P. oceanica plants were collected in June 2014 in front of the port of Banyuls-sur-mer, south of France (42°28′54.12″N, 3° 8′11.99″E; 2 m deep). To maximize the recovery of the bacterial strains associated with P. oceanica (epiphytes and endophytes strains), we combined two protocols of isolation. The first protocol consisted of cutting a clean piece of leaves or rhizomes, after observations of epibionts presence under the stereomicroscope, to isolate epiphytic strains. A second protocol was used to isolate endophytic strains (Zinniel et al., 2002). This protocol included a series of surface sterilizations with first 99% ethanol, second 3.125% NaOCl, and third a second 99% ethanol step before a final and gentle washing with sterile water.

The different pieces of Posidonia (rhizomes and leaves) were ground with a blender. Serial dilutions (1 to 10⁻³) in presterilized seawater were performed, and 100 μL was plated on various media. All these media were reconstituted with 75% seawater: BG11 medium, R2A medium and Starch Casein Agar (SCA) medium. Petri plates were incubated at 25°C until growth of the colonies occurred and new colonies stopped appearing. Colonies were isolated and purified based on their different morphotypes on their respective media. Each bacterial strain was then stored at −80°C in a 35% glycerol solution and in a 5% DMSO solution in Marine Broth (MB, Difco™ BD). The strains were deposited in the Banyuls Bacterial Culture Collection (WDCM911) http://collection.obs-banyuls.fr/.

The bacterial identification was based on their partial 16SrRNA gene sequencing. All the sequences were deposited in Genbank (see Supplementary Table 1). The obtained 16S rRNA gene sequences were compared using the BLAST algorithm against the sequences of the EzTaxon server (http://eztaxon-e.ezbiocloud.net) database (Kim et al., 2012). Phylogenetic analyses were conducted using MEGA 6 software (Kim et al., 2012). All alignments were performed with ClustalW (Larkin et al., 2007) and adjusted manually. Phylogenetic trees were constructed using the Neighbor-Joining (NJ) method with p-distance correction and 1,000 bootstrap replicates.

Bacterial culture supernatants were tested in the AHL biosensor assay following previously described protocols using Pseudomonas putida and Escherichia coli-based biosensors. Briefly, P. putida F117 (pRK-C12; Km^r; ppuI:mpt) was used for the detection of long-chain AHLs (>8 carbons in the acyl side chain) (Andersen et al., 2001; Girard et al., 2017), and E. coli MT102 (pJBA132) for the detection of short-chain AHLs (≤8 carbons in the acyl side chain) (Riedel et al., 2001; Girard et al., 2017). E. coli MT102 and P. putida F117 were cultivated in Luria-Bertani (LB) Broth (Sigma L3022) overnight with continuous shaking (200 rpm) at 37°C supplemented with tetracycline (25 μL/mL) and at 30°C supplemented with gentamicin (20 μL/mL), respectively. An overnight culture of each biosensor strain was inoculated in fresh LB medium (dilution 1/50). This fresh biosensor culture was dispensed into 96-well microplates (150 μL/well for bacterial supernatant evaluation and 180 μL/well for extracts or fractions). To evaluate bacteria for their ability to produce Al-1, they were grown on MB medium 24 h at 25°C with agitation (120 rpm). Culture supernatants were collected after centrifugation (10 min at 10,000 rpm) and added to the microplate wells containing the biosensor strains (50 μL/well, in triplicate). Extracts or fractions were dissolved in DMSO/LB (25:75, v/v) and added to each well in triplicate (20 μL/well). Microplates were incubated at 30°C and 37°C, depending of optimum growth temperature of the selected biosensor strain, without shaking. After 0, 5, and 24 h of incubation, fluorescence was determined with a Victor1420 Multilabel Counter (Perkin-Elmer) at an excitation wavelength of 485 nm and a detection wavelength of 535 nm. The optical density (OD) at 620 nm was also measured to monitor biosensor cell growth. Negative controls were biosensor cultures with sterile LB medium and without extract. Biosensor cultures with added commercial AHLs (C6-HSL for E. coli MT102 and oxo-C10-HSL for P. putida F117) were used as positive controls.

A mutant strain of Vibrio harveyi, BB170, was used to detect the presence of AI-2 in the culture supernatants (Bassler et al., 1997; Taga, 2005). Briefly, the V. harveyi reporter strain was grown at 30°C in Autoinducer Bioassay (AB) medium. The overnight culture was then diluted 1:5,000 into fresh medium, and 180 μL of the diluted cells was added to wells containing 20 μL of the culture supernatants of the bacteria to be tested for AI-2 production. The 96-well plates were incubated at 30°C. Every hour, luminescence was measured using a Victor1420 Multilabel Counter (Perkin-Elmer) until the medium control sample produced the least amount of light. The Al-2 production is reported as the fold induction of the reporter strain divided by the background when only medium was added to the reporter. All assays were performed in duplicate.

The AHL biosensor E. coli MT102 was also used to highlight the quorum quenching ability of the bacterial strains. The biosensor was cultivated in Luria-Bertani (LB) Broth (Sigma L3022) overnight with continuous shaking (200 rpm) in the same conditions (temperature, antibiotics) used for the quorum-sensing bioassay. An overnight culture of the biosensor strain was inoculated into fresh LB medium (dilution 1/25). This fresh biosensor culture was dispensed into 96-well microplates (150 μL/well). Bacteria were grown on MB medium supplemented with an AHL (100 μL of C6-AHL or oxo-C10-HSL (both 1 mM) in 100 mL of MB for the assay with E. coli MT102 and P. putida F117, respectively) for 24 h at 25°C with agitation (120 rpm). Culture supernatants were collected after centrifuging for 10 min at 10,000 rpm and added to the microplate containing the biosensor strains (50 μL/well in triplicate). Extracts or fractions were dissolved in DMSO/LB (25:75, v/v) and added to each well in triplicate (20 μL/well). The microplates were analyzed as previously describe above (detection of AI 1 bioassay). The negative controls were biosensor cultures with sterile MB medium supplemented with AHL. Two strains of the Banyuls Bacterial Culture Collection (BBCC116 and BBCC2153) were used as positive controls.

The ability of strains to produce biofilm has been evaluated using a crystal violet-based protocol previously published (Stepanovic et al., 2000; O'Toole, 2011). Strains from fresh agar plates were inoculated in 5 mL of Marine Broth and incubated for 24 h at 25°C. Biofilm formation was conducted in 96-well flat bottom plates. To increase biofilm formation (Moreira et al., 2013), culture strains were diluted with fresh medium until a final OD630 nm of 0.2 was reached. The microplate wells were filled with 100 μL of diluted culture and 100 μL of a 4.5% glucose solution, whereas only broth with glucose served as a control to check the sterility of the medium and to test for non-specific attachment. Ruegeria mobilis (strain MOLABBCC1144, Banyuls Culture Collection) was used as a negative control, and Sulfitobacter brevis (strain BBCC1436, Banyuls Culture Collection) was used as a positive control. Each inoculated plate was incubated for 48 h at 25°C under static conditions. After incubation, the medium was gently removed, and the wells were washed with 200 μl of sterile phosphate-buffered saline (PBS, pH 7.2) to remove all planktonic cells while preserving the integrity of the biofilm. The remaining bacteria that were attached to the bottom of wells were stained for 15 min at room temperature with 100 μL of 0.2% crystal violet solution prepared in 20% ethanol (v/v). Subsequently, the unbound crystal violet stain was removed, and the wells were rinsed three times with sterile water before adding 200 μL of a solution of 10% acetic acid, 50% ethanol, and 40% distilled water (v/v/v) to each well. The absorbance was measured at 540 nm using a microplate reader [Victor1420 Multilabel Counter (Perkin-Elmer)]. We classified isolates according to their ability to form biofilms, and following Stepanovic et al. (2000), a cut-off value was calculated for the negative control using the following formula: ODc = Average OD of negative control + 3 × SD (Standard Deviation) of negative control. Based on the calculated cut-off OD, strains were classified into the following categories: non-biofilm producers: (OD < ODc), weak biofilm producers: (0Dc < OD < 2 × 0Dc), moderate biofilm producers: (2 × ODc < OD < 4 × 0Dc), and strong biofilm producers: (OD > 4 × 0Dc) (Stepanovic et al., 2000).

The bacterial strains that demonstrated positive results for the production of AHLs were grown in MB broth at 25°C with shaking (120 rpm). Cultures were realized in 6 1-L Erlenmeyer flasks, each containing 500 mL of Marine Broth (MB) 2216 (BD Difco, Sparks, MD). After 48 h, ethyl acetate (EtOAc, 200 mL) was added to each flask, and the ground mixture was shaken for 12 h at room temperature (150 rpm). After dehydration on anhydrous Na₂SO₄ and filtration, the organic phase was evaporated to dryness. The ethyl acetate extracts were treated prior to analysis using solid-phase extraction (Phenomenex Strata C18, 55 1.1m). The stationary SPE phase was activated with 100 mL of acetonitrile. The cartridge was loaded with 100 mg of each extract, which was solubilized in 2 mL of acetonitrile. The elution was performed with 50 mL of acetonitrile to eliminate the most lipophilic metabolites. The eluates were dried and dissolved in DMSO.

The treated extracts were then fractionated by liquid chromatography on a Phenomenex Luna C18 (5 mm, 250 × 21.20 mm, 100 Å) column with a separation HPLC system with 2 Varian Prep Star pumps, a manual injector, a Dionex Ultimate 3000 RS variable wavelength detector and a Dionex Ultimate 3000 fraction collector. The solvent was made of gradient grade H₂O and CH₃CN (70:30 for 3 min, followed by a 12-min linear gradient from 70:30 to 0:100, followed by 100% CH₃CN for 10 min) and pumped at a flow rate of 20 mL.min⁻¹. The eluents were monitored at 214, 254, 274, and 280 nm and were collected between 3 and 25 min (1 fraction/min, 22 fractions). The fractions were dried with a genevac HT-4X system and were dissolved in 100 μL DMSO to evaluate the presence of AHL with the biosensors and diluted.

UHPLC-MS analyses were performed with a Dionex Ultimate 3000 UHPLC-HESI HRMS Q Exactive focus (Thermo Scientific) system controlled by Thermo Xcalibur software (Thermo scientific). Elution was performed on a Hypersil gold C18 (2.1 × 150 mm) column with 1.9 μm particle size (Thermo Scientific) maintained at 50°C; solvent A consisted of water with 0.1% formic acid, and solvent B was acetonitrile with 0.1% formic acid. The gradient profile for chromatography was as follows: 100% solvent A for 5 min, a linear increase in solvent B to 100% over 5 min, hold at 100% solvent B for 5 min, and then equilibrate with 100% solvent A for 2 min. The flow rate was maintained at 0.8 mL.min⁻¹, and the injection volume was 5 μL of the fractions diluted to 1 mg.mL⁻¹. The detection using ESI was performed in positive mode. The ion source parameters were: capillary voltage 3 kV, aux gas flow rate 20, sheath gas flow rate 75, spray current 4 μA, capillary temperature 350°C, heater temperature 450°C, and S-lens RF 40. Nitrogen was used as the nebulizing gas in the ion trap source.

The MS/MS method was designed to perform an MS1 full-scan (scan range m/z 130–900) alternating with a MS/MS scan [AIF, all-ion fragmentation, scan range 60–600 m/z, normalized-collision energy (NCE) 25]. The mass resolution was set at 35,000, the AGC targets were 1 × 10⁶ and 5 × 10⁴ for the full scan and AIF modes, respectively, and injection time was 40 ms.

Standard AHLs (0.1 mg/mL in DMSO, 2 μL injected), were analyzed to determine the fragmentation patterns and retention time under our chromatographic conditions. As previous studies reported (Morin et al., 2003; Patel et al., 2016), we observed some characteristic fragments arising from fragmentation of the lactone ring. A daughter ion with m/z 102.055 is always observed, and three others ions with m/z 56.050, 74.061, and 84.045 are usually detected. These four daughter ions were chosen as the specific ions to indicate the presence of HSL-type compounds. To confirm the AHL identification, a second step of MS analyses was performed using SIM (resolution 3,500) and dd-ms² (resolution 17,500) modes. Each ion of interest was isolated using an isolation width of 3.0 m/z and fragmented with a resolution of 17 500 and a collision energy of 20 eV. The presence of the daughter ion with 102.055 m/z and at least one other daughter ion that is characteristic of the lactone ring were needed to confirm AHL identification. This confirmation has also been conducted using a combination of 25 commercially available AHL standards with acyl chains that vary in their length, the nature of substituent on their third carbon (hydroxy or oxo group) and the presence of unsaturation (see more details of our methods in Girard et al., 2017 and Supplementary Table 2).

Sixty bacterial strains were isolated from P. oceanica (16SrRNA genes Accession numbers KY787128 to KY787185, KY697665 and KY697666, see Supplementary Table 1). Thirty-nine were from leaves (only epiphytic strains), and 21 were from rhizomes (both epiphytic and endophytic strains). The identified strains are affiliated with distinct groups of bacteria representing a broad spectrum of diversity, including Alphaproteobacteria (72%), Gammaproteobacteria (20%), Flavobacteriia (3%) Bacilli, and Actinobacteria (5%). Rhodobacteraceae-affiliated isolates were dominate among the Alphaproteobacteria fraction (55% of Alphaproteobacterial isolates), while isolates affiliated with the genus Marinomonas sp. dominated the Gammaproteobacteria (14%) (Figures 1, 2). With the second isolation protocol, we focused on the endophytic bacteria that were present in the Posidonia rhizomes. A similarly broad diversity pattern was observed with the endophytic strains, which belong to Alphaproteobacteria (70%), Gammaproteobacteria (20%), and Flavobacteriia (10%). The collected bacteria were also affiliated with a large diversity of genera, including Labrenzia sp. (17 strains), Cohaesibacter sp. (3 strains), Ruegeria sp. (3 strains), Pseudophaeobacter sp. (4 strains), Nisaea sp. (4 strains), Marinomonas sp. (4 strains), Vibrios sp. (3 strains), Micrococcus sp. (2 strains), Aquimarina sp. (2 strains), Microbulbifer sp. (1 strain), Alteromonas sp. (1 strain) Pseudovibrio sp. (1 strain), Flavivirga sp. (1 strain), Sphingopyxis sp. (3 strains), Sphingorhabdus sp. (1 strain), Kordiimonas sp. (1 strain), and Erythrobacter sp. (1 strain). Eight strains were found to be phylogenetically too distant from the known isolates to be affiliated with a specific genus with enough confidence (Figure 1).

Among the 60 isolated strains, 6 were identified as AHL-producing isolates: 3 strains were double positive (responded to the P. putida F117 and the E.coli MT102 reporter strains), suggesting the production of long and short chain AHLs, and 3 strains gave rise only to a positive signal with the E. Coli MT102 reporter strain, revealing the production of short chain AHLs (Figure 1, and Datasheet S1). All six AHL-producing strains were isolated from both the rhizomes and leaves of P. oceanica and were affiliated with Alphaproteobacteria: 3 strains are related to to Pseudophaeobacter sp., 2 strains appeared phylogenetically related to the genus Cohaesibacter sp. and 1 strain of Labrenzia sp.

Among the whole isolates, the AI-2 biosensor V. harveyi BB170 responded positively to the supernatant of 19 isolated strains (Figures 1, 2, and Datasheet S1). The autoinducer secretion was measured by calculating the increased induction of bioluminescence using data collected at the optimal time point, i.e., the moment at which the control sample produced the least amount of light (Taga, 2005). A total of 6 strains were found to be strong producers of AI-2 and exhibited a >10-fold induction of bioluminescence, while 13 strains were weak producers with an induction between 5- and 10-fold. The AI-2 producers were found among isolates from Alphaproteobacteria (78%), Gammaproteobacteria (11%), and Flavobacteriia (11%). Among the Alphaproteobacteria, the genus Labrenzia seemed to be a large producer of AI-2, with 12 strains able to activate the AI-2 biosensor among the 17 that were isolated. Five of these strains were isolated from rhizomes, and 14 were associated with the leaves.

The bacterial ability to interfere in AHL-based quorum sensing was evaluated with the E. Coli MT102 reporter strain. Five strains (9%) produced quorum quenching signals. Four strains were related to Alphaproteobacteria (affiliated with the genera Kordiimonas, Erythrobacter, Sphingorhabdus, and an unknown Rhodospirillaceae). Additionally, one strain affiliated with Gammaproteobacteria (related to the genus Vibrio) was positive in our quorum quenching bioassay. Three of these strains were isolated in the leaves of Posidonia, and two were sampled from the rhizomes (Figures 1, 2 and Datasheet S1).

A crystal violet-based microplate assay was used to evaluate the capacity of microbial strains to attach to abiotic surfaces. We classified our isolates based on the calculated cut-off OD, resulting in the following categories: non-biofilm producers (OD < 3), weak biofilm producers (0.3 < OD < 0.6), moderate biofilm producers (0.6 < OD < 1.2), and strong biofilm producers (OD >1.4). A total of 67% of the P. oceanica-associated strains were able to produce biofilm (75% related to Alphaproteobacteria). Among them 11 strains are weak biofilm producers, 15 are moderate producers and 14 are strong biofilm producers (Figures 1, 2 and Datasheet S1). In contrast, 19 strains did not produce a biofilm. We observed that closely phylogenetically related strains responded differently to this biotest. For example, among the16 isolated strains related to the Labrenzia genus, 3 did not produce biofilm, while 3, 6, and 4 of the isolated strains related to the Labrenzia genus were either weak, moderate or strong biofilm producers, respectively. Interestingly, OD values above 1 were observed for the 4 strains that are related to Pseudophaeobacter sp. (and isolated from P. oceanica leaves), while the 4 strains related to Nisaea sp. were unable to produce biofilm. All strains identified as AHL producers were also, according to our bioassays, able to produce biofilms, while 58% of the strains involved in AI-2-based quorum sensing were able to synthesize a biofilm (Figures 1, 2 and Datasheet S1).

We combined HPLC fractionation and UHPLC-HRMS analyses to identify the diversity of the AHLs secreted by the 6 bacterial strains that our biosensors identified as positive. Our method consists of a non-targeted UHPLC-MS/MS analysis to identify AHLs in microbial samples. Our chromatographic method yielded good separation, and the mass accuracy for all standards was within 3 ppm (parts per million). The analytical method includes full-scan MS with successive all-ion fragmentation (AIF) MS/MS scan allowing the acquisition high resolution data of the parent ion and daughter ion masses in a single chromatographic run. As described by Patel et al. (2016), four characteristic fragments originating from the lactone ring in AHLs were detected in almost all of the MS/MS spectra of the AHL standards (Girard et al., 2017). One daughter ion was abundant and always present in fragmentation patterns: the 102.055 m/z ion, which corresponds to the lactone ring itself; three other ions that arose from the fragmentation of the lactone ring (the 56.050, 74.061, and 84.045 m/z fragments) were usually observed. These daughter ions were sought in the MS/MS scans of bacteria fractions to detect HSL-type compounds. All AHLs were also confirmed with commercially available standards.

Thus, we determine that the 3 strains related to Pseudophaeobacter sp. (BBCC2174, 2175, and 2176) produced 5–6 different AHLs with acyl side chains harboring 10–16 carbons with unsaturated regions or/and hydroxyl substitution (Table 1). The two Cohaesibacter strains, BBCC2161 and BBCC2195, as well as the Labrenzia strain BBCC2184 produce respectively 6, 5, and 6 AHLs with acyl chain harboring between 8 and 16 carbons; in which unsaturated regions or/and hydroxyl substitution could be found too. Because of the use of our non-targeted UHPLC-MS/MS method allowing the detection of known and novel AHLs and because of the presence in MS/MS spectra of the daughter ions characteristic of the lactone ring, we also suspect the presence of many other HSL-type compounds. However, the confirmation of such result is highly difficult. Especially, the Nuclear Magnetic Resonance step requires large amounts of pure compounds, while AHLs are emitted at very low concentrations (Table 1).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.