Effect of marennine on the rearing medium and microbiota of Mytilus edulis larvae and its protective effect after exposure to the pathogenic Vibrio splendidus

ABSTRACT Opportunistic pathogens have been associated with yield-limiting factors in bivalve hatcheries. Numerous natural compounds are being investigated for their beneficial effects and potential to enhance larval resistance without requiring antibiotics. One of those is the use of marennine, a blue pigment, originating from the diatom Haslea ostrearia, which has demonstrated a positive effect on larvae survival. The aim of this study was to highlight the protective effect of marennine on Mytilus edulis larvae during bacterial challenges in relation to a potential modification of the marennine-treated larvae microbiota. D-larvae and post-larvae were exposed for 96 h to Vibrio splendidus (106 cell mL−1) with and without mareninne (500 µg L−1). The presence of marennine increased the survival rate of D-larvae exposed to the pathogen. The molecular analysis of the larvae microbiota diversity indicated a modification in the D-larval microbiota’s richness related to survival rates of larvae. Ultimately, our study sheds light on the importance of the larval microbiota in pathogen resistance during the bivalve rearing process.


Introduction
During the last few decades, fish and shellfish farming industries have experienced an important increase in their production rates in response to the growing seafood demand worldwide.From 1960 to 2013, the world's marine products consumption per capita doubled, increasing from an average of 9.9 kg to 20.5 kg per year (FAO 2022).While the proportion of sustainably exploited fishery stocks is demonstrating a decreasing trend, reaching 64.6%, the share of aquaculture products in global seafood consumption shows an increasing trend reaching 49.2% in 2020 (FAO 2022).To meet the demand, producers need to have access to a large and stable supply of juveniles.Thus, hatcheries' popularity increases due to the high spatiotemporal variability of the natural recruitment success (Helm, Bourne, and Lovatell 2004).
The success of hatcheries, and ultimately the bivalve production rates depend on the larvae survival in the rearing system.Bacterial infections are known to be a major bottleneck for hatchery-reared juveniles by causing massive mortality (mass mortality events).Many opportunistic bacterial pathogens from the genera Vibrio, Pseudomonas, Aeromonas, and Roseovarius were reported to be linked to mass mortality events in bivalve hatcheries (Eggermont et al. 2017;Paillard, Le Roux, and Borrego 2004;Tubiash, Colwell, and Sakazaki 1970).These organisms exert their pathogenicity only in specific environmental or larvae physiological conditions that are still poorly understood.The seemingly random nature of mass mortality events is a major obstacle to achieve steady production rates.Understanding and therefore controlling disease outbreaks in hatcheries is crucial to assure the stability of the hatchery production.
Prior to the 1980s, antibiotics were regularly used by producers in bivalve hatcheries to reduce the impact of bacterial infection on their production (Asmani et al. 2016).However, the development of antibiotic resistance (Defoirdt, Sorgeloos, and Bossier 2011;Falaise et al. 2016), the risk of transmission of these resistant strains in the food web (Sapkota et al. 2008) and even the horizontal transfer of antibiotic resistance genes from aquatic bacteria to human pathogens (Heuer et al. 2009) led to strict regulations in several countries.The use of antibiotics is therefore not a sustainable solution for controlling the occurrence of mass mortality events in bivalve hatcheries.
Probiotics and natural bioactive molecules have been proposed as alternatives to antibiotics in animal farming (Balcazar et al. 2006;Beaz-Hidalgo et al. 2010;Defoirdt, Sorgeloos, and Bossier 2011;Falaise et al. 2016;Gastineau et al. 2012).A wide range of organisms, or their cellular components, were tested as probiotics for their positive effect in aquaculture (Irianto and Austin 2002).For example, Chilean scallop larvae (Argopecten purpuratus) were able to complete the larval pelagic phase without any antibiotic treatment when exposed to inhibitor-producing bacterial strains (Vibrio sp.C33, Pseudomonas sp.11 and Bacillus sp.B2) (Riquelme et al. 2001).The strains Pseudoalteromonas sp.X153 (Longeon et al. 2004) and Phaeobacter gallaeciensis X34 (Genard et al. 2014) also demonstrated a positive effect on cultured scallop larvae (Pecten maximus) survival.Azadirachtin, an extract from the neem tree (Azadirachta indica), also demonstrated a beneficial effect acting as an immunostimulant on goldfish (Carassius auratus) and increased their survival rate when challenged against Aeromonas hydrophila (Kumar et al. 2013).Recently, marennine, a blue-green pigment produced by the diatom Haslea ostrearia, was suggested as an interesting bioactive molecule for bivalve hatcheries (Turcotte et al. 2016).This natural pigment has shown promising results by reducing the mortality rate of blue mussels larvae (Mytilus edulis) challenged with V. splendidus.The mode of action underlying the beneficial effects of marennine in bivalves farming seems related to a membrane stiffening mechanism in Gram negative bacteria (Tardy-Laporte et al. 2013), like Vibrios splendidus (Bouhlel et al. 2021).Improvements in the use of feed, immunostimulation, antibacterial activity, alteration of the microbial metabolism, and competitive exclusion are some of the proposed, but poorly documented, hypothetical modes of action of probiotics in aquaculture (Irianto and Austin 2002;Prado, Romalde, and Barja 2010).Unraveling the role of marennine as a water-additive in bivalve hatcheries is crucial in order to safely expand its utilization.
In this study, we tested the hypothesis that the interaction of marennine with the larval microbiota could be a potential mode of action contributing to the observed beneficial effects on bivalve larvae survival.The artificial addition of marennine in the rearing medium might modify the bacterial community within the rearing medium itself and induce a change in the bacterial community recruited by the larvae to form their microbiota.In addition, marennine could interfere with the quorum sensing system within the larvae resulting in a modification of the microbiota formation (Kalia 2013).
Nowadays, it is well known that the taxonomic diversity, the abundance, and the physiological structure of the microbiota affect its host health condition (Laterza et al. 2016;Lopez et al. 2014;Marchesi et al. 2016).In the case of bivalves, it has been suggested that a shift in the structure and the specific diversity of the microbiota in adults might prevent bacterial pathogen to settle within the host (Froelich and Oliver 2013).These microbiota modifications might lead to a better pathogen resistance in larvae and contribute to the beneficial effect of marennine addition observed in bivalve hatcheries.Therefore, interaction between probiotics or bioactive molecules with the natural larvae microbiota, the conditions prevailing in the rearing medium and the diversity of bacterial communities might contribute to the observed protective effect of marennine on bivalve larvae in hatcheries.
The aim of this study was to investigate the importance of the larval microbiota in hatchery-reared bivalve larvae when exposed to the opportunistic pathogenic bacteria V. splendidus in the presence or absence of marennine.Blue mussel larvae, the most important shellfish aquaculture production in Canada from 1995Canada from to 2016Canada from (2015)), were exposed to the opportunistic bacterial pathogen V. splendidus during bacterial challenge experiments with and without marennine added to the rearing medium.The bacterial community in the rearing medium and the blue mussel larval microbiota were characterized under different bacterial contamination and treatment conditions in order to better understand the importance of the larval microbiota in the prevention of mass mortality events with marennine as a natural water-additive.The bacterial community abundance in the rearing medium was characterized by flow cytometry, and the diversity of communities in the rearing medium and larval microbiota were investigated using the denaturing gradient gel electrophoresis (DGGE) technique.Taken together, these analyses have helped to shed light on the host-pathogen-microbiota interplay during the rearing process of marine bivalve's larvae in hatcheries.

Marennine solution
Marennine was obtained from H. ostrearia culture produced as described in Gastineau et al. (2014) and Turcotte et al. (2016) and the pigment was purified by the method of Pouvreau et al. (2006) to obtain a solution in nanopure water (pH 7.2) that was autoclaved.The solution was filtered on 0.2 µm pore-sized cellulose acetate membrane before determining the solution concentration by spectrophotometry at 656 nm with the specific extinction coefficient of 12 L g −1 cm −1 according to Turcotte et al. (2016).

Bacterial culture condition
Vibrio splendidus 7SHRW, a strain isolated from the Gulf of St. Lawrence (Qc, Canada) (Mateo et al. 2009), was grown overnight in 10 mL of salty LB medium (25 g L −1 NaCl, pH 7.2, autoclaved) at room temperature prior to each experiment (bacterial growth kinetic and bacterial challenges).This strain is recognized for its ability to infect blue mussel larvae (Turcotte et al. 2016).After incubation, cells were centrifuged at 3,000 g for 5 min and the cell pellet was washed twice in sterile physiological water (9 g L −1 NaCl, pH 7.2, autoclaved).Then, bacterial cells were suspended in sterile physiological water to obtain a stock solution at 10 9 cell mL −1 .

V. Splendidus growth kinetic
The effect of marennine on V. splendidus growth was assessed by spectrophotometry.The three final concentrations of marennine used were 100 µg L −1 , 500 µg L −1 , 1000 µg L −1 .Negative controls without bacterial cells were used for each of the concentrations tested.A marennine-free positive control with bacteria was also included.Each condition was replicated five times.The experiment was performed in a 96-wells microplate in a final volume of 200 µL.Each well contained 100 µL of salty LB 2X medium (50 g L −1 NaCl, pH 7.2, autoclaved), 50 µL of Vibrio cells at 10 6 cell mL −1 (or physiological water (9 g L −1 NaCl, pH 7.2, autoclaved) and 50 µL of marennine (or nanopure water (pH 7.2) as a control) at desired concentrations.Cell growth was estimated by measuring the suspension's OD at 595 nm every 30 min during 48 h.The OD reads were performed automatically with a SpectraMax 190 (Molecular devices, CA, USA) plate reader running the SoftMax Pro software.Growth curves were obtained by plotting OD values against time.Specific growth rate was estimated from the slope of the ln-transformed curves as recommended by Beaulieu et al. (2015).Growth inhibition was calculated by comparing the OD values of the curves from each tested marennine concentration after 24 h to the OD value of the positive control.

Blue mussel larvae rearing conditions
Spawning adults from a pure population of M. edulis (Tremblay et al. 2011) were obtained from St. Peter's Bay (Prince Edward Island, Canada; 46.4281°N, 62.6422°W) in June 2017.The latter were used to produce gametes and ultimately larvae, following the usual protocol used by R. Tremblay's laboratory (Turcotte et al. 2016).Spawning was induced by thermal shocks from 5°C to 20°C and D-larvae (9 days-old) and metamorphosed post-larvae (29 daysold) were both used for bacterial challenges.All the rearing process was carried out at the UQAR/ISMER Station aquicole de Pointe-au-Père (Rimouski, Qc, Canada).

Experimental design
Bacterial challenges were carried out at the UQAR/ISMER marine microbial ecology laboratory (Rimouski, Qc, Canada) to assess the larvae's microbiota response to the exposure to the opportunistic bacterial pathogen V. splendidus and marennine.The experiments were performed in 3 Fernbach flasks (Thermo Fisher Scientific, USA) for each treatment applied at room temperature.A concentration of 10 larvae mL −1 was used during the bacterial challenges in a volume of 2.5 L.
Both D-larvae and metamorphosed post-larvae were exposed to four different treatments (three replicates of each): a control without V. splendidus or marennine (C), a pathogen treatment with only V. splendidus at a final concentration of 10 6 cell mL −1 (V), a marennine treatment with only marennine at a final concentration of 500 µg L −1 (M) and a treatment in which larvae were exposed to both V. splendidus and marennine at the same concentrations used in V and M treatments (MV).All treatments were performed in UV-treated and filtered (0.2 µm pore-sized filter) seawater from the St. Lawrence Estuary (Qc, Canada) having a salinity of 23.09‰ for the first experiment (D-larvae) and 23.11‰ for the second experiment (post-larvae).For the purpose of this article, treatments C and M will be referred to as unchallenged larvae and treatments V and MV will be referred to as challenged larvae.
Before the challenging experiments, the initial bacterial communities in the seawater inhabiting the rearing medium in the culture tanks (T0) were characterized by sampling 1 L of the seawater used to fill the tanks in triplicates that were filtered on a 0.2 µm pore-sized Durapore filter (Ø 47 mm) and frozen at −80°C before further analyses.The larval microbiota prior to the challenge experiment was characterized to assess the initial communities (T0).Approximately 10 000 larvae (10 larvae mL −1 ) were collected on 50 µm mesh, gently washed with sterile seawater (filtered on 0.2 µm and autoclaved) and immediately frozen at −80°C until DNA extraction.During the challenge experiments, two samples of 1 L of each flask were collected after 1 h and 96 h of exposition after gently homogeneous mixing and filtered on 50 µm mesh to collect the larvae.The collected larvae were treated as already described and frozen at −80°C.A sample of 4 mL of the 50 µm mesh filtered water was fixed in the dark for 15 min with 0.2% glutaraldehyde at pH 7 and then frozen at −80°C until further bacterial abundance analyses.Finally, the remaining water was filtered on a 0.2 µm pore-sized Durapore filter (Ø 47 mm) to collect the bacteria present in the rearing medium and frozen at −80°C for DNA extraction.

Larval performance estimation
After 96 h, three samples of 10 mL of rearing medium containing at least 30 individuals were taken to assess the larvae survival rate.The larval survival rate (%) was obtained by calculating the ratio (number of live larvae/total number of larvae) between the control (C) and the treatments (M, V, and MV).The control survival rate was set as 100%.In addition, the shell length of 30 live larvae was measured to monitor the potential effect of marennine on the larval growth during the experiments.The larvae were examined and photographed under 100× with a microscope Olympus B×41 coupled to an Evolution VF camera and use of Image Pro Plus software v5.1 (Media Cybernetics, Silver Spring, MD, USA).The mean of the three counts from each tank was used for statistical analyses.

Bacteria analyses
The total abundance of bacteria in the rearing medium was assessed by flow cytometry using a CytoFlex Flow Cytometer (Beckman Coulter Inc., Mississauga, Canada).Frozen samples were thawed at room temperature and then stained in the dark with 0.3 µL of SYBR Green I (10 000X, Invitrogen, Thermo Fisher Scientific, USA).Fluorescent beads (Fluoresbrite YG microsphere 1 µm, Polysciences) were added to each sample prior to their analyses as an internal standard (Lebaron et al. 2002).The sample volume analyzed was determined by weighing the tubes before and after the analyses.Then, the abundance was determined with the latter volume and the number of counted events.Data analyses were performed with the FCSalyzer software (version 0.9.14, Free Software Foundation Inc., Boston, USA).Total bacteria were detected by plotting the green fluorescence recorded at 530 nm (FL1) versus the side angle light scatter (SSC).
Total DNA was extracted from larvae and the filters containing the bacteria present in the rearing medium using the E.Z.N.A. Mollusc DNA kit (Omega Bio-Tek, Norcross, USA) according to the manufacturer's instructions.Prior to each DNA extraction, larvae were gently washed with physiological water and crushed in a sterile 1.5 mL Eppendorf tube containing 350 µL of ML1 buffer from the extraction kit.The filters were cut into pieces and then transferred into a sterile 1.5 mL Eppendorf tube also containing 350 µL of ML1 buffer from the extraction kit.The extracted DNA from the larvae contained the DNA from the larvae itself, the bacteria within the larvae (microbiota), and the bacteria attached to the shell of the larvae.
The bacterial 16S rDNA gene was amplified by PCR using the universal primers 341F-GC (5'-CGC-CCG-CCG-CGC-CCC-GCG-CCC-GTC-CCG-CCG-CCC-CCG-CCC-GCC-TAC-GGG-AGG-GGA-GAG-3') and 907 R (5'-CCG-TCA-ATT-CMT-TTG-AGT-TT-3') from Schäfer and Muyzer (2001).The mix was composed of 5 µL of 10× PCR buffer (QIAGEN, Hilden, DE), 200 µM of DNTPs (VWR, Radnor, USA), 50 pmol of each primer, 1 U of HotStart Taq polymerase (QUIAGEN, Hilden, DE), 200 ng of DNA, and sterile water (q.s.water 50 µL).Amplifications were performed in triplicates and then pooled to minimize the effect of PCR biases (Perreault et al. 2007).Briefly, a 500 bp fragment from the V4 region coding for the 16S sub-unit of the bacterial ribosome was amplified with the following PCR conditions: an initial denaturation at 94°C for 1 min, followed by 20 cycles consisting of a denaturation step at 94°C for 1 min, an annealing step at 65°C for 1 min (touchdown of −0.5°C per cycle) and an extension step at 72°C for 3 min.Following these steps, there was 15 cycles consisting of a denaturation step at 94°C for 1 min, an annealing step at 55°C for 1 min, and an extension step at 72°C for 3 min.Finally, a last extension step at 72°C for 3 min.Amplicons were then purified using the MiniElute columns (QIAGEN, Hilden, DE).Purified amplicons were then kept at −20°C until bacterial diversity analyses.
The bacterial diversity was assessed using the PCR-DGGE technique described in Schäfer and Muyzer (2001).Analyses were performed with a DGGE-4001-Rev-B (C.B.S. Scientific Company, CA, USA) following Schäfer and Muyzer (2001) recommendations.A denaturing gradient from 30% to 70% was used to allow a good discrimination of operational taxonomic units (OTUs).The migration was performed at 100 V for 16 h and temperature of 60°C.After migration, gels were stained with SYBR Gold at a final concentration of 3× (10,000×, Invitrogen, Thermo Fisher Scientific, USA) during 1 h in the dark.A photograph was taken of each gel using UV light (AlphaImager HP, Alpha Innotech, CA, USA).

Statistical analyses
For each model, residuals were screened for normality using the normal probability plot and then tested using the Shapiro-Wilk's test (Table S1 in supplementary material).Homogeneity of variance was graphically assessed using residuals, and larval survival rates were compared using a one-way analysis of variance (ANOVA) with the treatment (C, M, V, and MV) as a factor.The shell lengths were compared using a mixed model setting the 30 replications of the measure as the random effect to account for the pseudoreplications, replicates were the three tanks used per treatment and the factor the four treatments.The bacterial abundances in the rearing medium were compared by performing a two-way analysis of variance (ANOVA) with treatment and sampling time as factors.
DGGE profiles were analyzed using the Phoretix 1D Software (Non-linear Dynamics, Newcastle upon Tyne, UK) to obtain a presence-absence matrix from the detection of OTUs.Because the triplicates were similar, further analyses were performed with a single sample for each treatment for each larval stage.The latter matrix was then used to calculate distances between sample's profiles from the Jaccard dissimilarity index.Hierarchical analyses were performed using the unweighted pair group method with arithmetic mean (UPGMA) algorithm to form clusters based on the previously calculated dissimilarity index.The fingerprints of the communities from the rearing medium and the D-larvae microbiota after 96 h of exposition were compared to investigate the resemblance between both communities.The initial communities (T0) were also compared with each treatment.The proportions of matches (%) in terms of the number of shared unweighted operational taxonomic unit (OTUs) between samples were retained.
The variation of the number of conserved, gained, and lost OTUs for the D-larvae in the treatments M, V, and MV in regard to the control (C) and the initial community (T0) after 96 h of exposition were used to identify variations of OTUs among treatments in regard to the match mismatch analysis presented previously.The total number of OTUs for each treatment, the number of unique OTUs in each treatment and the number of common OTUs between treatments were assessed.

V. Splendidus growth kinetic
When exposed to the four different marennine concentrations, the lag time of V. splendidus ranged from 6.8 h to 7.2 h and the specific growth rate ranged from 0.17 h −1 to 0.19 h −1 (Table 1).No differences were found for the specific growth rate of V. splendidus among tested concentrations (chi-square = 0.55, p = .91,df = 3) in regard to the lag time (chi-square = 6.09, p = .11,df = 3).The exposition to different concentrations of marennine did not result in inhibition of V. splendidus growth.All calculated growth inhibitions (%) were under 10%, and therefore marennine concentrations under 1000 µg L −1 were considered to be non-effective on the growth of this bacteria (Table 1).

Larval growth and larval survival rate
The mean shell length measured on the D-larvae after 96 h was 183.5 µm.There was no difference of the shell's length among treatments for the D-larvae (F 3,8 = 0.78, p = .54).In contrast, significant survival rate variation of D-larvae among treatments were observed after 96 h (near 25%; F 3,8 = 16.44,p < .001; Figure 1), with a significant lower value for Vibrio treatment (V, p < .001).No difference was detected among the survival rates of the postlarvae from the different treatments with maximal variation observed lesser than 20% (F 3,8 = 1.93, p = .20).

Abundance
The cell abundance in the rearing medium was significantly different among sampling time and treatment and the interaction of these two factors (Table 2).In the flasks of unchallenged larvae (C and M), the abundance of bacteria after 1 h of exposition was under 0.5 × 10 5 cell mL −1 and increased threefold to fivefold after 95 h for both treatments (Figure 2).In the flasks containing Vibrio (V and MV), bacteria abundance after 1 h of exposition was over 7 × 10 5 cell mL −1 , translating the artificial addition of pathogen cells to attain a final concentration of 10 6 cell mL −1 , and decreased to less than 2.5 × 10 5 cell mL −1 after 96 h (Figure 2).The presence of marennine did not affect the bacteria abundance measured in the flasks after 96 h of exposition as the values are more similar than the V treatment (Figure 2).

Microbiota
After 96 h of exposition, the D-larvae microbiota genetic fingerprints were separated into two clusters.The first cluster composed of the fingerprint from the initial larval microbiota (T0), the control (C), and the MV treatment, and the second one composed of the fingerprint from the V and M treatments (Figure 3).For the post-larvae microbiota, the fingerprints of the challenged larvae (V and MV) are clustered with the unchallenged marennine-treated post-larvae microbiota (M) (Figure 3).The total number of OTUs decreased from 26 in T0 to 20 in the MV treatment.The total number of OTUs increased from 20 (T0) to 35 in Table 2. Two-way ANOVA's results for the cell abundance analyses for both larval stages (D-larvae and post-larvae) with treatments (C, M, V, and MV) and the sampling time (1 h and 96 h) as factors and the interaction between both factors.V and 41 in M in the microbiota fingerprints from the second cluster (Figure 3).In the rearing medium, the fingerprint of the MV treatment formed a cluster with the control (C), both with a decreased total number of OTUs representing, respectively, 20 and 26 compared to T0 which had 43 OTUs in total (Figure 3).Notably, both communities from the rearing medium and the microbiota from the MV treatment have a total of 20 OTUs (Figure 3) from which only 11 were common to the rearing medium and the microbiota.For the D-larvae and the post-larvae, the treatment M is less dissimilar from the initial communities (T0) in the rearing medium with 28.57% and 66.68%, respectively (Figure 3).

Match-mismatch between the rearing medium and the larval microbiota
The proportion of matches between the community in the rearing medium and the community composing the microbiota of the D-larvae was 34.2% for T0 and 34.4% for the treatment C. This proportion was more important for the treatment M and V being, respectively, 67.5% and 41.9%.The proportion of matches in the MV treatment was the lowest with 26.9% (Figure 4).

Comparison of microbial communities
The simultaneous comparison of the initial community (T0), the control (C), and each treatment (M, V, and MV) allowed determining which OTUs were lost, conserved, and gained among the different treatments.The treatment M, in regard to the larval microbiota and the rearing medium, was the one that lost the less OTUs that were either shared between T0 and C or unique to the treatment C (Figure 5).The number of gained or conserved OTUs was dissimilar between the D-larvae microbiota and the rearing medium.For example, in the treatment M, 22 OTUs were conserved from the initial community in the rearing medium and only four in the larval microbiota.In terms of lost OTUs, the treatments V and MV were similar, except for the OTUs lost that were unique to the initial microbiota (T0).The V treatment conserved 7 OTUs from T0 and the MV treatment conserved only 1 OTU from T0 (Figure 5).This difference is less important in the rearing medium for the same treatments (Figure 5).In the rearing medium, treatment M gained 10 unique OTUs while both V and MV gained 5 unique OTUs each (Figure 5).The difference in the number of unique OTUs gained is even greater than in the microbiota for the treatment M.There were 19 unique OTUs found in the larval microbiota from the M treatment, 13 unique OTUs in the treatment V and 4 in the treatment MV (Figure 5).

Discussion
The future of the development of shellfish aquaculture is often related to stable production juveniles potentially having controlled genetic characteristics.However, the larval rearing process can create an environment conducive to the development of opportunistic bacterial pathogens like V. splendidus.Controlling these pathogens in the rearing environment is crucial to ensure a stable juvenile production.As new alternative methods to the harmful usage of antibiotics in aquaculture, marennine, a biomolecule extracted from a diatom is becoming an interesting way of limiting the impact of pathogens (Falaise et al. 2016;Gastineau et al. 2014;Turcotte et al. 2016).In our study, we confirm the beneficial effects of marennine on blue mussels larvae challenged with V. splendidus and investigated the marennine effect in regard to the response of the bacterial communities in the system.

Antibacterial activity of marennine on V. splendidus 7SHRW
Our results do not show a direct antibacterial effect of marennine at a concentration of 500 µg L −1 and therefore an antimicrobial effect on the pathogen V. splendidus itself is unlikely to contribute to the observed beneficial effects documented by Turcotte et al. (2016).It has been previously reported that marennine demonstrates an inhibitor effect on the growth of several Vibrio species in previous experiments.Falaise et al. (2016) assessed the sensitivity of various species from the Vibrio genus (V.tasmaniensis, V. aestuarianus, V. coralliitycus, and V. tubiashii) to marennine at different concentrations ranging from 1 mg L −1 to 100 mg L −1 .Results showed concentration-dependant inhibitor effects on the growth kinetics of all the tested species for marennine concentrations higher than the one used in our experiments.Turcotte et al. (2016) and Falaise et al. (2016) observed that marennine had a toxic effect on M. edulis D-larvae at a concentration as low as 1000 µg L 1 .The marennine concentration used for this study was therefore reduced to 500 µg L −1 to avoid the toxic effect of the pigment on the larvae during the rearing procedure and still exert the protective effect.The mode of action of marennine at low concentration exerting a protective effect seems related to a membrane stiffening mechanism, without affecting the bilayer integrity (Bouhlel et al. 2021).

Observed beneficial effects of marennine during bacterial challenges
The exposition of the larvae to marennine during bacterial challenges did not result in a measurable inhibition or an enhancement of the larval growth unlike the results obtained by Turcotte et al. (2016).The effect of marennine on larval growth might only be observable after a longer exposure to the bioactive molecule Turcotte et al. (2016).
The effect of marennine on the D-larvae survival after 96 h of exposition was similar to the results published by Turcotte et al. (2016) with no significant difference between the survival rate of C and the corresponding treatments M and MV.In the same way, the survival rate of the marennineuntreated challenged D-larvae with Vibrio (V) was lower than the other treatments.Thus, our results confirm that marennine had a beneficial effect on the larval survival rate of the D-larvae when challenged against the opportunistic pathogen V. splendidus.The survival rates from the bacterial challenge experiment conducted with post-larvae were not different between the treatments including C. The lethal effect from the exposition to a pathogen known to cause mass mortality in the rearing systems was not observed for that ontogenic stage, suggesting a better resistance to pathogens after metamorphosis, as suggested in the studies of (Bassim et al.;Bassim et al. 2014).
The beneficial effect of marennine was only observed when exposing mussel D-larvae to V. splendidus and marennine at 500 µg L 1 for 96 h.The larval survival rate increased without a significant difference in the final shell length indicating that the beneficial effect of marennine is unlikely to be due to a change in the physiological state of the larvae that impacts the shell's growth.However, Turcotte et al. (2016) observed an accumulation of 40% more triacylglycerol in blue mussel larvae exposed to marennine for 20 days.This trend suggests that marennine has a slight effect on the larvae physiological state, but the authors were not able to link this physiological change to an effect of marennine on neither the feeding behavior of the larvae nor the accumulation of lipids reserves.A change in the larval microbiota might be another explanation for the observed beneficial effects of marennine on the larval survival rate against a pathogen than a direct influence of the biomolecule on the host's physiology.
Even though marennine did not show an antimicrobial effect on a pure culture of V. splendidus, the effects of the same bioactive molecule might differ when exposing it to a complex community.It could modify the taxonomic diversity or the cell abundance of the indigenous bacterial communities already in place by directly interacting with the bacteria or by modifying the conditions in the rearing medium thus modifying the bacterial assemblage.In the rearing medium of the unchallenged larvae (C and M), the increase of the bacterial abundance between 1 h and 96 h of exposition is most likely due to the normal development of the bacterial community in UV-treated and filtered seawater used to perform the experiments.The rearing medium was probably colonized by the microorganisms from the rejection of larvae's feces and microorganisms attached to the shell of the larvae.
In the rearing medium of the challenged larvae (V and MV), the high abundance of bacterial cells after 1 h of exposition is most likely due to the artificial addition of V. splendidus cells.Notably, the abundance of bacterial cells decreased after 96 h to reach the same abundance detected in the rearing medium of the other treatments (C and M), suggesting a potential ingestion of the suspended cells by the larvae.Dubert et al. (2016) demonstrated that shellfish larva tissues were colonized within 2 h during a challenge experiment against the bacterial pathogen V. splendidus.The observed beneficial effects of marennine on challenged larvae survival rate might come from a modification of the diversity of bacteria composing the larval microbiota and the larvae themselves, hence the main objective of this study.
The analysis of the bacterial diversity in the rearing medium and the larval microbiota from the larvae exposed to marennine and the pathogen revealed high dissimilarity between C and the treatments (M, V, and MV) and between each treatment.These results suggest that each treatment from the challenge experiment induced a microbial modification (a different balanced state compared to microbial communities from healthy individuals) in both the rearing medium and the larval microbiota.The modification that occurred in the larval microbiota from the challenged larvae (V) most likely induced microbial dysbiosis, which is known to occur in invertebrates' microbiota when exposed to pathogenic Vibrio sp (Rungrassamee et al. 2016;Xia et al. 2018).During an infection, opportunistic bacterial pathogens like V. splendidus outcompete other taxa and cause a microbial imbalance modifying the interactions between the host, the microbial communities (in the rearing medium and the microbiota), and the environment.
Interestingly, the exposition of mussel larvae to marennine seems to be related to a modification of the bacterial community in the rearing medium and in the larval microbiota as well when compared to the control.The latter results suggest that marennine has an impact on the environmental conditions of the rearing medium, thus modifying the community contained within.A change in the bacterial community in the rearing medium or in the larval physiological state due to the use of marennine might result in an observable change in the selection of the bacterial taxa recruited to form the larval microbiota.Surprisingly, the treatment combining an exposition to V. splendidus and to marennine seemed to exert an important dissimilarity of the bacterial diversity of both the rearing medium and the larval microbiota compared to the other treatments.In other words, the simultaneous exposition to the bacterial pathogen and marennine (MV) resulted in a different bacterial community compared to those sampled from the treatments M and V.A synergetic effect between V. splendidus and marennine might contribute to the observed beneficial effect of marennine on the larval survival rate of the challenged marennine-treated D-larvae.
Assessing the link between the microbial diversity in the rearing medium and the larval microbiota allows us to determine how the bacterial community in the rearing medium influences the composition of the larval microbiota or vice versa.The proportion of match-mismatch of OTUs detected in the samples from the rearing medium and the D-larvae microbiota assess how the link between those two communities' changes under different conditions.Surprisingly, the treatments M and V seemed to increase the resemblance between the rearing medium and the larval microbiota and the treatment MV seemed to decrease that proportion slightly under the one found in C and the initial time point (T0).Somehow, the combination of the exposition of D-larvae to V. splendidus and marennine (MV) resulted in a weaker link between the rearing medium bacterial community and the larval microbiota.Because mussel larvae are filter feeders, a modification of the bacterial community in the rearing medium should influence the composition of the larval microbiota assuming that larvae harvest and maintain environmental bacteria to form their microbiota.In contrast, our data suggest that mussel larvae maintain a uniquely composed microbial community as their microbiota.This trend in our data should be taken carefully since there was no replication.However, it is known that mussels maintain a different bacterial community from the rearing environment as their microbiota.Prieur (1982) presented data on the differences between the cultivable bacterial community of the normal microflora of M. edulis larvae and the rearing medium showing a clear dissimilarity between the two communities.More recently, many studies using cultivation-free molecular techniques showed that the link between the surrounding environment (natural waters or rearing mediums) and shellfish microbiota is not as direct as previously thought (Chauhan et al. 2018).
Comparing the sampled bacterial communities of each treatment (M, V, and MV) after 96 h of exposition with the initial community (T0) and the community found in the control (C) after 96 h allows characterizing the microbial imbalance in terms of OTUs gained, conserved and lost.In the rearing medium, marennine seemed to cause an important gain of unique OTUs and the conservation of a more important proportion of OTUs compared to the other treatments, which are similar.This similarity between the bacterial communities from the treatment V and MV suggests that the synergetic effect previously mentioned do not take place in the rearing medium but rather in the larval microbiota.The community from the challenged marenninetreated D-larvae microbiota (MV) gained less unique OTUs and conserved less OTUs from the initial microbiota (T0) compared to the two other treatments (M and V).Surprisingly, a different pattern in the rearing medium was observed compared to the larval microbiota.These data support the idea of a potential synergetic effect of marennine with the opportunistic bacterial pathogen V. splendidus on the microbial diversity of the D-larvae resulting in an observable beneficial effect on the larval survival rate.The influence of the composition of one's microbiota on the host health status has been extensively studied in many types of organisms (Zilber-Rosenberg and Rosenberg 2008).The presence of marennine in the rearing medium of the D-larvae exposed to the pathogen might have modified the larval microbiota's via direct interactions of marennine with the microbial cells or by modifying the microbial recruitment by the larvae, resulting in a modification of the larvae sensibility toward V. splendidus infections.

Conclusion
Our study demonstrated that marennine, a bioactive molecule extracted from the diatom H. ostrearia culture, did have a beneficial effect on the larval survival rate of blue mussel D-larvae when challenged with the opportunistic bacterial pathogen V. splendidus.Moreover, our data clearly demonstrate that the exposition of D-larvae to marennine and pathogen was accompanied by a modification of the microbiota in the rearing medium and in the larval.Notably, only the survival rate of the challenged D-larvae was lower indicating that post-larvae are able to survive the exposition to V. splendidus.Marennine seems to be useful preferentially with the larval stages before the metamorphosis.These results are of interest for the development of marennine-based prophylactic treatments and optimize its application to prevent mass mortality events caused by the opportunistic bacterial pathogen V. splendidus in blue mussel hatcheries.
Our data strongly suggest that a coupled effect of marennine and V. splendidus contributes to the increase of the larval survival rate of the D-larvae possibly involving the larval microbiota.A better understanding of the mode of action of the bioactive molecule marennine is crucial to expand and regulate its usage in shellfish aquaculture.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
The work was supported by the National Science and Engineering Research Council of Canada [299100]; Fonds de Recherche Nature et Technologies ([2014-RS-171171].

Figure 2 .
Figure 2. Bacterial abundance in the rearing medium after 1 h and 96 h of exposition to each treatment (C, M, V and MV); standard deviation is shown with error bars; letters indicate groups formed by the Tukey's HSD post-hoc analysis and one set of letters (a, b and c) were assigned to the analysis of D-larvae while another set (d, e and f) were assigned to the post-larvae analysis.

Figure 3 .
Figure 3. Dendrograms of the genetic fingerprints of the bacterial communities sampled in the microbiota and the rearing medium of blue mussel D-larvae and post-larvae at the beginning of the experiment (T0) and exposed to four different treatments (C, M, V and MV) during 96 h.The cluster analysis was based on the Jaccard coefficient similarity indicator and the dendrograms were constructed with the UPGMA algorithm using the vegan package (version 2.5-1) built for R (version 3.5).Numbers are the total number of OTUs recorded in each treatment.

Figure 4 .
Figure 4. Proportion of matches against mismatches in the comparison of the communities of the D-larvae rearing medium and the larval microbiota for each treatment (C, M, V and MV) and the initial communities (T0).

Figure 5 .
Figure 5. Simultaneous comparison of the initial community (T0) and the control (C) after 96 h of exposition with all other treatments (M, V and MV) in regard of a) the unique OTUs Gained in each treatment, b) the Conserved OTUs between T0 and the control (C), c) the Unique OTUs from the initial community (T0) and d) the unique OTUs from the control (C).

Table 1 .
Effect of marennine on V. splendidus 7SHRW lag time, specific growth rate, and growth inhibition (standard error is shown between parentheses).