Metabolic synthesis of clickable glutathione for chemoselective detection of glutathionylation

Article abstract of DOI:10.1021/ja503946q

Glutathionylation involves reversible protein cysteine modification that regulates the function of numerous proteins in response to redox stimuli, thereby altering cellular processes. Herein we developed a selective and versatile approach to identifying glutathionylation by using a mutant of glutathione synthetase (GS). GS wild-type catalyzes coupling of γGlu-Cys to Gly to form glutathione. We generated a GS mutant that catalyzes azido-Ala in place of Gly with high catalytic efficiency and selectivity. Transfection of this GS mutant (F152A/S151G) and incubation of azido-Ala in cells efficiently afford the azide-containing glutathione derivative, γGlu-Cys-azido-Ala. Upon H₂O₂ treatment, clickable glutathione allowed for selective and sensitive detection of glutathionylated proteins by Western blotting or fluorescence after click reaction with biotin-alkyne or rhodamine-alkyne. This approach affords the efficient metabolic tagging of intracellular glutathione with small clickable functionality, providing a versatile handle for characterizing glutathionylation.

Full text of DOI:10.1021/ja503946q

Communication
pubs.acs.org/JACS
Metabolic Synthesis of Clickable Glutathione for Chemoselective
Detection of Glutathionylation
Kusal T. G. Samarasinghe, Dhanushka N. P. Munkanatta Godage, Garrett C. VanHecke,
and Young-Hoon Ahn*
Department of Chemistry, Wayne State University, Detroit, Michigan 48202, United States
S
* Supporting Information
biotin is likely to interfere with enzyme-mediated (de)-
ABSTRACT: Glutathionylation involves reversible pro-
tein cysteine modification that regulates the function of
numerous proteins in response to redox stimuli, thereby
altering cellular processes. Herein we developed a selective
and versatile approach to identifying glutathionylation by
using a mutant of glutathione synthetase (GS). GS wild-
type catalyzes coupling of γGlu-Cys to Gly to form
glutathione. We generated a GS mutant that catalyzes
azido-Ala in place of Gly with high catalytic efficiency and
selectivity. Transfection of this GS mutant (F152A/
S151G) and incubation of azido-Ala in cells efficiently
afford the azide-containing glutathione derivative, γGlu-
Cys-azido-Ala. Upon H2O2 treatment, clickable gluta-
thione allowed for selective and sensitive detection of
glutathionylated proteins by Western blotting or fluo-
rescence after click reaction with biotin-alkyne or rhod-
amine-alkyne. This approach affords the efficient metabolic
tagging of intracellular glutathione with small clickable
functionality, providing a versatile handle for characterizing
glutathionylation.
glutathionylation. Moreover, direct incubation of biotinylated
glutathione with cells can increase cellular thiol content, thus
altering the redox state. In proteomics, the biotin switch method
has been used with glutaredoxin in the reducing step8 and the
similar resin-capture approach identified numerous glutathiony-
lated proteins.9 While promising, the complexity may arise from
this indirect identification method. Recently, tagging glutathione
with biotin-spermine by E. coli glutathionylspermidine synthe-
tase was developed, identifying many target proteins.10 However,
this analogue of glutathione is also bulky because of the presence
of biotin-spermine.
Here we report the selective and versatile approach to
characterizing glutathionylation: i.e., metabolic tagging of
glutathione with an azide or alkyne functional group by
engineering glutathione synthetase (GS) (Figure 1). Gluta-
edox regulation is a fundamental mechanism that controls
Rnumerous cellular processes, including proliferation,
inflammation, and apoptosis.1 Reactive oxygen species (ROS),
which were once thought to be toxic byproducts of respiration,
are now believed to be physiological signaling molecules that are
cautiously generated for regulating multiple signaling pathways.2
Importantly, one major mechanism of action of ROS is the
oxidation of protein cysteine residue into various oxoforms, such
as disulfide, sulfenic, sulfinic, and sulfonic acid, by which protein
function can be turned on or off.3 The reactivity and stability of
individual oxoforms appear different and their functional roles
may vary accordingly. Among various oxoforms, glutathionyla-
tion is disulfide formation of a protein cysteine residue with
glutathione. Glutathionylation serves as an important regulatory
switch that mediates the effect of ROS in normal redox signaling
and pathological diseases.4,5
There is a growing list of redox-sensitive proteins regulated by
glutathionylation of which identifications were made possible by
various biochemical approaches. Current methods include an
antibody pull-down approach in which a glutathione antibody is
used to detect or enrich target proteins of glutathionylation.
However, existing antibodies appear to suffer from low specificity
and sensitivity.4,6 Alternatively, biotinylated glutathione is used
to pull-down glutathionylated proteins,7 but the bulky nature of
Figure 1. Identifying glutathionylation using “clickable” glutathione
metabolically synthesized by glutathione synthetase mutant that
catalyzes azido-Ala in place of Gly.
thione, an abundant tripeptide made of γGlu-Cys-Gly, is
biosynthesized by two concerted ATP-dependent enzymes, γ-
glutamylcysteine ligase (GCL) and GS.11 GCL, the rate-limiting
enzyme of glutathione biosynthesis, ligates Glu and Cys to form
γGlu-Cys. Subsequently, GS catalyzes coupling of γGlu-Cys to
Gly to form glutathione. We envisioned that the active site of GS
can be engineered to incorporate an azide- or alkyne-containing
Gly derivative in place of Gly. In this way, mutant GS could afford
clickable glutathione in cells that can undergo glutathionylation
in response to ROS stimulus. The subsequent bio-orthogonal
click reaction with biotin or fluorophore would provide an
effective handle for detecting and enriching target proteins of
glutathionylation. Notably, the limiting amino acid for
glutathione synthesis is Cys;12 thus our approach is less likely
Received: April 20, 2014
Published: July 31, 2014
© 2014 American Chemical Society
11566
dx.doi.org/10.1021/ja503946q | J. Am. Chem. Soc. 2014, 136, 11566−11569

Journal of the American Chemical Society
Communication
to alter the total concentration of thiol or glutathione in cells.
Interestingly, the major degrading enzyme of glutathione, such as
γ-glutamyl transpeptidase, appears to retain the specificity to
glutathione by mainly interacting with the γGlu residue of
glutathione rather than the Gly residue.13 Furthermore, there are
a few structural variations of glutathione reported in plants, all of
which possess the modification of Gly replaced by β-Ala, Ser, or
Glu.14 This may imply that the modification of Gly in glutathione
can be tolerated for its metabolism in cells.
Table 1. Kinetic Data of Amino Acid Substrates for GS WT
and Mutants
a
The crystal structure of human GS (hGS) (PDB: 2HGS)
bound to glutathione and ADP predicts potential amino acids
(F152 and S151) that are in close proximity to the Gly residue of
glutathione (Figure 2). It appears that both F152 and S151 block
Figure 2. Active site of glutathione synthetase bound with glutathione
and ADP. Note F152 and S151 (cyan) in close proximity to Gly in
glutathione.
a
ND: not determined due to low activity. PG: propargyl-Gly. HPG:
homo-propargyl-Gly.
any side chain introduced in the position of Gly in glutathione.
This suggests that mutation of F152 in combination with S151 is
likely to provide the possibility of accommodating the azide- or
alkyne-containing Gly derivative (Figure S1).
fold higher selectivity over Met. In addition to azido-Ala,
propargyl Gly (PG) and homopropargyl Gly (HPG) were
catalyzed by GS M4 with a similar or lower kcat/Km to that for
azido-Ala. Collectively, these kinetic data indicate that the GS M4
double mutant can selectively catalyze azido-Ala to synthesize
“clickable” glutathione with similar catalytic efficiency to GS WT
synthesizing “endogenous” glutathione.
To find a GS mutant that efficiently catalyzes a clickable Gly
derivative, His-tag hGS wild-type (WT) and mutants were
expressed and purified in high purity (>95%) and a good yield
(∼6 mg/1 L culture) (Figure S2). Subsequently, the kinetic assay
for monitoring ATP consumption was measured, as reported
previously.15 The results are summarized in Table 1 and Figure
S3. Interestingly, GS F152G single mutant (M1) catalyzes azido-
Ala most efficiently with the highest kcat/Km among all the amino
acids tested. The small side chain of Gly and big side chains of
Met and Val make them poor substrates with high Km and/or low
kcat. This indicates that the best substrate for F152G mutant
(M1) is azido-Ala. However, the kcat/Km of azido-Ala for GS
F152G (M1) is still 25 times lower than the kcat/Km of Gly for GS
WT. Encouragingly, this is significantly improved in the GS
F152G/S151G double mutant (M2). Azido-Ala in this double
mutant (M2) shows only 5-fold lower kcat/Km versus Gly for GS
WT. Importantly, Gly and Ala are poorly catalyzed by this double
mutant (M2), but Met is catalyzed with similar kcat/Km to that for
azido-Ala. This indicates that GS F152G/S151G (M2) improved
the Km for azido-Ala, but it lost the selectivity of azido-Ala over
Met. In molecular size, azido-Ala appears slightly smaller than
Met.16 Thus, we reasoned that mutation of F152G/S151G (M2)
may have made a relatively large area that can accommodate not
only azido-Ala but also Met, thereby giving the poor selectivity.
In support of this, the GS S151G mutant (M3) shows high
selectivity for azido-Ala over Met, albeit with a high Km. We
further reasoned that mutation of F152A together with S151G
(M4) may have enough space for azido-Ala but not for Met.
Indeed, F152A/S151G (M4) catalyzes azido-Ala the most
efficiently with the highest kcat/Km (630.6 min−1 mM−1), which
is only 1.7-fold lower than the kcat/Km of Gly for GS WT (1095.0
min−1 mM−1). In addition, GS M4 catalyzes azido-Ala with 2.6-
Next, we sought to monitor the biosynthesis of azido-
glutathione, a glutathione derivative containing azido-Ala in
place of Gly, in cells. Interestingly, transient transfection of
FLAG-tagged GS M4 to HEK293 cells induced overexpression
of the mutant about 5−10 times higher than endogenous GS WT
(Figure S4). A high ratio of GS mutant (M4) over GS WT may
facilitate the flux of γGlu-Cys dipeptide toward the generation of
azido-glutathione. After incubation of azido-Ala (1 mM) for 20 h,
cells were lysed and the lysates free from proteins were injected
into LC-MS. Encouragingly, cells expressing GS M4 mutant
incubated with azido-Ala showed a strong mass ion peak of m/z
363, which corresponds to the mass of azido-glutathione (Figure
3b). At the same time, a large mass ion peak of m/z 308,
corresponding to endogenous glutathione, was also detected. In
contrast, control cells without GS M4 and azido-Ala showed only
a mass ion peak of m/z 308 but no ion peak of 363 (Figure 3a).
To confirm the identity of the peaks, we added fluorescein-
iodoacetamide (FL-IA) during lysis. This will conjugate all
glutathione with FL-IA by thiol-alkylation. LC-MS analysis
detects a mass shift to m/z 695 and 750, which are in agreement
with masses of endogenous glutathione and azido-glutathione
conjugated with FL-IA, respectively (Figure S5). The incubation
of azido-Ala alone without GS M4 in cells showed a mass of m/z
695 only, corresponding to endogenous glutathione, demon-
strating that GS M4 is absolutely necessary for azido-glutathione
biosynthesis (Figure S5). These data indicate the effective
biosynthesis of clickable glutathione by GS M4 in cells.
11567
dx.doi.org/10.1021/ja503946q | J. Am. Chem. Soc. 2014, 136, 11566−11569

Journal of the American Chemical Society
Communication
and a reducing agent, since a reducing agent will reduce the
disulfide in glutathionylation. Azido-Ala or H2O2 alone without
GS M4 did not produce a signal with streptavidin in Western blot
(lanes 1−4, Figure 4a), indicating that azido-Ala is not
Figure 3. Analysis of azido-glutathione by LC-MS. (a) Without and (b)
with transfection of GS M4 and incubation of azido-Ala (1 mM) in
HEK293 cells for 20 h, or (c) after transfection of GS M4 and increasing
the incubation time of azido-Ala (1 mM), lysates are collected and
injected into LC-MS directly or after modification with fluorescein-
iodoacetamide (FL-IA). Masses corresponding to endogenous (black)
and azido-glutathione (pink) with or without modification by FL-IA are
extracted and overlaid.
Figure 4. Azido-glutathione generated by transfection of GS M4 and
incubation of azido-Ala can sensitively identify glutathionylation. (a)
After transfection of GS M4, cells were incubated with azido-Ala (1
mM) for 20 h and then treated with H2O2 for 15 min. Lysates were
subjected to click reaction with biotin-alkyne, then resolved by SDS-
PAGE, and probed as shown above. Samples in lanes 9−12 were treated
with DTT before loading on gel. (b) Same as (a) except with treatment
of increasing concentrations of H2O2.
Our kinetic data indicate that GS M4 has a higher selectivity
for azido-Ala, but it can catalyze Gly, Ala, or Met to synthesize the
corresponding glutathione derivatives. As predicted, when
purified GS M4 and γGlu-Cys were mixed with equal amounts
of azido-Ala, Met, and Gly, in the presence of ATP, LC-MS
analysis detected peaks of glutathione derivatives containing
azido-Ala, Met, and Gly in a ratio of ∼4:2:1, respectively (Figure
S6). However, surprisingly, LC-MS analysis of cell lysates in
Figure 3b did not detect any mass of modified glutathione
containing Met or Ala (Figure S5). We suspect that incubation of
a higher concentration of azido-Ala (1 mM) over Met (0.2 mM)
or other amino acids (0.1−0.2 mM) in a typical medium may
have increased the selectivity of azido-Ala over other amino acids
in cells. In addition, despite the concern that the azide
functionality can be reduced to an amine group in cells,16 the
mass corresponding to the reduced form of azido-glutathione
was not detected in LC-MS analysis (Figure S5).
Interestingly, quantification of the mass ion peaks in Figure 3b
showed an ∼2:1 ratio of endogenous glutathione to azido-
glutathione. But, the amount of azido-glutathione was dependent
on incubation time. After transfection of GS M4 and incubation
of azido-Ala for 6, 20, and 40 h, LC-MS detected a gradual
increase of azido-glutathione to approximately 10%, 30%, and
60% of endogenous glutathione, respectively (Figure 3c). This
gradual increase of azido-glutathione may reflect the slow rate of
glutathione biosynthesis in nonstressed cells.17 Interestingly, the
measurement of the total thiol concentration in lysates by a thiol-
reactive fluorogenic bromobimane showed similar thiol concen-
trations in samples with and without GS M4 and azido-Ala
(Figure S4b). This is consistent with the fact that GCL is the rate-
limiting enzyme in glutathione production and GS over-
expression does not increase the total amount of glutathione.18
Our clickable glutathione provides a versatile handle that can
be utilized for identification of glutathionylated proteins. After
transfection of GS M4 and incubation of azido-Ala, cells were
treated with hydrogen peroxide (H2O2) (1 mM) for 15 min. The
resulting cell lysates were subjected to a click reaction with
biotin-alkyne and probed with streptavidin-peroxidase. Note that
the click reaction proceeded with Cu(I)Br instead of Cu(II)SO4
incorporated into cellular proteins (lanes 1 vs 3). Neither GS
M4 alone nor GS M4 with H2O2 gave any signal (lanes 5−6). In
contrast, H2O2 treatment after introducing GS M4 and azido-Ala
generated numerous strong bands (lane 7 vs 8), which
disappeared when treating samples with DTT (lanes 9−12).
The click reaction with rhodamine-alkyne provided similar
results while facilitating in-gel fluorescence detection (Figure
S7a). Encouragingly, when an increasing dose of H2O2 (0.1−1.0
mM) was treated for 15 min, numerous bands were detected
sensitively even at a low concentration of H2O2 (0.1 mM)
(Figure 4b). Concentrations of azido-Ala higher than 1 mM did
not enhance the signal of glutathionylation significantly (Figure
S7b). These results demonstrate that selective and sensitive
detection can be made by clickable glutathione generated by the
GS M4 mutant without significant alteration of the redox state.
Azido-glutathione coupled with biotin-alkyne or fluorophore-
alkyne can be used for chemoselective purification or cellular
imaging of glutathionylated proteins, respectively. Indeed, the
use of streptavidin-agarose enabled purification and enrichment
of glutathionylated proteins, which were readily detected by
silver stain (Figure S8a). In addition, the combination of
immunoprecipitation, click reaction, and Western blotting
allowed for detecting glutathionylation of individual proteins,
such as protein tyrosine phosphatase 1B19 and heat shock protein
909,10 (Figure S8b).
We also attempted to directly visualize glutathionylated
proteins in cells by fluorescence. HEK293 cells overexpressing
GS M4 were incubated with azido-Ala and treated with H2O2 (1
mM) for 15 min. Cells were then fixed with methanol and
washed with PBS to remove excess glutathione in cells.20 The
subsequent click reaction with Alexa-Fluor 647 (AF 647)-alkyne
detected a strong fluorescence signal (Figure 5e), which was
widespread in cells when costained with DAPI (Figure 5d, f). In
contrast, control cells without GS M4 did not produce any
11568
dx.doi.org/10.1021/ja503946q | J. Am. Chem. Soc. 2014, 136, 11566−11569

Journal of the American Chemical Society
Communication
ASSOCIATED CONTENT
* Supporting Information
Enzyme kinetics, LC-MS analysis, fluorescence detection of
glutathionylation, and experimental procedures. This material is
available free of charge via the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*E-mail: yahn@chem.wayne.edu.
■
Notes
Figure 5. Azido-glutathione can selectively detect glutathionylation by
fluorescence imaging. Cells without (a−c) or with (d−f) transfection of
GS M4 were incubated with azido-Ala (1 mM) for 20 h and then treated
with H2O2 (1 mM) for 15 min. After fixation, cells were subjected to
click reaction with AF647-alkyne and monitored for DAPI stains (a, d)
or glutathionylation (b, e). Images of DAPI stain and glutathionylation
were overlaid in (c, f).
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by Wayne State University Start-up
funds and a WSU University Research Grant. We thank Dr.
Miriam Greenberg laboratory for use of the fluorescence
microscope.
REFERENCES
■
fluorescence signal after the click reaction with AF647-alkyne
(Figure 5a−c). These results demonstrate that clickable
glutathione can selectively visualize the localization of gluta-
thionylated proteins in cells.
(1) Jacob, C., Winyard, P. G., Eds. Redox Signaling and Regulation in
Biology and Medicine; Wiley-VCH: Weinheim, 2009.
(2) (a) Sundaresan, M.; Yu, Z. X.; Ferrans, V. J.; Irani, K.; Finkel, T.
Science 1995, 270, 296. (b) Owusu-Ansah, E.; Banerjee, U. Nature 2009,
461, 537.
(3) Paulsen, C. E.; Carroll, K. S. ACS Chem. Biol. 2010, 5, 47.
(4) Grek, C. L.; Zhang, J.; Manevich, Y.; Townsend, D. M.; Tew, K. D.
J. Biol. Chem. 2013, 288, 26497.
(5) Gallogly, M. M.; Mieyal, J. J. Curr. Opin. Pharmacol. 2007, 7, 381.
(6) Fratelli, M.; Gianazza, E.; Ghezzi, P. Expert Rev. Proteomics 2004, 1,
365.
(7) Sullivan, D. M.; Wehr, N. B.; Fergusson, M. M.; Levine, R. L.;
Finkel, T. Biochemistry 2000, 39, 11121.
(8) Lind, C.; Gerdes, R.; Hamnell, Y.; Schuppe-Koistinen, I.; von
Lowenhielm, H. B.; Holmgren, A.; Cotgreave, I. A. Arch. Biochem.
Biophys. 2002, 406, 229.
(9) Su, D.; Gaffrey, M. J.; Guo, J.; Hatchell, K. E.; Chu, R. K.; Clauss, T.
R.; Aldrich, J. T.; Wu, S.; Purvine, S.; Camp, D. G.; Smith, R. D.; Thrall,
B. D.; Qian, W. J. Free Radical Biol. Med. 2014, 67, 460.
(10) Chiang, B. Y.; Chou, C. C.; Hsieh, F. T.; Gao, S.; Lin, J. C.; Lin, S.
H.; Chen, T. C.; Khoo, K. H.; Lin, C. H. Angew. Chem., Int. Ed. 2012, 51,
5871.
(11) Dickinson, D. A.; Forman, H. J. Biochem. Pharmacol. 2002, 64,
1019.
(12) Wu, G.; Fang, Y. Z.; Yang, S.; Lupton, J. R.; Turner, N. D. J. Nutr.
2004, 134, 489.
(13) Keillor, J. W.; Castonguay, R.; Lherbet, C. Methods Enzymol.
2005, 401, 449.
(14) Skipsey, M.; Davis, B. G.; Edwards, R. Biochem. J. 2005, 391, 567.
(15) Vergauwen, B.; De Vos, D.; Van Beeumen, J. J. J. Biol. Chem. 2006,
281, 4380.
(16) Kiick, K. L.; Saxon, E.; Tirrell, D. A.; Bertozzi, C. R. Proc. Natl.
Acad. Sci. U.S.A. 2002, 99, 19.
(17) Lu, S. C. FASEB J. 1999, 13, 1169.
(18) Grant, C. M.; Maclver, F. H.; Dawes, I. W. Mol. Biol. Cell 1997, 8,
1699.
Although small, the azide functionality in azido-glutathione
may interfere with redox enzyme-mediated (de)-
glutathionylation. Among several redox enzymes that are
implicated in glutathionylation, glutaredoxin 1 (Grx1) is a well-
known enzyme that is shown to catalyze the removal of
glutathionylation.5 Lysates containing glutathionylated proteins
by azido-glutathione were incubated with purified Grx1 in vitro
(Figure S9a). Lysates without Grx1 treatment showed numerous
bands. However, most signals completely disappeared when
Grx1 was added for 30 min, indicating Grx1 can efficiently
catalyze the removal of azido-glutathione. Similarly, when H2O2
was washed out after treatment in cells, a rapid deglutathiony-
lation was observed in a time-dependent manner (Figure S9b),
suggesting that glutathionylation by azido-glutathione can be
efficiently reduced in cells.
Our kinetic assay showed the similar kcat/Km of PG or HPG
versus azido-Ala for GS M4. Thus, we investigated the use of PG
and HPG for detecting glutathionylation with transfection of GS
M4, despite the concern that HPG is a good Met surrogate for de
novo protein synthesis16 and PG is an inhibitor of cystathionine γ-
lyase. When PG or HPG was incubated in cells overexpressing
GS M4, LC-MS mass analysis showed that GS M4 can effectively
synthesize the clickable glutathione containing PG or HPG
(Figure S10). However, the subsequent gel analysis showed that
HPG was used for de novo protein synthesis, which prevented the
selective detection of glutathionylation, and PG did not provide
the sensitive detection of glutathionylation when compared to
azido-Ala (Figure S11).
In summary, we developed a GS mutant for selective tagging of
glutathione with clickable functionality. Azido-glutathione
generated in situ was effective for sensitive detection of
glutathionylation. A major advantage of clickable glutathione
over existing approaches is the small versatile azide handle, which
can couple to various applications. It not only can conjugate with
biotin for purification in proteomics but also can click with
fluorophores for imaging applications. Also, azido-glutathione
may be used to identify small molecule substrates of glutathione
S-transferase, as shown in a recent report.21 Such efforts could
significantly contribute to an enhanced understanding of the
roles of glutathione and glutathionylation in redox regulation.
̈
(19) Barrett, W. C.; DeGnore, J. P.; Konig, S.; Fales, H. M.; Keng, Y. F.;
Zhang, Z. Y.; Yim, M. B.; Chock, P. B. Biochemistry 1999, 38, 6699.
̈
(20) Soderdahl, T.; Enoksson, M.; Lundberg, M.; Holmgren, A.;
̈
Ottersen, O. P.; Orrenius, S.; Bolcsfoldi, G.; Cotgreave, I. A. FASEB J.
2003, 17, 124.
(21) Feng, S.; Zhang, L.; Adilijiang, G.; Liu, J.; Luo, M.; Deng, H.
Angew. Chem., Int. Ed. 2014, 53, 7149.
11569
dx.doi.org/10.1021/ja503946q | J. Am. Chem. Soc. 2014, 136, 11566−11569