Professional Documents
Culture Documents
ABSTRACT
The relative antioxidant efficiency, expressed as electron donating capability, of a large
series of carotenoids has been studied using computational chemistry method at the hybrid
density functional theory with the B3LYP functional and 6-31G(d,p) basis set. Their reactivity
toward nine different radicals has been modeled as well as the electron transfer between pairs
of carotenoids, one of which is present as a radical cation. The influence of the solvent
polarity has also been studied. Torulene was found to be the most easily oxidized carotenoid,
followed by lycopene. This higher reactivity is proposed in the present work for the first time,
and the potential implications of such a finding are discussed. Since torulene has not been
previously studied, compared to other carotenoids in terms of oxidation potentials, further
experimental studies are suggested in order to confirm or reject this prediction. Ionization
potential seems to be a magnitude calculable at low computational cost that correctly predicts
the relative ease of oxidation in a series of carotenoids.
INTRODUCTION
Carotenoids (Car) are versatile organic molecules, which are of great relevance to living
systems. They serve multiple purposes for their producers as well as for their consumers.
Among their diverse functions [1], their free-radical-scavenging antioxidant properties stand
out [2,3]. However, their broadly defined antioxidant activity strongly depends on a mixture
valuable information for a particular application. This work will focus on the electron-transfer
(ET) processes, although hydrogen abstraction and adduct formation are also viable
The relative importance of these reaction channels will depend on diverse factors,
including the nature of the reacting free radical and the structural features of the carotenoid
[4,5] and, in biological systems, its location and orientation within the membrane [5].
Generally speaking, the antioxidant properties of carotenoids are related to their excellent
ability to deactivate excited states and to their high electron donation capability. Although
electron donation can immediately deactivate harmful radicals, the antioxidant mechanism
as Car•+. In addition, redox pairs involving Car/Car •+ conjugates [6] could regenerate the
parent molecule and eliminate the pro-oxidant concerns related to carotenoid radical cations.
The electron-transfer reaction (I) will clearly be favored when the R groups are electron
withdrawing. For example, the reaction of the trichloromethyl peroxyl radical (CCl 3O2) with
β-carotene [7] and several other carotenoids [8] has been suggested to involve electron
transfer to generate the corresponding anion and carotene radical cation. Everett and co-
workers [9] have also shown that β-carotene is oxidized by NO2• in solution via reaction I. In
a recent study using benzyl peroxyl radicals, it was concluded that carotenoids scavenge
peroxyl radicals, which are not highly reactive, by adduct formation and not by electron
transfer, while reaction III seems to be much less important than reaction II [10]. These results
are in line with a previous review article showing that carotenoids react with free radicals by
addition and/or electron-transfer reactions, with the reaction channel distribution varying
On the other hand, it seems obvious that there must be a relationship between the structure
of carotenoids and their reactivity toward free radicals. There are some reports addressing this
issue. Jeevarajan et al. [11] found that carotenoids substituted with electron donating groups
are more easily oxidized than those with electron accepting substituents. Mortensen and
Skibsted [12] have studied the reactions of eight carotenoids with phenoxyl radicals. These
authors report the following order of reactivity: lycopene (LYC) > β-carotene (BC) >
zeaxanthin (ZEA) > lutein (LUT) > echinenone (ECH). They also found that canthaxanthin
(CAN) and β-apo-8'-carotenal (APO) hardly react, while asthaxanthin (ASTA) does not react
at all. Edge et al. [13] have studied ET reactions between different pairs of carotenoids with
similar outcomes. They found that LYC is the most easily oxidized followed by BC > ZEA >
In addition, El-Agamey and McGarvey [14] have recently shown that the profile of
carotenoid radical products formed depends on the polarity of the solvent medium. According
to these authors, in nonpolar solvents, only addition radicals are formed, while in polar
The main aim of this work is to provide a computational strategy that allows reliable
carotenoids and oxygen carotenoids. Their reactivity toward nine different radicals has been
modeled through reaction I, as well as the electron transfer between pairs of carotenoids, one
of which is present as a radical cation. The influence of the solvent polarity has also been
investigated.
COMPUTATIONAL METHODS
For all of the modeled species, full geometry optimizations have been carried out using
the B3LYP hybrid density functional and the 6-31G(d,p) basis set, while frequency
optimized at the same level. Unrestricted calculations were used for open-shell systems, and
local minima were identified by the number of imaginary frequencies (NIMAG = 0). All of
the electronic calculations were performed with the Gaussian 98 suite of programs.
Thermodynamic corrections at 298 K were included in the calculation of the relative energies.
Solvent effects were included by using the polarizable continuum model (PCM) with water
and benzene as the solvents for polar and nonpolar environments, respectively. The Gibbs free
energies of reaction for an aqueous environment were calculated using a mixed discrete-
continuum model with two water molecules explicitly included, with a surrounding
continuum of bulk solvent for carotenoids + free radical reactions. The corresponding
energies were improved by single point (SP) calculations at the B3LYP/6-311++G(d,p) level
of theory.
Since there is previous knowledge about the relative antioxidant activity of seven
carotenoids [12], they have been included in the present work in order to test the reliability of
the results proposed here for the whole studied series. The structures of all modeled
carotenoids are shown in Tables 1 and 2. The simplest approach to study the ease of ET
processes from different molecules is to analyze their ionization potentials (IP). The IP of an
n-electron system (X), calculated at a given level of theory, implies the following energy
difference
EXn-1(gn-1) is the energy of the (n - 1)-electron ionic species calculated at a geometry gn-1.
Accordingly, to calculate IPs, two separate geometry optimizations are needed, performed
on the neutral and radical cationic species. The IP calculated this way is known as an adiabatic
Another approximation, known as the vertical IP, is to optimize the geometry of the X
species and then to perform the calculation of the radical cation at this very geometry. This
approach implies also two calculations but only one geometry optimization and one single
point to obtain the electronic energy of the Xn-1 species. A special case of a vertical IP is that
obtained within the framework of the Koopman’s theorem approximation. In this particular
case, the vertical IP is evaluated as minus the energy of the molecular orbital of the neutral
system from which an electron is removed, usually the highest occupied molecular orbital
(HOMO), provided that the potential is defined in such a way that it vanishes at infinity. This
is a very simple, relatively inexpensive way to estimate the IP from a single calculation on the
neutral system.
Even though the fast development of computers in the last few decades has remarkably
enlarged the potential use of computational quantum chemistry for modeling molecular
systems of increasing size at reliable levels of calculation, carotenoids can still be considered
find some parameters that describe their relative ease of electron transfer in a proper way, at
least qualitatively, and that, at the same time, could be calculated at a reasonable
computational cost. With that purpose in mind, the correlations of the adiabatic IP versus the
Koopman IP and the vertical IP versus the Koopman IP have been tested for all of the studied
carotenoids in polar and nonpolar media. As Figure 1 shows, there is a good correlation in all
of the tested cases. However, the agreement of Koopman’s IP with the more accurate ones is
better in polar solvents. According to these results, Koopman’s IP can be used to qualitatively
compare the ease of electron donation capability among a series of similar compounds. In the
specific case of carotenoids, it can be used as the most economical criteria (computationally
As the values in Table 3 show, the ionization potentials calculated within the vertical and
adiabatic approaches are lower for polar than those for nonpolar media, as it is expected since
a charged species is formed. In general, this effect was found to be less important for
hydrocarbon carotenoids than that for oxygen carotenoids. These results agree with those
reported by Sliwka et al. [15] who found that electron-transfer processes are enhanced in the
presence of water, especially for hydrophilic carotenoids. Koopman’s IPs do not show this
experimentally validated tendency; therefore, they are good enough to qualitatively compare
the ease in electron donation for a series of carotenoids, but they do not properly describe the
influence of the solvent in such process, at least when the continuum PCM model is used.
Among the different approaches used to calculate ionization potentials, special attention
should be paid to that referred to as the vertical IP. The ET reactions can be assumed to
mainly occur by electron tunneling from one species to another. Since electrons are much
faster than nuclei, the electronic transitions can be considered to take place in fixed nuclear
configurations. This process is expected to occur very fast, in such a way that at the moment
of the transfer, there is no time for geometry relaxation. Accordingly, while adiabatic IPs can
be related to the energy involved in the completion of the reaction, vertical IPs can be
associated with the energetic barrier. Vertical IPs can then be considered as an upper limit of
the activation barrier. According to these assumptions, vertical IPs would be better criteria to
predict which carotenoid would more readily transfer one of its electrons to another species
and then to relatively order a series of carotenoids in terms of their oxidation potentials. To
confirm this hypothesis, it is necessary to compare vertical and adiabatic IPs with the
available experimental data, which indicates the following order of reactivity in terms of the
ease of oxidation: LYC > BC > ZEA > LUT > ECH > CAN > ASTA [12] and LYC > BC >
ZEA > LUT > CAN > APO > ASTA [13]. For this particular sets of carotenoids, their relative
order, based on vertical IPs, is LYC > BC > ZEA > LUT > ECH > CAN > APO > ASTA,
which is in perfect agreement with the experimental findings. For this subset of carotenoids,
the order is the same in both solvents. On the other hand, the order based on adiabatic IPs is
LYC > BC > ZEA > LUT > ECH > APO > CAN > ASTA, which inverts the relative reactivity
pairs of carotenoids, their relative order, in term of oxidation potential, is shown in Figure 2
based on vertical IPs. This is a schematic representation, standing only for the order of the
ease of electron transfer. However, for some of these pairs, the difference may be small,
leading to slow reactions (e.g., DIH/ECH pair). The relative ease of electron transfer between
carotenoid pairs has been provided in polar and nonpolar media since these reactions can be
studied independently of their hydrophilicity only under special circumstances [12, 13].
However, in living systems, carotenoids are not likely to always react either in pure lipid or
a better knowledge of their behavior in such extreme environments, which is hard to obtain
from experiments, could be helpful since the physicochemical assessment of the carotenoid
radical scavenging behavior in aqueous solutions might be a key property in the formulation
of parental therapeutics.
Torulene was found as the most easily oxidized carotenoid in polar and nonpolar
environments. This can be a relevant finding since it implies that TOR is expected to repair
other damaged carotenoids through redox reactions between pairs Car2/Car1•+, regardless of
carotenoids, a more detailed study has been performed. The energy evolution associated with
the completion of ET reactions between carotenoid conjugated pairs has been computed as the
ΔG0ET = G(Car1•+, gn-1) + G(Car2, gn) - G(Car1, gn) - G(Car2•+, gn-1) (2)
where gn represents the relaxed geometry of the neutral carotenoid and gn-1 that of the radical
cation.
BC, LYC, and TOR have been modeled as the repairing carotenoids (Car 2). They have
been chosen based on the following criteria: BC is the most representative carotenoid of the
series; LYC has been predicted as the Car most easily oxidized from all previously studied
Car; and TOR has not been comparatively studied before, but according to our results, it
show that the processes involving electron transfer from neutral TOR to any of the other
studied radical cations are energetically favored in terms of Gibbs free energies, that is, they
are exergonic processes (ΔG < 0). Electron transfers from LYC are mostly exergonic, with the
logical exception of its reaction with TOR. When the electron transfer is modeled from BC,
the calculated ΔGs predict it as able to repair other Car•+ species than TOR, LYC, SAP, and
ZEA.
The possible electron-transfer reactions in nonpolar and polar environments have also
been modeled for carotenoids in Tables 1 and 2 and for the following radicals: hydrogen
peroxyl: HOO• (R1), methyl peroxyl: CH3OO• (R2), methoxyl: CH3O• (R3) benzyl peroxyl:
C6H5CH2OO• (R4), phenoxyl: C6H5O• (R5), acyl peroxyl: CH3C(O)OO• (R6), phenylacetyl
peroxyl: C6H5C(O)OO• (R7), trichloromethyl peroxyl: CCl3OO (R8), nitrite: NO2• (R9).
ΔG0ET = G(Car1•+, gn-1) + G(Ox-, gn+1) - G(Car, gn) - G(Ox•, gn) (3)
where Ox represent the oxidant species, that is, free radicals. The values obtained that way for
nonpolar media were found to be highly endergonic (ΔG from 25 to 80 kcal/mol) and,
accordingly, very unlikely to occur. On the other hand, the presence of polar solvent
drastically increased the reactivity of carotenoids toward the studied free radicals, through
electrontransfer reactions. Accordingly, it can be stated that in nonpolar media, the studied
radicals and probably most oxygenated radicals are bound to react by adduct formation
instead of electron transfer since the ionic products cannot be solvated, whereas in polar
media, carotenoid radical cations and the corresponding anions can be stabilized by the
surrounding molecules. In biological systems, these charged species could be solvated at the
lipid-water interface.
The carotenoids’ reactions with HOO•, CH3OO•, and C6H5-CH2OO• remain endergonic in
aqueous phase but with ΔG0ET more than 50 kcal/mol lower than those corresponding to
benzene as the solvent. On the other hand, the reactions of CH 3O•, C6H5O•, CH3C(O)OO•,
C6H5C(O)OO•, CCl3OO•, and NO2• become exergonic (Table 5). Accordingly, among the
peroxyl radicals, only those with electron withdrawing groups are able to react with
carotenoids, via electron-transfer reaction. Thus, peroxyl radicals with electron donating
groups are expected to react via adduct formation. All of these findings nicely agree with the
available experimental results. The most energetically favored of the modeled reactions are
those involving trichloromethyl peroxyl, nitro, acyl peroxyl, and phenylacetyl peroxyl
The calculations of the Gibbs free energy of electron transfers from carotenoids to free
radicals are relatively expensive since frequency calculations of the corresponding Car, Car•+,
oxygenated free radical, and its anion are necessary in order to add thermodynamic
corrections to the electronic energies. Since they are rather time-consuming calculations, it
would be useful to find another magnitude that could be calculated at a lower computational
cost and that retains the correct order of relative reactivity. Ionization potentials have been
tested for that purpose. Figure 3 shows the correlations of the ΔG of reaction and the
carotenoids’ ionization potentials calculated using the different approaches discussed above
for their reaction with the acyl peroxyl radical. Even though, for the sake of simplicity, their
reactions with the other studied radicals are not shown, the same tendency was obtained in all
of the cases. Regardless of the approach chosen for computing the ionization potentials, good
correlations were found with the Gibbs energies of reaction. However, and as it was expected,
The only carotenoid that seems to have higher antioxidant efficiency than LYC, expressed
as electron donating capability, is TOR. This seems to be the case in both polar and nonpolar
media and in Car + free radical reactions as well as in redox reactions between carotenoid
pairs. This new finding suggests that this Car could be of great use for repairing other
oxidation damaged carotenoids and can be easily explained in terms of the length of the
polyene chain. TOR has 13 conjugated bonds, that is, the largest number among all of the
modeled structures, while LYC and BC have 11, but in the case of BC, two of them are in
cyclohexane rings, which are nonplanar with the rest of the molecular backbone, reducing the
effective length of the conjugated system. The larger reactivity of BC compared to that of
ZEA, however, cannot be explained by the length of the conjugated chain but by the presence
of two OH’s, which comparatively reduces the stability of the carotenoid radical cation. In the
case of ASTA and CAN, the presence of the two carbonyl groups at the end of the conjugated
CONCLUSIONS
for polar and nonpolar solvents. The same order was obtained for Car + oxygenated free
radical and Car2 + Car1 •+ reactions. Ionization potentials seem to be capable of predicting the
The nuclear reorganization energy associated with ET reactions has been calculated in a
very simple but apparently efficient way that allows computing of free energy barriers and, at
least, relative rate coefficients in good agreement with the experimental values. TOR is
predicted as the carotenoid with the highest antioxidant efficiency from the modeled set,
expressed as the electron donating capability, in polar and nonpolar media and in Car + free
radical reactions as well as in carotenoid pair redox reactions. The general agreement between
different calculated magnitudes and the corresponding available experimental data supports
the predictions from this work. Since TOR was found to be the most easily oxidized from all
carotenoid, relative to other carotenoids would be of great interest, especially relative to LYC,
which up to date has been thought of as the most efficient one. Such experimental studies
might also help to validate the accuracy of the methodology used in this work for describing
electron-transfer reactions. They might also support the utility of this kind of calculations in
the design of new antioxidants with increasing efficiency in terms of electron donating
capabilities, which might potentially help to prevent or repair oxidative damage in living
systems.
REFERENCES
1. van den Berg, H., Faulks, R., Fernando Granado, H., Hirschberg, J., Olmedilla, B.,
Sandmann, G., Southon, S. and Stahl, W., 2000, J. Sci. Food Agric., 80, 880.
2. Krinsky, N. I., 1993, Annu. Rev. Nutr., 13, 561.
3. Edge, R., McGarvey, D. J. and Truscott, T., 1997, J. Photochem. Photobiol., 41, 189.
4. Everett, S. A., Dennis, M. F., Patel, K. B., Maddix, S., Kundu, S. C. and Willson, R. L.,
1996, J. Biol. Chem., 271, 3988.
5. Woodall, A. A., Britton, G. and Jackson, M., 1997, J. Biochim. Biophys. Acta, 1336, 575.
6. Larsen, E., Abendroth, J., Partali, V., Schulz, B., Sliwka, H. R. and Quartey, E. G. K.,
1998, Chem. Eur. J., 4, 113.
7. Packer, J. E., Mahood, J. S., Mora-Arellano, V. O., Slater, T. F., Willson, R. L. and
Wolfenden, R. S., 1981, Biochem. Biophys. Res. Commun., 98, 901.
8. Hill, T. J., Land, E. J., McGarvey, D. J., Schalch, W., Tinkler, J. H. and Truscott, T. G.,
1995, J. Am. Chem. Soc., 117, 8322.
9. Everett, S. A., Kundu, S. C., Maddix, S. and Willson, R. L., 1995, Biochem. Soc. Trans.,
23, 230.
10. Mortensen, A., 2002, Free Radical Res., 36, 211
11. Jeevarajan, A. S., Khaled, M. and Kispert, L. D., 1994, J. Phys. Chem., 98, 1111.
12. Mortensen, A. and Skibsted, L. H., 1997, J. Agric. Food Chem., 45, 2970.
13. Edge, R., Land, E. J., McGarvey, D., Mulroy, L. and Truscott, T. G., 1998, J. Am. Chem.
Soc., 120, 4087.
14. El-Agamey, A. and McGarvey, D. J., 2003, J. Am. Chem. Soc., 125, 3330.
15 Sliwka, H. R., Melø, T-B., Foss, B. J., Abdel-Hafez, S. H., Partali, V., Nadolski, G.,
Jackson, H. and Lockwood, S. F., 2007, Chem.Eur. J., 13, 4458.
Figure 2. Relative order of carotenoids in terms of oxidation potentials in nonpolar and polar solvents; ■ ref 12,
□ ref 13
Figure 1. Correlation between vertical and Figure 3. Correlations between the Gibbs free
adiabatic ionization potentials with Koopman’s IP, energy of electrontransfer reactions (Car + R7) and
for polar and nonpolar media. carotenoids’ IP calculated using different
approaches.
Table 1. Carotenoids Previously Studied by Experimental Techniques and Modeled in the Present
Work
Symbol Formula Structure and Name
O
OH
ASTA C40H52O4
HO
Astaxanthin
CHO
APO C30H40O
β-apo-8'-carotenal
O
CAN C40H52O2
Canthaxanthin
OH
LUT C40H56O2
HO
Lutein
OH
ZEA C40H52O2
HO
Zeaxanthin
BC C40H56
All-trans-β-carotene
LYC C40H56
Lycopene
Table 2. Carotenoids Modeled in the Present Work and with No Previous Data on Their Electron-
Transfer Reactions
Symbol Formula Structure and Name
DIH C40H58
7,7' Dihydro-β-carotene
O
Table 2.
ECH C40H54O
Echinenone
OH
HO
NOS C40H56O4 OH
HO
Nostoxanthin
OCH3
OKE C41H54O2
O
Okenone
SAP C40H56O2 OH
HO
Saproxanthin
TOR C40H54
Torulene
OH
CSAN C40H56O3
O
HO
Capsanthin
OH
CRUB C40H56O4 O
OH
Capsorubin
(Continued)
Symbol Formula Structure and Name
OH
CRY C40H56O2
O
Cryptocapsin
HO
NEO C40H56O4 O
HO OH
Neoxanthin
O
VIO C40H56O4 O
OH
HO
Violaxanthin
OH O
MIT C40H52O4
HO O
Mytiloxanthinone
Table 5. Gibbs Free Energies (kcal/mol)a of Carotenoidsa Electron-Transfer Reactions with Different Radicals in
a Polar Environment (Solvent = Water)
R1 R2 R3 R4 R5 R6 R7 R8 R9
ASTA 9.26 1.55 -10.84 4.11 2.71 -14.46 -11.03 -20.69 -14.50
APO 8.73 1.02 -11.37 3.58 2.18 -14.99 -11.56 -21.22 -15.03
CAN 9.02 1.31 -11.08 3.86 2.46 -14.71 -11.27 -20.93 -14.75
LUT 5.78 -1.92 -14.32 0.63 -0.77 -17.94 -14.51 -24.17 -17.98
ZEA 3.96 -3.75 -16.14 -1.20 -2.59 -19.77 -16.33 -25.99 -19.81
BC 4.48 -3.23 -15.62 -0.67 -2.07 -19.24 -15.81 -25.47 -19.28
LYC 2.07 -5.64 -18.03 -3.08 -4.48 -21.65 -18.22 -27.88 -21.69
DIH 5.59 -2.11 -14.51 0.44 -0.96 -18.13 -14.70 -24.36 -18.17
ECH 5.81 -1.90 -14.29 0.66 -0.74 -17.91 -14.48 -24.14 -17.95
NOS 5.69 -2.01 -14.40 0.54 -0.86 -18.03 -14.60 -24.25 -18.07
OKE 6.31 -1.40 -13.79 1.16 -0.24 -17.41 -13.98 -23.64 -17.45
SAP 3.96 -3.75 -16.14 -1.20 -2.59 -19.77 -16.33 -25.99 -19.81
TOR 1.52 -6.19 -18.58 -3.64 -5.03 -22.21 -18.77 -28.43 -22.25
CSAN 8.84 1.13 -11.26 3.69 2.29 -14.88 -11.45 -21.11 -14.92
CRUB 14.03 6.32 -6.07 8.87 7.47 -9.70 -6.26 -15.92 -9.74
CRY 8.42 0.71 -11.68 3.26 1.87 -15.30 -11.87 -21.53 -15.34
NEO 4.98 -2.73 -15.12 -0.18 -1.58 -18.75 -15.32 -24.97 -18.79
VIO 6.38 -1.33 -13.72 1.23 -0.17 -17.34 -13.91 -23.57 -17.38
MYT 9.92 2.21 -10.18 4.76 3.37 13.80 --10.37 -20.03 -13.84
a
Energies: SP/B3LYP/6-311++G(d,p); geometries: B3LYP/6-31G(d,p); thermodynamic corrections: B3LYP/3-21G.