Relationship between hydrophobicity and antioxidant ability of "phenolipids" in emulsion: A parabolic effect of the chain length of rosmarinate esters

Article abstract of DOI:10.1021/jf904119v

The polar paradox predicts that hydrophobic antioxidants are more active in emulsions than their hydrophilic homologues, thus assuming a linear dependency between hydrophobicity and antioxidant capacity. In contrast, we formulate in this paper an alternative hypothesis assuming a possible nonlinear dependency. To verify this so-called "nonlinear hypothesis", the antioxidant capacity of a homologous series of rosmarinic acid and its alkyl esters (methyl, butyl, octyl, dodecyl, hexadecyl, octadecyl, and eicosyl) was evaluated using a newly developed conjugated autoxidizable triene (CAT) assay. It appeared that the antioxidant capacity increases as the alkyl chain is lengthened, with a maximum for the octyl chain, after which further chain extension leads to a collapse in antioxidant capacity. This nonlinear effect was discussed in relation to the "cutoff effect" generally observed in studies using cultured cells. This new hypothesis may provide a better understanding of the antioxidant behavior of phenolics in emulsion which is a key to develop new antioxidant strategies to protect lipid substrates from oxidation. Moreover, the lipophilization with medium chain appeared as a promising way to enhance the antioxidant capacity of phenolics since octyl rosmarinate was three times more effective than rosmarinic acid which is already one of the most powerful known phenolic antioxidant. Finally, this work paves the way for systematic investigation of the chain length effect to design new "phenolipids" in a rational fashion.

Full text of DOI:10.1021/jf904119v

J. Agric. Food Chem. 2010, 58, 2869–2876 2869
DOI:10.1021/jf904119v
Relationship between Hydrophobicity and Antioxidant Ability
of “Phenolipids” in Emulsion: A Parabolic Effect of the Chain
Length of Rosmarinate Esters
†
.
ꢀPEZ
M
M
ICKAEL
L
AGUERRE
,
†,‡ LUIS J. Lo
G
IRALDO,† JEꢀROME
L
ECOMTE
,
ꢀ
ARIA-CRUZ
F
IGUEROA-ESPINOZA
,
†,§ BRUNO
B
AREA,† JOCHEN
,†
W
EISS
,
‡
E
RIC A. DECKER
,
AND
P
IERRE
V
ILLENEUVE
*
†CIRAD, UMR IATE, Montpellier, F-34398, France, ‡Department of Food Science,
University of Massachusetts, Chenoweth Laboratory, 100 Holdsworth Way, Amherst,
Massachusetts 01003, §IRC Montpellier SupAgro 1101, av. Agropolis BP 5098 34093,
Montpellier Cedex 05, France, and Department of Food Science and Biotechnology,
University of Hohenheim, 70599 Stuttgart, Germany
The polar paradox predicts that hydrophobic antioxidants are more active in emulsions than their
hydrophilic homologues, thus assuming a linear dependency between hydrophobicity and antioxi-
dant capacity. In contrast, we formulate in this paper an alternative hypothesis assuming a possible
nonlinear dependency. To verify this so-called “nonlinear hypothesis”, the antioxidant capacity of a
homologous series of rosmarinic acid and its alkyl esters (methyl, butyl, octyl, dodecyl, hexadecyl,
octadecyl, and eicosyl) was evaluated using a newly developed conjugated autoxidizable triene
(CAT) assay. It appeared that the antioxidant capacity increases as the alkyl chain is lengthened,
with a maximum for the octyl chain, after which further chain extension leads to a collapse in
antioxidant capacity. This nonlinear effect was discussed in relation to the “cutoff effect” generally
observed in studies using cultured cells. This new hypothesis may provide a better understanding of
the antioxidant behavior of phenolics in emulsion which is a key to develop new antioxidant
strategies to protect lipid substrates from oxidation. Moreover, the lipophilization with medium chain
appeared as a promising way to enhance the antioxidant capacity of phenolics since octyl
rosmarinate was three times more effective than rosmarinic acid which is already one of the most
powerful known phenolic antioxidant. Finally, this work paves the way for systematic investigation of
the chain length effect to design new “phenolipids” in a rational fashion.
KEYWORDS: Antioxidant; phenolic compound; phenolipid; rosmarinic acid; lipophilization; lipid
oxidation; emulsion; conjugated autoxidizable triene assay; partition; polar paradox; cutoff effect;
nonlinear hypothesis
INTRODUCTION
in bulk oil and dispersed lipid systems such as micellar media,
emulsions, liposomes, and even whole tissues. Regarding oil-in-
water emulsion, the polar paradox proposed that hydrophobic
antioxidants are more active than their hydrophilic homologues
(ceteris paribus), thus implicitly assuming a linear dependency
between the antioxidant capacity and the hydrophobicity in such
a multiphasic system. Then, Frankel et al. (4) strengthened this
model introducing the interfacial oxidation concept according to
which the higher effectiveness of hydrophobic antioxidants in oil-
in-water emulsions would be due to their tendency to concentrate
at the interfacial membrane where the oxidation is supposed to
occur, while more hydrophilic antioxidants would tend to parti-
tion more into the aqueous phase where they would be ineffective.
Based upon this polar paradox theory, some authors suggested to
design new lipophilized surface active antioxidant molecules,
especially phenolics, to improve their ability to counteract lipid
oxidation in emulsions and other dispersed lipid systems (5).
These functionalized molecules resulting from the grafting of
Lipid oxidation in emulsified systems plays an important role
for cosmetic and food products, since many of them exist either
partlyorwhollyasemulsion, orhave beenin anemulsified state at
some time during their existence (1). However, the reaction
mechanisms of lipid oxidation, though extensively investigated
since the beginning of the 20th century, are still not completely
understood, especially in connection with antioxidants in dis-
persed systems such as emulsions. Such knowledge is nevertheless
highly desirable to develop new antioxidant strategies to protect
nutritional lipids from oxidative alterations. To the best of our
knowledge, one of the first rationale models to explain antioxi-
dant effect in lipids was suggested by Porter (2), and then
formulated in the polar paradox hypothesis (3). This hypothesis
described paradoxes in the effectiveness of antioxidant substances
*Corresponding author. E-mail: villeneuve@cirad.fr. Phone: (33)
(0)4 67 61 55 18. Fax: (33) (0)4 67 61 55 15.
© 2010 American Chemical Society
Published on Web 02/04/2010
pubs.acs.org/JAFC

2870 J. Agric. Food Chem., Vol. 58, No. 5, 2010
Laguerre et al.
flasks each containing 5 mL of alcohol (methanol, 123.44 mmol;
n-butanol, 54.64 mmol; n-octanol, 31.905 mmol; n-dodecanol, 22.46
mmol; n-hexadecanol, 16.95 mmol; n-octadecanol, 15.09 mmol and
n-eicosanol, 13.6 mmol). The reaction mixtures were stirred (orbital
shaker, 250 rpm, 55-70 °C) prior to the addition of the catalyst, the
strongly acidic sulfonic resin Amberlite IR-120H (5% w/w - total weight of
both substrates) previously dried at 110 °C for 48 h. The water generated
during the reaction was removed by absorption on molecular sieves
(40 mg/mL) added to the medium. Samples (20 μL) were regularly
withdrawn from the reaction medium then mixed with 980 μL of
methanol, filtered (0.45 μm syringe filter Millex-FH, Millipore Corpora-
tion Bedfork, MA), and finally analyzed by reverse phase HPLC with UV
detection at 328 nm. After complete (4-21 days) conversion of rosmarinic
acid into the corresponding ester, the latter was purified in a two step
procedure. First, a liquid-liquid extraction using hexane and acetonitrile
was achieved to remove the excess of alcohol. Then, the remaining traces of
the alcohol and rosmarinic acid were eliminated by flash chromatography
on a CombiFlash Companion system (Teledyne Isco Inc., Lincoln, NE).
Separation was achieved on a silica column using an elution gradient of
hexane and ether (20% to 100% in 35 min). The yield of purified esters,
obtained as pale yellow to yellow amorphous powders, was calculated
from calibration curves previously established with pure compounds. Pure
Figure 1. Chemical structure of rosmarinic acid and its alkyl esters.
a lipid on a phenolic moiety (coined as “phenolipids”) can be
prepared via different synthesis strategies such as esterifica-
tion (6-13), amidation (14) or etherification (15, 16). However,
it has been demonstrated in oil-in-water emulsion that lipophili-
zation of phenolics is not always advantageous in terms of
antioxidant capacity (9, 17, 18), thus suggesting that not all
antioxidant behave in a manner proposed by the polar paradox
theory. These contrasting results highlighted the fact that the
behavior of phenolic antioxidants is governed by more complex
phenomena than expected, and lead us to postulate that the
dependency between the antioxidant capacity and the hydropho-
bicity does not necessarily follow a linear trend (19). Indeed, in a
previous study performed in oil-in-water emulsion and using a
complete homologous series of chlorogenic acid and its alkyl
esters (methyl, butyl, octyl, dodecyl, hexadecyl, octadecyl and
eicosyl) we observed that antioxidant capacity increases as
the alkyl chain is lengthened, with a threshold for the dodecyl
chain, beyond which further chain extension leads to an antioxi-
dant capacity collapse (19). This result suggested that the chain
length affects antioxidant capacity in a nonlinear manner, at least
in our system, providing a new model to understand the behavior
of phenolic antioxidants in emulsions and to design promising
new phenolipids.
1
esters and rosmarinic acid were then fully characterized by ESI-MS, H
NMR and 13C NMR as previously described by Lecomte et al. (13).
Preparation of Tocopherol-Free Tung Oil Samples. The polar
compounds of tung oil (including β-tocopherol) were removed by passing
25 mL of a 200 mg/mL tung oil solution in hexane, followed by 25 mL of
pure hexane through an alumina column prepared as follows: 25 g of
alumina in hexane was introduced into a glass column, and the excess
hexane was eliminated until it rose to the alumina surface. After complete
removal of tocopherols, hexane was evaporated under vacuum at 35 °C
using a rotatory evaporator equipped with a vacuum pump (Laborport,
KNF Neuberger GmbH, Freiburg, Germany). It is worth noting that all
experiments must be carried out under shelter from light, as much as
possible. Finally, the stripped tung oil was aliquoted into sealed brown
glass tubes, then flushed with nitrogen, and stored at -18 °C until use.
The use of disposable stripped tung oil aliquots (i.e., 15 aliquots for
15 microplates) avoids the necessity for successive withdrawals from one
aliquot and minimizes any eventual oxidation.
Conjugated Autoxidizable Triene (CAT) Assay Protocol. Anti-
oxidant capacity of rosmarinic acid and its alkyl esters was measured using
the CAT procedure developed by Laguerre et al. (20) with slight improve-
ments (19, 21) allowing the analysis of a larger panel of molecules from
hydrophilic to lipophilic ones. Briefly, phenolics solutions were prepared
as follows: a methanol solutionof rosmarinic acid, its alkyl esters orTrolox
(reference) was prepared at the desired concentration. Then, various
volumes of this solution (25, 50, 75, and 100 μL) were added to 24.9 mL
of phosphate buffer solution (PBS) pH 7.2, and then completed to 25.0 mL
with pure methanol (75, 50, 25, and 0 μL, respectively). In this way, all
buffered solutions of phenolics contain the same methanol volume
(100 μL), which allows any eventual bias among samples to be avoided.
All buffered solutions of phenolics were prepared extemporarily and
50 μL/well of these solutions were transferred using a multichannel
micropipet into a UV-Star 96-well microplate (Greiner, Frickenhausen,
Germany), a well adapted plastic-based microplate for spectral measure-
ment in the UV-domain (absorbance at 273 nm = 0.03). The microplate
was then prewarmed and stirred in a thermostatted shaker (PHMT Grant
Instruments Ltd., Shepreth, England) for 5 min at 37 °C and 1200 rpm.
Twenty-five milliliters of PBS (pH 7.2) containing 34 μM Brij 35
(neutral emulsifier, estimated MW = 1198 g/mol) was added to 5 mg of
stripped tung oil in a brown glass flask. Afterward, it is crucial to premix
this mixture by stirring it for 10 s using a Vortex apparatus, before its
homogenization in an Ultra Turrax homogenizer (Janke & Kunkel,
Staufen, Germany) at approximately 2400 rpm for 90 s. This tung oil-
in-PBS emulsion (100 μL) was then added to well. To improve repeat-
ability, the microplate was thenimmediately prewarmed and shaken, inthe
absence of light, in a thermostatted shaker (PHMT Grant Instruments
Ltd.) at 37 °C for 1 min at 1200 rpm.
The goal of the present study was to strengthen our under-
standing of the relationship between antioxidant capacity and
hydrophobicity using a new homologous series consisting of
rosmarinic acid and its alkyl esters (methyl, butyl, octyl, dodecyl,
hexadecyl, octadecyl and eicosyl, Figure 1) (13). Such a study
performed with both hydrophilic and lipophilic antioxidants was
possible using a newly developed conjugated autoxidizable triene
(CAT) assay (20) performed in tung oil-in-water emulsion with
slight improvements (19, 21). The partitioning behavior of
phenolics into this dispersed system using various emulsifier
1
concentrations was also investigated, along with the H NMR
data of those compounds.
MATERIALS AND METHODS
Chemical. Tung oil (averaged MW = 872 g/mol) was purchased from
Aldrich (ref 440337). Sunflower oil was purchased in a local supermarket.
Rosmarinic acid, phosphate buffer solution (PBS) pH 7.2, polyoxy-
ethylene(23)lauryl ether (Brij 35, estimated MW = 1198 g/mol) were
purchased from Sigma (Saint Quentin, France). All solvents used were
HPLC or analytical grade and were also purchased from Sigma. Trolox
was from Acros Organic (Geel, Belgium), and finally 2,20-azobis(2-
amidinopropane) dihydrochloride (AAPH) from Wako Chemical
(Neuss, Germany).
Synthesisof RosmarinateEsters. The chemoenzymatic esterification
of rosmarinic acid to obtain rosmarinic esters was carried out following the
procedure described by Lecomte et al. (13). Briefly, the chemical esteri-
fication of rosmarinic acid (56 μmol) was carried out in sealed brown
Fifty microliters of a freshly prepared AAPH solution in PBS (4 mM)
was then added to each well with a multichannel micropipet. Finally, each
well contained 200 μL of the final mixture consisting of 115 μM stripped

Article
J. Agric. Food Chem., Vol. 58, No. 5, 2010 2871
tung oil, 17 μM Brij 35, 1 mM AAPH, and various concentrations of
phenolics (from 0.01 to 0.55 μM) in PBS. The progress of reactions was
immediately monitored by recording the decrease in absorbance at
273 nm. The possibility to use direct spectrophotometric measurements
suggests that our emulsion is likely to be a microemulsion whose droplet
diameter is small enough to avoid light scattering. Afterward, measure-
ments were performed each minute for 5 h at 37 ( 0.1 °C, with 5 s stirring
€
before each measure, using a Saffire 2 microplate reader (Tecan, Grodig,
Austria) equipped with Magellan software. Three independent analyses
using different microplates were carried out.
Results Expression of the Conjugated Autoxidizable Triene
Assay. To normalize data, the raw absorbance signal was transformed
in relative absorbance according to the equation
relative absorbance ¼ At=A0
ð1Þ
where At and A0 are absorbances read at times t and 0 min, respectively. It
is worth mentioning that if the measurement is not rapid enough after
AAPH addition, the A0 for the blank (without antioxidant) may be lower
than that of the sample containing antioxidant. In this case, to normalize
A0, the experimental A0 for the blank can be replaced with the A0 of
samples in eq 1.
The area under the curve (AUC) corresponding to relative absorbance
decay was then calculated as
Figure 2. Kinetic of relative absorbance bleaching of stripped tung oil-in-
water emulsion in the absence or presence of 0.1 μM of phenolics. The
reaction mixture contained 115 μM stripped tung oil, 17 μM Brij 35, and 1
mM AAPH, in PBS pH 7.2 at 37 °C. R0 (0), R1 (2), R4 ((), R8 (9), R12
(O), R16 (]), R18 (4), R20 (*), and phenolic-free control (b).
AUC ¼ 1 þ A1=A0 þ A2=A0 þ : : : þ A299=A0 þ A300=A0
ð2Þ
The net protection area provided by an antioxidant sample was then
calculated using the difference between the AUC in the presence of an
antioxidant sample (AUCsample) and the AUC of the blank (AUCblank),
the latter consisting of the same mixture without antioxidant.
Trolox was used as a reference for antioxidant capacity measurements.
Thus, the antioxidant capacity of a sample relative to Trolox (CAT value)
is given as
paradox) assumes that hydrophobic antioxidants are more active
in oil-in-water emulsion than their hydrophilic homologues,
which can be implicitly interpretated as the fact that the relation-
ship follows a linear shape. In contrast, we formulate in this paper
an alternative hypothesis assuming that the relationship between
antioxidant capacity and hydrophobicity is not necessarily linear,
which can lead to a better rationale to design new phenolipids.
Antioxidant Capacity Measurement Using the CAT Method.
Taking advantage of the possibility to assess both hydrophilic
and hydrophobic compounds in the CAT assay, the antioxidant
capacity of a homologous series of rosmarinic acid and its alkyl
esters (Figure 1) was evaluated with the protocol developed by
Laguerre et al. (20) with slight improvements (19, 21). Except for
octadecyl and eicosyl esters, it appeared that all tested phenolics
exhibited noticeable antioxidant properties for all tested concen-
trations (Figure 2, data only shown for 0.1 μM). The calculation
of their net AUC from eq3 showed that all phenolics act ina dose-
dependent manner (Figure 3), which is a prerequisite to calculate
their CAT value expressed as moles of Trolox equivalent per mole
of test compound through eq 3. Strong differences were observed
between phenolics, with a maximal antioxidant capacity for the
octyl ester (15.31 ( 1.33 TE), followed by butyl rosmarinate
(10.33 ( 1.39 TE) > methyl rosmarinate (8.37 ( 0.60 TE) >
dodecyl rosmarinate (7.67 ( 0.76 TE) > rosmarinic acid (5.21 (
0.25 TE) . hexadecyl rosmarinate (1.53 ( 0.26 TE) > octadecyl
rosmarinate (0.65 ( 0.11 TE) ∼ eicosyl rosmarinate (0.54 (
0.13 TE) (Figure 4). First of all, the lipophilization appeared
to induce a drastic modulation of the antioxidant capacity.
Rosmarinic acid is known as one of the most powerful phenolic
antioxidants since it encompass two catechol (o-diphenol) moie-
ties which is an important feature in terms of bond dissociation
energy of the phenolic hydroxyl group (O-H BDE) (22). In this
context, the grafting of an octyl chain leads to a phenolipid three
times more effective than rosmarinic acid, demonstrating the
interest to modulate the hydrophobicity of phenolics through
lipophilization reaction. Moreover, we recently demonstrated
that the lipophilization of rosmarinic acid does not necessarily
lead to an increase of its free radical scavenging stoichiometry
CAT value ¼ ½ðAUCsample -AUCblankÞ=ðAUCTrolox -AUCblankÞꢀ
ꢁ½mol of Trolox=mol of sampleꢀ
ð3Þ
where (AUCsample - AUCblank) and (AUCTrolox - AUCblank) are the net
protection areas in the presence of a sample and Trolox, respectively. It
must be mentioned that Trolox was used as internal standard (i.e.,
analyzed on the same microplate). Finally, the CAT value is expressed
as moles of Trolox per mole of tested compound (Trolox equivalents).
Antioxidant Partitioning Measurements. The sunflower oil-in-
water emulsions (2 wt %/wt) were prepared using Brij 35 as emulsifier.
The final concentrations of emulsifier were 0, 0.017, 0.17, 1, 2, and 10 mM.
The procedure was the following: 25 mL of PBS (pH 7.2) containing the
previous concentration of Brij 35 were added to 0.5 g of sunflower oil in a
brown glass flask. Afterward, the solution was vortexed for 10 s before its
homogenization using an Ultra Turrax homogenizer (Janke & Kunkel,
Staufen, Germany) at approximately 2400 rpm for 90 s. To minimize lipid
and phenolic oxidation, 100 μM EDTA was added to emulsion sample. All
phenolics were dissolved in methanol and added to the emulsion at a final
concentration of 25-100 μM. Afterward, the previous solutions were kept
for 24 h in the dark at room temperature. To measure concentration of
phenolics in the continuous phase of the emulsion, 1 mL of emulsion and
0.5 mL of PBS were mixed in a centrifuge tube and centrifuged at 10,000g
at 4 °C for 40 min. After centrifugation, 1 mL of the continuous aqueous
phase (lower layer) of the emulsion was withdrawn with a syringe. This
procedure of PBS addition, centrifugation and collection ofthe continuous
phase of the emulsion was repeated three times. After the final centrifuga-
tion, the continuous phase was filtered through a 0.20 μm syringe filter to
remove any residual emulsion droplets. The concentration of rosmarinic
acid and its alkyl esters was finally evaluated using HPLC with a detection
at 328 nm as previously described by Lecomte et al. (13).
RESULTS AND DISCUSSION
Today, the development of new antioxidants such as pheno-
lipids is hampered by the lack of a reliable theory describing and
predicting the antioxidant properties of phenolic antioxidants in
emulsified systems. The only rational model (known as polar

2872 J. Agric. Food Chem., Vol. 58, No. 5, 2010
Laguerre et al.
(apart from our previous article (19)) may explain why some
lipophilization strategies lead sometimes to an unexpected
decrease of the antioxidant effect, especially those using long
aliphatic chains. In other words, these results along with those we
previously reported with chlorogenate esters (19) show that it is
not relevant to simply graft any fatty moiety and highlight the
prime importance of searching the optimal chain effect to design
the fittest phenolipids.
Interestingly, the nonlinear effect observed with the CAT assay
is consistent with current literature regarding biological activity,
with numerous studies documenting this type of response with
homologous series of amphiphilic molecules as antimicrobial
(23, 24), anesthetic (25, 26), cytotoxic (27), or spermicide (28)
agents. This particular phenomenon was called “cutoff effect”
and can be defined for a homologous series of hydrocarbon chain
surface active compounds, as the tendency for various biological
activitiestoincreasestepwise(or staystable) with increasingchain
length up to a critical point, beyond which these activities
suddenly collapse. It is worth pointing out that the first part of
the trend (below the threshold) does not always follow the same
trend from a study to another one, while after such a threshold is
reached, the trend does always follow a sudden collapse. This
definition of the cutoff effect perfectly fits with the parabolic
shape depicted in Figure 4. In terms of mechanism of action,
different hypotheses have been put forward, most of them dealing
with a specific interaction between drug(s) and the cell membrane.
However, our cutoff effect was observed in a nonliving system (i.
e., oil-in-water emulsion) which suggests that this phenomenon
may originate froma simple physicochemical process. In our case,
among numerous other putative mechanisms of action, we
investigated the hypothesis that this effect may come either from
(i) an intrinsic modulation of the H-donation ability by the chain
length through a π-stacking process, or from (ii) an interaction
between phenolics, water and lipids, involving possibly a micelli-
zation process as already assumed in a similar case (19).
Figure 3. Net protection AUC (AUCsample - AUCblank) vs concentration
(μM) for rosmarinic acid, its alkyl esters, and Trolox (reference). Con-
centration of phenolics has been checked by HPLC and corrected.
H-Donation Ability, π-Stacking and 1H NMR. In terms of
H-donation ability, the grafting of an alkyl chain on rosmarinic
acid should not theoretically affect the H-donation ability via a
mesomeric effect since the carboxylic and the two phenolic groups
are not conjugated. However, a self-association may occur
between the two aromatic rings via a π-stacking caused by
intramolecular overlapping of the p-orbital system which is
supposed to lead to an increase of the free radical scavenging
ability. It is well-known that a π-π interaction of two phenolic
rings can take place between two phenolic molecules (29). Such a
self-assembly process can also take place between two aromatic
rings present in the same molecule. To the best of our knowledge,
this latter type of π-stacking (so-called intramolecular π-stacking)
is only documented for antocyanins acylated by phenolic
acids (30). Of particular interest is that rosmarinic acid and its
alkyl esters satisfy basic requirements to undergo a folding
through intramolecular π-stacking. Indeed, the two phenolic rings
(A and B) are linked by a suitable spacer, which allows the
molecule to fold in such a way that the two rings can interact
together. Others requirements involve the protection from
hydration, and the planarity of aromatic rings to optimize the
contact area between themselves. In an attempt to explain the
difference observed for antioxidant capacity between rosmarinic
acid and its alkyl esters, it can be hypothesized that, in the case of
the rosmarinic acid, the π-stacking configuration (called config-
uration 1) may be in competition with another configuration.
This latter (called configuration 2) would involve the establish-
ment of a H-bond from the hydroxyl of the free COOH function
and the oxygen of one of the two phenolic hydroxyl groups in the
B ring. Let us consider the putative case where the better
Figure 4. CAT value (moles of Trolox per mole of tested compound) of
rosmarinic acid and its alkyl esters vs the alkyl chain length.
against the 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical (13).
This shows again that the use of two different methodologies can
lead to opposite conclusions regarding the interest of lipophili-
zation. One can also notice that noncompetitive free radical
scavenging capacity methods (such as DPPH test) do not contain
oxidizable substrate (such as lipid), and consequently do not
provide a competitive reaction scheme, in which antioxidant and
substrate compete for oxidizing species. Thus, unlike the CAT
assay, the DPPH methodology does not accurately indicate the
ability of a compound to inhibit oxidation.
Of particular interest was the fact that the dependence between
the alkyl chain length and the antioxidant capacity followed a
parabolic shape with a maximum for the octyl ester (Figure 4).
This effect can be divided into two virtual domains: a first part (i)
with a quasi-linear increase of the antioxidant capacity with the
elongation of the alkyl chain, and a second part (ii) for which the
antioxidant capacity collapses with increasing hydrophobicity.
The fact that such a nonlinear dependence was never reported

Article
J. Agric. Food Chem., Vol. 58, No. 5, 2010 2873
Figure 5. Effectofthealkylchainlength onthepartition behaviorofphenolicsina mixture of sunflower oil and PBS pH 7.2 without or with0.017, 0.17, 1, 2, and
10 mM Brij 35 used as emulsifier.
antioxidant capacity for methyl, butyl, octyl, and dodecyl ros-
marinates compared to free rosmarinic acid is due to the fact that
the configuration 1 is the main configuration for the esters,
whereas equilibrium between configuration 1 and 2 takes place
for rosmarinic acid. In this case, configuration 1 should confer
better antioxidant capacity than configuration 2 (all other things
being equal), since the phenolic hydroxyls are more susceptible to
establish H-bond with another phenolic hydroxyl group (and
consequently to decrease O-H BDE of the molecule by stabiliz-
ing the phenoxy radical), than with the carboxylic hydroxyl
group. Such a difference in the intramolecular configuration
between rosmarinic acid and its alkyl esters should lead to a
significant displacement of the chemical shifts of their aromatic
used in the CAT assay), the longer the chain length, the lower
their level in the water phase is, until octyl chain which totally
partitions in the discontinuous phase (oil þ interfacial mem-
brane). Then, further increments in the chain length do not
significantly change the partitioning behavior. This result was
expected and is consistent with literature reports about partition-
ing behavior of phenolics in oil-in-water emulsion (18, 19).
However, a little increase in the concentration of Brij 35 from
17 to 170 μM lead to a drastic change, with the appearance of a
weak cutoff effect with a threshold for the dodecyl ester, above
which, surprisingly, the longer the chain length, the higher their
level in the aqueous phase is. Interestingly, we found the same
result with chlorogenate esters in the same system apart from
the concentration of Brij 35 which was 17 μM. It has not escaped
our notice that the specific cutoff effect in the partitioning
behavior with a threshold for the dodecyl chain (in both rosmari-
nate and chlorogenate cases) may be due the fact that the
hydrophobic tail of Brij 35 emulsifier consists of a dodecyl
chain. In virtue of their amphiphilic properties, the emulsifier
accumulates at the oil-water interface in emulsions and thus is
considered to dominate the properties of the interface in terms
of solubilization capacity of antioxidants (17). Indeed, this
structural analogy could enable a good alignment between
dodecyl rosmarinate and Brij 35 hydrophobic tails. However,
regardless of the mechanism of action, the cutoff effect for a
dodecyl chain in the partitioning behavior does not explain the
cutoff effect for an octyl chain in the antioxidant properties of
rosmarinate esters. This must be connected with the fact that
some authors unexpectedly demonstrated that the partitioning
behavior of antioxidants is not always a good predictor of the
antioxidant effectiveness (18).
1
protons as indicated by H NMR analysis (31). However, the
investigation of our 1H NMR spectra of rosmarinic acid and its
alkyl esters (13) did not show such a displacement. Indeed, the
chemical shift for rosmarinic acid was 7.08, 6.80-6.78,
7.04-7.02, 6.70-6.69, 6.67-6.65, and 6.56-6.54 δ/ppm for the
aromatic protons in position 2, 5, 6, 20, 50, and 60, respectively,
whereas their chemical shifts for rosmarinate esters were 7.07,
6.79-6.77, 7.02-7.00, 6.68-6.67, 6.66-6.64, and 6.52-6.50
δ/ppm, respectively. Finally, although the occurrence of a
π-stacking between the two aromatic rings seems to be likely
for these phenolics, no difference in such a process between
rosmarinic acid and its alkyl esters can be deduced from 1H
NMR data.
Partitioning Behavior of Phenolics at Low Emulsifier Concen-
tration. Besides intrinsic H-donation ability, the antioxidant
capacity of phenolics in heterogeneous systems such as emulsions
is supposed to be also governed by their interaction with the
environment, especially their partitioning behavior between the
different phases (4) and their diffusibility toward reactive centers
(i.e., oxidizable substrate, free radicals, . . .) (22). In order to gain
insight about a possible mechanism of action, the partitioning of
rosmarinic acid and its alkyl esters was studied in a mixture of
sunflower oil and PBS (pH 7.2). It appeared in Figure 5 that, with
or without 17 μM Brij 35 (same Brij 35 concentration as in the
CAT assay, but different from the Brij 35/phenolic molar ratio
Besides partitioning, it is worth noting that increased hydro-
phobic interactions with the environment may lower the diffusion
of long chain rosmarinates to the reaction centers, resulting in a
reduced antioxidant capacity, as previously suggested by Stock-
mann et al. (17) on a homologous series of gallate esters. In
connection, according to Fendler (32), the mobility of the hydro-
carbon chain is reduced with increasing its length. One can also

2874 J. Agric. Food Chem., Vol. 58, No. 5, 2010
Laguerre et al.
and Laguerre et al. (19) showed the same partition behavior. For
instance, the Brij 700 emulsifier was able to increase the partition-
ing of propyl gallate into the aqueous phase of an oil-in-water
emulsion in a rapid and concentration-dependent manner (37).
Other related study (38) showed that lipid hydroperoxides can be
carried from the discontinuous phase to the aqueous one by Brij
76 micelles in a same way as phenolic antioxidants. Taken
together, these studies (including the present one) suggest that
above CMC, an emulsifier can drastically affect the partition of
crucial components in the lipid oxidation pathways by either
bringing them together so they can react or isolating them and
thus preventing their interaction. Accordingly, the micellization
of surfactant may partly shift the oxidation process from a pure
interfacial phenomenon to a micellar one. We feelthat this may be
an important point, since many oil-in-water emulsions contain
more surfactant than is needed to completely saturate the emul-
sion droplet surface (37).
Finally, the polar paradox is based on the observations that
hydrophobic antioxidants are more active in emulsion than
hydrophilic ones, while these latter are more efficient in bulk oil
than their hydrophobic homologues. Historically, this hypoth-
esis was put forward to show that it is not relevant to extra-
polate uncritically effectiveness data for antioxidants from bulk
oil to emulsion (or micelles, membrane and whole tissue). To be
clear, the present results along with those previously reported
for chlorogenate esters (19) do not put into question this part of
the polar paradox which was, in the field of lipid oxidation, one
of the greatest achievements of recent decades. Our results
rather show that the dependence in emulsion between the
hydrophobicity and the antioxidant capacity is nonlinear and
characterized by a strong cutoff effect for the octyl ester, which
cannot be anticipated nor explained on the basis of the polar
paradox theory. This finding may be crucial for two reasons.
The first one is that the cutoff (or nonlinear) hypothesis (if
confirmed with other methods, other oxidation inductors, and
other phenolipids) can lead to a much better fundamental
understanding of the behavior of phenolic antioxidants in
dispersed systems which is a prerequisite to master the current
antioxidant strategies. Indeed, at least in the CAT system, our
study suggests that the polar paradox theory is not valid for
medium and long chain lengths. The second important impact
of our nonlinear hypothesis may be the establishment of a
rational rule to design and customize new promising antiox-
idants, such as phenolipids, to optimize the protection of
nutritional lipids in multiphasic systems. We recently noted
with interest the publication by Medina et al. (33) of a supporting
evidence for our nonlinear hypothesis, since they found in oil-in-
water emulsion that “the presence of a short-medium lipophilic
chain (acetate, butyrate or octanoate) improved the antioxidant
efficiency of hydroxytyrosol [. . .], but longer alkyl chain (laurate)
maintained or even decreased their antioxidant activity”. In
addition, we recently observed with chlorogenate esters in human
fibroblasts (to be published elsewhere) the same antioxidant
cutoff effect with a threshold for the dodecyl chain as already
reported in oil-in-water emulsion (19), suggesting that this beha-
vior may have important biological consequences. In terms of
mechanism of action, neither 1H NMR data nor partitioning
studies elucidated the origin of this effect, which remains as
unclear for us as for other studies where such a behavior has been
observed. Further experiments varying both antioxidant models
and methodologies, especially the way to induce oxidation (natural
one, transient metal, azo compounds . . .), must be achieved to
confirm or infirm this nonlinear hypothesis, especially the collapse
beyond a threshold chain length, and finally to elucidate its
mechanism of action.
Figure 6. Emulsifier (Brij 35) concentration effect on the partition behavior
of phenolics in a mixture of sunflower oil and PBS pH 7.2. R0 ([), R1 (4),
R4 (b), R8 (0), R12 (]), R16 (O), R18 (9), R20 (2).
notice that steric hindrance induced by increasing the chain length
can render difficult the contact between long chain esters and
AAPH-derived free radicals, and may consequently be involved
in the cutoff effect observed in this study. In addition, the
partitioning behavior of phenolics in the presence of 17 μM
(Figure 5) may also suggest that increasing the hydrocarbon chain
from C8 could drive the antioxidant away from the interface and
into the emulsion droplet core where the phenolics would be a
poor antioxidant. Interestingly, this putative effect of the alkyl
chain length has already been suggested by Medina et al. (33) to
explain why the octyl ester of hydroxytyrosol exhibits higher
antioxidant capacity than the dodecyl one in fish oil-in-water
emulsion.
Effect of an Excess of Emulsifier on the Partioning Behavior of
Phenolics. Above 17 μM, increasing concentrations of Brij 35
(1, 2, and 10 mM) drastically impact the partition behavior of
phenolics and lead to the appearance and then the stepwise
disappearance of a partition cutoff trend (Figure 5). The depiction
of the phenolic concentration in the aqueous phase as a function
of the Brij 35 concentration (Figure 6) clearly showed that there is
a global tendency for esters (but not for free rosmarinic acid
which is already located in the aqueous phase) to return in the
water phase with increasing emulsifier concentration, especially
above 0.017 mM. As already reported by Laguerre et al. (19) in a
similar case with chlorogenate esters, the simplest hypothesis to
explain this trend is that above the critical micelle concentration
(CMC, 0.1-0.3 mM in water at room temperature (34,35)), Brij 35
micelles behave as vehicle for phenolics by means of comicelles to
carry them to the water phase. Indeed, the solubilization of
phenolics can take place in the palisade layer of the Brij 35
micelles. The formation of micelle with Brij 35 is very likely above
CMC since the area of its hydrophilic head (23 polyoxyethylene
groups) is much larger than the area of its hydrophobic tail (lauryl
chain). Consideringthe packing parameter (p) ofIsraelachvili (36)
it is quite evident that pBrij35 is less large than 1/3. This implies
that, just above the CMC, there would be formation of micelle
rather than other lipid polymorphs such as vesicles or bilayer
sheets for which a p comprised between 1/2 and 1 is theoretically
required.
Regarding the carrying of phenolics, the pioneer work of
Richards et al. (37), and more recently those of Yuji et al. (9)

Article
J. Agric. Food Chem., Vol. 58, No. 5, 2010 2875
ꢀ
(19) Laguerre, M.; Lopez Giraldo L. J.; Lecomte, J.; Figueroa-Espinoza,
LITERATURE CITED
ꢀ
M. C.; Barea, B.; Weiss, J.; Decker, E. A.; Villeneuve, P. Chain length
(1) McClements D. J. Lipid-based emulsions and emulsifiers. In Food
Lipids. Chemistry, Nutrition, and Biotechnology, 3rd ed.; Akoh, C. K.,
Min, D. B., Eds.; CRC Press: Boca Raton, 2008; pp 63-97.
(2) Porter, W. L. Recent trends in food applications of antioxidants. In
Autoxidation in food and biological systems; Simic, M. G., Karel, M.,
Eds.; Plenum Press: New York, 1980; pp 295-365.
(3) Porter, W. L.; Black, E. D.; Drolet, A. M. Use of polyamide
oxidative fluorescence test on lipid emulsion: contrast in relative
effectiveness of antioxidants in bulk versus dispersed systems.
J. Agric. Food Chem. 1989, 37, 615–624.
(4) Frankel, E. N.; Huang, S. W.; Kanner, J.; German, J. B. Interfacial
phenomena in the evaluation of antioxidants: bulk oils vs. emulsions.
J. Agric. Food Chem. 1994, 42, 1054–1059.
(5) Figueroa-Espinoza, M.-C.; Villeneuve, P. Phenolic acids enzymatic
lipophilization. J. Agric. Food Chem. 2005, 53, 2779–2787.
(6) Guyot, B.; Gueule, D.; Pina, M.; Graille, J.; Farines, V.; Farines, M.
Enzymatic synthesis of fatty esters in 5-caffeoyl quinic acid. Eur.
J. Lipid Sci. Technol. 2000, 102, 93–96.
affects antioxidant properties of chlorogenate esters in emulsion: the
cut-off theory behind the polar paradox. J. Agric. Food Chem., 2009,
57, 11335–11342.
ꢀ
ꢀ
(20) Laguerre, M.; Lopez Giraldo, L. J.; Lecomte, J.; Barea, B.; Cambon,
E.; Tchobo, P. F.; Barouh, N.; Villeneuve, P. Conjugated autoxidiz-
able triene (CAT) assay: a novel spectrophotometric method for
determination of antioxidant capacity using triacylglycerol as ultra-
violet probe. Anal. Biochem. 2008, 380, 282–290.
ꢀ
(21) Laguerre, M.; Lopez Giraldo, L. J.; Piombo, G.; Figueroa-
Espinoza, M. C.; Pina, M.; Benaissa, M.; Combe, A.; Rossignol
Castera, A.; Lecomte, J.; Villeneuve, P. Characterization of olive
leaf phenolics by ESI-MS and evaluation of their antioxidant
capacities by the CAT assay. J. Am. Oil Chem. Soc. 2009, 86,
1215–1225.
ꢀ
(22) Laguerre, M.; Lecomte, J.; Villeneuve, P. Evaluation of the
ability of antioxidant to counteract lipid oxidation: existing
methods, new trends and challenges. Prog. Lipid Res. 2007, 46,
244–282.
(7) Trujillo, M.; Mateos, R.; Collantes de Teran, L.; Espartero, J. L.;
Cert, R.; Jover, M.; Alcudia, F.; Bautista, J.; Cert, A.; Parrado, J.
Lipophilic hydroxytyrosyl esters. Antioxidant activity in lipid
matrices and biological systems. J. Agric. Food Chem. 2006, 54,
3779–3785.
ꢁ
ꢀ
ꢁ ꢁ
ꢀ
(23) Devı
´
nsky, F.; Kopecka-Leitmanova, A.; Sersen, F.; Balgavy, P.
Cut-off effect in antimicrobial activity and in membrane perturba-
tion efficiency of the homologous series of N,N-dimethylamine
oxides. J. Pharm. Pharmacol. 1990, 42, 790–794.
(24) Birnie, C. R.; Malamud, D.; Schnaare, R. L. Antimicrobial evalua-
tion of N-alkyl betaines and N-alkyl-N,N-dimethylamine oxides
with variations in chain length. Antimicrob. Agents Chemother. 2000,
44, 2514–2517.
(25) Meyer, K. H.; Hemmi, H. Beitrage zur theorie der narkose. III.
Biochem. Z. 1935, 277, 39–71.
(26) Nakahiro, M.; Arakawa, O.; Nishimura, T.; Narahashi, T. Potentia-
tion of GABA-induced Cl- current by a series of n-alcohols
disappears at a cutoff point of a longer-chain n-alcohol in rat dorsal
root ganglion neurons. Neurosci. Lett. 1996, 205, 127–130.
(27) Locatelli, C.; Rosso, R.; Santos-Silva, M. C.; de Souza, C. A.;
ꢀ
(8) Lopez Giraldo, L. J.; Laguerre, M.; Lecomte, J.; Figueroa-Espinoza,
ꢀ
M. C.; Barouh, N.; Barea, B.; Villeneuve, P. Lipase-catalyzed
synthesis of chlorogenate fatty esters in solvent-free medium.
Enzyme Microb. Technol. 2007, 41, 721–726.
ꢀ
(9) Yuji, H.; Weiss, J.; Villeneuve, P.; Lopez Giraldo, L. J.; Figueroa-
Espinoza, M. C.; Decker, E. A. Ability of surface-active antioxidants
to inhibit lipid oxidation in oil-in-water emulsion. J. Agric. Food
Chem. 2007, 55, 11052–11056.
(10) Mateos, R.; Trujillo, M.; Pereira-Caro, G.; Madrona, A.; Cert, A.;
Espartero, J. L. New lipophilic tyrosyl esters. Comparative antiox-
idant evaluation with hydroxytyrosyl esters. J. Agric. Food Chem.
2008, 56, 10960–10966.
(11) Kanjilal, S.; Shanker, K. S.; Rao, K. S.; Reddy, K. K.; Rao, B. V. S.
K.; Kumar, K. B. S.; Kantam, M. L.; Prasad, R. B. N. Chemo-
enzymatic synthesis of lipophilic ferulates and their evaluation for
antioxidant and antimicrobial activities. Eur. J. Lipid Sci. Technol.
2008, 110, 1175–1182.
Licınio, M. A.; Leal, P.; Bazzo, M. L.; Yunes, R. A.; Creczynski-
´
Pasa, T. B. Ester derivatives of gallic acid with potential toxicity
toward L1210 leukemia cells. Bioorg. Med. Chem. 2008, 16, 3791–
3799.
(28) Wong, Y. L.; Curfman, C. L.; Doncel, G. F.; Hubieki, M. P.;
Dudding, T. C.; Savle, P. S.; Gandour, R. D. Spermicidal, anti-HIV,
and micellar properties of di- and trihydroxylated cationic surfac-
tants. Tetrahedron 2002, 58, 45–54.
ꢀ
(12) Lopez Giraldo, L. J.; Laguerre, M.; Lecomte, J.; Figueroa-
ꢀ
Espinoza, M. C.; Barea, B.; Weiss, J.; Decker, E. A.; Villeneuve,
(29) Brouillard, R.; Wigand, M.-C.; Dangles, O.; Cheminat, A. pH and
solvent effects on the copigmentation reaction of malvin with
polyphenols, purine and pyrimidine derivatives. J. Chem. Soc.,
Perkin Trans. 2 1991, 1235–1241.
P. Stationary and kinetic study of antiradical activity of chloro-
genic acid and its alkyl esters by DPPH method. J. Agric. Food
Chem. 2009, 57, 863–870.
ꢀ
ꢀ
(13) Lecomte, J.; Lopez Giraldo, L. J.; Laguerre, M.; Barea, B.;
Villeneuve, P. Synthesis, characterization and free radical scavenging
properties of rosmarinic acid fatty esters. J. Am. Oil Chem. Soc.,
[Online early access]. DOI: 10.1007/s11746-010-1543-8. Published
Online: Jan 23, 2010.
(30) Dangles, O.; Saito, N.; Brouillard, R. Kinetic and thermo-
dynamic control of flavylium hydration in the pelargonidin-
cinnamic acid complexation. Origin of the extraordinary flower
color diversity of Pharbitis nil. J. Am. Chem. Soc. 1993, 115,
3125–3132.
(14) Hunneche, C. S.; Lund, M. N.; Skibsted, L. H.; Nielsen, J. Antioxidant
activity of a combinatorial library of emulsifier-antioxidant bioconju-
gates. J. Agric. Food Chem. 2008, 56, 9258–9268.
(15) Pereira-Caro, G.; Madrona, A.; Bravo, L.; Espartero, J. L.; Alcudia,
F.; Arturo Cert, A.; Mateos, R. Antioxidant activity evaluation of
alkyl hydroxytyrosyl ethers, a new class of hydroxytyrosol deriva-
tives. Food Chem. 2009, 115, 86–91.
(31) Dangles, O.; Brouillard, R. Polyphenol interactions. The copigmen-
tation case: thermodynamic data from temperature variation and
relaxation kinetics. Medium effect. Can. J. Chem. 1992, 70, 2174–
2189.
(32) Fendler, J. H. Membrane mimetic chemistry; John Wiley & Sons: New
York, 1982.
ꢀ
(33) Medina, I.; Lois, S.; Alcantara, D.; Lucas, R.; Morales, J. C. Effect
(16) Madrona, A.; Pereira-Caro, G.; Mateos, R.; Rodrıguez, G.; Trujillo,
´
of lipophilization of hydroxytyrosol on its antioxidant activity in fish
oils and fish oil-in-water emulsions. J. Agric. Food Chem. 2009, 57,
9773–9779.
ꢀ
~
M.; Fernandez-Bolanos, J.; Espartero, J. L. Synthesis of hydroxy-
tyrosyl alkyl ethers from olive oil waste waters. Molecules 2009, 14,
1762–1772.
(34) Becher, P. In Nonionic Surfactants; Schick, M. J., Ed.; Marcel Dekker:
New York, 1967.
€
(17) Stockmann, H.; Schwarz, K.; Huynh-Ba, T. The influence of various
emulsifiers on the partitioning and antioxidant activity of hydroxy-
benzoic acids and their derivatives in oil-in-water emulsions.
J. Am. Oil Chem. Soc. 2000, 77, 535–542.
(35) Zakharova, L. Y.; Valeeva, F. G.; Ibragimova, A. R.; Zakharov,
V. M.; Kudryavtseva, L. A.; Elistratova, Y. G.; Mustafina, A. R.;
Konovalov, A. I.; Shtykov, S. N.; Bogomolova, I. V. Properties
of a sodium dodecyl sulfate-Brij 35 binary micellar system
and their effect on the alkaline hydrolysis of O-ethyl-O-p-
nitrophenylchloromethylphosphonate. Colloid J. 2007, 69, 718–
725.
ꢀ
(18) Sasaki, K.; Alamed, J.; Weiss, J.; Villeneuve, P.; Lopez Giraldo, L. J.;
Lecomte, J.; Figueroa-Espinoza, M.-C.; Decker, E. A. Relationship
between the physical properties of chlorogenic acid esters and their
ability to inhibit lipid oxidation in oil-in-water emulsions. Food
Chem. 2010, 118, 830–835.

2876 J. Agric. Food Chem., Vol. 58, No. 5, 2010
Laguerre et al.
(36) Israelachvili, J. N. Intermolecular and surface forces; Academic Press:
London, U.K., 1992.
micelles and alter lipid oxidation rates in emulsions. J. Agric. Food
Chem. 2002, 50, 5445–5449.
(37) Richards, M. P.; Chaiyasit, W.; McClements, D. J.; Decker, E. A.
Ability of surfactant micelles to alter the partitioning of phenolic
antioxidants in oil-in-water emulsions. J. Agric. Food Chem. 2002,
50, 1254–1259.
Received for review November 23, 2009. Revised manuscript received
ꢀ
January 15, 2010. Accepted January 19, 2010. Luis Javier Lopez
Giraldo thanks the support from Programme Alβan, the European
Union Programme of High Level Scholarships for Latin America,
scholarship No. E05D055786CO.
(38) Nuchi, C. D.; Hernandez, P.; McClements, D. J.; Decker, E. A.
Ability of lipid hydroperoxides to partition into the surfactant