Introduction

Harmful algal blooms (HABs) have experienced an increase in their occurrence, intensity, and geographical distribution on a global scale, resulting in adverse environmental, health, and socioeconomic impacts (Marampouti et al., 2021). HABs have a considerable impact on human health as a result of direct exposure to volatile toxins or by toxic seafood consumption (Burkholder et al., 2018). From an ecological point of view, the expansion of HABs can result in the erosion of biodiversity, because they cause massive mortality of marine species and they are generally monospecific in nature (Chai et al., 2020). In coastal areas, understanding the biological interactions that control toxic algal blooms is therefore a major ecological challenge. Among HABs, a number of Alexandrium species have been placed on the list of invasive Mediterranean species. Among them, Alexandrium pacificum is a flagellated eukaryotic unicellular organism that form with Alexandrium tamarense and Alexandrium fundyense the “Alexandrium tamarense” complex (Hadjadji et al., 2020). Since 1998, A. pacificum (former A. catenella) was monitored by Rephy survey in the Thau lagoon (French Mediterranean) because it produces paralytic shellfish toxins (PSTs) resulting of paralytic shellfish poisoning (PSP) syndrome.

Laania (Laanaia et al., 2013) showed that in Thau lagoon, a water temperature around 20°C for several days and organic and inorganic nutrients in sufficient concentrations are parameters favoring the developemnt of A. pacificum, whose massive blooms occur in autumn.

Algal blooms are seasonal events resulting in a rapid increase in the concentration of a species of algae in an aquatic environment. Depending on the phytoplankton species, the tolerance range of physicochemical parameters is different and influences the time of appearance of blooms (Leblad et al., 2020).

Interestingly, although the collapse of phytoplankton blooms has been previously attributed to viruses (Pal et al., 2020), some ecological studies have suggested an important role of algicidal bacteria (Su et al., 2007; Wang et al., 2010), such as some Vibrio spp. (Li et al., 2014; Wang et al., 2020).

Vibrio (class γ-proteobacteria) are common microorganisms in marine systems worldwide (Baker-Austin et al., 2017; Mavian et al., 2020), where they are important components of the food chain, particularly in biodegradation, nutrient regeneration and biogeochemical cycles (Oberbeckmann et al., 2012). Vibrio is one of the most studied bacterial taxa due to their ubiquity in coastal marine systems and their capacity to cause infections in humans and animals, leading sometimes to epizootic or zoonotic epidemics (LeRoux et al., 2015; Mavian et al., 2020). Vibrio are extremely adaptable to their environment (Johnson, 2013). The main factors influencing their occurrence and distribution in water are temperature, salinity, nutrient availability (Wang et al., 2020), multiple strategies such as biofilm formation on biotic and abiotic surfaces (Espinoza-Vergara et al., 2020), or interactions with a multitude of other organisms such as eukaryotic predators (Drebes Dörr and Blokesch, 2020) or plankton (Lopez-Joven et al., 2018) are used by Vibrio in the environment. There is also evidence that global climate change has increased Vibrio-associated illnesses affecting humans and animals (Brumfield et al., 2021; Muhling et al., 2017). However, the drivers and dynamics of Vibrio survival and propagation in the marine environment are not yet fully understood.

A substantial number of research articles have highlighted the potential of γ-proteobacteria to exert algicidal activity against dinoflagellates, supporting the hypothesis that γ-proteobacteria such as Vibrio play a role in the control of algae blooms in situ (Coyne et al., 2022). However, the mechanisms behind Vibrio-driven algal lysis in the environment remain to be elucidated. Particularly, it is unclear how in the water column, algicidal compounds secreted by bacteria can concentrate around the algae to exert their lytic effect.

Based on observations from the natural environment showing a potent relationship between Vibrio and Alexandrium algae bloom events, this study aim to determine in vitro, the main factors implicated in this relationship. Using a combination of biochemical, proteomic, genetic, and fluorescence microscopy approaches, we explored the role of algal toxicity, bacterial pathogenicity and the quorum sensing pathway on this relationship and showed the important role of nutrient stress and the iron uptake pathway in this unique Vibrio/Alexandrium predator/prey interaction.

Methods

Quantification of Alexandrium algae and Vibrio bacteria in the environment

Seawater samples were collected in the Thau Lagoon (southern France, a shallow Mediterranean ecosystem open to the sea (Abadie et al., n.d.), during spring and autumn 2015. Briefly, samples were collected from the subsurface (-50 cm) near an oyster table at a phytoplankton surveillance site (part of the REPHY network, N 43°26.058’ and E 003°39.878’). Once a week during spring and autumn 2015, during field sampling campaigns, 20 L of water was filtered on board through a 180 μm pore-size nylon membrane. At the laboratory, according to Lopez-Joven et al. (Lopez-Joven et al., 2018) seawater was fractionated into two size classes as follows: 2 L of the above filtrate was filtered through a 0.8 μm pore-size polycarbonate Whatman Nuclepore membrane to obtain organisms in the 0.8–180 μm range corresponding to plankton-associated Vibrio and living Alexandrium forms. Then, the filtrate from the 0.8 µm membrane was filtered again through a 0.2 μm pore-size polycarbonate Whatman Nuclepore membrane until the membrane was saturated. Alexandrium cells, ranging from 25 to 40 µm, belong to the microphytoplankton and are therefore retained in the 0.8–180 µm fraction. Any Vibrio cells potentially associated with or attached to Alexandrium cells will also be retained in this fraction. Vibrio cells are approximately 0.5-0.8 µm-thick. The fraction between 0.2 and 0.8 µm therefore includes the free-living Vibrio. The bacterial population collected on 0.8-µm-pore-size filters was designated the particle-associated community, and the population on 0.2-µm-pore-size filters was designated as the free-living community. Membranes (in triplicate) were then conserved in 500 µL of 100% EtOH at -20°C. Environmental DNA (eDNA) was extracted from the MF Millipore membrane using the Macherey-Nagel NucleoSpin Tissue Kit and resuspended in 100 µL of water. The samples were then stored at -20°C after eDNA quantity and purity were assessed using a NanoDrop system (NanoDrop Technologies, Wilmington, DE, USA). PCR amplification reactions were done on a Roche LightCycler 480 Real-Time thermocycler (qPHD platform, University of Montpellier, France).using specific primer pairs (Table 3). Typically, the reactions contained 1 µL of template DNA (the DNA concentration for all samples varied from 1 to 40 μg mL⁻¹), 0.5 µL of each primer (3.33 µM) and 4 µL of reaction mixture (SYBR Green Master Mix) in a total volume of 6 μL. The reaction parameters were as follows: 5 minutes at 95°C (initial denaturation) and 40 cycles of 10 s at 95°C (denaturation), 10 s at the corresponding hybridization temperature (Table 3) and 10 s at 72°C (elongation). Melting curve profiles were generated by increasing the temperature from 65°C to 95°C at 0.5°C s^-1. Amplification products were analysed using LightCycler software (Roche Diagnostics). Vibrio spp. and A. pacificum and A. tamarense were quantified by constructing calibration curves based on DNA from the V. atlanticus LGP32 reference strain (former V. tasmaniensis LGP32) and from the A. pacificum reference strain (ACT03: A. catenella strain isolated form Thau in 2003) and the A. tamarense reference strain (ATT07: A. tamarense isolated from Thau in 2007) (not shown).

Strains and growth conditions

Vibrio strains. Wild-type and isogenic mutants of Vibrio atlanticus LGP32 (Table 1) were used in this study. Deletion-mutants included ΔluxR, ΔluxP, ΔluxM and ΔpvuB isogenic strains. The ΔpvuB mutant was constructed here by allelic exchange as described previously by Le Roux (Le Roux et al., 2007). We also used V. atlanticus LGP32 carrying the pSW3654T-GFP plasmid (Le Roux et al., 2007), hereafter referred to as V. atlanticus LGP32-GFP. Bacterial strains were grown at 22 ± 1°C in Zobell medium (0, 38 µM iron (III)). When needed, 25 μg mL^-1 chloramphenicol (Cm) was added to cultures of V. atlanticus LGP32-GFP (Le Roux et al., 2007). Oligonucleotides used for RNA expression analysis are specified in table 3.

Table 1.

Phytoplancton strains. Non-axenic phytoplankton species (Table 2) were grown in batch culture in enriched natural sea water (ENSW, 6,55 µM iron (III)) with a salinity of 36 practical salinity units (PSU) at 22 ± 1°C under cool white fluorescent illumination (100 µmol photons m^-2 s^-1) and a 12 h:12 h light:dark cycle (Harrison et al., 1980). The algae were used for experiments in their exponential growth phase.

Table 2.

Co-culture assay

For each tested phytoplankton species, 2.10⁴ cells harvested in their exponential growth phase were placed in 20 mL of ENSW medium in a 50 mL suspension culture flask (Cellstar® PS, Greiner bio-one). After incubation for 24 h at 22 ± 1°C under cool white fluorescent illumination, 40 µL of Vibrio strains grown for 12, 36, 60 and 156 h in Zobell medium or in Zobell medium supplemented with Fecl₃ (6 µM iron (III)) or with H₃BO₃ (0.47 mM) or the corresponding culture supernatant were added to the algal culture. After incubation at 22 ± 1°C under cool white fluorescent lights, living, non-swimming, group-attacked and lysed phytoplankton cells were counted in a sedimentation chamber under an inverted microscope. The number of lysed cells corresponded to algae showing disrupted membranes. Non-swimming algae were not counted as lysed cells. For Vibrio analysis, 100 µL of a 1:10 serial dilution mixture in ENSW (from 10^-2 to 10^-10) was plated on Vibrio Selective TCBS (thiosulfate-citrate-bile salts-sucrose) agar (in triplicate). After incubation for 24 h at 22 ± 1°C, the number of living Vibrio cells was determined by counting colony-forming units (CFUs). Data are the means of three independent experiments.

Microscope observations

The dynamic of the interaction between A. pacificum ACT03 and V. atlanticus LGP32, which is an oyster pathogen isolated from the French Atlantic coast (Gay et al., 2004) and present in the Thau Lagoon, south of France (Lopez-Joven et al., 2018)) was surveyed. As the algal strain used in the study is not axenic, means that additional bacteria, other than the V. atlanticus LGP32, are present in the experiments. To elucidate the interaction without thoroughly accounting for the non-axenic cells, interaction was observed under a Zeiss Axio upright fluorescence microscope equipped with an AxioCamMRm 2 digital microscope camera using V. atlanticus LGP32 tagged with green fluorescent protein (GFP). Lasers were used at excitation wavelength (λex) 488 nm for GFP (emission wavelengths (λem): 505–530 nm) and λex 532 nm for plankton chloroplasts (λem 560–630 nm). Images were taken sequentially to avoid cross-contamination between fluorochromes. Sequences of images were merged during the Vibrio-Alexandrium interaction using ZEN 2012 (blue edition) software. Interaction events between Vibrio strains and phytoplankton strains were also observed under a Leica TCS SPE confocal laser scanning system connected to a Leica DM 2500 upright microscope camera (Montpellier RIO Imaging Platform, University of Montpellier, France).

Comparative proteomic analysis

Vibrio sampling and protein extraction. V. atlanticus LGP32 was grown for 60 h at 22°C in artificial seawater (high nutrient stress, cond. 1) or 12 h at 22°C in Zobell media (low nutrient stress, cond. 2). After 10 min centrifugation at 8000 rpm, crude protein extracts of V. atlanticus LGP32 in each culture condition (triplicates) were obtained by sonication on ice at 20% amplitude for 20 s in 200 µL of ice-cold denaturing buffer (7 M urea, 2 M thiourea, 4% CHAPS in 30 mM Tris-HCl, pH 8.5) and clarified by centrifugation at 2000 x g, 15 min, 4°C. The protein concentration of the supernatant was estimated using the 2D Quant Kit (Cytiva™, MERCK) and samples were stored at -80°C until use.

Bi-dimensional gel electrophoresis (2D gel). Proteins extracts were individually analysed on 2D gel electrophoresis (6 gels per condition each corresponding to different biological replicates). To do so, 100 µg of proteins from each extract was added to rehydration buffer (7 M urea, 2 M thiourea, 4% CHAPS, 65 mM DTT) for a total volume of 350 µL. They were then individually loaded onto 17 cm isoelectric focusing strips (Bio-Rad) with a stabilized non-linear pH ranging from 3 to 10. Due to the high complexity of the protein profile in the acidic part (left) of the gel pH 3–10 (Fig. S3), we conducted additional ‘close-up’ analyses in gels using 17 cm isoelectric focusing strips (Bio-Rad) with a narrower, stabilized pH gradient ranging from 4 to 7. Strips were rehydrated passively for 5 h at 22°C, followed by active rehydration for 14 h under a 50 V current at 22°C (to help large proteins enter the strips). Thereafter, isoelectrofocusing was carried out using the following programme: 50 V for 1 h, 250 V for 1 h, 8000 V for 1 h and a final step at 8000 V for a total of 140 000 V.h with a slow ramping voltage (quadratic increasing voltage) at each step. Focused proteins were reduced by incubating the strip twice in equilibration buffer (1.5 M Tris, 6 M urea, 2% SDS, 30% glycerol; bromophenol blue, pH 8.8) containing DTT (130 mM) at 55°C. Then, they were alkylated by incubation with equilibration buffer containing iodoacetamide (135 mM) on a rocking agitator (400 rpm) at room temperature protected from light. Proteins were also separated according to their molecular weight (second dimension) on 12% acrylamide/0.32% piperazine diacrylamide gels run at 25 mA per gel for 30 min followed by 75 mA per gel for 8 h using a Protean II XL system (Bio-Rad). Gels were stained using an MS-compatible silver staining protocol and scanned using a ChemiDoc MP Imaging System (Bio-Rad) associated with Image Lab software version 4.0.1 (Bio-Rad).

Comparative bioinformatics analysis of 2D gels. Twelve gels (six per condition) were selected for comparative analysis on PD-Quest v. 7.4.0 (Bio-Rad) to identify changes in protein abundance between the proteomic profiles of V. atlanticus LGP32 cultured in contrasting nutrient conditions (ENSW/Zobell). Spots whose mean intensity across six replicates per strain was two times higher or lower than those from the other strain, with a P <0.01 (Mann-Whitney U-test), were considered significantly different in terms of abundance between the two conditions (quantitative difference). Differentially represented spots were then excised from the gels, destained, trypsin-digested and the obtained peptides were identified by tandem mass spectrometry (MS-MS) using the PISSARO platform facility (University of Rouen, France). To identify protein(s) present in each spot, the obtained peptides were compared with V. atlanticus LGP32 reference genome (https://vibrio.biocyc.org/). The genes whose peptides matched strongly were retrieved and used for an BLASTx query against non-redundant databases to determine the protein identity of the best match. A gene was considered as strongly matched when at least two peptides matched the sequence with a coverage of> 6%. Their theoretical isoelectric point (pI) and molecular weights were also calculated using the Expasy server (https://www.expasy.org/) to compare them with the location of the spot on the gel. Altogether, these complementary analyses made it possible to characterize the protein identity of each spot with confidence.

Gene expression analysis

Vibrio sampling and RNA extraction. V. atlanticus LGP32 was grown in Zobell media for 36 h and 60 h at 22°C (decline phase of growth, nutrient stress) or for 12 h at 22°C (exponential grow phase, poor nutrient stress). Total RNA was isolated from V. atlanticus LGP32 using the standard TRIzol method (Invitrogen Life Technologies SAS, Saint-Aubin, France) and then treated with DNase (Invitrogen) to eliminate genomic DNA contamination. After sodium acetate precipitation, the quantity and quality of the total RNA were determined using a NanoDrop spectrophotometer and agarose gel electrophoresis. Following heat denaturing (70°C for 5 minutes), reverse transcription was performed using 1 μg of RNA prepared with 50 ng μL^-1 oligo-(dT) 12–18 mer in a 20-μL reaction containing 1 mM dNTPs, 1 unit μL^-1 RNAseOUT and 200 units μL^-1 Moloney murine leukaemia virus reverse transcriptase (M-MLV RT) in reverse transcriptase buffer, according to the manufacturer’s instructions (Invitrogen Life Technologies SAS, Saint-Aubin, France).

PCR amplification. Amplification reactions were analysed using a Roche LightCycler 480 Real-Time thermocycler (Bio-Environnement platform, University of Perpignan, France). In this study, several PCR primer pairs were designed using Primer3 software (optimal primer size: 20 bases; Tm: 60°C; primer GC%: 50; 2GC clamp and product size range: 150–200 bp) and calibrated with V. atlanticus LGP32 genomic DNA (Table 1). To determine the qPCR efficiency of each primer pair used, standard curves were generated using seven serial dilutions of genomic DNA (10^-2, 10^-3, 10^-4, 10^-5, 10^-5, 10^-7 and 10^-8) (not shown); the qPCR efficiencies of the tested genes varied between 1.85 and 2.08 (Table 3). For gene expression, reverse transcription was performed with 1 µg of total RNA using random hexamers and SuperScript IV reverse transcriptase (Invitrogen). The total qPCR reaction volume was 10 μL and consisted of 5 μL of cDNA (diluted 1:5), 2.5 μL of SensiFAST SYBR No-ROX Mix (Bioline) and 100 nM or 300 nM PCR primer pair (Table 3). The reaction parameters were as follows: 2 min at 95°C (initial denaturation) and 40 cycles of 5 s at 95°C (denaturation), 10 s at 59°C (annealing) and 20 s at 72°C (elongation). The specificity of each PCR was checked by measuring fluorescent signals during melting curve analysis (PCR product heated from 65°C to 95°C continuously and slowly at 0.1°C s⁻¹). Relative expression was calculated by normalization to the expression of two constitutively expressed housekeeping genes, namely, 6PKF (VS_2913) and CcmC (VS_0852), using the delta-delta threshold cycle (ΔΔCt) method(Pfaffl, 2001).

Table 3.

Detection of quorum-sensing signalling molecules

Vibrio culture. To detect the QS molecules (AI-2, AI-1 and CAI-1) V. atlanticus LGP32 was grown in Zobell media for 12 h (exponential growth phase, control) and 60 h (decline phase of growth, nutrient stress).

AI-2 analysis. Bioluminescence assay using the QS bioluminescent of Vibrio campbellii MM32 (luxN::Cm, luxS::Tn5Kan) was used to detect AI-2 molecules in culture supernatants. Briefly, Vibrio cultures were centrifuged at 17,000 x g for 10 min, and the resulting supernatants were filtered on 0.22 μm. Then 20 μL of the filtrates were mixed with 180 μL of V. campbellii MM32 diluted 1:5000 then incubated at 30°C and 100 rpm. Luminescence and cell density (OD620) were collected in triplicate and analysed according to Tourneroche et al.(Tourneroche et al., 2019).

AI-1 and CAI-1 extraction and LC-MS analysis. Chemical analyses were conducted with a Q Exactive Focus Orbitrap System coupled to an Ultimate 3000™ ultrahigh-performance liquid chromatography (UHPLC) system (Thermo Fisher Scientific) according to Rodrigues et al. (Rodrigues et al., 2022). Briefly, ethyl acetate (2 mL) was added into each culture (2 mL). This mixture was shaken overnight at room temperature (150 rpm). The two phases were then separated and the aqueous phase was extracted once again. The two obtained organic phases were pooled and the solvent was evaporated under vacuum. The crude extracts were dissolved in 500 µL LC-MS grade methanol for analysis. All experiments were conducted in triplicate. Analyses of extracts and standards (3 μL injected) were performed in electrospray positive ionization mode in the 50–750 m/z range in centroid mode. The parameters were as follows: spray voltage: 3 kV; sheath flow rate: 75; aux gas pressure: 20; capillary temperature: 350°C; heater temperature: 430°C. The analysis was conducted in Full MS data-dependent MS2 mode (Discovery mode). Resolution was set to 70,000 in Full MS mode, and the AGC (automatic gain control) target was set to 1x10⁶. In MS2, resolution was 17,500, AGC target was set to 2x10⁵, isolation window was 0.4 m/z, and normalized collision energy was stepped to 15, 30 and 40 eV. The UHPLC column was a Phenomenex Luna Omega Polar C18 1.6 μm, 150 x 2.1 mm. The column temperature was set to 42°C, and the flow rate was 0.4 mL min^-1. The solvent system was a mixture of water (A) with increasing proportions of acetonitrile (B), with both solvents modified with 0.1% formic acid. The gradient was as follows: 1% B 3 min before injection, then from 1 to 15 min, a gradient increase of B up to 100% (curve 5), followed by 100% B for 5 min. The flow was injected into the mass spectrometer starting immediately after injection. All data were acquired and processed using FreeStyle 1.5 software (Thermo Fisher Scientific).

Chemicals and solvents. N-acyl-homoserine lactones (AHL) were obtained from Cayman Chemical (Ann Arbor, MI, USA). Stock solutions were obtained by dissolving standards in methanol or dichloromethane (C18-AHL) at a concentration of 1 mg mL^-1 and stored at -80°C. Standard solutions for UHPLC-high-resolution tandem mass spectrometry (HRMS) analyses were prepared by diluting each individual standard solution with methanol in order to obtain a concentration range from 2000 to 20 ng mL^-1. LC-MS grade methanol, acetonitrile and formic acid were purchased from Biosolve (Biosolve Chimie, Dieuze, France), analytical-grade ethyl acetate was obtained from Sigma-Aldrich. Pure water was obtained from Elga Purelab Flex System (Veolia LabWater STI, Antony, France).

Nature of lytic compounds secreted by V. atlanticus LGP32

To determine the nature of the lytic compounds secreted, V. atlanticus LGP32 grown for 60 h in Zobell media at 22 °C was filtered through a 10 kDa membrane (Amicon^® Ultra-4 filter unit). The eluate containing molecules with MW below 10 KDa was then incubated in a water bath at 100°C for 30 min. Boiled filtrates (0.1 mL) were subsequently used to inoculate A. pacificum (ACT03 strain) cultures, then lytic activity was observed under the Leica TCS SPE confocal laser scanning system. Zobell media with the same treatment was used as control.

Statistical analysis

Environmental data. Statistical analyses were performed using R 3.6.3 software. The relationship between Alexandrium and Vibrio was explored separately in spring and autumn. We used a generalized linear model specifying a Gaussian family. For spring, the dataset for salinity and temperature was complete (14 periods of observation). One influential period was detected and removed from the dataset whenever no dead Alexandrium cells were observed. The effects of explanatory variables such as log10 (Vibrio+1), salinity and temperature were centred, reduced and tested as fixed effects with a linear relationship. Model selection was performed using the Akaike information criterion corrected for small sample size (AIC_c). Models were considered different whenever the difference between their AIC_c value and the lowest AIC_c value (ΔAIC_c) was lower than 2 (Burnham and Anderson, 2002). Alexandrium distribution and model residuals were checked for normal distribution assumptions (QQ plot and Shapiro-Wilk test). For autumn, the dataset was complete for 10 periods of observation. Salinity and temperature were missing for three periods. We explored the relationship between Alexandrium and Vibrio alone using the method detailed above for spring.

In vitro data. Statistical analyses were performed using one-way ANOVA (analysed by pair) followed by Tukey’s test (Statistica 10.0 software, StatSoft, Maison-Alfort, France). *P <0.05, **P <0.01, ***P <0.001

Results

Concomitant occurrence of A. pacificum ACT03, A. tamarense ATT07 and free-living Vibrio spp. in the Thau lagoon

In the spring and autumn of 2015, in the Thau Lagoon (Fig. 1A), we detected Alexandrium algae (A. pacificum ACT03 and A. tamarense ATT07, both alive and in degraded cell forms) and free-living Vibrio, but no plankton-associated Vibrio were observed (Fig. 1B, Data S1). Using model selection based on AICc, we found no significant relationship between Alexandrium (A. pacificum ACT03, A. tamarense) and Vibrio Spp. abundances in autumn. This result is consistent with the difficulty that Vibrio has in growing at temperatures below 20°C and with the complex interacting factors driving bloom dynamics (Laanaia et al., 2013). Interestingly, in spring 2015, the mean densities of Alexandrium and of free-living Vibrio were positively correlated. The lowest AICc was obtained with the model explaining degraded form of Alexandrium density based on the free Vibrio density (Fig. 1C). Given that, this model is not so different from the model with only the intercept, but better than any other linear combination with other potentially interfering drivers, such as temperature and salinity (Fig. 1D), we searched for evidence of a relationship between Vibrio and Alexandrium by studying their interaction in vitro.

Figure 1.

Vibrio atlanticus LGP32 feeds on Alexandrium pacificum ACT03

To investigate whether Alexandrium interacts with Vibrio, we incubated in Enriched Natural SeaWater (ENSW) A. pacificum ACT03 with V. atlanticus LGP32 previously grown for 12 hours in Zobell media. In interaction with V. atlanticus LGP32, A. pacificum ACT03 cell abundance decreased significantly from 2.10 x 10⁴ cells mL^-1 to 1.07 x 10⁴ cells mL^-1 after 48 h (Fig. 2A), while the V. atlanticus LGP32 concentration grew significantly after 26 h of interaction, reaching a maximum peak density of 7.67 x 10⁷ CFU mL^-1 at 34 h (Fig. 2B). In the control experiment where A. pacificum ACT03 was cultured alone in ENSW, the algal concentration remained stable over time (Fig. 2A) and no bacteria were on the corresponding TCBS plates (Vibrio selective medium). In the control where V. atlanticus LGP32 was cultured alone in ENSW, the concentration of the bacteria tended to increase during the first 24 h (from 2.0 x 10⁵ to 7.0 x 10⁵ CFU mL^-1) and then decreased at 48h, below the initial starting concentration (Fig. 2B). These results show that the interaction between V. atlanticus LGP32 and A. pacificum ACT03 leads to a decline in the algal population and promotes the growth of V. atlanticus LGP32. This suggests that V. atlanticus LGP32 is able to feed on A. pacificum ACT03.

Figure 2.

V. atlanticus LGP32 performs group-attacks on A. pacificum ACT03

Epifluorescence microscopy observation of GFP-labelled V. atlanticus LGP32 (previously grow in Zobell medium) in interaction showed that A. pacificum ACT03 cells that had lost their motility were attacked individually by groups of V. atlanticus LGP32 before being lysed (Movie 1). Attacks were extremely rapid, with empty thecae (algal envelopes) observed in the medium after less than 60 s (Movie 2). During the attack, V. atlanticus LGP32 did not invade the algal cell but remained clustered on the cell surface (Fig. 3A2).

Figure 3.

Group attack of A. pacificum ACT03 is activated by V. atlanticus LGP32 starvation

In 2002, Martin hypothesized that nutritional stress induces bacteria to lyse algae (Martin, 2002). To test this hypothesis, we monitored V. atlanticus LGP32 behaviour in response to starvation (Fig. 3). We observed that V. atlanticus LGP32 in exponentially growth phase (12 h of culture in Zobell medium) did not interact with A. pacificum ACT03 cells for the first hour of contact (Fig. 3A, B, C). In contrast, V. atlanticus LGP32 in the decline phase (36 h of culture in Zobell medium) induced a significant decrease in the number of motile algae cells by 8.9% after 15 min and by 43.3% after 60 min (Fig. 3A). This phenomenon corresponded to the degradation and/or disruption of algal flagella (Movie 3). The flagella no longer functioned correctly, which caused erratic swimming of the algae (Video 3, left cell). This was followed by a complete cessation of swimming. When the flagellum detached from the algae (Video 3, right cell), the attack occurred. With starved V. atlanticus LGP32 (60 h of culture in Zobell medium), algae immobilization was fast and significant (91.4% in 15 min, Fig. 3A), and algae were attacked individually being targeted by groups of V. atlanticus LGP32 (Movie 4). The percentage of cells attacked and killed peaked at 30% after 15–30 min of contact (Fig. 3B) and then decreased. After 1 h, attacks had stopped with approximately 40% of the algal cells still alive (Fig. 3C). Although it remains unclear whether the attacks occur during a specific phase of growth, it is evident that the cells are already weakened before attack as they have all lost their flagella. An old-starved culture of V. atlanticus LGP32 (126 h) significantly immobilized A. pacificum ACT03 cells within a few minutes, with lysis occurring immediately (Fig. 3A, C), making it impossible to detect attacks by V. atlanticus LGP32 (Fig. 3B). The lysis phase corresponded to initial vesicle formation followed by the bursting of A. pacificum ACT03 cells (Movie 5) and was induced by the old-starved culture supernatant of V. atlanticus LGP32 (Fig. S1).

We next tested whether this lytic effect was mediated by thermostable molecule (s) secreted by Vibrio. The culture supernatant of starved culture of V. atlanticus LGP32 (36 h) filtered through a 10 kDa membrane and then incubated at 100°C for 30 min still possessed its lytic properties, indicating that the algicidal compounds produced by V. atlanticus LGP32 are small thermostable molecules unlikely to be lytic enzymes, or lysins able to digest the algae cell. However, these experimental observations clearly show the key role of nutrient limitation in triggering the ‘group-attack’ behaviour and the secretion of lytic compounds of V. atlanticus LGP32.

Group attack occurs on A. pacificum ACT03 in exponential phase of growth

Here, we wondered whether the live/dead status of algae is important for V. atlanticus LGP32-mediated group attacks targeting. To this aim, A. pacificum ACT03 in exponential growth phase was first exposed for 30 min to supernatant from 60 hours starved V. atlanticus LGP32 Zobell media that induced 25% lysis of A. pacificum ACT03 cells and next to the corresponding V. atlanticus LGP32 cells. Group attacks were observed on non-degraded A. pacificum ACT03 cells, but not on lysed cells. This result is similar to what is observed on the Movie 1 with flash attacks only on immobilized, but not degraded A. pacificum ACT03 cells (red), and not on lysed cells (green). In addition, no group attacks occurred on cells from an old A. pacificum ACT03 culture (1-month culture). Together, with the very short duration of group attacks (Movie 1), these results indicate that V. atlanticus LGP32 group-attacked exponentially growing cells of A. pacificum ACT03, but not decomposing algae, suggesting that this behaviour is not just an opportunistic response of heterotrophic bacteria to organic substrates.

Group attack is independent of quorum sensing

Considering the coordinated action of V. atlanticus LGP32 in attacks on A. pacificum ACT03, we tested whether the group-attack process depended on the key physiological mechanism that regulates many functions in marine microbial cells, quorum sensing (QS) (Lami, 2019; Papenfort and Bassler, 2016), a type of cell-cell communication. Although QS is a cell-density-dependent mechanism, our results showed no attack from a 12 h culture of V. atlanticus LGP32 up to a concentration of 4.10⁶ Vibrio mL^-1 (Fig 4A). Attacks were only observed with V. atlanticus LGP32 from a 60 h culture at low concentration of 5.10³ Vibrio mL^-1 to the highest concentration tested of 5.10⁵ Vibrio mL^-1 (Fig 4A), consistent with the hypothesis that the attacks were independent of Vibrio density.

Figure 4.

The analysis of the expression of genes involved in the known QS pathways in Vibrio cell (Fig. S3A), highlighted that only the AI-2 pathway was induced during nutrient stress of V. atlanticus LGP32, because only the expression of the AI synthase (LuxS) and its receptor (LuxP) increased significantly (Fig. 4B; ANOVA p <0.05). This was confirmed by a QS bioluminescence assay, which showed a AI-2 molecules (unquantified) in the Zobell culture supernatant of V. atlanticus LGP32 after 60 h of culture but not after 12 h of culture and not in the ENSW supernatant of V. atlanticus LGP32 after 12 or 60 h of culture. UHPLC-HRMS/MS provided no evidence of detectable HAI-1) and CAI-1 in any experiments. By targeted mutagenesis of key genes involved in QS pathways ΔluxM (HAI-1 production), ΔluxS (AI-2 production) and ΔluxR (high-density QS master regulator) did not lead to any change in the group-attack behaviour of V. atlanticus LGP32 (Fig. 4C). Taken together these results showed that group-attack by V. atlanticus LGP32 is not coordinated by QS.

Group attack related to the availability of iron

The comparative analysis of the proteome of V. atlanticus LGP32 incubated 60 h in artificial seawater (ENSW) versus V. atlanticus LGP32 grown 12 h in Zobell nutrient-rich medium revealed 10 proteins modulated by nutrient stress (Fig. S2). The two most down-regulated proteins correspond to be ß-ketoacyl-(acyl-carrier-protein) synthase II (-22-fold in ENSW compared to Zobell), a key regulator of bacterial fatty acid synthesis, and the dihydroorotase (-6.6-fold in ENSW compared to Zobell), an enzyme essential for pyrimidine biosynthesis and thus bacterial proliferation and growth. The low expression of these proteins in ENSW is consistent with V. atlanticus LGP32 nutritional starvation. The most up-regulated protein in starved V. atlanticus LGP32, with an increase of more than 6-fold, was glucosamine-6-phophate deaminase, an enzyme involved in bacterial energy metabolism probably necessary for its survival. Among the other up -regulated proteins, one was an iron siderophore-binding protein (Spot 4413, Fig. S2A) corresponding to the vibrioferrin outer membrane receptor PvuB, whose gene is part of the pvu operons involved in iron transport (Fig. S3B). Interestingly, the corresponding gene pvuB as well as the vibrioferrin membrane receptor gene pvuA2 (Fig. S3B) were both significantly induced in Vibrio under nutrient stress (Fig. 4B; ANOVA p <0.01) but not the one involved in the vibrioferrin biosynthesis, pvsA (Fig. 4B). Remarkably, among the 10 proteins identified by proteomic analysis only V. atlanticus LGP32 mutant lacking pvuB failed to attack A. pacificum ACT03 (Fig. 4C; ANOVA p <0.001). In the absence of the pvuB gene, V. atlanticus LGP32 was unable to attack in group A. pacificum ACT03. In addition, V. atlanticus LGP32 cells that had been washed with ENSW to remove their culture supernatant metabolites also failed to attack A. pacificum ACT03 (Fig. 4C; ANOVA p <0.001), which is congruent with the hypothesis that attacks depend on the V. atlanticus LGP32 vibrioferrin transport system. Finally, group-attacks increased significantly, when FeCl₃ was added to the Vibrio culture medium (Fig. 4D) but not with H₃BO₄ (Fig. 4D). Taken together, those results are consistent with the hypothesis that attacks are regulated by iron.

Group attack is a Vibrio spp. behavior specific to Alexandrium spp

To evaluate the dinoflagellates specificity of the group-attack behaviour, a selection of Vibrio spp. was co-cultured with a selection of dinoflagellate strains commonly found in the Mediterranean Sea. The results showed that, among the Vibrio spp. tested (pathogenic or not) all, under nutrient stress, were able to secrete algicidal compounds, immobilize, ‘group-attack’ and lyse A. pacificum ACT03 cells (Table 1) and no tropism linked to their pathogenesis for fish of invertebrates was observed (Table 1). Among the sixteen dinoflagellates species tested, all eight Alexandrium spp. (non-toxic and paralytic shellfish toxin (PST) producers) and Gymnodinium catenatum (PST producer) were immobilized, ‘group-attacked’ and lysed by V. atlanticus LGP32 (Table 2), but no tropism linked to PSTs was revealed (Table 2).

Discussion

Predation is a widespread mode of interaction for survival in the natural world (Finke and Denno, 2004; Sinclair et al., 2003). Among predators, predatory bacteria are found in a wide variety of environments, and like bacteriophages and predatory protists, they have been reported to prey exclusively on other bacteria (Johnke et al., 2014; Perez et al., 2016). Considering predator as a free organism that feeds at the expense of another, this study is the first evidence of the capacity of some Vibrio to develop a predatory strategy against an alga. This behaviour differs from parasitism, because the survival of Vibrio is not exclusively dependent on the algae.

The strategy developed by Vibrio to kill algae may be reminiscent of strategies previously described in the prokaryotes (Johnke et al., 2014). As shown in Movie 1, the interaction between V. atlanticus LGP32 and A. pacificum ACT03 proceeds in three stages (Fig. 5). The first stage, the ‘immobilization stage’, recalls the strategy used by Streptomyces to immobilize its prey (Kumbhar et al., 2014) based on the secretion of algicidal metabolites that disrupt the flagella. The second stage, the ‘attack stage’, resembles the ‘wolf-pack attack’ strategy described for Myxococcus xanthus and Lysobacter (Martin, 2002; Perez et al., 2014). V. atlanticus LGP32 also surrounds A. pacificum ACT03 cells at high density for a very short time, but does not invade the algal cell. This phenomenon is comparable to that of bacteria clustering around lysed ciliate cells (Blackburn et al., 1998). The third stage, the ‘killing stage’, is similar to that of epibiotic bacterial predators, which induces the lysis of bacteria from the outside (Rashidan and Bird, 2001). Overall, these observations show that V. atlanticus LGP32 is able of wolf-pack hunting behaviour. Results showed that quorum sensing does not control algal predation by Vibrio, in contrast to iron acquisition systems, which play a major role in regulating this phenotype, as we demonstrated genetically. Indeed, our quorum sensing mutants exhibited the same predatory phenotype as the wild-type Vibrio atlanticus LGP32, whereas mutants impaired in iron uptake completely lost this phenotype. However, quorum sensing and iron acquisition are sometimes interconnected in Vibrio (McRose et al., 2018). For instance, in Vibrio vulnificus, the production of vulnibactin (a siderophore) is controlled by AI-2 (Kim and Shin, 2011). Whether the production of AI-2 by V. atlanticus LGP32, induced during group attacks of A. pacificum ACT03, promotes the production of vibrioferrin (Lami, 2019) remains to be investigated.

Figure 5.

We further show that the group-attack behaviour of V. atlanticus LGP32 depends on iron availability. Iron is an essential element for growth in most organisms, including phytoplankton (Martin and Fitzwater, 1988) and bacteria (Neilands, 1981), and its concentration in seawater is known to be very low with measurements in the open ocean surface at 0.2 nM and in deep waters at 0.6 nM (Millero, 1998).

Moreover, Its low solubility in seawater limits its availability (Bruland et al., 1994; Wu and Luther, 1994). To acquire iron, bacteria have developed systems based on the secretion (and subsequent uptake) of iron-chelating siderophores to obtain this element from the environment (Amin et al., 2009a). Therefore, many Vibrio spp. produce a siderophore known as vibrioferrin, which is synthesized and secreted by proteins encoded by the pvsABCDE gene cluster (Fig. S3). Boron being known to be a regulator or capable of being transported by vibrioferrin (Romano et al., 2013; Weerasinghe et al., 2013) we tested its potential involvement in the interaction but no effect was evidenced here. For iron-vibrioferrin uptake, Vibrio parahaemolyticus uses a membrane siderophore receptor, called PvuA, which is coupled to an inner membrane ATP-binding cassette (ABC). This ABC transporter system comprised of proteins encoded by the pvuABCDE gene cluster (Fig. S3) is required for transporting the siderophore across the inner membrane (Tanabe et al., 2003). Siderophores are not only iron carriers but also important regulators of virulence (Miethke and Marahiel, 2007) and mediators of bacterial interaction with phytoplankton (Amin et al., 2009b; Kramer et al., 2020). We showed here a pivotal role of iron in the interaction between V. altlanticus LGP32 and A. pacificum ACT03. This mirrors the mutualistic interaction observed between Gymnodinium catenatum and Marinobacter (Amin et al., 2009b). In fact, in natural settings, the co-occurence of Marinobacter and G. catenatum is suggested to depend on a mutually beneficial utilization of iron and carbon resources (Bolch et al., 2011). As in the present study, iron seems to play a key role in the interaction. Indeed, the labile iron released through the photolysis of ferric chelates with vibrioferrin providing a crucial iron source for phytoplankton, which need substantial amounts of iron to support carbon fixation through photosynthesis (Amin et al., 2009a; Yang et al., 2021). This fixed carbon, in turn, sustains the growth of both the phytoplankton and their bacterial counterparts (Amin et al., 2009b; Kramer et al., 2020). Future study could demonstrate further the role of vibrioferrin in group attack, by adding iron-saturated vibrioferrin to algae-Vibrio co-cultures.

In the natural environment, associations between bacteria and algae have already been observed (Lopez-Joven et al., 2018; Miller et al., 2005; Rosales et al., 2022; Xu et al., 2022). Although we do not have definitive proof of a predator/prey interaction in situ due to confounding environmental variables, environmental data acquired in the Thau Lagoon showed a correlation between Alexandrium and Vibrio occurrence, suggesting that such interaction could occur in nature. Evolution of such a predatory behavior would provide an important ecological advantage to Vibrio to obtain nutrients in environment, where Alexandrium spp. and Gymnodinium catenatum form blooms.

With more than 30 species distributed all over the world (Anderson et al., 2012; Hallegraeff et al., 2021), Alexandrium spp. and Gymnodinium spp. considered as invasive species by the Delivering Alien Invasive Species Inventories for Europe (http://www.europe-aliens.org)), could play an unexpected and important role in maintaining, structuring and regulating Vibrio populations in the ecosystem. In turn, Vibrio could contribute to the regulation and control of their blooms.

To conclude, this study reveals the capacity of some Vibrio spp. to be facultative predators that hunt specific algae. In the current context of climate change favourable to their development monitoring the invasive algae, Alexandrium spp. and Gymnodinium catenatum should be considered not only for their potent harmful effect on humans and animals, but also as a potential source of nutrients for the expansion of Vibrio, particularly pathogenic species (Lemire et al., 2015).

Data availability

All data on environmental study are included in the article and/or supporting information files. All other study data are available on request from the corresponding author.

Acknowledgements

The authors thank F. Le Roux and Y. Labruche for their gift of Vibrio mutants; E. Garcés, S. Turki, H. Frehi, A. Bouquet and W. Medhioub for their contribution in obtaining Mediterranean phytoplankton strains; K. Escoubeyrou and D. Stien for advice on HRMS/MS data; V. Diakou-Verdin and E. Jublanc from the Montpellier RIO Imaging Platform for access to microscope facilities (www.mri.cnrs.fr); P. Clair from the Montpellier qPHD platform and the Technoviv platform of the UPVD for access to PCR facilities; N. Brunet, A. Lang, A. Payelleville, T. Milhau and M. Leroy for their technical assistance; The IFREMER LER-LR laboratory for access to the Thau Lagoon; the BIO2MAR platform at the Observatoire Océanologique de Banyuls for providing access to UHPLC-HRMS/MS facilities.

Additional information

Funding

European VIVALDI grant 678589, French EC2CO ROSEOCOM grant and IFREMER ALGOVIR grant.

Additional files

Video 1. Dynamics of Vibrio atlanticus LGP32-Alexandrium pacificum ACT03 interaction. GFP-tagged V. atlanticus (small green cells); living A. pacificum (large red cells); degraded A. pacificum (large green cells) filmed under an epifluorescence microscope.

Video 2. Second-by-second timing of Vibrio Atlanticus LGP32 attacking Alexandrium pacificum ACT03. GFP-tagged V. atlanticus (small green cells); A. pacificum living cell (large red cells) filmed under an epifluorescence microscope.

Video 3. Degradation and disruption of Alexandrium pacificum ACT03 flagella. Effect of Vibrio supernatant on the first stage of the interaction filmed under a confocal microscope.

Video 4. Group attacks of Vibrio atlanticus LGP32 on target Alexandrium pacificum ACT03. Vibrio, filmed under a confocal microscope, attacks in groups one immobilized Alexandrium cell then moves on to attack — still as a group — another cell without touching the other whole cells, suggesting active communication between Vibrio cells. V. atlanticus LGP32 (small cells); A. pacificum ACT03 (large cells).

Video 5. Vesicle formation and bursting of an Alexandrium pacificum ACT03. Direct effect of Vibrio supernatant on Alexandrium after 126 h of culture filmed under a confocal microscope

Data S1.