to the C-
termini of four different target proteins: histone H2B,
Tomm20, α-actinin, and myosin light chain (MLC). Mamma-
lian cell lines expressing the fusion proteins were successfully
labeled with A-TMP-fluorescein v2.0 in 10 min and were
characterized by both microscopy and in-gel fluorescence
analysis. These results demonstrate the generality of the v2.0
covalent TMP-tag for live cell imaging.
At first, we aimed for labeling of an abundant cellular target.
We chose histone H2B, an essential nuclear protein which has
been intensively investigated, as the first target. HEK 293T cells
transiently transfected with plasmids encoding H2B-
eDHFR:L28C fusion protein were incubated with 1 μM A-
TMP-Fl v2.0 for 10 min. After staining, cells were washed twice
with media and imaged by confocal microscopy. Distinct
nucleic distribution of fluorescence was observed in transfected
cells, with chromosomal patterns observed in a number of cells
(Figure 4A). No significant background cytosol staining was
detected. These observations indicated that A-TMP-Fl v2.0 was
able to selectively bind to H2B-eDHFR:L28C fusion proteins
with rapid kinetics in live cells.
To confirm that the labeling reaction was covalent, HEK
293T cells expressing the H2B-eDHFR:L28C fusion protein
were treated with 1 μM A-TMP-Fl v2.0 for 10 min, 30 min, or
3 h. After staining, cells were lysed and analyzed by SDS-PAGE
and in-gel fluorescence (Figure 4B). A major band with a green
channel emission was detected as the expected 35 kD H2B-
eDHFR:L28C-A-TMP-Fl v2.0 conjugate, while no detectable
background binding was observed in nontransfected cells
(Figure S6). The labeling products were further confirmed by
Western blot analysis using Anti-H2B antibody. Several minor
bands with lower molecular weight were also detected in cells
expressing the H2B-eDHFR:L28C fusion protein, particularly
in longer incubation. These bands are likely to be the
degradation products of labeled H2B, as the control experiment
showed undetectable background staining of endogenous
proteins.
four designed eDHFR:Cys variants eDHFR:A19C,
eDHFR:N23C, eDHFR:L28C, and eDHFR:R52C were over-
expressed via the T7 promoter using the corresponding E. coli
expression vectors and purified using Ni-NTA spin columns.
The proteins were judged to be more than 95% pure according
to Coomassie staining of SDS-PAGE gels. The endogenous
cysteines of eDHFR, Cys85, and Cys152 were mutated to
serines in these vectors to minimize possible cross reactivity
with the engineered Cys nucleophile. A-TMP-biotin v2.0 was
used as the tag in the conditional screening due to its better
solubility in PBS buffer compared with the hydrophobic
protected A-TMP-Fl.
Next, the in vitro labeling kinetics were determined using a
SDS-PAGE gel shift assay to identify the fastest eDHFR/A-
TMP pair (Figure 3B). Following the conditions reported in
our previous study,20 10 μM A-TMP-biotin was reacted with 5
μM eDHFR:Cys variant in PBS buffer, and the reaction mixture
was quenched at appropriate intervals with 6X SDS and then
subjected to SDS-PAGE. Gel shifts were produced due to
covalent modification of the eDHFR:Cys varients, simplifying
analysis of the reaction progression as described in Figure 3.
We first tested the reaction between eDHFR:L28C and A-
TMP-biotin v2.0. The labeling reaction was near quantitative
after 3 h, and the time required for 50% labeling was
determined to be 17 min. We then tested the effect of
NADPH to this labeling reaction, as NADPH is a native
cofactor of eDHFR. In the presence of 50 μM NADPH,
estimated to be the cellular concentration of NADPH,33,34 the
reaction between eDHFR:L28C and A-TMP-biotin v2.0 was
accelerated with a half-life of 8 min. Further screening found
13695
dx.doi.org/10.1021/ja303374p | J. Am. Chem. Soc. 2012, 134, 13692−13699

Journal of the American Chemical Society
Article
Figure 5. Labeling of diffused protein targets with covalent A-TMP-Fl
v2.0. Three different proteins, Tomm20, MLC, and α-actinin, were
successful labeled and imaged in two different mammalian cell lines.
(A) Microscopic imaging of A-TMP-tag v2.0 labeling. HEK293T cells
(for Tomm20) or mouse embryonic fibroblast (MEF) cells (for MLC
and α-actinin) transiently cotransfected with vectors encoding POI-
eDHFR:L28C and H2B-mCherry fusion proteins, respectively, were
incubated with 1 μM A-TMP-fluorescein v2.0 in media for 10 min,
washed twice with media, and directly imaged using confocal and
differential interference contrast (DIC) microscopy. Fluorescein was
excited at 488 nm, mCherry was excited at 594 nm. Scale bars are 25
μm. (B) In-gel fluorescence analysis of A-TMP-Fl v2.0 labeling. The
cells transfected with corresponding POI-eDHFR:L28C vectors were
harvested after 10 min incubation with 1 μM A-TMP-fluorescein v2.0,
lysed, and analyzed by SDS-PAGE and in-gel fluorescence scanning
with an excitation laser at 488 nm. These results show the target
versatility of the second-generation covalent TMP-tag for live cell
protein labeling.
Figure 4. Labeling of H2B with the covalent A-TMP-fluorescein v2.0
in HEK293T cells. (A) Microscopic evidence of successful labeling of
H2B tagged with eDHFR:L28C (H2B-eDHFR:L28C) by A-TMP-
fluorescein v2.0 in human embryonic kidney (HEK) 293T cells.
HEK293T cells transiently transfected with a vector encoding the
H2B-eDHFR:L28C fusion protein were incubated with 1 μM A-TMP-
fluorescein v2.0 in the appropriate media for 10 min, washed twice,
and then directly imaged using confocal fluorescence and differential
interference contrast (DIC) microscopy. Fluorescein was excited at
488 nm. Scale bars are 50 μm. (B) Zoom-in view of the fluorescence
image shown in (A). (C) In-gel fluorescence and Western blot analysis
of the labeling reaction. Labeled HEK 293T cells were lysed and then
analyzed by SDS-PAGE and in-gel fluorescence scanning with an
excitation laser at 488 nm. Together, these results provide evidence
that A-TMP-fluorescein v2.0 labels eDHFR:L28C tagged H2B rapidly,
selectively, and covalently.
TMP-Fl v2.0 for 10 min, respectively. In both experiments
H2B-mCherry was cotransfected as a counter stain. After
washing with fresh media, cells were imaged with a confocal
microscope. While noncovalent TMP-tag exhibits a similar
labeling specificity in live cell imaging experiments with the
covalent TMP-tag, the labeling pattern could not be
distinguished 12 h after paraformaldehyde fixation. In contrast,
the labeling of Tomm20-eDHFR:L28C with A-TMP-Fl v2.0
withstands the fixation protocol, which potentially facilitates the
applications of novel microscopic studies requiring long
acquisition time.
Finally, we prepared A-TMP-Atto 655 and A-TMP-Dapoxyl
to demonstrate the adaptability of second-generation covalent
TMP-tag over novel fluorophores for potential advanced
imaging applications. Atto 655 has been demonstrated as an
ideal organic fluorophore for live-cell super-resolution imaging
due to its unique cellular-environment-compatible photo-
switching mechanism.14 After 3 h incubation of 1 μM A-
TMP-Atto 655 with HEK 293T cells transiently expressing
H2B-eDHFR:L28C or plasma membrane-targeted
eDHFR:L28C (PMLS-eDHFR:L28C),9 selective labeling
could be observed in both cases with confocal microscopy
(Figure 5A). Dapoxyl dye, since its invention,35 has been
gaining growing attention due to its large and environmentally
sensitive Stokes shift.36 A-TMP-Dapoxyl was tested in labeling
experiments with H2B-eDHFR:L28C as well as plasma
membrane targeted eDHFR:L28C. Organelle-specific fluores-
cence images were obtained using confocal microscopy with a
Our next goal was to test the versatility of the second-
generation covalent TMP-tag for labeling diffuse cellular
protein targets. Tomm20, a mitochondrial localized protein,
was chosen as an organelle target in HEK 293T cells. Myosin
light chain (MLC) and α-actinin, two cytoplasmic proteins,
were chosen as cell skeleton targets in fibroblast cells. Cells
expressing eDHFR:L28C fusions were treated with 1 μM A-
TMP-Fl v2.0 for 10 min and then examined using a confocal
microscope. In each scenario, H2B-mCherry fusion protein was
cotransfected. The fluorescence images indicate that all three
eDHFR:L28C fusion proteins could be successfully labeled
(Figure 5A). To further characterize the labeling specificity,
cells expressing the different eDHFR:L28C fusions were lysed
after 10 min treatment of 1 μM A-TMP-Fl v2.0. The lysates
were subjected to SDS-PAGE and analyzed by in-gel
fluorescence with a 488 nm laser. In all cases, a single main
fluorescent band of the expected molecular weight was
observed (Figure 5B), indicating a rapid labeling with high
specificity among all the examined cellular protein targets in
different mammalian cell lines.
With A-TMP-Fl v2.0 in hand, we performed labeling
experiments to evaluate the performance of this covalent
TMP-tag compared to noncovalent TMP-tag (Figure 6). HEK
293T Cells transiently expressing Tomm20-eDHFR and
Tomm20-eDHFR:L28C were labeled with TMP-Fl17 and A-
13696
dx.doi.org/10.1021/ja303374p | J. Am. Chem. Soc. 2012, 134, 13692−13699

Journal of the American Chemical Society
Article
405 nm excitation laser (Figure 7B). To test the labeling
efficiency of the second-generation covalent TMP-tag, HEK
Figure 6. Comparative studies of labeling with noncovalent TMP-tag
and covalent TMP-tag. (A) HEK 293T cells expressing Tomm20-
eDHFR were labeled with TMP-Fl as in Figure 5A. After labeling, cells
were imaged using confocal microscope (Live panel) or fixed with 4%
paraformaldehyde in PBS for 10 min followed by washing with PBS for
12 h before imaged (Fixed panel). (B) HEK 293T cells expressing
Tomm20-eDHFR:L28C were labeled with A-TMP-Fl and examined
with and without fixation treatment as in (A). Scale bars are 25 μm.
Figure 7. Labeling of cellular protein targets with A-TMP-Atto655 and
A-TMP-Dapoxyl. (A) HEK293T cells transiently expressing H2B-
eDHFR:L28C or PMLS-eDHFR:L28C were incubated with 1 μM A-
TMP-Atto 655 in media for 3 h, washed twice, and imaged using
confocal and differential interference contrast (DIC) microscopy. Atto
655 was excited at 633 nm. (B) A-TMP-Dapoxyl was tested under the
same conditions as in (A), except Dapoxyl was excited at 405 nm.
Scale bars are 25 μm. (C) Western blot analysis of the labeling
efficiency. HEK293T cells transiently expressing eDHFR:L28C-6X His
were incubated with 1 μM A-TMP-Dapoxyl in media for 10 min, 30
min, and 3 h before being lysed and analyzed by SDS-PAGE/Western
blot with 6X His antibody.
293T cells expressing eDHFR:L28C-6X His were incubated
with media containig 1 μM A-TMP-Dapoxyl. At certain time
points, cells were harvested, lysed, analyzed by SDS-PAGE, and
blotted with Anti-6X His. According to the band shift, near-
quantitative labeling was achieved under the condition of 3 h
incubation (Figure 7C). These labeling assays pave the way
toward the development of novel biophysical, physiological,
and multicolor pulse−chase applications with TMP-tag.
designed for labeling purified proteins, cannot provide the
desired selectivity for labeling proteins within the cell. More
recently introduced bio-orthogonal reactions, such as the
copperless click reaction37 or the photoinduced reactions
which are triggered by UV-irradiation,38 require unnatural
amino acid incorporation,39 and are technically demanding
and/or damaging to cells. In comparison, the noncovalent
eDHFR-TMP interaction specifically accelerates the covalent
reaction between the engineered Cys on eDHFR and the
acrylamide electrophile. This approach, conceptually resem-
bling the biomolecule-templated organic reactions,40 expands
the scope of bioconjugation reactions as well as synthetic
biology.
From an engineering point of view, proximity-induced
reactions are facile implements for the development of novel
chemical tags. In the case of the covalent TMP-tag, the
specificity between ligand (A-TMP-probe) and receptor
(eDHFR) is guaranteed by high-affinity enzyme−inhibitor
recognition as opposed to heavy-metal chelations (FlAsH/
ReAsH)5 or additional enzyme-catalyzed reactions (PRIME).10
Significantly, the covalent TMP-tag, which is based on high-
affinity binding, exhibits superior specificity and efficiency that
enables intracellular protein labeling with minimal background.
Using similar approaches, the vast pool of bioactive natural
DISCUSSION
■
Together these results establish that, by improving the design
and hence the reaction half-life of our covalent TMP-tag, the
v2.0 covalent TMP-tag is now a robust and general reagent for
live cell imaging. The covalent TMP-tag design was improved
by optimizing the spatial positioning of the Michael addition
pairthe engineered Cys residue on eDHFR and the
acrylamide conjugated to the A-TMP-probe heterotrimer.
While numerous chemical tags have been reported, our v2.0
covalent TMP-tag is one of the few examples that is selective
enough to enable high signal-to-noise imaging of intracellular
proteins, particularly diffuse, cytoplasmic proteins. Impressively,
this speed and selectivity is achieved not from enzyme catalysis,
but rather from ligand−receptor binding followed by a
proximity-induced organic reaction using a well-chosen mild
electrophile. To date, the v2.0 covalent TMP-tag has been
successful for a variety of protein targets and mammalian cell
lines, and we expect it to be broadly useful to the community
for imaging a wide range of proteins in living cells.
From a chemical perspective, proximity-induced reactions
offer a combination of reactivity and specificity, which are both
critical for protein labeling in live cells. Traditional protein-
conjugation reagents such as maleimide electrophiles, which are
13697
dx.doi.org/10.1021/ja303374p | J. Am. Chem. Soc. 2012, 134, 13692−13699

Journal of the American Chemical Society
Article
Present Address
products and hit compounds from combinatorial libraries could
be potentially engineered into orthogonal chemical tags based
on proximity-induced reactions.41
§Present address: Environmental Protection Agency, Wash-
ington, DC 20004, United States.
With an in vitro labeling half-life of 8 min, the second-
generation covalent TMP-tag is significantly improved over our
first-generation design.20 Although the reaction rate is slower
than the suicide-substrate-based tags, e.g. SNAP tag,6 we
consider the rate difference of little practical significance given
that it typically requires 10 min to over an hour to label
proteins in living cells, with uptake of the organic fluorophore
considered to be the rate-limiting step. If needed, however, the
labeling reaction kinetics could likely be further optimized by
either molecular engineering of the small-molecule ligand or
directed evolution of eDHFR, or both. Notably, an advantage
to a chemical tag based on high-affinity binding is that it does
not require the high concentration of ligand−probe conjugate
necessary with enzyme-based chemical tags, where KM’s
typically range from micromolar to millimolar, leading to
high background noise from unbound fluorophore and
necessitating extensive washing steps.
The second-generation covalent TMP-tag reported here is
seen as a pressing improvement of the TMP-tag toward
advanced protein-labeling applications. With its improved
labeling kinetics and well-demonstrated cellular behavior, one
might be able to track single-protein molecules inside a cell42
with a fluorophore of high photon count. Moreover, the
viability and robustness of the second-generation covalent
TMP-tag point the way to multicolor protein labeling using
orthogonal chemical tags.
Notes
The authors declare the following competing financial
interest(s): V.W.C. and M.P.S. hold patents on the TMP-tag
technology, and the technology is licensed and commercialized
by Active Motif.
ACKNOWLEDGMENTS
■
We acknowledge Julia Sable, Rohitha Sriramaratnam, Reka
Letso, Kenichi Shimada and Yao Zong Ng for experimental
assistance and Tracy Y. Wang for helpful discussions. This work
was supported by the National Institutes of Health (RC1
GM091804 and U54 GM087519 to V.W.C). S.G. was
supported by a National Defense Science and Engineering
Graduate fellowship.
REFERENCES
(1) Jing, C.; Cornish, V. W. Acc. Chem. Res. 2011, 44, 784.
(2) Hinner, M. J.; Johnsson, K. Curr. Opin. Biotechnol. 2010, 21, 766.
■
(3) Fernan
9, 929.
́ ́
dez-Suarez, M.; Ting, A. Y. Nat. Rev. Mol. Cell. Biol. 2008,
(4) Giepmans, B. N. G.; Adams, S. R.; Ellisman, M. H.; Tsien, R. Y.
Science 2006, 312, 217.
(5) Griffin, B. A.; Adams, S. R.; Tsien, R. Y. Science 1998, 281, 269.
(6) Keppler, A.; Gendreizig, S.; Gronemeyer, T.; Pick, H.; Vogel, H.;
Johnsson, K. Nat. Biotechnol. 2003, 21, 86.
(7) Gautier, A.; Juillerat, A.; Heinis, C.; Correa, I. R., Jr.;
Kindermann, M.; Beaufils, F.; Johnsson, K. Chem. Biol. 2008, 15, 128.
(8) Los, G. V.; Encell, L. P.; McDougall, M. G.; Hartzell, D. D.;
Karassina, N.; Zimprich, C.; Wood, M. G.; Learish, R.; Ohana, R. F.;
Urh, M.; Simpson, D.; Mendez, J.; Zimmerman, K.; Otto, P.; Vidugiris,
G.; Zhu, J.; Darzins, A.; Klaubert, D. H.; Bulleit, R. F.; Wood, K. V.
ACS Chem. Biol. 2008, 3, 373.
(9) Miller, L. W.; Cai, Y.; Sheetz, M. P.; Cornish, V. W. Nat. Methods
2005, 2, 255.
(10) Uttamapinant, C.; White, K. A.; Baruah, H.; Thompson, S.;
́ ́
Fernandez-Suarez, M.; Puthenveetil, S.; Ting, A. Y. Proc. Natl. Acad.
Sci. U.S.A. 2010, 107, 10914.
(11) Hoskins, A. A.; Friedman, L. J.; Gallagher, S. S.; Crawford, D. J.;
Anderson, E. G.; Wombacher, R.; Ramirez, N.; Cornish, V. W.; Gelles,
J.; Moore, M. J. Science 2011, 331, 1289.
(12) Long, M. J. C.; Pan, Y.; Lin, H.-C.; Hedstrom, L.; Xu, B. J. Am.
Chem. Soc. 2011, 133, 10006.
(13) Zou, P.; Ting, A. Y. ACS Chem. Biol. 2011, 6, 308.
(14) Wombacher, R.; Heidbreder, M.; van de Linde, S.; Sheetz, M.
P.; Heilemann, M.; Cornish, V. W.; Sauer, M. Nat. Methods 2010, 7,
717.
CONCLUSION
■
By improving the reaction geometry of our covalent TMP-tag,
we now have a v2.0 covalent TMP-tag that is a robust cellular
reagent. This v2.0 covalent TMP-tag is an important addition
to the limited arsenal of orthogonal covalent chemical tags
available for multicolor imaging. Because our covalent TMP-tag
is based on a modular organic reaction rather than a specific
enzyme modification, we expect to be able to more readily
build additional features into the covalent TMP-tag and
generate new orthogonal tags simply by extending the Michael
addition reaction to other drug−receptor pairs. While used here
for live cell imaging, the covalent TMP-tag can be used broadly
as a biotin−avidin surrogate for in vitro applications or in other
applications of chemical dimerizers in live cells. Beyond the
utility of proximity-induced reactivity for chemical tag
engineering, the excellent reactivity and specificity of the
proximity-induced Michael addition in a live cell illustrates the
potential of organic chemistry for synthetic biology.
(15) Baccanari, D. P.; Daluge, S.; King, R. W. Biochemistry 1982, 21,
5068.
ASSOCIATED CONTENT
* Supporting Information
■
(16) Mattews, D. A.; Bolin, J. T.; Burridge, J. M.; Filman, D. J.; Volz,
K. W.; Kaufman, B. T.; Beddell, C. R.; Champness, J. N.; Stammers, D.
K.; Kraut, J. J. Biol. Chem. 1985, 260, 381.
(17) Calloway, N. T.; Choob, M.; Sanz, A.; Sheetz, M. P.; Miller, L.
W.; Cornish, V. W. ChemBioChem 2007, 8, 767.
(18) Gallagher, S. S.; Jing, C.; Peterka, D. S.; Konate, M.;
Wombacher, R.; Kaufman, L. J.; Yuste, R.; Cornish, V. W.
ChemBioChem 2010, 11, 782.
(19) Cai, Y.; Rossier, O.; Gauthier, N. C.; Biais, N.; Fardin, M.-A.;
Zhang, X.; Miller, L. W.; Ladoux, B.; Cornish, V. W.; Sheetz, M. P. J.
Cell Sci. 2010, 123, 413−423.
(20) Gallagher, S. S.; Sable, J. E.; Sheetz, M. P.; Cornish, V. W. ACS
Chem. Biol. 2009, 4, 547.
S
Model estimation of the optimal linker length between TMP
and acrylamide; synthetic procedures and characterization of
each compound; vector construction, protein expression, and
purification procedure; in vitro alkylation protocol and gel
pictures; cell culture, transfection and labeling protocol; live cell
imaging protocol and in-gel fluorescence scanning/Western
blotting protocol. This material is available free of charge via
the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
vc114@columbia.edu
■
(21) Breslow, R.; Corcoran, R.; Dale, J. A.; Liu, S.; Kalicky, P. J. Am.
Chem. Soc. 1974, 96, 1973.
13698
dx.doi.org/10.1021/ja303374p | J. Am. Chem. Soc. 2012, 134, 13692−13699

Journal of the American Chemical Society
Article
(22) Breslow, R.; Corcoran, R. J.; Snider, B. B.; Doll, R. J.; Khanna, P.
L.; Kaleya, R. J. Am. Chem. Soc. 1977, 99, 905.
̈
(23) Powers, J. C.; Asgian, J. L.; Ekici, O. D.; James, K. E. Chem. Rev.
2002, 102, 4639.
(24) Fry, D. W.; Bridges, A. J.; Denny, W. A.; Doherty, A.; Greis, K.
D.; Hicks, J. L.; Hook, K. E.; Keller, P. R.; Leopold, W. R.; Loo, J. A.;
McNamara, D. J.; Nelson, J. M.; Sherwood, V.; Smaill, J. B.; Trumpp-
Kallmeyer, S.; Dobrusin, E. M. Proc. Natl. Acad. Sci. U.S.A. 1998, 95,
12022.
(25) Evans, M. J.; Cravatt, B. F. Chem. Rev. 2006, 106, 3279.
(26) Levitsky, K.; Boersma, M. D.; Ciolli, C. J.; Belshaw, P. J.
ChemBioChem 2005, 6, 890.
(27) Krusemark, C. J.; Belshaw, P. J. Org. Biomol. Chem. 2007, 5,
2201.
(28) Sawaya, M. R.; Kraut, J. Biochemistry 1997, 36, 586.
(29) Polshakov, V. I.; Smirnov, E. G.; Birdsall, B.; Kelly, G.; Feeney, J.
J. Biomol. NMR 2002, 24, 67.
(30) Maestro, version 9.1; Schrodinger, LLC: New York, NY, 2010.
̈
(31) DeLano, W. L. The PyMOL Molecular Graphics System, DeLano
Scientific: Palo Alto, CA, 2002.
(32) Izumi, S.; Urano, Y.; Hanaoka, K.; Terai, T.; Nagano, T. J. Am.
Chem. Soc. 2009, 131, 10189.
(33) Veech, R. L.; Eggleston, L. V.; Krebs, H. A. Biochem. J. 1969,
115, 609.
(34) Reiss, P. D.; Zuurendonk, P. F.; Veech, R. L. Anal. Biochem.
1984, 140, 162.
(35) Diwu, Z.; Zhang, C.; Klaubert, D. H.; Haugland, R. P. J.
Photochem. Photobiol. A 2000, 131, 95.
(36) Pham, W; Pantazopoulos, P.; Moore, A. J. Am. Chem. Soc. 2006,
128, 11736.
(37) Agard, N. J.; Prescher, J. A.; Bertozzi, C. R. J. Am. Chem. Soc.
2004, 126, 15046.
(38) Lim, R. K. V.; Lin, Q. Acc. Chem. Res. 2011, 44, 828.
(39) Hao, Z.; Hong, S.; Chen, X.; Chen, P. R. Acc. Chem. Res. 2011,
44, 742.
(40) Li, X.; Liu, D. R. Angew. Chem., Int. Ed. 2004, 43, 4848.
(41) Choi, S.; Ong, D. S. T.; Kelly, J. W. J. Am. Chem. Soc. 2010, 132,
16043.
(42) Li, G. -W.; Xie, X. S. Nature 2011, 475, 308.
13699
dx.doi.org/10.1021/ja303374p | J. Am. Chem. Soc. 2012, 134, 13692−13699