Received 25 February 2016; accepted 24 April 2016; published 27 April 2016
Eutrophic and hypertrophic freshwater bodies worldwide are habitable environments for cyanobacteria  . As the unicellular cyanobacterium, Microcystis can be collected comparatively easily from lakes and ponds. It has been proposed as a model for studying the ecological patterning in free-living bacteria  . Ecological patterning may affect the concept of species through historical and/or contemporary environmental processes such as local selection, indicating that there are such ecologically distinct populations based on the Stable Ecotype Model   . To investigate the concept of bacterial ecological species, the bacterial ribosomal RNA gene region is utilized  . However, it is difficult to differentiate distinct ecotypes of Microcystis according to the internal transcribed spacer sequence of 16S-23S rRNA  .
Whole genome sequences of 12 genera of cyanobacteria have been decoded to date. The entire and draft genome sequences of Microcystis aeruginosa strains NIES-843  and PCC 7806  , respectively, have been established. Notably, 11.8% of the genome of NIES-843 was occupied by mobile elements: insertion sequences and miniature inverted-repeat transposable elements  . This indicates that rearrangements of the genome, including the deletion and mutation of genes, have frequently occurred in Microcystis. Moreover, the genus is known to produce natural products via secondary metabolism. The gene for a cyclic heptapeptide microcystin (mcy) whose production involves nonribosomal peptide synthetase (NRPS) and polyketide synthase was identified in M. aeruginosa K-139   . A highly conserved mcy gene cluster was confirmed among the Microcystis  -  . Moreover, there were hotspots for the insertion and deletion of a fragment between a non-coding region of the ORF gene and the area adjacent to the cyclic heptapeptide microcystin and micropeptin gene (mcn) clusters among the genus    .
We cloned and sequenced the genes mcnABCE, responsible for producing the heptadepsipeptide micropeptin belonging to the cyanopeptolin class, in M. aeruginosa K-139 and demonstrated that NRPS is involved in micropeptin production by means of mcn gene-knockout  . In addition, according to a comparative analysis of the mcn genes, we revealed the loss of a recombination event in relation to a halogenase gene, designated mcnD, within the mcn gene cluster as well as proposed a clear evolutional history for the absence of a transcriptional region in mcnD. Therefore, we concluded that the mcnD gene does not contribute to the production of non-ha- logenated MCNs  . Regarding NRPS-related halogenases, Cadel-Six et al.  proposed that the ancestral Microcystis had a mcn gene cluster containing a halogenase that was subsequently lost. Strain PCC 9812, which was isolated in Lake Mendota, United States, produces non-halogenated-MCN, but has an intact mcnD  . On the other hand, non-halogenated-MCN producing Microcystis strains, K-139, NIES-102, NIES-103, and S-70, which were isolated in Lake Kasumigaura, Japan, and NIES-90, which was isolated from Lake Kawaguchi, Japan, were also identified (Table 1). Although we revealed that strain B-35, which strain possesses an intact mcnD, retained the mcnA, mcnB, mcnC, and mcnE genes according to genomic Southern hybridization analysis  , we have not identified the MCN compound in strain B-35 to date (T. Nishizawa and M. Shirai, unpublished data). To identify the transcription of the mcn gene, a reverse transcribed-PCR analysis was carried out. For the reverse transcribed-PCR analysis, cells in a logarithmic growth phase were harvested, total RNA was isolated from M. aeruginosa B-35 and K-139 using hot phenol, and cDNA was prepared as described previously  . The reverse transcribed reaction was performed as described  . The primers sets for the mcnA and mcnD genes were F-mcnA-RT (5’-CGCCCAAAAATGTCACC-3’) and R-mcnA (5’-AAGGGGAAATCTTGGGC-3’) based on the mcnAK-139 gene (accession number, AB481215) and F-hal-RT (5’-GGCGAATCAATCTTTACATCG-3’) and R-hal-RT (5’-TCACTTACCAATTGCCTC-3’) based on the mcnDB-35 gene (AB481216), respectively. PCR (30 µL) was performed under the following condition; 3 min at 95˚C, then 25 cycles of 95˚C (30 s), 58˚C (30 s), and 72˚C (45 s), and 72˚C (2 min). In this study, no transcription of mcnA and mcnD in strain B-35 was found though the transcription of a NRPS-related microcystin biosynthetic gene was observed (data not shown). From these results and our previous investigation of genomic Southern hybridization  , it was suggested that the mcn gene cluster of strain B-35 may be pseudogenes. Therefore, strain B-35 is the first Microcystis strain which does not produce MCN although it has the mcn gene cluster including the halogenase gene.
In this study, a detailed comparison of the mcn-related halogenase gene of Microcystis species was carried out. According to a comparative analysis of strains K-139 and B-35 based on mcnC and mcnE, we propose that a loss of three DNA fragments occurred through recombination in a mosaic-like manner and then a nontranslated region containing a part of mcnD arose in strain K-139  . To examine the deletion of the intact mcnD gene among the genus Microcystis, we selected 22 sequences of mcnC-mcnE from the MCN-producing Microcystis strains (Table 1). The acceleration of the rate of nucleotide substitutions in the mcnC-mcnE gene region among the genus Microcystis was calculated based on a phylogenetic analysis. Nucleotide sequences of the mcnC-E region were aligned using Clustal W  . The bootstrap test was performed on 500 replicates. A neighbor-joining (NJ) tree was constructed  and evolutionary distances were computed using the Tajima’s test in MEGA 4
Table 1. Summary of cyanopeptolin-producing Microcystis and Anabaena strains.
aIsolated in Lake Kasumigaura. bIsolated in Lake Kawaguchi. cn.d. not determined. dChlorinated cyanopeptolin.
program  . When a phylogenetic analysis of the 92-bp sequence (downstream non-coding region) using 21 strains excluding strain NIVA-CYA 172/5  was conducted with a non-coding region between downstream of apdC involved in halogenation of the depsipeptide anabaenopeptilide and upstream of adpD in the heterocyst Anabaena sp. strain 90  as an outgroup, these 92-bp sequences formed one group (data not shown). Then, a phylogenetic analysis was performed based on the non-coding region of 170-bp, which is mosaic sequence traces, between mcnC and mcnE except for the 92-bp fragment in strain K-139  . In this phylogenetic analysis, the apdC gene was used as an outgroup. The phylogenetic relationship shown in Figure 1(a) clearly divided into two clades, mosaic sequence traces (clade I) and the intact mcnD gene (clade II). Interestingly, strain K-139 placed outside of clade I (Figure 1(a)). Except for strain K-139, the relative rate of substitution was examined at a number of sites (170-bp) using the shortest branch of strain PCC 9622 in clade I and the longest branch of strain B-35, which has a 54-bp insertion sequence upstream of mcnD  , in clade II. The number of nucleotide substitutions per site was 0.112:0.018 (6-folds, χ2 = 11.64, P < 0.001). When strain K-139 was used instead of strain PCC 9622, the number was 0.064:0.035 (χ2 = 1.47, P = 0.23). Otherwise, excluding the outgroup of adpC, we estimate the average evolutionary divergence of the 170-bp and 92-bp sequence pairs to be 0.097 and 0.045, respectively. These investigations indicate an acceleration of the rate of nucleotide substitutions in clade I. Moreover, the rate increased in the truncated mcnD gene region rather than the non-coding region (92-bp) between mcnD and mcnE.
Figure 1. Phylogenetic relationship determined using the neighbor-joining method (a) and a splits decomposition analysis (b) of the mcnC-E region of Microcystis strains. Only bootstrap values exceeding 70% are shown. There were a total of 170 positions in the final dataset.
To determine the genetic relationship of strain K-139 to both clades, a splits decomposition analysis was carried out using the 170-bp sequence. Recombination was investigated by a split decomposition analysis using SplitsTree Version 4  with default settings (uncorrected P method) and 500 bootstrap replicates. As shown in Figure 1(b), the 170-bp non-coding region of strain K-139 placed in between the two groups. Phylogenetic analyses revealed that the truncated mcnD gene of strain K-139 is a unique sequence, indicating the strain to have a unique non-halogenated-MCN genotype. In addition, lower divergence of the 170-bp non-coding region was found within the non-halogenated-MCN genotypes isolated in Lake Kasumigaura and Lake Kawaguchi, Japan except strain B-47 in group I (Figure 1(b)). Our results indicate that the region between mcnC and mcnE is the evolutionary hotspot among non-halogenated MCN-producing Microcystis.
Conclusively, the diversified sequences of the intact and truncated mcnD gene may imply that halogenated MCN- and non-halogenated MCN-producing Microcystis occur individually. In the present study, the existence of a truncated mcnD-possessing Microcystis peculiar to Lake Kasumigaura was showed and it is assumed that truncated mcnD such as strain K-139 is purged. Our results show new insight into the ecological patterning of widespread Microcystis species. Further isolation of mcy and mcn-possessing Microcystis strains is needed to elucidate the ecological species of Microcystis that have adapted genetically to local ecosystems.
We thank Dr. Takashi Kitano (Ibaraki University College of Engineering) for critical comments about evolutionary divergence and valuable discussions. We thank Dr. Akito Nishizawa of the laboratory of M.S. for technical assistance.