A specificity determinant for phosphorylation in a response regulator prevents in vivo cross-talk and modification by acetyl phosphate

Abstract

Bacterial two-component systems (TCSs) sense stimuli and transduce signals intracellularly through phosphotransfer between cognate histidine kinases (HKs) and response regulators (RRs) to alter gene expression or behavioral responses. Without high phosphotransfer specificity between cognate HKs and RRs, cross-phosphorylation or cross-talk between different TCSs may occur and diminish responses to appropriate stimuli. Some mechanisms to reduce cross-talk involve HKs controlling levels of cognate RR phosphorylation. Conceivably, some RRs may have evolved HK-independent strategies to insulate themselves from cross-talk with acetyl phosphate (AcP) or other small phosphodonor metabolites. Initial steps in flagellar biosynthesis in Campylobacter jejuni stimulate phosphotransfer from the FlgS HK to the FlgR RR to promote σ⁵⁴-dependent flagellar gene expression. We discovered that the FlgR C-terminal domain (CTD), which commonly functions as a DNA-binding domain in the NtrC RR family, is a specificity determinant to limit in vivo cross-talk from AcP. FlgR lacking the CTD (FlgR_ΔCTD) used FlgS or AcP as an in vivo phosphodonor and could be reprogrammed in ΔflgS mutants to respond to cellular nutritional status via AcP levels. Even though exclusive AcP-mediated activation of FlgR_ΔCTD promoted WT flagellar gene expression, proper flagellar biosynthesis was impaired. We propose that the FlgR CTD prevents phosphotransfer from AcP so that FlgR is solely responsive to FlgS to promote proper flagellar gene expression and flagellation. In addition to mechanisms limiting cross-talk between noncognate HKs and RRs, our work suggests that RRs can possess domains that prevent in vivo cross-talk between RRs and the endogenous metabolite AcP to ensure signaling specificity.

Cellular signal transduction systems link extracellular or intracellular stimuli to appropriate output responses. Prokaryotic organisms often are useful models for understanding aspects of signal transduction that may be applicable to higher organisms. Bacteria commonly use two-component systems (TCSs) to mediate responses to specific conditions. A basic TCS consists of a sensor histidine kinase (HK) that autophosphorylates upon detection of a specific signal (1). The phosphorylated histidine of the HK serves as a phosphodonor for autophosphorylation by the cognate response regulator (RR). The phosphorylated RR can then alter gene expression or a behavioral response. TCSs generally do not exist in isolation in a bacterial cell, but among a signaling network of up to as many as 200 different TCSs, depending on the species. Despite structural similarity between many TCSs, phosphotransfer specificity between cognate HK and RR pairs is high (2). In vivo cross-phosphorylation, or cross-talk, between noncognate HK and RRs is usually maintained at a minimum. If cross-talk between two TCSs occurs, correct responses to specific signals may be diminished or inhibited.

Mechanisms exist to insulate a TCS from cross-talk and ensure that intrasystem signal transduction fidelity is preserved. In many TCSs, specific amino acids mediate molecular recognition between cognate HK and RR pairs (3–6). In addition, some HKs are bifunctional with a phosphatase activity to reduce levels of phosphorylated cognate RRs. These bifunctional HKs control activity of the RR by reducing phosphorylation that may occur via cognate HKs, noncognate HKs, or low molecular-weight phosphodonors such as acetyl phosphate (AcP) (1, 7–10). Together, these mechanisms contribute to phosphotransfer specificity in some TCSs.

The FlgSR TCS of Campylobacter jejuni is required for expression of the σ⁵⁴ regulon, which mainly includes flagellar rod and hook genes (11, 12). Initiation of signal transduction through FlgSR is dependent upon components of the flagellar type III secretion system (T3SS) (13, 14). After autophosphorylation of H141 of the cytoplasmic FlgS HK, FlgR autophosphorylates on D51 of the receiver domain, using the phosphohistidine of FlgS as a substrate. Phosphorylation of FlgR is required for σ⁵⁴-dependent gene expression. Because the flagellar T3SS exports rod and hook proteins out of the cytoplasm, FlgSR links T3SS formation to expression of genes encoding substrates secreted by the T3SS to synthesize the organelle (14).

Most NtrC-like RRs possess an essential C-terminal DNA-binding domain (CTD) that interacts with target promoters to activate gene expression. Mutant RRs lacking the CTD fail to stimulate WT levels of σ⁵⁴-dependent gene expression under normal conditions, even in the presence of the cognate HK (15–17). In contrast, FlgR lacking its CTD (FlgR_ΔCTD) activated WT levels of σ⁵⁴-dependent gene expression in the presence of FlgS (13). However, the activities of WT FlgR and FlgR_ΔCTD in ΔflgS mutants differed. Whereas WT FlgR without FlgS did not stimulate σ⁵⁴-dependent gene expression, FlgR_ΔCTD in a ΔflgS mutant activated expression of the σ⁵⁴ regulon. However, the level of gene expression in ΔflgS flgR_ΔCTD was ∼20% of that of C. jejuni with a WT FlgSR TCS. Because FlgR_ΔCTD in the ΔflgS mutant required the phosphorylated D51 residue to activate gene expression (13), FlgR_ΔCTD must have autophosphorylated via a noncognate phosphodonor in this mutant. Thus, these results question the role of the FlgR CTD in DNA binding and suggest that the CTD may serve an alternative function to limit phosphotransfer specificity or cross-talk to FlgR.

We explored the role of the CTD of FlgR in activation of σ⁵⁴-dependent flagellar gene expression. By conducting genetic, biochemical, and physiological studies, we discovered that unlike most NtrC-like RRs, the CTD of FlgR has a DNA-binding activity that is not essential for expression of target genes under physiological conditions. Instead, we show that the CTD is a specificity determinant for phosphorylation that expressly limits the ability of FlgR to autophosphorylate in vivo using the small molecular-weight phosphodonor AcP. Without the CTD, metabolic processes that alter AcP levels influence FlgR activation, but complete reliance of FlgR on AcP for WT levels of gene expression hindered flagellation. As a result of the CTD limiting cross-talk of FlgR with AcP, FlgR activation is coupled to FlgS and a step in flagellar biosynthesis to promote proper gene expression and flagellation for optimal fitness. Whereas previously known mechanisms to eliminate intersystem cross-talk mostly involved activities of HKs, our work identified a domain within a RR that specifically prevents cross-talk with the endogenous central metabolite AcP.

Results

DNA Binding by FlgR Is Not Required for σ⁵⁴-Dependent Gene Expression.

C. jejuni FlgR is a member of the NtrC family of RRs. As such, FlgR contains an N-terminal receiver domain modified by phosphorylation on D51 and a central ATPase domain necessary for oligomerization and interactions with σ⁵⁴ (Fig. 1A). Within the CTD of most NtrC-like RRs is a helix-turn-helix (HTH) motif essential for DNA binding and function as a transcriptional regulator (15, 16). This HTH binds enhancer-like sites upstream of target promoters to promote WT levels of gene expression under normal conditions. Unlike most other NtrC-like RRs, the FlgR CTD lacks a strongly predicted HTH motif. Compared with WT C. jejuni with an intact FlgSR, a flgR_ΔCTD mutant (which produces WT FlgS) promoted similar or slightly higher levels of expression of σ⁵⁴-dependent genes (Table S1) (13). These results suggested either that the FlgR CTD does not bind DNA, or that any DNA-binding activity of the CTD is not essential for FlgR-dependent gene expression. Therefore, we explored the role of the CTD in the function of FlgR as a transcriptional regulator.

Fig. 1.

Analysis of the requirement of FlgR/DNA interactions for σ⁵⁴-dependent flagellar gene expression. (A) Domain organization of E. coli NtrC, C. jejuni WT FlgR, and C. jejuni FlgR_ΔCTD. The phosphorylated aspartic acids (D or D51; P denotes phosphoryl group) in receiver domains are indicated. (B and C) EMSAs to analyze P_flgDE2 DNA binding by FlgR proteins. (B) WT FlgR and FlgR_ΔCTD were used (from left to right) at 0, 0.1, 0.25, 0.5, 0.75, and 1 μM. (C) Increasing ratios of unlabeled P_flgDE2 DNA (1:1, 5:1, or 10:1) or P_gyrA DNA (1:1 or 10:1) to labeled P_flgDE2 DNA were incubated with 1 μM of FlgR proteins. (D) Expression of flgDE2::astA with 5′ promoter truncations in WT C. jejuni or flgR_ΔCTD, ΔrpoN (Δσ⁵⁴), or ΔflgR mutants. flgDE2::astA expression levels are reported as arylsulfatase units ± SD. All transcriptional fusions were on plasmids, except for the fusion at the native chromosomal flgDE2 locus (chr). The 5′ ends of fusions on plasmids are indicated relative to the transcriptional start site (arrow). Black boxes indicate the −24 and −12 σ⁵⁴ binding sites essential for σ⁵⁴-dependent gene expression.

We analyzed FlgR interactions with the σ⁵⁴-dependent flgDE2 promoter (P_flgDE2) by EMSAs. Purified WT FlgR bound promoter DNA encompassing –302 to +79 bases relative to the flgDE2 transcriptional start site in a dose-dependent manner (Fig. 1B) (18). Binding by WT FlgR was specific, because excess unlabeled P_flgDE2 DNA, but not P_gyrA DNA (a FlgR-independent promoter), competed for binding (Fig. 1C). In contrast, FlgR_ΔCTD only bound P_flgDE2 at the highest protein concentration examined, but binding was nonspecific, because both unlabeled P_flgDE2 and P_gyrA DNA reduced residual binding of FlgR_ΔCTD to labeled P_flgDE2 DNA (Fig. 1C).

Because our results suggested that the FlgR CTD binds DNA but is not essential for expression of the σ⁵⁴ regulon, we hypothesized that the 5′ end of a C. jejuni FlgR- and σ⁵⁴-dependent promoter may only need to begin with the −24 and −12 σ⁵⁴ binding sites. Thus, we tested if upstream promoter DNA, which usually contains binding sites for a NtrC-like RR, may be removed and not alter expression of C. jejuni σ⁵⁴-dependent genes. Therefore, we analyzed the ability of C. jejuni strains producing WT FlgS and either WT FlgR or FlgR_ΔCTD to express flgDE2::astA transcriptional fusions with 5′ truncations of P_flgDE2. Because the flagellar gene fliK overlaps P_flgDE2, and chromosomal deletions upstream of P_flgDE2 would create fliK mutants that alter expression of the σ⁵⁴ regulon in C. jejuni (19), we analyzed expression of flgDE2::astA with 5′ promoter truncations on plasmids in C. jejuni strains.

Because WT FlgR, but not FlgR_ΔCTD, specifically bound P_flgDE2 DNA from −302 to +79 (Fig. 1 B and C), we considered this promoter fragment potentially sufficient for flgDE2 expression. For all analyses in this work, WT flgR and flgR_ΔCTD were expressed from the native chromosomal flgR locus. As shown in Fig. 1D, WT C. jejuni and the flgR_ΔCTD mutant, which both produced WT FlgS, expressed chromosomal- or plasmid-borne flgDE2::astA with base −302 as the 5′ end of P_flgDE2. More expression was noted in the flgR_ΔCTD mutant than in WT C. jejuni for plasmid-borne flgDE2::astA, which may indicate some DNA conformational changes that artificially promote slightly more expression with FlgR_ΔCTD. Regardless, expression of P_flgDE2 was dependent on both a FlgR protein and σ⁵⁴ (encoded by rpoN; Fig. 1D). Removal of DNA up to base −29 (six bases before the essential σ⁵⁴ binding sites) did not significantly reduce expression of flgDE2::astA in WT C. jejuni or the flgR_ΔCTD mutant. However, deletion of DNA up to base −13 (which removes the essential −24 site for σ⁵⁴ binding) eliminated expression of flgDE2::astA in all strains regardless of the FlgR protein produced (Fig. 1D). Therefore, we concluded that a minimal σ⁵⁴-dependent promoter in C. jejuni likely includes only σ⁵⁴ binding sites at the 5′ end. Furthermore, the FlgR CTD bound DNA in σ⁵⁴-dependent promoters, but DNA binding by FlgR was not essential for gene expression.

The Acetogenesis Pathway Influences in Vivo Activation of FlgR_ΔCTD in the Absence of FlgS.

In strains producing FlgS, FlgR_ΔCTD promoted equal to modestly higher expression of most σ⁵⁴-dependent flagellar genes relative to WT FlgR (Figs. 1D and 2 and Table S1). In a ΔflgS mutant, WT FlgR did not activate σ⁵⁴-dependent flagellar gene expression (Fig. 2) (11). In contrast, FlgR_ΔCTD activated expression of σ⁵⁴-dependent genes without FlgS (Fig. 2) (13). However, the level of gene expression promoted by FlgR_ΔCTD without FlgS was ∼20% of the level of C. jejuni with a WT FlgSR TCS (Fig. 2). Furthermore, FlgR_ΔCTD required phosphorylation of D51 in the receiver domain to activate gene expression in a ΔflgS mutant (13). These results indicated that the FlgR CTD may limit in vivo phosphotransfer to FlgR by noncognate HKs or other phosphodonors, in addition to a nonessential DNA-binding activity.

Fig. 2.

Identification of darkhelmet Tn mutants with reduced FlgR_ΔCTD activity in ΔflgS mutants. Arylsulfatase assay examining expression of the σ⁵⁴-dependent flaB::astA transcriptional fusion in C. jejuni strains grown on MH agar. The level of flaB::astA expression in each strain is relative to WT C. jejuni producing the WT FlgSR TCS, which was set to 100 units. The FlgS and FlgR proteins produced in each strain are indicated. ΔCTD indicates FlgR_ΔCTD. Dashes indicate deletions of respective genes. Tn mutants are in the C. jejuni ΔastA ΔflgS flgR_ΔCTD flaB::astA background. Error bars indicate SDs. (Inset) Acetogenesis pathway as outlined in E. coli with minor modifications (20), and predicted to be intact in C. jejuni (24, 25).

To search for in vivo phosphodonors that activate FlgR_ΔCTD in the absence of FlgS, we performed transposon (Tn) mutagenesis with the darkhelmet Tn in C. jejuni ΔflgS flgR_ΔCTD flaB::astA and identified seven of ∼3,750 Tn mutants with seven- to 180-fold reductions in expression of the σ⁵⁴-dependent flaB::astA transcriptional reporter (Fig. 2). All mutants contained a Tn in genes likely affecting the acetogenesis pathway, a multistep process that converts pyruvate to ATP and acetate (reviewed in ref. 20). An intermediate in this pathway is AcP, a low molecular-weight phosphodonor often used to autophosphorylate many RRs in vitro, and a few RRs in vivo (8, 21–23). Six independent Tn insertions were within pta, encoding phosphotransacetylase, which reversibly converts acetyl-CoA (Ac-CoA) and inorganic phosphate to CoA and AcP (Fig. 2). One mutant contained a Tn in sdaA, encoding serine dehydratase, which converts serine to pyruvate upstream of the pathway. Immediately downstream of pta on the C. jejuni chromosome is ackA, encoding acetate kinase, which reversibly converts AcP and ADP to acetate and ATP in the acetogenesis pathway. Thus, Tn insertions in pta may have had polar effects on ackA expression and eliminated AcP production altogether. Mutants with Tn insertions in genes encoding HKs were not identified in this screen. These results suggested that fluctuations in AcP biosynthesis from the acetogenesis pathway may have directly influenced in vivo phosphorylation of FlgR_ΔCTD and σ⁵⁴-dependent flagellar gene expression.

Initial studies suggested that the acetogenesis pathway is intact in C. jejuni as in Escherichia coli (20, 24, 25). To analyze possible in vivo AcP-mediated activation of FlgR_ΔCTD, C. jejuni mutants predicted to produce different intracellular levels of AcP when grown on Mueller–Hinton (MH) agar were made: ΔackA, high AcP; Δpta, low AcP; and Δpta ΔackA, negligible AcP. Although we could not directly measure AcP levels in C. jejuni, our data described below suggested that the C. jejuni mutants produced similar trends in AcP levels as respective E. coli mutants (22, 26, 27). When ackA was deleted in the ΔflgS flgR_ΔCTD mutant, flaB::astA expression increased 4.5-fold and was 70% of that observed in WT C. jejuni with an intact FlgSR TCS (Table 1). Stepwise decreases in AcP levels by mutating pta alone and both pta and ackA reduced flaB::astA expression to negligible levels in the ΔflgS flgR_ΔCTD mutant (Table 1). Mutation of the phosphorylated D51 residue in FlgR_ΔCTD abolished all effects of the acetogenesis pathway on flaB::astA expression in the absence of FlgS (Table 1). In contrast to the ΔflgS flgR_ΔCTD mutant, flaB::astA expression in C. jejuni producing WT FlgR in a ΔflgS mutant was only slightly increased by deleting ackA (Table 1). We next examined if the acetogenesis pathway could influence FlgR or FlgR_ΔCTD activity in the presence of FlgS. Deletion of ackA did not increase flaB::astA expression in WT C. jejuni (Table 1). However, FlgR_ΔCTD-stimulated flaB::astA expression in the presence of FlgS increased 18% when ackA was deleted (Table 1). These results verified that the acetogenesis pathway significantly modulated FlgR_ΔCTD activity for promoting σ⁵⁴-dependent gene expression, both in the presence and absence of FlgS.

Table 1.

Effect of the acetogenesis pathway on activation of FlgR proteins and σ⁵⁴-dependent gene expression

	FlgS/FlgR protein produced		pta ackA genotype
flgS flgR genotype	FlgS	FlgR	WT	ΔackA	Δpta	ΔptaA ΔackA
WT	WT	WT	100 ± 6.9*	96.4 ± 3.8	ND†	ND
ΔflgR	WT	–	0.3 ± 0.1	ND	ND	ND
flgRΔCTD	WT	FlgRΔCTD	98.3 ± 3.2	116.0 ± 5.9	ND	ND
ΔflgS	–	WT	0.3 ± 0.0	1.8 ± 0.6	0.3 ± 0.1	0.1 ± 0.0
ΔflgS flgRΔCTD	–	FlgRΔCTD	15.3 ± 2.3	70.6 ± 14.4	4.3 ± 0.5	0.3 ± 0.1
ΔflgS flgRD51AΔCTD	–	FlgR D51AΔCTD	0.2 ± 0.1	0.2 ± 0.0	0.3 ± 0.1	0.2 ± 0.2

*Arylsulfatase assay examining flaB::astA expression after growth on MH agar. The level of flaB::astA expression (±SD) in each strain is relative to WT C. jejuni producing the WT FlgSR TCS, which was set to 100 units.

^†Not determined.

The CTD Is a Specificity Determinant That Limits FlgR Autophosphorylation via AcP.

Our results suggested that in vivo FlgR_ΔCTD autophosphorylation via AcP is likely enhanced relative to WT FlgR. Thus, we monitored the in vitro ability of FlgR proteins to autophosphorylate with AcP as the sole phosphodonor. In our assays, we observed modification of both proteins with Ac[³²P], with at least a two- to threefold greater level of autophosphorylation of FlgR_ΔCTD compared with WT FlgR (Fig. 3A and Fig. S1). Autophosphorylation of both RRs via AcP was increased as protein levels increased and was inhibited by mutation of the phosphorylated D51 residue. Enhanced modification of FlgR_ΔCTD was specific for AcP, because phosphotransfer from FlgS to FlgR_ΔCTD was ∼25–70% less than WT FlgR over time (Fig. 3B). Therefore, the CTD specifically limited cross-talk and phosphotransfer from AcP to FlgR.

Fig. 3.

In vitro phosphorylation of FlgR proteins. (A) Autophosphorylation of FlgR proteins with Ac[³²P] or [γ³²P]ATP. WT FlgR or FlgR_ΔCTD (50 and 100 pmol) or respective D51A mutants (100 pmol) are indicated. (B) Phosphotransfer from FlgS to FlgR proteins. Six picomoles of WT FlgR or FlgR_ΔCTD were mixed with autophosphorylated ³²P-FlgS and removed after 15–120 s.

Reprogramming Activation of FlgR_ΔCTD by AcP Through Metabolism.

AcP-mediated activation of FlgR_ΔCTD suggested that deletion of the CTD allowed FlgR to directly respond to the nutritional status of the cell through the acetogenesis pathway, in addition to responding to steps in flagellar biosynthesis via signal transduction through FlgS. Therefore, we tested if altering physiology of C. jejuni ΔflgS mutants would stimulate FlgR_ΔCTD more than WT FlgR to augment both flagellar gene expression and flagellar biosynthesis. In addition, we analyzed whether FlgR or FlgR_ΔCTD activity could be modulated by altering physiology even in the presence of FlgS.

To determine if FlgR could be reprogrammed to respond to metabolic cues, we grew C. jejuni strains on Campylobacter defined media (CDM) with increasing pyruvate concentrations to increase intracellular AcP levels via the acetogenesis pathway. No changes in flaB::astA expression were observed in WT C. jejuni with an intact FlgSR TCS by deleting ackA or by increasing pyruvate levels in the media (Fig. 4). However, we observed a 9–25% increase in flaB::astA expression when strains producing WT FlgS and FlgR_ΔCTD were grown in increasing concentrations of pyruvate to stimulate AcP biosynthesis (Fig. 4). These results suggested that even in the presence of WT FlgS, FlgR_ΔCTD, but not WT FlgR, is responsive to AcP, which results in augmented levels of gene expression.

Fig. 4.

FlgS-independent activation of FlgR_ΔCTD by altering C. jejuni physiology. Arylsulfatase assay examining flaB::astA expression after growth on CDM alone (white bars), CDM with 50 mM (gray bars), or 100 mM (black bars) excess sodium pyruvate. The level of flaB::astA expression in each strain is relative to WT C. jejuni producing the WT FlgSR TCS grown on CDM alone, which was set to 100 units. The FlgS, FlgR, and AckA proteins produced in each strain are indicated. ΔCTD indicates FlgR_ΔCTD. Dashes indicate deletions of respective genes. Error bars indicate SDs.

We next examined flaB::astA expression upon modulating the physiology of C. jejuni strains lacking FlgS. As a result of deletion of flgS, AcP is presumably the only possible phosphodonor for FlgR proteins. flaB::astA expression in ΔflgS flgR_ΔCTD after growth on CDM alone was only 6% of that of WT C. jejuni with an intact FlgSR TCS grown on the same media (Fig. 4). However, upon increasing pyruvate concentrations, flaB::astA expression increased ∼14-fold (Fig. 4); this enhanced FlgS-independent, pyruvate-stimulated flaB::astA expression was 81–88% of that of WT C. jejuni grown in similar levels of pyruvate. When ackA was deleted, the baseline level of flaB::astA expression in C. jejuni ΔflgS flgR_ΔCTD grown in CDM increased (Fig. 4). In higher pyruvate concentrations, flaB::astA expression in the ΔflgS flgR_ΔCTD ΔackA mutant equaled and even surpassed WT C. jejuni with an intact FlgSR TCS by 26% (Fig. 4). Deletion of both pta and ackA prevented activation of FlgR_ΔCTD and flaB::astA expression regardless of pyruvate concentrations, presumably due to low AcP levels in this mutant (Fig. S2).

In contrast, C. jejuni producing WT FlgR without FlgS was not responsive to increasing pyruvate levels as measured by flaB::astA expression (Fig. 4). However, by presumably increasing intracellular AcP levels through deletion of ackA, an appreciable increase in FlgS-independent gene expression by WT FlgR was observed, which was further augmented two- to threefold by increasing pyruvate levels (Fig. 4). However, flaB::astA expression levels were dramatically lower with AcP-dependent activation of WT FlgR than FlgR_ΔCTD.

We next determined if levels of flagellation could be changed via AcP-mediated activation of WT FlgR or FlgR_ΔCTD by altering C. jejuni physiology. We monitored the number of flagella on individual bacteria after growth on CDM alone or CDM with excess pyruvate. Approximately 93–97% of WT C. jejuni cells produced a single flagellum at one or both poles when grown on CDM or CDM with pyruvate or with deletion of ackA (64–75% with a flagellum at each pole, 21–29% with a flagellum only at one pole; Table 2 and Table S2). Similar levels of flagellation were observed in the flgR_ΔCTD mutant producing WT FlgS, regardless of pyruvate levels in media or the presence of ackA (Table S2). Considering that pyruvate augmented FlgR_ΔCTD-dependent flagellar gene expression but not flagellation in the presence of FlgS (Fig. 4 and Table S2), these findings suggested that other factors besides expression of flagellar rod and hook proteins are rate-limiting steps in flagellation. We next analyzed if growth on pyruvate could stimulate WT FlgR or FlgR_ΔCTD in ΔflgS mutants to result in maximal flagellation, despite lacking the ability to use FlgS to link expression of the σ⁵⁴ regulon to flagellar T3SS formation. Even though we observed pyruvate-stimulated activation of WT FlgR in the absence of FlgS that slightly increased flagellar gene expression (Fig. 4), this strain was aflagellated under all growth conditions, with or without ackA mutation. For the ΔflgS flgR_ΔCTD mutant, most cells were aflagellated after growth on MH or CDM agar. However, growth on CDM with excess pyruvate resulted in ∼30% of cells producing flagella (Table 2). Upon deletion of ackA, the population of flagellated ΔflgS flgR_ΔCTD cells increased to 56% with pyruvate supplementation (Table 2). Despite increases in flagellation, the number of ΔflgS flgR_ΔCTD cells producing two flagella (the predominant WT phenotype) was always lower than WT C. jejuni. Thus, removal of the CTD reprogrammed FlgR to respond to the nutritional status of C. jejuni via AcP levels and resulted in similar levels of gene expression as WT C. jejuni with an intact FlgSR TCS (Fig. 4). However, coupling FlgR activation to the flagellar T3SS through the cognate FlgS HK promoted optimal flagellation.

Table 2.

Flagellation due to AcP-dependent activation of FlgR proteins by altering C. jejuni physiology

		No. of flagella per bacterium*
Strain	Media	2	1	0
WT	CDM	64 ± 7	29 ± 7	6 ± 0
WT	CDM + pyruvate	67 ± 1	27 ± 6	6 ± 5
ΔflgS flgRΔCTD	CDM	0 ± 0	1 ± 0	99 ± 0
ΔflgS flgRΔCTD	CDM + pyruvate	4 ± 1	26 ± 4	71 ± 3
ΔflgS flgRΔCTD ΔackA	CDM	0 ± 0	5 ± 3	95 ± 3
ΔflgS flgRΔCTD ΔackA	CDM + pyruvate	8 ± 3	48 ± 8	44 ± 11

*The percentage of the population producing a single flagellum at both poles (two flagella), a single flagellum at one pole (one flagellum), or no flagella (zero) after growth on CDM or CDM with 100 mM excess sodium pyruvate are indicated. Data are the average of two experiments ± SD.

In Vivo Activation of FlgR_ΔCTD by AcP Is Not Reduced by a Potential FlgS Phosphatase Activity.

In addition to functioning as kinases, many HKs are bifunctional with a phosphatase activity for their cognate RRs. If conditions favor the HK possessing a net phosphatase activity, the level of the phosphorylated cognate RR decreases. The HK can dephosphorylate the RR regardless of the original phosphodonor used to autophosphorylate the RR. To date, a potential FlgS phosphatase activity has not directly been examined. If one exists, then in vivo AcP-mediated activation of FlgR may be prevented both by the CTD and FlgS.

Flagellar T3SS components such as FlhA are required to stimulate FlgSR signal transduction and transcription of σ⁵⁴-dependent flagellar genes (11, 14). A flgR_ΔCTD mutant was able to express flaB::astA in the absence of FlhA and/or FlgS, due to AcP-mediated activation of FlgR_ΔCTD (Fig. 5) (14). More importantly, the level of flaB::astA expression in the ΔflhA flgR_ΔCTD mutant (which produces FlgS with low autokinase activity) was at least the same as the ΔflgS flgR_ΔCTD mutant. Thus, FlgS when inactive as a kinase failed to protect FlgR_ΔCTD from cross-talk with AcP and reduce gene expression. Furthermore, these findings demonstrate the importance of the CTD in eliminating cross-talk of FlgR with AcP and influences from metabolism. As such, the CTD prevents AcP-mediated activation of FlgR and unnecessary flagellar gene expression and protein production in a bacterium unable to synthesize flagella due to a deficiency in producing a flagellar T3SS. Thus, the CTD is a key specificity determinant in the FlgSR TCS to prevent cross-talk between FlgR and endogenous metabolic phosphodonors.

Fig. 5.

Examination of an in vivo FlgS phosphatase activity for AcP-activated FlgR_ΔCTD. Arylsulfatase assay examining flaB::astA expression after growth on MH agar. The level of flaB::astA expression in each strain is relative to WT C. jejuni producing the WT FlgSR TCS, which was set to 100 units. The FlgS, FlgR, and FlhA proteins produced in each strain are indicated. ΔCTD indicates FlgR_ΔCTD. Dashes indicate deletions of respective genes. Error bars indicate SDs.

Discussion

Maintaining signaling fidelity within a TCS is essential for promoting the correct type and degree of response to particular stimuli. In many TCSs, HKs serve as phosphodonors for cognate RRs to result in proper output responses. If mechanisms do not exist to limit specificity of communication of each HK to a cognate RR, intersystem cross-talk between two TCSs may cause signaling interference and obstruct responses to activating stimuli. Mechanisms identified so far that promote intrasystem signaling fidelity and limit intersystem cross-talk include molecular recognition between cognate HK and RR pairs and phosphatase activities of some HKs that control levels of phosphorylation of cognate RRs (1, 3–6).

We identified a domain in a RR that limits in vivo cross-talk specifically from the metabolite AcP, rather than from a noncognate HK. This specificity determinant in C. jejuni FlgR is located in the CTD, which in most other NtrC-like RRs functions in an essential DNA-binding activity (15–17). Without the CTD, FlgR used AcP as an in vivo phosphodonor for autophosphorylation to stimulate gene expression, even in the presence of the cognate FlgS HK. Furthermore, FlgR_ΔCTD could be reprogrammed to respond to the nutritional status of the cell by altering the acetogenesis pathway and presumably AcP levels to result in levels of expression of the σ⁵⁴ regulon that equaled and exceeded WT C. jejuni with an intact FlgSR TCS. AcP serves as an in vivo phosphodonor for autophosphorylation of many RRs (9), but AcP-dependent activation of RRs is often negated by phosphatase activities of some cognate HKs. However, AcP has been shown to be the sole in vivo phosphodonor of a few RRs that appear to lack cognate HKs, including Borrelia burgdorferi Rrp2, which is also an NtrC-like RR (22, 23). Though our data strongly suggests that AcP directly serves as a phosphodonor for FlgR_ΔCTD, an alternative, yet remote, possibility is that high AcP levels stimulate a noncognate HK that weakly promotes phosphotransfer to FlgR. However, we have not identified another HK that influences FlgR activity in various genetic screens we have conducted in C. jejuni.

A major question arising from this work is how the CTD limits in vivo cross-talk between AcP and FlgR. Barbieri et al. (28) recently suggested a possible explanation for how phosphotransfer from small phosphodonors to some RRs may be influenced by interdomain interactions within a RR. By examining members of the OmpR/PhoB RR family, correlations between the in vitro ability of a protein to autophosphorylate and the degree of interdomain interactions were found. More extensive contacts between the receiver and effector domains of a RR were proposed to stabilize an inactivate state that is a barrier to autophosphorylation using small phosphodonors. Thus, interdomain interactions may naturally reduce substantial phosphotransfer from small phosphodonors to many RRs in TCS.

In light of this study, we attempted to determine structural differences between WT FlgR and FlgR_ΔCTD that may explain how the CTD reduces or prevents AcP-dependent FlgR_ΔCTD autophosphorylation. However, like many NtrC family members, the FlgR proteins were refractory to crystallization for structural analyses. Docking of atomic models of individual domains of Salmonella typhimurium NtrC have provided the most complete structure possible for this family of RRs (29). In this model, the CTD does not appear to be near to or interact with the N-terminal domain. Therefore, considerable doubt exists that the CTD of FlgR may directly obstruct the receiver domain, thereby limiting its ability to use AcP as a phosphodonor. It is possible that deletion of the CTD of FlgR may cause small structural changes in the protein, which results in a receiver domain with a conformation that more readily accepts AcP as a phosphodonor.

Even though we created physiological conditions that promoted AcP-dependent activation of FlgR_ΔCTD to result in levels of flagellar gene expression equivalent to WT C. jejuni with an intact FlgSR TCS, linking gene expression to the nutritional status of the cell did not result in WT flagellation. An ordered expression of subsets of flagellar genes is required for proper flagellar biosynthesis. Linking FlgR activation via FlgS to the flagellar T3SS appears to allow for correct temporal expression of σ⁵⁴-dependent genes for flagellation. Because the σ⁵⁴ regulon encodes rod and hook proteins, expression of these genes is necessary only after the T3SS has formed so that ordered protein secretion occurs for efficient flagellar biosynthesis. AcP-dependent activation of FlgR does not allow for precise temporal gene regulation relative to T3SS formation, and consequently reduced flagellation in C. jejuni. Whereas C. jejuni would rely on environmental nutrient composition for AcP-mediated activation of FlgR for motility, FlgS activation of FlgR allows the bacterium to produce flagella independently of exogenous factors, which likely enhances in vivo fitness of C. jejuni in hosts. Furthermore, the CTD allows FlgR to remain insensitive to AcP and influences from metabolism, preventing unnecessary flagellar gene expression and protein synthesis when flagella cannot form due to incomplete initial stages of flagellar biosynthesis.

We also found that FlgR activated WT levels of gene expression in the absence of DNA binding by the CTD or without DNA upstream of a σ⁵⁴-dependent promoter. These results support the hypothesis that FlgR may initiate transcription of some σ⁵⁴-dependent genes without binding DNA. We propose that FlgR may interact in a soluble state with σ⁵⁴ in a RNA polymerase holoenzyme that is bound to target promoters, negating the need for FlgR to be tethered directly to DNA. A similar mechanism has been proposed for a FlgR homolog in Helicobacter pylori, which naturally lacks a CTD (30). These features of FlgR proteins in different bacteria may be significant in expanding mechanisms of transcriptional initiation by the NtrC family of RRs, because many of these proteins promote only limited transcription in the absence of DNA binding under physiological conditions (16, 17).

Some HKs are bifunctional, with net autokinase and phosphatase activities that control the level of cognate RR phosphorylation, depending upon different conditions. In a flagellar T3SS mutant, the autokinase activity of FlgS is low, but AcP-mediated activation of FlgR_ΔCTD was not reduced. In the absence of a significant in vivo phosphatase activity of FlgS, these data demonstrate the importance of the CTD as a major specificity determinant to limit cross-talk between FlgR and AcP when conditions are not conducive for flagellation. Because this specificity determinant resides in the common DNA-binding domain of other NtrC family members, the possibility exists that this domain reduces AcP-mediated activation in these RRs to maintain intrasystem signaling fidelity. Furthermore, our findings suggest that phosphorylated regulators in other bacteria and more complex eukaryotic systems may possess domains with similar activities in providing insulation from cross-talk with small phosphodonor metabolites.

Materials and Methods

Information regarding growth of C. jejuni 81–176 strains is provided in SI Materials and Methods. Methods for generating chromosomal mutants and transposon mutants, plus a list of strains and plasmids used in this study, are located in SI Materials and Methods and Tables S3 and S4. A detailed description of experimental methods is also provided in SI Materials and Methods.