Pelagibacterales and Rickettsiales

Pelagibacterales and Rickettsiales are two evolutionary orders within the bacterial class Alphaproteobacteria that exemplify contrasting ecological pathways to genome reduction—free-living marine oligotrophic streamlining versus obligate intracellular parasitism—and stand at the center of phylogenomic reconstructions of the proto-mitochondrial endosymbiont. Pelagibacterales (historically designated the SAR11 clade, typified by Candidatus Pelagibacter ubique) comprises free-living heterotrophic marine bacterioplankton that constitute 25% to 50% of all microbial cells in the euphotic zone of the global oceans (~2.4 × 10²⁸ cells globally). Rickettsiales (including the families Rickettsiaceae, Anaplasmataceae, and Midichloriaceae, typified by Rickettsia prowazekii, Orientia tsutsugamushi, Ehrlichia, and the arthropod reproductive endosymbiont Wolbachia) comprises obligate intracellular bacteria that replicate inside eukaryotic host cells and include vectors of epidemic typhus and Rocky Mountain spotted fever.

Despite their opposite lifestyles, both orders evolved small, compact circular chromosomes (~1.1 to 1.5 Mb), elevated adenine-thymine (AT) nucleotide composition (65% to 72% AT), and strongly biased proteomes enriched in AT-encoded amino acids (Phe, Tyr, Met, Ile, Asn, Lys, abbreviated FYMINK). Because eukaryotic mitochondrial DNA and mitochondrion-derived nuclear genes share this exact AT/FYMINK compositional bias and accelerated substitution rates, early maximum-likelihood phylogenetics studies repeatedly grouped mitochondria with either Rickettsiales or Pelagibacterales. Subsequent development of site-heterogeneous substitution models (LG+C60+F+G4) and Dayhoff-6 amino-acid recoding demonstrated that the apparent affinity between Pelagibacterales, Rickettsiales, and mitochondria is an artifact of compositional convergence and long-branch attraction (LBA), clarifying how evolutionary models must separate ecological convergence from vertical phylogenetic signal during eukaryogenesis.

Although Candidatus Pelagibacter ubique (strain HTCC1062, 1,308,759 bp, 1,354 open reading frames) and Rickettsia prowazekii (Madrid E strain, 1,111,523 bp, 834 protein-coding genes) possess similarly small genomes compared with metabolically versatile free-living Alphaproteobacteria such as Bradyrhizobium diazoefficiens (~9.1 Mb), the population-genetic forces driving their genome reduction are diametrically opposed:

1. Selection-Driven Streamlining in Pelagibacterales (N_e ≈ 10⁹–10¹⁰): Under the genome streamlining hypothesis (Giovannoni et al., 2005), Pelagibacter inhabits ultra-oligotrophic ocean surface waters depleted in dissolved nitrogen and phosphorus. Because its effective population size (N_e) is vast, purifying selection is extraordinarily efficient (N_es ≫ 1 even for minuscule selective coefficients s) and eliminates every non-essential nucleotide to minimize the energetic and elemental (N and P) cost of DNA replication and cell division. Consequently, the 1.31 Mb P. ubique genome contains zero pseudogenes, zero transposons or insertion sequences, zero introns, a single ribosomal RNA operon, and a median intergenic spacer length of just 3 base pairs (~97% coding density). Despite its small size, P. ubique retains complete prototrophic pathways for synthesizing 19 of the 20 amino acids (requiring exogenous reduced sulfur as dimethylsulfoniopropionate or methionine, and glycine/serine precursors) alongside proteorhodopsin, a retinal-binding light-driven proton pump that supplements oxidative phosphorylation during carbon starvation.
2. Drift-Driven Reductive Evolution in Rickettsiales (N_e ≈ 10²–10⁴): In contrast, Rickettsia prowazekii, Anaplasma, and Wolbachia experience severe transmission bottlenecks as they pass between arthropod vectors (lice, ticks, mites) and vertebrate or insect host cells. These bottlenecks drastically reduce N_e, amplifying genetic drift and triggering Muller's ratchet—the irreversible stochastic accumulation of slightly deleterious mutations in asexual populations lacking horizontal recombination. Because the host cytosol provides preformed amino acids, nucleoside triphosphates, and cofactors, bacterial biosynthetic operons become neutral targets for mutational inactivation. The 1.11 Mb R. prowazekii genome is therefore 24% non-coding (median intergenic spacer ~150 bp), littered with decaying pseudogenes undergoing gradual mutational deletion (AT mutational bias), and stripped of glycolysis, amino-acid biosynthesis, and de novo nucleotide synthesis. To survive, Rickettsia imports host cytosolic adenosine triphosphate (ATP) in exchange for bacterial ADP using five paralogous Tlc (NTT) ATP/ADP translocases.

Why did Pelagibacterales and Rickettsiales both converge on ~70% AT composition when their evolutionary regimes are opposite? In Rickettsia and endosymbionts, loss of DNA repair enzymes (such as mutY/mutM or uracil-DNA glycosylase) permits the universal spontaneous cytosine deamination mutation bias (C:G → T:A) to drift to fixation under low N_e. In Pelagibacter, by contrast, AT base pairs require 7 fewer nitrogen atoms per base pair than GC base pairs (an A:T pair contains 7 N atoms vs. 8 N atoms in a G:C pair, plus lower GTP synthesis overhead), so natural selection in nitrogen-starved surface oceans actively favors AT-rich codons.

Regardless of its cause, shifting genomic GC content below 30% systematically overwrites the first and second codon positions of protein-coding genes, driving genome-wide replacement of GC-encoded amino acids (GARP: Glycine GGN, Alanine GCN, Arginine CGN, Proline CCN) with AT-encoded amino acids (FYMINK: Phenylalanine UUY, Tyrosine UAY, Methionine AUG, Isoleucine AUH, Asparagine AAY, Lysine AAR). Because mitochondrial genomes and fast-evolving mitochondrial proteins also exhibit elevated FYMINK proportions (~35–42% vs. ~20–24% in balanced GC genomes), standard time-homogeneous, site-homogeneous amino-acid substitution matrices (such as WAG, JTT, or LG) misinterpret shared FYMINK residues at homologous alignment columns as synapomorphies inherited from a recent common ancestor, pulling mitochondria, Pelagibacterales, and Rickettsiales together via long-branch attraction (LBA).

Resolving the true placement of Pelagibacterales, Rickettsiales, and mitochondria within the phylogenetic tree of Alphaproteobacteria required three methodological advances in phylogenomics (Viklund et al., 2012;Rodríguez-Ezpeleta & Embley, 2012; Martijn et al., 2018; Muñoz-Gómez et al., 2022):

1. Site-Heterogeneous Mixture Models (CAT-GTR and LG+C60+F+G4): Standard empirical matrices (LG) assume that all 20 amino acids are potentially acceptable at every alignment site. In folded proteins, each structural site tolerates only a small subset of biochemically compatible residues (typically 2 to 5 amino acids). Applying Bayesian CAT-GTR (Lartillot & Philippe) or maximum-likelihood LG+C60 (which models site-specific equilibrium frequency profiles as a mixture of 60 empirical biochemical classes) accounts for site-specific saturation and breaks the artificial Pelagibacterales + Rickettsiales ("HoloRickettsiales") grouping, placing Pelagibacterales instead within the marine subclass Caulobacteridae alongside Rhodobacterales and Caulobacterales.
2. Dayhoff-6 and SR4 Amino-Acid Recoding: By recoding the 20-letter amino-acid alphabet into 6 physicochemical Dayhoff categories—(AGPST), (DENQ), (HKR), (ILMV), (FWY), and (C)—or the 4-state SR4 alphabet, substitutions between GC-rich and AT-rich codons within the same biochemical group (such as Arginine CGN ↔ Lysine AAR in the basic HKR class, or Valine GUN ↔ Isoleucine AUH in the hydrophobic ILMV class) are masked out, eliminating GARP ↔ FYMINK compositional attraction while retaining deep trans-group historical signal.
3. Stationary Site Trimming (χ² Compositional Filtering) and Metagenomic Taxon Sampling: Removing the top 20–40% most compositionally heterogeneous alignment sites and expanding taxon sampling across deep oceanic metagenome-assembled genomes (MAGs) revealed that:

• Pelagibacterales is a derived, secondarily streamlined marine lineage nested high within core Caulobacteridae, not a basal alphaproteobacterium or mitochondrial sister group.
• Rickettsiales occupies a deep branch within Alphaproteobacteria, where intracellularity evolved independently of mitochondria (further supported by the non-homologous bacterial Tlc ATP importer in Rickettsia versus the eukaryotic SLC25 ATP exporter in mitochondria).
• The Proto-Mitochondrial Lineage: In comprehensive multi-gene analyses incorporating marine metagenomes (Martijn et al., 2018; Muñoz-Gómez et al., 2022), mitochondria branch either as a deep sister lineage to nearly all extant Alphaproteobacteria ("Protomitochondria", near the basal magnetotactic order Magnetococcales) or as the sister clade to the marine order Iodidimonadales at the base of the Alphaproteobacteria radiation.

1. Giovannoni, S. J., et al. "Genome streamlining in a cosmopolitan oceanic bacterium." Science 309, 1242–1245 (2005). https://doi.org/10.1126/science.1114057
2. Andersson, S. G. E., et al. "The genome sequence of Rickettsia prowazekii and the origin of mitochondria." Nature 396, 133–140 (1998). https://doi.org/10.1038/24094
3. Martijn, J., Vosseberg, J., Guy, L., Offre, P., & Ettema, T. J. G. "Deep mitochondrial origin outside the sampled alphaproteobacteria." Nature 557, 101–105 (2018). https://doi.org/10.1038/s41586-018-0059-5
4. Muñoz-Gómez, S. A., et al. "Site-and-branch-heterogeneous analyses of an expanded dataset favour mitochondria as sister to known Alphaproteobacteria." Nature Ecology & Evolution 6, 253–262 (2022). https://doi.org/10.1038/s41559-021-01638-2