Exploring radical migration pathways in peptides with positional isomers, deuterium labeling, and molecular dynamics simulations

Article abstract of DOI:10.1007/s13361-012-0540-6

One of the keys for understanding radical directed dissociation in peptides is a detailed knowledge of the factors that mediate radical migration. Peptide radicals can be created by a variety of means; however, in most circumstances, the originally created radicals must migrate to alternate locations in order to facilitate fragmentation such as backbone cleavage or side chain loss. The kinetics of radical migration are examined herein by comparing results from ortho-, meta-, and para-benzoyl radical positional isomers for several peptides. Isomers of a constrained cyclic peptide generated by several orthogonal radical initiators are also probed as a function of charge state. Cumulatively, the results suggest that small changes in radical position can significantly impact radical migration, and overall structural flexibility of the peptide is also an important controlling factor. A particularly interesting pathway for the peptide RGYALG that is sensitive to ortho versus meta or para substitution was fully mapped out by a suite of deuterium labeled peptides. This data was then used to optimize parameters in molecular dynamics-based simulations, which were subsequently used to obtain further insight into the structural underpinnings that most strongly influence the kinetics of radical migration. [Figure not available: see fulltext.]

Full text of DOI:10.1007/s13361-012-0540-6

B American Society for Mass Spectrometry, 2013
J. Am. Soc. Mass Spectrom. (2013) 24:524Y533
DOI: 10.1007/s13361-012-0540-6
FOCUS: DISTONIC IONS: RESEARCH ARTICLE
Exploring Radical Migration Pathways in Peptides
with Positional Isomers, Deuterium Labeling,
and Molecular Dynamics Simulations
Xing Zhang, Ryan R. Julian
Department of Chemistry, University of California, Riverside, CA 92521, USA
Abstract. One of the keys for understanding radical directed dissociation in
peptides is a detailed knowledge of the factors that mediate radical migration.
Peptide radicals can be created by a variety of means; however, in most
circumstances, the originally created radicals must migrate to alternate locations
in order to facilitate fragmentation such as backbone cleavage or side chain loss.
The kinetics of radical migration are examined herein by comparing results from
ortho-, meta-, and para-benzoyl radical positional isomers for several peptides.
Isomers of a constrained cyclic peptide generated by several orthogonal radical
initiators are also probed as a function of charge state. Cumulatively, the results
suggest that small changes in radical position can significantly impact radical
migration, and overall structural flexibility of the peptide is also an important controlling factor. A particularly
interesting pathway for the peptide RGYALG that is sensitive to ortho versus meta or para substitution was
fully mapped out by a suite of deuterium labeled peptides. This data was then used to optimize parameters in
molecular dynamics-based simulations, which were subsequently used to obtain further insight into the
structural underpinnings that most strongly influence the kinetics of radical migration.
Key words: Kinetics, ECD, Photodissociation, Iodine
Received: 23 October 2012/Revised: 8 November 2012/Accepted: 9 November 2012/Published online: 30 January 2013
abundant radicals, so called because the number of hydrogen
atoms exceeds the number required for an even electron
molecule. Most free radicals (including all discussed in the
Introduction
ree radicals play various important roles in biology in
F
both useful and destructive contexts [1]. For example,
previous paragraph) are hydrogen deficient species, where
essentially a hydrogen atom is missing relative to an even
electron molecule. Collision induced dissociation (CID) based
methods have been recently developed to generate hydrogen
deficient radical peptides in the gas phase through various
modification strategies, including oxidative dissociation of
peptide-metal complexes [3] homolytic cleavage of labile
bonds such as nitrate ester [4], peroxycarbamate [5], azo [6],
nitroso [7], tempo [8], etc. One typical concern in generating
radical peptides by CID is that charge directed backbone
dissociation processes compete with desired radical generation,
which may lead to suppression of radical yield in some
situations. In comparison, photodissociation (PD) is capable of
generating radical biomolecules via dissociative excited state
chemistry that results in homolytic bond cleavage at specific
chromophores. In particular, 266 nm UV photons can
selectively cleave carbon–iodine (C–I), carbon–sulfur, or
sulfur–sulfur bonds in modified peptides and proteins [9–11].
This method works independently of molecular size, and a
single radical in a whole protein can easily be generated with
nitric oxide is a small endogenous diatomic radical that
regulates blood flow. Radical chemistry is also key in the
synthesis of nucleotides by ribonucleotide reductase for
eventual synthesis of DNA. Reactive oxygen species such as
hydroxyl and peroxide radicals are utilized beneficially by
the immune system to attack invading organisms. On the
other hand, reactive oxygen species are also byproducts of
respiration or can be generated by other exogenous sources,
and can harmfully attack lipids, proteins, and DNA [2].
Radical chemistry also plays an important role in mass
spectrometry, where information about sequence and structure
can be derived from the radical driven fragmentation of
peptides or proteins via methods such as electron capture
dissociation (ECD) or electron transfer dissociation (ETD).
Both of these methods (at least initially) generate hydrogen-
Electronic supplementary material The online version of this article
(doi:10.1007/s13361-012-0540-6) contains supplementary material, which
is available to authorized users.
Correspondence to: Ryan R. Julian; e-mail: ryan.julian@ucr.edu

X. Zhang and R. R. Julian: Kinetics Radical Migration
525
high yield. The motivation for creating such radical species is Experimental
that the fragmentation chemistry of hydrogen deficient radicals
Materials
differs from that of even electron species, which enables unique
experiments to be carried out. Radical directed dissociation of
biomolecules in the gas phase can directly identify post-
translational modification sites, distinguish peptide epimers,
and even probe protein tertiary structure in vacuo [10, 12–14].
Radical directed dissociation (RDD) in hydrogen defi-
cient radical peptides typically involves migration of the
radical from the initial position where it was created to the
site where dissociation is ultimately observed. These radical
migration reactions, which all involve hydrogen atom trans-
fers, are influenced by both thermodynamics and kinetics.
The enthalpy of any given radical migration reaction can be
easily estimated by determination of the relevant carbon–
hydrogen (C–H) bond dissociation energies (BDEs) of both
the radical donor and acceptor. Large BDEs indicate highly
reactive radicals which will be enthalpically favored to
migrate to sites with lower BDEs in the absence of kinetic
constraints. Previously, we have shown that, on average, the
relative abundance of backbone fragments and side chain
losses observed for a series of peptides correlated well with
the relevant BDEs; radical directed dissociation was favored
for those sites with low C–H BDEs [15]. However, radical
migration within any individual peptide will be kinetically
influenced by the relevant activation energies, which are a
function of three-dimensional structure and sequence-depen-
dent steric constraints. For example, in very small, rigidly
constrained peptides, high barriers to radical migration prevent
radical isomerization and lead to different fragmentation
patterns [16, 17]. On the other hand, similar fragmentation
patterns have been observed in small flexible peptides,
indicating that radical migration can also be facile [18, 19].
In the present work, we focus on kinetic constraints
rather than thermodynamics. More specifically, we have
examined a series of isomers where dissociation was
investigated as a function of radical initiation site. For
example, ortho-, meta-, and para- benzoyl radical
modified peptide isomers reveal that even minor changes
in radical initiation site can significantly influence
subsequent dissociation pathways. A cyclic peptide yields
virtually identical fragmentation patterns for three distinct
radical isomers in the +1 charge state, yet the +2 charge
state of the same peptide produces highly distinct
dissociation spectra for all three isomers. In order to
obtain more detailed structural insight into these results,
we carried out molecular dynamics simulations. Deuteri-
um labeling was used to experimentally map out one
complete migration pathway, which was used to calibrate
the molecular dynamics simulations. Subsequent investi-
gation revealed that structural flexibility accounts for the
different results in the +1 versus +2 charges states of the
cyclic peptide. Alignment of the radical donors and
acceptors is the primary controlling factor that distin-
guishes the results from ortho versus meta and para
radical isomers.
Peptides (RGYALG, RYLPT, cyclo-(RGDyK)) were purchased
from American Peptide Company (Sunnyvale, CA, USA); 2-
(hydroxymethyl-iodobenzoate)-18-crown-6 ether and N-
hydroxysuccinimide (NHS) activated iodo-benzoyl ester were
synthesized previously [20]. Sodium iodide, chloramine-T,
sodium metabisulfite, and other organic solvents were purchased
from Thermo-Fisher Scientific Pittsburgh, PA, USA. α-d2-
Glycine, α-d1-alanine, iso-propyl-d7-leucine and α-d1-leucine
were purchased from CDN Isotopes (Pointe-Claire, Quebec,
Canada), while β-d2-tyrosine was purchased from Cambridge
Isotope Laboratories (Andover, MA, USA). Fmoc-protected
amino acids were purchased from Anaspec (San Jose, CA,
USA) or LC Sciences (Houston, TX, USA). Wang resin and 2,6-
dichlorobenzoyl chloride (DCB) (Acros Organics, NJ, USA),
fluorenylmethyloxycarbonyl-O-succinimide (Fmoc-oSu) and
tetramethylammonium hexafluorophosphate (HCTU) (Chem-
Pep Inc., Miami, FL, USA), isopropylsilane (TIPS) (Sigma-
Aldrich, St. Louis, MO, USA) were also purchased.
Peptide Derivatization
Peptides were labeled at the N-terminus with NHS activated
iodo-benzoyl ester (2× molar ratio) in 1:3 borate buffer
(0.2 M, pH 8.5):dioxane solution for 30 min at 40 °C. The
side product at arginine and tyrosine side chains were
removed by incubation in 1 M hydroxylamine (adjusted pH
to 8–9 by NaOH) for 4 h. Tyrosine in cyclo-(RGDyK) was
iodinated with sodium iodide and chloramine-T at room
temperature, after 1 min reaction time, excess sodium
metabisulfite was added to quench the reaction. Stoichio-
metric quantities of reagents (1:2:1:2 molar ratio of peptide/
sodium iodide/chloramine-T/sodium metabisulfite) were
used to limit the iodination extent and produce mainly
mono-iodinated tyrosine. A reversed phase cartridge
(Michrom Bioresources, Inc., Auburn, CA, USA) was used
for peptide desalting and purification.
Synthesis of Deuterium Labeled RGYALG
Fmoc protected amino acids (e.g., α-d2-glycine, β-d2-
tyrosine, etc.) were synthesized by reaction with 1× Fmoc-
oSu and 1× NaHCO3in 1:1 dioxane: H2O solution overnight.
After dissolving reaction mixture in 5 % HCl, the product
was taken up using ethyl acetate and washed with 0.1 M
HCl and water. Following drying with anhydrous Na2SO4,
organic solvent was evaporated away and the protected
amino acid was crystallized. Peptides were synthesized by
solid phase methodology using Wang-resin and Fmoc
chemistry. HCTU and piperidine was used for carboxyl
group activation and amine group deprotection, respectively,
whereas DCB was used for adding the first residue to Wang
resin. Peptides were cleaved from the resin with 95:2.5:2.5
TFA:H2O:TIPS.

526
X. Zhang and R. R. Julian: Kinetics Radical Migration
Photodissociation Mass Spectrometry
was used for all calculations with the dielectric constant, ε0
1.0. Extended cutoffs were used for dipole–dipole interac-
tions. Energy minimization was performed using the Polak-
Ribiere conjugate gradient method with a derivative conver-
gence criterion of 0.05 kJ/Å·mol. Conformational searches
with the Monte Carlo multiple minimum algorithm and
enhanced torsional sampling were used with a maximum
number of 10,000 steps; 100,000 conformations were saved
with an interval of 0.1 ps during 10 ns molecular dynamics.
We chose 500 K–750 K as molecular dynamics temperature
to explore the peptide potential energy surface. These values
are comparable to previous MD calculations for peptides in
the gas phase and work well in our systems [25].
Peptides (10 μM) were dissolved in 50:50 H2O:CH3OH with
0.1 % acetic acid and electrosprayed into an LTQ linear
quadrupole ion trap mass spectrometer (Thermo Fisher
Scientific, San Jose, CA, USA). Fifty μM 2-(hydroxy-
methyl-iodobenzoate)-18-crown-6 ether was mixed with
10 μM peptide to form peptide-crown complexes; 266 nm
photons were generated from the fourth harmonic generation
of a Nd:YAG laser (Continuum, Santa Clara, CA, USA).
The back plate of the instrument was modified with a quartz
window to transmit UV photons into the linear ion trap.
Laser pulses were synchronized to the isolation step via a
digital delay generator. The isolation window for precursor
ions in collision induced dissociation and photodissociation
was set between 3 to 10m/z with an activation value q00.25
and default activation time 30 ms. For experiments utilizing
deuterium labeled peptide to track the radical migration, the
precursor ion isolation window was set at 1m/z. Activation
parameters were held constant for experiments where
similarity scores are calculated.
Results and Discussion
Fragmentation of Ortho-, Meta-,
and Para-Iodobenzoyl Modified Peptide
Radical Isomers
Radical isomers with the initial radical site at the ortho-,
meta-, or para-position of the benzoyl group in N-iodoben-
zoyl-RGYALG were generated by 266 nm UV photodisso-
ciation to selectively cleave the photolabile C–I bonds and
leave σ radicals on the benzene rings. The radical ions were
subsequently re-isolated and subjected to CID as shown in
Figure 1a-c. The results are typical for dissociation of a
hydrogen deficient radical peptide [i.e., side chain losses
(e.g., –43L, –56L, –106Y)] and selective backbone frag-
mentation (a3 only) are observed [15]. Consecutive side
chain losses are also observed in low abundance (e.g., –
56L–CO2, –43L-CO2) suggesting multistep radical migra-
tions occur to a lesser degree. Although the observed
fragmentation channels for the ortho-, meta-, or para-
benzoyl radical modified RGYALG isomers are very
similar, there is some variation in the degree to which each
dissociation channel is populated. In particular, the relative
intensity for the a3 ion from the ortho-radical isomer is
significantly more abundant than is observed for the meta-
and para-radical isomers. A similarity score (S, defined in
experimental methods) is used to quantitatively characterize
the similarity between two fragmentation patterns [21]. An S
score of 1 implies that two spectra are identical with the
same distributions of relative intensities; scores lower than 1
indicate less similarity. The reported similarity scores
conservatively have uncertainties of 0.05. Similarity scores
are indicated with brackets between each pair of spectra in
Figure 1. The meta-isomer (Figure 1b) is fairly similar (S0
0.85) to the para isomer (Figure 1c), whereas the ortho-
isomer is more distinct relative to both other spectra (S0
0.74, 0.59 for para and meta, respectively). The lower scores
for the ortho-isomer are primarily due to significantly
increased peak intensity for the a3 and leucine side chain
loss fragments.
Fragmentation Pattern Similarity Score
ꢀ
ꢀ
ꢀ
ꢀ
P
1
2
ImðiÞ ꢀ ImðiÞ
Im1 ðiÞ þ Im2 ðiÞ
i
P
S ¼ 1 ꢀ
i
Of all ions (i), Im1 and Im2 are intensities of an ion at m/z 0
“m” in two spectra excluding precursor ions [21]. A similarity
score (S) value of 1 represents two spectra that are exactly the
same. An in-house JAVA program was used to automatically
calculate S from the peak lists of two mass spectra.
C–H Bond Dissociation Energy Calculations
GaussView 5.0 was used to build N-methyl-benzoylamide,
which was used to model benzoic acid labeled peptides.
Isodesmic reactions [22] were used to calculate C–H bond
dissociation energies (BDEs) at ortho-, meta-, and para-
positions of N-methyl-benzoylamide with C6H5-H (472 kJ/
mol) [23] as reference molecule. Energy calculations were
performed at the B3LYP/6-31G(d) level of theory using
Gaussian 09 ver. 8.0.
Molecular Modeling
MacroModel, ver. 9.9, (Schrödinger, LLC, New York, NY,
USA) molecular mechanics software package was used to
build benzoic acid modified RGYALG and cyclo-(RGDyK)
and perform all molecular modeling simulations. Arginine
side chains were protonated first, followed by lysine side
chains for higher charge states. The OPLS force field [24]
As an approximation to benzoic acid modified RGYALG,
the BDEs for the ortho-, meta-, and para- C–H bonds in N-

X. Zhang and R. R. Julian: Kinetics Radical Migration
527
Figure 1. (a–c) CID spectrum of radical peptide N-benzoyl-RGYALG in the +1 charge state with initial radical at (a) ortho-, (b)
meta-, or (c) para-position of benzoyl labeling group at N-terminus. (d–f) CID spectrum of radical peptide N-benzoyl-RYLPT in
the +1 charge state with initial radical at (d) ortho-, (e) meta-, or (f) para-position of benzoyl labeling group at N-terminus. Side
chain losses are abbreviated (i.e., –43L represents 43 Da side chain loss from leucine. Similarity scores (S) are indicated for
each pair. The relevant radical precursor for each spectrum is indicated in the inset box, and precursor ions are indicated by a
downward arrow
methyl-benzoylamide were calculated to be 461, 473, and
474 kJ/mol, respectively. These benzene radical isomers
have very similar BDEs and are predicted to be very reactive
(e.g., the C-H BDE values of all tertiary carbons in 20 amino
acids range from 319 to 399 kJ/mol) [22]. Thermodynamics
are therefore not anticipated to dictate the behavior of these
radical isomers to any significant extent. The nature of our
experiments, which utilize low energy collisional activation,
should also favor kinetic products. The variations in
fragmentation observed in Figure 1 are therefore most likely
due to the influence of positional isomerism on relevant
transition state geometries that dictate radical migration. The
details of migration in this system were mapped out with
deuterium labeling and molecular mechanics calculations
that are reported further below.
obviously more distinct spectrum as shown in Figure 1d
(S00.54, 0.54). This is reminiscent of the results from
the N-benzoyl-RGYALG isomers, where the ortho-
isomer also exhibited the greatest difference. Taken
together, these results suggest that the ortho-isomers are
most distinct in terms of ability to influence the kinetics
of radical migration. This observation is not unexpected
given that both the meta- and para-isomers yield
radicals, which are on the end of the benzoyl addition,
whereas the ortho-isomer yields a radical that is directed
back towards the point of attachment. Although it is
clearly possible in some cases to accommodate the para-
to meta- shift by modest rotation or shifting of the
benzoyl group, it is not possible to achieve similar
alignment with the ortho-isomer.
Figure 1d–f show the CID spectra of radical isomers
for another peptide, N-benzoyl-RYLPT. Side chain losses
(e.g., –56L, –106Y) along with few radical initiated
backbone fragmentations (e.g., a2) are observed. The
spectrum for the meta-isomer shown in Figure 1e is very
similar to that for the para-isomer in Figure 1f (S00.99).
Thus for RYLPT, the para and meta positional isomers
are essentially identical in terms of influence on radical
migration kinetics. In contrast, the ortho-isomer yields an
Fragmentation of Cyclic Peptide Radical Isomers
A series of radical isomers for a small cyclic peptide cyclo-
(RGDyK) (the small cap y indicates D-tyrosine) were
subjected to collisional activation and the results are shown
in Figure 2. These radical isomers were produced by several
methods, and the radical initiator is identified within each
spectrum. Figure 2a shows CID of the +1 cyclo-(RGDyK)•

528
X. Zhang and R. R. Julian: Kinetics Radical Migration
Figure 2. CID spectrum of cyclo-(RGDyK)• in the +1 (a, b, c) and +2 (d, e, f) charge state with the radical introduced by ortho-
iodobenzoyl modified 18-crown-6 based noncovalent radical delivery (a, d), para-iodobenzoyl modified 18-crown-6 (b, e) and
tyrosine side chain iodination (c, f). Labeling of the spectra follows the same pattern established in Figure 1
radical isomer, which was produced by intermolecular
radical migration [15]. In this method, a noncovalent
complex between a peptide and a radical precursor (i.e.,
ortho-iodobenzoyl derivatized 18-crown-6) is isolated in the
ion trap. The iodine atom in the radical precursor is then
selectively removed by photodissociation. The resulting
phenyl radical is very reactive and abstracts a hydrogen
atom from the peptide. Subsequent activation steps isolate
and then fragment the peptide radical. In Figure 2b, a para-
iodobenzoyl derivatized 18-crown-6 radical precursor was
Scheme 1. Radical migration pathway and β scission mechanism for generation of a3 fragment from N-benzoyl-RGYALG with
the initial radical at ortho-position of benzoyl group

X. Zhang and R. R. Julian: Kinetics Radical Migration
529
used. In addition to the crown ether method, chemical
iodination at tyrosine side chain was used as described
previously [9]. Figure 2c shows the dissociation spectrum of
cyclo-(RGDy•K) radical generated by tyrosine iodination.
Most peaks in Figure 2a–c correspond to radical directed
side chain losses from lysine (–•NH2, –43K, –58K, –71K),
tyrosine (–106Y) and aspartic acid (–44D) residues. No
backbone dissociation is detected, which is not surprising
because at least two backbone bonds would need to be
cleaved for fragmentation to be observable. A rich fragmen-
tation pattern originating from the lysine side chain indicates
that the radical can freely migrate to almost every carbon on
its aliphatic chain [26]. Although the arginine side chain
provides a similar target for radical attack, no side chain
losses from arginine are observed. The arginine side chain is
also the most probable location for protonation, which may
enhance kinetic barriers via restricted conformational flexi-
bility and could also complicate detection if the side chain
loss retained the proton.
In Figure 2a and b, the initial sites where the radical
transfers to the peptide are unknown. Given the steric
differences caused by the ortho versus para substitution, it
is unlikely that the initial sites are the same for Figure 2a and
b. Similarly, it is unlikely that radical transfer via abstraction
of an aromatic tyrosine hydrogen atom will occur due to
enthalpic considerations [22]. Therefore, the precursor ions
in Figure 2a, b, and c should all be different isomers of
cyclo-(RGDyK)•. The striking similarity between all three
spectra obtained by activation of the different isomers is
therefore quite unexpected. Analogous experiments con-
ducted on RGYALG yielded significantly less similar
spectra (see supporting information).
The dissociation spectra for cyclo-(RGDyK)• isomers in
Figure 2a–c have extremely high similarity scores (0.94,
0.97, and 0.93), indicating very little difference in fragmen-
tation between the various isomers. This suggests that
barriers to radical migration via numerous pathways for this
peptide are likely low and comparable in magnitude. From
an enthalpic standpoint, the radical sites which yield the
fragments observed in Figure 2a–c are similar (BDEs
ranging from 390 to 416 kJ/mol) [22]. A flat thermodynamic
surface also favors a flat kinetic surface because endother-
mic reactions cannot contribute to kinetic barriers. Since the
radicals initially start at different locations but ultimately
yield very similar fragmentation patterns, the results further
suggest that multiple migrations are occurring, which allows
equilibration of the various isomers. The cyclic nature of the
peptide restricts the overall available conformational space;
however, this may actually facilitate interactions between the
flexible lysine side chain and the remainder of the peptide,
allowing migration to occur freely.
observed in the +1 charge state. Interestingly, the similarity
scores also change substantially (0.59, 0.50, 0.70), indicating
that kinetic barriers are preventing the free migration
observed in the +1 charge state. The addition of the second
charge should occur on the side chain of lysine, which is by
far the most basic available site. Intramolecular solvation of
the additional charged site and Coulombic repulsion be-
tween the charges will likely lead to significant stiffening of
the peptide structure and restriction of side chain flexibility,
which will both serve to increase kinetic barriers to
migration. This hypothesis is supported by molecular
modeling which will be discussed further below.
Pinpointing Radical Migration with Deuterium
Isotope Labeling
In Figure 1a–c, the a3 fragment intensity serves as a
diagnostic feature to distinguish between the ortho and meta
or para positional isomers. An abundant a3 ion is only
formed for the ortho-isomer, suggesting favorable kinetics
exist for this migration pathway. In order to explore the
structural features that lead to these favorable kinetics in
greater detail, it is necessary to determine the exact
migration route. It has been well established previously that
immediately prior to formation of the a3 ion, the radical must
migrate to the tyrosine β carbon as shown in Scheme 1c [15,
27]. However, the radical could arrive at the tyrosine β
carbon via direct migration or through a multistep route. To
determine the exact radical migration pathway, we prepared
a series of deuterium isotope labeled peptides. Collisional
activation of deuterium labeled tyrosine (Tyr β d2) yields an
abundant a3(d1) ion as shown in Figure 3a. Direct migration
of the initial radical would lead to deuterium on the N-
terminus and generation of an a3(d2) ion. The results are not
consistent with direct migration but rather indicate that
deuterium has migrated to a site within the ALG residues. It
should be mentioned that some amount of direct migration
cannot be completely excluded because a small quantity of
a3(d2) is observed in Figure 3a, but direct migration is
clearly not the primary pathway. The results suggest that
initial radical at the N-terminus migrates to an intermediate
site within ALG first, followed by subsequent migration to
the tyrosine β carbon.
In an attempt to determine which intermediate site within
the ALG residues the ortho- benzoyl radical migrates to,
additional deuterium labeled peptides (i.e., Gly6 α d2, Leu α
d1, Leu d7 isopropyl, Ala α d1) were synthesized and the
results are shown in Figure 3b–e, respectively. None of the
labeling sites at Gly6 or Leu are attacked by the benzoyl
radical, otherwise an a3(d1) ion would be observed as the
major peak in the isotope distribution. However, the
observation of a3(d1) in Figure 3e suggests that the Ala α
deuterium was abstracted directly as shown in Scheme 1a.
Successive radical migration from the Ala α carbon to the
Tyr β carbon is expected to be a facile process through a
transition state with a six-membered ring. Our previous work
The situation changes dramatically when another charge
is added to the peptide. Analogous experiments were
conducted on the +2 charge state and are shown in
Figure 2d, e, and f. The relative intensities of several side
chain losses are notably different compared with those

530
X. Zhang and R. R. Julian: Kinetics Radical Migration
Figure 3e is inconsistent with the proposed two step
migration mechanism (Scheme 1) suggesting the existence
of other minor migration routes not involving the Ala α
carbon. The intensity ratio between a3(d0) and a3(d1) is 1:
2.5 in Figure 3e, suggesting the major pathway is the one
outlined in Scheme 1. Kinetic isotope effects further support
radical migration involving the Ala α hydrogen. The
intensity of the a3 ion in Figure 3e decreases significantly
compared to that of the unlabeled peptide (see Figure 1a),
which is consistent with a kinetic isotope effect disfavoring
abstraction of deuterium.
Figure 4 shows the dissociation spectra of meta- and
para-benzoyl radical RGYA(α d1)LG isomers. In contrast to
ortho-isomer in Figure 3e, observation of a3(d0) ion as the
major isotope peak in Figure 4 indicates that the radical
migration via the Ala α carbon is not a major pathway to
generate the a3 fragment for these two isomers. The relative
intensity ratio between a3(d0) and a3(d1) is 1 : 0.63 for the
meta-radical isomer (Figure 4a) and 1 : 0.47 for the para-
isomer (Figure 4b). Compared with the 1: 2.5 ratio for the
ortho-isomer in Figure 3e, the extent of radical migration to
Ala α occurs with the following preference: ortho 9 meta 9
para. This is consistent with the spectra in Figure 1a–c
showing that ortho-radical isomer has the largest relative
intensity of a3. This order is found to be related to peptide
3D structure based on molecular dynamics calculations as
discussed below.
Molecular Modeling
Conformational searches and molecular dynamics (MD)
simulations were carried out to determine in greater detail
Figure 3. CID spectrum of N-benzoyl-RGYALG (+1) with
radical at the ortho-position of benzoyl group and deuterium
labeling at (a) Tyr β d2, (b) Gly6 α d2, (c) Leu α d1, (d) Leu γ
d7 (isopropyl), and (e) Ala α d1. Down arrows indicate
precursor ions
has shown that Tyr β radical can migrate from the carbon α
position at an adjacent glycine residue [27]. Similarly, the
Ala α radical should also be able to migrate to adjacent Tyr β
carbon (Scheme 1b).
In order to determine if there are additional intermediate
sites between the initial ortho- benzoyl radical and the Ala α
carbon (Scheme 1a), the a3(d1) isotope peak in Figure 3e
was re-isolated and dissociated to identify the position of the
deuterium. A benzoyl fragment containing a deuterium
(Supplementary Figure S2) suggests that the Ala α deuteri-
um was abstracted directly by the benzoyl radical. Data from
deuterium labeling at the Gly2 α position also suggests that
Gly2 α hydrogens are not involved in the major fragmen-
tation pathway for a3 ion (Supplementary Figure S3). It is
worth noting that observation of the a3(d0) isotope peak in
Figure 4. CID spectrum of N-benzoyl-RGYA(α d1)LG (+1)
radical isomers with deuterium labeling at Ala α d1 and
radical at the meta- position (a) and para- position (b) of the
benzoyl group. Downward arrows indicate precursor ions

X. Zhang and R. R. Julian: Kinetics Radical Migration
531
the influence of peptide structure and dynamics on radical
migration. Force field parameters for odd-electron radicals
are not readily available, so saturated peptides were
constructed instead. Distances between hydrogen atoms
representing potential radical donor and acceptor sites were
monitored in lieu of calculating distances involving actual
radical sites.
Figures 5b–d show the conformations obtained during
molecular dynamics with the smallest distance from the Ala
α hydrogen to the ortho- (Figure 5b), meta- (Figure 5c), or
para- (Figure 5d) benzoyl hydrogen. All distances are
smaller than 3 Å and should be satisfactory for radical
migration; however, distance is not the only structural
parameter that can influence radical migration kinetics.
Hydrogen transfer preferentially proceeds via a linear or
near-linear pathway in the transition state [28]. The
alignment of bonds involved in potential hydrogen transfer
from benzoyl ortho-hydrogen to the Ala α carbon
(Figure 5b) is near-linear and thus optimal for hydrogen
transfer. In comparison, the meta- and para-isomers have
unfavorable nonlinear orientations and much higher activa-
tion barriers are expected as a consequence. The average
results are also consistent with these snapshots, with a
significantly higher fraction of the ortho-hydrogen isomer
exhibiting near linear alignment (see Supplementary
Figure S5).
N-Benzoyl-RGYALG
The lowest potential energy structure for N-benzoyl-
RGYALG obtained from a molecular mechanics conforma-
tional search is shown in Figure 5a. Interestingly, the N-
terminal benzoyl group is very close to the alanine residue,
which should kinetically facilitate radical migration from the
benzoyl group to the Ala α carbon. In an attempt to
understand why the ortho-benzoyl radical isomer favors
radical migration to the Ala α carbon more than the other
two isomers, the distances from the Ala α hydrogen to
benzoyl ortho-, meta-, and para-hydrogens, were monitored
during a 10 ns MD simulation at 500 K. Because hydrogen
transfer can only occur when the donor is in very close
proximity to the acceptor, the probability of residue contact
at a distance below 4 Å was calculated for 100,000 sampled
structures (Supplementary Figure S4). The ortho-benzoyl
hydrogen has the highest probability for close contact with
the Ala α hydrogen (26 %), whereas the probabilities for the
meta- and para-hydrogens are 8.6 % and 3.6 %, respective-
ly. This trend is consistent with the experimental data on the
extent of radical migration to the Ala α carbon which
follows the same order: ortho 9 meta 9 para.
Cyclo-(RGDyK) in the +1 and +2 Charge State
The fragmentation similarities for cyclo-(RGDyK) radical
isomers in Figure 2 vary with charge state. As suggested
above, this may be caused by restricted conformational
space in the +2 charge state, which may inhibit radical
migration. We carried out MD simulations for both charge
states and monitored several distances between side chains.
Figure 6 shows the statistical analysis of distances from
100,000 conformations sampled during 10 ns of MD at
750 K. Distances from the tyrosine ortho-hydrogen (the
initial radical site in tyrosine-iodinated peptide) to the lysine
δ-hydrogen are quite different between the +2 (Figure 6a)
and +1 (Figure 6b) charge states. In the +1 charge state, the
peptide exhibits a very broad distribution of distances,
suggesting fairly unbiased occupation of the conformational
space. For the +2 charge state, two distinct populations are
clearly observed. It should be mentioned that a slightly
greater number of structures with close contact is observed
in the +2 charge state, although the corresponding lysine
side chain losses are not more abundant for the +2 charge
state (Figure 2c and f). This may indicate that direct
migration is not the primary mechanism that generates these
fragments. The primary purpose of the data in Figure 6a and
b is to illustrate potential differences in conformational
space, not to elucidate migration pathways (as is the case in
Supplementary Figures S4 and S5).
In Figure 6c and d, the distances between the arginine δ
hydrogen and the lysine δ hydrogen are shown. Both side
chains are potential intermediate radical migration sites. In
the +2 charge state, the protonation sites are most likely at
lysine and arginine which leads to greater separation in
space (96 Å) as shown in Figure 6c. In comparison, there is
some contact (distance below 4 Å) between lysine and
arginine side chains in the +1 charge state (Figure 6d), which
again suggests larger conformational space which should be
Figure 5. (a) The lowest potential energy structure of N-
benzoyl-RGYALG resulting from a Monte Carlo multiple
minimum conformational search; (b-d) structures with the
smallest distance from the Ala α hydrogen to the ortho-
benzoyl hydrogen, meta-benzoyl hydrogen, or para-benzoyl
hydrogen obtained during 10 ns of MD simulations at 500 K
are shown in (b), (c), (d), respectively

532
X. Zhang and R. R. Julian: Kinetics Radical Migration
Figure 6. Histograms of monitored distances for cyclo-(RGDyK) in the +1 and +2 charge state during 10 ns molecular
dynamics at 750 K from 100,000 sampled conformations. Distances between the ortho-hydrogen in tyrosine side chain and the
δ-hydrogen in lysine are plotted in (a) for the +2 charge state and (b) for +1. Distances between the δ-hydrogen in arginine side
chain and the δ-hydrogen in lysine side chain are plotted in (c) for the +2 charge state and (d) for +1
beneficial for radical migration. These MD simulations are
consistent with the hypothesis that conformational flexibility
and multiple migrations are the primary sources of the
experimentally observed differences in dissociation between
the +1 and +2 charge states as shown in Figure 2.
proteins are expected to have restricted conformational space
due to high levels of intramolecular interactions, which
should (in general) lead to larger kinetic barriers for
migration. Results from this work suggest that radical
migrations in proteins should be highly structure dependent,
which should be beneficial for structure investigation.
Conclusions
Acknowledgments
The present results illustrate that radical migrations in
peptides are not unrestricted but can be significantly
influenced by peptide 3D structure. Information on radical
migration pathways can be derived from various approaches
including evaluation of fragmentation pattern similarities
among positional isomers, deuterium isotope labeling, and
molecular mechanics calculations. For example, site specific
deuterium labeling in RGYALG reveals that the initial
ortho-benzoyl radical at N-terminus prefers migration to the
tyrosine β carbon indirectly via the alanine α carbon as an
intermediate site. However, this radical migration pathway is
kinetically disfavored for meta- and para-radical isomers as
a result of steric hindrance. The charge state dependent
fragmentation pattern similarities among cyclo-(RGDyK)•
isomers indicate that radical migration pathways are sensi-
tive to the peptide structural flexibility. A relatively wide
conformational space in the +1 charge state allows sufficient
residue contact to facilitate radical migration, whereas
narrower conformational space in the +2 charge state results
in higher energy barriers to radical migrations. Folded
The authors gratefully acknowledge funding from the
National Science Foundation (CHE-074748).
References
1. Dean, R.T., Fu, S.L., Stocker, R., Davies, M.J.: Biochemistry and
pathology of radical-mediated protein oxidation. Biochem. J. 324, 1–18
(1997)
2. Hawkins, C.L., Davies, M.J.: Generation and propagation of radical
reactions on proteins. BBA Bioenergetics 1504, 196–219 (2001)
3. Chu, I.K., Rodriquez, C.F., Lau, T.C., Hopkinson, A.C., Siu, K.W.M.:
Molecular radical cations of oligopeptides. J. Phys. Chem. B 104, 3393–
3397 (2000)
4. Headlam, H.A., Mortimer, A., Easton, C.J., Davies, M.J.: beta-scission
of C-3 (beta-carbon) alkoxyl radicals on peptides and proteins: A novel
pathway which results in the formation of alpha-carbon radicals and the
loss of amino acid side chains. Chem. Res. Toxicol. 13, 1087–1095
(2000)
5. Masterson, D.S., Yin, H.Y., Chacon, A., Hachey, D.L., Norris, J.L.,
Porter, N.A.: Lysine peroxycarbamates: Free radical-promoted peptide
cleavage. J. Am. Chem. Soc. 126, 720–721 (2004)
6. Hodyss, R., Cox, H.A., Beauchamp, J.L.: Bioconjugates for tunable
peptide fragmentation: Free radical initiated peptide sequencing
(FRIPS). J. Am. Chem. Soc. 127, 12436–12437 (2005)

X. Zhang and R. R. Julian: Kinetics Radical Migration
533
7. Hao, G., Gross, S.S.: Electrospray tandem mass spectrometry analysis
of S- and N-nitrosopeptides: Facile loss of NO and radical-induced
fragmentation. J. Am. Soc. Mass Spectrom. 17, 1725–1730 (2006)
8. Lee, M., Kang, M., Moon, B., Oh, H.B.: Gas-phase peptide sequencing
by TEMPO-mediated radical generation. Analyst 134(8), 1706–1712
(2009)
9. Ly, T., Julian, R.R.: Residue-specific radical-directed dissociation of
whole proteins in the gas phase. J. Am. Chem. Soc. 130, 351–358
(2008)
10. Diedrich, J.K., Julian, R.R.: Site-specific radical directed dissociation of
peptides at phosphorylated residues. J. Am. Chem. Soc. 130, 12212–
12213 (2008)
11. Agarwal, A., Diedrich, J.K., Julian, R.R.: Direct elucidation of disulfide
bond partners using ultraviolet photodissociation mass spectrometry.
Anal. Chem. 83(17), 6455–6458 (2011)
18. Wee, S., Mortimer, A., Moran, D., Wright, A., Barlow, C.K., O'Hair,
R.A.J., Radom, L., Easton, C.J.: Gas-phase regiocontrolled generation
of charged amino acid and peptide radicals. Chem. Commun. 40, 4233–
4235 (2006)
19. Song, T., Ng, D.C.M., Quan, Q.A., Siu, C.K., Chu, I.K.: Arginine-
facilitated alpha- and pi-radical migrations in Glycylarginyltryptophan
radical cations. Chem Asian J 6, 888–898 (2011)
20. Ly, T., Zhang, X., Sun, Q.Y., Moore, B., Tao, Y.Q., Julian, R.R.: Rapid,
quantitative, and site-specific synthesis of biomolecular radicals from a
simple photocaged precursor. Chem. Commun. 47, 2835–2837 (2011)
21. Dong, N.P., Liang, Y.Z., Yi, L.Z.: Investigation of scrambled ions in
tandem mass spectra. Part 1. Statistical characterization. J. Am. Soc.
Mass Spectrom. 23, 1209–1220 (2012)
22. Moore, B.N., Julian, R.R.: Dissociation energies of X–H bonds in
amino acids. Phys. Chem. Chem. Phys. 14, 3148–3154 (2012)
23. Blanksby, S.J., Ellison, G.B.: Bond dissociation energies of organic
molecules. Acc. Chem. Res. 36, 255–263 (2003)
12. Tao, Y., Quebbemann, N.R., Julian, R.R.: Discriminating D-amino acid-
containing peptide epimers by radical-directed dissociation mass
spectrometry. Anal. Chem. 84(15), 6814–6820 (2012)
13. Ly, T., Julian, R.R.: Elucidating the tertiary structure of protein ions in
vacuo with site specific photoinitiated radical reactions. J. Am. Chem.
Soc. 132, 8602–8609 (2010)
24. Jorgensen, W.L., Maxwell, D.S., TiradoRives, J.: Development and
testing of the OPLS all-atom force field on conformational energetics
and properties of organic liquids. J. Am. Chem. Soc. 118, 11225–11236
(1996)
14. Zhang, X., Julian, R.R.: Investigating the gas phase structure of KIX
with radical directed dissociation and molecular dynamics: Retention of
the native structure. Int. J. Mass Spectrom. 308, 225–231 (2011)
15. Sun, Q.Y., Nelson, H., Ly, T., Stoltz, B.M., Julian, R.R.: Side chain
chemistry mediates backbone fragmentation in hydrogen deficient
peptide radicals. J. Proteome Res. 8, 958–966 (2009)
16. Chu, I.K., Zhao, J., Xu, M., Siu, S.O., Hopkinson, A.C., Siu, K.W.M.:
Are the radical centers in peptide radical cations mobile? The
generation, tautomerism, and dissociation of isomeric α-carbon-centered
triglycine radical cations in the gas phase. J. Am. Chem. Soc. 130,
7862–7872 (2008)
25. Tsaprailis, G., Somogyi, A., Nikolaev, E.N., Wysocki, V.H.: Refining
the model for selective cleavage at acidic residues in arginine-
containing protonated peptides. Int. J. Mass Spectrom. 195, 467–479
(2000)
26. Wee, S., O’Hair, R.A.J., McFadyen, W.D.: Comparing the gas-phase
fragmentation reactions of protonated and radical cations of the
tripeptides GXR. Int. J. Mass Spectrom. 234, 101–122 (2004)
27. Ly, T., Julian, R.R.: Tracking radical migration in large hydrogen
deficient peptides with covalent labels: Facile movement does not equal
indiscriminate fragmentation. J. Am. Soc. Mass Spectrom. 20, 1148–
1158 (2009)
17. Ng, D.C.M., Song, T., Siu, S.O., Siu, C.K., Laskin, J., Chut, I.K.:
Formation, isomerization, and dissociation of α-carbon-centered and π-
centered glycylglycyltryptophan radical cations. J. Phys. Chem. B 114,
2270–2280 (2010)
28. Moran, D., Jacob, R., Wood, G.P.F., Coote, M.L., Davies, M.J., O'Hair,
R.A.J., Easton, C.J., Radom, L.: Rearrangements in model peptide-type
radicals via intramolecular hydrogen-atom transfer. Helv. Chim. Acta
89, 2254–2272 (2006)