An Iron-Dependent Alcohol Dehydrogenase Is Involved in Ethanol Metabolism of Aromatoleum aromaticum

Abstract

The NAD⁺-dependent alcohol dehydrogenase AdhB from Aromatoleum aromaticum EbN1 belongs to family III of Fe-dependent alcohol dehydrogenases. It was recombinantly produced in Escherichia coli and biochemically characterized, showing activity only with ethanol or n-propanol. The enzyme contained substoichiometric amounts of Fe, Zn, and Ni and a yet unidentified nucleotide-like cofactor, as indicated by mass spectrometric data. As suggested by its narrow substrate spectrum and complementation of a related species to growth on ethanol, the most probable physiological function of AdhB is the oxidation of short aliphatic alcohols such as ethanol or n-propanol. The enzyme also exhibits a very high tolerance to ethanol and n-propanol, showing moderately substrate-inhibited Michaelis–Menten kinetics up to concentrations of 20% (v/v). AdhB can also be applied biotechnologically to convert acetate to ethanol in coupled enzyme assays with the tungsten enzyme aldehyde oxidoreductase, showing activity with either another aldehyde or pre-reduced benzyl viologen as electron donors.

1. Introduction

The betaproteobacterial species Aromatoleum aromaticum is highly flexible in growing on many different substrates under either aerobic or denitrifying conditions. In addition to various aromatic compounds, it also accepts aliphatic substrates, e.g., several amino acids, organic acids, aldehydes, ketones, or alcohols. The pathways involved in either aerobic or anaerobic degradation of many of these substrates have been identified in recent years, and many of the participating enzymes have been biochemically characterized. Moreover, the available genome sequence allowed the correlation of the respective enzymes with their coding genes and their induction and regulation processes in the presence or absence of the respective substrates to be investigated [1,2,3,4].

In a previous study, we have observed that a constructed A. aromaticum strain lacking the pdh gene for a substrate-specific phenylacetaldehyde dehydrogenase involved in anaerobic phenylalanine (Phe) degradation showed highly retarded growth on Phe, but evolved back to almost the previous growth rate by overexpressing the aldB gene for another aldehyde dehydrogenase, which carried a mutation of a single amino acid (Y460C) and forms an apparent operon with the adhB gene coding for an Fe-dependent alcohol dehydrogenase [5]. The original aldB gene product was recently characterized as NAD⁺-dependent aldehyde dehydrogenase exhibiting high activities with acetaldehyde or propionaldehyde and lower activities with benzaldehyde or phenylacetaldehyde. In contrast, the Y460C variant showed strongly reduced activities with acetaldehyde or propionaldehyde and lost its activity with benzaldehyde, but retained the same activity with phenylacetaldehyde as wild-type AldB [6]. Therefore, it has been assumed that the physiological role of AldB is the oxidation of the aldehyde intermediates in the degradation pathway of short aliphatic alcohols [6,7], while the co-expressed adhB gene in the same operon might code for the corresponding alcohol dehydrogenase. However, the corresponding AdhB protein has not yet been investigated for its biochemical properties, prompting us to purify and characterize the protein.

The alcohol dehydrogenases (EC 1.1.1.1) are NAD(P)⁺-dependent enzymes oxidizing a wide range of aliphatic or aromatic alcohols to the corresponding aldehydes (or catalyzing the reverse reaction with NAD(P)H as cofactor). They are present in all organisms and typically consist of homodimers or tetramers of 45–60 kDa subunits. Most of the known ADHs are currently affiliated to three major families: family I consists of the medium- or long-chain zinc-containing ADHs, the most studied group in vertebrates; family II, the “short-chain dehydrogenases and reductases” (SDR) do not harbor a metal cofactor; and family III is represented by the iron-dependent or iron-activated enzymes containing a transition metal ion (mostly Fe²⁺) in the active center [8,9,10,11]. The alleged alcohol dehydrogenase AdhB of A. aromaticum is affiliated with the iron-dependent ADHs of family III [12,13], representing the only enzyme of this family encoded in the genome [1].

2. Materials and Methods

Cloning, heterologous gene expression, and preparation of cell-free extracts. The gene adhB (ebA4623) from A. aromaticum strain EbN1 was amplified via PCR from chromosomal DNA using appropriate primers (AdhB_for AAGCTCTTCAATGAGCACGACGACTTTCTTCATCC and AdhB_rev AAGCTCTTCACCCCAGCGCGCCGCGGAAGATCGCC) and cloned into the broad host-range vectors pASG105-mobori or pASG103-mobori [12], using the “Stargate” cloning system (IBA Lifesciences, Göttingen, Germany). The resulting plasmids code for fusion proteins of AdhB with N- or C-terminal Twin-Strep-tag sequences. The enzymes were subsequently produced in E. coli DH5α, which was grown in LB medium at room temperature and induced with added anhydrotetracycline as reported previously [5]. Cells were harvested by centrifugation and resuspended in two volumes of 10 mM Tris/HCl, pH 7.5, containing 0.1 mg/mL DNase I. Cell-free extracts were prepared by sonification at 4 °C, followed by ultracentrifugation (100,000× g, 60 min). AdhB was exclusively present in the soluble fractions.

Protein purification and characterization. Cell-free extracts with overproduced AdhB were applied on a Streptactin affinity column (IBA Lifesciences, Göttingen, Germany) and eluted with 0.2 mg/mL desthiobiotin, as reported before [5,14]. Native molecular masses of AdhB were determined by gel filtration on a calibrated Superdex-200 10/300 column (Cytiva, Marlborough, MA, USA), using 100 mM Tris/HCl, pH 7.8, 100 mM KCl, 5% glycerol as buffer, by Ferguson plot analysis after native polyacrylamide gel electrophoresis (6–10% polyacrylamide gels) [15] and by crosslinking analysis with glutardialdehyde, as described previously [16]. The buffer of the purified proteins was exchanged into protein storage buffer without the respective eluent (30% glycerol, 150 mM NaCl, 25 mM Tris/Cl, pH 7.9). Proteins were stored at −20 °C until further use.

Cofactors were extracted from the purified AdhB protein by acid treatment with HCl and removing the precipitated protein by centrifugation. The supernatant was then analyzed by UV–Vis spectroscopy. Further standard protein analytic techniques, such as SDS-PAGE and concentration determinations, were performed as described in Coligan et al. [17].

Metal contents of protein fractions and controls were analyzed by inductively coupled plasma mass spectrometry (ICP-MS), as described previously [18]; protein concentrations were determined as described in Bradford [19]. The extracted cofactor was subjected to LC-MS/MS analysis [16,20].

LC-MS/MS measurements. LC-MS/MS measurements were performed on an Orbitrap ID-X (Thermo Scientific, Waltham, MA, USA) connected to a Vanquish HPLC system (Thermo Scientific). The chromatographic separation was performed using a SeQuant ZIC-pHILIC column (150 × 2.1 mm, 5 μm particle size, peek coated, Merck, Darmstadt, Germany) connected to a guard column of similar specificity (20 × 2.1 mm, 5 μm particle size, Phenomoenex, Torrance, CA, USA) a constant flow rate of 0.1 mL/min with a mobile phase comprised of 10 mM ammonium acetate in water, pH 9, supplemented with medronic acid to a final concentration of 5 μM (phase A) and 10 mM ammonium acetate in 90:10 acetonitrile to water, pH 9, supplemented with medronic acid to a final concentration of 5 μM (phase B) at 40° C [21]. The injection volume was 1 µL. The mobile phase profile consisted of the following steps and linear gradients: 0–1 min constant at 90% B; 1–9 min from 90 to 40% B; 9 to 10 min constant at 40% B; 10–10.1 min from 40 to 90% B; 10.1 to 20 min constant at 90% B. 503.22, 541.175, 755.333, 793.289, 831.245.

Orbitrap ID-X was used in negative mode. Ionization was performed using a high-temperature electro spray ion source at a static spray voltage of 2500 V (negative), Sheath gas at 35 (Arb), Auxiliary Gas at 7 (Arb), and Ion transfer tube and Vaporizer at 300 and 275 °C. Data-dependent MS2 measurements were conducted applying an Orbitrap mass resolution of 120,000 using quadrupole isolation in a mass range of 500–1000 and combining it with a high-energy collision-induced dissociation (HCD). HCD was performed on five ions of interest, predefined in a target list (503.22, 541.175, 755.333, 793.289, 831.245) according to prior full scan runs, applying a mass tolerance of 25 ppm for precursor selection with a relative collision energy of 30%. Fragments were detected using the ion-trap mass analyzer. Acquired data was analyzed qualitatively by generation of extracted ion chromatograms applying a mass tolerance of 5 ppm using QualBrowser (Thermo Scientific).

Enzymatic assays and product analysis. Enzyme activity was routinely assayed in 100 mM HEPPS buffer at pH 8.5. The same buffer was used at pH 7.5 to 9.0, while a 100 mM PIPES buffer was used at pH 6.0 to 7.5. Activity was measured in a continuous photometric assay by directly following the formation or consumption of NADH at 340 nm (ε = 6.22 mM⁻¹ cm⁻¹) or 365 nm (ε = 3.4 mM⁻¹ cm⁻¹) [22], starting with 0.5 mM NAD⁺ or NADH, respectively. Because the various batches of purified enzyme showed different specific activities, the assays contained 40–200 µg of AdhB. The reactions were started by adding the respective substrates ethanol, n-propanol, acetaldehyde, or propionaldehyde (1 mM). Also tested, but not converted, were benzyl alcohol, butanol, 2-phenoxyethanol, pentanol, methanol, isopropanol, benzaldehyde, phenylacetaldehyde, and formaldehyde. For supplementation of Fe²⁺ ions, 100 µM iron-(II)-sulfate was used in the enzymatic assays. Tolerance to higher ethanol or n-propanol concentrations was tested accordingly. Plotting and evaluation of the assays and curve fitting was performed by the Prism 10.2.3 program package (GraphPad Software, Boston, MA, USA). Inactivation of AdhB was fitted using an equation for exponential decay, according to $v=\left(V^0-Plateau\right)\ast e^{-K \left[substrate\right]}+Plateau$; Michaelis–Menten enzyme kinetics with included substrate inhibition was calculated using the equation $v= {\displaystyle \frac{V_{max}\ast \left[substrate\right]}{K_m+\frac{\left[substrate\right]}{1+\frac{\left[substrate\right]}{K_i}}}}$ [23].

AdhB was used in a coupled assay with AOR, as demonstrated with BADH [24]. Briefly, 5–20 µg/mL of purified AOR and AdhB was mixed with 5 mM benzyl viologen (pre-reduced with Ti^III-citrate) or 5 mM benzaldehyde, respectively, 1 mM NADH, and 10 mM acetic acid in 100 mM HEPPS buffer at pH 8.0 under anaerobic conditions (H₂:N₂ atmosphere; 2.5%:97.5%) and incubated at room temperature. Samples were drawn every 30 min and incubated at 70 °C for 5 min before detection of the metabolites to denature the enzymes AOR and AdhB.

Benzaldehyde and benzoate concentrations were determined via HPLC as published previously [24]. Ethanol was detected by the Megazyme ethanol detection kit (Wicklow, Ireland) following the manufacturer’s instructions, or by HPLC-RID on an Agilent Infinity 1260 HPLC system (Agilent Technologies, Inc., Santa Clara, CA, USA) using a Rezex-ROA-Organic Acid H⁺ (8%) column (50 × 7.8 mm; Phenomenex, Torrance, CA, USA) at 60 °C. Samples for HPLC analysis were collected for each time point and diluted in a 1:1 ratio with sample buffer (0.005 N sulfuric acid in ddH₂O). Aliquots of 5 µL were analyzed using 0.005 N sulfuric acid in ddH₂O as mobile phase in isocratic mode with a flow rate of 0.6 mL/min. Detection in RID was performed at 40 °C (retention times: ethanol 3.6 min, acetaldehyde 2.4 min, acetic acid 2.2 min).

Recombinant gene expression and growth conditions in Aromatoleum spp. The broad host-range plasmid pASG103-mobori containing the adhB gene was transformed into the conjugation-competent E. coli strain WM3064 and conjugated into A. evansii strain KB740, as previously described [25]. The obtained recombinant A. evansii cells were grown on minimal TA media with individual carbon sources supplemented [25,26]; 5 mM succinate and varying concentrations of ethanol were used as carbon sources (1% ethanol corresponds to 17 mM). Strains under aerobic conditions were grown shaking at 250 rpm at 28 °C, strains under denitrifying conditions as standing cultures at 28 °C. Expression of adhB in recombinant A. evansii strains was induced by adding 200 ng/mL anhydrotetracyclin [25].

Sequence similarity and structure comparison. Proteins related to AdhB were identified by BLASTp searches against the protein database and tBLASTn searches (BLAST 2.17.0 version) against the core_nucleotide database. Moreover, the positions of AdhB and related proteins were determined by constructing a similarity tree with various other members of the Fe-dependent alcohol dehydrogenase/dehydroquinate synthase superfamily (conserved domains category cl02872), which were aligned using Clustal Omega (www.ebi.ac.uk/Tools/msa/clustalo and http://avermitilis.ls.kitasato-u.ac.jp/clustalo, accessed on 17 June 2025) with bootstrap values calculated in 1000 replications. A neighbor-joining tree was constructed based on the alignment, using the Program iTOL (http://itol.embl.de/).

An Alphafold structure of AdhB is available from the UniProt database (Q5P150), which was structurally aligned to the closest known relatives with solved structures (PDB numbers 3OWO, 2BI4, 1RRM), using ChimeraX 1.9.

3. Results

3.1. AdhB Is an Alcohol Dehydrogenase for Small Aliphatic Alcohols

AdhB containing either an N- or C-terminal strep tag fusion was overproduced and purified from recombinant E. coli cells containing expression plasmids with the cloned adhB gene (ebA4623), using affinity chromatography. Homogeneous preparations were obtained with yields of 5 mg protein (L of culture)⁻¹ of the N-terminally or 10.5 mg protein L⁻¹ of the C-terminally tagged variant (from 2 L of culture). Because of the higher yield, the C-terminally tagged variant was used for the further experiments. Starting with 786 mg of protein from 5 g wet cell mass, 8.6 mg of purified AdhB was obtained after affinity chromatography. While no activity was recorded in the cell extract, the various batches of purified protein showed specific activities of 16–90 mU (mg protein)⁻¹ (1 U equals 1 µmol min⁻¹) using 1 mM ethanol and NAD⁺ as substrates. Assays with NADP⁺ replacing NAD⁺ showed no activity. The enzyme was not affected by oxygen; therefore, all assays with AdhB were performed in a normal atmosphere.

AdhB consisted of a single subunit migrating between 35 and 40 kDa in SDS-PAGE, which fits to the expected mass of 40.4 kDa of the adhB gene product (including the strep-tag, Figure 1A). Different native masses were determined by gel filtration, Ferguson plot, and crosslinking analysis (Figure 1B,C and Figure S1). Gel filtration showed two peaks for AdhB, which corresponded to more abundant dimeric and less abundant tetrameric complexes (75 and 177 kDa, respectively; Figure 1C and Figure S1C). In contrast, the Ferguson plot only showed a dimeric complex (Figure S1), while the crosslink analysis showed the presence of dimeric, tetrameric, and hexameric complexes (Figure 1B and Figure S11). In the crosslink experiment, the oligomer masses were determined slightly smaller than expected, which may be explained by aberrant migration behavior in PAGE (Figure S1A). Therefore, we assume that AdhB consists primarily of homodimers, which also aggregate to tetramers or hexamers to some extent. The different methods used appear to detect the different oligomers to various extents.

The UV–Vis spectrum of purified AdhB showed an unusually broad maximum around 270 nm and some residual absorption at 305 to 360 nm (Figure 1D). The latter feature may be expected from an enzyme affiliated with the family III of Fe-dependent alcohol dehydrogenases, which often show a small peak at 330 nm due to the bound Fe²⁺ ion [27]. In contrast, the broad maximum ranging from 260 to 280 nm indicates the presence of an unknown cofactor, as recently observed for a benzyl alcohol dehydrogenase (BaDH) affiliated to the Zn-dependent family I after recombinant expression in E. coli [16]. As reported for BaDH, the cofactor can be extracted by acidic precipitation of the protein. After removing the protein by centrifugation, the supernatant exhibits an absorption maximum of 260 nm, suggesting a nucleotide-like molecule binding to the enzyme. We tried to evaluate its nature by high-resolution LC-MS/MS analysis of the supernatant and detected single-charged m/z ions of 831.2449, 793.2892, 755.3335, 514.1757, and 503.2201 in full scan runs. Fragmentation analysis targeting those ions revealed consistently appearing fragments of 251.2077 and 79.9770 m/z (see Supplementary Materials, Figure S2). As of now, we cannot correlate these data with any known molecule. Remarkably, some identical masses (particularly of the fragments) were previously recorded for the molecule bound to BaDH [16], suggesting that carrying this molecule his may be a common phenomenon for recombinantly expressed proteins in E. coli. After subtracting the spectrum of the supernatant from the AdhB spectrum, the maximum is at 280 nm. The expected absorption value based on the protein sequence lies between the value as measured for AdhB and that of the difference spectrum (expected absorption of 2.62 vs. recorded values of 2.78 before and 2.28 after subtracting the absorption of the extracted cofactor).

Elemental analysis by ICP-MS was performed with an enzyme batch with a specific activity of 23 mU (mg protein)⁻¹, showing the presence of 0.1 Fe, 0.06 Ni, and up to 0.13 Zn, as well as up to 1.9 P per subunit (

Table 1. ICP-MS analysis of AdhB. Values are given in atoms of element (subunit of AdhB)⁻¹.

	Desalted AdhB
Mg	0.01
P	1.88
Ca	0.03
Mn	0.02
Fe	0.10
Co	0.00
Ni	0.06
Cu	0.01
Zn	0.13
Se	0.00
Mo	0.00
W	0.00

). This is consistent with the affiliation of AdhB with the family of Fe-dependent alcohol dehydrogenases, although Fe appears to be present at only 10% occupancy and may be partially substituted with Ni or Zn. Supplementation of Fe²⁺ to the growth medium of recombinant cells, to the enzyme preparations, or the assay buffers did not show a beneficial effect on enzyme activity. The presence of 1.9 P per subunit is consistent with a potential nucleotide-like cofactor bound to the enzyme.

The enzyme was assayed for activity with various alcohols and aldehydes, as well as NAD⁺/NADH or NADP⁺/NADPH. AdhB only showed activity with NAD⁺ or NADH and the substrate spectrum was restricted to the oxidation of ethanol or n-propanol (specific activities 23 and 25 mU (mg protein)⁻¹) and to the reduction of acetaldehyde or propionaldehyde (specific activities 161 and 123 mU (mg protein)⁻¹), using the respective substrates at 1 mM and the same batch of purified AdhB. Substrates not converted by AdhB include methanol, n-butanol, n-pentanol, benzyl alcohol, 2-phenylethanol, isopropanol, formaldehyde, benzaldehyde, or phenylacetaldehyde. The pH dependency of the enzyme was determined in forward and reverse directions with a different, more active batch of AdhB, using ethanol/NAD⁺ and acetaldehyde/NADH as substrates. Ethanol oxidation rates decreased from 145 to 90 mU (mg protein)⁻¹ with increasing pH values from pH 6.0 to 9.0, whereas the rates of acetaldehyde reduction increased twofold from 250 to 500 mU (mg protein)⁻¹ from pH 6.0 to 8.5 and decreased again to 360 at pH 9.0 (Figure 2).

An enzyme kinetic analysis was performed for ethanol or n-propanol oxidation as well as for acetaldehyde and propionaldehyde reduction, using AdhB with 23 mU (mg protein)⁻¹ specific activity. As shown in Figure 3, the data fitted to the Michaelis–Menten equation revealed very similar behavior for the C2 and C3 substrates (

Table 2. Enzyme kinetics of purified AdhB. Values of k_cat refer to a single subunit.

Substrate	Ethanol	Acetaldehyde	n-Propanol	Propionaldehyde
App. Vmax [U/mg]	121.3	160.1	129.8	158.1
app. kcat [s−1]	79.7	105.1	85.2	103.8
app. Km [mM]	4.4	0.2	4.2	0.3
kcat/Km [mM−1 s−1]	18.1	525.5	20.3	346.1

). AdhB showed similar apparent V_max values for aldehyde reduction and for alcohol oxidation, but the calculated apparent K_m values were much higher for alcohol oxidation than for aldehyde reduction (Table 2).

The discrepancy of the apparent K_m values of forward and backward reaction results in higher catalytic efficiencies (k_cat/K_m) of AdhB for aldehyde reduction than for alcohol oxidation. This feature of AdhB may provide a “safety valve” to prevent the production of high concentrations of toxic aldehydes.

3.2. Tolerance to Alcohols

To evaluate its applicability even in highly concentrated alcohol solutions, AdhB was tested for its tolerance to high ethanol or n-propanol concentrations. To this end, alcohol oxidation assays with NAD⁺ have been set up with concentrations up to 40% (v/v; equal to 6.9 M for ethanol). Remarkably, the enzyme appeared to be hardly affected by the alcohols up to 3.5 M (20% v/v), exhibiting Michaelis–Menten-like kinetics with slight substrate inhibition. Only at higher concentrations did the activity of AdhB decrease exponentially with increasing alcohol concentrations, but still showed 9% of its maximum activity at 6.9 M (v/v) of ethanol and 22% of the maximum with n-propanol (Figure 4). The apparent K_m value for n-propanol was still close to that obtained from the study with lower substrate concentrations, while the apparent K_m value for ethanol (24.9 mM) was six-fold higher, probably caused by the fewer data points obtained at low concentrations in this study. Moreover, very weak conventional substrate inhibition (K_i = 18–25 M) has been observed in the concentration range between 0.5 and 3 M with both substrates, which is only visible in analyses with high enough alcohol concentrations (Figure 4).

3.3. AdhB Enables Aromatoleum evansii to Grow on Ethanol as a Carbon Source

Utilization of ethanol as a sole carbon source has been reported for A. aromaticum EbN1, but the closely related species A. evansii KB740 has been reported as being unable to grow on this substrate [26,28]. A comparison of the respective genomes indeed showed that the aldB-adhB operon or comparable genes somewhere else on the chromosome are completely missing in A. evansii (accession numbers GCA_000025965.1; GCA_012910805.1).

Since A. aromaticum should be able to grow with ethanol as a carbon source aerobically as well as under denitrifying conditions [26,28,29,30,31], we used the strain as a positive control for a complementation study in A. evansii. Since both species grow well on succinate, we used this substrate for additional control experiments (Figure 5A). We confirmed that A. aromaticum grows very well aerobically with either succinate or ethanol at different concentrations, ranging from 0.5 to 2% (w/v), while A. evansii only showed growth on succinate, but not on ethanol. However, if A. evansii expresses the adhB gene from A. aromaticum from an introduced plasmid, it shows an even better aerobic growth behavior on ethanol than the latter. In both cases, growth rates were slightly reduced in the experiments with 2% ethanol, probably indicating substrate toxicity effects (Figure 5). Under denitrifying conditions, the two control strains showed the same behavior for growth on ethanol (growth of A. aromaticum, no growth of A. evansii), but the recombinant A. evansii cells carrying the adhB gene did not reproducibly grow on ethanol. Because A. evansii lacks the co-expressed aldB gene, which codes for the subsequent aldehyde dehydrogenase [5,6], it needs to employ an endogenous aldehyde dehydrogenase for growth on alcohols, which apparently suffices under aerobic growth conditions, but may become limiting under denitrifying conditions, explaining our mixed results in anaerobic growth of the recombinant strain on ethanol.

3.4. Application of AdhB in Coupled Enzyme Reactions

We used AdhB to develop a potential application for enzyme-catalyzed conversion of acetate to ethanol, which may be used as a biofuel component [14,25]. For this purpose, we established a coupled reaction cascade of AdhB with the tungsten enzyme AOR, which catalyzes the oxidation of many different aldehydes to acids with NAD⁺ or BV as electron acceptors, but also the simultaneous reduction of organic acids to the corresponding aldehydes and of NAD⁺ to NADH with appropriate electron donors of low redox potential, such as Ti^(III)-citrate, Eu^(II)-EDTA, reduced hexa- or tetramethylviologen, or hydrogen [18,24,32]. We already demonstrated the feasibility of a similar reaction in a previous report by reducing benzoate to benzyl alcohol with AOR and a benzyl alcohol dehydrogenase, which was driven by hydrogen as reductant [16,24]. Instead of using H₂, we tried two alternative electron donor systems for powering the reaction in this study: either benzaldehyde as a sacrificial aldehyde, or benzyl viologen pre-reduced by Ti^(III)-citrate (Figure 6). The reductants were expected to allow AOR to reduce acetate to acetaldehyde and NAD⁺ to NADH, while AdhB further reduces acetaldehyde to ethanol at the expense of re-oxidizing NADH to NAD⁺. We indeed observed significant ethanol formation in both setups, demonstrating that either electron donor enables AOR to reduce acetic acid to acetaldehyde and to recycle NAD⁺ to NADH.

The system driven by benzaldehyde showed a decrease in benzaldehyde to about half of the starting concentration within 60 min, while ethanol was simultaneously formed at about half the rate of benzaldehyde decay (Figure 6A). During the next 60 min, only a very small increase in ethanol and no further decrease in benzaldehyde were recorded, indicating that the system reached an equilibrium state. Remarkably, the NADH concentration stayed almost the same during the first 60 min, but showed some decrease during the second 60 min. This indicates that NADH is regenerated with high efficiency during the first phase via AOR-mediated benzaldehyde oxidation and only shows net contribution to ethanol production during the second phase (Figure 6A).

The system driven by pre-reduced benzyl viologen (5 mM) showed continuous production of ethanol over 120 min, coupled with a slight increase in NADH concentration. Since we have confirmed previously that AdhB is not able to use viologens as electron donors, reduced benzyl viologen apparently acted as an electron donor for AOR, enabling it to reduce acetate to acetaldehyde and residual NAD⁺ to NADH and providing AdhB with both substrates required for ethanol synthesis (Figure 6B).

4. Discussion

We report here on AdhB from A. aromaticum, an NAD⁺-dependent enzyme of the iron-dependent ADH family III, which exhibits a very small substrate range, only catalyzing the oxidation of ethanol and n-propanol or the reduction of the respective aldehydes, while it is not active on larger aliphatic, aromatic, or secondary alcohols. Some ADHs of family III show similar substrate restrictions, such as Adh II from Zymomonas mobilis [33], while others turn over many more different substrates [34,35,36]. As previously reported for other recombinant ADHs of family III [27], the enzyme shows a relatively low specific activity and an iron content as low as 0.1 atoms per subunit. Therefore, only a fraction of the enzyme appears to be in the active, metal-bound state, even if it may also be active with Zn or Ni, which have been detected in substoichiometric amounts as well. The active-site Fe²⁺ could not be complemented with added Fe²⁺, suggesting that metal incorporation does not occur spontaneously and may require special conditions, which are not met in the heterologous expression host.

AdhB shows a homodimeric composition, resembling the homodimeric compositions of Adh II from Zymomonas mobilis [37,38] and 1,2-propanediol dehydrogenase from E. coli [39], which represent the closest related enzymes with known structures. However, the AdhB dimers also appear to assemble into larger complexes to some extent, as we observed tetrameric or even hexameric species in our gel filtration and crosslinking studies. The latter feature indicates that the quaternary structure AdhB may be in a state of flux between a dimeric state and the decameric state of 1,3-propanediol dehydrogenase from Klebsiella pneumoniae [40], probably caused by relatively weak interactions between the dimers. Both AdhB and Adh II also appear to retain significant activity at high alcohol concentrations [41], with AdhB still showing normal enzyme kinetics up to 20% alcohol concentrations (v/v). While the UV–Vis spectrum of AdhB showed no evidence of a bound NAD⁺ cofactor, the presence of an unknown nucleotide-like cofactor exhibiting an absorption maximum at 260 nm was detected. While we have not been able to identify this cofactor yet, we have previously encountered the same situation in another alcohol dehydrogenase from A. aromaticum EbN1 after recombinant production in E. coli [12]. Therefore, the presence of this cofactor may be a common feature for recombinant NAD(P)-binding proteins produced in E. coli. We presume that this feature became evident in working with AdhB or BaDH, because both of these proteins are completely devoid of any Trp residue, and only the recombinant variants contain the two Trp present in the twin-strep tags. This results in unusually low absorption values at 280 nm (extinction coefficients of AdhB: 8.94 and 19.94 mM⁻¹ cm⁻¹ without and with twin-strep tag, respectively, for BaDH see [16]), which become more distorted by a cofactor absorbing at 260 nm than in an enzyme with a more usual Trp content.

BLAST analysis revealed that from the sequenced strains of the genus Aromatoleum (NCBI taxid 551759), only Aromatoleum aromaticum strains EbN1 and pCyN1 and A. buckelii strain U120 contain adhB orthologues and degrade ethanol. The same trait is found in a few close relatives affiliated with the species Thauera aromatica or T. chlorobenzoica (see

Table 3. Growth on ethanol and presence of the adhB gene in the genus Aromatoleum and close relatives. +, positive; −, negative; ND, not determined.

Organism	Growth on Ethanol	adhB Gene Present (% Protein Identity)
A. aromaticum EbN1	+	+ (100)
A. aromaticum pCyN1	+	+ (100)
A. bremense PbN1	−	−
A. petrolei ToN1	+	−
Aromatoleum sp. strain EB1	ND	−
A. toluolicum T	+	−
A. diolicum 22Lin	−	−
A. evansii KB740	−	−
A. buckelii U120	+	+ (99)
A. anaerobium LuFRes1	+	−
A. tolulyticum Tol-4	+	−
A. toluvorans Td21	+	−
A. toluclasticum MF63	ND	−
Azoarcus indigens VB32	ND	−
Az. communis SWub3	+	−
Az. olearius BH72	+	−
Thauera aromatica K172/AR-1	+	+ (90)
T. chlorobenzoica 3CB1	ND	+ (90)
T. aromatica SP/LG356	ND	+ (89)

), indicating that adhB enables ethanol degradation in these strains. The other taxonomically described strains of the genera Aromatoleum, Thauera, or Azoarcus are lacking orthologues of the adhB gene in their genomes, although some of them still grow on ethanol [28,42] (see Table 3). Each of the latter strains contains at least two genes coding for uncharacterized “alcohol dehydrogenases” or “zinc-binding dehydrogenases”, which may substitute for adhB in ethanol oxidation. A. evansii KB740 is one of the strains devoid of any gene coding for AdhB or any other Fe-ADH and does not grow on ethanol, although it contains 14 genes for predicted alcohol dehydrogenases from other families (coding for three Zn-dependent ADHs, six SDR, three S-(hydroxymethyl)glutathione dehydrogenases, one acryloyl-CoA reductase, and one quinone oxidoreductase). We show here that recombinant expression of adhB from A. aromaticum is sufficient to enable A. evansii to grow on ethanol, confirming the proposed physiological role of the gene.

Synthetic pathways. Finally, we show here that AdhB may be applied as an auxiliary enzyme in synthetic pathways converting acetate or propionate to the corresponding alcohols, which may be used as fine chemicals or biofuels. An advantage of performing this reaction by coupling alcohol dehydrogenases with the tungsten enzyme AOR comes from the high versatility of AOR, both in regard to the range of acids reduced and different available sources of reducing equivalents [32]. We have shown previously that benzoate is converted to benzyl alcohol by AOR coupled to a benzyl alcohol dehydrogenase (BaDH), using H₂ as the sole reductant for AOR, which exhibits H₂-oxidizing hydrogenase side reactivity. This affords both reduction of benzoate to benzaldehyde and of NAD⁺ to NADH by AOR, while BaDH reduces benzaldehyde further with NADH as reductant, closing the reaction cycle [24]. AOR has also been used in a different setup to reduce acids in an electrochemical cell with electric current as reductant and hexamethyl viologen as a very low-potential redox mediator (E°’ = −610 mV) [43]. In this paper, we show that coupled assays with AOR and AdhB also convert acetate to ethanol with different reductants: either sacrificial aldehydes or even pre-reduced benzyl viologen (BV). While the first reaction may have been expected due to almost equal redox potentials of any acid/aldehyde pairs, the considerable production of ethanol with reduced BV has been unexpected because of the relatively high standard redox potential of BV (E°’ = −374 mV). Other AORs have been reported to require more redox-negative electron donors such as reduced tetramethyl viologen (E°’ = −536 mV) or methyl viologen (E°’ = −450 mV) [44,45,46] to afford acid reduction to the respective aldehyde. However, these considerations do not readily apply to the reaction observed here, which continues by reducing the aldehyde further to the alcohol. The calculated standard potentials of the redox pairs acetate/acetaldehyde (E°’ = −588 mV) and acetaldehyde/ethanol (E°’ = −202 mV) are almost equidistant to that of BV, resulting in an only slightly endergonic overall reaction according to the following equation:

Acetate + 4 BV^•+ + 5 H⁺ → Ethanol + 4 BV²⁺ + H₂O, ΔG°’ = +8.1 kJ mol⁻¹.

Since the coupled reactions have been set up under conditions favoring the desired direction (e.g., with high acid concentrations), the calculated overall energetics is consistent with the observed BV_red-dependent production of ethanol. We assume that the structural details of the particular AOR used in these experiments contribute to its flexibility in accepting reductants with very different redox potentials. AOR from A. aromaticum consists of a basal FAD-containing AorC subunit, which forms a complex with several AorAB protomers, each containing a tungsten cofactor and five Fe₄S₄ clusters. [34]. The AorAB protomers are stacked on top of AorC and each other, forming a filament-like quaternary structure where all redox cofactors are connected by electrically conductive chains of Fe₄S₄ clusters (“nanowires”) [47]. This may result in re-distributing the transferred electrons from an electron donor of intermediate potential, as in reduced BV (−374 mV), to acceptors exhibiting more positive potentials (−320 mV for NAD⁺/NADH, and −202 mV for acetaldehyde/ethanol) and those with more negative potential (−588 mV for acetate/acetaldehyde). We propose that the actual redox state of AOR can be fine-tuned by either filling up or depleting the Fe₄S₄ clusters of the nanowires with electrons, enabling even relatively weak electron donors like BV_red to initiate the reaction, provided it is allowed by the overall thermodynamics. In the case of the coupled assay, the exergonic reduction reactions of NAD⁺ (by AOR) and of the acetaldehyde intermediate (by AdhB) will then drive the endergonic reduction of acetate to acetaldehyde: AOR is predicted to first reduce most of the available NAD⁺ to NADH, producing a high enough NADH/NAD⁺ ratio to establish the low overall redox potential required for acid reduction. With AOR alone, this leads to an equilibrium state with very low aldehyde concentrations [16], but in the presence of an appropriate ADH, the aldehyde reacts with NADH, taking both compounds out of the equilibrium and driving the reaction forward towards reducing the acid to alcohol. Since the reactions occurring with the two enzymes are interdependent, this reaction system may be regarded as a special form of electron bifurcation [48].

In further refinement steps, the ethanol yield produced by the coupled system of AOR and AdhB may be increased further, both by optimizing the biochemical parameters and by adding mechanical modules for continuous alcohol extraction, e. g. by pervaporation [49], membrane distillation [50], or vapor-phase membrane filtration [51].

Structure prediction of AdhB. An Alphafold prediction of the AdhB structure is available (Q5P150), which was compared to known structures of other Fe-ADHs by structural alignment. The closest structural matches have been found with Adh II from Zymomonas mobilis (3OWO; [37]) or lactaldehyde reductase FucO from E. coli (1RRM or 2BI4; [39,52]). RMSD values were 0.754, 1.131, and 1.275 Å, respectively. An overlay of AdhB with Adh II is shown in Figure 7, indicating highly similar structural properties of the enzymes, such as the metal-binding residues of the active site, Asp194, His198, His263, and His277. Additional amino acids surrounding the active site cavity and implied in the substrate specificity of Adh II are completely conserved, including nonpolar (Phe149, Ile151, Ala162, Phe254, Leu259, and Ala361) and polar residues (His 267, Asp360, and Cys362). His 267 has been assigned a role in substrate binding and proton abstraction (with support from Asp361) and has been proposed to correctly position the substrates and restrict the entrance of larger substrates, together with Phe149 and Phe254 [37]. Interestingly, most of these residues are also conserved in other members of the family that are specific for small alcohols, e.g., AdhE of E. coli [53]. Moreover, residues implicated in NAD⁺ binding in Adh II or other related enzymes are highly conserved in AdhB, such as a Gly-rich pyrophosphate-binding motif at residues 96–99, a Thr138/Thr139 motif, and Leu 179 involved in binding the adenosine end of NAD⁺, or Thr147/Phe149 binding the nicotinamide residue. Finally, AdhB also carries a conserved Asp39 residue, which is considered the major determinant for NAD⁺ selectivity, while the NADP⁺-dependent enzymes within the superfamily usually contain a Gly at this position [37].

Similarity tree of the Fe-ADH superfamily. The relative position of AdhB in the Fe-ADH/dehydroquinate synthase superfamily (cl02872 in the conserved domains database) was investigated by analyzing the sequences of the enzyme and some related proteins together with representatives of all other subcategories of the superfamily. AdhB is affiliated to category cd08188, which includes other enzymes specific for short aliphatic alcohols like Adh II from Z. mobilis [54], but also 1,3-propanediol dehydrogenases involved in glycerol fermentation [55]. The structurally related lactaldehyde reductases (a.k.a. 1,2-propanediol dehydrogenases) are affiliated to the closely neighboring category cd08176 (Figure 8). Figure 8 also shows the principal division of the superfamily into 28 categories of “actual” Fe-ADHs, whereas three more basal categories contain the dehydroquinate and 2-deoxy-inosose synthases (DHQS-like; cd08169), glycerol-1-phosphate dehydrogenases (Gro1pDH-like; cd08549), and glycerol dehydrogenases (GroDH-like; cd08550). All four metal-binding residues are conserved in most of the Fe-ADH categories, except for the three basal groups and the four closest related categories (cd14864, 14866, 08192, and 08177). The three basal groups have completely or partially lost the first His of the metal-binding residues, and the Asp appears to be replaced by Glu in the DHQS-like proteins (cd08169). Category cd14864 represents proteins of unknown function from Spirochaetes containing none of the conserved residues, while the proteins of cd14866 are from halophilic bacteria, which may have adapted to high salt by replacing the Asp with Asn, as well as the first and last His with Lys and Gln, respectively. Finally, the maleylacetate reductases (MAR; cd08177) and the MAR-like proteins (cd08192) contain Asn and Arg, respectively, instead of Asp. Since all these exceptions are located at the base of the tree, AdhB is embedded in a part of the tree that likely contains exclusively active ADHs.

5. Conclusions

We show in this study that AdhB from A. aromaticum is an NAD⁺-dependent ethanol dehydrogenase of the iron-dependent family III, which only converts ethanol or n-propanol and apparently is involved in the degradation of these alcohols. Expression of the corresponding gene even complements the related species A. evansii for growth on ethanol. The enzyme shows similar or better kinetic parameters for forward and reverse reactions than many other characterized ADHs of family III [36,56,57], and exhibits significantly lower K_m values involved in aldehyde reduction than in alcohol oxidation. The same pattern has also been observed for the highly similar orthologue Adh II from Z. mobilis, but with about ten-fold lower K_m and ten-to-seventy times higher V_max values [54]. The relatively low V_max values of AdhB are apparently due to the low abundance of Fe in the purified enzyme. Furthermore, AdhB shows surprisingly high alcohol tolerance. Finally, we show that the enzyme can be used in a coupled reaction with the tungstoenzyme AOR to reduce acetate to ethanol.