Garlic: Source of the ultimate antioxidants - Sulfenic acids

Article abstract of DOI:10.1002/anie.200804560

(Chemical Equation Presented) The medicinal properties of garlic, thought to derive at least in part from the antioxidant activity of its sulfur-containing secondary metabolites, have been recognized for hundreds of years. The ability of garlic to scavenge peroxyl radicals can be accounted for in terms of the action of transient sulfenic acids, which are predicted to react by diffusion-controlled five-center proton-coupled electron transfer (see scheme and transition state).

Full text of DOI:10.1002/anie.200804560

Angewandte
Chemie
DOI: 10.1002/anie.200804560
Radical Scavengers
Garlic: Source of the Ultimate Antioxidants—Sulfenic Acids**
Vipraja Vaidya, Keith U. Ingold, and Derek A. Pratt*
Garlic, onion, and other members of the Allium spp., long
believed to be of great medicinal benefit, contain up to about
5
% dry weight of nonprotein sulfur amino acid secondary
metabolites, such as (+)-S-allyl-l-cysteine S-oxide (alliin,
[
1]
1
). Alliin, which is found predominantly in garlic, is cleaved
constants for H-atom transfer from hydrocarbons to peroxyl
by alliinase upon the homogenization of garlic to form
ammonium pyruvate and 2-propenesulfenic acid [2, Eq. (1)].
The latter compound undergoes self-condensation to yield the
radicals are much lower than those reported for allicin-
inhibited autoxidation reactions, even when highly stabilized
carbon-centered radicals are formed. For example, the rate
constant for the reaction of peroxyl radicals with methyl
linoleate to give a highly delocalized pentadienyl radical is
[
2]
diallyl thiosulfinate allicin [3, Eq. (2)]. Allicin provides
garlic with its odor and flavor, and is believed responsible for
[
3]
[4]
À1 À1 [7]
its health benefits, often ascribed to antioxidant activity.
only 60m s . Second, carbon-centered radicals generally
undergo diffusion-controlled reactions with O2 to yield
peroxyl radicals, which continue to propagate the autoxida-
[
8]
tion chain reaction.
Amorati and Pedulli were similarly mystified by the
[
9]
suggested mechanism, which prompted them to investigate
the ability of diallyl disulfide (and allyl methyl disulfide) to
inhibit the controlled autoxidation of cumene and styrene. As
expected, they found these compounds to be ineffective
inhibitors, with second-order rate constants for the reaction
À1 À1
with peroxyl radicals of approximately 1m s . These results
[
5]
In a recent series of papers, Okada et al. presented
results of kinetic studies aimed at elucidating the mechanism
of allicinꢀs antioxidant activity. Therein, inhibited autoxida-
tion of methyl linoleate (ML) and cumene yielded inhibition
rate constants (kinh) for the reaction of allicin with methyl
reinforced our view that a mechanism other than the H-atom
transfer shown in Equation (3) must be responsible for the
radical-trapping activities of 3 and 4, but nevertheless a
mechanism that requires the S(O)SCH CH=CH and
2
2
S(O)SCH Ph moieties, respectively.
2
5
[10]
linoleate- and cumene-derived peroxyl radicals of 1.6 ꢁ 10
In 1972, Koelewijn and Berger
demonstrated that
3
À1 À1
[5b,6]
and 2.6 ꢁ 10 m s , respectively.
Since previous struc-
thermally unstable dialkyl sulfoxides were effective inhibitors
ture–activity studies had indicated that the S(O)SCH CH= of hydrocarbon autoxidation at 608C, because they decom-
2
CH2 moiety was essential for the antioxidant activity of
posed by a Cope elimination to yield a sulfenic acid [and an
olefin, Eq. (4)]. Indeed, from careful inhibited autoxidation
reactions it was estimated that 2-methyl-2-propanesulfenic
acid reacts with peroxyl radicals with a rate constant greater
[
5a]
allicin,
Okada et al. suggested a mechanism involving
abstraction of the allylic H atom adjacent to the divalent
sulfur atom [Eq. (3)]. Subsequent studies with the dibenzyl
thiosulfinate 4 [Eq. (3)] from Petiveria alliacae L. afforded
similar results, which prompted the suggestion that the
mechanism for its reaction with peroxyl radicals was abstrac-
tion of the analogous benzylic H atom.
7
À1 À1
than 10 m s , which suggests that sulfenic acids are among
[11]
the most potent classes of peroxyl-radical-trapping agents.
Unfortunately, few studies on sulfenic acids as antioxidants
have been reported in the intervening years.
The suggested mechanism for the scavenging of peroxyl
radicals by thiosulfinates is unlikely for two reasons. First, rate
[*] Dr. V. Vaidya, Prof. D. A. Pratt
Department of Chemistry, Queen’s University
90 Bader Lane, Kingston, ON, K7L3N6 (Canada)
Fax: (+1)613-533-6669
E-mail: pratt@chem.queensu.ca
Thiosulfinates also undergo Cope elimination to form
sulfenic acids, along with thioaldehydes or thioketones. This
process does not require elevated temperatures, because the
Dr. K. U. Ingold
National Research Council of Canada, Ottawa (Canada)
[1]
[
**] We are grateful for the support of the Natural Sciences and
Engineering Research Council of Canada and the Ontario Ministry
of Innovation. D.A.P. acknowledges the support of the Canada
Research Chairs program.
SÀS bond in a thiosulfinate is much weaker than the SÀC
bond in a sulfoxide. Cope elimination is even more facile for
allyl (and benzyl) thiosulfinates, such as allicin (and 4),
because of the weak b CÀH bond of the allyl (benzyl) moiety.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200804560.
Indeed, allicin is known to undergo Cope elimination readily
Angew. Chem. Int. Ed. 2009, 48, 157 –160
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
157

Communications
at room temperature to give 2-propenesulfenic acid (2) and
[
1]
thioacrolein [Eq. (5)].
Hence, we considered it probable that the antioxidant
activity ascribed to allicin (and garlic, in general) is actually
due to the trapping of peroxyl radicals by 2-propenesulfenic
acid—both its decomposition product [Eq. (5)] and its
precursor [Eq. (2)]! Likewise, a-toluenesulfenic acid is prob-
ably the inhibitor in autoxidation reactions involving the
analogous dibenzyl thiosulfinate 4.
If 2-propenesulfenic acid is responsible for the trapping of
peroxyl radicals in allicin-inhibited autoxidation reactions,
then the inhibition should be less effective under reaction
conditions which retard the decomposition of allicin. It is well
documented that hydrogen-bond-donor (HBD) solvents
[1,12]
retard allicin decomposition.
Accordingly, the half-life of
allicin (50 mm) in chlorobenzene was found to be approx-
imately 1 h, but very little of the allicin (a few percent)
decomposed in 1 h when hexafluoroisopropanol (HFPA,
0
.15m) was added to the solution in chlorobenzene (Fig-
ure 1b). Most importantly, when HFPA was added to the
autoxidation reaction mixture, the induction period (previ-
[
5]
ously ascribed to allicin itself) was eliminated (see Fig-
ure 1a). No such solvent effect would be observed if the
[5]
trapping of peroxyl radicals involved the (suggested)
abstraction of an allylic H atom. Clearly, allicin is not directly
responsible for the inhibition of ML autoxidation; instead, an
[13]
allicin decomposition product must be responsible.
We also explored the effects of added hydrogen-bond-
acceptor (HBA) solvents on allicin-inhibited autoxidation
reactions. Although the addition of CH CN (1m) to a solution
3
Figure 1. a) Thermally initiated (azobisisobutyronitrile (AIBN), 40 mm)
autoxidation of methyl linoleate (91 mm) at 378C in chlorobenzene
of allicin in chlorobenzene had little effect on the rate of
decomposition of allicin (Figure 1b), it markedly reduced the
ability of allicin to inhibit the autoxidation of ML (Figure 1a).
Thus, the use of freshly purified (HPLC) allicin in chloro-
containing allicin (50 mm) and HFPA (0.15m, *), CH
no additive (~). b) Decomposition of allicin (50 mm) at 378C in
CN (1m,
&
), or
3
chlorobenzene containing HFPA (0.15m, *), CH CN (1m,
&
), or no
3
additive (~).
benzene in the presence of CH CN (1m) led to an almost
3
eightfold increase in the rate of inhibited autoxidation, and
[14]
the stoichiometric factor dropped from n = 0.95 to 0.69.
Since allicin itself is not the radical-trapping antioxidant in
these autoxidation reactions, we cannot derive an inhibition
rate constant (kinh) from the autoxidation data in the custom-
[
9]
ary way; however, the drop in antioxidant activity in HBA
solvents is informative. This type of solvent effect is well
documented for reactions of peroxyl radicals with H-atom
[
15]
donors that are HBDs, such as phenols. The decomposition
product of allicin, 2-propenesulfenic acid, is expected to be a
strong HBD that will form strong hydrogen bonds with
Scheme 1. Mechanistic scheme to explain the observed increase in the
rate of allicin-inhibited autoxidation in the presence of acetonitrile.
CH CN (Scheme 1) and thus lead to a decrease in the rate of
3
the inhibited autoxidation, as observed.
Although these experiments demonstrate that 2-prope-
nesulfenic acid is highly likely to be the peroxyl-radical
scavenger in allicin-inhibited autoxidation reactions, the
reason that sulfenic acids appear to be such effective
phenols, X = Ar), is the key thermodynamic parameter
connected with the rate of reaction of XOH with peroxyl
radicals, it would be desirable to compare the OÀH BDE of
2-propenesulfenic acid and other sulfenic acids with the OÀH
[
10]
antioxidants
is currently unclear. Since the OÀH bond-
BDEs of their isoelectronic cousins, the hydroperoxides
(which react with peroxyl radicals much more slowly, with
dissociation enthalpy (BDE) of an H-atom donor, XOH (e.g.,
1
58
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 157 –160

Angewandte
Chemie
2
3
À1 À1
rate constants of approximately 10 –10 m s ). Unfortu-
nately, no sulfenic acid OÀH BDE has ever been reported. As
the reliable determination of a sulfenic acid OÀH BDE
appeared to be experimentally impractical (owing to the
transient nature of most known sulfenic acids), we calculated
these BDEs by using a method known to accurately predict
[
16]
OÀH BDEs, the complete-basis-set approach of Petersson
[17]
and co-workers (Table 1).
À1
Table 1: OÀH BDEs (in kcalmol ) calculated by the CBS-QB3 method
for some sulfenic acids and hydroperoxides.
[
a]
OÀH BDE
OÀH BDE
HSO-H
MeSO-H
73.1
68.4
68.6
68.6
68.6
HOO-H
MeOO-H
87.3
86.2
86.2
84.8
86.1
C H CH SO-H
C H CH OO-H
2
3
2
2
3
2
tBuSO-H
tBuOO-H
C H CH SO-H
C H CH OO-H
6
5
2
6
5
2
[18]
À1
[
a] G3 calculations yielded 72.8 and 68.2 kcalmol for HSO-H and
MeSO-H, respectively.
Figure 2. a) Cisoid and b) transoid transition-state structures for the
HSOH/COOH reaction. c) The four highest occupied molecular orbitals
of the structure in (a) showing the overlap between the sulfenic acid
S atom and the internal peroxyl O atom.
Table 1 reveals two important features of the OÀH BDEs
of sulfenic acids. First, the values are much lower than those
for the analogous hydroperoxides. For the simplest members
of the series (hydrosulfenic acid and hydrogen peroxide), the
À1
Table 2: Activation energies (in kcalmol ) calculated by the CBS-QB3
[
a]
À1
method for H-atom transfer between sulfenic acids and peroxyl
radicals.
difference is 14.2 kcalmol . For the alkanesulfenic acids and
alkyl hydroperoxides, the difference is even larger: almost
À1
1
8 kcalmol for methane-, 2-propene-, and a-toluenesul-
E
a
(cisoid TS)
E (transoid TS)
a
fenic acids and the corresponding hydroperoxides. Thus, the
sulfur atom imposes two effects: It stabilizes the radical to a
greater extent, largely by delocalizing the unpaired electron
onto itself (the unpaired spin distribution is close to 50:50
between the oxygen and sulfur atoms in sulfinyl radicals and
HSO-H/COOH
HSO-H/COOMe
MeSO-H/COOMe
6.7
7.7
4.6
4.3
4.2
13.6
15.1
11.0
10.9
9.8
CH =CHCH SO-H/COOMe
2
2
tBuSO-H/COOMe
7
0:30 between the terminal and nonterminal oxygen atoms in
[a] Determined from a common hydrogen-bonded prereaction complex.
peroxyl radicals), and this results in greater interactions with
substituents that are bonded to the sulfur atom. This effect
leads to a greater difference between hydrosulfenic acid and
HOMO-1 in the cisoid (Figure 2c) and transoid TS structures
are very similar, HOMO-2 and HOMO-3 are quite different
in the two TS geometries: There is significant bonding overlap
between the sulfenic acid sulfur atom and the internal oxygen
atom of the peroxyl radical in the cisoid structure, an
interaction that is absent in the transoid structure. This
bonding overlap suggests that the electron to be transferred
from the sulfenic acid to the peroxyl radical is partly localized
on the peroxyl radical in the cisoid TS and thus indicates a
mechanism based on proton-coupled electron transfer
À1
the alkanesulfenic acids (ca. 4.5 kcalmol ) than between
hydrogen peroxide and the alkyl hydroperoxides (ca. 1 kcal
À1
mol ). The OÀH BDEs of sulfenic acids are among the
weakest known and comparable to those of hydroxylamines,
À1
such as TEMPO-H (70.7 kcalmol ; TEMPO = 2,2,6,6-tetra-
[
19, 20]
methylpiperidin-1-oxyl).
To provide additional insight into the reactions of sulfenic
acids with peroxyl radicals, we also calculated the transition-
state (TS) structures (Figure 2) and associated activation
energies of some representative reactions (Table 2). Two TS
structures were identified: one with a cisoid geometry with
respect to the oxygen atoms between which the H atom is
being transferred (Figure 2a) and one with a transoid
geometry (Figure 2b). The cisoid TSs were found to be
[
21]
(PCET).
As in other PCETreactions,
[
21,22]
H-atom transfer between
a sulfenic acid and a peroxyl radical is predicted to involve
initial formation of a hydrogen-bonded complex, RSOH···
COOR’. For the reactions we investigated, in which R and R’
were alkyl groups, these complexes lie some 4.5 to 5.0 kcal
À1
lower in energy by approximately 6–7 kcalmol than the
À1
transoid TSs (Table 2). The highest occupied molecular
orbitals (HOMOs) of the cisoid and transoid TS structures
mol lower in energy than the separated reactants. This result
implies that the corresponding TSs are slightly lower in
energy than the separated reactants. These reactions are,
therefore, predicted to be diffusion-controlled, consistent
(
which comprise the four possible combinations of the two p*
orbitals on the sulfinyl and peroxyl fragments between which
the H atom is transferred) reveal why the cisoid geometry is
favored. Whereas the singly occupied (SO) HOMO and
[
10]
with Koelewijn and Bergerꢀs estimate that the rate constant
for the reaction of 2-methyl-2-propanesulfenic acid with
Angew. Chem. Int. Ed. 2009, 48, 157 –160
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
159

Communications
7
À1 À1
tetralyl peroxyl radicals was > 10 m s . Thus, sulfenic acids
[9] R. Amorati, G. F. Pedulli, Org. Biomol. Chem. 2008, 6, 1103.
10] P. Koelewijn, H. Berger, Recueil Trav. Chim. Pays-Bas. 1972, 91,
1275.
11] For other highly reactive radical-trapping antioxidants, see: M.
Wijtmans, D. A. Pratt, L. Valgimigli, G. A. DiLabio, G. F.
Pedulli, N. A. Porter, Angew. Chem. 2003, 115, 4506; Angew.
Chem. Int. Ed. 2003, 42, 4370; D. A. Pratt, G. A. DiLabio, G.
Brigati, G. F. Pedulli, L. Valgimigli, J. Am. Chem. Soc. 2001, 123,
4625.
[
[
are very probably the most potent of all peroxyl-radical-
[
23,24]
trapping antioxidants.
In conclusion, we suggest that the peroxyl-radical-trap-
ping activity of garlic is primarily due to 2-propenesulfenic
acid formed by the decomposition of allicin. It is highly likely
that the decomposition of thiosulfinates from other Allium
spp. (e.g., onion) gives rise to sulfenic acids that are equally
reactive towards peroxyl radicals. Since the reactions of
alkanesulfenic acids with peroxyl radicals are predicted to be
diffusion-controlled and to occur by a common proton-
coupled-electron-transfer mechanism, both the abundance
and stability of the thiosulfinate precursors may account for
the different antioxidant activities of extracts of different
species.
[
[
12] F. Freeman, Y. Kodera, J. Agric. Food Chem. 1995, 43, 2332.
13] Owing to its transient nature, 2-propenesulfenic acid is normally
detected indirectly, for example, following its reaction with ethyl
propiolate to yield ethyl (Z)-2-thioformyl-4-pentenoate S-
[1]
oxide. In our inhibited autoxidation reactions, the addition of
ethyl propiolate to the reaction mixture did not prevent
inhibition, presumably since the reaction of 2-propenesulfenic
acid with peroxyl radicals is much faster than its addition to ethyl
propiolate.
[
10]
[
14] Values of n < 1 led to the suggestion
that sulfenic acids
Received: September 16, 2008
Revised: October 10, 2008
undergo other reactions, such as reaction with oxygen, during
autoxidation processes.
Published online: November 28, 2008
[
[
15] G. Litwinienko, K. U. Ingold, Acc. Chem. Res. 2007, 40, 222.
16] See, for example: P. Mulder, H.-G. Korth, D. A. Pratt, G. A.
DiLabio, L. Valgimigli, G. F. Pedulli, K. U. Ingold, J. Phys.
Chem. A 2005, 109, 2647.
Keywords: allicin · antioxidants · garlic · lipid peroxidation ·
peroxyl radicals · proton-coupled electron transfer
.
[
17] J. A. Montgomery, Jr., J. W. Ochterski, G. A. Petersson, J. Chem.
Phys. 1994, 101, 5900.
[
1] E. Block, Angew. Chem. 1992, 104, 1158; Angew. Chem. Int. Ed.
Engl. 1992, 31, 1135.
[18] L. A. Curtiss, K. Raghavachari, P. C. Redfern, V. Rassolov, J. A.
Pople, J. Chem. Phys. 1998, 109, 7764.
[
2] C. J. Cavallito, J. H. Bailey, J. S. Buck, J. Am. Chem. Soc. 1945,
[19] L. R. Mahoney, G. D. Mendenhall, K. U. Ingold, J. Am. Chem.
6
7, 1032; A. Stoll, E. Seebeck, Adv. Enzymol. 1951, 11, 377.
Soc. 1973, 95, 8610.
À1 [19]
[
[
[
3] E. Block, Sci. Am. 1985, 252, 94.
4] See, for example, references 11–20 in Ref. [5b].
5] a) Y. Okada, K. Tanaka, I. Fujita, E. Sato, H. Okajima, Redox
Rep. 2005, 10, 96; b) Y. Okada, K. Tanaka, I. Fujita, E. Sato, H.
Okajima, Org. Biomol. Chem. 2006, 4, 4113; c) Y. Okada, K.
Tanaka, I. Fujita, E. Sato, H. Okajima, Org. Biomol. Chem. 2008,
[20] The measured value of 71.8 kcalmol
must be decreased by
À1
1.1 kcalmol because of a revised heat of formation of (E)-
[
16]
azobenzene.
[21] J. M. Mayer, D. A. Hrovat, J. L. Thomas, W. T. Borden, J. Am.
Chem. Soc. 2002, 124, 11142.
[22] G. A. DiLabio, K. U. Ingold, J. Am. Chem. Soc. 2005, 127, 6693.
[23] In contrast, the PCET pathway for hydroperoxide/peroxyl self-
exchange reactions is characterized by much higher activation
6, 1097.
[
6] Such a large difference in the rate constants of reactions of
radical-trapping antioxidants with secondary and tertiary per-
oxyl radicals is unprecedented. Also, the inhibition rate con-
stants were based on the assumption that one molecule of allicin
reacts with one chain-carrying peroxyl radical in the autoxida-
tion of each substrate (stoichiometric factor: n = 1), an assump-
tion that is invalid on the basis of the actual inhibition periods
observed,
approximately 0.3.
7] J. A. Howard, Adv. Free-Radical Chem. 1972, 4, 49.
8] Appropriately substituted carbon-centered radicals, such as that
derived from HP-136, are an exception; see: J. C. Scaiano, A.
Martin, G. P. A. Yap, K. U. Ingold, Org. Lett. 2000, 2, 899.
À1
energies, for example, 9.0 kcalmol for the MeOOH/COOMe
couple (or 6.0 kcalmol from the separated reactants, in very
À1
[
25]
good agreement with experiment ); therefore, the rate con-
stants for these reactions are much smaller (generally on the
2
3
À1 À1
order of 10 –10 m
s ).
[24] Sulfinic and sulfonic acids, the oxidation products of sulfenic
acids, would not be expected to be good peroxyl-radical
scavengers, since the lone pair of electrons on the sulfur atom
would either be too low in energy (former) or unavailable
(latter) for the five-center PCET reaction.
[
5b]
which correspond to stoichiometric factors of
[
[
[25] J. H. B. Chenier, J. A. Howard, Can. J. Chem. 1975, 53, 623.
1
60
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 157 –160