Vibrio fischeri Amidase Activity Is Required for Normal Cell Division, Motility, and Symbiotic Competence

Peptidoglycan (PG) is a critical microbe-associated molecular pattern (MAMP) that is sloughed by cells of V. fischeri during symbiotic colonization of squid. Specifically, this process induces significant remodeling of a specialized symbiotic light organ within the squid mantle cavity.

ABSTRACT

N-Acetylmuramoyl-l-alanine amidases are periplasmic hydrolases that cleave the amide bond between N-acetylmuramic acid and alanine in peptidoglycan (PG). Unlike many Gram-negative bacteria that encode redundant periplasmic amidases, Vibrio fischeri appears to encode a single protein that is homologous to AmiB of Vibrio cholerae. We screened a V. fischeri transposon mutant library for strains altered in biofilm production and discovered a biofilm-overproducing strain with an insertion in amiB (VF_2326). Further characterization of biofilm enhancement suggested that this phenotype was due to the overproduction of cellulose, and it was dependent on the bcsA cellulose synthase. Additionally, the amiB mutant was nonmotile, perhaps due to defects in its ability to septate during division. The amidase mutant was unable to compete with the wild type for the colonization of V. fischeri’s symbiotic host, the squid Euprymna scolopes. In single-strain inoculations, host squid inoculated with the mutant eventually became colonized but with a much lower efficiency than in squid inoculated with the wild type. This observation was consistent with the pleiotropic effects of the amiB mutation and led us to speculate that motile suppressors of the amiB mutant were responsible for the partially restored colonization. In culture, motile suppressor mutants carried point mutations in a single gene (VF_1477), resulting in a partial restoration of wild-type motility. In addition, these point mutations reversed the effect of the amiB mutation on cellulosic biofilm production. These data are consistent with V. fischeri AmiB possessing amidase activity; they also suggest that AmiB suppresses cellulosic biofilm formation but promotes successful host colonization.

IMPORTANCE Peptidoglycan (PG) is a critical microbe-associated molecular pattern (MAMP) that is sloughed by cells of V. fischeri during symbiotic colonization of squid. Specifically, this process induces significant remodeling of a specialized symbiotic light organ within the squid mantle cavity. This phenomenon is reminiscent of the loss of ciliated epithelium in patients with whooping cough due to the production of PG monomers by Bordetella pertussis. Furthermore, PG processing machinery can influence susceptibility to antimicrobials. In this study, we report roles for the V. fischeri PG amidase AmiB, including the beneficial colonization of squid, underscoring the urgency to more deeply understand PG processing machinery and the downstream consequences of their activities.

INTRODUCTION

Vibrio fischeri is the exclusive colonist of the squid Euprymna scolopes (1). An array of V. fischeri characteristics facilitate this unique ability, such as bioluminescence, motility, and biofilm formation (2–4). In this study, we identified an additional requirement for colonization by identifying and evaluating a biofilm-overproducing strain of V. fischeri carrying a transposon insertion in a gene encoding an N-acetylmuramoyl-l-alanine amidase (AmiB) homolog.

Amidases are Zn²⁺-cofactored, often periplasmic enzymes that comprise one of many types of peptidoglycan (PG)-hydrolyzing enzymes in bacteria (5). Their primary activity is to cleave the amide bond between the peptide side chain and N-acetylmuramic acid. Thus, peptidoglycan hydrolysis by N-acetylmuramoyl-l-alanine amidases is a crucial step in bacterial cell division (6). Bacteria tightly regulate this process in order to encourage cytokinesis and avoid cell lysis. Cytokinesis is achieved by a collection of proteins referred to as the “divisome,” which includes amidases, Z-ring-associated proteins such as FtsA, and amidase regulators (e.g., EnvC) that are believed to recruit amidases to the septum (7, 8). An additional layer of amidase regulation in Salmonella occurs through the activation of the two-component envelope stress response proteins CpxA and CpxR; CpxR then binds to promoters of the redundant amidases amiA and amiC to enhance expression (9). Similarly, in Pseudomonas aeruginosa, the cell division defects displayed by an amiB mutant are suppressed by the activation of the Cpx-like envelope stress response (10).

In Escherichia coli, four paralogous genes, amiA, amiB, amiC, and amiD, appear to act in a hierarchical manner to regulate peptidoglycan cleavage (6, 7). P. aeruginosa also possesses these amidase homologs but apparently requires only amiB and amiC for proper cell division (11). In support of its function in peptidoglycan cleavage, AmiB in P. aeruginosa possesses C-terminal LysM peptidoglycan-binding domains (10). In contrast, Vibrio species such as Vibrio cholerae and V. anguillarum possess a single divisome-related amidase activity encoded by amiB (12, 13). Rashid et al. reported that an amiB mutant of V. cholerae was defective in cell division and growth. Consistent with these pleiotropic effects, the amiB mutant was unable to compete with the wild type for intestinal colonization of mice. Notably, both EnvC and NlpD can promote AmiB activation in V. cholerae; thus, only when both activators are removed will cells of V. cholerae display the cell division phenotype of the amiB mutant (14).

The possibility that the influence of AmiB on host colonization extends beyond peptidoglycan cleavage in cell division presents an intriguing area of study. For example, there is abundant evidence that amidase activity influences biofilm formation in diverse bacteria. Núñez et al. reported an Azotobacter amidase mutant with defects in alginate production and suggested a link between alginate production and cell wall recycling (15). Yoshida and Kuramitsu identified a lytR mutant of the etiological agent of dental caries, Streptococcus mutans, that was deficient in biofilm production. Notably, LytR regulates genes required for amidase expression in S. mutans (16). Similarly, in V. cholerae, AmiB appears to control the switch between smooth colonies and the exopolysaccharide-producing rugose phenotype (12). Biofilm formation by cells of V. fischeri facilitates the colonization of squid light-organ tissue (4). In V. fischeri, biofilms are constructed by one of two sets of polysaccharide gene loci, syp and bcs. The importance of Syp-type biofilm in the ability of V. fischeri to colonize squid is established (4); however, the importance of bcs-encoded (cellulosic) biofilm in squid colonization is unknown. Bassis and Visick suggested that cellulosic biofilm formation might be essential in the marine environment (17). Similarly, while the regulation and role of syp-dependent biofilm formation are well characterized, far less is known about bcs. Within the bcs locus, bcsA, bscB, and bcsZ encode cellulose synthase and endoglucanase activities that are needed for cellulosic biofilm formation. Cyclic diguanylate (c-di-GMP) is required to activate the cellulose synthase, while inhibition is achieved by the cyclic diguanylate-digesting phosphodiesterase (PDE) BinA (17, 18). Phenotypically, the loss of PDE activity results in cells that overproduce cellulose and form red colonies when grown on differential medium containing Congo red dye. Taken together, cellulosic biofilm formation is greatly influenced by c-di-GMP and perhaps some as-yet-undescribed environmental conditions.

In this study, we confirm several traditional roles for AmiB in V. fischeri (14, 19) and describe novel roles in cellulosic biofilm regulation and squid colonization. Furthermore, we identify and characterize suppressor mutants that arose in an amiB mutant population, uncovering a function for a previously unknown gene, VF_1477.

RESULTS

Disruption of amiB results in increased biofilm production.

A screen of 2,000 V. fischeri mutants resulted in 78 with ≥2-fold increased or ≤2-fold reduced biofilm formation as indicated by crystal violet staining. Of these mutants, one of the more substantially altered was strain C10, which produced excess biofilm. Sequence analysis revealed that C10 contained an insertion in VF_2326. This gene encoded an amino acid sequence sharing all major functional domains and approximately 62% identity (79% similarity) across nearly the entire sequence with V. cholerae AmiB (Fig. 1A). Based on this homology and our evidence described below, we conclude that VF_2326 is the V. fischeri amiB homolog. Quantitation of crystal violet staining of biofilms produced by C10 showed that this mutant had an ∼18-fold increase in staining relative to the wild-type strain. Similarly, colonies of C10 lacked the smooth-colony phenotype of its ES114 parent but rather exhibited noticeably bumpy architecture relative to ES114 (Fig. 1B). The expression of amiB in strain C10 restored biofilm formation to wild-type levels (Fig. 1C), confirming that the disruption of amiB was responsible for the increased biofilm phenotype.

FIG 1.

The amidase is a V. cholerae homolog that controls biofilm production. (A) V. fischeri AmiB shares major functional domains with the V. cholerae homolog. AMIN, domain. (B) Comparison of crystal violet-stained biofilms and colony architecture between the amiB mutant (C10) and the V. fischeri wild type (ES114). (C) Comparison of biofilms in wild-type and C10 strains carrying pAmiB with or without amiB induction by IPTG. (D) Comparison of biofilms in C10 in which the Syp biofilm regulator RscS is overexpressed on pRscS or in combination with mutations in either gene required for syp or bcs biofilm formation. Strains tested include C10, C10 carrying a vector (pKV69), C10/pRscS (pKG11), C10 ΔbcsA (KV8442), and C10 ΔsypQ (KV8443). The specific biofilm density is indicated above each tube and represents the average (± standard error) for three separate trials.

Because biofilm formation by V. fischeri depends on two distinct polysaccharides, symbiosis polysaccharide (SYP) and cellulose (bcs) (20), we next asked which one(s) was affected by amiB disruption (Fig. 1D). To evaluate SYP-dependent biofilm formation, we (i) overexpressed the positive regulator RscS and (ii) disrupted the structural gene sypQ. Neither manipulation substantially affected biofilm formation by C10. In contrast, when we disrupted bcsA, one of the genes for cellulose synthase, the distinctive amiB mutant phenotype was abolished. These data suggest that AmiB inhibits cellulose-dependent biofilm formation by V. fischeri.

Confirmation that excess biofilm in the amiB mutant depends on bcsA.

A more typical assay for cellulose-dependent biofilm formation is Congo red binding (17, 21). Therefore, we next assessed Congo red binding by the amiB mutant. When strain C10 was streaked onto Congo red-containing medium (CRM), the resulting cell growth was stained a dark hue of red relative to ES114, supporting the conclusion that the biofilm of C10 is primarily composed of cellulose (Fig. 2A). Furthermore, this increased staining was complemented by expressing amiB in trans in C10, restoring the wild-type phenotype. Consistent with the crystal violet staining results, Congo red staining was not disrupted when the amidase mutation was combined with sypQ (Fig. 2B). Similarly, it was also not disrupted by mutation of cpxR, which in Salmonella encodes a protein that promotes the transcription of the amidase-encoding genes amiA and amiC (9). However, the deletion of bcsA in the amiB mutant background resulted in a loss of Congo red staining, supporting the conclusion that the increased biofilm phenotype of the amiB mutant was due to cellulose. Consistent with that conclusion, the expression of bcsA in trans restored the excess cellulosic biofilm phenotype to the amiB-bcsA mutant.

FIG 2.

The amiB mutant overproduces cellulosic biofilm. (A) Congo red staining reveals that the biofilm formed by the amiB mutant is composed of cellulose. Strains tested include ES114, C10 (amiB::mini-Tn), and C10/pAmiB grown either with or without added IPTG to induce amiB expression. (B) Confirmation that increased cellulosic biofilm in the amiB mutant requires bcsA and not sypQ or cpxR. Strains tested include ES114, C10, C10 ΔsypQ (KV8443), C10 ΔcpxR (KV8614), C10 ΔbcsA (KV8442), and C10 ΔbcsA (KV8442)/pBcsA grown either with or without added IPTG to induce bcsA expression. Where IPTG was required (e.g., expression from pAmiB and pBcsA), IPTG was added to broth containing cells prior to streaking onto the plates. Shown are typical results from three separate trial plates.

Disruption of amiB results in cells with altered morphology and arrangement.

Others have reported cells of planktonic amiB mutants displaying an aberrant morphology (10, 14). Similarly, planktonic cells of the V. fischeri amidase mutant (C10) formed long chains, in contrast to the separated cells of the wild-type strain (Fig. 3A and B). Notably, cells of the wild type ranged from 1 to 2 μm in length, while chained cells of the mutant ranged from 4 to 5 μm (Fig. 3B), and the depth of the invagination is sufficient to clearly see each chained cell individually. Despite this phenotype, the growth curves of the amiB mutant showed no substantial defect relative to the wild-type parent (data not shown). The chaining phenomenon was also evident in biofilm-associated C10 cells (Fig. 3F to H). In addition, many of the biofilm-associated C10 cells appeared to produce “blebs” of either cellular contents or secreted polysaccharide (Fig. 3F and G, arrows). Combining the amiB mutation with deletions in either cpxR or bcsA did not affect the cell arrangement relative to strain C10 (Fig. 3C, D, G, and H); however, blebs were not evident in combination with the bcsA mutation (Fig. 3D and H). These data indicate that the increased biofilm of the C10 mutant is not due to chaining per se but is due to some other consequence of the loss of amidase function.

FIG 3.

The amiB mutant is defective in cell arrangement. Comparisons of crystal violet-stained cells of planktonic (A to D) and biofilm-associated (E to H) V. fischeri strains are shown. Strains tested include ES114, C10, C10 ΔcpxR (KV8614), and C10 ΔbcsA (KV8442). The arrows in panels F and G point to a representative bleb. The removal of bcsA, but not cpxR, appears to eliminate blebbing (G and H). Each panel shows a typical result from 10 randomly viewed fields in three separate trials. Bars = 5 μm.

Timing of AmiB activity.

To determine how quickly AmiB can exert its activity, we evaluated the ability of the C10 mutant chains to be separated into individual cells following the induction of AmiB production. We found that when amiB expression was induced from a plasmid, the chaining phenotype of strain C10 was largely lost within 15 min (Fig. 4). These data indicate that (i) AmiB activity (and, thus, likely the amiB complementation plasmid) seems to be distributed throughout the length of the chained cells and (ii) despite the long length of the chained cells, with potentially physiologically distinct properties, AmiB could act quickly to separate the individual cells.

FIG 4.

Kinetics of the loss-of-chaining phenotype following IPTG induction of the complemented amiB mutant. Each panel shows a typical result from three trials in which 10 random fields of crystal violet-stained planktonic cells were viewed at the indicated times following IPTG induction. Strains tested include ES114, C10 (amiB::mini-Tn), and C10/pAmiB grown either with or without added IPTG to induce amiB expression.

AmiB activity is regulated by two LytM-containing regulators.

In V. cholerae, the activity of AmiB is regulated by two LytM (peptidase M23) family proteins, EnvC and NlpD (14). We thus wondered whether the same would be true for V. fischeri, which possesses homologs of EnvC and NlpD (VF_2351 and VF_2068) that each share ≥65% similarity and all major functional domains with their V. cholerae counterpart (Fig. 5A). As with V. cholerae, the inactivation of either of these regulators was not sufficient to abolish AmiB activity. The inactivation of both regulators, however, produced a chaining phenotype similar to that of the amiB mutant (Fig. 5B). Furthermore, the chaining phenotype of the envC nlpD double mutant could be suppressed by the overexpression of amiB in trans. Finally, the double mutant also produced colonies with a darker hue of red on Congo red medium, phenocopying the amiB mutant (data not shown). Together, these data suggest that the activity of AmiB is controlled by EnvC and NlpD.

FIG 5.

AmiB activity is regulated by EnvC and NlpD. (A) Both V. fischeri regulators (VF_2068 and VF_2351) share major functional domains with their respective V. cholerae homologs (VC0395_A0061 and VC0395_A2736). (B) Planktonic cells of each strain were stained with crystal violet to visualize the effects on cellular arrangement. Strains tested include ES114, C10 (amiB::mini-Tn), envC::mini-Tn5 (PFC), nlpD::mini-Tn5 (PFD), envC::mini-Tn5 ΔnlpD (PFCD), and envC::mini-Tn5 ΔnlpD (PFCD), induced to express amiB from pAmiB with IPTG. Each panel shows a typical result from 10 randomly viewed fields in three separate trials.

Disruption of amiB results in a squid colonization defect.

Given the cell-chaining and biofilm phenotypes of the amiB mutant, we asked if AmiB was important for symbiotic colonization. Notably, because turbidity measurements (i.e., optical density at 600 nm [OD₆₀₀]) did not correlate well with CFU per milliliter, we conducted parallel experiments in which sets of squid were exposed to a different dilution of cells. Indeed, in squid exposed to 1:1 inoculations of the amidase mutant (C10) and wild-type V. fischeri strain ES114, C10 was completely outcompeted by the wild type for squid colonization. In all three competition trials, 100% of isolates from squid homogenates were the wild type (Fig. 6A). However, in single-strain squid inoculation experiments, in which colonization efficiency was based on luminescence emission from squid, strain C10 gradually achieved a peak of an ∼60% colonization efficiency by 48 h, whereas the wild type achieved 100% colonization by 24 h and maintained this level of colonization for at least 48 h (Fig. 6B).

FIG 6.

Squid colonization defect in the amiB mutant (C10). (A) Relative proportions of C10 to wild-type cells in each squid light organ 24 h after exposure to the bacteria, represented by closed black circles. (B) Kinetics of single-strain colonization of squid by strain C10. Each open circle represents bioluminescence emission by a squid colonized by strain C10 for up to 96 h. Each closed triangle represents bioluminescence emission by a squid colonized by the wild type (WT) (ES114) for up to 48 h. Black lines represent peak relative light units (RLU) of aposymbiotic squid. The number of colonized squid was determined by comparing RLU emissions between aposymbiotic squid and those exposed to strain C10. Results shown are typical for three independent trials.

Disruption of amiB results in loss of motility.

The severe squid colonization defect led us to wonder if the amiB mutant has a defect in motility, as motility is a critical colonization determinant. Indeed, when the amidase mutant (C10) was spotted onto semisolid Tris-buffered seawater with MgSO₄ (TBS-MgSO₄), the cells failed to migrate from the initial spot after a 12-h incubation, while the wild-type strain migrated across the plate in about 6 h (Fig. 7A). The expression of amiB in trans alleviated this motility defect, indicating that AmiB is indeed required for V. fischeri motility. This result is consistent with those reported previously (19). Intriguingly, by 24 h, “flares” of motility appeared from the original C10 spot (Fig. 7B), suggesting that a mutation(s) arose that could suppress the motility defect of C10. Indeed, when we collected cells from individual flares, purified them, and retested them for motility, we found that these strains (S1 to S4) had increased motility relative to C10, although they did not have the same rate of motility as ES114 (Fig. 7C). This intermediate motility phenotype, and the fact that they retained the erythromycin resistance of their C10 parent, suggested that the amiB mutation was intact and that these strains had acquired an additional mutation that promoted motility.

FIG 7.

The amiB mutant is defective in motility. (A) Comparison of wild-type and amiB mutant (C10) motilities. The motility defect of C10 can be complemented with pAmiB but only when IPTG is added to induce the expression of amiB. (B) Motility “flares” (arrow) emerging 12 h after spotting strain C10 on soft agar. (C) Motility phenotypes of 4 suppressor mutants isolated from the flares (S1 to S4). Results shown are typical for three independent trials.

Acquired motility in the amiB mutant is the result of suppressor mutations in VF_1477.

To determine the cause of the increased motility in strains S1 to S4, we performed whole-genome sequencing. Each of these strains had a mutation in VF_1477, a gene of unknown function, but the presence of Sel1 repeats and a DnaJ domain (E = 1.9e−8) suggests that it might be a chaperone in the stress response (22, 23). Notably, VF_1477 is conserved in other Vibrio species (Table 1). Two of these mutations were expected to disrupt VF_1477 function (S2 and S3), as they result in premature stop codons, while the consequences of the other two mutations (S1 and S4) are less clear. In mutant S1, a nonpolar valine was replaced with an acidic glutamic acid in a region of the protein with very low homology to an anti-sigma factor K domain (E value = 0.25). In mutant S4, an acidic glutamic acid replaced a nonpolar glycine in a region of the protein with high homology to a Sel1 repeat domain (E = 1.7e−10). To further understand the role of VF_1477, we deleted this gene in both the wild-type and amiB mutant backgrounds and assessed both motility and the cell-chaining phenotype (Fig. 8). Both the motility and cell arrangement of the VF_1477 mutant were identical to those of the wild type. Consistent with the findings for the suppressor mutants, the ΔamiB ΔVF_1477 double-deletion mutant exhibited partially increased motility relative to the amiB single mutant but did not increase motility to the level of the wild-type strain. This increase in motility could be complemented by the expression of VF_1477 in cis. In contrast to the results observed for motility, disruption of VF_1477 did not alter the microscopic appearance of the cells: the single mutant phenocopied the wild-type strain, while the double mutant phenocopied the amiB single mutant. Finally, neither disruption of nor complementation with VF_1477 altered the observed “blebbing” of the amiB mutant.

TABLE 1.

Genetic characterization of amiB suppressor mutants S1 to S4

Strain	Mutation (gene; genotype)
ES114	Wild type
C10	VF_2326 (amiB); amiB::mini-Tn5
S1	VF_2326 (amiB); amiB::mini-Tn5
	VF_1477; 443T>A (Val→Glu)
S2	VF_2326 (amiB); amiB::mini-Tn5
	VF_1477; 302,303delAC
S3	VF_2326 (amiB); amiB::mini-Tn5
	VF_1477; 79G>T (Glu→stop)
S4	VF_2326 (amiB); amiB::mini-Tn5
	VF_1477; 245G>A (Gly→Glu)

FIG 8.

Effects of amiB and VF_1477 on motility and cell arrangement. Disruption of VF_1477 partially restores the defect in motility (top) but not in cell arrangement in the amiB mutant (bottom). These results were determined using wild-type strain ES114 and mutants ΔVF_1477 (KV8711), ΔamiB ΔVF_1477 (KV8717), and ΔamiB ΔVF_1477 complemented with VF_1477 driven by a cellobiose-inducible promoter (KV8756). The plus symbol indicates the addition of 0.1% cellobiose. Results for motility are typical for three independent trials. Each panel shows a typical result from 10 randomly viewed fields.

VF_1477 plays a role in biofilm formation.

Because motility and biofilm formation are typically inversely controlled (18, 24), and because disruption of VF_1477 increases motility, we wondered if VF_1477 could control biofilm formation. Indeed, the loss of VF_1477 alone almost completely eliminated biofilm production; levels of crystal violet-stained material were ∼10-fold lower than those for the wild-type parent (Fig. 9). Consistent with the findings in Fig. 1, the loss of amiB resulted in a significant increase in biofilm formation; this increase was also abolished by disruption of VF_1477, resulting in wild-type-like levels of biofilm formation. The expression of VF_1477 in cis complemented the double mutant, restoring the levels of biofilm production to that of the amiB single mutant. These data thus provide new insight into this novel consequence of amiB disruption.

FIG 9.

Effects of amiB and VF_1477 disruption on biofilm formation. Biofilm formation of the following strains was assessed using the crystal violet assay: amiB::mini-Tn (C10), ΔVF_1477 (KV8711), ΔamiB ΔVF_1477 (KV8717), ΔamiB ΔVF_1477 complemented with VF_1477 driven by a cellobiose-inducible promoter (KV8756), and wild-type strain ES114. The plus symbol indicates the addition of 0.1% cellobiose. The specific biofilm density is indicated above each tube and represents the average (± standard error) for three replicates.

DISCUSSION

In this study, we investigated the pleiotropic effects of an amiB mutation in V. fischeri in order to better understand how amidases influence physiology and host colonization. Analysis of the V. fischeri genome revealed a single amidase gene encoding an AmiB homolog with a conserved signal peptide cleavage site, suggesting that it is secreted by the Sec system like the other AmiB homologs. Consistent with this discovery, the inactivation of amiB, or both genes encoding its positive regulators in V. fischeri (e.g., nlpD and envC), was sufficient to disrupt normal cell division. Furthermore, the inactivation of amiB virtually abolished motility. The defective motility displayed by the amiB mutant is consistent with its inability to compete with the wild type for squid colonization. However, despite the motility defect, we were surprised that the mutant could still colonize squid in single-strain inoculation studies albeit with a significant delay. The delayed colonization by the amiB mutant was consistent with the observation of motile suppressor mutants emanating from the point of inoculation on a motility plate. Notably, whole-genome sequence analysis of several motile suppressors revealed single nucleotide variations (SNVs) in a specific gene of unknown function (VF_1477). Our studies also revealed that the inactivation of amiB resulted in a profound increase in cellulosic, but not Syp, biofilm production. Indeed, an amiB bcsA double mutation completely abolished biofilm production.

Shared genomic synteny among the amiB homologs in V. fischeri, E. coli, and V. cholerae (data not shown) provided preliminary evidence that they might share functions. However, there are differences between these species as to the specific types of amidases present and their role in the cell. For example, E. coli possesses three periplasmic amidases (AmiA, AmiB, and AmiC), and P. aeruginosa possesses two (AmiA and AmiB), while Vibrio species possess only one (AmiB) (10, 13, 14, 25). Not surprisingly, in E. coli, no single amidase is responsible for the chaining phenotype (6). In fact, the periplasmic amidases have different substrate preferences. AmiC preferentially removes tetrapeptide fragments from cross-linked peptidoglycan fragments, while AmiA and AmiB appear to prefer to cleave pentapeptides (6). Adin et al. reported that V. fischeri possesses homologs of E. coli peptidoglycan synthesis and recycling pathways, with minor differences in peptidoglycan processing homologs (26). Our study confirmed that differences exist between the peptidoglycan processing machineries of E. coli and V. fischeri. Other studies reported that the lipoprotein NlpD and the protease EnvC activate amidase activity in E. coli and V. cholerae (14, 27, 28). In E. coli, EnvC activates AmiA and AmiB, while NlpD activates AmiC. However, in V. cholerae, both EnvC and NlpD are required to activate AmiB. V. fischeri possesses homologs of both LytM motif-containing proteins, EnvC and NlpD, and not unexpectedly, an envC nlpD double mutant formed chains like the amiB mutant. These findings suggest a conserved function for EnvC and NlpD in AmiB regulation in these two Vibrio species.

Our data support and extend the findings of Brennan et al. that the loss of AmiB in V. fischeri results in defective cell division (i.e., chaining) (19). Yakhnina et al. pointed out subtle differences between the chaining phenotypes of E. coli and P. aeruginosa, such as the depth and positioning of cell constrictions in chained cells. E. coli chains seemed to possess deeper, more frequent invaginations, resulting in more clearly defined, shorter cells, while cells in chains of P. aeruginosa appear longer, with shallower cleavage, resulting in chains that more closely resemble filaments (10). In this study, V. fischeri chains appear to be more similar to those of P. aeruginosa in cell length but more similar to E. coli in the depth of invaginations. Thus, while V. fischeri resembles P. aeruginosa with respect to the major purpose of AmiB in cell division, subtle differences in the chaining phenotype suggest that different accessory proteins might be required for proper cell division.

The loss of AmiB resulted in a mutant with increased biofilm production. Bassis and Visick identified a “bacterial cellulose biosynthesis” (bcs) cluster in V. fischeri and reported its importance in cellulosic biofilm production (17). When we deleted the cellulose synthase (bcsA) in the amiB mutant, the resulting double mutant failed to produce biofilm. In addition, cells of the double mutant ceased to produce a bleb that could be excess exopolysaccharide, similar to what was reported for Grimontia hollisae (29). These data suggested that the loss of AmiB perturbs the regulation of the bcs operon, resulting in excess biofilm formation. Previous work supports a potential role for amidases in biofilm formation in a wide range of bacteria (12, 15, 16). For example, in Azotobacter vinelandii, the loss of the ampDE operon, in which ampD encodes a cytosolic amidase, results in increased alginate biofilm production (15). Apparently, AmpD and AmpE negatively regulate algD, which encodes an alginate synthase. Interestingly, the ampDE mutant also forms chains, like the various amiB mutants. In many types of bacteria, cells switch between smooth and rugose (rough) colony morphologies (29–31); in the rugose phenotype, exopolysaccharide production is turned “on.” Rashid et al. screened a V. cholerae transposon insertion library for strains that were defective in the transition from a smooth to a rugose colony morphology (12). Those authors reported several mutants with transposon insertions in genes required for lipopolysaccharide (LPS) biosynthesis as well as genes within the Vibrio polysaccharide operon. Consistent with our findings, those authors also reported that an amiB mutant was defective in cell morphology, motility, and making the switch to a rugose phenotype. In G. hollisae, an abundance of elongated cells in rugose colonies led the authors to conclude that disrupted cell division enhances exopolysaccharide production (29). Taken together, these data support a model in which cell division is tightly linked to exopolysaccharide production in a diversity of microbes. In Salmonella enterica serovar Typhimurium, two of the Tat-secreted amidases, AmiA and AmiC, appear to be regulated by the “envelope stress response” two-component system CpxR/CpxA (9). Similarly, in V. cholerae, mutation of amiB activated the Cpx two-component system (32). In E. coli, CpxR and CpxA regulate amidase activity and also appear to regulate cellulosic biofilm production (33). In this case, the activation of the envelope stress response leads to increased levels of the phosphorylated response regulator CpxR and repression of the cellulosic biofilm activator cyclic di-GMP. V. fischeri possesses homologs of cpxR and cpxA. Thus, we hypothesized that the loss of amiB and subsequent defects in cell division might somehow be connected to the envelope stress response and c-di-GMP production. However, the cellulosic biofilm phenotype of an amiB cpxR double mutant was indistinguishable from that of the amiB mutant (C10), suggesting that the biofilm phenotype of the AmiB mutant might not be due to the envelope stress response. Future work on the mechanism by which AmiB contributes to the regulation of cellulosic biofilm will hopefully shed light on the role of bcs genes in V. fischeri ecology.

We also discovered that strain C10 is severely defective in symbiotic colonization of squid. In addition to motility, several features of V. fischeri contribute to it being the sole colonist of E. scolopes light organs, such as Syp-type biofilm production and optimal levels of bioluminescence (2, 3, 34, 35). However, we did not detect a bioluminescence defect in strain C10 (data not shown). In competition experiments, the amiB mutant was completely outcompeted by the wild type for colonization of the squid light organ. However, when strain C10 alone was exposed to squid, the cells were able to colonize albeit after a significant delay not seen in the wild type. In Salmonella, repression of cellulosic biofilm promotes host colonization (36, 37). Conversely, enhanced production of cyclic di-GMP, the positive regulator of cellulosic biofilm, interferes with flagellum-based motility in V. cholerae and Salmonella (18, 36). Specifically, Zorraquino et al. reported a dual purpose for cyclic di-GMP in blocking flagellar rotation through YcgR and BcsA and cellulose production. In the latter case, cellulose fibrils physically inhibit flagellar rotation (18). These findings are consistent with the excess cellulosic biofilm and diminished motility phenotypes demonstrated by strain C10. In support, the ΔamiB ΔVF_1477 double mutant displayed diminished biofilm production compared to the amiB single mutant and was significantly more motile. Taken together, our findings suggest that cellulose production in V. fischeri, like Salmonella, might play an indirect role in host colonization, possibly through its effect on motility. Perhaps, by defining a set of natural conditions in which cellulosic biofilm production is significantly expressed, we might eventually determine the full extent of its role in the ability of V. fischeri to persist in the environment and colonize squid.

We hypothesize that the temporal increase in the colonization efficiency of the amiB mutant might be due to the evolution of motile suppressor mutants such as those that we identified with SNVs in gene VF_1477. Analysis of the amino acid sequence using SignalP revealed a signal peptide cleavage site, suggesting that it most likely is secreted through the Sec system (data not shown). The presence of Sel1 domains in the encoded protein might be required for the assembly of macromolecular complexes (22). Sel1-containing proteins appear to play diverse roles in a variety of pathogens, including regulating genes within the Fur regulon of Neisseria meningitidis (38). In Vibrio species, including V. fischeri, the Sel1-containing protein MotX is part of its sodium-driven flagellar motor (39). In this study, the deletion of VF_1477 in the nonmotile amiB mutant partially restored motility, suggesting a possible role for VF_1477 in the functioning of flagellar motor proteins. However, our current methodology is not sensitive enough to detect an obvious phenotype when VF_1477 is disrupted in otherwise wild-type cells. For now, we can conclude that the role of VF_1477 in motility is apparent only when the system is perturbed (e.g., when amiB is disrupted). For the future, we plan to study the colonization phenotype of the VF_1477 mutant as well as test the hypothesis that interactions between MotX and the VF_1477 protein are important in flagellar assembly and function. Hussa et al. reported that the flagellar regulator FlrC also appears to regulate biofilm formation (40). Furthermore, Visick et al. revealed that FlrA, FliQ, and possibly flagella in general interfere with cellulosic biofilm formation in V. fischeri (41).

Our study highlights the pleiotropic effects of amidase activity on motility, cellulosic biofilm formation, and symbiotic colonization of squid by cells of V. fischeri. The purpose of cell division proteins in host colonization and as targets for antimicrobials is a burgeoning area of study. Furthermore, the ways in which V. fischeri cells use cellulosic biofilm are not known, but the findings presented here connecting cellulose production, proper cell division, and motility provide new perspectives to study this potentially important aspect of V. fischeri biology.

MATERIALS AND METHODS

Strains and media.

Strains, plasmids, and primers used in this study are listed in Tables 2, 3, and 4, respectively. The parent V. fischeri strain used in this study (e.g., ES114) was isolated from E. scolopes (2). Mutants of this strain were constructed either by triparental mating using E. coli strain CC118 λpir carrying helper plasmid pEVS104 (42) or by TfoX-mediated transformation (43). Gene alterations resulted from either allelic replacement or transposon insertion as described previously (41, 44). For the resolution of the FLP recombination target (FRT)-flanked antibiotic resistance cassettes, flippase plasmid pKV496 was used (41). Mutants containing the Tn5 insertion were verified by sequencing DNA that flanked the insertion (44). For the complementation of either amiB or bcsA, the gene (including the ribosome-binding site through the stop codon) was amplified and inserted into separate pAKD601B plasmids (45). The expression of amiB in pAKD601B was controlled by the addition of 100 μM isopropyl-β-d-thiogalactopyranoside (IPTG), which we determined to be an optimal dose for gene expression (data not shown). For genetic manipulations, cells of V. fischeri were grown on LB-salt (LBS) medium as described previously (46). The liquid medium for biofilm assays (biofilm minimal medium [BMM]) was a modified HEPES minimal medium that contained artificial seawater, 0.1% (wt/vol) ammonium chloride, 0.0058% K₂HPO₄, 10 μM ferrous ammonium sulfate, 100 mM HEPES (pH 7.5), and 0.03% Casamino Acids. Biofilm production was also visualized on solid LBS agar supplemented with 40 μg/ml Congo red and 15 μg/ml Coomassie blue (LBS-CR). For squid colonization assays, V. fischeri cells were grown in seawater tryptone (SWT) medium (2) prior to inoculation into seawater containing squid. Where necessary, the following antibiotics were added to the growth medium: kanamycin (100 μg/ml), erythromycin (2.5 μg/ml), chloramphenicol (5 μg/ml), tetracycline (15 μg/ml), and trimethoprim (10 μg/ml). All incubations were carried out at 24°C to 28°C.

TABLE 2.

V. fischeri strains constructed and/or used in this study^a

Strain	Genotype, characteristic(s), and/or purpose	Reference
C10	amiB::mini-Tn5 Ermr; amidase mutant	This study
C10/RscS	C10+/pRscS Cmr; mutant expressing biofilm regulator RscS	This study
C10+	C10/pAKD601::amiB; complemented AmiB mutant	This study
ES114	Wild type	2
KV7894	ΔbcsA; cellulose synthase mutant	20
KV8069	ΔsypQ::FRT-Cmr; glycosyltransferase mutant	20
KV8233	IG (yeiR-glmS)::Ermr-trunc Trimr	41
KV8442	ΔbcsA amiB::mini-Tn5	This study
KV8443	ΔsypQ::FRT-Cmr amiB::mini-Tn5	This study
KV8578	ΔcpxR::FRT-Trimr; envelope stress response regulator mutant	This study
KV8614	amiB::mini-Tn5 Ermr ΔcpxR::FRT-Trimr	This study
KV8692	ΔamiB::FRT-Trimr	This study
KV8693	ΔVF_1477::FRT-Ermr; putative tetratricopeptide repeat protein mutant	This study
KV8711	ΔVF_1477::FRT	This study
KV8717	ΔamiB::FRT-Trim ΔVF_1477::FRT-Ermr	This study
KV8756	ΔamiB::FRT ΔVF_1477::FRT IG (yeiR-glmS)::Pcel VF_1477; amiB VF_1477 double mutant complemented with VF_1477 driven by a cellobiose-inducible promoter	This study
KV8737	ΔamiB::FRT ΔVF_1477::FRT	This study
PFC	envC::mini-Tn5 Ermr; amidase regulator mutant	This study
PFD	nlpD::mini-Tn5 Ermr; amidase regulator mutant	This study
PFCD	envC::mini-Tn5 Ermr nlpD::FRT	This study
PFCD/pAmiB	envC::mini-Tn5 Ermr ΔnlpD::FRT/pAmiB; amiB-complemented envC nlpD double mutant	This study

^aIG, intergenic region between the genes in parentheses.

TABLE 3.

Plasmids constructed and/or used in this study^a

Plasmid	Characteristic(s)	Reference
pAKD601B	lacIq; IPTG-inducible promoter; Knr	45
pAmiB	pAKD601B containing the coding sequence of VF_2326 (amiB)	This study
pBcsA	pAKD601B containing the coding sequence of bcsA	This study
pEVS170	Mini-Tn5 delivery plasmid	48
plostfoX	TfoX overexpression; Cmr	43
plostfoX-Kan	TfoX overexpression; Knr	49
pRscS (pKG11)	RscS overexpression; Tetr	4
pKV69	Vector for pRscS; Tetr	50
pKV494	pJET + FRT-Ermr	41
pKV496	pJET + flippase gene	41
pKV503	pJET + glmS sequences	41
pKV518	pJET + trunc Ermr; Pcel	41
pMLC2	pJET + FRT-Trimr	41

^aErm^r, erythromycin resistance; Cm^r, chloramphenicol resistance; Trim^r, trimethoprim resistance; Kn^r, kanamycin resistance.

TABLE 4.

Primers used in this study

Primer	Sequencea	Purposeb
cL1	CCATACTTAGTGCGGCCGCCTA	Amplify antibiotic resistance marker
cL2	CCATGGCCTTCTAGGCCTATCC	Amplify antibiotic resistance marker
ΔcpxR-up-F	TCCTGCATAGCTTGGTGATC	Delete cpxR
ΔcpxR-up-R	taggcggccgcactaagtatggCAGAGCTGTCAGTTCAATATC	Delete cpxR
ΔcpxR-down-F	ggataggcctagaaggccatggTTTGTAGAGGAGTAATCGGTG	Delete cpxR
ΔcpxR-down-R	TAACGAGCATGGTAACTAAGAG	Delete cpxR
ΔamiB-Up-F	GAAATTAGTGAATAATGGAACGC	Delete amiB
ΔamiB-Up-R	taggcggccgcactaagtatggAAGAACAACAAATAGCGTATACC	Delete amiB
ΔamiB-Down-F	GGATAGGCCTAGAAGGCCATGGATTCCAACCAGTTAAGGAACAAGG	Delete amiB
ΔamiB-down-R	CCAATCGACTTAATGCAATACGC	Delete amiB
ΔnlpD-Up-F	CGAAGTATGGTTAATCGCTATT	Delete nlpD
ΔnlpD-Up-R	taggcggccgcactaagtatggATTTGCTTTGTTGCCGATATC	Delete nlpD
ΔnlpD-Down-F	ggataggcctagaaggccatggGTTTTGGCTTAGCAACCGGTT	Delete nlpD
ΔnlpD-Down-R	CTCTAGTACGAGCAGTGTACG	Delete nlpD
ΔVF_1477-Up-F	ATTGGCATCGTGATTCCGTG	Delete VF_1477
ΔVF_1477-Up-R	taggcggccgcactaagtatggAACAGCAAGTAAAACAGAACGGATC	Delete VF_1477
ΔVF_1477-Down-F	ggataggcctagaaggccatggCGATAACGTACACGGAATACGAG	Delete VF_1477
ΔVF_1477-Down-R	GCTCCAATTGCAATTGGTCTG	Delete VF_1477
amiBF	CTTGAGTTTATGGGGATCCGTGA	Complement amiB and envC nlpD
amiBR	GATCCCCGATCCATTATCACGGA	Complement amiB and envC nlpD
bcsAF	TGAGTTAGATAAGCAGTGC	Complement bcsA
bcsAR	CATTGTCACTATGAGCTGT	Complement bcsA
Ins-Up-F*	AAGAAACCGATACCGTTTACG	Complement VF_1477
c-VF1477-F	ggataggcctagaaggccatggaggaggtTTATATTTATGATCCGTTCTGTTTTACTTG	Complement VF_1477
c-VF1477-R	taggcggccgcactaagtatggaCTCGTATTCCGTGTACGTTATCG	Complement VF_1477
Ins-Down-F	TCCATACTTAGTGCGGCCGCCTA	Complement VF_1477
Ins-Down-R	GGTCGTGGGGAGTTTTATCC	Complement VF_1477
VIPCR-F	CCTAGAGCGGCCGCAGA	iPCR
VIPCR-RBio	Biotin-ACTGGCCGTCGTTTTACAG	iPCR
MoSeq-F	AGATGTGTATAAGAGAC	Tn localization

^aAll sequences are shown in the 5′-to-3′ direction. Lowercase letters indicate “tails” that are not homologous to the target being amplified.

^bTn, transposon.

Transposon mutagenesis of V. fischeri.

A transposon mutant library of V. fischeri was constructed as described previously (44). Two thousand mutants were streak purified and screened for alterations in biofilm formation using the crystal violet assay (see below). Mutants with at least a 2-fold variation in biofilm production (over- or underproduction) relative to wild-type strain ES114 were sequenced to localize the transposon insertion.

Determining the site of Tn5 insertion in V. fischeri strains.

Genomic DNA was extracted from biofilm mutants, cut with Sau3AI, and self-ligated to form closed circular DNA fragments. Next, fragments were subjected to inverse PCR using primers VIPCR-F (5′-CCTAGAGCGGCCGCAGA-3′) and VIPCR-RBio (biotin-5′-ACTGGCCGTCGTTTTACAG-3′) to amplify DNA flanking the transposon insertion. The resulting PCR amplicons were sequenced using primer MoSeq-F (5′-AGATGTGTATAAGAGAC-3′) and a pyrosequencer to identify the flanking DNA.

Biofilm assays. (i) Crystal violet.

V. fischeri cells were grown in BMM at 28°C with shaking overnight. Biofilms formed in tubes or on glass slides were then stained with crystal violet. The cell density (OD₆₀₀) of each culture was determined, and the tubes were thoroughly rinsed with sterile water to remove nonadhering cells and then filled with 1% crystal violet for 5 min to stain biofilms. For the qualitative determination of biofilm density, nonadherent crystal violet was rinsed with water, and biofilms were photographed. For the quantitative determination of biofilm density, biofilm-associated crystal violet was eluted with 95% ethanol. The OD₆₀₀ of the eluate was then measured with a spectrophotometer. The specific biofilm density was determined by dividing the OD₆₀₀ of the biofilm eluate by the OD₆₀₀ of the planktonic cell density.

(ii) Colony architecture.

We also used colony appearance as an indirect measure of biofilm formation. V. fischeri strains were grown overnight at 28°C in LBS medium. The cultures grown overnight were subcultured, grown until mid-log phase, and diluted to an OD₆₀₀ of 0.2. Aliquots were spotted onto LBS agar and incubated for 24 h at 24°C, and the resulting growth (referred to as a “colony” for simplicity) was imaged with a Zeiss Stemi 2000-C dissecting microscope.

(iii) Congo red.

Cellulosic biofilm production was visualized by streaking cells on LBS-CR plates. After incubation, streaks were overlaid with white paper to transfer the cells for easier visualization of red dye accumulation indicative of cellulose production, as described previously (20). When needed, broth cultures were supplemented with IPTG prior to streaking onto Congo red plates.

(iv) Determining cell arrangement and morphology.

Planktonic cells of V. fischeri grown with shaking overnight in LBS broth at 28°C were aseptically smeared onto glass slides with a loop, air dried, heat fixed, and then stained with crystal violet for 1 min. Excess crystal violet was then rinsed off, and cells were visualized by bright-field microscopy. To visualize biofilm-associated cells, V. fischeri cells were grown in BMM at 28°C with gentle shaking overnight in 50-ml conical tubes containing a sterilized glass slide. Slides were removed from the tubes, rinsed to remove nonadhering cells, and then flooded with 1% crystal violet for 5 min. Nonadherent crystal violet was removed with sterile water, and stained biofilms were visualized by bright-field microscopy.

(v) Loss-of-chaining kinetics in the mutant due to complementation.

One flask each of the wild type and mutant and two flasks of the mutant complemented with pAKD601::amiB were grown in LBS medium with shaking. When the flasks reached an OD₆₀₀ of ∼1.0, an aliquot of IPTG was added to one of the flasks of the complemented mutant. Subsequently, an aliquot was removed from each flask to determine cell arrangement and morphology as described above; this was repeated every 5 min for 1 h.

(vi) Motility assay.

Cells of V. fischeri were grown at 28°C with shaking in LBS medium to mid-log phase (OD₆₀₀ = ∼0.3 to 0.4). A 10-μl aliquot of cells was spotted onto TBS-MgSO₄ soft agar and monitored for migration from the spot over the course of 6 h as described previously (47).

(vii) Whole-genome sequencing to characterize suppressor mutants.

Genomic DNA was fragmented by sonication in iced deionized H₂O six times for 15 s at 5 W using an F60 Sonic Dismembrator (Fisher Scientific) to produce ∼200-bp fragments for Illumina library construction. Genomic DNA adaptors were diluted to 15 mM prior to ligation. Sequencing was performed using an Illumina NextSeq genome analyzer at the University of Georgia Facility. Assembled reads were compared to the resequenced and published ES114 genomes as references using Geneious software.

(viii) Squid colonization assays.

The symbiotic competence of mutant V. fischeri strains was determined either singly or in competition with the wild type. For single-strain inoculations, cells were grown to mid-log phase in SWT medium and then inoculated into separate bowls of seawater containing squid to achieve ∼3,000 CFU/ml. After 3 h, squid were moved to V. fischeri-free seawater and monitored for luminescence over several days. For competition experiments, squid were exposed to an ∼1:1 ratio of mutant to wild-type cells for 3 h. Squid were moved into V. fischeri-free seawater and then homogenized after 24 h to determine the ratio of wild-type to mutant cells in each light organ. The relative competitive index (RCI) was determined by dividing the wild-type-to-mutant ratio for each squid by the ratio for the inoculum. Log-transformed data were used to calculate the average RCI and determine statistical significance.

Data availability.

The raw sequencing reads are available via the Sequence Read Archive (BioProject no. PRJNA674884).