Exploring the transcriptome of luxI− and ΔainS mutants and the impact of N-3-oxo-hexanoyl-L- and N-3-hydroxy-decanoyl-L-homoserine lactones on biofilm formation in Aliivibrio salmonicida

Background Bacterial communication through quorum sensing (QS) systems has been reported to be important in coordinating several traits such as biofilm formation. In Aliivibrio salmonicida two QS systems the LuxI/R and AinS/R, have been shown to be responsible for the production of eight acyl-homoserine lactones (AHLs) in a cell density dependent manner. We have previously demonstrated that inactivation of LitR, the master regulator of the QS system resulted in biofilm formation, similar to the biofilm formed by the AHL deficient mutant ΔainSluxI−. In this study, we aimed to investigate the global gene expression patterns of luxI and ainS autoinducer synthases mutants using transcriptomic profiling. In addition, we examined the influence of the different AHLs on biofilm formation. Results The transcriptome profiling of ΔainS and luxI− mutants allowed us to identify genes and gene clusters regulated by QS in A. salmonicida. Relative to the wild type, the ΔainS and luxI− mutants revealed 29 and 500 differentially expressed genes (DEGs), respectively. The functional analysis demonstrated that the most pronounced DEGs were involved in bacterial motility and chemotaxis, exopolysaccharide production, and surface structures related to adhesion. Inactivation of luxI, but not ainS genes resulted in wrinkled colony morphology. While inactivation of both genes (ΔainSluxI−) resulted in strains able to form wrinkled colonies and mushroom structured biofilm. Moreover, when the ΔainSluxI− mutant was supplemented with N-3-oxo-hexanoyl-L-homoserine lactone (3OC6-HSL) or N-3-hydroxy-decanoyl-L-homoserine lactone (3OHC10-HSL), the biofilm did not develop. We also show that LuxI is needed for motility and for repression of EPS production, where repression of EPS is likely operated through the RpoQ-sigma factor. Conclusion These findings imply that the LuxI and AinS autoinducer synthases play a critical role in the regulation of biofilm formation, EPS production, and motility.


INTRODUCTION
Quorum sensing (QS) is a widespread mechanism in bacteria that employs autoinducing chemical signals in response to cell density. A variety of classes of QS chemical signals have been identified in different bacteria. Gram-negative bacteria usually employ N-acyl homoserine lactones (AHLs) which contain a conserved homoserine lactone (HSL) ring and an amide (N)-linked acyl side chain. The acyl groups identified to date range from 4 to 18 carbons in length (Fuqua, Parsek & Greenberg, 2001;Swift et al., 2001;Whitehead et al., 2001). AHL-mediated QS was originally discovered in the marine bacterium Aliivibrio fischeri, which was found to regulate bioluminescence through the lux operon in a cell-density dependent manner (Nealson & Hastings, 1979). The bacterium A. fischeri controls luminescence via the QS systems LuxS/LuxPQ, LuxI/LuxR, and AinS/AinR, where LuxS, LuxI, and AinS are the AHL autoinducer synthases (Lupp & Ruby, 2004Lupp et al., 2003).
The marine bacterium Aliivibrio salmonicida, is known to cause cold-water vibriosis in Atlantic salmon (Salmo salar), rainbow trout (Oncorhynchus mykiss), and captive Atlantic cod (Gadus morhua) (Egidius et al., 1981(Egidius et al., , 1986Holm et al., 1985). The genome sequence of A. salmonicida revealed five QS systems of which three are similar to those of A. fischeri, the LuxS/PQ, LuxI/R, and AinS/R (Hjerde et al., 2008). In A. salmonicida the LuxS/PQ and AinS/R systems transduce the information from the autoinducers AI-2 and 3OHC10-HSL to the histidine phosphotransferase protein LuxU and finally to the response regulator LuxO. The level of phosphorylated LuxO depends on the autoinducer concentrations. The phosphorylated LuxO controls the expression of small regulatory RNAs Qrr that together with the RNA chaperon Hfq, destabilize the transcript of the master regulator LitR.
In addition to regulating bioluminescence, AHLs are also involved in several physiological processes in bacteria such as production of virulence factors, drug resistance, motility, and biofilm formation (Abisado et al., 2018;Whitehead et al., 2001). AHL-mediated QS is involved in all stages of biofilm formation from attachment to maturation and dispersal in a number of bacterial species (Emerenini et al., 2015;Fazli et al., 2014;Guvener & McCarter, 2003;Hmelo, 2017;Huber et al., 2001;Pratt & Kolter, 1998;Whitehead et al., 2001;Yildiz & Visick, 2009). In many Vibrio species development of biofilm and rugose colony morphology dependent on exopolysaccharide (EPS) production (Yildiz & Visick, 2009). In A. salmonicida the EPS is produced by an operon known as the symbiosis polysaccharide (syp) operon, which is regulated by LitR via RpoQ sigma factor (Hansen et al., 2014;Khider, Willassen & Hansen, 2018). Mutation in LitR of A. salmonicida resulted in strains with wrinkled colonies and three-dimensional biofilm formation (Bjelland et al., 2012). Similar to A. salmonicida, inactivation of HapR, the master regulator of Vibrio cholerae, resulted in a regulatory state mimicking low cell density (LCD) conditions, where the mutant produced more EPS compared to wild type (Zhu & Mekalanos, 2003). In contrast to the two species mentioned above, Vibrio parahaemolyticus and Vibrio vulnificus, form biofilm and opaque colonies at high cell density (HCD). The inactivation of the master QS regulators, OpaR and SmcR, resulted in translucent colonies with decreased EPS production (Lee et al., 2013;McCarter, 1998).
We have previously shown that AinS and LuxI in A. salmonicida are responsible for the production of eight AHLs that are involved in regulation of biofilm formation (Hansen et al., 2015). When both luxI and ainS synthases genes were inactivated, no AHL production was observed in A. salmonicida and the double mutant (DainSluxI -) produced a biofilm similar to the biofilm of DlitR mutant (Hansen et al., 2015). In the present work, we aimed to understand the complex regulation of biofilm formation, EPS production and motility using transcriptomic profiling on the DainS and luxImutants. At HCD, inactivation of luxI had a global effect on the transcriptome and resulted in nearly 500 differentially expressed genes (DEGs), whereas deletion of ainS only resulted in 29 DEGs under the same conditions. Genes involved in motility and EPS production were among the DEGs in the luxImutant, which may explain the observations that this mutant lacks flagella, is non-motile and produces rugose colonies. The DainS mutant showed DEGs associated with phosphorylation and was not involved in regulating colony rugosity. Exposing the DainSluxIdouble mutant to 3OC6-HSL (LuxI product) or 3OHC10-HSL (AinS product), resulted in restoration of wild type traits and no biofilm formation was observed indicating that both LuxI/R and AinS/R systems are important in the regulation of biofilm formation.

Bacterial strains and supplements
Bacterial strains used in this study are listed in Table 1. A. salmonicida LFI1238 strain and the A. salmonicida mutants were grown from a frozen glycerol stock on blood agar base no. 2 (Oxoid, Basingstoke, UK) with a total concentration of 5% blood and 2.5% NaCl (BA2.5) or in Luria Bertani broth (Difco; BD Diagnostics, Berkshire, UK) with a total concentration of 2.5% NaCl (LB2.5).
A seawater-based medium (SWT) used for all assays consists of five g/L of bacto peptone (BD, Biosciences, San Jose, CA, USA), three g/L of yeast extract (Sigma-Aldrich, Darmstadt, Germany), and 28 g/L of a synthetic sea salt (Instant Ocean, Aquarium Systems, Blacksburg, VA, USA).
The green fluorescence protein (GFP) constitutive plasmid pVSV102 and helper plasmid pEVS104 propagated in Escherichia coli, DH5alpir and CC118lpir, respectively.

Culture conditions
A. salmonicida strains were cultivated from single colonies in two ml (LB2.5) at 12 C, 220 rpm for 2 days (primary culture). The primary cultures were diluted 1:20 and grown at 12 C, 220 rpm for an additional day (secondary cultures). GFP tagged strains were selected on BA2.5 supplemented with 150 ml/ml kanamycin.
The E. coli strains were cultivated in LB or LA (Luria Agar) containing 1% NaCl (LB1 and LA1, respectively) and incubated at 37 C and 220 rpm.

Sample collection
The overnight secondary cultures of DainS, luxI -, and A. salmonicida LFI1238 wild type were diluted to OD 600 ¼ 0.05 (optical density measured at 600 nm) in a total volume of 70 ml SWT media supplemented with 2.5% sea salt. The cultures were grown further at 8 C and 220 rpm in 250 ml baffled flasks. A total of 10 ml of the grown cultures were collected at LCD OD 600 ¼ 0.30 (early logarithmic phase) and 2.5 ml was collected at HCD OD 600 ¼ 1.20 (late exponential phase). The collected samples were harvested by centrifugation at 13,000Âg for 2 min at 4 C (Heraeus 3XR; Thermo Scientific, Waltham, MA, USA) and preserved in RNAlater (Invitrogen, Carlsbad, CA, USA). The preserved cultures were stored at -80 C until RNA extraction. Total RNA isolation and rRNA depletion The total RNA was extracted from the cell pellets following the standard protocols provided by the manufacturer (Epicenter, Madison, WI, USA). Ribosomal rRNA was removed from the samples using Ribo-Zero rRNA Removal kit for bacteria (Illumina, San Diego, CA, USA) following the manufacturer's instructions. The quality of total RNA before and after depletion was determined using the Agilent 2100 Bioanalyzer with the RNA 600 Nano and RNA 600 Pico chips, respectively (Agilent Technologies, Santa Clara, CA, USA).

RNA sequencing and data analysis
The RNA-sequencing libraries were constructed using the TruSeq Stranded mRNA Library Prep Kit (Illumina, San Diego, CA, USA), and sequenced at the Norwegian Sequencing Center with the Illumina NextSeq 500 system using Mid-Output Kit v2 for a 75-cycle sequencing run. The sequencing quality of FASTQ files was assessed using FastQC (http://www. bioinformatics.babraham.ac.uk/projects/fastqc). Further analysis of the RNA-Seq data was performed using EDGE-pro v1.0.1 (Magoc, Wood & Salzberg, 2013) and DESeq2 (Love, Huber & Anders, 2014). EDGE-pro was used to align the reads to the A. salmonicida LFI1238 genome (Hjerde et al., 2008) and to estimate gene expression levels. Differences in gene expression between wild type and luxI -, and wild type and DainS mutants were determined using DESeq2. DESeq2 first estimates size factors for each gene, then estimate the dispersion by fitting this to a model using the negative binomial distribution. Finally, DESeq2 performs a statistical test to see whether there is evidence for the observed differential expression between wild type and mutant genes, which is reported as a p-value. Log2 fold changes of the genes were recalculated to x differential expression values (i.e., DainS/wt) and genes were defined as significantly DEGs based on a p-value 0.05 and fold change values of !2 and -2 equal to log 2 fold 1 and -1. tRNA and rRNA reads were filtered out before analysis.
PCA plots are automatically generated by DESeq2 and were used for quality control of the biological replicates. DESeq2 normalize differences in gene expression patterns to compute a distance matrix. The X-and Y-axes in the PCA plot correspond to a mathematical transformation of these distances so that data can be displayed in two dimensions (Love, Huber & Anders, 2014).
The RNA sequence data presented in this study have been deposited in the European Nucleotide Archive (www.ebi.ac.uk/ena) under study accession number PRJEB29457 for DainS and luxIand accession number PRJEB28385 for A. salmoncida wild type.
High-performance liquid chromatography tandem mass spectrometry assay

Preparation of bacterial supernatants for AHL measurements
Two biological replicates were used for all A. salmonicida strains. The overnight secondary cultures were diluted to an OD 600 ¼ 0.05 in a total volume of 60 ml SWT media supplemented with 2.5% sea salt. The cultures were grown further at 8 C and 220 rpm in 250 ml baffled flasks for 50 h. Cells from one ml were harvested from each culture using 13,000Âg (Heraeus Fresco 21; Thermo Scientific, Waltham, MA, USA) for 2 min at 4 C. The supernatants were acidified with 1M HCl before threefold ethyl acetate extraction as previously described (Purohit et al., 2013). The ethyl acetate phase was dried using a rotary vacuum centrifuge (CentriVap; Labconco, Kansas City, MO, USA) for 15 min at 40 C and then redissolved in 150 ml of 20% acetonitrile containing 0.1% formic acid and 775 nM of the internal standard 3OC12-HSL.

Detection of AHL profiles using a mix of HPLC-MS/MS and full scan HR-MS analysis
The detection of AHL was adapted from the methods described previously (Hansen et al., 2015). Briefly, the samples (20 ml) were injected into an Ascentis Express C18 5 cm Â 2.1 mm, 2.7 mm reverse phase column (Supelco, Sigma-Adrich, Darmstadt, Germany) using an Accela autosampler (Thermo Scientific, Waltham, MA, USA). The elution was performed using an Accela pump (Thermo Scientific, Waltham, MA, USA) with an acetonitrile gradient in 0.1% formic acid and consisted of 5% acetonitrile for 18 s, followed by a linear gradient up to 90% acetonitrile over 222 s, and finally 90% acetonitrile for 60 s. The column was re-equilibrated for 60 s with 5% acetonitrile in 0.1% formic acid before the next sample was injected. Flow rate was 500 ml/min for all steps.
The separated compounds were ionized in positive ion electrospray using the following settings: sheath gas flow rate 70, auxiliary gas flow rate 10, sweep gas flow rate 10, spray voltage +4.50 kV, capillary temperature 330 C, capillary voltage 37 V, and tube lens 80 V.
The ionized components where detected using an LTQ Orbitrap XL (Thermo Scientific, Waltham, MA, USA) run in either MS/MS low resolution mode or full scan HRMS mode. C4 AHLs are difficult to detect using full scan HR-MS analysis due to co-eluting isobaric compounds seen in some samples, so these components together with 3OC6 and 3OHC6 where measured using high-performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) using the LTQ part of the LTQ orbitrap XL. The rest of the compounds were measured using Full Scan HR-MS analysis. The C4's, 3OHC6, and 3OC6 elute early in the chromatogram and where measured in two segments each with three scan events. Segment 1 ran from 0 to 0.88 min, with the following scan events. The system was calibrated with a mixture of 15 AHLs including the internal standard 3OC12-HSL. The ion chromatograms were analyzed using the Xcalibur v. 2.1.0 software package (Thermo Scientific, Waltham, MA, USA). The mass window was set to eight parts per million. The limit of detection and the limit of quantification for the different AHLs were calculated as previously described (Purohit et al., 2013).

Construction of GFP tagged A. salmonicida strains
A. salmonicida mutants used in this study were constructed previously (Bjelland et al., 2012;Hansen et al., 2015). The DainS mutant is a complete deletion of the ainS gene resulting from a double cross-over event, luxIis an insertional mutant constructed by cloning an internal part of the luxI gene into a suicide vector pNQ705. The insertional mutant is a result of a single cross-over event and is chloramphenicol resistant, DainSluxIis the DainS complete deletion mutant with an insertional mutation in the luxI gene using a luxI-pNQ705 plasmid (chloramphenicol resistant) and DlitR is a complete deletion of the litR gene using a suicide plasmid and a double cross-over event. All mutants were tagged with GFP using tri-parental mating as described previously (Khider, Willassen & Hansen, 2018). Briefly, the pVSV102 plasmid carrying the gene coding for GFP and kanamycin was transferred from E. coli DH5alpir to the mutant strains using the conjugative helper strain CC118lpir harboring pEVS104 helper plasmid. Donor and helper cells were grown to mid-log phase (OD 600 ¼ 0.7) in LB1. Recipient strains (A. salmonicida) were grown to early stationary phase (OD 600 ¼ 1.2) in LB2.5. The donor, helper, and recipient strains were harvested (13,000Âg for 1 min) and washed twice with LB1 before they were mixed in 1:1 ratio and spotted onto BA2.5 plates, followed by overnight incubation at 16 C. The spotted cells were re-suspended in LB2.5 and incubated for 24 h at 12 C with agitation (220 rpm). The potential tagged strains were selected on BA2.5 supplemented with 150 ml/ml kanamycin after 5 days. The tagged strains were confirmed using Nikon Eclipse TS100 Inverted Fluorescence Microscope (Nikon, Melville, NY, USA).

Static biofilm assay
The biofilm assay was performed as described previously (Hansen et al., 2014;Khider, Willassen & Hansen, 2018). Briefly, the overnight secondary cultures were grown to an OD 600 of 1.3. The secondary cultures were further diluted 1:10 in SWT and a total volume of 300 ml of culture was added to each well of flat-bottomed, non-tissue culture-treated Falcon 24-well plates (BD, Biosciences). For each mutant and the wild type, final concentrations of 1,400 ng/ml of 3OC6-HSL, 100 ng/ml of 3OHC10-HSL, 197 ng/ml of 3OC8, 100 ng/ml of C8, or 400 ng/ml of C6 were added separately to each well. The concentrations of the AHLs were selected based on those A. salmonicida produced in "in vitro" experiments (Hansen et al., 2015). The plates were incubated statically at 8 C, for 72 h and the biofilm was visualized using a Nikon Eclipse TS100 Inverted Fluorescence Microscope (Nikon) at 10Â magnification and photographed with Nikon DS-5Mc (Nikon) camera. The biomasses of the biofilms were indirectly quantified using crystal violet. The medium was removed and 300 ml of 0.1% (wt/vol) crystal violet in H 2 O was added. The plates were incubated at room temperature for 30 min. The crystal violet stain was removed by flipping the plates gently. The wells were then washed twice with 0.5 ml of H 2 O. The plates were air dried overnight and the biofilm was dissolved in 0.5 ml of 96% ethanol with agitation (250 rpm) overnight. The dissolved biofilm was diluted 1:10 in 96% ethanol and transferred to a 96-well plate (100 ml/well). The absorbance was measured at 590 nm (Vmax kinetic microplate reader; Molecular Devices, LabX, Midland, ON, Canada).

Soft agar motility assay
The motility assay was performed using SWT soft agar plates containing 0.25% agar as previously described (Khider, Willassen & Hansen, 2018). Briefly, the secondary overnight cultures were diluted to an OD 600 of 0.4 and three ml of each culture was spotted onto the soft agar plates and incubated at 8 C for 5 days. The degree of motility for each strain was monitored every 24 h for 5 days by measuring the diameters of spreading halos on the soft agar plate.

Colony morphology and adhesion assay
The colony morphology assay was carried out as described previously (Hansen et al., 2014;Khider, Willassen & Hansen, 2018). In short, a 250 ml aliquot was harvested from each secondary overnight culture by centrifugation and the pellet was re-suspended in 250 ml SWT. Two microliters of each culture was then spotted onto SWT agar plates and incubated at 8 C for 14 days. The colonies were viewed microscopically with Zeiss Primo Vert and photographed with AxioCam ERc5s (Zeiss, Berlin, Germany) at 4Â magnification. Adhesion was examined by using the same 14 day old colonies to test their ability to adhere to the SWT agar plates. The assay was performed by touching the colonies using a sterile plastic loop as previously described (Bjelland et al., 2012;Khider, Willassen & Hansen, 2018). The adherence grading was only recorded as "Non adhesive" for smooth and non adherent colonies, "Weak" for slightly adherent and "Strong" for colonies that were impossible to separate from the SWT agar plate.

Scanning electron microscopy
The overnight secondary cultures of A. salmonicida strains were fixed with 2.5% (wt/vol) glutaraldehyde and 4% formaldehyde in PHEM-buffer and incubated for one day at 4 C.
A total of 100 microliters of each sample was mounted on a poly-L-lysine coated coverslip for 5 min. Coverslips were washed three times with PHEM buffer before they were postfixed in 1% (wt/vol) Osmium tetroxide (OsO 4 ). Samples were washed an additional three times with PHEM buffer. All samples were dehydrated with a graded series of ethanol solutions at room temperature for 5 min. The samples were dried using hexamethyldisilazane as a drying agent and left to dry in a desiccator overnight before being mounted on aluminum stubs using carbon tape and silver paint. The samples were coated with gold-palladium using a Polaron Range Sputter Coater (Polaron, Quorum Technologies Ltd, East Sussex, UK). Pictures were obtained with Ziess Zigma Scanning Electron Microscopy (Ziess, Berlin, Germany).
All biological assays were carried out in biological triplicates, unless otherwise indicated. The assays were performed in two to three independent experiments to validate the results. The difference between the mutants relative to the wild type in AHL production, biofilm formation, and motility migration zones were calculated using student's t-test. A p-value 0.05 was regarded as significant.

AHL profiling of A. salmonicida in SWT medium
In our previous studies, AHL profiling of A. salmonicida LFI1238 and mutants thereof were performed after growth in LB2.5 medium. However, since SWT medium is required for biofilm formation we first wanted to know whether a change of medium would affect the AHL profiles of A. salmonicida wild type and the mutants.
The different A. salmonicida strains (LFI1238, DlitR, DainS, luxI -, and DainSluxI -) were grown in SWT medium at 8 C for 50 h (OD 600 ∼ 2.0) before samples were harvested and analyzed using HPLC-MS/MS. The A. salmonicida wild type and mutants showed AHL profiles ( Table 2) that were similar to the profiles after growth in LB (Hansen et al., 2015) with the exception of C4 and 3OC4. Thus, the wild type and the DlitR AHL profiles consisted of six AHLs, where the 3OC6-HSL was the most abundant. No AHLs were detected in the DainSluxIsupernatant. The luxImutant produced only 3OHC10-HSL, and the DainS mutant produced the remaining five AHLs. Compared to the wild type, the DlitR mutant produced lower concentrations of the 3OC6-HSL and 3OHC10-HSL confirming that LitR is a positive regulator of these two AHLs also after growth in SWT medium. Our previously reported results (Hansen et al., 2015) and the AHL profiling presented above, have shown that litR deletion significantly influenced the production of 3OC6-HSL (LuxI product) and 3OHC10-HSL (AinS product) compared to the wild type. Therefore, we wished to investigate the possible effects of 3OC6-HSL and 3OHC10-HSL on biofilm formation. The different AHLs were added separately to the SWT medium and A. salmonicida strains were allowed to form biofilm at 8 C for 72 h. As shown in Fig. 1A, the biofilm formation of DainSluxIwas totally inhibited when supplemented with either 3OHC10-HSL or 3OC6-HSL. The DainS, luxIand the wild type do not form a biofilm (Hansen et al., 2015), and no clear morphological differences in the biofilm formation was observed when treated with 3OHC10-HSL or 3OC6-HSL (Fig. 1A). The mushroom structured DlitR biofilm remained unchanged after the addition of AHLs. This shows that LuxI-3OC6-HSL and AinS-3OHC10-HSL functions through LitR, and downregulation of biofilm formation cannot be achieved when litR is inactivated (Fig. 1A). Next, the biomasses of treated and untreated biofilms were quantified using crystal violet. Relative to the untreated control samples, the addition of either 3OHC10-HSL or 3OC6-HSL significantly decreased the biomass of DainSluxIbiofilm (p-value 0.05). Quantitation of treated and untreated DlitR, LFI1238, DainS, and luxIshowed no significant changes (Fig. 1B). The treatment of A. salmonicida wild type and the mutant strains with other AHLs (C6, C8, and 3OC8) did not interfere with the biofilm formation (data not shown). This may indicate that these AHLs are not involved in the regulation of biofilm and have other functions in A. salmonicida.
luxImutant forms adherent wrinkled colonies in A. salmonicida To determine whether any of the A. salmonicida QS systems (lux and/or ain) are involved in the formation of wrinkled colony morphology, the luxI -, DainS, and the double mutant DainSluxIwere allowed to form colonies on SWT plates at 8 C. As shown in Fig. 2, the DainS mutant formed smooth colonies indistinguishable from those formed by the wild type. This may indicate that ainS is not required for formation of rugosity. The luxIand DainSluxImutants formed wrinkled colonies after 14 days of incubation similar to the colonies formed by DlitR. The wrinkled colonies formed by the luxI -, DainSluxI -, and DlitR mutants were found to be adhesive on the SWT agar at 8 C. The DainS mutant behaved similarly to the wild type and produced non-adhesive smooth colonies under the same conditions (Table S1).
Expression profiles of A. salmonicida luxIand DainS mutants revealed differentially expressed genes related to QS Transcriptome data of A. salmonicida wild type, luxIand DainS mutants In order to gain a better understanding of the roles of LuxI and AinS work in the QS system, samples from A. salmonicida LFI1238 wild type, luxIand DainS mutants were extracted at early logarithmic (OD 600 ¼ 0.3) and late exponential (OD 600 ¼ 1.2) phases from three independent cultures. The transcriptome expression profiles (luxIand DainS) of these cultures were compared to the A. salmonicida LFI1238 wild type. The total assembled transcriptome of A. salmonicida wild type LFI1238 generated an average of 9.87 million reads at LCD and 9.56 million at HCD. The average of mapped reads to the reference genome (A. salmonicida LFI1238) was 88.7% at LCD and 91.4% at HCD, with an average mapping coverage of 140.6 and 141.0, respectively. The total assembled transcriptome of luxIgenerated an average of 11.5 million reads at LCD and 9.18 million at HCD. The average of mapped reads to the reference genome (A. salmonicida LFI1238) was 94.5% at LCD and 92.9% at HCD, with an average mapping coverage of 109.5 and 85.2, respectively. The detailed transcriptome data of DainS also showed average of mapped reads to the reference genome of 93.9% and 92.8% at LCD and HCD, respectively, with average mapping coverage of 88 at LCD and 100.8 at HCD (Table S2).
To control for technical variations between biological replicates PCA analysis on the expression data from HCD vs. LCD was performed using DESeq2. The biological replicates clustered together well and were distinct between HCD and LCD. The transcriptome profile of A. salmonicida luxImutant relative to the wild type The expression profiling of luxImutant relative to the wild type LFI1238 revealed 494 and 446 DEGs at LCDs and HCDs, respectively, that fell into various functional gene classes adapted from MultiFun (Serres & Riley, 2000) (Fig. 3). The luxI -DEGs were distributed almost equally between the large and small A. salmonicida chromosomes with 292 DEGs at LCD (59%) and 259 DEGs at HCD (55%). Among the DEGs at LCD 366 were downregulated and 128 were upregulated (Table S3). While at HCD 224 genes were downregulated and 222 genes were upregulated (Table S4). Many of the DEGs we discovered were genes organized in operons. One of the significantly downregulated genes is the rpoQ sigma factor (VSAL_II0319), organized in an operon of seven genes (VSAL_II0319-VSAL_II0325). The whole operon was differentially expressed at HCD and five genes of the operon were differentially expressed at LCD with fold change values  (Table 3). The luxImutant formed wrinkled colony morphology on SWT plates. The rugosity is associated with the enhanced production of EPSs, which requires the expression of syp operon (18 genes) in A. salmonicida (Hansen et al., 2014;Khider, Willassen & Hansen, 2018). The syp operon (VSAL_II0295-VSAL_II0312) is located on chromosome II and organized in four transcription units (Fig. 4A). Our data at HCD demonstrated that 11 genes in the syp operon were significantly upregulated with fold change values ranging from 9.55 to 2.04, the remaining genes of the syp operon with fold change values ranging from 1.01 to 1.97 were filtered out due to the predominant criteria for identifying DEGs (fold change !2 and -2, p-value 0.05) ( Table 3).
Among the upregulated genes that fell into the cell envelope (surface structures) functional group were genes associated with the tight adherence (Tad) loci also known as tad operon, that consists of 13 genes located on chromosome II of A. salmonicida genome (Fig. 4B). Among the genes of the tad operon with highest level of expression were VSAL_II0366 (83.8-fold change at LCD and 151.5-fold change at HCD) and VSAL_II0377 (57.3-fold change at LCD and 39.5-fold change at HCD) coding for fimbrial proteins, Flp/Fap pilin component and type IV leader peptidase, respectively. The remaining genes of the tad operon were also upregulated in the luxImutant relative to the wild type at both cell densities, and are listed in detail in Table 3. Within the same functional group (cell envelope), we were able to identify four downregulated DEGs (VSAL_I0471,  VSAL_I0473, and VSAL_I0479) at LCD and one at HCD (VSAL_I0476) associated with type IV pilus.
In several bacteria, QS has been shown to regulate motility and flagellar synthesis (Kim et al., 2007;Ng & Bassler, 2009). The expression profile of the luxImutant revealed genes associated with motility and chemotaxis (59 DEGs at LCD and 57 DEGs at HCD) (Tables S3 and S4). The greatest transcript abundance at LCD and HCD were regulatory genes flrA (VSAL_I2312) encoding sigma 54-dependent transcription regulator, flrB (VSAL_I2311) coding for two-component system, sensor histidine kinase and flrC (VSAL_I2310) coding for response regulator. Other genes coding for flagellin subunits and flagellar basal body rod, ring, hook, and cap proteins, were also downregulated in the luxImutant relative to the wild type at both cell densities. Additionally, genes coding for methyl-accepting chemotaxis proteins and motor proteins as MotA and MotB were downregulated in the luxImutant (Fig. 5).
The remaining DEGs of the luxI -/wt transcriptome were mostly genes with unknown functions, transport/binding proteins, extrachromosomal/foreign DNA, and small RNAs functional groups (Fig. 3). Values indicated in bold are genes filtered out from the analysis due to fold change values (FC) below 2 and !-2 (not significantly expressed).
The transcriptome profile of A. salmonicida DainS relative to the wild type The transcriptome of DainS relative to the wild type revealed fewer DEGs compared to the luxI -. At LCD, a total of 20 DEGs (8 up-and 12 downregulated) were identified. At HCD we were able to identify 29 DEGs where 8 were upregulated and 21 genes were downregulated (Tables S5 and S6). The DEGs fell into 10 functional groups (Fig. S1). At LCD and in the absence of AHLs, the ain system of V. harveyi act as kinase and serve as phosphoryl-donors to LuxU, which in turn phosphorylates LuxO (Freeman & Bassler, 1999). The DainS transcriptome demonstrates an upregulation in genes responsible for phosphorylation. The DEGs with significant expression level relative to the wild type was phosphorelay protein LuxU (VSAL_I1875) with a fold change values of 2.22 and 2.37 at LCD and HCD, respectively. Whereas the luxO gene (VSAL_I1874) was not differentially expressed (1.14-fold change at LCD and 1.62-fold change at HCD) and thus not considered further.  tides (1998701-1999579). In Vibrios flagellar genes fall into four different hierarchical classes; Class I encode the regulatory protein FlrA, which together with sigma factor 54 controls expression of Class II genes. Class II proteins FlrB and FlrC are important for controlling transcription of Class III genes necessary for synthesis of hook, basal body, and filaments. Class II sigma factor 28 (FliA) regulates transcription of Class IV genes associated with the production of motor components (Brennan et al., 2013;Norsworthy & Visick, 2013;Millikan & Ruby, 2003Stewart & McCarter, 2003). Among the upregulated genes that fell into the surface structures functional group was the tad operon (VSAL_II0366-VSAL_II0378). The VSAL_II0366 gene coding for fimbrial protein showed a fold change values of 2.82 and 4.24 at LCD and HCD, respectively. VSAL_II0367 coding for Flp/Fap pilin component and type IV leader peptidase was identified among upregulated genes at LCD only (Table S5). Among the 21 genes that were downregulated at HCD, the DEGs with highest fold change values were assigned to amino acid biosynthesis functional group including the sulfate adenyltransferase subunit 1 and 2 encoded by VSAL_I0421 and VSAL_I0420, respectively (Table S6).

LuxI controls motility in A. salmonicida LFI1238
The flagellum is required for motility of bacteria, mediating their movements toward favorable environments and avoiding unfavorable conditions (Utada et al., 2014;Zhu, Kojima & Homma, 2013). Since the expression profile demonstrated that a large group of flagellar biosynthesis and assembly genes are regulated by the lux system, we wished to analyse the motility behavior of the QS mutants (luxI -, DainS, and DainSluxI -), using a soft motility assay.
Our results showed that inactivation of luxI resulted in a non-motile strain, where the size of the spotted colony (2.0 mm) did not change, indicating no migration from the site of inoculation (Figs. 6A and 6B). AinS was shown to negatively regulate motility in A. fischeri (Lupp & Ruby, 2004), and similarly, we assessed the impact of ainS deletion on motility of A. salmonicida. Compared to the wild type, which showed motility zones of 26.6 ± 0.57 mm, the DainS showed an increased motility, where migration through the soft agar resulted in motility zones of 30.3 ± 0.57 mm. Similarly, the DainSluxIdouble mutant also demonstrated an increased motility compared to the wild type with motility zones of 31.3 ± 1.15 mm (Table S7). In order to determine whether the strains analyzed by soft motility assay possess or lack flagella, the wild type and the constructed mutants were visualized by scanning electron microscopy. The DainS and DainSluxImutants produced several flagella similar to the wild type. As expected the luxImutant is non-motile and lacks flagella (Fig. 6C).

DISCUSSION
Acyl-homoserine lactones have been identified in many vibrio and aliivibrio species including A. salmonicida (Buchholtz et al., 2006;García-Aljaro et al., 2008;Purohit et al., 2013;Valiente et al., 2009), which showed to produce a broad range of AHLs through LuxI and AinS synthases (Hansen et al., 2015). However, there is still limited understanding of the biological advantages of this AHL diversity in the QS mechanism. In this study, we have demonstrated the influence of luxI and ainS on the global gene regulation and the impact of AHLs on several phenotypic traits related to QS in order to understand the complex network of signal production and regulation in A. salmonicida.
The ability to form rugose colonies and biofilm are often correlated features in vibrios (Casper-Lindley & Yildiz, 2004;Yildiz & Schoolnik, 1999;Yildiz et al., 2004), where wrinkled colony phenotype is generally associated with enhanced EPS production (Yildiz & Schoolnik, 1999). Likewise, in A. salmonicida colony wrinkling (rugosity) and biofilm formation requires the expression of syp genes responsible for the production of EPS (Hansen et al., 2014;Khider, Willassen & Hansen, 2018). In the study presented here, we show that luxImutant exhibited a strong wrinkling colony morphology, indicating an enhanced polysaccharide production. Earlier studies demonstrated that inactivation of the AHL synthase (luxI homologous) in several bacteria caused a reduction in both AHL and EPS production (Koutsoudis et al., 2006;Molina et al., 2005;Von Bodman, Bauer & Coplin, 2003). However, here we show that inactivation of luxI in A. salmonicida enhanced the EPS production and resulted in wrinkled colonies. Unlike the luxImutants, the DainS mutant formed smooth colonies, similar to the wild type. These findings are further confirmed by the transcription analysis, which revealed upregulation of 11 syp genes and downregulation in the rpoQ gene in the luxImutant relative to the wild type. Whereas no DEGs associated with EPS production or rpoQ were found in the DainS transcriptome. The sigma factor RpoQ functions downstream of LitR and is known to be a strong repressor of syp in A. salmonicida (Khider, Willassen & Hansen, 2018). Although LitR was shown to be a positive regulator of rpoQ (Khider, Willassen & Hansen, 2018), the gene (litR) was not found among the DEGs of the luxIor DainS transcriptome. We have previously proposed that the expression of rpoQ may be independent of LitR and that its regulatory function can be altered by environmental conditions (Khider, Willassen & Hansen, 2018). Hence, our results indicate that the negative regulation of EPS through rpoQ is controlled directly or indirectly by LuxI, where this regulation may involve other genes and transcription factors independent of LitR, AinS or the 3OHC10-HSL production. Interestingly, the active AinS and thereby the production of 3OHC10-HSL did not downregulate the rugosity of the luxImutant through the LitR-RpoQ pathway. This can be explained by the late production of the 3OHC10-HSL in A. salmonicida (Hansen et al., 2015). We have reported previously that the concentration of 3OHC10-HSL is very low at OD 600 ¼ 1.2 (the OD 600 of transcriptomics analysis and morphology assay), which may not be strong enough to induce the LitR-RpoQ cascade involved in the repression of syp genes. On the other hand, the DainS mutant appeared to have wild type colony morphology, indicating a stronger regulatory effect of the LuxI on the RpoQ leading to repression of EPS (via syp) and downregulation of rugosity. These results suggest that the luxImutant is locked into a regulatory state in the bacterial life cycle that requires the EPS production, whereas the DainS is locked into a different regulatory state, where the EPS production is not necessary. LitR is suggested to link AinS/R and LuxS/PQ systems to LuxI/R systems in A. salmonicida, where its deletion influenced the production of AinS and LuxI AHLs. When both luxI and ainS were inactivated simultaneously, biofilm and colony morphology similar to DlitR mutant was formed (Hansen et al., 2015). A simple explanation for this observation is the deficiency in AHL production, leading to litR inactivation (Bjelland et al., 2012;Hansen et al., 2014), and thereby no repression of biofilm or colony rugosity is achieved. Furthermore, the exogenous addition of either 3OHC10-HSL (AinS signal) or 3OC6-HSL (LuxI signal) to DainSluxI -, completely inhibited biofilm formation. We have previously shown that the disruption of either EPS or other matrix components (e.g., proteins, lipoproteins, and eDNA), disrupts the mature biofilm formation in A. salmonicida (Hansen et al., 2014;Khider, Willassen & Hansen, 2018).
While the DainS mutant did not produce neither mature biofilm nor wrinkled colonies, introduction of luxI mutation to a DainS background, resulted in strains (DainSluxI -) with three-dimensional biofilm architecture and wrinkled colonies. These data suggest that these two systems regulate biofilm formation synergistically, where the effect of AinS and LuxI AHLs is operated through a common pathway as previously reported (Hansen et al., 2015). The results presented here show that both systems function to either promote or repress production of EPS and other matrix components. However, the lux system is believed to be essential for the production of EPS rather than the ain system (as discussed above). Studies showed that one key function of EPS involves the attachment of cells to different substratum, which is the initial step in biofilm formation (Vu et al., 2009). For example in V. cholerae, the EPS production is the first step in biofilm formation as cells switch from motile planktonic state to being non-motile and surface attached (Silva & Benitez, 2016). Likewise, we suggest that the non-motile luxImutant, increases EPS production to mediate the initial steps in biofilm formation, whereas ainS is neither fully activated nor required at this time. This suggests that lux system may operate at a lower threshold cell density than ain system, which is more essential at later stages of biofilm development, mainly the maturation into three-dimensional mushroom structure. With our results, we expand the previously suggested model, to include luxI and ainS and their proposed role in regulating biofilm formation and colony rugosity. In the model Figure 7 The proposed model of QS system in A. salmonicida LFI1238. The autoinducers synthases LuxS, LuxI, and AinS produces AI-2 and eight AHLs that are transported across the outer (OM) and inner membrane (IM) (Hansen et al., 2015). At high cell density AHLs and AI-2 are accumulated to reach a critical concentration to be sensed by their receptors LuxPQ, AinR, or LuxRs. It is still unknown which AHLs bind the LuxRs and is illustrated with a question mark. The Al-2 binds LuxPQ and 3OHC10-HSL binds AinR, which in turn induces a dephosphorylation cascade, resulting in LitR activation. Although due to a frame shift mutation within luxP, the LuxS/PQ pathway may not be active. The expressed LitR, activates the production of the AinS AHL (3OHC10-HSL) and the expression of downstream rpoQ gene. The increased RpoQ levels represses syp operon leading to biofilm disruption and inhibition of colony rugosity. Moreover, LitR represses other matrix components, through a pathway that remain unknown (Khider, Willassen & Hansen 2018). LitR together with LuxRs are proposed to regulate luxI. The expressed LuxI mediate the production of seven AHLs and represses syp genes via RpoQ either directly or indirectly through unknown factors. Blue arrows and red lines with bar end indicate pathways of positive and negative regulation, respectively, and may consist of several steps. The thicker, empty arrows indicate the resulting phenotypes.
Full-size  DOI: 10.7717/peerj.6845/ fig-7 presented in Fig. 7, we propose that as cell density rises 3OHC10-HSL binds AinR receptor, resulting in activation of LitR, which in turn regulates the production of AinS AHL. The activated LitR leads to a repression on other matrix components required for building a mature biofilm through a mechanism that remain unknown. The activated LitR also leads to increased levels of RpoQ, resulting in repression of EPS through syp genes. It has been proposed that luxI is activated by both LitR and LuxRs. The active LuxI synthesizes seven AHLs and represses syp operon via RpoQ most probably independently of the AinS-LitR pathway. Hence, the production of EPS (via LuxI) and other matrix components (via AinS) appears to play an important role in building the three-dimensional architecture of the biofilm. In summary, our results further support the hypothesis that biofilm formation is a LCD dependent phenotype, when neither LuxI nor AinS AHLs are present. As AHLs accumulate at HCDs, the biofilm is dispersed, indicating that AHL-mediated QS in A. salmonicida is involved in the dispersal step of the biofilm cycle. Although further investigations are needed to support this hypothesis. The deletion of luxI was shown to influence expression of 500 genes in the A. salmonicida. A similar global regulation of QS regulon has also been observed in Pseudomonas aeruginosa, where around 600 genes were believed to be regulated by las and rhI QS systems (Wagner et al., 2003). The most pronounced regulation in the luxImutant was observed for genes involved in motility and chemotaxis, exhibiting a significantly low expression level. The regulatory mechanism of motility in A. salmonicida remain poorly understood, however, the motility genes are organized in a similar fashion to A. fischeri (Karlsen et al., 2008), where flagellar genes are often grouped into different hierarchical classes (Fig. 5). We propose that the loss of motility and flagellation is associated with the elevated level of EPS production by the luxImutant. Inverse regulation of EPS production and flagellum has been observed in several other microorganisms (Burdman et al., 1998;Garrett, Perlegas & Wozniak, 1999;Prigent-Combaret et al., 1999). In V. cholerae O139, strains with inactivated flagellar genes (e.g., fliK, flhB, fliF, fliE, and fliR) exhibited rugose colony morphology, while mutations in genes coding for motor proteins (motB and motY) did not display V. cholerae strains with a rugose colony morphology (Watnick et al., 2001). We also recently showed that RpoQ in A. salmonicida regulates motility and EPS production inversely, where the DrpoQ mutant exhibited a strong colony rugosity and reduced motility. The transcriptome of DrpoQ revealed a downregulation in a number of flagellar biosynthesis genes mainly flaA (Khider, Willassen & Hansen, 2018;Khider et al., 2019). Our luxItranscriptomics results demonstrated DEGs that fell into all hierarchical classes (Fig. 5) including flagellar and motor genes. Thus, it is unclear at which regulatory level LuxI affects motility genes and which motility genes may be associated with the rugose phenotype. Nevertheless, our lack of ability to complement the luxImutant makes it difficult to exclude other factors influencing flagellar synthesis and/or motility and a question remains to be investigated is whether there is a relationship between the loss of motility and the rugose colony morphology in A. salmonicida. LitR, is a positive regulator of ainS in A. salmonicida (Hansen et al., 2015). Thus, not surprisingly, we found that DainS displayed an increased motility compared to the wild type, similar to that reported for DlitR (Bjelland et al., 2012). The defect build the mature biofilm. Furthermore, we identified DEGs associated with motility were regulated by LuxI. These results add a new knowledge to the nature of the QS mechanism of A. salmonicida, however further investigations are needed to understand the regulation and complexity of this mechanism.