Early bioenergetic evolution

Table 2 shows which genes those are, by annotation for the case of 106 genes that are present in at least one member each of 11/11 archaebacterial groups and in at least one member each of 20/30 higher prokaryotic taxa and that share more than 30 per cent amino acid identity.

The list depicts what sorts of functions might have been present as genetically encoded functions in the common ancestor, allowing generously for loss and keeping in mind that we cannot readily tell how much LGT has contributed to those gene distributions. With those caveats, the list tends to reflect a last common ancestor that had ribosomes, the genetic code, bits and pieces of cofactor biosynthesis, bits and pieces of amino acid biosynthesis, bits and pieces of nucleotide biosynthesis, and a fully fledged ATPase that could convert an ion gradient into chemically accessible high-energy bonds. The paucity of obvious components that would generate an ion gradient is striking. The list in table 2 is remarkably consistent with the view that the genes listed arose in an environment (an alkaline hydrothermal vent), where geochemically produced proton gradients existed and only had to be tapped. The presence of several tRNA modifying enzymes in the list is also congruent with our arguments about modified bases.

The simplicity of the acetogen and methanogen bioenergetic pathways, not involving quinones or cytochromes and being replete in FeS centres, as well as their occurrence in strictly anaerobic chemolithoautotrophs, tends to speak in favour of their antiquity. An alternative view [211] has it that menaquinone (MK)-based-dependent proton gradient generation is the ancestral state of membrane bioenergetics, with a good portion of that argument resting on the wide distribution of MK among archaebacterial and eubacterial cells, and the occurrence of several ion translocating enzymes of bioenergetic membranes in some archaebacterial and eubacterial lineages. There are several problems with that view, though. Three problems with the argument for the ‘antiquity based on ubiquity’ of MK is that (i) methanogens do not have it at all, except derived methanogens that have acquired the genes for MK biosynthesis from eubacteria [231], (ii) members of the sulfolobales and acidobales synthesize benzothiophenes (quinone derivatives), but have no known MK pathway [234], and (iii) the distribution of the MK biosynthesis pathways across archaebacteria is generally sparse (figure 7), whereby tests of the two possibilities underlying that sparse distribution, differential loss versus lineage-specific acquisitions, have not been reported. Another problem with the view that quinone-cytochrome-based ion-pumping systems are ancestral is that haems are lacking in all methanogens except the evolutionarily derived group of the methanosarcinales and that the archaebacteria haem biosynthesis pathway is unrelated to and arose independently from the eubacterial pathway [130,131]. We have plotted the distribution of the two haem and the two MK biosynthetic pathways across groups in figure 7.

Besides the exceptions mentioned earlier, an interesting observation is the absence of both MK pathways in some quinone-containing organisms as for instance thermotogae and tenericutes. This might suggest the existence of an additional MK biosynthetic pathway. Of course one could also argue that non-enzymatic haem synthesis, which does work [238], is the ancestral state used by the common ancestor. However, a more problematic aspect of the cytochromes first model is that, in that view, both methanogens and acetogens are seen as very late and highly derived evolutionary lineages [212], whereas isotope evidence has it that methanogens are ancient [164] and comparative physiology would have it that the core biochemistry of both methanogens and acetogens is ancient [44,73,160].

If our premise is correct that acetogens and methanogens that lack cytochromes and quinones are the most ancient kinds of cells, and that the primordial organisms did not have basically everything and could not do basically everything in the bioenergetic sense, a question arises, namely who invented cytochromes and respiratory chains? Some acetogens have cytochromes, but at least those of Moorella thermoacetica [239] are lateral acquisitions, probably from eubacterial sulfate reducers, as sequence comparisons readily reveal. Some methanogens also have cytochromes but they are also lateral acquisitions [231]. The deltaproteobacterial sulfate reducers have regular cytochromes b, as well as multi-haem cytochromes c [172] synthesized via the alternative archaeal haem pathway. Sulfate reducers also have MK, but they also have some things in common with methanogens in terms of Hdr-related proteins. Sulfate reducers do not, however, have membrane integral cytochrome bc or cytochrome b6f complexes [172], and it is not yet clear how they use their MK pool, although a participation of electron confurcation, as discussed earlier, seems possible [197]. The circumstance that sulfate reduction is mainly a cytosolic process and that it likely involves flavin-based electron bifurcation (or confurcation) speaks for its antiquity. Indeed, given that there was surely no lack of sulfur substrates for sulfate reducers [179] at the vent–ocean interface, autotrophic sulfate reducers that use the acetyl-CoA pathway could, in principle, harbour a physiology that is nearly as ancient as that of acetogens and methanogens, requiring merely the additional evolutionary invention of cytochromes and quinones. The links between sulfur metabolism and methanogenesis point to the antiquity of both [163].

The prokaryotes who invented cytochrome bc type containing respiratory chains probably also invented the Q-cycle [240], the membrane phase and quinone-based analogue of flavin-based electron bifurcation, which is a cytosolic process so far. Among prokaryotes, flavins are far more universal than quinones. This suggests a possible sequence of events in the early evolution of energy conservation: (i) thioester-dependent substrate-level phosphorylations, (ii) chemiosmotic harnessing and the universality of ATP as energy currency, (iii) harnessing of Na+ gradients generated by H+/Na+ antiporters, (iv) flavin-based bifurcation-dependent ion gradient generation, (v) quinone-based (and, eventually, Q-cycle containing) proton gradient generation involving membrane integral cytochrome complexes and bona fide respiratory chains. All of these processes ultimately depend, even today, upon CO2 reduction with low-potential Fdred (generated either chemosynthetically or photosynthetically), placing a reaction of the type ‘reduced iron → reduced carbon’ at the beginning of bioenergetic evolution, as outlined in figure 8. The evolutionary advent of quinones affected chemiosmotic pumping efficiency in at least two ways: (i) via quinone-dependent electron bifurcation in the Q cycle, both in cytochrome bc1 and in cytochrome b6f complexes [243], and (ii) in complex I. The recent structure of complex I reveals how redox-dependent quinone movement introduces conformational changes across several adjacent antiporter subunits, causing them to pump in concert [244]. This generates a greater H+/2e– stoichiometry than the hydrogenase-related precursors of complex I which, despite a similar subunit composition [202], pump without the help of quinones.

Here, we have specified some links between the energy releasing chemistry of a Hadean alkaline hydrothermal vent and the energy metabolism of particular groups of modern microbes. Energetic aspects of the origin of life are less widely discussed than RNA-oriented genetic aspects, and one can approach the problem either from a very general perspective, without linking early energetic models to modern bioenergetic configurations, or from a more specific perspective that aims to forge tangible connections between ancient chemical environments and modern microbial physiology. The latter approach requires stating some premises about what is ancient in modern biochemistry, and the further one delves into the physiology of acetogens and methanogens that lack cytochromes, the more ancient they look. Moreover, the more closely we compare their chemistry to processes at alkaline hydrothermal vents, the stronger the similarities become, and in terms of their antiquity the sulfate reducers lag not far behind. The circumstances that the acetogens and methanogens share the Wood–Ljungdahl pathway, the most ancient of six CO2 fixation pathways known [73], that they have cytochrome- and quinone-lacking forms (the only such groups among chemiosmotic prokaryotes known), that they are replete with FeS proteins [50,60]) as seen in figure 5, that they live in habitats hardly different from those on the early Earth [11,30], that their energy metabolism is centred around flavin-based electron bifurcation [143]—the soluble precursor of the Q-cycle—and that the energetic pillar of their energy metabolism is low-potential-reduced ferredoxin, are all best interpreted in our view as evidence reflecting their ancient and primordial nature.

From the standpoint of energetics, it appears possible that life could have evolved from gases (H2, CO2, CO, N2, H2S, SO2) that reacted, with the help of transition metals, at the solid catalyst phase to produce aqueous organic compounds. Current views on the changes in free energy of biological systems, which are necessarily negative [14,29,39], as well as changes in entropy, which are close to zero [245,246], are compatible with that view. The antiquity of anaerobic chemolithoautotrophs seems as evident today as it did 40 years ago [44]. The ubiquity of chemiosmotic coupling as the ultimate source of net energy conservation [151] throughout the microbial world has, by comparison, come as a surprise. In the search for life's start, it is that peculiar energetic attribute of living things—the strict dependence upon chemiosmotic coupling—that makes alkaline hydrothermal vents [9] special. They go a long way towards closing the gap between acetogens, methanogens and the elements on early Earth.

We thank Rolf Thauer, Wolfgang Buckel, Manuela Pereira and Lee Hansen for comments on the draft, Bettina Siebers and Christiane Dahl for helpful discussions, and Re-Young Yu for help with preparing the manuscript. For funding, N.L. thanks the Leverhulme Trust, J.F.A. thanks the Leverhulme Trust for research grant F/07 476/AQ, I.C.P thanks PTDC/QUI-BIQ/100591/2008 funded by FCT, Portugal and FEDER program, and W.F.M. thanks the European Research Council.