H. akashiwo, S. costatum, and P. donghaiense were obtained from the Algal Species Management Center, School of Ocean and Earth, Xiamen University (China). The algal cultures were grown in f/2 medium (Guillard, 1975) at 20∼22°C under a 12:12 h light: dark cycle with the light intensity maintained at 50 μmol photons m^–2 s^–1.

The algicidal bacterium D8 was originally isolated from water samples that were collected from Xiamen’s first wharf in Fujian Province, China, during a H. akashiwo bloom. It was grown in STA medium (Hetharua et al., 2018) at 28∼30°C with a shaking speed of 150 rpm. Isolation and purification of strain D8 were carried out according to a previous study (Zhang et al., 2018).

Bacterial D8 was characterized by its 16S rRNA gene. The 16S rRNA gene was amplified by PCR using primers 27F and 1492R. The PCR product of the 16S rRNA gene was purified using the TaKaRa MiniBEST DNA purification kit (TaKaRa Bio Inc., Dalian, China) and then sequenced. The sequence was submitted to EzBioCloud¹ for sequence alignment. Phylogenetic analysis was performed using MEGA version 5.0 software and neighbor-joining analysis. The cell morphology was observed by SEM (JSM-6390, JEOL Co., Tokyo, Japan) and TEM (JEM-2100HC, JEOL Co., Tokyo, Japan).

To investigate the algicidal mode, the cell-free supernatant of the D8 culture (20 ml) was collected by centrifugation at 6,000 rpm for 10 min and then filtered through a 0.22 μm membrane. The cell pellet was resuspended in 20 ml of sterile f/2 medium for the subsequent analysis of direct algicidal activity. The different fractions of the D8 culture were added to the algal cultures (10 ml) at a concentration of 5% (v/v). After 48 h of treatment, the algicidal rate of each fraction was calculated by the following formula:

where N₀ represents the cell number of the algal cultures measured immediately after treatment, and N_t represents the cell number of the algal cultures at different treatment times.

Strain D8 was cultured in 500 ml of STA medium for 72 h, and then the supernatant was obtained by centrifugation at 8,000 rpm for 10 min. The pH of the D8 supernatant was adjusted to 2.0 with 6 M HCl, and the solution was kept at 4°C overnight. Then, the precipitate was collected by centrifugation and extracted once with 20 ml of chloroform:methanol (2:1) and twice with 20 ml methanol. The methanol was removed using a rotary evaporator to obtain the D8 crude extract. The D8 crude extract (20 mg) was dissolved in 1 ml of DMSO for further analysis. To assess the algicidal activity of the D8 crude extract, 2.5, 5, 10, 15, or 20 μg/mL D8 crude extract was inoculated into H. akashiwo cultures, and the algicidal activity was calculated according to the above formula.

To assay the photosynthetic response of H. akashiwo exposed to the D8 crude extract, the Chl a content, carotenoid content, maximum quantum yield of photosystem (PS) II (Fv/Fm) and relative electron transport rate (rETR) of the algal cells were measured after treatment with D8 crude extract. The algal cultures (100 ml) were treated with 2.5, 5, or 10 μg/ml of D8 crude extract, and 10 ml samples were collected after 3, 6, 12, and 24 h of exposure. The Chl a and carotenoids in the collected samples were extracted with 2 ml of 95% alcohol. The pigment levels were calculated by measuring the absorbance at 665, 645, and 470 nm, and applying the following equations:

where A₆₆₅, A₆₄₅, and A₄₇₀ represent absorbance values at wavelengths of 665, 645, and 470 nm, respectively, and C_{Chlorophyll a} represents the Chl a content.

The Fv/Fm and rETR values of treated samples (2 ml) were measured using a multiple excitation-wavelength modulated chlorophyll fluorometer (Multi-Colour-PAM, Walz, Oberschwaben, Germany) according to our previous study (Zhang et al., 2018).

Fluorescein diacetate (FDA) was used to assess the viability of H. akashiwo. FDA is a non-polar molecule and can freely enter living cells, where it is digested by esterase to produce fluorescein. Fluorescein is a polar molecule and cannot penetrate the cell membrane. Therefore, digested fluorescein accumulates in living cells and emits green fluorescence. The ratio of fluorescein-positive cells to total cells indicates cell viability. To determine the effect of D8 crude extract on cell viability, algal cells were treated with D8 crude extract at final concentrations of 2.5, 5, and 10 μg/ml. After treatment for 5, 30, 60, and 180 min, the algal cells were collected by centrifugation and stained with 20 μg/mL FDA (F809625, Beijing Huawei Ruike Chemical Co., Ltd, Peking, China) for 5 min in the dark at room temperature. Fluorescence was analyzed on a flow cytometer (BD LSRFortessa, Becton, Dickinson and Company, New Jersey, USA) using 488 nm excitation and a 525 nm bandpass filter for digested fluorescein detection. For each sample, approximately 5,000 cells were analyzed. Fluorescence images were collected by a fluorescence microscope (Leica DM4B, Lecia, Wetzlar, Germany).

Algal cells treated with 5 μg/ml D8 crude extract for 0, 0.5, 3, 6, 12, and 24 h were separately collected by centrifugation at 1,000 rpm for 5 min. The collected cells were fixed with 2.5% (v/v) glutaraldehyde overnight at 4°C and then washed once with f/2 medium. The fixed samples were observed using SEM (SUPRA 55 SAPPHIRE, Zeiss Optical Instruments Inc., Jena, Germany) according to our previous work (Zhang et al., 2018).

D8 crude extracts were further separated and purified by chromatographic methods using an octadecylsilyl (ODS) column (particle size 50 μm, Φ3.0 × 30 cm) with a loading volume of 2 ml (100 mg/ml D8 crude extract of methanol) at room temperature. Acetonitrile (mobile phase A) and 0.05% TFA/water (v/v) (mobile phase B) were used as the solvent system. The column was first washed with the initial mobile phase (80% A + 20% B) at a flow rate of 3 ml/min, and 25 ml aliquots were collected in test tubes. After filling the 8th tube, 100% A was used to elute the algicidal components. The eluate signal was detected using a spectrophotometer at a wavelength of 205 nm. Algicidal components were characterized by an analytical high-performance liquid chromatograph (Waters 2545, Waters Corporation, Milford, MA, USA) equipped with a reversed-phase HPLC column (Sunfire C18, Waters Corporation, USA, Φ4.6 mm × 50 mm, particle size 5 μm) at a flow rate of 1 ml/min.

Further purification and preparation were performed using a SilGreen C18 column (GH0525010C18A, Beijing Green Baicao Technology Development Co. Ltd, Peking, China, Φ10 mm × 100 mm, particle size 5 μm) at a flow rate of 4.5 ml/min. The eluate was monitored at 205 nm by preparative liquid chromatography (LC-20AP, Shimadzu, Kyoto, Japan). The mobile phase was 85% acetonitrile (0.05% TFA), and an isometric elution method was used. The target fraction was collected and evaporated under vacuum by a rotary evaporator (EYELA OSB-2100, Tokyo Rikakikai Co., Ltd, Japan) at 45°C.

Purified compounds were analyzed using a high-resolution LC-MS/MS instrument (Q-Exactive, Thermo Fisher Scientific Inc., Waltham, USA). LC–MS was conducted using an Orbitrap MS instrument equipped with an electrospray ionization (ESI) source. The mass spectrometer was operated in positive ion mode using ESI under the following conditions: a spray voltage of 3800 V and a capillary temperature of 320°C.

One-dimensional (¹H-NMR, ¹³C-NMR) spectrograms of purified compounds were recorded using an NMR spectrometer (Bruker AV600, Bruker, Zurich, Switzerland) at 600 and 150 MHz for ¹H and ¹³C NMR, respectively, with DMSO-d6 (δ_H 2.50 and δ_C 39.5 ppm) as the solvent. ¹H-NMR spectra were obtained at 600 MHz with 40 scans, a frequency resolution of 0.09 Hz, and a relaxation delay time of 1 s using a spectral width of 12019.2 Hz. ¹³C-NMR spectra were obtained at 150 MHz with 1500 scans, a frequency resolution of 0.55 Hz and a relaxation delay time of 2 s using a spectral width of 36057.7 Hz.

To assess the algicidal activity of purified components, various concentrations (0.5, 1, 2, 3, 4, 5, 10, 15 20, 30, or 60 μg/ml) of each component were inoculated into H. akashiwo, S. costatum, and P. donghaiense cultures, respectively. The cell density of algal cultures was assessed and recorded using a Countstar analyzer (ALIT, Shanghai, China) every 12 h for 48 h after the treatment with each component. The algicidal activity was calculated according to the above formula. Treatments with the same volume of DMSO served as the control. All treatments were performed in triplicate.

All of the experimental data were collected in triplicate and are presented as the means ± standard deviations. The statistical significance of the difference in parameters between the control and treatment groups was analyzed using GraphPad Prism 8.0 with two-way ANOVA, for which P < 0.05 was considered to indicate statistical significance. The LC₅₀ values were calculated by probit analysis in SPSS.

Morphological identification showed that the algicidal strain D8 formed white and smooth colonies when plated. SEM observation showed that the cells of strain D8 were short and rod-shaped and lacked flagella (Figures 1A,B). PCR amplification of the 16S rRNA gene (1,546 bp, GenBank accession number: MW479447) and sequencing revealed that D8 shared the highest similarity (99.86%) with B. tequilensis KCTC 13622^T (Figure 1C). The results indicated that strain D8 belonged to the genus Bacillus, and thus we named it B. tequilensis D8. The algicidal activity analysis showed that both the D8 culture and the cell-free supernatant exhibited strong algicidal activities (above 85%) against H. akashiwo after treatment for 24 h, and the activities exceeded 95% after treatment for 48 h (Figure 1D). The bacterial cells alone exhibited less than 6% algicidal activity. This result indicates that strain D8 killed algal cells by secreting active substances into the supernatant.

In addition to H. akashiwo, strain D8 showed high algicidal activities against Prorocentrum donghaiense, Thalassiosira pseudonana, and Skeletonema costatum. However, it exhibited low algicidal activities against species belonging to Cyanophyta and did not kill species belonging to Chlorophyta (Supplementary Table 1).

The stability of the algicidal activity of the D8 supernatant was evaluated at various temperatures and light intensities. Supplementary Figure 1 shows that the algicidal activity of the D8 supernatant was stable from −80 to 121°C and was not affected by light exposure. The results suggest that the algicidal substances produced by strain D8 exhibit good temperature and light stability and have potential applications in the biocontrol of HABs.

To explore the chemical nature of the algicidal substances, the D8 supernatant was first precipitated by HCl or NaOH. The precipitate and supernatant were then separately collected by centrifugation. Supplementary Figure 2 shows that only the 5% D8 supernatant and 5% acid precipitate exhibited obvious algicidal activities, while the 5% HCl-precipitated supernatant, 5% alkali precipitate, and 5% NaOH-precipitated supernatant showed no significant algicidal effects, indicating that the main algicidal substances were present in the acid precipitate.

The acid precipitate was further purified to obtain the D8 crude extract. The algicidal activity measurements showed that treatment with 2.5, 5, 10, 15, and 20 μg/ml D8 crude extract resulted in algicidal rates of 57.4, 78.5, 95.7, 97.2, and 100%, respectively, after 3 h (Figure 2A), indicating that the D8 crude extract had strong algicidal activity against H. akashiwo in a short period of time and that the algicidal effect of the D8 crude extract was concentration dependent.

To monitor the events involved in the cell damage to H. akashiwo induced by the D8 crude extract, the morphological changes of the algal cells were observed using SEM. The algal cells in the control group were intact, and the plastids were arranged neatly on the cell surface (Figure 2B). After treatment with the D8 crude extract for 0.5 and 3 h, the algal cell morphology was severely damaged, the cell membrane was ruptured, the plastids had spilled out, and finally, whole cells collapsed (Figure 2B; Supplementary Video 1). When the algal cells were treated with 10 μg/ml D8 crude extract, the resulting dynamic changes showed that the membrane was attacked and rapidly disintegrated, and the cell contents were immediately released (Supplementary Video 1). The above results suggested that the D8 crude extract could rapidly target the cell membrane and lyse algal cells in a short time.

Fluorescein diacetate (FDA) was used to assess the viability and membrane permeability of H. akashiwo. As shown in Figure 3A, the control emitted strong fluorescence, while the fluorescence intensity of algal cells treated with 5 or 10 μg/ml D8 crude extract for 5 min was significantly weaker than that of the control. The results indicated that the D8 crude extract instantly damaged the plasma membrane and eliminated esterase activity. Figure 3B shows that the proportions of fluorescence-positive cells decreased by 71.6, 83.6, and 89.3% after exposure to 2.5, 5, and 10 μg/ml D8 crude extract for 60 min, respectively. This result indicated that the proportion of damaged cells increased significantly with increasing concentrations of D8 crude extract.

According to the above results, the D8 crude extract primarily targets the membrane system of H. akashiwo. Photosynthetic electron transformation is dependent on the intact thylakoid membrane. To explore the effect of the D8 crude extract on photosynthesis in H. akashiwo, the Chl a content, carotenoid content, Fv/Fm and rETR were investigated. As shown in Figures 4A,B, the Chl a and carotenoid levels in the algal cells were markedly decreased after treatment with different concentrations of the D8 crude extract for 3, 6, and 12 h and remained at low levels after 24 h of treatment with 5 and 10 μg/ml D8 crude extract; however, the Chl a and carotenoid levels began to recover after 24 h of treatment with 2.5 μg/ml D8 crude extract, suggesting that the toxicity of the D8 crude extract was attenuated and the photosynthetic activity of the remaining algal cells resumed.

The Fv/Fm and rETR values of H. akashiwo cells treated with D8 crude extract showed similar changes. After treatment for 0.5 and 3 h, the Fv/Fm and rETR values of all the treatment groups decreased markedly, especially after treatment with 10 μg/ml D8 crude extract. The Fv/Fm values decreased to only 3% of that of the control, and the rETR values were less than 5% of that of the control, indicating that photosynthesis in the algal cells was damaged. However, with increasing treatment time, the Fv/Fm and rETR values of algal cells in all the treatment groups recovered significantly, and the recovery rate of the low-concentration group was faster than that of the high-concentration group. The rETR values of algal cells in the 2.5, 5, and 10 μg/ml treatment groups recovered to approximately 90, 60, and 30% of that of the control, respectively, after treatment for 6 h (Figures 4C,D). The Fv/Fm values of the 2.5, 5, and 10 μg/ml treatment groups recovered to 94, 86, and 56% of that of the control, respectively, after treatment for 12 h. The results suggested that the D8 crude extract could dramatically inhibit and destroy the photosynthesis of H. akashiwo, but the remaining algal cells could self-restore and recover their photosynthetic function after the toxicity of the D8 crude extract was attenuated.

An oil drain ring showed that there was high correlation between the concentration of the D8 crude extract and the diameter of the oil drain ring (Supplementary Figure 3), indicating that there were biosurfactants in the D8 crude extract. To further identify the substances in the D8 crude extract that played a role in algal cell lysis, we used HPLC to isolate and purify algicidal components. As shown in Supplementary Figure 4, the D8 crude extract contained 4 main components. Algicidal experiments showed that components 2, 3, and 4 had obvious algicidal effects, but component 1 did not. According to the molecular weights in Supplementary Figure 4, components 2, 3, and 4 may be a class of homologues, as the differences between the molecular weights of the three substances are 14, which is equivalent to the mass of a CH₂ group.

To further identify the structures of the three components, we used Q-Exactive high-resolution liquid chromatography-mass spectrometry to explore their specific structures. The full scan mass spectrometry (MS1) of the three compounds showed a series of ion peaks of [M + H]⁺ (at m/z 1008, 1022, and 1036) and [M + Na]⁺ (at m/z 1030, 1044, and 1058) (Figures 5A, 6A, 7A). Interestingly, these ion data were very similar to those reported for surfactin-type lipopeptides in Bacillus. However, due to the existence of many surfactin isoforms, secondary mass spectrometry (MS2) was used to further understand the amino acid compositions of these three components. The [M + Na]⁺ ion peak was chosen for further tandem mass spectrometry analysis. The MS2 spectra of the three compounds showed a series of ion peaks that indicate characteristic surfactin-type peptide fragmentations (Figures 5B, 6B, 7B). The MS2 data of these compounds generated a series of y-type ions and b-type ions, allowing amino acid sequence analysis. The series of y⁺ ions at m/z 836→707→594→481→382→267 represented cleavage along the peptide bonds due to the loss of Glu, Leu, Leu, Val, and Asp from the C-terminus of a C13 β-OH fatty acid, whereas the b⁺ ions at m/z 1030→917→804→689→590→477→364→235 indicated the loss of Leu, Leu, Asp, Val, Leu, Leu, Leu, and Glu from the N-terminus (Figures 5C, 6C, 7C; Supplementary Table 2). According to the rules of cleavage, the amino acid compositions of the three components were identical. Therefore, it could be inferred that the amino acid sequence, starting from the C-terminus, was Glu₁, Leu/IIe₂, Leu/IIe₃, Val₄, Asp₅, Leu/IIe₆, and Leu/IIe₇. However, because leucine and isoleucine have the same molecular weight, NMR technology was needed to further determine the amino acid composition.

Although the molecular weights of leucine and isoleucine are the same, their terminal methyl positions are different, so they exhibit obvious differences in their one-dimensional NMR spectra. In the ¹³C NMR spectrum, the chemical shift of the terminal methyl group in leucine was approximately 21–23 ppm, while that of isoleucine was significantly different: the chemical shift of one methyl group was 11.2 ppm, while that of the other was 15.6 ppm. The ¹³C NMR spectra (Supplementary Table 3 and Supplementary Figures 5–7) of the three algicidal components showed that they all had only peaks with chemical shifts of 22-23 ppm but no peaks with chemical shifts of 11.2 and 15.6 ppm, which suggested that they contained 4 leucines but no isoleucine. In addition, the δ_C of 169.4–174.4 ppm indicated representative signals for carboxyl groups.

We also found some characteristic peaks in the ¹H-NMR spectra (Supplementary Table 3 and Supplementary Figures 5–7). More specifically, these data included a δ_H of 1.20 ppm for a broad singlet peak for a fatty acid chain, δ_H of 4.04–4.95 ppm for multiplet peaks for the C_α of amino acids, and δ_H of 7.76–8.43 ppm for singlet and doublet peaks for “-NH-” units. By combining the data of the ¹H-NMR spectra, ¹³C-NMR spectra and mass spectra and referring to related articles (Tang et al., 2007; Ma and Hu, 2015), the chemical shifts of the ¹H-NMR spectra and ¹³C-NMR spectra of the three algicidal components were collected and are shown in Table 1. In summary, we elucidated that the three components were surfactin-C13, surfactin-C14 and surfactin-C15. The chemical structures of these three algicidal components are shown in Figure 8.

To evaluate the algicidal activities of the three surfactin homologues, we measured their LC₅₀ values against three HAB-causing species, namely, H. akashiwo, S. costatum, and P. donghaiense. The results (Table 1) showed that surfactin-C13 and surfactin-C14 exhibited high algicidal activities against the three algal species. The LC₅₀ values of surfactin-C13 for the three algal species were less than 3 μg/ml. The LC₅₀ values of surfactin-C14 for H. akashiwo and S. costatum were less than 3 μg/ml, while the value for P. donghaiense was higher than 5 μg/ml. The 12 h-LC₅₀ of surfactin-C14 for P. donghaiense was 14.82 μg/ml, and the 24 h-LC₅₀ was 5.31 μg/ml. The algicidal activities of surfactin-C15 were weaker than those of surfactin-C13 and surfactin-C14. Surfactin-C15 had high algicidal activity against S. costatum and no algicidal effect on P. donghaiense. Although surfactin-C15 had an algicidal effect on H. akashiwo, the LC₅₀ value was higher than 50 μg/ml. The results suggested that the fatty acid chain length probably changed the algicidal activity of the surfactin homologues. S. costatum was the most sensitive to all three surfactins.