Solvent-free enzymatic synthesis of 1,2-dipalmitoylgalloylglycerol: Characterization and optimization of reaction condition

Article abstract of DOI:10.1016/j.foodchem.2020.128604

A novel diacylglycerol-based galloyl structured lipid, 1,2-dipalmitoylgalloylglycerol (DPGG), was synthesized using the enzymatic transesterification of propyl gallate (PG) and tripalmitin under solvent-free condition. An immobilized and commercially available food-grade Candida antarctica lipase B, Lipozyme 435, was used as the biocatalyst. The reaction variables that affect the yield of DPGG were optimized using a 3³ full factorial design. At 70 °C, DPGG was obtained at a yield of 33.0 ± 2.0% with PG conversion at 44.8 ± 1.8% when the following condition was used: 25 substrate molar ratio of tripalmitin to PG, 120 h reaction time, and 25% enzyme load relative to the total substrate weight. The structure of reaction product was elucidated using Fourier-transform infrared spectroscopy (FT-IR), electrospray ionization high-resolution accurate-mass tandem mass spectrometry (ESI-HRAM-MS/MS), and 1D and 2D nuclear magnetic resonance spectroscopy (NMR). The effects of different lipases and galloyl donors/acceptors on the transesterification were also investigated.

Full text of DOI:10.1016/j.foodchem.2020.128604

Food Chemistry xxx (xxxx) xxx
Contents lists available at ScienceDirect
Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem
Solvent-free enzymatic synthesis of 1,2-dipalmitoylgalloylglycerol:
Characterization and optimization of reaction condition
*
Siyu Zhang, Joseph R. Hyatt, Casimir C. Akoh
Department of Food Science and Technology, University of Georgia, Athens, GA, USA
A R T I C L E I N F O
A B S T R A C T
Keywords:
A novel diacylglycerol-based galloyl structured lipid, 1,2-dipalmitoylgalloylglycerol (DPGG), was synthesized
using the enzymatic transesterification of propyl gallate (PG) and tripalmitin under solvent-free condition. An
immobilized and commercially available food-grade Candida antarctica lipase B, Lipozyme® 435, was used as the
1
,2-Dipalmitoylgalloylglycerol
Lipase-catalyzed transesterification
Structure elucidation
3
biocatalyst. The reaction variables that affect the yield of DPGG were optimized using a 3 full factorial design.
Solvent-free reaction
◦
At 70 C, DPGG was obtained at a yield of 33.0 ± 2.0% with PG conversion at 44.8 ± 1.8% when the following
condition was used: 25 substrate molar ratio of tripalmitin to PG, 120 h reaction time, and 25% enzyme load
relative to the total substrate weight. The structure of reaction product was elucidated using Fourier-transform
infrared spectroscopy (FT-IR), electrospray ionization high-resolution accurate-mass tandem mass spectrometry
(
ESI-HRAM-MS/MS), and 1D and 2D nuclear magnetic resonance spectroscopy (NMR). The effects of different
lipases and galloyl donors/acceptors on the transesterification were also investigated.
1
. Introduction
hydroxycinnamic acids have been intensively studied via enzymatic or
chemical routes, including caffeic acid, ferulic acid, and p-coumaric
acid, and both hydrophilic (water-soluble) and lipophilic (oil-soluble)
derivatives have been successfully obtained (Compton, Laszlo, & Ber-
how, 2000; Compton, Laszlo, & Evans, 2012; Sun & Hu, 2017; Sun,
Wang, & Zhu, 2017; Wang & Shahidi, 2014). However, only few reports
on the modification of hydroxybenzoic acid derivatives are available
and they used complicated chemical synthesis (Pollastri, Porter, McIn-
tosh, & Simon, 2000; Schmidt, Carrigan, & DeWolf, 2006; Song & Xiao,
1988; Takasago, Horikawa, & Masuyama, 1976). As a hydroxybenzoic
acid, gallic acid (GA) has one of the highest antioxidant capacities
among the phenolic acids that commonly occur in plants (Kikuzaki,
Hisamoto, Hirose, Akiyama, & Taniguchi, 2002; Ozgen, Reese, Tulio,
Scheerens, & Miller, 2006). This high antioxidant capacity of GA is
partially attributed to the high number (three) of the hydroxyl groups
attached to the benzene ring, which also makes it susceptible to oxida-
tion and other chemical reactions. As a result, protection steps for GA are
required during chemical synthesis (Takasago et al., 1976), compli-
cating the synthesis and purification processes. Furthermore, the toxic
reagents used in chemical synthesis may also preclude the use of the
reaction products for human consumption.
Plant derived phenolic compounds have attracted substantial inter-
est owing to their beneficial, functional, and nutritional properties,
which include antioxidant and antimicrobial activities (Lee et al., 2009).
Phenolic acids, as a subclass of phenolic compounds, are potent anti-
oxidants that universally exist abundantly in plants (Rice-Evans, Miller,
&
Paganga, 1996). The synthetic antioxidants currently used in foods,
such as butylated hydroxyanisole (BHA), butylated hydroxytoluene
BHT), tert-butyl hydroquinone (TBHQ), and alkyl gallates, are sus-
(
pected of being carcinogenic (BHA, BHT and TBHQ) (Stamatis, Sereti, &
Kolisis, 1999), and contact allergenic/estrogenic (alkyl gallates) (Ama-
dasi, Mozzarelli, Meda, Maggi, & Cozzini, 2008; Crontn, 1980). Due to
the adverse health effects of synthetic antioxidants, researches have
been focused on the potential use of natural antioxidants, such as
phenolic acids, in food products. As a class of secondary phenolic me-
tabolites, phenolic acids function as antioxidants in plants, performing
the roles of reducing agents, free radical scavengers, and quenchers of
singlet oxygen formation (Ghasemzadeh & Ghasemzadeh, 2011).
However, the solubility of most phenolic acids in water or oil is usually
low, which limits their applications in foods.
Hydroxybenzoic and hydroxycinnamic acids are two types of
phenolic acids that are widely distributed in plants. The modifications of
The idea of producing glycerol-based lipophilic derivatives of GA
was first implemented in 1950 with the purpose of using it in fish oils as
*
Corresponding author.
E-mail addresses: sz25438@uga.edu (S. Zhang), jrh1@uga.edujrh, 1@uga.edu (J.R. Hyatt), cakoh@uga.edu (C.C. Akoh).
https://doi.org/10.1016/j.foodchem.2020.128604
Received 2 July 2020; Received in revised form 4 November 2020; Accepted 7 November 2020
Available online 17 November 2020
0
308-8146/© 2020 Elsevier Ltd. All rights reserved.
Please cite this article as: Siyu Zhang, Food Chemistry, https://doi.org/10.1016/j.foodchem.2020.128604

S. Zhang et al.
Food Chemistry xxx (xxxx) xxx
antioxidants (Bittenbender, 1950). Compared with alkyl gallates, 1,2-
dipalmitoylgalloylglycerol (DPGG), as well as some other diac-
ylglycerol- and monoacylglycerol-based GA derivatives, showed higher
protective effects on various oils (Bittenbender, 1950). However, the
purification and structure characterization were not carried out due to
the technology limitation at the time. Many more functions and appli-
cations of DPGG have been reported in the last 20 years, such as self-
adhesion, bilayer and liposome forming, drug delivery, and biosensing
applications (Fotoran et al., 2019; Kang et al., 2014; Pollastri et al.,
DPGG consists of one gallate moiety esterified at the terminal position of
the glycerol backbone with each of the other two hydroxyl groups
esterified with one palmitate. Lipases and feruloyl esterases have been
reported capable of catalyzing the esterification/transesterification of
phenolic acids and glycerol or glycerol-based lipids (Compton et al.,
2000; Matsuo et al., 2008; Sun & Hu, 2017; Zeng et al., 2014). Compared
with feruloyl esterases, lipases possess a much broader substrate range
and are able to catalyze a wide range of reactions (Fojan, Jonson,
Petersen, & Petersen, 2000), making it more likely to overcome the
steric hindrance caused by the galloyl moiety, and more suitable for
catalyzing the esterification/transesterification involving GA and its
derivatives. Also, lipases are more stable and active in nonaqueous
environment, more commercially available, and have a preference for
hydrophobic substrates (Bornscheuer, 2002).
2
000; Schmidt & DeWolf, 2004). But the chemical synthesis methods
used in previous studies involved either using high reaction temperature
Bittenbender, 1950), which causes decarboxylation of GA, or the use of
toxic solvents/reagents and complicated reaction and purification steps
Pollastri et al., 2000; Schmidt et al., 2006). Recent studies applied more
(
(
intricate methods using protecting groups for aromatic hydroxyl groups
and forming acyl chloride to facilitate the esterification (Pollastri et al.,
Herein, we report a simple selective enzymatic method for producing
DPGG, using a commercially available food-grade Candida antarctica
lipase B (CALB), Lipozyme® 435, as the biocatalyst in solvent-free
media. As described in Fig. 1, the transesterification of propyl gallate
2
000; Schmidt et al., 2006). As an alternative to these methods, a one-
step solvent-free enzymatic synthesis method could be more suitable
for DPGG synthesis. In contrast to chemical synthesis, enzymatic syn-
thesis is selective and does not require toxic solvents/reagents or high
reaction temperature, and therefore is more suitable for producing
bioactive compounds for use in foods, cosmetics, and pharmaceuticals.
In our previous studies, GA was converted to a more hydrophilic
form, 1-o-galloylglycerol (GG), through the enzymatic glycerolysis of PG
using glycerol as both reactant and solvent (Zhang & Akoh, 2019; 2020).
A similar scheme could be used to produce lipophilic GA derivatives.
3
(PG) and tripalmitin was proposed as the synthesis scheme. A 3 full
factorial design was used to optimize the synthesis conditions, including
the substrate ratio, reaction time, and enzyme load. The structure of
reaction product was characterized using Fourier-transform infrared
spectroscopy (FT-IR), electrospray ionization high-resolution accurate-
mass tandem mass spectrometry (ESI-HRAMMS/MS), and 1D and 2D
nuclear magnetic resonance spectroscopy (NMR). The effects of different
lipases and galloyl donors/acceptors on the transesterification were also
–
Fig. 1. Reaction scheme for 1,2-dipalmitoylgalloylglycerol (DPGG) synthesis and the possible side reactions. Palmitate: R = ꢀ (C
–
O)ꢀ (CH
2
)14 ꢀ CH
3
.
2

S. Zhang et al.
Food Chemistry xxx (xxxx) xxx
investigated and reported.
min, then an isocratic elution with solvent C for 7 min. Throughout the
sequence, the UV spectra (from 190 nm to 400 nm) of the eluates were
recorded. Under these conditions, the retention times were: GA, 6.1 min;
PG, 15.2 min; monopalmitoylgalloylglycerol(s) (MPGG), 25.2 min;
monopalmitin, 26.9 min; DPGG, 30.1 min; dipalmitin, 32.3 min; tri-
palmitin, 37.8 min (Fig. 2). The yield of DPGG, conversion of PG, hy-
drolysis of PG, and reaction selectivity were calculated according to the
following equations:
2
. Materials and methods
2
.1. Chemicals and reagents
n-Propyl gallate (≥98%) was purchased from HiMedia Laboratories
(
Nashik, India). Tripalmitin (≥85%) was purchased from Tokyo Chem-
ical Industry (Tokyo, Japan). GA, methyl gallate (MG), and tricaprylin
97–98%) was purchased from Sigma-Aldrich (St. Louis, MO, USA).
Moles of DPGG form ed
(
Yield (\%) = ) =
(1)
(2)
(3)
(4)
Moles of PG at t = 0 h
Lipozyme® RM IM (lipase from Rhizomucor miehei, immobilized on an
anionic exchange resin), Lipozyme® TL IM (lipase from Thermomyces
lanuginosus and immobilized on a non-compressible silica gel carrier),
and Lipozyme® 435 (recombinant lipase B from Candida antarctica,
expressed in Aspergillus niger, and immobilized on a macroporous hy-
drophobic resin) were purchased from Novozymes North America, Inc.
Moles of PG consum ed
Conversion (\%) = ) =
Hydrolysis (\%) = ) =
Reaction selectivity =
Moles of PG at t = 0 h
Moles of GA form ed
Moles of PG at t = 0 h
(
Franklinton, NC, USA). All solvents used in this study were HPLC grade
Moles of DPGG form ed
Moles of PG consum ed
and purchased from local chemical suppliers. All chemicals were used
without further treatment or purification.
2
.2. Enzymatic transesterification
2.4. Structural determination and characterization
Closed-cap Reacti-Vial™ (Thermo Fisher Scientific, Waltham, MA,
The structure of DPGG was analyzed as described below. A Nicolet
Nexus FT-IR 1100 spectrometer (Thermo Fisher Scientific Co. Ltd.,
Waltham, MA, USA) equipped with a ZnSe attenuated total reflection
USA) with a volume of 10 mL and an internal cone was used to conduct
the transesterification of PG and tripalmitin. After various amounts of
◦
tripalmitin were added into the reaction vials and melted at 70 C, 125
(ATR) attachment was used to collect the FT-IR spectra of samples ( max
ν
ꢀ 1
μ
mol (26.5 mg) of PG were added and dissolved under a constant ni-
was reported in cm ). GA, MG, PG, and tripalmitin were also analyzed
for structural comparison with DPGG. Before each experiment, the in-
strument was purged with nitrogen for at least 10 min. After 10 mg of
each sample were dissolved in 1 mL methanol (GA, MG, and PG) or
chloroform (DPGG and tripalmitin), a drop of the solution was placed
onto the ZnSe crystal, then pressed using the attached accessory after
evaporating the solvent. The spectra were collected from 650 to 4,000
trogen flow. After PG was dissolved, the enzyme was added to the
mixtures. The reaction mixtures were flushed with nitrogen, incubated
◦
at 70 C, and stirred at 200 rpm using a Reacti-Therm™ heating and
stirring module (Thermo Fisher Scientific, Waltham, MA, USA) equip-
ped with aluminum heating blocks and triangular magnetic stirring bars.
Samples (10 L) were withdrawn every 24 h and diluted to 1 mL with 1:1
μ
ꢀ 1
ꢀ 1
isopropyl alcohol:chloroform (v/v) for further analysis. The reaction
temperature was chosen considering the melting point of tripalmitin
cm with a resolution of 4 cm and 32 scans. The data were processed
using Omnic Specta software (Thermo Fisher Scientific, Waltham, MA,
USA) and the KnowItAll® informatic system (Bio-Rad Laboratories,
Hercules, CA, USA).
◦
◦
(
65–67 C) and DPGG (69.3 C) (Pollastri et al., 2000), thermal stabil-
ities of GA and PG (Boles, Crerar, Grissom, & Key, 1988; Santos et al.,
012), and the reported enzyme activities in similar reactions (Sun &
Hu, 2017; Sun, Zhu, & Bi, 2014).
2
NMR spectroscopy analysis was done with the sample dissolved in
CDCl3 with 0.1% tetramethylsilane (TMS) as internal standard at a
-
1
1
13
1
1
concentration of 10 mg L . H, C, H– H gradient correlation spec-
1
13
2
.3. Reaction mixture analysis
troscopy (gCOSY), H– C gradient heteronuclear single quantum
1 13
coherence (gHSQC), H– C gradient heteronuclear multiple quantum
1
13
Diluted samples withdrawn from the reaction mixture were analyzed
coherence (gHMQC), and H– C gradient heteronuclear multiple bond
◦
by thin layer chromatography (TLC) on 250
μ
m silica gel G plates
correlation (gHMBC) spectra were recorded at 25 C using a Varian
(
Uniplate, Analtech, Newark, DE, USA). A mixture of methanol/chlo-
Unity Inova 500 MHz NMR Spectrometer (Varian Inc., Palo Alto, CA,
USA) equipped with a 5-mm hydrogen-carbon-nitrogen room tempera-
ture probe. The chemical shifts of DPGG were reported in parts per
roform (1:9, v/v) was used as the mobile phase. The TLC plates were
directly visualized under 254 and 365 nm ultraviolet (UV) light for
′
′
1
preparative purposes or after spraying with 0.2% (w/w) 2 ,7 -dichloro-
fluorescein methanol solution. The Rf values were PG, 0.61; DPGG, 0.82;
dipalmitin, 0.88; and tripalmitin, 0.98. The band corresponding to
DPGG on the preparative plates was recovered and extracted with
acetone. The solvent was then evaporated under a nitrogen flow and the
product was further analyzed using FT-IR, NMR spectroscopy, and ESI-
HRAM-MS/MS.
million (δ/ppm). For H NMR analysis, the scan range was set to vary
from ꢀ 1.0 to 8.5 ppm with 8 scans. For gCOSY, the scan range was from
ꢀ 1.0 to 8 ppm with 512 t1 increments and 4 scans per t1 increment. For
1
3
gHSQC, the C spectrum width ranged from ꢀ 10 to 190 ppm, and the
1
H spectrum width ranged from ꢀ 1.0 to 8.5 ppm with 128 t1 increments
1
3
and 32 scans per increment. For gHMBC, the C spectrum width ranged
1
from ꢀ 15 to 225 ppm, and the H spectrum width ranged from ꢀ 1.0 to
1
An Agilent 1260 Infinity HPLC system (Santa Clara, CA, USA)
equipped with a diode-array detector (DAD) scanned at 280 nm was
used for quantification purposes. A Sedex 85 low-temperature evapo-
8.5 ppm with 200 t1 increments and 32 scans per increment. The H
1
1
3
3
spectra of gHSQC and gHMBC were both acquired at 500 MHz. The
spectra of gHSQC and gHMBC were acquired at 125 MHz. The
C
C
◦
rative light-scattering detector (LT-ELSD) operated at 3.5 bar and 40 C
spectra were acquired indirectly from the gHSQC and gHMBC spectra.
was used to monitor the hydrolysis of tripalmitin. An Agilent Zorbax
The chemical shifts were assigned based on the 1D and 2D NMR spectra
1
StableBond C18 column (5
μ
m, 80 Å, 250 × 4.6 mm, Santa Clara, CA,
as follows: H NMR (500 MHz, CDCl
3
, TMS) δ: 7.21 (s, 2H, armo. H),
◦
USA) was used at a controlled temperature of 35 C. For each sample, an
6.10 (s, 3H, –OH), 5.48–5.38 (m, 1H, >CH–O–palmitate), 4.47–4.32 (m,
3H, ꢀ CH –O–gallate, ꢀ (CH–H)–O–palmitate), 4.24 (dd, J = 12.0, 6.0
Hz, 1H, ꢀ (CH–H)–O–palmitate), 2.36–2.30 (m, 4H, ꢀ OCOCH –), 1.62
(d, –CH –), 1.30–1.21 (m, 48H,
10.5 Hz, 4H, ꢀ OCOCH
ꢀ (CH 12–CH ), 0.88 (t, J = 6.9 Hz, 6H, ꢀ CH
O, CDCl
, TMS) δ: 173.87 (CΔ1 palmitoyl–O–CH
ꢀ 1
aliquot of 20
μL was injected and eluted at a flow rate of 0.7 mL min
.
2
The mobile phase consisted of 1% acetic acid (v/v) in water (A),
methanol (B), and isopropyl alcohol (C). The samples were eluted using
a linear gradient from 90% (v/v) A to 0% A (v/v) in 20 min, followed by
a consecutive linear gradient from 100% (v/v) B to 100% C (v/v) in 13
2
J
=
2
2
1
3
2
)
3
3
); C NMR (125 MHz,
–), 173.64 (CΔ1
D
2
3
2
3

S. Zhang et al.
Food Chemistry xxx (xxxx) xxx
◦
Fig. 2. HPLC chromatograms of the reaction mixture after 120 h reaction time under the following reaction condition: 70 C, substrate ratio at 25/1 (tripalmitin/PG,
mol/mol), an enzyme load of 25% (relative to the total weight of substrates). (a) HPLC-ELSD chromatogram of the reaction mixture; (b) HPLC-DAD chromatogram of
the reaction mixture at 280 nm with UV spectra of gallic acid (GA), propyl gallate (PG), monopalmitoylgalloylglycerol(s) (MPGG), and 1,2-dipalmitoylgalloylgly-
cerol (DPGG).
Table 1
Experimental design and the effects of the substrate ratio, enzyme load, and reaction time on the yield of 1,2-dipalmitoylgalloylglycerol (DPGG), hydrolysis of propyl
gallate (PG), conversion of PG, and reaction selectivity during the transesterification of PG and tripalmitin using Lipozyme® 435 as the catalyst.
a
b
c
Run
Substrate Ratio
Enzyme Load (%)
Reaction time (h)
Yield (%)
Hydrolysis (%)
Conversion (%)
Reaction selectivity
1
2
3
4
5
6
7
8
9
25
15
5
25
25
25
15
15
15
5
24
14.3 ± 0.3
13.2 ± 1.9
11.8 ± 3.7
3.7 ± 0.4
3.7 ± 0.6
6.1 ± 1.8
2.6 ± 1.7
0.5 ± 0.1
0.4 ± 0.1
25.7 ± 1.5
21.8 ± 2.8
22.8 ± 2.3
5.8 ± 2.0
6.0 ± 1.2
10.4 ± 4.0
3.4 ± 0.4
1.0 ± 0.3
0.9 ± 0.2
33.0 ± 2.1
31.1 ± 2.1
27.3 ± 2.3
8.9 ± 2.1
7.9 ± 2.1
9.9 ± 5.6
3.9 ± 0.9
1.0 ± 0.3
0.9 ± 0.1
3.7 ± 0.7
5.6 ± 0.8
8.2 ± 1.3
3.2 ± 0.3
4.8 ± 1.2
10.2 ± 1.6
6.3 ± 0.3
2.8 ± 0.6
7.2 ± 1.4
6.0 ± 0.9
8.9 ± 0.1
15.9 ± 1.3
4.9 ± 0.9
6.6 ± 0.6
15.1 ± 1.0
5.3 ± 0.3
3.0 ± 0.6
12.3 ± 2.3
8.1 ± 1.8
9.6 ± 2.8
16.5 ± 3.9
4.4 ± 1.2
5.7 ± 1.3
16.5 ± 2.6
6.1 ± 0.8
3.8 ± 1.1
9.1 ± 2.1
19.4 ± 0.3
19.5 ± 2.0
23.0 ± 6.2
7.3 ± 0.1
0.737 ± 0.023
0.673 ± 0.054
0.517 ± 0.101
0.502 ± 0.071
0.416 ± 0.020
0.330 ± 0.041
0.236 ± 0.119
0.132 ± 0.008
0.049 ± 0.013
0.770 ± 0.035
0.684 ± 0.024
0.555 ± 0.051
0.517 ± 0.131
0.456 ± 0.069
0.375 ± 0.057
0.336 ± 0.022
0.202 ± 0.027
0.060 ± 0.004
0.737 ± 0.050
0.717 ± 0.058
0.586 ± 0.054
0.625 ± 0.024
0.550 ± 0.018
0.331 ± 0.101
0.338 ± 0.027
0.169 ± 0.037
0.081 ± 0.013
24
24
25
15
5
24
24
9.0 ± 1.7
24
18.2 ± 4.0
10.6 ± 2.0
4.1 ± 1.1
25
15
5
24
5
24
5
24
9.0 ± 1.8
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
25
15
5
25
25
25
15
15
15
5
72
33.5 ± 3.1
31.9 ± 3.2
41.2 ± 0.8
11.0 ± 1.1
13.0 ± 0.7
27.0 ± 6.3
10.1 ± 0.6
4.6 ± 1.0
72
72
25
15
5
72
72
72
25
15
5
72
5
72
5
72
14.5 ± 3.3
44.8 ± 1.8
43.5 ± 4.3
47.0 ± 7.6
14.2 ± 3.0
14.3 ± 3.5
28.3 ± 8.6
11.4 ± 1.9
6.0 ± 1.7
25
15
5
25
25
25
15
15
15
5
120
120
120
120
120
120
120
120
120
25
15
5
25
15
5
5
5
11.5 ± 2.5
All values are expressed as mean ± standard deviation, n = 3.
a
Numbers were run randomly.
b
Tripalmitin to PG, mol/mol.
c
Relative to the weight of total substrates.
4

S. Zhang et al.
Food Chemistry xxx (xxxx) xxx
palmitoyl–O–CH<), 166.13 (galloyl–COO–), 143.60 (galloyl C3,
5
),
conversion of GG was also less than optimal after reacting for 120 h
(data not shown). Moreover, GG is not commercially available, which
makes this strategy even less desirable. Meanwhile, when the trans-
esterification of PG and tripalmitin (Fig. 1) was conducted, DPGG was
the major reaction product and decent PG conversions were observed
(data not shown). Similar results were reported in previous studies when
caffeic and ferulic acids were incorporated into monoacylglycerols and
diacylglycerols (Sabally, Karboune, St-Louis, & Kermasha, 2006, 2007;
Sun & Hu, 2017; Sun, Song, Bi, Yang, & Liu, 2013; Sun, Wang, & Zhu,
2017; Sun et al., 2008). Compared with the other two synthesis strate-
gies, transesterification showed more favorable results. Thus, the
transesterification of PG with tripalmitin was chosen as the route of
synthesis for the production of DPGG.
1
21.08 (galloyl C
2.96 (ꢀ CH
1
), 110.08 (galloyl C2, 6), 69.21 (>CH–O–palmitate),
6
2
–O–gallate), 62.37 (ꢀ CH
palmitate), 29.57 (C
palmitate), 25.03 (CΔ3 palmitate), 22.76 (C
palmitate).
2
–O–palmitate), 34.22 (CΔ2
13 palmitate), 29.10
palmitate), 14.08
palmitate), 34.1, 31.91 (C
ω
3
ω ω
5-
(
C
C
ω
ω
4
1
ω
2
(
The same sample used for NMR spectroscopy was also used for ESI-
HRAM-MS/MS analysis. Analysis was performed by direct infusion in a
Thermo Orbitrap Elite mass spectrometer (Thermo Fisher Scientific,
Waltham, MA, US) with an ESI interface using negative-ion mode. For
MS, the instrument was set to scan from 175 to 1525 mass-to-charge (m/
z). For tandem MS analysis, collision-induced dissociation was chosen as
the fragmentation method using the same mass spectrometer and the
spectra were collected from 200 to 800 m/z.
3
.2. Structural characterization of 1,2-dipalmitoylgalloylglycerol
2
.5. Experimental design and statistical analysis
The transesterification of PG and tripalmitin was performed in a
3
◦
A 3 full factorial design (Table 1) consisting of three independent
closed system with melted tripalmitin at 70 C as both the reactant and
solvent. An immobilized and commercially available food-grade lipase,
Lipozyme® 435, was used as the biocatalyst. Along the course of
transesterification (Fig. 1), three new peaks appeared on the HPLC-DAD
chromatogram while the PG peak decreased (Fig. 2). The peaks at 6.1
min (GA) and 30.1 min (DPGG) were predominant among the reaction
products. No change in total peak area for HPLC-DAD spectra was
observed. The HPLC-ELSD chromatogram revealed that the hydrolysis of
tripalmitin occurred, forming dipalmitin (32.3 min) and monopalmitin
(26.9 min). Dipalmitin was the major hydrolysis product of tripalmitin,
while some of the monopalmitin already existed in the tripalmitin ma-
terial as impurity before the reaction. Further analysis showed all three
new peaks in the HPLC-DAD chromatogram share similar UV spectra
with absorption peak around 275 nm (Fig. 2). Notably, the small MPGG
peak at 25.2 min might consist of both 1-monopalmitoyl-3-galloylgly-
cerol and 2-monopalmitoyl-3-galloylglycerol. No formation of GG
occurred. After the reaction mixture was separated using TLC, the
structure of the band corresponding to DPGG was elucidated using FT-
IR, ESI-HRAM-MS/MS, and NMR spectroscopy.
factors (enzyme load, substrate ratio, and reaction time) was used to
optimize the synthesis conditions and to evaluate the main and inter-
action effects of reaction variables on the reaction. The yield of DPGG,
the conversion of PG, the hydrolysis of PG, and the reaction selectivity
were used as responses with an equal level of importance. The inde-
pendent factors and levels were as follows: enzyme load (5, 15, and 25%;
relative to the weight of total substrates), substrate ratio (5, 15, and 25;
tripalmitin/PG, mol/mol), and reaction time (24, 72, and 120 h).
Statistical analysis of the obtained results was conducted using JMP®
software (version 15, SAS Institute, Inc., Cary, NC, USA) and SPSS Sta-
tistics (version 25, IBM Corp., Armonk, NY, USA). The results were
expressed as mean values ± standard deviation (SD). To determine the
optimal reaction conditions and to evaluate the effects of reaction var-
iables, the responses were subjected to analysis of variance (ANOVA)
tests with 2-way interactions. The results were expressed as least-
squares means (LS mean, also called marginal means) ± the standard
error of mean (SEM). Tukey’s honest significant difference test (Tukey
HSD) was used to determine the occurred differences when varying each
reaction parameter and the level of significance (p < 0.05) among them.
Empirical polynomial quadratic equations were applied to correlate
each response variable to the independent variables using regression
analysis. The general form of the model is shown in Eq. (5):
The detailed report of the FT-IR spectra and the interpretation are
shown in Supplementary material (Supplementary material
Figs. S1–S3). The peaks in the spectra were assigned according to Coates
(2000) spectral interpretations and using KnowItAll® informatic sys-
ꢀ
1
tem. A broad band within 3,100–3,700 cm region typical of the O
–
H
4
4
4
4
∑
∑
∑ ∑
2
stretching mode for hydroxy was observed in the DPGG spectrum
(Supplementary material Fig. S1), which could be attributed to the three
hydroxy groups in the gallate moiety of the molecule. Similar bands
were also shown in GA, MG, and PG spectra (Supplementary material
Y = β +
βiX
i
+
β X +
βijX
i
X
j
(5)
0
ii
i
i = 1
i = 1
i = 1 j = 1
in which Y is the predicted response, X
i
and X
j
are independent factors,
ꢀ
1
β
0
is the intercept, β
i
is the linear coefficient, βii is the quadratic coeffi-
Fig. S2). An additional broad peak around 3000 cm in the GA spec-
trum, could be attributed to the O H stretching of the COOH. The lack
cient, and βij is the interaction coefficient.
–
ꢀ 1
All experiments were carried out at least in triplicate. The experi-
mental results were expressed as mean ± SD while the predictive results
were given as mean ± SEM.
of such peak above 3000 cm in the tripalmitin spectrum indicates the
absence of free hydroxy groups. The strong sharp peaks from 2800 to
ꢀ
1
3000 cm
Fig. S2d) spectra corresponded with the various C
long acyl chains. The most distinguishable feature of the DPGG spectrum
in the DPGG and tripalmitin (Supplementary material
–
H stretching of the
3
. Results and discussion
ꢀ
1
is the three closely located peaks around 1,700 cm , which could be
explained by the existence of three types of ester linkages in the mole-
3
.1. Preliminary study
ꢀ 1
could be ascribed to the C–
O
cule. The first peak at 1,738 cm
–
Several strategies, including alcoholysis, direct esterification, and
stretching in the ester linkage between palmitate and the glyceryl pri-
ꢀ 1
transesterification, were carried out in the attempt to produce DPGG
using Lipozyme® 435. For the alcoholysis of PG with a mixture of mono-
and dipalmitin/distearin, zero or trace amounts of diacylgalloylglycerol
or monoacylgalloylglycerol were obtained after a 120-h reaction time
under various reaction conditions. These results could be explained by
the steric hindrance between monoacylglycerols/diacylglycerols and
PG, as well as the competing reaction to form triacylglycerols and GA.
When attempting to produce DPGG via the direct esterification of GG
with palmitic acid using GG produced from a previous study (Zhang &
Akoh, 2020), MPGG was the predominant product while the overall
mary hydroxyl group. A peak located at 1736 cm in the tripalmitin
–
spectrum indicated the same C O function. The second peak in the
–
ꢀ 1
cluster in the DPGG spectrum, 1712 cm , and in the tripalmitin spec-
trum, 1728 cm 1, represented the ester linkage between glyceryl and
ꢀ
palmitate at the sn-2 position of the glycerol. The third peak in the
ꢀ
1
cluster of DPGG spectrum, 1695 cm , confirmed the existence of the
ester linkage in between the gallate moiety and the primary alcohol
function of the glycerol moiety, which shifted from the typical 1,740
ꢀ 1
cm location by the double bond conjugated with the aromatic ring
–
structure. The C O stretching modes were detected at 1698, 1688, and
–
5

S. Zhang et al.
Food Chemistry xxx (xxxx) xxx
692 cm 1 for PG, GA, and MG, respectively. The 1,2,3,5-substituted
ꢀ
3.4. Effect of substrate ratio
1
aromatic ring stretching pattern was also observed for DPGG, PG, GA,
and MG.
Unlike the previously reported glycerolysis reaction using PG (Zhang
& Akoh, 2019), increasing the molar ratio of acyl acceptor did not
necessarily lead to increased yield. As shown in Fig. 3a and 3b, although
changing substrate ratio had minimal effect on yield (with slightly lower
yield for the substrate ratio at 15), it did have a significant effect on
hydrolysis. When the substrate molar ratio of tripalmitin to PG increased
to 15 or 25, the hydrolysis significantly decreased, leading to the
increased reaction selectivity and the decreased PG conversion. Similar
results were observed in a previous study, when the molar ratio of tri-
acylglycerols to ethyl ferulate exceeded 3, no change in yield was
noticed with slightly higher hydrolysis at the lower substrate ratio (Sun
et al., 2017). Notably, when the molar ratio of tripalmitin to PG was used
at 5, a turbid reaction mixture was obtained due to the limited solubility
of PG in oils. Thus, testing the effect of higher PG to tripalmitin molar
ratio was difficult in this solvent-free reaction condition. In addition, the
commercial prices of PG are usually much higher than oils (including
tripalmitin), so a high tripalmitin to PG substrate molar ratio would not
significantly increase the cost. Moreover, a significantly higher reaction
selectivity was obtained when the substrate molar ratio increased from 5
or 15 to 25, meaning less PG was wasted at the side reactions. Without
being wasted on the side reactions, the residual PG could be easily
recovered and reused. Therefore, considering all the effects of substrate
ratio on the yield, hydrolysis, conversion, and reaction selectivity, 25:1
tripalmitin/PG (mol/mol) was chosen as the optimal reaction substrate
ratio.
ESI-HRAM-MS and -MS/MS were used to determine the elemental
composition and structure of the reaction product (Supplementary ma-
1
terial Figs. S4 and S5). In MS (Supplementary material Fig. S4), the
-
1
molecular ion [Mꢀ H] was found at 719.51015 m/z, which was
ꢀ
71 9
H O as the chemical formula for [DPGG-H]
-1
consistent with C42
calculated 719.51036 m/z, error = -0.29 ppm). The tandem ESI-HRAM-
(
-
1
MS data (Supplementary material Fig. S5) showed the [Mꢀ H] peak at
2
7
19.51030 m/z (calculated 719.51036 m/z, error = -0.08 ppm). In MS ,
the most abundant fragment generated from the molecular ion was
ꢀ
ꢀ
2
)
2
55.23290 m/z, which was attributed to [palmitoyl + O] (C16
H
31
O
ions (calculated 255.23295 m/z, error = -0.20 ppm). The second
2
abundant fragment in the MS was 463.27014, which corresponded to
ꢀ
ꢀ
7
) ion (calculated 463.27013 m/
[
DPGG ꢀ palmitoyl ꢀ H
2
O] (C26
H
39
O
2
z, error = 0.22 ppm). The MS results confirmed the structure of the
1
reaction product while the MS determined its elemental composition.
However, the ESI-HRAM-MS/MS results could not definitively confirm
the position of the galloyl moiety in the glycerol backbone.
Considering the results of the H- and 13C-unidimensional NMR, as
well as 2D gCOSY, gHSQC, and gHMBC NMR (Supplementary material
Figs. S6–S10), the comprehensive structural information of DPGG could
1
1
be obtained. The H spectrum (Supplementary material Fig. S7) showed
that five hydrogen atoms from the glycerol moiety had three different
chemical shifts, which indicated that the formed ester had an asym-
metrical structure (gallate was not esterified at the secondary hydroxyl
group of glycerol). Furthermore, three aromatic hydroxyl groups
attached to the benzene ring bearing two identical hydrogen atoms,
were consistent with the galloyl structure. The results of gCOSY NMR
disclosed the connectivity of the hydrogen atoms, providing additional
evidence for the hydrogen arrangement of DPGG (Supplementary ma-
terial Fig. S8). The 13C NMR results showed that the three carbon atoms
of glycerol moiety had three different chemical shifts, which provided
further evidence that the formed compound had an asymmetrical
structure. The gHSQC spectra (Supplementary material Fig. S9)
demonstrated the one-bond correlation of the carbon atoms and
hydrogen atoms in the DPGG molecule. Combining all the information
above, the complete and definitive structural arrangement of palmitate
and gallate moieties on the glycerol backbone could be elucidated. The
long-range coupling between carbons and protons was investigated
using gHMBC NMR spectroscopy (Supplementary material Fig. S10).
The long-range correlations acquired from this experiment, revealed and
confirmed the position for the carbonyl group of the palmitate and
gallate on the glycerol moiety. Regarding all the available 1D NMR and
3.5. Effect of enzyme load
As shown in Fig. 3c and 3d, the highest LS means for the yield,
conversion, and reaction selectivity were obtained at 25% enzyme load
relative to the total substrates (w/w). As shown in Fig. 3c, when a 25
substrate molar ratio (tripalmitin/PG) was used, varying the enzyme
load from 5% to 25%, both the yield of DPGG and reaction selectivity
increased significantly, while 25% and 30% enzyme loads showed no
significant difference. The ANOVA tests (Fig. 3d) confirmed the trend for
the effect of enzyme from 5% to 25% of the total substrate weight,
significant increases in yield, hydrolysis, conversion, and reaction
selectivity was observed. Even though more hydrolysis happened, the
yield of DPGG increased more significantly, leading to a higher reaction
selectivity when enzyme load increased from 5% or 15% to 25% of the
total substrate weight. With all the data available, 25% (relative to the
weight of total substrates) was chosen as the optimal enzyme load.
2
D NMR data, our data indicated that when using Lipozyme® 435 as a
3.6. Effect of reaction time
biocatalyst, the major reaction product, DPGG, has one galloyl moiety
replacing a palmitate on the primary hydroxyl group of glycerol. The
lack of gallate esterified at the secondary hydroxyl group of glycerol
could be attributed to the great steric hindrance of such configuration
In the ANOVA tests, the highest LS mean for the yield (13.8 ± 0.5%)
and conversion (24.6 ± 0.9%) were obtained with a 120-h reaction time.
As shown in Fig. 4a, when the reaction time decreased from 120 h to 72
h or 24 h, yield, conversion, and reaction selectivity significantly
decreased. Notably, the change in yield, hydrolysis, conversion, and
reaction selectivity became less significant when the reaction time
exceeded 96 h (Fig. 4a and 4b). Although no significant difference in
reaction selectivity was found when the reaction time extended from
120 h to 144 h, a minor downward trend was observed. Considering the
increased hydrolysis caused by the increased reaction time, and to
prevent the possible oxidation of gallates, 120 h was chosen as the
optimal reaction time.
(
Zhang & Akoh, 2020).
3
.3. Full factorial design
The mean values ± SD for the yield of DPGG, conversion and hy-
drolysis of PG, and reaction selectivity from the full factorial design are
3
shown in Table 1. The 3 full factorial design required 27 runs with each
condition repeated trice. Compared with other fractional factorial de-
signs, such as response surface methodology and the Taguchi method, 33
full factorial design contains all the experimental conditions when the
same number of the independent variables are used, providing a much
higher degree of freedom than the other designs (Raki ´c , Kasagi ´c -
Vujanovi ´c , Jovanovi ´c , Jan ˇc i ´c -Stojanovi ´c , & Ivanovi ´c , 2014). Each re-
action variable and their interaction could be estimated independently.
Therefore, the effects of each independent factor were investigated and
analyzed in terms of LS means ± SEM.
3.7. Model fitting
Polynomial quadratic models with secondary-degree interactions
3
were selected to fit the data obtained using the 3 full factorial design
(Table 1), as described in Eq. (5). For each response (the yield of DPGG,
the conversion of PG, the hydrolysis of PG, and the reaction selectivity),
6

S. Zhang et al.
Food Chemistry xxx (xxxx) xxx
Fig. 3. Effects of the reaction variables on the yield of 1,2-dipalmitoylgalloylglycerol (DPGG), hydrolysis of propyl gallate (PG), conversion of PG, and reaction
selectivity. (a) Effect of substrate ratio (tripalmitin to PG, mol/mol) when the enzyme load (relative to the weight of total substrates) was fixed at 25/1; (b) effect of
substrate ratio (tripalmitin/PG, mol/mol) determined by three-way analysis of variance (ANOVA) with second-order interaction and expressed as least-squares means
(
LS mean) ± the standard error of mean (SEM); (c) effect of enzyme load (relative to the weight of total substrates) when a 25 substrate ratio (tripalmitin to PG, mol/
mol) was used; and (d) effect of enzyme load (relative to the weight of total substrates) analyzed using ANOVA with second-order interactions and expressed as LS
means ± SEM. Different letters indicate significant statistical difference at p-value < 0.05 in Tukey’s honest significant difference (HSD) tests.
the regression models are given as follows:
reaction selectivity, 94.35% of the variability in the response could be
2
explained (R = 0.9435). Even all the indicators suggested the proposed
2
2
2
Yield (\%)) = ꢀ 0.40 ꢀ 0.04X
1
+ 4.37X
2
2
+ 0.26X
1
3
+ 1.09X + 5.06X
1
models were satisfactory, accuracy checks were still necessary to
generate adequate models. In this respect, the validity of the model was
tested by comparing the experimental results obtained with three time-
course reactions (each reaction performed in triplicate) with the pre-
dicted values (Fig. 4c). The experimental values fell within the 95%
confidence intervals of the predicted values from the models, thus
confirming the validity of the model. Thus, taking all four responses into
consideration, the optimal reaction condition was determined to be: 25
substrate molar ratio (tripalmitin/PG), 25% enzyme load relative to the
total substrates (w/w), and reaction time of 120 h. As illustrated in
Fig. 4c, under the optimal condition, the models predicted a 31.68 ±
2
ꢀ
0.01X + 0.29X
1
X
+ 0.02X
X
3
+ 0.17X
2
X
3
(6)
3
2
2
Hydrolysis (\%)) = 1.69 ꢀ 2.19X
1
ꢀ 0.20X
2
+ 0.23X
3
+ 3.19X ꢀ 0.24X
1
2
2
ꢀ
0.01X ꢀ 0.99X
1
X
2
ꢀ 0.04X
1
X
3
+ 0.05X
2
X
3
(7)
3
2
2
Conversion (\%)) = 1.86 ꢀ 2.76X
1
+ 4.16X
2
+ 0.45X
3
+ 5.00X + 5.62X
1
2
2
ꢀ
0.01X ꢀ 0.87X
1
X
2
ꢀ 0.01X
1
X
3
+ 0.24X
2
X
3
(8)
3
2
2
Selectivity (\%)) = 0.41 + 0.09X
1
+ 0.25X
2
+ 0.01X
3
ꢀ 0.02X ꢀ 0.03X
1
2
2
ꢀ
0.01X ꢀ 0.01X
1
X
2
+ 0.01X
1
X
3
ꢀ 0.01X
2
X
3
(9)
3
1
.02% DPGG yield, 7.10 ± 0.79% hydrolysis of PG, 41.76 ± 1.74% PG
conversion (all expressed as mean ± SEM), and 0.79 ± 0.02 reaction
selectivity, while the experimental values were 33.02 ± 2.10%, 8.13 ±
in which X
1
is the substrate molar ratio (tripalmitin/PG), X
2
is the
4
is the
enzyme load relative to the total weight of substrates (%), and X
reaction time (h).
1
.79%, 44.83 ± 1.85%, 0.74 ± 0.05 (all expressed as mean ± SD) for the
actual yield, hydrolysis, conversion, and reaction selectivity, respec-
tively. The good agreements between the actual and predicted values
validated the model.
The ANOVA tests (Table 2) showed all four models were significant
at 99.99% probability levels, suggesting that the variation accounted for
by the model was significantly greater than the residual variation with
less than a 0.01% probability that the model F-value could occur owing
2
to the noise. The coefficient (R ) of the regression model indicated the
3
.8. Effect of acyl donor/acceptor, other lipases, and alcohol removal
level of response that could be explained by the proposed model. For
DPGG yield, 94.66% of the variability in the response could be explained
After the optimal reaction parameters were chosen, the effects of
2
(
R
= 0.9466). For PG hydrolysis, 82.64% of the variability in the
other reaction variables were analyzed under the selected condition. As
shown in Fig. 4d, using the same Lipozyme® 435, the yield and con-
version were significantly decreased when MG was used instead of PG.
2
response could be explained (R = 0.8264). For PG conversion, 90.99%
2
of the variability in the response could be explained (R = 0.9099). For
7

S. Zhang et al.
Food Chemistry xxx (xxxx) xxx
Fig. 4. Time-course analyses of the transesterification of propyl gallate or methyl gallate with tripalmitin or tricaprylin under different reaction conditions: (a) effect
of reaction time analyzed using three-way analysis of variance (ANOVA) with second-order interaction and expressed as least-squares means (LS mean) ± the
standard error of mean (SEM); (b) yield of 1,2-dipalmitoylgalloylglycerol (DPGG), hydrolysis of propyl gallate (PG), conversion of PG, and reaction selectivity when a
2
5 substrate molar ratio (tripalmitin/PG) was used and the enzyme load (relative to the weight of total substrates) was fixed at 25/1; (c) validation of the proposed
models under the optimal reaction conditions (25/1 tripalmitin/PG by mol, 120 h reaction time, and 25% enzyme load relative to the total weight of substrates); and
d) effect of acyl/acceptor on yield and conversion (reaction condition was fixed at 25/1 tripalmitin/PG substrate molar ratio and 25% enzyme load to the total
(
weight of substrates). Different letters indicate significant statistical difference at p-value < 0.05 in Tukey’s honest significant difference (HSD) tests.
Table 2
Analysis of variance (ANOVA) table for the predictive models pertaining to the yield of 1,2-dipalmitoylgalloylglycerol (DPGG), hydrolysis of propyl gallate (PG),
conversion of PG, and reaction selectivity during the transesterification of PG and tripalmitin using Lipozyme® 435 as the catalyst.
Response
Yield
Source
Degree of freedom
Sum of squares
Mean square
F-value
p-value
Model
Residual
Total
9
7700.92
434.14
855.66
6.12
139.94
< 0.0001*
71
80
8135.06
2
2
Root mean square error (RMSE) = 2.4728, R = 0.9466, adjusted R = 0.9399
Hydrolysis
Conversion
Selectivity
Model
Residual
Total
9
1246.39
261.79
138.49
3.69
37.56
79.64
< 0.0001*
< 0.0001*
< 0.0001*
71
80
1508.18
2
2
Root mean square error (RMSE) = 7.7701, R = 0.8264, adjusted R = 0.8044
Model
Residual
Total
9
12719.96
1259.95
1413.33
17.75
71
80
13979.91
2
2
Root mean square error (RMSE) = 4.2126, R = 0.9099, adjusted R = 0.8985
Model
Residual
Total
9
3.89
0.23
0.43
0.01
131.71
71
80
4.12
2
2
Root mean square error (RMSE) = 0.0573, R = 0.9435, adjusted R = 0.9363
*Indicates statistically significant at 95% of confidence level.
No DPGG formation was detected when GA was used as the acyl donor.
Turbid reaction mixtures were observed when using GA or MG instead of
PG. These results could be attributed to the combined effects of low
solubility/concentration, steric hindrance, and electrostatic interactions
between the substrates and enzyme.
of tripalmitin, tricaprylin was used as acyl acceptor and reaction me-
dium. The HPLC chromatogram of the reaction mixture is shown in
supplementary material (Fig. S11). Compared with tripalmitin, a higher
PG conversion and diacylgalloylglycerol (dioctanoylgalloylglycerol or
dicapryloylgalloylglycerol, DCGG) yield were obtained (Fig. 4d), which
could be attributed to the preference of CALB for saturated medium-
chain fatty acids (Karabulut, Durmaz, & Hayaloglu, 2009; Xiao et al.,
The effect of acyl acceptor was tested using tricaprylin in a similar
approach. With all the other reaction parameters kept the same, instead
8

S. Zhang et al.
Food Chemistry xxx (xxxx) xxx
2
015). DPGG and DCGG are different in the fatty acid chain length,
References
which may cause physical and chemical property changes. Further
research is needed regarding the property and application of DCGG.
Besides Lipozyme® 435, two other food-grade lipases, Lipozyme®
RM IM and Lipozyme® TL IM, were also evaluated using the pre-
established reaction condition. After 120 h, no DPGG was detected for
either of the lipases, indicating that these two lipases were not suitable
for catalyzing this reaction. Similar results were reported in previous
studies, CALB is generally considered to possess high transesterification
and esterification activities toward phenolic acids (Compton et al., 2000;
Sun et al., 2013; Zhang & Akoh, 2019).
Amadasi, A., Mozzarelli, A., Meda, C., Maggi, A., & Cozzini, P. (2009). Identification of
xenoestrogens in food additives by an integrated in silico and in vitro approach.
Chemical Research in Toxicology, 22(1), 52–63. https://doi.org/10.1021/tx800048m.
Bittenbender, C. D. (1950). Gallic acid ester antioxidant for fish oils. Commercial Fisheries
Review, 12(2), 1–18.
Boles, J. S., Crerar, D. A., Grissom, G., & Key, T. C. (1988). Aqueous thermal degradation
of gallic acid. Geochimica et Cosmochimica Acta, 52(2), 341–344. https://doi.org/
10.1016/0016-7037(88)90089-0.
Bornscheuer, U. T. (2002). Microbial carboxyl esterases: Classification, properties and
application in biocatalysis. FEMS Microbiology Review, 26(1), 73–81. https://doi.org/
1
0.1111/j.1574-6976.2002.tb00599.x.
Coates, J. (2000). Interpretation of infrared spectra, a practical approach. In R. A. Meyers
(Ed.), Encyclopedia of analytical chemistry (pp. 10815–10837). Hoboken, NJ: John
Wiley and Sons.
Compton, D. L., Laszlo, J. A., & Berhow, M. A. (2000). Lipase-catalyzed synthesis of
ferulate esters. Journal of the American Oil Chemists’ Society, 77(5), 513–519. https://
doi.org/10.1007/s11746-000-0082-9.
In previous reports, removal of the produced alcohol showed an
encouraging effect on the conversion of acyl donor (Sun et al., 2017,
2
014). Conversely, no significant difference was observed between
closed and open systems (Fig. 4.d). n-Propyl alcohol was not the by-
product of the transesterification. Removing it did not affect the yield
of DPGG. However, n-propyl alcohol was the by-product of side re-
actions, and an increase in PG conversion and decrease in reaction
selectivity were expected, but not observed. The relative high boiling
Compton, D. L., Laszlo, J. A., & Evans, K. O. (2012). Antioxidant properties of feruloyl
glycerol derivatives. Industrial Crops and Products, 36(1), 217–221. https://doi.org/
1
0.1016/j.indcrop.2011.09.009.
Crontn, E. (1980). Lipstick dermatitis due to propyl gallate. Contact Dermatitis, 6(3),
13-214.
2
Fojan, P., Jonson, P. H., Petersen, M. T. N., & Petersen, S. B. (2000). What distinguishes
an esterase from a lipase: A novel structural approach. Biochimie, 82(11),
1033–1041.
◦
point of n-propyl alcohol (97 C) might limit its evaporation under at-
mospheric pressure, thus, no effect was observed. Considering the
possible oxidation of PG, a closed system should be used.
Fotoran, W. L., Müntefering, T., Kleiber, N., Miranda, B. N. M., Liebau, E., Irvine, D. J., &
Wunderlich, G. (2019). A multilamellar nanoliposome stabilized by interlayer
hydrogen bonds increases antimalarial drug efficacy. Nanomedicine: Nanotechnology,
Biology and Medicine, 22, Article 102099.
4
. Conclusions
Ghasemzadeh, A., & Ghasemzadeh, N. (2011). Flavonoids and phenolic acids: Role and
biochemical activity in plants and human. Journal of Medicinal Plants Research, 5(31),
A novel solvent-free enzymatic method for synthesizing a lip-
6
697–6703.
ophilized GA derivative, DPGG, was established. DPGG was synthesized
via the transesterification of PG and tripalmitin, using an immobilized
and commercially available food-grade CALB, Lipozyme® 435, as the
biocatalyst. Under the optimal condition, the selective synthesis pro-
duced DPGG at a 33.02% yield with reaction selectivity at 0.74. The
hydrolysis of PG was identified as the major side reaction. Compared
with the chemical synthesis methods, the established enzymatic method
is simple and does not require any harmful chemical reagent, making it
more suitable for use in foods, cosmetics, and pharmaceuticals.
Kang, D. H., Jung, H.-S., Ahn, N., Yang, S. M., Seo, S., Suh, K.-Y., et al. (2014). Janus-
Compartmental Alginate Microbeads Having Polydiacetylene Liposomes and
Magnetic Nanoparticles for Visual Lead(II) Detection. ACS Applied Materials &
Interfaces, 6(13), 10631–10637. https://doi.org/10.1021/am502319m.
Karabulut, I., Durmaz, G., & Hayaloglu, A. A. (2009). Fatty acid selectivity of lipases
during acidolysis reaction between oleic acid and monoacid triacylglycerols. Journal
of Agriculture and Food Chemistry, 57(21), 10466–10470. https://doi.org/10.1021/
jf902816e.
Kikuzaki, H., Hisamoto, M., Hirose, K., Akiyama, K., & Taniguchi, H. (2002). Antioxidant
properties of ferulic acid and its related compounds. Journal of Agriculture and Food
Chemistry, 50(7), 2161–2168. https://doi.org/10.1021/jf011348w.
Lee, O.-H., Lee, B.-Y., Lee, J., Lee, H.-B., Son, J.-Y., Park, C.-S., Shetty, K., & Kim, Y.-C.
(
2009). Assessment of phenolics-enriched extract and fractions of olive leaves and
CRediT authorship contribution statement
their antioxidant activities. Bioresource Technology, 100(23), 6107–6113. https://doi.
org/10.1016/j.biortech.2009.06.059.
Matsuo, T., Kobayashi, T., Kimura, Y., Tsuchiyama, M., Oh, T., Sakamoto, T., &
Adachi, S. (2008). Synthesis of glyceryl ferulate by immobilized ferulic acid esterase.
Biotechnology Letters, 30(12), 2151–2156. https://doi.org/10.1007/s10529-008-
Siyu Zhang: Conceptualization, Methodology, Formal analysis,
Investigation, Data curation, Writing - original draft. Joseph R. Hyatt:
Investigation, Writing - review & editing. Casimir C. Akoh: Conceptu-
alization, Methodology, Supervision, Writing - review & editing, Fund-
ing acquisition.
9
814-2.
Ozgen, M., Reese, R. N., Tulio, A. Z., Scheerens, J. C., & Miller, A. R. (2006). Modified
,2-azino-bis-3-ethylbenzothiazoline-6-sulfonic acid (ABTS) method to Measure
2
antioxidant capacity of selected small fruits and comparison to ferric reducing
antioxidant power (FRAP) and 2,2‘-diphenyl-1-picrylhydrazyl (DPPH) methods.
Journal of Agricultural and Food Chemistry, 54(4), 1151–1157.
Declaration of Competing Interest
Pollastri, M. P., Porter, N. A., McIntosh, T. J., & Simon, S. A. (2000). Synthesis, structure,
and thermal properties of 1,2-dipalmitoylgalloylglycerol (DPGG), a novel self-
adhering lipid. Chemistry and Physics of Lipids, 104(1), 67–74. https://doi.org/
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
1
0.1016/S0009-3084(99)00110-3.
Raki ´c , T., Kasagi ´c -Vujanovi ´c , I., Jovanovi ´c , M., Jan ˇc i ´c -Stojanovi ´c , B., & Ivanovi ´c , D.
2014). Comparison of full factorial design, central composite design, and box-
(
behnken design in chromatographic method development for the determination of
fluconazole and its impurities. Analytical Letters, 47(8), 1334–1347. https://doi.org/
Acknowledgement
1
0.1080/00032719.2013.867503.
Rice-Evans, C. A., Miller, N. J., & Paganga, G. (1996). Structure-antioxidant activity
This work is supported by Enzymatic Modification of Lipids [project
accession no. 1024219] from the USDA National Institute of Food and
Agriculture. The authors would like to thank Food Science Research,
University of Georgia for partially supporting this work. Special thanks
to Dr. Dennis R. Phillips and Dr. Chau-Wen Chou from the Proteomics
and Mass Spectrometry Facility, University of Georgia, for assistance
with MS analysis, and Dr. John Glushka from the Complex Carbohydrate
Research Center NMR spectroscopy facility, University of Georgia, for
assisting with NMR analysis.
relationships of flavonoids and phenolic acids. Free Radical Biology and Medicine, 20
(
7), 933–956. https://doi.org/10.1016/0891-5849(95)02227-9.
Sabally, K., Karboune, S., St-Louis, R., & Kermasha, S. (2006). Lipase-catalyzed
transesterification of dihydrocaffeic acid with flaxseed oil for the synthesis of
phenolic lipids. Journal of Biotechnology, 127(1), 167–176. https://doi.org/10.1016/
j.jbiotec.2006.06.010.
Sabally, K., Karboune, S., St-Louis, R., & Kermasha, S. (2007). Lipase-catalyzed synthesis
of phenolic lipids from fish liver oil and dihydrocaffeic acid. Biocatalysis and
Biotransformation, 25(2-4), 211–218. https://doi.org/10.1080/
1
0242420701379916.
Santos, N. A., Cordeiro, A. M., Damasceno, S. S., Aguiar, R. T., Rosenhaim, R., Carvalho
Filho, J. R., et al. (2012). Commercial antioxidants and thermal stability evaluations.
Fuel, 97, 638–643. https://doi.org/10.1016/j.fuel.2012.01.074.
Appendix A. Supplementary data
Schmidt, R., Carrigan, J. G., & DeWolf, C. E. (2006). Synthesis and characterization of a
series of novel phenol- and polyphenol-based glycerolipids. Canadian Journal of
Chemistry, 84(10), 1411–1415. https://doi.org/10.1139/v06-102.
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.foodchem.2020.128604.
9

S. Zhang et al.
Food Chemistry xxx (xxxx) xxx
Schmidt, R., & DeWolf, C. E. (2004). Monolayer Behavior of 1,2-Dipalmitoylgalloylgly-
Sun, S., Zhu, S., & Bi, Y. (2014). Solvent-free enzymatic synthesis of feruloylated
structured lipids by the transesterification of ethyl ferulate with castor oil. Food
Chemistry, 158, 292–295. https://doi.org/10.1016/j.foodchem.2014.02.146.
Takasago, M., Horikawa, K., & Masuyama, S. (1976). Studies on derivatives of aromatic
hydroxycarboxylic acids as antioxidant. I. J. Jpn. Oil. Chem. Soc., 25(1), 16–20.
https://doi.org/10.5650/jos1956.25.16.
Wang, J., & Shahidi, F. (2014). Acidolysis of p-coumaric acid with omega-3 oils and
antioxidant activity of phenolipid products in in vitro and biological model systems.
Journal of Agriculture and Food Chemistry, 62(2), 454–461. https://doi.org/10.1021/
jf404140v.
Xiao, Z., Hou, X., Lyu, X., Zhao, J.-y., Xi, L., Li, J., & Lu, J. R. (2015). Enzymatic synthesis
of aroma acetoin fatty acid esters by immobilized Candida antarctica lipase B.
Biotechnology Letters, 37(8), 1671–1677. https://doi.org/10.1007/s10529-015-1834-
0.
Zeng, Y., Yin, X., Wu, M.-C., Yu, T., Feng, F., Zhu, T.-D., & Pang, Q.-F. (2014). Expression
of a novel feruloyl esterase from Aspergillus oryzae in Pichia pastoris with
esterification activity. Journal of Molecular Catalysis B: Enzymatic, 110, 140–146.
https://doi.org/10.1016/j.molcatb.2014.10.002.
cerol, a Synthetic Lipid with Strong Cohesive Properties. Langmuir, 20(8),
3
284–3288. https://doi.org/10.1021/la036099f.
Song, Z., & Xiao, Z. (1988). The preparation of 1-glycerol gallate and study on its
antioxidation properties. Chemistry and Industry of Forest Products, 3, 001.
Stamatis, H., Sereti, V., & Kolisis, F. N. (1999). Studies on the enzymatic synthesis of
lipophilic derivatives of natural antioxidants. Journal of the American Oil Chemists’
Society, 76(12), 1505. https://doi.org/10.1007/s11746-999-0193-1.
Sun, S., & Hu, B. (2017). Enzymatic preparation of novel caffeoyl structured lipids using
monoacylglycerols as caffeoyl acceptor and transesterification mechanism.
Biochemical Engineering Journal, 124, 78–87. https://doi.org/10.1016/j.
bej.2017.05.002.
Sun, S.-D., Shan, L., Liu, Y.-F., Jin, Q.-Z., Zhang, L.-X., & Wang, X.-G. (2008). Solvent-free
enzymatic preparation of feruloylated monoacylglycerols optimized by response
surface methodology. Journal of Agriculture and Food Chemistry, 56(2), 442–447.
https://doi.org/10.1021/jf0728911.
Sun, S., Song, F., Bi, Y., Yang, G., & Liu, W. (2013). Solvent-free enzymatic
transesterification of ethyl ferulate and monostearin: Optimized by response surface
methodology. Journal of Biotechnology, 164(2), 340–345. https://doi.org/10.1016/j.
jbiotec.2013.01.013.
Zhang, S., & Akoh, C. C. (2019). Solvent-free enzymatic synthesis of 1-o-galloylglycerol
optimized by the Taguchi method. Journal of the American Oil Chemists Society, 96(8),
877–889. https://doi.org/10.1002/aocs.12229.
Sun, S., Wang, P., & Zhu, S. (2017). Enzymatic incorporation of caffeoyl into castor oil to
prepare the novel castor oil-based caffeoyl structured lipids. Journal of Biotechnology,
Zhang, S., & Akoh, C. C. (2020). Enzymatic synthesis of 1-o-galloylglycerol:
Characterization and determination of its antioxidant properties. Food Chemistry,
2
49, 66–72. https://doi.org/10.1016/j.jbiotec.2017.03.022.
3
05, 125479. https://doi.org/10.1016/j.foodchem.2019.125479.
1
0