5-Cyanovaleramide production using immobilized Pseudomonas chlororaphis B23

Article abstract of DOI:10.1016/S0968-0896(99)00157-1

A biocatalytic process for the hydration of adiponitrile to 5- cyanovaleramide has been developed which can be run to higher conversion, produces more product per weight of catalyst, and generates significantly less waste products than alternate chemical processes. The biocatalyst consists of Pseudomonas chlororaphis B23 microbial cells immobilized in calcium alginate beads. The cells contain a nitrile hydratase (EC 4.2.1.84) which catalyzes the hydration of adiponitrile to 5-cyanovaleramide with high regioselectivity, and with less than 5% selectivity to byproduct adipamide. Fifty-eight consecutive batch reactions with biocatalyst recycle were run to convert a total of 12.7 metric tons of adiponitrile to 5-cyanovaleramide. At 97% adiponitrile conversion, the yield of 5-cyanovaleramide was 13.6 metric tons (93% yield, 96% selectivity), and the total weight of 5-cyanovaleramide produced per weight of catalyst was 3150 kg/kg (dry cell weight).

Full text of DOI:10.1016/S0968-0896(99)00157-1

Bioorganic & Medicinal Chemistry 7 (1999) 2239±2245
5
-Cyanovaleramide Production using Immobilized Pseudomonas
chlororaphis B23
a
a
a
a
Eugenia C. Hann, Amy Eisenberg, Susan K. Fager, Neal E. Perkins, F. Glenn
a
a
a
b
Gallagher, Susan M. Cooper, John E. Gavagan, Barry Stieglitz,
b
a,
Susan M. Hennessey and Robert DiCosimo *
aDuPont Life Sciences, Experimental Station, PO Box 80328, Wilmington, DE 19880-0328, USA
bDuPont Agricultural Products, Experimental Station, PO Box 80402, Wilmington, DE 19880-0402, USA
Received 26 September 1999
AbstractÐA biocatalytic process for the hydration of adiponitrile to 5-cyanovaleramide has been developed which can be run to
higher conversion, produces more product per weight of catalyst, and generates signi®cantly less waste products than alternate
chemical processes. The biocatalyst consists of Pseudomonas chlororaphis B23 microbial cells immobilized in calcium alginate beads.
The cells contain a nitrile hydratase (EC 4.2.1.84) which catalyzes the hydration of adiponitrile to 5-cyanovaleramide with high
regioselectivity, and with less than 5% selectivity to byproduct adipamide. Fifty-eight consecutive batch reactions with biocatalyst
recycle were run to convert a total of 12.7 metric tons of adiponitrile to 5-cyanovaleramide. At 97% adiponitrile conversion, the
yield of 5-cyanovaleramide was 13.6 metric tons (93% yield, 96% selectivity), and the total weight of 5-cyanovaleramide produced
per weight of catalyst was 3150 kg/kg (dry cell weight). # 1999 Elsevier Science Ltd. All rights reserved.
1
5,16
Introduction
acid was also important. Bacterial strains of Bacillus,
15,16
15,16
15,16
Bacteridium,
Brevibacterium,
Micrococcus,
Pseudomonas,17 and Acinetobacter have each been
reported to produce a regioselective aliphatic nitrile
hydratase, and a variety of these strains were evaluated
as a catalyst for hydration of ADN to 5CVAM.
18
The ®rst step in the manufacture of a new herbicide
required the conversion of adiponitrile (ADN) to 5-
cyanovaleramide (5CVAM). Chemical catalysts for the
1
±3
hydration of nitriles to amides,
such as Raney cop-
4
5
per or manganese dioxide, produced signi®cant quan-
tities of adipamide (ADAM) in addition to 5CVAM at
high conversions of ADN (Scheme 1). The amount of
catalyst waste produced per kilogram of product was
also signi®cant when employing many of these chemical
catalysts. As an alternative to chemical catalytic meth-
ods, the use of enzymatic catalysis for the production of
Whole microbial cells of Pseudomonas chlororaphis B23,
®rst isolated and characterized by H. Yamada and co-
workers,1
Chemical Industry Co. (now merged with Mitsubishi
9,20,21
and subsequently employed by the Nitto
Rayon Co.) for the commercial production of acrylam-
ide from acrylonitrile,2
2±25
were found to be a highly
5
CVAM from ADN was examined.
regioselective catalyst for the production of 5CVAM.
This biocatalyst was superior in both enzyme stability
and productivity of 5CVAM (kg 5CVAM/kg biocata-
lyst/h) to all other whole-cell microbial catalysts which
were examined. The demonstration of a commercial-
scale process for the production of 5CVAM using
immobilized P. chlororaphis B23 cells is now described.
Aliphatic nitriles are readily hydrated to the corre-
sponding amides by the nitrile hydratase (EC 4.2.1.84)
of a variety of bacteria and fungi.
tion of an o-cyanocarboxylic acid amide from an ali-
phatic dinitrile, a regioselective nitrile hydratase was
required to prevent signi®cant production of the corre-
sponding diamide. When employing a bacterial cell as
catalyst, the absence of amidase activity that could
result in further hydrolysis of the amide to carboxylic
6
±14
For the prepara-
Results and Discussion
Storage stability of P. chlororaphis B23 nitrile hydratase
activity
Key words: Adiponitrile; 5-cyanovaleramide; nitrile hydratase; Pseu-
domonas chlororaphis B23; immobilization.
The nitrile hydratase activity of P. chlororaphis B23 cells
suspended in aqueous bu€er was found to be most
*
8
Corresponding author. Tel.: +1-302-695-9416; fax: +1-302-695-
114
0968-0896/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(99)00157-1

2
240
E. C. Hann et al. / Bioorg. Med. Chem. 7 (1999) 2239±2245
Scheme 1.
ꢀ
stable at ca. 5 C, and was increasingly less stable with
mixture was only 0.45 and 0.52 M, respectively, so batch
reactions which were run to produce greater than
0.45 M 5CVAM started as two-phase mixtures of aqu-
ꢀ
19
increasing temperature from 5 to 25 C. Because the
B23 cell nitrile hydratase rapidly and irreversibly lost
activity at a pH below pH 6, a bu€er was required to
maintain the pH of the cell suspension between 6.5 and
ꢀ
eous cell suspension and ADN. At 5 C, the solubility of
byproduct ADAM was only 37±42 mM in 1 to 1.5 M
5CVAM. The nitrile hydratase activity of B23 cells in
these reactions was inhibited by sodium butyrate; for
7
.5, which was the optimum pH range for stabilization
19
of the enzyme activity. Phosphate bu€er could not be
employed, as it reacted with calcium ions used to cross-
link sodium alginate during cell immobilization, produ-
cing insoluble calcium phosphate and causing
dissolution of the alginate beads. Millimolar concentra-
tions of organic acids (e.g. 20 mM n-butyric acid) had
previously been shown to markedly improve the stabi-
lity of B23 nitrile hydratase activity. P. chlororaphis
B23 cell suspensions (50% wet cell weight (wcw) sus-
pensions) containing 0.5% sodium propionate (w/v) at
pH 6.2 lost over 50% of initial nitrile hydratase speci®c
ꢀ
the production of 0.75 M 5CVAM at 5 C, the reaction
rate decreased 41% as the concentration of sodium
butyrate was increased from 10 to 40 mM.
ꢀ
Six consecutive reactions were run at 5 C to convert
0.40 M ADN (in a single-phase aqueous solution) to
5CVAM with a recycle of the unimmobilized cells; the
time required for complete conversion of ADN
increased three-fold over the course of the six reactions
as the cells lost activity. At complete conversion of
ADN, the ratio of 5CVAM to ADAM was ꢁ100:1 in
each of the six reactions. If a reaction was allowed to
continue after complete conversion of ADN, 5CVAM
was further hydrolyzed to ADAM at ca. 1±2% of the
initial rate of 5CVAM production. Consecutive batch
1
9
ꢀ
activity after three weeks at 5 C. Cell suspensions at 10mg
ꢀ
wcw/mL were stable at 5 C for as long as 6 weeks in the
presence of 23mM sodium butyrate at pH 7.0. Cell sus-
pensions at 200 mg wcw/mL in 23mM sodium butyrate
were much less stable, suggesting that higher butyrate or
propionate concentrations were required for suspensions
containing cells at or above 200 mg wcw/mL.
ꢀ
reactions were next performed at 5 C to produce a ®nal
concentration of 1.5 M 5CVAM by starting with a two-
phase mixture of aqueous cell suspension (containing
the solubility limit of ADN) and ADN; only the ®rst
reaction gave complete conversion of ADN, and an
attempt to reuse the unimmobilized cells in a second
reaction resulted in incomplete conversion of ADN at
extended reaction times. Only cell debris (no intact cells)
was recovered from this second reaction, indicating that
the unimmobilized cells were readily lysed in the two-
phase mixture of ADN and aqueous bu€er, with con-
comitant loss of nitrile hydratase activity. Addition of
ADN over the course of a reaction to produce 1.5 M
5CVAM resulted in only a modest improvement in cell
stability relative to complete addition of ADN at the
start of a reaction.
The nitrile hydratase activity of B23 cell paste which
was isolated by centrifugation and subsequently frozen
ꢀ
Similar stabilities were observed for frozen 38 and
ꢀ
at � 20 C or � 80 C was stable for at least 12 months.
5
4 wt% cell suspensions (wcw) in 0.5% sodium propio-
nate (pH 6.2). Cell breakage or lysis after one freeze±
thaw cycle was minimal, with only 1±5% of the total
nitrile hydratase activity present measured in assays as
cell-free enzyme activity. A second and third freeze±
thaw of frozen B23 cells resulted in extracellular nitrile
hydratase activity of 9 and 13% of total activity,
respectively, indicating an increase in permeabilization
of the B23 cell wall and membrane with increasing
number of freeze±thaw cycles. B23 cells subjected to one
freeze±thaw cycle and subsequently used in the prepara-
tion of B23/alginate beads (see below) produced a cata-
lyst with superior selectivity to 5CVAM in batch recycle
reactions than B23/alginate beads made with cells which
had not undergone at least one freeze±thaw cycle.
B23 cells must be rapidly separated from the product
mixture after complete conversion of ADN to prevent
over-conversion of 5CVAM to ADAM, and ®ltration
or centrifugation for catalyst separation and recovery
could not be performed on a commercial scale in a suf-
®ciently short time. Several methods of cell immobiliza-
tion were examined to produce a catalyst which could
be readily separated from the product mixture, and
which would also improve the stability of the whole cells
against lysis in a two-phase mixture of ADN and aqu-
eous bu€er. An immobilization method was required
that was simple to perform on a large scale, and pro-
duced an immobilized catalyst that had a high speci®c
activity (IU/g catalyst), and was stable to the reaction
conditions employed. Immobilization by entrapment in
agarose26 or carrageenan could not be utilized, since
Optimization of reaction conditions using unimmobilized
and immobilized B23 cells
ꢀ
ADN hydration reactions were run at 5 C with unim-
mobilized B23 cells to determine the stability of the
microbial nitrile hydratase activity. The concentration of
5
CVAM produced in these reactions was varied from 0.40
ꢀ
to 1.50M. The measured solubility of ADN at 5 C in
either the initial 20mM sodium butyrate/5 mM calcium
chloride bu€er (pH 7) or in 0.38M 5CVAM product
27

E. C. Hann et al. / Bioorg. Med. Chem. 7 (1999) 2239±2245
2241
the nitrile hydratase activity of B23 cells was rapidly
deactivated at the temperatures required for immobili-
zation in these matrices (35±50 C). Immobilization on
between pH 6.5 and 7.5, reactions were typically run in
a pH range of 6.8±7.2. The dependence of B23/alginate
bead speci®c activity on temperature was also measured
(Fig. 2); an increase in reaction temperature from 5 to
ꢀ
the surface of porous inorganic supports (such as Celite
or porous silica) would produce a catalyst with a low
speci®c activity relative to gel entrapment, and in stirred
batch reactions, rapid loss of activity due to attrition of
the particle surface during mechanical mixing of the
ꢀ
10 C resulted in an increase in catalyst speci®c activity
(and for a given catalyst loading, reaction rate) of ca.
60%. Speci®c activity also increased with decreasing
diameter of the B23/alginate beads (measured over a
range of 3.5 mm±2.0 mm diameter), indicating that the
reaction rate was limited at least in part by the rate of
di€usion of ADN into the alginate bead.
28
suspended catalyst could occur. Immobilization in
either polyacrylamide gel (PAG)29 or calcium alginate
produced a catalyst with high speci®c activity, but under
identical reaction conditions the B23/alginate catalyst
was more stable, and produced more product per weight
of catalyst, than B23/PAG catalyst. B23 cells were
immobilized in calcium alginate by ®rst preparing a
suspension of 20% wcw (5% dry cell weight (dcw)) cells
30
B23/alginate beads (ca. 2.5 mm dia.) were used as cata-
lyst for the production of either 0.75 or 1.5 M 5CVAM
in up to 60 consecutive batch reactions with catalyst
ꢀ
recycle at 5 C and in 23 mM butyrate/5 mM calcium
in an aqueous mixture containing sodium alginate
1
chloride (pH 7). Using a 7.5 wt % catalyst charge of
B23/alginate beads to produce 0.75 M 5CVAM, the
ratio of 5CVAM/ADAM averaged ca. 28:1 at complete
conversion of ADN. When a similar set of recycles was
(Pronova Protanal LF 10/60, 2.75% w/w ®nal con-
centration) and sodium butyrate (23 mM ®nal con-
ꢀ
centration, pH 7) at 5 C, then the resulting cell
ꢀ
suspension was ®ltered through consecutive 230 and 140
micron steel mesh ®lters prior to adding the suspension
dropwise with stirring to a solution of 0.20 M calcium
run at 10 C, the initial reaction time decreased from 3 to
2 h, but the ratio of 5CVAM/ADAM decreased to ca.
ꢀ
14:1. Repeating the hydration reactions at 5 and 10 C
ꢀ
chloride/23 mM sodium butyrate (pH 7) at 5 C. After
with unimmobilized cells showed no di€erence in selec-
tivity to 5CVAM at the two temperatures (5CVAM/
ADAM of >100:1). The decrease in 5CVAM selectivity
with increasing temperature when using B23/alginate
beads was most likely due to a larger increase in the
cellular nitrile hydratase activity relative to the increase
in the rate of di€usion of 5CVAM out of the alginate
beads, which resulted in the production of more
ADAM. Recycle reactions to produce 0.75 M 5CVAM
the resulting beads were allowed to harden in this solu-
tion for 18 h, the beads were washed twice with 23 mM
sodium butyrate/5 mM calcium chloride bu€er and
ꢀ
stored at 5 C in this same bu€er for up to 135 days with
no signi®cant loss of activity. The inclusion of 5 mM
calcium ion in both storage and reaction mixture bu€er
maintained the calcium-crosslinked form of the alginate
gel.
ꢀ
were also run at 5 C in the absence of 23 mM sodium
The dependence of speci®c activity (ADN IU/gram cat-
ꢀ
butyrate (substituting 20 mM Tris bu€er), and the
activity of the beads was observed to decrease over time
much more rapidly than for recycle reactions run within
the presence of butyrate, demonstrating the stabilizing
e€ect of butyrate on enzyme activity.
alyst) of the B23/alginate beads on pH at 5 C was
determined (Fig. 1). The speci®c activity decreased mark-
edly below pH 7, and was fairly constant from pH 7.5±9.
As the optimum pH for B23 nitrile hydratase stability was
Figure 1. Activity of B23/alginate beads as a function of pH (catalyst
activity at pH 9=100%). Reactions were run with pH monitoring to
produce 0.75 M 5CVAM in two-phase reaction mixtures containing
Fig. 2. Dependence of B23/alginate bead speci®c activity (ADN IU/g)
on temperature. Reactions were run to produce 0.75 M 5CVAM in two-
phase reaction mixtures containing ADN and 7.5% (w/w) B23/alginate
beads in 23 mM sodium butyrate/5 mM calcium chloride (pH 7).
ADN and 7.5% (w/w) B23/alginate beads in 23 mM sodium butyrate/
ꢀ
5
mM calcium chloride (pH 6±9) at 5 C.

2
242
E. C. Hann et al. / Bioorg. Med. Chem. 7 (1999) 2239±2245
Fifty consecutive batch reactions to produce 1.5 M
ꢀ
mixture was supersaturated with ADAM. When reac-
tions were run to produce ꢁ 1.5 M 5CVAM, no pre-
cipitation of ADAM was observed in these reaction
mixtures if the catalyst was immediately separated from
the product mixture at the end of the reaction. Some
B23/alginate beads recovered after use in 60 consecutive
batch reactions to produce ꢁ 1.5 M 5CVAM had
ADAM precipitated within the beads, but a signi®cant
decrease in catalyst activity for these catalyst beads was
not observed.
5CVAM at 5 C were run using 7.5 wt% B23/alginate
catalyst beads; after 50 reactions, the catalyst pro-
ductivity was 1910 g 5CVAM/g dcw B23 cells. After the
®
rst reaction, the ®nal concentration of 5CVAM in
subsequent batch reactions exceeded 1.7 M, as con-
secutive reactions were run by decanting the product
mixture from the previous reaction, and adding fresh
bu€er and ADN to the catalyst and remaining product
mixture from the previous batch reaction (ca. 16 wt% of
the initial reaction mixture). The ratio of 5CVAM to
ADAM with alginate beads made using frozen/thawed
B23 cells decreased from 36:1 to ca. 24:1 after the ®rst
several catalyst recycles (Fig. 3), while alginate beads
prepared using B23 cells that had not undergone a
freeze/thaw cycle produced a ®nal ratio of 5CVAM to
ADAM of ca. 15:1 under these same conditions. This
di€erence in selectivity was most likely due to an
increase in permeability of the cells to 5CVAM, caused
by the freezing and thawing of the cells prior to immo-
As an alternative to running consecutive batch reac-
tions, the use of a ®xed-bed packed column containing
B23/alginate beads was also examined. Packed-column
reactors containing B23/alginate beads were run at 5
ꢀ
and 10 C while feeding 0.40 M ADN (ca. the solubility
ꢀ
limit of ADN in reaction bu€er). At 5 C, the average
yields of 5CVAM and ADAM produced over the ®rst
800 h of continuous operation were 85 and 10%,
respectively, with a 95% conversion of ADN (5CVAM:
ADAM=8.5:1); after 800 h, the catalyst began to slowly
3
1,32
bilization.
selectivity was an additional bene®t of storing B23 cell
This signi®cant improvement in 5CVAM
ꢀ
lose activity. At 10 C, the column began to lose activity
ꢀ
suspensions frozen at � 80 C prior to immobilization in
after only 430 h, and produced up to 14% ADAM.
Because of the lower selectivity to 5CVAM in the ®xed-
bed column reaction, and the possibility of ADAM
precipitation and plugging of the column, the batch
reaction was chosen for scale-up.
alginate beads.
A time course for the production of 1.5 M 5CVAM
using B23/alginate beads at 5 C is illustrated in Figure
ꢀ
; at the reaction endpoint, the yields of 5CVAM and
4
ADAM were 95 and 3.5% respectively, with 1.2%
ADN remaining. This reaction was repeated, but the
reaction was allowed to continue past the point of
complete conversion of ADN until ADAM precipitated
from the product mixture; under these reaction condi-
tions, complete conversion of ADN occurred within 4 h,
and ADAM was ®rst observed to precipitate in the
reaction mixture between 7 and 8 h. The concentration
of ADAM reached ca. 110±120 mM before precipita-
tion, which was more than twice the measured solubility
limit for ADAM, and may indicate that the product
Commercial-scale preparation of 5CVAM
Scale-up of production of B23/alginate catalyst beads
was accomplished using a 7-inch diameter stainless steel
``nozzle'' equipped with 288 20-gauge hollow stainless
steel tubes, 3.5 cm in length. The B23 cell/alginate sus-
ꢀ
pension was pumped at 3 to 7 C through consecutive
230 and 140 mm steel screens to remove particulates,
then through the nozzle, and the resulting droplets col-
lected and hardened in 0.20 M calcium chloride/23 mM
ꢀ
sodium butyrate bu€er for 18 h at 5 C. The catalyst
Figure 3. Ratio of 5CVAM/ADAM produced in consecutive batch
reactions with recycle of B23/alginate beads. Reactions were run to
produce 1.5 M 5CVAM in two-phase reaction mixtures containing
Figure 4. Time course for the production of 1.5 M 5CVAM in a two-
phase reaction mixture containing ADN and 7.5% (w/w) B23/alginate
beads in 23 mM sodium butyrate/5 mM calcium chloride (pH 7.0) at
ADN and 7.5% (w/w) B23/alginate beads in 23 mM sodium butyrate/
ꢀ
ꢀ
5
mM calcium chloride (pH 7.0) at 5 C.
5 C: ADN (*), 5CVAM (~), ADAM (&).

E. C. Hann et al. / Bioorg. Med. Chem. 7 (1999) 2239±2245
2243
sion, produced more product per weight of catalyst, and
generated signi®cantly smaller amounts of byproducts
and waste products than alternate chemical processes
which were examined. In particular, the ratio of catalyst
waste to 5CVAM produced in the process was very low;
only 0.006 kg of catalyst waste was produced per kg
5
CVAM, and the catalyst waste was 93% by weight
water. The reaction produced ca. 190 g 5CVAM per
liter of reaction volume, and puri®cation and recovery
of the product for use in the next process step was
readily performed.
Experimental
Materials and methods
All materials were obtained from commercial sources
and used as received unless otherwise noted. Water used
in catalyst preparation and reactions was distilled and
deionized. Sodium alginate (Protonal1 LF 10/60) was
obtained from Pronova Biopolymers. Pseudomonas
chlororaphis B23 cell suspensions obtained from the
Figure 5. Reaction time (&), average reaction temperature (~), and
weight % 5CVAM (*) for 58 consecutive batch reactions with catalyst
recycle to produce at total of 13.6 metric tons of 5CVAM. Reactions
were run in two-phase reaction mixtures containing ADN and 5.8%
(
®nal weight catalyst/weight reaction mixture) B23/alginate beads in
ꢀ
2
3 mM sodium butyrate/5 mM calcium chloride (pH 7) at 5 C.
Nitto Chemical Industry Co. were stored frozen at
ꢀ
�
80 C. Wet cell weights of whole cell catalysts
beads were washed twice with 23 mM butyrate/5 mM
calcium chloride bu€er and stored in this same bu€er at
employed in reactions or assays were obtained from cell
pellets prepared by centrifugation of fermentation broth
or cell suspensions in bu€er. Dry cell weights were
determined by lyophilization of wet cells, and the ratio
of dry cell weight to wet cell weight for all cells was
typically 0.25:1. Reaction temperatures were thermo-
stated using recirculating temperature baths. The per-
cent recovery of ADN and the percent yields of ADAM
and 5CVAM were based on the initial concentration of
ADN present in the reaction mixture, and was deter-
mined by HPLC using refractive index detection and a
Supelcosil LC-18 DB column (25 cmÂ4.6 mm dia.) with
20 mM Tris bu€er at pH 7.5 in 2.5% methanol/water as
mobile phase.
ꢀ
5
C until needed.
The ®rst commercial-scale production of 5CVAM using
B23/alginate beads was performed by converting a total
of 12.7 metric tons of ADN in 58 consecutive 400-gallon
batch reactions with catalyst recycle. The reaction time,
average reaction temperature, and wt%5CVAM pro-
duced in the reactions are illustrated in Figure 5. At
97% overall conversion of ADN, the combined yield of
recovered 5CVAM was 13.6 metric tons (93% yield).
The catalyst productivity was 3150 kg of 5CVAM pro-
duced per kg of B23 cells (dcw). The total weight of
product mixture produced was 70.7 metric tons, which
contained 19.2 wt% 5CVAM, 0.99 wt% ADAM, and
Nitrile hydratase assay of B23 cells. Into a 20 mL glass
vial equipped with magnetic stir bar was added 10 mL
of an aqueous solution containing 0.20 M adiponitrile,
23 mM sodium butyrate, and 50 mM potassium phos-
phate (pH 7) at 25 C. With stirring, 0.100 mL of a B23
cell suspension (10.0 mg wet cell wt/mL) in an aqueous
bu€er containing 23 mM sodium butyrate and 50 mM
0.52 wt% ADN. The product was isolated from the
reaction mixture by removal of water by distillation,
ꢀ
then dissolution of the resulting oil at >65 C in
ꢀ
methanol, which precipitated the byproduct ADAM as
well as calcium and butyrate salts. The methanolic
5CVAM solution was used directly in the next process
ꢀ
step for herbicide synthesis. The reaction time for reac-
tion number 56 (Fig. 5) was allowed to extend for 4 h
past the endpoint of the reaction, then two more reac-
tions were run to determine if the increased production
and precipitation of ADAM in reaction number 56
produced a loss of catalyst activity in subsequent reac-
tions; no increase in reaction time for the ®nal two
batch reactions was observed.
potassium phosphate (pH 7) at 5 C was added to the
vial. After 15 min, a 0.200 mL aliquot of the reaction
mixture was added to 20 ml of 1 N HCl (to stop the
reaction) and 0.200 mL of 20 mM N-methylpropiona-
mide in water (HPLC standard) in a 1.5 mL micro-
centrifuge tube. The contents of the tube were mixed and
then centrifuged to pellet the cells, then the supernatant
was analyzed for 5CVAM by HPLC. The speci®c activity
(
IU/g) of the B23 cells was then calculated from the
reaction rate in micromoles 5CVAM per minute and the
weight of B23 cells employed in the assay.
Conclusions
A biocatalytic process for the highly regioselective
hydration of ADN to 5CVAM has been demonstrated
using P. chlororaphis B23 cells immobilized in calcium
alginate beads. The reaction was run to higher conver-
Preparation of P. chlororaphis B23/calcium alginate
beads. To a 250 mL media bottle was added 90.2 mL of
38 mM sodium butyrate (pH 7), and the bottle placed in
ꢀ
water bath at 25 C. Using an overhead stirrer, 4.125 g

2
244
E. C. Hann et al. / Bioorg. Med. Chem. 7 (1999) 2239±2245
of sodium alginate was slowly added to the bottle with
stirring, and the resulting mixture heated brie¯y to 80 C
added and the reaction repeated. After the desired
number of recycle reactions were completed, the ®nal
product mixture was separated from the catalyst beads
by ®ltration, and the catalyst beads washed twice by
ꢀ
with stirring to dissolve the alginate. The resulting
ꢀ
cell suspension (53.8% wet cell wt) was slowly added
solution was cooled to 5 C, then 55.7 g of frozen B23
ꢀ
stirring with reaction bu€er for 15 min at 5 C. The
ꢀ
cells were fully dispersed and suspended in the mixture
ꢀ
with stirring to the alginate solution at 5 C. After the
catalyst was then stored in reaction bu€er at 5 C until
the next set of recycle reactions were started.
ꢀ
vacuum for ca. 1 h at 5 C. The mixture was then
at 5 C, the cell suspension was degassed under partial
ꢀ
pumped at 2 mL/min through consecutive 230 and
Scale-up of B23/alginate bead preparation and 5-
cyanovaleramide production. Into a 100-gallon reactor
was charged 140 L of 23 mM sodium butyrate bu€er
(pH 6.8±7.2), then 6.42 kg of sodium alginate was added
140 mm steel mesh ®lters to remove particulates, then
through a 20 gauge needle (5 mm length, blunt tip) while
ꢀ
ꢀ
maintaining the temperature of the mixture at 3±7 C.
and the resulting mixture heated to 80 C for 1 h with
ꢀ
87.2 kg of frozen P. chlororaphis B23 cell slurry (54%
The resulting droplets of B23 cell/alginate suspension
produced by the needle were collected in a 1 L beaker
containing 488 mL of a stirred aqueous hardening solu-
tion containing 0.20 M calcium chloride and 23 mM
stirring. The resulting solution was cooled to 10 C, then
wcw/slurry weight) was added with stirring while main-
ꢀ
taining the temperature between 3 and 7 C. After the
ꢀ
sodium butyrate (pH 7) at 5 C. A distance of at least
20 cm between the needle tip and the hardening solution
®nal addition of frozen cell slurry, the suspension was
stirred until all the cells were suspended, then the cell
suspension was degassed under partial vacuum for an
additional 2 h. The resulting cell slurry was maintained
was required for the formation of spherical gel beads.
After completion of the addition, the resulting B23/
alginate beads were stirred in the hardening solution at
ꢀ
at 3±7 C while pumping through a 230 and 140 mm ®l-
ꢀ
5 C for 2 h, then the hardening solution was replaced
with an additional 488 mL of hardening solution at 5 C
ter, then through a 7-inch diameter stainless steel nozzle
equipped with 288 20-gauge hollow stainless steel tubes
(3.5 cm in length) attached to the top of a 300-gallon
reactor containing 757 L of 0.20 M calcium chloride/23
ꢀ
and the bead suspension mixed an additional 16 h at
ꢀ
5 min with 5 mM calcium chloride, 23 mM sodium
5
1
C. The catalyst beads were washed three times for
ꢀ
mM sodium butyrate hardening bu€er at 5 C. After the
ꢀ
butyrate at 5 C, and stored in this same bu€er at
dropwise addition of the cell slurry was complete, the
resulting B23 alginate beads were stirred for an addi-
tional 2 h, then the stirring was stopped and the hard-
ening bu€er decanted. An equivalent volume of fresh
ꢀ
5
3
C until ready for use (®nal bead diameter: 2.5±
.0 mm).
ꢀ
Assay of B23/alginate beads. Into a 50 mL Erlenmeyer
ask equipped with magnetic stir bar was weighed 1.88 g
hardening bu€er at 5 C was added to the catalyst beads
¯
and the mixture stirred an additional 16 h. The hard-
ening bu€er was again decanted, and the catalyst beads
washed twice with ca. 400 L of 23 mM sodium butyrate/
of B23/alginate beads and 20.98 mL of aqueous bu€er
containing 23 mM sodium butyrate and 5 mM calcium
ꢀ
.15 mL (2.048 g, 18.9 mmol) of adiponitrile, and the
ꢀ
chloride (pH 7) at 5 C. To the ¯ask was then added
2
5 mM calcium chloride bu€er (pH 7.0) at 5 C to yield
201 kg of B23/alginate beads. The beads were stored in
ꢀ
ꢀ
resulting mixture stirred at 5 C in a thermostated tem-
perature bath. Aliquots (0.100 mL) of the reaction were
removed at 1, 5, 10 and 15 min and immediately mixed
with 5 mL of 2 N HCl (to stop the reaction) and
the wash bu€er at 5 C until needed.
Into a 500-gallon reactor was charged 1007 L of 23 mM
sodium butyrate/5 mM calcium chloride bu€er (pH 7) at
ꢀ
218 kg (2015 mol) of adiponitrile. The resulting mixture
was stirred at 5 C under nitrogen, and samples were
0.100 mL of 0.20 M N-methylpropionamide (HPLC
5 C and 55.4 kg of B23/alginate beads, followed by
standard), and the resulting solution analyzed by HPLC
for 5-cyanovaleramide. The speci®c activity (IU/g) of
the B23/alginate beads was then calculated from the
reaction rate (micromoles 5CVAM per min) and the
weight of catalyst beads employed in the assay.
ꢀ
withdrawn at regular intervals and analyzed by HPLC
to determine the progress of the reaction. When the
conversion of adiponitrile was ꢁ 97% (ca. 4 h initial
reaction time), the stirring was stopped and ca. 90% of
the product mixture was decanted. The reactor was
immediately charged with 1007 L of reaction bu€er and
218 kg (2015 mol) of adiponitrile, and the reaction repe-
ated. After running 13 consecutive batch reactions to
establish a reaction baseline, a ®nal additional 20.4 kg of
B23/alginate beads was added to the fourteenth con-
secutive batch reaction (5.8 wt % ®nal catalyst weight/
weight reaction mixture). Each decanted product mix-
Recycle reactions using B23/alginate beads. Into a
125 mL Erlenmeyer ¯ask equipped with magnetic stir
bar was placed 7.5 g of B23/alginate beads, 75.3 mL of
aqueous bu€er containing 23 mM sodium butyrate and
5
ꢀ
16.49 g, 0.152 mol) of adiponitrile. The resulting mix-
ture was stirred at 5 C in a thermostated temperature
bath. The progress of the reaction was monitored by
removal of 0.100 mL aliquots of reaction mixture, which
were mixed with 5 mL of 1.0 N HCl (to stop the rxn) and
mM calcium chloride (pH 7) at 5 C, and 17.2 mL
(
ꢀ
ꢀ
ture was brie¯y heated to 40 C to deactivate any cata-
lyst activity that was present. Fifty-eight consecutive
batch reactions produced 13.6 metric tons of 5CVAM,
and the yields of 5-cyanovaleramide and adipamide
from the combined product mixtures were 93 and
4%, respectively, with 3% recovery of unconverted
adiponitrile.
0.100 mL of 0.20 M N-methylpropionamide (HPLC
standard), and then analyzed by HPLC. When the con-
version of adiponitrile was at least 97%, 83.25 mL of the
product mixture was decanted and 66.05 mL of reaction
ꢀ
bu€er at 5 C and 17.2 mL (16.49 g) of adiponitrile was

E. C. Hann et al. / Bioorg. Med. Chem. 7 (1999) 2239±2245
2245
Isolation and puri®cation of 5-cyanovaleramide. A 2 L
sample of product mixture from a commercial-scale
reaction (190 g 5-cyanovaleramide/L) was continuously
extracted with 1.5 L of ethyl acetate. The ethyl acetate
fraction was then concentrated by rotary evaporation to
produce ca. 696 g of a wet white solid, which was not
dried further to remove remaining ethyl acetate, but was
divided into two equal portions by weight. Each portion
3. Schaefer, F.C. In The Chemistry of the Cyano Group;
Rappoport, Z., Ed.; Interscience: New York, NY 1970;
Chapter 6, pp 256±262.
4. Sugiyama, K.; Miura, H.; Nakano, Y.; Suzuki, H.;
Matsuda, T. Bull. Chem. Soc. Jpn. 1987, 69, 453±456.
5
1
6
7
. Onuoha, N. I.; Wainwright, M. S. Chem. Eng. Commun.
984, 29, 1±12.
. Holland, H. L. Curr. Opin. Chem. Biol. 1998, 2, 77±84.
. Sugai, T.; Yamazaki, T.; Yokoyama, M.; Ohta, H. Biosci.
was dissolved in a solution of 200 mL of methanol and
ꢀ
Biotech. Biochem 1997, 61, 1419±1427.
8. Meth-Cohn, O.; Wang, M.-X. J. Chem. Soc., Perkin Trans.
1 1997, 3197±3204.
8
00 mL of toluene and heated to 60 C in a water bath
with stirring. Undissolved solids were removed from the
resulting warm solution by ®ltration, and the ®ltrate
9. Ho
Commun. 1990 648±650
0. Blakey, A. J.; Colby, J.; Williams, E.; O'Reilly, C. FEMS
È
nicke-Schmidt, P; Schneider, M.P. J. Chem. Soc., Chem.
ꢀ
was then cooled to 5 C. Addition of 1.4 L of ethyl ether
1
to the cold ®ltrate with stirring immediately precipitated
a white solid, which was collected by vacuum ®ltration,
Microbiology Lett. 1995, 129, 57±62.
11. Crosby, J.; Moilliet, J.; Parratt, J. S.; Turner, N. J. J.
Chem. Soc., Perkin Trans. 1 1994, 1679±1687.
ꢀ
and washed with two 200 mL portions of cold (5 C) ethyl
ether. The combined ether precipitates were combined
and dried to constant weight under vacuum to yield 376 g
1
2. De Raadt, A.; Klempier, N.; Faber, K. T.; Griengl, H. J.
Chem. Soc., Perkin Trans. 1 1992, 137±140.
3. Yokoyama, M.; Sugai, T.; Ohta, H. Tetrahedron: Asym-
(
99% yield) of 99% pure 5CVAM (0.7% ADAM and
1
0
.3% ADN by HPLC). A portion of this 5CVAM was
metry 1993, 4, 1081±1084.
14. Cohen, M. A.; Sawden, J.; Turner, N. J. Tetrahedron Lett.
1990, 31, 7223±7226.
ꢀ
further puri®ed by continuous extraction at 100 C with
toluene, followed by separation of the toluene fraction
and concentration by rotary evaporation. The resulting
white solid was recrystallized from ethyl acetate at
15. Ingvorsen, K.; Yde, B.; Godtfredsen, S. E.; Tsuchiya, R. T.
Ciba Found. Symp. 1988, 140 (Cyanide Compd. Biol.), 16±31.
ꢀ
ꢀ
1
16. Andresen, O.; Godtfredsen, S.E. European Patent
�
20 C to yield puri®ed 5CVAM: mp 66.0±66.9 C; H
1
1
5
78,106B1, 1993.
7. DiCosimo, R.; Stieglitz, B.; Fallon, R. D. U.S. Patent
728556 A 980317.
NMR (300 MHz, CDCl ) d 6.24 (bs, 1H), 5.99 (bs, 1H),
3
2
.40 (t, J=6.7 Hz, 2H), 2.28 (t, J=6.9 Hz, 2H),1.85±1.67
1
3
(
m, 4H); C NMR (75 MHz, CDCl ) d 174.8, 119.6,
3
18. Saito, O.; Kawakami, K. Jpn. Kokai Tokkyo Koho JP
2154692 A2 900614.
3
1
4.6, 24.9, 24.4, 17.0; IR 3414, 3217, 2959, 2885, 2245,
�
0
1
+
677, 1621 cm ; MS (EI) m=z 127 (MH , 7), 110 (4), 86
19. Nagasawa, T.; Nanba, H.; Ryuno, K.; Takeuchi, K.;
Yamada, H. Eur. J. Biochem. 1987, 162, 691±698.
20. Yamada, H.; Ryuno, K.; Nagasawa, T.; Enomoto, K.;
Watanabe, I. Agric. Biol. Chem. 1986, 50, 2859±2865.
(9), 82 (11), 59 (100), 54 (31); HRMS calcd for
+
C H N O (MH ) 127.0871, found 127.0871.
6
10
2
2
Chem. 1982, 46, 1183±1189.
1. Asano, Y.; Yasuda, T.; Tani, Y.; Yamada, H. Agric. Biol.
Acknowledgements
2
2. Ashina, Y.; Suto, M. Bioprocess Technol. 1993, 16, 91±107.
3. Kobayashi, M.; Nagasawa, T.; Yamada, H. Trends Bio-
2
technol. 1992, 10, 402±408.
The authors gratefully acknowledge the contributions of
L. Winnie Wagner, Robert D. Fallon, Mark J. Nelson,
David L. Anton, Ron Grosz, Umesh Hattikadur, John
Freudenberger, Rafael Shapiro, Onorato Campopiano,
and the Agricultural Products Semiworks (DuPont),
Coreen R. B. Reed (Pronova Biopolymer, Inc.), and Tet-
suro Horinouchi and Katsumi Nakamura (Mitsubishi
Rayon Co.).
2
4
4. Nagasawa, T.; Ryuno, K.; Yamada, H. Experientia 1989,
5, 1066±1070.
25. Ryuno, K.; Nagasawa, T.; Yamada, H. Agric. Biol. Chem.
1988, 52, 1813±1816.
26. Nilsson, K.; Brodelius, P.; Mosbach, K. Methods Enzymol.
1
2
987, 135, 222±230.
7. Chibata, I.; Tosa, T.; Sato, T.; Takata, I. Methods Enzy-
mol. 1987, 135, 189±198.
8. Monsan, P.; Durand, G.; Navarro, J. M. Methods Enzy-
mol. 1987, 135, 307±318.
2
References
2
1
9. Skryabin, G. K.; Koshcheenko, K. A. Methods Enzymol.
987, 135, 198±216.
1
2
. Izumi, Y. Catal. Today 1997, 33, 371±409.
. Larock, R.C. Comprehensive Organic Transformations: A
30. Bucke, C. Methods Enzymol. 1987, 135, 175±189.
31. Felix, H. Anal. Biochem. 1982, 120, 211±234.
32. Oliveira, D. E.; Santos Neto, L. C.; Panek, A. D. Anal.
Biochem. 1981, 113, 188±192.
Guide to Functional Group Preparations; VCH: New York,
989; p 993.
1