The energy-converting hydrogenase Ech2 is important for the growth of the thermophilic acetogen Thermoanaerobacter kivui on ferredoxin-dependent substrates

ABSTRACT Thermoanaerobacter kivui is the thermophilic acetogenic bacterium with the highest temperature optimum (66°C) and with high growth rates on hydrogen (H2) plus carbon dioxide (CO2). The bioenergetic model suggests that its redox and energy metabolism depends on energy-converting hydrogenases (Ech). Its genome encodes two Echs, Ech1 and Ech2, as sole coupling sites for energy conservation during growth on H2 + CO2. During growth on other substrates, its redox activity, the (proton-gradient-coupled) oxidation of H2 may be essential to provide reduced ferredoxin (Fd) to the cell. While Ech activity has been demonstrated biochemically, the physiological function of both Ech’s is unclear. Toward that, we deleted the complete gene cluster encoding Ech2. Surprisingly, the ech2 mutant grew as fast as the wild type on sugar substrates and H2 + CO2. Hence, Ech1 may be the essential enzyme for energy conservation, and either Ech1 or another enzyme may substitute for H2-dependent Fd reduction during growth on sugar substrates, putatively the H2-dependent CO2 reductase (HDCR). Growth on pyruvate and CO, substrates that are oxidized by Fd-dependent enzymes, was significantly impaired, but to a different extent. While ∆ech2 grew well on pyruvate after four transfers, ∆ech2 did not adapt to CO. Cell suspensions of ∆ech2 converted pyruvate to acetate, but no acetate was produced from CO. We analyzed the genome of five T. kivui strains adapted to CO. Strikingly, all strains carried mutations in the hycB3 subunit of HDCR. These mutations are obviously essential for the growth on CO but may inhibit its ability to utilize Fd as substrate. IMPORTANCE Acetogens thrive by converting H2+CO2 to acetate. Under environmental conditions, this allows for only very little energy to be conserved (∆G′<–20 kJ mol−1). CO2 serves as a terminal electron acceptor in the ancient Wood-Ljungdahl pathway (WLP). Since the WLP is ATP neutral, energy conservation during growth on H2 + CO2 is dependent on the redox metabolism. Two types of acetogens can be distinguished, Rnf- and Ech-type. The function of both membrane-bound enzyme complexes is twofold—energy conversion and redox balancing. Ech couples the Fd-dependent reduction of protons to H2 to the formation of a proton gradient in the thermophilic bacterium Thermoanaerobacter kivui. This bacterium may be utilized in gas fermentation at high temperatures, due to very high conversion rates and the availability of genetic tools. The physiological function of an Ech hydrogenase in T. kivui was studied to contribute an understanding of its energy and redox metabolism, a prerequisite for future industrial applications.

A cetogens thrive on the conversion of hydrogen (H 2 ) + carbon dioxide (CO 2 ) to acetic acid (1).This presents an energetic problem since acetogenesis from H 2 + CO 2 at low ambient H 2 partial pressures (>1 µM) only yields a free energy of about -20 kJ mol -1 , close to the thermodynamic limit of life (2,3).Acetogens respire CO 2 , using the reductive Acetyl-Coenzyme A (Acetyl-CoA) pathway or Wood-Ljungdahl pathway (WLP) as the terminal electron accepting pathway (1).Physiologically, they must couple this reduction of CO 2 to the conservation of energy in the form of a chemiosmotic ion gradient since reductive acetogenesis from H 2 + CO 2 primarily does not involve substrate-level phosphorylation.In the WLP, initially, one ATP must be invested to bind formate to tetrahydrofolate, which is reclaimed in the acetate kinase reaction during catabolic conversion of the product of the WLP, acetyl-Coenzyme A (4) to acetate.As a consequence, the WLP in conjunction with the formation of acetate is ATP-neu tral, and it has been a long-standing question how autotrophic acetogenesis can be coupled to energy conservation.Beyond acetate kinase, no further reaction allows for substrate-level phosphorylation, and since all enzymes of the pathway are cytoplasmic, none of the reactions can be directly coupled to the build-up of a transmembrane ion gradient (∆µ i ).Moreover, to facilitate acetate formation, eight reducing equivalents are required to reduce two molecules of CO 2 , and the electron donors for the reductases and dehydrogenases in the WLP differ among species (5)(6)(7).During growth on H 2 + CO 2 , electrons need to be transferred from H 2 to ferredoxin (Fd), the donor for the CODH, and putatively, to NADPH and/or NADH, which are involved in reducing CO 2 to formate and methenyl-THF to methyl-THF (3,5,8,9), requiring a complex redox network of different oxidoreductases.The enigma of redox metabolism and energy conservation during growth on H 2 + CO 2 have been solved for Acetobacterium woodii.In this mesophilic acetogen (T OPT 30°C), H 2 is oxidized by two hydrogenases.Interestingly, part of the H 2 is oxidized by a soluble enzyme complex, the hydrogen-dependent carbon dioxide reductase HDCR (10), a soluble enzyme that directly reduces CO 2 to formate in the first step of the methyl branch of the WLP.The remaining six reducing equivalents needed to reduce formate to a methyl group in the methyl branch of the WLP, and CO 2 to [CO] in the carbonyl branch are provided by the electron-bifurcating hydrogenase HydABCD (10).However, there is a redox imbalance since the remaining reactions of the WLP need 2 NADH and 1 Fd red , and HydABCD provides equal amounts of NADH and Fd red .A membrane-bound multisubunit enzyme, the Rnf complex, uses the excess Fd red and reduces NAD + , and the free energy of this exergonic reaction is conserved as sodium ion-dependent membrane potential (11,12).Thus, the complex fulfills both, redox homeostasis and energy conservation (13).
The thermophilic model acetogen Thermoanaerobacter kivui (T OPT 66°C) does not possess a Rnf complex.Instead, its genome has two gene clusters for energy-converting hydrogenases (Ech), ech1 (nine genes) and ech2 (eight genes) as sole membrane-bound putative respiratory enzymes (14).Ech hydrogenases catalyze the reversible oxidation of Fd red with protons as electron acceptors (yielding molecular H 2 ) coupled to the build-up of a transmembrane ion gradient in archaea (15)(16)(17)(18).Ech from T. kivui was demonstrated to carry out this chemiosmotic energy conservation (19), but it was not clear whether Ech1 or Ech2 or both complexes were responsible for the observed activity.The redox and energy metabolism of T. kivui is also distinctly different from that of A. woodii since many key enzymes have different cofactor specificities (3).The electron-bifurcating hydrogenase HydABC and methylene-THF dehydrogenase are NADPH-dependent (3).Moreover, the organism possesses a transhydrogenase Nfn (NADPH:Fd NADH oxidor eductase).According to the current model, either Ech may therefore occur in two forms, membrane-bound or in a membrane complex with the methylene-THF reductase (MetFV), and depending on the substrate, the complexes are proposed to operate in the direction of Fd oxidation or reduction (3).Since the genetic methods for T. kivui (20) allow the overproduction of tagged proteins (7), Ech2 was recently purified and biochemically characterized (21).The enzyme was not found in a super-complex with MetFV and was shown to be dependent on ferredoxin, thereby translocating protons into liposomes (21).The function of Ech2 in redox and energy metabolism, however, has not been elucidated.
Here, we describe the deletion of the complete operon encoding Ech2 in T. kivui.The phenotype of the deletion mutant was studied in growth experiments and in resting cells to elucidate the metabolic function of the enzyme complex.Surprisingly, Ech2 was not essential during growth with sugar substrates and on H 2 + CO 2 .The results are discussed in the context of the current bioenergetic model of T. kivui.

Generation of an Ech2 deletion mutant
To resolve the metabolic function of Ech2 in energy and redox metabolism of T. kivui, a markerless deletion of the entire 6.77 kb gene cluster encoding the membranebound hydrogenase Ech2 was performed using methods for genome manipulations, as described by Basen et al. (20).The uracil-auxotrophic strain T. kivui TKV002 (TKV_MB002; ∆pyrE) (20) was transformed with the plasmid pE2TK02, which contained pyrE and upstream (0.9 kb) and downstream (1.0 kb) flanking regions (UFR and DFR, respectively) of the ech2 gene cluster (14).Two consecutive steps were performed.First, the plasmid was integrated into the genome, allowing the transformants to grow on minimal media since the plasmid contained pyrE (Fig. 1A).In the second step, selection against pyrE was carried out using 5-fluoro orotic acid (5-FOA, selection 2).Of the ten colonies grown in the presence of 5-FOA, seven were revertants that had lost the pyrE gene, but not the ech2 cluster, the other three lost the entire ech2 cluster, as intended (Fig. 1B).Sequencing and qPCR confirmed the genotype of the resulting deletion mutant strain (T.kivui ∆ech2D-ech2F; TKV_c19750-TKV_c19680; ∆pyrE).The strain was designated T. kivui TKV_MB050 and will be called ∆ech2 mutant herein.

The ∆ech2 mutant grows on sugar substrates and H 2 + CO 2
Ech hydrogenases are proposed to have different roles during growth with sugar substrates and H 2 + CO 2 in T. kivui.As membrane-bound enzyme complexes, they couple two fundamental metabolic functions, the oxidation/reduction of redox carriers (Fd ox /Fd red and H + /H 2 ) which is either driven by or generating a transmembrane proton gradient (13).In T. kivui, sugars are presumably taken up by PTS systems (22), and the sugar phosphates are then oxidized in glycolysis (yielding NADH) and by pyruvate:ferre doxin oxidoreductase (yielding Fd red ) to acetate (Fig. 2).The reduced electron carriers NADH and (part of the) Fd red are oxidized by a transhydrogenase NfnAB, producing NADPH, part of which is oxidized by the methylene THF dehydrogenase, and another part is oxidized concomitantly with Fd red by the electron-bifurcating hydrogenase HydABC (3).This leaves internally produced H 2 , part of which is used to activate CO 2 to formate in the HDCR reaction.T. kivui is the only acetogen to date besides A. woodii (23) with a characterized HDCR for direct reduction of CO 2 to formate using electrons from H 2 .The HDCR is extremely active, abundant in the cells (24), and essential for autotrophic growth (25).Recently, HDCR of T. kivui has been demonstrated to form bundles of nanowire-like structures attached to one cell pole (24).To balance the redox metabolism on H 2 + CO 2 and sugars, H 2 must be oxidized and Fd ox reduced at the expense of a transmembrane proton gradient, as demonstrated for the purified Ech2 complex (21).
Since it is unclear whether Ech1 can oxidize Fd red (unpublished observation, V. Müller), we anticipated Ech2 to be essential for Fd ox reduction during growth on sugars as described above, and to be the coupling site for energy conversion during growth on H 2 + CO 2 (Fig. 3A).We performed gene expression analysis by RT-qPCR and found that ech2D as representative gene of the ech2 operon was expressed comparable to the housekeeping gene gyrase (TKV_c00100, gyrB), in cells grown with glucose (1.4-fold) and in cells grown with H 2 +CO 2 (1.9-fold) (Table 1).ech1A was present at a higher level, however, was only slightly upregulated in the presence of H 2 + CO 2 (in wild type 7.6-fold vs. gyrB, in ∆ech2 6.8-fold vs. gyrB) compared to glucose (in wild type 4.6-fold vs. gyrB, in ∆ech2 3.6-fold vs. gyrB).This result is in contrast to the previously reported 6-fold and 16-fold higher expression of ech1 and ech2, respectively, with the growth of the wild type on H 2 + CO 2 , which may be growth phase dependent (19).As expected, in the mutant ech2D was not detected.Growth experiments using different sugars as substrates were performed with the ∆ech2 mutant and wild type.Interestingly, only slight differences in growth of the ∆ech2 mutant, compared to the wild type, were observed on glucose, fructose, and the sugar alcohol mannitol.The doubling times of wild type and the ∆ech2 mutant during growth on complex medium with glucose (1.29 h ± 0.12 vs. 1.27 h ± 0.08) (Fig. 4A), fructose (1.49 h ± 0.03 vs. 1.21 h ± 0.03), and mannitol (1.75 h ± 0.14 vs. 1.47 h ± 0.13) (Fig. 5) were similar, with the ∆ech2 mutant being significantly faster on the latter two substrates (Fig. 6).Consistent with this result, previous studies demonstrated doubling times of 1.24-2.4h on glucose (24, 25), 1.3 h on fructose (20) and 2 h on mannitol (22) for wild type cells grown on complex medium.Growing cells of wild type and ∆ech2 mutant metabolized glucose in a homoacetogenic manner, with 2.43 ± 0.23 mol and 2.43 ± 0.28 mol acetate produced from one mol glucose, respectively (Fig. 4B).This corresponds to the previously reported conversion of one mol glucose to 2.5 ± 0.24 mol (26) and 2.6 ± 0.1 mol acetate by T. kivui (27).Other acetogens, such as Moorella thermoacetica (28) and Acetobacterium woodii (29) produce between 2.5 and 2.7 mol acetate per glucose and fructose.In addition to growth experiments on complex medium, growth was also investigated on a defined medium containing uracil.Wild type and ∆ech2 mutant  grown in defined medium with uracil had comparable doubling times on glucose as well (1.57h ± 0.01, wt; 1.60 h ± 0.05, ∆ech2; Fig. 7A) and mannitol (2.01 h ± 0.17, wt; 1.77 h ± 0.20, ∆ech2; Fig. 7B).This shows that growth of the ∆ech2 mutant did not depend on additional substrates or growth factors supplied by the yeast extract in the complex medium.The acetogens Clostridium ljungdahlii and A. woodii both have an Rnf complex instead of an Ech complex, and deletion of the rnf gene cluster had different phenotypes for both species.The ∆rnf mutant of C. ljungdahlii showed growth inhibition and reduced ATP synthesis on fructose (30).While the ∆rnf mutant of A. woodii displayed no inhibition of growth on fructose, a change in substrate to acetate ratio was observed.Instead of producing 2.55 mol acetate from one fructose in the wild type, only 2.06 mol acetate was produced, indicating an inhibition of electron flow toward the WLP (31).The ∆ech2 mutant of T. kivui, presented here, did not show any growth inhibition or change in substrate to acetate ratio, as compared to the wild type.Therefore, we conclude that T. kivui does not rely on Ech2 during growth on sugars, although the metabolic model suggests that either Ech1 or Ech2 or both are essential for H 2 oxidation in redox metabolism on sugar substrates (3), as described above.For consumption of H 2 + CO 2 , one or both Ech complexes are assumed to be necessary (Fig. 3A), not only for redox homeostasis but also as the only site for chemios motic energy conservation coupled to oxidation of excess Fd red (3,14,19).Much to our surprise, the ∆ech2 mutant grew similar to the wild type on H 2 + CO 2 (66/33, vol/vol, 3 atm), with doubling times of 1.24 h ± 0.09 (wild type) vs. 1.30h ± 0.04 (∆ech2 mutant) in complex media (Fig. 8A).This growth was faster than the previously reported doubling times of 1.75-2.5 h (26).Interestingly, during growth on H 2 + CO 2 , acetate is built up at different rates, with the ∆ech2 mutant accumulating acetate almost twice as fast as the wild type in the exponential phase (10.4 mM h −1 ± 1.0, wt; vs.17.3 mM h −1 ± 2.7, ∆ech2; Fig. 8B).Resting cells of the wild type and the ∆ech2 mutant, however, produced acetate from H 2 + CO 2 at a similar rate (Fig. 8C), indicating a higher turnover rate of H 2 + CO 2 only during growth.This observation suggests that to compensate for the same growth rate or biomass buildup, the ∆ech2 mutant consumes more H 2 than the wild type based on the production of acetate during growth.Higher substrate to product conversion rates and lower yields (Y substrate ) have been observed for Escherichia coli and Pseudomo nas taiwanensis if respiratory chain complexes such as the cytochrome b oxidases or NADH dehydrogenases are uncoupled (32,33).
However, this is not easy to explain mechanistically with the current bioenergetic model of T. kivui (3), which requires one of the Ech complexes to oxidize Fd red and provide H 2 to the cell (14).If MetFV is Fd-dependent, coupled to Ech1 and building up a H + -gradient, the non-complexed Ech (putatively Ech2) oxidizes H 2 to provide Fd red to the WLP, at the expense of part of the H + -gradient (3) (Fig. 3A).Since we demonstrated that Ech2 is not essential for growth on H 2 + CO 2 , one option is, that Ech1 is responsible for both Fd ox reduction and energy conservation, however, an Fd-dependent activity of Ech1 has not yet been demonstrated.The second option is depicted in Fig. 3A where H 2 may be oxidized by HDCR to provide Fd red in addition to formate (34).This, however, is an endergonic reaction.Since HDCR has been found membrane-associated (24), the reaction may be driven by a membrane gradient, although a necessary membranespanning motif has not been demonstrated as part of the complex.If HDCR runs in reverse and produces Fd red , this would benefit energy conservation which would be inconsistent with the ∆ech2 mutant producing less biomass per acetate.

Growth on Fd-dependent substrates is impaired
In addition to sugar and H 2 + CO 2 , T. kivui can grow on pyruvate (26), and it has been adapted to grow on 100% CO (35).In contrast to all other substrates tested, pyruvate and carbon monoxide (CO) are substrates whose oxidation solely provides Fd red .While pyruvate is oxidized by pyruvate:Fd oxidoreductase (POR) in T. kivui (7) there are two CO dehydrogenases in T. kivui that catalyze a Fd-dependent CO oxidation, CODH/ACS, and the monofunctional CooS (36).The latter has been demonstrated to be responsible for CO oxidation during growth on the gas (36), whereas the CODH/ACS as the key enzyme in the WLP, mainly reduces CO 2 to CO (4).Fd red from CO or pyruvate oxidation would have to be re-oxidized by either of the Ech complexes depending on the assumed pathway model (7,14).Under the assumption that Ech1 cannot oxidize Fd red , deletion of Ech2 should lead to a redox imbalance and should thus inhibit growth on CO.
In accordance with this logic, we recently demonstrated that resting cells of the ∆ech2 mutant are unable to produce acetate from CO (37).To study the effect of the ech2 deletion on growth with CO, we tried to adapt the ∆ech2 mutant to CO.For the adaption of T. kivui for growth on CO as the only substrate, several weeks are required (35).Therefore, wild type and ∆ech2 mutant were first cultured on H 2 /CO 2 in addition to CO (3 atm; 44/22/33; vol/vol/vol) (Fig. 9A).Subsequently, the cells were transferred to CO/N 2 atmosphere (3 atm; 14/86; vol/vol) (Fig. 9B), where they were cultured for several days.The wild type grew to an OD 600 of 0.057 after 14 days, whereas the ∆ech2 mutant remained at 0.023.To test the adaption and viability of the cells, wild type and ∆ech2 mutant were passaged after 23 days.7 days after the transfer, the wild type had grown to an OD 600 of 0.12 while the ∆ech2 mutant stopped growing at an OD 600 of around Acetate was determined by gas chromatography (n = 4).0.04 (Fig. 7B).This low increase in OD 600 of the ∆ech2 mutant is most likely due to the yeast extract in the medium, which can be utilized by T. kivui (25).From these findings, we concluded that T. kivui depends on Ech2 for catabolic conversion and growth on CO.
To test growth on the second Fd-dependent substrate, pyruvate, cultures of T. kivui wild type and ∆ech2 mutant were grown on glucose and subsequently transferred to complex medium with pyruvate as the only substrate.The wild type grew immediately after inoculation, whereas the ∆ech2 mutant did not grow for 48 to 72 h after inoculation, but ultimately reached an OD 600 of 0.464 ± 0.045 after 96 h.As a control, we inoculated the ∆ech2 mutant on a medium with glucose at the same time and with this substrate, the strain grew immediately (Fig. 10A).After four passages on pyruvate, however, the ∆ech2 mutant showed a similar doubling time as the wild type (wt, 1.61 h ± 0.03; 1.53 h ± 0.05, ∆ech2 mutant) (Fig. 10B).Interestingly, resting cells of the wild type and the ∆ech2 mutant that were not previously adapted to pyruvate, were able to utilize pyruvate at a similar rate (−17.23 mM/h, wt vs. −17.72mM/h, ∆ech2) (Fig. 11A), but with different acetate (+12.45 mM/h, wt; vs. +7.68mM/h, ∆ech2 mutant) and formate (+3.43 mM/h, wt; +1.84 mM/h ∆ech2 mutant) production rates.After 24 h, the wild type and ∆ech2 mutant had consumed 80.16 ± 2.90 mM and 87.11 ± 3.24 mM pyruvate, respectively, and produced 78.09 ± 3.30 mM (wt) and 71.05 ± 2.01 mM (∆ech2) acetate, respectively.In addition to acetate, the wild type produced 14.73 ± 0.86 mM formate, while the ∆ech2 mutant produced 19.50 ± 1.80 mM formate after 24 h (Fig. 11B).Of interest here is that for the production of 1 mM acetate the wild type required 1.02 mM pyruvate, whereas the ∆ech2 mutant required 1.23 mM of pyruvate, indicating a reduced electron flux toward the WLP.The carbon and electron balance during pyruvate conversion was not closed.This change in acetate yield from pyruvate was likely due to an unidentified product.This was previously observed by Leigh et al. (26), who reported that 0.93 mM pyruvate was necessary to produce 1 mM acetate in the wild type.These data demonstrate that T. kivui depends on Ech2 for the oxidation of Fd red , to a different extent for different substrates.Neither Ech1 nor any other enzyme may substitute for Ech2 for CO utilization.Meanwhile, during growth on pyruvate, substrate conversion is slowed down and growth is initially impaired, but-putatively through a mutation in the genome-the ∆ech2 mutant may be adapted to grow on pyruvate.During growth on H 2 + CO 2 or sugars, no growth inhibition, not even initially, was observed.This aggregate of observations implies that there must be another enzyme involved in Fd red oxidation coupled to H 2 production.Importantly, this activity has not been demonstrated for Ech1 (Volker Müller, pers.communication), and therefore, we propose that the alternative activity of HDCR may substitute for Ech2 (Fig. 12).The HDCR of T. kivui is described to oxidize H 2 , or Fd red provided by CODH during CO consumption to fix CO 2 ; thus, the HCDR could oxidize Fd red to reduce H + (10,34).It is unclear, why this affects pyruvate metabolism more than sugar and H 2 + CO 2 metabolism since the predicted direction of Ech2 is Fd ox reduction with the latter two substrates (3), which is thermodynamically more difficult.

Adaptation to CO involves a conserved SNP in HDCR
It is also unclear, why CO metabolism is completely inhibited, while growth on pyruvate is only initially impaired.Putatively, CODH may form an essential complex with Ech2 (23), or HDCR, that is CO sensitive, may carry a mutation toward a lower CO sensitiv ity, simultaneously affecting its interaction with Fd.To investigate this hypothesis, we performed a SNP analysis on the T. kivui strain originally adapted to CO (35), as well as on a strain adapted to CO in our hands.A ∆pyrE CO-adapted strain with a large horizontal gene transfer has already been described (38).In our laboratory, the wild type was re-adapted to CO, as described originally in Weghoff (35), and four single colonies from the adapted culture were picked and sent for SNP sequencing.Sequencing results from an additional strain, adapted to low-temperature growth rather than to CO (39; data set available in the Sequence Read Archive (SRA) under the accession number SRR25301688), were included to identify mutations already present in the laboratory's wild-type strain, or resulting from adaptations to the media/culture conditions used in the laboratory, rather than specifically to the presence of CO.Reads from the original sequencing run (35) were re-analyzed using the same BreSeq analysis pipeline as the other samples to rule out spurious alignment artifacts.
The ∆pyrE-CO strain had 313 identified SNPs, but most (265) were in the area of the previously characterized HGT and were removed for this analysis.All other strains had around 60 identified SNPs.After filtering out all SNPs present in the re-analyzed wild type or in the low temperature adapted strain (un-related to CO-adaptation), and after removing all silent and all intergenic mutations more than 100 bp upstream of a gene, a total of 69 unique SNPs remained that could be associated with the CO-adapted phenotype.A subset of the most interesting SNPs is shown in Table 2. Evidently, a high concentration of mutations was present in acsA and cooC, two genes important for the function of the CODH/ACS complex.Two distinct mutations result in almost complete disruption of the CooC protein, while the two other mutations lead to single amino acid substitutions near the N-terminus.All three mutations present in the acsA gene lead to early frame shifts disrupting the resulting protein, but each mutation is only present in a small subset of the population (between 5% and 20% of reads).A previous study in strains of the acetogen Eubacterium limosum following adaptation to CO found almost identical disruption mutations in cooC2, which the authors believe are possible because cooC is non-essential when nickel is plentiful, as is the case in most synthetic laboratory media (40).However, the complete disruption of acsA observed in our data is surprising since the CODH/ACS complex is essential for acetogenesis.Adapted strains of E. limosum also exhibited mutations in acsA (40), and both acsA and acsB (41), but in all cases these were single amino acid substitutions.That three acsA disruption mutations occurred independently in our adapted strains suggests they do confer some advantage during growth on CO but the low prevalence in the population (low percentage of reads) could be indicative of a balance between whatever benefit the mutations provide and the detrimental effect of disrupting such an important gene.
Given the known toxicity of CO to the metal centers of hydrogenases, it was unsurprising to see mutations in genes for the electron-bifurcating hydrogenase (hydA1), subunits of the energy-converting hydrogenases Ech1 (ech1C) and Ech2 (ech2E and promoter of ech2D), as well as subunits of the hydrogen-dependent carbon dioxide reductase (fdhD and hycB3), and the transhydrogenase (nfnB) (Table 2).As with the cooC gene, there are several frame shifts, but most disrupt only the C-terminus of the relevant gene.For example, two independent mutations remove roughly the last 10% of amino acids from HycB3, and one mutation disrupts only 2% (the last eight amino acids) of NfnB.The largest disruption among the hydrogenases is in HydA1, where two-thirds of the protein remains intact.
The mutations leading to disruption of the C-terminus of HycB3 are of particular interest.The mutations identified here introduce frame shifts at codons 164 and 168 of HycB3, while previous research found that terminating the protein at codon 160 disrupts the oligomeric structure of the HDCR complex, and results in a single hetero-hexametric subunit that is still functional but has reduced activity (to 33% of the wild-type's activity for formate production from H 2 + CO 2 , and 18% for H 2 production from formate) (24,34).HycB3 contains four 4Fe-4S clusters for transferring electrons within the HDCR complex, but the C-terminal truncations described here come after the iron-sulfur coordinating cysteine residues.
As mentioned previously, we hypothesize that a side reaction of the HDCR allows for partial compensation for the loss of Fd red re-oxidation normally carried out by Ech2 during growth on pyruvate.It seems likely that the disruption of the HDCR oligomeric complex caused by the HycB3 mutation in the CO-adapted strains also disrupts this side reaction.If this mutation is essential for CO-adaptation (and it is worth noting that all five sequenced CO-adapted strains had a mutation in hycB3), then growth on CO and deletion of Ech2 are mutually exclusive: the Ech2 mutant requires the oligomeric HDCR side reaction to re-oxidize Fd, while the CO-adapted strains require its monomeric form (possibly because of a structural shift that protects from CO-toxicity).Since the hycB3

Growth conditions
Thermoanaerobacter kivui strains were routinely grown anaerobically at 65°C in a modified DSMZ171 medium.Defined medium contained 50 mM Na For transformation, T. kivui was cultured as described previously (20).Defined medium agar with 1.5% Bacto agar BD (Difco, BD Life Sciences, Heidelberg, Germany) was prepared, autoclaved, and cooled down below 65°C to add substrates.If required, uracil or 5-fluoroorotic acid (5-FOA) was supplemented to minimal medium agar, and the agar is then poured into 20 mL petri dishes containing a cell suspension.The petri dishes holding liquid agar were transferred into an anoxic box (Coy Laboratory Products, Grass Lake, MI) with a former gas H 2 /N 2 [5%/95%; (vol/vol)] atmosphere to solidify.The petri dishes were put into an incubator jar containing a palladium catalyst and calcium chloride as water absorbent.The jar was sealed inside the anoxic box, and the gas atmosphere was adjusted to an overpressure of 5 × 10 4 Pa with Protadur C20 [80%/20% (vol/vol) N 2 /CO 2 ].The jar was transferred for incubation at 65°C.After the incubation period, the jar was taken out and allowed to cool down to room temperature before opening.

Genetic manipulation of Thermoanaerobacter kivui
For generation of the ech2 mutant strain, an integrating plasmid pEch2Tk02 was constructed according to Basen et al. (20).The plasmid is derived from pMBTKv0022, which enables markerless deletions of genes via two independent homologous recombination events.0.5 kbp upstream (UFR) and downstream (DFR) flanking regions of the ech2 gene cluster (TKV_c19680-TKV_c19750) were amplified, using the primer pairs 2E02a/2E02b and 2E02c/2E02d (Fig. 1).Both fragments were fused via PCR using the primers 2E02a/2E02d.The resulting UFR/DFR fragment and pMBTKv0022 were diges ted with BamHI and XbaI.Plasmid and fragment were ligated via T4 ligase, and the resulting plasmid pE2TK02 was transformed into the ∆pyrE strainTKV_MB002 (20,25).Two selection rounds were performed to generate ∆ech2 according to 20 (20).First, the transformants were plated on a minimal medium agar (1.5%) without uracil to enforce the integration of the plasmid encoding pyrE and the UFR and DFR.Grown colonies were transferred to minimal medium.To enforce the loss of the plasmid encoding pyrE (including ech2), 5 mM 5-FOA and 40 mM uracil were supplemented to minimal medium agar.Grown uracil-auxotrophic colonies were transferred into a minimal medium containing 50 mM uracil and screened by PCR for the loss of ech2 using primers O2a/O2d afterward.The loss of the ech2 operon was verified by qPCR.

Resting cell experiment
Resting cells of T. kivui were prepared as described previously (37) under strict anoxic conditions in an anoxic glove box (Coy Laboratory Products, Grass Lake, MI) containing a reducing gas atmosphere [H 2 /N 2 ; 5%/95%; (vol/vol)] atmosphere.Cells were cultivated in 250-500 mL complex medium in 1 L bottles (SCHOTT AG, Mainz, Germany), and harvested in the late exponential growth phase by centrifugation at 4,000 rpm, 4°C for 15 min.The cells were then first washed in imidazole buffer (50 mM imidazole, 20 mM MgSO4, 20 mM KCl, 4 µM resazurin, 2 mM DTE, pH 7.0), subsequently resuspended in imidazole buffer to a final OD 600 of 2 and finally transferred to 50 mL serum bottles, which were then sealed with butyl rubber stoppers.The gas H 2 /N 2 atmosphere [5%/95%; (vol/vol)] in the serum bottles was replaced by flushing with N 2 gas to remove the H 2 .Finally, the headspace was changed to CO/N 2 [2 × 10 5 Pa, 20/80 (vol/vol)] in the experiment toward CO conversion, and to CO 2 /N 2 [1.1 × 10 5 Pa, 20/80 (vol/vol)] in the experiment toward pyruvate conversion.

Metabolite analysis
The concentrations of glucose and pyruvate, formate, and acetate were determined by high-performance liquid chromatography.For the sample preparation, cells were centrifugated at 13,000 rpm for 10 min at 4°C and 250 µL of supernatant was filled into a 1.5 mL polypropylene tube and 5 µL H 2 SO 4 (50%) was added to the supernatant.The sample was mixed and centrifuged at 13,000 rpm for 10 min at 4°C.Afterward, 200 µL of the sample was transferred into an HPLC vial.Separation of sugars, alcohols, and organic acids was performed using a 300 × 8 mm column packed with organic acid resin (CS-Chromatography Service GmbH, Langerwehe, Germany) on a Shimadzu (Kyoto, Japan) LC20 system, equipped with a SIL-20AC autosampler (10 µL sample injection), a LC-20AD pump a CTO-20AC column oven (30°C), RID-10A refractive index detector and a UV-Vis detector, at a flow rate of 0.6 mL/min (5 mM H 2 SO 4 as eluent).
Acetate concentrations were determined by gas chromatography.For the sample preparation, cells were centrifuged at 13,000 rpm for 10 min at 4°C and 450 µL of supernatant was filled into a 1.5 mL polypropylene tube and 50 µL 2 M phosphor acid was added to the supernatant.The sample was mixed and centrifuged at 13,000 rpm for 10 min at 4°C.Next, 450 µL of the supernatant was mixed with 50 µL of 2 M phosphoric acid and 550 µL of H20 dd and transferred to a GC vial.Gas chromatography was performed using an 8860 GC System (Agilent, Santa Clara, United States), equipped with a 6 Ft 1/8 2 mm HayeSep P 60/80 UM column and a flame ionization detector (FID).The FID operates at 250°C, with an airflow (synthetic air) of 300 mL/min, an H2 flow of 30 mL/min, and a make-up flow (N 2 ) of 25 mL/min.As carrier gas served pure N 2 (Westfalen AG, Münster, Germany) at a flow rate of 10 mL/min.0.5 µL of sample was injected at 195°C using a PPZ inlet with a flow rate of 10 mL/min.A temperature gradient was applied by starting from 120°C for 3 min followed by an increase of 5 °C/min to 145°C for 3 min, followed by an increase of 10 °C/min to 170°C for 0.5 min.In the last step, the temperature was increased by 30 °C/min to 200°C.Data analysis was performed with the OpenLAB CDS EZChrom software (Agilent, Santa Clara, United States).

RNA extraction and qPCR
RNA extraction was performed as described in the protocol of the RNeasy Protect Kit and RNase-Free DNase Set (QIAGEN GmbH, Hilden, Germany).A cDNA Synthesis Kit (Biozym Scientific GmbH, Hessisch Oldendorf, Germany) was used to generate cDNA and the following qPCR was run with Blue S'Green qPCR Mix Separate ROX (Biozym Scientific GmbH, Hessisch Oldendorf, Germany) and analyzed with qTOWER³ (Analytik Jena GmbH, Jena, Germany).Primers (Table 3) used in this study were ordered from Merck KGaA (Darmstadt, Germany).The housekeeping gene for qPCR was gyrAB.

FIG 1 (
FIG 1 (A) Deletion of the entire operon encoding Ech2 via two independent homologous recombination events using plasmid pEch2TK02.Plasmid pEch2Tk02 contained homologous regions for integration at the 5′ or 3′ region flanking the ech2 operon (UFR, upstream flanking region; DFR, downstream flanking region) into the parent strain, T. kivui TKV002 (TKV_MB002) lacking pyrE; and ech2 operon replacement via homologous recombination.For methodological details, see Material and Methods.(B) 1% Agarose gel stained with midori green (Biozym, Hessisch Oldendorf, Germany).The loss of ech2 (6,770 bp) was verified by PCR using primers binding outside of the ech2 cluster.Shown is the electrophoretic separation of the DNA fragments from the PCRs using cells of T. kivui wild type (WT) (expected fragment size, 8.8 kbp) and ∆ech2 mutant (expected fragment size, 2.0 kbp).

FIG 2
FIG 2 Model of acetogenesis from glucose in Thermoanaerobacter kivui (A) wild type (3) and (B) ech2 mutant.(A) Glu cose oxidation provides the required reducing equivalents for the reduction of CO 2 in the WLP.The electron-bifurcating transhydrogenase Nfn concomitantly oxidizes NADH and Fd red from glucose and pyruvate oxidation to provide NADPH to the WLP and to the electron-bifurcating hydrogenase (HydABC) that oxidizes NADPH and Fd red .The produced H 2 is used by HDCR in the WLP.To balance redox carriers, the energy-conserving hydrogenase (Ech complex) oxidizes H 2 to reduce ferredoxin (Fd ox ).(B) Hypothetical takeover of Ech2 function by the HDCR.Ech1-MetFV is assumed to translocate 1 + x H + across the membrane.The generated proton motive force is used to synthesize ATP via ATP synthase.

FIG 3
FIG 3 Model of acetogenesis from H 2 + CO 2 in Thermoanaerobacter kivui (A) wild type (3) and (B) ech2 mutant.(A) H 2 oxidation provides the required reducing equivalents for the reduction of CO 2 in the WLP.The electron-bifurcating hydrogenase (HydABC) and the energy-conserving hydrogenase (Ech complex) oxidize H 2 to reduce ferredoxin (Fd ox ) and NADP + to Fd red and NADPH.(B) In the ech2 mutant, HDCR may catalyze the Fd-dependent H 2 oxidation.Ech1-MetFV is assumed to translocate 1 + x H + across the membrane.The generated proton motive force is used to synthesize ATP via ATP synthase.

FIG 4 (FIG 5
FIG 4 (A) Representative growth curve of the T. kivui Δech2 mutant (gray) and the wild type, strain DSM 2030 (black) with 25 mM glucose.A representative growth curve is shown.(B) Concentrations of glucose (continuous line) and acetate (dashed line).All experiments were performed on complex medium at 65°C and 160 rpm (n = 3).

FIG 8 (
FIG 8 (A) Representative growth curve of the T. kivui strain Δech2 mutant (gray) and the wild-type strain DSM 2030 (black) in the presence of 3 atm H 2 /CO 2 (66/33 vol/vol).H 2 /CO 2 was refilled every hour to a pressure of 3 atm.(B) Acetate concentration of wild type (black) and Δech2 mutant (gray) in media of growing cells or (C) resting cells.Resting cells were incubated for 24 h in a minimal medium and 3 atm H 2 /CO 2 (66/33 vol/vol).All experiments were performed on a complex medium at 65°C and 160 rpm.

FIG 9 (
FIG 9 (A) Representative growth curve of ∆ech2 mutant on H 2 /CO 2 /CO.Growth of the wild type (black) and ∆ech2 mutant (gray) in the presence of 3 atm H 2 /CO 2 /CO.All experiments were performed on complex medium at 65°C and 160 rpm.Gas mix H 2 /CO 2 /CO (44/22/33; vol/vol/vol; [2 atm H 2 /CO 2 (66/33; vol/vol) plus 1 atm pure CO]).(B) Increase of OD 600 of Δech2 mutant and the wild-type strain, DSM 2030 in the presence of CO/N 2 (3 atm; 14/86 vol/vol).The Δech2 mutant and the wild type were inoculated with an OD 600 of around 0.01 (passage 1; solid line), while passage 2 (dashed line) was inoculated with an OD 600 of 0.001 after 23 days from passage 1.All experiments were performed on complex medium at 65°C and 160 rpm (n = 3).

FIG 11 FIG 12
FIG 11 Resting cell experiment of the wild type (black) and ech2 mutant (gray).Incubated for 24 h in Tris-HCL buffer with ca. 100 mM pyruvate at 65 C. (A) Concentration changes in pyruvate (solid line), acetate (dashed line), and formate (dotted line) over 3 h.(B) Concentration at the beginning and after 24 h of pyruvate (filled column), acetate (striped column), and formate (tiled column).Acetate and formate concentrations at the beginning of the experiment were below 0.75 mM, while almost all pyruvate was consumed within 24 h.

TABLE 1
Expression levels of ech1A and ech2D compared to the expression level of the housekeeping gene gyrB (2 -∆Ct ) in cells of the T. kivui ech2 mutant and the wild type (DSM2030) as determined by RT-qPCR.The cells were grown in complex medium with either 25 mM glucose or with H 2 /CO 2 (66/33, vol/vol, 2 atm), and the fourth and seventh rows show the differential expression level (2 -∆∆Ct ) a a The ech2D was not detectable (n.d.) in the ∆ech2 mutant.Therefore, no differential expression level could be calculated.

TABLE 2
Selected SNPs detected in T. kivui strains adapted to CO, and the percent of reads in each strain that contains a given SNP is not necessary for growth on pyruvate, the cells are eventually able to grow on pyruvate, although after a substantial lag phase. disruption (20)O 4 , 50 mM NaH 2 PO 4 , 1.3 mM K 2 HPO 4 , 1.6 mM KH 2 PO 4 , 3.8 mM NaCl, 2.9 mM NH 4 Cl, 0.8 mM (NH 4 ) 2 SO 4 , 0.4 mM MgSO 4 , 3.6 µM FeSO 4 , 5.6 µM CaCl 2 , 1% trace element solu tion DSMZ141, 1% mL vitamin solution DSMZ141.The complex medium additionally contained 2 g L −1 yeast extract (AppliChem GmbH, Darmstadt, Germany).Preparation of the medium was performed as previously described(20).Media was purged with Protadur C20 [80% / 20% (vol/vol) N 2 /CO 2; Westfalen AG, Münster] and autoclaved.If not described otherwise, cells were grown with 25 mM glucose, 25 mM mannitol or 25 mM fructose or 50 mM or 100 mM pyruvate as substrate, each added from sterile, anoxic stock solutions.Autotrophic cultures were grown on H 2 /CO 2 [66%/33%; (vol/vol); 2 × 10 5 Pa, Westfalen AG, Münster] or carbon monoxide (2 × 10 5 Pa, Westfalen AG, Münster).If not described otherwise, all growth experiments were carried out on complex medium under strict anoxic conditions using 100 mL or 200 mL serum bottles filled with 40 mL medium, and sealed with butyl rubber stoppers.For all growth experiments, cultures were inoculated from pre-cultures grown on the same substrate.Growth was monitored by measuring the absorption at 600 nm (OD 600 ).