October 2, 2007

The Genus Gluconobacter Oxydans: Comprehensive Overview of Biochemistry and Biotechnological Applications

By De Muynck, Cassandra Pereira, Catarina S S; Naessens, Myriam; Parmentier, Sofie; Et al

ABSTRACT The genus Gluconobacter comprises some of the most frequently used microorganisms when it comes to biotechnological applications. Not only has it been involved in "historical" production processes, such as vinegar production, but in the last decades many bioconversion routes for special and rare sugars involving Gluconobacter have been developed. Among the most recent are the biotransformations involved in the production of L-ribose and miglitol, both very promising pharmaceutical lead molecules. Most of these processes make use of Gluconobacter 's membrane-bound polyol dehydrogenases. However, recently other enzymes have also caught the eye of industrial biotechnology. Among them are dextran dextrinase, capable of transglucosylating substrate molecules, and intracellular NAD-dependent polyol dehydrogenases, of interest for co-enzyme regeneration. As such, Gluconobacter is an important industrial microbial strain, but it also finds use in other fields of biotechnology, such as biosensor-technology. This review aims to give an overview of the myriad of applications for Gluconobacter, with a special focus on some recent developments. KEYWORDS polyol dehydrogenase, biosensors, dextran, L-ribulose, co-enzyme regeneration, bacterial cellulose, L-sorbose

Abbreviation: 2KGADH, 2-keto-D-gluconate dehydrogenase; 2KGR, 2- keto-D-gluconate reductase; 5KGR, 5-keto-D-gluconate reductase; 2KLG, 2-keto-L-gulonic acid; ArDH, D-arabitol dehydrogenase; Cyt.c, c-type cytochrome; Cyt.O, o-type cytochrome; DDase, dextran dextrinase; DHA dihydroxyacetone; FAD, flavine adenine dinucleotide; FDH, D-fructose dehydrogenase; G4, maltotetraose; G4H, maltotetraitol; GADH, D-gluconate dehydrogenase; GDH, D-glucose dehydrogenase; GHK, D-glucose hexokinase; GIc, D-glucose; GIyDG, glycerol dehydrogenase; HIV, human immunodeficiency virus; MDH, D- mannitol dehydrogenase; mGlyDH, membrane fractions containing GlyDH; MPDH, major polyol dehydrogenase; NAD^sup +^, nicotinamide adenine dinucleotide; NADH, reduced nicotinamide adenine dinucleotide; NADP^sup +^, nicotine adenine dinucleotide phosphate; NADPH, reduced nicotine adenine dinucleotide phosphate; NCBI, National Center for Biotechnology Information; ORF, open reading frame; pGlyDH, purified GlyDG; PQQ, pyrroloquinoline quinine; RLDG, ribitol dehydrogenase; SKDH, shikimate dehydrogenase; SLDH, D-sorbitol dehydrogenase; TCA tricarboxylic acid cycle; UDP, uridine diphosphate; Q^sub 10^, ubiquinone-10.


Gluconobacter oxydans is one of the most frequently used microorganisms in industrial biotechnology. Its unique capacity to incompletely oxidize polyol substrates has led to numerous production processes for the synthesis of compounds such as vitamin C, (keto)gluconic acids, dihydroxyacetone and vinegar. Nowadays, these processes are being further optimized whilst new processes for the synthesis of compounds such as L-ribulose, D-tagatose, miglitol and chiral aldehydes and acids are being developed. Because of their unique properties, Gluconobacter oxydans' polyol dehydrogenases also find application in biosensor technology and co-enzyme regeneration. Besides the many applications of its polyol dehydrogenases, Gluconobacter oxydans is also of interest for the production of bio- polymers (mainly cellulose and dextran) and for the production of transglucosylating enzymes (dextran dextrinase). The very recent sequencing of the Gluconobacter oxydans genome has had a major impact on the understanding of its metabolism and respiratory chain. With these new insights, the development of new and renewed applications of this strain in industrial biotechnology and biosensor technology will doubtlessly follow.


The genus Gluconobacter belongs to the family of the Acetobacteraceae and in the 1980s (De Ley and Swings, 1984) the many species belonging to the Gluconobacter genus were reduced to one single species, G. oxydans, subdivided into four subspecies: subsp. oxydans, suboxydans, industrius and melanogenus. However, in the last decade it became clear that they must be considered as separate biological entities. Consequendy, following the latest version of Bergey's Manual of Systematic Bacteriology, four different species belong to the genus Gluconobacter, namely G. oxydans, G. asaii, G. cerinus and G. frateurii (Sievers and Swings, 2005) and according to Taxonomy Browser NCBI (http://www.ncbi.nlm.nih.gov/, 11/12/2006), there are seven: G. oxydans, G. cerinus and G. frateurii and G. albidus, G. krungthepensis, G. thailandicus and Gluconobacter sp. NBRC 3243. Furthermore, Acetobacter suboxydans, Acetobacter oxydans and G. suboxydans were classified as G. oxydans. However, recently a phylogenetic tree constructed from 16S rRNA sequence analysis indicated the presence of five clusters, corresponding, respectively, to the major five species of the genus Gluconobacter, namely G. albidus, G. cerinus, G. frateurii, G. oxydans (type species), and G. thailandicus (Takahashi et al., 2006). The type strain of G. asaii was included in the G. cerinus cluster and several other re-classifications are proposed. These results suggest that the phenotypic differences among Gluconobacter species are ambiguous and the species definition must be re-evaluated. In this review, the original names mentioned in the cited papers are used. G. oxydans gets its species name from oxys, Latin for "sharp, acidic," and dans, "giving" (Sievers and Swings, 2005).

G. oxydans is a Gram negative ellipsoidal- to rodshaped bacterium, which occurs as a single cell and/or in pairs, rarely in chains. Enlarged irregular cell forms may occur, so-called involution forms. These involution forms can differ from long filamentous cells to large swollen sac formed cells and other aberrant cell forms. These bacteria can be motile or nonmotile; if motile, the cells possess 3 to 8 flagella. Endospores are not formed. The species is chemoorganotrophic and an obligatory aerobe, having a stricdy respiratory type of metabolism with oxygen as the terminal electron acceptor (De Ley and Swings, 1984; Sievers and Swings, 2005).

In nature, this bacterial species can mainly be found in sugary niches such as flowers and fruits. Therefore, it can be isolated from garden soil, fruit, sake, cider, beer and wine, where it causes spoilage of these products. The species is responsible for pink disease in pineapple, rot on apples and pears, and it is found in massive numbers on honeybees. Interestingly, a G. oxydans strain isolated from patulin-contaminated apples was capable of degrading patulin to a less-toxic compound, ascladiol (Ricelli et al., 2007). Gluconobacter is not known to have any pathogenic effect towards humans or animals (De Ley and Swings, 1984).

II.1 Metabolism of Gluconobacter Oxydans

The pentose phosphate pathway constitutes the most important catabolic route for the breakdown of sugars and polyols to carbon dioxide in Gluconobacter (De Ley and Swings, 1984). Glycolysis is absent in Acetobacteraceae due to lack of phosphofructokinase (Sievers and Swings, 2005), and Gluconobacter is not able to overoxidize acetic acid to CO2 and H2O due to an incomplete tricarboxylic acid cycle, lacking succinate dehydrogenase (Sievers and Swings, 2005). Furthermore, all genes encoding the enzymes of the pentose phosphate and Entner-Doudoroff pathways have been identified. However, some important enzymes of the latter pathway were not found, and in this respect, the pentose phosphate pathway remains the most plausible route in Gluconobacter.

Therefore, the Gluconobacter genus has the capability to take up and channel several polyols, sugars and sugar derivates into the pentose phosphate pathway. In most cases the substrates are first phosphorylated by specific kinases and are further degraded by dehydrogenases and isomerases. One key question is how the NAD(P)H produced in the dehydrogenation reactions is reoxidized. Genomic data point to a potential answer: a transhydrogenase enzyme linking NAD(P)H to proton translocation may be responsible for maintaining this redox balance (Prust et al., 2005). The structure, function and role of this critical linking enzyme merit considerable further study. Besides this padiway for sugar and alcohol oxidation, Gluconobacter is capable of incompletely oxidizing these compounds with membrane-bound dehydrogenases and the products (aldehydes, ketones and acids) are excreted into the medium (Deppenmeier etal, 2002). This makes the strain very interesting for biotechnological applications.

Within the genome of G. oxydons, no open reading frame (ORF) could be assigned that potentially encodes phosphoenolpyruvate synthase or other phosphoenol pyruvate-synthesizing enzymes, indicating that this species cannot produce Ce-sugars via gluconeogenesis. Therefore, the production of glucose, for example, from pentoses or glycerol must depend on the pentose phosphate pathway (Prust etal., 2005).

Gluconobacter possesses an incomplete set of tricarboxylic acid cycle (TCA) enzymes (De Ley and Swings, 1984); in particular, the lack of succinate dehydrogenase prevents the full operation of this metabolic patiiway (Greenfield and Claus, 1972), which is confirmed by the absence of the genes encoding for this enzyme in the G. oxydons genome. This fact can be used to distinguish Gluconobacter strains from those belonging to the genus Acetobacter, which possess a complete set of TCA enzymes that function in terminal oxidation (Asai, 1968). Consequently, Acetobacter fully oxidizes acetate to CO2 and H2O whereas Gluconobacter 'can not. Greenfield and Claus (1972) have shown that the purpose of a partial TCA cycle functioning in Gluconobacter is primarily for the biosynthesis of glutamate, aspartate, and succinate. The existence of a complete set of glycolytic enzymes in Gluconobacter is as yet not completely elucidated (Macauley et al., 2001). Furthermore, the presence of a total set of these enzymes does not necessarily mean that glycolysis occurs. This is because all of the glycolytic enzymes, except phosphofructokinase, can also be used in either the pentose phosphate pathway or the Entner-Doudoroff pathway (Asai, 1968).

Furthermore, genomic analyses showed that G. oxydons contains metabolic pathways for the de novo synthesis of all nucleotides, amino acids, phospholipids and most vitamins. Surprisingly, the metabolic pathways and regulatory mechanisms of Gluconobacter are not yet fully elucidated, despite its industrial relevance for several decades.

II.1.1 Respiratory Chain of Gluconobacter Oxydans

The composition of the electron transport system in G. oxydans is still a matter of debate. It has been stated diat the respiratory chain contains ubiquinone-10 (Qio). tw[degrees] types off-type cytochromes (Cyt. c), and two types of o-type cytochromes (Cyt. 0), being involved in this order in an NADH oxidase system, but cytochrome d or a do not occur. Moreover, evidence has been presented recently that there is a cytochrome c oxidase of the aa^sub 3^-type present (Deppenmeier et al., 2002).

The peculiar properties of the respiratory chain of G. oxydans (Matsushita et al., 1987, 1989) suggest diat it is branched at the ubiquinone level, with cyanidesensitive and -insensitive terminal oxidases, of which the cyanide-sensitive one is cytochrome 0 (Matsushita et al., 1994). Takeda and Shimizu (1991) suggest that the cyanide-insensitive terminal oxidase may contain some cytochrome c activity. However, according to Matsushita et al. (1994), G. oxydans lacks cytochrome c as the terminal oxidase system; this controversy points to an as yet unknown enzyme involved in the cyanideinsensitive terminal oxidase system.

The sequencing of the G. oxydans 621H genome in 2005 (Prust et al., 2005) helped to clarify the electron transport system of the species. Two operons were found that encode quinol oxidases of b0^sub 3^-type and of bdtype; the latter might represent the cyanide- insensitive terminal oxidase mentioned above. Genes coding for a cytochrome c oxidase are absent from the genome of G. oxydans 62 IH. Nevertheless, genes encoding a ubiquinohcytochrome c oxidoredutase {bccomplex) were identified. However, the function of the corresponding protein complex remains unclear due to the fact diat reduced cytochrome c can not be reoxidized, due to the absence of cytochrome c oxidase (Prust etal., 2005).

Hence, the core system is rather simple, consisting of a non- proton-translocating NADH: ubiquinone oxidoreductase and two quinol oxidases of b0^sub 3^-type and of M-type. The organism lacks a proton-translocation NADH: ubiquinone oxidoreductase and cytochrome c oxidase. Therefore, its ability to translocate protons in the course of redox reactions is rather limited (Prust et al., 2005).

In addition to the typical respiratory complexes mentioned above, G. oxydans possesses several membrane-bound dehydrogenases that channel electrons into the respiratory chain as well as many intracellular oxidoreductases. It has been shown that membrane- bound dehydrogenases transfer electrons to ubiquinone, which acts as electron donor for the quinol oxidases. The limited proton translocation ability of this simple respiratory chain may serve to reduce the potential inhibition of membrane redox reactions tliat would occur following changes in transmembrane potential, and may thus allow incomplete oxidation to continue in the presence of abundant substrate (McNeil and Harvey, 2005).

Genomic data also revealed the presence of a membrane-bound transhydrogenase, which foresees the necessary protons for an FiFn- type ATP synthase, capable of generating ATP (Prust et al., 2005).

II. 1.2 lntracytoplasmic Membrane Formation

The cells of G. oxydans typically differentiate after exponential growth by forming complex intracytoplasmatic membranes (ICM) (Batzing and Claus, 1973). They are formed at the polar regions of the cell by invagination of the cytoplasmic membrane at the end of active cell division (White and Claus, 1982), with an additional increase in polar ribosomal material and free lipid content (Heefher and Claus, 1976). This cytological difference between the stationary phase and exponentially growing cultures was shown to increase membrane-bound dehydrogenase activity and consequentially the rate of polyol oxidation (White and Claus, 1982). According to Claus (1975), intracytoplasmatic membrane formation may be due to low phosphorylation efficiency, or it may form as a response to environmental stress.

II.2 Genetics of Gluconobacter Oxydans

In the last decade, a large number of studies have been published on the genetics of G. oxydans. In 2005, the complete genome of G. oxydans 62 IH was sequenced (Prust et al., 2005). This is of great importance for the further understanding of the overall Gluconobacter metabolism, particularly to obtain detailed insight into the oxidative potential of the bacteria and to elucidate the mechanisms of incomplete oxidations of industrial importance (Prust et al., 2005).

Gluconobacter oxydans was found to have a small circular chromosome size (2,702 kb) (Verma et al., 1997) with a G+C content of 61%, which can be related to the small range of metabolic activities (Brubaker, 1991). Furthermore, five plasmids were identified in this bacterium (163, 27, 15, 13 and 3 kb). Therefore the total size of the genome is 2,922 kb (Prust et al., 2005). The coding region of the genome is about 89.9% with 2664 identified ORF's of which 1877 have been assigned a function.

The G. oxydans genome contains a large number of repeated DNA elements, which are known to be involved in genomic rearrangements; 82 insertion sequences and 103 transposase genes were identified (Prust et al., 2005). Some of these copies are partially deleted, and therefore, supposed to be defective. Nevertheless, most of the insertion sequences appear to have functional copies and are perhaps responsible for the genetic instability leading to deficiencies in some physiological properties as observed in a variety of acetic acid bacteria, for example the sudden loss of growdi, vigor or bioconversion activity (Kondo and Horinouchi, 1997, 1997). Several stuthes have revealed the presence of plasmids in acede acid bacteria (Fukaya et al., 1985). From the 36 G. oxydans strains tested, 23 contained one or more plasmids. Additionally, G. oxydans ATCC 621 contained two plasmids, with a molecular weight of 3.1 and 4.0 x 10^sub p^ DaIton. However, as mentioned above, the G.oxydans 621H strain of which the genome was sequenced contains five plasmids and none of the sequences show homology to known plasmids of other G. oxydans strains (Tonouchi et al., 2003).

On the basis of the genomic sequence and analysis, it is now possible to describe more extensively the process of incomplete oxidation and the physiology of strains performing this process. Furthermore, research into the genetic diversity of G. oxydans should be intensified, as physiologically similar G. oxydans strains possess genomes of varying sizes (from 2,240 to 3,780 kb) (Verma et al., 1997) and variable composition, which widens the spectrum of possible biotechnological applications of this organism (McNeil and Harvey, 2005).


Acetic acid bacteria, particularly the genus Gluconobacter, are unsurpassed among microorganisms in their capacity to incompletely oxidize a large variety of carbohydrates and related compounds (Deppenmeier et al., 2002). Previous investigations showed diat the responsible enzymes from these bacteria can be classified into two major groups, capable of dehydrogenation via two independent padiways (Arcus and Edson, 1956; Kersters et al., 1965). These two enzyme systems differ in subcellular location, functioning within the cell, and substrate specificity. The first group contains membrane-particle-associated and cytochrome-linked dehydrogenases, with maximal oxidizing capacity at pH 5.5, following a very strict substrate-specificity, referred to as the Bertrand-Hudson rule. The second group is formed by cytoplasmic soluble polyol dehydrogenases with optimal activity at pH 8.0 and with a substrate specificity diat does not conform to a simple rule. It is now clear that the enzymes of the first group are associated with the cytoplasmic membrane, excreting their products into the fermentation medium, whereas the products of the second group, containing at least six different soluble polyol dehydrogenases, with each a unique specificity and NAD(P)+ as co-enzyme, are readily dissimilated in the general metabolic pathways (Kersters etal., 1965; Kulhanek, 1989).

III.1 Membrane-bound Dehydrogenases

In the last decades, a number of membrane-bound polyol dehydrogenases (first group) of acetic acid bacteria have been described, oxidizing D-glucose, ethanol, ribitol, D-sorbitol, L- sorbose, glycerol, D-arabitol and many other polyol-containing molecules (Table 1). This has led to a large number of biotechnological applications (Deppenmeier et al., 2002), some of which will be briefly discussed here. However, it must be pointed out diat many controversies still exist related to the enzyme systems of Gluconobacter species. Matsushita and co-workers (2003) suggest that many of these polyol dehydrogenases may be reduced to one major glycerol dehydrogenase, involved in the oxidation of almost all sugar alcohols with "Bertrand-Hudson configuration" by Gluconobacter sp. Since the essential polyol structure required for these oxidations is glycerol, they suggest that the major polyol dehydrogenase would ideally be called glycerol dehydrogenase (EC (Matsushita et al., 2003). As can be seen from Table 1, many of these membrane-bound polyol dehydrogenases use PQQ_as a prosthetic group to transfer reducing equivalents to the electron transport system. Mutants that are unable to produce PQQcan dierefore no longer grow on polyols like D-mannitol, D-glucose or glycerol as the sole energy source, showing that in G. oxydans (621H) PQQ_ is essential for growth with these substrates (Holscher and Gorisch, 2006).

III. 2 Intracellular, NAD(P)-Dependent Dehydrogenases

In the cytosolic fraction various kinds of NAD(P)dependent dehydrogenases predominate, most of which show the same reaction as the dehydrogenases in the cytoplasmic membranes under different reaction conditions (Adachi et ed., 1999). These NAD(P)dependent enzymes have no evidence of bioenergy production during the growth of acetic acid bacteria (Adachi et al, 2001).

One apparent difference between membrane-bound and cytosolic dehydrogenases is that optimal substrate oxidation by cytosolic NAD(P)-linked dehydrogenases is observed in fairly alkaline pH regions in the range pH 8.0-11.0, while membrane-bound dehydrogenases catalyze the reaction at an optimal acidic pH between pH 3.0-6.0 (Adachi et al., 2001).

Kersters et al. (1965) described different soluble polyol dehydrogenases: NADP-specific xylitol dehydrogenase, NAD-linked ?- erythro dehydrogenase, NAD-linked ?-xylo dehydrogenase, NADP-linked Dlyxo dehydrogenase, and two NADP-linked gluconate dehydrogenases. The specificity of the oxidation capacity by the different enzymes was examined and the oxidation products were identified.

Adachi et al. (1980) crystallized and characterized a NADP- dependent D-glucose dehydrogenase (EC from Gluconobacter suboxydans IFO 12528. The physiological role of this enzyme in vivo is the oxidation of D-glucose with NADP+. The NADPH formed is spontaneously oxidized to NADP+ by the "old yellow enzyme" (OYE and NADPH dehydrogenase) existing predominantly in Gluconobacter sp. The same "old yellow enzyme" was described as being important for the regeneration OfNADP+ for the aldehyde dehydrogenase (EC and glucose-6-phosphate dehydrogenase (Figure 1) (Adachi etal., 1980).

In acetic acid bacteria, different NAD(P)-dependent sugar alcohol dehydrogenases are reported. A comparison is given in Table 2.

Shaw and Bygrave (1966) reported the presence of a NAD-mannitol, a NADP-mannitol and a NADsorbitol dehydrogenase in Acetobacter suboxydans ATCC 621. Shaw and Bygrave (1966) and Adachi et al. (1999) purified an NAD-dependent (EC and NADP-dependent (EC mannitol dehydrogenase from Gluconobacter suboxydans IFO 12528. Both enzymes catalyze the oxidation of D-mannitol to D- fructose and the reduction of D-fructose to Dmannitol with NAD(H) or NADP(H), respectively. Also, a membrane-bound D-mannitol dehydrogenase (EC has been purified from acetic acid bacteria (Oikawa et al., 1997). Interestingly, the complete 3D structure of none of these dehydrogenases has been published, except for the NAD-xylitoI dehydrogenase from Gluconobacter oxydans (Ehrensberger et al., 2006). To understand and alter the cosubstrate specificity of XDH, Ehrensberger et al. determined the crystal structure of the Gluconobacter oxydans holoenzyme to 1.9angstrom resolution. The structure revealed that NAD^sup +^ specificity is largely conferred by Asp38, which interacts with the hydroxyls of the adenosine ribose. Met39 stacked under the purine ring was also located near the 2' hydroxyl group. Based on the location of these residues and on sequence alignments with related enzymes of various cosubstrate specificities, they constructed a double mutant (D38S/ M39R) that was able to exclusively use NADP^sup +^ instead of NAD^sup +^, with no loss of activity.

TABLE 1 Summary of Membrane-Bound Dehydrogenases Purified from Gluconobacter Species

FIGURE 1 Role of the old yellow enzyme in Gluconobacter (substrate oxidation is catalyzed by a NADP-dependent aldehyde dehydrogenase, b NADP-dependent glucose dehydrogenase, c glucose-6- phosphate dehydrogenase, d 6-phosphogluconate dehydrogenase; OYE = old yellow enzyme).

III.3 Glucose Oxidizing Systems

Gluconobacter can oxidize glucose via two alternative pathways: the first includes uptake, intracellular oxidation and dissimilation by oxidation via the pentose phosphate padiway. The second consists of the direct oxidation in the periplasmic space by membranebound dehydrogenases (Matsushita et al., 1994; Olijve and Kok, 1979; Silberbach et al., 2003).

Thus, in the latter metabolic pathway, glucose can be direcdy oxidized by the membrane-bound glucose dehydrogenase to gluconate; and this can be further oxidized to 2-keto-D-gluconate and 2,5- diketo-D-gluconate by the membrane-bound enzymes gluconate dehydrogenases (GADH) and 2-keto-D-gluconate dehydrogenase (2KGADH), respectively (Matsushita et al., 1994). Gluconobacter species are also able to accumulate 5-keto-D-gluconate in the culture medium (Shinagawa et al., 1983).

TABLE 2 Comparison of NAD(P)-Dependent Sugar Alcohol Dehydrogenases from Acetic Acid Bacteria

When taken up in the cytoplasm, glucose can also be converted to gluconate by a cytosolic glucose dehydrogenase and then further dissimilated in the pentose phosphate cycle. Additionally, in the cytoplasm, D-gluconate can be oxidized by 2-keto-D-gluconate reductase (2KGR) and 5-keto-D-gluconate reductase (5KGR). The cytosolic 2KGR can also reduce 2-ketoD-gluconate back to gluconate.

FIGURE 2 Pathways of glucose oxidation in G. oxydans; GDH; glucose dehydrogenase; GADH; gluconate dehydrogenase; 2KGADH; 2- keto-D-gluconate dehydrogenase; 2KGR; 2-keto-D-gluconate reductase; 5KGR; 5-keto-D-gluconate reductase; MPDH; major polyol dehydrogenase.

However, gluconate production by G. oxydans has been shown to be mainly due to the activity of the membrane bound GDH, as its activity is 30-fold higher tiian diat of the cytosolic GDH (Pronk et al., 1989). The various known padiways of glucose metabolism in Gluconobacter oxydans are summarized in Figure 2.

III.3.1 Formation of D-gluconic Acid

The membrane-bound GDH catalyzes the direct oxidation of D- glucose to D-gluconate at the surface of the cytoplasmic membrane and the pH optimum is around 6.0 (Matsushita et al., 1994). The enzyme binds noncovalendy and tighdy to the prosthetic group pyrroloquinoline quinone (PQQ) (Ameyama et al., 1981) and the electron acceptor in the membrane is ubiquinone (Matsushita et al., 1994). Gluconic acid and its salts are used for the removal, decomposition, or prevention of mineral deposits and the prevention of cloudiness and scaling by calcium compounds in beverages. Applications in the textile and tanning sector and in the pharmaceutical, fodder and concrete industries have also been reported (Hommel and Ahnert, 2000). Therefore, D-gluconic acid is one of the top-ten organic chemicals produced from sugar with an annual production of around 100,000 tons per annum (Lichtenthaler, 2006).

The microbial oxidation of glucose to D-gluconic acid by G. oxydans is subject to substrate and product inhibition (Velizarov and Beschkov, 1998; Velizarov etal., 1997). Substrate (D-glucose) concentrations higher than 0.5 M (S^sub crit^) caused an extended lag phase in batch fermentations, although the final cell density remained unchanged. High substrate concentrations could most probably be tolerated due to the development of cell adaptation mechanisms during the lag and early exponential phase. Product inhibition was found to occur when D-gluconic acid concentrations reached a level higher than 0.7 M (P^sub crit^). This somewhat low toxic effect is probably due to D-gluconic acid being a weak acid (pKa = 3.62) and being further metabolized by G. oxydans to ketogluconic acids (Velizarov and Beschkov, 1998).

III.3.2 Production of 2,5-Diketogluconic Acid

The conversion of D-glucose into 2,5-diketogluconic acid by G. oxydans is mediated by three membrane-bound NADP-independent dehydrogenases (glucose dehydrogenase, gluconate dehydrogenase and 2- ketogluconate dehydrogenase) (Gupta et al., 2001) (Figure 2). The 2,5-diketogluconate may be converted to 2-keto-L-gulonate by stereospecific reduction using 2,5-diketogluconic acid reductase from Cotynebacterium sp. (Sonoyama etal., 1982). The 2-keto-L- gulonate is the penultimate intermediate in industrial production of ascorbic acid, known as vitamin C. However, Grindley et al. developed a more elegant method for producing 2-keto-L-gulonate directly from D-glucose, by cloning the gene encoding for 2,5- diketogluconate reductase from Corynebacterium sp. into Erminia citreus, which naturally oxidizes D-glucose to 2,5-diketogluconate (Grindley etal., 1988).

III.3.3 Production of 5-Ketogluconic Acid

5-Ketogluconic acid is a precursor of L(+)-tartaric acid. G. oxydans DSM 3503 is capable of converting glucose into 5- ketogluconic acid via gluconic acid. Klasen and co-workers (1995) have purified a NADP-dependent cytosolic 5-keto-D-gluconate reductase (5KGR) from G. oxydans, concluding that this enzyme is responsible for 5-keto-D-gluconate accumulation in the culture medium. However, its optimum alkaline pH is not favorable for 5- keto-D-gluconate production in the oxidative fermentation by acetic acid bacteria, most of which happens under fairly acidic conditions (Matsushita etal., 1994). Moreover, if the 5-ketoD-gluconate production is carried out in the cytoplasm, the strain has to incorporate the substrate, D-gluconate, into the cells and then pump out the product, 5-keto-Dgluconate, under the expense of considerable amounts of bio-energy. According to Shinagawa (1999), such energy wasting looks unfavorable for the proposed mechanism of oxidative fermentation. Recently, Shinagawa et al. (1999) have shown that a membrane suspension from G. suboxydans can produce 5-keto-D- gluconate as well as 2-keto-D-gluconate under acidic conditions in the absence of NADP^sup +^, and also that exogenous addition of PQQ and CaCl^sub 2^ stimulate 5-keto-D-gluconate formation, thus suggesting that the accumulation of 5-keto-D-gluconate involves a membrane-bound quinoprotein. Furthermore, it seems that Gluconobacter has two types of gluconate dehydrogenases: one 2-keto- D-gluconate producing flavoprotein and one quinoprotein glycerol (major) dehydrogenase involved in 5-keto-D-gluconate production. These two enzymes compete with each other to oxidize D-gluconate, thus selective expression of either dehydrogenase can improve the production of either of the keto-D-gluconates (Elfari et al., 2005; Matsushita et al, 2003). In G. oxydans MF1, a mutant strain in which the membrane-bound gluconate-2-dehydrogenase complex was inactivated, formation of the undesired 2-keto-D-gluconate was absent. This mutant strain uniquely accumulates high amounts of 5- keto-D-gluconate in the culture medium (Elfari et al, 2005; Merfort et al., 2006). The production rate of this mutant was further increased by equipping the strain with plasmids allowing the overproduction of the soluble and membrane-bound 5-keto-D-gluconate forming enzymes (Merfort et al., 2006). III.4 Ethanol Oxidizing System

In Gluconobacter species, similar to glucose, ethanol can be oxidized by two different metabolic pathways: one includes the transportation through the cytoplasmatic membrane and oxidation by cytosolic alcohol and aldehyde dehydrogenases with further dissimilation, whereas the other proceeds by the direct oxidation in the periplasmic space by membrane-bound dehydrogenases (Matsushita et al., 1994).

III.4. 1 Production of Vinegar

Vinegar contains at least 4 g of acetic acid per 100 ml and not more than 0.5% of ethanol, together with small amounts of glycerol, esters, sugars and salts. A clear distinction should be made between vinegar, the condiment, and acetic acid, the industrial organic chemical (Swings, 1992).

Vinegar is produced with Acetobacter in a two-step reaction: (1) the oxidation of ethanol to acetaldehyde with alcohol dehydrogenase; (2) the oxidation of acetaldehyde to acetic acid with aldehyde dehydrogenase (Macauley etal, 2001). Gluconobacter has not been used extensively in the vinegar industry, despite its inability to further oxidize acetic acid, since it is slower in ethanol oxidation when compared to Acetobacter (De Ley and Swings, 1984). Nevertheless, the production of gluconic acid by Gluconobacter would be beneficial to the industrial process because it is reported to add flavor to the vinegar (Macauley et al., 2001).

III.5 Sorbitol Oxidizing System

The metabolic pathways of D-sorbitol, L-sorbose and their metabolites in Gluconobacter strains are depicted in Figure 3. The membrane bound NAD(P)-independent D-sorbitol dehydrogenase (1) is believed to be responsible for the efficient production of L- sorbose from D-sorbitol by oxidative fermentation (Park et al., 1994). The particulate L-sorbose dehydrogenase (2) catalyzes the oxidation of L-sorbose to L-sorbosone. The NAD(P)-dependent L- sorbosone dehydrogenase (3) is involved in the oxidation of L- sorbosone to 2-keto-L-gulonic acid (2KLG) (Hoshino et al., 1991). The NADPH-dependent L-sorbosone reductase (4) is able to reduce L- sorbosone back to L-sorbose (Hoshino et al., 1990). Sugisawa et al. (1991) and Adachi et al. (1999) purified an NADPH-linked L-sorbose reductase (5) from the cytosol fraction of Gluconobacter melanogenus N44-1 and Gluconobacter melanogenus IFO 3294, respectively. The enzyme catalyzes the oxidoreduction between D-sorbitol and L- sorbose in the presence of NADP^sup +^ or NADPH. Shinjoh et al. (2002) reported that the NADPH-dependent L-sorbose reductase of L- sorbose producing Gluconobacter suboxydans IFO 3291 contributes to intracellular L-sorbose assimilation. In vivo the enzyme functions mainly as an L-sorbose-reducing enzyme and not as a D-sorbitol- oxidizing enzyme. Adachi et al. (1999) crystallized an NAD- dependent Dsorbitol dehydrogenase (6) from Gluconobacter suboxydans IFO 3257. The physiological role of the enzyme is the oxidation of D- sorbitol to D-fructose in the cytoplasmic fraction.

FIGURE 3 The metabolic pathway of D-sorbitol, L-sorbose and their metabolites in Gluconobacter strains (1: membrane-bound D-sorbitol dehydrogenase, 2: membrane- bound L-sorbose dehydrogenase, 3: NAD(P)- dependent L-sorbosone dehydrogenase, 4: NADPH-dependent L-sorbosone reductase, 5: NADPH-dependent L-sorbose reductase, 6: NAD-dependent D-sorbitol dehydrogenase) (adapted from Shinjoh et al. (2002)).

III.5.1 Membrane-Bound Sorbitol Dehydrogenase and the Production of Vitamin C

At present, commercially manufactured L-ascorbic acid (Vitamin C) is still mainly synthesized via the seven-step Reichstein-Grussner process using D-glucose as a starting material (Hancock and Viola, 2002). The process involves six chemical steps and one fermentation step, for the oxidation of D-sorbitol to L-sorbose.

The latter biochemical step is catalyzed by G. oxydans sorbitol dehydrogenase. Chemical oxidation of D-sorbitol would result in the racemization of L-sorbose and would reduce the yield by half. Notwithstanding the fact that the G. oxydans biotransformation features both substrate and product inhibition (De Wulf et al., 2000; Srivastava and Giridhar, 1998; Srivastava and Lasrado, 1998), modern strains giving exceptionally high conversion yields of almost 100% have been developed (Hancock and Viola, 2002). Hence, this is still the most significant and economically important industrial process involving the use of Gluconobacter. The world production of L-ascorbic acid was estimated at 80,000 tons per annum in 2002 with a global market in excess of US$ 600 million and an annual growth rate of 3-4[degrees]/o (Hancock and Viola, 2002). Moreover, as a result of its production in the vitamin C process, L-sorbose, produced from D-sorbitol by G. oxydans, is the cheapest large-scale accessible L-sugar, with an annual production of 60,000 tons per year and a bulk market price of 7.5 euro per kg (Lichtenthaler, 2006).

During the past 50 years, there have been a lot of attempts to find alternative methods for L-ascorbic acid synthesis that could compete with the ReichsteinGriissner process from an economical point of view. In the last decades, a number of publications appeared describing biotransformations for the synthesis of other Reichstein intermediates (Hancock and Viola, 2002). For example, the bio-oxidation of D-sorbitol (Okazaki et al., 1969) or L-sorbose (Kitamura and Perlman, 1975; Tsukada and Perlman, 1972) to 2-keto-L- gulonic acid (2-KLG), using the combination of several polyol dehydrogenases from G. oxydans, have frequently been investigated. Furthermore, Cerestar/BASF has developed a process in which sorbitol is directly fermented to 2-keto-L-gulonic acid (Buchholz et al., 2005). As mentioned earlier, 2,5-diketogluconate produced by G. oxydans can also be converted to 2-keto-L-gulonate (Sonoyama et al., 1982), by 2,5-diketogluconate reductase from Corynebacterium sp. or by the Grindley-pathway (Grindley et al., 1988). Furthermore, various continuous biocatalytic systems, based on the Reichstein procedure, have been proposed as well as direct biosynthesis of L- ascorbic acid in eukaryotes, such as micro-algae and yeasts (Hancock and Viola, 2002; Sauer etal., 2004).

III.5.2 Membrane-Bound Sorbitol Dehydrogenase and the Production of D-Tagatose

D-Tagatose is a low-calorie bulk sweetener, with 92% of the sweetness of sucrose but less than half the calories (Levin etal., 1995). In order to produce D-tagatose from cheaper resources, attempts have been made to produce it via bioconversion. The main focus lies on isomerization of D-galactose to D-tagatose with L- arabinose isomerase (Cheetham and Wootton, 1993; Kim etal, 2002; Oh et al, 2006; Roh et al., 2000) and on oxidation of galactitol to D- tagatose (Huwig et al, 1997; Manzoni et al., 2001; Muniruzzaman et al, 1994; Rollini and Manzoni, 2005). For the latter, Rollini and Manzoni (2005) succeeded in adapting the sorbitol dehydrogenase of G. oxydans for the dehydrogenation of galactitol. This enzyme activity was a minor activity of SDH and could be increased by adapting the strain repeatedly to a glycerol + galactitol medium. Glycerol is needed during the induction phase, probably because glycerol produces a higher and more stable intracytoplasmic membrane content. The non-adapted cells initially showed a higher affinity for sorbitol (37 mM) than for galactitol (98 mM). The galactitol adaptation procedure determined a decrease in sorbitol affinity (57 mM) and an increase in that related to galactitol (21 mM). With these modifications, 4.4 g L^sup -1^ of D-tagatose could be produced starting from 22.5 g L^sup -1^ galactitol.

FIGURE 4 Coenzyme recycling by the coupled-substrate and the coupled-enzyme method (E: enzyme).

III. 5.3 Intracellular Sorbitol Dehydrogenase and Co-Enzyme Regeneration Systems

In many industrial fields, the need for safer and purer products leads to the use of ever more selective production processes. Due to their ability to perform reactions in a stereo- and regiospecific manner, whilst functioning under mild reaction conditions, enzymes are increasingly used for chiral synthesis. Oxidoreductases catalyze many of the reactions that are difficult to perform by conventional chemistry and are therefore of main interest for biotransformations. The major drawback of these enzymes lies in the fact diat they usually need co-enzymes for their catalyzing activity. For the majority of these enzymes, this co-enzyme is nicotinamide adenine dinucleotide (NAD(H)) (Faber, 1997). Thus, when NAD-dependent oxidoreductases are used in industrial processes, there has to be some form of supplying the enzymes with NAD+. Since NAD+ is a very expensive molecule, regeneration is the better option. Regeneration of NAD^sup +^ for example may be accomplished by (electro-photo) chemical, enzymatic or microbial catalysis.

In the case of enzymatic regeneration, the two different strategies are called "coupled-substrate" and "coupled-enzyme" types (Figure 4). In the coupledsubstrate process, the coenzyme required for the transformation of the main substrate is constandy regenerated by the addition of a second auxiliary substrate, which is transformed by the same enzyme, but in the opposite direction. The coupled-enzyme method uses two parallel redox reactions- conversion of the main substrate and coenzyme recycling- these are catalyzed by two different enzymes. In such systems, the enzyme of choice for NADH regeneration is formate dehydrogenase (FDH, E.C., which catalyzes the oxidation from formate to CO2 witii the concomitant reduction of NAD^sup +^ to NADH. The reaction is irreversible and the equilibrium can easily be shifted by flushing out the formed CO2, dius the coupled reaction is well dislocated towards the product side.

In diis context, the production of NAD-dependent D-sorbitol dehydrogenase from G. oxydans was optimized in our laboratory (Parmentier et al., 2005). However, the cell-free extracts used also contained a NAD-dependent D-mannitol dehydrogenase. This cellfree extract could be used to produce D-mannitol and D-sorbitol enzymatically from D-fructose. Efficient coenzyme regeneration was accomplished by formate dehydrogenase-mediated oxidation of formate into CO2 (Figure 5).

Coenzymes, which are oxidized or reduced during in vitro incubation with wheat flour and/or yeast, can also be regenerated with microbial enzyme systems. During yeast rising of dough in the baking process, ethanol is formed as a result of the fermentation of the sugars, present in and liberated from flour. Ethanol can be used as a substrate by the alcohol dehydrogenase present in wheat flour, resulting in the reduction of NAD^sup +^to NADH + H^sup +^. The NADH + H^sup +^ can then be re-oxidized by the use of mannitol dehydrogenase (Parmentier, 2005). Baking tests with added coenzyme regenerating systems were performed by the audiors to evaluate the effect of these components on the bread volume and on the coenzyme concentrations. With a D-mannitol dehydrogenase based regenerating system, a slight increase in bread volume was observed. However, the regenerating effect in flour was larger with the D-mannitol dehydrogenase of Leuconostoc pseudomesenteroides than with the same enzyme from Gluconobacter oxydans. Thus, coenzyme regeneration systems are effective in (in vivo/in pono) coenzyme regeneration, as can be concluded from dough extractions (Parmentier, 2005).

FIGURE 5 Schematic representation of reactions Involved In D- mannitol/D-sorbitol production and enzymatic conversion between D- fructose (*) and D-mannitol ([white circle]) + D-sorbitol ([black triangle down]), Na-formate ([black triangle down]).

III.6 Ribitol Oxidizing System

In 1934, Reichstein and co-workers were the first to notice the bioconversion of ribitol to ribulose by an unidentified acetic acid bacterium (Reichstein, 1934). Recently, Adachi et al. (2001) showed diat ribitol oxidation is carried out by two different enzymes; one NAD(P)-dependent cytosolic enzyme and another membrane-bound NAD(P)- independent enzyme. The cytosolic enzyme oxidized ribitol to Dribulose, whereas the membrane-bound oxidized diis substrate to L- ribulose. L-ribulose was accumulated in the medium, as the direct oxidation product catalyzed by membrane-bound ribitol dehydrogenase according to the Bertrand-Hudson rule. According to Adachi etal. (2001), L-ribulose can be further incorporated into the cytoplasm in several ways when there is need for carbonand energy sources. This incomplete ribitol oxidation reaction is coupled to the respiratory chain of the bacteria and the electrons generated in substrate oxidation are sequentially transferred to the terminal electronacceptor in the cytoplasmic membranes generating bio-energy (Adachi et al., 2001). For the Gluconobacter membrane-bound ribitol dehydrogenase, the optimum pH was found to be 5.0 and no reactivity was noted at alkaline pH regions (Arcus and Edson, 1956).

L-ribulose is an important chiral lead molecule used for the synthesis of, among others, L-ribose, a high-value rare sugar used in the preparation of antiviral drugs. These drugs-nucleoside- analogues-gain importance in the treatment of severe viral diseases, such as those caused by the HIV or hepatitis virus. The production of L-ribulose is performed by a biocon version-type process in which the aeration level is the most critical factor (De Muynck et al, 2006). In a well- aerated system, production of L-ribulose via dehydrogenation of ribitol with G. oxydans could be reached with a maximal conversion rate of 15.6 g L^sup -1^ h^sup -1^. Furthermore, the cells can quantitatively convert a 300 g L^sup -1^ solution within 30 hours. Another important factor is the harvesting time of the cells used in the bioconversion. Cells harvested during the exponential phase showed much lower bioconversion capacities than cells harvested in the early stationary phase (De Muynck et al., 2006) (Figure 6).

Gluconobacter cells typically differentiate by forming intracytoplasmatic membranes after exponential growth on polyols (White and Claus, 1982). White and Claus proved that cells containing a higher development of intracytoplasmatic membranes possess improved oxidation power. Moreover, the specific activity of particulate-bound dehydrogenases from early stationary-phase cells, thus containing intracytoplasmatic membranes, was twice that obtained from a particulate fraction prepared from exponential- phase cells lacking intracytoplasmatic membranes. These authors' results suggest that increased respiratory activity of early stationary-phase cells is caused both by intracytoplasmatic membranes formation and increased synthesis (or activity) of polyol dehydrogenases found in these membranes.

FIGURE 6 Effect of the culture age on the bioconversion of ribitol to L-ribulose (cell concentration 1 g CDW L^sup -1^, ribitol initial concentration 50 g L^sup -1^, pH 7.0).

These observations clarify the results obtained with ribitol, given that the early stationary phase was indeed the cell phase that showed improved oxidation levels (59% ribitol removal). Therefore, the early-stationary phase cells should best be used in the bioconversion. The slighdy lower level of oxidation observed for the stationary-phase cells, when compared to the level measured for the early stationary phase, might be due to some loss of cell viability.

III.7 Glycerol Dehydrogenase and the Production of Dihydroxyacetone

Dihydroxyacetone (DHA) is an oxidation product of glycerol, produced by the action of membranebound glycerol dehydrogenase. DHA is used as a sun tanning agent and serves as an intermediate for the synthesis of various organic chemicals and surfactants (Claret et al., 1994). The responsible enzyme, purified by Ameyama et al. (1985) from G. industrius, was identified as an enzyme that shows broad substrate specificity toward many kinds of polyhydroxyl alcohols including glycerol, D-mannitol, D-sorbitol, Darabitol, ribitol, propylene glycol and meso-erythritol, but not ethanol, aliphatic aldehydes, D-glucose, D-fructose, D-gluconate or 2-keto-D- gluconate (Lapenaite et al., 2005; Tkac et al., 2000). Recent studies of two different membrane-bound quinoproteins, D-arabitol (ArDH) and D-sorbitol (SDH) dehydrogenases purified from Gluconobacter suboxydans IFO 3257 and IFO 3255, respectively, showed that diese quinoproteins are identical and responsible for almost all sugar alcohol dehydrogenations in Gluconobacter species (Matsushita et al., 2003). As mentioned earlier, these authors suggest that this main polyol dehydrogenase should be referred to as glycerol dehydrogenase.

For the production of DHA, glycerol was most rapidly oxidized by this enzyme at pH 7.5 to 8.0 and a maximal total activity was observed at the end of the exponential phase of bacterial growth. The enzyme was found to be a quinoprotein that included PQQ as the prosthetic group.

High concentrations of glycerol were shown to inhibit G. oxydans growth by interfering with the cell division mechanism. Glycerol caused cell deformation to long filaments instead of the short rods that are usually observed (Ohrem and Merck, 1996). Ohrem and Voss (1996) disproved earlier assumptions on glycerol inhibition of DHA production (Claret et al., 1992) but demonstrated that DHA irreversibly damaged G. oxydans cells (Ohrem and Merck, 1996; Ohrem and Voss, 1996). Recent work of Bauer et al. (2005) circumvented this problem by using a two-stage reactor system, in which product formation was observed up to a maximum DHA concentration of 220 kg m^sup -3^. They also showed that G. oxydans was reversibly growth- inhibited during the repeated fed-batch mode in a range of 80-160 kg m^sup -3^ DHA. The G. oxydans cultures lost their regeneration capability at DHA concentrations above 160 kg m^sup -3^ (Bauer et al., 2005).

III.8 NADP-Dependent Shikimate Dehydrogenase and Shikimate Production

Recently, an intracellular NADP-dependent shikimate dehydrogenase (SKDH, E.C. was purified from Gluconobacter oxydans IFO 3244. The enzyme consists of a single protein unit and requires NADP exclusively to catalyze the reaction between shikimate and 3- dehydroshikimate (Adachi et al., 2006). SKDH can be used for the synthesis of shikimate from 3-dehydroshikimate, which in turn can be produced from quinate by the successive action of two enzymes: quinoprotein quinate dehydrogenase and 3-dehydroquinate dehydratase, in the cytoplasmic membranes of acetic acid bacteria (Adachi et al., 2006). Hence, the proposed production process requires two systems: first an oxidative fermentation to produce 3-dehydroshikimate from quinate via 3-dehydroquinate with dried Gluconobacter cells or membrane fractions, and secondly, a conversion of 3- dehydroshikimate to shikimate by intracellular SKDH. Since the latter reaction consumes NADPH, this co-enzyme is elegantly regenerated by the action of intracellular glucose dehydrogenase. This process might contribute to the efficient synthesis of shikimate, which is needed for the production of a large number of antibiotics, alkaloids and herbicides, as well as for the synthesis of oseltamivir, a product aimed at protecting people from pandemic flu infections (Adachi et al., 2006). IV. ENANTIOSELECTIVE BIOCATALYSIS WITH G. OXYDANS

The unique ability of G. oxydans cells to regioselectively and rapidly oxidize polyols and other carbohydrates has recently widened its application field. Several reports on the use of G. oxydans for the enantioselective oxidation of (pro)chiral alcohols and sugar derivatives have appeared in the literature during the past few years (Svitel et al., 2006). A summary of these reports is given in Table 3 and some of the most recent examples are discussed below.

Primary alcohols, such as 2-chloropropanol, 2-phenylpropanol and 2-phenyledianol, can easily be oxidized to their corresponding aldehydes and acids by using G. oxydans (Gandolfi et al., 2002; Gandolfi et al., 2004; Romano et al., 2002). It seems that the bulkiness of the substituents may have an effect on the reaction velocity and enantioselectivity (Romano et al., 2002). For example, 2-chloropropanol is oxidized to a lower extent but with a higher enantiomeric excess than 2-phenylpropanol. Further investigation of the selectivity of the oxidations revealed that the stereo bias is mainly dependent on the substrate and not on the strain used. Interestingly, when higher diols are used as a substrate, only the primary alcohol group is oxidized and the oxidation of 1,4-diols, such as 1,4-nonandiol, can lead to the formation of lactones (Romano et al., 2002).

A problem that is often encountered in the microbial oxidation of these alcohols is the toxicity of the products formed. Furthermore, when using higher concentrations of the substrate, the enantioselectivity of the reaction drops as a result of the simultaneous action of different dehydrogenases (Gandolfi et al., 2002). Different strategies can be used to circumvent these problems: (1) the addition of a second phase (for example an adsorbing resin or an organic solvent) to partition the substrate and/or product and lower its effective concentration in the aqueous phase; (2) the use of additives able to complex the substrate reversibly (e.g. cyclodextrins); and (3) the immobilization of the whole cells on a solid support (Gandolfi et al., 2002). Zigova et al. (2000) could increase the oxidation of butanol to butyric acid with G. oxydans CCM 1783 up to 93.3% of the theoretical yield with a productivity of 0.66 g l^sup -1^ h^sup -1^. In order to decrease the inhibitory effect of the product and to achieve higher butyrate productivity, they studied the possibility of butyric acid extraction from the production medium witii Hostarex A327 in oleylalcohol (Zigova et al., 2000). In 2002, Leon et al. (2002) reported on the stereoselective oxidation of 2-methyl-1,3- propanediol to (R)-beta-hydroxyisobutyric acid (HIBA). Here, an aqueous/organic biphasic system was used in order to maintain a low concentration of the (toxic) product in the aqueous phase. By using trioctyl phosphine oxide (TOPO) in isooctane, which shifted the reaction in the right direction and made it possible to recover the product in situ, a production of 20 mg per gram fresh cell weight per hour could be reached (Leon et al., 2002).

TABLE 3 Summary of "New" Substrates for Dehydrogenation by G. Oxydans

Gandolfi et al. (2002) on the other hand, could increase the rate, substrate tolerance and enantioselectivity of the oxidation of (R, S)-2-phenyl-1-propanol to (S)-2-phenylpropionic acid with A. aceti by adding cyclodextrins to the reaction medium. A four-fold increase in the reaction rate (45% molar conversion after 6 h) was achieved by using methyl cyclodextrin, with enantiomeric excess above 97%. The same oxidation with cells immobilized in calcium alginate could increase the reaction less dramatically. However, in another report (Gandolfi et al., 2004) these researchers used a combination of cell immobilization and fed-batch fermentation successively to increase the yield of the oxidation of 2-phenyl-1- ethanol to phenylacetic acid. A final yield of 23 g phenylacetic acid per liter could be reached by repeatedly adding 2.5 g of the substrate every 24 h during 9 days.

G. oxydans cells can also be used for the selective oxidation of hydroxyl groups in carbohydrate derivatives. Landis et al. (2002) successively oxidized N-butylglucamine to 6-deoxy-6-butylamino sorbose. By performing the reaction at low temperature (12-15 [degrees]C), and optimizing the other parameters of the procedure, they reached a bioconversion yield of approximately 95%. Likewise, 1- amino-1-deoxy-D-sorbitol can be dehydrogenated to 6-amino-6-deoxy-L- sorbose and N-hydroxyethyl-1-amino-1-deoxy-D-sorbitol can be dehydrogenated to its corresponding L-sorbose derivative by bioconversion with G. oxydans (Schedel, 2000). The aminopolyols are not converted into intermediates of central metabolism and do not promote growth of G. oxydans. Thus, these biotransformations have to be carried out as a two-step process. In the first phase, G. oxydans is grown on D-sorbitol to produce biomass. In the second phase, the resting cells oxidize the aminopolyols (Deppenmeier et al., 2002).

The 6-amino-L-sorbose can easily be converted to 1- deoxynojirimycin by catalytic hydrogenation; likewise the N- hydroxyethyl and N-butyl derivatives can be reduced to N- hydroxyethyl-1-deoxynojirimycin (miglitol) and to N-butyl-1- deoxynojirimycin, respectively (Landis et al., 2002; Schedel, 2000). These nojirimycin-derivatives are glucose analogues (Figure 7), which inhibit alpha-glucosidases. As such, miglitol is used for the treatment of non-insulin-dependent type II diabetes mellitus (Asano, 2003; Deppenmeier et al., 2002). The N-butyl-1-deoxynojirimycin, on the other hand, is considered an effective drug in the treatment of type I Gaucher disease, a lysosomal storage disorder caused by deficient lysosomal beta-glucocerebrosidase activity (Asano, 2003; Butters et al., 2005). Together with 1-deoxynojirimycine, N-butyl-1- deoxynojirimycin has been considered as an antiviral drug against HIV and/or hepatitis viruses. However, in vivo tests did not indicate practical use as anti-HIV agents (Asano, 2003).

FIGURE 7 Chemical structures of D-glucose and nojirimycin- derivatives.


Gluconobacter dehydrogenases are considered ideal for use in biosensors, due to their ability to stoichiometrically oxidize a wide range of substrates, the products of which accumulate in the medium (Reshetilov, 1996). Biosensors are usually an inexpensive and very specific alternative for monitoring biochemical reactions, requiring very low sample preparation (Reshetilov et al., 1997; Tkac et al., 2000).

V.1 Biosensors with Whole G. Oxydans Cells

Whole cell containing biosensors have the advantage that they are less expensive and they are simple to construct, as no enzyme purification is needed (Reshetilov et al., 1997). Moreover, enzymes are more stable in their natural environment (the cell), which also provides them with the necessary co-enzymes and activators (Tkac et al., 2000). This also makes these types of biosensors more robust and stable. On the other hand, whole cell biosensors are less specific, due to the presence of a whole set of enzymes, and they usually require a longer response time (Reshetilov et al., 1997; Tkac et al., 2000).

Whole G. oxydans cells containing biosensors have been constructed for the monitoring of ethanol (Hikuma et al., 1995; Reshetilov et al., 1998; Tkac et al., 2003) and xylose concentrations (Reshetilov et al., 1997), for sensing glucose (Lee et al., 2002; Reshetilov et al., 1998; Svitel et al., 1998), for carbohydrate determination in lignocellulose hydrolysate fermentations (Tkac et al., 2000), and are generally considered promising tools for the monitoring of sugar, aldose and polyalcohol concentrations (Lusta and Reshetilov, 1998; Reshetilov et al., 1998). Disaccharide-sensing membranes have been prepared by co- immobilization of G. oxydans with cells of Saccharomyces cerevisiae for sucrose determination or with permeabilized cells of Kluyveromyces marxianus for lactose determination (Svitel et al., 1998). However, it is noteworthy that biosensors for glucose, sucrose, lactose and ethanol are well-established tools in biochemical analysis. The latter are constructed with D-glucose oxidase (E.C. for measuring D-glucose, invertase and mutarotase in combination with the same D-glucose oxidase for determination of sucrose, D-galactose oxidase (E.C. for lactose and alcohol oxidase (E.C. for ethanol measurements (www.ysilifesciences.com).

The major limitation of whole cell biosensors is often their inability to distinguish between different substrates of interest (Lobanov et al., 2001). Recently, several options were investigated to improve the selectivity of these biosensors (Tkac et al., 2003), in particular sensors for measuring ethanol in the presence of glucose. Microbial sensors, consisting of G. oxydans cells immobilized on Clark-type amperometric oxygen electrodes, apparently exhibit nearly complete additivity for total glucose plus ethanol concentrations from 0.0 to 0.6 mM (Reshetilov et al., 1998). Given this principle, analyses could be established by comparing the output given by the biosensor, and the concentration of either glucose or ethanol measured independently with another, more selective sensor (Lobanov et al., 2001; Reshetilov et al., 1998). Combining these data using chemometrics and multivariate calibration techniques makes it possible to make accurate estimates of both glucose and ethanol concentrations. A more recent solution, by which ethanol concentrations can be determined effectively in the presence of high amounts of glucose, is by using the size exclusion effect of a cellulose acetate membrane (Tkac et al., 2002). The electrode, to which the bacterial cells are attached, is covered with a cellulose acetate membrane that hampers the diffusion of glucose, whereas ethanol diffusion is possible. With this membrane, an ethanol/ glucose selectivity ratio of 541.1 was reached. This system also works well for glycerol, as the ethanol/glycerol selectivity ratio is 1366.6 (Tkac et al., 2002). V.2 Biosensors with Purified Dehydrogenases

Biosensors can be made more selective by using (partially) purified enzymes instead of whole cells. However, due to the low stability and the limited commercial availability of purified dehydrogenases, only a few analytical systems using these enzymes have been reported to date (Lapenaite et al., 2005; Tkac et al., 2001).

D-fructose dehydrogenase (FDH) has been under investigation as a candidate for a D-fructose biosensor since 1996 (Garcia et al., 1996). As FDH is a quinoprotein, there is no need to supply the system with additional cofactors; the PQQ. is tightly bound to the apo-enzyme. However, there is a need of providing the system with mediators, such as hexacyanoferrate(III) or ferrocene. As these molecules are water soluble, their leaching from the system forms an important drawback. By adding the FDH onto an electrode covered with a cellulose acetate film embedded with ferrocene, the ferrocene leakage could be efficiently prevented (Tkac et al., 2001). Such a biosensor does not suffer from ascorbate interference, has a rather low detection limit (3-7 [mu]M depending on the amount of ferrocene added) and a fairly good stability (40% of initial activity after 8 h of continuous use) (Tkac et al., 2001).

Glycerol dehydrogenase (GIyDH) of G. oxydans has been under investigation for use in biosensors developed for the monitoring of glycerol and triglyceride during wine fermentations (Lapenaite et al., 2005). As this enzyme is not commercially available (contrary to FDH), membrane fractions (mGlyDH) of G. oxydans sp. 33, as well as purified enzyme (pGlyDH) solutions, were used.

The membrane fraction sensors had a greater affinity to the substrate than the ones with purified enzyme. This is probably due to greater diffusion restrictions that can be attributed to the material structure of the membrane fractions; since hydrophobic cell membrane adsorbs better on the surface of the hydrophobic carbon surface and creates a natural surrounding of the immobilized enzyme (Lapenaite et al., 2005). Furthermore, the membrane fractions were less sensitive to Co^sup ++^, Cd^sup ++^ and Ni^sup ++^, probably because the membrane lipids play a protective role.

In order to prevent the leaching out of mediator, aromatic mediators were polymerized onto the electrode surface. However, this increased the Km of the biosensors as the immobilized mediator cannot get through to the deeper layers of GlyDH and thus only communicates with the first layer.

Due to the low stability of these sensors, they can not easily be used for the on-line monitoring of (wine) fermentations. However, they might be applied in single use biosensors or in biosensors with a precalibration step (Lapenaite et al., 2005).


VI.1 Bacterial Cellulose

Cellulose is nature's most abundant macromolecule, recognized as the major component of plant biomass, but also as a representative of microbial extracellular polymers (De Wulf et al., 1996). Bacterial cellulose is synthesized by several bacterial genera, of which Acetobacter xylinum is the most efficient producer. Recently, a Gluconobacter oxydans strain capable of producing bacterial cellulose has also been isolated (Jia et al., 2004). The produced cellulose proved to be identical to that produced by A. xylinum, however, yields were much lower.

Although bacterial cellulose has been produced by static culture 0oris and Vandamme, 1993), agitated culture is more suitable for commercial-scale production because higher production rates can be achieved (De Wulf et al., 1996). With the aim of improving bacterial cellulose production in agitated culture, the screening of suitable strains (Masaoka et al., 1993; Seto et al., 1997), improvement of the producing organism (De Wulf et al., 1996; Ishkawa et al., 1995) and cultivation conditions (Chao et al., 2001; De Baets et al., 1997; Kouda et al., 1997; Oikawa et al., 1995; Vandamme et al., 1998) have been extensively investigated.

Recently, bacterial cellulose production could be enhanced by disrupting the PQQ-dependent glucose dehydrogenase (GHD) gene. Since GDH and bacterial cellulose synthesis are in competition for glucose, knocking out this gene creates a better substrate yield for bacterial cellulose production. More specifically, the glucose conversion to bacterial cellulose could be increased by 70%. Secondly, ethanol supplementation to this mutant could further increase the production to 7 g L^sup -1^. This is probably due to the fact that ethanol can act as an energy source for ATP generation, supplying glucose hexokinase (GHK) with sufficient energy. Glucose hexokinase is the responsible enzyme for initiation of glucose metabolism and a member of the ATPdependent hexose kinases. Thus, supplying this enzyme with