Glycerol Dehydrogenase from Gluconobacter Industrius

Agric. Biol Chem., 49 (4), 1001 -1010, 1985 1001

Solubilization, Purification and Properties of Membrane-bound Glycerol Dehydrogenase from Gluconobacter industrius Minoru Ameyama,Emiko Shinagawa, Kazunobu Matsushita and Osao Adachi Laboratory of Applied Microbiology, Department of Agricultural Chemistry, Faculty of Agriculture, Yamaguchi University, Yamaguchi 753, Japan Received July 30, 1984

Membrane-bound glycerol dehydrogenase was solubilized and purified about 100-fold from the membraneof Gluconobacter industrius IFO 3260 grown on a glycerol-glutamate medium. Solubilization of the enzyme was successfully achieved by use of 0.5% dimethyldodecylamineoxide in 0.05 m Tris-HCl, pH 8.0. Alcohol dehydrogenase and D-glucose dehydrogenase, which were abundantly formed in the same bacterial membrane, were eliminated on solubilization. Glycerol dehydrogenase was further purified through fractionation with polyethylene glycol 6000. The enzymeshowed a broad substrate specificity and various kinds of polyhydroxyl alcohols, in addition to glycerol, were rapidly oxidized in the presence of 2,6-dichlorophenolindophenoi and phenazine methosulfate as the electron acceptor but NADand NADPwere inert. The enzyme was proved to be a quinoprotein in which pyrroloquinoline quinone functioned as the prosthetic group.

The first report on microbial oxidation of localization of the oxidase system in cells of G. glycerol to dihydroxyacetone was by Bertrand liquefaciens and found that the oxidation of with a strain capable of L-sorbose fermen- glycerol and raeso-erythritol was carried out by tation.1* Two types of glycerol dehydrogenase the particulate fraction. Unlike glycerol oxi- have been reported, one is pyridine nucleotide dase from fungi,13) the pyridine nucleotide dependent and the other is independent and independent glycerol dehydrogenase in the membrane-bound.As to the former type of bacterial membranedid not require oxygen as enzymeyielding either glyceraldehyde or di- an electron acceptor but was linked to the hydroxyacetone, NAD-dependent glycerol de- electron transport system of the organism. hydrogenase has been partially purified from Thus the localization and enzymatic properties the soluble fractions of Escherichia coii,2) of the membrane-bound glycerol dehydro- Acetobacter suboxydans and Klebsiella pneu- genase yielding dihydroxyacetone of acetic monia.^ Recently, Yamada et al.4'~6) have acid bacteria have remained to be elucidated. crystallized NAD-dependent glycerol dehy- On the other hand, production of dihydroxy- drogenase yielding dihydroxyacetone from acetone by growing cells or resting cells of Cellulomonas sp. and characterized its enzy- acetic acid bacteria has actually advanced. In matic and structural properties in detail. recent report by Chibata et a/.,14~16) fermen- NADP-dependent glycerol dehydrogenase tative and enzymatic production of dihydroxy- yielding dihydroxyacetone has been partially acetone by acetic acid bacteria was exam- purified from various origins.7~10) King and ined to increase the productivity in fermen- Cheldelinn) reported that an enzyme catalyz- tation processes which remained low until ing glycerol oxidation yielding dihydroxy- recently. In spite of such production and usage acetone independent of ATPor NADwas of dihydroxyacetone, characterization of the present in the cell free extract of Gluconobacter membrane-bound glycerol dehydrogenase of suboxydans. De Ley and Dochy12) studied the acetic acid bacteria has not been achieved. In 1002 M. Ameyamaet al. this paper, we describe the purification and enzyme solution and lOO'/miol of glycerol in a total properties of the membrane-bound glycerol volume of 1.0ml. The reaction was initiated at 25°C by the addition of glycerol and terminated by adding 0.5 ml dehydrogenase of acetic acid bacteria. At the of ferric-Dupanol reagent. Then, 3.5ml of water was same time, it is also reported that the mem- added to the reaction mixture. The resulting Prussian- brane-bound glycerol dehydrogenase con- blue color was measured with a photometer at 660nm tains a novel prosthetic group, pyrrolo- after standing at 25°C for 20min. One unit of enzyme quinoline quinone (PQQ), as in the cases activity was defined as the amount of enzyme catalyzing of alcohol, aldehyde, and D-glucose dehydro- the oxidation of 1 ^mol of glycerol per min under these conditions; 4.0 absorbance units equaled 1 pimol of glyc- genases of acetic acid bacteria.17) erol oxidized. Alcohol dehydrogenase in Table I was assayed under essentially the same conditions as for glyc- MATERIALS AND METHODS erol dehydrogenase, except that the substrate was changed to ethanol. Materials. The yeast extract was a kind gift from Assaying of the glycerol dehydrogenase with the solubi- Oriental Yeast Industry Co. Dimethyldodecylamineoxide, lized enzymefrom the membranefraction was performed dodecylsulfobetaine and dodecylbetaine were kind gifts spectrophotometrically at 25°C by measuring the decrease from Kao-Atlas Co. Lauroylsarcosine was a product of in absorbance at 600nmof DCIPwith PMS. Once the Wako Pure Chemicals Ind., and /?-octyl-glucoside a enzymewas solubilized from the membrane,the enzyme product of Sigma Chemicals Co. All other chemicals activity was no longer linked to ferricyanide as the electron used in this study were commercial products of analyti- acceptor in the oxidation of glycerol. The enzyme activi- cal grade. Authentic PQQ was prepared according to ty was expressed in /rniol of reduced DCIP per min, the method described previously.i8) with 15-OmM"1as the extinction coefficient at pH 8.0. The reaction mixture contained lOO^umolof Tris-HCl or Microbial strains and culture conditions. The bacterial potassium phosphate buffer, pH 8.0, 0.6/xmol of DCIP, strains of the genus Gluconobacter listed in Table I were 0.6//mol of PMS, 100/imol of glycerol and the enzyme generously donated by the Institute for Fermentation, solution in a total volume of 3.0ml. One unit of enzyme activity was defined as the amount of enzyme catlyzing the described.19>Osaka (IFO). TheStock compositioncultures wereofpreparedthe cultureas previouslymedium oxidation of 1 jumol of glycerol per min. containing glycerol and the method for cultivation were Protein determination. The protein content was de- also essentially the same as described previously.19) In the case of a large scale culture, 30 liters of the glycerol termined by the method of Lowry et al.21) with bovine serum albumin as the standard. The modified method mediumwas incubated in a 50-liter jar fermentor at 30°C described by Dulley and Grieve22) was employed for for about 24hr under vigorous agitation (500rpm) and aeration (25 to 30 liters per min). Bacterial cells were samples which contained detergent. harvested at the late exponential phase as soon as the jbH of the culture medium had decreased to 5.2. Identification of reaction products. The reaction product of glycerol oxidation was identified by thm-layer chro- Preparation of the cell homogenate and membrane frac- matography with acetone as the developing solvent by tion. All operations were performed below 4°C in essen- essentially the same method as that described by tially the same way as described previously.19) The pre- Yamada et al.14) Detection of dihydroxyacetone was cipitated membranefraction on ultfacentrifugation was performed by spraying with 1% /?-anisidine-HCl in n- washedwith buffer solution to removecytoplasmic ma- butanol followed by standing for 10min at 110°C. Dihy- terials by repeating the ultracentrifugation. droxyacetone gave a dar^-brown spot on a yellow back- ground with an Rfvalue of 0.75~0.80. Authentic dihy- Assay ofglycerol dehydrogenase. As for the other mem- droxyacetone and glyceraldehyde were chromato- brane-bound dehydrogenases reported elsewhere,20' graphed on the same plate. Glyceraldehyde showed an the enzyme activity of glycerol dehydrogenase could be Rf value of 0.3 to 0.4 with relatively faint coloring assayed by colorimetry in the presence of potassium ferri- with /?-anisidine. Reaction products such as D-fruc- cyanide or spectrometry in the presence of 2,6-dichlo- tose, L-sorbose and D-xylulose were also identified by rophenolindophenol (DCIP) and phenazine methosul- thin-layer chromatography with dual solvent systems fate (PMS). Hereafter, the potassium ferricyanide assay of phenol-NH3 and «-butanol-acetic acid-water (4 : 1 : was employed for routine assaying as has already been 5). Detection of such compounds was also performed reported.20) The rate of reduction of ferricyanide to ferro- with /7-anisidine. Alternatively, D-fructose was identified cyanide correlated with the ratio of glycerol oxidation. enzymatically with D-fructose dehydrogenase for The reaction mixture contained 10/miol of potassium which only D-fructose was available as the substrate.19) ferricyanide, 0.5 ml of citrate-phosphate buffer, pH 5.0, L-Sorbose was detected by reverse reaction of NADP-de- Membrane-boundGlycerol Dehydrogenase 1003 pendent sorbitol dehydrogenase23)24) prepared from G. abundantly in most strains of Gluconobacter. suboxydans var. a. IFO 3254 in the presence of NADPH D-Glucose dehydrogenase activity was also at pH 6.5. D-Fructose was inert for the NADP-depend- predominant throughout these strains al- ent sorbitol dehydrogenase. Ketoses formed by the en- zymereaction were further detected by the cysteine- though the data are not shown. In G. in- carbazole reaction.25) dustrius, the level of alcohol dehydrogenase was exceptionally low under the culture con- Extraction and purification of the prosthetic group. To ditions employed. Thus the use of G. industrius the purified enzymesolution, cold methanol was added to 90% and then the solution was stirred gently for 2 days at IFO 3260 for further purification of glycerol 4°C. Coagulated proteins were separated by precipitation dehydrogenase was advantageous from the in a high-speed centrifuge at 10,000xg for lOmin. The standpoint of elimination of other enzyme supernatant was subjected to evaporation in vacuo at 30°C proteins. It was obvious that for glycerol oxi- in a rotary evaporator to removemethanol. The pale- dation for dihyxoyacetone production it was yellow solution was then applied to a DEAE-SephadexA- 25 column (1 x 10cm), which had been equilibrated with advantageous to use such strains as G. cerinus, G. gluconicus and G. suboxydans as has already 0.002m potassium phosphate, pH 7.0. After washing the been shown.14~16) columnwith the same buffer, the columnwas treated with the same buffer containing KC1. A minor contaminant was almost completely removed with 0.1 m KC1. When the Time course of enzyme formation by G. elution was monitored by reading the absorbance at dual wavelengths of 249nmand 330nm, the major intensity indus trius was found in the chromatographed fraction eluted with The time course of enzyme formation in 1 mKC1, though a minor elution peak was observed with cells of G. industrius IFO 3260 during growth 0.5m KC1. The fractions showing PQQ activity were was examined. As shown in Fig. 1, maximal pooled and concentrated to a minimal volume (about total enzyme activity of glycerol dehydro- 2 ml). The concentrated solution was separated from solid KC1in a highspeed centrifuge and applied to a column of genase wasobserved at the end of the exponen- Sephadex G-25 (1 x 180cm), which had been prepared in tial phase of bacterial growth. When the bac- distilled water. Fractions were collected by 3ml at a flow terial growth and the change in pH of the rate of one drop per 15 sec. culture mediumwere monitored, termination

Analytical procedures. Absorption spectra were re- of the growth was found to coincide with the corded with a Hitachi 200-10 spectrophotometer. fall in pH of the culture medium. Both the Fluorescence spectra were recorded with a Hitachi 650- maximumenzyme activity and the highest 10S fluorescence spectrophotometer at 25°C with exci- specific activity were found after 20 to 24hr tation at 365nmand emission at 465nm under essentially incubation when the pH of the mediumhad the same conditions as reported previously.18) decreased to 5.2. The culture period for max- imum enzyme formation could be conve- Enzymatic identification of the prosthetic group. Enzymatic identification of PQQ was performed by niently determined by only checking the pH measuring activation of apo-glucose dehydrogenase as of the culture medium. reported elsewhere.17) Solubilization of glycerol dehydrogenase RESULTS All operations weire carried out at 0 to 4CC, unless otherwise stated. Themembranefrac- Distribution of glycerol dehydrogenase in tion was prepared from cells at the late expo- Gluconobacter nential phase. This fraction was suspended in The enzyme activities of glycerol dehydro- 50mMTris-HCl, pH 8.0, and the protein con- genase and alcohol dehydrogenase were sur- tent was adjusted to lOmg/ml. To the suspen- veyed with cell homogenates of various strains sion, various detergents were added to the final of Gluconobacter as shown in Table I. Both concentration of 0.5% followed by gentle stir- enzyme activities were assayed at their pH ring for 60min. Glycerol dehydrogenase ac- optima, which had been determined pre- tivity was assayed with DCIP and PMS. As viously. Alcohol dehydrogenase was produced can be seen in Table II, detergents effective for 1004 M. Ameyamaet al.

TabLe I. Distribution of Glycerol Dehydrogenase and Alcohol Dehydrogenase in the Genus Gluconobacter Each strain was cultured in three shaking flasks (3Q0ml of the medium) at 30°C for 24hr. Both enzyme activities were assayed with the cell homogenateat each pH optimumas indicated using ferricyanide as an electron acceptor. GLDHand ADHdonate glycerol dehydrogenase and alcohol dehydrogenase, respectively. A ssa y ed T o ta l S p ec ifi c S tra in E n z ym e p H u n its ac tiv ity

G. a lb i du s IF O 3 25 1 G L D H 5.0 82.6 0. 89 A D H 5.0 2 5 1. 3 2.71 G. al bi d us I F O 32 5 3 G L D H 4.5 11 2.6 1.4 5 A D H 5.0 24 6.8 3. 1 7 G. ce ri n us IF O 3 26 2 G L D H 5.0 9 8.6 1.4 6 A D H 4.5 14 7.6 2. 1 8 G . ce r in us IF O 3 26 4 G L D H 5.0 1 5 3. 1 0. 87 A D H 5.0 52 7.5 2. 99 G. ce ri n us I F O 32 6 5 G L D H 4.5 5 3.2 0.6 1 A D H 5.0 434 .9 4. 98 G. ce ri n us I F O 32 6 8 G L D H 4.5 1 7 0. 1 .12 A D H 4.5 52 7.4 3 .46 G . gl u co ni cu s I FO 3 1 71 G L D H 4.5 1 4 6. 1 1. 1 6 A D H 4.5 4 0 3. 1 3. 2 1 G . gl u co ni c us I F O 32 8 5 G L D H 4.5 89 .4 0 .98 A D H 4.5 20 7. 6 2 .27 G . gl u co n ic us IF O 3 2 86 G L D H 4.5 1 29. 2 1. 3 5 A D H 4.0 58 2.0 6 .06 G . i n du st ri us I FO 3 26 0 G L D H 4.5 3 1. 4 0 .36 A D H 4.5 4.8 0 .05 G . su b o x yd a n s I F O 12 5 2 8 G L D H 5.0 13 3.2 0. 91 A D H 6.0 6 2 9.5 4. 31 G . s u b ox y d an s IF O 3 17 1 G L D H 5.0 9 9.5 0 .89 A D H 4.0 3 12. 1 2 .7 9 G . su b o x yd a n s I F O 32 6 9 G L D H 5.0 10 2.3 1. 2 8 A D H 4.5 19 1. 3 2 .39 G . su b o x yd a n s I F O 32 9 0 G L D H 4.5 99 .4 1. 0 3 A D H 5.0 142 .6 1. 4 7 G . s u b o xy d a n s I F O 3 2 91 G L D H 5.0 13 1. 6 1. 13 A D H 4.5 3 72 .4 3. 18 G . s u bo x yd an s va r. a I FO 32 5 8 G L D H 4. 5 3 1. 1 0 .3 0 A D H 4. 5 14 3.9 1. 3 9

enzymesolubilization were found to be such amineoxide was varied for enzyme solubiliza- detergents as lauroylsarcosine, dodecylsulfo- tion and the level of solubilized enzyme in the betaine, lauroylbetaine or dimethyldodecyl- centrifuged supernatant was examined. When amineoxide. Most of these detergents showed the detergentwas added at 0.1, 0.5 and 1.0% to similarities in their molecular structures and the membranesuspension and solubilization ionic characters, being amphoteric detergents. of the enzymewas performed as above, about Of the tested detergents, dimethyldodecyl- 70% of the original enzyme activity found in amineoxide was the most effective for the the membranewas recovered on solubilization purpose. with 0.5% detergent but less than 10% of the enzymeactivity was found in the supernatant Effect of detergent concentration on enzyme with the other concentrations. The solubilized solubilization and stability enzymecould be stored with no concomitant The concentration of dimethyldodecyl- loss of enzyme activity at -20°C for half a Membrane-bound Glycerol Dehydrogenase 1005

Table III. Summaryof Purification of Glycerol Dehydrogenase from Gluconobacter industrius IFO 3260 <° VR..V g I T o ta l T o ta l S p ec ifi c «0.6__« \,/\\ |35-o.s>; S te p u n its p ro te in a c tiv ity ( V )( un i ts ) ( m g) ( u ni ts / mg ) y /o J 0.4-- k Nx S« I f I //\_ ^^O _oi.O 30.2-5 ,/7 ~^~ '1 M em b r an e 1,0 3 0 1, 9 8 5 0 .5 2 10 0 S o lu b iliz ed su p . 8 9 2 5 7 6 1. 5 5 8 7 P E G * fra c tio n 2 5 3 5 2 2 5 nI---''? I I I o|n 0 12 24 36 48 60 Culture period (hr) PEG, polyethyleneglycol 6000. Fig. 1. Time Course ofFormation of Membrane-bound Glycerol Dehydrogenase during a Culture of G. industrius. HC1, pH 8.0, in which a neglegible amount of Three shaking culture flasks (300 ml of the medium)were harvested at 12hr intervals. Bacterial growth was deter- alcohol dehydrogenase and D-glucose dehy- mined from the turbidity of the culture broth at 660nm. drogenase was solubilized. On the other hand, The pH was checked before centrifugation with a pH a fair amount of D-glucose dehydrogenase was meter. Enzymeactivity was measured with the cell ho- solubilized when potassium phosphate was mogenateusing ferricyanide as the electron acceptor. used.

Table II. Effect of Detergents on Solu- BILIZATION OF GLYCEROL DEHYDROGENASE Purification of glycerol dehydrogenase The membrane fraction obtained from 30 g Themembranefraction was treated with various kinds of G. industrius cells was suspended in 50mM of detergents at 0.5% as indicated at 4°C for 1 hr. After centrifugation, the enzyme activity in the supernatant Tris-HCl buffer, pH 8.0, and the protein con- and the precipitate was assayed under the standard assay centration was adjusted to lOmg/ml. To the conditions. The activity was expressed as the relative suspension, dimethyldodecylamineoxide was activity to that of the untreated membrane. added to 0.5% followed by gentle stirring for E n z ym e a c tiv ity in 60min. After centrifugation at 68,000 x g for D e te rg e n t 60min, the resulting supernatant was frac- S u p er n a ta n t P re cip ita te tionated with polyethylene glycol 6000. The T r i t o n X - 1 0 0 1 4 . 5 % 1 5 . 1 % precipitate obtained with 15 to 25% poly- N a - D e o x y c h o l a t e 6 . 0 1 7 . 7 ethylene glycol 6000 was suspended in 50mM B rij 5 8 0 .8 1 4 .5 Tris-HCl buffer, pH 8.0, containing 0.1% de- L u b r o l W X 0 . 6 9 . 7 f i - O c t y l - g l u c o s i d e 2 2 . 7 1 8 . 8 tergent and.10% glycerol. The insoluble ma- N a - D o d e c y l s u l f a t e 2 0 . 0 0 . 1 terials were removed by centrifugation at N a -L a u ro ylsa rco sin e 4 3 .6 3 .6 65,000xg for 60min. As shown in Table III, L a u ro y lb eta in e 5 5 .6 1.0 glycerol dehydrogenase was purified about D i m e t h y l d o d e c y l a m i n e o x i d e 6 5 . 3 1 2 . 0 100-fold with a yield of 25% through this simple purification procedure. year. When a diluted enzyme solution was Purity of the final enzymepreparation made with buffer solution containing 0.1% The purity of the purified enzyme prepara- dimethyldodecylamineoxide and 10% glycerol, tion could not be estimated by conventional the enzymecould be stored in a refrigerator for methods including gel electrophoresis, gel fil- more than two weeks. The solubilization of tration and analytical ultracentrifugation due glycerol dehydrogenase was much affected by to its severe hydrophobicity. Inactivation of the buffer solution employedfor enzymesolu- the enzyme proceeded rapidly with the for- bilization. The most favorable solubilization mation of a huge aggregate in the absence of of the enzyme was obtained with 50mMTris- 0.1% dimethyldodecylamineoxide and 10% 1006 M. Ameyamaet al.

Table IV. Substrate Specificity of enzyme as summarized in Table IV. The en- Glycerol Dehydrogenase zyme catalyzed oxidation of glycerol as well as The enzymeactivity was measured with various sub- that of other polyhydroxyl alcohols such as strates as indicated at the final concentration of 33mM. The membranefraction, the solubilized supernatant and meso-Qrythvitol and D-arabitol at similar rates. the finally purified enzyme as indicated in Table III were D-Sorbitol, D-mannitol, propylene glycol, used as the enzyme. The results are expressed as the adonitol and dulcitol were also oxidized by the relative activity tq that with glycerol. enzyme but at relatively low rates. NADor E nz ym e a ct i vi ty (% ) a ss ay e d wi t h NADPand molecular oxygen were not available as the electron aceptor. PMS-DCIPwas M e m b ra n e S o lu b iliz ed P u rifi e d the most effective electron acceptor for oxida- fra c tio n su p e rn a ta n t e n zy m e tion of substrates. The oxidation product of G l y c e r o 1 1 0 0 1 0 0 1 0 0 glycerol was identified as dihydroxyacetone by D -S o r b ito 1 5 2 4 4 4 5 thin-layer chromatography. The reaction D -M an n ito 1 3 5 2 3 2 6 products on D-sorbitol oxidation and D-man- D - A r a b i t o 1 1 4 5 1 3 0 1 3 6 D u lc ito 1 4 5 3 0 29 nitol oxidation were identified as L-sorbose X y lito l and D-fructose, respectively. These two prod- A d o n ito 1 5 0 3 5 4 2 ucts showed a slight difference in Rf values m es o -E ry th rito 1 1 1 2 12 5 1 1 1 on thin-layer chromatography when exam- P r o p y l e n e g l y c o 1 1 8 5 4 4 5 4 D - F r u c t o s e 1 0 0 ined with dual solvent systems. L-Sorbose for- D -G lu c o se 2 39 mation was further confirmed by the observa- D - G l u c o n a t e 1 5 7 tion that the reaction product became avail- 2 -K eto -D -glu c o n a te able as the substrate of NADP-dependent E th a n o l , 15 A ce ta ld eh y d e 2 0 sorbitol dehydrogenase in the presence of NADPH. D-Fructose yielded on D-mannitol oxidation was also confirmed by it serving as the substrate of D-fructose dehydrogen- glycerol. During the enzyme purification, con- ase.19) On the basis of these results, the en- tamination by other species of membrane- zyme purified here was found to be different bound dehydrogenases which occur in the G. from pyridine nucleotide dependent polyal- industrius membrane sucn as alcohol, d- cohol dehydrogenases of G. oxydans reported gluuose, D-fructose and aldehyde dehydro- by Kersters et al.26) It was also concluded genases, was completely eliminated as shown that the glycerol dehydrogenase from G. in Table IV. The oxidation rate of glycerol industrius was a different enzyme species compared to that of other polyhydroxyl al- from D-sorbitol dehydrogenase purified from cohols such as raeso-erythritol, D-arabitol or d- G. suboxydans var. a.27) The former showed sorbitol settled down to a constant level a broad substrate specificity for various poly- throughout the purification, while the enzyme hydroxyl alcohols while the latter was only activity toward D-glucose, D-fructose or d- specific to D-sorbitol and other polyhy- gluconate disappeared as shown in Table IV. droxyl alcohols were not oxidized other than Considering the magnitude of purification of D-mannitol which was oxidized to an ex- the enzymepreparation with the abundance of tent of 5% relative to D-sorbitol oxidation. the enzyme in the bacterial membrane,the Moreover, as will be mentioned later, the enzymethus purified could be separated from prosthetic group of the former was identified other impurities. as PQQ, but that of the latter has been identified as bound flavin.27) Substrate specificity The substrate specificity of glycerol dehy- Catalytic properties of the purified enzyme drogenase was examined with the purified Catalytic properties of the enzyme were Membrane-bound Glycerol Dehydrogenase 1007 surveyed with the purified enzyme prepara- 1.5 r- -|1-5 tion. Glycerol was most rapidly oxidized by the enzyme at pH 7.5 to 8.0, with DCIP and I' o 'I PMS as the electron acceptor. The optimum pH found with the membrane fraction at ll.0- Y -10I acidic pH was abolished by solubilization with detergents. On standing at various pHs at 4°C, 15w - - 13w the enzymewas stable at an alkaline pH such as 8.5 to 10 in the presence of detergent and å So.5- fo -°-5"| glycerol. The apparent Michaelis constant for various substrates was determined and Km values of 34, 22 and 19mMwere obtained with glycerol, D-arabitol and raeso-erythritol, respectively. The Km value for glycerol was o 10 L^^^^X&J20 30 40 o50 determined to be 2.3mMwith the membrane Fraction number (3ml/tube) fraction. On solubilization by detachment of Fig. 2. Elution Pattern of the Isolated Prosthetic Group the enzyme from the bacterial membrane, the on a Sephadex Column. affinity to the substrate became lower. Such Aconcentrated solution of the isolated prosthetic group (2ml) was applied to a Sephadex G-25 column (1x phenomenawere often seen with other species 180cm) and 3ml fractions were collected. Elution was of membrane enzymes. Judging from the monitored by reading the absorbance at 249nm and severe hydrophobicity of glycerol dehydro- 330nm. genase, as the enzyme might be protected by excess detergent to maintain the molecular structure stable and active in a solution, such i-*or~^] i r Kmvalue variation was not an unexpected ,J :: l\f\'-" results. 0.8 " " / II \ "*°

0.4-I - - / U \-20g Prosthetic group Whenthe membranewas treated with 2mM EDTA, pH 7.5, glycerol dehydrogenase ac- Jo.O8i-n 1 1 1 I r 1 1 1 1^i80% tivity was bleached completely. After removal 0.08-\ - - /I f\ -805 of excess EDTAby centrifugation, the enzyme activity of the EDTA-treated membrane was 0.04-I : - / / \ -40 specifically restored to the original level of the 0.02- \ ' ' I / V2° membraneonly by the addition ofPQQin the oL i^~T--I V- i i A>i i lo presence of magnesium ions. These prelim- 240 380 480 800 220 340 460 580 inary observations suggesting that the glyce- Wavelength (nm) rol dehydrogenase could be a quinoprotein Fig. 3. Absorption Spectra and Fluorescence Spectra of were confirmed by the purified enzymeprepa- the Isolated Prosthetic Group and Authentic PQQ. ration. The purified enzyme (5mg, having a Absorption spectra and fluorescence spectra are shownin specific activity of 50units/mg protein) was the left column and the right column, respectively. Upper treated with methanol to isolate the prosthetic frame, fraction number 40 of the gel filtration shown in group. The gel filtration pattern of the isolated Fig. 2; bottom frame, authentic PQQ. prosthetic group on a Sephadex G-25 column is shown in Fig. 2. These chromatographic shown in Fig. 3, and compared with those of profiles seemed to be similar to those of au- authentic PQQ. These results were the same thentic PQQ. The absorption and fluorescence as the recent results such as polyethylene gly- spectra of fraction number 40 were recorded as col dehydrogenase.28) It was obvious that the 1008 M. Ameyamaet al. spectra were identical with those of authentic tivity was found, while glycerol dehydrogenase PQQ. Thus glycerol dehydrogenase was con- was normally formed. To use this strain for cluded to be a quinoprotein like the other further purification of glycerol dehydrogenase dehydrogenases from acetic acid bacteria such was advantageous due to the less contami- as alcohol, aldehyde and D-glucose dehydro- nation by alcohol dehydrogenase in the glyc- genases.19) At the same time, the capability as erol dehydrogenase preparation. the prosthetic group for D-glucose dehydro- The culture conditions for the bacterium genase was also confirmed with chromato- were also critical for enzyme formation. The graphed fractions 30 to 45 which showed ab- maximalenzymeformation was observed at sorbance at 330nm. The PQQcontents of the late exponential phase where cell growth these fractions were also determined with was almost completed and the pH of the apo-D-glucose dehydrogenase, and 0.5fig culture medium had fallen to 5.2. By monitor- PQQ/ml and 1.5fig PQQ/ml were calculated ing the pH of the culture medium, the cells for fraction numbers 35 and 40, respective- could be harvested at the right time. ly- The next advantage was the use of a favorable detergent for the purpose. A series of DISCUSSION detergents having the dodecyl moiety were available, which had been developed as am- In this study, membrane-bound glycerol de- photeric detergents. Other detergents such a hydrogenase was solubilized from the mem- series of Triton, Tween, Span or Brij which brane of G. industrius IFO 3260 grown on are extensively used for the solubilization glycerol medium. The solubilization and prop- of membraneproteins were not so available erties of many kinds of membrane-bound for the solubilization of glycerol dehydro- dehydrogenases from acetic acid bacteria and genase. other Gram-negative bacteria have been re- Solubilization with dimethyldodecylami- ported.19'29) As to such dehydrogenases, for neoxide and fractionation with polyethylene glycerol dehydrogenase, in spite of its in- glycol 6000 gave 100-fold purification. There dustrial significance, solubilization, purifi- was no criterion to prove the purity of the cation and enzymatic properties have re- finally purified enzyme due to its severe hy- mained to be elucidated. Due to its severe drophobicity. However, it can be reasonably hydrophobicity, solubilization of the enzyme concluded that the enzyme was highly purified, was extremely difficult and no further in- since other enzyme activities were no longer vestigation has been performed so far. The observed when examined. As the enzyme oxi- coexistence of many kinds of membrane- dized many kinds of polyhydroxyl alcohols, it bound enzymes in the bacterial cytoplasmic was necessary to check whether such oxidation membrane and separation of such cosolubi- was catalyzed by a single enzyme or other lized enzymes from glycerol dehydrogenase contaminating species ofenzymes. The enzyme have been critical for the investigation of oxidized polyhydroxyl alcohols but not glycerol dehydrogenase. ethanol, aldehyde, D-glucose, D-fructose, d- First we surveyed the enzyme activities of gluconate or 2-keto-D-gluconate indicating alcohol and glycerol dehydrogenases through- that the enzymepreparation wasat least not out the strains of Gluconobacter. As shown in contaminated by enzymes catalyzing oxidation Table I, both the enzyme activities were as- of these substrates. Thus glycerol dehydro- sayed at the respective pH optima after check- genase was defined as an enzymethat showed a ing with individual strains. Lots of alcohol broad substrate specificity toward many kinds dehydrogenase was found throughout the of polyhydroxyl alcohols including glycerol, d- strains tested. With G. industrius IFO 3260, mannitol, D-sorbitol, D-arabitol, adonitol, pro- exceptionally low alcohol dehydrogenase ac- pylene glycol and m^o-erythritol. The differ- Membrane-boundGlycerol Dehydrogenase 1009 ence from D-sorbitol dehydrogenase from G. here has not been conducted for other well suboxydans21) was also confirmed. Oxidation established PQQ enzymes.17) Because such a of D-mannitol by the enzyme gave D-fruc- chromatographic procedure has only recently tose. The enzyme could be identical with d- been developed in studies on microbial pro- mannitol dehydrogenase (EC 1.1.2.2)3O) that duction of PQQ.18) has not been purified. Recently, Ueda et al.31) reported D-arabitol oxidation by immobilized Acknowledgment. The authors wish to express their thanks to Misses H. Takamatsu, K. Mieno, J. Yano and G. suboxydans and they referred to the enzyme Y. Takasaki, and Mr. T. Kyo, for their technical assis- as D-arabitol dehydrogenase. This enzyme tance in this study. could also be the same as the glycerol dehydrogenase discussed here. In an earlier report REFERENCES by Kersters et al.,26) the contribution of many kinds of pyridine nucleotide dependent en- 1) G. Bertrand, Comp. Rend. Scan. Acad. Sci., 126, 762 (1898). zymes in the cytoplasmic fraction yielding 2) R. E. AsnisandA. F. Brodie, /. Biol. Chem., 203, 153 corresponding products was indicated. How- (1953). ever, in our present sense on the membrane- 3) E. C. C. Lin and B. Magasanik, /. Biol. Chem., 235, bound dehydrogenase localized in the peri- 1820 (1960). plasmic space or outer surface of the cyto- 4) M. Yamada, A. Nagano and Y. Tani, Agric. Biol. plasmic membrane and contributing to the Chem., 44, 471 (1980). 5) H. Yamada, A. Nagano, H. Nishise and Y. Tani, rapid production of various oxidation prod- Agric. Biol. Chem., 46, 2333 (1982). ucts in so high concentrations, the mem- 6) H. Nishise, A. Nagano, Y. Tani and H. Yamada, brane-bound glycerol dehydrogenase should Agric. Biol. Chem., 48, 1603 (1984). be directly catalyzing in the process in dihy- 7) A. Ben-Amots and M. Avron, FEBS Lett., 29, 153 droxyacetone formation and the contribu- (1973). tion of NAD(P)-dependent cytoplasmic en- 8) H. Dutler,J. L. vanderBaan, E. Hochuli, Z. Kis, K. E. Taylor and V. Prelog, Eur. J. Biochem., 75, 423 zymes should be ruled out. The difference (1977). found between their pH optima, the cytoplas- 9) M. Viswanath-Reddy, J. E. Pyle and H. B. Howe, Jr., mic enzymes showed optima at alkaline pH J. Gen. Microbiol, 107, 289 (1978). 10) A. W. Kormann, R. O. Hurar and T. G. Flynn, while the membrane-bound enzyme showed an Biochim. Biophys. Acta, 258, 49 (1972). optimum at acidic pH, and high content of ll) T. E. King and V. H. Cheldelin, /. Biol. Chem., 198, such membrane-bound enzymes, were sup- 127 (1952). porting evidence for the above. On the basis of 12) J. De Ley and R. Dochy, Biochim. Biophys. Acta, 40, its catalytic properties, glycerol dehydrogenase 277 (1960). 13) T. Uwajima and O. Terada, Agric. Biol. Chem., 44, could be available for microdetermination of 2039 (1980). glycerol. Unlike for the other pyridine nucle- 14) S. Yamada, K. Nabe, N. Izuo and I. Chibata, /. otide dependent dehydrogenases, no reverse Ferment. Technol., 57, 215 (1979). reaction occurred on glycerol oxidation by the 15) S. Yamada, K. Nabe, N. Izuo and I. Chibata, J. membrane-bound enzyme as has been report- Ferment. Technol., 57, 221 (1979). 16) K. Nabe, N. Izuo, S. Yamada and I. Chibata, Appl. ed previously.20) Environ. Microbiol., 38, 1056 (1979). Finally, the enzyme was found to be a 17) M. Ameyama, K. Matsushita, Y. Ohno, E. quinoprotein which included PQQas the pros- Shinagawa and O. Adachi, FEBS Lett., 130, 179 thetic group. The isolated prosthetic group (1981). fraction was identical with that obtained from 18) M. Ameyama, M. Hayashi, K. Matsushita, E. Shinagawa and O. Adachi, Agric. Biol. Chem., 48, other species of quinoproteins. In addition to 561 (1984). spectral and enzymatic identification of the 19) M. Ameyama, E. Shinagawa, K. Matsushita and O. prosthetic group with authentic PQQ,chro- Adachi, J. Bacteriol., 145, 814 (1981). matographic identification was performed. 20) M. Ameyama, "Methods in Enzymol," Vol. 89, ed. The chromatographic identification reported by W. A. Wood, Academic Press Inc., New York, 1010 M. Ameyamaet al.

1982, pp. 20-29. Ameyama, Agric. Biol Chem., 46, 135 (1982); E. 21) O. H. Lowry,N.J. Rosebrough,A. L. FarrandR.J. Shinagawa and M. Ameyama, "Methods in Randall, /. Biol. Chem., 193, 265 (1951). Enzymol," Vol. 89, ed. by W. A. Wood, Academic 22) J. R. Dulley and P. A. Grieve, Anal. Biochem., 64, Press Inc., New York, 1982, pp. 141~145. 136 (1975). 28) F. Kawai, H. Yamanaka, M. Ameyama, E. 23) J. T. Cummins, T. E. King and V. H. Cheldelin, J. Shinagawa, K. Matsushita and O. Adachi, Agric. Biol. Chem., 224, 323 (1957). Biol Chern., 49, 1071 (1984). 24) J. T. Cummins, V. H. Cheldelin and T. E. King, /. 29) O. Adachi, Nippon Nogeikagaku Kaishi, 53, R77 Biol. Chem., 226, 301 (1957). (1979). 25) Z. Dische and E. Borenfreund, J. Biol. Chem., 192, 30) A. C. Arcus and N. L. Edson, Biochem. /., 64, 385 583 (1951). (1956). 26) K. Kersters, W. A. Wood and J. De Ley, /. Biol. 31) S. Ohmomo, Y. Tozawa and K. Ueda, /. Ferment. Chem., 240, 965 (1965). TechnoL, 61, 373 (1983). 27) E. Shinagawa, K. Matsushita, O. Adachi and M.