Phylogenetic relationships between toxic and non-toxic strains of the genus Microcystis based on 16S to 23S internal transcribed spacer sequence

Abstract

16S to 23S ribosomal DNA internal transcribed spacer sequences of 47 strains of the genus Microcystis were determined. Derived maximum likelihood and DNA distance trees indicated that Microcystis can be divided into three clusters. The first cluster included toxic and non-toxic strains, the second only toxic ones, and the third only non-toxic ones. The tree topologies were not necessarily correlated with morphospecies distinction or phycobilin pigment composition, and one genotype may have more than one morphotype. Phylogenetic analysis based on intergenic spacer sequences was thought to be effective for understanding relationships among closely related species and strains.

Cyanobacteria, Microcystis, Internal transcribed spacer, Phylogeny, Microcystin

1 Introduction

The cyanobacterium Microcystis is a water bloom-forming organism, which can produce the cyclic heptapeptide hepatotoxin, microcystin. This toxin causes a variety of human illnesses and is responsible for deaths in native and domestic animals [1]. There are several morphological types (morphotypes) in the genus, and each is equivalent to a species (morphospecies). Komárek [2] distinguished six morphospecies in Japanese waters: M. viridis (A. Brown) Lemmermann, M. wesenbergii (Komárek) Komárek in Kondratieva, M. aeruginosa (Kützing) Kützing, M. novacekii (Komárek) Compère, M. ichthyoblabe Kützing and M. flos-aquae (Wittrock) Kirchner. Watanabe [3] recognized all these morphospecies except M. flos-aquae, and regarded M. flos-aquae as small colonies of M. ichthyoblabe with a homogeneous cell arrangement. Watanabe [3] reported that M. viridis and M. aeruginosa produce microcystins, while M. wesenbergii (Japanese strains) and M. novacekii produce no microcystin, and M. ichthyoblabe includes toxic and non-toxic strains. However, a toxic strain of M. wesenbergii and a non-toxic strain of M. aeruginosa were included in a study by Neilan et al. [4]. Accordingly M. aeruginosa, M. ichthyoblabe and M. wesenbergii are known to include both toxic and non-toxic strains. Neilan et al. [4] performed a phylogenetic analysis of toxic and non-toxic Microcystis strains based on 16S rDNA sequences, and reported that 16S rDNA was useful for delineating toxic and non-toxic strains of Microcystis. Otsuka et al. [5] found, however, that five morphospecies of Microcystis were so closely related in terms of 16S rDNA sequence that they may be integrated into one species, and concluded that the 16S rDNA sequence is insufficiently variable for phylogenetic analysis of these organisms at the species level. DNA sequences of the 16S to 23S internal transcribed spacer region (16S–23S ITS) are known to be more variable and exhibit significant differences in sequence and length [6, 7]. It has been shown that the information obtained from analysis of this region is useful for species differentiation [6]. Previous studies on cyanobacterial rRNA operons revealed variation in length and sequence of the 16S–23S ITS using both restriction fragment length polymorphism (RFLP) on PCR amplification products and DAF (DNA amplification fingerprinting) [8–10], but the sequence analysis has not yet been examined. The aim of this study is to establish whether or not correlations exist between microcystin production, morphological characteristics and phyletic line, based on the 16S–23S ITS sequences of Microcystis strains.

2 Materials and methods

2.1 Cyanobacterial cultures

Details of the strains examined in the present study, which included five morphospecies, are shown in Table 1. Microcystis sp. 4A3, 4B3, T1-4, and T17-1 contain phycoerythrin [11]. The strains whose designations begin with NIES and TAC were obtained from the National Institute for Environmental Studies (Tsukuba, Japan) and Tsukuba Algal Collection in the Department of Botany, National Science Museum (Tsukuba, Japan), respectively. The other strains were isolated during the present study. M. flos-aquae was regarded as a form of M. ichthyoblabe according to Watanabe [3], except for those NIES strains where M. flos-aquae was regarded as a form of M. aeruginosa. Cultures were maintained in MA medium [12] at 20°C under a 12:12 h light/dark cycle with a photon flux density of about 30 µmol m⁻² s⁻¹ provided by daylight fluorescent lamps.

1

List of strains used in the present study

Strain	Origin	Microcystin	Phycoerythrin
Microcystis sp. 4A3	Hubei, P.R. China	−	+
Microcystis sp. 4B3	Hubei, P.R. China	−	+
Microcystis sp. T1-4	Bangkok, Thailand	−	+
Microcystis sp. T17-1	Chon Buri, Thailand	−	+
M. viridis TAC17	Ibaraki, Japan	+	−
M. viridis TAC44	Ibaraki, Japan	+	−
M. viridis TAC45	Ibaraki, Japan	+	−
M. viridis TAC78	Fukui, Japan	+	−
M. viridis TAC92	Hokkaido, Japan	+	−
M. viridis TAC93	Hokkaido, Japan	+	−
M. wesenbergii NIES104	Tokyo, Japan	−	−
M. wesenbergii NIES111	Ibaraki, Japan	−	−
M. wesenbergii TAC38	Ibaraki, Japan	+	−
M. wesenbergii TAC52	Nagano, Japan	−	−
M. wesenbergii TAC57	Nagano, Japan	−	−
M. wesenbergii CL5	Hubei, P.R. China	+	−
M. wesenbergii NC1	Shiga, Japan	−	−
M. wesenbergii NC 2	Shiga, Japan	−	−
M. wesenbergii NC 3	Shiga, Japan	−	−
M. wesenbergii NC 4	Shiga, Japan	−	−
M. wesenbergii NC 5	Shiga, Japan	−	−
M. aeruginosa NIES44	Ibaraki, Japan	−	−
M. aeruginosa TAC71	Ishikawa, Japan	+	−
M. aeruginosa TAC86	Shimane, Japan	+	−
M. aeruginosa TAC87	Shimane, Japan	+	−
M. aeruginosa TAC170	Tokyo, Japan	Unknown	−
M. aeruginosa CL1	Hubei, P.R. China	+	−
M. aeruginosa CL3	Hubei, P.R. China	+	−
M. aeruginosa CC1	Neinengo, P.R. China	−	−
M. aeruginosa T20-1	Nakhornpathom, Thailand	−	−
M. aeruginosa TC6	Bangkok, Thailand	−	−
M. aeruginosa TC8	Bangkok, Thailand	−	−
M. novacekii TAC20	Ibaraki, Japan	−	−
M. novacekii TAC65	Nagano, Japan	−	−
M. novacekii TAC66	Nagano, Japan	−	−
M. novacekii BC18	Scotland, United Kingdom	−	−
M. novacekii CC2	Hubei, P.R. China	−	−
M. novacekii T20-3	Nakhornpathom, Thailand	+	−
M. novacekii TL2	Chon Buri, Thailand	+	−
M. ichthyoblabe TAC48	Nagano, Japan	−	−
M. ichthyoblabe TAC91	Hokkaido, Japan	−	−
M. ichthyoblabe TAC125	Hokkaido, Japan	+	−
M. ichthyoblabe TAC136	Fukuoka, Japan	−	−
M. ichthyoblabe TAC138	Fukuoka, Japan	−	−
M. ichthyoblabe NL1	Ibaraki, Japan	−	−
M. ichthyoblabe TC2	Bangkok, Thailand	−	−
M. ichthyoblabe TC9	Bangkok, Thailand	−	−

Strain marked with ‘+’ produces microcystin(s) and one with ‘−’ does not. It was not confirmed that a product of M. aeruginosa TAC170 is a microcystin (Sano, T. and Kaya, K., unpublished data).

2.2 PCR amplification and sequencing

Primers for amplification of the DNA region including 16S–23S ITS were MSR-S2f (forward), 5′-TCAGGTTGCTTAACGACCTA-3′, corresponding to position 610–629 of Escherichia coli 16S rDNA numbering, and 242r (reverse), 5′-(G/T)TTCGCTCGCC(A/G)CTAC-3′, corresponding to position 257–242 of E. coli 23S rDNA numbering. The primers for sequencing were 1505f (forward), 5′-CCAGTGAAGTCGTAACAAGG-3′, corresponding to position 1486–1505 of E. coli 16S rDNA numbering, and 115r (reverse), 5′-GGGTT(T/G/C)CCCCATTCGG-3′, corresponding to position 130–115 of E. coli 23S rDNA numbering. The primer 1505f and the primers 242r and 115r were according to Neilan [13] and Lane [14], respectively. Primer MSR-S2f was designed in the present study to be Microcystis-specific based on a comparison of cyanobacterial 16S rDNA sequences available in the EMBL database. The specificity of this primer was not tested in vitro, but was checked using the BLAST [15] search of databases. All the primers were commercially synthesized by Nippon Flour Mills (Kanagawa, Japan). PCR template DNAs was extracted according to Palinska et al. [16]. The methods for DNA amplification and sequencing of the 16S–23S ITS region were the same as those described previously [5].

2.3 Alignment and phylogenetic analyses

The sequence alignment data were obtained using CLUSTAL W version 1.6 [17] and then converted to a distance matrix. A phylogenetic tree was reconstructed from the distance matrix using the neighbor-joining algorithm of CLUSTAL W version 1.6, with multiple substitutions corrected by the method of Jukes and Canter, and positions with gaps excluded, and the seed number for random number generator and numbers of bootstrap trials were set to 111 and 1000, respectively. Maximum likelihood analysis (ML) was carried out using the program package MOLPHY version 2.3b3 [18] based on the same alignment data after excluding positions with gaps. Maximum likelihood distance matrix was calculated using NucML, and the initial neighbor-joining tree was reconstructed by NJdist in the MOLPHY. The maximum likelihood tree was finally obtained using NucML with the R (local rearrangement search) option base on the HKY model, and local bootstrap probabilities (LBPs) were estimated by the resampling of estimated log-likelihood (RELL) method [19].

3 Results and discussion

With the exception of M. viridis, all morphospecies included both toxic and non-toxic strains. The existence of a toxic strain of M. novacekii is the first record for this species.

The PCR amplification products of both axenic and non-axenic strains were sequenced. The primers utilized in this study enabled sequencing of both strands of 16S–23S ITS. The length of this region was about 360 bp, 74 bp of which corresponded to the gene for tRNA-Ile. Nine strains belonging to three different morphospecies, M. aeruginosa, M. viridis and M. wesenbergii, shared identical sequences, as did the other four strains belonging to two different morphospecies, M. aeruginosa and M. wesenbergii (Fig. 1). Since the 16S–23S ITS sequence is more variable than 16S rDNA [6, 7], the 16S sequence of strains sharing 100% 16S–23S ITS sequence would identify, or extremely close. These strains might be one genotypic species (genospecies), but exhibiting more than one morphotype. Otsuka et al. [5] concluded that some strains belonging to different morphospecies may be integrated into one species, and Palinska et al. [16] mentioned that more ‘ecophenic’ and/or phenotypic forms have probably been described than genospecies exist. Both these views were supported by the present study.

1

(a) Maximum likelihood tree and (b) DNA distance tree of the Microcystis strains. Outgroups (Synechocystis sp. PCC6803 and Trichodesmium sp. NIBB1067) are omitted. Local bootstrap probabilities are indicated at nodes. Strains in each rectangle have an identical sequence. A strain marked with a triangle produces microcystin(s). It was not confirmed that a product of M. aeruginosa TAC170 is a microcystin. Accession numbers in the DDBJ, EMBL, and GenBank databases are Microcystis sp. 4A3, AB015357; 4B3, AB015358; T1-4, AB015359; T17-1, AB015360; M. viridis TAC17, AB015398; TAC44, AB015399; TAC45, AB015400; TAC78, AB015401; TAC92, AB015402; TAC93, AB015403; M. wesenbergii NIES104, AB015387; NIES111, AB015388; TAC38, AB015389; TAC52, AB015390; TAC57, AB015391; CL5, AB015392; NC1, AB015393; NC2, AB015394; NC3, AB015395; NC4, AB015396; NC5, AB015397; M. aeruginosa NIES44, AB015361; TAC71, AB015362; TAC86, AB015363; TAC87, AB015364; TAC170, AB015365; CL1, AB015381; CL3, AB015382; CC1, AB015383; T20-1, AB015384; TC6, AB015385; TC8, AB015386; M. novacekii TAC20, AB015374; TAC65, AB015375; TAC66, AB015376; BC18, AB015377; CC2, AB015378; T20-3, AB015379; TL2, AB015380; M. ichthyoblabe TAC48, AB015366; TAC91, AB015367; TAC125, AB015368; TAC136, AB015369; TAC138, AB015370; NL1, AB015371; TC2, AB015372; TC9, AB015373; Synechocystis sp. PCC6803, D90916; and Trichodesmium sp. NIBB1067, X72871.

The constructed phylogenetic trees, based on either a distance or ML algorithm, are shown in Fig. 1. Sequence variation of the 16S–23S ITS was found to be more variable than the corresponding 16S rDNA sequences. However, the similarity of sequences determined for this study was still extremely high (93.3–100% after excluding positions with gaps) which is the reason for the low bootstrap probabilities obtained for both trees. Although the tree topologies differed from one another, both trees had the following three clusters in common. Cluster I included all M. novacekii and M. ichthyoblabe strains and most M. aeruginosa strains, Cluster II included all M. viridis strains, two M. wesenbergii strains and one M. aeruginosa strain, and Cluster III included most M. wesenbergii strains, one M. aeruginosa strain and one phycoerythrin-containing strain, Microcystis sp. T17-1. The other strains containing phycoerythrin were included in Cluster I. The three phylogenetic clusters did not correspond to differences in morphological characteristics. All the strains of Cluster II were toxic, all the strains of Cluster III were non-toxic, and Cluster I included both toxic and non-toxic strains.

Watanabe [3] used the term ‘M. aeruginosa complex’ for M. aeruginosa, M. novacekii and M. ichthyoblabe, since the properties of these morphospecies are obscure. This complex generally corresponded to Cluster I. In the study presented here, M. viridis was the only morphospecies found to be monophyletic, while all other morphospecies were polyphyletic. M. aeruginosa strains were found in all clusters, which indicates that members of all phylogenetic lineages could potentially have morphological characteristics of the M. aeruginosa type. It was also shown that strains belonging to Cluster I had a higher level of sequence variation compared to those within the other cluster Kato et al. [20] reported that M. aeruginosa strains (equivalent to the M. aeruginosa complex in the present study) differed markedly at the allozyme level, which was supported by the present results. Kato et al. [20] also concluded, however, that M. viridis and M. wesenbergii are both well-established species based on allozyme tests. This view was not fully supported by the present study, since both of these morphospecies shared identical sequences with other morphospecies. Kato et al. [20] reported that M. wesenbergii TAC38, TAC52, and TAC57 shared the same morphospecies-specific genotype. Here, we showed that TAC38 had identical 16S–23S ITS sequence to all M. viridis strains, and was therefore included in a different cluster to TAC52 and TAC57. These results agreed with Otsuka et al. [11] who showed that 16S rDNA sequence of M. wesenbergii TAC38 was identical to M. viridis strains and not to M. wesenbergii TAC52 and TAC57. Although Clusters II and III both include more than one morphospecies, the two clusters seem to be homogeneous in terms of microcystin production. Among all strains, strains with identical sequences were homogeneous for toxin production, being either toxic or non-toxic. This result agrees with the report that a basic difference between toxic and non-toxic strains of M. aeruginosa is the presence of one or more genes coding for microcystin synthetases [21]. Toxin production was, however, polyphyletic in distribution from the trees reconstructed from 16S–23S ITS sequences, which could be seen also in the alignment (partially shown in Fig. 2). Furthermore, since some other cyanobacteria (e.g. Oscillatoria agardhii) produce microcystin, the production of this substance originated from a common ancestral cyanobacterium, and the observed heterogeneous distribution of the toxic and non-toxic strains may be the result of gene deletions at a number of times throughout evolution. Phycoerythrin-containing Microcystis strains were found in both Cluster I and Cluster III, and no clear division could be seen between strains with and without phycoerythrin. In the previous study [5], we indicated that the five known morphospecies and the phycoerythrin-containing strains may possibly be integrated into one species due to the high sequence similarities of the 16S rDNA. Difference in phycobilin pigment composition may be regarded as an intra-species-specific variation.

2

16S–23S ITS partial (5′-end) alignment of selected strains of Microcystis in the present study. The upper sequences correspond to positions 1–60 of the consensus sequence, and the lower to positions 61–120. Characters representing nucleotide bases are according to IUPAC codes, and ‘*’ denotes a gap and ‘–’ denotes the same base as the consensus sequence of the eight strains. Boxed characters are those bases which were variable. Grey shaded strains are the toxic strains.

In conclusion, the validity of the current taxonomy of Microcystis chiefly based on morphological characteristics was not supported by phylogenetic analysis presented here based on 16S–23S ITS sequence comparisons. Phycoerythrin-containing Microcystis is polyphyletic and the difference in phycobilin pigment composition can be regarded as an intra-species-specific variation. The present result indicated that the basic difference between toxic and non-toxic strains of Microcystis has a genetic background.

Acknowledgements

This work was supported by Science and Technology Agency, Japan.