Calcium-Responsive Diguanylate Cyclase CasA Drives Cellulose-Dependent Biofilm Formation and Inhibits Motility in Vibrio fischeri

ABSTRACT The marine bacterium Vibrio fischeri colonizes its host, the Hawaiian bobtail squid, in a manner requiring both bacterial biofilm formation and motility. The decision to switch between sessile and motile states is often triggered by environmental signals and regulated by the widespread signaling molecule c-di-GMP. Calcium is an environmental signal previously shown to affect both biofilm formation and motility by V. fischeri. In this study, we investigated the link between calcium and c-di-GMP, determining that calcium increases intracellular c-di-GMP dependent on a specific diguanylate cyclase, calcium-sensing protein A (CasA). CasA is activated by calcium, dependent on residues in an N-terminal sensory domain, and synthesizes c-di-GMP through an enzymatic C-terminal domain. CasA is responsible for calcium-dependent inhibition of motility and activation of cellulose-dependent biofilm formation. Calcium regulates cellulose biofilms at the level of transcription, which also requires the transcription factor VpsR. Finally, the Vibrio cholerae CasA homolog, CdgK, is unable to complement CasA and may be inhibited by calcium. Collectively, these results identify CasA as a calcium-responsive regulator, linking an external signal to internal decisions governing behavior, and shed light on divergence between Vibrio spp.

in which red fluorescent protein (RFP) production is controlled by a c-di-GMP-dependent riboswitch (30) (Fig. 1A). Calcium supplementation significantly increased RFP levels in biosensor-containing wild-type (WT) strain ES114 ( Fig. 1B and C). This increase was dose dependent ( Fig. 1B and C), indicating that increasing calcium causes an increase in intracellular c-di-GMP.
The V. fischeri genome contains 50 genes that encode proteins with GGDEF, EAL, or HD-GYP domains, predicted to synthesize, degrade, and/or bind c-di-GMP (31)(32)(33). To determine whether any of these genes was responsible for the calcium-dependent increase in c-di-GMP, we performed a preliminary screen evaluating biosensor activity of strains carrying single deletions of each gene. While several mutants exhibited altered responses to calcium, Calcium-Responsive Diguanylate Cyclase CasA ® we report here our findings for one gene, the putative diguanylate cyclase gene VF_1639 (Fig. 1D), which we termed casA, for calcium-sensing protein A. In contrast to the WT, a DcasA mutant exhibited minimal changes in RFP levels up to 100 mM CaCl 2 ( Fig. 1B and C), suggesting that CasA is critical for calcium-dependent increases in c-di-GMP.
CasA inhibits motility in the presence of calcium. To explore whether loss of the calcium-dependent increase in c-di-GMP in the DcasA mutant translated into c-di-GMP-related phenotypes, we evaluated motility, a phenotype known to be influenced by calcium levels (25). In unsupplemented motility agar, the DcasA mutant phenocopied WT migration over time (Fig. 2). In the presence of calcium, WT migration decreased in a dose-dependent manner, while migration of the DcasA mutant remained constant and unaffected by concentrations up to 100 mM (Fig. 2). As migration depends on both growth and motility, we evaluated growth of the two strains to determine if differences in migration could be attributed to growth defects or advantages. In the presence of either 40 mM or 100 mM calcium, cells entered stationary phase sooner, but the WT and DcasA strains were equally affected (see Fig. S1 in the supplemental material), indicating that any differences in migration are likely due to motility rather than growth. These data suggest that V. fischeri senses calcium and, in a manner that requires CasA, alters its migration accordingly.
Calcium enters V. fischeri cells independent of CasA. V. fischeri exhibits dose-dependent calcium phenotypes that are lost in a DcasA mutant ( Fig. 1 and 2), suggesting the possibility that a DcasA mutant could be defective in calcium uptake. To evaluate if calcium enters cells, and if this uptake depends on CasA function, we expressed aequorin, a cytoplasmic photoprotein that emits light dependent on calcium binding (34,35), in both the parent (Dlux) and DcasA (Dlux) mutant strains. Upon exposure to 40 mM calcium, the parent strain exhibited a spike in light production, indicating that changes in exogenous calcium are reflected intracellularly (Fig. 3). The DcasA strain behaved similarly to the parent strain, albeit displaying a minor decrease in activity. Thus, the inability of a casA mutant to respond to calcium cannot be attributed to an entry defect. Additional work will be necessary to understand the dynamics of calcium uptake; in this study, we focused on elucidating the role of CasA in the signaling network that leads to control of motility and biofilm formation.
CasA contributes to calcium-dependent biofilm formation. V. fischeri biofilms are induced by calcium (20), but it is unknown if these biofilms depend on CasA. We thus assessed biofilm formation by the WT, DcasA, and complemented DcasA strains grown with increasing amounts of calcium using a shaking liquid biofilm assay. In the presence of 10 mM CaCl 2 , as previously seen (20), the WT formed a small cellulose-associated surface-adherent ring, which became more robust with increasing calcium concentrations (Fig. 4A).  Conversely, the DcasA mutant was severely attenuated for ring formation, regardless of calcium levels. Complementation restored the DcasA mutant to a WT phenotype (Fig. 4A). Quantification using crystal violet staining supported the conclusion that the DcasA mutant produced significantly less adherent biomass than both the WT and complemented strains ( Fig. 4B and Fig. S2A).
Because loss of CasA disrupted cellulose-dependent ring formation, we evaluated the same strains using Congo red, which can bind cellulose. On unsupplemented medium, the strains bound similar amounts of dye, resulting in streaks the same shade of red, while a cellulose mutant (DbcsA) failed to bind Congo red, resulting in yellow streaks (Fig. 4C, left). On plates with calcium, however, the DcasA mutant produced yellow streaks, phenocopying the DbcsA mutant, while the WT and complemented DcasA strains streaks were red (Fig. 4C, right). Quantification using ImageJ supported these visual differences (36) (Fig. S2B). Thus, casA is required for calcium-induced cellulose production. Additionally, while polysaccharide can decrease motility through steric hindrance (37), the DbcsA mutant migrated the same as WT, regardless of calcium (Fig. S3), suggesting that cellulose is not responsible for the calcium-mediated motility inhibition.
CasA mediates calcium-dependent increase in bcs transcription. Because calcium induces bcs transcription (20), we hypothesized that this effect could depend on CasA. Indeed, calcium induced significant bcsQ transcription in the WT but not in the DcasA mutant ( Fig. 5A), indicating that CasA is necessary for the calcium-dependent increase in bcs transcription. However, as CasA lacks DNA binding domains (Fig. 1D), it most likely acts indirectly.
VpsR was previously linked to the control of cellulose in V. fischeri (38); thus, we hypothesized that it regulates bcs transcription. Indeed, bcsQ transcription remained low when vpsR was deleted, regardless of calcium, suggesting that VpsR is necessary for bcsQ transcription (Fig. 5A). Complementation restored bcsQ transcription and the calcium-mediated increase in transcription (Fig. 5B). Thus, both bcsQ transcription and its induction by calcium require VpsR.
To determine if CasA was epistatic to VpsR, we overexpressed vpsR in a DcasA DvpsR double mutant. vpsR overexpression increased bcsQ transcription but did not induce a response to calcium, indicating that CasA is needed for the calcium-dependent increase in bcs transcription (Fig. 5B). Additionally, vpsR transcription was not induced by calcium (Fig. S4A), indicating that control by calcium occurs at a different level. Of note, vpsR transcription was unaffected by a casA mutation, and it is negatively autoregulated (Fig. S4B to D). Finally, like the DbcsA mutant, the DvpsR mutant phenocopied the WT strain for motility in response to calcium (Fig. S3), further suggesting that the CasA-mediated inhibition of motility is independent of cellulose. Overall, these data suggest that VpsR is epistatic to CasA for bcs transcription but that the calcium-dependent increase in transcription requires CasA. Calcium-Responsive Diguanylate Cyclase CasA ® CasA is a functional DGC. CasA contains two domains, an N-terminal periplasmic sensory domain and a C-terminal GGDEF domain (Fig. 1D), suggesting that CasA may both sense a signal(s), such as calcium, and respond accordingly by altering c-di-GMP synthesis. To probe CasA function, we evaluated complementation by hemagglutinin (HA) epitope-tagged point mutant alleles driven by a constitutive promoter and introduced into the chromosome of the DcasA mutant relative to the WT-CasA-complemented strain (DcasA/casA 1 ). All the resulting variants were produced (Fig. S5A).
First, DGC activity was evaluated by generating a G410A substitution in the second glycine of the functional GGEEF motif. c-di-GMP biosensor experiments revealed that the CasA-G410A-expressing strain phenocopied its DcasA parent, failing to increase cdi-GMP levels in response to calcium ( Fig. 6A and Fig. S5C). Furthermore, the CasA-G410A variant failed to complement the Congo red and motility phenotypes ( Fig. 6B and C and Fig. S2C). Together, these data suggest that c-di-GMP production is necessary for the calcium-dependent biofilm and motility phenotypes controlled by CasA.
N-terminal sensory domain controls CasA function. The N-terminal domain of CasA contains dCache-1 (calcium and chemotaxis) and MCP-like (methyl-accepting chemotaxis protein) sensory domains, suggesting a sensory function (Fig. 1D). However, CasA lacks conserved motifs typical of such sensory domains, such as N, F, G1, G2, and G3 boxes (39,40). Thus, to identify residues potentially important for function, we aligned CasA with homologous proteins from closely related Vibrio spp., via BLAST (41), and chose several highly conserved residues to mutate, with a focus on aspartates and glutamates because calcium is known to bind such residues: D111, G231, D236, and E293 ( Fig. S6A) (42). Complementation experiments revealed two classes of phenotypes, described below.
First, one variant, CasA-D111A, exhibited increased activity while maintaining a response to calcium. Specifically, in the absence of calcium, this strain produced substantially higher levels of c-di-GMP (Fig. 7A, light blue bars, and Fig. S5D) and increased red color on Congo red ( Fig. 7B and Fig. S2D). In the presence of 10 mM CaCl 2 , this strain exhibited decreased migration compared to that of the DcasA/casA 1 strain (Fig. 7C), indicating a productive response to calcium.
Second, three of the substitutions (G231A, D236A, and E293A) diminished the apparent calcium responsiveness of CasA. All three variants lost the ability to increase c-di-GMP in response to calcium and without calcium maintained similar (G231A and D236A) or slightly higher (E293A) c-di-GMP levels than the DcasA/casA 1 strain ( Fig. 7A and Fig. S5D). However, these variants phenocopied the DcasA/casA 1 strain on Congo red agar containing calcium ( Fig. 7B and Fig. S2D), indicating that they must retain some function. In support of this conclusion, they also exhibited intermediate motility phenotypes, with increased migration compared to that of the DcasA/casA 1 strain but diminished migration relative to the DcasA mutant at either 10 or 20 mM CaCl 2 (Fig. 7C). Combining the D236A and E293A substitutions resulted in a strain with slightly increased migration compared to that of each single mutant strain (Fig. S7), suggesting that this combination of substitutions decreases CasA activity but is insufficient to render CasA completely inactive. Overall, these experiments reveal four sensory domain residues that impact CasA function and calcium-dependent phenotypes, suggesting that this domain may be responsible for sensing calcium as a signal.
CasA is sufficient for the response to calcium. While CasA is responsible for several calcium-dependent phenotypes, it was unclear whether CasA directly senses calcium as a signal or if it requires a partner(s) to sense and respond to calcium. To explore the sufficiency of CasA in response to calcium, we used Escherichia coli as a heterologous system with a high-copy-number plasmid that expressed casA. Low calcium levels exerted minimal effects on c-di-GMP levels in E. coli under our conditions ( Fig. 8A and B). However, when casA was overexpressed, addition of 10 mM calcium resulted in a significant increase in cdi-GMP ( Fig. 8A and B). Furthermore, casA-overexpressing E. coli failed to migrate in soft agar when calcium was added ( Fig. 8C and Fig. S8A). The vector control strain did migrate in the presence of calcium, albeit slower than in its absence, ultimately reaching the same diameter as in the absence of calcium, while the casA-overexpressing strain did not Calcium-Responsive Diguanylate Cyclase CasA ® progress. The calcium-induced delay in migration was not due to a growth defect (Fig. S8B). These data suggest that CasA alone is sufficient to sense calcium and inhibit motility, presumably by producing c-di-GMP.
To determine if this response (i) was dependent on the ability of CasA to make c-di-GMP and (ii) required a functional sensory domain, we evaluated the G410A enzymatic domain and D236A sensory domain variants. E. coli that expressed either variant phenocopied the vector control strain, with no calcium-dependent change in c-di-GMP levels or migration ( Fig. 8 and Fig. S8A). These data suggest that both the periplasmic sensory domain and DGC enzymatic domain are required for CasA to sense and respond to calcium. Thus, CasA appears to be a novel calcium sensor that controls cellulose-dependent biofilm formation via a calcium-mediated induction of c-di-GMP.
CdgK and CasA respond differently to calcium. V. fischeri CasA exhibits significant homology to V. cholerae CdgK; when the CasA sequence was used to probe the V. cholerae genome by BLAST, CdgK (VC1104) was the top hit, and vice versa for CdgK and the V. fischeri genome (Fig. S6B). Additionally, the genes for these two proteins are immediately adjacent to a gene (encoding tRNA-dihydrouridine synthase) that is conserved in both genomes, suggesting that they could be orthologs. CdgK is one of a set of DGCs that work upstream of VpsR to activate vps transcription (13,43). These similarities prompted us to explore if cdgK could complement the DcasA mutant by expressing cdgK in the chromosome of a DcasA mutant. This strain produced CdgK in both the presence and absence of calcium (Fig. S5B). However, it exhibited significantly decreased c-di-GMP levels in response to calcium (Fig. 9A). This suggested that CdgK is an active DGC when expressed in V. fischeri and that this activity may be negatively modulated by exogenous calcium. Consistent with these findings, the cdgK-expressing strain showed increased Congo red binding relative to controls in the absence of calcium (Fig. 9B), but with calcium supplementation, it failed to bind Congo red, phenocopying the DcasA parent ( Fig. 9B and Fig. S2E).
Despite the impact on c-di-GMP levels and Congo red binding, cdgK expression exerted no effect on motility, with or without calcium (Fig. S9). However, when we omitted the motility-promoting magnesium supplement (25), we found that it had masked the impact of cdgK on migration. Whereas the migration patterns of the WT, DcasA, and casA 1 /DcasA strains mimicked those seen on plates containing magnesium (Fig. 6), the cdgK-expressing DcasA strain behaved differently (Fig. 9C). The cdgKexpressing strain exhibited significantly less migration in the absence of calcium, but its migration matched that of the uncomplemented DcasA parent strain when calcium was added, indicating a lack of CdgK activity under calcium conditions (Fig. 9C). Together, the observed increases in c-di-GMP levels and Congo red binding and decrease in motility of the cdgK-expressing strain in the absence of calcium suggest the expression of an active DGC, while the relative lack of observed activity in the presence of calcium suggests that CdgK activity may be inhibited by calcium.

DISCUSSION
This study investigated how calcium changes V. fischeri behavior. We determined that increasing concentrations of calcium resulted in increased cellulose-dependent biofilm formation and decreased motility, corresponding to increasing concentrations of c-di-GMP. CasA, one of 33 putative DGCs encoded by V. fischeri, was solely responsible; the Calcium-Responsive Diguanylate Cyclase CasA ® observed effects were lost in the absence of CasA, and heterologous expression of casA in E. coli similarly resulted in calcium-dependent phenotypes. For biofilm formation, the calcium-dependent increase in bcs transcription required both CasA and VpsR. Finally, the V. cholerae homolog, CdgK, did not complement CasA but instead exhibited an inverse response, with increased activity in the absence of calcium.
c-di-GMP is a widespread bacterial signal that permits many organisms to change their behavior in response to varied internal and external signals. Calcium was one of the first signals identified as an activator of c-di-GMP through inhibition of PDE activity in Komagataeibacter xylinus, even before the discovery of c-di-GMP in 1987 (8,44,45). In the intervening decades, the connection between calcium and c-di-GMP has been investigated in a variety of organisms and signaling pathways. For example, in Mycobacterium tuberculosis, calcium alters PDE activity, affecting growth and survival during macrophage infection (46)(47)(48). In the c-di-GMP-activated Lap systems of Pseudomonas aeruginosa and Legionella pneumophila, calcium activates the protease LapG, which promotes biofilm dispersal (49,50). Migration on motility agar with or without 10 mM calcium. The y axis represents migration distance, and the x axis represents added calcium. Bacteria migrate slower in the presence of calcium, independent of growth ( Fig. S8B), so endpoints were chosen when the vector control (VC) reached approximately 40 mm in diameter, which was at 15 h and 19 h for 0 mM and 10 mM calcium, respectively. The horizontal dotted line represents the diameter of the VC with 0 mM calcium to facilitate a comparison of strain migration. Overexpression of casA resulted in significantly less migration distance relative to that of the vector control strain (**, P = 0.003; ****, P = 0.0001). Neither casA-G410A nor casA-D236A migrated significantly differently than the vector control. Strains shown contain one of the indicated plasmids: pKV69 (VC), pAT100 (casA), pAT101 (casA-G410A), pAT102 (casA-D236A) in E. coli strains GT115 (A and B), or AJW678 (C). Error bars represent standard deviation. Each assay was performed independently at least three times.
Tischler et al.

®
Additionally, similar to what we observed in this study, calcium increases intracellular c-di-GMP levels in V. vulnificus (19). Our work adds to this literature by identifying CasA as a DGC whose activity is induced in response to calcium. Other DGCs and PDEs responsive to calcium in a variety of organisms are likely to be uncovered as the connection between calcium and c-di-GMP continues to be explored.
The architecture of CasA includes a putative N-terminal periplasmic sensory domainputatively involved in binding calcium-and a well-conserved C-terminal cytoplasmic GGDEF enzymatic domain responsible for c-di-GMP production. Residues in both of these domains are required for function. Protein prediction programs identify both dCache-1 (calcium and chemotaxis) and MCP-like (methyl-accepting chemotaxis protein) domains in the N terminus. Class I MCPs have a periplasmic ligand binding domain (LBD), with another, functional domain in the cytoplasm (51), matching the general structure of CasA. A screen of calcium-binding proteins identified aspartates and glutamates as the most common calcium-coordinating residues (42). Consistent with those known interactions, Nterminal CasA residues D236 and E293 likely contribute to sensing/binding calcium in some way, as mutating these residues resulted in partial loss of the calcium response. Conversely, mutation of D111 allowed for an increase in basal CasA function without disrupting the ability to sense/respond to calcium. Structural analysis of CasA could provide insight into where and how calcium binds to the N-terminal LBD. Calcium-Responsive Diguanylate Cyclase CasA ® CasA responds to the calcium signal by synthesizing c-di-GMP to control cellulose production. A wide variety of bacterial species synthesize cellulose polysaccharide, leading to a diversity of bcs operon structures, genes, cellulose products, and regulatory mechanisms. But throughout this heterogeneity, bacterial cellulose remains inextricably linked to c-di-GMP, as all identified BCS complexes contain a BcsA subunit with a PilZ c-di-GMP binding domain (52). E. coli and Salmonella spp. are the best-studied organisms with type II bcs operons, the same kind found in V. fischeri. In V. fischeri, regulation starts at the transcriptional level, where bcs transcription depends on VpsR and is significantly increased by calcium due to the DGC CasA. Without CasA, calcium has no impact on either the internal levels of c-di-GMP or bcs transcription. Conversely, in E. coli and Salmonella spp., bcs transcription is thought to be constitutively active and therefore unregulated (53). However, posttranscription regulation of cellulose by c-di-GMP is well established in these organisms, with c-di-GMP binding to and allosterically activating BcsA promoting synthesis and binding to BcsE, activating BcsG for postsynthetic modification (54,55). While this has not yet been investigated, c-di-GMP almost certainly regulates cellulose synthesis in V. fischeri by binding to and activating the predicted BcsA and BcsE proteins, perhaps due to CasA-relayed calcium and/or additional signals. The PDE BinA is known to impact cellulose polysaccharide and represents another candidate for cellulose regulation at the synthesis or postsynthetic level (56).
Regulation of the bcs locus in V. fischeri contains parallels to regulation of other Vibrio polysaccharide loci and is often directly inverse to regulation in V. cholerae. First, calcium affects transcription of a major polysaccharide locus in at least three different Vibrio spp., activating bcs and brp transcription in V. fischeri and V. vulnificus, respectively, and downregulating vps transcription in V. cholerae (18)(19)(20). Calcium acts to increase c-di-GMP in both V. fischeri and V. vulnificus, activating CasA in V. fischeri to increase bcs transcription (19). The specific calcium-dependent mechanism activating brp transcription is yet unknown but also dependent on c-di-GMP (19). Conversely, in V. cholerae, vps transcription is decreased by calcium through downregulation of vpsR transcription by the calcium-sensing two-component system CarRS (18). Second, the homologous transcriptional regulators VpsR in both V. fischeri and V. cholerae autoregulate their own transcription, albeit in opposite directions, with V. fischeri VpsR negatively autoregulating and V. cholerae VpsR engaging in positive autoregulation (57). Third, the homologous DGCs CasA and CdgK respond oppositely to calcium, with CasA increasing and CdgK decreasing c-di-GMP levels in the presence of calcium, paralleling the effects on bcs and vps transcription. In V. cholerae, VpsR activity is increased by c-di-GMP (11,12). This is likely to be the case for the V. fischeri protein as well; if so, this could account for CasA-mediated induction of bcs transcription in response to calcium. Overall, these differences in regulation speak to the evolutionary divergence between species, with the same signal allowing each organism to adapt to its environmental niches without major functional changes.
In summary, this study identifies CasA as a calcium-sensing DGC in V. fischeri, responsible for inducing cellulose-dependent biofilm formation and inhibiting motility in response to calcium. CasA is required for a calcium-dependent increase in bcs transcription, connecting c-di-GMP to transcriptional control of a type II bcs locus. The V. cholerae homolog, CdgK, and CasA are not equivalent, but the enzymatic activity of both proteins seems to be modulated by calcium, providing insight into how related species can adapt to their specific niches.
Molecular techniques and strain construction. Mutations in ES114 were generated through TfoXmediated transformation (63)(64)(65)(66). Briefly, ;500-bp segments upstream and downstream of genes of interest were PCR amplified using high-fidelity KOD polymerase (Novagen, EMD Millipore), and PCR splicing by overlap extension (SOE [67]) was used to fuse segments to an antibiotic cassette as described previously (68). The fused product was amplified and transformed into the recipient V.  DcasA::FRT DvpsR::FRT-Spec r attTn7::PbcsQ-lacZ Derived from KV9866 using pCMA26 (20) This study KV9929 DcasA::FRT-Spec r IG::PvpsR-lacZ TT KV9573 with gKV8920 This study a Abbreviations: HA, HA epitope tagged; IG, intergenic between yeiR and glmS (adjacent to the Tn7 site); FRT, the antibiotic cassette was resolved using Flp recombinase, leaving a single FRT sequence. b Derivation of strains constructed in this study. TT, TfoX-mediated transformation of a tfoX-overexpressing version of the indicated strain with the indicated genomic DNA (gDNA) or with a PCR-SOE product generated using the indicated primers and templates.
fischeri strain (typically ES114) carrying a TfoX-overproducing plasmid (plostfoX [64], plostfoX-Kan [64], or pJJC4 [66]), and recombinant cells were selected on media containing the appropriate antibiotic. The allelic replacement was confirmed by PCR with outside primers using Promega Taq polymerase (Table 3). After the initial deletion was made, genomic DNA (gDNA) was isolated from the recombinant strains using the Quick-DNA Miniprep plus kit (Zymo Research) and used to introduce the mutation into other desired strain backgrounds. Insertion at the Tn7 site was performed via tetraparental mating (69) between the V. fischeri recipient and three E. coli strains, carrying the conjugal plasmid pEVS104 (70), the Tn7 transposase plasmid pUX-BF13 (71), and pCMA26 (20). Insertions were also introduced adjacent to the Tn7 site at the intergenic (IG) region between yeiR and glmS as previously described (68). These insertions were

TABLE 3 Primers used in this study
Primer no. Sequence a 975 CCTCACCCCAGATGGTTTGGCA made using the PCR amplification and SOE method described above, with genes of interest fused to an upstream Erm r cassette for selection, driven by the constitutive PndrR promoter, and containing an idealized ribosome binding site (RBS). In some cases, the antibiotic resistance cassette was removed from V. fischeri deletion mutants using Flp recombinase, which acts on Flp recombination target (FRT) sequences to delete the intervening sequences, as has previously been shown (72). Overexpression plasmids were constructed by amplifying genes of interest from the IG region of V. fischeri strains, and the resulting PCR product was ligated into the pJET1.2 blunt cloning vector (Thermo Fisher), transformed into chemically competent E. coli DH5a, and selected using Amp. The resulting plasmids were sequenced (Integrated DNA Technologies), purified using the plasmid miniprep kit (Zymo Research), and transformed into chemically competent E. coli FT115 containing either the pFY4535 biosensor or AJW678. The apoaequorin plasmid was synthesized by GenScript using sequence from Aequorea victoria for apoaequorin, clone UTAEQ04, and inserted into plasmid pVSV105 (73). c-di-GMP biosensor assay. Relative c-di-GMP levels were assessed in either LBS or LB broth at 24 or 28°C for V. fischeri and E. coli, respectively, and cultures inoculated from single colonies contained added calcium chloride as indicated. The biosensor was not selected using antibiotics, as it contains a toxin/antitoxin system (30), but Amp was used to select for overexpression plasmids in E. coli strains. Samples were diluted 1:1,000 in phosphate-buffered saline (PBS) and assessed via flow cytometry on a LSRFortessa (BD Biosciences). Forward scatter (FSC) and side scatter (SSC) were collected in a log scale with a threshold of 200, and AmCyan and phycoerythrin (PE)-Texas Red channels were used to measure AmCyan and RFP, respectively. Data were analyzed using FlowJo 10, gating first on live cells as determined by FSC and SSC, then AmCyan to confirm singlets, and finally RFP to assess relative c-di-GMP. This resulting population was used to create representative histograms, and the geometric mean fluorescence intensity (MFI) of each curve was used to quantify and compare samples. Data were graphed using GraphPad Prism 6 and analyzed via linear regression or one-way analysis of variance (ANOVA) as indicated.
Aequorin assay. The DluxCDABEG strain (parent) and its DcasA derivative carrying pAT103 were grown overnight at 28°C in LBS with Cm. The overnight cultures were subcultured 1:100, grown until mid-log phase, and collected and centrifuged. Samples were washed two times and resuspended in PBS, and a limiting amount of coelenterazine (Nanolight Technology) was added to a final concentration of 5 mM. The samples were vortexed and incubated in the dark for 1 h and washed in PBS, and the optical density at 600 nm (OD 600 ) was measured. Samples were normalized to an OD of 0.4 in a white 96-well plate and incubated in the dark for 10 min. The baseline luminescence of the samples was measured with a delay of 1 s for 5.9 min in a luminometer (Veritas microplate luminometer; Turner Biosystems). Calcium was added to a final concentration of 40 mM, and the luciferase activity was measured with a delay of 1 s for approximately 17 min 40 s. The limiting amount of coelenterazine is responsible for the temporal appearance of light production. This assay was performed at least three separate times.
Motility assay. Single colonies were inoculated in either TBS or TB for V. fischeri and E. coli, respectively. Cultures were grown overnight shaking at 28°C, subcultured 1:100 in fresh broth, and incubated with shaking until exponential growth phase. Cultures were normalized to a final OD 600 of 0.2, and 10-ml aliquots were spotted onto soft-agar motility plates supplemented with the desired concentrations of CaCl 2 . E. coli plasmids were maintained with Amp throughout. Plates were incubated at 28°C, and the diameter of each zone of migration was measured and imaged at 4 h unless indicated otherwise for V. fischeri and 15 h and 19 h for E. coli. Pictures were taken using an iPhone 11 front-facing camera. Data were analyzed using one-way ANOVA in GraphPad Prism 6. Each experiment utilizing this assay was performed at least three independent times.
Shaking biofilm assay. To assess calcium-induced biofilm formation under shaking liquid conditions, LBS broth containing between 10 and 100 mM calcium chloride (as indicated) was inoculated with single colonies of V. fischeri strains and grown with shaking overnight at 24°C. For these experiments only, test tubes (13 by 100 mm) were used with a culture volume of 2 ml of LBS broth. For crystal violet staining, 200 ml of a 1% crystal violet solution was added to cultures for 30 min. Tubes were washed with deionized H 2 O and destained with ethanol. OD 600 was measured using a Synergy H1 microplate reader (BioTek). Pictures were captured via an iPhone 12 minicamera, and data are representative of at least 3 independent experiments. Linear regression analysis was performed in GraphPad Prism 6.
Congo red assay. Bacteria were streaked onto LBS plates containing Congo red and Coomassie blue dyes (40 mg ml 21 and 15 mg ml 21 , respectively), and 40 mM calcium as indicated, and grown overnight at 24°C. To better visualize color differences, cells were transferred onto white paper in a replica plating-like approach by briefly smoothing the paper onto the agar plate and then lifting it off (68). The result was photographed with an iPhone 12 minicamera. For quantification, 10-ml aliquots of culture normalized to an OD of 0.2 were spotted and grown as described above. Spots were compared by assessing the gray values via ImageJ (36).
b-Galactosidase assay. Strains carrying a lacZ reporter fusion to the bcsQ or vpsR promoter were grown in duplicate at 24°C in LBS containing calcium chloride as indicated. Strains were subcultured into 20 ml of fresh medium in 125-ml baffled flasks, and samples were collected after 4 h of growth. OD 600 was measured, and cells were resuspended in Z buffer and lysed with chloroform. The b-galactosidase activity of each sample was assayed as described previously (74) and measured using a Synergy H1 microplate reader (BioTek). Assays were performed at least 2 independent times and analyzed via oneway ANOVA in GraphPad Prism 6.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, PDF file, 0.1 MB. FIG S1, PDF file, 0.04 MB.