Function of multiple heme c moieties in intramolecular electron transport and ubiquinone reduction in the quinohemoprotein alcohol dehydrogenase-cytochrome c complex of Gluconobacter suboxydans.

Alcohol dehydrogenase (ADH) of acetic acid bacteria functions as the primary dehydrogenase of the ethanol oxidase respiratory chain, where it donates electrons to ubiquinone. ADH is a membrane-bound quinohemoprotein-cytochrome c complex which consists of subunits I (78 kDa), II (48 kDa), and III (14 kDa) and contains several hemes c as well as pyrroloquinoline quinone as prosthetic groups. To understand the role of the heme c moieties in the intramolecular electron transport and the ubiquinone reduction, the ADH complex of Gluconobacter suboxydans was separated into a subunit I/III complex and subunit II, then reconstituted into the complex. The subunit I/III complex, probably subunit I, contained 1 mol each of pyrroloquinoline quinone and heme c and exhibited significant ferricyanide reductase, but no Q1 reductase activities. Subunit II was a triheme cytochrome c and had no enzyme activity, but it enabled the subunit I/III complex to reproduce the Q1 and ferricyanide reductase activities. Hybrid ADH consisting of the subunit I/III complex of G. suboxydans ADH and subunit II of Acetobacter aceti ADH was constructed and it had showed a significant Q1 reductase activity, indicating that subunit II has a ubiquinone-binding site. Inactive ADH from G. suboxydans exhibiting only 10% of the Q1 and ferricyanide reductase activities of the active enzyme has been isolated separately from active ADH (Matsushita, K., Yakushi, T., Takaki, Y., Toyama, H., and Adachi, O (1995) J. Bacteriol. 177, 6552-6559). Using these active and inactive ADHs and also isolated subunit I/III complex, we performed kinetic studies which suggested that ADH contains four ferricyanide-reacting sites, one of which was detected in subunit I and the others in subunit II. One of the three ferricyanide-reacting sites in subunit II was defective in inactive ADH. The ferricyanide-reacting site remained inactive even after alkali treatment of inactive ADH and also after reconstituting the ADH complex from the subunits, in contrast to the restoration of Q1 reductase activity and the other ferricyanide reductase activities. Thus, the data suggested that the heme c in subunit I and two of the three heme c moieties in subunit II are involved in the intramolecular electron transport of ADH into ubiquinone, where one of the two heme c sites may work at, or close to, the ubiquinone-reacting site and another between that and the heme c site in subunit I. The remaining heme c moiety in subunit II may have a function other than the electron transfer from ethanol to ubiquinone in ADH.

Alcohol dehydrogenase (ADH) 1 of acetic acid bacteria, consisting of the genera Acetobacter and Gluconobacter, catalyzes the first step of acetic acid production, oxidation of ethanol to acetaldehyde. ADH is a quinohemoprotein-cytochrome c complex bound to the periplasmic side of the cytoplasmic membrane and functions as the primary dehydrogenase in the ethanol oxidase respiratory chain, where ADH oxidizes ethanol by transferring electrons to ubiquinone embedded in the membrane phospholipids. The resulting ubiquinol is oxidized by terminal ubiquinol oxidase, cytochrome o or a 1 (1). ADH has been purified from five strains and it consists of subunits I, II, and III (1)(2)(3)(4), except for one ADH purified from Acetobacter polyoxogenes which consists only of subunits I and II (5). ADH contains pyrroloquinoline quinone (PQQ) (6) and several heme c moieties in subunits I and II (1). The genes encoding subunits I and II have been cloned and sequenced from several sources including Acetobacter aceti (7,8), A. polyoxogenes (9), and Acetobacter pasteurianus (10). Takeda et al. (11) have also cloned the gene encoding the CO-binding cytochrome c from Gluconobacter suboxydans, which is identical to subunit II of ADH. These genetic data suggest that subunit I is a typical secretory protein with a cleavable signal sequence which has significant homology to the putative PQQ-binding motif found in the methanol dehydrogenase ␣ subunit, and a heme c binding motif, and that subunit II is also a secretory protein with three heme c binding motives.
Coupled with ethanol oxidation, ADH reduces phenazine methosulfate, dichlorophenolindophenol, or ferricyanide as an artificial electron acceptor in vitro (12). Since ferricyanide reacts with heme components having a high redox potential, the heme c sites in the ADH complex should reduce ferricyanide. Furthermore, ADH reacts with several ubiquinone homologues and also with native ubiquinone in proteoliposomes (1). To couple with the reduction of ubiquinone, an electron from ethanol must be transferred inside the ADH complex, where PQQ and several heme c moieties may be involved in the electron transfer and thus in the reduction of ubiquinone. Furthermore, ADH is involved in the CN-insensitive by-pass oxidase system of the G. suboxydans respiratory chain (13,14) and may mediate electron transfer from another primary dehydrogenase, glucose dehydrogenase, to ferricyanide (15). Thus, ADH appears to have several additional functions in vivo, besides the oxidation of ethanol to acetaldehyde.
To understand why there are so many prosthetic groups and how the intramolecular electron is transported in the ADH complex, we separated and reconstituted individual subunits from the ADH complex. In addition, during the course of the investigation, inactive ADH was isolated from G. suboxydans (16), which has at least 10 times lower activity, although there are no differences in the subunit composition or prosthetic groups. Thus, we also studied the kinetic properties of active and inactive ADH and the reactivation of inactive ADH. The results indicated that subunit I of ADH is a quinohemoprotein which contains one molecule each of PQQ and heme c, that subunit II contains three heme c moieties which are responsible for ubiquinone reduction and that the four heme c sites of ADH are separately involved in the various ferricyanide reductase activities of the ADH complex. Furthermore, based on the results obtained in this study, the intramolecular electron transport of the ADH complex to ubiquinone is discussed.

Materials
Monoclonal antibodies against the subunit I of ADH of G. suboxydans were prepared as described (17). Ubiquinone homologues (Q) were supplied by Eizai Co., Tokyo. DEAE-or CM-Toyopearl, which was used as a medium performance ion-exchanger, was from Tosoh Co. (Tokyo). Phenyl-Sepharose and Ampholine (pH 3.5-10.0 for IEF) were purchased from Pharmacia LKB. PQQ was from Wako Chemical Co. (Osaka). An immunoblotting kit and prestained marker proteins were obtained from Bio-Rad. The polyvinylidene difluoride microporous membrane (PVDF) was obtained from Millipore. High performance liquid chromatography marker proteins and pI marker proteins were supplied by Oriental Yeast Co. Ltd. (Osaka). All other materials were of reagent grade and obtained from commercial sources.

Bacterial Strains, Plasmids, and Growth Conditions
The bacterial strains and plasmids used in this study are listed in Table I. These organisms were cultivated at 30°C with rotary shaking (200 rpm). Acetic acid bacteria were maintained on agar slants containing 1.5% agar, 0.5% CaCO 3 , and potato medium (18). Cells maintained on the agar slant were inoculated into 5 ml of the potato medium and shaken for 24 h as seed cultures. G. suboxydans was grown on sugarrich (19) or sorbitol medium (20). Acetobacter species including A. aceti and A. pasteurianus were grown on glycerol medium (18). When necessary, 50 mM potassium phosphate buffer (KPB) was added to adjust the pH of the medium. The seed culture was inoculated into 100 ml of the respective medium in a 500-ml Erlenmeyer flask, which was then shaken for 16 -24 h. For large scale cultivation, 100 ml of the culture was transferred into 1.5 liters of the same medium in a 3-liter Erlenmeyer flask and when necessary, further transferred into 20 liters of the same medium in a 50-liter jar fermentor.
Escherichia coli and Pseudomonas aeruginosa were grown in LB medium. P. aeruginosa was also grown on a minimal medium composed of 4.52 g of KH 2 PO 4 , 11.76 g of K 2 HPO 4 , 3.0 g of (NH 4 ) 2 SO 4 , 0.5 g of MgCl 2 ⅐6H 2 O, 15 mg of CaCl 2 , 15 mg of FeSO 4 ⅐7H 2 O, 7.8 g of CuSO 4 ⅐5H 2 O, 10 g of H 3 BO 3 , 10 g of MnSO 4 ⅐4H 2 O, 125 g of ZnSO 4 ⅐7H 2 O, and 10 g of MnO 3 in 1 liter of distilled water, supplemented with 0.5% (v/v) ethanol or 0.5% (w/v) sodium gluconate as the sole carbon source. Antibiotics when added, were routinely used at the final concentrations as follows: 100 g/ml kanamycin and 25 g/ml tetracycline for acetic acid bacteria and P. aeruginosa; and 50 g/ml kanamycin and 12.5 g/ml tetracycline for E. coli.

Preparation of A. pasteurianus and P. aeruginosa Strains
Harboring the Plasmid Containing adh Gene Plasmids, pAA025 (21) and pRK2013 (22), and A. pasteurianus NP2503 (21) (Table I) were supplied by Dr. Masao Fukuda (Department of Bioengineering, Nagaoka University of Techology). E. coli HB101 was transformed with the plasmids by a standard CaCl 2 procedure (23). The transformants were screened on an LB plate containing tetracycline or kanamycin. The plasmid pAA025 was transferred from E. coli to A. pasteurianus NP2503 or P. aeruginosa IFO 3445 by the triparental mating method using pRK2013 as a helper plasmid (7). The resulting transconjugants were isolated on plates of glycerol medium containing 1% acetic acid or of the minimal medium supplemented with gluconate, respectively, both of which contained tetracycline. The transconjugated strains were termed A. pasteurianus 2503C or P. aeruginosa 3445A. These strains were cultivated in glycerol medium or the minimal medium supplemented with 0.5% ethanol, respectively, and both contained tetracycline.

Preparation of the Membrane Fraction
Cells were harvested by centrifugation at 9,000 ϫ g for 10 min, and washed twice with 50 mM KPB (pH 6.0). The washed cells were resuspended at about 1 g of wet cells per 5 ml of 50 mM KPB (pH 6.0), and passed twice through a French press (American Instrument Co.) at 16,000 psi. After centrifugation at 9,000 ϫ g for 10 min to remove intact cells, the supernatant were ultracentrifuged at 86,000 ϫ g for 90 min to obtain the membrane fraction.
Purification of ADH, Subunit I/III Complex, and Subunit II from G. suboxydans ADH was purified essentially as described (1,2) with some modifications as follows. The membrane fraction was suspended in 10 mM KPB (pH 6.0) at a protein concentration of 20 mg/ml, and Triton X-100 was added to the suspension at a final concentration of 1.0% (w/v). After an incubation at 4°C for 60 min, solubilized ADH was recovered by ultracentrifugation and dialyzed against 5 mM KPB (pH 6.0) containing 0.1% Triton X-100. The dialyzate was applied to a DEAE-Toyopearl column (about 5 mg of protein per 1-ml of bed volume) equilibrated with the same buffer. ADH was eluted with a linear gradient consisting of 5-bed volumes each of 5 and 50 mM KPB (pH 6.0), both of which contained 0.1% Triton X-100. Rose red fractions having ADH activity were eluted at around 20 -30 mM buffer. The active fractions were collected and dialyzed against 5 mM acetate buffer (pH 5.0) containing 0.1% Triton X-100. The dialyzate was applied to a CM-Toyopearl column (about 5 mg of protein per 1-ml bed volume) equilibrated with the same buffer. After washing with 5 mM buffer, the column was further washed with 5-bed volumes of 40 mM acetate buffer (pH 5.0) where a cytochrome c was eluted as a purified subunit II. ADH was eluted with a linear gradient consisting of 5-bed volumes each of 40 and 100 mM acetate buffer (pH 5.0) and an enzyme fraction having relatively high ADH activity was eluted as a purified ADH complex, at the midpoint of the gradient. All buffer systems used until this step contained 0.1% Triton X-100. The detergent was omitted from the system at this point. Another ADH fraction having relatively low enzyme activity was eluted with a linear gradient consisting of 5-bed volumes each of 100 and 200 mM acetate buffer (pH 5.0) followed by 5-bed volumes of 200 mM buffer. The enzyme, subunit I/III complex, was eluted almost at the end of the gradient. When necessary, the latter ADH fraction was further purified as follows. Solid ammonium sulfate was added to the fraction at a final concentration of 30% saturation, and the suspension was applied to a Phenyl-Sepharose column (about 5 mg of protein per 1-ml bed volume) equilibrated with 50 mM KPB (pH 6.0) containing 30% saturated ammonium sulfate and 2 mM CaCl 2 . An orange-colored active fraction was eluted with the same buffer without ammonium sulfate.

Purification of the Inactive ADH of G. suboxydans and A. aceti ADH from A. pasteurianus 2503C Strain
Inactive ADH of G. suboxydans was purified from the membranes of G. suboxydans grown on sugar-rich medium as described (16). ADH of A. aceti was purified from the membranes of A. pasteurianus 2503C strain, harboring pAA025 including the gene encoding ADH subunits I and II of A. aceti, as described (1).

Purification of Subunit II of A. aceti ADH
P. aeruginosa 3445A strain harboring pAA025 was grown to the late-logarithmic phase on minimal medium containing ethanol. The membranes were prepared from about 20 g of wet cells, and suspended in 5 mM KPB (pH 6.0) at a protein concentration of around 10 mg/ml. Triton X-100 was added to the membrane suspension at a final concentration of 1.0% (w/v), followed by standing with stirring at 4°C for 30 min. The solubilized supernatant was recovered by ultracentrifugation at 86,000 ϫ g for 90 min, and dialyzed overnight against 50-fold volumes of 5 mM KPB (pH 6.0) containing 0.1% Triton X-100. The dialyzate was applied to a DEAE-Toyopearl column (about 5 mg of protein per 1-ml of bed volume) which had been equilibrated with the same buffer. The enzyme was eluted with a linear gradient of 3-bed volumes each of 5 and 50 mM KPB (pH 6.0) containing 0.1% Triton X-100, after washing the column with 3-bed volumes of 5 mM buffer containing the detergent. Orange-colored cytochrome was eluted around 40 mM buffer. The fractions containing cytochrome c were collected, concentrated by ultrafiltration using a UP-20 membrane (Advantec Toyo), and dialyzed overnight against 50 volumes of 5 mM acetate buffer (pH 5.0) containing 0.1% Triton X-100. After dialysis and centrifugation (10,000 ϫ g for 10 min), the sample was applied to a CM-Toyopearl column (2.4 mg of protein per 1-ml of bed volume) equilibrated with the same buffer. After washing the column with 5-bed volumes of the buffer containing 0.1% Triton X-100, the cytochrome c was eluted with a linear gradient of 10-bed volumes each of 5 and 100 mM acetate buffer (pH 5.0) containing detergent. Cytochrome c eluted around 20 mM buffer concentrations was collected separately from another cytochrome c (20-kDa protein) successively eluted, and concentrated by ultrafiltration. The sample was dialyzed against 10 mM KPB (pH 6.0) containing 0.1% Triton X-100, and used as subunit II of A. aceti ADH. It contained 3.2 M heme c with about 30% impurities.

Preparation of Polyclonal Antibodies Raised against CO-binding Cytochrome c of G. suboxydans and ADH of A. aceti
CO-binding cytochrome c 553 from G. suboxydans and ADH of A. aceti were purified as described (1,24). Antibodies against both proteins were prepared as follows. About 1 mg of the cytochrome or ADH was injected subcutaneously into a rabbit after emulsifying with an equal volume of complete adjuvant. One month later, about 0.5 mg of the cytochrome or ADH was mixed with an equal volume of incomplete adjuvant and the rabbit was given a booster injection. Another 10 days later, about 30 ml of blood was collected from the rabbit, left at room temperature for 4 h and centrifuged at 3,000 ϫ g for 15 min to remove red cells. The supernatant containing the antiserum was used as a polyclonal antibody for CO-binding cytochrome c or the ADH of A. aceti.

Enzyme Assays
Ferricyanide reductase activity of ADH was measured colorimetrically using potassium ferricyanide as an electron acceptor as described (12,16). Ferricyanide reductase activity was also measured spectrophotometrically in the reaction mixture (1 ml) containing buffer, potassium ferricyanide, and enzyme solution. The reaction was started by adding 10 mM ethanol and the absorbance at 417 nm was followed. Enzyme activity was defined as the amount of enzyme oxidizing 1 mol of substrate per min, calculated from a millimolar extinction coefficient of potassium ferricyanide of 1.0 mM Ϫ1 . Q 1 reductase activity of ADH was measured spectrophotometrically by following the decrease of absorbance at 275 nm at 25°C in a reaction mixture (1 ml) consisting of appropriate amounts of enzyme, 10 mM ethanol, 50 M Q 1 , and McIlvaine buffer (pH 4.5), as described (1). One unit of these activities was defined as the amount of enzyme oxidizing 1 mol of ethanol per min.
Analysis of N-terminal Sequence-Purified ADH was applied to SDS-PAGE using glycine-free buffers 200 mM Tris-HCl (pH 8.5) and 100 mM Tris-Tricine containing 0.1% SDS (pH 8.25) as the anode and cathode buffers, respectively. The proteins in the gel were transferred electrophoretically onto a PVDF membrane in 10 mM CAPS-NaOH buffer (pH 11.0) containing 10% methanol for 9 h at 25 to 50 mA. The transferred membrane was stained with Coomassie Brilliant Blue R-250 and destained with 50% methanol. The visible bands were excised and applied to a peptide sequence analyzer (Shimadzu).
Measurement of PQQ Content-Purified enzyme solution was mixed with 9 volumes of methanol. After incubation at 25°C for 30 min, this solution was centrifuged at 3,000 ϫ g for 10 min. The supernatant was evaporated and used as PQQ extract. The PQQ content was measured enzymatically using the membrane fraction containing apo-glucose dehydrogenase from E. coli K12 strain as described (27).
Heme c Contents-Heme was measured from dithionite-reduced minus ferricyanide-oxidized difference spectrum of its pyridine hemochrome with a dual-wavelength spectrophotometer. The pyridine hemochrome was prepared by mixing the sample with a final concentrations of 20% (v/v) pyridine and 0.2 N NaOH. The heme content was calculated by a millimolar extinction coefficient of 24.3 (549 -535 nm).
Protein Content-The protein content was determined by the modified Lowry method (28). Bovine serum albumin was used as the standard protein.

Isolation of Subunit I/III
Complex and Subunit II of ADH from G. suboxydans-ADH purified by DEAE-Toyopearl column was dialyzed against pH 5 buffer, then applied to a CM-Toyopearl column to separate three cytochrome c fractions (Fig.  1). The first cytochrome fraction that eluted at 40 mM buffer had CO-binding ability, according to the CO-reduced minus reduced difference spectrum (data not shown), but no ADH activity. The second fraction that eluted at 70 -80 mM buffer showed a relatively high ADH activity. The third fraction that eluted at 200 mM buffer had lower ADH activity. As shown in SDS-PAGE (Fig. 2), these fractions consisted of a single peptide of 48 kDa, three bands of 78, 48, and 14 kDa, and two peptides of 78 and 14 kDa. Thus, although the principle of this separation was unclear, the ADH complex was separated into three fractions in the CM-Toyopearl column chromatography. The second fraction was the same as the reported native ADH complex (2), and the first and third fractions seemed to correspond to subunit II and subunit I/III complex of ADH, respectively.
CO-binding cytochrome c 553 has been purified from G. suboxydans where the cytochrome can be solubilized from the membrane with 0.2% Triton X-100 and separated from ADH by CM-cellulose column chromatography (24). Since the first cytochrome c fraction exhibited CO-binding ability and almost the same molecular weight and heme c contents as the cytochrome (see below), it seems to be similar to the CO-binding cytochrome c 553 . Therefore, the immunocross-reactivity of the first cytochrome c fraction with the antibody raised against the CO-binding cytochrome c 553 was examined by immunoblotting (Fig. 2). The antibody cross-reacted at the same intensity with the cytochrome of the first fraction and also with the second subunit in ADH complex of the second fraction but not with the third fraction. Immunoblotting confirmed that the third fraction contained the subunit I present in the ADH complex (Fig.  2). Thus, it was shown that the ADH complex can be separated into subunit II, which is identical to the CO-binding cytochrome c 553 , ADH complex, and subunit I/III complex by CM-Toyopearl column chromatography. The pI values of the ADH complex, subunit I/III complex, and subunits I and II were also determined by isoelectrofocusing to be 5.1, 5.3 (5.5 in the apoform), 6.4, and 4.7, respectively.
Characterization of Subunit I/III Complex and Subunit II of G. suboxydans ADH-The contents of the prosthetic groups, PQQ and heme c, in ADH complex, subunit I/III complex, and subunit II were determined (Table II), and their contents were estimated based on their relative molecular masses of 140, 92, and 48 kDa with the ADH complex, subunit I/III complex, and subunit II, respectively. The ADH and subunit I/III complexes contained about 0.6 mol of PQQ per mol. Heme c was present at 3.5, 0.74 and 2.5 mol/mol of the ADH complex, subunit I/III complex, and subunit II, respectively. In addition to subunit II, as shown by heme-stained SDS-PAGE (Fig. 2), the heme c moiety was detected in subunit I but not in subunit III. The absorption spectra of these fractions are shown in Fig. 3. The ADH and subunit I/III complexes were completely reduced whereas subunit II was oxidized. ADH complex exhibited the same absorption spectrum as the purified ADH (2), having ␣, ␤, and ␥ peaks of 553, 522, and 417 nm. The absorption spectra of subunit II was similar to that of CO-binding cytochrome c 553 (24), which exhibited ␣, ␤, and ␥ peaks of 553, 522, and 418 nm, respectively, in the reduced state and a ␥ peak of 410 nm in the oxidized form. The subunit I/III complex exhibited absorption peaks at 551, 522, and 416 nm.
Although the N-terminal amino acids of subunits I and II were blocked by some modifications and thus could not be determined, the N-terminal amino acid sequence of subunit III was determined without deblocking, to be Gln-Asp-Gln-Leu-Gly-Ala-Pro-Val-Gly.
Reconstitution of ADH Activity from the Separated Subunits-The first fraction, subunit II, did not exhibit any ADH activity, while the third fraction, subunit I/III complex, showed a relatively weak ADH activity of around 100 units/mg at pH 5.0. In contrast to the ADH complex acting at a broad pH range from acidic to neutral pH, the subunit I/III complex exhibited ADH activity only at acidic pH (Fig. 4). Notably, the subunit I/III complex showed no Q 1 reductase activity although it had ferricyanide reductase activity (Table II).
ADH complex was reconstituted from the isolated subunits by mixing subunit I/III complex and subunit II in 10 mM KPB (pH 6.0) containing 0.1% Triton X-100 and incubating it at 25°C for 20 min. ADH activities, ferricyanide reductase activities at pH 5.0 and pH 7.0, and Q 1 reductase activity at pH 5.0, of subunit I/III complex were titrated with subunit II, in which the enzyme activities were measured following holoenzyme formation with both PQQ and Ca 2ϩ . As the added subunit II was increased, ferricyanide reductase activity increased slightly at pH 5.0 and drastically at pH 7.0 and most importantly, ubiquinone reductase activity was recovered to almost the same level as that of the native ADH complex (Fig. 5). In the reconstitution experiments, the enzyme activity of the reconstituted ADH seemed to reflect that of the subunit I/III complex used. Since subunit I/III complex was so unstable that the activity was difficult to maintain constantly during storage, the activity of the reconstituted ADH varied largely among experiments even if the holoenzyme was formed with PQQ (see Fig. 4 and Fig. 5). Nonetheless, the ratio between Q 1 reductase activity and ferricyanide reductase activity at pH 7.0 of the reconstituted enzyme was constant through the study. When the molar ratio was calculated based on the heme contents of the subunits where subunit I/III complex and subunit II were estimated to contain 1 and 3 mol of heme c, respectively, the activities were saturated with 0.5-1.0 mol of subunit II per mol of subunit I/III complex. In high performance liquid chromatography gel filtration (data not shown), the reconstituted enzyme was eluted at the same position as the native ADH. This was faster than subunit I/III complex, suggesting that subunit II binds with subunit I/III complex at an equimolar ratio to form the ADH complex. Considering that the reconstituted enzyme consisted of a one to one ratio of both subunits, it seems that the reconstituted activity can also be saturated at a ratio of roughly 1 mol of subunit II per 1 mol of subunit I/III complex. One specific ferricyanide reductase activity of the native ADH, which functions at acidic to neutral pH regions, was not functional in the reconstituted ADH (Fig. 4). This also shows the pH profiles of the ferricyanide reductase activities of the reconstituted ADH.
Construction of Hybrid ADH from Subunit I/III Complex of

FIG. 1. Dissociation of ADH complex into subunit II and subunit I/III complex in CM-Toyopearl column chromatography.
The supernatant (245 mg of protein) solubilized with 1.0% Triton X-100 (TX) was applied to a 50-ml of DEAE-Toyopearl column and ADH was eluted with a linear gradient as described under "Experimental Procedures." Fractions exhibiting ADH activity (27 mg of protein) were pooled and dialyzed as described under "Experimental Procedures," then applied to a 7-ml CM-Toyopearl column. The enzymes were eluted in 40 mM acetate buffer (pH 5.0) containing 0.1% Triton X-100, then by a linear gradient from 40 to 100 mM acetate buffer (pH 5.0) containing 0.1% Triton X-100 and by another linear gradient from 100 to 200 mM acetate buffer (pH 5.0) without the detergent. ADH activity (⌬) was measured by ferricyanide reductase assay (pH 5.0). Elution of the protein (E) and cytochrome (q) was measured at 290 and 420 nm, respectively. At the point indicated as Tx-free, the buffer system was exchanged to that excluding the detergent.

G. suboxydans ADH and Subunit II of A. aceti ADH-
The affinity for Q 1 between ADH from G. suboxydans and that from A. aceti IFO 3284 largely differs (1). Therefore, if a hybrid ADH can be prepared from the subunits of both strains, the subunit containing Q-site could be identified. Since plasmid pAA025 encodes the genes for subunits I and II, but not subunit III, of ADH of A. aceti K6033 (21), transformants harboring this plasmid may produce whole ADH complex or part of the subunits   3. Absorption spectra of subunit II, subunit I/III complex, and ADH purified from G. suboxydans. Triton X-100 included in ADH and subunit II was depleted as described (1). First, each spectrum (broken lines) was taken with subunit II (0.3 mg/ml), subunit I/III complex (0.58 mg/ml), and ADH (0.24 mg/ml), then taken again after adding a few grains of borohydride (solid lines).

FIG. 4. The pH profiles for the ferricyanide reductase activities of ADH, subunit I/III complex, and the reconstituted ADH.
Ferricyanide reductase activity was measured in McIlvaine buffer at pH 3.5 to 8.0 using active ADH (A) subunit I/III and reconstituted ADH complexes (B) as described under "Experimental Procedures." The reconstituted ADH was prepared by mixing subunit II and subunit I/III complex at a heme c ratio of 3 (mol/mol) as described under "Experimental Procedures." In panel A, the thin lines indicate the ideal values of four ferricyanide-reacting sites, and the respective numbers correspond to those mentioned under "Discussion." In panel B, the ferricyanide reductase activity of subunit I/III complex (triangles) and the reconstituted ADH complex (circles) is indicated. and thus may be useful for the purpose described above. As shown in Fig. 6, when this plasmid was transconjugated into the ADH-deficient strain, A. pasteurianus NP2503, the transconjugant A. pasteurianus 2503C, produced whole ADH complex, probably because the mutant strain retains the ability to produce subunit III, but not subunits I and II. On the other hand, when the transconjugant, P. aeruginosa 3445A, was prepared with the same plasmid, the strain produced only subunit II of A. aceti ADH. Although the reason for this is not yet clear, the host strain, P. aeruginosa, may not have any genes for the ADH of acetic acid bacteria and thus subunit I of ADH encoded in the plasmid might not be produced properly without subunit III, which is not present in the plasmid. Thus, the ADH complex and subunit II of A. aceti K6033 were purified from the membranes of these transconjugants, A. pasteurianus 2503C and P. aeruginosa 3445A, respectively.
To construct a hybrid ADH, we attempted to reconstitute ADH from subunit I/III complex of G. suboxydans ADH with subunit II of A. aceti ADH, and the kinetics for Q 1 reductase activity were compared with those of whole ADH complexes of A. aceti K6033 and G. suboxydans. When the subunit I/III complex was titrated with the subunit II, ADH activity was gradually increased but not saturated, even when excess subunit II was added to the subunit I/III complex (data not shown). This implies that affinity of the interaction between subunit I/III complex and subunit II from different origins is not so high. Importantly, however, Q 1 reductase activity could also be reproduced in the "hybrid ADH" as well as the ferricyanide reductase activities at pH 5.0 and 7.0. Thus, kinetics of Q 1 reductase activity can be compared between native complex and hybrid ADH complex (Table III). Affinity for Q 1 of ADH from G. suboxydans was high (K m ; 32-40 M) while that of native ADH from A. aceti was relatively low (K m ; 204 M). The K m value for Q 1 of the hybrid ADH (205 M) was comparable to that of A. aceti native ADH. Thus, the results suggested that the ubiquinone-binding site of ADH is present in subunit II of ADH.
Kinetic Characterization in the Subunits of Active and Inactive ADHs-An inactive ADH has been detected and purified, separate from the active (native) enzyme, in the membranes of the cells grown on acidic pH, and it has enzyme activities that are 10 times lower than those of active ADH (16). Like active ADH as described above, inactive ADH was also partially dissociated into the subunit I/III complex and subunit II. Although the subunit I/III complex from inactive ADH exhibited less ferricyanide reductase activity, it was re-activated by holoenzyme formation with PQQ and Ca 2ϩ to the level with the subunits obtained from active ADH. Thus, the K m values for electron acceptors, ferricyanide and Q 1 , were determined and compared with active and inactive ADHs, and also with the subunit I/III complex derived from inactive ADH (Table IV). When ferricyanide reductase activity was measured at pH 5.0 and 7.0, active ADH exhibited two significantly distinct K m values for ferricyanide at either pH, 0.09 and 0.40 mM at pH 5, and 0.47 and 4.5 mM at pH 7. Two of these K m values (0.40 and 0.47 mM) seemed to be identical. On the other hand, inactive ADH exhibited only one K m value at both pH values which were almost the same as one of two of the K m values for active ADH: only the low value (0.09 mM) at pH 5.0 and only the high value (more than 2 mM) at pH 7.0, in which the saturation curve became sigmoidal against ferricyanide concentrations so that V max could not be obtained. Furthermore, although subunit I/III complex exhibits ferricyanide reductase activity only at pH 5, the K m value for ferricyanide was also the same as that of ADH complex. In addition to these activities, an additional K m value for ferricyanide was detected with ADH complex at pH 3.5, in which the K m value with active ADH was below 20 M. The value is so high that a real K m value could not be  1 and 2), of P. aeruginosa 3445A grown on ethanol-minimal medium (lanes 3 and 4), and of the same strain on LB medium (lanes 5 and 6). Lanes 1, 3, and 5 contain membrane fractions (50 g of protein each) and lanes 2, 4, and 6 contain soluble fractions (50 g of protein each). Lane A contains purified G. suboxydans ADH (1.5 g of protein). determined from the usual steady state kinetics. Although the K m value at pH 3.5 was not determined with inactive ADH, the enzyme also had this high affinity site that reacted with ferricyanide, judging from the pH profile (see Fig. 7). On the other hand, both active and inactive ADHs exhibited the same K m value for Q 1 . Thus, it was suggested that inactive ADH was defective in one of the ferricyanide-reacting sites, which is a middle affinity (K m ; ϳ0.4 mM) site working at acidic to neutral pH regions with the active ADH, although it has a normal ubiquinone-reacting site. Incubation of enzyme with alkali conditions causes a conformational change in inactive ADH (16). As shown in Fig. 7, the alkali treatment restored several enzyme activities of inactive ADH. Ferricyanide reductase activity at neutral pH regions was restored to about 80% of the activity of the active ADH, while only 50% of the ferricyanide reductase activity was restored around pH 5. Thus, inactive ADH could not restore one of the ferricyanide reductase activities detected in active ADH even after exposure to alkali. On the other hand, the Q 1 reductase activity of inactive ADH was almost completely restored to ϳ90% of the activity of the active ADH by the same procedure. DISCUSSION ADH of acetic acid bacteria is a highly sophisticated enzyme complex composed of subunits I (78 kDa), II (48 kDa), and III (14 kDa). In this study, from the ADH of G. suboxydans, subunit I was isolated as a complex with subunit III, and subunit II was isolated as a free form. The subunit I/III complex exhibited ferricyanide reductase activity only at acidic pH but not Q 1 reductase activity, whereas subunit II had no activity. The electron flow of ADH from ethanol to ubiquinone was reproduced by reconstituting subunit I/III complex with subunit II, indicating that subunit I/III complex, probably subunit I, is responsible for the dehydrogenation of ethanol. By sequence homology with the methanol dehydrogenase of methylotrophs (29,30) and the alcohol dehydrogenase of Comamonas testosteroni, 2 as well as by the presence of a heme c-binding motif in their amino acid sequences, 2 subunit I of ADH complex should have PQQ and heme c as the prosthetic group. Actually, this study showed that subunit I/III complex contained 1 mol each of PQQ and heme c and functioned as the dehydrogenase. Thus, subunit I can be classified as a quinohemoprotein ADH termed type II ADH (31), which includes ADHs from C. testosteroni (32), Pseudomonas putida (ADHs IIB and IIG; 26), and Rhodopseudomonas acidophila (33), as well as polyvinyl alcohol dehydrogenase from Pseudomonas sp. VM15C, 3 all of which have 1 mol each of PQQ and heme c and a relative molecular mass of around 70 kDa. Subunit II of ADH was shown to be identical to cytochrome c 553 isolated from the membranes of G. suboxydans (24), which had been thought to contain 2 mol of heme. However, the amino acid sequence of the cytochrome c deduced from the DNA sequence has suggested that there are three heme c-binding motives (11). The heme determination of the purified subunit II or ADH in this study actually showed that subunit II contained three heme c moieties. This notion has also been confirmed by redox titration with subunit II, which shows the cytochrome c behaving as three one-electron carriers. 4 Thus it can be concluded that the ADH complex contains a total of four heme c moieties, one in subunit I and three in subunit II.
Data obtained using active and inactive ADHs and the isolated subunit I/III complex in this study indicate that these four heme c moieties in the ADH complex can be distinguished by their kinetic differences with ferricyanide, since four specific ferricyanide-reacting sites were detected. The first site functions with high affinity at acidic pH, the second with low affinity at neutral pH, the third with extremely high affinity at acidic pH, and the fourth with middle affinity over a range of pH. Since the first ferricyanide-reacting site (high affinity at acidic pH) was detected even in subunit I/III complex ( Fig. 4 and Table IV), it may be located at the heme c site in subunit I and termed heme c site I (see Fig. 4). Thus, the other three ferricyanide-reacting sites should locate at or near one of the three heme c moieties in subunit II, in which the second, third, and fourth ferricyanide-reacting sites are tentatively termed inactive (E) ADHs and the alkali-treated inactive ADH (q), as described under "Experimental Procedures." Right panel, ferricyanide (ferri at pH 5 and 7) and Q 1 reductase activities were also measured with active and inactive ADHs and the alkali-treated inactive ADH (alkali inactive), as described under "Experimental Procedures."

TABLE IV Kinetic parameters of active and inactive ADHs and subunit I/III complex
The K m values (mM) for ferricyanide were determined spectrophotometrically (for sites I, II 2 , and II 3 ) or colorimetrically (for II 1 site), and those for Q 1 were spectrophotometrically, as described under "Experimental Procedures." The ferricyanide-reacting sites, I, II 1 , II 2 , and II 3 , correspond to those in Fig. 4 and are also described under "Discussion." Ferricyanide Q 1 , pH 5.0 pH 3.5, II 2 pH 5.0 pH 7.0 heme c sites II 1 , II 2 , and II 3 , respectively (see Fig. 4). Inactive ADH and also the reconstituted ADH complex may lack one of the ferricyanide-reacting sites, namely the fourth site with middle affinity working at broad pH regions, the II 3 site. One of the heme c moieties in inactive ADH remains oxidized and is not reduced with ethanol, although the individual subunits seemingly remain intact (16). Inactive ADH can be activated by alkali treatment, where, despite the Q 1 reductase activity being almost completely recovered, the fourth ferricyanide-reacting site, II 3 , remained unrecovered. This is consistent with the notion that the oxidized heme c moiety of inactive ADH remains oxidized after exposure to alkali (16). Thus, these data suggested that the ubiquinone reductase activity of ADH can function properly irrespective of whether the fourth ferricyanide site works or not. Therefore, the heme c site II 3 would not be functioning in the pathway of electron transport from ethanol to ubiquinone within the ADH complex. Thus other heme c moieties (I, II 1 , and II 2 ) should function for intraand inter-subunit electron transport within subunits I or II. Furthermore, this study showed that inactive ADH, except for missing one ferricyanide-reacting site, kept the same K m values for ferricyanide and also for Q 1 as active ADH. Although inactive ADH has an electron transfer rate of only 10% of active ADH (16), the electrons from ethanol at the PQQ site in subunit I should be effectively extracted in inactive ADH and thus even in subunit I/III complex alone, like active ADH. Thus, we speculate that in inactive ADH, an improper interaction between subunit II and subunit I/III complex impairs efficient intersubunit electron transport in the ADH complex.
This study also showed that Q 1 reductase activity can be reproduced by reconstituting subunit II to the subunit I/III complex and furthermore, its kinetics for Q 1 in a hybrid reconstituted ADH complex reflected the feature of the original ADH from which subunit II was derived. These results indicated that the ubiquinone-reacting site of ADH is located in subunit II. The ubiquinone site would be very close to either the second or third ferricyanide-reacting sites (II 1 or II 2 site) since three heme c sites (I, II 1 , and II 2 ) may be involved in the electron transport to ubiquinone in ADH as described above and sites II 1 and II 2 are present in subunit II. It cannot be determined at this moment, which should be the actual site or close to the ubiquinone-reacting site, because we could not obtain any evidence indicating a relationship between the ubiquinone-reacting site and the ferricyanide-reacting sites in the ADH of G. suboxydans. Thus, to understand whether the II 1 or the II 2 site is related to the ubiquinone-reacting site, we are searching for some specific inhibitors of Q 1 reductase activity and also the ferricyanide reductase activity of G. suboxydans ADH.
Thus, we speculate that electrons extracted from ethanol at the PQQ site may be transferred via heme c site I in subunit I to either heme c site II 1 or II 2 in subunit II, then to the ubiquinone site, which may also be at or near either of heme c sites II 1 or II 2 . If so, the physiological function of the heme c site II 3 in subunit II remains to be elucidated. The respiratory chain of G. suboxydans branches at the site of ubiquinone, with CN-sensitive terminal oxidase and -insensitive by-pass oxidase, of which the former is cytochrome o (19) and the latter may be constituted at least partly with subunit II of ADH (13,14,34) which may connect the quinone pool to the by-pass oxidase (15). We found that ADH can oxidize ubiquinol and the ubiquinol-ferricyanide oxidoreductase activity works at somewhere other than the ubiquinone-reacting site, but which has similar affinity to ferricyanide as the II 3 site. 4 Thus, the II 3 site may be involved in the electron transport from ubiquinol to the CN-insensitive by-pass oxidase independent of the intramolecular electron transport from ethanol to ubiquinone.