Lipase-catalyzed synthesis of ferulate esters

Article abstract of DOI:10.1007/s11746-000-0082-9

Four commercially available esterases were screened for their ability to esterify ferulic acid (4-hydroxy-3-methoxy cinnamate). Novozym 435 was found to be the only one of those screened to convert ferulic acid to ethyl and octyl ferulate at 20 and 14% yields, respectively. The highest percentage conversion was obtained using a 1:1 mole ratio of alcohol to ferulic acid in stirred batch reactions in anhydrous 2-methyl-2-propanol at 60 °C using one equivalent (wt/wt based on ferulic acid) of Novozym 435. Increased water content and a higher alcohol/ethyl ferulate ratio had adverse effects on the lipase-catalyzed esterification. The Novozym 435 activity was tested in less polar solvents (anhydrous toluene and hexane) by monitoring the alcoholysis of ethyl ferulate with 1-octanol, which resulted in a 50% yield of octyl ferulate. The alcoholysis was improved to 83% by applying a 16 mm Hg vacuum for 5 min every 24 h to remove the ethanol co-product. The optimal alcoholysis parameters were applied to the alcoholysis of ethyl ferulate with monoolein and the transesterification with triolein. The transesterification of ethyl ferulate with triolein in anhydrous toluene produced a combined 44% yield of ferulyl monoolein and ferulyl diolein, a 20% greater yield than that obtained for alcoholysis using monoolein. The highest yield, 77%, of ferulyl monoolein and ferulyl diolein was achieved using a threefold excess of neat triolein. The lipase-catalyzed transesterification of ethyl ferulate with triolein appears to be a technically feasible route to ferulyl-substituted acylglycerols, which are potentially useful sunscreen ingredients.

Full text of DOI:10.1007/s11746-000-0082-9

Lipase-Catalyzed Synthesis of Ferulate Esters
a,
a
b
David L. Compton *, Joseph A. Laszlo , and Mark A. Berhow
a
b
Biomaterials Processing Research Unit, and Bioactive Agents Research Unit,
USDA, ARS, NCAUR, Peoria, Illinois 61604
ABSTRACT: Four commercially available esterases were atures (200–250°C) (3). Enzymatic interesterifications, how-
screened for their ability to esterify ferulic acid (4-hydroxy-3- ever, offer the advantages of milder reaction conditions, a
®
methoxy cinnamate). Novozym 435 was found to be the only
wider variety of synthetic substrates, and regioselective speci-
ficity toward the acyl groups of the TAG (4). The industrial
uses of lipase-catalyzed reactions are not limited to nutritional
one of those screened to convert ferulic acid to ethyl and octyl
ferulate at 20 and 14% yields, respectively. The highest percent-
age conversion was obtained using a 1:1 mole ratio of alcohol
and pharmaceutical applications, but also include the synthe-
to ferulic acid in stirred batch reactions in anhydrous 2-methyl-
sis of soaps and surfactants, improvement in lumber process-
2
-propanol at 60°C using one equivalent (wt/wt based on fer-
ing, leather manufacturing, and flavor production in the dairy
and related industries (3,4). This work examines the technical
feasibility of using lipase-catalyzed reactions to synthesize
ulic acid) of Novozym 435. Increased water content and a
higher alcohol/ethyl ferulate ratio had adverse effects on the li-
pase-catalyzed esterification. The Novozym 435 activity was
tested in less polar solvents (anhydrous toluene and hexane) by structured lipids with properties suitable for use as possible
monitoring the alcoholysis of ethyl ferulate with 1-octanol, sunscreen reagents.
which resulted in a 50% yield of octyl ferulate. The alcoholysis
was improved to 83% by applying a 16 mm Hg vacuum for 5
min every 24 h to remove the ethanol co-product. The optimal
alcoholysis parameters were applied to the alcoholysis of ethyl
ferulate with monoolein and the transesterification with triolein.
The transesterification of ethyl ferulate with triolein in anhy-
drous toluene produced a combined 44% yield of ferulyl
monoolein and ferulyl diolein, a 20% greater yield than that ob-
tained for alcoholysis using monoolein. The highest yield, 77%,
of ferulyl monoolein and ferulyl diolein was achieved using a
Sunscreens function either as ultraviolet (UV) filters or
blocks. UV blocks, such as TiO, ZnO, and derivatives of other
metal oxides, form a physical barrier that scatters UV light
(
5). These UV blocks offer the most comprehensive sun-
screen protection, blocking the full spectrum of UVA
400–320 nm) and UVB (320–290 nm) light. The particulate
(
nature of these formulations, however, often leaves notice-
able residues when applied to the skin, which are cosmeti-
cally unacceptable to the consumer. The most commonly used
threefold excess of neat triolein. The lipase-catalyzed transes- sunscreens are UV filters, which are typically organic com-
terification of ethyl ferulate with triolein appears to be a techni- pounds incorporated (2–15%) into topical formulations (6,7).
cally feasible route to ferulyl-substituted acylglycerols, which A disadvantage of UV filters is that each organic compound
are potentially useful sunscreen ingredients.
has a limited range of maximal UV absorptivity, rendering
each reagent better suited for either UVA or UVB protection,
Paper no. J9361 in JAOCS 77, 513–519 (May 2000).
but not both. The advantage of the UV-filtering molecules,
KEY WORDS: Alcoholysis, cinnamates, ethyl ferulate, lipase,
however, is that they can be engineered to provide sunscreens
®
monoolein, Novozym 435, transesterification, triacylglycerols,
with desirable physical appearance, solubility, and water-re-
triolein.
sistant properties (7).
Although no longer used today, benzyl cinnamate formu-
There is a growing interest in modifying fats and oils to form lated as an emulsion with benzyl salicylate was used as a sun-
structured lipids with specific properties for nutritional and screen as early as 1928 (6). Today, cinnamic acid derivatives
pharmaceutical applications. Recent reviews have outlined are the most widely used UVB-absorbing chemicals in sun-
the strategies for synthesizing tailor-made fats and oils and screen formulations, with four derivatives approved for use
their desired properties (1,2). These strategies have included in the United States and 17 approved for use in Europe (6).
blending, distillation, fractionation, hydrogenation, interester- The cinnamates’ unsaturated C=C bond adjacent to the aro-
ification with chemical catalysts, and, more recently, inter- matic ring allows for a continuous, conjugated π-system
esterification with biocatalysts. Chemical interesterifications throughout the molecule. An electron can be delocalized
of triacylglycerols (TAG) for industrial applications are typi- throughout the π-system by photoexcitation with energy cor-
cally performed using inorganic catalysts at elevated temper- responding to ~305 nm. Most common cinnamic acids and
short-chain esters are water soluble, limiting their usefulness
as waterproof sunscreens. Cinnamic acid derivatives there-
fore have been designed with long-chain hydrocarbons (e.g.,
*
1
To whom correspondence should be addressed at USDA, ARS, NCAUR,
815 N. University St., Peoria, IL 61604.
E-mail: comptond@mail.ncaur.usda.gov
octyl-p-methoxy cinnamate), which renders them water-in-
Copyright © 2000 by AOCS Press
513
JAOCS, Vol. 77, no. 5 (2000)

5
14
D.L. COMPTON ET AL.
soluble and suitable for waterproof sunscreens. The –OCH3 to stir at 25, 40, or 60°C, and 10-µL aliquots were periodi-
group of octyl-p-methoxy cinnamate acts as an electron-re- cally taken for analysis.
leasing group to improve the electron excitation process (6).
Alcoholysis of EF with 1-octanol. In a 25-mL Schlenk
In this work, we report a lipase-catalyzed route to ferulyl- tube, EF (111.1 mg, 0.5 mmol) was dissolved in 5 mL of t-
substituted TAG. Ethyl ferulate (EF), the ethyl ester of ferulic butanol or 5 mL of toluene at 60°C. 1-Octanol (78 µL, 0.5
acid (FA; 4-hydroxy-3-methoxy cinnamate) that is widely mol) was added against a nitrogen flow followed by the en-
distributed in plants, possesses a UV absorbance maximum at zyme (110 mg). The reactions were allowed to stir at 60°C,
~
320 nm. Several lipases were screened to determine their ac- and 10-µL aliquots were periodically taken for analysis.
tivity toward the ferulyl compounds. The most active lipase
Alcoholysis of EF with MO. MO (48 mg, 0.14 mmol) was
was used to facilitate the transesterification of EF with triolein transferred from the reagent bottle by dissolving it in 2 mL of
TO). TO is a water-insoluble TAG. When combined, the fer- hexane and pipetting the clear colorless solution into a 25-mL
(
ulyl-substituted oleins provide the UV absorptivity of a cin- Schlenk tube. The reagent bottle was washed with an addi-
namate ester with the water-insoluble properties of a lipid. tional 2 mL of hexane, which also was added to the Schlenk
The lipase-catalyzed transesterifications of EF with TO could tube. The hexane was removed from the Schlenk tube by vac-
potentially lead to new sunscreen products, synthesized from uum, leaving a clear, colorless oil that was kept in vacuo for
natural products, while providing a value-added use for veg- an additional 15 min. EF (30 mg, 0.14 mmol) was added to
etable oils.
the Schlenk tube, and the stirred contents were dissolved in 5
mL of toluene at 60°C. Novozym 435 (100 mg) was added,
and 100-µL aliquots were periodically taken for analysis.
Transesterification of EF with TO. TO (447 mg, 0.5 mmol)
EXPERIMENTAL PROCEDURES
®
Enzymes. Novozym 435 (Candida antarctica) with a 1–2% was dissolved in 5 mL of toluene in a 25-mL Schlenk tube at
®
(w/w) water content and Lipozyme IM (Mucor miehei) were 60°C. EF (111 mg, 0.5 mmol) was added to the reaction mix-
®
from Novo Nordisk (Franklinton, NC). Chirazyme L-1 ture followed by Novozym 435 (110 mg). The suspension
(
(
porcine pancreas) was from Roche Molecular Biochemicals was stirred at 60°C, and 100-µL aliquots were periodically
Indianapolis, IN ). α-Chymotrypsin (bovine pancreas) at- taken for analysis.
tached to macroporous acrylic beads was purchased from
Sigma-Aldrich (St. Louis, MO). All enzymes were used as mg, 0.5 mmol) was dissolved in 1.24 g of TO (1.4 mmol) at
obtained from the supplier without modification.
60°C to give a clear, colorless, and viscous solution.
The reaction also was performed without solvent. EF (111
Chemicals. FA (Sigma-Aldrich) and ethyl 4-hydroxy-3- Novozym 435 (110 mg) was added, and 20-µL aliquots were
methoxy cinnamate (EF, Sigma-Aldrich) were kept desic- periodically taken for analysis.
cated under a static vacuum. Monoolein (MO), diolein (DO),
and oleic acid (OA) were from Sigma-Aldrich and stored at Separation Products (San Jose, CA) HPLC system consisting
4°C. TO and ethyl oleate (EO) were from Nu-Chek-Prep of a Spectra System AS3000 autosampler, a Spectra System
HPLC analysis. Analyses were performed using a Thermo
–
(Elysian, MN) and stored at –4°C.
P4000 pump, a Spectra System UV6000LP detector, an All-
Anhydrous (<0.005% water) toluene and t-butanol (2-methyl- tech (Deerfield, IL) 500 Evaporative Light-Scattering Detec-
®
2
-propanol) were purchased in Sure/Seal bottles (Sigma- tor (ELSD), and a Prodigy C column (5 µm, 250 × 4.6 mm;
8
Aldrich). Ethanol and 1-octanol were dried over activated mo- Phenomenex, Torrance, CA). Solvents were filtered using
lecular sieves and stored under a nitrogen atmosphere. Methanol, Whatman 0.45-µm nylon membrane filters (Sigma-Aldrich)
acetone, acetonitrile, glacial acetic acid, and 1-butanol were and degassed using a Thermo Separation Products SCM 1000
high-performance liquid chromatography (HPLC) grade. All Membrane Degasser.
were obtained from Sigma-Aldrich. Water used for reactions was
For determination of FA and ferulate esters, the 10-µL
deionized and filtered through a 0.22 µm nylon filter.
aliquots taken from the esterification and alcoholysis reac-
General considerations. Reactions were performed under a tions were diluted to 2 mL with 50% aqueous ethanol and fil-
nitrogen atmosphere using standard Schlenk line techniques in tered through 0.45 µm Gelman Acrodisc 13LC PVDF syringe
25-mL Schlenk tubes (1.5 × 15 cm) containing a Spinplus stir- filters (Sigma-Aldrich). Injections of 10 µL were eluted from
bar (3/8 × 3/8 in.) from Sigma-Aldrich. The reaction tempera- the column at a flow rate of 1.5 mL/min using a binary gradi-
ture was controlled using a water-jacketed oil bath heated/ ent of solvent A [263 mL water, 1 mL glacial acetic acid, 11
cooled by an external circulating water bath (Precision Scien- mL 1-butanol (3%), 70 mL methanol (20%)] and solvent B
tific Inc., Chicago, IL). The oil bath was mounted over a mag- [93 mL water, 1 mL glacial acetic acid, 11 mL 1-butanol
netic stir plate. The quantities of enzyme added to reactions rep- (3%), and 245 mL methanol (70%)]. The elution sequence
resent the combined mass of the enzyme and its support.
was 5 min isocratic flow of 75:25 (A/B), 2 min linear gradi-
Esterification of FA. In a 25-mL Schlenk tube, FA (97 mg, ent to 100% (B), 18 min isocratic flow of 100% (B), 2 min
.5 mmol) was dissolved in 5 mL of t-butanol at 25, 40, or linear gradient to 75:25 (A/B), followed by 3 min of isocratic
0°C. Typically, ethanol (29 µL, 0.5 mmol) or 1-octanol (78 elution with 75/25 (A/B). The eluate was monitored at 300
0
6
µL, 0.5 mmol) was added against a nitrogen flow followed by nm using the Spectra System UV6000LP detector.
the addition of enzyme (100 mg). The reactions were allowed
For determination of TO and its derivatives, the 100- and
JAOCS, Vol. 77, no. 5 (2000)

LIPASE-CATALYZED SYNTHESIS OF FERULATE ESTERS
515
2
0-µL aliquots taken from the glycerolysis reactions were di- in t-butanol. The FA to EF conversion was 1, 5, and 20% at
luted to 1 mL with acetone and filtered through 0.45 µm Gel- 25, 40, and 60°C, respectively, after 312 h. Figure 1 shows
man Acrodisc 13LC PVDF syringe filters. Injections of 10 µL the effect of reaction conditions on the time course of the con-
were eluted from the column at a flow rate of 1.5 mL/min version of FA to EF or OF at 60°C in stirred batch reactions
using an isocratic flow of 40:60 (vol/vol) acetone (containing under nitrogen. At a 0.1 M concentration of substrates, equi-
1
% glacial acetic acid)/acetonitrile. The eluate was monitored librium was not reached after 312 h, but the extent of conver-
at 360 nm using the Spectra System UV6000LP detector fol- sion was greater for EF than OF throughout the time course
lowed by the Alltech 500 ELSD (nitrogen; 2.0 slpm, 70°C). of the reaction. The equilibria of lipase-catalyzed esterifica-
The detectors were plumbed in series with a void volume re- tions have been observed by Habulin et al. (11) to be influ-
sulting in a 0.1 s delay in peak retention time for species de- enced by the alcohol molar mass; For example, butanol pro-
tected by both methods. It should be noted that the ELSD re- vided a higher equilibrium conversion of OA to ester than dec-
sponse is not linear and that peak intensities and areas are anol. Decreasing the concentration of reactants 10-fold to
species dependent; therefore, the tabulated data are intended 0.01 M ethanol and FA allowed equilibrium to be achieved
for qualitative comparisons only.
within 312 h, but resulted in a lower FA conversion (8%). The
HPLC–atmospheric pressure chemical ionization–mass conversion of FA to EF after 312 h was reduced to 10 and 4%
spectroscopy (HPLC–APCI–MS) analysis. HPLC–APCI–MS when water was added to the reaction at concentrations of 0.5
analyses were performed on samples prepared for HPLC and 1.0% (vol/vol), respectively (Fig. 1). This decrease in FA
analysis. Elution conditions were identical to those described conversion most likely results from a shift in equilibrium to-
previously except glycerolysis reaction samples were eluted ward hydrolysis rather than any adverse effects the water may
using an isocratic flow of 40:60 (vol/vol) methanol (contain- have on the enzyme (12,13). Doubling the ethanol concentra-
ing 1% glacial acetic acid)/acetonitrile. Mass spectral analy- tion of the reaction resulted in a slightly lower conversion to
ses were performed using a Finnigan (San Jose, CA) MAT EF, 18%, compared to the conversion obtained from the reac-
LCQ mass spectrometer with a direct liquid APCI interface.
tion performed with a 1:1 alcohol to FA ratio, 20% (data not
shown). A lack of mass-action response has been observed
previously for the Novozym 435-catalyzed esterification of
glucuronic acid with cinnamic alcohol where excess alcohol
RESULTS AND DISCUSSION
Esterification of FA. The esterification of FA was studied as a did not improve the ester yield (14). The best conversion of
model reaction to optimize conditions for subsequent experi- FA to EF was obtained using a 1:1 molar ratio of 0.1 M alco-
ments. Lipozyme IM (M. miehei), Chirazyme L-1 (porcine hol to 0.1 M FA at 60°C; therefore, these reaction conditions
pancreas), Novozym 435 (C. antarctica), and α-chy- were used for subsequent experiments.
motrypsin (bovine pancreas) were screened for their ability
to esterify FA (Scheme 1) to produce EF or n-octyl ferulate exhibit higher activities in nonpolar solvents; however, the
OF). Novozym 435 was found to be the only enzyme to cat-
Alcoholysis of EF. As discussed above, lipases typically
(
alyze the esterification of FA with ethanol or 1-octanol in t-
butanol. Autoclaved samples of Novozym 435 did not facili-
tate the esterification of FA; therefore, the catalytic activity is
attributed to the enzyme and not its acrylic support. The es-
terification of FA was not detected in reactions performed in
hexane and toluene. Hydrophobic solvents generally support
higher enzymatic activity (8–10); however, the solubility of
FA was limited in these nonpolar solvents. Therefore, the es-
terifications were performed at 25, 40, and 60°C (manufac-
turer’s suggested maximum temperature for Novozym 435)
FIG. 1. Esterification of 0.1 M ferulic acid with 0.1 M ethanol or 0.1 M
®
1
●
-octanol in t-butanol catalyzed by Novozym 435 (100 mg) at 60°C.
: ethyl ferulate; ●: octyl ferulate; ●: ethyl ferulate (0.5% water added
to reaction medium), ●: ethyl ferulate (1% water added to reaction
medium); ●: ethyl ferulate (reaction performed with 0.01 M ferulic acid
and 0.01 M ethanol). Percentage yields were calculated based on total
peak area of substrates and products detected by high-performance liq-
uid chromatography analysis.
SCHEME 1
JAOCS, Vol. 77, no. 5 (2000)

5
16
D.L. COMPTON ET AL.
poor solubility of FA in hexane and toluene limited the study action medium refluxed slightly at 16 mm Hg and 60°C but
of its esterifications to using t-butanol as a solvent. The abil- never boiled. Figure 2 shows that a 30% increase in yield
ity of Novozym 435 to effect the transesterification of ferulyl (~80% final yield) of OF was achieved after 312 h when the
substrates in less-polar solvents was investigated using EF. vacuum was applied. An alternative method used to improve
The alcoholysis of EF with 1-octanol (1:1 molar ratio) was the yield of OF was to increase the amount of enzyme used.
performed in stirred batch reactions in anhydrous toluene at Doubling the amount of Novozym 435 increased the yield of
60°C. Figure 2 shows that a 50% conversion of EF to OF was OF to 75% after 312 h (Fig. 2). Considering the cost of the
achieved after 312 h. The water concentration in the reac- enzyme, however, the application of a periodic vacuum to in-
tions, based on the water content of Novozym 435 (maximum crease the yield of OF is preferable.
of 2% by wt) and anhydrous toluene (0.005% by vol) as ob-
Transesterification of EF. The goal of this research was to
tained from the manufacturers, was 0.027 M (EF concentra- synthesize structured lipids containing one or more substi-
tion was 0.10 M). Under these conditions the unwanted hy- tuted ferulyl moieties for possible use as ingredients in sun-
drolysis of EF to FA was observed to be <1%, and no effort screen formulations. The results discussed above provide the
was made to study enzyme performance at lower water con- optimal parameters for the Novozym 435-catalyzed alcoholy-
centrations.
sis of EF. Experiments involved the reaction of EF with TO
A few factors may contribute to the higher conversion (1:1 mol ratio) catalyzed by 1 equivalent of Novozym 435
50%) of EF to OF compared to the 14% conversion of FA to (w/w based on EF) in 5 mL of anhydrous toluene at 60°C. The
(
OF. First, the more hydrophilic solvent, t-butanol, used in the reaction reached equilibrium after 72 h. The chromatogram
FA esterification may be more likely to remove essential obtained from the HPLC analysis of the reaction shows nine
water from the enzyme (9), lowering its activity. The addition peaks detected by ELSD. The peaks are illustrated in Figure
of water to this reaction promotes hydrolysis and does not re- 3, which depicts the HPLC analysis of the transesterification
hydrate the enzyme. Also, hydrophobic solvents are less performed in neat TO (discussed below). Table 1 lists the
likely to interact with the protein structure, thus preventing peak assignments, corresponding retention times (R ), and rel-
t
denaturation of the enzyme (9). Finally, the reaction is con- ative yields of the reaction products. The peaks correspond-
ducted at 60°C under a positive pressure of nitrogen (830 mm ing to EF, MO, OA, EO, DO, and TO were identified based
Hg). This may allow the ethanol formed as the alcoholysis co- on retention times determined from standards. This left four
product to be driven from the reaction when the rubber stop- unidentified UV-absorbing (UVA) species with retention
per is removed during sampling, thus allowing the equilib- times of 1.8, 2.4, 2.5, and 4.4 min.
rium to favor OF. To see if the removal of ethanol from the
HPLC–APCI–MS was used to determine that these four
reaction medium would increase the production of OF, the reaction products were MO or DO derivatives containing fer-
same experiment was performed while applying a partial vac- ulyl moieties. Scheme 2 shows the structures of the reactants
uum (16 mm Hg) to the reaction for 5 min every 24 h. The re- and the UVA products of interest. The major ion correspond-
ing to the peak at R = 1.8 min was 251 m/z and is ferulyl glyc-
t
+
erol (FG) with a major ion value of [M + H − H O] . An ion
2
FIG. 2. Alcoholysis of 0.1 M ethyl ferulate with 0.1 M 1-octanol in an-
hydrous toluene catalyzed by Novozym 435 (110 mg). ●: octyl feru-
late; ● octyl ferulate (16 torr vacuum applied to reaction for 5 min every
2
4
4 h); ●: octyl ferulate (reaction performed with 220 mg of Novozym
35). Percentage yields were calculated based on total peak area of sub-
FIG. 3. High-performance liquid chromatography analysis of the
Novozym® 435 (110 mg)-catalyzed transesterification of ethyl ferulate
(0.5 mmol) in neat triolein (1.4 mmol) after 144 h recorded using the
evaporative light-scattering detector.
strates and products detected by high-performance liquid chromatogra-
phy analysis.
JAOCS, Vol. 77, no. 5 (2000)

LIPASE-CATALYZED SYNTHESIS OF FERULATE ESTERS
517
TABLE 1
Peak Assignments and Relative Yields for the Alcoholysis and Transesterification Products of Ethyl Ferulate (EF)
Relative peak areasa
Rt (min)
Peak
Major ion (m/z)b
MOc
TOc
TO (neat)d
1
2
2
2
2
3
4
4
1
.8
.0
.4, 2.5
.6
.8
.2
.4
.7, 4.9
5.2
FG
EF
251
—
515
—
—
—
796
—
—
2.6 (44%)e
1.0 (32%)
2.4 (24%)
2.2
—
0.4
—(0%)
1.0
—
0.4 (19%)
1.0 (37%)
1.3 (32%)
—
0.2
1.3
0.6 (12%)
1.2
4.1
— (10%)
1.0 (13%)
3.6 (29%)
—
2.5
4.6
8.1 (48%)
9.0
40.2
FMO
MO
OA
EO
FDO
DO
TO
aPeaks were obtained using the evaporative light-scattering detector, and the areas are relative for each reaction based on
EF.
b
The major ions detected by high-performance liquid chromatography–atmospheric pressure chemical ionization–mass
+
+
spectroscopy were identified as follows: ferulylglycerol (FG), [M + H − H O] ; ferulylmonoolein (FMO), [M + H − H O] ;
2
2
+
ferulyl diolein (FDO), [M] ; monoolein, MO; oleic acid, OA; ethyl oleate, EO; diolein, DO; triolein, TO.
c
®
Alcoholysis and transesterification of 0.1 M EF with 0.1 M MO and TO in 5 mL toluene catalyzed by 110 mg of Novozym
4
35 for 72 h.
d
e
Transesterification of 0.36 M EF in 1.4 mL TO catalyzed by 110 mg of Novozyme 435 for 144 h.
Percentage yields for the ultraviolet-A (UVA) species produced in each reaction were calculated based on the total peak
area recorded using the UV6000LP detector.
corresponding to diferulyl glycerol (427 m/z, [M + H −
The transesterification of EF is dependent upon the re-
H O] ) was not observed. The loss of H O from a monoacyl- moval of oleate groups from TO to form MO and DO, which
+
2
2
glycerol during HPLC–APCI–MS analysis was expected and then can react with the EF-enzyme intermediate to form FMO
has been previously documented (15). Similarly, the peaks at and FDO. Novozym 435 provides enough water (0.025 M in
R = 2.4 and 2.5 min possessed major ions of 515 m/z [M + H the 5-mL toluene reactions) to promote the partial hydrolysis
t
+
−
H O] and were identified as isomers of ferulyl monoolein of TO to form MO and DO. The addition of more water to in-
2
(
FMO). It is believed that the peaks at R = 2.4 and 2.5 min crease the formation of MO and DO was not attempted be-
t
are the result of at least two FMO isomers, most likely the 1- cause, as shown above, additional water is deleterious in these
ferulyl-3-monoolein and 1-ferulyl-2-monoolein. The forma- reactions. Instead, ethanol (0.05 M) was added to the reaction
tion of a third isomer, 2-ferulyl-1-monoolein, is improbable in an attempt to enhance the rate of TO to MO and DO con-
due to the size of the ferulyl moiety and the steric restrictions version, giving EO as a by-product. After 24 h with ethanol
of the sn-2 position of the lipid. The last UVA species (R = present, twice as much DO and EO formed. The amount of
t
4
.4 min) formed from the glycerolysis corresponds to a major FMO and FDO formed after 24 h, however, was similar with
+
ion of 796 m/z, [M] , and was identified as ferulyl diolein or without ethanol added. When the reactions reached equi-
FDO). The presence of a single peak for FDO is consistent librium (72 h), the amount of FMO, FDO, and all other prod-
with the formation of a single isomer, most probably 1-feru- ucts formed are comparable for both reactions. These results
lyl-2,3-diolein.
suggest that, although initially more DO can be made avail-
(
SCHEME 2
JAOCS, Vol. 77, no. 5 (2000)

5
18
D.L. COMPTON ET AL.
able for the glycerolysis of EF, the rate-limiting step most $2,000/kg for Novozym 435). The cost efficiency of the reac-
likely involves the formation of, or the reaction with, the EF- tion could be greatly improved by reusing the lipase for sev-
enzyme intermediate.
eral reactions. The catalytic stability of Novozym 435 was
An alternative to partially hydrolyzing TO was perform- demonstrated by repeating the transesterification of EF with
ing the alcoholysis of EF using MO, which has available –OH neat TO with previously used lipase. The transesterification
groups. The reaction conditions were identical to the reaction of EF with neat TO, described above, reached equilibrium
using TO, and equilibrium was reached after 72 h. Compar- after 144 h resulting in a combined FMO and FDO yield of
ing the results of the two reactions (Table 1) shows that using 77%. After equilibrium was reached, the TO was decanted,
TO for the transesterification of EF produces 8% more FMO and fresh TO was added to the reaction vessel. Residual quan-
and 12% FDO, which is not formed in the alcoholysis reac- tities of EF and reaction products from the previous reaction
tion of EF and MO. In addition, the alcoholysis using MO were determined by HPLC and subtracted from the yields ob-
forms 25% more of unwanted transesterification product FG, tained from the second reaction. EF was added to initiate the
which lacks an oleate moiety. It was observed that the MO al- reaction, which stirred for an additional 120 h. HPLC analy-
coholysis results in the formation of a significant quantity of sis revealed that the second glycerolysis produced 31% FMO
DO after 2 h, which is slowly consumed as the reaction pro- and 43% FDO, a combined yield of 74%. The residual
ceeds (data not shown). The large quantity of glycerol formed amounts of EF and FG after 120 h were 17 and 9%, respec-
from the MO to DO conversion is then available to sequester tively. The slightly higher quantity of unreacted EF present at
ferulyl groups to form FG. It is reasonable to assume that a equilibrium (17%) compared to the quantity of unreacted EF
smaller quantity of glycerol is formed during TO glyceroly- present at equilibrium in the original glycerolysis (13%) is at-
sis, which would involve the removal of two oleate groups tributed to the shorter reaction time. These results indicate
from the lipid. Thus, less glycerol is available to form FG. that Novozym 435 remains active at 60°C for weeks and is
Overall, using TO for the Novozym 435-catalyzed transester- able to catalyze multiple glycerolysis reactions.
ification of EF in toluene is more efficient, resulting in a 44%
conversion of EF to the desired FMO and DMO products,
with 56% being converted to FG or left unreacted.
ACKNOWLEDGMENTS
The authors thank Mary M. Hallengren and Leslie J. Smith for their
excellent technical assistance.
If ever developed commercially, the transesterification of
EF would be more attractive if the cost of the solvent, solvent
waste, and residual solvent in the product could be avoided.
Therefore, we decided to determine the efficiency of perform-
ing the transesterification of EF in neat TO. A sufficient quan-
REFERENCES
tity of TO was needed to allow the Novozym 435 to be stirred
as a suspension, and it was found that 0.5 mmol EF was con-
veniently soluble in 1.4 mmol of TO (1.36 mL) at 60°C. The
transesterification of EF in neat TO required 144 h to reach
equilibrium, double the time required for the reaction per-
formed in toluene. This is attributed to the higher viscosity of
the neat TO reaction and the subsequent slower mass transfer
of the EF. The relative peak areas of the species detected
using the ELSD (Fig. 3) and the percent yields of FMO and
FDO obtained using the UV6000LP detector are reported in
Table 1. The combined yield of FMO and FDO was 77%,
leaving just 23% of the EF unreacted or converted to FG. This
is a 33% increase in the combined FMO and FDO yield com-
pared to the yield obtained during the transesterification of
EF with TO performed in toluene. As indicated in Table 1, the
relative peak area of FG produced during the transesterifica-
tion of EF in neat TO could not be calculated because an in-
dividual peak was not detected by ELSD (Fig. 3). The yield
of FG (10%), however, was determined based on the peak
area recorded by the UV6000LP detector. Overall, the trans-
1. Willis, W.M., R.W. Lencki, and A.G. Marangoni, Lipid Modifi-
cation Strategies in the Production of Nutritionally Functional
Fats and Oils, Crit. Rev. Food Sci. Nutr. 38:639–674 (1998).
2
3
4
. Gunstone, F.D., Movements Towards Tailor-Made Fats, Prog.
Lipid Res. 37:277–305 (1998).
. Gandhi, N.N., Applications of Lipase, J. Am. Oil Chem. Soc.
74:621–633 (1997).
. Schmid, R.D., and R. Verger, Lipases: Interfacial Enzymes with
Attractive Applications, Angew. Chem. Int. Ed. 37:1608–1633
(1998).
5
. Fairhurst, D., and M.A. Mitchnick, Particulate Sun Blocks: Gen-
eral Principles, in Sunscreens: Development, Evaluation, and
Regulatory Aspects, 2nd edn., edited by J.L. Nicholas, N.A.
Shaath, and M.A. Pathak, Marcel Dekker, Inc., New York, 1997,
pp. 313–352.
6
. Shaath, N.A., Evolution of Modern Sunscreen Chemicals, Ibid.,
pp. 3–33.
7. Shaath, N.A., Quality Control of Sunscreens, Ibid., pp. 657–676.
8. Wehtje, E., and P. Adlercreutz, Water Activity and Substrate
Concentration Effects on Lipase Activity, Biotechnol. Bioeng.
5
5:798–806 (1997).
9
. Cernia, E., C. Palocci, and S. Soro, The Role of the Reaction
Medium in Lipase-Catalyzed Esterifications and Transesterifi-
cations, Chem. Phys. Lipids 93:157–168 (1998).
esterification of EF in neat TO provides the most efficient 10. Valiverty, R.H., G.A. Johnston, C.J. Suckling, and P.J. Halling,
Solvent Effects on Biocatalysis in Organic Systems: Equilib-
route to ferulyl-acylglycerols, producing the highest com-
rium Position and Rates of Lipase-Catalyzed Esterification,
Biotechnol. Bioeng. 38:1137–1143 (1991).
1. Habulin, M., V. Krmelj, and Z. Knez, Synthesis of Oleic Acid
Esters Catalyzed by Immobilized Lipase, J. Agric. Food. Chem.
44:338–342 (1996).
bined yield of FMO and FDO with the absence of residual
solvent, but at the expense of doubling the reaction time.
One factor limiting the use of lipase-catalyzed reactions
on an industrial scale is the cost of the enzyme (e.g.,
1
JAOCS, Vol. 77, no. 5 (2000)

LIPASE-CATALYZED SYNTHESIS OF FERULATE ESTERS
519
1
1
1
2. Chulalaksananukul, W., J.S. Condoret, P. Delorme, and R.M. 15. Neff, W.E., and W.C. Byrdwell, Characterization of Model Tri-
Willmot, Kinetic Study of Esterification by Immobilized Lipase
in n-Hexane, FEBS Lett. 276:181–184 (1990).
3. Marty, A., W. Chulalaksananukul, R.M. Willmot, and J.S. Con-
doret, Kinetics of Lipase Catalyzed Esterification in Supercriti-
acylglycerol (triolein, trilinolein, and trilinolenin) Autooxida-
tion Products via High-Performance Liquid Chromatography
Coupled with Atmospheric Pressure Chemical Ionization Mass
Spectroscopy, J. Chromatogr. A 818:169–186 (1998).
cal CO , Biotechnol. Bioeng. 39:273–280 (1992).
2
4. Otto, R.T., U.T. Bornscheuer, H. Scheib, J. Pleiss, C. Sydatk,
and R.D. Schmid, Lipase-Catalyzed Esterification of Unusual
Substrates: Synthesis of Glucuronic Acid and Ascorbic Acid (vi-
tamin C) Esters, Biotechnol. Lett. 20:1091–1094 (1998).
[Received August 23, 1999; accepted January 31, 2000]
JAOCS, Vol. 77, no. 5 (2000)