Acetogenesis and the Wood-Ljungdahl Pathway of CO₂ Fixation

I. Summary

Conceptually, the simplest way to synthesize an organic molecule is to construct it one carbon at a time. The Wood-Ljungdahl pathway of CO₂ fixation involves this type of stepwise process. The biochemical events that underlie the condensation of two one-carbon units to form the two-carbon compound, acetate, have intrigued chemists, biochemists, and microbiologists for many decades. We begin this review with a description of the biology of acetogenesis. Then, we provide a short history of the important discoveries that have led to the identification of the key components and steps of this usual mechanism of CO and CO₂ fixation. In this historical perspective, we have included reflections that hopefully will sketch the landscape of the controversies, hypotheses, and opinions that led to the key experiments and discoveries. We then describe the properties of the genes and enzymes involved in the pathway and conclude with a section describing some major questions that remain unanswered.

II. Introduction

In 1945 soon after Martin Kamen discovered how to prepare ¹⁴C in the cyclotron, some of the first biochemical experiments using radioactive isotope tracer methods were performed to help elucidate the pathway of microbial acetate formation [1]. Calvin and his coworkers began pulse labeling cells with ¹⁴C and using paper chromatography to identify the ¹⁴C-labeled intermediates, including the phosphoglycerate that is formed by the combination of CO₂ with ribulose diphosphate, in what became known as the Calvin-Benson-Basham pathway. However, though simple in theory, the pathway of CO₂ fixation that is used by M. thermoacetica proved to be recalcitrant to this type of pulse-labeling/chromatographic analysis because of the oxygen sensitivity of many of the enzymes and because the key one-carbon intermediates are enzyme bound. Thus, identification of the steps in the Wood-Ljungdahl pathway of acetyl-CoA synthesis has required the use of a number of different biochemical, biophysical, and bioinorganic techniques as well as the development of methods to grow organisms and work with enzymes under strictly oxygen-free conditions.

III. Importance of acetogens

A. Discovery of acetogens

Acetogens are obligately anaerobic bacteria that use the reductive acetyl-CoA or Wood-Ljungdahl pathway as their main mechanism for energy conservation and for synthesis of acetyl-CoA and cell carbon from CO₂ [2, 3]. An acetogen is sometimes called a “homoacetogen” (meaning that it produces only acetate as its fermentation product) or a “CO₂-reducing acetogen”.

As early as 1932, organisms were discovered that could convert H₂ and CO₂ into acetic acid (Equation 1) [4]. In 1936, Wieringa reported the first acetogenic bacterium, Clostridium aceticum [5, 6]. M. thermoacetica [7], a clostridium in the Thermoanaerobacteriaceae family, attracted wide interest when it was isolated because of its unusual ability to convert glucose almost stoichiometrically to three moles of acetic acid (Equation 2) [8].

$$2{\text{CO}}_2+4{\mathrm{H}}_2\rightleftharpoons {\text{CH}}_3{\text{COO}}^{-}+{\mathrm{H}}^{+}+2{\mathrm{H}}_{2}0\mathrm{\Delta }{\mathrm{G}}^0’=-95\text{kJ}/\text{mol}$$ (1)

$${\mathrm{C}}_6{\mathrm{H}}_{12}{\mathrm{O}}_6\to 3{\text{CH}}_3{\text{COO}}^{{-}_{-}}+3{\mathrm{H}}^{+}\mathrm{\Delta }{\mathrm{G}}^0’=-310.9\text{kJ}/\text{mol}$$ (2)

B. Where acetogens are found and their environmental impact

Globally, over 10¹³ kg (100 billion US tons) of acetic acid is produced annually, with acetogens contributing about 10% of this output [2]. While most acetogens like M. thermoacetica are in the phylum Firmicutes, acetogens include Spirochaetes, δ-proteobacteria like Desulfotignum phosphitoxidans, and acidobacteria like Holophaga foetida. Important in the biology of the soil, lakes, and oceans, acetogens have been isolated from diverse environments, including the GI tracts of animals and termites [9, 10], rice paddy soils [11], hypersaline waters [12], surface soils [13, 14], and deep subsurface sediments [15]. Acetogens also have been found in a methanogenic mixed population from an army ammunition manufacturing plant waste water treatment facility [16] and a dechlorinating community that has been enriched for bioremediation [17].

For organisms that house acetogens in their digestive systems, like humans, termites, and ruminants [18–20], the acetate generated by microbial metabolism is a beneficial nutrient for the host and for other microbes within the community. In these ecosystems, acetogens can compete directly with hydrogenotrophic methanogenic archaea, or interact syntrophically with acetotrophic methanogens that use H₂ and CO₂ to produce methane [20]. In the termite gut, acetogens are the dominant hydrogen sinks [21], and it has been proposed that acetate is the major energy source for the termite [22]. Like methanogens, acetogens act as a H₂ sink, depleting H₂ that is generated in anaerobic environments during the natural biodegradation of organic compounds. Since the build-up of H₂ inhibits biodegradation by creating an unfavorable thermodynamic equilibrium, acetogens enhance biodegradative capacity by coupling the oxidation of hydrogen gas to the reduction of CO₂ to acetate.

Methanogens are the dominant hydrogenotrophs in many environments since methanogens have a lower threshold for H₂ than acetogens [23] and since the energy yield from the conversion of CO₂ and H₂ to methane is greater than that for conversion to acetate [24, 25]. Under such conditions, acetogens must often resort to other metabolic pathways for growth and, thus, have a highly diverse metabolic menu that comprises the biodegradation products of most natural polymers like cellulose and lignin, including sugars, alcohols, organic acids and aldehydes, aromatic compounds, and inorganic gases like CO, H₂, and CO₂. They also can use a variety of electron acceptors, e.g., CO₂, nitrate, dimethylsulfoxide, fumarate, and protons.

IV. Importance of the Wood-Ljungdahl pathway

A. The Wood-Ljungdahl pathway in diverse metabolic pathways

The Wood-Ljungdahl pathway (Figure 1) is found in a broad range of phylogenetic classes, and is used in both the oxidative and reductive directions. The pathway is used in the reductive direction for energy conservation and autotrophic carbon assimilation in acetogens [26–28]. When methanogens grow on H₂ + CO₂, they use the Wood-Ljungdahl pathway in the reductive direction (like acetogens) for CO₂ fixation [29, 30]; however, they conserve energy by the conversion of H₂ + CO₂ to methane (Equation 2). Given that hydrogenotrophic methanogens assimilate CO₂ into acetyl-CoA, it is intriguing that they do not make a mixture of methane and acetate. Presumably this is governed by thermodynamics, since the formation of methane is −36 kJ/mol more favorable (Equation 3) than acetate synthesis. Aceticlastic methanogens [31] exploit this advantageous equilibrium to generate metabolic energy by interfacing the Wood-Ljungdahl pathway to the pathway of methanogenesis (Equation 3). In this reverse direction, the combined actions of acetate kinase [32, 33] and phosphotransacetylase [34] catalyze the conversion of acetate into acetyl-CoA. Sulfate reducing bacteria also run the Wood-Ljungdahl pathway in reverse and generate metabolic energy by coupling the endergonic oxidation of acetate to H₂ and CO₂ (the reverse of equation 1) to the exergonic reduction of sulfate to sulfide (ΔG⁰’ = −152 kJ/mol), with the overall process represented by Equation 4 [35–37].

$${\text{CO}}_2+4{\mathrm{H}}_2\rightleftharpoons {\text{CH}}_4+2{\mathrm{H}}_{2}0\mathrm{\Delta }{\mathrm{G}}^0’=-131\text{kJ}/\text{mol}$$ (2)

$${\text{CH}}_3{\text{COO}}^{-}+{\mathrm{H}}^{+}\rightleftharpoons {\text{CH}}_4+{\text{CO}}_2\mathrm{\Delta }{\mathrm{G}}^{\mathrm{o}}’=-36\text{kJ}/\text{mol}{\text{CH}}_4$$ (3)

$${{\text{SO}}_4}^{2-}+{\text{CH}}_3{\text{COO}}^{-}+2{\mathrm{H}}^{+}\rightleftharpoons {\text{HS}}^{-}+2{\text{CO}}_2+2{\mathrm{H}}_{2}0\mathrm{\Delta }{\mathrm{G}}^{\mathrm{o},}=-57\text{kJ}/\text{mol}$$ (4)

Figure 1.

The Wood-Ljungdahl pathway. “H₂” is used in a very general sense to designate the requirement for two electrons and two protons in the reaction.

B. Historical Perspective: Key stages in elucidation of the Wood-Ljungdahl pathway

1. Isolation of Moorella thermoacetica (f. Clostridium thermoaceticum) and demonstration by isotope labeling studies that acetogens fix CO₂ by a novel pathway

Chapter 1 of the story of the Wood-Ljungdahl pathway begins with discovery of C. aceticum, the first isolated organism that was shown to grow by converting hydrogen gas and carbon dioxide to acetic acid [5, 6]. Unfortunately, this organism was “lost”, and the baton was passed to Moorella thermoacetica, which was named Clostridium thermoaceticum when it was isolated in 1942 [38] and was so-called until fairly recently when the taxonomy of the genus Clostridium was revised [7]. Thus, M. thermoacetica became the model acetogen, while C. aceticum hid from the scientific community for four decades, until G. Gottschalk, while visiting Barker’s laboratory, found an old test tube containing spores of the original C. aceticum strain and reactivated and reisolated this strain [39, 40]. Although M. thermoacetica did not reveal its potential to grow on H₂ + CO₂ until 1983 [41], the recognition of a homoacetogenic fermentation of glucose [38], implied that it could use the reducing equivalents from glucose oxidation to convert CO₂ to acetate.

Wood and others embarked on the characterization of this pathway at about the same time that Calvin began to study the CO₂ fixation pathway that bears his name [42, 43]. As with the Calvin cycle, isotope labeling studies provided important information regarding the mechanism of CO₂ fixation. When ¹⁴CO₂ was used, approximately an equal concentration of ¹⁴C was found in the methyl and carboxyl positions of acetate, leading Barker and Kamen to suggest that glucose fermentation occurred according to Equation 5 and Equation 6, where “*” designates the labeled carbon [1]. Because ¹⁴CO₂ was used as a tracer in these experiments, one could not distinguish a product that contained an equal mixture of ¹⁴CH₃¹²COOH/¹²CH₃¹⁴COOH from one containing doubly labeled ¹⁴CH₃¹⁴COOH. Wood used mass spectroscopy to analyze the acetate produced from a fermentation containing a significant percentage (i.e., 25%) of ¹³CO₂ and could conclusively demonstrate that the CO₂ is indeed incorporated nearly equally into both carbons of the acetate, confirming equation 6 [44]. It was subsequently shown that carbons 3 and 4 of glucose are the source of the CO₂ that is fixed into acetate [45]. The Embden-Meyerhof-Parnas pathway converts glucose to two molecules of pyruvate, which undergo decarboxylation to form two molecules of acetyl-CoA and, thus, two molecules of acetate (Equation 5). The third molecule of acetic acid is generated by utilizing the eight-electrons released during glucose oxidation to reduce the two molecules of CO₂ released by pyruvate decarboxylation (Equation 6). Figure 2 describes the level of understanding of the Wood-Ljungdahl pathway in 1951 in a scheme reproduced from a landmark paper by H.G. Wood [44], which was published in the same year that Wood and Utter suggested that the mechanism of acetate synthesis could constitute a previously unrecognized mechanism of autotrophic CO₂ fixation [46].

$${\mathrm{C}}_6{\mathrm{H}}_{12}{\mathrm{O}}_6+2{\mathrm{H}}_2\mathrm{O}\to 2{\text{CH}}_3\text{COOH}+8\mathrm{H}+2{\text{CO}}_2$$ (5)

$$8\mathrm{H}+2^{*}{\text{CO}}_2{\to }^{*}{\text{CH}}_3^{*}\text{COOH}+2{\mathrm{H}}_{2}0$$ (6)

Figure 2.

Working model of the Wood-Ljungdahl pathway circa 1951.

It is ironic that M. thermoacetica became the model organism for elucidating the Wood-Ljungdahl pathway of autotrophic CO₂ fixation, since it was only known as a heterotroph until 1983 [41]. However, as mentioned above, C. aceticum appeared to have been lost, and Acetobacterium woodii (the next acetogen could be cultured on H₂ and CO₂) was not isolated until 1977 [47]. The anaerobic methods described by Balch et al. [48] have since led to the isolation of many anaerobes, including many acetogens, that can grow autotrophically, and over 100 acetogenic species, representing 22 genera, have so far been isolated [49].

Based on the microbiological, labeling and mass spectrometric results described above and further ¹⁴CO₂ labeling experiments that ruled out the Calvin cycle and the reductive citric acid cycle [50], it seemed clear that homoaceogenesis involved a novel CO₂ fixation pathway. As described in detail below, the “C1” and “X-C₁” precursors shown in Figure 2 were found to involve C₁ units bound to H₄folate, to cobalamin, and to a nickel center in a highly oxygen-sensitive enzyme. Thus, elucidation of the biochemical steps in the Wood-Ljungdahl pathway took many years and required and the use of many biochemical and biophysical methods.

2. Demonstration that corrinoids and tetrahydrofolate are involved in the Wood-Ljungdahl Pathway

Chapter 2 in the history of the Wood-Ljungdahl Pathway is the discovery of the role of corrinoids and tetrahydrofolate (H₄folate). Corrinoids contain a tetrapyrrolic corrin ring with a central cobalt atom, which can exist in the 1+, 2+, and 3+ oxidation states. Cobalamin (as vitamin B₁₂) had been identified in the mid-1920’s as the antipernicious anemia factor [51, 52] and had only recently been isolated [53, 54] and crystallized [54] at the time of the early pulse labeling studies of the CO₂ fixation pathway. The structure of cobalamin was not determined until 1956 [55]. The initial indication that a corrinoid could be involved in this pathway was based on two key findings: (1) intrinsic factor (known to inhibit vitamin B₁₂-dependent reactions) inhibits incorporation of the methyl group of methyl-B₁₂ into the C-2 of acetate, and (2) in the presence of pyruvate (as an electron and carboxyl group donor), the radioactive methyl group of ¹⁴C-methylcobalamin is incorporated into the methyl group of acetate [56] (Equation 7). These results suggested to Wood and coworkers why they had been unable to isolate intermediates in deproteinated solutions by pulse labeling with ¹⁴CO₂ - the labeled intermediates were bound to cofactors on proteins!

$${}^{14}\text{C}{\text{H}}_3-\text{cobalamin}+{\text{CH}}_3\text{COCOOH}+{\mathrm{H}}_2\mathrm{O}{\to }^{14}{\text{CH}}_3-\text{COOH}+{\text{CH}}_3-\text{COOH+}{\mathrm{H}}^{+}+\text{cobalamin}$$ (7)

It was then shown that pulse labeling M. thermoacetica cells with ¹⁴CO₂ led to the formation of ¹⁴CH₃-labeled corrinoids; furthermore, when the ¹⁴CH₃-corrinoids were added to cell extracts in the presence of pyruvate and CoA, ¹⁴C-labeled acetate was formed [57]. Based on these results, it was proposed that H₄folate and cobalamin, shown as [Co] in Figure 3, were involved in this novel CO₂ fixation pathway [58]. This proposal was based partly on an analogy with methionine synthase, which requires both H₄folate [59] (as methyl-H₄folate) and a corrinoid (as methylcobalamin) [60, 61] to convert homocysteine to methionine. Thus, it was postulated [62] that CO₂ is converted to HCOOH and then to CH₃-H₄folate via the same series of H₄folate-dependent enzymes (vide infra) that had been studied in Clostridium cylindrosporum and Clostridium acidi-urici by Rabinowitz and coworkers [63]. Subsequently, M. thermoacetica cell extracts were shown to catalyze conversion of the methyl group of ¹⁴CH₃-H₄folate (isolated from whole cells that had been pulse labeled with ¹⁴CO₂) [64] to ¹⁴C-acetate [65]. As described below, Ljungdahl and coworkers isolated and characterized the H₄folate-dependent enzymes that catalyze the conversion of formate to methyl-H₄folate [66]. Although a role for cobalamin in acetogenesis was described around 1965 [56, 58], the required cobalamin-containing enzyme was isolated nearly two decades later [67] - the first protein identified to contain both cobalamin and an iron-sulfur cluster [68]. The properties of the H₄folate- and cobalamin-dependent enzymes involved in the pathway are described below.

Figure 3.

The pathway of CO₂ fixation into acetyl-CoA circa 1966, modified from [58].

H₄folate and the H₄folate-dependent enzymes involved in the acetyl-CoA pathway play key roles in one-carbon transfers for a number of essential cell functions (synthesis of serine, thymidylate, purines, and methionine), as they do in all bacteria and eukaryotes; however, because they are involved in a key catabolic pathway in acetogenic bacteria, they are found at levels 1000-fold higher and with ca. 100-fold higher specific activity than in other organisms.

3. Isolation of CO dehydrogenase/acetyl-CoA synthase and the elusive corrinoid protein and determination of their roles in the Wood-Ljungdahl pathway

Between 1980 and 1985, all components of the Wood-Ljungdahl pathway were purified and their roles were identified. In 1981, Drake, Hu, and Wood isolated five fractions from M. thermoacetica that together could catalyze the conversion of ¹⁴CH₃-H₄folate and pyruvate to acetylphosphate (Equation 8), and purified four of these components to homogeneity. The purified enzymes were pyruvate ferredoxin oxidoreductase, ferredoxin, methyltransferase, and phosphotransacetylase, while the fifth fraction (called Fraction F3) contained several proteins [69]. If CoA was substituted for phosphate, acetyl-CoA was the product (Equation 9) and phosphotransacetylase was not required.

$${}^{14}\text{C}{\text{H}}_3{\mathrm{H}}_4\text{folate}+{\text{CH}}_3{\text{COCOO}}^{-}+2{\text{HPO}}_4^{-}{\to }^{14}{\text{CH}}_3-{\text{COPO}}_4^{-}+{\text{CH}}_3-{\text{COPO}}_4^{-}+{\mathrm{H}}_2\mathrm{O}+{\mathrm{H}}_4\text{folate}$$ (8)

$$\begin{array}{l}{}^{14}\text{C}{\text{H}}_3-{\mathrm{H}}_4\text{folate}+{\text{CH}}_3{\text{COCOO}}^{-}+2{\text{CoAS}}^{-}{\to }^{14}{\text{CH}}_3-\text{COSCoA}+{\text{CH}}_3-\text{COSCoA}+{\mathrm{H}}_2\mathrm{O}+\\ {\mathrm{H}}_4\text{folate}\hfill \end{array}$$ (9)

At approximately the same time that Wood’s group was characterizing the five components, Gabi Diekert and Rolf Thauer discovered CO dehydrogenase (CODH) activity in M. thermoacetica [70] and established a growth requirement for nickel [71]. Drake et al. partially purified CODH and provided direct evidence that it contains Ni, using ⁶³Ni [72]. CODH catalyzes the oxidation of CO to CO₂ (Equation 10) and transfers the electrons to ferredoxin, or a variety of electron acceptors [73, 74]. Hu et al. then demonstrated that CODH was a component of Fraction F3 and that CO could substitute for pyruvate (Equation 11), and, in this case, only three components were required for acetyl-CoA synthesis: Fraction F3, ferredoxin, and methyltransferase [75]. It was not yet clear whether CODH was a required component or an impurity in Fraction F3, and it was considered that it might act as an electron donor in the reaction. Hu et al. also discovered that Fraction F3 alone could catalyze an exchange reaction between CO and the carbonyl group of acetyl-CoA, as shown in Equation 12 [75]. Discovery of this exchange reaction enabled studies that led to the discovery of the role of CODH in acetyl-CoA synthesis (see below).

$$\text{CO}+{\mathrm{H}}_2\mathrm{O}\rightleftharpoons {\text{CO}}_2+2{\mathrm{e}}^{-}+2{\mathrm{H}}^{+}\mathrm{\Delta }{\mathrm{E}}_{\mathrm{o}}^{’}=-540\text{mV}$$ (10)

$${}^{14}\text{C}{\text{H}}_3{\mathrm{H}}_4\text{folate}+\text{CO}+{\text{CoAS}}^{-}{\to }^{14}{\text{CH}}_3-\text{COSCoA}+{\mathrm{H}}_4\text{folate}$$ (11)

$$\text{CO}+\mathrm{C}{\mathrm{H}}_{3^{-}}{}^{14}\text{CO}-\text{SCoA}{\rightleftharpoons }^{14}\text{CO}+\mathrm{C}{\mathrm{H}}_3-\text{CO}-\text{SCoA}$$ (12)

The scheme shown in Figure 4 stirred up the Ljungdahl laboratory, where one of the authors (SWR) was a graduate student. The concept of a bound form of formate was under intense discussion in a number of laboratories. As described in Figure 4, CODH was proposed to generate [HCOOH], which might react with the methylated corrinoid protein and CoA to generate acetyl-CoA [75]. For example, as described below, a carboxymethyl-corrinoid was being discussed as a key intermediate in CO₂ fixation. Early in his graduate program at the University of Georgia, SWR had the opportunity to work for a few weeks in the Wood laboratory at Case Western University to learn how to isolate the various fractions and to assay the total synthesis of acetate.

Figure 4.

Scheme circa 1982 proposing roles for the corrinoid protein and CODH in anaerobic autotrophic CO₂ fixation, modified from Fig. 2 of Hu et al. [75].

The proposal related to the red arrow shown in Figure 4 became a focus of intense discussion in the Ljungdahl laboratory and a series of studies were initiated to test the hypothesis that the [HCOOH] might directly serve as the source of the formyl group of formyl-H₄folate, thus bypassing formate dehydrogenase [75] (Ljungdahl’s favorite enzyme).

Following the aphorism of Efraim Racker ("Don't waste clean thinking on dirty enzymes"), we aimed to purify all the enzymes (including all the H₄folate-dependent enzymes, formate dehydrogenase, methyltransferase, and “Fraction F3”) required for the synthesis. Many of these had previously been isolated; however, we wished to also use purified CO dehydrogenase, an enzyme that was known to be extremely oxygen sensitive, and the only reported purification had yielded an impure enzyme that rapidly lost activity [72]. Having visited the Wood laboratory, SWR recognized that they probably had been unsuccessful with the purification because the anaerobic set-up (as described in [69]) was insufficiently rigorous. As students in Ljungdahl’s laboratory, we had experience with the strictly anaerobic methods (including an anaerobic chamber) required for purifying formate dehydrogenase and had purified many of the H₄folate-dependent enzymes. So, we set up a small anaerobic chamber in a cold room (maintained at 10 °C) in which the normal light bulb was replaced with a red light in case the reaction proved to be light sensitive. With the anaerobic set-up, the requirement for formate dehydrogenase in the pathway was established [76]; therefore, the red arrow shown in Figure 4 could be erased and replaced with the formate dehydrogenase-and formyl-H₄folate synthetasecatalyzed reactions, as shown in Figure 5, and as had been proposed in 1966. Furthermore, we were elated that the first attempt at purifying CODH yielded a homogeneous enzyme with specific activity that was 10-fold higher than had ever been measured [73]. Although CODH was indeed very oxygen sensitive, it was not sensitive to light or heat and, when purified and stored under strictly anaerobic conditions, the enzyme was remarkably stable. The CODHs from M. thermoacetica [73, 77] and Acetobacterium woodii [78] were characterized as two-subunit nickel-iron-sulfur metalloenzymes. However, the role of CODH in the acetyl-CoA pathway remained obscure.

Figure 5.

Scheme circa 1983 after re-establishment of formate dehydrogenase and formyl-H₄folate synthetase in the pathway.

Besides CODH, another component of Fraction F3 was the elusive corrinoid enzyme, which Hu purified and showed to undergo methylation to form ¹⁴CH₃-corrinoid when incubated with the methyltransferase and ¹⁴CH₃-H₄folate [75]. Furthermore, in the presence of Fraction F3, CO, and CoA, the radioactive methyl group of the ¹⁴CH₃-labeled corrinoid protein was converted to ¹⁴C-acetyl-CoA (Equation 13).

$${}^{14}\text{C}{\text{H}}_3-[\text{Co}]-\text{Enzyme}+\text{CO}+{\text{CoAS}}^{-}{\to }^{14}{\text{CH}}_3-\text{COSCoA}+[\text{Co}]-\text{Enzyme}$$ (13)

Several key questions remained. The roles of CODH and the corrinoid protein in the pathway were unclear. The chemical nature of [HCOOH] was unknown. It also remained to be elucidated how the methyl, carbonyl, and coenzyme A groups were combined in generating the C-C and C-S bonds of acetyl-CoA.

a) Identification of the Corrinoid Protein as Methyl Carrier and CODH as the acetyl-CoA synthase: Bioorganometallic Chemistry

Regarding the roles of CODH and the corrinoid enzyme in acetyl-CoA formation, it was hypothesized (although incorrectly, vide infra) that CO and the methyl group combine on the corrinoid protein to generate a carboxymethyl- or perhaps an acetyl-corrinoid intermediate that would then be cleaved by CoA to generate acetyl-CoA. This hypothesis of the corrinoid protein acting as the site of acetate assembly was partly based on the finding that M. thermoacetica extracts catalyze the formation of acetate from carboxymethylcobalamin and the observation that the conversion of ¹⁴CO₂ to a ¹⁴C-labeled corrinoid, which undergoes photolysis to generate products that had been previously identified as the photolysis products of carboxymethylcobalamin [57]. Furthermore, in the mid-1980’s, it was unclear whether the methyl group reacted as an anion as a cation. Since alkylcobamides had been considered to act as biological Grignard reagents [79] (although we now recognize that the alkyl group is actually transferred as an carbocation, i.e., CH₃ + [80]), a mechanism of acetate synthesis involving attack of the methyl carbanion on a carboxy group, forming an acetoxycorrinoid was proposed [81]. This mechanism was based partly on the finding that M. thermoacetica extracts catalyzed the incorporation of approximately 50% of the deuterium from CD₃-H₄folate or [CD₃]-methylcobalamin to trideuteromethyl-acetate. The proposal of a corrinoid-catalyzed methyl radical mechanism also seemed feasible since photolysis of methylcobalamin (in solution) in the presence of a high CO pressure (31 atm.) was shown to generate acetyl-cobalamin via a radical mechanism [82]. Thus, it seemed reasonable that the assembly of acetate occurred on a corrinoid protein though it was unknown whether the synthesis occurred from an acetylcobalt, acetoxycobalt, or a carboxymethylcobalt intermediate. Evidence for an acetyl-intermediate was provided by experiments demonstrating that when the methylated corrinoid protein was incubated with CO, acetate (albeit, low amounts) was formed; however, in the presence of CoA, acetyl-CoA was formed [75].

Clarification of the actual roles of the corrinoid protein and CODH exemplifies a remark attributed to Albert von Szent-Györgyi: "Very often, when you look for one thing, you find something else". As noted above, Hu et al. had discovered that Fraction F3 alone could catalyze an exchange reaction between CO and the carbonyl group of acetyl-CoA (Equation 12) [75]. This assay was much less complicated than the typical assay for acetyl-CoA synthesis, which involved ¹⁴CH₃-H₄folate, CO (or pyruvate and pyruvate ferredoxin oxidoreductase), CoA, methyltransferase, the corrinoid protein, plus Fraction F3. Because this assay required the disassembly and reassembly of acetyl-CoA, SWR surmised that purifying the component required in the exchange would uncover the key missing component(s) required for acetyl-CoA synthesis. Since it was thought that the assembly of acetyl-CoA occurred on the corrinoid protein (above), it came as a surprise that the purified corrinoid protein could not catalyze this exchange [83]. This was interpreted to mean that some component of Fraction F3 plus the corrinoid protein was required. However, the shocking results were that CO dehydrogenase was the only required enzyme and that addition of the corrinoid protein had no effect [83]. Thus, as depicted in Figure 6, it was concluded that the role of CODH is to catalyze the assembly of acetyl-CoA from CO, a bound methyl group, and CoA, and that the corrinoid protein serves to accept the methyl group from CH₃-H₄folate and then transfer it to ACS. As it so happened, what was found was more exciting than what was being looked for. Because CODH played the central role in this new scheme, the enzyme was renamed acetyl-CoA synthase (ACS) [83]. It took several years to recognize that CO oxidation and acetyl-CoA synthesis are catalyzed at separate metal centers [84] and this bifunctional enzyme is now called CODH/ACS [85].

Figure 6.

Scheme circa 1985 showing that CODH is ACS, the central enzyme in the pathway, and that the corrinoid enzyme is a methyl carrier. The red arrows designate the reactions involved in the CO/acetyl-CoA exchange. Modified from [83].

However, the nature of the “C1” shown in Figure 6 (equivalent to [HCOOH] in other schemes) was unknown. Furthermore, it was not yet recognized that CO not only serves as a precursor of the carbonyl group of acetyl-CoA, but it is actually generated from CO₂ or pyruvate as an intermediate in the acetyl-CoA synthesis [86]. The nature of “C1” or “[HCOOH]” is described in the next section.

b) Identification of the enzyme-bound precursor of the carbonyl group of acetyl-CoA

Figure 4–Figure 6 portray a “C1” or “HCOOH” intermediate, which designates a bound form of CO that derives from CO₂, CO, or the carboxyl group of pyruvate. To attempt to identify this intermediate, various biochemical spectroscopic studies were initiated (see [87–90] for reviews). CODH was incubated with CO and was found to generate an EPR-detectible intermediate [91, 92] that exhibited hyperfine splitting from both ⁶¹Ni and ¹³CO, indicating that an organometallic Ni-CO complex had been formed. This EPR signal eventually was shown to be elicited from a complex between CO and a nickel iron-sulfur cluster and was thus called the “NiFeC species” [93]. The electronic structure of this center is discussed below. However, the identity of this EPR-active species as the active carbonylating agent has been actively debated [87–89, 94], which is also discussed below.

V. General aspects of the Wood-Ljungdahl pathway

A. Coupling of acetogenesis to other pathways allows growth on diverse carbon sources, electron acceptors, electron donors

Acetogens can use a wide variety of carbon sources and electron donors and acceptors. One-carbon compounds that M. thermoacetica and most acetogens can use for growth include H₂ + CO₂, CO, formate, methanol, and methyl groups from many methoxylated aromatic compounds. In addition, M. thermoacetica can grow on sugars, two-carbon compounds (glyoxylate, glycolate, and oxalate), lactate, pyruvate, and short-chain fatty acids. Besides CO₂, electron acceptors include nitrate, nitrite, thiosulfate, and dimethylsulfoxide. Information about the heterotrophic growth characteristics and electron donors/acceptors are described in more detail below.

B. The Wood-Ljungdahl pathway and the emergence and early evolution of life

Based on the patterns of ¹²C/¹³C isotopic fractionation, the sedimentary carbon record indicates the emergence of autotrophy soon after the earth became habitable, ca. 3.8 billion years ago [95]. The isotopic fractionation pattern of anaerobic organisms using the Wood-Ljungdahl pathway suggests that they may have been the first autotrophs, using inorganic compounds like CO and H₂ as an energy source and CO₂ as an electron acceptor approximately 1 billion years before O₂ appeared [96].

Acquisition of the core Wood-Ljungdahl genes would allow a wide variety of organisms, including archaea and a broad range of bacterial phyla (Firmicute, Chlorflexi, and Deltaproteobacteria), to fix CO and CO₂ by the reductive acetyl-CoA pathway. The simple strategy of the Wood-Ljungdahl pathway of successively joining two one-carbon compounds to make a two-carbon compound has been envisioned to be the earliest form of metabolism [97].

C. Methods used in elucidation of the pathway

As mentioned above, both ¹⁴C- and ¹³C-isotope labeling experiments were key in demonstrating that this pathway is a novel mechanism of CO₂ fixation. The ¹³C-labeling experiments involved mass spectroscopy using a tall column that Wood constructed in the stairwell of the biochemistry department at what was then known as Western Reserve University (“Case-” was appended in 1967). However, these pulse-labeling methods did not elucidate the intermediates in the Wood-Ljungdahl pathway. As one can see in Figure 1, the only products from ¹⁴CO₂ that would build up in solution are formate and acetate. Various cofactors are used in enzymes in the Wood-Ljungdahl pathway. These include H4folate, cobalamin, [4Fe-4S] clusters, and some unusual metal clusters, including two nickel-iron-sulfur clusters. Eventually, ¹⁴CH₃-H₄folate and ¹⁴CH₃-corrinoids were identified and shown to be incorporated into the methyl group of acetyl-CoA, but both of these methylated cofactors are bound to enzymes and do not build up in solution.

The hypothesis that H₄folate-dependent enzymes were involved was confirmed by isolation of the enzymes and demonstration that the reactions elucidated in other organisms indeed occur in M. thermoacetica. One of the unusual features of this pathway is that a number of intermediates (CH₃-Co, CH₃-Ni, Ni-CO, acetyl-Ni) are bound to transition metals. Thus, although paper chromatography was not appropriate for isolating intermediates, much more powerful methods that can focus directly on the active site metal centers were enlisted to characterize the properties of the cobalt, nickel, and iron-sulfur clusters in the CFeSP and CODH/ACS. The various oxidation states of these transition metals in the enzyme active sites have been established by electrochemical and spectroscopic methods (EPR, Mossbauer, UV-visible, Resonance Raman, EXAFS, XANES) and the rates of interconversion among the various intermediates have been studied by coupling spectroscopy to rapid mixing methods (stopped flow, chemical quench, rapid freeze quench). Density functional theory is also being used to test various mechanistic hypotheses. The existence of organometallic species has attracted a number of inorganic and bioinorganic chemists, who are developing models of the enzyme active sites to provide a better understanding of the roles of the metals in these intriguing reactions. This combination of approaches has spurred our understanding of the pathway and enriched our understanding of metal-based biochemistry.

D. Insights from the genome sequence

1. What it takes to be an acetogen

The genome of M. thermoacetica was recently sequenced and annotated. This first acetogenic genome to be sequenced is a 2.6 Megabase genome and 70 % of the genes have been assigned tentative functions [98]. Based on the 16S rRNA sequence, M. thermoacetica was classed as a Clostridium within the Thermoanaerobacteriaceae family [7]; after all, it was first known as Clostridium thermoaceticum. Homoacetogenic microbes are widely distributed among a few members of many phyla including Spirochaetes, Firmicutes (e.g., Clostridia), Chloroflexi, and Deltaproteobacteria, clearly indicating that acetogenesis is a metabolic, not a phylogenetic trait. Thus, homoacetogenesis is a rare occurrence within any phylum. For example, two close relatives of M. thermoacetica in the Thermoanaerobacteriaceae family, Thermoanaerobacter tengcongensis [99] and Thermoanaerobacter ethanolicus (draft sequence available at JGI) can grow on various sugars and starch [100], but the other two anaerobes lack acetyl coenzyme A synthase (ACS) and are not homoacetogenic. Among all the genes in the available Firmicutes, Chlorflexi, and Deltaproteobacteria genomic sequences, only acsC and acsD, which encode the two subunits of the CFeSP, co-occur with acetyl-CoA synthase (acsB) gene and are not present in sequences that lack acsB [98]. Surprisingly, acetogens do not seem to require a special set of electron transport-related genes, indicating that acetogens have co-opted electron transfer pathways used in other metabolic cycles.

2. Pathways for heme and cobalamin biosynthesis

In order to grow strictly autotrophically, an organism, of course, should not require any vitamins or cofactors. However, it is likely that in nature, some vitamins could be provided by crossfeeding. M. thermoacetica does require one vitamin, nicotinic acid [101]; however, it contains all the other genes needed to convert nicotinic acid to NAD [98]. This organism also has genes encoding both the de novo and salvage pathways for H₄folate synthesis and both the complete anaerobic branch of corrin ring synthesis and the common branch of the adenosyl-cobalamin pathway [98]. M. thermoacetica could be considered to be a cobalamin factory, generating over twenty different types of cobalamin or precursors that total 300 – 700 nmol per gram of cells [102, 103].

VI. Description of the Wood-Ljungdahl Pathway

Figure 1 shows the key reactions in the Wood-Ljungdahl pathway of CO₂ fixation described to consist of an Eastern (red coloring scheme) and a Western (blue) branch [104]. One molecule of CO₂ undergoes reduction by six electrons to a methyl group in the Eastern branch, while the Western branch involves reduction of the other CO₂ molecule to carbon monoxide, and condensation of the bound methyl group with CO and coenzyme A (CoA) to make acetyl-CoA. Acetyl-CoA is then either incorporated into cell carbon or converted to acetyl phosphate, whose phosphoryl group is transferred to ADP to generate ATP and acetate, the main growth product of acetogenic bacteria (for reviews see [26, 105, 106]).

A. The Eastern or Methyl Branch of the Wood-Ljungdahl Pathway

The genes encoding enzymes in the Eastern branch of the pathway are scattered around the M. thermoacetica genome. For example, the genes encoding 10-formyl-H₄folate synthetase, and the bifunctional 5,10-methenyl-H₄folate cyclohydrolase/5,10-methylene-H₄folate dehydrogenase are far from each other and are at least 300 genes away from the acs gene cluster.

1. Formate Dehydrogenase (Moth_2312-Moth_2314; EC 1.2.1.43)

The first reaction in the conversion of CO₂ to CH₃-H₄folate is the two-electron reduction of CO₂ to formate (Equation 14), which is catalyzed by a tungsten- and selenocysteine-containing formate dehydrogenase [107–109]. CO₂, rather than bicarbonate, is the substrate [110–112]. The first enzyme shown to contain tungsten [113], the M. thermoacetica FDH is an α₂β₂ enzyme (M_r = 340,000) that contains, per dimeric unit, one tungsten, one selenium, and ~ 18 irons and ~ 25 inorganic acid-labile sulfides in the form of iron-sulfur clusters [109]. The α subunit contains the selenocysteine residue, which is encoded by a stop codon in most contexts, and is responsible for binding the molybdopterin cofactor. The acetogenic formate dehydrogenase must catalyze a thermodynamically unfavorable reaction: the reduction of CO₂ to formate (E^o’ = −420 [114]) with electrons provided by NADPH, with a half cell potential for the NADP⁺/NADPH couple of −340 mV. The purified enzyme catalyzes the NADP⁺-dependent oxidation of formate with a specific activity of 1100 nmol min⁻¹mg ⁻¹ (k_cat = 6.2 s⁻¹). Another selenocysteine-containing formate dehydrogenase (Moth_2193) was located in the genome that has been proposed to be part of a formate hydrogen lyase system [98].

$$\text{NADPH}+{\text{CO}}_2+{\mathrm{H}}^{+}\rightleftharpoons {\text{NADP}}^{+}+\text{HCOOH}\mathrm{\Delta }{\mathrm{G}}^{{\mathrm{o}}_{\text{'}}}=+21.5\text{kJ}/\text{mol}[115]$$ (14)

In the M. thermoacetica FDH, the tungsten is found as a tungstopterin prosthetic group, like the molybdopterin found in the Mo-FDHs and in xanthine oxidase, sulfite oxidase, and nitrate reductase [116, 117]. The M. thermoacetica genome encodes a surprisingly large number of molybdopterin oxidoreductases in the dimethylsulfoxide reductase and xanthine dehydrogenase families [98]. Based on the original [118] and a reinterpreted [119] crystal structure, the mechanism for the Mo-Se-FDH involves formate displacing the Se-Cys residue as it binds to the oxidized Mo(VI) center. Then, two-electron transfer from the substrate to the Mo(VI) generates Mo(IV) and CO₂, which is released. Finally, the Se-Cys re-ligates to the Mo and two electrons are transferred through a [4Fe-4S] cluster to an external acceptor, reforming the oxidized enzyme for another round of catalysis. It is likely that the W-Se enzyme follows the same catalytic mechanism.

2. 10-Formyl-H₄folate synthetase (Moth_0109, EC. 6.3.4.3.)

In the reaction catalyzed by10-formyl-H₄folate synthetase, formate undergoes an ATP-dependent condensation with H₄folate, forming 10-formyl-H₄folate (Equation 15). The formyl-H₄folate product is then used in the biosynthesis of fMet-tRNAfMet (which is converted to N-formyl-methionine) and purines, or dehydrated and reduced by the succeeding H₄folate-dependent enzymes (cyclohydrolase and dehydrogenase) to generate 5,10-methylene-H₄folate, which is used in the biosynthesis of amino acids and pyrimidines [120] or acetate in acetogens.

$$\text{HCOOH}+\text{ATP}+{\mathrm{H}}_4\text{folate}\rightleftharpoons 10-\text{HCO}-{\mathrm{H}}_4\text{folate}+\text{ADP}+\text{Pi}\mathrm{\Delta }{\mathrm{G}}^{{\mathrm{o}}_{\text{'}}}=-8.4\text{kJ}/\text{mol[121]}$$ (15)

Like all bacterial formyl-H₄folate synthetases that have been studied, the M. thermoacetica enzyme is homotetrameric, with identical substrate binding sites on four 60 kDa subunits [120, 122]. This enzyme has been purified, characterized, and sequenced from a number of sources including M. thermoacetica [123–127], and the structure of the M. thermoacetica enzyme is known [128]. In higher organisms, this enzyme exists as one of the activities of a trifunctional C₁-H₄folate synthase (also containing a cyclohydrolase and a dehydrogenase) [129, 130], whereas the bacterial formyl-H₄folate synthetases are monofunctional and share similar properties [131].

The M. thermoacetica enzyme has been heterologously and actively expressed in E. coli [123]. Based on steady-state kinetic and equilibrium isotope exchange studies, the synthetase appears to follow a random sequential mechanism [125] involving a formyl-phosphate intermediate that suffers nucleophilic attack by the N¹⁰ group on H₄folate to form the product [132]. Formation of 10-formyl-H₄folate by the synthetase from M. thermoacetica occurs with a k_cat value of 1.4 s⁻¹ [133]. Monovalent cations activate the enzymes from M. thermoacetica and C. cylindrosporum by decreasing the K_m for formate to ~0.1 mM [126, 127].

3. 5,10-methenyl-H₄folate cyclohydrolase (EC 3.5.4.9.) and 5,10-methylene-H₄folate dehydrogenase (EC 1.5.1.5., NADP; EC 1.5.1.15,NAD) (Moth_1516)

The next two steps in the Ljungdahl-Wood pathway are catalyzed by 5,10-methenyl-H₄folate cyclohydrolase (Equation 16) and 5,10-methylene-H₄folate dehydrogenase (Equation 17). While the cyclohydrolase and dehydrogenase are part of a bifunctional protein in M. thermoacetica, they are monofunctional proteins in other acetogens, e.g., Clostridium formicoaceticum and Acetobacterium woodii [134–136]. As mentioned above, they are part of the trifunctional C₁-synthase in eukaryotes.

$$10-\text{formyl}-{\mathrm{H}}_4\text{folate}+{\mathrm{H}}^{+}\rightleftharpoons 5,10-\text{methenyl}-{\mathrm{H}}_4\text{folate}+{\mathrm{H}}_2\mathrm{O}\mathrm{\Delta }{\mathrm{G}}^{{\mathrm{o}}_{\text{'}}}=-35.3\text{kJ}/\text{mol}[121]$$ (16)

$$\begin{array}{l}\text{NAD(P)H}+5,10-\text{methenyl}-{\mathrm{H}}_4\text{folate}\rightleftharpoons 5,10-\text{methenyl}-{\mathrm{H}}_4\text{folate}+\text{NAD(P)}\mathrm{\Delta }{\mathrm{G}}^{{\mathrm{o}}_{\text{'}}}=-4.9\\ \text{kJ}/\text{mol}[121]\hfill \end{array}$$ (17)

The cyclohydrolase reaction strongly favors cyclization, with an equilibrium constant for 5,10-methenyl-H₄folate formation of 1.4 × 10⁶ M⁻¹ [137]. The reverse reaction (hydrolysis) occurs at a rate of ~ 250 s⁻¹ (35 °C, pH 7.2) for the C. formicoaceticum enzyme [136]. The NAD(P)H-dependent reduction of methenyl-H₄folate to form 5,10-methylene-H₄folate is catalyzed by 5,10-methylene-H₄folate dehydrogenase (Equation 17). The E^o' for the methenyl-/methylene-H₄folate redox couple is −295 mV vs. the standard hydrogen electrode (SHE) [121]; thus, reduction by NAD(P)H is thermodynamically favorable. Generally, microbes contain a bifunctional cyclohydrolase-dehydrogenase. The monofunctional dehydrogenase has so been identified in C. formicoaceticum [136], yeast [138], and A. woodii [135]. One rationale for the bifunctional enzyme may be to protect the highly labile methenyl-H₄folate from hydrolysis and channel it [139, 140] to the active site of the dehydrogenase for reduction to the relatively more stable methylene-H₄folate. There are both NAD⁺- and NADP⁺-dependent forms of this enzyme, yet the NADP-dependent enzyme is most common. A. woodii [135], Ehrlich ascites tumor cells [141], and yeast [142] contain the NAD⁺-dependent form. Crystal structures have been determined for the monofunctional NAD⁺-dependent yeast dehydrogenase [142], the bifunctional E. coli dehydrogenase/ cyclohydrolase [143], and the dehydrogenase/cyclohydrolase domains of the human trifunctional C₁ synthase [144].

The dehydrogenase reaction has been probed mostly in the direction of 5,10-methenyl-H₄folate formation, which is the reverse of the physiological reaction. The A. woodii dehydrogenase catalyzes both the forward and reverse reactions with k_cat values of ~1600 s⁻¹ [135]. With the M. thermoacetica dehydrogenase, oxidation of 5,10-methylene-H₄folate by NADP⁺ occurs according to a ternary complex mechanism with a specific activity of 360 s⁻¹ at 37 °C [¹⁴⁵] (720 s⁻¹ at 60 °C, D.W. Sherod, W.T. Shoaf, and L.G. Ljungdahl, unpublished).

4. 5,10-methylene-H₄folate reductase (Moth_1191, EC 1.1.99.15)

The last step in the eastern branch is catalyzed by 5,10-methylene-H₄folate reductase (Equation 18), which in acetogens is an oxygen-sensitive octomeric (αβ)₄ enzyme consisting of 35 and 26 kDa subunits [146, 147], while the Peptostreptococcus productus [148] and E. coli [149] enzymes are multimeric consisting of identical 35 kDa subunits. The mammalian enzyme is ~75 kDa, containing a C-terminal extension that binds the allosteric regulator S-adenosyl-L-methionine, which decreases the activity of the reductase by up to 50,000-fold [150]. Studies of the mammalian enzyme are of particular importance because the most common genetic cause of mild homocysteinemia in humans is an A222V polymorphism in the reductase [150, 151]. Among these characterized reductases, all contain FAD but only the acetogenic enzymes contain an iron-sulfur cluster [146, 147, 149]. The acetogenic enzyme uses reduced ferredoxin as an electron donor [146, 147], while the mammalian, E. coli, and P. productus enzymes use NAD(P)H. With the acetogenic proteins, pyridine nucleotides are ineffective electron carriers for the reaction in either direction [146]. The standard reduction potential for the methylene-H₄folate/CH₃-H₄folate couple is −130 mV vs. SHE [152]; thus, this reaction is quite exergonic with either NADH or ferredoxin as electron donor.

$$\begin{array}{l}2{\mathrm{e}}^{-}+{\mathrm{H}}^{+}5,10-\text{methenyl}-{\mathrm{H}}_4\text{folate}\rightleftharpoons 5-\text{methenyl}-{\mathrm{H}}_4\text{folate}\mathrm{\Delta }{\text{G}}^{{\mathrm{o}}_{\text{'}}}=-39.2\text{kJ}/\text{mol}\text{with}\\ \text{NAD(P)H}[121],-57.3\text{kJ}/\text{mol}\text{with}{\mathrm{H}}_2\text{as}\text{electron}\text{donor}\hfill \end{array}$$ (18)

The mammalian and E. coli enzymes have been more extensively studied than the acetogenic enzyme [150, 153]. The catalytic domain of all methylene-H₄folate reductases consists of an α₈β₈ TIM barrel that binds FAD in a novel fold and the Ala222 residue, whose polymorphic variant results in hyperhomocysteinemia is located near the bottom of the cavity carved out by the barrel [154].

The acetogenic reductase uses a ping-pong mechanism to catalyze the oxidation of CH₃-H₄folate with benzyl viologen with a k_cat of 300–350 s⁻¹ at its optimal temperature (35 °C and 55 °C for the C. formicoaceticum and M. thermoacetica enzymes, respectively) [146, 147]. The E. coli and pig-liver enzymes also use a ping-pong mechanism and exhibit a significantly lower value of k_cat than the acetogenic proteins [149, 155]. For the mammalian enzyme, the first half reaction, the stereospecific reduction of FAD by the pro-S hydrogen of NADPH to form FADH₂, is rate limiting during steady-state turnover [153, 155]. In the second half reaction, 5,10-methylene-H₄folate undergoes reduction by bound FADH₂ to form CH₃-H₄folate, as hydrogen is stereospecifically transferred to the more sterically accessible face of the pteridine and before undergoing exchange with solvent [153, 156]. It is proposed that during the reduction, the imidazolium ring of 5,10-methylene-H₄folate opens to form an iminium cation followed by tautomerization [157–159].

An unsolved question is how acetogens conserve energy by use of the Wood-Ljungdahl pathway. Thauer et al. [160] suggested that energy conservation could occur by linking an electron donor (e.g., H₂/hydrogenase, CO/CODH, or NADH dehydrogenase) to a membrane-associated electron transport chain, which would in turn donate electrons to methylene-H₄folate and the 5,10-methylene-H₄folate reductase. In this scenario, electron transport could result in the generation of a transmembrane proton potential coupled to ATP synthesis. In A. woodii, one of the steps involving methylene-H₄folate reductase, the methyltransferase or the CFeSP was shown to generate a sodium motive force across the membrane, which is linked to energy conservation [161]. However, since the reductase seems to be located in the cytoplasm in M. thermoacetica [146] and in other organisms, it appears that this is not the energy-conserving step in the Wood-Ljungdahl pathway [162]. On the other hand, the location of the 5,10-methylene-H₄folate reductase gene very near genes encoding a hydrogenase and heterodisulfide reductase (which are directly downstream from the acs gene cluster), suggests a possible mechanism of energy conservation and proton translocation involving the reductase and the membrane-associated hydrogenase and/or heterodisulfide reductase.

B. The Western or Carbonyl Branch of the Wood-Ljungdahl Pathway

1. Genes related to the Western branch of the pathway

While the genes encoding the Eastern branch of the Wood-Ljungdahl pathway are ubiquitous and dispersed on the genome, the Western branch genes are unique to organisms that use the Wood-Ljungdahl pathway and are co-localized in the acs gene cluster (Figure 7) [163]. Interestingly, their functional order in the pathway often matches the gene order: (1) carbon monoxide dehydrogenase (CODH), (2) acetyl-CoA synthase (ACS), (3) and (4) the two subunits of the corrinoid iron-sulfur protein (CFeSP), and (5) methyltransferase (MeTr). The acs gene cluster also includes a gene encoding an iron-sulfur protein of unknown function (orf7), and two genes, cooC and acsF, that are homologous to the Rhodospirillum rubrum cooC, which is required for nickel insertion into carbon monoxide dehydrogenase [164].

Figure 7.

Arrangement of Wood–Ljungdahl pathway genes in the acs gene cluster and in the chromosome. A. Arrangement of Wood–Ljungdahl pathway genes on the circular chromosome of M. thermoacetica. The numbering shows kilobase pairs from the origin of replication. B. The acs gene cluster that contains core Wood–Ljungdahl pathway genes discussed in the text. From [98].

Surprisingly, the M. thermoacetica genome sequence reveals an additional carbon monoxide dehydrogenase (Moth_1972), which is similar to an uncharacterized carbon monoxide dehydrogenase from Clostridium cellulolyticum and to CODH IV from Carboxydothermus hydrogenoformans [165], and is unlinked to the acs gene cluster.

2. Enzymology and bioinorganic chemistry of the Western branch of the pathway

a) Methyl-H₄folate:CFeSP Methyltransferase (MeTr) (Moth_1197,EC 2.1.1.X)

MeTr catalyzes the transfer of the methyl group of methyl-H₄folate to the cobalt center of the corrinoid iron-sulfur protein (CFeSP) (Equation 19). This reaction forms the first in a series of enzyme-bound bioorganometallic intermediates in the Wood-Ljungdahl pathway (methyl-Co, methyl-Ni, Ni-CO, acetyl-Ni), a strategy that is a novel feature of the Wood-Ljungdahl pathway. Equation 19 uses the term cobamide, instead of cobalamin, because the CFeSP contains methoxybenzimidazolylcobamide, not dimethylbenzimidazolylcobamide (cobalamin), although the enzyme is fully active with cobalamin [166]. This reaction is similar to the first step in the reaction mechanism of cobalamin-dependent methionine synthase [80, 167]. As noted in Equation 19, cobalt undergoes redox changes during the reaction and, as in other cobalamin-dependent methyltransferases [168], Co(I) is the only state that can catalyze methyl transfer, forming an alkyl-Co(III) product. The redox chemistry involving the CFeSP will be covered in the next section. Here we will focus on the structure and function of MeTr.

$$\text{Cob}(\mathrm{I})\text{amide}+\text{methyl}-{\mathrm{H}}_4\text{folate}\rightleftharpoons \text{methyl}-\text{Cob}(\text{III})\text{amide}+{\mathrm{H}}_4\text{folate}$$ (19)

While MeTr catalyzes methyl transfer from methyl-H₄folate, various cobalamin-dependent methyltransferases are found in biology, which react with a wide range of natural methyl donors, including methanol, methylamines, methyl thiols, and aromatic methyl ethers [168]. Structural and kinetic studies have been performed to attempt to understand how the methyltransferases catalyze the methyl transfer reaction. A key issue is that the methyl group requires activation, since the bond strengths of the C-O, C-N, or C- bonds in methanol, methylamines, and methyl thiols are quite large (356, 305, and 272 kJ/mol, respectively). As discussed in a recent review [168], there are at least three ways that MeTr enzymes activate the methyl donor: general acid catalyzed protonation of the N5 of pterins in methyl-H₄folate (and probably in methyltetrahydromethanopterin) through a H-bonding network, Lewis acid catalysis using a Zn active site near the cobalamin, and covalent catalysis using a novel amino acid (pyrrolysine). Protonation of the N-5 group of N5-methyl-H₄folate would lead to electrophilic activation of the methyl group. Thus, one of the goals of structural studies described below was to identify how substrates bind and undergo activation in the MeTr active site.

All of these methyltransferases bind the methyl donor within an α/β TIM barrel structure, as observed in the crystal structures of the methanol binding protein MtaB [169], the methyl-H₄folate binding domain of methionine synthase [170], the methylamine binding protein MtmB [171], and the methyl-H₄folate:CFeSP methyltransferase, shown in Figure 8 [172, 173]. In all of these proteins, the methyl donor binds within the cavity formed by the TIM barrel, as shown in (Figure 8B). The red surface indicates the negative charge in the cavity, which complements positive charges on methyl-H₄folate, which is tightly bound by hydrogen bonds involving residues D75, N96, and D160 that are conserved between MeTr and methionine synthase and have been referred to as the “pterin hook” [172]. Another conserved feature among the various MeTr structures is the lack of an obvious proton donor near the N-5 group. How does the methyl group then undergo activation?

Figure 8.

A. Structure of MeTr. Methyl-H₄folate and base-off cobalamin were modeled into the active site of MeTr, with the methyl group of methyl-H₄folate shown in green. From Fig. 5, [172]. B. The active site cavity of MeTr, based on the structure of the MeTr-methyl-H₄folate binding site. From Fig. 5, [173]. C. Hydrogen bonding network near the N-5 of methyl-H₄folate in the crystal structure of the binary complex of MeTr with methyl-H₄folate [173].

All of the cobalamin-dependent methyltransferases appear to use a similar principle to activate the methyl group that will be transferred: to donate positive charge to the heteroatom (O, N, or S) attached to the methyl group. This mechanism of electrophilic activation of the methyl group is supported by studies of the transfer of methyl groups in solution in inorganic models. Quaternary amines, with a full positive charge on nitrogen, do not require activation, as shown by studies of methyl transfer from a variety of quaternary ammonium salts to Co(I)-cobaloxime [174], trimethylphenylammonium cation to cob(I)alamin [175], and dimethylaniline at low pH to Co(I)-cobyrinate [176]. Furthermore, Lewis acids, such as Zn(II), can catalyze methyl transfer [177].

The mechanism of protonation of the N⁵ group of methyl-H₄folate, which leads to electrophilic activation of the methyl group, has been extensively studied in methionine synthase and the M. thermoacetica MeTr by transient kinetic and NMR measurements of proton uptake [178, 179], by measurements of the pH dependencies of the steady-state and transient reaction kinetics [167, 180], and by kinetic studies of variants that are compromised in acid-base catalysis [173]. These studies cumulatively show that general acid catalysis is involved in the methyl transfer reaction, but whether this occurs in the binary or ternary complex, or perhaps in the transition state for the methyl transfer reaction remains under discussion.

If protonation of the N⁵ group of methyl-H₄folate is key to the mechanism of methyl transfer, one would expect that the crystal structure would uncover the residue that acts as the general acid catalyst. However, in the crystal structures of the binary complexes of CH₃-H₄folate bound to methionine synthase [170] and the methyl-H₄folate:CFeSP methyltransferase [173], no obvious proton donor is within H-bonding distance of the N5 position of CH₃-H₄folate, and the only amino acid located near enough to N5 to participate in H-bonding is the side chain of Asn199, which is a conserved residue among all methyltransferases (Figure 8) [173]. Although this is an unlikely proton donor, upon binding CH₃-H₄folate, Asn199 swings by ~ 7 Å from a istant position into the H-bonded location shown in Figure 8C. Furthermore, the N199A variant has a mildly lower (~20-fold) affinity for CH₃-H₄folate, but a marked (20,000–40,000) effect on catalysis, suggesting that Asn199 plays an important role in stabilizing a transition state or high-energy intermediate for methyl transfer [173]. These experiments are consistent with the involvement of an extended H-bonding network in proton transfer to N5 of the folate that includes Asn199, a conserved Asp (Asp160), and a water molecule. Thus, although Asn199 is certainly not a chemically suitable proton donor, it becomes part of an extended H-bonding network that includes several water molecules that are also conserved in the methionine synthase structure.

The lack of a discernable proton donating residue is seen in a number of enzymes, including methionine synthase [170], dihydrofolate reductase [181], and purine nucleoside phosphorylase [182]. An extended H-bond network that includes water molecules, a Glu residue, and an Asn residue has been proposed to protonate the purine N7 in the transition state of the reaction catalyzed by purine nucleoside phosphorylase [183]. Thus, it has been speculated that the Asn residue in MeTr plays a key role in the transition state for methyl transfer and would slightly shift its position to allow both the carboxamide oxygen and nitrogen atoms to engage in H-bonding interactions with the pterin [173]. This dual H-bonding function could rationalize the placement of Asn at this key position in the methyltransferases, i.e., a typical general acid would be less apt in this bifurcated H-bonding stabilization of the transition state.

Most of the information related to the MeTr in this section dealt with binding and activation of its smaller substrate, the methyl group donor. We will now address the methyl acceptor, which is a cobamide that is tightly bound to an 88 kDa heterodimeric protein, the CFeSP.

b) Corrinoid iron-sulfur protein (CFeSP) (Moth_1198, Moth_1201,EC 2.1.1.X)

Twenty years after the discovery (described above) that cobalamin is involved in the pathway of anaerobic CO₂ fixation, Hu et al. partially purified a corrinoid protein that accepts the methyl group of CH₃-H₄folate to form a methylcorrrinoid species that, when incubated with CO, CoA, and a protein fraction containing CODH activity, is incorporated into the methyl group of acetyl-CoA [67] (Equation 20). When this 88 kDa protein was purified to homogeneity and characterized by various biophysical methods, it was found to contain an iron-sulfur cluster in addition to the corrinoid; therefore, it was named the corrinoid iron-sulfur protein (CFeSP) [68]. Besides this CFeSP protein involved in acetogenesis and its homolog in methanogens [184], the only other corrinoid proteins so far known to contain an iron-sulfur cluster are the dehalogenases, which catalyze the reductive removal of the halogen group in the process linked to growth of microbes on halogenated organics [185, 186]. The structure of the CFeSP (Figure 9A) is divided into three domains: the N-terminal domain that binds the [4Fe-4S] cluster, the middle TIM-barrel domain, and the C-terminal domain (also adopts a TIM barrel fold), which interacts with the small subunit to bind cobalamin [187]. The two TIM barrel domains of the CFeSP appear to be related to the MeTr and it has been proposed that the C-terminal domain undergoes conformational changes to alternatively bind MeTr/methyl-H₄folate and ACS (Figure 9B).

$$\text{Co}(\mathrm{I})-\text{CFeSP}+\text{methyl}-{\mathrm{H}}_4\text{folate}\rightleftharpoons \text{methyl}-\text{Co}(\text{III})-\text{CFeSP}+{\mathrm{H}}_4\text{folate}$$ (20)

$$\text{methyl}-\text{Co}(\text{III})-\text{CFeSP}+\text{Ni}-\text{ACS}\rightleftharpoons \text{Co}(\mathrm{I})-\text{CFeSP}+\text{methyl}-\text{Ni}-\text{ACS}$$ (21)

.

Figure 9.

Structure of the CFeSP. (A) From Figure 1 of [187], AcsC and AcsD are the large and small subunits, respectively, and Cba is the corrinoid cofactor. This figure shows the middle and C-terminal domains of the large subunit, while the N-terminal domain that contains the [4Fe-4S] cluster was disordered and could not be modeled. An expanded view of the [4Fe4S] cluster is shown in the upper right hand corner. (B) Modified from Fig. 3 of [187], proposed conformational changes of the CFeSP during methyl transfer from methyl-H₄folate/MeTr to the corrinoid (MT-Conform 1) to a resting state (Cap-Conform shown in Fig. A), and then from methyl-corrinoid to the A cluster on ACS (MT-Conform 2). The reductive activation conformation is also depicted (Red-Conform).

Cobalt in cobalamin can access the (I), (II) and (III) redox states. As shown in Equation 20, the Co center cycles between the Co(I) and methyl-Co(III) states during the reaction, with the Co(I) state being required to initiate catalysis, as established by stopped-flow studies with the CFeSP [180] and with methionine synthase [188]. In the Co(I) state, cobalt has a d⁸ configuration. In B₁₂ and related corrinoids, Co(I) is a supernucleophile [189, 190] and is weakly basic, with a pKa below 1 for the Co(I)-H complex [191]. Protein-bound Co(I) is also highly reducing with a standard reduction potential for the Co(II)/(I) couple below −500 mV [192–194]. This high level of reactivity comes with a price - Co(I) can easily undergo oxidative inactivation to the Co(II) state; for example, in MeTr and in methionine synthase, the Co(I) center succumbs to the 2+ state once in every 100–2,000 turnovers [195–197].

Resuscitation of inactive MeTr requires reductive activation, which occurs by the transfer of one electron from the [4Fe-4S] cluster within the large subunit (product of the acsC gene) of the CFeSP [197]. This cluster has a reduction potential of −523 mV, which is slightly more negative than that of the Co(II)/Co(I) couple of the cobamide (−504 mV) in the CFeSP, making reductive activation a thermodynamically favorable electron-transfer reaction [193]. The [4Fe-4S] cluster can accept electrons from a low-potential ferredoxin as well as directly from CO/CODH, H₂/hydrogenase, or pyruvate/pyruvate ferredoxin oxidoreductase [197]. It has been demonstrated that the cluster is involved in reductive activation, but does not participate directly in the methyl transfer from methyl-H₄folate or to the Ni center in ACS [198, 199]. Unlike the CFeSP, in methionine synthase, the electron donors (methionine synthase reductase in mammals [200], flavodoxin in E. coli [188]) are flavins, with a higher redox potential that that of the Co(II)/(I) couple; therefore, a reductive methylation system involving S-adenosyl-L-methionine is used to drive the uphill reactivation [192].

Although the redox potential of −504 mV for the Co(II)/Co(I) couple of the cobamide in the CFeSP is fairly low, it is well above the standard potential for the same couple in cobalamin in solution (~-610 mV) [201]. It appears that control of the coordination state of the Co(II) center plays a key role in increasing the redox potential into a range that would make the reduction feasible for biologically relevant electron donors. Instead of having a strong donor ligand like imidazole or benzimidazole, which is found in most other corrinoid proteins, the Co center in the CFeSP is base-off and the only axial ligand is a weakly coordinated water molecule [187, 202]. Interestingly, removing the dimethylbenzimidazole ligand is an intermediate step in the reductive activation of the cobalt center in methionine synthase [203] and in the electrochemical reduction of Co(II) to the Co(I) state of B₁₂ in solution [201]. Thus, the CFeSP appears to have evolved a mechanism to facilitate reductive activation that can be explained by some well-studied electrochemical principles.

Formation of the first organometallic intermediate (methyl-Co) in the Wood-Ljungdahl pathway precedes methyl group transfer from the methylated CFeSP to a NiFeS cluster in acetyl-CoA synthase (ACS) (Equation 21). This reaction and the associated carbonylation and methyl migration to form an acetyl-ACS intermediate and then thiolysis to form acetyl-CoA are the subject of the next section.

c) CO dehydrogenase/acetyl-CoA synthase (CODH/ACS)

The stepwise formation of a series of organometallic intermediates (methyl-Co, Ni-CO, methyl-Ni, acetyl-Ni) is one of the novel features of the Wood-Ljungdahl pathway. The key enzyme in this pathway, CODH/ACS, has been recently reviewed [204–206]. Figure 1 shows that when acetogens grow on H₂ + CO₂, one molecule of CO₂ is reduced to CO by CODH, which becomes the carbonyl group of acetyl-CoA, and another CO₂ is reduced to formate, which serves as the precursor of the methyl group of acetyl-CoA. Under heterotrophic growth conditions, e.g., sugars, CO₂ and electrons are generated from the decarboxylation of pyruvate by PFOR (see below). When CO is the growth substrate, one molecule of CO must be converted to CO₂, which is then reduced to formate for conversion to the methyl group of acetyl-CoA, while another molecule of CO can be incorporated directly into the carbonyl group.

Both CO and CO₂ are unreactive without a catalyst, but the enzyme-catalyzed reactions involving these one-carbon substrates are fast, with turnover numbers as high as 40,000 s⁻¹ at 70 °C reported for CO oxidation by the Ni-CODH from Carboxydothermus hydrogenoformans [207]. Ni-CODHs are classified into two groups: the monofunctional enzyme that functions physiologically in catalyzing CO oxidation (Equation 10), allowing microbes to take up and oxidize CO at the low levels found in the environment, while CODH in the bifunctional protein functions to generate CO by the reverse of Eq. 10 and to couple to ACS to generate acetyl-CoA (Equation 11).

(1) CO dehydrogenase structure and function

Regardless of whether the Ni-CODH is monofunctional or part of the CODH/ACS bifunctional CODH/ACS complex, the CODH components are homodimeric enzymes that contain five metal clusters, including two C-clusters, 2 B-clusters, and a bridging D-cluster as diagrammed in Figure 10. The B-cluster is a typical [4Fe-4S]^2+/1+ cluster, while the D-cluster is a [4Fe-4S]^2+/1+ cluster that bridges the two identical subunits, similar to the [4Fe-4S]^2+/1+ cluster in the iron protein of nitrogenase. Spectroscopic studies and metal analyses identified four of the five metal clusters present in CODH [91, 92, 208–210], but did not identify the D-cluster and were unable to predict the correct arrangement of metals at the catalytic site for CO oxidation, the C-cluster. The entire complement of metal clusters and their locations were provided by the crystal structures of the monofunctional Ni-CODHs [211, 212]. A nearly superimposable structure including identical metal clusters were found for the CODH component of the bifunctional CODH/ACS [213, 214]. Buried 18 Å below the protein surface, as shown in Figure 11, the C-cluster can be described as a [3Fe-4S] cluster that is bridged to a binuclear NiFe site, in which the iron in the binuclear cluster is bridged to a His ligand. Because of its redox state, this iron has been called Ferrous Component II (FCII). The C. hydrogenoformans C-cluster may have a sulfide bridge between Ni and FCII [215, 216]; however, recent studies may have ruled out the catalytic relevance of this bridge [217, 218]. Furthermore, there is evidence for a catalytically important persulfide at the C-cluster [219].

Figure 10.

From Figure 1 of [87].The two subunits are shown with light and dark wire tracings. Electrons generated by the oxidation of CO at the C and C’ clusters are transferred to the internal redox chain in CODH, consisting of the B (and B’) and D clusters. The D cluster, located at the interface between the two subunits, is proposed to transfer electrons to the electron transfer protein (CooF), which is coupled to hydrogenase.

Figure 11.

A. The C-cluster of CODH and (B) the A-cluster of ACS. From [204].

As shown in Figure 12, the C-cluster can exist in four redox states (C_ox, C_red1, C_int, and C_red2), which differ by one electron. C_ox is an inactive state and can undergo reductive activation to the C_red1 form [220, 221]. Reaction of the active C_red1 form of CODH with CO, generates the C_red2 state [84], which is considered to be two electrons more reduced than C_red1.

Figure 12.

Redox states of the CODH catalytic center, the C-cluster. From Figure 2 of [335].

The CODH reaction mechanism has been reviewed [204]. Following a ping-pong mechanism, CODH is reduced by CO in the “Ping” step and the reduced enzyme transfers electrons to an external redox mediator like ferredoxin (or a specialized ferredoxin-like protein, CooF, in R. rubrum) in the “Pong” step. CooF then couples to a membrane hydrogenase as shown in Fig. 10, or to other energy requiring cellular processes. Recent direct electrochemical studies of CODH linked to a pyrolytic graphite electrode show complete reversibility of CO oxidation and CO₂ reduction without any overpotential; in fact, at low pH values, the rate of CO₂ reduction exceeds that of CO oxidation [221].

The CODH mechanism described in Figure 13 was derived from NMR studies that will be described below [222] and from crystallographic, kinetic (steady-state and transient), and spectroscopic studies (reviewed in [87]). The reaction steps are analogous to those of the water-gas shift reaction in that the mechanism includes a metal-bound carbonyl, a metal bound hydroxide ion, and a metal-carboxylate, which is formed by attack of the M-OH on M-CO. Elimination of CO₂ either leaves a metal-hydride or a two-electron-reduced metal center and a proton. CODHs have a weak CO-dependent hydrogen evolution activity that might suggest a metal-hydride intermediate [223–225]. However, a key difference between the water-gas shift and the enzymatic reaction is that H₂ is the product of the nonenzymatic reaction, whereas protons and electrons are the product of the CODH reaction. This indicates that in the enzyme, electron transfer is very rapid relative to H₂ evolution, perhaps because of the placement of the B and D clusters as electron acceptors, like a wire, between the C-cluster and the site at which external electron carriers bind. Furthermore, the enzyme uses distinct pathways for delivery/egress of protons and electrons.

Figure 13.

Proposed mechanism of CO oxidation at the C-cluster of CODH, from Figure 4 of [222]. B₁ and B₂ designate active site bases. Recent information about early stages in CO oxidation was gleaned through NMR studies of a rapid reaction involving the interconversion between CO and bound CO₂ [222]. See the text for a detailed explanation of each step.

The sequential steps in the CODH mechanism have been discussed in a recent review [74], so here we will only summarize earlier studies and focus on work published since 2004. Not shown in Figure 13 are two proposed channels that connect the C-cluster to substrates in the solvent: a hydrophobic channel for CO and a hydrophilic water channel [212]. NMR experiments indicate that movement of gas molecules through the hydrophobic channel is very rapid, i.e., ~33,500 s⁻¹ at 20 °C [222], which would extrapolate to 10⁶ s⁻¹, if the rate doubles with every 10 °C rise in temperature. In fact, migration of CO to the C-cluster occurs at diffusion-controlled rates even at 20 °C (3.3 × 10⁹ M-¹ s⁻¹ at the published K_d of 10 µM).

As shown in Step 1, CO appears to bind to Ni, based on crystallographic [211, 212], FTIR [226], and X-ray absorption studies of cyanide (a competitive inhibitor) binding to CODH [227]. However, the M-CO has not yet been observed in any crystal structure, and there is also evidence that CN⁻ binds to iron [228]. It is likely that water (or hydroxide) binds to the special iron (ferrous component II, above), based on ENDOR studies [229], which would lower the pKa for water and facilitate formation of an active hydroxide, as described for other enzymes [230]. A recent crystal structure shows a bridging OH⁻ in a sample prepared at mild redox potentials [227], so perhaps the OH⁻ can equilibrate between bridging and Fe-bound states. The nonbridging hydroxide would be expected to be the most active nucleophilic state for attack on the Ni-CO.

Following substrate binding in this proposed bimetallic mechanism, the bound hydroxide would be poised to rapidly attack the Ni-CO to generate a bound carboxyl group, which is shown in Figure 13 to be bridged between Ni and Fe, based on the recent structure of bound CO₂ in the C. hydrogenoformans CODH [218]. The intermediacy of a bound CO₂ intermediate is consistent with NMR experiments described in the next paragraph and with FTIR studies in which IR bands assigned to M-CO disappear as bands for metal-carboxylates (1724 and 1741 cm⁻¹) and CO₂ (2278 cm⁻¹ in the ¹³CO sample) appear [226].

The first two steps in the CODH reaction were followed by NMR experiments with the monofunctional C. hydrogenoformans enzyme [222]. When CODH is incubated with ¹³CO, the¹³CO linewidth markedly increases. This linewidth broadening was concluded to arise from a chemical exchange between solution ¹³CO and a bound form of ¹³CO₂ (presumably the bridged form shown in Figure 13). Thus, the ¹³CO exchange broadening mechanism involves steps 1 and 2 of Figure 13, and, although it is in the slow exchange regime of the NMR experiment, occurs with a rate constant (1080 s⁻¹ at 20 °C) that is slightly faster than the rate of CO oxidation at 20 °C. The ¹³CO exchange broadening is pH independent, unlike the overall CO oxidation reaction, which has a well defined pKa value of 6.7, suggesting the presence of an internal proton reservoir that equilibrates with solvent more slowly than the rate of CO exchange with bound CO₂ [222]. A Lys and several His residues were earlier suggested as acid-base catalysts [211, 219], and might account for the internal proton reservoir, depicted as B1 and B2 in Figure 13.

Since a ping-pong mechanism requires that electron acceptors bind only after CO₂ is released, presumably the protein is in a two-electron reduced state when CO₂ is bound. Then in slightly slower reactions, CO₂ is proposed to dissociate (step 3) and protons and electrons are transferred to solvent (steps 4, 5).

(2) The final steps in acetyl-CoA synthesis catalyzed by CODH/ACS

By the principle of microreversibility, CO₂ reduction should occur by a direct reverse of the steps shown in Figure 13. This reverse reaction generates CO as a key one-carbon intermediate in the Wood-Ljungdahl pathway [86]. Thus, as shown in Figure 1, the bifunctional CODH/ACS, encoded by the acsA/acsB genes, is a machine that converts CO₂, CoA, and the methyl group of the methylated CFeSP to acetyl-CoA, which is a precursor of cellular material (protein, DNA, etc.) and a source of energy.

Acetyl-CoA synthesis is catalyzed by ACS at the A-cluster, which consists of a [4Fe-4S] cluster that is bridged via a cysteine residue (Cys509 in M. thermoacetica) to a Ni site (called the proximal Ni, Ni_p) that is linked to the distal Ni ion (Ni_d) in a thiolato- and carboxamido-type N₂S₂ coordination environment [213, 214, 231], as shown in Figure 11B.

One of the proposed mechanisms of acetyl-CoA synthesis is shown in Figure 14. Two competing mechanisms for acetyl-CoA synthesis have been proposed, which differ mainly in the electronic structure of the intermediates: one proposes a paramagnetic Ni(I)-CO species as a central intermediate [74] and the other (the “diamagnetic mechanism”) proposes a Ni(0) or, as recently proposed, a spin-coupled [4Fe-4S]¹⁺-Ni_p¹⁺ intermediate [88, 232]. In this review, a generic mechanism is described that emphasizes the organometallic nature of this reaction sequence.

Figure 14.

(A) Proposed mechanism of acetyl-CoA synthesis at the A-cluster of ACS, from Figure 3 of [206]. See the text for details.

Before the first step in the ACS mechanism, CO migrates from its site of synthesis at the C-cluster of CODH to the A-cluster of ACS through a 70 Å channel [213, 233–235]. By determining the structure of CODH crystals incubated with high pressures of Xe, which has a molecular size that approximates that of CO, a series of hydrophobic gas binding pockets have been located within this tunnel [235]. The residues that make up these cavities are conserved among the CODH/ACSs in their hydrophobic character, but not in sequence identity. One of the Xe binding sites is within 4 Å of the binickel center and likely represents a portal near the A-luster for CO, just prior to its ligation to the metal center, shown in Figure 14, reminiscent of the “inland lake”, which has been suggested as the entry point for O₂ into the active site of the copper-containing amine oxidase [236, 237].

Although Figure 14 simplistically shows an ordered mechanism, recent pulse-chase studies demonstrate that binding of CO and the methyl group in Steps 1 and 2 occurs randomly as shown in Figure 15 [238]. As mentioned above, after CO travels through the intersubunit channel, it binds to the Ni_p site in the A-cluster to form an organometallic complex, which according to the paramagnetic mechanism, is the so-called “NiFeC species” that has been characterized by a number of spectroscopic and kinetic approaches [74]. The electronic structure of the NiFeC species is described as a [4Fe-4S]²⁺ cluster linked to a Ni¹⁺ center at the Ni_p site, while Ni_d apparently remains redox-inert in the Ni²⁺ state [89]. Another view is that the NiFeC species is not a true catalytic intermediate in acetyl-CoA synthesis, but is an inhibited state, and that a Ni(0) state is the catalytically relevant one [88, 232], as recently discussed [204]. Gencic and Grahame recently suggested a spin-coupled [4Fe-4S]¹⁺ cluster linked to a Ni¹⁺ site as the active intermediate [239], although no spectroscopic studies were reported in the paper describing this proposal. Evidence for a such a spin-coupled state was recently provided by Mossbauer spectroscopy, and Tan et al. suggest that the methyl group of the methylated CFeSP could be transferred as a methyl cation (see the CFeSP section above) to the spin-coupled Ni_p ¹⁺-[4Fe-4S]¹⁺ center to form a CH₃-Ni_p²⁺-[4Fe-4S]²⁺ state [240]. One possibility is that the paramagnetic state and the spin-coupled state are in equilibrium and that the paramagnetic state is favored with CO as the first substrate, while the spin-coupled state is favored when the methyl group binds first. If so, a redox shuttle is required since the spin-coupled cluster is one-electron more reduced than the NiFeC species.

Figure 15.

Random mechanism of acetyl-CoA synthesis, from Figure 3 of [238]. See the text for details.

Following transfer of the methyl group of the methylated CFeSP apparently to the Ni_p site of the A-cluster [241–244], Step 3 involves carbon-carbon bond formation by condensation of the methyl and carbonyl groups to form an acetyl-metal species. Experimental evidence for an acetyl-enzyme intermediate was proposed because an exchange reaction between CoA and acetyl-CoA, which involves C-S bond cleavage and resynthesis, was found to occur significantly faster than another exchange reaction between CO and the carbonyl group of acetyl-CoA, which involves cleavage and resynthesis of both C-C and C-S bonds [245–247]. Thus, many cycles of cleavage and re-synthesis of the C-S bond of acetyl-CoA can occur for each cycle of C-C bond synthesis/cleavage. Further evidence for an acetyl-ACS intermediate was recently provided by burst kinetic studies [239].

In the last step, CoA binds to ACS triggering thiolysis of the acetyl-metal bond to form the C-S bond of acetyl-CoA, completing the reactions of the Western branch of the Wood-Ljungdahl pathway. The CoA binding site appears to be within the second domain of ACS, which includes residues 313–478, based on evidence that Trp418 undergoes fluorescence quenching on binding CoA [248]. Furthermore, this domain contains six Arg residues near Trp418, and CoA binding is inhibited by phenylglyoxal, which is a rather specific Arg modification reagent [83].

C. Phosphotransacetylase (E.C. 2.3.1.8) and acetate kinase (E.C.2.7.2.1.)

Phosphotransacetylase catalyzes the conversion of acetyl-CoA to acetyl-phosphate. Although the M. thermoacetica phosphotransacetylase has been purified and characterized enzymatically [249], the gene encoding this protein could not be easily annotated because the protein sequence has not been determined and, when the genome sequence was first determined, no candidates were found that were homologous to known phosphotransacetylase genes. Fortuitously, a novel phosphotransacetylase (PduL) involved in 1,2-propanediol degradation was recently identified in Salmonella enterica [250] that is homologous to two M. thermoacetica genes (Moth_1181 and Moth_0864). Thus, one of these genes is likely to encode the M. thermoacetica phosphotransacetylase that is involved in the Wood-Ljungdahl pathway. Located only ten genes downstream from the gene encoding methylene-H₄folate reductase, Moth_1181 is a likely candidate. However, the phosphotransacetylase associated with the Wood-Ljungdahl pathway needs to be unambiguously identified by sequencing the active protein, because phosphotransacetylase is a member of a rather large family of CoA transferases.

Conversion of of acetyl-phosphate to acetate is catalyzed by acetate kinase, which is encoded by the Moth_0940 gene. This gene is distant from, and thus is not linked to Moth_1181, Moth_0864, or the acs gene cluster.

D. The Secret to Metabolic Diversity in Acetogens: Coupling of the Wood-Ljungdahl to Other Pathways

1. Pyruvate ferredoxin oxidoreductase (PFOR) (EC1.2.7.1,Moth_0064)

Pyruvate:ferredoxin oxidoreductase (PFOR) catalyzes the oxidative cleavage of pyruvate and attachment of the two-carbon fragment (carbons 2 and 3) to CoA, forming acetyl-CoA, reducing ferredoxin, and eliminating CO₂, (Equation 22) [251–256]. A fairly recent review on PFOR is available [257]. There are five enzymatic activities that can oxidize pyruvate. While PFOR transfers its electrons to a low-potential reductant (ferredoxin or flavodoxin), pyruvate dehydrogenase reduces NAD to NADH (also generating acetyl-CoA), and pyruvate oxidase transfers its electrons to oxygen to make hydrogen peroxide and acetyl-phosphate. Pyruvate decarboxylase, which is a key enzyme in ethanol fermentation, retains its electrons in the substrate to generate acetaldehyde and pyruvate formate lyase transfers the electrons to the carboxyl group to generate formate and acetyl-CoA. Among these five pyruvate metabolizing enzymes, only pyruvate formate lyase does not use thiamine pyrophosphate (TPP).

PFORs have been isolated from several bacteria and the enzymes from M. thermoacetica [69, 86], D. africanus [258–260] have been best characterized. In addition, the enzymes from Pyrococcus [261–263] and from the methanogenic archaea, Methanosarcina barkeri [264] and Methanobacterium thermoautotrophicum [265] have been studied. PFOR plays an important role in the oxidation of acetogens since it catalyzes the first step in the metabolism of pyruvate, supplied as a growth substrate. PFOR is also essential in sugar metabolism, since it couples the Embden-Meyerhof-Parnas pathway to the Wood-Ljungdahl pathway. Furthermore, for anaerobes like methanogens and acetogens that fix CO₂ by the Wood-Ljungdahl pathway, PFOR is also a pyruvate synthase that catalyzes the conversion of acetyl-CoA to pyruvate, the first step in the incomplete reductive tricarboxylic acid cycle (TCA, see below) [30, 266].

PFOR is an ancient molecule that apparently predated the divergence of the three kingdoms of life (prokarya, archaea, and eukarya). All archaea appear to contain PFOR and it is widely distributed among bacteria, and even some anaerobic protozoa like Giardia [267]. A heterotetrameric enzyme, like the archaeal PFORs, has been proposed to be a common ancestor [254, 268] that underwent gene rearrangement and fusion to account for the hetero- and homodimeric enzymes. These are represented by COG0674, COG1013, and COG1014, which are found as separate alpha, beta, and gamma subunits in some organisms [98]. The ancestral δ subunit is an 8 Fe ferredoxin-like domain (COG1014) that binds two [4Fe-4S] clusters, while the beta subunit (COG1013) binds the essential cofactor TPP and an iron-sulfur cluster. The PFOR that has been purified from M. thermoacetica is a homodimer (a fusion of all four COGs), with a subunit mass of 120 kDa [269]. Five other sets of genes belonging to the same COGs as PFOR could encode authentic pyruvate:ferredoxin oxidoreductases, but they could also encode other proteins in the oxoglutamate oxidoreductase family, like 2-ketoisovalerate oxidoreductase, indolepyruvate oxidoreductase, and 2-ketoglutarate oxidoreductase.

$$\text{pyruvate}+\text{CoA}+\text{Fd}(\text{ox})\to {\text{CO}}_2+\text{acetyl}-\text{CoA}+\text{Fd}(\text{red})$$ (22)

The PFOR mechanism is summarized in Figure 16. As with all TPP-dependent enzymes, PFOR forms an "active acetaldehyde" intermediate, as first proposed by Breslow in 1957 [270], which involves the formation of a hydroxyethyl adduct between the substrate and the C-2 of the thiazolium of TPP. This is followed by one-electron transfer to one of the three [4Fe-4S] clusters to generate a radical intermediate. The radical intermediate was first identified in the early 1980’s, and was considered to be a pi radical with spin density delocalized over the aromatic thiazolium ring as shown in Figure 16 [252, 271]. However, based on the crystal structure, this intermediate was proposed to be a novel sigma-type acetyl radical [272]. However, spectroscopic studies recently provided unambiguous evidence that negated the sigma radical formulation and provided strong support for the pi radical model [273, 274]. The [4Fe-4S] cluster to which the HE-TPP radical is coupled also was identified by recent spectroscopic studies [273, 274]. Rapid freeze quench EPR and stopped flow studies demonstrated that this radical forms and decays significantly faster than the k_cat value of the steady-state reaction (5 s⁻¹ at 10°C for the M. thermoacetica enzyme), demonstrating its catalytic competence as an intermediate in the PFOR reaction mechanism [275, 276]. One of the key remaining questions in the PFOR mechanism is how CoA stimulates decay of the HE-TPP radical intermediate by at least 100,000-fold [276]. Several hypotheses have been proposed [257, 276].

Figure 16.

PFOR Mechanism showing the ylide nucleophile and the HE-TPP anionic and radical intermediates.

Perhaps one reason that PFOR, instead of pyruvate dehydrogenase, is used in acetogens and other anaerobes is that PFOR uses low-potential electron-transfer proteins like ferredoxin and flavodoxin, which is likely to make the pyruvate synthase reaction feasible as the first step for converting acetyl-CoA into cell material. The requisite electron donor for the PDH complex, NADH, is much too weak an electron source to reduce acetyl-CoA to pyruvate (the NAD/NADH half reaction is 200 mV more positive than the acetyl-CoA/pyruvate couple). However, one should consider the possibility that even the ferredoxin-coupled PFOR reaction requires some type of additional driving force for the energetically uphill synthesis of pyruvate, as suggested for some other systems. In methanogens, the pyruvate synthase reaction appears to be linked to H₂ oxidation by the membrane-associated Ech hydrogenase and to require reverse electron transfer [277]. The M. barkeri enzyme was shown to catalyze the oxidative decarboxylation of pyruvate to acetyl-CoA and the reductive carboxylation of acetyl-CoA with ferredoxin as an electron carrier [278]. Yoon et al. have also studied the pyruvate synthase reaction of PFOR from Chlorobium tepidum [279], which, like the methanogens, links to the incomplete TCA cycle to convert oxaloacetate (derived from pyruvate by the action of pyruvate carboxylase or the linked activities of phosphoenolpyruvate (PEP) synthetase and PEP carboxylase) into malate, fumarate, succinate, succinyl-CoA, and alpha-ketoglutarate [266].

2. Incomplete TCA cycle

The tricarboxylic acid (TCA) cycle is used oxidatively by aerobic organisms to generate reducing equivalents that eventually couple to the respiratory chain to generate ATP. Many autotrophic anaerobes use the TCA cycle in the reductive direction to incorporate the acetyl group of acetyl-CoA into cell carbon and to generate metabolic intermediates. The reductive TCA cycle is also used for autotrophic growth by some green sulfur bacteria [280] and Epsilonproteobacteria [281]. The reductive TCA cycle has three enzymes that are distinct from the oxidative pathway enzymes: 2-oxoglutarate:ferredoxin oxidoreductase, fumarate reductase, and ATP citrate lyase [280, 282]. In some anaerobes, like acetogens, the reverse TCA cycle is incomplete (Figure 17). The incomplete TCA cycle is used in both the oxidative and reductive directions to generate metabolic intermediates [283, 284], like α-ketoglutarate and oxaloacetate for amino acid synthesis.

Figure 17.

Incomplete TCA cycle allowing conversion of acetyl-CoA to cellular intermediates. Dashed arrows represent enzymes that not identified in the M. thermoacetica genome. From [98].

As shown in Figure 17, pyruvate:ferredoxin oxidoreductase can catalyze the formation of pyruvate from acetyl-CoA and CO₂ (see secton IV D1). Generally in the reductive TCA cycle, pyruvate is converted to phosphoenolpyruvate, which then undergoes carboxylation to generate oxaloacetate, which is reduced to fumarate. However, since homologs of phosphoenolpyruvate carboxylase are not found in the M. thermoacetica genome, M. thermoacetica may be able to convert pyruvate to malate, using a malate dehydrogenase homologous to the NADP-dependent malic enzyme of E. coli (encoded by maeB). Although this enzyme generally functions in the malate decarboxylation direction, the pyruvate carboxylation reaction is thermodynamically favorable [25], and expression of the other E. coli malate dehydrogenase (encoded by maeA) has been shown to allow malate formation from pyruvate in a pyruvate-accumulating E. coli strain [285]. The genome of M. thermoacetica appears to lack any homolog of known fumarate reductases; therefore, either M. thermoacetica lacks this enzyme, or it is encoded by a novel gene. On the other hand, the closely related acetogens Clostridium formicoaceticum and Clostridium aceticum can reduce fumarate to succinate [286].

The pathway from succinate to citrate appears to be present in M. thermoacetica. The presence of several sets of genes predicted to encode 2-oxoacid:ferredoxin oxidoreductases [98] and of a likely citrate lyase gene open the possibility that the complete reductive TCA cycle can be used by M. thermoacetica, or that M. thermoacetica has lost this ability during its evolution, with loss of fumarate reductase. M. thermoacetica also encodes a putative citrate synthase, which could allow generation of compounds from citrate to succinate in the oxidative direction.

3. Various cobalamin-dependent methyltransferases allow growth on diverse methyl group donors

Acetogenic bacteria are able to grow on a variety of methyl donors (methanol, aromatic O-methyl ethers, and aromatic O-methyl esters) by coupling different methyltransferase systems to the Wood-Ljungdahl pathway, as shown in Figure 18. The cobalamin dependent methyltranferases have recently been reviewed [80, 168]. Each of the acetogenic methyltransferase systems conforms to a three-module arrangement for catalysis plus a separate module for reductive activation. The three catalytic components comprise a binding site for the methyl donor, a corrinoid binding module, and a module for binding the methyl acceptor. The methyl group is transferred from the donor to the corrinoid, generating an intermediate methyl-Co species, from which the methyl group is transferred to methyl acceptor. The reductive activation module comes into play when the Co center undergoes oxidation. Because the Co^2+/1+ couple has such a low redox potential, reductive activation often requires energy input in the form of ATP or adenosylmethionine (as in methionine synthase). In the Wood-Ljungdahl pathway, the methyl donor module is the methyltransferase that binds methyl-H₄folate, the corrinoid binding module is the CFeSP, and the A-cluster in ACS is the methyl group acceptor module. The reductive activation module for the CFeSP, as described above, does not require extra energy input (i.e., ATP, etc.) because the one-electron reductant is a low-potential [4Fe-4S cluster] within the large subunit of the CFeSP. It is likely that the modular structural arrangement of methyl donors and acceptors allows for diverse metabolism of methyl groups, with different metabolic modules interfacing at the level of methyl-H₄folate.

Figure 18.

Coupling of various methyl donors to the Wood-Ljungdahl pathway. The methyltransferase systems involved in transferring the methyl groups from aromatic methyl ethers (Mtv) or methanol (Mta) to methyl-H₄folate have been identified by genomic and enzymatic studies. See the text for details. Ar-O-CH₃ designates the aromatic methyl ether that serves as the methyl donor, and Ar-OH signifies the alcohol, which is the demethylation product.

Many of the aromatic compounds that are utilized are products of the degradation of lignin, which is a polymer that constitutes about 25% of the earth’s biomass and is composed of hydroxylated and methoxylated phenylpropanoid units. Acetogenic bacteria have a special propensity for metabolism of methoxylated aromatics and this property has been exploited to selectively isolate acetogens [287]. M. thermoacetica has been shown to utilize at least 20 methoxylated aromatics [288].

Metabolism of the various methyl donors involves three successive two-electron oxidations of one of the methyl groups, generating six electrons that are used by CODH to convert three mol of CO₂ to CO, each of which then reacts at the ACS site with a methyl group and CoA to make acetyl-CoA. The methyltransferase systems that catalyze transfer of the methyl groups of methanol and methoxylated aromatic compounds to tetrahydrofolate have been identified by genomic and biochemical studies. A three-component system in M. thermoacetica that initiates the metabolism of methoxylated aromatic compounds is encoded by the mtvABC gene cluster (identified in the genomic sequence as Moth_0385-7 and Moth_1316-8) [98]. This system has been purified and identified biochemically by enzymatic function and protein sequence analysis [289]. A similar system has been identified in Acetobacterium dehalogenans, which also is linked to an ATP-dependent activating protein that can accept electrons from a hydrogenase [290, 291]. The Mtv system is coupled to the Wood-Ljungdahl pathway as shown in Figure 18. M. thermoacetica also encodes several homologs of MtvA, MtvB and MtvC that may used for growth on methyl donors other than vanillate and methanol [98].

Many acetogens can use methanol as a methyl donor by expressing the methanol methyltransferase system (MtaABC), encoded by Moth_1208-9 and Moth_2346 in the M. thermoacetica genome [98]). The M. thermoacetica methanol:methyl-H₄folate system was recently characterized and the crystal structure of the MtaC component was determined [292].

4. Heterotrophic Growth

In environments where both acetogens and methanogens are found, methanogens are usually the dominant hydrogenotrophs (see section IB). Acetogenic bacteria can grow in these environments because they have the ability to use a great variety of carbon sources and electron donors and acceptors. A long-recognized characteristic of acetogens is their ability to convert sugars stoichiometrically to acetate [38, 293]. A typical acetogen can use most of the substrates shown in Figure 19. M. thermoacetica, for example, can grow heterotrophically on fructose, glucose, xylose, ethanol, n-propanol, n-butanol [294], oxalate and glyoxylate [295], glycolate [296], pyruvate [293], and lactate [297], and acetate yields from growth on these substrates are consistent with oxidation to CO₂, and formation of acetate from all available electrons.

Figure 19.

Redox couples that can be used by acetogens. Figure taken from [336]. CO₂ reduction to acetate is one of many possible electron-accepting processes.

Genes encoding all enzymes of the Embden-Meyerhof-Parnas pathway pathway and the oxidative branch of the pentose phosphate pathway [98] are found in the M. thermoacetica genome. Thus, one mole of xylose is converted to 2.5 moles of acetate. It seems likely that fructose is metabolized by the same pathway as glucose, since growth yields on the two substrates are very similar [298]. Ethanol, propanol, and butanol are oxidized to acetate, propionate, and butyrate, respectively, [294] and genes encoding alcohol dehydrogenases and aldehyde:ferredoxin oxidoreductases have been located on the M. thermoacetica genome. When M. thermoacetica is grown on lactate, Moth_1826 appears to be the lactate dehydrogenase that catalyzes the conversion of lactate to pyruvate. Pyruvate is then metabolized using pyruvate:ferredoxin oxidoreductase (PFOR) and the Wood-Ljungdahl pathway enzymes, as described above.

The two-carbon substrates oxalate, glyoxylate and glycolate can be oxidized by two, four and six electrons, respectively. Thus, during growth on these substrates and CO₂, for each mole of acetate produced, approximately two moles of glyoxylate or four moles of oxalate are oxidized, and four moles of glycolate are oxidized to make three moles of acetate [295]. It is not clear from the genome sequence what pathway is used by M. thermoacetica for growth on the two-carbon substrates oxalate, glyoxylate, and glycolate, because only partial pathways known to be involved in glycolate metabolism in other organisms are present in M. thermoacetica [98]. Thus, post-genomic studies will be required to elucidate the pathways for growth on these compounds.

The Wood-Ljungdahl pathway serves two functions in acetogenic bacteria: as an electron-accepting, energy-conserving pathway, and as a pathway for carbon assimilation. Yet, as shown in Figure 19, reduction of CO₂ to acetate by the Wood-Ljungdahl pathway is only one of many electron-accepting pathways that can be used by acetogens. Respiration with higher redox potential acceptors than CO₂, such as nitrate, could provide more energy to the cell, thus it is not surprising that nitrate is preferred over CO₂ as electron acceptor for M. thermoacetica [299]. Accordingly, in undefined medium (with yeast extract to provide precursors for cell carbon), the growth of M. thermoacetica on a variety of electron donors (H₂, CO, formate, vanillate, ethanol, and n-propanol) produces more biomass with nitrate as electron acceptor than with CO₂ [300].

The presence of nitrate blocks carbon assimilation by the Wood-Ljungdahl pathway, by transcriptional regulation of genes encoding Wood-Ljungdahl enzymes, at least under some culture conditions, and by decreasing levels of cytochrome b in membranes [300, 301]. A similar down-regulation of acetogenesis has been seen in nitrite-supplemented cultures. Because of this repression, some substrates (e.g., methanol, vanillate, formate, CO, and H₂ + CO₂) cannot be used by M. thermoacetica for growth in nitrate media that lacks yeast extract (provides precursors for cell carbon). Glyoxylate is the only electron donor that has been shown to support growth with nitrite. In these cultures, ammonia was the reduced end-product, and CO₂ reduction to acetate only occurred in stationary phase, and cytochrome b was absent during the exponential growth phase [302].

Oxalate is a unique substrate in that M. thermoacetica preferentially reduces CO₂ over nitrate in the presence of oxalate, and produces acetate during exponential growth on oxalate, CO₂ and nitrate during exponential growth, [303]. Some substrates, including sugars, pyruvate, oxalate and glyoxylate support growth in basal medium with nitrate. M. thermoacetica can also grow in basal medium with nitrate if vanillate, which provides pre-formed methyl groups, and CO are supplied together.

To our knowledge, the ability of M. thermoacetica to make acetate in the presence of electron acceptors besides nitrate and nitrite has not been tested. In contrast to the tight regulation of acetogenesis by these electron acceptors in M. thermoacetica, other organisms have been shown to use other electron acceptors simultaneously with CO₂. With either fructose or H₂ as the electron source, A. woodii can reduce caffeate simultaneously with reduction of CO₂ to acetate, even though the reduction of caffeate is more favorable (ΔE_o’ = +32 mV for caffeate reduction vs. SHE, by analogy to the fumarate/succinate couple). However, methanol or betaine, when provided as the electron donors, inhibit caffeate reduction during acetogenesis [304]. Peptostreptococcus productus and Clostridium formicoaceticum both reduce fumarate to succinate simultaneously with CO₂ reduction to acetate [305, 306], although in P. productus, both activities are stimulated by increasing the levels of CO₂ in the medium [305]. Exogenous CO₂ levels can also affect utilization of other substrates; for example, C. formicoaceticum does not convert fructose to acetate in the absence of added CO₂ [307]. P. productus directs electron flow away from the Wood-Ljungdahl pathway when the CO₂ concentration is low, and although some electron donors were used more efficiently at high CO₂ concentrations, increasing the CO₂ concentration during growth on xylose decreased the cell yield, indicating that acetogenesis may not be the most energetically favorable metabolic pathway for this organism [305]. Clearly, acetogens have well-developed strategies for adjusting their growth to environments that compromise their ability to use the Wood-Ljungdahl pathway.

E. Reverse acetogenesis: Acetate catabolism by methanogens and sulfate reducers

As mentioned in the introduction, several classes of anaerobes other than acetogens, including methanogens and anammox bacteria, use the Wood-Ljungdahl pathway of CO₂ reduction to allow them to grow autotrophically by generating cell carbon from H₂ + CO₂. Furthermore, reverse acetogenesis enables the use of acetate as a carbon and electron source; however as indicated by Equation 1, acetate oxidation to H₂ + CO₂ is thermodynamically unfavorable. Therefore, the reaction has to be coupled to a thermodynamically favorable process, like methanogenesis or sulfate reduction.

The pathway of acetate catabolism by methanogens is shown in Figure 20 (color-coded as in Figure 1) (for review see [31]). Acetoclastic methanogens convert acetate in the growth medium to acetyl-CoA by the actions of acetate kinase and phosphotransacetylase. Then, the C-C and C-S bonds of acetyl-CoA are cleaved to release CoA and CO, which is converted to CO₂ by CODH, leaving a methyl-ACS intermediate on the A-cluster. Continuing the reverse of the acetogenesis pathway, the methyl group is transferred to the corrinoid iron-sulfur protein component of the methanogenic CODH/ACS (or acetyl-CoA decarbonylase synthase) complex. Two consecutive methyltransferase reactions then are involved in transfer of the methyl group from the methyl-[Co] intermediate to CoM to generate methyl-SCoM, which, in the presence of Coenzyme B, is converted to methane and the heterodisulfide product (CoB-SS-CoM) by the nickel metalloenzyme, methyl-SCoM reductase. In the overall process, the methyl and carboxyl groups of acetate are converted to methane and CO₂, respectively, while the two electrons released in the oxidation of CO to CO₂ are used by heterodisulfide reductase to reduce the heterodisulfide product of the MCR reaction back to the free thiolate state.

Figure 20.

Acetoclastic methanogenesis: coupling methanogenesis to the Wood-Ljungdahl pathway (reverse acetogenesis). MCR, methyl-SCoM reductase; HDR, heterodisulfide reductase

An interesting example of metabolic coupling of acetogenic and methanogenic metabolism is observed in co-cultures of acetogens (growing on sugars) and methanogens. The reducing equivalents produced during glycolysis and pyruvate oxidation are transferred to the methanogen, perhaps as H₂, to reduce CO₂ to methane [308]. Thus, per mol of glucose, two mol of acetate and one mol each of CO₂ and CH₄ are produced. Furthermore, when acetogens are co-cultured with acetate utilizing methanogenic strains, the three mol of acetate produced from glycolysis are converted to three mol each of CO₂ and CH₄ [309]. This metabolic interdependence was described for a co-culture of the mesophiles, A. woodii and Methanosarcina barkeri, and for a thermophilic co-culture [310, 311]. The physiology, biochemistry, and bioenergetics related to the syntrophic growth of acetogens with other organisms has been recently reviewed [312]. General principles include the requirement for a syntrophic association when free energy for the metabolic process is very low; for example, coupling anaerobic methane oxidation to sulfate reduction (with a free energy associated with this process of only -18 kJ/mol). On the other hand, various sulfate reducers are capable of coupling acetate oxidation to H₂S production as described in Figure 21 [35–37], a process for which the free energy is much more significant (Equation 4).

Figure 21.

Coupling reverse acetogenesis to sulfate reduction.

VII. Energy metabolism associated with acetogenesis

It has been long recognized that autotrophic growth by the Wood-Ljungdahl pathway must be linked to an energy-generating anaerobic respiratory process, since during autotrophic growth, there is no net ATP synthesis by substrate-level phosphorylation. However, the chemiosmotic pathway(s) that is connected to the Wood-Ljungdahl pathway has not been identified. Evidence for chemiosmotic ATP synthesis has been found in studies with A. woodii, M. thermoacetica, and Moorella thermoautotrophica (formerly, Clostridium thermoautotrophicum [313]), which is physiologically similar to M. thermoacetica [314].

A decade before it was shown to grow autotrophically, growth yields of M. thermoacetica on glucose and fructose were calculated to be higher than one would expect for ATP production by substrate level phosphorylation alone, and it was proposed that additional ATP is generated by an anaerobic respiratory process [315]. The high growth yields measured when A. woodii is cultured on fructose or caffeate also supported a chemiosmotic energy conservation mechanism for this acetogen [316]. Electron transport-linked ATP synthesis in M. thermoacetica received further support when b-type cytochromes and a menaquinone were identified in its membranes during growth on CO₂ and glucose [317].

In A. woodii and M. thermoacetica, the F₁F₀ ATP synthases responsible for ATP synthesis have been isolated. The F₁F₀ ATPase from M. thermoacetica was shown to be similar in subunit composition, and pattern of inhibition and stimulation by metal cations, anions, and alcohols to ATPases from mitochondria, chloroplasts, and other bacteria that use them for energy generation [318]. Evidence for ATP formation dependent on a proton gradient in M. thermoacetica includes increased ATP formation following an extracellular drop in pH, which was inhibited by ATPase inhibitor dicyclohexylcarbodiimide (DCCD) [318].

The Wood-Ljungdahl pathway enzyme(s) that are linked to transmembrane proton gradient formation is not known. The enzymes needed for acetate synthesis have been found to be soluble after preparation of cell extracts with a French pressure cell, but after gentler osmotic lysis of M. thermoautotrophica cells, much of the CODH and methylenetetrahydrofolate reductase activities remained in the membrane fractions [319]. Incubation of CODH- and methylenetetrahydrofolate reductase-containing membranes with CO or with dithionite caused eduction of the b-type cytochromes present in the membranes (the redox potentials of these cytochromes are −200 mV and −48 mV). When CO-reduced membranes were treated with CO₂, the lower potential cytochrome was partially re-oxidized [319]. Further studies with M. thermoautotrophica membranes showed that electrons from CO can reduce cytochrome b₅₅₉, followed by menaquinone, followed by cytochrome b₅₅₄. Based on these studies, and on the redox potential for the methylenetetrahydrofolate/methyltetrahydrofolate couple (−120 mV), Ljungdahl’s group proposed an electron transport chain (Figure 22) involving the following sequential steps: oxidation of CO (CO+H₂O ⇌CO₂2e−+2H⁺, ΔE_o’ = −540 mV) coupled to formation of H₂ or NADH by a flavoprotein, reduction of cytochrome b₅₅₉, which could then reduce methylenetetrahydrofolate or menaquinone and cytochrome b₅₅₄, which would finally reduce rubredoxin (ΔE_o’ ≈ 0 mV vs. SHE [320]) [321]. Driven by the oxidation of CO with ferricyanide, a proton motive force that could drive ATP synthesis and amino acid transport was measured in M. thermoacetica membrane vesicles [322]. However, which step would extrude protons in the proposed electron transport chain is still not known.

Figure 22.

Proposed electron transport chain from work done in Moorella species (modified from [98]). The specific proton translocating step and which specific electron carriers are coupled to the Wood-Ljungdahl pathway are not known. The cytochromes may be parts of larger protein complexes (i.e. cytochrome bd oxidase and formate dehydrogenase).

A cytochrome bd oxidase has recently been identified in M. thermoacetica [323]. Although M. thermoacetica is a strict anaerobe, it can tolerate traces of oxygen [324]. Furthermore, M. thermoacetica membranes catalyzed NADH-dependent O₂ uptake that is apparently coupled to a membrane-associated electron transport chain. Duroquinol- and quinol-dependent O₂ uptake activities increased in membrane preparations that were enriched for cytochrome bd oxidase, and it was proposed that the cytochrome bd oxidase is involved in protection from oxidative stress. Based on the genome sequence, the cytochrome b in this complex could be identified as cytb₅₅₉, which had been identified by Ljungdahl’s group many years earlier [323]. Another cytochrome (b₅₅₄) is part of a gene cluster annotated as formate dehydrogenase, although the substrate specificity of this protein is not clear from the sequence. Both of these annotations support the possible role of these cytochromes in electron transport coupled to energy conservation.

While some acetogens like M. thermoacetica seem to use a cytochrome-based proton-coupled electron transfer pathway, others, like A. woodii lack membrane-bound cytochromes, but, contain membrane-bound corrinoids that have been suggested, in analogy to the corrinoid-containing, Na⁺ -pumping methyltetrahydromethanopterin: coenzyme M methyltransferases of methanogenic archaea, to be involved in energy conservation ([325] and reviewed in [162]). A. woodii growth on fructose, methanol, or H₂ + CO₂, is dependent on the sodium concentration in the growth medium, and growth on fructose at low sodium levels decreased acetate production from 2.7 to 2.1 moles of acetate per mole of fructose, which is consistent with inhibition of synthesis of the third molecule of acetate from fructose by the Wood-Ljungdahl pathway (as described for acetogenesis from glucose in section II B1) [161]. Experiments with resting A. woodii cells showed that acetate production from H₂ + CO₂ is also dependent on sodium ions and that these cells can produce a strong Na⁺ gradient in which concentrations inside cells are as much as forty-fold lower than outside, corresponding to a transmembrane chemical potential of −90 mV [161]. Furthermore, Na⁺-dependent ATP formation by A. woodii cells is inhibited by the F₁F₀ ATPase inhibitor DCCD [326, 327]. Na⁺ is taken up into inverted membrane vesicles, coupled with acetogenesis [328]. The Na⁺ dependent ATPase was purified and shown to be an unusual F₁F₀ ATPase [329], with two types of rotor subunits, e.g., two bacterial F(0)-like c subunits and an 18 kDa eukaryal V(0)-like c subunit [330, 331]. The Na⁺ dependent step was shown to be in the methyl branch of the Wood-Ljungdahl pathway, and it has suggested that the responsible enzyme is either methylenetetrahydrofolate reductase or the methyltransferase/corrinoid protein involved in methyl group transfer from tetrahydrofolate to ACS.

A possible respiratory pathway for growth on caffeate as an electron acceptor has been recently identified in A. woodii [332] (reviewed in [333]). Membrane-bound ferredoxin:NAD⁺ oxidoreductase activity was found, and is postulated to be the activity of a protein complex homologous to the Rnf complex of Rhodobacter capsulatus. Genetic evidence for such a complex in A. woodii has been shown [332]. The possible Rnf complex has also been hypothesized as the Na⁺-translocating protein for growth by the Wood-Ljungdahl pathway. Reduced ferredoxin, from hydrogenase or PFOR would be used by Rnf to reduce NAD⁺, which could provide electrons to the methyl branch of the pathway.

In summary, both H⁺- and Na⁺-dependent acetogens produce electrochemical gradients that can be used for ATP synthesis by F₁F₀ ATPases. In A. woodii, ATP synthesis from a transmembrane Na⁺ gradient is linked to acetogenesis by some step in the methyl branch of the Wood-Ljungdahl pathway, possibly at methylenetetrahydrofolate reductase, or via an Rnf ferredoxin:NAD⁺ oxidoreductase. In M. thermoacetica and M. thermoautotrophica, methylenetetrahydrofolate reductase and CO dehydrogenase could be linked to energy conserving steps in the Wood-Ljungdahl pathway. The possible electron transport chain in Moorella is summarized in Figure 22, which was modified from [322].

The genome studies of M. thermoacetica give some insight into possible energy conserving pathways [98]. Homologs of NADH dehydrogenase I and archaeal heterodisulfide reductase are present in the genome. NADH dehydrogenase in E. coli, as in other bacteria, has 13 subunits. As discussed in the genome paper [98], the genes encoding NADH dehydrogenase are indicated in Figure 23 to be encoded by three gene clusters: Moth_0977-87, which encodes ten of the thirteen NADH dehydrogenase subunits; and Moth_1717-9 and Moth_1886-6, which appear to be redundant and encode the other three subunits (NuoE, F, and G in E. coli). Moth_0979, which is in the middle of the first gene cluster, has no homolog in E. coli. Other genes identified that may play important roles in electron transfer linked proton translocation are Moth_2184-90, with significant homology to the genes encoding the subunits of E. coli’s hydrogenase 4, which has been proposed to catalyze proton translocation [334]. These various sets of genes could encode partial or complete membrane-bound complexes that would function in anaerobic respiration by generating a transmembrane proton gradient for use by the F₁F₀ATP synthase. Localization of genes encoding one homolog of heterodisulfide reductase, a hydrogenase, and 5,10-methylene-H₄folate reductase near the acs gene cluster are consistent with a suggestion that the reductase and a hydrogenase might link to CODH to generate a proton motive force, driving anaerobic respiration [160], as suggested in Figure 23. This figure is depicted as a speculative and highly incomplete scheme to underline our lack of knowledge in this area. Hopefully, the M. thermoacetica genome sequence [98], which contains a number of potential electron transfer components with as yet unknown functions, will aid in the identification of the various electron acceptors in the pathways shown in this scheme. Furthermore, there is evidence that in M. thermoautotrophica, CO dehydrogenase and methyl-H₄folate reductase, which were isolated from the soluble fraction of the cell extract (above in the Wood-Ljungdahl pathway enzyme section), can associate with membranes [319]. The presence of multiple gene clusters encoding possible membrane-bound respiratory complexes could be explained by the ability of M. thermoacetica to use many electron acceptors as different respiratory proteins could be induced by different growth substrates.

Figure 23.

Protein complexes possibly involved in proton transfer in M. thermoacetica.

VIII. Prospective: Questions that remain unanswered

Many questions that remain about the biochemistry and bioenergetics of acetogenesis and of the Wood-Ljungdahl pathway have become more tractable with the sequencing of the first acetogenic genome. Some of these questions are reiterated here, with the details provided above.

A number of questions remain about the enzymology of the Wood-Ljungdahl pathway. Although it has been clearly shown that general acid catalysis is involved in the methyl transfer reactions of the MeTr involved in the Wood-Ljungdahl pathway and in methionine synthase, it needs to be firmly established whether there is a generally applicable mechanism of proton transfer (i.e., in the binary or ternary complex, or in the transition state for the methyl transfer reaction), or if perhaps these enzymes use different mechanisms. Biochemical, biophysical, and structural studies of the various proposed conformational states of the MeTr:CFeSP and CFeSP:CODH/ACS complexes. Although the structure of the CFeSP clearly defined the two TIM barrel domains including the corrinoid binding site, it is important to determine the protein structure around the [4Fe-4S] cluster, which was not resolved.

Although there is ample spectroscopic evidence for Ni-CO intermediates on CODH and ACS, neither of these has been observed by X-ray crystallography. Furthermore, no crystal structures have been reported for the methylated or the CoA-bound states of ACS. It is important to resolve the apparently conflicting results supporting the binding of the CODH competitive inhibitor CN⁻ to Ni or Fe. Evidence for a spin-coupled state on ACS was recently provided by Mossbauer spectroscopy and there are proposals for a Ni(0) state. Although the Ni(0) seems rather unlikely, it is important to further characterize and to establish whether or not there is any physiological relevance for the spin-coupled state.

The role of one CODH, which is part of the acs gene cluster as part of the CODH/ACS machine is well understood; however, enzymatic and gene expression studies are required to elucidate the function of the other CODH (Moth_1972), which is unlinked to this gene cluster.

A number of questions remain about the coupling of acetyl-CoA formation by the Wood-Ljungdahl pathway to the reactions and pathways that are coupled to it. The phosphotransacetylase associated with the Wood-Ljungdahl pathway needs to be unambiguously identified by sequencing the active protein. It is not clear how binding of CoA accelerates electron transfer among the internal iron-sulfur clusters by ~100,000-fold. With respect to the pyruvate synthase reaction, it is not clear how acetogens accomplish the energetically unfavorable carboxylation of acetyl-CoA; for example, if reverse electron transfer is involved in driving this reaction and, if so, what is the mechanism by which this is accomplished. In addition, whether M. thermoacetica has a complete or incomplete reductive TCA cycle is unclear.

It is unknown what sequence of reactions M. thermoacetica uses for growth on the twocarbon substrates oxalate, glyoxylate, and glycolate, because only partial pathways known to be involved in glycolate metabolism in other organisms are present in M. thermoacetica; thus, biochemical and perhaps gene expression studies will be required to characterize these pathways.

Perhaps the most important unsolved question related to acetogenesis and the Wood-Ljungdahl pathway is how acetogens conserve energy for growth. A proposed mechanism of energy conservation and proton translocation involving the 5,10-methylene-H₄folate reductase and the membrane-associated hydrogenase and/or heterodisulfide reductase. It is likely that the acetogens that contain membrane-bound cytochromes (like M. thermoacetica) will have some components of the energy generation system in common with those that lack cytochromes (like A. woodii) and some that are distinct. For the non-cytochrome containing acetogens, there are important questions about the role of the Rnf complex described above [332]. Is it an ion (Na⁺) pump? Is it a coupling site in acetogens, and, if so, how is it coupled to the Wood-Ljungdahl pathway? What is the role of the membrane bound corrinoids in the acetogens that lack cytochromes, and does the corrinoid-dependent methyltransferase catalyze Na⁺ translocation?