One-pot synthesis of L-fructose using coupled multienzyme systems based on rhamnulose-1-phosphate aldolase

Article abstract of DOI:10.1021/jo030021m

Two methods have been developed for the highly efficient enzymatic synthesis of L-fructose: one is based on rhamnulose-1-phosphate aldolase and acid phosphatase using racemic glyceraldehyde and dihydroxyacetone phosphate as substrates; the other is to gen

Full text of DOI:10.1021/jo030021m

On e-P ot Syn th esis of L-F r u ctose Usin g
Cou p led Mu ltien zym e System s Ba sed on
Rh a m n u lose-1-p h osp h a te Ald ola se
SCHEME 1. P r ep a r a tion of l-F r u ctose (1) Sta r tin g
fr om r a c-Glycer a ld eh yd e (r a c-3) a n d DHAP (2)
Dirk Franke, Timothy Machajewski,1
Che-Chang Hsu, and Chi-Huey Wong*
Department of Chemistry and the Skaggs Institute for
Chemical Biology, The Scripps Research Institute,
0550 North Torrey Pines Road, La J olla, California 92037
1
wong@scripps.edu
Received J anuary 14, 2003
Abstr a ct: Two methods have been developed for the highly
efficient enzymatic synthesis of L-fructose: one is based on
rhamnulose-1-phosphate aldolase and acid phosphatase
using racemic glyceraldehyde and dihydroxyacetone phos-
phate as substrates; the other is to generate enantiomeri-
cally pure L-glyceraldehyde in situ from glycerol for the aldol
reaction, using galactose oxidase catalyzed oxidation of
glycerol in the presence of catalase. Using this four-enzyme
system, enantiomerically pure L-fructose was obtained.
Using the more expensive dihydroxyacetone phosphate, the
yield was 55% after purification.
We reported earlier the enzymatic synthesis of L-
fructose (1) from dihydroxyacetone phosphate (2, DHAP)
and L-glyceraldehyde (L-3),1
2,13
which is carried out by a
multienzyme system comprising rhamnulose-1-phos-
phate aldolase (RhaD) and acid phosphatase (AP) using
14
stereospecific aldol addition reaction by this aldolase.
Although 1 is produced rapidly in moderate yield, it
suffers from two limitations. First, L-3 is not com-
mercially available. Although it can be synthesized with
osmium-catalyzed asymmetric dihydroxylation of pro-
1
3
tected acrolein or glycol cleavage of ascorbic acid
derivatives,15 preparation of L-3 on gram scales is inef-
ficient and the use of toxic heavy metals places a limit
on its applications. Additionally, L-3 is thermodynami-
Unnatural monosaccharides have many useful applica-
tions. L-Fructose, for example, is known as a nonnutritive
2
3
sweetener, an inhibitor of various glycosidase, and an
insecticide for ants and house flies.4 In addition, un-
natural monosaccharides are potentially useful as chiral
building blocks for the synthesis of biologically active
16
cally metastable and decomposes easily. Second, the
intermediate L-fructose-1-phosphate (4) was isolated as
a barium salt, and additional steps were required to
remove and recycle the toxic metal salts. We report
herein two improved methods for the synthesis of 1, each
using a coupled enzyme system, that overcome these
limitations.
Our initial strategy was to use racemic glyceraldehyde
(rac-3) and utilize the stereoselectivity of the aldolase to
preferentially make L-fructose (1) (Scheme 1).
5
compounds. Given the importance of these monosaccha-
rides, the development of their synthesis has been the
subject of considerable recent interest.
L-Fructose (1) has previously been prepared via
chemical6 and chemoenzymatic
-8
9-11
synthesis. Whereas
the enzyme-based methods afford very low yields and
often require relatively expensive starting materials,
attempts to chemically synthesize 1 have been character-
ized by tedious multistep sequences and low selectivity.
Thus, the overall low yields and difficulty in scaling up
L-fructose production have prevented further evaluation
of its applications.
DHAP (2) and rac-3 were condensed using RhaD to
yield L-fructose-1-phosphate (4). Conducting phosphate
ester hydrolysis with acid phosphatase (AP) without
isolation allowed us to avoid the use of barium salts. No
diastereomers of 1 were found as byproducts of the
1
condensation and hydrolysis steps ( H NMR). Chromato-
*
Corresponding author.
1) Current address: Chiron Corporation, Emeryville, CA 94608-
916.
2) Levin, G. V.; Zehner, L. R.; Saunders: J . P.; Beadle, J . R. Am.
J . Clin. Nutr. 1995, 62, 1161.
3) Muniruzzaman, S.; Pan, Y. T.; Zeng, Z.; Atkins, B.; Izumori, K.;
Elbein, A. D. Glycobiology 1996, 6, 795.
4) Levin, G. V.; Zehner, L. R. Eur. Pat. EP 397,027 (Cl. A01N43/
graphic purification was used to remove salts and
remaining D-glyceraldehyde (D-3) to give pure 1 in 64%
yield based on 2.17
As a refinement of this method we sought to produce
L-glyceraldehyde (L-3) in situ from an inexpensive start-
ing material and couple this reaction to the aldolase-
(
2
(
(
(
0
8), 14. Nov 1990; US Patent 6,166,193, 24. Nov 1992; Chem. Abstr.
991, 114, 242827q.
1
(
5) Gy o¨ rgyde a´ k, Z.; Pelyv a´ s, I. Monosaccharide Sugars-Chemical
(12) Alajarin, R.; Garcia-J unceda, E.; Wong, C.-H. J . Org. Chem.
1995, 60, 4294.
Synthesis by Chain Elongation, Degradation, and Epimerization;
(13) Henderson, I.; Sharpless, K. B.; Wong, C.-H. J . Am. Chem. Soc.
1994, 116, 558.
Commun. 1992, 8, 610.
(14) Fessner, W. D.; Badia, J .; Eyrisch, O.; Schneider, A.; Sinerius,
G. Tetrahedron Lett. 1992, 33, 5231.
(7) Chen, C. C.; Whistler, R. L. Carbohydr. Res. 1988, 175, 265.
(8) Gizaw, Y.; BeMiller, J . N. Carbohydr. Res. 1995, 266, 81.
(9) Mayo, J . W.; Anderson, R. L. Carbohydr. Res. 1968, 8, 344.
(10) Dhawale, M. R.; Szarek, W. A.; Hay, G. W.; Kropinski, A. M.
(15) Hubschwerlen, C. Synthesis 1986, 962-964.
(16) Baer, E.; Fischer, H. O. L. J . Am. Chem. Soc. 1939, 61, 761.
(17) Given the possibility of recovering D-3 and using the racem-
1
6
Carbohydr. Res. 1986, 155, 262.
11) Itoh, H.; Izumori, K. J . Ferment. Bioeng. 1996, 81, 354.
ization reaction to regenerate rac-3, it is in theory possible to
completely transform rac-3 into 1.
(
1
0.1021/jo030021m CCC: $25.00 © 2003 American Chemical Society
Published on Web 07/26/2003
6
828
J . Org. Chem. 2003, 68, 6828-6831

SCHEME 2. En zym a tic Oxid a tion of Glycer ol (5)
to Effor t L-Glycer a ld eh yd e (L-3) Usin g HL-ADH
a n d a NAD Regen er a tion System w ith r-Keto
Glu ta r ic Acid (7) a n d Glu DH
TABLE 1. Kin etic P a r a m eter s for th e Oxid a tion of
L-F r u ctose (1), DHAP (2), L-Glycer a ld eh yd e (L-3),
D-Glycer a ld eh yd e (D-3), L-F r u ctose-1-p h osp h a te (4),
Glycer ol (5), a n d Ga la ctose (ga l) w ith GOa se in 50 m M
P ota ssiu m P h osp h a te Bu ffer a t p H 6.7 a n d 23 °C
+
substrate
Kma [mm]
kcata [min-1
]
kcat/Kmb [mm-1 min-1
]
5
2
1
1150 ( 200 7.7 ( 0.5
0.0067 ( 0.0010
1.0 ( 0.2
4.2 ( 0.8
4.3 ( 0.2
e0.09
, L-3, D-3, 4 nqc
nq
c
gal
55 ( 10
11 ( 1
0.20 ( 0.04
a
The error intervals resulted from nonlinear regression of the
Lineweaver-Burk plot. b The error interval was calculated ac-
cording to Gauss’ error propagation law:
2
2
kcat
1
Km
kcat
∆
)
∆kcat
+
2∆Km
Km
x(
)
(K
)
m
catalyzed condensation with DHAP (2). The enzymatic
oxidation of glycerol (5) to L-3 has been reported on
analytical scales by two approaches: with horse liver
alcohol dehydrogenase (HL-ADH)18 and galactose oxidase
c
Not quantifiable.
HRP (5.5-7.5)24 was compatible with GOase for the
working coupled assay.
(
GOase) from Dactylium dendroides.19
+
The HL-ADH reaction requires NAD for activity, and
Determination of the kinetic parameters of the oxida-
tion of glycerol (5) and galactose indicated that the kcat
because NAD is too expensive for stoichiometric use,20
a cofactor regeneration system would be required for
preparative synthesis of L-3. This can be accomplished
by the L-glutamic dehydrogenase (GluDH) catalyzed
reduction of R-keto glutaric acid (7) to give glutamic acid
+
2
5
values for both substrates are quite similar (Table 1),
but the K value for 5 was 20 times higher than for the
m
natural substrate galactose. Quantification of product
formation proved that 5 was oxidized with only 1 equiv
of oxygen and neither L-glyceraldehyde (L-3) nor D-
glyceraldehyde (D-3) was accepted as a substrate for
GOase. L-Fructose (1) or L-fructose-1-phosphate (4) was
not accepted as a substrate at concentrations up to 5 m.
(
6) (Scheme 2).
However, two reasons render HL-ADH unattractive for
our purposes. HL-ADH is known to operate at a pH
optimum of 8.0, which precludes the possibility of coup-
ling the oxidation to the aldolase-catalyzed condensation
that would require more acidic conditions. In addition,
stoichiometric quantities of 7 would be necessary for
2
6
We have found that in addition to dihydroxyacetone,
its phosphorylated analogue DHAP (2) is accepted as a
substrate for GOase, resulting in the formation of an
+
27
NAD recycling and 1 would have to be separated from
unidentified product.
6
.
For establishing a preparative route toward the pro-
duction of L-fructose (1) from glycerol (5) with GOase and
RhaD, two different strategies were investigated.
In the first strategy the formation of L-glyceraldehyde
In contrast, GOase does not require the addition of a
cofactor and also operates at a similar pH as RhaD,
having a pH-optimum of around 6.7,21 but being also
active over a wide pH range from 5.5 to 8.0 (see
supplemental information). This would allow for the
simultaneous production and condensation of L-glycer-
aldehyde (L-3). This reaction, however, has never been
illustrated on preparative scales. Fungal GOase has
recently been produced with Pichia pastoris22 as an active
fusion protein and with Escherichia coli as a mutant
generated by directed evolution,23 making it possible to
economically produce large quantities of the enzyme.
To combine the GOase-catalyzed oxidation of glycerol
(
L-3) and condensation with DHAP (2) was conducted
simultaneously (Scheme 3, Strategy I).
Although the K value for GOase using 5 as a sub-
m
strate was 1.15 M, for enhanced diffusion rates, we chose
a concentration of 183 mM for 5 to produce L-3 in situ,
which corresponds to an activity of only 14% vmax. To
compensate for the low turnover rate, a GOase concen-
-1
tration of 5 U mL was used, which was significantly
-1
higher than the concentration of RhaD (0.8 U mL ). To
prevent inactivation of the enzyme, we supplemented the
system with catalase in excess in order to destroy
hydrogen peroxide, which was found to inactivate the
system.
(5) with RhaD condensation, we established an analytical
method for determining the kinetic parameters of GOase
reaction with different substrates. The released hydrogen
peroxide in the GOase reaction was used to quantify the
enzymatic activity through horseradish peroxidase (HRP)
catalyzed oxidation of 2,2′-azino-bis[3-ethylbenzthiazo-
Combining glycerol (5), DHAP (2), GOase, catalase,
and RhaD resulted in the formation of L-fructose-1-
1
28
phosphate (4) ( H NMR, no other products identified).
Simultaneously, the disappearance rate for DHAP (2)
2 2
line-6-sulfonic acid] (ABTS) with H O . The pH range of
-
1
was 17.5 mM h (enzymatic quantification of 2). Here
(18) Wong, C.-H.; Matos, J . R. J . Org. Chem. 1985, 50, 1992.
19) Klibanov, A. M.;. Alberti, B. N.; Marletta, M. A. Biochem.
(
Biophys. Res. Commun. 1982, 108, 804.
20) Wong, C.-H.; Whitesides, G. M. Enzymes in Synthetic Organic
Chemistry; Pergamon Press: Oxford, 1994.
(24) Bolscher, B. G. J . M.; Zoutberg, G. R.; Cuperus, R. A.; Wever,
R. Biochim. Biophys. Acta 1984, 784, 189.
(25) Kinetic parameters calculated with GraFit, Version 4.0; Erith-
acus Software Limited: U.K., 1998.
(26) Mendonca, M. H.; Zancan, G. T. Arch. Biochem. Biophys. 1987,
252, 507.
(
(
(
21) Driscoll, J . J .; Kosman, D. J . Biochemistry 1987, 26, 3429.
22) Whittaker, M. M.; Whittaker, J . W. Protein Expression Purif.
2
000, 20, 105.
23) Sun, L.; Petrounia, I. P.; Yagasaki, M.; Bandara, G.; Arnold,
F. H. Protein Eng. 2001, 14, 699.
(
(27) The possible product phosphoric acid mono(2,3-dioxo-propyl)
1
ester could not be identified using H NMR.
J . Org. Chem, Vol. 68, No. 17, 2003 6829

SCHEME 3. On e-P ot Syn th esis of L-F r u ctose (1)
Sta r tin g fr om Glycer ol (5) a n d DHAP (2) Usin g
Cou p led En zym a tic System s w ith GOa se, Ca ta la se,
Rh a D, a n d AP
starting materials the product is formed with three new
chiral centers in only one synthetic operation.
In summary, we have described two methods for the
preparation of L-fructose (1). The condensation of racemic
glyceraldehyde (rac-3) proceeds in high yield and elimi-
1
3
nates the necessity to use toxic osmium or palladium
catalysts15 to introduce the required oxygenation. In addi-
tion, the use of toxic barium for precipitation was elimi-
nated, thereby providing an efficient and environmentally
friendly process for the production of enantiomerically
pure 1. Additionally, the stereoselective oxidation of
glycerol (5) by galactose oxidase, coupled with the aldo-
lase reaction in situ, provides 1 in a relatively high yield.
Exp er im en ta l Section
E n zym e Sou r ces. Acid phosphatase type XA (EC 3.1.3.2,
from sweet potato), galactose oxidase (EC 1.1.3.9, from Dactyl-
ium dendroides, partially purified), horseradish peroxidase type
VI (EC 1.11.1.7, from horseradish), glycerol-3-phosphate dehy-
drogenase type I (EC 1.1.1.8, from rabbit muscle), alcohol
dehydrogenase (EC 1.1.1.1, from baker’s yeast), and catalase (EC
1.11.1.6, from bovine liver) were purchased commercially. All
enzymes have been used as received and have not been desalted.
Rhamnulose-1-phosphate aldolase (EC 4.1.2.19) was isolated
from E. coli as described previously.29
Assa y for Mon itor in g Ald ol Con d en sa tion w ith Rh a D.
Aldol condensation was monitored indirectly by the decrease in
DHAP concentration. This was done by reduction with NADH-
dependent glycerol-3-phosphate dehydrogenase in 50 mM potas-
sium phosphate buffer (pH 7.5) at 23 °C, while monitoring the
decrease of absorbance at 340 nm.
again, the dephosphorylation of 4 could be done with AP
in the presence of the other enzymes without prior
isolation. Unfortunately, after silica gel purification, only
a 5% yield of L-fructose (1) could be obtained from DHAP
(2). This can most likely be attributed to the reaction of
DHAP (2) with GOase. Increasing the concentration of
-
1
RhaD to 1.7 U mL (factor 2) or glycerol (5) to concen-
trations of 0.4, 0.8, or 1.2 M (factor 2, 4, or 6) did not
result in any significant additional formation of L-fructose-
Assa y for GOa se a ctivity by Cou p lin g w ith HRP /ABTS
Assa y. The determination of substrate acceptance for GOase
and Michaelis-Menten parameters was done by quantification
of hydrogen peroxide with a coupled HRP/ABTS assay at 23 °C.
1
-phosphate (4). Comparing the high kcat/K
GOase with DHAP (2) as a substrate to that with glycerol
5), we concluded that spatiotemporal separation of
m
value of
(
GOase reaction and RhaD condensation is essential to
prevent consumption of 2.
-1
Reaction mixture contained (final volume 500 µL): 500 µg mL
(0.91 mM) ABTS, 4.3 mg L-1 (0.75 U mL-1) HRP, 840 mg L-1
-
1
(
5.0 U mL ) GOase, 50 mM potassium phosphate buffer (pH
In a second attempt (Scheme 3, Strategy II) L-glycer-
aldehyde (L-3) was synthesized with GOase using cata-
lase and free oxygen prior to condensation with DHAP
6
.7). For the determination of kinetic parameters for D-galactose
-
1
as a substrate, the GOase concentration was 250 mg L (1.5 U
mL ). The reaction was initiated by addition of the substrate.
-1
(2). The formation of L-3 could be quantified in an end-
The absorption was monitored over a time period of 10 min at
a wavelength of 410 nm, which is the absorption maximum for
point assay using the NADH-dependent reaction of the
aldehyde with alcohol dehydrogenase. Alternatively,
Schiff’s reagent was used in a photometrically end-point
assay for quantification, which was more accurate but
+
oxidized ABTS . The pH operating experiments for oxidation of
glycerol (5) and D-galactose were done in 50 mM 2-(N-morpholi-
no)ethanesulfonic acid (MES, 5.5 e pH e 6.5) and 50 mM 3-(N-
morpholino)propanesulfonic acid (MOPS, 6.5 e pH e 8.0). The
pH was adjusted with 10 N NaOH.
-
1
-1
less sensitive. Using 75 U mL GOase and 93 U mL
catalase, 50 mM L-3 could be produced from 0.87 M
glycerol (5) within a 48 h process time. With regard to a
theoretical yield of 96% (not including the oxygen recy-
cling) this corresponds to a relative yield of 6%. For the
elimination of the remaining GOase activity, the proteins
were heated. We found that L-3 is stable in aqueous
reaction mixture at pH 6.7, when heated to 100 °C over
a time period of 10 min (alcohol dehydrogenase test,
Schiff’s test). Condensation of L-3 thus produced with
DHAP (2), dephosphorylation, and purification gave
L-fructose (1) in a highly pure form and 55% yield. We
concluded that the production of 1 using GOase is
favorable as a result of the low cost of the starting
material and the possibility of integrating the four-
enzyme system into a continuous process with only one
separation step. It is also remarkable that from achiral
Assa y for Qu a n tifica tion of L-Glycer a ld eh yd e (L-3) w ith
Alcoh ol Deh yd r ogen a se. l-Glyceraldehyde (L-3) was quantified
using an enzymatic assay based on the aldehyde reduction with
NADH-dependent alcohol dehydrogenase from baker’s yeast. The
assay was performed in 50 mM potassium phosphate buffer (pH
-1
-1
7
.0) using 3.4 mg mL (990 units mL ) alcohol dehydrogenase.
The rate of decrease of absorption at 340 nm was measured 2
min after addition of the enzyme. Linear correlation of the rate
and concentration of substrate was observed at up to 5 mM L-3
for the final concentration.
Assa y for Qu a n tifica tion of L-Glycer a ld eh yd e (L-3) w ith
Sch iff’s Test. For the quantification of l-glyceraldehyde (L-3)
using Schiff’s test, 10 µL of aqueous sample containing up to 83
mM L-3 was dissolved in 0.4 mL of Schiff’s reagent for aldehydes.
Absorption was monitored at a wavelength of 560 nm over a
time period of 30 min while incubating at 23 °C. Maximal
absorption was measured after approximately 15 min. The
concentrations of L-3 were calculated from the maximal absorp-
tion using a reference calibration curve with rac-3 and acetal-
dehyde as substrates.
(28) We found that crude GOase resulted in the formation of a
product mixture, while 4 was produced only when partially purified
GOase was used. Therefore, partially purified enzyme was used in all
further experiments.
(29) Garcia-J unceda, E.; Shen, G.-J .; Sugai, T.; Wong, C.-H. Bioorg.
Med. Chem. 1995, 3, 945.
6
830 J . Org. Chem., Vol. 68, No. 17, 2003

Im p r oved Syn th esis of L-F r u ctose (1) fr om Glycer ol (5)
in a Th r ee-Sta ge En zym a tic P r ocess. GOase (3750 units) and
catalase (4650 units) were added to a solution of 5 mL (44 mmol)
glycerol (5) in 45 mL of 50 mM potassium phosphate buffer (pH
this correlates with a yield of 55%. 1H NMR, 13C NMR, and
optical rotation were in complete agreement with authentic
samples.31
6
.7). The mixture was carefully shaken at 23 °C for 48 h while
vented to the atmosphere. The formation of L-glyceraldehyde (L-
) in the reaction mixture was monitored using the Schiff test.
Ack n ow led gm en t. A Feodor-Lynen fellowship of
the Alexander-von-Humboldt Foundation to D.F. is
gratefully acknowledged. We are grateful to Drs. Tom
Tolbert, Lac Lee, and J ungie Liu and Mrs. Sarah
Hanson for stimulating discussions.
3
-1
After a process time of 48 h, 4.5 g L (50 mM) L-3 were achieved.
For inactivation of remaining GOase the solution was heated
to 100 °C for 5 min using a microwave oven and immediately
cooled to 23 °C with a water bath. All nonsoluble ingredients
were removed by filtration and pH 6.8 was adjusted (1 N NaOH);
Su p p or tin g In for m a tion Ava ila ble: General experimen-
tal information; procedure for the synthesis of 1 from rac-3;
procedure for the synthesis of 1 from 5 with a two-stage
enzymatic process; Lineweaver-Burk plots for the determi-
nation of kinetic constants for GOase with galactose, glycerol,
and DHAP as substrates; pH rate profile for the oxidation of
glycerol and galactose with GOase; standard curve for the
quantification of glyceraldehyde with Schiff’s reagent; calcula-
tion of theoretical achievable yield for oxidation of glycerol;
2
4 mg (3.4 mmol) DHAP (2),30 3.8 mg (28 µmol) zinc(II) chloride,
and 35 units RhaD were added. The mixture was shaken at 23
C for 3 days. The pH was adjusted to 4.7 with 1 N HCl, and
00 µL (110 units) AP in aqueous solution was added. For
hydrolysis the reaction mixture was shaken at 37 °C for another
6 h. After cooling to 23 °C the pH was adjusted to 7.0 with 1 N
°
1
1
NaOH, and the solution was diluted with 150 mL of methanol.
The white precipitate was removed by filtration through Celite.
All volatiles were removed in vacuo, and the yellow residue was
1
and H NMR spectra of 1 generated from rac-3 and 5,
further purified using silica gel chromatography with CHCl
3
/
respectively. This material is available free of charge via the
Internet at http://pubs.acs.org.
MeOH (2:1). Fractions containing 1 were pooled and concen-
trated in vacuo, and 350 mg (1.9 mmol) of 1 was obtained as a
white solid. With regard to DHAP (2) as the starting material,
J O030021M
(
30) J ung, S.-H.; J eong, J .-H.; Miller, P.; Wong, C.-H. J . Org. Chem.
(31) Authentic 1 was prepared from L-3 following the method
described in ref 12.
1
994, 59, 7182.
J . Org. Chem, Vol. 68, No. 17, 2003 6831