Prior to undertaking studies on identifying proteins that are specific for different cyanobacterial clades, it was necessary to determine the branching pattern of sequenced cyanobacteria in phylogenetic trees. Although detailed phylogenetic studies have been previously reported for a limited numbers of cyanobacteria 41112, sequence information for many other genomes has become available in the past 2-3 years (see Table 1). Hence, it was necessary to carry out phylogenetic studies on all of these cyanobacteria to determine their branching pattern. The phylogenetic trees are now commonly constructed based on concatenated sequences for large number of proteins 4111231. Their main advantage is that because they are based on large numbers of characters derived from many independent proteins, they are generally considered to provide a better reflection of organismal phylogeny than trees based on any single gene or protein, where the observed relationship could be affected by various factors including lateral gene transfer, differences in evolutionary rates among species, long branch attraction effect, etc. 32. However, it should be recognized that the trees based on concatenated sequences, due to the possibility of their lumping together gene sequences with discordant evolutionary histories, can sometime result in unreliable inferences 323334. In the present work, phylogenetic trees were constructed based on a combined sequence alignment for 44 widely distributed proteins (see additional file 1) from 44 cyanobacterial species/isolates for which sequence information was available (see Materials and Methods). Most of these proteins carry out important housekeeping functions, and they are universally present in various species 35, making them a good choice for phylogenetic analysis.

A rooted maximum likelihood (ML) distance tree based on the combined sequences for these proteins is shown in Fig. 1 and a neighbour-joining (NJ) tree for the same dataset is provided as additional file 2. A number of distinct clades of cyanobacteria were observed in both these trees. Very similar branching patterns and the grouping of cyanobacterial species in various clades have been observed in earlier studies based on other large and independent datasets of protein sequences 41112, giving confidence in the observed results. One of the observed clades, referred to here as Clade A, consists of Gloebacter violaceus and Synechococcus sps. (JA-3-3Ab and JA2-3-B'a). The ML and NJ tree differ from each other in the branching position of this clade. In the ML tree, the Clade A species/strains formed the deepest branching lineage within cyanobacteria. In contrast, in the NJ tree, the cyanobacteria were divided into two main clades at the deepest level and the Clade A formed the outermost branch of one of these clades, separated from all other species/strains by a long branch (additional file 2). However, the branching of Clade A in this position is not reliable, as in our recent studies based on the same dataset of protein sequences but with smaller numbers of cyanobacteria, the clade A species/strains branched in the same position as seen here in the ML tree 23. The deep branching of Clade A species/strains has also been observed in a number of earlier studies based on different datasets of protein sequences 4611122336373839. Further strong and independent evidence that the Clade A species/strains constitutes the earliest branching lineage within sequenced cyanobacteria is provided by our recent identification of several conserved indels in broadly distributed proteins (viz. 18 aa insert in DNA polymerase I, 4-5 aa insert in the tryptophan synthase beta chain, 4 aa insert in tryptophanyl-tRNA synthetase and a 2 aa insert in the DNA polymerase III) 23. The indicated conserved inserts in these proteins are commonly shared by all other sequenced cyanobacteria, but they are lacking in Clade A as well as all other phyla of bacteria 23. The species distributions of these conserved indels strongly indicate that these synapomorphies were introduced in a common ancestor of various other cyanobacteria after the branching of Clade A. In a recent proposal for the classification of cyanobacteria, the thylakoids lacking Gloebacterales are placed into a separate subclass (Gloebacterophycidae) 15. It is unclear whether the Synechococcus sps. (JA-3-3Ab and JA2-3-B'a), which group with G. violaceus, also lack thylakoids or not.

Most other cyanobacteria could be grouped into two main clades in these trees. One of these clades (designated here as Clade B) is comprised of diverse cyanobacteria including Thermosynechococcus, Acaryochloris, as well as other cyanobacterial groups such as Chroococcales (Synechocystis/Crocosphaera/Microcystis/Cyanothece), Nostocales (Nostoc/Nodularia/Anabaena) and Oscillatoriales (Trichodesmium/Lynbya)15. Within Clade B, a subclade comprising of the Chroococcales, Nostocales and Oscillatoriales is also observed in both ML and NJ trees (Fig. 1 and additional file 2). The other main clade (clade C) is composed entirely of different strains/isolates of marine unicellular Prochlorococcus and Synechococcus cyanobacteria. This latter clade has been referred to as the Syn/Pro clade 4 and it corresponds to the subclass Synechococcophycidae in the proposal by Hoffman et al. 15. Within clade C, different Prochlorococcus and Synechococcus strains/isolates were not completely separated from each other. In particular, two of the Prochlorococcus strains, MIT 9303 and MIT 9313, branched within the Synechoccous strains/isolates, in both ML and NJ trees (Fig. 1 and additional file 2). Similar polyphyletic branching of these strains has been observed in earlier studies 1223. However, in both these trees, one subclade of Prochlorococcus strains, which is referred to as the low B/A ecotype subgroup 4041, was separated from all others Prochlorococcus strains by a long-branch. The branching position of the freshwater unicellular cyanobacterium Synechococcus elongatus (strains PCC 6301 and PCC 7942), although it appeared as a deep branching lineage of Clade C, was uncertain in these trees (discussed later).

These phylogenetic trees provide a framework for identifying proteins that are specific for either all cyanobacteria or their different well-resolved clades. Based upon earlier studies, within any given group of bacteria or organisms, signature proteins are present at various phylogenetic depths 252728424344. Hence, to identify proteins that are specific for different main clades of cyanobacteria, Blastp searches were carried out on each ORF in the genomes of the following 6 cyanobacteria: Synechococcus sp. WH8102, Synechocystis sp. PCC6803, Nostoc sp. PCC7120, Synechococcus sp. JA-3-3Ab, Prochlorococcus sp. MIT9215 and Pro. marinus subsp. marinus str. CCMP1375. These cyanobacteria are present at the tips of various clades in phylogenetic trees (Fig. 1 and additional file 2). Hence, blast searches with the proteins in them should enable us to identify proteins that are specific for various main clades of cyanobacteria at different phylogenetic depths. The results of these studies are summarized below.

Blast searches on the above genomes have identified 39 proteins that are specific for cyanobacteria and which are present in virtually all of the sequenced genomes (Table 2a). Thirty-three of these proteins are present in all sequenced cyanobacteria (Table 2a) whereas the remaining 6 (marked with *) are missing in 1-2 isolated species/strains. The homologs of some of these proteins are also found in a few algae or plants. Because of their specific presence in practically all cyanobacteria, but generally no other bacteria, these proteins could be regarded as the cyanobacterial signature proteins. The number of cyanobacterial signature proteins identified in the present work is much smaller than those reported in earlier studies 2930. However, this difference is mainly due to the large increase in the number of sequenced cyanobacterial as well as other genomes in the past few years. In earlier work, we have also described 15 conserved indels in broadly distributed proteins that are distinctive characteristics of all available cyanobacteria and which are not found in any other bacterial groups/phyla 2223.

These analyses have also identified 5 proteins whose homologs are present in all other cyanobacteria, except those from Clade A (Table 2b). Based upon solely the genomic distributions of these proteins, it is difficult to interpret whether the genes for these proteins first evolved in a common ancestor of all cyanobacteria followed by their loss in Clade A species/strains, or they originally evolved in a common ancestor of the Clade B and C cyanobacteria after the branching of Clade A. However, based upon the results of phylogenomic analyses, and more importantly the species distribution patterns of several conserved indels in widely distributed proteins that provide evidence that the Clade A is ancestral to other cyanobacteria 23, the most parsimonious explanation for the observed distribution of these genes is that they first evolved in a common ancestor of the Clade B and C cyanobacteria, as indicated in Fig. 2. Table 2c lists 13 other proteins for which high scoring homologs are present in all (or most) cyanobacteria from Clades A and B, but which are lacking in Clade C strains/isolates. Because of the deep branching of Clade A, it is likely that the genes for these proteins also first evolved in a common ancestor of cyanobacteria, followed by their loss in an ancestor of Clade C. The alternate possibility that the Clade A and B cyanobacteria shared a common ancestor exclusive of Clade C is not supported by the species distribution pattern of conserved indels in several proteins, as noted above. Blast searches with proteins in the genome of Synechococcus sp. JA-3-3Ab have also identified 14 proteins that are specific for the Clade A cyanobacteria (additional file 3). The Clade A species/strains can also be distinguished from other cyanobacteria based upon a 15 aa conserved insert in the protein synthesis elongation factor-G that is specific for this clade 23.

The Clade B comprises the majority of known cyanobacteria except the unicellular marine cyanobacteria (Clade C) and some deep branching cyanobacteria (see Fig. 1). This clade as defined in our work includes all of the species/strains from the orders Chroococcales, Nostocales and Oscillatoriales as well as the deeper branching cyanobacteria, A. marina and Thermosyn. elongatus. Of these latter cyanobacteria, Acaryochloris is unique in containing chlorophyll d as its primary photosynthetic pigment 45, whereas Thermosynechococcus is a unicellular thermophilic cyanobacterium 46. Our analyses have identified 38 proteins that are uniquely shared by all or most of the species/strains from this clade. Two of the Synechococcus strains viz. PCC7002 and PCC7335, also consistently appeared in this group and of these Synechococcus PCC7002, for which sequence information was available from various cyanobacteria, branched with the Chroococcales in phylogenetic trees (Fig. 1 and additional file 2).

The branching position of Syn. elongatus (strains PCC 6301 and PCC 7942) is not resolved in phylogenetic trees 411123747. It generally branches in between the Clades B and C species/strains in phylogenetic trees (Fig. 1, additional file 2) 23. Our analyses have identified 22 proteins, which in addition to various Clade B cyanobacteria are also present in Syn. elongatus (Table 3b). It is known from earlier work that a number of cyanobacteria contain split DnaE protein due to the presence of intervening inteins 464849. Examination of DnaE gene/protein from various cyanobacteria indicates that the split DnaE proteins are found in all of the Clades B cyanobacteria as well as Syn. elongatus, whereas all other species/strains from clade A and C do not contain split DnaE 4(Gupta, R. S., results not shown). This rare genetic characteristic together with the various proteins in Table 3b suggests that Syn. elongatus and Clade B cyanobacteria probably shared a common ancestor exclusive of other cyanobacteria.

Within Clade B, the cyanobacterial species/strains belonging to the orders

N

O

C

Within Clade B, the heterocyst-forming cyanobacteria form a monophyletic group (subclass Nostocophycidae) 6104750. We recently described two conserved indels (a 4 aa insert in the PetA protein, a precursor of the apocytochrome f, and a 5 aa insert in the ribosomal protein S3) that are specific for these bacteria 23. In the present work, blast searches on the genome of Nostoc sp. PCC7120 have identified 65 proteins that are uniquely shared by all of the sequenced Nostocales species/strains (Nostoc, Anabaena and Nodularia) (Table 4d and additional file 5). Fifty-eight additional protein listed in the additional file 5 are also specific for this order, but they are missing in 1-2 species/strains. These proteins provide potential molecular signatures for the Nostocales order (Nostocophycidae subclass).

The cyanobacteria such as Synechocystis, Microcystis, Crocosphaera and Cyanothece, belonging to the order Chroococcales, form another well-defined clade in phylogenetic trees (see Fig. 1 and additional file 2) 411123747. A 1 aa insert in a highly conserved region of the RecA protein is also specific for these cyanobacteria 23. This insert is also present in Synechococcus sp. PCC7002, which branches with this clade in the phylogenetic trees (see Fig. 1 and additional file 2) 447. In this work, we have identified 8 proteins that are uniquely present in various sequenced Chroococcales species/strains (Table 4c). The evolutionary stages where the genes for these proteins have likely evolved are indicated in the interpretive diagram (Fig. 2).

The Clade C is comprised of different strains/isolates of marine Prochlorococcus and Synechococcus 4041515253. We have recently described a number of conserved indels in widely distributed proteins that are specific for all of the species/strains from Clade C 23. These signatures include a 3 aa insert in the RNA polymerase beta subunit, a 2 aa insert the proteins KsgA, a 6 aa insert in tyrosyl-tRNA synthetase, a 2 aa insert in the tRNA (guanine-N1-)-methyltransferase, a 1 aa insert in the RNA polymerase β' subunit and a 12 aa insert in the DNA polymerase I 23. These signature indels are not found in the Clades A or B cyanobacteria or other phyla of bacteria. Additionally, they are also absent in Syn. elongatus as well as Synechococcus sps. PCC7002 and PCC7335. Another example of a signature insert that is specific for Clade C species/strains is presented in Fig. 3. In this case, a 6 aa insert in a flavoprotein is commonly present in all Clade C species/strains, but absent from all other cyanobacteria as well as other bacteria. This latter observation indicates that this indels is an insert in the Clade C species/strains. Interestingly, this insert and also several of the other Clade C signature indels are also present in Cyanobium sp. PCC7001 (Fig. 3), supporting its placement within the Clade C (Fig. 2) 415.

Our blast analyses on proteins from the genomes of Synechococcus sp. WH8102, Prochlorococcus sp. MIT9215 and Pro. marinus subsp. marinus str. CCMP1375 have identified 60 proteins that are uniquely shared by virtually all of the species/strains from Clade C cyanobacteria (Table 5a). These signature proteins provide further evidence and molecular markers indicating the distinctness of Clade C. Eight additional proteins in Table 5b are also specific for Clade C cyanobacteria, but they are absent in all of the low B/A ecotype Prochlorococcus strains, indicating that the genes for these proteins were lost from a common ancestor of the low B/A clade.

As noted earlier, in phylogenetic trees, the branching position of Syn. elongatus is not resolved. In our analyses, we have come across only 3 proteins (marked with ⁺ in Table 5a) that are uniquely found in Clade C species/strains as well as Syn. elongatus. This is in contrast to 22 proteins that are uniquely shared by Clade B cyanobacteria and Syn. elongatus (Table 3b). These observations in conjunction with the unique presence of split DnaE genes in Clade B cyanobacteria and Syn. elongatus make a strong case that Syn. elongatus is more closely related to the Clade B cyanobacteria than to the Clade C species/strains.

The two genera, Prochlorococcus and Synechococcus, which make up most of the Clade C cyanobacteria, differ from each other in important respects, particularly with regard to the main pigments in their light harvesting systems 4041. In contrast to various Synechococcus strains/isolates and most other cyanobacteria, which contain chlorophyll a and phycobiliproteins as the major pigments in their photosynthetic systems, all Prochlorococcus strains/isolates utilize divinyl chlorophyll a and both mono and divinyl chlrophyll b as the main pigments in their light-harvesting systems 4041. Further, while Synechococcus isolates are ubiquitous in different aquatic environments including estuarine, coastal and offshore waters 53, Prochlorococcus strains are mainly found in warm oligotrophic oceanic settings 40. Among the sequenced cyanobacteria, Prochlorococcus strains/isolates have the smallest genomes (see Table 1). Although Prochlorococcus are indicated to be polyphyletic in phylogenetic analyses (with strains MIT 9303 and MIT 9313 branching within the Synechococcus strains/isolates; see Fig. 1 and additional file 2) 122333, our blast searches have identified 19 proteins that are uniquely shared by all or most of the Prochlorococcus strains (Table 6b). These results indicate that despite their polyphyletic branching in phylogenetic trees, all Prochlorococcus strains/isolates form a monophyletic clade, which is in accordance with their distinctive photosynthetic pigments composition. In this work, we also describe a 2 aa conserved insert in the protein heme oxygenase that is also exclusively present in various Prochlorococcus strains (Fig. 4). The unique presence of this insert in various Prochlorococcus strains provides further evidence that this group is monophyletic. The enzyme heme oxygenase, which contains this conserved insert, plays an important role in the biosynthesis of photosynthetic pigments phyto-chromobilin and phycobilins 54. Because Prochlorococcus are unique in terms of their photosynthetic pigment composition, it is of much interest to determine the functional significance of this conserved indel.

If Prochlorococcus strains/isolates form a monophyletic lineage, then one expect that other cyanobacteria that are part of Clade C might also share many unique proteins in common. Indeed, our blast searches have identified 14 proteins that are uniquely present in various other cyanobacteria (mostly Synechococcus strains) that are part of Clade C (Table 6a). It should be mentioned that for several of these proteins, blast hits indicating significant similarity are also found for Cyanobium sp. PCC7001 and Paulinellla chromatophora, indicating that these cyanobacteria are also part of the Clade C. The grouping of Cyanobium sp. PCC7001 with Clade C is also supported by the conserved indel in the flavoprotein (see Fig. 3).

As noted above, in phylogenetic trees based on concatenated protein sequences Prochlorococcus str. MIT9303 and MIT9313 branch within the various Synechococcus strains/isolates (Fig. 1 and additional file 2). Earlier phylogenetic studies by Rocap et al. 41 based on the 16S-23S rDNA spacer region indicate that these two strains (high B/A clade IV) form the deepest branching isolates of this genus. Further, in contrast to other sequenced Prochlorococcus strains, whose G+C content range from 30-39%, the strains MIT9303 and MIT9313 have much higher G+C content (~50%) (see Table 1). Our blast analyses, in addition to identifying many proteins that are unique to various Synechococcus strains/isolates, have also identified 22 proteins that are specifically present in all of the Clade C Synechococcus strains as well as in Prochlorococcus MIT9303 and MIT9313 (additional file 6a). At the same time, we have come across 37 proteins that are uniquely found in all other sequenced Prochlorococcus strains, but which are missing in MIT9303 and MIT9313 (additional file 6b). In addition, we have also identified a 1 aa deletion in a conserved region of the protein protochlorophyllide oxidoreductase (POR) that is uniquely shared by all other Prochlorococcus strains except MIT9303 and MIT9313 (Fig. 5). The enzyme POR is responsible for catalyzing light driven reduction of protochlorophyllide to chlorophyllide - a key regulatory reaction in the chlorophyll biosynthetic pathway 55. Hence, it is again of much interest to understand the functional significance of this conserved indel. The rare genetic change leading to this indel likely occurred in a common ancestor of various Prochlorococcus strains after the branching of MIT9303 and MIT9313 (Fig. 2). These observations, in conjunction with the branching pattern of these strains in phylogenetic trees, provide evidence that these two Prochlorococcus strains comprise the deepest branching group (high B/A clade IV) 41 within the Prochlorococcus genus, exhibiting closest relationship to the Synechococcus strains/isolates.

Earlier studies have led to the division of Prochlorococcus strains/isolates into two physiologically distinct groups (high B/A and low B/A ecotypes), based upon the ratios of chlorophyll b and a2 in their light-harvesting systems and their ability to grow at different light intensities 404156. Of these two groups, strains from the high B/A ecotype, which have larger ratio of chlorophyll b/a₂ are able to grow at extremely low irradiance, whereas those from the low- B/A ecotype containing lower ratio of chlorophyll b/a₂ are unable to grow under these conditions. The low- B/A ecotype strains instead are adapted to growth at high light intensities, where the growth of high B/A ecotype strains is inhibited. The strains from these two ecotypes also differ in terms of their sensitivity to copper and their ability to use nitrite or nitrate as nitrogen sources 4157. In phylogenetic trees, the low B/A ecotype Prochlorococcus isolates (viz. MIT9515, CCMP1986, MIT9312, MIT9215, MIT9301 and AS9601) formed a distinct subclade that was well separated from all other Clade C species/strains by a long-branch and 100% bootstrap score (Fig. 1 and additional file 2)2341. We have also described two conserved indels (viz. a 5 aa deletion in leucyl-tRNA synthetase and 1 aa insert in the Ffh protein) that are uniquely shared by all of the low B/A ecotype Prochlorococcus strains 23. In the present work, we have identified 67 proteins that are exclusively found in all of the sequenced strains from the low B/A ecotype clade (additional file 7a). Seventy-two proteins listed in the additional file 7b are also specific for this clade, but they are missing in 1-2 of the strains/isolates. These signature proteins and indels together with the distinct branching of the low B/A strains in phylogenetic trees provide strong evidence that this group of Prochlorococcus strains are phylogenetically, physiologically and molecularly distinct from all other Prochlorococcus strains. Based upon species distribution patterns of various cyanobacteria-specific proteins, evolutionary stages where the genes for these proteins likely evolved are indicated in the interpretive diagram in Fig. 2.

Phylogenetic analyses were carried out on a set of 44 proteins involved in important housekeeping functions that are present in most organisms (see Additional file 1) 35. Blast searches with these proteins revealed that their homologs were present in all 34 sequenced cyanobacterial genomes (listed in Table 1), the two outgroup species (Bacillus subtilis and Staphylococcus aureus), as well as 10 other cyanobacteria (viz. Crocosphaera watsonii WH8501, Cyanothece sp. CCY0110, Lyngbya sp. PCC8106, Microcystis aeruginosa PCC7806, Nodularia spumigena CCY9414, Syenchococcus sp. WH5701, Syenchococcus sp. BL107, Syenchococcus sp. RS9917, Syenchococcus sp. RS9916 and Syenchococcus sp. WH7805). Hence, sequence information for all of these cyanobacteria was included in our analyses. The multiple sequence alignments for these proteins were created using the ClustalX 1.83 program 77 and they were concatenated into a single large file. This unedited sequence alignment was imported into the Gblocks 0.91b program to remove poorly aligned regions 78. This program was used with default settings except that allowed gap position parameter was changed to half. The resulting final alignment of 16834 amino acid sites was used for phylogenetic analyses. A neighbour-joining (NJ) tree based on 1000 bootstrap replicates was constructed by the Kimura model 79 using the TREECON 1.3b program 80. The maximum-likelihood (ML) analysis was carried out using the WAG+F model with gamma distribution of evolutionary rates with four categories using the TREE-PUZZLE program with 10000 puzzling steps 81.

The Blastp searches were carried out on each ORF in the genomes of Synechococcus sp. WH8102, Synechocystis sp. PCC6803, Nostoc sp. PCC7120, Synechococcus sp. JA-3-3Ab, Prochlorococcus sp. MIT9215 and Prochlorococcus marinus subsp. marinus str. CCMP1375 to identify proteins that are uniquely present in various clades of cyanobacteria seen in the phylogenetic trees (Fig. 1). The blast searches were performed against all organisms (i.e. non-redundant (nr) database) using the default parameters, without the low complexity filter 82. The proteins that were of interest were those where either all significant hits were from the indicated groups of cyanobacteria, or which involved a large increase in E values from the last hit belonging to a particular clade to the first hit from any other bacteria/cyanobacteria and the E values for the latter hits were >1e^-04, indicating weak similarity that could occur by chance. Higher E values are often significant for smaller proteins as the magnitude of the E value depends upon the length of the query sequence 82. Hence, the lengths of the query proteins and those of various hits were also taken into consideration when analyzing the results of these studies. In most cases, the lengths of various significant hits were very similar to those of the query proteins. Some proteins, which in addition to cyanobacteria were also found in the plants/plastids, or in an isolated species from some other groups (noted appropriately), were also retained. The proteins, which were uniquely found in a given species or strain were not examined in this work. For all cyanobacterial proteins that are specific for various clades or subgroups, their accession numbers, any information regarding cellular functions, and protein lengths, were tabulated and are presented. Identification of new conserved indels that are specific for cyanobacterial clades was carried out as described in our earlier work 2223.