Specific adducts formed through a radical reaction between peptides and contact allergenic hydroperoxides

Article abstract of DOI:10.1021/tx9003352

The first step in the development of contact allergy (allergic contact dermatitis) includes the penetration of an allergy-causing chemical (hapten) into the skin, where it binds to macromolecules such as proteins. The protein - hapten adduct is then recognized by the immune system as foreign to the body. For hydroperoxides, no relevant hapten target proteins or protein - hapten adducts have so far been identified. In this work, bovine insulin and human angiotensin I were used as model peptides to investigate the haptenation mechanism of three hydroperoxide haptens: (5R)-5-isopropenyl-2-methyl-2- cyclohexene-1-hydroperoxide (Lim-2-OOH), cumene hydroperoxide (CumOOH), and 1-(1-hydroperoxy-1-methylethyl) cyclohexene (CycHexOOH). These hydroperoxides are expected to react via a radical mechanism, for which 5,10,15,20-tetraphenyl- 21H,23H-porphine iron(III) chloride (Fe(III)TPPCl) was used as a radical initiator. The reactions were carried out in 1:1 ethanol/10 mM ammonium acetate buffer pH 7.4, for 3 h at 37 °C, and the reaction products were either enzymatically digested or analyzed directly by MALDI/ TOF-MS, HPLC/MS/MS, and 2D gel electrophoresis. Both hydroperoxide-specific and unspecific reaction products were detected, but only in the presence of the iron catalyst. In the absence of catalyst, the hydroperoxides remained unreacted. This suggests that the hydroperoxides can enter into the skin and remain inert until activated. Through the detection of a Lim-2-OOH adduct bound at the first histidine (of two) of angiotensin I, it was confirmed that hydroperoxides have the potential to form specific antigens in contact allergy.

Full text of DOI:10.1021/tx9003352

Chem. Res. Toxicol. 2010, 23, 203–210
203
Specific Adducts Formed through a Radical Reaction between
Peptides and Contact Allergenic Hydroperoxides
Theres Redeby,†,‡ Ulrika Nilsson,‡ Timothy M. Altamore,† Leopold Ilag,‡
Annalisa Ambrosi,‡ Kerstin Broo,† Anna Bo¨rje,*,† and Ann-Therese Karlberg†
Department of Chemistry, Dermatochemistry and Skin Allergy, Medicinal Chemistry, UniVersity of Gothenburg,
Gothenburg, Sweden, and Department of Analytical Chemistry, Stockholm UniVersity, Stockholm, Sweden
ReceiVed September 16, 2009
The first step in the development of contact allergy (allergic contact dermatitis) includes the penetration
of an allergy-causing chemical (hapten) into the skin, where it binds to macromolecules such as proteins.
The protein-hapten adduct is then recognized by the immune system as foreign to the body. For
hydroperoxides, no relevant hapten target proteins or protein-hapten adducts have so far been identified.
In this work, bovine insulin and human angiotensin I were used as model peptides to investigate the
haptenation mechanism of three hydroperoxide haptens: (5R)-5-isopropenyl-2-methyl-2-cyclohexene-1-
hydroperoxide (Lim-2-OOH), cumene hydroperoxide (CumOOH), and 1-(1-hydroperoxy-1-methylethyl)
cyclohexene (CycHexOOH). These hydroperoxides are expected to react via a radical mechanism, for
which 5,10,15,20-tetraphenyl-21H,23H-porphine iron(III) chloride (Fe(III)TPPCl) was used as a radical
initiator. The reactions were carried out in 1:1 ethanol/10 mM ammonium acetate buffer pH 7.4, for 3 h
at 37 °C, and the reaction products were either enzymatically digested or analyzed directly by MALDI/
TOF-MS, HPLC/MS/MS, and 2D gel electrophoresis. Both hydroperoxide-specific and unspecific reaction
products were detected, but only in the presence of the iron catalyst. In the absence of catalyst, the
hydroperoxides remained unreacted. This suggests that the hydroperoxides can enter into the skin and
remain inert until activated. Through the detection of a Lim-2-OOH adduct bound at the first histidine
(of two) of angiotensin I, it was confirmed that hydroperoxides have the potential to form specific antigens
in contact allergy.
not been shown to react via nucleophilic-electrophilic interac-
tions. Instead, hydroperoxides are believed to form covalent
bonds though a radical mechanism (1, 2). It has therefore been
discussed that hydroperoxides do not form specific antigens but
cause a general hypersensitivity, through an unspecific oxidation
reaction. In contrast, both clinical and experimental studies have
demonstrated specific responses to the hydroperoxides inves-
tigated (10). The involvement of carbon, alkoxy, and peroxyl
radicals has been demonstrated using radical trappers and cross-
reactivity studies (11, 12). However, a radical reaction pathway
of hydroperoxides for haptenation of proteins has so far not
been confirmed in studies on hapten-protein/peptide interactions.
The aim of the present study was to investigate the role of
radicals in the formation of hapten adducts. This was done by
investigating the products formed in the reaction between the
known sensitizers (5R)-5-isopropenyl-2-methyl-2-cyclohexene-
1-hydroperoxide (1, Lim-2-OOH), cumene hydroperoxide (2,
CumOOH), and 1-(1-hydroperoxy-1-methylethyl) cyclohexene
(3, CycHexOOH)(Figure 1) and peptides modeling skin proteins.
Introduction
Contact sensitization and its clinical manifestation, allergic
contact dermatitis (ACD1), are caused by low molecular weight
compounds (haptens) which are able to penetrate into the skin.
It is generally accepted that the haptens must bind to high
molecular weight compounds in the skin to become im-
munogenic, and the formation of immunogens is normally
considered to take place between an electrophilic hapten and
nucleophilic moieties in amino acid side chains of skin proteins
(1). In many cases, the prerequisite for this interaction is the
chemical activation of the compounds to more reactive species
either by autoxidation on exposure to air or by metabolic
activation (2).
Fragrance compounds are the second most common cause
of contact allergy after nickel (3). The monoterpenes limonene
from citrus and linalool from lavender, or otherwise obtained
by chemical synthesis, are the most commonly used fragrance
compounds in perfumes found in every day products (4). They
are themselves very weak allergens but are activated to strong
sensitizers by autoxidation (5-9). Their primary oxidation
products, the hydroperoxides, are the major sensitizers in the
oxidation mixture formed on air exposure. Hydroperoxides have
Experimental procedures
Caution: The following chemicals are sensitizing, thus skin
contact should be aVoided: limonene-2-hydroperoxide, cumene
hydroperoxide, cyclohexene hydroperoxide, and carVone.
* To whom correspondence should be addressed. Dermatochemistry and
Skin Allergy, Department of Chemistry, University of Gothenburg, Ke-
miva¨gen 10, SE-412 96 Gothenburg, Sweden. Phone +46 31 772 4725.
Fax +46 31 772 3840. E-mail: aborje@chem.gu.se.
† University of Gothenburg.
Chemicals and Biochemicals. Bz-N-His-OMe was purchased
from Bachem (Bubendorf, Switzerland). (5R)-2-Methyl-5-(prop-
1-en-2-yl)cyclohex-2-enone (R-carvone, 98%), cumene hydroper-
oxide (CumOOH) Luperox CU90, 5,10,15,20-tetraphenyl-21H,23H-
porphine iron(III) chloride (Fe(III)TPPCl), human angiotensin I,
bovine insulin, trypsin, chymotrypsin, proteinase K, Boc-Tyr-OMe,
and the MALDI matrix R-cyano-4-hydroxycinnamic acid (HCCA)
‡ Stockholm University.
1 Abbreviations: ACD, allergic contact dermatitis; Fe(III)TPPCl, 5,10,15,20-
tetraphenyl-21H,23H-porphine iron(III) chloride.
10.1021/tx9003352  2010 American Chemical Society
Published on Web 11/23/2009

204 Chem. Res. Toxicol., Vol. 23, No. 1, 2010
Redeby et al.
curtain gas, 30 psi; collision gas, 6 psi; ion source gas 1, 30 psi;
ion source gas 2, 40 psi; ion spray voltage, 4500 V; temperature,
200 °C; and collision energy, 30 or 40 V. Analyses were also
performed on corresponding control mixtures.
Reactions of Hydroperoxides with Angiotensin I, Insulin,
Boc-Tyr-OMe and Bz-N-His-OMe. The following stock solutions
of angiotensin I, insulin, Boc-Tyr-OMe, BZ-N-His-OMe, Lim-2-
OOH, CycHexOOH, CumOOH, and Fe(III)TPPCl were prepared:
peptides (1 or 10 mg/mL) in 10 mM ammonium acetate (pH 7.4),
amino acids (1 mg/mL) in EtOH, hydroperoxide (10 mg/mL) in
EtOH, and Fe(III)TPPCl (0.034 mg/mL) in EtOH. New stock
solutions of the hydroperoxides were made directly before starting
each reaction while all other stock solutions were used either fresh
or after storage at -20 °C. Reaction mixtures of peptide/hydro-
peroxide/Fe(III)TPPCl or amino acid/hydroperoxide/Fe(III)TPPCl
were prepared from the stock solutions at molar ratios of 1:10:0.1
(if not otherwise stated in the text) with a total reaction volume of
0.10-1.0 mL. The final solvent ratio in the mixtures was 1:1 (v/v)
EtOH/buffer. The reaction was initiated by adding Fe(III)TPPCl
and then allowed to proceed at 37 °C for up to 180 min in darkness.
In parallel to every reaction, three control mixtures were prepared,
containing peptide/hapten, peptide/Fe(III)TPPCl, or peptide alone.
Equivalent control mixtures were prepared for the amino acids. The
reaction mixtures were analyzed directly after the reaction time
specified in the text and figures or stored at -20 °C. Analysis of
the peptide reaction was performed as described above using
MALDI/TOF-MS, while the amino acid reactions were analyzed
using LC/MS.
Native PAGE of Insulin Reaction. The volumes of the reaction-
and control mixtures were reduced to half using vacuum centrifuga-
tion (Savant SpeedVac, Techtum Lab, Umeå), where after the
mixtures were diluted 1:1 (v/v) with 10 mM ammonium acetate
buffer (pH 7.4). Each well on a 16% native polyacrylamide gel
was then loaded with10 µL of 1:2 (v/v) sample/sample buffer,
resulting in a total peptide load of 3.0 µg per well. The buffers
used were running buffer (10 mM Tris base with glycine (3% w/v),
sample buffer (glycerol (30% v/v) with running buffer (30% v/v)
and bromophenol blue (about 0.01% w/v)), stacking buffer (470
mM Tris base pH adjusted to 6.7 with H3PO4), and separation buffer
(1.5 M Tris base, pH adjusted to 8.9 with HCl). Visualization was
done with Coomassie stain, and the gel was photographed (Canon
EOS 400D).
SDS-PAGE of Insulin Reaction. The volumes of the reaction
and control mixtures were reduced by half by vacuum centrifugation
as in the case of the native PAGE experiments. The samples were
then diluted 1:1 (v/v) with 10 mM ammonium acetate buffer (pH
7.4). Each sample (10 µL) was mixed with 3.0 µL of sample buffer:
63 mM Tris at pH 6.8, glycerol (10% v/v), SDS (2.5% w/v), and
bromophenol blue (0.02% w/v). The samples were then boiled for
3 min before loading them on a 15% polyacrylamide gel with
running buffer: 50 mM Tris, glycine (3% w/v), and SDS (0.1%
w/v). Visualization was done with Coomassie stain, and the gel
was photographed (Canon EOS 400D).
Enzymatic Digestion of Angiotensin I Reaction. Prior to
digestion, the angiotensin I reaction and control mixtures were
evaporated to dryness through vacuum-centrifugation (Savant
SpeedVac, Techtum Lab, Umeå). Samples were then redissolved
in each digestion buffer (see below) to a volume of 100-500 µL.
For each experiment, three different enzymatic digestions were
performed using trypsin, chymotrypsin, or proteinase K. Stock
solutions of the proteases were 0.01 mg/mL trypsin in 50 mM
ammonium bicarbonate (pH 8), 2 mg/mL chymotrypsin in 1 mM
HCl and 2 mM CaCl2, and 1 mg/mL proteinase K in 0.5 M
ammonium acetate (pH 7.8) and 100 mM CaCl2. The digestions
were then run in the following way: tryptic digestion of angiotensin
I (1:50 w/w) in 0.5 M ammonium acetate (pH 7.8) and 100 mM
CaCl2 at 37 °C for 24 h. Chymotryptic digestion of angiotensin I
(1:50 w/w) in 0.5 M ammonium acetate (pH 7.8) and 100 mM
CaCl2 at 30 °C for 24 h. Proteinase K digestion of angiotensin I
(1:20 w/w) in 10 mM ammonium acetate (pH 7.4) and 5 mM CaCl2
at 37 °C for 18 h. Digested samples were desalted by ZipTip C18
Figure 1. Chemical structures of compounds referred to in this article.
(5R)-5-isopropenyl-2-methyl-2-cyclohexene-1-hydroperoxide (1, Lim-
2-OOH), cumene hydroperoxide (2, CumOOH), and 1-(1-hydroperoxy-
1-methylethyl) cyclohexene (3, CycHexOOH), and carvone (4).
were obtained from Sigma-Aldrich (Stockholm, Sweden). The
Sequazyme Peptide Mass Standards Kit Calibration mixture 2 from
Applied Biosystems (Stockholm, Sweden) was used for the
MALDI/TOF-MS calibration. The molecular weight markers used
for native PAGE were PageRuler Unstained Protein Ladder
#SM0661 and for SDS-PAGE PageRuler #SM 1811 (6.5-200
kDa), both from Fermenta (Helsingborg, Sweden). The C18 ZipTips
and the Synergy 185 system used for purification of water were
from Millipore (Billerica, USA). Organic solvents and other
chemicals were from Sigma-Aldrich or BioRad (Sundbyberg,
Sweden). Bovine insulin contains two amino acid chains with intra-
and intermolecular disulfide bonds. They are marked here by
superscripts: GIVEQC1C2ASVC1SLYQLENYC3N (R-chain) and
FVNQHLC2GSHLVEALYLVC3GERGFFYTPKA (ꢀ-chain). The
amino acid sequence of human angiotensin I is NRVYIHPFHL.
The theoretical average m/z value of the quasi-molecular ion [M
+ H]+ is 5733.49 amu for insulin and 1296.69 amu for angiotensin
I.
Synthesis of Lim-2-OOH and CycHexOOH. The syntheses of
Lim-2-OOH (12) and CycHexOOH (11) were performed as reported
in the literature.
Stability of Lim-2-OOH. Lim-2-OOH (7.5 mg, 44.58 µmol)
was dissolved in base-washed CDCl3 (0.6 mL) and placed in an
NMR tube. A stock solution of the Fe(III)TPPCl (3.1 mg, 4.40
µmol) in base-washed CDCl3 (1.5 mL) was added (150 µL) to the
NMR tube at t ) 0. A control solution without catalyst was also
prepared. The degradation of Lim-2-OOH was then followed over
time by 1H NMR at 400 MHz on a JEOL Eclipse+ 400 instrument.
The endocyclic olefinic proton was used as a measure of the
progress of the reaction.
MALDI/TOF-MS Analysis of the Reaction between
Hydroperoxides and Angiotensin I or Insulin. Analyses were
performed on a Voyager DE STR MALDI/TOF instrument (Applied
Biosystems) using a 337 nm laser, with a 3 ns pulse width, and a
20 Hz repetition rate. Samples were mixed 1:1 (v/v) with matrix
solution, and 1 µL of this mixture was applied onto the MALDI
target. Instrument settings were optimized for the calibration mix
at m/z ranges 100-2000, 700-9000, or 1400-9000, in positive
reflector mode. Analyses were also performed on the corresponding
control mixtures.
LC/MS Analysis of the Reaction between LimOOH and
Boc-Tyr-OMe or Bz-N-His-OMe. The amino acid reactions and
control mixtures were analyzed with LC/MS using an HPLC system
(Shimadzu, Japan) consisting of two pumps (LC-10 ADvp), a
degasser (DGU-14 A), a system controller (SCL-10 Avp), and an
autoinjector (SIL-10ADvp) coupled to a triple quadrupole API 2000
mass spectrometer (Applied Biosystems/MDS Sciex, Canada).
Typically, 20 µL of a 500 µL reaction or control mixture was
injected onto a C18 100 × 2.1 mm YMC HPLC column and
separated at a flow rate of 250 µL/min. The mobile phases were
(A) 1 mM formic acid in water and (B) 1 mM formic acid in
acetonitrile. A 20 min linear gradient from 10% B to 60% was
employed, followed by an increase to 100% B in 1 min. The
analytes were ionized by positive ESI with the following settings:

Peptide Hydroperoxide Adducts in ACD
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 205
amount of angiotensin I. No adducts could be detected with
angiotensin I and CumOOH or CycHexOOH.
The angiotensin I-LimOOH adduct at m/z 1444 corresponds
to the molecular weight of angiotensin I+148. The reaction
mixture containing this adduct was subjected to LC/MS/MS to
elucidate which amino acid(s) in the angiotensin I sequence had
been modified. However, the amount of +148 adduct was too
low to obtain fragmentation information. In order to increase
the concentration of adduct for this analysis, attempts were made
to improve the adduct yield by increasing the amount of Lim-
2-OOH in the reaction mixture up to 1:100:0.1 (peptide/
hydroperoxide/Fe(III)TPPCl), by varying the reaction temper-
ature (37, 25, and 4 °C) and increasing the reaction time (5 h,
24 h, and 14 days). However, no increase in adduct amount
was obtained. Therefore, the angiotensin I reaction mixture was
digested by trypsin, chymotrypsin, and proteinase K, respec-
tively, and then subjected to MALDI/TOF-MS. The MALDI/
TOF data of these digestions are shown in Table 1.
Angiotensin I-Hydroperoxide Polymerization Products. A
decrease in the amount of angiotensin I was observed for all
three hydroperoxide reaction mixtures in the presence of iron
porphyrin; however, it was not as pronounced as in the insulin
reaction (described below). The [M + H]+ ion of angiotensin I
remained the main peak in the mass spectra from all reaction
mixtures even after 3 h (Figure 2a). Typical results for all three
hydroperoxide reactions are represented by the results from the
experiments with Lim-2-OOH (Figure 2b), where singly charged
peptide dimer and trimer peaks are observed at m/z 2590 and
3884, respectively. These molecular weights correspond to the
loss of two hydrogens for each formed angiotensin-angiotensin
bond. These oligomers were formed at detectable levels im-
mediately after mixing the reagents (data not shown). No
haptenated oligomers were observed in the reaction mixture.
In the control samples of angiotensin I, neither adducts,
polymers, nor reduction in peak intensity over time was detected.
Figure 2. MALDI-TOF mass spectra of products from reactions
between angiotensin I, Lim-2-OOH, and Fe(III)TPPCl (1:10:0.1 molar
ratio) after 3 h of reaction time. Data were acquired at instrument
settings of m/z range (a) 700-9000 or (b) 1400-9000. The sample in
a was desalted by ZipTip C18 prior to analysis. The asterisk indicates
unmodified angiotensin I at m/z 1296.55. Arrows indicate the angio-
tensin I+148 adduct at m/z 1444.66. The angiotensin I dimer was
detected at m/z 2590.59 and angiotensin I trimer at (average) m/z
3884.51.
Lim-2-OOH and Boc-Tyr-OMe or Bz-N-His-OMe Ad-
ducts. The amino acids tyrosine and histidine were individually
reacted with Lim-2-OOH in the presence of Fe(III)TPPCl, and
the reaction products were investigated for an amino acid +148
adduct by LC/MS. Both the amine and carboxyl ends of the
amino acids were protected by derivatization to mimic their
structure and reactivity as part of a protein chain. No +148 or
other adducts were detected for the tyrosine, instead significant
tyrosine dimerization was observed (data not shown). For
histidine, a +148 adduct was detected at relatively high amounts
in the presence of Fe(III)TPPCl, and this adduct was not detected
at the histidine control samples. Mass spectra of unreacted
histidine and the histidine +148 adduct are shown in Figure 3.
Besides the +148 adduct, two peaks corresponding to histidine
+150 were detected. The intensities of these peaks were
approximately 10 times higher than that for the +148 adduct
(data not shown).
prior to MALDI/TOF-MS analysis, following instructions from the
manufacturer. A matrix solution of 4 µL (saturated HCCA in 1:1
(v/v) ACN:0.1% TFA) was used to elute the material from the
ZipTips, followed by 1 µL of 100% ACN to ensure maximum
peptide recovery. MALDI/TOF-MS analysis was performed as
described above.
Results
Control samples were prepared for every reaction mixture,
excluding either hydroperoxide or Fe(III)TPPCl, or both. In the
absence of Fe(III)TPPCl, no appreciable reaction of Lim-2-OOH
was observed over 24 h (data not shown).
Angiotensin I-Carvone Reaction Products. It has previ-
ously been shown that Lim-2-OOH is converted into the hapten
carvone in the presence of iron porphyrin (12). Thus, the reaction
pathway for the formation of the angiotensin I +148 adduct
with Lim-2-OOH could possibly proceed directly from the
formation of carvone. To determine if the angiotensin I +148
adduct is unique for Lim-2-OOH or if it is developed via
carvone, a set of additional experiments with and without
Fe(III)TPPCl were made in the presence of carvone instead of
Lim-2-OOH. No adducts or polymers were detected, neither in
the carvone reaction mixture nor in the control after 3 h at 37
°C (data not shown).
Stability of Lim-2-OOH in the Presence of Fe(III)TPPCl.
The decomposition of Lim-2-OOH in the presence of Fe(III)T-
PPCl was explored by 1H NMR. It showed that all Lim-2-OOH
was converted into carvone (∼70%) and carveol (∼30%) within
100 min at 25 °C (see Supporting Information). Therefore,
removal of unreacted hydroperoxide before peptide analysis in
the reaction mixtures was determined to be unnecessary.
Angiotensin I-Lim-2-OOH Adducts. Of the three hydrop-
eroxides, the only angiotensin I-hydroperoxide adduct detected
was in the reaction mixture with Lim-2-OOH, at m/z 1444
(Figure 2a). From the peak areas in the MALDI spectra, this
adduct was estimated to be formed at maximum 1% of the total

206 Chem. Res. Toxicol., Vol. 23, No. 1, 2010
Redeby et al.
Table 1. MALDI-TOF-MS Analysis of Digested Products from the Reaction between Angiotensin I and Lim-2-OOH in the
Presence of Fe(III)TPPCla
enzyme
trypsin
peptide sequence
theoretical [M + H]+
detected [M + H]+
VYIHPFHL
1025.6
1025.5
1173.3
1189.3
288.3
VYIHPFHL(+hapten)
VYIHPFHL(+hapten+O)
NR
1025.6 + 148.1 ) 1173.7
1025.6 + 148.1 + 16 ) 1189.7
289.2
NRVYIHPF
NRVYIHPF
NRVYIHPF(+hapten)
RVYIHPFH
1045.6
1045.6
1045.6 + 148.1 ) 1193.7
1068.6
1046.5
1046.3
1194.5
1068.4
784.4
chymotrypsin
proteinase K
RVYIHP
784.4
NRVY
IHPF (or PFHL)
PF
551.3
513.3
263.1
1045.6
552.2
513.2
263.1
1046.6
1194.6
1210.6
552.3
513.3
401.1
NRVYIHPF
NRVYIHPF(+hapten)
NRVYIHPF(+hapten+O)
NRVY
IHPF (or PFHL)
HPF (or PFH)
1045.6 + 148.1 ) 1193.7
1045.6 + 148.1 + 16 ) 1209.7
551.3
513.3
400.2
a The full sequence of angiotensin I is NRVYIHPFHL.
Insulin-Hydroperoxide Reaction Products. Aliquots from
the reaction mixtures of insulin with 100 molar excess of the
three hydroperoxides were taken out at different times to monitor
the product formation. Figure 4 shows the mass spectra
immediately after mixing the reagents (Figure 4a, c, and e) and
after 3 h (Figures 4b, d, and f). The arrows indicate one broad
peak or clusters of just detectable peaks, about m/z 5860-5920,
corresponding to the addition of one hydroperoxide molecule
to the insulin. For all three hydroperoxides, the insulin signal
was markedly decreased in the reaction mixtures over time
compared to that in the control mixtures. After 3 h of reaction
time, no or a very small insulin peak was detected using
extensive sweet spot searching and high laser intensity (Figure
4b, d, and f). This means that only barely detectable amounts
of unreacted insulin remained in the reaction mixture after 3 h.
Furthermore, the signal corresponding to the mass of the
insulin-hydroperoxide adducts decreased or disappeared after
3 h (Figure 4b, d, and f), which indicates that these products
are consumed in a further reaction. As in the reactions with
angiotensin I, neither adduct formation nor decrease in insulin
signal was observed in any of the control samples. The reaction
mixtures were also analyzed with MALDI/TOF-MS both at
higher and lower m/z ranges, but no signals could be detected.
There were no signs of either large molecular weight polymers
or smaller degradation products.
Native PAGE was used to further investigate the insulin
reaction products when they were still in oxidized and nonde-
natured forms (Figure 5). While distinctive bands of intact
insulin were observed for all controls, the reaction mixtures (with
both Fe(III)TPPCl and one of the three hydroperoxides)
exhibited a continuous smear in their individual lanes, revealing
an apparently broad range of high molecular weight products.
Part of the sample in the reaction mixtures did not even enter
past the stacking part of the gel (e.g., Figure 5, top of the lane
with insulin + Fe(III)TPPCl + CumOOH). These insulin
reaction products are likely aggregates and/or polymers which
have molecular weights above 200 kDa.
only covalent structures resistant to reduction remained. The
only discrete protein bands distinguished on the gel cor-
responded to the comigration of the two intact and reduced
monomer insulin R and ꢀ chains (Figure 6). Thus, the reaction
products did not include any covalently bound insulin oligomers
resistant to reduction. However, when both Lim-2-OOH and
Fe(III)TPPCl were included in the reactions, a faint broadening
of the insulin R-ꢀ-bands of up to approximately an addition
of 3000 Da to the insulin chains could be seen in the lanes for
all molar ratios. Consequently, in this case there were covalently
modified insulin reaction products of a polydisperse nature
which remained after both denaturation and reduction. MALDI/
TOF-MS analyses of the same reduced samples did not provide
any additional information since the individual species within
the range of the modified insulin were at amounts below the
detection limit.
Discussion
Three hydroperoxides, Lim-2-OOH, CumOOH, and Cy-
cHexOOH, were reacted in darkness with angiotensin I or the
protein insulin using Fe(III)TPPCl as a radical initiator. The
results showed that there are two possible routes of reactions:
peptide-hapten binding and/or polymerization but that no
reaction will take place without the addition of the radical
initiator.
Angiotensin I. When angiotensin I was reacted with Lim-
2-OOH in the presence of Fe(III)TPPCl, a small but distinctive
peak at m/z 1444 was detected. This corresponds to the
formation of a +148 monohaptenated adduct (Figure 2a). There
were also peaks corresponding to dimers and trimers of
angiotensin I (Figure 2b), but no haptenated dimer or trimer
could be detected under the applied analytical conditions. The
same relative amounts of angiotensin I dimers and trimers were
also detected in the CycHexOOH and CumOOH reaction
mixtures (data not shown). However, monohaptenated adducts
could not be detected in the CycHexOOH and CumOOH
reaction mixtures.
The possible presence of covalently bound insulin oligomers
was examined by subjecting reaction mixtures of Lim-2-OOH/
insulin at molar ratios of 100:1, 10:1, and 1:1 to SDS-PAGE.
The denaturing conditions of SDS-PAGE would dissociate any
noncovalent insulin aggregate. Left to be separated on the gel
were covalently bound species, with the exception of, e.g.,
disulfide bonds, as the reducing conditions would ensure that
The amount of angiotensin I +148 adduct in the Lim-2-OOH
experiment did not permit any structure determination by LC/
MS/MS. Therefore, the angiotensin I reaction mixture was
digested by three different enzymes to elucidate which amino
acid had been modified. From the data in Table 1, it can be
concluded that the hapten was bound somewhere in the sequence

Peptide Hydroperoxide Adducts in ACD
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 207
Figure 4. MALDI-TOF mass spectra of products from reactions among
insulin, hydroperoxide, and Fe(III)TPPCl (1:100:0.1 molar ratio) using
the hydroperoxides Lim-2-OOH (a,b), CumOOH (c,d) or or Cy-
cHexOOH (e,f). Samples were analyzed immediately after mixing (a,
c, and e) and again after 3 h (b, d, and f). The asterisk indicates
unmodified insulin at m/z 5733 (average). Arrows indicate one broad
peak or clusters of several minor peaks, in the range of m/z 5860-5920.
Figure 5. Native PAGE of insulin/hydroperoxide/Fe(III)TPPCl (1:10:
0,1 molar ratio) reaction mixtures and control samples after 3 h of
reaction. I ) insulin, T ) Fe(III)TPPCl, C ) CumOOH, H )
CycHexOOH, and L ) Lim-2-OOH.
other amino acid side chains should not be reactive toward the
Lim-2-OOH hapten. In the trypsin and proteinase K digestions,
an oxidized hapten bound to the peptides was observed in the
mass spectra (Table 1). These oxidized adducts (angiotensin I
+148 +16) were not detected in the chymotrypsin digest or in
the nondigested sample. Thus, they were probably formed during
the trypsin and proteinase K digestions and not from the reaction
between Lim-2-OOH and angiotensin I.
When tyrosine and histidine were separately subjected to
reactions with Lim-2-OOH and any +148 adduct formation was
examined by LC/MS/MS, only the histidine reaction mixture
showed a positive result for a +148 adduct using LC/MS/MS
(Figure 3b). The adduct formed exhibited a high stability toward
MS/MS, requiring a collision energy as high as 40 V to fragment
the quasi-molecular ion. The high stability could be due to a
formed conjugated double bond within the structure. As shown
in Figure 3d, the most intense fragments from the +148 adducts
were the benzoyl ion (m/z 105) from the protected N-terminal,
m/z 362 that corresponds to the neutral loss of the methyl ester
Figure 3. LC/MS analysis of histidine/Lim-2-OOH/Fe(III)TPPCl (1:
10:0.1 molar ratio) reaction products. Extracted ion chromatograms of
(a) unmodified His at m/z 274 which has not reacted (indicated by an
asterisk) and (b) the His+148 adduct at m/z 422 (indicated by an arrow).
Mass spectra after MS/MS fragmentation of (c) unreacted histidine
(collision energy 30 V) and (d) the histidine +148 adduct (collision
energy 40 V).
VYIHPF, of which only tyrosine and histidine were considered
as likely candidates since pure hydrocarbon structures of the

208 Chem. Res. Toxicol., Vol. 23, No. 1, 2010
Redeby et al.
in angiotensin I is a bit more exposed due to the neighboring
proline, while the second histidine is sterically hindered by the
surrounding Phe and Leu side chains.
As the hydroperoxide had to be activated by Fe(III)TPPCl
to participate in any reaction, the involvement of a radical
mechanism was strongly suspected. To further investigate the
importance of a radical reaction involvement, a reaction between
angiotensin I and carvone (4, Figure 1) was studied. Carvone
is the major decomposition product of Lim-2-OOH and a known
hapten (15) which reacts with proteins through an electrophilic-
nucleophilic Michael addition mechanism. No peptide-hapten
adducts or polymers were detected in the reaction mixture of
carvone or its control. This shows that the reaction mechanism
for protein-hapten binding of Lim-2-OOH with angiotensin I
is not achieved through a Michael addition reaction. Thus, there
is a strong indication the angiotensin I +148 adduct and the
histidine +148 adduct are formed through a radical pathway.
Furthermore, the molecular weights of the dimer and trimer of
angiotensin I (Figure 2) correspond to the loss of two hydrogens
for each peptide-peptide bond formed rather than the loss of
water. It is, therefore, likely that the polymers are also formed
through a radical mechanism. This is supported by the observa-
tion that tyrosine forms dimers after exposure to hydrogen
peroxides and lipid hydroperoxides (16). Interestingly, hemo-
globin in red blood cells forms dimers and trimers through
dityrosine binding when exposed to hydrogen peroxide, and
similar to our results, these species are formed only in the
presence of both hydrogen peroxide and the iron-containing
heme group (17). In our experiment with protected tyrosine,
significant dimerization was also observed.
Figure 6. SDS-PAGE of insulin/Lim-2-OOH/Fe(III)TPPCl reaction
mixtures at three different molar ratios of Lim-2-OOH, 1:(100, 10, or
1):0.1 after 3 h of reaction. I ) insulin, T ) Fe(III)TPPCl, and L )
Lim-2-OOH. Control samples, without Fe(III)TPPCl (T) were pre-
formed for every Lim-2-OOH concentration.
Even though the amino acids involved in the binding of the
observed angiotensin I polymers are currently unidentified, it
is clear that the formation of the same molecular weight peptide
polymers is independent of the hydroperoxide used.
We have in previous animal experiments investigated if cross-
reactions between Lim-2-OOH, CycHexOOH, and CumOOH
could be obtained but found no such reactions (10). We have
also patch tested dermatitis patients with known contact allergy
to one hydroperoxide with a structurally different hydroperoxide
but found no increase in the number of positive reactions to
this second hydroperoxide compared to that obtained by testing
patients without contact allergy to hydroperoxides (10). In a
subsequent study, we observed that (4R)-4-isopropenyl-1-
methyl-2-cyclohexene-1-hydroperoxide was significantly more
sensitizing compared to Lim-2-OOH in experimental studies
and also gave more positive patch test reactions in dermatitis
patients with known contact allergy to oxidized limonene
containing both hydroperoxides (18). Thus, neither the human
data nor the animal studies support the hypothesis that hydro-
peroxides form nonspecific antigens but rather indicate that
hydroperoxides form specific immunogenic complexes in con-
tact allergy. Therefore, the polymerization alone is unlikely to
account for the hydroperoxide specificity in the hapten forma-
tion, and its role in developing ACD remains uncertain. Instead,
the specificity of hydroperoxides must, therefore, be due to
haptenated adducts such as angiotensin I +148 or similar, with
most of the original hydroperoxide structure still attached to
peptide or protein.
Figure 7. Possible structures of the [M + H]+ quasi-molecular ions,
m/z 424 and m/z 422, respectively, of the His+150 and His+148 species
identified in the reactions of Lim-2-OOH with Fe(III)TPPCl in the
presence Bz-N-His-OMe.
group (-60 Da from the C-terminal) and m/z 214, which is the
histidine skeleton. The fragmentation pattern in the low mass
region for the adduct is very similar to the spectrum of the
unreacted histidine in Figure 3c, indicating that the histidine
structure in the adduct is unchanged. It was also observed that
the histidine +150 adducts required less collision energy for
fragmentation than the +148 adduct. This indicates that the
+148 adduct, which has lost two hydrogens compared to the
+150 adduct, most likely has formed a conjugated double bond,
located in the terpene moiety (Figure 7).
From previously conducted radical reaction experiments with
protected cystein, only cystein adducts of molecular weights
+150 were observed when Lim-2-OOH was reacted in the
presence of Fe(III)TPPCl (13). These adducts correspond to the
addition of a carvone moiety to cystein.
Angiotensin I has two histidine units in its sequence; it is
interesting to note that all data obtained here show that only
the first histidine (from the N-terminal end) is haptenated by
Lim-2-OOH. This reactivity differs from what is observed for
angiotensin I in the nucleophilic addition reaction with styrene
oxide, where both histidines are equally reactive (14). This
difference in reactivity maybe caused by a variation in the
availability of the different histidines because of the dissimilar
steric environment of the reacting histidine. The first histidine
Insulin. To further test the reactivity of hydroperoxides and
to test the possibility to form immunogenic complexes with
proteins, the hydroperoxides were reacted with insulin in the
presence of a radical initiator (Fe(III)TPPCl). Although, no
adduct formation with CumOOH and CycHexOOH was ob-
served for angiotensin I, peaks corresponding to the binding of

Peptide Hydroperoxide Adducts in ACD
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 209
one hydroperoxide moiety to one insulin were detected for all
three investigated hydroperoxides immediately after mixing
(Figure 4a, c, and e). These data suggest that a unique
hydroperoxide-insulin adduct is formed for each specific
hydroperoxide. This initial adduct formation is very fast, and
over time, the adducts disappear, and a substantial amount of
polymerization of insulin is detected. The polymerization
products formed with insulin could not be detected using
MALDI/TOF-MS, indicating that the main products are highly
polydisperse and/or difficult to ionize. This was confirmed by
the native PAGE analysis (Figure 5), where in each reaction
mixture containing both hydroperoxide and Fe(III)TPPCl, the
original insulin band is lost in favor of a continuous smear.
Evidently, insulin displays a range of large molecular weight
reaction products. Insulin has the ability to form cytotoxic
noncovalent aggregates (19) and as seen in Figure 6, most of
the original insulin bands are recovered at reducing and
denaturing conditions. Whether the large molecular weight
insulin products are polymers and/or noncovalent aggregates
has not been further investigated. However, most of the original
R- and ꢀ-chains of insulin were recovered under the reducing
conditions of the SDS-PAGE (vide infra), indicating that
polymerization is in part caused by the formation of disulfide
bridges.
Further indications of insulin modifications of up to an
additional molecular weight of about 3000 Da for the R- and
ꢀ-chains of insulin can be observed as weak smearing of the
peptide bands in the SDS-PAGE reaction mixtures containing
Lim-2-OOH and Fe(III)TPPCl (Figure 6). These modifications
of the R- and ꢀ-chains would correspond to one or more Lim-
2-OOH moieties. Analogous to angiotensin I, the histidines of
the insulin ꢀ-chain could have been modified. There is also a
possibility that the redox conditions of the models systems could
generate free cystein or thiyl radicals that would react with Lim-
2-OOH or its major degradation product carvone. In a previously
conducted study, it was shown that cystein can be modified by
Lim-2-OOH via a thiol-ene radical mechanism (13).
via a radical reaction since hapten adducts were observed only
in the presence of a radical initiator. It is thus reasonable to
expect that intact hydroperoxide molecules can penetrate into
the skin and remain there until activated, and then react via a
radical mechanism. On the basis of the present observations
and the results from a previous study (10) where no cross-
reactivity between the investigated hydroperoxides in animal
experiments and clinical studies was observed, we conclude that
the specificity of the immunogenic response in ACD is caused
by specific hapten-protein complexes formed from the hydro-
peroxides. In parallel to the specific reactions, a polymerization
reaction was detected for angiotensin I and insulin for all
hydroperoxides. The role of the polymerization reaction in ACD
is presently unclear but may play an important role in the overall
toxicity of hydroperoxides in the skin.
Acknowledgment. We thank Johan Redeby for technical
assistance with the figures. We are also grateful to Per-Olof
Edlund at Biovitrum, Stockholm, for running some of the LC/
MS/MS analyses and for helpful discussions. This work was
performed within the Go¨teborg Science Centre for Molecular
Skin Research in cooperation with the Department of Analytical
Chemistry, Stockholm University.
Supporting Information Available: Reaction of Lim-2-OOH
in the presence of Fe(III)TPPCl as followed by 1H NMR. This
material is available free of charge via the Internet at http://
pubs.acs.org.
References
(1) Divkovic, M., Pease, C. K., Gerberick, G. F., and Basketter, D. A.
(2005) Hapten-protein binding: from theory to practical application
in the in Vitro prediction of skin sensitization. Contact Dermatitis 53,
189–200.
(2) Karlberg, A. T., Andresen Bergstro¨m, M., Bo¨rje, A., Luthman, K.,
and Nilsson, J. L. G. (2008) Allergic contact dermatitis: formation,
structural requirements and reactivity of skin sensitizers. Chem. Res.
Toxicol. 21, 53–69.
(3) Thyssen, J. P., Linneberg, A., Menne, T., and Johansen, J. D. (2007)
The epidemiology of contact allergy in the general population:
prevalence and main findings. Contact Dermatitis 57, 287–299.
(4) Rastogi, S. C., Heydorn, S., Johansen, J. D., and Basketter, D. A.
(2001) Fragrance chemicals in domestic and occupational products.
Contact Dermatitis 45, 221–225.
(5) Matura, M., Goossens, A., Bordalo, O., Garcia-Bravo, B., Magnusson,
K., Wrangsjo, K., and Karlberg, A. T. (2002) Oxidized citrus oil (R-
limonene): a frequent skin sensitizer in Europe. J. Am. Acad. Dermatol.
47, 709–714.
It should be noted that any adduct formed in these model
systems does not necessarily correspond to the immunogenic
complex formed in vivo. According to theory, the formation of
a stable immunogenic complex is required to elicit an immune
response. This is accomplished by the covalent binding of the
hapten to a protein (20, 21). Even if the structures of the
modified insulin are unknown, the results strongly suggest that
Lim-2-OOH in the presence of a radical initiator can modify a
protein and thus form a stable immunogenic complex that
eventually results in an immune response.
(6) Karlberg, A. T., Boman, A., and Melin, B. (1991) Animal experiments
on the allergenicity of d-limonene: the citrus solvent. Ann. Occup.
Hyg. 35, 419–426.
One can speculate that the polymerization reaction is part of
an overall radical reaction initiated by the presence of hydro-
peroxide and an initiator such as Fe(III)TPPCl. Large amounts
of radicals are formed under these conditions (12). Elevated
levels of radicals formed from hydroperoxides in the skin will
increase the oxidative stress by the consumption of antioxidants
(22, 23). Furthermore, the radical originating from hydroper-
oxides would increase the possibility of radical formation on
macromolecules such as proteins. Reaction of protein radicals
with molecular oxygen results in the formation of reactive
oxygen species (ROS) that will deplete antioxidant reserves even
further. Thus, the action of hydroperoxides in ACD results in
increased oxidative stress due to the formation of high amounts
of radicals, which would aid the progression of ACD.
(7) Matura, M., Sköld, M., Börje, A., Andersen, K. E., Bruze, M., Frosch,
P., Goossens, A., Johansen, J. D., Svedman, C., White, I. R., and
Karlberg, A. T. (2005) Selected oxidized fragrance terpenes are
common contact allergens. Contact Dermatitis 52, 320–328.
(8) Sköld, M., Börje, A., Matura, M., and Karlberg, A. T. (2002) Studies
on the autoxidation and sensitizing capacity of the fragrance chemical
linalool, identifying a linalool hydroperoxide. Contact Dermatitis 46,
267–272.
(9) Sköld, M., Börje, A., Harambasic, E., and Karlberg, A. T. (2004)
Contact allergens formed on air exposure of linalool. Identification
and quantification of primary and secondary oxidation products and
the effect on skin sensitization. Chem. Res. Toxicol. 17, 1697–1705.
(10) Bråred Christensson, J., Matura, M., Ba¨cktorp, C., Bo¨rje, A., Nilsson,
J. L. G., and Karlberg, A. T. (2006) Hydroperoxides form specific
antigens in contact allergy. Contact Dermatitis 55, 230–237.
(11) Lepoittevin, J. P., and Karlberg, A. T. (1994) Interactions of allergenic
hydroperoxides with proteins: a radical mechanism. Chem. Res.
Toxicol. 7, 130–133.
(12) Johansson, S., Gime´nez-Arnau, E., Grøtli, M., Karlberg, A.-T., and
Bo¨rje, A. (2008) Carbon and oxygen centered radicals are equally
important haptens of allylic hydroperoxides in allergic contact der-
matitis. Chem. Res. Toxicol. 21, 1536–1547.
Conclusions
We propose that the formation of specific immunogenic
hapten-protein complexes from hydroperoxides can proceed

210 Chem. Res. Toxicol., Vol. 23, No. 1, 2010
Redeby et al.
(13) Johansson, S., Redeby, T., Altamore, T. M., Nilsson, U, and Bo¨rje,
A. (2009) Mechanistic proposal for formation of specific immunogenic
complexes via a radical pathway: a key step in allergic contact
dermatitis to olefinic hydroperoxides. Chem. Res. Toxicol. 22, 1774–
1781.
(14) Ferranti, P., Carbone, V., Sannolo, N., Fiume, I., Milone, A., Ruoppolo,
M., Gallo, M., and Malorni, A. (1992) Study of interaction of styrene
oxide with angiotensin by mass-spectrometry. Carcinogenesis 13,
1397–1401.
(15) Nilsson, A., Ga¨fvert, E., Salvador, L., Luthman, K., Bruze, M.,
Gruvberger, B., Nilsson, J. L. G., and Karlberg, A. T. (2001)
Mechanism of the antigen formation of carvone and related R,
ꢀ-unsaturated ketones. Contact Dermatitis 44, 347–356.
(16) Giulivi, C., Traaseth, N. J., and Davies, K. J. A. (2003) Tyrosine
oxidation products: analysis and biological relevance. Amino Acids
25, 227–232.
(17) Giulivi, C., and Davies, K. J. A. (2001) Mechanism of the formation
and proteolytic release of H2O2-induced dityrosine and tyrosine
oxidation products in hemoglobin and red blood cells. J. Biol. Chem.
276, 24129–24136.
(19) Grudzielanek, S., Velkova, A., Shukla, A., Smirnovas, V., Tatarek-
Nossol, M., Rehage, H., Kapurniotu, A., and Winter, R. (2007)
Cytotoxicity of Insulin within its Self-assembly and Amyloidogenic
Pathways. J. Mol. Biol. 370, 372–384.
(20) Rustemeyer, T., van Hoogstraten, I. M. V., von Blomberg, B. M. E.,
and Scheper, R. J. (2006) Mechanisms in Allergic Contact Dermatitis,
in Contact Dermatitis (Frosch, P. J., Menne´, T., and Lepoittevin, J. P.,
Eds.) pp 11-43, Springer-Verlag, Berlin, Germany.
(21) Lepoittevin, J. P., and Roberts, D. W. (1998) Hapten-Protein Interac-
tions, in Allergic Contact Dermatitis, The Molecular Basis (Lepoit-
tevin, J. P., Basketter, D., Goossens, A., and Karlberg, A. T., Eds.)
pp 81-111, Springer, Berlin, Germany.
(22) Murray, A. R., Kisin, E. R., Kommineni, C., Vallyathan, V.,
Castranova, V., and Shvedova, A. A. (2007) Pro/antioxidant status
and AP-1 transcription factor in murine skin following topical exposure
to cumene hydroperoxide. Carcinogenesis 28, 1582–1588.
(23) Shvedova, A. A., Kisin, E. R., Murray, A. R., Kommineni, C.,
Castranova, V., Mason, R. P., Kadiiska, M. B., and Gunther, M. R.
(2002) Antioxidant balance and free radical generation in vitamin
E-deficient mice after dermal exposure to cumene hydroperoxide.
Chem. Res. Toxicol. 15, 1451–1459.
(18) Christensson, J. B., Johansson, S., Hagvall, L., Jonsson, C., Börje,
A., and Karlberg, A. T. (2008) Limonene hydroperoxide analogues
differ in allergenic activity. Contact Dermatitis 59, 344–352.
TX9003352