Genome fluctuations in cyanobacteria reflect evolutionary, developmental and adaptive traits

Abstract

Background: Cyanobacteria belong to an ancient group of photosynthetic prokaryotes with pronounced variations in their cellular differentiation strategies, physiological capacities and choice of habitat. Sequencing efforts have shown that genomes within this phylum are equally diverse in terms of size and protein-coding capacity. To increase our understanding of genomic changes in the lineage, the genomes of 58 contemporary cyanobacteria were analysed for shared and unique orthologs.

Results: A total of 404 protein families, present in all cyanobacterial genomes, were identified. Two of these are unique to the phylum, corresponding to an AbrB family transcriptional regulator and a gene that escapes functional annotation although its genomic neighbourhood is conserved among the organisms examined. The evolution of cyanobacterial genome sizes involves a mix of gains and losses in the clade encompassing complex cyanobacteria, while a single event of reduction is evident in a clade dominated by unicellular cyanobacteria. Genome sizes and gene family copy numbers evolve at a higher rate in the former clade, and multi-copy genes were predominant in large genomes. Orthologs unique to cyanobacteria exhibiting specific characteristics, such as filament formation, heterocyst differentiation, diazotrophy and symbiotic competence, were also identified. An ancestral character reconstruction suggests that the most recent common ancestor of cyanobacteria had a genome size of approx. 4.5 Mbp and 1678 to 3291 protein-coding genes, 4%-6% of which are unique to cyanobacteria today.

Conclusions: The different rates of genome-size evolution and multi-copy gene abundance suggest two routes of genome development in the history of cyanobacteria. The expansion strategy is driven by gene-family enlargment and generates a broad adaptive potential; while the genome streamlining strategy imposes adaptations to highly specific niches, also reflected in their different functional capacities. A few genomes display extreme proliferation of non-coding nucleotides which is likely to be the result of initial expansion of genomes/gene copy number to gain adaptive potential, followed by a shift to a life-style in a highly specific niche (e.g. symbiosis). This transition results in redundancy of genes and gene families, leading to an increase in junk DNA and eventually to gene loss. A few orthologs can be correlated with specific phenotypes in cyanobacteria, such as filament formation and symbiotic competence; these constitute exciting exploratory targets.

Figure 1

Phylogeny of 58 cyanobacteria based on a concatenated alignment of core orthologs. Maximum likelihood phylogenetic tree based on a concatenated alignment of 285 single-copy orthologs present in all genomes. The cyanobacteria with fully developed heterocysts with a regular intercalary pattern distribution are indicated by the dark green shaded box while those with terminal or undeveloped heterocysts are indicated by the light green shaded box. Specific phenotypes are shown by the coloured boxes next to the organism names. Numbers at nodes indicate bootstrap values (when < 100). Bar, 0.4 expected substitutions per site.

Figure 2

Amount and functional distribution of paralogs in cyanobacterial genomes. To the left a plot of proportion of protein groups containing paralogs (x-axis) and total number of paralogous gene copies (colour of circles). Numbers in the right margin of the plot show the average number of gene copies for paralogs. Organisms are ordered according to their position in the phylogenetic tree. To the right a heat map of the functional distribution of paralogs classified according to the Cluster of orthologous groups (COG) categories. Abbreviations for functional categories are as follows: t, a subset of L containing 26 transposase categories; O, posttranslational modification, protein turnover and chaperones; S, function unknown; L, replication, recombination and repair; P, inorganic ion transport and metabolism; T, signal transduction mechanisms; M, cell wall/membrane/envelope biogenesis; R, general function prediction only; J, translation, ribosomal structure and biogenesis; V, defence mechanisms; U, intracellular trafficking, secretion and vesicular transport; G, carbohydrate transport and metabolism; K, transcription; C, energy production and conversion; Q, secondary metabolites biosynthesis, transport and catabolism; E, amino acid transport and metabolism; N, cell motility; H, coenzyme transport and metabolism; F, nucleotide transport and metabolism; D, cell cycle control, cell division and chromosome partitioning; I, lipid transport and metabolism. Categories are ordered by maximum percentage.

Figure 3

Genome sizes among 58 cyanobacteria. The bars denote the genome size in Mbp and the green, grey and red portions of the bars represent nucleotides in coding, non-coding and pseudogene regions, respectively.

Figure 4

Parsimony reconstruction of ancestral genome sizes. Genome sizes are indicated (in Mbp) at specific nodes and in the right margin for the extant genomes. A single event of genome reduction appears to have occurred in the ancestor of the unicellular and mostly marine cyanobacteria (Prochlorococcus/Synechococcus). In contrast, the history of Clade 1 cyanobacteria involves a mix of genome expansions and reductions. The most notable expansion events are evident in the genome of Acaryochloris marina (Acam), Trichodesmium erythraeum (Trie), Nostoc punctiforme PCC 73102 (Nosp) and Microcoleus chthonoplastes (Micc). Reductions in this clade include Cylindrospermopsis raciborskii (Cylr), Raphidiopsis brookii (Rapb), Synechococcus sp. PCC 7002 (Syn7002), Synechocystis sp. PCC 6803 (Scys6803) and cyanobacterium UCYN-A (Ucyn). Note that the genome size of the obligate symbiont 'Nostoc azollae' 0708 (NoAz) has not changed considerably since the ancestor shared with Cylr and Rapb although the pseudogenization in NoAz is considerable (Figure 3).

Figure 5

Relationships between genome sizes and paralog number. A and B) Diagnostic plots of absolute values of standardized independent contrasts in genome size (A) and number of paralogs (B) versus their standard deviations (square roots of sums of branch lengths) for Clade 1 (closed circles) and Clade 2 (open circles) cyanobacteria. The plots shows that contrasts within Clade 1 are, on average, larger in magnitude (Wilcoxon tests, two-tailed p < 0.0001). This indicates that genome sizes and gene-copy numbers have evolved at a higher rate in Clade 1 species. In A and B, a plus sign denotes the contrast between Clade 1 and Clade 2, which was excluded from the tests. C) Phylogenetically correct regression analysis through the origin of contrasts in number of paralogous gene copies vs. positivized contrasts in genome size for Clade 1. The plots show a strong positive correlation between genome size and number of duplicated genes (r = 0.691, two-tailed p < 0.0001).

Figure 6

Functional distribution of genes in the cyanobacterial genomes. The distribution of functions for the most recent common ancestor (MRCA) of cyanobacteria is shown at the top. Clade 1 and Clade 2 cyanobacteria are shaded in green and blue in the phylogeny, respectively. At the bottom of the plot is the distribution for chloroplasts (averaged over 189 genomes with a total of 17,216 genes). Note that the chloroplast plot has its own colour key (bottom). COG categories are ordered according to maximum contribution in a single genome. Some Clade 1 cyanobacteria appear to have a higher distribution of genes related to Replication (L), Signal transduction mechanisms (T) and Cell motility (N), while functions related to Amino acid metabolism (E) and Posttranslational modification and chaperones (O) are more abundant in Clade 2. Pseudogenes have been excluded from this analysis. COG abbreviations are as in Figure 2.

Figure 7

Parsimony reconstruction of ancestral states of the filamentous and nitrogen-fixing phenotypes. Equivocal inference are indicated with bi-coloured branches. Genome sizes (in Mbp) is indicated at key nodes of loss/gain. Both characters show a homoplastic pattern with a possible single common origin of the trait followed by one (A) or many (B) losses.