2.1 Affinity Calculations

Evaluating affinities requires writing overall reactions leading to the formation of proteins from the inorganic constituents of the mixed fluid. Take for example the overall synthesis reaction for the uncharged polypeptide chain of a cell-surface glycoprotein in M. jannaschii given by

(4)

This protein has 530 amino acid residues so the reaction coefficients on everything but the protein are increased by approximately that factor compared to the synthesis reactions for single amino acids (see below and Table S3). Equilibrium constants (K) for reactions of this type were evaluated for all 1,787 encoded proteins in M. jannaschii with the CHNOSZ software package (Dick, 2019) using estimates of standard Gibbs energies of formation of the proteins as functions of temperature and pressure (Dick et al., 2006); for these calculations, the pressure was set to 250 bar (Shock & Canovas, 2010). The group additivity scheme used for these estimates also predicts the charge on each protein as a function of T, P, and pH. Updated values for the methionine sidechain (LaRowe & Dick, 2012) and the glycine sidechain and protein backbone group (Kitadai, 2014) were used in these calculations; the latter update is associated with a more favorable (less endergonic) polymerization reaction compared to previous estimates (Dick et al., 2006). Standard Gibbs energies were also computed as a function of temperature and pressure for CO₂(aq), H₂(aq), H₂S(aq) (Shock et al., 1989), NH₄⁺, H⁺ (Shock et al., 1997), and H₂O (Shock et al., 1992). Activity products (Q) for the same set of reactions were calculated using the activities of inorganic species from the mixing calculations of Shock and Canovas (2010) and the activity of each protein set to 10⁻³. The resulting values of K and Q were combined with Equation 3 to compute affinities for the 1,787 M. jannaschii proteins.

2.2 Computational Uncertainties

Molecular crowding in cells is expected to have large effects on the activity coefficients of proteins (Minton, 2001). These effects are difficult to precisely quantify for all proteins coded by a genome, but as a rule of the thumb become larger with increasing molecular mass. The activity coefficient of a 100 kDa protein may reach 10² to 10⁴ in experiments with dextran gel used to simulate crowded conditions of cells (Minton, 2001). The activity of the proteins used in this study is 10⁻³. This is a conservative value, as it is higher than values used by other authors (e.g., 8.7 × 10⁻⁶ for the concentration of an average protein molecule in E. coli (Amend et al., 2013) and 10⁻⁹ for activities of biomolecules used to calculate Gibbs energy of biomass synthesis (LaRowe & Amend, 2016)).

Owing to its effect on Q in Equation 3, a higher value of the activity of the protein in Reaction (4) would be associated with a lower calculated affinity; that is, the overall synthesis reaction would become less exergonic. For purposes of illustration, we can consider an increase in activity of the protein of 10 orders of magnitude as an extreme example. Because the reaction coefficient on the protein is 1, this would result in an affinity difference of −10*2.303RT = −68.6 kJ (mol protein)⁻¹ at 85°C, the optimal growth temperature of M. jannaschii (Jones et al., 1983). This is much smaller than the range of calculated affinities for protein synthesis at this temperature, which is on the order of 1–100 MJ (mol protein)⁻¹ (see Figure 3a below). Therefore, even a potentially large uncertainty in the actual activities of proteins in cells has a relatively minor effect on the calculated affinities.

3.1 Affinities of Methanogenesis and Amino Acid Synthesis

A summary of calculated chemical affinities for autotrophic methanogenesis is shown in Figure 1a for conditions created as a 350 °C fluid typical of the Rainbow hydrothermal vent field on the Mid-Atlantic Ridge mixes with seawater. The compositions of endmember hydrothermal fluids, seawater, and mixed fluids taken from Shock and Canovas (2010) for Rainbow Field on the Mid-Atlantic Ridge and Endeavour Field on the Juan de Fuca Ridge are provided in Tables S1 and S2 in Supporting Information S1. As noted previously (McCollom, 2007; Shock & Canovas, 2010) the hydrothermal fluid at Rainbow contains less methane than equilibrium with its hydrogen and CO₂ contents requires. Although methanogenesis is thermodynamically feasible for the endmember hydrothermal fluid at Rainbow, its temperature is far higher than the currently known upper temperature limit of life, so the state of redox disequilibrium at lower temperatures in mixed hydrothermal fluid and seawater is of greater interest for understanding microbial ecosystems. Redox disequilibrium was accounted for in the mixing calculations of Shock and Canovas (2010) by allowing the reaction H₂(aq) + 0.5O₂(aq) = H₂O to go to equilibrium, but suppressing all other redox reactions, including that between CO₂ and CH₄. In this scenario, the affinities for methanogenesis and overall biosynthesis reactions are maximized at intermediate temperatures, around 50–150°C, depending on the host rock type of the hydrothermal system (Shock & Canovas, 2010). Therefore, it is not the hydrothermal fluid by itself that is responsible for driving methanogenesis, but the large disequilibrium that results from the mixing between hydrothermal fluid and seawater. This is of even greater importance for basalt-hosted systems like Endeavour, where methanogenesis is not favorable in the highest temperature fluids (unlike Rainbow), but becomes thermodynamically favorable in the mixing zone.

As shown in Figure 1a, once the 350°C fluid at Rainbow begins to mix with seawater, energy from methanogenesis is increasingly abundant even as the temperature of the mixture drops as more and more seawater is added. Ultimately, at very low temperatures attained through advanced extents of mixing, these calculations show that methanogenesis becomes thermodynamically unfavorable (the affinity for Reaction (1) becomes negative) as the oxidation state of the mixture becomes dominated by seawater. The solid curve in Figure 1a represents the affinity computed for a fixed activity of CH₄ (10⁻³), and the dashed curve indicates that computed for activities of CH₄ taken from the mixing calculations of Shock and Canovas (2010), which range from 10^−4.6 in seawater to 10^−2.6 in 350°C hydrothermal fluid. It can be seen that quite similar values of affinity are obtained whether the activity of CH₄ is taken from the mixing calculations or held constant. This occurs because of the overall reduced state of the mixed fluid and the large reaction coefficient on H₂; as a result, the activity of H₂ is the major control on the affinity of methanogenesis and on overall biosynthesis reactions (Shock & Canovas, 2010). Because of this, and the general lack of information about actual biomolecular concentrations in hydrothermal systems, it is reasonable to assign fixed values for activities of biomolecules in order to compare their potential for synthesis in the hydrothermal vent environment. The following calculations for amino acids and proteins use the same value of activity throughout (10⁻³).

These results can be understood by considering the interplay of changes in equilibrium constants and activity products as cold, oxidized seawater mixes with hot, reduced hydrothermal fluids. Even if pressure is held constant, which is appropriate for the zone of mixing between hydrothermal fluids and seawater, the composition-independent equilibrium constant is still a function of temperature and depends on changes in the standard Gibbs energies of formation of the compounds in the reactions. The temperature in a mixture of seawater and hydrothermal fluid depends on the relative proportions of the two fluids, which means that temperature and composition are explicitly linked (McCollom & Shock, 1997; Shock & Canovas, 2010). It follows that the composition-dependent activity product (Q in Equation 3) will also change with temperature.

Both negative and positive affinities for amino acid synthesis reactions are attained during fluid mixing as shown in Figure 1b. As an example, the reaction to form valine is given by

(5)

and analogous reactions for the 20 common amino acids are listed in Table S3 in Supporting Information S1. The affinity for these reactions can be evaluated with Equation 3 using equilibrium constants calculated with the revised Helgeson-Kirkham-Flowers equations of state (Dick et al., 2006; Shock et al., 1989, 1992, 1997) and activity products obtained from values of pH and the activities of CO₂(aq), NH₄⁺, and H₂(aq) returned by the mixing calculations, together with a choice for the activity of amino acids in this case arbitrarily set to 10⁻³. This is higher than the value of 10⁻⁶ used by other authors (Shock & Canovas, 2010) and represents a more conservative choice since the calculated affinities are lower (less positive). All of the curves in Figure 1b maximize with decreasing temperature, indicating that conditions in the undiluted hydrothermal fluid at Rainbow are not conducive to amino acid synthesis, but that mixtures with seawater are. Not all of the affinities for amino acids reach positive values in the mixture of hydrothermal fluid and seawater (Shock & Canovas, 2010). In the results shown in Figure 1b, affinities for glycine, serine, and asparagine stay negative, while those for aspartic acid barely attain positive values. In contrast, affinities for valine, leucine, tryptophan, and phenylalanine reach large negative values during mixing. Synthesis of these compounds would lead to the release of energy in the same way that methanogenesis (Reaction 1) releases energy at these conditions.

Affinities for both methanogenesis and biosynthesis are considerably lower in basalt-hosted hydrothermal systems, such as the Endeavour segment of the Juan de Fuca Ridge. The main reason for this is large differences in hydrogen concentration; the 345°C hydrothermal fluid at Endeavour has 0.62 mmolal H₂, compared to 16 mmolal H₂ in the 350 °C hydrothermal fluid at Rainbow (Shock & Canovas, 2010). Nevertheless, because of the redox disequilibrium that can occur during mixing with seawater, methanogenesis is predicted to be thermodynamically feasible between about 100–270°C (Figure 1a). On the other hand, the affinities for the synthesis of all amino acids during mixing at Endeavour remain negative at all temperatures (Figure 1c).

3.2 Carbon Oxidation States of Amino Acids and Proteins

Reasons why calculated affinities of various amino acids are so different during mixing of hydrothermal fluids and seawater include differences among the Gibbs energies of the amino acids and differences in the stoichiometric reaction coefficients of the various overall amino acid synthesis reactions, of which Reaction (5) is one example. In turn, the reaction stoichiometries reflect the average oxidation state of the carbon (Z_C) in the various amino acids given by (Dick & Shock, 2011)

(6)

where the subscripted n's represent the numbers of each element in a structural formula of an organic compound, and Z stands for the overall charge on the molecule. Values of Z_C for amino acids are shown along the abscissa of Figure 2a, where it can be seen that they range from −1.0 to 1.0. There are nine protein-forming amino acids with positive values of Z_C and nine with negative values of Z_C, as well as two with Z_C = 0. It can be anticipated that the Z_C values for proteins will fall somewhere in this same range depending on the stoichiometric abundances of each amino acid. Figure 1b reveals that the amino acids with negative values of Z_C tend to have the largest positive affinity values. So, just as reduction of carbon dioxide to methane by methanogens releases energy in seawater-hydrothermal fluid mixtures, so can the reduction of carbon dioxide and its reaction with ammonia to form amino acids, especially if the average oxidation state of carbon in the amino acid is relatively low.

The fact that many affinities for amino acid synthesis in ultramafic-hosted seafloor hydrothermal habitats are positive at temperatures ≤200°C means that hyperthermophilic autotrophs can synthesize these biomolecules with no net thermodynamic cost. This leads us to hypothesize that the overall synthesis of proteins is also accompanied by a release of energy; in effect that proteins represent a lower-energy configuration of C, H, N, O, and S than the inorganic forms of the same elements in mixtures of hydrothermal fluids and seawater. Testing this hypothesis starts with assembling protein compositions, which can be done by translating genomic data into protein sequences.

In this study, we evaluated sequences for the 1,787 proteins coded for in the genome of M. jannaschii. The relative abundances of the amino acids in these 1,787 proteins are plotted along the ordinate of Figure 2a. Note that some of the most abundant amino acids have the most negative values of Z_C. In addition, each sequence was converted into an elemental formula for each protein so that values of Z_C could be calculated and so that reactions from inorganic starting compounds could be written. The elemental abundances and Z_C values for all the proteins are listed in Data Set S1. A histogram of the Z_C values for all 1,787 proteins encoded in the genome of M. jannaschii is shown in Figure 2b. The mean value of Z_C is −0.250, and the distribution is asymmetric with a smaller peak at low Z_C, suggesting the presence of many proteins enriched in hydrophobic residues with negative values of Z_C. As shown below, the groups of proteins with high and low Z_C values have different predicted energetics of synthesis.

3.3 Chemical Affinities of Overall Protein Synthesis

Calculated affinities for protein formation become positive as mixtures of seawater and hydrothermal fluid in the Rainbow vent field reach temperatures <150°C (Figure 3a). At optimal growth conditions for M. jannaschii of 85°C (Jones et al., 1983), affinities of overall synthesis reactions analogous to Reaction (4) for all but one of the 1,787 encoded proteins are positive. Keeping in mind that temperature and composition are explicitly linked, these results show that the conditions generated during fluid mixing at ultramafic-hosted submarine hydrothermal systems are highly conducive to the formation of all of the proteins used by M. jannaschii. Many of the overall reactions at these conditions release tens of megajoules of energy per mole of protein, which scales with the overall size of the protein. It should be noted that decreasing the activities of the proteins from the value of 10⁻³ used here would move the curves up in Figure 3a, indicating greater affinities of overall synthesis for the lower concentrations that characterize rare and low-abundance proteins.

On a per-residue basis, proteins with relatively more negative values of Z_C tend to have more positive affinities (Figure 3b). This is completely analogous to the behavior of the affinities of amino acids depicted in Figure 1b. Furthermore, as Z_C becomes more negative, the amount of energy released correspondingly increases. The affinities of all of the proteins become negative at higher temperatures, in line with the trends for individual amino acids (Figure 1b). At about 120°C, the proteins are nearly equally split between those with positive and negative affinities. Above this temperature, overall protein synthesis under the conditions at Rainbow is generally unfavorable.

It can be useful to think of affinity as a surface whose height depends on both temperature and hydrogen activity; a mixing trajectory is a single curve on that surface. Endeavour and Rainbow are represented by two such curves. Whereas the affinities for protein synthesis at Endeavour remain below zero at all temperatures, those at Rainbow cross zero at about 120°C. Based on this energetic assessment, higher-temperature growth might be expected to be promoted by more H₂ than Rainbow has. This is a factor to consider when looking for life at higher temperatures than the current record of 122°C for the growth of a hydrogenotrophic methanogen in the laboratory (Takai et al., 2008). However, because of the greater expected range of possible vent fluid compositions than the examples considered here, the findings for Rainbow in the present study should not be taken as an independent prediction of the upper temperature limit of life.

3.4 Energetics of Amino Acid Synthesis Overcome the Polymerization Penalty

Although the overall protein synthesis reactions are highly exergonic in the mixed seawater–hydrothermal fluid between about 10 and 100°C, it should be noted that the polymerization of amino acids to form a polypeptide chain is endergonic over the temperature range considered here (e.g., Amend et al., 2013; Kitadai, 2014). The energetics for the overall protein synthesis reactions remain favorable because of the strong drive to synthesize amino acids, which releases more energy than the polymerization reactions require. Once again, it should be stressed that the actual mechanisms of biosynthesis have energetic costs in terms of ATP that are not measured here. For instance, the affinity of Reaction (4) calculated using activities of precursors in mixed seawater–hydrothermal fluid (Shock & Canovas, 2010) and 10⁻³ for the activity of the protein is 17.4 MJ/mol at 85°C. This corresponds to the sum of affinities for the formation of the amino acids and the polymerization reaction as follows. To calculate the affinity for the synthesis of all 530 amino acids in this protein (first arrow in the scheme below), the equilibrium constants for each amino acid reaction (Table S3 in Supporting Information S1) were computed from standard Gibbs energies of the amino acids and other species, and the activities of amino acids were set to 10⁻³ and combined with activities of other species for the mixed fluid at Rainbow (see above) to calculate the affinities. The affinity for each amino acid was multiplied by the frequency of that amino acid in the protein sequence and summed to produce the first value (37.8 MJ/mol). To calculate the affinity of polymerization (second arrow in the scheme below), the equilibrium constant was calculated from standard Gibbs energies of amino acids and of the protein using group additivity (Dick et al., 2006) and combined with activities of amino acids and the protein set to 10^–3 and unit activity of H₂O, resulting in a negative affinity (−20.4 MJ/mol).

The affinities in this example are calculated for non-ionized amino acids and protein and yield an overall affinity for protein synthesis of 17.4 MJ/mol. The ionization of the sidechain and terminal groups of the protein makes a relatively small positive contribution to the affinity at the temperature and pH value predicted from the mixing model so that the synthesis of the ionized protein from inorganic precursors has an overall affinity of 19.7 MJ/mol. The methods for calculating protein ionization state and its contribution to Gibbs energy were described previously (Dick et al., 2006) and were used for all the other calculations in this paper.

3.5 Broader Implications of Exergonic Biomolecular Synthesis

Taken together, the affinity values for the formation of amino acids and proteins reveal that fluid mixing at ultramafic-hosted submarine hydrothermal systems produces conditions that are conducive to the synthesis of biomolecules, whereas overall biosynthesis can be endergonic in basalt-hosted systems. Thus, the conditions in some highly reduced submarine hydrothermal systems are far more suitable for biosynthesis than more familiar conditions at the Earth's surface.

Autotrophic methanogens are, by definition, catalysts for methane production. They depend on the preexistence of conditions that are thermodynamically favorable for methane generation. They also depend on sluggish rates for the abiotic reduction of CO₂ (McDermott et al., 2015; Wang et al., 2018). Fluid mixing at submarine hydrothermal systems is one way in which such conditions are generated. Conventionally, autotrophs are thought to obtain the energy required to drive anabolism from the dissimilatory processes that they catalyze; these dissimilatory processes are often classified as energy metabolism in contrast to assimilatory processes that are biomass building (Fike et al., 2016). The results shown here indicate that the overall process of protein synthesis releases energy in some of the environments where thermophilic autotrophic methanogens live, and suggest that conditions may be found near ultramafic-hosted vents where entire microbial cells are thermodynamically stable relative to their external environment.

The distinction between high affinities of protein synthesis at Rainbow and negative affinities at Endeavour is directly linked to the supply of H₂, which provides the reducing power to drive the reactions forward. Similarly, it has been shown that the calculated Gibbs energies of total biomolecular synthesis in a model of E. coli cells can become negative under sufficiently reducing conditions with CO₂ as a precursor, but that the overall biosynthesis reactions are thermodynamically unfavorable under more oxidizing conditions (LaRowe & Amend, 2016). The thermodynamic calculations presented here suggest that the microbial ecology of basalt-hosted (relatively oxidized) and ultramafic (relatively reduced) vent systems should be vastly different. Because of the lower energetic requirements for biosynthesis in ultramafic systems, they could be populated by organisms with greater ranges of metabolic capabilities. This prediction based on energetic comparisons is supported by field observations that suggest increased bacterial, but not archeal, diversity is associated with higher-H₂ vent environments (Perner et al., 2007). Moreover, multiple studies have identified a preferential distribution of methanogens in Rainbow and other ultramafic-hosted hydrothermal systems, in contrast to Endeavour and other basalt-hosted systems where methanogens are relatively rare (Flores et al., 2011; Roussel et al., 2011; Ver Eecke et al., 2012; Zhou et al., 2009).

There are large stepwise energy costs, usually measured in numbers of ATP hydrolyzed, associated with the enzyme-guided reactions leading to the production of amino acids and proteins. Those costs cannot be assessed by the approach taken here, and remain in place regardless of the energetics of the overall synthesis reactions. This comparison reveals that the large magnitude of the overall release of energy during protein synthesis in ultramafic systems may be able to overstep the stepwise energy requirements, at least in principle. If so, in mixing zones of ultramafic-hosted submarine hydrothermal systems autotrophy is compelled by its habitat.

We acknowledge that our model for the energetics of overall biosynthesis does not capture the cellular processes that incur ATP costs. Nevertheless, the possibility for metabolic flexibility and consequent differences in ATP usage depending on the availability of metabolic precursors is well known for heterotrophic model organisms (Akashi & Gojobori, 2002). For autotrophs, which are even more dependent on the availability of inorganic starting materials, it, therefore, seems plausible that the ATP requirements for biomass synthesis would differ considerably between relatively oxidized and reduced submarine vent environments even for the same overall cellular composition.

Implications of these observations are diverse. Living in an oxidizing atmosphere, we expect that microbes consume energy-rich compounds in order to make energy-expensive biomolecules, so biosynthesis that releases energy to its surrounding environment is unfamiliar at least, and may challenge some assumptions. In habitats where biomolecules are actually more stable than the inorganic feedstocks that flow from the environment, catalysis could be decoupled from energy conservation. We should, at least, anticipate metabolic pathways in autotrophic methanogens that are unusual compared to heterotrophic microbes from near-surface environments.

One observation in line with this reasoning is that methanogenesis in many methanogens does not operate as a linear pathway, but as a cycle made possible by electron bifurcation that couples endergonic and exergonic steps of the cycle (Kaster et al., 2011). Furthermore, as intermediates are diverted from the methanogenesis cycle to be used for biosynthesis they are replenished via an anaplerotic mechanism that uses electrons only from H₂ (Lie et al., 2012). Therefore, the methanogenesis cycle can partially sustain biosynthesis in addition to ATP production. The possibility that reducing conditions permits an exergonic anabolism (without coupling to catabolism) was proposed previously (Smith & Morowitz, 2016). Our results provide theoretical support for such a possibility and therefore complement microbiological studies that reveal the types of mechanisms that may operate when energy is released by biosynthesis (Kaster et al., 2011; Lie et al., 2012).