Growth of the acetogenic bacterium Acetobacterium woodii on glycerol and dihydroxyacetone

enzymatic analysis we present the ﬁ rst metabolic and bioenergetic schemes for glycerol and DHA utilization in A. woodii .


Introduction
Acetogenic bacteria are a specialized group of strictly anaerobic bacteria that are facultative autotrophs (Diekert and Wohlfarth, 1994;Drake et al., 2008;Müller and Frerichs, 2013). Carbon dioxide is reduced to acetate with electrons derived from molecular hydrogen or carbon monoxide in the Wood-Ljungdahl pathway (WLP), the only pathway of carbon dioxide fixation that is coupled to net synthesis of ATP thus enabling an autotrophic life style (Ragsdale and Pierce, 2008;Poehlein et al., 2012;Schuchmann and Müller, 2014). The WLP also conveys enormous metabolic flexibility to acetogens . During heterotrophic growth, the WLP serves as an electron sink and many substrates can be oxidized with the electrons shuffled to the WLP. Redox balancing between the oxidation module and the WLP is achieved by the Rnf complex (ferredoxin $ NAD) (Westphal et al., 2018), the Ech complex (ferredoxin $ protons) (Schölmerich and Müller, 2019)), the electron bifurcating hydrogenase (hydrogen $ NAD + ferredoxin) (Schuchmann and Müller, 2012) and the transhydrogenases Nfn and Stn (NADPH $ NAD + ferredoxin; Mock et al., 2015;Kremp et al., 2020). By transferring the electrons to the WLP, sugar oxidation yields the highest amount of ATP (4.3 mol ATP/mol hexose) in fermenting bacteria and thus makes it possible for acetogens to compete with other bacteria in anoxic ecosystems (Müller, 2008).
In the oxidation module, molecular hydrogen, carbon monoxide, sugars, primary and secondary alcohols, carbonic acids or methyl groups are oxidized (Bertsch and Müller, 2015;Schuchmann et al., 2015;Bertsch et al., 2016;Trifunovi c et al., 2016;Dönig and Müller, 2018;Kremp et al., 2018;Lechtenfeld et al., 2018). In recent years, many of these pathways have been elucidated on a molecular level. However, little is known on the ability to grow on C3 sugars such as glycerol or dihydroxyacetone (DHA). The latter has not been described as a growth substrate for acetogens. Glycerol has been described to promote growth of Acetobacterium and Eubacterium species (Eichler and Schink, 1984). We have followed up these observations and analysed glycerol metabolism in Acetobacterium woodii. The unexpected result was that cells did grow after the first transfer but could not be subcultured on glycerol. This led us to analyse whether the oxidation product of glycerol, DHA, is a growth substrate for A. woodii.

Results
Acetobacterium woodii is able to use glycerol as sole carbon and energy source In 1984, Eichler & Schink reported the isolation of a new Acetobacterium strain, Acetobacterium carbinolicum. In the course of the characterization of the strain, the substrates utilized were determined and the data summarized in a table ('+' for growth, 'À' for no growth). The same table included A. woodii strain NZva16 as a control. Both species were reported to grow on glycerol but growth curves were not presented. Since glycerol is a cheap feedstock and would be interesting for acetogenic conversions in biotechnological applications, we decided to follow up this observation and to study glycerol metabolism in the acetogenic model strain A. woodii DSM1030. After transfer of A. woodii from a preculture grown on fructose to complex medium containing 5 mM glycerol as carbon and energy source, cells grew with a rate of 0.12 h À1 to a final OD of 0.36 (Fig. 1.). The medium contained 0.02% yeast extract that enabled poor growth up to an OD 600 of only 0.13 (Fig. 1). When an aliquot of the glycerol-grown culture was used to inoculate (20%) a second culture, growth was much slower (μ = 0.06 h À1 ) and the final OD was only 0.21. After the third transfer, there was no growth anymore. We were puzzled by this effect and tried to adapt A. woodii on glycerol by decreasing the fructose concentration in the preculture and increasing the glycerol concentration in the main culture (5, 10, 50 mM). We also precultivated cells on fructose + glycerol (5 mM each) or used ethylene glycol or DHA as substrate for the preculture. A change of the medium to minimal medium with and without bicarbonate/ CO 2 , to phosphate buffered medium with and without bicarbonate/CO 2 , to complex medium containing tryptone (2 g L À1 ), to complex medium with 120, 180 or 240 mM bicarbonate, or to the medium used by Eichler and Schink (1984) did not prevent the effect. Still, A. woodii could not be subcultured on glycerol.
To determine whether glycerol was actually consumed in the first transfer, we determined the glycerol concentration over time in the culture. After transfer from a preculture containing only 2 mM fructose to a culture containing 5 mM glycerol, glycerol consumption started immediately (Fig. 2) and acetate was produced simultaneously. When glycerol was exhausted, growth ceased. Glycerol (4.5 mM) was utilized with a rate of 0.61 ± 0.05 mM h À1 , 5.1 mM acetate were produced and the pH in the medium stayed constant over time. This experiment clearly demonstrates that glycerol is used as carbon and energy source by A. woodii.
A pduAB deletion mutant grows stably on glycerol The genome of A. woodii encodes a propanedioldehydratase (PduCDE) that is encapsulated by a bacterial microcompartment (Schuchmann et al., 2015). This gene cluster is apparently also induced during growth on other substrates (Schuchmann et al., 2015). Transcriptome analysis with cells grown on glycerol (see below) revealed a twofold (log2) overexpression of the pdu genes. We speculated that the propanediol dehydratase may also act on glycerol, leading to the dead end-product 3-hydroxypropionaldehyde, a broadspectrum antibiotic also known as Reuterin (Talarico et al., 1988). To check this possibility, we took advantage of a ΔpduAB stain in which not only the genes pduAB but also the upstream promoter region was deleted; this mutant will be described elsewhere. Indeed, the ΔpduAB mutant could be subcultured on glycerol with rates and yields not decreasing over time (Fig. 3A). Using this strain, we determined the concentration dependence for Growth of Acetobacterium woodii after the first transfer to medium containing glycerol. 5% of a preculture grown on 2 mM fructose was used to inoculate 5 ml CO 2 /bicarbonate-buffered medium containing yeast extract [0.2% (w/v)] under a N 2 /CO 2 [80/20% (v/v)] atmosphere. The cultures were cultivated at 30 C in the absence (■) or presence of 5 mM glycerol (▲). Growth was followed by measuring the optical density (OD) at 600 nm. All data points are mean ± SEM; N = 3 independent experiments. growth on glycerol. As seen in Fig. 3B, the final yield and growth rate were dependent on the glycerol concentration. By adding more glycerol, the growth rate increased and reached a maximum at 5 mM glycerol with 0.13 ± 0.01 h À1 . Higher concentrations led a slight decrease, until 500 mM glycerol. By using more than 500 mM the growth rate was decreased drastically until only slight growth was observed at 3 M glycerol (Fig. 3B). The final yield was maximal at 50-100 mM and declined slightly thereafter. Surprisingly, even with 500 mM glycerol the ΔpduAB stain was able to grow fine with slightly reduced growth rate of 0.127 ± 0,004 h À1 and a final OD 600 of 1.6 ± 0.05. Deletion of pduAB did not only lead to stable growth on glycerol but also to a higher tolerance towards glycerol.
Acetobacterium woodii is able to use DHA as sole carbon and energy source Dihydroxyacetone phosphate (DHAP) is a likely intermediate of glycerol metabolism. To address whether A. woodii is able to grow on DHA, cells were precultured on complex medium with 2 mM fructose and then transferred into complex medium with 10 mM DHA as carbon and energy source. As can be seen in Fig. 4, cells started to grow without an apparent lag phase and a growth rate Glycerol utilization by Acetobacterium woodii after the first transfer to glycerol-containing media. 5% of a preculture grown on 2 mM fructose was used to inoculate 250 ml CO 2 /bicarbonate-buffered medium under a N 2 /CO 2 [80/20% (v/v)] atmosphere and 4.5 mM glycerol as a sole carbon and energy source. The cultures were cultivated at 30 C and at the time points indicated 2 ml samples were taken. Concentrations of glycerol (▼), acetate (♦) and pH (▲) were measured as described in Material and Methods, OD 600 (■) was determined photometrically. All data points are mean ± SEM; N = 3 independent experiments. A. 5% of a preculture grown on 20 mM fructose was used to inoculate 5 ml CO 2 /bicarbonate-buffered medium containing yeast extract [0.2% (w/v)] and 20 mM glycerol under a N 2 /CO 2 [80/20% (v/v)] atmosphere. The first (•), second (■) and third transfer (▲) were cultivated at 30 C. B. 5% of a fully adapted preculture on 20 mM glycerol was used to inoculate 5 ml CO 2 /bicarbonate-buffered medium containing different amounts of glycerol and yeast extract [0.2% (w/v)] under a N 2 /CO 2 [80/20% (v/v)] atmosphere. Growth was followed by measuring the optical density (OD) at 600 nm. Growth yield and final OD 600 were plotted against the glycerol concentration. These data points represent three different biological replicates. of 0.15 h À1 to a final optical density of 0.68. At the third transfer, the growth rate was slightly increased and the final OD was increased to 1.07.
To determine the optimal substrate concentration, cells were grown at different DHA concentrations. At 10 mM DHA, the growth rate was 0.16 and the final OD 600 only 1.1 (Fig. 5A). Increasing DHA concentrations led to a decrease in growth rates but to an increase in final yield; the growth rate was optimal at 5 mM DHA whereas the highest yield was obtained at 50 mM DHA (Fig. 5B). With 10 mM DHA as substrate, DHA consumption and acetate formation had a lag phase of about 5 h. Then, DHA decreased with a rate of 0.41 ± 0.03 mM h À1 whereas acetate was produced with a rate of 0.42 ± 0.04 mM h À1 (Fig. 6). Ethanol was not produced. At the end of growth, 7.8 mM of DHA was converted to 8.6 mM acetate plus biomass, corresponding to.
Resting cells also converted DHA to acetate (Fig. 7). In the presence of NaCl in the buffer, 8.5 mM DHA were oxidized with a rate of 4.4 mM h À1 to finally 10.4 mM acetate, giving a acetate to DHA ratio of 1.2.

The WLP is involved in DHA utilization
The data presented so far are in line with the hypothesis that electrons coming from DHA oxidation are transferred to the WLP for disposal. The WLP in A. woodii requires Na + (Heise et al., 1989) and indeed, in the absence of additional Na + , acetate formation as well as DHA utilization was drastically reduced (Fig. 7). A hydrogenase mutant of A. woodii was not able to grow on fructose anymore because the reducing equivalents from glycolysis can no longer be disposed by the WLP (Wiechmann et al., 2020). Same is true for growth on DHA which is impaired in the ΔhydBA mutant (data not shown). Moreover, a Δrnf mutant (Westphal et al., 2018) lacking the energy conserving Rnf complex did also not grow on DHA. These experiments underline the notion that the WLP is essential as electron disposal module for growth on DHA.

Identification of genes involved in DHA and glycerol utilization
Glycerol could be taken up by a transporter and then oxidized to DHA by glycerol dehydrogenases followed by phosphorylation of DHA by a kinase. Alternatively, glycerol could be phosphorylated by a glycerol kinase to glycerol-3-phosphate followed by oxidation to DHAP. Inspection of the genome sequence revealed a gene possibly encoding a glycerol facilitator (Awo_c29540), two genes encoding potential glycerol dehydrogenases (Awo_c24450, Awo_c06880), five genes encoding glycerol-3-phosphate dehydrogenases (Awo_c01050, Awo_c09230, Awo_c12570, Awo_c12750, Awo_c17750), six genes encoding potential glycerol kinases (Awo_c09050, Awo_c12560, Awo_c12770, Awo_c16650, Awo_c23340, Awo_c23380) and three genes encoding potential DHA kinase (Awo_c20330 -Awo_c20310). Thus, the pathways for entry of glycerol and DHA could not be proposed unambiguously from the genome sequence. Therefore, we studied gene expression in DHA-and glycerol-cultivated cells on a genome-wide level using transcriptomics. Interestingly, genes with the highest degree of differential regulation are apparently not involved in glycerol metabolism but in ion homeostasis (Table S1). Nitrogenase was upregulated ≈5 fold as well as nickel, iron and molybdate transporter. As mentioned above, also pdu genes were highly upregulated. Among the gene clusters potentially related to glycerol metabolism was only Awo_c12730 -c12770 that encodes for five different proteins (Fig. 8). Awo_c12770 encodes a glycerol kinase with 68.8% identity to the GlpK from Escherichia coli, Awo_c12760 is a glycerol-3-phosphate responsive antiterminator, Awo_c12750 a glycerol-3-phosphate dehydrogenase with 23.4% similarity to GlpA from E. coli, Awo_c12740 is a FAD-dependent and Awo_c12730 is a Fe-containing oxidoreductase that are 20.2% and 23.5% similar to GlpB and GlpC from E. coli (Fig. 8). This cluster was not only upregulated Adaptation of Acetobacterium woodii to grow on dihydroxyacetone. 5% of a preculture grown on 2 mM fructose was used to inoculate 5 ml CO 2 /bicarbonate-buffered medium under a N 2 /CO 2 [80/20% (v/v)] atmosphere and 10 mM dihydroxyacetone as carbon and energy source. The first (■) and third adaptation step (▲) were cultivated at 30 C and growth was followed by measuring the optical density (OD) at 600 nm. All data points are mean ± SEM; N = 3 independent experiments. A. 5% of a preculture grown on 10 mM dihydroxyacetone was used to inoculate 5 ml CO 2 /bicarbonate-buffered medium under a N 2 /CO 2 [80/20% (v/v)] atmosphere. The cultures were cultivated at 30 C in absence (■) or presence of 10 mM dihydroxyacetone (▲). Growth was followed by measuring the optical density (OD) at 600 nm. B. The growth rate μ (■) and final optical density at 600 nm (OD 600 ,▲) were plotted against the dihydroxyacetone concentration. All data points are mean ± SEM; N = 3 independent experiments. Dihydroxyacetone utilization by Acetobacterium woodii. 5% of a preculture grown on 10 mM dihydroxyacetone was used to inoculate 250 ml CO 2 /bicarbonate-buffered medium under a N 2 /CO 2 ([80/20% (v/v))) atmosphere and 10 mM dihydroxyacetone as carbon and energy source. The cultures were cultivated at 30 C and at the time points indicated 2 ml samples were taken. Concentrations of dihydroxyacetone (▲) and acetate (▼) were measured as described in Experimental procedures, OD 600 (■) was determined photometrically. All data points are mean ± SEM; N = 3 independent experiments. during growth on glycerol, but had very high transcript levels (Table S3). There was another potential glycerol kinase gene upregulated (Awo_c16650) but the transcript levels were $ 50 fold lower. Awo_c12730-c12770 is annotated as a glpK3 operon, interestingly, there is a similar operon, glpK2 (Awo_c12560-12,590) that is not upregulated (Table S3). Same holds for the potential glycerol facilitator (Awo_c29540), solitary glycerol-3-phosphate dehydrogenases (Awo_c01050, Awo_c09230, Awo_c17750), glycerol kinases (Awo_ c09050, Awo_c23340, Awo_c23380) and glycerol dehydrogenases (Awo_c06880). In sum, these data suggest phosphorylation of glycerol followed by oxidation to DHAP, catalysed be the glpK3 operon gene products as most likely pathway.
In the presence of glycerol, genes for uptake and phosphorylation of fructose (Awo_c03330, Awo_c03340) were highly downregulated as well as other genes involved in sugar metabolism (Table S2). As expected, genes encoding enzymes of the WLP and the respiratory chain were hardly altered, if at all, the trend was a slight downregulation during growth on glycerol (Table S3).
Cells grown on DHA did not upregulate genes involved in ion acquisition but otherwise the response was very similar to cells grown on glycerol (Tables S4-S6). Genes possibly involved in uptake and phosphorylation of DHA are Awo_c09000 -c09020 and Awo_c20310 -c20330 that each encode similar set of genes, but they were not differentially transcribed during growth on glycerol or DHA. However, transcript levels for the second cluster were much higher (≈10-fold, Table S6).

Genes possibly involved in glycerol and DHA oxidation
The utilization of DHA is well described in bacteria (Magasanik et al., 1953;Gutnick et al., 1969;Lin, 1976;Forage and Lin, 1982) and there are different ways to phosphorylate DHA to DHAP, an intermediate of glycolysis. At least to our knowledge, DHA has not been described as carbon and energy source for acetogenic bacteria. Here we describe that the model organism A. woodii grows on DHA. The growth rate (μ = 0.16 h -1 ) is comparable to growth on 10 mM fructose (μ = 0.17 h À1 ). DHAP is a central intermediate of glycolysis and the unique feature of DHA metabolism is its uptake and phosphorylation. The genome sequence of A. woodii encodes proteins with high similarity to a DHA kinase that consists of three soluble subunits. They are encoded in one potential operon comprising three genes (Awo_c20330 -Awo_c20310). Awo_c20330 is 996 bp long and encodes for DhaK that is only separated 46 bp from Awo_c20320, that is 639 bp long and encoding DhaL. Of note, 4 bp upstream is Awo_c20310, which has a length of 348 bp and encodes for DhaM. DhaK has a molecular mass of 34.98 kDa and DhaL of 22.55. They are similar to the N-and C-terminal halves of the ATPdependent DHA kinase ubiquitious in bacteria, plants and animals (Daniel et al., 1995;Gutknecht et al., 2001), respectively. Like in A. woodii, the kinase is encoded in E. coli by two genes, dhaK and dhaL. DhaK and DhaL of E. coli and A. woodii share a 50.6% and 45.4% similarity. The phosphoryl group donor for DhaKL in E. coli is phosphoenolpyruvate, not ATP, with DhaM acting as donor. DhaM of E. coli is 51.5 kDa and consists of three domains, EIIA, Hpr and EI (Gutknecht et al., 2001). Each of the domains has an active site histidine that is phosphorylated by the general PTS proteins EI and Hpr; the EII-domain of DhaM then transfers the phosphoryl group to DhaKL (Gutknecht et al., 2001). DhaM of A. woodii (Awo_c20310) is only 13.1 kDa and similar to the EIIdomain of DhaM Ec . Therefore, it is likely that the phosphoryl group of PEP is transferred via the general PTS proteins EI and Hrp to DhaM; DhaM is at the end of the phosphoprotein relay and phosphorylates DHA by DhaKL. General PTS proteins are encoded in the genome of A. woodii but no DHA-PTS. There are only three phosphotransferase systems annotated in the genome of A. woodii: PtsG of the glucose family, FruA/B of the Fru family and SrlA/E, SrB and SrlE of the sorbitol family. We could not find data in the literature that any of these are known to transport DHA as well. Therefore, DHA transport in A. woodii remains an open question. One potential candidate for DHA uptake could be the glycerol uptake facilitator GlpF (Awo_c29540). Unlike in E. coli, the up-and downstream genes are apparently not involved in DHA or glycerol metabolism.
From the data presented here, it seems that phosphorylation of glycerol to glycerol-3-phophate followed by an oxidation to DHAP is the most likely pathway for activating and channelling glycerol into glycolysis mediated by the products of the glpK3 gene cluster, the kinase is a three-component (glpABC) glycerol-3-phosphate dehydrogenase. Since other pathways may also be present as indicated by the genome analysis, transcriptomics and enzyme measurements, the pathway postulated is not unambiguously. It also may be possible that different feeding pathways for glycerol are active simultaneously. The role of different pathways in glycerol metabolism requires future mutational analysis.

Glycerol utilization in the pduAB mutant
The pdu gene cluster of A. woodii encodes a diol dehydratase along with proteins forming a bacterial microcompartment (Schuchmann et al., 2015;Chowdhury et al., 2020). In contrast to Salmonella, in which the system is only involved in propanediol metabolism, the pdu-encoded BMS's are produced in A. woodii during growth on several substrates such as fructose, several diols, H 2 + CO 2 and ethanol (Schuchmann et al., 2015) which raises the question about the nature of the inducer/repressor for gene activation that is not answered yet. Obviously, the pdu expression was induced during growth on glycerol. We hypothesize that the diol dehydratase dehydratases glycerol to 3-hydroxypropionaldehyde, a toxic compound with antimicrobial activities (Talarico et al., 1988). This hypothesis is in line with the observation that deletion of pduAB led to stable growth and high tolerance of A. woodii on glycerol.
The DHA and potential glycerol metabolism in A. woodii From the data presented here, we propose the following metabolic pathway of acetogenesis from DHA and glycerol. Two molecules of DHA are taken up and phosphorylated to DHAP via the dihydroxyacetone kinase pathway. DHAP is then converted to glyceraldehyde-3-phosphate by the triosephosphate isomerase which is further metabolized to PEP and pyruvate. Pyruvate is then decarboxylated to acetyl-CoA and further converted to acetate. The generated 2 CO 2 , 2 Fd red and 2 NADH are used in the WLP to generate one additional acetate. In addition, 0.3 ATP are produced by the respiratory chain (Schuchmann and Müller 2014). This metabolic scenario is consistent with the observation that Na + is required for DHA oxidation and that DHA oxidation is impaired in mutants with defects in WLP genes. Oxidation of 1 mol DHA yields 2.15 mol of ATP corresponding to the following equation (Fig. 9A): This is the highest C3 substrate:ATP ratio for acetogenic bacteria. On the other hand, with glycerol as a substrate the energy yield is different. The initial phosphorylation of glycerol leads to an investment of ATP. The further oxidation of glycerol-3-phosphate to DHAP leads to additional 2 NADH, and to reoxidize the excess NADH, additional CO 2 has to be taken up and reduced to acetate via the WLP. In this scenario 0.25 mol ferredoxin have to be reduced with NADH as reductant to get the reduced ferredoxin required for CO 2 reduction to CO. To drive ferredoxin reduction with NADH as reductant, reversed electron transport is required, energized by ATP hydrolysis (Fig. 9B). In sum, glycerol is converted to acetate according to: 2 glycerol þ 1 CO 2 þ 3:85 ADP þ 3:85 P i ! 3:5 acetate þ 3:85 ATP Experimental procedures

Cultivation of A. woodii
A. woodii DSMZ 1030 (WT or ΔpduAB) was cultivated at 30 C in complex medium (Heise et al., 1989). Minimal media did not contain yeast extract, and the amount of KH 2 PO 4 and NH 4 Cl was increased to 0.2 and 1.35 g L À1 respectively, selenite tungsten solution to 1.5 ml L À1 and 10 μg ml À1 of D/L-pantothenic acid was added. For some experiments, the medium described by Eichler and Schink (1984) was used. Fructose (2-20 mM), ethylene glycol (50 mM), DHA (10 mM) or glycerol (5-50 mM) was used as growth substrate. Since the deletion strain is uracil auxotrophic, uracil was supplemented to the media with a final concentration of 50 mg L À1 . Growth was monitored by measuring the optical density at 600 nm (OD 600 ).

Preparation of cell suspensions
All buffers and media were prepared using the anaerobic techniques described previously (Hungate, 1969;Bryant, 1972) and all preparation steps were performed at room temperature in an anaerobic chamber (Coy Laboratory Products, Grass Lake, MI; filled with 95%-98% N 2 and 2%-5% H 2 ) under strictly anoxic conditions. A. woodii was grown with DHA as substrate to an OD 600 of 0.6-0.7, harvested by centrifugation (8.000Âg, 10 min) and washed and resuspended in imidazole buffer (50 mM imidazole-HCl, 20 mM MgSO 4 , 20 mM KCl, 2 mM DTE, 4 μM resazurin, pH 7.0). The protein concentration of the cell suspension was measured as described (Schmidt et al., 1963). Resting cell experiments were done in 20 ml imidazole buffer (50 mM imidazole-HCl, 20 mM MgSO 4 , 20 mM KCl, 2 mM DTE, 4 μM resazurin, 60 mM KHCO 3 , pH 7.0) under a N 2 /CO 2 atmosphere. Cells were added to a final protein concentration of 1 mg ml À1 . These cell suspensions were incubated at 30 C for 5 min and the experiment was started by the addition of 200 μl DHA (1 M). For the determination of the substrate and product concentrations, 1 ml samples were taken at the indicated time points, cells were removed by centrifugation (14.000Âg, 1 min) and the supernatant was stored at À20 C for further analysis.

Preparation of cell-free extracts
Cells grown on fructose, glycerol or DHA were cultivated till mid exponential growth phase (OD 600 = 1.25 for fructose, 0.71 for DHA and 0.2 for Glycerol) and were harvested aerobically by centrifugation (8.000Âg, 10 min) and washed and resuspended in 3 ml buffer A (50 mM Tris, 20 mM MgSO 4 , pH 7.5). Cells were disrupted by passing two times through a 'French Press' cell (SLM Aminco, SLM instruments, USA) at 100 MPa. Cells and cell debris were removed by centrifugation (13.000 rpm 10 min, 4 C). The protein concentration was measured according to Bradford (1976). The cell-free extract was stored at 4 C until further use.

Determination of enzyme activities
All measurements were performed at 30 C under oxic conditions. To measure the electron transfer from glycerol to NAD + in cell-free extracts from A. woodii grown on different substrates, different amounts of cell-free extracts and 3 mM NAD + were added to the buffer (0.1 M Na 4 P 2 O 7 , pH 8.0). The reaction was started by addition of 30 mM glycerol and monitored by following the absorption at 340 nm. NADH-dependent DHA reduction was measured under identical conditions, with 30 mM DHA as electron acceptor. For electron transfer from NADH to DHAP 0.45 mM NADH were added to the buffer (0.3 M triethanolamine, pH 7.6). The reaction was started by addition of 0.5 mM DHAP.
To determine glycerol kinase activity 30 mM glycerol and 3 mM ATP were added to the buffer (0.2 M glycine, 20.9 g ml À1 hydrazine, 1.8 mM MgCl 2 , pH 9.8). ADP produced was determined by a coupled optical-enzymatic assay by adding 160 μl premixed 'Reaction Start solution' ('Glycerol Assay Kit' (Megazyme, Bray, Ireland) following the instructions of the manufacturer, but suspension 4 (Glycerol kinase) was omitted. The ADP produced was used by phosphoenolpyruvate kinase and to generate pyruvate from phosphoenolpyruvate. Pyruvate was then reduced to lactate by a lactate dehydrogenase and NADH oxidation was monitored at 340 nm.
The change in absorption during the electron transfer and the extinction coefficient of NADH/NAD + (ε = 6.2 mM À1 cm À1 ) were used to calculate the apparent enzymatic activities in the cell-free extracts. 1 Unit (U) is defined as 1 μmol substrate utilized per minute.

Determination of acetate, glycerol and DHA
For acetate determination 400 μl of the samples were mixed with 50 μl phosphoric acid (2 M), 50 μl 1-propanol (200 mM) and 500 μl acetone (13.6 M). The samples were analysed by gas chromatography on a Gas chromatograph 7890B GC Systems (Agilent Technologies, Santa Clara, California, USA) with an DB-Wax column (30 m Â 0.25 mm, Agilent Technologies, Santa Clara, California, USA). The following temperature profile was used: 60 C 3 min followed by a temperature gradient to 180 C with 10 C min À1 followed by a hold at 180 C for 5 min. Acetate was analysed with a flame ionization detector at 250 C. The peak areas were proportional to the concentration of each substance and calibrated with standard curve.
Glycerol concentration was measured using a 'Glycerol Assay Kit' (Megazyme, Bray, Ireland) following the instructions of the manufacturer. The DHA concentrations were measured as described (Sponholz and Wünsch, 1980).

Transcriptome analyses
For transcriptome analyses, wild type cells were grown on 10 mM DHA or during the first adaptation step with 5 mM glycerol and were harvested in the exponential growth phase with OD 600 of 0.31 and 0.15 respectively. All the following steps were carried out as described previously . Genes with a log 2 -fold change (FC) of +2/À2 and a P-adjust value < 0.05 were considered differentially expressed. The sequence data have been submitted to the SRA database. transhydrogenase essential to reversibly link cellular NADH and ferredoxin pools in the acetogen Acetobacterium woodii. J Bacteriol 200: e00357-00318. Wiechmann, A., Ciurus, S., Oswald, F., Seiler, V.N., and Muller, V. (2020) It does not always take two to tango: "Syntrophy" via hydrogen cycling in one bacterial cell. ISME J 14: 1561-1570.

Supporting Information
Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Table S1. The most upregulated genes of A. woodii during the first transfer on glycerol. Table S2. The most downregulated genes of A. woodii during the first transfer on glycerol. Table S3. Transcript abundance of potential genes involved in glycerol metabolism of A. woodii. Table S4. The most upregulated genes of A. woodii during growth on dihydroxyacetone. Table S5. The most downregulated genes of A. woodii during growth on dihydroxyacetone. Table S6. Transcript abundance of potential genes involved in dihydroxyacetone metabolism of A. woodii.