Respiratory Heme A-Containing Oxidases Originated in the Ancestors of Iron-Oxidizing Bacteria

Abstract

Respiration is a major trait shaping the biology of many environments. Cytochrome oxidase containing heme A (COX) is a common terminal oxidase in aerobic bacteria and is the only one in mammalian mitochondria. The synthesis of heme A is catalyzed by heme A synthase (CtaA/Cox15), an enzyme that most likely coevolved with COX. The evolutionary origin of COX in bacteria has remained unknown. Using extensive sequence and phylogenetic analysis, we show that the ancestral type of heme A synthases is present in iron-oxidizing Proteobacteria such as Acidithiobacillus spp. These bacteria also contain a deep branching form of the major COX subunit (COX1) and an ancestral variant of CtaG, a protein that is specifically required for COX biogenesis. Our work thus suggests that the ancestors of extant iron-oxidizers were the first to evolve COX. Consistent with this conclusion, acidophilic iron-oxidizing prokaryotes lived on emerged land around the time for which there is the earliest geochemical evidence of aerobic respiration on earth. Hence, ecological niches of iron oxidation have apparently promoted the evolution of aerobic respiration.

Introduction

Aerobic respiring organisms contain terminal oxygen reductases for energy metabolism at different oxygen levels in the environment. The evolutionary origin of these oxygen-consuming enzymes is unknown (Castresana and Saraste, 1995; Pereira et al., 2001; Han et al., 2011; Sharma and Wikström, 2014). A common class of terminal oxidases is heme A-containing proton pumping cytochrome oxidase (COX), which has a relatively low affinity for oxygen (Han et al., 2011; Degli Esposti et al., 2019) and consequently must have evolved during or after the Great Oxygenation Event (GOE), which produced stable levels of oxygen on earth (Konhauser et al., 2011). COX of the family A of Heme Copper Oxygen Reductases (HCO) is widespread in all kingdoms of life (Castresana and Saraste, 1995; Pereira et al., 2001; Han et al., 2011; Degli Esposti, 2020) due to extensive Lateral Gene Transfer (LGT) (Nelson-Sathi et al., 2015; Soo et al., 2017). Subsequently bacterial COX became the cytochrome c oxidase of mitochondrial organelles. The prokaryotes that initially evolved COX have remained elusive (Pereira et al., 2001; Soo et al., 2017; Degli Esposti et al., 2019; Matheus Carnevali et al., 2019) and, consequently, the origin of aerobic respiration is a major unresolved problem. To address the question of COX origin, we have studied the phylogeny of two proteins specifically required for COX biosynthesis and performed new and expanded analyses on the phylogeny of the major protein subunit of COX. The latter analysis is focused on family A oxidases since the phylogenetic distribution of family B oxidases, which also have heme A, is rather patchy among bacteria. Indeed, it is possible that family B derived from some ancestral family A oxidase (Degli Esposti, 2020). Combined with other findings, our results suggest that family A COX has evolved in ancestors of extant acidophilic iron-oxidizing bacteria.

COX in mitochondria and aerobic bacteria is a multi-protein intrinsic membrane complex with several metal prosthetic groups. Subunit I, COX1, contains two heme A molecules (Figures 1A,B), denoted cytochrome a and a₃ in the assembled enzyme, and one copper atom, Cu_B. COX2 contains two copper atoms in a binuclear center, Cu_A (Figure 1A). The redox reaction of molecular oxygen reduction at the heme a₃-Cu_B center is coupled to the generation of an electrochemical gradient via conserved proton-conducting channels (Pereira et al., 2001; Ferguson and Ingledew, 2008; Han et al., 2011; Sharma and Wikström, 2014; Degli Esposti, 2020). This gradient drives ATP synthesis and various other energy-demanding functions in the cell.

FIGURE 1

Biosynthesis of COX requires multiple proteins that catalyze the formation of heme A, or are involved in the insertion of the metal cofactors and overall assembly of the enzyme in the membrane (Figure 1A). Heme A as a prosthetic group is only found in respiratory oxygen reductases (Hederstedt, 2012). The biosynthesis of heme A from protoheme IX (heme B) involves two enzymes (Figure 1A; Mogi et al., 1994; Hederstedt, 2012). First heme O synthase, CtaB/Cox10/CyoE, catalyzes formation of heme O and then heme A synthase, CtaA/Cox15, converts heme O into heme A. Synthesis of heme A Figure 1B requires ambient oxygen levels (Brown et al., 2002) that were attained in primordial earth only after the GOE. In bacteria, the genes for COX proteins and those for heme A synthesis and COX assembly factors show modularity, i.e., the genes are frequently clustered in the chromosome (Figure 1C). Based on the assumption that COX proteins and protein factors specifically required for COX biosynthesis have coevolved, we analyzed the phylogeny of bacterial CtaA to obtain insights on how COX has evolved. The results prompted us to establish a revised expanded classification of heme A synthases and develop a scheme for their evolution based upon phylogenetic data, which was compared to that of COX1 and the COX-specific assembly factor CtaG to define the likely origin of COX.

Materials and Methods

Phylogenetic Analysis

Database searches and genome scanning were conducted by iterative BlastP (Basic Local Alignment Search Tool for Proteins) searches as detailed in the Supplementary Material. Briefly, wide searches expanded to 5000 hits were usually performed with the DELTABlast program using the BLOSUM62 substitution matrix (Boratyn et al., 2012). Integrated searches were expanded in granular detail to recognize established conserved domains of the (super) family to which it may belong (as shown in the NCBI protein website)¹ (Lu et al., 2020) preferentially using BlastP and PSI-BLAST searches restricted to 500 hits. Classification of HCO has been undertaken using the bioinformatic classifier available at http://www.evocell.org/hcoevocell.org (Sousa et al., 2011).

Maximum likelihood (ML) trees were initially produced with the MEGA5.2 program to elaborate BLAST outputs and preliminary phylogenetic frameworks (Degli Esposti et al., 2020), with a variety of substitution models (generally the WAG model for multitopic membrane proteins and the Dayhoff or BLOSUM62 substitution matrix for membrane-anchored, predominantly water-exposed proteins, and discrete gamma distribution of four categories to account for evolutionary rate differences among sites and lineages, allowing some sites to be evolutionarily invariable (Tamura et al., 2011). The MEGA program was also used to transform ML trees into ultrametricized trees similar to BEAST tree outputs using the condensed option, with a cut-off of 50% bootstrap support.

Phylogenetic inference was routinely undertaken using either Bayesian or ML probabilistic approaches. We reconstructed ML trees using the more sophisticated IQ-Tree program (Nguyen et al., 2015), normally from its server http://iqtree.cibiv.univie.ac.at/ (Trifinopoulos et al., 2016) with the ultrafast bootstrap option of 1000 replicates and the LG model as in previous publications (Degli Esposti et al., 2019; Spang et al., 2019). Additionally, we used mixture models of amino acid substitution, such as C20 (Le et al., 2008) and EX_EHO (Le and Gascuel, 2010). We found that the EX_EHO model tended to produce trees with stronger support than those obtained with the LG model for CtaA and COX1 proteins, but not with caa3_CtaG proteins, probably because of their limited conservation. ML trees were reconstructed also with the program PhyML 3.0 (Guindon et al., 2010), and generally run from the online platform http://www.phylogeny.fr/index.cgi using the WAG model and the Shimodaira–Hasegawa (SH)-like statistical support analysis. We used the program FigTree 1.4.4² for visualizing trees obtained with various methods.

We preferentially used the BEAST 2.6.2 package for Bayesian phylogenetic inference (Drummond et al., 2006) because of its information-rich outputs (Lee et al., 2013; Drummond and Bouckaert, 2015). Although predominantly utilized for divergence studies of animal species with genes including mitochondrial COX1-3 (Lee et al., 2013; Ratnasingham and Hebert, 2013), BEAST analysis has been applied also to study the molecular evolution of bacterial redox proteins (Khadka et al., 2018; Bouckaert et al., 2019). Manually curated alignments were first loaded into the BEAUti 2 app of the package to prepare.xml files containing the detailed settings for the phylogenetic analysis run with the BEAST program (Bouckaert et al., 2019). Routinely, such settings included: four gamma categories with shape 0.3 (or other empirically estimated values) and proportion of invariant sites from 0.01 to 0.1, depending on the protein and its taxonomic sampling; BLOSUM62 or WAG as substitution models; Relaxed Clock Log model and the gamma option for the Yule birth model (other priors were left in their default setting); at least 2 million iterations for the Markov chain of Monte Carlo analysis with 5 pre-burnin. Trees were stored every 1000 iterations and then usually reduced by 10% burnin to generate Maximal Clade Credibility (MCC) output files with minimal posterior value of 0.1 using the TreeAnnotator app (Drummond and Bouckaert, 2015). MCC trees were graphically elaborated with the FigTree program, which produced a wealth of quantitative data for the various branches (Lee et al., 2013; Khadka et al., 2018), or visualized using the program DensiTree 2 (Bouckaert and Heled, 2014; Drummond and Bouckaert, 2015). In a few cases, we also used the program MrBayes v.3.0b4 as in reference (Degli Esposti et al., 2020).

Protein Sequence Analysis

Given that the quality of phylogenetic trees heavily depends upon the accuracy of the protein sequence alignments upon which they are reconstructed, we performed an in depth analysis of the sequence variation of each protein to guide its proper alignment. This analysis was undertaken by exploiting the versatility of the MEGA programs after importing whole sequences or alignments of multiple sequences downloaded directly from the BlastP searches. An initial alignment of the protein was built with a minimal set of 20 sequences using either the ClustalW or the MUSCLE algorithm within the MEGA5 program (Tamura et al., 2011). The alignments thus obtained contained several gaps that were often unnecessary to maximize for local sequence similarity, as verified by visual inspection; such gaps were then removed along manual refinements conducted with iterative rounds of implementation that were aided by the inclusion, whenever possible, of protein sequences for which 3D information is currently available. The alignments were then progressively enlarged to include sequences that were representative of different prokaryotic taxa in which the protein was found, with additional refinements to accommodate local sequence variations. Short residue gaps that were needed to properly align a single sequence were subsequently deleted along detailed manual refinements of local sequence similarity and congruent hydropathy profile.

The enlarged alignments thus refined were used to build phylogenetic trees encompassing most major molecular variants, as well as the overall taxonomic distribution of any protein studied. Some of the alignments are available upon request. Sequences that produced long branches or displayed high substitution rates were identified by statistical analysis as described in Supplementary Material. Such sequences were subsequently removed, often substituted by sequences clustering in the same subclade that did not display equivalent branch aberrations. Then the set of aligned sequences for a given protein was reduced to simplify tree presentation without altering the tree topology found with larger alignments.

The CtaA, caa3_CtaG and COX1 proteins studied here have multiple membrane-spanning segments (TM). Consequently, we have applied extensive hydropathy analysis to all proteins analyzed, using both the TMpred server https://embnet.vital-it.ch/software/TMPRED_form.html and the TMHMM Server v. 2.0 https://hsls.pitt.edu/obrc/index.php?page=URL1164644151. These programs utilize complementary methodologies that help define the ends of predicted TM (Möller et al., 2001). This sequence analysis was combined with the available 3D structural information for COX subunits (Iwata et al., 1995; Svensson-Ek et al., 2002), and CtaA (Niwa et al., 2018), to define the TM regions and other topological features in distant protein homologs, as in the case of proteins from iron-oxidizers. Membrane topology was graphically rendered with the program TOPO2³ and then used as a platform for building the protein models.

FIGURE 2

Other Methods

Genome completeness was evaluated as described previously (Degli Esposti et al., 2019), or using information available in the GTNB database (Parks et al., 2018).

Results and Discussion

Iron-Oxidizing Bacteria Appear to Have an Ancestral Form of Heme A Synthase

Heme A synthase is required for the biosynthesis of the characteristic heme prosthetic groups in COX. The heme A synthase enzyme protein belongs to the super-family of Cox15-CtaA, cl19388, members of which are widespread among prokaryotes (He et al., 2016). The 3D structure of Bacillus subtilis CtaA has been determined by X-ray diffraction crystallography (Niwa et al., 2018) and that of Aquifex aeolicus analyzed by cryogenic electron microscopy (Zeng et al., 2020). B. subtilis CtaA has 8 TM and two extended extracellular loops, ECL1 and ECL3, each of which contains a conserved pair of cysteine residues linked by a disulfide bond (Lewin and Hederstedt, 2016; Niwa et al., 2018; Figure 2A). Overall, the CtaA protein comprises two nearly superimposable 4-helical bundle domains, with the C-terminal domain binding a b-type heme group via two conserved histidine residues, while the N-terminal domain seems to contain the catalytic site for conversion of heme O into heme A (Figure 2).

Many members of the Cox15-CtaA superfamily lack cysteine pairs in extracellular loops, for example Rhodobacter capsulatus CtaA (Figure 2B) and the eukaryotic Cox15 homologs. These variants have been called type 2 to distinguish them from type 1 proteins represented by B. subtilis CtaA (He et al., 2016). In recent explorative work on CtaA sequences (Degli Esposti et al., 2020) we found more variants, as presented in the expanded classification system (Table 1) and in the unrooted phylogenetic tree of Supplementary Figure 1 (left). We also discovered a potential root for the phylogeny of CtaA in the Domain of Unknown Function 420 (DUF420) that has four TM. CtaM of Staphylococcus aureus, Bacillus anthracis and B. subtilis contain the DUF420 domain and are required for assembly of active COX, but dispensable for heme A synthesis (Hammer et al., 2016; von Wachenfeldt et al., 2021).

Our comprehensive new classification of heme A synthases (Table 1) is based upon the integration of the molecular features found in all variants present in current NCBI protein resources, including recently added metagenome data, and the consistent phylogenetic pattern of the variants that emerged from multiple approaches of phylogenetic analysis (detailed in the Supplementary Material). The Bayesian tree shown in Figure 3A represents a condensed view of the current phylogenetic distribution of CtaA proteins in prokaryotes, encompassing diverse representatives not only of type 1 and 2, but also members of two variants that we defined recently (Degli Esposti et al., 2020). One such variant is a derivative of type 1, called type 1.5. The other variant could not be fitted in either type 1 or type 2 and hence is called type 0 (Table 1). The type 0 CtaA proteins are very divergent in sequence but contain the invariant residues that are considered crucial for activity of heme A synthases (Hederstedt, 2012) (Figure 2B) and their genes are often present at the end of either the rus operon encoding COX subunits (Figure 1C) or the cytochrome bo₃ operon of acidophilic Fe²⁺-oxidizers (Appia-Ayme et al., 1999; Quatrini et al., 2009; Issotta et al., 2018).

FIGURE 3

Type 0 CtaA consistently forms the basal branch in phylogenetic trees comprising all types of CtaA proteins (Figures 3A, 4 and Supplementary Figures 2–5). This pattern suggests a scheme for the molecular evolution of CtaA proteins (Figure 3B), which fundamentally follows the branching order of Bayesian (ultrametric) trees obtained from comprehensive alignments of the bacterial variants of CtaA. We hypothesize that the gene for a four TM DUF420 protein might have been duplicated and fused to form the gene for the ancestor of all current CtaA proteins (Figure 3B). Type 0 CtaA may constitute the extant descendant of such an ancestral protein, from which the type 2 and type 1 CtaA branched off. This branching might have occurred simultaneously, forming the sister clades that are frequently observed in phylogenetic trees (Figures 3, 4 and Supplementary Figure 3; He et al., 2016; Degli Esposti et al., 2020), or in a rapid sequence of differentiation, in which the cysteine pairs of current type 1.1 CtaA might have been acquired gradually, as suggested by some Maximum Likelihood (ML) trees (Supplementary Figure 4). The latter possibility is sustained by the occurrence of type 1.0 CtaA proteins having only the first cysteine pair and a long ECL1 comparable to that of type 2 CtaA (Table 1). These proteins are clearly different from type 1.0 proteins such as that of S. aureus, which most likely derive from a recent loss of the cysteine pair in ECL3, since they retain the same extracellular loops and closely cluster with type 1.1 CtaA proteins of related taxa such as B. subtilis (Table 1, Figures 3, 4 and Supplementary Figures 3–5). Similarly, type 1.5 CtaA likely derives from secondary losses of both cysteine pairs from type 1.1 proteins of Proteobacteria. As known from experiments with B. subtilis CtaA, the two cysteine residues in ECL1, but not those in ECL3, are important for heme A synthase activity (Lewin and Hederstedt, 2016) and the size of ECL3 can be changed without loss of enzyme activity (Lewin and Hederstedt, 2008).

FIGURE 4

In summary, our detailed analysis [see Supplementary Material for the evaluation of potential problems arising from Long Branch Attraction, LBA (Brinkmann et al., 2005; Philippe et al., 2005; Bleidorn, 2017)] strongly suggests that type 0 CtaA constitutes the ancestral form of heme A synthase, or a relative thereof (Figure 3B). Next, we explored the current taxonomic distribution of these proteins that, following our rationale, would define extant taxa possessing early divergent forms of COX.

Taxonomic Distribution of Ancestral CtaA and Ecological Niches for Bacteria With This Type of CtaA

Blast searches with type 0 CtaA protein sequences against the entire nr database produced significant hits with closely related proteins and a few outgroup proteins. Currently there are about 100 recognized type 0 CtaA proteins, the majority of which is coded by the genomes of prokaryotes that share niches of acid soil or hydrothermal environments (Quatrini and Johnson, 2018; Supplementary Figure 6). The taxa of Proteobacteria with acidophilic iron- and sulfur-oxidizing physiology (Acidithiobacillus, Acidiferrobacter and Acidihalobacter spp.) contain the gene for type 0 CtaA as the sole heme A synthase. Intriguingly, several Acidithiobacillus species that lack the rus operon with COX genes also have type 0 CtaA orthologs (Acuña et al., 2013; Issotta et al., 2018). In these taxa, which include At. caldus, At. sulfuriphilus and At. thiooxidans, the gene for type 0 CtaA is usually found at the end of operons encoding orthologs of cytochrome bo₃ ubiquinol oxidases. Identification of cytochrome a in the membranes of At. thiooxidans and the isolation of a cytochrome aa₃ ubiquinol oxidase from the same bacterium (Sugio et al., 2006) suggest that these Acidithiobacillus spp. have heme A-containing quinol oxidases similar to B. subtilis cytochrome aa₃. The distribution of genes for type 0 CtaA is scattered on the chromosome of other Proteobacteria that are not strongly acidophilic.

Biochemical studies on COX have been reported for several thermoacidophilic archaea that often share the same acidic environments with Acidithiobacillus spp., i.e., Sulfolobus (Lübben and Morand, 1994; Castresana and Saraste, 1995; Yarzábal et al., 2004) and Ferroplasma (Yarzábal et al., 2004; Castelle et al., 2015; Blake et al., 2016). However, biochemical information is not currently available for the other prokaryotes that contain type 0 CtaA. Consequently, we had to apply deductive approaches of genomic and sequence analysis to rationalize the peculiar taxonomic distribution of type 0 CtaA and understand its possible relationship with the evolution of COX.

Archaea May Be Excluded From the Ancestry of Heme A-Containing Oxidases

Given the common ecophysiology of the organisms having type 0 CtaA, two alternative hypotheses can explain their current distribution (Supplementary Figure 6). The first hypothesis is that archaea first evolved CtaA and consequently heme A-containing oxidases and then passed the genes to bacteria. The second hypothesis is that ancestors of extant acidophilic bacteria evolved CtaA and COX first and their genes were transferred by LGT to archaeal lineages sharing their same environment. We examined the first hypothesis thoroughly, considering recent reports of Thermoplasmatales oxidases forming the basal branch in COX1 trees (Golyshina et al., 2016; Spang et al., 2019), which would strengthen the idea of archaeal ancestry for COX (Hemp and Gennis, 2008; Ducluzeau et al., 2014). Our phylogenetic analysis focused on archaeal COX proteins of family A. The COX1 proteins of the phylum Ca. Marsarchaeota (Jay et al., 2018) are fusion proteins with a highly divergent form of COX3 and contain two extra TM at the N-terminal end (Supplementary Table 1). This is an ancestral feature shared with diverse bacterial COX1 sequences (Degli Esposti, 2020), but not seen in other archaeal COX1 proteins such as SoxM, which also are COX1-3 fusion proteins (Pereira et al., 2001; Komorowski et al., 2002). Conversely, COX proteins of Cuniculiplasmataceae have been considered to be ancestral to the whole family A (Golyshina et al., 2016). We confirmed the basal position of these variants and found an extra TM at their N terminal end, but noted that genomes of Cuniculiplasmataceae spp. do not contain a gene for CtaA. Hence, their COX likely represents an intermediate in the transition between family A and B oxidases.

Once mapped upon a robust phylogenetic tree of family A and B oxidases (Figure 5), the distribution of an extra TM at the N terminal end and the variants of both the K- and D-channels for proton pumping (Supplementary Figure 7) indicate that the ancestor of archaeal heme A-containing oxidases diverged after the separation of Cuniculiplasmataceae from the ancestral lineage connecting to family B (node 2 in Figure 5). If this hypothetical proto-archaeal lineage contained the ancestor of all heme A-containing oxidases, then the feature of an extra TM at the N terminal end would have been lost at least five times along their evolution. The same ancestor must have possessed the proton conducting K-channel, because the canonical form of this channel is present in both major lineages departing from its node (Figure 5 and Supplementary Figure 7). Consequently, the K-channel would have been independently lost seven times along the evolution of family A oxidases (Figure 5). Moreover, the presence of type 0 CtaA in organisms belonging to both major clades of archaeal family A oxidases implies multiple losses, if its gene were present in the archaeal ancestor (Figure 5). The appraisal of the predictions and implications applied to the tree and data in Figure 5 indicates that the second hypothesis, namely that current archaeal genes for type 0 CtaA proteins have been acquired from bacteria via multiple LGT events, is much more plausible than the first hypothesis stating the opposite. Indeed, several works have suggested or reported LGT events from bacteria to archaea, especially for proteins associated with bioenergetics such as COX (Boucher et al., 2003; Blank, 2009, 2019; Nelson-Sathi et al., 2012, 2015; López-García et al., 2015; Wolfe and Fournier, 2018). Together with our integrated analysis (Figure 5), this evidence for LGT essentially excludes the first hypothesis.

FIGURE 5

Although Thermoplasmatales proteins appear to occupy the most basal branch of COX1 in phylogenetic trees (Figure 6, see also Supplementary File COX1refined100IQTreeEX_EHOFigTreefull.pdf), their clade exhibits a very long branch (Figure 6 and Supplementary Figure 8). An equivalent long branch was observed in ML trees obtained with various methods and tended to cluster together with long branches of bacterial COX1-3 fusion proteins such as those of Thermus spp. (Figure 6, cf. Supplementary Figure 8). As mentioned earlier in connection with CtaA, attraction between protein clades with long branches is a recurrent artifact in phylogenetic analysis (Brinkmann et al., 2005; Philippe et al., 2005; Bleidorn, 2017). Our analysis suggests that the basal position of Thermoplasmatales COX1-3 proteins in trees encompassing most COX subclades analyzed here (Figure 6), as well as in previously published phylogenetic trees (Spang et al., 2019), likely derives from LBA artifacts. Such problems probably arise from the different evolutionary rate of archaeal and bacterial proteins (Blank, 2009, 2019; Da Cunha et al., 2017; Cavalier-Smith and Chao, 2020). Moreover, the basal position of Thermoplasmatales COX1 in phylogenetic trees (Figure 6 and Supplementary Figure 8a) does not reflect their phylogenetic position in species trees of archaea (Blank, 2009; Colman et al., 2018; Wolfe and Fournier, 2018), which is consistent with the relatively recent age of these thermoacidophilic prokaryotes (Blank, 2009; Colman et al., 2018).

FIGURE 6

Toward a Resolution of the Phylogeny of Bacterial COX1

Given that archaea can be excluded from the ancestry of COX, we next focused our work on bacterial COX to resolve its phylogeny, following the reductionist approaches presented in the Supplementary Material. We started from the observation that no bacterial genome could be found without the concomitant presence of CtaA and family A COX1 genes. We have analyzed all forms of heme A-containing oxidases in currently available genome sequences (Supplementary Table 1) but focus here on those belonging to family A for several reasons. This family is by far the most widespread of the HCO superfamily among living organisms and it encompasses the majority of prokaryotic lineages (see Degli Esposti, 2020 and references therein). In several bacterial taxa, the gene for heme A synthase is associated with the gene cluster of the COX subunits of family A HCO (Degli Esposti, 2020), while this is rarely the case for family B oxidases (Degli Esposti et al., 2019). Furthermore, prokaryotic genomes that encode a single terminal oxidase related to family B, as members of the Cuniculiplasmataceae, do not possess a gene for heme A synthase (Figure 5). Finally, and most importantly, family B oxidases do not present the ancestral features of an extra TM at the N-terminal end and residues typical of the K^C-channel that are shared by family A and C oxidases (Degli Esposti, 2020 – see also Supplementary Figure 7).

The phylogeny of bacterial COX1 is much more complicated than that of archaeal COX1, fundamentally because family A oxidases have a very broad phylogenetic distribution encompassing diverse bacterial lineages (Degli Esposti, 2020; Pereira et al., 2001). We have analyzed all COX1 variants found in available genome sequences and found that the great majority of the variants can be divided in about twenty bacterial subclades (Supplementary Table 1). Split gene clusters typically found in late diverging Actinobacteria, such as Corynebacterium spp., have been previously found to cluster with one of the subclades described in Supplementary Table 1 (Degli Esposti, 2020), and were not analyzed here. After this initial selection, we progressively removed the subclades with long branches or fast substitution rate, that were present in phylogenetically broad trees such as that in Figure 6. This strategy was integrated with a novel approach to build COX1 alignments encompassing the N-terminal region so as to minimize LBA artifacts in assessing the phylogeny of bacterial COX1, as described in the dedicated section of Supplementary Material. Using multiple methods of phylogenetic inference, we ultimately concluded that the Bayesian tree shown in Figure 7A represents a most robust phylogeny of bacterial COX1. The tree presents an early separation of the A1 and the A2 type of COX1 from the basal clade containing COXI with 2 additional TM at the N-terminal end (these proteins are abbreviated 2TM) of various lineages, including iron-oxidizing bacteria. This phylogeny was confirmed in ML trees obtained with the same refined alignment (Supplementary Figures 9–12). Moreover, an equivalent topology emerged in phylogenetic trees obtained without family C paralogs as outgroup (Figures 7B,C). Hence, the choice of outgroup proteins used in previous phylogenetic trees does not affect the consistent basal position of the proteins with the ancestral 2TM extra feature, which include all those of iron-oxidizers that have type 0 CtaA.

FIGURE 7

Acidophilic Iron-Oxidizers Have an Ancestral Form of the COX Assembly Protein CtaG

CtaG is required for the assembly of Cu_B in COX1 (Figure 1A). It must be noted that there are two different types of proteins in bacteria that are called CtaG. One is a homolog of eukaryotic Cox11 and present in, for example, P. denitrificans. We present an analysis of this type of protein and its phylogenetic distribution in a dedicated section of the Supplementary Material. The other type is named caa3_CtaG and present in, for example, B. subtilis (Bengtsson et al., 2004). The caa3_CtaG constitutes the focus of our analysis for several reasons. It is a membrane protein with multiple TM (Figure 8A) and its function is intertwined with that of CtaA to form the heme a₃-Cu_B oxygen-reacting center in COX1. Moreover, the gene for caa3_CtaG often clusters with genes for COX in the chromosome (Figure 1C) of taxa that have the deep branching form of COX1 with the 2TM extra feature (Supplementary Tables 1, 2). Considering all these points, the molecular evolution of caa3_CtaG is likely connected with the early evolution of COX in bacteria, while that of Cox11 homologs may be more relevant to the bacterial ancestry of mitochondrial proteins (Degli Esposti et al., 2019).

FIGURE 8

There is no available structure for caa3_CtaG nor any paralog. The taxonomic distribution of caa3_CtaG is narrower than that of CtaA. To discern the molecular evolution of caa3_CtaG proteins, we undertook a systematic phylogenetic analysis of all the proteins that show the caa3_CtaG domain and have 6 to 9 TM (Supplementary Table 2). The most common structure is similar to that of B. subtilis caa3_CtaG, with 7 TM (Figure 8A). There are several conserved amino acid residues at the predicted positive (outer side) side of the membrane and which might function as Cu ligands (Degli Esposti et al., 2020), among which aspartate residue 249 appears to be invariant. The number of TM does not necessarily correlate with the phylogenetic position of caa3_CtaG proteins, which consistently cluster in three major clades (Figure 8B and Supplementary Figures 13, 14).

The possible phylogeny of caa3_CtaG proteins is shown by the tree of Figure 8B in its simplest and most robust form. This tree was reconstructed from an alignment of the sequences of 40 diverse proteins that represent the taxonomic breadth of the major clades of caa3_CtaG (Supplementary Table 2) while minimizing the presence of long branches, as described in the dedicated section of the Supplementary Material. The caa3_CtaG proteins of Acidithiobacillus spp. and Acidiferrobacter spp., which are predicted to have 8TM (one additional TM at the N-terminal end), show a different set of potential copper ion ligands than those in most other caa3_CtaG proteins (Figure 8C). However, these proteins maintain the invariant aspartate 249 (B. subtilis caa3_CtaG numbering) at the positive side of the membrane and are recognized as members of the caa3_CtaG superfamily. Using alignments that included also the N-terminal part of caa3_CtaG proteins, which has been routinely trimmed before (as in the alignment used to generate the tree in Figure 8B, cf. [Degli Esposti et al., 2020]), we obtained similar phylogenetic trees (Figure 8D and Supplementary Figure 14). Hence, the basal position of caa3_CtaG of iron-oxidizing Proteobacteria such as Acidithiobacillus spp. was confirmed in multiple ways, consistent with the molecular evolution of CtaA (Figure 3B).

Conclusion

This work presents converging evidence suggesting that COX of extant acidophilic bacteria, in particular iron-oxidizing Proteobacteria such Acidithiobacillus and Acidiferrobacter spp., may be the closest to primitive heme A-containing respiratory oxidases. Protein factors specifically involved in biosynthesis of COX have presumably coevolved with COX1, as reflected by the clustering of their genes with those encoding COX proteins. The results of our in depth analysis of two proteins required for the assembly of the oxygen-reacting center in COX, CtaA for the biogenesis of heme A and caa3_CtaG for Cu_B, strongly support the inferred phylogeny of bacterial COX. Namely, CtaA and caa3_CtaG proteins of iron-oxidizing Proteobacteria form the basal branch in phylogenetic trees obtained with different inference methods and programs of phylogenetic analysis. Once invented, the advantageous bioenergetic capacity of respiration with molecular oxygen via heme A-containing oxidases evidently spread to various prokaryotes (Han et al., 2011; Wikström and Springett, 2020).

Geochemical and ecological evidence sustains the COX phylogeny that emerges from our data. Lithotrophic bacteria, such as Acidithiobacillus spp. and other acidophilic iron-oxidizers can release abundant levels of Cu ions by bioleaching of common crust rocks (Quatrini and Johnson, 2018; Degli Esposti et al., 2020). Cu bioavailability often limits COX biogenesis in aquatic environments, especially in oceans where Cu ion concentrations are normally low (Degli Esposti et al., 2019). The earliest geochemical evidence for bacterial respiration points to ancestral iron-oxidizing bacteria similar to extant Acidithiobacillus spp. as responsible for the acid leaching of soil crust rocks containing metals such as Cu, Cr and Co, which then were washed away from emerged land producing rich deposits in ocean sediments (Konhauser et al., 2011). Metal leaching driven by bacterial respiration lasted only a few hundred million years during and after the GOE (Konhauser et al., 2011). Once surface pyrite minerals were consumed by intense oxidation, COX genes had spread laterally to soil dwelling bacteria with faster growth capacity than the ancestors of extant iron oxidizers. To conclude, the early evolution of COX was apparently promoted by the availability of relatively high levels of oxygen produced locally by Cyanobacteria, combined with the availability of surface pyrite rock material in primordial earth environments.

References

• 1

  AcuñaL. G.CárdenasJ. P.CovarrubiasP. C.HaristoyJ. J.FloresR.NuñezH.et al (2013). Architecture and Gene Repertoire of the Flexible Genome of the Extreme Acidophile Acidithiobacillus caldus.PLoS One8:e78237. 10.1371/journal.pone.0078237

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2

  Appia-AymeC.GuilianiN.RatouchniakJ.BonnefoyV. (1999). Characterization of an operon encoding two c-type cytochromes, an aa(3)-type cytochrome oxidase, and rusticyanin in Thiobacillus ferrooxidans ATCC 33020.Appl. Environ. Microbiol.654781–4787. 10.1128/aem.65.11.4781-4787.1999

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3

  BengtssonJ.von WachenfeldtC.WinstedtL.NygaardP.HederstedtL. (2004). CtaG is required for formation of active cytochrome c oxidase in Bacillus subtilis.Microbiology150415–425. 10.1099/mic.0.26691-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4

  BlakeI. R. C.AnthonyM. D.BatesJ. D.HudsonT.HunterK. M.KingB. J.et al (2016). In situ Spectroscopy Reveals that Microorganisms in Different Phyla Use Different Electron Transfer Biomolecules to Respire Aerobically on Soluble Iron.Front. Microbiol.7:1963.

  • Google Scholar

• 5

  BlankC. E. (2009). Not so old Archaea - the antiquity of biogeochemical processes in the archaeal domain of life.Geobiology7495–514. 10.1111/j.1472-4669.2009.00219.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6

  BlankC. E. (2019). Phylogenomic dating—a method of constraining the age of microbial taxa that lack a conventional fossil record.Astrobiology9173–191. 10.1089/ast.2008.0247

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 7

  BleidornC. (2017). Phylogenomics.Cham: Springer International Publishing, 173–193.

  • Google Scholar

• 8

  BoratynG. M.SchäfferA.AgarwalaR.AltschulS. F.LipmanD. J.MaddenT. L. (2012). Domain enhanced lookup time accelerated BLAST.Biol. Direct.7:12. 10.1186/1745-6150-7-12

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9

  BouckaertR. R.HeledJ. (2014). DensiTree 2: Seeing trees through the forest.BioRxiv2012:012401. 10.1101/012401

  • CrossRef
  • Google Scholar

• 10

  BouckaertR.VaughanT. G.Barido-SottaniJ.DuchêneS.FourmentM.GavryushkinaA.et al (2019). BEAST 2.5: An advanced software platform for Bayesian evolutionary analysis.PLoS Comput. Biol.15:e1006650.

  • Google Scholar

• 11

  BrinkmannH.van der GiezenM.ZhouY.De RaucourtG. P.PhilippeH. (2005). An empirical assessment of long-branch attraction artefacts in deep eukaryotic phylogenomics.System. Biol.54743–757. 10.1080/10635150500234609

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12

  BrownK. R.AllanB. M.DoP.HeggE. L. (2002). Identification of novel hemes generated by heme A synthase: evidence for two successive monooxygenase reactions.Biochemistry4110906–10913. 10.1021/bi0203536

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13

  BoucherY.DouadyC. J.PapkeR. T.WalshD. A.BoudreauM. E.NesbøC. L.et al (2003). Lateral gene transfer and the origins of prokaryotic groups.Ann. Rev. Genet.37283–328. 10.1146/annurev.genet.37.050503.084247

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14

  CastelleC. J.RogerM.BauzanM.BrugnaM.LignonS.NimtzM.et al (2015). The aerobic respiratory chain of the acidophilic archaeon Ferroplasma acidiphilum: A membrane-bound complex oxidizing ferrous iron.Biochim. Biophys. Acta1847717–728. 10.1016/j.bbabio.2015.04.006

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15

  CastresanaJ.SarasteM. (1995). Evolution of energetic metabolism: the respiration-early hypothesis.Trends Biochem. Sci.20443–448. 10.1016/s0968-0004(00)89098-2

  • CrossRef
  • Google Scholar

• 16

  Cavalier-SmithT.ChaoE. E. (2020). Multidomain ribosomal protein trees and the planctobacterial origin of neomura (eukaryotes, archaebacteria).Protoplasma257621–753. 10.1007/s00709-019-01442-7

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17

  ColmanD. R.PoudelS.HamiltonT. L.HavigJ. R.SelenskyM. J.ShockE. L.et al (2018). Geobiological feedbacks and the evolution of thermoacidophiles.ISME J.12225–236. 10.1038/ismej.2017.162

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 18

  Da CunhaV.GaiaM.GadelleD.NasirA.ForterreP. (2017). Lokiarchaea are close relatives of Euryarchaeota, not bridging the gap between prokaryotes and eukaryotes.PLoS Genet.13:e1006810. 10.1371/journal.pgen.1006810

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19

  Degli EspostiM. (2020). On the evolution of cytochrome oxidases consuming oxygen.Biochim. Biophys. Acta Bioenerg.1861:148304. 10.1016/j.bbabio.2020.148304

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20

  Degli EspostiM.Garcia-MezaV.GomezA. C.Moya-BeltranA.QuatriniR.HederstedtL. (2020). Heme A-containing oxidases evolved in the ancestors of iron oxidizing bacteria.BioRxiv2020:968255. 10.1101/2020.03.01.968255v2

  • CrossRef
  • Google Scholar

• 21

  Degli EspostiM.MentelM.MartinW.SousaF. L. (2019). Oxygen Reductases in Alphaproteobacterial Genomes: Physiological Evolution From Low to High Oxygen Environments.Front. Microbiol.10:499.

  • Google Scholar

• 22

  DengX.DohmaeN.NealsonK. H.HashimotoK.OkamotoA. (2018). Multi-heme cytochromes provide a pathway for survival in energy-limited environments.Sci. Adv.4:eaao5682. 10.1126/sciadv.aao5682

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23

  DucluzeauA. L.Schoepp-CothenetB.van LisR.BaymannF.RussellM. J.NitschkeW. (2014). The evolution of respiratory O₂/NO reductases: an out-of-the-phylogenetic-box perspective.J. R. Soc. Interface11:20140196. 10.1098/rsif.2014.0196

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 24

  DrummondA. J.BouckaertR. R. (2015). Bayesian evolutionary analysis with BEAST.Cambridge,: Cambridge University Press.

  • Google Scholar

• 25

  DrummondA. J.HoS. Y.PhillipsM. J.RambautA. (2006). Relaxed phylogenetics and dating with confidence.PLoS Biol.4:e88. 10.1371/journal.pbio.0040088

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26

  FergusonS. J.IngledewW. J. (2008). Energetic problems faced by micro-organisms growing or surviving on parsimonious energy sources and at acidic pH: I. Acidithiobacillus ferrooxidans as a paradigm.Biochim. Biophys. Acta17771471–1479. 10.1016/j.bbabio.2008.08.012

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 27

  GolyshinaO. V.KublanovI. V.TranH.KorzhenkovA. A.LünsdorfH.NechitayloT. Y.et al (2016). Biology of archaea from a novel family Cuniculiplasmataceae (Thermoplasmata) ubiquitous in hyperacidic environments.Sci. Rep.6:39034.

  • Google Scholar

• 28

  GuindonS.DufayardJ. F.LefortV.AnisimovaM.HordijkW.GascuelO. (2010). New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0.Syst. Biol.59307–321. 10.1093/sysbio/syq010

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29

  HammerN. D.Schurig-BriccioL. A.GerdesS. Y.GennisR. B.SkaarE. P. (2016). CtaM Is Required for Menaquinol Oxidase aa₃ Function in Staphylococcus aureus.mBio7:4.

  • Google Scholar

• 30

  HanH.HempJ.PaceL. A.OuyangH.GanesanK.RohJ. H.et al (2011). Adaptation of aerobic respiration to low O₂ environments.Proc. Natl. Acad. Sci. USA10814109–14114. 10.1073/pnas.1018958108

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31

  HeD.FuC. J.BaldaufS. L. (2016). Multiple Origins of Eukaryotic cox15 Suggest Horizontal Gene Transfer from Bacteria to Jakobid Mitochondrial DNA.Mol. Biol. Evol.33122–133. 10.1093/molbev/msv201

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32

  HederstedtL. (2012). Heme A biosynthesis.Biochim. Biophys. Acta Bioenerg.1817920–927.

  • Google Scholar

• 33

  HempJ.GennisR. B. (2008). Diversity of the heme-copper superfamily in archaea: insights from genomics and structural modeling.Results Prob. Cell Diff.451–31. 10.1007/400_2007_046

  • CrossRef
  • Google Scholar

• 34

  IssottaF.Moya-BeltránA.MenaC.CovarrubiasP. C.ThyssenC.BellenbergS.et al (2018). Insights into the biology of acidophilic members of the Acidiferrobacteraceae family derived from comparative genomic analyses.Res. Microbiol.169608–617. 10.1016/j.resmic.2018.08.001

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35

  IwataS.OstermeierC.LudwigB.MichelH. (1995). Structure at 2.8 A resolution of cytochrome c oxidase from Paracoccus denitrificans.Nature376660–669. 10.1038/376660a0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 36

  JayZ. J.BeamJ. P.DlakićM.RuschD. B.KozubalM. A.InskeepW. P. (2018). Marsarchaeota are an aerobic archaeal lineage abundant in geothermal iron oxide microbial mats.Nat. Microb.3732–740. 10.1038/s41564-018-0163-1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 37

  KhadkaR.ClothierL.WangL.LimC. K.KlotzM. G.DunfieldP. F. (2018). Evolutionary History of Copper Membrane Monooxygenases.Front. Microbiol.9:2493.

  • Google Scholar

• 38

  KomorowskiL.VerheyenW.SchäferG. (2002). The archaeal respiratory supercomplex SoxM from S. acidocaldarius combines features of quinole and cytochrome c oxidases.J. Biol. Chem.3831791–1799.

  • Google Scholar

• 39

  KonhauserK. O.LalondeS. V.PlanavskyN. J.PecoitsE.LyonsT. W.MojzsisS. J.et al (2011). Aerobic bacterial pyrite oxidation and acid rock drainage during the Great Oxidation Event.Nature478369–373. 10.1038/nature10511

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 40

  LeS. Q.GascuelO. (2010). Accounting for Solvent Accessibility and Secondary Structure in Protein Phylogenetics Is Clearly Beneficial.Syst. Biol.59277–287. 10.1093/sysbio/syq002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41

  LeS. Q.LartillotN.GascuelO. (2008). Phylogenetic mixture models for proteins.Phil. Trans. R. Soc. B3633965–3976. 10.1098/rstb.2008.0180

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 42

  LeeM. S.SoubrierJ.EdgecombeG. D. (2013). Rates of phenotypic and genomic evolution during the Cambrian explosion.Curr. Biol.231889–1895. 10.1016/j.cub.2013.07.055

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 43

  LewinA.HederstedtL. (2008). Promoted evolution of a shortened variant of heme A synthase in the membrane of Bacillus subtilis.FEBS Lett.5821330–1334. 10.1016/j.febslet.2008.03.015

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 44

  LewinA.HederstedtL. (2016). Heme A synthase in bacteria depends on one pair of cysteinyls for activity.Biochim. Biophys. Acta Bioenerg.1857160–168. 10.1016/j.bbabio.2015.11.008

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 45

  López-GarcíaP.ZivanovicY.DeschampsP.MoreiraD. (2015). Bacterial gene import and mesophilic adaptation in archaea.Nat. Rev. Microbiol.13447–456. 10.1038/nrmicro3485

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 46

  LuS.WangJ.ChitsazF.DerbyshireM. K.GeerR. C.GonzalesN. R.et al (2020). CDD/SPARCLE: the conserved domain database in 2020.Nucleic Acids Res.48D265–D268.

  • Google Scholar

• 47

  LübbenM.MorandK. (1994). Novel prenylated hemes as cofactors of cytochrome oxidases. Archaea have modified hemes A and O.J. Biol. Chem.26921473–21479. 10.1016/s0021-9258(17)31828-8

  • CrossRef
  • Google Scholar

• 48

  Matheus CarnevaliP. B.SchulzF.CastelleC. J.KantorR. S.ShihP. M.SharonI.et al (2019). Hydrogen-based metabolism as an ancestral trait in lineages sibling to the Cyanobacteria.Nat. Commun.10:463.

  • Google Scholar

• 49

  MöllerS.CroningM. D.ApweilerR. (2001). Evaluation of methods for the prediction of membrane spanning regions.Bioinformatics17646–653. 10.1093/bioinformatics/17.7.646

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 50

  MogiT.SaikiK.AnrakuY. (1994). Biosynthesis and functional role of haem O and haem A.Mol. Microbiol.14391–398. 10.1111/j.1365-2958.1994.tb02174.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51

  Nelson-SathiS.DaganT.LandanG.JanssenA.SteelM.McInerneyJ. O.et al (2012). Acquisition of 1,000 eubacterial genes physiologically transformed a methanogen at the origin of Haloarchaea.Proc. Natl. Acad. Sci. USA10920537–20542. 10.1073/pnas.1209119109

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52

  Nelson-SathiS.SousaF. L.RoettgerM.Lozada-ChávezN.ThiergartT.JanssenA.et al (2015). Origins of major archaeal clades correspond to gene acquisitions from bacteria.Nature51777–80. 10.1038/nature13805

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 53

  NguyenL. T.SchmidtH. A.von HaeselerA.MinhB. (2015). Q. IQ-TREE: A fast and effective stochastic algorithm for estimating maximum likelihood phylogenies.Mol. Biol. Evol.32268–274. 10.1093/molbev/msu300

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 54

  NiwaS.TakedaK.KosugiM.TsutsumiE.MogiT.MikiK. (2018). Crystal structure of heme A synthase from Bacillus subtilis.Proc. Natl. Acad. Sci. USA11511953–11957.

  • Google Scholar

• 55

  NorrisP. R.FalagánC.Moya-BeltránA.CastroM.QuatriniR.JohnsonD. B. (2020). Acidithiobacillus ferrianus sp. nov.: an ancestral extremely acidophilic and facultatively anaerobic chemolithoautotroph.Extremophiles24329–337. 10.1007/s00792-020-01157-1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 56

  ParksD. H.ChuvochinaM.WaiteD. W.RinkeC.SkarshewskiA.ChaumeilP. A.et al (2018). A standardized bacterial taxonomy based on genome phylogeny substantially revises the tree of life.Nat. Biotechnol.36996–1004. 10.1038/nbt.4229

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 57

  PereiraM. M.SantanaM.TeixeiraM. (2001). A novel scenario for the evolution of heme-copper oxygen reductases.Biochim. Biophys. Acta1505185–208. 10.1016/s0005-2728(01)00169-4

  • CrossRef
  • Google Scholar

• 58

  PhilippeH.ZhouY.BrinkmannH.RodrigueN.DelsucF. (2005). Heterotachy and long-branch attraction in phylogenetics.BMC Evol. Bio5:50.

  • Google Scholar

• 59

  QuatriniR.Appia-AymeC.DenisY.JedlickiE.HolmesD. S.BonnefoyV. (2009). Extending the models for iron and sulfur oxidation in the extreme acidophile Acidithiobacillus ferrooxidans.BMC Genom.10:394. 10.1186/1471-2164-10-394

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 60

  QuatriniR.JohnsonD. B. (2018). Microbiomes in extremely acidic environments: functionalities and interactions that allow survival and growth of prokaryotes at low pH.Curr. Opin. Microbiol.43139–147. 10.1016/j.mib.2018.01.011

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 61

  RatnasinghamS.HebertP. D. (2013). A DNA-based registry for all animal species: the Barcode Index Number (BIN) system.PLoS One8:e66213. 10.1371/journal.pone.0066213

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 62

  SharmaV.WikströmM. (2014). A structural and functional perspective on the evolution of the heme-copper oxidases.FEBS Lett.5883787–3792. 10.1016/j.febslet.2014.09.020

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 63

  SooR. M.HempJ.ParksD. H.FischeW. W.HugenholtzP. (2017). On the origins of oxygenic photosynthesis and aerobic respiration in cyanobacteria.Science3551436–1440. 10.1126/science.aal3794

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 64

  SousaF. L.AlvesR. J.Pereira-LealJ. B.TeixeiraM.PereiraM. M. (2011). A bioinformatics classifier and database for heme-copper oxygen reductases.PLoS One6:e19117. 10.1371/journal.pone.0019117

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 65

  SpangA.StairsC. W.DombrowskiN.EmeL.LombardJ.CaceresE. F.et al (2019). Proposal of the reverse flow model for the origin of the eukaryotic cell based on comparative analyses of Asgard archaeal metabolism.Nat. Microb.41138–1148. 10.1038/s41564-019-0406-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 66

  SugioT.HisazumiT.KanaoT.KamimuraK.TakeuchiF.NegishiA. (2006). Existence of aa₃-type ubiquinol oxidase as a terminal oxidase in sulfite oxidation of Acidithiobacillus thiooxidans.Biosci. Biotech. Biochem.701584–1591. 10.1271/bbb.50623

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 67

  Svensson-EkM.AbramsonJ.LarssonG.TörnrothS.BrzezinskiP.IwataS. (2002). The X-ray crystal structures of wild-type and EQ(I-286) mutant cytochrome c oxidases from Rhodobacter sphaeroides.J. Mol. Biol.321329–339. 10.1016/s0022-2836(02)00619-8

  • CrossRef
  • Google Scholar

• 68

  TamuraK.PetersonD.PetersonN.StecherG.NeiM.KumarS. (2011). MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods.Mol. Biol. Evol.282731–2739. 10.1093/molbev/msr121

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 69

  TrifinopoulosJ.NguyenL. T.von HaeselerA.MinhB. Q. (2016). W-IQ-TREE: a fast online phylogenetic tool for maximum likelihood analysis.Nucleic Acid Res.44W232–W235.

  • Google Scholar

• 70

  von WachenfeldtC.HallgrenJ.HederstedtL. (2021). YtkA (CtaK) and YozB (CtaM) function in the biogenesis of cytochrome c oxidase in Bacillus subtilis.Mol. Microbiol.2021:14701. 10.1111/mmi.14701

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 71

  WikströmM.SpringettR. (2020). Thermodynamic efficiency, reversibility, and degree of coupling in energy conservation by the mitochondrial respiratory chain.Commun. Biol.3:451.

  • Google Scholar

• 72

  WolfeJ. M.FournierG. P. (2018). Horizontal gene transfer constrains the timing of methanogen evolution.Nat. Ecol. Evol.2897–903. 10.1038/s41559-018-0513-7

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 73

  YarzábalA.Appia-AymeC.RatouchniakJ.BonnefoyV. (2004). Regulation of the expression of the Acidithiobacillus ferrooxidans rus operon encoding two cytochromes c, a cytochrome oxidase and rusticyanin.Microbiology1502113–2123. 10.1099/mic.0.26966-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 74

  ZengH.ZhuG.ZhangS.LiX.MartinJ.MorgnerN.et al (2020). Isolated Heme A Synthase from Aquifex aeolicus Is a Trimer.mBio11e2615–e2619.

  • Google Scholar