Pathways and Bioenergetics of Anaerobic Carbon Monoxide Fermentation

Carbon monoxide can act as a substrate for different modes of fermentative anaerobic metabolism. The trait of utilizing CO is spread among a diverse group of microorganisms, including members of bacteria as well as archaea. Over the last decade this metabolism has gained interest due to the potential of converting CO-rich gas, such as synthesis gas, into bio-based products. Three main types of fermentative CO metabolism can be distinguished: hydrogenogenesis, methanogenesis, and acetogenesis, generating hydrogen, methane and acetate, respectively. Here, we review the current knowledge on these three variants of microbial CO metabolism with an emphasis on the potential enzymatic routes and bio-energetics involved.

Keywords: water–gas shift reaction, syngas, carboxydotrophic, hydrogenogenesis, methanogenesis, acetogenesis

Carbon monoxide is a tasteless, odorless gas, best known for its toxic characteristics. It is part of the global carbon cycle, is involved in assimilatory and dissimilatory pathways of numerous microorganisms and was found to act as a signal molecule in mammals (Ryter and Otterbein, 2004). Additionally, it is speculated to be an important substrate for early life on earth (Miyakawa et al., 2002). CO is chemically formed during oxygen limited combustion of carbon materials, and can be biologically generated via cellular metabolism (Eikmanns et al., 1985; Ryter and Otterbein, 2004) or enzymatic degradation of heme (Chin and Otterbein, 2009). A large part of naturally generated CO is assumed to be formed via photochemical reactions (Weinstock and Niki, 1972). Other sources contributing to the atmospheric CO level are: volcanic activity, forest fires, and over the last two centuries industrial activity. Cumulative, these activities result in the production of approximately 2.6 petagram (Pg) CO per year (Khalil and Rasmussen, 1990).

Due to the foreseen depletion of fossil energy sources and consideration on environmental impact of current chemical industry, alternative sustainable technologies are being developed. A bio-based economy is considered one of the sustainable solutions for the growing resource depletion, and has potential to replace petroleum industries in the future. However, the hydrolysis of non-food-competing, ligno-cellulosic biomass limits the application of bio-based technologies, leaving a significant amount of the initial substrate unused (Hamelinck et al., 2005). Gasification of the carbohydrate material, forming syngas, is a potential way of gaining access to all material of the original source, mainly in the form of CO, H2, and CO2. A large spectrum of carbon sources can fuel this technology, from coal, tar, and gas to ligno-cellulosic biomass, all kinds of municipal waste, and digester sludge. As syngas mainly consists of CO, H2, and CO2, autotrophic, CO-tolerant microorganisms are required as biocatalysts for fermentation of this gas. Numerous microorganisms have shown to utilize CO as a substrate, producing organic compounds such as acetate, ethanol, 2,3-butanediol, and butyrate (Latif et al., 2014; Tiquia-Arashiro, 2014; Dürre and Eikmanns, 2015). In addition, methanogenesis and hydrogenogenesis using syngas as a substrate could have large implications for biofuel production. Even though production of interesting chemicals from syngas is possible, the exact metabolism of syngas conversion is not fully understood. This limits the optimization of potential production strains and, thus, the development of bio-based production processes.

Different aspects of biological conversion of CO and syngas have been reviewed in the past decade (Henstra et al., 2007b; Oelgeschlager and Rother, 2008; Bengelsdorf et al., 2013; Sokolova and Lebedinsky, 2013; Latif et al., 2014; Dürre and Eikmanns, 2015). However, a large part of these reviews focus on industrially relevant species and their application potential, while only few address fundamental aspects.

Two types of CO metabolism can be distinguished: respiratory and fermentative (Table 1), the former involving an exogenous electron acceptor, whereas the latter utilizes internally generated intermediates as electron acceptor. A relatively well studied example of respiratory CO metabolism is CO oxidation coupled to oxygen reduction (Meyer and Schlegel, 1983; Meyer et al., 1986). Other, less characterized, electron acceptors that have been identified for carboxydotrophic growth are sulfate (Parshina et al., 2005a, 2010), anthroquinone disulfonate (AQDS; Henstra and Stams, 2004), fumarate (Henstra and Stams, 2004), and ferrihydrite (Slobodkin et al., 2006). The main focus of this review is on fermentative CO metabolism, distinguishing hydrogenogenesis, acetogenesis, and methanogenesis with a special emphasis on the potential enzymatic routes and bio-energetics involved. Classifying hydrogenogenesis as a fermentative process is debatable because protons can be considered exogenous acceptors. However, as protons are present in any microbial environment and are generated from the substrate water during hydrogenogenesis, hydrogenogenic metabolism is considered here as fermentative.

Both aerobic and anaerobic CO metabolism process the CO molecule via the enzyme: carbon monoxide dehydrogenase (CODH). Here, we only consider the anaerobic CODHs, which differ from the aerobic CODHs by structure and the presence of nickel–iron clusters in their active centers (Jeoung et al., 2014). About 6% of all known microbial genomes contain at least one [Ni–Fe] CODH gene sequence, out of which 43% contain at least two, suggesting a more widespread anaerobic CO-utilizing capability than assumed before (Techtmann et al., 2012). It has been shown that CODH genes cluster according to function, instead of clustering by phylogeny, suggesting horizontal gene transfer events have led to the establishment of the CODH gene in the different microbial genomes (Techtmann et al., 2012). CODHs from different organisms have been purified and characterized, including the ones from three relatively well studied anaerobic carboxydotrophic organisms: Rhodospirillum rubrum (Drennan et al., 2001), Carboxydothermus hydrogenoformans (Dobbek et al., 2001), and Moorella thermoacetica (Doukov et al., 2002; Darnault et al., 2003). All the structures of these anaerobic CODHs contain iron-sulfur center (Hu et al., 1996; Dobbek et al., 2001). Additionally, these CODHs contain nickel as a cofactor, for binding and coordinating CO in the active site (Dobbek et al., 2001; Drennan et al., 2001). Other divalent metals were found in the active center of anaerobic CODHs, however, only the nickel containing enzymes were observed to be active in CO conversion (Darnault et al., 2003). CODH can be mono- or bi-functional, both enabling the organism to utilize CO for the energy metabolism. The bifunctional CODH is associated with an acetyl-CoA synthase (ACS), and additionally has a role in carbon fixation, catalyzing the condensation of CO, CoA-SH, and a methyl-group into acetyl-CoA.

The bacterial CODH and ACS are connected via a hydrophobic tunnel (Maynard and Lindahl, 1999; Seravalli and Ragsdale, 2000; Lindahl, 2002), preventing CO from being a toxic intermediate in the metabolism of the cell (Doukov et al., 2002).

The redox potential of the CO/CO2 pair (E0 = –520 mV), is lower than that of H2/H+ (E0 = –414 mV), which has significant implications for the metabolism. While a metabolism driven by hydrogen requires bifurcation mechanisms to reduce ferredoxin (E0 = –400 mV) (Buckel and Thauer, 2013), CO can solely drive this reaction. However, the more negative redox potential of CO poses a challenge for the redox balance of the organism. Therefore, efficient cofactor re-oxidizing pathways are required to avoid the cell from becoming completely reduced. Additionally, hydrogenases are considered to be a weak point in CO metabolism, as hydrogen metabolism is often observed to be rapidly inactivated upon CO exposure (Purec et al., 1962; Daniels et al., 1977; Genthner and Bryant, 1982; Bertsch and Müller, 2015). However, [Ni–Fe]-hydrogenases were found to be less sensitive to CO than [Fe–Fe] or iron-only hydrogenases (Adams, 1990b; De Lacey et al., 2007). Some microorganisms possess [Ni–Fe] hydrogenases that are highly tolerant to CO, such as Rhodospirillum rubrum (Fox et al., 1996b) and Pyrococcus furiosus (Adams, 1990a). CO is known to strongly bind to metals via a process called back bonding, which is also considered to be the mechanism of toxicity (Jeoung et al., 2014).

Coupling CO oxidation to proton reduction is, conceptually seen, one of the simplest mechanisms of biological energy conservation. This reaction, also known as the water–gas shift reaction, results in the formation of hydrogen and CO2 (Table 1). The reaction was found to be completed by three enzymes: CODH, an electron transfer protein and an energy converting hydrogenase (EcH). CO is oxidized via the CODH complex, and electrons are transferred to a “ferredoxin-like” electron carrier. Oxidation of this electron carrier can be coupled to proton reduction via an EcH complex, producing hydrogen and simultaneously generating an ion motive force (Hedderich and Forzi, 2004). Besides being involved in hydrogenogenic metabolism, EcH enzymes also play a role in sugar fermentation (Sapra et al., 2003) and methanogenesis (Thauer et al., 2010). Several microorganisms which hydrogenogenically metabolize CO have been isolated; most of them are thermophiles (Table 2). Two microorganisms conserving energy via the water–gas shift reaction have been rather well studied, the mesophilic Rhodospirillum rubrum (Kerby et al., 1995) and the thermophilic Carboxydothermus hydrogenoformans (Svetlichny et al., 1991).

Rhodospirillum rubrum, Rubrivivax gelatinosa, and Rhodopseudomonas palustris are photosynthetic bacteria and the only known mesophiles capable of efficiently conserving energy from the water–gas shift reaction (Table 2). In R. rubrum the CO-dependent metabolism is regulated via a heme-protein, which acts as CO sensor (CooA) and controls transcription of the enzymatic machinery required for CO dependent growth (Roberts et al., 2001). The genes controlled by CooA in R. rubrum are arranged in two gene clusters: cooFSCTJ and cooMKLXUH. The first gene cluster codes for the active CODH (cooS), electron carrier (cooF) and a nickel inserting complex (cooCTJ) (Kerby et al., 1997), whereas the latter codes for a six subunit EcH complex. The CODH structure of R. rubrum has been resolved to 2.8Å, and is similar to the CODH of anaerobes such as C. hydrogenoformans and M. thermoacetica (Dobbek et al., 2001; Drennan et al., 2001). Electrons from CO oxidation are transferred to an iron–sulfur protein (CooF), which shuttles the electrons to the EcH complex. The CooF complex is tightly associated with the CODH, and was shown to be reduced upon CO exposure (Ensign and Ludden, 1991). Other electron carriers, such as other native ferredoxins from R. rubrum, were ineffective in mediating electron transfer from CODH to the hydrogenase (Ensign and Ludden, 1991). This suggests that the CooF subunit is highly specific for electron transfer from CODH to the hydrogenase. Not only is the R. rubrum CODH efficient in converting CO to CO2, also its CO-induced hydrogenase is well adapted to CO dependent growth (Bonam et al., 1989). The EcH of R. rubrum consists of 6 subunits, of which two subunits, CooH and CooL, are similar to the ones found in some [Ni–Fe] hydrogenases.

In addition, all six subunits show high similarity with complex I NADH:oxidoreductases, which are involved in proton translocation coupled to NADH oxidation (Fox et al., 1996a,b). During activity assays, the EcH is found to function optimally in presence of CODH:CooF, which is theorized to promote forming and maintaining a stable complex (Singer et al., 2006). The CO-induced hydrogenase of R. rubrum is highly CO tolerant, and only shows signs of inhibition above 60% CO in the headspace (Fox et al., 1996b). Despite the seemingly efficient water–gas shift metabolism in R. rubrum, autotrophic growth on solely CO as a carbon source is very slow (Dashekvicz and Uffen, 1979). R. rubrum requires small amounts of yeast extract and acetate as a carbon source to grow efficiently. R. gelatinosus and R. palustris exhibit a similar hydrogenogenic CO metabolism as R. rubrum. However, in contrast to R. rubrum, these bacteria were able to perform the water–gas shift reaction and grow on CO as a sole carbon source, but merely in presence of light (Jung et al., 1999; Maness et al., 2005). Growth was significantly slowed down for R. gelatinosus in the dark, which was not assessed for R. palustris. The growth rate of R. gelatinosus increased significantly after addition of malate as a carbon source (Maness et al., 2005). So, despite efficient energy conservation via the water–gas shift reaction, as shown in the presence of organic carbon sources, autotrophic growth seems very energy intensive for these phototrophic bacteria. The relatively slow growth on CO as a sole carbon source is likely due to the use of the energy demanding Calvin-cycle, which can be considered the main carbon-fixation mechanism in these phototrophic bacteria. All three isolated mesophilic, hydrogenogenic carboxydotrophs are phototrophs.

However, it is unclear why the trait of hydrogenogenic CO-utilization among mesophiles is exclusive to this group.

Among thermophilic hydrogenogens, C. hydrogenoformans is one of the best studied. This bacterium was first thought to only grow fermentatively on CO or pyruvate as substrate, but was later shown to be capable of respiratory growth with CO as well (Henstra and Stams, 2004). With five different CODHs encoded in its genome, it is one of the few organisms known to have multiple CODH types, which is likely related to its exceptional growth capabilities on CO (Wu et al., 2005). C. hydrogenoformans uses a CODH–CooF–EcH complex, which is highly similar to the system found in R. rubrum (Soboh et al., 2002). In contrast to the mentioned mesophilic phototrophs, C. hydrogenoformans is capable of efficient autotrophic growth, using solely CO as energy and carbon source. This characteristic might be assigned to the presence of the Wood–Ljungdahl pathway, which in contrast to the Calvin-cycle is not as energy demanding. Additionally, a turnover rate of 31000 s-1 was found for the CODHII of C. hydrogenoformans (Svetlitchnyi et al., 2001), allowing for fast generation of reduction equivalents and thus a quick energy metabolism. Upon increased hydrogen and carbon dioxide pressure, acetate is produced from CO by C. hydrogenoformans. This suggests acetogenic use of the Wood–Ljungdahl pathway could act as a backup for its hydrogenogenic metabolism (Henstra and Stams, 2011). C. hydrogenoformans and related thermophilic species are suggested to fulfill an important role in the volcanic environments they originate from, ensuring CO concentrations are kept below toxic levels, making life of other non-CO tolerant organisms possible (Techtmann et al., 2009). Furthermore, it seems horizontal gene transfer events have played an important role in the establishment of CO-driven hydrogenogenic metabolism in these environments (Techtmann et al., 2012; Sant’Anna et al., 2015).

When assessing the distribution of isolated hydrogenogenic carboxydotrophic microorganisms (Table 2), thermophilic isolates seem to be more prevalent than mesophilic ones, which contrasts with the solubility of gaseous substrates at elevated temperatures. Temperature increase has two effects on dissolved gases: decreased gas solubility and increased gas diffusion rates. In a hydrogenogenic metabolism the microorganisms use a gaseous substrate, subsequently producing a gaseous product. The thermodynamics of this metabolism thus relies on the concentration of two gases, which is indirectly related to the diffusion rate of these gases. The Km values of the CODH in R. rubrum, C. hydrogenoformans, and the acetogen M. thermoacetica, are in the order of 0.032, 0.018, and 0.01 mM, respectively (Raybuck et al., 1988; Jeon et al., 2005; Seravalli and Ragsdale, 2008). The maximal solubility of CO in water is approximately 1.6 to 0.38 mM, in the range of 273 to 353 K, respectively. Assuming a Km of 0.03 mM and applying simple Michaelis–Menten kinetics, the associated CODH reaction rate at these CO concentrations goes from 98 to 93% of Vmax (Figure 1). This suggests that carboxydotrophic microorganisms are not significantly limited by the maximal solubility of CO at elevated temperatures. The CO diffusion coefficient, an indication of the diffusion rate of the gas, is 2.0 × 10-5 cm2/s in water at 298 K. Compared to 298 K, the estimated diffusion constant at 333 K is two times larger and three times larger at 353 K (approximated by the Stokes–Einstein equation, using the dynamic viscosity of water, Figure 1). This suggests that at increased temperatures, CO is more rapidly supplied to the microorganisms.

Via the same mechanism, the temperature indirectly affects the degree of accumulation of hydrogen in the near vicinity of the microorganism, allowing two times faster removal of H2 at 333 K, and three times at 353 K. This suggests thermophilic hydrogenogenic metabolism suffers less from hydrogen accumulation to thermodynamically unfavorable levels when compared to mesophilic conditions. Therefore, we hypothesize that CO driven hydrogenogenic metabolism is more favorable at higher temperatures when compared to lower temperatures, giving rise to the currently observed temperature distribution of carboxydotrophic hydrogenogenic isolates (Table 2). If this potential advantage also translates into an increased energy yield, and thus a higher growth rate with increasing temperatures is unclear as numerous factors influence the growth rate.

Formation of acetate from CO was first reported by Lynd et al. (1982). Since then many acetogens have been discovered to utilize CO, both homo-acetogenic organisms (i.e., generating solely acetate) and mixed-product acetogenic organisms (i.e., generating a mixture of end-products; Table 3). Large differences can be observed between the growth rates and yields of CO-grown acetogens, indicating differences in their ways of energy conservation and pathways utilized (Table 3). The Wood–Ljungdahl pathway is the central metabolism for acetogenic growth (Ragsdale and Pierce, 2008). In contrast to the Calvin-cycle, the reductive TCA-cycle or the 3-hydroxypropionate cycle, the Wood–Ljungdahl pathway can be used to conserve energy. The Wood–Ljungdahl pathway consists of two branches, which in total require eight reduction equivalents and one ATP to form acetyl-CoA from two CO2. During formation of acetate the ATP investment is regained by substrate level phosphorylation, but no net ATP is gained, thus requiring an ion motive force to conserve net energy. Complex I related RnF complexes have been identified as the cation extruding mechanism in many acetogens, linking ferredoxin oxidation to NAD reduction, simultaneously forming either a sodium or proton gradient (Biegel et al., 2011). Upon consumption of this gradient, ATP is formed from ADP and phosphate via an ATPase (Müller, 2003). The availability of ferredoxin can thus be considered the driving force of energy generation in the autotrophic acetogenic metabolism. It is assumed that in hydrogenotrophic acetogens electron-bifurcating hydrogenases are required to generate reduced ferredoxin (Poehlein et al., 2012; Buckel and Thauer, 2013).

As CO can directly reduce ferredoxin, these bifurcating systems seem redundant during CO-driven growth, but might be utilized to correctly distribute reduction equivalents over the metabolism. Four steps in the Wood–Ljungdahl pathway require input of electrons for fixation of two CO2. The type of cofactors utilized in each of these steps differs per enzyme and per microorganism, making it impossible to propose a general metabolism for acetogenic CO metabolism. This distinct use of cofactors can explain part of the differences observed in yield and growth rate of different acetogens (Table 3). Recently, the energy metabolism for autotrophic growth on H2/CO2 for Acetobacterium woodii, M. thermoacetica, and Clostridium ljungdahlii was reviewed (Schuchmann and Müller, 2014). Here, we assess and compare the CO metabolism of these three species, which are the current model organisms for acetogenic metabolism.

Moorella thermoacetica is one of the best studied homo-acetogenic bacteria able to utilize CO. This organism differs from other acetogenic strains in the sense that it does not possess an RnF complex, leaving the mechanism of cation transport unknown (Pierce et al., 2008). Either an EcH complex or cytochromes are proposed to perform the build-up of an ion motive force. Several cytochromes have been found in Moorella species that are potentially active in proton transport (Gottwald et al., 1975). However, the role of these electron carrier proteins in acetogenic metabolism has never been experimentally shown. Additionally, a role of these cytochromes in respiratory metabolism is likely to exist, as growth with nitrate and nitrite was shown to be possible (Drake and Daniel, 2004). An EcH complex is coded for in the genome of M. thermoacetica (Huang et al., 2012), making cation export via this enzyme a possibility. Based on presence of an EcH complex, an energy metabolism for growth on H2/CO2 has been proposed for M. thermoacetica (Schuchmann and Müller, 2014). The plausible metabolism is based on the assumption that methylenetetrahydrofolate reductase, responsible for the exergonic reduction of methylenetetrahydrofolate to methyltetrahydrofolate, is somehow coupled to energy conservation. As all enzymes of the Wood–Ljungdahl pathway are considered soluble, none of these is expected to be involved in generation of an ion motive force. Crude membrane extraction methods were thought to be the cause of finding all the enzymes in the soluble fraction, and more gentle extraction methods confirmed membrane attachment of methylenetetrahydrofolate reductase (Hugenholtz and Ljungdahl, 1989).

Also in the acetogen Blautia producta, methylenetetrahydrofolate reductase was found to be loosely attached to the cellular membrane, supporting a potential role in energy conservation (Wohlfarth et al., 1990). Further experimental evidence for a direct role of this enzyme in energy conservation has never been found though. A recent theory is that the methylenetetrahydrofolate reductase has bifurcation activity, coupling the oxidation of two NADH molecules to the reduction of ferredoxin and methylenetetrahydrofolate (Huang et al., 2012). The hypothesis for a bifurcating function of this enzyme in M. thermoacetica is supported by the following arguments: i) the enzyme was found to be associated with subunits similar to F420 non-reducing hydrogenases, also found in archaea for bifurcational reduction of CoM–CoB (Huang et al., 2012), ii) in several acetogens the complex was found to contain flavin (Clark and Ljungdahl, 1984; Wohlfarth et al., 1990), which is thought to be essential for bifurcation of the electrons over the two acceptors (Herrmann et al., 2008; Thauer et al., 2008; Kaster et al., 2011; Nitschke and Russell, 2012; Buckel and Thauer, 2013). The flavin is theorized to donate one electron to a high potential acceptor, leaving the flavin at a “red hot” flavosemiquinone state, capable of reducing a low potential acceptor (Buckel and Thauer, 2013). This can be repeated for another two electrons, obtaining two fully reduced products. In case of methylenetetrahydrofolate reductase, the high potential acceptor would be methylenetetrahydrofolate (E0 = –117 mV) whereas ferredoxin (E0 = –400 mV) would be the low potential acceptor. Assuming a bifurcating function of methylenetetrahydrofolate reductase and a proton translocation ratio of 1:1 per hydrogen formed by the EcH complex, the energy yield of H2/CO2 grown M.

thermoacetica was suggested to be 0.5 ATP per acetate formed (Schuchmann and Müller, 2014). Applying a model using similar assumptions, the metabolism on CO is expected to yield 1.5 ATP per acetate formed (Figure 2). The suggested three times increase in energy yield matches with the increased observed growth yield of M. thermoacetica with CO (Table 3). The same yield increase is observed in the related organism M. thermoautotrophica, which is thought to exhibit a similar metabolism (Table 3).

Initially, the acetogen Acetobacterium woodii was reported to grow homo-acetogenically on CO as a sole energy source (Genthner and Bryant, 1987). However, recently it was shown that the organism can only utilize CO in co-fermentation with either hydrogen or formate (Bertsch and Müller, 2015). Additionally, A. woodii was shown to produce ethanol when the pressure of CO in the headspace was over 25 kPa (Bertsch and Müller, 2015). These contradictory observations might be explained by the rich undefined medium (UM) used in the initial study, which makes it possible that the organism has co-fermented CO with other substrates, such as formate (Bertsch and Müller, 2015). This is further supported by findings of the initial study that it was not possible to grow A. woodii with solely CO on defined medium (Genthner and Bryant, 1987). For A. woodii, a metabolism on H2/CO2 was proposed based on genomic data (Poehlein et al., 2012), and was later adapted suggesting a gain of ∼0.3 ATP per acetate formed (Schuchmann and Müller, 2014). A. woodii is suggested to contain a non-bifurcating methylenetetrahydrofolate reductase, using only NADH to form methyltetrahydrofolate (Schuchmann and Müller, 2014). Growth on H2/CO2 for A. woodii was reported with a generation time of approximately 6 h. Presence of increasing levels of CO negatively affected the growth rates on H2/CO2, and became almost fully inhibited above 15 kPa CO (Bertsch and Müller, 2015). A. woodii efficiently co-fermented CO (25 kPa) with formate at a generation time of approximately 5.5 h. Levels up to 50 kPa CO stimulated growth when co-fermenting formate, however, higher CO pressures caused a decrease in growth rate. Growth remained possible up to a maximally tested pressure of 100 kPa CO (Bertsch and Müller, 2015). The inhibitory effect of CO on the A.

woodii metabolism is suggested to be related to its formate dehydrogenase (Bertsch and Müller, 2015), which is associated with a [Fe–Fe] hydrogenase (Schuchmann and Müller, 2013). Additionally, bifurcational [Fe–Fe] hydrogenases present in A. woodii can be a bottleneck in utilization of CO (Figure 3A). In the presence of formate, CO-inhibited enzymes are expected not to be required, facilitating the use of CO as a substrate (Figure 3B).

Unlike M. thermoacetica, which contains an EcH, RnF-utilizing acetogens cannot directly couple ferredoxin oxidation to proton reduction. Therefore, to prevent a completely reduced state of the cell, RnF-containing acetogens need to stoichiometrically couple CO oxidation to the Wood–Ljungdahl pathway. When compared to hydrogenotrophic acetogenesis, acetogenic growth on CO often leads to the formation of additional alcohols, hydrogen, or fatty acids. These by-products are likely formed due to the more strongly reduced environment created by CO. An example is the CO metabolism of E. limosum, which produced solely acetate when grown on H2/CO2, but generates a mixture of acetate and butyrate when grown on CO (Jeong et al., 2015). The best studied pathway for maintaining redox balance during acetogenic growth on CO is solventogenesis. The model organism for this type of fermentation is C. ljungdahlii, which is known for its fast growth rate and solventogenic production characteristics on CO (Köpke et al., 2010). In contrast to A. woodii and M. thermoacetica, the enzymes from the Wood–Ljungdahl pathway in C. ljungdahlii have not been purified and tested for cofactor specificity. Therefore, the proposed metabolism of C. ljungdahlii on H2/CO2 was partly based on genomic data (Schuchmann and Müller, 2014). The proposed metabolism on H2 is to yield a minimum of 0.13 ATP per acetate formed, and can go up to 0.63 ATP per acetate, depending on which cofactors are utilized in each of the steps. Based on the assumptions of the minimal H2 metabolism, a model for CO driven acetogenic growth can be proposed for C. ljungdahlii, and related bacteria (Figure 4). The model shows a yield of about 1.125 ATP per acetate formed (Figure 4), which is further reduced when formation of side products is taken into account.

Despite the fact that the energy yield per acetate formed is less as proposed for M. thermoacetica, the generation time of C. ljungdahlii is shorter. Two mechanisms might contribute to this enhanced growth rate: a bifurcational formate dehydrogenase, and the up-regulation of re-oxidizing reactions. A related bacterium, Clostridium autoethanogenum, was found to highly express a formate dehydrogenase associated with a [Fe–Fe]-bifurcating hydrogenase. This formate dehydrogenase is suggested to use one mole of NADPH and one mole of ferredoxin to reduce two moles of CO2 to formate (Wang et al., 2013). Despite the sensitivity of [Fe–Fe]-hydrogenases to CO, levels in the cell are assumed to be kept low enough for the hydrogenase to function. The utilization of both NADPH and ferredoxin results in re-oxidation of these two important cofactors, and prevents over-reduction of the cell. However, loss of ferredoxin in re-oxidizing reactions reduces its capacity to act as a driving force for cation export via the RnF complex, lowering the energy yield of the overall metabolism. C. ljungdahlii codes for a similar formate dehydrogenase complex in its genome, and might thus utilize a similar system during growth on CO. Additionally, solventogenic reactions play a role in maintaining redox balance. Ethanol production in C. ljungdahlii occurs via two pathways: a direct or an indirect pathway (Köpke et al., 2010, 2011). The direct pathway forms ethanol via acetaldehyde directly from acetyl-CoA, utilizing an acetaldehyde/alcohol dehydrogenase complex. This pathway is expected to largely reduce the overall energy yield as no ATP is generated via acetate formation. The indirect pathway does not omit the ADP phosphorylation step, maintaining the energy conservation via acetate synthesis.

Acetate is theorized to subsequently be reduced with ferredoxin to acetaldehyde via an aldehyde oxidoreductase. This enzyme was shown to be expressed in CO grown C. ljungdahlii, and its expression is stimulated by the addition of external acids (Xie et al., 2015). Ethanol is subsequently formed from acetaldehyde via an alcohol dehydrogenase, utilizing additional reducing equivalents such as NADH or NADPH (Köpke et al., 2010). Energy conservation linked to ethanol formation from H2/CO2 in C. autoethanogenum is also expected to run via this indirect pathway (Mock et al., 2015). Additionally, a similar pathway could be responsible for conversion of different carboxylic acids into alcohols, as observed in mixed cultures exposed to syngas (Liu et al., 2014).

As the suggested energy yield per acetate formed from CO for C. ljungdahlii is 1.125 ATP (Figure 4), 0.125 ATP is generated up to acetyl-CoA formation. This suggests net energy can be yielded by this organism from any product formed from acetyl-CoA which does not require further investment of ATP. This could include compounds such as C4-carboxylic acids, lactate, fatty acids, and a variety of alcohols, which are considered interesting end products of bio-based processes (Dürre and Eikmanns, 2015).

Methane can be formed anaerobically from different substrates, such as H2/CO2, formate, methanol, acetate, or methylamines. The physiology and bioenergetics of different types of methanogenic metabolism have been reviewed before (Thauer et al., 2008). CO is a less studied substrate of methanogenesis, and was first reported by Fischer et al. (1931). It has thereafter been shown to be utilized in the metabolism of several methanogenic strains (Table 4). Methanogenic growth on CO as sole substrate appears not to be very efficient, as only three species have been reported to grow while producing methane: Methanobacterium thermoautotrophicus, Methanosarcina acetivorans, and Methanosarcina barkeri. To gain insight in CO utilization by different groups of methanogens, two main types of methanogenic metabolism are of interest: hydrogenotrophic and aceticlastic methanogenesis. CO is an intermediate in both types of methanogenic metabolism: playing a role in anabolism or catabolism of acetyl-CoA. It is therefore not surprising that genes coding for CODH in methanogens are mainly located in the genome as neighbor of an ACS (Techtmann et al., 2012). Some methanogens do, however, possess more than one CODH, which are not all associated with an ACS complex. Examples are Methanothermobacter marburgensis and Methanococcus jannaschii. M. jannaschii has a CODH which is in located the same operon as a hydrogenase, suggesting that hydrogenogenic carboxydotrophic metabolism is possible. However, many methanogenic strains have never been tested for utilization of CO as a substrate, and their growth potential on CO is therefore unknown. Unlike in many bacteria, the routes for electron transfer from CODH to the rest of the metabolism are not well established for archaea. In general, “ferredoxin-like” proteins are proposed as the acceptor molecules for CODH complexes.

This is confirmed for some methanogens, such as Methanosarcina thermophila (Terlesky and Ferry, 1988; Abbanat and Ferry, 1991) and Methanosarcina barkeri (Fischer and Thauer, 1990), which require ferredoxin to perform CO-dependent reactions. However, cell-free extract of M. thermoautotrophicus exhibited F420-reducing activity in presence of CO, while ferredoxin of Clostridium pasteurianum was not reduced (Daniels et al., 1977). F420 was also observed to be reduced by purified CODH from Methanosaeta concilii, indicating potential ability of this enzyme to reduce this cofactor (Jetten et al., 1989). However, ferredoxin was not tested as acceptor for the CODH complex of M. concilii, and thus cannot be excluded as acceptor. When assuming E = –500 mV for ferredoxin under physiological conditions (Buckel and Thauer, 2013), it can be expected that ferredoxin is an ideal acceptor for electrons from CO (E0 = –520 mV). Transfer of electrons to F420 (E0 = –380 mV) would result in additional energy loss, and would require bifurcation processes to generate reduced ferredoxin. Ferredoxins have in general been shown to be interchangeable over long phylogenetic ranges, as indicated by the similar characteristics of plant and bacterial ferredoxins (Tagawa and Arnon, 1968). However, ferredoxin from spinach did not form a complex with the CODH of M. thermoacetica and was not reduced by this bacterial CODH (Drake et al., 1980; Shanmugasundaram and Wood, 1992). Additionally, the observation that the CODH from R. rubrum could only effectively donate electrons to the tightly associated CooF complex (Ensign and Ludden, 1991), suggests not all ferredoxins are efficient in receiving in electrons from a CODH complex.

Therefore, the observation that some archaeal CODHs do not transfer electrons to non-native ferredoxin might be related to a difference in characteristics, such as structure and location of the special nickel–iron–sulfur clusters present in the CODH and the possibly different mid-point redox potentials of the ferredoxin.