Solvent-Free Enzymatic Synthesis of 1-o-Galloylglycerol Optimized by the Taguchi Method

Article abstract of DOI:10.1002/aocs.12229

Gallic acid (GA) and its lipophilic forms, alkyl gallates, have been widely used in several industrial fields as antioxidants. However, the potential harmful effects of alkyl gallates, such as estrogenic effects, limit their application and raise safety c

Full text of DOI:10.1002/aocs.12229

J Am Oil Chem Soc (2019)
DOI 10.1002/aocs.12229
ORIGINAL ARTICLE
Solvent-Free Enzymatic Synthesis of 1-o-Galloylglycerol
Optimized by the Taguchi Method
Siyu Zhang1 · Casimir C. Akoh1
Received: 21 February 2019 / Revised: 17 April 2019 / Accepted: 24 April 2019
© 2019 AOCS
Abstract Gallic acid (GA) and its lipophilic forms, alkyl
gallates, have been widely used in several industrial fields as
antioxidants. However, the potential harmful effects of alkyl
gallates, such as estrogenic effects, limit their application and
raise safety concerns. The glycerol ester of GA, 1-o-
galloylglycerol (GG), has not been reported to cause adverse
health effects. Owing to the steric and electron-donating
effects of GA, lipase-catalyzed synthesis of GG has not been
successfully achieved. In this work, glycerol ester of GA, GG,
was successfully synthesized for the first time by the enzy-
matic transesterification of glycerol and n-propyl gallate (PG).
GG was synthesized with an immobilized and commercially
available food-grade lipase (Lipozyme® 435) under solvent-
free (no extra solvent) conditions at atmospheric pressure and
nitrogen flow. The effects of the reaction conditions, including
the reaction temperature, substrate molar ratio, reaction time,
and enzyme load, were optimized using the Taguchi method
and regression analysis. The structure of the product was
elucidated using Fourier-transform infrared spectroscopy (FT-
IR), electrospray ionization high-resolution accurate-mass tan-
dem mass spectrometry (ESI-HRAM-MS/MS), and 1D and
2D nuclear magnetic resonance spectroscopy (NMR). Under
the optimal conditions, GG was synthesized in 67.1 ꢀ 1.9%
yield at 50ꢁC, 25:1 (glycerol:PG) substrate molar ratio,
120 hours, and 23.8% enzyme load relative to the total weight
of the substrates.
J Am Oil Chem Soc (2019).
Introduction
Gallic acid (GA) is a natural phenolic compound that is
abundant in various plants, such as tea leaves, gallnuts,
hazelnuts, and grapes (Masoud et al., 2012). It is a phenolic
secondary metabolite that functions as an antioxidant in
plants. Antioxidants are important components of food,
cosmetic, and pharmaceutical products, because they slow
down the oxidation of unsaturated lipids and the associated
generation of unpleasant flavors, thereby preserving prod-
uct quality. The widely used synthetic antioxidants in food
products, such as butylated hydroxyanisole (BHA), butyl-
ated hydroxytoluene (BHT), tert-butyl hydroquinone
(TBHQ), and alkyl esters of GA, are suspected of being
carcinogenic (BHA, BHT, and TBHQ) (Stamatis et al.,
1999) and contact allergenic/estrogenic (alkyl esters of GA)
(Gamboni et al., 2013; Ter Veld et al., 2006). The interest
in using natural antioxidants, such as GA, is increasing
owing to the potential beneficial effects against inflamma-
tory and cardiovascular diseases as well as neuroprotective
effects (Locatelli et al., 2013; Nayeem and Smb, 2016).
However, the application of GA as an antioxidant has been
limited because of its poor solubility in aqueous and nonpo-
lar media.
Keywords 1-o-galloylglycerol ꢂ lipase-catalyzed
alcoholysis ꢂ solvent-free ꢂ Taguchi method
As a hydroxybenzoic acid derivative, the solubility of
GA in both aqueous and nonpolar media is not satisfactory,
which limits its effectiveness in water/oil systems. GA is
basically insoluble in oil and fat, whereas the solubility of
GA in water is highly dependent on the temperature and
requires a long time (4 hours) to reach equilibrium at room
temperature (Daneshfar et al., 2008). Glycosylation and
* Casimir C. Akoh
cakoh@uga.edu
1
Department of Food Science & Technology, University of
Georgia, Athens, GA 30602, USA
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
lipophilization are two methods that have been proposed to
solve the solubility problem of GA in aqueous and oil sys-
tems, respectively (Nam et al., 2017; Otto et al., 2000).
However, their products may not be suitable for use as
common antioxidants in food products. For example,
β-glucogallin, the glycosylation product of GA, is an aldose
reductase inhibitor, which is a medicine for diabetic eye
disease (Puppala et al., 2012). As products of the
lipophilization of GA, alkyl gallates may have an estro-
genic effect by acting as antagonists of estrogen-response
elements; such an effect is associated with the chain length
of the alkyl group (Ter Veld et al., 2006). Moreover,
numerous studies have indicated that alkyl gallates could
cause contact dermatitis and stomatitis (Gamboni et al.,
2013). Furthermore, alkyl gallates are susceptible to hydro-
lysis during digestion, releasing alkyl alcohol in the gastro-
intestinal tract (EFSA Panel on Food Additives and
Nutrient Sources Added to Food, 2014), which could be a
health concern.
The glycerol ester of GA, 1-o-galloylglycerol (GG), could
be a better alternative to both β-glucogallin and alkyl gallates.
As a glycerol monoester, GG is not only readily soluble in
aqueous media, it can also be transformed into a lipophilic
form through well-established chemical or new enzymatic
esterification to esterify one or two of the remaining hydroxyl
groups with fatty acid(s) (Sun et al., 2015). Even without fur-
ther modification, GG could be directly used in vegetable oil
products and has a better antioxidant effect than propyl gallate
(PG) and BHT (Song and Xiao, 1988). GG has not been
reported to have any adverse health effects and would only
release glycerol aside from GA during digestion, which is
much less toxic than alkyl alcohol (released from alkyl gal-
lates during digestion). Furthermore, the gastrointestinal side
effect of GA could be avoided by esterification of the carbox-
ylic acid group (Ravelo et al., 2016). Notably, GG is a natural
component of many plants, such as Rheum rhabarbarum
(Nonaka and Nishioka, 1983). However, the natural occur-
rence of GG is low. An efficient synthesis method for GG is
necessary for its potential application in the food, pharmaceu-
tical, and cosmetic industries.
side reactions and limits the formation of by-products,
which leads to high purity of the reaction products, fewer
reaction intermediates and purification steps, and a more
environmentally friendly and green reaction route
(Figueroa-Espinoza et al., 2013). Although the reaction
time is usually longer, the cost of using enzymes is higher,
and inhibition/deactivation issues related to the reaction
conditions need to be considered, enzymatic reaction is still
the best choice for meeting the stringent requirements that
are necessary for food applications (Buisman et al., 1998).
The enzymatic synthesis of GG is limited due to the ste-
ric hindrance and electrostatic interaction between the
bulky phenyl ring of GA and the active site of the enzyme,
like some lipases (Tamayo et al., 2012). The enzymatic
syntheses of the glycerol esters of benzoic, ferulic, caffeic,
and 3,4-dihydroxyphenylacetic acids, and ibuprofen have
been achieved in recent years (Compton et al., 2000;
Ravelo et al., 2016; Sun and Hu, 2017; Tamayo et al.,
2012). Particularly, the synthesis and application of the
glycerol esters of ferulic acid have been extensively stud-
ied. Not only 1-feruloyl-sn-glycerol, but glyceryl diferulate,
feruloyl monoacylglycerol, feruloyl diacylglycerol, and
glyceryl triferulate have also been successfully synthesized
for use as antioxidants (Compton et al., 2012; Hollande
et al., 2018; Sun et al., 2015), as building blocks for poly-
mer chemistry (Pion et al., 2013), or sunscreen ingredients
(Compton et al., 2000), in ionic liquids, hydrophobic sol-
vents, and under solvent-free conditions. Because of the
structure of GA, its enzymatic esterification by commer-
cially available enzymes was considered unfeasible
(Figueroa-Espinoza et al., 2013; Figueroa-Espinoza and
Villeneuve, 2005; Otto et al., 2000). To our knowledge, the
enzymatic synthesis of GG has not yet been achieved
before. Candida antarctica lipase B (CALB) has a funnel-
like active center with a very short lid, which is similar to
esterases, and is inclined to adapt to more voluminous acids
such as benzoic acid and its derivatives (Tamayo et al.,
2012). Lipozyme® 435 is a commercially available food-
grade immobilized form of CALB (Willett and Akoh,
2018). It has high thermostability and is capable of catalyz-
ing reaction in a solvent or in solvent-free media (Willett
and Akoh, 2018). Based on its properties, Lipozyme®
435 appears to be a good choice for enzymatically synthe-
sizing GG for use in food products.
The synthesis of GG was first achieved by chemical syn-
thesis (Artamonov et al., 1999; Nonaka and Nishioka,
1983; Song and Xiao, 1988). The chemical synthesis of
GG involves multiple protection steps of the hydroxyl
groups in both glycerol and GA, and the yield of GG was
less than 70% (Song and Xiao, 1988). Without protecting
the hydroxyl groups, the yield of GG in the trans-
esterification reaction of methyl gallate and glycerol was
less than 12% (Artamonov et al., 1999). Although they are
simple and rapid, chemical reactions are nonselective and
result in unwanted by-products, in turn leading to low
yields and the need for complex purification steps (Roby,
2017). In contrast, the enzyme selectivity minimizes the
Herein, we report an efficient enzymatic synthesis
method for producing GG catalyzed by a commercially
available food-grade enzyme in solvent-free media under
atmospheric pressure. As depicted in Scheme 1, alcoholysis
(transesterification) of PG with glycerol is proposed as the
route of synthesis. The Taguchi method (Taguchi, 1987)
and regression analysis were used to optimize the synthesis
conditions, including the reaction temperature, substrate
ratio, reaction time, and enzyme load. The structure of the
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
Scheme 1 Synthesis of GG based on the alcoholysis of PG with glycerol
product was determined by Fourier-transform infrared spec-
troscopy (FT-IR), electrospray ionization high-resolution
accurate-mass tandem mass spectrometry (ESI-HRAM-
MS/MS), and 1D and 2D nuclear magnetic resonance spec-
troscopy (NMR).
module (Thermo Fisher Scientific, Waltham, MA, USA)
equipped with aluminum heating blocks and a triangular mag-
netic stirring bar. After PG was dissolved, different percentage
loads of Lipozyme® 435 were added relative to the substrate
weight. Samples (2 μL) were withdrawn every 24 hours then
diluted to 1 mL with methanol for HPLC analysis.
Materials and Methods
Chemicals and Reagents
Reaction Mixture Analysis
Samples from the reaction mixture were withdrawn and
diluted 10-fold with methanol. Sample aliquots were ana-
lyzed by thin layer chromatography (TLC) on 250 μm sil-
ica gel G (Uniplate, Analtech, Newark, DE, USA).
Separation was achieved using methanol/chloroform (1:3,
v/v) as the mobile phase. The TLC plates were directly
visualized under 254 nm ultraviolet (UV) light for prepara-
tive purposes or after spraying with 0.2% 20,70-
dichlorofluorescein dissolved in methanol (w/w). The band
corresponding to GG on the preparative plates was recov-
ered and extracted with methanol. The solvent was then
evaporated under a nitrogen flow and the product was fur-
ther analyzed using FT-IR and NMR spectroscopy. The Rf
values were: glycerol, 0.54; GA, 0.77; GG, 0.66; and
PG, 0.96.
The reaction was monitored by HPLC on an Agilent1260
Infinity system (Agilent Technologies, Santa Clara, CA,
USA), equipped with a diode-array detector (DAD) scanned
at 280 nm and a Sedex 85 low-temperature evaporative light-
scattering detector (LT-ELSD) operated at 3.3 bar and 35ꢁC.
A reverse phase C18 column (Ultrasphere ODS, 5 μm,
250 × 4.6 mm) was used at a controlled temperature of 35ꢁC
and. DAD was used to quantify PG, GA, and GG, whereas
LT-ELSD was used to monitor glycerol. Considering that the
UV–Vis spectra of PG, GA, and GG were identical, the com-
pounds were quantified using the area normalization method.
The separation was achieved by injecting a 20 μL sample at a
flow rate of 0.7 mL min−1 using a binary eluent as the mobile
phase, which consisted of solvent A/water with 1% acetic acid
(v/v), and solvent B/methanol. The elution sequence consisted
of a linear gradient from 90% (v/v) A to 0% A (v/v) in
30 min, followed by an isocratic elution of solvent B for
10 min. Throughout the whole sequence, the UV spectra
n-Propyl gallate (≥98%) was purchased from HiMedia Lab-
oratories (Nashik, India). Glycerol was purchased from
Hoefer Inc. (San Francisco, CA, US). Lipozyme® RM IM
(lipase from Rhizomucor miehei, immobilized on an
anionic exchange resin, with specific activity of 442 inter-
esterification unit g−1), Lipozyme® TL IM (lipase from
Thermomyces lanuginosus and immobilized on
a
noncompressible silica gel carrier, with specific activity of
250 interesterification unit g−1), Novozym® 435 (lipase B
from Candida antarctica, immobilized on a macroporous
acrylic resin, with specific activity of 10,000 propyl laurate
unit g−1), and Lipozyme® 435, a food-grade lipase (recom-
binant lipase B from C. antarctica, expressed in Aspergil-
lus niger, and immobilized on a macroporous hydrophobic
resin, with specific activity of 8000 propyl laurate unit g−1
)
were purchased from Novozymes North America, Inc.
(Franklinton, NC, USA). Nitrogen (Ultra High Purity 5.0
Grade) was purchased from the local branch of Airgas, Inc.
(Athens, GA, USA). All solvents were of high performance
liquid chromatography (HPLC) grade and purchased from
local chemical suppliers. All the chemicals were used with-
out further purification.
Enzymatic Transesterification
The transesterification of PG and glycerol was performed in
10 mL reaction vials with an internal cone. After 1 mmol of
PG was added into the reaction vials, various amounts of
glycerol were added. The reaction mixtures were heated at
various temperatures and stirred at 200 rpm under a constant
nitrogen flow using a Reacti-Therm™ heating and stirring
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
Fig. 1 HPLC results of the reaction mixtures. Reaction conditions: 50ꢁC, glycerol/PG = 25:1 (mol/mol), enzyme load 25% (relative to the total
weight of the substrates). (a) HPLC-ELSD analysis of the reaction mixture at 0 hour, (b) HPLC-DAD analysis of the reaction mixture at 0 hour
with detection at 280 nm, and (c) HPLC-DAD analysis of the reaction mixture at 120 hours with detection at 280 nm
(from 190 to 400 nm) were recorded for analysis of the peak
purity. Under these conditions, the retention times were: glyc-
erol, 4.0 min; GA, 6.4 min; GG, 8.0 min; and PG, 17.6 min
(Fig. 1). The yield of GG, the conversion of PG, and the hydro-
lysis rate were calculated according to the following equations:
ESI-HRAM-MS/MS Analysis
Eluted HPLC-DAD peaks were collected and dried under a
nitrogen atmosphere. For ESI-HRAM MS/MS analysis, the
sample was dissolved in methanol. Analysis was performed
by direct infusion in a Thermo Orbitrap Elite mass spec-
trometer (Thermo Fisher Scientific) with an ESI interface
using the negative-ion mode. For MS, the instrument was
set to scan from 100 to 1000 mass-to-charge (m/z). For tan-
dem MS analysis, collision-induced dissociation was cho-
sen as the fragmentation method, Fourier-transform mass
spectrometry spectra were collected from 65 to 500 m/z.
The capillary temperature was set at 200ꢁC, the source volt-
age was −2.99 kV, and the vaporizer temperature was
223.80ꢁC.
Moles of GG formed
Moles of PG at t = 0 hour
Yield ð%Þ =
ð1Þ
Moles of PG consumed
Moles of PGat t = 0hour
Conversion ð%Þ =
Hydrolysis ð%Þ =
ð2Þ
Moles of GA formed
Moles of PG at t = 0 hour
ð3Þ
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
1
FT-IR and NMR Analyses
13C spectrum width ranged from 0 to 160 ppm, and the H
spectrum width ranged from 0 to 9 ppm with 128 t1 incre-
ments and 16 scans per increment. For gHMBC, the 13C
spectrum width ranged from 0 to 190 ppm, and the 1H spec-
trum width ranged from 0 to 9 ppm with 256 t1 increments
The FT-IR spectra were collected using a Thermo Nicolet
Nexus FT-IR 1100 spectrometer (Thermo Fisher Scientific)
equipped with an ZnSe attenuated total reflection attachment
(νmax is reported in cm−1). After 10 mg of the samples were
dissolved in 1 mL methanol, a drop of the solution was
placed onto the ZnSe crystal. Spectra were collected after
evaporation of the solvent. Spectral data were collected from
500 to 4000 cm−1 with a resolution of 4 cm−1 and 32 scans.
The data were processed using Omnilab software.
1
and 56 scans per increment. The H spectra of gHSQC and
gHMBC were both acquired at 300 MHz. The 13C spectra of
gHSQC and gHMBC were acquired at 75 MHz. The chemi-
cal shifts of GG were assigned based on the information
1
from both the 1D and 2D NMR spectra (Fig. 3). H NMR
(300 MHz, D2O) δ: 7.04 (2H, s, H8.), 4.27–4.14 (2H, m,
For NMR spectroscopy analysis, 10 mg of the sample col-
lected by TLC or HPLC was dissolved in 1 mL deuterium
H1), 4.12–4.03 (1H, m, H2), and 3.99–3.70 (2H, m, H4), 13
C
NMR (75 MHz, D2O) δ: 65.72 (C1), 69.54 (C2), 62.48 (C4),
120.41 (C7), 109.90 (C8), 144.58 (C9), 138.32 (C11), and
168.12 (C6).
1
oxide. H, 13C, gradient correlation spectroscopy (gCOSY),
1H–13C gradient heteronuclear single-quantum coherence
1
(gHSQC), and H–13C gradient heteronuclear multiple bond
correlation (gHMBC) spectra were recorded at 25ꢁC using a
Varian Mercury-300 (MHz) NMR system (Agilent Technol-
ogies) equipped with a 5 mm broad band room temperature
probe, and chemical shifts were reported in parts per million
(δ/ppm). The chemical shifts were referenced to the residual
peak for HDO (δH 4.79 ppm) and the acetic acid residual
peak (δH/δC 2.00/20.2, 175.21 ppm). For 1H NMR analysis,
the scan range was set to vary from 0 to 9 ppm with 16 scans.
The 13C spectra were obtained indirectly from the gHSQC
and gHMBC spectra. For gCOSY, the HDO peak was
purged, and the scan range was from 0 to 9 ppm with 800 t1
increments and 4 scans per t1 increment. For gHSQC, the
Experimental Design and Statistical Analysis
To achieve the highest yield of GG, the effects of four reac-
tion variables, i.e., the reaction temperature (T, ꢁC), substrate
ratio (glycerol/PG, mol/mol), enzyme load (percentage of
enzyme relative to the total substrate weight), and reaction
time (t, hour), were analyzed and optimized using the
Taguchi method. An L18 (34) orthogonal array (Table 1) was
adapted from the standard Taguchi method (Taguchi, 1987).
As a robust approach for fractional factorial experimental
design, the Taguchi method provides an efficient algorithm
to find the optimum design for engineering, which allows
Table 1 Experimental design and the effects of the reaction temperature, substrate ratio, reaction time, and enzyme load on the yield of GG,
conversion of PG, and hydrolysis
Trials
T (ꢁC)
Substrate ratio
t (hours)
Enzyme load (%)
Yield (%)
Conversion (%)
Hydrolysis (%)
1
50
50
50
65
65
65
80
80
80
50
50
50
65
65
65
80
80
80
5
15
25
5
24
120
72
5
15
25
25
5
1.7 ꢀ 1.4
41.6 ꢀ 1.6
47.2 ꢀ 1.0
1.8 ꢀ 0.1
9.2 ꢀ 2.2
17.1 ꢀ 2.0
1.2 ꢀ 0.1
13.8 ꢀ 1.6
16.7 ꢀ 2.6
5.4 ꢀ 2.6
29.6 ꢀ 3.1
19.9 ꢀ 0.5
2.0 ꢀ 0.6
11.4 ꢀ 3.4
29.7 ꢀ 2.2
1.1 ꢀ 0.1
14.1 ꢀ 3.5
5.6 ꢀ 1.6
5.4 ꢀ 2.1
46.2 ꢀ 2.0
53.7 ꢀ 0.9
4.5 ꢀ 0.4
10.8 ꢀ 2.4
19.9 ꢀ 2.5
5.3 ꢀ 0.3
18.2 ꢀ 2.8
17.5 ꢀ 2.0
10.3 ꢀ 3.3
34.5 ꢀ 3.7
23.8 ꢀ 1.2
4.6 ꢀ 1.6
14.2 ꢀ 4.3
34.7 ꢀ 2.7
5.1 ꢀ 0.1
17.7 ꢀ 3.6
6.8 ꢀ 1.8
2.8 ꢀ 0.5
3.7 ꢀ 0.2
4.8 ꢀ 0.4
2.0 ꢀ 0.8
1.4 ꢀ 0.5
1.6 ꢀ 0.4
3.1 ꢀ 0.2
2.7 ꢀ 0.3
1.1 ꢀ 0.1
4.0 ꢀ 0.9
3.9 ꢀ 0.7
2.0 ꢀ 0.2
1.5 ꢀ 0.3
2.0 ꢀ 0.8
3.9 ꢀ 0.7
2.8 ꢀ 0.1
2.1 ꢀ 0.2
0.8 ꢀ 0.1
2
3
4
120
72
5
15
25
5
6
24
15
15
25
5
7
72
8
15
25
5
24
9
120
120
72
10
11
12
13
14
15
16
17
18
5
15
25
5
15
25
25
5
24
72
15
25
5
120
72
15
15
25
5
24
15
25
120
72
All the results are average of triplicate reactions ꢀ SD.
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
the reduction of the number of experiments required to opti-
mize a process (Mohan et al., 2009). The reaction temper-
ature (50, 65, and 80ꢁC), substrate ratio (5, 15, and 25),
enzyme load (5%, 15%, and 25%), and reaction time
(24, 72, 120 hours) were selected based on the reports in
the literature involving the use of CALB to catalyze
the esterification/transesterification of phenolic acids
(Nam et al., 2017; Sun and Hu, 2017). As shown in
Table 1, 18 reaction variables were tested (each in tripli-
cate), and each variable was evaluated using an orthogo-
nal design.
Results and Discussion
Structural Characterization of GG
The alcoholysis of PG with glycerol (Scheme 1) was per-
formed in an open system under a nitrogen flow in a
solvent-free system, catalyzed with an immobilized and
commercially available food-grade lipase (Lipozyme®
435). The HPLC and spectra of the reaction mixture
detected by DAD and/or ELSD at the start and at the end
of reaction are shown in Fig. 1. During the course of the
reaction, a decrease in the PG and the formation of GG
were observed. The same UV absorption peak was
observed at 275 nm for PG, GG, and GA, which agreed
with the previous data (Artamonov et al., 1999). Notably,
there was a small peak with a retention time of 16.2 min
that also had the same UV absorption pattern (Fig. 1c),
indicating that it may have been glyceryl digallate.
The GG product was further characterized using FT-IR,
ESI-HRAM-MS/MS, and NMR spectroscopy. A broad
band within 3100–3700 cm−1 region typical of the O–H
stretching mode of the hydroxy was observed in the FT-IR
spectrum (spectra not shown). This was ascribed to the
unesterified hydroxyl groups of the glycerol moiety and
hydroxyl group in the gallate part of the molecule follow-
ing formation of the ester function. The peak at 1699 cm−1
indicated the existence of the ester linkage in GG, which
shifted from the typical signal at 1740 cm−1 owing to the
influence of the aromatic ring. A sharp signal at 1610 cm−1
indicated the existence of an aromatic structure bearing
hydroxyl groups in the ortho positions, which was consis-
tent with the arrangement of the hydroxyl groups in
GG. The presence of a pair of sharp bands at 1208 and
1137 cm−1 indicated the asymmetrical C–O of the ester
linkage.
ESI-HRAM-MS/MS analysis was used to determine the
elemental composition and structure of the new compounds
(Fig. 2). In MS1, the molecular ion [M-H]−1 peak found at
243.0508 m/z was consistent with C10H1407 as the chemical
formula for the molecular ion (calculated 243.0510 m/z,
error = −0.82 ppm). The tandem ESI-HRAM-MS data
(generated in the negative-ion mode) showed an [M-H]−1
peak at 243.0510 m/z (calculated 243.0510 m/z,
error = 0 ppm). In MS2, the most abundant fragment gener-
ated from the molecular ion was 169.0144 m/z, which was
attributed to a GA molecule with one hydrogen atom lost
(calculated 169.0142 m/z, error = 1.18 ppm), confirming
bond formation between GA and glycerol. Other less abun-
dant peaks could also be found from GG by the loss of log-
ical neutral species (Fig. 2). As shown in Fig. 2, the MS2
results confirmed the structure of the product and demon-
strated that the final compound generated by enzymatic
esterification was GG. However, the ESI-HRAM-MS/MS
Statistical analysis of the experimental results was per-
formed using JMP® software (version 13.2.0, SAS Insti-
tute, Inc., Cary, NC, USA). The results were expressed as
mean values ꢀ SD. To determine the optimal reaction
conditions and evaluate the effects of the reaction vari-
ables, the yield of GG and the signal-to-noise ratio (S/N,
dB) were analyzed by ANOVA with 2-way interactions.
The results were expressed as least-squares means
(LS mean) ꢀ SEM. Tukey’s honest significant difference
test (Tukey HSD) and the LSD test were used to deter-
mine the differences that occurred when varying each
reaction parameter and the level of significance
(P < 0.05). The S/N ratio is a measure of the robustness
that can be used to detect the control factor settings for a
minimum impact of noise on a response (Willett and
Akoh, 2018). The larger the S/N is, the higher the yield,
as shown in Eq. 4.
!
n
X
1
S=NðdBÞ = −10log
yi−2
ð4Þ
n i = 1
in which yi represents the experimental value of the ith
experiment and n represents the number of replications.
The experimental results were also analyzed by regres-
sion analysis using a polynomial quadratic equation to cor-
relate each response variable with the independent
variables, using Eq. 5:
4
4
4
4
X
X
XX
Y = β0 +
βiXi +
βiiX2i +
βijXiXj
ð5Þ
i = 1
i = 1
i = 1 j = 1
in which Y is the predicted response, Xi and Xj are indepen-
dent factors, β0 is the intercept, βi is the linear coefficient,
βii is the quadratic coefficient, and βij is the interaction
coefficient.
To select the model with the best fit, the forward step-
wise minimum Bayesian information criterion (BIC) para-
meter/model selection method was used. The analyses were
performed using three-dimensional diagrams constructed
for the selected model. The ANOVA test, level of signifi-
cance (P < 0.05), and statistically fitting (R2) were used to
evaluate the model. Time-course reactions were also used
for model validation.
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
Fig. 2 ESI-HRAM-MS/MS analysis of GG
results could not confirm the position of the ester function
on the glycerol moiety of GG.
Results showed that both C1 and C4 had two protons
attached directly, while only one proton was directly
attached to C2. Furthermore, C1 and C4 showed different
chemical shifts, which indicates the ester linkage was
formed at the primary hydroxyl group of the glycerol moi-
ety. Assuming the ester linkage had formed at the C2
hydroxy, the chemical shifts of C1 and C4 would be the
same. The gHMBC experiments were used to investigate
the long-range coupling between carbon atoms and proton
atoms. Fig. 3d shows the results acquired by gHMBC
experiment configured for an 8-Hz coupling. The long-
range correlations acquired from this experiment, particu-
larly between the carbonyl group and H1, were consistent
with the ester linkage of GG formed at the primary
hydroxyl group of the glycerol moiety. Regarding the 1D
NMR and 2D NMR data of the product, our data indicated
that when using Lipozyme® 435 as a biocatalyst, GA was
esterified only on the primary hydroxyl group of glycerol.
The results of the 1H- and 13C-unidimensional NMR
together with 2D gCOSY, gHSQC, and HMBC NMR pro-
vided evidence that the primary hydroxyl group of glycerol
was esterified and formed a linkage with the GA moiety.
The 1H spectrum (Fig. 3a) showed that five hydrogen
atoms from the glycerol moiety had three different chemi-
cal shifts, which indicated that the formed ester had an
asymmetrical structure. Furthermore, only two hydrogen
atoms were attached to the benzene ring, and these two
1
hydrogens had the same chemical shift in the H NMR
spectrum, which confirmed the structure of the esterified
GA moiety. The results of gCOSY NMR unveiled the con-
nectivity of the hydrogen atoms, providing additional evi-
dence of the hydrogen arrangement of GG (Fig. 3b). The
13C NMR results indicated that the three carbon atoms of
the glycerol moiety had three different chemical shifts,
which provided further evidence that the formed ester had
an asymmetrical structure. The gHSQC spectra (Fig. 3c)
displayed the one-bond correlation of the carbon atoms and
hydrogen atoms in the GG molecule. Combined with the
information obtained by 1H NMR spectroscopy, the carbon
atoms of the glycerol moiety were of particular interest.
Taguchi Method
The mean values of the yield, conversion, and hydrolysis
were calculated according to Eqs. 1–3, respectively. The
S/N ratios were calculated using the method of “larger is
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
Fig. 3 NMR spectra and chemical shift assignment of GG: (a) 1H NMR spectrum, (b) gCOSY spectrum, (c) gHSQC spectrum, and (d) gHMBC
spectrum
the better” in Eq. 4. For all 18 experimental combinations
described in Table 1, each condition was repeated trice.
Because the experimental design was orthogonal, all speci-
fied parameters could be estimated independent of any
other. Therefore, the effects of each factor were separated
out in terms of the S/N together with the mean response,
reported in LS means ꢀ SEM (Fig. 4).
quadratic trendlines with R-values equal to 1, suggesting
the effect of temperature could be fitted into a quadratic
model. To investigate the yield at lower temperature, time-
course reactions were performed at 35, 40, 45, and 50ꢁC,
with other conditions optimized (enzyme load of 23.8%,
substrate ratio of 25). As shown in Fig. 5a, lower conver-
sions were obtained at 35, 40, and 45ꢁC after reaction for
120 hours, which was compatible with the Arrhenius law.
An increased yield could also result from a decrease in the
viscosity in the reaction medium and, consequently, from
the reduction of mass transfer resistances. Considering the
above data, 50ꢁC was selected as the optimum reaction
temperature.
Effect of Reaction Temperature
As shown in Fig. 4a, both the yield and S/N were the
highest at 50ꢁC, whereas the yield at 60ꢁC was significantly
higher than the yield at 80ꢁC, which indicated that the yield
of GG decreased with increasing temperature from 50 to
80ꢁC. The highest LS mean for the yield (31.3 ꢀ 1.0%)
and S/N (29.7 ꢀ 1.7 dB) were obtained at 50ꢁC. The
results could be explained by partial thermal deactivation
of the enzyme at the higher temperatures. In addition, the
yield of GG and S/N in this range of temperature tested fits
Effect of the Substrate Ratio
The highest LS mean yield (31.8 ꢀ 1.7%) and S/N
(30.3 ꢀ 1.0 dB) were obtained at a 25:1 (glycerol/PG,
mol/mol) substrate ratio. As shown in Fig. 4b, when the
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
Fig. 4 Effects of the (a) reaction temperature on yield, (b) substrate ratio on yield, (c) enzyme load on yield, and (d) reaction time on yield.
Different letters indicate significant difference at P < 0.05 with Tukey’s HSD and LSD tests
substrate ratio of glycerol to PG reduced from 25:1 to 15:1
or 5:1, both the yield of GG and S/N significantly decreased.
These results showed that increasing the proportion of glyc-
erol improved the conversion rate of PG, which could be
accredited to the excessive amount of glycerol serving as a
reactant and reaction medium for the reaction. Notably,
hydrolysis increased when increasing the substrate ratio (data
not shown). Therefore, 25:1 glycerol/PG (mol/mol) was cho-
sen as the optimum reaction substrate ratio. Similar to the
effect of temperature, the GG yield and S/N in the range of
substrate ratios tested fit quadratic trendlines with R-values
equal to 1, suggesting that the effect of varying the substrate
ratio could be fitted into a quadratic model.
and 26.1 ꢀ 1.4 dB, respectively. As shown in Fig. 4c,
when the enzyme load decreased from 25% to 5%, both the
yield of GG and S/N significantly decreased. Although the
LS mean yield and S/N were significantly higher at 25%,
the difference between 15% enzyme load and 25% enzyme
load was not significant. Furthermore, similar to the effect
of the temperature/substrate ratio, the GG yield and S/N
ratio in the tested range of enzyme load fit quadratic
trendlines with R-values equal to 1, suggesting that the
effect of the enzyme load could be fitted into a quadratic
model. However, the trendlines of yield and S/N suggest
that a point exists between a 25% and 15% enzyme load
with a higher yield and S/N ratio. Thus, an optimal enzyme
load of 23.8% was predicted by regression analysis. The
three levels of enzyme loads (15%, 23.8%, and 25%) were
tested in time-course reactions in triplicate (with the other
conditions optimized) to determine the optimal enzyme
load (Fig. 5b). The results showed that the highest yield
Effect of Enzyme Load
The highest LS mean of yield and S/N were obtained at
25% enzyme load of total substrates (w/w), 22.7 ꢀ 1.2%
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
Fig. 5 Time-course reactions under different conditions: (a) low temperature (fixed-reaction conditions: glycerol/PG = 25:1, and enzyme load
23.8%), (b) three levels of enzyme load (fixed-reaction conditions: 50ꢁC, glycerol/PG = 25:1), (c) validation of the proposed model (the reaction
conditions listed are the reaction temperature, substrate ratio, and enzyme load, respectively), and (d) under the optimal conditions (50ꢁC,
glycerol/PG = 25:1, enzyme load 23.8%)
was achieved with an enzyme load of 23.8%, which was
selected as the optimal enzyme load. The decreased yield
and S/N ratio at a high enzyme concentration could be
explained by enzyme aggregation (Sun and Hu, 2017).
time. Similar to the effect of temperature, the yield of GG
and S/N in the tested range of reaction time fits quadratic
trendlines with R-values equal to 1, suggesting that the
effect of the reaction time could be fitted to a quadratic
model.
Effect of Reaction Time
Regression Analysis and Determination of Optimal
Conditions
The highest LS mean yield (24.4 ꢀ 1.0%) and S/N
(24.9 ꢀ 1.8 dB) were obtained at a 120-hour reaction time.
As shown in Fig. 4d, when the reaction time decreased
from 120 to 72 hours or 24 hours, both the yield of GG and
S/N significantly decreased. Considering that hydrolysis
increased with an increasing reaction time (data not
shown), and to prevent the conversion of monoesters to
diesters, 120 hours was chosen as the optimum reaction
Based on the characteristics of the effects of four factors, a
polynomial quadratic model with two-way interactions was
selected, as described in Eq. 5. The coefficients of the pro-
posed model were calculated based on the experimental
responses. The fitted model that is expressed in coded vari-
ables is represented by Eq. 6:
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
Table 2 ANOVA results observed when fitting the experimental data
to the proposed model
statistically significant when using the F-test at a 99.99%
probability level, which suggested that the variation
accounted for by the model was significantly greater than
the residual variation. There was less than a 0.01% proba-
bility that the model F-value could occur owing to noise.
Likewise, the coefficient (R2) of the regression model indi-
cated that over 96.7% of the variability in the response
could be explained by the predicted second-order polyno-
mial equation given above. An accuracy check is necessary
to generate an adequate model. In this respect, the validity
of the model was established by comparing the experimen-
tal results obtained with three time-course reactions (each
reaction performed in triplicate) with the predicted values
(Fig. 5c). The actual yields fell within the 95% confidence
intervals of the predicted yields. These results confirmed
the validity of the model.
The mutual effects of the reaction temperature, substrates
ratio, reaction time, and enzyme load on the yield of GG were
also illustrated in the three-dimensional diagrams (Fig. 6).
The effects of reaction conditions shown in these figures are
consistent with the results obtained from the Taguchi method.
Combining the results of the three-dimensional diagrams with
the results obtained in the previous sections, the optimal
Source
Degree of Sum of
Mean
F-value P-value
freedom
squares
square
Model
Residual
Total
10
43
53
9830.10 983.01
126.29
<0.0001*
334.70
7.78
10,164.81
Root mean square error (RMSE) = 2.7899, R2 = 0.9671
*Statistically significant at 99.99% of confidence level.
Yield ð%Þ = −12:3628−0:0247X1 + 0:2184X2
+ 0:4272X3 + 2:0673X4 −0:0279X22 −0:0007X32
−0:0635X24 −0:0070X1X3 + 0:0182X2X3
+ 0:0383X2X4
ð6Þ
in which X1 is the reaction temperature (ꢁC), X2 is the sub-
strate ratio (glycerol/PG, mol/mol), X3 is the reaction time
(hours), and X4 is the enzyme load (%).
The quality of the fitted model was confirmed by
ANOVA, as demonstrated in Table 2. As shown in
Table 2, the sum of squares related to the regression was
Fig. 6 Three-dimensional diagrams showing the effects of two parameters (as indicated) on the GG yield. For each plot, the other two factors
were fixed under the optimal conditions. The optimal reaction conditions were a temperature of 50ꢁC, a glycerol/PG = 25:1 (mol/mol), an
enzyme load of 23.8% (relative to the total weight of the substrates), and a reaction time of 120 hours
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
Table 3 Alcoholysis of PG and glycerol catalyzed by different
lipases
nonaqueous media, the use of hydrophobic solvents gener-
ally results in a higher product yield and conversion
(Buisman et al., 1998; Figueroa-Espinoza and Villeneuve,
2005). However, the solubilities of PG and glycerol are
rather low in apolar solvents. t-Butanol has the highest log
P value (0.839) among the solvents that could sufficiently
dissolve the substrates (PG and glycerol). After 120 hours
of reaction, less than 1.0% product yield was obtained
(Table 3). This could be attributed to enzyme deactivation
by t-butanol. Similar results were also found in a previous
study, where higher yields were observed in a solvent-free
system when using CALB as a catalyst for the esterification
of cinnamic acid derivatives and aliphatic alcohols
(Stamatis et al., 1999).
Time (hours)
Yield (%)
Lipozyme® Lipozyme® Novozym® Lipozyme®
RM IMa
TL IMa
435a
435 in t-butanol
24
48
72
96
120
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
23.7 ꢀ 3.7
42.2 ꢀ 1.5
57.2 ꢀ 5.2
72.5 ꢀ 5.9
75.0 ꢀ 2.5
0.2 ꢀ 0.1
0.3 ꢀ 0.1
0.4 ꢀ 0.1
0.5 ꢀ 0.1
0.5 ꢀ 0.1
All results are average of triplicate reactions ꢀ SD.
ND, not detected.
a Reactions were carried out without using extra solvent.
conditions were determined to be: 23.8% enzyme load rela-
tive to the total substrates (w/w), substrate ratio
glycerol/PG = 25:1 (mol/mol), a reaction temperature of
50ꢁC, and a reaction time of 120 hours. The actual yield of
GG, conversion of PG, and the hydrolysis rate were con-
firmed by a time-course reaction in triplicate under the opti-
mal conditions. The results are illustrated in Fig. 5d. After a
reaction time of 120 hours, the yield of GG was
67.1 ꢀ 1.9%, the conversion of PG was 73.6 ꢀ 1.5%, and
the hydrolysis rate was 5.1 ꢀ 1.3%. According to the model,
the predicted optimal yield was 59.9–69.2%, which agreed
with the actual yield. The good agreement between the actual
and predicted values validated the model.
Conclusion
A novel method for synthesizing GG by the enzymatic
transesterification of PG and glycerol was established. GG
was synthesized using an immobilized commercially avail-
able food-grade lipase, Lipozyme® 435, in a solvent-free
system at atmospheric pressure under a nitrogen flow. High
PG conversion and GG yield with low hydrolysis were
obtained under the optimized conditions. Compared with
chemical synthesis of GG, the enzymatic strategy does not
require the presence of a strong acid or base as the catalyst
(Artamonov et al., 1999; Nonaka and Nishioka, 1983; Song
and Xiao, 1988). In this study, we confirmed that the enzy-
matic preparation of galloylated structured lipids was feasi-
ble, and that a green and efficient enzymatic method for the
production of GG was developed. This product (GG) may
have potential for use as an antioxidant and an active ingre-
dient in the food, cosmetic, and pharmaceutical fields.
Other Lipases and Solvent Medium
Lipozyme® RM IM, Lipozyme® TL IM, and Novozym®
435 were used separately as catalysts under the same reac-
tion conditions determined in the previous section. As shown
in Table 3, the highest yield was achieved using Novozym®
435 as the catalyst, while no GG formation was observed
when using Lipozyme® RM IM or Lipozyme® TL IM, indi-
cating that these two lipases are not suitable for catalyzing
this reaction. Previous studies with phenolic acids/esters as
substrates showed that only CALB could catalyze the reac-
tion with alcohols in moderate to good yields (Buisman
et al., 1998), which is in agreement with the results obtained
in this work. Although the yield achieved using Novozym®
435 (75.0 ꢀ 2.5%) was higher than the one obtained using
Lipozyme® 435 (67.1 ꢀ 1.9%), it should be noted that
Novozym® 435 is not a food-grade enzyme. Therefore,
Lipozyme® 435 would be a better choice to synthesize GG
for applications in food, cosmetic, and pharmaceutical fields.
To check the potential influence of extra solvent in
the medium, the reaction was also carried out in 10 mL t-
butanol using the values previously optimized for the rest
of the variables. When using CALB (Lipozyme® 435) in
Acknowledgments The authors would like to thank Food Science
Research, the University of Georgia for supporting this work. Special
thanks go to Dr. Dennis R. Phillips and Dr. Chau-Wen Chou from the
UGA PAMS facility for assistance with MS analysis, and Dr. John
Glushka from the CCRC NMR spectroscopy facility of UGA for
assisting NMR analysis.
Conflict of Interest The authors declare that they have no conflicts
of interest.
References
Artamonov, A., Nigmatullina, F., Aldabergenova, M.,
&
Dzhiembaev, B. Z. (1999) Synthesis of α-monoglycerides of aro-
matic acids. Chemistry of Natural Compounds, 35:404–408.
Buisman, G., Van Helteren, C., Kramer, G., Veldsink, J.,
Derksen, J., & Cuperus, F. (1998) Enzymatic esterifications of
functionalized phenols for the synthesis of lipophilic antioxidants.
Biotechnology Letters, 20:131–136.
J Am Oil Chem Soc (2019)

J Am Oil Chem Soc
Compton, D. L., Laszlo, J. A., & Berhow, M. A. (2000) Lipase-
catalyzed synthesis of ferulate esters. Journal of the American Oil
Chemists’ Society, 77:513–519.
Compton, D. L., Laszlo, J. A., & Evans, K. O. (2012) Antioxidant
properties of feruloyl glycerol derivatives. Industrial Crops and
Products, 36:217–221.
Daneshfar, A., Ghaziaskar, H. S., & Homayoun, N. (2008) Solubility
of gallic acid in methanol, ethanol, water, and ethyl acetate. Journal
of Chemical & Engineering Data, 53:776–778.
Otto, R. T., Scheib, H., Bornscheuer, U. T., Pleiss, J., Syldatk, C., &
Schmid, R. D. (2000) Substrate specificity of lipase B from Can-
dida antarctica in the synthesis of arylaliphatic glycolipids. Journal
of Molecular Catalysis B: Enzymatic, 8:201–211.
Pion, F., Reano, A. F., Ducrot, P. H., & Allais, F. (2013) Chemo-
enzymatic preparation of new bio-based bis-and trisphenols: New
versatile building blocks for polymer chemistry. RSC Advances, 3:
8988–8997.
Puppala, M., Ponder, J., Suryanarayana, P., Reddy, G. B.,
Petrash, J. M., & LaBarbera, D. V. (2012) The isolation and charac-
terization of β-glucogallin as a novel aldose reductase inhibitor
from Emblica officinalis. PLoS One, 7:e31399.
Ravelo, M., Esteban, J., Ladero, M., & García-Ochoa, F. (2016) Enzy-
matic synthesis of ibuprofen monoglycerides catalyzed by free
EFSA Panel on Food Additives and Nutrient Sources Added to Food
(ANS). (2014) Scientific opinion on the re-evaluation of propyl gal-
late (E 310) as a food additive. EFSA Journal, 12:3642–3688.
Figueroa-Espinoza, M. C., Laguerre, M., Villeneuve, P.,
&
Lecomte, J. (2013) From phenolics to phenolipids: Optimizing anti-
oxidants in lipid dispersions. Lipid Technology, 25:131–134.
Figueroa-Espinoza, M. C., & Villeneuve, P. (2005) Phenolic acids
enzymatic lipophilization. Journal of Agricultural and Food Chem-
istry, 53:2779–2787.
Gamboni, S. E., Palmer, A. M., & Nixon, R. L. (2013) Allergic con-
tact stomatitis to dodecyl gallate? A review of the relevance of posi-
tive patch test results to gallates. Australasian Journal of
Dermatology, 54:213–217.
Hollande, L., Domenek, S., & Allais, F. (2018) Chemo-enzymatic
synthesis of renewable sterically-hindered phenolic antioxidants
with tunable polarity from lignocellulose and vegetal oil compo-
nents. International Journal of Molecular Sciences, 19:3358–3370.
Locatelli, C., Filippin-Monteiro, F. B., & Creczynski-Pasa, T. B.
(2013) Alkyl esters of gallic acid as anticancer agents: A review.
European Journal of Medicinal Chemistry, 60:233–239.
Masoud, M. S., Hagagg, S. S., Ali, A. E., & Nasr, N. M. (2012) Syn-
thesis and spectroscopic characterization of gallic acid and some of
its azo complexes. Journal of Molecular Structure, 1014:17–25.
Mohan, S. V., Raghavulu, S. V., Mohanakrishna, G., Srikanth, S., &
Sarma, P. (2009) Optimization and evaluation of fermentative
hydrogen production and wastewater treatment processes using data
enveloping analysis (DEA) and Taguchi design of experimental
(DOE) methodology. International Journal of Hydrogen Energy,
34:216–226.
Nam, S. H., Park, J., Jun, W., Kim, D., Ko, J.-A., El-Aty, A. A., …
Yang, K. Y. (2017) Transglycosylation of gallic acid by using Leu-
conostoc glucansucrase and its characterization as a functional cos-
metic agent. AMB Express, 7:224–234.
Nayeem, N., & Smb, A. (2016) Gallic acid: A promising Lead mole-
cule for drug development. Journal of Applied Pharmacy, 8:
213–216.
Nonaka, G., & Nishioka, I. (1983) Tannins and related compounds.
X. Rhubarb (2): Isolation and structures of a glycerol gallate, gallic
acid glucoside gallates, galloylglucoses and isolindleyin. Chemical
and Pharmaceutical Bulletin, 31:1652–1658.
Candida antarctica lipase
medium. RSC Advances, 6:69658–69669.
B in a toluene–glycerol biphasic
Roby, M. H. H. (2017) Synthesis and characterization of phenolic
lipids. In M. Soto-Hernández, M. P. Tenango, & R. García-Mateos
(Eds.), Phenolic compounds-natural sources, importance and
applications (pp. 89–116). London, England: InTech.
Song, Z., & Xiao, Z. (1988) The preparation of 1-glycerol Gallate and
study on its Antioxidantion properties. Chemistry and Industry of
Forest Products, 3:9–18.
Stamatis, H., Sereti, V., & Kolisis, F. (1999) Studies on the enzymatic
synthesis of lipophilic derivatives of natural antioxidants. Journal
of the American Oil Chemists’ Society, 76:1505–1510.
Sun, S., & Hu, B. (2017) A novel method for the synthesis of glyceryl
monocaffeate by the enzymatic transesterification and kinetic analy-
sis. Food Chemistry, 214:192–198.
Sun, S., Hu, B., Qin, F., & Bi, Y. (2015) Comparative study of soy-
bean oil and the mixed fatty acids as acyl donors for enzymatic
preparation of feruloylated acylglycerols in ionic liquids. Journal of
Agricultural and Food Chemistry, 63:7261–7269.
Taguchi, G. (1987) System of experimental design: Engineering
methods to optimize quality and minimize costs (Vol. 1). White
Plains, NY: UNIPUB/Kraus International.
Tamayo, J., Ladero, M., Santos, V., & Garcia-Ochoa, F. (2012) Ester-
ification of benzoic acid and glycerol to α-monobenzoate glycerol
in solventless media using an industrial free Candida antarctica
lipase B. Process Biochemistry, 47:243–250.
Ter Veld, M. G., Schouten, B., Louisse, J., van Es, D. S., van der
Saag, P. T., Rietjens, I. M., & Murk, A. J. (2006) Estrogenic
potency of food-packaging-associated plasticizers and antioxidants
as detected in ERα and ERβ reporter gene cell lines. Journal of
Agricultural and Food Chemistry, 54:4407–4416.
Willett, S. A., & Akoh, C. C. (2018) Application of Taguchi method
in the enzymatic modification of menhaden oil to incorporate capric
acid. Journal of the American Oil Chemists’ Society, 95:299–311.
J Am Oil Chem Soc (2019)