The bacterial strains that were used in this study are listed in Table 1. E. coli strains were routinely grown in Luria-Bertani (LB) broth or agar at 37°C in an ambient atmosphere. Media were supplemented with antibiotics at the following concentrations when appropriate: ampicillin (Amp), 50 μg/ml; chloramphenicol (Cm), 30 μg/ml. C. jejuni was grown in Mueller-Hinton (MH) broth or agar at 42°C in a tri-gas incubator with an atmosphere of 10% O₂, 10% CO₂, and 80% N₂.

Sequential deletions of the C-terminal region of CsrA were constructed by using inverse PCR with the primers listed in Supplementary Table S1. PCR primers csrA-5.R, csrA-10.R, csrA-15.R, csrA-20.R, and csrA-25.R were used in conjunction with primer csrA-del.F on pET-20b-CsrA-His₆ template (Fields et al., 2016), to generate derivatives lacking 5, 10, 15, 20, and 25 C-terminal amino acids of CsrA (designated CsrAΔ5, Δ10, Δ15, Δ20, and Δ25, respectively). In each case a stop codon was introduced to avoid translation of the His₆ tag. Deletion mutants of CsrA were overexpressed in E. coli strain BL21(DE3)pLysS. Cells were grown at 37°C in LB broth containing 50 μg/ml Amp and 30 μg/ml Cm. When cells reached mid-log phase (OD₆₀₀ = 0.5), the expression of the proteins was induced with 0.5 mM IPTG for 3 h.

CsrA-WT-His₆, CsrA-V51A-His₆, and CsrA-N55A-His₆ were purified as described previously (El Abbar et al., 2019). FliW-His₆ protein was produced in E. coli strain BL21(DE3)pLysS. Cells transformed with pET-20b-FliW-His₆ (Li et al., 2018) were grown in Luria-Bertani broth (LB) containing ampicillin and chloramphenicol at 37°C. When the culture reached OD₆₀₀ = 0.6, the expression of the protein was induced with 0.5 mM IPTG for 4 h. Cells were collected by centrifugation and the cell pellet was resuspended in buffer A: 50 mM Tris-HCl, pH 7.4, 300 mM NaCl, 5 mM β-mercaptoethanol, lysed by passing twice through a French Press (Thermo), and centrifuged at 10,000 × g for 10 min. The pellet was subsequently dissolved in buffer B: 50 mM Tris-HCl, pH 7.4, 8 M urea, 5 mM β-mercaptoethanol, and the sample was centrifuged at 15,000 × g to remove insoluble material. Supernatant was mixed with Ni-NTA chromatography resin (Ni-NTA Agarose, Qiagen). After protein binding for 1 h in 4°C, the resin was washed 3 times with 10 resin volumes of buffer B and protein was eluted with buffer C: 50 mM Tris-HCl, pH 7.4, 8 M urea, 5 mM β-mercaptoethanol, 300 mM imidazole. The protein was then diluted to 1 mg/ml with buffer C and refolded by dialysis against buffer E: 50 mM Tris-HCl, pH 7.4, 300 mM NaCl, 0.4 M arginine, 1 mM EDTA, 1 mM glutathione, 0.1 mM glutathione disulfide, and subsequently against buffer F: 20 mM Tris-HCl, pH 7.4, 300 mM NaCl, 5 mM β-mercaptoethanol. Final protein purity was achieved by gel filtration on HiPrep Sephacryl S-100 HR column (GE Healthcare) in buffer F.

Purified FliW-His₆ protein in 50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 0.01% Tween-20 was incubated with 10 μl of Dynabeads^TM His-Tag Isolation & Pulldown (Invitrogen) for 10 min, and the beads were subsequently washed four times with the same buffer. E. coli BL21(DE3)pLysS cells expressing wild type CsrA or CsrA proteins containing C-terminal deletions were resuspended in 20 mM sodium phosphate, pH 7.5, 150 mM NaCl, 0.01% Tween-20, and lysed by sonication. Insoluble material was removed by centrifugation at 12,000 × g and lysates were then incubated with FliW-coated beads for 20 min at room temperature. Beads were subsequently washed four times with 300 μl of the same buffer. Protein complexes were eluted from magnetic beads with 50 mM sodium phosphate, pH 7.5, 150 mM NaCl, 300 mM imidazole.

For Western blot analysis, samples were separated on a 15% SDS-PAGE gel and then electroblotted onto PVDF membranes (Millipore). Membranes were probed overnight with 1:200 dilution of primary antibodies against CsrA in TBS-T (Tris-buffered saline, 0.05% Tween 20). After washing twice with TBS-T, horseradish peroxidase-conjugated secondary goat anti-rabbit antibodies (Bio-Rad, 1:20,000 in TBS-T) were applied. Blots were washed twice and incubated in SuperSignal West Pico Chemiluminescent substrate (Thermo Fisher Scientific). Specific bands were visualized using G:Box imager (Syngene).

To identify the FliW binding site of CsrA, we used a combination of SDM and the bacterial two-hybrid system described by Karimova et al. (1998) based on reconstitution of adenylate cyclase activity. We first performed SDM to create substitution mutations in which each of the 75 amino acids of CsrA was converted to alanine, with the exception of two native alanine residues (A30 and A36) that were converted to valines. A Q5 SDM kit (NEB) was used for all SDM experiments. For CsrA mutants at amino acid positions 40–60, csrA was amplified using PCR with the primers csrA-KpnI F and csrA-HindIII R (Supplementary Table S1), using template DNA of CsrA point mutations previously cloned into pET20b (El Abbar et al., 2019). The PCR products were digested with KpnI, HindIII, and DpnI and cloned into pT18. To create mutations at CsrA amino acid positions 11–39 and 61–67, we used the appropriate SDM primers (Supplementary Table S1) with template DNA of csrA WT cloned into pT18. For CsrA mutations at amino acid positions 1–10 and 68–75, mutations were designed in the forward PCR primers for 1–10 and in the reverse PCR primers for 68–75. These primers were combined with PCR primers csrA-HindIII.R or csrA-KpnI.F, respectively (Supplementary Table S1), and used to amplify csrA from C. jejuni 81–176 chromosomal DNA. Each PCR product was then cloned into pT18. The fliW gene was amplified from 81–176 chromosomal DNA using primers fliW-PstI.F and fliW-BamHI.R (Supplementary Table S1), and was then cloned into pT25 using PstI and BamHI. We performed bacterial two-hybrid experiments as described (Karimova et al., 1998) to detect the interaction of FliW with WT or mutant CsrA proteins. Briefly, the interaction of FliW with WT or mutant CsrA proteins was assessed by co-expressing CsrA constructs cloned into pT18 with FliW cloned into pT25 in the E. coli host strain Sp850 (Karimova et al., 1998). Co-expression of vectors pT18 and pT25 lacking inserts was used as a negative control. β-galactosidase assay (Miller, 1972; Karimova et al., 1998) was performed to measure the degree of interaction between CsrA and FliW.

Purified FliW-His₆ protein was mixed with CsrA-His₆ wild-type or mutant proteins in SEC buffer (20 mM sodium phosphate, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 2 mM DTT) at an equimolar ratio for a final protein concentration of 1 mg/ml, and kept on ice for 30 min. 100 μl of each sample was loaded on a Superdex 200 10/300 GL column (GE Healthcare Life Sciences), and separation was performed in SEC buffer at a flow rate of 0.4 ml/min.

CsrA (or its ortholog RsmA) proteins from diverse bacterial species were aligned using ClustalOmega at EMBL-EBI (Sievers et al., 2011; Li et al., 2015). C. jejuni (WP_002852854.1), H. pylori (WP_120911520.1), W. succinogenes (WP_011138784.1), N. profundicola (WP_127679601.1), S. denitrificans (WP_011372074.1), T. pallidum (WP_010882102.1), B. subtilis (WP_014478124.1), C. difficile (WP_022619991.1), C. botulinum (WP_003367280.1), Geobacillus thermodenitrificans (WP_011 888204.1).

To model interaction of C. jejuni CsrA with FliW, we used the only available crystal structure of CsrA containing a C-terminal extension from G. thermodenitrificans [PDB: 5DMB (Altegoer et al., 2016)]. Molecular graphics and analyses were performed with UCSF Chimera (Pettersen et al., 2004).

Alignment of CsrA proteins from C. jejuni and E. coli shows both conserved and divergent features (Figure 1A; Fields and Thompson, 2012). Both CsrA proteins are composed of five β-strands. CsrA from C. jejuni is longer (75 amino acids), and its activity is modulated by FliW (Dugar et al., 2016; Radomska et al., 2016; Li et al., 2018). In contrast, E. coli CsrA is shorter (61 amino acids) (Figure 1A) and its activity is regulated by binding to sRNAs (Jin et al., 2002; Thomson et al., 2006; Babitzke and Romeo, 2007; Fakhry et al., 2017; Janssen et al., 2018; Potts et al., 2019). As the C-terminus of C. jejuni CsrA is longer than the analogous region of CsrA proteins of E. coli and other bacteria, whose activities are regulated by sRNAs rather than by FliW (Fields and Thompson, 2012), and adjacent to the most C-terminal β-strand (β₅) that is involved in RNA binding in both E. coli and C. jejuni (Figure 1A; Mercante et al., 2006; El Abbar et al., 2019), we hypothesized that FliW would bind to this region. We therefore used inverse PCR to construct a series of deletions in the C. jejuni 81–176 csrA gene cloned into an expression vector (Fields et al., 2016), resulting in a deletion of 5, 10, 15, 20, and 25 amino acids of the C-terminus (lacking a C-terminal His₆ tag, Figure 1A). WT and C-terminally deleted proteins were expressed in E. coli, and cell lysates containing these proteins were used in pulldown assays with FliW-His₆ immobilized on magnetic beads (Figure 1B). Bound CsrA was detected using anti-CsrA antibodies. CsrA-WT bound to FliW (Li et al., 2018), as did mutant CsrA proteins lacking the C-terminal 5, 10, and 15 amino acids (Figures 1A,B). However, the deletion of 20 amino acids from the C-terminus of CsrA almost completely abrogated binding, suggesting that at least part of the FliW binding site was located in the region around amino acid 56 of C. jejuni CsrA. A CsrA protein containing a 25 amino acid deletion was highly unstable and could not be used in these assays (data not shown).

To further refine the FliW binding site, we used a combination of SDM and a bacterial two-hybrid system. In this system, potentially interacting proteins are constructed as fusion proteins with two domains (T18 and T25) of the B. pertussis adenylate cyclase toxin (Karimova et al., 1998). Positive interaction reconstitutes cAMP synthesis and results in β-galactosidase activity in the E. coli reporter strain Sp850 (Karimova et al., 1998). We first cloned the C. jejuni csrA gene into one of the two hybrid vectors (pT18) and fliW in the other (pT25) (Figure 2A). Compared to a control strain containing pT18 and pT25 only, co-expression of pT18-csrA and pT25-fliW generated significant β-galactosidase activity, indicating positive interaction in this system (Figure 2B). We next performed SDM on C. jejuni csrA cloned into pT18, changing each amino acid to alanine (except for two native alanine residues, which were changed to valines). Each site-directed mutant was introduced into E. coli Sp850 containing pT25-fliW. β-galactosidase activities showed that 73 of 75 site-directed mutants had interactions of CsrA and FliW that were not significantly different than WT (Figure 2B). However, CsrA-V51A and CsrA-N55A showed a significant reduction in β-galactosidase activity, demonstrating that these two amino acids are required for binding of FliW to CsrA (Figure 2B). These amino acids lie just C-terminally to amino acids involved in RNA binding by C. jejuni CsrA (Figure 2B; El Abbar et al., 2019).

To confirm that CsrA V51 and N55 were involved in binding FliW, we evaluated the protein-protein interactions of FliW with WT and mutant CsrA, using two in vitro methods. In size-exclusion chromatography experiments, we passed purified FliW-His₆ and CsrA-His₆ (WT, V51A, or N55A) either alone, or mixed in equimolar ratios over a SEC column. FliW-His₆ eluted at a retention volume of 16.9 ml (Figure 3A). Mixture of FliW-His₆ and CsrA-WT-His₆ resulted in a stable heterotetramer of CsrA₂-FliW₂ (Li et al., 2018), with a retention volume of 15.1 ml (Figure 3A). FliW-His₆ and CsrA-V51A-His₆ gave mixture of protein complex and unbound proteins, indicating weaker interaction (Figure 3B). No complex was formed between FliW-His₆ and CsrA-N55A-His₆ in the equimolar mixture of proteins, indicating lack of protein-protein interaction (Figure 3C). These results are consistent with the findings from two-hybrid experiments (Figure 2B). To further determine the affinities of CsrA WT, V51A, and N55A to FliW, we performed binding assays using surface plasmon resonance. Purified FliW-His₆ was immobilized onto the chip surface and different CsrA-His₆ (WT, V51A, or N55A) concentrations were run over the chip surface to obtain binding parameters (Figures 3D–F). Data followed a bivalent analyte binding model, confirming that one molecule of CsrA dimer is able to bind two molecules of FliW. The K_D for the first ligand binding site of CsrA-WT was calculated to be 9 × 10^–9 M, showing strong binding (Figure 3D). As expected from the SEC experiment, the affinity of CsrA-V51A to FliW was reduced 6-fold (K_D = 5.2 × 10^–8 M), while binding of CsrA-N55A was decreased 10-fold (K_D = 9.3 × 10^–8 M) (Figures 3E,F). Taken together, our results from the two-hybrid assay and both in vitro binding assays indicate that amino acids V51 and N55 are crucial for full FliW binding by CsrA.

We next examined CsrA proteins for conservation of these FliW-binding amino acids among diverse bacteria expressing CsrA proteins with extended C-terminal regions and containing FliW. Comparison of these proteins showed that N55 was identical in CsrA proteins from all species examined (Figure 4A). V51 was identical in five of the species examined, while in the other five species this position was occupied by the conservative substitution isoleucine.

A crystal structure of C. jejuni CsrA is not currently available. To model CsrA-FliW interactions, we used the published structure of CsrA from G. thermodenitrificans (Altegoer et al., 2016; Figure 4B). The CsrA proteins from G. thermodenitrificans and C. jejuni are 38% identical/60% similar, and both have extended C-terminal regions compared to E. coli CsrA (Fields and Thompson, 2012). Validity of this model was further emphasized by the fact that CsrA amino acid residues V51 and N55 correspond to isoleucine and asparagine in the CsrA ortholog from G. thermodenitrificans, respectively. According to this structure, a symmetrical homodimer of CsrA binds two molecules of FliW on the opposite sides of the molecule. The amino acids with roles in FliW binding lie within the first α-helix C-terminally adjacent to the RNA-binding site. These are distinct from residues that were shown previously to play a role in mRNA binding by C. jejuni CsrA, which mainly occupy positions within the β-strands in the core of the CsrA homodimer (Figures 1A, 4B; El Abbar et al., 2019). They are however predicted to be located adjacent to the RNA-binding groove at the edge of the CsrA dimer β-sheet, which in this model is adjacent to the aforementioned α-helix (Figure 4B). As amino acids involved in interaction with FliW seem not to be necessary for mRNA binding, we propose that FliW inhibition of CsrA regulatory activity does not rely on direct competition for the CsrA RNA-binding site.