Exploring the hologenome concept in marine bivalvia: haemolymph microbiota as a pertinent source of probiotics for aquaculture


  • Florie Desriac,

    1. Université de Brest, EA3882, Laboratoire Universitaire de Biodiversité et Ecologie Microbienne (LUBEM), IFR148 ScInBioS, Institut Universitaire de Technologie, Quimper, France
    Search for more papers by this author
  • Patrick Le Chevalier,

    1. Université de Brest, EA3882, Laboratoire Universitaire de Biodiversité et Ecologie Microbienne (LUBEM), IFR148 ScInBioS, Institut Universitaire de Technologie, Quimper, France
    Search for more papers by this author
  • Benjamin Brillet,

    1. Université de Brest, EA3882, Laboratoire Universitaire de Biodiversité et Ecologie Microbienne (LUBEM), IFR148 ScInBioS, Institut Universitaire de Technologie, Quimper, France
    Search for more papers by this author
  • Ivan Leguerinel,

    1. Université de Brest, EA3882, Laboratoire Universitaire de Biodiversité et Ecologie Microbienne (LUBEM), IFR148 ScInBioS, Institut Universitaire de Technologie, Quimper, France
    Search for more papers by this author
  • Benoît Thuillier,

    1. Institut départemental d'analyses de conseil et d'expertise en hygiène alimentaire eau et environnement et santé animale, Quimper Cedex, France
    Search for more papers by this author
  • Christine Paillard,

    1. Université de Brest, CNRS, IRD, Ifremer, UMR 6539 Laboratoire des Sciences de l'Environnement Marin (LEMAR), Institut Universitaire Européen de la Mer (IUEM), Technôple Brest Iroise, Plouzané, France
    Search for more papers by this author
  • Yannick Fleury

    Corresponding author
    1. Université de Brest, EA3882, Laboratoire Universitaire de Biodiversité et Ecologie Microbienne (LUBEM), IFR148 ScInBioS, Institut Universitaire de Technologie, Quimper, France
    • Correspondence: Yannick Fleury, Laboratoire Universitaire de Biodiversité et d'Ecologie Microbienne EA3882, Institut Universitaire de Technologie, Université de Brest, 6 rue de l'université, 29334 Quimper Cedex, France. Tel.:+33 (0) 298 641 935; fax: +33 (0) 298 641 969;

      e-mail: yannick.fleury@univ-brest.fr

    Search for more papers by this author


Haemolymph-associated microbiota of marine bivalves was explored for antibacterial activity against important aquaculture pathogens. A collection of 843 strains were cultured from the haemolymph of four bivalve species (Crassostrea gigas, Mytilus edulis, Pecten maximus and Tapes rhomboides) collected by deep-sea diving in the Glenan Archipelago (France). Cell-free culture supernatants were investigated for antibacterial activity using the well-diffusion assay. About 3% of haemolymph-associated cultivable bacteria displayed antibacterial activity toward Gram-negative pathogens. Among the active bacteria, Pseudoalteromonas strains exhibited the highest antibacterial activity. The cell-free culture supernatant of one of them, named hCg-51, was able to inhibit the growth of bacterial pathogens even after drastic dilution (1 : 1024). Hemocyte survival was not significantly altered in the presence of the haemolymph-associated strains assayed. Moreover, a dose-dependent beneficial effect on hemocyte survival rates was observed with the hCg-51 strain. These results suggest that haemolymph microbiota may participate in bivalve protection and therefore confer a health benefit on the host. As a result, the results highlight bivalve haemolymph microbiota as a promising novel source for aquaculture probiotics. This work also gives a first insight into the contribution of the haemolymph-associated microbiota as part of the bivalve ‘hologenome’.


The ‘hologenome’ concept was introduced by Zilber-Rosenberg & Rosenberg (2008). It considers the host and its associated microorganisms (microbiota) a super-organism (holobiont) and therefore the true evolutionary unit of selection. This concept is based on (1) existing symbiotic relationships between all animals or plants and several microorganisms; (2) the transmission of the symbiotes; (3) the benefits of the symbiotic association between the host and the microorganisms; and (4) the genetic plasticity enhancement of the holobiont, through change in the microbiotic composition under environmental pressure (Zilber-Rosenberg & Rosenberg, 2008). The ‘hologenome’ theory strengthens the probiotic concept. Microbiota may form a microbial shield that could limit the settlement of pathogens by competition for resources and/or antimicrobial compound production and/or stimulation of the host immune system (Oelschlaeger, 2010). Microbiota antimicrobial compounds seem to play a key role in control of the microbial community and health recovery (Dobson et al., 2012). As environmental pressures such as climate changes can disturb the microbial shield, allowing epizootic events in marine invertebrates, antimicrobial compounds from autochthonous probiotics could be a powerful tool to restore the microbial shield and protect the host from pathogens (Desriac et al., 2010). Marine invertebrates and especially bivalves may be considered pertinent animal models since they are filter-feeders and so have to face large numbers of microorganisms. Furthermore, the well-accepted presence of microorganisms in the haemolymph of healthy bivalves tends to indicate that this ecosystem could contribute to host haemostasis. The haemolymph-associated microbiota has been mainly investigated for pathogens, whereas its composition and role remain obscure. We have recently isolated antimicrobial compound-producing strains from oyster haemolymph, suggesting that microbiota may confer a health benefit on the host (Defer et al., 2013). In this study, we have explored the cultivable haemolymph-associated bacteria in four bivalves (oyster, clam, mussel and scallop) for their antimicrobial activity. The most potent ones were also investigated for hemocyte cytotoxicity. Results are clearly in line with the hologenome concept. Moreover, they suggest that haemolymph-associated bacteria are a potential source of aquaculture probiotics.

Materials and methods

Bivalve sampling

To limit the impact of anthropic pressure, bivalve specimens were collected by deep-sea diving in the Glenan Archipelago (47°43′N, 4°01′W, WGS84 system), a Natura 2000 area (FR5300023), during winter 2009 and spring 2010. Selected species were the oyster (Crassostrea gigas), the blue mussel (Mytilus edulis), the scallop (Pecten maximus), and the pink clam (Tapes rhomboides).

Haemolymph of each individual was collected aseptically by inserting a 25-gauge needle attached to a 1-mL syringe directly into the adductor muscle. For C. gigas, haemolymph was collected from the pericardium. A volume ranging from 0.5 to 1 mL of haemolymph was drawn from each mollusc and placed in ice to prevent the hemocyte aggregation. Each sample was microscopically examined to check the presence of healthy hemocytes.

Bacteria isolation

Checked haemolymph (50 μL) was spread onto Marine agar Petri dishes using a Wasp® automated spiral plater (AES Lab). Plates were further incubated for 72 h at 18 °C. To isolate as many different bacteria as possible, 1–10 macroscopically distinguishable colonies were picked and subcultured in Marine Broth for 48 h at 18 °C with shaking (100 r.p.m.). Bacterial purity was assessed by streaking on Marine Agar. For long-term storage, sterile glycerol was added to 1 mL bacterial culture (25% v/v) in cryogenic vials that were stored at −80 °C.

Antimicrobial assay

Cell-free supernatants coming from culturable haemolymph-associated bacteria were assayed for antibacterial activity against a panel of 12 aquaculture pathogens (Table 1). After growth (72 h, 18 °C, 100 r.p.m.), the culture supernatant (1 mL) was collected by centrifugation (6000 g for 10 min at 4 °C) and filtration (0.22 μm, SFCA serum Filter Unit, Nalgene). To detect antibacterial activity, the well-diffusion method was used (Wiegand et al., 2008; Defer et al., 2013). Specific agar medium according to bacterial target was inoculated with an 8-h-old culture broth of the indicator strain to a bacterial concentration of 1.106 CFU mL−1. Wells (diameter 4 mm) were punched into the agar medium and cell-free supernatants (20 μL) or controls (Marine Broth for negative control and polymyxin B sulphate and Nisaplin® at 1 mg mL−1 as positive control against respectively Gram-negative and Gram-positive target bacteria) were created. After an overnight incubation at optimal growth temperature of the indicator strain, antibacterial activity was revealed by an inhibition halo around wells. When antibacterial activity was detected, a second antibacterial assay in liquid medium was performed to define minimal inhibitory concentrations in standard 96-well microtiter plates (Wiegand et al., 2008; Defer et al., 2013). Briefly, target bacteria in exponential growth state (1 × 106 CFU mL−1) were incubated with serial twofold dilutions (in sterilized Marine Broth) of active cell-free supernatant and incubated for 48 h at optimal growth temperature. Sterile as well as growth and inhibition controls (Polymyxin B at 100 μg mL−1) were carried out. The activity was expressed as a function of protein concentrations (μg mL−1) determined using BC Assay Kit (Interchim) according to the manufacturer's instructions and as a function of the highest dilution factor of cell-free supernatant that inhibited 100% of the target strain growth. The target bacteria panel was broadened. Five other strains of Vibrio were included: Vibrio pectenecidae A365, V. coralliilyticus CIP107925, V. tubiashii CIP102760, V. parahaemolyticus and V. harveyi ORM4.

Table 1. Aquaculture pathogens chosen as target bacteria to screen the collection of haemolymph-associated bacteria for antibacterial activity
Target bacteriaStrainGrowth conditionsPathogen ofDisease or symptomReferences
  1. NRRL, Agricultural Research Service Culture Collection; ATCC, American Type Culture Collection; CIP, Collection of Institut Pasteur; TSB, Tryptone Soy Broth from Fisher Scientific Bioblock; TSB*, TSB + NaCl (1.5%); MB, Marine Broth (Difco).

Gram-positive bacteria
Carnobacterium maltaromaticum NRRL B-14829TSB30 °CSalmonidCarnobacteriosisLeisner et al. (2007)
Lactococcus garviae ATCC43921TSB30 °CSalmonidLactoccocisVendrell et al. (2006)
Vagococcus salmoninarum 18–96TSB30 °CTroutVagococcosisRuiz-Zarzuela et al. (2005)
Gram-negative bacteria
Aeromonas caviae CIP 7616TSB30 °CCarp-catfish- troutHaemorrhagic disease, furunculosisKozińska (2007)
Aeromonas hydrophila CIP 7614TSB30 °CCarp-catfish-troutHaemorrhagic disease, furunculosisSeshadri et al. (2006)
Aeromonas salmonicida ATCC14174TSB30 °CCarp-catfish-troutHaemorrhagic disease, furunculosisKozińska (2007)
Listonella anguillarum ATCC 43305TSB*25 °CFish-bivalve-crustaceanVibriosisFrans et al. (2011)
Salmonella enterica CIP 8297TSB30 °CHuman food bornSalmonellosisMartinez-Urtaza et al. (2004)
Vibrio alginolyticus CIP 103360MB30 °CCrustaceanVibriosisWang (2011)
Vibrio splendidus  MB18 °CMolluscVibriosisKesarcodi-Watson et al. (2009)
Vibrio tapetis CECT 4600MB18 °CClamsBrown Ring DiseasePaillard et al. (2004)
Yersinia ruckeri ATCC 29473TSB30 °CSalmonidsEnteric RedMouthTobback et al. (2007)

DNA extraction and 16S PCR

The bacterial isolates expressing antibacterial activity were selected for a phylogenetic analysis based on 16S rRNA gene sequences. DNA was extracted as previously described (Godon et al., 1997) and 16S rRNA gene was amplified using two universal primers, W18 : 9F and W20 : 1462R, yielding 1000–1500 pb PCR products (Godon et al., 1997). The PCR mixture was carried out according to the manufacturer's instructions (PCR Master Mix Promega®). The following PCR conditions were used: initial denaturation at 94 °C for 4 min, followed by 35 cycles at 94 °C for 1 min, 52 °C for 1 min and 72 °C for 1 min and a final elongation step at 72 °C for 10 min. The PCR products were analyzed on agarose (1.2%) gel electrophoresis and sequenced by GATC Biotech (Germany). Sequences were compared with the GenBank nr/nt database by blastn to identify their closest match. To construct trees, an alignment with the first five hit blast 16S sequences of each strain was made, using clustalw2 (Larkin et al., 2007). Phylogenetic trees were built using mega 5 program package (Tamura et al., 2011).

Cytotoxicity assay

The cytotoxicity activity was estimated for three active strains isolated from oyster haemolymph. The two antimicrobial compound-producing strains, named hCg-6 and hCg-42, isolated from oyster haemolymph in a previous study (Defer et al., 2013), were also investigated for hemocyte cytotoxic effect. The experimental procedure was as described previously (Delaporte et al., 2003). Briefly, the haemolymph of about 30 C. gigas was withdrawn, pooled and filtered through an 80-μm mesh. A 19-h-long contact was established at 18 °C between hemocytes and bacteria in cytometry tubes. Several concentrations of bacteria were evaluated (ratio bacteria/bivalve hemocytes 25/1, 50/1, 100/1). A control was done using incubated hemocytes in sterile seawater. Bacteria were washed three times with sterile seawater and suspended in an appropriate volume of sterile seawater before contact. Following incubation, propidium iodide (1% v/v) was added and hemocytes incubated in the dark for an additional 30 min. Samples were then analyzed with a FACS-Calibur™ flow cytometer (Becton Dickinson). The measures were obtained after 30 s with a low flow rate. The three replicate data collected were then statistically analyzed by a one-way anova, with P-error level set at 0.05.

Antibiotic sensitivity

The sensitivity to antibiotics was determined by a disc-diffusion method according to the AFNOR NF U47-106 instructions, with Marine Agar plate as medium due to marine bacteria cultivability. Antibiotics tested were amoxicillin (25 μg), colistin (50 μg), enroflaxin (5 μg), florfenicol (30 μg), flumequin (30 μg), tetracycline (30 UI) and trimethroprim/sulphamethoxazole (1.25/23.75 μg). Results were observed after an 18–20-h incubation at 18 °C.

Results and discussion

Culturable haemolymph-associated bacterial collection

The haemolymph from oysters, clams, mussels and scallops were spread onto non-selective Marine Agar. A great disparity in culturable haemolymph-associated bacteria was observed intra host species (data not shown). Haemolymph bacterial concentrations below the lower limit of detection (i.e. 102 CFU mL−1) were more frequently observed in mobile bivalve (75% of P. maximus and 51% of Tapes rhomboides collected) than in haemolymph from fixed bivalves (9% of C. gigas and M. edulis collected). Excluding these extreme bacterial concentrations, the highest average bacterial concentration was detected in M. edulis haemolymph and the lowest one in P. maximus (Table 2).The culturable haemolymph-associated bacterial concentrations were shown to be individual- and species-dependent (Table 2). This may be the result of various environmental concentrations (Olafsen et al., 1993) as well as bivalve physiological characteristics. Moreover, growth conditions (MB medium and incubation temperature) may clearly impact the bacterial growth rate and/or select some marine species (Gram et al., 2010). A total of 843 haemolymph-associated strains were isolated from the bivalve haemolymph sampling (Table 2). They were named according to their origin and the number of the isolate. For instance, the hCg-1 strain was the first strain isolated from C. gigas haemolymph.

Table 2. Collection of haemolymph-associated bacteria and antibacterial spectrum of their cell-free supernatantsThumbnail image of
  • (−), No inhibitory activity was detected. (+), (++) and (+++), Target bacteria inhibition was detected. Radius (r) was respectively r < 2 mm for (+), 2 mm < r < 4 mm for (++) and r > 4 mm for (+++). After cryogenic storage, haemolymph-associated strains in grey were not cultivable anymore. In black, strains were cultivable but lost (normal text) or retained (bold) antibacterial activity. Nisaplin® and polymyxin were used at 1 mg mL−1 as positive control respectively for Gram-positive and Gram-negative targets. hCg-, hPm-and hMe- refer to bacteria isolated respectively from the haemolymph of Crassostrea gigas, Pecten maximus and Mytilus edulis.

  • Carnobact. maltarom., Carnobacterium maltaromaticum; Lacto.garviae, Lactococcus garviae; Vag. salmon., Vagococcus salmoninarum; Aeromonas hydro., hydrophyla; salmon., salimonicida; List. anguill., Listonella anguillarum; Salmo. enterica, Salmonella enterica; Vibrio algino., alginolyticus; splend., splendidus.

  • No. means number of animal collected.

  • Screening for antibacterial activity

    The 843 isolates were screened for antibacterial activity against 12 target bacteria by the well-diffusion assay. Among these, 26 isolates (about 3%) showed a clear inhibition zone around wells for at least one target strain (Table 2). The antibacterial activity was exclusively directed against Gram-negative bacteria, mostly of the Vibrio genus. Such selectivity of activity differs from the antibacterial spectra usually described during marine antibiotic screenings. Indeed, Gram-positive target bacteria generally appear to be more sensitive (Hughes & Fenical, 2010; Wilson et al., 2010). Our data suggest that haemolymph-associated bacteria may constitute a specific microbiota tolerated and/or selected by host for their ability to inhibit pathogens.

    On the other hand, despite a high number of isolates, no strain isolated from clam haemolymph demonstrated antibacterial activity. The target bacteria spectrum and/or the growth conditions [nutrients (Spanggaard et al., 2001) and/or temperature (Zhang et al., 2012) or bacterial presence in the surroundings (Mearns-Spragg et al., 1998; Dusane et al., 2011)] may explain these results. Nonetheless, the potential of bivalve microbiota as a source of antimicrobial compounds is evident, although underexplored.

    The cryogenic storage resulted in total loss of cultivability for five strains (hMe-15, -22, -82, -119 and -131) and the cell-free supernatant of a further nine strains did not exhibit antibacterial activity (hCg-60, -78, -111 and-114, hPm-100 and -102, hMe-34, -43 and -273). Such loss of cultivability or bioactivity after storage is frequently described with marine bacteria (Gram et al., 2010) and discussed (Hazen et al., 2010; Vynne et al., 2011).

    Antibacterial potency evaluation

    The antibacterial activity of the 12 bioactive strains remaining was quantified using a 96-well micro-titration method (Wiegand et al., 2008). Insofar as the chemical nature of the active compounds was unknown, MICs were expressed as a function of maximal dilution factor of the supernatant that exerted a total inhibition of pathogen growth. MICs were also expressed in protein concentration (Table 3). All the hCg strains and hMe-187 and -253 supernatants were able to inhibit 100% of bacterial growth of at least one pathogen when diluted at least 64-fold. Moreover, eight haemolymph-associated isolates inhibited at least five different species among the seven Vibrio species included in the panel and one or more other bacteria, suggesting a real potential for antibacterial treatment in aquaculture farming, since Vibrio species are pathogenic for fish (Toranzo et al., 2005), molluscs (Verschuere et al., 2000) and crustaceans (Wang, 2011).

    Table 3. Antibacterial potency of bivalve haemolymph-associated strains expressed as maximum dilution factor of cell-free supernatant that result in 100% bacterial growth inhibition
     hCg-hPm-hMe-Polymyxin μg mL−1
    Target bacteria23 4851108109269597187223238253 
    1. Serial twofold dilutions were realized in Marine Broth. Data are the results of triplicates. hCg-, hPm- and hMe- represent bacteria isolated respectively from the haemolymph of Crassostrea gigas, Pecten maximus and Mytilus edulis. Minimal supernatant protein concentration (μg mL−1) that inhibited 100% of target growth is indicated in brackets.

    Aeromonas caviae 128 (5.1)128 (5.1)128 (5.6)128 (5.6)16 (44.7)16 (45.9)16 (37)4 (125)32 (18.6)2 (354)32 (31.3)4
    Aeromonas hydrophila 128 (5.1)128 (5.1)256 (2.8)16 (44.8)16 (44.7)16 (45.9)16 (37)4 (125)64 (9.3)2 (354)32 (31.3)1
    Aeromonas salmonicida 2 (356)16 (37)4 (125)4 (148.8)4 (177)2 (500.5)ND
    Listonella anguillarum 512 (2.6)256 (2.3)512 (1.4)32 (22.4)4 (148)4 (125)4 (177)2 (237.5)32 (31.3)ND
    Vibrio coralliitycus 64 (10.3)32 (18.3)128 (5.6)4 (178.8)4 (148)2 (250)8 (74.4)16 (62.6)ND
    Vibrio harveyi 16 (41.1)4 (146)64 (11.1)8 (125.1)32
    Vibrio parahaemolyticus 16 (44.5)––2 (297.5)2 (500.5)ND
    Vibrio pectenecidae 512 (2.6)128 (4.6)512 (1.4)32 (22.4)4 (148)4 (125)16 (37.2)64 (15.6)ND
    Vibrio splendidus 512 (2.6)256 (2.3)1024 (0.7)16 (44.8)64 (11.2)16 (45.9)16 (37)8 (75.5)32 (18.6)4 (177)64 (15.6)2
    Vibrio tapetis 512 (2.6)256 (2.3)1024 (0.7)4 (179.5)64 (11.2)32 (23.0)32 (18.5)8 (75.5)64 (9.3)4 (177)4 (118.8)64 (15.6)
    Vibrio tubiashii 64 (10.3)32 (18.3)128 (0.7)8 (89.4)4 (148)2 (250)4 (148.8)2 (354)16 (62.6)ND

    Active haemolymph-associated bacteria identification

    The active haemolymph-associated strains, hCg-23, -48, -51, -108, -109, hPm-26, hMe-95, -223, -253 and -273, were identified by 16S rRNA gene amplification as members of the Gammaproteobacteria class (Fig. 1) belonging to either the Alteromonadales (89%) or the Vibrionales orders (11%). Such affiliation was also observed in antimicrobial screening of marine bacteria and in previously described probiotics used in bivalve hatcheries (Zheng et al., 2005; Gram et al., 2010; Prado et al., 2010; Wilson et al., 2010; Flemer et al., 2012). Vibrio genus has been described to be a natural flora in bivalve and crustacean haemolymph (Faury et al., 2004; Gay et al., 2004; Gomez-Gil et al., 2010). The antibacterial as well as probiotic ability of this genus has been described (Riquelme et al., 1997, 2001; Mansson et al., 2011; Silva-Aciares et al., 2011).

    Figure 1.

    Neighbour-joining tree indicating the phylogenetic relationships inferred from partial 16S rRNA gene sequences of strains hMe, hPm and hCg within the two orders of the Gammaproteobacteria phyla Alteromonadales and Vibrionales. Filled squares indicate strains isolated from haemolymph (this study and Defer et al., 2013) and empty circles indicate previously described probiotic strains. Bootstrap values (expressed as percentage of 1000 replications) > 50% are shown at branching point. Filled circles indicate that the corresponding nodes were also recovered in trees generated with the maximum parsimony and the maximum-likelihood algorithms. The Enterobacteriales member Escherichia coli 2012K11 (position 119–1269) was used as outgroup. Scale bar: 0.005 substitutions per nucleotide.

    Nine strains, hCg-23, -48, -51, -108, -109, hMe-95, -223, -253 and -273, were affiliated with the Pseudoalteromonas genus. One set of bacteria, hCg-23, -48, -51, -108 and -109, formed a new cluster in the Pseudoalteromonas genus (Fig. 1). This genus is well-known to produce a wide variety of biologically active secondary metabolites (Kalinovskaya, 2004; Bowman, 2007). Within this genus, the strains Pseudoalteromonas haloplanktis INH, Pseudoalteromonas sp. X153 and Pseudoalteromonas sp. D41 were shown to protect or enhance the survival rate of Agropecten purpuratus, P. maximus and C. gigas, respectively (Riquelme et al., 1997; Longeon et al., 2004; Kesarcodi-Watson et al., 2012). Some of the haemolymphatic strains are within lineages that are phylogenetically distinct from known probiotic strains and may have unique probiotic properties (e.g. secondary metabolites).

    As no antibacterial activities have ever been described in species closely related to the isolated strains, we postulate that the antibacterial compounds produced by these strains have not been described to date. Nonetheless, the hPm-26 bacterial strain isolated from P. maximus haemolymph was affiliated within the genus Thalassomonas. To our knowledge, this is the first report describing antibacterial activity from the recent genus Thalassomonas.

    Hemocyte protection of the haemolymph-associated bacteria

    Due to their potent antibacterial activities, three strains of Pseudoalteromonas were investigated for their impact on oyster hemocyte survival in vitro. Our goal was to control the safety of hCg strains toward hemocytes. Indeed, although the bivalves collected were healthy, there was no guarantee that a high concentration of the bacteria would not result in hemocyte death. Hemocytes were incubated with up to 5 × 108 CFU mL−1 for 19 h. Recently isolated Pseudoalteromonas strains hCg-6 and hCg-42 from oyster haemolymph were also analyzed. These strains produced antimicrobial peptide in haemolymph in an in vitro assay (Defer et al., 2013).

    Hemocyte/bacteria mixes did not exhibit any morphological changes, whatever the ratio used and the strain assayed, when examined using flow cytometry (Supporting Information, Fig S1). After 3 h, hemocyte mortality in sterile seawater was quantified at 5.2% (± 0.7) (data not shown). A study on Crassostrea virginica hemocyte viability showed that around 3–5% of hemocytes died when incubated in sterile seawater (Allam & Ford, 2006). The hemocyte death observed herein (15% after 19 h incubation in seawater) is in agreement with the short lifespan of bivalve hemocytes described previously (Binelli et al., 2009). Moreover, after a 19-h-long incubation of hemocyte in the presence of hCg strains, flow cytometry analyses revealed that (1) no additional hemocyte mortality was detected with strains hCg-6 and hCg-42, suggesting that these strains have no opportunistic behaviour, whatever the hemocyte/bacteria ratio used; and (2) a significant reduction of hemocyte mortality with strains hCg- 23, -51 and -108 (Table 4).

    Table 4. Haemocyte mortality observed after a 19-h-long incubation time with different bacterial concentrations of strains hCg-6, hCg-42, hCg-23, hCg-48 and hCg-51
    Ratio bacteria/hemocytehCg-6hCg-42hCg-23hCg-51hCg-108
    1. Values not sharing the same superscript letter (a–g) are statistically different (anova,< 0.05). The strains hCg-6 and -42 were isolated from oyster haemolymph in a previous study (Defer et al., 2013). Means ± SD, n = 3.

    2. ND, not determined.

    0 : 1 (control)15.48 ± 1.27a15.48 ± 1.27a15.48 ± 1.27a15.48 ± 1.27a15.48 ± 1.27a
    25 : 1 12.39 ± 3.1a14.56 ± 2.89a9.88 ± 0.21b4.06 ± 0.38c6.71 ± 0.56f
    50 : 113.46 ± 1.27a17.12 ± 4.75a10.09 ± 0.47b3.19 ± 0.37d5.91 ± 0.99f, g
    100 : 111.96 ± 1.8a11.64 ± 1.94aND2.15 ± 0.29e5.09 ± 1.04g

    Interestingly, hemocyte mortality was significantly decreased in the presence of strain hCg-51 in a concentration-dependent manner. To our knowledge, no similar results showing a dose-dependent reduction of hemocyte mortality by bacterial strain or metabolites have been published to date. Future experiments will investigate phagocytic activity, adherence, and nitric oxide synthesis of hemocytes. These preliminary in vitro experiments support the beneficial role of bivalve microbiota in stimulating and/or protecting hemocyte cells. These results suggest that the haemolymphatic microbiota may play a role in host immunity and homeostasis. As a result, haemolymph microbiota may represent a potential source for aquaculture probiotics.

    Antibiotic sensitivity

    Major molluscan pathogens such as Vibrio were shown to harbour a high number of mobile genetic elements (Hazen et al., 2010), showing their abilities to integrate elements that can increase their capacity to colonize an ecological niche. As antibiotics used in prophylaxis were banned to limit the development of bacterial resistance, antibiotic substitutes such as probiotics should not harbour antibiotic-resistant genes (Saarela et al., 2000; Nair et al., 2012). We therefore investigated the hCg-strains to ensure their antibiotic sensitivity to the common antibiotic used in aquaculture. No resistance to antibiotics was observed for the five tested strains except for the tetracycline antibiotic (Table 5). The medium used (Marine agar) appears to be unsuitable for tetracycline diffusion due to antibiotic co-precipitation with the salts observed. Nevertheless, the recommended medium for antibiotic sensitivity assay (i.e. Mueller–Hinton, AFNOR NF U47-106) was unsuited to hCg strains, as no bacteria grew on it.

    Table 5. Antibiotic sensitivity of the active haemolymph-associated strains from Crassostrea gigas
    1. S and R means respectively sensitive and resistant. Inhibition diameters in millimeter measured around each disc are in brackets. For antibiotic concentrations see 'Materials and methods'. hCg designates the strains isolated from Crassostrea gigas haemolymph.

    AmoxicillinS (25)S (31)S (35)
    TetracyclineR (13)R (6)R (14)
    Trimethroprim/sulphamethoxazoleS (28)S (29)S (34)
    ColistineS (25)S (25)S (26)
    FlumequineS (38)S (36)S (34)
    EnrofloxacinS (37)S (35)S (31)
    FlorfenicolS (39)S (31)S (26)

    To conclude, we have shown that some culturable haemolymph-associated bacteria can exhibit (1) potent antibacterial activity against some bacterial pathogens in aquaculture; (2) no significant cytotoxic effect on hemocytes but rather a reduction in hemocyte mortality; and (3) sensitivity to the main antibiotic used in aquaculture. Insofar as such strains may confer a health benefit to the host, they may be considered potential probiotics. A combined strategy using antibacterial screening, hemocyte viability and antibiotic sensitivity may allow us to focus on a reduced number of haemolymphatic strains for in vivo experiments.

    As a result, the haemolymphatic microbiota, to which little attention has been given, represents a potential source for future aquaculture probiotics and may be used to renew the antimicrobial arsenal. The bioactive molecules, as well as the dynamics of haemolymph colonization and the ability of strains to protect bivalves from infection are being investigated.


    Thanks are given to Dr J. L. Nicolas (Ifremer) for the generous gift of V. pectenecidae A365, coralliilyticus CIP107925, tubiashii CIP102760, parahaemolyticus and harveyi ORM4, to Estelle Bellanger-Thuillier for her technical assistance, and to Hervé Bourdon for manuscript corrections. F.D. was supported by a ‘Quimper-communauté’ grant for PhD thesis. This work was partly funded by the region Bretagne (Biprobio project, ARED 6227).