Algal Auxins Decrease the Virulence of Vibrio Campbellii, a Major Pathogen of Aquatic Organisms

Environmental Microbiology (2017) 19(5), 1987–2004 doi:10.1111/1462-2920.13714

Indole signalling and (micro)algal auxins decrease the virulence of Vibrio campbellii, a major pathogen of aquatic organisms

Qian Yang,1 Gde Sasmita Julyantoro Pande,1 indole-3-acetamide, which were produced by various Zheng Wang,2 Baochuan Lin,2† Robert A. Rubin,3 (micro)algae sharing the aquatic environment with Gary J. Vora2 and Tom Defoirdt1,4* V. campbellii, have a similar effect as observed for 1Laboratory of Aquaculture & Artemia Reference indole. Auxins might, therefore, have a significant Center, Ghent University, Gent, Belgium. impact on the interactions between vibrios, (micro)al- 2Center for Bio/Molecular Science & Engineering, Naval gae and higher organisms, with major ecological and Research Laboratory, Washington, DC, USA. practical implications. 3Mathematics Department, Whittier College, Whittier, CA, USA. 4Center for Microbial Ecology and Technology (cmet), Ghent University, Gent, Belgium. Introduction Vibrio campbellii and related species belonging to the Har- Summary veyi clade of vibrios, are marine Gram-negative bacteria that can infect a wide range of wild and cultured aquatic Vibrios belonging to the Harveyi clade are major organisms (both vertebrates and invertebrates), leading to pathogens of marine vertebrates and invertebrates, significant losses in the aquaculture industry worldwide causing major losses in wild and cultured organisms. (Austin and Zhang, 2006; Ruwandeepika et al., 2012; Despite their significant impact, the pathogenicity Dong et al., 2016). Despite their role as major pathogens mechanisms of these bacteria are not yet completely of marine animals, the pathogenicity mechanisms of these understood. In this study, the impact of indole signal- bacteria are not yet fully understood. The factors that have ling on the virulence of Vibrio campbellii was been considered to be involved in pathogenicity include investigated. Elevated indole levels significantly biofilm formation, swimming motility and the production of decreased motility, biofilm formation, exopolysac- a variety of extracellular products (Karunasagar et al., charide production and virulence to crustacean 1994; Austin and Zhang, 2006; Ruwandeepika et al., hosts. Indole furthermore inhibited the three-channel 2012; Yang and Defoirdt, 2015). quorum sensing system of V. campbellii, a regulatory Since virulence factors are often costly metabolic prod- mechanism that is required for full virulence of the ucts, their expression usually is under strict regulatory pathogen. Further, indole signalling was found to control. One of the regulatory mechanisms controlling the interact with the stress sigma factor RpoS. Together production of virulence factors in V. campbellii is quorum with the observations that energy-consuming pro- sensing, cell-to-cell communication with secreted signalling cesses (motility and bioluminescence) are molecules (Defoirdt et al., 2008). V. campbellii strain ATCC downregulated, and microarray-based transcriptom- BAA-1116 (previously designated as V. harveyi;Linet al., ics demonstrating that indole decreases the 2010) is a model bacterium in quorum sensing expression of genes involved in energy and amino research and has been found to contain a three-channel acid metabolism, the data suggest that indole is a quorum sensing system that uses three kinds of signalling starvation signal in V. campbellii. Finally, it was molecules: harveyi autoinducer 1 (HAI-1; Cao and Meigh- found that the auxins indole-3-acetic acid and en,1989), autoinducer 2 (AI-2; Chen et al., 2002) and cholerae autoinducer 1 (CAI-1; Higgins et al., 2007). Previ- Received 13 January, 2017; accepted 25 February, 2017. *For cor- ous work in our lab has shown that the three-channel respondence. E-mail [email protected]; Tel. 132 (0)9 264 quorum sensing system is required for full virulence of V. 59 76; Fax 132 (0)9 264 62 48. †Present address: Defense Threat Reduction Agency, Chemical and Biological Technologies, Fort campbellii toward different aquatic hosts (Defoirdt and Sor- Belvoir, VA, USA. geloos, 2012; Pande et al., 2013).

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd 1988 Q. Yang et al. In recent years, more compounds have been added to its ecological perspective, algal-bacterial signalling mecha- the arsenal of bacterial signalling molecules, and one of nisms (e.g., using indoles) are of significant interest to these compounds is indole (Lee and Lee, 2010). Indole is aquaculture as they might be manipulated to steer the produced by tryptophanase (TnaA), which reversibly con- microbial activity in aquaculture systems in order to verts tryptophan into indole, pyruvate and ammonia (Lee decrease the incidence of diseases (Natrah et al., 2014). et al., 2007). Although a large variety of both Gram- Given the ecological and economical importance of V. positive and Gram-negative bacteria (more than 85 spe- campbellii as a major pathogen of aquatic organisms, the cies including many pathogens) have long been known to previous reports documenting the impact of indole on produce indole, knowledge of its role as a signalling mole- virulence-related phenotypes and virulence of vibrios, and cule is relatively recent (Lee et al., 2015b). Some bacteria the potential impact of microalgal metabolites on this pro- that lack tryptophanase, and therefore do not produce cess, we sought to determine the impact of indole indole, for example, Pseudomonas aeruginosa,also signalling on the virulence of V. c amp be ll ii and to investi- respond to the presence of extracellular indole (Chu et al., gate whether the indole analogues indole-3-acetic acid 2012). Diverse biological responses to indole have recently and indole-3-acetamide induce a similar response as been revealed, including spore formation (Stamm et al., indole. 2005), drug resistance (Hirakawa et al., 2005), plasmid stability (Chant and Summers, 2007), persister formation Results (Vega et al., 2012), motility (Bansal et al., 2007), lipopoly- Impact of indole on the virulence of V. campbellii toward saccharide production (Han et al., 2011) and biofilm gnotobiotic brine shrimp larvae and conventionally formation (Mueller et al., 2009). Additionally, indole has reared giant river prawn larvae been reported to interfere with acylhomoserine lactone (AHL)-based quorum sensing in a number of Gram- To examine whether indole could decrease the virulence of negative bacteria including Acinetobacter oleivorans, Chro- wild type V. c amp be ll ii, we performed a standardized chal- mobacterium violaceum, Pseudomonas chlororaphis and lenge test with gnotobiotic brine shrimp larvae. The Serratia marcescens (Kim and Park, 2013; Hidalgo- addition of 50 and 100 mM of indole to the brine shrimp Romano et al., 2014). To date, however, a role for indole rearing water (single addition at the start of the experi- signalling in vibrios has only been documented in V. ch ol - ment) significantly increased the survival of the challenged erae and very recently in V. anguillarum.InV. cholerae larvae (Supporting Information Table S1). However, indole strain SIO, indole activated Vibrio polysaccharide (VPS) appeared to be toxic to the larvae at higher concentrations production and biofilm formation, whereas it decreased since significant mortality was observed in larvae (both motility (Mueller et al., 2009). In V. anguillarum strain challenged and unchallenged) exposed to 200 mMindole. NB10, by contrast, exopolysaccharide production was Therefore, in a second experiment, V. c amp be ll ii was incu- decreased by indole, whereas motility was not affected (Li bated in the presence of indole at 50, 100 and 200 lM, et al., 2014). Furthermore, in V. anguillarum,indole respectively, after which the cultures were washed to signalling and the stress sigma factor RpoS are intercon- remove the indole prior to inoculation into the brine shrimp nected as deletion of rpoS increased indole production rearing water. This way, we could eliminate any direct (Li et al., 2014). effect of indole to the larvae, enabling us to unambiguously Vibrios share the aquatic environment with (micro)algae, determine the impact of indole on (the virulence of) V. and both groups of organisms are known to interact with campbellii. A significantly increased survival of brine each other in various ways (ranging from mutualistic to shrimp larvae challenged with indole-pretreated V. ca mp - antagonistic) (Mayali and Azam, 2004; Amin et al., 2015). bellii was observed (when compared with larvae While these interactions are most likely coordinated by sig- challenged with untreated V. c amp be ll ii), indicating that nalling mechanisms between bacteria and algae, little is indole indeed decreases the virulence of V. ca mp be ll ii known with respect to signalling mechanisms that occur (Table 1). Importantly, all treatments received the same between algae and bacteria (Zhou et al., 2016). Interest- inoculum (105 cells per ml brine shrimp rearing water), and ingly, various (micro)algae produce indole analogues such there were no differences between treatments with respect as indole-3-acetic acid and indole-3-acetamide (Support- to the density of V. campbellii in the brine shrimp rearing ing Information Fig. S1) (Stirk et al., 2013; Lu and Xu, water after 1 and 2 days of challenge (Supporting Informa- 2015). Although several bacteria (predominantly plant- tion Fig. S2), indicating that all larvae had been exposed to associated bacteria and the human pathogens E. coli and the same density of vibrios. P. aeruginosa) have also been reported to respond to Based on these promising results, we went further to these natural indole analogues (Spaepen and Vanderley- investigate the effect of indole on the survival of challenged den, 2011; Melander et al., 2014), their impact on marine larvae of the giant river prawn (Macrobrachium rosenber- bacteria has not been investigated thus far. In addition to gii), a commercially important aquaculture species. Similar

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 Indole and auxins control the virulence of vibrios 1989

Table 1. Survival of gnotobiotic brine shrimp larvae after 2 days of challenge with V. campbellii ATCC BAA-1116 (average 6 standard deviation of 3 shrimp cultures).

Treatment Survival (%)1 No challenge 100 6 0d V. campbellii 53 6 6a b V. campbellii [50 mM indole]pretreatment 72 6 8 c V. campbellii [100 mM indole]pretreatment 90 6 5 c V. campbellii [200 mM indole]pretreatment 93 6 3

V. campbellii was either untreated or pretreated with indole prior to inoculation into the brine shrimp rearing water. “Control” refers to unchallenged larvae that were otherwise treated in the same way as the challenged larvae. 1Values with a different superscript letter are significantly different from each other (One way ANOVA with Tukey’s post-hoc test; P < 0.01). to what we observed in brine shrimp larvae, the survival of giant river prawn larvae challenged with indole-pretreated V. ca mp be ll ii was significantly higher than that of larvae challenged with untreated V. campbellii (Table 2). None of the treatments had any effect on the growth of the surviv- ing larvae, as manifested by the larval stage index (LSI).

Impact of indole on biofilm formation and exopolysaccharide production by V. campbellii

In order to determine the mechanism by which indole affects the virulence of V. c amp be ll ii ,weinvestigatedits impact on virulence factor production. Indole has been reported to control biofilm formation and exopolysaccharide production in V. cholerae SIO (Mueller et al., 2009) and V. anguillarum NB10 (Li et al., 2014). Therefore, we determined the impact of indole on biofilm formation in V. campbellii by crystal violet staining, and found that indole decreased biofilm levels, with a fivefold decrease at 200 mM (Fig. 1a). Since exopolysaccharide production is one of the major factors affecting biofilm formation, we subsequently investigated the impact of indole on Fig. 1. Impact of indole on (a) biofilm and (b) exopolysaccharide levels on polystyrene 96-well plates and (c) swimming motility of V. Table 2. Survival and growth (as manifested by the larval stage campbellii on soft agar plates. The error bars represent the index, LSI) of conventionally reared giant river prawn larvae after 6 standard deviation of three independent experiments for biofilm and days of challenge with V. campbellii ATCC BAA-1116 (average 6 exopolysaccharide, and six replicates for swimming motility. standard deviation of 5 prawn cultures). Different letters indicate significant differences (One way ANOVA with Tukey’s post hoc test; P < 0.01). Treatment Survival (%)1 LSI No challenge 100 6 8c 4.4 6 0.5A V. campbellii 51 6 10a 4.4 6 0.5A exopolysaccharide levels, and found that the addition of b A V. campbellii [100 mM indole]pretreatment 70 6 6 4.4 6 0.5 indole also significantly decreased exopolysaccharide production, with a sevenfold decrease at 200 mM (Fig. 1b). V. campbellii was either untreated or pretreated with indole prior to inoculation into the giant river prawn rearing water. “Control” refers Importantly, indole at these concentrations shows no effect to unchallenged larvae that were otherwise treated in the same way on the planktonic growth of V. campbellii (Supporting Infor- as the challenged larvae. mation Fig. S3). 1The survival in the non-challenged treatment (83% 6 7%) was set at 100% and the other treatments were normalized accordingly. Val- The Vibrio polysaccharide (vps) genes are responsible ues with a different superscript letter are significantly different from for exopolysaccharide production and biofilm formation in each other (One way ANOVA with Tukey’s post-hoc test; P < 0.01). vibrios, and the regulators vpsT and vpsR have been

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 1990 Q. Yang et al. reported before to be controlled by indole in V. ch ol er ae significantly decreased in the presence of indole (Support- SIO (Mueller et al., 2009). VpsR is the main activator of ing Information Table S2). Further experiments showed Vibrio polysaccharide production in V. cholerae,andVpsT that indole significantly decreased mRNA levels of the lat- is a secondary activator that acts synergistically with VpsR eral flagellar regulator lafK and the flagellin lafA from early (Yildiz et al., 2001). The V. c amp be ll ii ATCC BAA-1116 stationary phase onward, with a 2.0- and 2.7-fold differ- vpsT and vpsR genes show 73% and 70% nucleotide ence between indole-treated and untreated cells in homology with the V. cholerae analogues respectively. We stationary phase respectively (Table 3). determined the impact of indole on the expression of the vpsT and vpsR genes in V. ca mp be ll ii by quantitative reverse transcriptase PCR and found that the addition of Interaction between the stress sigma factor RpoS and indole resulted in approximately twofold decreased mRNA indole signalling in V. campbellii levels of vpsT and vpsR from early stationary phase We previously reported that the alternative sigma factor onward (Table 3). RpoS controls indole production in V. anguillarum NB10, with an rpoS deletion mutant showing a significantly increased production when compared with the wild type (Li Impact of indole on the swimming motility of et al., 2014). To determine whether this regulation also V. campbellii exists in V. ca mp bell ii, we investigated the indole production Since motility plays a major role in the virulence of V. in V. campbellii wild type and an rpoS1 rpoS2 double dele- campbellii (Yang and Defoirdt, 2015), we subsequently tion mutant (the V. campbellii ATCC BAA-1116 genome determined the effect of indole on the swimming motility of contains two functional rpoS genes). Consistent with what V. campbellii on soft agar. Indole was found to significantly has been reported for V. anguillarum, indole production decrease the swimming motility in a concentration- was also significantly higher in the DrpoS1 Drpos2 mutant dependent manner, with a 2.7-fold decrease at 200 mM than in the wild type at all time points tested (Fig. 2a). Fur- (Fig. 1c). We further investigated the effect of indole on the thermore, the expression level of the indole synthase gene expression of 10 selected genes involved in flagellar motili- tnaA was significantly higher in the DrpoS1 Drpos2 mutant ty in V. ca mp be ll ii (Yang and Defoirdt, 2015). A first than in the wild type at all sampling points (threefold at 6 h; screening revealed that there was no significant impact on fourfold at 12 h and threefold at 24 h; Fig. 2b). the genes involved in the synthesis of the polar flagellum RpoS has been reported to be involved in the response and in chemotaxis, whereas the mRNA levels of to indole in V. ch ol er ae (Mueller et al., 2009), and we found genes involved in the synthesis of lateral flagella were that rpoS mRNA levels in V. anguillarum NB10 are higher

Table 3. Relative mRNA levels of Vibrio polysaccharide production regulators, lateral flagellar genes, stationary phase sigma factors and the quorum sensing master regulator in V. campbellii ATCC BAA-1116 during incubation in LB35 broth in the absence and presence of 100lM indole (average 6 standard deviation of three independent cultures).

Relative mRNA levels (fold)1

12 h (early 6 h (exponential phase) stationary phase) 24 h (stationary phase) Gene Function Non-treated 100 lM indole Non-treated 100 lM indole Non-treated 100 lM indole Vibrio polysaccharide (VPS) production regulators vpsR Main activator of VPS production 1.0 6 0.2 1.1 6 0.3 2.0 6 0.3 1.0 6 0.4** 3.4 6 0.2 1.4 6 0.3** vpsT Secondary activator of VPS production 1.0 6 0.1 0.9 6 0.2 1.5 6 0.1 0.8 6 0.1** 3.6 6 0.4 2.1 6 0.1***

Motility genes lafA Lateral flagellar flagellin 1.0 6 0.2 0.8 6 0.2 4.9 6 0.7 3.6 6 0.4** 18.2 6 1.7 6.7 6 1.0*** lafK Lateral flagellar regulator 1.0 6 0.2 0.8 6 0.4 4.5 6 0.3 2.7 6 0.4** 16.4 6 1.5 8.3 6 0.6**

Regulators rpoS1 Stationary phase sigma factor, copy 1 1.0 6 0.2 1.3 6 0.6 1.7 6 0.3 4.0 6 0.4** 11.4 6 2.3 27.8 6 3.8** rpoS2 Stationary phase sigma factor, copy 2 1.0 6 0.1 1.3 6 0.2 1.2 6 0.4 2.1 6 0.1* 4.6 6 0.8 10.2 6 1.1*** luxR Three-channel quorum sensing system 1.0 6 0.1 1.0 6 0.1 1.0 6 0.2 0.9 6 0.2 1.2 6 0.3 0.6 6 0.1** master regulator

For each gene, the mRNA level in untreated V. campbellii at the 6 h time point was set at 1 and the levels in the other samples were normalized accordingly using the 22DDCT method. The RNA polymerase A subunit gene (rpoA) was used as an internal control. 1Asterisks indicate significant differences when compared with non-treated V. campbellii at the respective time point (independent samples T-test; *P < 0.05; **P < 0.01; ***P < 0.001).

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 Indole and auxins control the virulence of vibrios 1991 high cell density in order to activate quorum sensing- controlled bioluminescence, after which indole was added at 50, 100 and 200 lM, respectively, and bioluminescence was measured after 1 h. Indole significantly inhibited the quorum sensing-regulated bioluminescence in wild type V. campbellii at all concentrations tested (Fig. 3). We further sought to confirm that the effect of indole on bioluminescence was due to interference with the quorum sensing system by determining the impact of indole on bioluminescence of mutant LuxO D47E containing plasmid pAKlux1(Defoirdtet al., 2012). This strain contains a point mutation in luxO which renders the quorum sensing system inactive, and plasmid pAKlux1 contains the Photo- rhabdus luminescens bioluminescence operon under the control of a constitutive promoter. Hence, bioluminescence is independent of the three-channel quorum sensing system in this strain. Indole had no effect on the bioluminescence of LuxO D47E pAKlux1 at any of the concentrations tested (Fig. 3), indicating that the inhibition of bioluminescence in wild type V. c amp be ll ii was indeed caused by interference with the three-channel quorum sensing system. We recently proposed the specific quo-

rum sensing inhibitory activity AQSI as a new parameter to reliably identify quorum sensing inhibitors based on experiments with signal molecule reporter strains (Yang et al.,

2015; Defoirdt, 2016). The AQSI values obtained for indole Fig. 2. Indole production in V. campbellii wild type and the rpoS1 were 32, 49 and 28 at 50, 100 and 200 mM respectively. rpoS2 double deletion mutant. (a) Indole levels (bars) and cell In order to determine at what stage in the quorum sens- densities (lines) of V. campbellii during growth in LB35 medium. (b) Relative mRNA levels of the indole biosynthesis gene tnaA. The ing signal transduction cascade indole is acting, we used tnaA mRNA level in the wild type at 6 h was set at 1 and the other mutants LuxO D47A and Dhfq. These mutants show con- 2DDCT data points were normalized accordingly using the 2 method. stitutive luminescence irrespective of cell density and The RNA polymerase A subunit gene (rpoA) was used as an internal control. For both panels, error bars represent the standard signal molecule concentration, and consequently, blocking deviation of three independent V. campbellii cultures. Asterisks of luminescence in these mutants would indicate that indicate a significant difference when compared with wild type V. indole interacts with a quorum sensing signal transduction campbellii at the respective time point (independent samples T-test; ** P < 0.01, *** P < 0.001). component located downstream of the mutated one (Defoirdt et al., 2007). A scheme of the V. campbellii three- upon exposure to exogenous indole as well (Supporting channel quorum sensing system is presented in Fig. 4 for Information Fig. S4). Consistent with this, we found that further information. Indole was found to block lumines- from early stationary phase onward, the mRNA levels of cence in LuxO D47A and Dhfq at all the concentrations both rpoS1 and rpoS2 were significantly higher in V. camp- tested, although the effect was less pronounced in the bellii upon exposure to 100 mM exogenous indole (Table 3). Dhfq mutant (Fig. 3). This indicates that indole blocks the three-channel quorum sensing system by interfering with quorum sensing signal transduction. Consistent with this, Impact of indole on the three-channel quorum sensing we found that indole decreases the mRNA levels of the system of V. campbellii master regulator of the three-channel quorum sensing signal transduction cascade, LuxR (Table 3). Indole has been reported to interfere with quorum sensing in various bacteria (Kim and Park, 2013; Hidalgo-Romano Impact of indole on the V. campbellii transcriptome et al., 2014), and because all phenotypes that were affected by indole in V. campbellii are also controlled by its We used microarray analyses to determine whether in three-channel quorum sensing system, we went further to addition to the genes identified in our activity and RT determine whether indole interfered with the activity of the qPCR assays, other genes are differentially expressed in three-channel quorum sensing system, using biolumines- the presence of exogenous indole. We compared profiles cence as a readout. Wild type V. campbellii was grown to from wild type V. ca mp be ll ii grown for 24 h with or without

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 1992 Q. Yang et al.

Fig. 3. Impact of indole on bioluminescence of V. campbellii wild type, constitutively luminescent quorum sensing signal transduction mutants LuxO D47A and Dhfq, and mutant LuxO D47E containing plasmid pAKlux1, in which bioluminescence is independent of quorum sensing. For each strain, bioluminescence in the control treatment was set at 100% and the other treatments were normalized accordingly. The error bars represent the standard deviation of three replicates. Different letters indicate significant differences (One way ANOVA with Tukey’s post hoc test; P < 0.01).

Fig. 4. Schematic representation of the quorum sensing system of Vibrio campbellii. The LuxM, LuxS and CqsA enzymes synthesise the autoinducers HAI-1, AI-2 and CAI-1 respectively. These autoinducers are detected at the cell surface by the LuxN, LuxQ and CqsS two- component receptor proteins respectively. Detection of AI-2 by LuxQ requires the periplasmic protein LuxP. In the absence of autoinducers (left), the receptors autophosphorylate and transfer phosphate to LuxO via LuxU. Phosphorylation activates LuxO, which together with r54 activates the production of five small regulatory RNAs (sRNAs). The sRNAs promote translation of the master regulator AphA and inhibit translation of the master regulator LuxRVh. In the presence of high concentrations of the autoinducers (right), the receptor proteins switch from kinases to phosphatases, which results in dephosphorylation of LuxO. Dephosphorylated LuxO is inactive and therefore, the sRNAs are not formed, AphA is not translated and LuxRVh is translated. AphA and LuxR are transcriptional regulators that (either individually or together) affect the transcription of many target genes.

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 Indole and auxins control the virulence of vibrios 1993 the addition of 100 mM indole and found that the levels of blocked the three-channel quorum sensing system of V. 64 transcripts differed by 1.5-fold or more, 21 of which dif- campbellii since they blocked quorum sensing-regulated fered twofold or more (Table 4). About 17 transcripts were bioluminescence in wild type V. c amp be ll ii and had no found to be more abundant (6 of which more than twofold), effect on quorum sensing independent bioluminescence in whereas 47 transcripts were less abundant (15 of which strain LuxO D47E pAKlux1 (Fig. 5c and d). The AQSI val- twofold or more). The largest difference in mRNA levels ues for indole-3-acetic acid were >39, 19 and >37 at 50, were observed for fliR, which encodes a protein involved in 100 and 200 mM, respectively, and the values for indole-3- export of lateral flagellar flagellin (88-fold more abundant in acetamide were 13, 9 and 14 at 50, 100 and 200 mM the presence of indole). The differentially expressed tran- respectively. Similar to indole, both auxins interfered with scripts comprised a variety of functions. Several of them quorum sensing signal transduction since they blocked are linked to cell wall and membrane biogenesis and trans- luminescence of the constitutively luminescent LuxO D47A port mechanisms (Table 4). When considering the COG mutant. However, in contrast to what we observed for classification groups of differentially expressed transcripts, indole, the auxins did not affect the bioluminescence of the the largest fraction was found to be involved in metabolic hfq deletion mutant, indicating that the exact mechanism processes, mainly amino acid metabolism and transport of interference is slightly different. and energy production and conversion (Supporting Infor- mation Table S3). Most of the genes in these categories Discussion were found to be less abundant in the presence of 100 mM Indole has recently been proposed as an intercellular sig- indole (16/16 and 9/12 for amino acid metabolism and nal molecule in bacteria, affecting various behaviours (Lee transport, and energy production and conversion et al., 2015b). In the present study, we demonstrated that respectively). indole controls the virulence of V. campbellii in a highly None of the genes we identified to be differentially controlled model system with gnotobiotic brine shrimp lar- expressed by reverse transcriptase qPCR were detected vae. Gnotobiotic conditions are important in this respect as such in the microarray analyses. This is most probably since many bacteria are capable of producing high levels related to a higher sensitivity of reverse transcriptase of indole (Lee and Lee, 2010) and therefore, the microbiota qPCR when compared with microarrays (which we have that is naturally associated with the host might confuse the also observed previously in studies on the regulation of vir- experiments. We observed significantly decreased viru- ulence factors by the three-channel quorum sensing lence of V. c amp be ll ii in this model system when the system). Nevertheless, several transcripts from the lucifer- pathogen was pretreated with indole at 50 mMormore, ase operon (luxB, luxD, luxE and luxG) were found to be and this was confirmed in another model system using less abundant in the presence of indole (Table 4), which is conventionally reared giant river prawn larvae. The results consistent with our observation that indole blocks the obtained in this study are consistent with our previous three-channel quorum sensing system of V. c amp be ll ii .In observation that indole decreases the virulence of V. fact, transcripts of the quorum sensing master regulator luxR were found to be significantly less abundant in the anguillarum NB10 toward sea bass (Dicentrarchus labrax) microarray analysis as well. However, the difference was larvae (Li et al., 2014). Therefore, a decreased virulence below the 1.5-fold threshold (1.3-fold; P 5 0.001). upon exposure to elevated indole levels seems to be con- served among distantly related marine vibrios that are pathogenic to various aquatic organisms. The pathogen Impact of indole analogues produced by (micro)algae was pretreated with indole in order to avoid a direct effect on virulence and virulence-related phenotypes in of indole on the host, which would confuse interpretation of V. campbellii the results. Indeed, indole has been reported to exert a Various (micro)algae are known to produce the auxin plant direct effect on higher organisms, for example, by increas- hormones indole-3-acetic acid and indole-3-acetamide, ing epithelial cell tight-junction resistance (Bansal et al., and therefore, we went further to test the impact of these 2010), and we previously found indole to have a beneficial auxins on the phenotypes that were affected by indole in V. effect on gnotobiotic sea bass (Dicentrarchus labrax)lar- campbellii. Both compounds showed a similar effect as vae (Li et al., 2014). In this study, we also observed a indole, since pretreatment with auxins significantly direct (though negative) effect of indole on brine shrimp lar- decreased the virulence of V. c amp be ll ii toward gnotobiotic vae as significant mortality was observed in brine shrimp brine shrimp larvae (Table 5). Both compounds also larvae exposed to 200 mMindole. decreased biofilm (Fig. 5a) and exopolysaccharide levels In order to determine the mechanism(s) by which indole (Fig. 5b). In contrast to indole, however, the auxins had no reduces the virulence of V. campbellii, we evaluated the effect on swimming motility of V. c amp be ll ii (Supporting impact of indole on several important virulence factors that Information Table S4). Finally, both auxins also specifically are affected by indole in other vibrios (Table 6). Indole

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 1994 Q. Yang et al.

Table 4. Vibrio campbellii genes differentially expressed in the presence of 100mM indole, as determined by microarray analyses (fold difference).

NCBI Accession numbers Annotation Fold difference P value More abundant in the presence of 100 mM indole M892_01150 VIBHAR_02455 Choline/carnitine/betaine transporter 16.1 0.000 M892_03895 VIBHAR_01857 Trimethylamine-N-oxide reductase 1.7 0.000 M892_03905 VIBHAR_01855 NapE 1.8 0.000 M892_05930 VIBHAR_01423 Rhomboid family integral membrane protein 1.5 0.000 M892_06530 VIBHAR_01300 Polar flagellin (FlaC) 1.5 0.000 M892_16980 VIBHAR_02782 Coenzyme A disulfide reductase 2.0 0.001 M892_17235 VIBHAR_02727 Cytochrome C nitrite reductase subunit c552 1.9 0.001 M892_24170 VIBHAR_05451 Cytochrome c-type protein (TorC) 1.6 0.000 M892_18365 VIBHAR_06844 Thiamine-phosphate pyrophosphorylase 4.3 0.001 M892_18810 VIBHAR_06739 Major facilitator transporter multidrug resistance protein D 1.5 0.001 M892_19395 VIBHAR_06599 Alpha-1,2-mannosidase 1.9 0.001 M892_22310 VIBHAR_05888 Permease 1.6 0.000 M892_24415 VIBHAR_05392 Membrane protein 1.8 0.000 M892_25985 VIBHAR_05043 Probable phage protein 2.3 0.001 M892_26245 VIBHAR_04987 Glyoxylase I family protein 3.2 0.001 M892_26355 VIBHAR_04965 Flagellar biosynthesis protein (FliR) 88.0 0.001 M892_26630 VIBHAR_04905 Membrane-bound lytic murein transglycosylase D (FlgO) 1.5 0.001 Less abundant in the presence of 100 mM indole M892_01345 VIBHAR_02414 Tricarboxylic transporter (TctC) 21.6 0.000 M892_01680 VIBHAR_02342 NAD-glutamate dehydrogenase 21.6 0.000 M892_01855 VIBHAR_02302 Cbb3-type cytochrome oxidase, cytochrome c subunit 21.5 0.000 M892_01860 VIBHAR_02300 Cytochrome C oxidase, subunit (CcoQ) 21.5 0.000 M892_01890 VIBHAR_02294 Universal stress protein (UspE) 21.9 0.001 M892_01940 VIBHAR_02284 NAD-dependent deacetylase 22.0 0.000 M892_02445 VIBHAR_02179 2-Keto-4-pentenoate hydratase 21.5 0.000 M892_02450 VIBHAR_02178 Dioxygenase 21.6 0.000 M892_02455 VIBHAR_02177 4-Hydroxyphenylpyruvate dioxygenase 22.0 0.000 M892_03130 VIBHAR_02029 Hypothetical protein 26.2 0.000 M892_03135 VIBHAR_02028 Cysteine desulfurase 22.3 0.000 M892_05360 VIBHAR_01551 Amino acid ABC transporter permease (HisQ) 21.9 0.001 M892_06250 VIBHAR_01355 2-Oxoglutarate dehydrogenase, E1 component 24.8 0.000 M892_07505 VIBHAR_01084 Sodium/alanine symporter 21.5 0.000 M892_07710 VIBHAR_01039 Peroxidase, AhpC/Tsa family 21.5 0.000 M892_08325 VIBHAR_00913 Carbamoyl phosphate synthase, large subunit 22.1 0.000 M892_08945 VIBHAR_00784 Inorganic pyrophosphatase 22.0 0.000 M892_09655 VIBHAR_00638 TonB-dependent receptor, iron transport 21.5 0.000 M892_11530 VIBHAR_00225 RNA polymerase, beta subunit (RpoB) 21.6 0.000 M892_11765 VIBHAR_00169 Acyl-CoA synthetase 21.9 0.000 M892_11805 VIBHAR_00160 Sodium/solute symporter 21.9 0.000 M892_11810 VIBHAR_00159 RNA polymerase subunit sigma-32 21.7 0.000 M892_12230 VIBHAR_00076 Succinyldiaminopimelate aminotransferase 21.5 0.000 M892_12410 VIBHAR_00043 N-acetyl-gamma-glutamyl-phosphate reductase 22.1 0.000 M892_12415 VIBHAR_00042 Acetylglutamate kinase 21.7 0.000 M892_12420 VIBHAR_00041 Argininosuccinate synthase 22.0 0.000 M892_12425 VIBHAR_00040 Bifunctional argininosuccinate lyase/N-acetylglutamate synthase 22.3 0.000 M892_12965 VIBHAR_03647 Aspartate carbamoyltransferase 22.6 0.000 M892_13885 VIBHAR_03440 Bifunctional aconitate hydratase 2/2-methylisocitrate dehydratase 21.8 0.000 M892_14740 VIBHAR_03271 Na(1)-translocating NADH-quinone reductase subunit E 21.7 0.001 M892_16230 VIBHAR_02949 Oligopeptide ABC transporter, periplasmic protein 21.5 0.000 M892_18990 VIBHAR_06695 Malate synthase 21.6 0.000 M892_20780 VIBHAR_06256 Transaldolase B 22.1 0.000 M892_20830 VIBHAR_06243 Myristoyl-ACP-specific thioesterase (LuxD) 21.8 0.000 M892_20840 VIBHAR_06241 Alkanal monooxygenase beta chain (LuxB) 21.5 0.000 M892_20845 VIBHAR_06240 Long-chain-fatty-acid luciferin-component ligase (LuxE) 21.6 0.000 M892_20850 VIBHAR_06239 NAD(P)H-dependent FMN reductase (LuxG) 21.5 0.000 M892_21955 VIBHAR_05973 Glycine dehydrogenase protein P 22.0 0.001 M892_21975 VIBHAR_05968 Glycine cleavage system protein T2 21.6 0.000 M892_23285 VIBHAR_05670 Moxr-like ATPase AAA 21.6 0.001

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 Indole and auxins control the virulence of vibrios 1995

Table 4. cont.

NCBI Accession numbers Annotation Fold difference P value M892_23875 VIBHAR_05531 Arginine transporter permease subunit (ArtM) 21.7 0.000 M892_23880 VIBHAR_05530 Arginine transporter permease subunit (ArtQ) 21.9 0.000 M892_23885 VIBHAR_05529 Arginine ABC transporter substrate-binding protein 21.8 0.000 M892_24220 VIBHAR_05438 Hypothetical protein 21.8 0.000 M892_24280 VIBHAR_05422 Pterin-4-alpha-carbinolamine dehydratase 25.4 0.000 M892_24325 VIBHAR_05412 Alcohol dehydrogenase 21.6 0.000 M892_24525 VIBHAR_05368 Poly(3-hydroxyalkanoate) synthetase (PhaC) 22.0 0.000 mRNA levels of genes that are listed were significantly different between cells grown in the presence of indole, when compared with cells grown in the absence of indole, and the difference was at least 1.5-fold. Fold change in expression is represented as the ratio between the average log2 values for cells grown in the presence of indole and cells grown in the absence of indole (average values of 5 experiments).

decreased biofilm formation, exopolysaccharide production effect (up- vs. downregulation) is different for different spe- and motility of V. campbellii, and these effects were con- cies. This might reflect differences in the life styles of the firmed at the transcriptional level by reverse transcriptase bacteria. Indeed, when compared with marine vibrios, vari- qPCR targeting key genes involved in these phenotypes. ous regulatory responses have been reported to be The decreased exopolysaccharide production could, opposite in the freshwater human pathogen V. cholerae together with the decreased flagellar motility, explain the (e.g., the downregulation of virulence factors and biofilm decrease in biofilm formation induced by indole in V. ca mp - formation by the multichannel quorum sensing system; Mil- bellii. Indeed, both phenotypes play an important role in ton, 2006). The observation that indole decreased the biofilm formation in bacteria (Petrova and Sauer, 2012). motility of V. ca mp be ll ii and the expression of genes The decreased biofilm formation and exopolysaccharide involved in flagellar biosynthesis is consistent with what production we observed here in V. c amp be ll ii is consistent has been reported for V. cholerae (Mueller et al., 2009), E. with what we observed before in V. anguillarum (Li et al., coli (Lee et al., 2007), Salmonella enterica serovar Typhi- 2014) and with what has been reported for E. coli (Lee murium (Nikaido et al., 2012) and P. aeruginosa (Lee et al., et al., 2007; Bansal et al., 2007). However, these results 2009a). In contrast, indole had no impact on motility of V. are in contrast with what has been reported for V. cholerae anguillarum in our previous study (Li et al., 2014). SIO, where deletion of the indole synthase gene tnaA We investigated the interaction of indole signalling with resulted in decreased biofilm formation, and this was other key regulatory mechanisms, and found that indole restored upon addition of 350 mM indole (Mueller et al., interferes with quorum sensing-regulated bioluminescence 2009). Hence, although indole consistently alters biofilm in V. c amp be ll ii . This activity could be specifically attributed formation in various bacterial species, the direction of the to inhibition of the three-channel quorum sensing system

since the specific quorum sensing-disrupting activity AQSI of indole was higher than 10, which is comparable to the Table 5. Survival of gnotobiotic brine shrimp larvae after 2 days of most specific quorum sensing inhibitors revealed previous- challenge with V. campbellii ATCC BAA-1116 (average 6 standard deviation of 3 shrimp cultures). ly (Yang et al., 2015). Consistent with this, we found that indole decreased the mRNA levels of the quorum sensing Treatment Survival (%)1 master regulator LuxR. The decreased luxR mRNA levels No challenge 100 6 3d can explain the impact of indole on swimming motility, bio- V. campbellii 52 6 3a film formation, and bioluminescence since all these b V. campbellii [50 mM indole-3-acetic acid]pretreatment 73 6 8 phenotypes are controlled by the three-channel quorum b V. campbellii [100 mM indole-3-acetic acid]pretreatment 77 6 6 c sensing system (Anetzberger et al., 2009; Yang and V. campbellii [200 mM indole-3-acetic acid]pretreatment 85 6 5 b Defoirdt, 2015). The inhibition of quorum sensing by indole V. campbellii [50 mM indole-3-acetamide]pretreatment 72 6 10 c V. campbellii [100 mM indole-3-acetamide]pretreatment 83 6 3 in V. c amp be ll ii contrasts with what has been reported for c V. campbellii [200 mM indole-3-acetamide]pretreatment 87 6 3 V. ch ol er ae SIO, where no relation between indole sensing and quorum sensing was observed (Mueller et al., 2009). V. campbellii was either untreated or pretreated with indole-3-acetic acid or with indole-3-acetamide prior to inoculation into the brine Further research will be needed in order to determine how shrimp rearing water. “Control” refers to unchallenged larvae that indole is detected by vibrios and to unravel the indole sens- were otherwise treated in the same way as the challenged larvae. ing mechanism (including the identification of the 1The survival in the non-challenged treatment (98% 6 3%) was set at 100% and the other treatments were normalised accordingly. Val- molecular mechanism by which the indole signal feeds into ues with a different superscript letter are significantly different from the complex quorum sensing signal transduction cascade each other (One way ANOVA with Tukey’s post-hoc test; P < 0.01). of V. campbellii). As far as we know, the indole sensing

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 1996 Q. Yang et al.

Fig. 5. Impact of the auxins indole-3-acetamide and indole-3-acetic acid on biofilm formation, exopolysaccharide production, and quorum sensing in V. campbellii. (a) Biofilm and (b) exopolysaccharide levels of V. campbellii on polystyrene 96-well plates. Error bars represent the standard deviations of three independent experiments. Different letters indicate significant differences (One way ANOVA with Tukey’s post hoc test; P < 0.01). Impact of the auxins (c) indole-3-acetamide and (d) indole-3-acetic acid on bioluminescence of V. campbellii wild type, constitutively luminescent quorum sensing signal transduction mutants LuxO D47A and Dhfq, and mutant LuxO D47E containing plasmid pAKlux1, in which bioluminescence is independent of quorum sensing. For each strain, the bioluminescence in the control treatment was set at 100% and the other treatments were normalized accordingly. The error bars represent the standard deviation of three replicates. mechanism has not been elucidated for any bacterium, We further determined the interaction between indole although the transcriptional regulator SdiA has been sensing and the alternative sigma factor RpoS since the hypothesized to be central in the indole signalling cascade latter has previously been reported to regulate the produc- in E. coli (Lee et al., 2009b) and the DnaK suppressor pro- tion of indole in E. coli (Lelong et al., 2007) and V. tein DksA has been suggested to be involved in the indole anguillarum (Li et al., 2014). In agreement with what we signalling cascade in V. cholerae SIO (Mueller et al., observed in V. anguillarum,theV. campbellii DrpoS1 2009). The V. c amp be ll ii ATCC BAA-1116 genome does DrpoS2 double mutant produced higher indole levels than not contain a homologue of sdiA, but it does contain a the wild type, and this was reflected in a higher expression dksA homologue (GenBank accession no. M892_13860), of the indole synthase tnaAintheDrpoS1 DrpoS2 mutant. showing 82% identity at the nucleotide level with V. chol- However, these results are opposite to what has been erae dksA. Despite repeated attempts, we were not able to revealed in E. coli, in which RpoS enhanced indole produc- obtain a dksA deletion mutant (which would have enabled tion (Lelong et al., 2007). We also found that indole in turn us to determine whether DksA is involved in indole sensing increased expression of rpoS in wild type V. ca mp be ll ii, in V. c amp be ll ii), and this might indicate that this gene is which is also consistent with what we observed in V. essential for the viability of V. c amp be ll ii. Nevertheless, the anguillarum NB10. This is in contrast with what has been fact that all virulence factors that we found to be affected reported for V. cholerae SIO, where rpoS was downregu- by indole are also controlled by the three-channel quorum lated by indole (Mueller et al., 2009). We also noticed that sensing system suggests that the decreased virulence V. campbellii and V. anguillarum produce lower indole lev- might be a consequence of the inhibition of the three- els (60–150 mMat24h)thanV. c hol er ae (600 mMat24h). channel quorum sensing system. Indeed, previous Given the fact that RpoS is known as the stress sigma research at our laboratories revealed that an active three- factor in bacteria, we hypothesize that indole induces a channel quorum sensing system is required for full viru- stress response in V. ca mp be ll ii. Several other pieces of lence toward various hosts (Defoirdt and Sorgeloos, 2012; evidence support this hypothesis. First, indole is produced Pande et al., 2013). from tryptophan and this reaction also generates pyruvate

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 Indole and auxins control the virulence of vibrios 1997

Table 6. Comparison of the impact of indole between different species of Gram-negative bacteria.1

Species and strain Virulence Biofilm Motility rpoS Quorum sensing References Acinetobacter oleiovorans DR1 ## # Kim and Park (2013) Agrobacterium tumefaciens C58 # Kim and Park (2013) Agrobacterium tumefaciens2 "# Lee et al. (2015a) Chromobacterium violaceum ATCC 12472 ##Hidalgo-Romano et al. (2014) Escherichia coli K-12 ## Lee et al. (2008) Escherichia coli O157:H7 ## Bansal et al. (2007), Lee et al. (2011) Pseudomonas aeruginosa PAO1 "# # Lee et al. (2009a), Lee et al. (2011) Edwardsiella tarda LTB-4 "" Han et al. (2011) Serratia marcescens ATCC 13880 ##Hidalgo-Romano et al. (2014) Vibrio anguillarum NB10 ##5 " Li et al. (2014) Vibrio campbellii ATCC BAA-1116 ###"# This study Vibrio cholerae SIO "##5 Mueller et al. (2009)

1" and #: increased and decreased activity/expression at increasing levels of indole respectively; 5: no impact. 2Strain not mentioned.

(Lee et al., 2015b). Consequently, tryptophan can serve as possible at this moment to unambiguously determine at an energy source under nutrient limitation. Because indole which stage exactly indole and auxins feed into this path- is produced in the same reaction, it can serve as a reliable way. Based on our observations, we can derive that indole starvation indicator. Furthermore, the microarray analysis feeds into the system downstream of the sRNA-hfq stage, indicated that the majority of transcripts that were differen- whereas auxins interfere at an earlier stage downstream of tially expressed are involved in energy production and LuxO (presumably the sRNA-hfq stage). However, this will conversion, and in amino acid metabolism and transport. also mask any intereference with a signal transduction Moreover, most of these genes were downregulated in the stage located further downstream. Therefore, we cannot presence of indole. Finally, the inhibition of the three- exclude that in addition to the sRNA-hfq stage, the auxins channel quorum sensing system also resulted in also feed into the system at the same stage as indole. decreased motility and bioluminescence, both of which are The auxin concentrations that affected gene expression high energy-consuming processes (McCarter, 2004; Miya- in this study are several orders of magnitude higher than shiro and Ruby, 2012). those reported in the marine environment (Amin et al., Vibrios, including those belonging to the Harveyi clade, 2015). However, auxin concentrations will be significantly are known to be able to associate with (micro)algae (Neal- higher inside an algal bloom, and particularly in the phyco- son and Hastings, 2006; Manset et al., 2013; Main et al., sphere, where local auxin concentrations are much higher 2015), which (like terrestrial plants) are known to produce than in the bulk due to diffusive boundary layers (Jonsson various indole analogues as auxin hormones (Stirk et al., et al., 2009; Amin et al., 2015). Auxins might therefore have 2013; Lu and Xu, 2015). Auxins have been reported to a significant impact on the interactions between vibrios and affect various phenotypes in terrestrial bacteria (Spaepen (micro)algae, and given the fact that auxins are produced and Vanderleyden, 2011). However, despite the potential by many marine (micro)algal taxa (Stirk et al., 2013; Lu and ecological significance as cross-kingdom signals in the Xu, 2015), this phenomenon might not be uncommon in aquatic environment, the impact of auxins that are known to the marine environment. The downregulation of biofilm for- be produced by (micro)algae on aquatic bacteria has thus mation by auxins might be a mechanism to prevent far not been investigated. We found that the auxins indole- extensive colonization of (micro)algae by marine vibrios, a 3-acetic acid and indole-3-acetamide showed similar effects mechanism that has previously also been suggested for as indole. Indeed, pretreatment of V. campbellii with auxins quorum sensing-disrupting brominated furanones produced resulted in a similar decrease in mortality of challenged by a red seaweed (Defoirdt et al., 2007). brine shrimp larvae, and biofilm formation and exopolysac- In addition to the ecological significance of cross- charide production in the pathogen were also decreased to communication between aquatic bacteria and (micro)al- the same extent as was observed for indole at the same gae, these observations also offer an explanation for the concentrations. However, despite these similarities, there beneficial effect that is associated with microalgae in aqua- were also differences: in contrast to indole, auxins had no culture. Indeed, microalgae are an important constituent of effect on motility of V. campbellii and on bioluminescence of the so-called greenwater aquaculture systems, in which the hfq deletion mutant. The quorum sensing signal trans- the animals are cultured in water containing high levels of duction cascade of V. campbellii is highly complex, microalgae (Natrah et al., 2014). Empirical evidence has including several feedback loops (Fig. 4). Therefore, it is not indicated that there is a lower incidence of disease in these

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 1998 Q. Yang et al. kinds of rearing systems (De Schryver et al., 2014), which Belgium) were hydrated in 18 ml of filter-sterilized tap water might be explained by the presence of (relatively) high lev- for 1 h. Sterile cysts and larvae were obtained by decapsula- els of indole analogues in the digestive tract of animals tion according to Marques et al. (2005). Briefly, 660 mlof NaOH (32%) and 10 ml of NaOCl (50%) were added to the that have ingested large amounts of microalgae. Hence, hydrated cyst suspension to facilitate decapsulation. The pro- the selection and application of microalgae (or alga- cess was stopped after 2 min by adding 14 ml of Na2S2O3 bacteria consortia) that are capable of producing high lev- (10 g l21). Filtered (0.22 mm) aeration was provided during the els of auxins could be an interesting and novel strategy to reaction. The decapsulated cysts were washed with filtered control bacterial disease in aquaculture that is based on (0.22 mm) and autoclaved artificial seawater (containing 35 g 21 ecological interactions between aquatic (micro)organisms. l of instant ocean synthetic sea salt, Aquarium Systems, However, before any therapy that is based on the interfer- Sarrebourg, France). The cysts were re-suspended in a 50 ml tube containing 30 ml of filtered, autoclaved seawater and ence with indole signalling can be applied in practice, hatched for 28 h on a rotor (4 min21)at288Cwithconstantillu- further research will be needed in order to select the best mination (2000 lux). The axenity of cysts was verified by agent, that is, a micro-organism, a microbial metabolite inoculating one ml of culture water into 9 ml of marine broth (e.g., auxin) or a synthetic indole analogue. Further, as is and incubating at 288C for 24 h. After 28 h of hatching, the case with any other novel therapy, potential negative batches of 30 larvae were counted and transferred to fresh, side effects (including toxicity to the host, negative impacts sterile 50 ml tubes containing 30 ml of filtered and autoclaved on non-target micro-organisms, and resistance develop- seawater. Finally, the tubes were returned to the rotor and kept at 288C. All manipulations were performed in a laminar ment) should be carefully considered. flow to maintain sterility of the cysts and larvae. Experimental procedures Brine shrimp challenge tests Indoles The impacts of indole and indole analogues on the virulence Indole, indole-3-acetic acid and indole-3-acetamide were used of V. campbellii were determined in a standardized challenge in this study (Supporting Information Fig. S1). The compounds test with gnotobiotic brine shrimp larvae. Unless mentioned were dissolved in dimethyl sulfoxide (DMSO) at 10 mM. All the otherwise, V. campbellii was incubated with or without indole chemicals were purchased from Sigma-Aldrich (Bornem, Bel- or auxins at 50, 100 and 200 lM respectively, and cultures gium). Compounds were further diluted in DMSO, and in all were washed with phosphate-buffered saline (pH 7.4) prior to experiments, all treatments received the same volumes of inoculation into the brine shrimp rearing water at 105 CFU DMSO. ml21. The challenge tests were performed as described by Defoirdt et al. (2005) with some modifications. A suspension Bacterial strains and growth conditions of autoclaved LVS3 bacteria (Verschuere et al., 1999) in filtered and autoclaved seawater was added as feed at the start Five V. campbellii strains were used in this study; wild type of the challenge test at 107 cells ml21. Brine shrimp cultures, strain ATCC BAA-1116 and four strains that were derived from to which only autoclaved LVS3 bacteria were added as feed, this wild type: JAF483, BNL258, JAF548 pAKlux1 and were used as controls. The survival of the larvae was counted DrpoS1DrpoS2. Strains JAF483 (LuxO D47A) and BNL258 48 h after the addition of the pathogens. Each treatment was (Dhfq) contain mutations in the quorum sensing system ren- carried out in triplicate and each experiment was repeated dering it locked in the high cell density configuration (Freeman twice to verify the reproducibility. In each test, the sterility of and Bassler, 1999; Lenz et al., 2004). Strain JAF548 (LuxO the control treatments were checked at the end of the chal- D47E) contains a point mutation in luxO which renders the lenge by inoculating 1 ml of rearing water to 9 ml of marine quorum sensing system inactive. Plasmid pAKlux1 containing broth and incubating the mixture for 2 days at 288C. the Photorhabdus luminescens bioluminescence operon under the control of a constitutive promoter, was conjugated into this strain (Defoirdt et al., 2012). Hence, strain JAF548 pAKlux1 Giant freshwater prawn challenge test produces bioluminescence that is independent of the quorum Giant freshwater prawn experiments were performed as sensing system. DrpoS1DrpoS2 is a double deletion mutant in described in Pande et al. (2013). Briefly, prawn broodstock which both copies of the stress sigma factor RpoS are inacti- maintenance was performed according to Cavalli et al.(2001) vated (Wang et al., 2012). Unless otherwise stated, all strains and water quality parameters were adjusted according to New 21 were cultured in Luria-Bertani medium containing 35 g l of (2003). The larvae were obtained from a single oviparous sodium chloride (LB35)at288C under constant agitation (100 female breeder. A matured female, which had just completed 21 min ). Cell densities were measured spectrophotometrically its pre-mating moult, was mated with a hard-shelled male as at 600 nm. described before (Baruah et al., 2009). The female with fertil- ized eggs was then maintained for 20–25 days to undergo Axenic hatching of brine shrimp larvae embryonic development. When fully ripe (indicated by dark grey colour of the eggs), the female was transferred to a Two hundred milligrams of high-quality hatching cysts of Arte- hatching tank (30 l) containing slightly brackish water (contain- 21 mia franciscana (EGVR Type; INVE Aquaculture, Baasrode, ing6gl Instant Ocean synthetic sea salt, Aquarium System

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 Indole and auxins control the virulence of vibrios 1999

Inc., Sarrebourg, France). The water temperature was main- Swimming motility assay tained at 288C by a thermostat heater. After hatching, the The swimming motility assay was performed on soft agar newly hatched larvae with yolk were left for 24 h in the hatch- (LB plates containing 0.2% agar) as described previously ing tank. The next day, prawn larvae with absorbed yolk were 35 (Yang and Defoirdt, 2015). Indole was added to the agar distributed in groups of 25 larvae in 200 ml glass cones con- after autoclaving and immediately before pouring the agar taining 100 ml fresh autoclaved brackish water (12 g l21 into petri plates. V. campbellii was grown overnight in LB synthetic sea salts). The glass cones were placed in a rectan- 35 broth, and 5 ml aliquots (OD 5 1.0) were inoculated into gular tank containing water maintained at 288C and were 600 the center of the soft agar plates. Plates were incubated for provided with aeration. The larvae were fed daily with 5 Arte- 24 h, after which the diameters of the motility halos were mia nauplii/larva and acclimatized to the experimental measured. All assays were done with freshly prepared conditions for 24 h. During the experiments, water quality media in six replicates. parameters were kept at minimum 5 mg l21 dissolved oxygen, maximum 0.5 mg l21 ammonium-N and maximum 0.05 mg l21 nitrite-N. Determination of biofilm and exopolysaccharide levels Prawn larvae were challenged with wild type V. campbellii 6 or pre-treated V. campbellii by adding the strains at 10 CFU Biofilm formation was quantified by crystal violet staining, 21 ml to the culture water on the day after first feeding. Survival as described previously (Stepanovic´ et al., 2007). In brief, was counted daily in the treatment challenged to wild type V. an overnight culture of V. campbellii was diluted to an OD600 campbellii and the challenge test was stopped when more of 0.1 in LB35 broth with or without indole and indole ana- than 50% mortality was achieved in this treatment (in order to logues, and 200 ml aliquots of these suspensions were have enough larvae remaining for the growth measurement). pipetted into the wells of a polystyrene 96 well plate. Then At this time point, larval survival was determined in all treat- the bacteria were allowed to adhere and grow without agita- ments by considering that only those larvae presenting tion for 24 h at 288C. After that, the cultures were removed movement of appendages were alive. and the wells were washed three times with 300 llsterile The growth parameter larval stage index (LSI) was deter- physiological saline to remove all non-adherent bacteria. mined according to Maddox and Manzi (1976) on five The remaining attached bacteria were fixed with 150 llof randomly sampled larvae from each cone and calculated as: 99% methanol per well for 20 min, after which the methanol was removed and plates were air-dried. Then, biofilms were LSI 5 R Si =N stained for 15 min with 150 ll per well of a 1% crystal violet solution (Pro-lab Diagnostics, Richmond Hill, ON, Canada). where, Si is the stage of the larva (i 5 1–12); N is the number Excess stain was rinsed off by placing the plate under run- of larvae examined. ning tap water, and washing was continued until the washings were free of the stain. After the plates were air dried, the dye bound to the adherent cells was resolubilized Bacterial growth assays with 150 ll of 95% ethanol per well, and absorbance was For the bacterial growth assays, V. campbellii was grown over- measured at 570 nm. Sterile medium served as negative control, and the reported values are blank-corrected (OD night in LB35 broth at 288C. After that, the culture was diluted 570 of the blank was 0.132 6 0.002). to an OD600 of 0.1 in fresh LB35 broth, without indole and with indole (at 50, 100 and 200 lM respectively). The cultures For the quantification of exopolysaccharides, Calcofluor were grown in 200 ml volumes in 96-well plates at 288Cfor white staining (Sigma-Aldrich) was used as described previ- 24 h, and the turbidity at 600 nm was monitored every hour ously (Natrah et al., 2011). In brief, wells were rinsed after using a Multireader machine (Infinite M200, TECAN, Austria). 24 h biofilm formation and 100 ll phosphate buffered saline Growth curves were determined for three independent containing 0.5 ll 5 mM Calcofluor white staining dye was cultures. added to the wells. After 60 min, fluorescence (excitation 405 nm and emission 500 nm) was measured with a Multi- reader (Infinite M200, TECAN, Austria). Sterile medium Quantification of indole served as negative control, and the reported fluorescence values are blank-corrected (fluorescence of the blank was V. campbellii cultures were grown for 24h in LB35 medium 2662 6 64). (reaching OD600 of 1) and re-inoculated (10% v/v) into fresh LB35 medium. Cells were harvested at different time points and centrifuged at 8000 x g for 5 min. The concentration of Bioluminescence assays indole in the supernatants was measured as described previously (Li et al., 2014) by mixing 500 ml of supernatant with 500 Bioluminescence assays were performed as described previ- ml of Kovac’s reagent. After vortexing, the top 200 mlwere ously (Defoirdt et al., 2012). V. campbellii strains were grown removed and the OD571 was measured. The indole concentra- overnight and diluted to an OD600 of 0.1. Indole and analogues tion in each sample was determined based on a standard were added at 50, 100 and 200 lM respectively. The cultures curve using synthetic indole (Sigma-Aldrich). At least three dif- were further incubated at 288C with shaking, and biolumines- ferent V. campbellii cultures were sampled for each strain at cence was measured after 1 h with a Tecan Infinite 200 each time point. microplate reader (Tecan, Mechelen, Belgium).

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 2000 Q. Yang et al.

Specific quorum sensing-inhibitory activity AQSI reaction mixture was incubated for 5 min at 258C followed by 60 min at 428C. The reaction was terminated by heating at The specific quorum sensing-inhibitory activity of the com- 708C for 5 min and then cooled to 48C. cDNA samples were pounds at a given concentration was calculated as described checked by PCR and stored at 2208C for further use. previously (Yang et al., 2015):

%InhibitionQS2regulated Real-time PCR A 5 (1) QSI %Inhibition QS2independent Real-time PCR was used to quantify the expression level of all with the genes and was performed with the MaximaVR SYBR Green/ROX qPCR Master Mix (Fermentas, Fisher Scientific, Erembodegem, Belgium) as described previously (Yang and % Inhibition : percentage inhibition of QS- QS-regulated Defoirdt, 2015). The reaction was performed in a StepOneTM regulated bioluminescence in wild type V. campbellii Real-Time PCR System thermal cycler (Applied Biosystems, % Inhibition : percentage inhibition of QS- QS-independent Gent, Belgium) in a total volume of 25 ml, containing 12.5 mlof independent bioluminescence of V. campbellii JAF548 23 SYBR green master mix, 300 nM of forward and reverse pAKlux1. primers and 2 ml of template cDNA. The thermal cycling con- sisted of an initial denaturation at 958C for 10 min followed by For inhibition values below 1%, a value of 1% is used in the 40 cycles of denaturation at 958C for 15 s and primer anneal- > calculations, and the AQSI values are reported as “calculated ing and elongation at 608C for 1 min. Dissociation curve A .” Compounds are considered as false positives if A is QSI QSI analysis was performed to check for the amplification of untar- lower than 2 at all concentrations tested, whereas they are geted fragments. Data acquisition was performed with the considered as specific quorum sensing inhibitors if they cause TM StepOne Software. a significant inhibition of quorum sensing-regulated bioluminescence and if AQSI is higher than 10 at one of the concentrations tested (Yang et al., 2015; Defoirdt, 2016). Real-time PCR data analysis (22DDCT method) Real-time PCR data were analyzed as described previously RNA extraction (Yang and Defoirdt, 2015). The real-time PCR was validated by amplifying serial dilutions of cDNA synthesized from 1 mgof The mRNA was extracted as described previously (Yang and RNA isolated from bacterial samples. Serial dilutions of cDNA Defoirdt, 2015). V. campbellii strains were grown in triplicate in were amplified using gene specific primers. DC (average C LB broth with or without 100 lMindoleat288C for 24 h. T T 35 value of target-average C value of rpoA) was calculated for RNA was extracted with the SV Total RNA Isolation System T the different dilutions and plotted against the cDNA concentra- (Promega, Leiden, The Netherlands) according to the manu- tion. The slope of the graph was almost equal to 0 for all of the facturer’s instructions. The RNA quantity was measured target nine genes. Therefore, the amplification efficiency of ref- spectrophotometrically (NanoDrop Technologies, Wilmington, erence and the target genes was considered to be equal. DE, USA) and adjusted to 200 ng ml21 in all samples. The Based on this precondition, real-time PCR data were analyzed RNA integrity was checked by Agarose Gel Electrophoresis using the 22DDCT method (Livak and Schmittgen, 2001). The and the RNA samples were stored at 2808C. expression of the target genes was normalized to the endoge-

nous control (rpoA) by calculating DCT: Primers DCt 5 Ct target — Ct rpoA The primers used for RT-qPCR analysis are listed in Support- ing Information Table S5. Specific primers were designed and expressed relative to a calibrator strain by calculating using the software PRIMER PREMIER version 5.00 (Premier DDCt: Biosoft International, Palo Alto, CA USA), with predicted prod- uct sizes in the 100–200 bp range. The RNA polymerase A DDCt 5 DCt — Ct calibratior subunit (rpoA) mRNA was used as an endogenous control (Defoirdt et al., 2007). Unless otherwise indicated, untreated V. campbellii ATCC BAA-1116 was used as a calibrator, and all other treatments were normalized accordingly. The relative expression was Reverse transcription then calculated as Reverse transcription was performed as described previously 2DDCt with the RevertAidTM H minus First strand cDNA synthesis kit Relative expression 5 2 (Fermentas Gmbh, Baden-Wu€rttemberg, Germany) according to the manufacturer’s instructions (Yang and Defoirdt, 2015). Briefly, a mixture of 1 mgRNAand1ml random hexamer prim- Microarray-based transcriptomics er solution was mixed first. Then, 8 ml of reaction mixture containing 4 mlof53 reaction buffer, 2 ml of the dNTP mix, 20 The quality and quantity of RNA samples were determined units of ribonuclease inhibitor, 200 units of RevertAidTM H using RNA 6000 Nano kit with 2100 Bioanalyzer (Agilent Tech- minus M-MuLV Reverse Transcriptase was added. The nologies, Inc., Santa Clara, CA) and NanoDrop ND2000

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 Indole and auxins control the virulence of vibrios 2001

(Thermo Scientific Inc., Waltham, MA) spectrophotometer. Acknowledgements The RNA samples were accepted for mRNA enrichment if the The authors thank Bonnie Bassler for providing us with the V. A260/A280 ratios were between 1.9 and 2.1, the ratio of intensities of the 23S and 16S rRNA were 1.5–2.5:1 and the RNA campbellii wild type and quorum sensing mutants. This work integrity number was between 7 and 10. Bacterial mRNA was was funded by the China Scholarship Council, the Special isolated from 4.0 to 8.9 mg of total RNA using the MICROBE- Research Fund of Ghent University (BOF-UGent), the Scien- xpressTM kit (Ambion) according to the manufacturer’s tific Research Fund of Flanders (FWO), and the Office of protocol. The quantity of mRNA samples were determined by Naval Research via US Naval Research Laboratory core NanoDrop ND2000 (Thermo Scientific Inc., Waltham, MA) funds. spectrophotometer. RNA labelling were carried out in 30 ml reaction volume con- Authors’ contributions taining 1 mg of mRNA and diluted polyA RNA control TD designed the project. QY, GJV and TD designed the (Affymetrix) using KREATECH’s aRNA Labelling Kit (KREA- experiments. QY, GSJP, ZW, BL and GJV performed TECH Biotechnology, Amsterdam, the Netherlands) according to the manufacture’s recommended protocol. The labelled experiments. RAR, ZW, GJV and TD performed the in sil- mRNA was hybridized to a custom Affymetrix microarray ico analyses. QY, RAR and TD performed statistics. QY (520694F) according to standard protocols (Affymetrix, Santa and TD wrote the article. All authors approved the final Clara, CA, USA). All hybridizations incubated for 16 h at 498C manuscript. in the GeneChipVR Hybridization Oven 640 at 60 rpm and the microarrays were then washed and stained with the Gen- References eChipVR Fluidics Station 450 and scanned using the GeneChipVR Scanner 7G (Affymetrix). Hybridization signal Amin, S.A., Hmelo, L.R., van Tol, H.M., Durham, B.P., intensities were analyzed with the GeneChipVR Operating Soft- Carlson, L.T., Heal, K.R., et al. (2015) Interaction and sig- ware (GCOS) to generate raw image files (.DAT) and nalling between a cosmopolitan phytoplankton and associ- simplified image files (.CEL) with intensities assigned to each ated bacteria. Nature 522: 98–101. of the corresponding probe positions. In all, data from 10 Anetzberger, C., Pirch, T., and Jung, K. (2009) Heterogeneity microarrays, comprising five control replicates and five indole- in quorum sensing-regulated bioluminescence of Vibrio har- treated replicates, were obtained. veyi. Mol Microbiol 73: 267–277. Differential expression between control and indole-treated Austin, B., and Zhang, X.H. (2006) Vibrio harveyi: a significant genes was determined in the following manner. First, only pathogen of marine vertebrates and invertebrates. Lett Appl those probe sets containing 11 probes (comprising 5951 of Microbiol 43: 119–124. the 12,717 probe sets on the custom microarray) were includ- Bansal, T., Englert, D., Lee, J., Hegde, M., Wood, T.K., and ed in the analysis. Those probe amplitudes were then log- Jayaraman, A. (2007) Differential effects of epinephrine, transformed and the results were normalized within-chip by norepinephrine, and indole on Escherichia coli O157: H7 conversion to standard scores (z-scores). Probe-level differen- chemotaxis, colonization, and gene expression. Infect tial expression analysis was then carried out in accordance Immun 75: 4597–4607. with Rubin (2009). More specifically, for each probe set, analy- Bansal, T., Alaniz, R.C., Wood, T.K., and Jayaraman, A. sis of variance (ANOVA) was performed at each probe (2010) The bacterial signal indole increases epithelial-cell position, and the resulting “signed P-value” was retained. The tight-junction resistance and attenuates indicators of inflam- median of the 11 signed P-values was then used as the mea- mation. Proc Natl Acad Sci USA 107: 228–233. sure for determining if the gene was differentially expressed. Baruah, K., Cam, D.T.V., Dierckens, K., Wille, M., Defoirdt, T., Sorgeloos, P., and Bossier, P. (2009) In vivo effects of single Under the null hypothesis that the gene is not differentially or combined N-acyl homoserine lactone quorum sensing expressed, signed ANOVA P-values are uniformly distributed signals on the performance of Macrobrachium rosenbergii on the interval [21, 1] and their median has a beta distribution larvae. Aquaculture 288: 233–238. (with defining parameters a 5 b 5 6 in the case of 11 probe Cao, J.G., and Meighen, E.A. (1989) Purification and structur- genes). Thus we were able to determine a parametric differen- al identification of an autoinducer for the luminescence sys- tial expression P-value for each gene and to adjust those tem of Vibrio harveyi. JBiolChem264: 21670–21676. gene-level P-values for the multiple hypothesis testing condi- Cavalli, R.O., Lavens, P., and Sorgeloos, P. (2001) Reproduc- tion using standard false discovery rate methods. The gene tive performance of Macrobrachium rosenbergii female in designations and annotations utilized are from the Naval captivity. J World Aquacult Soc 32: 60–67. Research Laboratory’s V. campbellii ATCC BAA-1116 genome Chant, E.L., and Summers, D.K. (2007) Indole signalling con- sequencing effort (GenBank accession numbers CP006605.1, tributes to the stable maintenance of Escherichia coli multi- CP006606.1 and CP006607.1) and the expression profiling copy plasmids. Mol Microbiol 63: 35–43. data can be found in the GenBank Gene Expression Omnibus Chen, X., Schauder, S., Potier, N., Van Dorsselaer, A., repository (accession number GSE80484). Pelczer,I.,Bassler,B.L.,et al. (2002) Structural identification of a bacterial quorum-sensing signal containing boron. Statistical analyses Nature 415: 545–549. Chu, W., Zere, T.R., Weber, M.M., Wood, T.K., Whiteley, M., Data analysis was carried out using the SPSS statistical soft- Hidalgo-Romano, B., et al. (2012) Indole production pro- ware (version 15). motes Escherichia coli mixed-culture growth with

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 2002 Q. Yang et al.

Pseudomonas aeruginosa by inhibiting quorum signaling. Karunasagar, I., Pai, R., Malathi, G.R., and Karunasagar, I. Appl Environ Microbiol 78: 411–419. (1994) Mass mortality of Penaeus monodon larvae due to De Schryver, P., Defoirdt, T., and Sorgeloos, P. (2014) Early antibiotic-resistant Vibrio harveyi infection. Aquaculture mortality syndrome outbreaks: a microbial management 128: 203–209. issue in shrimp farming?. PLoS Pathog 10: e1003919. Kim, J., and Park, W. (2013) Indole inhibits bacterial quorum Defoirdt, T. (2016) Specific antivirulence activity, a new con- sensing signal transmission by interfering with quorum cept for reliable screening of virulence inhibitors. Trends sensing regulator folding. Microbiology 159: 2616–2625. Microbiol 34: 527–529. Lee, J.H., and Lee, J. (2010) Indole as an intercellular signal in Defoirdt, T., and Sorgeloos, P. (2012) Monitoring of Vibrio har- microbial communities. FEMS Microbiol Rev 34: 426–444. veyi quorum sensing activity in real time during infection of Lee, J., Jayaraman, A., and Wood, T.K. (2007) Indole is an brine shrimp larvae. Isme J 6: 2314–2319. inter-species biofilm signal mediated by SdiA. BMC Micro- Defoirdt, T., Bossier, P., Sorgeloos, P., and Verstraete, W. biol 7: 42. (2005) The impact of mutations in the quorum sensing sys- Lee, J., Zhang, X.S., Hegde, M., Bentley, W.E., Jayaraman, tems of Aeromonas hydrophila, Vibrio anguillarum and Vib- A., and Wood, T.K. (2008) Indole cell signaling occurs pri- rio harveyi on their virulence toward gnotobiotically cultured marily at low temperatures in Escherichia coli. Isme J 2: Artemia franciscana. Environ Microbiol 7: 1239–1247. 1007–1023. Defoirdt, T., Miyamoto, C.M., Wood, T.K., Meighen, E.A., Lee, J., Attila, C., Cirillo, S.L.G., Cirillo, J.D., and Wood, T.K. Sorgeloos, P., Verstraete, W., and Bossier, P. (2007) The (2009a) Indole and 7-hydroxyindole diminish Pseudomonas natural furanone (5Z)24-bromo-5-(bromomethylene)23- aeruginosa virulence. Microb Biotechnol 2: 75–90. butyl-2(5H)-furanone disrupts quorum sensing-regulated Lee, J., Maeda, T., Hong, S.H., and Wood, T.K. (2009b) gene expression in Vibrio harveyi by decreasing the DNA- Reconfiguring the quorum-sensing regulator SdiA of binding activity of the transcriptional regulator protein LuxR. Escherichia coli to control biofilm formation via indole and Environ Microbiol 9: 2486–2495. N-acylhomoserine lactones. Appl Environ Microbiol 75: Defoirdt, T., Boon, N., Sorgeloos, P., Verstraete, W., and 1703–1716. Bossier, P. (2008) Quorum sensing and quorum quenching Lee, J.H., Cho, M.H., and Lee, J. (2011) 3-Indolylacetonitrile in Vibrio harveyi: lessons learned from in vivo work. Isme J decreases Escherichia coli O157:H7 biofilm formation and 2: 19–26. Pseudomonas aeruginosa virulence. Environ Microbiol 13: Defoirdt, T., Benneche, T., Brackman, G., Coenye, T., 62–73. Sorgeloos, P., and Scheie, A.A. (2012) A quorum sensing- Lee, J.H., Kim, Y.G., Baek, K.H., Cho, M.H., and Lee, J. disrupting brominated thiophenone with a promising thera- (2015a) The multifaceted roles of the interspecies signalling peutic potential to treat luminescent vibriosis. PLoS ONE 7: molecule indole in Agrobacterium tumefaciens. Environ e41788. Microbiol 17: 1234–1244. Dong, X., Wang, H., Xie, G., Zou, P., Guo, C., Liang, Y., and Lee, J.H., Wood, T.K., and Lee, J. (2015b) Roles of indole as Huang, Y. (2016) An isolate of Vibrio campbellii carrying the an interspecies and interkingdom signaling molecule. pirVP gene causes acute hepatopancreatic necrosis dis- Trends Microbiol 23: 707–718. ease. Emer Microb Infect 6: e2. Lelong, C., Aguiluz, K., Luche, S., Kuhn, L., Garin, J., Freeman, J.A., and Bassler, B.L. (1999) A genetic analysis of Rabilloud, T., et al. (2007) The Crl-RpoS regulon of Escheri- the function of LuxO, a two-component response regulator chia coli. Mol Cell Proteom 6: 648–659. involved in quorum sensing in Vibrio harveyi. Mol Microbiol Lenz, D.H., Mok, K.C., Lilley, B.N., Kulkarni, R.V., Wingreen, 31: 665–677. N.S., and Bassler, B.L. (2004) The small RNA chaperone Han, Y., Yang, C.-L., Yang, Q., Qi, Z., Liu, W., Xu, Z.-H., et al. Hfq and multiple small RNAs control quorum sensing in Vib- (2011) Mutation of tryptophanase gene tnaA in Edward- rio harveyi and Vibrio cholerae. Cell 118: 69–82. siella tarda reduces lipopolysaccharide production, antibiot- Li, X., Yang, Q., Dierckens, K., Milton, D.L., and Defoirdt, T. ic resistance and virulence. Environ Microbiol Rep 3: 603– (2014) RpoS and indole signaling control the virulence of 612. Vibrio anguillarum towards gnotobiotic sea bass (Dicen- Hidalgo-Romano, B., Gollihar, J., Brown, S.A., Whiteley, M., trarchus labrax)larvae.PLoS ONE 9: e111801. Valenzuela, E., Jr., Kaplan, H.B., et al. (2014) Indole inhibi- Lin, B., Wang, Z., Malanoski, A.P., O’grady, E.A., Wimpee, tion of N-acylated homoserine lactone-mediated quorum C.F., Vuddhakul, V., et al. (2010) Comparative genomic signalling is widespread in Gram-negative bacteria. Microbi- analyses identify the Vibrio harveyi genome sequenced ology 160: 2464–2473. strains BAA-1116 and HY01 as Vibrio campbellii. Environ Higgins, D.A., Pomianek, M.E., Kraml, C.M., Taylor, R.K., Microbiol Rep 2: 81–89. Semmelhack, M.F., and Bassler, B.L. (2007) The major Vib- Livak, K.J., and Schmittgen, T.D. (2001) Analysis of relative rio cholerae autoinducer and its role in virulence factor pro- gene expression data using real-time quantitative PCR and duction. Nature 450: 883–886. the 22DDCT method. Methods 25: 402–408. Hirakawa, H., Inazumi, Y., Masaki, T., Hirata, T., and Lu, Y., and Xu, J. (2015) Phytohormones in microalgae: a new Yamaguchi, A. (2005) Indole induces the expression of mul- opportunity for microalgal biotechnology?. Trends Plant Sci tidrug exporter genes in Escherichia coli. Mol Microbiol 55: 20: 273–282. 1113–1126. Maddox, M.B., and Manzi, J.J. (1976) The effects of algal Jonsson, P.R., Pavia, H., and Toth, G. (2009) Formation of supplements on static system culture on Macrobrachium harmful algal blooms cannot be explained by allelopathic rosenbergii (de Man) larvae. Proc World Maticult Soc 7: interactions. Proc Natl Acad Sci USA 106: 1177–1182. 677–698.

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 Indole and auxins control the virulence of vibrios 2003

Main, C.R., Salvitti, L.R., Whereat, E.B., and Coyne, K.J. Pathogenesis, virulence factors and virulence regulation of (2015) Community-level and species-specific associations vibrios belonging to the Harveyi clade. Rev Aquacult 4: 59– between phytoplankton and particle-associated Vibrio spe- 74. cies in Delaware’s island bays. Appl Environ Microbiol 81: Spaepen, S., and Vanderleyden, J. (2011) Auxin and Plant- 5703–5713. Microbe Interactions. Cold Spring Harb Perspect Biol 3: Manset, K.J.V., Azanza, R.V., and Onda, D.F.L. (2013) Algicid- a001438. al bacteria from fish culture areas in Bolinao, Pangasian, Stamm, I., Lottspeich, F., and Plaga, W. (2005) The pyruvate Northern Philippines. JEnvironSciManagSI1: 11–20. kinase of Stigmatella aurantiaca is an indole binding Marques, A., Dinh, T., Ioakeimidis, C., Huys, G., Swings, J., protein and essential for development. Mol Microbiol 56: Verstraete, W., et al. (2005) Effects of bacteria on Artemia 1386–1395. franciscana cultured in different gnotobiotic environments. Stepanovic´, S., Vukovic´, D., Hola, V., Bonaventura, G.D., Appl Environ Microbiol 71: 4307–4317. Djukic´, S., C´ irkovic´,I.,et al. (2007) Quantification of biofilm Mayali, X., and Azam, F. (2004) Algicidal bacteria in the sea in microtiter plates: overview of testing conditions and prac- and their impact on algal blooms. JEukaryotMicrobiol51: tical recommendations for assessment of biofilm production 139–144. by staphylococci. Apmis 115: 891–899. McCarter, L.L. (2004) Dual flagellar systems enable motility Stirk, W.A., Ord€ og,€ V., Novak, O., Rolcˇık, J., Strnad, M., under different circumstances. J Mol Microbiol Biotechnol 7: Balint, P., et al. (2013) Auxin and cytokinin relationships in 18–29. 24 microalgal strains. JPhycol49: 459–467. Melander, R.J., Minvielle, M.J., and Melander, C. (2014) Con- Vega, N.M., Allison, K.R., Khalil, A.S., and Collins, J.J. (2012) trolling bacterial behavior with indole-containing natural Signaling-mediated bacterial persister formation. Nat Chem products and derivatives. Tetrahedron 70: 6363–6372. Biol 8: 431–433. Milton, D.L. (2006) Quorum sensing in vibrios: complexity for Verschuere, L., Rombaut, G., Huys, G., Dhont, J., Sorgeloos, diversification. Int J Med Microbiol 296: 61–71. P., and Verstraete, W. (1999) Microbial control of the culture Miyashiro, T., and Ruby, E.G. (2012) Shedding light on biolu- of Artemia juveniles through preemptive colonization minescence regulation in Vibrio fischeri. Mol Microbiol 84: by selected bacterial strains. Appl Environ Microbiol 65: 795–806. 2527–2533. Mueller, R.S., Beyhan, S., Saini, S.G., Yildiz, F.H., and Wang, Z., O’shaughnessy, T.J., Soto, C.M., Rahbar, A.M., Bartlett, D.H. (2009) Indole acts as an extracellular cue reg- Robertson, K.L., Lebedev, N., et al. (2012) Function and ulating gene expression in Vibrio cholerae. J Bacteriol 191: regulation of Vibrio campbellii proteorhodopsin: acquired 3504–3516. phototrophy in a classical organoheterotroph. PLoS ONE 7: Natrah, F.M.I., Ruwandeepika, H.A.D., Pawar, S., e38749. Karunasagar, I., Sorgeloos, P., Bossier, P., et al.(2011) Yang, Q., and Defoirdt, T. (2015) Quorum sensing positively Regulation of virulence factors by quorum sensing in Vibrio regulates flagellar motility in pathogenic Vibrio harveyi. harveyi. Vet Microbiol 154: 124–129. Environ Microbiol 17: 960–968. Natrah, F.M.I., Bossier, P., Sorgeloos, P., Yusoff, F.M., and Yang, Q., Scheie, A.A., Benneche, T., and Defoirdt, T. Defoirdt, T. (2014) Significance of microalgal–bacterial inter- (2015) Specific quorum sensing-disrupting activity (AQSI) actions for aquaculture. Rev Aquacult 6: 48–61. of thiophenones and their therapeutic potential. Sci Rep Nealson, K.H., and Hastings, J.W. (2006) Quorum sensing on 5: 18033. aglobalscale:massivenumbersofbioluminescentbacteria Yildiz, F.H., Dolganov, N.A., and Schoolnik, G.K. (2001) VpsR, in milky seas. Appl Environ Microbiol 72: 2295–2297. a member of the response regulators of the two-component New, M.B. (2003) Farming freshwater prawns: a manual for regulatory systems, is required for expression of vps bio- the culture of the giant river prawn, Macrobrachium rose- synthesis genes and EPSETr-associated phenotypes in nbergii. FAO Fisheries Technical Paper No. 428. Food and Vibrio cholerae O1 El Tor. J Bacteriol 183: 1716–1726. Agriculture Organisation, Rome, Italy. Zhou, J., Lyu, Y., Richlen, M.L., Anderson, D.M., and Cai, Z. Nikaido, E., Giraud, E., Baucheron, S., and Yamasaki, S. (2016) Quorum sensing is a language of chemical signals (2012) Effects of indole on drug resistance and virulence of and plays an ecological role in algal-bacterial interactions. Salmonella enterica serovar Typhimurium revealed by Crit Rev Plant Sci 35: 81–105. genome-wide analyses. Gut Pathog 4: 5–18. Pande, G.S.J., Natrah, F.M.I., Sorgeloos, P., Bossier, P., and Defoirdt,T.(2013)TheVibrio campbellii quorum sensing Supporting information signals have a different impact on virulence of the bacterium Additional Supporting Information may be found in the towards different crustacean hosts. Vet Microbiol 167: online version of this article at the publisher’s web-site: 540–545. Petrova, O.E., and Sauer, K. (2012) Sticky situations: key Table S1. Survival of gnotobiotic brine shrimp larvae after 2 components that control bacterial surface attachment. days of challenge with V. campbellii ATCC BAA-1116 J Bacteriol 194: 2413–2425. (average 6 standard error of 4 shrimp cultures), with or Rubin, R.A. (2009) A first principles approach to differential without adding indole to the brine shrimp rearing water at expression in microarray data analysis. BMC Bioinform 10: the start of the experiment. 292. Table S2. Relative mRNA levels of Vibrio polysaccharide Ruwandeepika, H.A.D., Jayaweera, T.S.P., Bhowmick, P.P., production regulators, flagellar and chemotaxis genes, the Karunasagar, I., Bossier, P., and Defoirdt, T. (2012) stress sigma factors and the quorum sensing master

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004 2004 Q. Yang et al. regulator in V. campbellii after 24 h incubation in the Fig. S1. Structures of the compounds used in this study. absence and presence of 100 lM indole. Fig. S2. Density of wild type V. campbellii, either or not pre- Table S3. Vibrio campbellii genes differentially expressed in treated with indole prior to inoculation, in the brine shrimp the presence of 100 mM indole, as determined by microar- rearing water after 24 and 48 h of challenge. ray analyses. Fig. S3. Growth of wild type V. campbellii in LB35 medium, Table S4. Impact of the auxins indole-3-acetamide and indole- with and without indole. 3-acetic acid on motility of Vibrio campbellii on soft agar. Fig. S4. Schematic representation of the quorum sensing Table S5. Specific primers used for reverse transcriptase system of Vibrio campbellii. qPCR.

VC 2017 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 19, 1987–2004