Bioconversion of N-butylglucamine to 6-deoxy-6-butylamino sorbose by Gluconobacter oxydans

Article abstract of DOI:10.1021/op0255128

Gluconobacter oxydans has the unique ability to regioselectively and rapidly oxidize sorbitol and other erythro saccharides. In this report a new process is described by which N-butylglucamine is regioselectively oxidized by the organism. A largescale process is described by which N-butylglucamine can be converted to an intermediate (6-deoxy-6-butylaminosorbose) which can be readily converted to N-butyldeoxynojirimycin by catalytic hydrogenation. The primary process variables of temperature, pH, and added acids and salts were investigated in laboratory bioreactors. Since degradation of the sorbose product was rapid above room temperature, significant enhancement of the selectivity was achieved by lowering the temperature at which the bioconversion was run. The optimum temperature for this conversion was 12-15°C. The pH maximum of the bioconversion was 5.5-6.0. However, the small gain in rate relative to pH 5.0 was at least offset by the increase in degradation of the product at the higher pH. Nitrate salts of N-butylglucamine could replace chloride salts, but sulfate, acetate, and phosphate salts could not. Sulfate in particular led to inhibition of the conversion, while phosphate and acetate led to increased degradation. At temperatures in the range of 12-15°C, pH of around 5.0 and substrate concentrations of 0.2 M, Gluconobacter oxydans catalyzed bioconversion to 6-deoxy-6-butylaminosorbose with yields approaching 95%. These conditions were used to scale this process to 5500-L scale.

Full text of DOI:10.1021/op0255128

Organic Process Research & Development 2002, 6, 547−552
Bioconversion of N-Butylglucamine to 6-Deoxy-6-butylamino Sorbose by
Gluconobacter oxydans
Bryan H. Landis,* Joseph K. McLaughlin, Robert Heeren, Roy W. Grabner, and Ping T. Wang
Bioprocess and Formulation Technology, Pharmacia, 700 Chesterfield Parkway North, St. Louis, Missouri 63198, U.S.A.
Abstract:
N-butylglucamine to an intermediate suitable for reduction
to N-butyldeoxynojirimycin was demonstrated on a labora-
tory scale and selected as the best available technology for
scale-up. A broad range of operating conditions was evalu-
ated to demonstrate the technical feasibility and performance.
This report describes in detail five primary bioconversion
parameters: temperature, pH, N-butylglucamine (NBG)
concentration, addends, and type of acid (counterion).
Stabilization of the oxidized product by the butyl group was
Gluconobacter oxydans has the unique ability to regioselectively
and rapidly oxidize sorbitol and other erythro saccharides. In
this report a new process is described by which N-butylglu-
camine is regioselectively oxidized by the organism. A large-
scale process is described by which N-butylglucamine can be
converted to an intermediate (6-deoxy-6-butylaminosorbose)
which can be readily converted to N-butyldeoxynojirimycin
by catalytic hydrogenation. The primary process variables of
temperature, pH, and added acids and salts were investigated
in laboratory bioreactors. Since degradation of the sorbose
product was rapid above room temperature, significant en-
hancement of the selectivity was achieved by lowering the
temperature at which the bioconversion was run. The optimum
temperature for this conversion was 12-15 °C. The pH
maximum of the bioconversion was 5.5-6.0. However, the small
gain in rate relative to pH 5.0 was at least offset by the increase
in degradation of the product at the higher pH. Nitrate salts of
N-butylglucamine could replace chloride salts, but sulfate,
acetate, and phosphate salts could not. Sulfate in particular led
to inhibition of the conversion, while phosphate and acetate led
to increased degradation. At temperatures in the range of 12-
a
found to arise from increasing the pK of the amino nitrogen
(NBG vs glucamine) and providing steric crowding, lessening
the ability of nucleophilic nitrogen to interact with the
carbonyl carbon of the oxidized product.
The microorganism used for the bioconversion was
Gluconobacter oxydans, a member of the family Aceto-
3
bacteraceae. This microorganism is used industrially to
prepare sorbose from sorbitol (glucitol) and dihydroxyacetone
from glycerol. G. oxydans has also been used to prepare
4
deoxynojirimycin (DNJ) by first inactivating the basic amino
nitrogen, performing the bioconversion with G. oxydans, then
removing the inactivating group and reducing the unalkylated
aminosorbose. All of these bioconversions are catalyzed
by membrane-bound polyol dehydrogenases (a portion of
these dehydrogenases may be soluble), one of which is
sorbitol dehydrogenase. Sorbitol dehydrogenase is believed
to be the enzyme responsible for catalyzing the bioconversion
of N-butylglucamine to 6-deoxy-6-butylaminosorbose. Ad-
ditional details and extensions may be found elsewhere.5
1
5 °C, pH of around 5.0 and substrate concentrations of 0.2
M, Gluconobacter oxydans catalyzed bioconversion to 6-deoxy-
-butylaminosorbose with yields approaching 95%. These
6
conditions were used to scale this process to 5500-L scale.
Introduction
Experimental Section
N-Butyldeoxynojirimycin (N-butyl DNJ) is an inhibitor
of glycosidase activity, and as such was clinically evaluated
N-Butylglucamine. N-Butylglucamine (NBG) was ob-
tained from Pharmacia, Skokie, IL, U.S.A. Material was
prepared by reductive amination of glucose with n-butyl-
amine. High-purity NBG was isolated from these reactions
by crystallization. Bioconversion solutions were prepared by
suspending NBG in water and adjusting the pH with an acid
to below 6, simultaneously dissolving the NBG as the salt.
If magnesium sulfate or other salts were to be added, it was
added at this point, the pH was then adjusted to the target
pH (usually 4.9 to 5.1), and the volume was adjusted to
achieve the target concentration of NBG. For the pH-rate
studies, the pH of a stock solution was adjusted to various
1
by Pharmacia R&D as a potential therapeutic agent against
retroviral infections, in particular acquired immune deficiency
syndrome (AIDS). The activity of this compound was
determined in vitro to prevent infection of H9 cells by the
2
AIDS virus, and these results were promising enough to
encourage Pharmacia to pursue the development of processes
to make this compound in quantity for clinical trials and
commercial production. At least four processes were given
extensive consideration, three of which combined bioprocess
technology with chemical synthesis. A novel, high-efficiency
process based on the selective microbial oxidation of
(3) Krieg, N. R., Ed. Bergey’s Manual of Systematic Bacteriology; Williams
*
To whom correspondence should be addressed. Telephone: 01 636-737-
& Wilkins: Baltimore/London, 1984; Vol. 1, p 275.
6
760. Fax: 01 636 737 7281. E-mail: bryan.h.landis@pharmacia.com.
(4) Kinast, G.; Schedel, M. U.S. Patent 4,246,345, 1981; Kinast, G.; Schedel,
M. U.S. Patent 4,266,025, 1981; Schroeder, T.; Stube, M. U.S. Patent
4,806,650, 1989; Kinast, G.; Schedel, M.; Koebernick, W. U.S. Patent
4,405,714, 1983.
(5) Grabner, R. W.; Landis, B. H.; Wang, P. T.; Scaros, M.; Prunier, M.
U.S. Patent 5,916,784, 1999.
(
1) Schweden, J.; Borgmann, C.; Legler, G.; Bause, E. Arch. Biochem. Biophys.
986, 248, 335-340.
1
(
2) Karpas, A.; Fleet, G. W. J.; Dwek, R. A.; Petursson, S.; Namgoong, S. K.;
Ramsden, N. G.; Jacob, G. S.; Rademacher, T. W. Proc. Natl. Acad. Sci.
U.S.A. 1988, 85, 9229-9233.
1
0.1021/op0255128 CCC: $22.00 © 2002 American Chemical Society
Vol. 6, No. 4, 2002 / Organic Process Research & Development
•
547
Published on Web 05/16/2002

target pH values, and 50 mL aliquots were removed for the
individual tests.
sodium borohydride were used, little of the N-butyldeoxy-
nojirimycin was observed. The procedure used 50 µL of
Gluconobacter oxydans Cells. Gluconobacter oxydans
4
sample, 50 µL of 2% NH OH (v/v from concentrated) to
(DSM 2003 or ATCC 621) cells were prepared using either
make the pH approximately 9, and 200 µL of aqueous NaBH
4
batch fermentation or chemostat fermentation. In either case,
the protocols were defined to ensure full differentiation and
activity of the Gluconobacter cells. The cells were cultivated
in a liquid medium containing 24 g/L yeast extract (Red Star
Specialty Products, Milwaukee, WI), 60-120 g/L sorbitol
(5 mg/mL). After reduction for 1 h at room temperature,
excess borohydride was destroyed with glacial acetic acid.
After evaporation of the water, acetylation was accomplished
with acetic anhydride and either pyridine or triethylamine.
While the GC method gave the best resolution of products
from starting materials, it suffered from variable efficiency
of the chemical reduction step and could not detect the DBS
directly. The IEX HPLC gave the least resolution. Overall
the ion-pairing chromatography with PAD gave the best
results.
6
(
Roquette, Gurnee, IL), 3 g/L succinic acid (Mallinckrodt
Inc., Chesterfield, MO), and 0.5 g/L Ucon LB625 antifoam
Union Carbide Chemicals & Plastics, S. Charleston, WV).
(
Initial pH was adjusted to 6.3 with sodium hydroxide. The
three-stage fermentation batch process for production of G.
oxydans was initiated by inoculating 1 mL stock culture into
Preparation of N-Butyldeoxynojirimycin. In the fol-
lowing example, the bioconversion product (6-deoxy-6-
butylaminosorbose) was catalytically hydrogenated to N-butyl-
DNJ. In a manner similar to the above, NBG solutions were
prepared and bioconverted, except that the concentration of
NBG was varied from 20 to 80 g/L. After 48 h, the percent
conversions were all approximately 90%. The cells from each
were removed by centrifugation, and aliquots of the super-
natants were combined and filtered to remove all debris. This
clear, light-yellow solution was decolorized with activated
charcoal and placed in a Parr hydrogenator with 4%
palladium on charcoal catalyst and 50 psig hydrogen at
ambient temperature. After 2 h, 20 min the reduction was
complete (approximately stoichiometric consumption of
hydrogen), and the catalyst was filtered off. The filtrate was
applied to a column of cation-exchange resin (Dowex 50
mm × 8 mm, 200 mesh), washed with water and absolute
methanol; the product was then eluted in batch mode with
methanol:ammonium hydroxide (3:1). The eluate was evapo-
rated to a thick oil in vacuo and recrystallized from methanol/
acetone. Yield from starting NBG to N-butyl-DNJ was 53%.
GC of the crude oil indicated 88% purity for the N-butyl-
DNJ. After one crystallization the purity was 99.5% by GC.
0
.5 L growth medium in a 2.8 L Fernbach. The culture was
incubated in a shaker at 30 °C and 200 rpm for 24 h. The
primary seed was then transferred to a secondary fermenter
at either 10- or 800-L scale for 12 to 24 h. Production
fermentation was carried out in a 40000-L fermenter. The
final cell concentration at harvest was measured by optical
density at 550 nm and was always greater than 6. The cells
were isolated by centrifugation and washed with chilled 20
mM MgSO
as cell pastes or 40% suspensions in 20 mM MgSO
30 °C. For short periods of time (2 or 3 h), cells were also
stored refrigerated.
4
. These were used immediately or stored frozen
4
at
-
Shake Flask Experiments. The substrate (NBG) solu-
tions (if necessary) were incubated at the desired temperature
in baffled 500-mL shake flasks. G. oxydans cells were added,
either freshly prepared or thawed (or stored refrigerated), as
a cell paste or 40% cell suspension. The target was 40 g of
G. oxydans (wet) cell paste (approximately 8 g/L dry cell
weight) per liter of NBG solution. No aeration was supplied
other than that achieved by rotation at 100-120 rpm. The
working volumes for shake flasks were typically 50 mL. The
shake flask experiments were limited to 30 °C and ambient
room temperature, generally 22-23 °C. Because of these
relatively higher temperatures, degradation (vide infra) of
the 6-deoxy-6-butylaminosorbose (DBS) was substantial.
Spinner Flask Experiments. Spinner flasks were used
to perform bioconversions below room temperature. The
solutions were incubated at the desired temperature in a 500-
mL spinner flask. G. oxydans cells were added, either freshly
prepared or thawed (or stored refrigerated), as a cell paste
or 40% cell suspension. The target was 40 g of G. oxydans
Results and Discussion
The initial phase of this project was to evaluate experi-
mentally the concept of a simple, novel process to produce
N-butyl-DNJ. G. oxydans cells had been shown to catalyze
conversion of NBG to a material which could be converted
to N-butyl-DNJ by reduction. Although the yields were low
(<5%), these experiments indicated the feasibility of this
bioconversion.
Three aspects of the NBG bioconversion needed to be
optimized. The first was percent conversion. The percent
conversion was the percent of the starting NBG which
disappeared during the bioconversion reaction. The second
aspect was selectivity. Selectivity was the proportion of the
NBG which was converted and was recovered as DBS. Since
this impacted yield, maximizing selectivity was important.
The third was the rate of conversion. Early rates of
conversion were on the order of a few percent per day, and
the target was to complete the conversion (97%) in 24 h.
Conversion. The original conditions tried for the conver-
(
wet) cell paste (approximately 8 g/L dry cell weight) per
liter of NBG solution. The suspension was aerated at 1 vvm
vol/vol/min) by means of a gas dispersion stone and the
(
solution agitated. The working volumes were typically 200
mL. Antifoam was added to control foaming once cells were
added.
Analytical. Samples were clarified and analyzed by ion-
pairing chromatography, with pulsed amperometric (PAD)
detection, IEX HPLC, or GC after reduction with NaBH
and acetylation with acetic anhydride. This last method gave
substantial reduction of the DBS back to NBG. If excessive
4
6
sion of NBG were based on the literature. It was originally
(6) White, S. A.; Claus, G. W. J. Bacteriology 1982, 150, 934-943.
believed that oxalate or succinate was required for effective
5
48 Vol. 6, No. 4, 2002 / Organic Process Research & Development
•

Figure 2. Rate of bioconversion of NBG to DBS as a function
of NBG concentration.
Figure 1. Arrhenius plots for bioconversion of NBG to DBS
and degradation of DBS; (9) bioconversion, ([) degradation.
Table 2. Effect of counterion on selectivity of the
bioconversion of NBG to DBS at room temperature
Table 1. Effect of acid used to titrate NBG to pH 5 on the
conversion to DBS at 30 °C
counterion
% conversion
% selectivitya
acid
% conversion
chloride
nitrate
sulfate
phosphate
acetate
>99
>99
30
60
75
97
>99
>99
70
oxalic acid
30
60
<10
>90
succinic acid
ascorbic acid
hydrochloric acid
50
a
Sum of furanose and piperidinose forms.
conversion of NBG. When oxalic acid was used to titrate
the NBG free base to pH 5 for the bioconversion, the percent
conversions at 30 °C were low (Table 1). In addition, other
organic di- and triprotic acids gave poor percent conversions.
However, addition of magnesium salts along with any of
these organic acids improved the percent conversion. This
suggested that these organic acids might have been chelating
a metal ion essential for the bioconversion. When hydro-
chloric acid was substituted for the organic acids, in the
presence of 20 mM magnesium sulfate, the conversion
proceeded to at least 90%. This dramatic improvement
indicated complete conversion was possible, and after
removal of the magnesium salts as well and lowering the
temperature (see below), conversions approaching 100%
were attained.
Selectivity. The selectivities, however, were lower than
desired. It was observed during the bioconversions that the
absorbance at 280 nm increased dramatically. It was believed
that this UV absorbance increase was due to degradation (via
piperidinose intermediates) to a pyridine derivative. Since
the degradation reaction was not an enzyme-catalyzed
reaction, and was independent of cells, it was proposed that
its temperature-dependence would be greater than that of the
biochemical oxidation. In Figure 1 are superimposed Ar-
rhenius plots for the degradation and bioconversion reactions.
The temperature dependence of the degradation was steep,
with an energy of activation of about 20 kcal/mol. The
temperature dependence of the bioconversion was clearly
much shallower than that of the degradation. This indicated
that a significant lessening of the degradation could be
realized by lowering the temperature a few degrees without
greatly diminishing the rate of conversion. When shake-flask
experiments were performed at room temperature rather than
of conversion and selectivity was in the range of 12-15 °C.
However, shake-flask experiments were still performed at
room temperature.
The effect of other acids (counterions) was also examined
at room temperature (Table 2). In these tests it was observed
that titrating the NBG free base with nitric acid to pH 5 gave
essentially the same results in terms of conversion and
selectivity as using hydrochloric acid. However, titrating with
sulfuric acid, while giving good selectivity, seemed to inhibit
the bioconversion, the percent conversion proceeding to only
a few percent. Phosphoric acid and acetic acid gave
somewhat better conversions than sulfuric acid, but the
selectivities were much diminished. With these two counter-
ions the degradation increased, suggesting that they promoted
formation of the piperidinose structure of DBS (vide infra).
Rate of Conversion. With the percent conversion and
selectivity improved, it was possible to investigate the cause
of the slow conversion rate. Relative to the published rates
6
,7
of conversion for sorbitol and carbobenzyloxyglucamine,
the rate of conversion of NBG was much slower. In Figure
2
are shown initial rates of conversion for NBG as a function
of substrate concentration. In Figure 2 of ref 6, the profile
for sorbitol had the appearance of an enzyme-catalyzed
reaction, giving saturation at high substrate concentrations
and half-maximal rate at about 0.1 M. The concentration
dependence for NBG, on the other hand, was first order, with
no apparent saturation. For simple enzyme-catalyzed reac-
tions, the region where first-order kinetics is obeyed is where
the substrate concentration is much less than the enzyme-
substrate dissociation constant. This suggested that this
dissociation constant was considerably increased for NBG
relative to that for sorbitol, probably due to the positive
charge on the amino group of NBG at pH 5. To give rates
30 °C, the selectivity increased from less than 70% to about
8
5-90%. At lower temperatures, the selectivity increased a
(
7) Yang, F.-L.; Wu, S.-H. J. Chin. Chem. Soc. 1999, 46, 707-714, reported
little more. The temperature giving the best balance of rate
K
m
for sorbitol of 19 mM.
Vol. 6, No. 4, 2002 / Organic Process Research & Development
•
549

Effect of the Butyl Group on Stability. In Figure 5 are
compared bioconversions of glucamine and NBG. As the
bioconversion of NBG proceeded, the correctly oxidized
product was obtained in high yield. As the bioconversion of
glucamine proceeded, little, if any, of the desired oxidation
product was obtained. After lyophilization and reconstitution
of the oxidized products from each, the UV absorbance of
the glucamine bioconversion was more than twelve times
greater than that of NBG bioconversion. This suggested that
the butyl group was stabilizing the oxidized product relative
to the glucamine system.
The butyl group stabilized the DBS relative to amino-
sorbose in two ways. First, the butyl group made the
amino nitrogen a stronger base. Titration curves for NBG,
glucamine and DBS are shown in Figure 6. The butyl group
Figure 3. Effect of pH on the bioconversion of NBG to DBS;
4 h incubation with G. oxydans; (0) DBS concentration; (2)
selectivity; ([) NBG concentration.
2
%
sufficient to complete the bioconversion in 24 h, the levels
of G. oxydans added to the shake flasks and bioreactors had
to be severalfold higher than the levels achieved in a typical
G. oxydans fermentation. A second contribution to increasing
the rate was ensuring that fermentations to provide G.
oxydans cell mass reached late log-phase growth. It had been
a a
increased the pK of the amino group from 9.23 (K )
-10
-10
5
.95 ( 0.22 × 10 ) to 9.59 (K
a
) 2.59 ( 0.16 × 10 ).
a
Titration of isolated DBS indicated a pK of 8.16. While the
pK
it was not unreasonable to believe that a similar increase in
the pK due to the butyl group occurred for the oxidized
products as with the starting substrates, projecting a pK for
a
of the 6-deoxy-6-aminosorbose could not be determined,
a
6
noted by Claus and co-workers that the membrane-bound
a
sorbitol dehydrogenase increased at that point (due to
membrane invagination). The qualification that cells be
grown to late log phase was incorporated as part of the cell
mass fermentation protocols.
the 6-deoxy-6-aminosorbose of about 7.8. Since the stronger
base would be more highly protonated at the mildly acidic
pH of the bioconversion, there would be less nucleophilic
amine to compete with the hydroxyl for the carbonyl. This
would shift the equilibria shown in Figure 4 more in the
direction of the sorbofuranose for the DBS system than for
the aminosorbose system. The other contribution from the
butyl group was steric interference. The butyl group will
provide some steric crowding around the nitrogen, with the
result that attack at the carbonyl by the nitrogen will be
inhibited. This would again shift the equilibria in favor of
the sorbofuranose for the butylaminosorbose system relative
to the aminosorbose system. By shifting the equilibria toward
the stable sorbofuranose structure, less piperidinose would
be formed, with the result that the degradation rate would
be slowed. This in turn led to the bioconversion results shown
in Figure 5.
pH Dependence. The pH dependence of the bioconver-
sion is shown in Figure 3. The overall shape of the pH-rate
6,7
profile was similar to that for sorbitol, with a broad
maximum centering on pH 5.5-6.0. Although the curves
were similar, there was a fundamental difference between
sorbitol and NBG. That is, the oxidized product for sorbitol
(sorbose) was stable, whereas that for NBG had been shown
to decompose. As the pH was increased there was a decrease
in selectivity, most likely caused by increased degradation
as the proportion of piperidinose form of DBS increased (see
below). The selectivity decreased with increasing pH, the
average linear decrease in selectivity being 3-4% per pH
unit. Although there did appear to be some increase in rate
to be gained by increasing the pH of the bioconversion
somewhat above pH 5.0, this was offset by the increased
degradation. While lowering the pH below 5.0 might have
benefited the selectivity slightly, the rate decreased enough
to render this insubstantial. For these reasons the pH was
maintained at 5.0.
In the literature there have been numerous attempts to
8,9
quantify electronic and steric effects of substituents.
10
According to Fujita and co-workers, equilibrium constants
3
for reactions of amines (hydrogen bonding with CHCl ,
2
charge transfer with I plus ester reactions) might be
described by
Structure of DBS. Structures for the bioconversion are
shown in Figure 4. NBG was oxidized by G. oxydans to
c
s
log K ) F*∑
σ* + a E (R ) + c
∑
i
i
13
DBS (structure A). However, neither C NMR nor infrared
spectroscopy indicated the presence of significant amounts
of carbonyl (data not shown). Rather, the molecule cyclized
through the 5-hydroxyl to form the sorbofuranose, structure
B. At alkaline pH, chromatographic evidence suggested the
presence of a second structure, probably a piperidinose
structure such as structure C. Since the DBS could be reduced
where K was the equilibrium constant, σ* was an electronic
c
coefficient, E
s
i
was the steric coefficient and F* and a were
constants of the particular reaction. For unsubstituted amines
σ* was 0.490, for n-butylamine it was -0.130; for unsub-
c
stituted amines E
s
was 0.32, for n-butylamine it was -0.70.
(8) Gallo, R. Progress in Physical Organic Chemistry; John Wiley & Sons:
2
at pH 5 by H with Pd/C catalyst to N-butyl-DNJ, at least
New York, 1983; Vol. 14, pp 115-1163 and references therein.
9) Taft, R. W., Jr. In Steric Effects in Organic Chemistry; Newman, M. S.,
Ed.; Wiley: New York, 1956; pp 556 and ff.
some portion of the DBS had to exist in this form, the
reduction going to completion as the equilibria generated
additional piperidinose. It was this piperidinose, however,
which led to degradation, forming an aromatic product.
(
(
10) (a) Takayama, C.; Fujita, T.; Nakajima, M. J. Org. Chem. 1979, 44, 2871-
879. (b) Bogatkow, S. V. et al. Org. React. USSR 1969, 6, 436, cited by
Takayama in ref 10a.
2
550
•
Vol. 6, No. 4, 2002 / Organic Process Research & Development

Figure 4. Equilibria proposed for the bioconversion of NBG to DBS.
Figure 6. Titration of NBG, DBS and glucamine; ([) NBG
titration; (0) glucamine titration; (2) DBS concentration.
Figure 5. Appearance of sorbose products as a function of
time during bioconversion of the polyols at room temperature;
Inc., North Chicago, Illinois. The scaled process utilized a
40000-L fermenter to produce the G. oxydans cells in batch
fermentation. The cells were concentrated with BTUX 510
continuous discharge disc-stack centrifuge to 2000 L and
(9) DBS concentration; (0) NBG concentration; (]) glucamine
concentration; ([) aminosorbose concentration.
Relative to the unsubstituted glucamine system, these
coefficients predicted that the equilibrium constant for
n-butylglucamine system should be 4-fold lower due to
4
washed with 40 000 L of 20 mM MgSO . The washed cells
were reconcentrated and stored under refrigeration until used
for the bioconversion. The bioconversion (Figure 7) was
performed at approximately 5500-L scale at 220 g/L NBG,
14 g/L dry cell weight (dcw) G. oxydans. The bioconversion
was completed in 48 h yielding over 700 kg of DBS.
electronic effects (compare 2.3-fold for the change in pK
and 10-fold lower due to steric effects. Overall, this predicted
0-45-fold lower equilibrium constant for the n-butyl
a
)
4
derivative than for the unsubstituted compound. This pre-
dicted large decrease in the equilibrium constant for attack
of the basic amino nitrogen on the carbonyl carbon (C-5)
was consistent with the increased stability of the bioconver-
sion product obtained from NBG relative to that obtained
from glucamine and the equilibria shown in Figure 4.
Scale-Up. The process was scaled up to produce material
for clinical trials at a toll manufacturer, Abbott Laboratories,
Conclusions
By including one regioselective bioconversion step, it was
possible to simplify synthesis of N-butyldeoxynojirimycin
and utilize inexpensive starting materials. The key to
implementing this technology, however, was the realization
of the contribution of competing reactions and the relative
Vol. 6, No. 4, 2002 / Organic Process Research & Development
•
551

Figure 7. Bioconversion of NBG to DBS at 5500-L scale, 15 °C, 220 g/L NBG; (O) % conversion; (0) DBS concentration; (2) NBG
concentration.
energies of activation of the two reactions. Significant
enhancement of the selectivity was therefore achieved by
lowering the temperature at which the bioconversion was
run. The optimum temperature for this system was 12-15
Acknowledgment
J. Scanlon, who labored long and hard to optimize ion-
pair analytical chromatography for this biooxidation, pro-
1
vided analytical support. H NMR and its interpretation was
°
C. The kinetics of the bioconversion appeared to be first-
order, suggesting that the introduction of the charged
butylammonium group (at pH 5) increased the K for this
substrate. The pH maximum of the bioconversion rate was
.5-6.0. However, the small gain in rate relative to pH 5.0
was at least off-set by the increase in degradation of the
-deoxy-6-butylaminosorbose at the higher pH. It appeared
accomplished by P. Finegan. Synthesis of N-butylglucamine
was carried out by M. Scaros, M. Prunier and co-workers.
The assistance of all of these individuals greatly facilitated
this effort. We are grateful for their assistance.
m
5
6
Received for review January 29, 2002.
OP0255128
that nitrate salts of N-butylglucamine could replace chloride
salts, but sulfate, acetate, and phosphate salts could not.
552
•
Vol. 6, No. 4, 2002 / Organic Process Research & Development