L-Erythrulose Production by Oxidative Fermentation Is Catalyzed by PQQ-Containing Membrane-Bound Dehydrogenase

Biosci. Biotechnol. Biochem., 66 (2), 307–318, 2002

L-Erythrulose Production by Oxidative Fermentation is Catalyzed by PQQ-Containing Membrane-bound Dehydrogenase

Duangtip MOONMANGMEE,† Osao ADACHI,‡ Emiko SHINAGAWA,* Hirohide TOYAMA, Gunjana THEERAGOOL,** Napha LOTONG,** and Kazunobu MATSUSHITA

Department of Biological Chemistry, Faculty of Agriculture, Yamaguchi University, Yamaguchi 753-8515, Japan *Department of Chemical and Biological Engineering, Ube National College of Technology, Tokiwadai, Ube 755-8555, Japan **Department of Microbiology, Faculty of Science, Kasetsart University, Bangkok 10900, Thailand

Received August 10, 2001; Accepted September 20, 2001

Thermotolerant Gluconobacter frateurii CHM 43 was alkaline pHs such as 9.0–10.5. L-Erythrulose reduction selected for L-erythrulose production from meso- was found at pH 6.0 with NADH as coenzyme. Judging erythritol at higher temperatures. Growing cells and the from the catalytic properties, the NAD-dependent en- membrane fraction of the strain rapidly oxidized meso- zyme in the cytosolic fraction was regarded as a typical erythritol to L-erythrulose irreversibly with almost pentitol dehydrogenase of NAD-dependent and the en- 100% of recovery at 379C. L-Erythrulose was also zyme was independent of the oxidative fermentation of produced e‹ciently by the resting cells at 379Cwith L-erythrulose production. 85% recovery. The enzyme responsible for meso- erythritol oxidation was found to be located in the Key words: acetic acid bacteria; membrane-bound cytoplasmic membrane of the organism. The EDTA- meso-erythritol dehydrogenase; NAD- resolved enzyme required PQQ and Ca 2＋ for L- dependent meso-erythritol de- erythrulose formation, suggesting that the enzyme hydrogenase; L-erythrulose reductase; catalyzing meso-erythritol oxidation was a quino- oxidative fermentation protein. Quinoprotein membrane-bound meso- erythritol dehydrogenase (QMEDH) was solubilized In relation to sugar and sugar alcohol metabolism and puriêd to homogeneity. The puriêd enzyme in acetic acid bacteria, we have characterized many showedasinglebandinSDS-PAGEofwhichthe membrane-bound dehydrogenases localized in the molecular mass corresponded to 80 kDa. The optimum cytoplasmic membranes.1) Membrane-bound de- pH of QMEDH was found at pH 5.0. The Michaelis hydrogenases catalyze substrate oxidation, yielding constant of the enzyme was found to be 25 mM for an oxidation product by which the oxidative fermen- meso-erythritol as the substrate. QMEDH showed a tations have the actual function such as acetate fer- broad substrate speciˆcity toward C3-C6 sugar alcohols mentation, D-gluconate fermentation, L-sorbose in which the erythro form of two hydroxy groups exist- fermentation, and so on. In the cytosolic fraction of ed adjacent to a primary alcohol group. On the other the same organisms, diŠerent kinds of NAD(P)- hand, the cytosolic NAD-denpendent meso-erythritol dependent dehydrogenases have been indicated.1) The dehydrogenase (CMEDH) of the same organism was NAD(P)-dependent dehydrogenases characteristical- puriêd to a crystalline state. CMEDH showed a ly catalyze reduction of the oxidized product after it molecular mass of 60 kDa composed of two identical is incorporated into the cytoplasm. In our previous subunits, and an apparent sedimentation constant was study,2) membrane-bound dehydrogenses related to 3.6 s.CMEDH catalyzed oxidoreduction between meso- pentitol oxidation in acetic acid bacteria were report- erythritol and L-erythrulose. The oxidation reaction was ed, indicating that they are responsible for the oxida- observed to be reversible in the presence of NAD at tive fermentation and for which NAD-dependent

‡ To whom correspondence should be addressed. Laboratory of Applied Microbiology, Department of Biological Chemistry, Faculty of Agriculture, Yamaguchi University, Yamaguchi 753-8515, Japan; Tel: +81-83-933-5857; Fax: +81-83-933-5820; E-mail: osao＠ agr.yamaguchi-u.ac.jp † On leave from the Department of Microbiology, Faculty of Science, King Mongkut's University of Technology Thonburi, Bangkok 10140, Thailand. A part of this paper was presented in the annual meeting of the Japan Society for Bioscience, Biotechnology, and Agrochemistry held in Kyoto from March 24 to 27, 2001. The abstract paper appears in the Nippon Nogeikagaku Kaishi, 75, 215 (2001). 308 D. MOONMANGMEE et al. cytosolic enzymes have no use. Regarding oxidative vation was done aerobically at 309Cor379Cina fermentations of C4 sugar alcohols, no information 500-ml side-armed Erlenmeyer ‰ask. Bacterial has been accumulated, though some earlier reports growth was monitored by measuring the turbidity by on meso-erythritol oxidation by aerobic bacteria a Klett-Summerson photoelectric colorimeter with a were made.3–4) Since L-erythrulose is not available red ˆlter. readily from commercial sources, it is important to investigate the fermentation proˆle of L-erythrulose Preparation of L-erythrulose. Areactionmixture production, identiˆcation of the enzyme responsible obtained from incubation of meso-erythritol with for meso-erythritol oxidation, and puriˆcation and resting cells of thermotolerant Gluconobacter CHM characterization of the responsible enzyme, provid- 43 at 379C for 6 h was centrifuged at 9,000×g for ing basic information for L-erythrulose production. 10 min to remove the cells. The resultant supernatant As has already been reviewed,1) oxidative fermenta- was freeze-dried overnight and mixed with 1z tion is only catalyzed by the membrane-bound trichloroacetic acid to removed remaining protein in dehydrogenases localized on the outer surface of the the sample. After a brief centrifugation, the solution cytoplasmic membranes of acetic acid bacteria. With was put onto a Dowex-50 W column (2×80 cm) and respect to cytosolic NAD(P)-dependent meso- eluted with distilled water. Fractions were collected erythritol dehydrogenase (CMEDH), only one en- of every 100 drops and L-erythrulose was measured zyme has been reported as L-erythrulose reductase by the resorcinol test and phenol sulfuric method es- (EC 1.1.1.162) from beef liver and L-erythrulose sentially by the same method described previously.2) reduction to meso-erythritol predominated over The pooled L-erythrulose fraction was then freeze-d- meso-erythritol oxidation.5) It is also reported that ried overnight. Purity of L-erythrulose was measured NADPH is more reactive than NADH in L-erythru- by thin layer chromatography using a silica gel 60 lose reduction with the mammalian enzyme. plate (Merck, Darmstadt, Germany) with a solvent In this paper, Gluconobacter frateurii CHM 43 system of n-butanol:ethanol:distilled water (4:1:1, was screened among thermotolerant Gluconobacter vWv). A solution of 0.5z triphenyl tetrazolium chol- and mesophilic strains as high L-erythrulose producer ride was used as the coloring agent. from meso-erythritol when grown at 379C. Enzymes catalyzing meso-erythritol oxidation from the mem- Preparation of membrane and cytosolic fraction. brane fraction of NAD-independent and from the G. frateurii CHM 43 was used throughout this study. cytosolic fraction of NAD-dependent meso-erythritol The culture conditions were the same as described in dehydrogenases of the organism were puriêd to our previous paper used for L-ribulose production by homogeneity. The puriêd meso-erythritol dehydro- acetic acid bacteria.2) The bacterial strain was cul- genase from the membrane fraction was identiêd as tured in 30 l of the medium in a 50-l jar fermentor. a quinoprotein, which is the enzyme responsible for The cultivation was done at 309C for 20 h under the oxidative fermentation, but the NAD-dependent vigorous aeration. Bacterial cells were harvested with meso-erythritol dehydrogenase was independent of a Sharples centrifuge (Carl Padberg, model GLE, the oxidative fermentation. Further characterization Germany) and suspended in 2 mM Tris-HCl (pH 7.5) of the two enzymes in catalytic and physicochemical containing 10 mMD-sorbitol. The bacterial cell sus- properties are also conducted. pension chilled in ice-cold water was passed twice through a French pressure cell press (SIM Aminco, Materials and Methods Spectronic Instruments, Inc., Rochester, NY, USA) at 16,000 lbWin2. To remove intact cells and cell Chemicals. NAD, NADP, NADH, NADPH, and debris, the suspension was centrifuged at 9,000×g at yeast extract were kind gifts from Oriental Yeast Co., 49C for 15 min. Separation of the membrane fraction Tokyo. Mydol 10 was a kind gift from Kao Co., from the cytosolic fraction was done by ultracen- Tokyo. Other chemicals used were from commercial trifugation (Hitachi model 55P-72) at 150,000×g at sources of guaranteed grade unless otherwise stated. 49C for 60 min. For the EDTA (ethylenediamine- N,N,N?,N?-tetraacetic acid, disodium salt) treatment Microorganisms and culture conditions. Meso- experiment, the membrane fraction was homo- philic Gluconobacter strains used in this study were genized with 20 mM Tris-HCl (pH 7.5) while 10 mM obtained from the Institute for Fermentation, Osaka, potassium phosphate buŠer (pH 6.0) was used for (IFO). Thermotolerant Gluconobacter strains were enzyme solubilization. isolated from Thailand and characterized as has been reported previously.6) Culture medium used for EDTA treatment and reactivation of the enzyme screening for L-erythrulose-producing strains was activity. The membrane solution of which the protein composed of 1z meso-erythritol, 0.1z yeast ex- concentration was adjusted to 10 mgWml was incu- tract, and 0.1z polypeptone. For the resting cell bated twice with 10 mM EDTA for 30 min with gentle preparation, a potato medium was used.7) The culti- stirring at 49C. The membrane fraction was precipi- meso-Erythritol Oxidation by Thermotolerant Gluconobacter 309 tated by centrifugation and resuspended with 20 mM mg of protein. Tris-HCl (pH 7.5) to remove excess EDTA. The same procedure was repeated two times. All centrifu- SDS-polyacrylamide gel electrophoresis (SDS- gations were done by an ultracentrifuge operated at PAGE). SDS-PAGE was done on 12.5z (wWv) slab 150,000×g at 49C for 60 min. Reactivation of meso- gel by the methods described by Laemmli.9) Before erythritol dehydrogenase activity which had been application, samples were treated with 6z (w Wv) SDS resolved by EDTA-treatment was done by the and 0.1 mM dithiothreitol at 609C for 30 min. The addition of 5 mM pyrroloquinoline quinone (PQQ), following calibration proteins (Bio-Rad, Hercules, andWor 5 mM calcium chloride to the membrane sus- CA, U.S.A.) with the indicated molecular mass were pension and incubated at 259C for 30 min. meso- used as references: phosphorylase b (110 kDa), Erythritol dehydrogenase activity was measured by bovine serum albumin (77 kDa), ovalbumin using either potassium ferricyanide or phenazine (50.5 kDa), carbonic anhydrase (35.2 kDa), soybean methosulfate (PMS)-2,6-dichlorophenolindophenol trypsin inhibitor (29.1 kDa), and lysozyme (DCIP) as artiˆcial electron acceptors. (21.2 kDa). Gels were stained with Coomassie brilli- ant blue (CBB R-250). Assay of enzyme activity. All enzyme assays were done in a ˆnal volume of 1.0 ml at 259C. The stan- Analytical ultracentrifugation. Analytical ultra- dard assay for measurement of membrane-bound centrifugation was done by a Hitachi model SCP85H meso-erythritol dehydrogenase activity was done ultracentrifuge at 209C throughout measurements. using potassium ferricyanide or PMS-DCIP as the Estimations of the sedimentation coe‹cient was artiˆcial electron acceptors. The standard assay done by the methods of sedimentation velocity,10) mixture contained 10 mmol potassium ferricyanide or which was operated by a combination of a Hitachi 0.2 mmol PMS in the presence of 0.11 mmol DCIP, UV scanner (ABS-7), an absorption scanner, and a 100 mmol meso-erythritol, and enzyme solution in UC processor (DA-7). 0.8 ml of McIlvaine buŠer (pH 5.0). The reaction was started by the addition of enzyme solution. The Identiˆcation of reaction products. Reaction resulting Prussian-blue color with potassium ferric- products by the puriêd enzyme were identiêd by yanide was measured colorimetrically at 660 nm incubating 100 mmol substrate; glycerol, meso- (HitachimodelU-2000)andPMS-DCIPassaywas erythritol, ribitol, D-arabitol, D-sorbitol, and D-man- traced by measuring the decreasing absorbance at nitol, 0.6 mmol PMS in McIlvaine buŠer (pH 5.0) for 600 nm (Hitachi model 200–10), respectively. One 10 h at room temperature in the dark. The reaction unit of the enzyme activity was deˆned as the amount product formed was then identiêd by a combination of enzyme catalyzing oxidation of 1 mmol of meso- of enzymatic method and thin layer chromatography erythritol per min under these conditions. The (Silica gel 60), which was done by a solvent system as speciˆc activity was deˆned as units of enzyme activi- described above and the Rf value was compared with ty per mg of protein. Protein content was measured an authentic standard. according to a modiˆcation of Lowry's method8) using bovine serum albumin as the protein standard. Results and discussion The enzyme activity of NAD-dependent meso- erythritol dehydrogenase was measured by a routine Screening of L-erythrulose producing Gluconobac- method used in common for NAD(P)-dependent en- ter strains zymes by recording the rate of increase of NADH at Mesophilic and thermotolerant Gluconobacter 340 nm with meso-erythritol as the substrate in a strains were used for screening of L-erythrulose reaction mixture at 259C. The reaction mixture (1 ml) production. Growth of the organisms in the meso- contained 50 mmol of meso-erythritol, 50 mmol of erythritol medium at 309C was observed in all tested glycine-NaOH (pH 9.5), 0.5 mmol of NAD, and an species of Gluconobacter except for G. cerinus had appropriate amount of enzyme. L-Erythrulose reduc- very poor growth. Oxidative fermentation of meso- tion was measured in a reaction mixture (1 ml) con- erythritol by growing cells of thermotolerant G. taining 50 mmol L-erythrulose, 100 mmol of potassi- frateurii CHM 43 at 379Cwasfoundtobehigher um phosphate buŠer (pH 6.0), 0.2 mmol of NADH, than that of 309C. In contrast, it clearly showed that and the enzyme. One unit of the enzyme activity was at higher temperatures, mesophilic Gluconobacter deˆned as the amount of enzyme catalyzing 1.0 mmol produced L-erythrulose lower than that of ther- of NADH formation and 1.0 mmol of NADH reduc- motolerant strains (Table 1). tion in meso-erythritol oxidation and L-erythrulose G. frateurii CHM 43 was proved to be one of ther- reduction, respectively. An optical absorption motolerant Gluconobacter strains that can grow well 1z coe‹cient of Ecm 280 nm＝10.0 was tentatively used for even at higher temperatures such as 379C, though the protein concentration measurement. The speciˆc they show the optimum growth at 33–359C.6) This activity was deˆned as units of enzyme activity per strain was preliminary classiêd as G. frateurii ac- 310 D. MOONMANGMEE et al.

Table 1. Production of L-Erythrulose from meso-Erythritol by Thermotolerant and Mesophilic Gluconobacter

L-Erythrulose (z) Strain 309C379C

1 day 2 days 3 days 1 day 2 days 3 days

G. frateurii CHM 1 93 88 87 68 51 49 G. frateurii CHM 8 79 85 99 44 37 21 G. frateurii CHM10898989654747 G. frateurii CHM16667166878987 G. frateurii CHM33898285564746 G. frateurii CHM43796666949695 G. frateurii CHM52897777896460 G. frateurii CHM54958989735858 G. suboxydans var. a IFO3254938241523124 G. suboxydans var. a IFO3255393224590 0 G. suboxydans var. a IFO3258626049241810 G. asaii IFO 3275 85 87 95 29 21 12 G. oxydans IFO 3292 80 81 40 32 23 23 G. melanogenus IFO 3294 89 88 84 51 42 40

Culture broth containing L-erythrulose was reacted with resorcinol reagent and heated at 909C for 20 min. The resulting solution was then measured spec- trophotometrically at absorbance 490 nm. L-Erythrulose content was calculated using puriˆed L-erythrulose as the standard.

Fig. 1. Course of L-Erythrulose Production by Growing Cells of G. frateurii CHM 43 at 309C(A)and379C(B).

cording to Mason and Claus11) and others biochemi- that the optimal production of L-erythrulose at cal characteristics as has been reported elsewhere.6) higher temperatures was better than those at lower Comparison of meso-erythritol oxidase activity temperatures. The oxidation product from meso- among thermotolerant Gluconobacter and meso- erythritol formed by the action of Gluconobacter is philic Gluconobacter strains showed that, unlike the based on the fact that this tetritol has the erythro other, G. frateurii CHM 43 had meso-erythritol form of two secondary hydroxyl groups adjacent to oxidase activity higher than glycerol oxidase activity the primary alcohol moeity.22) That is known as Ber- (data not shown). Moreover, this strain accumulated trand-Hudson's rule. the highest L-erythrulose after 24–36 h of cultivation when grown at 379C though the total cell mass was Course of L-erythrulose production by growing less than those observed at 309C. Thus, judging from cells the enzyme activity, productivity, and growth Growth of G. frateurii CHM 43 at 309C was faster characteristics, G. frateurii CHM 43 was selected for than that of 379C but it was interesting that produc- further study. It was obvious that pH of the culture tion of L-erythrulose at 379C was higher than at 309C broth was decreased from 6.5 to 4.0–3.5 after 24– (Fig. 1). The strain produced L-erythrulose to 90z 36 h of cultivation, suggesting that this oxidative after 24 h at 309C, however, it was decreased by 34z fermentation had taken place by the action of after 48 h incubation, while it was maintained at membrane-bound dehydrogenase of the organism of almost 100z when grown at 379C. From these which the optimum pH for oxidation reaction is results it can be explained that at 309C the microor- shown to be in the acidic region.2,7,12–21) It appeared ganism used both meso-erythritol and L-erythrulose meso-Erythritol Oxidation by Thermotolerant Gluconobacter 311

Fig, 2. L-Erythrulose Production by Resting Cells of G. frateurii CHM 43 at 379C. (A) EŠects of resting cells concentrations. Resting cells concentrations indicated as protein content was incubated with meso-erythritol for 2 h at 379C. (B) Course of L-erythrulose formation. Resting cells (1.6 mg protein) was incubated with meso-erythritol for the period indicated. L-Erythrulose formed in the reaction mixture was measured by the resorcinol method as described in Materials and Methods. as growth substrates. It can be said that, at higher temperatures were counted by higher conversion temperatures, the organism highly oxidized meso- rate, shorter incubation time, and reduced cost for erythritol to L-erythrulose in order to generate energy cooling expenses when it comes to actual manufac- for survival under the stressful conditions of higher turing of L-erythrulose. The appearance of puriˆed L- temperatures. Similar control of the oxidative fer- erythrulose solution after chromatographic separa- mentation for producing a higher yield of oxidation tion and subsequent freeze-dry gave a clear pale- product by stressing the bacterial growth at higher yellow syrup. The solution thus obtained showed a 23 temperatures was successful for L-sorbose produc- speciˆc rotation of +12.29at [a]D ,whichwasin 30 18 tion and D-fructose production (Adachi, O., unpub- good accordance with [a]D +11.319 and [a]D lished observations). +11.49of reported authentic L-erythrulose.4,24)

L-Erythrulose production with resting cells Localization of meso-erythritol dehydrogenase Resting cells converted meso-erythritol to L- meso-Erythritol dehydrogenase activity of G. erythrulose and the conversion rate was proportional frateurii CHM 43 was found to be located absolutely to not only cell concentrations but also incubation in the cytoplasmic membrane the same as other period (Fig. 2). The production e‹ciency or conver- typical membrane-bound dehydrogenases in acetic sion rate by resting cells was quite high and reached acid bacteria.1) meso-Erythritol dehydrogenase used almost 100z after 5 h incubation at 379C. Large potassium ferricyanide as well as PMS-DCIP as elec- scaled production of L-erythrulose could be done tron acceptors in vitro. Using potassium ferricyanide practically by incubating meso-erythritol with the as the electron acceptor, meso-erythritol dehydro- resting cells aerobically at 379C. After removal of genase activity was detected only in the acidic region cells from the reaction mixture, the reaction mixture at pH 5.0, while the enzyme activity showed a double was freeze-dried and puriˆed by Dowex-50 W pH optimum, a major peak at pH 5.0 and a minor column chromatography. This calcium form of the peak at pH 9.0, when PMS-DCIP was used as the cation exchange column chromatography has been electron acceptor (Fig. 3) like other membrane- reported to be useful for separation of diŠerent su- bound dehydrogenases.25) These results were in good gars from corresponding sugar alcohols.23) L-Erythru- agreement with those reported by De Ley and lose was eluted as a single peak which could be Dochy26) reporting that only cell debris and small detected with resorcinol reagent or phenol sulfuric particles of the cytoplasmic membrane oxidizes acid reagent. As the result, meso-erythritol given to meso-erythritol to the corresponding keto-derivative, the resting cells reaction mixture completely disap- L-erythrulose, but not for protoplasts. peared after 6 h incubation. L-Erythrulose formation was quite high, over 85z, which was quantitatively Reactivation of enzyme activity and L-erythrulose the same as those obtained after 9 days cultivation at production 289C as reported by Whistler and Underko‰er.4) The Prosthetic groups of membrane-bound dehydro- advantages of L-erythrulose production at higher genases of acetic acid bacteria are divided into ‰avin 312 D. MOONMANGMEE et al. adenine dinucleotide (FAD) covalently bound or pyr- lent cation such as Ca2+ contributes to binding of roloquinoline quinone (PQQ).1) In order to identify PQQ to the apo-enzyme. The divalent cation is essen- the prosthetic group of meso-erythritol dehydro- tial for enzyme activity of quinoproteins, which can genase, the membrane fraction was ˆrst treated with be removed easily by EDTA.27,28) While FAD is cova- chelating agent; EDTA. meso-Erythritol dehydro- lently bound to the enzyme it is di‹cult to remove it genase activity was much decreased after EDTA by such chelating agents.29) We reported previously treatment, while exogenous addition of PQQ or PQQ that EDTA treatment allowed the membrane-bound and CaCl2 reactivated the enzyme activity to the glycerol dehydrogenase of G. industrius IFO 3260 to original level. These results suggested that mem- be the apo-enzyme, and the subsequent exogenous brane-bound meso-erythritol dehydrogenase was addition of PQQ to the membrane led reactivation of shown to have PQQ as the prosthetic group the enzyme.30) Quinoprotein glycerol dehydrogenase (Fig. 4-A). In quinoprotein dehydrogenase, a diva- activity was recovered by the addition of cations such as Mg 2+,Ca2+,Ni2+ or Co 2+.Ca2+ and Ni 2+ seemed to be the most eŠective for reactivation of the enzyme activity. Characterization of meso-erythritol dehydrogenase as a quinoprotein was done by thin layer chromatography. As shown in Fig. 4-B, increased L- erythrulose production was observed when PQQ and

CaCl2 were added to the native membrane. When the membranefractionwastreatedwithEDTA,L- erythrulose production was much reduced, while L- erythrulose production was resumed to the original

level by the addition of PQQ and CaCl2 to the EDTA-treated enzyme. Thus, it was concluded that L-erythrulose production was catalyzed by PQQ containing membrane-bound meso-erythritol dehydrogenase. Therefore, it was identiˆed as a member of the membrane-bound dehydrogenases Fig. 3. Optimum pH of Membrane-bound meso-Erythritol containing PQQ as prosthetic groups like methanol Dehydrogenase from G. frateurii CHM 43 using Potassium 31) Ferricyanide () and Phenazine Methosulfate (PMS)-2,6- dehydrogenase (EC 1.1.99.8) in methylotrophs, 15,17) Dichlorophenolindophenol (DCIP) () as the Electron Accep- alcohol dehydrogenase (EC 1.1.99.8), aldehyde tors. dehydrogenase (EC 1.2.99.3),32) and glycerol

Fig. 4. EŠects of PQQ and CaCl2 on the Membrane-bound meso-Erythritol Dehydrogenase and L-Erythrulose Production before and after EDTA Treatment. (A) meso-Erythritol dehydrogenase activity before and after EDTA treatment.

Reactivation of enzyme by the addition of PQQ and CaCl2 was done as described in Materials and Methods. (B) Thin layer chromatogram of reaction product. Reaction mixture consisted of 100 mmol meso-erythritol, 400 ml membrane fraction, and 500 ml McIlvaine buŠer pH 5.0. The reaction mixture was then incubated at 259C with shaking for 2 h. Sample was obtained by ultracentrifugation at 150,000×g for 1 h. Lane 1, puriˆed L-erythrulose; lane 2, meso-erythritol was incubated with native membrane; lane 3, native membrane which had been reactivat-

ed with PQQ and CaCl2 before reacted with meso-erythritol; lane 4, meso-erythritol was reacted with EDTA-treated membrane; lane 5,

reactivation of EDTA-treated membrane with PQQ and CaCl2 before incubation with meso-erythritol. meso-Erythritol Oxidation by Thermotolerant Gluconobacter 313

Table 2. EŠects of PQQ and CaCl2 on the Membrane-bound meso-Erythritol Dehydrogenase

Enzyme preparations Relative activity (z) (1) Native membrane 54

(2) Native membrane+PQQ+CaCl2 100 (3) Native membrane+Mydol 10 21

(4) Native membrane solubilized+PQQ+CaCl2 45 (5) Holoenzyme solubilized 86

(2) PQQ and CaCl2 were added to the native membrane at the concentrations of 5 mM and 5 mM, respectively, and left at room temperature for 30 min before assaying enzyme activity. (3) Enzyme was solubilized from the native membrane with 1z Mydol 10.

(4) After the enzyme was solubilized with 1z Mydol 10, PQQ and CaCl2 were added to the enzyme solution at the same concentrations as above.

(5) After PQQ and CaCl2 were added to the native membrane at the same concentrations as above, enzyme was solubilized with 1z Mydol 10.

dehydrogenase (EC 1.1.99) in acetic acid bacteria,21) made before the solubilization step by the addition of

D-glucose dehydrogenase (EC 1.1.99.17) in acetic PQQ and CaCl2 to the membrane suspension. acid bacteria,20) pseudomonads,33) Acinetobacter Solubilization was done using 1.5z Mydol 10 in the calcoaceticus,34) and Escherichia coli.28) presence of 0.1 M KCl. The membrane suspension was stirred gently for 1 h and centrifuged at 150,000 Solubilization conditions ×g for 1 h. The solubilized fraction was then dia- Since meso-erythritol dehydrogenase was located lyzed overnight against 10 mM sodium acetate buŠer on the cytoplasmic membrane, solubilization of the (pH 5.0) containing 5 mM CaCl2 and 0.3z Mydol 10 enzyme was done using 1z various detergents in the (buŠer A) for 6 h. Dialyzed enzyme was then presence of 0.1 M KCl as the chaotropic agent. En- centrifuged to remove precipitate at 9,000×g for zyme solubilization was done with the original native 20 min. The obtained supernatant was put onto a membrane as it was prepared as well as the native DEAE-Toyopearl column (2.5×10 cm) previously membrane to which both PQQ and CaCl2 had been equilibrated with buŠer A and the column was added to make it a holoenzyme as described in washed with the same buŠer with ‰ow rate of Materials and Methods. Among detergents tested, 1.5 mlWmin. QMEDH activity was passed through Mydol 10, an alkyl glucoside detergent, was selected the column together with quinoprotein alcohol de- for the purpose. meso-Erythritol dehydrogenase hydrogenase.15,17) The fractions containing enzyme could be solubilized with recovery of 80–86z from activity were pooled and then put onto a CM- the membrane fraction when it had been brought to cellulose column (1×5 cm) previously equilibrated the holoenzyme (Table 2). Enzyme activity solubi- with 10 mM sodium acetate buŠer (pH 4.5) contain- lized with the original native membrane was about ing 5 mM CaCl2 and 0.3z Mydol 10. After the 21z, indicating that membrane-bound meso- column was washed with the same buŠer, elution of erythritol dehydrogenase existed partially as apo- the enzyme was done with a linear gradient by in- enzyme. creasing concentration of CaCl2 to 75 mM at a ‰ow Enzyme activity assayed with solubilized enzyme rate of 0.6 mlWmin. Fractions having enzyme activity using potassium ferricyanide was much lower than eluted at about 10–15 mM CaCl2 were combined and that of PMS-DCIP. It was suggested that the deter- put into a dialysis tube to concentrate the enzyme by gent used might be dissociated heme component in embedding in sucrose powder. As the results the puri- the membrane during enzyme solubilization, since êd enzyme was clearly separated from quinoprotein potassium ferricyanide seemed to be less eŠective as alcohol dehydrogenase at diŠerent CaCl2 concentra- an electron acceptor once the enzyme has been tion (Fig. 5). A summary of the enzyme puriˆcation solubilized. Thus, the following steps after enzyme is shown in Table 3. The puriêd enzyme had a solubilization were required to use only PMS-DCIP speciˆc activity of 7 unitsWmg with meso-erythritol as for enzyme activity measurement. the substrate. Though the recovery of the enzyme from CM-cellulose column chromatography was not Puriˆcation of membrane-bound meso-erythritol high, separation of QMEDH from quinoprotein al- dehydrogenase (QMEDH) cohol dehydrogenase was successful after many times Membrane-bound meso-erythritol dehydrogenase of trials of column chromatography. The puriêd from thermotolerant G. frateurii CHM 43 was puri- meso-erythritol dehydrogenase gave one band on êd to homogeneity as below. All operations were SDS-PAGE and the molecular mass measurement by done at 49C. The membrane fraction of the microor- SDS-PAGE gave a relative molecular mass of 80 kDa ganism was homogenized with 10 mM potassium (Fig. 6). phosphate buŠer (pH 6.0) giving a protein concentration of about 10 mgWml. Holoenzyme formation was 314 D. MOONMANGMEE et al.

Table 3. Puriˆcation Summary of Membrane-bound meso-Erythritol Dehydrogenase

Total activity* Total protein Speciˆc activity Recovery Puriˆcation Step (units) (mg) (unitsWmg) (z) (fold) Native membrane 490 548 0.9 100 1 Holoenzyme formation 678 550 1.2 138 1 Solubilization 641 170 3.8 131 4 DEAE-Toyopearl 507 52 9.8 103 11 CM-cellulose 47 7 6.7 10 8

*Enzyme activity was assayed in McIlvaine buŠer, pH 5.0, using PMS-DCIP as the electron acceptor.

Table 4. Substrate Speciˆcity of Membrane-bound meso- Erythritol Dehydrogenase

Relative activity (z)assayedwith Substrate Membrane fraction Puriˆed enzyme

meso-Erythritol 100 100 DL-Threitol 15 0 Glycerol 93 66 Ribitol 86 41 D-Arabitol 129 107 L-Arabitol 31 0 Xylitol 14 0 D-Sorbitol 100 62 D-Mannitol 93 45 Dulcitol 7 0 Fig. 5. Chromatography of Quinoprotein Membrane-bound myo-Inositol 21 0 meso-Erythritol Dehydrogenase of G. frateurii CHM 43 on CM- Xylose 32 0 Cellulose Column. D-Glucose 68 0 Pooled meso-erythritol dehydrogenase from DEAE- D-Fructose 13 0 Toyopearl column chromatography was put onto a CM- L-Sorbose 15 0 cellulose column (1×5 cm) previously equilibrated with 10 mM 2-Keto-D-Gluconate 11 0 sodium acetate buŠer (pH 4.5) containing 0.3z Mydol 10 and Acetaldehyde 88 0

5mM CaCl2. Elution of enzyme was made by a linear gradient of Ethanol 71 0

CaCl2 as indicated.

Catalytic properties of puriêd QMEDH The puriêd QMEDH had the optimum pH at 5.0 for meso-erythritol oxidation, the same optimum pH as observed with membrane fraction. Nitroblue tetrazolium (NBT), PMS and DCIP can be available as electron acceptors for QMEDH while potassium ferricyanide was not as electron acceptor suggesting thattheenzymewasheme-free.Theapparent Michaelis constant for meso-erythritol was 25 mM. QMEDH showed a broad substrate speciˆcity toward C3-C6 sugar alcohols such as glycerol, meso- erythritol, ribitol, D-arabitol, D-sorbitol and D-mannitol as shown in Table 4. QMEDH did not oxidize DL-threitol because it has the threo form of hydroxyl groups adjacent to the primary alcohol group. QMEDH well oxidized meso-erythritol and D- arabitol glycerol with relatively the same high rates while ribitol and D-mannitol were oxidized at the rate 40–65z those of meso-erythritol. QMEDH oxidized Fig. 6. SDS-PAGE of Quinoprotein Membrane-bound meso- Erythritol Dehydrogenase from G. frateurii CHM 43. (2R, 3R)-(„)-2,3-butanediol with that of one third of SDS-PAGE. Lane 1, marker proteins; lane 2, solubilized meso-erythritol but did not oxidize S,S-(+)-2,3- fraction; lane 3, DEAE-Toyopearl pooled; lane 4, puriêd butanediol (data not shown). It was suggested that meso-erythritol dehydrogenase from CM-cellulose. Each lane the enzyme readily oxidized only the erythro form was loaded with 10 mgprotein. and R conˆguration of sugar alcohols and of alcohol, meso-Erythritol Oxidation by Thermotolerant Gluconobacter 315 respectively. QMEDH reported here was diŠerent in many respects from puriêd alcohol dehydrogenase,15) D-sorbitol dehydrogenase,29) D-fructose dehydrogenase,7) D-mannitol dehydrogenase,35) and D-gluconate dehydrogenase.18) Identiˆcation of reaction product was done with glycerol, meso-erythritol, ribitol, D-arabitol, D-sorbitol, and D-mannitol as the substrates in the presence of PMS and the corresponding oxidation products, dihydroxyacetone, L-erythrulose, L-ribulose, D-xylulose, L-sorbose, and D-fructose, respectively, were identiêd by a combination of enzymatic methods and thin-layer chromatography.

Puriˆcation of cytosolic NAD-dependent meso- erythritol dehydrogenase (CMEDH) To the cytosolic fraction (850 ml) which was separated from the membrane as described above was added b-mercaptoethanol to 5 mM and put on a DEAE- cellulose column (5×35 cm) that had been equilibrated with 2 mM KPB (pH 7.2). After the column was washed with KPB containing 50 mM KCl, elution of CMEDH was done with KPB containing 0.2 M KCl. When KCl in KPB was increased more than 0.25 M, NAD-dependent D-sorbitol dehydrogenase came out as has been reported previously.36) There was no CMEDH activity with the NAD- Fig. 7. Puriˆed NAD-Dependent meso-Erythritol Dehydro- genase from G. frateurii CHM 43. dependent D-sorbitol dehydrogenase. To the pooled (A) Photopicture of the crystalline enzyme. enzyme fraction of CMEDH (680 ml), ammonium (B) Sedimentaion patterns of crystalline enzyme. sulfate was added to 0.4 saturation (22.6 g ammoni- Photographs were taken every 20 min as indicated after reach- um sulfateW100 ml) and the precipitate appeared was ing 60,000 rpm. The enzyme solution containing 17.5 mg pro- removed by a conventional centrifugation. Ammoni- teinWml was used. um sulfate was further added to 0.6 saturation by in- (C) SDS-PAGE. Lane 1, marker proteins; lane 2, NAD-dependent meso- creasing 12 g ammonium sulfate per 100 ml of the erythritol dehydrogenase (10 mg protein) was put on. original enzyme solution. The precipitate was collected by centrifugation and dialyzed against 2 mM KPB thoroughly. The dialyzed enzyme was put on a trifuge and recrystallization of CMEDH was done DEAE-Sephadex A-50 column (1×20 cm) which had under the similar manner. Photopicture of the been equilibrated with 2 mM KPB. After the column recrystallized enzyme is shown in Fig. 7-A. Summary was washed with KPB containing 50 mM KCl, the of CMEDH puriˆcation is shown in Table 5 and column was treated by a linear gradient set by 75 mM CMEDH was puriˆed by 72-fold from the cytosolic KCl (500 ml) and 250 mM KCl (500 ml) and every fraction of G. frateurii CHM 43. 10 ml of fraction was collected. CMEDH came out from the column where KCl concentration cor- Physicochemical and catalytic properties of responded to about 120 mM.CMEDH was concen- CMEDH trated by ammonium sulfate precipitation. After the When analyzed in an analytical ultracentrifuga- precipitate was dissolved in a minimum volume of tion, the enzyme showed a single sedimentation peak KPB, the enzyme solution was passed through a as shown in Fig. 7-B. An apparent sedimentation column of Sephadex G-200 (2×180 cm) and 2.0-ml coe‹cient was measured to be 3.6 s. Molecular mass fractions were collected. When the elution of the en- measurement by SDS-PAGE gave an apparent zyme was monitored by intensity of absorbance at molecular mass of 28 kDa (Fig. 7-C). CMEDH came 280 nm, elution pattern showed slightly asymmetric out from a Sephadex G-75 column (1×150 cm) peak, though elution of the enzyme activity gave a almost at the same position of NADP-dependent D- symmetric pattern. CMEDH was precipitated with mannitol dehydrogenase (EC 1.1.1.138) from G. ammonium sulfate and dissolved in a small volume suboxydans IFO 12528.37) Thus, it was reasonable to of KPB. Impurities were separated by crystallizing conclude that the molecular mass of CMEDH had to CMEDH by ammonium sulfate as the precipitant. be 60 kDa and it was composed of two identical The ˆrst crystals were collected by a table top cen- subunits. In enzyme puriˆcation, enzyme activity was 316 D. MOONMANGMEE et al.

Table 5. Summary of Enzyme Puriˆcation of NAD-Dependent meso-Erythritol Dehydrogenase

Total protein Total activity* Speciˆc activity Yield Puriˆcation Step (mg) (units) (unitsWmg) (z) (fold) Cytosolic fraction 16,500 770 0.046 100 1 DEAE-cellulose 8,600 795 0.092 103 2 Sephadex A-50 220 480 2.18 62 47 Sephadex G-200 120 350 2.91 45 63 Recrystallized 52 171 3.30 22 72

*Enzyme activity was assayed in 50 mM glycine-NaOH, pH 9.5, by measuring the increase of absorbance of NADH using meso-erythritol as the substrate. assayed with respect to meso-erythritol dehydro- Acknowledgments genase using meso-erythritol as the substrate. meso- Erythritol was rapidly oxidized to L-erythrulose un- We wish to express our sincere thanks to Dr. Y. der alkaline pHs such as 9.0–10.5 in the presence of Akakabe of our faculty for his kind measurement of NAD, but NADP was inert as the coenzyme diŠerent speciˆc rotation of the oxidation product, L-erythru- from mammalian enzyme.5) Apparent Michaelis con- lose, and Dr. S. Tanaka of our faculty for his kind stants for meso-erythritol and NAD were measured contribution to photopicture processings of crystal- to be 33 mM and 2.8 mM, respectively. Glycerol, D- line enzyme. We thank Mr. Y. Fujii for his helpful mannitol, and D-sorbitol were oxidized by the en- discussion in membrane-bound enzyme puriˆcation. zyme at 13, 68, and 18z of the relative rate to meso- A part of this work was done by collaboration in a erythritol oxidation. It was interesting to see that L- Core University Program between Yamaguchi arabitol, ribitol and xylitol were rapidly oxidized by University and Kasetsart University supported by the the enzyme, while D-arabitol oxidation (126z of Scientiˆc Cooperation Program agreed by the Japan relative rate) was almost comparable to meso- Society for the Promotion of Science (JSPS) and the erythritol oxidation. The maximum reaction rate National Research Council of Thailand (NRCT). measured with L-arabitol, ribitol, and xylitol were found to be 32.0, 34.6, and 21.0 unitsWmg of protein, References nearly ten times higher than that with meso- erythritol. Thus CMEDH puriêd from the cytosolic 1) Matsushita, K., Toyama, H., and Adachi, O., fraction in this study can be regarded as a typical Respiratory chains and bioenergetics of acetic acid pentitol dehydrogenase of NAD-dependent. It is a bacteria. in `Àdvances in Microbial Physiology'', clear contrast in the substrate speciˆcity of the mem- Vol. 36, ed. Rose, A. H. and Tempest, D. W., Aca- brane-bound enzyme to the cytosolic enzyme. Ac- demic Press, Ltd., London, pp. 247–301 (1994). cording to Table 4, QMEDH well oxidized D-arabitol 2) Adachi, O., Fujii, Y., Ano, Y., Moonmangmee, D., Toyama, H., Shinagawa, E., Theeragool, G., but not L-arabitol, ribitol, or xylitol among pentitols Lotong, N., and Matsushita, K., Membrane-bound tested, while CMEDH oxidized D-arabitol but at sugar alcohol dehydrogenase in acetic acid bacteria lower rate. catalyzes L-ribulose formation and NAD-dependent On the other hand, L-erythrulose was reduced to ribitol dehydrogenase is independent of the oxidative meso-erythritol by CMEDH in the presence of fermentation. Biosci. Biotechnol. Biochem., 65, NADH with a pH optimum around 6.0 (data not 115–125 (2001). shown). Apparent Michaelis constants for L-erythru- 3) Bertrand, G., Novel preparation of L-erythrulose by loseandNADHweremeasuredtobe10mM and meso-erythritol dehydrogenation. Compt. Rend. 0.1 mM, respectively. These kinetic constants also in- Sean. Acad. Sci. (in French), 130, 1472–1475 (1900). dicate that CMEDH works mainly in reduction of L- 4) Whistler,R.L.andUnderko‰er,L.A.,Theproduc- erythrulose to meso-erythritol, as similarly concluded tion of l-erythrulose by the action of Acetobacter suboxydans upon erythritol. J. Am. Chem. Soc. 60, with mammalian NADP-dependent L-erythrulose reductase.5) It can be concluded that there is no 2507–2508 (1938). 5) Uehara, K., Tanimoto, T., and Sato, H., Studies on speciˆc NAD(P)-dependent meso-erythritol de- D-tetrose metabolism IV. Puriˆcation and some L hydrogenase in the organism. The relative rate of - properties of D-erythrulose reductase from beef liver. erythrulose reduction was nearly ten times higher J. Biochem., 75, 333–345 (1974) than that of meso-erythritol oxidation in the 6) Moonmangmee, D., Adachi, O., Ano, Y., cytoplasm. Therefore, NAD-dependent CMEDH in Shinagawa, E., Toyama, H., Theeragool, G., the cytosolic fraction may contribute to the reduction Lotong, N., and Matsushita, K., Isolation and of ketopentitol to pentitol as similarly discussed with characterization of thermotolerant Gluconobacter ribitol dehydrogenase.2) Only QMEDH is responsible strains catalyzing oxidative fermentation at higher for the oxidative fermentation producing L-erythru- temperatures. Biosci. Biotechnol. Biochem., 64, lose. 2306–2315 (2000). meso-Erythritol Oxidation by Thermotolerant Gluconobacter 317 7) Ameyama, M., Shinagawa, E., Matsushita, K., 20) Ameyama, M., Shinagawa, E., Matsushita, K., and and Adachi, O., D-Fructose dehydrogenase of Adachi, O., D-Glucose dehydrogenase of Gluconobacter industrius: puriˆcation, characteriza- Gluconobacter suboxydans: solubilization, puriˆca- tion, and application to enzymatic microdetermination and characterization. Agric. Biol. Chem., 45, tion of D-fructose. J. Bacteriol., 145, 814–823 (1981). 851–861 (1981). 8) Dully, J. R. and Grieve, P. A., A simple technique 21) Ameyama, M., Shinagawa, E., Matsushita, K., and for eliminating interference by detergents in the Adachi, O., Solubilization, puriˆcation and proper- Lowry method of protein determination. Anal. ties of membrane-bound glycerol dehydrogenase Biochem., 64, 136–141 (1975). from Gluconobacter industrius. Agric. Biol. Chem., 9) Laemmli, U. K., Cleavage of structural proteins dur- 49, 1001–1010 (1985). ing the assembly of the head of bacteriophage T4. Na- 22) Hann,R.M.,Tilden,E.B.,andHudson,C.S.,The ture (London), 227, 680–685 (1970). oxidation of sugar alcohols by Acetobacter subox- 10) Chervenka, C. H., Determination of sedimentation ydans. J. Am. Chem. Soc., 60, 1201–1203 (1938). rates. in `À Manual of Methods for the Analytical 23) Angyal, S. J., Bethell, G. S., and Beveridge, R. J., Ultracentrifuge'', ed. Chervenka, C. H., Spinco The separation of sugars and of polyols on cation- Division of Beckman Instruments Inc., Palo Alto, exchange resins in the calcium form. Carbohydr. California, pp. 23–37 (1970). Res., 73, 9–18 (1979). 11) Mason, L. M. and Claus, G. W., Phenotypic charac- 24) Muller, von H., Montigel, C., and Reichstein, T., teristics correlated with deoxyribonucleic acid se- Reine l-erythrulose (l-2-keto-tetrose). Helv. Chim. quence similarities for three species of Gluconobac- Acta, 20, 1468–1473 (1937). ter. G. oxydans (Henneberg 1897) De Ley 1961; G. 25) Shinagawa, E., Matsushita, K., Toyama, H., and frateurii sp. nov., and G. asaii sp. nov. Int. J. Syst. Adachi, O., Production of 5-keto-D-gluconate by Bacteriol., 39, 174–184 (1989). acetic acid bacteria is catalyzed by pyrroloquinoline 12)Shinagawa,E.Chiyonobu,Y,Adachi,O.,and quinone (PQQ)-dependent membrane-bound D- Ameyama, M., Distribution and solubilization of gluconate dehydrogenase. J. Mol. Cat. B: Enzymatic, particulate gluconate dehydrogenase and particulate 6, 341–350 (1999). 2-ketogluconate dehydrogenase in acetic acid bacter- 26) De Ley, J. and Dochy, R., On the localisation of oxi- ia. Agric. Biol. Chem., 40, 475–483 (1976). dase systems in Acetobacter cells. Biochim. Biophys. 13) Ameyama, M., Tayama, K., Miyagawa, E., Shinaga- Acta, 40, 277–289 (1960). wa, E., Matsushita, K., and Adachi, O., A new en- 27) Ameyama, M., Nonobe, M., Shinagawa, E., zymatic microdetermination procedure for ethanol Matsushita, K., and Adachi, O., Method of enzymat- with particulate alcohol dehydrogenase from acetic ic determination of pyrroloquinoline quinone. Anal. acid bacteria. Agric. Biol. Chem., 42, 2063–2069 Biochem., 151, 263–267 (1985). (1978). 28) Ameyama, M., Nonobe, M., Hayashi, M., 14) Ameyama, M., Tayama, K., Shinagawa, E., Shinagawa, E., Matsushita, K., and Adachi, O., Matsushita, K., and Adachi, O., New enzymatic de- Mode of binding of pyrroloquinoline quinone to apo- termination of D-gluconate with particulate D- glucose dehydrogenase. Agric. Biol. Chem., 49, gluconate dehydrogenase. Agric. Biol. Chem., 42, 1227–1231 (1985). 2347–2354 (1978). 29) Shinagawa, E., Matsushita, K., Adachi, O., and 15) Adachi, O., Miyagawa, E., Shinagawa, E., Matsushi- Ameyama, M., Puriˆcation and characterization of ta, K., and Ameyama, M., Puriˆcation and proper- D-sorbitol dehydrogenase from membrane of ties of particulate alcohol dehydrogenase from Gluconobacter suboxydans var. a. Agric. Biol. Acetobacter aceti. Agric. Biol. Chem., 42, 2331–2340 Chem., 46, 135–141 (1982). (1978). 30) Adachi, O., Matsushita, K., Shinagawa, E., and 16) Shinagawa, E., Matsushita, K., Adachi, O., and Ameyama, M., Enzymatic determination of pyrrolo- Ameyama, M., Membrane-bound D-gluconate de- quinoline quinone with a quinoprotein glycerol hydrogenase of Serratia marcescens: puriˆcation and dehydrogenase. Agric. Biol. Chem., 52, 2081–2082 properties. Agric. Biol. Chem., 42, 2355–2361 (1978). (1988). 17) Adachi, O., Tayama, K., Shinagawa, E., Matsushita, 31) Anthony, C. The bacterial oxidation of methane, K., and Ameyama, M., Puriˆcation and characteriza- methanol, formaldehyde, and formate. in ``The tion of particulate alcohol dehydrogenase from Biochemistry of Methylotrophs'', ed. Anthony, C. Gluconobacter suboxydans. Agric. Biol. Chem., 42, Academic Press, New York, pp. 152–194 (1982). 2045–2056 (1978). 32) Ameyama, M., Matsushita, K., Ohno, Y., 18) Matsushita, K., Shinagawa, E., Adachi, O., and Shinagawa, E., and Adachi, O., Existence of a novel Ameyama, M., Membrane-bound D-gluconate de- prosthetic group, PQQ, in membrane-bound, elec- hydrogenase: its kinetic properties and a reconstitu- tron transport chain-linked, primary dehydrogenases tion of gluconate oxidase. J. Biochem., 86, 249–256 of oxidative bacteria. FEBS Lett., 130, 179–183 (1979). (1981). 19) Adachi, O., Tayama, K., Shinagawa, E., Matsushita, 33) Matsushita, K. and Ameyama, M., D-Glucose K., and Ameyama, M., Puriˆcation and characteriza- dehydrogenase from Pseudomonas ‰uorescens,mem- tion of membrane-bound aldehyde dehydrogenase brane-bound. in ``Methods in Enzymology'', Vol. 89. from Gluconobacter suboxydans. Agric. Biol. ed. Wood, W. A., Academic Press, Ltd., New York, Chem., 43, 503–515 (1980). pp. 149–154 (1982). 318 D. MOONMANGMEE et al. 34) Duine,J.A.,Frank,J.,Jr.,andvanZeeland,J.K., 36) Adachi, O., Toyama, H., Theeragool, G., Lotong, Glucose dehydrogenase from Acinetobacter calcoa- N., and Matsushita, K., Crystallization and proper- ceticus: a `quinoprotein'. FEBS Lett., 108, 443–446 ties of NAD-dependent D-sorbitol dehydrogenase (1979). from Gluconobacter suboxydans IFO 3257. Biosci. 35) Oikawa, T., Nakai, J., Tsukagawa, Y., and Soda, K., Biotechnol. Biochem., 63, 1589–1595 (1999). A novel type of D-mannitol dehydrogenase from 37) Adachi, O., Toyama, H., and Matsushita, K., Crys- Acetobacter xylinum: occurrence, puriˆcation, and talline NADP-dependent D-mannitol dehydrogenase basic properties. Biosci. Biotechnol. Biochem., 61, from Gluconobacter suboxydans IFO 12528. Biosci. 1778–1782 (1997). Biotechnol. Biochem., 63, 402–407 (1999).