The alternative sigma factor RpoQ regulates colony morphology, biofilm formation and motility in the fish pathogen Aliivibrio salmonicida

Aliivibrio salmonicida belongs to the Vibrionaceae family, which is widely distributed in the environment, mainly in the aquatic habits. Members of this family may exist in symbiotic or pathogenic relations with their hosts [1]. According to current taxonomy, A. salmonicida belongs to the Aliivibrio genus together with its three most closely related species Aliivibro logei, Aliivibrio wodanis and Aliivibrio fischeri [2].

A. salmonicida causes cold water vibriosis or Hitra disease in farmed Atlantic salmon (Salmo salar L), Atlantic cod (Gadus morhua) and rainbow trout (Oncorhynchus mykiss). The disease occurs mainly during late autumn and winter seasons when the seawater temperature is below 12 °C. A. salmonicida is a gram-negative psychrophilic bacterium with a rod shape and nine polar flagella for motility and colonization [3–5].

Members of the Vibrionaceae family use quorum sensing (QS) for cell-to-cell communication to regulate gene expression in response to cell density by secretion and sensing of extracellular signals called auto-inducers (AIs). As the bacterial population density increases, AIs accumulate in the environment. When the AI concentration increases above a certain threshold, the bacteria detect this and modulate gene expression [6, 7]. N-acyl homoserine lactones (AHLs) are the major class of AIs in gram-negative bacteria, and were first described in A. fischeri [8, 9] and Vibrio harveyi [10]. The QS systems in A. fischeri control properties such as motility, squid colonization and bioluminescence [11–13]. A. fischeri has two AHL based systems, LuxI/LuxR and AinS/AinR, which are primarily responsible for regulating bioluminescence and colonization factors [14]. In addition to the LuxI/LuxR and AinS/AinR systems, A. fischeri has the LuxS/LuxPQ QS system [14, 15]. LuxI is responsible for the synthesis of the autoinducer N-3-(oxo-hexanoyl)-homoserine lactone (3-oxo-C6-HSL) which binds the cytoplasmic receptor LuxR. LuxR then functions as a transcription activator for the luciferase luxICDABE operon [16]. LuxS and AinS synthesize signal molecules which are sensed by LuxPQ and AinR, respectively. The two signal systems work in parallel and convey the signal responses to LuxU-LuxO. At low cell density when AIs are not produced, LuxPQ and AinR act as kinases and relay phosphates to LuxU, which in turns phosphorylates LuxO. Phosphorylated LuxO activates the transcription of qrr which binds and destabilizes the mRNA of the master QS regulator LitR [12, 15, 17]. At high cell density, the AI produced by AinS (C8-HSL) accumulates in the environment and results in dephosphorylation of LuxO. When LuxO is dephosphorylated, the qrr level decreases and allows LitR translation. In turn, LitR activates the transcription of luxR which contributes to bioluminescence [12, 13].

A. salmonicida has three QS systems similar to those in A. fischeri: LuxS/LuxPQ, LuxI/LuxR and AinS/AinR [18]. LuxI is responsible for the synthesis of a total of seven AHLs, while AinS synthesizes only one AHL. This AHL diversity may suggest a complex sensing system which allows more fine-tuned responses to changes in the environment [19]. A. salmonicida does not produce bioluminescence per se [20], but regulates activities such as virulence, motility, colony morphology, adhesion, and biofilm formation by QS in a temperature dependent manner [21, 22].

Sigma factors are essential dissociable subunits of prokaryotic RNA polymerase that control promoter recognition and transcription initiation [23, 24]. Primary sigma factors (RpoD, σ⁷⁰ family) direct transcription from the promoters of genes required for basic cellular functions. In addition to the primary sigma factors, bacteria have a variable number of alternative sigma factors whose activities increase in response to certain environmental conditions or stress [25].

Several alternative sigma factors have been identified or predicted in vibrios and aliivibrios [26], and recently a divergent copy of a putative RpoS-like sigma factor was identified in A. fischeri and named RpoQ due to its activation by the AinS/AinR QS system [27]. RpoQ was later found to regulate bioluminescence, motility and chitinase activity in A. fischeri through LuxO via LitR [28]. Pfam analysis of RpoQ identified four conserved domains (σ⁷⁰ regions) where all were significant except for region 3. Phylogenetic analysis further revealed that region 3 in RpoQ is clearly divergent from the corresponding region in RpoD and RpoS [27, 28]. This less conserved region 3 is involved in binding the core RNA polymerase and recognition of the extended −10 promoter [29]. An RpoS-like sigma factor (RpoX) lacking region 3 has been described in Vibrio alginolyticus, and shown to be involved in biofilm formation and stress responses [30].

A. salmonicida strain LFI1238 encodes an rpoQ homolog (VSAL_II0319) similar to the one in A. fischeri [18, 28]. In a previous study we analyzed the transcriptomes of an A. salmonicida ΔlitR mutant and the isogenic wild-type strain LFI1238. The rpoQ gene was found to be downregulated in the ΔlitR mutant [31] suggesting that LitR is a positive regulator of rpoQ in A. salmonicida. In the work presented here we have studied the impact of this putative RpoS-like sigma factor in A. salmonicida with regard to different phenotypic traits such as biofilm formation, motility and colony morphology.

Our previous studies show that A. salmonicida LitR is involved in regulating a number of activities that may be important for host interactions [22], and by using microarray we identified a number of genes regulated by LitR [31]. The regulation of LitR on downstream genes could proceed either directly or indirectly. One of the genes found to be regulated by LitR was the rpoQ gene (VSAL_II0319). We therefore sought to analyze the role of RpoQ in the different phenotypes known to be regulated by LitR and QS in A. salmonicida. To this end we constructed an in-frame deletion mutant (ΔrpoQ) of the wild-type strain LFI1238 by removing 254 of the 294 amino acids in RpoQ. A complementation mutant (ΔrpoQ_c) was constructed to verify whether the observed phenotypes were due to the mutation of rpoQ. We do not expect the in-frame deletion (ΔrpoQ) or the insertion (ΔrpoQc) of rpoQ to have any polar effect(s) on downstream genes. However, it should be noted that this possibility cannot be excluded since the expression of the downstream genes in the operon was not analyzed in this work. Since temperature is an important factor involved in regulating AHL production and phenotypes related to QS in A. salmonicida [22, 31], the experiments were performed at different temperatures (4–16 °C).

RpoQ shows temperature dependent rugose colony morphology

The ability to form rugose colonies and biofilm are often correlated features in vibrios [38–40], and a rugose colony phenotype usually indicates high production of exopolysaccharides [39].

To compare colony morphologies of the wild-type LFI1238, and the ΔrpoQ and ΔlitR mutants a spot colony assay was performed on SWT agar incubated at different temperatures (4 to 14 °C). The LFI1238 produced smooth colony morphology at all temperatures as previously reported [31]. The ΔrpoQ mutant started to form wrinkled colonies after 7 days of incubation, and at day 12 a strong rugose colony morphology with wrinkled edges was observed after growth at 4 and 8 °C (Fig. 1). When incubated at 12 °C, the ΔrpoQ colony remained smooth in the central part whereas the edges became wrinkled. No wrinkling was observed for ΔrpoQ at 14 °C. The ΔlitR mutant was used as positive control [31] and, compared to ΔrpoQ, it showed a weaker rugose colony morphology. A strong ΔlitR rugose colony morphology similar to the wrinkled ΔrpoQ colonies was observed after 3 weeks (Additional file 2: Figure S2). As previously reported the wrinkling of ΔlitR colonies is absent after growth at 14 °C [31].

The wrinkled colonies formed by the ΔrpoQ and ΔlitR mutants were found to be adhesive on the SWT agar, and the adhesiveness was stronger at low temperatures (4 to 8 °C). No colonies were adhesive after growth at 14 °C (Additional file 3: Table S1). The complementary strain (ΔrpoQc) behaved similar to the wild-type and produced non-adhesive, smooth and creamy colonies at all temperatures.

RpoQ is involved in biofilm formation

In order to investigate whether rpoQ is involved in biofilm formation, the ΔrpoQ mutant was allowed to form biofilm in SWT medium at different temperatures using static conditions (Additional file 4: Figure S3). To better visualize the biofilm, GFP-tagged strains were used. The tagged strains were constructed by transferring a constitutive GFP expressing plasmid (pVSV102) into the different mutants and the wild-type strain. As shown in Fig. 2, ΔrpoQ produced a biofilm at 8 and 14 °C, which could be clearly visualized after 72 h. Little or no biofilm was observed at 16 °C for the different strains. The biofilm produced by the ΔrpoQ mutant does not show large mushroom shaped structures similar to those produced by ΔlitR (Fig. 2 and [31]); instead the ΔrpoQ mutant formed a more regular and flat biofilm with smaller micro-colonies and structures. Above the microscopically visual ΔrpoQ biofilm structures is a thick and slimy extracellular matrix without or with few embedded bacteria (Additional file 5: Figure S4). The complementary strain ΔrpoQc behaved similar to the wild-type, whereas the double mutant ΔlitR-rpoQ⁻ produced a biofilm with mushroom structure similar to the one produced by the ΔlitR mutant (Additional file 4: Figure S3).

RpoQ regulates motility in A. salmonicida

The flagellum is required for motility of bacteria, mediating their movements towards favorable environments or away from harmful conditions [41, 42]. Previous studies have shown that A. salmonicida is more motile at 12 °C than at 4 °C, and that LitR is a negative regulator of motility [22]. Here we analyzed the influence of RpoQ on the motility of A. salmonicida at different temperatures (4 to 16 °C). Deletion of rpoQ resulted in a strain with reduced motility compared to the wild-type and the ΔlitR mutant at all tested temperatures (Fig. 3 and Additional file 6: Table S2). After 5 days of incubation at 4 °C the ΔrpoQ mutant was almost non-motile and the motility zone was only 6.0 ± 1.0 mm. At higher temperatures (8 to 16 °C) the motility of the ΔrpoQ mutant was between 36 and 51% compared to the motility of wild-type. Hence, the incubation temperature did not seem to affect the regulatory effect of RpoQ on the motility. Similar to the wild-type and ΔlitR, the ΔrpoQ mutant shows highest motility at 14 °C. The ΔrpoQc behaved similar to the wild-type (Fig. 3a and b).

RpoQ is a negative regulator of biofilm

RpoQ is believed to function downstream of LitR in the QS cascade in A. salmonicida [31], and as shown above, deletion of rpoQ resulted in increased biofilm formation (Fig. 2). Hence, it was of interest to examine the influence of overexpressing rpoQ on the ΔlitR biofilm formation. For this purpose, the control vector (pTM214) and the inducible rpoQ vector (pTM214-rpoQ) were separately transferred to the ΔlitR mutant strain and the wild-type LFI1238 by conjugation. The biofilm assay was performed as before in SWT medium (4 to 16 °C) but with 1 mM IPTG to induce expression of rpoQ.

As shown in Fig. 4, overexpression of rpoQ disrupted or inhibited the biofilm formation produced by ΔlitR (ΔlitR-Ptrc-rpoQ at 4 to 14 °C) leaving small aggregates in the wells, whereas the ΔlitR biofilm formation was unaffected by the presence of the control vector (ΔlitR-pTM214) at all temperatures. Biofilm formation does not occur at 16 °C, and hence no effects of the overproduced rpoQ was observed. Neither was any changes observed when rpoQ was overexpressed in wild-type cells (LFI1238-Ptrc-rpoQ) (Fig. 4).

Overexpression of RpoQ decreases motility in A. salmonicida

In the experiments performed above we show that rpoQ is required for full wild- type motility at all temperatures (Fig. 3) and that overexpression of rpoQ has a negative effect on the biofilm forming ability of the ΔlitR mutant (Fig. 4). It therefore was of interest to analyze if overexpressed rpoQ also affected the motility of the wild-type and the ΔlitR mutant. As shown in Fig. 5, overexpression of rpoQ repressed the motility in both strains. Most notable, overexpression of rpoQ in the wild-type resulted in a completely non-motile strain when incubated at 4 and 8 °C, and the size of the spotted LFI1238-Ptrc-rpoQ colony (5 mm) did not change at any of the two temperatures during the 5 days of the experiment (Fig. 5a and b). At 12, 14 and 16 °C small motility zones (7–9 mm) were observed for LFI1238-Ptrc-rpoQ showing that overexpression of rpoQ in the wild-type does not result in complete shutdown of the motility at these temperatures. Overexpression of rpoQ in the ΔlitR also resulted in clearly diminished motility zones at all temperatures (Fig. 5b and Additional file 7: Table S3).

Bacteria continually face changes in their environment such as temperature fluctuations, nutrient accessibility and pH changes. In order to adapt to these changes and often challenging conditions, bacteria have developed various responses. Alternative sigma factors such as RpoS provide a main line of responses to changes in the environment by altering gene transcription [43, 44]. Several studies have shown a connection between RpoS and QS in different vibrios [38, 45–48]. When Cao et al. (2012) described the alternative sigma factor RpoQ in A. fischeri a homologue was only found in A. salmonicida [28]. However, since then the genomes of A. wodanis [49] and A. logei (A.logei S5–186 GeneBank accession no. AJY02000108.1) have become available. Analysis show that they also encode an RpoQ homolog with four conserved domains (σ⁷⁰ regions 1–4). RpoQ of A. salmonicida shares a high amino acid sequence identity (99%) with its homolog in A. logei whereas the amino acid sequence identity is 72% with A. fischeri and 69% with A. wodanis. Region 2 and region 4 of the putative RpoQ are well conserved between the four species, whereas region 3 is less conserved (Additional file 8: Figure S5).

RpoQ is regulated by LuxO through LitR in A. fischeri [28]. Similarly, our previous microarray results suggested that LitR is a positive regulator of RpoQ in A. salmonicida [31]. In the study presented here, we show that RpoQ is involved in regulation of colony morphology, adhesion, biofilm and motility similar to LitR. However, since RpoQ is suspected to act downstream of the master regulator LitR, one can expect that the ΔlitR mutant expresses phenotypes that are independent of RpoQ regulation.

The ΔrpoQ mutant demonstrated a stronger and an earlier onset of the rugose colony morphology as compared to the ΔlitR mutant. A rugose colony phenotype usually develops when the bacteria produce high amounts of polysaccharides, suggesting that more polysaccharides are made by the ΔrpoQ mutant. We know from our previous work that LitR represses the expression of the symbiosis polysaccharide (syp) operon, and that inactivation of syp (sypC, sypP and sypQ) in the ΔlitR mutant results in smooth colonies [31]. Hence, it is likely that LitR performs its activity on syp through RpoQ and that activation of RpoQ leads to a strong(er) repression of syp. The weaker rugose colony morphology of the ΔlitR mutant may be due to low levels of LitR-independent rpoQ expression, consistent with our previous microarray results that show expression of rpoQ in the ΔlitR mutant [31]. Hence, some repression of syp via RpoQ probably occurs in the ΔlitR mutant. Whereas in the ΔrpoQ mutant there is zero expression of rpoQ resulting in no or low syp repression and stronger rugosity.

Both LitR and RpoQ are negative regulators of biofilm formation in A. salmonicida. However, the ΔrpoQ mutant formed a biofilm morphologically different from the ΔlitR mutant. The biofilm produced by the ΔrpoQ was less mature and relatively flat and compact, without the large mushroom structures exhibited by the ΔlitR mutant. Additionally, the biofilm produced by ΔrpoQ contained a heavy and slimy extracellular matrix substance above the biofilm cells attached to the substratum (Additional file 5: Figure S4). This slimy matrix is likely due to high amounts of polysaccharides (i.g. syp expression as discussed above) that are common components of the extracellular matrix of biofilms, together with proteins and eDNA [50]. When we previously analyzed the ΔlitR biofilm we found that major components were polysaccharides and proteins, and by using electron microscopy we were able to see a network of fibers that connected biofilm cells together. The microarray analysis identified, in addition to syp, some lipoprotein, pili, flagella, and curli genes that were upregulated in the ΔlitR mutant [31]. Hence, LitR may repress some lipoproteins or filament structures needed to build up this mushroom-shaped biofilm architecture. Thus, one explanation for the observed biofilm morphology of the ΔrpoQ mutant may be that polysaccharide production is obtained through expression of syp, whereas expression of a functional LitR down-regulates genes involved in building mushroom shaped structures. When we inactivated syp in the ΔlitR mutant we found that although the rugose colony morphology reverted to wild-type morphology (smooth), some biofilm formation still occurred when using SWT medium [31]. Indeed, the biofilms produced by the ΔlitRsyp⁻ mutants resembles the biofilm produced by ΔrpoQ but without the slimy extracellular matrix. We therefore believe that the pathway through which LitR represses genes responsible for building the mushroom-shaped structures is different from the pathway through which LitR represses syp (via RpoQ) resulting in rugose colony morphology (Fig. 6).

Both mutants produce biofilms that are loosely attached; however, in contrast to the ΔlitR biofilm, the ΔrpoQ biofilm is not able to withstand the washing steps required after staining with crystal violet. To our knowledge RpoQ, has not been shown to be involved in biofilm formation of A. fischeri or any other aliivibrios. However, studies have shown that RpoS is able to enhance or repress biofilm formation in E. coli and other bacteria [51–53]. Additionally, RpoS has been shown to be involved in cell attachment and the maturation of biofilm [30, 54, 55], and inactivation RpoX in V. alginolyticus results in cells with decreased ability to form biofilm [29]. Likewise, inactivation of rpoQ in A. salmonicida may have reduced the ability of the bacteria to attach to the abiotic surface and to build a mature biofilm. Another explanation is that the ΔrpoQ biofilm contains a higher amount of a heavy, extracellular, slimy polysaccharide matrix that tears the biofilm away from the substratum when the medium or wash solutions is being poured out or a combination of both.

Thus, as shown in Fig. 6 we propose that RpoQ and LitR function in the same pathway, where RpoQ functions downstream of the LitR and is involved in repression of biofilm and the wrinkled colony morphology in A. salmonicida. The negative regulation cascade of extracellular polysaccharide matrix from LitR to the syp operon is probably operated through RpoQ, either directly or indirectly. The phenotypes are likely regulated in a cell density manner as previously discussed, where the development of a mushroom shaped biofilm structures and wrinkled colony morphology are initiated when neither AinS or LuxI AHLs are present at low cell density [19, 22, 31]. At high cell density when AHLs are produced, LitR represses genes required for building a mature biofilm structure, and activates rpoQ leading to repression of syp.

Inactivation of either rpoQ or litR had the opposite effect on motility in A. salmonicida. Unlike the ΔlitR mutant, which is more motile than the wild-type strain, the ΔrpoQ mutant exhibited significantly reduced motility. The complementary strain ΔrpoQc showed wild-type motility, suggesting that the termination of motility is due to rpoQ deletion and not to other factors. Reduced motility due to disruption of sigma factors has been reported for other bacteria, and inactivation of rpoS in Y. pseudotuberculosis results in decreased motility due to downregulation of the flagella master regulatory gene flhDC [51]. Thus, RpoQ may work in a similar manner by altering transcription of genes responsible for flagellar assembly or flagellar biosynthesis in A. salmonicida. Flagellum-mediated motility is important for specific stages of biofilm formation and surface attachment in several bacteria [56–58], and disruption of flagella biosynthesis is known to decrease attachment and alter biofilm architecture [59–62]. For example, loss of motility in E. coli affected the biofilm architecture, where poorly motile strains formed flatter biofilms compared to highly motile strains, which displayed more mature vertical biofilm structures [63]. Thus, it is tempting to speculate that the decreased motility of the ΔrpoQ mutant resulted in cells with reduced ability to attach and form mature biofilms.

Furthermore, overexpression of rpoQ resulted in non-motile wild-type cells and ΔlitR cells with reduced motility. These results are similar to those obtained with A. fischeri, where the overexpression of rpoQ in the wild-type and ΔlitR mutant resulted in non-motile strains [28]. The finding that both deletion and overexpression of rpoQ in A. salmonicida resulted in bacteria with reduced motility is interesting, but at the same time difficult to interpret. We know that RpoQ functions downstream of LitR and that LitR is a negative regulator of motility at high cell density [22, 31]. Thus, we may have expected to observe a similar effect on motility when we knocked out rpoQ. However, the ΔrpoQ mutant show decreased motility compared to the wild-type indicating that RpoQ is a positive regulator of motility (Fig. 6). This may suggest that at low cell densities some litR independent expression of rpoQ occurs and that RpoQ activates genes involved in flagellar biosynthesis. As the cell population increases litR will be expressed leading to increased levels of RpoQ. High RpoQ levels (overexpression of rpoQ) then turns down motility probably by acting as an activator of genes involved in down regulation of the flagellar apparatus. Hence, RpoQ probably controls genes responsible for both promoting or repressing motility depending on growth phase, environmental conditions and stress factors. Our results show that regulation of motility in A. salmonicida is complex similar to other vibrios [64] and probably involves several regulatory genes and factors, which is still unrevealed.

Temperature is an important factor in developing cold-water vibriosis and for production of AHLs in A. salmonicida. When the bacteria is grown at temperature above the disease limit (16 °C), the production of AHLs is nearly absent [19]. Our results from the biofilm and colony morphology assays show that the ΔrpoQ mutant behaves as the wild-type strain and the ΔlitR mutant when the assays are performed at 16 °C, and neither of the strains forms rugose colonies or biofilm at this temperature. This shows that RpoQ, similar to LitR, represses formation of biofilm and rugose colonies more at low temperatures (4–14 °C), and at 16 °C the effect of the rpoQ deletion is absent with regard to these phenotypes. Interestingly, this temperature effect was not observed when the motility of the ΔrpoQ mutant was analyzed, and at 16 °C the motility of the ΔrpoQ mutant was still clearly reduced compared to the wild-type. This implies that RpoQ is expressed and is able to regulate motility in A. salmonicida at temperatures above the limit for developing cold water vibriosis, and at conditions when AHL concentrations are expected to be low.

In this work we have shown that the alternative sigma factor RpoQ regulates motility, colony morphology and biofilm formation in A. salmonicida. This broad range of different phenotypes suggests that RpoQ is involved in a regulatory hierarchy influencing expression of a large panel of genes. Overexpression of RpoQ led to disruption of the biofilm produced by ΔlitR, paralyzed the motility of the wild-type A. salmonicida and caused a reduction in ΔlitR motility. These findings confirm that the RpoQ is a novel factor in the QS and functions downstream of the LitR. However, further studies are needed to understand exactly how LitR and RpoQ work together or independently to regulate the QS dependent phenotypes investigated here, and to identify genes regulated by RpoQ.