Vitamin E chemistry. studies into initial oxidation intermediates of α-tocopherol: Disproving the involvement of 5a-C-centered "chromanol methide" radicals

Article abstract of DOI:10.1021/jo062553j

(Chemical Equation Presented) Contrary to concepts handed down in the literature from the early days of vitamin E research, one-electron oxidation of vitamin E does not involve 5a-C-centered radicals. A combined approach of analytical techniques, in particular electron paramagnetic resonance spectroscopy (EPR), organic synthesis of special derivatives, isotopic labeling, kinetic studies, and computational chemistry was used to re-evaluate the one-electron and two-electron oxidation chemistry of α-tocopherol (α-toc). EPR in combination with 5a-¹³C-labeled compounds provided no indication of the involvement of 5a-C-centered radicals. Oxidation of special tocopherol derivatives were used to disprove the occurrence of 5a-C-centered one-electron intermediates. Additionally it was shown that those vitamin E reactions that were commonly evoked to plead for the involvement of C-centered tocopheryl radicals actually proceeded via heterolytic, i.e., non-radical, intermediates. The results will help to clear widely spread misunderstandings about the chemistry of vitamin E and will have mechanistic implications for the synthesis of tocopherol-based supramolecular structures and 5a-substituted α-tocopherol derivatives.

Full text of DOI:10.1021/jo062553j

Vitamin E Chemistry. Studies into Initial Oxidation Intermediates
of r-Tocopherol: Disproving the Involvement of 5a-C-Centered
“Chromanol Methide” Radicals
Thomas Rosenau,*,† Elisabeth Kloser,† Lars Gille,*,‡ Francesco Mazzini,§ and
Thomas Netscher|
Department of Chemistry, UniVersity of Natural Resources and Applied Life Sciences, Muthgasse 18,
A-1190 Vienna, Austria, Research Institute for Biochemical Pharmacology and Toxicology, UniVersity of
Veterinary Medicine Vienna, Veterina¨rpl. 1, A-1210 Vienna, Austria, Dipartimento di Chimica e Chimica
Industriale, UniVersita` di Pisa, Via Risorgimento 35, 56126 Pisa, Italy, and Research and DeVelopment,
DSM Nutritional Products, P.O. Box 3255, CH-4002 Basel, Switzerland
thomas.rosenau@boku.ac.at; lars.gille@Vu-wien.ac.at
ReceiVed December 13, 2006
Contrary to concepts handed down in the literature from the early days of vitamin E research, one-
electron oxidation of vitamin E does not involve 5a-C-centered radicals. A combined approach of analytical
techniques, in particular electron paramagnetic resonance spectroscopy (EPR), organic synthesis of special
derivatives, isotopic labeling, kinetic studies, and computational chemistry was used to re-evaluate the
one-electron and two-electron oxidation chemistry of R-tocopherol (R-toc). EPR in combination with
5a-13C-labeled compounds provided no indication of the involvement of 5a-C-centered radicals. Oxidation
of special tocopherol derivatives were used to disprove the occurrence of 5a-C-centered one-electron
intermediates. Additionally it was shown that those vitamin E reactions that were commonly evoked to
plead for the involvement of C-centered tocopheryl radicals actually proceeded via heterolytic, i.e., non-
radical, intermediates. The results will help to clear widely spread misunderstandings about the chemistry
of vitamin E and will have mechanistic implications for the synthesis of tocopherol-based supramolecular
structures and 5a-substituted R-tocopherol derivatives.
1. Introduction
R-tocopheryl acetate, but in fact it is a generic descriptor of all
tocol and tocotrienol derivatives exhibiting qualitatively the
vitamin E activity (as determined by specific biological tests)
of R-tocopherol,1-4 i.e., covering the four tocopherols and the
four tocotrienols (R-, â-, γ-, and δ-homologues).
Many reports on vitamin E chemistry date back to the 1950s
to 1970s of the 20th century, and most of them are dealing with
R-tocopherol chemistry or with converting non-R-tocopherols
Vitamin E, first reported barely one century ago, is the
biologically most important fat-soluble antioxidant and has
become a commodity product and bulk chemical in the
meantime. The major, large-scale application of vitamin E is
animal nutrition, and many pharmaceutical, health care, and
cosmetic products contain the substance. In the consumer notion
vitamin E is connected with terms such as antioxidant, radical
scavenging, and anti-aging. Usually, and due to the dominance
of the R-homologue in all kinds of applications, the term vitamin
E is widely used synonymous with R-tocopherol or even
(1) IUPAC-IUB Eur. J. Biochem. 1982, 123, 473-475. Nomenclature
of Tocopherols and Related Compounds.
(2) IUPAC-IUB Commission on Biochemical Nomenclature (CBN) Arch.
Biochim. Biophys. 1974, 165, 1-8.
(3) Baldenius, K. U.; Von dem Bussche-Hu¨nnefeld, L.; Hilgemann, E.;
Hoppe, P.; Stu¨rmer, R. In Ullmann’s Encyclopedia of Industrial Chemistry;
VCH Verlagsgesellschaft: Weinheim, 1996; Vol. A27, pp 478-488, 594-
597.
† University of Natural Resources and Applied Life Sciences.
‡ University of Veterinary Medicine.
§ Universita` di Pisa.
| DSM Nutritional Products.
(4) Netscher, T. Chimia 1996, 50, 563-567.
10.1021/jo062553j CCC: $37.00 © 2007 American Chemical Society
Published on Web 03/29/2007
3268
J. Org. Chem. 2007, 72, 3268-3281

Vitamin E Chemistry
SCHEME 1. Structure and Atom Numbering of
r-Tocopherol (1), Formation of the r-Tocopheroxyl Radical
and Its Major Resonance Forms (2 and 2′), and Their
Dominant Radical Coupling Products 3 and 4a
into the R-congener. After that time, the research focus shifted
somewhat to biological and medicinal aspects of tocopherols.
The chemical basics were largely handed down from that time
and after some time were seen as well established. As an
example, the peculiar reactivity of the 5-methyl group in
R-tocopherol (1) had been explained by the ill-defined and
inapplicable Mills-Nixon effect5 for more than six decades.
Nowadays, the matter finally has been clarified, and this
hypothesis was replaced by the modern strain-induced bond
localization (SIBL) explanation.6,7
The fact that the interest in tocopherols shifted away from
purely chemical aspects may also explain why theories and
reactions that might appear questionable today were passed
around for a longer time by citation, rewriting, and incorporation
into reviews.
This paper is concerned with a particular aspect of vitamin
E chemistry or, correctly, R-tocopherol chemistry: the nature
of primary intermediates in the one-electron oxidation of the
compound, i.e., R-tocopherol-derived radicals. In particular, the
involvement of C-centered radicals at 5a-C is critically studied.
Participation of 5a-C in radical processes was proposed in the
early years of vitamin E chemistry and has become widely
spread and accepted since then, not by appearance of additional
proofs but by repeated citation and the lack of necessity to
recheck that chemistry in detail. With the use of R-tocopherol
and R-tocopherol-type model compounds in supramolecular
chemistry and in 5a-C-linked derivatives, the question of the
exact chemistry of early R-tocopherol oxidation intermediates
became of crucial interest. In the present study, we set out to
clarify that chemistry by a combined experimental and theoreti-
cal approach, applying electron paramagnetic resonance spec-
troscopy (EPR) and other analytical techniques, as well as
synthesis of special derivatives, isotopic labeling, kinetic studies,
and computational chemistry.
a Here and in the following the term “R ) C16H33” refers to the
(2R,4′R,8′R)-isomer.
acceptor. The second process involves one-electron oxidation
to a radical cation, which can be observed by EPR spectroscopy
at low temperatures under certain conditions,15,16 followed by
loss of a proton to form 2. Also, the direct subsequent chemistry
of the R-tocopheroxyl radical 2 is well established and no matter
of debate. The spin density of 2, as a classic phenoxyl radical,
is mainly concentrated at oxygen O-6, which is the major
position for coupling with other C-centered radicals, leading to
chromanyl ethers 3. The spin density is also increased at ortho-
and para-positions (5-C, 7-C, 8a-C) of the aromatic ring.
Coupling with other radicals, especially O-centered ones,
proceeds mainly at the para-position (C-8a), leading to differ-
ently 8a-substituted chromanones 4 (Scheme 1).
2. Results and Discussion
2.1. R-Tocopherol and Its Primary Oxidation Products-
the Alleged Involvement of 5a-C-Centered Radicals. The
chromanoxyl radical 2 is the primary homolytic (one-electron,
radical) oxidation product of R-tocopherol (1). Its formation and
occurrence is comprehensively supported and confirmed by
EPR8-14 and ENDOR15 experiments. Radical 2 can be formed
by two pathways (Scheme 1). The first one involves loss of a
hydrogen atom and is largely dependent on the hydrogen atom
(5) (a) Mills, W. H.; Nixon, I. G. J. Chem. Soc. 1930, 2510-2525. (b)
See also a later review: Badger, G. M. Quart. ReV. Chem. Soc. 1951, 5,
147.
(6) Rosenau, T.; Ebner, G.; Stanger, A.; Perl, S.; Nuri, L. Chem. Eur. J.
2005, 11, 280-287.
(7) Rosenau, T.; Stanger, A. Tetrahedron Lett. 2005, 46, 7845-7848.
(8) Matsuo, M.; Matsumoto, S.; Ozawa, T. Org. Magn. Res. 1983, 21,
261-264.
(9) Burton, G. W.; Doba, T.; Gabe, E. J.; Hughes, L.; Lee, F. L.; Prasad,
L.; Ingold, K. U. J. Am. Chem. Soc. 1985, 107, 7053-7065.
(10) Boguth, W.; Niemann, H. Biochim. Biophys. Acta. 1971, 248, 121-
130.
The occurrence of a 5a-C-centered tocopherol-derived radical
5, often called a “chromanol methide” radical, had been
postulated in literature articles17-24 dating back to the early days
(16) Lehtovuori, P.; Joela, H. Phys. Chem. Chem. Phys. 2002, 4, 1928-
1933.
(17) Inglett, G. E.; Mattill, H. A. J. Am. Chem. Soc. 1955, 77, 6552-
6554.
(18) Knapp, F. W.; Tappel, A. L. J. Am. Oil Chem. Soc. 1961, 38, 151-
156.
(19) Skinner, W. A. J. Med. Chem. 1967, 10, 657-661.
(20) Skinner, W. A.; Alaupovic, P. J. Org. Chem. 1963, 28, 2854-2858.
(21) Nilsson, J. L. G.; Daves, D. G.; Folkers, K. Acta Chim. Scand. 1968,
22, 207-218.
(22) Fujimaki, M.; Kanamaru, K.; Kurata, T.; Igarashi, O. Agr. Biol.
Chem. 1970, 34, 1781-1786.
(23) Goodhue, C. T.; Risley, H. A. Biochem. Biophys. Res. Commun.
1964, 17, 549-553.
(11) Mukai, K.; Tsuzuki, N.; Ouchi, S.; Fukuzawa. K. Chem. Phys. Lipids
1982, 30, 337-345.
(12) Matsuo, M.; Matsumoto, S. Lipids 1983, 18, 81-86.
(13) Tsuchiya, J.; Niki, E.; Kamiya, Y. Bull. Chem. Soc. Jpn. 1983, 56,
229-232.
(14) Eloranta, J.; Ha¨ma¨la¨inen, E.; Ma¨kela¨, R.; Keka¨la¨inen, U. Acta Chem.
Scand. A 1983, 37, 383-391.
(15) Mukai, K.; Tsuzuki, N.; Ishizu, K.; Ouchi, S.; Fukuzawa. K. Chem.
Phys. Lipids 1981, 29, 129-135.
J. Org. Chem, Vol. 72, No. 9, 2007 3269

Rosenau et al.
SCHEME 3. Hypothetical Radical Mechanism for the
Formation of 5a-r-Tocopheryl Benzoate (7) by Reaction of
r-Tocopherol (1) with Dibenzoyl Peroxide
SCHEME 2. Structure of the Literature-Postulated
5a-C-Centered Radical 5 as a Tautomer of 2 and Structures
of the “Favored” (6) and “Disfavored” (6a)
r-Tocopherol-Derived o-Quinone Methides
SCHEME 4. Formation of r-Tocopherol Ethano-dimer 8 as
the Result of a Hypothetical Radical Recombination of Two
Radicals 5
of vitamin E research, which have been cited or supposedly
reconfirmed later.25-28 In some accounts, this radical has been
described in the literature as being a resonance form (canonic
structure) of the tocopheroxyl radical, which of course is
inaccurate. If indeed it exists, radical 5 represents a tautomer
of chromanoxyl radical 2, being formed by a chemical reaction,
namely, a 1,4-shift of one 5a-proton to the 6-oxygen, and not
just by a “shift of electrons” as in the case of resonance
structures (Scheme 2).
In all accounts mentioning R-tocopherol-derived C-centered
radicals, the spin density was described to be centered at 5a-C
but not at alternative carbons, such as 7a-C or 8b-C. Actually,
the occurrence of 5a-C-radicals was concluded from two facets
of vitamin E chemistry: on the one hand by experimental
observations that seemed to support 5a-C-radicals, and on the
other hand by analogy to the chemistry of the R-tocopherol-
derived o-quinone methide (oQM) 6. These theoretical and
experimental considerations will be briefly addressed next to
explain the starting situation for our studies.
Alleged “Theoretical Evidence” for R-Tocopherol-Derived
5a-C-Radicals. o-Quinone methide 6, a very frequent interme-
diate in tocopherol chemistry, is the product of two-electron
oxidation processes. It is formed with large preference for 5a-C
over 7a-C, oQM formation at the latter position leading to 6a
(Scheme 2). This peculiar reaction behavior of R-tocopherol
has been attributed to the so-called Mills-Nixon effect. Even
though this rationale was shown to be wrong and inapplicable
to the R-tocopherol case,6 it haunts the pertinent literature as a
blurred theory accounting for the preference of 5a-C in all kinds
of reactions. From the observed favoring of the 5a-methyl group
over the 7a-methyl group upon formation of an o-quinone
methide, it was concluded that for hypothetic R-tocopherol-
derived C-centered radicals the situation must be similar: the
radical was supposed to be preferentially centered at 5a-C and
not at 7a-C.
chemistry. The first one is the formation of 5a-substituted
derivatives in the reaction of R-tocopherol with radicals and
radical initiators. The most prominent example is here the
reaction of 1 with dibenzoyl peroxide leading to 5a-R-tocopheryl
benzoate (7) in fair yields,17 so that a “typical” radical
recombination mechanism was postulated (Scheme 3). Similarly,
low yields of 5a-alkoxy-R-tocopherols were obtained by oxida-
tion of R-tocopherol with tert-butyl hydroperoxide or other
peroxides in inert solvents containing various alcohols ranging
from methanol to cholesterol,29,30 although the involvement of
5a-C-centered radicals in the formation mechanism was not
evoked for explanation in these cases. Also, oxidation of
R-tocopherol in the presence of methyl linoleate in methanol
provided the 5a-methoxy derivative.31
The second observation cited as evidence for a radical
mechanism involving radical 5 is the frequent occurrence of
ethano-dimer 8, proposed to proceed by recombination of two
5a-C-centered radicals 5 (Scheme 4).27,32,33
The third fact that seemed to argue in favor of the occurrence
of 5 was the observation that reactions of R-tocopherol under
typical radical conditions, i.e., in the presence of radical initiators
in inert solvents or under irradiation, provided also large amounts
Alleged “Experimental Evidence” for R-Tocopherol-
Derived 5a-C-Radicals. Basically, three reactions were evoked
to support the occurrence of 5a-C-centered radicals in tocopherol
(24) Skinner, W. A. Biochem. Biophys. Res. Commun. 1964, 15, 469-
472.
(25) Fukuzawa, K.; Gebicki, J. M. Arch. Biochem. Biophys. 1983, 226,
242-251.
(29) Suarna, C.; Southwell-Keely, P. T. Lipids 1989, 24, 56-60.
(30) Suarna, C.; Southwell-Keely, P. T. Lipids 1998, 23, 137-139.
(31) Yamauchi, R.; Kato, K.; Ueno, Y. Lipids 1988, 23, 779-783.
(32) Rousseau-Richard, C.; Richard, C.; Martin, R. FEBS Lett. 1988,
233, 307-310.
(33) Nilsson, J. L. G.; Daves, D. G.; Folkers, K. Acta Chim. Scand. 1968,
22, 200-206.
(26) Tsuchija, J.; Niki, E.; Kamiya, Y. Bull. Chem. Soc. Jpn. 1983, 56,
229-232.
(27) Matsuo, M.; Matsumoto, S.; Iitaka, Y.; Niki, E. J. Am. Chem. Soc.
1989, 111, 7179-7185.
(28) Kamal-Eldin, A.; Appelqvist, L. A. Lipids 1996, 31, 671-701.
3270 J. Org. Chem., Vol. 72, No. 9, 2007

Vitamin E Chemistry
TABLE 1. Coupling Constants for Selected Positions in the
Chromanoxyl Ring System Predicted by Computations on the
B3LYP/6-31G(d,p) Level (Absolute Values)a
SCHEME 5. Hypothetical Disproportionation of Two
r-Tocopherol-Derived Radicals 2 and 5 in the Absence of
Other Coreactants To Account for the Formation of Typical
Two-Electron Oxidation Products (r-Tocopherol
Spiro-dimer 9)
position
nucleus
coupling constants (G)
6a
5a
5a
5
7a
7a
7
8b
8b
8
8a
4a
17O
10.81
6.49
3.32
12.12
4.52
2.53
8.66
1.68
0.68
7.21
13.90
8.20
3 H (av)
13C
13C
3 H (av)
13C
13C
3 H (av)
13C
13C
13C
13C
a In non-labeled compounds, only proton couplings are experimentally
detectable by EPR spectroscopy.
densities at the respective methyl carbon atoms. The compu-
tational results show no evidence whatsoever for a special
preference of 5a-C in forming a radical. The spin density values
indicate where radical attack or recombination reactions will
occur; for the methyl groups these processes mean H-atom
abstraction. The H-atom abstraction will be favored in the order
5a-CH3 > 7a-CH3 . 8b-CH3 but will be rather unlikely as
compared to reactions at 6-O and the aromatic ring carbons.
of two-electron oxidation products such as oQM 6 and its spiro-
dimerization product 9.21,32,33 This was taken as support of a
disproportionation reaction involving tocopheroxyl radical 2 and
its tautomer 5 (Scheme 5), affording one molecule of oQM 6
(oxidation) and regenerating one molecule of 1 (reduction). The
term “disproportionation” was used here to describe a one-
electron redox process with concomitant transfer of a proton,
i.e., basically an H-atom transfer from 5 to 2. In the following,
we would like to present an EPR study combined with labeling
experiments, which clearly speak against the involvement of
5a-C-centered radicals in tocopherol chemistry. This is followed
by evidence for the non-radical course of the above reactions
in Schemes 3-5 and explanation of their actual mechanism.
2.2. EPR, Computational, and Isotopic Labeling Study.
From the above-mentioned experimental observations and
theoretical considerations the question arises why 5a-C-centered
radical 5, if indeed in equilibrium with tocopheroxyl radical 2,
as often proposed, cannot be reported by EPR spectroscopy. In
a first step, we approached this question by quantum chemical
calculations.34 The spin distribution data obtained by these
calculations predicted high spin density for the phenolic oxygen
atom of chromanoxyl radicals; however, this cannot be directly
observed by EPR in natural 16O compounds (see Table 1).
Similarly high values were found for the carbon atoms of the
aromatic ring, especially in para- and ortho-positions, which
is in accordance with the observed chemistry, i.e., occurrence
of coupling products at 6-O, 5-C, and 8a-C. The electron
densities at all substituent positions, such as the three methyl
groups 5a-CH3, 7a-CH3, and 8b-CH3, as well as the methylene
group 4-CH2, are significantly lower. The hypothetic coupling
values for 13C atoms at the 5a, 7a, and 8b positions were
predicted with 3.32, 2.53, and 0.68 G. Hydrogen coupling
constants for the methyl groups, averaged by their free rotation,
were found to be 6.49, 4.52, and 1.68 G for 5a-methyl, 7a-
methyl, and 8b-methyl, respectively. Large hydrogen coupling
constants in the methyl groups are associated with high spin
H-atom abstraction from the 5a-methyl group in tocopheroxyl
radical 2 would formally produce a biradical as a resonance
form of oQM 6 but not the 5a-C-centered radical 5, which has
an intact phenolic OH group. Radical 5 would rather be formed
by a tautomerism, i.e., a proton shift from the methyl group to
the oxygen, a [1,4]-sigmatropic proton shift that proceeds via a
cyclic transition state according to a concerted mechanism.
Computations readily showed that radical 5 was energetically
largely disfavored as compared to tocopheroxyl radical 2 by
0.029 H (78.10 kJ mol-1). This is in accordance with the EPR
results (see below) that provided no evidence as to the presence
of C-centered radicals, by giving the neat phenoxyl spectrum
of 2. However, the occurrence of 5 in small amounts and very
low steady-state concentrations cannot be excluded just from
this observation.
Moreover, computations and theoretical considerations pro-
vided clear evidence against a preference of position 5a over
position 7a as radical 2 is still an aromatic system. The radical
centered at 5a-C (structure 5) was calculated to be only
insignificantly more stable by 1.55 kJ mol-1 than the one
centered at 7a-C (structure 5a). In contrast, in the case of the
quinoid oQM structures the involvement of 5a-C in oQM
formation is clearly energetically favored over 7a-C (and of
course over 8b-C). This was demonstrated experimentally and
smoothly explained by the strain-induced bond localization
(SIBL) approach6 (Scheme 6). Further confirmation comes from
computational results showing the 5a-oQM (6) to be more stable
by 21.65 kJ/mol than the 7a-oQM (6a). This clearly proved that
the conclusion by analogy as described in the introductory
section is wrong: the observed and experimentally well-
documented favoring of the 5a-C position in the formation of
o-quinone methides, i.e., quinoid structures as the result of a
two-electron oxidation, does not allow the conclusion that the
situation is similar for hypothetic R-tocopherol-derived C-
centered radicals, i.e., aromatic radical structures resulting from
one-electron oxidation. Thus, there are no theoretical consid-
(34) Xie, C. P.; Lahti, P. M.; George, C. Org. Lett. 2000, 2, 3417-
3420.
J. Org. Chem, Vol. 72, No. 9, 2007 3271

Rosenau et al.
SCHEME 6. Comparison of Computational Data (∆E, kJ
mol-1) for the Two Tautomeric o-Quinone Methides (6 and
6a) and the Hypothetic Two Tautomeric Radicals (5 and 5a)
Derived from the r-Tocopherol Model Compound
2,2,5,7,8-Pentamethylchroman-6-ol (PMC, R ) Me)a
a The data for the analogous species derived from R-tocopherol (1) can
be assumed to be nearly identical, since the influence of the side chain is
effectively cancelled out in relative values.
FIGURE 1. 13C-Labeled and non-labeled starting phenols and EPR
spectra of the respective phenoxyl radicals generated from 200 mM
phenol solutions in benzene by flash photolysis: (A) non-labeled
R-tocopheroxyl (2) from 1; (B) 5a-13C-labeled R-tocopheroxyl from
1*-5a; (C) non-labeled pentamethyl-chromanoxyl from PMC; (D) 5a-
13C-labeled pentamethylchromanoxyl from PMC*-5a.
erations supporting a dominance of radical 5 over the analogous
radical at 7a-C, contrary to the above-mentioned conventional
notion.
TABLE 2. Coupling Constants for the Hyperfine Splitting of EPR
Spectra from 13C Labeled and Non-labeled r-Tocopherol
(Chromanol) Compounds
Thus, if radicals at 5a-C and 7a-C are involved in tocopherol
chemistry, then products of both species would have to be
expected. The fact that in reality products of 5a-C are highly
preferred over those of 7a-C can already be seen as an indirect
proof of the underlying chemistry not being radical by nature.
To verify the theoretical predictions experimentally, we
synthesized R-tocopherol (1, 2R,4′R,8′R stereoisomer) 13C-
labeled at either 5a-C (1*-5a) or 7a-C (1*-7a) and also the 5a-
13C-labeled derivative (PMC*-5a) of its truncated model
compound 2,2,5,7,8-pentamethylchroman-6-ol (PMC) (see Fig-
ure 1 and Experimental Section). The syntheses involved a
Mannich reaction with 13C-paraformaldehyde and morpholine
leading to the 4-morpholino-(13C-methyl) derivatives which were
subsequently reduced by NaBH3CN to the 13C-methyl products.
From the labeled compounds the corresponding radicals were
generated by flash photolysis in benzene. The spectra of the
5a-13C-labeled radicals in comparison to non-labeled material
are shown in Figure 1.
The comparison of the spectra clearly showed the additional
coupling of the 13C atom in the labeled derivatives, and
otherwise slightly different sets of coupling constants, which
required detailed examination. For the analysis of the EPR
spectra, the WINSIM program was used,35 which iteratively
optimizes the coupling constants using the values from quantum
chemical calculations as starting conditions. The quality of the
obtained fits was assessed by regression coefficients calculated
by the simulation program. For labeled and unlabeled toco-
pheroxyl and chromanoxyl radicals the data are given in
Table 2.
no. of
set spin spins
2
5a-13C-2 7a-13C-2 PMC• 5a-13C PMC•
1
2
3
4
5
6
R
0.5
0.5
0.5
0.5
0.5
0.5
3
3
3
1
1
1
6.016
4.527
0.942
1.463
1.435
0.057
0.917
5.835
4.541
0.912
1.419
1.558
0.984
0.922
5.912
4.498
0.919
1.246
1.538
0.929
0.753
6.009
4.627
0.983
1.553
1.552
0.110
0.987
5.817
4.542
0.881
1.407
1.561
1.016
0.925
methyl groups, as well as with the benzylic 4-CH2 group of the
pyrano ring, which is in agreements with previous findings.36
Analogous experiments with the 13C-labeled compounds resulted
in almost identical coupling constants for those groups. How-
ever, because of the additional coupling of the 13C atom with
its nuclear spin of 1/2, a changed hyperfine splitting was observed
and iteratively calculated (Figure 1B and D; see also Table 2).
The additional coupling constant amounted to about 1 G
irrespective of whether the 5a- or the 7a-position had been
labeled. To perform spectral simulations with a uniform set of
coupling nuclei, an additional hypothetic coupling was intro-
duced also for non-labeled compounds. The respective simula-
tions always converged to values close to 0 G for this additional
coupling, proving its absence in unlabeled compounds and thus
confirming our assignment. As expected, the spin distribution
at other positions of the molecule is nearly unaffected by the
labeling. The quantum chemical calculations predict for both
positions 5a-C and 7a-C considerably higher coupling constants
For non-labeled R-tocopherol and non-labeled PMC radicals
characteristic couplings were found with the 5a-, 7a-, and 8b-
(36) Gregor, W.; Grabner, G.; Adelwo¨hrer, C.; Rosenau, T.; Gille, L. J.
Org. Chem. 2005, 70, 3472-3483.
(35) Duling, D. R. J. Magn. Reson., Ser. B 1994, 104, 105-110.
3272 J. Org. Chem., Vol. 72, No. 9, 2007

Vitamin E Chemistry
SCHEME 7. General Mechanism for the Spin Trapping of
C-Centered Radical Species (Primary Alkyl Radicals) by
PBNa
a The hyperfine structure of the spin adducts is usually dominated by
the coupling of the â-proton and of nitrogen.
than experimentally observed, this discrepancy probably being
related to the basis set chosen, which was not especially
optimized for calculations of hyperfine couplings.
FIGURE 2. Formation and decay of chromanoxyl radicals formed
from PMC by flash photolysis in benzene. The experiment was
performed with 33 mM PMC alone (0) or with 33 mM PMC and 33
mM PBN together (4). Each data point is a mean ( SD of three
separate measurements.
To verify the hypothetic involvement of C-centered radicals
arising from R-tocopherol or PMC, the spin trapping technique
in combination with EPR was employed. It is well-known that
nitroxide spin traps efficiently scavenge C-centered radicals in
the absence of oxygen. For the reaction of aliphatic carbon-
centered radicals with the spin trap N-tert-butyl-R-phenyl-nitrone
(PBN), high rate constants of about 107 L mol-1 s-1 were
reported.37 Also the formation of PBN spin adducts with
aromatic C-centered radicals is evident from the numerous
reports on spin adducts of phenyl radicals.38-40 So far no PBN
adducts with chromanoxyl radicals have been described. There-
fore, if C-centered R-tocopherol-derived or PMC-derived radi-
cals are produced in the presence of PBN, it is safe to assume
that they would be efficiently trapped (cf. Scheme 7). Moreover,
if the 5a-C-centered radical 5 indeed existed in a tautomeric
equilibrium with the tocopheroxyl radical 2, trapping products
of 5 would accumulate as the radical is constantly removed out
of the equilibrium by trapping and regenerated according to the
equilibrium constant.
SCHEME 8. Photolysis of Benzyl Bromide (r-13C-Benzyl
Bromide) in the Presence of PBN in Benzene and Formation
of the Corresponding Benzyl Radical Adducts (10 and 10*,
Respectively)
the decay curves should show significant differences, with the
decay of the chromanoxyl being noticeably accelerated by the
trap constantly removing the C-centered radical out of the
equilibrium with the chromanoxyl radical. Thus, the findings
lead to the conclusion that carbon-centered chromanol radicals
do not exist in equilibrium with the chromanoxyl radical.
However, to make this argumentation safe and secure, it must
be demonstrated that such 5a-C-centered radicals can indeed
be trapped readily by PBN once they are formed. Only in this
case can the absence of trapping products really be taken as
proof of absence of the underlying radicals.
To address the topic of detectability of C-centered radicals
derived from PMC or R-tocopherol by PBN trapping, the
generation of the radical species under similar reaction condi-
tions, i.e., by flash photolysis in benzene at ambient temperature,
was required. In a first step, benzyl bromide was employed as
a rather simple model compound with a photolabile bond for
carbon (benzyl) radical formation. Despite its structural simplic-
ity, the model possesses the basic structural elements of radical
5: the aromatic nucleus and a benzylic position to simulate the
5a-position in R-tocopherol/PMC (Scheme 8).
The trapping experiment was performed by irradiating a
solution of PMC in benzene in either the presence or absence
of PBN as the spin trap inside the cavity of the EPR
spectrometer. In the first phase, a flash before each EPR scan
was triggered until steady state was reached after about 90 s.
Subsequently, the decay of the chromanoxyl radical was
observed without irradiation (Figure 2). The decay curves were
identical within the error limits of the measurement, independent
of the presence or absence of PBN. Analysis of the half-lives
of the corresponding decay curves by a simple first-order decay
model gave half-lives of 37.2 ( 2.0 s for PMC alone and 35.0
( 2.5 s for PMC with equimolar amounts of PBN present, the
decay rates measured in three sets of experiments being
statistically identical. In addition, after the complete decay of
the chromanoxyl radical, no PBN spin adduct of carbon-centered
radicals was observed (spectra not shown), although such
adducts are known to have long half-lives of hours and even
days.41 In the presence of carbon-centered radicals, the corre-
sponding spin adducts should have been readily detectable, and
(37) Greenstock, C. L.; Wiebe, R. H. Can. J. Chem. 1982, 60, 1560-
1564.
The EPR spectra obtained by photolysis of benzyl bromide
in the presence of PBN are shown in Figure 3A. The simulation
of the respective hyperfine splitting of the benzyl radical adduct
of PBN (10) gave aN ) 14.480 ( 0.014 G, aHâ ) 2.455 (
0.021 G, and a1/2 ) 0.124 ( 0.009 G (n ) 2). Repetition of the
experiment with R-13C-labeled benzyl bromide afforded the
corresponding labeled adduct 10* (Figure 3B), the coupling
(38) Janzen, E. G.; Coulter, G. A.; Oehler, U. M.; Bergsma, J. P. Can.
J. Chem. 1982, 60, 2725-2733.
(39) Kandror, I. I.; Gasanova, R. G.; Freidlina, R. K. Tetrahedron Lett.
1976, 14, 1075-1078.
(40) Alekperov, G. A.; Zubarev, V. E.; Slovetskii, D. I. Khim. Vys. Energ.
2006, 13, 188-189.
(41) Janzen, E. G.; Chen, G.; Bray, T. M.; Reinke, L. A.; Poyer, J. L.;
McCay, P. B. J. Chem. Soc., Perkin Trans. 2 1993, 11, 1983-1989.
J. Org. Chem, Vol. 72, No. 9, 2007 3273

Rosenau et al.
FIGURE 4. EPR spectra of spin adducts obtained from the UV
irradiation of chromanol derivatives in the presence of PBN (100 mM)
in benzene: (A) 630 mM 12, (B) 630 mM 12*; (C) 630 mM 11; (D)
630 mM 11*.
TABLE 3. Coupling Constants of PBN Spin Adducts Obtained
from UV Irradiation of Labeled and Non-labeled Chromanol
Derivativesa
FIGURE 3. EPR spectra of spin adducts obtained upon UV irradiation
of benzyl bromide (900 mM) and PBN (100 mM) in benzene using
310 nm cut off filter: (A) adduct from benzyl bromide (10); (B) adduct
from R-13C-benzyl bromide (10*).
aN (g)
14.765 ( 0.007 2.365 ( 0.007 1.229 ( 0.013 0.066 ( 0.076
12* 14.585 ( 0.021 1.950 ( 0.042 0.930 ( 0.000 3.969 ( 0.042
11
a
Hâ (G)
a1/2,other (G)
a1/2,other (G)
12
SCHEME 9. PBN-Trapping of 5a-C-centered Radicals
Generated by Photolysis of the Corresponding
5a-Bromo-derivatives (12, 12*) and Lack of Radical
Generation from the Phenyl Acetates (11, 11*)
13.628 ( 0.004 1.907 ( 0.037
11* 13.627 ( 0.017 1.881 ( 0.050
a All data are means ( standard deviation of 2-3 independent experi-
ments.
bromine according to a non-radical, two-step oxidation-addition
mechanism.42 The bromo derivative was subsequently acetylated
with acetic anhydride under acid catalysis.
Upon irradiation of these compounds under otherwise identi-
cal conditions the EPR spectra as shown in Figure 4 were
obtained. Simulation of the respective EPR spectra afforded the
sets of coupling constants given in Table 3. Irradiation of
solutions containing non-labeled (11) and labeled PMC acetate
(11*) in the presence of PBN provided two similar EPR spectra
(Figure 4C and D). Simulations revealed nearly identical
coupling constants (Table 3), clearly showing the absence of
an additional coupling constant for the 13C-labeled analogue.
Thus, the 5a-position of the acetylated chromanols did not
couple with the unpaired electron in the observed spin adducts,
and thus the presence of a 5a-carbon centered radical of the
acetylated chromanols was safely excluded. The nature of the
spin adducts in these experiments was not completely elucidated,
but the presence of phenyl adducts from benzene is likely. The
slightly different coupling constants compared with those
observed by irradiation of PBN in benzene alone might be due
to polarity changes, such as the presence of PMC-Ac, cf.
Experimental Section. The presence of another trapped radical
species cannot be ruled out, even though this species was
definitely no 5a-C-radical, which would show the mentioned
additional coupling because of the labeling.
constants being aN ) 14.497 ( 0.016 G and aHâ ) 2.443 (
0.012 G with an additional coupling constant a13C ) 3.614 (
0.008 G (n ) 2). From these coupling constants and also from
the simple appearance of the spectra, the effect of the labeling
became obvious. The data unambiguously demonstrated the
formation of a C-centered radical, the benzyl radical, under the
conditions chosen.
In the case of the 6-O-acetyl-5a-bromo derivatives of PMC,
both labeled (12*) and in natural isotopic composition (12), the
5a-C-centered radicals were readily trapped with PBN (Figure
4A and B). The additional coupling in the case of 12* was an
unambiguous proof of the formation of 5a-C-centered radicals
in that particular case of photochemical generation from suitable
benzyl bromide precursors. The value of the additional spin
This approach toward identification of C-centered radicals
was transferred to the phenolic compounds of interest. For this
purpose, the O-acetyl derivatives of non-labeled and labeled
PMC (11 and 11*, respectively) were synthesized by acid-
catalyzed esterification with acetic anhydride. Similarly, the 6-O-
acetyl-5a-bromo derivative of PMC (Scheme 9) was synthesized,
both with natural isotopic composition at 5a-C (12) and 5a-
13C-labeled (12*). In a first step, PMC was quantitatively
converted into 5a-bromo-PMC by treatment with elemental
(42) Rosenau, T.; Habicher, W. D. Tetrahedron 1995, 51, 7919-7926.
3274 J. Org. Chem., Vol. 72, No. 9, 2007

Vitamin E Chemistry
coupling constant of about 3.9 G originating from the 13C
nucleus is similar to the additional coupling constant observed
with 13C-labeled benzyl bromide (3.6 G); see Scheme 8 and
Figure 3. These spectra showed distinctly different splitting
patterns for the 13C-labeled compound in comparison to the
unlabeled molecule.
SCHEME 10. Major Products of the Reaction of
r-Tocopherol (1) with Dibenzoyl Peroxide; Alternative
Heterolytic Formation Pathway for 5a-r-Tocopheryl
Benzoate (7) without Involvement of 5a-C-Centered Radicals
and Its Proof by Trapping of o-quinone Methide
Intermediate 6
In summary, it was demonstrated on the one hand that
chromanol-derived 5a-carbon-centered radicals, if formed from
suitable precursor compounds, can readily and easily be detected
by PBN-trapping in combination with EPR. In our study such
radicals were generated by breakage of the photolabile bond in
5a-bromo derivatives. The ease and apparent sensitivity of
trapping, on the other hand, allows the reliable conclusion that
no such radicals were formed if no corresponding trapping
products were found, as in the case of R-tocopherol (1), PMC,
R-tocopheryl acetate, or PMC acetate (11). The experiments thus
provided sufficient evidence against the occurrence of R-toco-
pherol-derived C-centered radical species, such as 5, and against
a tautomeric equilibrium between tocopheroxyl radical 2 and
benzylic radical 5.
2.3. Disproving the “Experimental Evidence” for the
Occurrence of C-Centered Radical 5. The formation of 5a-
R-tocopheryl benzoate (7) upon reaction of R-tocopherol (1)
with dibenzoyl peroxide has usually been taken as “solid proof”
of the involvement of 5a-C-centered radicals in tocopherol
chemistry (see Scheme 3). However, there are already some
general considerations that render a radical reaction pathway
to 7 questionable. First, the yields of 5a-R-tocopheryl benzoate
(7) can be as high as 30-40%, which seems astonishingly high
for a nonselective radical process. Second, it is clear that radicals
are involved in the overall process, since dibenzoyl peroxide is
a typical radical initiator. It undergoes homolytic cleavage of
the O-O bond to form two benzoyloxy radicals that subse-
quently fragment into CO2 and phenyl radicals. By reaction of
R-tocopherol with these radicals, tocopheroxyl radical 2 with
its resonance form 2′ is formed, which undergoes typical radical
coupling reactions at positions 6-O and 8a-C (cf. 3 and 4 in
Scheme 1, with R′ and R′′ being phenyl and benzoyl). Thus,
mainly R-tocopheryl phenyl ether (13) and R-tocopheryl ben-
zoate (14) are formed by reaction of 2 with benzoyl(oxy) and
phenyl radicals, respectively. Also, smaller amounts of 8a-
benzoyloxytocopherone (15) and 8a-phenyloxytocopherone (16)
were found, originating from recombination of those two radicals
with the 8a-C-centered resonance form 2′. Interestingly, for
position 5a only benzoate substitution and no phenyl substitution
was found. Even if one assumes a slower reaction of the
hypothetical tautomeric 5a-C-radical 5 with the phenyl radical
than with the benzoyloxy radical as in the case of 8a-C, there
is no obvious reason why large amounts of benzoyl coupling
product but absolutely no 5a-phenyl coupling product was
detected.
phenyl derivative and readily explains the observed products
without having to involve radical 5.
To conclusively disprove the involvement of that species, we
conducted the reaction of R-tocopherol with dibenzoyl peroxide
in the presence of a large excess of ethyl vinyl ether used as a
solvent component. Radical polymerization and recombination
reactions of ethyl vinyl ether are disfavored and relatively slow.
Therefore, if 5a-R-tocopheryl benzoate (7) was formed ho-
molytically according to Scheme 3, the presence of ethyl vinyl
ether should have no large influence on the product distribution.
However, if 7 was formed heterolytically according to Scheme
10, the intermediate oQM 6 would be readily trapped by ethyl
vinyl ether in a hetero-Diels-Alder process with inverse electron
demand,44 thus drastically reducing the amount of 7 formed.
Exactly the latter outcome was observed experimentally. In fact,
using a 10-fold excess of ethyl vinyl ether relative to R-toco-
pherol and azobis(isobutyronitrile) (AIBN), no 5a-R-tocopheryl
benzoate (7) at all was formed but rather the corresponding
trapping product 17 was produced, whereas the coupling
products of R-tocopheroxyl radical 2, i.e., compounds 13 and
14 and its 8a-C-resonance structure, i.e., compounds 15 and
16, were found as in the absence of the trap. Thus, it was shown
that the formation of 7 proceeded via oQM 6 without involve-
ment of 5a-C-centered radical 5, and the alleged conclusiveness
of the AIBN reaction pro radical 5 was disproved.
(43) In the case of a stepwise process (loss of one electron and one
proton), an phenoxonium cation intermediate might well be involved, which
has recently been shown to be quite stable: Lee, S. B.; Willis, A. C.;
Webster, R. D. J. Am. Chem. Soc. 2006, 128, 9332-9333. Also the
conversion of p-quinoid into o-quinoid structures in the chemistry of
R-tocopherol involves such an intermediate, e.g., the reaction of p-tocopheryl
quinone with acetyl halide or trimethylsilyl halide to O-protected 5a-halo-
tocopherols: (a) Dallacker, F.; Eisbach, R.; Holschbach, M. Chem. Ztg.
1991, 115, 113-116. (b) Rosenau, T.; Habicher, W. D. Tetrahedron Lett.
1997, 38, 5959-5960.
(44) This trapping reaction has been frequently applied in vitamin E
chemistry, its kinetics and products being well-known; cf. for instance: (a)
Rosenau, T.; Potthast, A.; Elder, T.; Lange, T.; Sixta, H.; Kosma, P. J.
Org. Chem. 2002, 67, 3607-3614. (b) Rosenau, T.; Potthast, A.; Elder, T.;
Kosma, P. Org. Lett. 2002, 4, 4285-4288. (c) References 6 and 41.
Starting from such considerations, we were proposing an
alternative, heterolytic formation mechanism for 7, which does
not involve 5a-C-centered radical species 5: the initiator-derived
radical products generate R-tocopheroxyl radicals (2) from
R-tocopherol (1). The radicals 2 are further oxidized to
o-quinone methide 6 in a formal H-atom abstraction,43 thereby
converting benzoyloxy radicals to benzoic acid and phenyl
radicals to benzene. The generated oQM 6 will add benzoic
acid in a [1,4]-addition process, whereas it cannot add benzene
in such a fashion. This pathway accounts for the observed
occurrence of benzoate 7 and simultaneous absence of a 5a-
J. Org. Chem, Vol. 72, No. 9, 2007 3275

Rosenau et al.
SCHEME 11. Formation of Ethano-dimer 8 by Reduction
of Spiro-dimer 9 in Different Reaction Systems;
5a-C-Centered Radicals Are Not Involved in This Process
starting R-tocopherol, could in theory act as the reductant.
However, incorporation of the spiro-dimer with tert-butylhy-
droperoxide gave epoxidation products and hydroperoxides, but
no ethano-dimer. Also R-tocopherol itself did not reduce the
spiro-dimer to the ethano-dimer. However, when we incorpo-
rated the spiro-dimer with double molar amounts of AIBN and
R-tocopherol, 22% of the ethano-dimer was found, whereas with
only AIBN no reduction took place. This demonstrated that the
reduction of the spiro-dimer by intermediate tocopheroxyl
radicals indeed proceeded. When we used double molar amounts
of 2,6-dimethylphenol/AIBN as a source of phenoxyl radicals,
as much as 46% of the spiro-dimer was converted into the
ethano-dimer, besides polymeric material from dimethylphenol.
It should be noted that the yield of ethano-dimer from spiro-
dimer could not be increased proportionally with the amount
of phenoxyl radicals generated by AIBN/phenol. With increasing
overall radical concentrations, side reactions generally become
more prominent, and also the ethano-dimer will be increasingly
reoxidized to the spiro-dimer. These results unambiguously
demonstrated that ethano-dimer 8 in hydroperoxide reaction
mixtures of R-tocopherol was not formed by recombination of
5a-C-centered radical 5, but according to a more complex
pathway involving the reduction of the spiro-dimer by R-toco-
pheroxyl or similar phenoxyl radicals (Scheme 11). Thus, also
the second alleged proof of the occurrence of 5 in tocopherol
chemistry did not hold.
A second observation, used to support the occurrence of an
R-tocopherol-derived 5a-C-centered radical, is the formation of
R-tocopherol ethano-dimer 8 as the result of radical recombina-
tion of two molecules of radical 5 (see Scheme 4). The major
components of typical reaction mixtures of R-tocopherol with
hydroperoxides in inert solvents are 6-O-tocopheryl ethers and
8a-tocopherones (3 and 4 in Scheme 1), spiro-dimer 9, ethano-
dimer 8, spiro-trimers,30 and epoxidized products.26,28,29,45,46
Typical conditions, which we also used in our experiments, are
an equimolar ratio of tert-butylhydroperoxide and R-tocopherol
in chloroform at room temperature. Under these conditions,
ethano-dimer 8 was obtained in 2.4% yield. Already the
relatively large amount of ethano-dimer 8 produced (2.4% means
that actually 4.8% of the starting R-tocopherol were converted
into the dimer) seems to collide with the proposed pathway,
since the recombination of two hypothetical 5a-C-centered
radicals would be expected to be a rather rare event. If the
formation of 8 indeed proceeded by radical recombination, then
both the increase of competing radicals and high dilution should
disfavor the recombination process and decrease the amount of
8 formed. However, repetition of the reaction in the 10-fold
amount of solvent did not result in a yield drop: 2.5% of the
ethano-dimer was formed. We repeated the reaction with the
initial concentrations but with a 10-fold amount of hydroper-
oxide, which was now used in ratio of 10:1 relative to 1, and
the amount of ethano-dimer even increased from 2.4% to 3.2%.
Finally, the reaction was conducted with both a 10-fold amount
of solvent and a 10:1 reagent ratio, affording 3.5% of the ethano-
dimer. Thus, neither dilution nor the presence of competitive
radicals had a depressing effect on the yield of 8, rendering a
radical formation pathway highly unlikely.
The last reaction commonly evoked to support the involve-
ment of radical species 5 in tocopherol chemistry is the
disproportionation of two molecules into the phenol R-toco-
pherol and the o-quinone methide 6 (Scheme 5), the latter
immediately dimerizing into spiro-dimer 9. This dimerization
is actually a hetero-Diels-Alder process with inverse electron
demand. It is largely favored, which is also reflected by the
fact that spiro-dimer 9 is an almost ubiquitous product and
byproduct in vitamin E chemistry.47,48 The disproportionation
mechanism was proposed to account for the fact that in reactions
of tocopheroxyl radical 2 generated without chemical coreac-
tants, i.e., by irradiation, the spiro-dimer 9 was the only major
product found.
As all of our evidence gathered so far disproved the existence
of 5, an alternative pathway (Scheme 12) was proposed, which
did not involve the dubious 5a-C-centered radical 5 but instead
the two well-documented and theoretically sound structures
tocopheroxyl 2 and 2′ as its major canonic form. According to
general tocopherol chemistry (cf. Scheme 1), 2 and 2′ will
recombine in the absence of other coreactants to afford a labile
8a-R-tocopheryl-tocopherone (18), which undergoes [1,4]-
elimination to afford R-tocopherol (1) and o-quinone methide
6, by analogy to other 8a-tocopherones.49-51 oQM 6, once
formed, will immediately dimerize into 9 in inert media. Thus,
this pathway explains the observed product readily on the basis
of general tocopherol chemistry without the need to evoke the
5a-C-centered radical 5.
Starting from these results, we proposed an alternative
pathway (Scheme 11), according to which the ethano-dimer 8
is formed by reduction of the spiro-dimer 9. This reduction
process is well-known in tocopherol chemistry; ethano-dimer
and spiro-dimer have been shown to establish a reversible redox
pair. In the present hydroperoxide reaction system, three species,
the hydroperoxide, intermediate R-tocopheroxyl radicals, and
(47) Schro¨der, H.; Netscher, T. Magn. Reson. Chem. 2001, 39, 701-
708.
(48) Schudel, P.; Mayer, H.; Metzger, J.; Ru¨egg, R.; Isler, O. HelV. Chim.
Acta 1963, 46, 636-649.
(45) Matsumoto, S.; Matsuo, M.; Iitaka, Y.; Niki, E. J. Chem. Soc., Chem.
Commun. 1986, 14, 1076-1077.
(46) Suarna, C.; Baca, M.; Southwell-Keely, P. T. Lipids 1992, 27, 447-
453.
(49) Du¨rckheimer, W.; Cohen, L. A. Biochem. Biophys. Res. Commun.
1962, 9, 262-265.
(50) Omura, K. J. Org. Chem. 1989, 54, 1987-1990.
(51) Kohar, I.; Southwell-Keely, P. T. Redox Rep. 2002, 7, 251-255.
3276 J. Org. Chem., Vol. 72, No. 9, 2007

Vitamin E Chemistry
SCHEME 12. Pathway for the Observed
SCHEME 13. Disproportionation of Phenoxyl Radicals
Derived from Thiotocopherol 19; 5a-C-Centered Radicals
Are Not Involved in This Process
Disproportionation of Tocopheroxyl Radical 2 into
r-Tocopherol (1) and r-Tocopherol Spiro-dimer (9);
5a-C-Centered Radicals Are Not Involved in This Process
Intermediate 18 proved to be rather elusive.52 In fact an
intermediate of similar polarity as compared to 9 was detected
by TLC in irradiation mixtures of 1, but several standard
isolation and purification attempts failed. Evidently, the com-
pound was decomposed both on silica gel and on all acid types
of alumina upon chromatographic purification. Finally, we used
a column of finely powdered potassium carbonate, which upon
elution with hexane bound the excess of starting material while
letting the tocopherone pass. The reaction had to be performed
directly in C6D6 as the NMR solvent. CDCl3 as the solvent
caused immediate degradation of 18 likely as a result of traces
of DCl generated, and so did any attempt of complete solvent
removal, even at lower temperatures. The HMBC spectrum of
18 showed characteristic typical long range (5J) connectivities
of the methyl group protons at 2.11 and 2.12 ppm (the 5a and
7a methyl groups in the tocopheryl moiety in 18, corresponding
to 13C resonances at 11.8 and 12.0 ppm) to a carbon at 96.4
ppm (the quinone ketal in the tocopherone part of 18). Heating
of the solution of 18 to temperatures above 50 °C or contact
with traces of acids caused immediate formation of R-tocopherol
(1) and spiro-dimer 9, confirming the proposed elimination-
dimerization mechanism (Scheme 12). Upon degradation of 18,
no 5a-(R-tocopheryl)-R-tocopherol was found, hypothetically
formed by [1,4]-addition of R-tocopherol onto o-quinone
methide 6, which showed that the dimerization of 6 was by far
preferred over that competitive addition path.
tocopherone 18, where the link between the ketal carbon and
the aroxy-oxygen is the weakest bond undergoing ready
cleavage, the most fragile link in thioketal 20 is the C-S
linkage. If intermediate 20 was formed, products of this cleavage
should be detectable. Indeed, these products were found:
reaction of 19 under otherwise identical irradiation conditions
provided 8a-(R-thiotocopheryl)-thiotocopherone 20 by analogy
to 8a-(R-tocopheryl)-tocopherone 18 formed from R-tocopherol
(1). By heating this intermediate to 50 °C or by treatment with
acid the thioketal 20 was immediately degraded into a complex
product mixture with the main components benzothiepine 21
and bisaryl ether 22. By irradiation of thiotocopherol 20 at 50
°C benzothiepine 21 (33%) and bisaryl ether 22 (11%) were
directly formed along with smaller amounts of oligomers and
polymers, at a conversion of 28%. The intermediacy of 18 and
20, respectively, and the subsequent degradation of those
intermediates into the respective products confirmed the mech-
anism presented in Scheme 12. Thus, the alleged dispropor-
tionation proceeds a stepwise process (see Scheme 12) and not
just an alteration of oxidation states as the term suggests. It
involves as the first step a radical coupling of the O-centered
radical 2 with its major resonance form 2′ to give 8a-tocopheryl-
tocopherone (18), which is transformed into the final products
by purely heterolytic processes. The net outcome is the observed
disproportionation of two tocopheroxyl radicals 2 into phenol
1 (reduction) and a quinoid structure 6 or its dimerization
product 9, respectively (oxidation). 5a-C-Centered radical 5 as
hypothetic tautomer of R-tocopheroxyl radical 2 is not involved
in the whole sequence. Hence, also the third “proof” seemingly
supporting the occurrence of radical 5 was shown to be false.
A direct proof of the alternative mechanism according to
Scheme 12 involved thiotocopherol 19.53-56 The compound was
selected on the basis of the following consideration: the
phenoxyl radical generated from this substance would afford
8a-C-thioketal 20 as the primary radical coupling product (see
Scheme 13) by analogy to 8a-tocopherone 18. In contrast to
(52) For the 8a-methoxy-tocopherone. see: Omura, K. J. Org. Chem.
1989, 54, 1987-1990. For other 8a-alkoxy-tocopherones, see: (a) Liebler,
D. C.; Baker, P. F.; Kaysen, K. L. J. Am. Chem. Soc. 1990, 112, 6995-
7000. (b) Winterle, J.; Dulin, D.; Mill, T. J. Org. Chem. 1984, 49, 491-
495.
3. Conclusion
(53) Valashek, I. E.; Shakova, M. K.; Samokhvalov, G. I. Zh. Org. Khim.
1982, 18, 2497-2500.
In the present paper it has been shown that there is no
theoretical evidence for the 5a-C-centered radical 5. In literature
accounts, the preference of 5a-C over 7a-C in quinoid structures
(often wrongly referred to as the Mills-Nixon effect) has
mistakenly been transferred to the case of radical species.
However, as the Mills-Nixon effect does not explain the
(54) Robillard, B.; Hughes, L.; Slaby, M.; Lindsay, D. A.; Ingold, K. U.
J. Org. Chem. 1986, 51, 1700-1704.
(55) Zahalka, H. A.; Robillard, B.; Hughes, L.; Lusztyk, J.; Burton, G.
W. J. Org. Chem. 1988, 53, 3739-3745.
(56) Robillard, R.; Ingold, K. U. Tetrahedron Lett. 1986, 27, 2817-
2820.
J. Org. Chem, Vol. 72, No. 9, 2007 3277

Rosenau et al.
behavior of quinoid R-tocopherol-derived structures correctly
and was thus replaced by the strain-induced bond localization
(SIBL) approach,6,7 it can, of course, not be used for other
R-tocopherol-derived structures. However, even if one accepted
the Mills-Nixon theory, it is only applicable to quinoid
structures, two-electron oxidation products of R-tocopherol, and
cannot be used for radical systems, i.e., one-electron oxidation
products. Computations clearly show the favoring of 5a-C in
quinoid R-tocopherol-derived systems according to the SIBL
concept and equally well disprove any favoring of 5a-C in the
R-tocopherol-derived radicals.
The theoretical facts against formation of radical 5 as primary
oxidation intermediate of R-tocopherol were complemented by
analytical evidence. By combination of EPR spectroscopy with
isotopic labeling (13C), it was demonstrated that 5a-C-centered
radicals were not formed by one-electron oxidation of R-toco-
pherol or as a tautomer of tocopheroxyl radical 2. Whereas 5a-
C-centered radicals were readily generable by alternative
pathways, such as photochemical cleavage of 5a-bromo deriva-
tives, and were neatly detected by spin trapping, a similar
approach afforded no trapping products at all starting from
R-tocopherol. Hence, a tautomeric equilibrium between radical
2 and radical 5 was excluded.
Three reactions are commonly cited to support the occurrence
of the chromanol methide radical 5 in vitamin E chemistry: (1)
the reaction of R-tocopherol (1) with AIBN in inert solvents
affording 5a-R-tocopheryl benzoate (7) as a major product, (2)
the formation of the ethano-dimer (8) upon reaction of 1 with
hydroperoxides in inert solvents, and (3) the alleged dispropor-
tionation of the tocopheroxyl radical 2 with radical 5 leading
to regenerated R-tocopherol (1) and o-quinone methide 6 in the
form of its spiro-dimer 9.
It was shown that all three processes actually follow alterna-
tive mechanisms that do not involve radical 5. For each of those
reactions the corrected pathway was proven by trapping
methodology, kinetic studies, oxidation of special R-tocopherol
derivatives, or combinations of those methods. Thus, finally,
all alleged evidence in favor of the formation of the chromanol
methide radical 5 from R-tocopherol (1) or R-tocopheroxyl (2)
was refuted.
The questions whether 5a-C-centered radicals exist in the
oxidation chemistry of R-tocopherol and whether mechanisms
proposed in the early days of vitamin E research are correct
might appear academic at a first glance, but as soon as one
recalls the immense medical, physiological, and economic
importance of tocopherol compounds, the significance of an
exact knowledge about their basic chemistry becomes obvious,
and so this study should be seen as an attempt to advance general
understanding in tocopherol chemistry. However, the results of
this study have also a direct and immediate impact on two fields
of tocopherol research: the synthesis of 5a-substituted deriva-
tives for medicinal or pharmacological applications57 and the
synthesis of polytocopherols as antioxidants, organic magnets,
and conductors in material science.55 For both fields the presence
or, more precisely, absence of 5a-C-centered radicals derived
from R-tocopherol is a key issue, as those species would imply
highly undesired effects, such as side reactions with biologically
active compounds or cell structures in the first case or cross-
linking reactions and chain breakage in the second case. By
analogy to the Mills-Nixon theory in tocopherol chemistry
having been replaced by the SIBL concept recently, the
condoned involvement of 5a-C-centered radicals 5 in oxidation
reactions of R-tocopherol must be revised. The alternative
heterolytic pathways as outlined in this study, shown to afford
the observed “radical” products, will have to find their general
acceptance in tocopherol chemistry.
4. Experimental Section
4.1. EPR Experiments. Sample Preparation. Solutions of PBN
and the respective tocopherols/chromanols were prepared in dry
benzene. The solutions were filled into quartz tubes (5.5 mm o.d.)
and connected via a valve to a vacuum manifold, which was flushed
with argon prior to the experiment. Under light protection, samples
were degassed by five consecutive freeze-thaw cycles. Finally,
the sample tube including the valve was disconnected from the
vacuum manifold and transferred to the EPR spectrometer.
EPR Measurements. Measurements of EPR spectra were
performed on a Bruker ESP 300E equipped with a TE102 standard
cavity. The sealed sample tubes were inserted into the resonator of
the EPR spectrometer, and the recording of the spectra was started
either immediately after flash illumination (12 W flash energy) or
after initiation of continuous irradiation by a mercury HBO 50 lamp.
Instrumental settings for measurement of chromanoxyl radicals were
as follows: 9.79 GHz; microwave power, 2 mW; receiver gain,
6.3 × 105; modulation frequency, 100 kHz; modulation amplitude,
0.2 G; scan rate, 343 G/min; time constant, 81 ms; temperature,
298 K; scans, 20. For spin trapping experiments: microwave
frequency, 9.2 GHz; microwave power, 2 mW; receiver gain, 5 ×
105; modulation frequency, 100 kHz; modulation amplitude, 0.25
G; scan rate, 85 G/min; time constant, 81 ms; temperature, 298 K;
scans, 10. Control experiments in the solvent benzene revealed that
despite a 310 nm cutoff filter used during irradiation a partial
photolysis of benzene leading to phenyl radicals and their corre-
sponding spin adducts cannot be avoided. This spin adduct was
simulated by the coupling constants aN ) 14.301 ( 0.044 G and
aHâ ) 1.826 ( 0.029 G (n ) 3).
Simulation of EPR Spectra. EPR spectral files from the ESP
300E were imported into the WINSIM program.34 As starting
approximation, data from quantum chemical calculations were used
and subsequently fitted to the experimental spectra until at least a
regression coefficient of >0.9 was achieved. Always different sets
of coupling constants were tested, and the best model was chosen
for iterations. The presence of couplings from 13C-labeled atoms
was verified by simulation of EPR spectra from unlabeled radicals
with a respective dummy coupling constant, which approached a
value close to zero.
4.2. Computations. For full geometry optimization the widely
employed B3LYP hybrid method was used, as implemented through
the GAUSSIAN58 and SPARTAN59 program packages, which
includes a mixture of HF and DFT exchange terms and the gradient-
corrected correlation functional of Lee, Yang, and Parr60,61 para-
metrized by Becke,62,63 along with the double-ú split valence basis
(58) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A.; Stratman,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Barone, V.; Cossi, M.;
Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski,
J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J.
V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.;
Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng,
C. Y.; Nananyakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.;
Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon, M.;
Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.6; Gaussian Inc.:
Pittsburgh, 1998.
(59) Spartan Pro; Wavefunction, Inc.: Irvine, CA, 2004.
(60) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. 1988, B37, 785-789.
(61) Miehlich, B.; Savin, A.; Stoll, H.; Preuss, H. Chem. Phys. Lett. 1989,
157, 200-206.
(57) Rosenau, T. In Encyclopedia of Vitamin E; Preedy, V., Watson R.,
Eds.; CABI Publishing: Oxford, 2006; Chapters 3 and 6.
(62) Becke, A. D. Phys. ReV. 1988, A38, 3098-3100.
3278 J. Org. Chem., Vol. 72, No. 9, 2007

Vitamin E Chemistry
set 6-31+G*,64,65 which includes diffuse functions. Transition states
and minima were confirmed by analysis of the calculated vibrational
spectrum and by intrinsic reaction coordinate analysis. For all
transition states the number of imaginary frequencies was 1; for
all minimum geometries it was 0.
dihydro-6-hydroxy-2,2,7,8-tetramethyl-1(2H)-benzopyran66 (1.870
g, 9.06 mmol) in 6 mL of absolute ethanol was added via syringe
and heated to reflux for 10 h. After cooling to -4 °C the precipitated
product was recrystallized from ethanol to provide 5-morpholino-
(13C-methyl)-2,2,7,8-tetramethyl-chroman-6-ol (2.46 g, 89%) as a
1
4.3. Synthesis of 5a-13C-R-Tocopherol (1*-5a). Mannich
Reagent. To morpholine (1.05 g, 12.06 mmol) was added portion-
wise 13C-paraformaldehyde (380 mg, 12.26 mmol, 1.02 equiv),
stirring in such a manner that the temperature reached a maximum
of 80 °C (exothermic reaction). The reaction was stirred for an
additional 3 h at 80 °C, obtaining a colorless solution with a small
white solid (mp 135-136 °C). H NMR: δ 1.27 (s, 6H, 2a-CH3
and 2b-CH3), 1.76 (t, 2H, 3-CH2, J ) 6.8 Hz), 2.10 (3H, s, 7a-
CH3), 2.14 (s, 3H, 8b-CH3), 2.56-2.63 (m, 6H, 4-CH2 and N-CH2-
CH2), 3.64 (d, 2H, 13CH2, JC,H ) 134.4 Hz), 3.74 (s, 4H,
O-CH2),10.56 (s, br, 1H, OH). 13C NMR: δ 11.7 and 11.9 (7a-
CH3 and 8b-CH3), 20.8 (4-CH2), 26.6 (2a-CH3 and 2b-CH3), 33.0
(3-CH2), 52.8 (N-CH2), 56.7 (5a-13C, 99% s, 1% m), 66.7 (O-CH2),
72.3 (2-C), 114.4 (d, 5-C, JCC ) 47.5 Hz,), 115.8 (C-4a), 122.7
and 125.2 (C-7 and C-8), 144.5 and 148.5 (C-6 and C-8a).
5a-13C-2,2,5,7,8-Pentamethylchroman-6-ol (PMC*-5a). A so-
lution of 5-morpholino(13C-methyl)-2,2,7,8-tetramethyl-chroman-
6-ol (0.597 mmol) in 2-butanol (2 mL) was heated to 70 °C and
NaBH3CN (4.776 mmol) was added. The mixture was heated to
reflux for 2 h and quenched with 2 M HCl while cooling in an ice
bath. The aqueous phase was extracted repeatedly with hexane. The
organic layers were combined, washed with saturated NaHCO3
solution and brine, and dried over MgSO4, and the solvent was
removed in vacuo. The crude product was purified by column
chromatography (EtOAc/hexane, v/v 1:25) to give 5a-13C-2,2,5,7,8-
pentamethylchroman-6-ol PMC*-5a (120 mg, 90% yield): mp 93-
96 °C. 1H NMR: δ 1.31 (s, 6H, 2a-CH3 and 2b-CH3), 1.81 (t, 2H,
1
amount of nonreacted paraformaldehyde. According to H NMR
data, the solution composition was 13C-morpholinomethanol (A,
MW ) 117) and 13C-dimorpholinomethane (B, MW ) 186) in an
1:1.1 ratio, corresponding to a 34:66 w/w ratio, and minor amounts
of polymeric components coming from paraformaldehyde. The
solution (referred to as “Mannich reagent” in the following) was
1
used for further manipulations without purification. H NMR: δ
2.48 (m, 8 H, NCH2CH2 in B), 2.68 (m, 4 H, NCH2CH2 in A),
2.87 (d, 2 H, JCH ) 138.8 Hz, 13CH2 in B), 3.68 (m, 12 H,
NCH2CH2O in A and B), 4.03 (d, 2 H, JCH ) 151.1 Hz, 13CH2OH).
5-Morpholino(13C-methyl)-(2R,4′R,8′R)-γ-tocopherol. Man-
nich reagent (340 mg) was added to (2R,4′R,8′R)-γ-tocopherol (420
mg, 1 mmol), and the resulting mixture was stirred at 80 °C for 4
h. After cooling to room temperature the mixture was diluted with
TBME (20 mL) and washed with H2O until neutral. The organic
layer was dried over K2CO3. After evaporating the solvent, the
residue was purified by column chromatography (hexane/EtOAc,
v/v 5:1) affording 5a-13C-5-morpholinomethyl-(2R,4′R,8′R)-γ-to-
copherol (480 mg, 93% yield) as dark yellow oil. 1H NMR: δ 0.7-
1.9 (m, 38 H, 3-CH2, 2a-CH3 and C16H33 chain), 2.09 (s, 3 H, 8b-
CH3), 2.13 (s, 3 H, 7a-CH3), 2.54-2.61 (m, 6 H, NCH2CH2 and
4-CH2), 3.62 (d, 2 H, 1JCH ) 134.2 Hz, N-13CH2-Ar), 3.73 (t, 4 H,
J ) 4.2 Hz, NCH2CH2O), 10.6 (s, br, 1 H, OH). 13C NMR: δ
12.0, 12.1, 19.4, 19.8, 20.7, 21.1, 22.7, 22.8, 23.7, 24.5, 24.8, 28.1,
31.6, 32.7, 32.8, 37.3, 37.4, 37.5, 37.6, 39.4, 39.9, 52.8, 56.7, 66.8,
74.3, 114.4, 116.0, 122.6, 125.2 (d, JCC ) 53.12 Hz), 144.4, 148.5.
3
3-CH2, J ) 6.9 Hz), 2.14 (d, 2H, 5a-13CH3, JC,H ) 126.5 Hz),
2.14 (s, 3H, 7a-CH3), 2.18 (s, 3H, 8b-CH3), 2.64 (t, 2H, 4-CH2, J
) 6.9 Hz), 4.21 (s, 1H, OH). 13C NMR: δ 11.2 (5a-13C, 99% s,
1% m), 11.7 (8b-CH3), 12.1 (7a-CH3), 21.0 (4-CH2), 26.7 (2a-CH3
and 2b-CH3), 33.0 (3-CH2), 72.5 (2-C), 117.1 (4a-C), 118.2 (d, JCC
) 43.5 Hz, 5-C) 121.1 (7-C), 122.5 (8-C), 144.6 (6-C), 145.7 (8a-
12
C). Anal. Calcd for
75.93; H, 9.04.
C
13CH20O2: C, 75.98; H, 9.11. Found: C,
13
6-Acetoxy-5a-13C-2,2,5,7,8-pentamethylchroman (11*). To a
solution of PMC*-5a (0.92 g, 4.164 mmol) in dry dichloromethane
(30 mL) were added acetic anhydride (2.36 mL, 25.0 mmol) and
three drops of concentrated sulfuric acid. The mixture was stirred
in a closed flask at room temperature overnight, quenched with
saturated NaHCO3 solution, stirred for another 10 min, and extracted
with dichloromethane. The combined organic layers were washed
two times with brine and dried over MgSO4, and the solvent was
removed in vacuo. The crude product was purified by crystallization
from EtOAc/hexane, v/v 1:10, to give 11* (710 mg, 65% yield):
5a-13C-(2R,4′R,8′R)-R-Tocopherol (1*-5a). To a solution of
5-morpholino(13C-methyl)-(2R,4′R,8′R)-γ-tocopherol (470 mg, 0.91
mmol) in iBuOH (5 mL) was added NaBH3CN (260 mg, 4.1 mmol,
4.5 equiv), and the resulting mixture was refluxed under stirring
for 4 h. Then 3 M HCl was added (8 mL) and the acid aqueous
phase was extracted with Et2O (3 × 10 mL). The combined organic
phases were washed with NaHCO3 (sat.) and H2O and dried over
Na2SO4. After evaporating the solvent, the residue was purified by
column chromatography (hexane/EtOAc, v/v 9:1), affording 5a-
13C-(2R,4′R,8′R)-R-tocopherol (370 mg, 94% yield) as a yellow oil.
1H NMR: δ 0.7-1.9 (m, 38 H, 3-CH2, 2a-CH3 and C16H33 chain),
1
mp 100-102 °C. H NMR: δ 1.23 (s, 6H, 2a-CH3 and 2b-CH3),
3
1.71 (t, 2H, 3-CH2, J ) 6.8 Hz), 1.91 (d, 2H, 5a-13CH3, JC,H
)
127.2 Hz), 1.95 (s, 3H, 7a-CH3), 2.02 (s, 3H, 8b-CH3), 2.25 (s,
3H, CH3CO), 2.54 (t, 2H, 4-CH2, J ) 6.8 Hz). 13C NMR: δ 11.2
(5a-13CH3, 99% s, 1% m). Anal. Calcd for 12C1513CH22O3: C, 72.97;
H, 8.42. Found: C, 73.31; H, 8.49.
2.15 (s, 3 H, 8b-CH3), 2.19 (s, 3 H, 7a-CH3), 2.14 (d, 2 H, 1JCH
)
126.3 Hz, 5a-13CH3), 2.63 (t, J ) 6.6 Hz, 2 H, 4-CH2), 4.27 (s, 1
H, OH). 13C NMR: δ 11.2, 11.5, 12.3, 19.7, 19.8, 20.7, 21.0, 22.6,
22.7, 23.8, 24.5, 24.7, 27.9, 31.5, 32.6, 32.7, 37.3, 37.4, 37.5, 37.6,
39.3, 39.7, 74.6, 117.6, 118.8 (d, JCC ) 45.08 Hz), 121.1, 122.6,
144.6, 145.6. NMR data are consistent with those from the
literature.56
6-Acetoxy-5-(bromo-13C-methyl)-2,2,7,8-tetramethyl-chro-
man (12*). Elemental bromine (147 µL, 2.862 mmol) in dry hexane
(60 mL) was added quickly to a solution of PMC*-5a (2.806 mmol)
in the same solvent (150 mL). The mixture was stirred at room
temperature for 1.5 h in a closed flask, and then the formed HBr
was removed by evaporation under reduced pressure while the
mixture was still stirred. The solvent was removed in vacuo to
obtain 5-(bromo-13C-methyl)-2,2,7,8-tetramethylchroman-6-ol quan-
titatively, which was subsequently dissolved in dry dichloromethane.
Acetic anhydride (1.59 mL, 16.836 mmol) and three drops of
concentrated sulfuric acid were added, and the mixture was stirred
in a closed flask at room temperature overnight. The mixture was
quenched with water, stirred for another 10 min, and extracted with
hexane/dichloromethane (v/v 1:2). The combined organic layers
were washed with saturated NaHCO3 solution and two times with
brine and dried over MgSO4, and the solvent was removed in vacuo.
4.4. Synthesis of 5a-13C-2,2,5,7,8-Pentamethylchroman-6-ol
(PMC*-5a), 6-Acetoxy-5-(bromo-13C-methyl)-2,2,7,8-tetrameth-
ylchroman (11*), and 6-Acetoxy-5a-13C-2,2,5,7,8-pentamethyl-
chroman (12*). 5-Morpholino(13C-methyl)-2,2,7,8-tetramethyl-
chroman-6-ol. A mixture of 13C-paraformaldehyde (327 mg,
10.88 mmol) and morpholine (1.81 mL, 13.05 mmol) was heated
to 70 °C and a solution of the γ-tocopherol model compound 3,4-
(63) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652.
(64) Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta 1973, 28, 213-
222.
(65) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon,
M. S.; DeFrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654-3665.
(66) Prepared according to literature, giving literature-consistent NMR
data: Yenes, S.; Messeguer, A. Tetrahedron 1999, 55, 1411-1422.
J. Org. Chem, Vol. 72, No. 9, 2007 3279

Rosenau et al.
TABLE 4. Reduction of Spiro-dimer 9 to Ethano-dimer 8 by
Different Phenol/Radical Initiator Systems
The crude product was purified by column chromatography (EtOAc/
hexane, v/v 1:10) to give 6-acetoxy-5-(bromo-13C-methyl)-2,2,7,8-
tetramethylchroman 12* (750 mg, 78% overall yield): mp 118-
radical initiator
phenol
yield of 8 [%]
1
121 °C. H NMR: δ 1.31 (s, 6H, 2a-CH3 and 2b-CH3), 1.83 (t,
t-BuOOH, 5 mmol
none
none
0
0
23
0
22
28
16
26
21
46
44
3
2H, 3-CH2, J ) 6.9 Hz), 2.01 (s, 3H, 7a-CH3), 2.11 (s, 3H, 8b-
R-tocopherol (1), 5 mmol
R-tocopherol (1), 5 mmol
none
R-tocopherol (1), 5 mmol
R-tocopherol (1), 10 mmol
R-tocopherol (1), 5 mmol
2,6-dimethyl-phenol, 5 mmol
2,6-dimethyl-phenol, 5 mmol
2,6-dimethyl-phenol, 10 mmol
2,6-dimethyl-phenol, 20 mmol
CH3), 2.39 (s, 3H, CH3CO), 2.79 (t, 2H, 4-CH2, J ) 6.9 Hz), 4.39
t-BuOOH, 5 mmol
AIBN, 5 mmol
AIBN, 5 mmol
AIBN, 5 mmol
AIBN, 10 mmol
AIBN, 5 mmol
AIBN, 10 mmol
AIBN, 10 mmol
AIBN, 20 mmol
(d, br, 2H, 13CH2Br, JC,H ) 151.6 Hz). 13C NMR: δ 25.67 (5a-13-
12
CH3, 99% s, 1% m). Anal. Calcd for
C
15
13CH21BrO3: C, 56.15;
H, 6.18; Br, 23.34. Found: C, 55.96; H, 6.23.
4.5. Studies into the Reaction of R-Tocopherol with Dibenzoyl
Peroxide. Reaction of R-Tocopherol with Dibenzoyl Peroxide.
In an Ar atmosphere, a degassed solution of dibenzoyl peroxide
(0.870 g, 3.6 mmol) in dry chloroform (30 mL) was quickly added
to a degassed solution of R-tocopherol (1.29 g, 3.0 mmol) in the
same solvent (15 mL). The mixture was placed into a water bath
at 40 °C and stirred for 1 h. The mixture was cooled to room
temperature, and the solvent was concentrated in vacuo to a volume
of about 3 mL. The yellow, oily residue was chromatographed on
neutral alumina (hexane/EtOAc, v/v 5:1) to afford in the order of
elution R-tocopheryl phenyl ether (13, 18%), R-tocopheryl benzoate
(14, 22%), 8a-phenyltocopherone (16, 8%), 8a-benzoyloxytoco-
pherone (15, 4%), and 5a-benzoyloxy-R-tocopherol (7, 34%) along
with minor byproducts that were not separated. The procedure was
repeated three times; the above yields are averaged. An aqueous
workup must be avoided as the tocopherones are quite unstable
compounds, and compound 7 is degraded in basic media. The
analytical data of products 13-16 were consistent with previously
published data67 and are therefore not repeated here. The data of 7
are listed for ready comparison.
dimer (8). Reaction of R-Tocopherol with tert-Butylhydroper-
oxide. tert-Butylhydroperoxide was used as ∼5.5 M water-free
solution in decane as distributed by Sigma-Aldrich. In an Ar
atmosphere, a solution of tert-butylhydroperoxide in decane (1.8
mL, ∼10 mmol) was added at once to a solution of R-tocopherol
(4.30 g, 10.0 mmol) in hexane (50 mL). The mixture was placed
into an oil bath at 70 °C and stirred for 5 h. The content of ethano-
dimer 8 was determined by HPLC. For preparative separation and
purification of this compound, the solution was evaporated to a
volume of about 5 mL and chromatographed on basic alumina,
eluting the non-phenolic compounds (spiro-dimer 9, tocopheryl
ethers, tocopherones) with (hexane/EtOAc, v/v 9:1). Elution with
hexane/EtOAc, v/v 4:1 provided dimer 8 as a yellow oil (210 mg,
2.4%), followed by nonreacted R-tocopherol (8%) and phenolic 5a-
substituted R-tocopherols. The reaction was repeated according to
the above procedure in 500 mL of hexane, affording 8 (215 mg,
2.5% yield). Using 18 mL of tert-butylhydroperoxide solution under
otherwise identical conditions (50 mL of hexane), 275 mg (3.2%)
of 8 was obtained. With the same amount of hydroperoxide in 500
mL of hexane, the yield of 8 was similar (302 mg, 3.5%).
Benzoic Acid 6-Hydroxy-2,7,8-trimethyl-2-(4,8,12-trimethyl-
tridecyl)-chroman-5-ylmethyl ester (5-benzoyloxy-R-tocopherol,
1
7). H NMR: δ 0.72-1.89 (m, 38 H, 3-CH2, 2a-CH3 and C16H33
chain), 2.10 (s, 3H, 7a-CH3), 2.15 (s, 3H, 8b-CH3), 2.60 (t, 2H, 3J
) 6.8 Hz, 4-CH2), 5.30 (s, 2H, 5a-CH2), 6.95 (m, 5H, ArCH), 10.23
(s, 1H, OH). 13C NMR: δ 11.9 (8b-CH3), 12.4 (7a-CH3), 17.9 (4-
CH2), 23.6 (2a-CH3), 33.6 (3-CH2), 59.4 (5a-CH2), 74.2 (2-C), 114.8
(4a-C), 115.5 (5-C), 122.0 (7-C), 123.4 (8-C), 127.1 (4-C in Ph),
129.6 (2-C and 6-C in Ph), 130.6 (1-C in Ph), 132.6 (3-C and 5-C
in Ph), 144.6 (6-C), 145.5 (6-C), 166.3 (COO). Anal. Calcd for
C21H24O4: C, 74.09; H, 7.11. Found : C, 74.23; H, 7.18.
Reaction of R-Tocopherol with Dibenzoyl Peroxide in the
Presence of Trapping Agents. In an Ar atmosphere, a solution of
dibenzoyl peroxide (0.870 g, 3.6 mmol) in dry, degassed chloroform
(30 mL) was quickly added to a degassed solution of R-tocopherol
(1.29 g, 3.0 mmol) in chloroform (10 mL) and ethyl vinyl ether
(20 mL). The mixture was placed into a water bath at 40 °C and
stirred for 1 h. Every 15 min, 5 mL of ethyl vinyl ether was added.
The mixture was cooled to room temperature, and the solvent was
concentrated in vacuo to a volume of about 10 mL. The yellow,
oily residue was chromatographed on neutral alumina (hexane/
EtOAc, v/v 9:1) to afford in the order of elution R-tocopheryl phenyl
ether (13, 14%), the trapping product of the o-quinone methide
(17, 38%), R-tocopheryl benzoate (14, 17%), 8a-phenyltocopherone
(16, 8%), and 8a-benzoyloxytocopherone (15, 6%), along with
minor byproducts that were not separated. The procedure was
repeated two times; the above yields are averaged. An aqueous
workup must be avoided as in the above case. More polar eluant
will cause interference with polymerization products of the vinyl
ether upon chromatographic separation.
1,2-Bis(5-γ-tocopheryl)-ethane (R-tocopherol ethano-dimer,
1
8). H NMR: δ 0.7-1.9 (m, 38 H, 3-CH2, 2a-CH3 and C16H33
3
chain), 2.12 (s, 4 × 3H, 7a-CH3 and 8b-CH3), 2.65 (t, 4H, J )
6.7 Hz, 4-CH2), 3.70-4.30 (s, br, 2H, OH). 13C NMR: δ 12.2 (8b-
CH3), 12.4 (7a-CH3), 20.6 (4-CH2), 23.7 (2a-CH3), 24.8 (5a-CH2),
33.8 (3-CH2), 75.5 (2-C) 117.1 (4a-C), 121.5 (5-C), 121.7 (7-C),
122.8 (8-C), 144.6 (6-C), 145.8 (8a-C). Anal. Calcd for C58H98O4
(859.42): C, 81.06; H, 11.49. Found: C, 81.15; H, 11.44.
An authentic sample of 8 was prepared by refluxing 5a-bromo-
R-tocopherol with the 8-fold molar amount of Fe(CO)9 in hexane
for 1 h. Yields were quantitative, requiring no chromatographic
purification. Anal. Calcd for C58H98O4 (859.42): C, 81.06; H, 11.49.
Found: C, 81.12; H, 11.53.
Reduction of Spiro-dimer 9 by Intermediate Phenoxyl
Radicals to Ethano-dimer 8. To a solution of R-tocopherol spiro-
dimer 9 (0.20 g 0.23 mmol) in hexane were quickly added a solution
of a radical initiator (10 mL of hexane, 0.5 mmol) and the solution
of a phenol in the same solvent (10 mL pro 0.5 mmol). The mixture
was placed into an oil bath at 70 °C and stirred for 1 h. The content
of ethano-dimer 8 was determined by HPLC. The types and amounts
of initiator and phenols used together with the yields of ethano-
dimer 8 obtained are given in Table 4.
4.7. Studies into the Disproportionation of Tocopheroxyl
Radicals. Irradiation of R-Tocopherol in Inert Solvents. A
solution of R-tocopherol (2.15 g, 5 mmol) in dry, perdeuterated
benzene (50 mL) was irradiated for 10 h by a mercury HBO 200
W lamp under external cooling at 10 °C. At this temperature, the
mixture was concentrated to a volume of 5 mL and chromato-
graphed at room temperature on powdered anhydrous K2CO3 using
C6D6 as the eluant. After minor amounts of tocopheryl phenyl ether
and other non-identified non-pehnolic byproducts, 8a-R-tocopheryl-
tocopherone (18) was eluted. The fraction containing 18 was
concentrated at 0 °C to about 0.5 mL and directly measured by
8-Ethoxy-3,5,6-trimethyl-3-(4,8,12-trimethyl-tridecyl)-1,2,3,8,9,-
10-hexahydro-pyrano[3,2-f]chromene (17). The NMR and mass
spectroscopic data of product 17 were consistent with previously
published data.41,68 Anal. Calcd for C33H56O3: C, 79.14; H, 11.27.
Found : C, 79.08; H, 11.34.
4.6. Studies into the Formation of R-Tocopherol Ethano-
(67) Rosenau, T.; Habicher, W. D. J. Prakt. Chem. 1996, 338, 647-
653.
(68) Bolon, D. A. J. Org. Chem. 1970, 35, 3666-3671.
3280 J. Org. Chem., Vol. 72, No. 9, 2007

Vitamin E Chemistry
NMR. The amount of 18 formed (32 mg, 1.4%) was calculated
from the amount of 1 and 9 formed by acid treatment.
At 28% conversion of thiotocopherol 19, yields of benzothiepin
21 and thiol 22 were 209 mg (33%) and 72 mg (11%), respectively.
Preparation of Thiotocopherol 19. Thiotocopherol 19 was
prepared according to the literature51,53 in satisfactory purity. The
NMR data agreed with one literature report53 but showed some
small deviations from another one.54
Irradiation of Thiotocopherol 19 in Inert Solvents at 10 °C.
Irradiation of thiotocopherol 19 and purification of the reaction
intermediate 20 (38 mg, 1.7%) was performed according to the
above procedure.
Acknowledgment. The financial support of the Fonds zur
Fo¨rderung der wissenschaftlichen Forschung (FWF), project
P17428 to T.R., is gratefully acknowledged. The authors thank
Dr. A. Hofinger for acquisition of th NMR spectra and DI G.
Ebner for practical assistance.
Irradiation of Thiotocopherol 19 at 70 °C. Working according
to the above irradiation procedure but at 50 °C afforded 21 and 22
as degradation products of 20 directly. Also thermal treatment of
20 at 80 °C for 1 min or treatment with 1 drop of TFA at room
temperature provided a compound mixture with 21 and 22 as the
main components. A solution of thiotocopherol 19 (2.23 g, 5 mmol)
in dry cyclohexane (50 mL) was irradiated for 10 h by a mercury
HBO 200 W lamp at 50 °C. The greenish mixture was concentrated
to a volume of 5 mL and chromatographed on basic alumina using
an hexane/EtOAc eluant (v/v 5:1) deacidified by filtration over K2-
CO3 shortly before use. After two minor non-identified byproducts
and nonreacted starting material, benzothiepin 21 eluted as a yellow
oil. Changing the eluant to hexane/EtOAc eluant (v/v 3:1) afforded
the thiol 22 as colorless waxy solid. The reaction was repeated
three times with the retrieved starting material of the previous run.
Supporting Information Available: Preparation and NMR data
(1H, 13C) of 7-morpholino(13C-methyl)-(2R,4′R,8′R)-â-tocopherol
and 7a-13C-(2R,4′R,8′R)-R-tocopherol (1*-7a). NMR (1H, 13C) and
purity data of 8a-R-tocopheryl-tocopherone (18), 2,5,7,8-tetra-
methyl-2-(4,8,12-trimethyltridecyl)-thiochroman-6-ol (19), 8a-R-
thiotocopheryl-thiotocopherone (20), 3,7,8-trimethyl-6-[2,5,7,8-
tetramethyl-2-(4,8,12-trimethyltridecyl)-thiochroman-6-yloxy]-3-
(4,8,12-trimethyltridecyl)-1,3,4,5-tetrahydrobenzo[c]thiepin-9-ol (21),
and 3-(3-mercapto-3,7,11,15-tetramethylhexadecyl)-2,5,6-trimethyl-
4-[2,5,7,8-tetramethyl-2-(4,8,12-trimethyltridecyl)-thiochroman-6-
yloxy]-phenol (22). Computational details for radicals 2, 5, and 5a
as well as o-quinone methides 6 and 6a. This material is available
free of charge via the Internet at http://pubs.acs.org.
JO062553J
J. Org. Chem, Vol. 72, No. 9, 2007 3281