Tuning fresh: radiation through rewiring of central metabolism in streamlined bacteria

Abstract

Most free-living planktonic cells are streamlined and in spite of their limitations in functional flexibility, their vast populations have radiated into a wide range of aquatic habitats. Here we compared the metabolic potential of subgroups in the Alphaproteobacteria lineage SAR11 adapted to marine and freshwater habitats. Our results suggest that the successful leap from marine to freshwaters in SAR11 was accompanied by a loss of several carbon degradation pathways and a rewiring of the central metabolism. Examples for these are C1 and methylated compounds degradation pathways, the Entner-Doudouroff pathway, the glyoxylate shunt and anapleuretic carbon fixation being absent from the freshwater genomes. Evolutionary reconstructions further suggest that the metabolic modules making up these important freshwater metabolic traits were already present in the gene pool of ancestral marine SAR11 populations. The loss of the glyoxylate shunt had already occurred in the common ancestor of the freshwater subgroup and its closest marine relatives, suggesting that the adaptation to freshwater was a gradual process. Furthermore, our results indicate rapid evolution of TRAP transporters in the freshwater clade involved in the uptake of low molecular weight carboxylic acids. We propose that such gradual tuning of metabolic pathways and transporters toward locally available organic substrates is linked to the formation of subgroups within the SAR11 clade and that this process was critical for the freshwater clade to find and fix an adaptive phenotype.

Figure 1

Evolutionary relationships between marine and freshwater SAR11 genomes. Unrooted maximum likelihood phylogenetic tree of concatenated ribosomal protein sequences from single-cell genomes and isolate genomes of the SAR11 clade. In addition to bootstrap values as inferred by maximum likelihood, the second (large and bold) number at the branching points show proportion of protein trees with corresponding branching pattern. These proportions were generated by using the Robinson–Foulds metric summarizing 518 orthologous protein trees that had at least one representative in each of the seven subclades (for details, see Supplementary Table S1). Freshwater SAR11 (subgroup IIIb/LD12) are indicated in red, whereas marine SAR11 subgroups in other colors.

Figure 2

Nine protein trees with evolutionary inconsistency when compared with the ribosomal tree. (a–i) Unrooted phylogenetic trees of orthologous protein clusters with alternative tree topologies (that is, sister clade to IIIb/LD12 is not IIIa) when compared with the ribosomal tree (Figure 1).

Figure 3

Central carbon metabolism and other relevant metabolic pathways identified in SAR11 genomes. Glycogenesis, TCA cycle and glyoxylate shunt are shown in the center of the plot with adjunction pathways to either side such as the pentose phosphate and ED pathway. The color of the arrows indicates genes encoded in at least one genome of each SAR11 subgroup (see legend for details). The presence and absence of genes was determined within the IMG system based on automated and manual annotations. Detailed results are in Supplementary Tables S6–S9.

Figure 4

Closest blast hits against refseq and rarefaction curves of orthologous SAR11 protein clusters. (a) Bar charts revealing top hits from BLASTP against all available genomes in refseq. Hits are classified based on their taxonomic affiliation with higher taxonomic levels removed from lower taxonomic levels (that is, SAR11 hits are removed from the alphaproteobacterial bin, with alphaproteobacterial hits removed from the proteobacterial bin and so on). Black bars indicate hits of genes specific to freshwaters as inferred by orthologous clustering. (b) Detailed representation of freshwater-specific orthologous protein clusters and their closest hits classified on their taxonomic affiliation with higher taxonomic levels removed from lower taxonomic levels. (c) Results from a rarefaction analysis of orthologous protein clusters after normalization to account for the partial single-cell genomes.