Structure and Molecular Phylogeny of sasA Genes in Cyanobacteria: Insights into Evolution of the Prokaryotic Circadian System
http://www.100md.com
分子生物学进展 2004年第8期
* Osteoporosis Research Center and Department of Biomedical Sciences, Creighton University
Institute of Evolution, University of Haifa, Mount Carmel, Haifa, Israel
E-mail: dvornyk@creighton.edu.
Abstract
Cyanobacteria are the simplest organisms known to have a circadian system. In addition to the three well-studied kai genes, kaiA, kaiB, and kaiC, an important element of this system is a two-component sensory transduction histidine kinase sasA. Using publicly available data of complete prokaryotic genomes, we performed structural and phylogenetic analyses of the sasA genes. Results show that this gene has a triple-domain structure, and the domains are under different selective constraints. The sasA gene originated in cyanobacteria probably through the fusion of the ancestral kaiB gene with a double-domain, two-component sensory transduction histidine kinase. The results of the phylogenetic analyses suggest that sasA emerged before the kaiA gene, about 3,000–2,500 MYA, and has evolved in parallel with the evolution of the kaiBC cluster. The observed concordant patterns of the sasA and kaiBC evolution suggest that these genes might compose an ancient KaiBC–SasA-based circadian system, without the kaiA gene, and that such a system still exists in some unicellular cyanobacteria.
Key Words: sasA ? circadian system ? prokaryotes ? evolution ? cyanobacteria
Introduction
Circadian rhythmicity in regulation of endogenous physiological processes is a characteristic feature of eukaryotic organisms (Sweeney 1987; Pittendrigh 1993). Cyanobacteria are the simplest organisms known to manifest circadian clock (Kondo and Ishiura 1999). A cluster encoding three genes, kaiA, kaiB, and kaiC, was identified in the unicellular cyanobacterium, Synechococcus elongatus PCC 7942, as a key component of the circadian system (Ishiura et al. 1998). The clock genes are ubiquitous in cyanobacteria (Lorne et al. 2000; Dvornyk, Vinogradova, and Nevo 2003). They have been shown to play an important role in high adaptiveness of these prokaryotes to various environmental conditions (Ouyang et al. 1998; Johnson, Golden, and Kondo 1998; Johnson and Golden 1999; Dvornyk, Vinogradova, and Nevo 2002).
Although the three kai genes are essential in maintaining circadian oscillation (Ishiura et al. 1998; Iwasaki et al. 2002; Kitayama et al. 2003; Xu, Mori, and Johnson 2003), a number of other genes have been determined to participate in the input and output pathways of the cyanobacterial circadian system (Tsinoremas et al. 1996; Kutsuna et al. 1998; Katayama et al. 1999; Schmitz et al. 2000; Katayama et al. 2003). One of the latter is sasA, a kaiC-interacting sensory histidine kinase, which is important for robust circadian rhythmicity (Iwasaki et al. 2000; Kageyama, Kondo, and Iwasaki 2003).
Recently we showed that the kai genes have quite a different evolutionary history and occur also in some archaea and proteobacteria. The kaiA and kaiB genes originated in cyanobacteria, and the three-gene cluster, kaiABC, evolved about 1,000 MYA (Dvornyk, Vinogradova, and Nevo 2003). However, since the cyanobacterial circadian system also includes genes other than kai, the question is raised: how has this system evolved?
In the present work, we comprehensively studied structure and sequence diversity of the sasA genes from cyanobacteria. We also tried to reconstruct their phylogeny and to compare it to those of the kaiB and kaiC genes in order to find out when and how the sasA genes became involved into the regulation of circadian rhythmicity in cyanobacteria.
Materials and Methods
DNA and Protein Sequences
The annotated and homologous sequences of the sasA, kaiB, and kaiC genes were retrieved from GenBank by using the gapped BLASTP and PSI-BLAST (Altschul et al. 1997) and the respective amino acid sequences of Synechococcus elongatus PCC 7942 (GenBank accession Q06904, BAA37102, and BAA37103) as the probes. The kai sequences of only those species, which have the sasA genes, were used in the analyses. The retrieved protein and nucleotide sequences were aligned using CLUSTALW (Thompson, Higgins, and Gibson 1994) and manually adjusted based on structural considerations (Ishiura et al. 1998; Iwasaki et al. 2000). Alignments of the nucleotide sequences were modified manually according to the respective amino acid alignments. These multiple alignments are available upon request. The list of the used sequences is given in table 1.
Table 1 Sequences and Species Used in the Study.
Phylogenetic Analysis
The rate of nonsynonymous nucleotide substitutions per nonsynonymous site (dN) was calculated using the modified Nei-Gojobori method (Nei and Gojobori 1986) with Juke-Cantor correction for multiple substitutions at the same site and the transitions/transversions ratio equal to 1.8. The MEGA 2.1 software (Kumar et al. 2001) was used for the computations of dN.
In the comparative phylogenetic analysis, we proceeded from the evidence that the kaiB and kaiC genes, after fusion into in the kaiBC cluster, have evolved as a unit rather than independently (Dvornyk, Vinogradova, and Nevo 2003). The phylogenetic trees for the sasA genes and kaiBC cluster were obtained using respective amino acid sequences and a neighbor-joining (NJ) algorithm (Saitou and Nei 1987) with Poisson correction as implemented in the MEGA 2.1 software (Kumar et al. 2001). The same method and software were used to perform a phylogenetic reconstruction for the kaiB genes and KaiB-like domain of the sasA genes. Statistical significance of the nodes in all derived trees was evaluated by the bootstrap procedure with 1,000 replications.
Analysis of Positive Selection
Comparing synonymous (silent, dS) and nonsynonymous (amino acid-changing, dN) substitution rates in protein-coding genes provides important information for understanding molecular evolution. The nonsynonymous/synonymous rate ratio, = dN/dS, measures selective pressure at the protein level. The values of = 1, <1, and >1, indicate neutral evolution, purifying selection, and positive selection, respectively. To test the sasA genes for the presence of natural selection, we used the site-specific models, which do not average dN/dS ratio over sequences but allow it to vary among amino acid sites (Nielsen and Yang 1998; Yang et al. 2000). The models were pairwise compared by the likelihood ratio test (LRT). This was done by comparing the log-likelihood values with = 1 constrained and without such constraint. If the null hypothesis = 1 is correct, twice the log-likelihood difference between the two models (2) asymptotically has a 2 distribution with df = 1. All computations were performed by PAML software (Yang 1997).
Analysis of Functional Divergence
There are two main types of functional divergence. Type I functional divergence after gene duplication results in altered functional constraints (i.e., different evolutionary rate) between duplicate genes, whereas type II results in no altered functional constraints but radical change in amino acid property between them (e.g., charge, hydrophobicity, etc.) (Gu 1999).
The type I functional divergence between the different clades of the sasA gene tree was analyzed with the method proposed by Gu (Gu 1999). We estimated a rate correlation between the members of the different clades, or a coefficient of functional divergence, , as implemented in DIVERGE v.1.04 software (Gu and Vander Velden 2002). Using this method, we also identified critical amino acid residues that may be responsible for the observed functional divergence.
Results
Structure and Sequence Polymorphism of sasA Genes in Cyanobacteria
Screening available prokaryotic genomes using BLAST search revealed the large number of sasA homologs in many prokaryotic taxa. This was expected, because sasA belongs to the large superfamily of two-component sensory transduction histidine kinases widely distributed in prokaryotes (Nagaya, Aiba, and Mizuno 1993). Genes of this superfamily often have a three-domain structure (Dutta, Qin, and Inouye 1999). One of the domains is histidine kinase A phosphoacceptor domain (HisKA), another is related to the family of histidine kinase-like ATPases (HATPase_c). The former is an element of the two-component signaling systems, consisting of a histidine protein kinase that senses a signal input and a response regulator that mediates the output (Stock, Robinson, and Goudreau 2000). The latter belongs to the ATP binding superfamily that includes diverse protein families such as DNA topoisomerase II, molecular chaperones Hsp90, DNA-mismatch-repair enzymes, phytochrome-like ATPases, and histidine kinases (Dutta and Inouye 2000). The third domain is a sensor domain (Dutta, Qin, and Inouye 1999).
In contrast to the other two-component sensory transduction histidine kinases, the sensor domain of the sasA kinases is homologous to the kaiB genes (Iwasaki et al. 2000) (fig. 1A). Importantly, the BLAST search did not reveal the sasA homologs with the KaiB-like domain in genomes of prokaryotes, other than cyanobacteria, including those that possess the kaiBC cluster transferred from cyanobacteria (Dvornyk, Vinogradova, and Nevo 2003).
FIG. 1. Amino acid sequence alignment of the three domains of the sasA genes. Black- and gray-shaded backgrounds indicate different degree of conservation (black is the most conserved). The upper numbers indicate positions in the alignment of the full sequences; the numbers on the right indicate positions in the respective sequences. (A) KaiB-like domain. Putative hydrophobic cores are underlined. Amino acid residues, whose mutations in KaiB alter circadian periods, are marked by arrows. (B) HisKA domain. Putative autophosphorylation histidine residue is marked by block arrow. (C) HATPase_c domain. Putative ATP binding sites and Mg2+ binding site are marked with "#" and "+," respectively. G-X-G-motifs are marked with square brackets
The structure of the KaiB-like domain of the SasA protein is shown in figure 1A. It is 86 amino acid residues long (corresponds to residues 14–99 in the S. elongatus PCC 7942 SasA sequence) and has a number of putative motifs and highly conserved residues. The first motif (positions 36–42 in the alignment, fig. 1A) is six or seven residues long, and at least five are hydrophobic. This motif is one of two putative hydrophobic cores of this domain. Another corresponds to the positions 87–97 of the alignment and contains at least eight hydrophobic residues (fig. 1A). Comparing these motifs against public databases revealed no homologs. Aligning the KaiB-like domain of SasA with the KaiB proteins (the alignment not shown) indicated that both putative hydrophobic cores in KaiB are also highly conserved, although they differ by sequence from those of the KaiB-like domain. Motifs of the KaiB-like domain contain conserved amino acid residues, whose mutations in KaiB alter circadian periods (Iwasaki et al. 2000). Those are L40 and R/K102 in the alignment (fig. 1A). In addition, several other residues, which are fully conserved in the KaiB-like domain and KaiB, can be identified. In figure 1A, those correspond to three proline residues (positions 80, 92, and 99), L94, and G107. Interestingly, all these conserved residues are nonpolar.
The HisKA and HATPase_c domains are 59–71 and 111–130 amino acid residues long, respectively (fig. 1B, C). They have a number of conserved characteristic features. The HisKA domain contains putative autophosphorylation histidine residue (Iwasaki et al. 2000) (fig. 1B); the HATPase_c domain has several ATP binding sites, an Mg2+ binding site, and two G-X-G motifs (fig. 1C). The motifs are located in loops defining the top and bottom of the ATP binding pocket (Obermann et al. 1998).
The three domains of the sasA genes differ by the number of accumulated nonsynonymous substitutions. As the data in table 2 suggest, the HATPase_c domain is the most conserved among the three (dN = 0.412 ± 0.046), the KaiB-like domain has dN = 0.614 ± 0.068, whereas the HisKA domain is the least conserved (dN = 0.840 ± 0.112). The regions not belonging to any of the three domains have dN = 0.836 ± 0.077.
Table 2 Patterns of Nonsynonymous Nucleotide Substitutions (dN) in the Different Domains of the sasA Genes.
Comparative Phylogeny of sasA and kai Genes in Cyanobacteria
All sasA genes found in cyanobacteria occurred in a single copy. The resulting consensus tree shows that the sasA genes are separated into two clades (fig. 2). Clade A comprises sasA from the species, which have all three kai genes. Another clade includes the sasA genes from cyanobacteria, which possess either all the three clock oscillators (Synechococcus sp. WH 8102) or only two of them, kaiB and kaiC (Prochlorococcus strains). The trees of the kaiBC cluster and sasA genes from the same species have similar topologies (fig. 2). In particular, two main clades of each tree, A and B, consist of the same species. Importantly, of two kaiBC clusters of Synechocystis sp. PCC 6803, one that is not in the cluster with the kaiA gene (Syncys3), forms a branch discordant with the tree topology of the sasA genes.
FIG. 2. Concordant NJ phylogenetic trees of the sasA genes and kaiBC cluster from cyanobacteria. Bootstrap values <50% are not shown. The kaiBC sequences are denoted according to the respective kaiC sequences (table 1)
The sasA genes from the different branches indicate different patterns of nonsynonymous nucleotide substitutions in their domains (table 2). The sasA genes of clade A have significantly higher dN values (P 0.05) in all the three domains than do the genes of clade B.
The phylogeny of the kaiB genes and KaiB-like domains of the sasA genes from the studied species show their clear separation into four clades, all with very high bootstrap support for each clade (fig. 3). Three clades (B1, B2, and B3) correspond to those previously reported for the kaiB genes (Dvornyk, Vinogradova, and Nevo 2003). The fourth clade, B4, comprises the KaiB-like domains and is fairly diverged from the others.
FIG. 3. Rooted NJ tree of the kaiB genes and the KaiB-like domains of the sasA genes. Bootstrap values <50% are not shown. For the designations of the genes see table 1
Natural Selection in Evolution of sasA Genes
We tested whether the sasA genes experienced positive selection during their evolution. First we compared two models: the one-ratio model, which assumes the same parameter for the entire tree (1 = –9976.02), and the free-ratio model, which assumes a different parameter for each branch in the tree, (0 = –11108.85). Since the free-ratio model employs 18 parameters for 18 branches, while the one-ratio model assumes one, the LRT can be performed with 2 distribution and df = 17. The test yielded 2 = 2(1 – 0) = 2265.66 at P < 0.0001 suggesting rejection the free-ratio model and, consequently, one dN/dS ratio for all lineages. For further comparison, three pairs of models forming three LRTs were used: M0 (one ratio) and M3 (discrete), M1 (neutral) and M2 (selection), M7 (beta) and M8 (beta&) (Nielsen and Yang 1998; Yang et al. 2000). The first of the three comparisons is, in fact, a test for variable among sites. It gave significantly better results (2 = 416.0, P < 0.0001 with df = 4) for M3 (discrete) with K = 3 site classes. Two other comparisons yielded significantly better estimates for the models assuming positive selection, M2 and M8, respectively (table 3). For M1-M2 comparison, 2 = 30.02, P < 0.0001 with df = 2, and for M7-M8, 2 = 7.36, P = 0.025 with df = 2. M2 and M8 models produced largely consistent results regarding sites, which are probably under positive selection. Both models indicated the same five sites, 196C, 253S, 264G, 322R, and 340S to be likely under diversifying selection (table 3). On the other hand, the data of this analysis are in favor that vast majority of the sites are under strong purifying selection.
Table 3 Parameter Estimates and Log-Likelihood Values under Different Models of Variable Ratios Among Sites.
Functional Divergence of sasA Genes
The coefficient of functional divergence between clades A and B of the sasA genes is = 0.24 ± 0.10, implying that the altered functional constraint (or altered evolutionary rate) between them is significant (2 = 6.02, P = 0.014 at df = 1). However, the baseline of the site-specific profile estimated by the posterior probability P(S1|X) (Gu 1999) is 0.15–0.25 (fig. 4). Only six sites (1.4% of total sites) have P(S1|X) > 0.5. Interestingly, all these sites are conserved in clade A but are highly variable in clade B. The fact that most sites have the posterior probability <50% is in favor of their functional similarity between the two clades of the sasA genes.
FIG. 4. The site-specific profile for prediction of critical amino acid residues responsible for the Type I functional divergence between A and B clades of the sasA genes, measured by the posterior probability of being functionally divergent at each site
Discussion
SasA Structure and Polymorphism in Terms of the Functional Assignment of the Gene
Along with the three Kai proteins, the SasA protein was shown to be an essential part of the cyanobacterial circadian system (Iwasaki et al. 2000). It forms heteromultimeric complexes with KaiC in a circadian manner (Kageyama, Kondo, and Iwasaki 2003). The KaiB-like domain plays a central role in the SasA-KaiC interaction and maintenance of robust circadian oscillation in cyanobacteria (Iwasaki et al. 2000). In this interaction, the KaiB-like and HisKA domains were shown to take part. The former may interact with either of the two kaiC domains, CI or CII, similarly to KaiB, whereas the latter affect kaiBC promoter activity through autophosphorylation (Iwasaki et al. 2000; Kageyama, Kondo, and Iwasaki 2003).
The new functional assignment of the sasA gene was likely among the major factors, which shaped the observed patterns of its structure and polymorphism. Nuclear magnetic resonance (NMR) analysis of the first 105 amino acid residues of SasA, which cover the whole KaiB-like domain, indicates a secondary structure of ?a?aa (break), and preliminary structures suggest a thioredoxin-like fold (Klewer et al. 2002). Given that KaiB and KaiB-like domain of SasA interact with KaiC (Iwasaki et al. 2000), the six amino acid residues, which are highly conserved in both, are likely critical for maintaining the protein structure necessary for this physical association.
Origin and Evolution of sasA Genes and Implications into Development of the Circadian System in Cyanobacteria
The sasA gene originated in cyanobacteria apparently through the fusion of a two-component histidine kinase and ancestral kaiB gene to form the currently observed triple-domain structure. There are a number of facts supporting this conclusion. First, no triple-domain homologs of the sasA genes with the KaiB-like sensor domain occur in other prokaryotes, even those possessing the kai genes laterally transferred from cyanobacteria (Dvornyk, Vinogradova, and Nevo 2003). Theoretically, it is possible that the laterally transferred kaiB gene might fuse with the two-component histidine kinase in the other prokaryotes (e.g., proteobacteria) to form sasA, which then was transferred back to cyanobacteria. However, this scenario seems unlikely, given that there has been no evidence so far for the sasA homologs in prokaryotes, other than cyanobacteria.
Second, the kaiB genes were shown to originate in cyanobacteria (Dvornyk, Vinogradova, and Nevo 2003). Along with the improbability of the above scenario, this is in favor of that the fusion of kaiB and a two-component histidine kinase to form sasA probably occurred in cyanobacteria. This assumption is further supported by recent data of a complete genome sequence of an early evolved cyanobacterium, Gloeobacter violaceus, indicating that this species lacks the kaiA, kaiB, and sasA genes (Nakamura et al. 2003) and thus suggesting all these genes appeared at the later stages of cyanobacterial evolution.
Third, the similarity of the sasA and kaiBC trees (fig. 2) suggests that the former have likely evolved as an essential component of the cyanobacterial circadian system, together with the kaiBC cluster. Given that both KaiB-like domain and KaiB physically interact with KaiC in a circadian fashion (Ditty, Williams, and Golden 2003), it implies that any evolutionary changes in the respective genes, which may somehow alter a protein structure, should be concordant to ensure a possibility of this interaction.
Cyanobacterial genomes may carry several copies of the kaiBC cluster (e.g., Synechocystis sp. PCC 6803, which has the two). However, as follows from the results of the comparative phylogenetic analysis (fig. 2), probably only one of these copies (Syneys1) is involved into regulation of circadian oscillation, whereas another (Synesys3) appears to be fairly diverged to hold this function.
Some incongruence between the kaiBC and sasA trees within each of the two clades may suggest lateral transfers of either kaiBC cluster or sasA gene within the clades. However, the transfers of the sasA genes, if they took place, likely did not occur between A and B clades (fig. 2). This fact implies that the genes of each clade have specific functional and selective constraints, which make their transfers between the clades maladaptive. Indeed, as seen in table 2, the sasA genes of clade B are much higher conserved than those of clade A (average dN = 0.249 ± 0.019 and 0.541 ± 0.030, respectively). The most significant differences were observed in the HisKA domain, giving dN = 0.716 ± 0.095 for clade A and dN = 0.197 ± 0.044 for clade B. The observed differences in the rates of nonsynonymous nucleotide substitutions between the clades may suggest that, although the sasA genes are functionally similar (fig. 4), their KaiB-like and HisKA domains have different functional importance. In fact, only a few critical amino acid residues are involved in the functional divergence that precludes lateral transfers of the sasA genes between the clades (fig. 4). Importantly, although the sasA genes of clade A are generally less conserved than those of clade B, evolutionary rates at all these critical sites shifted in favor of their conservation in clade A. These conserved sites may be related to emerging KaiA as an input circadian regulator and a resultant subsequent change of the SasA function to mediation of mainly circadian output.
An intriguing question is: when did the fusion of kaiB and the histidine kinase occur and sasA gene appear? As the phylogenetic tree in figure 3 suggests, predecessors of the kaiB genes experienced a number of duplications during the early stages of their evolution, and the fate of the duplicated genes was different. Comparison of this tree with the previously reported data on evolution of the kaiB genes and kaiBC cluster (Dvornyk, Vinogradova, and Nevo 2003) makes it possible to reconstruct evolutionary history of the sasA genes and estimate the time of their origin.
First, sasA appeared most likely before the origin of the kaiBC cluster. An immediate predecessor of the gene, which then became a KaiB-like domain of sasA, did not experience any duplication before the fusion. Otherwise, B4 lineage of the KaiB-like domains (fig. 3) would have related lineages that are not a case. This fact suggests that the fusion resulting in a triple-domain sasA likely occurred soon after the first duplication, whereas the lineage leading to the kaiB genes has experienced a few more duplications (fig. 3). Such a scenario may explain the larger divergence of the KaiB-like domain from the kaiB genes. However, based on the currently available data, it is hardly possible to determine exactly when the triple-domain sasA gene emerged. These considerations, together with the estimates for the time of the origin of the kaiB genes and appearance of the kaiBC cluster between 3,500 and 2,320 MYA (Dvornyk, Vinogradova, and Nevo 2003), let us assume that the sasA gene evolved about 3,000–2,500 MYA. This is only a rough approximation, which is essentially based on the results of another molecular clock analysis. Given the observed heterogeneity in the rates of evolution, the estimates of the time of the sasA origin may carry considerable error.
So far, only the KaiABC-SasA-based circadian system, which includes the kaiA, kaiBC cluster and sasA, has been comprehensively studied in cyanobacteria using S. elongatus PCC 7942 as a model species (Ditty, Williams, and Golden 2003). Based on the results of the phylogenetic analysis of the kai genes, we recently hypothesized that a simpler system without kaiA may also exist (Dvornyk, Vinogradova, and Nevo 2003). The present study provides evidence for this hypothesis. Such a system is probably characteristic to cyanobacteria evolutionary older than genus Synechococcus, e.g., Prochlorococcus.
The obtained results give important implications for evolution of the cyanobacterial circadian system. Apparently, an ancestral circadian system consisted of the kaiB and kaiC genes, which were not in a cluster. The sasA gene supposedly evolved as a universal input-output regulator (Iwasaki et al. 2000) enhancing performance of this system. Later, kaiB and kaiC fused in the kaiBC cluster and sasA likely became an indispensable part of this KaiBC–SasA-based circadian system.
The next significant step in the evolution of the prokaryotic circadian system was emergence of the kaiA gene and, respectively, formation of the KaiABC-SasA system. In this system, the KaiA protein acts as an input circadian regulator of KaiC autophosphorylation (Williams et al. 2002), and SasA primarily controls a clock output and regulates downstream clock-controlled processes (Iwasaki et al. 2000). The appearance of kaiA probably conferred some selective advantage under certain conditions, because the KaiABC-SasA system remains functioning even after disruption of one of its components, the sasA gene (Iwasaki et al. 2000). The appearance of kaiA probably altered the evolutionary constraints and resulted in the observed functional divergence between the sasA genes of clades A and B as well as between respective kaiBC clusters (figs. 2 and 4).
Using the results obtained in this study and the previously reported data on evolution of the kai genes (Dvornyk, Vinogradova, and Nevo 2003), we can considerably update the evolutionary scenario of the cyanobacterial circadian system as presented in figure 5.
FIG. 5. Timeline of major events in evolution of the cyanobacterial circadian system. The time scale is not proportional. Extinct genes are designated as dashed boxes. Unclustered kaiB and kaiC genes are shaded. The elements of the KaiBC–SasA- and KaiABC–SasA-based circadian systems are connected by dashed arrows. pkaiC = predecessor of the kaiC gene, sdkaiC = single-domain kaiC, ddkaiC = double-domain kaiC, pkaiB = predecessor of the kaiB gene, TCHK = two-component sensory transduction histidine kinase
The fact that the sasA genes occur only in cyanobacteria and do not in other prokaryotes, even those possessing laterally transferred kai genes and kaiBC cluster, brings about one more important conclusion: these kai genes have probably undergone a functional alteration and are not circadian by their function anymore. Similarly, both the kaiB and kaiC genes scattered in the cyanobacterial genomes and extra copies of kaiBC not residing in the kaiABC cluster (e.g., Syneys3 in Synechocystis sp. PCC 6803) have probably significantly diverged from their original circadian function. However, this assumption is based solely on the results of the current phylogenetic analysis and needs to be verified experimentally.
Recently, it was shown that under permanent UV stress the kaiBC cluster experiences frequent duplications of adaptive significance and has very high mutation rate (Dvornyk, Vinogradova, and Nevo 2002; Dvornyk and Nevo 2003). However, it remains unknown how the other components of the circadian system respond to the stress. Are the sasA and kaiA genes duplicated as well? Do they have the same mutation rate?
Origin and evolution of the sasA genes provide an example of "fine evolutionary tuning" of the cyanobacterial circadian system. The evolutionary evidence for existence of the KaiBC-SasA-based system in cyanobacteria gives researchers a few intriguing questions to be answered. What happened to cyanobacteria 1,000 MYA to make the kaiA gene appear? How does the circadian system without this gene work? Answering these and other questions will give further insights into the mechanisms of evolutionary and physiological processes, which make cyanobacteria such highly successful generalists.
Acknowledgements
We thank Dr. Xun Gu (University of Iowa) and Prof. Outi Savolainen (University of Oulu, Finland) for their comments on the early draft of the manuscript. We are grateful to Dr. Laura Katz and two anonymous reviewers, whose suggestions helped to improve the article. This work was supported in part by the National Institutes of Health (grant R01 GM60402-01A1) and the State of Nebraska Cancer and Smoking Related Disease Research Program.
Literature Cited
Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang, W. Miller, and D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucl. Acids Res. 25:3389-3402.
Ditty, J. L., S. B. Williams, and S. S. Golden. 2003. A cyanobacterial circadian timing mechanism. Annu. Rev. Genet. 37:513-543.
Dutta, R., and M. Inouye. 2000. GHKL, an emergent ATPase/kinase superfamily. Trends Biochem. Sci. 25:24-28.
Dutta, R., L. Qin, and M. Inouye. 1999. Histidine kinases: diversity of domain organization. Mol. Microbiol. 34:633-640.
Dvornyk, V., and E. Nevo. 2003. Genetic polymorphism of cyanobacteria under permanent natural stress: a lesson from the "Evolution Canyons.". Res. Microbiol. 154:79-84.
Dvornyk, V., O. N. Vinogradova, and E. Nevo. 2002. Long-term microclimatic stress causes rapid adaptive radiation of kaiABC clock gene family in a cyanobacterium, Nostoc linckia, from the "Evolution Canyons" I and II, Israel. Proc. Natl. Acad. Sci. USA 99:2082-2087.
Dvornyk, V., O. N. Vinogradova, and E. Nevo. 2003. Origin and evolution of circadian clock genes in prokaryotes. Proc. Natl. Acad. Sci. USA 100:2495-2500.
Gu, X. 1999. Statistical methods for testing functional divergence after gene duplication. Mol. Biol. Evol. 16:1664-1674.
Gu, X., and K. Vander Velden. 2002. DIVERGE: phylogeny-based analysis for functional-structural divergence of a protein family. Bioinformatics 18:500-501.
Ishiura, M., S. Kutsuna, S. Aoki, H. Iwasaki, C. R. Andersson, A. Tanabe, S. S. Golden, C. H. Johnson, and T. Kondo. 1998. Expression of a gene cluster kaiABC as a circadian feedback process in cyanobacteria. Science 281:1519-1523.
Iwasaki, H., T. Nishiwaki, Y. Kitayama, M. Nakajima, and T. Kondo. 2002. KaiA-stimulated KaiC phosphorylation in circadian timing loops in cyanobacteria. Proc. Natl. Acad. Sci. USA 99:15788-15793.
Iwasaki, H., S. B. Williams, Y. Kitayama, M. Ishiura, S. S. Golden, and T. Kondo. 2000. A kaiC-interacting sensory histidine kinase, SasA, necessary to sustain robust circadian oscillation in cyanobacteria. Cell 101:223-233.
Johnson, C. H., and S. S. Golden. 1999. Circadian programs in cyanobacteria: adaptiveness and mechanism. Annu. Rev. Microbiol. 53:389-409.
Johnson, C. H., S. S. Golden, and T. Kondo. 1998. Adaptive significance of circadian programs in cyanobacteria. Trends Microbiol. 6:407-410.
Kageyama, H., T. Kondo, and H. Iwasaki. 2003. Circadian formation of clock protein complexes by KaiA, KaiB, KaiC, and SasA in cyanobacteria. J. Biol. Chem. 278:2388-2395.
Katayama, M., T. Kondo, J. Xiong, and S. S. Golden. 2003. ldpA encodes an iron-sulfur protein involved in light-dependent modulation of the circadian period in the cyanobacterium Synechococcus elongatus PCC 7942. J. Bacteriol. 185:1415-1422.
Katayama, M., N. F. Tsinoremas, T. Kondo, and S. S. Golden. 1999. cpmA, a gene involved in an output pathway of the cyanobacterial circadian system. J. Bacteriol. 181:3516-3524.
Kitayama, Y., H. Iwasaki, T. Nishiwaki, and T. Kondo. 2003. KaiB functions as an attenuator of KaiC phosphorylation in the cyanobacterial circadian clock system. EMBO J. 22:2127-2134.
Klewer, D. A., S. B. Williams, S. S. Golden, and A. C. LiWang. 2002. Sequence-specific resonance assignments of the N-terminal, 105-residue KaiC-interacting domain of SasA, a protein necessary for a robust circadian rhythm in Synechococcus elongatus. J. Biomol. NMR 24:77-78.
Kondo, T., and M. Ishiura. 1999. The circadian clocks of plants and cyanobacteria. Trends Plant Sci. 4:171-176.
Kumar, S., K. Tamura, I. Jakobsen, and M. Nei. 2001. MEGA2: Molecular Evolutionary Genetics Analysis software. Arizona State University, Tempe.
Kutsuna, S., T. Kondo, S. Aoki, and M. Ishiura. 1998. A period-extender gene, pex, that extends the period of the circadian clock in the cyanobacterium Synechococcus sp. strain PCC 7942. J. Bacteriol. 180:2167-2174.
Lorne, J., J. Scheffer, A. Lee, M. Painter, and V. P. Miao. 2000. Genes controlling circadian rhythm are widely distributed in cyanobacteria. FEMS Microbiol. Lett. 189:129-133.
Nagaya, M., H. Aiba, and T. Mizuno. 1993. Cloning of a sensory-kinase-encoding gene that belongs to the two-component regulatory family from the cyanobacterium Synechococcus sp. PCC7942. Gene 131:119-124.
Nakamura, Y., T. Kaneko, and S. Sato, et al. (19 coauthors). 2003. Complete genome structure of Gloeobacter violaceus PCC 7421, a cyanobacterium that lacks thylakoids. DNA Res. 10:137-145.
Nei, M., and T. Gojobori. 1986. Simple methods for estimating the numbers of synonymous and nonsynonymous nucleotide substitutions. Mol. Biol. Evol. 3:418-426.
Nielsen, R., and Z. Yang. 1998. Likelihood models for detecting positively selected amino acid sites and applications to the HIV-1 envelope gene. Genetics 148:929-936.
Obermann, W. M., H. Sondermann, A. A. Russo, N. P. Pavletich, and F. U. Hartl. 1998. In vivo function of Hsp90 is dependent on ATP binding and ATP hydrolysis. J. Cell Biol. 143:901-910.
Ouyang, Y., C. R. Andersson, T. Kondo, S. S. Golden, and C. H. Johnson. 1998. Resonating circadian clocks enhance fitness in cyanobacteria. Proc. Natl. Acad. Sci. USA 95:8660-8664.
Pittendrigh, C. S. 1993. Temporal organization: reflections of a Darwinian clock-watcher. Annu. Rev. Physiol. 55:16-54.
Saitou, N., and M. Nei. 1987. The neighbor-joining method: A new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406-425.
Schmitz, O., M. Katayama, S. B. Williams, T. Kondo, and S. S. Golden. 2000. CikA, a bacteriophytochrome that resets the cyanobacterial circadian clock. Science 289:765-768.
Stock, A. M., V. L. Robinson, and P. N. Goudreau. 2000. Two-component signal transduction. Annu. Rev. Biochem. 69:183-215.
Sweeney, B. M. 1987. Rhythmic phenomena in plants. Academic Press, San Diego, Calif.
Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680.
Tsinoremas, N. F., M. Ishiura, T. Kondo, K. Tanaka, H. Takahashi, C. H. Johnson, and S. S. Golden. 1996. A sigma factor that modifies the circadian expression of a subset of genes in cyanobacteria. EMBO J. 15:2488-2495.
Williams, S. B., I. Vakonakis, S. S. Golden, and A. C. LiWang. 2002. Structure and function from the circadian clock protein KaiA of Synechococcus elongatus: A potential clock input mechanism. Proc. Natl. Acad. Sci. USA 99:15357-15362.
Xu, Y., T. Mori, and C. H. Johnson. 2003. Cyanobacterial circadian clockwork: roles of KaiA, KaiB and the kaiBC promoter in regulating KaiC. EMBO J. 22:2117-2126.
Yang, Z. 1997. PAML: a program package for phylogenetic analysis by maximum likelihood. CABIOS 15:555-556.
Yang, Z., R. Nielsen, N. Goldman, and A.-M. K. Pedersen. 2000. Codon-substitution models for heterogeneous selection pressure at amino acid sites. Genetics 155:431-449.(Volodymyr Dvornyk*, Hong-)
Institute of Evolution, University of Haifa, Mount Carmel, Haifa, Israel
E-mail: dvornyk@creighton.edu.
Abstract
Cyanobacteria are the simplest organisms known to have a circadian system. In addition to the three well-studied kai genes, kaiA, kaiB, and kaiC, an important element of this system is a two-component sensory transduction histidine kinase sasA. Using publicly available data of complete prokaryotic genomes, we performed structural and phylogenetic analyses of the sasA genes. Results show that this gene has a triple-domain structure, and the domains are under different selective constraints. The sasA gene originated in cyanobacteria probably through the fusion of the ancestral kaiB gene with a double-domain, two-component sensory transduction histidine kinase. The results of the phylogenetic analyses suggest that sasA emerged before the kaiA gene, about 3,000–2,500 MYA, and has evolved in parallel with the evolution of the kaiBC cluster. The observed concordant patterns of the sasA and kaiBC evolution suggest that these genes might compose an ancient KaiBC–SasA-based circadian system, without the kaiA gene, and that such a system still exists in some unicellular cyanobacteria.
Key Words: sasA ? circadian system ? prokaryotes ? evolution ? cyanobacteria
Introduction
Circadian rhythmicity in regulation of endogenous physiological processes is a characteristic feature of eukaryotic organisms (Sweeney 1987; Pittendrigh 1993). Cyanobacteria are the simplest organisms known to manifest circadian clock (Kondo and Ishiura 1999). A cluster encoding three genes, kaiA, kaiB, and kaiC, was identified in the unicellular cyanobacterium, Synechococcus elongatus PCC 7942, as a key component of the circadian system (Ishiura et al. 1998). The clock genes are ubiquitous in cyanobacteria (Lorne et al. 2000; Dvornyk, Vinogradova, and Nevo 2003). They have been shown to play an important role in high adaptiveness of these prokaryotes to various environmental conditions (Ouyang et al. 1998; Johnson, Golden, and Kondo 1998; Johnson and Golden 1999; Dvornyk, Vinogradova, and Nevo 2002).
Although the three kai genes are essential in maintaining circadian oscillation (Ishiura et al. 1998; Iwasaki et al. 2002; Kitayama et al. 2003; Xu, Mori, and Johnson 2003), a number of other genes have been determined to participate in the input and output pathways of the cyanobacterial circadian system (Tsinoremas et al. 1996; Kutsuna et al. 1998; Katayama et al. 1999; Schmitz et al. 2000; Katayama et al. 2003). One of the latter is sasA, a kaiC-interacting sensory histidine kinase, which is important for robust circadian rhythmicity (Iwasaki et al. 2000; Kageyama, Kondo, and Iwasaki 2003).
Recently we showed that the kai genes have quite a different evolutionary history and occur also in some archaea and proteobacteria. The kaiA and kaiB genes originated in cyanobacteria, and the three-gene cluster, kaiABC, evolved about 1,000 MYA (Dvornyk, Vinogradova, and Nevo 2003). However, since the cyanobacterial circadian system also includes genes other than kai, the question is raised: how has this system evolved?
In the present work, we comprehensively studied structure and sequence diversity of the sasA genes from cyanobacteria. We also tried to reconstruct their phylogeny and to compare it to those of the kaiB and kaiC genes in order to find out when and how the sasA genes became involved into the regulation of circadian rhythmicity in cyanobacteria.
Materials and Methods
DNA and Protein Sequences
The annotated and homologous sequences of the sasA, kaiB, and kaiC genes were retrieved from GenBank by using the gapped BLASTP and PSI-BLAST (Altschul et al. 1997) and the respective amino acid sequences of Synechococcus elongatus PCC 7942 (GenBank accession Q06904, BAA37102, and BAA37103) as the probes. The kai sequences of only those species, which have the sasA genes, were used in the analyses. The retrieved protein and nucleotide sequences were aligned using CLUSTALW (Thompson, Higgins, and Gibson 1994) and manually adjusted based on structural considerations (Ishiura et al. 1998; Iwasaki et al. 2000). Alignments of the nucleotide sequences were modified manually according to the respective amino acid alignments. These multiple alignments are available upon request. The list of the used sequences is given in table 1.
Table 1 Sequences and Species Used in the Study.
Phylogenetic Analysis
The rate of nonsynonymous nucleotide substitutions per nonsynonymous site (dN) was calculated using the modified Nei-Gojobori method (Nei and Gojobori 1986) with Juke-Cantor correction for multiple substitutions at the same site and the transitions/transversions ratio equal to 1.8. The MEGA 2.1 software (Kumar et al. 2001) was used for the computations of dN.
In the comparative phylogenetic analysis, we proceeded from the evidence that the kaiB and kaiC genes, after fusion into in the kaiBC cluster, have evolved as a unit rather than independently (Dvornyk, Vinogradova, and Nevo 2003). The phylogenetic trees for the sasA genes and kaiBC cluster were obtained using respective amino acid sequences and a neighbor-joining (NJ) algorithm (Saitou and Nei 1987) with Poisson correction as implemented in the MEGA 2.1 software (Kumar et al. 2001). The same method and software were used to perform a phylogenetic reconstruction for the kaiB genes and KaiB-like domain of the sasA genes. Statistical significance of the nodes in all derived trees was evaluated by the bootstrap procedure with 1,000 replications.
Analysis of Positive Selection
Comparing synonymous (silent, dS) and nonsynonymous (amino acid-changing, dN) substitution rates in protein-coding genes provides important information for understanding molecular evolution. The nonsynonymous/synonymous rate ratio, = dN/dS, measures selective pressure at the protein level. The values of = 1, <1, and >1, indicate neutral evolution, purifying selection, and positive selection, respectively. To test the sasA genes for the presence of natural selection, we used the site-specific models, which do not average dN/dS ratio over sequences but allow it to vary among amino acid sites (Nielsen and Yang 1998; Yang et al. 2000). The models were pairwise compared by the likelihood ratio test (LRT). This was done by comparing the log-likelihood values with = 1 constrained and without such constraint. If the null hypothesis = 1 is correct, twice the log-likelihood difference between the two models (2) asymptotically has a 2 distribution with df = 1. All computations were performed by PAML software (Yang 1997).
Analysis of Functional Divergence
There are two main types of functional divergence. Type I functional divergence after gene duplication results in altered functional constraints (i.e., different evolutionary rate) between duplicate genes, whereas type II results in no altered functional constraints but radical change in amino acid property between them (e.g., charge, hydrophobicity, etc.) (Gu 1999).
The type I functional divergence between the different clades of the sasA gene tree was analyzed with the method proposed by Gu (Gu 1999). We estimated a rate correlation between the members of the different clades, or a coefficient of functional divergence, , as implemented in DIVERGE v.1.04 software (Gu and Vander Velden 2002). Using this method, we also identified critical amino acid residues that may be responsible for the observed functional divergence.
Results
Structure and Sequence Polymorphism of sasA Genes in Cyanobacteria
Screening available prokaryotic genomes using BLAST search revealed the large number of sasA homologs in many prokaryotic taxa. This was expected, because sasA belongs to the large superfamily of two-component sensory transduction histidine kinases widely distributed in prokaryotes (Nagaya, Aiba, and Mizuno 1993). Genes of this superfamily often have a three-domain structure (Dutta, Qin, and Inouye 1999). One of the domains is histidine kinase A phosphoacceptor domain (HisKA), another is related to the family of histidine kinase-like ATPases (HATPase_c). The former is an element of the two-component signaling systems, consisting of a histidine protein kinase that senses a signal input and a response regulator that mediates the output (Stock, Robinson, and Goudreau 2000). The latter belongs to the ATP binding superfamily that includes diverse protein families such as DNA topoisomerase II, molecular chaperones Hsp90, DNA-mismatch-repair enzymes, phytochrome-like ATPases, and histidine kinases (Dutta and Inouye 2000). The third domain is a sensor domain (Dutta, Qin, and Inouye 1999).
In contrast to the other two-component sensory transduction histidine kinases, the sensor domain of the sasA kinases is homologous to the kaiB genes (Iwasaki et al. 2000) (fig. 1A). Importantly, the BLAST search did not reveal the sasA homologs with the KaiB-like domain in genomes of prokaryotes, other than cyanobacteria, including those that possess the kaiBC cluster transferred from cyanobacteria (Dvornyk, Vinogradova, and Nevo 2003).
FIG. 1. Amino acid sequence alignment of the three domains of the sasA genes. Black- and gray-shaded backgrounds indicate different degree of conservation (black is the most conserved). The upper numbers indicate positions in the alignment of the full sequences; the numbers on the right indicate positions in the respective sequences. (A) KaiB-like domain. Putative hydrophobic cores are underlined. Amino acid residues, whose mutations in KaiB alter circadian periods, are marked by arrows. (B) HisKA domain. Putative autophosphorylation histidine residue is marked by block arrow. (C) HATPase_c domain. Putative ATP binding sites and Mg2+ binding site are marked with "#" and "+," respectively. G-X-G-motifs are marked with square brackets
The structure of the KaiB-like domain of the SasA protein is shown in figure 1A. It is 86 amino acid residues long (corresponds to residues 14–99 in the S. elongatus PCC 7942 SasA sequence) and has a number of putative motifs and highly conserved residues. The first motif (positions 36–42 in the alignment, fig. 1A) is six or seven residues long, and at least five are hydrophobic. This motif is one of two putative hydrophobic cores of this domain. Another corresponds to the positions 87–97 of the alignment and contains at least eight hydrophobic residues (fig. 1A). Comparing these motifs against public databases revealed no homologs. Aligning the KaiB-like domain of SasA with the KaiB proteins (the alignment not shown) indicated that both putative hydrophobic cores in KaiB are also highly conserved, although they differ by sequence from those of the KaiB-like domain. Motifs of the KaiB-like domain contain conserved amino acid residues, whose mutations in KaiB alter circadian periods (Iwasaki et al. 2000). Those are L40 and R/K102 in the alignment (fig. 1A). In addition, several other residues, which are fully conserved in the KaiB-like domain and KaiB, can be identified. In figure 1A, those correspond to three proline residues (positions 80, 92, and 99), L94, and G107. Interestingly, all these conserved residues are nonpolar.
The HisKA and HATPase_c domains are 59–71 and 111–130 amino acid residues long, respectively (fig. 1B, C). They have a number of conserved characteristic features. The HisKA domain contains putative autophosphorylation histidine residue (Iwasaki et al. 2000) (fig. 1B); the HATPase_c domain has several ATP binding sites, an Mg2+ binding site, and two G-X-G motifs (fig. 1C). The motifs are located in loops defining the top and bottom of the ATP binding pocket (Obermann et al. 1998).
The three domains of the sasA genes differ by the number of accumulated nonsynonymous substitutions. As the data in table 2 suggest, the HATPase_c domain is the most conserved among the three (dN = 0.412 ± 0.046), the KaiB-like domain has dN = 0.614 ± 0.068, whereas the HisKA domain is the least conserved (dN = 0.840 ± 0.112). The regions not belonging to any of the three domains have dN = 0.836 ± 0.077.
Table 2 Patterns of Nonsynonymous Nucleotide Substitutions (dN) in the Different Domains of the sasA Genes.
Comparative Phylogeny of sasA and kai Genes in Cyanobacteria
All sasA genes found in cyanobacteria occurred in a single copy. The resulting consensus tree shows that the sasA genes are separated into two clades (fig. 2). Clade A comprises sasA from the species, which have all three kai genes. Another clade includes the sasA genes from cyanobacteria, which possess either all the three clock oscillators (Synechococcus sp. WH 8102) or only two of them, kaiB and kaiC (Prochlorococcus strains). The trees of the kaiBC cluster and sasA genes from the same species have similar topologies (fig. 2). In particular, two main clades of each tree, A and B, consist of the same species. Importantly, of two kaiBC clusters of Synechocystis sp. PCC 6803, one that is not in the cluster with the kaiA gene (Syncys3), forms a branch discordant with the tree topology of the sasA genes.
FIG. 2. Concordant NJ phylogenetic trees of the sasA genes and kaiBC cluster from cyanobacteria. Bootstrap values <50% are not shown. The kaiBC sequences are denoted according to the respective kaiC sequences (table 1)
The sasA genes from the different branches indicate different patterns of nonsynonymous nucleotide substitutions in their domains (table 2). The sasA genes of clade A have significantly higher dN values (P 0.05) in all the three domains than do the genes of clade B.
The phylogeny of the kaiB genes and KaiB-like domains of the sasA genes from the studied species show their clear separation into four clades, all with very high bootstrap support for each clade (fig. 3). Three clades (B1, B2, and B3) correspond to those previously reported for the kaiB genes (Dvornyk, Vinogradova, and Nevo 2003). The fourth clade, B4, comprises the KaiB-like domains and is fairly diverged from the others.
FIG. 3. Rooted NJ tree of the kaiB genes and the KaiB-like domains of the sasA genes. Bootstrap values <50% are not shown. For the designations of the genes see table 1
Natural Selection in Evolution of sasA Genes
We tested whether the sasA genes experienced positive selection during their evolution. First we compared two models: the one-ratio model, which assumes the same parameter for the entire tree (1 = –9976.02), and the free-ratio model, which assumes a different parameter for each branch in the tree, (0 = –11108.85). Since the free-ratio model employs 18 parameters for 18 branches, while the one-ratio model assumes one, the LRT can be performed with 2 distribution and df = 17. The test yielded 2 = 2(1 – 0) = 2265.66 at P < 0.0001 suggesting rejection the free-ratio model and, consequently, one dN/dS ratio for all lineages. For further comparison, three pairs of models forming three LRTs were used: M0 (one ratio) and M3 (discrete), M1 (neutral) and M2 (selection), M7 (beta) and M8 (beta&) (Nielsen and Yang 1998; Yang et al. 2000). The first of the three comparisons is, in fact, a test for variable among sites. It gave significantly better results (2 = 416.0, P < 0.0001 with df = 4) for M3 (discrete) with K = 3 site classes. Two other comparisons yielded significantly better estimates for the models assuming positive selection, M2 and M8, respectively (table 3). For M1-M2 comparison, 2 = 30.02, P < 0.0001 with df = 2, and for M7-M8, 2 = 7.36, P = 0.025 with df = 2. M2 and M8 models produced largely consistent results regarding sites, which are probably under positive selection. Both models indicated the same five sites, 196C, 253S, 264G, 322R, and 340S to be likely under diversifying selection (table 3). On the other hand, the data of this analysis are in favor that vast majority of the sites are under strong purifying selection.
Table 3 Parameter Estimates and Log-Likelihood Values under Different Models of Variable Ratios Among Sites.
Functional Divergence of sasA Genes
The coefficient of functional divergence between clades A and B of the sasA genes is = 0.24 ± 0.10, implying that the altered functional constraint (or altered evolutionary rate) between them is significant (2 = 6.02, P = 0.014 at df = 1). However, the baseline of the site-specific profile estimated by the posterior probability P(S1|X) (Gu 1999) is 0.15–0.25 (fig. 4). Only six sites (1.4% of total sites) have P(S1|X) > 0.5. Interestingly, all these sites are conserved in clade A but are highly variable in clade B. The fact that most sites have the posterior probability <50% is in favor of their functional similarity between the two clades of the sasA genes.
FIG. 4. The site-specific profile for prediction of critical amino acid residues responsible for the Type I functional divergence between A and B clades of the sasA genes, measured by the posterior probability of being functionally divergent at each site
Discussion
SasA Structure and Polymorphism in Terms of the Functional Assignment of the Gene
Along with the three Kai proteins, the SasA protein was shown to be an essential part of the cyanobacterial circadian system (Iwasaki et al. 2000). It forms heteromultimeric complexes with KaiC in a circadian manner (Kageyama, Kondo, and Iwasaki 2003). The KaiB-like domain plays a central role in the SasA-KaiC interaction and maintenance of robust circadian oscillation in cyanobacteria (Iwasaki et al. 2000). In this interaction, the KaiB-like and HisKA domains were shown to take part. The former may interact with either of the two kaiC domains, CI or CII, similarly to KaiB, whereas the latter affect kaiBC promoter activity through autophosphorylation (Iwasaki et al. 2000; Kageyama, Kondo, and Iwasaki 2003).
The new functional assignment of the sasA gene was likely among the major factors, which shaped the observed patterns of its structure and polymorphism. Nuclear magnetic resonance (NMR) analysis of the first 105 amino acid residues of SasA, which cover the whole KaiB-like domain, indicates a secondary structure of ?a?aa (break), and preliminary structures suggest a thioredoxin-like fold (Klewer et al. 2002). Given that KaiB and KaiB-like domain of SasA interact with KaiC (Iwasaki et al. 2000), the six amino acid residues, which are highly conserved in both, are likely critical for maintaining the protein structure necessary for this physical association.
Origin and Evolution of sasA Genes and Implications into Development of the Circadian System in Cyanobacteria
The sasA gene originated in cyanobacteria apparently through the fusion of a two-component histidine kinase and ancestral kaiB gene to form the currently observed triple-domain structure. There are a number of facts supporting this conclusion. First, no triple-domain homologs of the sasA genes with the KaiB-like sensor domain occur in other prokaryotes, even those possessing the kai genes laterally transferred from cyanobacteria (Dvornyk, Vinogradova, and Nevo 2003). Theoretically, it is possible that the laterally transferred kaiB gene might fuse with the two-component histidine kinase in the other prokaryotes (e.g., proteobacteria) to form sasA, which then was transferred back to cyanobacteria. However, this scenario seems unlikely, given that there has been no evidence so far for the sasA homologs in prokaryotes, other than cyanobacteria.
Second, the kaiB genes were shown to originate in cyanobacteria (Dvornyk, Vinogradova, and Nevo 2003). Along with the improbability of the above scenario, this is in favor of that the fusion of kaiB and a two-component histidine kinase to form sasA probably occurred in cyanobacteria. This assumption is further supported by recent data of a complete genome sequence of an early evolved cyanobacterium, Gloeobacter violaceus, indicating that this species lacks the kaiA, kaiB, and sasA genes (Nakamura et al. 2003) and thus suggesting all these genes appeared at the later stages of cyanobacterial evolution.
Third, the similarity of the sasA and kaiBC trees (fig. 2) suggests that the former have likely evolved as an essential component of the cyanobacterial circadian system, together with the kaiBC cluster. Given that both KaiB-like domain and KaiB physically interact with KaiC in a circadian fashion (Ditty, Williams, and Golden 2003), it implies that any evolutionary changes in the respective genes, which may somehow alter a protein structure, should be concordant to ensure a possibility of this interaction.
Cyanobacterial genomes may carry several copies of the kaiBC cluster (e.g., Synechocystis sp. PCC 6803, which has the two). However, as follows from the results of the comparative phylogenetic analysis (fig. 2), probably only one of these copies (Syneys1) is involved into regulation of circadian oscillation, whereas another (Synesys3) appears to be fairly diverged to hold this function.
Some incongruence between the kaiBC and sasA trees within each of the two clades may suggest lateral transfers of either kaiBC cluster or sasA gene within the clades. However, the transfers of the sasA genes, if they took place, likely did not occur between A and B clades (fig. 2). This fact implies that the genes of each clade have specific functional and selective constraints, which make their transfers between the clades maladaptive. Indeed, as seen in table 2, the sasA genes of clade B are much higher conserved than those of clade A (average dN = 0.249 ± 0.019 and 0.541 ± 0.030, respectively). The most significant differences were observed in the HisKA domain, giving dN = 0.716 ± 0.095 for clade A and dN = 0.197 ± 0.044 for clade B. The observed differences in the rates of nonsynonymous nucleotide substitutions between the clades may suggest that, although the sasA genes are functionally similar (fig. 4), their KaiB-like and HisKA domains have different functional importance. In fact, only a few critical amino acid residues are involved in the functional divergence that precludes lateral transfers of the sasA genes between the clades (fig. 4). Importantly, although the sasA genes of clade A are generally less conserved than those of clade B, evolutionary rates at all these critical sites shifted in favor of their conservation in clade A. These conserved sites may be related to emerging KaiA as an input circadian regulator and a resultant subsequent change of the SasA function to mediation of mainly circadian output.
An intriguing question is: when did the fusion of kaiB and the histidine kinase occur and sasA gene appear? As the phylogenetic tree in figure 3 suggests, predecessors of the kaiB genes experienced a number of duplications during the early stages of their evolution, and the fate of the duplicated genes was different. Comparison of this tree with the previously reported data on evolution of the kaiB genes and kaiBC cluster (Dvornyk, Vinogradova, and Nevo 2003) makes it possible to reconstruct evolutionary history of the sasA genes and estimate the time of their origin.
First, sasA appeared most likely before the origin of the kaiBC cluster. An immediate predecessor of the gene, which then became a KaiB-like domain of sasA, did not experience any duplication before the fusion. Otherwise, B4 lineage of the KaiB-like domains (fig. 3) would have related lineages that are not a case. This fact suggests that the fusion resulting in a triple-domain sasA likely occurred soon after the first duplication, whereas the lineage leading to the kaiB genes has experienced a few more duplications (fig. 3). Such a scenario may explain the larger divergence of the KaiB-like domain from the kaiB genes. However, based on the currently available data, it is hardly possible to determine exactly when the triple-domain sasA gene emerged. These considerations, together with the estimates for the time of the origin of the kaiB genes and appearance of the kaiBC cluster between 3,500 and 2,320 MYA (Dvornyk, Vinogradova, and Nevo 2003), let us assume that the sasA gene evolved about 3,000–2,500 MYA. This is only a rough approximation, which is essentially based on the results of another molecular clock analysis. Given the observed heterogeneity in the rates of evolution, the estimates of the time of the sasA origin may carry considerable error.
So far, only the KaiABC-SasA-based circadian system, which includes the kaiA, kaiBC cluster and sasA, has been comprehensively studied in cyanobacteria using S. elongatus PCC 7942 as a model species (Ditty, Williams, and Golden 2003). Based on the results of the phylogenetic analysis of the kai genes, we recently hypothesized that a simpler system without kaiA may also exist (Dvornyk, Vinogradova, and Nevo 2003). The present study provides evidence for this hypothesis. Such a system is probably characteristic to cyanobacteria evolutionary older than genus Synechococcus, e.g., Prochlorococcus.
The obtained results give important implications for evolution of the cyanobacterial circadian system. Apparently, an ancestral circadian system consisted of the kaiB and kaiC genes, which were not in a cluster. The sasA gene supposedly evolved as a universal input-output regulator (Iwasaki et al. 2000) enhancing performance of this system. Later, kaiB and kaiC fused in the kaiBC cluster and sasA likely became an indispensable part of this KaiBC–SasA-based circadian system.
The next significant step in the evolution of the prokaryotic circadian system was emergence of the kaiA gene and, respectively, formation of the KaiABC-SasA system. In this system, the KaiA protein acts as an input circadian regulator of KaiC autophosphorylation (Williams et al. 2002), and SasA primarily controls a clock output and regulates downstream clock-controlled processes (Iwasaki et al. 2000). The appearance of kaiA probably conferred some selective advantage under certain conditions, because the KaiABC-SasA system remains functioning even after disruption of one of its components, the sasA gene (Iwasaki et al. 2000). The appearance of kaiA probably altered the evolutionary constraints and resulted in the observed functional divergence between the sasA genes of clades A and B as well as between respective kaiBC clusters (figs. 2 and 4).
Using the results obtained in this study and the previously reported data on evolution of the kai genes (Dvornyk, Vinogradova, and Nevo 2003), we can considerably update the evolutionary scenario of the cyanobacterial circadian system as presented in figure 5.
FIG. 5. Timeline of major events in evolution of the cyanobacterial circadian system. The time scale is not proportional. Extinct genes are designated as dashed boxes. Unclustered kaiB and kaiC genes are shaded. The elements of the KaiBC–SasA- and KaiABC–SasA-based circadian systems are connected by dashed arrows. pkaiC = predecessor of the kaiC gene, sdkaiC = single-domain kaiC, ddkaiC = double-domain kaiC, pkaiB = predecessor of the kaiB gene, TCHK = two-component sensory transduction histidine kinase
The fact that the sasA genes occur only in cyanobacteria and do not in other prokaryotes, even those possessing laterally transferred kai genes and kaiBC cluster, brings about one more important conclusion: these kai genes have probably undergone a functional alteration and are not circadian by their function anymore. Similarly, both the kaiB and kaiC genes scattered in the cyanobacterial genomes and extra copies of kaiBC not residing in the kaiABC cluster (e.g., Syneys3 in Synechocystis sp. PCC 6803) have probably significantly diverged from their original circadian function. However, this assumption is based solely on the results of the current phylogenetic analysis and needs to be verified experimentally.
Recently, it was shown that under permanent UV stress the kaiBC cluster experiences frequent duplications of adaptive significance and has very high mutation rate (Dvornyk, Vinogradova, and Nevo 2002; Dvornyk and Nevo 2003). However, it remains unknown how the other components of the circadian system respond to the stress. Are the sasA and kaiA genes duplicated as well? Do they have the same mutation rate?
Origin and evolution of the sasA genes provide an example of "fine evolutionary tuning" of the cyanobacterial circadian system. The evolutionary evidence for existence of the KaiBC-SasA-based system in cyanobacteria gives researchers a few intriguing questions to be answered. What happened to cyanobacteria 1,000 MYA to make the kaiA gene appear? How does the circadian system without this gene work? Answering these and other questions will give further insights into the mechanisms of evolutionary and physiological processes, which make cyanobacteria such highly successful generalists.
Acknowledgements
We thank Dr. Xun Gu (University of Iowa) and Prof. Outi Savolainen (University of Oulu, Finland) for their comments on the early draft of the manuscript. We are grateful to Dr. Laura Katz and two anonymous reviewers, whose suggestions helped to improve the article. This work was supported in part by the National Institutes of Health (grant R01 GM60402-01A1) and the State of Nebraska Cancer and Smoking Related Disease Research Program.
Literature Cited
Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang, W. Miller, and D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucl. Acids Res. 25:3389-3402.
Ditty, J. L., S. B. Williams, and S. S. Golden. 2003. A cyanobacterial circadian timing mechanism. Annu. Rev. Genet. 37:513-543.
Dutta, R., and M. Inouye. 2000. GHKL, an emergent ATPase/kinase superfamily. Trends Biochem. Sci. 25:24-28.
Dutta, R., L. Qin, and M. Inouye. 1999. Histidine kinases: diversity of domain organization. Mol. Microbiol. 34:633-640.
Dvornyk, V., and E. Nevo. 2003. Genetic polymorphism of cyanobacteria under permanent natural stress: a lesson from the "Evolution Canyons.". Res. Microbiol. 154:79-84.
Dvornyk, V., O. N. Vinogradova, and E. Nevo. 2002. Long-term microclimatic stress causes rapid adaptive radiation of kaiABC clock gene family in a cyanobacterium, Nostoc linckia, from the "Evolution Canyons" I and II, Israel. Proc. Natl. Acad. Sci. USA 99:2082-2087.
Dvornyk, V., O. N. Vinogradova, and E. Nevo. 2003. Origin and evolution of circadian clock genes in prokaryotes. Proc. Natl. Acad. Sci. USA 100:2495-2500.
Gu, X. 1999. Statistical methods for testing functional divergence after gene duplication. Mol. Biol. Evol. 16:1664-1674.
Gu, X., and K. Vander Velden. 2002. DIVERGE: phylogeny-based analysis for functional-structural divergence of a protein family. Bioinformatics 18:500-501.
Ishiura, M., S. Kutsuna, S. Aoki, H. Iwasaki, C. R. Andersson, A. Tanabe, S. S. Golden, C. H. Johnson, and T. Kondo. 1998. Expression of a gene cluster kaiABC as a circadian feedback process in cyanobacteria. Science 281:1519-1523.
Iwasaki, H., T. Nishiwaki, Y. Kitayama, M. Nakajima, and T. Kondo. 2002. KaiA-stimulated KaiC phosphorylation in circadian timing loops in cyanobacteria. Proc. Natl. Acad. Sci. USA 99:15788-15793.
Iwasaki, H., S. B. Williams, Y. Kitayama, M. Ishiura, S. S. Golden, and T. Kondo. 2000. A kaiC-interacting sensory histidine kinase, SasA, necessary to sustain robust circadian oscillation in cyanobacteria. Cell 101:223-233.
Johnson, C. H., and S. S. Golden. 1999. Circadian programs in cyanobacteria: adaptiveness and mechanism. Annu. Rev. Microbiol. 53:389-409.
Johnson, C. H., S. S. Golden, and T. Kondo. 1998. Adaptive significance of circadian programs in cyanobacteria. Trends Microbiol. 6:407-410.
Kageyama, H., T. Kondo, and H. Iwasaki. 2003. Circadian formation of clock protein complexes by KaiA, KaiB, KaiC, and SasA in cyanobacteria. J. Biol. Chem. 278:2388-2395.
Katayama, M., T. Kondo, J. Xiong, and S. S. Golden. 2003. ldpA encodes an iron-sulfur protein involved in light-dependent modulation of the circadian period in the cyanobacterium Synechococcus elongatus PCC 7942. J. Bacteriol. 185:1415-1422.
Katayama, M., N. F. Tsinoremas, T. Kondo, and S. S. Golden. 1999. cpmA, a gene involved in an output pathway of the cyanobacterial circadian system. J. Bacteriol. 181:3516-3524.
Kitayama, Y., H. Iwasaki, T. Nishiwaki, and T. Kondo. 2003. KaiB functions as an attenuator of KaiC phosphorylation in the cyanobacterial circadian clock system. EMBO J. 22:2127-2134.
Klewer, D. A., S. B. Williams, S. S. Golden, and A. C. LiWang. 2002. Sequence-specific resonance assignments of the N-terminal, 105-residue KaiC-interacting domain of SasA, a protein necessary for a robust circadian rhythm in Synechococcus elongatus. J. Biomol. NMR 24:77-78.
Kondo, T., and M. Ishiura. 1999. The circadian clocks of plants and cyanobacteria. Trends Plant Sci. 4:171-176.
Kumar, S., K. Tamura, I. Jakobsen, and M. Nei. 2001. MEGA2: Molecular Evolutionary Genetics Analysis software. Arizona State University, Tempe.
Kutsuna, S., T. Kondo, S. Aoki, and M. Ishiura. 1998. A period-extender gene, pex, that extends the period of the circadian clock in the cyanobacterium Synechococcus sp. strain PCC 7942. J. Bacteriol. 180:2167-2174.
Lorne, J., J. Scheffer, A. Lee, M. Painter, and V. P. Miao. 2000. Genes controlling circadian rhythm are widely distributed in cyanobacteria. FEMS Microbiol. Lett. 189:129-133.
Nagaya, M., H. Aiba, and T. Mizuno. 1993. Cloning of a sensory-kinase-encoding gene that belongs to the two-component regulatory family from the cyanobacterium Synechococcus sp. PCC7942. Gene 131:119-124.
Nakamura, Y., T. Kaneko, and S. Sato, et al. (19 coauthors). 2003. Complete genome structure of Gloeobacter violaceus PCC 7421, a cyanobacterium that lacks thylakoids. DNA Res. 10:137-145.
Nei, M., and T. Gojobori. 1986. Simple methods for estimating the numbers of synonymous and nonsynonymous nucleotide substitutions. Mol. Biol. Evol. 3:418-426.
Nielsen, R., and Z. Yang. 1998. Likelihood models for detecting positively selected amino acid sites and applications to the HIV-1 envelope gene. Genetics 148:929-936.
Obermann, W. M., H. Sondermann, A. A. Russo, N. P. Pavletich, and F. U. Hartl. 1998. In vivo function of Hsp90 is dependent on ATP binding and ATP hydrolysis. J. Cell Biol. 143:901-910.
Ouyang, Y., C. R. Andersson, T. Kondo, S. S. Golden, and C. H. Johnson. 1998. Resonating circadian clocks enhance fitness in cyanobacteria. Proc. Natl. Acad. Sci. USA 95:8660-8664.
Pittendrigh, C. S. 1993. Temporal organization: reflections of a Darwinian clock-watcher. Annu. Rev. Physiol. 55:16-54.
Saitou, N., and M. Nei. 1987. The neighbor-joining method: A new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406-425.
Schmitz, O., M. Katayama, S. B. Williams, T. Kondo, and S. S. Golden. 2000. CikA, a bacteriophytochrome that resets the cyanobacterial circadian clock. Science 289:765-768.
Stock, A. M., V. L. Robinson, and P. N. Goudreau. 2000. Two-component signal transduction. Annu. Rev. Biochem. 69:183-215.
Sweeney, B. M. 1987. Rhythmic phenomena in plants. Academic Press, San Diego, Calif.
Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680.
Tsinoremas, N. F., M. Ishiura, T. Kondo, K. Tanaka, H. Takahashi, C. H. Johnson, and S. S. Golden. 1996. A sigma factor that modifies the circadian expression of a subset of genes in cyanobacteria. EMBO J. 15:2488-2495.
Williams, S. B., I. Vakonakis, S. S. Golden, and A. C. LiWang. 2002. Structure and function from the circadian clock protein KaiA of Synechococcus elongatus: A potential clock input mechanism. Proc. Natl. Acad. Sci. USA 99:15357-15362.
Xu, Y., T. Mori, and C. H. Johnson. 2003. Cyanobacterial circadian clockwork: roles of KaiA, KaiB and the kaiBC promoter in regulating KaiC. EMBO J. 22:2117-2126.
Yang, Z. 1997. PAML: a program package for phylogenetic analysis by maximum likelihood. CABIOS 15:555-556.
Yang, Z., R. Nielsen, N. Goldman, and A.-M. K. Pedersen. 2000. Codon-substitution models for heterogeneous selection pressure at amino acid sites. Genetics 155:431-449.(Volodymyr Dvornyk*, Hong-)