Mutational Analysis of Vibrio fischeri c-di-GMP-Modulating Genes Reveals Complex Regulation of Motility

ABSTRACT

The symbiont Vibrio fischeri uses motility to colonize its host. In numerous bacterial species, motility is negatively controlled by cyclic-di-GMP (c-di-GMP), which is produced by diguanylate cyclases (DGCs) with GGDEF domains and degraded by phosphodiesterases with either EAL or HD-GYP domains. To begin to decode the functions of the 50 Vibrio fischeri genes with GGDEF, EAL, and/or HD-GYP domains, we deleted each gene and assessed each mutant’s migration through tryptone broth salt (TBS) soft agar medium containing or lacking magnesium (Mg) and calcium (Ca), which are known to influence V. fischeri motility. We identified 6, 13, and 16 mutants with altered migration in TBS-Mg, TBS, and TBS-Ca soft agar, respectively, a result that underscores the importance of medium conditions in assessing gene function. A biosensor-based assay revealed that Mg and Ca affected c-di-GMP levels negatively and positively, respectively; the severe decrease in c-di-GMP caused by Mg addition correlates with its strong positive impact on bacterial migration. A mutant defective for VF_0494, a homolog of V. cholerae rocS, exhibited a severe defect in migration across all conditions. Motility of a VF_1603 VF_2480 double mutant was also severely defective and could be restored by expression of “c-di-GMP-blind” alleles of master flagellar regulator flrA. Together, this work sheds light on the genes and conditions that influence c-di-GMP-mediated control over motility in V. fischeri and provides a foundation for (i) assessing roles of putative c-di-GMP-binding proteins, (ii) evaluating other c-di-GMP-dependent phenotypes in V. fischeri, (iii) uncovering potential redundancy, and (iv) deciphering signal transduction mechanisms.

IMPORTANCE Critical bacterial processes, including motility, are influenced by c-di-GMP, which is controlled by environment-responsive synthetic and degradative enzymes. Because bacteria such as Vibrio fischeri use motility to colonize their hosts, understanding the roles of c-di-GMP-modulating enzymes in controlling motility has the potential to inform on microbe-host interactions. We leveraged recent advances in genetic manipulation to generate 50 mutants defective for putative c-di-GMP synthetic and degradative enzymes. We then assessed the consequences on motility, manipulating levels of magnesium and calcium, which inversely influenced motility and levels of c-di-GMP. Distinct subsets of the 50 genes were required under the different conditions. Our data thus provide needed insight into the functions of these enzymes and environmental factors that influence them.

INTRODUCTION

The second messenger cyclic diguanylate (c-di-GMP) is a key regulator that influences a wide array of pathways, including motility, biofilm formation, and virulence (1–3). Central to c-di-GMP signaling are enzymes, diguanylate cyclases (DGCs) and phosphodiesterases (PDEs), that synthesize and degrade this small molecule, respectively, often in response to environmental signals. Completing the signal transduction pathway, c-di-GMP binds proteins or RNA, thus stimulating or inhibiting a downstream response.

Bioinformatics and biochemical work have revealed key domains and motifs that allow these enzymes and regulators to carry out their functions (4–8). Functional DGCs generally contain GGDEF domains with a GG(D/E)EF active site, although other variations in this motif may be permissive for function (e.g., reference 9). A subset of DGCs also contains an I-site (inhibitory site), RxxD, that is positioned five residues upstream of the GG(D/E)EF motif; c-di-GMP binding at this position causes feedback inhibition of the DGC (10). Two general classes of PDEs are known, those that contain an EAL domain with an ExLxR motif and those that contain an HD-GYP domain with HD and GYP motifs (4, 11). Some proteins contain both GGDEF and EAL domains and may have dual function, exhibiting both DGC and PDE activity and switching those activities in response to a signal(s). Other GGDEF- and/or EAL-containing proteins are degenerate, lacking clear GGDEF or EAL motifs and/or having incomplete domains. In some cases, degenerate proteins that lack enzymatic activity retain the ability to bind c-di-GMP, thus maintaining a role in a signaling pathway. In particular, I-site-containing degenerate DGCs can bind c-di-GMP. In addition to degenerate enzymes, many other types of proteins appear competent to bind c-di-GMP, including a class of proteins that contain the PilZ domain (12). Finally, many GGDEF, EAL, and/or HD-GYP proteins also contain predicted sensory domains that likely influence their activities.

Some bacteria possess no or only a few proteins that control c-di-GMP levels, while others contain numerous such enzymes. In general, Vibrio species encode large numbers of these proteins, with the pathogens Vibrio cholerae, Vibrio parahaemolyticus, and Vibrio vulnificus predicted to contain 62, 62, and 92 DGC/PDE enzymes, respectively (https://www.ncbi.nlm.nih.gov/Complete_Genomes/c-di-GMP.html) (2, 5, 12, 13). The symbiont V. fischeri encodes 50 proteins with GGDEF/EAL/HD-GYP annotations (14–16). The large numbers of these proteins suggest that c-di-GMP signaling provides key advantages to this species.

V. fischeri serves as an important model for a variety of biological phenomena, including bioluminescence, biofilm formation, and symbiosis. This marine bacterium forms an exclusive symbiosis with Euprymna scolopes, a Hawaiian squid (reviewed in references 17 and 18). By colonizing the squid, V. fischeri gains nutrients and a protected environment in exchange for its bioluminescence, which E. scolopes uses to evade predation. Newly hatched, uncolonized juvenile squid are colonized by V. fischeri present in the surrounding seawater in a series of events that include those that are known to be affected by c-di-GMP signaling, biofilm formation, and motility. Notably, mutants defective for biofilm formation or motility are deficient in colonizing squid.

Because of its importance in squid colonization and the ease with which it can be studied, motility of V. fischeri has been probed in a number of studies (e.g., references 19–22). In laboratory culture, V. fischeri motility is affected by the presence of the cations magnesium (Mg) and calcium (Ca) (23). Addition of 35 mM Mg (the amount present in seawater) permits V. fischeri to migrate through a soft agar medium more rapidly than in its absence. Mg exerts that effect by increasing the number of flagella per cell (23). The impact of Ca varies, dependent on the concentration. At concentrations of 10 mM and below, Ca supplementation increases bacterial migration through soft agar, albeit to a lesser extent than that caused by Mg (23). At higher concentrations, notably 40 mM and above, Ca exerts a negative effect on migration. Investigation into the role of Mg in promoting V. fischeri motility led to the identification of MifA (VF_0989) and MifB (VF_A0959), two DGCs that contributed to the poor migration of V. fischeri in the absence of Mg (24). More recently, the DGC CasA (encoded by VF_1639) was shown to control the response to high levels of Ca: mutation of casA permitted V. fischeri to migrate to the same extent that occurred in the absence of high calcium, rather than exhibiting the reduced migration of the wild-type strain (25). Together, these studies revealed the importance of cations in controlling V. fischeri motility and identified roles for three specific DGCs in motility under different conditions.

Here, we take a global approach to probing the functions of putative DGC and PDE enzymes in the physiology of V. fischeri, focusing specifically on bacterial motility using the sensitive soft agar migration assay. We leveraged recent advances in tools for rapid genetic manipulation of V. fischeri (26) to generate mutations in each of the 50 genes encoding proteins with GGDEF, EAL, and HY-GYP domains in wild-type squid symbiont strain ES114. These studies revealed putative DGCs and PDEs with significant roles in controlling V. fischeri motility. Furthermore, by manipulating Mg and Ca levels to alter environmental conditions, we uncovered additional c-di-GMP regulators responsive to particular conditions. Finally, our studies demonstrated a conservation in function between V. fischeri and V. cholerae for certain DGCs and PDEs.

RESULTS AND DISCUSSION

Functional grouping of 50 c-di-GMP-related genes in V. fischeri.

V. fischeri strain ES114, a symbiont of the squid E. scolopes, contains 50 genes encoding GGDEF, EAL, or HY-GYP proteins in its genome (14–16). To begin characterizing this extensive c-di-GMP regulatory network, we used bioinformatics tools to identify putative associated sensory domains (see Fig. S1 in the supplemental material), to assess whether each may be catalytically active or degenerate (based on the conservation of motifs and the presence/absence of intact domains) (Table 1 to 4), and to predict where each may be localized in the cell (Fig. S2). We also compared these enzymes to those encoded by V. cholerae (27), whose extensive regulatory network has been assessed previously (e.g., reference 28) (Table 5). These analyses permitted us to classify these proteins into four putative functional groups: (i) DGCs, (ii) PDEs, (iii) dual-function DGC/PDEs, and (iv) nonenzymatic, putative c-di-GMP-binding proteins.

TABLE 1.

Putative DGCs

Protein and motif informationa				Migration resultsb
Protein	RxxD	GG(D/E)EF	Vc homolog	TBS-Mg	TBS	TBS-Ca
VF_0596		SGEEF	VC_0900 (CdgG)	−	−	−
VF_0989 (mifA)	RSID	GGEEF			+	+
VF_1200		GGDEF			+	+
VF_1245		GGEEF
VF_1350	KEND	GGEEF
VF_1515		GGDEF	VC_1376 (CdgM)
VF_1561	KSTD	GGEEF	VC_1185
VF_1639 (CasA)	RDSD	GGEEF	VC_1104 (CdgK)			+
VF_2261	RETD	GGEEF			+
VF_2362	RQSD	GGDEF
VF_A0056	RSRD	GGDEF
VF_A0057	FEND	SGDEF
VF_A0152	ESDD	GGDEF
VF_A0155	NQDD	GGDEF				+
VF_A0276		GGEEF				+
VF_A0323		GGDEF
VF_A0342	RVND	GGEEF
VF_A0343	RGDD	GGEEF
VF_A0368		GGEEF			−
VF_A0381		GGEEF			+	+
VF_A0398		GGEEF
VF_A0476		GGDEF				−
VF_A0567	RAND	AGDEF
VF_A0692		GGEEF
VF_A0796	RNTD	GGEEF
VF_A0959		GGDEF		+	+	+
VF_A0976		GGDEF				−
VF_A1012	RKTD	GGEEF

^aListed are the V. fischeri protein, the putative I-site (RxxD), and/or the GGDEF motif (the former 5 bp upstream of GGDEF motif), with residues that depart from the consensus shown in bold and the V. cholerae homolog (see Table 5 for detailed information).

^bSymbols reflect migration results of the mutant relative to those the wild-type parent, as follows: −, decreased and +, increased.

TABLE 2.

Putative PDEs^a

Protein and motif informationa			Migration resultsb
Protein	ExLxR/HD-GYP	Vc homolog	TBS-Mg	TBS	TBS-Ca
VF_0087	ECLMR	VC_0130 (CpdA)		−
VF_0091	EALIR	VC_2750
VF_1603	ELLFR	VC_1851	−	−	−
VF_2480	ELLFR	VC_0137 (CdgJ)	−	−	−
VF_A0344	EVLAR
VF_A0526	EVLLR		−	−	−
VF_A0551	EALLR	VC_A0536
VF_A0706	EALLR
VF_A0879	EALSR
VF_A1014 (PdeV)	EALIR
VF_A1038 (BinA)	EALLR	CdgC
VF_A1076	EALLR
VF_1367	HD-GY
VF_A0506	HD-GYP

^aListed are the V. fischeri protein, the putative ExLxR or HD-GYP motif, and the V. cholerae homolog (see Table 5 for detailed information). VF_0087, VF_0091, and VF_A1038 have apparently degenerate GGDEF motifs, as follows, respectively: GIAEW, GGAEF, and GSSDI. VF_1367 lacks the P of the HD-GYP motif.

^bSymbols reflect migration results of the mutant relative to those the wild-type parent, as follows: −, decreased.

TABLE 3.

Putative dual DGC/PDE proteins

Protein and motif informationa					Migration resultsb
Protein	RxxDa	GGDEF	ExLxR	Vc homolog	TBS-Mg	TBS	TBS-Ca
VF_0094		GGDEF	EALIR	VC2750			−
VF_0494		GGDEF	EALVR	VC_0653 (RocS)	−	−	−
VF_0985		GGDEF	EVLSR			+
VF_A0244	RKED	GGDEF	EALIR
VF_A0475	RKED	GGDEF	EALLR

^aListed are the V. fischeri protein, the putative I-site (RxxD) and/or GGDEF motif (the former 5 bp upstream of GGDEF motif), the ExLxR motif, and the V. cholerae homolog (see Table 5 for detailed information).

^bSymbols reflect migration results of the mutant relative to those the wild-type parent, as follows: −, decreased and +, increased.

TABLE 4.

Putative nonenzymatic GGDEF and EAL proteins

Protein and motif informationa				Migration resultsb
Protein	GG(D/E)EF	ExLxR	Vc homolog	TBS-Mg	TBS	TBS-Ca
VF_0355 (MshH/CsrD)	YQSDF	ELFSR	VC_0398 (MshH)			−
VF_A0216		QVSPH
VF_A1166 (LapD)	NSSEF	EVFSA	VC_A1082 (LapD)

^aListed are the V. fischeri protein, the residues that align with the GGDEF and/or ExLxR motif, and the V. cholerae homolog (see Table 5 for detailed information). The residues that depart from consensus are shown in bold.

^bColors reflect migration results of the mutant relative to the wild-type parent, as follows: −, decreased.

TABLE 5.

Comparisons of V. fischeri proteins with V. cholerae proteins

Protein		Expect	Identity (%)	Similarity (%)	Gaps (%)
V. fischeri	V. cholerae
VF_0087	VC_0130 (CpdA)	0	359/829 (43%)	511/829 (61%)	3/829 (0%)
VF_0091	VC_2750	0	295/806 (37%)	474/806 (58%)	16/806 (1%)
VF_0094	VC_2750	0	391/846 (46%)	564/846 (66%)	13/846 (1%)
VF_0355	VC_0398 (MshH)	0	329/642 (51%)	453/642 (70%)	2/642 (0%)
VF_0494	VC_0653 (RocS)	0	439/676 (65%)	546/676 (80%)	5/676 (0%)
VF_0596	VC_0900 (CdgG)	4.00E−135	212/513 (41%)	336/513 (65%)	2/513 (0%)
VF_1515	VC_1376 (CdgM) (AdrA_2)	4.00E−117	186/478 (39%)	288/478 (60%)	3/478 (0%)
VF_1561	VC_1185	9.00E−129	178/431 (41%)	275/431 (63%)	5/431 (1%)
VF_1603	VC_1851	5.00E−175	233/397 (59%)	302/397 (76%)	0/397 (0%)
	VC_0137 (CdgJ)	4.00E−126	190/409 (46%)	274/409 (66%)	12/409 (2%)
VF_1639 (CasA)	VC_1104 (CdgK)	2.00E−87	145/492 (29%)	266/492 (54%)	11/492 (2%)
VF_2480	VC_0137 (CdgJ)	1.00E−150	208/404 (51%)	296/404 (73%)	2/404 (0%)
	VC_1851	1.00E−118	178/403 (44%)	267/403 (66%)	4/403 (0%)
VF_A0551	VC_A0536	7.00E−68	113/255 (44%)	157/255 (61%)	3/255 (1%)
VF_A1038 (BinA)	VC_A0785 (CdgC)	0	279/618 (45%)	407/618 (65%)	7/618 (1%)
VF_A1166 (LapD)	VC_A1082 (LapD)	3.00E−161	216/379 (57%)	286/379 (75%)	0/379 (0%)
V. fischeri protein 1	V. fischeri protein 2
VF_0091	VF_0094	0	335/824 (41%)	512/824 (62%)	17/824 (2%)
VF_1603	VF_2480	1.00E−130	186/402 (46%)	278/402 (69%)	4/402 (0%)
VF_A0342	VF_A0343	2.00E−85	156/434 (36%)	246/434 (56%)	16/434 (3%)

Group 1 (Table 1) encompasses 28 putative DGCs that contain a conserved GGDEF domain with a fully or largely conserved GG(D/E)EF motif. We included in this group three proteins that contain a noncanonical residue in the first amino acid of the GGDEF motif. One contains an alanine in that position (VF_A0567), while two contain a serine (VF_0596 and VF_A0057). Functional DGCs with these substitutions have been reported previously. For example, V. cholerae VC_A0965 contains an alanine in that position, yet it is a fully functional DGC (9), while Pectobacterium atrosepticum expresses an active DGC that contains a serine (29). However, CdgG (VC_0900), a homolog of VF_0596 (Table 5) that contains a serine instead of glycine in that location, lacks DGC activity and inhibits rather than promotes c-di-GMP-dependent phenotypes in V. cholerae (30); thus, the inclusion of this or the other two V. fischeri enzymes with noncanonical GGDEF motifs in this class may need to be revisited at a later time. In addition to VF_0596, three other putative DGCs exhibit significant similarity to V. cholerae proteins, VF_1515 (VC_1376, CdgM), VF_1561 (VC_1185), and CasA (VF_1639; VC_1104, CdgK) (Table 5). Finally, 10 have putative I-sites (RxxD) that are conserved, while 5 others contain a properly positioned D but another residue in place of the R, and the remaining 13 lack a putative I-site.

Group 2 (Table 2) includes 14 putative phosphodiesterase proteins. Of these, 12 have a conserved EAL domain with a fully conserved ExLxR motif; three of these also contain a degenerate GGDEF domain or motif, making them unlikely to be catalytically active as DGCs. Five of the EAL-domain-containing proteins are similar to those found in V. cholerae (Table 5): VF_0087 (VC_0130, CpdA), VF_0091 (VC_2750), VF_1603 (VC_1851), VF_2480 (VC_0137, CdgJ), and VF_A0551 (VC_A0536). Of note, both VF_0091 and, below in group 3, VF_0094 are similar to VC_2750 and to each other; while VF_0091 contains a putative GGDEF domain, it lacks an intact GGDEF motif, containing instead a GGAEF motif that is expected to lack DGC activity. VF_1603 and VF_2480 also exhibit substantial similarity to each other; in this case, each is most similar to a distinct V. cholerae protein (VC_1851 and VC_0137, respectively) (Table 5). Also included in this group are two proteins that contain an HD-GYP domain. VF_A0506 contains HD and GYP amino acids with spacing similar to that found in other HD-GYP proteins (31), while VF_1367 lacks the P of the GYP portion of the HD-GYP motif.

Group 3 (Table 3) includes five proteins that contain both GGDEF and EAL domains with intact motifs, supporting their assignment as putative dual-function DGC/PDEs; two of these also contain intact I-sites. VF_0094 and VF_0494 exhibit significant sequence similarity to V. cholerae proteins VC_2750 and VC_0653 (RocS [32]), respectively (Table 5).

Finally, group 4 (Table 4) includes three proteins with degenerate domains and/or motifs, suggesting that these proteins lack enzymatic activity but may function as c-di-GMP-binding proteins. One of these is VF_A1166, or LapD, a homolog of the Pseudomonas LapD protein that binds c-di-GMP, permitting it to sequester the LapG protease responsible for cleaving a surface adhesin (reviewed in reference 33); VF_A1166/LapD appears to function in the same manner (34), as does its V. cholerae homolog, VC_A1082 (35). Another is VF_0355 or MshH (CsrD), which is a homolog of V. cholerae VC_0398 (Table 5); in V. cholerae, this protein functions to inhibit biofilm formation under specific conditions (36). The last protein in this group is VF_A0216, which contains a response regulator receiver (REC) domain and a degenerate EAL domain.

Our list of 50 proteins and their grouping into functional categories differs in three ways from that proposed by Galperin et al. (https://www.ncbi.nlm.nih.gov/Complete_Genomes/c-di-GMP.html) (2, 5, 12, 13). First, due to errors in the original sequence of the ES114 genome that were later uncovered (14, 15), the genes encoding VF_A0242, VF_A0243, and VF_A0244 were resolved into a single gene, VF_A0244, which encodes a dual GGDEF/EAL protein. Second, protein VF_A0477, listed as a GGDEF domain protein, actually lacks this domain, and thus we have not included it here. Finally, several proteins contain either incomplete domains or domains with degenerate motifs, and thus we have assigned those proteins to categories different from those listed on the website.

Mutants defective for c-di-GMP-related genes exhibit normal growth.

To begin to understand the role of c-di-GMP in the physiology of V. fischeri strain ES114, we generated deletion mutants of the 50 c-di-GMP-related genes, regardless of whether they were predicted to encode catalytically active enzymes. In addition, we also generated double and/or triple deletions of genes contained within the putative operons, VF_A0056-57, VF_A0342-343-344, and VF_A0475-476, that consist of multiple DGC/PDE genes. We selected for gene replacement mutants using one of several different antibiotic resistance cassettes and then resolved the cassettes using Flp recombinase acting on FRT sequences flanking each resistance gene as described previously (26). A few of these single mutants have been reported separately (lapD [VF_A1166] and pdeV [VF_A1014] [34], VF_1200 [37], and casA [VF_1639] [25]), and, as noted in the introduction, a few other genes have been characterized previously, including mifA, mifB, and binA (24, 38). Of the 55 mutants generated (including combinations), all exhibited normal growth characteristics in a rich medium (Fig. S3). Together, these data suggest that these proteins function to regulate cellular processes other than normal growth and metabolism under these conditions. In subsequent experiments, we assess the impact of these mutations on motility using the soft agar assay, which relies on not only motility but also growth (and chemotaxis). These growth experiments indicate that any motility defect we discover is unlikely to be due to a growth defect.

Six c-di-GMP-related proteins contribute to migration in magnesium-supplemented medium.

We assessed the consequences of these mutations on bacterial motility by inoculating the mutants onto a tryptone-based soft agar medium supplemented with magnesium (TBS-Mg), which is known to promote V. fischeri motility (23). We measured the area of the migrating cells after a specific incubation period (4.5 h) and normalized the migration area to that of the wild-type strain on the same plates. Under these TBS-Mg conditions, only six mutants exhibited notable migration differences: two DGCs, three PDEs, and one dual DGC/PDE. The remainder of the mutants (including mutants defective for multiple genes within putative operons) migrated similarly to the wild-type strain (no deviation from wild-type migration greater than 15%), potentially because magnesium is a powerful motility stimulant (23) (Fig. 1 and Fig. S4).

FIG 1.

Migration of mutant strains on TBS-Mg. Migration of strains with significant changes relative to that of wild-type strain ES114 on TBS-Mg soft agar plates. (A and B) Graphs of the ratio of mutant migration relative to that of ES114. (A) DGCs. (B) PDEs and dual function. (C) Representative pictures of mutants with significant changes in migration and/or patterns relative to ES114, following growth on TBS-Mg. Pictures taken after 4.5 h. WT, wild-type strain ES114; the VF/VFA number indicates the gene that is deleted in the mutant. Error bars represent standard deviation. Data from a representative experiment performed with three technical triplicates are shown. The migration of each mutant was compared to the average migration of the wild-type strain present on the same soft agar plates and analyzed using an unpaired t test. ****, P < 0.0001; ***, P ≤ 0.001.

Given that DGC-produced c-di-GMP inhibits motility, we expected that DGC mutants would exhibit increased motility. Of the DGC mutants, however, only the mifB mutant (VF_A0959) consistently exhibited increased migration, and it was only modestly increased relative to that of the wild-type (WT) strain (Fig. 1A). Of note, and as previously reported, both the mifA and mifB mutants lacked the robust center spot observed for wild-type strain ES114 (24) (Fig. 1C). The underlying signal transduction pathway(s) responsible for the formation of this center spot, a densely populated region of poorly swimming cells, remains unknown. We observed a similar phenomenon for the VF_1200 mutant (Fig. 1C), which migrated about 10% more than the wild-type strain (Fig. S4), but not for any other mutant. These data suggest that the center spot can be attributed to the activities of these three proteins; loss of any one is sufficient to eliminate this phenotype, presumably by rendering the culture more uniformly motile under these conditions.

The other DGC mutant exhibited a substantial decrease in migration, ΔVF_0596 (Fig. 1A and Fig. S4A). Although a decrease in migration is not consistent with the predicted function of VF_0596, a similar phenomenon was reported for its V. cholerae homolog, CdgG (30); the two proteins exhibit 41% amino acid identity and 65% similarity (Table 5) across the length of the proteins, with both containing a C-terminal GGDEF domain. It will be of interest in the future to determine if these unusual regulators share other characteristics and/or regulons. At the current time, it is unclear why loss of VF_0596 would be detrimental for migration, but a few other DGC mutants also exhibited modestly decreased migration (less than 15% deviation from the wild-type migration) (Fig. S4B).

We expected PDE mutants to exhibit decreased migration due to increases in c-di-GMP levels, and indeed, this was the case. Mutants lacking the putative PDEs VF_1603, VF_2480, or VF_A0526 exhibited migration defects, with the ΔVF_2480 mutant defect being the most severe of these (Fig. 1B and C). Of note, a single predicted dual-function protein, VF_0494, was critical for migration; the observed defect of the VF_0494 mutant was the strongest of any of the 50 mutants tested (Fig. 1B and C).

Thirteen genes are involved in controlling migration on TBS medium.

Given that relatively few mutants exhibited phenotypes on TBS-Mg, we wondered if additional phenotypes would be observed if the strains were inoculated on TBS soft agar medium lacking Mg. ES114 migrates considerably more slowly through this medium, and thus we increased the incubation time to 6 to 7 h. We found that, indeed, 13 of the set of 55 strains (including operon mutants) exhibited altered migration under these conditions, compared to 6 in TBS-Mg (Fig. 2 and Fig. S5). The phenotypes of DGC mutants lacking VF_0596 or MifB (VF_A0959) were similar on the two media, albeit with a larger difference on TBS for the latter (Fig. 2A). Moreover, the TBS conditions revealed altered migration phenotypes for five additional DGC mutants; those lacking MifA (VF_0989) exhibited a striking increase, while those lacking VF_1200, VF_2261, or VF_A0381 had more modest but significantly increased migration (Fig. 2A). Additionally, the ΔVF_A0368 mutant exhibited modestly decreased migration (Fig. 2A). For the PDE mutants, the decreased migration seen for strains lacking VF_1603, VF_2480, or VF_A0526 on TBS-Mg also occurred on TBS. A small but significant decrease in migration was observed for PDE mutant ΔVF_0087 (Fig. 2B). Finally, the dual DGC/PDE mutant lacking VF_0494 retained a migration defect (albeit slightly less severe), while a mutant lacking VF_0985 consistently exhibited a small but significant increase in migration (Fig. 2C). None of the other mutants, including those defective for multiple genes within putative operons, exhibited a consistent defect in migration (greater or less than 15% of wild-type migration) (Fig. S5).

FIG 2.

Migration of mutant strains on TBS. Migration of strains with significant changes relative to that of wild-type strain ES114 on TBS soft agar plates. (A and B) Graphs of the ratio of mutant migration relative to that of ES114. (A) DGCs. (B) PDEs and dual function. (C) Representative pictures of mutants with significant changes in migration relative to ES114, following growth on TBS. Pictures taken after 6.5 h. WT, wild-type strain ES114; the VF/VFA number indicates the gene that is deleted in the mutant. Error bars represent standard deviation. Data from a representative experiment performed with three technical triplicates are shown. The migration of each mutant was compared to the average migration of the wild-type strain present on the same soft agar plates and analyzed using an unpaired t test. ****, P < 0.0001; ***, P ≤ 0.001; **, P ≤ 0.01.

Sixteen c-di-GMP-related proteins influence migration in calcium-supplemented medium.

Finally, we assessed the consequences of these mutations on the ability of V. fischeri to migrate through TBS soft agar supplemented with 10 mM calcium chloride (TBS-Ca). Supplementation with 10 mM calcium promotes migration relative to unsupplemented TBS medium (23) but to a lesser extent than when magnesium is added. As a result, we evaluated migration following incubation on soft agar plates for approximately 5.5 h. Like TBS, the TBS-Ca condition also revealed a large number of mutants (16) to be defective (Fig. 3 and Fig. S6), but there was only a partial overlap between the two conditions.

FIG 3.

Migration of mutant strains on TBS-Ca. Migration of strains with significant changes relative to that of wild-type strain ES114 on TBS-Ca soft agar plates. (A and B) Graphs of the ratio of mutant migration relative to that of ES114. (A) DGCs. (B) PDEs, dual function, and degenerate. (C) Representative pictures of mutants with significant changes in migration relative to ES114, following growth on TBS-Ca. Pictures taken after 5.5 h. WT, wild-type strain ES114; the VF/VFA number indicates the gene that is deleted in the mutant. Error bars represent standard deviation. Data from a representative experiment performed with three technical triplicates are shown. The migration of each mutant was compared to the average migration of the wild-type strain present on the same soft agar plates and analyzed using an unpaired t test. ****, P < 0.0001; ***, P ≤ 0.001; **, P ≤ 0.01; *, P ≤ 0.05.

Five of the DGC mutants exhibited the same relative migration pattern on TBS-Ca as they did on TBS (mutants deleted for VF_0596, mifA [VF_0989], VF_1200, VF_A0381, or mifB [VF_A0959]) (Fig. 3A). Two DGC mutants with phenotypes on TBS failed to exhibit phenotypes on TBS-Ca (defective for VF_2261 or VF_A0368). Of note, the TBS-Ca condition uncovered phenotypes for five additional DGCs not observed on TBS: casA (VF_1639; a phenotype reported by reference 25), VF_A0155, VF_A0276, VF_A0476, and VF_A0976 (Fig. 3A).

For the PDE mutants (Fig. 3B), the behaviors of the VF_2480, VF_1603, and VF_A0526 mutants on TBS-Ca were similar to those observed on TBS, although the relative severity of the defect of the former two mutant strains was different: the VF_2480 mutant exhibited a greater migration defect on TBS-Ca (and on TBS-Mg) than the VF_1603 mutant, whereas the migration defects of the two mutants on TBS were opposite, with the VF_1603 mutant migrating worse than the VF_2480 mutant.

As with the other conditions, a mutant defective for the putative dual regulator VF_0494 migrated poorly on TBS-Ca; this mutant was not only consistently defective across the media types, but it also exhibited the most severe defects of all the strains under all conditions. The small increase in migration observed on TBS for putative dual regulator mutant ΔVF_0985 was lost on TBS-Ca, while small defects for mutants of VF_0094 (a putative dual regulator) and VF_0355 (a putative nonenzymatic protein) were observed (Fig. 3B). None of the other mutants exhibited consistent, significant changes in migration on TBS-Ca relative to that of the wild-type control (Fig. S6). Altogether, TBS-Ca conditions revealed phenotypes for 16 mutants, similar in number to but not fully overlapping that seen with the TBS conditions; both of these conditions were superior to TBS-Mg, which uncovered only 6 mutants with phenotypes. Thus, the use of different medium conditions provided valuable insights into gene function and permitted us to uncover more subtle—or potentially condition-specific—phenotypes.

Inhibition of motility at high calcium occurs primarily via CasA.

Because the calcium conditions provided the greatest insight by identifying the greatest number of mutants with phenotypes, we extended our experiments for a subset of strains by increasing the amount of calcium in the motility agar. Relative to its absence, addition of 10 mM calcium increases migration of ES114, whereas larger amounts, such as 40 mM calcium, impair it (23). We recently reported that a mutant lacking the DGC CasA (VF_1639) migrates robustly through soft agar containing calcium at levels up to 100 mM and produces c-di-GMP in response to calcium in a heterologous microbe (Escherichia coli) (25). While the casA mutant was one of the nine DGC and dual mutants that exhibited increased abilities to migrate through TBS-Ca (10 mM) (Fig. 3), it did not exhibit the greatest increase. Rather, a strain deleted for VF_A0155 exhibited the greatest increase, with the ΔmifA and ΔVF_1200 mutants both migrating more than the ΔcasA mutant. Thus, we wondered whether these mutants would also migrate robustly at increased concentrations of Ca. We tested a subset of these mutants (mifA [VF_0989], VF_1200, casA [VF_1639], VF_A0155, VF_A0381, and mifB [VF_A0959]) on media containing larger amounts of calcium. Despite the substantial migration observed for the VF_A0155 mutant on TBS-Ca (10 mM), neither this nor the other tested mutants phenocopied the casA mutant for migration on larger amounts of Ca (20, 30, or 40 mM) (Fig. 4). These data provide additional support for the key role of CasA in responding to high levels of calcium (25) and suggest that the 10 mM condition may provide distinct information relative to higher calcium levels.

FIG 4.

Migration of selected mutant strains on TBS plates containing different concentrations of Ca. Areas of migration for mutants lacking the following proteins were assessed on TBS containing (A) 20 mM Ca, (B) 30 mM Ca, or (C) 40 mM Ca. Migration on TBS with 20 or 30 mM Ca was assessed at 5.5 h, while migration on TBS with 40 mM was assessed at 6 h. (D) Representative pictures of strains on TBS-Ca (30 mM) at 5.5 h. WT, wild-type strain ES114; the VF/VFA number indicates the gene that is deleted in the mutant. Error bars represent standard deviation. A representative experiment, one of at least two performed, is shown and incorporates three technical replicates. Statistical analyses were done using one-way ANOVA and Dunnett’s test for multiple comparison. ****, P < 0.0001; ***, P ≤ 0.001; *, P ≤ 0.05; ns, not significant.

c-di-GMP levels influenced by Mg and Ca.

Using a c-di-GMP riboswitch biosensor in which the gene for red fluorescent protein (RFP) is positively controlled by c-di-GMP (39, 40), we previously reported that increasing amounts of calcium caused increasing levels of c-di-GMP in a medium distinct from that used here (25). Because ES114 migrates at different rates under the three conditions in this study, TBS, TBS-Mg, and TBS-Ca (23), we wondered whether these conditions resulted in altered levels of c-di-GMP. Levels of RFP, the output of the c-di-GMP biosensor, were modestly but significantly diminished by addition of low levels of Ca (10 mM) relative to unsupplemented TBS, consistent with the slightly increased ability of ES114 to migrate through TBS-Ca (10 mM) relative to TBS (23) (Fig. 5A). In contrast, addition of higher levels of Ca (20 and 30 mM) to TBS increased levels of the biosensor, as seen previously using different medium conditions (25). These data provide an explanation for the differential migration under different amounts of Ca (23).

FIG 5.

Effect of calcium and magnesium on c-di-GMP levels. RFP levels (reflective of c-di-GMP levels) in cells carrying the c-di-GMP biosensor pFY4535 were measured by flow cytometry following growth in TBS supplemented with (A) 10, 20, or 30 mM CaCl₂ (TC10, TC20, and TC30, respectively) or (B) 1 or 35 mM MgSO₄ (TM1 and TM35, respectively) and compared to unsupplemented TBS (T) and each other. Cells were gated on double positive AmCyan/RFP, and the number of cells was normalized to mode. Geometric mean (GeoMean) of the resulting curves was used for quantification. For panels A and B, each of the dots represents biological replicates and the error bars represent standard deviation; the TBS data are the same in the two panels. Statistics were performed by one-way ANOVA (**, P < 0.005; ****, P < 0.0001).

We next probed the impact of the addition of Mg. The 35 mM concentration of Mg has been historically used in V. fischeri motility experiments because this amount approximates that found in seawater and is thus biologically relevant for this marine microbe. We found that addition of 35 mM Mg resulted in substantially decreased levels of RFP, whereas addition of a smaller amount (1 mM Mg) trended lower but did not reach the level of significance (Fig. 5B). These data thus add Mg to the list of signals that control c-di-GMP levels in V. fischeri and provide an explanation for the ability of Mg to robustly promote V. fischeri flagellation and motility (23).

Ca-induced motility defect can be suppressed by flrA point mutation.

In V. cholerae, the master flagellar regulator, FlrA, is subject to control by c-di-GMP (41). FlrA variants with substitutions in the c-di-GMP-binding residues R135 (R135H) and R176 (R176H) exhibit decreased binding to c-di-GMP and thus permit increased motility for specific strains grown under conditions of high c-di-GMP (41). V. fischeri FlrA is similarly required for motility (20, 26), and the R135 and R176 residues are conserved, but the role of c-di-GMP in controlling its activity has not been reported. Thus, we generated strains that expressed wild-type FlrA, FlrA-R135H, or FlrA-R176H from a nonnative location in the chromosome and evaluated their ability to complement the nonmotile flrA mutant. Indeed, expression of each of the three versions of FlrA restored wild-type migration to the flrA mutant on TBS-Mg (Fig. 6A). Moreover, when the strains were inoculated onto TBS-Ca (10 mM) soft agar, the two point-mutant derivatives were far superior in their ability to promote migration relative to the wild-type-complemented strain (Fig. 6B). These data indicate that Ca, presumably due to induction of higher levels of c-di-GMP, inhibits migration via an impact on FlrA activity and that each of the point mutations rendered V. fischeri FlrA “blind” to c-di-GMP. At 20 mM Ca, while the ability of these variants to promote migration relative to wild-type FlrA remained superior, the overall migration of the strains was reduced compared to the 10 mM Ca condition (Fig. 6B and C), indicating that there are likely other processes being affected by the high levels of c-di-GMP. For example, growth of V. fischeri is reduced by high levels of Ca (25).

FIG 6.

Migration of strains containing “c-di-GMP-blind” alleles of flrA. Left: areas of migration of strains containing “c-di-GMP-blind” alleles of flrA were assessed on TBS soft agar containing (A) 35 mM Mg, (B) 10 mM Ca, or (C) 20 mM Ca. Right: Corresponding representative images for ΔflrA flrA⁺ and ΔflrA flrA-R135H strains. TBS-Mg plates were assessed at 4.5 h, while TBS-Ca (10 mM and 20 mM) plates were assessed at 5.5 h. Error bars represent standard deviation. A representative experiment, one of at least two performed, is shown and incorporates three technical replicates. Statistical analyses were done using one-way ANOVA and Dunnett’s test for multiple comparison. ****, P < 0.0001; *, P ≤ 0.05; ns, not significant.

In V. cholerae, observing the effect of these mutations on bacterial motility requires the additional disruption of genes involved in polysaccharide biosynthesis (vps) (41). However, disruption of additional genes was not necessary to observe an effect in V. fischeri. Rather, a simple manipulation of the growth media to induce higher levels of c-di-GMP was sufficient to observe an effect of these point mutations on V. fischeri migration.

Double and triple mutants exhibit more severe motility defects on TBS-Mg.

Finally, we pursued the construction of a limited number of strains with multiple mutations (other than the operon mutants described above) in an effort to generate strains with enhanced phenotypes. We focused our efforts on the investigation of three genes, VF_0494, VF_1603, and VF_2480. The first was of interest because of the consistent and severe defect observed for the ΔVF_0494 mutant across the different media. Similarly, mutants lacking either of the other two genes VF_1603 and VF_2480 exhibited consistent, albeit milder, migration defects as well. Because the proteins of these genes share 46% identity and 69% similarity (Table 5), we wondered if a double mutant would exhibit a greater defect in migration and found that it did (Fig. 7A and B). Furthermore, whereas derivatives in which the ΔVF_0494 mutation was combined with either ΔVF_1603 or ΔVF_2480 retained their ability to migrate, albeit poorly, a combination mutant deleted for all three genes failed to migrate within the 5-h time course (Fig. 7A and B), although migration did occur following prolonged incubation.

FIG 7.

Migration of strains containing double and triple mutations. Areas of migration were assessed on TBS soft agar containing 35 mM Mg (and, for panel C, chloramphenicol). (A and B) Migration was assessed for strains containing double and triple deletions of VF_0494, VF_1603, and VF_2480 at 4.5 h. (C) Migration of the triple mutant (ΔVF_2480 ΔVF_1603 ΔVF_0494) containing either the PDE overexpression plasmid pKV302 or its vector control pKV69 was assessed at 4.5 h; (D) Migration of the single, double, and triple mutants containing or lacking flrA-R135H as indicated was assessed after 6 h. WT, wild-type strain ES114; the VF/VFA number indicates the gene that is deleted in the mutant; the triple (ΔVF_2480 ΔVF_1603 ΔVF_0494) is also shown as ΔΔΔ. Error bars represent standard deviation. A representative experiment, one of at least two performed, is shown and incorporates three technical replicates. Statistical analyses were done using one-way ANOVA and Dunnett’s test for multiple comparison. For panel C, one-way ANOVA and Sidak’s multiple-comparison test were used. For panel D, the comparison was made to the ΔVF_0494 mutant. ****, P < 0.001.

To determine if the observed defects in migration could be attributed to high levels of c-di-GMP, we evaluated the ability of a PDE (VF_0087), introduced on plasmid pKV302, to restore wild-type migration; this PDE has been effective in promoting V. fischeri migration in other contexts (37). Indeed, pKV302 restored some migration of the triple mutant relative to that of the vector control; however, it did not permit wild-type levels of migration (Fig. 7C). It is possible that this PDE could not sufficiently remove the c-di-GMP to fully permit migration, or else that there are other consequences due to the loss of one or more of these proteins. To further understand the consequences of the losses of VF_0494, VF_1603, and VF_2480, we introduced the “c-di-GMP-blind” flrA-R135H allele into the single ΔVF_0494 mutant as well as the double VF_1603/VF_2480 mutant. Migration of the double mutant was robustly restored (Fig. 7D), suggesting that these two genes affect c-di-GMP levels that in turn control FlrA activity. However, migration was only weakly restored to the ΔVF_0494 mutant by the presence of the flrA-R135H; this is similar to our observations that this allele was not sufficient to restore motility in high calcium and suggests that either (i) the c-di-GMP produced by the ΔVF_0494 mutant acts at levels other than FlrA activity to control motility or (ii) VF_0494 has a distinct function in motility besides controlling c-di-GMP levels. Future work will distinguish between these possibilities.

Overall conclusions.

This work identified roles in motility for a subset of 50 V. fischeri genes with putative GGDEF, EAL, and HD-GYP domains. Of note, the use of media supplemented with or lacking added Mg or Ca provided additional insights into the requirements for specific genes in this phenotype (Fig. 8). These cations influence the levels of c-di-GMP corresponding to their impact on V. fischeri’s ability to migrate through soft agar; in particular, we report here the dramatic effect that Mg supplementation exerts over c-di-GMP levels, consistent with its known ability to promote motility of V. fischeri (23). We found that the inhibition of motility caused by the addition of Ca can be suppressed by expression of either of two “c-di-GMP-blind” alleles of the master regulator flrA. Finally, we identified three genes (VF_0494, VF_1603, and VF_2480) that play central roles (potentially somewhat redundant, in the case of VF_1603 and VF_2480) in controlling the ability of V. fischeri to migrate through soft agar.

FIG 8.

Model of genes required for V. fischeri motility under different conditions. Venn diagram depicting the three conditions under which bacterial migration was assessed, TBS-Mg, TBS, and TBS-Ca, and the genes whose loss affects motility under those conditions. The colors of the boxes represent the grouping of the indicated gene: orange, DGC; blue, PDE; green, dual; and gray, nonenzymatic. The color of the text represents the direction of the impact of mutation: up, black; down, white.

Liu et al. performed similar work in V. cholerae, with the disruption of 52 GGDEF- and/or EAL-domain-encoding genes (28). That study reported that only 7 of the 52 mutants exhibited altered motility, with three decreased (cdgJ, rocS, and cdgG mutants) and four increased (cdgH, cdgK, cdgL, and cdgD mutants) under the assay conditions, LB soft agar incubated at 30°C; this small number is similar to that which we observed on TBS-Mg conditions (six mutants). V. fischeri carries homologs for four of those seven genes. Indeed, mutants defective for the homologs of V. cholerae cdgJ (VF_2480), rocS (VF_0494), cdgG (VF_0596), and cdgK (casA [VF_1639]) behaved the same, with decreased or increased migration in soft agar (albeit for ΔcasA only when calcium was present), suggesting a conservation of function across species. In V. cholerae, cdgJ exerted the greatest impact, while the effect of rocS disruption was milder, whereas the opposite was true for mutants of the V. fischeri homologs, with disruption of the rocS homolog VF_0494 exerting the greatest impact on migration. Of note, although V. cholerae carries cdgJ and a cdgJ-like gene, VC_1851, homologous to V. fischeri genes VF_2480 and VF_1603, respectively, a motility defect was not reported for a ΔVC_1851 mutant. It will be of interest to see if a V. cholerae mutant defective for cdgJ and VC_1851 correspondingly exhibits a greater defect in migration than the cdgJ single mutant.

This work focused on GGDEF, EAL, and HD-GYP domain-carrying proteins, so it did not incorporate the study of putative c-di-GMP-binding elements, such as PilZ domain genes (12, 42), riboswitches (43), or the relatively newly identified MshEN domain found in the c-di-GMP-binding MshE (44, 45). V. fischeri carries four PilZ domain genes, at least one putative riboswitch, and MshE (VF_0362) (https://www.ncbi.nlm.nih.gov/Complete_Genomes/c-di-GMP.html) (2, 5, 12, 13). With the foundational work described here, the impact of these and other putative c-di-GMP-binding proteins and riboswitches can be analyzed more readily using mutants either with single or multiple DGC or PDE gene deletions or with the addition of specific amounts of Mg or Ca. Together, this work sheds light on the genes and conditions that influence c-di-GMP-mediated control over motility in V. fischeri and provides a foundation for (i) assessing roles of putative c-di-GMP-binding proteins, (ii) evaluating other c-di-GMP-dependent phenotypes in V. fischeri, (iii) uncovering potential redundancy, and (iv) deciphering signal transduction mechanisms.

MATERIALS AND METHODS

Bioinformatic analyses.

Each of the 50 proteins previously identified as containing GGDEF and/or EAL or HY-GYP domains (see reference 16 and https://www.ncbi.nlm.nih.gov/Complete_Genomes/c-di-GMP.html) (2, 5, 12, 13) was submitted to BLASTP with the default settings (46). The graphical summary and corresponding alignments (47–49) were used to identify possible domains, both c-di-GMP-related (GGDEF, EAL, HD-GYP) domains and others found in the database. These putative domains (see Fig. S1 in the supplemental material) are largely the same as identified previously (16). Analysis of the presence of the GGDEF, EAL, HD-GYP, and I-site motifs were performed using the alignments and the known conserved amino acids. To determine the possible membrane-associated regions (Fig. S2), we submitted protein sequences to TMPred (50) or Protter (51). Finally, to evaluate the similarity between V. fischeri and V. cholerae proteins, V. fischeri sequences were submitted to BLASTP using V. cholerae O1 biovar El Tor strain N16961 (27) in the organism search set. The corresponding V. cholerae proteins were similarly submitted to BLASTP, selecting V. fischeri ES114 (14, 15) in the organism search set; the numbers listed in Table 5 were taken from the resulting alignments in this analysis.

Strains and media.

V. fischeri strains were derived from ES114 (52) and are listed in Table S1. They were routinely cultured in LB-salt (LBS; 1% tryptone, 0.5% yeast extract, 2% sodium chloride, and 50 mM Tris [pH 7.5] [19]) liquid or solidified with agar to a final concentration of 1.5%, with some exceptions: LB (1% tryptone, 0.5% yeast extract, 1% sodium chloride) liquid or solidified with agar to a final concentration of 1.5% was used when certain antibiotics specified below were used. For motility experiments, TBS (1% tryptone, 2% sodium chloride) and TBS solidified with 0.25% agar alone or supplemented with 35 mM magnesium sulfate (Mg) or 10 mM calcium chloride (Ca) were used; other amounts of Mg and Ca were used in specific experiments as indicated. We note that the source of tryptone can affect this phenotype, and thus we consistently used Difco (Gibco) tryptone for all motility experiments. To promote consistency between experiments, the same volume (25 mL) was pipetted into each petri dish on the day prior to a given experiment. For genetic manipulation, cells were grown in TMM (100 mM Tris [pH 7.5], 300 mM NaCl, 0.1% ammonium chloride, 10 mM N-acetylglucosamine, 50 mM MgSO₄, 10 mM KCl, 10 mM CaCl₂, 0.0058% K₂HPO₄, 10 μM ferrous ammonium sulfate) as described below to generate competent cells. Antibiotics were added as needed to the following final concentrations: chloramphenicol (Cm), 1 μg mL⁻¹; erythromycin (Em), 2.5 μg mL⁻¹; gentamicin (Gm), 5 μg mL⁻¹; kanamycin (Kn), 100 μg mL⁻¹ (V. fischeri) or 50 μg mL⁻¹ (E. coli); spectinomycin (Sp), 200 μg mL⁻¹; tetracycline (Tc), 2.5 μg mL⁻¹; and trimethoprim (Tm), 2.5 μg mL⁻¹. LB was used instead of LBS with Sp or Gm supplementation.

E. coli strains were used for conjugation to introduce plasmids into V. fischeri. Plasmids used are listed in Table S2 and include conjugal plasmid pEVS104 (53), Flp recombinase plasmid pKV496 (26), or TfoX-producing plasmids plostfoX (54), plostfoX-Kan (55), or pJJC4 (56). E. coli strains were cultured in LB. Antibiotics were added as needed to the following final concentrations: Cm, 12.5 μg mL⁻¹; Kn, 50 μg mL⁻¹. Thymidine was added as needed to a final concentration of 0.3 mM for use with E. coli thymidine auxotrophic strains.

Construction of mutants.

Strains were constructed as described previously (26, 57) using plasmids and primers listed in Tables S2 and S3, respectively. Briefly, sequences (about 500 bp) upstream and downstream of the gene of interest were amplified using EMD Millipore KOD polymerase in a PCR with primers containing linker sequences. A third PCR product, containing an antibiotic resistance cassette, was amplified from plasmid templates (Table S2) with primers 2089 and 2090 (Table S3). The three PCR products were fused together using a splicing by overlap extension (SOE) technique (58). The composite DNA fragments were introduced by transformation into ES114 containing a tfoX-overexpression plasmid. Competent cells were generated and transformations were carried out as described previously (57), with selection for the antibiotic resistance cassette. The deletion of interest was confirmed using Taq polymerase in a PCR with the outside primers. In some cases, mutants were first generated in an intermediate strain, such as a tfoX-carrying Δqrr1 mutant (56), and genomic DNA from the resulting strain was used in a subsequent transformation to introduce the gene replacement mutation into ES114. Similarly, strains carrying more than one deletion were generated by introducing genomic DNA generated from a marked (with an antibiotic resistance cassette) mutant of interest into a tfoX-overexpressing recipient strain carrying one or more other mutations. Antibiotic resistance markers were flanked by FRT sequences and were resolved using Flp recombinase expressed from pKV496, as described previously (26, 57), to reduce the concern of a polar effect due to the insertion and to readily permit the construction of strains with multiple deletions. This approach leaves a small scar at the deletion site.

Construction of flrA alleles.

PCR was used to introduce point mutations into the flrA gene as follows. To generate PflrA-flrA-R135H, strain KV8290 (26), which carries an flrA deletion (and an FRT scar) and an flrA complementation cassette in the intergenic region between yeiR and glmS (VF_2380-VF_2372), was used as a template in two separate reactions, with primer pairs 2290 & 2256 and 2255 & 1487. The resulting PCR products were combined and used in a PCR SOE reaction to generate full-length flrA with the point mutation, which was subsequently introduced by transformation into a tfoX-overexpressing derivative of KV8232 (26). A similar approach was used to generate PflrA-flrA-R176H, using primer pairs 2290 & 2258 and 2257 & 1487. When strains carrying the complementation constructs were obtained, genomic DNA was isolated and used to transform a tfoX-overexpressing derivative of the ΔflrA strain KV8148 or similarly introduced into the strains with mutations in VF_0494, VF_1603, and/or VF_2480.

Evaluation of growth.

Growth of all strains in the present study was assessed in 96-well plates by the Molecular Devices SpectraMax iD3 multi-mode microplate reader. Kinetic absorbance mode was used to chart strains over 16 h, with reads every 14.5 min and medium shaking for 10 s between reads. Absorbance was measured at a wavelength of 600 nm. The data were then plotted using Excel.

Assessment of motility.

Strains were grown overnight (14 to 16 h) with shaking in TBS medium at 28°C and then subcultured into fresh TBS and incubated with shaking at 28°C for ~3 h. The growth of the strains was estimated by measuring the optical density at 600 nm (OD₆₀₀). The OD₆₀₀ was normalized to 0.2 using TBS as a diluent. An aliquot (10 μL) of each strain was spotted in triplicate onto different petri plates containing TBS soft agar (0.25% agar) with or without supplementation (generally with 35 mM MgSO₄ or 10 mM CaCl₂ but with other concentrations as indicated; technical replicates). Following 4 to 7 h of incubation at 28°C, all plates were photographed. For all motility experiments, each strain was assessed as described above at least two times (two biological replicates each with three technical replicates); those that are shown in Fig. 1 to 3 were performed at least three times or more. For the experiments shown in Fig. 1 to 3 and the corresponding supplemental data Fig. S4 to S6, subsets of the strains were assessed on different days; as a result, each mutant strain was compared to the average migration of the wild-type strain present on the same plates. For the other motility experiments, a representative experiment with three technical replicates performed with all the indicated strains on the same day is shown; at least two such experiments were performed.

Motility spot analysis.

Images were analyzed using Fiji (ImageJ) software. Spot area was measured using the oval selection tool to surround spots at their outermost growth (at the edge of rings when present, or visible growth when rings were less/not apparent); in some cases, the outermost growth was faint. The consistent scale used was 15.4737 pixels/mm, and this was set from the diameter of one spot measured by hand. All images analyzed were taken from the same distance, with 33.3% zoom. For the evaluation of the 50 single mutants and 5 operon mutants, the migration of each mutant was normalized to the migration of the wild-type strain ES114 present on the same plate by dividing mutant migration area by average wild-type migration area. No contrast adjustments or other digital manipulations were used to process images.

c-di-GMP biosensor measurements.

Strains were grown with shaking overnight in TBS at 28°C and subcultured the following morning in various conditions (TBS, TBS with 10, 20, or 30 mM CaCl₂, or TBS with 1 or 35 mM MgSO₄) for 5 h. The biosensor was selected using gentamicin throughout growth. Ten microliters of the subculture was added to 1 mL phosphate-buffered saline (PBS). Samples were then analyzed by flow cytometry on the BD Sciences LSRFortessa system. A threshold of 500 was set for forward scatter (FSC) and side scatter (SSC) in log scale. AmCyan and RFP were visualized using two channels, AmCyan and phycoerythrin (PE)-Texas Red, respectively. Data were compiled using BD Sciences FlowJo version 10.8 and GraphPad Prism version 9.1.2. FSC and SSC gating determined live cells. Double positive AmCyan and RFP cells were further analyzed by creating histograms for RFP. The resulting curves were quantified by the geometric mean fluorescence intensity (MFI) for comparison after normalization to mode. Statistical analyses were performed using one-way analysis of variance (ANOVA).

Statistical measurements.

Statistical analyses for motility experiments were done in GraphPad Prism 9. Assessing the large number of strains and conditions required many experiments over numerous days. To normalize the large number of results derived from the very sensitive soft agar migration assay, each of three technical replicates of a given mutant strain was compared to the average migration of the wild-type strain present on the same plates to obtain a ratio of mutant migration relative to that of the wild-type strain. If the mutant migration did not consistently deviate from wild-type migration by more than 15%, the data were included in the supplemental data section. For all mutants, an unpaired t test was performed to determine the significance of the deviation of each mutant’s migration from that of the average migration of the wild-type strain present on the same three plates. For experiments shown in Fig. 4, 6, and 7, statistical tests included one-way analysis of variance (ANOVA) and Dunnett’s test for multiple comparisons. For Fig. 7C, one-way ANOVA and Sidak’s multiple-comparison test was used.