Contact with the CsrA Core Is Required for Allosteric Inhibition by FliW in Bacillus subtilis

CsrA is a small dimeric protein that binds RNA and is one of the few known examples of transcript-specific protein regulators of translation in bacteria. A protein called FliW binds to and antagonizes CsrA to govern flagellin homeostasis and flagellar assembly. Despite having a high-resolution three-dimensional structure of the FliW-CsrA complex, the mechanism of noncompetitive inhibition remains unresolved. Here, we identify FliW residues required for antagonism and we find that the residues make a linear connection in the complex from initial binding interaction with CsrA to a critical salt bridge near the core of the CsrA dimer. We propose that the salt bridge represents an allosteric contact that distorts the CsrA core to prevent RNA binding.

ABSTRACT

The RNA-binding protein CsrA is a posttranscriptional regulator encoded by genomes throughout the bacterial phylogeny. In the gammaproteobacteria, the activity of CsrA is inhibited by small RNAs that competitively sequester CsrA binding. In contrast, the firmicute Bacillus subtilis encodes a protein inhibitor of CsrA called FliW, which noncompetitively inhibits CsrA activity but for which the precise mechanism of antagonism is unclear. Here, we take an unbiased genetic approach to identify residues of FliW important for CsrA inhibition and these residues fall into two distinct spatial and functional classes. Most loss-of-function alleles mutated FliW residues surrounding the critical regulatory CsrA residue N55 and abolished interaction between the two proteins. Two loss-of-function alleles, however, mutated FliW residues near the CsrA core dimerization domain and maintained interaction with CsrA. One of the FliW alleles reversed a residue charge to disrupt a salt bridge with the CsrA core, and a compensatory charge reversal in the CsrA partner residue restored both the salt bridge and antagonism. We propose a model in which the initial interaction between FliW and CsrA is necessary but not sufficient for antagonism, and for which salt bridge formation with, and deformation of, the CsrA core domain is likely required to allosterically abolish RNA-binding activity.

IMPORTANCE CsrA is a small dimeric protein that binds RNA and is one of the few known examples of transcript-specific protein regulators of translation in bacteria. A protein called FliW binds to and antagonizes CsrA to govern flagellin homeostasis and flagellar assembly. Despite having a high-resolution three-dimensional structure of the FliW-CsrA complex, the mechanism of noncompetitive inhibition remains unresolved. Here, we identify FliW residues required for antagonism and we find that the residues make a linear connection in the complex from initial binding interaction with CsrA to a critical salt bridge near the core of the CsrA dimer. We propose that the salt bridge represents an allosteric contact that distorts the CsrA core to prevent RNA binding.

INTRODUCTION

CsrA is an RNA-binding protein and posttranscriptional regulator in bacteria (1). Primarily an inhibitor of translation, CsrA dimers bind to hairpin structures in a transcript such that binding often occludes the Shine-Dalgarno ribosome binding site (2–5). In enteric gammaproteobacteria, CsrA pleiotropically controls translation of a wide range of genes involved in physiological transitions during pathogenesis (6–10). Repression is relieved by the expression of small RNAs (sRNA) with multiple CsrA-binding sites that act as competitive inhibitors and sequester CsrA (11–15). In the Gram-positive bacterium Bacillus subtilis, however, CsrA appears dedicated to flagellar regulation and inhibits translation of perhaps only a single gene, encoding flagellin (16–18). The flagellin transcript is the most abundant transcript in the cell and thus would be difficult to outcompete if CsrA was regulated by a competitive mechanism (19). Instead, CsrA is stoichiometrically antagonized by the protein FliW (17, 18, 20, 21). FliW seems to be the ancestral form of CsrA regulation and is conserved in the spirochaetes and epsilonproteobacteria (22–26).

Up to two molecules of FliW can bind to a CsrA dimer, but only a single FliW is necessary to inhibit RNA binding by a poorly understood mechanism (18, 21). Genetic analysis has indicated that FliW does not bind to the same surface of CsrA that binds RNA and that mutation of a remote residue, CsrA N55, abolishes both inhibition and FliW interaction (20). Allosteric inhibition was invoked, as mutation of one residue buried in the core of the CsrA dimer, CsrA^I14, abolished FliW inhibition but preserved FliW interaction (20). A three-dimensional structure of the FliW₂-CsrA^dimer was resolved, but conformational changes in CsrA were unclear, as the structure of the CsrA^dimer alone or the CsrA^dimer bound to RNA has not been resolved for comparison (21). Thus, even with a high-resolution structure of the inhibited complex, it is not clear how FliW binding antagonizes CsrA-RNA interaction. A model was proposed in which a series of negatively charged residues in an unstructured loop of FliW might electrostatically repel the negatively charged backbone of RNA and prevent CsrA binding (21). Although both structural and genetic data suggested that FliW binding was necessary but not sufficient to inhibit CsrA, the mechanism of FliW-mediated inhibition remains unknown.

Here, we show that mutation of the negatively charged series of residues in the FliW loop did not abolish FliW activity and thus the mechanism of inhibition was likely not electrostatic repulsion. FliW activity was also not reduced by deletion of residues from loop regions near the RNA-binding pocket, and thus the mechanism of inhibition was likely not steric occlusion. Instead, an unbiased genetic approach to identify loss-of-function alleles of FliW found two types of mutations that abolished inhibition of CsrA. The first type of mutation altered residues that were either proximal to, or in contact with, CsrA N55 and disrupted interaction between the two proteins. The second type of mutation altered residues that were proximal to core of the CsrA^dimer but preserved FliW-CsrA interaction. Combined, the residues required for inhibition formed a linear series of connections in the structure, suggesting that binding of FliW provokes a conformational change that distorts either one or both of the CsrA^dimer RNA-binding pockets.

RESULTS

Electrostatic and steric inhibition likely does not contribute to FliW antagonism of CsrA.

Genetic and structural analysis indicated that FliW inhibited CsrA binding to RNA by interacting with a surface of CsrA that does not overlap the RNA-binding site (Fig. 1) (20, 21). One hypothesis for the mechanism of inhibition invoked a loop of the FliW protein with a strong negative charge, which when modeled with CsrA (from a distantly related organism) bound to RNA suggested possible electrostatic interference with the RNA backbone (Fig. 1A) (21). To determine whether the negative loop of FliW was important for the release of flagellin (hag) transcript from CsrA, we simultaneously mutated all five negatively charged residues to alanine (fliW^neg5A) and phenotypically tested for the ability to antagonize CsrA. Deletion of fliW leads to a nonswarming phenotype, as uninhibited CsrA restricts flagellin synthesis to levels insufficient for flagellar assembly (Fig. 2A) (17). Ectopic expression of an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible wild-type copy of fliW restored motility by restoring CsrA antagonism (Fig. 2A). Expression of the fliW^neg5A allele restored motility to a fliW deletion background in a manner indistinguishable from the expression of the wild type (Fig. 2A). We conclude that the negative charges within the loop are not required for FliW to inhibit CsrA. We further conclude that FliW inhibits CsrA through a mechanism other than electrostatic repulsion.

FIG 1.

Residues altered by FliW loss-of-function mutations localize to two distinct regions in the FliW-CsrA complex. Shown are three-dimensional structures of the FliW-CsrA complex from Geobacillus thermodenitrificans (PDB 5DMB) (21) projected from the side view (left) and top view (right). The FliW proteins are colored blue and the CsrA dimer is colored lavender. “RNA” indicates the region in which RNA would otherwise be bound. (A) Location of the residues constituting the “negative loop.” FliW negatively charged residues within the loop (E71, D73, D75, E76, and E80) are shown in space fill and colored green. (B) Location of the residues that were deleted in both the “negative loop” shown in space fill and colored green and the “C-terminal loop” shown in space fill and colored gray. (C) Location of the residues constituting the “CsrA N55 proximal cluster.” FliW residues within the cluster that were mutated (P26, F58, N108, A111, and Q124) are shown in space fill, colored cyan, and indicated by a caret. CsrA residue N55 is shown in space fill and colored red. (D) Location of the residues constituting the “CsrA core proximal cluster.” FliW residues within the cluster that were mutated (A122 and K123) are shown in space fill, colored orange, and indicated by a caret. Note that FliW residue 122 of the G. thermodenitrificans is a glycine and not an alanine. CsrA residue D41 is shown in space fill and colored red. The location of each residue in FliW which, when mutated, conferred a loss-of-function mutation can be found individually depicted in Fig. S5 in the supplemental material.

FIG 2.

Loss-of-function mutations in fliW reduce or abolish swarming motility. Quantitative swarm expansion assays of artificially induced fliW alleles are shown. Each point is the average of three replicates. FliW expression was induced throughout growth and swarming by the addition of 1 mM IPTG. The following strains were used to generate the panels: DK3237 (fliW; black circles), DK2371 (fliW amyE::P_IPTG-fliW; white circles), and DK7342 (fliW amyE::P_IPTG-fliW^{neg5A(E71A,D73A,D75A,E76A,E80A)}; gray circles) (A); DK2665 (fliW-csrA; black diamonds), DK7567 (fliW-csrA amyE::P_IPTG-fliW-csrA; white diamonds), and DK8324 (fliW-csrA amyE::P_IPTG-fliW-csrA^N55D; gray diamonds) (B); DK8236 (fliW-csrA amyE::P_IPTG-fliW^L20P-csrA; black squares), DK8244 (fliW-csrA amyE::P_IPTG-fliW^F28S-csrA; white squares), and DK8237 (fliW-csrA amyE::P_IPTG-fliW^A111P-csrA; gray squares) (C); DK8245 (fliW-csrA amyE::P_IPTG-fliW^N108T-csrA; black squares) and DK8326 (fliW-csrA amyE::P_IPTG-fliW^N108Y-csrA; white squares) (D); DK8238 (fliW-csrA amyE::P_IPTG-fliW^Q124H-csrA; black diamonds), DK8325 (fliW-csrA amyE::P_IPTG-fliW^Q124R-csrA; white diamonds), and DK8345 (fliW-csrA amyE::P_IPTG-fliW^Q124H-csrA^N55D; gray diamonds) (E); and DK7762 (fliW-csrA amyE::P_IPTG-fliW^K123E-csrA; black diamonds), DK7761 (fliW-csrA amyE::P_IPTG-fliW-csr^AD41K; white diamonds), and DK7717 (fliW-csrA amyE::P_IPTG-fliW^K123E-csrA^D41K; gray diamonds) (F).

While the negatively charged residues in the loop were not important for FliW inhibition, the loop region might instead promote release of the hag transcript through steric occlusion. To test the role of steric occlusion, the negatively charged loop was shortened by deletion of two different pairs of the negatively charged residues (FliW^Δ71,73 and FliW^Δ75–76) (Fig. 1B). Expression of both truncated proteins in a fliW deletion background restored motility to wild-type levels, suggesting that the length of the negatively charged loop was irrelevant (Fig. S1). Reexamination of the three-dimensional FliW-CsrA complex identified a second loop of FliW near the C terminus that might cause a potential steric clash with the RNA-binding pocket (Fig. 1B; indicated by space filled and gray-colored residues). To further test the role of steric occlusion, the C-terminal loop was shortened by deletion of two different pairs of residues (FliW^Δ128–129 and FliW^Δ130–131) (Fig. 1B). Again, expression of both truncated proteins in a fliW deletion background restored motility to wild-type levels, suggesting that the length of the C-terminal loop was also irrelevant (Fig. S1). It is difficult to eliminate a steric occlusion mechanism without a three-dimensional structure of B. subtilis CsrA bound to RNA, but our results suggest that the wild-type lengths of both the negatively charged and the C-terminal loop of FliW are not required for inhibition of CsrA.

Two regions of FliW are required for FliW antagonism of CsrA.

To explore FliW-mediated antagonism in an unbiased approach, random loss-of-function mutations in FliW were isolated which rendered the protein incapable of inhibiting CsrA. The phenotypic basis for the genetic screen was the aforementioned test for swarming motility (Fig. 2A) (17). To identify alleles of fliW that could not inhibit CsrA, the IPTG-inducible fliW construct was mutagenized by error-prone PCR and introduced into the fliW deletion background. The pool was then screened for loss-of-function alleles by gridding individual colonies onto swarm agar and then seeking colonies that failed to restore motility upon induction. Over 3,000 colonies were screened and 92 colonies were found to be nonmotile, suggesting they carried loss-of-function alleles in the IPTG-inducible copy of fliW.

One way in which an allele of fliW would fail to complement the native fliW deletion is if the IPTG-inducible copy failed to produce stable FliW protein. Indeed, Western blot analysis indicated that 77 of the 92 candidate isolates failed to produce wild-type levels of protein either due to defects in the IPTG-inducible promoter or due to mutations in FliW that reduced protein stability, and all such mutants were discarded from further analysis. The inducible fliW construct was sequenced in all 15 of the remaining mutants that produced levels of FliW protein that were either equal to or greater than the amount produced by wild type (Fig. S2 and S4A). After discarding sibling sequences, two alleles had single point missense mutations (fliW^Q124R and fliW^N108Y), and six alleles had double missense mutations (fliW^L20P,A122V, fliW^I25T,P26S, fliW^F28S,S53P, fliW^F58S,Q124H, fliW^N108T,M120T, and fliW^A111P,K123E).

Previous work indicated that FliW antagonizes CsrA in a 1:1 stoichiometric ratio and the level of FliW to CsrA is maintained through translational coupling (18). To further validate the loss-of-function alleles of fliW while maintaining the appropriate stoichiometric ratio of FliW to CsrA, all alleles were rebuilt as single mutations in a construct that expressed both fliW and csrA from an IPTG-inducible promoter and integrated at an ectopic site (amyE::P_IPTG-fliW-csrA), in a fliW-csrA double-deletion background (Fig. 2B to F; Fig. S3 and S4B). Both the fliW-csrA double-deletion mutant and the double-deletion mutant that was complemented with an IPTG-induced copy of wild-type fliW-csrA were proficient for motility (Fig. 2B). Motility was lost, however, when wild-type fliW and an allele of csrA that cannot be inhibited by FliW (csrA^N55D) were coinduced (Fig. 2B). Next, each loss-of-function fliW mutation was tested for the ability to maintain motility in the fliW-csrA double-deletion background to determine which, if any, of the mutations were sufficient to abolish FliW activity (Fig. 2B to F; Fig. S3A to F).

The 14 reconstructed mutants were separated based on their effect on motility. Six of the mutated proteins (FliW^L20P, FliW^I25T, FliW^F28S FliW^S53P, FliW^M120T, and FliW^Q124H) had little to no effect on FliW function as indicated by minor changes in motility proficiency and were discarded from further study (Fig. 2C; Fig. S3A to E). FliW activity was abolished, however, in the remaining eight mutants (FliW^P26S, FliW^F58S, FliW^N108T, FliW^N108Y, FliW^A111P, FliW^A122V, FliW^K123E, and FliW^Q124R) as indicated by the loss of motility (Fig. 2C to F; Fig. S3A and F). We note that two positions, N108 and Q124, were mutated twice with two different amino acid substitutions, and while N108 was allele independent, Q124 was allele specific such that an arginine substitution abolished activity but a histidine substitution did not (Fig. 2D and E). Ultimately, 8 missense alleles that conferred loss of function to FliW were retained for further study (Fig. S5). We conclude that the ability of FliW to antagonize CsrA can be abrogated or reduced by a variety of single point mutations located throughout the protein.

To separate the remaining mutations, the location of each mutation was mapped onto the FliW structure within the 3-dimensional FliW₂-CsrA^dimer complex (21) (PDB 5DMB). Most of the fliW alleles involved changes to one or more residues (P26, F58, N108, A111P, and Q124R) that clustered near residue N55 of CsrA, which was previously identified to be required for inhibition by, and binding of, FliW (Fig. 1C) (20). Residues A122 and K123, however, were located in a region of FliW that was not near residue N55 of CsrA but rather were proximal to the CsrA core dimerization domain (Fig. 1D). Based on structural analysis, the residues required for inhibition were spatially segregated into two groups, one of which we will refer to as the “CsrA N55 proximal cluster” (Fig. 1C) and the other we will refer to as the “CsrA core proximal cluster” (Fig. 1D). We hypothesized that the two different FliW clusters of residues might be defective in CsrA inhibition for different reasons.

Binding of FliW is necessary but not sufficient to inhibit CsrA.

One way in which residues of FliW might be required for CsrA inhibition is if the residues were required for FliW-CsrA interaction. To test for FliW-CsrA interaction in vivo, each of the wild-type and mutant fliW alleles were artificially expressed with csrA at an ectopic site on the chromosome in a fliW-csrA-hag triple mutant background and treated with the cross-linker formaldehyde to trap protein interactions (Fig. 3). Note that FliW also interacts with Hag, and hag was deleted in this background to emphasize FliW-CsrA interaction. The lysates were then resolved on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and detected by Western blotting. In the absence of cross-linker, wild-type FliW and CsrA both ran at their predicted molecular weights of 16 kDa and 8 kDa, respectively (Fig. S6). In the presence of cross-linker, however, an additional higher molecular weight band of 24 kDa appeared that contained both FliW and CsrA, trapping a 1:1 FliW-CsrA complex, as previously reported (Fig. 3) (20). Moreover, the 24 -kDa band was severely reduced when FliW was expressed with the csrA^N55D allele previously reported to abolish interaction between the two proteins (Fig. 3) (20). Expression of wild-type CsrA with the FliW alleles of the CsrA N55 proximal cluster, FliW^P26S, FliW^F58S, FliW^N108T, FliW^N108Y, FliW^A111P, and FliW^Q124R, also reduced or abolished the 24-kDa FliW-CsrA complex, suggesting that, like CsrA N55 itself, they were required for protein-protein interaction (Fig. 3 and Fig. S6).

FIG 3.

Mutations of FliW CsrA N55 proximal residues abolish CsrA interaction in vivo, while mutations of FliW CsrA core proximal residues do not. Whole-cell lysates were grown in the presence of 0.1 mM IPTG, cross-linked with 0.3% formaldehyde, and subjected to Western blot analysis separately using primary antibodies against FliW (top) and CsrA (bottom). Samples were prepared from strains containing a triple deletion of fliW, csrA, and hag at their respective native sites (fliWcsrAhag). The introduction of the hag deletion was to abolish FliW-Hag interaction and thus reduce the complexity of the interacting species. Translational coupling of fliW and csrA was maintained by coexpressing both proteins from the same IPTG-inducible promoter integrated at an ectopic site (amyE::P_IPTG-fliW-csrA). The csrA and fliW allele is expressed at the top of the panel using either “+” for wild type or the substitution as indicated. The locations of the CsrA monomer (CsrA₁), FliW monomer (FliW₁), and FliW-CsrA monomer complex (FliW₁CsrA₁) are indicated. The CsrA dimer (CsrA₂) is not observed, likely because the interaction is too tight to permit formaldehyde cross-linking in this analysis. An additional species (labeled ??) is observed in cells expressing FliW^K123E, but the molecular nature of this species is unknown. The following strains were used to generate lysates for both panels. DK7974 (fliW-csrA-hag amyE::P_IPTG-fliW-csrA), DK8303 (fliW-csrA-hag amyE::P_IPTG-fliW-csrA^N55D), DK8282 (fliW-csrA-hag amyE::P_IPTG-fliW^P26S-csrA), (fliW-csrA-hag amyE::P_IPTG-fliW^F58S-csrA), DK8304 (fliW-csrA-hag amyE::P_IPTG-fliW^N108T-csrA), DK8113 (fliW-csrA-hag amyE::P_IPTG-fliW^N108Y-csrA), DK8330 (fliW-csrA-hag amyE::P_IPTG-fliW^A111P-csrA), DK8241 (fliW-csrA-hag amyE::P_IPTG-fliW^A122V-csrA), DK8286 (fliW-csrA-hag amyE::P_IPTG-fliW^K123E-csrA), and DK8108 (fliW-csrA-hag amyE::P_IPTG-fliW^Q124R-csrA (left); DK8379 (fliW-csrA-hag amyE::P_IPTG-fliW-csrA^D41K) and DK8378 (fliW-csrA-hag amyE::P_IPTG-fliW^K123E-csrA^D41K) (right). The uncropped image of each Western blot can be seen in Fig. S6 in the supplemental material.

To further test the interactions between the FliW alleles above and CsrA in vitro, a biochemical protein pulldown experiment followed by Western blotting was performed (Fig. 4A). Glutathione S-transferase (GST)-tagged CsrA (GST-CsrA) was purified and loaded onto glutathione-Sepharose resin and incubated with either purified untagged wild-type FliW or mutant protein. Two candidate mutants, FliW^N108Y and FliW^Q124R, were chosen because both were altered in residues mutated twice each with two separate substitutions (FliW^N108Y/T and FliW^Q124R/H), and the mutated residues were highly conserved compared to other FliW homologs (Fig. 5B). After mixing CsrA with FliW, the supernatant was retained, the pellet was washed 4 times, and both proteins were eluted by boiling the pelleted resin in SDS-PAGE loading buffer (Fig. 4A; Fig. S7A to C). Wild-type FliW was retained in the pellet containing GST-CsrA and only accumulated in the supernatant when the amount of FliW was in stoichiometric excess (Fig. 4A; Fig. S7A). In contrast, both FliW^N108Y and FliW^Q124R were found abundantly in the supernatant and were absent from the GST-CsrA pellet fraction (Fig. 4A; Fig. S7B and C). We conclude that FliW^N108Y and FliW^Q124R do not interact with CsrA either in vivo or in vitro, and that residues N108 and Q124 are both required for interaction. We infer that, like the two representative alleles, the cluster of FliW residues proximal to CsrA N55 are all likely important for the interaction between the two proteins.

FIG 4.

Differential binding of loss-of-function FliW alleles to CsrA^WT and CsrA^D41K in vitro. A protein pulldown assay using purified GST-CsrA^WT or GST-CsrA^D41K proteins at the indicated amounts was loaded onto a glutathione-Sepharose column and incubated with the indicated amounts of purified FliW^WT, FliW^N108Y, FliW^Q124R, or FliW^K123E protein in the presence of BSA. Samples were subjected to Western blot analysis and probed with primary antibodies to FliW and CsrA. “Supernatant” indicates the proteins that failed to bind to the beads, and “pellet” indicates the proteins that remained bound to the beads following a series of washes. (A) Testing the ability of FliW^WT and FliW loss-of-function mutations to interact with GST-CsrA^WT. (B) Testing the ability of FliW and FliW^K123E to interact with GST-CsrA^D41K. Note that in some instances CsrA can be seen in the supernatant lanes and this is due to resin being accidentally collected while taking the supernatant sample. A representative CsrA blot is shown at the top of each panel. The associated CsrA blot for each FliW allele can be seen in Fig. S7 in the supplemental material.

FIG 5.

The residues important for CsrA binding and antagonism by FliW form a linear series of connections in space. (A) Structural presentation of FliW residues and CsrA residues identified here and reported previously (20, 21). The CsrA dimer is colored in lavender and FliW proteins are colored in blue. “RNA” indicates the region in which RNA would otherwise be bound. Side view (left) and a zoomed-in view from the top (right) are provided. The FliW residues Q124 (space filled cyan) and K123 (space filled orange) important for CsrA binding and inhibition, respectively, and CsrA residues N55 and D41 (space filled red) required for binding and inhibition, respectively, mapped onto the FliW-CsrA complex structure from G. thermodenitrificans (PDB 5DMB; 21). CsrA residue I14 (space filled red) is also indicated, as mutation to a methionine at this position previously showed that changes to the CsrA core region abolished FliW inhibition but not binding (20). (B) The following organisms were used for a multiple sequence alignment of both FliW (top) and CsrA (bottom): Bacillus subtilis (Bsu; NCIB 3610), Bacillus halotolerans (Bha; ATCC 25096), Geobacillus thermodenitrificans (Gth; NG-80), Treponema pallidum (Tpa; SS14), Thermotoga lettingae (Tle; TMO), Campylobacter jejuni (Cje; ATCC 700819), Helicobacter pylori (Hpy, J99), Clostridium difficile (Cdi; CD13), and Borrelia burgdorferi (Bbu; B31). Positions mutated in Bacillus subtilis FliW or CsrA are indicated above the multiple sequence alignment such that the colored circles match the colored residues in the structural analyses shown in Fig. 1, in panel A, and in Fig. S5. Green circles indicate residues in the FliW “negative loop” (boxed in green); gray circles indicate residues in the “C-terminal loop” that were deleted; cyan circles indicate residues in FliW that were mutated near the residue N55 of CsrA “CsrA N55 proximal cluster”; orange circles indicate residues that were mutated in FliW near the CsrA core “CsrA core proximal cluster”; red circles indicate residues in CsrA that are important for either FliW binding (N55) or FliW inhibition (I14 and D41).

The loss of interaction between CsrA and the FliW alleles FliW^N108Y and FliW^Q124R could be the result of subtle folding issues that did not result in gross protein instability. To assess the folding state of each mutated protein in vitro, wild-type protein and both mutant proteins were subjected to circular dichroism analysis. There was little difference between the circular dichroism spectra of wild-type FliW protein and either FliW^N108Y or FliW^Q124R, suggesting that the structure of each was largely intact (Fig. S8). While in vitro analysis does not directly report on in vivo folding states, combined with in vivo protein stability, we infer that the mutant defect in CsrA interaction is not due to either gross or subtle changes in FliW structure. We further conclude that the loss of interaction between FliW and CsrA is due to the disruption of residues critical for the interaction.

Structural analysis indicated that residue Q124 of FliW appeared to form hydrogen bonds with residue N55 of CsrA, suggesting that these two residues form an interaction couple (Fig. 5A). Mutation of either position to a charged residue disrupted interaction, and we wondered whether interaction could be restored if both FliW^Q124R and CsrA^N55D were expressed in the same cell to form an alternate salt bridge interaction. Simultaneous expression of both mutant alleles failed to support motility in a fliW-csrA double-deletion background, suggesting that replacement of hydrogen bonding interaction with an artificial salt bridge was incapable of restoring productive interaction between the two proteins (Fig. 2E). Expression of FliW^Q124H and FliW^Q124K with CsrA^N55D was similarly insufficient to restore motility (Fig. S3G). Perhaps consistent with the inability to artificially restore interaction, we noted that each of these residues are invariant in a sample of FliW-CsrA-encoding organisms taken from throughout the bacterial phylogeny (Fig. 5B). We conclude that glutamine at position 124 of FliW and asparagine at position 55 of CsrA are critical for the interaction, likely because the two residues make direct contact with one another.

Inhibition of CsrA by FliW depends on the formation of a core-proximal salt bridge.

Residues mutated within the FliW CsrA N55 proximal cluster reduced or abolished interaction with CsrA, but mutation of residues in the CsrA core proximal cluster preserved interaction with CsrA to levels comparable to that of the wild type by in vivo cross-linking (Fig. 3; Fig. S6). To test for interaction in vitro, the FliW^K123E protein was purified, mixed with GST-CsrA in a protein pulldown assay, and detected by Western blotting (Fig. 4A). The FliW^K123E protein was predominantly found in the supernatant with little to no FliW visible in the GST-CsrA pellet (Fig. 4A; Fig. S7D). Thus, mutation of the FliW residue K123 to a glutamate did not strictly abolish interaction with CsrA, as the interaction could be trapped by a chemical cross-linker, but the residue appeared to be required for stabilizing the complex, as the mutant protein was unable to remain in contact with CsrA in a purified in vitro system. We conclude that the residues mutated in the FliW CsrA core proximal cluster behave differently from residues mutated within the FliW CsrA N55 proximal cluster and are likely necessary beyond simple protein interaction.

Structural analysis indicated that the positively charged FliW residue K123 might form a salt bridge with the negatively charged CsrA residue D41 (Fig. 1D; Fig. 5A). We note that the FliW^K123E allele reversed the charge at that position, perhaps resulting in loss of contact with CsrA residue D41 by electrostatic repulsion. Consistent with the idea that the salt bridge is important, motility was inhibited when a form of CsrA was expressed in which residue D41 was changed to an oppositely charged lysine by site-directed mutagenesis (CsrA^D41K) (Fig. 2F). As with FliW^K123E, CsrA^D41K appeared to maintain FliW-CsrA interaction in vivo when trapped by chemical cross-linking (Fig. 3; Fig. S6) and poorly retained wild-type FliW in the in vitro protein pulldown assays (Fig. 4B; Fig. S7E). Motility was restored, however, when both charge-reversed alleles were expressed simultaneously, suggesting that both the salt bridge formation and FliW antagonism had been restored (Fig. 2F). Moreover, combined expression of both mutants also showed interaction in in vivo cross-linking detected by Western blotting (Fig. 3; Fig. S6), and strong interaction between the two alleles was restored in in vitro protein pulldown (Fig. 4B; Fig. S7F). We conclude that the formation of a salt bridge between FliW and CsrA near the CsrA core region is important for antagonism. We infer that this contact plays an allosteric role secondary to the initial interaction event. Finally, we note that the salt bridge appears to be conserved only in the Bacillales but clearly tolerates alteration such that different core contacts could conceivably substitute in other organisms (Fig. 5B).

DISCUSSION

CsrA is one of a few proteins in bacteria that binds to mRNA and regulates translation (1–5). Some bacteria, such as the gammaproteobacteria, relieve CsrA inhibition by expressing competitive inhibitor small RNAs with iterations of the CsrA-binding site that act by sequestration (11–15). Most other bacteria, however, antagonize the RNA-binding activity of CsrA by expressing the protein FliW (17, 22–26). Genetic and structural analysis supports the notion that FliW antagonism functions differently than sRNA inhibition, in that CsrA is antagonized noncompetitively (18, 20, 21). Despite having a high-resolution structure of the inhibited complex, the mechanism of FliW-mediated CsrA inhibition remains unclear. Here, we argue against electrostatic and steric conflict, and we describe mutants of FliW defective for antagonism of CsrA that fall into two spatial and functional classes. Combined with previous results, the alleles presented here allow us to expand on a model of FliW interaction that leads to the allosteric inhibition of CsrA.

The foundational observations governing the model of FliW antagonism are as follows. First, FliW and CsrA directly interact, and a highly conserved CsrA residue N55 is critical; when N55 is mutated, both antagonism and interaction are simultaneously abolished (20, 21). Second, two molecules of FliW may bind to one CsrA dimer, but only one molecule of FliW is needed to inhibit interaction with RNA (18, 21). Third, the structure of the FliW-CsrA complex has been solved, and although it represents the antagonized state, there is little indication of perturbation to, or occlusion of, the RNA-binding pocket (21). Importantly, interaction is necessary but not sufficient for antagonism, as a mutation of a particular residue within the CsrA core (I14) to a bulkier methionine abolished antagonism while retaining the ability to bind FliW (20). How alteration of CsrA residue I14 abolishes antagonism is unclear but seems to suggest that the core conformation of CsrA is important. Thus, FliW binds to a surface of CsrA that is remote from the active site, but antagonism requires an additional step.

To add to these observations, we took an unbiased approach and identified residues of FliW that are required for inhibition. One group of required FliW residues included several proximal to the regulatory CsrA residue N55 (Fig. 1C). Similar to mutation of CsrA residue N55 itself, mutation of the proximal FliW residues abolished both antagonism and interaction between the two proteins, as observed in both in vivo and in vitro assays (Fig. 3 and 4A). FliW residue Q124, in particular, appears to make direct contact with CsrA residue N55 in the FliW-CsrA complex, and we note that both residues appear to be invariant among different homologs in genomes where both proteins are encoded (Fig. 5) (21). Moreover, mutation of these residues was allele-independent and substitutions that could potentially restore interaction between these two residues were not tolerated (Fig. 2C; Fig. S3G). We suspect that the intolerance is likely due to the number of conserved FliW residues required within the CsrA N55 proximal cluster, each of which seems to contribute to contact and may therefore exclude nonnative combinations of position 124 of FliW with position 55 of CsrA (Fig. 1C and 5B). Whatever the case, we conclude that the FliW residues that make up the CsrA N55 proximal cluster are required for antagonism because they mediate the initial interaction between the two proteins.

Another group of required FliW residues were proximal to the CsrA core region (Fig. 1D). Mutation of the core-proximal residues abolished antagonism but behaved more like the CsrA^I14M allele in that they maintained the ability to bind FliW (20). One residue in particular, FliW K123, appeared to form a salt bridge connection with CsrA residue D41, and mutation of either residue to the opposite charge maintained complex formation between the two proteins in vivo when trapped by chemical cross-linking (Fig. 3 and 5A). We note that each charge-reversed allele decreased complex stability in vitro, but the reduced stability could have been due to the introduction of nonnative electrostatic repulsion (Fig. 4). Charge reversal on both proteins simultaneously, however, restored FliW-mediated inhibition, suggesting that the precise residues were less important than the restoration of the quaternary structure interaction at that position (Fig. 2F and 4B). Consistent with allele flexibility, we note that the residues at FliW position 123 and CsrA position 41 vary in genomes that express both proteins and, indeed, the salt bridge seems to be particular to the Bacillales (Fig. 5B). The interaction with the CsrA core domain seems to be what is important and this could be accomplished by different residues at perhaps slightly different locations in other organisms.

The combined genetic data indicate that the required residues of B. subtilis FliW connect two distinct regions of CsrA as a linear chain (Fig. 5A). We suggest that initial binding is mediated by direct interaction between CsrA residue N55 and FliW residue Q124. The adjacent residue of FliW K123 then forms a salt bridge with CsrA D41, resulting in allosteric inhibition of RNA-binding activity. The mechanism for allosteric regulation during the second step of FliW antagonism is still unclear. The allosteric salt bridge could change the FliW structure, but a change in FliW conformation upon binding is unlikely, as the structure of FliW on its own is the same as when it is bound to CsrA (21). Instead, we favor a model in which CsrA changes conformation because the linear series of amino acids that mediate the inhibition are in line with the CsrA core residue I14 (Fig. 5A). We note that the CsrA^I14M allele phenocopies disruption of the salt bridge and we speculate that changing to a bulkier methionine prevents deformation of, and allosteric conformational change through, the core domain following FliW binding (20). In toto, our results provide further evidence that FliW antagonism occurs in two steps, with the second step playing an allosteric role.

MATERIALS AND METHODS

Strains and growth conditions.

B. subtilis strains were grown in lysogeny broth (LB) (10 g tryptone, 5 g yeast extract, 5 g NaCl per liter) or on LB plates fortified with 1.5% Bacto agar at 37°C. When appropriate, antibiotics were included at the following concentrations: 10 μg/ml tetracycline (tet), 100 μg/ml spectinomycin (spec), 5 μg/ml chloramphenicol (cm), 5 μg/ml kanamycin (kan), and 1 μg/ml erythromycin plus 25 μg/ml lincomycin (mls). Isopropyl-β-d-thiogalactopyranoside (IPTG; Sigma) was added to the medium at the indicated concentration when appropriate.

For the quantitative swarm expansion assay, strains were grown to mid-log phase (optical density at 600 nm [OD₆₀₀] = 0.3 to 1.0) and concentrated to 10 OD₆₀₀ in phosphate-buffered saline (PBS) at pH 7.4 (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 2 mM KH₂PO₄) containing 0.5% India ink (Higgins). LB plates containing 0.7% Bacto agar with or without various concentrations of IPTG were dried for 10 min in a laminar flow hood, centrally inoculated with 10 μl of the cell suspension, dried for another 10 min, and incubated at 37˚C in a humid chamber. Swarm radius was measured along the same axis every 30 min.

Strain construction.

All constructs were either introduced into a 3610-derived natural competent strain DK1042 (27) or first introduced by natural competence into a domesticated strain PY79 or a 3610-derived competent strain cured of the pBS32 plasmid DS2569 (27) and then transferred to the 3610 background using SPP1-mediated generalized phage transduction (28). Briefly, SPP1-mediated transduction was performed by generating a lysate on B. subtilis grown in TY medium (1% tryptone, 0.5% yeast extract, 0.5% NaCl, 10 mM MgSO₄, and 1 mM MnSO₄). Recipient strains were grown to stationary phase in TY, 1 ml was diluted into 9 ml TY, and 25 μl (for streptomycin) lysates was added, followed by incubation at room temperature for 30 min and then selection on the respective antibiotic at 37°C overnight. For transductions in which spectinomycin resistance was selected for, 10 mM sodium citrate was added to the selection plate. All strains used in this study are listed in Table 1. All primers used in this study are listed in Table S1 in the supplemental material. All plasmids used in this study are listed in Table S2.

TABLE 1.

Strains

Strain	Genotype
3610	Wild type
DK1042	comIQ12L (27)
DK1483	comIQ12L ΔfliW-csrA-hag (20)
DK2371	ΔfliW amyE::Physpank-fliW spec (17)
DK2665	comIQ12L ΔfliW-csrA (18)
DK3237	comIQ12L ΔfliW
DK4205	comIQ12L csrAN55D (18)
DK6425	comIQ12L ΔfliW amyE::Physpank-fliWN108Y spec
DK6426	comIQ12L ΔfliW amyE::Physpank-fliWF28S,S53P spec
DK6428	comIQ12L ΔfliW amyE::Physpank-fliWN108T,M120T spec
DK7254	amyE::Physpank-fliWE71A,D73A,D75A,E76A,E80A spec
DK7342	comIQ12L ΔfliW amyE::Physpank-fliWE71A,D73A,D75A,E76A,E80A spec
DK7386	comIQ12L ΔfliW amyE::Physpank-fliWF58S,Q124H spec
DK7387	comIQ12L ΔfliW amyE::Physpank-fliWL20P,A122V spec
DK7389	comIQ12L ΔfliW amyE::Physpank-fliWQ124R spec
DK7567	comIQ12L ΔfliW-csrA amyE::Physpank-fliW-csrA spec
DK7714	amyE::Physpank-fliWK123E-csrAD41K spec
DK7717	comIQ12L ΔfliW-csrA amyE::Physpank-fliWK123E-csrAD41K spec
DK7741	comIQ12L ΔfliW amyE::Physpank-fliWA111P,K123E spec
DK7742	comIQ12L ΔfliW amyE::Physpank-fliWI25T,P26S spec
DK7747	amyE::Physpank-fliW-csrAD41K spec
DK7761	comIQ12L ΔfliW-csrA amyE::Physpank-fliW-csrAD41K spec
DK7762	comIQ12L ΔfliW-csrA amyE::Physpank-fliWK123E-csrA spec
DK7974	comIQ12L ΔfliW-csrA-hag amyE::Physpank-fliW-csrA spec
DK7997	amyE::Physpank-fliW-csrAN55D spec
DK8085	amyE::Physpank-fliWQ124R-csrA spec
DK8090	amyE::Physpank-fliWN108Y-csrA spec
DK8108	comIQ12L ΔfliW-csrAhag amyE::Physpank-fliWQ124R-csrA spec
DK8113	comIQ12L ΔfliW-csrAhag amyE::Physpank-fliWN108Y-csrA spec
DK8206	amyE::Physpank-fliWL20P-csrA spec
DK8207	amyE::Physpank-fliWA111P-csrA spec
DK8208	amyE::Physpank-fliWQ124H-csrA spec
DK8209	amyE::Physpank-fliWI25T-csrA spec
DK8210	amyE::Physpank-fliWP26S-csrA spec
DK8211	amyE::Physpank-fliWA122V-csrA spec
DK8212	amyE::Physpank-fliWF58S-csrA spec
DK8213	amyE::Physpank-fliWS53P-csrA spec
DK8214	amyE::Physpank-fliWF28S-csrA spec
DK8215	amyE::Physpank-fliWN108T-csrA spec
DK8216	amyE::Physpank-fliWM120T-csrA spec
DK8236	comIQ12L ΔfliW-csrA amyE::Physpank-fliWL20P-csrA spec
DK8237	comIQ12L ΔfliW-csrA amyE::Physpank-fliWA111P-csrA spec
DK8238	comIQ12L ΔfliW-csrA amyE::Physpank-fliWQ124H-csrA spec
DK8239	comIQ12L ΔfliW-csrA amyE::Physpank-fliWI25T-csrA spec
DK8240	comIQ12L ΔfliW-csrA amyE::Physpank-fliWP26S-csrA spec
DK8241	comIQ12L ΔfliW-csrA amyE::Physpank-fliWA122V-csrA spec
DK8242	comIQ12L ΔfliW-csrA amyE::Physpank-fliWF58S-csrA spec
DK8243	comIQ12L ΔfliW-csrA amyE::Physpank-fliWS53P-csrA spec
DK8244	comIQ12L ΔfliW-csrA amyE::Physpank-fliWF28S-csrA spec
DK8245	comIQ12L ΔfliW-csrA amyE::Physpank-fliWN108T-csrA spec
DK8246	comIQ12L ΔfliW-csrA amyE::Physpank-fliWM120T-csrA spec
DK8282	comIQ12L ΔfliW-csrA-hag amyE::Physpank-fliWP26S-csrA spec
DK8283	comIQ12L ΔfliW-csrA-hag amyE::Physpank-fliWA122V-csrA spec
DK8284	comIQ12L ΔfliW-csrA-hag amyE::Physpank-fliWF58S-csrA spec
DK8286	comIQ12L ΔfliW-csrA-hag amyE::Physpank-fliWK123E-csrA spec
DK8303	comIQ12L ΔfliW-csrA-hag amyE::Physpank-fliW-csrAN55D spec
DK8304	comIQ12L ΔfliW-csrA-hag amyE::Physpank-fliWN108T-csrA spec
DK8324	comIQ12L ΔfliW-csrA amyE::Physpank-fliW-csrAN55D spec
DK8325	comIQ12L ΔfliW-csrA amyE::Physpank-fliWQ124R-csrA spec
DK8326	comIQ12L ΔfliW-csrA amyE::Physpank-fliWN108Y-csrA spec
DK8330	comIQ12L ΔfliW-csrAhag amyE::Physpank-fliWA111P-csrA spec
DK8335	amyE::Physpank-fliWQ124R-csrAN55D spec
DK8336	amyE::Physpank-fliWQ124K-csrA spec
DK8337	amyE::Physpank-fliWQ124K-csrAN55D spec
DK8345	comIQ12L ΔfliW-csrA amyE::Physpank-fliWQ124R-csrAN55D spec
DK8346	comIQ12L ΔfliW-csrA amyE::Physpank-fliWQ124K-csrA spec
DK8347	comIQ12L ΔfliW-csrA amyE::Physpank-fliWQ124K-csrAN55D spec
DK8373	amyE::Physpank-fliWQ124H-csrAN55D spec
DK8378	comIQ12L ΔfliW-csrAhag amyE::Physpank-fliWK123E-csrAD41K spec
DK8379	comIQ12L ΔfliW-csrAhag amyE::Physpank-fliW-csrAD41K spec
DK8380	comIQ12L ΔfliW-csrA amyE::Physpank-fliWQ124H-csrAN55D spec
DK8638	amyE::Physpank-fliWΔ75–76 spec
DK8641	amyE::Physpank-fliWΔ71,73 spec
DK8662	comIQ12L ΔfliW amyE::Physpank-fliWΔ75–76 spec
DK8671	comIQ12L ΔfliW amyE::Physpank-fliWΔ71,73 spec
DK8692	amyE::Physpank-fliWΔ128–129 spec
DK8693	amyE::Physpank-fliWΔ130–131 spec
DK8717	comIQ12L ΔfliW amyE::Physpank-fliWΔ128–129 spec
DK8718	comIQ12L ΔfliW amyE::Physpank-fliWΔ130–131 spec
DS2569	ΔpBS32 (27)
PY79	sfp0 swrAFS

In-frame deletions.

To generate the ΔfliW in-frame markerless deletion construct in the DK1042 background, plasmid pJP87 (17) was introduced and plasmid integration was selected for by mls resistance at 37°C. Plasmid pJP87 encodes a temperature-sensitive origin that is active at room temperature but not at 37°C. To evict the plasmid, the strain was subsequently incubated at a room temperature overnight without antibiotic selection. Cells were then serially diluted and plated on LB agar at 37°C. Individual colonies were patched on LB plates and LB plates containing mls to identify mls-sensitive colonies that had evicted the plasmid. Deletion of fliW was verified by PCR using primers 1541/1544.

Generation of the fliW mutant pool.

To generate a pool of fliW mutants, primer pair 861/862 was used to amplify the fliW reading frame, using DK2371 chromosomal DNA as a template and Expand polymerase with Expand buffer 2 (Roche). The mutant PCR library was first transformed into PY79 by natural competence, and phage lysates were generated from the resulting transformants. The phage lysates were used to transduce strains of B. subtilis for screening of loss-of-function alleles.

P_hyspank-fliW^neg5A construct.

Site-directed mutations were introduced into the P_hyspank-fliW spec construct in a stepwise fashion using DK2371 genomic DNA (gDNA), amplified using primer pairs 861/6805 and 862/6806 for fliW^E71A,D73A, 861/5871 and 862/5872 for fliW^D75A,E76A, and 861/6707 and 862/6708 for fliW^E80A. The two fragments were assembled together by isothermal “Gibson” assembly (29) and introduced into laboratory strain PY79 by natural competence. Mutations were verified prior to introduction of the next set of mutations by amplifying the appropriate gDNA with primer pair 6275/689 and sequenced using primers 1541 and 1542. Once all mutations were confirmed, the construct was introduced into the appropriate strain backgrounds by SPP1-mediated transductions (28).

P_hyspank-fliW truncation constructs.

Site-directed mutations were introduced into the P_hyspank-fliW spec construct in a stepwise fashion using DK2371 genomic DNA, amplified using primer pairs 861/7422 and 862/7423 for fliW^Δ71,73, 861/7418 and 862/7419 for fliW^Δ75–76, 861/7441 and 862/7442 for fliW^Δ128–129, and 861/7443 and 862/7444 for fliW^Δ_130–131. The two fragments were assembled together by isothermal “Gibson” assembly (29) and introduced into laboratory strain PY79 by natural competence. Mutations were verified by amplifying the appropriate gDNA with primer pair 6275/689 and sequenced using primers 1541 and 1542. Correct constructs were introduced into the appropriate strain backgrounds by SPP1-mediated transductions (28).

P_hyspank-fliW-csrA mutation constructs.

Site-directed mutations to the P_hyspank-fliW-csrA spec construct were generated using DK7567 genomic DNA, amplified using primer pairs 861/7098 and 862/7099 for fliW^L20P, 861/7102 and 862/7103 for fliW^A111P, 861/7104 and 862/7105 for fliW^K123E, 861/7110 and 862/7111 for fliW^Q124H, 861/7112 and 862/7113 for fliW^I25T, 861/7114 and 862/7115 for fliW^P26S, 861/7116 and 862/7117 for fliW^A122V, 861/7118 and 862/7119 for csrA^D41K, 861/7163 and 862/7164 for fliW^F58S, 861/7165 and 862/7166 for fliW^S53P, 861/7256 and 862/7257 for fliW^Q124R, 861/7258 and 862/7259 for fliW^F28S, 861/7260 and 862/7261 for fliW^N108T, 861/7262 and 862/7263 for fliW^M120T, 861/7266 and 862/7267 for fliW^N108Y, and 861/7311 and 862/7312 for fliW^Q124K. The two fragments were assembled together by isothermal “Gibson” assembly (29), introduced into laboratory strain PY79 by natural competence, and further introduced into the appropriate strain backgrounds by SPP1-mediated transductions (28). Mutations were verified by amplifying the appropriate gDNA with primer pair 6275/689, and sequenced using primers 1541 and 1544. Construct P_hyspank-fliW^K123E-csrA^D41K was generated using DK7762 chromosomal DNA and amplified using primer pairs 861/7118 and 862/7119; construct P_hyspank-fliW^Q124R-csrA^N55D was generated using DK8324 chromosomal DNA and amplified using primer pairs 861/7256 and 862/7257; construct P_hyspank-fliW^Q124H-csrA^N55D was generated using DK8324 chromosomal DNA and amplified using primer pairs 861/7110 and 862/7111; and construct P_hyspank-fliW^Q124K-csrA^N55D was generated using DK8324 chromosomal DNA and amplified using primer pairs 861/7311 and 862/7312. The previous four constructs were assembled and introduced into the appropriate strain backgrounds as above.

P_hyspank-fliW-csrA^N55D construct.

To generate P_hyspank-fliW-csrA^N55D, a fragment containing fliW-csrA^N55D was amplified by using DK4025 as a template and primer pair 1541/1544 and was digested with NheI/SphI. The fragment was ligated into the NheI and SphI sites of pDR11 containing an ampicillin and spectinomycin resistance cassette, to generate pRO102. pRO102 was introduced into the laboratory strain PY79 by natural transformation, then further introduced into the appropriate strain background using SPP1-mediated transduction (28).

GST-CsrA^D41K expression vector.

To generate a translational fusion of CsrA^D41K to the GST tag, a fragment containing csrA^D41K was amplified by using DK7761 as a template and primer pair 2140/2141 and was digested with BamHI/EcoRI. The fragment was ligated into the BamHI and EcoRI sites of pGEX-2TK containing an ampicillin resistance cassette, to create pRO114.

His-SUMO-FliW^mutant expression vectors.

To generate a translational fusion of FliW^N108Y, FliW^Q124R, and FliW^K123E to the His-SUMO tag, a fragment containing the aforementioned mutations was amplified by using DK8326 (fliW^N108Y), DK8325 (fliW^Q124R), and DK7762 (fliW^K123E) as templates with primer pair 2230/7194 and was digested with SapI/XhoI. The fragment was ligated into the SapI and XhoI sites of pTB146 containing an ampicillin resistance cassette, to create pRO115, pRO106, and pRO107, respectively.

GST-CsrA^WT and GST-CsrA^D41K protein purification.

The GST-CsrA^WT and GST-CsrA^D41K protein expression vectors pSM6 and pRO114, respectively, were transformed into Rosetta gami Escherichia coli, grown to an optical density at 600 nm (OD₆₀₀) of ∼0.7 in 500 ml of Luria-Bertani broth, induced with 1 mM IPTG, and grown for 3 h at 37°C. Cells were pelleted and resuspended in lysis buffer (25 mM Tris-HCl [pH 8.0], 1 mM dithiothreitol [DTT], 1 mM EDTA, 150 mM NaCl, and 0.5 mM phenylmethylsulfonyl fluoride [PMSF]), and frozen at –80°C overnight. The frozen cell pellet was thawed and lysed by sonication. Lysed cells were ultracentrifuged at 14,000 rpm for 30 min at 4°C. Cleared supernatant was combined with glutathione-Sepharose (GE Healthcare) and incubated for 3 h at 4°C. The bead/lysate mixture was poured onto a 1-cm separation column (Bio-Rad), the resin was allowed to pack, and the mixture was washed with wash buffer (25 mM Tris-HCl [pH 8.0], 1 mM DTT, 1 mM EDTA, 250 mM NaCl, 10% glycerol, and 0.5 mM PMSF). GST-CsrA bound to the resin was then eluted using GST elution buffer (25 mM Tris-HCl [pH 8.5], 20 mM glutathione, 1 mM DTT, 1 mM EDTA, 250 mM NaCl, 10% glycerol, and 0.5 mM PMSF). Eluted fractions were separated by SDS-PAGE and Coomassie stained to verify purification of the GST-CsrA fusion, and the appropriate fractions were pooled and concentrated to ∼2 ml. Final purification of the GST-CsrA protein was conducted via size exclusion chromatography on a Superdex 75 16/60 (GE Healthcare) column using CsrA gel filtration buffer (20 mM Tris-HCl [pH 8.0], 200 mM NaCl, 10% glycerol, and 1 mM EDTA [pH 8.0]) and fractions were concentrated and stored at −80°C. Concentrations of GST-CsrA and GST-CsrA^D41K were determined by Bradford assay (Bio-Rad).

His₆-SUMO-FliW^WT and -FliW^mutant protein purification.

The His₆-SUMO-FliW^WT, FliW^N108Y, FliW^K123E, and FliW^Q124R protein expression vectors pSM12, pRO115, pRO107, and pRO106, respectively, were transformed into Rosetta gami E. coli. Cells were grown to ∼0.7 OD₆₀₀ in 500 ml of terrific broth, induced with 1 mM IPTG, and grown overnight at 16°C. Cells were pelleted and resuspended in CsrA lysis buffer (100 mM Tris-HCl [pH 8.0] and 400 mM NaCl), treated with lysozyme, and lysed by sonication. Lysed cells were centrifuged at 14,000 × g for 30 min. Cleared supernatant was combined with Ni-nitrilotriacetic acid (Ni-NTA) resin (Novagen) and incubated for 1 h at 4°C. The bead/lysate mixture was poured onto a 1-cm separation column (Bio-Rad), the resin was allowed to pack, and was washed with CsrA wash buffer (50 mM Tris-HCl [pH 8.0], 200 mM NaCl, and 10% glycerol). His₆-SUMO-FliW^{WT/N108Y/K123E/Q124R} (referring to all four proteins) bound to the resin was then eluted using a stepwise elution of CsrA wash buffer with 5, 15, and 500 mM imidazole. Eluted proteins were separated by SDS-PAGE and Coomassie stained to verify purification of the respective His₆-SUMO-FliW^{WT/N108Y/K123E/Q124R} proteins and appropriate fractions were pooled and concentrated to ∼2 ml. His₆-SUMO-FliW^{WT/N108Y/K123E/Q124R} proteins were further cleaned by size exclusion chromatography on a Superdex 75 16/60 (GE Healthcare) column using CsrA gel filtration buffer (20 mM Tris-HCl [pH 8.0], 200 mM NaCl, and 10% glycerol) and fractions were separated by SDS-PAGE and Coomassie stained to verify purified His₆-SUMO-FliW^{WT/N108Y/K123E/Q124R} proteins. Purified His₆-SUMO-FliW^{WT/N108Y/K123E/Q124R} proteins were combined with ubiquitin ligase (protease) and cleavage buffer and incubated overnight at 4°C to cleave the SUMO tag (30). The cleavage reaction was combined with Ni-NTA beads, incubated for 2 h at 4°C, and centrifuged to pellet the resin. Removal of the SUMO tag was verified by SDS-PAGE and Coomassie staining. Supernatants for purified FliW^{WT/N108Y/K123E/Q124R} proteins were further cleaned by size exclusion chromatography on a Superdex 75 16/60 (GE Healthcare) column using CsrA gel filtration buffer (20 mM Tris-HCl [pH 8.0], 200 mM NaCl, and 10% glycerol) and fractions were separated by SDS-PAGE and Coomassie stained to verify purified FliW^{WT/N108Y/K123E/Q124R} proteins. FliW^{WT/N108Y/K123E/Q124R} proteins were stored at −80°C. Concentration of FliW^{WT/N108Y/K123E/Q124R} proteins were determined by Bradford assay (Bio-Rad).

Western blotting.

B. subtilis strains were grown in LB to OD₆₀₀ of ∼1.0; then 1 ml of sample was harvested by centrifugation, resuspended to OD₆₀₀ of 10 in lysis buffer (20 mM Tris [pH 7.0], 10 mM EDTA, 1 mg/ml lysozyme, 10 μg/ml DNase I, 100 μg/ml RNase I, and 1 mM PMSF) and incubated for 60 min at 37°C. Lysate (10 μl) was mixed with 2 μl of 6× SDS loading dye. Samples were separated by 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The proteins were electroblotted onto nitrocellulose and developed with either anti-FliW (1:20,000) (20) or anti-SigA (1:80,000; generous gift of Masaya Fujita, University of Houston) and a 1:10,000 dilution of secondary antibody (horseradish peroxidase (HRP)-conjugated goat anti-rabbit immunoglobulin G). The immunoblots were developed using the Immun-Star HRP developer kit (Bio-Rad).

In vivo cross-linking.

B. subtilis was grown to mid-log phase at 37°C in LB broth supplemented with 0.1 mM IPTG. A 10-ml aliquot of culture was harvested by centrifugation, and resuspended in 10 ml of phosphate-buffered saline (PBS) at pH 7.4 (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 2 mM KH₂PO₄). Formaldehyde (0.3% final concentration) was added to samples and rocked at room temperature for 1 h. Next, 0.75 ml of 2 M glycine was added to samples and rocked at room temperature for 10 min to quench the cross-linker. Samples were centrifuged and resuspended to OD₆₀₀ of 50 in lysis buffer (20 mM Tris-HCl [pH 7.0], 10 mM EDTA, 1 mg ml⁻¹ lysozyme, 10 μg ml⁻¹ DNase I, and 1 mM PMSF) and incubated at 37°C for 1 h. Lysed samples were diluted 1:10 for samples probed with α−FliW antibody. Loading dye (6× SDS) was added and incubated for 15 min at room temperature. Samples were assayed following the Western blotting protocol described above. Blots were probed with either anti-CsrA (1:10,000) or anti-FliW (1:20,000) antibodies (20).

GST-CsrA^WT/D41K and FliW^WT/mutant interaction pulldown assay.

Glutathione-Sepharose beads were washed with T(0.1) buffer (25 mM Tris-HCl [pH 8.0], 20% glycerol, 100 mM NaCl, 1 mM DTT, 1× protease inhibitor cocktail [Roche], 1 mg/ml bovine serum albumin [BSA; Sigma]). Washed beads (60 μl) were mixed with 60 μl of 0.25 μM GST-CsrA^WT/D41K protein and rotated on a Labquake at 4°C for 2 h. Next, the beads bound to GST-CsrA^WT/D41K protein were centrifuged at 1,000 rpm for 2 min and the pellet was washed twice with T(1.0) buffer (25 mM Tris-HCl [pH 8.0], 20% glycerol, 1 M NaCl, 1 mM DTT, 1× protease inhibitor cocktail [Roche], and 1 mg/ml BSA [Sigma]) and again twice with T(0.1) buffer. Then, 60 μl of increasing concentrations of FliW^{WT/N108Y/K123E/Q124R} (0.125, 0.167, 0.25, 0.333, and 0.5 μM) proteins was added to the washed beads bound to GST-CsrA^WT/D41K and rotated on a Labquake at 4°C for 2 h. The samples were centrifuged at 1,000 rpm for 2 min and 40 μl of the supernatant was saved. The pellet was washed 4 times with T(0.1) buffer. The supernatant and pellet fractions were subjected to SDS-PAGE analysis and Western blot analysis following the protocols described above. Blots were probed with either anti-CsrA (1:40,000) or anti-FliW (1:60,000) antibodies (20).

Circular dichroism.

A Jasco J-715 circular dichroism machine was used to measure the different FliW proteins at 25°C. Each protein sample was scanned 5 times using a 1-mm path length quartz cuvette. Each FliW^WT, FliW^N108Y, and FliW^Q124R protein was extensively dialyzed into 10 mM sodium phosphate buffer (pH 7.5).

Structural analysis.

The FliW-CsrA 3-dimensional structure (PDB 5DMB) (21) was analyzed using Chimera version 1.13.1.