Analysis of the hli gene family in marine and freshwater cyanobacteria

Abstract

Certain cyanobacteria thrive in natural habitats in which light intensities can reach 2000 μmol photon m⁻² s⁻¹ and nutrient levels are extremely low. Recently, a family of genes designated hli was demonstrated to be important for survival of cyanobacteria during exposure to high light. In this study we have identified members of the hli gene family in seven cyanobacterial genomes, including those of a marine cyanobacterium adapted to high-light growth in surface waters of the open ocean (Prochlorococcus sp. strain Med4), three marine cyanobacteria adapted to growth in moderate- or low-light (Prochlorococcus sp. strain MIT9313, Prochlorococcus marinus SS120, and Synechococcus WH8102), and three freshwater strains (the unicellular Synechocystis sp. strain PCC6803 and the filamentous species Nostoc punctiforme strain ATCC29133 and Anabaena sp. (Nostoc) strain PCC7120). The high-light-adapted Prochlorococcus Med4 has the smallest genome (1.7 Mb), yet it has more than twice as many hli genes as any of the other six cyanobacterial species, some of which appear to have arisen from recent duplication events. Based on cluster analysis, some groups of hli genes appear to be specific to either marine or freshwater cyanobacteria. This information is discussed with respect to the role of hli genes in the acclimation of cyanobacteria to high light, and the possible relationships among members of this diverse gene family.

1 Introduction

The major peripheral light-harvesting complex (LHC) of cyanobacteria is the water-soluble phycobilisome, which is comprised of tetrapyrrole-bound phycobiliproteins and non-pigmented linker polypeptides[1]. In vascular plants, the major LHC is composed primarily of the integral membrane Lhc polypeptides that contain three transmembrane helices and bind chlorophylls a and b. The Lhc polypeptides are encoded by a family of genes that generally contains more than 10 individuals [2,,4]. However, there are more distantly related Lhc genes that comprise the Lhc extended gene family. Polypeptides encoded by this extended family include the early light-inducible proteins (ELIPs), the four transmembrane helix-containing polypeptide PsbS or PSII-S[5], and polypeptides that have one or two putative transmembrane helices 6,7]. Several genes have been identified on cyanobacterial genomes that encode single-helix members of the Lhc extended gene family 8,9]. These genes have been designated hli (high light inducible; protein designation HLIPs)[8] or scp (small cab-like proteins; protein designation Scps)[9]. While hli genes were first noted in Synechococcus sp. strain PCC7942[10], they were subsequently identified in other cyanobacteria[11], red algae[12] and vascular plants[6].

The pattern of expression of hli genes in cyanobacteria and vascular plants is similar to that of the genes encoding ELIPs; hli mRNAs and encoded polypeptides accumulate under conditions that result in the absorption of excess excitation energy, including exposure to high irradiance, nitrogen starvation and low temperature [1,6,8]. Synechocystis sp. strain PCC6803 deleted for all four of its hli genes was shown to be photosensitive, and under strong illumination the cells lost all variable fluorescence and died[13]. By analogy to vascular plant ELIPs, a number of functions have been suggested for the cyanobacterial HLIPs. They may associate with pigments, perhaps transiently, serving as chlorophyll carriers[11], function in the dissipation of excess absorbed light energy within antennae complexes 6,14], or modulate the biosynthesis of chlorophyll[15]. Essentially all evidence suggests that photosynthetic organisms need HLIPs under stressful, often growth-limiting conditions.

Specific Prochlorococcus species thrive in high-light, nutrient-poor environments that characterize the surface waters of the open oceans, while others grow at greater depths in lower-light and higher-nutrient environments 16,17]. Although phylogenetic analyses, based on 16S rDNA and rpoC sequences, have established that Prochlorococcus is a cyanobacterial genus, it does not contain the light harvesting phycobilisomes typical of most cyanobacteria[18]. Instead, the major antennae pigment complex contains chlorophylls a and b associated with polypeptides that are similar to CP43, a polypeptide integral to the core of photosystem II that binds chlorophyll a[19].

An understanding of the light and nutrient habitats in which specific cyanobacterial strains thrive, and recent acquisition of complete or near complete sequence information for several cyanobacterial genomes, have made it attractive to explore both intra- and inter-species relationships among cyanobacterial hli genes. The genomes of seven cyanobacterial strains have been sequenced at the Kazusa DNA Research Institute, Japan; the Joint Genome Institute (JGI), USA and the Genoscope, France, and the sequences of other cyanobacterial genomes are nearly complete. Three of these cyanobacteria, the unicellular Synechocystis strain PCC6803 (SC), the filamentous Anabaena (Nostoc) sp. strain PCC7120 (AN) and Nostoc punctiforme strain ATCC29133 (NT) grow in freshwater habitats. The marine species for which complete genome information is available are represented by the high-light ecotype Prochlorococcus marinus strain MED4 (PM) and three species that grow in low/moderate light, P. marinus strain MIT9313 (PL), P. marinus strain SS120 (PS) and Synechococcus sp. strain WH8102 (SN) 17,20]. In this report we analyze the cyanobacterial hli gene family and discuss the differences in this gene family among the seven different cyanobacterial strains for which complete genome sequences are available.

2 Materials and methods

2.1 Genome data

Sequences of the genomes of Synechocystis and Anabaena were downloaded from the EMBL web site (http://www.ebi.ac.uk/genomes/); those of the Prochlorococcus strains MED4 (version of 12/19/2001), MIT9313 (version of 01/25/2002), and Synechococcus strain WH8102 (version of 01/26/2002) from ftp sites of the JGI (ftp://ftp.jgi-psf.org/pub/JGI_data/Microbial/prochlorococcus/final.011129; ftp://ftp.jgi-psf.org/pub/JGI_data/Microbial/prochlorococcusII/final.010823; ftp://ftp.jgi-psf.org/pub/JGI_data/Microbial/synechococcus/final.010910). Contig sequences of the genome of N. punctiforme (version of 01/25/2002) were downloaded from the JGI ftp site: (ftp://ftp.jgi-psf.org/pub/JGI_data/Microbial/nostoc/010409/). The genomic sequence of Prochlorococcus SS120 was downloaded from http://www.sb-roscoff.fr/Phyto/ProSS120. The data has been provided freely by the US DOE Joint Genome Institute for use in this publication/correspondence only.

2.2 Gene detection

The hli genes were identified by similarity searches using the four hli genes of SC as query sequences (gene identifiers: ssl1633(hliC), ssr1789(hliD), ssl2542(hliA), ssr2595(hliB) in Cyanobase) against complete cyanobacterial genome sequences translated in their six reading frames using the tBLASTn program[21]. Because of the very small size of these genes, the tBLASTn program was operated with a default E-value threshold of 10 and no filter for low-complexity regions. In a second step, a multiple alignment of the hli genes was performed using ClustalX software[22] in order to build a profile Hidden Markov Model (profile HMM) with the hmmbuild program of the HMMER 2 package (http://hmmer.wustl.edu/)[23]. This profile HMM was calibrated with the hmmcalibrate program and used with the hmmsearch program (HMMER 2 package) to identify hli genes that were not detected by tBLASTn. All three of these programs were operated with the default options. Using the HMMER package five new hli genes were detected in PL (hli05, 06, 07, 08, 09) and two in SN (hli07 and hli08).

2.3 Sequence clustering

Clustering of hli genes was achieved using the GeneRAGE algorithm (http://www.ebi.ac.uk/research/cgg/services/rage/)[24], which groups sequences based on their similarity. We chose to use BLASTp to detect similarity between Hli polypeptide sequences. The choice of a threshold value is critical to obtain the optimal ratio between specificity and sensitivity. To determine the optimal threshold, an initial all-against-all comparison of protein sequences was made using BLASTp. After analysis of these results, an E-value cut-off of 10⁻¹³ was chosen as the threshold.

2.4 Conservation of regions neighboring hli genes

To examine whether regions surrounding the hli genes were conserved among the different genomes, orthologous relationships between open reading frames (ORFs) flanking the hli genes were investigated. Initially, ORFs flanking the hli genes were identified using the Artemis software package (http://www.sanger.ac.uk/Software/Artemis/). Each of these flanking ORFs was then compared against the complete cyanobacterial genome sequences using tBLASTn. Whenever the hli genes of two different genomes were found to have similar neighboring ORFs, the neighboring ORFs were used for reciprocal comparisons using tBLASTn; ORFs that satisfied the criterion of reciprocal best hit for each other were considered to be orthologs.

3 Results and discussion

Multiple hli genes are present on the genomes of all seven cyanobacterial strains examined in this study (Table 1). Based on this analysis, the four marine cyanobacteria PM, PL, PS and SN have 22, 9, 13 and 8 hli genes, respectively, while the three freshwater cyanobacteria SC, AN and NT have 4, 8 and 9 hli genes, respectively. These 73 hli genes (Table 1) were analyzed to help understand the significance of the large number of hli genes on the genome (especially in PM), to determine possible relationships among the different hli genes and to evaluate whether the related clusters of hli genes represent group-specific (e.g. present only in the marine cyanobacteria or Prochlorococcus species etc.) and/or possible functional classes.

1

hli genes detected in seven fully sequenced cyanobacterial genomes

Genome	Gene	Start	Stop	Size (aa)	GeneRAGE cluster
PM	hli01	292 721	292 575	49	6
PM	hli02	320 885	321 115	77	8
PM	hli03	634 018	634 170	51	7
PM	hli04	713 298	713 194	35	14
PM	hli05	697 543	697 695	51	12
PM	hli06	702 237	702 341	35	14
PM	hli07	702 344	702 550	69	10
PM	hli08	702 553	702 813	87	12
PM	hli09	702 844	702 969	42	15
PM	hli10	706 726	706 851	42	16
PM	hli11	713 054	713 191	46	12
PM	hli12	983 410	983 306	35	14
PM	hli13	777 735	777 469	89	17
PM	hli14	962 255	962 106	50	10
PM	hli15	968 466	968 720	85	24
PM	hli16	1 274 111	1 274 215	35	14
PM	hli17	1 274 218	1 274 427	70	10
PM	hli18	1 274 427	1 274 687	87	12
PM	hli19	1 274 718	1 274 843	42	15
PM	hli20	1 600 841	1 600 662	60	5
PM	hli21	1 388 883	1 389 014	44	12
PM	hli22	1 389 022	1 389 276	85	15
PL	hli01	413 647	413 847	67	5
PL	hli02	72 191	72 409	73	6
PL	hli03	228 052	228 201	50	7
PL	hli04	120 778	121 041	88	8
PL	hli05	746 073	745 927	49	9
PL	hli06	572 474	572 674	67	10
PL	hli07	744 236	744 114	41	11
PL	hli08	572 319	572 459	47	12
PL	hli09	572 833	572 937	35	13
PS	hli01	1 388 013	1 388 129	39	15
PS	hli02	1 489 980	1 489 831	50	7
PS	hli10	1 387 372	1 387 512	47	19
PS	hli11	1 387 549	1 387 653	35	14
PS	hli12	1 397 334	1 397 438	35	14
PS	hli13	1 387 253	1 387 369	39	20
SN	hli01	489 613	489 413	67	21
SN	hli02	1 840 235	1 840 432	66	8
SN	hli03	2 114 476	2 114 625	50	7
SN	hli04	2 299 673	2 299 924	84	2
SN	hli05	1 389 330	1 389 683	118	22
SN	hli06	1 152 224	1 152 361	46	6
SN	hli07	403 771	403 553	73	23
SN	hli08	830 619	830 443	59	5
SC	hli01	701 350	701 138	71	2
SC	hli02	982 968	983 180	71	2
SC	hli03	398 188	398 361	58	3
SC	hli04	1 141 803	1 142 015	71	4
AN	hli01	1 006 982	1 006 782	67	1
AN	hli02	607 714	607 499	72	2
AN	hli03	2 836 843	2 836 676	56	3
AN	hli04	6 277 367	6 277 543	59	4
AN	hli05	3 686 003	3 686 203	67	1
AN	hli06	3 686 251	3 686 451	67	1
AN	hli07	4 499 702	4 499 526	59	4
AN	hli08	531 645	531 526	40	4
NTa	hli01	81 354	81 175	60	4
NTa	hli02	20 142	20 354	71	2
NTa	hli03	181 687	181 854	56	3
NTa	hli04	15 331	15 122	70	2
NTa	hli05	77 669	77 878	70	1
NTa	hli06	77 923	78 123	67	1
NTa	hli07	238 527	238 324	68	1
NTa	hli08	95 265	95 089	59	4
NTa	hli09	107 200	107 024	59	4
PS	hli03	1 426 441	1 426 352	30	18
PS	hli04	118 186	118 332	49	6
PS	hli05	88 528	88 283	82	8
PS	hli06	1 297 723	1 297 986	88	17
PS	hli07	1 097 580	1 097 786	69	10
PS	hli08	1 097 458	1 097 562	35	14

Start and stop positions for each gene are specified. Abbreviations used are: PM, P. marinus strain MED4; PL, P. marinus strain MIT9313; PS, P. marinus strain SS120; SN, Synechococcus sp. strain WH8102; SC, Synechocystis sp. strain PCC6803; AN, Anabaena sp. strain PCC7120; NT, N. punctiforme strain ATCC29133. Marine cyanobacteria are in bold. For each gene the size of the corresponding protein and the GeneRAGE cluster number is shown.

^aThe Nostoc sequence is incomplete, so start and stop data represent information from individual contigs (hli01: contig 483; hli02: contig 397; hli03: contig 507; hli04: contig 485; hli05: contig 480; hli06: contig 480; hli07: contig 509; hli08: contig 476; hli09: contig 486).

The genome of PM, the high-light ecotype of marine Prochlorococcus, encodes at least 22 hli genes. The number of hli genes on the PM genome is significantly greater than the number present on the genomes of the low-light ecotypes PL and PS (which have nine and 13 hli genes, respectively), the marine Synechococcus SN (which has eight hli genes) and the freshwater species SC, AN and NT (which have between four and nine hli genes). In SC, the hli gene family is required for survival in high light. A mutant of SC lacking one or two copies of the hli gene survives, but it is at a disadvantage relative to wild-type cells as evaluated by growth competition experiments performed in high light. If all four of the SC hli genes are disrupted, the cells die following exposure to high light[13]. These results suggest that HLIPs act in a cumulative manner to sustain the cells in high light, although there may also be requirements for particular gene products under specific environmental conditions. The significant increase in the number of hli genes on the genome of the Prochlorococcus high-light ecotype, in spite of the fact that this strain has the smallest genome amongst the seven genomes analyzed, may reflect a requirement for the additional gene products in coping with the persistent high-light conditions associated with the ocean surface[17]. Conversely, the smaller number of hli genes in PL, PS, SN and the freshwater strains, which are adapted to low/moderate-light growth conditions, is consistent with an important role for HLIPs in habitats in which the organisms are under persistent excitation pressure 25,26].

The arrangement of 22 hli genes in PM is shown in Fig. 1; the hli genes are scattered throughout the genome, with some gene clustering in particular regions. There are two instances in which two hli genes are contiguous (hli 11 and hli 12, and hli 21 and hli 22). In addition, two other regions of the genome contain four tandemly arranged hli genes (hli 06-09 shown as A and hli 16-19 shown as B in Fig. 1). The four tandemly arranged genes in region A are flanked by hli 05 and hli 10 (these genes are 4.5 kb and 3.7 kb distant from region A, respectively). Strikingly, the two clusters of tandemly arranged genes each cover 1.3 kbp and represent exact duplications; i.e. the sequence of hli 06-09 is identical to that of hli 16-19 at the nucleotide level. The duplicated 1.3-kbp region is comprised exclusively of four hli genes plus a small ORF of 59 codons (Fig. 1). This ORF has no similarity to other ORFs in the public databases, and thus may not represent a protein product. If this is the case, the duplication has resulted in the exclusive doubling of just the hli genes. A cursory examination of the PM genome indicates that this 1.3-kb region is the only exact duplication in excess of 1 kb in the genome. Interestingly, while two other hli genes (hli 04 and hli1 2) are identical in their predicted amino acid sequences, they are not identical at the nucleotide level.

1

Position of the 22 hli genes detected in the genome of P. marinus strain MED4. Gene groups A and B correspond to exactly duplicated regions containing the four hli genes (hli 06-09 and hli 16-19). The arrangement of the duplicated genes in groups A and B are shown on the right. Note that hli 06, hli 07 and hli 08 (group A) and hli 16, hli 17 and hli 18 (group B) are overlapping (see text for details).

Since identification of the duplication of the hli gene clusters was based on genome sequence information, it was possible that the exact nucleotide match observed was generated by a computational artifact that yielded improper assembly results. To establish whether or not the putative duplication was an artifact, primers within the duplicate regions paired with specific primers flanking the duplications were constructed and used for PCR. The results of these experiments demonstrated the presence of the duplicate sequences at two distinct genomic locations (data not shown, Stephanie Stillwagen, DOE JGI, Walnut Creek, CA, USA, personal communication). Identical sequences within these hli gene clusters may reflect a very recent duplication of this locus, or the occurrence of a copy correction mechanism in the cell that maintains identity between the two sequences (although there is no experimental evidence to support such a mechanism). As more bacterial genomes are being sequenced and analyzed, it will be interesting to examine them for the presence of exact sequence duplications and gene duplications. The significance and mechanisms for creation and maintenance of these duplications is not yet understood although genome-wide studies suggest that duplicate genes are subject to specific selection and may not evolve at the same rate[27].

The arrangement of the clustered genes in the 1.3-kbp duplicate region is striking; the last nucleotide of the stop codon (TAA) for hli06 and hli07 is the first nucleotide of the start codon (ATG) of hli07 and hli08, respectively (Fig. 1). This type of overlapping gene arrangement was demonstrated to be important for coordinating the expression of the overlapping trpA and trpB genes in the Escherichia coli trp operon[28]. Other examples of translational coupling have also recently been noted in Prochlorococcus MED4 (phoB–PhoR) and in the pta–ack bicistronic operon of Corynebacterium glutamicum16,29]. Analysis of expression from the hli gene clusters would identify populations of polycistronic mRNAs that are transcribed from these genes, and may reveal how environmental conditions influence both the levels and distribution of distinct polycistronic transcripts.

The small size of the hli genes, the finding that some members of the gene family represent exact duplications, and the somewhat low degree of conservation among the different Hli proteins make classical sequence-based phylogenetic approaches difficult. In particular, bootstrap values of phylogenetic trees obtained by various methods (e.g. maximum likelihood) are very low (data not shown), raising serious concerns about the robustness of the observed relationships. To gain insights into the relationships among the 73 putative hli genes, we used the GeneRAGE program (Table 2). This program is a robust algorithm for quickly and accurately clustering large protein datasets into families and subfamilies[24]. Although no single program can give definitive answers regarding the relationship between genes and organisms, it does set the stage for a more complete analysis and provides information for hypothesis generation and evaluation. A number of observations resulting from these analyses are discussed below.

2

Cluster analysis of the 73 hli genes using the GeneRAGE program (see Section 2)

Cluster	Genome	Gene
1	AN	hli01
1	AN	hli05
1	AN	hli06
1	NT	hli05
1	NT	hli07
2	SN	hli04
2	SC	hli01
2	SC	hli02
2	AN	hli02
2	NT	hli02
2	NT	hli04
3	SC	hli03
3	AN	hli03
3	NT	hli03
4	SC	hli04
4	AN	hli04
4	AN	hli07
4	AN	hli08
4	NT	hli01
4	NT	hli08
4	NT	hli09
5	PM	hli20
5	PS	hli09
5	PL	hli01
5	SN	hli08
6	PM	hli01
6	PS	hli04
6	PL	hli02
6	SN	hli06
7	PM	hli03
7	PS	hli02
7	PL	hli03
7	SN	hli03
8	PM	hli02
8	PS	hli05
8	PL	hli04
8	SN	hli02
9	PL	hli05
10	PM	hli07/17
10	PM	hli14
10	PS	hli07
10	PL	hli06
11	PL	hli07
12	PM	hli05
12	PM	hli08/18
12	PM	hli11
12	PM	hli21
12	PL	hli08
13	PL	hli09
14	PM	hli04/12
14	PM	hli06/16
14	PS	hli08/11
14	PS	hli12
15	PM	hli09/19
15	PM	hli22
15	PS	hli01
16	PM	hli10
17	PM	hli13
17	PS	hli06
18	PS	hli03
19	PS	hli10
20	PS	hli13
21	SN	hli01
22	SN	hli05
23	SN	hli07
24	PM	hli15

Genes were aligned using the BIOEDIT program. Marine cyanobacteria are in bold.

The 73 hli genes were separated into 24 clusters (these clusters contain up to seven genes, although 11 of the clusters contain a single gene representative). The clustering analysis clearly shows a strong divergence between marine and freshwater species (Fig. 2). This may indicate an early separation of the marine and freshwater cyanobacteria and the generation of divergent hli gene clusters within these environmentally distinct groups. However the strong divergence may also reflect very distinct evolutionary pressures that are associated with the markedly different environments in which these organisms are able to thrive.

2

Alignment of hli genes in specific clusters of freshwater species (A) and marine species (B) using the GeneRAGE program. Dots mark residues that are identical to the top sequence in each cluster and dashes represent gaps. Clusters that have only a single representative are not shown.

3.1 Fresh water species

Clusters 1–4 contain all 22 hli genes of the freshwater species, AN, NT and SC (Fig. 2A and Table 2). Of these, clusters 2, 3 and 4 contain at least one representative from each species; while cluster 1 contains three genes each from NT and AN (AN_hli 01, AN_hli05 AN_hli06 and NT_hli05, NT_hli06, NT_hli07). Cluster 2 contains two representatives from NT and SC each (NT_ hli02 and NT_hli04, SC_hli01 and SC_hli02) and one from AN (hli02) and SN (hli04), cluster 3 contains a single representative from each of the freshwater species (SC_hli03, AN_hli03, NT_hli03) and cluster 4 contains seven genes, with three each from NT and AN and one from SC (hli 04). Based on nearest neighbor analysis, AN_hli02 and NT_hli 02 (both in cluster 2), AN_hli04 and NT_hli01 (both in cluster 4), and AN_hli03 and NT_hli03 (both in cluster 3) all share neighboring genes, as do NT_hli04 and SC_hli02 (both in cluster 2) (Fig. 3). These results suggest the following features of the hli gene families:

3

Conserved genes in the neighborhood of hli genes arranged according to the GeneRAGE clusters. ORFs that are painted with the same color are considered orthologs. For clusters 5–8 only genes for which the assignment of a gene name is unambiguous have been labeled (e.g. rpoC2). For clusters 2–4 the Cyanobase Gene Identifier for flanking genes of Synechocystis or Anabaena have been labeled for identification of flanking genes.

1. 1

   There may be three basic hli gene ‘forms’ in the freshwater species represented by the sequences in clusters 2–4. Furthermore, strong sequence similarity among pairs of polypeptides representative of the different freshwater species, encoded by genes within these groups, combined with nearest neighbor analyses with respect to these genes, maybe suggestive of orthologous relationships among specific members of these gene clusters. Whether the genes in clusters 2, 3, and 4 have distinct functions is still unclear. Recent evidence suggests that a severe phenotype is associated only with a mutant in which all four hli genes are inactivated; mutants lacking one or two hli genes do not exhibit this phenotype which may be indicative of a redundancy or overlapping gene functions[13].

2. 2

   AN and NT have multiple copies of closely related hli genes within clusters 1 and 4 (and cluster 2 for NT). However, in SC there appears to have been only one recent duplication (hli01 and hli02 for which the encoded amino acid sequences are 87% identical; 94% similar). Clusters 4 and 1 have three representatives each from AN and NT, but one or none from SC, respectively. The significance of this is unclear since all three species grow in relatively low-light environments. However, both NT and AN are multi-cellular and developmentally complex (both species can differentiate nitrogen-fixing heterocysts which contain a highly modified photosynthetic apparatus that does not evolve oxygen). It would be particularly interesting to follow expression patterns of the hli genes under different growth conditions (e.g. low-nitrogen, high-light) as well as to examine regions upstream of the hli coding regions to identify conserved sequence elements.

3. 3

   Only one marine cyanobacterial HLIP sequence, SN_hli04, clusters with the sequences from the freshwater organisms (within cluster 2). The amino acid similarity between SN_hli04 and the most closely related gene in cluster 2, NT_hli02, is 43%. The significance of this clustering is not apparent at this time.

4. 4

   A comparison of sequences in NT, AN and SC demonstrates that there is little conservation of genes that flank the hli genes between the filamentous and unicellular cyanobacteria (Fig. 3). There is evidence that there have been rearrangements of genomes in some freshwater cyanobacteria, including SC and there is also evidence for recent transposition events in SC 30,31]. This makes it less surprising that neighborhood conservancy is not maintained between NT/AN and SC.

3.2 Marine species

Clusters 5–8 all contain a single representative from each of the four marine species analyzed (Fig. 2B). Furthermore, these genes also all share flanking gene neighbors (Fig. 3). This may indicate that these four gene clusters are essential for all marine species (somewhat similar to the case in freshwater species where there are three clusters that have at least one representative from each of the species). Although the phylogenetic relationships among the gene groupings are difficult to evaluate, based on sequence similarity, the marine gene clusters 6–8 are possibly most closely related to the freshwater gene clusters 2 and 3. It is unclear if the apparent similarities between these groups reflect the specific functions of the proteins, but it would be interesting to explore this possibility by examining the influence of diverse environmental conditions on the expression of genes within these clusters.

Five clusters (10, 12, 14, 15 and 17) contain multiple genes from the marine species but not all of the species are represented in these clusters (Fig. 2B). Furthermore, 11 clusters (9, 11, 13, 16 and 18–24) each contain only one representative gene. This indicates that there are several marine hli genes (three from PL (hli05, hli07, hli09); two from PM (hli10 and hli15); three from PS (hli3, hli10 and hli13); three from SN (hli01, hli05 and hli07)) that have diverged to the point of not being grouped together using the GeneRAGE program. Clusters with multiple family members from one species may represent an evolutionary trend toward duplication of specific genes; however until more experimental evidence is available, it is not possible to draw any direct conclusions based simply on sequence similarities.

In Fig. 4, 23 out of the 44 hli genes within the Prochlorococcus species (PL, PM and PS) are aligned using ClustalX. It is quite striking that within this group the C-terminus of the HLIPs maintain a strongly conserved motif TGQIIPGF/IF. This motif is not conserved in clusters 5, 6, 7 and 8 (which contain one representative from each of the marine cyanobacterial species) or in clusters 17 and 18. It is also notably missing in all of the freshwater strains. The fact that this motif is only present in a subset of the Prochlorococcus HLIPs may be indicative of specialized function.

4

Comparison of all the hli sequences of Prochlorococcus strains that share the conserved C-terminus motif (TGQIIPGI/FF). Dots mark residues that are identical to the top sequence and dashes represent gaps.

3.3 Concluding remarks

We have attempted an analysis of the large hli gene family that is ubiquitous in all cyanobacterial species examined so far (as well as in other groups). This study was motivated by the initial observation that there was an apparent over-representation of hli genes in PM, the high-light adapted marine strain. The presence of a very large hli gene family in PM is consistent with recent results showing that a mutant of SC lacking all four copies of the hli gene was unable to survive in high light[13], raising the obvious question of whether the number of hli genes in an organism could be correlated with adaptation/acclimation to the light environment.

To attempt to answer this question, we took a bioinformatics approach to analyze the 73 hli genes identified in seven recently sequenced cyanobacterial strains. There are a number of problems associated with the use of small genes to construct a phylogeny. The construction of cyanobacterial phylogenies, is particularly problematic since there are issues related to the ancient history of this group; over the course of evolution cyanobacteria may have experienced both lateral gene transfer and the formation of gene mosaics 32,33]. Furthermore, it is even difficult to determine which bacteria are most closely related to cyanobacteria; some sister groups are considered to be the Deinococcales and spirochetes and more recent analyses suggests a relationship with low GC Gram-positive bacteria such as Halobacterium and Aquifex aeolicus[34,,36]. To avoid these potential pitfalls we used a clustering analysis method (GeneRAGE) that does not make any assumptions about phylogenetic relatedness between genes.

One of the most obvious results from the analyses presented above suggests that there is a significant distinction between hli genes in the marine and freshwater strains. Since there has not yet been an extensive analysis of genes across various cyanobacterial species, it is useful to compare this data in the context of some recent molecular phylogenetic studies of various cyanobacterial species using 16S rRNA sequence data 37,38]. Honda et al. used 16S rRNA sequences from a variety of 44 different freshwater and marine strains to determine evolutionary lineages within the cyanobacteria. Based on maximum likelihood and neighborhood joining trees generated with these data, they suggested that there were at least seven different evolutionary lineages within the cyanobacteria. These trees also indicate that unicellular and filamentous species may be closely arranged on the same branch of the tree, but that freshwater and marine strains are often well separated. This is consistent with our results where, for example, unicellular SC hli genes are much more closely related (based on clustering) to the hli genes from freshwater, filamentous species than to the hli genes from unicellular marine strains. The generation of phylogenies based on 16S rRNA sequences was often taken as the benchmark for phylogenetic analyses. However with the explosion of information associated with the generation of complete genome sequences from a variety of prokaryotes, a number of individual genes within these genomes can be used to help establish phylogenetic relationships among the cyanobacteria. Differences in phylogenetic relationships that are obtained when different genes are used to evaluate such relationships suggest that the generation of a single, consistent evolutionary tree may not be easy, especially since multiple pressures imposed by specific environments may differentially influence the apparent rates at which specific genes evolve. In the case of the hli genes, the evolutionary pressure for this gene family to evolve and adapt to high-light conditions may create a phylogeny that is not necessarily consistent with a 16S rRNA tree. This raises the interesting possibility of analyzing and classifying a variety of molecular markers potentially indicative of specific environmental pressures (high-light or nutrient-stress). Analyses which focus on the proliferation or drastic reduction of genes in a particular adapted ecotype or species (for instance, the proliferation of hli genes in a high-light-adapted strain versus in low-light-adapted ecotypes or the steep reduction in two-component regulatory systems in marine species relative to freshwater species may allow us to gain an insight into environmental selection pressures, and the extent to which such pressures has shaped the individual cyanobacterial species. Since we now have a large data base of information from a range of different cyanobacteria that have adapted to very different ecological niches, this orientation represents an attractive approach for future work.

Acknowledgements

We thank F. Partensky the coordinator of the Prochlorococcus SS120 genome sequencing project and the Genoscope, France (M. Salanoubat) for providing us with access to the genome sequence prior to publication. This study is partly supported by the MARGENES program (EU contract xxx). A.D. is supported by a doctoral fellowship from Région Bretagne. The cooperation between Stanford and Roscoff is supported by an NSF–CNRS bilateral grant.