Vibrio fischeri is the exclusive symbiont residing in the light organ of the squid Euprymna scolopes. To understand the genetic requirements for this association, we searched a library of V. fischeri transposon insertion mutants for those that failed to colonize E. scolopes. We identified four mutants that exhibited severe defects in initiating colonization. Sequence analysis revealed that the strains contained insertions in four different members of a cluster of 21 genes oriented in the same direction. The predicted gene products are similar to proteins involved in capsule, exopolysaccharide or lipopolysaccharide biosynthesis, including six putative glycosyltransferases. We constructed mutations in five additional genes and found that they also were required for symbiosis. Therefore, we have termed this region syp, for symbiosis polysaccharide. Homologous clusters also exist in Vibrio parahaemolyticus and Vibrio vulnificus, and thus these genes may represent a common mechanism for promoting bacteria–host interactions. Using lacZ reporter fusions, we observed that transcription of the syp genes did not occur under standard laboratory conditions, but could be induced by multicopy expression of sypG, which encodes a response regulator with a predicted σ54 interaction domain. This induction depended on σ54, as a mutation in rpoN abolished syp transcription. Primer extension analysis supported the use of putative σ54 binding sites upstream of sypA, sypI and sypM as promoters. Finally, we found that multicopy expression of sypG resulted in robust biofilm formation. This work thus reveals a novel group of genes that V. fischeri controls through a σ54-dependent response regulator and uses to promote symbiotic colonization.
The symbiotic colonization of the Hawaiian bobtail squid Euprymna scolopes by the marine bioluminescent bacterium Vibrio fischeri serves as a model for studying bacteria–host interactions (Ruby, 1999; Nyholm and McFall-Ngai, 2004). Like well-studied pathogenic associations, the formation of this symbiosis requires the bacterium to express traits that allow it to enter and multiply within a host organ. Furthermore, colonization by V. fischeri promotes developmental changes within the host, including tissue remodelling (Foster and McFall-Ngai, 1998; Foster et al., 2000).
Factors involved in establishing the exclusive relationship between V. fischeri and its invertebrate host have been examined from the perspective of both organisms. The surface of the symbiotic organ, known as the light organ, includes ciliated epithelial appendages that project into the body cavity of the squid and help direct the flow of the seawater towards this organ (Montgomery and McFall-Ngai, 1993). V. fischeri cells in the seawater then aggregate in squid-secreted mucus on the surface of the light organ (Nyholm et al., 2000). In addition to V. fischeri, other Gram-negative bacteria, such as Vibrio parahaemolyticus, also exhibit the capability to aggregate on the light organ (Nyholm et al., 2000). However, if both V. fischeri and V. parahaemolyticus are introduced into the seawater, V. fischeri rapidly becomes the dominant species in the mixed bacterial aggregate (Nyholm and McFall-Ngai, 2003). These data suggest that V. fischeri cells contribute to the observed specificity of the interaction.
After 2–3 h of aggregation, during which bacteria–bacteria and bacteria–host signalling likely occurs (Nyholm et al., 2000; Lupp and Ruby, 2005), V. fischeri cells enter the light organ. First, they migrate into one of six openings known as pores (Nyholm et al., 2000). This migration requires bacterial motility: non-motile bacteria aggregate but fail to migrate into the light organ (Graf et al., 1994; Nyholm et al., 2000; Millikan and Ruby, 2003). Next, they passage through ducts that contain nitric oxide, which may provide protection against colonization by non-symbionts (Davidson et al., 2004). Finally, they reach nutrient-filled crypts, where they multiply to high cell density and induce bioluminescence (Ruby and Asato, 1993; Graf and Ruby, 1998). Mutants defective for structural or regulatory genes required for light production (luxA, luxR and luxI) reach this stage of colonization, but subsequently fail to persist at wild-type levels within the light organ (Visick et al., 2000). Within the crypts, additional signalling occurs between the two organisms. Specifically, colonization by V. fischeri triggers apoptosis and subsequent regression of the surface ciliated appendages and a decrease in duct size (Foster and McFall-Ngai, 1998; Foster et al., 2000; Kimbell and McFall-Ngai, 2004). These changes presumably reduce the likelihood of colonization by other bacteria.
Besides motility and luminescence, other bacterial factors are required for various stages of symbiosis. Colonization requires the outer membrane protein OmpU for initiation, the metabolic enzyme Pgm for growth to high cell density, and the nitrogen and siderophore regulator GlnD for persistence (Graf and Ruby, 2000; Aeckersberg et al., 2001; DeLoney et al., 2002). In addition to these factors, a number of regulators have been identified as important for symbiotic colonization. These include quorum-sensing regulators (AinS, LuxS and LuxO), members of the class of two-component regulators (RscS, GacA and FlrA) and an alternative sigma factor (σ54, encoded by rpoN). Other than RscS, for which targets have not been identified, these regulators control multiple factors known or predicted to play roles in symbiotic colonization (Visick and Skoufos, 2001; Lupp et al., 2003; Whistler and Ruby, 2003; Lupp and Ruby, 2004; 2005; Wolfe et al., 2004). Two of these regulators, FlrA and LuxO, are predicted to work with σ54 to control expression of flagella and bioluminescence respectively (Lupp et al., 2003; Millikan and Ruby, 2003). Indeed, rpoN mutants exhibit defects in motility and bioluminescence regulation, as well as in nitrogen metabolism and biofilm formation (Wolfe et al., 2004). Mutants defective for rpoN fail to initiate symbiotic colonization; however, whether this defect is due solely to the loss of motility exhibited by these strains or also to other σ54-dependent phenotypes, such as biofilm formation, remains unclear.
Genes controlled by σ54-containing RNA polymerase require transcription factors (such as FlrA and LuxO) to promote formation of an open, initiation-competent, complex by RNA polymerase (Reitzer and Schneider, 2001). Often, the transcription factor is a two-component response regulator activated by phosphorylation in response to an environmental signal received by a membrane-associated sensor kinase. The result is a tightly controlled, environment-specific regulation of gene expression.
In this article, we report the identification of a novel cluster of genes required for symbiotic initiation by V. fischeri. These genes are co-ordinately regulated by σ54 and a previously uncharacterized response regulator, SypG. Multicopy expression of sypG resulted in a dramatic, σ54-dependent, enhancement of biofilm formation. This cluster, which includes putative genes for polysaccharide biosynthesis, is strongly conserved in the pathogens V. parahaemolyticus and V. vulnificus and thus may define a new paradigm for bacteria–host interactions in Vibrio species.
Isolation of symbiosis-defective mutants
To identify novel bacterial genes required for establishing symbiosis, we screened a transposon (Tn)-mutagenized library of V. fischeri using newly hatched juveniles of E. scolopes as described previously (Visick and Skoufos, 2001; DeLoney et al., 2002). Briefly, we inoculated animals with individual mutant strains. We measured successful colonization by monitoring bacterial bioluminescence, produced when V. fischeri reaches high cell density in the symbiotic light organ. We then determined colonization levels by homogenizing the animals and counting the resulting number of colony-forming units (cfu). Mutants defective in symbiotic initiation were further confirmed using additional animals.
Using this assay, we identified four independent mutant strains, from among about 600 screened, that exhibited severe defects in initiating symbiotic colonization. None of the animals exposed to these strains achieved symbiotic luminescence above background levels within 17 h post inoculation (Fig. 1A). In contrast, 90% of the animals exposed to the parent strain, ESR1, began to emit light within 10 h, and all 10 animals inoculated with ESR1 produced detectable levels of light by 17 h (Fig. 1A). In culture, the mutant bacteria grew like their parent on a variety of media. They also exhibited normal kinetics of bioluminescence induction and similar overall levels of light relative to the parent strain (data not shown). These data suggested that the lack of symbiotic luminescence reflected either a delay in or an absence of colonization. In support of this prediction, we found that most animals exposed to the Tn mutants remained uncolonized while the few that were colonized contained less than 1% of the symbionts found in animals inoculated with ESR1 (Fig. 1B). Because the majority of the animals remained uncolonized, these data indicated that the defect in initiation was severe.
We next investigated whether these mutant strains were defective for known symbiotic factors. Non-motile strains of V. fischeri fail to colonize (Graf et al., 1994; Millikan and Ruby, 2003; Wolfe et al., 2004), while hypermotile strains, if they colonize, do so with a significant delay (Millikan and Ruby, 2002). The four Tn mutants exhibited migration indistinguishable from that of their parent strain in a tryptone-based soft agar medium (data not shown). Symbiotic initiation also requires the outer membrane protein, OmpU (Aeckersberg et al., 2001); however, all four Tn mutants contained outer membrane protein profiles similar to that of the parent strain (data not shown). These data suggested that the Tns had inserted into genes not previously characterized as symbiosis determinants of V. fischeri.
Identification of a novel symbiotic gene cluster
We further characterized these novel colonization-defective mutants by examining the number and location of the Tn insertions by Southern analysis, cloning and sequencing. Each of the four mutant strains contained a single Tn insertion (data not shown; see Experimental procedures). Each insertion mapped to a distinct and previously uncharacterized gene within a cluster of 21 genes oriented in the same direction on chromosome 2 of V. fischeri(Fig. 2).
blast analyses (Altschul et al., 1990; 1997) and motif searches (Marchler-Bauer and Bryant, 2004) indicated that many of the genes in this cluster exhibited weak similarities to those involved in lipopolysaccharide (LPS), capsule or polysaccharide biosynthesis or export (Table 1). Six predicted proteins exhibited sequence similarities to glycosyltransferases, enzymes that transfer a sugar moiety from an activated donor (typically nucleotide sugar derivatives) to a specific acceptor (Kikuchi et al., 2003; Zhang et al., 2003). Target acceptors could be a growing carbohydrate chain of LPS, capsule or glycoproteins (Upreti et al., 2003). In addition, four genes were predicted to encode regulatory proteins such as two-component regulators. Based on these and other analyses (see below), we designated the genes in this locus as syp, for symbiosis polysaccharide.
Table 1. Putative functions of the V. fischeri Syp proteins and VFA1038–1040.
. The transposons inserted into sypC, sypD, sypJ and sypN, shown in bold.
. Protein sequences were submitted to blastp. The most significant matches over the length of the protein are listed, along with the e-value to the motif. No conserved domains were noted for VFA1019. The alignments spanned over 70% of the protein sequence for all of the Syp proteins, except SypJ (51.7%).
Sulphate transporter and anti-sigma factor antagonist
Outer membrane protein and related peptidoglycan-associated (lipo) proteins
Wza, periplasmic protein involved in polysaccharide export
Mrp, ATPases involved in chromosome partitioning
Response regulator receiver domain
Sigma factor PP2C-like phosphatases
Signal transduction histidine kinase
Response regulator receiver domain
Sigma-54 interaction domain
Response regulator receiver domain
HTH_8, bacterial regulatory protein, Fis family
Glycosyl transferases group 1
Glycosyl transferases group 1
Glycosyl transferases group 1
RfbX, membrane protein involved in the export of O-antigen and teichoic acid
RfaL, lipid A core-O-antigen ligase and related enzymes
GumC, uncharacterized protein involved in exopolysaccharide biosynthesis
Glycosyltransferase, probably involved in cell wall biogenesis
WczJ, sugar transferase involved in lipopolysaccharide synthesis
DUF1, domain of unknown function with GGDEF motif
COG3310, uncharacterized protein conserved in bacteria
To verify the importance of the syp cluster, we constructed derivatives of the wild-type strain, ES114, with mutations in this region. We cloned internal portions of specific genes into a suicide vector and introduced the constructs into V. fischeri. The resulting insertional mutants were confirmed by Southern analysis (see Experimental procedures). The ability of each mutant to initiate symbiotic colonization was compared with that of the wild-type strain. We used sypN as a control for this approach, as this gene was disrupted in one of the original Tn mutants and thus was not expected to colonize successfully. Indeed, like the original mutant, the insertional mutant of sypN colonized poorly (Fig. 3). Similarly, mutations in sypL, sypO, sypP, sypQ and sypR resulted in substantial defects (more than three orders of magnitude) in symbiotic colonization. In contrast, mutations in two downstream genes, VFA1038 (Fig. 2; Table 1) and VFA1043 (a putative ABC transporter), did not affect colonization of V. fischeri cells (Fig. 3). These data supported our hypothesis that this region functions in symbiotic colonization and suggested that the syp cluster does not extend to VFA1038 or beyond.
Identification of Syp homologues in related Vibrio species
Significant similarities of the Syp proteins to others in the database were largely limited to predicted proteins found in two marine pathogens, V. parahaemolyticus (Makino et al., 2003) and V. vulnificus (Chen et al., 2003). Further analysis using the TIGR database (http://www.tigr.org/) revealed that homologues of almost all of the syp genes were present and similarly clustered in the above pathogens but not in Vibrio cholerae (Heidelberg et al., 2000) (Table S1). The syp cluster was located on the small chromosome of V. fischeri, but on the larger of two chromosomes in both V. parahaemolyticus and V. vulnificus (Chen et al., 2003; Makino et al., 2003; Ruby et al., 2005).
The three genes (VFA1038–1040) immediately downstream of the syp cluster, although apparently conserved in V. parahaemolyticus, V. vulnificus and V. cholerae (Table S1), were not found in proximity to the other 18 genes in these organisms. The lack of clustering of the VFA1038–1040 homologues in V. parahaemolyticus and V. vulnificus suggests that these genes may not contribute to the function of this locus. This hypothesis is supported by our observation that a mutant defective for VFA1038 is not deficient for colonization. Therefore, at this time, we designate only first 18 genes of the V. fischeri cluster as syp. Similarly, a homologue of VFA1019, the gene immediately upstream of and transcribed divergently to the cluster genes, exists in V. parahaemolyticus (e = 1.2e−298) and potentially in V. vulnificus (e = 1.9e−22); however, these genes were unlinked to the respective clusters. Our preliminary data suggested that this gene also is not required for colonization (B.T. Grublesky and K.L. Visick, unpublished data).
In V. fischeri, the G+C content of the syp cluster genes ranged from 34% to 42% and averaged 37.6%, a value similar to the overall G+C content of V. fischeri genome (38.85%). For V. parahaemolyticus and V. vulnificus, the values of the syp homologues also reflected that of the average G+C content. Thus, the syp cluster does not appear to represent an ‘island’.
Transcriptional regulation of syp genes by SypG and σ54
Nine of the open reading frames in the syp cluster appear to overlap their upstream neighbours, suggesting that these genes may be co-ordinately regulated. Close examination revealed larger gaps (∼140 bp to 400 bp) between four sets of genes within the cluster as well as between sypA and the divergent gene VFA1019 (Fig. 2). We examined these five intergenic regions and found four potential promoters that could be recognized by RNA polymerase containing the alternative sigma factor, σ54(Fig. 4A). Unlike other sigma factors, σ54 recognizes sequences at positions −24 and −12 relative to the transcriptional start site. The consensus sequence of the σ54 binding site (mrNrYTGGCACG-N4-TTGCWNNw) is well conserved in a number of organisms (Barrios et al., 1999). The most important nucleotides in the consensus sequence (GG-N10-GC) (Barrios et al., 1999) are absolutely conserved at three of the four predicted promoters in the cluster (Fig. 4A).
Transcription by σ54-containing RNA polymerase requires an interaction with an activator protein, often a response regulator whose ability to promote transcription is limited to specific environmental or cellular conditions. Sequence analysis of the syp cluster revealed a putative σ54-dependent response regulator, SypG. This putative regulator was highly conserved in V. parahaemolyticus, V. vulnificus and V. cholerae (Table S1). Using Multiple Em for Motif Elicitation (MEME) analysis (see Experimental procedures), we identified a conserved 22 bp sequence in each of the four intergenic regions, upstream of the putative σ54 binding site (Fig. 4A). This sequence could potentially serve as a binding site for an activator. Both the putative σ54 binding sites and 22 bp sequences also are conserved in V. parahaemolyticus and V. vulnificus (data not shown).
To investigate the transcriptional control of the syp cluster, we used as reporter strains two of the Tn mutants, sypD and sypN, each of which contained a promoterless lacZ gene oriented correctly for syp-dependent transcription. Surprisingly, we observed no β-galactosidase activity from these strains under a variety of media and growth conditions (data not shown). We therefore hypothesized that SypG, in concert with σ54, might control transcription of the syp genes in V. fischeri. Although response regulators typically depend on activation (by phosphorylation) from cognate sensor kinases, this requirement often can be overridden by multicopy expression of the response regulator. Thus, to investigate our hypothesis, we cloned a wild-type copy of sypG under the control of the lac promoter on a multicopy plasmid. We then introduced the resulting construct, pEAH40, into the Tn10lacZ reporter strains and observed the formation of blue colonies on X-gal-containing media, suggesting that SypG could induce syp transcription. We quantified this induction by measuring β-galactosidase activity, and found that multicopy expression of sypG increased the activity 37- to 70-fold over that of the vector control (Fig. 5). To determine whether the ability of SypG to induce syp transcription required the activity of σ54, as predicted, we constructed double mutants (containing a mutation in rpoN and the lacZ reporter in either sypD or sypN). Indeed, a mutation in rpoN abolished the SypG-dependent activation of syp transcription (Fig. 5). These data suggest that SypG activates syp cluster expression in a σ54-dependent manner.
Finally, to provide further support for the use of the identified σ54 binding sites, we determined the transcriptional start sites for sypA, sypI and sypM using primer extension experiments. Briefly, mRNA, isolated from wild-type cells containing multicopy sypG, was hybridized to a radiolabelled primer complementary to the gene of interest and subjected to reverse transcription. The resulting cDNA product was compared with a sequencing ladder (Fig. 4B). This analysis revealed two sypA-specific transcripts, with the primary band located 14 bp downstream from the putative σ54 binding site. Similarly, a sypI transcript initiated 12 bp downstream of the putative σ54 binding site. Finally, one of the sypM transcripts initiated at 13 bp downstream of the binding site. As these transcripts initiated within the range of start sites (8–16 bp) typical for σ54-dependent transcription (Barrios et al., 1999), these data further support a role for σ54 in promoting transcription of the syp genes.
Formation of biofilms under syp-inducing conditions
In nature, bacteria often exist in biofilms in which bacterial cells are embedded within a matrix of extracellular polysaccharides. Due to the putative functions of the syp genes in polysaccharide biosynthesis, we hypothesized that the syp cluster may function to alter the surface properties of V. fischeri to enhance biofilm formation. We therefore asked whether syp cluster members contributed to biofilm formation by V. fischeri. Not surprisingly, given the lack of syp transcription in culture, we noticed little difference in biofilm formation between the syp mutant strains and their parent when grown in static culture (data not shown). However, the addition of the sypG multicopy plasmid caused a significant, 3.5-fold, increase (P = 0.0041) in the ability of the parent strain, ESR1, to produce a biofilm in static culture (Fig. 6A and C). The sypG-expressing plasmid caused a similar increase in biofilm formation in all of the syp mutants except sypC, suggesting that sypC plays role in enhancing biofilm formation. Finally, consistent with our β-galactosidase results, a mutation in rpoN eliminated the sypG-dependent induction of biofilm formation, further supporting the roles of these two regulators in controlling the syp cluster.
During the course of our β-galactosidase experiments, we observed that cells containing multicopy sypG, when grown with shaking, produced a substantial ring of cells around the test tube. This large biofilm ring seemed inconsistent with the relatively modest increase in biofilm production we observed using the static culture assay. We therefore examined biofilms formed by these strains during growth with shaking. We found that these conditions dramatically increased the magnitude of the sypG- and rpoN-dependent biofilm formation (Fig. 6B and D). The parent strain containing multicopy sypG exhibited a > 30-fold increase relative to the vector control and a > 12-fold increase relative to that achieved under static conditions. This increase in stainable material stemmed from an increase in adherent cells: we observed a 7.8-fold increase of adherent cells when they contained the sypG plasmid, relative to the vector control. Among syp mutant strains, biofilms of the sypC, and to a lesser extent, sypJ, mutants were substantially reduced relative to their parent strain. In contrast, mutants defective for sypD or sypN exhibited biofilm formation similar to that of the parent strain. Finally, the formation of biofilms under these conditions depended on a functional copy of rpoN. These data further demonstrate the dependence of syp biofilm formation on SypC, SypG and σ54, and provide an in vitro function for the syp cluster members.
Identification of a novel cluster that is required for symbiotic colonization
In this study, we identified four Tn mutants of V. fischeri that exhibited severe defects in initiating symbiotic colonization of E. scolopes. The Tns in the four strains mapped to four genes within a cluster of genes, termed syp, that encodes proteins with putative functions in LPS or capsule biosynthesis. Our investigations of these mutants in culture revealed no defects in traits known to be important for symbiotic colonization, including motility and luminescence. Insertional mutations in five other genes within this locus similarly caused severe initiation defects, supporting the hypothesis that this locus represents a novel gene cluster that functions in symbiotic initiation.
Transcriptional control of the syp genes
Two of the syp mutants contained insertions in which a promoterless lacZ reporter was oriented correctly for syp-dependent transcription. Surprisingly, we observed no β-galactosidase activity under any of a variety of media and growth conditions. Through a bioinformatics approach, we identified both SypG, a putative σ54-dependent transcriptional activator encoded by the syp cluster, and potential binding sites for RNA polymerase carrying σ54. Multicopy expression of sypG induced transcription of the two lacZ reporters (in sypD and sypN) by 37- to 70-fold. This induction depended on σ54, as mutations in rpoN abolished β-galactosidase activity by these strains. Similarly, multicopy expression of sypG in a wild-type strain resulted in sypA, sypI and sypM transcripts that initiated between 12 and 14 based downstream of putative σ54 consensus sequences, a common range of initiation sites for such promoters (Barrios et al., 1999). In addition, in preliminary experiments, we have fused the putative sypP promoter region to the lacZ reporter, introduced it in single copy in the chromosome, and similarly observed a sypG-dependent induction of transcription (E.A. Hussa and K.L. Visick, unpubl. data). Together, these data support the existence of multiple operons within syp that depend on both SypG and σ54.
Control by σ54-containing RNA polymerase is an effective method for keeping transcription turned off: because the σ54-bound RNA polymerase fails to promote transcription without the aid of a regulatory protein, transcription of target genes is largely off until the appropriate, activated transcription factor binds and interacts with the polymerase. For syp, this presumably occurs through the binding of the two-component response regulator, SypG, to its target, perhaps the conserved 22 bp sequence we identified in four intergenic regions. We postulate that a colonization signal transduced by a sensor kinase activates SypG and turns on syp transcription to permit symbiotic initiation. A sensitive assay for gene expression in the host, such as recombination-based in vivo expression technology (RIVET) (Camilli and Mekalanos, 1995; Angelichio and Camilli, 2002), will be needed to determine whether the syp genes are induced early in symbiotic colonization. Consistent with the predicted importance of sypG in colonization, our preliminary results suggest that a sypG mutant initiates colonization poorly (E.A. Hussa, T. O’Shea and K.L. Visick, unpubl. data).
If SypG functions as a response regulator that is activated during symbiotic colonization, then what is the identity of its cognate sensor kinase? Sensor kinases typically detect a particular environmental signal and respond by autophosphorylating and subsequently serving as a phospho-donor to a response regulator. Just upstream of sypG are two additional putative two-component regulator genes: sypE, which likely encodes a response regulator, and sypF, which encodes a putative sensor kinase. The SypF sensor could potentially function with SypG, or, alternatively, with both SypE and SypG in a complex phosphorelay such as that which controls capsule biosynthesis in Escherichia coli (the rcs system) (Takeda et al., 2001). Another possibility is the sensor kinase, RscS, that we have previously identified as essential for symbiotic initiation (Visick and Skoufos, 2001). The rscS gene is not located within the syp cluster, and to date, we have not identified its cognate response regulator nor any members of its regulon. Clear identification of the cognate sensor kinase for SypG may require simulation of the host conditions that induce syp transcription, or the construction of constitutively active sensor kinase genes.
Role for SypG in promoting biofilm formation
We found that multicopy expression of sypG substantially enhanced biofilm formation by V. fischeri. Biofilms, dynamic mixtures of cells encased within an extracellular matrix, are being intensively studied in part due to their importance in medicine (Parsek and Singh, 2003; Parsek and Fuqua, 2004). Work in other organisms has shown that the biofilm structures formed by a single organism may vary substantially depending on the exact environmental condition, perhaps due to the environment-specific expression of factors that contribute to the extracellular matrix. For example, experiments in Pseudomonas aeruginosa indicate that decreased oxygen availability is a factor in robust biofilm formation (Yoon et al., 2002). Our studies in V. fischeri revealed that multicopy sypG expression enhanced biofilm formation to a greater extent when the cells were grown with shaking, relative to static growth. We thus hypothesize that aeration plays a role in biofilm formation under these conditions. Alternatively, it is possible that another difference between the two conditions, such as fluid distribution, which plays a role in Bordetella biofilms (Mishra et al., 2005), or the relative amounts of cell growth, accounts for the differences observed. Future work will address these possibilities.
Multicopy sypG expression also substantially increased biofilm formation by mutants defective for sypD, sypJ and sypN, but not sypC. Multiple explanations are possible. For example, SypG could control a locus, in addition to syp, that promotes biofilm formation. Alternatively, SypG-mediated overexpression of sypC, which encodes a protein with similarity to those involved in capsule export (KpsD, OtnA and Wza) (Bik et al., 1996; Arrecubieta et al., 2001; Nesper et al., 2003), could result in inappropriate export of a biofilm-promoting factor. It is also possible that the syp genes encode redundant functions with respect to biofilm formation, and thus multiple mutations are necessary to counteract the effect of multicopy sypG. Regardless of the explanation, these data make it clear that, in contrast to what has been assumed from this and previous studies of biofilm formation by wild-type strains (Wolfe et al., 2004), V. fischeri is not a poor biofilm former; rather, like pathogenic bacteria, this symbiont can produce substantial biofilms, under the appropriate conditions.
Function of the syp gene cluster
Our biofilm results suggest that, under conditions in which the syp genes are expressed, V. fischeri cells are producing an extracellular matrix that enables them to adhere to a glass surface and, perhaps, to a host tissue. The tentative functional assignments for Syp proteins in capsule, exopolysaccharide or LPS synthesis, macromolecules that contribute to biofilm formation by other bacteria, are consistent with that hypothesis. Our studies to date have yielded no differences in the overall carbohydrate levels produced by multicopy sypG or vector containing cells (E.S. Yip and K.L. Visick, unpubl. data). This result is similar to recent studies of P. aeruginosa psl (polysaccharide synthesis locus) – a locus that, like syp, contains multiple putative glycosyltransferase genes (Frideman and Kolter, 2004; Jackson et al., 2004; Matsukawa and Greenberg, 2004): a comparison of matrix carbohydrate levels in a psl mutant relative to the wild-type strain revealed similar overall yields (Matsukawa and Greenberg, 2004). Mutations in the psl locus result in an altered carbohydrate composition; this could potentially be true for syp as well. Recently, we have identified another phenotype consistent with potential exopolysaccharide production: prolonged static growth of V. fischeri with multicopy sypG results in the formation of a pellicle at the air/liquid interface, a phenotype that we do not observe in cultures of wild-type or vector-control cells (E.S. Yip and K.L. Visick, unpublished). We anticipate that investigation of this pellicle will provide insights into the nature of the product of the syp cluster.
Homologues of syp genes in related Vibrio species
Our analysis of the syp gene cluster of V. fischeri revealed similar clusters, with the same gene order, in the related human pathogens, V. parahaemolyticus and V. vulnificus. Neither of these two species is competent to colonize E. scolopes, although at least one strain of V. parahaemolyticus can adhere to the surface of the squid light organ (Nyholm et al., 2000). We speculate that the syp cluster may be used by all of these Vibrio species in the early stages of colonization with their hosts, either marine invertebrates, or, potentially, human hosts. If so, then comparative studies of the three loci will aid our understanding of such bacteria–host interactions. Although there is high overall conservation, a number of individual genes exhibit significant sequence divergence; these differences could potentially account for the observed differences in host range. Further study of the role and regulation of the syp cluster may enhance our knowledge not only of the communication between V. fischeri and its symbiotic host but also of the signals and responses leading to initiation of pathogenic associations.
Bacterial strains and media
Vibrio fischeri strains used in this study are listed in Table 2. Strain ES114 (Boettcher and Ruby, 1990) was used as the wild-type strain, and ESR1, a rifampin-resistant (RifR) derivative of ES114 (Graf et al., 1994), was used as the parent strain in Tn mutagenesis and mutant analyses. E. coli strains DH5α and CC118λpir (Herrero et al., 1990) were used as hosts for cloning and conjugation. V. fischeri strains were grown on SWT medium (0.5% tryptone, 0.3% yeast extract, 210 mM NaCl, 35 mM MgSO4, 7 mM CaCl2 and 7 mM KCl) for all colonization experiments and Hepes-minimal medium (MM) (Ruby and Nealson, 1977) supplemented with 0.2% glucose and 0.3% Casamino acids for β-galactosidase and biofilm experiments. For motility assays, V. fischeri cells were spotted on tryptone-based soft agar medium containing 0.25% Bacto-Agar (Difco, Detroit, MI). Antibiotics were added to media, where appropriate, to the following final concentrations: ampicillin (Amp), 100 µg ml−1 for E. coli; chloramphenicol (Cm), 25 µg ml−1 for E. coli and 2.5 µg ml−1 for V. fischeri; erythromycin (Em), 150 µg ml−1 for E. coli and 5 µg ml−1 for V. fischeri; tetracycline (Tet), 15 µg ml−1 for E. coli and 30 µg ml−1 for V. fischeri.
Table 2. Strains and plasmids used or constructed in this study.
pCR2.1-TOPO + 1.8 kb fragment of VFA1019-sypA intergenic DNA
pCR2.1-TOPO + 294 bp fragment of sypH-sypI intergenic DNA
pCR2.1-TOPO + 398 bp fragment of sypL-sypM intergenic DNA
1.4 kb BamHI/SacI fragment from pKV124 + 1.6 kb SspI/XmnI fragment from pEVS122; CmR, oriR6K, oriT, lacZ
pEVS122 (EcoRI) + 386 bp internal fragment of sypL; EmR
pEVS122 (EcoRI) + 406 bp internal fragment of sypN; EmR
pEVS122 (EcoRI) + 375 bp internal fragment of sypO; EmR
pEVS122 (EcoRI) + 364 bp internal fragment of sypQ; EmR
pEVS122 (EcoRI) + 341 bp internal fragment of sypR; EmR
Plasmid and mutant constructions
Transposon mutagenesis was performed using the mini-Tn10lacZ delivery plasmid pKV124 (Visick and Skoufos, 2001). Cloning of the Tn and flanking DNA was carried out as described previously (Visick and Skoufos, 2001), using the origin of replication and Cm-resistance gene contained within the Tn. All plasmids were constructed using standard molecular biology techniques, with restriction and modifying enzymes from New England Biolabs (Beverly, MA) or Promega (Madison, WI). To construct syp and rpoN mutants of V. fischeri, we first amplified, by polymerase chain reaction (PCR), an internal fragment of specific genes using DNA oligonucleotides (Table S2) from MWG Biotech (High Point, NC). The PCR fragments were then cloned into cloning vector pCR2.1-TOPO (Invitrogen, Carlsbad, CA) and subcloned into one of two ‘suicide’ plasmids [pEVS122 (Dunn et al., 2005) or pKV194], which do not replicate but integrate into genomic DNA via homologous recombination between the inserted V. fischeri sequence and the chromosome. Triparental matings were performed and potential insertional mutants were selected on the appropriate antibiotic-containing plates.
To screen for Tn mutants defective in colonizing the light organ, V. fischeri mutant strains were inoculated into artificial seawater (Instant Ocean; Aquarium Systems, Mentor, OH) containing individual newly hatched juvenile squid. After 3 h incubation, juvenile squid were washed and transferred to symbiont-free seawater. Colonization of the light organ was monitored by measuring luminescence over the course of 18 h in a scintillation counter as described previously (Ruby, 1996). To quantify the number of bacteria present in the light organ, juvenile squid were homogenized, serially diluted and plated on SWT agar. The limit of detection is 14 bacterial cells per squid.
ESR1 and the Tn mutants were diluted 1:100 from overnight cultures and grown in SWT/IO medium [0.5% tryptone, 0.3% yeast extract, 0.3% glycerol and 43 g l−1 Instant Ocean (Aquarium Systems) (Stabb et al., 2004)]. Samples were taken for luminescence and optical density measurements over the course of 4 h. A TD-20/20 luminometer (Turner Designs, Sunnyvale, CA) was used to determine the level of bioluminescence.
To examine the outer membrane protein (OMP) profiles of the mutant strains, cellular protein extracts were obtained as described previously (Aeckersberg et al., 2001). Upon extraction, OMP-enriched fractions were separated by electrophoresis on a 8% SDS-polyacrylamide gel and visualized by staining with either Coomassie brilliant blue (Sigma, St Louis, MO) or SYPRO Red protein gel stain (Molecular Probes, Eugene, OR).
Southern blot analysis
We analysed both the original Tn mutants and our constructed insertional mutants by Southern blot experiments. Chromosomal DNA was extracted, digested and separated by a 0.6% agarose gel, transferred onto a nylon membrane (Hybond XL; Amersham-Pharmacia Biotech, Piscataway, NJ) and cross-linked using UV light. Detection of DNA fragments was performed by using the Boehringer Mannheim DIG DNA labelling kit (Roche Molecular Biochemicals, Indianapolis, IN) as previously described (Visick and Skoufos, 2001) using either Tn or vector probes. We found that all Tn mutant strains carried a single Tn insertion. The insertion in sypC also contained the delivery vector integrated into the chromosome at the site of the insertion. Similarly, for the ES114-derived syp insertional mutants, the resulting banding pattern showed that each contained a single insertion of suicide plasmid vector at the appropriate location in the chromosome.
Twenty-five millilitres of cultures of strain ES114 containing either pEAH40 or pKV69 were grown in MM for 24 h at 28°C. The cells were lysed in GITCN lysis buffer (Totten and Lory, 1990), and mRNA was isolated by ultracentrifugation through a cesium chloride gradient. Oligonucleotide primers (VFA1020PER, VFA1028PER and VFA1032PER; see Table S2) complementary to sypA, sypI and sypM sequences were radiolabelled via T4 polynucleotide kinase (USB) and [γ32-P]-ATP (Amersham-Pharmacia Biotech, Piscataway, NJ). The labelled primers were hybridized to 32 µg of mRNA and incubated with Moloney murine leukaemia virus (MMLV) reverse transcriptase (Stratagene, LaJolla, CA) and nucleotides per the manufacturer's instructions. Primer extension products were visualized upon separation on a 6% polyacrylamide gel. To determine the start of transcription, the extension product was compared with a sequencing ladder generated from pKV189 (sypA), pKV190 (sypI) or pKV191 (sypM) using Sequenase Version 2.0 DNA Sequencing Kit (USB) with the same primer used for the primer extension.
To measure the expression of syp genes, Tn reporter strains which carried either pKV69 or pEAH40 were grown in MM with shaking for 24 h; β-galactosidase activity was measured as described (Miller, 1972). Total protein concentrations were determined by the method of Lowry et al. (1951). All experiments were performed in triplicate.
To quantify biofilm formation, we used methods modified from Djordjevic et al. (2002). Briefly, all V. fischeri strains were grown in triplicate in MM. Overnight cultures were diluted to an optical density at 600 nm (OD600) ∼0.1 and then incubated in test tubes with and without shaking for 24 and 30 h respectively. We quantified biofilm by adding 1 ml of 1% crystal violet to each culture. After 30 min of staining, all liquid was removed from the tubes and each was rinsed 10 times with dH2O. Tubes were then dried by aspiration and 4 ml of 100% ethanol was added to each for destaining. Samples were destained for 1 h with vortexing every 15 min. Each crystal violet-containing sample was diluted, if necessary, and quantified by measuring its absorbance at OD600. In parallel, samples were grown over the same time-course, vortexed and the optical density was determined to assess growth of the strains. Under these experimental conditions, strains containing multicopy sypG exhibited a decreased growth yield (and increased biofilm formation) relative to vector controls. Statistical analysis was performed using the Student's t-test.
To quantify viable cells within biofilms we inoculated cells as described above to generate biofilms in triplicate. After incubation, we removed the liquid from the culture without staining and rinsed tubes with 70% artificial seawater (Instant Ocean; Aquarium Systems, Mentor, OH) 10 times each. We then added 3 ml of 70% artificial seawater to each tube and 1 mm glass beads. After 15 min of incubation with vortexing every 5 min, we plated dilutions of biofilm material and calculated viable cell recovery.
We would like to thank Eric Stabb for generous donation of plasmid vectors in advance of publication and Therese O’Shea for the construction of syp mutant plasmids. We also thank Kati Geszvain, Therese O’Shea, Jon Visick and Alan Wolfe for their insightful experimental suggestions and critical review of the article. This work was supported by NIH Grant GM59690 awarded to K.L.V.