Chimeric inheritance and crown-group acquisitions of carbon fixation genes within Chlorobiales: Origins of autotrophy in Chlorobiales and implication for geological biomarkers.

The geological record of microbial metabolisms and ecologies primarily consists of stable isotope fractionations and the diagenetic products of biogenic lipids. Carotenoid lipid biomarkers are particularly useful proxies for reconstructing this record, providing information on microbial phototroph primary productivity, redox couples, and oxygenation. The biomarkers okenane, chlorobactane, and isorenieratene are generally considered to be evidence of anoxygenic phototrophs, and provide a record that extends to 1.64 Ga. The utility of the carotenoid biomarker record may be enhanced by examining the carbon isotopic ratios in these products, which are diagnostic for specific pathways of biological carbon fixation found today within different microbial groups. However, this joint inference assumes that microbes have conserved these pathways across the duration of the preserved biomarker record. Testing this hypothesis, we performed phylogenetic analyses of the enzymes constituting the reductive tricarboxylic acid (rTCA) cycle in Chlorobiales, the group of anoxygenic phototrophic bacteria usually implicated in the deposition of chlorobactane and isorenieretane. We find phylogenetically incongruent patterns of inheritance across all enzymes, indicative of horizontal gene transfers to both stem and crown Chlorobiales from multiple potential donor lineages. This indicates that a complete rTCA cycle was independently acquired at least twice within Chlorobiales and was not present in the last common ancestor. When combined with recent molecular clock analyses, these results predict that the Mesoproterzoic lipid biomarker record diagnostic for Chlorobiales should not preserve isotopic fractionations indicative of a full rTCA cycle. Furthermore, we conclude that coupling isotopic and biomarker records is insufficient for reliably reconstructing microbial paleoecologies in the absence of a complementary and consistent phylogenomic narrative.

Introduction

Lipid biomarkers are geochemically stable molecular remnants of organisms, frequently used to trace paleoenvironmental proxies, or as “fossils” indicating the presence of specific groups of organisms [1-3]. Phototrophic bacteria, including oxygenic cyanobacteria and anoxygenic green sulfur bacteria (GSB, order Chlorobiales) and purple sulfur bacteria (PSB, order Chromatiales), have especially significant biomarker records in the form of carotenoid pigment derivatives. These can indicate periods of major planetary change, such as great oceanic anoxic events [4].

The first appearance of sulfur bacteria in the biomarker record is from the 1.64 Ga Barney Creek Formation, which contains carotenoids including okenane, chlorobactane, and isorenieratane [5, 6]. This implies sulfur bacteria were a substantial part of aquatic microbial ecologies by this time, consistent with the presence of a stratified ocean with widespread euxinic conditions [7]. However, the genes encoding the biosynthesis of aromatic carotenoids and their derivatives have also been identified in cyanobacteria, which occupy more oxygenated water columns [8-10]. Carotenoid biomarkers within Neoproterozoic marine sediments (1000–542 Ma) have shown to be primarily comprised of cyanobacterial renierapurpurane, with small amounts of the Chlorobiaceae-associated biomarker isorenieratane. Furthermore, cyanobacteria producing synechoxanthin have been shown to be capable of producing both isorenieratene and, in principle, chlorobactene, given their enzymatic repertoire [11]. This brings the evidence for GSB at 1.64 Ga BCF formations into question, as well as the practice of using these biomarkers as a calibration for dating Chlorobiales in molecular clock analyses, since cyanobacteria could potentially be the source of the carotenoid derivatives found in these deposits [12, 13].

A distinct, comparable source of information for tracing microbial autotrophy can be found within the geochemical record of carbon fixation, which can further inform reconstructions of microbial paleoecology. There are six known modern biological pathways used by organisms to fix CO2. Most fractionate carbon differently, variably offsetting carbon isotope (δ13C) values of inorganic and organic carbon in the isotopic record [14, 15]. Through this bulk signaling of δ13C sources, a cycle’s input can be mathematically inferred by isotopic mass balance of its components [16].

Of the known carbon fixation pathways, the Calvin-Benson-Bassham (CBB) cycle and the rTCA cycle are the most prevalent. The taxonomic distributions of these pathways are polyphyletic, consistent with deep evolutionary histories of horizontal gene transfer(HGT) [16, 17]. The CBB cycle is found in both photoautotrophic bacteria such as cyanobacteria and purple sulfur bacteria, as well as some chemolithoautotrophs; it fixes three molecules of CO2 for the synthesis of one 3-phoshoglycerate molecule using ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCo). The rTCA cycle is present in a diverse group of autotrophic microbes, including most members of Chlorobiales, sulfate-reducing bacteria, hyperthermophilic bacteria, and Crenarchaeota [18, 19]. The complete cycle fixes three molecules of CO2 for the synthesis of one 3-carbon pyruvate molecule through 9 steps that also produce several metabolic intermediates [20-22] (Fig 1). Except for ATP-citrate lyase, which drives the reductive cycling [23], and pyruvate: ferredoxin oxioreductase and 2-oxoglutarate:ferredoxin oxioreductase (OGOR) needed to reduce CO2, these enzymes are also used within the oxidative TCA cycle, which has been shown to operate within Chlorobiales under some growth conditions [24]. However, a complete oxidative TCA cycle is not present in obligate anaerobes, as aerobic respiratory chains are presumably necessary for the efficient re-oxidation of NADH [25].

Fig 1

Enzymatic map of the rTCA cycle.

Enzymes catalyzing each reaction (red), inputs of carbon dioxide (blue) and intermediate compounds (black) are shown. Steps are numbered in the direction of reactions, starting with the conversion of oxaloacetate to malate via malate dehydrogenase.

Recently, an alternative rTCA cycle has been characterized that uses two different enzymes to convert citrate into citryl-CoA, instead of ATP citrate lyase [26]. This was initially observed in Hydrogenobacter thermophilus but is also present in other members of Aquificeae [27]. Anaerobes with partial TCA cycles also use α-ketoglutarate:ferredoxin oxioreductase, and fumarate reductase instead of succinate dehydrogenase, similar to the rTCA cycle [25, 28], and pyruvate: ferredoxin oxioreductase is used in some carbohydrate fermentation pathways [29]. Thus, the phyletic distribution of genes found in the rTCA cycle is complex, and potentially indicative of several different metabolic processes other than autotrophy.

The order Chlorobiales consists of a group of unclassified photoheterotrophic thermophiles and two families: Chlorobiaceae that contains five genera (Chlorobium, Chlorobaculum, Chloroherpeton, and Prosthecochloris) and the newly proposed family Chloroherpetonaceae with species of Chloroherpeton [30, 31]. The 16S rRNA phylogenies of this group recover close relationships to Ignavibacterales (within the phylum Chlorobi) and the Bacteriodetes, which are often grouped together along with Firmicutes in a CFB superphylum [32, 33] (Fig 2). With the exception of the unclassified thermophiles, Chlorobiales are strictly anaerobic, obligate photoautotrophs that use the rTCA cycle. Within this group, Chlorobium, Chlorobaculum, and Prosthecochloris genera form a clade that synthesizes aromatic carotenoid compounds. Metagenomic analyses and culturing of novel strains has revealed more aerobic and photoheterotrophic members of Chlorobi lacking enzymes for autotrophic carbon fixation [34-37]. As shown in Fig 2, some of these uncultured genome species, such as Candidatus Thermochlorobacter aerophylum [34, 35], Chlorobium sp. 445 and Chlorobium sp. GBChlB35 group as a distinct clade within Chlorobiales. This distribution of physiological traits in Chlorobiales may be explained by vertical inheritance and loss, or by HGT, depending on the inferred phylogenies of the individual gene families providing these functions.

Fig 2

16S rRNA tree of Chlorobiales and Ignavibacterium (outgroup) showing taxonomic distributions of a selection of traits.

Labeled brackets indicate the taxonomic distribution of phototrophy and autotrophy within this group. The green box indicates groups known to synthesize aromatic carotenoid lipid biomarkers.

If the constituent enzymes of the rTCA cycle were acquired within crown Chlorobiales, rather than the stem lineage, then ancestral autotrophy cannot be inferred for this order. Subsequently, diagnostic lipid biomarkers more ancient than the inferred age of crown Chlorobiales should not show isotopic fractionations indicative of rTCA cycle carbon fixation. Conversely, a phylogenetic signal showing clear acquisition of all rTCA enyzmes within stem Chlorobiales would be supportive of a much older history of autotrophy within this group and would predict isotopic fractionations of even the oldest preserved biomarkers to be similar to those of modern autotrophic Chlorobiales. To investigate this, we performed phylogenetic analyses of rTCA cycle enzymes in Chlorobiales to map their evolutionary histories. Overall, our investigations reveal a mixed history of vertically inherited and horizontally transferred rTCA cycle genes across multiple groups of Chlorobiales (Table 1). These complex histories support a hypothesis of patchwork enzymatic inheritances from multiple and diverse HGT donors, implying the complete rTCA cycle, and therefore autotrophy, was not present in the ancestral stem group of Chlorobiales.

Table 1

Summary of evolutionary histories of carbon fixation genes in Chlorobiales.

rTCA Step	Enzyme Name	Inferred History	Sibling Groups	Outgroup Carbon Metabolism
1	malate dehydrogenase	Chlorobiales: stem HGT	Rhodothermus marinus	Mixed
2	fumarate hydratase	(A)Chlorobiales: stem HGT/vertical inheritance (B)C.thalissium: orthologous displacement	(A) Bacteroidetes, Ignavibacterium, Plantomycetaeceae (B) Oceanospiralles bacterium	Heterotrophs
3	fumarate reductase	Chlorobiales: stem HGT	Candidatus Lambdaproteobacteria, Geobacter, Desulfuromonadales bacterium	Mixed
4	succinyl-CoA synthetase	Chlorobiales: vertical inheritance	Bacteroidetes, Ignavibacterium	Heterotrophs
5	2oxoglutarate: ferredoxin oxidoreductase	Chlorobiales: stem HGT from Bacteroidetes	Balneolaceae, Bacteroidetes	Heterotrophs
6	isocitrate dehydrogenase	(A)Chlorobiales: HGT (B)C.thalissium/ Photoheterotrophs: HGT	(A) Deltaproteobacteria, Nitrospira/ Nitrospinae (B) Armatimonadetes, Caldithrix	Mixed
7	aconitate hydratase	(A)Chlorobiales: HGT (B)C.thalassium/ Thermochlorobacter: HGT	(A) Campylobacterales (B)Ignavibacteria, Spirochaetes	(A) Autotrophs (B) Heterotrophs
8	ATP citrate lyase	Chlorobiales: stem HGT	Nitrospirae/ Nitrospinae	Mixed
9	pyruvate: erredoxin oxidoreductase	Chlorobiales: HGT	Bacteroidetes, Ignavibacteria	Mixed

Table notes: Italicized species names refer to taxon source of query sequence for respective groupings

Materials and methods

Genomic data retrieval

Query sequences for rTCA cycle enzymes were selected from the sequenced Chlorobium thalissium ATCC 35110 genome (CP001100.1) retrieved from the NCBI database (S1 Table). In cases where the protein ortholog from Chloroherpeton thalissum did not recover monophyly with other Chlorobiaceae, orthologous query sequences were taken from Chlorobium tepidum TLS (AE006470.1).

BLAST search for homologous enzymes

The Basic Local Alignment Search Tool (BLAST) was used to compare queries with homologous sequences using default search parameters. The BLASTp cutoff for homologous identification was set at 250 taxa maximum, with sequence identity ≥ 32% These searchers were not exhaustive for all rTCA homologs, but rather exhaustively identified all Chlorobiales homologs and extensively sampled sequences representing potential HGT donor lineages and/or distant vertically inherited sequences from outgroup clades.

Phylogenetic analysis

Protein sequences were aligned using the multiple sequence alignment tool MAFFT v.7.245 (FFT-NS-2, BLOSUM6) [38]. Maximum-likelihood phylogenetic trees were generated using IQTree run with the Bayes Information Criterion (BIC) test (S1 Table). Support for bipartitions was inferred using rapid bootstraps (1000 replicates) and SH-aLRT tests. Additional phylogenetic analysis of the ATP citrate lyase dataset was performed using Phylobayes v. 4.171, using C20 site specific profiles and a convergence criterion cutoff of 0.3 for all parameters, with a 20% chain burn-in. Trees were rooted at midpoints and visualized in FigTree v.1.4.4.

Results

rTCA cycle protein phylogenies show a complex history of HGT

Maximum-likelihood phylogenies of rTCA pathway proteins within Chlorobiales show surprisingly diverse evolutionary origins (Table 1). Several show monophyly indicative of acquisition within stem Chlorobiales (malate dehydrogenase, pyruvate ferredoxin, ATP citrate lyase, succinyl-CoA synthetase, OGOR, fumarate reductase). Other enzymes show polyphyletic distributions, indicative of multiple HGT acquisitions from different donor lineages (isocitrate dehydrogenase, fumarate hydratase, aconitate hydratase). Several enzymes were subsequently lost within the photoheterotrophic Chlorobiales clade (aconitate hydratase, pyruvate: ferredoxin oxidoreductase, fumarate reductase, ATP-citrate lyase), likely following divergence from the Chloroherpeton lineage.

Several general trends can be identified in the placement of these groups within broader protein diversity, based on the outgroup sequences within each tree. There are three scenarios wherein the inheritance of an rTCA cycle protein could be inferred for crown Chlorobiales. First, trees where the outgroups are similar to those of the species tree, such as other Chlorobi (e.g., Ignavibacteria and/or Bacteroidetes) suggest that these proteins were ancestral within the Chlorobiales lineage, and subsequently vertically inherited. Of the rTCA pathway proteins, only succinyl-CoA synthetase appears to have a history consistent with this interpretation. Second, trees where Chlorobiales sequences are on long branches without any closely related outgroups and crown group monophyly is preserved (malate dehydrogenase, pyruvate:ferredoxin, ATP citrate lyase). These enzymes were likely acquired by stem Chlorobiales via HGT, but current taxonomic sampling and/or lack of phylogenetic signal prevent an unambiguous identification of a putative donor lineage. Third, trees showing vertical inheritance within crown Chlorobiales following an HGT from a clear donor group to stem Chlorobiales. Only OGOR preserves this signal. The remaining tree topologies are complicated by HGTs within crown group lineages, so that Chlorobiales is polyphyletic within the tree (fumarate hydratase, isocitrate dehydrogenase, aconitate hydratase). Either these genes were acquired in stem Chlorobiales and a subsequent orthologous displacements occurred, or the original acquisitions within crown Chlorobiales were independent.

Malate dehydrogenase

The ML tree for malate dehydrogenase shows a broad taxonomic range of bacteria, with 11 distinct phyla represented in 242 taxa. The tree is consistent with acquisition of malate dehydrogenase in the Chlorobiales stem ancestor, and subsequent vertical inheritance. The sibling groups are diverse, including both autotrophic and heterotrophic representatives of Gemmatimonadetes, Deltaproteobacteria, Acidobacteria, and Bacteroidetes, with poorly supported relationships between most groups (Fig 3). This indicates a complex history of HGT between phyla, preventing identification of a likely HGT donor to stem Chlorobiales.

Fig 3

Maximum likelihood (ML) tree of malate dehydrogenase homologs.

(A) midpoint-rooted tree with collapsed clades labeled with taxonomic group names. (B) Higher resolution tree showing crown Chlorobiales. Support values indicate approximate likelihood ratio test (aLRT)/ bootstrap (100 replicates). Major bipartitions with bootstrap (BS) support are labeled. The full tree contains 242 taxa. Collapsed clades are labeled with taxonomic group names. Color bars or boxes to the right of the trees indicate autotrophic (green), heterotrophic (orange), or undetermined (white) carbon metabolisms. Clades including multiple carbon metabolisms are indicated with diagonally hatched boxes.

Fumarate hydratase

Genes encoding fumarate hydratase do not recover the monophyly of Chlorobiales, with the ortholog from Chloroherpeton thalissium more closely related to homologs from other bacterial groups. This is consistent with two possible evolutionary scenarios: (1) ancestral presence and vertical inheritance within Chlorobiales (with the exception of C.thalissium) in which the gene underwent subsequent orthologous displacement; (2) independent acquisition via two independent HGT events following divergence of C.thalissium from the ancestor lineage of other Chlorobiales. In either scenario, the C.thalissium copy of the gene was likely acquired from within gammaproteobacteria; some members of Chloroflexi and Acidobacteria appear to have acquired this gene within gammaproteobacteria as well (Fig 4A). Chlorobium sp. 445 and Candidatus Thermochlorobacter, which usually group with C.thalissium on other trees, do not for this gene, which further indicates an orthologous displacement for C.thalissium after its divergence from the photoheterotrophic lineages. The broad and sparse taxonomic distribution of sequences grouping with crown Chlorobiales (Fig 4B) prevents the identification of a likely HGT donor group. or an inference of vertical inheritance in stem Chlorobiales, even though some Ignavibacteria sequences are present.

Fig 4

Maximum likelihood (ML) trees of fumarate hydratase homologs.

(A-B) Two groups of sequences related to Chlorobiales query genes were identified, shown as midpoint rooted trees with collapsed clades labeled with tax¬onomic group names. (C) Higher resolution tree of crown Chlorobiales sequences from B. Support values indicate approximate likelihood ratio test (aLRT)/ bootstrap (100 replicates). Major bipartitions with bootstrap (BS) support are labeled. The full trees contain 250 and 242 taxa, respectively. Color bars to the right of the tree indicate autotrophic (green), heterotrophic (orange), or undetermined (white) carbon metabolisms. Clades including multiple carbon metabolisms are indicated with diagonally hatched boxes.

Fumarate reductase

The ML tree of fumarate reductase homologs recovers the monophyly of Chlorobiales with the exception of the thermophilic photoheterotrophs Thermochlorobacter, and Chlorobium sp.445. Observed sequences are closely related to those from Candidatus Lambdaproteobacteria, Geobacter, and Desulfuromonadales bacterium (Fig 5). These are distantly related by a long branch to other bacterial sequences, including members of Firmicutes, Actinobacteria, and other Proteobacteria, indicative of an HGT into stem Chlorobiales, although the donor group cannot be discerned. The short branch separating the Chlorobiales and proteobacterial sequences indicates a history of relatively recent HGT involving stem Chlorobiales, although the absence of a polarizing outgroup prevents the direction of HGT from being established. A stem Chlorobiales lineage could have been the primary HGT donor to this group of Proteobacteria, or vice versa. Both groups could, alternatively, have been HGT recipients from an unsampled or extinct lineage, potentially explaining the long empty branch preceding this grouping.

Fig 5

Maximum likelihood (ML) tree of fumarate reductase homologs.

(A) midpoint-rooted tree with collapsed clades labeled with taxonomic group names. (B) Higher resolution tree of crown Chlorobiales and closely related proteobacterial sequences. Support values indicate approximate likelihood ratio¬¬ test (aLRT)/ bootstrap (100 replicates). Major bipartitions with bootstrap (BS) support are labeled. The full tree contains 248 taxa. Color bars to the right of the tree indicate autotrophic (green), heterotrophic (orange), or undetermined (white) carbon metabolisms. Clades including multiple carbon metabolisms are indicated with diagonally hatched boxes.

Succinyl-CoA synthetase

The ML tree for succinyl-CoA synthetase recovers the monophyly of Chlorobiales with a well-supported sibling group including Cytophagales, Marinifilaceae, other Bacteroidetes, and Ignavibacteria (Fig 6). While this outgroup contains similar taxa to the species tree outgroup of Chlorobiales, the phylogeny shows a mixing of Ignavibacteriales and Bacteroidetes groups that prevents inference of vertical inheritance in the Chlorobiales stem ancestor, or identification of an HGT donor group. Photoheterotrophic Chlorobiales have also retained this gene, indicating that its presence is not strictly indicative of a functional rTCA cycle. Succinyl-CoA may be involved in any number of additional metabolic functions, including heme synthesis, ketone metabolism, and full or partial oxidative TCA cycles [39]. In bacteriochlorophyll biosynthesis, succinate/succinyl-CoA can be carboxylated by a partial rTCA cyle to create 2-oxoglutate as precursor molecules, providing a pre-adaptation in the form of a selective advantage to photoheterotrophic lineages [35, 40]. More distantly related outgroups on the tree also show extensive HGT between bacterial groups, including Calditrichaeota, Firmicutes, Acidobacteria, and Proteobacteria.

Fig 6

Maximum likelihood tree of succinyl-CoA synthetase homologs.

(A) midpoint-rooted tree with collapsed clades labeled with taxonomic group names. (B) Higher resolution tree of crown Chlorobiales. Support values indicate approximate likelihood ratio test (aLRT)/ bootstrap (100 replicates). Major bipartitions with bootstrap (BS) support are labeled. The full tree contains 248 taxa. Color bars to the right of the tree indicate autotrophic (green), heterotrophic (orange), or undetermined (white) carbon metabolisms. Clades including multiple carbon metabolisms are indicated with diagonally hatched boxes.

2-oxoglutarate: Ferredoxin oxidoreductase

The tree for 2-oxoglutarate: ferredoxin oxidoreductase alpha subunit recovers the monophyly of Chlorobiales, grouping within a taxonomically broad range of Bacteroidetes (Fig 7). The placement of the specific orders Chitinophaga, Flavobacteriales, Sphingobacteriales, and Balneolaceaes within the tree suggests similarity to published species trees of Bacteroidetes [33], and is further indicative of a likely HGT from within Bacteroidetes. The ML tree does not recover the sibling grouping of C.thalassium and thermophilic photoheterotrophic lineages, suggesting a possible misrooting of Chlorobiales in this tree.

Fig 7

Maximum likelihood (ML) tree of 2-oxoglutarate: Ferredoxin oxidoreductase homologs.

(A) midpoint-rooted tree with collapsed clades labeled with taxonomic group names. (B) Higher resolution tree of crown Chlorobiales. Support values indicate approximate likelihood ratio test (aLRT)/ bootstrap (100 replicates). Major branches with bootstrap (BS) support are labeled with respective values. Tree A contains 243 taxa. Color bars to the right of the tree indicate autotrophic (green), heterotrophic (orange), or undetermined (white) carbon metabolisms. Clades including multiple carbon metabolisms are indicated with diagonally hatched boxes.

Isocitrate dehydrogenase

The isocitrate dehydrogenase phylogeny recovers a polyphyletic placement of sequences within Chlorobiales, with Chlorobium, Pelodictyon, and Chlorobaculum genera and Prosthecochloris genera forming two distinct groups within one set of identified homologs (Fig 8A) and C.thalassium, Thermochlorobacter, and Chlorobium sp.445 group within another set of identified homologs (Fig 8B), providing clear evidence of multiple HGT events for this gene family. Specific HGT donor groups to Chlorobiales cannot be clearly identified from these trees. The placement of closely related Chlorobiales genera in Fig 8A is consistent with vertical inheritance in this group, with subsequent multiple HGTs to other bacterial groups, including Deltaproteobacteria, Desulfobulbaceae bacterium, and Marinifilaceae. These distinct gene histories suggest that the clades within Chlorobiales shown in Fig 8C and Fig 8D acquired isocitrate lyase independently after they diverged.

Fig 8

Maximum likelihood (ML) trees of isocitrate dehydrogenase homologs.

Two groups of sequences related to Chlorobiales query sequences were identified, shown as midpoint rooted trees with collapsed clades labeled with taxonomic group names (A-B). (C-D) Higher resolution trees of crown Chlorobiales groups. Support values indicate approximate likelihood ratio test (aLRT)/ bootstrap (100 replicates). Major bipartitions with bootstrap (BS) support are labeled. A and B trees contain 250 and 232 taxa, respectively. Color bars to the right of the tree indicate autotrophic (green), heterotrophic (orange), or undetermined (white) carbon metabolisms. Clades including multiple carbon metabolisms are indicated with diagonally hatched boxes.

Aconitate hydratase

Aconitate hydratase sequences within Chlorobiales are polyphyletic with C. thalassium, Thermochlorobacter, and remaining taxa each grouping with different sets of bacterial sequences. The main group (Fig 9C) places within Epsilonproteobacteria, suggesting this is the HGT donor group to the major clade of crown Chlorobiales as well as the HGT donor to a subset of Aquificales, a frequency observed HGT partner with Epsilonproteobacteria. However, long branches and sparse taxonomic representation prevent a reliable rooting and clear identification of a donor clade within Epsilonproteobacteria. The absence of aconitate hydratase within other photoheterotrophic lineages can be explained by two possible scenarios: (1) initial acquisition in the common ancestor lineage of Chloroherpetonaceae, with subsequent loss in the photoheterotroph ancestor; or (2) acquisition by C. thalassium after the divergence of the photoheterotroph ancestor lineage. Both scenarios include a subsequent independent acquisition by Thermochlorobacter.

Fig 9

Maximum likelihood (ML) tree of aconitate hydratase homologs.

Two groups of sequences related to Chlorobiales query sequences were identified, shown as midpoint rooted trees with collapsed clades labeled with taxonomic group names (A-B). A higher resolution tree for Chlorobiales sequences in tree B is shown in (C). Support values indicate approximate likelihood ratio test (aLRT)/bootstrap (100 replicates). Major bipartitions with bootstrap (BS) support are labeled. A and B trees contain 249 and 250 taxa, respectively. Color bars to the right of the tree indicate autotrophic (green), heterotrophic (orange), or undetermined (white) carbon metabolisms. Clades including multiple carbon metabolisms are indicated with diagonally hatched boxes.

In the case of (1) the photoheterotroph clade can be inferred to be ancestrally autotrophic, as all rTCA cycle genes would be present in their common ancestor; in the case of (2) their common ancestor would be inferred to be lacking a complete rTCA cycle, and therefore extant photoheterotrophic lineages would represent a continuation of an ancestral metabolic state. However, this latter scenario is less likely, as all rTCA enzymes but aconitate hydratase would be ancestrally present, a “nearly complete” rather than merely a “partial” rTCA cycle that would not have a known modern analog.

ATP citrate lyase

More than other enzymes, ATP citrate lyase is considered diagnostic for the presence of rTCA cycle carbon fixation, as it plays the crucial step of citrate cleavage into acetyl-CoA and oxaloacetate [41]. A variant of this process is found in Aquificae, where cleavage is catalyzed in tandem by citryl-CoA synthetase and citryl-CoA lyase [42]. These enzymes share a distant common ancestry indicated by sequence homology as well as conserved structural elements to the two subunits of ATP citrate lyase in Chlorobium limicola, specifically, a two-helix stalk and β-hairpins [43] (S2 Fig).

In our results, all autotrophic Chlorobiales form a single group in the ML phylogeny of both the α (Fig 10) and β (S1 Fig) subunits of this protein, which includes Nitrospirae and Nitrospinae sequences, sibling to the deeply branching heterotrophic Chlorobium sp. 445 lineages (Fig 10B). The direction of HGT between Chlorobiales and Nitrospirae/Nitrospinae is unclear, as the recovered root of Chlorobiales is inconsistent with the species tree and varies between subunits. However, the α subunit sequence within Chlorobium sp. 445 is only partial, missing 167 amino acids (S2 Fig). This absence in the alignment may impose tree reconstruction artifacts, resulting in the spurious rooting of this group, complicating HGT inference. To test this, we re-ran the ML tree using only the sequence region present in Chlorobium sp. 445. This recovered the same phylogenetic placement of sibling groups Nitrospirae/Nitrospinae with respect to Chlorobiales. Furthermore, Bayesian inference, an approach more resistant to long branch attraction artifacts, also recovered the same rooting for both subunits (S3 Fig). If not mis-rooted, this tree topology could also be explained by an ancestral loss of ATP citrate lyase in the photoheterotroph ancestor, followed by a secondary acquisition in Chlorobium sp. 445, either from the Nitrospirae/Nitrospinae lineage, or an HGT donor group common to both, although this is a less parsimonious scenario.

Fig 10

Maximum likelihood (ML) tree of ATP citrate lyase alpha subunit homologs.

(A) Midpoint-rooted tree with collapsed clades labeled with taxonomic group names. (B) Higher resolution tree showing crown Chlorobiales and closely related sequences in Nitrospirae/Nitrospinae. Support values indicate approximate likelihood ratio test (aLRT)/ bootstrap (100 replicates). Major clades with bootstrap (BS) support are labeled with respective values. The full tree contains 247 taxa. Color bars to the right of the tree indicate autotrophic (green), heterotrophic (orange), or undetermined (white) carbon metabolisms. Clades including multiple carbon metabolisms are indicated with diagonally hatched boxes.

A mapping of this missing region to the crystal structure of the ATP citrate lyase enzyme structure shows a missing a stabilizing β hairpin region (S2 Fig), suggesting that this partial version may be nonfunctional, or retain a function unrelated to carbon fixation. This may also be evidence of independent loss of rTCA carbon fixation capability within other photoheterotrophic lineages, such as Thermochlorobacter. Still, without knowing if the truncated variant is functional, it brings the viability of ATP citrate lyase as a marker for autotrophy into question.

Other published phylogenies of ATP citrate lyase shows a history consistent with our findings, with vertical inheritance in Chlorobiales, but concluding that Nitrospirae and Nitrospinae horizontally transferred these genes to Chlorobiales, albeit with low statistical support [44]. The long branch separating these sequences from other taxa in the tree obscures the ancestry of ATP citrate lyase within the Chlorobiales stem lineage, and no clear HGT donor group can be inferred. The absence of other closely related sequences suggests an acquisition from unsampled or extinct groups. Further analyses are thus required to resolve the true history of these genes.

Pyruvate: Ferredoxin oxidoreductase

The ML tree for pyruvate: ferredoxin oxidoreductase shows very sparse taxonomic outgroups separated by a long branch indicative of independent HGT events. Distantly related groups include Bacteroidetes, Ignavibacteria, autotrophic Cyanobacteria, Chloroflexi, Verrucomicrobia, and Acidobacteria (Fig 11A). This enzyme is one of only three that are present in the rTCA cycle and not the oxidative TCA cycle [24] and was apparently lost within the photoheterotrophic Chlorobiaceae lineages, as they do not appear with other Chlorobiales in the gene tree (Fig 11B). Due to branch lengths, a putative donor group cannot be established.

Fig 11

Maximum likelihood (ML) tree of pyruvate: Ferredoxin oxidoreductase homologs.

(A) Midpoint-rooted tree with collapsed clades labeled with taxonomic group names. (B) Higher resolution tree showing crown Chlorobiales and other closely related sequences. Support values indicate approximate likelihood ratio test (aLRT)/ bootstrap (100 replicates). Major clades with bootstrap (BS) support are labeled with respective values. The full tree contains 247 taxa. Color bars to the right of the tree indicate autotrophic (green), heterotrophic (orange), or undetermined (white) carbon metabolisms. Clades including multiple carbon metabolisms are indicated with diagonally hatched boxes.

rTCA cycle proteins do not trace the evolutionary history of autotrophy

Additional evidence for the patchwork assembly of the rTCA cycle within Chlorobiales is provided by the taxonomic distribution of autotrophy across species represented in each protein tree. Surprisingly, in the majority of cases the most closely related outgroup sequences are not from autotrophic species, suggesting that these proteins have an alternative function in these groups (fumarate hydratase, succinyl-CoA synthetase, OGOR). Presumably, before the complete rTCA pathway was present, these proteins performed similar functions in Chlorobiales as part of a photoheterotrophic metabolism.

Further evidence of the potential utility of incomplete sets of rTCA cycle proteins is provided by the retention of several of these proteins within photoheterotrophic Chlorobiales; in some cases, it appears these proteins were acquired after the divergence from other Chlorobiaceae (fumarate hydratase, isocitrate lysase, aconitate hydratase) while others were vertically inherited within the crown group (malate dehydrogenase, succinyl-CoA synthetase, 2-oxoglutarate: ferredoxin oxidoreductase). In all these cases, the genes present are also utilized in the oxidative TCA cycle, thus potentially explaining their retention in the divergent thermophiles. Elucidating their functions within these groups may provide clues to the ancestral functions of these genes in stem Chlorobiales, before a complete rTCA cycle was present. The inferred presence of ATP citrate lyase in stem Chlorobiales, in the absence of a complete and functional rTCA cycle, further suggests that this enzyme is not, on its own, diagnostic for autotrophy in this group, as is observed for other heterotrophic microbes [45, 46].

Discussion

A Chimeric origin of rTCA cycle genes and autotrophy within Chlorobiales

Our results show a striking diversity of the evolutionary histories of rTCA cycle enzymes within Chlorobiales, suggesting that carbon fixation has a complex and relatively recent evolutionary history within this group (Fig 12). There is strong evidence for the independent acquisition of several enzymes via HGT from different donors, both to the stem group, and to different lineages within Chlorobiales following their crown divergence. While current taxon sampling and the inherent limitations of phylogenetic inference prevent, in most cases, the identification of specific HGT donor groups for these genes, their collective histories are unambiguously incompatible with the alternative hypothesis of vertical inheritance from stem Chlorobiales. This interpretation is consistent with previous phylogenetic trees in which the inheritance of rTCA enzymes could not be narrowed down to a specific donor group or transfer event [44, 47].

Fig 12

16S rRNA tree of Chlorobiales and Ignavibacterium (outgroup) with gene histories of rTCA proteins mapped.

Arrows indicate inferred acquisitions of specific enzymes, and subsequent losses are indicated with crosses with enzyme names coded by color. Shaded fields indicate inferred carbon metabolisms for different clades. Divergence times labelled by respective nodes [13].

In several cases, C. thalassium and thermophilic organisms have orthologs of rTCA genes not closely related to those in other members of Chlorobiales, clearly indicating a more recent history of HGT for one or more subgroups. From these observations, two potential hypotheses emerge regarding the evolution of the rTCA cycle in Chlorobiales: (1) the full rTCA cycle was acquired in the stem lineage, with some enzymes replaced via additional HGTs in descendant crown group lineages; or (2) the full rTCA cycle was only assembled later, within crown Chlorobiales, with independent acquisition of some enzymes by multiple lineages. We find the latter explanation to be more parsimonious. Only for the fumarate hydratase tree is (1) a reasonable hypothesis (Fig 4A). When Candidatus Thermochlorobacter and Chlorobium sp. 445 appear with C. thalassium in the isocitrate lysase (Fig 8A) and aconitate hydrase (Fig 9A) trees, this implies (2) an independent acquisition via two HGT events following the divergence within the family of Chlorobi. These trees each have variable outgroups which include both autotrophic and heterotrophic bacteria/archaea.

rTCA cycle genes within photoheterotrophic lineages

A functional rTCA cycle was likely originally present in the common ancestor of photoheterotrophic thermophilic Chlorobiales lineages. In several gene trees showing HGT within crown Chlorobiales, these group together with C. thalassium, indicating a likely shared ancestral state of autotrophy. Other genes present in a complete rTCA path were apparently lost in thermophilic members, consistent with their photoheterotrophic physiology. For example, Candidatus Thermochlorobacter aerophilum and Chlorobium sp. GBChlB do not have genes encoding ATP citrate lyase, an essential enzyme in the pathway (Fig 10B). Alternative genes to ATP citrate lyase that may rescue rTCA cycle function (citryl-CoA synthetase and citryl-CoA lyase) [42] are also absent. The retention of a partial ATP citrate lyase within Chlorobium sp. 455 further suggests that these genes were independently lost within thermophilic lineages following their diversification.

Other genes of the rTCA cycle were apparently retained in this group, suggesting functional roles independent of carbon fixation. It is also possible that 2-oxoglutarate:ferredoxin and pyruvate:ferredoxin might be involved in mixotrophic growth in an incomplete rTCA path in these lineages [34, 35, 48].

Molecular clocks inform the history of carbon fixation within Chlorobiales

The existence of the rTCA cycle as a primordial carbon fixation pathway, mediated by non-enzymatic, inorganic catalysts, is a hotly debated issue in origins of life research [20, 49, 50]. It has also been proposed that an ancestral rTCA was present in an autotrophic last universal common ancestor (LUCA) [51, 52]. However, the phylogenomic signal needed to establish this deep ancestry may be impossible to fully elucidate, given the relatively shallow taxonomic distributions of rTCA cycle constituent proteins, and high frequency of HGT between phyla, as evidenced by reconstructed gene trees. Nevertheless, the clear history of rTCA cycle proteins within Chlorobiales, and the fidelity of their inheritance in several subgroups, permits the potential dating of carbon fixation within this clade. Recent molecular clock studies have produced age estimates for major divergences within Chlorobiales informed by large protein sequence datasets, cyanobacterial fossil calibrations, and HGT constraints [13]. These ages show crown Chlorobiales diversifying between 1977 Ma and 1607 Ma, and subsequent diversification of Chlorobiaceae between 1300 Ma and 988 Ma (Fig 12. While these age uncertainties are large and will likely be revised in future studies with additional constraints, they are consistent with BCF lipid biomarkers at 1.64 Ga most likely being produced after the crown diversification of Chlorobiales but before the diversification of major extant autotrophic groups. These phylogenetic analyses therefore raise the intriguing possibility that BCF lipid biomarkers were produced by early diverging lineages that did not yet have a complete rTCA cycle. If that is indeed the case, they would not preserve C isotopic fractionations consistent with Chlorobiales carbon fixation under rTCA (Δδ13C -20 to -10%) [53], but rather should show fractionations more like those observed for extant photoheterotrophic lineages, and/or BCF bulk organic carbon (Table 2). Furthermore, an isotopic shift in preserved biomarkers to modern values reflective of carbon fixation via an rTCA cycle should be expected to occur before 988 Ma [13]. These interpretations may be complicated by the existence of unsampled or extinct Chlorobiales stem groups that preserved an ancestral heterotrophic metabolism long after the emergence of autotrophy within more derived groups; however, it is unlikely that such groups would be ecologically abundant following selection favoring autotrophy in these environments. Furthermore, aromatic carotenoid synthesis is absent within Chloroherpeton and photoheterotroph lineages, suggesting that the most parsimonious history of these genes would have them acquired after the crown divergence of Chlorobiaceae. This argues against the BCF biomarkers being sourced from stem group Chlorobiales, unless there were subsequent losses of biomarker synthesis genes within the crown group.

Table 2

Isotopic differences by carbon fixation pathways and taxa [54–64].

Pathway	Taxa	Δδ13C%
CBB	Cyanobacteria, Eukaryota, Proteobacteria (Alpha,Beta,Gamma), Oschillochloridaceae	−10 → −35
rTCA	Chlorobiales, Aquificales, Nitrospirae, Nitrospinae	−2 → −23
Wood-Ljungdahl	Euryarcheota (methanogens), Planctomycetes, Deltaproteobacteria, Spirochaetes	−5 → −80

Table notes: Sourced as Δδ13C = δ13Cbiomass − δ13Ccarbon source calculation. Listed taxa are not diagnostic of full pathway distribution, but major groups of relevance. There is a 25% mean deviation between inorganic and organic carbon isotopes [65].

Autotrophy in Chlorobiales: Adaptation to a changing Mesoproterozoic world?

The independent assembly of a complete rTCA cycle for carbon fixation in at least two major lineages of Chlorobiales is potentially indicative of broad changes in environmental conditions and ecological niches during the Mesoproterozoic. Because of their unique metabolisms and chemical requirements, Chlorobiales are useful indicators of past environmental conditions [66], most notably for the changes in sulfur abundance and redox state through Earth history. The Mesoproterozoic was a time of complex interactions between sulfur and oxygen cycles, potentially spurring the metabolic evolution of anoxygenic phototrophs such as Chromaticeae and Chlorobiales. Redox-sensitive element abundances and stable isotopes indicate progressive oxygenation of the ocean–atmosphere system, along with a significant increase in marine sulfate inventory during this time [1, 2, 67–69]. This increased oxygenation of the atmosphere promoted weathering of pyrite and delivery of sulfate to aquatic environments, subsequently increasing the prevalence of euxinic conditions via microbial sulfate reduction [70, 71]. At the same time, increased oxygenation of surface waters would drive anoxygenic phototrophs deeper in water columns where these anoxic conditions could persist; these lower light environments would select for more efficient light-harvesting systems, such as chlorosomes [72], and more energy efficient metabolic pathways for processing carbon. This may also explain why Chlorobiales and Chromatiales don’t have shared carbon fixation pathways despite their similar physiology and environmental occurrence. Chromatiales are more oxygen-tolerant and can exist higher in the water column where light is more available for energy-intensive carbon fixation via the CBB cycle, requiring seven ATP per molecule of synthesized pyruvate, compared with only two for the rTCA cycle [41].

In such a scenario, the HGT donors of rTCA cycle genes to Chlorobiales would also have inhabited these deeper aquatic environments. Unfortunately, the phylogenies of rTCA cycle proteins do not identify putative HGT donor groups with sufficient resolution to evaluate this hypothesis with the present dataset. Future environmental sequencing investigations of stratified marine environments may discover missing microbial diversity that better resolves these gene tree histories; alternatively, these microbial groups and their metabolisms may have been unique to stratified Proterozoic marine conditions, and have since gone extinct, or have lost these genes. While this scenario may explain why Chlorobiales acquired carbon fixation via rTCA cycle genes rather than CBB cycle genes, it does not, on its own, explain why the shift to autotrophy occurred during this time. One possibility may once again relate to increasing oxygen levels. If aquatic aerobic/sulfate reducing metabolisms also experienced a diversification and ecological expansion at this time, then, presumably, more organic carbon would be microbially oxidized, leaving less available for assimilatory heterotrophic processes [73]. These conditions would therefore favor more widespread autotrophy, even among energy-limited organisms such as Chlorobiales.

Conclusion

Our proposed scenario for the history of autotrophy within Chlorobiales infers early members of this group were heterotrophs and that autotrophy was independently acquired within crown Chlorobiales lineages through a chimeric history of acquisition of rTCA genes. Therefore, early members of Chlorobiales fractionated carbon differently than extant autotrophic members that have a complete rTCA cycle, a transition that is predicted to be apparent within the isotopic record of lipid biomarkers. This specifically implies that a “GSB-like” carbon isotopic fractionation within preserved aromatic carotenoid biomarkers should not be the standard of evidence to infer Chlorobiales as their biological source. Rather, we would expect that older carotenoid material, such as that obtained from the BCF, should instead show fractionations consistent with heterotrophy, closer to that of bulk organic carbon within the system. This is especially important given the proposed alternative cyanobacterial origin for these lipids, which may be falsely inferred in the absence of observing the expected “rTCA” signature fractionation. Molecular clock studies of Chlorobiales constrain the timing of these metabolic evolutionary events, providing the means to integrate these genomic and geochemical records to establish a well-resolved evolutionary ecology of Chlorobiales. This work demonstrates how phylogenomic analysis can provide an independent source of information for interpreting the lipid biomarker record, and the importance of integrating phylogenomic data with stable carbon isotope analysis in inferring the evolutionary history of microbial metabolisms.

Enzymes, NCBI accessions, Genome IDs, alignment, taxonomic diversity, and ML tree model for query sequences.

(PDF)

Click here for additional data file.

Maximum likelihood (ML) tree of ATP citrate lyase beta subunit homologs.

(A) midpoint-rooted tree with collapsed clades labeled with taxonomic group names. (B) Higher resolution tree showing crown Chlorobiales and closely related sequences in Nitrospira/Nitrospinae. Support values indicate approximate likelihood ratio test (aLRT)/ bootstrap (100 replicates). Major clades with bootstrap (BS) support are labeled with respective values. Color bars to the right of the tree indicate autotrophic (green), heterotrophic (orange), or undetermined (white) carbon metabolisms.

(TIF)

Click here for additional data file.

Protein Structural map of Chlorobium Limicola ATP Citrate Lyase enzyme.

Structural and labeled functional regions regions are colored corresponding to bars above sequence alignments. The amino acid alignment includes are sequences from C.thalassium, C.limicola, Chlorobium sp.445. [23].

(TIF)

Click here for additional data file.

Bayesian Inference (BI) consensus trees of ATP citrate lyase alpha (A, C) and beta (B, D) subunit homologs.

Trees depicted without (A, B) and with (C, D) Nitrospinae metagenomic sequences from groundwater metagenomic biosampling included in the alignment depicted with arrow. Collapsed clades labeled with taxonomic group names. Support values show consensus posterior probabilities.

(TIF)

Click here for additional data file.

Transfer Alert

This paper was transferred from another journal. As a result, its full editorial history (including decision letters, peer reviews and author responses) may not be present.

10 Jun 2022

Submitted filename: Paoletti & Fournier Report.pdf

Click here for additional data file.

1 Aug 2022

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: N/A

Reviewer #2: Yes

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copy edit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The manuscript of Paoletti and Fournier explores the phylogenetic history of the reverse TCA cycle in the order Chlorobiales. Biomarkers, mostly derived from aromatic carotenoids once believed to be specific markers for phototrophiic sulfur bacteria, have been observed in the rock record extending back about 1.64 Ga. However, recent studies have shown that aromatic carotenoids are also produced by many modern-day cyanobacteria. To properly interpret isotopic fractionation of carbon in biomarkers requires knowledge of the carbon fixation pathways in use by organisms that produced the biomarkers. Thus, this is a timely study that will be of interest to a wide audience. Comments for the authors follow.

1. Lines 31 to 33. Considering that renieratene and isorenieratene have recently been shown to be produced by cyanobacteria that can also produce synechoxanthin, okenone may be the only reliable biomarker associated with (purple) sulfur bacteria. Although chlorobactene has not been shown to be produced by cyanobacteria yet, any cyanobacterium that can synthesize isorenieratene could in principle make chlorobactene, assuming the carotene desaturase could function with gamma-carotene as substrate.

We thank the reviewers for pointing out this detail, and now expand our discussion of these biomarkers in Cyanobacteria to mention the possibility of ancient cyanobacterial sources of chlorobactene. We have added the text:

“Carotenoid biomarkers within Neoproterozoic marine sediments (1000-542 Ma) have been shown to be primarily comprised of cyanobacterial renierapurpurane, with small amounts of the Chlorobiaceae-associated biomarker isorenieratane. Furthermore, cyanobacteria producing synechoxanthin have been shown to be capable of producing both isorenieratene and, in principle, chlorobactene, given their enzymatic repertoire (cite?). This brings the evidence for GSB at 1.64 Ga BCF formations into question, as well as the practice of using these biomarkers as a calibration for dating Chlorobiales in molecular clock analyses, since cyanobacteria could potentially be the source of the carotenoid derivatives found in these deposits12,13” in lines 16-24 of page 2.

2. Lines 50-51. The CBB cycle also occurs in many chemolithoautotrophic bacteria, not just photoautotrophs.

The reviewer is indeed correct, we now clarify this detail:

Page 2, line: 36-39: “The CBB cycle is found in both photoautotrophic bacteria such as cyanobacteria and purple sulfur bacteria, as well as some chemolithoauthotrophs; it fixes three molecules of CO2 for the synthesis of one 3-phoshoglycerate molecule using ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCo).”

3. Lines 71-73. I am not sure of the “official” state of taxonomy concerning the order Chlorobiaceae, but I think there are only three genera (see Imhoff 2003) now: Prosthecochloris, Chlorobium and Chlorobaculum. Pelodictyon (note spelling) and Ancalochloris are no longer considered to be valid taxa. The status of Chloroherpeton is less clear, but it is probably the type genus of a separate family, because Chloroherpeton are metabolically distinct from all other green sulfur bacteria in that they can only oxidize sulfide to the level of sulfur/polysulfide.

The reviewer is correct. The taxonomic designations used in our original manuscript were based on (Imhoff, 2003). The author of this work has since published an updated classification schema for GSB solely using the genera Chlorobium, Chlorobaculum, Prosthecochloris, and Chloroherpeton. The change is noted in our reference #30 on page 17 and in the manuscript page 3, lines 63-65.

There has been some recent debate due from new isolates, particularly from environmental metagenomic sequences, grouping within the more metabolically distinct Chloropeton genera for further revisions to this taxonomic labeling, such as creating distinct families of Chloroherpetonaceae and Thermochlorobacteriacea (Liu et. al., 2012). However, since these labels have not been adopted in the larger field yet and much debate remains on the physiologies of characterized photoheterotrophs in the order Chlorobiales, we retain the taxonomic labels suggested by Imhoff. The taxonomic labels in our datasets and resulting phylogenies reflect those in use by NCBI at the time of the analysis, as these are automatically populated by our scripts.

4. Lines 73-75. The authors should probably cite Liu et al., 2012, Front. Microbiol. 3, 185. Ignavibacterium album is an aerobe that can grow as a facultative or fermentative anaerobe as well. It has a full aerobic electron transfer chain and has been demonstrated to grow by aerobic respiration.

The reviewer is correct–however, Ignavibacteriales, while the sister clade to Chlorobiales, is still quite evolutionary distant, and the relevance of their extant energy metabolisms to the ancestral state of autotrophy vs. heterotrophy in Chlorobiales is not immediately evident. Making such a connection would require a significant ancestral trait reconstruction effort, which is beyond the scope of the work presented here, and does not directly bear on the evolutionary history of the much more derived groups within Chlorobiales that we investigate.

5. Lines 77 to 79. Thermochlorobacter spp. has been cultured but not axenically, only in enrichment cultures (Tank et al. 2017). It is clearly a full aerobe with a complete electron transfer chain to match, with metabolic properties similar to those of Chloroacidobacterium. Thermochlorobacter and related taxa are unable to oxidize sulfide at all, do not fix carbon dioxide, but have a photosynthetic apparatus similar to that of GSB. These bacteria are not “sulfur bacteria” at all, unless assimilatory sulfate reduction is included in the definition, but then just about all bacteria would be sulfur bacteria.

The reviewer’s comment points out the importance in clarifying the distinction between “Green sulfur bacteria” as a phenotypic descriptor, vs. its use as an interchangeable clade name for “Chlorobiales”. In order to avoid the possible confusion between these two usages, we now only use the order name “Chlorobiales” to refer to the members of this group including the photoheterotroph clade and the family name “Chlorobiaceae” when referring to traditional Chlorobi “green sulfur bacteria” . We still use “GSB” in the introduction when referring to previously published work.

6. Lines 71 to 81. The authors do not present any analysis of the current state of the 16S rRNA phylogeny of GSB, nor do they present any phylogenetic trees based upon shared proteins among these organisms. This makes the interpretation of the trees presented for the rTCA cycle enzymes difficult for those less familiar with the taxonomy of green bacteria. A 16S rRNA tree finally appears in Figure 11.

Our understanding is that the evolutionary relationships between these groups are well-established, and agree with our own 16S tree presented here. We agree with the reviewer, that clearly communicating this species tree early in the manuscript will aid the reader in interpreting the gene trees. In the introduction, we now include a brief description of the current phylogeny of GSB, and a figure (lines 80-84) showing these relationships on pg.3 , lines 68-71 & 76-79-113:

“With the exception of the unclassified thermophiles, Chlorobiales are strictly anaerobic, obligate photoautotrophs that use the rTCA cycle. Within this group, Chlorobium, Chlorobaculum, and Prosthecochloris genera form a clade that synthesizes aromatic carotenoid compounds…This distribution of physiological traits in Chlorobiales may be explained by vertical inheritance and loss, or by HGT, depending on the inferred phylogenies of the individual gene families providing these functions.”

Fig 2. 16S rRNA tree of Chlorobiales and Ignavibacterium (outgroup) showing taxonomic distributions of a selection of traits. Labeled brackets indicate the taxonomic distribution of phototrophy and autotrophy within this group. The green box indicates groups known to synthesize aromatic carotenoid lipid biomarkers.

7. The importance of the last several points is that the earliest diverging members of the Chlorobiales and the sister taxon Ignavibacteria is that these organisms were aerobes, not anaerobes as once generally believed. This has important implications for how the authors interpret their trees and the information derived from them. Even if the authors do not agree with these data, they must present the context for their interpretation, so I would suggest that the authors begin with a section describing protein and 16S rRNA trees to establish the relationships of the organisms under consideration with other bacteria.

We now begin with such a section that clearly provides the evolutionary framework for these lineages.

8. Lines 109-111. Again, considering that I. album and Thermochlorobacter are aerobes, and that all GSB have retained the genes for at least one terminal cytochrome/quinol oxidase, it seems likely that the GSB are derived from aerobic ancestors that gained phototrophy and many other traits based on HGT.

The reviewer proposes an interesting evolutionary scenario that would reconstruct aerobic respiration as the ancestral metabolism for the Ignavibacteriales/Chlorobiales clade. While this is certainly possible, our work here is concerned with the evolution of carbon fixation within phototrophic Chlorobiales (an apomorphy-defined clade including extant Chlorobiales and their phototrophic ancestors). Reconstructing earlier evolutionary stages (pre-phototrophy) is an interesting and complex question that would require ancestral reconstruction of many gene families and traits, that is beyond the scope of the work presented here, and does not bear directly on the question of photoheterotrophy vs. photoautotrophy in ancient Chlorobiales. In other words, an early ancestral aerobic metabolism within stem Chlorobiales is not an “alternative” hypothesis to be presented, but an independent hypothesis that does not directly bear on subsequent evolutionary scenarios for carbon fixation that may have occurred within phototrophic Chlorobiales, and so does not aid in discriminating between them.

9. Figure 3. Chloracidobacterium thermophilum is not an autotroph. It is a microaerophilic heterotroph that mostly consumes amino acids, especially branched chain amino acids. It likely uses the partial reverse TCA cycle to carboxylate succinyl-CoA to make 2-oxoglutarate.

We thank the reviewer for pointing this out, and have updated Figure 3 to reflect this, as well as other figures where Acidobacteria were incorrectly labeled as autotrophs (Fig 2, Fig 3, Fig 5, Fig 7, Fig 10). All members of this phyla that have been cultured are heterotrophs (Kielak et. al., 2016).

10. Line 181. Not exactly. Thermochlorobacter doesn’t have fumarate reductase because it has an oxidative TCA cycle, not a reductive one:

The reviewer is correct in their interpretation. We have now expanded this section to specifically point out that the gene encoding fumarate reductase was apparently lost in the ancestor lineage of Thermochlorobacter and related photoheterotrophs on pg 7 lines 202-205:

“The ML tree of fumarate reductase homologs recovers the monophyly of Chlorobiales with the exception of the thermophilic photoheterotrophs Thermochlorobacter, and Chlorobium sp.445. Observed sequences are closely related to those from Candidatus Lambdaproteobacteria, Geobacter, and Desulfuromonadales bacterium (Fig 5).”

11. Figure 5B. This figure largely replicates what I expect a 16S rRNA tree would look like, and probably a tree based on concatenated core proteins as well. In panel A, I am not sure whether there are any autotrophic Acidobacteria, but I don’t think there are any.

This error has been noted and fixed as stated in comment #9.

12. Lines 206-210. Organisms that can degrade branched chain amino acids as carbon and nitrogen sources can do so by making succinate/succinyl-CoA. This can be carboxylated to produce 2-oxoglutarate by the partial reverse TCA cycle to supply precursors for Chl/BChl biosynthesis. This is very beneficial to organisms that are unable to grow as full autotrophs.

We thank the reviewer for this interesting detail. We now include a brief comment noting this possible connection on pg.7, lines 233-236: “In bacteriochlorophyll biosynthesis, succinate/succinyl-CoA can be carboxylated by a partial rTCA cyle to create 2-oxoglutate as precursor molecules, providing a pre-adaptation in the form of a selective advantage to photoheterotrophic lineages34,73.”

13. Figure 6A. Not all Chlorobiales are autotrophs (see panel B below).

The error in the figure labeling the Chlorobiales collapsed branch as all autotrophs has been noted and changed in panel A to correctly depict it a mixed grouping.

14. Line 218 to 225. Presumably this refers to Figure 6, but panel 6B looks like a 16S rRNA tree to me with early diverging Chloroherpeton and Thermochlorobacter sequences. What is misrooted about this?

The tree is described as misrooted because, in contrast to the 16S tree and other published phylogenies, the photoheterotrophs do not group with Chloroherpeton, but are the most deeply diverging lineage. The expected root would place Chloroherpeton together with the photoheterotrophs as a clade, with other members of Chlorobiales as the sibling group. We believe this is already clearly stated in the text on pg 8 lines 263-265.

“The ML tree does not recover the sibling grouping of C.thalassium and thermophilic photoheterotrophic lineages, suggesting a possible misrooting of Chlorobiales in this tree.”

15. Figure 9A. Chlorobium sp. 445 should be orange, not green (see panel B).

Noted and updated in figure.

16. Lines 305 to 307. Seems that the ancestral state was an oxidative rather than a reductive TCA cycle, so gain more likely than loss?

In this particular case, the gene in question (ATP citrate lyase) has experienced deletions potentially indicative of a loss of function. Since other photoheterotrophs lack this gene entirely, we infer this to be a remnant that points to ancestral presence of more rTCA genes within this lineage. The remaining TCA genes in this group are now part of a partial pathway that is likely oxidative. This C. sp. 445 sequence groups basally with other Chlorobiales, consistent with a shared common ancestry and thus loss in other photoheterotropic lineages, rather than an independent gain, although more complex histories are certainly possible. We now qualify our observations by mentioning this scenario:

Lines 339-342, Page 10: “If not mis-rooted, this tree topology could also be explained by an ancestral loss of ATP citrate lyase in the photoheterotroph ancestor, followed by a secondary acquisition in Chlorobium sp. 445, either from the Nitrospirae/Nitrospinae lineage, or an HGT donor group common to both, although this is a less parsimonious scenario.”

17. Lines 330 to 332. Is pyruvate dehydrogenase present in the aerobic heterotrophs?

As far as our results show, pyruvate dehydrogenase is not present in the Chlorobiales heterotrophs, which we believe is demonstrated by our tree Fig 11 and now made clearer in the text pg. 10, lines 373-374: “This enzyme is one of only three that are present in the rTCA cycle and not the oxidative TCA cycle24 and was apparently lost within the photoheterotrophic Chlorobiaceae lineages, as they do not appear with other Chlorobiales in the gene tree (Fig 10B).”

18. Figure 11. This is an interesting figure, but it could perhaps be made still more informative by including some other events as well. For example, all the organisms in the gray shading acquired the genes for type-1 reaction centers, FMO bacteriochlorophyll a binding protein, and the genes for bacteriochlorophyll c biosynthesis and chlorosome assembly. The Chlorobiaceae gained the ability to oxidize sulfur/polysulfide to sulfate, and Chlorobaculum gained the ability to oxidize thiosulfate to sulfate.

The acquisition of these additional traits are indeed important and interesting parts of the evolution of Chlorobiales. In fact, our figure already denotes the acquisition of type-1 reaction centers (PS1), and the acquisition of bacteriochlorophyll synthesis and binding proteins is implied by the “Photoheterotrophy” label. We agree that “chlorosomes” is a valuable addition to this figure, as the role of chlorosomes is specifically discussed in the manuscript. We now see that “PS1” is insufficiently descriptive. Therefore, we have updated this figure to more precisely and less redundantly describe the crown Chlorobiales box as containing “type-1 reaction centers, bacteriochlorophyll, chlorosomes”. Other physiological traits relating to sulfur oxidation are less directly connected to the results and narrative of the paper, and we have opted to not include these, to keep the figure as simple as readable as possible.

19. Most (but not all) members of the Chlorobiaceae can synthesize chlorobactene or isorenieratene; a few seem to have lost CrtQ. Chloroherpeton and Thermochlorobacter cannot make aromatic carotenoids, and that seems to be trait that was acquired relatively late by GSB. However, exactly when this trait was acquired is not clear. All strains in the shaded area, as well as I. album, can synthesisize carotenoids, which are essential for organisms that use chlorophylls to perform phototrophy in the presence of oxygen.

We agree, our current description as “carotenoid pigments” was insufficiently precise. We agree, our current description as “carotenoid pigments” was insufficiently precise. We have removed this label in the discussion figure (Fig. 12). We now include this correctly labeled group

"aromatic carotenoid biosynthesis" in the newly added overview Figure 2. We have also added a section to the discussion to clarify how the taxonomic distribution of biomarker synthesis impacts these hypotheses:

Lines 490-495 Page 13: “Furthermore, aromatic carotenoid synthesis is absent within Chloroherpeton and photoheterotroph lineages, suggesting that the most parsimonious history of these genes would have them acquired after the crown divergence of Chlorobiaceae. This argues against the BCF biomarkers being sourced from stem group Chlorobiales, unless there were subsequent losses of biomarker synthesis genes within the crown group.”

Reviewer #2: This work presents an extensive phylogenetic study of all enzymes involved in the rTCA cycle of Chlorobiales and closely related species. the work is framed within the context of the origin of autotrophy within this group and its potential implications for the use of the ~1.64 biomarkers traditionally used as a calibration point for molecular clock analyses.

From a phylogenetic perspective, the work is well done and comprehensive. I am not aware of other studies that have analyzed all the enzymes in this pathway and, therefore, it adds an important piece to reconstruct its evolutionary history. The methodology used to infer the phylogenies is standard and appropriate. However, it would have been useful to see a discussion on the accuracy of all the phylogenies used as basis for the conclusions drawn. Prokaryote phylogenies are notoriously sensitive to many factors, from alignment length to species composition. Most bootstrap values, especially those of the chlorobiales/sister group node, are relatively high, which suggest these trees can be used with a decent amount of confidence. However, I think it is risky to draw such wide-ranging conclusions without even discussing potential pitfall in phylogenetic reconstruction.

The reviewer is correct in that tree inferences are often limited by the accuracy of phylogenetic reconstruction, as impacted by the available sequence data and models we have. These pitfalls relate to interpreting topology, when alternative topologies not recovered by the phylogeny may also be potentially correct, leading to different conclusions. In the work presented here, for many of these gene families, the two hypotheses being compared are (1) shared common ancestry and vertical inheritance of the gene within crown group Chlorobiales; vs. (2) independent acquisition via HGT by different sub-groups within Chlorobiales, and no shared common ancestry within crown group Chlorobiales. In the cases where we propose (2), the homologs found within different groups of Chlorobiales are so distantly related that it is extremely unlikely that the observed polyphyly is a result of phylogenetic uncertainty or reconstruction error. In other words, even with the limitations of phylogenetic reconstruction, these gene histories adequately discriminate between these two hypotheses. However, the inferred HGT donor groups in these cases may be impacted by the kinds of uncertainty mentioned by the reviewer. Therefore, we now clarify that both taxon sampling and phylogenetic reconstruction limitations may impact the exact placement of these groups, limiting our ability to conclusively identify HGT donors or trace the deeper history of the acquisitions of these gene families.

Page 11, lines 415-418: “While current taxon sampling and the inherent limitations of phylogenetic inference prevent, in most cases, the identification of specific HGT donor groups for these genes, their histories are unambiguously incompatible with the alternative hypothesis of vertical inheritance from stem Chlorobiales.”

My major concern, that I think can be addressed easily with some clarifications, is the connection of the history of rTCA and autotrophy with the validity of the 1.64 Ga biomarker. Lines 38-40 of the introduction state that the use of these biomarkers as a calibration for dating Chlorobiales is “into question” but I was not able to see how the results presented in this work may or may not support the use of these biomarkers. This is extremely important because these biomarkers are among the very few available for prokaryote molecular clock analyses and, therefore, any alteration of their use would have profound impacts on the field.

Thank you for pointing out that additional clarity is needed here. We expanded our conclusions to now more explicitly make predictions and emphasize the significance of our findings with respect to interpreting the carotenoid biomarker record:

Lines 544-552, Page 14: “This specifically implies that a “GSB-like” carbon isotopic fractionation within preserved aromatic carotenoid biomarkers should not be the standard of evidence to infer Chlorobiales as their biological source. Rather, we would expect that older carotenoid material, such as that obtained from the BCF, should instead show fractionations consistent with heterotrophy, closer to that of bulk organic carbon within the system. This is especially important given the proposed alternative cyanobacterial origin for these lipids, which may be falsely inferred in the absence of observing the expected “rTCA” signature fractionation. “

Overall, I find that this work adds interesting information on the rTCA cycle evolution (provided a discussion on phylogenetic accuracy is added, as mentioned above) but its framing for its relevance to the timing of chlorobiales to be unclear. I would encourage the authors to either reconsider this framework, since the work would stand on its own without needing the molecular clock angle, or clarify its implication for the use of the 1.64 Ga biomarkers.

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Submitted filename: Responce to Reviewers(1).docx

Click here for additional data file.

9 Sep 2022

14 Sep 2022

Reviewer #1 Comment: The authors have done an excellent job of addressing my extensive

comments, with only a single exception. I don't think Chloroherpeton belongs in the Chlorobiaceae. It

should be a 2nd family, Chloroherpetonaceae, in the Chlorobiales. Other than this, I am satisfied with the

modifications made to the original manuscript, which I believe has been improved.

Response: We thank reviewer one for this point. After considering newer literature on the classification system for

Chlorobi along with NCBI taxonomic identifications, we have decided to place Chloroherpeton in the

family Chloroherpetonaceae, separate from Chlorobiaceae. Besides altering taxonomic names in the

manuscript where required and relabeling Figure 12, we believe this change does not further alter our

findings or conclusions.

Submitted filename: Response to Reviewers.pdf

Click here for additional data file.

19 Sep 2022

Chimeric Inheritance and Crown-Group Acquisitions of Carbon Fixation Genes within Chlorobiales: Origins of autotrophy in Chlorobiales and implication for geological biomarkers

PONE-D-22-08377R2

Dear Dr. Fournier,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Chih-Horng Kuo, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

22 Sep 2022

PONE-D-22-08377R2

Chimeric inheritance and crown-group acquisitions of carbon fixation genes within Chlorobiales: Origins of autotrophy in Chlorobiales and implication for geological biomarkers

Dear Dr. Fournier:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Chih-Horng Kuo

Academic Editor

PLOS ONE