Despite the apparently large diversity of pigmentation existing among marine Synechococcus, these can be partitioned into only three major types based on the phycobiliprotein composition of the rods: type 1 representatives have only PC, type 2 have PC and PEI and type 3 have PC, PEI and PEII. Type 3 can be further subdivided into four subtypes (3a through 3d) based on the ratio of the two chromophores (PEB and PUB) bound to PEs, a ratio that can be low, medium, high or variable. Figure 1a illustrates these different pigment types or subtypes and their corresponding colors. The 11 fully sequenced marine Synechococcus strains cover the whole range of PBS pigmentation known so far in this group 62333. Pigment type 1 is represented by the blue-green, PE-lacking strains WH5701 and RS9917. These strains absorb light optimally in the wavelength range 600-660 nm, that is, red and orange light (Figure 1b). The genome of the fuchsia pink WH7805 strain (pigment type 2) contains a single set of PE genes encoding a PEI-like complex, as detailed below. The whole cell absorption maximum of this form of PE devoid of PUB (A_max = 570 nm, corresponding to yellow-green light) is red-shifted relative to other PEs (Figure 1b).

All strains displaying pigment type 3 possess both PEB and PUB chromophores. Subtypes 3a through 3c differ from one another in their whole cell ratio of PUB to PEB (hereafter PUB:PEB), as assessed by their fluorescence excitation maxima (F_{495 nm} : F_{550 nm}) with emission at 580 nm (Table 1). Note that the use of this fluorescence excitation ratio is preferable to using the corresponding absorption ratio (A_{495 nm}: A_{550 nm}) to characterize these different subtypes in vivo, since the carotenoids zeaxanthin and β-carotene have a notable contribution to the wavelength range of the PUB absorption peak (Figure 1b). The PUB:PEB can be either low (approximately 0.4) in type 3a strains such as WH7803, medium (approximately 0.8) in type 3b strains such as RCC307 or high (>1.7) in type 3c strains such as in WH8102 and CC9605 (Table 1). Depending on this ratio, PBSs of these strains preferentially harvest either green light (550 nm) or blue-green light (495 nm) (Figure 1b). Finally, pigment type 3d includes strains with a variable PUB:PEB (0.7-1.7), depending on whether these cells are grown under white/green or blue light 3334. These type IV chromatic adapters include the strains CC9311, RS9916, BL107 and CC9902 as well as a number of other strains that have not yet been sequenced (including WH8020, M16.17, M11.1, RCC61 (a.k.a. Minos 11) and RS9911; Table 1 and data not shown). To this suite of pigment types can be added a 'moderately high' PUB:PEB subtype (PUB:PEB approximately 1.2), represented by strain WH8103 and which is indistinguishable by eye from, and included within, type 3c (Figure 1a). Though as yet unsequenced, the genome of WH8103 has been screened, in part, by suppression subtractive hybridization 35.

The color and specific absorption properties of whole Synechococcus cells (Figure 1) are mainly determined by the major phycobiliprotein form occurring in the PBS rods. Isolated PC and/or PE complexes have been previously characterized in a few marine Synechococcus strains, including WH7803, WH7805, WH8102, WH8103 and the chromatic adapters WH8020 (under white light only), M11.1 and M16.17 13161718193436, as summarized in Table 1. In order to explore further the diversity and possible combinations of these phycobiliproteins in the different Synechococcus pigment types, we have used sucrose density gradients and isoelectric focusing to isolate PC, PEI and/or PEII from a number of other strains and then have determined their optical properties (Figures 2 and 3 and Table 1).

The PC present in WH5701 and RS9917, which formed a sky blue band on isoelectric focusing gels (not shown), had absorption (A_max = 621 nm) and fluorescence (F_max = 648 nm) properties typical of C-PC (Figure 2a), that is, known to bind only PCB chromophores 15. We also found C-PC in the PE-containing, PUB-lacking strain WH8018, whereas WH7805 (which, like WH8018, displays pigment type 2) is known to possess R-PCIII 17. R-PCIII has a molar PCB:PEB of 2:1, like the R-PCI of red algae, but a different spectrum, with an A_max at 555 nm and a shoulder at 590 nm 17. Our isolation protocol did not allow us to obtain a pure PC fraction from any of the PEII-containing strains, because the PC band was always contaminated by variable amounts of PEII. It is known, however, that Synechococcus sp. WH7803, like WH8020 and WH8103, possesses a R-PCII type PC with a molar PEB:PCB of 2:1; it has absorption peaks at 533, 554 and 615 nm and maximal fluorescence emission at 646 nm 16.

Several types of PEI can be distinguished based on their different optical properties. The major phycobiliprotein found in WH7805 and WH8018, a PEI-like phycobiliprotein, exhibited an A_max at 556 nm and an F_max at 577 nm (Figure 2b). We have called it PEI-A* to distinguish it from the PEI-A found in Synechococcus strains displaying the 3a and 3b pigment types. PEI-A has blue-shifted optical properties (A_max = 550 nm; F_max = 572 nm; Figure 2c) compared to PEI-A*, though both forms bind only PEB chromophores. PEI-B, which has a molar PUB:PEB of 2:3 18, has been found in all strains exhibiting pigment type 3c examined thus far, as well as in some chromatic adapters, including M11.1 and M16.17 34. It has maximal absorption at 493 and 563 nm and fluorescence at 573 nm (Figure 2d).

Similarly, one can distinguish three optical types of PEII differing by their PUB:PEB. All have two absorption maxima (or at least shoulders) around 495 nm and 550 nm, due to the two chromophores they bind, and a maximal fluorescence emission around 565 nm. PEII-A (Figure 3a) is found only in Synechococcus pigment type 3a, including WH7803 18, Almo03 and RS9912 (this study). Its molar PUB:PEB is most likely 1:5, although the cysteinyl site to which the sole PUB chromophore is bound (either α-75 or β-50/61) has not yet been ascertained. PEII-B (Figure 3b) is found in RCC307 (Table 1) and in all white light-grown chromatic adapters that have been screened thus far, including WH8020 18, M11.1, M16.17 34 and RS9916 (this study). Its molar PUB:PEB is 2:4. PEII-C (Figure 3c) is found in Synechococcus pigment type 3c, including WH8103 18, WH8102 19, Oli31 and CC9605 (this study) as well as in the blue light-grown chromatic adapters 34. The molar PUB:PEB of this PEII has been shown to be 4:2 18.

After careful annotation, we compared PBS gene complement (Additional data file 1) and organization in the 11 different genomes. One remarkable trait of marine Synechococcus is that most of the PBS genes are gathered into a few gene clusters 1937. As in several other cyanobacteria, a first small cluster groups together four AP core genes, in the order apcE-A-B-C, while two other core genes, apcD and apcF (encoding the minor α-B and β-18 AP subunits, respectively) have no PBS gene in their close vicinity. Most of the PBS rod genes are located in a much larger cluster, the size of which increases with the complexity of the rod structure from approximately 9-10 Kbp in pigment type 1 up to 27-28.5 Kbp in chromatic adapters (Figure 4). Interestingly, the gene organization in this region is very similar for strains belonging to a given pigment type. It is also similar between the chromatic adapters and the medium PUB:PEB strain RCC307.

In most genomes, the 5'-end of the PBS rod gene region starts with a short gene of unknown function (unk1). In RCC307, however, the unk1 ortholog is found elsewhere in the genome. The 3'-end of the region consists of a well conserved gene predicted to encode a low molecular weight phosphotyrosine phosphatase. In the blue-green, PE-lacking strains, the rest of the region is mainly composed of two identical cpcB-A operons encoding the C-PC α- and β-subunits and of genes encoding three rod linkers, one rod-core linker and two types of phycobilin lyases (CpcT and CpcE/F; see below). Both RS9917 and WH5701 have an additional cpcB gene copy outside the PBS rod gene region but, surprisingly, no additional cpcA.

A part of the PC gene cluster found in the blue-green strains (cpcCI-D-B-A-CII) is replaced in the fuchsia pink strain WH7805 by a set of 19 genes, likely involved in the synthesis and regulation of a PEI-like complex (Figure 2). The pebA and pebB genes, located at the 3'-end of this insertion, are known to be involved in the synthesis of PEB chromophores 38.

This PE region can also be found in all PEII-containing strains, but it is interrupted by an additional sub-region containing 5 to 9 genes, between the PE regulator cpeR 39 and the putative lyase gene cpeY in WH7803 (or cpeZ in the other strains). This inserted sub-region includes genes encoding the PEII α- and β-subunits, two phycobilin lyases, one linker polypeptide and two or three uncharacterized proteins. In addition, all PEII-containing strains have, upstream of cpcGII, an ortholog of aplA. Its product, AplA, which shows homology to the AP α-subunit (ApcA), was recently described in Fremyella diplosiphon as belonging to a new class of cyanobacterial photosensors of unknown function 40.

In the following sections, we have analyzed more specifically the phyletic profile (that is, the different patterns of occurrence of orthologs in the set of Synechococcus genomes) and characteristics of three gene categories: genes encoding linker polypeptides (Table 2), genes encoding putative phycobilin lyases (Table 3) and genes of unknown function specifically located in the PBS rod gene region and, therefore, potentially involved in PBS metabolism or regulation (Table 4).

The core-membrane linker L_CM, encoded by apcE, possesses three predicted repeat (or linker-like) domains in all marine Synechococcus except strains CC9311 and RS9916, in which L_CM has four such domains. RCC307 has the shortest L_CM sequence (953 amino acids) compared to the other strains due to shorter Arm2 and Arm3 regions (see 1541 for details on L_CM domains). Besides the PC-associated linker genes found in the rod gene region of both blue-green strains (Figure 4), WH5701 has a third cpcC homolog (cpcCIII) located elsewhere in the genome that potentially encodes a chimeric protein since it has an extended carboxyl terminus showing strong similarity to CpcD. None of the PE-containing strains possesses cpcC and cpcD homologs. In all marine Synechococcus genomes, the rod-core linker gene cpcGII is found in the PBS rod region whereas cpcGI is found outside this cluster. A third cpcG gene copy, which we refer to as cpcGIII, is present elsewhere in the genomes of BL107, CC9902, CC9311 and CC9605.

The total number of putative PE-associated linker genes varies from zero in the blue-green strains to six in the group constituted by the chromatic adapters and RCC307 (Table 2 and Figure 4). The location of the mpeE linker gene appears more variable than the other PEII genes, as it can be found either in the PBS rod gene region (for example, upstream of cpcGII in CC9311 or downstream of cpcGII in RS9916) or a few genes upstream of this region (in RCC307, BL107 and CC9902) or even in a totally different location of the genome (in CC9605).

Surprisingly, the PEII-lacking strain WH7805 possesses a homolog of mpeD, a gene known to encode a chimeric protein made of a PEII-associated linker (amino terminus) and a PEI-associated CpeD-like linker (carboxyl terminus) 19. However, closer examination of the amino-terminal part of this protein in WH7805 reveals a relatively low similarity with other MpeD sequences and a notable deletion of the region corresponding to amino acids 43-59 in Synechococcus sp. WH8102 19 that is conserved in all other MpeD sequences (Additional data file 2). This region includes two cysteinyl residues involved in linking a PUB chromophore via a Δ2,3 double bond, a type of chromophorylation typical of PEII-associated linker polypeptides 21. Synechococcus sp. WH7803 also lacks the mpeC gene, which encodes the distal PEII-associated linker polypeptide in other strains 1921. Finally, both chromatic adapters and RCC307 have, outside the PBS core region, an additional gene potentially encoding a PEII-associated linker (Table 3). In phylogenetic trees made with all PEII linkers (Additional data file 3), these sequences are both related to the amino terminus of MpeD but are split between two distinct gene clusters, one gathering BL107, CC9311 and CC9902, which we propose to name MpeF, and the other gathering RS9916 and RCC307, which we propose to name MpeG.

In order to compare further the linker composition of marine Synechococcus strains and determine whether they are all present in the PBSs, we performed a lithium dodecyl sulphate (LiDS)-PAGE analysis of intact PBSs. The Coomassie stained gel shown in Figure 5 displays the PBS proteins of two to three strains per pigment type. For WH7803 and RCC307, a Tris-tricine running buffer provided a better separation of the linker polypeptides than Tris-glycine (Figure 5, right). For strains WH5701, WH7805, WH7803, RCC307 and RS9916, all linker polypeptide bands (except ApcC and CpcD, which are not detectable under these electrophoresis conditions) were cut out from the gel and then identified by mass spectrometry (Table 2). In all five strains, the upper band proved to be the core-membrane linker L_CM, often accompanied by its degradation product L_CM', making a band of lower apparent molecular weight. As expected, RS9916, which has an extended apcE gene sequence, possesses the L_CM band of lowest electrophoretic mobility. Although the rod-core linker CpcGI was systematically present in all four strains, no CpcGII was detected by mass spectrometry, suggesting either that the cpcGII gene is expressed at a much lower level than cpcGII or that CpcGII is not present in the PBS fraction of these strains. It is worth noting though that we previously observed CpcGII (co-migrating with CpcGI) in a PBS fraction from Synechococcus sp. WH8102 19. Interestingly, we identified all three predicted PC rod linkers in WH5701, including the CpcCD-like protein, which is not found in the RS9917 genome. Furthermore, all PEII linkers predicted in WH7803, RCC307 and RS9916 were detected by mass spectrometry, except the products of the mpeF gene of RS9916 and of the mpeG gene of RCC307 (Table 1). This suggests that either these two potential linker genes are not expressed in our standard culture conditions or their products are undetectable on Coomassie-stained LiDS-PAGE gels due to some inherent biochemical properties.

Four types of phycobilin lyases, enzymes involved in the chromophorylation of phycobiliproteins, have been characterized so far. One of these, the heterodimeric CpcE/F complex, reversibly ligates a PCB molecule to Cys-84 of the α-subunit of C-PC 4243. Two genes with strong homology to the characterized cpcE and cpcF genes of Synechococcus spp. PCC 7942 44 and PCC 7002 45 are found near the 3'-end of the rod gene region in 7 out of the 11 marine Synechococcus genomes. We have called these cpcE-F in the two C-PC-containing strains (RS9917 and WH5