ANALYTICAL

BIOCHEMISTRY
Analytical Biochemistry 313 (2003) 216–225
www.elsevier.com/locate/yabio

Protein surface mapping by chemical oxidation: Structural

analysis by mass spectrometry
Joshua S. Sharp,a,b Jeffrey M. Becker,a and Robert L. Hetticha,b,*
a
Graduate School of Genome Science and Technology, The University of Tennessee-Oak Ridge National Laboratory,
1060 Commerce Park, Oak Ridge, TN 37830-8026, USA
b
Organic and Biological Mass Spectrometry Group, Chemical Sciences Division Oak Ridge National Laboratory,
P.O. Box 2008, Oak Ridge, TN 37831-6131, USA

Received 27 June 2002

Abstract

The solvent-accessible surface area of proteins is important in biological function for many reasons, including protein–protein
interactions, protein folding, and catalytic sites. Here we present a chemical technique to oxidize amino acid side chains in a model
protein, apomyoglobin, and subsequent elucidation of the effect of solvent accessibility on the sites of oxidation. Under conditions
of low protein oxidation (zero to three oxygen atoms added per apomyoglobin molecule), we have positively identified five oxidation
sites by liquid chromatography–tandem mass spectrometry and high-resolution Fourier transform mass spectrometry. Our results
indicate that all oxidized amino acids, with the exception of methionine, have highly solvent-accessible side chains, but the rate of
oxidation may not be dictated solely by solvent accessibility and amino acid identity.
Ó 2003 Elsevier Science (USA). All rights reserved.

Keywords: Mass spectrometry; Electrospray; Protein derivatization; Oxidation; Fenton chemistry; Collisional dissociation

Mass spectrometry is a rapidly emerging tool for
protein characterization. The high-sensitivity and high-
throughput capabilities of mass spectrometry make it an
excellent technique for fast protein analyses of many
types, including structural analyses. Several methods
have been developed to probe higher-order protein
structure by mass spectrometry. Solution-phase hydro-
gen–deuterium exchange is a commonly used technique
to study solvent-accessible surfaces and hydrogen
bonding of proteins (e.g., [1,2]). Hydrogen–deuterium
exchange is a powerful technique that can probe the
backbone of any protein residue. Analysis of hydrogen–
deuterium exchange data is often very complicated,
though, with the effects of hydrogen bonding, solvent
accessibility, and complex back-exchange kinetics all
contributing to the observed exchange rates (e.g., [3]).
Ion mobility has also been used as a rapid method to
study protein surface areas in the gas phase (e.g., [4]).
However, previous research suggests that, while some
*
Corresponding author. Fax: +865-576-8559.
E-mail address: hettichrl@ornl.gov (R.L. Hettich).
facets of protein structure are retained in the gas phase,
the overall structure of the protein has little or no re-
lationship to solution-phase structure [5,6].
Protein chemical modification is widely used for
studying protein structure. The recent explosion in the
field of biological mass spectrometry has led to the de-
velopment of many chemical modification techniques.
One method that has been demonstrated is the use of
specific chemical modification reagents to examine cer-
tain amino acids with specific reactive side chains [7–9].
The major drawback to this method is that, to probe a
wide variety of sites, a panel of reagents and reactions
must be used, increasing the time for analysis of a pro-
tein. More nonspecific chemical modification methods
are preferable due to their ability to probe different
amino acid side chains simultaneously. One nonspecific
method demonstrated to effectively probe protein sur-
faces utilizes methyl bromide [10]. Unfortunately, me-
thyl bromide is a highly toxic and carcinogenic gas, and
great care must be taken when working with the reagent.
The use of a synchrotron X-ray beamline has been
proposed to generate hydroxyl radicals very rapidly in

0003-2697/03/$ - see front matter Ó 2003 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0003-2697(02)00612-7
 J.S. Sharp et al. / Analytical Biochemistry 313 (2003) 216–225
solution. The hydroxyl radicals oxidize certain amino
acid side chains, and then mass spectrometry is used to
determine the sites of oxidation [11–15]. Good correla-
tion was demonstrated in most cases between oxidation
and solvent accessibility, and the value of the technique
for probing protein folding and unfolding has been
demonstrated [11,15]. Unfortunately, there are some
practical difficulties with a synchrotron X-ray approach.
Standard low-flux or pulse X-ray sources are not ideal
for these studies due to potential damage to the analyte
protein [14], and access to a synchrotron beamline is not
readily available to all researchers. Also, while hydroxyl
radicals do probe a variety of amino acid side chains, the
range of amino acids that are readily oxidized by syn-
chrotron radiolysis is limited. Previous studies of oxi-
dation of small peptides found that the order of amino
acid reactivity is cysteine, methionine
phenylalanine,
tyrosine, tryptophan > proline > histidine, leucine [12].
The same set of reactive residues was identified by use of
another hydroxyl radical-based technique using high-
electrospray needle voltages with oxygen as the nebu-
lizer gas [16]. However, this technique invokes questions
about the structure of the protein in the aerosol spray
and creates difficulties in subsequent tryptic digestion.
An alternate approach to generate hydroxyl radicals
for protein solvent-accessible surface studies is the use of
chelated iron to catalyze the formation of hydroxyl
radicals from hydrogen peroxide. Previous use of this
method was limited to utilizing the hydroxyl radicals to
cleave the protein backbone, followed by determination
of the sites of cleavage by gel electrophoresis [17]. Gel
electrophoresis is a low-resolution, low-accuracy mass
measurement technique, limiting the quality of infor-
mation gained from solvent-accessible surface analyses
by this method. Adding to the inherent inaccuracy of gel
electrophoresis is the fact that amino acid side chain
oxidation reactions occur on a shorter time scale than
backbone reactions, resulting in the addition of an un-
known number of oxygens to the protein [18]. Conse-
quently, the fragments resulting from oxidative cleavage
of the protein backbone have a range of oxygens at-
tached to the amino acid side chains, altering the frag-
mentsÕ mass and electrophoretic mobility by an
unknown amount and reducing the accuracy of oxida-
tive cleavage site determinations.
We report an experimental protocol to use chemically
generated hydroxyl radicals to probe the solvent-acces-
sible surface areas of apomyoglobin by determination of
the sites of side chain oxidation by mass spectrometry.
This method has the benefits of requiring only com-
monly available chemicals and providing fast reaction
times. Apomyoglobin was used because its secondary
and tertiary structures have been extensively character-
ized by NMR [19–22], it is commercially available at
high purity and is easily digested, and it has no pros-
thetic groups to complicate chemistry or mass spectro-
217
metric analysis. We propose the results from this proof-
of-principle experiment as evidence that selective amino
acid side chain oxidation by chemically generated hy-
droxyl radicals can be used for the rapid analysis of
protein solvent-accessible surface areas.
Materials and methods
Chemical oxidation of protein solutions
Two stock solutions were prepared for the chemical
oxidation reactions. Stock solution 1 consisted of
20 mM sodium ascorbate and 50 mM sodium phosphate
at pH 6.5. Stock solution 2 consisted of 6 lM
NH4 FeðSO4 Þ2 , 13 lM EDTA, and 50 mM sodium
phosphate at pH 6.5. All of these reagents were pur-
chased from Fisher Scientific (Pittsburgh, PA) and used
as supplied without further purification. The stock so-
lutions and the hydrogen peroxide reagent (3% H2 O2 ;
commercially available at local drugstore) were used
according to a protocol adapted from Heyduk et al. [18].
Lyophilized apomyoglobin (Sigma–Aldrich, St. Louis,
MO) was reconstituted at 200 lM concentration in
50 mM sodium phosphate. For reactions, ascorbate and
the iron–EDTA solutions were added in a 1:10 dilution
(stock volume:total volume) to the buffered apomyo-
globin. Finally, the H2 O2 was added to a 0.3% final
concentration to initiate the oxidation. To quench the
reaction, an equal volume of 2 M Tris (pH 5) was added.
Protein oxidation determination by electrospray Fourier
transform ion cyclotron resonance mass spectrometry
(ES-FT-ICR-MS)1
To analyze the extent of apomyoglobin oxidation, the
oxidized apomyoglobin was extracted using a C4 ZipTip
(Millipore, Bedford, MA) according to manufacturerÕs
protocol. The protein was eluted into a total of 60–80 lL
of acetonitrile with 0.1% trifluoroacetic acid (TFA). All
solvents were purchased from Sigma–Aldrich at the
highest purity available and used as supplied without
further purification. Deionized water ð18 MXÞ was pre-
pared in-house with a Millipore Mill-Q water purifica-
tion system (Millipore).
All ES-FT-ICR mass spectra were acquired with an
IonSpec (Irvine, CA) 9.4-Tesla HiRes electrospray Fou-
rier transform ion cyclotron resonance mass spectrome-
ter, as described previously [23]. Ions were generated with
an Analytica electrospray source (1–2 lL=minÞ, accu-
mulated in an external hexapole, transferred into the
1
Abbreviations used: ES-FT-ICR-MS, electrospray Fourier trans-
form ion cyclotron resonance mass spectrometry; SORI-CAD, sus-
tained off-resonance irradiation collision-activated dissociation; TFA,
trifluoroacetic acid.
218 J.S. Sharp et al. / Analytical Biochemistry 313 (2003) 216–225
high-vacuum region with a quadrupole lens system, and
then detected in the cylindrical analyzer cell of the mass
spectrometer. Because ion detection was achieved in an
ultralow-vacuum regime (2  1010 TorrÞ, a broadband
mass resolution of about 150,000 (full width of peak at
half-maximum) at m=z 1000 was possible. Mass scale
calibration was accomplished with ubiquitin. The high-
resolution mass measurement enables isotopic resolution
of multiply charged ions. Thus, the charge state of a
multiply charged ion can be determined solely by its
isotopic spacing [24].
Deconvoluted molecular mass spectra were generated
with the IonSpec software from the electrospray mass
spectra by multiplying the masses of the electrospray
ions by their respective charges and then subtracting the
masses of the protons added. This ‘‘unfolds’’ the mul-
tiply charged ion mass spectrum into a more easily in-
terpreted molecular mass spectrum. Errors in mass
measurement for the multiply charged ions will be scaled
proportional to the charge in the calculation of the
molecular masses in the deconvoluted mass spectra. By
calibrating on the calculated values of the most abun-
dant isotopic peaks of six different charge states (7þ to
12þ ) of bovine ubiquitin, the deconvoluted molecular
mass spectrum yielded a measured molecular mass for
ubiquitin that was within 0.030 Da of the calculated
value. This calibration procedure generally provides a
molecular mass measurement accuracy of 63 ppm for
most proteins up to at least 20 kDa. Measurements of
lower-mass peptides, such as those obtained from a
proteolytic digestion, usually yield molecular mass ac-
curacy in the few millimass range (61 ppm).
Ion collisional dissociation was conducted by isolat-
ing an ion of interest (either a peptide or a protein)
within the analyzer cell of the mass spectrometer and
then accelerating the ion into an argon target gas under
sustained off-resonance irradiation collision-activated
dissociation (SORI-CAD) [25]. For the SORI-CAD
experiments, the ion excitation was accomplished with
an rf pulse (1 kHz lower in frequency than the ionÕs
cyclotron frequency) applied for 2 s in duration and with
an amplitude in the range of 1–4 V (peak-to-peak). A
pulsed valve was used to admit the argon collision gas
into the high-vacuum region to a maximum pressure of
about 5  106 Torr during the ion excitation step. A
base pressure of about 1  109 Torr was reestablished
prior to ion detection.
Trypsin digestion of oxidized apomyoglobin
For tryptic digest analyses, the oxidized apomyoglo-
bin was digested with sequencing-grade modified trypsin
following the manufacturerÕs protocol (Promega, Mad-
ison, WI). The tryptic digest was then desalted using a
C18 Sep-Pak Lite (Waters, Milford, MA) according to
the manufacturerÕs protocol. The peptides were eluted
into a total of 1–1.5 mL of acetonitrile with 0.1% TFA.
The peptides were then ionized by electrospray and
analyzed by ES-FT-ICR-MS using SORI-CAD as de-
scribed above and LC-MS/MS as described below.
LC-MS/MS analyses of oxidized apomyoglobin tryptic
digests
For LC-MS/MS analyses, 200 lL of the Sep-Pak Lite
eluant was dried to completion in a centrifugal evapo-
rator (Savant Instruments, Holbrook, NY) and resus-
pended in 10 lL 0.1% TFA. All LC-MS/MS experiments
were performed on an Ultimate HPLC system (LC
Packings, San Francisco, CA) interfaced to an LCQ-
Deca ion trap mass spectrometer (Thermo Finnigan,
San Jose, CA) equipped with an electrospray source.
The HPLC was operated at a flow rate of 4 lL=min
using a plug-in to the Xcalibur software provided by LC
Packings. The resuspended tryptic digest was loaded on
a C18 (218MS5.315, 300 lm i.d.  15 cm, 300 A  with 3-
lm particles) reverse-phase column (Vydac, Hesperia,
CA). The analytical column was connected to the in-
jection port and to the electrospray source with 100-lm-
i.d. fused silica. Injections were made manually with a
10-lL low-dispersion loop. The sample was washed over
the column for 5 min in 95% water, 5% acetonitrile, 0.1%
formic acid. A 1-h linear gradient from 100% Buffer A
(95% water, 5% acetonitrile, 0.1% formic acid) to 100%
Buffer B (30% water, 70% acetonitrile, 0.1% formic acid)
was then run, followed by a 5-min wash of 100% Buffer
B. The LCQ was operated in the data-dependent mode
with dynamic exclusion activated. TurboSEQUEST was
then executed, each time with a +16 Da differential
modification on one of the 20 standard amino acids and
a +14 Da differential modification on proline as previ-
ously described in synchrotron radiolysis studies [12,16].
MS/MS spectra identified as arising from possible oxi-
dized peptides were analyzed manually.
Calculation of solvent-accessible areas of apomyoglobin
Although the secondary and tertiary structures of
apomyoglobin have been studied extensively by NMR, it
is difficult to use this information directly for side chain
solvent-accessibility calculations. Therefore, initial side
chain solvent-accessibility calculations were performed
using the X-ray crystal structure of holomyoglobin and
then adjusted qualitatively according to the NMR
structure of apomyoglobin. To determine whether oxi-
dized residues were solvent accessible, the program
GETAREA 1.1 [26] was utilized with the X-ray crystal
structure of horse heart myoglobin (1YMB) [27]. The
default parameters for radius of the water probe and de-
fault atomic radii and atomic solvation parameters were
used for all calculations. Solvent accessibilities for apo-
myoglobin were calculated on an area-per-residue basis.
 J.S. Sharp et al. / Analytical Biochemistry 313 (2003) 216–225 219

Contrary to the supposition that native apomyoglo-
bin is a molten globule, NMR results definitely reveal
that this protein is primarily a folded globular protein
that closely resembles the native holomyoglobin in ter-
tiary structure [19–22,28]. NMR is a high-resolution
method for determination of protein structure and in
many ways is thought to be more indicative of the
physiological protein structure than is X-ray crystal-
lography. The most notable deviation in the two forms
of myoglobin is a contiguous region (residues 82–101)
that is conformationally disordered in the absence of the
heme moiety. Detailed NMR measurements indicate
that, with the exception of the F helix, the helices are
fully formed even in the absence of the heme, although
there is some indication of fraying toward the C-termi-
nal end of the H helix. The {1 H}–15 N heteronuclear
NOEs in both the native apomyoglobin and the holo-
myoglobin are characteristic of a folded globular pro-
tein, with the only exception being limited backbone
flexibility at the termini and in the local regions of the
C-D and G-H loops [20].
Results
Chelated iron-catalyzed oxidation of apomyoglobin
Solutions of apomyoglobin were oxidized for varying
amounts of time using chelated iron and hydrogen
peroxide to generate solution-phase hydroxyl radicals,
which then react with the amino acid side chains to
oxidize the protein. Chelated iron does not bind
specifically to proteins, helping to ensure the nonspeci-
ficity of the oxidation reaction [17,18,29]. Analysis of the
oxidation products by ES-FT-ICR-MS showed that
oxidation of apomyoglobin occurs in a time-dependent
manner, and several amino acid side chain oxidation
events can be detected before significant backbone
cleavage occurs. Fig. 1 shows the deconvoluted mass
spectra of four time points in the presence of a 10-fold
dilution of catalyst to slow the rate of oxidation. The
zero time point was collected after performing the re-
action in the presence of 1 M Tris buffer and allowed to
sit on the bench top for 1 h, showing that the presence of
1 M Tris is sufficient to quench the oxidation reaction.
Note that there are no oxidation events at the zero time
point; the observed packet of ions is due to the naturally
occurring 13 C isotopic distribution. A mass shift of
15.995 Da per oxidation event is characteristic of oxi-
dation events. The increase in oxidation events per
molecule is clearly time dependent, with cleavage of the
protein backbone by the hydroxyl radicals occurring at
4.5 min (data not shown).
One concern with the development of this technique
is ensuring that the oxidation reactions probe the native
structure of the protein. Multiple oxidations could
conceivably alter the structure of the protein. Therefore,
we used direct-infusion ES-FT-ICR-MS to monitor the

Fig. 1. Deconvoluted mass spectra of a time course of apomyoglobin oxidation. The reactions were quenched at 90-s intervals and spectra obtained
by ES-FT-ICR-MS, showing that the number of oxygen atoms added to each molecule of apomyoglobin is time dependent. At 4.5 min, in addition to
the oxidation observed, significant fragmentation has begun to occur (data not shown).
220 J.S. Sharp et al. / Analytical Biochemistry 313 (2003) 216–225
oxidation status of the intact protein and chose a time
point that represented a small number of oxidation
events per protein molecule distributed across the reac-
tive, solvent-accessible side chains. Thus, we could help
ensure that the oxidation events themselves do not in-
validate the data gathered by altering the structure of
the protein with initial oxidation events and then oxi-
dizing a site that is buried in the native structure. No
evidence of fragmentation or cross-linking is observed in
this mass spectrum of oxidized apomyoglobin.
Analysis of oxidation sites by tandem mass spectrometry
To determine which amino acid side chains are sol-
vent accessible, we determined the sites of oxidation by
both direct-infusion ES-FT-ICR-MS with SORI-CAD
fragmentation and reverse-phase capillary LC-MS/MS
in data-dependent mode. The use of high-resolution ES-
FT-ICR-MS allows for high-accuracy mass measure-
ments to ensure the identity of oxidized peptides, while
LC-MS/MS is a robust, partially automated method for
interrogating the complex mixture to assign oxidation
sites. Full sequence coverage of nonoxidized peptides
was obtained by ES-FT-ICR-MS for all peptides that
did not contain one of the two methionines in the se-
quence. Oxidized peptides containing the two methio-
nines were detected, in addition to oxidized peptides
containing either Phe151 or Trp7 and Leu11. The high
resolution and mass accuracy of the ES-FT-ICR-MS
measurements ensured that the peptides were identified
and correctly assigned. Oxidized peptides containing all
sites identified by LC-MS/MS were also found by ES-
FT-ICR-MS. Listed in Table 1 are the residues that we
were able to identify as sites of oxidation by tandem
mass spectrometry. Oxidized peptides were identified by
their masses compared to an in silico digestion of apo-
myoglobin using PROWL [30]. The site of oxidation
was then determined by tandem mass spectrometry,
looking for a neutral mass loss 16 Da more than the
expected neutral mass loss as determined by PROWL.
As the peptide is fragmented in the mass spectrometer,
cleavage at the amide bond is the most prevalent mode
of fragmentation. Fragments which are charged on the
Table 1
Sites of apomyoglobin oxidation
Measured Peptide
mass (Da)
1830.889 1–16
1830.889 1–16
1494.716 48–56
1517.660 119–133
956.454 146–153
Oxidized peptides detected by ES-FT-ICR-MS and the sites of oxidation on each determined by LC-MS/MS. The calculated masses were
determined using PROWL [30]. The measured and calculated masses all represent the deprotonated peptide. The side chain solvent-accessible areas
were calculated using GETAREA 1.1 [26] using the X-ray structure of horse heart myoglobin (1YMB) [27].
N-terminal portion of the peptide are termed b-type ions
and are numbered from the N terminus of the parent
peptide. Fragments which are charged on the C-terminal
portion of the peptide are termed y-type ions and are
numbered from the C terminus of the parent peptide
[31]. As the oxidation occurs by free radical chemistry,
which is notorious for producing various side products,
high-resolution and high-accuracy analyses increase the
confidence with which we can assign oxidation sites. ES-
FT-ICR-MS allows a high-resolution confirmation by
exact mass to the data generated by the ion trap. Also,
the data generated in the ion trap are often of moderate
quality due to the low signal to noise ratio when ex-
amining oxidized peptide CAD spectra. The data gen-
erated by ES-FT-ICR-MS allow us to confirm the fact
that the identified peptide is indeed oxidized.
Fig. 2 shows the MS/MS spectrum that identifies
Phe151 as an oxidized residue. The presence of an
abundant nonoxidized b5 ion with no (b5 + O) ion
present shows that the peptide is oxidized at or C-ter-
minal to Phe151. The presence of an abundant (b6 + O)
ion with no nonoxidized b6 ion detected shows that
oxidation of this peptide occurred exclusively at Phe151.
The additional presence of b4 and (b7 + O) further
confirms this assignment. As calculated from the crystal2
structure of horse heart myoglobin, Phe151 has 20:75 A 
side chain solvent-accessible area.
Table 2 summarizes the MS/MS spectra that reveal
the sites of oxidation. In the case of peptide 119–133, the
CAD spectrum allowed assignment of Met131 as the
sole oxidation site. The presence of the (y3 + O) ion with
the nonoxidized y2 ion shows that the oxygen is present
on Met131. The lack of nonoxidized y3 ion and the lack
of a (y2 + O) ion show that the oxidation is exclusively
on Met131. In the X-ray crystal structure of horse
myoglobin, Met131 is completely buried in the interior
of the protein and has 0:00 A 2 of solvent-accessible side
chain surface area. Although methionine oxidation
might not be expected in this case, this particular amino
acid can be oxidized by other processes, as will be dis-
cussed below.
Surprisingly, the fragments from CAD of peptide 1–
16 allow for the assignment of Trp7 and Leu11 as two
Calculated oxidized Site of oxidation by Surface area by
mass (Da) LC-MS/MS X-ray model ðA2 Þ
1830.889 Trp7 13.42
1830.889 Leu11 40.44
1494.723 Met55 5.74
1517.656 Met131 0.00
956.460 Phe151 20.75
 J.S. Sharp et al. / Analytical Biochemistry 313 (2003) 216–225 221

Fig. 2. A representative ion trap MS/MS spectrum used to determine sites of oxidation. The presence of an abundant nonoxidized b5 ion with no
(b5 + O) ion present shows that the peptide is oxidized at or C-terminal to Phe151. The presence of an abundant (b6 + O) ion with no nonoxidized b6
ion detected shows that oxidation of this peptide occurred exclusively at Phe151. All oxidation site assignments were performed in a similar manner.

Table 2
MS/MS fragmentation of oxidized peptides
Peptide Detected b ions Detected y ions Oxidized residues
1-GLSDGEWQQVLNVWGK-16 b9, b9 + O, b10, b10 + O, y5, y6, y6 + O, y7, y7 + O, y9, y9 + O, W7, L11
b11 + O, b14 + O, b15 + O y10 + O, y12 + O, y13 + O, y14 + O
51-TEAEMKASEDLKK-63 b2 y8, y9 + O, y11 + O M55

119-HPGDFGADAQGAMTK-133 b7, b8, b9 y2, y3 + O, y5 + O, y6 + O M131

146-YKELGFQG-153 b4, b5, b6 + O, b7 + O y5 + O, y6 + O F151
MS/MS fragments from oxidized peptides. Four oxidized peptides were found by LC/MS/MS. The MS/MS spectra were analyzed manually and
assignments made for various fragment ions. The fragments assigned are listed along with the assignments of the oxidation sites made from the
fragmentation patterns.

distinct oxidation sites in a single peptide with only one
oxidation event per peptide. The presence of both oxi-
dized and nonoxidized fragment ions for many b and y
ions shows that there is more than one oxidation site on
the peptide. In the y ion series, y14 through y10 are
found only as oxidized fragments, while y9 and (y9 + O)
are present, showing that the N-terminal oxidation site
is at Trp7. The y6 and (y6 + O) ions are also both
present, while only the nonoxidized y5 ion is present,
showing that the C-terminal oxidation site is Leu11.
According to the crystal structure of horse myoglobin,
the side chain of Trp7 has 13:42 A 2 of solvent-exposed
surface area, while the side chain of Leu11 has 40:44 A2 .
The fragments from the CAD of peptide 51–63 were
used to determine the oxidation of Met55 of apomyo-
globin. The presence of an abundant ðy11 þ OÞþ2 ion
and a ðy9 þ OÞþ2 ion with no unoxidized ðy9Þþ2 present
shows that all oxidation events detected at this peptide
must occur at or C-terminal to Met55. However, the
presence of a ðy8Þþ2 ion with no ðy8 þ OÞþ2 ion detected
shows that the sole site of oxidation on this peptide oc-
curs at Met55. As shown in Table 1, according to the X-
ray structure of myoglobin, Met55 has 5:74 A 2 side chain
solvent-accessible surface area, which is relatively small.
A calculation based on the X-ray crystal structure of
holomyoglobin of the solvent-accessible side chain areas
222 J.S. Sharp et al. / Analytical Biochemistry 313 (2003) 216–225

of all residues in apomyoglobin that are chemically
susceptible to modification by hydroxyl radicals is pre-
sented in Fig. 3. The residues that we found to be oxi-
dized are not the most solvent accessible according to
the structure of holomyoglobin. A partial NMR struc-
ture of sperm whale apomyoglobin, however, shows that
three regions of apomyoglobin are disordered [22],
causing the residues in these disordered regions to be
extremely solvent accessible. As shown in Fig. 3, Phe151
and Met55 are directly in disordered regions, causing
their side chain solvent accessibilities to be much higher
than the calculations based on the X-ray crystal struc-
ture of holomyoglobin suggested.
Trp7 and Leu11 are not in a disordered region, and
yet were found to be oxidized by our method. As shown
in Fig. 4, Trp7 and Leu11 are protected from solvent in
holomyoglobin by a region that is disordered in apo-
myoglobin. Thus, in apomyoglobin, Trp7 and Leu11
have a much greater solvent-accessible side chain
area than that calculated from the X-ray structure of

Fig. 3. Side chain solvent accessibilities of all amino acid residues in apomyoglobin that are very highly reactive (cysteine or methionine), highly
reactive (phenylalanine, tyrosine, or tryptophan), reactive (proline), or poorly reactive (histidine or leucine) [12]. Residues determined to be oxidized
by tandem mass spectra are labeled with black triangles. The vertical axis represents the side chain solvent-accessible area in A2 as calculated from the
crystal structure of holomyoglobin, PDB identifier 1YMB [27]. Plotted across the top of the graph are dotted bars representing disordered regions as
determined from the partial NMR structure of sperm whale apomyoglobin [22].

Fig. 4. The X-ray crystal structure of holomyoglobin, PDB identifier 1YMB [27]. Shown in green is the side chain Trp7; shown in red is the side chain
Leu11. The blue areas are all regions determined by the partial NMR structure of sperm whale apomyoglobin to be disordered [22].
 J.S. Sharp et al. / Analytical Biochemistry 313 (2003) 216–225
holomyoglobin. Therefore, the oxidation sites identified
by our method are shown by established structural
methods to be highly solvent accessible in the structure
of apomyoglobin.
Discussion
Computational methods for the determination of
protein structure are rapidly increasing in speed and
accuracy; however, their accuracy is still not fully reli-
able, especially when analyzing proteins with no known
homologs with solved structures [32]. One method for
improving the accuracy of computational models is
through the use of experimental constraints to the pro-
tein structure. Partial NMR datasets have been applied
to improve the quality of computational predictions
[33]. Another such constraint that can be used is solvent
accessibility information of certain amino acids, which
help to define the solvent-accessible surface areas of the
folded protein [34]. Therefore, a technique which can
provide experimental constraints in a high-throughput
manner would be very valuable to the realization of an
accurate, high-throughput structural proteomics proto-
col based on computational structure prediction.
With the exception of methionines, all of the residues
oxidized and identified in this study are highly solvent
accessible according to the NMR structure of apomyo-
globin. The data presented above (most notably the
complete lack of false positives) provide evidence that
the use of chemically generated hydroxyl radicals as
probes to determine protein surface residues is feasible,
with the exception of methionine oxidation. The short
reaction time (30 s) along with relatively short digestion
and analysis times suggest that this method would be
useful as a rapid analysis of certain aspects of protein
structure. The fact that Met131 has essentially no sol-
vent-accessible area, even considering the disordered
regions in apomyoglobin, and yet is still oxidized sug-
gests that there is a different mechanism for the oxida-
tion of methionine that is not dictated by solvent
accessibility. These data are supported by observations
in studies with lysozyme using synchrotron radiation
that also found buried methionines to be oxidized [14].
Kiselar et al. [13] postulated that methionines are oxi-
dized by an intramolecular radical transfer mechanism,
in which a hydroxyl radical would first abstract a hy-
drogen from a solvent-accessible reactive residue. Then,
another side chain proximate to the radical site would
donate an electron to the solvent-accessible radical. That
solvent-accessible residue would then pick up a proton,
transferring the radical to the electron-donating group.
Molecular oxygen would then oxidize the radical, which
would not necessarily be solvent accessible [13]. Met131,
the only buried methionine in the sequence, is in close
proximity to both Leu11 and Trp7, which were both
223
shown to be oxidized. Therefore, we propose that a
similar radical transfer mechanism is responsible, at
least in part, for the oxidation of Met131. In addition,
peptides with oxidized methionines can be detected in
ES-FT-ICR-MS spectra of tryptic digests of apomyo-
globin that have undergone no oxidation reactions (data
not shown), suggesting that methionine oxidation also
occurs, at least in part, during or after digestion.
According to the X-ray structure of horse myoglobin,
Leu11 has 40:44 A 2 of solvent-accessible side chain
surface area. The previous observation that the main
region in holomyoglobin that partially protects Trp7
and Leu11 from solvent is disordered in apomyoglobin
shows that the actual solvent-accessible regions of the
Trp7 and Leu11 side chains are greater than those cal-
culated from holomyoglobin. Even so, Phe43 is within a
disordered region and has a higher chemical reactivity
than leucine. Phe41 should also have a higher solvent
accessibility than Leu11, as the group is actually within
a disordered region; however, no oxidation of Phe43 is
observed. The fact that an amino acid residue with a
higher chemical reactivity within a disordered region is
not oxidized while Leu11 is oxidized suggests either that
Phe43 is not fully exposed to solvent even within a dis-
ordered region or that a factor other than solvent ac-
cessibility and amino acid identity play a role in the rate
of oxidation.
While only limited oxidation was performed in this
initial experiment, the failure to observe the oxidation of
certain residues which one would expect to be oxidized
by no means invalidates the approach presented in this
report. All residues (except methionine) that were found
to be oxidized are highly solvent accessible in the native
solution-phase structure of apomyoglobin as determined
by NMR [22]. The data presented in this study suggest
that the technique presented here may have applications
in a variety of structural analyses, including protein
folding and protein–protein interaction studies. X-ray
synchrotron studies, which use the same hydroxyl radi-
cal probe, have been utilized to examine protein un-
folding [15] and protein structure [14]. A chemical means
for generating the hydroxyl radical probe should allow
for more widespread use of this rapid structural tech-
nique. Although this method has many advantages,
there still are some uncharacterized issues at this point.
For example, it is conceivable that the high concentra-
tions of iron salt, EDTA, and ascorbate could induce
conformational changes in the analyte protein. Previous
studies using Fe–EDTA-catalyzed reactions to cleave
the protein backbone and determination of cleavage
sites by gel electrophoresis have thus far agreed with
experimental data gathered by other methods [17,29].
Therefore, existing empirical evidence suggests that the
chemical oxidation method presented here accurately
probes the surface of proteins. It is also obvious that this
chemical oxidation reaction, which is identical to that
224 J.S. Sharp et al. / Analytical Biochemistry 313 (2003) 216–225
employed for the X-ray synchrotron hydrolysis experi-
ments, probes only a limited range of amino acids and
may not be successful for extensively detailed protein
surface mapping. Also, because of the nature of the
hydroxyl radical-generating reaction, this method is not
applicable to metal-containing proteins. The data gen-
erated by mass spectrometry can establish unambigu-
ously the amino acid side chain that is oxidized; to
obtain a more complete description of the surface of a
protein, however, it may be necessary to increase the
numbers of oxygen atoms attached to each protein
molecule. Additional studies need to be conducted to
determine the amount of oxidation events that can be
tolerated by a protein before the protein begins to lose
its native structure. Further testing of other model
proteins is underway in our laboratory and is designed
not only to examine the factors listed above but also to
evaluate the generic utility and kinetics of this oxidation
method to probe solvent-accessible areas.
Acknowledgments
Research was conducted under the Center for
Structural and Molecular Biology Research Program at
Oak Ridge National Laboratory (ORNL) and was
sponsored by the U.S. Department of Energy, Office of
Biological and Environmental Research. The authors
are grateful to Nathan VerBerkmoes, Keiji Asano, and
Karin Keller for technical assistance with LC-MS/MS
experiments. J.S.S. acknowledges financial support from
the Graduate School of Genome Science and Technol-
ogy, University of Tennessee-Oak Ridge National
Laboratory. ORNL is operated and managed by the
University of Tennessee-Battelle, LLC, for the U.S.
Department of Energy under Contract DE-AC05-
00OR22725.
References
[1] N. Yamada, E. Suzuki, K. Hirayama, Identification of the
interface of a large protein–protein complex using H/D exchange
and Fourier transform ion cyclotron resonance mass spectrome-
try, Rapid Commun. Mass Spectrom. 16 (2002) 293–299.
[2] D.L. Smith, Y. Deng, Z. Zhang, Probing the non-covalent
structure of proteins by amide hydrogen exchange and mass
spectrometry, J. Mass Spectrom. 32 (1997) 135–146.
[3] F. Wang, X. Tang, Conformational heterogeneity of stability of
apomyoglobin studied by hydrogen/deuterium exchange and
electrospray ionization mass spectrometry, Biochemistry 35
(1996) 4069–4078.
[4] E.R. Badman, C.S. Hoaglund-Hyzer, D.E. Clemmer, Monitoring
structural changes of proteins in an ion trap over approximately
10–200 ms: unfolding transitions in cytochrome c ions, Anal.
Chem. 73 (2001) 6000–6007.
[5] D.M. Horn, K. Breuker, A.J. Frank, F.W. McLafferty, Kinetic
intermediates in the folding of gaseous protein ions characterized
by electron capture dissociation mass spectrometry, J. Am. Chem.
Soc. 123 (2001) 9792–9799.
[6] M. Gustafsson, W.J. Griffiths, E. Furusjo, J. Johansson, The
palmitoyl groups of lung surfactant protein C reduce unfolding
into a fibrillogenic intermediate, J. Mol. Biol. 310 (2001) 937–950.
[7] F. Zappacosta, P. Ingallinella, A. Scaloni, A. Pessi, E. Bianchi, M.
Sollazzo, A. Tramontano, G. Marino, P. Pucci, Surface topology
of minibody by selective chemical modifications and mass
spectrometry, Protein Sci. 6 (1997) 1901–1909.
[8] D. Suckau, M. Mak, M. Przybylski, Protein surface topology-
probing by selective chemical modification and mass spectrometric
peptide mapping, Proc. Natl. Acad. Sci. USA 89 (1992) 5630–
5634.
[9] C.E. Hopkins, P.B. OÕConnor, K.N. Allen, C.E. Costello, D.R.
Tolan, Chemical-modification rescue assessed by mass spectrom-
etry demonstrates that gamma-thia-lysine yields the same activity
as lysine in aldolase, Protein Sci. 11 (2002) 1591–1599.
[10] A. Scaloni, P. Ferranti, G. De Simone, G. Mamone, N. Sannolo,
A. Malorni, Probing the reactivity of nucleophile residues in
human 2,3-diphosphoglycerate/deoxy-hemoglobin complex by
aspecific chemical modifications, FEBS Lett. 452 (1999) 190–194.
[11] S.D. Maleknia, C.Y. Ralston, M.D. Brenowitz, K.M. Downard,
M.R. Chance, Determination of macromolecular folding and
structure by synchrotron X-ray radiolysis techniques, Anal.
Biochem. 289 (2001) 103–115.
[12] S.D. Maleknia, M. Brenowitz, M.R. Chance, Millisecond radio-
lytic modification of peptides by synchrotron X-rays identified by
mass spectrometry, Anal. Chem. 71 (1999) 3965–3973.
[13] J.G. Kiselar, S.D. Maleknia, M. Sullivan, K.M. Downard, M.R.
Chance, Hydroxyl radical probe of protein surfaces using
synchrotron X-ray radiolysis and mass spectrometry, Int. J.
Radiat. Biol. 78 (2002) 101–114.
[14] S.D. Maleknia, J.G. Kiselar, K.M. Downard, Hydroxyl radical
probe of the surface of lysozyme by synchrotron radiolysis and
mass spectrometry, Rapid Commun. Mass Spectrom. 16 (2002)
53–61.
[15] S.D. Maleknia, K.M. Downard, Unfolding of apomyoglobin
helices by synchrotron radiolysis and mass spectrometry, Eur. J.
Biochem. 268 (2001) 5578–5588.
[16] S.D. Maleknia, M.R. Chance, K.M. Downard, Electrospray-
assisted modification of proteins: a radical probe of protein
structure, Rapid Commun. Mass Spectrom. 13 (1999) 2352–2358.
[17] E. Heyduk, T. Heyduk, Mapping protein domains involved in
macromolecular interactions: a novel protein footprinting ap-
proach, Biochemistry 33 (1994) 9643–9650.
[18] T. Heyduk, N. Baichoo, E. Heyduk, Hydroxyl radical footprint-
ing of proteins using metal ion complexes, Met. Ions Biol. Syst. 38
(2001) 255–287.
[19] D. Eliezer, P.E. Wright, Is apomyoglobin a molten globule?
Structural characterization by NMR, J. Mol. Biol. 263 (1996)
531–538.
[20] D. Eliezer, J. Yao, H.J. Dyson, P.E. Wright, Structural and
dynamic characterization of partially folded states of apomyoglo-
bin and implications for protein folding, Nat. Struct. Biol. 5
(1998) 148–155.
[21] J.T. Lecomte, Y.H. Kao, M.J. Cocco, The native state of
apomyoglobin described by proton NMR spectroscopy: the A–
B–G–H interface of wild-type sperm whale apomyoglobin, Pro-
teins 25 (1996) 267–285.
[22] J.T. Lecomte, S.F. Sukits, S. Bhattacharya, C.J. Falzone,
Conformational properties of native sperm whale apomyoglobin
in solution, Protein Sci. 8 (1999) 1484–1491.
[23] T. Uchiki, R. Hettich, V. Gupta, C. Dealwis, Characterization of
monomeric and dimeric forms of recombinant Sml1p-histag
protein by electrospray mass spectrometry, Anal. Biochem. 301
(2002) 35–48.
[24] D.M.Z. Horn, F.W. McLafferty, Automated de novo sequencing
of proteins by tandem high-resolution mass spectrometry, Proc.
Natl. Acad. Sci. USA 97 (2000) 10313–10317.
 J.S. Sharp et al. / Analytical Biochemistry 313 (2003) 216–225
[25] J.W. Gauthier, T.R. Trautman, D.B. Jacobson, Sustained off-
resonance irradiation for CAD involving FTMS. CAD technique
that emulates infrared multiphoton dissociation, Anal. Chim.
Acta 246 (1991) 211–225.
[26] R. Fraczkiewicz, W. Braun, Exact and efficient analytical
calculation of the accessible surface areas and their gradients
for macromolecules, J. Computational Chem. 19 (1998) 319–
333.
[27] S.V. Evans, G.D. Brayer, High-resolution study of the three-
dimensional structure of horse heart metmyoglobin, J. Mol. Biol.
213 (1990) 885–897.
[28] A. Fontana, M. Zambonin, P. Polverino de Laureto, V. De
Filippis, A. Clementi, E. Scaramella, Probing the conformational
state of apomyoglobin by limited proteolysis, J. Mol. Biol. 266
(1997) 223–230.
[29] N. Baichoo, T. Heyduk, Mapping conformational changes in a
protein: application of a protein footprinting technique to cAMP-
225
induced conformational changes in cAMP receptor protein,
Biochemistry 36 (1997) 10830–10836.
[30] D. Fenyo, W. Zhang, B.T. Chait, R.C. Beavis, Internet-based
analytical chemistry resources: a model project, Anal. Chem. 68
(1996) 721A–726A.
[31] P. Roepstorff, J. Fohlman, Proposal for a common nomenclature
for sequence ions in mass spectra of peptides, Biomed. Mass
Spectrom. 11 (1984) 601.
[32] Y. Xu, D. Xu, O.H. Crawford, Einstein, F. Larimer, E.
Uberbacher, M.A. Unseren, G. Zhang, Protein threading by
PROSPECT: a prediction experiment in CASP3, Protein Eng. 12
(1999) 899–907.
[33] Y. Xu, D. Xu, O.H. Crawford, J.R. Einstein, A computational
method for NMR-constrained protein threading, J. Computa-
tional Biol. 7 (2000) 449–467.
[34] Y. Xu, D. Xu, Protein threading using PROSPECT: design and
evaluation, Proteins 40 (2000) 343–354.