Idiosyncratic genome evolution of the thermophilic cyanobacterium Synechococcus at the limits of phototrophy

Abstract

Thermophilic microorganisms are expected to have smaller cells and genomes compared with mesophiles, a higher proportion of horizontally acquired genes, and distinct nucleotide and amino acid composition signatures. Here, we took an integrative approach to investigate these apparent correlates of thermophily for Synechococcus A/B cyanobacteria, which include the most heat-tolerant phototrophs on the planet. Phylogenomics confirmed a unique origin of different thermotolerance ecotypes, with low levels of continued gene flow between ecologically divergent but overlapping populations, which has shaped the distribution of phenotypic traits along these geothermal gradients. More thermotolerant strains do have smaller genomes, but genome reduction is associated with a decrease in community richness and metabolic diversity, rather than with cell size. Horizontal gene transfer played only a limited role during Synechococcus evolution, but, the most thermotolerant strains have acquired a Thermus tRNA modification enzyme that may stabilize translation at high temperatures. Although nucleotide base composition was not associated with thermotolerance, we found a general replacement of aspartate with glutamate, as well as a dramatic remodeling of amino acid composition at the highest temperatures that substantially differed from previous predictions. We conclude that Synechococcus A/B genome diversification largely does not conform to the standard view of temperature adaptation. In addition, carbon fixation was more thermolabile than photosynthetic oxygen evolution for the most thermotolerant strains compared with less tolerant lineages. This suggests that increased flow of reducing power generated during the light reactions to an electron sink(s) beyond carbon dioxide has emerged during temperature adaptation of these bacteria.

Keywords: adaptation; cyanobacteria; genomes; horizontal gene transfer; hot springs; thermophile.

Figure 1

Genome-wide maximum likelihood SynAB phylogeny reconstructed from a concatenated alignment of 118 301 amino acid sites derived from 404 single-copy orthologous gene sequences. The tree was outgroup-rooted with Synechococcus strain Nb3U1. Bootstrap values of 100% are indicated by closed circles, and values greater than 95% are indicated by open circles. Numbers at nodes indicate gCF and internode certainty values, respectively. Tree scale is in units of amino acid substitutions per site.

Figure 2

(A) Temperature dependence of growth rate for representative SynAB strains from each clade. Error bars are standard errors for triplicate cultures. (B) Temperature dependence of maximal carbon fixation rate P_m. Error bars are standard deviations. (C) Temperature dependence of the rate constant α (the initial slope of carbon fixation increase with increase in irradiance). Error bars are standard deviations. (D) Temperature dependence of maximal oxygen evolution rate P_m. Error bars are standard deviations. For both carbon fixation and oxygen evolution experiments, cells had been grown at optimal temperature prior to assay. Color-coding for B-D as for panel A.

Figure 3

(A) SynAB genome size versus sample collection temperature (N = 47 genomes with BUSCO coverage >90%). Mean genome size for clade V/VI strains was 80% of that of clade I (mean ± SE: 2.4 ± 0.03 versus 3.0 ± 0.03 Mb). Number of KEGG pathway orthologs (KOs) in each genome versus temperature for (B) global metabolic pathways, (C) secondary metabolite synthesis, and (D) ABC transporters. Dashed lines are least square regression lines.

Figure 4

Lotus plot indicating the SynAB core genome (white), unique genes for individual clades, and uniquely shared genes for representative gene pairs.

Figure 5

(A) Maximum likelihood phylogeny of the trmH tRNA methyltransferase gene for Thermus and SynAB clade V/VI strains reconstructed with the TPM3 + F + G4 model selected by ModelFinder and 1000 bootstrap replicates. The tree is outgroup-rooted with sequence data for Meiothermus taiwanensis (GenBank accession CP021130.1). Bootstrap values of 100% are indicated by closed circles, and values greater than 90% are indicated by open circles. Tree scale is in units of nucleotide substitutions per site. (B) GC content of trmH at first, second and third codon positions.

Figure 6

Heat maps of GC content and amino acid compositions for SynAB strains. Percent ranges for heat map gradient: GC (58–61); D (4–4.5); E (6.2–6.7); R (7–7.5); K (2.7–3.2); L (12.3–13); I (4.4–4.9); V (6.8–7.3); M (1.6–1.9); F (3.3–3.6); W (1.6–1.9); Y (2.5–2.8); A (9.7–10.3); S (5.3–5.8); T (4.4–4.9); Q (5.4–5.9); N (2.3–2.6); H (1.8–2.1); C (0.9–1.2); P (6.2–6.5); G (7.6–8.1).