Journal of the American Chemical Society
Article
cancer cells.41 To evaluate whether our labeling strategy is
applicable to nonhistone substrates, we specifically constructed
and expressed the truncated form of EZH2 (1−500) in
Escherichia coli, with a shake-flask yield of approximately 2 mg/
L. The protein’s purity and size were confirmed by SDS-PAGE
(Figure S16). Reaction of the EZH2 fragment under the same
conditions as those for histone proteins resulted in a similar
type of tagging, as evident by Figure 3C. Taken together, these
observations displayed the steric-free in vitro labeling and
tagging of a range of known protein substrates using the FTDR
reaction, which could not be achieved by the known CuAAC
method.
For labeling proteins in living cells, azide or alkyne analogs
of fatty acids have been exploited, which were metabolized
intracellularly into CoA derivatives.3−5,10 For an increase of
their cellular delivery, pro-metabolites with esters masking the
polar carboxylate group constituted an effective strategy in
recent years.42 Nevertheless, the intrinsic sterics of these pro-
metabolites resulted in varied and suboptimal labeling
results,10,42 sometimes requiring extensive structural optimiza-
tion.42 To explore the utility of our probing system for
studying acetylation in the cellular level, we designed the
fluorinated version of pro-metabolite, ethyl fluoroacetate
(Figure 4A). Given the in vivo toxicity of fluoroacetate,21 we
first evaluated the cell cytotoxicity and found that this pro-
metabolite exhibited minimal toxicity with doses up to 2 mM
after 12 h of incubation (Figure S17). Additional LC-MS/MS
studies indeed confirmed its conversion by enzymes to
fluoroacetyl−CoA in live cells (Figure S18), which, taken
together with the observed minimal toxicity in cell lines, could
support the applicability of ethyl fluoroacetate and its CoA
metabolite to studies at the cellular level.
With confidence in the safety profile of our pro-metabolite,
we started by treating it with two representative cell lines,
HeLa and HEK293, followed by subsequent FTDR with the
TAMRA−SH probe (15) as the second step for fluorescent
detection (Figure 4A). Treatment with the azido-modified pro-
metabolite42 and the TAMRA−alkyne (16) for CuAAC
chemistry was performed in parallel as a control, wherein
weak signals were previously reported.42 Direct microscope
imaging studies (Figure 4B) revealed much stronger intra-
cellular labeling and tagging with TAMRA following the
FTDR-based approach, suggesting a drastically improved,
more complete profiling of acetylation substrates. We also
observed significant fluorescence not only in the nucleus but
also in the cytoplasm, which may indicate the successful
labeling of both histones and nonhistone proteins. The little
background signals emitted from the cells treated with only 15
(step 2) further demonstrated the specificity of the developed
−SH probes. Following similar procedures, we also tested the
labeling of the cell lysates after metabolic incorporation of
fluorine reporters. Lysates following FTDR or CuAAC
mediated ligation with TAMRA were separated on SDS-
PAGE and visualized by CBB to confirm an equal amount of
protein loading (Figure 4C). Yet, multiple labeled protein
bands spanning a wide range of molecular weights were
observed only for the lysates of cells that underwent a
complete two-step process of FTDR (Figure 4C). This
discovery was consistent with the microscope imaging results.
Furthermore, pretreatment of the cells with A-485 (potent and
selective p300/CBP inhibitor)43,44 (Figure 4C) before the
two-step process of FTDR also resulted in weaker labeling
intensity, which is consistent with the previously reported
results using HATi.10,42,45−49 In addition, we exploited the
concurrent incubation with HDAC inhibitor cocktails and
observed slightly decreased labeling (Figure S19), which was
consistent with literature reports,9,10,42 suggesting that
incorporation of F-acetylation needs prior deacetylation of
intrinsically acetylated lysine residues to make the lysine
available. Thus, blocking the removal of wild-type acetylation
prevents metabolic incorporation of F-acetylation.9,10,42 Taken
together, these observations fully supported that FTDR
allowed for profiling the proteome-wide substrates of
acetylation from the cellular contexts.
To further validate that the FTDR-based two-step metabolic
labeling occurs on acetylation protein substrates, we performed
histone extraction (Figure 5A) and confirmed the existence of
TAMRA labeling on the primary acetylation substrates,
histones (Figure 5B). As controls, both the treatment with
HAT inhibitors and the competition with acetate have resulted
in decreased TAMRA−SH probe labeling, suggesting that the
FTDR-based two-step labeling is acetyltransferase-dependent
and relies on F−Ac−CoA metabolites. To test whether the
FTDR-based labeling can be used to enrich these protein
substrates, we treated cell lysates with the biotin−SH probe in
step 2 and pulled down the labeled proteins (Figure 5A).
Western blot analysis confirmed the presence of known
Figure 4. Cellular evaluation of FTDR-based tagging with TAMRA-
SH probe. (A) Scheme for cellular pro-metabolite incorporation (1
mM, 6 h, at 37 °C, step 1) and protein substrate detection (step 2).
(B) Fluorescent microscopy of fixed and permeabilized cells that were
stained by Hoechst 33342 (blue) and TAMRA probes (red); scale
bars: 25 μm. (C) Cell lysate protein labeling by pro-metabolites and
detection by TAMRA probes (red); left panel, PAGE gel stained by
CBB; right panel, in-gel fluorescent detection. “C” is the positive
control for CuAAC, which randomly labeled lysines on BSA with
azide−NHS ester, followed by CuAAC-mediated conjugation with
TAMRA−alkyne. “HATi” indicates the addition of HAT inhibitor (A-
485) prior to step 1.
1344
J. Am. Chem. Soc. 2021, 143, 1341−1347