Degradation of Acetaldehyde and Its Precursors by Pelobacter carbinolicus and P. acetylenicus

Pelobacter carbinolicus and P. acetylenicus oxidize ethanol in syntrophic cooperation with methanogens. Cocultures with Methanospirillum hungatei served as model systems for the elucidation of syntrophic ethanol oxidation previously done with the lost “Methanobacillus omelianskii” coculture. During growth on ethanol, both Pelobacter species exhibited NAD+-dependent alcohol dehydrogenase activity. Two different acetaldehyde-oxidizing activities were found: a benzyl viologen-reducing enzyme forming acetate, and a NAD+-reducing enzyme forming acetyl-CoA. Both species synthesized ATP from acetyl-CoA via acetyl phosphate. Comparative 2D-PAGE of ethanol-grown P. carbinolicus revealed enhanced expression of tungsten-dependent acetaldehyde: ferredoxin oxidoreductases and formate dehydrogenase. Tungsten limitation resulted in slower growth and the expression of a molybdenum-dependent isoenzyme. Putative comproportionating hydrogenases and formate dehydrogenase were expressed constitutively and are probably involved in interspecies electron transfer. In ethanol-grown cocultures, the maximum hydrogen partial pressure was about 1,000 Pa (1 mM) while 2 mM formate was produced. The redox potentials of hydrogen and formate released during ethanol oxidation were calculated to be EH2 = -358±12 mV and EHCOOH = -366±19 mV, respectively. Hydrogen and formate formation and degradation further proved that both carriers contributed to interspecies electron transfer. The maximum Gibbs free energy that the Pelobacter species could exploit during growth on ethanol was −35 to −28 kJ per mol ethanol. Both species could be cultivated axenically on acetaldehyde, yielding energy from its disproportionation to ethanol and acetate. Syntrophic cocultures grown on acetoin revealed a two-phase degradation: first acetoin degradation to acetate and ethanol without involvement of the methanogenic partner, and subsequent syntrophic ethanol oxidation. Protein expression and activity patterns of both Pelobacter spp. grown with the named substrates were highly similar suggesting that both share the same steps in ethanol and acetalydehyde metabolism. The early assumption that acetaldehyde is a central intermediate in Pelobacter metabolism was now proven biochemically.

Pelobacter spp. grown with the named substrates were highly similar suggesting that both share the same steps in ethanol and acetalydehyde metabolism. The early assumption that acetaldehyde is a central intermediate in Pelobacter metabolism was now proven biochemically.
either Methanospirillum hungatei or Methanobrevibacter arboriphilus, were studied to understand the biochemistry of syntrophic ethanol oxidation. In this context, the genome of P. carbinolicus revealed interesting gene loci: a gene cluster for comproportionating hydrogenases (gene loci Pcar_1602-Pcar_1605 and Pcar_1633-Pcar_1936) as described previously for Thermotoga maritima [25], and various acetylating or non-acetylating acetaldehyde dehydrogenases. Some of these putative non-acetylating acetaldehyde: ferredoxin oxidoreductases were annotated as either molybdenum-or tungsten-dependent enzymes, and their possible involvement in ethanol metabolism remained unclear.
In the present study, we determined the activities of the cytoplasmic enzymes that are supposed to be involved in ethanol degradation by P. acetylenicus and P. carbinolicus, following the original metabolic concept of M. omelianskii [26,27] and subsequent studies [28][29][30]. Since initial experiments indicated that growth of Pelobacter species/M. hungatei cocultures was affected by tungsten and molybdenum availability the dependence of activity and proteome patterns on these trace metals were studied. We inventoried the most abundant soluble enzymes by 2D PAGE and peptide mass-fingerprinting and identified key players in the metabolism by native activity staining. The relative importance of either hydrogen or formate as syntrophic electron shuttles is still a matter of debate. Highly sensitive detection methods allowed to quantify both potential electron carriers in comprehensive cultivation experiments of both Pelobacter species grown on acetoin or ethanol. Finally, the simultaneously measured concentration of educts and products of ethanol oxidation helped to elucidate the energetics of ethanol oxidation.

Enzyme activities
The specific enzyme activities detected in both P. acetylenicus and P. carbinolicus after syntrophic growth with ethanol were highly similar (Table 1). Cytoplasmic protein fractions oxidized ethanol with NAD + to acetaldehyde. This alcohol dehydrogenase activity increased at higher pH or if assayed in the (thermodynamically favored) reductive direction. Acetaldehyde was oxidized further to acetyl-CoA by a coenzyme A-dependent acetylating acetaldehyde dehydrogenase activity. Cytoplasmic fractions also exhibited phosphotransacetylase and acetate kinase activity. These enzymes converted acetyl-CoA to equimolar amounts of acetate and ATP via substrate level phosphorylation. In addition to this reaction sequence, a non-acetylating acetaldehyde dehydrogenase (acetaldehyde: benzyl viologen oxidoreductase) activity was detected as well. This enzyme oxidizes acetaldehyde directly to acetate yielding an electron pair of low redox potential able to reduce benzyl viologen (BV) or methyl viologen (MV). NAD + was not a suitable electron acceptor for this reaction.
In bacteria with syntrophic lifestyle, hydrogenase or formate dehydrogenase activities are expected to release electron carriers such as H 2 and/or formate to the syntrophic partners. No hydrogenase and negligible formate dehydrogenase activity with NAD + as electron acceptor was found in either Pelobacter species. However, hydrogen or formate reduced benzyl viologen and methyl viologen at high rates. If P. carbinolicus was cocultivated with the hydrogen-only scavenging Methanobrevibacter arboriphilus [31,32], formate dehydrogenase activity of P. carbinolicus was 100-fold lower than after cocultivation with M. hungatei.
In order to identify enzymes that were specifically induced after growth with ethanol, enzymes were also assayed in P. carbinolicus cells after axenic growth on acetoin. The pathway of acetoin degradation had been worked out in the past for this bacterium [7,8]. Activities of alcohol dehydrogenase, acetylating acetaldehyde dehydrogenase and phosphotransacetylase were detected. Additionally, significantly high activities of non-acetylating acetaldehyde dehydrogenase (acetaldehyde: benzyl viologen oxidoreductase) and hydrogenase activity were found although both are not necessary for acetoin degradation. The activity of the initial enzyme of the acetoin pathway, acetoin: DCPIP oxidoreductase [8], was high after growth with acetoin and far lower after growth with ethanol. For the first time, P. carbinolicus was grown axenically on acetaldehyde and enzyme activities were assayed. Acetoin: DCPIP oxidoreductase was still active since the inoculum had been pre-cultivated on acetoin. Alcohol dehydrogenase was also active whereas acetylating acetaldehyde dehydrogenase, the key enzyme of substrate level phosphorylation, exhibited only minor activity. Instead, a nonacetylating acetaldehyde dehydrogenase activity higher than in acetoin-grown cells was detected. The enzyme systems necessary for syntrophic electron transfer, hydrogenase and formate dehydrogenase, were also present after growth with acetaldehyde.
Initial growth experiments (see S1 Fig.) revealed that molybdenum and tungsten are essential trace elements for the P. acetylenicus/M. hungatei coculture. In the absence of both trace elements, growth was extremely slow. Cocultures grew best if 100 nM tungstate but no molybdenum was added. Cultures supplied with 150 nM molybdate but no tungstate grew slower during the exponential growth phase than tungstate-supplied cultures. In particular non-acetylating acetaldehyde dehydrogenases and formate dehydrogenases are known to depend on tungsten and/or molybdenum-containing cofactors [33][34][35][36]. Thus, cytoplasmic fractions of P. carbinolicus cells grown on ethanol with either 150 nM molybdate or 300 nM tungstate were assayed for these activities ( Table 2). Both activities were largely increased after growth in tungstate-rich medium. Since the non-acetylating acetaldehyde dehydrogenase activity in tungstate-free medium was still high, molybdenum might be a suitable substitute for tungsten although the tungsten-dependent enzymes appeared to be more active. On the contrary, the formate dehydrogenase activity was tungstate-induced to a larger extent (100-to 2700-fold), indicating that tungsten is essential for this enzyme and cannot easily be substituted by molybdenum. Noteworthy, the hydrogenase activity tested as a molybdenum-or tungsten-independent reference activity was the same under both growth conditions.

Soluble proteome analysis and activity staining
A possible presence of two types of acetaldehyde dehydrogenases, either molybdenum-or tungsten-dependent, has been discussed in the past [33,37]. Therefore, we wanted to track down differences in protein expression patterns during growth on ethanol with the named trace element concentrations. As depicted in Fig. 1, simple SDS PAGE of soluble protein extract mainly differed in the expression of two bands which were identified as molybdenum-dependent (Pcar_0220, at 120 kDa, Fig. 1 A) or tungsten-dependent acetaldehyde dehydrogenases (Pcar_0665 or Pcar_0456, at 65 kDa, Fig. 1 B), respectively. Pcar_0665 and Pcar_0456 share 75.4% sequence identity and are likely to carry the same physiological function. Protein expression in standard medium (12 nM W plus 150 nM Mo) was similar to tungstate-rich medium without molybdenum (see S2 Fig.).
A differential analysis of 2D gels in Fig. 2 confirmed the overexpression of both tungsten-dependent isoforms in tungstate-rich media. After growth with only molybdate the large Pcar_0220 was expressed. The spot size was smaller than expected because, according to our experience, this PAGE method discriminates proteins larger than 80 kDa. Besides non-acetylating acetaldehyde dehydrogenase enzymes, 2D PAGE also made putative subunits of a formate dehydrogenase gene cluster (Pcar_0833-0835) detectable which was less abundant. Furthermore, an additional acetylating acetaldehyde dehydogenase (Pcar_2758) was expressed in tungstate-rich medium.
Besides the differentially expressed proteins, 2D PAGE revealed highly abundant and constitutively expressed proteins such as alcohol dehydrogenases (Pcar_0251 and Pcar 0255), an acetylating acetaldehyde dehydrogenase (Pcar_1246), a glutamine synthetase (Pcar_1501, likely responsible for ammonia uptake) and two subunits of a putative hydrogenase (Pcar_1633 and Pcar_1634).
To link activity and protein identification, we applied our recently described activity staining method [38] to cytoplasmic fractions of ethanol-grown P. carbinolicus cells (see S3 Fig., Table S1). In all experiments, several spots or lanes were stained per gel strip, indicating that the native protein complexes decayed to varying extent. However, all staining reactions finally led to the same gene locus, confirming the PAGE results and the annotation of the non-acetylating acetaldehyde dehydrogenase (Pcar_0220 with molybdate and Pcar_0665 or Pcar_0456 with tungstate), hydrogenase (Pcar_1633/Pcar_1635 or Pcar_1604/ Pcar_1605), and formate dehydrogenase (likely Pcar_0834/Pcar_0835). This links the proteome analyses to the viologen-staining activities determined in the cytoplasmic fraction in Table 1 and 2. Interestingly, hydrogenase and alcohol dehydrogenase gene loci were found repoducibly in the same spots which might indicate that these proteins constitute one complex in vivo.

Formation of hydrogen or formate during syntrophic growth
Growth, substrate turnover and formation of hydrogen or formate by Pelobacter species during (syntrophic) growth on acetaldehyde, ethanol or acetoin was evaluated in a 14 days time course ( Fig. 3 and S4 Fig.). Moreover, we report about growth of these species on volatile acetaldehyde ( Fig. 3 A and Fig. S4). The product formation curves indicate a dissimilation pathway in which acetaldehyde is disproportionated to equimolar amounts of ethanol and acetate. The data allow a comparison of the formate and the hydrogen pool sizes. As depicted for P. carbinolicus in Fig. 3 B, hydrogen partial pressure and formate concentration rose fast during the first hours of syntrophic cultivation with ethanol while only minor amounts of ethanol were consumed. After this burst to about 1,000 Pa or 1 mM hydrogen and 2 mM formate, both concentrations decreased slowly during methanogenic consumption. Concentrations of hydrogen and formate increased again after the methanogen was inhibited by addition of 4 mM bromoethanesulfonate plus 20 mM ethanol. The rate of formation of both electron carriers was exponential and converged a limit. Cultivation of P. acetylenicus on ethanol with or without bromoethanesulfonate inhibition showed similar results (see S4 Fig.).
P. acetylenicus and P. carbinolicus both can grow without a methanogenic partner with acetoin as carbon and energy source, yielding acetate and ethanol as products [4,5]. Formation of hydrogen during acetoin or ethylene glycol degradation in pure culture has been reported before [39]. Our investigations confirm that in the absence of a methanogenic partner hydrogen and formate were produced in parallel to ethanol, reaching the same maxima ( Fig. 3 C). P. carbinolicus degraded acetoin in syntrophic coculture as well, producing acetate as the sole product (Fig. 3 D). Growth proceeded in two phases: During a first phase of 48 hours, exponential growth and product formation as in pure culture were observed. In a second phase, the accumulated ethanol was oxidized to acetate as observed with ethanol-degrading cocultures. The course of hydrogen and formate concentrations followed again the ethanol concentration. Bromoethanesulfonate inhibition and ethanol addition caused an increase of hydrogen and formate concentration that came close to the observed limits (Fig. 3 D). The dependence of hydrogen and formate production on the apparent ethanol concentration was observed also in acetaldehyde-grown cultures.
In all cultivation experiments, hydrogen and formate were formed in parallel, even though the turnover of formate was always slightly faster than that of hydrogen. The pool sizes of these electron carriers were within the same order of magnitude. If the headspace was small (330 ml as in ethanol or acetoin cultivations, Fig. 3 B, C, D), the pool size of hydrogen was relatively small. With a bigger headspace (1500 ml as in the acetaldehyde set-up, Fig. 3 A), more electrons were transferred to protons to form hydrogen.
The maximum hydrogen partial pressures and formate concentrations in all cultures allowed the calculation of the corresponding redox potentials using the Nernst equation. The redox potential of hydrogen production was E H2 52358¡12 mV, and the redox potential for formate synthesis was E HCOOH 52366¡19 mV, assuming a bicarbonate concentration of 60 mM. Thus, both syntrophic electron carriers were released at similar energy levels. The culture-specific redox potentials did not differ between P. acetylenicus and P. carbinolicus cultures nor between cultivation substrates. Furthermore, simultaneous monitoring of educt and product concentrations in our cultures allowed the calculation of the Gibbs free energy of ethanol oxidation to acetate and hydrogen according to Thauer et al. [40] in ethanol-or acetoin-fed cultures to be DG'5235 …228 kJ per mol.

Discussion
In the present study, key enzymes of ethanol and acetaldehyde metabolism were assayed in two Pelobacter species, together with hydrogenase and formate dehydrogenase activities which act as electron carriers to the methanogenic partner organism. The observed similarity of enzyme activities in P. acetylenicus and P. carbinolicus agrees with the close phylogenetic relationship and other biochemical similarities of these two species [1]. Thus, both species are likely to use the same set of enzymes for ethanol and acetaldehyde metabolism.

NAD + -dependent ethanol oxidation
Both Pelobacter species oxidized ethanol with NAD + as electron acceptor, just as the ethanol-oxidizing S organism in the M. omelianskii coculture did [26]. However, the standard redox potential of electrons released by ethanol oxidation (2196 mV) is inadequate to reduce NAD + at a standard redox potential of2320 mV, a problem discussed also before [20]. Nonetheless, this unfavorable reaction equilibrium prevents an accumulation of the toxic intermediate acetaldehyde which has been reported to attack DNA and proteins [41][42][43].

Oxidation of acetaldehyde
Our enzyme assays revealed that acetaldehyde can be oxidized by two different enzymes, an acetylating acetaldehyde dehydrogenase and a non-acetylating acetaldehyde dehydrogenase (acetaldehyde: benzyl viologen oxidoreductase). Acetyl-CoA formation from acetaldehyde and subsequent substrate-level phosphorylation follows the original concept of ethanol oxidation [20,24,28,44] and could easily explain the mode of ATP synthesis and carbon assimilation. Two acetylating acetaldehyde dehydrogenase gene loci could be identified: Pcar_1246 is constitutively expressed and Pcar_2758 was found additionally in fast-growing, tungstate-supplied cultures. Expression and activity of these acetylating acetaldehyde dehydrogenases is in agreement with the detected activities of the subsequent substrate-level phosphorylation enzymes phosphotransacetylase and acetate kinase as determined before [30].
However, the non-acetylating acetaldehyde dehydrogenase activity was constitutively expressed as well. The gene loci Pcar_0456 and Pcar_0665, annotated as tungsten-dependent acetaldehyde: ferredoxin oxidoreductases, were expressed after growth in standard medium (12 nM W plus 150 nM Mo) and were even more expressed in tungstate-rich medium (Fig. 1, Fig. 2 and S2 Fig.). Pcar_0665 and Pcar_0456 share 75.4% sequence identity and are likely to carry the same physiological function. Their annotation as tungsten-dependent enzymes is credible since both enzymes share 34% sequence identity with the wellinvestigated ferredoxin-and tungsten-dependent acetaldehyde dehydrogenase of Pyrococcus furiosus [45]. The well conserved binding motifs of the iron-sulfur clusters, the bound Mg 2+ and the tungstopterins give further evidence of functional similarities (data not shown). These enzymes were substituted by a molybdenum-dependent isoenzyme (Pcar_0220) if molybdenum but no tungsten was available.
So far, only two mesophilic bacteria were found to be able to express either molybdenum-or tungsten-dependent isoenzymes according to trace metal availability: Desulfovibrio gigas [33] and Eubacterium acidaminophilum [46]. Other examples of tungsten-dependent acetaldehyde dehydrogenases were found in thermophilic bacteria and archaea [47,48]. The use of tungsten-dependent isoenzymes in the metabolism of Pelobacter species remains unclear. Tungsten cofactors have been reported to catalyse electron transfer reactions at very low redox potential, such as ATP-independent benzoyl-CoA reduction [49]. A tungsten-dependent acetaldehyde dehydrogenase could exploit the entire redox potential of the acetaldehyde/acetate couple (E˚'52580 mV [47]) for ferredoxin reduction and possibly allow for faster acetaldehyde turnover. Coexpression of an additional (tungsten-independent) acetylating acetaldehyde dehydrogenase (Pcar_2758) in P. carbinolicus in tungstate-rich medium might help to balance the acetylating and non-acetylating path of acetaldehyde use. Increased substrate turnover can account for the preference of P. carbinolicus for tungsten-dependent acetaldehyde dehydrogenases. However, tungstate uptake requires specialised systems and is ATP-consuming in the presence of molybdate since tungstate and molybdate have similar physicochemical properties [50][51][52]. Growth in molybdate-free medium might facilitate tungstate import and with this also expression of tungsten-dependent enzymes. Both arguments could help to explain why P. acetylenicus grows faster but not to higher density in tungstate-only medium compared to standard medium (S1 Fig.). The question which one of both acetaldehyde-oxidising pathways is more important for dissimilation of acetaldehyde and its precursors cannot be answered until the acetaldehyde: benzyl viologen oxidoreductase is assayed with its natural electron acceptor. Cell-free extracts of acetaldehyde-grown cells exhibited both activities. In the original concept of acetoin degradation, acetaldehyde was oxidized with NAD + , and the formed NADH was used by the NADH-consuming alcohol dehydrogenase to reduce further acetaldehyde to ethanol [6,7]. However, the non-acetylating acetaldehyde dehydrogenase was also active in cells grown on acetaldehyde or acetoin, indicating an unknown function of this enzyme ( Table 1).

Formation of hydrogen and formate as interspecies electron shuttles
Expression of a comproportionating hydrogenase as proposed by Schut and Adams for the gene loci Pcar_1633-1636 (writing mistake in reference [25] corrected) and Pcar_1602-1605 has been confirmed in the present study. The respective gene loci were the only hydrogenase candidates found in PAGE and activity staining experiments. No NAD + -dependent hydrogenase activity was detectable. Hydrogenase activity could be measured with the artificial electron acceptors benzyl viologen and methyl viologen which both can substitute for ferredoxin. Proof of such comproportionation with in vivo electron acceptors will be a task for future research. Notably, an NADH-and ferredoxin-dependent hydrogenase was found also in the ethanol-oxidizing S organism of the M. omelianskii coculture, which could be interpreted nowadays as a comproportionating hydrogenase activity as well [26].
The measured maximum hydrogen partial pressure (about 1,000 Pa) requires a minimum redox potential of about 2358 mV. A similar redox potential could be calculated for formate synthesis (E<2366 mV). With regard to the detected maximum concentrations, only simultaneous oxidation of NADH and a ferredoxin-like protein (E˚'<2410 mV [53]) would allow for exergonic hydrogenase and formate dehydrogenase reactions. Thus, we expect that the detected tungsten-dependent formate dehydrogenase encoded in gene cluster Pcar_0833-0835 is a comproportionating enzyme as well, even though the formate dehydrogenase exhibited NAD + -reducing activity in a reverse assay. All genes encoding formate dehydrogenase subunits show very high sequence identity with the genes of the putative comproportionating hydrogenase gene cluster Pcar_1633-1636 (omitting Pcar_1635 whose function remains unclear). An electron-comproportionating formate dehydrogenase complex has been reported recently for Clostridium autoethanogenum [54]. The identified formate dehydrogenase was strictly tungsten-dependent and could not be replaced by molybdenum-dependent isoenzymes, as found in Desulfovibrio species [35,36].
In our experiments, hydrogen could replace formate as syntrophic electron shuttle during ethanol oxidation. Cocultures of P. carbinolicus and Methanobrevibacter arboriphilus which cannot use formate as electron donor [31,32] grew at similar rates and to similar densities as cocultures with M. hungatei (S5 Fig.). Growth of Pelobacter carbinolicus or Pelobacter acetylenicus on ethanol solely by interspecies hydrogen transfer in cocultures with M. arboriphilus, as well as axenic growth in a culture vessel continuously sparged with nitrogen have also been reported earlier [28,39,77]. Enzyme assays revealed that formate dehydrogenase activity in P. carbinolicus was lowered by a factor of 100 if formate was not consumed (Table 1). Simultaneous hydrogen and formate turnover (Fig. 3, S4 Fig.) and energetically equivalent maximum concentrations (in connection with synthesis redox potentials, see above) measured in cocultures with M. hungatei prove that hydrogen and formate are used equally and probably simultaneously as electron carriers. However, Pelobacter species might prefer to use hydrogen as sole electron carrier if tungsten is scarce.

Syntrophic acetaldehyde oxidation as an ecological niche
The enzyme activity pattern of P. carbinolicus cells cultivated with ethanol, acetaldehyde or acetoin (Table 1) revealed a certain degree of substrate adaptation. Nonetheless, few activities could be found which are actually not necessary for the respective metabolism: acetoin: DCPIP oxidoreductase after syntrophic cultivation with ethanol, non-acetylating acetaldehyde dehydrogenase and hydrogenase after axenic cultivation with acetaldehyde or acetoin. A differential 2D PAGE of ethanol-and acetoin-grown cells (data not shown) rendered similar (minor) differences in the expression of these proteins. Obviously, these enzymes are not strictly regulated because the overall spectrum of utilizable substrates of these bacteria is rather limited and comprises only substrates that are all closely related to acetaldehyde. This view is supported also by the intermediate accumulation of side products which are obviously caused also by the comparably high substrate concentrations used.
The medium was prepared in 4 or 10 l flasks and distributed to gas-tight infusion bottles under a headspace of N 2 /CO 2 (80:20). Ethanol and acetoin were added from 1 M stock solutions. Cultures were incubated at 28-30˚C in the dark. Cell growth was followed microscopically and by measuring the optical density of 1 ml samples at 578 nm against water or dithionite-reduced medium (Hitachi spectrophotometer 100-40).
Growth experiments with P. acetylenicus plus M. arboriphilus or M. hungatei M1h were also done in rubber-stopped Hungate tubes filled with 10 ml brackish water medium that were inoculated with 0.1 ml of both pure cultures grown on acetoin or CO 2 /H 2 (20%/80%, 1 bar overpressure), respectively. Growth at 30˚C was detected as optical density at 600 nm in a tube spectrophotometer (M107, Camspec).

Growth under trace element limitation
Rubber-stoppered 60 ml-serum bottles that were cleaned by acid and base treatment were extensively flushed with a mixture of N 2 /CO 2 (80%/20%) to remove oxygen, and autoclaved with 100 ml water. Each bottle was filled by syringes with 30 ml medium prepared without tungstate and molybdate. All trace element limitation experiments were done with medium from the same batch. Each growth condition was checked in triplicate. The specially prepared medium contained 100 nM tungstate or 150 nM molybdate or both trace elements. Precultures for those experiments were grown without tungstate or molybdate once to avoid introduction of tungstate or molybdate to the media by the inoculum. For determination of acetate and ethanol concentrations, 200 ml samples were obtained for HPLC. Methane was measured by gas chromatography at the beginning and the end of the growth experiment.

Monitoring substrate turnover and hydrogen production
Hydrogen/formate production and substrate turnover in cultures of Pelobacter species cultivated on ethanol and acetoin was investigated in triplicate in halffilled infusion bottles with approx. 250 ml medium and 330 ml headspace. Growth experiments with cocultures were done with M. hungatei strain M1h as syntrophic partner. Cultures with acetaldehyde as substrate were run semi-sterile in stirred half-filled 1250 ml-infusion bottles with approx. 500 ml medium containing an inoculum of 20% from an acetoin-grown pre-culture. Before inoculation, 500 ml Ti(III)nitrilotriacetic acid (95 mM) was added to eliminate traces of oxygen. Acetaldehyde was added by connecting the headspace of the infusion bottle to a second bottle filled with 2 M anoxic acetaldehyde solution in distilled water via 30 cm gas-tight tubing (5 mm in diameter). The overall headspace was approx. 1500 ml. This set-up was chosen to feed the culture continuously through the gas phase with the volatile acetaldehyde without overdosing this toxic compound. Acetaldehyde cultures were discarded under a fume hood to minimize exposure to ill-smelling thioaldehyde and its derivatives [61,62]. Hydrogen gas concentrations were measured in triplicate by gas chromatography. These gas concentrations were related to the culture volumes to calculate absolute pool sizes, and related to the liquid culture volume. Formate, acetate, acetoin, and ethanol concentrations were determined in triplicate by HPLC. Methanogenesis was inhibited by addition of 4 mM 2-bromoethanesulfonate.
Preparation of cell-free extracts of P. acetylenicus/M. hungatei Cells grown in 1-l cultures were harvested by centrifugation as described before [60]. Cell pellets were washed twice in anoxic 50 mM potassium phosphate buffer, pH 7.5, containing 3 mM dithiothreitol (DTT). Cell pellets were resuspended in 5 ml of the same buffer, transferred to an 8-ml serum vial sealed with a butyl rubber stopper, shock-frozen in liquid nitrogen, and stored at 220˚C. All steps were performed under strictly anoxic conditions in an anoxic glove box [60]. Cell suspensions were lysed by addition of a total of 60 units mutanolysin and 0.2 mg DNaseI per ml cell suspension. Mutanolysin has proven to be a good lysing agent for syntrophic cocultures, as it leaves cell walls of archaea intact and only lyses cells of the fermenting bacteria [63]. Cell suspensions were then incubated anoxically at 37˚C for 60 min. Non-lysed cells and debris were removed by centrifugation at 3,0006g for 20 min. The supernatant (cell-free extract) was stored in serum vials under N 2 on ice. Cell-free extracts of P. acetylenicus used in enzyme assays were obtained from cocultures with M. hungatei strain M1h.

Preparation of cell-free extracts from P. carbinolicus/M. hungatei cells
Cell-free extracts of P. carbinolicus were prepared using similar techniques as for P. acetylenicus. Harvested cell pellets were washed in anoxic 20 mM Tris-HCl (pH 7.3, with medium salts). Pure cultures were lysed directly. In coculture, P. carbinolicus cells were separated from the methanogenic partner M. hungatei by Percoll gradient centrifugation as described [38]. Solutions of Percoll (Sigma) in harvesting buffer were layered in centrifuge tubes in descending order (concentrations 70%, 65%, 60% and 55%) and the resuspended cell pellet was poured on top. Tubes were centrifuged at 10,0006g for 60 min in a swing-bucket rotor. The upper fraction containing P. carbinolicus was recovered with a syringe, and purity was checked by light microscopy (purity .99.5%). Cells were washed in harvesting buffer and centrifuged at 16,0006g for 15 min.

Enzyme assays
All enzyme assays were run anoxically in rubber-stoppered 1-ml cuvettes at 30˚C in a Hitachi spectrophotometer 100-40 connected to an analogous recorder (BBC Goerz Metrawatt SE 120) as described before [60]. Assays were run in triplicate. One enzyme unit was defined as 1 mmol substrate consumed or product formed per min and mg protein under the respective assay conditions. Benzyl viologen and methyl viologen were regarded as one-electron-acceptors.

Protein electrophoresis and activity staining
One-dimensional SDS PAGE was done according to Laemmli [72]. Gel preparation, protein sample preparation, protein electrophoresis and twodimensional (2D) SDS PAGE were performed according to our previously described protocols [38]. For PAGE analyses, P. carbinolicus cocultivated with the genome-sequenced M. hungatei strain JF1 were used to exclude possible misidentification of proteins. No contaminants were ever detected. Following observations in initial proteomic experiments, usage of Protease Inhibitor Cocktail P 8465 (Sigma-Aldrich) was necessary. Cell-free extracts from P. carbinolicus were desalted by gel filtration with a PD-10 column (GE Healthcare Life Sciences). Isoelectric focusing was extended to a total of 60,000 Vh.
Cytoplasmic extract of P. carbinolicus cells cultivated with ethanol was used to identify hydrogenases and formate dehydrogenases in activity staining experiments as described previously [38]. Activities of non-acetylating acetaldehyde dehydrogenases were assayed as for formate dehydrogenase but with 5 mM acetaldehyde as starting reagent. A rather long response time for this enzyme has been observed before [33] which also applied for P. carbinolicus activities in cuvette and activity staining test. Hence, the reaction was facilitated by adding 2 ml of freshly prepared, anoxic dithionite solution (about 50 mM).

Analytical methods
Protein concentrations were measured according to Bradford [73].
Ethanol and acetate concentrations were measured by ion-exclusion HPLC as described before [60,74]. The HPLC system comprised an autoinjector (Gilson 234; Abimed), a high-pressure pump (LC-10AT vp, Shimadzu), an Aminex HPX-87H column (BioRad) heated to 40˚C and a refraction index detector (RID-10A, Shimadzu). 5 mM H 2 SO 4 was used as eluent at a flow rate of 0.6 ml/min. Acetoin, acetate, and formate were determined using a Shimadzu Prominence HPLC system equipped with a photodiode array detector and an Aminex HPX-87H column (BioRad). 5 mM H 3 PO 4 was used as eluent at a flow rate of 1 ml/min. The column was heated to 60˚C. Acetate and formate were detected at 200¡4 nm, acetoin at 274¡4 nm. This method was developed to increase the sensitivity of formate detection down to 10 mM. The measured concentrations of acetate were similar for both HPLC systems, which was useful as an internal reference.
Methane was quantified by gas chromatography with a Carlo Erba GC 6000 (Carlo Erba) with a flame-ionization detector by injection of 100 ml sample from the headspace of cultures with gas-tight syringes. Nitrogen was used as carrier gas on a packed CarboSieve column heated to 120˚C. Hydrogen was measured by gas chromatography with a Peak Performer 1 RCP (Peak Laboratories) using a reductive gas detector. The two inbuilt columns were heated to 105˚C, the detector unit to 265˚C. Nitrogen was used as carrier gas. The injected sample size was 100 ml. Besides syringe injection, a gas-tight, self-made pump-through system allowed automatic and highly reproducible injection directly from the culture headspace.

Peptide mass fingerprinting and database search
Protein bands of interest were excised and analyzed by peptide mass fingerprinting-mass spectrometry at the Proteomics Facility of Konstanz University as described previously [38]. The MASCOT engine (Matrix Science) [75] was used to match each peptide fingerprint against a local database of predicted protein sequences of the annotated genomes of P. carbinolicus (NCBI Reference Sequence: NC_007498.2; US DOE Joint Genome Institute; described in [76]) or M. hungatei JF1 (NCBI Reference Sequence NC_007796.1; US DOE Joint Genome Institute) to verify cell separation. With the applied procedure, score values of 50 or more are regarded as significant.