A combined spin trapping/EPR/mass spectrometry approach to study the formation of a cyclic peroxide by dienolic precursor autoxidation

Article abstract of DOI:10.1039/b921694d

The spontaneous addition of air oxygen to a dienolic compound, yielding a cyclic peroxide, was followed by spin trapping (ST) combined with EPR spectroscopy and mass spectrometry (MS). Using two different nitrones, the ST/EPR study allowed the detection of the spin adduct of a radical intermediate, and the radical centre in the addend was identified after similar experiments performed with two different ¹³C-labelled analogues of the substrate. The media were also submitted to electrospray ionisation, in both positive and negative modes, for structural characterisation of the spin adducts by tandem mass spectrometry. This allowed the structure of the hydroxylamine derivatives of the nitroxides formed to be identified. Following these results, a mechanism pathway was proposed for this autoxidation.

Full text of DOI:10.1039/b921694d

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
A combined spin trapping/EPR/mass spectrometry approach to study the
formation of a cyclic peroxide by dienolic precursor autoxidation†
Mathilde Triquigneaux,a Laurence Charles,a Christiane Andre´-Barre`sb and Be´atrice Tuccio*a
Received 19th October 2009, Accepted 17th December 2009
First published as an Advance Article on the web 21st January 2010
DOI: 10.1039/b921694d
The spontaneous addition of air oxygen to a dienolic compound, yielding a cyclic peroxide, was
followed by spin trapping (ST) combined with EPR spectroscopy and mass spectrometry (MS). Using
two different nitrones, the ST/EPR study allowed the detection of the spin adduct of a radical
intermediate, and the radical centre in the addend was identified after similar experiments performed
with two different 13C-labelled analogues of the substrate. The media were also submitted to
electrospray ionisation, in both positive and negative modes, for structural characterisation of the spin
adducts by tandem mass spectrometry. This allowed the structure of the hydroxylamine derivatives of
the nitroxides formed to be identified. Following these results, a mechanism pathway was proposed for
this autoxidation.
netic resonance (EPR) methodology.7 The ST technique relies on
Introduction
the fast addition of a transient radical to a diamagnetic spin trap
G-factors are natural compounds acting as phytohormones,
(usually a nitrone or a nitroso compound) to yield a longer-lived
growth regulators and intervening in plant defence in Eucalyptus
paramagnetic spinadduct(anitroxide) that can be detected by con-
ventional EPR spectroscopy (see Scheme 2). Analysing the EPR
grandis and other Myrtaceae.1 Several natural or synthetic G-
factors showed interesting antiplasmodial activities related to the
spectrum obtained gives information about the addend structure.
presence of a peroxide bridge,2 also found in a large series of
Using 2-methyl-2-nitrosopropane (MNP) and diethyl (2-methyl-
other antimalarial compounds (e.g. artemisinin, BO7, arteflene
1-oxido-3,4-dihydro-2H-pyrrol-2-yl) phosphonate (DEPMPO) as
spin traps, carbon-centred radical intermediates formed during
or trioxaquines).3 Since the cyclic peroxide 2 could be formed
after oxygen uptake from air by the dienolic system 1a (which
1a autoxidation were detected. However, we did not gain much
is in equilibrium with a dienone form), this autoxidation was
information about the structure of the radical trapped using the
used as the key step in the synthesis of various G-factors (see
Scheme 1a).4 The reaction is quantitative, selective, performed at
sole EPR spectroscopy, and both structures 3 and 5, corresponding
to two different reaction pathways, were then postulated for the
room temperature in very mild conditions, even in the solid state,
intermediate initially formed (see Scheme 1a).
producing no waste and maximising atom economy. It could thus
Recently, we described several applications of ST coupled with
mass spectrometry (MS) for free radical characterisation.8 The
feasibility of using tandem mass spectrometry (MS/MS) for
direct identification of nitroxide spin adducts electrosprayed from
complex mixtures, without prior chromatographic separation,
was thus demonstrated. This method also brought valuable
information about diamagnetic products derived from the
nitroxide initially formed. The purpose of this work was to
proceed with the study of the mechanism of 3O2 addition on
1a using a combined ST/EPR/MS approach. With the help of
this powerful tool, our aim was i) to confirm with non-cyclic
nitrones the ST/EPR results previously obtained with MNP and
DEPMPO, ii) to discriminate the different conceivable reaction
pathways and iii) to elucidate the structures of various reaction
intermediates. Therefore, two nitrones, the commercially available
4-{(E)-[tert-butyl(oxido)imino]methyl}pyridine 1-oxide (POBN),
and the 31P-labelled trinitrone diethyl {1-[(Z)-(3,5-bis{(E)-
[[1(diethoxyphosphoryl)-1-methylethyl](oxido)imino]methyl}-
constitute an environmentally benign green chemical process to
synthesise cyclic peroxides. Since then, efforts have been made to
elucidate the mechanism of this intriguing reaction.4,5 It proceeds
in the dark by adding oxygen from air, without any activation or
sensitiser. A photo- or a base-catalysed oxidation, the intervention
of singlet dioxygen or an electron transfer between O2 and the
substrate are all processes that have been discarded based on
various experiments. On the other hand, this peroxidation was
inhibited by a catalytic amount of the radical scavenger tert-
butanethiol, which led us to examine the possibility that 1a could
behave as a radicaloid in a reaction with triplet oxygen avoiding the
spin barrier. Although such a mechanism is unusual, several other
“forbidden” additions of triplet oxygen to singlet substrates have
been reported.6 A preliminary study of the autoxidation of 1a into
2 was performed using the spin-trapping (ST)/electron paramag-
aUniversite´s Aix-Marseille I, II & III–CNRS, UMR 6264: Laboratoire
Chimie Provence, Spectrome´tries Applique´es a` la Chimie Structurale, 13397
Marseille Cedex 20, France; Fax: (+33) 491282897; Tel: (+33) 491288743.
E-mail: beatrice.tuccio-lauricella@univ-provence.fr
benzylidene)(oxido)amino]-1-methylethyl}phosphonate
(TN),
were employed to trap radical intermediates formed during
the autoxidation of three dienolic precursors, i.e. 1a and its
13C-labelled analogues 1b and 1c (see Schemes 1 and 2). Then,
both conventional X-band EPR detection and electrospray
ionisation mass spectrometry (ESI-MS) were employed to
bUniversite´ Paul Sabatier–CNRS, UMR 5068: Synthe`se et Physico-Chimie
de Mole´cules d’Inte´reˆt Biologique, 31062 Toulouse Cedex, France
† Electronic supplementary information (ESI) available: ESI-MS/MS of
1a and of 18, in positive and negative modes. See DOI: 10.1039/b921694d
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 1361–1367 | 1361
©

View Article Online
Scheme 1 a) Formation of the G3-factor 2 by spontaneous addition of dioxygen on the dienolic precursor 1a according to two possible pathways A and
B, and structures of the biradical intermediates 3–6 postulated for this reaction; b) formulae of the 13C-labelled analogues of 1a considered in this study.
prepared according to procedures described previously and stored
under argon to avoid autoxidation.5b The cyclic peroxide 2 was
obtained after air oxidation of 1a.4
Spin trapping
Spin adducts were produced by mixing the dienolic precursor
(1a–1c, 0.2 mol dm-3) and the spin trap (0.2 mol dm-3 POBN
or 0.1 mol dm-3 TN) in oxygenated benzene. The system was
allowed to react for 1 h, and the sample was deoxygenated by
nitrogen bubbling before EPR analysis. For each experiment,
300 mL of reaction medium was prepared. An aliquot (ca. 30 mL)
was transferred into a glass pipette closed with a septum for EPR
analysis, while 50 mL was diluted in methanol (1/10, vol/vol) for
MS analysis.
EPR spectroscopy
EPR assays were carried out at room temperature by using a
Bruker EMX spectrometer operating at X-band with 100 kHz
modulation frequency, and equipped with an NMR gaussmeter for
magnetic field calibration. The instrument settings were as follows:
non-saturating microwave power, 20 mW; modulation amplitude,
Scheme 2 Spin trapping of a radical Y∑ using the nitrones POBN or TN
0.15 mT; receiver gain, 1 ¥ 106; time constant, 1.28 ms; scan time,
and structures of the nitroxide spin adducts formed.
120 s; scan width, 10–20 mT. Hyperfine coupling constant (hfcc)
and g-tensor values were obtained after computer simulation of
detect and characterise the spin adducts formed as well as their
derivatives.
the EPR spectra using the software elaborated by Duling.10
Mass spectrometry
Experimental
High-resolution MS and MS/MS experiments were performed
with a QStar Elite mass spectrometer (Applied Biosystems SCIEX,
Concord, ON, Canada) equipped with an electrospray ionisation
source. In the positive ion mode, the capillary voltage was set at
5500 V and the cone voltage at 80 V. For experiments performed
in the negative ion mode, the capillary voltage was -4500 V
and the cone voltage was -70 V. In this hybrid instrument, ions
were measured using an orthogonal acceleration time-of-flight
(oa-TOF) mass analyser. A quadrupole was used for selection
Materials
POBN was purchased from Sigma-Aldrich (St. Louis, MO).
Methanol (SDS, Peypin, France), acetonitrile and benzene
(Sigma-Aldrich) were of the highest grade of purity commercially
available and used as received. Poly(ethylene glycol) with Mn
400 g mol-1 (Sigma-Aldrich) was used as an internal standard
for accurate mass measurements. TN was synthesised and purified
as described elsewhere.9 The dienolic precursors 1a, 1b and 1c were
1362 | Org. Biomol. Chem., 2010, 8, 1361–1367
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
of precursor ions to be further submitted to collision induced
dissociation (CID) in MS/MS experiments. In MS, accurate mass
measurements were performed using two reference ions from a
poly(ethylene glycol) internal standard, according to a procedure
described elsewhere.11 The precursor ion was used as the reference
for accurate measurements of product ion m/z ratio in MS/MS
spectra. In the instrument, air was used as the nebulising gas
(10 psi) whereas nitrogen was used as the curtain gas (20 psi)
as well as the collision gas. Collision energy was set according
to the experiments. Instrument control, data acquisition and data
processing of all experiments were achieved using Analyst software
(QS 2.0) provided by Applied Biosystems. Sample solutions were
introduced in the ionisation source at a 5 mL min-1 flow rate using
a syringe pump.
Results and discussion
ST/EPR study
Two linear aldonitrones, POBN and TN, were used to trap free
radicals produced during 1a–1c autoxidation. In order to simplify
the notation, the nitroxide formed by trapping a radical Y∑ by
POBN or TN will be denoted POBN-Y or TN-Y, respectively (see
Scheme 2). A standard EPR spectrum of a POBN-Y nitroxide
shows six lines due to hyperfine couplings of the unpaired electron
with the nitrogen and the b-hydrogen nuclei. In the case of TN-Y,
this sextuplet is split by a large extra coupling with the phosphorus
nucleus, resulting in a twelve line EPR spectrum. The hyperfine
coupling constants (hfccs) with the nitrogen, the b-hydrogen and
the b-phosphorus nuclei will be denoted aN, aH and aP, respectively.
Preparing a benzene solution containing 1a, POBN and O2
allowed us to record the spectrum given in Fig. 1a, which revealed
the presence of two nitroxides. The major species EPR signal
exhibited six lines (g = 2.00619, aN = 1.64 mT and aH = 0.16 mT,
82%) and could be assigned to a carbon-centred radical adduct
denoted POBN-C, on the basis of the results previously obtained
with DEPMPO7 and by comparison with literature data.12 The
minor signal (g = 2.00660, aN = 1.36 mT and aHb = 1 mT,
18%) showed a much lower aH value, often characteristic of
an oxygen-centred radical adduct.13 Complementary experiments
were performed to establish whether this adduct was formed
Fig. 1 Experimental EPR signals (black full lines) obtained in deoxy-
genated benzene after 1 h of reaction between molecular oxygen and (a)
1a (0.2 mol dm-3) in the presence of POBN (0.2 mol dm-3), and (b) 1b
(0.2 mol dm-3) in the presence of TN (0.1 mol dm-3). Their superimposed
simulations (red dotted lines) led to the following parameters: signal (a)
82% POBN-C (g = 2.00619, aN= 1.64 mT and aH = 0.16 mT) and 18%
POBN-OR (g = 2.00660, aN = 1.36 mT and aHb = 0.1 mT); signal (b)
100% TN-13C (g = 2.00624, aN = 1.55 mT, aH = 0.19 mT, aP = 4.26 mT
13 C
and a = 1.14 mT).
to POBN. First, the EPR spectra of its various spin adducts
usually provide more information about the addend, because
the strong hfcc aP is very sensitive to the nature of the radical
trapped. In addition, the presence of three nitrone functions on
TN greatly enhances the radical trapping rates, without leading
to the formation of a significant amount of di- or tri-nitroxide
when TN concentration is higher than 0.02 mol dm-3.9 Mixing 1a
(0.2 mol dm-3) and TN (0.1 mol dm-3) in oxygenated benzene gave
rise to an intense twelve line EPR spectrum, showing the presence
of a single mono-nitroxide. Its analysis resulted in the following
parameters: g = 2.00624, aN = 1.55 mT, aH = 0.19 mT and aP =
4.26 mT. This nitroxide was identified as the TN spin adduct of
a radical centred on a sterically hindered carbon, i.e. TN-C, by
comparison with the literature data.9,14 It is worth mentioning that
the nitroxide TN-OR was never observed in these experiments, and
two hypotheses were envisaged to explain this result. The first one
relates to kinetic criteria. Just after the initial addition of O2 to 1a,
two competitive reactions should be considered: the ring closure,
leading to 2, and the spin trapping of a radical intermediate. The
formation of RO∑, originating from the cyclic peroxide bridge
break, necessitates the presence in the medium of a significant
amount of 2. With the nitrone POBN, the trapping kinetics were
slow enough to allow the generation of RO∑ and its subsequent
addition to the nitrone, and POBN-OR was detected as a minor
by-product. Since the rates of free radical trapping by TN are much
higher,9,15 the amount of RO∑ in the medium could be too low to be
∑
after trapping a peroxyl radical RO2 , as suggested in the study
realised with DEPMPO,7 or an alkoxyl radical RO∑, formed
after homolytic ring opening in the final cyclic peroxide.5 POBN
(0.2 mol dm-3) and 2 (0.2 mol dm-3) were dissolved in acetonitrile
(2 being poorly soluble in benzene) and the medium was passed
through a metallic needle in order to add metal traces. EPR
analysis of the medium revealed the presence of a single nitroxide,
with g = 2.00660, aN = 1.36 mT and aHb ≤ 0.12 mT. Though the
lines were too broad to allow an accurate determination of aH,
these EPR parameters resemble very much that of the oxygen-
centred radical adduct resulting from the reaction between 1a
and O2 in the presence of POBN. Therefore, we can reasonably
assign the minor species detected in the former experiments to
POBN-OR, a by-product formed after trapping an alkoxyl radical
originating from a O–O bond break in 2, rather than to a peroxyl
radical adduct.
The same kind of experiment was then conducted with the spin
trap TN, which presents two main advantages when compared
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 1361–1367 | 1363
©

View Article Online
observed by ST/EPR. Note however that no signal was detected
when a solution containing TN (0.1 mol dm-3), 2 (0.2 mol dm-3)
and metal traces in acetonitrile was analysed by EPR, which goes
against this first hypothesis. Alternatively, trace metal impurities
eventually present in the medium could form inactive complexes
with the phosphonate groups borne on TN, which would hamper
the peroxide ring opening reaction. Whatever it may be, these
results refute the hypothesis of a peroxyl radical trapping.
The experiments performed with both POBN and TN confirm
the formation of carbon-centred radical intermediates during
1a autoxidation and produces further evidence of a radical
pathway for this reaction. At this point however, the actual
mechanism of this reaction remains ambiguous because the adduct
characteristics as indicated by EPR did not reveal much about
the exact structure of the species trapped. In particular, the two
possible pathways A and B depicted in Scheme 1 would allow the
formation of the intermediate isomers 3–6. In order to identify
the actual radical intermediate, further spin trapping experiments
were performed using the 13C-labelled precursors 1b and 1c. The
EPR spectrum recorded from an oxygenated benzene solution
containing 1b and POBN, revealed the presence of two nitroxides
unpaired electron on a tertiary carbon, the cyclisation reaction in
the resonance form 6 seems to be a much faster process than the
trapping reaction.
This ST/EPR study is fully consistent with the pathway B for
this autoxidation reaction, but it does not necessarily imply that
the pathway A did not occur simultaneously. Neither 3 nor 4 was
ever observed in these ST/EPR experiments, but it could not be
concluded that these alternative intermediates were not created
during the endoperoxidation reaction without considering the
kinetic aspects. Actually, if the ring closure yielding 2 in pathway
A is much faster than the trapping reactions, no spin adduct
derived from either 3 or 4 would ever be detected. Note also that,
in a previous work concerning the search for new antimalarial
cyclic peroxides related to G-factor series, cyclic ether alcohols
and hydroperoxides were obtained besides cyclic peroxides after
autoxidation of the dienol intermediate step.17 The formation of
these unexpected compounds were explained by evolution of the
postulated biradical intermediate 3 following pathway A. Pursuing
this hypothesis, a mechanism was proposed in which biradical
species participate but in this special case could evolve differently
from the sole formation of cyclic peroxides as described in previous
reports.17 Taking these results into account, the present ST/EPR
study only proved that the pathway B occurred in our experimental
conditions, but does not allow pathway A to be discarded as an
alternative or even as a main process.
only. The minor species (g = 2.00660, aN = 1.36 mT and aHb
=
0.10 mT, 10%) could be clearly identified as the spin adduct POBN-
OR, formed after trapping the free radical resulting from the ring
opening in the final cyclic peroxide. The predominant EPR signal
(90%) showed twelve lines, because of an extra hyperfine coupling
with the 13C nucleus in the b-position towards the nitrogen, with
the following parameters: g = 2.00619, aN = 1.64 mT, aH = 0.16 mT
Mass spectrometry study
and a C = 1.14 mT. These data clearly suggest that the 13C carbon
Compounds generated during the reaction of the dienol-O2
biradical with POBN or TN could not be precisely defined
by EPR spectroscopy, which focuses on the detection of the
sole paramagnetic function, and does not provide information
about the global structure. More particularly, a predominant
mono-radical carbon-centred spin adduct was detected, with no
indication about the rearrangement of the oxygen part in the bi-
radical intermediate. Therefore, the ST/EPR/MS approach8 was
used to characterise the spin adduct structures.
13
is the radical centre that added to the nitrone function during the
trapping reaction, and the spin adduct detected was thus denoted
POBN-13C. The same kind of results were obtained with the spin
trap TN. Reacting dioxygen on 1b in the presence of TN in benzene
gave rise to the intense EPR signal shown in Fig. 1b. Its analysis
revealed the presence of a single nitroxide which, on the basis of
its EPR parameters (g = 2.00624, aN = 1.55 mT, aH = 0.19 mT,
13
aP = 4.26 mT and a = 1.14 mT) could be identified as TN-
C
13C, the spin adduct obtained after trapping a radical intermediate
centred on the 13C atom. All these results clearly point to the
fact that the spin adducts detected, i.e. POBN-13C and TN-13C,
came from the initial formation of the biradical 5 only. In both
Since no MS study of 1a has ever been described, this compound
was first analysed by ESI-MS/MS, to establish a referenced
fragmentation behaviour to be further used in the structural
characterisation of spin adducts. The positive mode ESI-MS
spectrum showed a peak at m/z 237.1 which could be assigned
to the protonated dienol, [1a + H]+. Upon CID, this precursor
ion gave rise to a main product ion at m/z 139.1 as well as three
other minor products at m/z 209.1, m/z 167.1 and m/z 121.1 (see
Electronic Supplementary Information, Fig. S1). Accurate mass
measurements allowed us to propose the fragmentation pathways
depicted in Scheme 3a. Elimination of a CO molecule from [1a +
13
adducts, the a C values were higher than those frequently observed
for a splitting due to a 13C in the b-position towards a nitroxide
nitrogen. Since this hyperfine coupling depends on the dihedral
angle between the planes CbCaN and CaNpz, substituents may
greatly modify a C, as shown for instance by Rockenbauer et al.16
13
In the case of POBN-13C and TN-13C, the steric hindrance due to
the bulky radical captured by the two PBN-type nitrones should
yield a particular conformation in which the Ca-13C bond is almost
+
H]+ would produce m/z 209.1524 (C13H21O2 , +5.7 ppm) from
13
orthogonal to the nitroxide plane CaNO, resulting in the high a
values determined.
which the loss of a 2-methylprop-1-en-1-one (C4H6O) would give
rise to m/z 139.1117 (C9H15O+, -0.3 ppm). Alternatively, this m/z
139.1 product ion could be formed after the precursor ion has first
C
The same kind of experiment was also performed with the 13C-
analogue 1c with both POBN and TN. The spin adducts thus
detected showed EPR spectra identical to those obtained with
the non-labelled precursor 1a, indicating that the carbon-centred
radical trapped did not correspond to the intermediate 4. In all
these experiments, we never observed any spin adduct derived
from 6, regardless of the nitrone or of the dienolic precursor used.
Though the mesomeric intermediates 5 and 6 both present an
+
eliminated the C4H6O moiety to yield m/z 167.1062 (C10H15O2 ,
+2.7 ppm), then the neutral CO. Finally, dehydration of m/z 139.1
would produce m/z 121.1012 (C9H13+, +0.2 ppm) via a cyclisation
mechanism. In the negative ion mode, the deprotonated dienol
could be observed at m/z 235.1 in the ESI-MS spectrum. CID
experiments revealed that [1a - H]- dissociates via a main pathway,
consisting of two successive eliminations of a methyl radical to
1364 | Org. Biomol. Chem., 2010, 8, 1361–1367
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
signal at m/z 429.2, assigned to the deprotonated hydroxylamine
[16 - H]-.
A similar approach was followed to scrutinise the ESI mass
spectra obtained in the case of the spin adduct TN-C. The same
structural assumptions were proposed for derivatives of the spin
adduct involving TN and 5, that is, compounds 11–14 resulting
from the addition of a H∑ on the peroxyl radical moiety (m/z
1011.4), the rearrangement in diol (m/z 995.4) and bimolecular
association (m/z 1247.4 and m/z 1279.4). Again, none of these
species could be mass detected and, as in the case of POBN,
a peak at m/z 978.4 in positive mode ESI-MS and a signal at
m/z 976.4 in the negative ion mode suggested the detection of
the hydroxylamine 18 diagnostic of the formation of 17 during
spin trapping experiments (see Scheme 4). These assignments were
mainly based on m/z values measured in the MS spectra and low
abundance of the targeted ions did not allow good quality accurate
mass measurements.11 MS/MS experiments were thus required to
validate the structure proposed in Scheme 4.
As can be seen in Fig. 2a, CID of [16 + H]+ at m/z 431.3
produced two main fragment ions at m/z 195.1 and m/z 139.1.
+
Assuming a C24H35N2O5 composition for the precursor ion to
calibrate this MS/MS spectrum, accurate mass measurement
+
indicates that m/z 195.1124 (C10H15N2O2 , +1.7 ppm) would
correspond to the protonated POBN molecule, arising from the
elimination of the 1a neutral from the precursor ion. The m/z
139.1 product ion (m/z 139.0480, C6H7N2O2 , +15.8 ppm) would
Scheme 3 Fragmentation pattern of (a) [1a + H]+ and (b) [1a - H]-.
+
thus result from the loss of a 2-methylpropene from m/z 195.1, a
typical dissociation reaction observed during CID of protonated
POBN. MS/MS experiments performed on the deprotonated
hydroxylamine 16 were shown to provide complementary infor-
mation to data obtained in the positive ion mode. Indeed, the main
product ion generated upon CID of [16 - H]- (Fig. 2b) could be
∑
respectively produce m/z 220.1110 (C13H16O3 -, -2.7 ppm) and
∑
m/z 205.0868 (C12H13O3 -, +0.8 ppm), as illustrated in Scheme 3b.
The first CH3 loss appears as an exception to the even-electron
rule18 but could be explained by the radicaloid character of 1a in
its deprotonated form, as illustrated in Scheme 3b.
∑
-
The reaction medium in which the nitroxide adduct POBN-
C was produced and detected in EPR was then submitted to
electrospray ionisation after diluting the organic solution in
methanol. The so-obtained ESI-mass spectrum was scrutinised
for signals diagnostic of the different protonated spin adducts
that could be envisaged to be formed during spin trapping of
the intermediate 5. Indeed, various rearrangements of 5, prior or
after spin trapping, can be assumed. A first derivative 7 could be
formed after reduction of the peroxyl radical of the nitroxide spin
adduct into hydroperoxide to generate a protonated compound
at m/z 464.3 in ESI-MS. An alcohol group could also be formed
upon evolution of the peroxyl part, generating the geminated-diol
compound 8 which should be detected as a protonated molecule
at m/z 448.3. Finally, bimolecular rearrangements have also been
envisaged, leading to molecules 9 and 10 that would be respectively
detected at m/z 699.4 and at m/z 667.4 in their protonated
forms. The structure of these proposed derivatives are detailed
in Scheme 4. Neither the intact POBN/5 adduct nor any of the
proposed derivatives could be detected during mass spectrometric
experiments. Alternatively, the carbon position related to both
hydroxyl group and peroxyl radical in 5 could rearrange in a
carbonyl group, giving rise to a protonated nitroxide expected at
m/z 430.3 in mass spectrometry. Actually, the positive mode ESI
mass spectrum showed a peak at m/z 431.3, which was assigned
to the protonated hydroxylamine, [16 + H]+, presumably derived
from the spin adduct 15 (see Scheme 4). This result was further
supported in negative mode MS experiments which showed a
assigned to a deprotonated 1a molecule (m/z 235.1304, C14H19O3 ,
+15.4 ppm) formed after the elimination of a POBN neutral from
the precursor ion. This assignment was further supported by data
obtained at higher collision energy (60 eV, laboratory frame) which
showed the typical m/z 220.1 and m/z 205.1 product ions of
[PG3 - H]- (vide supra). It could thus be concluded from these
MS/MS data that the ionic species observed at m/z 431.1 in
the positive mode and at m/z 429.2 in the negative mode were
respectively the protonated and the deprotonated form of the
hydroxylamine 16.
A similar approach was followed to validate the structure
proposed for the hydroxylamine 18 (see Electronic Supplementary
Information, Figure S2). In positive mode, a unique product
+
ion (m/z 742.3089, C30H55N3O12P3 , -13.0 ppm) was observed
after the dissociation of [18 + H]+, which could be assigned
to a protonated TN nitrone formed after elimination of a 1a
moiety from the precursor ion. The intense peak measured at
-
m/z 235.1320 (C14H19O3 , +8.6 ppm) in the MS/MS spectrum
of [18 - H]-, would arise from the elimination of a TN neutral
leading to a deprotonated 1a molecule. As in the case of 16,
the complementary data obtained while performing MS/MS
experiments in both polarity modes allows us to validate the
structure proposed for the hydroxylamine 18. Detection of the
hydroxylamine species in mass spectrometry would indicate that
the paramagnetic species observed in EPR experiments were 15
and 17. These nitroxides could result from the trapping of 5
followed by a fast conversion of the peroxyl radical moiety into a
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 1361–1367 | 1365
©

View Article Online
Scheme 4 a) Various structures 7–14 hypothesised for the spin adducts POBN-C and TN-C and discarded after ESI-MS experiments and b) structures
15–18 proposed for POBN-C and TN-C spin adducts and for their hydroxylamine derivatives based on ESI-MS results.
Conclusions
This work clearly demonstrates that the cyclic peroxide synthesis
by autoxidation of the dienolic precursor 1a corresponds to a
radical process, thereby confirming the main result obtained in
our preliminary ST/EPR study.7 Whatever the nitrone, a single
mono-radical nitroxide, i.e. a carbon-centred radical spin adduct,
was EPR-detected in the ST experiments performed, and the use
of 13C-labelled precursors allowed us to identify the radical centre
in the addend. The ST/MS technique, used without preliminary
Scheme 5 Mechanism assumed for the formation of 15.
chromatographic purification, provided a powerful tool, comple-
mentary to EPR, for the identification of the radical intermediates.
In particular, the structure of the spin adducts POBN-C and TN-C
could be elucidated after MS/MS experiments. All the EPR and
MS results obtained led us to conclude that the conversion of the
dienolic precursor 1a into the G-factor 2 could follow the pathway
B (see Scheme 1) in the experimental conditions considered
here, though pathway A could not be dismissed at present.
Finally, this combined ST/EPR/MS approach was demon-
strated to be a powerful tool for mechanism studies in radical
chemistry.
∑
ketone function. This release of HO2 could not occur in 5 before
the spin trapping process. In this case, this process would also take
place in the absence of spin trap, which contradicts the quantitative
formation of the cyclic peroxide 2 from 1 autoxidation. Following
all these results, the mechanism assumed for the formation of the
spin adducts detected is illustrated by Scheme 5 in the case of
POBN-C.
1366 | Org. Biomol. Chem., 2010, 8, 1361–1367
This journal is
The Royal Society of Chemistry 2010
©

View Article Online
3 (a) R. Haynes and S. Vonwiller, Acc. Chem. Res., 1997, 30, 73; (b) A.
Robert and B. Meunier, Chem. Soc. Rev., 1998, 27, 273; (c) C. Jefford,
S. Kohmoto, D. Jaggi, G. Timari, J.-C. Rossier, M. Rudaz, O. Barbuzzi,
D. Ge´rard, V. Berger, P. Kamalaprija, J. Mareda, G. Bernardinelli, I.
Manzanares, C. Canfield, S. Fleck, B. Robinson and W. Peters, Helv.
Chim. Acta, 1995, 78, 647; (d) W. Hofeinz, H. Burgin, E. Gocke, R.
Jacquet, R. Masciadri, S. Schmid, H. Stohler and H. Urwyler, Trop.
Med. Parasitol., 1994, 45, 261; (e) O. Dechy-Cabaret, F. Benoit-Vical,
A. Robert and B. Meunier, ChemBioChem., 2000, 4, 283.
4 M. Gavrilan, C. Andre´-Barre`s, M. Baltas, T. Tzedakis and L. Gorri-
chon, Tetrahedron Lett., 2001, 42, 2465–2468.
5 (a) F. Najjar, F. Fre´ville, F. Desmoulin, L. Gorrichon, M. Baltas, H.
Gornitzka, T. Tzedakis and C. Andre´-Barre`s, Tetrahedron Lett., 2004,
45, 6919; (b) C. Andre´-Barre`s, F. Najjar, A. L. Bottalla, S. Massou,
C. Zedde, M. Baltas and L. Gorrichon, J. Org. Chem., 2005, 70, 6921;
(c) C. Lacaze-Dufaure, F. Najjar and C. Andre´-Barre`s, THEOCHEM,
2007, 803, 17.
6 (a) L. Sy and G. Brown, Tetrahedron, 2002, 58, 897; (b) R. Schobert, R.
Stehle and H. Milhous, J. Org. Chem., 2003, 68, 9827; (c) C. Ba¨cktorp,
J. Tobias, Johnson Wass, I. Panas, M. Sko¨ld, A. Bo¨rje and G. Nyman,
J. Phys. Chem. A, 2006, 110, 12204; (d) C. Bowes, D. Montecalvo and
F. Sondheimer, Tetrahedron Lett., 1973, 14, 3181; (e) D. Barton, R.
Haynes, G. Leclerc, P. Magnus and I. Menzies, J. Chem. Soc., Perkin
Trans. 1, 1975, 2055; (f) M. Laguerre, J. Dunogues and R. Calas,
J. Chem. Res., 1978, 295; (g) A. Krebs, H. Schmalstieg, O. Jarchow
and K.-H. Klaska, Tetrahedron Lett., 1980, 21, 3171.
7 F. Najjar, C. Andre´-Barre`s, R. Lauricella, L. Gorrichon and B. Tuccio,
Tetrahedron Lett., 2005, 46, 2117.
8 (a) B. Tuccio, R. Lauricella and L. Charles, Int. J. Mass Spectrom.,
2006, 252, 47; (b) I. El Hassan, L. Charles, R. Lauricella and B.
Tuccio, New J. Chem., 2008, 32, 680; (c) M. Triquigneaux, B. Tuccio,
R. Lauricella and L. Charles, J. Am. Soc. Mass Spectrom., 2009, 20,
2013.
Fig. 2 ESI-MS/MS of a) [16 + H]+ at 15 eV and b) [16 - H]- at 15 eV.
9 V. Roubaud, H. Dozol, C. Rizzi, R. Lauricella, J. C. Bouteiller and B.
Tuccio, J. Chem. Soc., Perkin Trans. 2, 2002, 958.
Acknowledgements
10 D. R. Duling, J. Magn. Reson., Ser. B, 1994, 104, 105.
11 L. Charles, Rapid Commun. Mass Spectrom., 2008, 22, 151.
12 P. Tordo, Electron Paramagn. Reson., 1998, 16, 116, and references
therein.
L. Charles acknowledges support from Spectropole, the Analytical
Facility of Aix-Marseille University, by allowing a special access
to the instruments purchased with European Funding (FEDER
OBJ2142-3341).
13 G. Buettner, Free Radical Biol. Med., 1987, 3, 259, and references
therein.
14 (a) B. Tuccio, A. Zeghdaoui, J. P. Finet, V. Cerri and P. Tordo, Res.
Chem. Intermed., 1996, 22, 393; (b) C. Rizzi, S. Marque, F. Belin, J. C.
Bouteiller, R. Lauricella, B. Tuccio, V. Cerri and P. Tordo, J. Chem.
Soc., Perkin Trans. 2, 1997, 2513.
Notes and references
15 R. Lauricella, A. Allouch, V. Roubaud, J. C. Bouteiller and B. Tuccio,
Org. Biomol. Chem., 2004, 2, 1304.
16 A. Rockenbauer, L. Korecz and K. Hideg, J. Chem. Soc., Perkin Trans.
2, 1993, 2149.
17 V. Bernat, N. Saffon, M. Maynadier, H. Vial and C. Andre´-Barre`s,
Tetrahedron, 2009, 65, 7372–7379.
18 F. W. McLafferty, Org. Mass Spectrom., 1980, 15, 114.
1 E. Ghisalberti, Phytochemistry, 1996, 41, 7.
2 (a) F. Najjar, M. Baltas, L. Gorrichon, Y. Moreno, H. Vial, T. Tzedakis
and C. Andre´-Barre`s, Eur. J. Org. Chem., 2003, 3335; (b) F. Najjar,
L. Gorrichon, M. Baltas, H. Vial, T. Tzedakis and C. Andre´-Barre`s,
Bioorg. Med. Chem. Lett., 2004, 14, 1433; (c) F. Najjar, L. Gorrichon,
M. Baltas, H. Vial and C. Andre´-Barre`s, Org. Biomol. Chem., 2005, 3,
1612.
This journal is
The Royal Society of Chemistry 2010
Org. Biomol. Chem., 2010, 8, 1361–1367 | 1367
©