Correlation between cyanopeptolin gene clusters and peptides
The presence of two highly similar NRPS gene clusters (oci) in NIVA CYA 116 and NIES 205, and the production of nearly identical peptides by the two strains corroborate the association between the oci gene cluster and cyanopeptolin-1138 proposed by Rounge et al . This association is further substantiated by high degree of similarity to the cyanopeptolin gene cluster in Anabaena (apd), where the functional relationship between genes and peptides has been confirmed by a gene knock-out study  - as well as similarity to the Microcystis cyanopeptolin gene cluster (mcn) .
Global dispersal and distribution of cyanopeptolin genes
Based on the genomic regions studied here, two Planktothrix strains, NIVA CYA 116 and NIES 205, appear to be closely related despite the geographical separation. This is in accordance with the sequence comparison of 16S rDNA  identifying identical 16S rDNA sequences in Japan, China, The Netherlands, UK, Finland, Sweden and Norway, and thus may indicate a global distribution of closely related Planktothrix strains. Since Lake Årungen in Norway host international rowing competitions, a co-transport of this Planktothrix genotype with rowing equipment may be feasible. The data presented here do not allow any conclusions about global distribution without a more thorough analysis. The highly specific differences observed in the oci gene clusters are, independently of geographic distributions, intriguing. Our analyses indicate that the differences to some extent are due to positive selection at specific amino acid positions.
Variation in peptide content due to lack of specificity in the A-domains?
Previous studies have shown that lack of specificity in A-domains leads to activation of several amino acids with similar properties, thus giving rise to the synthesis of a series of related peptides from a single NRPS system . Ile/Leu/Val activating A-domains have been reported in lichenysin biosynthesis , and fengycin synthetase  among others. It is likely that the 116-Oci-A5- and A7-domains can activate Leu, Ile and Val and that the 116-Oci-A1- and A3-domains, that mainly activates Htyr, also can activate Ile and Leu. Consequently, 116-Oci is responsible for production of all seven cyanopeptolin detected in NIVA CYA 116 in this study. Likewise, 205-Oci probably is responsible for all oscillapeptin variants. The biological significance of a single NRPS complex giving rise to several peptide variants is yet to be determined.
Six of the seven binding pockets signatures of corresponding A-domains in NIES 205 and NIVA CYA 116 are identical (Table 1). If the different peptide profiles observed in the two strains are due to genetic differences in the NRPS genes, they are likely to be due to differences not involving the amino acids constituting the binding pocket signatures. LC-MS-MS-analyses were performed on strains cultivated on the same media, but we cannot completely exclude substrate availability as a contributory cause of variable peptide amount and peptide profile in the strain.
Module exchange and amino acid configuration
Over a stretch of total of 30 kb including the ABC transporter, the 116-oci and 205-oci gene clusters are remarkable similar, except for the modules encoding the T2-(E)-C2 domains. Too low sequence similarity is found between the whole T2-(E)-C2 modules in NIVA CYA 116 and NIES 205 to make a reliable alignment, suggesting that in one of these strains an entire module may have been exchanged through recombination. The E-domain trees (Figure 2) show a close relationship between cyanobacterial E-domains.
Sequence similarity to other E-domains and the distinctive flanking C (Figure 3) and T [Additional file 1 figure 5] domains observed by phylogenetic analysis indicate that the Oci-E-domain is an active epimerase, and are responsible for epimerization of Htyr to D-configuration. The configuration of the amino acids in cyanopeptolin-1138 were not determined however, a D-Htyr in oscillapeptin E and a putative L-Htyr in cyanopeptolin-1138 might explain the small difference between the oligopeptides with regard to polarity observed by HPLC analysis, as reported by Rounge et al. .
Interestingly, in the corresponding region of the Mcn cyanopeptolin synthetase in Microcystis the McnA-T1 and McnB-C2 include motifs suggesting association with an E-domain . In this case, however, no E-domain is present.
Sequence conservation and selection within cyanopeptolin modules
The two cyanopeptolin gene clusters (205-oci and 116-oci) are highly similar also at the third codon position. The first part (ABC-transporter, the spacer, GA-, T1-, S-, and C1-domains) and last part (C4-, A4-, T5-, C5-, A5-, T6-, C6-, M-, T7-, C7-, A7-, T8- and TE domains) of the Planktothrix cyanopeptolin gene cluster are nearly identical, despite the geographical distance separating the strains. Mechanisms for such sequence conservation may be frequent homology-driven genetic exchange within a genotype, leading to homogenization - in line with the general models suggested by Rudi et al. , Gogarten et al. and Papke et al. . Or alternatively sequence conservation may be due to low evolutionary rates caused by purifying selection or very short time of independent evolution.
Analysis of segregating sites and rates of nonsynonymous and synonymous nucleotide substitutions (Ka/Ks) indicate that module 3 (T3-, C3- and A3-domains) is different from the remaining domains by displaying higher substitution rates and signs of positive selection at several sites (Ka/Ks higher than 1). This is the module responsible for incorporation of the amino acid at position AA3 in the peptide.
According to data from Itou et al , a single amino acid replacement in the AA3 position of oscillapeptin E and F alters the protease inhibitory profile, indicating that this position could be pivotal for the inhibitory activity of cyanopeptolins. Positive selection in the third module could thus be expected to increase the adaptability of the inhibitory- or other putative functions of cyanopeptolin.