Introduction

The ATP-binding cassette (ABC) systems are involved in the transportation of a wide variety of substances as well as in cellular processes and their regulation. The main parts of this system, the ABC transporters, are also the most common and important part of the active transport systems which consume energy in the transport process and play an essential role in the flow of substances in cells (Higgins 1992). It has been well documented that various substrates, including metabolic nutrition (e.g., amino acid, maltose, phosphonate, ferric iron, and polyamine), multiple antibiotics, and secreted protein (e.g., lipids, polysaccharides, and alpha-hemolysin) are transported by ABC systems. Based on the transport direction, they can be classified into import systems and export systems (Atkins et al. 2006; Cornillot et al. 2002; Davidson et al. 2008). The import systems are responsible for accumulating small ions and macromolecules within the cell, and are found only in prokaryotes. The export systems are involved in effluxing noxious substances and the targets of membrane components out cell, and can be found in all living organisms. Recently, genome-wide analysis of many completely sequenced genomes has revealed that ABC systems occupy a large proportion of encoded proteins, sometimes they accounting for 3–7% in different organisms (Ren and Paulsen 2005). For example, Escherichia coli were revealed to have 79 ABC systems, representing about 1.8% of total number of genes (Linton and Higgins 1998). In prokaryotes, a typical ABC transporter consists of three components: the extracellular high-affinity substrate-binding protein, the transmembrane protein, and the ATP-binding protein. Among them, the ATP-binding protein, hydrolyzing ATP to provide free energy, is characterized to be the most conserved part of ABC transporter (Tomii and Kanehisa 1998).

Cyanobacteria, dating back 2.5–3.5 billion years, constitute a single but large taxonomic and phylogenetic group within the domain of eubacteria (Chisholm et al. 1988). Among them, Synechococcus and Prochlorococcus have made great contributions to earth’s photosynthetic biomass. To date, only a limited number of ABC systems in Synechococcus and Prochlorococcus have been investigated in detail. However, these characterized transporters offered new insights into their important physiological roles in cyanobacteria. In the carbon fixation process, the accumulation of inorganic carbon into the cytosol is dependent on CO2 and HCO3 − ABC transporters (Badger and Price 2003). In marine conditions with phosphate in low concentration, Synechococcus and Prochlorococcus rely on ABC transporters of PstSACB and PhnDEC to uptake phosphate as inorganic phosphorus sources and phosphonates/phosphites as organic phosphorus sources, respectively (Su et al. 2007). Moreover, synthesis of enzymes such as coenzyme vitamin B12 and nitrile hydratase relies on a high-affinity uptake transporter system encoded by cbiOOQ (Rodionov et al. 2006). Phylogenetic analysis can be instructive to highlight the evolutionary history of given protein families. However, a comprehensive survey on the composition and evolutionary mechanism of ABC systems in Synechococcus and Prochlorococcus still does not exist.

Currently, the breakthrough of high-throughput genome sequencing technologies brings us a great opportunity to extensively sequence Synechococcus and Prochlorococcus genomes. At present, there are 27 Synechococcus and Prochlorococcus genomes with different physiological features available on the Integrated Microbial Genomes (IMG) database (http://img.jgi.doe.gov). Such resources have offered a golden opportunity to explore the evolutionary mechanism of ABC systems in Synechococcus and Prochlorococcus and further their functional implication. In this study, we performed comparative genomic and molecular evolutionary analyses, aiming at investigating the repertoire and evolutionary mechanisms of ABC systems in Synechococcus and Prochlorococcus. As a result, different strains were found to contain a divergent number of ABC systems, and in particular that fresh-water living Synechococcus contain more number of ABC systems than those of marine ones (marine-living Synechococcus and Prochlorococcus). Molecular evolutionary analysis indicated that lineage-specific gene expansion and gene duplication may be the important forces driving the variability of ABC systems in fresh-water Synechococcus and such an expansion may be related to their ecological tolerance. At the domain level, several ABC systems were found to be fused with many additional domains after the divergence from their common ancestor. In addition, the characterized features of the conserved ABC systems present in all marine-living Synechococcus and Prochlorococcus are also discussed.

Materials and Methods

Data Resource

The predicted proteins from 27 Synechococcus and Prochlorococcus genomes were obtained from the IMG database v.2.5 (http://img.jgi.doe.gov/cgi-bin/pub/main.cgi). The 27 genomes are Prochlorococcus marinus MIT9301, P. marinus MIT9215, P. marinus AS9601, P. marinus MIT9312, P. marinus MIT9515, P. marinus CCMP1986, P. marinus NATL2A, P. marinus NATL1A, P. marinus CCMP1375, P. marinus MIT9211, P. marinus MIT9313, P. marinus MIT9303, S. sp. RCC307, S. sp. WH 7803, S. sp. WH7805, S. sp. CC9311, S. sp. WH8102, S. sp. CC9902, S. sp. CC9605, S. sp. BL107, S. sp. RS9917, S. sp. RS9916, S. elongatus PCC6301, S. elongatus PCC7942, Thermosynechococcus elongatus BP-1, S. sp. JA-2-3Ba, and S. sp. JA-3-3Ab.

Identification and Annotation of Putative ABC Systems

To identify ABC systems, the ABC_tran domains (the ATP-binding domain of ABC transporters) of a set of well-characterized and putative ABC system protein sequences (obtained from the KEGG database release 44.0, http://www.genome.jp/kegg/ortholog/tab02010.html) were used as query sequences in a local BLASTP search against the proteins of 27 genomes. A stringent E-value of E−10 was used to filter BLAST hits. Putative ABC systems identified through this process were added to the query set to perform another round of BLASTP searches. The procedure was continued until no new protein could be found. ABC systems in other cyanobacterial genomes, such as Gloeobacter, Crocosphaera, Synechocystis, Trichodesmium, Anabaena, Nostoc, and Nodularia, were identified in the same way.

After an ABC system was identified, the hmmpfam program implemented in the HMMER package 2.3.2 (http://hmmer.janelia.org) was employed to predict its domain architecture through searching against the Pfam database release 22.0 (http://pfam.janelia.org) with an E-value of 0.01. The ABCISSE database (http://www.pasteur.fr/recherche/unites/pmtg/abc/database.html) provides comprehensive information of ABC systems, such as ABC transporter classification and predicted function based on similarity against proteins with experimentally derived function. Similar to other studies (Atkins et al. 2006; Nicolas et al. 2007), assignment of specific substrate to each ABC system was mainly carried out by comparing to the ABCISSE database using BLASTP searches with a stringent E-value of E−50. Since the ABC system and its substrate-binding protein often form an operon, the neighbor genes of each ABC system and the annotated information of best hit genes from the KEGG database were also considered in order to increase the robustness of the substrate prediction.

Multiple Alignment and Phylogenetic Analysis

The multiple alignments of ABC_tran domain sequences were conducted with the ClustalX 1.83 program (http://www.ebi.ac.uk/clustalw) using default parameters. The phylogenetic tree was reconstructed using the Neighbor-Joining method implemented in the QuickTree 1.1 program (http://www.sanger.ac.uk/Software/analysis/quicktree/). The reliability of the tree was evaluated with 1,000 replicates of a bootstrapping test and its visualized operation was carried out using the MEGA 3.1 program (http://www.megasoftware.net/).

Results and Discussion

Uneven Distribution of ABC Systems in Synechococcus and Prochlorococcus

To identify all ABC systems, we performed exhaustive BLASTP searches of the Synechococcus and Prochlorococcus genomes with the ABC_tran domains of a set of well-characterized and putative ABC transporters. Across all 27 strains, a total of 934 putative ABC systems were identified (Fig. 1 and Table S1). Synechococcus and Prochlorococcus genera were revealed to have a close phylogenetic relationship with relatively compact genomes (Urbach et al. 1992). Interestingly, different Synechococcus and Prochlorococcus strains were found to contain divergent number of ABC systems, especially that fresh-water Synechococcus contain a greater number than those of marine ones (marine-living Synechococcus and Prochlorococcus). All Prochlorococcus strains encode less than or equal to 38 ABC systems. Similarly, all marine-living Synechococcus strains also have a limited number of ABC systems (usually below 41). On the contrary, the fresh-water Synechococcus contain much more ABC systems (Mann–Whitney U-test, P < 0.01). Across the genome of Synechococcus sp. JA-2-3Ba, a total of 58 putative ABC systems were identified, representing the most abundant strain in all 27 cyanobacterial genomes.

Fig. 1
figure 1

Distribution of ABC systems across the phylogenetic tree of Synechococcus and Prochlorococcus. The phylogenetic tree was constructed based on their 16 s rRNA using the Neighbor-Joining method. The reliability of the tree was evaluated with 1,000 replicates of bootstrapping test and only high bootstrap value scores (>50%) were indicated on the branches. In addition, each strain is followed by its living habitat, total number of ORFs, as well as absolute and relative number of ABC systems

Full size image

Genome size has been indicated to have an important effect on the divergent number of protein families, and it is well known that the increase in bacterial genome size is accompanied with the accumulation of paralogous protein families (Konstantinidis and Tiedje 2004). Taking genome size effect into account, we calculated the relative number of ABC systems (ORFs divided by the putative number of ABC systems). It is shown that the percent of ABC systems in fresh-water Synechococcus is greater than 1.6%, whereas in marine strains it is less than or equal to 1.6%. These statistics demonstrate that the relative number of ABC systems in fresh-water Synechococcus is still significantly higher than that in marine ones (Mann–Whitney U-test, P < 0.01).

The statistical analysis was made on 27 available Synechococcus and Prochlorococcus genomes. Among them, however, only five genomes are fresh-water living strains, which may cause bias in the analysis. At present, beside Synechococcus and Prochlorococcus, there are also many other cyanobacterial genomes available, including Crocosphaera watsonii WH8501, Synechocystis sp. PCC6803, Trichodesmium erythraeum IMS101, Anabaena variabilis ATCC29413, Anabaena sp. PCC7120, Nostoc punctiforme PCC73102, Acaryochloris marina MBIC11017, Cyanothece sp. ATCC 51142, Cyanothece sp. PCC 7424, Cyanothece sp. PCC 7425, and Cyanothece sp. PCC 8801. When we included them into analysis (Table S2), the statistic result still indicated that fresh-water living cyanobacteria contain more ABC systems than those of marine ones (Mann–Whitney U-test, P < 0.01). For example, marine-living strain C. watsonii WH8501 with 6,000 ORFs was found to have only 45 ABC systems, representing only 0.75% of total number of genes. Similarly, another marine-living cyanobacterium, T. erythraeum IMS101, possesses 47 ABC systems which only account for 0.9% of its large genome (5,124 ORFs). However, fresh-water or soil-living strains encode more ABC systems, such as Synechocystis sp. PCC6803 (54/3,626), A. variabilis ATCC29413 (96/5,764), Anabaena sp. PCC7120 (98/6,217), and N. punctiforme PCC73102 (95/7,745).

Compared with marine ecosystem, the fresh-water ecosystem is considered to be a less stable environment that has frequently changing light and temperature, unstable availability of nutritious substances, and more unpredictable threats. In order to survive in such an unstable environment and construct a long-term evolutionary selective advantage, it is indicated that the fresh-water living cyanobacteria have to encode more genes within some specific gene families. In our previous studies, several gene families related to defense mechanisms, signal transduction, and transcription regulation have already been confirmed to have distinctive numbers in cyanobacteria living in different environments. Examples for these gene families including restriction–modification-related proteins (Zhao et al. 2006), Ser/Thr-specific protein kinases (Zhang et al. 2007), purine nucleotide cyclases (Wu et al. 2008), and transcription factors (Wu et al. 2007). Here, ABC systems may also be involved in such a strategy and their essential role of actively exchange substances with the surrounding environment may have made great contributions to adaptation in ever changing condition.

Lineage Expansion of ABC Systems in Fresh-Water Synechococcus

To explore the evolutionary history of ABC systems in both fresh-water Synechococcus and marine-living strains, a phylogenetic tree of ABC systems from all 27 Synechococcus and Prochlorococcus strains was constructed based on ATP-binding domains of ABC systems using the Neighbor-Joining algorithm. The reason for using ATP-binding domains of ABC systems rather than the whole coding sequences to perform phylogenetic analysis lies in the fact that many ABC systems are multi-domain proteins with different sequence lengths and could not be aligned with high confidence. On the contrary, the ATP-binding domain has been proved to be considerably conserved across all the ABC transporter structure domains (Davidson et al. 2008; Linton and Higgins 1998). In addition, their lengths in various proteins are quite similar (about 220 amino acid residues) and could provide enough information for constructing a reliable phylogenetic tree. From the phylogenetic tree (Fig. 2 and Table S1), it is clear that most of the ABC systems were grouped into many separate families with high bootstrap supports (>50%). Meanwhile, members of each ABC system subfamily are found to have similar substrates, providing additional support for the results of phylogenetic tree construction (Table S1).

Fig. 2
figure 2

Phylogenetic tree of all of the ABC systems in Synechococcus and Prochlorococcus. The phylogenetic tree is constructed based on the ABC_tran domains of ABC systems. Strains names are cut off for brevity on the phylogenetic tree using the Neighbor-Joining method. The branches of 15 common ABC systems families are marked in deep green; the branches of the less conserved groups are marked in light green; the branches of expanded fresh-water-specific Synechococcus and marine-living strains are separately marked in deep blue and light blue, respectively. Representative families were displayed out labeled with family name followed by putative substrate in bracket

Full size image

Careful observation of interior branches in each cluster showed that there are 15 “indispensable” ABC system families universally present in all Synechococcus and Prochlorococcus strains (Fig. 2). These families represent a large proportion (15/25) of total number of ABC systems in the genome of P. marinus CCMP1375, and as much as 25.86% (15/58) of the ABC systems in S. sp. JA-2-3Ba, which has the largest number of ABC systems. Obviously, these ABC system families should be derived from the same common ancestor. Therefore, they can be considered as a core set of ABC systems conserved in all the Synechococcus and Prochlorococcus, and are supposed to be associated with the basic physiological functions. Among these 15 families, there are 5 involved in the uptake of phosphate [mineral and organic import (MOI) family], phosphonates [phosphonates phosphonates and phosphites (PHN) family], ferric iron (MOI family), cobalt [CBU subfamily of CBY (cobalt uptake) family], and oligopeptides [oligopeptides and nickel (OPN) family]; one very long chain fatty acids (FAE) family related to the export of fatty acids; one o228 family related to the export of antimicrobial peptide; three drugs, peptides, lipids (DPL) families in charge of multidrug resistance; and one iron–sulfur center biogenesis (ISB) family related to Fe–S cluster assembly (Table S1). Due to limited functional information, the substrates of the other four conserved families remain unknown.

Interestingly, in addition to these common conserved ABC systems, many fresh-water Synechococcus-specific families without counterparts in marine-living strains were observed (Fig. 2). These families are found to be able to transport substrates quite different from marine-living strains. For example, it was found that one fresh-water-specific cluster, belonging to the MOI family, contains proteins from all five fresh-water strains (including tlr1690, CYB_0238, CYA_0470, Synpcc7942_1680, and syc2410_d). The member of this cluster has the ability to import thiosulfate or sulfate, and such ability is not found in other marine-living strains. Further, they exhibit weak phylogenetic connection (bootstrap value < 50%) with neighbor clusters related to the transport of either ferric iron or oligosaccharides and polyols. Another case is that seven genes (CYA_2437, CYB_1396, CYB_0390, CYA_1583, CYB_1651, CYA_2888, and CYA_1919) are grouped into a single cluster (belonging to the MOS family) which is supposed to be in charge of monosaccharide import. In addition, a group of genes consisting of CYB_1887, CYA_0147, and tll0453 are found to be similar with the Cell division family, and turned out to be the only group of genes belonging to this family in analyzed Synechococcus and Prochlorococcus strains.

Meanwhile, gene duplication events can also be observed in fresh-water Synechococcus-specific ABC system families. For example, in a fresh-water-specific cluster with five ABC systems, CYA_2882, CYB_1657, and tlr1473 are found to have an extra duplicated copy of CYB_2400, CYA_1050, and tlr1901 (multiple sequence alignment is shown in Fig. S1). Similarly, we found that, in the HAA family with eight members, three pairs (tlr0053 and tlr2006, syc0724_c and syc1614_d, and Synpcc7942_0815 and Synpcc7942_2492) of them have experienced an ancient duplication event. This may result in an enhanced ability to import branched-chain amino acids for these three fresh-water Synechococcus strains. These results indicate that lineage-specific expansion of ABC systems genes combined with gene duplication appear to be the major contributors to the great divergence of the numbers of ABC systems between marine and fresh-water living cyanobacteria. The existence of large number of ABC systems with additional new functions could imbue the fresh-water cyanobacteria with the extra possibility of using external nutrition and strong defense.

Besides the frequent occurrence of fresh-water-specific ABC systems, three marine-specific groups were also observed in the phylogenetic tree (Fig. 2). The first one is assigned to the YLU family with an unknown function. The member of the second group contains two ATP-binding domains, belonging to the DNA repair and drug (UVR) family. The third group members are highly similar to the lipid A or glycerophospholip export (LIP) subfamily of the DPL family which is mainly in charge of lipid A export. The existence of marine-specific ABC systems indicated that marine-living cyanobacteria had also formed ABC systems for certain kinds of physiological functions, which may be developed as an adaptation strategy in order to survive in the open ocean.

Domain Organization of ABC Systems in Synechococcus and Prochlorococcus

Previous studies indicated that ABC systems can recruit a great variety of additional domains with distinct functions (Biemans-Oldehinkel et al. 2006). To get a comprehensive understanding of the functional role of ABC systems in Synechococcus and Prochlorococcus, we investigated their domain architectures. Through the Pfam domain assignment, we found that they exhibited diverse domain architectures (Fig. 3 and Table S3), implying versatile functions of ABC systems in Synechococcus and Prochlorococcus. Across all the 934 ABC systems, besides the ABC_tran domain, as many as 13 types of non-ABC_tran domains were detected. The ABC_tran domains and these 13 additional domains result in 17 organization patterns. Among these patterns, the majority of them (788/934) are either just single or tandem ABC_tran domain(s) without additional domains or one ABC_tran domain fused with one basic ABC transporter element, including ABC_membrane, ABC_membrane_2, or ABC2_membrane domain. Interestingly, the single ABC_tran domains are found to be more prevalent in Synechococcus strains than in Prochlorococcus strains, as well as more in fresh-water living strains than marine-living strains. The other 146 ABC systems are fused with various additional domains and exhibit complex architectures.

Fig. 3
figure 3

Domain architectures of ABC systems in Synechococcus and Prochlorococcus. The domain architecture was predicted against the Pfam database using the hmmpfam program implemented in HMMER package with an E-value of 0.01. The number appearing at right of the domain architecture indicates the number of ABC systems with this kind of domain architecture. In addition, the left line chart indicates the relative divergence of domain architecture in Prochlorococcus and Synechococcus and the right line chart shows the relative divergence of domain architecture in marine-living and fresh-water living strains

Full size image

The fusion of additional domains to ABC systems can imbue them with various functional capabilities (Fig. 3). Actually, functions of some of these additional domains have already been clarified by experimental approaches. The CBS domain, which is a well-known sensor for ionic strength that controls the transport activity (Mahmood et al. 2006), is found to be associated with four ABC systems in P. marinus MIT9303 (P9303_16991), Synechococcus sp. RS9917 (RS9917_03548), Synechococcus sp. JA-2-3Ba (CYB_0580), and Synechococcus sp. JA-3-3Ab (CYA_2231). It is suggested that the increased concentration of endocellular ions was first sensed by the CBS domain, and then ABC system (mediating the uptake of compatible solutes such as glycine betaine) was activated, thereby inverting the osmotic shrinkage trend of the cell. Transport-associated oligonucleotide or oligosaccharide-binding domains (TOBE and TOBE_2) were formerly documented to be involved in the recognition of small ligands, such as molybdenum and sulfate. We found that, in cyanobacteria, TOBE and TOBE_2 domains containing ABC systems show high similarity with ABC importers of ferric iron, oligosaccharide, polyol, and polyamine. Otherwise, several ABC systems were found to be encompassed by enzyme domains involved in catalytic roles. Peptidase_C39 domain was mainly observed in prokaryotes and functioned by cleaving off double-glycine type signal peptides during maturation (Havarstein et al. 1995). It might be possible that ABC transporters with peptidase_C39 domain have a dual function: removal of the signal peptide from its substrate followed by translocation of its substrate across the cytoplasmic membrane. Two ABC systems in P. marinus MIT9515 (P9515_19031) and P. marinus CCMP1986 (PMM1712) were also found to fuse with APS_kinase domain, which may catalyze the phosphorylation of adenylylsulfate to 3′-phosphoadenylylsulfate, and thereby display the dual role of transporter and kinase.

Cyanobacteria were revealed to have a larger number of signal transduction modules in order to adapt to various environmental conditions (Ashby and Houmard 2006; Zhang et al. 2007). In this study, many new ABC systems were found to be fused with signal transduction modules, including forkhead-associated (FHA) and cNMP_binding (cyclic nucleotide-binding) domains (Fig. 3). FHA is a phosphopeptide recognition domain often found in protein kinases and transcription factors (Durocher et al. 1999). We found that four ABC systems (P9303_28591, tlr1253, tll1329, and tll1340) were fused with this domain. An ABC system Rv1747 with similar domain architecture (FHA + FHA + ABC_tran) from Mycobacterium tuberculosis was well characterized to interact with the Ser/Thr protein kinase through phosphorylation (Molle et al. 2004). The cNMP_binding domain is linked to bind small-molecules, in particular the second messengers (cAMP and cGMP). It is likely that the activity of these ABC systems is regulated by outside signals and thereby controls the exchange of substances with the outside environment in response to environmental or metabolic changes.

Evolution of protein families is not only well known to involve gene gain and loss, but also by domain recruitment (Pasek et al. 2006). To elucidate the evolutionary mechanism of domain architectures of ABC systems, we investigated the domain architectures of all ABC systems in Synechococcus and Prochlorococcus across their phylogenetic tree. It can be seen that the domains of ABC systems are generally organized in the same way within each subfamily, suggesting that the domain architectures of most ABC systems subfamilies in Synechococcus and Prochlorococcus were fixed before their divergence (Table S1). These fixed domain architectures occupy a large amount in all the identified ABC systems, such as single ABC_tran, ABC_tran + ABC_membrane, and ABC_tran + ABC_membrane_2.

However, there are several interesting exceptional cases, indicating the occurring of domain recruitment during the evolution of ABC systems in Synechococcus and Prochlorococcus (Fig. 4). As an example, most of ABC systems related to osmoprotectant have only the ABC_tran domain, whereas four orthologous ABC systems (RS9917_03548, P9303_16991, CYA_2231, and CYB_0580) are fused with an extra CBS domain in the C-terminal (Fig. 4a). Likewise, P9515_19031 and PMM1712 with an additional APS domain were distinctive from the other proteins in the same group, suggesting the occurrence of domain recruitment in both P9515_19031 and PMM1712 (Fig. 4b). The orthologs of hemolysin exporters were commonly organized in the form of ABC_membrane + ABC_tran. However, several members were attached by a cNMP_binding domain and two of them (P9303_01291 and WH7805_08511) are also inserted by a Peptidase_C39 domain between the cNMP_binding and ABC_membrane (Fig. 4d). For the FHA-containing ABC systems, it is obvious that tll1329 and tll1340 were formed by gene duplication and tll1340 was fused by an additional FHA domain after the gene duplication event (Fig. 4c). The above observations indicated that the widely distributed domain architectures were formed in the ancestor of Synechococcus and Prochlorococcus, whereas the ABC systems with complicated additional domains might be formed by the recruitment of additional domains. The selective advantage of these domain architectures may lie in the increased efficiency by integrating different physiological functions through the co-expression of the fused domains. It is suggested that these complex domain architectures might be a reflection of physiological diversities of cyanobacteria and their strong ability to adapt to various environmental conditions.

Fig. 4
figure 4

Cases of domain fusion occurred during the evolution of ABCs in Synechococcus and Prochlorococcus. The phylogenetic tree was constructed based on the ABC_tran domains of ABC systems using the Neighbor-Joining method. The reliability of each branch was tested by 1,000 bootstrap replications and branches with bootstrap support >50% had been marked. The family name and putative substrate in bracket is labeled above the phylogenetic tree. Corresponding domain architectures are shown on the right of the branches

Full size image

The Common Repertoire of ABC Systems in Marine-Living Synechococcus and Prochlorococcus

After all ABC systems were identified, we were interested in identifying the core ABC systems (ABC systems universally presented) in marine-living Synechococcus and Prochlorococcus, which live in relatively similar habitats and have a close phylogenetic relationship. We performed this analysis under the belief that it provides considerable insights into the evolution of common repertoire of ABC systems in marine-living Synechococcus and Prochlorococcus. As the species were restricted in these strains, a total number of 19 core ABC system families were detected (Table 1 and Table S4). According to the annotated information of the best hits from the ABCISSE database and the KEGG database as well as their flanking genes, the function of 14 ABC system families can be assigned and grouped into 3 categories: import system, export system, and cellular process (Table 1). Due to limited functional information, physiology roles of the other five ABC systems still remain unknown.

Table 1 The core ABC transporter families present in all marine-living Synechococcus and Prochlorococcus
Full size table

Among the 14 annotatable core ABC system families, 6 appear to serve as import system. The first two systems (PstSACB and PhnDEC) are involved in phosphate utilization (Su et al. 2007). The PstSACB system is assigned to the MOI family and consists of four components: a substrate-binding protein (PstS), two transmembrane proteins (PstC and PstA), and an ATP-binding protein (PstB). This system is responsible for the active transport of inorganic phosphate into the cell. The PhnDCE system, belonging to the PHN family, forms an operon with a substrate-binding protein (PhnD), a transmembrane protein (PhnE), and an ATP-binding protein (PhnC). This system utilizes phosphonate or phosphite as alternative sources of phosphorus. With these two high-affinity Pi transporters, cyanobacteria may be able to efficiently use phosphate, phosphonate, or phosphite as phosphorus sources under Pi limited conditions. The second universal ABC system contains an ATP-binding protein FutC (belonging to MOI family), which was related to the uptake of ferric iron. Iron plays crucial roles in physiological and morphological aspects of cyanobacteria, since its deficiency may cause a decreased amount of light-harvesting phycobilisomes and thylakoids (Webb et al. 2001). In the previous studies, the ferric ABC system FutABC was identified as a major contributor to ferric iron import through the plasma membrane. It contains genes futA1 and futA2 (encodes substrate-binding protein), futB (encodes transmembrane protein), and futC (encodes ATP-binding protein) (Katoh et al. 2001).

Cobalt is a central component in the corrin ring of vitamin B12, which is a cofactor involved in various important enzymatic reactions, such as the methyl group transfer and the rearrangement reaction (Rodionov et al. 2006). Y179 (CBU-like systems) subfamily of the CBY family might be the main system for high-affinity uptake of cobalt from the surrounding environment, where the nutrient is inconstantly available. Moreover, a manganese import system MntCAB (MntC as substrate-binding protein, MntA as ATP-binding protein, and MntB as transmembrane protein) was found to be present in all marine-living Synechococcus and Prochlorococcus strains. Several reports have demonstrated that manganese is required for photosynthetic growth in plants and cyanobacteria for its specific repertoire in the water oxidizing activity of photosystem II (Hillier and Babcock 2001). The last import system, identified as the OPN family, is likely to be involved in the uptake catabolites of exogenous proteins and/or recycling of cell–wall peptides, thus providing the sources of nitrogen and metabolic energy (Detmers et al. 2001).

ABC exporters in bacteria are characterized by secreting various molecules, such as peptides, lipids, hydrophobic drugs, polysaccharides, and related proteins (including hemolysin, heme-binding protein, and alkalineprotease). Here, six ABC system families are found to be the exporters present in marine-living Synechococcus and Prochlorococcus. Among them, an exporter family, belonging to the DIP (lipid A or glycerophospholip export) subfamily of the DPL family (drugs, peptides, lipids), encodes proteins homologous to the MsbA subfamily of ABC systems. In E. coli, MsbA system is well investigated to transport precursors, such as lipid A and glycerophospholipids, during the lipopolysaccharide biosynthesis to protect the membrane from certain kinds of chemical attack (Zhou et al. 1998). In addition, two ABC families are highly similar with the LLP subfamilies (LIP-like exporters) of the DPL (drugs, peptides and lipids) family. The fourth one is assigned to the DRR (polyketide drug resistance) subfamily of the DRA family, which is likely to confer to cells resistance to doxorubicin, daunorubicin, oleandomycin, and mithramycin by catalyzing the active discharge of the toxins across the membrane. The fifth export family is the DevA protein, belonging to the o228 family. Interestingly, gene DevA together with flanking DevB and DevC, form a DevABC operon, is conserved in all marine-living Synechococcus and Prochlorococcus. It is suggested that this operon is responsible for exporting glycolipids or other related enzymes which are necessary to the formation of the laminated layer (Fiedler et al. 1998). Based on the sequence similarity, the sixth export family is designated as the FAE (very long chain fatty acids) family. The proteins of this ABC system family may be related to the transport of fatty acids and/or fatty acyl-CoAs to peroxisome.

Besides transport, more studies have shown that certain ABC system families can also be related to cellular processes (Atkins et al. 2006; Cornillot et al. 2002; Davidson et al. 2008), such as DNA repair, translation, or regulation of gene expression. In this study, we found that two ABC system families might be involved in cellular process rather than the transport function. The first one is the sufC gene (belonging to the ISB family), which is located between sufB and sufD. The SufBCD complex involves in the iron–sulfur cluster biogenesis as a source of energy required by iron acquisition (Nachin et al. 2003). It is indicated that the SufBCD complex presents unambiguously within the cytoplasm and enhances the cysteine desulfurase activity of SufSE complex. The second one is less well documented, and is classified into the REG (gene expression regulation) subfamily of antibiotic resistance and translation regulation (ART). The defined function of this subfamily has been proven by the yeast protein GCN20 (Vazquez de Aldana et al. 1995). It has been proven that the interaction between GCN0 and GCN1 could activate the kinase GCN2 which then phosphorylates the eukaryotic translation initiation factor eIF2. This leads to a rise of the translation of the transcriptional activator GCN4 in amino acid-starved cells. Furthermore, a highly homologous ABC transporter ChvD in Agrobacterium tumefaciens was found to be inactivated in mutants picked up for reduced transcription of the virA and virG genes (Liu et al. 2001). Thus, it is likely that the REG subfamily of cyanobacteria ABC transport also plays certain roles in the regulatory process of gene expression.

In summery, in this study, uneven phylogenetic distribution of ABC systems in marine and fresh-water cyanobacteria strains was found, and indicated by phylogenetic analysis, the important driving forces of this uneven distribution may be the lineage-specific gene expansion and duplication. In addition, at the domain level, many additional domains were found to fuse with ATP-binding domains, giving rise to the versatile functions to ABC systems in cyanobacteria. Finally, 19 ABC systems (14 systems with detailed information) conserved in all marine-living Synechococcus and Prochlorococcus were identified. In conclusion, the comprehensive survey of ABC systems in Synechococcus and Prochlorococcus provides novel insights into their potential evolutionary mechanism which may be the starting point for further investigation of their physiological roles.