Energy Conservation Associated with Ethanol Formation from H2 andCO2 in Clostridium autoethanogenum Involving Electron Bifurcation

Johanna Mock,a Yanning Zheng,a Alexander P. Mueller,b San Ly,b Loan Tran,b Simon Segovia,b Shilpa Nagaraju,b Michael Köpke,b

Peter Dürre,c Rudolf K. Thauera

Max Planck Institute for Terrestrial Microbiology, Marburg, Germanya; LanzaTech Inc., Skokie, Illinois, USAb; Institute of Microbiology and Biotechnology, University of Ulm,Ulm, Germanyc

ABSTRACT

Most acetogens can reduce CO2 with H2 to acetic acid via the Wood-Ljungdahl pathway, in which the ATP required for formateactivation is regenerated in the acetate kinase reaction. However, a few acetogens, such as Clostridium autoethanogenum, Clos-tridium ljungdahlii, and Clostridium ragsdalei, also form large amounts of ethanol from CO2 and H2. How these anaerobes witha growth pH optimum near 5 conserve energy has remained elusive. We investigated this question by determining the specificactivities and cofactor specificities of all relevant oxidoreductases in cell extracts of H2/CO2-grown C. autoethanogenum. Theactivity studies were backed up by transcriptional and mutational analyses. Most notably, despite the presence of six hydroge-nase systems of various types encoded in the genome, the cells appear to contain only one active hydrogenase. The active [FeFe]-hydrogenase is electron bifurcating, with ferredoxin and NADP as the two electron acceptors. Consistently, most of the otheractive oxidoreductases rely on either reduced ferredoxin and/or NADPH as the electron donor. An exception is ethanol dehydro-genase, which was found to be NAD specific. Methylenetetrahydrofolate reductase activity could only be demonstrated with arti-ficial electron donors. Key to the understanding of this energy metabolism is the presence of membrane-associated reducedferredoxin:NAD� oxidoreductase (Rnf), of electron-bifurcating and ferredoxin-dependent transhydrogenase (Nfn), and of acet-aldehyde:ferredoxin oxidoreductase, which is present with very high specific activities in H2/CO2-grown cells. Based on thesefindings and on thermodynamic considerations, we propose metabolic schemes that allow, depending on the H2 partial pressure,the chemiosmotic synthesis of 0.14 to 1.5 mol ATP per mol ethanol synthesized from CO2 and H2.

IMPORTANCE

Ethanol formation from syngas (H2, CO, and CO2) and from H2 and CO2 that is catalyzed by bacteria is presently a much-dis-cussed process for sustainable production of biofuels. Although the process is already in use, its biochemistry is only incom-pletely understood. The most pertinent question is how the bacteria conserve energy for growth during ethanol formation fromH2 and CO2, considering that acetyl coenzyme A (acetyl-CoA), is an intermediate. Can reduction of the activated acetic acid toethanol with H2 be coupled with the phosphorylation of ADP? Evidence is presented that this is indeed possible, via both sub-strate-level phosphorylation and electron transport phosphorylation. In the case of substrate-level phosphorylation, acetyl-CoAreduction to ethanol proceeds via free acetic acid involving acetaldehyde:ferredoxin oxidoreductase (carboxylate reductase).

Clostridium autoethanogenum (1) is industrially used in fer-mentations of syngas (H2/CO/CO2) or industrial off-gases,

like basic oxygen furnace (BOF) gas from steel manufacturing (2,3). It is a close relative of Clostridium ljungdahlii (4) and Clostrid-ium ragsdalei (5). The 16S RNA sequences between the threestrains are 99% similar; also, the gene arrangement and sequencein the Wood-Ljungdahl operon are almost identical (6). For C.autoethanogenum (7, 8) and C. ljungdahlii (9), complete genomesequences are available, and a comparative genomic analysis sug-gested some key differences (7). C. autoethanogenum can grow onCO as the sole carbon and energy source, forming mainly ethanoland acetic acid but also 2,3-butanediol, lactic acid, and some H2 asfermentation products (6, 10). This organism can also grow on H2

and CO2, albeit at lower rates and to lower cell concentrationsthan on CO. A characteristic of C. autoethanogenum, C. ljungdah-lii, and C. ragsdalei is that they grow optimally between pH 5 andpH 5.5 and that, besides acetic acid, they are able to form ethanolduring growth on H2 and CO2 (see reactions 1 and 2, below) (11).The three clostridia differ in this respect from the well-studiedAcetobacterium woodii, which has a pH growth optimum near 7and reduces CO2 with H2 to acetic acid (reaction 1) but not to

ethanol (reaction 2) (12). The three Clostridium species also differfrom A. woodii in that A. woodii is dependent on sodium ions,whereas the three clostridial species are not, and in that A. woodiiis unable to grow on CO in minimal medium (13) unless culturesare provided with formate (14). A common feature, however, isthe lack of cytochromes and of menaquinone, which distinguishthese acetogens from Moorella thermoacetica, which contains cy-tochromes and menaquinone (12, 15). M. thermoacetica, a ther-

Received 22 May 2015 Accepted 29 June 2015

Accepted manuscript posted online 6 July 2015

Citation Mock J, Zheng Y, Mueller AP, Ly S, Tran L, Segovia S, Nagaraju S, Köpke M,Dürre P, Thauer RK. 2015. Energy conservation associated with ethanol formationfrom H2 and CO2 in Clostridium autoethanogenum involving electron bifurcation.J Bacteriol 197:2965–2980. doi:10.1128/JB.00399-15.

Editor: W. W. Metcalf

Address correspondence to Rudolf K. Thauer, thauer@mpi-marburg.mpg.de.

Copyright © 2015, American Society for Microbiology. All Rights Reserved.

doi:10.1128/JB.00399-15

September 2015 Volume 197 Number 18 jb.asm.org 2965Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

mophile, can also grow on H2 and CO2 and form mainly aceticacid (16–18).

2 CO2 � 4 H2 → CH3COO� � H� � 2 H2O;

�G°� � �95 kJ ⁄ mol acetate (1)

2 CO2 � 6 H2 → CH3CH2OH � 3 H2O;

�G°� � �105 kJ ⁄ mol ethanol (2)

Under standard conditions (pH2 and pCO2 each at 105 Pa,[acetate] at 1 M, pH 7), ethanol formation from H2 and CO2 ismore exergonic than acetic acid formation from H2 and CO2.However, the opposite is true under physiological conditions,where the pH2 is lower than 104 Pa. This is because the free energychange of reaction 2, with six H2 molecules, decreases with the H2

concentration more than the free energy change of reaction 1,with four H2 molecules. The advantage of forming ethanol istherefore the avoidance of acidification and not based in thermo-dynamics.

It has recently been elucidated how A. woodii growing on H2

and CO2 conserves energy (12). The scheme shown in Fig. 1Aillustrates that in A. woodii, CO2 is reduced with H2 to acetic acidvia the Wood-Ljungdahl pathway,which involves formyltetrahy-drofolate (formyl-H4F), acetyl coenzyme A (acetyl-CoA), andacetyl-phosphate as “energy-rich” intermediates. The ATP re-quired for formate activation to formyl-H4F is quantitatively re-generated in the acetate kinase reaction. Net ATP formation isachieved via electron transport phosphorylation involving themembrane-associated complexes RnfA to -G and F1Fo ATP syn-thase. The exergonic reduction of NAD� with reduced ferredoxincatalyzed by the RnfA to -G complex is associated with the buildupof an electrochemical sodium ion potential (1 Na� translocatedper electron) (19), which in turn drives the phosphorylation ofADP via the F1Fo ATP synthase complex (3.3 Na� translocated per

ATP) (20). The reduced ferredoxin is regenerated via reduction offerredoxin and NAD� with 2 H2 catalyzed by the electron-bifur-cating and NAD-dependent [FeFe]-hydrogenase HydABCD (21).NADH is reoxidized via reduction of methenyl-H4F to methyl-ene-H4F and of methylene-H4F to methyl-H4F, reactions whichare catalyzed by NAD-specific methylene-H4F dehydrogenase(22) and NAD-specific methylene-H4F reductase (23), respec-tively. CO2 reduction to formate is catalyzed by a cytoplasmicformate-hydrogen-lyase complex (24). Via this scheme (Fig. 1A),a net 0.3 mol ATP is generated per mol of acetic acid formed fromH2 and CO2 (12). The low ATP gain of 0.3 mol ATP/mol of acetateis consistent with an H2 threshold concentration near 0.2% in thegas phase reported for the acetogen (25–27). However, growthyields corrected for maintenance (Ymax) of nearly 7 g of cells/molof acetate (26, 28) indicate that the ATP gain (n) could be some-what higher: Ymax � k·n·YATP

theor, with k representing the the yieldcoefficient and YATP

theor representing the theoretical amount ofcells per mole of ATP (29). YATP

theor has been calculated to beabout 15 g of cells/mol of ATP when starting with acetate and CO2

as carbon sources and H2 as electron donor (30–33). The yieldcoefficient k is generally less than 1 and usually between 0.4 and 0.8(29).

A. woodii does not form ethanol when growing on H2 and CO2

(12, 34). However, when growing on glucose, this bacterium canform ethanol in addition to acetic acid, indicating that the organ-ism has the potential to reduce acetyl-CoA to ethanol. Both theacetaldehyde dehydrogenase (CoA acetylating) and the ethanoldehydrogenase were found to be NAD specific (35). The CoA-linked acetaldehyde dehydrogenase has been characterized (36).The free energy change associated with reaction 2 at H2 concen-trations above 0.1% would, in principle, allow ATP formation.The inability to form ethanol during growth on H2/CO2 must

FIG 1 Energy metabolism of Acetobacterium woodii growing on H2 and CO2 at pH 7. (A) Metabolic scheme showing how acetic acid is formed from H2 and CO2

and which steps are associated with energy conservation (12). (B) Metabolic scheme assuming ethanol is formed from H2 and CO2, for which the overall reactionis thermodynamically feasible. However, ethanol formation from CO2/H2 does not occur, although enzymes required for ethanol formation are present, but withthese the ATP gain is negative as predicted with the scheme. For explanations, see the text. Hyd, electron-bifurcating and NAD-dependent [FeFe]-hydrogenase;Rnf, membrane-associated and energy-conserving reduced ferredoxin:NAD� oxidoreductase; ADH, alcohol dehydrogenase; ALD, coenzyme A-dependentacetaldehyde dehydrogenase.

Mock et al.

2966 jb.asm.org September 2015 Volume 197 Number 18Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

therefore be due to factors other than thermodynamics. A possibleexplanation is given in Fig. 1B, in which the scheme shown in Fig.1A has been expanded to include ethanol formation from acetyl-CoA. Via these reactions, the ATP gain (moles of ATP per mole ofethanol) in A. woodii would be negative (�0.1 mol ATP/mol eth-anol).

In contrast to A. woodii, C. autoethanogenum forms ethanol inaddition to acetic acid when growing on H2 and CO2 (1). Ourstudy addresses the questions of why and how ethanol can beformed. We identified five enzymes involved in ethanol formationfrom H2 and CO2 that are present in H2/CO2-grown C. autoetha-nogenum cells but, based on the genome sequence (27), are notpresent in A. woodii: an electron-bifurcating and NADP-depen-dent [FeFe]-hydrogenase (Hyt) (catalyzing reaction 3, below), anelectron-bifurcating and ferredoxin-dependent transhydrogenase(Nfn) (reaction 4), an NADP-specific methylene-H4F dehydroge-nase (FolD) (reaction 5), a methylene-H4F reductase (MetFV) forwhich the physiological electron donor has not yet been found(reaction 6), and an acetaldehyde:ferredoxin oxidoreductase(AOR) (reaction 7).

The finding that H2/CO2-grown cells of C. autoethanogenumcontain only an electron-bifurcating and NADP-dependent[FeFe]-hydrogenase (Hyt) (reaction 3) was unexpected, becausein vivo, NADP generally has a lower redox potential (NADP�/NADPH, �1) than NAD (NAD�/NADH, �1) (37), and thereforeNAD rather than NADP should be the thermodynamically fa-vored electron acceptor of electron-bifurcating hydrogenases(38). In vivo, the redox potential of ferredoxin in acetogens isprobably near �400 mV (39), while the in vivo redox potential ofNAD is near –300 mV (40) and that of NADP is near – 360 mV(37).

Nagarajan and colleagues stated that “a thorough understand-ing of various factors governing the metabolism, in particular en-ergy conservation mechanisms, is critical for metabolic engineer-ing of acetogens for targeted production of desired chemicals”(41), and this increased understanding was the motivation for ourperformed work.

MATERIALS AND METHODSChemicals. NAD�, NADP�, NADH, NADPH, FAD, FMN, thiamine py-rophosphate, coenzyme A, pyruvate, glyceraldehyde-3-phosphate (GAP),acetaldehyde, H4F, benzyl viologen (BV), dithiothreitol (DTT), and phos-photransacetylase from Bacillus stearothermophilus were obtained fromSigma-Aldrich Chemie GmbH (Taufkirchen, Germany). Methyl-H4F wasobtained from Schircks Laboratories (Jona, Switzerland). Methylene-H4Fwas prepared from H4F and formaldehyde by spontaneous reaction. H2,N2, and CO with highest purity (99.996%) were sourced from Messer(Düsseldorf, Germany) and Air Liquide Speciality Gases USA LLC (Hous-ton, TX). [14C]K2CO3 (1 mCi in 4 ml H2O; 9.25 MBq/ml) was obtainedfrom Hartmann Analytic (Braunschweig, Germany). Ferredoxin (Fd)(42) and ferredoxin-dependent monomeric [FeFe]-hydrogenase (43)were purified from Clostridium pasteurianum DSM 525. Ferredoxin of C.pasteurianum harbors two [4Fe-4S] clusters, each being reducible by oneelectron and both having almost identical redox potentials near �400 mV(44).

Clostridium autoethanogenum. C. autoethanogenum DSM 10061 wasobtained from the Deutsche Sammlung von Mikroorganismen undZellkulturen GmbH, Braunschweig, Germany. The Gram-positive aceto-gen was grown strictly anaerobically at pH 5.3 and 37°C with either H2/CO2, fructose, or CO as carbon and energy sources. Growth was followedturbidometrically at 600 nm. An optical density at 600 nm (OD600) of 1was found to correspond to about 0.4 g (dry mass) of cells per liter.

Growth on H2/CO2. C. autoethanogenum was grown in a 1.3-litercontinuous retentostat culture (45) using the BioFlo110 system from NewBrunswick Scientific, in which a portion of the cells was recycled using ahollow fiber membrane (0.2-�m pore size, 1,200-cm2 surface area) fromGE Healthcare. The dilution rate for the cells was 0.5 per day, and that forthe metabolites was 4.9 per day. Mineral salts medium (0.4 g MgCl2 ·6H2O, 0.294 g CaCl2 · 6H2O, 0.15 g KCl, 0.12 g NaCl per liter) was sup-plemented per liter with 10 ml of a vitamin solution (20 mg biotin, 20 mgfolic acid, 10 mg pyridoxine hydrochloride, 50 mg thiamine, 50 mg ribo-flavin, 50 mg nicotinic acid, 50 mg D-(�)-pantothenate, 50 mg vitaminB12, 50 mg p-aminobenzoic acid, 50 mg lipoic acid per liter). In additionto the usual trace elements (100 �M FeCl2, 2 �M CoCl2, 5 �M NiCl2, 5�M ZnCl2, 2 �M MnCl2, and 2 �M Na2MoO4), the medium contained 2�M selenite (Na2SeO3) and 2 �M tungstate (Na2WO4). Ammonium ac-etate (3.083 g/liter) and hydrogen sulfide (0.3 ml of a 0.2 M Na2S solutionper liter per hour) served as nitrogen and sulfur sources, respectively. Thecultures were continuously gassed at a rate of up to 350 ml per min with65% H2, 9.2% N2, 23% CO2 as sole energy and carbon sources. The pHwas continuously recorded and maintained at pH 5.3 by the addition ofNH4OH. Metabolites were measured by high-performance liquid chro-matography (HPLC), and headspace gas was analyzed using a Micro gaschromatograph (46). The steady-state cell concentration in the continu-ous culture of C. autoethanogenum was about 1.83 g (dry mass) of cells/liter, that of produced acetate was 7.48 g/liter (0.125 M), and that ofproduced ethanol was 6.3 g/liter (0.14 M) (concentration in the broth notconsidering stripping, by which some ethanol is lost). Before harvest, theculture was adjusted from pH 5.3 to pH 6 with K2CO3 and subsequentlychilled in an ice water bath. The culture was centrifuged in an anaerobicchamber in 1-liter bottles at 5,000 � g for 10 min. The pelleted cells wereresuspended in 30 ml of 50 mM potassium phosphate (pH 7.0) containing10 mM DTT, pelleted again by centrifugation, snap-frozen in liquid ni-trogen, and stored at �80°C under N2.

Growth on fructose. C. autoethanogenum was grown on fructose (4.5g/liter) at pH 5.7 (initial pH) in a 0.2-liter batch culture in a sealed 1-literSchott bottle with 100% N2 as the gas phase. The mineral salts mediumwas identical to that used for growth on H2/CO2. It was bufferedwith 20 gMES [2-(N-morpholino) ethanesulfonic acid; pKa 6.15] added per liter.Under these conditions, mainly acetic acid (2 g/liter at the time of harvest)and much less ethanol (0.2 g/liter at the time of harvest) were formed asproducts. Growth was linear rather than exponential, probably becauseconditions were not as controlled as in bioreactors; for example, the pH inthe culture continuously dropped. At the time of harvest, the doublingtime was near 15 h and the cell concentration was 0.64 g (dry mass) ofcells/liter.

Growth on CO. C. autoethanogenum was grown at pH 5 in a 1.2-litercontinuous culture at a dilution rate of 1.8 per day (td � 9.2 h) (46). Thecultures were continuously gassed at a final rate of 350 ml per min with abasic oxygen furnace (BOF) gas stream (composition, 42% CO, 36% N2,20% CO2, and 2% H2; collected at Glenbrook steel mill in New Zealand).The pH was continuously recorded and maintained at pH 5 by the addi-tion of NH4OH. Several days after steady state was reached, the cells wereharvested as described previously (46). Since the cell and metabolite con-centrations fluctuated considerably for reasons not yet fully understood,one batch of cells was harvested in September 2012 and one batch washarvested in October 2012 for enzyme activity analysis. The enzyme ac-tivities reported in 2013 (46) were those of the cells harvested in Septem-ber 2012, and the properties of the electron-bifurcating hydrogenase re-ported in the same paper (46) were those of the enzyme purified from thecells harvested in October 2012.

Preparation of cell extracts. Frozen cells of C. autoethanogenum (2 gwet mass) were suspended in 3.25 ml of anoxic 50 mM potassium phos-phate (pH 7.0) containing 2 mM DTT, 5 �M FAD, and 5 �M FMN. Afteradding 0.5 mg DNase I and 0.5 mM MgCl2, the cell suspension was passedthrough a prechilled French pressure cell three times at 120 MPa. Beforethis step, the French pressure cell was first flushed with N2 for 15 min and

Ethanol Formation from H2 and CO2

September 2015 Volume 197 Number 18 jb.asm.org 2967Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

then washed three times with anoxic buffer. Unbroken cells and cell debriswere removed by centrifugation at 20,000 � g and 4°C for 30 min. Thesupernatant was used for enzyme assays. The membrane fraction wasobtained by centrifugation at 150,000 � g and 4°C for 60 min (46).

Specific activity measurements. Enzyme activities were measured un-der strictly anoxic conditions at 37°C in 1.5-ml anaerobic cuvettes[NAD(P), ferredoxin, and viologen dye reduction) or 6.5-ml anaerobicserum bottles (H2 formation and CO2 reduction) sealed with rubber stop-pers and filled with 0.8 ml assay mixture and N2 (100%), H2 (100%), orCO (100%) at 1.2 � 105 Pa as the gas phase. After the start of the reactionwith enzyme, the reduction of NADP� or NAD� and the oxidation ofNADPH or NADH were monitored spectrophotometrically at 340 nm(ε � 6.2 mM�1 cm�1) or at 380 nm (ε � 1.2 mM�1 cm�1). C. pasteur-ianum ferredoxin reduction was monitored at 430 nm (εox-red 13.1mM�1 cm�1), and benzyl viologen reduction was monitored at 578 nm(ε � 8.6 mM�1 cm�1). The reduction of 14CO2 to [14C]formate wasfollowed by quantifying 14C radioactivity in a Beckman LS6500 liquidscintillation counter (Fullerton, CA). One unit was defined as the transferof 2 micromoles of electrons per minute. Protein levels were determinedusing the Bio-Rad protein assay (Munich, Germany) with bovine serumalbumin as the standard (46).

Hydrogenase (reaction 3). For determination of hydrogenase specificactivity, the assay mixtures contained 100 mM Tris-HCl (pH 7.5), 2 mMDTT and, where indicated, 25 �M ferredoxin, 1 mM NAD�, or 1 mMNADP�. The gas phase was 100% H2.

Ferredoxin-dependent transhydrogenase (reaction 4). For ferre-doxin-dependent transhydrogenase specific activity measurements, theassay mixtures contained 100 mM morpholinopropanesulfonic acid(MOPS)-KOH (pH 7.0), 10 mM 2-mercaptoethanol, 10 �M FAD, 0.5mM NADP�, 40 mM glucose-6-phosphate, and 2 U glucose-6-phosphatedehydrogenase (NADPH-regenerating system), 10 mM NAD�, andabout 25 �M ferredoxin. N2 was the gas phase.

Reduced ferredoxin:NAD(P)� oxidoreductase. For determinationof reduced ferredoxin:NAD(P)� oxidoreductase activity, the assay mix-tures contained 100 mM potassium phosphate (pH 7.0), 2 mM DTT, 15�M ferredoxin, and 1 mM NAD� or NADP�. The gas phase was 100%CO for the reduction of ferredoxin via the ferredoxin-dependent CO de-hydrogenase in the cell extract (19).

CO dehydrogenase. The assay mixtures for CO dehydrogenase activ-ity measurements contained 100 mM Tris-HCl (pH 7.5), 2 mM DTT, andabout 30 �M ferredoxin, 1 mM NAD�, or 1 mM NADP�. The gas phasewas 100% CO.

Formate-hydrogen lyase. When the formation of H2 from formate(formate hydrogen lyase activity) was followed, the assay mixtures con-tained 100 mM Tris-HCl (pH 7.5), 2 mM DTT, and 20 mM sodiumformate. The gas phase was 100% N2. The serum bottles were continu-ously shaken at 200 rpm to ensure H2 transfer from the liquid phase intothe gas phase. After start of the reaction with cell extract, gas samples (0.2ml) were withdrawn every 1 min, and H2 was quantified by gas chroma-tography (46).

When the reduction of CO2 with H2 to formate was measured, theassay mixtures contained 100 mM potassium phosphate (pH 7.0), 2 mMDTT, and 30 mM [14C]K2CO3 (24,000 dpm/�mol). The gas phase was100% H2. The serum bottles were continuously shaken at 200 rpm toensure equilibration of the gas phase with the liquid phase. After start ofthe reaction with cell extract, 100-�l liquid samples were withdrawn every1.5 min and added to a 1.5-ml safe-seal microtube containing 100 �l of150 mM acetic acid to stop the reaction by acidification. The 200-�l mix-ture was then incubated at 40°C for 10 min with shaking at 1,400 rpm in aThermomixer (type 5436; Eppendorf, Germany) to remove all 14CO2,leaving behind the [14C]formate formed. Subsequently, 100 �l of the mix-ture was added to 5 ml of Quicksave A scintillation fluid (Zinsser Analytic,Frankfurt, Germany) and analyzed for 14C radioactivity in a BeckmanLS6500 liquid scintillation counter (Fullerton, CA).

Formate dehydrogenase. The assay mixtures for determination offormate dehydrogenase activity contained 100 mM Tris-HCl (pH 7.5), 2mM DTT, 20 mM sodium formate and, where indicated, 25 �M ferre-doxin, 1 mM NADP�, or 1 mM NAD�. The gas phase was 100% N2.

Methylene-H4F dehydrogenase (reaction 5). For methylene-H4F ac-tivity determinations, the assay mixtures contained 100 mM MOPS-KOH(pH 6.5), 50 mM 2-mercaptoethanol, 0.4 mM H4F, 10 mM formaldehyde,and 0.5 mM NADP� or 0.5 mM NAD�. The gas phase was 100% N2. Afterthe start of the reaction with the enzyme, the formation of NAD(P)H andmethenyl-H4F was followed at 350 nm using an ε of 30.5 mM�1 cm�1 [5.6mM�1 cm�1 for NAD(P)H plus 24.9 mM�1 cm�1 for methenyl-H4F].

Methylene-H4F reductase (reaction 6). For measurement of methyl-ene-H4F reductase activity, the assay mixtures contained 100 mM Tris-HCl (pH 7.5), 20 mM ascorbate, 10 �M FAD, 20 mM benzyl viologen, and1 mM methyl-H4F. Before the start of the reaction with enzyme, benzylviologen was reduced to an A578 of 0.3 with sodium dithionite. The gasphase was 100% N2.

When the oxidation of NAD(H) was to be followed, the assay mixturecontained potassium phosphate (pH 7.0), 1 mM methylene-H4F, 0.5 mMNADH or NADPH (as indicated), and about 25 �M ferredoxin. The gasphase was 100% N2.

Acetaldehyde:ferredoxin oxidoreductase (reaction 7). The assaymixtures for determination of acetaldehyde:ferredoxin oxidoreductaseactivity contained 100 mM Tris-HCl (pH 7.5), 2 mM DTT, 1.1 mM acet-aldehyde, and about 25 �M ferredoxin. The gas phase was 100% N2.

Alcohol dehydrogenase (reaction 8). For measurement of the specificactivity of alcohol dehydrogenase, the assay mixtures contained 100 mMpotassium phosphate (pH 6), 2 mM DTT, 1.1 mM acetaldehyde, and 1mM NADPH or 1 mM NADH. The gas phase was 100% N2.

Acetaldehyde dehydrogenase (CoA acetylating) (reaction 9). The as-say mixture for measurement of acetaldehyde dehydrogenase activitycontained 100 mM Tris-HCl (pH 7.5), 2 mM DTT, 1.1 mM acetaldehyde,1 mM coenzyme A, and 1 mM NADP� or 1 mM NAD�. The gas phasewas 100% N2.

Pyruvate:ferredoxin oxidoreductase (reaction 10). For determina-tion of pyruvate:ferredoxin oxidoreductase activity, the assay mixturecontained 100 mM Tris-HCl (pH 7.5), 2 mM DTT, 10 mM pyruvate, 1mM coenzyme A, 0.1 mM thiamine pyrophosphate, 25 �M ferredoxin,and 1 mM NADP� or 1 mM NAD�. The gas phase was 100% N2.

Glyceraldehyde phosphate dehydrogenase. For assay of glyceralde-hyde phosphate dehydrogenase activity, the assay mixture contained 50mM Tricine-NaOH (pH 8.5), 10 mM 2-mecaptoethanol, 10 mM potas-sium phosphate, 1 mM glyceraldehyde-3-phosphate, and 1 mM NADP�

or 1 mM NAD�. The gas phase was 100% N2.RNA-seq analyses. Transcriptome sequencing (RNA-seq) analaysis

was performed on 20-ml fermentation samples, either from a continuousculture sparged with 42% CO, 36% N2, 20% CO2, and 2% H2 or frombatch cultures grown on 65% H2 and 23% CO2 (N2 as the balance) in aclosed system (serum bottles). The samples were centrifuged at 4,000 rpmfor 10 min at 4°C. The supernatants were discarded, and the pellets werestabilized by the addition of 5 ml of RNAlater (Ambion Inc.). Total RNAwas isolated from the cell pellets by using a RiboPure bacteria kit (AmbionInc.) according to the manufacturer’s standard protocol. DNA was re-moved using the Turbo DNA-free kit (Ambion Inc.), and RNA qualitywas assessed by using a 2100 Bioanalyzer (Agilent Technologies). RNAconcentration was determined with a Nanodrop 2000 apparatus (ThermoFischer Scientific, Waltham, MA). Ribodepletion was conducted using theMicrobe Express kit (Ambion Inc.). cDNA libraries were prepared andsequenced by standard procedures using either SOLiD2 or Illumina Hi-Seq-2000 sequencing technology. Processed reads were mapped to a ref-erence assembly, displayed, and manually inspected in the Geneious ge-nome browser, v7.0.3 (Biomatters). Transcript abundance was estimatedbased on the fragments per kilobase of exon per million fragmentsmapped (FPKM).

Mock et al.

2968 jb.asm.org September 2015 Volume 197 Number 18Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

Construction of strains with disrupted genes. Genes were disruptedby using the ClosTron system (47). Constructs were designed using theWeb-based ClosTron design tool, and intron-targeting plasmids weresynthesized by the company DNA 2.0 (Menlo Park, CA). Transformationof C. autoethanogenum was achieved by bacterial conjugation using Esch-erichia coli strain HB101 (R702) as donor according to previously pub-lished methods (48, 49). Briefly, a 2-ml culture of E. coli strain HB101(R702) bearing the target plasmid was grown to an OD600 of 0.5, washedwith phosphate-buffered saline, and mixed with 0.2 ml of a C. autoetha-nogenum culture grown to an OD600 of 0.5. The cell mixture was spottedonto 1.5% agar plates containing 5 g fructose/liter and incubated at 37°Cfor 24 h under 1.7 � 105 Pa gas (gas composition: 42% CO, 36% N2, 20%CO2, and 2% H2) in gas-tight jars. Cells were subsequently transferredonto agar plates containing 5 g fructose/liter, 10 mg trimethoprim/liter forcounterselection of E. coli, and antibiotics for selection of the plasmid andgroup II intron insertion (15 mg thiamphenicol/ml and 5 mg clarithro-mycin/ml, respectively). Single colonies were restreaked and thenscreened using colony PCR (50).

Nucleotide sequence accession numbers. The genome sequences ofC. autoethanogenum and C. ljungdahlii have been deposited withGenBank under the accession numbers NC_022592.1 and NC_014328.1,respectively.

RESULTS

Clostridium autoethanogenum was grown on H2/CO2 at a ratio of2.82 (65% H2, 23% CO2, 9% N2) in continuous culture at pH 5.3,37°C, in a 2-liter bioreactor (1.3-liter volume) equipped with amembrane system which allowed the retention of 90% of the cells.The apparent dilution rate that the cells experienced was 0.5 day�1

(td � 33 h), and the apparent dilution rate of the metabolites was4.9 day�1. The steady-state concentration of cells was kept near1.83 g/liter, that of acetate was 7.5 g/liter (0.125 M), and that ofethanol was 6.3 g/liter (0.134 M). Under these growth conditions,the formation of 2,3-butanediol and/or lactic acid, productsformed when CO is the substrate, was not observed. Besides eth-anol and acetic acid, the HPLC analyses revealed no other fermen-tation products. The culture produced (per day) 0.9 g cells (0.04mol C)/liter, 36 g (0.6 mol) acetic acid/liter (0.23 �mol min�1 mgof cells�1), and 31 g (0.65 mol) ethanol/liter (0.25 �mol min�1 mgof cells�1). From these data and considering the stoichiometriesgiven by reactions 1 and 2, a growth yield of about 0.22 g of cellsper mol H2 and a specific rate of H2 consumption of 2.4 �molmin�1 mg of cells�1 were calculated.

The specific rate of H2 consumption indicated that in cell ex-tracts of the H2/CO2-grown organism, hydrogenase should have aspecific activity near 4 �mol min�1 mg of protein�1 and that theoxidoreductases involved in CO2 reduction to acetate and ethanolshould have specific activities of about 0.8 �mol min�1 mg ofprotein�1 (assuming 50% of the cell mass to be protein). For al-most all of the enzymes, this was found to be the case (Table 1),with exception of the specific activity of energy-conserving re-duced ferredoxin:NAD� oxidoreductase (Rnf), which was only0.1 �mol min�1 mg of protein�1. The reason for the low Rnfspecific activity could be that Rnf is oxygen sensitive and difficultto assay (19). This interpretation was substantiated by the findingthat the rnf genes in C. autoethanogenum belong to those mosthighly expressed (see below), indicating a catabolic role. The spe-cific activity of the electron-bifurcating hydrogenase was only 0.7�mol ferredoxin reduced by H2 per min per mg of protein, equiv-alent to 1.4 �mol H2 oxidized per min per mg of protein (reaction3), which is only one-third of that expected. The specific activity ofthe electron-bifurcating ferredoxin-dependent transhydrogenase

(Nfn) was only 0.1 �mol min�1 mg of protein�1 (Table 1), butthis low specific activity was consistent with the proposed functionof this enzyme (see Discussion).

In Table 1, the specific activities of the oxidoreductases foundin cell extracts of H2/CO2-grown cells of C. autoethanogenum arecompared with those found in fructose-grown and CO (BOF gas)-grown cells. At the time of harvest, the fructose-grown cells in thebatch culture had a doubling time of near 15 h, fermented fructoseat a specific rate of 0.02 �mol min�1 mg of cells�1, and producedacetic acid a specific rate of about 0.06 �mol min�1 mg of cells�1.The rate of ethanol formation was less than 10% of that of aceticacid. 2,3-Butanediol was not formed in significant amounts. Atthe time of harvest, the CO-grown cells in the continuous culturehad a doubling time near 5 h, generating ethanol at a specific rateof 0.2 �mol min�1 mg of cells�1. Besides ethanol, acetic acid (0.1�mol min�1 mg of cells�1) and 2,3-butanediol (0.12 �mol min�1

mg of cells�1) were formed. CO was consumed at a specific rate ofabout 2 �mol min�1 mg of cells�1 (46).

Hydrogenase (Hyt). Cell extracts of H2/CO2-grown cells cata-lyzed the reduction of ferredoxin with H2 only in the presence ofNADP� and vice versa (Table 1), indicating the presence of anelectron-bifurcating, NADP- and ferredoxin-dependent hydroge-nase catalyzing reaction 3. NAD� was not reduced by H2 in eitherthe absence or presence of ferredoxin. Only in the extracts of fruc-tose- and CO-grown cells was there some hydrogenase activitywith ferredoxin alone, indicating the additional presence of a non-electron-bifurcating hydrogenase specific for ferredoxin.

2 H2 � Fdox � NADP�º Fdred2� � NADPH � 3 H�

(3)

The electron-bifurcating, NADP- and ferredoxin-dependent[FeFe]-hydrogenase was recently purified from CO-grown C. au-toethanogenum cells and characterized as an electron-bifurcating[FeFe]-hydrogenase composed of the subunits HytABCDE1E2(HytA-E) and catalyzing the reversible reduction of NADP� andFdox with 2 H2 (reaction 3) (46). HytA is the H-cluster-harboringsubunit, HytB is an iron-sulfur flavoprotein harboring the NADPbinding site, and the other subunits are iron-sulfur proteins. Thehydrogenase forms a tight complex with the selenium- and tung-sten-dependent formate dehydrogenase FdhA (HytA-E/FdhA).The hydrogenase (HytABCDE1E2) is encoded by the genesCAETHG_2794-99 (hytCBDE1AE2), which form a transcriptionunit (46). RNA-seq analyses revealed that these genes belong tothe most highly expressed genes in C. autoethanogenum grown onCO-rich BOF gas (42% CO, 36% N2, 20% CO2, and 2% H2), withall genes among the 100 highest-expressed genes. The expressionanalysis revealed higher expression of the genes encoding the sub-units HytE1, -A, and -E2 (405 FPKM average across all genes) overthe subunits HytC, -B, and -D (312 FPKM) (Fig. 2). The geneswere also highly expressed during growth of C. autoethanogenumon H2/CO2 (65% H2, 9.2% N2, 23% CO2) and fructose, as de-duced from available transcriptomics data sets for C. ljungdahliigrown on H2/CO2 or fructose (41).

Besides the structural genes, the other genes required for[FeFe]-hydrogenase maturation, hydE (CAETHG_1691; 47FPKM), hydF (CAETHG_2063; 21 FPKM), and hydG(CAETHG_0339; 148 FPKM) are present and show expressionunder all conditions tested in C. autoethanogenum as well as in C.ljungdahlii (41).

The genome of C. autoethanogenum harbors genes for two

Ethanol Formation from H2 and CO2

September 2015 Volume 197 Number 18 jb.asm.org 2969Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

other multisubunit [FeFe]-hydrogenases (CAETHG_1576-78and CAETHG_3569-71), which have subunit structures and com-positions very similar to those of the electron-bifurcating andNAD- and ferredoxin-dependent [FeFe]-hydrogenase from Ther-motoga maritima (51), A. woodii (21), M. thermoacetica (52), and

Ruminococcus albus (53). While during growth of C. autoethano-genum on CO-rich BOF gas CAETHG_3569-71 is hardly ex-pressed (2 FPKM), CAETHG_1576-78 is the second-highest ex-pressed hydrogenase in C. autoethanogenum but at a significantlylower level (35 FPKM) than the characterized electron-bifurcating

TABLE 1 Oxidoreductase activities in cell extracts of C. autoethanogenum grown on either H2/CO2, fructose, or CO

Oxidoreductase Substratesa

Sp actb (U/mg of protein) in cells grown on:

H2/CO2 Fructose COc

Hydrogenase (electron bifurcating) H2 � NADP� � Fdox 0.7 1.4 1.1/2.0H2 � NAD� � Fdox �0.1 �0.1 �0.01H2 � Fdox �0.1 0.2 0.01/0.5H2� NADP� �0.1 0.07 0.2H2� NAD� �0.1 0.04 �0.1

Transhydrogenase (Nfn; ferredoxin dependent) NADPH � NAD� � Fdox 0.1 1.1 0.7/1.0NADPH � Fdox 0.01 0.1 0.1

Fdred2�:NAD� OR (Rnf)d CO � Fdox � NAD� 0.1 0.2 0.6

CO � Fdox � NADP� 0.01 0.14 0.01/0.6

CO DH CO � Fdox 5.6 7.0 2.7/2.3CO � NADP� 0.01 0.03 0.1CO � NAD� 0.01 0.02 0.2

Formate-H2 lyase HCOO� � H� (H2 formation) 0.9 3.8 2.4/7.2HCOO� � Fdox � H� 0.9 4.0 2.5HCOO� � Fdox � NADP� � H� 0.9 3.5 2.5H2 � CO2 (formate formation) 1.0 4.0 2.5

Formate DH HCOO� � NADP� � Fdox 0.4 1.2 1.1/1.6HCOO� NAD� � Fdox 0.02 �0.01 �0.1HCOO� � Fdox �0.01 0.7 0.01/0.7HCOO� � NADP� 0.1 0.05 0.2HCOO� � NAD� �0.1 0.2 �0.1

Methylene-H4F DH Methylene-H4F � NADP� 0.7 2.4 1.5/5.9Methylene-H4F � NAD� �0.01 �0.01 �0.01

Methylene-H4F reductase Methyl-H4F � BV 2.8 1.0 1.9/1.9Methylene-H4F � NADPH �0.1 �0.1 �0.1Methylene-H4F � NADH �0.1 �0.1 �0.1Methylene-H4F � NAD(P)H � Fdox 0.1 0.1 0.1

Acetaldehyde:Fd OR Acetaldehyde � Fdox 8 1.5 1.9/20Alcohol DH Acetaldehyde � NADH 0.2 0.1 0.2/0.2

Acetaldehyde � NADPH �0.01 0.05 0.15/0.2

Acetaldehyde DH Acetaldehyde � CoA � NADP� 0.1 0.03 0.1/0.01Acetaldehyde � CoA � NAD� 0.04 0.08 0.06/0.1

Pyruvate:Fd OR Pyruvate � CoA � Fdox 0.2 0.6 0.2/0.8Pyruvate � CoA � NAD(P)� 0.04 0.02 0.01

GAPDH GAP � Pi � NAD� 0.1 1.6 0.4/0.15GAP � Pi � NADP� 0.03 0.2 0.01/0.2

a Underlined portions indicate substrates or products whose reduction, oxidation, or formation was monitored. Boldface indicates the relevant substrates and specific activities. Theactivities were determined at 37°C in cell extracts (20,000 � g supernatant) prepared with a French press. For pH and substrate concentrations, see Materials and Methods. DH,dehydrogenase; OR, oxidoreductase; Fd, ferredoxin from C. pasteurianum; BV, benzyl viologen; H4F, tetrahydrofolate; GAP, glyceraldehyde-3-phosphate.b One unit equals 2 micromoles of electrons transferred per minute.c BOF gas was the source of CO and was composed of 42% CO, 36% N2, 20% CO2, and 2% H2. Cells were from continuous cultures that were harvested in September 2012 (andalso October 2012 when two values are reported).d Ferredoxin was reduced by CO via the CO dehydrogenase activity present in the cell extracts. In cell extracts prepared with a French press, most of the activity was associated withthe soluble fraction. When the cell extracts were obtained by lysis of the cells with lysozyme, almost 100% of the activity was membrane associated.

Mock et al.

2970 jb.asm.org September 2015 Volume 197 Number 18Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

and NADP-dependent [FeFe]-hydrogenase. CAETHG_1576-78is absent in C. ljungdahlii.

The genome of C. autoethanogenum harbors genes fortwo monosubunit [FeFe]-hydrogenases, CAETHG_0110 andCAETHG_3841. These generally use ferredoxin as the electronacceptor/donor, are not electron bifurcating, and show onlyvery low expression (5 to 9 FPKM). There is also a third gene,CAETHG_0119, annotated as an [FeFe]-hydrogenase, but itlacks the sequence segments involved in H-cluster iron bindingand can therefore not encode a functional [FeFe]-hydrogenase.

Genes for an [NiFe]-hydrogenase (CAETHG_0861-62) arealso present and clustered with a gene for a [NiFe]-hydrogen mat-uration protease (CAETHG_0860). No expression for these threegenes was found under any condition investigated (�0.1 FPKM).This has also been reported for transcriptomic data sets for C.ljungdahlii grown on either H2/CO2 or fructose (41). Consistently,not all the genes required for [NiFe]-hydrogenase maturation arepresent. Only the genes hypECDF (CAETHG_0372-0369) werefound, whereas the genes hypAB appeared to be absent. Neverthe-less, the hypECDF genes were expressed at a reasonable level (35FPKM).

None of the genes of the six putative hydrogenases contain atwin arginine (TAT) motif sequence or a transmembrane helixmotif sequence, indicating that the hydrogenases are cytoplasmicenzymes.

Based on the genetic outfit (see above), C. autoethanogenumshould be able to synthesize five different functional hydrogenases.This contrasts with the finding that during growth on H2/CO2, CO-rich BOF gas, or fructose, apparently only the electron-bifurcatingand NADP-dependent [FeFe]-hydrogenase HytABCDE1E2 is active(Table 1). Consistently, the hyt genes could not be disrupted when weemployed the ClosTron technology (47), whereas, with exception ofCAETHG_3841, the other [FeFe]- and [NiFe]-hydrogenase genescould be disrupted without a negative affect on growth in CO-richBOF gas.

Ferredoxin-dependent transhydrogenase (Nfn). Extracts ofH2/CO2-grown cells catalyzed the reduction of ferredoxin withNADPH only in the presence of NAD� and vice versa (Table 1),

indicating the presence of an electron-bifurcating and ferredoxin-dependent transhydrogenase (Nfn) catalyzing reversible reaction4 (54, 55). Unlike the genes in C. kluyveri (55), the respective genesnfnA and nfnB are fused into one gene in C. autoethanogenum(CAETHG_1580) and C. ljungdahlii (CLJU_c37240). The nfngene in CO-grown C. autoethanogenum belongs to the most highlyexpressed genes (926 FPKM average across all genes). Heterolo-gous expression of this gene from C. autoethanogenum in E. coliyielded cells with Nfn activity (unpublished result).

Fdred2� � NADH � 2 NADP� � H�º Fdox � NAD�

� 2 NADPH (4)

Fdred2�:NAD� oxidoreductase (Rnf). Cell extracts of H2/

CO2-grown C. autoethanogenum catalyzed the reduction of NAD�

with reduced ferredoxin (Fdred2� � NAD� � H� º Fdox �

NADH). Reduced ferredoxin was regenerated via the carbon mon-oxide dehydrogenase reaction (CO � H2O � Fdox º CO2 �Fdred

2� � 2 H�). The Rnf activity was associated with the mem-brane fraction (150,000 � g pellet) (data not shown). NADP�

could not substitute for NAD� as the electron acceptor (Table 1).The activity was independent of the presence of sodium ions.

Rnf is generally composed of 6 different subunits (RnfA to -G),of which three are integral membrane proteins. In the genome ofC. autoethanogenum, the genes are clustered in one operon, rnf-CDGEAB (CAETHG_3227-32), 193 bases downstream of thegene rseC (CAETHG_3226). The rnf genes in CO-grown cells be-long to the most highly expressed genes (932 FPKM average acrossall genes). One of the genes encoding the Rnf complex in C. ljung-dahlii has been deleted and this resulted in reduced growth onsugar and no growth on H2/CO2 or CO (56).

Carbon monoxide dehydrogenase. The specific activity of theenzyme in cell extracts of H2/CO2-grown cells was 5.6 U/mg andthus considerably higher than in those of CO-grown cells (Table1). This can be explained by the fact that during growth of C.autoethanogenum on H2/CO2 or fructose, the function of the en-zyme is to catalyze the endergonic reduction of CO2 to CO (E0=��520 mV) with reduced ferredoxin (E0= � �400 mV), a step inthe total synthesis of acetate from 2 CO2, whereas during growth

FIG 2 Expression of the hyt-fdh gene cluster and surrounding genes during growth in continuous culture on CO-rich BOF gas (42% CO, 36% N2, 20% CO2, and2% H2). The growth conditions were the same as those for the CO-rich BOF gas-grown cells analyzed for enzyme activity in Table 1. However, RNA isolation andenzyme activity measurements were performed from different harvests. The graph indicates coverage of reads from the RNA-seq experiments, with FPKM valuesfor the indicated fragments. HytABCDE1E2, electron-bifurcating and NADP-dependent [FeFe]-hydrogenase.

Ethanol Formation from H2 and CO2

September 2015 Volume 197 Number 18 jb.asm.org 2971Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

on CO the function of the enzyme is to catalyze the exergonicoxidation of CO to CO2 with ferredoxin. The Haldane equationpredicts that the catalytic efficiency (kcat/Km) of the enzyme tooxidize CO divided by the catalytic efficiency of the enzyme toreduce CO2 is equal to the equilibrium constant (57, 58).

The C. autoethanogenum genome harbors three genes (one ofwhich is split) that encode Ni-carbon monoxide dehydrogenases(CAETHG_1620-21, CAETHG_3005, and CAETHG_3899). Ofthese, CAETHG_1620-21 encodes the CO dehydrogenase (AcsA)in complex with acetyl-CoA synthase (AcsB; CAETHG_1608)within the Wood-Ljungdahl gene cluster (CAETHG_1606-21).Interestingly, the gene encoding AscA consists of two open read-ing frames in C. autoethanogenum, whereas there is only a singlegene in C. ljungdahlii (CLJU_c37670) or C. ragsdalei. Of the twoCO dehydrogenases outside the Wood-Ljungdahl gene cluster,CAETHG_3005 forms an operon with genes encoding a putative[4Fe-4S]-ferredoxin protein and a reductase. The same arrange-ment is found in the other strains.

Of the three CO dehydrogenases genes, the split gene(CAETHG_1620-21) that encodes the enzyme in complex withthe acetyl-CoA synthase (1,586 FPKM) and the geneCAETHG_3005 for CO dehydrogenase show very high expression(1,129 FPKM average across all genes), while the geneCAETHG_3899 for CO dehydrogenase is hardly expressed (0.9FPKM) during growth on CO. In C. ljungdahlii, the homologue ofCAETHG_3005 (CLJU_c09110) was found to be highly upregu-lated under lithoautotrophic growth conditions (growth on H2/CO2) compared to that under organo-heterotrophic growth con-ditions (growth on fructose) (41), while the opposite was reportedin a study by Tan et al. (59).

In this respect, it is of interest that the genome of Methanosar-cina acetivorans harbors one gene for CO dehydrogenase in a clus-ter with a gene for acetyl-CoA synthase and an orphan gene for COdehydrogenases, either one of which can be disrupted withoutaffecting growth of the methanogen on CO, methanol, or acetate,indicating that each of the two CO dehydrogenases can function invivo in both directions (60, 61).

Formate-hydrogen lyase. The cell extracts of H2/CO2 growncells catalyzed the formation of CO2 and H2 from formate and thereduction of CO2 with H2 to formate (HCOO� � H�º CO2 �H2) at similar specific rates. The specific rates were not affected bythe addition of ferredoxin or NAD(P), indicating that CO2 wasdirectly reduced by H2. The enzyme complex catalyzing the for-mate-hydrogen lyase reaction has been purified from CO-grown cells and characterized (46). It is composed of 7 differentsubunits, namely, FdhA and HytABCDE1E2. The subunitFdhA (CAETHG_2789) is a selenium- and tungsten-contain-ing formate dehydrogenase, and HytABCDE1E2 is the elec-tron-bifurcating and NADP- and ferredoxin-dependent hydro-genase described above, catalyzing reaction 3. The gene fdhA forthe selenium-dependent formate dehydrogenase clusters withgenes for formate dehydrogenase biosynthesis (moeA, mob,and fdhD; CAETHG_2790-93). The cluster is located directlyupstream of the gene cluster hytCBDE1AE2 for the electron-bifurcating [FeFe]-hydrogenase HytABCDE1E2. It has beenshown that the two clusters are transcribed separately (46),which was confirmed by the RNA-seq analysis (Fig. 2).

Formate dehydrogenase. Cell extracts of H2/CO2-grown C.autoethanogenum catalyzed the NADP�-dependent reduction offerredoxin with formate (Table 1). NAD� could not substitute for

NADP�. The reduction of NADP� and of ferredoxin with for-mate proceeded in a 1:1 stoichiometry and was tightly coupled (2HCOO� � NADP� � Fdoxº 2 CO2 � NADPH � Fdred

2� �H�). The electron-bifurcating formate dehydrogenase has beenpurified from CO-grown C. autoethanogenum and found to becomposed of the subunits FdhA and HytABCDE1E2. It is thus thesame enzyme complex (HytA-E/FdhA) that catalyzes the reduc-tion of ferredoxin and NADP� with 2 H2 (reaction 3) and theformate-hydrogen lyase reaction.

In the genome of C. autoethanogenum, there are several genesfor formate dehydrogenase: the aforementioned CAETHG_2789(fdhA) as well as CAETHG_0084 encode two selenium-containingenzymes, and CAETHG_2988 encodes a selenium-free enzyme.Of these, the gene for the formate dehydrogenase that forms acomplex with the Hyt hydrogenase is most highly expressed (580FPKM) (Fig. 2).

None of the formate dehydrogenases appear to have a“periplasmic” location, based on the lack of an N-terminal twinarginine motif. This is in contrast to Moorella thermoacetica, inwhich one of three formate dehydrogenases is predicted to have a“periplasmic” location (52, 54, 62).

Methylene-H4F dehydrogenase (FolD). Cell extracts of H2/CO2-grown C. autoethanogenum catalyzed the reduction ofNADP� with methylene-H4F (reaction 5) at a specific rate of 0.7U/mg. The activity with NAD� was below the detection limit. Thehighest specific activity (5.7 U/mg) was found in CO-grown cells,and an intermediate specific activity (2.4 U/mg) was detereminedfor fructose-grown cells, in agreement with the finding thatgrowth on H2/CO2 was slowest (td � 33 h) and on CO it was fastest(td � 5 h).

NADPH � methenyl-H4F � H�º NADP� � methylene-H4F(5)

In the genome of C. autoethanogenum, only one gene(CAETHG_1616), for a methylene-H4F dehydrogenase, is found.It is predicted to encode a cytoplasmic bifunctional cyclohydro-lase/dehydrogenase without a prosthetic group and is located in acluster of genes that encode many of the other enzymes involvedin the Wood-Ljungdahl pathway, including methylene-H4F re-ductase (CAETHG_1614-15).

Methylene-H4F reductase (MetFV). Cell extracts of H2/CO2-grown cells catalyze the reduction of benzyl viologen with methyl-H4F (reaction 6). The extracts did not catalyze the oxidation ofNADPH or NADH with methylene-H4F in either the absence orthe presence of oxidized ferredoxin (Table 1). It should be notedthat the cell extracts catalyzed a slow methylene-H4F-dependentreduction of ferredoxin in the absence or presence of NAD(P)H.This activity was probably due to the high formaldehyde:ferre-doxin oxidoreductase activity present in the cell extracts, consid-ering that methylene-H4F is synthesized by a spontaneous reac-tion of H4F with formaldehyde and that methylene-H4F alsospontaneously dissociates into H4F and formaldehyde. The form-aldehyde:ferredoxin oxidoreductase activity is a side activity of theacetaldehyde:ferredoxin oxidoreductase (63–65), which wasfound to have high specific activities in the cell extracts (Table 1).

2e� � methylene-H4F � 2 H�ºmethyl-H4F (6)

In the genome of C. autoethanogenum, only one set of genesfor methylene-H4F reductase is found, namely, metFV(CAETHG_1614-15), which is adjacent to the gene folD for meth-

Mock et al.

2972 jb.asm.org September 2015 Volume 197 Number 18Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

ylene-H4F dehydrogenase (CAETHG_1616). Both genes are re-ported to be cotranscribed in C. ljungdahlii (41), and the same wasconfirmed in the RNA-seq analysis for C. autoethanogenum (datanot shown). Heterologous expression of the metF gene from C.autoethanogenum in Escherichia coli revealed that the subunitMetF contains one FMN rather than FAD, the usual cofactor ofmethylene-H4F reductases. Bioinformatic analyses indicated thatMetF lacks an NAD(P) binding site and favors FMN as a cofactor.MetF exhibited high methylene-H4F reductase activity with ben-zyl viologen only when produced together with the iron-sulfurzinc protein MetV, which in part shows sequence similarity toMetF (23, 62). The nature of the electron donor(s) is still uncer-tain (see Discussion).

Acetaldehyde:ferredoxin oxidoreductase. Cell extracts of H2/CO2-grown cells exhibited very high acetaldehyde:ferredoxin ox-idoreductase (AOR) activity (Table 1) that catalyzes the reversiblereaction 7. The high specific activity (8 U/mg) can probably beexplained by the fact that the activity was tested in the direction offerredoxin (E0=� �400 mV) reduction with acetaldehyde (E0=��580 mV) rather than in the physiological direction of acetic acidreduction with reduced ferredoxin (see the expanation given forCO dehydrogenase).

Fdred2� � CH3COO� � 3 H�º Fdox � CH3CHO � H2O

(7)

We tried to also determine in the cell extracts the activity in thephysiological direction by measuring the acetate-dependent oxi-dation of NADH in the presence of ferredoxin and CO (100% inthe gas phase). In this assay, the acetaldehyde:ferredoxin oxi-doreductase reaction is coupled with the ferredoxin-dependentcarbon monoxide dehydrogenase reaction and NAD-dependentethanol dehydrogenase reaction. Unfortunately, the backgroundactivity was too high for an accurate determination, one likelyreason being that the cell extracts contained acetate, which thecells accumulate during growth (see Discussion).

The genome of C. autoethanogenum encodes two putativeAORs (CAETHG_0092 and CAETHG_0102) that share 79%identity on the amino acid level and are both highly expressed(1,558 and 353 FPKM) when the bacterium is grown on CO. Theacetaldehyde:ferredxoxin oxidoreductases are molybdenum- ortungsten iron-sulfur proteins that have been characterized forother acetogenic bacteria and shown to catalyze the reversible re-duction of acetic acid to acetaldehyde (63–65). A role in ethanolformation from acetic acid during growth of C. autoethanogenumon CO has therefore been proposed (46). The high specific activityof acetaldehyde:ferredoxin oxidoreductase in extracts of H2/CO2-grown cells (Table 1) indicates that the enzyme also plays a criticalrole in the formation of ethanol from H2 and CO2. Gene expres-sion profiles obtained with C. ljungdahlii point in the same direc-tion (66).

In addition to the two genes for the two acetaldehyde:ferre-doxin oxidoreductases, the genome of C. autoethanogenum con-tains two genes annotated as aldehyde oxidoreductases. Carefulinspection revealed, however, that the two genes most probablyencode xanthine dehydrogenases rather than acetaledehyde:ferre-doxin oxidoreductases.

Alcohol dehydrogenase and acetaldehyde dehydrogenase.Cell extracts of H2/CO2-grown cells catalyzed the reduction ofacetaldehyde with NADH rather than with NADPH (reaction 8)and the reduction of NADP� with acetaldehyde in a CoA-depen-

dent reaction (reaction 9). Some acetaldehyde dehydrogenase ac-tivity (CoA acetylating) was found, also with NAD� (Table 1).

CH3CHO � NADH � H�º CH3CH2OH � H2O � NAD�

(8)

CH3CO-CoA � NAD(P)H � H�º CH3CHO � NAD(P)�

� CoA (9)

In the genome of C. autoethanogenum, there are two adjacentopen reading frames (CAETHG_3747 and CAETHG_3748) forbifunctional alcohol dehydrogenase/acetaldehyde dehydrogenase(CoA acetylating; AdhE). It is not yet known whether these en-zymes are NAD or NADP dependent. In C. ljungdahlii, the genesfor the putative bifunctional aldehyde/alcohol dehydrogenases,adhE1 (CLJU_c16510) and adhE2 (CLJU_c16520), are arrangedthe same way and have been deleted individually or together. In C.ljungdahlii, deletion of adhE1 but not adhE2 diminished ethanolproduction from fructose with proportional carbon recovery inacetate. The double deletion mutant had a phenotype similar tothat of the adhE1-deficient strain (67). RNA-seq data indicatedthat only the adhE1 (61 FPKM) and not the adhE2 gene (0.4FPKM) was transcribed in C. autoethanogenum, and data from C.ljungdahlii (41) suggest the same, explaining the observed effect.

Additionally, the genome of C. autoethanogenum containsopen reading frames for three putative aldehyde dehydrogenases(CoA acetylating) that catalyze reaction 9 (CAETHG_1819,CAETHG_1830, and CAETHG_3287). All three are clusteredwith genes for microcompartment proteins and genes for ethanol-amine or propanediol utilization. Genes CAETHG_1819 andCAETHG_3287 share 98% sequence identity on the nucleotidelevel. In each of the two microcompartment protein gene clusters,there is also a gene for an alcohol dehydrogenase (CAETHG_1813and CAETHG_3279). The coenzyme specificities have not beendetermined for any of these dehydrogenases.

Besides the two adhE genes and the two genes for alcohol de-hydrogenase in the microcompartment protein gene clusters,there are additional genes annotated as alcohol dehydroge-nases. Of these orphan alcohol dehydrogenases, three havebeen purified and characterized in detail. These are theprimary:secondary alcohol dehydrogenase of C. autoethano-geum encoded by CAETHG_0553 (10), as well as two butanoldehydrogenases of C. ljungdahlii encoded by CLJU_c24880 (ho-mologue of CAETHG_0555) and CLJU_c39950 (homologue ofCAETHG_1841) (68). All three alcohol dehydrogenases havebeen shown to have activity with acetaldehyde/ethanol and toprefer NADPH rather than NADH as a cofactor. It is thus un-likely that they play a major role in ethanol formation duringgrowth on H2/CO2, where mainly NADH-dependent alcohol de-hydrogenase activity was measured in cell extracts (Table 1).

Pyruvate:ferredoxin oxidoreductase. Cell extracts of H2/CO2- or CO-grown C. autoethanogenum catalyzed the CoA-de-pendent reduction of ferredoxin with pyruvate at a specific rate of0.2 U/mg and of cell extracts of fructose-grown cells at a specificrate of 0.6 U/mg (Table 1). During growth on H2/CO2 or on CO,the function of the enzyme was anabolic and it catalyzed the re-duction of acetyl-CoA plus CO2 to pyruvate (E0= � �500 mV)with reduced ferredoxin (E0= � �400 mV) (reaction 10), whichunder standard conditions is an endergonic reaction.

CH3CO-CoA � CO2 � Fdred2� � H�º CH3COCOO�

� Fdox � CoA (10)

Ethanol Formation from H2 and CO2

September 2015 Volume 197 Number 18 jb.asm.org 2973Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

During growth on fructose, the function of the enzyme iscatabolic and it catalyzes the reverse reaction. In the genomeof C. autoethanogenum, there are two copies of the genefor pyruvate:ferredoxin oxidoreductase (CAETHG_0928 andCAETHG_3029).

Glyceraldehyde-phosphate dehydrogenase. Cell extracts cat-

alyzed the readily reversible phosphate-dependent reduction ofNAD� rather than that of NADP� with GAP (reaction 11). Thespecific activities (Table 1) are sufficient to account for the purelyanabolic function of the enzyme when C. autoethanogenum isgrown on H2/CO2 (0.1 U/mg) or CO (0.4 U/mg) and for thepurely catabolic function when the bacterium is grown on fruc-tose (1.6 U/mg) (Table 1).

GAP � Pi � NAD�º 1,3-bis-phosphoglycerate�

� NADH � H� (11)

The genome of C. autoethanogenum contains two open readingframes for GAP dehydrogenase (GAPDH; CAETHG_1760 andCAETHG_3424), one of which probably is NAD specific and theother NADP specific. This is indicated by the finding that in ex-tracts of H2/CO2- and fructose-grown cells, there was someGAPDH activity also with NADP (Table 1).

DISCUSSIONH2 activation. The results indicated that during growth of C. au-toethanogenum on H2 and CO2, probably only one of the six hydro-genases encoded by the genome is active, namely, the electron-bifur-cating, NADP- and ferredoxin-dependent [FeFe]-hydrogenaseHytABCDE1E2 catalyzing reaction 3. This hydrogenase appears toprovide the cells not only with electrons for the reduction of NADPand ferredoxin but also, together with formate dehydrogenase, withelectrons for the reduction of CO2, as shown in Fig. 3 and 4. Thehydrogenase differs in several respects from the electron-bifurcat-ing [FeFe]-hydrogenase HydABCD present in H2/CO2-grown A.woodii: HydABCD is NAD rather than NADP specific, is com-posed of four rather than of six different subunits, and is notinvolved in CO2 reduction to formate (12).

An electron-bifurcating, NAD- and ferredoxin-dependent hy-

FIG 3 Scheme of the metabolism of Clostridium autoethanogenum growing onH2 and CO2 at pH 5.3, assuming that only acetic acid is formed and that themethylene-H4F reductase involved is electron bifurcating and NAD specific.For explanations, see the text. Hyt, electron-bifurcating and NADP-dependent[FeFe]-hydrogenase; Fdh, formate dehydrogenase; Nfn, electron-bifurcatingand ferredoxin-dependent transhydrogenase; Rnf, membrane-associated andenergy-conserving reduced ferredoxin:NAD� oxidoreductase.

FIG 4 Scheme of the metabolism of Clostridium autoethanogenum growing on H2 and CO2 at pH 5.3, assuming that only ethanol is formed and that themethylene-H4F reductase involved is electron bifurcating and NAD specific. (A) Ethanol formation via acetyl-CoA reduction to acetaldehyde. (B) Ethanolformation via acetic acid reduction to acetaldehyde. For explanations, see the text. Hyt, electron-bifurcating and NADP-dependent [FeFe]-hydrogenase; Fdh,formate dehydrogenase; Nfn, electron-bifurcating and ferredoxin-dependent transhydrogenase; Rnf, membrane-associated and energy-conserving reducedferredoxin:NAD� oxidoreductase; ADH, alcohol dehydrogenase; ALD, CoA-dependent acetaldehyde dehydrogenase; AOR, acetaldehyde:ferredoxinoxidoreductase.

Mock et al.

2974 jb.asm.org September 2015 Volume 197 Number 18Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

drogenase activity was observed in neither H2/CO2-grown cellsnor in fructose- nor CO-grown cells (Table 1). However, such ahydrogenase (most likely CAETHG_3569-71) could have a func-tion when the cells are grown at lower H2 concentrations than theH2 concentration used to grow the cells analyzed for enzyme ac-tivity. At an H2 concentration below 1%, the coupled reduction ofNADP and ferredoxin with H2 is thermodynamically no longerpossible (see the introduction), because then the redox potentialsof the 2H�/H2 couple and that of the NADP�/NADPH couple arealmost identical, whereas the redox potential of the NAD�/NADH couple is still more positive by 60 mV.

Methylene-H4F reduction to methyl-H4F. There is a secondimportant difference between C. autoethanogenum and A. woodii.In A. woodii, the reduction of methylene-H4F to methyl-H4F iscatalyzed by an NAD-specific methylene-H4F reductase com-posed of the subunits MetF, MetV, and RnfC2 (23). MetF containsFMN and is the site of methylene-H4 reduction, RnfC2 containsFMN and iron-sulfur clusters and is the site of NADH oxidation,and MetV is an iron-sulfur protein that mediates electron transferfrom RnfC2 to MetF. In C. autoethanogenum, only the genes forMetFV and not those for RnfC2 are present, which is in agreementwith the observation that in cell extracts, methylene-H4F reduc-tase activity is only found with viologen dyes (Table 1).

It has been proposed (9) that in C. ljungdahlii the subunitsMetFV form a complex with the electron transfer flavoproteins(EtfAB) and that the complex of MetFV and EtfAB mediates thereduction of methylene-H4 to methyl-H4F (E0=� �200 mV) (69)with NADH (E0= � �320 mV) and couples this exergonic reac-tion with the endergonic reduction of ferredoxin (E0= � �400mV) with NADH (reaction 12). The proposal is substantiated bythe finding that in anaerobic bacteria the complexes of EtfAB withbutyryl-CoA dehydrogenase (43, 70), with caffeyl-CoA reductase(71) and with lactate dehydrogenase (72) catalyze electron-bifur-cating reactions, in which the exergonic reduction of crotonyl-CoA, caffeyl-CoA and pyruvate, respectively, with NADH arecoupled with the endergonic reduction of ferredoxin with NADH.In these complexes EtfAB is assumed to be the site of electronbifurcation (38, 70).

Methylene-H4F � 2 NADH � Fdox → methyl-H4F

� 2 NAD� � Fdred2� (12)

MetFV alone catalyzes only the reduction of methylene-H4Fwith reduced viologen dyes (62), and EtfAB alone only catalyzesthe reduction of viologen dyes with NADH (70). Both activitiesare found in cell extracts of C. autoethanogenum but a ferredoxin-dependent reduction of methylene-H4F with NADH (reaction 12)or NADPH has not yet been demonstrated, maybe because MetFVand EtfAB only form a loose complex which dissociates in the cellextracts. In the genome of C. autoethanogenum, the genes metFV(CAETHG_3551-52), of which there exists no paralogues, are notcolocated with the genes etfAB, of which there exist five paral-ogues: CAETHG_0115-0116 (clustered with an FMN/FAD-containing dehydrogenase and a reductase); CAETHG_0245-0246 (clustered with an FMN/FAD binding protein);CAETHG_1785-86 (clustered with an acyl-CoA dehydroge-nase, transferase, and ligase); CAETHG_1868-69 (clusteredwith a FixC protein); and CAETHG_3471-72 (clustered withan FMN/FAD binding protein). It could also be that in vivomethylene-H4F reductase forms a complex with both EtfAB andcarbon monoxide dehydrogenase and that only this supercomplex

couples the exergonic reduction of methylene-H4F to methyl-H4Fwith the endergonic reduction of CO2 to CO (Fig. 3 and 4). Afurther possibility is that an iron-sulfur protein other than ferre-doxin or a specific ferredoxin is the low-potential electron accep-tor in reaction 12 and that therefore activity with ferredoxin fromC. pasteurianum was not found.

As explained below, it is also possible that methylene-H4F re-ductase in C. autoethanogenum is not electron bifurcating and iseither NAD or NADP specific, as EtfAB can be involved in electrontransfer from NADH to a more positive electron acceptor withoutelectron bifurcation, as has been reported for the propionyl-CoAdehydrogenase/EtfAB complex from Clostridium propionicum(38). Finally, it has to be considered that in C. autoethanogenum,depending on the growth conditions (low or high H2 concentra-tions) (Table 2), MetFV could form complexes with different dia-phorases (flavoproteins like EtfAB and RnfC2) that catalyze thereduction of dyes with NAD(P)H. Dependent on the diaphorase,the complex could be electron bifurcating or not electron bifur-cating.

Acetic acid reduction to acetaldehyde. There are three otherimportant differences between C. autoethanogenum and A. woodiiwith respect to their oxidoreductase composition. In C. autoetha-nogenum, the methylene-H4F dehydrogenase is NADP specific,whereas this enzyme in A. woodii is NAD specific (12). In C. au-toethanogenum, there is an electron-bifurcating and ferredoxin-dependent transhydrogenase (Nfn) (46) which is not present in A.woodii (27). Most importantly, in C. autoethanogenum there is anactive acetaldehyde:ferredoxin oxidoreductase (AOR) which isnot found in A. woodii (27). A putative formaldehyde:ferredoxinoxidoreductase gene in A. woodii lacks the part for the importantC-terminal domain (Awo_c12420).

In C. autoethanogenum, acetaldehyde:ferredoxin oxidoreduc-tase can catalyze the direct reduction of acetic acid (pK �4.7) toacetaldehyde (63–65) without prior activation of the acetic acid toacetyl-CoA, despite the fact that under standard conditions (1 Mconcentrations of substrates and products except for the protonconcentration of 10�7 M) the reduction of acetate to acetaldehyde(E0= � �580 mV) with reduced ferredoxin (E0= � �400 mV) isendergonic by �35 kJ/mol (31). The bioreduction of carboxylicacids has recently been reviewed (73).

The internal pH of C. autoethanogenum was determined to benear 6 when grown on BOF gas at pH 5.3, based on measurementsof the intracellular and extracellular acetate concentrations (un-published results). An internal pH of near 6 has also been observedin other anaerobes when they are grown at an extracellular pHnear 5 (29, 74–78). At pH 6, the thermodynamics of direct acetatereduction are more favorable by �17.1 kJ/mol, since in reaction 7three protons are consumed. Thus, at pH 6, reaction 7 already isexergonic when the intracellular acetate concentration is 1,000-fold higher than the intracellular acetaldehyde concentration.Such a high ratio is easily reached in the cell when the intracellularpH is 6 and the extracellular is pH 5. Then the intracellular acetateconcentration will be 10 times higher than the extracellular acetateconcentration, since undissociated acetic acid can freely diffusethrough the cytoplasmic membrane but not the anion. For exam-ple, at an extracellular acetate concentration of 0.01 M (0.6 g/li-ter), the intracellular acetate concentration will be 0.1 M andreaction 7 will be already exergonic at an acetaldehyde concentra-tion of 0.1 mM. However, the intracellular acetaldehyde concen-tration is most probably much lower than 0.1 mM due to the

Ethanol Formation from H2 and CO2

September 2015 Volume 197 Number 18 jb.asm.org 2975Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

exergonic reduction of acetaldehyde to ethanol (E0=� �200 mV)with NADH (E0=� �320 mV).

Many bacteria contain a monocarboxylate permease for ace-tate transportation into the cells when the extracellular pH is high,the acetate concentration is low, and thus the acetic acid concen-tration is very low (79–81). This indicates that diffusion of aceticacid into the cells can become rate limiting. In the genome of C.autoethanogenum, a gene for such a transporter is not apparent.

The acetaldehyde:ferredoxin oxidoreductase has been shownto only catalyze the reduction of the undissociated acetic acid(pKa � 4.7) rather than the dissociated acetate (64, 73). At anintracellular pH of 6 and an intracellular acetate concentration of0.1 M, as assumed in the example above, the intracellular concen-tration of acetic acid is 5 mM and thus in the order of the reportedKm of 6 mM acetic acid for the acetaldehyde:ferredoxin oxi-doreductase (64).

With a recombinant strain of Pyrococcus furiosus it was recentlydirectly demonstrated that the acetaldehyde:ferredoxin oxi-doreductase is involved in ethanol formation from acetic acid(82).

Aldehyde:ferredoxin oxidoreductase are tungsten- or molyb-denum iron-sulfur proteins (63–65). It has therefore been consid-ered that ethanol formation from H2 and CO2 in C. autoethano-genum could be steered via the tungstate and/or molybdateconcentration. However, since also the formate dehydrogenasecatalyzing the first step in the Wood-Ljungdahl pathway is a tung-sten- and/or moldybdenum iron-sulfur protein (46), this provednot to be possible.

Calculation of ATP gains and H2 thresholds. For the calcula-tion of ATP gains (mole ATP per mole of acetate or ethanol

formed from H2 and CO2), four reactions have to be considered:(i) the formyl-H4F synthetase reaction in which 1 ATP is con-sumed; (ii) the acetate kinase reaction, in which 1 ATP is formedby substrate level phosphorylation; (iii) the reduced ferredoxin:NAD� oxidoreductase reaction, in which two protons are electro-genically translocated via the Rnf complex; (iv) the ATP synthe-tase (F1Fo) reaction, in which, depending on the number ofsubunits in the c-ring between 3 and 4 vectorial protons are re-quired to drive the synthesis of 1 ATP (83). The c-ring of the ATPsynthase of Clostridium paradoxum has been shown to be com-posed of 11 identical subunits, indicating that in clostridia 3.66(11/3) protons are required to drive the synthesis of 1 ATP (84,85). This stoichiometry was used in all our calculations (Fig. 3 and4; Table 2). It is a little bit higher than that used in the metabolicscheme for the energy metabolism of A. woodii (Fig. 1), whoseATP synthase most probably requires 3.33 (10/3) sodium ions for1 ATP (20), and it is a little bit lower than the stoichiometry of 4(12/3) generally assumed for the ATP synthase of bacteria (12). Allthree stoichiometries are compatible with the conclusions drawnin the following for the energy metabolism of C. autoethanoge-num.

From the ATP gains, H2 thresholds can be calculated (86),considering that to drive the synthesis of 1 ATP under the irrevers-ible conditions of the living cell, between �60 kJ/mol and �80kJ/mol are required (29, 31).

Different scenarios. In the following discussion, we first as-sume that in C. autoethanogenum, methylene-H4F reductase cat-alyzes the electron-bifurcating reaction 12 with NADH as electrondonor. The metabolic scheme in Fig. 3 additionally assumes thatonly acetic acid is formed as the product from H2 and CO2. All the

TABLE 2 ATP gains predicted from metabolic schemes for C. autoethanogenuma

Scenario and fermentationproduct

Methylene-H4Freductase

Acetaldehyde dehydrogenase(CoA acetylating)

Acetaldehyde:ferredoxinoxidoreductase ATP gain

Theoretical H2

threshold (%)

Scenario 1Acetate 2 NADH � Fdox � � 1 �10Ethanol 2 NADH � Fdox � � 0.5 1Ethanol 2 NADH � Fdox � � 1.2 �10

Scenario 2Acetate 2 NADPH � Fdox � � 0.4 �1Ethanol 2 NADPH � Fdox � � �0.04 0Ethanol 2 NADPH � Fdox � � 0.7 �1

Scenario 3Acetate NADH � � 0.4 �1Ethanol NADH � � �0.03 0Ethanol NADH � � 0.7 �1

Scenario 4Acetate NADPH � � 0.14 �0.1Ethanol NADPH � � �0.3 0Ethanol NADPH � � 0.4 �1

a ATP gains (in moles of ATP per mole of product) were predicted from the metabolic schemes for C. autoethanogenum (see Fig. 3 and 4 for illustration of results with Clostridiumautoethanogenum growing on H2 and CO2). In all scenarios, H2 activation is assumed to be catalyzed by electron-bifurcating and NADP-dependent [FeFe]-hydrogenase (HytA-E),and ethanol formation from acetaldehyde is assumed to be catalyzed by a NAD-dependent alcohol dehydrogenase. Either a NADP-dependent CoA-acetylating acetaldehydedehydrogenase or an acetaldehyde:ferredoxin oxidoreductase is involved in acetaldehyde formation. In scenario 1 (see Fig. 3 and 4), the methylene-H4F reductase is assumed to beelectron bifurcating and NAD dependent. In scenario 2, methylene-H4F reductase is assumed to be electron bifurcating and NADP dependent. In scenario 3, methylene-H4Freductase is assumed to be non-electron bifurcating and NAD dependent. In scenario 4, methylene-H4F reductase is assumed to be non-electron bifurcating and NADP dependent.From the ATP gains, the H2 thresholds were calculated. The H2 threshold is the minimal H2 concentration in the gas phase in equilibrium with the liquid phase at which thefermentation is sufficiently exergonic to allow for the predicted ATP gain (see Discussion). �, involved; �, not involved; �, somewhat below threshold value; �, somewhat abovethreshold value.

Mock et al.

2976 jb.asm.org September 2015 Volume 197 Number 18Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

enzymes required are present in H2/CO2-grown cells (Table 1).The scheme predicts that per mole of acetate about 1 ATP is gen-erated (ATP gain), for which, under the conditions in the growingcells, at least �60 kJ per mol acetate are required (see above). AtpH 5.3, an acetate concentration in the growth medium near 0.1M and a CO2 concentration in the gas phase of 23% (growthconditions of the cells described in the Results section), acetateformation from H2/CO2 (reaction 1) is associated with a free en-ergy change of about �60 kJ/mol, when the H2 concentration inthe gas phase in equilibrium with the liquid phase is somewhatbelow 10% (the theoretical H2 threshold) (Table 2). The muchsmaller ATP gain per mol acetate in A. woodii (0.3 mol ATP/molacetate) (Fig. 1) allows this acetogen to grow at much lower H2

concentrations (at about 0.1%), which can explain why A. woodiiis much easier to cultivate on H2/CO2 than C. autoethanogenum,with an ATP gain near 1 mol ATP/mol acetate and an H2 thresholdof �10% (Table 2).

Schemes of the energy metabolism of C. autoethanogenum aregiven in Fig. 4A and B, under the assumption of an electron-bifurcating, NAD- and ferredoxin-dependent methylene-H4F re-ductase and that only ethanol is formed as the product from H2

and CO2. The scheme in Fig. 4A involves ethanol formation via anNADP-specific acetaldehyde dehydrogenase (CoA acetylating)and NAD-specific ethanol dehydrogenase, and the scheme in Fig.4B involves acetaldehyde:ferredoxin oxidoreductase and NAD-specific ethanol dehydrogenase (Table 1). Ethanol formation viaacetyl-CoA reduction to acetaldehyde is associated with the for-mation of 0.5 mol ATP, requiring reaction 2 to be exergonic by atleast �30 kJ/mol. Under the growth conditions described in theResults section, this is the case when the H2 concentration in thegas phase in equilibrium with the liquid phase is somewhere near1% (Table 2). Ethanol formation via acetic acid reduction to ac-etaldehyde involving acetaldehyde:ferredoxin oxidoreductase isassociated with the net formation of 1.2 mol ATP, requiring reac-tion 2 to be exergonic by at least �70 kJ/mol. Under the growthconditions described in the Results section, this is the case whenthe H2 concentration in the gas phase in equilibrium with theliquid phase is higher than 10% (Table 2).

The metabolic schemes in Fig. 3 and 4 predict that per 4 H2

only 0.25 ferredoxin (Fig. 3) and per 6 H2 only 0.25 ferredoxin(Fig. 4A) or 0.75 ferredoxin (Fig. 4B) are reduced in the transhy-drogenase reaction catalyzed by Nfn. This stoichiometry can ex-plain why only relatively low specific activities of this enzyme werefound in H2/CO2-grown cells of C. autoethanogenum (Table 1).

In Table 2, scenarios are also considered under the assumptionthat methylene-H4F reduction is electron bifurcating but NADPrather than NAD specific (Table 2, scenario 2) and that methyl-ene-H4F reduction is not electron bifurcating and is either NADspecific (Table 2, scenario 3) or NADP specific (Table 2, scenario4). Notably, the formation of acetic acid from H2 and CO2 isalways associated with net ATP synthesis regardless of the meth-ylene-H4F reductase involved. This is also the case for ethanolformation from H2 and CO2 when ethanol is formed via aceticacid reduction to acetaldehyde involving acetaldehyde:ferredoxinoxidoreductase. However, the ATP gains are much lower than thescenario where the methylene-H4F reductase is assumed to beelectron bifurcating and NAD specific (Table 2). Lower ATP gainswould allow C. autoethanogenum to grow at lower H2 concentra-tions (H2 thresholds), which could be of advantage, but whichwould only be possible if the electron-bifurcating [FeFe]-hydro-

genase would be NAD� rather than NADP� specific, based on thethermodynamic reasons discussed above. The fact that it is ex-tremely difficult to grow C. autoethanogenum on H2 and CO2 ar-gues against a low H2 threshold, although the difficulties may haveother reasons.

More revealing is, however, the finding that ethanol formationfrom H2 and CO2 via acetyl-CoA reduction to acetaldehyde in-volving CoA-linked acetaldehyde dehydrogenase is only associ-ated with energy conservation when the methylene-H4F reductaseis assumed to be electron bifurcating and NAD specific (Fig. 4A),while in any other case the ATP gain is negative (ATP is net con-sumed rather than formed) (Table 2). Therefore, we conclude thatat least in the case of ethanol formation from H2 and CO2 viaacetyl-CoA reduction to acetaldehyde, the methylene-H4F reduc-tase must be electron bifurcating and NAD specific, because onlyin this case would the ATP gain be positive and able to supportgrowth.

There is a caveat, however, for this scenario: it is possible thatethanol formation from H2 and CO2 in C. autoethanogenum doesnot at all involve acetyl-CoA reduction to acetaldehyde and thatthe CoA-linked acetaldehyde dehydrogenase (AdhE) activityfound in the H2/CO2-grown cells (Table 1) is only there to allowthe cells to reuse the ethanol formed by coupling ethanol oxida-tion to acetate to the reduction of 2 CO2 to acetate when theethanol concentration is high and the H2 concentration is low.Indeed, under the latter conditions, the ethanol concentration incultures of C. autoethanogenum growing on BOF gas can decreaseagain (unpublished results) and, similarly, butanol oxidation tobutyrate has been reported in C. ljungdahlii (9, 68). Growth onethanol and CO2 has also been observed for A. woodii (35). Acounterargument is that ethanol formation from CO-rich BOFgas appears to be favored when the pH in the growth medium ispH 6 rather than pH 5.5 (87, 88), i.e., conditions under whichethanol formation via acetic acid reduction to acetaldehyde is lessfavored (see above). However, these findings were not backed by amore recent study that showed that during growth of C. autoetha-nogenum on CO at pH 4.7, mainly ethanol was formed, whereas atpH 6 ethanol and acetate were generated in almost equal amounts(89).

When C. autoethanogenum is grown on H2 and CO, both H2

and CO are consumed (90), sometimes resulting in a decrease inthe H2 concentration in the gas phase to below the concentrationof 1%. This observation does not necessarily argue for a low H2

threshold during growth on H2 and CO2, since under mixotrophicgrowth conditions the H2 threshold can be lower, which has beenclearly shown for A. woodii growing on H2, CO2, and lactate (26).

It was mentioned above that at H2 concentrations below 1%the coupled reduction of NADP and ferredoxin with H2 is ther-modynamically no longer possible and that therefore at suchlow H2 concentrations an electron-bifurcating, NAD- andferredoxin-dependent [FeFe]-hydrogenase (HydABC) or oneof the nonbifurcating ferredoxin-only-dependent [FeFe]-hy-drogenases should come into action. Such scenarios were notdiscussed because they involve too many unknowns.

Conclusion. In a recent review on the bioenergetics of aceto-gens (12), it was stated that “open questions in the bioenergetics ofC. ljungdahlii (growing on H2 and CO2) concern the electron ac-ceptor of the hydrogenase (either ferredoxin and NAD or ferre-doxin and NADP), the electron donor for the formate dehydro-genase (ferredoxin, hydrogen, or ferredoxin and NADPH), and

Ethanol Formation from H2 and CO2

September 2015 Volume 197 Number 18 jb.asm.org 2977Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

whether or not the methylene-H4F reductase is electron bifurcat-ing and reduces ferredoxin.” Our results indicate that in C. auto-ethanogenum, a close relative of C. ljungdahlii, ferredoxin andNADP are the electron acceptors for H2 oxidation (reaction 3) andthat CO2 is reduced with H2 in a reaction that does not involveferredoxin and/or NAD(P)H. But the third open question,whether or not the methylene-H4F reductase is electron bifurcat-ing and “reduces ferredoxin,” remains to be answered. In Moorellathermoacetica, there is indirect evidence for an electron-bifurcat-ing and NAD-specific methylene-H4F reductase (62).

ACKNOWLEDGMENTS

This work was supported by the Max Planck Society and the Fonds derChemischen Industrie. We thank the following investors in LanzaTech’stechnology: Stephen Tindall, Khosla Ventures, Qiming Venture Partners,Softbank China, the Malaysian Life Sciences Capital Fund, Mitsui, Prim-etals, CICC Growth Capital Fund I, L.P., and the New Zealand Superan-nuation Fund.

REFERENCES1. Abrini J, Naveau H, Nyns EJ. 1994. Clostridium autoethanogenum, sp.

nov., an anaerobic bacterium that produces ethanol from carbon monox-ide. Arch Microbiol 161:345–351. http://dx.doi.org/10.1007/BF00303591.

2. Bengelsdorf FR, Straub M, Dürre P. 2013. Bacterial synthesis gas (syn-gas) fermentation. Environ Technol 34:1639 –1651. http://dx.doi.org/10.1080/09593330.2013.827747.

3. Daniell J, Köpke M, Simpson SD. 2012. Commercial biomass syngas fer-mentation. Energies 5:5372–5417. http://dx.doi.org/10.3390/en5125372.

4. Tanner RS, Miller LM, Yang D. 1993. Clostridium ljungdahlii sp. nov., anacetogenic species in clostridial rRNA homology group I. Int J Syst Bacte-riol 43:232–236. http://dx.doi.org/10.1099/00207713-43-2-232.

5. Saxena J, Tanner RS. 2011. Effect of trace metals on ethanol productionfrom synthesis gas by the ethanologenic acetogen, Clostridium ragsdalei. JInd Microbiol Biotechnol 38:513–521. http://dx.doi.org/10.1007/s10295-010-0794-6.

6. Köpke M, Mihalcea C, Liew F, Tizard JH, Ali MS, Conolly JJ, Al-SinawiB, Simpson SD. 2011. 2,3-Butanediol production by acetogenic bacteria,an alternative route to chemical synthesis, using industrial waste gas.Appl Environ Microbiol 77:5467–5475. http://dx.doi.org/10.1128/AEM.00355-11.

7. Brown SD, Nagaraju S, Utturkar S, De Tissera S, Segovia S, Mitchell W,Land ML, Dassanayake A, Köpke M. 2014. Comparison of single-molecule sequencing and hybrid approaches for finishing the genome ofClostridium autoethanogenum and analysis of CRISPR systems in indus-trial relevant Clostridia. Biotechnol Biofuels 7:40. http://dx.doi.org/10.1186/1754-6834-7-40.

8. Utturkar S, Klingeman DM, Bruno-Barcena JM, Chinn MS, GrundenAM, Köpke M, Brown SD. 2015. Sequence data for Clostridium autoetha-nogenum using three generations of sequencing technologies. Sci Data2:150014. http://dx.doi.org/10.1038/sdata.2015.14.

9. Köpke M, Held C, Hujer S, Liesegang H, Wiezer A, Wollherr A,Ehrenreich A, Liebl W, Gottschalk G, Dürre P. 2010. Clostridium ljung-dahlii represents a microbial production platform based on syngas. ProcNatl Acad Sci U S A 107:13087–13092. http://dx.doi.org/10.1073/pnas.1004716107.

10. Köpke M, Gerth ML, Maddock DJ, Mueller AP, Liew F, Simpson SD,Patrick WM. 2014. Reconstruction of an acetogenic 2,3-butanediol path-way involving a novel NADPH-dependent primary-secondary alcohol de-hydrogenase. Appl Environ Microbiol 80:3394 –3403. http://dx.doi.org/10.1128/AEM.00301-14.

11. Phillips JR, Clausen EC, Gaddy JL. 1994. Synthesis gas as substrate for thebiological production of fuels and chemicals. Appl Biochem Biotechnol45– 46:145–157.

12. Schuchmann K, Müller V. 2014. Autotrophy at the thermodynamic limitof life: a model for energy conservation in acetogenic bacteria. Nat RevMicrobiol 12:809 – 821. http://dx.doi.org/10.1038/nrmicro3365.

13. Sharak Genthner BR, Bryant MP. 1987. Additional characteristics ofone-carbon-compound utilization by Eubacterium limosum and Aceto-bacterium woodii. Appl Environ Microbiol 53:471– 476.

14. Bertsch J, Muller V. 19 June 2015. CO metabolism in the acetogen Ace-tobacterium woodii. Appl Environ Microbiol http://dx.doi.org/10.1128/AEM.01772-15.

15. Gottwald M, Andreesen JR, LeGall J, Ljungdahl LG. 1975. Presence ofcytochrome and menaquinone in Clostridium formicoaceticum and Clos-tridium thermoaceticum. J Bacteriol 122:325–328.

16. Daniel SL, Hsu T, Dean SI, Drake HL. 1990. Characterization of the H2-and CO-dependent chemolithotrophic potentials of the acetogens Clos-tridium thermoaceticum and Acetogenium kivui. J Bacteriol 172:4464 –4471.

17. Frostl JM, Seifritz C, Drake HL. 1996. Effect of nitrate on the autotrophicmetabolism of the acetogens Clostridium thermoautotrophicum and Clos-tridium thermoaceticum. J Bacteriol 178:4597– 4603.

18. Sakai S, Nakashimada Y, Yoshimoto H, Watanabe S, Okada H, NishioN. 2004. Ethanol production from H2 and CO2 by a newly isolated ther-mophilic bacterium, Moorella sp. HUC22-1. Biotechnol Lett 26:1607–1612. http://dx.doi.org/10.1023/B:BILE.0000045661.03366.f2.

19. Hess V, Schuchmann K, Müller V. 2013. The ferredoxin: NAD� oxi-doreductase (Rnf) from the acetogen Acetobacterium woodii requires Na�

and is reversibly coupled to the membrane potential. J Biol Chem 288:31496 –31502. http://dx.doi.org/10.1074/jbc.M113.510255.

20. Matthies D, Zhou WC, Klyszejko AL, Anselmi C, Yildiz O, Brandt K,Müller V, Faraldo-Gomez JD, Meier T. 2014. High-resolution structureand mechanism of an F/V-hybrid rotor ring in a Na�-coupled ATP syn-thase. Nat Commun 5:5286. http://dx.doi.org/10.1038/ncomms6286.

21. Schuchmann K, Müller V. 2012. A bacterial electron-bifurcating hydro-genase. J Biol Chem 287:31165–31171. http://dx.doi.org/10.1074/jbc.M112.395038.

22. Ragsdale SW, Ljungdahl LG. 1984. Purification and properties of NAD-dependent 5,10-methylenetetrahydrolate dehydrogenase from Acetobac-terium woodii. J Biol Chem 259:3499 –3503.

23. Bertsch J, Oppinger C, Hess V, Langer JD, Müller V. 2015. A hetero-trimeric NADH-oxidizing methylenetetrahydrofolate reductase from theacetogenic bacterium Acetobacterium woodii. J Bacteriol 179:1681–1689.http://dx.doi.org/10.1128/JB.00048-15.

24. Schuchmann K, Müller V. 2013. Direct and reversible hydrogenation ofCO2 to formate by a bacterial carbon dioxide reductase. Science 342:1382–1385. http://dx.doi.org/10.1126/science.1244758.

25. Seitz HJ, Schink B, Pfennig N, Conrad R. 1990. Energetics of syntrophicethanol oxidation in defined chemostat cocultures. 1. Energy requirementfor H2 production and H2 oxidation. Arch Microbiol 155:82– 88.

26. Peters V, Janssen PH, Conrad R. 1998. Efficiency of hydrogen utilizationduring unitrophic and mixotrophic growth of Acetobacterium woodii onhydrogen and lactate in the chemostat. FEMS Microbiol Ecol 26:317–324.http://dx.doi.org/10.1111/j.1574-6941.1998.tb00516.x.

27. Poehlein A, Schmidt S, Kaster AK, Goenrich M, Vollmers J, Thürmer A,Bertsch J, Schuchmann K, Voigt B, Hecker M, Daniel R, Thauer RK,Gottschalk G, Müller V. 2012. An ancient pathway combining carbondioxide fixation with the generation and utilization of a sodium ion gra-dient for ATP synthesis. PLoS One 7:e33439. http://dx.doi.org/10.1371/journal.pone.0033439.

28. Seitz HJ, Schink B, Pfennig N, Conrad R. 1990. Energetics of syntrophicethanol oxidation in defined chemostat cocultures. 2. Energy sharing inbiomass production. Arch Microbiol 155:89 –93.

29. Thauer RK, Morris JG. 1984. Metabolism of chemotrophic anaerobes:old views and new aspects, p 123–168. In Kelly PD, Carr NG (ed), Themicrobe 1984, pt II. Cambridge University Press, Cambridge, UnitedKingdom.

30. Decker K, Jungermann K, Thauer RK. 1970. Energy production inanaerobic organisms. Angew Chem Int Ed Engl 9:138 –158. http://dx.doi.org/10.1002/anie.197001381.

31. Thauer RK, Jungermann K, Decker K. 1977. Energy conservation inchemotrophic anaerobic bacteria. Bacteriol Rev 41:100 –180.

32. Stouthamer AH. 1979. The search for correlation between theoretical andexperimental growth yields, p 1– 47. In Quayle JR (ed), Microbial bio-chemistry, vol 21. University Park Press, Baltimore, MD.

33. Nethe-Jaenchen R, Thauer RK. 1984. Growth yields and saturation con-stant of Desulfovibrio vulgaris in chemostat culture. Arch Microbiol 137:236 –240. http://dx.doi.org/10.1007/BF00414550.

34. Balch WE, Schoberth S, Tanner RS, Wolfe RS. 1977. Acetobacterium, anew genus of hydrogen-oxidizing, carbon dioxide-reducing, anaerobicbacteria. Int J Syst Bacteriol 27:355–361. http://dx.doi.org/10.1099/00207713-27-4-355.

Mock et al.

2978 jb.asm.org September 2015 Volume 197 Number 18Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

35. Buschhorn H, Dürre P, Gottschalk G. 1989. Production and utilizationof ethanol by the homoacetogen Acetobacterium woodii. Appl EnvironMicrobiol 55:1835–1840.

36. Buschhorn H, Dürre P, Gottschalk G. 1992. Purification and propertiesof the coenzyme A-linked acetaldehyde dehydrogenase of Acetobacteriumwoodii. Arch Microbiol 158:132–138. http://dx.doi.org/10.1007/BF00245216.

37. Bennett BD, Kimball EH, Gao M, Osterhout R, Van Dien SJ, Rabinow-itz JD. 2009. Absolute metabolite concentrations and implied enzymeactive site occupancy in Escherichia coli. Nat Chem Biol 5:593–599. http://dx.doi.org/10.1038/nchembio.186.

38. Buckel W, Thauer RK. 2013. Energy conservation via electron-bifurcating ferredoxin reduction and proton/Na� translocating ferre-doxin oxidation. Biochim Biophys Acta 1827:94 –113. http://dx.doi.org/10.1016/j.bbabio.2012.07.002.

39. Bar-Even A. 2013. Does acetogenesis really require especially low reduc-tion potential? Biochim Biophys Acta 1827:395– 400. http://dx.doi.org/10.1016/j.bbabio.2012.10.007.

40. de Graef MR, Alexeeva S, Snoep JL, de Mattos MJT. 1999. The steady-state internal redox state (NADH/NAD) reflects the external redox stateand is correlated with catabolic adaptation in Escherichia coli. J Bacteriol181:2351–2357.

41. Nagarajan H, Sahin M, Nogales J, Latif H, Lovley DR, Ebrahim A,Zengler K. 2013. Characterizing acetogenic metabolism using a genome-scale metabolic reconstruction of Clostridium ljungdahlii. Microb CellFact 12:118. http://dx.doi.org/10.1186/1475-2859-12-118.

42. Schönheit P, Wäscher C, Thauer RK. 1978. Rapid procedure for purifi-cation of ferredoxin from clostridia using polyethyleneimine. FEBS Lett89:219 –222. http://dx.doi.org/10.1016/0014-5793(78)80221-X.

43. Li F, Hinderberger J, Seedorf H, Zhang J, Buckel W, Thauer RK. 2008.Coupled ferredoxin and crotonyl coenzyme A (CoA) reduction withNADH catalyzed by the butyryl-CoA dehydrogenase/Etf complex fromClostridium kluyveri. J Bacteriol 190:843– 850. http://dx.doi.org/10.1128/JB.01417-07.

44. Smith ET, Feinberg BA. 1990. Redox properties of several bacterial ferre-doxins using square wave voltammetry. J Biol Chem 265:14371–14376.

45. Tappe W, Tomaschewski C, Rittershaus S, Groeneweg J. 1996. Cultiva-tion of nitrifying bacteria in the retentostat, a simple fermenter with in-ternal biomass retention. FEMS Microbiol Ecol 19:47–52. http://dx.doi.org/10.1111/j.1574-6941.1996.tb00197.x.

46. Wang SN, Huang HY, Kahnt J, Mueller AP, Köpke M, Thauer RK.2013. NADP-specific electron-bifurcating [FeFe]-hydrogenase in a func-tional complex with formate dehydrogenase in Clostridium autoethanoge-num grown on CO. J Bacteriol 195:4373– 4386. http://dx.doi.org/10.1128/JB.00678-13.

47. Heap JT, Kuehne SA, Ehsaan M, Cartman ST, Cooksley CM, Scott JC,Minton NP. 2010. The ClosTron: mutagenesis in Clostridium refined andstreamlined. J Microbiol Methods 80:49 –55. http://dx.doi.org/10.1016/j.mimet.2009.10.018.

48. Purdy D, O’Keeffe TAT, Elmore M, Herbert M, McLeod A, Bokori-Brown M, Ostrowski A, Minton NP. 2002. Conjugative transfer of clos-tridial shuttle vectors from Escherichia coli to Clostridium difficile throughcircumvention of the restriction barrier. Mol Microbiol 46:439 – 452. http://dx.doi.org/10.1046/j.1365-2958.2002.03134.x.

49. Williams DR, Young DI, Young M. 1990. Conjugative plasmid transferfrom Escherichia coli to Clostridium acetobutylicum. J Gen Microbiol 136:819 – 826. http://dx.doi.org/10.1099/00221287-136-5-819.

50. Plourde-Owobi L, Seguin D, Baudin MA, Moste C, Rokbi B. 2005.Molecular characterization of Clostridium tetani strains by pulsed-field gelelectrophoresis and colony PCR. Appl Environ Microbiol 71:5604 –5606.http://dx.doi.org/10.1128/AEM.71.9.5604-5606.2005.

51. Schut GJ, Adams MW. 2009. The iron-hydrogenase of Thermotoga ma-ritima utilizes ferredoxin and NADH synergistically: a new perspective onanaerobic hydrogen production. J Bacteriol 191:4451– 4457. http://dx.doi.org/10.1128/JB.01582-08.

52. Wang S, Huang H, Kahnt J, Thauer RK. 2013. A reversible electron-bifurcating ferredoxin- and NAD-dependent [FeFe]-hydrogenase (Hyd-ABC) in Moorella thermoacetica. J Bacteriol 195:1267–1275. http://dx.doi.org/10.1128/JB.02158-12.

53. Zheng YN, Kahnt J, Kwon IH, Mackie RI, Thauer RK. 2014. Hydrogenformation and its regulation in Ruminococcus albus: involvement of anelectron-bifurcating [FeFe]-hydrogenase, of a non-electron-bifurcating[FeFe]-hydrogenase, and of a putative hydrogen-sensing [FeFe]-

hydrogenase. J Bacteriol 196:3840 –3852. http://dx.doi.org/10.1128/JB.02070-14.

54. Huang H, Wang S, Moll J, Thauer RK. 2012. Electron bifurcationinvolved in the energy metabolism of the acetogenic bacterium Moorellathermoacetica growing on glucose or H2 plus CO2. J Bacteriol 194:3689 –3699. http://dx.doi.org/10.1128/JB.00385-12.

55. Wang S, Huang H, Moll J, Thauer RK. 2010. NADP� reduction withreduced ferredoxin and NADP� reduction with NADH are coupled via anelectron-bifurcating enzyme complex in Clostridium kluyveri. J Bacteriol192:5115–5123. http://dx.doi.org/10.1128/JB.00612-10.

56. Tremblay PL, Zhang T, Dar SA, Leang C, Lovley DR. 2012. The Rnfcomplex of Clostridium ljungdahlii is a proton-translocating ferredoxin:NAD� oxidoreductase essential for autotrophic growth. mBio 4(1):e00406-12. http://dx.doi.org/10.1128/mBio.00406-12.

57. Noor E, Bar-Even A, Flamholz A, Reznik E, Liebermeister W, Milo R.2014. Pathway thermodynamics highlights kinetic obstacles in central me-tabolism. PLoS Comput Biol 10:e1003483. http://dx.doi.org/10.1371/journal.pcbi.1003483.

58. Noor E, Flamholz A, Liebermeister W, Bar-Even A, Milo R. 2013. A noteon the kinetics of enzyme action: a decomposition that highlights thermo-dynamic effects. FEBS Lett 587:2772–2777. http://dx.doi.org/10.1016/j.febslet.2013.07.028.

59. Tan Y, Liu J, Chen X, Zheng H, Li F. 2013. RNA-seq-based comparativetranscriptome analysis of the syngas-utilizing bacterium Clostridiumljungdahlii DSM 13528 grown autotrophically and heterotrophically. MolBiosyst 9:2775–2784. http://dx.doi.org/10.1039/c3mb70232d.

60. Matschiavelli N, Oelgeschlager E, Cocchiararo B, Finke J, Rother M.2012. Function and regulation of isoforms of carbon monoxide dehydro-genase/acetyl coenzyme A synthase in Methanosarcina acetivorans. J Bac-teriol 194:5377–5387. http://dx.doi.org/10.1128/JB.00881-12.

61. Rother M, Oelgeschlager E, Metcalf WW. 2007. Genetic and proteomicanalyses of CO utilization by Methanosarcina acetivorans. Arch Microbiol188:463– 472. http://dx.doi.org/10.1007/s00203-007-0266-1.

62. Mock J, Wang SN, Huang HY, Kahnt J, Thauer RK. 2014. Evidencefor a hexaheteromeric methylenetetrahydrofolate reductase inMoorella thermoacetica. J Bacteriol 196:3303–3314. http://dx.doi.org/10.1128/JB.01839-14.

63. Huber C, Caldeira J, Jongejan JA, Simon H. 1994. Further characteriza-tion of 2 different, reversible aldehyde oxidoreductases from Clostridiumformicoaceticum, one containing tungsten and the other molybdenum.Arch Microbiol 162:303–309. http://dx.doi.org/10.1007/BF00263776.

64. Huber C, Skopan H, Feicht R, White H, Simon H. 1995. Pterincofactor, substrate specificity, and observations on the kinetics of thereversible tungsten-containing aldehyde oxidoreductase from Clostrid-ium thermoaceticum: preparative reductions of a series of carboxylatesto alcohols. Arch Microbiol 164:110 –118. http://dx.doi.org/10.1007/BF02525316.

65. White H, Huber C, Feicht R, Simon H. 1993. On a reversible molybde-num-containing aldehyde oxidoreductase from Clostridium formicoaceti-cum. Arch Microbiol 159:244–249. http://dx.doi.org/10.1007/BF00248479.

66. Xie BT, Liu ZY, Tian L, Li FL, Chen XH. 2015. Physiological response ofClostridium ljungdahlii DSM 13528 of ethanol production under differentfermentation conditions. Bioresour Technol 177:302–307. http://dx.doi.org/10.1016/j.biortech.2014.11.101.

67. Leang C, Ueki T, Nevin KP, Lovley DR. 2013. A genetic system forClostridium ljungdahlii: a chassis for autotrophic production of biocom-modities and a model homoacetogen. Appl Environ Microbiol 79:1102–1109. http://dx.doi.org/10.1128/AEM.02891-12.

68. Tan Y, Liu JJ, Liu Z, Li FL. 2014. Characterization of two novel butanoldehydrogenases involved in butanol degradation in syngas-utilizing bac-terium Clostridium ljungdahlii DSM 13528. J Basic Microbiol 54:996 –1004. http://dx.doi.org/10.1002/jobm.201300046.

69. Wohlfarth G, Diekert G. 1991. Thermodynamics of methylenetetrahy-drofolate reduction to methyltetrahydrofolate and its implications forthe energy metabolism of homoacetogenic bacteria. Arch Microbiol 155:378 –381.

70. Chowdhury NP, Mowafy AM, Demmer JK, Upadhyay V, Koelzer S,Jayamani E, Kahnt J, Hornung M, Demmer U, Ermler U, Buckel W.2014. Studies on the mechanism of electron bifurcation catalyzed by elec-tron transferring flavoprotein (Etf) and butyryl- CoA dehydrogenase(Bcd) of Acidaminococcus fermentans. J Biol Chem 289:5145–5157. http://dx.doi.org/10.1074/jbc.M113.521013.

71. Bertsch J, Parthasarathy A, Buckel W, Müller V. 2013. An electron-

Ethanol Formation from H2 and CO2

September 2015 Volume 197 Number 18 jb.asm.org 2979Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

bifurcating caffeyl-CoA reductase. J Biol Chem 288:11304 –11311. http://dx.doi.org/10.1074/jbc.M112.444919.

72. Weghoff MC, Bertsch J, Müller V. 2015. A novel mode of lactate metab-olism in strictly anaerobic bacteria. Environ Microbiol 17:670 – 677. http://dx.doi.org/10.1111/1462-2920.12493.

73. Napora-Wijata K, Strohmeier GA, Winkler M. 2014. Biocatalytic reduc-tion of carboxylic acids. Biotechnol J 9:822– 843. http://dx.doi.org/10.1002/biot.201400012.

74. Riebeling V, Thauer RK, Jungermann K. 1975. Internal-alkaline pHgradient, sensitive to uncoupler and ATPase inhibitor, in growing Clos-tridium pasteurianum. Eur J Biochem 55:445– 453. http://dx.doi.org/10.1111/j.1432-1033.1975.tb02181.x.

75. Cook GM. 2000. The intracellular pH of the thermophilic bacteriumThermoanaerobacter wiegelii during growth and production of fermentationacids. Extremophiles 4:279–284. http://dx.doi.org/10.1007/s007920070014.

76. Gottwald M, Gottschalk G. 1985. The internal pH of Clostridium aceto-butylicum and its effect on the shift from acid to solvent formation. ArchMicrobiol 143:42– 46. http://dx.doi.org/10.1007/BF00414766.

77. Menzel U, Gottschalk G. 1985. The internal pH of Acetobacterium wier-ingae and Acetobacter aceti during growth and production of acetic acid.Arch Microbiol 143:47–51. http://dx.doi.org/10.1007/BF00414767.

78. Booth IR, Krulwich TA, Padan E, Stock JB, Cook GM, Skulachev V,Bennett GN, Epstein W, Slonczewski JL, Rowbury RJ, Matin A, FosterJW, Poole RK, Konings WN, Schäfer G, Dimroth P. 1999. The regula-tion of intracellular pH in bacteria, p 19 –37. In Chadwick D, Cardew G(ed), Bacterial response to pH, vol 221. Novartis Foundation Symposium,London, England.

79. Hosie AHF, Allaway D, Poole PS. 2002. A monocarboxylate permease ofRhizobium leguminosarum is the first member of a new subfamily of trans-porters. J Bacteriol 184:5436 –5448. http://dx.doi.org/10.1128/JB.184.19.5436-5448.2002.

80. Gimenez R, Nunez MF, Badia J, Aguilar J, Baldoma L. 2003. The geneyjcG, cotranscribed with the gene acs, encodes an acetate permease inEscherichia coli. J Bacteriol 185:6448 – 6455. http://dx.doi.org/10.1128/JB.185.21.6448-6455.2003.

81. Welte C, Kroninger L, Deppenmeier U. 2014. Experimental evidence ofan acetate transporter protein and characterization of acetate activation in

aceticlastic methanogenesis of Methanosarcina mazei. FEMS MicrobiolLett 359:147–153. http://dx.doi.org/10.1111/1574-6968.12550.

82. Basen M, Schut GJ, Nguyen DM, Lipscomb GL, Benn RA, Prybol CJ,Vaccaro BJ, Poole FL, II, Kelly RM, Adams MW. 2014. Single geneinsertion drives bioalcohol production by a thermophilic archaeon. ProcNatl Acad Sci U S A 111:17618 –17623. http://dx.doi.org/10.1073/pnas.1413789111.

83. Allegretti M, Klusch N, Mills DJ, Vonck J, Kühlbrandt W, Davies KM.2015. Horizontal membrane-intrinsic alpha-helices in the stator a-sub-unit of an F-type ATP synthase. Nature 521:237–240. http://dx.doi.org/10.1038/nature14185.

84. Ferguson SA, Keis S, Cook GM. 2006. Biochemical and molecular char-acterization of a Na�-translocating F1Fo-ATPase from the thermoalkali-philic bacterium Clostridium paradoxum. J Bacteriol 188:5045–5054. http://dx.doi.org/10.1128/JB.00128-06.

85. Meier T, Ferguson SA, Cook GM, Dimroth P, Vonck J. 2006. Structuralinvestigations of the membrane-embedded rotor ring of the F-ATPasefrom Clostridium paradoxum. J Bacteriol 188:7759 –7764. http://dx.doi.org/10.1128/JB.00934-06.

86. Thauer RK, Kaster AK, Seedorf H, Buckel W, Hedderich R. 2008.Methanogenic archaea: ecologically relevant differences in energyconservation. Nat Rev Microbiol 6:579 –591. http://dx.doi.org/10.1038/nrmicro1931.

87. Cotter JL, Chinn MS, Grunden AM. 2009. Influence of process param-eters on growth of Clostridium ljungdahlii and Clostridium autoethanoge-num on synthesis gas. Enzyme Microb Technol 44:281–288. http://dx.doi.org/10.1016/j.enzmictec.2008.11.002.

88. Kim YK, Hong S, Kyoo KK. 2011. Effect of pH on growth and ethanolproduction of Clostridium ljundahlii. Appl Chem Eng 22:562–565.

89. Abubackar HN, Veiga MC, Kennes C. 2015. Carbon monoxide fermen-tation to ethanol by Clostridium autoethanogenum in a bioreactor with noaccumulation of acetic acid. Bioresour Technol 186:122–127. http://dx.doi.org/10.1016/j.biortech.2015.02.113.

90. Younesi H, Najafpour G, Mohamed AR. 2005. Ethanol and acetateproduction from synthesis gas via fermentation processes using anaerobicbacterium, Clostridium ljungdahlii. Biochem Eng J 27:110 –119. http://dx.doi.org/10.1016/j.bej.2005.08.015.

Mock et al.

2980 jb.asm.org September 2015 Volume 197 Number 18Journal of Bacteriology

on March 2, 2020 by guest

http://jb.asm.org/

Dow

nloaded from