Protein signatures (molecular synapomorphies) that are distinctive characteristics of the major cyanobacterial clades

Phylogenomic/phylogenetic analyses on cyanobacteria

Genome sequences for a large number of cyanobacteria have become available in recent years (Table 1⇑). Phylogenetic analysis on these cyanobacteria was carried out based on a concatenated sequence alignment for 45 widely distributed proteins (see Methods). A rooted NJ tree based on this dataset is shown in Fig. 1⇓ and an ML tree based on these sequences is provided as Supplementary Fig. S1, available in IJSEM Online. The branching patterns of cyanobacteria in these trees are very similar and these results are comparable to those observed in other recent studies based on different large datasets of protein sequences (Sánchez-Baracaldo et al., 2005; Shi & Falkowski, 2008; Swingley et al., 2008a). In both NJ and ML trees, a clade consisting of Gloeobacter violaceus and Synechococcus spp. (JA-3-3Ab and JA2-3-B′a) (referred to here as clade A) showed the deepest branching within the cyanobacteria. The deep branching of clade A species has also been observed in a number of earlier studies (Giovannoni et al., 1988; Honda et al., 1999; Wilmotte & Herdman, 2001; Seo & Yokota, 2003; Sánchez-Baracaldo et al., 2005; Shi & Falkowski, 2008; Swingley et al., 2008a). According to a recent proposal for the classification of cyanobacteria (Hoffmann et al., 2005), the Gloeobacterales are placed into a separate subclass (Gloeobacterophycidae). Most other cyanobacteria in our dataset can be grouped into two major clades. One of these clades (designated clade B) comprises diverse cyanobacteria such as Thermosynechococcus, Synechocystis, Crocosphaera, Acaryochloris, Trichodesmium, Nostoc and Anabaena spp., and it includes both the subclasses Oscillatariophycidae and Nostocophycidae of cyanobacteria (Hoffmann et al., 2005). In contrast, the other main clade (clade C) is composed entirely of different strains/isolates of Prochlorococcus and Synechococcus. This latter clade corresponds to the subclass Synechococcophycidae (Hoffmann et al., 2005), except that our analysis indicates that Acaryochloris is not part of this clade or subclass. Based on their genetic distances (Figs 1⇓ and 2⇓), the clade C isolates are genetically not as diverse as the taxa in clade B. Within clade C, two subclades can be distinguished. One of these subclades consisted entirely of the Prochlorococcus strains/ecotypes, whereas the other subclade mainly comprised various marine Synechococcus strains, except for the branching of Prochlorococcus marinus MIT9303 and MIT9313 within them. In clade B, a number of subclades including those corresponding to Nostoc/Anabaena (order Nostocales) and Synechocystis/Crocosphaera (order Chroococcales) were resolved with high statistical support (i.e. supported by >90 % puzzled quartets or bootstrap samples). The branching position of Synechococcus elongatus was uncertain in these trees; in the ML tree (Supplementary Fig. S1), it branched in between clades B and C, whereas in the NJ tree (Fig. 1⇓), it branched with clade B, although this association lacked statistical support.

An NJ tree for these cyanobacteria was also constructed based on 16S rRNA gene sequences (Fig. 2⇑). This tree consisted of only two main clades, one consisting of the clade A and B cyanobacteria as well as Synechococcus elongatus, whilst the clade C species/strains/isolates comprised the other clade. In contrast to the tree based on the concatenated protein sequences (Fig. 1⇑), the deep branches in the 16S rRNA gene-based tree, including that indicating the branching of Thermosynechococcus elongatus with clade A, were not supported statistically. These trees provide us with a framework for understanding the evolutionary significances of various conserved indels identified in this work.

Usefulness of signature indels (synapomorphies) for taxonomic/evolutionary studies

The shared derived characters that are unique to particular groups or clades of organisms (i.e. synapomorphies) provide an important means of identifying various monophyletic clades and also for understanding how these clades are related to each other. In the past, this approach has largely been employed by using morphology and other observable traits (Sneath & Sokal, 1973; Sneath, 2001). However, such traits are often either plesiomorphic (i.e. a particular character is not limited to a given group) or exhibit homoplasy (the derived character state has evolved independently in the given group of organisms), limiting their utility as phylogenetic or taxonomic markers. In recent years, the availability of genome sequences has led to the discovery of important molecular characteristics that are uniquely shared by different groups of organisms and provide important means for their grouping and for evolutionary studies. The conserved indels of defined lengths that are present in gene/protein sequences at specific positions, and which are uniquely shared by particular groups of organisms, form an important class of such characters (synapomorphies) and have proven very useful in clarifying many important evolutionary relationships (Rivera & Lake, 1992; Baldauf & Palmer, 1993; Gupta, 1998; Rokas & Holland, 2000). When a conserved indel of defined length and sequence is found specifically in a particular group of species, then the simplest and most parsimonious explanation for this observation is that these species shared a common ancestor, in which the rare genetic change responsible for this indel first occurred and was then passed on to various descendants (Rivera & Lake, 1992; Gupta, 1998; Rokas & Holland, 2000). However, such changes can also occur due to lateral gene transfers between or among species (Boucher et al., 2003; Zhaxybayeva et al., 2006); hence, it is important to exclude this possibility by comparing the inference obtained by the signature approach with that derived from other independent means, such as phylogenetic or phylogenomic methods. I describe below many conserved indels or molecular synapomorphies that are distinctive characteristics of various clades of cyanobacteria.

Conserved indels that are specific for cyanobacteria and their deepest-branching clades

Previously, we described 14 conserved indels in 10 proteins that are specific for all cyanobacterial homologues (Gupta et al., 2003). Since that time, sequence information for many cyanobacteria and a large number of other bacteria has become available. Hence, it was of importance to check the species specificities of these signatures. Of these 14 indels, 13 (including the 6, 7 and 28 aa inserts in the DNA helicase II protein, 14 aa insert in ADP-glucose pyrophosphorylase, 3 aa insert in FtsH protein, 11–13 aa insert in phytoene synthase, 5 aa insert in EF-Tu, 2 and 7 aa deletions in the ribosomal S1 protein, 2 aa insert in SecA protein, 6 aa insert and 1 aa deletion in inosine-5′-monophosphate dehydrogenase and 1 aa insert in the major sigma factor) were present in all sequenced (available) cyanobacteria, but not found in any other bacteria (results not shown). In this work, I have identified two additional conserved indels, viz. a 3 aa deletion in carbamoylphosphate synthase (Supplementary Fig. S2) and a 3 aa insert in prolyl-tRNA synthetase (Supplementary Fig. S3), that are also specific for cyanobacteria and are not found in any other bacteria. In prolyl-tRNA synthetase, a 2 aa insert is also found in various chlamydial species in the same position where the 3 aa insert is present in cyanobacteria. Earlier studies have indicated the presence of chlamydia-like genes in green plants and algae, but not in cyanobacteria (Horn et al., 2004; Huang & Gogarten, 2007; Moustafa et al., 2008). In a phylogenetic tree for ProRS sequences, the chlamydiae and cyanobacteria do not group together (results not shown), indicating that the inserts in these two groups, which are of different lengths, have originated independently. Because of their cyanobacterial specificities, the genetic changes responsible for these molecular signatures probably occurred in a common ancestor of cyanobacteria, as indicated in Fig. 3⇓.

In contrast to the above indels, the 18 aa insert in DNA polymerase I (Pol I), identified previously (Gupta et al., 2003), was present in all other cyanobacteria except those corresponding to clade A (Fig. 4⇓). This indel was also absent in various other bacteria, indicating that this synapomorphy is an insert in other cyanobacteria that was introduced after the divergence of clade A (see Fig. 3⇑). In addition to the Pol I signature, we have identified several other conserved indels that support a similar relationship. These include a 4–5 aa insert in the tryptophan synthase beta chain (Supplementary Fig. S4), a 4 aa insert in tryptophanyl-tRNA synthetase (TrpRS) (Supplementary Fig. S5) and a 2 aa insert in the DNA polymerase III gamma/tau subunits (encoded by the dnaX gene) (Supplementary Fig. S6). The conserved inserts in these proteins are commonly shared by all other cyanobacteria, but they are absent in the clade A species/strains as well as other bacteria. These signatures/synapomorphies support the view that the clade A species/strains constitute the deepest-branching lineage within cyanobacteria (Fig. 3⇑), which is in accordance with the phylogenetic tree based upon our protein dataset (Fig. 1⇑; Supplementary Fig. S1) and also with earlier phylogenomic studies based on other datasets of proteins (Honda et al., 1999; Turner et al., 1999; Sánchez-Baracaldo et al., 2005; Shi & Falkowski, 2008; Swingley et al., 2008a). Interestingly, the insert in the tryptophan synthase beta chain (Supplementary Fig. S4), which is absent in the clade A cyanobacteria but found in all other cyanobacteria, is also found in various plant and algal homologues. The shared presence of this synapomorphy in the plant/algal homologues and various cyanobacteria, except those from clade A, provides evidence that the plant/algal homologues of this protein are derived from a cyanobacterium not belonging to clade A.

We have also identified a 15 aa insert in a highly conserved region of the protein synthesis elongation factor-G (EF-G) that is found only in the clade A cyanobacteria (Fig. 5⇓). This synapomorphy provides a molecular marker for the clade A species/strains (subclass Gloeobacterophycidae) and, in conjunction with other signatures described above (Fig. 3⇑), provides evidence that this group is distinct from all other cyanobacteria. A number of conserved inserts in other proteins are specifically present in the two Synechococcus strains JA-3-3Ab and JA2-3-B′a that are part of clade A. These include a 12 aa insert in the protein serine hydroxymethyltransferase (GlyA), an 8 aa insert in the RNA polymerase beta subunit (RpoB), a 6 aa insert in the ribosomal L2 protein and a 2 aa insert in the prolyl-tRNA synthetase (Supplementary Figs S7–S10).

Signature indels for the clade C cyanobacteria

As indicated earlier, clade C is composed entirely of different strains/isolates and ecotypes of Prochlorococcus and Synechococcus. Of these, Synechococcus spp. are ubiquitous in different aquatic environments, including estuarine, coastal and offshore waters (Palenik et al., 2006). In contrast, Prochlorococcus spp. are found mainly in warm, oligotrophic oceanic settings (Rocap et al., 2003). We have identified a number of conserved indels in widely distributed proteins that are specific for the clade C species/strains (subclass Synechococcophycidae) and also for one of its subgroups. The signatures that are specific for clade C include a 3 aa insert in the RNA polymerase β subunit (RpoB; Fig. 6⇓), a 2 aa insert in the protein KsgA (Supplementary Fig. S11) that carries out essential dimethylation of the 16S rRNA, a 6 aa insert in the tyrosyl-tRNA synthetase (Supplementary Fig. S12), a 2 aa insert in the tRNA (guanine-N1-)-methyltransferase (Supplementary Fig. S13) and a 1 aa insert in the RNA polymerase β′ subunit (RpoC; Supplementary Fig. S14). The inserts in all of these proteins are present in all of the clade C species/strains (sequences for some closely related strains of Prochlorococcus and Synechococcus strains are not shown in these alignments), but they are not found in other cyanobacteria, including Acaryochloris marina, which has also been suggested to belong to the subclass Synechococcophycidae (Hoffmann et al., 2005). Another prominent signature for clade C (a 12 aa insert) is found in Pol I (Supplementary Fig. S15), which, as indicated previously, also contains a large insert that is specific for various cyanobacteria except those from clade A (Fig. 4⇑). In contrast to the other clade C-specific signatures, the insert in Pol I is lacking in Synechococcus WH5701, which lies in the middle of the clade C species/strains. To account for the absence of the Pol I insert in Synechococcus WH5701, one must postulate that either the Pol I gene in this cyanobacterium was acquired laterally from some other bacterium lacking the insert, or that the insert in this case has somehow been deleted/lost. Based upon the available information, we are unable to distinguish between these possibilities. Of the inserts that are specific for the clade C cyanobacteria, the homologues of RpoB are also present in various plastids and green plants, and this insert is not found in any of them (Fig. 6⇓). Likewise, the plant homologues of the protein tRNA (guanine-N1-)-methyltransferase also lack the insert in this protein (Supplementary Fig. S13). These observations suggest that the plastid/plant homologues of these proteins have not originated from clade C cyanobacteria.

Within clade C, among the Prochlorococcus isolates, two phylogenetically and physiologically different subgroups or ecotypes (high-B/A and low-B/A) have been identified (Moore et al., 1998; Rocap et al., 2002). The strains from these two ecotypes differ in terms of the relative ratios of chlorophyll b and a2 in their light-harvesting systems. Strains from these two subgroups also differ in their ability to grow at different light intensities and copper toxicities and to use nitrite or nitrate as nitrogen sources (Ferris & Palenik, 1998; Rocap et al., 2002). In the phylogenetic trees shown in Figs 1⇑ and 2⇑, the low-B/A Prochlorococcus isolates formed a distinct subclade within clade C that was separated from all other clade C species/strains by a long branch length and a 100 % bootstrap score. We have identified two conserved inserts that are specific for the low-B/A subgroup of Prochlorococcus marinus isolates. In the protein leucyl-tRNA synthetase, which is essential for protein synthesis, a 5 aa deletion is present that is specific for the various low-B/A subgroup Prochlorococcus strains (MIT9515, CCMP1986, MIT9312, MIT9215, MIT9301 and AS9601), but not found in any other cyanobacteria (Supplementary Fig. S16). Another 1 aa insert in a conserved region of the signal-recognition particle protein Ffh or SRP54 is also specific for the low-B/A subgroup of Prochlorococcus strains (Supplementary Fig. S17). Additionally, in the TrpRS signature shown in Supplementary Fig. S5, all of the low-B/A subgroup Prochlorococcus marinus strains contained a 3 aa insert, whilst a 4 aa insert was found in other cyanobacteria. These signature indels provide evidence that the low-B/A subgroup of Prochlorococcus marinus strains is phylogenetically and molecularly distinct from all other clade C species/strains. The evolutionary stages where genetic changes responsible for these signature sequences probably occurred are indicated in Fig. 7⇓.

Conserved indels that are specific for the clade B cyanobacteria

A number of conserved indels have been identified that are specific for the clade B cyanobacteria, which encompass the majority of known cyanobacteria (Honda et al., 1999; Turner et al., 1999; Castenholz, 2001; Wilmotte & Herdman, 2001; Sánchez-Baracaldo et al., 2005; Swingley et al., 2008a). Within clade B, heterocyst-forming cyanobacteria form a monophyletic group (subclass Nostocophycidae) (Turner et al., 1999; Adams, 2000; Wilmotte & Herdman, 2001; Hoffmann, 2005; Rajaniemi et al., 2005). These heterocystous cyanobacteria include both Nostocales (non-branching and dividing always in a plane at a right angle to the long axis of the trichome) and Stigonematales (showing true branching) (Rippka et al., 1979; Wilmotte & Herdman, 2001; Rajaniemi et al., 2005). We have identified two conserved indels that are specific for these cyanobacteria. In the PetA protein, which is a precursor of the apocytochrome f, a 4 aa insert in a conserved region is present in various Nostocales species/strains (Nostoc, Anabaena and Nodularia) as well as in Mastigocladus laminosus, which belongs to the order Stigonematales (Fig. 8⇓). The unique shared presence of this insert in these cyanobacteria provides further evidence that they form a monophyletic group. Another conserved insert (5 aa) specific for the Nostocales is found in ribosomal protein S3 (Supplementary Fig. S18). The sequence information for this protein is not available for Stigonematales. In ribosomal protein S3, in the same position where the 5 aa insert is found in the Nostocales, an 8 aa insert is present in many green plants, but not in any of the plastid homologues (Supplementary Fig. S19). The different lengths and sequences of these inserts indicate that they are of independent origin.

Cyanobacteria such as Synechocystis, Microcystis, Crocosphaera and Cyanothece, etc., which belong to the order Chroococcales, form another well-defined clade in phylogenetic trees (Figs 1⇑ and 2⇑) (Honda et al., 1999; Turner et al., 1999; Sánchez-Baracaldo et al., 2005; Shi & Falkowski, 2008; Swingley et al., 2008a). This clade has been referred to as the SPM clade in earlier work (Turner et al., 1999; Sánchez-Baracaldo et al., 2005) and it forms a part of the subclass Oscillatariophycidae in the proposal by Hoffmann et al. (2005). We have identified a 1 aa insert in a highly conserved region of the RecA protein that is commonly shared by all of these cyanobacteria (Fig. 9⇓). This insert is also present in Synechococcus PCC7002, which branches with this clade in the phylogenetic trees (Fig. 2⇑) (Turner et al., 1999; Sánchez-Baracaldo et al., 2005). In phylogenetic trees based on different datasets, Trichodesmium erythraeum generally branches in between the Nostocales and the Chroococcales, often as an outgroup of the Nostocales clade (Figs 1⇑ and 2⇑) (Honda et al., 1999; Turner et al., 1999; Sánchez-Baracaldo et al., 2005; Shi & Falkowski, 2008; Swingley et al., 2008a). We have identified a 4 aa deletion in translation-initiation factor IF-2 that, within cyanobacteria, is uniquely present in various Nostocales, as well as in Trichodesmium erythraeum and Lyngbya PCC8106 (Supplementary Fig. S19), both of which belong to the order Oscillatoriales. The shared presence of this signature in the Nostocales and Oscillatoriales provides further evidence that these two orders of cyanobacteria are close relatives, as also seen in phylogenetic trees (Figs 1⇑ and 2⇑). Interestingly, another conserved signature consisting of a 7 aa insert in the α subunit of riboflavin synthase is uniquely shared by various Chroococcales and Oscillatoriales species/strains (Supplementary Fig. S20). A close relationship between these two cyanobacterial orders, which are grouped together in the subclass Oscillatariophycidae (Hoffmann et al., 2005), has also been reported in earlier work (Turner et al., 1999). These molecular signatures support an intermediate placement of the Oscillatoriales in between the orders Nostocales and Chroococcales, as indicated in Fig. 7⇑.

Three other large signature indels provide evidence that cyanobacteria from the orders Nostocales, Oscillatoriales and Chroococcales shared a common ancestor, exclusive of other cyanobacteria for which sequence information is presently available (Fig. 7⇑). In the DnaE protein, a 19 aa insert is present specifically in all available sequences from these orders, but not in any other cyanobacteria or bacteria (Fig. 10⇑). Likewise, in the protein GDP-mannose pyrophosphorylase, a 13 aa deletion is specifically present in these cyanobacteria, to the exclusion of all others (Supplementary Fig. S21). Another prominent indel of about 22–27 aa in the protein NAD(P)H-quinone oxidoreductase subunit D (Supplementary Fig. S22) is also specific for this clade of cyanobacteria. Most cyanobacteria contain at least two homologues of this protein (Melo et al., 2004) and this insert is present in only one of the two homologues from these groups of cyanobacteria, but not in other cyanobacteria. The evolutionary stages where genetic changes responsible for these conserved indels have probably occurred are indicated in Fig. 7⇑.