A Campylobacter jejuni sensor phosphatase-driven two-component system integrates metabolic phosphodonors as cues in signal transduction

Abstract

Canonical bacterial two-component signal transduction systems (TCSs) detect cues by sensors with opposing kinase and phosphatase activities to alter the level of phosphorylation of cognate response regulators to mediate responses. We previously identified the Campylobacter jejuni BumSR TCS as the founding member of a bacterial TCS family in which the sensor solely functions as a phosphatase. Sensing specific intestinal metabolites inhibits BumS dephosphorylation of phospho-BumR (P-BumR) to impact BumR as a transcriptional regulator of genes influencing host colonization. Since BumS lacks kinase activity, BumR must depend upon a non-cognate phosphodonor in the bacterium to form P-BumR. Through a genetic screen and selection, physiological assays, and biochemical analysis, we identified acetyl phosphate (AcP) and carbamoyl phosphate (CP) as natural in vivo phosphodonors for BumR. In C. jejuni, AcP and CP are products of metabolic pathways fueled by amino acids favored by the bacterium as carbon sources for growth. Producing and utilizing AcP and CP as bona fide BumR phosphodonors allows BumSR to integrate different types of inputs for signal transduction. Microbiota-generated gut metabolites are cues for BumS to control its phosphatase activity for P-BumR and inform about the spatial location of C. jejuni in host intestines. In contrast, AcP and CP are cues for BumR that directly influence P-BumR levels and activity and inform about the richness of favored carbon sources in intestinal niches for optimal energy generation and metabolism. Our study reveals how a bacterial TCS strategically integrates information from multiple cues through both essential components for optimal signal transduction.

Introduction

Bacterial two-component signal transduction systems (TCSs) sense specific cues to execute an output response that alters the behavior of the bacterium so it can adapt to different conditions. A canonical TCS utilizes a sensor histidine kinase (HK) that detects one or more specific cues to autophosphorylate on a histidine residue (1–3) (Fig S1AB). A cognate response regulator (RR) uses the phosphohistidine on the HK as a phosphodonor to autophosphorylate a specific aspartate, which changes the activity of the RR to promote a response (1, 2). For RRs that are transcriptional regulators, phosphorylation alters the ability of RRs to bind DNA and change the level of transcription of target genes (4, 5). Many HKs also possess an opposing phosphatase activity to dephosphorylate its cognate RR (6, 7). This phosphatase activity reduces crosstalk between the cognate RR and other HKs or phosphometabolites in the bacterial cell and signaling noise (8). As such, bacterial TCSs have evolved as finely-tuned signal transduction systems so that a precise response occurs for a specific cue.

We recently identified the Campylobacter jejuni BumSR TCS as the first member of a family of sensor phosphatase-driven TCSs (9). In this TCS family, the sensor lacks kinase activity and only functions as a phosphatase to dephosphorylate the cognate RR (9–12) (Fig S1CD). We found that the C. jejuni BumSR TCS directs a response to specific microbiota-derived short-chain and branched-chain fatty acid metabolites such as butyrate, isobutyrate, and isovalerate that are enriched in the lower intestinal tract of human and avian hosts, which are favored niches for colonization by the bacterium (9, 13–17). Detection of these cues by BumS ultimately alters the activity of the BumR RR to either activate or repress transcription of specific C. jejuni genes involved in colonization of hosts. In vitro biochemical assays revealed that in the absence of cues, BumS dephosphorylated phospho-BumR (P-BumR) (9, 13). However, upon sensing isobutyrate or isovalerate, BumS phosphatase activity for P-BumR was inhibited (13). Further analysis revealed that P-BumR has higher DNA-binding activity for target promoters than BumR to repress transcription of some genes like peb3, encoding a transporter for phosphate-containing metabolites such as 3-phosphoglycerate (18, 19), or increase transcription of other genes like Cjj81176_0438 and Cjj81176_0439, encoding a gluconate hydrogenase complex for energy generation (20). Peb3, Cjj81176_0438, Cjj81176_0439, and other members of the BumSR regulon are required for WT levels of commensal colonization of the chick intestinal tract (9, 20). Furthermore, we and others have shown that the BumSR TCS is required for optimal commensal colonization of chicks and BumR is essential for infection of human volunteers (9, 13, 21). We proposed that detection of specific lower intestinal metabolites by BumS allows C. jejuni to spatially discern its location in the lower intestinal tract of hosts and then alter gene expression via P-BumR to assist the bacterium in colonizing this preferred area of the intestinal tract.

For this family of bacterial sensor phosphatase-driven TCSs (9–12), employing a sensor that only functions as a phosphatase creates a problem for the cognate RR in that it must use a non-cognate cellular phosphodonor to autophosphorylate and change its activity to execute a response. Non-cognate HKs or phosphometabolites generated in metabolism such as acetyl-phosphate (AcP), carbamoyl phosphate (CP), or Z nucleotides have been shown to serve as in vitro phosphodonors for some RRs (22–32). However, this type of crosstalk with non-cognate phosphodonors is not typically observed in WT cells due to the specificity of interactions of HK-RR pairs, the innate phosphatase activity of a HK for its RR, and fluctuations in phosphometabolite concentrations. For at least the E. coli UvrY and Salmonella SirA RRs, AcP is a bona fide in vivo phosphodonor (29, 33), with Z nucleotides also influencing phosphorylation of UvrY (34).

In order to understand signal transduction mechanisms of a sensor phosphatase-driven TCS, we conducted a genetic screen, a genetic selection, biochemical analyses, and physiological assays to identify bona fide non-cognate phosphodonors in the C. jejuni cell for the BumR RR of the BumSR TCS. Results from our study revealed that AcP and CP, which are generated by catabolism of specific amino acids favored by C. jejuni, are phosphodonors for BumR. Furthermore, we demonstrate that BumR transcriptional regulatory activity is influenced by the availability and metabolism of these favored amino acids. These results indicate that the evolutionary design of the sensor phosphatase-driven BumSR TCS allows the TCS to integrate multiple inputs, one through the BumS sensor phosphatase directly detecting a cue to inform of its geographical location in the intestines to change its phosphatase activity for its cognate BumR RR and another to inform about nutrient richness in niches that contributes to autophosphorylation of BumR to alter its ability to mediate an output response. These findings expand the information processing and signaling strategies that are available to bacterial TCSs.

Results

Campylobacter jejuni HKs are not required for BumR phosphorylation.

A possible phosphodonor in the C. jejuni cell that may influence autophosphorylation of BumR and its activity in executing a response is a non-cognate HK of another TCS. TCS HKs can contribute to phosphorylation of non-cognate RRs in vitro, although this type of signaling crosstalk is rarely observed in vivo due to the endogenous phosphatase activity of HKs for their cognate RRs to preserve signaling fidelity (7, 8, 35–38). C. jejuni produces six TCS HKs each with their own cognate RR to control various responses: PhoS (phosphate homeostasis), RacS and DccS (colonization processes), FlgS (flagellar assembly), CprS (growth and metabolism), and CheA (chemotaxis responses) (39–44).

We investigated whether these HKs contributed to P-BumR formation in C. jejuni. To indirectly measure levels of P-BumR, we examined P-BumR activity for repressing peb3::astA expression via a transcriptional arylsulfatase reporter assay in WT C. jejuni and isogenic mutants. P-BumR is at its highest level in ΔbumS due to lack of BumS phosphatase activity, which causes maximal repression of peb3::astA expression (Fig. 1; Fig. S1CD; (9)). Therefore, we insertionally inactivated each gene encoding a different HK in C. jejuni ΔbumS for more sensitive detection of a drop in P-BumR levels. Inactivation of a non-cognate HK that contributed to P-BumR formation in ΔbumS would result in a reduction in P-BumR levels and P-BumR-mediated repression to cause increased peb3 expression relative to C. jejuni ΔbumS. However, the level of peb3::astA expression was unchanged when genes for five of the HKs were mutated in ΔbumS (Fig. 1). We did observe increased repression of peb3::astA in C. jejuni ΔbumS ΔracS relative to ΔbumS. While this finding indicated that loss of RacS did not reduce P-BumR levels or activity, peb3 transcription may also be regulated by a RacS-dependent pathway. These data suggest that none of the other six C. jejuni non-cognate HKs contribute to BumR phosphorylation. Thus, BumR likely relies upon another type of endogenous phosphodonor in the C. jejuni cell.

Figure 1. Analysis of C. jejuni non-cognate TCS HKs as potential phosphodonors for BumR.

Arylsulfatase reporter assay of peb3::astA expression in WT C. jejuni, ΔbumS, or ΔbumS also lacking one of six C. jejuni TCS HKs. Strains were grown in Brucella media alone. Level of peb3::astA expression in all strains is relative to WT, which was set to 1. Results from a representative assay with each strain tested in triplicate are presented. Error bars indicate standard deviations of the average level of peb3::astA expression. Statistical significance in peb3::astA expression compared to C. jejuni ΔbumS was calculated by ANOVA and Dunnett’s multiple comparison test (*, P < 0.05).

A genetic selection identifies CP as a potential BumR phosphodonor.

We developed a genetic selection to identify a BumR phosphodonor by creating a transcriptional reporter that fused a promoterless chloramphenicol-resistant gene to the peb3 promoter (peb3::cat). Replacement of peb3 on the WT C. jejuni chromosome with peb3::cat allowed for an indirect measurement of P-BumR levels through chloramphenicol resistance. Deletion of bumS from this reporter strain or growth of the strain on media with exogenous 12.5 mM butyrate (which inhibits BumS phosphatase activity) increased P-BumR levels to repress expression of peb3::cat and cause sensitivity to 5 μg/ml chloramphenicol (Fig. 2A and 2B). In contrast, deletion of bumR resulted in full derepression of peb3::cat and chloramphenicol resistance (Fig. 2B).

Figure 2. Characterization of C. jejuni peb3 reporter strains.

(A) Diagram of BumSR-mediated control of peb3::astA and peb3::cat reporter activity. Increased isobutyrate, isovalerate, or butyrate inhibits BumS phosphatase activity for P-BumR causing higher levels of P-BumR and low expression of peb3::astA or peb3::cat (for chloramphenicol sensitivity, Chl^S). In conditions without these gut metabolites, BumS dephosphorylated P-BumR to increase BumR levels and derepress peb3::astA or peb3::cat expression (for chloramphenicol resistance, Chl^R). (B) Comparison of ΔbumS, ΔbumR, and WT C. jejuni reporter strains Brucella media without (top row) or with (bottom row) 12.5 mM butyrate. Agar plates in bottom row contained X-S (left) to monitor arylsulfatase production via peb3::astA expression, no chloramphenicol (middle), or 5 μg/ml chloramphenicol (right) to monitor peb3::cat expression.

We exploited the heightened chloramphenicol sensitivity of C. jejuni ΔbumS peb3::cat for use as a reporter strain in a genetic selection to identify genes contributing to P-BumR phosphodonors. Disruption of a gene required to form a BumR phosphodonor in ΔbumS peb3::cat would presumably reduce P-BumR levels, derepress peb3::cat expression, and select for mutants with increased chloramphenicol resistance. For this approach, we performed in vitro transposon (Tn) mutagenesis with the ripley Tn (harboring a kanamycin-resistance gene) and ΔbumS peb3::cat genomic DNA. C. jejuni ΔbumS peb3::cat was naturally transformed with the transposed DNA and mutants with potential Tn insertions were selected for on Brucella agar with 7-10 μg/mL of chloramphenicol and 100 μg/mL kanamycin. We isolated 37 chloramphenicol-resistant ΔbumS peb3::cat ripley mutants out of ~25,000 generated Tn mutants. The majority of Tn insertions were found within three genes in the bumSR locus and an adjacent genomic region. Isolating ten mutants with Tn insertions in bumR served as a control since eliminating bumR derepressed peb3::cat expression to allow for growth on chloramphenicol (Fig 2B). We also isolated seven mutants with Tn insertions in Cjj81176_1487 (a gene encoding a protein of unknown function) and six mutants with Tn insertions in the downstream gene carA. These two adjacent genes are upstream of and separated from bumSR by Cjj81176_1485. carA encodes the small subunit of carbamoyl phosphate synthetase (45). CarA with CarB, the large subunit of the carbamoyl phosphate synthase complex, forms carbamoyl phosphate (CP) (46–48). CarA converts glutamine into glutamate and ammonia, which with ATP and bicarbonate is a substrate for CarB to form CP that can feed pathways for arginine or pyrimidine synthesis (Fig 3A; (48)). CP has been shown to serve as an in vitro phosphodonor for some RRs (22, 23, 27, 32), therefore making it a possible candidate as an endogenous in vivo BumR phosphodonor. Since Tn insertions in Cjj81176_1487 may have caused polar effects on transcription of carA, we focused on CP synthesized by carA and carB as a potential phosphodonor for BumR.

Figure 3. Identification of carbamoyl phosphate pathway in influencing BumR activity.

(A) Predicted carbamoyl phosphate (CP) pathway C. jejuni. The carbamoyl phosphate synthase complex consists of CarA and CarB that form CP. CP can be diverted to form citrulline (by ArgF) and then arginine (by ArgG and ArgH). Alternatively, CP can be used to form UTP or CTP through the pyrimidine biosynthesis pathway. (B) Production of BumR in WT C. jejuni and mutants. Immunoblot analysis of the level of BumR in whole-cell lysates of WT and mutant strains after growth in Brucella media. RpoA served as a control for protein loading. (C) Expression of peb3 in WT C. jejuni and mutants lacking BumS, BumR, or CP production. Strains were grown in Brucella media alone. Level of expression in all strains is relative to WT, which was set to 1. Results from a representative assay with each strain tested in triplicate. Error bars indicate standard deviations of the average level of gene expression. Statistical significance in peb3 expression compared to WT C. jejuni (*, P < 0.05) or ΔbumR (**, P < 0.05) was calculated by ANOVA and Dunnett’s multiple comparison test.

Although carA and the downstream bumSR operon are predicted to possess independent promoters, we ensured that ripley Tn insertions in carA were not affecting bumR expression. We first recreated a carA mutant by naturally transforming WT C. jejuni with genomic DNA from an isolated Tn mutant. We then measured BumR levels via immunoblotting in both WT C. jejuni and the carA::Tn strains. BumR levels remained similar between WT C. jejuni and the carA mutant (Fig. 3B). Thus, interruption of carA in the isolated ripley mutants was likely causing alteration of BumR activity rather than reducing levels of BumR protein for peb3::cat depression.

To validate that mutation of carA impacted peb3 derepression, we compared the level of peb3 transcription via qRT-PCR in WT C. jejuni, ΔbumS, ΔbumR, and carA::Tn. In comparison to WT C. jejuni, peb3 was maximally repressed 3.4-fold in ΔbumS (due to high levels of P-BumR) and maximally derepressed in ΔbumR to cause a 44.1-fold increase in transcription (Fig. 3C). We observed a 5.1-fold increase in peb3 expression in carA::Tn relative to WT, suggesting that mutation of carA likely decreased CP and P-BumR levels for partial derepression of peb3 transcription. We also created a C. jejuni ΔcarB mutant and observed that it had a similar level of peb3 derepression as the carA mutant, further supporting CP as a phosphodonor for BumR (Fig. 3C). However, in the absence of CP in the carA and carB mutants, peb3 was not fully derepressed like in ΔbumR. These findings suggest that in a C. jejuni mutant deficient for CP production, another phosphodonor is likely present to contribute to P-BumR levels to repress peb3 expression.

A genetic screen identifies AcP as an additional phosphodonor for BumR

To identify other potential BumR phosphodonors, we modified a previous genetic screening strategy originally used to identify the BumSR TCS (9). In this previous screen, we employed a peb3::astA transcriptional reporter to identify C. jejuni Tn mutants that failed to repress peb3::astA expression (monitored by increased production of arylsulfatase encoded by astA) in the presence of 12.5 mM butyrate in Campylobacter defined medium (9). Arylsulfatase production in bacterial colonies was monitored by a chromogenic substrate (5-bromo-4-chloro-3-indolyl sulfate; X-S), with arylsulfatase-positive colonies forming a blue pigment whereas negative colonies remain white (Fig. 2B). This strategy led to the identification of darkhelmet Tn insertions in bumS and bumR (9). We performed a similar transposon mutagenesis and screening strategy, but isolated darkhelmet mutants in the WT C. jejuni background on Brucella agar with 12.5 mM butyrate and X-S since this media is richer in nutrients than Campylobacter defined media, allowing for potentially wider diversity of recovered mutants. We identified 79 mutants out of over 10,000 total mutants that formed colonies with increased blue pigment, suggesting failure to fully repress peb3::astA expression in the presence of butyrate (Table 1). Seven mutants had Tn insertions in bumS and 13 had Tn insertions in bumR, which verified our screening strategy.

Table 1.

Identification of C. jejuni darkhelmet Tn mutants with impaired peb3 repression.

Site of Tn Insertion†	Number of Independent Mutants‡	Putative or Known Function
bumS	6	TCS sensor phosphatase
bumR	9	TCS response regulator
pta	18	phosphotransacetylase for AcP synthesis
sdaA	4	l-serine ammonia-lyase
sdaC	8	serine transporter
cheA	1	TCS histidine kinase for chemotaxis
Cjj81176_1428	3	capsular polysaccharide methyltransferase
Cjj81176_1430	1	capsular polysaccharide epimerase
Cjj81176_1431	1	capsular polysaccharide glycosyltransferase
Cjj81176_1434	1	capsular polysaccharide sugar transferase
Cjj81176_1436	6	capsular polysaccharide glycosyltransferase
kpsC	1	capsular polysaccharide Kdo transferase
kpsE	1	capsular polysaccharide transporter
kpsS	1	capsular polysaccharide Kdo transferase
Cjj81176_0316	1	major facilitator superfamily transporter
Cjj81176_1022	3	periplasmic protein
rpsO	1	30S ribosomal protein S15
panC	1	pantothenate synthetase

^†darkhelmet Tn mutants were identified by light blue to blue colony color compared to white colonies with normally repressed peb3::astA reporter expression on Brucella medium with 12.5 mM butyrate

^‡darkhelmet Tn inserted in a different location in gene for each mutant.

We isolated 25 individual mutants with Tn insertions in pta, which encodes phosphotransacetylase in the acetogenesis pathway that converts acetyl-CoA to acetyl phosphate (AcP) (Table 1 and Fig. 4A). AcP, like CP, has been used as an in vitro phosphodonor for other RRs and has been reported to be a bona fide phosphodonor in a bacterial cell for only E. coli UvrY and Salmonella SirA (22–31). We previously showed that BumR can autophosphorylate on D58 in vitro in the presence of AcP (9, 13, 49). These data are evidence supporting AcP could be an in vivo phosphodonor for BumR in the C. jejuni cell to contribute to P-BumR levels.

Figure 4. Identification of serine catabolism and the acetogenesis pathway in influencing BumR activity.

(A) The serine catabolism and acetogenesis pathways of C. jejuni. Serine can be transported by SdaC into the C. jejuni cytoplasm and then deaminated by SdaA to produce pyruvate. Lactate uptake and metabolism by LctP and the LutABC complex can also result in pyruvate. Pyruvate can be diverted to the TCA cycle, but during overflow metabolism, pyruvate can be shunted to the pyruvate oxidoreductase (POR) and converted into acetyl-CoA (Ac-CoA) to enter the acetogenesis pathway. Phosphotransacetylase (Pta) converts Ac-CoA to acetyl-phosphate (AcP), which can then be converted to acetate and ATP by AckA. (B) Expression of peb3 in WT C. jejuni and mutants. Strains were grown in Brucella media alone. Level of expression in all strains is relative to WT, which was set to 1. Results from a representative assay with each strain tested in triplicate. Error bars indicate standard deviations of the average level of gene expression. Statistical significance in peb3 expression compared to WT C. jejuni (*, P < 0.05) or ΔbumR (**, P < 0.05) was calculated by ANOVA and Dunnett’s multiple comparison test.

We also isolated four mutants with Tn insertions in sdaA and eight mutants with Tn insertions in sdaC. SdaC is the main serine transporter in C. jejuni whereas SdaA deaminates serine to form pyruvate (50–52) (Fig 4A). C. jejuni metabolism mainly relies on amino acids, rather than sugars, as carbon sources, with serine being one of the six most favored amino acids removed from the environment by the bacterium for metabolism (53–55). Pyruvate produced from deamination of serine by SdaA can then be converted to oxaloacetate to enter the TCA cycle or converted to acetyl-CoA by the pyruvate oxidoreductase (POR) to enter the acetogenesis pathway (56). In the acetogenesis pathway, AcP is produced by Pta from acetyl-CoA (Fig. 4A; (57)). Thus, identifying mutants with Tn insertions in sdaC and sdaA suggests that serine catabolism may contribute to AcP production to alter P-BumR levels and expression of the bumSR regulon.

In addition to these mutants, we also isolated a mutant with a Tn insertion in cheA (Table 1). However, mutation of cheA did not influence peb3::astA expression (Fig. 1), which suggested that this cheA::Tn mutant was a false positive. We also identified several mutants with Tn insertions in genes presumably involved in capsular polysaccharide biosynthesis (Table 1). It is uncertain how disruption of these genes could affect P-BumR levels or peb3 expression and further investigation was not pursued.

Considering that SdaC, SdaA, and Pta may be linked together in a pathway for AcP generation, we created deletions in each gene in WT C. jejuni to validate their role in influencing peb3 expression. Deletion of pta, sdaA, or sdaC caused 7- to 10.1-fold derepression of peb3 expression, verifying that these factors likely influence P-BumR levels to alter the activity of the protein as a transcriptional repressor for peb3 (Fig. 4B). In trans complementation of the ΔsdaA and ΔsdaC mutants with plasmids expressing a WT copy of each gene significantly restored repression of peb3 transcription (Fig. S2), verifying that SdaA and SdaC and presumably serine acquisition and catabolism were required for full transcriptional control of peb3 expression.

AckA has been shown to have reversible activity in vitro, which could lead to residual AcP levels in the absence of Pta (58). Therefore, we also examined a C. jejuni Δpta ΔackA mutant that lacks AcP. We observed a modest increase in peb3 expression in Δpta ΔackA compared to Δpta. Similar to the carA::Tn or ΔcarB mutants that presumably eliminated CP synthesis, abolishing AcP production alone did not fully derepress peb3 levels to that observed in ΔbumR (Fig. 3C and 4B).

We hypothesized that eliminating both CP and AcP generation could fully eliminate phosphodonors for BumR and P-BumR production. Therefore, we created a Δpta ΔackA carA::Tn mutant that should lack both AcP and CP synthesis. This triple mutant continued to produce BumR levels similar to WT (Fig. 3B), indicating that loss of AcP and CP synthesis did not affect bumR expression. We observed greater derepression of peb3 transcription in this mutant compared to the individual carA::Tn and Δpta ΔackA mutants (Fig 4B). Importantly, this degree of repression in Δpta ΔackA carA::Tn was comparable to ΔbumR that lacks BumR and fully derepresses peb3 expression (Fig 4B). These data indicate that CP and AcP together compose the panel of in vivo phosphodonors for BumR in the C. jejuni cell.

Disruption of the acetogenesis or carbamoyl phosphate synthase pathways could cause off-target side effects in C. jejuni physiology to alter peb3 transcription rather than specifically reducing phosphodonors to modify BumR and alter BumR activity. To differentiate between these possibilities, we compared peb3 expression in carA::Tn and Δpta ΔackA producing WT BumR or BumR_D58E. We previously showed that replacement of BumR D58, the residue that is autophosphorylated to increase activity of BumR as a transcriptional regulator (49), with a glutamate served as a phosphomimetic that caused BumR_D58E to be modestly constitutively active to repress peb3 expression regardless of BumS phosphatase activity (9). We reasoned that if altering CP or AcP levels targeted P-BumR levels rather than caused a general enhancement of peb3 expression independent of BumR, BumR_D58E would be insensitive to reduced AcP and CP levels and continue to repress peb3 transcription in the carA and Δpta ΔackA mutants. As expected, we observed BumR_D58E to continue to repress peb3 expression in carA::Tn and Δpta ΔackA, whereas peb3 expression was derepressed in Δpta ΔackA and carA::Tn producing WT BumR (Fig. 5). These data provide further support that CP and AcP are in vivo phosphodonors for BumR.

Figure 5. Analysis of BumR_D58E activity in cells lacking AcP and CP.

qRT-PCR analysis of peb3 transcription in WT C. jejuni or Δpta ΔackA and carA::Tn mutants producing WT BumR or BumR_D58E. Strains were grown in Brucella media alone. Level of expression in all strains is relative to WT, which was set to 1. Results from a representative assay with each strain tested in triplicate. Error bars indicate standard deviations of the average level of gene expression. Statistical significance in peb3 expression compared to WT C. jejuni (*, P < 0.05) or between Δpta ΔackA or carA::Tn producing BumR_D58E compared to respective strains producing WT BumR (**, P < 0.05) was calculated by ANOVA and Dunnett’s multiple comparison test.

CP enhances BumR binding to the peb3 promoter.

In previous work, we demonstrated that BumR is autophosphorylated on D58 in the presence of ³²P[AcP] in vitro (49), and this modification increased the in vitro ability of BumR to bind target DNA such as the peb3 promoter (9, 13, 49). For further confirmation that CP is a phosphodonor for BumR, we examined phosphorylation of D58 of BumR after incubation with Li-AcP (as a positive control) or Li-CP by mass spectrometry analysis due to the lack of ability to produce ³²P[CP]. This analysis revealed a phosphoryl group on D58 only in BumR samples incubated with AcP or CP (Table S1).

We next examined whether in vitro CP-dependent autophosphorylation of BumR altered the ability of BumR to bind the peb3 promoter, similar to our previous analysis of BumR autophosphorylation with AcP. For these assays, we incubated increasing concentrations of BumR with or without Li-AcP and Li-CP to promote autophosphorylation prior to addition of peb3 promoter DNA for binding analysis by electrophoretic mobility shift assays (EMSAs). In the absence of either phosphodonor, BumR bound the peb3 promoter at the highest concentrations of BumR used (0.75 to 3 μM), similar to our prior studies (9). However, incubation of BumR with AcP or CP promoted BumR binding to peb3 promoter DNA at lower BumR concentrations than BumR alone as evidence by increased levels of peb3 promoter DNA binding and less unbound peb3 promoter remaining (Fig. 6, compare lanes 3-5 with 0.19 - 0.75 μM BumR in each panel). Collectively, these findings indicate that like AcP, CP can serve as a phosphodonor for BumR to modulate BumR binding of the peb3 promoter.

Figure 6. Electrophoretic mobility shift assays for analysis of CP as a phosphodonor for BumR to increase BumR DNA-binding activity.

Recombinant BumR alone or after autophosphorylation with lithium potassium-AcP or lithium-CP was added at concentrations ranging from 0 to 3 μM prior to addition of radiolabeled peb3 promoter DNAs and electrophoresis. The position of unbound peb3 promoter DNA remaining after electrophoresis is indicated.

Glutamine represses peb3 transcription in a CarA and CarB-dependent manner.

Because glutamine is one of six amino acids favored as carbon sources for C. jejuni and is a substrate for carbamoyl phosphate synthase to produce CP (Fig. 3A; (53–55)), we hypothesized that C. jejuni may indirectly sense exogenous glutamine availability via autophosphorylation of BumR with CP generated by CarA and CarB. We predicted that increasing concentrations of glutamine in media could increase intracellular CP to augment P-BumR levels and repress peb3 expression. If so, deletion of carA or carB would eliminate any effect observed by increasing glutamine supplementation since these mutants lack CP production from glutamine. To test this hypothesis, we grew C. jejuni in Brucella media with or without 25 mM glutamine supplementation for 4 h and then monitored peb3 expression. We observed that peb3 expression was further repressed 5.3-fold in WT C. jejuni upon growth in media with increased glutamine (Fig. 7A). As we reported above, peb3 expression was derepressed in the carA and carB mutants due to the lack of CP (Fig 3C and 7A). Unlike with WT C. jejuni, glutamine supplementation did not cause any significant repression of peb3 expression in the carA and carB mutants. These results suggest that activity of the BumSR TCS in controlling gene expression is sensitive to exogenous glutamine converted by CarA and CarB into CP to modify BumR and influence its activity as a transcriptional regulator.

Figure 7. Influence of exogenous carbon sources on BumR activity.

(A) Analysis of peb3 transcription in WT C. jejuni and isogenic carA and carB mutants grown with or without exogenous glutamine supplementation. Strains were grown in Brucella media alone (black bars) or Brucella media with 25 mM glutamine (Gln; red bars). Level of expression in all strains is relative to WT, which was set to 1. Results from a representative assay with each strain tested in triplicate. Error bars indicate standard deviations of the average level of gene expression. Statistical significance in peb3 expression compared to WT C. jejuni (*, P < 0.05) or a strain grown upon glutamine supplementation compared to without glutamine (**, P < 0.05) was calculated by student’s t-test (*, P < 0.05). (B) Analysis of peb3 transcription in WT C. jejuni and isogenic sdaC, sdaA, pta, and ackA mutants grown with or without exogenous serine or pyruvate supplementation. Strains were grown in Brucella media alone (black bars), Brucella media with 25 mM serine (Ser; blue bars), or Brucella media with 25 mM pyruvate (Pyr; yellow bars). Level of expression in all strains is relative to WT, which was set to 1. Results from a representative assay with each strain tested in triplicate. Error bars indicate standard deviations of the average level of gene expression. Statistical significance in peb3 expression compared to WT C. jejuni (*, P < 0.05) or a strain grown upon glutamine supplementation compared to without serine or pyruvate (**, P < 0.05) was calculated by ANOVA and Dunnett’s multiple comparison test.

Exogenous serine and pyruvate cause repression of peb3 expression in C. jejuni.

Like glutamine, serine is one of six amino acids favored by C. jejuni as a carbon source (46, 53, 54). In C. jejuni metabolism, serine is a precursor to pyruvate, which can enter the acetogenesis pathway to generate AcP to serve as a phosphodonor for BumR (Fig 4A). We hypothesized that C. jejuni may sense exogenous serine and/or pyruvate levels through AcP generated from metabolism of these carbon sources and available to BumR to form P-BumR. To test this hypothesis, WT C. jejuni and isogenic mutants lacking sdaC, sdaA, pta, and ackA were cultivated in Brucella media alone or media with 25 mM serine or 25 mM pyruvate for 4 h prior to isolation of RNA for qRT-PCR analysis of peb3 expression. Although we did not observe a decrease in peb3 expression with serine supplementation in WT C. jejuni, we observed a modest 32% decrease in peb3 expression with addition of pyruvate to media (Fig. 7B). Despite serine not causing an expected increase in repression of peb3 transcription in WT C. jejuni, we did observe a 2.5-fold repression in peb3 transcription in C. jejuni ΔackA grown with increased serine supplementation (Fig. 7B). This increased repression is presumably due to increased AcP accumulating to augment P-BumR levels due to the lack of AckA (Fig. 4A). Likewise, we observed a 5-fold repression of peb3 transcription when ΔackA was grown in media supplemented with excess pyruvate (Fig. 7B).

We hypothesized that C. jejuni sdaC and sdaA mutants that are defective for serine transport and conversion to pyruvate, respectively, would be insensitive to serine supplementation to repress peb3 transcription (Fig 4A). As predicted, peb3 levels remained unchanged in the mutants upon growth with increased exogenous serine (Fig 7B). However, supplementation of these mutants with pyruvate, which is a product of serine catabolism by SdaA, restored repression of peb3 expression compared to the mutants when grown in Brucella media alone (Fig. 7B). Supplementation of C. jejuni Δpta with serine or pyruvate failed to alter peb3 expression suggesting that AcP synthesis downstream of serine and pyruvate catabolism is required by C. jejuni to sense these carbon sources, impact P-BumR formation, and modulate peb3 transcription. These data support that C. jejuni senses exogenous serine and pyruvate by monitoring cellular AcP levels.

Discussion

Bacterial TCSs are information-processing systems that link detection of various cues to execution of an output response. Although numerous cues are detected and different responses are modulated by these systems, the working components and mechanics of signal transduction within TCSs are fairly well conserved. In most canonical TCSs, a sensor HK detects one or more specific cues, which influence its inherent kinase or phosphatase activity to control the level of phosphorylation of a respective cognate RR to mediate a response (Fig S1AB). These canonical systems are designed to be insulated from signaling noise from other TCSs and spurious, non-specific high-energy metabolic phosphodonors. However, this canonical signaling mechanism is broken in a newly-revealed family of bacterial TCS exemplified by C. jejuni BumSR in which the sensor functions strictly as a phosphatase ((9–12); Fig S1CD). In these sensor phosphatase-driven systems, the phosphodonor must be a cellular factor outside the TCS. A RR employing an endogenous phosphodonor other than its cognate HK is unusual, but it is a consequence of the mechanics of a sensor-phosphatase driven TCSs like BumSR.

Through our genetic selection and screen, we identified carA and pta mutants with significant decreases in BumR-mediated repression of peb3 expression and presumably dysregulation of other BumSR regulon members. CarA and Pta are directly involved in the synthesis of CP and AcP, respectively, which are high-energy phosphometabolites often used as in vitro phosphodonors for analysis of TCS RRs (22–32). For at least the E. coli UvrY and Salmonella SirA RRs, AcP is a bona fide in vivo cellular phosphodonor (29, 33). Whereas mutations that disrupt each pathway individually caused partial derepression of peb3 expression, mutations that disrupt both pathways in C. jejuni phenocopied a ΔbumR mutant, indicating that BumR utilizes both CP and AcP as phosphodonors in the C. jejuni cell under in vitro growth conditions to modulate its activity as a transcriptional regulator. Furthermore, both phosphometabolites contributed to phosphorylation of D58 on BumR and modulation of the ability of BumR to bind target promoter DNA. These data, combined with the lack of evidence for any non-cognate TCS HK influencing BumR activity, implicate CP and AcP as true in vivo phosphodonors for BumR in the sensor-phosphatase driven BumSR TCS.

Finding CP and AcP as BumR phosphodonors has provocative implications for the type of information these metabolites convey to C. jejuni and how such a sensor-phosphatase driven TCS integrates different signaling inputs for accurate signal transduction to execute an output response. First, our findings suggest that AcP and CP are indirect cues informing the availability of environmental carbon sources. C. jejuni mostly depends on amino acids rather than sugars as carbon sources. Serine and glutamine are two of the six amino acids that are first removed from external sources for growth (53–55). Furthermore, serine and glutamine are precursors for synthesis of AcP and CP, respectively. In this study, we observed that C. jejuni can sense glutamine via BumR and the CP pathway to regulate peb3 expression. As such, deletion of carA and carB abolished the ability of glutamine to modulate BumR activity, indicating that CP synthesis is necessary for this regulatory effect. Therefore, we propose that intracellular CP levels increase in C. jejuni in intestinal niches rich in glutamine to directly increase P-BumR levels and cause expression of certain BumSR regulon members like peb3 to be repressed and others like Cjj81176_0438 and Cjj81176_0439 to increase (9, 13, 49).

We did encounter difficulties in observing that changing concentrations of serine impacted BumR activity in WT C. jejuni. However, in C. jejuni ΔackA, which should accumulate AcP, increasing exogenous serine levels caused significant repression of peb3 expression, which is consistent with a higher level of P-BumR. We also observed that increasing pyruvate levels enhanced BumR activity to repress peb3 transcription in both WT C. jejuni and the ΔackA mutant. These findings suggest that C. jejuni can respond to a serine- or pyruvate-rich environment through BumR with activity of the acetogenesis pathway. Additionally, mutations in sdaA, sdaC, or pta hindered the ability of C. jejuni to respond to serine as these genes are required for serine transport and serine conversion ultimately to AcP. However, ΔsdaA and ΔsdaC mutants remained sensitive to changing exogenous pyruvate levels as pyruvate is a downstream product of serine deamination by SdaA. Collectively, our data indicate that BumR activity is modulated directly by AcP and CP, which likely fluctuate depending on serine, pyruvate, and glutamine levels in various niches. Measuring AcP and CP levels directly in C. jejuni under varying conditions would provide further support for our data presented in this work, but these metabolites are volatile and difficult to assess in bacterial cells. Likewise, we were unable to directly monitor levels of P-BumR in C. jejuni cells, but our analysis of peb3 expression in various mutants and conditions correlated extremely well with predicted changes in P-BumR levels and activity. With these caveats, carbon sources like serine and glutamine appear to be signaling inputs for the BumSR TCS by their conversion into phosphometabolites that directly impact P-BumR levels and BumR transcriptional regulatory activity. Previous studies by us and others have shown that bumR, sdaA, and pta are essential for WT colonization of the natural avian host or infection of the human host, potentially highlighting the importance of serine catabolism and AcP generation on BumR activity for host colonization (9, 17, 21, 51, 52). Investigating the role of CarA and CarB in colonization of hosts would further support both CP and AcP as phosphodonors for BumR to modulate its activity as a transcriptional regulator during infection.

We propose that evolution of BumSR as a sensor-phosphatase driven TCS has facilitated development of a mechanism to integrate multiple signaling inputs through both BumS and BumR to finely tune a transcriptional response for colonization of ideal intestinal niches of hosts (Fig. 8). Through BumS, C. jejuni senses microbiota-derived short-chain and branched short-chain fatty acids such as butyrate, isobutyrate, and isovalerate that are abundant in the lower intestinal tract of avian and human hosts. Sensing these cues via BumS to impact its phosphatase activity for BumR assists the bacterium in spatially discerning between different host intestinal regions to specifically identify the ideal region within the lower intestinal tract to colonize. Meanwhile, through BumR, C. jejuni can sense AcP and CP as indirect measures of sufficient carbon sources like serine and glutamine in local niches for metabolism. By integrating these very different cues through BumS and BumR, the BumSR TCS can process multiple inputs for optimal signal transduction to colonize intestinal niches in multiple hosts (Fig 8).

Figure 8. Model for integration of multiple cues by the C. jejuni BumSR TCS for signal transduction for optimal modulation of gene expression.

Spatial cues, such as butyrate, isobutyrate, and isovalerate are sensed by BumS to inform C. jejuni of its geographical location in the intestinal tract of hosts as these microbiota-generated metabolites are most abundant in the lower intestinal tract of avian, animal, and human hosts. Sensing these metabolites by BumS leads to inhibition of its phosphatase activity for P-BumR. Metabolic cues, such as AcP and CP, are generated when serine, pyruvate, and glutamine are available in the environment. Fluctuations in these nutrients cause changes in AcP or CP levels, which can be sensed by BumR as natural, endogenous phosphodonors. Therefore, AcP and CP serve as metabolic cues of nutrient availability in the immediate niche occupied by C. jejuni. Both spatial and metabolic cues provide signaling inputs to modulate P-BumR levels, which causes changes in colonization and virulence factor expression.

Since our initial identification of BumSR as the first bacterial sensor-phosphatase driven TCS (9), at least three others have been identified in different species and we expect more members of this growing TCS family to be discovered (10–12). While the primary cues sensed by these sensor phosphatases have mostly been identified, the phosphodonors for the respective cognate RRs have not. These RRs may utilize high-energy phosphometabolites such as AcP or CP as phosphodonors like BumR or use another endogenous phosphometabolite or a non-cognate TCS HK. Thorough analysis of each system will be required to determine the phosphodonors for each respective RR and what type of additional information may be conveyed by this phosphodonor to the bacterium. The potential is high for these systems to have evolved similarly as BumSR in integrating multiple inputs, one through the sensor phosphatase and one through the RR to promote signal transduction and an optimal response.

TCS RRs are not normally considered to be sensors. However, many bacterial genomes encode orphan RRs that seemingly are not genetically or functionally linked to a cognate HK (59–61). For many of these orphan RRs, it is unclear whether they are modified by autophosphorylation, and if so, what is the source of the phosphodonor. Based on our work presented herein, it will be interesting to determine if some of these orphan RRs might similarly sense exogenous nutrients, the metabolic status of the cell, or other internal conditions through AcP, CP, or other high-energy phosphometabolites.

Beyond components of the CP and AcP biosynthesis pathways, our genetic screen identified Tn insertions in additional genes that appeared to impact the ability of BumR to repress peb3 expression. Many of these genes encode enzymes or components involved in capsular polysaccharide formation or transport in C. jejuni. Currently, we do not know how these factors may influence BumSR signal transduction. One reason that capsular polysaccharide synthesis may impact BumR activity is that the C. jejuni capsule contains O-methyl phosphoramidate, which could serve as a potential phosphodonor (62). Thus, these mutants may change levels of phosphoramidate-related metabolites in the C. jejuni cell. However, the level of peb3 expression in the Δpta ΔackA carA::Tn mutant, in which both AcP and CP synthesis are eliminated, is similar to that observed in C. jejuni ΔbumR that displays full peb3 derepression. Therefore, under the in vitro growth conditions used in this study, there is minimal room for an alternative phosphodonor to significantly impact BumR activity. Another possibility is that disruption of capsular polysaccharide synthesis could alter levels of certain metabolites that feed into AcP or CP pathways, which impact P-BumR levels and activity. Further investigation is necessary to validate whether these factors significantly impact BumR activity, and if so, how they impact BumSR signal transduction.

Our work has expanded the signaling and information processing strategies available for bacterial TCSs, especially those with non-canonical components like the sensor-phosphatase driven TCSs. While finding AcP and/or CP as phosphodonors for a TCS RR might not seem surprising on the surface since they historically have been used to modify RRs in vitro, finding them as bona fide in vivo phosphodonors for a specific TCS RR in WT bacterial cells with an intact array of TCS is rare. Furthermore, we show that AcP and CP convey specific information for C. jejuni regarding nutrient availability in the environment and metabolic efficiency within the bacterial cell. These factors are especially important for C. jejuni with an amino acid-based metabolism that favors utilization of serine and glutamine, the substrates for AcP and CP synthesis. A deep and thorough interrogation will be required to determine what types of information non-canonical phosphodonors can provide for unusual TCSs and perhaps orphan RRs when they are true signaling inputs.

Materials and Methods

Bacterial strains, plasmids, and growth.

All bacterial strains and plasmids used in this study are listed in SI Appendix, Tables S2 and S3, respectively. Construction of most C. jejuni strains and mutants are described in detail in SI Appendix. C. jejuni strains were routinely grown from freezer stocks in microaerobic conditions (10% CO₂, 5% O₂ and 85% N₂) created by a tri-gas incubator on Brucella or Mueller Hinton (MH) agar containing 10 μg/mL trimethoprim (TMP) at 37 °C for 48 h. Strains were then restreaked onto Brucella agar with TMP unless otherwise stated and grown for an additional 16 h. Agar was added to Brucella or Mueller Hinton broth to 1.7% (w/v) to create agar media. Antibiotics were added to media when needed at the following concentrations: 15 μg/mL chloramphenicol, 100 μg/mL kanamycin or 0.1, 0.5, 1, or 2 mg/mL streptomycin. E. coli DH5α, DH5α/pRK212.1, and BL21 (DE3) were grown on LB agar or LB broth containing 100 μg/mL ampicillin, 100 μg/mL kanamycin,12.5 μg/mL tetracycline, or 15 μg/mL chloramphenicol when necessary. C. jejuni strains were stored at −80 °C in a mixture of 85% Brucella or MH broth and 15% glycerol. E. coli strains were stored at −80 °C in a mixture of 80% LB broth and 20% glycerol.

Genetic selection strategy to identify BumR phosphodonors.

The promoterless cat gene encoding chloramphenicol acetyltransferase from pRY109 was inserted into BsaBI-digested pPML849 (pUC19::peb3) (9, 63). A plasmid was recovered and verified by sequencing to result in pDRH8016, which resulted in the creation of a transcriptional reporter with the promoter of peb3 driving expression cat to result in chloramphenicol resistance. pDRH8016 was then electroporated into DRH212 (81-176 rpsL^Sm) to create DRH8126 (81-176 rpsL^Sm peb3::’cat) (64).

To remove bumS from DRH8126, pDRH8181 was first generated by inserting a SmaI digested kan-rpsL from pDRH437 into the HapI site of pPML107 (pUC19::bumS) (39, 49). pDRH8181 was then electroporated into DRH8126 and a kanamycin-resistant transformant was isolated and verified to result in DRH8204 (81-176 rpsL^Sm peb3::cat bumS::kan-rpsL). This strain was then electroporated with pPML334 (pUC19::ΔbumS) and streptomycin-resistant, chloramphenicol-sensitive transformants were isolated (49). Deletion of bumS was verified by PCR to result in DRH8228 (81-176 rpsL^Sm ΔbumS peb3::cat).

The ripley Tn was created by cloning an EcoRI-HindIII fragment containing the solo Tn from pFalcon that was blunt ended with T4 DNA polymerase into EcoRI-HindIII digested pUC19 (64). A plasmid was recovered and verified by sequencing to result in pFalcon2. A 913 bp region containing the ori from pACYC184 was amplified by PCR with primers that added 5’ PstI and 3’ KpnI restriction sites to the DNA. This fragment was then digested with PstI and KpnI and ligated into the PstI- and KpnI sites in the solo Tn in pFalcon2. The resulting plasmid and transposon were designated pNostromo and ripley, respectively.

Chromosomal DNA from DRH8228 was purified and used in in vitro transposition with the ripley Tn contained in pNostromo (65). After DNA transposition and repair, transposed DNA was introduced into DRH8228 by natural transformation. Tn mutants were recovered on Brucella agar containing 7-10 μg/mL chloramphenicol and 100 μg/mL kanamycin. Putative transposon mutants with increased peb3::cat expression were isolated due to formation of chloramphenicol-resistant colonies. The site of the Tn insertion in these mutants was determined as previously described.

Genetic screening strategy to identify BumR phosphodonors.

To identify genes encoding factors involved in production of a BumR phosphodonor, we employed a similar genetic strategy previously used to identify bumS and bumR with some modifications (9). Briefly, chromosomal DNA from PML921 (81-176 rpsL^Sm ΔastA peb3::astA-kan) was purified and used in in vitro transposition reactions with the darkhelmet Tn in pSpaceball1 as previously described (9, 65). After DNA transposition and repair, transposed DNA was introduced into the reporter strain PML921 by natural transformation. Tn mutants were recovered on Brucella agar containing chloramphenicol, kanamycin, 12.5 mM sodium butyrate, and 35 μg/ml 5-bromo-4-chloro-3-indolyl sulfate. Putative transposon mutants with increased peb3::astA expression presumably due to lower levels of P-BumR were identified as dark blue colonies in contrast to the white or light blue colony phenotype of WT 81-176 rpsL^Sm astA peb3::astA-kan. The site of the Tn insertion in these mutants was determined by plasmid rescue as previously described (65).

AstA transcriptional reporter assays.

For creation of C. jejuni peb3::astA transcriptional reporter strains lacking specific TCS histidine kinases, pDRH426, pDRH9051, pKNG607, pKNG609, pKNG618, and pKNG634 were electroporated into PML912 (81–176 rpsL^Sm ΔastA ΔbumS peb3::astA-kan) (9, 39). Transformants were selected on MH agar with chloramphenicol and verified by colony PCR to result in NR101 (81–176 rpsL^Sm ΔastA ΔbumS phosS::cat-rpsL peb3::astA-kan), NR104 (81–176 rpsL^Sm ΔastA ΔbumS racS::cat-rpsL peb3::astA-kan), NR107 (81–176 rpsL^Sm ΔastA ΔbumS dccS::cat-rpsL peb3::astA-kan), NR110 (81–176 rpsL^Sm ΔastA ΔbumS flgS::cat-rpsL peb3::astA-kan), NR113 (81–176 rpsL^Sm ΔastA ΔbumS cprS::cat-rpsL peb3::astA-kan), and NR842 (81–176 rpsL^Sm ΔastA ΔbumS cheA::cat-rpsL peb3::astA-kan).

Arylsulfatase assays were performed to measure the level of expression of peb3::astA on the chromosome of WT and mutant reporter strains as previously described (39). Strains for arylsulfatase assays were first grown from freezer stocks and then each strain was restreaked on Brucella agar and grown for 16 h at 37 °C in microaerobic conditions. Arylsulfatase assays were performed with each strain in triplicate. The level of peb3::astA expression in each strain was calculated relative to the expression in WT C. jejuni ΔastA strain which was set to 100 units.

Semi-quantitative real-time RT-PCR (qRT-PCR) analysis.

After growth of C. jejuni strains from freezer stocks on Brucella agar with 10 μg/mL trimethoprim and appropriate antibiotics, strains were restreaked onto Brucella agar and grown for an additional 16 h. C. jejuni growth was then suspended from agar plates in PBS and diluted to an OD₆₀₀ of 0.1 into 25 mL of Brucella broth alone or Brucella broth supplemented with 25 mM serine, glutamine, or pyruvate. Strains were grown statically at 37 °C in microaerobic conditions for 4, 6, or 8 h to isolate RNA. Total RNA was extracted with TRIzol (Ambion) and RNA was treated with DNaseI (Invitrogen). RNA was diluted to a concentration of 5 ng/μL before analysis. A one-step qRT-PCR was performed using MultiScribe Reverse Transcriptase (Invitrogen) and PowerTrack SYBR Green Master Mix (Applied Biosystems) with the QuantStudio 3 system (Applied Biosystems) following the ΔΔCt method. secD mRNA detection was used as an endogenous control since transcript levels were consistent across strains and conditions, and similar to expression levels of target genes. mRNA transcript levels in DRH461 grown in Brucella alone served as WT controls to determine relative gene expression.

Expression and purification of recombinant proteins.

Glutathione S-transferase (GST)-BumR recombinant proteins were purified as previously described with slight modification (9, 49). Briefly, E. coli BL21 (DE3) was transformed with pPML165 and then grown in 2xYT medium to mid-log phase prior to induction with 300 μM IPTG. GST-BumR protein was purified from the soluble fraction with glutathione Sepharose beads (GE Healthcare). Following cleavage of the GST tag by thrombin and removal of thrombin by benzamidine Sepharose following manufacturer’s instructions (GE Healthcare), recombinant BumR was recovered. Glycerol was added to a final concentration of 10% and proteins were stored at −80 °C.

In vitro analysis of BumR autophosphorylation.

In vitro autophosphorylation of BumR was analyzed by mixing 3 μg of recombinant BumR with 4 μl of 5X reaction buffer (250 mM Tris-HCl, pH 7.6, 600 mM potassium acetate, 100 mM MgCl₂, 5 mM DTT, 50% glycerol). Reactions were brought up to 20 μl with dH₂O alone or with 500 mM of lithium potassium acetyl-phosphate (Li-AcP; Sigma) or lithium carbamoyl phosphate dibasic hydrate (Li-CP; Sigma) and then incubated for 1 h at 37 °C. Samples were digested with trypsin overnight and then analyzed by short reverse-phase LC-MS/MS by the Orbitrap Fusion Lumos mass-spectrometry platform to identify peptides with phospho-aspartates. Identified peptides were analyzed by Proteome Discoverer 3.0 and searched against the C. jejuni protein database from UniProt and the BumR protein sequence. The abundance (as calculated by the sum of intensities) of each modified tryptic BumR_F50-R71 peptide that contains D58 is reported.

Electrophoretic mobility shift assay (EMSA) to assess BumR binding to promoter DNA.

A DNA fragment containing the peb3 promoter from −286 bases upstream to +35 downstream of the start codon was amplified by PCR. EMSAs were performed based on modification of a previously published protocol (49). Briefly, 0 to 1.5 μM of WT BumR in the presence or absence of 50 mM Li-AcP or Li-CP was incubated with ³²P-labelled DNA at 42°C for 20 min as indicated. After electrophoresis of samples on 5% acrylamide gels, DNA binding by BumR was analyzed by exposure of dried gels to a Typhoon FLA 9500 phosphorimager according to manufacturer’s instructions (Amersham Biosciences).

Immunoblotting analysis of C. jejuni BumR.

After growing C. jejuni strains from frozen stocks, strains were restreaked on Brucella agar and grown for 16 h at 37 °C in microaerobic conditions. Cells were resuspended in PBS and diluted to OD₆₀₀ 0.8. For whole-cell lysates, 1 mL samples were centrifuged and washed once with 1 mL of PBS. Pellets were resuspended in 50 μl of 1x Laemmli buffer with 5% BME and boiled for 10 min. For detection of BumR, 7.5 μl of whole-cell lysate were separated on a 4-20% TGX gradient gels (Bio-Rad). For detection of RpoA, 12.5 μL of whole-cell lysate were separated similarly. Immunoblots were developed by first applying a 1:2000 dilution of M166 murine antisera for BumR and 1:2000 dilution of GP275 guinea pig antisera for RpoA to membranes for 2 h (49, 66). Appropriate horseradish peroxidase-conjugated goat antibodies at 1:5000 dilution were used as secondary antisera. Immunoblots were developed with a Western Lightning Plus ECL kit (Revvity) and imaged using the Bio-Rad ChemiDoc system.

Significance.

Bacterial TCSs are information-processing networks usually linking cue detection by sensor kinases to phosphorylation of response regulators to alter behavior. For Campylobacter jejuni BumSR, the BumS sensor functions only as a phosphatase, requiring the BumR response regulator to use non-cognate phosphodonor(s) to form P-BumR. We discovered that C. jejuni exploits AcP and CP from amino acid metabolism as in vivo BumR phosphodonors to expand types of cues processed by BumSR. BumS sensing gut metabolites and BumR sensing AcP and CP simultaneously inform about geographical locations and nutrient richness within intestinal niches for optimal colonization. Our findings may suggest how other sensor phosphatase-driven TCSs and orphan response regulators can use non-cognate phosphodonors to convey cellular and environmental information for signal transduction.

Data availability.

All protocols and data discussed in the paper are available in the main text and SI Appendix. All strains and plasmids generated in this report will be made promptly available to readers.