Quantification of protein sulfenic acid modifications using isotope-coded dimedone and iododimedone

Article abstract of DOI:10.1002/anie.201007175

Quantitative proteomics: The new technique mentioned in the title-in short, ICDID-enables quantification of sulfenic acid modifications in proteins (see picture). The approach permits S-hydroxylation site occupancy to be monitored at individual cysteines

Full text of DOI:10.1002/anie.201007175

Communications
DOI: 10.1002/anie.201007175
Redox Proteomics
Quantification of Protein Sulfenic Acid Modifications Using Isotope-
Coded Dimedone and Iododimedone**
Young Ho Seo and Kate S. Carroll*
Dedicated to Professor Carolyn R. Bertozzi on the occasion of her 45th birthday
Since its discovery almost 40 years ago,[1] S-hydroxylation
À
( SOH) of cysteine thiol side chains at active and allosteric
sites within proteins has emerged as a central post-transla-
tional modification.[2] At present, more than 200 transcription
factors, signaling proteins, metabolic enzymes, proteostasis
regulators, and cytoskeletal components that undergo sulfenic
acid modification have been identified.[3] Like phosphoryla-
tion, S-hydroxylation can be a dynamic and reversible post-
translational modification whereby hydrogen peroxide
(H2O2) and other reactive oxygen species (ROS) generated
during the “oxidative burst” that accompanies many receptor-
mediated signaling processes react with a thiolate anion to
form sulfenic acid, and a family of enzymes reduces these
modifications (or subsequent disulfides) in proteins
(Scheme 1).[4] Although protein sulfenic acids are often
Scheme 1. Oxidation states of protein cysteines that are implicated in
biological function. Both reactive oxygen and nitrogen species (ROS/
RNS) can oxidize the thiol side chain of cysteine to sulfenic acid,
which may be stabilized by the protein microenvironment or go on to
form other reversible (disulfides and glutathione conjugates) and
largely irreversible (sulfinic and sulfonic acid) species.
transient and labile, S-hydroxylation is significant in physio-
logical and pathophysiological events. For example, it has
been shown that this modification plays an essential role in
eukaryotic H2O2 sensing,[5] T-cell activation,[6] enzyme catal-
ysis,[7] as an intermediate in disulfide formation,[8] and
correlates with disease states.[9]
Given the rapidly expanding interest in this field, a
number of strategies have been developed to detect protein S-
hydroxylation.[10] Chemical probes based on the selective
reaction between 5,5-dimethyl-1,3-cyclohexanedione (dime-
done) and sulfenic acid are particularly well suited to monitor
this modification in proteins under biological conditions
(Scheme 2a). Using dimedone-based reagents, many S-
hydroxylated proteins have been observed by immunoblot
and identified by mass spectrometry (MS);[3] however, in vivo
sulfenylation levels are virtually unknown. Knowledge
regarding the degree or extent of S-hydroxylation within
Scheme 2. Chemoselective reactions of cyclic 1,3-diketone derivatives.
a) S-alkylation of a sulfenic acid by dimedone. b) S-alkylation of a thiol
proteins is essential for understanding the function and
by iododimedone.
[*] Dr. K. S. Carroll
Department of Chemistry, The Scripps Research Institute
130 Scripps Way, Jupiter, FL 33458 (USA)
Fax: (+1)561-228-2919
E-mail: kcarroll@scripps.edu
regulation of this oxidative post-translational modification.
In addition, the ability to quantify S-hydroxylation should
help overcome a major hurdle in the field—namely, the
prioritization of proteins within the sulfenome selected for
further characterization and functional analysis.
Dr. Y. H. Seo
Department of Pharmaceutical Sciences, University of Michigan
Ann Arbor, MI 49109 (USA)
Current methods permit the relative abundance of
oxidized or reduced forms to be estimated for intact proteins
using dimedone or 2-nitro-5-thiobenzoic acid (the reduced
form of Ellmanꢀs reagent) and MS.[11] Nonetheless, this
general approach is limited because it: 1) assumes that
different protein species will ionize with equal efficiency,
2) cannot analyze individual sites of modification within a
[**] The authors acknowledge funding from the Camile Henry Dreyfus
Teacher Scholar Award (to K.S.C.) and the American Heart
Association Scientist Development Award (0835419N to K.S.C.).
The authors also gratefully acknowledge helpful discussions with
Profs. U. Jakob and G. Micalizio.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201007175.
1342
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 1342 –1345

protein, and 3) is not applicable to peptide-based proteo-
mic strategies. Herein, we describe a new method that
overcomes these issues based on a class of reagents termed
isotope-coded dimedone and iododimedone (ICDID) for
the quantitative analysis of protein sulfenic acid modifi-
cations.
Our general strategy to quantify protein S-hydroxyl-
ation is motivated by the use of isotope-labeled reagents in
functional proteomics.[12] In the present case, the key
challenge is to devise a set of probes that generate
chemically identical proteins, but differ in the specific
mass of their label depending on the original redox state of
the cysteine residue. Since the selectivity of dimedone for
protein sulfenic acids is well established, our initial goal
was to transform this cyclic 1,3-diketone into a thiol-
reactive probe. From literature examples of thiourea S-
Figure 1. ESI-LC/MS and Western blot analysis shows that iododimedone
selectively labels the thiol group of C36 in C64S C82S Gpx3. a) The
molecular weight of mock-treated C64S C82S Gpx3 is 22736.8 Da corre-
sponding to intact, unmodified protein (top spectrum). The molecular
weight of C64S C82S Gpx3 incubated with iododimedone is 22874.6 Da
alkyation by 1,3-dicarbonyl halides,[13] we reasoned that (bottom spectrum), corresponding to C64S C82S Gpx3 with a single
dimedone adduct (Dm=138 Da). b) C64S C82S Gpx3 (50 mm) untreated or
oxidized with H2O2 (12.5, 25, 50, or 100 mm) followed by dimedone
(20 mm) or iododimedone (20 mm). The resulting samples were analyzed
by Western blot using an antibody that recognizes the protein–dimedone
installation of iodine at the 2-position of the dimedone
scaffold would furnish the requisite trapping reagent
(Scheme 2b; see also Figure S1 in the Supporting Infor-
mation).
adduct (top pannel). Total protein content was determined by an anti-His
With this plan in mind, we prepared iododimedone
according to established procedures,[14] and characterized
its reactivity with low-molecular weight biomolecules by
NMR spectroscopy and LC-MS under simulated physio-
logical conditions. We initially studied the chemical stability
of iododimedone. To this end, the reagent was prepared in
PBS/D2O/10% [D6]DMSO (pH 6, 7, or 8; PBS = phosphate-
buffered saline; DMSO = dimethylsulfoxide) and 1H NMR
spectra were collected over a 24 h period. In all cases,
iododimedone was stable at room temperature and no
degradation was observed (Figure S2). Next, we examined
whether iododimedone would react with glutathione (GSH),
a biologically relevant thiol-containing tripeptide. LC-MS
analysis of the reaction revealed a unique m/z 446 signal
corresponding to the expected alkylation product (Figure S3)
antibody (bottom panel).
peroxide-dependent increase in adduct formation. The exact
opposite trend was observed among reactions that contained
iododimedone owing to oxidation of C36 and the concomitant
loss of the thiol groups, as expected.
With qualitative results demonstrating the feasibility of
our approach in hand, we prepared an isotopically labeled
version of dimedone ([D6]dimedone) to distinguish between
the reaction product of a sulfenic acid and a thiol (i.e.,
D6 Da). Overall, the method we envisioned for ICDID
includes the following sequential steps (Scheme 3): 1) sulfenic
acids are derivatized with [D6]dimedone, 2) excess reagent is
removed and free thiols are labeled with iododimedone,
3) protein samples are proteolyzed, and 4) the resulting
peptides are separated and analyzed by LC-MS. The extent
or fraction of sulfenic acid modification at a particular
cysteine residue can be determined by dividing the heavy-
isotope labeled peak by the sum of the heavy and light-
1
and verified further by H chemical shifts of the cysteine b-
protons in GSH [[D6]DMSO, 400 MHz: d = 2.7 (dd, 1H),
2.6 ppm (dd, 1H)]. Finally, we tested iododimedone for
potential cross-reactivity with a range of other biological
nucleophiles, including the primary e-amine of the lysine side
chain; no undesired reaction was observed (Figure S4).
Next, we tested the ability of iododimedone to modify
cysteine residues within double (C64S C82S) and single
(C64S) mutant forms of recombinant glutathione peroxidase
Gpx3 from yeast.[5] Gpx3 protein was reduced with dithio-
threitol (DTT) and treated with iododimedone or DMSO
alone. The resulting ESI mass spectra acquired for intact
Gpx3 demonstrate selective and quantitative formation of the
expected single and double protein–dimedone adducts (Fig-
ure 1a and Figure S5). Having confirmed thiol-selectivity in a
prototype protein, we investigated the peroxide dependence
of dimedone or iododimedone labeling of C64S C82S Gpx3,
which harbors a single reactive cysteine at position 36 (C36).
Mutant Gpx3 was untreated or oxidized with 0.25 to 2
equivalents of H2O2 and then incubated with dimedone or
iododimedone. Samples were resolved by SDS-PAGE and
analyzed by Western blotting using an anti-dimedone anti-
body (Figure 1b).[9] Dimedone-treated Gpx3 exhibited a
Scheme 3. The ICDID strategy to quantify protein sulfenic acid modifi-
cations.
Angew. Chem. Int. Ed. 2011, 50, 1342 –1345
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1343

Communications
isotope labeled peak intensities in the mass spectrum. As a
[D6]dimedone and iododimedone tagging reactions proceed
to completion. In this regard, we note that no change in the
extent of peptide labeling was observed when the concen-
trations of probes or the length of incubation were increased,
indicating that the labeling reactions were complete. Third, no
further increase in the fraction of sulfenic acid modification at
C36 was observed at 100 mm H2O2. Since intermolecular
disulfide formation was not observed in MS of intact Gpx3,
and it is well established that the catalytic cysteine of thiol
peroxidases is prone to hyperoxidation in the presence of
ꢀ 100 mm oxidant,[15] a likely explanation is that excess H2O2
promotes oxidation of the sulfenic acid at C36 to sulfinic acid
(which is not a target for dimedone or iododimedone).
Consistent with this proposal: 1) the sulfinic acid modified
form of Gpx3 peptide 36–43 was observed at 100 mm H2O2,
and 2) simulated oxidation of C36 accurately reproduces
experimental observations (Figure S6).
proof-of-concept, we applied ICDID toward the quantifica-
tion of sulfenic acid modifications within Gpx3 and the
glycolytic enzyme, glyceraldehyde 3-phosphate dehydrogen-
ase (GAPDH), as detailed below.
We first investigated the approach with C64S C82S Gpx3,
which contains a single cysteine, C36. During the oxidative
part of the catalytic cycle, C36 thiolate reduction of H2O2
leads to sulfenic acid formation. For these experiments, C64S
C82S Gpx3 was untreated or oxidized with 0.25 to 2
equivalents of H2O2 and incubated with [D6]dimedone.
Small molecules were removed and the resulting protein
was treated with iododimedone. The labeled protein was
digested with trypsin and tagged peptides were identified and
quantified by LC-MS. Figure 2a shows the peroxide-depen-
dent changes observed for the molecular ions corresponding
To investigate the ability of ICDID to monitor sulfenic
acid modification at distinct sites within a protein we
performed the analogous set of experiments with C82S
Gpx3. This mutant protein harbors C36 and a second cysteine
residue at position 64 (C64), which is characterized by a high
pKa.[16] Like C64S C82S Gpx3, the fraction of sulfenic acid
modified peptide 36–43 paralleled the increase in H2O2 up to
50 mm (Figure S7). Compared to the single mutant protein,
the overall extent of oxidation was twofold lower. The
structure homology model of Gpx3 indicates that C64 is
located near C36, but biochemical studies show that these
residues do not form a disulfide bond.[5,16] Thus, it is possible
that C64 alters the active-site environment at C36 and
increases its propensity for hyperoxidation. The molecular
ion corresponding to m/z of 639 represents the reaction
product of iododimedone and C64 within peptide 58–67,
whereas no [D6]dimedone-tagged peptide 58–67 was
observed (Figure S7). The absence of sulfenic acid modifica-
tion at C64 is in excellent agreement with earlier studies of
Gpx3, and indicates that S-hydroxylation can be monitored at
discrete sites within a single protein.
Figure 2. Quantifying sulfenic acid modification of C64S C82S Gpx3.
a) C64S C82S Gpx3 (50 mm) was untreated or oxidized with H2O2 and
incubated with [D6]dimedone (20 mm). Small molecules were removed
and the resulting protein treated with iododimedone (50 mm). Labeled
protein was digested with trypsin and tagged peptides were identified
and quantified by LC-MS. The molecular ions corresponding to m/z of
541 and 544 are consistent with the mass of peptide 36–43 modified
by iododimedone (representing the thiol form) or [D6]dimedone
(representing the sulfenic acid form), respectively. b) Fraction of
Finally, we investigated sulfenic acid modification in
GAPDH. This enzyme is well known to undergo sulfenic
acid modification at its active-site cysteine (C149).[1,3,9]
However, in contrast to Gpx3, the reactive cysteine in
GAPDH is significantly less susceptible to hyperoxidation.[17]
For these experiments, GAPDH was untreated or oxidized
with 0.1 to 2 equivalents H2O2 and incubated with
[D6]dimedone. Small molecules were removed, the resulting
sample treated with iododimedone, and reaction products
were analyzed as illustrated in Scheme 3. Figure 3a shows the
peroxide-dependent changes observed for the molecular ions
corresponding to m/z of 992 and 995, which represent peptide
143–159 tagged by iododimedone and [D6]dimedone, respec-
tively. Figure 3b shows the fraction C149 (or C244) present in
the sulfenic acid form plotted as a function of peroxide
concentration. No sulfenic acid modification of C149 within
peptide 143–159 was detected in the absence of H2O2, as
expected. The addition of H2O2 led to an increase in the
fraction sulfenic acid modification at C149, which was
determined as 1.0 at 50 mm oxidant. By contrast, the cysteine
residue within peptide 232–245, C244, showed no
*
sulfenic acid detected at C36 ( ) plotted as a function of H2O2
concentration. The extent of sulfenic acid formation was determined
by dividing the heavy-isotope labeled peak by the sum of the heavy-
and light-isotope labeled peak intensities in the mass spectrum (i.e.,
[RSOH]/[RSH]+[RSOH]). Error bars represent Æs.d. calculated from
duplicate experiments.
to m/z of 541 and 544, which represent the masses of peptide
36–43 labeled by iododimedone (i.e., the thiol form) and
[D6]dimedone (i.e., the sulfenic acid form), respectively.
Figure 2b shows the fraction of C36 present in the sulfenic
acid form plotted as a function of peroxide concentration.
Several features of the ICDID strategy are immediately
apparent from the data reported in Figure 2. First, the fraction
of sulfenic acid modified peptide 36–43 parallels the increase
in H2O2 up to 50 mm. Second, at stoichiometric oxidant
concentration (i.e., 50 mm) approximately one-half of Gpx3
C36 is present in the sulfenic acid form. Accurate quantifi-
cation of sulfenic acid modification requires that the
1344
www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 1342 –1345

in vivo sulfenylation levels and dynamics, particularly in
growth factor pathways and cancer models.
Received: November 15, 2010
Published online: January 5, 2011
Keywords: protein modification · quantitative proteomics ·
.
S-hydroxylation · sulfenic acid · thiol oxidation
[1] L. V. Benitez, W. S. Allison, J. Biol. Chem. 1974, 249, 6234 – 6243.
[2] Reviewed in: a) C. Jacob, G. I. Giles, N. M. Giles, H. Sies,
Angew. Chem. 2003, 115, 4890 – 4907; Angew. Chem. Int. Ed.
2003, 42, 4742 – 4758; b) L. B. Poole, P. A. Karplus, A. Claiborne,
Annu. Rev. Pharmacol. Toxicol. 2004, 44, 325 – 347; c) K. G.
Reddie, K. S. Carroll, Curr. Opin. Chem. Biol. 2008, 12, 746 –
754.
[3] S. E. Leonard, K. G. Reddie, K. S. Carroll, ACS Chem. Biol.
2009, 4, 783 – 799.
[4] Reviewed in: a) L. B. Poole, K. J. Nelson, Curr. Opin. Chem.
Biol. 2008, 12, 18 – 24; b) C. E. Paulsen, K. S. Carroll, ACS
Chem. Biol. 2010, 5, 47 – 62.
[5] C. E. Paulsen, K. S. Carroll, Chem. Biol. 2009, 16, 217 – 225.
[6] R. D. Michalek, K. J. Nelson, B. C. Holbrook, J. S. Yi, D.
Stridiron, L. W. Daniel, J. S. Fetrow, S. B. King, L. B. Poole,
J. M. Grayson, J. Immunol. 2007, 179, 6456 – 6467.
[7] S. Boschi-Muller, S. Azza, S. Sanglier-Cianferani, F. Talfournier,
A. Van Dorsselear, G. Branlant, J. Biol. Chem. 2000, 275, 35908 –
35913.
Figure 3. Quantifying sulfenic acid modification of GAPDH. a) GAPDH
(25 mm) was untreated or oxidized with H2O2 and incubated with
[D6]dimedone (10 mm). Small molecules were removed and the result-
ing protein treated with iododimedone (20 mm). Labeled protein was
digested with trypsin and tagged peptides were identified and quanti-
fied by LC-MS. The peaks corresponding to m/z of 992 and 995
represent the reduced peptide and the sulfenic acid modified peptide,
*
[8] D. S. Rehder, C. R. Borges, Biochemistry 2010, 49, 7748 – 7755.
[9] Y. H. Seo, K. S. Carroll, Proc. Natl. Acad. Sci. USA 2009, 106,
16163 – 16168.
respectively. b) The fraction of sulfenic acid formation at C149 ( ) and
^
C244 ( ) plotted as a function of H2O2 concentration (see Figure S8
for MS data corresponding to peptide 232–245). Error bars represent
[10] Reviewed in: a) N. J. Kettenhofen, M. J. Wood, Chem. Res.
Toxicol. 2010, 23, 1633 – 1646; b) S. E. Leonard, K. S. Carroll,
Curr. Opin. Chem. Biol. 2010, doi:10.1016/j.cbpa.2010.11.012.
[11] D. S. Rehder, C. R. Borges, BMC Biochem. 2010, 11, 25.
[12] Recent examples: a) E. S. Underbakke, Y. Zhu, L. L. Kiessling,
Angew. Chem. 2008, 120, 9823 – 9826; Angew. Chem. Int. Ed.
2008, 47, 9677 – 9680; b) V. Sinha, G. T. Wijewickrama, R. E.
Chandrasena, H. Xu, P. D. Edirisinghe, I. T. Schiefer, G. R.
Thatcher, ACS Chem. Biol. 2010, 5, 667 – 680.
[13] Reviewed in: K. Schank, S. Buegler, H. Folz, N. Schott, Helv.
Chim. Acta 2007, 90, 1606 – 1649.
[14] B. Sreedhar, P. S. Reddy, M. Madhavi, Synth. Commun. 2007, 37,
4149 – 4156.
Æs.d. calculated from duplicate experiments.
[D6]dimedone labeling in the presence of H2O2 (Figure S8), in
accordance with earlier studies of thiol reactivity in
GAPDH.[1,3,9]
To summarize, we have presented ICDID, a new tech-
nique that enables quantification of protein sulfenic acid
modifications. Using several prototypes, we show that this
approach permits S-hydroxylation to be monitored at indi-
vidual cysteines within a single protein and is amenable to
peptide-based proteomic strategies. Another key feature of
this approach is that it allows the occupancy of the site of
modification to be determined. In some proteins, conversion
of sulfenic acid to other oxoforms may preclude absolute
quantification. However, our findings indicate that this
phenomenon is readily diagnosed by peroxide-dependence
studies in cases where the fraction of sulfenic acid modifica-
tion levels off below one. Moreover, we anticipate that this
technology can be extended to monitor changes in protein
disulfides by the inclusion of an additional chemical reduction
step.[18] Going forward, the ICDID approach should provide a
widely applicable means for the quantitative profiling and a
comparison of thiol redox status in physiological and patho-
logical conditions associated with oxidative stress. Ongoing
efforts are focused on extending this approach to profile
[15] H. A. Woo, S. W. Kang, H. K. Kim, K. S. Yang, H. Z. Chae, S. G.
Rhee, J. Biol. Chem. 2003, 278, 47361 – 47364.
[16] L. H. Ma, C. L. Takanishi, M. J. Wood, J. Biol. Chem. 2007, 282,
31429 – 31436.
[17] a) K. ꢁ. Szabꢂ, K. Line, P. Eggleton, J. A. Littlechild, P. G.
Winyard, Structure and Function of the Human Peroxiredoxin-
Based Antioxidant System, Wiley-VCH, Weinheim, 2009; b) C.
Maller, E. Schroder, P. Eaton, Antioxid. Redox Signaling 2011,
14, 49 – 60.
[18] For example, an additional chemical reduction step could be
included after step 2 in Scheme 3. The resulting free thiols could
then be labeled by iododimedone (or an isotopic version of the
iododimedone reagent) and further processed. At any given
cysteine, the fraction present in the thiol, sulfenic acid or
disulfide state could be calculated by comparing data obtained
using the modified procedure to data obtained from samples
processed without the additional reduction and alkylation steps.
Angew. Chem. Int. Ed. 2011, 50, 1342 –1345
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1345