Chemical reactivity and skin sensitization potential for benzaldehydes: Can Schiff base formation explain everything?

Article abstract of DOI:10.1021/tx300278t

Skin sensitizers chemically modify skin proteins rendering them immunogenic. Sensitizing chemicals have been divided into applicability domains according to their suspected reaction mechanism. The widely accepted Schiff base applicability domain covers aldehydes and ketones, and detailed structure-activity-modeling for this chemical group was presented. While Schiff base formation is the obvious reaction pathway for these chemicals, the in silico work was followed up by limited experimental work. It remains unclear whether hydrolytically labile Schiff bases can form sufficiently stable epitopes to trigger an immune response in the living organism with an excess of water being present. Here, we performed experimental studies on benzaldehydes of highly differing skin sensitization potential. Schiff base formation toward butylamine was evaluated in acetonitrile, and a detailed SAR study is presented. o-Hydroxybenzaldehydes such as salicylaldehyde and the oakmoss allergens atranol and chloratranol have a high propensity to form Schiff bases. The reactivity is highly reduced in p-hydroxy benzaldehydes such as the nonsensitizing vanillin with an intermediate reactivity for p-alkyl and p-methoxy-benzaldehydes. The work was followed up under more physiological conditions in the peptide reactivity assay with a lysine-containing heptapeptide. Under these conditions, Schiff base formation was only observable for the strong sensitizers atranol and chloratranol and for salicylaldehyde. Trapping experiments with NaBH₃CN showed that Schiff base formation occurred under these conditions also for some less sensitizing aldehydes, but the reaction is not favored in the absence of in situ reduction. Surprisingly, the Schiff bases of some weaker sensitizers apparently may react further to form stable peptide adducts. These were identified as the amides between the lysine residues and the corresponding acids. Adduct formation was paralleled by oxidative deamination of the parent peptide at the lysine residue to form the peptide aldehyde. Our results explain the high sensitization potential of the oakmoss allergens by stable Schiff base formation and at the same time indicate a novel pathway for stable peptide-adduct formation and peptide modifications by aldehydes. The results thus may lead to a better understanding of the Schiff base applicability domain.

Full text of DOI:10.1021/tx300278t

Article
pubs.acs.org/crt
Chemical Reactivity and Skin Sensitization Potential for
Benzaldehydes: Can Schiff Base Formation Explain Everything?
Andreas Natsch,* Hans Gfeller, Tina Haupt, and Gerhard Brunner
Givaudan Schweiz AG, Ueberlandstrasse 138, CH-8600 Duebendorf, Switzerland
S
* Supporting Information
ABSTRACT: Skin sensitizers chemically modify skin proteins
rendering them immunogenic. Sensitizing chemicals have been
divided into applicability domains according to their suspected
reaction mechanism. The widely accepted Schiff base
applicability domain covers aldehydes and ketones, and
detailed structure−activity-modeling for this chemical group
was presented. While Schiff base formation is the obvious reaction pathway for these chemicals, the in silico work was followed up
by limited experimental work. It remains unclear whether hydrolytically labile Schiff bases can form sufficiently stable epitopes to
trigger an immune response in the living organism with an excess of water being present. Here, we performed experimental
studies on benzaldehydes of highly differing skin sensitization potential. Schiff base formation toward butylamine was evaluated in
acetonitrile, and a detailed SAR study is presented. o-Hydroxybenzaldehydes such as salicylaldehyde and the oakmoss allergens
atranol and chloratranol have a high propensity to form Schiff bases. The reactivity is highly reduced in p-hydroxy benzaldehydes
such as the nonsensitizing vanillin with an intermediate reactivity for p-alkyl and p-methoxy-benzaldehydes. The work was
followed up under more physiological conditions in the peptide reactivity assay with a lysine-containing heptapeptide. Under
these conditions, Schiff base formation was only observable for the strong sensitizers atranol and chloratranol and for
salicylaldehyde. Trapping experiments with NaBH3CN showed that Schiff base formation occurred under these conditions also
for some less sensitizing aldehydes, but the reaction is not favored in the absence of in situ reduction. Surprisingly, the Schiff bases
of some weaker sensitizers apparently may react further to form stable peptide adducts. These were identified as the amides
between the lysine residues and the corresponding acids. Adduct formation was paralleled by oxidative deamination of the parent
peptide at the lysine residue to form the peptide aldehyde. Our results explain the high sensitization potential of the oakmoss
allergens by stable Schiff base formation and at the same time indicate a novel pathway for stable peptide-adduct formation and
peptide modifications by aldehydes. The results thus may lead to a better understanding of the Schiff base applicability domain.
quantify their sensitization potency based on in silico SAR-
modeling.6 In one such paradigm, chemicals are first classified
into so-called applicability domains according to their suspected
reaction mechanism, which is inferred from functional groups
present in the molecules.7 Experimental or in silico reactivity
indices and physicochemical parameters for chemicals within an
applicability domain are then used to model their in vivo
sensitization potential.8
One widely accepted applicability domain contains all
chemicals with the theoretical capability to form Schiff
bases.7,9,10 The sensitization potential of these molecules has
been modeled based on the Taft σ* and Σσ* constants used as
reactivity indices and cLogP as physicochemical parameter, and
a QSAR was developed which predicts the sensitization
potential of the chemicals in this applicability domain nicely.9
However, this was not followed up by experimental work to
prove Schiff base formation of these chemicals toward peptides
or proteins. However, using the peptide reactivity approach,
Gerberick et al.11 had screened a significant number of
aldehydes and ketones for their reactivity toward a test peptide
INTRODUCTION
■
Skin sensitizers are reactive exogenous molecules with the
ability to chemically modify skin proteins.1 The modified
proteins represent novel nonself epitopes to the immune
system and thus trigger a specific T-cell mediated immune
response. On the basis of this mechanistic understanding, the
relationship between reactivity of molecules toward proteins
and their sensitization potential has attracted increased interest
and was used to develop experimental assays to predict the skin
sensitization potential of chemicals to replace the current
animal tests.2 Gerberick et al.3 developed a peptide depletion
assay using different heptapeptides. This assay, later coined the
DPRA (direct peptide reactivity assay), has gone through
thorough prevalidation at ECVAM and it gives a rapid overall
estimate of reactivity of test chemicals toward peptides.
Chipinda et al.4 used nitrobenzenthiol as a simple surrogate
for a reactive cysteine in a peptide and developed a kinetic
profiling approach to compare sensitizers. A large database on
glutathione depletion has also been used to establish a
correlation between reactivity and skin sensitization.5 The
link between sensitization and reactivity has also led to the
development of a number of in silico approaches to predict
which chemicals are reactive and thus sensitizers and to
Received: June 18, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
Figure 1. Structure of the tested benzaldehydes. In parentheses is shown the sensitization potential as an EC3 value in the LLNA or, if available,
other information regarding the skin sensitization potential. For 13, 18, and 19, no data regarding the skin sensitization potential are available, but
these chemicals were included to complete the SAR study for reactivity.
with a lysine residue. The sensitizing aldehydes α-hexyl
cinnamic aldehyde, hydroxycitronellal, lilial, cyclamen aldehyde,
benzaldehyde, and farnesal did not lead to significant peptide
depletion,11 although the reactions were run at pH 10.2 to keep
the lysine residue in its unprotonated form and in the presence
of a large excess of test chemical (50-fold over peptide).
Similarly, for the same aldehydes we could not observe adduct
formation in a peptide reactivity assay with a heptapeptide
containing both Cys and Lys residues.12 These chemicals only
promoted the formation of disulfide bridges at the Cys residue.
However, under much more concentrated reaction conditions
(80 mM peptide and 200 mM test chemical), Reichhardt et
al.13 reported Schiff base formation with the ε-NH2-group of
Lys-Tyr for the two aldehydes formaldehyde and benzaldehyde.
Furthermore, supplementary information from Aleksic et al.14
reports Schiff base formation in a peptide reactivity assay for
cyclamen aldehyde, hydroxycitronellal, benzaldehyde, and α-
hexyl cinnamic aldehyde. The test conditions in this assay were
selected to favor Schiff base formation by applying a 100-fold
excess of test chemical and pH 10. Still, in these cases peptide
depletion was only marginal, and the reaction was monitored
by the more sensitive approach of following adduct formation.
Although Schiff base formation is an obvious reaction
mechanism for small aldehydes to modify skin proteins, there
is an open question which to our surprise has attracted little
attention: In water, Schiff base formation is a reversible reaction
reaching an equilibrium, and the Schiff base would be expected
to persist only as long as the free aldehyde is present in
significant quantities. Since modified peptides need to be
transported by migrating dendritic cells to local lymph nodes to
trigger the immune response, and since the locally and topically
applied aldehyde will not be present all along the migration
route and in the lymph node, it is not inherently clear how
Schiff bases could form sufficiently stable epitopes which are
not hydrolyzed prior to T-cell presentation.
In summary, there is a certain discrepancy between the
theoretical and in silico work postulating Schiff base formation
as a key mechanism for skin sensitization by aldehydes and
ketones and the experimental data on stable adduct formation
by the Schiff base mechanism in the now widely used peptide
reactivity assays such as the DPRA.
Benzaldehydes (BAs) represent an interesting class of
chemicals to study with respect to their skin sensitization
potential. This class comprises chloratranol (7) and atranol (8),
the allergens in oakmoss, for which very low thresholds for
elicitating immune responses in sensitized patients have been
reported and which are the true culprit of relatively frequent
positive patch test reactions of dermatitis patients to oakmoss
absolute.15,16 At the other extreme are vanillin (3) and
ethylvanillin (5), which are widely used in flavor and fragrance
compositions and which are generally considered nonsensitizers
in animal tests and in humans. BA (1) was for a long time
B
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
solution of fluram (Fluka, Buchs, Switzerland). The fluorescence due
to the reaction of the free amine groups with fluram was measured
(bottom-read fluorescence; excitation at 381 nm, emission at 470 nm;
Flexstation, Molecular Devices).
Peptide Reactivity Experiments. The Lys-peptide (Ac-RFAA-
KAA, MW 775.4) was obtained from Genscript Inc. (Piscataway, NJ,
USA). It has a purity of 98.1%. It was used at a final concentration of
0.5 mM in all experiments. All peptide reactivity tests at pH 7.5 were
run at 36 °C in 75% phosphate buffer (20 mM) and 25% acetonitrile
with shaking at 150 rpm. Different concentrations of test chemicals
dissolved in acetonitrile were used as specified in the Results section.
In order to perform in situ reduction of formed Schiff bases, the
reactions were performed in the presence of 50 mM sodium
cyanoborohydride (NaBH3CN).
To determine chemical modifications of peptides, reactions were
carried out in a final volume of 1 mL in HPLC vials. LC/MS analysis
was performed as described before on a Finnigan LCQ classic mass
spectrometer (Thermo Finnigan, San Jose, CA, U.S.A.) operated in
the ESI(+) mode.12 The ion source was operated at 5 kV, the sheath
gas flow rate was set to 60 arb and the aux gas flow rate to 20 arb. The
temperature of the capillary was kept at 220 °C. Mass spectra were
recorded from 400 to 2000 amu. A ZORBAX Eclipse XDB-C18
column, 2.1 mm ID, 150 mm, 5-Micron (Agilent Technologies) was
used. The mobile phase consisted of H2O and methanol, each
containing 0.1% formic acid (v/v). The solvent flow was 250 μL/min.
HPLC Separation of Modified Peptides. The peptide adducts
from the reaction of BA with the Lys-peptide were isolated by
reversed-phase HPLC on a YMC Pack ODS-AQ column (250 mm
length, 3.0 mm diameter, 3.0 μm particles) with a gradient from with
10%−100% methanol in 0.1% formic acid in water.
reported as a classical nonsensitizer based on animal tests but
was recently discovered to be a sensitizer in a human repeat
insult patch test (HRIPT; RIFM, personal communication).
This result contrasts with a published theoretical consideration
arguing for slow Schiff base formation for BA due to a
resonance loss effect.17 The broad span in in vivo sensitizing
potency in humans and the conflicting evidence from some
modeling studies made the BAs an interesting case to
experimentally study the Schiff base applicability domain in
more detail. Here, we thus report a SAR study for reactivity of
differently substituted BAs with butylamine followed by
detailed analysis of reactivity in a slightly modified DPRA
protocol, and we show that in water, Schiff base formation leads
to stable adducts only for o-hydroxy BAs. At the same time,
Schiff base formation appears to be the first reaction step in a
sequence involving oxidative events leading to stable peptide
adducts and peptide modifications by other benzaldehyde
derivatives and other sensitizing aldehydes.
MATERIALS AND METHODS
Caution: Some of the test chemicals are mild to strong skin sensitizers
in the LLNA, and any skin contact should be avoided.
■
Chemicals. All fragrance chemicals were obtained from Givaudan
Schweiz AG. Other BA derivatives were obtained from Sigma−Aldrich
(Buchs, Switzerland). Atranol and chloratranol were synthesized
according to the following procedure: O-Alkylation (MeI, K2CO3,
acetone)18 of 5-methyl-1,3-benzenediol (orcinol) led to orcinol
dimethyl ether (47%). Subsequent t-BuLi deprotonation followed by
quenching with DMF cleanly afforded atranol dimethyl ether (67%)
that was monochlorinated in good yield (85%) into chloroatranol
dimethyl ether using SO2Cl2 in CH2Cl2 (0 to 20 °C). De-O-
methylation of the two formyl diethers using BBr3 in CH2Cl2 (−70 to
20 °C) provided atranol and chloroatranol (13% and 80% yield,
respectively). The structures of all tested BAs are shown in Figure 1;
this Figure also gives the sensitization potential from the LLNA and if
available human evidence for sensitization risks. The LLNA data came
from different sources as indicated in Table S1 in Supporting
Information. The majority of LLNA and HRIPT data were kindly
provided by RIFM (Research Institute for Fragrance Materials).
Reactions with Butylamine Monitored by GC-FID and GC/
MS Analysis. The test chemicals (100 μL) dissolved at 2 mM in
acetonitrile were mixed with an equal volume of butylamine dissolved
at 20 mM in acetonitrile. After 6 h of incubation at 22 °C, the
reactions were extracted with 200 μL of sodium phosphate buffer (pH
7.5, 20 mM), 50 mg of NaCl, and 200 μL of MTBE. The organic
phase was analyzed with GC/MS. A combination of a Hewlett-Packard
6890 gas chromatograph and an Agilent 5973 MSD mass spectrometer
(70 eV EI mode, ion source temperature 230 °C, Agilent
Technologies, Santa Clara, CA, USA) was applied. A capillary column
of 12 m length × 0.22 mm i.d. coated with 0.25 μm of BPX5 (5%
phenyl polysilylphenylene-siloxane, SGE Analytical Science Pty Ltd.,
Ringwood, Australia) was used. Sample volumes of 1.0 μL were
injected in the splitless mode at an injector temperature of 230 °C.
Temperature of the column oven was initially set to 35 °C for 2 min
and subsequently increased by 20 °C/min to 240 °C, then by 35 °C/
min to 270 °C, and held for 3 min. For quantitative analysis, samples
were injected on an Agilent 6890N equipped with a flame-ionization
NMR Analysis of Modified Peptides. All NMR experiments
were performed on a Bruker Avance III 600 MHz FT-NMR
spectrometer equipped with a MicroCryoprobe 1.7 mm TCI (Triple
Inverse Cryoprobe) with a Z-Gradient coil. The sample (45 μg) was
dissolved in 35 μL of D2O 99.96% D (euriso-top, D215B). The
residual H2O signal was used as the chemical shift reference at 4.79
ppm. 1H NMR with presaturation: spectral width = 20.5 ppm,
acquisition time = 2.7 s, recycle time = 1.0 s, time domain = 64 K
points, size of real spectrum = 128 K points, window function =
Gaussian multiplication, line broadening −0.2, Gaussian factor 0.2,
number of scans = 16, attenuation for presaturation = 60.7 dB, and
temperature = 308 K. Gradient selected HSQC, spectral width = 10.8
ppm (1H) × 200.0 ppm (13C); acquisition time, 158 ms; and recycle
time, 2.0 s; time domain = 2 K (1H) × 512 (13C) points, size of real
spectrum = 1024 × 1024 points, window function = squared sine-bell
(phase shift = 2) in both dimensions, number of scans = 32, and
temperature = 308 K. Gradient selected HMBC: spectral width = 9.4
ppm (1H) × 200.0 ppm (13C), acquisition time = 181 ms, recycle time
= 2.0 s, time domain = 2 K (1H) × 512 (13C) points, size of real
spectrum = 2048 (1H) × 1024 (13C) points, window function = sine-
1
bell (phase shift = 4, H), squared sine-bell (phase shift = 2, 13C),
number of scans = 128, and temperature = 308 K. 1D-selective
TOCSY: spectral width = 9.4 ppm (1H) × 200.0 ppm (13C),
acquisition time = 2.7 s, recycle time = 2.0 s, time domain = 32 K
points, size of real spectrum = 32 K points, window function =
exponential multiplication, line broadening of 0.3, number of scans =
128, mixing time = 20−120 ms, and temperature = 298 K.
RESULTS
■
detector and a BGB-5 capillary column (BGB Analytik, Bockten,
̈
Switzerland) of 30 m length × 0.32 mm i.d. and 0.25 μm coating with
5% diphenyl−95% dimethylpolysiloxane. The column was held at 40
°C for 1 min and subsequently increased by 8 °C/min to 280 °C and
held at this temperature for 3 min.
Schiff Base Formation with Butylamine in Acetonitrile
and Analysis by GC and GC-FID. First, the reactivity of the
test chemicals dissolved at 1 mM in the presence of a 10-fold
excess of butylamine in ACN was studied. For 1, 2, 3, 5, 7, 8,
11, 12, 13, and 14, the reaction products were analyzed by GC-
MS. For all tested chemicals with the exception of 7 and 8, one
new chromatographic signal besides the parent BA signals was
observed. The molecular ions and the fragmentation patterns in
the associated mass spectra could be interpreted or directly
Reactions with Butylamine in a Dose−Response Analysis.
The test chemicals were dissolved at 10 mM in acetonitrile, and serial
dilutions down to 0.625 mM were prepared in microtiter plates. To
100 μL of these dilutions, an equal volume of butylamine dissolved at
0.5 mM in acetonitrile was added. Samples were incubated for 3 or 6 h
at 22 °C, and the reaction was stopped by adding 50 μL of a 4 mM
C
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
identified by reference mass spectra comparison (available for
parent BA 1, 3, and 5) as the respective Schiff base products.
For 7 and 8, only the Schiff base and no parent aldehyde were
detected, and as a byproduct, low levels of N-butylidene-1-
butanamine were observed.
Conversion rates after 6 h were determined by GC/FID for
the differentially substituted BAs in Figure 1 (Table 1). Schiff
Conversion rates for p-NO2-, p-alkyl- BAs, and BA were
inbetween.
Dose−Response Analysis to Quantify Reactivity with
Butylamine. The experimental procedure chosen above is well
suited to monitor formation of new products, but it is not ideal
for a quantitative structure−activity study of the reaction rate.
Since the reaction was stopped by organic extraction to
separate the free butylamine and since some of the test
chemicals have relatively high water solubility, the extraction
step is not quantitative. Furthermore, a large excess of
butylamine was used, which may, in general, facilitate the
reaction. We thus performed a dose−response analysis without
an extraction step and with determination of the free amine at
the end of a fixed incubation time with a fluorescent probe.
Dilutions of the test chemicals (0.312 mM to 5 mM) were
reacted with 0.25 mM butylamine, and the unbound butyl-
amine was quantified after 3 h and 6 h. Figure 2 presents the
logarithmic dose−response curves for compounds 1, 3, 4, 7, 10,
and 14. Table 2 presents all the data. For each chemical, linear
regression was performed from data plotted as shown in Figure
2. The slope, which is proportional to the rate constant,19 is
given in Table 2 for reactions stopped after 3 and 6 h. In
addition, the % depletion of 0.25 mM butylamine by 5 mM of
the test chemical is also listed, as these data may be easier to
read.
Table 1. Binding of Benzaldehydes by an Excess of
Butylamine
% free
aldehyde
% bound
aldehyde
a
no.
name
b
8
7
atranol
0
100
b
chloratranol
0
100
96.1 1.9
b
14 salicylaldehyde
benzaldehyde
3.9 1.9
b
1
35.9 11.9
47.6 1.6
54.4 3.4
59.7 6.9
61.5 2.1
81.6 2.9
82.2 2.1
64.1 11.9
52.4 1.6
45.6 3.4
40.3 6.9
38.5 2.1
18.4 2.9
17 2-bromo-5-hydroxy-benzaldehyde
15 4-nitrobenzaldehyde
10 4-methylbenzaldehyde
18 3-hydroxybenzaldehyde
16 p-chlorobenzaldehyde
19 3-methoxybenzaldehyde
17.8 2.1
b
2
9
3
5
4-hydroxybenzaldehyde
cuminic aldehyde
vanillin
84.7
86.5 4.4
88.1
2
15.3
2
13.5 4.4
b
2
11.9
2
b
b
ethylvanillin
92.8 3.7
94.2 1.2
95.1 0.1
96.4 0.1
96.8 1.7
7.2 3.7
5.8 1.2
4.9 0.1
3.6 0.1
3.2 1.7
In this study, the o-hydroxy BAs 14 (salicylaldehyde), 7
(atranol), and 8 (chloratranol) as well as 1 (benzaldehyde) had
the highest reactivity. Residual reactivity is observed for p-
methoxy BAs and no significant reactivity for p-hydroxy BAs
such as vanillin. The dramatic difference in reactivity between
these latter chemicals and chemicals such as atranol and
chloratranol supports the hypothesis that the differential ability
to bind amine nucleophiles may explain their highly differing
sensitization potential.
11 3-chloro-4-methoxybenzaldehyde
4
6
4-methoxybenzaldehyde
heliotropine
b
b
12 6-methoxynaphtalene-2-
carbaldehyde
13 veratrumaldehyde
97.9 1.9
2.1 1.9
a
1 mM of test chemical was reacted with 10 mM butylamine. The
peak areas from GC-FID analysis of the parent aldehyde and the
b
reaction product were integrated, and the sum was set to 100%. In
Peptide Reactivity Assays: Quantifying Schiff Base
Formation with Trapping by NaBH3CN. Next, we
performed a modified version of the DPRA assay to observe
Schiff base formation in a mainly aqueous milieu (75% buffer
and 25% ACN). We used a pH of 7.5 and a low amount of test
chemicals (1 mM) and trapped the formed Schiff bases by in
situ reduction with NaBH3CN. This low amount of test peptide
was selected since NaBH3CN trapping does facilitate the
reaction, and with higher amounts of test chemical, double
these cases, GC-MS analysis was used to verify that the structure of the
bound fraction indeed is the Schiff base.
base formation was almost quantitative for the o-hydroxy BAs
(i.e., the oakmoss allergens 7 and 8 as well as for 14)
Comparatively low levels of Schiff base formation were
observed for p-hydroxy BAs as compared to o-hydroxy BAs.
Low levels of Schiff bases were found for p-methoxy BAs.
Figure 2. Dose−response analysis of butylamine binding by benzaldehyde derivatives. Test chemical (0.3125−5 mM) was incubated with 0.25 mM
butylamine for (a) 3 h and (b) 6 h, and the unbound butylamine was determined with the amine-specific probe fluorescamine.
D
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
Table 2. Dose−Response Analysis of Butylamine-Binding by Benzaldehyde Derivatives
3 h incubation
6 h incubation
depletion at 5 mM
% depletion at 5 mM
slope
−3.20
−0.56
−0.49
−0.34
−0.31
−0.29
−0.26
−0.20
−0.21
−0.18
−0.04
−0.03
−0.03
−0.02
−0.02
slope
steep
14
1
salicylaldehyde
99.8 0.6
93.3 3.2
89.5 2.6
88 0.4
76.2 7.2
74.1 1.5
68.2 1.2
61.3 3.8
59.9 3.1
55.6 4.8
16.3 7.8
9.4 1.4
8.3 1.4
6.2 0.4
5.4 11.6
<2
97.2 3.6
95.4 2.6
91.3
benzaldehyde
−0.77
−0.52
−1.08
−0.52
−0.59
−0.50
−0.17
−0.51
−0.36
−0.07
−0.05
−0.05
−0.02
−0.07
−0.03
16
7
4-chlorobenzaldehyde
chloratranol
95.3 1.2
90.9 3.6
94.3 0.8
90.4 0.2
54.5 15
10
8
4-methylbenzaldehyde
atranol
19
15
17
9
3-methoxybenzaldehyde
4-nitrobenzaldehyde
2-bromo-5-hydroxy-benzaldehyde
cuminic aldehyde
90.8
0
81 5.9
25.9 11.1
15 1.3
14.9 3.4
8.8 1.6
26.4 18.6
9.4 4.4
<2
4
4-methoxybenzaldehyde
6-methoxynaphtalene-2-carbaldehyde
3-chloro-4-methoxybenzaldehyde
heliotropine
12
11
6
18
13
5
3-hydroxybenzaldehyde
veratrumaldehyde
no reaction
no reaction
no reaction
no reaction
ethylvanillin
<2
no reaction
no reaction
no reaction
2
4-hydroxybenzaldehyde
vanillin
<2
<2
3
<2
<2
adducts at the Lys and Arg residues in the Lys-peptide are
favored. Here, we were first interested in the SAR for
modifications at the Lys residue. Table 3 lists the results both
expressed as depletion of the Lys-peptide and relative formation
of a new peak with the theoretical mass of the reduced SB
(referred to as adduct-16). For all chemicals with the exception
of atranol and chloratranol, the reduced SB appears as the main
reaction product. Relatively high amounts were detected for 17.
Somewhat lower reactivity was apparent for benzaldehyde and
the m-substituted 21 and 19 and for the p-substituted
compounds 16, 9, 10, and 15. Similar to the results from the
butylamine experiments, the lowest reactivity under these
conditions was observed for the p-hydroxy and p-methoxy
derivatives such as vanillin. However, these quantitative results
must be treated with some caution: NaBH3CN is also able to
reduce the test chemical, and thus, we have two competing
reactions, and not a constant reservoir of test chemical is
present. These results indicate that SB formation can occur
with the Lys-peptide in the DPRA if the SBs are trapped in situ.
For atranol and chloratranol, a more complex picture of adduct
formation was observed in the presence of NaBH3CN. For
chloratranol, the main peak has the mass consistent with a
direct Schiff base (not reduced), while for atranol, several peaks
which could not be interpreted were observed.
Peptide Reactivity Assay: Peptide Modifications
Observed in the Absence of NaBH3CN. The most relevant
question of course is how BAs react in the presence of water
and in the absence of in situ reduction, as this would be closest
to the conditions we can expect in the living organism. We thus
performed peptide reactivity assays under modified DPRA
conditions (50 mM test chemical, 0.5 mM peptide, no in situ
reduction, and pH of 7.5 instead of 10.5 as in the original
DPRA). Interestingly, significant formation of an adduct with a
base ion of the theoretical m/z of a direct SB (labeled as
adduct-18) could only be detected for the o-hydroxy-aldehydes
7, 8, and 14 (Figure 3 shows the LC-MS analysis for atranol 7
as an example, and Table 4 lists the quantitative results). No
significant peptide depletion and no, or only traces, of reaction
products were observed for most of the m- and p-substituted
hydroxy- and methoxy BAs. These results would indicate that
Schiff base formation is not favored (i.e., the equilibrium is far
on the left side) in the aqueous DPRA conditions in the
Table 3. Formation of Reduced Schiff Bases with the Lys-
Peptide with NaBH3CN-Trapping
b
% adduct-16 [rel.
a
% depletion
34.2 1.3
%]
17 2-bromo-5-
hydroxybenzaldehyde
47.3 8.7
18 3-hydroxybenzaldehyde
16 4-chlorobenzaldehyde
19 3-methoxybenzaldehyde
13.6 2.2
13.5 1.4
13.4 2.3
10.8 1.5
10.5 1.2
10 5.6
23.7 3.3
21.8 2.3
30.5
3
8
1
7
9
atranol
complex adduct
17.6 3.2
benzaldehyde
chloratranol
cuminic aldehyde
complex adduct
9
5.3
23
3
10 4-methylbenzaldehyde
15 4-nitrobenzaldehyde
8.8 1.9
8.5 2.5
7.2 2.3
12.4 1.4
14.4 1.3
9.4 2.9
12 6-methoxynaphtalene-2-
carbaldehyde
14 salicylaldehyde
6.4 1.1
5.3 0.9
8.5 1.3
6.9 0.7
11 3-chloro-4-
methoxybenzaldehyde
6
5
3
heliotropine
ethylvanillin
vanillin
3.2 2.5
2.3 3.5
2.2 4.9
0.9 4.2
−0.2 6.4
−0.6 2.4
1.8 0.3
0.8 0.1
0.7 0.1
1.7 0.3
2.1 0.3
0.8 0.3
13 veratrumaldehyde
4
2
4-methoxybenzaldehyde
4-hydroxybenzaldehyde
a
0.5 mM peptide was reacted with 1 mM test chemical at pH of 7.5 in
the presence of 50 mM of NaBH3CN, and the % depletion of the
b
parent peptide was calculated The peak of the reaction product with
the theoretical mass of the reduced Schiff base (Mpeptide + Mtest_chemical
− 18 + 2) was integrated and quantified relative to the control peak of
the Lys-residue; since the response factor for the different adducts is
not known, this is an approximation. In most cases, the response factor
appears higher for the adducts.
E
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
Figure 3. LC-MS analysis of 50 mM atranol 8 incubated with 0.5 mM of the Lys-peptide at pH 7.5 for 24 h.
Table 4. Formation of Adducts and Peptide Modifications with the Lys-Peptide at pH 7.5 without NaBH3CN-Trapping
% peptide
a
adduct −18 [rel.
Lys-aldehyde [rel. adduct −2 [rel.
%] %]
b
depletion
%]
other adducts
7
8
chloratranol
atranol
60.5 0.3
91.7 12.8
n.d.
n.d.
several double adducts, mass 1080−
1264
60.1 0.2
95.4 3.9
n.d.
n.d.
several double adducts, mass 1060−
1074
16 4-chlorobenzaldehyde
14 salicylaldehyde
35.8 6.8
27.3 2.6
12.5 2.1
8.8 2.3
5.1 3.5
4.4 4.2
3.1 2.3
2.2 1.9
1.9 4.4
0.9 0.5
0.9 0.7
0.5 3.6
−0.2 4.6
−0.5 3.4
−0.5 3.3
n.d.
81 8.1
n.d.
199.4 7.5
n.d.
73.2 4.1
n.d.
9
1
cuminic aldehyde
benzaldehyde
19.7 3.1
39.6 11.6
19 2.4
n.d.
36.4
5
10 4-methylbenzaldehyde
n.d.
20.3 11.4
n.d.
12.2
12
4
11 3-chloro-4-methoxybenzaldehyde
n.d.
5
traces of peptide +28
5
3
ethylvanillin
vanillin
0.7
0
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
traces of adduct-16
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
traces of peptide +28 and adduct-16
traces of adduct-38
13 veratrumaldehyde
n.d.
18 3-hydroxybenzaldehyde
n.d.
6
heliotropine
n.d.
15 4-nitrobenzaldehyde
n.d.
4
2
4-methoxybenzaldehyde
4-hydroxybenzaldehyde
n.d.
low amount peptide +28
n.d.
12 6-methoxynaphtalene-2-
n.d.
carbaldehyde
19 3-methoxybenzaldehyde
−0.9 3.1
−3.8 3.1
n.d.
n.d.
n.d.
n.d.
n.d.
17 2-bromo-5-hydroxybenzaldehyde
0.3
0
a
b
0.5 mM of test peptide was incubated with 50 mM of test chemical for 24 h at pH 7.5. The peaks of the reaction products were quantified relative
to the control peak of the Lys-residue; since the response factor in the MS analysis for the different adducts is not known, this is a relative value.
absence of the trapping reagent with the exception of the
potent sensitizers with o-hydroxy groups.
However, we observed two unexpected chromatographic
peaks in the reactions with a number of different test chemicals,
in particular 1, 16, and the p-alkyl BAs 9 and 10: One identical
peak (further referred to as peptide-1) was present in
incubations with all these test chemicals. It contained two
base ions at m/z 775.4 and m/z 807.4 in the ESI+-spectrum
(see Figure 4). Because of the mass difference of 32 between
these two ions, we suspected that the ion at 807.4 could be due
to the addition of methanol from the eluent used in the
chromatographic separation, and LC-MS analysis was repeated
F
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
Figure 4. LC-MS analysis of 50 mM BA 1 incubated with 0.5 mM of the Lys-peptide at pH 7.5 for 24 h.
with ACN as eluent. Under these conditions, this peak solely
contained the ion at m/z 775.4 (data not shown). This mass
would correspond to the Lys-peptide having lost one mass unit.
We hypothesized that the ion at m/z 775.4 could correspond to
the aldehyde formed at the Lys-residue by oxidative
deamination and that the ion at m/z 807.4 would then
correspond to the hemiacetal of this aldehyde with methanol
formed during chromatography. This hypothesis will be further
substantiated below. The second peak observed in incubations
with BA 1 has a base ion at m/z 880.4. This ion corresponds to
the pseudomolecular ion of a peptide−benzaldehyde reaction
product [MProduct + H]+ with the formal interpretation of
[(Mpeptide+ MBenzaldehyde − 2) + H]+ or (775.4 + 106 − 2) + 1
and will be referred to as adduct-2. A similar product (i.e.,
adduct-2) was observed for a number of other test chemicals,
and these results are summarized in Table 4. We wondered
whether this unexpected adduct can only be formed from
aromatic aldehydes and thus tested the aliphatic skin sensitizing
aldehyde hydoxycitronellal under the same conditions. As
shown in Supporting Information, Figure S1, this chemical also
led to the formation of the peptide-1and the adduct-2 peaks; in
this case, the reaction was even quantitative. Hence, this
reaction pathway appears to be relevant for the reactivity of
different aldehydes when tested under neutral DPRA
conditions with the Lys-peptide, and we thus decided to
investigate it in greater detail.
Incubations with BA. To better understand the unexpected
adduct-2 between the Lys-peptide and benzaldehyde, we first
performed LC-MS2 analysis (Figure S2 in Supporting
Information). The fragments B4 (the largest fragment without
the Lys residue) was still observed in the adduct-2 peak,
whereas the fragments B5 and B6 containing the Lys residue do
carry a mass difference of +104, thus indicating that the peptide
does carry the modification at the Lys-residue. In the LC-MS2
analysis of the peptide-1 peak, the fragments B5 and B6 have
lost one mass unit, whereas B4 is unmodified, indicating that
this modification also occurred at the Lys residue (data not
shown).
We next separated the different modified peaks with
preparative HPLC and performed NMR experiments on the
purified fractions. Initially, the purified adduct-2 peak was over
90% pure, but it then rapidly reacted further to contain some
peptide-1 along with the apparent adduct-2 peak. This mixture
then remained stable over >2 weeks (as verified by LC-MS)
and was used for the NMR experiments.
Adduct-2 shows two signals in the 1H NMR spectrum, which
were key in its structural determination. One of them belongs
to a monosubstituted aromatic ring stemming from BA at 7.77
ppm; its doublet-like multiplicity stands for two ortho-protons
in an AA’MM’X system. The other one is a triplet (J = 6.8 Hz)
at 3.45 ppm. Their respective 13C NMR shifts extracted from a
HSQC experiment20 are 127.2 ppm for the aromatic proton
pair and 39.9 ppm for an aliphatic methylenic group. In the
HMBC experiment21 depicted in Figure 5, these two proton
Finally, we observed traces of a reaction product with the
pseudomolecular ion with the formal interpretation (775.4 +
Mtest chemical − 16) + 1 for vanillin and ethylvanillin. We
currently cannot interpret this result, but it indicates that these
molecules still do have a minimal peptide reactivity, but it
should be kept in mind that a large excess of test chemical was
used and that only very small traces of product were observed.
Structure Elucidation of the Modified Peptide
(Peptide-1) and the Peptide Adduct (Adduct-2) in
3
groups share JC,H couplings to a common carbon atom
resonating at 171.1 ppm. These latter results are fully
compatible with an aromatic amide of the type Ar-CO-NH-
CH2-R, the R group being the remaining modified lysine
residue from the heptapeptide. Furthermore, this HMBC
experiment leads to the two next carbons in the lysine chain,
both of methylenic type, at 22.7 and 28.2 ppm, still starting
G
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
Figure 5. gs-HMBC NMR spectrum of the HPLC purified fraction containing adduct-2. The arrows in the benzamide moiety indicate the two 3JC,H
coupling pathways corresponding to the cross-peaks a and b; both of them lead to the amide carbonyl. Two further cross-peaks arising from the ε-
protons at 3.45 ppm indicate the δ- and γ-neighbor-carbons at, respectively, 28.2 and 22.7 ppm.
from the key signal at 3.45 ppm and also indicated in Figure 5.
These two 13C NMR shifts are expected in γ- and δ-positions,
respectively, for such a benzamide, as well as 39.9 ppm for the
ε-carbon. These values were compared with the values we
measured for the commercially available N-ε-benzoyl-L-lysine in
the same positions, and they were found to be in very good
agreement: 22.0 (Cγ), 28.4 (Cδ), and 39.7 ppm (Cε) (data not
shown). However, any modification on the arginine residue can
be excluded since it has been found intact in the same HSQC/
HMBC experiments. Additionally, a series of 1D-selective
TOCSY NMR experiments22 were performed. The selective
excitation offset was set on the key triplet at 3.45 ppm since it is
located at the end of the ex-lysine unit, in the ε-position (Figure
6). By increasing the mixing time starting from 20 to 80 ms in
four experiments, the δ-, γ-, and β-protons and finally the
characteristic α-proton at 4.23 all appear successively on the
obtained spectra series. These latter findings as well as the
arguments stemming from the two key proton groups
demonstrate clearly that the lysine moiety has been trans-
formed into the corresponding benzamide in adduct-2.
Another component present in that mixture is the aldehyde
from the formal oxidative deamination of the Lys-peptide, as
well as its hydrate. The characteristic aldehyde proton has been
1
found at 9.61 ppm as a triplet (J = 1.5 Hz) in the H NMR
spectrum, which suggests that the first neighbor (δ-position) is
a methylenic pair of protons as would be expected for a lysine-
derived aldehyde. The aldehyde carbon (ε-position) resonates
at 208.0 ppm according to the HSQC experiment. The HMBC
experiment indicates a correlation to a carbon at 42.7 ppm
starting from the aldehyde proton; this is in line with a
methylene-carbon in δ-position. Further positions in the ex-
lysine chain are in important overlapping regions; therefore, the
1D-selective TOCSY NMR experiment was again applied to
explore the whole spin-system. This time, a selective excitation
H
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
Figure 6. 1D-selective TOCSY NMR series of the HPLC purified fraction containing adduct-2. The selective excitation offset was set on ε-protons at
3.45 ppm for all TOCSY traces (1−4). With a short mixing time of 20 ms as in trace 1, only the first neighbor-protons (Hδ) appear. By increasing the
mixing time in 20 ms steps in three further experiments, the γ-, the β-protons, and finally the characteristic α-proton at 4.23 appear all successively.
1
Trace 5 is the corresponding H NMR spectrum with presaturation on the residual water in D2O.
offset on the δ-protons was set; logically, one would expect to
start at the beginning of the spin system, at the aldehyde
proton, but the very low vicinal coupling constant of this
proton (1.5 Hz) will hamper the scalar coupling information
transmission along the whole spin-system. A mixing time of 120
ms was necessary to reach the α-proton at 4.24 ppm, as
depicted in the middle trace of Figure 7. The aldehyde derived
hydrate was also unveiled by a 1D-selective TOCSY experi-
ment; the selective excitation offset on the ε-proton at 4.96
ppm could point out all protons until reaching the α-proton, as
shown in the lower trace of Figure 7.
On the basis of these analytical data, we propose that the
initially formed Schiff base between BA and the Lys-peptide
undergoes spontaneous oxidation leading to the benzamide. In
a parallel reaction, the double bond of the Schiff base can
migrate and then hydrolyze, which leads to oxidative
deamination of the peptide. These reactions are summarized
in Scheme 1.
conserving their integral ratio. At a first view, these signals
match with the imine proton and the α-imine methylene
protons of the expected Schiff base. A more extensive analysis
with gs-HSQC and gs-HMBC experiments point out a 13C
NMR chemical shift of 173.2 ppm for the aromatic carbon
bearing the hydroxyl group of the former salicyl aldehyde,
which is approximately 10 ppm higher than expected. A control
experiment using butylamine instead of the Lys-peptide in
CD3CN shows a shift of 162.1 ppm for the same carbon.
Moreover, the α-imine aliphatic methylenic carbon resonates at
51.9 ppm versus 59.5 ppm in this control experiment. This may
indicate that we observe an enaminone tautomer in the reaction
with the Lys-peptide, which we could confirm by a further
control experiment still using butylamine and salicylic aldehyde
but with the buffered solvent mixture 75% D2O/25% CD3CN.
This later experiment showed shifts very similar to those of the
Lys-peptide adduct as exposed in Scheme 2.
NMR Analysis of the Reaction Products between
Salicylaldehyde and the Lys-Peptide or Butylamine.
Finally, the formation of the adduct occurring from
DISCUSSION
■
This study reveals a striking difference in the reactivity of
differently substituted BA derivatives toward amine nucleo-
philes, with o-hydroxybenzaldehydes having a high propensity
to form Schiff-bases which are stable in aqueous conditions.
Other benzaldehydes do form Schiff-bases which are labile to
hydrolysis, but the Schiff bases can act as an intermediate which
is further transformed to stable amide adducts. Thus, Schiff
1
salicylaldehyde and the Lys-peptide has been followed by H
NMR over time with a 100-fold excess of the aldehyde, in a
phosphate buffered solvent mixture 75% D2O/25% CD3CN.
After 15 min of reaction time, a singlet at 8.47 ppm as well as a
triplet at 3.80 ppm appear. Their relative integral ratio is
respectively 1:2. Both of them grow with time until 160 min,
I
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
Figure 7. 1D-selective TOCSY NMR series of the HPLC purified fraction containing besides the adduct-2 an aldehyde and its hydrate. In the middle
trace, the selective excitation offset was set on the δ-protons at 2.55 ppm. All protons of the aldehyde spin-system appear with a mixing time of 120
ms. With the same mixing time, the lower trace shows the protons Hδ, Hγ, Hβ, and finally Hα when the selective excitation offset was set on the ε-
1
proton of the hydrate. The upper trace is the corresponding H NMR spectrum with presaturation on the residual water signal in D2O.
Scheme 1. Possible Pathway of the Oxidative Decomposition of the Schiff Base of Benzaldehyde
Scheme 2. Comparison of Observed 13C-NMR Chemical Shifts at Key Positions in the Enaminone/Imine Tautomers of the
Former Salicyl Aldehyde
base formation is an important process to understand skin
sensitization by benzaldehydes, but unexpected reactions
downstream of the initial Schiff base adduct appear equally
important.
SAR for Reactivity with Amines and Relationship to
Sensitization Potential. Overall, the reactivity can be
summarized in descending order as o-hydroxy BAs >
unsubstituted BA > p-alkyl BAs > p-methoxy BAs > p-hydroxy
BAs. A similar trend is seen in the simple butylamine−
acetonitrile system and in the somewhat more biological
peptide reactivity conditions with trapping of the formed Schiff
bases. These differences can explain some of the experimentally
J
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
or clinically reported information on the skin sensitizing
potency of BAs. Thus, atranol and chloratranol, which are
formed in the classical processing of oak moss absolute, appear
to be the most potent chemicals to elicit allergic reactions to
fragrances.16 These two chemicals are also among the most
reactive ones in the amine reactivity assay, and they (along with
salicylaldehyde 14) are also the only BAs forming stable Schiff
bases in the aqueous conditions of the peptide reactivity assay.
Salicylaldehyde finds very little use in commercial fragrances,
but positive patch test reactions have been observed by
dermatologists.23 Our reactivity results confirm that the use of
this chemical should be limited. The pronounced effect of the
o-hydroxy group may be explained by the observed stable
enaminone tautomer and/or a favorable hydrogen bond
between the nitrogen atom and the keto-group of the
enaminone. This tautomeric form might render the SB of the
o-hydroxy BA itself reactive to additional nucleophilic attack.
However, in the case of the SB of 14 with the Lys-peptide,
addition of a Cys-containing peptide or propylthiol did not
change the amount of adduct-18 nor did we observe a new
reaction product due to a modification of this adduct (data not
shown).
weak sensitizer (EC3 = 25%), whereas 4 is nonsensitizing in the
LLNA (>25%) and a weak sensitizer in human tests (LOEL in
human repeat insult patch test of 4724 μg/cm2, which would
correspond to an LLNA EC3 of ca. 20%; RIFM, personal
communication). Hence, we would conclude that p-alkoxy BAs
do have a low reactivity and only a very weak sensitization
potential.
2-Bromo-5-hydroxy-benzaldehyde has a quite unique profile:
intermediate reactivity with butylamine and the highest amount
of reduced Schiff base with the peptide but no covalent adduct
in the absence of in situ reduction. Nevertheless, it has a quite
low LLNA EC3 value. This may indicate that in this case also
the unstable Schiff base is of importance.
Reactivity in Aqueous System. As outlined in the
Introduction, Schiff base formation is a reversible process,
and the equilibrium is highly dependent on (i) the
concentration of water, (ii) the amount of free aldehyde, and
(iii) pH/protonation of the amine group. Thus, it is probably
not surprising that no evidence for Schiff base formation was
found for most aldehydes under the classical peptide reactivity
conditions,11,12 unless reaction conditions were specifically
favored to facilitate Schiff base formation.14 One would
anticipate that conditions on the migratory routes of dendritic
cells away from the aldehyde-treated skin sites would favor
hydrolysis due to neutral pH, presence of water, and absence of
free aldehyde. Here, we confirmed that, with the notable
exceptions of 7, 8, and 14, Schiff base formation under the
aqueous peptide reactivity conditions can only be observed to a
significant extent by including a trapping reagent.
Yet the current results indicate that stable peptide
modification (aldehyde formation at Lys by oxidative
deamination) and stable adduct formation (amide formation)
is nevertheless possible under dilute aqueous conditions at
physiological pH. Interestingly, these modifications were mostly
observed for those aldehydes which also formed higher
amounts of Schiff bases in the trapping experiments, probably
indicating that Schiff base formation is the first step in the
reaction cascade, and we have to invoke oxidation of the Schiff
base to explain the results. Indeed, amide formation from Schiff
bases of benzaldehyde derivatives in the presence of molecular
oxygen had been observed,26 with the probable involvement of
the transient formation of a peroximidic acid. This process was
observed in neat preparations of the Schiff base and was
accelerated after the addition of a free radical initiator. In a
more biological system, the long-lived adduct of 4-oxo-2-
nonenal at lysine residues was shown to be a 4-ketoamide27
formed from an initial Schiff base adduct. However, in this
system no oxidation by molecular oxygen had to be invoked, as
amide formation was occurring under concomitant reduction of
the double bond. In addition, it appears to be a specific process
for the 4-oxo-2-enal substructure, and a mechanism with
intermediate formation of a pyrrol ring was proposed. Even if
the amide is the final stable product, the mechanism thus
appears to be significantly different.27 Amide formation from an
initial Schiff base between lysine residues and aliphatic
aldehydes had also been reported by Ishino et al.,28 in that
study H2O2 was needed to facilitate the reaction. Immuno-
logical evidence also indicated that the process occurs in vivo
under conditions of oxidative stress.
The stable Schiff base formation by o-hydroxy BAs is not
without precedent. The immunostimulating drug candidate
tucaresol (4[2-formyl-3-hydroxyphenoxymethyl] benzoic acid)
does contain this structural element and had been reported to
form stable Schiff bases with cell surface proteins.24
At the other extreme of the above scale of reactivity are the p-
hydroxy BAs. Schiff base formation in the butylamine−
acetonitrile system was detectable for these chemicals in the
presence of a high excess of butylamine, but in the dose−
response analysis, this group showed no significant reactivity. In
addition, only traces of the reduced Schiff bases were detected
in the peptide reactivity experiments with NaBH3CN trapping
for p-hydroxy BAs. This low reactivity is paralleled by the
absence of sensitization potential in animal tests and the low
incidence of positive sensitization data from diagnostic patch
testing (12 cases in almost 6000 patients tested for vanillin),
despite the very broad use of vanillin and ethylvanillin in
consumer products.23,25 The very low reactivity of these agents,
as well as of 4-hydroxybenzaldehyde, is in line with their low
degree of sensitization.
We did not for all BAs observe a correlation between
reported sensitization potential in animals and the ability to
form Schiff bases. BA 1, 4-chloro BA 16, and 4-nitro BA 15 are
negative in the LLNA; 1 has a clear sensitization potential in
humans (LOEL in human maximization test of 2760 μg/cm2,
which would correspond to an LLNA EC3 of around 10%;
RIFM, personal communication), and no human data are
available for 15 and 16. These chemicals have a significant
reactivity according to our results. Among the p-alkyl BAs, 9
was reported to be nonsensitizing at 10%, whereas 10 is almost
as strong according to the LLNA as atranol. Compound 10 is
only slightly more reactive than 9, and given the high structural
similarity and similar reactivity, this difference in the LLNA
might be an overestimation, and it should be kept in mind that
the compiled LLNA data in Figure 1 come from different
sources which may contribute to the variability of the animal
results. A more homogeneous data set would probably only be
obtained by retesting all chemicals in one laboratory, which
appears not to be warranted for ethical reasons.
Another surprising result is the formation of the aldehyde
functionality at the Lys-residue due to oxidative deamination.
This process would require migration of the Schiff base double
bond and subsequent hydrolysis. Aldehyde formation at the Lys
The p-alkoxy BAs have a slightly higher reactivity as
compared to the p-hydroxy BAs. Compound 6 is rated as a
K
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
residue in reactivity experiments with sensitizers had been
reported before upon reaction of a peptide with 2,5-dimethyl-p-
benzoquinonediimine and subsequent hydrolysis.29 However,
this appeared to be a rather specific case, as the Schiff base-like
adduct has a quinoid structure which can tautomerize to the
stable hydroquinone form upon migration of the Schiff base
double bond (similar to the reaction cascade in the oxidative
deamination of primary amines by topaquinone, the cofactor of
amine oxidases).30 Surprisingly, our results indicate that this
process also can happen upon reaction of the peptides with
simple aldehydes, which do not contain this redox system. The
formed peptide/protein aldehyde could in theory then lead to
immunogenic novel epitopes by intra- or intermolecular cross-
linking, yet these epitopes would not be specific to the
sensitizing aldehyde, and we would not expect them to lead to
clinically relevant specific allergies.
ACKNOWLEDGMENTS
■
We thank Peter Gygax and Felix Flachsmann for interesting
discussions and valuable suggestions for structure elucidation,
Fabian Kuhn for interpreting GC-MS data, Michael Rothaupt
for HPLC purification of peptide adducts, and the Research
Institute for Fragrance Materials (RIFM) for providing human
and animal data on the sensitization potential.
ABBREVIATIONS
■
ACN, acetonitrile; BA, benzaldehyde; DPRA, direct peptide
reactivity assay; ECVAM, European Center for the Validation
of Alternative Methods to animal testing; ESI, electrospray
ionization; HRRIPT, human repeat insult patch test; LLNA,
local lymph node assay; LOEL, lowest observed effect level;
MTBE, methyl tert-butyl ether; RIFM, Research Institute for
Fragrance Materials; SB, Schiff base
We have no full explanation of the reaction mechanism
leading to the oxidative degradation products of the Schiff base.
The most likely explanation is the formation of an intermediate
oxaziridine or nitrone (with the same molecular weight as the
amide), which further reacts to the amide and the aldehyde.
This would explain why the isolated peak with the mass of
adduct-2 was not fully stable in the beginning but reacted
further to form some of the peptide-1; it might initially contain
both the semistable oxaziridine, the nitrone, and the stable
amide. Indeed, the addition of oxone (KHSO5), which is
known to oxidize Schiff bases to the oxaziridines,31 to the
reaction of BA with the Lys-peptide led to higher yields of
adduct-2 and adduct-1 (data not shown).
The stable modifications and adducts occur at a physiological
pH. We currently do not know how important the context of
the peptide structure is, but we observed that the same
reactions do occur but with much lower efficacy for acetyl-
lysine (data not shown). We cannot predict whether the same
reactions will happen with proteins and in the living body, but if
they do, this would describe a mechanism of how aldehydes can
form stable immunogenic epitopes since after an equilibration
time, the amide-adduct was found to be stable in water in the
absence of the free aldehyde for over a week (data not shown).
Finally, a similar reaction was observed for the structurally
completely different aldehyde hydroxycitronellal, and it is
therefore not specific to benzaldehydes. To what extent amide
formation correlates to the sensitizing capacity of different
aldehydes is a field which warrants further study.
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) Gerberick, F., Aleksic, M., Basketter, D., Casati, S., Karlberg, A.
T., Kern, P., Kimber, I., Lepoittevin, J. P., Natsch, A., Ovigne, J. M.,
Rovida, C., Sakaguchi, H., and Schultz, T. (2008) Chemical reactivity
measurement and the predictive identification of skin sensitisers.
ATLA 36, 215−242.
(3) Gerberick, G. F., Vassallo, J. D., Bailey, R. E., Chaney, J. G.,
Morrall, S. W., and Lepoittevin, J. P. (2004) Development of a peptide
reactivity assay for screening contact allergens. Toxicol. Sci. 81, 332−
343.
(4) Chipinda, I., Ajibola, R. O., Morakinyo, M. K., Ruwona, T. B.,
Simoyi, R. H. Siegel, P. D. Rapid and simple kinetics screening assay
for electrophilic dermal sensitizers using nitrobenzenethiol. Chem. Res.
Toxicol. 23, 918-925.
(5) Schultz, T. W., Rogers, K., and Aptula, A. O. (2009) Read-across
to rank skin sensitization potential: subcategories for the Michael
acceptor domain. Contact Dermatitis 60, 21−31.
(6) Patlewicz, G., Dimitrov, S. D., Low, L. K., Kern, P. S., Dimitrova,
G. D., Comber, M. I., Aptula, A. O., Phillips, R. D., Niemela, J.,
Madsen, C., Wedebye, E. B., Roberts, D. W., Bailey, P. T., and
Mekenyan, O. G. (2007) TIMES-SS–a promising tool for the
assessment of skin sensitization hazard. A characterization with respect
to the OECD validation principles for (Q)SARs and an external
evaluation for predictivity. Regul. Toxicol. Pharmacol. 48, 225−239.
(7) Roberts, D. W., Patlewicz, G., Kern, P. S., Gerberick, F., Kimber,
I., Dearman, R. J., Ryan, C. A., Basketter, D. A., and Aptula, A. O.
(2007) Mechanistic applicability domain classification of a local lymph
node assay dataset for skin sensitization. Chem. Res. Toxicol. 20, 1019−
1030.
ASSOCIATED CONTENT
* Supporting Information
■
S
Source of the LLNA data; LC-MS analysis of the reaction
product between hydroxycitronellal and the Lys-peptide; and
LC-MS2 analysis of adduct-2, i.e., the peak with pseudomo-
lecular ion at 880 formed by incubating BA with the Lys-
peptide. This material is available free of charge via the Internet
at http://pubs.acs.org.
(8) Roberts, D. W., and Aptula, A. O. (2008) Determinants of skin
sensitisation potential. J. Appl. Toxicol. 28, 377−387.
(9) Roberts, D. W., Aptula, A. O., and Patlewicz, G. (2006)
Mechanistic applicability domains for non-animal based prediction of
toxicological endpoints. QSAR analysis of the schiff base applicability
domain for skin sensitization. Chem. Res. Toxicol 19, 1228−1233.
(10) Aptula, A. O., Patlewicz, G., and Roberts, D. W. (2005) Skin
sensitization: Reaction mechanistic applicability domains for structure-
activity relationships. Chem. Res. Toxicol. 18, 1420−1426.
(11) Gerberick, G. F., Vassallo, J. D., Foertsch, L. M., Price, B. B.,
Chaney, J. G., and Lepoittevin, J. P. (2007) Quantification of chemical
peptide reactivity for screening contact allergens: A classification tree
model approach. Toxicol. Sci. 97, 417−427.
AUTHOR INFORMATION
Corresponding Author
■
*Tel: ++41 44 824 21 05. Fax: ++41 44 824 29 26. E-mail:
andreas.natsch@givaudan.com.
Funding
(12) Natsch, A., and Gfeller, H. (2008) LC-MS-based character-
ization of the peptide reactivity of chemicals to improve the in vitro
prediction of the skin sensitization potential. Toxicol. Sci. 106, 464−
478.
This study was entirely funded by Givaudan Schweiz AG.
Notes
The authors declare no competing financial interest.
L
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Chemical Research in Toxicology
Article
(13) Reichardt, P., Schreiber, A., Wichmann, G., Metzner, G., Efer, J.,
and Raabe, F. (2003) Identification and quantification of in vitro
adduct formation between protein reactive xenobiotics and a lysine-
containing model peptide. Environ. Toxicol. 18, 29−36.
(14) Aleksic, M., Thain, E., Roger, D., Saib, O., Davies, M., Li, J.,
Aptula, A., and Zazzeroni, R. (2009) Reactivity profiling: Covalent
modification of single nucleophile peptides for skin sensitization risk
assessment. Toxicol. Sci. 108, 401−411.
formylation of 2,5-dimethyl-p-benzoquinonediimine. Chem. Res.
Toxicol. 20, 1155−1161.
(31) Schoumacker, S., Hamelin, O., Teti, S., Pecaut, J., and
Fontecave, M. (2005) Activation of oxaziridines by lewis acids:
application in enantioselective sulfoxidation. J. Org. Chem. 70, 301−
308.
(15) Bernard, G., Gimenez-Arnau, E., Rastogi, S. C., Heydorn, S.,
Johansen, J. D., Menne, T., Goossens, A., Andersen, K., and
Lepoittevin, J. P. (2003) Contact allergy to oak moss: search for
sensitizing molecules using combined bioassay-guided chemical
fractionation, GC-MS, and structure-activity relationship analysis.
Arch. Dermatol. Res. 295, 229−235.
(16) Johansen, J. D., Bernard, G., Gimenez-Arnau, E., Lepoittevin, J.
P., Bruze, M., and Andersen, K. E. (2006) Comparison of elicitation
potential of chloroatranol and atranol–2 allergens in oak moss
absolute. Contact Dermatitis 54, 192−195.
(17) Patlewicz, G., Basketter, D. A., Smith, C. K., Hotchkiss, S. A.,
and Roberts, D. W. (2001) Skin-sensitization structure-activity
relationships for aldehydes. Contact Dermatitis 44, 331−336.
(18) Braun, M. M., and U. Houben, S. (1990) Synthese von (R)-
Lasiodiplodin aus Polyhydroxybuttersaure. Liebigs Ann. Chem., 513−
̈
517.
(19) Roberts, D. W., and Natsch, A. (2009) High throughput kinetic
profiling approach for covalent binding to peptides: Application to
skin sensitization potency of michael acceptor electrophiles. Chem. Res.
Toxicol. 22, 592−603.
(20) Schleucher, J., Schwendinger, M., Sattler, M., Schmidt, P.,
Schedletzky, O., Glaser, S. J., Sorensen, O. W., and Griesinger, C.
(1994) A general enhancement scheme in heteronuclear multidimen-
sional NMR employing pulsed field gradients. J. Biomol. NMR 4, 301−
306.
(21) Cicero, D. O., Barbato, G., and Bazzo, R. (2001) Sensitivity
enhancement of a two-dimensional experiment for the measurement
of heteronuclear long-range coupling constants, by a new scheme of
coherence selection by gradients. J. Magn. Reson. 148, 209−213.
(22) Stott, K., Stonehouse, J., Keeler, J., Hwang, T. L., and Shaka, A.
J. (1995) Excitation sculpting in high-resolution nuclear magnetic
resonance spectroscopy: Application to selective NOE experiments. J.
Am. Chem. Soc. 117, 4199−4200.
(23) Uter, W., Geier, J., Frosch, P. Schnuch, A. Contact allergy to
fragrances: current patch test results (2005−2008) from the
Information Network of Departments of Dermatology. Contact
Dermatitis 63, 254-261.
(24) Chen, H., and Rhodes, J. (1996) Schiff base forming drugs:
mechanisms of immune potentiation and therapeutic potential. J. Mol.
Med. 74, 497−504.
(25) Hausen, B. M. (2001) Contact allergy to balsam of Peru. II.
Patch test results in 102 patients with selected balsam of Peru
constituents. Am. J. Contact Dermatitis 12, 93−102.
(26) Auret, J. B., Boyd, D. R., and Coulter, P. B. (1984) Autoxidation
reactions of imines to form oxaziridines and amides. J. Chem. Soc.,
Chem. Commun. 7, 463−464.
(27) Zhu, X., and Sayre, L. M. (2007) Long-lived 4-oxo-2-enal-
derived apparent lysine michael adducts are actually the isomeric 4-
ketoamides. Chem. Res. Toxicol. 20, 165−170.
(28) Ishino, K., Shibata, T., Ishii, T., Liu, Y. T., Toyokuni, S., Zhu, X.,
Sayre, L. M., and Uchida, K. (2008) Protein N-acylation: H2O2-
mediated covalent modification of protein by lipid peroxidation-
derived saturated aldehydes. Chem. Res. Toxicol. 21, 1261−1270.
(29) Eilstein, J., Gimenez-Arnau, E., Duche, D., Cavusoglu, N.,
Hussler, G., Rousset, F., and Lepoittevin, J. P. (2008) Sensitization to
p-amino aromatic compounds: Study of the covalent binding of 2,5-
dimethyl-p-benzoquinonediimine to a model peptide by electrospray
ionization tandem mass spectrometry. Bioorg. Med. Chem. 16, 5482−
5489.
(30) Eilstein, J., Gimenez-Arnau, E., Duche, D., Rousset, F., and
Lepoittevin, J. P. (2007) Mechanistic studies on the lysine-induced N-
M
dx.doi.org/10.1021/tx300278t | Chem. Res. Toxicol. XXXX, XXX, XXX−XXX