Indole decreases the virulence of the bivalve model pathogens Vibrio tasmaniensis LGP32 and Vibrio crassostreae J2-9

Indole signaling plays an important role in bacterial pathogenesis. In this study, the impact of indole on biofilm formation, swimming and swarming motility were explored in Vibrio tasmaniensis LGP32 and Vibrio crassostreae J2-9, two model pathogens of bivalves. The results showed that indole decreased swimming and swarming motility in both strains, and decreased biofilm formation in V. crassostreae J2-9. Furthermore, indole affected a large number of genes at RNA level, including genes related to metabolism, ABC transporters, flagellar assembly, chemotaxis, and response regulators. Finally, the bacterial virulence towards mussel larvae was decreased by pretreatment with indole in both V. tasmaniensis LGP32 and V. crassostreae J2-9. After 5 days, the survival rate of mussel larvae increased 2.4-fold and 2.8-fold in mussel larvae challenged with V. tasmaniensis LGP32 pretreated with 200 µM and 500 µM indole, respectively. The survival rate of mussel larvae increased 1.5-fold and 1.9-fold in mussel larvae challenged with V. crassostreae J2-9 pretreated with 200 µM and 500 µM indole, respectively. These data indicate that indole has a significant impact on the virulence of V. tasmaniensis LGP32 and V. crassostreae J2-9, and indole signaling could be a promising target for antivirulence therapy.

World mussel aquaculture production has been increasing steadily since the 1950s 1 , and currently 94% of the world mussel production comes from aquaculture 2 . In Europe, mussels are economically the most important aquaculture species as its production represents more than one-third of EU aquaculture production. The blue mussel (Mytilus edulis) is one of the main cultivated mussel species 3 , and it is known as a robust bivalve species. However, its larviculture appears to be highly susceptible to diseases and mass mortality occurs in dense larval cultures, which is a bottleneck for sustainable development of mussel culture 4 . These mortality events are most probably caused by bacterial pathogens such as vibrios 5,6 . Vibrios belonging to the Splendidus clade (i.e. Vibrio splendidus and closely related species) can infect a broad host range of aquaculture animals, including mussels 7 , oysters 8 , shrimp 9 , sea cucumber Apostichopus japonicus 10 and fish 11 , resulting in various diseases and high mortality and leading to high economic losses. Vibrio tasmaniensis LGP32 and Vibrio crassostreae J2-9 are two model bacterial pathogens of marine bivalves such as oyster and mussel [12][13][14][15] . It has been reported that V. tasmaniensis LGP32 and V. crassostreae J2-9 caused more than 80% and 70% mussel larvae mortality respectively after 5 days of challenge 7,15 .
Antibiotics often are the only effective agents that farmers have to protect their animals from bacterial infections. As a result of the frequent use of antibiotics in aquaculture, the adverse multifactorial effects of antibiotic resistance have increasingly been demonstrated and have become a serious problem for public health 16 . Therefore, there is an urgent need for a new strategy replacing the use of antibiotics 17 . Among the newly developed therapeutic strategies, antivirulence therapy has been proposed as a promising alternative. Instead of killing pathogens, antivirulence therapy aims at disarming pathogens in order to inhibit their virulence, thereby preventing them from attacking the host 18 . Amongst the most intensively studied targets for antivirulence therapy are bacterial cell-to-cell communication systems (quorum sensing) 19 . Vibrios typically contain a multichannel Impact of indole on swimming and swarming motility, and biofilm formation. In order to determine the mechanism of how indole affects the virulence of V. tasmaniensis LGP32 and V. crassostreae J2-9, we investigated its impact on swimming and swarming motility and biofilm formation. First, we tested the impact of indole on swimming motility. The results showed that indole inhibited the swimming motility of both V. tasmaniensis LGP32 and V. crassostreae J2-9 (Fig. 3A). After 2 days, the diameters of the swimming halos produced by cultures of V. tasmaniensis LGP32 and V. crassostreae J2-9 treated with 500 µM of indole were decreased to 17% and 49%, respectively, of those of untreated cultures. In the second experiment, swarming motility exhibited by V. tasmaniensis LGP32 and V. crassostreae J2-9 on 0.8% agar was also significantly reduced by indole (Fig. 3B). After 7 days, the swarming motility halos produced by V. tasmaniensis LGP32 were 21% and 27% smaller in the presence of 200 and 500 µM indole, respectively, when compared to untreated cultures. The swarming motility of V. crassostreae treated with 200 and 500 µM of indole was 26% and 41% smaller, respectively, than the motility of untreated cultures. In the third experiment, we determined the impact of indole on biofilm formation. Indole was found to significantly decrease the biofilm formation in V. crassostreae J2-9 but showed no effect on biofilm formation of V. tasmaniensis LGP32 (Fig. 3C). The biofilm in V. crassostreae J2-9 was reduced to 49% and 17% in the presence of 200 and 500 µM indole, respectively, compared with that of the non-treated control.
Transcriptional profiling of V. tasmaniensis LGP32 during indole treatment. Since the above results indicated that indole inhibits the production of several virulence factors, we sought to further determine the genes that are transcriptionally regulated in this context. Thus, we performed comparative RNA sequencing (RNA-seq) on strain LGP32 (whose genome is publicly available on NCBI) during growth in the presence or LGP32 and V. crassostreae J2-9. Bacteria were grown in LB 35 broth and the production of indole was measured using Kovac's reagent. The error bars represent the standard deviation of three independent experiments.  (Fig. 4). The Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis revealed that a large number of differentially expressed genes (DEGs) were related to metabolism, ABC transporters, flagellar assembly, chemotaxis, and response regulators (Table 1). Based on the transcriptomic analysis, several genes associated with flagellar motility were found to be downregulated in V. tasmaniensis LGP32 that was treated with indole, including genes encoding flagellin, polar flagellin B, and flagellar protein FlaG. Genes associated with methyl-accepting chemotaxis proteins, universal stress protein A, and carbon starvation protein A were also downregulated, indicating that the treatment with indole significantly decreased the expression of genes related to chemotaxis and stress response. Notably, one gene encoding a two-component system response regulator was also downregulated. Further, the expression of several flagellin related genes including flaB (encoding flagellin), flaE (encoding polar flagellin), and flaN (encoding polar flagellin B) were decreased, although the changes were not significant (fold change between 1.0 and 1.5; Table S1). In contrast, a substantial number of genes related to ABC transporter permease and ATP synthase were significantly upregulated ( Table 1).

Impact of indole on the expression of selected genes in V. tasmaniensis
LGP32 and V. crassostreae J2-9. To confirm the transcriptomic analysis in V. tasmaniensis LGP32, we further validated the results by RT-qPCR. The expression of seven genes were verified in V. tasmaniensis LGP32 treated with 200 and 500 µM indole, i.e. flaG (encoding flagellar protein), flaN (encoding flagellin), flaB (encoding polar flagellin), pflaB (encoding polar flagellin B), tnaA (encoding tryptophanase, which produces indole), cspA (encoding carbon starvation protein A) and tcS (encoding two-component system response regulator) (Fig. 5A). Consistent with the transcriptomic results, all of these genes were down-regulated (p < 0.01). qPCR also confirmed that genes associated with ABC transporter permease (ABC tpr encoding ABC transporter permease-related gene), ATP synthase (atpE encoding putative ATP synthase subunit C) were upregulated.
To further investigate the impact of indole on V. crassostreae J2-9, the genes fliF (encoding flagellar M-ring protein), flaG (encoding flagellar protein), flgE (encoding flagellar hook protein), cspA (encoding carbon starvation protein A) and pmbA (encoding metalloprotease PmbA) were tested by RT-qPCR. As shown in Fig. 5B, 500 µM indole also down-regulated the mRNA levels of genes related to motility including fliF, flaG, and flgE. These data were substantiated by the fact that indole could significantly inhibit the swimming and swarming motility of V. crassostreae J2-9.
Impact of indole on the virulence of vibrios towards mussel larvae. To explore the impact of indole on the virulence of V. tasmaniensis LGP32 and V. crassostreae J2-9 towards mussel larvae, vibrios pretreated with different concentrations of indole were added to mussel larvae. The pathogens were pretreated with indole, and washed before addition to the mussel larvae in order to exclude any direct effect of indole on the larvae. Both V. tasmaniensis LGP32 and V. crassostreae J2-9 showed to be pathogenic to mussel larvae, leading to significant mortality after 5 days of challenge (Fig. 6). However, the treatment with indole improved the survival of the lar-

Discussion
In the present study, we investigated the impact of indole on virulence factor production and virulence of V. tasmaniensis LGP32 and V. crassostreae J2-9 towards mussel larvae. We found that pretreatment of V. tasmaniensis LGP32 and V. crassostreae J2-9 with indole before inoculation into the mussel larvae rearing water resulted in decreased mortality when compared to the larvae that were challenged with untreated pathogens. The higher the concentration of indole, the lower the mortality of mussel larvae was. This indicated that elevated indole levels significantly decreased the virulence of V. tasmaniensis LGP32 and V. crassostreae J2-9. The results obtained in this study are consistent with our previous observation that indole decreases the virulence of Vibrio campbellii, V. parahaemolyticus, and V. harveyi strains towards gnotobiotic brine shrimp (Artemia franciscana) larvae 27,28 , and that of V. anguillarum NB10 towards sea bass (Dicentrarchus labrax) larvae 26 . All these data further indicated that indole signaling has the potential to be an effective target for antivirulence therapy in aquaculture. However, for practical applications we will need to find indole analogues that do not affect the cultured organisms because LGP32 and V. crassostreae J2-9. Data are presented as the mean ± SD of six replicates for swimming and swarming motility and three independent experiments for biofilm formation. For each strain, the swimming motility, swarming motility and biofilm formation in the control treatment was set at 100% and the other treatments were normalized accordingly. Different letters indicate significant differences (One way ANOVA with Tukey's post hoc test; P < 0.01). www.nature.com/scientificreports/ adding indole to the rearing water affects fish larvae (positive effect 26 ), brine shrimp larvae (negative effect 27 ) and mussel larvae (negative effect 29 ). Indole displays a diverse range of effects on bacterial physiology and metabolism 27,30 , yet its molecular targets and mechanisms of action remain obscure 24 . To determine the mechanisms by which indole reduces the virulence of V. tasmaniensis LGP32 and V. crassostreae J2-9, we evaluated the impact of indole on biofilm formation, swimming and swarming motility. Swimming motility (in liquids) and swarming motility (on wet surfaces) are common modes of motility, both are important throughout the bacterial infection 31 . Swimming motility enables bacteria to detect and pursue nutrients, and move toward environments of favorable conditions 32,33 . Swarming motility allows bacteria to rapidly colonize a surface, leading to the formation of biofilms 34 . Indole decreased swimming and swarming motility in both V. tasmaniensis LGP32 and V. crassostreae J2-9. These effects were confirmed at the transcriptional level by transcriptomic analysis and reverse transcriptase qPCR targeting key genes involved in these phenotypes. The expression of genes related to bacterial flagella and chemotaxis was repressed by indole. This result is similar to that obtained for Salmonella enterica serovar Typhimurium 35 and V. campbelli 27 in which indole repressed motility and the production of flagella. Further, indole also decreased the biofilm formation of V. crassostreae J2-9 but did not affect the biofilm formation of V. tasmaniensis LGP32. Indeed, the direction of the effect (up-or downregulation of biofilm formation) caused by indole is different in different bacterial species. For example, indole inhibited biofilm formation in Acinetobacter oleivorans 36 , Bdellovibrio bacteriovorus 37 , Pseudomonas aeruginosa 38 , E. coli 39 , S. enterica serovar Typhimurium 35 and V. campbellii 27 , while it increased the biofilm formation in Agrobacterium tumefaciens 40 and Burkholderia unamae 41 . Additionally, indole had no impact on the motility of V. anguillarum in our previous study 26 . These differences might reflect differences in the life styles of the bacteria.
According to the transcriptomic analysis, indole treatment elevated the expression of ATP synthase and genes involved in the ABC transporter permease in LGP32. Likewise, Kim et al. 42 conducted a transcriptome analysis for Pseudomonas putida KT2440 under indole treatment. They demonstrated that 47 genes were differentially expressed, of which 12 genes involved in chaperone and protease functions. Subsequent biochemical analyses showed that the presence of indole improved the membrane perturbation and promoted higher expression of genes associated with TCA cycle, resulting in the decrease of the ATP concentration inside cells. Our results suggest that a similar mechanism might be present in V. tasmaniensis LGP32.
Indole is a signaling molecule produced by many bacterial species and involved in intraspecies, interspecies, and interkingdom signaling 24 . To date, more than 85 bacterial species have been found to produce indole 23 . In this study, the concentration of indole produced reached about 200 µM and 250 µM in V. tasmaniensis LGP32 and V. crassostreae J2-9, respectively. It is well recognized that the accumulation of extracellular indole can be affected by environmental factors, such as the cell density, carbon sources, temperature and pH 43 . In this study, the levels of indole produced by V. tasmaniensis LGP32 and V. crassostreae J2-9 were dependent on cell density, which is similar to what has been reported for Edwardsiella tarda 44 , E. coli 45 , V. anguillarum 26 and V. campbellii 27 .
In conclusion, indole showed protection of blue mussel larvae against V. tasmaniensis LGP32 and V. crassostreae J2-9 without affecting bacterial growth. This is consistent with the concept of antivirulence therapy, which does not kill the pathogens 19 . Further, indole was found to control different phenotypes in V. tasmaniensis LGP32 and V. crassostreae J2-9, including biofilm levels, swimming and swarming motility and mRNA levels of genes responsible for these phenotypes. All of the results indicate that indole signaling has the potential to be used as a target for antivirulence therapy in blue mussel larviculture. Interestingly, when compared to conventional antibiotics, interfering with signaling mechanisms (like indole signaling) is expected to impose less selective pressure (no impact on growth) on bacteria to evolve resistance leading to a lower chance of resistance development (although it cannot be excluded at this moment). Indeed, we did not observe any trend towards resistance in our study, nor has resistance to indole been documented for any other bacterium.

Methods
Bacterial strains, culture conditions, and chemicals. Vibrio tasmaniensis LGP32 and V. crassostreae J2-9 were cultured in Luria-Bertani medium containing 35 g/L of sodium chloride (LB 35 ) at 28 °C under constant agitation (100 min −1 ). Cell densities were measured spectrophotometrically at 600 nm. Indole, purchased from Sigma-Aldrich (Belgium), was dissolved in methanol at 100, 200, 500 mM, respectively. In all experiments, all treatments received the same volume of methanol.

Quantification of indole. Vibrio tasmaniensis
LGP32 and V. crassostreae J2-9 cultures were grown overnight in LB 35 broth (reaching OD 600 of 1) and re-inoculated (1% v/v) into fresh LB 35 broth. Cell free supernatants from the cultures were obtained at the time intervals of 6, 12, 24 and 48 h by centrifugation at 8000×g, followed by filtration through 0.22 µm membrane filters. The concentration of indole in the supernatants was measured as described previously 26 by mixing 500 µL of supernatant with 500 µL of Kovac's reagent. After vortexing, the top 200 µL was removed and the OD 571 was measured. The indole concentration in each sample was determined based on a standard curve using synthetic indole (Sigma-Aldrich). At least three different cultures were sampled for each strain at each time point.

Impact of indole on bacterial growth.
To investigate the effect of indole on the growth of V. tasmaniensis LGP32 and V. crassostreae J2-9, overnight grown cells were inoculated into fresh LB 35 media at an initial OD 600 of 0.01. Indole was added at 0, 100, 200 and 500 µM, respectively. Then 200 µL aliquots of these suspensions were pipetted into the wells of a polystyrene 96-well plate and cultured in Tecan Infinate M200Pro plate reader at 28 °C for 40 h. The OD 600 of each sample was measured by a Tecan Infinate M200Pro plate reader every hour. Growth curves were determined for three independent cultures.

RNA extraction and quantitative reverse transcription PCR (RT-qPCR). Overnight cultured V.
tasmaniensis LGP32 and V. crassostreae J2-9 were diluted to 1% in fresh LB 35 medium supplemented with or without indole at 200 and 500 µM. Each sample was grown in triplicate at 28 °C for 6 h. RNA extraction and reverse transcriptase real-time PCR were performed as described previously 46 . Total RNA was extracted using the SV Total RNA Isolation System (Promega) and DNA contamination was eliminated with a DNase treatment (Thermo Scientific Rapid Out DNA Removal Kit). The RNA quantity was measured spectrophotometrically (NanoDrop Technologies) and adjusted to 200 ng/µL in all samples. The RNA quality was confirmed with Agarose Gel Electrophoresis and the RNA samples were stored at − 80 °C. The cDNA was synthesized from RNA with reverse transcriptase using the ProtoScript (r) II First Strand cDNA Synthesis Kit (New England biolabs) according to the manufacturer's instructions. Briefly, 1 µg RNA, 2 µL random primer and Nuclease-free H 2 O was mixed to a total volume of 8 µL. Then the samples were denatured for 5 min at 65 °C and put promptly on ice. After that, 10 μL of 2 × Reaction Mix and 2 μL of 10 × Enzyme Mix were added. The 20 μL cDNA synthesis reactions were incubated at 25 °C for 5 min, followed by 60 min at 42 °C and 80 °C for 5 min. Finally, cDNA samples were cooled to 4 °C, were checked by PCR and stored at − 20 °C for further use. Real-time qPCR was carried out in a StepOne ™ Real-Time PCR System thermal cycler (Applied Biosystems, Gent, Belgium) by using the Luna Universal qPCR Master Mix (New England biolabs). The primers for qPCR are listed in Table 2, and each assay was performed in triplicate, expression of 16S rRNA was used as the internal reference gene to normalize the expression of the genes in Table 2, with the primers of 933F and16SRTR1. The comparative threshold cycle method (2 −ΔΔCT ) 49 was used to analyze the relative mRNA level.
Challenge tests with blue mussel D-larvae. Challenge tests were performed as described previously 7 .
Wild-caught mature blue mussels were stimulated to spawn by thermal shocks in autoclaved seawater at 5 °C and 20 °C until gametes were released. Spawning males and females were transferred to sterile plastic cups containing 50 mL sterile seawater and allowed to spawn for 15 min. Sperm and eggs were collected and gently mixed at a 10:1 ratio in a beaker containing 1 L of sterile seawater. After the appearance of polar bodies, the eggs were gently rinsed with sterile seawater using a sterile 30 μm sieve to remove excess sperm. Fertilized eggs were incubated in 2 L of sterile seawater (max 100 eggs/mL) containing chloramphenicol, nitrofurazone and enrofloxacin (each at Figure 4. The difference between V. tasmaniensis LGP32 without and with 500 µM indole at the whole transcriptome level. (A) Volcano plot of differentially expressed genes (DEGs). Red indicates upregulated expression, blue indicates downregulated expression and black indicates no significantly differential expression. FC: fold change; FDR: false discovery rate. (B) Functional enrichment of differentially expressed genes on KEGG categorization in V. tasmaniensis LGP32 (control vs. 500 µM indole). The Rich factor is the ratio of differentially expressed gene numbers annotated in this pathway term to all gene numbers annotated in this pathway term. The higher the Rich factor, the higher the degree of pathway enrichment. The q value is the corrected p value (q = p*length(p)/rank(p)); a lower value indicates higher pathway enrichment. www.nature.com/scientificreports/ 10 mg/L). After two days of incubation, D-larvae were harvested on a sterile 60 μm sieve. The larvae were washed gently with sterile seawater to remove the antibiotics. Rinsed D-larvae were transferred to a beaker containing 1 L of sterile seawater and distributed uniformly using a plunger. Subsamples were taken to calculate the larval density, and the density was corrected to obtain a final concentration of 250 larvae/mL. All manipulations were performed under a laminar flow hood. Natural rifampicin-resistant mutants V. tasmaniensis LGP32 and V. crassostreae J2-9 were cultured overnight and transferred to fresh LB 35 , and further cultured with 200 µM and 500 µM indole for 6 h, respectively. The treatment without indole contained the same volume of methanol. One mL aliquots of the larval suspension in seawater supplemented with 10 mg/L tryptone and 5 mg/L yeast extract were subsequently transferred to 24-well plates. Rifampicin was added at 10 mg/L to avoid contamination. The final larval density was 200 larvae/mL. Natural rifampicin-resistant mutants of V. tasmaniensis LGP32 and V. crassostreae were inoculated into the rearing water at 10 5 cells/mL. Larvae to which no bacteria were added and that were otherwise treated in the same way as challenged larvae, were used as controls. Each treatment was performed in 24 replicates. The plates were incubated at 18 °C. Each day, four replicates per treatment were stained with lugol [5% (v/v)], and stained larvae were counted under a binocular microscope (Nikon Eclipse E 200, Nikon Instruments Europe). Larvae were considered alive when stained black by lugol, death if only parts of the larvae were stained or if shells were empty. Figure 5. The impacts of indole on mRNA levels of selected genes in V. tasmaniensis LGP32 (A) and V. crassostreae J2-9 (B). Bacteria were grown in LB 35 medium supplemented with indole at 0, 200 or 500 µM and cultured for 6 h, respectively. Cells were collected for total RNA extraction and used for real-time RT-PCR. The relative mRNA levels were normalized to that of 16S rRNA. The mRNA level of the treatment without indole was set at 1 and the other treatments were normalized accordingly using the ΔΔC T method. Error bars represent the standard deviation of three independent experiments. Different letters indicate significant differences (One way ANOVA with Tukey's post hoc test; P < 0.01). www.nature.com/scientificreports/