Introduction

Harmful algal blooms are an increasingly serious environmental problem for aquaculture, fisheries and public health in many coastal waters throughout the world (Anderson, 1997). Although there have been various biological and non-biological studies of the distribution and growth of flagellate populations (Erga and Heimdal, 1984; Karentz and Smayda, 1984; Haigh et al., 1992; Smayda, 1997; Miralto et al., 1999; Diehl et al., 2002; Ianora et al., 2004; Bruggeman and Kooijman, 2007), it is not always clear how certain phytoplankton species can come to dominate a community. Empirical approaches to studying the allelopathic effects and inter-specific interactions among phytoplankton have developed relatively slowly (Figueredo et al., 2007; Gentien et al., 2007; Strom, 2008).

Allelopathy refers to any direct or indirect inhibitory or stimulatory effect of one plant on another through the production of chemical secretions (Maestrini and Bonin, 1981; Rice, 1984). Most allelochemicals, including low-molecular weight (MW) organic acids, phenolic substances, alkaloids and terpenoids, are known as ‘plant secondary metabolites’—they exist only in plants and are not directly relevant to the maintenance of life. This is considered to be an important factor for elucidating the mechanism of phytoplankton succession. Both in situ and in vitro experiments have suggested that allelopathy plays a key role in the growth dynamics of red-tide blooms through inhibitory effects (Cembella, 2003; Legrand et al., 2003; Granéli and Hansen, 2006; Tillmann et al., 2007, 2008; Van Rijssel et al., 2008). However, a recent review of the allelopathy of phytoplankton (Granéli and Hansen, 2006) questioned whether many low-MW allelochemicals exist in the field at concentrations exceeding their action threshold. To confirm that allelopathy plays an ecological role, the existence of allelochemicals should be reported in relation to their threshold concentrations on a case-by-case basis.

The raphidophyte Heterosigma akashiwo and the diatom Skeletonema costatum form alternating blooms in various coastal waters (Pratt, 1966; Honjo et al., 1978; Honjo and Tabata, 1985). In in situ and in vitro experiments, high concentrations of Olisthodiscus luteus (H. akashiwo; Honjo, 1994; Smayda, 1998) inhibit S. costatum growth, whereas lower concentrations stimulate S. costatum growth (Pratt 1966). Similarly, Honjo et al. (1978) reported that H. akashiwo and S. costatum alternated in forming red-tide blooms in the Hakozaki Fishing Port. They found that filtrate from dense cultures of H. akashiwo, re-enriched with nutrients, inhibited S. costatum growth. Furthermore, Honjo and Tabata (1985) reported an apparent and reciprocal codominance between O. luteus and diatoms in a 70 m3 outdoor tank with flowing coastal water. Recently, we investigated growth interactions between H. akashiwo and S. costatum using bi-algal cultures under axenic conditions, and our findings indicated that the growth of either species could be suppressed by the unidentified allelochemicals of the other, depending on cell densities (Yamasaki et al., 2007).

In this study, we report that H. akashiwo produces allelopathic polysaccharide–protein complexes, and that high-molecular weight allelochemicals exist in the field at concentrations exceeding their action threshold during the H. akashiwo bloom. The study clearly demonstrates the significance of allelopathy by H. akashiwo in a multispecies phytoplankton community in the field.

Materials and methods

Algal species and culture conditions

Axenic strains of S. costatum (NIES-324) and H. akashiwo (NIES-10) were obtained from the National Institute for Environmental Studies (NIES), Japan. An axenic strain of Thalassiosira rotula (CCMP 1647) was obtained from the US Provasoli-Guillard National Center for Culture of Marine Phytoplankton (CCMP). Prorocentrum minimum was isolated from Hakozaki Fishing Port in 1996, and the axenic strain was obtained by repeatedly washing with a capillary pipette. All strains were tested for bacterial contamination by the fluorochrome 4′,6-diamidino-2-phenylindole staining method (Porter and Feig, 1980), and were verified as axenic. Cultures were maintained in 100-ml flasks containing 50 ml modified SWM-3 medium (Yamasaki et al., 2007) without the addition of calcium pantothenate, nicotinic acid, p-aminobenzonic acid, biotin, inositol, folic acid and thymine, at a salinity of 25 at 25 °C under 228 (±5) μmol m–2 s–1 of cool-white fluorescent illumination on a 12:12 h light/dark cycle. In addition, the modified SWM-3 medium was autoclaved (at 121 °C for 15 min) and contained a Tris (hydroxymethyl) aminomethane buffer (Wako Pure Chemical Industries, Ltd.) to prevent pH change during culture. Irradiance in the incubator was measured with a quantum scalar laboratory irradiance sensor (Biospherical Instruments, San Diego, CA, USA).

Allelopathic effect of H. akashiwo on phytoplankton growth

Heterosigma akashiwo was inoculated at a density of 102 cells per ml into 200-ml glass flasks (n=10) containing 100 ml modified SWM-3 medium. After incubation for 17 days (initial cell density, 4–5 × 105 cells per ml), each 100-ml sample from the 10 replicate flasks was combined and passed through a 5.0-μm pore-size membrane filter on a 47-mm polysulphone holder under gravity filtration. Then, 1000 ml filtrate was aerated to avoid an increase in pH and to supply dissolved inorganic carbon (DIC). After aeration, nutrients dissolved in Milli-Q water were added to 1000 ml filtrate to give the same final concentration as in the original modified SWM-3 medium. The filtrate was then passed through a 0.22-μm pore-size filter (Millipore, Billerica, MA, USA). The concentration of protein in the filtrate was determined using a BCA protein assay kit (Thermo Fisher Scientific Inc., Rockford, IL, USA). In addition, bovine serum albumin dissolved in modified SWM-3 medium was used as a standard for the determination of protein. This quantitation assay had three replicates.

A sample (0.5 ml) of each cell suspension of S. costatum, T. rotula, P. minimum and H. akashiwo at various inoculum cell densities (initial cell densities, 10, 102, 103 and 104 cells per ml) was added to 4.5 ml of each test solution (described in the above two paragraphs) in polystyrene test tubes (Evergreen Scientific, Los Angeles, CA, USA). As a control, each species was grown in the modified SWM-3 medium without the addition of any sample solution. This growth experiment had 16 test conditions, with four replicates per test condition. After the start of incubation, the in vivo fluorescence of each species was measured daily with an in vivo fluorometer (10-AU-005-CE; Turner Designs, Sunnyvale, CA, USA).

Purification of allelochemicals

For the culture, H. akashiwo was inoculated at a density of 100 cells per ml into 200-ml glass flasks (n=10) containing 100 ml modified SWM-3 medium. After 20 days, 100 ml samples of the culture from each flask were combined to give a total of 1000 ml. This was passed through a 5.0-μm pore-size membrane filter (Millipore) under gravity filtration, and then through a 0.45-μm pore-size membrane filter (Millipore). The filtrate was dialysed against deionized water for 3 days at 4 °C using dialysis membranes with a 3500-Da MW cutoff (MWCO; Spectrum Laboratories, Rancho Dominguez, CA, USA). After the dialysis, the inner solution was frozen at −80 °C and lyophilized with a Free Zone 2.5 freeze-drying apparatus (Labconco, Kansas City, MO, USA). The crude extract obtained from filtrates of H. akashiwo cultures was dissolved in 50 mM Tris-HCl buffer (pH 8.0), and passed through a 0.22-μm syringe filter (Millipore). A sample (2.5 ml) was subjected to a DEAE-Toyopearl Pack column (8 mm I.D. × 7.5 cm; Tosoh, Tokyo, Japan), which was equilibrated with 50 mM Tris-HCl buffer (pH 8.0), and was eluted with a 0–1 M gradient of NaCl. The flow rate was 1.0 ml min–1, and the eluate was monitored at 280 nm. The total sugar in 1.0 ml of each fraction was determined by the modified phenol-sulphuric acid method (Mckelvy and Lee, 1969). Fraction 5 (Figure 1a) obtained by the anion-exchange chromatography was pooled and dialysed against deionized water for 3 days at 4 °C using dialysis membranes with a 3500-Da MWCO (Spectrum Laboratories). After dialysis, the inner solution was frozen at −30 °C and lyophilized with a Free Zone 2.5 freeze-drying apparatus (Labconco). The freeze-dried sample was dissolved in 50 mM Tris-HCl buffer (pH 7.0) including 200 mM NaCl, and was passed through a 0.22-μm syringe filter (Millipore). The 300-μl sample was then subjected to a TSKgel G-4000SWXL column (7.8 mm I.D. × 30 cm; Tosoh), which was equilibrated with 50 mM Tris-HCl buffer (pH 7.0). The flow rate was 0.8 ml min–1, and the eluate was monitored at 280 nm. The total sugar in 0.8 ml of each fraction was determined by the modified phenol-sulphuric acid method (Mckelvy and Lee, 1969). In addition, the inhibitory effect of each fraction obtained from the anion-exchange chromatography and the gel-filtration chromatography on S. costatum growth was determined by the bioassay.

Figure 1
figure 1

Purification of H. akashiwo allelochemicals. (a) Anion-exchange chromatography of a preparation from the culture filtrate of H. akashiwo on a high-performance liquid chromatography (HPLC) system equipped with a DEAE-Toyopearl Pack column. (b) Effect of each fraction obtained from anion-exchange chromatography on the growth of S. costatum examined using a 48-well plate. This bioassay had three replicates. (c) Gel-filtration profile of the fraction (F-5) obtained from anion-exchange chromatography on a HPLC system equipped with a TSK gel G-4000SWXL column. The arrowhead (VBD) indicates the elution time of blue dextran (MW 2 × 106). (d) Effect of each fraction obtained from gel-filtration chromatography on the growth of S. costatum examined using a 48-well plate. This bioassay had three replicates. (e) SDS-PAGE of F-A conducted using 5% polyacrylamide gel. After electrophoresis, the gels were stained with either silver stain (lane 1–4) or PAS stain (lane 5–7). Lane 1, the molecular mass standard markers; lane 2, 5, F-A obtained from gel-filtration chromatography; lane 3, 6, positive control (horseradish peroxidase); and lane 4, 7, negative control (soybean trypsin inhibitor).

Full size image

Sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE)

The fraction-A (Figure 1c; F-A) obtained from gel filtration was analysed by SDS-PAGE on 5% gels according to the Laemmli method (Laemmli, 1970). After electrophoresis, the gels were silver-stained and periodic acid-Schiff (PAS)-stained for visualization of protein and glycoprotein bands, respectively.

For deglycosylation, the fraction-A (Figure 1c; F-A) was treated with glycopeptidase F (TaKaRa) under denaturing conditions at 37 °C for 24 h (Tarentino et al., 1985), and was treated with alkaline (Morelle et al., 1998), respectively. After deglycosylation with glycopeptidase F and alkaline treatments, each sample was analysed by SDS-PAGE as described above.

Amino-acid and sugar-composition analyses

The amino-acid and sugar compositions of fraction-A (Figure 1c; F-A) were analysed by the ninhydrin method (Moore and Stein, 1948) and post-column fluorometric detection (Mikami and Ishida, 1983), respectively.

Bioassays using 48-well plates

Skeletonema costatum cells in the stationary phase (10 to 12 × 105 cells per ml) were diluted to 104 cells per ml with modified SWM-3 medium. A sample (10 μl) of the S. costatum cell suspension was added to 990 μl of each sample solution to be tested in 48-well plates (initial cell density, 102 cells per ml). Each bioassay had three replicates. As a control, S. costatum was grown in modified SWM-3 medium without the addition of any sample solution. After incubation for 5 days, the cells in each of five 10-μl subsamples from each well were counted microscopically. If necessary, the samples were diluted by a factor of 10–20 with modified SWM-3 medium.

Antibody production against allelopathic polysaccharide–protein complexes (APPCs) in a rabbit

APPCs (F-A, Figure 1c) was purified as described above. The F-A fractions were pooled, dialysed and lyophilized. An adult New Zealand white rabbit was used to produce polyclonal antibodies against APPCs of H. akashiwo. Immunological cross-reaction of polyclonal antibody against 11 species of phytoplankters was examined by the dot-blot analysis. Details of methods are given in the Supplementary Figure 1.

Indirect fluorescent-antibody test (IFAT)

Skeletonema costatum (7 × 105 cells per ml), T. rotula (13 × 104 cells per ml) and P. minimum (4 × 105 cells per ml) were collected from each 10 ml culture by centrifugation, and were treated with 1 ml of APPCs (F-A, Figure 1c) solution dissolved in modified SWM-3 medium (100 μg per ml) as the antigen. After incubation for 90 min at 25 °C, the APPCs solution was removed from each tube and the cells were washed three times in modified SWM-3 medium. The modified SWM-3 medium was then removed, and the first antibody solution, which was produced by dissolving 100 μl antiserum in 900 μl-modified SWM-3 medium, was added. After incubation for 50 min at 25 °C, the first antibody solution was removed from each tube, and the cells were washed three times in modified SWM-3 medium. The modified SWM-3 medium was then removed, and the second antibody solution (fluorescein isothiocyanate-conjugated anti-rabbit γ-globulin IgG of goat; ZYMED Laboratories, South San Francisco, CA, USA), which was produced by dissolving 20 μl antiserum in 980 μl modified SWM-3 medium, was added. After incubation for 50 min at 30 °C under dark conditions, the second antibody solution was removed from each tube and the cells were washed three times in modified SWM-3 medium. They were then observed using a fluorescent microscope (BZ-9000; Keyence, Osaka, Japan) under the following conditions: excitation wavelength, 488 nm; absorption wavelength, 530 nm; exposure time, 0.17 s for S. costatum, T. rotula and P. minimum, and 0.33 s for H. akashiwo. In addition, H. akashiwo was tested using the method described above in the absence of APPCs solution.

Field study

Surface seawater was sampled at Hakozaki Fishing Port from May to June 2007. The sampling was conducted every 3 or 4 days, or daily during the H. akashiwo bloom. After sampling, the numbers of phytoplankters in each of the two 500-μl subsamples were counted microscopically without fixation. After sampling, a 10-ml portion of each 1000 ml sample was passed through a 0.22-μm pore-size filter, and frozen at −80 °C until use. APPCs (F-A, Figure 1c) of H. akashiwo in seawater samples were detected by dot-blot analysis.

A 300-ml seawater sample from each of the H. akashiwo and S. costatum blooms was passed through a 5.0-μm pore-size membrane filter (Millipore) under gravity filtration to remove particulates, including phytoplankton, and then through a 0.45-μm pore-size membrane filter (Millipore). These filtrates were dialysed and lyophilized as described above. A bioassay was conducted on each sampling day, to estimate the effect of the collected seawater samples on S. costatum growth. A 50-ml portion of each 1000 ml sample was passed through a 0.45-μm pore-size filter (Steriflip; Millipore) to remove microparticles. Nutrients dissolved in Milli-Q water were added to each 50 ml sample of filtrate to give the same final concentration as in the original modified SWM-3 medium. The filtrate was then passed through a 0.22-μm pore-size filter (Millipore). The bioassay was conducted using 48-well plates (Corning, Corning, NY, USA), and had three replicates. Crude extracts from the laboratory culture of H. akashiwo and red-tide seawater from H. akashiwo were dissolved in modified SWM-3 medium. The protein concentration in each sample was determined by a BCA protein assay kit (Thermo Fisher Scientific Inc.). Crude extracts from red-tide seawater samples from S. costatum (80 μg dry weight) were dissolved in 1 ml modified SWM-3. The effects of each sample on S. costatum growth were examined using 48-well plates.

Dot-blot analysis

A dry nitrocellulose sheet (Bio-Rad, Hercules, CA, USA) was pre-wetted in Tris-buffered saline (TBS; 20 mM Tris and 500 mM NaCl, pH 7.5), and the pre-wetted membrane was placed in the Bio-Dot SF apparatus (Bio-Rad). To ensure uniform binding of the antigen, the membrane was rehydrated with 100 μl TBS per well. After rehydration, each sample was applied at 5 μl per well, and the entire samples were filtered through the membrane under a gentle vacuum. Each sample well was then washed with 200 μl TBS under a gentle vacuum. After the wells had completely drained, the membrane was removed from the apparatus and placed in the blocking solution (3% gelatin-TBS). The membrane in the blocking solution was gently agitated using an orbital shaker platform and incubated for 40 min at room temperature (RT). After incubation, the blocking solution was removed, and TBS containing 0.05% Tween-20 (TTBS) was added to the membrane. The membrane was washed in TTBS for 10 min with gentle agitation at RT. After washing, the TTBS was removed and first antibody solution, which was a 1 per 100 dilution of the antiserum in antibody buffer (1% gelatin-TTBS), was added to the membrane. After incubation for 2 h under gentle agitation at RT, the unbound first antibody was removed by washing the membrane with TTBS for 5 min at RT, and then with fresh TTBS for 5 min at RT. The TTBS was removed and the second antibody solution, which was a 0.033 per 100 dilution of the antibody conjugate (goat anti-rabbit IgG (H+L) alkaline phosphatase (AP) conjugate; Bio-Rad) in 100 ml antibody buffer, was added. After incubation for 2 h under gentle agitation at RT, the conjugate solution was removed and the membrane was washed in TTBS for 5 min with gentle agitation at RT. The TTBS was then removed, and the membrane was washed with fresh TTBS for 5 min at RT. The membrane was then washed with TBS for 5 min with gentle agitation to remove residual Tween-20 from the surface. The TBS was removed and the membrane was immersed in colour development solution (Immun-Blot Assay Kit; Bio-Rad). The colour development was stopped by immersing the membrane in distilled water for 10 min with gentle agitation at RT.

Results and discussion

Allelopathic effect of H. akashiwo on phytoplankton growth

When S. costatum was inoculated at low-cell densities and exposed to the filtrate from dense culture of H. akashiwo, its growth was markedly inhibited. By contrast, S. costatum growth was not inhibited by the filtrate of H. akashiwo at all of the concentrations (37–84 μg protein per ml) when it was inoculated at high cell densities (Figure 2a, Supplementary Figure 2). Thus, it is likely that the allelopathic effects of H. akashiwo were both species-specific and dependent upon the cell density of the target species. This supported the finding of a previous study (Yamasaki et al., 2007), and suggested that allelopathy drives the alternation of the two species in the field. Similarly, the diatom T. rotula was also strongly inhibited depending upon the concentration of protein dissolved in the filtrate of H. akashiwo (Figure 2, Supplementary Figure 2). However, Prorocentrum minimum, which does occur naturally with H. akashiwo, was stimulated by the filtrate of H. akashiwo at all of the concentrations (37–84 μg protein per ml) and the target cell densities (10–104 cells per ml) tested (Figure 2, Supplementary Figure 2). This result was similar to previous studies, which showed that P. minimum was stimulated by organic substances such as humic acids (Kondo et al., 1990; Heil, 2005). Therefore, the stimulation effect of allelochemicals on P. minimum growth might be caused by the utilization of allelochemicals as nutrient sources. On the other hand, the filtrate of H. akashiwo at all of the concentrations (37–84 μg protein per ml) did not inhibit H. akashiwo at inoculum densities >102 cells per ml (Figure 2d, Supplementary Figure 2).

Figure 2
figure 2

Species-specific allelopathic effects of H. akashiwo filtrate. Allelopathic effect of H. akashiwo on phytoplankton growth [(a), S. costatum; (b), T. rotula; (c), P. minimum; (d), H. akashiwo]. Allelopathic effects were examined as described in the text, and the experiment had four replicates.

Full size image

Purification and chemical composition analysis of APPCs

The high-molecular weight fraction obtained from the filtrate of H. akashiwo culture was subjected to a DEAE-Toyopearl anion-exchange column chromatography. Based on the UV (280 nm)-monitored protein elution profile, five fractions (F-1 to F-5) were collected (Figure 1a). The elution pattern plotted by the phenol-sulphuric acid method indicated that F-1 and F-5 contained abundant sugar, whereas other fractions (F-2 to F-4) were almost sugarless (Figure 1a). When the allelopathic effect of each fraction was examined by the bioassay using S. costatum, its effect was exclusively detected in F-5, whereas no such effect was detected in other fractions (Figure 1b). For further purification and characterization, the fraction F-5 was dialysed against deionized water, lyophilized, dissolved in 50 mM Tris-HCl buffer (pH 7.0) and subjected to gel-filtration chromatography on a TSKgel G-4000SWXL column. The three fractions (F-A to F-C), all of which contain sugar, were collected (Figure 1c) and the allelopathic effect against S. costatum was examined. Among the three fractions, most of the allelopathic activity was detected in F-A (Figure 1d). As the active fraction (F-A) appeared in the void volume of TSKgel G-4000SWXL column, which has an exclusion limit of 106–107 Da, the molecular sizes of allelochemical substances produced by H. akashiwo were estimated to be >106 Da (Figure 1c). The extremely large molecular sizes of the allelochemicals (F-A) were also suggested by a silver-stained SDS-PAGE gel, which showed a broad band in the stacking gel, not migrating into the separation gel (Figure 1e, lane 2). The broad band was also visualized by PAS staining (Figure 1e, lane 5), suggesting that the allelochemicals are polydisperse glycoproteins, as is the case with mucin, which has molecular masses ranging from 2 × 106 Da to 5 × 107 Da (Bansil and Turner, 2006). We therefore proceeded to analyse chemical compositions of the allelochemicals (F-A) obtained at the step of gel filtration, without further purification.

Amino acid and sugar compositions of the allelochemicals (F-A) showed enrichment in Asx, Thr, Ser, Glx, Gly, Ala, Val, Leu (around 10 mol% per each amino acid) as protein constituents (Figure 3a), Gal (60 mol%) and Man (20 mol%) for sugar constituents (Figure 3b). Because Asx and Ser/Thr are relatively dominant amino acids (10 and 16 mol%, respectively), the presence of a great number of N- and O-linked glycans were expected as seen in macromolecular glycoproteins, such as mucin, collagen and proteoglycans. Contrary to expectation, the allelochemicals (in fraction F-A) were resistant to glycopeptidase F and alkaline treatments (data not shown). These observations suggest that the H. akashiwo allelochemicals are most likely macromolecular glycoproteins, but are distinct from any known glycoproteins, having characteristic chemical compositions (Figure 3). Although further biochemical analyses are needed to clarify the structural aspects of H. akashiwo allelochemicals, we tentatively designate these allelochemicals as APPCs (allelopathic polysaccharide–protein complexes).

Figure 3
figure 3

Chemical composition analysis of F-A. (a and b) Amino-acid (a) and sugar (b) compositions of F-A produced by H. akashiwo.

Full size image

All the allelochemicals reported to date in land and aquatic plants are small compounds (<104 Da) and most of them have been identified as secondary metabolites (Maestrini and Bonin, 1981; Rice, 1984; Cembella, 2003; Legrand et al., 2003; Granéli and Hansen, 2006). To our knowledge, this is the first report demonstrating the presence of extremely large allelochemicals (>106 Da). Thus, APPCs produced by H. akashiwo are novel members of allelochemicals, meaning that APPCs form a new category of allelopathic agents. Although chemical structures of APPCs and mechanisms by which APPCs express allelopathic functions are unclear, viscoelastic, colloidal and/or selectively adhesive properties of APPCs could be important features of certain exopolymeric APPCs, which would not be exerted by small allelochemicals.

Localization of APPCs on the H. akashiwo cells

To investigate localization of APPCs on the H. akashiwo cells, we developed an antibody against purified APPCs (F-A, Figure 1c). An indirect fluorescent-antibody test (IFAT) indicated that APPCs were present on the cell surface of H. akashiwo (Figure 4a). In addition, immunological cross-reaction of the second antibody against H. akashiwo was not detected under experimental conditions (Figure 4b). Furthermore, IFAT using the cell pellet of H. akashiwo indicated that APPCs were present in the glycocalyx (Yokote et al., 1985) of H. akashiwo (Figures 4c–e). On the other hand, APPCs were present not only in the glycocalyx and/or on the cell surface of H. akashiwo, but also in the filtrate when this species was cultured alone (Supplementary Figure 3). It is therefore likely that H. akashiwo continually releases APPCs from its cell surface into the environment.

Figure 4
figure 4

Localization of APPCs on the surface of H. akashiwo cells using the fluorescent-antibody method. IFAT was conducted as described in the text. (a) H. akashiwo was tested by IFAT without addition of APPCs. (b) As a control, H. akashiwo was tested by the addition of normal serum. Each cell and glycocalyx of H. akashiwo was tested by the IFAT (c–e); (c) Single cell and glycocalyx of H. akashiwo; (d) Antigen-positive part of glycocalyx; (e) Antigen-negative part of glycocalyx.

Full size image

Possible mechanism of growth inhibition by APPCs

An indirect fluorescent-antibody test (IFAT) indicated that the growth inhibition of diatoms could be induced by specific binding of APPCs to the cell surface of the target species (Figure 5). In addition, immunological cross-reaction of polyclonal antibody against S. costatum, T. rotula and P. minimum was not detected under experimental conditions (Figures 5d–f). The APPCs might therefore have bound to the cell surfaces of S. costatum and T. rotula (Figures 5a and b), which were inhibited by H. akashiwo (Figures 2a and b), but not P. minimum (Figure 5c), which was not inhibited (Figure 2c). These factors—that is, the amounts of APPCs produced for each cell and whether they bound to the cell surface—might therefore have acted as triggers, and induced the particular effects of allelopathy as evidenced by two physiological processes: photosynthesis and enzyme activity (Gross, 2003). Alternatively, it remains possible that the cells bound specifically with APPCs get clogged by the APPCs, and suffer from a reduced gas exchange.

Figure 5
figure 5

Binding specificity of H. akashiwo APPCs. Detection of APPCs from the cell surfaces of S. costatum, T. rotula and P. minimum using IFAT. IFAT was conducted as described in the text. S. costatum (a), T. rotula (b) and P. minimum (c) were exposed by APPCs and were tested by IFAT. As a control, S. costatum (d), T. rotula (e) and P. minimum (f) were tested without the addition of APPCs.

Full size image

However, it may be difficult to explain species-specific allelopathic effects of APPCs produced by H. akashiwo by such mechanisms. It is much more likely that the mode of action of APPCs is similar to growth factors (Mendelsohn and Baselga, 2003) in having specific receptors comparable with those on the surface of target mammalian cells. Further study will be required to clarify the mode of action of APPCs on the growth of phytoplankters.

Significance of APPCs in the field

A field study was carried out in Hakozaki Fishing Port, Fukuoka, Japan (latitude 34°24′58′′N, longitude 130°12′20′′W), in May and June 2007. To elucidate the role of allelopathy in the field, we evaluated whether several factors affected the growth dynamics of phytoplankton in Hakozaki Fishing Port. During the field study, most of the environmental factors were suitable for S. costatum growth (Supplementary Figures 4, 5); however, S. costatum declined from 28 to 30 May due to unknown factors apparently unrelated to allelopathy of H. akashiwo (Figure 6b).

Figure 6
figure 6

Detection of APPCs in the field during H. akashiwo bloom. (a and b) A bioassay was conducted on each sampling day using seawater samples (a), and phytoplankters were counted microscopically (b). (c) Dot-blot analysis conducted with two replicates using seawater samples. All samples were spotted at 5 μl per spot. (d) Effects of crude extracts from the laboratory culture of H. akashiwo and red-tide seawater from H. akashiwo on S. costatum growth examined using 48-well plates. (e) Effects of crude extracts from red-tide seawater samples from S. costatum on its growth examined using 48-well plates. This bioassay had three replicates (100% growth as that in control medium).

Full size image

Interestingly, APPCs were detected by a dot-blot analysis of three seawater samples taken from 7 to 9 June (Figure 6c), when S. costatum growth was inhibited according to the results of the bioassay (Figure 6a), but were below the detection limit of dot-blot analysis (Figure 6c, Supplementary Figure 3a). Similarly, APPCs were detected by a dot-blot analysis of axenic cultures of H. akashiwo at densities 1 × 105 cells per ml (Supplementary Figure 3b), at which H. akashiwo can inhibit S. costatum growth in the laboratory experiments (Yamasaki et al., 2007); these findings supported the conclusion drawn from the field study (Figures 6b and c) that APPCs can affect the growth of other phytoplankton in the field. A comparison of dot-blots from 40 μg ml–1 culture filtrate and from the dense bloom sampled on 8 June (Figure 6d) showed a similar stain density in both, indicating a similar amount of allelopathic activity, which was sufficient to suppress S. costatum growth. In fact, the inhibitory effect of the filtrate from the dense bloom was greater than that of the crude extract of axenic cultures of H. akashiwo (Figure 6d). However, two crude extracts collected from seawater samples during the S. costatum bloom did not affect its growth (Figure 6e). By contrast, S. costatum growth in the field and in the laboratory experiment recovered concurrently (compare Figures 6a and b) when H. akashiwo declined and APPCs were no longer detected by the dot-blot analysis (Figure 6c). Furthermore, we showed that the APPCs became deactivated over time (time-scale of days) (Supplementary Figure 6), which was comparable to the reported reduction of activity by the allelopathic material secreted by the red-tide dinoflagellate Karenia mikimotoi (Gentien et al., 2007). These results suggest that APPCs of H. akashiwo affect the growth of phytoplankton during dense H. akashiwo blooms.

Final conclusions

This is the first report that phytoplankton-secreted allelochemicals, which have previously been found in the laboratory, exist in nature at concentrations exceeding their action threshold. Allelopathic polysaccharide–protein complexes (APPCs, MW >106) produced by H. akashiwo (Figures 1 and 3), inhibited the growth of primary producers such as S. costatum and T. rotula (Figure 2, Supplementary Figure 2), but stimulated or were neutral to the harmful dinoflagellates, Prorocentrum spp. and Heterocapsa circularisquama (Figure 2, Supplementary Figure 7), during the dense H. akashiwo bloom (Figure 5b, Supplementary Figure 8). In addition to the traditional concept of allelopathy between phytoplankton species, APPCs produced by H. akashiwo may have another role. Shikata et al. (2008) reported that the density of H. akashiwo cysts in the bottom sediment of the Hakozaki Fishing Port sharply increased from the time of the peak to the decline phase of a bloom in the water. Thus, APPCs may not only contribute to growth competition with other species during the late exponential phase of the bloom as reported by Pratt (1966), Honjo et al. (1978), and Honjo and Tabata (1985), but also contribute to maintenance of monospecific H. akashiwo blooms through inhibitory effects on competitors. By reducing competition from organisms such as diatoms and attacks from predators, H. akashiwo succeeds in establishing dense cyst beds, which give the next generation an advantage by allowing extensive re-inoculation of the water column. Increases in the frequency and severity of harmful algal blooms have become a global-scale problem, associated with climate warming, (Peperzak, 2003) and eutrophication (Wu, 1999). Furthermore, the escalation of global environmental problems in the coming years will likely exacerbate the problem of harmful algal blooms including H. akashiwo to socio-economic activity, especially in developing countries (Wu, 1999). Our results suggest that, as harmful algal blooms increase in severity and frequency, the target-specific effects of APPCs produced by H. akashiwo will increasingly control the structure of plankton ecosystems.