Improved Stabilities of Labeling Probes for the Selective Modification of Endogenous Proteins in Living Cells and In Vivo

Article abstract of DOI:10.1002/asia.202100060

To date, various affinity-based protein labeling probes have been developed and applied in biological research to modify endogenous proteins in cell lysates and on the cell surface. However, the reactive groups on the labeling probes are also the cause of

Full text of DOI:10.1002/asia.202100060

DOI: 10.1002/asia.202100060
Full Paper
Improved Stabilities of Labeling Probes for the Selective
Modification of Endogenous Proteins in Living Cells and In Vivo
Kuan-Yu Lin+,[a] Chak Hin Lam+,[a] Xin-Hui Lin,[a] Jung-I Hsu,[a] Syuan-Yun Fan,[a]
Nitesh K. Gupta,[a] Yu-Chun Lin,[a] Boon Khoon Tee,[a] Jui-Ping Li,[d] Jen-Kun Chen,[d] and
Kui-Thong Tan*[a, b, c]
Abstract: To date, various affinity-based protein labeling
probes have been developed and applied in biological
research to modify endogenous proteins in cell lysates and
on the cell surface. However, the reactive groups on the
labeling probes are also the cause of probe instability and
nonselective labeling in a more complex environment, e.g.,
intracellular and in vivo. Here, we show that labeling probes
composed of a sterically stabilized difluorophenyl pivalate
can achieve efficient and selective labeling of endogenous
proteins on the cell surface, inside living cells and in vivo. As
compared with the existing protein labeling probes, probes
with the difluorophenyl pivalate exhibit several advantages,
including long-term stability in stock solutions, resistance to
enzymatic hydrolysis and can be customized easily with
diverse fluorophores and protein ligands. With this probe
design, endogenous hypoxia biomarker in living cells and
nude mice were successfully labeled and validated by in vivo,
ex vivo, and immunohistochemistry imaging.
Introduction
specific protein-ligand interaction, triggering a chemical reac-
tion via the proximity effect between the reactive group and an
amino acid near the binding site of the POI. To date, a variety of
chemically and photochemically reactive electrophiles have
been incorporated in the affinity-based protein labeling probes,
including activated esters, sulfonyl chlorides, epoxides, malei-
mides and α-halocarbonyls, as well as photo-activatable groups,
such as aryl azide, diazirin, and benzophenone.[3] In addition to
these reactive modules, Hamachi and co-workers have also
developed several cleavable electrophiles to create ligand-
directed traceless labeling probes. In this approach, the ligand
can be easily released from the binding site of the POI upon
labeling to allow the POI to retain its native activity and
function. Several other electrophiles have been employed in
the traceless labeling probes, including tosyl, acyl imidazole,
dibromophenyl benzoate, N-sulfonyl pyridone, and N-acyl-N-
alkyl sulfonamide.[4] Although these reactive moieties are
valuable for the development of protein labeling probes for
protein imaging, proteomics analysis, irreversible inhibition of
protein activity, and biosensor construction, the reactive nature
of such electrophiles is also the root cause of many selectivity
and stability problems with these probes.[5]
Protein modification with small molecules is an important
strategy to realize many biological and medical applications,
including the protein imaging in complex biological systems,
functional studies of individual proteins, and development of
new biopharmaceuticals.[1] To achieve selective chemical mod-
ification of target proteins in complex biological contexts, e.g.,
cells, tissues and animals, most strategies rely on protein/
peptide tags and bioorthogonal chemistry-based methods.[2]
Although these approaches are versatile and robust, the major
limitation is that they are only applicable to genetically-
encoded proteins, and cannot be used directly to study
endogenous proteins in living cells and in vivo.
In contrast, the chemical modification of the target protein
using affinity-based protein labeling probes can provide an
opportunity to detect and study endogenous proteins in the
crowded and multimolecular cellular environments. In a typical
design, the probe binds to the protein of interest (POI) via a
[a] K.-Y. Lin,+ C. Hin Lam,+ X.-H. Lin, J.-I Hsu, S.-Y. Fan, Dr. N. K. Gupta, Y.-C. Lin,
B. Khoon Tee, Prof. Dr. K.-T. Tan
Ideally, biocompatible electrophiles exhibiting fast reaction
are desirable for efficient protein labeling. However, such high
reactivity may produce significant nonspecific reactions due to
the complex nature of biological systems. Therefore, despite
the success of several affinity-based protein labeling probes for
the modification of endogenous proteins in live cells, covalent
modification of in vivo endogenous proteins remains a formida-
ble task and have yet to be applied directly for endogenous
protein of tumor cells in living animals.
In this paper, we report a novel phenyl ester moiety based
on a bulky difluorophenyl pivalate electrophile to create steri-
cally stabilized and highly selective affinity-based traceless
protein labeling probes for the covalent modification of
Department of Chemistry, National Tsing Hua University, 101 Section 2,
Kuang Fu Road, Hsinchu 30013 (Taiwan)
E-mail: kttan@mx.nthu.edu.tw
[b] Prof. Dr. K.-T. Tan
Frontier Research Center on Fundamental and Applied Sciences of Matters,
National Tsing Hua University, 101 Section 2, Kuang Fu Road, Hsinchu
30013 (Taiwan)
[c] Prof. Dr. K.-T. Tan
Department of Medicinal and Applied Chemistry, Kaohsiung Medical
University, Kaohsiung 807 (Taiwan)
[d] J.-P. Li, Dr. J.-K. Chen
Institute of Biomedical Engineering and Nanomedicine, National Health
Research Institutes, Miaoli 35053 (Taiwan)
[+] These authors contributed equally.
Supporting information for this article is available on the WWW under
https://doi.org/10.1002/asia.202100060
Chem Asian J. 2021, 16, 937–948
937
© 2021 Wiley-VCH GmbH

Full Paper
endogenous proteins in living cells and in vivo (Figure 1). The
phenyl esters can potentially modify lysine residues, which
occurs abundantly on protein surfaces and are nucleophilic
enough under physiological conditions. Previously, several
fluorescent probes based on phenyl ester have been applied in
the labeling of different protein targets.[4c,6] Despite their high
selectivity, the phenyl ester probes degraded rapidly in the
presence of esterase and were prone to hydrolysis. For broader
protein labeling applications, the phenyl ester probe must
exhibit both sufficient labeling and resistance to the catalytic-
and auto-hydrolysis in a complex environment, e.g. intracellular
and in vivo conditions.
Herein, we show that our design for an affinity-based
protein labeling probe based on a difluorophenyl pivalate
electrophile (Figure 2a and Figure 4a) can be customized easily
with various fluorophores and ligands to label selectively and
efficiently different target proteins, e.g. carbonic anhydrase
(CA), avidin and E.coli dihydrofolate reductase (eDHFR), in cell
lysates, living cells and living animals. Our probes exhibit long
term stability in aqueous buffer and stock solutions as well as
resistance to catalytic hydrolysis by esterase. Besides the
labeling of endogenous proteins in intracellular and in vivo
environments, we also showed that the difluorophenyl pivalate
probes can be used to quantify endogenously expressed cell
surface hCAIX and hCAXII proteins under normoxia and hypoxia
conditions.
ure 2a, R1=H, F, Cl and Br). Since the reported experimental pKa
values of phenol, ortho-difluorophenol, ortho-dichlorophenol
and ortho-dibromophenol are 9.9, 7.51, 6.78 and 6.67, respec-
tively, the probe reactivity should increase as the pKa of the
phenol decreases.[7] The stability was further optimized by using
a sterically hindered pivalate moiety to minimize autolysis and
enzymatic hydrolysis. Based on this principle, we designed five
different probes 1–5 containing a sulfonamide ligand and a
synthetic fluorescein dye. Human carbonic anhydrase II (hCAII)
and sulfonamide were chosen as the initial target-ligand pair
because their binding is known to be highly specific with Ki
value of about 260 nM.[8] The synthesis of probes 1–5 can be
accomplished in 7 steps, where the final step involves the
coupling of the sulfonamide ligand via Huisgen 1,3-dipolar
cycloaddition (Scheme S1–S9). This synthetic design facilitates
the rapid and simple interchange of the ligand for the labeling
of other target proteins in the future.
To examine which probe is the most effective for the
labeling of hCAII, the labeling reactions of probes 1–5 (5 μM)
and purified recombinant hCAII (5 μM) were conducted and
analyzed by SDS-PAGE in-gel fluorescence and coomassie
brilliant blue staining. As shown in Figure 2b, fluorescence was
clearly observed from the bands for hCAII (MW~32KDa, Fig-
ure S1) after a 2 hours incubation with 2 and 3, whereas weak
fluorescence were observed with 1, 4 and 5. The fluorescence
intensity was reduced dramatically when a competitive inhibitor
of hCAII, ethoxzolamide (EZA) was added, indicating that these
labeling reactions were driven by a specific ligand-protein
interaction. In contrast to 2 and 3, the fluorescent bands for 1, 4
and 5 remained very weak even after 4 hours of incubation.
The protein labeling level by each of the five probes was
quantified by measuring the mean fluorescence intensity of the
bands pixel-by-pixel using ImageJ software. The results showed
that the least sterically hindered 2 is the most efficient at
labeling hCAII. Based on the fluorescence intensity generated
by 2, about 40% and 60% of hCAII were labeled by 3 after 2
and 4 hours incubation. Comparatively, the hCAII protein label-
ing levels for 1, 4 and 5 were merely 2%, 6%, and 3% after
Results and Discussion
Molecular design and reactivity of the affinity-based phenyl
ester probes
The use of phenyl ester as a cleavable electrophilic group allows
the stability and reactivity of the probe to be optimized at the
phenol and the carboxylate sites, respectively. We sought to
modulate the reactivity of the probe by introducing ortho-
substituted dihalogen phenols as good leaving groups (Fig-
Figure 1. Schematic illustration of an affinity-based traceless protein labeling probe based on the difluorophenyl pivalate activated ester as a reactive
electrophile. In the presence of a specific protein-ligand interaction, an acylation reaction between the difluorophenyl pivalate group and an amino acid near
the active pocket of the target protein can occur via proximity effect.
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
938
© 2021 Wiley-VCH GmbH

Full Paper
Figure 2. Molecular design and reactivity analyses of the affinity-based phenyl ester probes to label hCAII protein. (a) Chemical structures of protein labeling
probes 1–5 for hCAII containing sulfonamide ligand and fluorescein dye. (b) Analysis of the covalent labeling of purified hCAII with 1–5. Reaction conditions:
°
hCAII (5 μM) and labeling probes (5 μM) were incubated in the absence or presence of 100 μM EZA in PBS buffer (pH 7.4) at 37 C. The gel was analyzed by in-
gel fluorescence imaging and stained with Coomassie brilliant blue (CBB).
4 hours of reaction. In general, the reactivity of the probes
increases with decreasing steric hindrance (e.g. 2 and 3) and
pKa values of the phenol leaving groups (e.g. 1 and 3).
However, the impact of lower pKa values on probe reactivity
can be offset by the increased steric hindrance caused by the
larger molecular size of chloro and bromo atoms (the CPK
model is shown in Figure S2). Thus, the difluorophenyl ester
moiety on 2 and 3 provides a good balance between electro-
philicity and steric hindrance to accomplish efficient protein
labeling. We also confirmed that the labeling of hCAII with the
phenyl ester probes occurs in a traceless manner by using
activity assay and MALDI-TOF MS analysis. In the activity assay,
we found that the enzymatic activity of hCAII to hydrolyze p-
nitrophenol acetate was not affected after labeling with 2
(Figure S3). As for the MALDI-TOF MS analysis, we were able to
observe the molecular peak of the released sulfonamide ligand
upon labeling of hCAII with 2 and 3 (Figure S4). From the
previous results, we believe that the labeling sites for hCAII by
probe 3 should also occur at Lys 190 and 192, respectively.[6b]
Stability, labeling kinetic and selectivity of the affinity-based
difluorophenyl pivalate probes
Maintaining the chemical integrity of small molecule probes in
stock solutions for long-term storage and freeze/thaw cycles is
a fundamental concern for the reproducibility of the results. It is
known that a certain proportion of chemical compounds
decompose over time when stored in DMSO, which is the most
common solvent employed.[9] To this end, we investigated the
stabilities of 2 and 3 stored in DMSO by HPLC analysis. The
results from the HPLC chromatogram showed that 3 can
maintain up to 98% purity after 12 months of storage in DMSO
°
(1 mM) at À 80 C with no signs of degradation (Figure 3a and
Figure S5). Furthermore, 3 was also stable in MeOH or to be
°
stored as a dry form at À 80 C for at least 18 months (Figure S6).
On the other hand, 2 decomposed slowly in DMSO, leaving
°
with only 90% purity after 4 months of storage at À 80 C
(Figure S7). In comparison, activated esters that are based on N-
hydroxysuccinimide (NHS, pKa=6.0) and pentafluorophenol
(pKa=5.5) are not stable in aqueous buffer and DMSO
solvent.[10]
Next, we evaluated the stabilities of 2 and 3 under the
labeling conditions and in the presence of esterase. Through
HPLC analysis, we determined that only 6% of 2 and 3% of 3
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
939
© 2021 Wiley-VCH GmbH

Full Paper
°
Figure 3. Stability, labeling kinetic and selectivity of 2 and 3. (a) HPLC trace of 3 on day 1 and after storage in DMSO at À 80 C for 12 months. (b) HPLC trace
°
analyses of 2 and 3 in the absence or presence of PLE (330 nM) in pH 7.4 PBS buffer at 37 C. Error bars were calculated from three measurements. (c) SDS-
°
PAGE analysis of hCAII labeling with 5 μM 3 in 6 mg/mL E. coli lysates at 37 C for 2 hours. (d) Time course of fluorescence increase for the reaction of hCAII
(2.5 μM each) with 2 and 3 (25 μM each) at 37 C in pH 7.4 PBS buffer. The inset shows the kinetic plot of the apparent reaction rate constant kobs (sÀ 1) versus
°
various concentrations of 2 and 3 (R2 =0.99).
°
were hydrolyzed after 12 hours of incubation at 37 C in PBS
buffer (Figure 3b and Figure S8). While 2 and 3 exhibit similar
stabilities under the labeling condition, their resistances to
hydrolysis by esterase are completely dissimilar. Upon the
addition of 330 nM (~1 U/mL) porcine liver esterase (PLE), 2
decomposed completely after 2 hours, whereas 96% of 3
remained after 4 hours of incubation (Figure S9). The ability of 3
to resist catalytic hydrolysis by PLE prompted us to investigate
the efficiency of 3 to label hCAII under high concentrations of
PLE (Figure S10). The results showed that even in the presence
of 10 μM PLE, 3 can still achieve about 80% of hCAII labeling. As
PLE is a reactive esterase with high catalytic efficiency (kcat/Km =
1×106 MÀ 1sÀ 1 for fluorescein diacetate), these results also
suggest that the probe should be able to resist catalytic
hydrolysis by intracellular and in vivo esterase.[11] Taken togeth-
er, we believe that the high stability of the probe in DMSO
stock solution and in the presence of esterase is due to the
synergetic effect of the difluorophenol’s moderate pKa value
(pKa=7.51) and the bulky pivalate group.
Having optimized probe 3 as a highly stable labeling
reagent, we attempted to evaluate its protein labeling
selectivity under crude conditions. The labeling reactions were
carried out with 5 μM 3 using cell lysates of E. coli (6 mg/mL)
with overexpressed hCAII (Figure 3c). A clear single fluorescent
band corresponding to hCAII appeared when the purified hCAII
and lysates were incubated with 3, which was not visible in the
presence of EZA. Although 2 is more efficient in labeling its
target protein, it also suffers from significant levels of non-
specific labeling in the crude mixture (Figure S11). In fact, this
problem is prevalent among protein labeling probes which
employed highly reactive electrophiles. Therefore, probe 3
which is based on the difluorophenyl pivalate moiety offers an
optimal balance between reactivity and selectivity to label the
target protein in complex biological environments.
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
940
© 2021 Wiley-VCH GmbH

Full Paper
Figure 4. (a) Chemical structures of protein labeling probes 6, 7 and 8 which are used for the labeling of hCAII, avidin and eDHFR, respectively. (b) SDS-PAGE
in-gel fluorescent analysis of the covalent labeling of hCAII, avidin and eDHFR (5 μM each) with equal concentration of 6, 7 and 8, respectively, in 5 mg/mL E.
°
coli lysates at 37 C for 2 hours. EZA, biotin and methotrexate (100 μM each) were used as an inhibitor for inhibiting covalent labeling of hCAII, avidin and
eDHFR, respectively.
To better understand the high stability and selectivity of 3
to label hCAII under esterase and lysate conditions, kinetic
analyses of the labeling reactions with hCAII were carried out
by monitoring the increase of fluorescence intensity over time
at SDS-PAGE gel (Figure S12). As expected, the less sterically
hindered 2 exhibits faster labeling kinetic with hCAII, achieving
full labeling within 1 hour (Figure 3d). For 3, full labeling was
accomplished after 6 hours of reaction. The second-order rate
constants (k2) for 2 and 3 were determined to be about 210 and
7.4 MÀ 1sÀ 1, respectively (Figure 3d, inset). With the given k2
values, the reaction rate of difluorophenyl pivalate is compara-
ble to the copper-catalyzed azide-alkyne cycloaddition as well
as LDAI and LDT chemistry developed by Hamachi and
coworkers.[3a,12] It is significant to note that with this labeling
rate, difluorophenyl pivalate protein probes are able to label
very low amounts of target protein, achieving protein detection
limit of around 1 ng (Figure S13). We also found that the
reaction kinetic is correlated with the binding affinity between
the ligand and the target protein. Due to the existence of a
weak binding between sulfonamide and BSA (Kd ~1–10 μM), the
k2 value of 3 with BSA was determined to be around
0.70 MÀ 1sÀ 1 which is about 10 times smaller than the reaction
with hCAII (Figure S14).[13] This indicates that difluorophenyl
pivalate probes can also be used to label proteins with weak
binding affinity.
Traceless labeling of hCAII, avidin and eDHFR proteins in cell
lysates
To show the modular nature of our difluorophenyl pivalate
probe design, we replaced the fluorescein and sulfonamide
with the probes that contain Cy5 dye and various other ligands,
e.g. sulfonamide, biotin and methotrexate, to generate probe 6,
7, and 8 for the labeling of recombinant hCAII, avidin (MW~
16KDa, monomer) and recombinant E.coli dihydrofolate reduc-
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
941
© 2021 Wiley-VCH GmbH

Full Paper
tase (eDHFR, MW~21KDa) proteins, respectively (Figure 4a). To
evaluate the protein labeling of the probes under crude
conditions, hCAII, avidin and eDHFR (5 μM each) were reacted
with equal concentration of 6, 7 and 8, respectively, in E. coli
This imaging result was validated by labeling HeLa cells that
overexpress transmembrane CFP-hCAII protein with probe 6
(Figure S18). The fluorescence from the Cy5 channel overlaid
very well with the CFP-hCAII protein expressed on the cell
surface. To characterize the extracellular CA isozymes of MCF7
and A549, the cells labeled with 3 were lysed and identified by
in-gel fluorescence analysis. The results showed that two
distinct fluorescent bands at~50 kDa and~44 kDa can be
observed clearly from the cell lysates of MCF7 and A549 cells
(Figure 5b and Figure S19). As the bands were not visible in the
presence of EZA, this indicates that the labeling by 3 is specific
for CA isozymes. We believe that the slower reactivity of
difluorophenyl pivalate group and the highly selective inter-
action of sulfonamide ligand with CA proteins are important
factors for the specific labeling in living cells. In line with the in-
gel fluorescence results, significant western-blot bands were
observed for MCF7 with anti-hCAIX antibody and for A549 with
anti-hCAXII (major band) and anti-hCAIX (minor band) anti-
bodies (Figure 5c). These results correlate with many previous
reports which indicate that the major basal extracellular CA
isozymes under normoxia conditions for MCF7 and A549 cells
are hCAIX and hCAXII, respectively.[14b,d] It is also important to
note that dual-color imaging of hCAIX protein can be accom-
plished when 3 and 6 were added sequentially to MCF7 cells
(Figure S20). This indicates that the difluorophenyl pivalate
probe can be a very useful protein labeling toolkit for biological
experiments that require double protein labeling to unveil
accurate and precise insights.
°
lysates (5 mg/mL) at 37 C for 2 hours. As shown in Figure 4b, a
clear single fluorescent band was observed for the target
protein which was not visible in the presence of the
corresponding inhibitor. To further validate the high selectivity
of the difluorophenyl pivalate probes in a medium containing
high protein concentration, we applied the protein labeling
strategy in fetal bovine serum (FBS) which has about 56 mg/mL
of total proteins as determined by BCA assay. The results
showed that as little as 6.25 ng of hCAII and 13 ng of avidin can
be labeled and detected in FBS (Figure S15). A similar detection
limit can be obtained when the labeling reaction was
performed in clean PBS buffer, indicating the consistent and
reliable performance of the probe to label target protein in
different environments (Figure S16). Due to the existence of a
weak binding between sulfonamide and BSA, labeling of BSA
can be observed when the amount of hCAII in FBS was low
(Figure S17). These results show that the difluorophenyl pivalate
moiety is a robust electrophile that can be easily modified by
incorporating different ligands and fluorescent dyes, and can
also be applied to mediums containing high concentration of
proteins.
Labeling, imaging and identification of basal extracellular
carbonic anhydrase isozymes
Carbonic anhydrases (CAs) form a family of enzymes that
catalyze the inter-conversion between carbon dioxide and
carbonic acid.[8] Depending on tissue distribution, cellular local-
ization and conditions, the role of the enzyme changes slightly.
For example, transmembrane-type human carbonic anhydrase
IX (hCAIX,~50 kDa) and XII (hCAXII,~44 kDa) are highly ex-
pressed under hypoxia condition in many tumor cell lines.[14]
Tumor hypoxia is linked with reduced efficacy of cancer
therapies which makes the two CAs valuable biomarkers for
preclinical and diagnostic imaging.[15] Currently, western blot
analysis is the standard method used to detect and identify the
different CA isozymes. We believe that the identity of extrac-
ellular CA isozymes can be rapidly revealed by labeling the
proteins with either 6 or 3, followed by molecular weight
identification using in-gel fluorescent imaging, thereby avoiding
the laborious western-blot analysis.
To show that our difluorophenyl pivalate probes are
sufficiently sensitive and selective to detect endogenous
proteins in living cells, 3 was applied for the imaging and
identification of basal transmembrane CA isozyme of MCF7 and
A549 cells under normoxia conditions. We first applied cell-
impermeable 3 to demonstrate the utility of the probe in
imaging extracellular carbonic anhydrases of MCF7 and A549
cells cultured under normoxia conditions. When the cells were
treated with 3, strong fluorescence along the surface of MCF7
and A549 cells was observed (Figure 5a). In contrast, negligible
fluorescence was observed when EZA was added to the cells.
Imaging, labeling and quantification of hypoxia-induced
carbonic anhydrase isozymes
To further demonstrate the importance of difluorophenyl
pivalate as a reactive module in protein labeling and biological
applications, 3 was used to label overexpressed extracellular
CAs of A549 cells under hypoxia conditions. To induce the
overexpression of CAs, A549 cells were cultured in the presence
of 200 μM CoCl2 for 24 hours to mimic the hypoxia.[16] When 3
was added to the hypoxia-mimetic A549 cells, strong
fluorescence along the cell surface was observed as compared
to the cells cultured under normoxia (Figure 6a). The 3-labeled
A549 cells were subsequently lysed and analyzed by in-gel
fluorescence to identify the type of overexpressed extracellular
CA isozymes (Figure 6b). The results showed that the expression
level of hCAXII under the hypoxia-mimetic condition was
increased by about 33% (Figure 6c). As a comparison, western-
blot analysis also showed that hypoxia-A549 cells expressed
about 30% higher level of hCAXII (Figure S21).
It is noteworthy to mention that the minor hCAIX protein of
A549 cells can also be detected if probe 6, with a brighter Cy5
dye, was used to label the cells (Figure S22). The in-gel
fluorescence analysis of the 6-labeled A549 cells showed that
the expression level for hCAIX and hCAXII under hypoxia-
mimetic conditions were increased by about 68% and 35%,
respectively. These results indicate that the expression of hCAIX
in A549 cells is more sensitive to the hypoxia condition, which
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
942
© 2021 Wiley-VCH GmbH

Full Paper
Figure 5. Live-cell imaging of MCF7 and A549 cells expressing endogenous extracellular hCAs. (i) and (iii) were treated with 3 only, while (ii) and (iv) were with
°
3 and 100 μM EZA. Images were taken after incubating the cells with 5 μM 3 for 2 hours at 37 C in DMEM medium. Scale bar: 40 μm. (b) SDS-PAGE in-gel
fluorescence analyses of the labeling of endogenous hCAs with 5 μM 3 in living MCF7 and A549 cells in the absence or presence of 100 μM EZA. The band
with a single asterisk (*) corresponds to hCAIX and the double asterisk (**) corresponds to hCAXII. (c) Western blotting analyses of the endogenous hCAs in
living MCF7 and A549 cells, detected with anti-hCAIX and anti-hCAXII antibodies.
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
943
© 2021 Wiley-VCH GmbH

Full Paper
Figure 6. Labeling of normoxia and hypoxia-induced carbonic anhydrase isozymes of A549 cells with 3. (a) Live cell images of normoxia and hypoxia-induced
A549 cells labeled with 5 μM 3 for 2 hours. (i) and (ii) were treated with 3 only while (iii) was treated with 3 and 100 μM EZA. All cellular images were taken on
the same day with identical microscope setup. Scale bar: 50 μm. (b) SDS-PAGE in-gel fluorescence analyses of the 3-labeled A549 cells under normoxia and
hypoxia-mimetic conditions and in the absence or presence of EZA inhibitor. M denotes the molecular weight marker with the upper and the lower band
corresponding to 75- and 25-KDa, respectively. (c) Quantitative analysis of hCAXII level of A549 cells cultured under normoxia and hypoxia-mimetic conditions.
The fluorescent intensity was determined from 3-labeled hCAXII in-gel image shown in 6b. Error bars were calculated from three independent measurements.
is in accordance with many previous findings.[14d,17] Thus, cell-
impermeable 3 and 6 can be harnessed as a new selective and
sensitive protein labeling probe to study the expression of
different transmembrane hCAs and the results provide the first
quantitative data of the level of hCAs under basal and
stimulated conditions. Owing to the modular nature of the
probe design, we believe that many other extracellular proteins
can also be studied by our affinity-based difluorophenyl
pivalate probes.
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
944
© 2021 Wiley-VCH GmbH

Full Paper
Intracellular endogenous protein labeling using
cell-permeable difluorophenyl pivalate probe
not only extracellular proteins but also intracellular targets with
strict specificity in live cells.
To demonstrate that our protein labeling probe design can also
be employed to label endogenous intracellular proteins, we
converted the cell-impermeable 3 to the permeable form by
introducing two acetates to the fluorescein dye to generate
3(OAc)2 (Figure 2a). When 0.5 μM 3(OAc)2 was added to living
In vivo labeling and imaging of tumor cells
Currently, there are many strategies available to achieve in vivo
protein labeling. However, almost all of the existing methods
require the incorporation of a genetic tag to facilitate protein
labeling. Although a protein labeling probe based on tosyl
electrophile has been demonstrated previously to achieve
in vivo protein modification, the probe can only label the highly
abundant hCAII protein of red blood cells in living mice.[18] To
date, covalent modification of low abundance endogenous
proteins in vivo remains challenging. To this end, we showed
that tumor cells expressing endogenous hCAXII protein bio-
marker in living mice can be successfully labeled by 6, which
were validated by in vivo, ex vivo, and in-gel fluorescence
imaging.
Probe 6 was intravenously injected via tail vein into the
tumor-bearing mice prepared by subcutaneous inoculation of
A549 cells. The fluorescent images in Figure 8a showed the
biodistribution of 6 in head, neck, and chest regions at 2, 4, and
24 hours after intravenous injection. By the second hour, the
probe remained in the blood circulatory system as observed by
the very strong fluorescent signals present in the head, neck,
and chest regions of mice. The probe can specifically target
either small (top row, 138 mm3) or large (bottom row, 469 mm3)
subcutaneous tumors developed by A549 cells in the xenograft,
°
MCF7 cells and incubated at 37 C for 3 hours, a strong
fluorescence signal was observed in the cytosol as well as on
the cell surface (Figure 7a). In contrast, weak fluorescent
emission was observed in the presence of ethoxzolamide. We
also validated this result by using 3(OAc)2 to label HeLa cells
that over-express cytosolic CFP-hCAII protein. The fluorescence
from the fluorescein channel overlaid very well with the
cytosolic CFP-hCAII protein (Figure S23). To assess the labeling
specificity of the probe, the cells labeled with 3(OAc)2 were
lysed and analyzed by in-gel fluorescence imaging (Figure 7b).
Despite the presence of various other proteins inside the cells,
only two distinctive fluorescence bands were observed at
around 50 and 30 KDa, corresponding to hCAIX and hCAII
proteins, respectively. The two fluorescent bands were sulfona-
mide-responsive, which were not observed in the presence of
EZA inhibitors. It is noteworthy to mention that treatment of
the drug-resistant MCF7 cells with Verapamil, an inhibitor of
drug efflux pump proteins, was important for the probe to
diffuse across the cell membrane. Taken together, these results
show that protein modification by difluorophenyl pivalate
probes occurs via a specific ligand-protein interaction to label
Figure 7. Labeling of endogenous intracellular proteins of MCF7 cells using probe 3(OAc)2. (a) Images of MCF7 cells after treatment with (i) 0.5 μM 3(OAc)2 for
°
3 hours at 37 C in DMEM medium or (ii) in the presence of 100 μM EZA. All cellular images were taken on the same day with identical microscope setup. Scale
bar: 20 μm. (b) SDS-PAGE in-gel fluorescence analyses of the labeling of living MCF7 cells with 0.5 μM 3(OAc)2 for 2 hours in the absence or presence of
100 μM EZA. For the labeling of intracellular hCAII, 50 μM Verapamil and 3(OAc)2 were added together to the cells.
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
945
© 2021 Wiley-VCH GmbH

Full Paper
Figure 8. Labeling and imaging of hCAXII in living mice with probe 6. (a) In vivo imaging and biodistribution of 6 within 24 hours after intravenous injection.
Red arrows indicate the accumulation of 6 in A549 tumor. (b) Ex vivo images of tumor, organs, and tissues at the 24 hour after injection of 6. Abbreviation: Ht
(heart), Kd (kidney), Lv (liver), Lg (lung), Spl (spleen), Ms (muscle), Sk (skin), Tm (tumor). (c) Immunohistochemistry staining of the 6-labeled tumor section by
using anti-hCAXII antibody. Images acquired with different magnifications (i) 20x objective, scale bar: 100 μm, (ii) 40x objective, scale bar: 20 μm. The Cy5 of
probe 6 are shown in red. The Alexa488 of the antibody are shown in green. The nuclear dye DAPI are shown in blue.
with significant accumulation in the tumor at 4 and 24 hours
after injection. By the fourth hour after injection, the apparent
difference between the small and large tumors could be
observed, where the large tumor presented >70% elevation of
fluorescent signal (5.7×109 vs. 3.3×109) than the smaller one.
24 hours after injection, there was an approximately 15%
increase in the fluorescent signal in the large tumor as
compared to that in the smaller tumor, suggesting majority of
probe 6 molecules were excreted through mice.
The mice (N=4) were sacrificed 24 hours after the injection
of 6 to verify the in vivo images. Organs and tissues were
collected for ex vivo imaging and the fluorescent intensities
were quantified to calculate the biodistribution (Figure 8b). The
ex vivo images showed that stronger fluorescent signals were
observed in the tumor rather than other organs, such as heart,
lung, liver, spleen, kidney, etc. Despite only 0.57% injected dose
(ID) presence in the tumor at 24 h (Table S1), the concentration
of 6 in the tumor (0.037%ID/g) is significantly higher than that
in other tissues and organs except for stomach (0.25%ID/g) and
intestine (0.46%ID/g). These results are reasonable because
carbonic anhydrase (CA) isozymes are also highly expressed in
gastrointestinal mucosa.[19] In contrast to other in vivo imaging
probes for CAs, probe 6 also exhibits superior tumor-to-liver (T/
L>3.6) and tumor-to-blood (T/B>12) ratios, indicating that the
probe is a specific tumor-targeting reagent to image the growth
of tumor.[20]
To confirm the in vivo results, an immunohistochemistry
experiment was performed by using anti-hCAXII antibody to
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
946
© 2021 Wiley-VCH GmbH

Full Paper
stain the tumor section. The results showed that fluorescence Acknowledgements
from the antibody co-localize very well with the Cy5 signals of
probe
6
(Figure 8c). To further validate that the high
We are grateful to the Ministry of Science and Technology (Grant
No.: 108-2113-M-007-028-MY3), Ministry of Education (“Aim for
the Top University Plan”; Grant No.: 107QR001I5) and National
Tsing Hua University (Grant No.: 108QI009E1), for financial
support. We also thank Hsin-Ru Wu of the Instrumentation Center
of National Tsing Hua University for HRMS measurements.
fluorescence observed in the tumor is due to the covalent
modification of hCAXII by probe 6, the tumor cells were
homogenized and analyzed together with 6-labeled A549 cells
by SDS-PAGE in-gel fluorescent imaging (Figure S24). Besides
the labeling of high abundance blood proteins: mice serum
albumin (SA) and carbonic anhydrase II (CAII), the gel also
showed a distinctive 44 KDa fluorescent band from the tumor
lysates which was verified by the A549 cell lysates to be the Conflict of Interest
hCAXII protein. These results suggested the ability of 6 to label
low abundant target protein in vivo. Thus, the affinity-based
protein labeling probes with a sterically stabilized difluorophen-
yl pivalate electrophile present a novel solution to resist auto
and catalytic hydrolysis yet exhibiting sufficient labeling rate to
enable protein modification in vivo.
The authors declare no conflict of interest.
Keywords: Protein labeling · Membrane proteins · Hypoxia ·
Fluorescent Imaging · Difluorophenyl ester
Conclusions
[1] a) S. Ziegler, V. Pries, C. Hedberg, H. Waldmann, Angew. Chem. Int. Ed.
Engl. 2013, 52, 2744–2792; b) G. Crivat, J. W. Taraska, Trends Biotechnol.
2012, 30, 8–16; c) K. Tsuchikama, Z. An, Protein Cell 2018, 9, 33–46.
[2] a) J. Lotze, U. Reinhardt, O. Seitz, A. G. Beck-Sickinger, Mol. BioSyst. 2016,
12, 1731–1745; b) H.-J. Chen, C. Y. Chew, E.-H. Chang, Y.-W. Tu, L.-Y. Wei,
B.-H. Wu, C.-H. Chen, Y.-T. Yang, S.-C. Huang, J.-K. Chen, I. C. Chen, K.-T.
Tan, J. Am. Chem. Soc. 2018, 140, 5224–5234; c) C. Jing, V. W. Cornish,
Acc. Chem. Res. 2011, 44, 784–792; d) C. D. Spicer, B. G. Davis, Nat.
Commun. 2014, 5, 4740.
[3] a) T. Tamura, I. Hamachi, J. Am. Chem. Soc. 2019, 141, 2782–2799;
b) D. P. Murale, S. C. Hong, M. M. Haque, J.-S. Lee, Proteome Sci. 2017,
15, 14; c) C. G. Parker, A. Galmozzi, Y. Wang, B. E. Correia, K. Sasaki, C. M.
Joslyn, A. S. Kim, C. L. Cavallaro, R. M. Lawrence, S. R. Johnson, I.
Narvaiza, E. Saez, B. F. Cravatt, Cell 2017, 168, 527–541.e529; d) Y.-C.
Chen, K. M. Backus, M. Merkulova, C. Yang, D. Brown, B. F. Cravatt, C.
Zhang, J. Am. Chem. Soc. 2017, 139, 639–642.
[4] a) T. Tamura, Y. Kioi, T. Miki, S. Tsukiji, I. Hamachi, J. Am. Chem. Soc.
2013, 135, 6782–6785; b) T. Miki, S.-H. Fujishima, K. Komatsu, K. Kuwata,
S. Kiyonaka, I. Hamachi, Chem. Biol. 2014, 21, 1013–1022; c) Y. Takaoka,
Y. Nishikawa, Y. Hashimoto, K. Sasaki, I. Hamachi, Chem. Sci. 2015, 6,
3217–3224; d) K. Matsuo, Y. Nishikawa, M. Masuda, I. Hamachi, Angew.
Chem. Int. Ed. 2018, 57, 659–662; Angew. Chem. 2018, 130, 667–670;
e) T. Tamura, T. Ueda, T. Goto, T. Tsukidate, Y. Shapira, Y. Nishikawa, A.
Fujisawa, I. Hamachi, Nat. Commun. 2018, 9, 1870.
[5] a) A. F. Kornahrens, A. B. Cognetta, D. M. Brody, M. L. Matthews, B. F.
Cravatt, D. L. Boger, J. Am. Chem. Soc. 2017, 139, 7052–7061; b) S. M.
Hacker, K. M. Backus, M. R. Lazear, S. Forli, B. E. Correia, B. F. Cravatt, Nat.
Chem. 2017, 9, 1181; c) S. Tsukiji, H. Wang, M. Miyagawa, T. Tamura, Y.
Takaoka, I. Hamachi, J. Am. Chem. Soc. 2009, 131, 9046–9054; d) M. H.
Wright, S. A. Sieber, Nat. Prod. Rep. 2016, 33, 681–708; e) B. D. Horning,
R. M. Suciu, D. A. Ghadiri, O. A. Ulanovskaya, M. L. Matthews, K. M. Lum,
K. M. Backus, S. J. Brown, H. Rosen, B. F. Cravatt, J. Am. Chem. Soc. 2016,
138, 13335–13343; f) J. S. Marvin, H. W. Hellinga, J. Am. Chem. Soc. 1998,
120, 7–11.
In summary, we have developed a novel affinity-based traceless
protein labeling probe approach with high selectivity, stability
and sensitivity to label target proteins in complex environments
of living cells and in vivo. The probes exhibit long term stability
in aqueous buffer and stock solutions, as well as resistance to
catalytic hydrolysis by esterase. As compared to all the other
protein labeling probes, our probe design features a unique
difluorophenyl pivalate as the activated module to accomplish
high selectivity, sensitivity and stability. The moderate pKa of
difluorophenol ensures the long term stability of the probe in
stock solutions and aqueous buffer, while the large steric
hindrance provided by the pivalate group imparts resistance to
both autolysis and catalytic hydrolysis. The introduction of a
large steric effect is crucial for the probe to reduce nonspecific
protein labeling which acts by decreasing the reactivity of the
activated phenyl ester. Therefore, chemical reaction induced by
the proximity effect can only occur via the specific protein-
ligand interaction. Owing to the modular design and the
versatile synthetic scheme, different fluorophores and ligands
can be attached easily on the probes to label various natural (or
engineered) proteins having nucleophilic amino acids situated
around the ligand binding site, as shown in the cases of avidin,
eDHFR and hCAs. With the cell-impermeable sulfonamide
probes, we have revealed the major CA proteins expressed on
the cell surface of MCF7 and A549 cells to be hCAIX and hCAXII
respectively. Furthermore, the in-gel analysis of the 6-labeled
tumor lysate also implies that covalent modification of tumor
protein in vivo can be achieved with our affinity-based protein
labeling probe. We believe that this difluorophenyl pivalate
probe approach can spearhead the development of novel
protein labeling reagents to become an important tool in drug-
discovery, medical diagnosis and basic biology research.
[6] a) C. C. Hughes, Y.-L. Yang, W.-T. Liu, P. C. Dorrestein, J. J. L. Clair, W.
Fenical, J. Am. Chem. Soc. 2009, 131, 12094–12096; b) H.-J. Chan, X.-H.
Lin, S.-Y. Fan, J. Ru Hwu, K.-T. Tan, Chem. Asian J. 2020, 15, 3416–3420.
[7] J. Han, F.-M. Tao, J. Phys. Chem. A 2006, 110, 257–263.
[8] V. M. Krishnamurthy, G. K. Kaufman, A. R. Urbach, I. Gitlin, K. L. Gudiksen,
D. B. Weibel, G. M. Whitesides, Chem. Rev. 2008, 108, 946–1051.
[9] a) M. E. Olson, D. Abate-Pella, A. L. Perkins, M. Li, M. A. Carpenter, A.
Rathore, R. S. Harris, D. A. Harki, J. Med. Chem. 2015, 58, 7419–7430;
b) Z. Blaxill, S. Holland-Crimmin, R. Lifely, J. Biomol. Screening 2009, 14,
547–556.
[10] a) O. Koniev, A. Wagner, Chem. Soc. Rev. 2015, 44, 5495–5551; b) G. T.
Hermanson, Bioconjugate Techniques, Elsevier Science, 2013.
[11] L. D. Lavis, T.-Y. Chao, R. T. Raines, Chem. Sci. 2011, 2, 521–530.
[12] Craig S. McKay, M. G. Finn, Chem. Biol. 2014, 21, 1075–1101.
[13] a) X. Tan, Z. Song, RSC Adv. 2014, 4, 3263–3271; b) R. Zini, P. d’Athis, A.
Hoareau, J. P. Tillement, Eur. J. Clin. Pharmacol. 1976, 10, 139–145.
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
947
© 2021 Wiley-VCH GmbH

Full Paper
[14] a) J. Chiche, K. Ilc, J. Laferrière, E. Trottier, F. Dayan, N. M. Mazure, M. C.
Brahimi-Horn, J. Pouysségur, Cancer Res. 2009, 69, 358–368; b) J. Jiang,
J. H. Zhao, X. L. Wang, X. J. Guo, J. Yang, X. Bai, S. Y. Jin, R. L. Ge,
Neoplasma 2015, 62, 456–463; c) Z. Chen, L. Ai, M. Y. Mboge, C. Tu, R.
McKenna, K. D. Brown, C. D. Heldermon, S. C. Frost, PLoS One 2018, 13,
e0199476; d) C. C. Wykoff, N. J. P. Beasley, P. H. Watson, K. J. Turner, J.
Pastorek, A. Sibtain, G. D. Wilson, H. Turley, K. L. Talks, P. H. Maxwell,
C. W. Pugh, P. J. Ratcliffe, A. L. Harris, Cancer Res. 2000, 60, 7075–7083.
[15] a) C. T. Supuran, Nat. Rev. Drug Discovery 2008, 7, 168; b) A. Nógrádi,
Am. J. Pathol. 1998, 153, 1–4; c) P. H. Watson, S. K. Chia, C. C. Wykoff, C.
Han, R. D. Leek, W. S. Sly, K. C. Gatter, P. Ratcliffe, A. L. Harris, Br. J.
Cancer 2003, 88, 1065–1070.
[19] a) T. Kleinke, S. Wagner, H. John, D. Hewett-Emmett, S. Parkkila, W.-G.
Forssmann, G. Gros, J. Exp. Biol. 2005, 208, 487–496; b) V. Endeward, T.
Kleinke, G. Gros, Comp. Biochem. Physiol., Part A Mol. Integr. Physiol.
2003, 136, 281–287; c) R. E. Fleming, S. Parkkila, A. K. Parkkila, H.
Rajaniemi, A. Waheed, W. S. Sly, J. Clin. Invest. 1995, 96, 2907–2913;
d) S.-i. Igarashi, Y. Kano, T. Nishita, H. Amasaki, M. Asari, J. Vet. Med. Sci.
1992, 54, 535–539.
[20] a) B. Bao, K. Groves, J. Zhang, E. Handy, P. Kennedy, G. Cuneo, C. T.
Supuran, W. Yared, M. Rajopadhye, J. D. Peterson, PLoS One 2012, 7,
e50860; b) J. Lau, Z. Liu, K. S. Lin, J. Pan, Z. Zhang, D. Vullo, C. T.
Supuran, D. M. Perrin, F. Benard, J. Nucl. Med. 2015, 56, 1434–1440.
[16] a) Y. Yuan, G. Hilliard, T. Ferguson, D. E. Millhorn, J. Biol. Chem. 2003,
278, 15911–15916; b) D. Wu, P. Yotnda, J. Visualization 2011, 2899.
[17] M. Ilie, V. Hofman, J. Zangari, J. Chiche, J. Mouroux, N. M. Mazure, J.
Pouysségur, P. Brest, P. Hofman, Lung Cancer 2013, 82, 16–23.
[18] S. Tsukiji, M. Miyagawa, Y. Takaoka, T. Tamura, I. Hamachi, Nat. Chem.
Biol. 2009, 5, 341.
Manuscript received: January 19, 2021
Revised manuscript received: February 23, 2021
Accepted manuscript online: February 25, 2021
Version of record online: March 9, 2021
Chem Asian J. 2021, 16, 937–948
www.chemasianj.org
948
© 2021 Wiley-VCH GmbH