An optimal feeding strategy with a l: l. 2 molar ratio of benzaldehyde: pyruvate has been developed to maximize the enzymatic production of the pharmaceutical intermediate R-phenylacetylcarbinol (PAC) based on a previously published model for this intermediate. The results of the simulation indicate with pyruvate decarboxylase (PDC) at an initial carboligase activity of 4 U ml(-1) that up to 730 mM PAC would be produced in 81 h. However, the experimental results show an appreciably lower maximum level of PAC (viz 300 mM) produced after only 54 h, although enzyme activity was maintained at similar or higher values than in the simulation results. It is possible that the increasing PAC concentrations and associated by-products (acetoin and acetaldehyde) have resulted in significant inhibition of PDC during the course of the biotransformation and future model development will need to include one or more product inhibition terms in its structure.

ABSTRACT Biotransformation plays an increasingly important role in the industrial production of fine chemicals due to its high product specificity and low energy requirement. One challenge in biotransformation is the toxicity of substrates and/or products to biocatalytic microorganisms and enzymes. Biofilms are known for their enhanced tolerance of hostile environments compared to planktonic free-living cells. Zymomonas mobilis was used in this study as a model organism to examine the potential of surface-associated biofilms for biotransformation of chemicals into value-added products. Z. mobilis formed a biofilm with a complex three-dimensional architecture comprised of microcolonies with an average thickness of 20 μm, interspersed with water channels. Microscopic analysis and metabolic activity studies revealed that Z. mobilis biofilm cells were more tolerant to the toxic substrate benzaldehyde than planktonic cells were. When exposed to 50 mM benzaldehyde for 1 h, biofilm cells exhibited an average of 45% residual metabolic activity, while planktonic cells were completely inactivated. Three hours of exposure to 30 mM benzaldehyde resulted in sixfold-higher residual metabolic activity in biofilm cells than in planktonic cells. Cells inactivated by benzaldehyde were evenly distributed throughout the biofilm, indicating that the resistance mechanism was different from mass transfer limitation. We also found that enhanced tolerance to benzaldehyde was not due to the conversion of benzaldehyde into less toxic compounds. In the presence of glucose, Z. mobilis biofilms in continuous cultures transformed 10 mM benzaldehyde into benzyl alcohol at a steady rate of 8.11 g (g dry weight)−1 day−1 with a 90% molar yield over a 45-h production period.

Bacteria, yeasts and filamentous fungi were screened for enantio-specific reduction of the α, β-unsaturated carbon bond in citral to produce citronellal. While a traditional aqueous screening system revealed only Zymomonas mobilis as positive, citronellal was produced in an aqueous/organic two liquid phase system by 11 of the 46 tested strains, which demonstrates the relevance of applying two-phase systems to screening strategies. Z. mobilis and Citrobacter freundii formed 1mM citronellal in 3h in the presence of a NADPH regenerating system and 20% (v/v) toluene. In comparison to these bacteria, the eukaryotic strains showed at least five-fold lower citral reductase activities. The bacterial strains produced preferentially the (S)-enantiomer of citronellal with e.e. values of >99% for Z. mobilis and 75% for Citrobacter freundii. In contrast the yeasts produced preferentially (R)-citronellal, i.e. Candida rugosa with an enantiomeric excess value of more than 98%. Many strains formed alcoholic by-products, viz. geraniol, nerol and citronellol. For Z. mobilis the production of these alcohols was suppressed in the presence of various organic solvents, e.g. toluene, and further decreased after EDTA addition.

A mathematical model for the enzymatic biotransformation of benzaldehyde and pyruvate to R-phenylacetylcarbinol (PAC) and its associated by-products has been developed using a schematic method devised by King and Altman [E.L. King, C. Altman, A schematic method of deriving the rate laws for enzyme catalysed reactions, J. Phys. Chem. 60 (1956) 1375–1378] for deriving the rate equations for a complex enzyme-catalysed reaction. PAC is the commercial intermediate for the production of ephedrine and pseudoephedrine. A combinatorial theorem was applied using Visual Basic to create all of the possible reaction patterns for a simplified form of the pyruvate decarboxylase (PDC) biotransformation mechanism. The rate equations for substrates, product, and by-products have been derived from the patterns for yeast PDC and combined with a deactivation model for PDC from Candida utilis. The batch biotransformation profile generated by the model validated previously for a data set at initial substrate concentrations 50–150mM benzaldehyde and 60–180mM pyruvate, provided an acceptable fit for published data at initial concentrations of 400mM benzaldehyde and 600mM pyruvate.

An octanol/aqueous two-phase process for the enzymatic production of (R)-phenylacetylcarbinol (PAC) has been investigated further with regard to optimal pH control and replacement of 2.5 M MOPS buffer by a low cost solute. The specific rate of PAC production in the 2.5 M MOPS system controlled at pH 7 was 0.60 mg U-1 h(-1) (reaction completed at 34 h), a 1.6 times improvement over the same 2.5 M MOPS system without pH control (0.39 mg U-1 h(-1) at 49 h). An improved stability of PDC was evident at the end of biotransformation for the pH-controlled system with 84% residual carboligase activity, while 23% of enzyme activity remained in the absence of pH control. Lowering the MOPS concentration to 20 mM resulted in a lower benzaldehyde concentration in the aqueous phase with a major increase in the formation of by-product acetoin and three times decreased PAC production (0.21 mg U-1 h(-1)). Biotransformation with 20 mM MOPS and 2.5 M DPG as inexpensive replacement of high MOPS concentrations provided similar aqueous phase benzaldehyde concentrations compared to 2.5 M MOPS and resulted in a comparable PAC concentration (92.1 g L-1 in the total reaction volume in 47 h) with modest formation of acetoin.

Pyruvate decarboxylase (PDC) catalyses the synthesis of asymmetric carbinols, e.g., chiral precursors for pharmaceuticals such as ephedrine and pseudoephedrine. The production of PDC by Candida utilis in a minimal medium was improved by manipulating the pH during fermentation in a 5 L bioreactor. At an aeration rate of 0.1 vvm with a stirrer speed of 300 rpm at constant pH 6, a specific PDC activity of 141 U/g dry cell weight (DCW) was achieved (average of two fermentations +/-13%). By allowing the yeast to acidify the growth medium from pH 6 to 2.9, the final specific PDC activity increased by a factor of 2.7 to 385 U/g DCW (average from 4 fermentations +/-16%). The effect of this pH drift on PDC production was confirmed by another experiment with a manual shift of pH from 6 to 3 by addition of 5 M sulfuric acid. The final PDC activity was 392 U/g DCW (average from two fermentations +/-5%). However, experiments with constant pH of 6, 5, 4, or 3 resulted in average specific activities of only 102 to 141 U/g DCW, suggesting that a transitional pH change rather than the absolute pH value was responsible for the increased specific PDC activity.

Zymomonas mobilis pyruvate decarboxylase (PDC) transformed acetaldehyde and benzaldehyde into (R)-phenylacetylcarbinol (PAC), the precursor for the synthesis of ephedrine and pseudoephedrine. Organic solvents were screened for a biphasic biotransformation with the enzyme in an aqueous phase and the toxic substrates delivered through the organic phase. In the absence of substrates a second phase of 1-pentanol, hexadecane or MTBE (methyl tertiary-butyl ether) stabilized the PDC activity in comparison to a control without added solvent. Organic phase solvents for optimal PAC production had partitioning coefficient (log P) values between 0.8 and 2.8 (production of more than 8 mg PAC/ U PDC), however there was no correlation between enzyme stability and log P. Best PAC formation was observed with the eight tested alcohols, which in contrast to the other solvents allowed lower initial concentrations of toxic acetaldehyde (54-81 mM) in the aqueous phase. 1-pentanol, 1-hexanol, and isobutanol resulted in the highest specific PAC production of 11 mg PAC /U PDC. Without the addition of an organic phase, only 1.2 mg/U was formed.

Biotransformation of benzaldehyde and pyruvate into (R)-phenylacetylcarbinol (PAC) catalysed by Candida utilis pyruvate decarboxylase (PDC) at low buffer concentration (20 mM MOPS) was enhanced by maintenance of neutral pH through acetic acid addition. PDC was very stable in this buffer (half-life 138 h at 6 degrees C), however a benzaldehyde emulsion (400 mM) caused rapid deactivation. The inclusion of 2M glycerol did not protect PDC from inactivation by benzaldehyde but initial rates were increased by 50% and the final PAC level was enhanced from 40 to 51 g l(-1). Low levels of by-products acetaldehyde (0.1-0.15 g l(-1)) and acetoin (1.1-1.3 g l(-1)) were formed in both the presence and absence of 2 M glycerol. Interestingly PDC was more stable towards benzaldehyde when pyruvate was present: no activity was lost during the first hour of biotransformation (2 M glycerol, benzaldehyde concentration decreased from 400 to 345 mM, pyruvate from 480 to 420 mM) but PDC was completely inactivated in less than 30 min when exposed to the same concentrations of benzaldehyde in the absence of pyruvate. Thus the enzyme in catalytic action was more stable than the resting enzyme.

Initial rate and biotransformation studies were applied to refine and validate a mathematical model for enzymatic (R)-phenylacetylcarbinol (PAC) production from pyruvate and benzaldehyde using Candida utilis pyruvate decarboxylase (PDC). The rate of PAC formation was directly proportional to the enzyme activity level up to 5.0 U ml−1 carboligase. Michaelis–Menten kinetics were determined for the effect of pyruvate concentration on the reaction rate. The effect of benzaldehyde followed the sigmoidal shape of the Monod–Wyman–Changeux (MWC) model. The biotransformation model, which also included a term for PDC inactivation by benzaldehyde, was used to determine the overall rate constants for the formation of PAC, acetaldehyde, and acetoin. These values were determined from data for three batch biotransformations performed over a range of initial concentrations (viz. 50–150 mM benzaldehyde, 60–180 mM pyruvate, 1.1–3.4 U ml−1 enzyme activity). The finalised model was then used to predict a batch biotransformation profile at 120/100 mM initial pyruvate/benzaldehyde (initial enzyme activity 3.0 U ml−1). The simulated kinetics gave acceptable fitting (R2=0.9963) to the time courses of these latter experimental data for substrates pyruvate and benzaldehyde, product PAC, by-products acetaldehyde and acetoin, as well as enzyme activity level.

Based on experimental data, a kinetic model for the deactivation of partially purified pyruvate decarboxylase (PDC) by benzaldehyde (0-200 mM) in MOPS buffer (2.5 M) has been developed. An initial lag period prior to deactivation was found to occur. With first order dependencies of PDC deactivation on exposure time and on benzaldehyde concentration, a reaction time deactivation constant of 2.64x10 -3 h(-1) and a benzaldehyde deactivation coefficient of 1.98x10 -4 mM(-1) h(-1) were determined for benzaldehyde concentrations up to 200 mM. The PDC deactivation kinetic equations established in this study are an essential component in an overall model being developed to describe the enzymatic biotransformation of benzaldehyde and pyruvate to produce the pharmaceutical intermediate ( R )-phenylacetylcarbinol ( R -PAC).

Recent progress in enzymatic (R)-phenylacetylcarbinol (PAC) production has established the need for low cost and efficient biocatalyst preparation. Pyruvate decarboxylase (PDC) added in the form of Candida utilis cells showed higher stability towards benzaldehyde and temperature in comparison with partially purified preparations. In the presence of 50 mM benzaldehyde and at 4 degrees C, a half-life of 228 h was estimated for PDC added as C. utilis cells, in comparison with 24 h for the partially purified preparation. Increasing the temperature from 4 to 21 degrees C for PAC production with C. utilis cells resulted in similar final PAC levels of 39 and 43 g l(-1) (258 and 289 mM), respectively, from initial 300 mM benzaldehyde and 364 mM pyruvate. The overall volumetric productivity was enhanced 2.8-fold, which reflected the 60% shorter reaction time at the higher temperature. Enantiomeric excess values of 98 and 94% for R-PAC were obtained at 4 and 21 degrees C, respectively, and benzyl alcohol (a potential by-product from benzaldehyde) was not formed.

Enzymatic biotransformations offer the potential for improved concentrations, yields and productivities in the production of chiral precursors for the pharmaceutical industry. The present review describes new developments for the production of R-phenylacetylcarbinol (R-PAC) from substrates pyruvate and benzaldehyde using an enzymatic process, and compares the results with those for traditional fermentative yeast biotransformations. With batch processes, using pyruvate decarboxylase (PDC) from yeast and filamentous fungi, 50g/l R-PAC was achieved in benzaldehyde emulsions, while concentrations in excess of 100g/l were determined in the organic phase for a two-phase octanol/aqueous process. Higher yields and productivities were found also for the enzymatic biotransformations.

Extracts of 14 filamentous fungi were examined regarding their potential for production of (R)-phenylacetylcarbinol [(R)-PAC], which is the chiral precursor in the manufacture of the pharmaceuticals ephedrine and pseudoephedrine. Benzaldehyde and pyruvate were transformed at a scale of 1.2 ml into PAC by cell-free extracts of all selected strains, covering the broad taxonomic spectrum of Ascomycota, Zygomycota and Basidiomycota. Highest final PAC concentrations were obtained with the extracts of Rhizopus javanicus and Fusarium sp. [78-84 mM (11.7-12.6 g/l) PAC within 20 h from initial substrate concentrations of 100 mM benzaldehyde and 150 mM pyruvate]. (R)-PAC was in about 90-93% enantiomeric excess. Rhizopus javanicus had the advantage of faster growth than Fusarium sp. Rhizopus javanicus mycelia were used as an example in a biotransformation process based on whole cells and benzaldehyde and glucose as substrates. The substrate pyruvate was generated through the fungal fermentation of glucose. Only 19 mM PAC (2.9 g/l) were produced within 8 h from 80 mM benzaldehyde. with evidence of significant benzyl alcohol production.

2-Oxo-1,2-dihydroquinoline 8-monooxygenase, an enzyme involved in quinoline degradation by Pseudomonas putida 86, had been identified as a class IB two-component nonheme iron oxygenase based on its biochemical and biophysical properties (B. Rosche, B. Tshisuaka, S. Fetzner, and F. Lingens, J. Biol. Chem. 270:17836-17842, 1995). The genes oxoR and oxoO, encoding the reductase and the oxygenase components of the enzyme, were sequenced and analyzed. oxoR was localized approximately 15 kb downstream of oxoO. Expression of both genes was detected in a recombinant Pseudomonas strain. In the deduced amino acid sequence of the NADH:(acceptor) reductase component (OxoR, 342 amino acids), putative binding sites for a chloroplast-type [2Fe-2S] center, for flavin adenine dinucleotide, and for NAD were identified. The arrangement of these cofactor binding sites is conserved in all known class IB reductases. A dendrogram of reductases confirmed the similarity of OxoR to other class IB reductases. The oxygenase component (OxoO, 446 amino acids) harbors the conserved amino acid motifs proposed to bind the Rieske-type [2Fe-2S] cluster and the mononuclear iron. In contrast to known class IB oxygenase components, which are composed of differing subunits, OxoO is a homomultimer, which is typical for class IA oxygenases. Sequence comparison of oxygenases indeed revealed that OxoO is more related to class IA than to class IB oxygenases. Thus, 2-oxo-1,2-dihydroquinoline 8-monooxygenase consists of a class IB-like reductase and a class IA-like oxygenase. These results support the hypothesis that multicomponent enzymes may be composed of modular elements having different phylogenetic origins.

The 2-oxo-1,2-dihydroquinoline 8-monooxygenase from Pseudomonas putida 86 comprises two components with four redox active sites necessary for activity. We present an EPR characterization of the iron-sulfur centres in the purified reductase and oxygenase component of this novel enzyme system. The oxygenase component was identified as a Rieske [2Fe2S] protein on the basis of its characteristic EPR spectrum with gz,y,x = 2.01, 1.91, 1.76 and gav = 1.893. The reductase component, an iron-sulfur flavoprotein, contained a [2Fe2S] cluster with gz,y,x = 2.03, 1.94, 1.89 and the average g-value (gav) of 1.953, typical of a ferredoxin-type centre. In redox titrations at pH 7, the midpoint potentials were determined to be −180 mV ± 30 mV and −100 mV ± 10 mV for the reductase and oxygenase component, respectively. A detailed comparison to other multicomponent enzyme systems is presented pointing out the EPR and redox properties of the Fes centres involved.

2-Oxo-1,2-dihydroquinoline 8-monooxygenase, which catalyzes the NADH-dependent oxygenation of 2-oxo-1,2-dihydroquinoline to 8-hydroxy-2-oxo-1,2-dihydroquinoline, is the second enzyme in the quinoline degradation pathway of Pseudomonas putida 86. This enzyme system consists of two inducible protein components, which were purified, characterized, and identified as reductase and oxygenase. The yellow reductase is a monomeric iron-sulfur flavoprotein (M(r), 38,000), containing flavin adenine dinucleotide and plant-type ferredoxin [2Fe-2S]. It transferred electrons from NADH to the oxygenase or to some artificial electron accepters. The red-brown oxygenase (M(r), 330,000) consists of six identical subunits (M(r), 55,000) and was identified as an iron-sulfur protein, possessing about six Rieske-type [2Fe-2S] clusters and additional iron. It was reduced by NADH plus catalytic amounts of reductase. For monooxygenase activity, reductase, oxygenase, NADH, molecular oxygen, and substrate were required. The activity was considerably enhanced by the addition of polyethylene glycol and Fe2+. 2-Oxo-1,2-dihydroquinoline 8-monooxygenase revealed a high substrate specificity toward 2-oxo-1,2-dihydroquinoline, since none of 25 other tested compounds was converted. Based on its physical, chemical, and catalytic properties, we presume 2-oxo-1,2-dihydroquinoline 8-monooxygenase to belong to the class IB multicomponent non-heme iron oxygenases.