Article

pubs.acs.org/jnp

Antioxidant Activity of Individual Steryl Ferulates from Various

Cereal Grain Sources
Dan Zhu, Antoni Sánchez-Ferrer,* and Laura Nyström*
Institute of Food, Nutrition and Health, Department of Health Sciences and Technology, ETH Zurich, Schmelzbergstrasse 9,
CH-8092 Zurich, Switzerland

ABSTRACT: Steryl ferulates (SFs) are a subclass of bioactive

lipids contributing to the health-promoting effects of whole
grains. Most related studies focus on γ-oryzanol, a SF mixture
from rice, since individual steryl ferulates are not commercially
available. There is little evidence that individual SFs may vary
in their bioactivity. The aim of this study was to evaluate the
antioxidant activity of eight individual SFs by determining their
radical scavenging capacity. Additional molecular properties of
the individual SFs were determined by molecular simulation in
order to identify correlations with their antioxidant activities. Our study demonstrates that individual SFs exhibit 1,1-diphenyl-2-
picrylhydrazyl radical, hydroxyl radical, and superoxide anion radical scavenging abilities with subtle differences that were highly
dependent on the kind of reaction taking place. The grouping of SFs by principle component analysis was mainly attributed to
molecular properties, not antioxidant activities. Solvation energy was significantly correlated with some experimental
observations. To our knowledge, this is the first study to evaluate the antioxidant activity of eight individual steryl ferulates from
different sources. Results of this work will provide better insight into the antioxidant activity of SFs and the health benefits of
whole grains.

S teryl ferulates (SFs) are the esters of phytosterols and

ferulic acid, which are present in the bran of some grains
such as rice (Oryza sativa L.), wheat (Triticum aestivum L.),
reaction of lipid oxidation as alkyl radicals.14 To date,
bioactivity studies of SFs have mostly been performed with
ORY due to the lack of individual SFs on the market. For
corn (Zea mays L.), and triticale (×Triticosecale Wittmack).1 instance, Kim et al. proved that ORY effectively improved flavor
They are bioactive lipids shown to possess health benefits such and oxidative stability of refrigerated cooked beef.15 Juliano et
as lowering cholesterol,2 inhibiting melanogenesis,3 and al. demonstrated that ORY prevented AMVN-triggered lip-
exhibiting antioxidant4 and anti-inflammatory5 activities. To operoxidation and improved the oxidative stability of oils.16
date, at least 21 different steryl ferulates, varying only in the Recently, ORY was also determined to exhibit 2,2′-azinobis(3-
type of sterol moiety, have been detected in various studies.6−9 ethylbenzothiazoline-6-sulfonic acid) cation radical (ABTS•+)
The different SFs investigated in this study are shown in Figure and superoxide anion radical (O2•−) scavenging activity, as well
1. Their total content, as well as the composition of individual as a strong inhibition effect on linoleic acid peroxidation.17
SFs, varied depending on the grain source, genotype, and Furthermore, there are indications that individual SFs may
environmental factors.10 In rice bran (total SFs, commonly vary in their antioxidant activity, and therefore, the sterol
known as γ-oryzanol (ORY), 1550−8400 μg·g−1 dry weight) composition may be an important aspect in defining the activity
the ratio of 4,4-dimethylsteryl ferulates (SFs 1 and 2) and 4- of SF mixtures. However, there are limited data available on
desmethylsteryl ferulates (SFs 3−7) is around 65:35.11 these differences in antioxidant effects. Xu et al. reported that
However, only 4-desmethylsteryl ferulates are found in wheat SFs 2 > 5 = 1 in preventing cholesterol oxidation.18
bran (SFs 3−6, total 297−584 μg·g−1) and corn bran (SFs 3−
Furthermore, in terms of preventing hydroperoxide formation
7, total 200−250 μg·g−1).7,12 SF 3 accounts for approximately
in methyl linoleate bulk oil systems, the SF mixture from wheat
60% of SFs in wheat, while in corn, the most abundant is SF 4,
and rye > SFs 6 and 8 > ORY and 1.4 Huang demonstrated that
representing approximately 70% of total SFs.1
SFs have been shown to prevent oxidation in various the antioxidant activity of SF 2 > 1 and SF 5 > ORY in an
biological systems. The mechanism of their antioxidant activity SVEC-10 mouse lymph endothelial cell model, using tert-butyl
results from the phenolic proton in the ferulic acid moiety, hydroperoxide (tBHP) as an oxidizing agent.19 Since their
which can be abstracted by any radical present in the media, antioxidant capacities differ considerably from one to another, it
and the resulting SF radical is stabilized by resonance along the is of great interest to evaluate the activities of individual SFs
π-electron system constituted by the aromatic ring and the more systematically.
carboxylate in para position to the phenol group.7,13 Some
researchers have also suggested that the SF radical might still Received: October 14, 2015
influence oxidation, for example, by interfering with the chain
© XXXX American Chemical Society and
American Society of Pharmacognosy A DOI: 10.1021/acs.jnatprod.5b00880
J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products Article

Figure 1. Molecular structures of (1) cycloartenyl ferulate, (2) 24-methylenecycloartanyl ferulate, (3) campestanyl ferulate, (4) sitostanyl ferulate,
(5) campesteryl ferulate, (6) sitosteryl ferulate, (7) stigmasteryl ferulate, (8) cholesteryl ferulate, (9) ferulic acid, (10) methyl ferulate, (11) ethyl
ferulate, and (12) a general sterol skeleton based on IUPAC-IUB 1989.

Table 1. Kinetic Parameters for the DPPH Radical Reactiona

absorbance (t → ∞)
−1 −1
compound k (μM ·s ) 1.67 μM 16.7 μM 60 μM
1 0.010 ± 0.004 1.301 ± 0.001 1.109 ± 0.003 0.72 ± 0.01
2 0.011 ± 0.002 1.283 ± 0.001 1.030 ± 0.004 0.82 ± 0.01
3 0.018 ± 0.004 1.198 ± 0.001 1.091 ± 0.005 0.79 ± 0.03
4 0.015 ± 0.002 1.156 ± 0.001 0.991 ± 0.004 0.64 ± 0.01
5 0.013 ± 0.002 1.273 ± 0.001 1.045 ± 0.004 0.70 ± 0.01
6 0.013 ± 0.001 1.298 ± 0.001 0.944 ± 0.004 0.64 ± 0.01
7 0.012 ± 0.002 1.144 ± 0.001 1.010 ± 0.004 0.72 ± 0.01
8 0.012 ± 0.002 1.245 ± 0.001 0.979 ± 0.005 0.74 ± 0.01
9 0.012 ± 0.001 1.232 ± 0.001 0.913 ± 0.005 0.53 ± 0.01
10 0.014 ± 0.002 1.008 ± 0.001 0.896 ± 0.005 0.57 ± 0.01
11 0.013 ± 0.002 0.992 ± 0.001 0.851 ± 0.003 0.58 ± 0.01
ORY 0.011 ± 0.006 1.246 ± 0.001 0.981 ± 0.003 0.69 ± 0.01
WB 0.014 ± 0.002 1.245 ± 0.001 0.992 ± 0.006 0.65 ± 0.01
pyrogallol 0.12 ± 0.01
a
Data are expressed as mean ± SEM (n = 3−9).

The aim of this study was to evaluate the antioxidant activity C6); and SFs 5, 6, 7, and 8 (with unsaturated sterol and a
of eight individual SFs. SFs 2−6 were purified from ORY and double bond at C5 and C6). Additionally, antioxidant effects of
wheat bran. SF 7 and SF 8, which occur only in trace amounts ferulic acid (9), methyl ferulate (10), ethyl ferulate (11), ORY,
in nature, were synthesized. SFs were divided into groups and the SF mixtures from wheat bran (WB) were evaluated.
according to structure similarity: SFs 1 and 2 (with methyl Antioxidant activity was determined by evaluating the
groups at C4 and C14 and a cyclopropyl ring at C9/C10); SFs scavenging capacity of 1,1-diphenyl-2-picrylhydrazyl (DPPH)
3 and 4 (with a saturated sterol, no double bond at C5 and radicals (DPPH•) in methanol, hydroxyl radical (•OH), and
B DOI: 10.1021/acs.jnatprod.5b00880
J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products Article

superoxide anion radical (O2•−) in an in vitro water where A0(DPPH) is the absorbance of DPPH• solution without
environment. Additional molecular properties of individual any antioxidant and A∞(DPPH) is the absorbance in the presence
SFs were determined by molecular simulation, in order to of an antioxidant when the reaction is finished (t → ∞).
correlate them with their antioxidant activities. As far as we Generally, the efficiencies of ferulates, as well as compound 9,
know, this study is the first to provide a comprehensive decreased with increasing concentrations (Figure 2). This could
comparison of the antioxidant activity of individual SFs.

■ RESULTS AND DISCUSSION

Scavenging Effect on DPPH Radical. DPPH• scavenging
measurement was carried out with antioxidants at concen-
trations of 1.67, 16.7, and 60.0 μM. The irreversible reaction is
presented by eq 1. The decrease of the concentration of
DPPH• ([DPPH•]) over time can be written as shown in eq 2,
following global second-order kinetics. In the reaction system,
the concentrations of DPPH• and product DPPH2 can be
determined from eqs 3 and 4 with a rate constant k, as well as
initial concentrations of DPPH• ([DPPH•]0) and SF ([SF]0).
The kinetics parameters, k value, and plateau value (the
absorption when t → ∞) were obtained from curve-fitting of
the absorbance vs time plots (Table 1).
k
DPPH• + SF → DPPH 2 + SF• (1)
Figure 2. Efficiency of DPPH radical scavenging.
•
d[DPPH ]
− = k[DPPH•][SF]
dt (2)
be explained by the lower relative amount of DPPH• per unit of
[DPPH•] = antioxidant in treatments with higher concentrations of
antioxidant, based on the constant concentration of DPPH•.
⎛ • ⎞ At the lowest antioxidant concentration of 1.67 μM, the
⎜ 1 − e([DPPH ]0 − [SF]0 )kt ⎟
•
[DPPH ]0 ⎜1 − differences in DPPH• scavenging efficiency among these
⎜ [DPPH•]0 ([DPPH•]0 − [SF]0 )kt ⎟
⎟
⎝ 1 − [SF] e ⎠ compounds were very clear. Ferulates 10 and 11 were observed
0 (3)
to have the highest efficiency. This may be partially explained
by the fast diffusion of such small molecules compared to the
[DPPH 2] = SFs (D = kBT/6πηR, where kB is the Boltzmann constant, T is
⎛ • ⎞ the absolute temperature, η is the viscosity of the medium, and
⎜
• 1 − e([DPPH ]0 − [SF]0 )kt ⎟ R is the radius of the solute molecule). Ferulates 10 and 11
[DPPH ]0 ⎜ • ⎟⎟
⎜1 − have much smaller radii than SF molecules; hence they had
[DPPH ] •
0 ([DPPH ] − [SF] )kt
⎝ e 0 0
⎠
[SF]0 (4) higher diffusion coefficients (D) and exhibited higher reaction
• efficiencies. However, compound 9, which also has the smallest
All of these antioxidants showed DPPH scavenging effect. radius of molecules studied, showed only a moderate efficiency,
Each antioxidant showed the same rate constant at different which could be explained by the strong solvent interaction
concentrations, confirming the kinetic order of the scavenging between the polar protic solvent (methanol) and the carboxylic
process. Rate constants of individual SFs decreased in the group of compound 9, especially when compound 9 was at very
following order: SFs 3 and 4 ≥ SFs 5, 6, 7, and 8 ≥ SFs 1 and low concentrations. Additionally, compound 9 may form
2. ORY, which contains mainly SFs 1 and 2, had a low rate noncovalent intermolecular interactions and dimerize via O−
constant, but with high experimental error. Additionally, WB H···O hydrogen bonds between carboxylic groups, which may
was found to show a higher rate constant than ORY, which was further restrict its diffusion compared to the small ferulates.
in agreement with the behavior of individual SFs. Furthermore, High efficiencies were demonstrated by SFs 7, 3, and 4, as
ferulates 10 and 11 had moderate and similar kinetics. compared with other SFs; however, according to the results of
However, compound 9, the smallest molecule in this study, the simulation studies, SFs 7, 3, and 4 did not have the smallest
did not exhibit any advantage in the scavenging process. radii. We therefore infer that other factors in the reaction may
Moreover, the positive control molecule, pyrogallol, was contribute to this observation, for example intra- or
observed to have a kinetic constant nearly 10-fold of that of intermolecular interactions or solvent interaction. When the
the ferulates and compound 9. Nevertheless, the differences in concentration of antioxidant was increased to 16.7 and 60 μM,
rate constants among ferulates and compound 9 in this reaction the differences in efficiencies among ferulates and compound 9
were very small. were less notable. Ferulates 10 and 11 and compound 9
The absorbance in the final state or plateau (t → ∞; A∞) was showed slightly higher efficiencies than the SFs at 16.7 μM;
also obtained from the curve-fitting. Moreover, the efficiency nevertheless, all the compounds had similar efficiencies at 60
(ε) of each antioxidant on DPPH• scavenging can be μM. This suggests that when the concentration of antioxidant
determined from was very low, the number of effective collisions was greatly
A 0(DPPH) − A∞ (DPPH) influenced by the solvent effect; meanwhile with higher
εDPPH = concentrations of antioxidant, the effective collisions could
A 0(DPPH)[SF]0 (5) mostly result from its high relative concentration.
C DOI: 10.1021/acs.jnatprod.5b00880
J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products Article

The DPPH• scavenging activities of SFs have also been Islam et al. also reported SFs 1, 2, and 6 (concentration of 40
reported in previous studies. Akiyama et al. reported similar μM, with 0.1% supporting solvent ethanol: DMSO 9:1) had
DPPH• scavenging activity for SFs 1, 2, 5, and 6 and similar •OH scavenging capacities, while compound 9 showed
compound 9 (>100 μM).20 Islam et al., who used a comparable significantly higher activity than the SFs.21 However, Juliano et
reaction system with ours, also found SFs 1, 2, and 6 and al. reported that ORY (concentrations of 1.65 and 16.5 μM,
compound 9 had similar DPPH• scavenging activities.21 In with 1% ethanol) had no •OH scavenging activity, when
another study, Kikuzaki et al. reported that SF 1, SF 2, and measured by its inhibition of p-nitrosodimethylaniline-trapping
ORY (20 μM) had the same DPPH• scavenging capacity, while of •OH.16 Given the differences in the reaction systems, our
compound 9 was slightly better than the SFs.22 Moreover, data are not directly comparable with their findings. Generally,
compound 9 has also been reported to be a better DPPH• all the antioxidants in our study were found to have similar
•
scavenger than SFs 6 and 8, as well as the SF mixture from rye OH scavenging abilities.
and wheat (17 μM).4 However, all of these studies only Scavenging of Superoxide Anion Radical. Scavenging
compared scavenging capacities after the reaction. To the best of O2•− was investigated with antioxidants at 40, 80, and 160
of our knowledge, ours is the first study to investigate reaction μM. This method examined the ability of antioxidants to
kinetics with DPPH•, which provide more information on compete with the probe nitrotetrazolium blue chloride (NBT)
individual SFs as DPPH• scavengers. in scavenging O2•−. The irreversible reaction of NBT with O2•−
Scavenging of Hydroxyl Radical. In our study, •OH in the system is described by eq 6. At the onset of the reaction,
scavenging capacity was determined using electron spin the concentration of O2•− ([O2•−]0) is much higher than that
resonance (ESR) with the spin trap DMPO method. The of NBT ([NBT]0); then the decrease in the concentration of
•
OH is very short-lived, and DMPO is commonly used to trap NBT ([NBT]) over time can be written, as shown in eq 7, as a
•
OH, as the DMPO-OH adduct has a half-life of 12−156 min pseudo-first-order kinetic process. Furthermore, the concen-
in neutral solutions.23 Due to the limitation of solubility, the tration of NBT and the corresponding product NBTH2 can be
highest concentration of SF studied was 15.0 μM. Generally, all determined from eqs 8 and 9 with rate constant k1 and [NBT]0
the compounds in this study scavenged the •OH concentration as parameters. The reaction of SF with O2•− in the system has
dependently (Table 2). At 1.5 μM, similar •OH scavenging the same kinetics as NBT, as shown in eqs 10 through 13. The
kinetic parameter, k1, of reaction 6 and the absorbance of
Table 2. Hydroxyl Radical Scavenging Activity (RSA)a NBTH2 when the reaction is finished (t → ∞; A∞) were also
obtained from the curve-fitting experimental measurements
RSA (%) (Table 3). The differences in k1 and A∞ reflect the competitive
compound 1.5 μM 2.5 μM 5.0 μM 15.0 μM ability of SFs in reacting O2•− with NBT.
1 23 ± 2 21 ± 2 27 ± 5 43 ± 3 k1
2 19 ± 3 24 ± 1 36 ± 3 45 ± 4 2O2•− + NBT → 2O2 + NBTH 2 (6)
3 23 ± 2 27 ± 2 34 ± 2 43 ± 2
d[NBT]
4 22 ± 3 25 ± 4 28 ± 2 40 ± 2 − = k1[O2•−]2 [NBT] ≈ k1[NBT]
5 21 ± 3 25 ± 3 29 ± 2 39 ± 3 dt (7)
6 20 ± 3 24 ± 3 31 ± 2 43 ± 1
[NBT] = [NBT]0 e−k1t (8)
7 23 ± 4 26 ± 4 30 ± 1 44 ± 2
8 24 ± 3 24 ± 3 30 ± 3 40 ± 3
[NBTH 2] = [NBT]0 (1 − e−k1t ) (9)
9 23 ± 3 26 ± 1 33 ± 3 40 ± 3
10 19 ± 5 31 ± 2 36 ± 1 43 ± 3 k2
11 19 ± 6 29 ± 4 36 ± 3 42 ± 3 O2•− + SF → O2 + SF• (10)
ORY 25 ± 1 25 ± 4 31 ± 3 42 ± 3
d[SF]
WB 23 ± 2 22 ± 1 33 ± 3 43 ± 1 − = k 2[O2•−][SF] ≈ k 2[SF]
dt (11)
a
Data are expressed as mean ± SD (n = 3).
[SF] = [SF]0 e−k 2t (12)

[SF•] = [SF]0 (1 − e−k 2t ) (13)

capacities (19−25%) were observed among the antioxidants
tested. At 2.5 μM, ferulate 10 (29%) was only slightly more As shown in Table 3, some of the rate constants of NBT
effective than SF 1 and WB (21% and 22%, respectively), and were lower with higher concentrations of antioxidant (SFs 2, 5,
other compounds did not differ from each other. While using a 6, 7, 8), suggesting these compounds interfere with NBT dose
higher concentration of 5 μM, ferulates 10 and 11 as well as SF dependently. The rate constant was dramatically lower (40%)
2 exhibited higher scavenging capacities (around 36%) than SF in the presence of SFs 7 and 8 at concentrations of 80 and 160
1 (27%). Nevertheless, at the highest concentration of 15.0 μM, μM. Meanwhile, for SFs 5 and 6, significantly lower rate
all the ferulates, as well as compound 9, showed very similar constants (30%) were observed only with the high concen-
activities (39−44%). In this study, compound 9, which has a tration of 160 μM. Moreover, for SF 2, the effect was gradual:
higher solubility than the ferulates, did not show any advantage 0%, 30%, and 45%, for concentrations of 40, 80, and 160 μM,
in scavenging •OH. In this water environment (pH around respectively. In the case of SF 1, which has a similar structure to
7.4), the carboxyl group of compound 9 could be deprotonated, SF 2, the rate was consistently low (30%) compared to the
leaving only the proton from the phenolic hydroxyl group NBT reaction at all three concentrations. Furthermore, SFs 3
available for •OH scavenging, which could explain why the and 4 demonstrated interference with the NBT reaction (by
stable, deprotonated form (carboxylate) may exhibit similar approximately 30%) only at concentrations of 40 and 160 μM.
behavior to other ferulates in this reaction system. On the other hand, for the SF mixtures, WB and ORY were
D DOI: 10.1021/acs.jnatprod.5b00880
J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products Article

Table 3. Kinetic Parameters for Superoxide Anion Radical Reactiona

k1 (min−1) absorbance (t → ∞)
compound 40 μM 80 μM 160 μM 40 μM 80 μM 160 μM
1 0.12 ± 0.01 0.14 ± 0.01 0.15 ± 0.01 1.8 ± 0.1 1.57 ± 0.06 1.48 ± 0.06
2 0.18 ± 0.01 0.13 ± 0.01 0.10 ± 0.01 1.68 ± 0.05 1.79 ± 0.09 1.80 ± 0.06
3 0.13 ± 0.03 0.19 ± 0.01 0.13 ± 0.01 1.89 ± 0.01 1.47 ± 0.03 1.57 ± 0.08
4 0.14 ± 0.02 0.21 ± 0.01 0.14 ± 0.01 1.85 ± 0.05 1.35 ± 0.02 1.47 ± 0.05
5 0.17 ± 0.01 0.18 ± 0.01 0.13 ± 0.01 1.65 ± 0.04 1.40 ± 0.03 1.19 ± 0.05
6 0.19 ± 0.01 0.19 ± 0.01 0.13 ± 0.01 1.64 ± 0.06 1.38 ± 0.04 1.16 ± 0.03
7 0.18 ± 0.01 0.11 ± 0.01 0.11 ± 0.01 1.68 ± 0.05 1.76 ± 0.07 1.78 ± 0.06
8 0.17 ± 0.01 0.11 ± 0.01 0.11 ± 0.01 1.76 ± 0.04 1.71 ± 0.08 1.82 ± 0.08
9 0.20 ± 0.01 0.14 ± 0.03 0.19 ± 0.02 1.27 ± 0.07 1.4 ± 0.2 1.03 ± 0.08
10 0.15 ± 0.01 0.15 ± 0.01 0.14 ± 0.02 1.90 ± 0.01 1.72 ± 0.06 1.6 ± 0.2
11 0.14 ± 0.01 0.16 ± 0.03 0.13 ± 0.02 1.90 ± 0.01 1.7 ± 0.1 1.6 ± 0.2
ORY 0.17 ± 0.01 0.15 ± 0.04 0.18 ± 0.01 1.60 ± 0.04 1.4 ± 0.2 1.33 ± 0.07
WB 0.21 ± 0.01 0.10 ± 0.01 0.20 ± 0.01 1.52 ± 0.03 1.75 ± 0.09 1.19 ± 0.02
NBT alone 0.188 ± 0.001 1.90 ± 0.01
a
Data are expressed as mean ± SEM (n = 3).

able to interfere with NBT only at a concentration of 80 μM, compounds. Compound 9, which is the most water-soluble
with approximately 45% reduction in the rate constant for WB antioxidant at pH 7.4, was found to have the highest efficiency.
and 20% for ORY. Additionally, both ferulates 10 and 11 Among the SFs, the SF mixture WB was the most effective,
exhibited a 20% difference in rate constants from the NBT followed by ORY. SFs 2, 5, 6, and 7 showed similar efficiency
reaction for all tested concentrations. However, compound 9 and were slightly more effective than SFs 1 and 8. Furthermore,
only slightly interfered with the NBT reaction at 80 μM. the weakest SFs were found to be SFs 3 and 4. However,
Nevertheless, SFs were able to effectively compete with NBT- ferulates 10 and 11, which consistently decreased the rate
trapping of O2•−. SFs 2, 7, and 8 appeared to have slightly constant of NBT by 20%, cannot scavenge O2•− at this
higher competitive abilities than the other individual SFs, concentration. From this, we deduce that ferulates could be
especially at the highest concentrations. considered as amphiphile molecules due to the hydrophilic
The absorbance of NBTH2 when the reaction was completed (phenolic hydroxyl group) and lipophilic (methyl, ethyl, or
(t → ∞; A∞) was also obtained from curve-fitting (Table 3), to sterol moiety) characteristics, and therefore self-assembly may
determine the amount of O2•− in the system. The efficiency (ε) occur in this environment. Their different potentials to
of each antioxidant in reducing O2•− can be determined from aggregate in the water/solvent mixture may induce their
eq 14: different efficiencies on radical scavenging. When applying a
A 0(NBTH2) − A∞ (NBTH2) higher concentration of 80 μM, all the antioxidants showed
εO2•− = O2•− scavenging activity. Compound 9, ORY, and SFs 3−6
A 0(NBTH2)[SF]0 (14) were observed to have similar efficiency and were slightly more
effective than the other compounds. Moreover, ferulates 10 and
where A0(NBTH2) is the initial absorbance of NBTH2 without the 11 showed relatively low efficiency at scavenging O2•−. At the
presence of antioxidant, and A∞(NBTH2) is the absorbance of concentration of 160 μM, the highest efficiency was still
NBTH2 after complete reaction with the antioxidant. The obtained with compound 9, followed by SFs 5 and 6, ORY, and
efficiencies of most of the antioxidants in this study were lower WB. Overall, among all individual SFs, SFs 5 and 6 generally
with higher concentrations (Figure 3). At the 40 μM level, the showed the highest efficiency to scavenge O2•− at all tested
differences in efficiency were very clear among these concentrations.
However, the O2•− scavenging effect of SFs was not observed
by Juliano et al.16 They reported that ORY at the concentration
of 10 μM had no scavenging activity in a system where O2•−
was produced by spontaneous autoxidation of FeCl2 in
morpholinepropanesulfonic acid buffer. Recently, Saenjum et
al. reported that ORY showed an O2•− scavenging effect (IC50
30 μM) in a phenazine methosulfate−β-nicotinamide adenine
dinucleotide system.17 Due to the different radical-generating
systems and concentrations of antioxidants employed, our data
are not directly comparable with their findings. To the best of
our knowledge, this is the first study to investigate and compare
the O2•− scavenging activity of individual SFs.
Simulation and Correlation Analysis. We performed
molecular simulations to explain the differences in antioxidant
activity from the point of view of the SF’s molecular properties.
The dipole moment and quantitative structure−activity
relationship parameters, i.e., logP, surface area, volume, and
Figure 3. Efficiency of superoxide anion radical scavenging. solvation energy (hydration), were obtained from the quantum
E DOI: 10.1021/acs.jnatprod.5b00880
J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products Article

Table 4. Molecular Properties from Molecular Simulations

Connolly surface area Connolly volume polarizability dipole moment molar mass
compound (Å2) (Å3) solvation energy (kcal·mol−1) logP (Å3) (D) (Da)
1 928 1734 −4.31 7.72 71 3.55 602.9
2 943 1776 −4.60 8.09 72 3.72 616.9
3 914 1705 −3.92 7.61 68 3.55 578.9
4 939 1754 −3.72 8.01 70 3.52 592.9
5 924 1706 −4.15 7.17 68 3.51 576.9
6 945 1748 −3.98 7.57 70 3.49 590.9
7 942 1745 −4.35 7.31 69 3.45 588.9
8 919 1672 −4.28 6.84 66 3.49 562.8
9 379 591 −14.1 −0.63 20 4.07 194.2
10 414 652 −8.77 −0.60 21 3.74 208.2
11 450 709 −8.03 −0.25 23 3.73 222.2

mechanics simulations (Table 4). Among the individual SFs, SF observed to be better antioxidants than 4,4-dimethylsteryl
8 was found to have the smallest volume, as well as the smallest ferulates in DPPH•4 and oil models,4,24,26 for which it was
logP value, while SF 2 had the highest dipole moment and suggested to be a relic of the negative effects from the dimethyl
solvation energy. Generally, the differences between SFs with groups at C4 as well as the cyclopropyl ring at C9/C19.
respect to these parameters were very small. Moreover, saturated 4-desmethylsteryl ferulates were observed
The grouping of SFs was performed by principal component to have slightly higher antioxidant activity during frying than
analysis (PCA) with variables of their antioxidant activities and those with a double bond at C5/C6.24 Nevertheless, some
molecular properties (Figure 4), but compounds 9−11 were observations also showed similar antioxidant activity for
not included for further analyses. The overview of SFs 1−8, as individual SFs in DPPH.20,21 In this study, generally, the
well as all the variables, is presented in a bi-plot (Figure 4A). difference in antioxidant activity of individual SFs was very
PC1 and PC2 explained 94% and 4% (in total 98%) of the small and depended greatly on the reaction models. In addition
variance. Considering the PCA scores (Figure 4B), four groups to the inner-molecular variations, the intermolecular inter-
could be proposed, namely, SFs 1 and 2, SFs 4, 6, and 7, SFs 3 actions, as well as interactions with the solvent (environment),
and 5, and SF 8. Correlation loadings (Figure 4C) showed that should also be considered. Thus, the same SF compound might
the molecular properties, except for solvation energy, had a show different scavenging properties depending on the
position at the right side of PC1, close to the 100% explained substrate studied, as well as the kind of radical scavenging
variance circle, indicating that these variables greatly contribute process taking place. Therefore, the antioxidant activity of SFs
to the SF groupings. However, the variables related to the SF’s should be reconsidered based on the reaction occurring, and
antioxidant activities were located near the inner ellipse (less with respect to the food product to be studied, along with other
than 50% of explained variance), indicating these variables had synergistic processes that might happen in natural settings.
little influence. Nevertheless, the variable of solvation energy
was very close to the position of antioxidant activity, suggesting
they are somehow correlated.
■ EXPERIMENTAL SECTION
General Experimental Procedures. Acetic acid (Ph Eur) and 1-
Pearson’s correlation was performed for compounds 1−8 to butanol (Ph Eur) were purchased from Merck, Darmstadt, Germany.
support the correlation loadings of PCA; compounds 9−11 are Acetone (≥99.9%), acetonitrile (ACN; ≥99.9%), cholesterol (99%),
not shown because they do not fall in the same linear tendency. 3,4-dihydro-2H-pyran (97%), 5,5-dimethyl-1-pyrroline N-oxide
Examining all the correlations between antioxidant activities (DMPO; ≥99%), 2,2-diphenyl-1-picrylhydrazyl (DPPH), ethylenedia-
minetetraacetic acid dipotassium salt dihydrate (EDTA; ≥99.0%),
and molecular properties of the eight SFs led us to propose that
hydrochloric acid (37%), hydrogen peroxide solution (H2O2; ≥35%),
solvation energy was more likely to influence antioxidant hypoxanthine (HPX; ≥99%), iron(II) sulfate heptahydrate (FeSO4·
activity (Figure 5), and the trend was that the higher the 7H2O; ≥99%), methanol (≥99.9%), nitrotetrazolium blue chloride
solvation energy, the better the antioxidant activity. Higher (NBT; 98%), pyrogallol (≥99%), 2,2,6,6-tetramethyl-1-piperidinyloxy
solvation energyless released energy when solvent molecules (TEMPO; ≥99%), p-toluenesulfonic acid monohydrate (≥98%),
bond to the SF surfaceled to less solvation of the SF trans-ferulic acid (≥99%), and xanthine oxidase (XOD) from bovine
molecule due to the number of established bonds and the milk (Grade IV) were obtained from Sigma-Aldrich, St. Louis, MO,
distance/angle between the solvent and the substrate. USA. Stigmasterol was bought from Research Plus, Bayonne, NJ, USA.
Cycloartenyl ferulate (≥99%; SF 1) was purchased from Wako, Osaka,
Furthermore, SF species were available to react with radicals Japan. γ-Oryzanol was obtained from CTC Organics, Atlanta, GA,
in the system. With respect to molecular properties, solvation USA. Methyl ferulate (99%) and ethyl ferulate (99%) were from Alfa
energy was identified as a significantly correlated parameter to Aesar, Ward Hill, MA, USA. Wheat bran was obtained from a
the scavenging properties only in some observations. commercial milling of a mixture of wheat varieties from Swissmill,
Previous studies also reported that different SFs, as well as Switzerland.
compound 9, vary in antioxidant activity. SFs were observed to Preparative HPLC (Merck-Hitachi, Japan) with an XBridge Prep
show less antioxidant activity than compound 9 in some Shield RP C18 column (5.0 μm, 10 × 250 mm, Waters, Ireland),
analytical HPLC (Agilent 1100, Germany) with an XBridge Shield C18
models, e.g., scavenging DPPH•4,22 or ABTS•+,24 which was
column (3.5 μm, 3 × 150 mm, Waters, Ireland), a U-2800 UV/vis
primarily explained by solubility or steric hindrance. However, spectrophotometer (Hitachi, Japan), and a MiniScope MS300
SFs were also found to be better antioxidants than compound 9 benchtop electron spin resonance spectrometer (ESR) (Magnettech,
in scavenging oxygen radical and in a cooked meat model Berlin, Germany) were used for analyses. 1H NMR experiments were
system.25 In some studies, 4-desmethylsteryl ferulates were carried out on a Bruker Avance spectrometer (Bruker BioSpin GmbH,

F DOI: 10.1021/acs.jnatprod.5b00880
J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products Article

Figure 4. Principal component analysis (PCA) of antioxidant activity and molecular properties for the individual SFs (1−8). (A) Bi-plot. (B) Scores
plot of SFs with Hotelling’s T2 ellipse at 5% of confidence. (C) Loadings plot of the variables: (a) antioxidant properties: kDPPH; εDPPH at 1.67, 16.7,
and 60 μM (εDPPH1, εDPPH2, and εDPPH3, respectively); kO2•− at 40, 80, and 160 μM (k O2_1, k O2_2, and k O2_3, respectively); εO2•− at 40,
80, and 160 μM (εO2_1, εO2_2, and εO2_3, respectively); •OH scavenging activity at 1.5, 2.5, 5.0, and 15.0 μM (OH1, OH2, OH3, and OH4,
respectively); (b) molecular properties: area, volume, polarizability, dipole moment, solvation, logP, and molar mass).

G DOI: 10.1021/acs.jnatprod.5b00880
J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products Article

Figure 5. Linear correlation among antioxidant activities (y-axis) and solvation energy (x-axis) of individual SFs (1−8) based on Pearson’s
correlation coefficients (r). The symbol *: the correlation is significant at the 0.05 level (2-tailed); **: significant at the 0.01 level (2-tailed); #:
significant at the 0.05 level (1-tailed).

Rheinstetten, Germany) operating at 400 MHz (1H) and using CDCl3 1, the only commercially available SF standard, was used as an external
as solvent and as internal standard. standard.
Synthesis of SF 7 and SF 8. A new two-step synthetic approach Eight SFs in total were used in this study: SFs 1, 2, 3, 4, 5, 6, 7, and
was designed for the synthesis of high-purity SFs 7 and 8. The first 8. The respective purities, reported as area percentage from the HPLC
step was Steglich esterification using the protected tetrahydropyranyl analysis, were 99%, 96%, 95%, 98%, 98%, 96%, 99%, and 97%, and the
ferulic acid and stigmasterol or cholesterol. The molecules obtained purity of the WB and ORY were 99% and 95%, respectively.
were the corresponding protected SF 7 and 8 derivatives, for which DPPH Radical Scavenging Activity Measurement. The
good yields (85%) were isolated. The second and final step was DPPH• scavenging activity method used in this study was modified
cleavage of the previous protected molecules, carried out in the from Nyström et al.4 The DPPH solution was freshly prepared in
presence of methanol and catalytic amounts of p-toluenesulfonic acid, methanol and was brought to a concentration of 112 μM in the
yielding the target molecules with very good yields (>99%). Both reaction system. The compounds 1−11, ORY, and WB were also
molecules were analyzed by NMR and HPLC-MS to confirm high dissolved in methanol with final concentrations of 1.7, 16.7, and 60.0
purity of the compounds synthesized according to the starting μM, respectively, in the reaction system. Pyrogallol in methanol (final
stigmasterol and cholesterol. concentration 66 μM) was used as the positive control. After mixing
SF 7: 1H NMR (400 Hz, CDCl3, δ) 7.59 (1H, d, ArCH, J = 15.0 the antioxidant and DPPH for 10 s, the absorbance at 517 nm was
Hz), 7.05 (2H, m, Ar), 6.90 (1H, d, Ar, J = 7.8 Hz), 6.27 (1H, d, recorded immediately and subsequently every 15 s for 5 min. Reaction
OCOCH, J = 15.0 Hz), 5.83 (1H, s, OH), 5.40 (1H, t, CH, J = kinetics were analyzed by fitting the absorbance vs time curves with
6.3 Hz), 5.15 (1H, dd, CH, J = 15.2 Hz, J = 8.5 Hz), 5.01 (1H, dd, Origin 9.0.
CH, J = 15.1 Hz, J = 8.5 Hz), 4.74 (1H, m, CHOCO), 3.92 (3H, s, Hydroxyl Radical Scavenging Activity Measurement. The
•
OCH3), 2.39 (2H, d, OCHCH2C, J = 3.1 Hz), 2.1−0.8 (38H, OH scavenging activity method used in this study was a modified
stigmasterol), 0.70 (3H, s, CH3) ppm. version of the methods reported by Cheng et al.29 and Faure et al.23
SF 8: 1H NMR (400 Hz, CDCl3, δ) 7.59 (1H, d, ArCH, J = 15.0 The •OH radical was generated by the Fenton reaction and measured
Hz), 7.05 (2H, m, Ar), 6.90 (1H, d, Ar, J = 7.8 Hz), 6.27 (1H, d, with the spin-trapping technique by ESR. The solutions were added in
OCOCH, J = 15.0 Hz), 5.85 (1H, s, OH), 5.40 (1H, t, CH, J = the following order: 30 μL of 250 mM spin trap DMPO, 30 μL of 1.0
6.3 Hz), 4.74 (1H, m, CHOCO), 3.92 (3H, s, OCH3), 2.39 (2H, d, mM H2O2, 30 μL of 1.0 mM EDTA, 22.5 μL of H2O, 7.5 μL of
OCHCH2C, J = 3.1 Hz), 2.1−0.8 (38H, cholesterol), 0.68 (3H, s, antioxidants in ACN at various concentrations or ACN alone as
CH3) ppm. control, and 30 μL of 1.0 mM FeSO4 to initiate the reaction. The
Extraction and Purification of SFs 2−6. Extraction and solutions of DMPO, H2O2, EDTA, and FeSO4 were prepared in Milli-
purification of SFs 2−6 from ORY and wheat bran was performed Q water. Compounds 1−11, ORY, and WB were each dissolved in
by the method we previously designed and reported.27 First, total ACN at four concentrations, 0.03, 0.05, 0.1, and 0.3 mM, thus leading
lipids from wheat bran were extracted at 50 °C with acetone, then to final concentrations of 1.5, 2.5, 5.0, and 15.0 μM in the respective
subjected to base−acid cleanup in order to eliminate neutral lipids. reaction systems. An aliquot of the reaction mixture was loaded in a 50
Subsequently, the residues from wheat bran and ORY were purified by μL micropipet, and the ESR spectra were recorded. The ESR
preparative HPLC using a UV detector at 325 nm and ACN−H2O− parameters were as follows: B0-field, 3350 G; sweep width, 100 G;
butanol−acetic acid (88:6:4:2, v/v/v/v) as eluent, with a 6.6 mL·min−1 steps, 4096; sweep time, 30 s; number of passes, 2; modulation
flow rate at 25 °C. SF 2 was collected only from ORY. WB and SFs 3 frequency, 1000 mG; microwave attenuation, 10 dB; and receiver gain,
and 4 were collected from the wheat bran. SFs 5 and 6 were collected 900. TEMPO in H2O (2 μM) was used as a daily reference standard
from both ORY and wheat bran. Purity and quantification of SFs were for the ESR instrument. The relative ESR signal was obtained by
also carried out with an analytical HPLC using a UV detector at 325 calculating the ratio of the peak-to-peak amplitude of the second
nm, and MS was used to confirm the high purity of the compounds singlet in the ESR signal of DMPO and the peak-to-peak amplitude of
according to the method previously reported.28 For quantification, SF the first singlet in the ESR signal of the TEMPO. The preparation and

H DOI: 10.1021/acs.jnatprod.5b00880
J. Nat. Prod. XXXX, XXX, XXX−XXX
Journal of Natural Products Article

measurement were operated at room temperature. The percentage of (6) Luo, H. F.; Li, Q.; Yu, S.; Badger, T. M.; Fang, N. J. Nat. Prod.
radical scavenging activity was calculated by RSA% = (h0 − hx)/h0 × 2005, 68, 94−97.
100, where h0 is the ESR signal in the control and hx is the relative ESR (7) Nyström, L., Analysis methods of phytosterols. In Analysis of
signal of the antioxidants. Antioxidant-Rich Phytochemicals; Xu, Z.; Howard, L. R., Eds.; Wiley-
Superoxide Anion Radical Scavenging Activity Measure- Blackwell: Oxford, 2012; Chapter 10, pp 313−351.
ment. The O2•− scavenging activity method applied was modified (8) Aladedunye, F.; Przybylski, R.; Rudzinska, M.; Klensporf-Pawlik,
from Saint-Cricq de Gaulejac et al.30 and Zhou et al.31 The O2•− was D. J. Am. Oil Chem. Soc. 2013, 90, 1101−1109.
generated by the enzymatic hypoxanthine−xanthine oxidase (HPX- (9) Nagy, K.; Actis-Goretta, L.; Redeuil, K.; Barron, D.; Fumeaux, R.;
XOD) system. The molecular probe NBT reacted with O2•−, forming Giuffrida, F.; Cruz-Hernandez, C.; Destaillats, F. J. Chromatogr. A
formazan (NBTH2), which could be detected at 560 nm. This method 2013, 1301, 162−168.
measures the ability of antioxidants to compete with NBT in (10) Bergman, C. J.; Xu, Z. Cereal Chem. 2003, 80, 446−449.
scavenging O2•−. In this method, 2 mM HPX, 0.56 U·mL−1 XOD, (11) Mandak, E.; Nyström, L. J. Agric. Food Chem. 2012, 60, 6123−
and 0.34 mM NBT were prepared in 50 mM phosphate buffer (PBS) 6130.
at pH 7.4. The blank solution contained 300 μL of PBS, 200 μL of (12) Norton, R. A. Cereal Chem. 1994, 71, 111−117.
NBT, and 500 μL of HPX. The solutions required for the reaction (13) Graf, E. Free Radical Biol. Med. 1992, 13, 435−448.
were added in the following order: 200 μL of NBT, 500 μL of HPX, (14) Kochhar, S. P. Eur. J. Lipid Sci. Technol. 2000, 102, 552−559.
50 μL of PBS, 50 μL of antioxidants in DMSO or DMSO alone for (15) Kim, J. S.; Suh, M. H.; Yang, C. B.; Lee, H. G. J. Food Sci. 2003,
control, and 200 μL of XOD to initiate the reaction. Compounds 1− 68, 2423−2429.
11, ORY, and WB were each dissolved in DMSO at concentrations of (16) Juliano, C.; Cossu, M.; Alamanni, M. C.; Piu, L. Int. J. Pharm.
0.8, 1.6, and 3.2 mM, leading to 40, 80, and 160 μM in the final 2005, 299, 146−154.
systems, respectively. The absorbance at 560 nm was recorded every (17) Saenjum, C.; Chaiyasut, C.; Chansakaow, S.; Suttajit, M.;
minute once the reaction started, for a total of 7 min. Reaction kinetics Sirithunyalug, B. J. Med. Plants Res. 2012, 6, 1070−1077.
were analyzed by fitting the absorbance vs time curves with Origin 9.0. (18) Xu, Z.; Hua, N.; Godber, J. S. J. Agric. Food Chem. 2001, 49,
Simulation. Structures were drawn using HyperChem 8.0.3 2077−2081.
software, and the semiempirical quantum mechanics method Austin (19) Huang, C. J. Potential functionality and digestibility of oryzanol
Model 1 (AM1), together with the Polak−Ribiere conjugate gradient as determined using in vitro cell culture models. Ph.D. Thesis,
algorithm with a root-mean-square (RMS) gradient of 0.05 kcal·Å−1· Louisiana State University, Baton Rouge, LA, 2003; pp 38−83.
mol−1, was used for the geometrical optimization of the molecules. (20) Akiyama, Y.; Hori, K.; Hata, K.; Kawane, M.; Kawamura, Y.;
The spin-pairing restricted Hartree−Fock (RHF) operators were used Yoshiki, Y.; Okubo, K. Luminescence 2001, 16, 237−241.
for neutral species, while unrestricted Hartree−Fock (UHF) operators (21) Islam, M. S.; Yoshida, H.; Matsuki, N.; Ono, K.; Nagasaka, R.;
were employed for radicals. The self-consistent field (SCF) Ushio, H.; Guo, Y.; Hiramatsu, T.; Hosoya, T.; Murata, T.; Hori, M.;
convergence limit was set at 10−5, and the accelerated convergence Ozaki, H. J. Pharmacol. Sci. 2009, 111, 328−337.
procedure was used. The dipole moment and some quantitative (22) Kikuzaki, H.; Hisamoto, M.; Hirose, K.; Akiyama, K.; Taniguchi,
structure−activity relationship (QSAR) parameters were obtained H. J. Agric. Food Chem. 2002, 50, 2161−2168.
from the quantum mechanics simulations. The Connolly surface area (23) Faure, A. M.; Andersen, M. L.; Nyström, L. Carbohydr. Polym.
and solvation energy (hydration) were evaluated using the TIP3P 2012, 87, 2160−2168.
water model, where the dimension of the solvent molecule was set to
(24) Winkler-Moser, J. K.; Hwang, H. S.; Bakota, E. L.; Palmquist, D.
1.4 Å radius.
A. Food Chem. 2015, 169, 92−101.
Data Analysis. Principle component analysis was conducted with
(25) Tan, Z.; Shahidi, F. J. Agric. Food Chem. 2011, 59, 12375−
Unscrambler X 10.1 (Oslo, Norway) to obtain a complete view of
12383.
comparisons of individual SFs with their antioxidant activities, as well
(26) Wang, T.; Hicks, K.; Moreau, R. J. Am. Oil Chem. Soc. 2002, 79,
as molecular properties. The relationships of variables were also
1201−1206.
measured using Pearson’s correlation coefficients with IBM SPSS
(27) Mandak, E.; Zhu, D.; Godany, T. A.; Nyström, L. J. Agric. Food
Statistics 19.0 (Armonk, NY, USA).

■
Chem. 2013, 61, 2446−2452.
(28) Zhu, D.; Brambilla, D.; Leroux, J.-C.; Nyström, L. Mol. Nutr.
AUTHOR INFORMATION Food Res. 2015, 59, 1182−1189.
Corresponding Authors (29) Cheng, Z.; Zhou, H.; Yin, J.; Yu, L. J. Agric. Food Chem. 2007,
*Tel: + 41 44 632 5340. Fax: + 41 44 632 1478. E-mail: antoni. 55, 3325−3333.
[email protected]. (30) Saint-Cricq De Gaulejac, N.; Provost, C.; Vivas, N. J. Agric. Food
*Tel: + 41 44 632 9165. Fax: +41 44 632 1123. E-mail:laura. Chem. 1999, 47, 425−431.
[email protected]. (31) Zhou, K.; Su, L.; Yu, L. L. J. Agric. Food Chem. 2004, 52, 6108−
6114.
Notes
The authors declare no competing financial interest.

■ ACKNOWLEDGMENTS
This study was financed by ETH Zurich and Chinese
Scholarship Council (CSC).

■ REFERENCES
(1) Seitz, L. M. J. Agric. Food Chem. 1989, 37, 662−667.
(2) Berger, A.; Rein, D.; Schafer, A.; Monnard, I.; Gremaud, G.;
Lambelet, P.; Bertoli, C. Eur. J. Nutr. 2005, 44, 163−173.
(3) Jun, H.-j.; Lee, J. H.; Cho, B.-R.; Seo, W.-D.; Kang, H.-W.; Kim,
D.-W.; Cho, K.-J.; Lee, S.-J. J. Nat. Prod. 2012, 75, 1706−1711.
(4) Nyström, L.; Mäkinen, M.; Lampi, A. M.; Piironen, V. J. Agric.
Food Chem. 2005, 53, 2503−2510.
(5) Hu, Y.; Xiong, L.; Huang, W.; Cai, H.; Luo, Y.; Zhang, Y.; Lu, B.
Food Funct. 2014, 5, 1150−1159.

I DOI: 10.1021/acs.jnatprod.5b00880
J. Nat. Prod. XXXX, XXX, XXX−XXX