Ecological Dynamics of Broad- and Narrow-Host-Range Viruses Infecting the Bloom-Forming Toxic Cyanobacterium Microcystis aeruginosa

ABSTRACT

Microcystis aeruginosa is predicted to interact and coexist with diverse broad- and narrow-host-range viruses within a bloom; however, little is known about their affects on Microcystis population dynamics. Here, we developed a real-time PCR assay for the quantification of these viruses that have different host ranges. During the sampling period, total Microcystis abundance showed two peaks in May and August with a temporary decrease in June. The Microcystis population is largely divided into three phylotypes based on internal transcribed sequences (ITS; ITS types I to III). ITS I was the dominant phylotype (66% to 88%) except in June. Although the ITS II and III phylotypes were mostly less abundant, these phylotypes temporarily increased to approximately equivalent abundances of the ITS I population in June. During the same sampling period, the abundances of the broad-host-range virus MVGF_NODE331 increased from April to May and from July to October with a temporary decrease in June, in which its dynamics were in proportion to the increase of total Microcystis abundances regardless of changes in host ITS population composition. In contrast, the narrow-host-range viruses MVG_NODE620 and Ma-LMM01 were considerably less abundant than the broad-host-range virus and generally did not fluctuate in the environment. Considering that M. aeruginosa could increase the abundance and sustain the bloom under the prevalence of the broad-host-range virus, host abundant and diverse antiviral mechanisms might contribute to coexistence with its viruses.

IMPORTANCE The bloom-forming toxic cyanobacterium Microcystis aeruginosa interacts with diverse broad- and narrow-host-range viruses. However, the dynamics of the Microcystis population (at the intraspecies level) and viruses with different host ranges remain unknown. Our real-time PCR assays unveiled that the broad-host-range virus gradually increased in abundance over the sampling period, in proportion to the increase in total Microcystis abundance regardless of changes in genotypic composition. The narrow-host-range viruses were considerably less abundant than the broad-host-range virus and did not generally fluctuate in the environment. The expansion and maintenance of the Microcystis bloom even under the increased infection by the broad-host-range virus suggested that highly abundant and diverse antiviral mechanisms allowed them to coexist with viruses under selective pressure. This paper expands our knowledge about the ecological dynamics of Microcystis viruses and provides potential insights into their coexistence with their host.

INTRODUCTION

Microcystis aeruginosa is one of the most pervasive and harmful cyanobacteria that can accumulate at the surface of the water, leading to the development of massive dense blooms in eutrophic freshwater ecosystems (1). Some Microcystis strains can produce the potent hepatotoxin microcystin (2–4), and thus their toxic blooms have caused fatal accidents for livestock, pets, and wildlife (5). Microcystis toxicity also poses serious problems for humans who use impaired water resources for dialysis treatment (6). Therefore, the World Health Organization proposes a drinking water guideline of 1 μg/L for microcystin (7). Thus, it is a global issue for maintaining safe water supplies to identify and characterize the environmental and biological factors that affect Microcystis abundance and composition (8).

One of the most unique features of M. aeruginosa is that they possess highly abundant and diverse antiviral defense systems in the genome (9–11). Microcystis CRISPR-based studies have revealed that blooms are composed of diverse populations with different CRISPR spacers acquired from foreign DNA fragments (12–14), suggesting that M. aeruginosa has frequently interacted with diverse viruses in the environment. Consistent with this observation, the number of Microcystis viral isolates has been gradually increasing since Ma-LMM01 was first isolated (e.g., MaMV-DC, PhiMa05, Mic1, Me-ZS1, ΦMHI42, Mwe-Yong1112-1, and Mae-Yong1326-1) (15–22). In addition to these isolates, we recently revealed the existence of Microcystis viruses classified into groups I (including Ma-LMM01), II, and III through a metagenomic approach (23). Based on the phylogenetic distribution of host strains with spacers against each group, Microcystis group II viruses are predicted to have a broad host range, whereas the group I and III viruses have a narrow host range (23). Indeed, the narrow host ranges of Ma-LMM01 and newly isolated Mic1 (15, 18, 24), each of which corresponds to the group I and III virus, corroborate the validity of our in silico predictions, although there was no direct evidence of group II viral host range because of the lack of corresponding viral isolates. Furthermore, it was recently shown that a novel siphovirus with high similarity to prophage-like sequences in M. aeruginosa NIES-88 existed in the environment (25).

Among these Microcystis viruses, the ecological dynamics of Ma-LMM01 have been intensively studied by quantification of gp091 (tail sheath) abundance (26). For example, previous studies revealed that a clear increase in Ma-LMM01 abundance was observed at the same time as total Microcystis abundance decreased (27) and that Ma-LMM01 infection was likely to occur in 0.01 to 2.9 cells/mL of the natural Microcystis population (28). The latter observation reflects the narrow host range of Ma-LMM01; this virus might have the potential to infect only a small part of the Microcystis population in the natural environment. Furthermore, the metatranscriptome approach in Lake Tai indicated the negative correlation between gp091 and lysogenic-associated genes, creating a debate on the existence of a lysogenic state for Ma-LMM01 (29). In contrast, our knowledge about the ecological dynamics of other phylogenetically different Microcystis viruses is limited to only a few studies conducted in Lake Erie and Lake Taihu (25, 30). Of these, the latter study based on RNA polymerase sequence (rpoB) demonstrated that Microcystis myovirus and siphovirus were positively correlated with host populations, including several morphospecies, that emerged in the early and late bloom, respectively (25). However, little is known about the relationships between the ecological dynamics of Microcystis population at the intraspecies level and those of its viruses with different host ranges.

In this study, we developed a real-time quantitative PCR (qPCR) assay to monitor the environmental dynamics of broad- and narrow-host-range Microcystis viruses. Furthermore, we investigated Microcystis population dynamics based on ITS type to assess the relevance of each viral type dynamics on the bloom during the same sampling period.

RESULTS

Seasonal changes in total and ITS phylotype abundances of M. aeruginosa.

Dense Microcystis blooms have occurred at Hirosawanoike Pond every year because of the eutrophication caused by nutrient influx from agricultural fields (28, 31, 32). We first investigated the total abundance of M. aeruginosa from May to October 2016. The phycocyanin intergenic spacer (PC-IGS) gene copy numbers of M. aeruginosa significantly increased from 4.64 × 10⁴ copies/mL in May to 1.51 × 10⁷ copies/mL in August (P value, 0.021 by Tukey-Kramer test) and then decreased in October (1.39 × 10⁶ copies/mL) (P value, 0.036; Fig. 1). Thus, the intensity of the Microcystis bloom increased from early summer to autumn at Hirosawanoike Pond.

FIG 1.

Seasonal changes in total abundances of Microcystis aeruginosa in the environment. Total abundance of M. aeruginosa was determined by quantitative PCR analysis. Error bars indicate standard deviation. *, Significant differences in the abundance compared with that in August (P < 0.05 by Tukey-Kramer test): in May (P = 0.021), June (P = 0.021), July (P = 0.024), and October (P = 0.036).

We further conducted Microcystis ITS amplicon analysis to monitor seasonal changes in their abundances at the intraspecies level in the environment. Our sampling of the Microcystis bloom generated a total of five amplicon samples from the Hirosawanoike Pond (Table S1 in the supplemental material). A total of 1,169,183 paired reads were obtained from these environmental samples (Table S1). After the quality control processes, an average of 183,596 merged reads were used for Microcystis ITS amplicon analysis (Table S1). On average, 105,762 Microcystis ITS sequences were extracted by using the ITS database derived from 159 Microcystis isolates (Table S1). These ITS amplicon sequences were largely divided into three groups (ITS types I to III), in accordance with the ITS phylotyping of Microcystis strains in previous studies (data not shown) (13, 33).

The Microcystis ITS I phylotype (14 operational taxonomic units [OTUs]) was the dominant phylotype in the environment (67 to 88% of total abundances) over the sampling period except in June (Fig. 2). The Microcystis ITS II phylotype (2 OTUs) accounted for 20 to 34% of total abundances from May to July and then became less abundant (0.4 to 2.9%) from August to October (Fig. 2). The Microcystis ITS III phylotype (5 OTUs) was present at a low abundance (3.2 to 13%) over the sampling period; however, its abundance rapidly increased up to 39% and 28% of total abundances in June and August, respectively (Fig. 2). Also, the standard deviation in the number of each ITS type reads per total assigned reads was lowest in June (Fig. 2).

FIG 2.

Temporal changes in abundances of Microcystis populations based on internal transcribed spacer (ITS) type in the environment. The percentages of Microcystis ITS I to III populations are shown in green, pink, and yellow, respectively. The top represents the standard deviations in the number of each ITS type reads per total assigned reads.

Design of real-time PCR primers for Microcystis viruses with different host ranges.

To investigate the relationships between the dynamics of M. aeruginosa and its broad-host-range viruses, we next identified the most abundant virus with a broad host range through the recruitment of virome reads collected in a previous study (23). Relative abundances of 4 broad-host-range viruses and 13 narrow-host-range viruses including Ma-LMM01 and MaMV-DC ranged from 31.5 to 50.1 and from 13.1 to 60.6 fragments per kilobase of exon per million mapped reads (FPKM) values, respectively (Fig. S1). We selected MVGF_NODE331 possessing the fourth most abundant spacers (23) as a target broad-host-range virus that was the most abundant in the environment. The narrow-host-range virus Ma-LMM01 was used for comparison, for which the quantification method was already established. This virus was the least abundant in the environment. Hence, we also selected MVGF_NODE620 as the most abundant target virus with a narrow host range.

Thereafter, the specific real-time PCR primer sets were designed to detect and quantify the broad-host-range virus MVGF_NODE331 and narrow-host-range virus MVGF_NODE620, respectively. For MVGF_NODE331 quantification, we checked the nucleotide diversity of the terminase gene through the recruitment of environmental sequences. As a result, we found a single nucleotide polymorphism (SNP) located at 608 bp despite its low genetic diversity (Fig. 3). Therefore, we used the degenerate primers including the SNP region in the following experiments to comprehensively quantify MVGF_NODE331 and its close relatives (Table 1). We confirmed the SNP region in the terminase gene products by Sanger sequencing peak. Meanwhile, no nucleotide polymorphisms were included in the primer region of the PadR family transcriptional regulator for MVGF_NODE620 quantification (Fig. 3 and Table 1). A BLAST search with the primer sequences showed no genomic cross-reactivity with other viruses in the current database as well as the assembled viral database from the Hirosawanoike Pond (23). Both primer pairs could amplify a single specific product, which was confirmed by gel electrophoresis (Fig. S2) and Sanger sequencing (data not shown). The amplification efficiencies of the qPCR assay targeting for the broad- and narrow-host-range viruses were 110% and 112%, respectively (R² ≥ 0.98; Fig. S3).

FIG 3.

Virome read mapping pattern and nucleotide polymorphisms of the targeted Microcystis viral genomic fragments. (A) MVGF_NODE331. (B) MVGF_NODE620. Each viral names were designated in a previous study (23). The top and bottom of each virus represent the read mapping patterns derived from the sample collected at 06:00 and 18:00 in the previous study (23). The visualization and normalization of read abundances were conducted for each targeted virus by the Integrative Genomics Viewer function (43). Red rectangles indicate the position of target region in each gene.

TABLE 1.

Primer sets for sequencing and real-time PCR assay used in this study

Primer	Sequence (5′ to 3′)a	References
188F	GCTACTTCGACCGCGCC	Yoshida et al. (45)
254R	TCCTACGGTTTAATTGAGACTAGCC	Yoshida et al. (45)
MITSamp-F	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGAAGGGAGACCTAATTCVGGT	Yoshida et al. (38)
MITSamp-R	GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGTTGCGGTCYTCTTTTTTGGC	Yoshida et al. (38)
g91DF1	CTGGGGTAATCAAGTTA	Kimura et al. (44)
g91DR3	CGGGTGGRGTTRMAYCYRCG	Kimura et al. (44)
MVGF331-F	GAGAAAGACTGGGAGCGATC	This study
MVGF331-R	ACCAYGCCTCAACCTTATCG	This study
MVGF620-F	CTTTTTCTAGCACCGCCTCT	This study
MVGF620-R	ATCCTTGTTTTGCAGGCTCT	This study

^aUnderline indicates overhang adapter sequence according to an Illumina MiSeq amplicon sequencing protocol.

Changes in abundance of Microcystis viruses with different host ranges.

Finally, we investigated the ecological dynamics of each Microcystis virus using viral fractions collected during the same sampling period to reveal differences in the relationships with total abundance and intraspecies population composition of M. aeruginosa. The broad-host-range virus MVGF_NODE331 decreased from 4.67 × 10⁵ copies/mL in May to 3.19 × 10³ copies/mL in June and then gradually increased up to 2.82 × 10⁵ copies/mL in October (Fig. 4). The narrow-host-range virus MVGF_NODE620 did not markedly fluctuate over the sampling period (9.23 × 10² to 4.04 × 10³ copies/mL) (Fig. 4). Likewise, Microcystis virus Ma-LMM01 with a narrow host range did not change in abundance and showed lower abundance than that of MVGF_NODE620 in the environment (1.28 × 10² to 6.34 × 10² copies/mL) (Fig. 4), which is consistent with FPKM value described above. Thus, the broad-host-range virus tended to be more dominant than the other two narrow-host-range viruses at the Hirosawanoike Pond over the sampling period except for June (Fig. 4). Meanwhile, an analysis of variance (ANOVA) test detected significant differences (P value, 0.034), but the Tukey-Kramer test for multiple comparisons did not show a significant difference among each data.

FIG 4.

Temporal changes in abundances of the targeted Microcystis viruses including Ma-LMM01 in the environment. Ecological dynamics of MVGF_NODE331, MVGF_NODE620, and Ma-LMM01 are shown in red, green, and black, respectively. The y axis represents the average copy numbers of each target gene in triplicate samples (except for the samples shown by open symbols, n = 2). Error bars indicate standard deviation.

DISCUSSION

This is the first report focusing on the relationship between the ecological dynamics of viruses and changes in total host abundance and population composition in terms of assumed host range differences. Our results using real-time PCR assays revealed that the abundance of the broad-host-range virus MVGF_NODE331 gradually increased over the sampling period, in proportion to the increase in total Microcystis abundance (Fig. 1 and 4). At this time, as supported by the standard deviation, the dominant ITS type of the Microcystis population temporarily declined to approximately equal rates to those of the other ITS type populations in June (Fig. 2); however, MVGF_NODE331 maintained its abundance in the environment at the same level as that in July (Fig. 4). Thus, the ecological dynamics of MVGF_NODE331 coincided with that of the whole host community regardless of the population composition, suggesting that broad-host-range viruses can proliferate in diverse host strains in the environment.

In contrast, both narrow-host-range viruses MVGF_NODE620 and Ma-LMM01 showed markedly lower abundance than that of the broad-host-range virus (Fig. 4). Also, these narrow-host-range viruses did not markedly fluctuate over the sampling period even though the total Microcystis abundance increased and the host population composition largely changed in June (Fig. 1, 2, and 4). However, previous studies indicated that these narrow-host-range viruses could rapidly increase or become temporarily dominant in the environment. For example, Microcystis virus Ma-LMM01 rapidly increased up to 1.1 × 10⁴ copies/mL in Lake Mikata (27). Likewise, the narrow-host-range virus MVGF_NODE620 showed the highest relative abundance at Hirosawanoike Pond in 2017 (Fig. S1) (23). Thus, the narrow-host-range viruses are thought to be prevalent if the specific host strain becomes dominant and thereby proliferates in the cells. This insight is also partly supported by the data that narrow-host-range viruses exhibited a broader range of FPKM values than that of the broad-host-range viruses (Fig. S1).

The results obtained in this study further indicated that M. aeruginosa could increase abundance and sustain dense bloom even under the increase of infection by the broad-host-range virus (Fig. 1 and 4). One of the reasonable explanations for the observations is that highly abundant and diverse antiviral mechanisms confer resistance to coexistence with its viruses under a negative frequency-dependent selection by viral infection. Indeed, Microcystis populations with different CRISPR spacers oscillated during the bloom, suggesting that this antiviral mechanism is one of the key factors in maintaining the population size under selective pressure (31). Likewise, it was proposed that nutrient loading and virion accidental methylation drive the excessive accumulation of restriction-modification (RM) system in the Microcystis genome, thereby allowing them with different sets of RM systems to coexist under the increased viral contact rate (11). The observation that Microcystis antiviral genes were expressed at the time of viral multiplication in the environment (23) also supported the importance of diverse antiviral mechanisms for the coexistence with viruses.

In conclusion, our data sets revealed differences in ecological dynamics between broad- and narrow-host-range viruses when the total abundance and population composition of M. aeruginosa were changed. These results will help our understanding of the coexistence of M. aeruginosa with viruses with different host ranges and provide insight into the potential importance of highly abundant and diverse antiviral mechanisms for expanding and maintaining the bloom.

MATERIALS AND METHODS

Sampling.

Seasonal freshwater samples were collected from the surface at an offshore point at Hirosawanoike Pond (35° 026′ N, 135° 690′ E) once per month from May to October 2016 except for September. For the amplicon analysis, the samples (100 mL) were gently sonicated followed by centrifugation at 3,000 rpm for 10 min. Cell pellets were stored at −20°C until DNA extraction. For quantification of viral abundance for each virus, 1.5 L of freshwater samples was prefiltered through a 142-mm 3.0-μm pore-size polycarbonate membrane filter (Millipore) and then filtered through 0.22-μm pore-size polycarbonate membrane filters or Sterivex units (Millipore). The filtrates (15 mL) were ultracentrifuged at 106,978 × g for 1.5 h, followed by resuspension in 200 μL of virus-free water and storage at −80°C.

Amplicon library preparation and sequencing.

Genomic DNA extraction from the cell pellets was performed using a combination of the potassium xanthogenate-sodium dodecyl sulfate and phenol-chloroform/isoamyl alcohol procedures as described previously (34, 35). The extracted DNA was used for the following PCR as a template. Microcystis ITS region was amplified using specific primer sets MITS-F and MITS-R designed in a previous study (36) with 5′-end overhang adapter sequences according to an Illumina MiSeq amplicon sequencing protocol (Table 1). The resulting amplicons were purified using Agencourt AMPure XP beads (Beckman Coulter) according to the manufacturer’s instructions. Amplicon libraries were sequenced using a MiSeq reagent kit v3 (2 × 300-bp read length; Illumina).

Amplicon data processing.

The adaptor sequences of the raw sequence reads were automatically removed by the Illumina MiSeq system, whereas the primer region was trimmed by VSEARCH 2.9.1 (37). The quality of each raw sequence file was checked using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Trimmomatic v0.38 (38) was used to remove ambiguous and low-quality reads by scanning the reads with a 50-base-wide sliding window (quality score threshold = 20). The trimmed reads were merged using VSEARCH 2.9.1 with default parameter (37). After that, we removed the merged reads that were shorter than 200 bp, included “N” nucleotides, or showed quality scores <20. Chimera sequences were also removed using VSEARCH 2.9.1 (37). The remaining merged reads were clustered into OTUs using VSEARCH 2.9.1 with a sequence identity threshold of 100% (37), and then singleton OTUs were discarded at this stage. The representative sequences of the remaining OTUs were assigned to the ITS database collected from 159 Microcystis isolates in the current database including a previous study (13) (e-value threshold, 1.0 × 10⁻⁵). The assigned Microcystis ITS sequences were aligned using the MAFFT program version 8.2.12 (39) and trimmed with trimAl version 1.4.1 (40). A maximum likelihood (ML) phylogenetic tree was constructed using RAxML version 8.2.12 (41). Bootstrap resamplings were conducted with 100 replications in the ML phylogenetic tree construction. The relative abundance of each Microcystis ITS type was calculated as a percentage of total Microcystis ITS sequences in a sample.

Selection of the target viruses.

For selection of the target viruses, we referred to the abundances of each Microcystis viruses revealed through a metagenomic approach in a previous study (23). Briefly, the quality-controlled reads were aligned to the metagenome-assembled Microcystis viral genomes (23) including Microcystis virus Ma-LMM01 using Bowtie 2 (42) with the option “-score-min L,0,-0.6.” The virome read abundances were normalized as fragments per kilobase of exon per million mapped reads (FPKM). We selected the most abundant broad- and narrow-host-range viruses as the target viruses (23).

Primer design and PCR amplification.

The specific primers for the broad- and narrow-host-range viruses were designed to amplify a 133-bp region of the terminase coding gene and 194-bp region of PadR family transcriptional regulator coding gene, respectively (Table 1). These genes are not shared within the same Microcystis viral group. Also, we mapped virome reads collected at 18:00 on 19 October and 06:00 on 20 October 2017 in a previous study onto the target viral genomes under the same conditions as described above, visualized the read mapping patterns with the Integrative Genomics Viewer (43), and then the degenerate primer sets were designed based on the environmental sequences of primer regions (Table 1). Thereafter, these sequences were queried against the National Center for Biotechnology Information (NCBI) nonredundant (nr) and the previously assembled viral contig (≥10 kb) database (23) to ensure there were no nonspecific matches with other viruses.

Viral DNA was extracted from the filtered samples using the previously described xanthogenate-SDS method (34). PCR amplification with the primer sets was performed using TaKaRa PCR Thermal Cycler Dice Touch (TaKaRa Bio Inc.). The PCR mixture contained 12.5 μL of GoTaq Green Master Mix (Promega), 0.4 μM each primer, 1 μL of the extracted DNA, and virus-free water. Each PCR amplification consisted of an initial denaturation at 94°C for 1 min, followed by 30 cycles: 94°C for 30 s, 60°C for 30 s, 78°C for 30 s; with a final extension at 72°C for 5 min. After purification using the Wizard SV Gel and PCR Clean-Up System (Promega), the dilution series of PCR products were used as standard samples in the real-time PCR assay. The obtained products were confirmed to be a single band of the appropriate size by 2% gel electrophoresis (Fig. S2) and validated by the Applied Biosystems 3130 Genetic Analyzer (Thermo Fisher Scientific) (data not shown).

Quantification of the predicted broad- and narrow-host-range viruses.

We conducted a real-time PCR assay with the designed primer sets to quantify the abundances of each target virus in the Hirosawanoike Pond from April to October 2016. The degenerate primer sets g91DF1 and g91 DR3 were also used for the quantification of Microcystis virus Ma-LMM01 (44), whereas the primer sets for phycocyanin intergenic spacer (PC-IGS) were used for the quantification of Microcystis total abundance (32, 45) (Table 1). Each real-time PCR mixture contained 12.5 μL of TB Green Premix Ex Taq II (TaKaRa), 0.4 μM each primer, 2 μL of the extracted DNA, and virus-free water. These assays were performed using the Thermal Cycler Dice Real-Time System (TaKaRa) under the same amplification conditions as described above. A standard curve for each target virus was used for quantification and plotted using threshold cycle values. We confirmed a positive log-linear correlation between the C_T values and serial dilutions of each standard sample (terminase; R² = 0.990, PadR family transcriptional regulator; R² = 0.983) (Fig. S3). The specificity of each amplified standard sample was also verified by the dissociation curves. The detection ranges of terminase and PadR family transcriptional regulator sequences were approximately 1.0 × 10² to 10⁷ and 1.0 × 10² to 10⁸ gene copy numbers, respectively (Fig. S3). Host and viral abundance data were applied to an ANOVA test followed by the Tukey-Kramer test to estimate the significant differences, respectively (P < 0.05).

Data availability.

The amplicon data sets were deposited in the DNA Data Bank of Japan (DDBJ) Sequence Read Archive under accession numbers DRR318264 to DRR318269.