High-throughput method for on-target performic acid oxidation of MALDI-deposited samples

Article abstract of DOI:10.1002/jms.1699

An information-rich on-target performic acid oxidation method, which is compatible with alkylation for differentiation of free cysteine versus disulfide-containing peptides, is described. On-target oxidation is achieved using performic acid vapor to oxidi

Full text of DOI:10.1002/jms.1699

Research Article
Received: 3 September 2009
Accepted: 30 October 2009
Published online in Wiley Interscience: 24 November 2009
(
www.interscience.com) DOI 10.1002/jms.1699
High-throughput method for on-target
performic acid oxidation of MALDI-deposited
samples
∗
Brad J. Williams, William K. Russell and David H. Russell
An information-rich on-target performic acid oxidation method, which is compatible with alkylation for differentiation of
free cysteine versus disulfide-containing peptides, is described. On-target oxidation is achieved using performic acid vapor to
oxidize disulfide-containing peptides and/or small proteins on the matrix-assisted laser desorption/ionization (MALDI) sample
deposits. The on-target oxidation method is preferred over solution-phase oxidation methods because (1) less sample handing
is required, (2) oxidation throughput is drastically increased and (3) ion suppression effects are reduced because performic acid
is not added directly to the MALDI spot. The utility of this method is demonstrated by simultaneous oxidation of multiple MALDI
sample deposits containing model disulfide-linked peptides, intact bovine insulin and a bovine ribonuclease A proteolytic
digest. Copyright ꢀc 2009 John Wiley & Sons, Ltd.
Supporting information may be found in the online version of this article.
Keywords: on-target oxidation; performic acid vapor; disulfide bonds; ribonuclease A; MALDI-MS
Introduction
teins prior to proteolytic digestion and strong cation exchange
(
SCX) chromatography.[ The oxidized cysteine-containing pep-
18]
Disulfide bonds play important roles in stabilizing and main-
tides undergo an elution shift owing to a decrease in peptide
solution-phase charge state, because low charge state peptides
are weakly retained on the SCX column. The resulting enriched
oxidized cysteine-containing peptides are then subjected to off-
line MALDI-MS/MS followed by peptide identification via database
searching ordenovo sequencing. Thismethodologyutilizesacom-
bination of oxidation, MS and database searching for analyzing
the cleaved disulfide bonds of bovine serum albumin. Because
performic acid cleaves the disulfide bonds prior to digestion, the
native disulfide bond pattern is lost; therefore, it is important to
maintain intact disulfide-linked peptides throughout proteolytic
digestion to characterize the disulfide bond connectivity within
proteins.
[
1,2]
taining native protein structure,
and mass spectrometry
(
MS) plays an increasingly important role in deciphering these
[
3]
[4–6]
processes, especially protein folding/unfolding kinetics.
Disulfide bond locations within proteins can be determined us-
ing a variety of MS-based methods, including tandem MS,
databasesearchalgorithms,
icaloxidation[
[
7,8]
[
9–12]
[13–15]
chemicalreduction,
chem-
16–18]
andmetalioncleavageofdisulfidebonds.[19,20]
Xu and coworkers developed a tandem MS search engine to sim-
plify database searches for disulfide-linked peptides fragment ion
spectra,[ but such spectra can be complicated by direct cleav-
age of the disulfide bond between interlinked disulfide peptides,
which results in a limited number of sequence-informative frag-
ment ions from each α- and β-chain peptide forming the disulfide
linkage.[ Disulfide-linked peptides can also be identified through
chemical reduction of the disulfide bond. For example, a reduc-
tive matrix-assisted laser desorption/ionization (MALDI) matrix,
12]
Although performic acid oxidation is widely used, there are sev-
eralchallengesassociatedwithsolution-phaseoxidationmethods.
For example, typical solution-phase oxidation methods are per-
7]
formed on lyophilized peptide/protein samples using reaction
[
14]
[18,21,22]
1,5-diaminonapthalene (1,5-DAN),
was used by Quinton and
times ranging from 2 to 4 h
and several steps are re-
coworkers to characterize disulfide-bridged peptides using liq-
uid chromatography (LC) coupled off-line with MALDI-MS/MS.[
This reductive MALDI matrix successfully reduces disulfide bonds
during the MALDI process; however, 1,5-DAN also yields a high
relative abundance of matrix clusters up to 1000 m/z, which com-
plicates the analysis of low molecular weight disulfide and/or
cysteine-containing peptides.
Chemical oxidation offers the following benefits over perform-
ing chemical reduction of the disulfide bond: (1) a 48-Da mass
shift occurs per cysteine residue, (2) cysteic acid (SO3H) enhances
negative ion formation and (3) the performic acid simultane-
ously cleaves the disulfide bond and chemically modifies each
peptide involved in the disulfide bridge. For example, performic
acid oxidation has been used to study disulfide-containing pro-
quired following oxidation to remove the performic acid reagent
15]
[18,21]
prior to MS analysis.
Some solution-phase oxidation meth-
ods have decreased reaction times to 10–30 min for oxidation
of cysteine/methionine-containing peptides; however, the ad-
ditional clean-up steps following oxidation make this method
less desirable.[ On-target oxidation methods using solution-
phase performic acid have also been successfully applied to
17]
[
21]
MALDI spots containing peptides and/or proteolytic digests.
∗
Correspondence to: David H. Russell, Department of Chemistry, Texas A&M
University,CollegeStation,TX77843, USA.E-mail: russell@mail.chem.tamu.edu
Department of Chemistry, Texas A&M University, College Station, TX 77843,
USA
J. Mass. Spectrom. 2010, 45, 157–166
Copyright ꢀc 2009 John Wiley & Sons, Ltd.

B. J. Williams, W. K. Russell and D. H. Russell
For example, we showed that directly adding the performic acid
solution to a prespotted MALDI deposit containing a bovine apo-
transferrin tryptic digest yields oxidized tryptic peptides; however,
the signal-to-noise (S/N) ratios of the oxidized peptide negative
ions decreased by fivefold because of ion suppression effects.
The overall amino acid sequence coverage also decreased by half
compared to the solution-phase oxidized sample. Thus, an on-
target performic acid oxidation method that can mitigate these
adverse effects would be highly beneficial. The ideal on-target
oxidation method would offer the following characteristics over
solution-phase oxidation methods: (1) simple experimental pro-
cedure with no post-oxidation sample cleanup; (2) decreased ion
suppression effects following oxidation; (3) improved through-
put over solution-phase methods and (4) maintain positive and
negative ion mass spectral quality following on-target oxidation.
This report describes an on-target performic acid oxidation
method using performic acid vapor to oxidize disulfide-containing
peptides/proteins deposited for MALDI-MS analysis. Brown and
Hartley have reported a similar approach using performic acid
vapor to oxidize disulfide-linked peptides on electrophoresis
paper.[ Thenovelaspectofourmethodisfoundwithintheability
to successfully oxidize multiple MALDI spots simultaneously using
performicacidvaporcontainedwithinareactionchamberwithout
sample cleanup. Reaction times for the on-target performic acid
oxidation method (10–60 min) are also shorter than traditional
solution-phase oxidation methods (2–4 h). We apply this on-
targetperformicacidoxidationmethodtoamodeldisulfide-linked
peptide, intact bovine insulin and a proteolytic digest of bovine
ribonuclease A. In addition, we discuss the advantages of this
method over solution-phase oxidation methods and show the
ability to elucidate disulfide-linked peptides from ribonuclease A.
MALDI matrix) to α-cyano-4-hydroxycinnamic acid (CHCA) matrix.
CHCA was prepared at 5 mg/ml in a 2 : 3 ddH2O : CH3OH solution
containing10 mM dihydrogenammoniumphosphate(NH4H2PO4)
and 1 µl was prespotted using the underlayer method.[
24]
Stock solutions of α-melanocyte stimulating hormone (Ac-
SYSMEHFRWGKPV-NH2) and α1-mating factor (WHWLQL) were
prepared at 10 µM in 10 mM formic acid. The resulting solutions
were then mixed 1 : 1 (v/v) with CH3OH and 1 µl (5 pmol)
was spotted onto the MALDI target as an overlayer to CHCA
matrix. These MALDI deposits were initially oxidized using the on-
◦
target oxidation methods described below (−20 C for 60 min).
Additional oxidation was performed on the same MALDI deposits
◦
using room temperature (+24 C) oxidation for an additional
60 min.
Astocksolution(1 mg/ml)of bovineinsulinwasdilutedto10 µM
in 10 mM formic acid. The insulin solution was then mixed 1 : 1 (v/v)
withCH OHand1 µl(5pmol)wasspottedontotheMALDItargetas
3
an overlayer to prespotted MALDI matrix. The on-target oxidation
experiments using intact insulin were performed with three
different MALDI matrices: CHCA, 2,4-dihydroxyacetophenone (2,4-
DHAP)and2,4-DHAPcombinedwithCHCAasdualMALDImatrices.
For the experiments with CHCA and 2,4-DHAP (10 mg/ml in 2 : 3
ddH O : CH OH with 10 mM NH H PO ), the prespotted MALDI
23]
2
3
4
2
4
deposit was re-dissolved with an overlayer of insulin (1 µl of
5 µM insulin in 1 : 1 (v/v) CH OH : 10 mM formic acid). For the
3
dual MALDI matrices experiment, 2,4-DHAP was spotted as an
underlayer followed by an overlayer of insulin, then subjected to
on-target oxidation. Following oxidation, the 2,4-DHAP + insulin
sample deposit was re-dissolved with an overlayer of CHCA prior
to MS analysis. The intact insulin experiments with the varied
MALDI matrices were oxidized using the on-target performic
acid oxidation methods described below, with the exception of
◦
performing oxidation at room temperature (+23 C) for 60 min.
Experimental Section
Ribonuclease A proteolytic digestion
Chemicals
All chemicals were purchased from Sigma-Aldrich (St Louis, MO)
unless otherwise noted. Sequencing grade-modified trypsin was
obtained from Promega (Madison, WI). Oxytocin (CYIQNCPLG-
NH2) was purchased from American Peptide Company (Sunnyvale,
CA). Hydrogen peroxide (35% w/w) and formic acid (99% w/w)
were obtained from Acros Organics (Morris Plains, NJ). HPLC
grade acetone, acetonitrile (CH3CN) and methanol (CH3OH)
were purchased from EMD Chemicals Inc. (Gibbstown, NJ).
All experiments were performed with 18-Mꢀ water (ddH2O),
purified using a water purification unit (Barnstead International,
Dubuque, IA).
Dual protease digestion (trypsin followed by chymotrypsin) of
bovine ribonuclease A (RNase A) was performed using the meth-
ods described by Xu and coworkers with minor modifications.
[
12]
Briefly, RNase A was dissolved at 1 mg/ml in a 25 mM ammonium
acetate buffer adjusted to pH 6.0 with 10 mM formic acid. RNase
A was initially digested with trypsin [1 : 50 (w/w) enzyme : protein
◦
ratio] for 4 h at 37 C followed by chymotrypsin addition (1 : 50
(w/w) enzyme : protein ratio) and incubated for an additional 4 h.
The RNase A proteolytic digest was alkylated following proteolytic
digestion with 5 µl of 100 mM 2-iodoacetamide and incubated in
the dark for 45 min at room temperature. Following alkylation, the
sample was subjected to cleanup using a 10-µl C18 resin pipette
tip. The RNase A proteolytic digest was eluted with a 60 : 40
CH3CN : ddH2O solution containing 0.1% formic acid and 1 µl of
the eluate was spotted as the overlayer to CHCA matrix.
Sample preparation for MALDI-MS
A 15-µM stock solution of native oxytocin was prepared in 10 mM
formic acid. Oxytocin was reduced by adding 2 µl of 50 mM tris(2-
◦
carboxyethyl)phosphine and incubating at 60 C for 60 min. The
On-target and solution-phase performic acid oxidation
synthetic mixture of native and reduced oxytocin (1 : 1 molar ratio
of native : reduced oxytocin) was reacted with 2-iodoacetamide
by increasing the solution pH from 3.0 to 7.1 with 40 µl of 25 mM
ammonium bicarbonate (NH4HCO3) and incubated in the dark
for 45 min at room temperature. The excess 2-iodoacetamide
and NH4HCO3 was removed using a 10-µl C18 resin pipette tip
Performic acid was prepared using hydrogen peroxide (35% w/w)
and formic acid (99% w/w) mixed at a 1 : 9 ratio, respectively, and
allowedtoageatroomtemperaturefor2 h.[ On-targetperformic
acid oxidation using performic acid vapor was performed by
placing 40 µl of a 1 : 1 (v/v) mixture of acetone : performic
acid directly on the MALDI target in an area that does not
contain sample (distance from the performic acid reagent to
the MALDI sample deposit is not critical). The MALDI target was
21]
(Millipore, Billerica, MA) and eluted with a 60 : 40 CH3CN : ddH2O
solution containing 0.1% formic acid. A 1-µl aliquot of the eluate
was spotted as the overlayer (analyte deposited on prespotted
www.interscience.wiley.com/journal/jms
Copyright ꢀc 2009 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2010, 45, 157–166

On-target oxidation of MALDI samples
◦
then covered in a plastic petri dish and placed in a −20 C freezer
for 10–60 min. Following on-target performic acid oxidation, the
MALDI plate was allowed to reach ambient temperature then
subjected to MALDI analysis without sample cleanup. Solution-
phase performic acid oxidation was performed using a 1 : 19 ratio
of hydrogen peroxide : formic acid as previously described.[ The
performic acid solution was used to resuspend vacuum-dried
peptide samples or RNase A proteolytic digests then allowed to
The disulfide-containing peptide oxytocin (CYIQNCPLG-NH2) is
usedtoillustratethereactionshowninScheme1(c),andasynthetic
mixture of native and chemically reduced oxytocin is used to show
the mixture described in Scheme 1(b). The synthetic mixture of
native and reduced oxytocin was reacted with 2-iodoacetamide,
and Fig. 1 contains (a) positive and (b) negative ion MALDI-MS
spectra for this sample taken after on-target performic acid
oxidation of the native and CAM-modified oxytocin mixture. Ion
25]
◦
+
react at 0 C for 4 h. Following oxidation, samples were diluted
signals for native oxytocin (m/z 1007.48 for [M + H] ) and signals
+
with 500 µl of ddH2O and vacuum dried. Samples were then
resuspended in ddH2O and mixed 1 : 1 (v/v) with CH3OH prior to
spotting on the MALDI target as an overlayer.
for alkylation products (m/z 1066.53 for [M + H + CAM] and m/z
+
1
123.55 for [M + H + 2CAM] ) are observed in the positive ion
mass spectrum after oxidation (Fig. 1(a)). These assignments were
confirmed by tandem MS (Supporting Information, Fig. S1(b,c)).
The intact disulfide bond of native oxytocin is cleaved and
MALDI-MS and MALDI-MS/MS
−
converted to cysteic acid, as evidenced by the [M − H + O6] ion
All MALDI-MS experiments were performed using a model
700 Proteomics Analyzer MALDI-TOF/TOF (Applied Biosys-
(
m/z 1103.37) in the negative ion mass spectrum (Fig. 1(b)). The
4
negative ion tandem mass spectrum for fully oxidized oxytocin
tems, Framingham, MA). The MALDI-MS data were acquired
using the reflectron detector in both positive and negative
ion modes using 1200 laser shots with external calibrants:
bradykinin (2–9) (PPGFSPFR) and adrenocorticotropic hormone
can be found in Supporting Information, Fig. S1(a). Following
oxidation, [M + H]+ ions of native oxytocin are still observed
in the positive ion mass spectrum (Fig. 1(a)), which suggests an
incompleteoxidationofnativeoxytocin. Experimentsarecurrently
underway to address the issues related to incomplete oxidation,
and preliminary data for these on-target oxidation optimization
studies are discussed below. Despite incomplete conversion to
the fully CAM-modified oxytocin, ion signals for [M + H +
(ACTH) (18–39) (RPVKVYPNGAENESAEAFPLEF). Two different in-
strument parameters were used for MS experiments with pep-
tides/proteolytic digests (500–4500 Da, 2100 Da focus mass) and
intact bovine insulin (700–6000 Da, 3500 Da focus mass). The
1
18
123
nonoxidized forms of RNase A proteolytic peptides, YPNCAY
+
−
CAM + O3] or [M − H + CAM + O3] are not observed in
the positive or negative ion mass spectra. Therefore, the single
CAM-modified oxytocin does not complicate the differentiation
between a peptide that contains two free cysteine residues from
one that contains an intralinked disulfide bond. Nonetheless, the
7
3
87
and VHESLADVQAVCSQK were used to internally calibrate
the negative ion mass spectrum of the RNase A proteolytic di-
gest. Collision-induced dissociation (CID) tandem MS spectra were
acquired using 20% greater laser power than the MS spectra ac-
quisition with atmosphere (medium pressure) as the collision gas
with 1 kV of collision energy. The average coefficient of variation
2
1
-iodoacetamide alkylation chemistry incorporated into Scheme
(c) allows a multiple free cysteine-containing peptide to be
(
CV) was calculated to be ± 6% for triplicate analysis of an RNase
differentiated from an intralinked disulfide-bridged peptide such
as oxytocin.
A proteolytic digest subjected to on-target oxidation. Normalized
ion signals were determined using a ratio of the single peptide ion
count to the total ion count for five oxidized RNase A peptides.
The normalized ion signals were then used to calculate the CV.
The effect of on-target oxidation on peptides that contain
other oxidizable amino acids, such as histidine, methionine,
tryptophan and tyrosine, was investigated using α-melanocyte
stimulating hormone (Ac-SYSMEHFRWGKPV-NH2) and α1-mating
factor (WHWLQL). Figure 2(a) contains a positive ion mass
spectrum for α-melanocyte stimulating hormone taken before
oxidation. Note that several oxidized peptide ion signals ([M +
Results and Discussion
The main motivation of this work is to develop a method to si-
multaneously scan complex mixtures for disulfide-linked peptides
and be able to distinguish between free cysteine and bridged
disulfide-containing peptides. Our strategy is to use thiol alky-
lation chemistry followed by on-target performic acid oxidation.
Scheme 1 illustrates the resulting performic acid oxidation prod-
uctsforboth(a) interlinkedand(b) intralinkeddisulfide-containing
peptides. For example, an interlinked disulfide-bridged peptide
will form oxidized products for both α- and β-chain peptides
+
+
H + O] , [M + H + O2] , etc.) are present in this spectrum.
These oxidized ion signals are a result of oxidation owing to air
◦
exposure over time. Following on-target oxidation (1 h at −20 C)
(
Fig. 2(b)), the intensities of the oxidized peptide ion signals do
+
not increase relative to the [M + H] ion, thus it appears that
on-target oxidation does not significantly oxidize other amino
acids under cold oxidation conditions. The same MALDI sample
◦
deposit was subjected to an additional hour of oxidation at +24 C
and the abundance of the oxidized ion signals do not increase
relative to the [M + H]+ ion signal for α-melanocyte stimulating
hormone, thus it appears that oxidation is not occurring (Fig. 2(c)).
As a control experiment, native oxytocin (CYIQNCPLG-NH2) was
spottedonaseparatesampledepositandoxidizedsimultaneously
on the same MALDI target, and the MALDI mass spectrum from
(Scheme 1(a)). The performic acid cleaves the disulfide bond to
convert each cysteine thiol to cysteic acid (SO3H), whereas an
intralinked disulfide-containing peptide will form a single peptide
containing two cysteic acid residues (Scheme 1(b)). Note that a
peptide containing two free thiol side chains will yield the same
product as Scheme 1(b). Incorporating an alkylation reagent, such
as 2-iodoacetamide, into the overall reaction scheme provides dif-
ferentiation between peptides with two free cysteine residues or
an intralinked disulfide bond. The carbamidomethyl (CAM) group
protects the free cysteine residues from oxidation, while the in-
tralinked disulfide-bridged peptide will be fully oxidized (Scheme
−
this sample contains a strong signal for the [M − H + O6]
ion (m/z 1103.44) (Fig. 2(d)), indicating the oxidation reaction
was successful. Similar experiments for α1-mating factor also
resulted in a minimal increase in the relative abundance of the
oxidizedpeptideionsignalsasaresultofoxidation(seeSupporting
Information,Fig. S2),thusweconcludethattheon-targetoxidation
1
(c)).
J. Mass. Spectrom. 2010, 45, 157–166
Copyright ꢀc 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/jms

B. J. Williams, W. K. Russell and D. H. Russell
Scheme 1. Performic acid oxidation products of (a) interlinked and (b) intralinked disulfide-linked peptides, and (c) a mixture of an intralinked
disulfide-linked peptide and the 2-iodoacetamide alkylated form.
Figure 1. (a) Positive ion and (b) negative ion mass spectrum of native and CAM-modified oxytocin after 10 min of on-target performic oxidation
◦
performed at −20 C.
is not complicated by reactions with H, M, W or Y amino acid
containing peptides.
Bovine insulin is an excellent model system to illustrate on-
Figure 3 contains the negative ion mass spectra of oxidized
intact insulin obtained using several different sample preparation
methods. The spectrum contained in Fig. 3(a) contains a strong ion
target performic acid oxidation of disulfide-containing proteins,
because it contains three disulfide bonds, two interchain disulfide
bonds, which link the α- and β-chains, and one intrachain disulfide
bond within the α-chain (see structure). Performic acid should
cleave both the intra- and interlinked disulfide bonds to fully
oxidize both α- and β-chains.
−
signalforthefullyoxidizedβ-chain[β−H+O6] atm/z3492.83,but
under these conditions ion signals for the fully oxidized insulin α-
chainarenotobservedineitherpositiveornegativeionspectra.On
www.interscience.wiley.com/journal/jms
Copyright ꢀc 2009 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2010, 45, 157–166

On-target oxidation of MALDI samples
Figure 2. Typical positive ion mass spectra for α-melanocyte stimulating hormone (Ac-SYSMEHFRWGKPV-NH2) (a) before and (b) after an hour of
◦
on-target performic acid oxidation at −20 C. Panel (c) represents an additional hour of on-target oxidation performed on the same sample deposit at
◦
+
24 C. Panel (d) contains a negative ion mass spectrum of a control experiment performed using native oxytocin (CYIQNCPLG-NH2) oxidized in parallel
with α-melanocyte stimulating hormone on the same MALDI target. Note the change in the m/z scale because of the low molecular weight of oxidized
oxytocin. Ion signals denoted with closed diamonds (ꢀ) represent contaminate ions arising from the CHCA matrix.
J. Mass. Spectrom. 2010, 45, 157–166
Copyright ꢀc 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/jms

B. J. Williams, W. K. Russell and D. H. Russell
Figure 3. Negative ion mass spectra after on-target oxidation of (a) intact insulin and (b) chemically reduced insulin using CHCA as the MALDI matrix
during oxidation. Panel (c) 2,4-DHAP was used as the MALDI matrix during on-target oxidation of intact insulin. Panel (d) 2,4-DHAP was used during
oxidation followed by an overlayer addition of CHCA, which we found to improve the yield of negative ions.
www.interscience.wiley.com/journal/jms
Copyright ꢀc 2009 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2010, 45, 157–166

On-target oxidation of MALDI samples
Table 1. Summary of negative ion MALDI-MS data resulting from on-target performic acid oxidation of RNase A tryptic/chymotryptic peptides
Label
Peptide sequence
[M − H]−obs
[M − H]−calc
Mass error
ppm
−28
−21
−15
−20
−18
−20
−14
−17
−10
0
8
8
5
1
1
9
6
5
1
8
2
2
NVACK92 + O3
A
B
C
D
E
F
G
H
I
580.2236
681.2732
736.1478
740.2736
776.2419
845.2562
854.3588
864.2386
1261.5394
1659.7649
580.2398
681.2875
736.1585
740.2883
776.2559
845.2733
854.3709
864.2535
1261.5521
1659.7646
AVCSQK87 + O3
CNQMM56 + O7
06
18
3
SITDCR111 + O3
YPNCAY123 + O3
NGQTNCY99 + O3
CKPVNTF72 + O3
6
2
CNQMMK57 + O7
31HIIVACEGNPY + O3
141
J
73VHESLADVQAVCSQK87 + O3
the other hand, ion signals at m/z 2463.08 are most likely partially
cleaved to form two peptides each containing a cysteic acid side
chain (Scheme 1(a)). Each oxidized peptide chain is then subjected
totandemMS, wheretheaminoacidsequencesof thetwopeptide
chains forming the disulfide bond can be determined.
−
oxidized α-chain, [α −H+O9 −17] calc = 2463.89 Da (−333 ppm
−
mass error) or [α − H + O8] calc = 2463.94 Da (−353 ppm mass
error), but we were unable to confirm these assignments by using
tandem MS owing to the low abundance of the signals. Note
that oxidation of chemically reduced insulin yields abundant ion
signals for both α- and β-chains, i.e. the fully oxidized insulin α-
Figure 4 contains negative ion mass spectra for the RNase
A trypsin/chymotrypsin digest + 2-iodoacetamide acquired
(a) before oxidation, (b) after on-target and (c) solution-phase
oxidation. Fourteen new peptide ion signals resulting from
performic acid oxidation are observed and those confirmed by
tandem MS are labeled A–J (Fig. 4(b)), while the four additional
ion signals too low in abundance to characterize by tandem MS
−
−
chain [α −H+O12] ion at m/z 2528.67 and β-chain [β −H+O6]
ion at m/z 3492.34 are observed in the negative ion mass spectrum
(Fig. 3(b)).
Figure 3(c) contains the negative ion mass spectra for oxidized
∗
intact insulin from 2,4-DHAP, and Fig. 3(d) contains similar spectra
obtained from 2,4-DHAP prepared using an overlayer of CHCA.
When on-target oxidation is performed on deposits of 2,4-DHAP,
are labeled with asterisks ( ). The tandem MS confirmed amino
acid sequences of the oxidized RNase A peptides are summarized
in Table 1 and the corresponding negative ion tandem mass
spectra are included in Supporting Information, Fig. S4. Several
−
the fully oxidized insulin α-chain [α − H + O12] and β-chain
−
52
57 93
99
[
β −H+O6] ions are observed in the negative ion mass spectrum
cysteine-containing peptides CNQMMK , NGQTNCY and
8
8
92
(
Fig. 3(b)). Note that the negative ion yields for both α- and β-
NVACK only appear after oxidation, which is mostlikely a result
chain oxidized forms are enhanced using 2,4-DHAP/CHCA sample
preparation. The signal-to-noise (S/N) ratio of the oxidized insulin
α-chain increases 1.5-fold for the dual MALDI matrix method
relative to using 2,4-DHAP as a MALDI matrix alone.
of oxidative cleavage of the intact disulfide-linked peptides. The
oxidizedaminoacidresiduesaredenotedinbolditalics. Proteolytic
5
2
57
peptides containing methionine residues (e.g. CNQMMK and
5
2
56
CNQMM ) were also oxidized owing to the incorporation of
It appears that efficient oxidation of intact insulin to form
oxidized α- and β-chains is related to greater accessibility of the
performic acid vapor to oxidize the insulin molecules within the
two additional oxygen atoms per methionine residue. Ten total
oxidized peptides accounting for all eight RNase A cysteine
residues involved in disulfide bonds were detected using on-
target oxidation, whereas a total of eight (six of these peptides
are involved in disulfide bridging) oxidized RNase A peptides
were detected using solution-phase oxidation. Comparing the
mass spectral quality of the on-target (Fig. 4(b)) and solution-
phase oxidation (Fig. 4(b)) method clearly illustrates an advantage
of the on-target oxidation method. At first glance, the on-
target oxidation method (Fig. 4(b)) appears to contain additional
chemical noise above m/z 1000; however, the majority of these
2,4-DHAP crystals. The presence of non-oxidized intact insulin
ions following oxidation could indicate that insulin molecules are
trapped within CHCA crystals, probably as inclusion complexes,[
whereas with 2,4-DHAP, the insulin molecules are located on
the 2,4-DHAP crystal surface and become more susceptible
to oxidation (Fig. 3(c)). Further investigation of different MALDI
matrices used for the on-target performic acid oxidation method
may provide additional insight to explain why some cysteine
and/or disulfide-containing peptides are more readily oxidized.[
RNase A is an excellent model protein to study disulfide bonds
in proteins because it contains eight cysteine residues, each
26]
27]
−
ion signals correspond to [M − H] ions for intact disulfide-linked
peptides that are present after oxidation. As for the solution-phase
oxidationmethod(Fig. 4(c)),theseionsignalsfromm/z 1000–2000
are not observed because these intact disulfide-linked peptides
have undergone oxidative cleavage to produce small oxidized
cysteine-containing peptides. In terms of S/N improvement, the
S/N ratios for the labeled ion signals increase an average 2.5-fold
for the on-target oxidation method relative to the solution-phase
oxidation method. A drastic S/N enhancement occurs for peptide
[
12,17,25]
present as disulfide bonds.
RNase A contains four native
disulfidebonds, threedisulfidebondsareinterlinkedandtheother
disulfide bond is intralinked. Our experimental design involves
performing proteolytic digestion under mildly acidic conditions
[
12]
(pH 6) to minimize disulfide scrambling
combined with 2-
iodoacetamide alkylation chemistry to protect the free cysteine
residues (i.e. disulfide-linked peptides that may become reduced
during digestion) from oxidation. Following proteolytic digestion,
the intact disulfide-containing peptides are then subjected to on-
target performic acid oxidation, where the disulfide bond(s) are
1
31
141
I ( HIIVACEGNPY
+ O3), where the ratio of the S/N values
for the on-target oxidation method improves ca. sixfold over
the solution-phase oxidation method. The decrease in relative
abundance for the solution-phase oxidation method could be
J. Mass. Spectrom. 2010, 45, 157–166
Copyright ꢀc 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/jms

B. J. Williams, W. K. Russell and D. H. Russell
Figure 4. Negative ion mass spectra for an RNase A trypsin/chymotrypsin + 2-iodoacetamide digest (a) before and (b) after on-target performic acid
◦
oxidation and (c) solution-phase oxidation. For this example, the on-target oxidation was performed for 1 h at −20 C and the solution-phase oxidation
◦
∗
was performed for 4 h at 0 C. The identities of the peptide ion signals labeled A–J are found in Table1. The peptides denoted with asterisks ( ) correspond
to new peptide ion signals resulting from performic acid oxidation not confirmed by tandem MS. The peptide ion signals denoted with closed diamonds
(
ꢀ) indicate chemical contaminates arising from the CHCA matrix.
attributed to ion suppression effects or sample loss associated
with post-oxidation vacuum drying prior to MS analysis. Overall,
the RNase A negative ion MS data suggest that the on-target
oxidation method is suitable for proteolytic protein digests.
Several disulfide-linked peptides can be elucidated from the
observed oxidized cysteine-containing peptides summarized in
Table 1. The correlation of the non-oxidized peptide masses and
thecalculateddisulfide-linkedpeptidemassesisshowninFig. 5(b).
Thepeptidesarearrangedaccording toincreasing masstoforman
arrayofcalculateddisulfide-linkedpeptidemasses.Allthedifferent
disulfide-linked peptide bridging possibilities are calculated from
the list of peptide m/z values accounting for the mass difference
of an interlinked disulfide bond. The calculated disulfide-linked
peptide masses that match an ion signal present in the positive
ion mass spectrum of the oxidized RNase A trypsin/chymotrypsin
digest (Fig. 5(a)) are indicated with a box around the number. For
example, an ion signal at m/z 1535.56 is present in the positive ion
mass spectrum that corresponds to the disulfide-linked peptide
www.interscience.wiley.com/journal/jms
Copyright ꢀc 2009 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2010, 45, 157–166

On-target oxidation of MALDI samples
Figure 5. (a) Positive ion mass spectrum of the RNase A trypsin/chymotrypsin digest + 2-iododactamide following on-target oxidation. The peptide ion
signals labeled with their corresponding m/z values indicate disulfide-linked peptides from RNase A elucidated from the oxidized negative ion mass
spectrum. Inset figure: expanded view of mass range 1315–1335 m/z. (b) Table correlating the observed oxidized cysteine-containing peptides with their
+
resulting candidate interlinked disulfide-bridged peptides. Shown are their calculated [M + H] values for their nonoxidized counterparts. The calculated
+
[
M + H] interlinked disulfide-bridged peptide masses highlighted with a box indicate matched m/z values in the above positive ion mass spectrum. The
calculated disulfide-linked peptides that do not match ion signals present in the positive ion mass spectrum were removed for clarity.
(118
YPNCAY / CKPVNTF ), which is a native disulfide linkage
123 66
72
peptides, intact insulin and a proteolytic digest of RNase A.
Cysteine, cystine and methionine amino acid side chains are
readilyoxidizedafteron-targetoxidation. Ontheotherhand, other
oxidizable amino acids such as histidine, tryptophan and tyrosine
are not significantly oxidized as a result of on-target oxidation.
The oxytocin experiments show the compatibility of the on-
target oxidation method with alkylation chemistry to distinguish
between a peptide containing two free cysteine residues from
a peptide with a single intralinked disulfide bond. The intact
insulin experiments illustrate the ability to oxidize small disulfide-
containing proteins and suggest that the MALDI matrix plays a
role in the oxidation process by the matrix crystals affecting the
ability of the performic acid vapor to reach the analyte. All four
native RNase A disulfide bonds were successfully elucidated from
the negative ion MALDI-MS data from the on-target performic acid
oxidation of an RNase A proteolytic digest.
The on-target oxidation method is advantageous over solution-
phase oxidation methods because of the decreased sample
handlingandreactiontime. Anothermajorbenefitoftheon-target
oxidation method is the ability to perform simultaneous oxidation
of several MALDI sample deposits. In addition, the positive and
negative ion mass spectral quality is not sacrificed following
oxidation.Theon-targetperformicacidoxidationmethodprovides
a very powerful technique to distinguish between free cysteine-
containing peptides and those peptides that contain disulfide
6
6
121
between Cys and Cys . This assignment was further confirmed
by tandem MS by observing several b-type fragment ions and the
signature fragment ions for an interlinked disulfide bond. The ion
signal at m/z 1677.66 is most likely the disulfide-linked peptide
8
4
87 131
141
+
(
CSQK / HIIVACEGPNY )withan[M+H] calc = 1677.77 Da
−67 ppm mass error) and a 170.09 Da mass difference (Ala-Val
(
residue mass = 170.11 Da) from the m/z 1847.75 ion signal. All
four native disulfide bonds from RNase A can be elucidated from
the oxidized cysteine-containing peptides resulting from the on-
target oxidation of the RNase A proteolytic digest. At this point we
are not claiming a method for complete disulfide bond elucidation
fromasingleMALDIdeposit;thereareseveralotherpossibleRNase
A disulfide-linked peptides that can be produced from different
proteolytic digestion products that were not identified. Therefore,
we plan to further expand on the disulfide bond elucidation
capability using this on-target oxidation method in combination
with LC-MALDI-MS to deconvolute the negative ion mass spectra
prior to oxidation.
Conclusions
On-target oxidation using performic acid vapor was achieved
for MALDI-deposited samples containing model disulfide-linked
J. Mass. Spectrom. 2010, 45, 157–166
Copyright ꢀc 2009 John Wiley & Sons, Ltd.
www.interscience.wiley.com/journal/jms

B. J. Williams, W. K. Russell and D. H. Russell
bonds. In addition, this method will make a drastic impact for
disulfide-linked peptide characterization from complex mixtures,
owing to the ability to rapidly screen an entire MALDI target for
sample deposits that contain enhanced negative ion signals, i.e.
these are MALDI spots that may contain candidate disulfide-
linked peptides. Hence, the motivation to combine the on-
target oxidation method with off-line separations (e.g. capillary
[11] S. Wefing, V. Schnaible, D. Hoffmann. SearchXLinks. A Program for
the Identification of Disulfide Bonds in Proteins from Mass Spectra,
Anal. Chem. 2006, 78, 1235.
[
12] H. Xu, L. Zhang, M.A. Freitas. Identification and Characterization of
Disulfide Bonds in Proteins and Peptides from Tandem MS Data by
Use of the MassMatrix MS/MS Search Engine, J. Proteome Res. 2008,
7, 138.
[
[
13] H.R. Morris, P. Pucci. A New Method for Rapid Assignment of S-S
Bridges in Proteins, Biochem. Biophys. Res. Commun. 1985, 126,
[
28,29]
electrophoresis or LC) coupled with MALDI-MS
to analyze
1122.
disulfide-linked peptides from complex proteomic mixtures.
14] Y. Fukuyama, S. Iwamoto, K. Tanaka. Rapid seqencing and disulfide
mapping of peptides containing disulfide bonds by using 1,5-
diaminonaphthalene as a reductive matrix, J. Mass Spectrom. 2006,
Acknowledgements
41, 191.
[
[
15] L. Quinton, K. Demeure, R. Dobson, N. Gilles, V. Gabelica, E.D. Pauw.
New Method for Characterizing Highly Disulfide-Bridged Peptides
in Complex Mixtures:Application to Toxin Identification from Crude
Venoms, J. Proteome Res. 2007, 6, 3216.
16] S.S. Thakur, P. Balaram. Fragmentation of Peptide Disulfides under
Conditions of Negative Ion Mass Spectrometry: Studies of Oxidized
Glutathione and Contryphan, J. Am. Soc. Mass Spectrom. 2008, 19,
This work was supported by the Department of Energy, Division of
Chemical Sciences, Offices of Basic Energy Sciences, BES DE-FG02-
0
4ER15520 and the National Institutes of Health, RR019587.
Supporting information
3
58.
Supporting information may be found in the online version of this
article.
[
[
17] Y. Sun, D.L. Smith. Identification of Disulfide-Containing Peptides
by Performic Acid Oxidation and Mass Spectrometry, Anal.Biochem.
1
18] J. Dai, J. Wang, Y. Zhang, Z. Lu, B. Yang, X. Li, Y. Cai, X. Qian.
Enrichment and Identification of Cysteine-Containing Peptides
from Tryptic Digests of Performic Oxidized Proteins by Strong
Cation Exchange LC and MALDI-TOF/TOF MS, Anal. Chem. 2005, 77,
988, 172, 130.
References
[
1] C.N. Pace, G.R. Grimsley, J.A. Thomson, B.J. Barnett. Conformational
Stability and Activity of Ribonuclease T1 with Zero, One, and Two
Intact Disulfide Bonds, J. Biol. Chem. 1988, 263, 11820.
7594.
[
19] H.P. Gunawardena, R.A.J. O’Hair, S.A. McLuckey. Selective Disulfide
Bond Cleavage in Gold (II) Cationized Polypeptide Ions Formed via
Gas-Phase Ion/Ion Cation Switching, J. Proteome Res. 2006, 5, 2087.
20] H. Lioe, M. Duan, R.A.J. O’Hair. Can metal ions be used as gas-
phase disulfide bond cleavage reagents? A survey of coinage
metal complexes of model peptides containing and intermolecular
disulfide bond, Rapid Commun. Mass Spectrom. 2007, 21, 2727.
21] J.E. Lucas. Chemical Oxidation of Tryptic Digests to Improve Sequence
Coverage in Peptide Mass Fingerprint Protein Identification, Masters
Thesis, Texas A&M University (College Station, TX), 2003.
22] R. Matthiesen, G. Bauw, K.G. Welinder. Use of Performic Acid
Oxidation to Expand the Mass Distribution of Tryptic Peptides,
Anal. Chem. 2004, 76, 6848.
23] J.R. Brown, B.S. Hartley. Location of Disulphide Bridges by Diagonal
Paper Electrophoresis, Biochem. J. 1966, 101, 214.
[
[
2] J.M. Thornton. Disulphide Bridges in Globular Proteins, J. Mol. Biol.
1
981, 151, 261.
[
3] J.J. Gorman, T.P. Wallis, J.J. Pitt. Protein Disulfide Bond
Determination by Mass Spectrometry, Mass Spectrom. Rev. 2002,
21, 183.
[
[
4] X. Yu, Z. Wu, C. Fenselau. Covalent Sequestration of Melphalan by
[
[
Metallothionein and Selective Alkyation of Cysteines, Biochemistry
1
995, 34, 3377.
5] J.L. Apuy, Z.-Y. Park, P.D. Swartz, L.J. Dangott, D.H. Russell,
T.O. Baldwin. Pulsed-Alkylation Mass Spectrometry for the Study of
ProteinFoldingandDynamics:DevelopmentandApplicationtothe
Study of a Folding/Unfolding Intermediate of Bacterial Luciferase,
Biochemistry 2001, 40, 15153.
[
[
[
6] J.L. Apuy, X. Chen, D.H. Russell, T.O. Baldwin, D.P. Giedroc.
Radiometric Pulsed Alkylation/Mass Spectrometry of the Cysteine
Pairs in IndividualZinc Fingers of MRE-BindingTranscription Factor-
24] R.D. Edmondson, D.H. Russell. Evaluation of Matrix-Assisted
Laser Desorption Ionization – Time-of-Flight Mass Measurement
Accuracy by Using Delayed Extraction, J. Am. Soc. Mass Spectrom.
1
(MFT-1) as a Probe of Zinc Chelate Stability, Biochemistry 2001,
1
996, 7, 995.
40, 15164.
[
[
25] C.H.W. Hirs. Performic Acid Oxidation, Methods Enzymol. 1966, 197.
26] R.C. Beavis, J.N. Bridson, Epitaxial protein inclusion in sinapic acid
crystals, J. Phys. D: Appl. Phys. 1993, 26, 442.
27] B.J. Williams, W.K. Russell, D.H. Russell. Evaluation of Experimental
Variables for the On-Target Oxidation of Cysteine-Containing
Peptides and Proteins, In preparation. 2009, (Submitted).
[
[
[
7] M.F. Bean, S.A. Carr. Characterization of Disulfide Bond Position in
Proteins and Sequence Analysis of Cystine-Bridged Peptides by
Tandem Mass Spectrometry, Anal. Biochem. 1992, 201, 216.
8] S.D. Patterson, V. Katta. Prompt Fragmentation of Disulfide-Linked
Peptides during Matrix-Assisted Laser Desorption Ionization Mass
Spectrometry, Anal. Chem. 1994, 66, 3727.
9] R. Craig, O. Krokhin, J. Wilkins, R.C. Beavis. Implementation of an
Algorithm for Modeling Disulfide Bond Patterns Using Mass
Spectrometry, J. Proteome Res. 2003, 2, 657.
[
[
28] B.J. Williams, W.K. Russell, D.H. Russell. Utility of CE-MS Data in
Protein Identification, Anal. Chem. 2007, 79, 3850.
[
29] G. Rosas-Acosta, W.K. Russell, A. Deyrieux, D.H. Russell, V.G. Wilson.
A Universal Strategy for Proteomic Studies of SUMO and Other
Ubiquitin-like Modifiers, Mol. Cell. Proteomics 2005, 4, 56.
[
10] C. Caporale,L. Bertini,P. Pucci,V. Buonocore,C. Caruso.CysMapand
CysJoin: Database and tools for protein disulphides localisation,
FEBS Lett. 2005, 579, 3048.
www.interscience.wiley.com/journal/jms
Copyright ꢀc 2009 John Wiley & Sons, Ltd.
J. Mass. Spectrom. 2010, 45, 157–166