BioprocessEngineering14 (z996)z9a 297 9 Springer-Verlag1996

Continuous production of L-tert-leucine in series of two enzyme

membrane reactors
Modelling and computer simulation

U. Kragl, D. Vasic-Racki, C. Wandrey

291

Abstract The L-tert-leucine synthesis was performed continu- The modelling in this work shows that the optimisation of
ously in series of two enzyme-membrane reactors by reductive the quantity of the enzyme used results in a minimisation of
amination of trimethylpyruvate with leucine dehydrogenase. the biocatalyst costs.
The necessary "native" cofactor NADH is regenerated with the
aid of a second enzyme, formate dehy-
drogenase. List of symbols
Considering detailed kinetic studies of initial reaction rates A volumetric enzyme activity U ml-1
under conditions relevant to the process a kinetic model was C concentration mmol litre- 1
developed. The model shows that the overall reaction rate is E enzyme concentration mg m l -
strongly inhibited by the reaction product. The reactor's kdes enzyme deactivation constant d-
models combine the mass balances and proposed kinetic Km Michaelis-Menten constant mmol litre- 1
equations. The model adequacy was verified by using it to Ki inhibition constant mmol litre-'
simulate the experiments and by comparing experimental and Me molecular weight of product g mol-
computed conversion, space-time yield and enzyme consump- STY space-time yield g Iitre- ~ d - 1
tion. t time min, h or d
The calculations for the three reactor's types (batch, single T temperature ~
CSTR and a cascade of two CSTRs in series) were compared. r residence time min, h or d
The results showed that a single CSTR is no favourable reactor vl reaction rate U mg-
configuration due to the very strong product inhibition. V,~ maximum reaction rate U mg-
Space-time yield drops from 560 g litre-~ day-~ in a batch VR reactor volume ml
reactor to 110 g litre-~ day-1 in a single CSTR at the highest
conversion of 98 %. At the conversion of 95 % the difference in Subscript
biocatalyst costs between batch and two CSTR in series is o feed
negligible. Therefore the use of two enzyme membrane
reactors in series was proposed. Abbreviations
CSTR Continuous Stirred Tank Reactor
EMR Enzyme Membrane Reactor
FOR Formate
FDH Formate Dehydrogenase
Received: 5 September 1995 LeuDH Leucine Dehydrogenase
NADH cofactor-NicotineAmideadenineDinucleotide
U. Kragl, C. Wandrey
Institut ffir Biotechnologie, Forschungszentrum Jiilich GmbH, NH4+ ammonia
D-52425 J/ilich, Germany TLEU L-tert-Leucine
TMP Trimethylpyruvate
D. Vasic-Racki
Faculty of Chemical Engineering, University of Zagreb, HR-41000
Zagreb, Croatia
1
Correspondence to: C. Wandrey Introduction
The continuous stirred tank reactor (CSTR) is by far one of the
The authors wish to thank Claudia Burichter and Ursula Mackfeld for simplest reactor configurations for continuous process realisa-
their skilful technical assistance. The authors gratefully acknowledge tion and is widely used in many biochemical reactions.
the financial support from the Bundesministerium fiir Bildung, The use of CSTRs as continuously operating enzyme
Wissenschaft, Forschung und Technologie (BMBF) for a scholarship
membrane reactors with enforced flow where the retention of
of Mr. C. Plavljanic, Faculty of Chemical Engineering, University of
Zagreb, Zagreb, Croatia, who supplied the software NLO used for the soluble enzyme in the reaction vessel is achieved by means of
mathematical calculations. We thank Prof. Dr. K. Drauz and Dr. A. an ultrafiltration membrane has been described earlier in detail
Bommarius, Degussa AG, Hanau, Germany for stimulating discussions [1]. An essential advantage of this system is that the reactor
and for providing samples of trimethylpyruvate. may be operated under sterile conditions without diffusion
 Bioprecess Engineering 14 (1996)
limitations. In this way continuous homogeneous biocatalysis
is possible with a very good exploitation of the biocatalyst.
While being advantageous for reactions with substrate inhibi-
tions as in the acylase process [2], this reactor is disadvantage-
ous in cases of severe product inhibition just like in the
synthesis of L-tert-leucine from trimethylpyruvic acid that
has been studied in detail elsewhere [1]. A system of
two membrane reactors in series has been proposed as
a solution for the product inhibition problem for the reaction
in question.
292
2 4
The reaction system
The L-tert-leucine synthesis was performed by reductive
amination of trimethylpyruvate with leucine dehydrogenase
(E.C.1.4.1.9) [3, 4, 5] (Figure 1). The necessary cofactor NADH
is regenerated by the aid of a second enzyme, formate
dehydrogenase (E.C.1.2.1.2) [6]. It could be demonstrated by
modelling in previous work [1] that high total turnover
numbers can be achieved also with "native" cofactor (the total
turnover number specifies how many moles of product are
formed per mole of consumed cofactor). Today the cofactor
costs are no longer a limiting factor for enzymatic redox
reactions anywhere. Therefore in this work we performed the
synthesis of L-tert-leucine only with the native cofactor NADH
at lowest possible cofactor concentrations in order to avoid an
excess of cofactor and to achieve the highest possible total
turnover numbers. On the other hand the concentration was
sufficiently high, so that the space-time yield would not drop in
the case of a slight deactivation.
3 balance equations for the ideal CSTR and proposed kinetic
Materials, methods and apparatus
Formate dehydrogenase was obtained by fermentation in the
Institute of Biotechnology, Forschungszentrum GmbH Jiilich,
Germany [6]. Leucine dehydrogenase, NAD + and trimethyl-
pyruvate were a gift from Degussa AG, Hanau, Germany. All
other chemicals were from Fluka, Buchs, Switzerland or Merck,
Darmstadt, Germany with the highest purity available.
Trimethylpyruvate was analysed by HPLC as described
elsewhere [7]. NADH was determined photometrically on
a photometer Shimadzu UV-160 at 340 nm (Shimadzu Europe,
Duisburg, Germany).
O
LeuDH
"COOH NH3 v +
trimethylpyruvate
NADH+H§ NAD+
HCOOH ~ "~v
FDH
Fig. 1. Reaction scheme
The enzyme membrane reactor consisted of a loop contain-
ing a circulation pump Watson Marlow 501 U, Rotterdam, The
Netherlands, heat exchanger and an ultrafiltration unit, Filtron
Minisette with a membrane having a cut off of 10000 Da,
Filtron, Karlstein, Germany. A dosing pump Pharmacia P 500
pump, Pharmacia LKB GmbH, Freiburg, Germany and a sterile
filter, Sartorius filter holder with micro filtration membrane
0.1 gm cellulose nitrate type 11358-47, Sartorius AG, G6ttin-
gen, Germany were used. A more detailed description can be
found in [8].
Modelling of reaction and single enzyme
membrane reactor
Considering detailed kinetic studies of initial reaction
rates with the enzymes leucine dehydrogenase and formate
dehydrogenase a mathematical model as presented in
Fig. 2 was developed. The kinetic measurements of initial
reaction rates were carried out under conditions relevant
to the process. In most respects, the individual model
equations were known from earlier studies [11. For the
enzyme preparations used for this study slightly different
kinetic parameters were obtained. An excess of ammonia and
formate was used, which increases the leucine dehydrogenase
activity. LeuDH is slightly inhibited by the non-natural
substrate trimethylpyruvate. This substrate acts also as a non-
competitive inhibitor for the formate dehydrogenase. In
comparison to leucine dehydrogenase, the formate dehydro-
genase shows lower activity and higher affinity towards the
substrates. The product L-tert-leucine shows a very strong
product inhibition on LeuDH.
Calculations based on the reactor model combining the mass
equations show that the substrate concentration influences the
overall reaction rate. Due to the very strong product inhibition
above 80% conversion, the reaction rate drops abruptly
(Fig. 3). The reaction rate can be increased below the 80% con-
version (Fig. 4) with increasing coenzyme concentration. So it
seems not reasonable to use a CSTR for the continuous
production of L-tert-leucine. Yet it could be shown in an
experiment that was carried out for two months in a single
enzyme membrane reactor that this simple approach is
suitable for production purposes. Nevertheless, to reduce the
influence of the product inhibitions, a plug flow behaviour
, ~ H2
H20
OOH
L-tert-leucine
CO2