Ferredoxin reduction by hydrogen with iron functions as an evolutionary precursor of flavin-based electron bifurcation

Significance

In the most ancient biochemical pathway known—the reductive acetyl-CoA pathway—the energy to fix CO₂ comes from geochemically produced H₂ gas, not from the sun. Cells that fix CO₂ with H₂ require an energetic trick: They split the electron pair in H₂, boosting the energy of one electron at the expense of the other to charge an ancient protein called ferredoxin with electrons. That requires enzymes and cofactors. How was ferredoxin charged before enzymes and cofactors arose? In alkaline, H₂-producing hydrothermal systems, the electron-donating potential of H₂ can reduce ferredoxins but requires the presence of a transition metal. We show that in water, raw iron will transfer electrons from H₂ to ferredoxin, uncovering a missing link in early metabolic evolution.

Abstract

Autotrophic theories for the origin of metabolism posit that the first cells satisfied their carbon needs from CO₂ and were chemolithoautotrophs that obtained their energy and electrons from H₂. The acetyl-CoA pathway of CO₂ fixation is central to that view because of its antiquity: Among known CO₂ fixing pathways it is the only one that is i) exergonic, ii) occurs in both bacteria and archaea, and iii) can be functionally replaced in full by single transition metal catalysts in vitro. In order to operate in cells at a pH close to 7, however, the acetyl-CoA pathway requires complex multi-enzyme systems capable of flavin-based electron bifurcation that reduce low potential ferredoxin—the physiological donor of electrons in the acetyl-CoA pathway—with electrons from H₂. How can the acetyl-CoA pathway be primordial if it requires flavin-based electron bifurcation? Here, we show that native iron (Fe⁰), but not Ni⁰, Co⁰, Mo⁰, NiFe, Ni₂Fe, Ni₃Fe, or Fe₃O₄, promotes the H₂-dependent reduction of aqueous Clostridium pasteurianum ferredoxin at pH 8.5 or higher within a few hours at 40 °C, providing the physiological function of flavin-based electron bifurcation, but without the help of enzymes or organic redox cofactors. H₂-dependent ferredoxin reduction by iron ties primordial ferredoxin reduction and early metabolic evolution to a chemical process in the Earth’s crust promoted by solid-state iron, a metal that is still deposited in serpentinizing hydrothermal vents today.

Autotrophic theories for life’s origins posit that the first organisms on Earth were CO₂-fixing microbes. In that view, relicts from primordial CO₂ fixation might have been preserved in modern metabolism, bringing CO₂ fixation pathways to bear the origin of metabolism and life (1). Among known routes of CO₂ fixation, the acetyl-CoA pathway is the most ancient (2). It is the only CO₂ fixation pathway known to operate in both archaea (methanogens) and bacteria (acetogens) (3) and the only one that is exergonic, releasing energy in the synthesis of acetyl-CoA (CH₃COSCoA) from H₂, CO₂, and coenzyme A (CoASH) via formate as an intermediate (4) according to

$$2{\text{CO}}_2+4{\mathrm{H}}_2+\text{CoASH}\to {\text{CH}}_3\text{COSCoA}+3{\mathrm{H}}_2\mathrm{O},$$

with ΔG_o' = –59 kJ/mol (2). Roughly 70% of the resulting acetyl-CoA pool is further converted to pyruvate (CH₃COCOO^–) by ferredoxin-dependent CO₂ fixation at the pyruvate synthase reaction

$$2{\text{Fd}}_{\text{red}}+{\text{CO}}_2+{\mathrm{H}}^{+}+{\text{CH}}_3\text{COSCoA}\to 2{\text{Fd}}_{\text{ox}}+\text{CoASH}+{\text{CH}}_3{\mathrm{\text{COCOO}}}^{–},$$

for the synthesis of amino acids, bases, cofactors, and cell mass (2). The enzymes of the acetyl-CoA pathway are replete with transition metals as electron carriers and catalysts (4–8), inorganic cofactors that are reasonably interpreted as relicts (9–12) from a geochemical setting (13) within which metabolism and life arose. In acetogens and methanogens growing on H₂ and CO₂, the acetyl-CoA pathway converts H₂ and CO₂ into formate (4), acetate, and pyruvate (2) via ~10 enzymatic reactions that require 14 cofactors and 127 proteins, including the cofactor biosynthesis pathways (14).

Newer findings reveal that under the conditions of H₂-producing (serpentinizing) hydrothermal vents (15, 16), the function of those 127 proteins can be replaced in toto by individual transition metal minerals such as Ni₃Fe or Fe₃O₄ in vitro. These catalysts not only convert CO₂ and H₂ with high specificity to formate, acetate, and pyruvate in the laboratory (17–20), they are deposited in serpentinizing hydrothermal vents in nature (21–23). They can even convert H₂ and CO₂ to pyruvate at concentrations of 200 µM (19), the physiological pyruvate concentration in the cytosol of acetogens growing on H₂ and CO₂ (24). Whether catalyzed by enzymes or metals, these ancient metabolic reactions go forward because in the reaction of H₂ with CO₂, the equilibrium lies on the side of reduced carbon compounds (25). Transition metal catalysts forge a functional link between ancient metabolism and modern serpentinizing hydrothermal systems, which synthesize formate via abiotic geochemical reactions in amounts sufficient to support microbial growth (26–30).

Though the reactions of the acetyl-CoA pathway naturally unfold from H₂ and CO₂ in the presence of transition metal catalysts in the laboratory, this does not, by itself, solve the problem of metabolic origins, because the biochemically catalyzed acetyl-CoA pathway as it occurs in cells presents a thermodynamic paradox. The CO₂-reducing enzymes of the acetyl-CoA pathway require reduced ferredoxin, Fd, as their electron donor, whereby Fd is reduced with electrons from H₂ via hydrogenases (Fig. 1). However, the midpoint potential of H₂ under standard physiological conditions, E_o′ = –414 mV, is more positive than the midpoint potential of low potential ferredoxins, roughly –450 to –500 mV in cells (31) as required for CO₂ fixation (E_o′ = –430 mV), making Fd reduction with H₂ a steeply endergonic process (Fig. 1A), especially in environments with low H₂ partial pressures (31). How do cells push electrons energetically uphill to Fd for CO₂ fixation?

Fig. 1.

Ferredoxin is the primordial one-electron carrier in metabolism. (A) Flavin-based electron bifurcation is required to reduce low potential ferredoxin for CO₂ fixation using electrons from H₂ (schematic). The flavoprotein and its high potential acceptor varies across organisms and pathways (31–42). (B) The structure of C. pasteurianum ferredoxin (PDB ID: 1CLF). Other renderings of ferredoxin are given in SI Appendix, Fig. S1. (C) The structure of the two forms of hydrogenase used to oxidize H₂ for Fd reduction and their catalytic metal clusters compared to Fd from C. pasteurianum. The structures are for MvhAGD, the [NiFe] hydrogenase from Methanothermococcus thermolithotrophicus (PDB ID: 5ODC) (36) and HydABC, the [FeFe] hydrogenase from Acetobacterium woodii (PDB ID: 8A5E) (37). The CN^– and aminodithiolate ligands are missing in the structure of the H-cluster of HydABC. One protomer of HydABC and one protomer of MvhAGD are shown, the biologically active complexes being a homodimer of heterotrimers and a homodimer of heterohexamers (including the heterodisulfide reductase), respectively. Under nickel limitation, some methanogens express a third form of hydrogenase with a unique iron-guanylylpyrinidol cofactor that transfers electrons from H₂ directly to methenyl-tetrahydromethanopterin without FeS cluster intermediates (8).

The discovery of flavin-based electron bifurcation (32, 33) revealed how the main groups of H₂-dependent autotrophs, methanogens and acetogens, reduce ferredoxin with H₂ (34, 35). The two electrons from H₂ are transferred via a hydrogenase (36, 37) to an enzyme-bound flavin, where they diverge along exergonic and endergonic paths, respectively. One electron is transferred to a high potential acceptor such as NAD⁺ or the heterodisulfide CoM-S-S-CoB, the other is transferred, in strict stoichiometry, to Fd as the low potential acceptor (Fig. 1A). Reduction of the high potential acceptor renders the overall Fd-reducing reaction exergonic (31–42). All flavin-based electron bifurcation redox couples known so far reduce Fd (39) or flavodoxins under Fe limitation (41). This indicates that the ancestral physiological function of flavin-based electron bifurcation was to supply reduced Fd as a low potential reductant for metabolism, in turn welding both flavin-based electron bifurcation and Fd into the foundation of autotrophic theories for the origin of metabolism because, in concert, they supply the electrons for FeS and FeNiS clusters in enzymes of the acetyl-CoA pathway (2–8, 31, 35, 43, 44).

While flavin-based electron bifurcation solves the bioenergetic problem of Fd reduction with H₂ in metabolism, it presents a paradox in prebiotic physiology (45, 46): Given the antiquity of the acetyl-CoA pathway, how was Fd reduced in early evolution i) before the origin of the more complicated Fe–Ni and Fe–Fe hydrogenases that oxidize H₂ (Fig. 1B) and ii) before the origin of the vastly more complicated multi-enzyme process of flavin-based electron bifurcation (31–42) (Fig. 1A) that is required for Fd reduction in H₂-dependent chemolithoautotrophs? Because of its universality among anaerobes, and by functional, structural, and biogenetic simplicity (Fig. 1B), Fd is clearly more ancient than the hydrogenases (Fig. 1C) that supply reduced ferredoxin as the reductant for CO₂ fixation. The FeS clusters of Fd are not only simpler in structure than hydrogenase active sites, which extract electrons from H₂ for Fd reduction, they also lack the small molecule ligands that are added to iron during hydrogenase active site maturation (47–49) (Fig. 1C) by maturases that arose subsequent to the hydrogenase structural protein (50). In early biochemical evolution, simpler mechanisms must have therefore preceded hydrogenases and flavin-based electron bifurcation as a means of supplying reduced ferredoxin, though such mechanisms have not been experimentally identified so far. Here, we investigate the ability of native transition metals to reduce ferredoxin with H₂ in the absence of flavins or electron-bifurcating enzymes.

Results

We used H₂ as the source of electrons for Fd reduction because it is the reductant that drove primary production on Earth prior to the origin of photosynthesis starting with the first signs of life 3.8 billion years ago (51, 52). To activate H₂ we used native metals and metal alloys that are naturally deposited in serpentinizing systems (21–23). Serpentinization entails geochemical redox reactions between Fe(II) minerals and H₂O circulating through the Earth’s crust (15, 16), generating H₂ as the source of energy and electrons for primary production by bifurcation-dependent anaerobic autotrophs that inhabit serpentinizing systems in the dark subsurface: acetogens (26, 53) and methanogens (27, 28). For assays, we used glass vials and 5 bar of H₂, corresponding to 3.8 mM H₂ in the aqueous phase, well within the range of H₂ concentrations measured in the effluent of Lost City (29, 54) and other serpentinizing hydrothermal systems (46, 55). The reduction of ferredoxin from Clostridium pasteurianum purified from Escherichia coli can be followed by the standard photometric assay (56) by following the difference in absorption at 420 nm (Fig. 2).

Fig. 2.

Photometric assay of Fd reduction. (A) UV-Vis spectra following the oxidation of 30 µM of reduced Fd from C. pasteurianum in 0.133 M phosphate buffer at pH 8.5 through air exposure for 300 s. Fd was previously reduced through addition of two equivalents of sodium dithionite (Na₂S₂O₄). (B) Magnified boxed region in A showing the difference in absorption of the [4Fe4S] clusters at 420 nm and the isosbestic point at 354 nm (56), used for normalization. The time course of Fd reduction is shown in SI Appendix, Fig. S2.

Previous studies showed that Fe⁰, Ni⁰, and Co⁰ can catalyze the reduction of NAD⁺ to NADH under mild hydrothermal conditions: 5 bar H₂, 40 °C (57). NAD⁺ is a stronger oxidant (E_o′ = –320 mV) than C. pasteurianum Fd (E_o′ = –412 mV) (58) and NAD⁺ reduction is a two-electron reaction involving hydride (H^–) transfer while ferredoxin reduction involves one-electron transfer to [4Fe4S] centers. A screen of transition metals revealed that Ni⁰, Co⁰, Mo⁰, NiFe, Ni₂Fe, Ni₃Fe, and Fe₃O₄ do not mediate H₂-dependent Fd reduction, whereas Fe⁰ reduces Fd in the presence of H₂ (Fig. 3). The concentration of Fd in reactions was 30 µM, slightly lower than the physiological concentration in anaerobes, where Fd concentrations are in the range of 80 to 400 µM (59).

Fig. 3.

Metals tested for H₂-dependent Fd reduction. Reaction of 30 µM Fd from C. pasteurianum with 0.1 mmol (A) native metal and (B) metal alloys and magnetite under a hydrogen atmosphere [5 bar] in 0.133 M phosphate buffer at pH 8.5. Normalized absorption of [4Fe4S] clusters in Fd is shown (Methods). Raw data for iron shown in SI Appendix, Fig. S3.

Ferredoxin reduction by Fe⁰ required roughly four hours at 40 °C (Fig. 4). Notably, C. pasteurianum Fd does not autoreduce under these conditions, as shown in Fig. 4 for reactions of Fd and H₂ without metal. This is also seen in reactions of Fig. 3 where no Fd reduction was detected. Ferredoxin reduction with Fe⁰ at pH 8.5 required H₂ and solid-state iron (Fig. 3). The lack of autoreduction is noteworthy, because FeNiS and FeS enzymes of the acetyl-CoA pathway (acetyl-CoA synthase/carbon monoxide dehydrogenase and pyruvate synthase) are able to activate H₂ via latent hydrogenase activities (60). However, C. pasteurianum Fd does not oxidize H₂, in agreement with the absence of hydrogenase activity for Fd from Clostridium thermoaceticum (60). The FeS clusters of C. pasteurianum Fd require an iron surface for reduction under the conditions tested here.

Fig. 4.

Time course of Fe⁰-promoted H₂-dependent Fd reduction. The lag time of about 1 h might reflect delayed formation of chemisorbed hydrogen on the iron surface (61, 62) or H₂-dependent formation of active sites on the metal surface. A similar lag time was observed for H₂-dependent NAD⁺ reduction on Fe⁰ (57).

Early studies detected an effect of pH on the reduction of FeS proteins from C. pasteurianum and other sources, though the effects were generally small in the pH range of 6 to 9 (63). Subsequent studies did not detect pH-dependent effects (58). The conditions in our investigation are aligned with serpentinizing systems, where the midpoint potential of H₂ oxidation is directly influenced by pH through the reaction

$${\mathrm{H}}_2\to 2{\mathrm{e}}^{-}+2{\mathrm{H}}^{+},$$

which is pulled to the right at alkaline pH via proton removal, resulting in very negative midpoint potentials. In actively serpentinizing hydrothermal systems, pH can exceed 10 and the midpoint potentials can exceed –700 mV (26, 30, 46, 55). We observed a tendency for the pH to increase with reaction time in runs with Fe⁰ despite the presence of 133 mM P_i buffer. To see whether the observed Fd reduction is due solely to a pH effect (altered midpoint potential of H₂ oxidation) on metal-dependent water reduction, native metals were tested at pH 10. For Ni⁰ and Co⁰, there was no observable Fd reduction with H₂ at pH 8.5 or pH 10 (Fig. 5B). At pH 8.5, Fd reduction required Fe⁰ and H₂, while at pH 10, there was also substantial Fd reduction without addition of exogenous H₂ (Fig. 5A). This was expected based on findings from NAD⁺ reduction using ²H labeling, where it was observed that at pH 10, Fe⁰ generates endogenous ²H^– from ²H₂O (57). The yield of reduced ferredoxin for Fe⁰ without H₂ at pH 10 relative to the yield without H₂ at pH 8.5 (Fig. 5A) might result from the more negative H₂ midpoint potential at increased pH. Reduction of Fd cannot be tested with ²H₂O in the same way as for NAD⁺ (57) because Fd accepts single electrons rather than hydride. Surface oxidation of Fe⁰ can generate soluble Fe²⁺ (64–66), which is not, however, a sufficiently strong reductant to reduce ferredoxin (Fe²⁺→Fe³⁺+ e^–, E_o′ = –200 mV) (67).

Fig. 5.

H₂-dependent Fd reduction at pH 10. Reaction of 30 µM Fd from C. pasteurianum with 0.1 mmol metal. (A) Native iron and (B) native cobalt and native nickel under a hydrogen atmosphere [5 bar] in 0.133 M phosphate buffer at pH 8.5 and pH 10. Normalized absorption of [4Fe4S] clusters in Fd is shown (Methods).

To further probe the effect of pH, we increased the buffer strength to 1.33 M P_i and incubated in the presence of H₂ at different pH values (Fig. 6). At pH 7, where the midpoint potential of the H₂ oxidation is E_o = –414 mV, close to the midpoint potential of Fd, the reaction is incomplete. At pH 6, the reaction is blocked, as expected. Ferredoxin reduced by H₂ over Fe⁰ is biologically active, it serves as a substrate to reduce NADP⁺ in the reaction catalyzed by ferredoxin:NADP oxidoreductase from spinach chloroplasts (SI Appendix, Fig. S4).

Fig. 6.

pH dependence of Fd reduction with H₂. Reaction of 30 µM Fd from C. pasteurianum with 0.1 mmol native iron under a hydrogen atmosphere [5 bar] in 0.133 M and 1.33 M phosphate buffer at pH 8.5. Normalized absorption of [4Fe4S] clusters in Fd is shown (Methods). The midpoint potential of H₂ was calculated using the Nernst equation.

Discussion

In water at 40 °C, H₂ can reduce C. pasteurianum Fd at a concentration of 30 µM in the presence of Fe⁰ within a few hours. The pH and partial pressure of H₂ used (5 bar, roughly 3.8 mM in solution according to Henry’s law) correspond to conditions in actively serpentinizing hydrothermal vents (29, 46, 54, 55). The 30 µM Fd concentration used is close to that in cells. We found that iron was effective in reducing Fd with H₂. Neither the Fe–Ni alloys tested nor magnetite were effective in Fd reduction. Though the mechanism of electron transfer from the heterogeneous Fe⁰ surface to Fd is obscure, it requires Fe in the solid state in these experiments and likely entails chemisorption for H₂ activation (61, 62) with electron transfer mechanisms possibly similar to those that operate between FeS clusters in proteins (68, 69). Because Fe⁰ conducts, H₂ oxidation and Fd reduction need not occur at the same site on iron particles. Several studies have used electrodes with externally applied potentials to promote prebiotic chemical reactions (70, 71). We applied no external potentials. At alkaline pH, serpentinization and Fe⁰ produce the diffusible reductant H₂ by interactions with water. The reduction of Fd took place via redox reactions with exogenous H₂ and Fe⁰ at pH 8.5 and without exogenous H₂ at pH 10. At pH 6, Fd reduction does not take place (Fig. 6). The reason why Fe⁰ reduces Fd in our reactions while the other metals tested do not (Fig. 3) is likely due to the more negative midpoint potential of Fe⁰ in reaction to the divalent ion relative to the other metals tested (57, 67) or to differences in H₂ chemisorption among metals tested (57), or both.

An Ancient Root to Modern Corrosion?

Though motivated by the study of early evolution, our findings bear upon a modern problem: the corrosion of iron by microbes that use Fe⁰ as an electron source. A number of H₂-dependent anaerobic microbes that use the acetyl-CoA pathway, most notably methanogens and acetogens, can use Fe⁰ as an electron source for growth (64–66). Iron oxidation results in iron dissolution as Fe²⁺ and corrosion. The molecular mechanisms by which acetogens and methanogens extract electrons from Fe⁰ are still unknown (66), but the transient formation of H₂ from oxide-generating reactions of iron and water—as suggested in a previous study of NAD⁺ reduction (57) and as we suspect for Fd reduction without H₂ at pH 10 (Fig. 6)—are widely discussed (64–66).

While it is possible that use of Fe⁰ as a source of electrons is a recent microbial adaptation to human metalworking technology, it is far more likely that this physiological capacity is ancient, a holdover from growth on ubiquitous Fe⁰ and iron alloys (21–23) in serpentinizing systems, which are as old as water on Earth (15, 16). This would be highly compatible with phylogenetic and physiological evidence for the age of acetogens and methanogens (26, 53, 72–74), lineages that can grow on iron. Native metals are deposited in hydrothermal systems when serpentinization and H₂ production are active. When H₂ production through serpentinization stops, deposited Fe⁰ is still available (21–23), which can serve as the sole electron source for several lineages of acetogens and methanogens (64–66). Our findings raise the possibility that FeS clusters localized in proteins at the cell surface mediate growth on iron as an electron source in organisms that do not use extracellular multiheme cytochromes (66) or extracellular flavins to access electrons.

No Requirement for Membranes or Ion Gradients.

The enzymatic process of flavin-based electron bifurcation requires neither membranes, ion gradients, nor membrane-integral proteins (31–42), hence it could have operated at a very early stage of biochemical evolution, before the origin of free-living cells and even before the origin of membranes. During the course of evolution, microbes evolved membrane-dependent protein complexes that generate reduced ferredoxin without flavin-based electron bifurcation. These include the energy-conserving hydrogenase Ech, which taps ion gradients at the plasma membrane to reduce Fd with H₂ (75, 76). It has been proposed that Ech-like functions might have been required for primordial ferredoxin reduction (77, 78), yet Ech is evolutionarily derived from soluble Fe–Ni hydrogenases (79), which require electron bifurcation for autotrophic function, such that Ech can hardly be an ancestral Fd-reducing agent. Our findings show that neither ion gradients, membranes, nor Ech-like functions are required by the far simpler process of Fd reduction in the presence of H₂ and iron.

A Metal:Protein Hybrid Intermediate.

In contrast to previous studies of inorganic-catalyzed prebiotic reactions (80), the substrate of our iron-dependent reaction is a protein with FeS clusters. Ferredoxin reduction with H₂ and Fe⁰ under serpentinizing conditions closes an important gap in early bioenergetic evolution, providing an example of a hybrid intermediate state in the evolution of metabolism subsequent to abiotic versions of the acetyl-CoA pathway (17–20, 81) yet prior to the origin of enzymes that perform flavin-based electron bifurcation (Fig. 7). The ancestral route of pyruvate synthesis from H₂ and CO₂ is readily catalyzed by a number of different transition metal catalysts, including Fe⁰, Ni⁰, Ni₃Fe, and Fe₃O₄ (17–20) that occur naturally in serpentinizing hydrothermal vents (Fig. 7A). With the advent of genes and proteins, primitive versions of the acetyl-CoA pathway (72, 82) comprising the Fd-dependent enzymes carbon monoxide dehydrogenase/acetyl-CoA synthase and pyruvate synthase (5–8) could replace inorganically catalyzed pyruvate synthesis, but the activity of the soluble enzymes remained tied to transition metals in the crust for the supply of reduced ferredoxin as a kind of electrochemical umbilical cord connecting H₂ activation (solid phase) to one-electron supply at enzymes (aqueous phase) (Fig. 7B).

Fig. 7.

Three phases in the evolution of autotrophy via the acetyl-CoA pathway. (A) Metal-catalyzed pyruvate synthesis from H₂ and CO₂ (17–20, 80) as the ancestral state of the acetyl-CoA pathway. (B) Iron-dependent ferredoxin reduction with H₂ as a source of the physiological reductant (reduced ferredoxin) for CODH/ACS (5, 7, 60) and PFOR (6, 34, 60) in an intermediate stage of physiological evolution before the origin of flavin-based electron bifurcation. (C) H₂ oxidation via Fe–Fe and Ni–Fe hydrogenases and flavin-based electron bifurcation for the synthesis of reduced ferredoxin in a fully soluble enzymatic system encoded by genes. The source of methyl groups for the ACS reaction in (B) can either be geochemical (72, 83) as in (A) or biochemical from CO₂ as in the extant acetyl-CoA pathway (2).

Given geochemical availability of H₂ from serpentinization and CO₂ from seawater, this hybrid intermediate state would be indefinitely sustainable, but metabolic evolution would remain bound to the Earth’s crust by the requirement for solid-state metals. Escape from the crust required a number of inventions, among them cellularization and the ability to fix CO₂ without the requirement for physical contact between cytosolic enzymes and metals in the crust. The origin of hydrogenases (36, 37, 50) and flavin-based electron bifurcation freed H₂-dependent ferredoxin reduction from solid-state contact, but coupled Fd reduction to the simultaneous reduction of stronger oxidants (Fig. 7C). These eventually included NAD⁺ via the electron-bifurcating Fe–Fe hydrogenase reaction of acetogens (35) or CoM-S-S-CoB via the MvhADG/HdrABC complex (Fe–Ni hydrogenase/heterodisulfide reductase) of methanogens (34), coupling Fd-dependent CO₂ reduction to stoichiometrically balanced energy metabolism in these H₂-dependent chemolithoautotrophs (39). The evolutionary innovation of flavin-based electron bifurcation enabled the use of H₂ as an electron donor for Fd reduction at cytosolic pH values near 7. Despite their divergent mechanisms of pumping for ATP synthesis (43, 44), the acetyl-CoA pathway links methanogens and acetogens (84), primordial members of the archaeal and bacterial lineages (72–74), with serpentinizing environments (15, 16, 85), which they still inhabit today (26–30, 53, 74). Our findings provide a crucial missing link in early metabolic evolution, closing a gap in autotrophic theories by demonstrating a functional intermediate state in the transition from geochemical to enzymatic catalysis in H₂-dependent CO₂ fixation (Fig. 7), as required if life arose at hydrothermal vents (13) involving the acetyl-CoA pathway (83).

In the Light of Origin Theories.

Do the present findings discriminate among theories for the origin of life? The main competing theories for origins encompass genetics first and metabolism first proposals. Genetics first theories posit, in current mainstream formulations, that RNA-based replicating systems preceded the origin of life-like metabolic pathways and that the RNA bases were synthesized with the help of atmospheric reactions giving rise to cyanide, C≡N^–, or C≡N-containing compounds that fueled RNA synthesis until metabolism arose (86–88). In genetics first theories, reactions between H₂ and CO₂ play no role, as the reduced carbon species that initiate primordial amino acid and nucleobase synthesis contain reduced carbon with highly reactive C≡N bonds (89–91). Because C≡N-containing compounds are absent in modern biosynthesis, genetics first models do not intersect microbial metabolism, such that flavin-based electron bifurcation, H₂-dependent CO₂ fixation, and the acetyl-CoA pathway do not address genetics first, and vice versa.

Metabolism first theories posit that organized and catalyzed chemical reaction networks having evident similarity to modern metabolic pathways are the starting point of life chemistry, from which genetic systems later emerged (92). In autotrophic theories, metabolism started from CO₂, just like it does in the primary production of modern ecosystems (1). Autotrophic theories that draw upon hydrothermal vents (13) and H₂ from serpentinization to reduce CO₂ via the acetyl-CoA pathway to pyruvate strictly require evolutionary precursors of electron bifurcation (45, 46), readily accommodating the present findings. From pyruvate, the central intermediate of biosynthesis (2, 93), the paths to the incomplete reductive reductive citric acid cycle (94), amino acids, nucleotide, and cofactor synthesis are short, entailing only about 400 reactions (95). Amino acid and nucleobase syntheses from biochemical starting compounds can take place without enzymes (80, 96, 97), as suggested in these pages in 1963: “primitive organic cofactors (along with inorganic ions) acting on surfaces were the original ‘bio’ catalysts, active long before specific polypeptides evolved” (92). As Fe⁰ presents a surface, our findings (Fig. 7) are in full accord with that depiction.

Origins theories that start with methane oxidation (98, 99) belong to the metabolism first category but not to autotrophic theories, because they start from CH₄ as the first carbon source (methanotrophy) as opposed to CO₂ (autotrophy). Methane requires strong oxidants such as nitrogen oxides, O₂, or H₂O₂ to enter primordial metabolism (99), precluding a requirement for H₂ (98) hence electron bifurcation. Wächtershäuser explicitly excluded H₂ as the primordial source of electrons for metabolism, because of the uphill nature of CO₂ reduction with H₂ (100) under standard conditions. Smith and Morowitz (101) concluded that the site of life’s origin must have performed electron bifurcation; our findings identify a simpler and functional environmental precursor. Other proposals involve electrogenesis, using externally applied voltages on transition metal electrodes to generate potentials required to reduce CO₂ without H₂ (71). These theories miss the impact of pH on H₂ oxidation.

Conclusion

Iron-mediated, H₂-dependent Fd reduction provides the function of electron bifurcation in water at H₂ partial pressures, temperatures, and pH that occur naturally at serpentinizing hydrothermal vents. It requires solid-state transition metal surfaces, with no need for sulfide minerals, membranes, ion gradients, redox cofactors, externally applied voltage, or oxidants other than Fd, presenting a simple (one gas, one solid) and energetically smooth transition in physiological evolution from abiotic to enzymatic use of H₂ from serpentinization to reduce Fd (Fig. 7). These findings point to an early phase of biochemical evolution in which proteins (ferredoxin), hence ribosomes and the code existed in the last universal common ancestor (72), but metabolic electron acquisition from H₂ during autotrophic origins was still tied to transition metals in serpentinizing systems in the Earth’s crust. Acetogens, methanogens, other organisms that inhabit modern serpentinizing systems (26–30, 53, 74), and chemolithoautotrophs that fuel their acetyl-CoA pathway with electrons from H₂, native iron (64–66) or phosphite (102) might harbor additional biochemical relicts from the earliest phases of physiological evolution.

Methods

Reactions were prepared under anaerobic conditions in a glovebox (GS 79821, GS Glovebox-System), using N₂-washed 0.133 M or 1.33 M phosphate buffer in HPLC-grade water (pH 6, 7, 8.5 or 10; potassium phosphate monobasic and sodium phosphate dibasic (Honeywell Fluka)) containing Fd, pipetted into 5 mL glass vials (beaded rim) with metal powders (no powder in controls) and a polytetrafluoroethylene (PTFE)-coated stirring bar. Vials were sealed by a crimp cap with PTFE-coated membrane. To allow gas exchange, a syringe needle was placed through the crimp cap membrane before the vials were placed in the high-pressure reactor.

Ferredoxin.

Aliquots of lyophilized ferredoxin from C. pasteurianum prepared in E. coli as described (55) were suspended in 3.5 mL of N₂-washed phosphate buffer per reaction with 100 nmol Fd distributed throughout the experimental run (one replica for Ni_xFe and Fe₃O₄ and three replicas for all other reactions) and the control. 0.8 mL of the solution was placed on top of 0.1 mmol iron powder (Fe⁰, 99.9+% metals basis, particle size <10 μm, Alfa Aesar), cobalt powder (Co⁰, metal basis, particle size 1.6 μm, Alfa Aesar), nickel powder (Ni⁰, metal basis, particle size 3 to 7 μm, Alfa Aesar), molybdenum powder (Mo⁰, 100 Mesh, 99.95% metal basis, Alfa Aesar), magnetite (Fe₃O₄, 97% trace metal basis, particle size 50 to 100 nm, Sigma-Aldrich), or Ni_xFe (synthesis based on (30), kindly provided by Tuğçe Beyazay, Max Planck Institute for Coal Research, Mülheim, Germany). The remaining volume of the Fd stock was used as a pre-reaction control for UV-Vis spectroscopy. The ferredoxin and hydrogenase structures in Fig. 1 were rendered with Pymol (103).

Reaction.

After pressurizing the reactor with either 5 bar of Ar gas (99.998%, Air Liquide) or 5 bar of H₂ gas (99.999%, Air Liquide), reactions were started and regulated by a temperature controller (BTC-3000, Berghof Products + Instruments GmbH). Reactions were performed at 40 °C. The reaction time spanned from 1 h to 4 h. Afterward, reactors were depressurized under anaerobic conditions and the samples (metal powders and solution) were transferred to 2-mL Eppendorf tubes and centrifuged for 15 min at 13,000 rpm (Biofuge Fresco, Hereaeus). Supernatants were analyzed.

Reduction with Sodium Dithionite.

To obtain standards of Fd, the protein was reduced using sodium dithionite (ACROS Organics, Geel). Ferredoxin was diluted in a N₂-washed 133 mM phosphate buffer solution (pH 8.5). Sodium dithionite was prepared in a similar solution at pH 8.6. The reduction was carried out with two moles of sodium dithionite per one mole of ferredoxin. In a 3-mL UV quartz cuvette with a small PTFE-coated stirring bar at 400 rpm, reduced Fd was re-oxidized by air, measuring a full scan of the UV-Vis spectrum every 30 s until the protein was re-oxidized.

UV-Vis Spectrophotometry.

Spectra were recorded in an Agilent Technologies Cary 300 UV-Vis Compact Peltier spectrophotometer using UV-quartz cuvettes covered with a rubber stopper. A 200 to 800 nm scan was taken for each sample. Absorption of the [4Fe4S] clusters in Fd was normalized to the absorbance of the isosbestic point of the spectrum at 354 nm (56). Data were analyzed with the Cary UV Workstation and Cary WinUV software. Midpoint potentials of H₂ were calculated with the Nernst equation for the standard hydrogen electrode.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

PDF data have been deposited in HHU ResearchData (https://doi.org/10.25838/d5p-51) (104).