Protein Engineering, Design & Selection vol. 23 no. 12 pp. 929–933, 2010
Published online October 29, 2010 doi:10.1093/protein/gzq089
Mutational analysis of phenylalanine ammonia lyase
to improve reactions rates for various substrates
Sebastian Bartsch and Uwe T.Bornscheuer 1
Department of Biotechnology and Enzyme Catalysis, Institute of
Biochemistry, Greifswald University, Felix-Hausdorff-Str. 4, D-17487
Greifswald, Germany
1
To whom correspondence should be addressed.
E-mail: [email protected]
Received June 24, 2010; revised September 3, 2010;
accepted September 27, 2010
Edited by Jacques Fastrez
Phenylalanine ammonia lyases (PAL) catalyze the
reversible, non-reductive amination of trans-cinnamic
acid to L-phenylalanine in the presence of high ammonia
concentrations. Since neither cofactor recycling nor
other additives are needed and by this asymmetric synthesis theoretical yields of 100% can be reached, it is an
interesting reaction for industrial processes. In this study
we demonstrate the superior properties of p-nitro-cinnamic acid ( p-n-CA) in the amination reaction using the
PAL from Petroselinum crispum ( pcPAL). By focuseddirected evolution, three mutants were identified
showing increased reaction rates and decreased substrate
inhibition. Together, the F137V mutant with p-n-CA
showed a 15-fold increased reaction rate compared with
the pcPAL WT with the natural cinnamic acid. The high
reaction rates were also proven in preparative scale
experiments. Activities towards other p-substituted
cinnamic acids showing different electronic effects of the
substituent were analyzed. Focused-directed evolution
around the carboxylic acid- and amine-binding
site always decreased PAL activity, due to a sensitive
H-bond network.
Keywords: amino acids/enzyme catalysis/molecular
modeling/protein design/substrate specificity
Introduction
The enzymatic synthesis of (non)natural, optically active
amino acids is widely used in industry and many enzymatic
routes have been developed using acylases, amidases, hydantoinases, aminotransferases, ammonia lyases and amino acid
dehydrogenases (Sonke et al., 2009). Among these processes,
only a few can be performed as asymmetric syntheses, in
which a prostereogenic substrate can be converted to the
optically pure amino acid at 100% theoretical yield. To date,
only aminotransferases and ammonia lyases have been
observed to catalyze asymmetric synthesis without the need
for cofactor recycling, implementation of dynamic kinetic
resolution strategies or the use of additional enzymes.
Common drawbacks of these two enzyme classes are low
enzyme stabilities, narrow substrate specificities and strong
substrate inhibition that leads to low reaction rates. Earlier
work on ammonia lyases has shown that amination of
(E)-cinnamic acid can be performed with phenylalanine
ammonia lyase (PAL) catalysis at high concentrations of
ammonia (Yamada et al., 1981). The aromatic amino acid
ammonia lyases catalyze the non-oxidative, reversible deamination of aromatic amino acids, including phenylalanine,
tyrosine and histidine, into the corresponding a,b-unsaturated
carboxylic acid. These enzymes share the catalytic prosthetic
4-methylidene imidazol-5-one group, which is constructed
through the autocatalytic cyclization of the Ala-Ser-Gly tripeptide and acts as the electrophile in the reaction mechanism (Schwede et al., 1999). Two possible reaction
mechanisms have been suggested in the literature: an E1cB
(Havir and Hanson, 1968; Hermes et al., 1985) and a
Friedel-Crafts like mechanism (Langer et al., 1995; Schuster
and Rétey, 1995).We previously proposed that both reaction
mechanisms are applicable to the deamination of aromatic
amino acids that depend on the residue corresponding to position 484 of PAL from Petroselinum crispum ( pcPAL)
(Fig. 1) (Bartsch and Bornscheuer, 2009).
The synthesis of several phenylalanine analogs with different substituents at the aromatic ring using PAL enzymes has
previously been shown (Gloge et al., 1998; Liu, 1999; Gloge
et al., 2000; Szymanski et al., 2009; Wu et al., 2009). These
compounds serve as essential building blocks that are used,
for example, in the synthesis of protease inhibitors for HIV
treatment (Liu, 1999).
Materials and methods
General
All chemicals were purchased from Fluka-Sigma-Aldrich
(München, Germany) and ABCR (Karlsruhe, Germany).
Protein expression
The codon optimized pcPAL gene (Baedeker and
Schulz, 1999) was expressed and purified as described previously (Bartsch and Bornscheuer, 2009). The desalting of
the His-Tag purified enzyme was performed using
Centriconw centrifugal filter units (Millipore, Schwalbach,
Germany).
Activity measurement
Enzyme activity and kinetic constants were determined by UV
spectroscopy in microtiter plates. The amination reaction was
assayed in 5 M ammonia solution at pH 10 adjusted by adding
sulfuric acid at 308C. Absorption was measured at a wavelength leading to absorption coefficients not higher than 1 ¼
1000 M21 cm21 to monitor initial rates using 10 mM substrate
concentrations. The deamination reaction was monitored in
Tris–HCl buffer (50 mM, pH 8.8, 308C). To directly compare
amination- and deamination reactions, both were monitored at
the same wavelength: 1: 1NH3-305 ¼ 679 M21 cm21, 1Tris-305 ¼
576 M21 cm21; 2: 1NH3-380 ¼ 562 M21 cm21, 1Tris-380 ¼
# The Author 2010. Published by Oxford University Press. All rights reserved.
For Permissions, please e-mail: [email protected]
929
S.Bartsch and U.T.Bornscheuer
25 mM were used with 2 – 3 mg purified enzyme (see main
text). Conversion was monitored using reversed phase highperformance liquid chromatography (HPLC) analysis
(Kromasil C18 colum (60 4 mm) in methanol/water:THF)
using heat-deactivated samples. The reaction mixture was
lyophilized and redissolved in 1 N aqueous HCl and loaded
on a column with activated ion-exchange resin (10 g Dowex
50 8, 50 – 100 mesh). The column was washed with water
and the product was eluted using 2 M ammonia solution.
Amino acid containing fractions (monitored by thin layer
chromatography and ninhydrin staining) were lyophilized.
NMR analysis confirmed product formation.
Determination of enantiomeric excess
Fig. 1 Active site of pcPAL (pdb-code: 1w27) (Ritter and Schulz, 2004)
with docked substrate and indicated binding pockets. The H-bond network of
most involved residues is shown in dashed lines.
534 M21 cm21; 3: 1NH3-315 ¼ 722 M21 cm21; 4: 1NH3-340 ¼
409 M21 cm21; 5: 1NH3-310 ¼ 353 M21 cm21, 1Tris-310 ¼
383 M21 cm21; 6: 1NH3-300 ¼ 336 M21 cm21, 1Tris-305 ¼
374 M21 cm21. Due to the low activity towards tyrosine, it
was necessary to use different wavelength to get increased
sensitivity: 7: 1NH3-380 ¼ 824 M21 cm21, 1Tris-310 ¼ 11
477 M21 cm21. The given errors are based on the standard
deviation of at least three independent protein expressions and
purifications. For determination of the kinetic constants,
Hanes-Woolf plots with 8 – 12 different substrate concentrations were used.
Enantiomer separations were performed on a Beckman
PACE-MDQ system with a photodiode array detector following the suggested protocol of the Beckman ‘highly sulfated
cyclodextrin test kit’. The capillary length was 30 cm, with a
distance of 20 cm to the detector and an inner diameter of
50 mm. After injection (0.3 psi for 4 s), the samples were
separated by applying 15 kV for 15 min using highly sulfated
b-cyclodextrins.
Computer modeling
The substrate binding of cinnamic acid and p-nitro cinnamic
acid was analyzed using YASARA (version 10.4.8) as
described earlier (Bartsch and Bornscheuer, 2009).
Results
In this study, we focused on the synthetically interesting
asymmetric amination process applied to a range of para-
Mutant library generation and expression
The site-directed saturation mutagenesis was performed following the QuikChangew protocol (Stratagene) using degenerate primers including an NNK codon. The following
primers were used: F137-NNK-FW: 50 -TG CAG AAA GAA
CTG ATC CGC NNK CTG AAC GCT GGT ATC TTC
GG-30 and F137-NNK-RV: 50 -CC GAA GAT ACC AGC
GTT CAG MNN GCG GAT CAG TTC TTT CTG CA-30 .
The PCR products were first transformed into chemocompetent XL10-gold cells (Stratagene) and after plasmid
preparation transformed into chemo-competent BL21(DE3)
cells already containing the pTF16-chaperone-plasmid
(TaKaRa). Clones were picked into 96-well microtiter plates
and after overnight culture 20 ml were transferred into 96
deep well plates for protein expression containing 1.2 ml LB
media with 0.5 mg/ml L-arabinose for induction of the chaperone expression (3 h, 378C, 1000 rpm). 1.2 mmol IPTG
were added per well to induce protein expression (20 h,
208C, 1000 rpm). Cells were precipitated and lysed adding
400 ml well21 lysis buffer (lysozyme 0.1 mg ml21, DNAseI
5 mg ml21) for 30 min at 308C at 1000 rpm followed by cell
disruption in microtiter plates in a nearly empty ultrasonic
bath.
Preparative biocatalysis
These were performed in a stirred solution of 20 ml 5 M
ammonia at 308C and pH 10 (adjusted by bubbling CO2
through the solution). Substrate concentrations of 5 or
930
Fig. 2 Cinnamic acid 1 and its analogs 2 –8 examined for amination by
pcPAL WT and its mutants. The corresponding amino acids are referred to
as 1a– 8a.
Engineering of PAL
Table I. Kinetic constants of pcPAL WT and F137 mutants in the deamination (L-Phe and L-p-n-Phe) and amination reaction (CA and p-n-CA).
Substrate
L-Phe (1a)
L-p-n-Phe (2a)
Enzyme
Vmax
(U mg21)
KM (mM)
Vmax
(U mg21)
KM (mM)
pcPAL WT
F137V
F137I
F137T
3.1 + 0.7
1.9 + 0.1
1.7 + 0.1
1.8 + 0.2
0.6 + 0.2
0.5 + 0.1
0.3 + 0.2
0.5 + 0.1
2.3 + 0.5
2.2 + 0.3
1.6 + 0.3
2.0 + 0.6
2.0 + 1.2
2.2 + 0.4
2.2 + 0.5
2.5 + 0.7
Substrate
CA (1)
Enzyme
Vmax
(U mg21)
KM (mM)
Vmax
(U mg21)
KM (mM)
pcPAL WT
F137V
F137I
F137T
0.3 + 0.1
0.9 + 0.2
0.7 + 0.4
0.8 + 0.1
0.2 + 0.1
0.2 + 0.1
0.3 + 0.2
0.4 + 0.1
2.6 + 0.5
4.5 + 0.8
3.8 + 1.1
3.1 + 0.9
2.8 + 1.0
4.2 + 0.7
3.1 + 1.9
2.5 + 1.0
p-n-CA (2)
substituted commercially available phenylalanine analogs
(Fig. 2).
The amination reaction
rate of 2 and a 3-fold higher reaction rate towards 1. The
deamination rates, however, were not increased in any of the
three mutants. The activity towards Phe (1a) was decreased
in all mutants and towards p-n-Phe (2a), no significant
changes in reaction rates were observed (Table I).
Biocatalysis
Preparative biocatalysis at the 0.5 mmol scale (25 mM substrate concentration, 3 mg purified enzyme) was performed.
The improved reaction rate of the F137V mutant towards 2
was confirmed by HPLC analysis (Fig. 3). After 6 h, the
F137V mutant enabled 82% conversion while the WT
showed only 62% and after 25 h final conversions of 90%
and 79%, respectively. Both WT and mutant showed high
enantioselectivity in the formation of L-p-nitro-Phe ( pcPAL
WT: 95%ee; F137V mutant: 97%ee).
Since the F137V mutant showed increased reaction rates
in the amination reaction, additional substituted cinnamic
acid and phenylalanine analogs were investigated (Table II).
Compared to the pcPAL WT, the reaction rates of the F137V
mutant for the amination reaction were doubled for
p-fluoro-CA (5; Vmax ¼ 0.25 U mg21) and were increased
In initial experiments, the amination process was catalyzed
by pcPAL WT in 5 M ammonia solution ( pH 10) and it was
found that p-nitro-cinnamic acid ( p-n-CA, 2) was converted
at a 9-fold higher reaction rate than cinnamic acid (CA, 1)
(Table I). Interestingly, the ratio between the reaction rates of
ammonia elimination/addition was inverted when comparing
Phe (1a/1) with the para-nitro substituted substrates (2a/2)
(Table I). While pcPAL showed a 10-fold higher deamination rate of Phe (1a) over the amination of CA (1), p-N-CA
(2) was aminated 10% higher than its corresponding amino
acid.
Focused-directed evolution
It is known that residue F137 in PAL plays an important role
in the substrate selectivity of ammonia lyases. F137 points
directly towards the aromatic ring of the substrate (Fig. 1,
Supplementary Fig. S1) and was reported to serve as the
selectivity switch that distinguishes PAL from tyrosine
ammonia lyases (TAL) (Louie et al., 2006). The corresponding position to F137 in the TAL enzymes is a histidine
residue and is thought to stabilize the p-hydroxy group of the
tyrosine. It was previously shown that PAL also accept tyrosine (7a) as substrate with very low reaction rates and that an
F137H mutant led to increased activity due to an additional
H-bond stabilizing the substrate in the active site (Louie
et al., 2006; Watts et al., 2006). Due to the non-natural and
bulky nitro-substituent found in compound 2 it was proposed
that saturation mutagenesis at position F137 of pcPAL would
allow access to an enzyme mutant with increased volume for
this extra steric residue that should enhance the reaction rate
(Fig. 1). From a saturation mutagenesis library at residue
137, we identified three different mutants showing increased
activity in the amination of p-n-CA (2): F137V, F137I and
F137T (Table I). All three mutants had a smaller side chain,
increasing the space available within the hydrophobicbinding pocket. Compared to pcPAL WT, the F137V mutant
showed a 1.7-fold higher reaction activity in the amination
Fig. 3 Preparative biocatalysis comparing conversions of p-n-CA using
pcPAL WT (triangles) and pcPAL-F137V (squares) as monitored by HPLC.
Table II. Kinetic constants for pcPAL WT and the pcPAL-F137V mutant
towards different substrates
Enzyme
pcPAL WT
pcPAL-F137V
Substrate
Vmax
(U mg21)
KM (mM)
Vmax
(U mg21)
KM (mM)
CA(1)
p-n-CA(2)
p-methyl-CA(3)
p-formyl-CA(4)
p-fluoro-CA(5)
p-trifluoromethyl-CA(6)
Coumarate(7)
p-amino-CA(8)
Phe(1a)
p-n-Phe(2a)
p-fluoro-Phe(5a)
p-trifluoro-methyl-Phe(6a)
Tyr(7a)
0.30 + 0.09
2.63 + 0.45
0.26 + 0.02
0.21 + 0.10
0.25 + 0.06
0.50 + 0.13a
0.58 + 0.13
,0.01
3.11 + 0.72
2.25 + 0.54
0.95 + 0.16
1.43 + 0.41
0.07 + 0.01
0.2 + 0.1
2.8 + 1.0
0.5 + 0.4
0.4 + 0.1
0.3 + 0.2
n.d.
12.6 + 1.1
n.d.
0.6 + 0.2
2.0 + 1.2
0.4 + 0.2
3.1 + 0.5
0.4 + 0.1
0.9 + 0.2
4.5 + 0.8
0.1 + 0.04
0.1 + 0.04
0.6 + 0.07
1.6 + 0.3a
0.5 + 0.1
,0.01
1.9 + 0.1
2.2 + 0.3
0.8 + 0.01
3.3 + 0.91
.0.01
0.2 + 0.02
4.2 + 0.7
0.1 + 0.03
0.1 + 0.1
0.7 + 0.6
n.d.
5.9 + 2.4
n.d.
0.5 + 0.1
2.2 + 0.4
0.4 + 0.1
2.5 + 0.5
n.d.
a
Specific activity at 2.5 mM substrate concentration due to strong substrate
inhibition.
931
S.Bartsch and U.T.Bornscheuer
3-fold when p-trifluoromethyl-CA (6; Vmax ¼ 0.5 U mg21)
was used as substrate.
In preparative scale experiments, the superior performance
of the F137V mutant using p-trifluoromethyl-CA (6) (5 mM)
was confirmed (Supplementary Fig. S2).
pcPAL WT showed significant substrate inhibition with
compound 6 at concentrations .2.5 mM (Supplementary
Fig. S3) and therefore specific activity at 2.5 mM rather than
Michaelis – Menten kinetics are given. With the F137V
mutant, however, the substrate inhibition was reduced and
reasonable activities even at 5 mM of 6 were observed
(Supplementary Fig. S3). In contrast, no substrate inhibition
was observed for the corresponding amino acid
p-trifluoromethyl-Phe (6a) up to 10 mM. This indicates that
the amino acid as the product of the amination reaction of 6
may be strongly bound in the active site. The p-methyl- (3)
and p-formyl- (4) substituted cinnamic acids were aminated
faster by the pcPAL WT than by the F137V mutant but at
low reaction rates. With p-amino-CA (8), however, bearing
an electron withdrawing substituent and having a positive
mesomeric effect, no conversion was found. The F137V
mutant showed very low deamination activity towards tyrosine (7a). In contrast, both the wild type (WT) and mutant
showed good activity with coumarate (7) (0.5 U mg21 (WT)
and 0.6 U mg21 (F137V)).
Focused-directed evolution of the carboxylic acid- and
amine-binding site
In addition to the para-substituted substrates we also
searched for mutants showing modifications at the carboxylic
acid group of the substrate. We previously identified the
F137H-mutant of pcPAL as being active towards the nonnatural substrate tyrosinol (Bartsch and Bornscheuer, 2009).
Due to more sensitive assay systems we continued searching
for mutants showing enhanced deamination activity towards
non-natural substrates using focused-directed evolution, followed by evaluation of the amination reaction efficacy. To
achieve this goal, a range of single and multiple site-directed
saturation mutagenesis libraries were predicted based on
computer-modeling focusing around the area of the carboxylic acid and amine-binding site (R354, D487, Q488,
Y110) of pcPAL (Fig. 1). In addition, a number of distinct
point mutations were investigated. Unfortunately, none of
these mutants showed enhanced activity and furthermore
most of them lost significant activity (data not shown). The
high sensitivity of this enzyme class to mutations in the
active site pocket has also been reported by other groups
(Cooke et al., 2009). Computer modeling studies showed a
sensitive hydrogen bond network surrounding the carboxylic
acid and amine-binding site of the substrate, which seems to
be easily destroyed by mutations as this hydrogen bond
network appears to be essential to maintain the catalytic
activity of the enzyme. Hence, no mutation could be identified leading to enhanced activity towards substrates with a
hydroxy-substituted carboxylic group or the natural substrates
by mutations around the carboxy- and amino-group-binding
site.
(Schuster and Rétey, 1995). It was, however, previously
reported that no activity was observed for the Rhodotorula
graminis PAL towards p-n-CA (2) (Liu, 1999). By using a
phenylalanine ammonia mutase (PAM), catalyzing the interconversion from a- and b-amino acids as well as ammonia
addition to phenylarylates, higher amination rates towards
p-n-CA (kCat ¼ 133 s21) were reported compared with cinnamic acid (1) (Szymanski et al., 2009; Wu et al., 2009). Wu
et al. observed strong electron withdrawing and mesomeric
effects of the p-nitro substituent which favored an
a-directing amination reaction. While the amination activity
of the PAM was two orders of magnitude lower than for the
ammonia lyase studied here, the effects of the substrate
nitro-group showed distinct similarity. Strong negative mesomeric and inductive effects led to a significant activation of
the Ca-atom of p-n-CA (2) for a nucleophilic attack (Fig. 4).
In previous studies, it was shown by measuring kinetic
isotope effects that the rate limiting step in the reaction
mechanism of the conversion of Phe to CA is the release of
CA and not the abstraction of the b – proton (Hermes et al.,
1985). By introducing the L138H-mutant, Viergutz and
Rétey changed the rate-limiting step from substrate release to
the b-H abstraction, effected by decreased interactions of the
aromatic ring in the hydrophobic-binding pocket (Viergutz
and Rétey, 2004). This observation might explain the
increased activity observed for the F137V-mutant in the
amination of 1. The product 1a is strongly bound by
H-bonds between R354 and the carboxylic acid, the newly
added amino group, E484 and by hydrophobic interactions of
the aromatic ring in the hydrophobic-binding pocket (Fig. 1)
(Bartsch and Bornscheuer, 2009). By decreasing the hydrophobic interactions, product 1a may be released faster from
the active site. The same effect could be responsible for the
observed decreased activity in the deamination of Phe.
In summary, we identified the F137V-mutant of pcPAL by
focused-directed evolution of the hydrophobic pocket
showing higher amination reaction rates for several substrates. In particular, p-nitro cinnamic acid (2) was shown to
be a superior substrate in the amination process. By combination of the F137V-mutant and p-n-CA (2), a 15 times
higher reaction rate was achieved as compared with the wildtype enzyme and its natural substrate cinnamic acid (1).
Using p-trifluoromethyl-CA (6), the mutation in the hydrophobic pocket was shown to alleviate substrate inhibition
effects. To our knowledge, all mutations reported for PAL
enzymes that generate or increase reaction rates have focused
Discussion
As described by Schuster et al., the p-nitro substitution
imparts an activating effect on the b-protons of p-n-Phe (2a)
932
Fig. 4 Charge distribution through the mesomeric effect in p-nitro cinnamic
acid.
Engineering of PAL
on residue 137 (Louie et al., 2006; Watts et al., 2006;
Bartsch and Bornscheuer, 2009) in the aromatic-binding site.
Successful mutations have not been found around the carboxylic acid- and amine-binding sites due to a sensitive and
essential hydrogen-bond network. With the pcPAL mutants
described here, significantly increased activities were
observed and transferred into preparative scale. Comparison
of the amination- and deamination reactions showed that
electron withdrawing substituents can significantly increase
the amination reaction rates, most likely due to increased
acidity of the Ha atoms.
Supplementary data
Supplementary data are available at PEDS online.
Acknowledgements
We thank Dr. Andrew Evitt and Dr. Dominique Böttcher (Institute of
Biochemistry, Greifswald University) for their useful discussions.
Funding
This work was supported by the Deutsche Bundesstiftung für
Umwelt (AZ 13197).
References
Baedeker,M. and Schulz,G.E. (1999) FEBS Lett., 457, 57– 60.
Bartsch,S. and Bornscheuer,U.T. (2009) Angew. Chem. Int. Ed., 48,
3362–3365. Corrigendum: Angew. Chem. Int. Ed. (2010), 49, 3860.
Cooke,H.A., Christianson,C.V. and Bruner,S.D. (2009) Curr. Opin. Chem.
Biol., 13, 453– 461.
Gloge,A., Langer,B., Poppe,L. and Rétey,J. (1998) Arch. Biochem. Biophys.,
359, 1– 7.
Gloge,A., Zoń,J., Kövári,A., Poppe,L. and Rétey,J. (2000) Chem. Eur. J., 6,
3386–3390.
Havir,E.A. and Hanson,K.R. (1968) Biochemistry, 7, 1904– 1914.
Hermes,J.D., Weiss,P.M. and Cleland,W.W. (1985) Biochemistry, 24,
2959–2967.
Langer,M., Pauling,A. and Rétey,J. (1995) Angew. Chem. Int. Ed., 34,
1464–1465.
Liu,W. (1999) Patent US 5981239.
Louie,G.V., Bowman,M.E., Moffitt,M.C., Baiga,T.J., Moore,B.S. and
Noel,J.P. (2006) Chem. Biol., 13, 1327– 1338.
Ritter,H. and Schulz,G.E. (2004) Plant Cell, 16, 3426–3436.
Schuster,B. and Rétey,J. (1995) Proc. Natl Acad. Sci. USA, 92, 8433– 8437.
Schwede,T., Rétey,J. and Schulz,G. (1999) Biochemistry, 38, 5355– 5361.
Sonke,T., Kaptein,B. and Schoemaker,H.E. (2009). In Hughes,A.B. (ed),
Amino acids, peptides and proteins in organic chemistry. Wiley-VCH,
Weinheim, pp. 77– 117.
Szymanski,W., Wu,B., Weiner,B., De Wildeman,S., Feringa,B.L. and
Janssen,D.B. (2009) J. Org. Chem., 74, 9152– 9157.
Viergutz,S. and Rétey,J. (2004) Chem. Biodivers., 1, 296–302.
Watts,K.T., Mijts,B.N., Lee,P.C., Manning,A.J. and Schmidt-Dannert,C.
(2006) Chem. Biol., 13, 1317–1326.
Wu,B., Szymanski,W., Wietzes,P., De Wildeman,S., Poelarends,G.J.,
Feringa,B.L. and Janssen,D.B. (2009) ChemBioChem, 10, 338– 344.
Yamada,S., Nabe,K., Izuo,N., Nakamichi,K. and Chibata,I. (1981) Appl.
Environ. Microb., 42, 773–778.
933