Can the Combination of Electrochemical

Regeneration of NAD+, Selectivity of

L-␣-Amino-Acid Dehydrogenase, and
Reductive Amination of ␣-Keto-Acid Be
Applied to the Inversion of Configuration
of a L-␣-Amino-Acid?

Agnès Anne,1 Christian Bourdillon,2 Sandra Daninos,1 Jacques Moiroux1

1
The Laboratoire d’Electrochimie Moléculaire, Unité Mixte de Recherche
Université–CNRS No. 7591, Université de Paris 7, Denis Diderot, 2 Place
Jussieu, 75251 Paris Cedex 05, France; telephone: +33 1 44 27 28 01; Fax:
+33 1 44 27 76 25; e-mail: moiroux@paris7.jussieu.fr
2
The Laboratoire de Technologie Enzymatique, UPRES-A No 6022,
Université de Technologie de Compiègne, BP 20529,
60205 Compiègne Cedex, France
Received 19 June 1998; accepted 5 December 1998

Abstract: The inversion of configuration of L-alanine can
be carried out by combining its selective oxidation in the
presence of NAD+ and L-alanine dehydrogenase, electro-
chemical regeneration of the NAD+ at a carbon felt an-
ode, and reductive amination of pyruvate, i.e., reduction
of its imino derivative at a mercury cathode, the reaction
mixture being buffered with concentrated ammonium/
ammonia (1.28M / 1.28M). The dehydrogenase exhibits
astonishing activity and stability under such extreme
conditions of pH and ionic strength.
The main drawback of the process is its slowness. At
best, the complete inversion of a 10 mM solution of L-
alanine requires 140 h. A careful and detailed quantita-
tive analysis of each of the key steps involved shows that
the enzyme catalyzed oxidation is so thermodynamically
uphill that it can be driven efficiently to completion only
when both the coenzyme regeneration and the pyruvate
reduction are very effective. The first condition is easily
fulfilled. Under the best conditions, it is the rate of the
chemical reaction producing the imine which controls
the whole process kinetically. © 1999 John Wiley & Sons,
Inc. Biotechnol Bioeng 64: 101–107, 1999.
Keywords: electrochemical; regeneration; NAD; amino
acid; configuration
INTRODUCTION
It is now well established that the electrochemical regen-
eration of enzymatically active NAD+ can be achieved quite
efficiently either directly at a carbon anode (Bonnefoy et al.,
1988; Fassouane et al., 1990; Laval et al., 1987) or mediated
by polyoxometalates (Essaadi et al., 1994; Keita et al.,
1995; Keita et al., 1996; Sadakane and Steckhan, 1998;
Steckhan, 1994), quinonoid compounds (Blankespoor and
Miller, 1984; Tse and Kuwana, 1978; Persson et al., 1985)
and their transition-metal complexes (Hilt et al., 1997a,b),
Correspondence to: Jacques Moiroux
or by enzymatic systems (Cosnier et al., 1997; Palmore et
al., 1998). The direct electrochemical regeneration can be
performed with a yield exceeding 99.99%, but requires an
anodic potential that may be too positive for the selective
oxidation of some substrates (Bonnefoy et al., 1988). The
overpotential can be lowered markedly by the mediation.
Thus, it is tempting to make continuous use of the electro-
chemical regeneration for preparative processes (Somers et
al., 1997).
The oxidation of secondary alcohols and amines by
NAD+ to ketones in the presence of the relevant dehydro-
genases is thermodynamically uphill (Lee and Whitesides,
1985). In practical reactors the sole driving force provided
by the electrochemical regeneration of the coenzyme is not
sufficient to draw the oxidation toward completion even
when the configuration of the system is optimized (Fassou-
ane et al., 1990; Lortie et al., 1992). The only effective
method is to remove the product as it forms (Lee and
Whitesides, 1985). Several examples of alcohol dehydroge-
nase catalyzed oxidations of alcohols by NAD+ to obtain
chiral products have been reported, various methods being
used for the regeneration of NAD+ (Hilt et al., 1997a,b;
Jones, 1985; Lee and Whitesides, 1985).
There exist only two possibilities of applying the regen-
eration of NAD+ to the preparation of a D-␣-alcohol or
amino acid. In the first case, it may be envisaged to proceed
to the selective oxidation of the L-enantiomer of a racemic
mixture in the presence of the L-enzyme and end up with a
mixture of the carbonyl product and the D-enantiomer. In
the second case, the selective oxidation of the L-enantiomer
is coupled to a non-enzymatically catalyzed reduction of the
ketone which produces both enantiomers. As a result, half a
molecule of L-enantiomer is transformed into half a mol-

© 1999 John Wiley & Sons, Inc. CCC 0006-3592/99/010101-07

ecule of D-enantiomer each time a molecule of NAD+ is
regenerated. Then one can start either with the L-enantiomer
or with the ketone and transform it completely into the sole
D-enantiomer. Previously, the inversion of configuration of
L-lactate was thus achieved, in the presence of L-lactate
dehydrogenase, by coupling the electrochemical regenera-
tion of NAD+ at an anode to the electrochemical reduction
of pyruvate at a cathode (Biade et al., 1992). In the present
work, we show that the same approach can be applied to the
inversion of configuration of a L-␣-amino acid provided
that the medium is buffered with concentrated NH3/NH4+
and that the enzyme is stable and still satisfactorily active
under such extreme conditions of pH and ionic strength. The
scheme of the process is given in Figure 1.
L-alanine is the starting ␣-amino acid. L-alanine dehy-
drogenase (L-AlaDH) is the enzyme which catalyzes its
oxidation into pyruvate. Like other ␣-keto carboxylates, py-
ruvate and its imine exist at equilibrium in the presence of
ammonia. We take advantage of the fact that the electro-
chemical reduction of the imine is markedly easier and pro-
duces the racemic 0.5 L-alanine + 0.5 D-alanine. Therefore,
the cathode potential can be tuned so as to achieve a selec-
tive reduction (Anne et al., 1994; Jeffery and Meisters,
1978; Lund, 1970).
The key steps of the reaction scheme involved in Figure
1 are the electrochemical regeneration of NAD+ which is
irreversible and rapid, provided that the anode potential is
positive enough (Biade et al., 1992; Bonnefoy et al., 1988;
Fassouane et al., 1990; Laval et al., 1987), the reversible
enzymatically catalyzed reaction, the reversible formation
of the imine, and the irreversible electrochemical reduction
of the latter (Anne et al., 1994). The flow of two moles of
electrons through the system provokes the transformation of
half a mole of L-alanine into the same quantity of D-alanine
while one mole of NAD+ undergoes a complete redox cycle.
Besides demonstrating that the inversion of configuration of
the amino acid is feasible in such a process, the quantitative
analysis of its kinetic behavior allows for the determination
of the steps which control the overall rate of D-enantiomer
production. As a result, the important parameters can be
identified, and it is possible to show how the rate could be
improved and thus, evaluate the optimized limit that could
Figure 1. Scheme of the process. The starting materials can be either
L-alanine, racemic D,L-alanine or pyruvate, NAD+ or NADH.
102 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 64, NO. 1, JULY 5, 1999
be reached practically. Such an optimized limit appears too
low for technical application.
MATERIALS AND METHODS
Materials
Enzymes and biochemicals, including L-alanine, D-alanine
and D, L-alanine were purchased from Sigma, Saint Quen-
tin Follonier, France, and used without further purification.
L-alanine dehydrogenase (EC 1. 4. 1. 1) and D-amino-acid
oxidase (EC 1. 4. 3. 3) and L-lactic acid dehydrogenase (EC
1. 1. 1. 27) were from Bacillus subtilis, porcine kidney, and
rabbit muscle, respectively. All other chemicals and the Na-
fion perfluorinated membranes separating the working elec-
trode and auxiliary electrode compartments inside each re-
actor were from Aldrich, Saint Quentin Follonier, France.
The NH3/NH4+ (1.28M / 1.28M) buffer was obtained by
dissolution of (NH4)2 SO4 and concentrated aqueous am-
monia (14.7M) from Prolabo in distilled water. The felt of
carbon fibers (RVG 4000) was obtained from Carbone-
Lorraine, Geneillies, France.
Apparatus
For the study of enzyme kinetics, the NADH concentration
was determined by measuring the absorbance at 340 nm (␧
⳱ 6320 M−1 cm−1) with a Hewlett-Packard HP 8452 spec-
trophotometer. The potentials of the working electrodes vs.
SCE electrodes were controlled with home-made potentio-
stats. The flow rates of the circulating solutions were con-
trolled with a Gilson Minipuls (Villius le Bel, France) 3
peristaltic pump equipped with a 4-channel head.
Reactors
The reactor in which NAD+ was regenerated was as previ-
ously described (Fassouane et al., 1990). It was used in the
vertical position, the solutions flowing from bottom to top.
The reduction reactor was identical but used in the horizon-
tal position, the carbon felt being replaced by the mercury
pool. The reactors were fed as previously described (Biade
et al., 1992).
Enzymatic Assays
The method for the enzymatic assay of L-alanine was es-
sentially similar to that described by Williamson (1985).
Exhaustive catalyzed oxidation was achieved in the pres-
ence of NAD+ and hydrazine in excess. The determination
was carried out as follows: 0.3 mL of sample containing
L-alanine and 2.5 mL of 0.2M sodium pyrophosphate buffer
containing 0.48M hydrazine and 1.25 mM EDTA (pH 9.2)
were mixed in a 1 cm spectrophotometric cell. After a 10
min standing at 25°C, NAD+ (0.1 mL of a 60 mM solution
in H2O) was added, and the contents were mixed again. The
enzymatic reaction was started by addition of the solution of
L-alanine dehydrogenase (0.1 mL of a 0.9 mg/mL solution).
L-alanine was quantified from the increase of absorbance at
340 nm after completion of the reaction (approximately 60
min). The enzymatic determination of L-alanine was unaf-
fected by the presence of D-alanine in the sample.
The enzymatic assay of D-alanine was performed accord-
ing to a previously described procedure (Gra␤l and Supp,
1985) that involves complete dioxygen D-amino acid oxi-
dase catalyzed oxidation to pyruvate followed by NADH
L-lactate dehydrogenase catalyzed reduction. Once cor-
rected for the blank test, the decrease in absorbance at 340
nm was used for the quantitative determination of D-alanine
which was unaffected by the presence of L-alanine.
RESULTS AND DISCUSSION
First, we investigated separately the thermodynamics and
the kinetics of each step of the process.
The Keto-Imine Transformation
The keto-imine equilibrium (Scheme 1) was characterized
thermodynamically and kinetically as described previously
in the case of ␣-keto-glutarate (Anne et al.,1994).
The equilibrium constant K1 ⳱ k1/k−1 can be determined
by means of cyclic voltammetry, the peak heights being
proportional to the imine and keto concentrations at equi-
librium (Anne et al., 1994). The peak potentials given in
Scheme 1 were measured at a potential scan rate of 0.2 V/s
in the NH3/NH4+ (1.28M / 1.28M) buffer (pH ⳱ 9.2). The
equilibrium constant and the rate constants k1 and k−1 can be
determined altogether through careful analysis of the cur-
rent vs. time dependence during the electrolysis of the so-
lution at a controlled potential at which only the imine is
reducible (Anne et al., 1994; Bard and Santhanam, 1970;
Bard and Solon, 1963). We found k1 ⳱ 3.3M−1 h−1, k−1 ⳱
16.5 h−1 and K1 ⳱ 0.2. Therefore, in the presence of am-
monia in excess, the ratio of the imine concentration over
the keto concentration is K1CNH3 and the pseudo-first-order
rate constant of imine formation is K1CNH3, CNH3 being the
ammonia concentration in M. The higher CNH3, the more
favored the imine formation. However, in the inversion pro-
cess, the reaction mixture must be buffered owing to the
poor stability of NAD+ in exceedingly basic media (Wong
ANNE ET AL.: INVERSION OF CONFIGURATION OF A L-␣-AMINO-ACID
and Whitesides,1981). We chose to operate in the NH3/
NAD4+ (1.28M / 1.28M) buffer (pH ⳱ 9.2) the counter ion
being sulfate. At a greater buffer concentration, the solution
becomes too viscous.
Rate of the Electrochemical Reduction of
the Imine
The flow-through reactor is of the same type as the one
already used for the inversion of L-lactate (Biade et al.,
1992). It has a volume vred of 10 mL. To test its efficiency,
a 2 mM solution of pyruvate in the NH3/NH4+ (1.28M /
1.28M) buffer (pH ⳱ 9.2) plus 0.5 mM NAD+ and 0.1
mg/mL L-AlaDH is pumped into the reactor at a flow rate
of 20 mL/h. At the output, D- and L-alanine (see Materials
and Methods) and L-lactate (Biade et al., 1992), are assayed
enzymatically. Pyruvate is assayed spectrophotometrically
(Biade et al., 1992), and by means of cyclic voltammetry
(Anne et al., 1994). No lactate is produced as long as the
mercury cathode potential does not exceed −1.35 V vs. the
KCl saturated calomel electrode (SCE) on the negative side.
At −1.35 V/SCE, both the D and L-alanine concentrations
are 0.16 mM, a result confirming that the reductive amina-
tion does take place and produces the racemic D,L-alanine
(concentration: 0.32 mM).
The rate of pyruvate reduction is: dnPRED/dt ⳱ 0.32 ×
10−3 × 20 × 10−3 ⳱ 6.4 × 10−6 mol/h.
It is the imine which is actually reduced. At equilibrium,
in the presence of 1.28M NH3, the imine (CI)e / pyruvate
(CP)e concentration ratio is (CI)e/(CP)e ⳱ K1CNH3 ⳱ 0.256
and (CI)e ⳱ 0.41 mM, (CP)e ⳱ 1.59 mM. The rate of the
imine → keto reaction k−1CIvred is 53 × 10−6 mol/h, i.e.,
much greater than the rate of electrochemical reduction
dnPred/dt. Therefore, the keto/imine transformation is al-
ways at equilibrium in the reduction reactor under such
circumstances and, at constant flow rate and cathode poten-
tial, the rate of electrochemical reduction can be considered
as pseudo-first-order (Bard and Santhanam, 1970; Bard and
Solon, 1963), i.e., dnPred/dt ⳱ Kred CP with Kred ⳱ 4 × 10−3
L/h. The current i is i ⳱ 2F dnPred/dt and (i/mA) ⳱
0.216(CP/mM).
As long as the reductive amination is kinetically con-
trolled by the electrochemical reduction of the imine, the
greater the flow rate, the greater Kred. Similarly, the smaller
the vred/A ratio (A ⳱ area of the mercury pool), the greater
Kred. The limiting situation is reached when the rate of
imine formation becomes the rate-determining step, then
Kred levels off at Kred,max ⳱ k1vred ⳱ 3.3 × 10−2 L/h. Rais-
ing vred at constant vred /A ratio would cause an equal in-
crease in the total volume of the reaction mixture vT which
ought to be kept as small as possible, as discussed later.
Rate of the Electrochemical Regeneration
of NAD+
The flow-through reactor is of the same type as the one
already used for the inversion of L-lactate (Biade et al.,
103
1992). The anode is made of carbon felt with an effective
surface area of 5500 cm2, the void volume vox filled with the
solution being 11 mL. To test the regeneration efficiency, a
0.5 mM solution of reduced coenzyme is pumped into the
reactor at a flow rate of 100 mL/h. At the output, the en-
zymatic assay of the solution ascertains that NAD+ is com-
pletely regenerated from its reduced forms which can be, in
the present case, NADH or the dimers produced by mono-
electronic reduction of NAD+ at a mercury cathode (Biade
et al., 1992). The electrochemical regeneration is as efficient
in the NH3/NH4+ (1.28M / 1.28M) buffer (pH ⳱ 9.2) as in
other buffers (Biade et al., 1992; Bonnefoy et al., 1988;
Fassouane et al., 1990; Laval et al., 1987), provided that the
anode potential is positive enough (> 0.4 V/SCE). In the
following, the rate of electrochemical regeneration of NAD+
will be always taken as very fast when compared to the
other key steps of the process.
Enzyme Kinetics
As previously mentioned, the enzyme catalyzed equilibrium
(Scheme 2) is thermodynamically unfavorable.
In the following, B ⳱ L-alanine, A ⳱ NAD+, P ⳱ py-
ruvate, Q ⳱ NADH, and Cj ⳱ concentration of species j.
The equilibrium constant K2 is defined as:
+ ously described for the inversion of L-lactate (Biade et
共CP兲eq共CQ兲eqCNH3CH
K2 =
共CB兲eq共CA兲eq
All the components of the system, including the enzyme, are
stable in the NH3/NH4+ (1.28M / 1.28M) buffer (pH ⳱ 9.2)
over a period exceeding 24 h.
At an enzyme concentration of 5 ␮g/mL, the equilibrium
is reached after 1 h and K2 in the medium is found to be K2
⳱ 8.0 × 10−14, all concentrations being expressed in M, a
value of the same order of magnitude as those found in other
buffers (Grimshaw and Cleland, 1981; Yoshida and Freese,
1964). The occurrence of the keto/imine equilibrium is
taken into account for the calculation of the pyruvate con-
centration at equilibrium in solution. For CH+ ⳱ 10−9.2 M
and CNH3 ⳱ 1.28M, that gives an apparent equilibrium
constant K2app ⳱ (CP)eq (CQ)eq /(CB)eq(CA)eq of 1.0 × 10−4.
When the enzyme concentration is lowered, it is possible
to obtain a linear variation of NADH (Q) absorbance at 340
nm over 10 min, starting either with the reactants A and B
(initial concentrations C°A and C°B) to measure the rate of
the forward reaction or with the products P and Q (initial
concentrations C°P and C°Q) to measure the rate of the
reverse reaction. In the first case, the rate of the appearance
of Q is measured at C°B ⳱ 16.7 mM and various C°A in the
104 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 64, NO. 1, JULY 5, 1999
1 to 10 mM range, or at C°A ⳱ 5 mM with C°B in the 5 to
50 mM range. Similarly, in the second case, the rate of
disappearance of Q is measured at C°P ⳱ 10 mM and 0.025
ⱕ C°Q ⱕ 0.2 mM or C°Q ⳱ 0.2 mM and 3 ⱕ C°P ⱕ 10
mM. The respective dependencies of the initial rates Vi2 and
Vi−2 of the forward and reverse reactions on C°A and C°B on
the one hand, and on C°P and C°Q on the other hand, can be
analyzed according to equations (1) and (2).
Vi2 Vf
= (1)
C°E KMA KMB KiaKMB
1+ + +
CA CB CACB
and
Vi–2 Vr
= (2)
C°E KMP KMQ KiqKMP
1+ + +
CP CQ CPCQ
Equations (1) and (2) fit well the relevant linear reciprocal
plots of the experimental data. The following quantitative
features of the enzyme kinetics can thus be derived: Vf ⳱ 25
and Vr ⳱ 60 ␮mol/min/mg enzyme; they characterize the
enzyme activity in the NH3/NH4+ buffer, and the constants
KMA ⳱ 3.0 mM, KMB ⳱ 12 mM, Kia ⳱ 0.01 mM, KMP ⳱
3.3 mM, KMQ ⳱ 0.12 mM and Kiq < 0.01 mM as well. In the
presence of 10 mM D-alanine, Vi−2 is unaffected and Vi2
undergoes a slight decrease of less than 5%.
Simultaneous Occurrence of the Enzymatically
Catalyzed and Electrochemically Driven Reactions
The oxidation and reduction reactors are coupled as previ-
al.,1992). The total volume vT circulating in the compart-
ments of the two working electrodes, the reservoir, and the
connecting tubes was minimized at vT ⳱ 38 mL. In a typical
experiment, the initial solution contains NAD+ (C°A ⳱ 0.5
mM), L-alanine (C°B ⳱ 10 mM) and L-AlaDH at a con-
centration C°E ⳱ 0.1 mg/mL. A quasi-steady-state is
reached after 1 h. Then, the electric currents through both
reactors have the same absolute value i. Aliquots of the
reaction mixture are taken out at various times t, and D-
alanine is thus enzymatically assayed so as to follow the
progress of its production. The measured D-alanine yield ␳
⳱ CD−ala/C°B and i vs. t plots are reproduced in Figure 2.
Quantitative Analysis of the Inversion Process
As the reductive amination is not very efficient the enzy-
matically catalyzed reaction is essentially driven by the fast
regeneration of NAD+. Within the oxidation reactor the ki-
netics is controlled by the enzymatically catalyzed reaction
which operates reversibly. Assuming that the enzyme ca-
talysis follows an ordered bibi mechanism, under the pres-
ent experimental conditions, its rate VE is given by Segel
(1975):
 Taking into account that CQ ⳱ 0, owing to the very fast
electrochemical regeneration of NAD+, and that (Vf /K2app)/
Vr ⳱ (25/60) × 104, equation (3) then reduces to:
VE K2appVrC°ACB
=
C°E KMQCP共1 + C°A Ⲑ Kia兲
Therefore, the rate of production of pyruvate dnPox/dt ⳱
voxVE within the oxidation reactor is:
dnPox K2appVrC°ACB
= v C° × 60 × 10−6
dt KMQCP共1 + C°A Ⲑ Kia兲 ox E
dnPox CB
i.e., = Kox
dt CP
with dnPox/dt in mol/h, va in mL and Vr in ␮mol/min.mg
enzyme, and:

K2appVrC°A voxC°E × 60 × 10−6

Kox = (4)
KMQ 共1 + C°A Ⲑ Kia兲
i.e., Kox ⳱ 3.3 × 10−8 mol/h in the present case.
At quasi-steady-state dnPox/dt ⳱ dnPred/dt, then the pyru-
vate concentration is given by:

Kox
CB
CP
= KredCP and CP = 冑 Kox
Kred
公CB
The rate of electrochemical reduction of pyruvate—dnP /dt
is:
dnP
− = KredCP = 公KredKox 公CB (5)
dt
and

i Ⲑ mA =
2F
3600 冉 冊
−
dnP
dt
× 1000 = 53.6

× 103 公KredKox 公CB (6)

Figure 2. (a) D-alanine production yield CD-ala/C°L-ala vs.time plot. (b) The rate of production of D-alanine dnD-ala/dt is half that of
Current i vs. time plot (continuous line). C°L-ala ⳱ 10 mM. C°E ⳱ 0.1
mg/mL. C°NAD ⳱ 0.5 mM, flow rate 20 mL/h. Dotted curves ⳱ computed
the electrochemical reduction of pyruvate, thus:
values with Kox ⳱ 3.3 × 10−8 mol/h and Kred ⳱ 4 × 10−3 L/h. (■):
experimental data. The dashed curve in (a) is computed with Kox,max ⳱ 3.3 dnD-ala 1 dnP 公KredKox 公CB
=− =
× 10−7 mol/h and Kred,max = 3.3 × 10−2 L/h (see Text). dt 2 dt 2
or
VE
= dCD-ala 公KredKox 公CB 公KredKox
C°E = −3
= k公CB, with k = ,
dt 2 × 10 vT 2 × 10−3vT
冉
VfVr CACB −
CPCQ
K2app 冊 vT being still expressed in mL. Keeping in mind that C°B ⳱
CB + CD-ala + CP + CI or C°B ⳱ CB + CD-ala + 1.256CP,
冉
Vr KiaKMB + KMBCA + KMACB + CACB +
KMACBCQ
Kiq
because the imine concentration CI is 0.256CP in the pres-
ence of 1.28 M NH3, then:

+
CACBCP
Kip
+冊 Vf
冉
K C + KMPCQ + CPCQ +
K2app MQ P
KMQCACP
Kia or
CB = C°B − CD-ala − 1.256 公Kox Ⲑ Kred 公CB
CB + 2␭公CB − 共C°B − CD-ala兲 = 0

+
CBCPCQ
Kib 冊 (3)
with
and
␭ = 0.628公Kox Ⲑ Kred
公CB = ␭共公1 + 共C°B − CD-ala兲 Ⲑ ␭2 − 1兲 (7)

ANNE ET AL.: INVERSION OF CONFIGURATION OF A L-␣-AMINO-ACID 105

The rate of production of D-alanine becomes:
= k␭共公1 + C°B Ⲑ ␭2 − CD-ala Ⲑ ␭2 − 1 兲
dCD-ala
dt
or
d共CD-ala Ⲑ C°B兲 k␭dt
=
公1 Ⲑ C°B + 1 Ⲑ ␭ 2
− CD-ala Ⲑ C°B␭ − 1 Ⲑ 公C°B
2 公C°B
Integration gives:
共公1 Ⲑ C°B + 1 Ⲑ ␭2 − 公1 Ⲑ C°B + 1 Ⲑ ␭2 − ␳ Ⲑ ␭2兲
冉 冊
1 公1 Ⲑ C°B + 1 Ⲑ ␭2 − 1 Ⲑ 公C°B
+ ln
公C°B 公1 Ⲑ C°B + 1 Ⲑ ␭2 − ␳ Ⲑ ␭2 − 1公C°B
kt
= (8)
2␭公C°B
Then equation (8) can be used for the computation of the
D-alanine production yield ␳ ⳱ CD-ala/C°B vs. t plot.
When CP is negligible compared to C°B −CD-ala, i.e.,
practically as long as ␳ ⱕ 0.9, equation (8) reduces to:
公KoxKred KoxKred 2
␳= t− t (9)
2vT公C°B 16vT2C°B
Computed curves are reproduced in Figure 2a.
Equation (6) also becomes:
i Ⲑ mA = 53.6 × 103␭公KredKoxC°B
共公1 Ⲑ C°B + 1 Ⲑ ␭2 − ␳ Ⲑ ␭2 − 1 Ⲑ 公C°B兲
Equation (8) gives t for each ␳ and the i vs. t plot can also
be computed as shown in Figure 2b.
Optimization of the Process
In Figure 2, the dotted curves which fit the measurements
were computed with the Kox and Kred values corresponding
to the chosen experimental conditions, namely the initial
concentrations C°E, C°A, C°B, the flow rate, the reactor’s
characteristics, and the electrodes potentials. In the present
case, Kox and Kred are 3.3 × 10−8 mol/h and 4 × 10−3 L/h,
respectively.
Figure 2 shows that there is very good agreement be-
tween the observed and simulated behaviors. Such an agree-
ment, which holds over a period of 20 d, ascertains the great
stability of both the enzyme and the coenzyme in the NH3/
NH4+ (1.28 M / 1.28 M) buffer (pH ⳱ 9.2). Taking into
account that the L-AlaDH molar weight is 228,000 (Grim-
shaw and Cleland, 1981; Yoshida and Freese, 1964), and
that the enzymatic system undergoes two redox cycles each
time one molecule of D-alanine is produced, the enzyme
turnover exceeds 3 × 107.
The remarkable fit between the measured and computed
data also justifies the approach, and the accompanying ap-
106 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 64, NO. 1, JULY 5, 1999
proximations, used for the quantitative analysis of the pro-
cess. Thus, it becomes possible to examine systematically
the effects of the various parameters that can be experimen-
tally adjusted so as to improve the productivity, i.e., de-
crease the time needed to reach a given D-alanine produc-
tion yield ␳ ⳱ CD-ala / C°B. The latter depends on k, ␭, and
C°B, i.e., on Kox, Kred, vT and C°B. Equations (5), (7), and (8)
indicate that ␳ would theoretically increase with decreasing
vT and/or increasing Kred and/or Kox. Practically, it is quite
difficult to reduce significantly the total volume vT of cir-
culating solution.
As mentioned earlier, the cathodic efficiency can be en-
hanced by increasing the flow rate and/or by increasing the
ratio of the mercury cathode area over the working com-
partment volume inside the reduction reactor. However, as
already pointed out, Kred cannot be raised above the limit
Kred,max ⳱ 3.3 × 10−2 L/h which corresponds to the kinetic
control by the rate of imine formation.
According to equation (4), Kox depends on K2app, vox, the
enzyme kinetic characteristics Vr, KMQ, Kia and the initial
concentrations C°A and C°E. The greater the pH the greater
the apparent equilibrium constant K2app. However, pH can-
not be increased because the stability of NAD+ decreases
dramatically above pH 9.2 (Wong and Whitesides, 1981).
Once the pH is settled the enzyme kinetic characteristics
cannot be changed. Any increase in the volume of the work-
ing electrode compartment vox of the oxidation reactor
would cause an equal increase in the total volume of circu-
lating solution vT. Due to the low value of Kia, Kox becomes
practically C°A independent once C°A ⱖ 0.5 mM and cannot
be improved by raising C°A above this level. Indeed, the
same i is measured at C°A ⳱ 1 mM. Therefore, the only
means of increasing Kox significantly is to raise C°E. We
checked out that i undergoes a twofold increase when C°E is
increased four times. Practically C°E can be increased 10
times, i.e. up to C° E ⳱ 1 mg/mL, that gives Kox ⳱ Kox,max
⳱ 3.3 × 10−7 mol/h.
The dependence of ␳ ⳱ CD-ala/C°B on C°B is given by
equation (8) and more explicitly by equation (9). Clearly,
at any time, the greater C°B the smaller ␳. If the times are
not too long, ␳ and CD-ala are even proportional to 1/√C°B
and √C°B respectively. Therefore, if raising CD-ala is the
main purpose, it can be achieved by increasing C°B. The
opposite conclusion holds if raising ␳ is the main purpose.
Keeping C°B ⳱ 10 mM, the dash-dotted curve in Figure
2a shows the ␳ growth that could be expected if the experi-
mental conditions, allowing for Kox ⳱ Kox,max and Kred ⳱
Kred,max, were fulfilled. Complete inversion of configuration
would then be obtained only after 140 h, quite a long time.
Obviously, even under the optimal conditions, the produc-
tivity remains too low. Therefore, the present process can-
not be suitable for general use on a preparative scale.
CONCLUSION
The L → D inversion of an ␣-amino-acid can be carried out
through the combination of the following electrochemically
driven reactions: selective oxidation of the L-enantiomer by
NAD+ in the presence of the L-␣-amino-acid dehydroge-
nase catalyst, electrochemical regeneration of the NAD+
coenzyme at a carbon felt anode, and reductive amination of
the ␣-keto-acid at a mercury cathode, the buffer of the re-
action mixture being made of concentrated ammonium/
ammonia (1.28M / 1.28M). The dehydrogenase exhibits as-
tonishing activity and stability under such extreme condi-
tions of pH and ionic strength; both remain apparently
unchanged after 20 d of functioning at room temperature.
A careful and detailed quantitative analysis of each of the
key steps involved shows that the enzyme catalyzed oxida-
tion of the L-enantiomer is so thermodynamically uphill that
it can be driven efficiently to completion only when both the
coenzyme regeneration and the ␣-keto-acid reduction are
very effective. The results confirm that the electrochemical
regeneration of NAD+ can be made as fast and efficient as
required. However, the reduction of the ␣-keto-acid must
produce only the ␣-amino-acid, i.e., it is the reduction of the
imine that must take place selectively. Then, under the best
conditions, the whole process becomes kinetically con-
trolled by the rate of imine formation whatever the nature of
the reducing reagent, even if it is a mercury cathode held at
a finely tuned potential as is the case in the present work.
Consequently, the process is irremediably slow. At best, the
complete inversion of configuration of a 10 mM solution of
L-alanine would require 140 h.
References
Anne A, Daninos S, Moiroux J, Bourdillon C. 1994. Electrochemical re-
duction of ␣-ketoglutarate in the presence of ammonia as a means of
achieving selectively the reductive amination to glutamate. Thermo-
dynamic and kinetic characteristics of the keto/imine equilibrium. New
J Chem 18:1169–1174.
Bard AJ, Santhanam KSV. 1970. Application of controlled potential cou-
lometry to the study of electrode reactions. In: Bard AJ, editor. Elec-
troanalytical chemistry, Vol. 4. New York: Marcel Dekker. p 215–232.
Bard AJ, Solon E. 1963. Secondary reactions in controlled potential cou-
lometry. III - Preceeding and simultaneous chemical reactions. J Phys
Chem 67:2326–2330.
Biade AE, Bourdillon C, Laval JM, Mairesse G, Moiroux J. 1992. Com-
plete conversion of L-lactate into D-lactate. A generic approach in-
volving enzymatic catalysis, electrochemical oxidation of NADH, and
electrochemical reduction of pyruvate. J Am Chem Soc 114:893–897.
Blankespoor RL, Miller LL. 1984. Electrochemical oxidation of NADH,
kinetic control by inhibition and surface coating. J Electroanal Chem
171:231–241.
Bonnefoy J, Moiroux J, Laval JM, Bourdillon C. 1988. Electrochemical
regeneration of NAD+. A new evaluation of its actual yield. J Chem
Soc Faraday Trans 1 84:941–950.
Cosnier S, Fontecave M, Innocent C, Nivière V. 1997. An original elec-
troenzymatic system: flavin reductase-riboflavin for the improvement
of dehydrogenase-based biosensors. Application to the amperometric
detection of lactate. Electroanalysis 9:685–688.
Essaadi K, Keita B, Nadjo L, Constant R. 1994. Oxidation of NADH by
oxometalates. J Electroanal Chem 367:275–278.
Fassouane A, Laval JM , Moiroux J, Bourdillon C. 1990. Electrochemical
regeneration of NAD in a plug-flow reactor. Biotechnol Bioeng 35:
935–939.
Gra␤l M, Supp M. 1985. D-alanine. In: Bergmeyer HU, editor. Methods of
enzymatic analysis, Vol. 8, 3rd ed. Weinheim: Verlag Chemie. p
336–340.
ANNE ET AL.: INVERSION OF CONFIGURATION OF A L-␣-AMINO-ACID
Grimshaw CE , Cleland WW. 1981. Kinetic mechanism of Bacillus subtilis
L-alanine dehydrogenase. Biochem 20:5650–5655.
Hilt G, Jarbawi T, Heineman WR, Steckhan E. 1997. An analytical study
of the redox behavior of 1,10-phenanthroline-5,6-dione, its transition-
metal complexes, and its N-monomethylated derivative with regard to
their as mediators of NAD(P)+ regeneration. Chem Eur J 3:79–88.
Hilt G, Lewall B, Montero G, Utley JHP, Steckhan E. 1997. Efficient
in-situ redox catalytic NAD(P)+ regeneration in enzymatic synthesis
using transition-metal complexes of 1,10-phenanthroline-5,6-dione
and its N-monomethylated derivative as catalyst. Liebigs Ann /Recueil
2289–2296.
Jeffery EA , Meisters A. 1978. Electrochemical synthesis of amino acids by
reductive amination of keto acids. I - Reduction at mercury electrodes.
Aust J Chem 31:73–78.
Jones JB. 1985. Enzymes as chiral catalysts. In: Morrison JD, editor.
Asymmetric synthesis, chiral catalysts, Vol. 5. New York: Academic
Press. p 309–339.
Keita B, Essaadi K, Nadjo L, Constant R, Justum Y. 1996. Oxidation
kinetics of NADH by heteropolyanions. J Electroanal Chem 404:
271–279.
Keita B, Essaadi K, Nadjo L, Desmadril M. 1995. Rate-limiting one-
electron transfer in the oxidation of NADH by polyoxometalates.
Chem Phys Lett 237: 411–418.
Laval JM , Bourdillon C, Moiroux J. 1987. The electrochemical regenera-
tion of NAD+ revisited. Biotechnol Bioeng 30:157–159.
Lee LG , Whitesides GM. 1985. Enzyme-catalyzed organic synthesis: A
comparison of strategies for in-situ regeneration of NAD from NADH.
J Am Chem Soc 107:6999–7008.
Lortie R, Fassouane A, Laval JM , Bourdillon C. 1992. Displacement of
equilibrium in electroenzymatic reactor for acetaldehyde production
using yeast alcohol dehydrogenase. Biotechnol Bioeng 39:157–163.
Lund H. 1970. Electrochemistry of the carbon-nitrogen double bond. In:
Patai S, editor. The chemistry of the carbon-nitrogen double bond.
New York: Wiley. p 536.
Palmore GTR , Bertschy H, Bergens SH, Whitesides GM. 1998. A metha-
nol/dioxygen biofuel cell that uses NAD+-dependent dehydrogenases
as catalysts: Application of an electro-enzymatic method to regenerate
nicotinamide adenine dinucleotide at low overpotentials. J Electroanal
Chem 443:155–161.
Persson B, Gorton L, Johansson G, Torstensson A. 1985. Biofuel anode
based on D-glucose dehydrogenase, nicotinamide adenine dinucleotide
and a modified electrode. Enzyme Microb Technol 7:549–552.
Sadakane M, Steckhan E. 1998. Electrochemical properties of polyoxo-
metalates as electrocatalysts. Chem Rev 98:219–237.
Segel IH. 1975. Enzyme kinetics. New York: Wiley. p 560–565.
Somers WAC, van Hartingsveldt W, Stigter ECA, van der Lugt JP. 1997.
Electrochemical regeneration of redox enzymes for continuous use in
preparative processes. TIBTECH 15:495–500.
Steckhan E. 1994. Electroenzymatic synthesis. In: Steckhan E, editor. Top-
ics in current chemistry, Electrochemistry V, Vol. 170. Berlin:
Springer-Verlag. p 84–107.
Tse DCS, Kuwana T. 1978. Electrocatalysis of dihydronicotinamide aden-
osine diphosphate with quinones and modified quinone electrodes.
Anal Chem 50:1315–1318.
Williamson DH. 1985. L-alanine. In: Bergmeyer HU, editor. Methods of
enzymatic analysis, Vol. 8, 3rd ed. Verlag Chemie: Weinheim. p
341–344.
Wong CH, Whitesides GM. 1981. Enzyme-catalyzed organic synthesis:
NAD(P)H cofactor regeneration by using 6-phosphate and the glucose
6-phosphate dehydrogenase from Leuconostoc mesenteroides. J Am
Chem Soc 103:4890–4899.
Yoshida A, Freese E. 1964. Purification and chemical characterization of
alanine dehydrogenase of Bacillus subtilis. Biochim Biophys Acta
92:33–43.
107