Illuminating the coevolution of photosynthesis and Bacteria

Abstract

Life harnessing light energy transformed the relationship between biology and Earth-bringing a massive flux of organic carbon and oxidants to Earth's surface that gave way to today's organotrophy- and respiration-dominated biosphere. However, our understanding of how life drove this transition has largely relied on the geological record; much remains unresolved due to the complexity and paucity of the genetic record tied to photosynthesis. Here, through holistic phylogenetic comparison of the bacterial domain and all photosynthetic machinery (totally spanning >10,000 genomes), we identify evolutionary congruence between three independent biological systems-bacteria, (bacterio)chlorophyll-mediated light metabolism (chlorophototrophy), and carbon fixation-and uncover their intertwined history. Our analyses uniformly mapped progenitors of extant light-metabolizing machinery (reaction centers, [bacterio]chlorophyll synthases, and magnesium-chelatases) and enzymes facilitating the Calvin-Benson-Bassham cycle (form I RuBisCO and phosphoribulokinase) to the same ancient Terrabacteria organism near the base of the bacterial domain. These phylogenies consistently showed that extant phototrophs ultimately derived light metabolism from this bacterium, the last phototroph common ancestor (LPCA). LPCA was a non-oxygen-generating (anoxygenic) phototroph that already possessed carbon fixation and two reaction centers, a type I analogous to extant forms and a primitive type II. Analyses also indicate chlorophototrophy originated before LPCA. We further reconstructed evolution of chlorophototrophs/chlorophototrophy post-LPCA, including vertical inheritance in Terrabacteria, the rise of oxygen-generating chlorophototrophy in one descendant branch near the Great Oxidation Event, and subsequent emergence of Cyanobacteria. These collectively unveil a detailed view of the coevolution of light metabolism and Bacteria having clear congruence with the geological record.

Keywords: Bacteria; carbon fixation; photosynthesis; phototrophy.

Fig. 1.

Phylogeny of Bacteria, (B)Chl biosynthesis enzymes, and reaction centers. (A) Bacteria species tree constructed through maximum-likelihood estimation (IQ-TREE2 with universal distribution mixture model UDM0064LCLR) based on a concatenated alignment of DNA-/RNA-binding marker proteins (see SI Appendix, Fig. S4 for full tree). The rooting is based on a maximum likelihood tree of Bacteria, Archaea, and Eukarya (SI Appendix, Fig. S2). Chlorophototrophic phyla are labeled with the well-recognized nomenclature and, in parentheses, the updated nomenclature. For other lineages, GTDB classification is used. Vertical inheritance of chlorophototrophy, as predicted from the right-hand gene trees, is shown as pink branches. The number of phyla included in grouped clades (triangles) is indicated for those containing more than one phylum. Chlorobi is a phylum-level lineage according to the International Classification for Nomenclature of Prokaryotes (ICNP) but a class (Chlorobia) under Bacteroidota in GTDB r214. Key bacterial speciations are indicated with yellow numbers. (B and C) Bayesian inferences of the phylogeny of (B) BChl and Chl synthases (BchG and ChlG) rooted with archaeal geranylgeranyl diphosphate synthase and (C) Mg-chelatase isoform 1 (BchH) and 2 (BchH2) rooted using each other. Speciation, duplication, and transfers were predicted and critical events are shown with yellow indicators and numbers corresponding to the bacterial species tree (A). Vertical inheritance inferred by this is shown as pink branches. Events with low agreement on the mapping to the species tree (<75% consensus) but with high agreement on the type of event (≥75% consensus) are indicated with question marks. Events mapped to the second-deepest branch in Cyanobacteria are indicated as “1^”. In cases where DTL reconciliation maps two sequential events to the same speciation (i.e., same number), we indicate the preceding event with a gray speciation number, suggesting that the event took place prior to the corresponding speciation. All branches have posterior probabilities ≥0.9 unless otherwise noted. See SI Appendix, Figs. S6 and S7A for full trees. (D) Maximum-likelihood trees of (Left) chlorophyllide oxidoreductase COR (BchYZ) and (Middle) protochlorophyllide oxidoreductase DPOR (BchNB) each rooted with uncharacterized DPOR BchNB-related proteins of Actinobacteria (see SI Appendix, Figs. S8 and S9 for full trees and SI Appendix, Figs. S10 and S11 for justification of rooting) and (Right) 3-(1-hydroxyethyl)-bacteriochlorophyllide dehydrogenase BchC rooted with uncharacterized non-phototroph-associated proteins of COG1063 (see SI Appendix, Fig. S12 for full tree). All branches have ultrafast bootstrap support ≥0.95 unless otherwise indicated. The hypothesized vertical inheritance shown is extrapolated from the evolutionary history of BChl synthase BchG and type II reaction centers. (E and F) Bayesian inferences of the phylogeny of (E) type I reaction centers and related PSII antenna proteins rooted using Ranger-DTL v2.0 and (F) type II reaction centers rooted using minimum ancestor deviation. Topologies and rootings of the reaction center trees are congruent with a tree constructed based on structural homology between type I and II reaction centers. Evolutionary events predicted by DTL reconciliation are shown as described for (B and C). See SI Appendix, Figs. S13–S16 for full trees.

Fig. 2.

Summary of chlorophototrophy evolution and estimated timings of critical transitions in Chl-associated Terrabacteria chlorophototrophy. (A) A tree constructed using Dali based on structural homology comparison of type I and type II reaction center structures [published or, for Vulcanimicrobiota RCII subunits, predicted using AlphaFold2-multimer (score: 0.91)]. Note that the Chlorohelix RCI was not included due to a low prediction score of 0.54. (B) The LCA of all extant chlorophototrophs, an ancient Terrabacteria anoxygenic chlorophototroph (LPCA), possessed two RCs (antennaed homomeric type I reaction center LCA and antenna-less homomeric type II reaction center LCA) and unknown (B)Chl-like pigment(s). PSI and PSII core subunits have dashed fills to differentiate from RCI and RCII respectively. Type I reaction centers have fused cores and antenna, which is indicated with a dashed boundary. Vertical transmission from LPCA toward its daughter lineages, Chl- and BChl-associated Terrabacteria, are shown as thick light green and light blue arrows. Evolutionary changes in reaction center structure are indicated with thin green and blue arrows. Changes via duplication are indicated with a pair of yellow circles. Horizontal transfers of reaction centers are shown as dotted lines. *(B)Chl’s with related structure, but with an unknown hydrophobic tail. (C) Estimated timings of emergences for the LCAs (solid bar) of Chl-associated Terrabacteria, Cyanobacteria, and PSI (PsaAB) and PSII core (PsbAD) and antenna (PsbBC) pre- and postduplication. The phylogenetic relationships of each reaction center subunit are shown as black lines connecting the bars indicating the respective estimated timings. The timings for the first common ancestors (dashed bar) are also shown for Chl-associated Terrabacteria and Cyanobacteria. The timing for the rise of atmospheric oxygen is also shown.

Fig. 3.

Phylogeny of chlorophototrophs and CBB cycle marker proteins, phosphoribulokinase (PRK), and ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) large and small subunits. (A) The left-hand tree is a schematic figure of chlorophototroph evolution starting from LPCA (yellow star) with photoautotrophy marked with a yellow circle and corresponding carbon fixation pathways. The yellow star indicates a clear phylogenetic divide between BChl-associated (blue) and Chl-associated (green) Terrabacteria. The lateral gain by non-Terrabacteria lineages is shown as dotted lines. (Inset) Maximum-likelihood estimation of the phylogeny of the protein family PF00485 including PRK hitherto typically associated with archaea (purple), oxygenic chlorophototrophs (green), and Proteobacteria (red) and other kinases with activity distinct from PRK (orange). The right-hand tree is a maximum-likelihood estimation of the phylogeny of the PRK clade including sequence clusters associated with Archaea and oxygenic phototrophs (indicated with a gray background in the Inset). Sequence groups (triangles) containing Terrabacteria chlorophototrophs are bolded and groups containing nonchlorophototrophic Terrabacteria are not. Clades containing photoautotroph sequences are marked with a yellow circle. Clades that branch within other phyla are noted on the right-hand side. (B) Maximum-likelihood estimation for the phylogeny of form I RuBisCO large subunit including the “Anaero” group and homomeric form I RuBisCO progenitors (I’, I”, and I-α), and form IV RuBisCO as an outgroup (represented as a left-facing arrow). All branches with ultrafast bootstrap support less than 0.95 are collapsed as polytomies. Basal branches connecting the tree to the outgroup sequences were truncated for visual purposes (broken line). (C) Maximum-likelihood estimation for the phylogeny of form I RuBisCO small subunits corresponding to panel (B) rooted with the “Anaero” group. See SI Appendix, Fig. S21 A–C for ungrouped trees. * Gene tree clades containing ≥5 phyla.

Fig. 4.

Pre- and post-LPCA evolution of pigment synthesis. (A–C) Maximum likelihood estimation of the phylogeny of (A) 8-vinyl reductase BciB, (B) 8-vinyl reductase BciA, and (C) geranylgeranyl reductase BchP (see SI Appendix, Figs. S24 and S25 for full trees). All branches have ultrafast bootstrap support ≥0.95 unless otherwise indicated. (D) Schematic diagram for the evolutionary relationship between extant chlorophototrophs (dotted line to non-Terrabacteria indicates lateral gene transfer) with the earliest traceable points for each pigment biosynthesis protein shown (teal indicates terrabacterial ancestry and red nonterrabacterial ancestry). For those that emerged through a more ancient gene duplication, the LCA is indicated as “protein A/protein B_CA”. The proteins that are predicted to have emerged prior to LPCA (black) are mapped to the branch prior to LPCA (dashed line). (E) Pigment biosynthesis pathways and proteins that mediate the reactions are shown. The proteins are colored as described in (D). Pink arrows indicate reactions that the LCA [star in (F)] may have mediated. (F) A time-calibrated phylogenetic tree of NifK-like members and CfbD (Ni-sirohydrochlorin a,c-diamide reductive cyclase) of the nitrogenase oxidoreductase protein family (see SI Appendix, Table S5 for calibration points and SI Appendix, Fig. S27B for full tree with the calibration points). The LCA of DPOR and COR subunits is indicated (star). Duplications are indicated with asterisks. The gold bars denote the 95% highest posterior density (HPD) intervals of node age.