Hydrothermal vents are often divided into two types — black and white smokers — but off ridge systems like Lost City belong to neither category because they do not emit “smoke.” Both the smoke of black and white smokers comes from metal sulfide and calcium and barium precipitation, but white smokers vent slower and at slightly lower temperatures (220–330°C) than black smokers, which is why a greater percentage of the metals precipitates in the chimneys and is not vented out causing the white color of the smoke, Lost City type vents have carbonate chimneys but no smoke (Rona et al., 1986; Kelley et al., 2005). A map showing the geographical distribution of some vents discussed here is shown in Figure 1.

Black smokers are hydrothermal fumaroles with abundant sulfides usually located close to or directly on top of the spreading zones of mid-ocean ridges. Because of their proximity to magma, the water in them can be up to over 400°C hot. They are also characterized by a low pH, even though black smokers that are hosted in ultramafic rocks can host serpentinization, such as the Rainbow and Logatchev hydrothermal fields on the mid Atlantic ridge (MAR) (Schrenk et al., 2013). These serpentinizing black smokers, in contrast to Lost City type vents, have higher temperatures, higher metal concentrations and an acidic pH (3–4), but unlike hydrothermal vents hosted in basaltic rock, they have high hydrogen and methane concentrations like other serpentinizing systems. The Rainbow and Logatchev systems are the best studied vents of this type. They also vent C2–C5 hydrocarbons in trace amounts (Charlou et al., 2002; Schrenk et al., 2013). Serpentinization does not happen in basalt-hosted vents because of the higher silica content of the basaltic rock (McCollom and Seewald, 2013).

Off ridge vents, or Lost City type vents, are characterized by carbonate chimneys, a high effluent pH (9 to 11) and moderate temperatures (between 28 to 116°C) compared to black or white smokers (Proskurowski et al., 2008; Seyfried et al., 2015). They are mainly represented by the first system of its kind discovered: the Lost City hydrothermal field. Lost City is located 750 m below sea level, but 4 km above the sea floor on the Atlantis massif (Kelley et al., 2005). Submarine systems that are comparable to Lost City are Prony Bay in New Caledonia (Monnin et al., 2014), the Old City hydrothermal field in the Indian ocean (Lecoeuvre et al., 2021) and the Shinkai Seep Field on the Mariana arc (Okumura et al., 2016).

Prony Bay is in much shallower seawater than Lost City (only 16 to 47 m below the surface) and its water source is fresh water, not crustal circulated seawater as in Lost City (Kelley et al., 2001; Quéméneur et al., 2014; Postec et al., 2015; Twing et al., 2022). Other important differences between shallow seawater vents and deep-sea vents are the access to sunlight, the influence of tidal forces and the different pressures (Price and Giovannelli, 2017). However, in terms of effluent parameters, Prony Bay is more similar to Lost City than the black smokers or the continental serpentinizing systems. It has elevated concentrations of H₂ and CH₄, a high pH and low venting temperatures (see Table 1) (Monnin et al., 2014).

The main difference in effluent chemical composition between the high temperature black smokers that serpentinize like the Rainbow and Logatchev hydrothermal fields and the low temperature Lost City hydrothermal field, is that Rainbow and Logatchev have higher concentrations of CO₂ and metals and acidic pH in their effluent whereas Lost City has a high pH and is depleted in metals and CO₂ (Proskurowski et al., 2008).

Continental serpentinizing systems have comparatively cool effluent under 100°C, and they also have hyperalkaline waters, although they are fed by fresh water (usually called meteoritic water, which is a complicated name for rainwater or snow) rather than altered sea water. Numerous continental serpentinization sites are known (Etiope et al., 2011b; Etiope and Schoell, 2014; Etiope, 2015). They are important analog sites to deep-sea serpentinizing sites because of their chemical similarities, and because they are easier to access. Like submarine serpentinizing systems, they also produce large amounts of H₂ and CH₄. They are the main source for abiotic gas on Earth (Etiope, 2017). Abiotic volatile hydrocarbons are still poorly understood but they can be generated by various inorganic mechanisms, including FTT. At some sites the amount of abiogenic methane can reach up to 90 vol% of the gas content (Etiope and Schoell, 2014), most of the remainder being H₂.

Abiogenic methane is widely discussed in the search for extraterrestrial life because methane is considered a potential biomarker, the sign of methanogenesis. But if Earth can generate large amounts of methane through abiotic processes, other wet, rocky planets and moons probably can as well, making methane itself problematic as a sign of life (Etiope, 2017). However, if biochemistry and life arose from exergonic H₂-dependent CO₂ reducing reactions, then abiotic methane serves as a proxy for the existence of geochemical conditions that are conducive to life’s origin: exergonic reactions of H₂ and CO₂ under continuously far from equilibrium conditions.

The distinction between biotic and abiotic methane in usually made by isotope fractionation (Etiope et al., 2013; Etiope and Schoell, 2014). Abiotic methane is enriched in ¹³C relative to biotic methane, which is isotopically light because of isotope discrimination in favor of ¹²C in the acetyl-CoA pathway (Blaser et al., 2013) of methyl synthesis from H₂ and CO₂ underpinning methanogenesis. However distinguishing biotic and abiotic methane by carbon isotope fractionation is not always straightforward (Horita and Berndt, 1999), which is why deuterium fractionation is also often taken into account (Etiope et al., 2011a) and often even combined (Miller et al., 2018) as shown in a study conducted in the Samail Ophiolite where it could be determined that biotic and abiotic methane mix (Nothaft et al., 2021).

Radiocarbon methods (¹⁴C) can also be used to determine the source of C in CH₄ from serpentinizing systems. In the upper atmosphere, ¹⁴C is constantly produced by neutron capture of ¹⁴N to generate ¹⁴C and a proton. The neutrons stem from cosmic radiation, the resulting ¹⁴C has a half-life of 5,700 years. There is no radiocarbon in CH₄ of Lost City effluent. That is, CH₄ from Lost City is “radiocarbon dead,” meaning that the carbon source cannot be bicarbonate from seawater, unless the circulation time exceeds 57,000 years (ten ¹⁴C half-lives) because seawater contains ¹⁴C that is constantly synthesized in the upper atmosphere. It is estimated that the entire volume of the ocean circulates through hydrothermal vents every 100,000–500,000 years (Fisher, 2005). The low CO₂ concentrations at Lost City together with its “radio-carbon-dead” CH₄ suggest that CH₄ in Lost City is likely derived from CO₂ of mantle origin that was reduced during serpentinization (Proskurowski et al., 2008).

Most of the carbon emitted from Lost City comes from the Earth’s mantle in the form of formate (Lang et al., 2012, 2018). Formate can be used as a substrate for methanogenesis in environments where CO₂ is limiting. Methanogens can readily convert formate into CO₂ and H₂ or use it to reduce coenzyme F₄₂₀ to F₄₂₀H₂, a flavin similar to FAD (Holden and Sistu, 2023). This allows them to access formate as a source of carbon and electrons for methanogenesis. This has been observed not only in Lost City but also in the Samail ophiolite (Fones et al., 2021). Formate forms abiotically under the high pH, reducing conditions of serpentinizing fluids. This can also be replicated in the laboratory (Preiner et al., 2020), where formate is typically the most abundant reduced carbon species under a variety of conditions (Belthle et al., 2022; Beyazay et al., 2023a,b). In Lost City, formate is the second most abundant carbon species after CH₄ and the second most available reductant after H₂ (Lang and Brazelton, 2020; Brazelton et al., 2022).

High concentrations of effluent H₂ are characteristic for serpentinizing systems, as summarized in Table 1. In terms of H₂ concentration, the non-serpentinizing Strytan hydrothermal field has the lowest out of all the systems where H₂ was detected. It is in the range of normal seawater. At first glance it might seem as if continental serpentinizing systems do not have high concentrations of H₂, however it should be mentioned that sometimes only the liquid phase was measured, not the actual gas phase emitted from the systems. In the Tekovira and Zambales ophiolite, where the gas phase composition was measured, H₂ can comprise 45 vol% of the emitted gas. High H₂ tends to correlate with high CH₄, but not strictly (see Supplementary Figure S1). Lost City and other serpentinizing systems of the mid Atlantic ridge (MAR) have high concentrations of CH₄ and H₂. The effluent at Prony Bay, a shallow sea vent with a fresh water source and similar microbial communities as continental ophiolites, has lower H₂ and CH₄ concentrations than Lost City of about a magnitude but still in significant amounts (Quéméneur et al., 2023). The Ashadze system on the MAR has very high H₂ concentration in its venting fluids of up to 26.5 mM. For CH₄ concentrations, many deep-sea hydrothermal vents, with a few exceptions, have a concentration of 1–3 mM. In continental serpentinizing sites CH₄ can reach between 20 vol% in New Zealand to 93 vol% of the gas phase in the Tekovira ophiolite (Etiope and Schoell, 2014). CH₄ is also extensively studied in continental serpentinization sites because they are the main contributor to abiotic gas (Etiope, 2017).

Alkaline serpentinizing systems are low in CO₂ because at high pH dissolved CO₂ turns into carbonate (CO₃^2–), which precipitate within the vent as Ca²⁺ and Mg²⁺ carbonates. This is the reason why hot and acidic hydrothermal vents such as Rainbow or Lucky Strike have very high CO₂ concentrations (up to 28 mM), while alkaline systems such as Lost City and Prony Bay harbor only have trace amounts of dissolved inorganic carbon. For continental serpentinization sites CO₂ was only detected in trace amounts in the gas phase. In the liquid phase CO₂ [often in the form of bicarbonate (HCO₃^–)] could sometimes be detected up to the mM range, but trace amounts are typical (Table 1). More complex organics than CH₄, when reported, were only detected in trace amounts both in ocean and continental serpentinization systems. Formate and acetate, when reported, were detected in Lost City in μM amounts and in the Conical Seamount of the Mariana Forearc in mM (formate) and μM (acetate) amounts, whereby the acetate at Lost City is likely biogenic (Lang et al., 2010).

Comparing different serpentinizing systems across studies is not straightforward because the sampling methods are often different and because of local heterogeneity: even in the same system values can differ greatly depending on where in the system the samples were taken. Cook et al. (2021) showed that the values measured in different samples, be it the temperature, the pH or the concentrations of the different compounds, can be vastly different depending on the sampling site. For example, if the groundwater, the well water or the surface water were sampled at three different sites — Tablelands, the Cedars, Aqua de Ney — the pH values could range between pH 8–9 in the surface water to pH 12–13 in the well water.

Although CO₂ was very abundant on the early Earth (Sossi et al., 2020), modern serpentinizing systems are generally low in CO₂ because of magnesium carbonate and calcium carbonate precipitation at high pH. The effluent of serpentinizing systems can however be rich in formate (HCOO^–). Formate is readily converted by simple inorganic catalysts (Preiner et al., 2020; Belthle et al., 2022; Beyazay et al., 2023a,b) and by microbes into H₂ and CO₂ via the near-equilibrium reaction:

with ΔG_o’ = –3.5 kJ·mol^–1 (Maden, 2000). Microbes have a number of different enzymes for that reaction. The formate to CO₂ converting reaction is catalyzed by the enzyme hydrogen-dependent CO₂ reductase, HDCR (Schuchmann and Müller, 2013), or by NADH-dependent formate dehydrogenase (FDH), which is the typical enzyme of the acetyl-CoA pathway in acetogens (Schuchmann and Müller, 2013). In methanogens the reaction is catalyzed via hydrogenases that convert H₂ into reduced ferredoxin and the activity of formyl-methanofuran dehydrogenase, which contains a FDH domain and 46 [4Fe4S] clusters (Wagner et al., 2016). A third kind of formate dehydrogenase, HylABC-Fdh2, performs a reaction called electron bifurcation in which an electron pair is split, with the individual electrons being transferred to different acceptors with different redox potentials (Buckel and Thauer, 2013; Buckel and Thauer, 2018). HylABC-Fdh2 converts formate into CO₂ with one of the electrons being transferred to NAD⁺ and the other being transferred to ferredoxin (Wang et al., 2013). Oxidation of two formate molecules via this reaction generates two CO₂, one NADH and two reduced ferredoxins. This bifurcating form of FDH is found in Bipolaricaulota, a deeply branching acetogen identified by metagenomics in the effluent of the hyperalkaline water of the Oman ophiolite (Colman et al., 2022). There are also cytochrome- dependent formate dehydrogenases (Sebban-Kreuzer et al., 1998). All formate dehydrogenases known so far, including HDCR, use Mo (in molybdopterin) or W (in tungstopterin) as a cofactor. The enzymatic mechanism of FDH is thought to involve binding of formate to Mo (or W) via its oxygen atoms (Hartmann et al., 2016), the abiotic mechanism is thought to involve bonding of the carbon atom in formate to Ni or Fe (Preiner et al., 2020).

The use of formate as a carbon source is a common theme in the effluent of serpentinizing systems. Although organisms have not yet been cultivated from hydrothermal vents that exhibit growth on formate, acetogens (Moon et al., 2021) and methanogens (Ferry et al., 1974) are known that use formate as their sole carbon and electron source, and some methanogens use formate as their sole carbon source (Costa et al., 2010; Lie et al., 2012). All indications are that formate based primary production is taking place at the submarine system of Lost City (Lang et al., 2018) and in some terrestrial serpentinizing systems including the Oman ophiolite (Colman et al., 2022), the Cedars (Suzuki et al., 2018), at Hakuba Happo (Nobu et al., 2023) and in the Prony Bay system, a freshwater fed submarine serpentinizing system that vents in seawater at a depth of about 50 m (Frouin et al., 2022), because there is no other carbon source reported to fuel primary production in these systems. For acetogens and methanogens living in these systems, formate is apparently the source of CO₂ and a source of electrons for primary production via the acetyl-CoA pathway, which generates reduced carbon to support the growth of fermenters that live from microbial cell mass (Schönheit et al., 2016) and other heterotrophs in the system.

The results of microbial sampling studies from different sites are not simple to compare directly because community composition within sites can vary, and at very small scales, depending on the specific position of the sample within the vent, effluent seawater mixing in the case of submarine vents, presence or absence of O₂ in the specific sample, the methods of analysis and other factors. For example, in the hyper-alkaline Strytan site, which has alkaline effluent but is not serpentinizing, anaerobes and the Wood Ljungdahl pathway of CO₂ fixation dominate in inner regions of vents while O₂-tolerant pathways dominate at the vent surfaces in contact with sea water (Twing et al., 2022). In the basement of Lost City alone, a broad diversity of phylotypes across sites sampled is observed (Brazelton et al., 2022). Methanogens are common in serpentinizing systems and can access carbon as formate (Fones et al., 2021). Quéméneur et al. (2023) showed that the abundance of certain prokaryotes is positively correlated to the H₂/CH₄ ratio in hyperalkaline springs and that bacteria that use H₂ as their main energy source are the most abundant among the microbial communities of Prony Bay.

A recent study by Frouin et al. (2022) stands out among surveys of microbial communities from different serpentinizing systems because the authors aimed to identify community and physiological properties that i) are shared among different serpentinizing systems and that ii) distinguish serpentinizing systems from hydrothermal vents in non-serpentinizing systems. They found that serpentinizing systems tend to be colonized by diverse communities, and that bacterial groups encountered often include acetogens (Firmicutes) and methanogens like Methanosarcinales (Archaea), as originally discovered at Lost City. Frouin et al. (2022) found that Prony Bay was dominated by the acetogen Bipolaricaulota, which was also the main inhabitant of the Oman ophiolite community recently characterized by Colman et al. (2022) on the other side of the globe. Abundance of Firmicutes, which includes many acetogens, has been shown to be proportional to dissolved H₂, however why that is still has to be determined (Quéméneur et al., 2023).

A common physiological theme in serpentinizing systems is the ubiquitous presence of enzymes for the acetyl-CoA pathway (or Wood-Ljungdahl pathway) of CO₂ fixation. The Wood-Ljungdahl pathway is considered to be the most ancient pathway of carbon fixation (Fuchs and Stupperich, 1985; Fuchs and Stupperich, 1986; Berg, 2011; Fuchs, 2011). It is still the only exergonic pathway of CO₂ fixation known. The reaction to the level of the energy-rich thioester

is exergonic with ΔG_o′ = –59.2 kJ/mol if 2[H] = H₂ and slightly endergonic with ΔG_o′ = +13.2 kJ/mol if 2[H] = NADH (Fuchs, 1994). The NADH forming reaction is slightly exergonic in the direction of CO₂, which is one reason that the pathway is so versatile among microbes (Fuchs, 1994; Zinder, 1994). It is the only pathway of CO₂ fixation that does not require net ATP input (Fuchs, 2011) and that occurs in both bacteria and archaea (Berg et al., 2010). It is a joined pathway of carbon and energy metabolism in which acetogens and methanogens obtain their ATP from ion gradients that they generate in the process of the exergonic reduction of CO₂ with electrons from H₂ at MtrA-H in methanogens (Thauer et al., 2008) and at Rnf in acetogens that lack cytochromes (Müller et al., 2018). In addition, acetogens and some methanogens can generate ATP in the acetyl-CoA pathway using substrate-level phosphorylation via acetyl-phosphate (Schöne et al., 2022). This kind of combined carbon and energy metabolism is so far unique to acetogens and methanogens. The acetyl-CoA pathway can be a dedicated pathway of CO₂ fixation (carbon metabolism) with no role in energy metabolism (ATP synthesis), for example in sulfate reducers that grow autotrophically, or in bacteria with phosphite-dependent energy metabolism (Mao et al., 2021). The acetyl-CoA pathway is a biochemical fossil that links acetogens and methanogens via functional, ecological and ancient evolutionary aspects (Martin, 2020; Schöne et al., 2022).

The diffusible carrier electrons for reduction reactions in serpentinizing systems, H₂, is present in overabundance. At Lost City the effluent H₂ concentration is on the order of 10 mmol/kg [0.5 to 14 mmol/kg: (Proskurowski et al., 2008)], which is about 5–6 orders of magnitude more H₂ than is required to support methanogenesis (Thauer et al., 2008). Because of the high alkalinity and high H₂ concentration, the calculated redox potential of serpentinizing effluent can be in the range of –800 mV (Boyd et al., 2020), sufficient for most biologically relevant reduction as shown in Figure 2, with the exception of the phosphate/phosphite couple, where the midpoint potentials become more negative with increasing pH (Bernhard Schink, pers. comm.). Alkalinity favors the dissociation of H₂ into protons and electrons by consuming protons from the reaction H₂ → 2e^– + 2H⁺. Electrons from H₂ enter metabolism via hydrogenase. All hydrogenases except the Fe hydrogenase of methanogens (Thauer et al., 2008; Shima et al., 2011), reduce 4Fe4S clusters and most of them generate reduced ferredoxin as the reaction product. The redox potential needed to reduce ferredoxin as the source of electrons in metabolism is on the order of –500 mV. Suzuki et al. (2018) found no hydrogenases in metagenomes from The Cedars, raising the possibility that other routes of ferredoxin reduction might be possible under such reducing conditions. Studies of microbes that grow on native iron and on electrodes suggest that side activities of extracellular enzymes might be able to substitute for hydrogenases (Deutzmann et al., 2015). Early studies by Menon and Ragsdale (1996) indicated that acetyl-CoA synthase and pyruvate synthase have latent hydrogenase side activity, and the H₂-producing side reaction of nitrogenase (Hu and Ribbe, 2016) is well known. All of those side activity hydrogenases have FeS or NiFeS clusters that are likely candidates for the moonlighting active site. Metals (Fe, Ni, Co) and metal oxides can serve as hydrogenases, activating H₂ by splitting it into two metal bound hydrogen atoms for CO₂ reduction (Preiner et al., 2020; Belthle et al., 2022; Beyazay et al., 2023a,b) or for NAD⁺ reduction (Henriques Pereira et al., 2022).

The highly reducing conditions of serpentinizing hydrothermal vents are key to origins because CO₂ has to be converted to organic carbon. There are, however, also alternative views about the source of energy at origins in serpentinizing hydrothermal systems. Duval et al. (2021), for example, argue that the energy required for the origin of metabolism was not exergonic H₂-dependent CO₂ reduction catalyzed by reduced transition metals as it occurs in the laboratory (Weiss et al., 2016; Preiner et al., 2020; Wimmer et al., 2021b; Beyazay et al., 2023a) and in acetogen and methanogen metabolism, but that instead the energy at origins stemmed from nitrate-dependent methane oxidation with the mixed valence iron oxide fougerite (green rust), acting as the catalyst (Duval et al., 2021) in alkaline hydrothermal systems. How energy release from methane oxidation is coupled to either prebiotic organic synthesis or carbon metabolism is not explained by Duval et al. (2021), changes in Gibbs free energy for the fougerite dependent reactions have not been presented, and abiotic laboratory versions of the fougerite dependent carbon assimilation reactions have not been reported. In laboratory versions of the acetyl-CoA pathway (Preiner et al., 2020; Beyazay et al., 2023a) and in acetogens and methanogens in vivo (Figure 3), energy for ATP synthesis is released in the exergonic H₂ dependent synthesis of acetate, pyruvate and methane from CO₂.

Frouin et al. (2022) looked carefully at sources and pathways for nitrogen but found no consistent trend. The source of nitrogen is still not well resolved in serpentinizing systems and might differ across systems. In very recent work, Shang et al. (2023a,b) found that experimental serpentinization reactions in the laboratory readily reduce N₂ to NH₃ in very large amounts and on time scales approaching 30 days. Cells are about 50% carbon and about 10% nitrogen by dry weight (Heldal et al., 1985), hence there has to be access to a nitrogen supply for primary production in serpentinizing systems. The simplest solution is that N₂ present in the water circulating through serpentinizing systems is reduced by nitrogenase. Most studies have detected nitrogenase genes among the microbiota of serpentinizing systems, though nitrate reduction can also be detected (Frouin et al., 2022). Trutschel et al. (2022) reported high concentrations of ammonia at Ney Springs. Ammonia has also been reported in the Oman ophiolite system (Rempfert et al., 2017), but in both cases the source of the ammonia is not resolved, with decaying organic matter within the system being the most widely discussed source. With the new report by Shang et al. (2023a), abiotic sources of NH₃ need to be considered in natural systems (Shang et al., 2023a).

The state of nitrogen in the primordial atmosphere was N₂ (Sossi et al., 2020). In an origins context, before the existence of biological N₂ fixation via nitrogenase, the source of nitrogen would have to be geochemical. Three sources are discussed: High pressure high temperature N₂ reduction by deep geochemical Haber-Bosch like processes (Brandes et al., 1998), low pressure low temperature FeS-mediated N₂ reduction processes (Dörr et al., 2003) and serpentinization-related geochemical processes. Ménez et al. (2018) reported the presence of abiotic tryptophan and other amino acids in the Atlantis massif that hosts the Lost City hydrothermal field. The exact route of synthesis is not yet known but might involve iron-dependent Friedel-Crafts like syntheses. The significance of amino acid synthesis in the crust for theories that life arose at hydrothermal vents is evident (Baross, 2018).

Lost City does not appear to emit abiotic amino acids in its effluent, but the terrestrial hydrothermal system Hakuba Happo does. Nobu et al. (2023) recently reported the presence of glycine (5 nM) in the effluent of Hakuba Happo which is likely of abiotic origin. Notably, there were no other amino acids in the effluent. Were glycine of biotic origin, the 19 other amino acids should be present. Some of the microbes that Nobu et al. characterized from Hakuba Happo, ca. Lithacetigenota, possessed genes for glycine reductase, an unusual selenoenzyme that synthesizes acetyl phosphate from glycine and P_i (Andreesen, 2004). The acetyl phosphate can be used for acetyl-CoA (carbon metabolism) or ATP synthesis via substrate level phosphorylation. Although the concentration of glycine in the Hakuba Happo effluent was low, it was sufficient to render the glycine reductase pathway thermodynamically favorable under the conditions of the vent. No archaea were detected in Hakuba Happo, but glycine reductase utilizing Lithacetigenota were detected at the Cedars (Nobu et al., 2023).

The glycine reductase variant of carbon assimilation and energy metabolism employed by Lithacetigenota starts with environmental glycine. It is distinct from the H₂-dependent glycine reductase pathway of autotrophic CO₂ fixation used by Desulfovibrio desulfuricans (Sánchez-Andrea et al., 2020). The findings of Ménez et al. (2018) indicate that geochemical nitrogen fixation to amino acids is possible. The findings of Nobu et al. (2023) suggest that it is ongoing today, continuously in the hyperalkaline serpentinizing environment of Hakuba Happo.

Sulfur is a trace element in cells. It is highly volatile in various oxidation states and readily eluted from rocks as H₂S or HS^– during serpentinization, but sulfate is also a common constituent of effluent and sulfate reducers are very common in serpentinizing systems (Reveillaud et al., 2016; Lang et al., 2018). Sulfur is an ancient substrate of energy metabolism (Rabus et al., 2006; Liu et al., 2012) and fulfills important catalytic functions in many cofactors (Kirschning, 2021). It is a component of ancient metabolism. Note however that the synthesis of formate, acetate and pyruvate from H₂ and CO₂ under simulated hydrothermal conditions using Ni₃Fe or Fe₃O₄ as catalysts requires no participation of sulfur (Preiner et al., 2020; Belthle et al., 2022; Beyazay et al., 2023a,b).

Today, nitrogen and sulfur compounds that occur in hydrothermal systems serve as terminal electron acceptors in anaerobic respiratory chains (Calisto and Pereira, 2021). At the origin of metabolism, CO₂ was arguably the most important electron acceptor because the synthesis of organic compounds was required in order to get metabolic pathways established. This underscores the evolutionary significance of the acetyl CoA pathway in acetogens and methanogens: it generates the starting compounds of metabolism while releasing energy that can be harnessed to synthesize ATP. In that sense, acetogenesis and methanogenesis are cytochrome-free and quinone-free anaerobic respirations that use CO₂ as the terminal acceptor in a reaction sequence that also generates a net gain of acetyl CoA. Anaerobic respirations that reduce sulfur and nitrogen compounds require siroheme, cytochromes and quinones, acetogenesis and methanogenesis do not, they are simpler in that regard. Coupled carbon and energy metabolism in the acetyl-CoA pathway using CO₂ as the electron acceptor probably preceded a differentiation into dedicated carbon metabolism and dedicated energy metabolism involving cytochrome dependent respiratory chains that use nitrogen and sulfur compounds as terminal acceptors.

Phosphate has always presented problems in prebiotic chemistry because of its low solubility in contact with calcium to generate apatite, which is more or less insoluble, leading to various suggestions, that certain kinds of carbonate rich lakes existed on the primordial Earth that maintained phosphate in solution (Toner and Catling, 2020). The recent findings of Frouin et al. (2022) from Prony Bay place the phosphate problem in a fundamentally different light. They estimate that about half of the microbes that they sampled from Prony Bay possess the genes required for the metabolism of phosphonates and they found that phosphonate metabolism was fairly common in microbial communities from serpentinizing systems. What are phosphonates? Phosphonates are compounds that contain a C–P bond and are so far only known to be synthesized by microbes via a pathway that starts from phosphoenolpyruvate. Phosphonates can have concentrations in open ocean water that account for up to 25% of dissolved organic phosphorus (Acker et al., 2022). Their exact function is still not known, but they might represent a means of sequestering phosphorus among community members (Metcalf et al., 2012). That proposal has however recently been challenged because the phosphonate synthesis and utilization genes are so common: phosphonates might also mediate biological interactions at the cell surface (Acker et al., 2022). The question posed by Frouin et al. (2022) is: What are microbes from serpentinizing systems doing with a high frequency of phosphonate metabolizing enzymes?

Moreover, in some of the Prony Bay metagenomes, the gene for an additional enzyme, NAD⁺-dependent phosphite oxidoreductase, ptxD, was often inserted into the phosphonate metabolizing operon (Frouin et al., 2022). The ptxD gene product converts phosphite and NAD⁺ to phosphate and NADH (Vrtis et al., 2001). Frouin et al. (2022) noted the ability of Prony Bay microbes to utilize reduced phosphorous compounds. They were mainly concerned with the question of what the source of phosphonates in the Prony Bay fluid might be. While about half of the genomes from the Prony Bay vents contained genes related to phosphonate metabolism, metabolizing operons represent a frequency that is significantly higher than average for marine microbes (Frouin et al., 2022), it is not orders of magnitude higher: up to 30 % of marine microbial genomes have the potential for phosphonate degradation (Acker et al., 2022).

New findings by Pasek et al. (2022) suggest that the phosphite oxidase gene, ptxD, that Frouin et al. find inserted into the phosphonate (phn) operon might hold an important clue. Why is phosphite relevant? Two reasons: First, phosphite is much more soluble than phosphate. One solution to the “phosphate problem” of early evolution is that phosphorus in the ancient oceans was present in the form of the much more soluble phosphite ion (Schwartz, 2006). Studies by Pasek and colleagues have suggested that phosphite might have been more prevalent on the early Earth than it is today. Archaean oceans may have contained substantial amounts of phosphite in addition to phosphate (Pasek et al., 2013; Herschy et al., 2018). If phosphite was present in early environments but phosphate was needed, enzymatic phosphite oxidation would have been needed. Might the presence of phosphonates in the ocean, with P⁺³ oxidation state (like phosphite) and oxidation to phosphate (P⁺⁵) in the cytosol, reflect a relic of ancestral marine environments? It is possible.

Second, where does phosphite in the environment come from such that so many microbes should have genes for phosphite metabolism (Schink et al., 2002)? In older papers, a source of ancient phosphite was originally suggested to be schreibersite from meteorite bombardments (Pasek et al., 2013). Schreibersite is a highly reactive phosphide mineral (Ni, Fe)₃P, that is indeed only known from meteorites (Pasek, 2008). But phosphite oxidation genes are commonplace today (Schink and Friedrich, 2000; Vrtis et al., 2001), being found in roughly 1.5 % of all genomes (Ewens et al., 2021), whereas huge schreibersite spreading impacts are rare at best. This clearly indicates that there are sources of phosphite in the environment that do not require bolide impacts.

The midpoint potential for the reduction of phosphate to phosphite is –690 mV at pH 7 and 25°C (Schink et al., 2002) (see Figure 2) but becomes more negative with higher pH (Bernhard Schink pers. comm.). Considering the measured and calculated reduction and midpoint potentials for serpentinizing systems is within the range of –435 to –830 mV, the reduction of phosphate to phosphite might not be within the range covered by H₂ in modern serpentinizing systems, […] but might lie within the range of serpentinization processes that generate H₂ on early Earth. This brings us to a question relevant for modern microbial communities and for life’s origins: do serpentinizing systems actually generate phosphite? The answer appears to be yes (Pasek et al., 2022), although phosphite has not been reported to be present directly in the effluent of serpentinizing systems. Pasek et al. (2022) measured phosphorus species in samples of rock from formations altered by serpentinization, they found that 20–50 % of the total P in some samples was phosphite (the remainder was phosphate). That corresponds to a substantial amount of phosphite stemming from serpentinization. The implications of i) genes for phosphite oxidation being present in microbes from serpentinizing systems (Frouin et al., 2022) and ii) phosphite being a component of rock in serpentinizing hydrothermal systems (Pasek et al., 2022) suggest that there is phosphite available in serpentinizing systems and that microbes that live there are using it. Could they be using this as a soluble source of P for enzymatic phosphate synthesis? If so, this would tie together some loose ends in phosphate metabolism and early evolution.

It is possible that genes for phosphite uptake and conversion into its biologically more useful form, phosphate, might reflect an ancient, possibly even ancestral state of phosphorus metabolism. As Buckel (2001) put it: “At the time of the origin of life, about 3.8 billion years ago, phosphites could have been more important than today.” This would mean that the highly exergonic conversion of phosphite into its biologically relevant form, phosphate, occurred inside cells. But in early chemical evolution, before there were enzymes, was there a role for phosphite? This has not yet been extensively studied. Incorporation of phosphate into the fabric of early protobiochemical synthesis still presents a few sticking points in early metabolic evolution (Martin, 2020), environmental phosphite offers some alternative entry routes (Buckel, 2001).

Finally, the relevance of phosphite is underscored by the circumstance that it is a growth substrate for microbes, specifically for anaerobic autotrophs (Schink and Friedrich, 2000). Schink et al. (2002) characterized Desulfotignum phosphitoxidans, that grew autotrophically with phosphite as the sole electron donor and CO₂ as the sole electron acceptor. The energy-yielding metabolic reaction was the synthesis of phosphate with some acetate being produced as well. D. phosphitoxidans can also grow as a sulfate reducer that uses the acetyl-CoA pathway. In the absence of sulfate, its metabolism was that of an acetogen with energy generated by CO₂ reduction to acetate but using phosphite instead of H₂ as a source of electrons (Schink et al., 2002). Another phosphite oxidizer was recently isolated, but it was an autotrophic phosphite specialist — the strict anaerobe grew as a phosphite-dependent autotroph, using only phosphite as the electron donor and only CO₂ as the electron acceptor (Mao et al., 2021).

Buckel (2001) suggested that the highly exergonic process of phosphite oxidation might be coupled to substrate level phosphorylation, which is thermodynamically possible, but has not yet been shown so far. Phosphite in the environment appears to be derived from serpentinization. It is the only reductant other than H₂ that is known to fuel chemolithoautotrophic growth using the acetyl-CoA pathway. This links phosphite to serpentinization and ancient acetogenic physiology. Phosphite has a sufficiently negative midpoint potential that electron bifurcation would not be needed to reduce ferredoxin in cells that use phosphite, but its enzymatic oxidation product is NADH, which would require electron bifurcation for ferredoxin reduction.

There are only two basic ways that cells conserve energy as high energy phosphate bonds: substrate level phosphorylation (Decker et al., 1970) and chemiosmotic energy harnessing using a rotor-stator ATP synthase (Walker et al., 1982; Walker, 2013). Serpentinizing systems provide a chemical environment that can support the origin of both forms of biological energy conservation.

Substrate level phosphorylation usually involves the generation of an acyl phosphate (or enol phosphate) bond during the oxidative breakdown of reduced carbon compounds. The classical example is the synthesis of the acyl phosphate bond in 1,3-bisphosphoglycerate at the reaction catalyzed by glyceraldehyde-3-phosphate dehydrogenase by oxidizing the aldehyde group on C1 of glyceraldehyde-3-phosphate with NAD⁺ to form a thioester bond between the enzyme and the substrate, which is cleaved via phosphorolysis, generating the acyl phosphate bond at C1 which can phosphorylate ADP to generate ATP. There are also reductive routes of SLP, for example the conversion of the thioester bond in acetyl-CoA from the (reductive) acetyl-CoA pathway to acetyl phosphate, which can also phosphorylate ADP. Although the direct synthesis of acyl phosphates from thioesters has so far not been reported, it has long been known that acyl phosphates can phosphorylate ADP without enzymes (Kitani et al., 1991). More recently, Whicher et al. (2018) have shown that thioacetate can readily react with P_i to form acyl phosphates without enzymes, although thioacids are not known in central carbon or energy metabolism. The synthesis of acetate and pyruvate from H₂ and CO₂ under serpentinizing vent conditions using catalysts synthesized at hydrothermal vents — awaruite and magnetite — is facile whereby there is furthermore enough energy released in the synthesis of acetate to energetically support SLP (Preiner et al., 2020; Belthle et al., 2022; Beyazay et al., 2023a,b). Reductive SLP also takes place via acetyl phosphate synthesis at the glycine reductase reaction (Andreesen, 2004) and the subsequent phosphorylation of ADP. Otherwise, net ATP synthesis involves either SLP via oxidation reactions of organic compounds (Müller, 2003; Martin and Thauer, 2017), which are thermodynamically unfavorable under the reducing conditions of serpentinizing systems, or it involves chemiosmotic ATP synthesis using the rotor-stator ATP synthase.

The rotor-stator ATP synthase is universal among cells. It generates ATP by the passage of protons (or Na⁺) from the outside of the cell through the ATP synthase to the inside of the cell. The passage of protons through the ion channel of the stator subunit a and the subsequent protonation and deprotonation of an acidic residue in each of the c subunits of the ring-shaped rotor sets the rotor of the ATP synthase in rotation, inducing conformational changes in the catalytic subunits of the “head,” changes that forge a phosphoanhydride bond between ADP and P_i and lead to release of ATP from the enzyme. Three ATP molecules are synthesized per complete 360° rotation of the head, this requires roughly 12 protons traversing the enzyme. This mechanism of ATP synthesis is universal in bacteria and archaea, it is furthermore reversible such that cells can expend ATP that they obtain from SLP to generate ion gradients for the import of small molecules (Franklin, 1986) or to generate reduced ferredoxin (Buckel and Thauer, 2013). The ion gradients that modern cells use to power chemiosmotic ATP synthesis are generated by membrane integral ion pumping proteins, or coupling sites, that pump protons (or Na⁺) from the inside of the cell to the outside. There are two ways that pumping can be achieved: i) with the help of conformational changes within membrane-integral domains, and ii) with the help of what Mitchell called vectorial chemistry (Mitchell, 1991). In the former, the general mechanism (conformational change) is the same as that for the ATP synthase, except that the energy for conformational change and pumping stems from exergonic redox reactions that are part of the overall bioenergetic reactions of the cell, for example methane synthesis from H₂ and CO₂ (in methanogens) or acetate synthesis from H₂ and CO₂ (in acetogens). The second, evolutionarily more advanced mechanism (vectorial chemistry) requires the existence of quinones or quinone analogs (Berry, 2002), membrane soluble redox factors that can accept an electron pair and two protons on the cytosolic side and transfer the electron pair to an acceptor on the outside of the cell, thereby transferring protons from the inside of the cell to the outside merely by the relative position of hydride donor and hydride acceptor across the membrane. Ancient lineages of methanogens and acetogens lack quinones. They generate their ion gradients solely via conformational changes in membrane proteins (pumping).

It is a very curious observation that the principle of harnessing ion gradients for ATP synthesis and the enzyme that does it, the rotor-stator ATP synthase, are as universally conserved across all bacteria and archaea as the ribosome and the genetic code itself. This indicates that ion gradient harnessing was present in LUCA. LUCA could use ion gradients. Yet there is no similar form of conservation to indicate that LUCA could generate ion gradients. But if LUCA arose and existed at a serpentinizing vent, as much other evidence summarized above indicates, there is also no need to assume that LUCA had to generate ion gradients by itself, because serpentinization generates geochemical ion gradients continuously and for free that remain stable for the duration of the serpentinization process. Recalling that Lost City has been actively serpentinizing for roughly 100,000 years (Ludwig et al., 2011; Denny et al., 2016), a stable geochemically generated pH gradient of the right polarity (alkaline, lower proton concentration on the inside) and sufficient magnitude (cells synthesize ATP with the help of a pH difference of about one pH unit) (Tran and Unden, 1998; Silverstein, 2014) would present a continuous environmental setting in which natural chemiosmotic potential existed and could be harnessed (Martin and Russell, 2007; Martin, 2012).

This natural chemiosmotic energy source would have allowed the ancestral ATP synthase to function but would have required the evolution of protein based pumping mechanisms in the ancestors of the archaea and the bacteria in order for LACA (Last Archaeal Common Ancestor), (a methanogen) and LBCA (Last Bacterial Common Ancestor) (an acetogen) to make the transition to the free-living state. Methanogens without cytochromes pump with energy derived from H₂-dependent CO₂ reduction. The pumping reaction takes place at the methyl transferase step catalyzed by the membrane protein MtrA-H, which in a cobalamine-dependent reaction transfers a methyl group from a nitrogen atom in methyl-H₄MPT to a sulfur atom in coenzyme M; the free energy of the reaction ΔG_o′ is –30 kJ·mol^–1 and pumps two Na⁺ ions from the inside of the cell to the outside (Thauer et al., 2008). This energy conserving reaction is universal among H₂-dependent methanogens and likely represents their ancestral state. In acetogens without cytochromes, the pumping reaction is energetically driven by H₂-dependent CO₂ reduction and is catalyzed by the membrane protein Rnf, which pumps Na⁺ from the inside of the cell to the outside while transferring electrons from reduced ferredoxin to NAD⁺ (Biegel et al., 2011) with a ΔG_o′ of –34.7 kJ·mol^–1 (Kuhns et al., 2020).

By coupling H₂-dependent CO₂ reduction to ion gradient formation, cells were able to generate their own ion gradients, were no longer dependent on geochemical ion gradients for ATP supply and could make the transition to the free-living state. The machinery to generate ion gradients by coupling exergonic reactions of H₂-dependent CO₂ reduction to ion pumping arose independently in the lineages leading to LBCA and LACA, because different steps were harnessed and unrelated enzymes were used. Note that both acetogens and methanogens require flavin based electron bifurcation to generate reduced ferredoxin for CO₂ reduction (Buckel and Thauer, 2018; Müller et al., 2018), but this is not required in CO₂ reduction promoted by metal catalysts (e.g., Ni₃Fe, Fe₃O₄) because of the very low midpoint potentials for H₂ oxidation that are generated by very high pH (Martin, 2022). A summary of energy inputs from serpentinization into the origin of biochemistry is given in Figures 3,4.