c-di-GMP-Dependent Regulation of Motility by comFB and comFC

Abstract

ComFB is encoded in the comF operon of Bacillus subtilis, situated between the genes for ComFA and ComFC. The latter two proteins are essential for natural transformation, whereas ComFB is dispensable. We show here that ComFB binds specifically and with high affinity to the second messenger c-di-GMP and that ComFB acts as a c-di-GMP receptor to inhibit swarming and swimming motility, apparently by interfering with flagellar activity. We show further that in the absence of ComFC, swarming is completely abrogated by a mechanism that requires FB. These results reveal a new c-di-GMP regulatory system that controls motility independently of MotI.

INTRODUCTION

Self-propelled bacterial movement through liquids and across surfaces is widely conserved despite the high energetic cost of synthesizing and operating the relevant organelles, typically flagella (1). This cost requires that the formation and function of flagella be tightly regulated, often through mechanisms involving the second messenger bis-(3′-5′)-cyclic diguanylic acid (c-di-GMP). c-di-GMP is ubiquitous in bacteria, regulating diverse activities beyond motility, including biofilm formation, exopolysaccharide synthesis, virulence, DNA replication, and cell size (2, 3). To fulfill its regulatory roles, c-di-GMP interacts with protein or RNA receptors. The receptors then act on phenotype-specific targets. For example, the B. subtilis (Bsu) receptor MotI, when bound to c-di-GMP, interacts with the flagellar stator protein MotA to uncouple flagella from the motor force that enables swimming (4).

Recently, a cyanobacterial protein with partial sequence and structural similarity to the Bacillus subtilis (Bsu) protein ComFB (FB) was shown to bind c-di-GMP (5). In fact, the FB of Bsu also binds c-di-GMP with specificity and high affinity (shown here and in (6)). While MotI is a PilZ-type c-di-GMP binding receptor with canonical RR and (N/D)hSXXG motifs (7), FB represents a new class of cyclic-dimeric nucleoside monophosphate-binding proteins, widespread in bacterial phyla (5, 6, 8). Although FB is dispensable for genetic transformation (9), it is embedded in the comF operon of Bsu, which encodes two proteins, ComFA (FA) and ComFC (FC), needed for the transport of transforming DNA to the cytoplasm (Fig. 1A) (9–12). Firmicute comF operons consistently encode FA and FC, but only some firmicutes encode FB; a search using eggNOG v6 (13) revealed that although 447 known firmicute species encode close orthologs of FB^Bsu, these orthologs are sporadically distributed. For example, even among the bacilli, FB is encoded by only a few close relatives of Bsu. At the same time, orthologs with partial sequence similarity to FB^Bsu are encoded in distantly related bacteria, including cyanobacteria, Treponema denticola, and Vibrio cholerae (5, 6). Although the structure of the Bsu FB has been solved (9), its role remained obscure. The discovery that it binds c-di-GMP suggested a regulatory role, and we have now shown that FB inhibits flagellar motility.

Figure 1.

Organization of the comF operon. (A) In Bsu, the comFA, comFB, and comFC open reading frames are transcribed from P_FA, a ComK-dependent promoter. P_FB, a ComK-independent promoter, drives comFB and FC, which overlap. The existence of P_FB is documented as discussed in the text. The +1 A-residue for P_FB is shown in boldface, and the proposed promoter sequence features are underlined. (B) Two examples of bacteria that encode ComFB in their comF operons, demonstrating the overlap of their comFB and comFC coding sequences.

In Bsu, the comF operon is activated by ComK, which drives the transcription of about 20 genes needed for transformation and about 80 additional genes with largely uncharacterized roles (14–16). The semi-dormant transformation-competent “K-state” conferred by ComK is established during the transition to the stationary growth phase in a minority of cells because ComK itself is bimodally expressed (17–21). The comF operon of Bsu is preceded by a ComK-dependent promoter (P_FA), which drives the transcription of comFA, comFB, and comFC (Fig. 1A) (10). A second promoter (P_FB) drives transcription of the comFB and comFC genes (Fig. 1A). P_B is not associated with the K-box sequences characteristic of ComK-dependent promoters (22), and is active in LB medium, where ComK is not produced. Two reports, both based on data from LB-grown cells, revealed the existence of the P_FB promoter. A tiled microarray study of global transcription, with a reported resolution of 22 bases, demonstrated an uptick in transcription during growth at position 3,642,139 in the Bsu genome, 32 bases upstream of the comFB open reading frame (23). Rend-seq (24, 25), with single-nucleotide resolution, showed the presence of a transcriptional start site nearby at chromosomal position 3,642,135 (https://rendseq.org/). Plausible SigA promoter sequences are located at an appropriate distance upstream from this start site (Fig. 1A).

Here, we report that FB binds to c-di-GMP with high affinity and specificity and that it inhibits swimming and swarming motility when bound to c-di-GMP, independently of MotI. The deletion of comFC, with which comFB is translationally coupled, prevents swarming, suggesting that the joint expression of the ComFB and ComFC proteins may play a role in the regulation of motility during growth and in the K-state.

RESULTS

ComFB binds c-di-GMP specifically and with high affinity

As noted above, orthologs of FB have been reported to bind cyclic dinucleotides, including c-di-GMP and c-di-AMP (5, 6, 8). To gain further insights into the affinity of Bsu FB for c-di-GMP, we heterologously purified FB in its native dimeric state (Fig. S1) and employed isothermal titration calorimetry (ITC). The raw titration data of c-di-GMP to FB were fitted using a single-site binding model to determine the average dissociation constant (K_D) for all available binding sites (Table 1, Fig. 2A). ITC experiments revealed that Bsu FB has a high affinity for c-di-GMP (K_D ~140 nM; Table 1), comparable to that of other high-affinity c-di-GMP binders, e.g., Pilz and FimX (26). In contrast, the FB K_D value for binding to c-di-AMP (Fig. 2B) (6), which Bsu also produces (27), is nearly 500-fold higher (Table 1), above the likely physiological concentration for this dinucleotide. Also, c-di-GMP competes efficiently with c-di-AMP (at a saturating concentration of 150 μM) for FB binding (Table 1) (6).

Table 1.

c-di-GMP binding to Bus FB as determined by ITC

Titrant	Competitor	Average KD (μM)a	ΔH (kcal mol−1)b	ΔG (kcal mol−1)c
c-di-GMP	----	0.14 ± 0.05	−9.6 ± 0.8	−9.4 ± 0.15
c-di-GMP	2 mM Zn2+	1.2 + 0.6	−5.7 ± 1.5	−12.2 ± 4.2
c-di-GMP (6)	150 μM c-di-AMP	0.24 ± 0.09	−6	−8.1 ± 1.8
c-di-AMP	----	69.1 ± 139	9.6	−5.68
c-di-GMP	150 μM GTP	0.11 + 0.09	−8.8 ± 0.5	−16.4 ± 0.1
c-di-GMP	500 μM GTP	0.2 ± 0.15	−8.1 ± 1.3	−9.2 ± 0.5
c-di-GMP	150 μM ATP	0.12 + 0.1	−9.2 ± 0.6	−13.5 ± 4.1
c-di-GMP	500 μM ATP	0.4 ± 0.16	−6.5 ± 1.25	−8.8 ± 0.2
c-di-GMP	150 μM ADP	0.25 + 0.17	−8.2 ± 0.7	−9.2 ± 0.7
c-di-GMP	500 μM ADP	0.3 ± 0.03	−7.4 ± 0.2	−8.9 ± 0.06
c-di-GMP	150 μM cGMP	1.9 + 0.1	−3.7 ± 1.5	−7.8 ± 0.3
c-di-GMP	150 μM cAMP	0.3 + 0.06	−6.9	−8.94
cGMP	----	1.1 + 0.7	−2.5 ± 0.9	−8.2 ± 0.4
cGMP	150 μM c-di-GMP	No binding
GTP	----	No binding
ATP	----
ADP	----	No binding
cAMP	----	No binding

The raw isothermal titration calorimetry (ITC) data were fitted using a one-site binding model for monomeric Bsu FB.

^aThe dissociation constant (K_D) values correspond to the mean of at least two independent experiments ± standard deviation. All titrations were performed in 50 mM Tris-HCl buffer (pH 7.9).

^bEnthalpy.

^cGibbs free energy. For competition binding assays, FB protein was preincubated with 150 μM or 500 μM of the ‘competitor’ nucleotide as indicated, followed by titration with c-di-GMP.

Fig. 2.

Figure 2. Isothermal titration calorimetry (ITC) analysis of c-di-GMP, cGMP, and c-di-AMP binding to FB, as indicated. Upper panels show the raw ITC data in the form of heat produced during the titration of the respective nucleotide on the FB protein; lower panels show the binding isotherms and the best-fit curves according to the one binding site model. (A-C) ITC analysis of c-di-GMP (A), c-di-AMP (B), and cGMP (C) binding to FB in the absence of competitors. (D-I) ITC analysis of c-di-GMP binding to FB in the presence of ATP (D), ADP (E), GTP (F), cAMP (G), cGMP (H), or Zn²⁺ (I).

To further explore the relation between c-di-GMP binding and that of other more abundant nucleotides (ATP, ADP, and GTP), and other second messengers (cGMP and cAMP), we tested the ability of c-di-GMP to bind FB in the presence of saturating concentrations of these compounds. GTP, ATP, and ADP could not bind to FB (Fig. S2), and their presence at 150 or 500 μM did not significantly affect the c-di-GMP affinity to FB (Table 1, Figs. S1 and S2). However, at the high concentration of 500 μM, the binding enthalpy of c-di-GMP was slightly reduced (Table 1; compare isotherm of Fig. 2A with 2D), causing a slight increase in the K_D values. The second messenger cAMP was not able to bind FB (Fig. S2G), but its presence at 150 μM triggered a trivial decrease in the c-di-GMP binding enthalpy to FB (Table 1), yielding weaker isotherms (compare the isotherm of Fig. 2A with 2G). In contrast, the second messenger cGMP could bind FB with a K_D of ~1.1 μM. The cGMP titration curve showed a complex biphasic binding curve; in the first injections, the enthalpy increased until a maximum was reached at a cGMP concentration of ~40 μM (Fig. 2C). Subsequently, the heat isotherm gradually decreased, indicating gradual saturation of FB by cGMP (Fig. 2C). Therefore, it seems that cGMP is unable to efficiently occupy the binding sites of FB. Consistent with this conclusion, the binding of cGMP was completely abolished in the presence of a saturating concentration of c-di-GMP (150 μM) (Fig S2H). Moreover, in the presence of 150 μM cGMP, c-di-GMP was able to compete efficiently and replace cGMP for FB binding, although the apparent K_D for c-di-GMP binding increased by more than 10-fold in the presence of cGMP (Table 1) as indicated by a decrease in the binding enthalpy (Table 1; compare isotherm of Fig. 2A with 2H). cGMP is produced by some bacteria (8, 28), but to our knowledge, has never been reported to exist in the Firmicutes, and its weak interaction with FB is of doubtful biological relevance. The crystal structure of Bsu FB contained a Zn²⁺ atom (one per monomer; PDB: 4WAI) that appeared to stabilize the protein (9) but could potentially interfere with c-di-GMP binding. Indeed, the binding assay performed in the presence of 2 mM ZnCl₂ resulted in a ~8-fold higher apparent K_D for c-di-GMP (Table 1, Fig. 2I). Collectively, competition studies show that GTP, ATP, ADP, and cAMP have negligible effects on c-di-GMP binding, while cGMP shows a modest competition effect, establishing the specificity of c-di-GMP binding to FB. These results indicate that FB can be regarded as a c-di-GMP binding receptor in Bsu.

The chromosomal organization of comFB in Bsu

In Bsu, the stop codon of comFB is embedded within the coding sequence of comFC with an overlap of 11 codons (Fig. 1B). A similar arrangement is found in closely related bacilli that encode FB and FC, as depicted for B. licheniformis in Fig. 1B. Such arrangements are often indicative of translational coupling (29) so that the translation of comFC may be augmented by, or even depend on, the translation of comFB.

ComFB inhibits swarming when bound to c-di-GMP

c-di-GMP binding receptors have been implicated in inhibiting swimming and swarming motility in several bacterial species (4, 30, 31). Although swimming through low-viscosity agar depends on flagellar activity, it also depends on chemotaxis (32), and a swimming assay cannot distinguish between these dependencies. Swarming, bacterial expansion across a solid surface, does not depend on chemotaxis per se (32) but rather on hyperflagellation, the association of individual cells into “rafts,” and the secretion of surfactin, which decreases surface tension. In most of the present study, we used the biologically relevant undomesticated NCIB 3610 background because domesticated strains have lost the ability to swarm (32).

To determine whether the overexpression of FB affected swarming motility, we expressed FB from the amyE locus of NCIB 3610 under the control of a constitutive promoter (P_c) unrelated to P_FA or P_FB, using either a weak or a strong ribosomal binding site (RBS). The P_c promoter was derived from the inducible P_hyperspac promoter but lacks the lacO operator site (33). The “weak” RBS construct (amyE::P_c(w)FB) employed the native FB ribosomal binding site, while the “strong” construct (amyE::P_c(s)FB) utilized the RBS of the highly expressed gapA gene. For these comparisons, we tested each construct both with and without the deletion of pdeH, which encodes the only c-di-GMP phosphodiesterase in Bsu (33, 34). In the absence of this enzyme, the cellular concentration of c-di-GMP increases (33). As reported previously (32), we confirmed that the inactivation of pdeH alone moderately decreased swarming (Fig. 3A). Although the amyE::Pc(w)FB construct did not affect swarming on its own, it exhibited a greater inhibitory effect when combined with ΔpdeH compared to ΔpdeH alone (Fig. 3A).

Figure 3.

FB, together with c-di-GMP, inhibits swarming. Each of the panels is a representative of at least three repeated swarming experiments, performed as described using 0.7% LB-agar plates. The expansion diameter of the cell sheet was measured at intervals and plotted against time. The symbols are consistent across all five panels, with empty symbols indicating ΔpdeH strains. MT: empty vector (pKB149 (33)) in amyE. The constructs amyE::P_c(w)FB and amyE::P_c(s)FB are described in the text. Δcyc denotes strains deleted for the three diguanlylate cyclases of Bsu (dgcP, dgcK, dgcW).

In contrast to these results, the amyE::P_C(s)FB construct partially inhibited swarming on its own and eliminated it when pdeH was deleted (Fig. 3B). Western blotting showed that expression from the weak construct was slightly stronger than from the native locus, while the strong construct produced about 100-fold more than the native locus (Fig. 4). (The samples loaded on the first three lanes of the gel shown to the right of Fig. 4, were diluted 20-fold before loading, and a 10-fold reduced volume was loaded compared to the gel shown to the left of Fig. 4). The blot also shows that the expression of FB from the overproduction construct was not increased when pdeH was eliminated, suggesting that the inhibition of swarming with excess FB is not due to enhanced FB expression but may be dependent on binding to c-di-GMP.

Fig. 4.

The expression of FB and its regulation. Western blots were performed with an anti-serum raised against purified FB. As a loading control, the tops of the membranes were cut off after blotting and developed with an antiserum raised against elongation factor G (EFG). We have noticed a consistent signal, indicated by arrows, at the position expected for a dimer of FB. The Δ(FB-FC) strain was included in each panel to confirm the identity of the FB signal. The vertical dashed lines indicate that intervening lanes were omitted from a gel. For the gel to the right of panel A, the first three lanes were loaded with lysates diluted 20-fold, and only 2 μl were loaded compared to 10–20 μl of undiluted lysate for the other lanes. This is evident from the weaker EFG signals. The empty vector strain produces a low level of FB, which is increased moderately in the amyE::P_c(w)FB strain and about 100-fold over the wild-type level in the amyE::P_c(s)FB strain.

To further investigate the role of c-di-GMP, we used the amyE::P_c(s)FB construct in a background lacking dgcW, dgcP, and dgcK, the three diguanylate cyclases of Bsu (33). In this c-di-GMP null strain (ΔCYC), the overproduction of FB did not affect swarming (Fig. 3C), and Western blotting showed that the absence of the cyclases did not reduce the amount of FB (Fig. 4). Since FB is a high-affinity c-di-GMP binding protein (Table 1), these results strongly support a dose-dependent inhibition of swarming by FB when bound to c-di-GMP. MotI is a c-di-GMP receptor that prevents motility by directly interacting with the flagellar stator protein MotA (4). To determine whether MotI mediates the inhibitory effect of FB, we placed the amyE::P_c(s)FB construct in a ΔmotI ΔpdeH background. As shown in Fig. 3D, the inhibitory effect of FB on swarming is not reversed when motI is deleted. Thus, Bsu encodes two c-di-GMP receptors that independently inhibit swarming.

FB inhibits the function of flagella, not their production.

Swarming requires hyperflagellation, surfactin, raft formation, and the expression of SwrA (32, 35). Swimming, the flagellar-dependent movement of cells through a low-viscosity medium, requires neither SwrA nor surfactin (36). Thus, swimming can be measured in domesticated strains, which are deficient in the production of these factors (32). As shown in Fig. 5A, the amyE::P_c(s)FB construct inhibited swimming motility in “wild-type” domesticated Bsu even without the inactivation of pdeH. This contrasts with the partial effect of the same construct in the swarming assay (Fig. 3B) and may be due to the presence of an intact swrA gene in the NCIB3610 background, which biases the cells towards increased production of flagella (37). However, when the three diguanylate cyclases were deleted in the amyE::P_c(s)FB strain, swimming was not inhibited (Fig. 5B). We conclude that FB does not inhibit motility by interfering with surfactin or chemotaxis but likely impacts either the production or the activity of flagella. Next, we addressed whether the overexpression of FB interferes with flagella formation. Labeling with maleimide-Alexa fluor 488 to visualize the flagella of wild-type and amyE::P_c-R^S-comFB ΔpdeH strains carrying the hag^T209C allele (38) revealed that the overproduction of FB, which prevents swimming, does not obviously interfere with flagellar production even when pdeH is deleted (Fig. 5C), although we cannot exclude subtle effects on the number of flagella.

Fig. 5.

FB inhibits swimming but not the production of flagella. (A and B) The indicated strains were tested for swimming in 0.3% agar. The images were collected when the empty vector (MT) control strains had reached or nearly reached the outer limits of the Petri dishes. (C) The two indicated strains, both carrying the hag^T209C allele, were reacted with Alexa Fluor 488 C5 maleimide (green), stained with the membrane dye FM4–64 (red), and imaged as described in Materials and Methods.

ComFC is required for swarming

The overlap of comFB and comFC suggests that the cognate proteins may be functionally related. To investigate the role of FC, we used a markerless construct in which the entire comFC coding sequence had been removed. This construct was derived from a ΔcomFC::erm deletion by removing the Erm-cassette with plasmid pDR244 (39). This was necessary because the cassette has a downward-facing promoter and readthrough from the P_FA promoter during the K-state has been shown to enhance production of FlgM, leading to the loss of Hag (flagellin) synthesis (40). Surprisingly, the markerless ΔcomFC strain did not swarm (Fig. 3E). For complementation of ΔcomFC, an FC-expressing construct was placed at the thrC locus, using the P_c promoter and the gapA RBS. An identical ectopic construct with a FLAG epitope at the C-terminus of FC confirmed that the fusion protein was expressed (Fig. S3). The thrC::Pc(s)FC construct fully complemented ΔcomFC for swarming, showing that the observed phenotype is due to the lack of FC protein (Fig. 3E). When both comFB and comFC were deleted, swarming was unaffected, suggesting that the swarming-deficient phenotype of ΔcomFC is dependent on FB. These results suggest that FC may normally mitigate the ability of FB to inhibit swarming.

DISCUSSION

FB represents a new class of c-di-GMP receptors (6, 8), joining three others encoded by Bsu that regulate aspects of cell physiology; MotI inhibits motility (4), while YdaK and YkuI have been proposed to regulate the production of uncharacterized extracellular polysaccharides (41, 42). The four known c-di-GMP receptors of Bsu are classified into distinct protein families; YdaK is a degenerate GGDEF protein lacking cyclase activity, YkuI is an EAL domain protein without phosphodiesterase activity, and MotI is a PilZ domain protein (33, 34).

The overexpression of FB, when it is bound to c-di-GMP, appears to inhibit flagellar activity. The available evidence argues strongly for these conclusions; the elimination of c-di-GMP degradation by inactivation of pdeH augments FB’s activity (Fig. 3A and B), FB has no effect on motility in a cyclase null background (Fig. 3C), FB is a specific and high-affinity c-di-GMP binding protein ((6) and Table 1), and no obvious diminution in flagellar abundance was noted when FB was overexpressed even when pdeH was eliminated (Fig. 5C). This study has also revealed that FC is required for swarming, perhaps acting to mitigate the effect of FB. Clearly, the overlapping comFB and comFC genes are part of a previously undetected mechanism for the regulation of motility.

The most intriguing aspects of FB and FC are their biological roles; why are they transcribed from the P_FB promoter in the absence of the K-state, and why is comFB transcribed in the K-state despite its dispensability for transformation (9)? Bsu encodes three known regulators of flagellar function: EpsE, MotI, and FB. EpsE contributes to the synthesis of a matrix polysaccharide and is transcriptionally activated as cells begin biofilm formation. In addition to its biosynthetic activity, EpsE, which does not appear to bind c-di-GMP, acts as a molecular clutch, most likely by binding to FliG, disrupting its interaction with MotA, and this activity benefits cells transitioning to a sessile state (38, 43). In contrast, MotI acts directly on MotA (4), but the rationale behind its activity and that of FB remains unclear. Since MotI and FB require c-di-GMP for their activity, they may moderate flagellar activity in response to elevated concentrations of the dinucleotide. For example, the repression of pdeH by Spo0A-P (34, 44), a regulator of various forms of development, will cause an increase in c-di-GMP as Spo0A becomes phosphorylated. This rise in c-di-GMP is expected to facilitate binding to FB by mass action, leading to reduced motility in coordination with Spo0A-P-regulated developmental processes. Given that the cellular concentration of c-di-GMP varies between cells of Bsu (44), the inhibition by FB (and MotI) may enable cells to reversibly and stochastically switch between motile and non-motile states while retaining their valuable flagella. It is noteworthy that the measured c-di-GMP binding affinities of MotI, FB, and YdaK are 0.011 μM, 0.14 μM, and 1.1 μM (Table 1 and (33)). It is possible that these differences in binding affinities allow the receptors to respond within overlapping but distinct ranges of c-di-GMP concentrations.

As noted above, the comF operon lies immediately upstream of a motility-related locus that encodes FlgM, the anti-sigma factor for SigD (45). Read-through from the ComK-dependent P_FA promoter in K-state cells has been shown to inhibit flagellar synthesis via the enhanced expression of flgM (40). It has also been demonstrated that flagellar activity favors transitions to the K-state by downregulating the phosphorylation of the response regulator DegU (46, 47). Thus, while flagellar activity facilitates entry into the K-state, further flagellar synthesis is restricted by read-through to flgM after cells have transcribed the comF operon. Thus, the K-state is linked to the regulation of both flagellar synthesis and function. Interestingly, a cyanobacterial ComFB ortholog binds c-di-AMP to control natural competence (48), underscoring the relationship of this receptor with transformation, albeit in this case with a different second messenger. Gaining a biological understanding of these connections will necessitate deeper insights into the FB-FC regulatory system, identifying the FB target for motility inhibition, and exploring the potential roles of FB in the control of the K-state itself.

MATERIALS AND METHODS

Strains, strain construction, and growth conditions.

Growth was in LB Medium, except that competent cells for transformation were grown as described (49). Antibiotics were added to solid media at the following concentrations: chloramphenicol (7 μg/ml), erythromycin and kanamycin (5 μg/ml), spectinomycin (100 μg/ml), and tetracycline (20 μg/ml). Two backgrounds were used as indicated: the domesticated parental strain (IS75) is a derivative of strain 168 and is auxotrophic for histidine, leucine, and methionine.

The parental undomesticated strain (BD5776) is a comI mutant derivative of NCIB 3610, gifted by D. Kearns. Constructs were generally introduced by transformation into both backgrounds, but in some cases by transduction with bacteriophage SPP1. Strains are listed in Table S1 and plasmids in Table S2. Construction of the ΔdgcK dgcW::erm dgcP::tet knockout strain in the IS75 background was initiated by transformation with genomic DNA from DK392 (dgcP::tet), a kind gift from D. Kearns, with selection for tetracycline. The Koo et al knockout collection (39) was used as a source of dgcW::erm and dgcK::erm deletions. dgcK::erm was first introduced into the dgcP::tet strain, and the erythromycin-resistance cassette was looped out using pDR244, as described (39). The resulting transformant was then transformed with dgcW::erm to produce the triple knockout erythromycin- and tetracycline-resistant strain (Table S1). For the insertion of constructs into amyE, plasmid DNA was linearized using ScaI and transformed into IS75 or BD5776 strains, selecting for spectinomycin resistance. The resulting transformants were tested for insertion in amyE using starch plates stained with iodine. For the insertion into thrC, plasmid DNA was cut with PvuI and transformed into the appropriate Bsu strain, selecting for erythromycin resistance, followed by tests for threonine auxotrophy and the absence of spectinomycin resistance.

Measurement of c-di-GMP binding to FB

Bacillus subtilis ComFB protein was heterologously expressed as an 8xHis-tagged protein in E. coli as described previously (6). Briefly, the 8xHis-tagged constructs were expressed by overnight cultivation of E. coli cells at 20°C in the presence of 0.5 mM IPTG and purified by immobilized metal affinity chromatography using Ni²⁺-Sepharose resin (Cytiva^™), followed by size exclusion chromatography on a Superdex 200 Increase 10/300 GL column (GE HealthCare, Munich, Germany), as described previously (50). Protein purity was assessed by Coomassie-stained SDS-PAGE, and protein concentrations were determined using Bradford assay.

Binding of c-di-GMP to FB protein was analyzed by isothermal titration calorimetry (ITC), as described previously (51). In a buffer composed of (50 mM Tris/HCl, pH 8.0, 300 mM NaCl, 0.55 mM EDTA), the ITC measurements were conducted using a MicroCal PEAQ-ITC instrument (Malvern Panalytical, Westborough, MA, USA), at 25 °C, with a reference power of 10 μcal/s. Dissociation constants K_D and ΔH values were calculated using the single binding site model with the MicroCal PEAQ-ITC Analysis Software (Malvern Panalytical). In the competition binding assays, the FB protein was incubated with 150 μM or 500 μM of (ATP, ADP, GTP, cAMP or cGMP) nucleotides and titrated against 1 mM of c-di-GMP.

Before ITC experiments were conducted, analytical size exclusion chromatography coupled to a multiangle light scattering (MALS) detector was carried out at room temperature, as described previously (50, 52, 53) using the ÄKTA purifier (GE Healthcare) on Superdex^™ 200 column. ASTRA software (Wyatt) was used for data analysis and molecular mass calculations using the MALS data. The elution volume was plotted against the UV signal and molecular mass.

Molecular constructs

amyE::P_c(w)FB:

The oligonucleotides FBWHinF (GCCATGAAGCTTCCGATGATGAGGAGCTG) and FBBamR (ATCGCCGGATCCGCGAATCACATAATAAACAGATC) were used in a PCR to isolate a comFB fragment containing the native RBS sequence that was then inserted between the HindIII and BamHI sites of pKB149 (33).

amyE::P_c(s)FB:

The oligonucleotide FBSHinF (GCCATGAAGCTTAAAGGAGGAAACAATCATGCTTGTCAATTCAAAAGA) was used in a PCR with FBBamR to produce a comFB fragment containing the gapA RBS that was inserted between the HindIII and BamHI sites of pKB149.

thrC::P_c(s)FB and thrC::P_c(s)FB3xFLAG:

These constructs were built using NEBuilder HiFi (New England Biolabs), and inserted into the HindIII site of pDG1664 for insertion in thrC (54). To obtain the Pc gapA RBS sequence, the oligonucleotides P_c(s)F (TGCTGCCTTCGGATCCTAGAAGCTTCTACACAGCCCAGTCCAG) and P_c(s)-R (CGTTTGACATGATTGTTTCCTCCTTTAAGCTTC) were used, with Pc(s)FB plasmid DNA as the template. To obtain the FC sequence, the oligonucleotides FC F (GGAAACAATCATGTCAAACGAAGCAGGAG) and FC-R (TAGGGTTATCGAATTCGATAAGCTTTTAGCTTCTGATCAAGGTAAAAG were used. For the equivalent FLAG-tagged construct, the FC-R-FLAG oligonucleotide (TAGGGTTATCGAATTCGATAAGCTTTCACTTGTCGTCATCGTC) was used. For the latter two PCR reactions, the template DNA was from wild-type Bsu and BD9425 (Table S1) genomic DNA, respectively.

ΔcomFC:

Strain BKE35450 (ΔcomFC::erm) (39) distributed by the Bacillus Genetic Stock Center, was the source of this deletion. After transforming into the appropriate backgrounds, the Errn-resistance cassette with its promoter was removed, using plasmid pDR244, as described (39).

Swarm and Swim tests.

Swarm expansion was measured using 0.7% LB agar Petri dishes as described by Gao et al (33). The plates were dried for about 10 minutes after spotting in the center with each strain and then incubated at 37 °C in a humidified incubator by placing a large water-filled tray on the bottom shelf. The diameter at the outermost edge of the swarm was measured at intervals. The swim assays were performed essentially as described by Hall et al (36). Petri dishes containing 0.3% LB agar were toothpick-inoculated with the desired strain and then incubated in a humidified incubator overnight at 33 °C, followed by an interval at 37 °C. The plates were imaged in a ChemiDoc MP Bio-Rad imager when the fastest-swimming strain had nearly reached the perimeter of its plate.

Western blotting.

Anti-FB Western blotting from 12.5% tricine-SDS-polyacrylamide gels (55) was carried out by semidry blotting onto nitrocellulose membranes. The blots were developed using HRP-coupled goat anti-rabbit secondary antiserum (for FB and EFG) or goat anti-mouse secondary antiserum (for FC-3xFLAG) from Abcam and ECL kits from Amersham. The images were recorded with a Bio-Rad ChemiDoc MP imager. For loading controls, the top of the membrane was cut off after blotting and developed separately with anti-elongation factor G (anti-EFG) antiserum, a kind gift from Jonathan Dworkin (Columbia University Medical School). The anti-FB polyclonal antiserum was raised in rabbits by the Thermofisher custom antibody service using FB protein, purified as described by the Burton lab (9), using the His-tagged expression construct kindly provided by B. Burton.

Flagellar labeling and microscopy

For the fluorescence microscopy of flagella filaments, cells were grown at 37 °C in LB broth until they reached an OD600 of 0.5–0.8. One milliliter of broth culture was harvested and resuspended in 50 μl of PBS containing 5 μg ml⁻¹ Alexa Fluor 488 C5 maleimide (Thermofisher), and I μg/ml FM4–64. This suspension was incubated in the dark at room temperature for 5 min. Cells were then washed gently twice with 1 ml of PBS and resuspended to an OD600 of 10. Samples were observed by spotting 1 μl of the suspension on an agarose pad. Images were collected using a Nikon Eclipse Ti microscope equipped with an Orca Flash 4.0 Digital camera (Hamamatsu) with a Nikon TIRF 1.45 NA Plan Neofluor 100 oil immersion objective. NIS-Elements AR (v 4.40, Nikon) software was used for image acquisition.

IMPORTANCE.

Bacterial motility is subject to tight regulation, and the second messenger c-di-GMP is often involved in the production and activity of flagella. Revealing the mechanisms of these regulatory pathways is broadly important for understanding bacterial motility and of c-di-GMP-related processes. We show that ComFB is a specific, high-affinity receptor for c-di-GMP that decreases the activity of flagella to control swarming and swimming motility in Bacillus subtilis.