Enzymatic synthesis of nylon-6 units in organic solvents containing low concentrations of water

Article abstract of DOI:10.1016/j.molcatb.2010.02.006

NylB′ carboxylesterase, which is 88% homologous to functional 6-aminohexanoate-dimer hydrolase (NylB) from Arthrobacter sp., possesses trace synthetic activity [0.0004μmolmin^-1mg^-1 (U/mg)] from 6-aminohexanoate (Ahx) to its oligomers in 90% tert-butyl alcohol. The synthetic activity and the ratio of the synthetic activity to the hydrolytic activity were significantly affected by amino acid substitutions at positions 181, 266 and 370. The synthetic activity was enhanced to 2.7U/mg by G181D-H266N substitutions, and the activity was further enhanced in the G181D-H266N-D370Y triple mutant to a level approximately 10⁴-fold greater than the parental carboxylesterase form (3.4U/mg), which was nearly equal to the ordinary hydrolytic activity in water (type A-mutants). Type A-mutants possessed more than 50% of the 6-aminohexanoate-linear dimer (Ald)-hydrolytic activity at 0-70% tert-butyl alcohol, but the synthetic reaction became predominant at 85-90% tert-butyl alcohol. In contrast, type B-mutants (G181E-H266N and G181N-H266N) possessed quite low levels of Ald-hydrolytic activity (<0.01U/mg) at 0-70% tert-butyl alcohol. However, both the hydrolytic and synthetic activities were enhanced at higher concentrations, and the maximum activity was obtained at 90% tert-butyl alcohol for both hydrolysis and synthesis. In a type C-mutant (R187S-F264C-D370Y), the Ald-hydrolytic activity was enhanced to approximately 80-fold that of the parental carboxylesterase, but the mutant barely demonstrated any synthetic activity. On the basis of the three-dimensional structure of the Ald-bound enzyme and a kinetic study for typical mutant enzymes, we propose that the efficient enzymatic syntheses of nylon-6 units were achieved by (i) stable binding of the 1st-Ahx at the N-terminal region with Asp181, (ii) interaction of the 2nd-Ahx at the C-terminal region with Tyr370, and (iii) motion of Tyr170 that generated a closed form in the catalytic center of Ald hydrolase.

Full text of DOI:10.1016/j.molcatb.2010.02.006

Journal of Molecular Catalysis B: Enzymatic 64 (2010) 81–88
Contents lists available at ScienceDirect
Journal of Molecular Catalysis B: Enzymatic
journal homepage: www.elsevier.com/locate/molcatb
Enzymatic synthesis of nylon-6 units in organic solvents containing low
concentrations of water
Yasuyuki Kawashimaa, Kengo Yasuhiraa, Naoki Shibatab, Yusuke Matsuuraa, Yusuke Tanakaa,
Masaaki Taniguchia, Yoshiaki Miyoshia, Masahiro Takeoa, Dai-ichiro Katoa,
Yoshiki Higuchib, Seiji Negoroa,∗
a Department of Materials Science and Chemistry, Graduate School of Engineering, University of Hyogo, 2167 Shosha, Himeji, Hyogo 671-2280, Japan
b Department of Life Science, Graduate School of Life Science, University of Hyogo, Hyogo 678-1297, Japan
a r t i c l e i n f o
a b s t r a c t
NylBꢀ carboxylesterase, which is 88% homologous to functional 6-aminohexanoate-dimer hydrolase
(NylB) from Arthrobacter sp., possesses trace synthetic activity [0.0004 ␮mol min−1 mg−1 (U/mg)] from
6-aminohexanoate (Ahx) to its oligomers in 90% tert-butyl alcohol. The synthetic activity and the ratio
of the synthetic activity to the hydrolytic activity were significantly affected by amino acid substitutions
at positions 181, 266 and 370. The synthetic activity was enhanced to 2.7 U/mg by G181D-H266N sub-
stitutions, and the activity was further enhanced in the G181D-H266N-D370Y triple mutant to a level
approximately 104-fold greater than the parental carboxylesterase form (3.4 U/mg), which was nearly
equal to the ordinary hydrolytic activity in water (type A-mutants). Type A-mutants possessed more
than 50% of the 6-aminohexanoate-linear dimer (Ald)-hydrolytic activity at 0–70% tert-butyl alcohol, but
the synthetic reaction became predominant at 85–90% tert-butyl alcohol. In contrast, type B-mutants
(G181E-H266N and G181N-H266N) possessed quite low levels of Ald-hydrolytic activity (<0.01 U/mg) at
0–70% tert-butyl alcohol. However, both the hydrolytic and synthetic activities were enhanced at higher
concentrations, and the maximum activity was obtained at 90% tert-butyl alcohol for both hydrolysis
and synthesis. In a type C-mutant (R187S-F264C-D370Y), the Ald-hydrolytic activity was enhanced to
approximately 80-fold that of the parental carboxylesterase, but the mutant barely demonstrated any
synthetic activity. On the basis of the three-dimensional structure of the Ald-bound enzyme and a kinetic
study for typical mutant enzymes, we propose that the efficient enzymatic syntheses of nylon-6 units
were achieved by (i) stable binding of the 1st-Ahx at the N-terminal region with Asp181, (ii) interaction
of the 2nd-Ahx at the C-terminal region with Tyr370, and (iii) motion of Tyr170 that generated a closed
form in the catalytic center of Ald hydrolase.
Article history:
Received 3 November 2009
Received in revised form 16 January 2010
Accepted 12 February 2010
Available online 18 February 2010
Keywords:
Enzymatic synthesis
Unnatural amide compound
6-Aminohexanoate-dimer hydrolase
Nylon oligomer
␤-Lactamase
Carboxylesterase
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
esterification reactions, using lipases [2–8]. Generally, enzymatic
synthesis of amides is less efficient than that of esters, although
Enzymatic catalysis in organic solvents has been recognized
as an effective tool to produce various esters and peptides [1].
In reactions such as RCOOH + H2NR’ (HO-R’) ꢀ RCO-NHR’ (RCO-
OR’) + H2O, it is possible to shift the thermodynamic equilibrium
of the reaction toward product formation rather than hydrolysis
by limiting the water concentration. Various attempts to produce
fatty acid esters, sugar esters, and polyesters in organic solvents
have been studied using the reverse reaction of hydrolysis or trans-
peptide synthesis using proteinases (such as pepsin, subtilisin,
trypsin, ␣-chymotrypsin, and thermolysin) in organic solvents has
been achieved [9–14].
In contrast to the extensive study of esters/peptides-syntheses,
the production of unnatural amides by enzymatic reactions has
scarcely been studied thus far. In the chemical processes that
condense hexamethylenediamine and adipic acid directly (nylon
6-6 production) or ring-cleavage polymerization of ␧-caprolactam
(nylon-6 production), reactions at high temperature (220–300 ◦C)
for >10 h are required to complete these polymerization reac-
tions. In the chemical synthesis of peptides, highly reactive
carboxylesters are prepared prior to the condensation reaction with
the amino group to enhance the reactivity of inactive carboxyl
groups with free amino groups at a lower temperature. Other-
wise, condensation-reagents such as carbodiimide are generally
Abbreviations: Ahx, 6-aminohexanoate; Ald, 6-aminohexanoate-linear dimer;
NylB, 6-aminohexanoate-dimer hydrolase; NylBꢀ, a protein with 88% homology to
NylB encoded by the plasmid pOAD2; Hyb-24, a NylB/NylBꢀ hybrid protein.
∗
Corresponding author. Tel.: +81 792 67 4891; fax: +81 792 67 4891.
E-mail address: negoro@eng.u-hyogo.ac.jp (S. Negoro).
1381-1177/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcatb.2010.02.006

82
Y. Kawashima et al. / Journal of Molecular Catalysis B: Enzymatic 64 (2010) 81–88
used [15–17]. However, the byproducts formed in the chemi-
cal processes are environmentally undesirable. Concerning the
enzymatic system responsible for the metabolism of unnatural
amide compounds, we have been studying the degradation of a
byproduct from the manufacture of nylon-6, 6-aminohexanoate
(Ahx)-oligomer (or nylon oligomer), by Arthrobacter sp. strain KI72
[18,19]. We found that the Ahx-linear dimer (Ald) is hydrolyzed by
Ald hydrolase (NylB) according to the following reaction.
H2N-(CH2)5CO-NH-(CH2)5-COOH + H2O → 2H2N-(CH2)5-COOH
6-Aminohexanoate-linear dimer (Ald)
6-Aminohexanote (Ahx)
Biochemical study revealed that strain KI72 also produces the
NylBꢀ protein, which is 88% homologous to NylB, and that NylBꢀ
possesses approximately 0.5% of the level of the Ald-hydrolytic
activity of NylB [20]. However, NylBꢀ is classified rather as a car-
boxylesterase, as the enzyme favors carboxylesters with short
acyl chains as substrates [21–26]. A NylB/NylBꢀ-hybrid (Hyb-24),
which was suitable for x-ray crystallographic analysis, contained
five amino acid replacements (T3A, P4R, T5S, S8Q, and D15G)
in the NylBꢀ protein and has been shown to possess levels of
Ald-hydrolytic activity that are similar to NylBꢀ [21,22] (Fig. 1).
On the basis of their three-dimensional structures, NylB/NylBꢀ
are classified as members of a penicillin-recognizing family of
serine-reactive hydrolases, and they utilize Ser112-Lys115-Tyr215
residues as catalytic triads [22,23]. Two substitutions (G181D and
H266N) in Hyb-24 have been shown to be sufficient to increase
the level of the Ald-hydrolytic activity to that of the wild-type
NylB enzyme (Fig. 1, see Hyb-24DN). In Hyb-24DN, Asp181-COO-
stabilizes the substrate binding by electrostatic interactions with
Ald-NH3+, while Asn266 makes suitable contacts with Ald and
improves the electrostatic environment in the N-terminal region of
Ald [23,24]. Moreover, the D370Y substitution in Hyb-24DN, which
stabilizes the C-terminal region of Ald, improves the kcat/Km value
of Hyb-24DN 5-fold by stabilizing the binding at the N-terminal and
C-terminal regions of the substrate [24]. Recently, we found that,
by using PCR-based directed evolution to optimize Ald-hydrolytic
activity from the common carboxylesterase (Hyb-24), there existed
an alternative conformation involving R187S-F264C-D370Y triple
substitutions, and that the activity was enhanced up to 80-fold
compared to the parental Hyb-24 in the directedly evolved Hyb-
S4M94 enzyme [25,26].
In most Ser-reactive hydrolases, the catalytic sites are exposed
to solvent [27–30]. However, in nylon-oligomer hydrolase (Hyb-
24DN and its related enzymes), Ald-binding induces a large
movement in the loop region (Asn167-Val177) and a flip-flop of
Tyr170, thereby resulting in the transition from an open to a closed
form. This induced fit motion removes water molecules from the
catalytic centers at the acylation stage [23]. In the dehydration
reaction required for amide synthesis, generation of a hydrophobic
environment in the catalytic center is expected to provide a suitable
environment for the functioning of the synthetic reaction. Thus,
the nylon-oligomer degrading enzymes provide us with a suitable
system to study the biochemical basis of amide-synthetic reactions.
In this study, we examined the possibility of producing unnatu-
ral amide compounds that contain a nylon-6 unit by using various
NylB/NylBꢀ mutants, and we have analyzed the structural require-
ments for effective amide-synthetic reactions.
Fig. 1. The effect of amino acid substitutions on enzyme activities. Hyb-24 is a
NylB/NylBꢀ hybrid protein constructed from conserved PvuII sites located 24 amino
acid residues downstream of the initiation codons and contains five amino acid
replacements (T3A, P4R, T5S, S8Q, and D15G) in the NylBꢀ protein. Open boxes
and shaded boxes represent the NylBꢀ and NylB regions, respectively. Amino acid
residues substituted in the mutant enzymes are shown as one-letter codes in the
open boxes. The synthesis of Ahx to Ald (assay S) and the hydrolysis of Ald to
Ahx (assay HB) were assayed in 90% tert-butyl alcohol/10% buffer A using purified
enzymes. The Ald-hydrolytic activities of the His-tagged purified enzymes (Hyb-
24NN, Hyb-24EN and Hyb-24NY) were assayed in buffer A using 10 mM Ald (assay
HA). The hydrolytic activities of the other His-tagged enzymes have been previ-
ously measured using “assay HA” [22–26]. On the basis of the synthesis/hydrolysis
activity ratio (assay S/assay HA), the Hyb-24 mutants were classified as type A
(ratio = 0.6–1.1), type B (ratio = 8–36), or type C (ratio < 0.01). Mutants that possessed
neither detectable synthetic activity nor hydrolytic activity were classified as type
D.
stitutions), we carried out site-directed mutagenesis using a
PCR-based “modification of restriction-site (MR)” method [31],
using the Hyb-24N gene (Hyb-24 having a H266N substitution) as
the template DNA and the specific primers shown in Table 2.
2.2. Enzyme purification and protein concentration
The mutant enzymes were expressed in Escherichia coli KP3998
using the pKP1500 vector and purified as described previously
[21,22]. In summary, the cells (grown in a 250 ml culture of Terrific
broth medium containing 25 mg of ampicillin) were harvested by
centrifugation, washed with buffer A (composition: 20 mM phos-
phate buffer containing 10% glycerol, pH 7.3), and resuspended
in 60 ml buffer A. The cells were lysed by sonication, the lysate
was centrifuged, and the supernatant was loaded onto a Hi-Trap
Q-Sepharose (GE Healthcare Bio-Science AB, Uppsala, Sweden) col-
umn (5 ml) equilibrated with buffer A. After washing with buffer
A, the enzyme was eluted with a total of 300 ml of the same buffer
containing a linear NaCl gradient from 0 to 0.5 M. The fractions con-
taining the enzyme were pooled and concentrated to a volume of
2. Materials and methods
2.1. Enzymes, plasmids and site-directed mutagenesis
The mutant enzymes and plasmids used in this study are listed
in Table 1 [21–26]. To obtain Hyb-24EN (Hyb-24 having G181E-
H266N substitutions), Hyb-24NN (Hyb-24 having G181N-H266N
substitutions) and Hyb-24NY (Hyb-24 having H266N-D370Y sub-

Y. Kawashima et al. / Journal of Molecular Catalysis B: Enzymatic 64 (2010) 81–88
83
Table 1
Enzymes and plasmids.
Abbreviations
Hyb-24
Characteristics
Reference
[21,22]
NylB (Wild-type 6-aminohexanoate-linear dimer (Ald) hydrolase from Arthrobacter sp.
KI72)/NylBꢀ (Wild-type carboxylesterase with the ␤-lactamase fold from Arthrobacter sp. KI72,
88% homology to NylB), a hybrid protein constructed from conserved PvuII sites located
24-amino acid residues downstream of the initiation codons (NylBꢀ containing
T3A/P4R/T5S/S8Q/D15G substitutions)
Hyb-24D
Hyb-24N
Hyb-24Y
Hyb-24 containing G181D substitution
Hyb-24 containing H266N substitution
Hyb-24 containing D370Y substitution
Hyb-24 containing G181D/H266N substitutions
Hyb-24 containing G181D/D370Y substitutions
Hyb-24 containing H266N/D370Y substitutions
Hyb-24 containing G181D/H266N/D370Y substitutions
Hyb-24 containing Y170F/G181D/H266N substitutions
Hyb-24 containing G181E/H266N substitutions
Hyb-24 containing G181H/H266N substitutions
Hyb-24 containing G181K/H266N substitutions
Hyb-24 containing G181N/H266N substitutions
An Ald-hydrolytic activity-enhanced mutant obtained from Hyb-24 by 4 cycles of mutagenic
PCR/DNA-shuffling (Hyb-24 containing A61V/A124V/R187S/F264C/G291R/G338A/D370Y
substitutions)
[25]
[24]
[25]
[23]
[25]
This study
[24]
[23]
This study
[22]
[22]
This study
[26]
Hyb-24DN
Hyb-24DY
Hyb-24NY
Hyb-24DNY
Hyb-24FDN
Hyb-24EN
Hyb-24HN
Hyb-24KN
Hyb-24NN
Hyb-S4M94
pHY3
A hybrid plasmid composed of the vector pKP1500 and a 1.4-kb EcoRI/HindIII fragment
containing the Hyb-24 gene
A hybrid plasmid composed of the vector pKP1500 and a 1.4-kb EcoRI/HindIII fragment
containing the Hyb-S4M94 gene
[21,22]
[26]
pS4M94
1 ml using a Centriprep YM-10 (Millipore Inc.). The concentrated
enzyme was applied onto a Sephacryl S-200 High-Resolution gel-
filtration (GE Healthcare Bio-Science AB, Uppsala, Sweden) column
that had been equilibrated with buffer A. The same buffer was used
to elute the enzyme. The fractions containing the Hyb-24 related
enzyme (approximately 1 ml) were pooled, and the enzyme solu-
tion was again applied onto a Hi-Trap Q-Sepharose column. The
enzyme was eluted as described above. All purification and con-
centration steps were carried out at 4 ◦C. Typically, a yield of 10 mg
of enzyme per 250 ml of culture was achieved.
For Hyb-24NN, Hyb-24EN and Hyb-24NY, a His-tagged region
was fused to the N-terminus of the mutant enzymes using the
pQE-80L vector (Qiagen GmbH, Hilden, Germany). The His-tagged
enzymes were expressed in E. coli JM109 and purified to homo-
geneity by conventional methods [22].
of the synthetic reaction, we assayed the enzyme activity using
“assay S”, except that various concentrations of Ahx were used.
Kinetic parameters (kcat and Km values) were evaluated by directly
fitting the Michaelis–Menten equation to the data using GraphPad
prism, version 5.01 (GraphPad, San Diego, CA USA). The kcat val-
ues were expressed as the turnover numbers per subunit (Mr of
the subunit: 42,000). As 10 mM Ahx generates 5 mM Ald stoichio-
metrically, the reverse reaction (i.e., the hydrolysis of Ald) in 90%
tert-butyl alcohol/10% buffer A was performed using 5 mM Ald (this
was chemically synthesized in our laboratory), and decreases in the
levels of Ald were quantified using HPLC (designated hereafter as
“assay HB”).
We have confirmed that adding the histidine tag to the N-
terminus of Hyb-24-related enzymes barely affected the original
Ald-hydrolytic activity. The hydrolytic activity of His-tagged puri-
fied enzymes (except for Hyb-24NN, Hyb-24EN and Hyb-24NY)
has been identified in water (buffer A without organic solvent)
for 10 mM Ald at 30 ◦C (assay HA) [22–26]. For Hyb-24NN, Hyb-
24EN and Hyb-24NY, the hydrolytic activity was similarly assayed
using the purified His-tagged enzymes using “assay HA”. We used
the activity ratio (assay S/assay HA) as a parameter to describe
the amide-synthetic ability and classification of the mutants (types
A–D).
The protein concentrations were assayed using the Lowry-Folin
method.
2.3. Enzymatic synthesis and hydrolysis of Ahx oligomers
Synthetic activities were assayed at 30 ◦C using 10 mM Ahx
(Nakarai tesque, Kyoto, Japan) in 90% tert-butyl alcohol contain-
ing 10% buffer A (Fig. 2A). For the qualitative detection of synthetic
activity, the samples were developed using a solvent mixture
(1-propanol:water:ethylacetate:ammonia = 24:12:4:1.3), and the
degradation products were detected by spraying with 0.2% nin-
hydrin solution [22]. For quantitative assays, 10 ␮l aliquots were
sequentially sampled, fractionated using a C18 reverse phase HPLC
column (TSK-GEL ODS-80Ts, TOSOH Co., Tokyo, Japan), and the
increase in Ald was monitored by following the absorbance at
210 nm (hereafter designated as “assay S”). To study the kinetics
One unit of enzyme activity was defined as the amount of
enzyme capable of catalyzing the synthesis (assay S) or hydrolysis
(assay HA and HB) of 1 ␮mol of amide linkage in 1 min.
2.4. The three-dimensional model of the protein structure
The structure of the Hyb-24DNY-A112/Ald complex (PDB ID
code: 2ZMA) [24] has previously been determined. The figure
showing the three-dimensional model was generated using the
“MolFeat” program (ver. 3.6, FiatLux Co., Tokyo, Japan).
Table 2
Primer DNA used for site-directed mutagenesis of the Hyb-24N gene.
3. Results
Mutation
Primer
Sequence
5ꢀ-AGCCGGCCGAGCGTTCGTGGGTCTGCACC-3ꢀ
5ꢀ-AGCCGGCCGAGCGGTTGTGGGTCTGCACC-3ꢀ
5ꢀ-GTGTAGGGATCGGGCCACG-3ꢀ
3.1. Optimization of the solvent system for efficient amide
synthesis
G181E
G181N
D370Y
RGmutE1
RGmutN1
RDmutY1
To achieve an efficient condensation reaction between free
amino- and carboxyl-group in organic solvents, the choice of a suit-
Site-directed mutagenesis was carried by a “modification of restriction-site (MR)”
method [31]. Mutated sites in the primer sequences are underlined.

84
Y. Kawashima et al. / Journal of Molecular Catalysis B: Enzymatic 64 (2010) 81–88
Fig. 2. TLC analysis of the amide synthesis catalyzed by Hyb-24DN. (A) The reaction scheme of Ald synthesis. (B) Enzyme reactions were performed in 90% solvent/10% buffer
A using Hyb-24DN (0.3 mg/ml) and 10 mM Ahx, and the synthesized products were analyzed by TLC. Slot 1, methyl alcohol; slot 2, ethyl alcohol; slot 3, n-propyl alcohol; slot
4, n-butyl alcohol; slot 5, tert-butyl alcohol; slot 6, dimethylsulfoxide. (C) Enzyme reactions were performed in 90% tert-butyl alcohol using Hyb-24DN (0.3 mg/ml) and 10 mM
Ahx. Reaction mixture (1 ␮l aliquot) was sequentially removed and spotted onto a TLC plate. (D) Enzyme reactions were performed in various concentrations (50–98%) of
tert-butyl alcohol dissolved in buffer A using Hyb-24DN (0.3 mg/ml) and 10 mM Ahx. After a 1 h reaction, the reaction mixture (1 ␮l aliquot) was spotted onto a TLC plate.
Authentic Ahx and Ald (10 mM each) were spotted as controls. Ahx (10 mM) was dissolved in 90% tert-butyl alcohol and analyzed to determine the effect of the solvent in
the sample on the mobility of Ahx (slot Ahx*). The positions of the spots for Ahx (A1), Ald (A2), Ahx-trimer (A3), and Ahx-tetramer (A4) are shown in panels B–D.
able solvent is often critical. Although the optimum pH for the
hydrolytic reaction of NylB is known to be pH 8–9 [32,33], the
enzyme activity is rather stable in 20 mM phosphate buffer con-
taining 10% glycerol at pH 7.3 (buffer A). Therefore, to check the
suitability of a potential solvent system, we analyzed the conver-
sion of Ahx to synthesized product using Hyb-24DN (0.3 mg/ml) in
various organic solvents (30 kinds) that were dissolved or emulsi-
fied in buffer A (Fig. 2).
TLC analysis of the reaction products revealed that syn-
thesized products were identified in reaction mixtures that
contained 90% of propyl alcohol (n-, iso-), butyl alcohol (n-, iso-
, tert-), or pentyl alcohol (n-, iso-) (Fig. 2B). In the reaction
mixture containing tert-butyl alcohol, Ald was formed within
a 1 min reaction, whereas linear Ahx trimers and tetramers
were formed sequentially (Fig. 2C), thereby demonstrating that
the oligomers were synthesized by exo-type modes. How-
ever, no detectable amounts of products were identified for
23 kinds of solvent systems, including mixtures that contained
90% of methanol, ethanol, 2-methoxyethanol, acetonitrile, glyc-
erin, acetone, ethyleneglycol, diethyleneglycol, triethyleneglycol,
polyethyleneglycol, dimethylsulfoxide, 2-ethylhexyl alcohol, n-
octyl alcohol, benzyl alcohol, n-hexane, n-heptane, chlorobenzene,
methyl-iso-butylketone, di-iso-butylketone, ethylenedichloride,
decahydronaphthalene, dioctylphthalate and di-n-butylbsebacate.
We found that the addition of these solvents to the enzyme
solution resulted in precipitation of the enzyme, thereby elim-
inating functional enzymatic activity from the solution. To
examine the concentration dependence of tert-butyl alcohol
on the enzyme activity, we performed the synthetic reaction
with Hyb-24DN (0.3 mg/ml) in various concentrations of tert-
butyl alcohol. We found that efficient product formation was
achieved within the concentration range of 85–92% alcohol, but
abruptly decreased at higher concentrations (especially at >93%)
(Fig. 2D).
The initial rates of synthetic activity of Hyb-24DN (0.01 mg/ml)
were determined by quantifying the increase of Ald using HPLC

Y. Kawashima et al. / Journal of Molecular Catalysis B: Enzymatic 64 (2010) 81–88
85
Fig. 3. HPLC analysis of amide synthesis catalyzed by Hyb-24DN. (A) The enzyme
reaction was carried out in 90% tert-butyl alcohol/10% buffer A using Hyb-24DN
(0.01 mg/ml) and 10 mM Ahx, and the initial rates of the synthetic reactions were
quantified by the increase of Ald (ꢁ) (assay S) and by the decrease of Ahx (ꢂ). (B)
The initial rates of the synthetic reactions in 90% organic solvent/10% buffer A using
Hyb-24DN (0.01 mg/ml) and 10 mM Ahx were quantified by the increase of Ald. (C)
The enzyme reaction was carried out using Hyb-24DN (0.3 mg/ml) in 90% tert-butyl
alcohol/10% buffer A using 10 mM Ahx, and the remaining Ahx in the reaction mix-
ture was quantified using HPLC. Amide synthesis was expressed as the percentage
of the amount of incorporated Ahx divided by the initial amount of Ahx.
(Fig. 3A). The observed reaction rates [2.72 ␮mol min−1 mg−1
(U/mg)] that were found in 90% tert-butyl alcohol agreed well
with the initial rates identified by the decrease of Ahx (2.55 U/mg).
Among the solvent systems using the three isomers of butyl alcohol
(90% each), the highest activity was found when using tert-butyl
alcohol. Only approximately 10% of the enzymatic activity was
identified when using n- and iso-butyl alcohol. Similarly, 40–50%
of the activity was found in n- and iso-pentyl alcohol (Fig. 3B). At
85% tert-butyl alcohol, the initial rates of the synthetic reaction
exceeded the rates of the hydrolytic reaction, but the activity dras-
tically decreased at 95% tert-butyl alcohol, due to precipitation of
the enzyme (Fig. 4A).
In 90% tert-butyl alcohol, the maximum yield (approximately
82%) was obtained within 30 min using Hyb-24DN (0.3 mg/ml),
and this conversion ratio was maintained stably even after extend-
ing the reaction time to 100 min (Fig. 3C). Moreover, the solvent
(90% tert-butyl alcohol) was mixed homogeneously with water and
did not form a biphasic layer. On the basis of these findings, we
established 90% tert-butyl alcohol/10% buffer A as the solvent sys-
tem for the standard assay for the synthetic reaction (designated
“assay S”).
Fig. 4. The effect of the concentration of tert-butyl alcohol on the initial rates of
the synthetic and hydrolytic reactions. Enzyme reactions were performed in vari-
ous concentrations of tert-butyl alcohol dissolved in buffer A using Hyb-24 mutants
(A–E) and 10 mM Ahx, and the initial rate of the synthetic reactions were quantified
by the increase of Ald (ꢃ). Enzyme reactions were performed in various concen-
trations of tert-butyl alcohol using Hyb-24 mutants (A–E) and 5 mM Ald, and the
initial rate of the hydrolytic reactions were quantified by the decrease of Ald (ꢁ).
(A) Hyb-24DN, (B) Hyb-24DNY, (C) Hyb-24NN, (D) Hyb-24EN, and (E) Hyb-S4M94.
The hydrolytic and synthetic activity at each tert-butyl alcohol concentration is also
shown in supplementary Tables S1 and S2, respectively.

86
Y. Kawashima et al. / Journal of Molecular Catalysis B: Enzymatic 64 (2010) 81–88
3.2. The effect of amino acid alterations on the amide-synthetic
activity
On the basis of the three-dimensional structures of Hyb-24 and
its mutants, we have proposed that (i) Asp181-COO− stabilizes the
binding of substrate through electrostatic interactions at the N-
terminal region of Ald, (ii) Asn266 makes suitable contacts with Ald
and improves the electrostatic environment, and (iii) Tyr370 sta-
bilizes the binding of Ald through hydrogen-bonding/hydrophobic
interactions at the C-terminal region of Ald [22–26]. Therefore, we
first examined the amide-synthetic activity for enzymes having
mutations at positions 181, 266 and/or 370 (Fig. 1). As expected,
trace synthetic activity (0.0004 U/mg) that was originally retained
in the parental Hyb-24 enzyme was significantly affected by substi-
tutions at these substrate-binding sites. Interestingly, as described
below, the ratio of the synthetic activity (in 90% tert-butyl alcohol)
to the hydrolytic activity (in water) (i.e., assay S/assay HA) varied
drastically among the mutants. From the ratio, we classified the
mutant enzymes as follows: ratio = 0.6–1.1 (type A); 8–36 (type
B); <0.01 (type C). Mutants that possessed neither synthetic nor
hydrolytic activity were classified as type D.
Type A: G181D, H266N and D370Y substitutions in Hyb-24
increased the synthetic activity additively up to 3.4 U/mg (approxi-
mately 104-fold increase from Hyb-24). It should be noted that the
synthetic/hydrolytic activity ratios were observed consistently to
be between 0.77 and 1.1 among the Hyb-24D, Hyb-24DN and Hyb-
24DNY mutants, thereby demonstrating that the level of synthetic
activity roughly correlated with that of the hydrolytic activity. Gen-
erally, the hydrolytic activity in the organic solvent/water mixed
system was expected to be lower than the activity in water. We
found that type A-mutants possessed more than 50% of the Ald-
hydrolytic activity in 0–70% tert-butyl alcohol, but the synthetic
reaction became predominant at 85–90% tert-butyl alcohol (Fig. 4A
and B). Hyb-24FDN (Hyb-24DN containing a Y170F substitution)
decreased both the synthetic and hydrolytic activity to approxi-
mately 2% of Hyb-24DN (0.05 U/mg), and the synthetic/hydrolytic
activity ratio was maintained at 1.0 (Fig. 1). This suggested that
Tyr170 plays an important role not only in the hydrolysis of amides,
but also in synthesis.
Type B: A single D181E and D181N substitution in Hyb-24DN
decreased the hydrolytic activity to 0.3% and 3.9% of the level
of Hyb-24DN, respectively (Fig. 1). However, the mutants still
possessed 17% and 51% of the level of the synthetic activity of
Hyb-24DN, respectively. Interestingly, type B-mutants (Glu181-
and Asn181-mutants) possessed quite low levels of Ald-hydrolytic
activity (<0.01 U/mg) at <70% tert-butyl alcohol (Fig. 4C and
D). However, both the hydrolytic and synthetic activities were
enhanced at higher concentrations, and the maximum activity was
obtained at 90% tert-butyl alcohol both for hydrolysis and synthe-
sis. In addition, it should be noted that the synthetic activity was
higher than the hydrolytic activity even at 70% tert-butyl alcohol,
although the specific synthetic activity was lower than the activity
of Hyb-24DN and Hyb-24DNY.
Type C: The hydrolytic activity of Hyb-24 (in water) was
enhanced 8-fold in the Hyb-24Y mutant by a single D370Y substi-
tution, and this has been demonstrated to improve Ald-binding at
the C-terminal region of Ald [25]. The activity was further enhanced
10-fold in the Hyb-S4M94 mutant [26]. Furthermore, we have
identified that efficient Ald-binding was due to Ser187-Cys264
(these are stabilizing at the N-terminal region) and Tyr370 (this
is stabilizing at the C-terminal region) [26]. Moreover, the enzyme
assay of a newly constructed Hyb-24NY mutant (Hyb-24 having
H266N-D370Y substitutions) demonstrated that the presence of
Asn266, which was an effective Ald-binding residue in Hyb-24DN,
increased the Ald-hydrolytic activity of Hyb-24Y approximately 4-
fold (Fig. 1). However, all Gly181-enzymes (Hyb-24Y, Hyb-24NY
Fig. 5. The effect of the concentration of Ahx on the synthetic activity. (A) The Ald-
synthetic activity was assayed using the purified enzymes using “assay S”, except
that various concentrations of Ahx were used. (ꢁ) Hyb-24DNY, (᭹) Hyb-24DN, (ꢁ)
Hyb-24EN, and (ꢃ) Hyb-24NN. (B) The kinetic parameters (the kcat and Km val-
ues) were evaluated by fitting the data to the Michaelis–Menten equation directly
using the program “GraphPad prism, version 5.01” (GraphPad, San Diego, CA, USA).
The kcat values are expressed as the turnover numbers per subunit (the Mr of the
subunit = 42,000).
and Hyb-S4M94) possessed hardly any synthetic activity in 90%
tert-butyl alcohol (<0.01 U/mg using “assay S”). In addition, enzyme
assays conducted using various concentrations of tert-butyl alco-
hol revealed that the Ald-hydrolytic activity of Hyb-S4M94 in 90%
tert-butyl alcohol was >50% of the activity in water, whereas the
synthetic activity was barely detectable in any concentration of
tert-butyl alcohol (Figs. 1 and 4E).
Type D: The amino acid substitutions to
a basic amino
acid residue at position 181 in Hyb-24DN (i.e., His181, Lys181)
decreased Ald-hydrolytic activity drastically to an undetectable
level (<0.1% of the level of Hyb-24DN). These mutations also
resulted in elimination of the amide-synthetic activity.
3.3. Kinetic study
We found that enzymatic amide synthesis in 90% tert-butyl
alcohol follows Michaelis–Menten kinetics for the type A mutants
(Hyb-24DN, Hyb-24DNY), although it was only possible to perform
kinetic studies at concentrations of Ahx lower than 10 mM due
to the limited solubility of Ahx in 90% tert-butyl alcohol (Fig. 5).
The kcat/Km value of Hyb-24DN (0.52 s−1 mM−1) was improved by
D370Y substitution in Hyb-24DNY (0.60 s−1 mM−1). In Hyb-24EN
(i.e., the Glu181-enzyme, a type B mutant), high affinity for Ahx was
retained (Km = 10.3 mM), while the kcat was approximately 18% that
of Hyb-24DN (Fig. 5B). In contrast, in Hyb-24NN (i.e., the Asn181-
enzyme, another type B mutant), the synthetic activity increased
linearly between 1.0 and 10 mM Ahx, and therefore, the Km value
was estimated to be much higher than the saturated Ahx concen-
tration (Fig. 5A). For the types C and D mutants, it was not possible
to measure the Km and kcat values for the synthetic reactions due
to their low activities. On the basis of the three-dimensional struc-
tures of the enzyme/Ald complex, we will discuss the structural

Y. Kawashima et al. / Journal of Molecular Catalysis B: Enzymatic 64 (2010) 81–88
87
requirements for efficient amide synthesis and the relationship
between the synthetic and hydrolytic activities.
4. Discussion
In the thermodynamic-controlled amide/ester synthesis, the
water content should ordinarily be maintained at a level lower than
0.5% to shift the reaction toward synthesis rather than hydroly-
sis. The optimal water content will depend on the stability of the
protein and the distribution of the water in the organic solvent
[9]. However, we found that the synthetic reaction of Hyb-24DN
and its mutants became even more favorable in the presence
of 15–30% water (Fig. 4). As the catalytic process should pass
through the same transition state in both the hydrolytic and
synthetic reactions, the rate of the synthetic reaction is rele-
vant to the mechanism of the hydrolytic reaction. If this catalytic
process is conserved among the series of Hyb-24 mutants, the
synthetic/hydrolytic activity ratio should be very similar among
the mutant enzymes. However, we found that a small number of
amino acid alterations in the 6-aminohexanoate-dimer hydrolase
(Hyb-24DN) affected the synthetic/hydrolytic activity ratios sig-
nificantly under standard assay conditions [i.e., the hydrolysis in
water assay (assay HA), the hydrolysis in 90% tert-butyl alcohol
assay (assay HB), and the synthesis in 90% tert-butyl alcohol assay
(assay S)].
In type A-mutants (Asp181-enzymes), G181D-H266N-D370Y
substitutions in the parental enzyme (Hyb-24) cumulatively
increased both the synthetic activity (in 90% tert-butyl alcohol)
(assay S) and the normal hydrolytic activity (i.e., in water, assay
HA), and the ratio was maintained between 0.6 and 1.1. We esti-
mate that the 1st-Ahx-COOH (i.e., the N-terminal Ahx-moiety in
Ald) and the 2nd-Ahx-NH2 (i.e., the C-terminal Ahx-moiety in Ald)
are localized close to the Ser112-Lys115-Tyr215 catalytic triads
during the dynamic motion of the enzyme complex (Fig. 6A and B).
Tyr170 appears to contribute to both the amide-synthetic reaction
and the Ald-hydrolytic reaction [23], as the Y170F substitution in
Hyb-24DN decreased the activity to approximately 2% of the level
of Hyb-24DN (see Fig. 1, Hyb-24FDN). In the type D-mutants (i.e.,
the Lys181/His181-enzymes), both the synthetic and hydrolytic
activities were found to be lower than the detection limit (<0.01%
the activity of Hyb-24DN). Therefore, electrostatic repulsion of the
positive charges at Lys181-NH3 and His181-imidazole N+ should
+
drastically diminish the substrate-binding ability. We estimate that
stable binding of the 1st-Ahx was achieved primarily by an elec-
trostatic stabilization effect from Asp181-COO−, even in a solvent
system containing 10% water. Moreover, in addition to the effect at
position 181, interaction with Asn266 and Tyr370 should enhance
the substrate-binding ability cumulatively, resulting in an increase
in the synthetic and hydrolytic activities (Fig. 6A and B). However,
it should be noted that the D370Y substitution has a more moder-
ate effect on the synthetic activity than on the hydrolytic activity.
That is, the kinetic study revealed that for the hydrolytic reac-
tion in water, the D370Y substitution improves the Ald-binding
of Hyb-24DN (Km = 27 mM) approximately 14-fold (Km = 2.0 mM
for Hyb-24DNY) and improves the kcat/Km value of Hyb-24DN
(0.34 s−1 mM−1) approximately 5-fold (kcat/Km = 1.58 s−1 mM−1 in
Hyb-24DNY) [24]. In contrast, for the synthetic reactions in 90% tert-
butyl alcohol, the D370Y substitution hardly affects the Km value
for Ahx (6.2–6.7 mM), and it increases the kcat/Km by only 1.2-fold
(Fig. 5). This result demonstrates that for the hydrolytic reaction,
the substrate binding was enhanced by dual effects for a single Ald
molecule both at the C- and N-terminal regions of the substrate,
whereas for the synthetic reaction, Asp181 and Tyr370 stabilize the
1st-Ahx and 2nd-Ahx independently, and the cooperative effect is
not estimated to be as significant.
Fig. 6. Enzyme–substrate interaction at the catalytic cleft of nylon-oligomer hydro-
lase. (A) The structure of the Hyb-24DNY-A112/Ald complex (PDB ID code: 2ZMA)
identified by X-ray crystallography [24]. Figures showing the three-dimensional
models were generated with the “MolFeat” program (ver. 3.6, FiatLux Co., Tokyo,
Japan). (B) A model showing the catalytic centers of Hyb-24DNY. The 1st-Ahx is
maintained stably by an interaction with Asp181-COO−, and the close position-
ing of the 2nd-Ahx-NH2 to the 1st-Ahx-COOH results in a condensation reaction.
(C) A model showing the catalytic centers of Hyb-24Y. Inefficient binding of the
1st-Ahx (marked by an X) resulted in the inability of the enzyme to catalyze the
amide-synthetic reaction.

88
Y. Kawashima et al. / Journal of Molecular Catalysis B: Enzymatic 64 (2010) 81–88
The type B-mutants (i.e., the Glu181/Asn181-enzymes) hardly
5. Conclusions
possessed any synthetic or hydrolytic activities in 0–70% tert-butyl
alcohol (Fig. 4C and D). However, both activities were significantly
increased at higher tert-butyl alcohol concentrations. Therefore, we
estimated that the increased hydrolytic activity in organic solvent
is correlated with the appearance of the synthetic activity. A single
D181N substitution in an Asp181-enzyme (Hyb-2) has also been
shown to decrease the affinity of the enzyme for Ald (Km = 80 mM)
[34]. Similarly, as stated above, we estimate that the Km value is
much higher than the saturated Ahx concentration in 90% tert-butyl
alcohol, as the synthetic activity increased linearly with Ahx con-
centrations in the range of 1.0–10 mM (Fig. 5). The low affinity of
Ahx in the Asn181-enzyme suggests that the electrostatic effect is
still predominant in reaction mixtures composed of 90% tert-butyl
alcohol, although the effects of the negative charge of Asp-COO−
would be weakened with increasing organic solvent concentra-
tion. In addition, because the Hyb-24NN possesses a high level of
synthetic activity (1.1 U/mg) using “assay S”, the enzyme should
possess a high kcat value for the synthetic reaction in 90% tert-butyl
alcohol. Therefore, minor structural alterations induced by the sol-
vent suitably improve the positioning of the Ahx molecule against
the catalytic residues.
We have shown that nylon-oligomer hydrolase efficiently cat-
alyzes a direct condensation reaction between a free amino-group
and a carboxyl-group to generate nylon-6 units. We propose that
the hydrophobic environment generated in the closed form of the
enzyme potentially provides a suitable environment for synthesiz-
ing amide compounds at a high efficiency, and this is preferable in
terms of industrial production.
Acknowledgments
This work was supported in part by a Grant-in-Aid for Scientific
Research (Japan Society for Promotion of Science) and grants from
the JAXA project.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.molcatb.2010.02.006.
References
We have previously reported that a D181E substitution in the
NylB-type protein (Hyb-2; Asp181-enzyme) decreases the affinity
of the enzyme for Ald (Km = 220 mM) [34]. In the Glu181-enzyme,
the longer side chain (compared to Asp) sterically hinders Ald-
binding and led to incorrect positioning of the amide linkage of
Ald against the catalytic residues. In contrast to the low hydrolytic
activity of Hyb-24EN in water (assay HA), increased Ald-hydrolytic
activity in 90% tert-butyl alcohol (assay HB) may imply that a minor
structural alteration in the catalytic cleft induced by organic solvent
allows for the effective binding of Ald. In addition, the kinetic study
of the synthetic reaction demonstrates that Hyb-24EN possesses a
relatively high affinity for Ahx (Km = ca. 10 mM), suggesting that the
enzyme holds Ahx stably in the catalytic cleft, although the kcat is
18% that of Hyb-24DN (Fig. 5).
[1] A. Schmid, J.S. Dordick, B. Hauer, A. Kiener, M. Wubbolts, B. Witholt, Nature 409
(2001) 258–268.
[2] U.T. Bornscheuer, FEMS Microbiol. Rev. 26 (2002) 73–81.
[3] M.E. Lowe, J. Lipid Res. 43 (2002) 2007–2016.
[4] U.T. Bornscheuer, C. Bessler, R. Srinivas, S.H. Krishna, Trends Biotechnol. 20
(2002) 433–437.
[5] F.J. Plou, M.A. Cruces, M. Ferrer, G. Fuentes, E. Pastor, M. Bernabé, M. Chris-
tensen, F. Comelles, J.L. Parra, A. Ballesteros, J. Biotechnol. 96 (2002) 55–66.
[6] S. Adachi, T. Kobayashi, J. Biosci. Bioeng. 99 (2005) 87–94.
[7] S. Namekawa, S. Suda, H. Uyama, S. Kobayashi, Int. J. Biol. Macromol. 25 (1999)
145–151.
[8] H. Fukuda, S. Hama, S. Tamalampudi, H. Noda, Trends Biotechnol. 26 (2008)
668–673.
[9] A. Zaks, A.M. Klibanov, J. Biol. Chem. 263 (1988) 3194–3201.
[10] F. Bordusa, Curr. Protein Pept. Sci. 3 (2002) 159–180.
[11] C. Lombard, J. Saulnier, J.M. Wallach, Protein Pept. Lett. 12 (2005) 621–629.
[12] D. Kühn, P. Dürrschmidt, J. Mansfeld, R. Ulbrich-Hofmann, Biotechnol. Appl.
Biochem. 36 (2002) 71–76.
In the type C-mutants (i.e., the Gly181-enzymes), the Ald-
hydrolytic activity was cumulatively enhanced by approximately
8-fold (Hyb-24Y), 34-fold (Hyb-24NY) and 80-fold (Hyb-S4M94)
as compared with the parental carboxylesterase (Hyb-24), while
retaining the common Gly181/Tyr370 residues. However, these
mutants hardly possessed any synthetic activity at any concentra-
tion of tert-butyl alcohol (Figs. 1 and 4E). We suggest therefore,
that inefficient binding of the 1st-Ahx maintains the enzyme struc-
ture in an open form and that the 2nd-Ahx is less likely to be
bound to the enzyme despite the presence of an effective Ald-
binding residue (Tyr370) interacting at the C-terminal region of
the substrate (Fig. 6C). These results demonstrate that, even if a cat-
alytic function is optimized for the forward reaction (hydrolysis),
the selected structure may not be suitable for the reverse reaction
(synthesis).
[13] A. Trusek-Holownia, J. Biotechnol. 102 (2003) 153–163.
[14] O.A. Adekoya, I. Sylte, Chem. Biol. Drug Design 73 (2009) 7–16.
[15] E. Valeur, M. Bradley, Chem. Soc. Rev. 38 (2009) 606–631.
[16] J.W. Bode, Curr. Opin. Drug Discov. Dev. 9 (2006) 765–775.
[17] F. Albericio, Curr. Opin. Chem. Biol. 8 (2004) 211–221.
[18] S. Negoro, Appl. Microbiol. Biotechnol. 54 (2000) 461–466.
[19] S. Negoro, Biopolymers 9 (2002) 395–415.
[20] H. Okada, S. Negoro, H. Kimura, S. Nakamura, Nature 306 (1983) 203–206.
[21] T. Ohki, N. Mizuno, N. Shibata, M. Takeo, S. Negoro, Y. Higuchi, Acta Crystallogr.
F 61 (2005) 928–930.
[22] S. Negoro, T. Ohki, N. Shibata, N. Mizuno, Y. Wakitani, J. Tsurukame, K.
Matsumoto, I. Kawamoto, M. Takeo, Y. Higuchi, J. Biol. Chem. 280 (2005)
39644–39652.
[23] S. Negoro, T. Ohki, N. Shibata, K. Sasa, H. Hayashi, H. Nakano, K. Yasuhira, D.
Kato, M. Takeo, Y. Higuchi, J. Mol. Biol. 370 (2007) 142–156.
[24] Y. Kawashima, T. Ohki, N. Shibata, Y. Higuchi, Y. Wakitani, Y. Matsuura, Y.
Nakata, M. Takeo, D. Kato, S. Negoro, FEBS J. 276 (2009) 2547–2556.
[25] T. Ohki, Y. Wakitani, M. Takeo, K. Yasuhira, N. Shibata, Y. Higuchi, S. Negoro,
FEBS Lett. 580 (2006) 5054–5058.
From the structural and functional analyses of the Hyb-24
mutants, we have summarized the enzyme/substrate interactions
responsible for efficient synthesis of nylon-6 units as follows:
[26] T. Ohki, N. Shibata, Y. Higuchi, Y. Kawashima, M. Takeo, D. Kato, S. Negoro,
Protein Sci. 18 (2009) 1662–1673.
[27] U.G. Wagner, E.I. Petersen, H. Schwab, C. Kratky, Protein Sci. 11 (2002) 467–478.
[28] M.A. McDonough, J.W. Anderson, N.R. Silvaggi, R.F. Pratt, J.R. Knox, J.A. Kelly, J.
Mol. Biol. 322 (2002) 111–122.
[29] M. Nukaga, S. Kumar, K. Nukaga, R.F. Pratt, J.R. Knox, J. Biol. Chem. 279 (2004)
9344–9352.
[30] A.R. Fersht, Enzyme Structure and Mechanism, 2nd ed., W.H. Freeman and Co.,
New York, 1984.
(i) The 1st-Ahx binding is facilitated by an interaction with
Asp181. Asn266 increases the 1st-Ahx-binding cooperatively
with Asp181.
[31] W. Ito, H. Ishiguro, Y. Kurosawa, Gene 102 (1991) 67–70.
[32] S. Kinoshita, T. Terada, T. Taniguchi, Y. Takene, S. Masuda, N. Matsunaga, H.
Okada, Eur. J. Biochem. 116 (1981) 547–551.
[33] K. Fujiyama, Y. Zhang, S. Negoro, I. Urabe, H. Okada, J. Ferment. Bioeng. 71 (1991)
298–302.
(ii) A 2nd-Ahx is stabilized by Tyr370. Movement of the flexible
loop region and rotation of Tyr170 generate the hydrophobic
environment suitable for the synthetic reaction.
(iii) The synthetic reaction proceeds by catalytic functions involv-
ing Ser112-Lys115-Tyr215. The catalysis is enhanced by the
cooperative functioning of Tyr170.
[34] H.S. Hatanaka, K. Fujiyama, S. Negoro, I. Urabe, H. Okada, J. Ferment. Bioeng. 71
(1991) 191–193.